a new emerging picoscale biotechnology

In: Protein Analysis Techniques Editors: Rakesh Sharma, Bharati D Shrinivas (C)2009 Innovations And Solutions, Inc. ___________________________________________________________________________

Lecture 3

SERUM PROTEOMICS FOR CANCER: FROM BASIC RESEARCH TO CLINICAL APPLICATION

Rakesh Sharma, Bharati D Shrinivas

ABSTRACT

Proteomic analysis based on the separation and identification of proteins/peptides is a po werful t ool for di scovering new b iomarkers. T wo-dimensional d ifferential gel electrophoresis ( 2D-DIGE), t he P roteinChip a rray®, C linProt®, a nd liq uid chromatography ar e k nown as s eparation t echnologies. Although t he d etection a nd identification of proteins/peptides could be performed using mass spectrometry (MS), the types o f M S used ar e d iverse an d t he ch aracteristics d epend o n t he ionization method (e.g., m atrix-assisted la ser d esorption/ionization [ MALDI], s urface-enhanced l aser desorption/ionization [ SELDI], a nd e lectrospray i onization [ ESI]) an d m ass an alyzer (e.g., t ime-of-flight [ TOF], qua drupole [ Q], and i on t rap [ IT]). The di verse pr oteomic systems that couple separation technologies with MS have been successfully applied to search for cancer biomarkers in biological fluid such as serum. Until now, many cancer biomarkers h ave be en r eported t o be i dentified f rom serum a nd pl asma. I n or der t o identify diagnostic biomarkers o f h epatocellular car cinoma ( HCC) an d ad ult T -cell leukemia (ATL), which are both human viral malignancies, we have been analyzing sera from patients using t he P roteinChip a rray®

-SELDI s ystem a nd M ALDI-TOF-TOF-MS. In t his r eview, we p resent a n o verview o f p roteomic t echnologies used t o s earch for biomarkers and cancer biomarkers discovered by these technologies, including our data on HCC and ATL. Cancer diagnosis using these proteomic biomarkers is also described.

Rakesh Sharma, Bharati D Shrinivas 2

INTRODUCTION Cancer r emains t he m ajor de vastating di sease t hroughout t he w orld. C urrently, i t i s

estimated th at c ancer k ills o ver 6 mill ion people pe r year worldwide, w ith over 10 m illion new c ases di agnosed e very year. D espite a dvances i n di agnostic i maging t echnologies, surgical m anagement, a nd t herapeutic m odalities, t he l ong-term s urvival i s poor i n m ost cancers. F or ex ample, t he f ive-year survival r ate is onl y 14% i n lung cancer and 4% in pancreatic cancer [1, 2]. The majority o f can cers ar e detected in t heir ad vanced s tages an d some h ave d istant metastases, w hich ar e t hought t o b e t he cau ses o f t he c urrent t reatment ineffectiveness.

It is widely accepted that early diagnosis and intervention are the best way to cure cancer patients [3, 4]. Cancer biomarkers provide diagnostic, prognostic, and therapeutic information about a particular cancer and show their ever-increasing importance in the early diagnosis of cancer [5, 6]. Over the past several decades, enormous efforts have been made to screen and characterize useful cancer b iomarkers. Several b iomarkers approved cl inically are available for e arly di agnosis a nd/or the s uccessful m onitoring of t reatment a nd r elapses a nd ha ve contributed s ignificantly t o reduce mortality r ates an d i ncrease o verall s urvival for can cers such as prostate cancer [5, 7]; however, unfortunately, most biomarkers are not satisfactory because o f t heir l imited s pecificity a nd/or s ensitivity [8–10]. T herefore, t here i s a n ur gent need to discover better potential biomarkers for cancer diagnosis in clinical practice.

Global an alyses h ave o pened n ew av enues f or can cer b iomarker d iscovery [11–13]. Genomics, t ranscriptomics, pr oteomics, a nd m etobolomics a re s tudies of a ll ge nes, transcripts, pr oteins, a nd metabolites, r espectively. From v arious s pecimens, such a s c ells, tissues, a nd body f luids, bi omarkers de tected di fferentially be tween c ancer pa tients a nd healthy in dividuals ha ve be en i nvestigated. In pa rticular, s erum/plasma i s a less-invasive specimen that is suited to cancer diagnosis. Because serum/plasma has a high protein content (60–80 m g/ml), a nalyzing all included proteins us ing pr oteomic a nalysis i s e fficient for diagnostic biomarker discovery.

Proteomic analysis is widely used to discover cancer biomarkers and is based on protein separation a nd i dentification t echnologies [ 14–17]. T o s earch f or di agnostic bi omarkers, proteins/peptides i n sp ecimen a re fi rst se parated w ith s eparation t echnologies. T hese expression pr ofiles a re t hen c ompared be tween c ancer a nd c ontrol gr oups. N ext, proteins/peptides di fferentially e xpressed unde r di sease c onditions c an be i dentified us ing mass spectrometry (MS). Using proteomic analysis of serum/plasma, many cancer biomarkers have been reported to be identified [15, 17]. Using proteomic technologies, we have also analyzed sera from patients with hepatocellular carcinoma (HCC) and adult T-cell leukemia (ATL), w hich a re both h uman v iral m alignancies [ 18, 19] . I n t his r eview, w e p resent a n overview of t he pr oteomic t echnologies us ed t o s earch f or bi omarkers a nd t he c ancer biomarkers discovered by these technologies, including our data on HCC and ATL. Cancer diagnosis using these proteomic biomarkers is also described.

Proteomic Analysis in Biomarker Discovery for Cancer Diagnosis 3

T able 1. C har acter istics of global analyses.

1. PROTEOMIC ANALYSIS OF SERUM/PLASMA IS EFFECTIVE TO SEARCH FOR CANCER DIAGNOSTIC MARKERS

There has been increasing interest in global analyses for the diagnosis of cancer [11–13].

Diagnostic biomarkers are currently being investigated throughout the world using genomic, transcriptomic, pr oteomic, and m etabolomic a nalyses. I n ge neral, a bi omarker w hose expression i s i ncreased o r decreased s pecifically u nder d isease c onditions i s r equired t o develop a new diagnostic method. Although specimen matrices such as serum/plasma, urine, and bi opsy t issues a re us ually a nalyzed, t he m ethod needs to be ch osen ap propriately depending on the developmental purposes, cancer types, and specimen.

1.1. Characteristics of Global Analyses and Diagnostic Availability of Biomarkers

Table 1 shows the characteristics of global analyses. Genomic analysis is based on DNA

sequencing of genes. By examining the whole genome sequence, we can obtain information about a ltered ge netic a nd e pigenetic e vents. S ingle nuc leotide pol ymorphisms ( SNPs) a re known a s ge netic events and thought t o be associated w ith disease s usceptibility and individual responses to drug treatment [20]. While many SNPs are silent with no direct effect on proteins as gene products, some affect the coding or regulatory (usually promoter) regions of a gene and change the function and expression of translated proteins. There is a report that SNPs o f th e liv er in testine-cadherin g ene ( CDH17) t rigger a berrant s plicing and a re associated with the developmental risk of HCC [21]. As epigenetic events, DNA methylation has b een s uggested t o b e as sociated w ith car cinogenesis [ 22]. Characteristically, D NA methylation does not change genetic information; however, it a lters the readability of DNA and results in gene inactivation. In particular, hypermethylation of CpG islands in promoter sequences is associated with the silencing of tumor suppressor genes and tumor-related genes by s ubsequent down-regulated e xpression of mRNA t ranscripts. S uppressors of c ytokine signaling 1 a nd 3 ( SOCS-1 and SOCS-3), w hich pl ay important r oles i n c ell gr owth a nd immune reaction, have been reported to be inactivated by hypermethylation in several cancers


[23]. From these f indings, i t i s s uggested t hat S NPs a nd D NA m ethylation c ould be us ed clinically as g enetic an d ep igenetic b iomarkers f or can cer d iagnosis, r espectively. Considering clinical applications, SNP and methylation arrays, by which it becomes easy to detect can cer-specific changes us ing extensive probe sets, might be useful. In addition, t he translated proteins whose expression, amino acid sequences, and function are altered by SNPs and methylation may also be available for diagnosis as protein biomarkers.

Transcriptomic analysis is based on DNA microarray and is used to investigate extensive gene ex pression at t he m RNA l evel [24–25]. G enerally, mRNAs ex tracted f rom cells an d tissues a re r everse-transcribed t o cDNAs and t hen hy bridized w ith c omplementary oligonucleotide DNA pr obes. The expression levels of each mRNA can be m easured by detecting hybridized cDNA. Because DNA probes for several tens of thousands of transcripts are fixed on the basal plate over a couple of centimeters, the expressions of all genes can be visualized s imultaneously i n j ust one experiment. I ndeed, us ing Genechip a rrays® (Affymetrix, Santa Clara, CA), the expression of about 50,000 transcripts can be examined. As part of the investigations into the ATL-developmental mechanism, Morishita’s group, in collaboration with us, applied DNA microarray to biomarker search [26]. The comparison of ATL cells derived from ATL patients with normal T-cells from healthy individuals showed that tumor suppressor in lung cancer 1 (TSLC1), which is a cell adhesion molecule, is over-expressed in ATL cells. While TSLC1 is a gene that associates with tissue invasion, its gene product (protein) may be also useful for ATL diagnosis. Cancer-specific genes found by DNA microarray analysis and these products could be used as diagnostic biomarkers [27–29].

Proteomic a nalysis ba sed o n t he s eparation a nd i dentification of pr oteins/peptides i s widely used to discover biomarkers, investigate cancer development, and to clarify the action mechanism of dr ugs [ 14–17]. T o s earch f or di agnostic bi omarkers, pr oteins/peptides i n a specimen are first separated with diverse separation technologies and the expression profiles are c ompared be tween di sease a nd c ontrol gr oups ( Chapter 2) . Next, pr oteins/peptides differentially expressed in disease s tatus can be identified us ing MS (Chapters 3 and 4). In clinical p ractice, n ovel d iagnostic m ethods ar e especially expected t o be developed us ing identified biomarkers (Chapter 5).

Metabolomics is the global study of all small molecular metabolites existing in cells and body fluids [30, 31]. Although metabolomic analysis consists of separation and identification processes a s i n pr oteomics, t he t arget of a nalysis i s s mall m olecular c ompounds but n ot proteins/peptides [32]. For metabolomic separation, gas chromatography (GC) [33], LC [34], and capillary electrophoresis (CE) [35] are mainly used. Compounds are identified with MS [33–35] and nuc lear m agnetic r esonance s pectroscopy ( NMR) [ 36, 37] , w hich a re coupled together with separation technologies. Metabolomic analysis contributes to the understanding of how metabolites a nd t heir c oncentrations c hange under ph ysiological c onditions. A t t he same time, application of th is analysis to the b iomedical f ield may lead to the d iscovery of novel drug candidates and diagnostic biomarkers [38, 39]. This analysis is most appropriate to search for biomarkers from urine, owing to the paucity of misplaced materials such as blood cells a nd pr oteins. R ecently, f rom more t han 2,000 m ass s pectra of ur ine metabolites, th e expression of several peaks has been reported to be increased in renal cell carcinoma [40].


T able 2. Diagnostic availability of biomar ker s identified by global analyses.

1.2. Advantages of Proteomic Analysis in the Search for Diagnostic Biomarkers

Genomic, t ranscriptomic, pr oteomic, metabolomic a nalyses pr ovide i nformation a bout

disease-specific genes (SNPs and methylation sites), mRNAs, proteins, and small molecular metabolites, respectively. Considering that these biomarkers are actually used for diagnosis in clinical p ractice, d eveloping m easurement s ystems f or t hem i s n ecessary. T he i mportant requirements f or di agnostic a vailability a re s implicity of t he de tection pr ocedure, hi gh throughput, specificity of biomarker detection, and good reproducibility.

Table 2 shows the diagnostic availability of biomarkers identified by global analyses. At present, protein/peptide biomarkers appear to best meet these requirements. In particular, the use of specific antibodies makes it possible to develop measurement systems such as enzyme-linked immunosorbent assay (ELISA), which is used widely in clinical practice (Chapter 5).

In contrast to proteomic biomarkers, genomic and transcriptomic biomarkers are difficult to make available for widely-used diagnostic systems at present. Genomic biomarkers such as SNPs a nd D NA m ethylation a re d etected by t roublesome pr ocedures, i ncluding D NA extraction from specimens and the amplification of target genes by polymerase chain reaction (PCR). Although its performance has been growing quickly recently, the throughput i s s till low for diagnostic a pplication. R ecently, a r apid S NP d etection s ystem na med s mart amplification process version 2 (SMAP 2) has reportedly been developed [41]; however, it is not c ommon yet in c linical p ractice. S imilarly, t ranscriptomic b iomarkers a re a lso p oorly utilized f or di agnosis. A lthough m RNA i s de tected by r everse t ranscriptase ( RT)-PCR o r Northern blotting, the method is tricky owing to mRNA flux. Collectively, it will take some time b efore t he s pecific d etection o f g enomic an d t ranscriptomic b iomarkers b ecomes common a s a c linical l aboratory t est; how ever, i f di fferential e xpression a nd S NPs c an be detected at the protein level, ELISA using specific antibodies will become a widely used diagnostic system.

Metabolomic b iomarkers ar e s mall molecular m etabolites. T hey ar e ex pected t o b e applied in the diagnostic system; however, metabolomic analysis is still under development,


in c ontrast t o o ther gl obal a nalyses. A lthough t he i dentification o f s mall molecular compounds is fundamental to the analysis, the database of mass and NMR spectra of metabolites i s not y et c omplete. I n a ddition, r eference c ompounds f or confirming identification are also lacking. Also, MS and NMR need extensive skills for their operation. Further, specific detection methods must be established with respect to each target compound; therefore, the development of a widely used diagnostic system using metabolic biomarkers is difficult. Recently, antibodies against small molecular compounds have been reported [42], thereby leading to the development of methods, although the specificity of antibodies may be low owing to few antigen determinants in the chemical structures of compounds.

As described above, measurement systems using protein biomarker appear to be effective for t he de velopment of di agnostic m ethods. C onsequently, i t i s r easonable t hat pr oteomic analysis using MS is focused on search for diagnostic biomarkers.

1.3. Specimen Early d iagnosis is very i mportant t o r educe c ancer m ortality. P rotein bi omarkers i n

cancer di agnosis c an be obtained mainly from serum/plasma, ur ine, and pa thologic t issues. Serum/plasma is a l ess-invasive specimen and is suitable for the early diagnosis. Because it constantly perfuses tissues, a certain abnormality in the living body is believed to be reflected in the expression and processing of contained proteins; for example, increased serum levels of prostate-specific antigen (PSA) and cancer antigen (CA) 125 are used to detect cancer in the prostate a nd ov ary, r espectively [43, 44] . Urine i s n on-invasive a nd us eful f or bi omarker search i n ad dition t o s erum/plasma. Wh ile u rine b iomarkers ar e ef ficient, es pecially f or urinary d iseases s uch as b ladder ca ncer [ 45], t hese co ncentrations i n ur ine un dergo m any changes depending on diet, intake of water and drugs, and diseases. This is the difficulty in urine a nalysis. P athologic t issues a re f requently us ed in t he de finitive di agnosis of s olid cancer s uch as HCC and kidney car cinoma; h owever, t he pa thologic di agnosis i s hi ghly invasive a nd uns uited t o e arly di agnosis. From t hese c haracteristics of s pecimens, i t i s suggested t hat p roteomic a nalysis o f s erum/plasma i s ef fective in s earching f or can cer diagnostic biomarkers.

High protein content (i.e., 60–80 mg/ml) in serum/plasma is attracting increasing interest in proteomic analysis. The expectation is that the characterization of thousands of individual proteins/peptides in serum/plasma will enable the discovery of a number of reliable disease biomarkers; however, t here is a di fficulty in the proteomic analysis of s erum/plasma. O ver 99% pr otein content i n s erum/plasma i s dom inated b y 30 pr oteins, s uch a s a lbumin, immunoglobulins, transferrin, ha ptoglobulin, complement C3, and l ipoproteins [46, 47] (Table 3). Serum/plasma proteins are present across an extraordinary dynamic concentration range that is likely to span more than 10 orders of magnitude. This dynamic range, namely the existence of hi ghly a bundant pr oteins, makes t he de tection o f l ow a bundant pr oteins ( the remaining 1%) extremely challenging. There are two remedies for this difficulty. The first is the removal of highly abundant proteins by affinity columns using antibodies against them. The s econd i s an alysis ai med at p eptides b ut not pr oteins. R egardless of t his difficulty, considering t hat b iomarkers d iscovered f rom s erum/plasma ar e d irectly a vailable f or l ess-invasive diagnosis, serum/plasma is an attractive specimen as the source of diagnostic biomarkers.


T able 3. H igh abundant proteins in plasma der ived from healthy per sons

a

Accession No. from Swiss-Prot knowledge base.


2. SEPARATION TECHNOLOGIES IN PROTEOMIC ANALYSIS Proteomic a nalysis ba sed on t he s eparation a nd i dentification of pr oteins/peptides is a

powerful t ool f or di scovering ne w bi omarkers. T o s earch f or di agnostic bi omarkers f rom serum/plasma, d etecting p roteins/peptides e xpressed d ifferentially in d isease is e ssential. Two-dimensional differential gel electrophoresis (2D-DIGE), ProteinChip array®, ClinProt®, and LC are typical separation technologies for biomarker discovery (Figure 1).

2.1. 2D-DIGE The gol d s tandard for protein separation i n p roteomic a nalysis i s 2 D-DIGE ( Figure 1A)

[48, 49] . First-dimensional s eparation is ach ieved b y i soelectric f ocusing, w hich s eparates proteins on t he basis of their charge. Namely, protein samples are applied to immobilized pH gradient (IPG) s trips. The IPG s trip is typically polyacrylamide gel in which the pH gradient (e.g., pH3–10) is immobilized. When the focusing voltage is applied to strips, the movement of a p rotein in a pH g radient i s determined solely b y i ts position relative to its isoelectric point (Figure 1A a). I f the pos ition of t he pr otein i n t he gr adient i s a t a pH t hat i s gr eater than i ts isoelectric point (pI), its net charge is negative, due to the net effect of dissociated carboxylate anions (negative charge) and uncharged amines in the protein; therefore, the protein migrates towards the positive electrode. If the position of the protein in the gradient is at a pH that is less than i ts p I, i ts ne t c harge i s pos itive, due t o t he ne t e ffect of unc harged c arboxylic acid a nd protonated amines (positive charge). The protein then migrates towards the negative electrode. Any migration continues until the protein migrates to the position in the pH gradient that has no net charge. On this principle, the protein is focused at its pI on the IPG strip.

Isoelectric f ocusing i s c oupled w ith s econd-dimensional s eparation, w hich e xploits sodium dodecylsulfate-polyacrylamide gel electrophoresis (SDS-PAGE) to separate proteins in accordance with molecular weight (Figure 1Ab). Under optimal conditions, the expression pattern of s everal t housands of i ndividual pr oteins c an be vi sualized s imultaneously on a single 2D gel. The resulting gel shows the proteins separated according to pI across the x-axis of the 2D display and according to molecular weight across the y-axis.

↓↓↓↓↓↓↓ ↓↓↓

(a) First-dimensional separation; Isoelectric focusing

Focusin

g T

ime

pH3 pH10

++ -C

=O

OH

NH2

H ＋C

=O

OH

NH2

H ＋＋ C

=O

O

NH2

−C

=O

O

NH2

−−

IPG strip

↑

pI

++ -

↑

C

=O

O

NH2

H ＋

−C

=O

O

NH2

H ＋＋

−−

→

→

(b) Second-dimensional separation; SDS-PAGE

Protein

IPG strip

Isoelectrically focused proteins↓ ↓↓

SDS-polyacrylamide gel →SDS-PAGE

pH3 pH10

Mo

lecu

lar

We

igh

t

High

Low

● ●

•

•

•

•●

●

•

●

●

●

●

•●

●•

•

••

•

••••

●

●•

•••

•

•••●

↓

↑

(A) 2D-DIGE: Protein sample is separated by isoelectric focusing (a) as first-dimensional separation, followed by SDS-PAGE (b) as second-dimensional separation. Protein spots in separation gels are visualized by CBB-, silver-, or fluorescent-staining.


Incubation

ProteinChip

Patients

2

3

4

1

Healthy

individuals

→→→

→

2

3

4

1→→

→

→

Sample

addition

■

□△

▲

▽▼◎

●

◉◇

◆

Just after sample addition

▼

◎

◆

□△

△

→

○

■

□△

▲

▽▼◎

●

◉◇

◆

▼

◎

◆

○

□△

△

○

Patients

Healthy individuals

Incubation→

Wash→

Wash→

Peak detection

→with MS

Peak detection

→with MS

■ ■

□

□

□

□

◎ ◎

◎ ◎○○ ○

▲

▼

▲

▼

◉

◉

◆

◆

▽

▽

◇

◇

△

△

△

△

◆

◆

●

●

Adsorption of proteins/peptides

having affinity with functional group

Functional group Hydrophobic, cationic, anionic,

metal ion presenting or hydrophilicProteins/peptides

◎ ◎○○ ○◉ ◇●

■ ■◎ ◎◉ ◇●

Removal of non-adsorbed

proteins/peptides

Pe

ak I

nte

nsity

mass

Pe

ak I

nte

nsity

mass

Pe

ak I

nte

nsity

mass

Biomarker peakBiomarker peak


↓

↓

↑

↑Differential analysis

(B) ProteinChip array®: Biological fluids, such as serum/plasma and urine, are treated with ProteinChip arrays®. Proteins/peptides in fluids are adsorbed on spot surfaces modified with a functional group (hydrophobic, cationic, anionic, metal ion presenting, or hydrophilic). After removing non-adsorbed proteins/peptides by washing, peaks derived from those adsorbed on the surfaces are detected with MS. For example, spectral comparison between diseased patients and healthy individuals could lead to the discovery of biomarker peaks.

■

□

△

▲

▽▼

◎

●

◉

◇

◆

▼

◎

◆

□

△

△

■

■

□

△

▲

▽

▼

◎

●

◉

◇◆

▼

◎

◆

□

△

△■

■ ◎

●◎

△◇

■

●

Magnetic beadsHydrophobic, cationic, anionic,

or metal ion presenting

Incubation→

■

□

△

▲

▽▼

◎

●

◉

◇

◆

▼

◎

◆

□

△

△

■

■

□

△

▲

▽

▼

◎

●

◉

◇◆

▼

◎

◆

□

△

△■

■

◎

●

◎△

◇

■

●

→Wash

■ ◎

●

◉

◇

◎

■

◎

●

◉

◇

◎

■

■

◎

●

◎

◇

■

●

→Elution

■

◎

●

◉◇◎

■

●◉◇◎

■

■

◎ ●

◎◇

■

●

Peak detection

with MS

→

Pe

ak I

nte

nsity

mass


↓P

ea

k I

nte

nsity

mass


↓

↑↑Differential analysis

Patients

Healthy individuals

Proteins/peptides

Mix of sample (serum)

and magnetic beads

Adsorption of

proteins/peptides

having affinity with

functional group

Removal of non-adsorbed

proteins/peptides

Harvest of adsorbed

proteins/peptides

(C) ClinProt®: Biological fluids are treated with ClinProt beads which have chromatographic surfaces (hydrophobic, cationic, anionic, or metal ion presenting). Proteins/peptides in fluids are adsorbed on bead surfaces. After washing, peaks derived from proteins/peptides adsorbed on bead surfaces are detected with MS.

g p g

Crude protein

extract

Digestion with

proteolytic enzyme

(e.g., trypsin)

→ Peptide mixture

HPLC or nano LC

→

InjectionStrong cation

exchanger

1st separation 2nd separation

Reverse phase

→ MS

Eluted peptides

(D) Shotgun proteomics using LC: Crude protein extract containing intact proteins is digested with proteolytic enzymes such as trypsin and then generated peptide mixture is subjected to HPLC or nano LC. The expression and identification of peptides in the eluate are assessed with MS.

Figure 1. Protein separation technologies for biomarker discovery. :


Protein profiles can be compared using sophisticated software packages to detect protein spots t hat ar e d ifferentially ex pressed b etween s amples. I n g eneral, p rotein s pots a re visualized by Coomassie brilliant blue (CBB) or silver staining [49]; however, CBB staining has low sensitivity and requires about 100 μg proteins per gel for spot detection, suggesting that th is s taining is n ot s uited to p roteomic a nalysis of v aluable s pecimens. Although detection by silver staining can be performed with a small amount of proteins (about 10 μg), irregular staining is often seen, leading to low reproducibility and quantification, which can be r esolved us ing f luorescent dyes [50, 51] . F luorescent 2D -DIGE has high sensitivity and enables m ultiple pr otein e xtracts t o be s eparated on t he s ame 2D ge l b y l abeling proteins using dy es w ith di fferent e xcitation a nd e mission w avelength pa irs. F urthermore, because proteins a re fl uorescently l abeled be fore e lectrophoresis, i t i s not necessary to consider t he staining i rregularity of ge ls. R ecently, f luorescent 2D -DIGE ha s b ecome pop ular f or proteomic analysis [51].

2.2. ProteinChip Array® ProteinChip array® (Bio-Rad Laboratories, Inc., CA, USA) has been successfully applied

to di agnose v arious m alignancies, s uch a s ov arian c ancer a nd ga stric c ancer [ 52, 53]. As shown i n F igure 1B , ProteinChip a rray® is a c onvenient s eparation to ol th at utilizes chromatographic properties and i ts surfaces can be modified with various functional groups (hydrophobic, c ationic, a nionic, m etal i on pr esenting a nd h ydrophilic). In t his m ethod, biological fluids, such as serum/plasma and urine, are directly applied to spot surfaces. After a s eries of bi nding a nd w ashing s teps, pr oteins/peptides a nd a ny c ontaminants t hat do not bind to the surface are removed. Protein mixtures retained on the surface are then analyzed with MS. In the binding step, the change of parameters (pH, salt concentration, and organic solvent concentration) in buffer solution alters the affinity between proteins/peptides and the surface, leading to the de tection of a variety of proteins/peptides. A lso, pretreatment of the specimen w ith pr otein de naturants ( e.g., urea) a nd de tergents ( e.g., CHAPS) a ffects the variation. B ecause d ifferential an alysis can b e p erformed w ith a s mall am ount o f serum/plasma (a few microliters), P roteinChip array® is useful in biomarker searches us ing valuable specimens; however, the l imitations of analysis include high cost and the fact that MS used in analysis is exclusive (Chapter 3.2).

2.3. ClinProt® ClinProt® (Bruker Daltonics, Bremen, Germany) is a purification method based on affinity

beads (Figure 1C) [54, 55] . This method uses chromatographic surfaces (hydrophobic, cationic, anionic, a nd metal i on pr esenting) on a n ou ter layer of m agnetic be ads. C ertain s ubsets of proteins/peptides, which have affinity to the surface, can be selectively purified, allowing unbound impurities to be removed by washing with buffers. Proteins/peptides bound to magnetic beads are then eluted and directly analyzed by MS. The technical performance of affinity bead purification is similar to that of ELISA, and it can be used to process many samples in parallel. In addition, the cost of analysis is low and data can be obtained with diverse types of MS.


Inlet Ion source Mass analyzer Detector Control system Data system

Vacuum system

Figure 2. Block diagram of MS. Sample inlet, ion source, mass analyzer, detector, vacuum system, control system, and data system are major components of MS. The mass analyzer, detector, and parts of the ion source are maintained under vacuum. The control system monitors and controls all parts of the instrument. Data are recorded by the data system.

2.4. Shotgun Proteomics Using LC LC i s a s eparation t echnology us ed c ommonly in t he i solation of v arious biological

molecules, i ncluding pr oteins/peptides. The L C el uate i s d irectly an alyzed u sing M S i n proteomic analysis [56, 57] (Figure 1D). Because intact proteins cannot be identified owing to the measurement l imits of MS, crude protein extract is digested in advance with proteolytic enzymes, such as t rypsin, with the a im of directly identifying the LC eluate. The generated peptides, de rived f rom a w ide v ariety of pr oteins i n e xtract, a re s ubjected t o L C a nd t hen sequences of pe ptides i n t he el uate ar e continuously determined us ing M S. S imilarly, a method that identifies digested peptides in complex mixtures using a combination of LC and MS i s r eferred t o a s “ shotgun pr oteomics”. T he na me i s de rived f rom D NA s hotgun sequencing, which is named for its analogy to the quasi-random firing pattern of a shotgun. Considering that LC eluate is directly analyzed with MS, in which involatile buffering agents and salts inhibit analysis, elution using a volatile buffer is necessary for peptide separation; therefore, elution with a w ater-acetonitrile gradient using reversed phase chromatography is mainly used to separate peptides prior to entering the MS. Furthermore, to enhance separation capacity, multidimensional LC ( MDLC), w hich c ombines t wo or more fo rms o f L C i n shotgun a nalysis, ha s thrived ov er t he pa st f ew years [58]. O ne of t he m ost w idely us ed combinations for MDLC is strong-cation exchange and reversed phase. In addition to the type of column, the flow rate is also important for separation. High performance LC (HPLC) and nano LC are generally used for peptide separation. In particular, nano LC is efficient for separating valuable specimens. In contrast to HPLC, its flow rate is extremely slow (typical flow r ate: 20 nl /min) [ 49]. This s low rate enables s amples t o b e s eparated i n a co ndensed state, indicating that analysis requires only a small amount of specimen.

3. MASS SPECTROMETRY Proteins/peptides separated using diverse technologies can be identified using MS, which

is now the gold standard for identification. In this chapter, MS used in proteomic analysis is described.


3.1. Fundamentals of MS A block di agram of a ba sic MS i s shown in Figure 2. T he s ample inlet, i on s ource, mass

analyzer, detector, vacuum s ystem, control s ystem, and d ata system a re the major components [49]. Control system monitors and controls the other components. The mass analyzer, detector and parts of the ion source are maintained under vacuum. MS equipment is mainly defined by details of the ion source and mass analyzer. Analytes introduced from the sample inlet a re ionized b y various ionization methods and then ions are separated by a mass analyzer. Masses and quantities of their ions are shown as the mass-to-charge ratio (m/z) and peak intensity, respectively.

3.2. Ionization Methods Ionization methods for biological macromolecules, such as proteins/peptides, were developed

in the 1980s. Matrix-assisted laser/ionization (MALDI) is a soft ionization technique used in the analysis of proteins/peptides, which are fragile when ionized by conventional methods [59, 60]. As s hown i n Figure 3A , pr oteins/peptides a re d issolved i n a s olution of a U V-absorbing compound, referred to as the “matrix”, and placed on sample stage. For peptide ionization, α-cyano-4-hydroxycinnamic a cid ( CHCA) i s w idely used as the matrix. A lso, p rotein is ionized using s innapic a cid ( SA) a nd 2,5 -dihydroxy acetophenone ( DHAP). A s t he s olvent d ries, t he matrix compound c rystallizes and proteins/peptides m olecules are i ncluded i n m atrix c rystals. Pulses of UV laser l ight a re used to vaporize the matrix and the included proteins/peptides a re carried into t he ga s pha se. B ecause most matrixes a re a cidic c ompounds a nd d ilute acid ( e.g., trifluoroacetic acid; TFA) is added to the solvent, ionization occurs by protonation in the acidic environment. W hile monomolecular p rotons a re no rmally a dded to a p rotein/peptide, multicharged i ons a re oc casionally ge nerated. T hese ge nerated i ons a re t hen i ntroduced i nto a mass a nalyzer. S urface-enhanced l aser desorption/ionization (SELDI) is a nother io nization method that modifies MALDI and is a specialized method for ProteinChip array® (Chapter 2.2).

Along w ith M ALDI, t he ionization m ethod us ed w idely i n pr oteomic a nalysis i s electrospray ionization (ESI) [61, 62]. Although the mechanism underlying ESI has not been fully clarified, its well-regarded model is illustrated in Figure 3B. In ESI, an acidic solution containing pr oteins/peptides i s s prayed di rectly i nto t he M S i nlet t hrough a s mall-diameter needle. A high, positive voltage is applied to this needle and droplets of the solution are sputtered f rom t he t ip. P rotons f rom a cidic c onditions gi ve t he dr oplets a pos itive c harge, causing then to move from the needle towards the negatively charged instrument (relative to the ne edle). D uring t his m ovement, the s ize of t he dr oplets i s r educed by e vaporation, a process t hat c an be a ided by a f low of ni trogen ga s a nd he at. Protonated pr oteins/peptides desorbed f rom dr oplets t o t he ga s pha se a re i ntroduced i nto t he m ass a nalyzer. Characteristically, a significant portion of ions produced by ESI is multiply charged. Because proteins/peptides dissolved in l iquid can be analyzed di rectly, ESI i s coupled together with LC and the eluate obtained by solvent gradient is analyzed in waves.


Laser light

H＋H＋

H＋H＋

H＋H＋

H＋H＋

H＋H＋

H＋H＋

H＋H＋

H＋H＋

Sample

stage

H＋H＋

Mass

analyzer

H＋

H＋

H＋H＋

H＋

H＋

H＋

H＋H＋

H＋

Double-charged ions

Proteins/peptides ions (single charge)

Matrix ions and neutrals

Crystallized matrix

with included

proteins/peptides

(A) MALDI: The matrix crystals, including proteins/peptides, are vaporized by irradiating pulses of UV laser light. Together, the included proteins/peptides are also carried into the gas phase. Ionization occurs by protonation in the acidic environment and then generated ions are introduced into the mass analyzer.

●

● ●●● ●

● ●

●●

●●●●

●●●

●●

● ●●LC→

＋＋＋

＋＋

＋＋＋＋

＋＋＋

＋

＋＋＋

＋＋＋

＋＋

＋＋＋＋

＋＋＋

＋

＋＋＋

＋＋＋

＋＋

＋＋＋＋

＋＋＋

＋

＋＋＋

＋＋＋＋＋＋＋＋＋

＋＋＋

＋＋＋＋＋＋＋＋＋

＋＋＋

＋＋＋＋＋＋＋＋＋＋＋

＋＋＋＋＋＋＋＋＋＋＋

＋＋＋＋＋＋＋＋＋＋＋

＋＋＋＋

＋＋＋＋

＋＋＋＋

＋＋＋＋

＋＋＋＋

＋＋＋＋

＋＋＋＋

＋＋＋＋

＋＋＋＋

＋＋＋＋

＋＋

＋＋＋＋

＋＋

＋＋

●＋●＋

●＋＋●＋

＋

●＋●＋

●＋＋＋●＋＋＋

●＋＋＋＋●＋

＋＋＋

●＋＋＋●＋

＋＋

●＋＋●＋

＋

●＋＋＋●＋＋＋

●＋＋＋＋●＋

＋＋＋

●＋＋●＋

＋

●＋●＋

●＋●＋

●＋＋●＋

＋

●＋＋●＋

＋＋

＋

＋＋

＋＋

Mass

analyzer→

Desolvated ionsCharged droplets

Heated desolvation region

Capillary column

effluent

High voltage

(+2 kV to +5 kV)

(B) ESI: A: solution that contains proteins/peptides is sprayed directly into the inlet of MS through a small-diameter needle. The solution is acidic and a positive voltage is applied to the needle. Droplets from the needle are given a positive charge by the addition of protons and then move towards the instrument. During this movement, the size of droplets is reduced by evaporation and then protonated proteins/peptides desorbed from droplets are introduced into the mass analyzer. In ESI, a significant portion of ions is multiply charged.

Figure 3. Ionization methods in proteomic analysis.


(a) Linear type

Laser

Sample

stage

→

Acceleration

region

↑+25 kV

↑+0 kV

↑Detector

Ion flight path in field-free region

Laser

Sample

stage

→

Acceleration

region

↑+25 kV

↑+0 kV

Ion flight path in field-free region

Reflectron

↑Detector

(b) Reflector type

(A) TOF mass analyzer: Ions are given a fixed amount of kinetic energy in the acceleration region and then travel in a field-free region. Because their velocities are inversely proportional to their m/z, ions with low m/z travel more rapidly than ions with high m/z. Using the time required for the ion to travel, its m/z is calculated. In addition to the linear type (a), the reflector type (b) is used as a mass analyzer. The use of a reflectron reduces accidental error in the velocity of traveling ions and increase of flight distance, leading to high resolution analysis.

+

+-

-+

-++

++

++

++--

++

Metal rods (quadrupole)

↑Ions

→Detector

(B) Q mass analyzer: Ions pass into the electrostatic field created by electric potentials applied to four parallel metal rods (quadrupole). Pre-selected ions make it through the inside diameter of the quadrupole and strike the detector (circular ion). All other ions are deflected out of the quadrupole (triangular ion).

Trapping

region

Ring electrode

Ring electrode

Endcap electrode

(Ion entrance)Endcap electrode

(Ion exit)

DetectorIons

(C) IT mass analyzer: All ions are initially trapped in the trapping region. Trapped ions are ejected out of the region and into the detector by changing the electrostatic field. Because all ions injected into the trap are sent to the detector, the detection sensitivity is extremely good.

Figure 4. Mass analyzers


3.3. Mass Analyzers Mass analyzers separate ions according to their mass by applying electric and magnetic

fields. Although the unit for their mass is the mass-to-charge ratio (m/z), care should be taken with the interchangeable use of “mass” and “m/z” because these values are not the same for any ion that is multiply charged. For example, a peptide ion with a mass of 1,000 Da that is doubly charged by the addition of two protons would be observed as an ion at 501 m/z. It is critical to remember this distinction, particularly when ESI, by which multicharged ions are mainly generated, is being used (Figure 3B). Time-of-flight (TOF), quadrupole (Q), and ion trap (IT) mass analyzers are used in MS for protein/peptides analysis [49].

The operating principles of the TOF mass analyzer are simple. As shown in Figure 4Aa, ions are first given a fixed amount of kinetic energy by acceleration in an electric field generated by a high voltage, typically + 20 kV t o + 30 kV ( for pos itive i ons such as peptides). Following acceleration, ions enter a field-free region where they travel at velocities based on a law. Namely, their velocities are inversely proportional to their m/z; therefore, ions with low m/z travel more rapidly than ions with high m/z. The time required for the ion to travel the length of the field-free region is measured and used to calculate the m/z of the ion. To obtain a higher resolution, while longer flight tubes or higher accelerating voltages are effective, an ion reflector, also referred to as a “reflectron”, is built into high resolution MS. As shown in Figure 4Ab, the reflectron is an ion mirror created by an electric field and reverses the flight path of the ion. In theory, ions with the same mass travel in a flight tube with the same velocity; however, in fact, there is accidental error in t he ve locity. F or example, ions w ith higher ve locity t han the t heoretical va lue penetrate the flight tube further. In the reflectron, higher-velocity ions take longer to reverse the flight path. This process reduces the error in the velocity. Also, another effect of a reflectron is the increased flight distance, which also leads to better resolution.

The Q mass analyzer consists of four parallel metal rods (quadrupole) through which ions pass (Figure 4B). When opposite and neighbor rods are subjected to electric potentials having the same and different polarities, respectively, an electrostatic f ield of specific s trength and frequency is created inside the quadrupole. The path of ions is influenced by the fluctuating electric field. Only ions of pre-selected m ass make it through the inside di ameter of the quadrupole and strike the detector (circular ion in Figure 4B). All other ions are deflected out of the quadrupole and never reach the detector (triangular ion in Figure 4B). Further, the Q mass analyzer can scan pre-selected masses at rates on the order of 1000 m/z per sec so that the acquisition of a mass spectrum between 200 m/z and 2000 m/z takes 1.8 s ec. This high-speed scan is suitable for acquiring data on LC, from which analytes are continuously eluted.

The IT mass analyzer works on the same physical principle as the Q mass analyzer. The term “ion trap” is derived from the fact that electrostatic fields are applied so that ions of all m/z are initially trapped in the trapping region (Figure 4C). Trapped ions are ejected out of the trapping region and into the detector by changing the electrostatic fields dependent on the m/z o f io ns. A n important c haracteristic o f I T is th at t he detection sensitivity is e xtremely good, because of the efficient t ransmission and utilization of ions formed by an ion source. The overall dimensions of the trap are such that the distance from the trapping region to the detector is only a few centimeters. In the course of the trapping and ejection process, all ions injected into the trap can ultimately be sent to the detector; therefore, the compact size of ion traps and the detection of all ions injected lead to good sensitivity.


3.4. Types of MS Used in Proteomic Analysis As described above, although the types of ion source and mass analyzer are diverse, MS

is composed of their various combinations; for example, MALDI-TOF-MS, SELDI-TOF-MS, ESI-Q-MS, a nd E SI-IT-MS a re us ed i n pr oteomic a nalysis. Furthermore, t o pe rform the measurement with good m ass accuracy and high resolution, tandem MS, in which multiple mass an alyzers ar e co nnected i n s eries, i s used w idely at p resent. Tandem M S, s uch a s MALDI-TOF-TOF-MS ( Figure 5 A), E SI-Q-TOF-MS ( Figure 5B ), a nd ESI-IT-TOF-MS (Figure 5C), is indispensable for the identification of proteins/peptides.

4. IDENTIFICATION OF PROTEINS/PEPTIDES USING MS The i dentification of pr oteins/peptides is an essential step in the f ield of p roteomics.

Examining protein expression patterns alone is, of course, important for diagnosis in various diseases; however, without identifying the proteins/peptides associated with diseases, it is not possible t o de lve i nto t heir f undamental c auses or t o develop c linical a pplications, s uch a s ELISA, t hat r equire a ntibodies a gainst bi omarkers. T he i dentification of pr oteins/peptides using MS is the key step in the development of clinical applications using proteomic biomarkers.

4.1. Protein Identification by Peptide Mass Fingerprinting Peptide m ass f ingerprinting ( PMF) i s a n a nalytical t echnique f or pr otein i dentification

and was developed in 1993 by several groups independently [63–67] (Figure 6A). In PMF, after protein separation, the protein of interest is first digested with proteolytic enzymes such as t rypsin. The m/z of generated pe ptides is accurately measured w ith MS. These act ual measurement v alues ar e t hen in silico compared w ith theoretical v alues of know n pr oteins registered in a database. A s da ta on know n proteins, registered proteins are t heoretically digested into peptides with enzymes and the absolute masses of these peptides are calculated. By comparing the actual measurement values of the interest protein with theoretical values of the e normous numbers of p roteins r egistered i n t he da tabase s tatistically, the be st-matched protein is searched. Databases of the National Center for Biotechnology Information (NCBI) and S wiss-Prot a re po pular f or identification. NCBI i ncludes m ost of t he publ ic dom ain sequence databases an d t he S wiss-Prot da tabase ha s a large a mount of a nnotation t ogether with the sequences to allow for easier understanding of the functionality of the identified proteins [ 68]. T he advantage of PMF i s t hat onl y t he m/z of ge nerated pe ptides h as t o b e examined. De novo peptide s equencing, w hich i s t ime-consuming, i s not ne cessary. A disadvantage is that the interest protein has to be registered in a database. Additionally, most PMF a lgorithms a ssume t hat t he pe ptides c ome f rom a s ingle pr otein [ 69]; t herefore, t he typical PMF samples are proteins isolated by 2D-DIGE and MDLC that have high separation capacity.


4.2. Peptide Identification by MS/MS Ion Search Mass analysis is essentially a separation of ions according to their m/z. Because tandem

MS h as m ultiple mass an alyzers t hat ar e co nnected i n s eries, t wo s tages o f m ass an alysis (MS/MS analysis) can be carried out in a s ingle experiment. An ion with a specific m/z i s selected i n t he f irst s tage of mass a nalysis a nd s tructural i nformation a bout s elected i on i s obtained in the second s tage. In the case of peptide ions, the information is the amino acid sequence of the peptide. At present, peptides with a molecular weight of less than ~3,000 can be routinely identified by MS/MS ion search using tandem MS, such as MALDI-TOF-TOF-MS, E SI-Q-TOF-MS, a nd E SI-IT-TOF-MS ( Figure 5) . M S/MS i on s earch i s a vailable t o identify peptides de tected by di fferential analyses us ing P roteinChip a rray®, ClinProt®, and shotgun proteomics, and to confirm the sequences of peptides used in protein identification by PMF.

MS/MS ion search i s ba sed on t he f ragmentation r eaction of peptides and by da tabase search [49, 70, 71] (Figure 6B). Namely, peptides ionized by ionization methods, including MALDI and E SI, ha ve excess i nternal energy. After a peptide of interest (parent i on) i s selected in the first mass analyzer, its peptide bonds are randomly cleaved by internal energy. While t he f ragmentation r eaction o ccurs i n i ons w ith excess e nergy, most i ons f ormed b y MALDI an d E SI h ave i nsufficient i nternal en ergy an d ar e s table t o d rive an y f ragment reactions. Thus, what is required for MS/MS analysis of peptides is a method to energize a stable p arent io n a fter it h as b een mass-selected and to induce f ragmentation r eaction. T he method used most often is collisionally induced dissociation (CID), in which a mass-selected ion is transmitted into a high-pressure region of the tandem MS where it undergoes a number of collisions with molecules of f iller ga s such a s a rgon. A s the ion collides with these gas molecules, a portion of the kinetic energy of the ion is converted into internal energy to make the ion uns table and drive the f ragmentation reaction. In the Q-TOF mass analyzer (Figure 5B), a fter a n io n is s elected in th e f irst s tage ( Q m ass a nalyzer), it is f ragmented in th e collision cell. Also, in the IT-TOF mass analyzer (Figure 5C), the trapping region of IT serves as the cell. In contrast, in MALDI-TOF-TOF-MS (Figure 5A), CID is not often used for fragmentation. T he s elected i on de cays i nto f ragments w ithout C ID a fter l eaving t he acceleration region of the ion source; therefore, this f ragmentation reaction is referred to as “post-source decay (PSD). In PSD, the reaction occurs mainly in the second stage of the mass analyzer (TOF2 in Figure 5A). The fragment ions produced are m/z analyzed in the second stage. B oth t he i ntact pe ptide m ass a nd f ragment i on m asses a re us ed f or pe ptide identification through database searching (Figure 6B).

5. CANCER BIOMARKER AND DIAGNOSIS Over t he p ast s everal d ecades, en ormous ef forts h ave b een m ade t o s creen and

characterize u seful can cer b iomarkers. A s cl inically i mportant b iomarkers, PSA, carcinoembryonic a ntigen ( CEA), α-fetoprotein ( AFP), C A125, C A15-3, and C A19-9 ha ve been identified [8–10]. Although they are commonly employed in clinical diagnosis, they are not satisfactory because of their limited sensitivity and/or specificity; for example, CA125 is widely us ed t o d etect ovarian can cer in post-menopausal women and i s t he b est cl inical


marker c urrently a vailable. Although CA125 l evels a re e levated in ~ 80% of pa tients w ith advanced s tage ovarian cancer, serum levels are increased in only 50–60% of patients with the ear ly stage [8, 72–74]. Early diagnosis of cancers, including ovarian cancer, remains difficult. There is therefore an urgent need to discover better potential biomarkers for cancer diagnosis in clinical practice.

Laser

Sample

stage

→

Acceleration

region

↑+25 kV

↑+0 kV

Reflectron

↑Detector

Electrostatic gate

for precursor ion

selection

Decision of the

precursor ion

takes place

TOF1TOF1 TOF2TOF2Fragmentation of

the mass-selected

precursor ion occurs

B. ESI-Q-TOF-MS

→

Reflectron

↑Detector

●

● ●

●

●● •

••

••

••

•

•• •

• •

•

•

••

●

● ●

●

●● •

••

••

••

•

•• •

• •

•

•

••

→ →Colision

cell→

ESI source Q mass analyzer TOF mass analyzer

(Reflector type)

MALDI source

C. ESI-IT-TOF-MS

→

Reflectron

↑

Detector

●

● ●

●

●● •

••

••

••

•

•• •

• •

•

•

••

→

ESI source

IT mass analyzer

TOF mass analyzer

(Reflector type)

→→ →••

•

•

•

• ••

••

••

Trapping

region

LC→

LC→

Figure 5. Tandem MS Tandem MS, in which multiple mass analyzers are connected in series, is used widely for analysis with good mass accuracy and high resolution. MALDI-TOF-TOF-MS (A), ESI-Q-TOF-MS (B), ESI-IT-TOF-MS (C) are well-known as tandem MS for proteomic analysis.


Isolation of interest protein using

separation technologies

Isolation of interest protein using

separation technologies

Interest proteins

2D-DIGE gel

●●

●●

●

●

●●

•

•

•

•••

•

•••

••

• •••• •

••

• •

••

→

Cutting

interest protein

Gel piece including

only interest protein

SDS-PAGE gel

Fraction including

only interest protein

•

••

•••

Separation column

LC (e.g., HPLC, nano LC)

→

→Digestion with

proteolytic enzymes

(e.g., trypsin)→

The m/z measurement of

generated peptides using MS

Isolated protein

Peptide derived

from isolated protein

Pe

ak I

nte

nsity

m/z→

Database search

→

Best-matched protein

(Identification)

1. Proteins registered in database are in silico digested with

the same proteolytic enzyme.

2.

3. Actual measurement values of peptides derived from interest

protein are compared with theoretical values of all proteins

registered in database.

Actual measurement values (m/z) of peptides

(A) PMF for protein identification. A protein isolated with SDS-PSGE, 2D-DIGE, and LC is digested with proteomic enzymes such as trypsin. The m/z of generated peptides is measured using MS. Actual measurement values (m/z) of the peptides are compared with theoretical values of those derived from all proteins registered in the database to find the best-matched protein.


Figure 6. (Continued)

↓

↓

↑↑

Selection of interest peptides

(amino acid sequence; A1-A

2-A

3------A

n-2-A

n-1-A

n)

Selection of interest peptides

(amino acid sequence; A1-A

2-A

3------A

n-2-A

n-1-A

n)

ProteinChip ClinProt

Patients

→

→

→

→

→

→

→

→

Healthy

indiv

iduals

○

●

△

▲◆

◎

▼

▽

◉■

□

◇

□

◉◎

◎◎

■

■

●

◎

○△

▼

◇

◎

■□

Peak I

nte

nsity

m/z

Biomarker peak

Differential analysisDifferential analysis

Patients

Healthy individuals

Peak I

nte

nsity


m/z

→

PMF

e.g., Biomarker peaks detected using ProteinChip array and ClinProt

Peaks used in database search for protein identification by PMF

(peaks circled in broken lines)

Peak I

nte

nsity

m/z

→Digestion

→Fragmentation of selected

peptide (parent ion) by CID

or PSD using tandem MS

An-2 An-1 AnAn-3A1 A2 A3 A4 ------ An-2 An-1 AnAn-3A1 A2 A3 A4 ------

A1A1An-2 An-1 AnAn-3A2 A3 A4 ------ An-2 An-1 AnAn-3A2 A3 A4 ------

＋

A1 A2A1 A2An-2 An-1 AnAn-3A3 A4 ------ An-2 An-1 AnAn-3A3 A4 ------

A1 A2 A3A1 A2 A3An-2 An-1 AnAn-3A4 ------ An-2 An-1 AnAn-3A4 ------

----

-

----

-

An-2 An-1 AnAn-2 An-1 AnAn-3A1 A2 A3 A4 ------ An-3A1 A2 A3 A4 ------

＋

＋

＋

＋

＋

＋

＋

＋

An-1 AnAn-1 AnAn-2An-3A1 A2 A3 A4 ------ An-2An-3A1 A2 A3 A4 ------

AnAnAn-2 An-1An-3A1 A2 A3 A4 ------ An-2 An-1An-3A1 A2 A3 A4 ------

＋

＋

＋

＋

Fragment ions

→

Detection of fragment ions

Peak I

nte

nsity

m/z

Fragment ions ←Parent ion

→

Actual measurement values (m/z) of parent ion and fragment ions

Database search

→

Best-matched peptides

(Identification)

2. Candidate peptides are in silico fragmented.

3. The m/z values of fragment ions are calculated (theoretical values).

4. Actual measurement values of fragment ions are compared

with theoretical values.

1. Peptides having the same m/z value as parent ion are searched from

database.

Parent ion

(B) MS/MS ion search for peptide identification. From spectra obtained by ProteinChip analysis, ClinProt analysis, and PMF, the peptide of interest is selected (peaks circled in broken lines). The selected peptide ion (parent ion) is fragmented by CID or PSD using tandem MS. After the m/z of the parent ion and its fragment ions is measured, these actual measurement values are compared with theoretical values from all peptides registered in the database to find the best-matched peptide.

Figure 6. Identification of proteins/peptides using MS


T able 4. C ancer biomar ker s identified by pr oteomic analysis of ser um/plasma.

5.1. Characteristics of Proteomic Cancer Biomarkers and Diagnostic Availability

As described in Chapters 2–4, diverse types of protein separation technology and MS are

proficiently combined in proteomic analysis. A lthough a variety of combined systems have


been us ed i n a nalysis, t he 2D -DIGE–MS sy stem a nd P roteinChip a rray®–SELDI-TOF-MS system are a t present t he main s tandards to s earch for new diagnostic biomarkers i n many cancers [14–17]. Many studies have reported proteomic biomarkers in various cancers [75–105].

↑

↑

ProteinChip(CM10)

Peak I

nte

nsity

m/z

Differential analysis

HCC

Non-HCC

→

4467 m/z

↓↓ 4470 m/z

↓7770 m/z

Peak I

nte

nsity

↓↓3890 m/z

4435 m/z↓3444 m/z

m/z

• Increased in HCC

4467, 4470, and 7770 m/z

• Decreased in HCC

3444, 3890, and 4435 m/z

→

Decision treeDecision tree

7770 m/zm/z of peak →Threshhold (peak intensity) → 6.0646

Non-HCCNon-HCC

HCCHCCNon-HCCNon-HCC

HCCHCC

HCCHCC

Non-HCCNon-HCC

Non-HCCNon-HCC

3890 m/z

4435 m/z

4470 m/z

7770 m/z

3444 m/z

4067 m/z

3.1838

1.0622

5.6677

4.5484

2.6786

1.4707

←Low High→

HCCHCC

Second analysis group

(HCC, N=29; non-HCC, N=33)

Sensitivity 83% (24/29)Specificity 76% (25/33)

HCC

(N=35)

2

3

4

1

Non-HCC

(N=44)

→→→

→

2

3

4

1→→→

→

First

analysis group

Figure 7. Classification of HCC and non-HCC using decision tree. The first analysis group (HCC, N=35; non-HCC, N=44) is subjected to ProteinChip array® (CM10) followed by obtaining spectra with SELDI. Six peaks (3444, 3890, 4435, 4467, 4470, and 7770 m/z) were expressed differentially in the HCC group compared to the non-HCC group. Using m/z and peak intensities of these peaks, the decision tree was built up for HCC diagnosis. To determine the accuracy and validity of the algorithm used, the decision tree was reevaluated using the second analysis group (HCC, N=29; non-HCC, N=33). The decision tree correctly diagnosed 83% (24 of 29, sensitivity) of patients with HCC and 76% (25/33, specificity) of patients without HCC.

Table 4 s hows can cer b iomarkers that h ave b een i dentified b y pr oteomic a nalysis of

serum/plasma. B iomarkers whose expression i s increased and/or decreased specifically in a particular cancer type have a g reat deal of potential in the diagnostic f ield; however, as the expression of α-antitrypsin has been reported to be increased in colorectal cancer [78], HCC [86], an d ovarian can cer [96], the expression of most biomarkers is increased and/or decreased i n multiple t ypes of c ancer. W hile pr oteins/peptides t hat a re e xpressed differentially in v arious c ancers co nstitute a l arge p ortion o f can cer b iomarkers, a f ew biomarkers are possibly specific to a particular cancer.

Neutrophil-activating peptide 2 (NAP-2) and Ras-related protein Rab-7B are expected to be specific biomarkers of HCC and lung cancer, respectively [ 87, 91]. NAP -2 is the major form of C XCL7, w hich i s a m ember of t he c hemokine f amily involved i n r egulating immunity a ngiogenesis, s tem-cell t rafficking, a nd m ediating or gan-specific m etastases o f cancers [106]. CXCL7 is translated as a propeptide and then cleaved to several smaller forms, including N AP-2. A lthough C XCL7 ha s be en r eported t o be a m arker of a dvanced myelodysplastic s yndromes (MDS), hematologic s tem c ell m alignancies in e lderly patients [107], no expression of NAP-2 is observed in tissues of other types of cancer. Furthermore, serum levels of NAP-2 are correlated with the clinicopathological features and prognosis of HCC [ 108]. Rab7B, w hich i s hi ghly e xpressed i n l ung c ancer, i s a m ember of the R as oncogene family and a small Rab GTPase that regulates vesicular t raffic f rom ear ly to late endosomal stages of the endocytic pathway. Rab7B is expressed in the heart, placenta, lung,


skeletal m uscle, a nd pe ripheral bl ood l eukocytes [ 109]; h owever, n o i ncrease o f i ts se rum levels h as b een reported i n c ancers ot her t han l ung c ancer. Consequently, t hese f indings suggest that N AP-2 an d R ab7B m ay b e s pecific b iomarkers o f H CC an d l ung can cer, respectively, although many clinical studies must be performed using sera from patients with a variety of diseases.

The expression of serum amyloid A (SAA) is increased in a v arious cancers (Table 4). Although proteomic analysis of serum/plasma identified SAA in HCC [89], lung cancer [91], nasopharyngeal carcinoma [94], ovarian cancer [99], and renal cancer [105], its increased expression ha s a lso been r eported i n ot her c ancers s uch a s ki dney, c olorectal, a nd pr ostate cancers, as well as in leukemias and lymphomas [110]. In addition, SAA is an apolipoprotein secreted in an acute phase of inflammation [111]; therefore, its increased expression may represent the epiphenomenon of cancers. While SAA is unlikely to be of much clinical use for diagnosing the onset of cancer, it may be important to monitor the stage of cancer progression and relapse. In nasopharyngeal cancer, SAA has been reported to be useful for monitoring its relapse [112].

Although most proteomic biomarkers a re not specific to a particular c ancer, a combination of t hose w ith know n di agnostic m ethods ha s be en r eported t o i mprove t he probability of cancer diagnosis [86, 101]. Using CA125 alone as the best clinical marker of ovarian can cer, p atients w ith ear ly s tage can cer ar e d iagnosed w ith a r eceiver o perating characteristic a rea under the curve (ROC AUC) of 0.613. In contrast, when four proteomic biomarkers (transthyretin, hemoglobin subunit β, apolipoprotein A1, a nd t ransferrin) a re combined with CA125, the diagnosis is performed with a ROC AUC of 0.955 [101]. The four biomarkers are highly abundant proteins and are not likely to be synthesized at the focal site. Thus, these biomarkers, even if they are not released directly by the tumor, could be used in combination w ith ot her markers, s uch a s CA125, t o i mprove t he di agnostic pr obability of early s tage ovarian cancer. In another instance, AFP is a biomarker for HCC diagnosis. By 2D-DIGE of sera from HCC patients followed by protein identification using MALDI-TOF-TOF-MS, heat-shock protein 27 (HSP27) was identified as an HCC biomarker [86] (Table 4). HSP27 i s a st ress-inducible c ytosolic pr otein t hat i s u biquitously pr esent i n m any nor mal tissues and has been reported to be over-expressed in many types o f cancer, such as breast cancer an d as trocytic b rain t umor [ 113, 11 4]. H SP27 is n ot a s pecific can cer b iomarker; however, i t c ould be de tected i n H CC patients w ith n egative A FP [ 86]. T his su ggests t hat HSP27 could be complementary to AFP in the diagnosis of HCC. Furthermore, combination diagnosis of HSP27 and AFP improves the diagnostic probability of HCC patients with small tumors (< 5 cm). By the same token, leucine-rich α2 glycoprotein (LRG) has been reported to serve as a complementary biomarker of CA19.9, which is a tumor marker commonly used for the diagnosis of pancreatic cancer [102].

5.2. Early Diagnosis of HCC Using ProteinChip Array® In cancer d iagnosis, ear ly detection b efore t he o nset o f cl inical s ymptoms i s very

important. Early diagnosis can lead to curative treatment, significantly improving prognosis. Diagnosing patients who present without symptoms, however, is difficult, because the tumor is s till s mall a nd th ere a re f ew le sions. Using th e ProteinChip a rray®–SELDI-TOF-MS


system, early diagnostic methods have been reported in several cancers, including our data on HCC [18, 115–117].

Approximately 170 million people worldwide are infected with hepatitis C virus (HCV), which when persistent can progress to HCC. Interferon (IFN) or combined IFN and ribavirin are cu rrently t he o nly ef fective t reatments for ch ronic h epatitis C , l eading to a d ecrease o f HCC occurrence [118, 119]; however, some patients do not receive IFN treatment or fail to recover f rom H CV e ven w ith I FN t reatment. In a ddition, a s ubset of i ndividuals r emains unaware that they are infected with HCV. Namely, in these patients, HCC may present only in the advanced stage. The prognosis of patients presenting with symptoms related to HCC is extremely poor. One requirement is the early detection of HCC before the onset of c linical symptoms.

We have been studying the developmental mechanism of HCC [120–122]. As part of the investigation, we sought to identify novel diagnostic markers of HCC using the ProteinChip array®–SELDI-TOF-MS system and to develop a diagnostic system using these markers. The decision t ree for HCC diagnosis was established us ing sera f rom pa tients with and without HCC (Figure 7) [18].

The f irst a nalysis gr oup ( HCC, N =35; non -HCC, N=44) w as s ubjected t o ProteinChip array® (CM10) f ollowed b y obt aining s pectra w ith S ELDI. S ix pe aks ( 3444, 3890, 4435 , 4467, 4470, and 7770 m/z) that were expressed differentially in the HCC group compared to the non-HCC group were used for diagnosing HCC. The decision t ree generated using m/z and pe ak i ntensities c orrectly c lassified 97% of H CC samples. Further, t o de termine t he accuracy a nd v alidity of t he a lgorithm us ed, t he decision t ree w as r eevaluated using t he second analysis group (HCC, N=29; non-HCC, N=33). The decision tree correctly diagnosed 83% ( 24 of 29, s ensitivity) of pa tients w ith H CC a nd 76% ( 25/33, s pecificity) of patients without HCC. Importantly, the accuracy of decision-tree diagnosis for HCC was higher than that of other known tumor markers such as AFP (sensitivity, 41%; specificity, 67%).

Pathogens SS

C3a

C3b

iC3b

C3

Cleavage Process of C3

SS

↑

↑↑

SSSS

C3fSSSS

C3c

C3dg

C3g C3d

Activation pathways

Malignant cells

Viral-infected cells

Bacteria

etc.

YY

Y●

●

●

Antibody

Non-self

material

■■

■■

■■

◇◇◇◇◇◇◇◇◇◇

◇◇◇◇◇◇◇◇◇◇

◇◇◇◇◇◇◇◇◇◇

Mannose-

containing

carbohydrate

Lectin pathway Lectin

pathway

Alternative pathway

Alternative pathway

Classical pathway Classical pathway →

→

→

→

→

→SSSS

Figure 8. Activating mechanism of complement system. When pathogens (e.g., malignant cells, viral-infected cells, and bacteria) exist in the living body, the system is activated through classical, alternative, and lectin pathways, with the result that they are cleared by opsonization and/or lysis. Classical, alternative, and lectin pathways are activated by an antibody bound to pathogens, certain non-self materials, and mannose-containing carbohydrates, respectively. Whichever pathway is activated, C3 is finally cleaved into C3a, C3b, iC3b, and C3f etc.


The most fundamental requirement for serum-based marker detection is the identification of carcinoma in an early stage when treatment has the greatest impact on prognosis. Decision-tree diagnosis was performed using serum samples taken from 7 patients 1 year before HCC development. S ix of t he 7 (86%) pa tients w ho l ater de veloped H CC w ere di agnosed w ith HCC using the decision tree. One year before development, HCC was s till undetectable by the current diagnostic methods. I t i s suggested that this decision t ree i s useful for the early diagnosis of HCC.

5.3. Proteomic Strategy for Discovering Specific Cancer Biomarkers As s hown i n T able 4 , m any can cer b iomarkers h ave b een i dentified b y s erum/plasma

proteomic an alysis. Wh ile s pecific b iomarkers s uch as NAP-2 an d Rab7B ar e d iscovered fairly infrequently [87, 91], most of these are unfortunately not specific to a particular cancer. In addition, the proteomic biomarker is mostly highly abundant proteins (Table 3). Certainly, non-specific a nd/or hi ghly expressed bi omarkers a re useful t o c omplement c onventional diagnosis methods [86, 101] and for monitoring cancer relapse [112]; however, because they are n ot n ecessarily r eleased d irectly f rom tumor t issues, t heir d iagnosis methods may h ave little s cientific b asis. D iscovering s pecific can cer b iomarkers that ar e b ased o n s cientific evidence are important for the development of novel diagnostic methods.

Proteomic a nalysis of l ow a bundant pr otein ( less t han 1% of s erum/plasma pr otein content) i n s erum/plasma may l ead t o di scovering s pecific bi omarkers. A lthough s hotgun proteomics us ing LC ( Figure 1D ) and t andem MS ( Figure 5B , C) ha s been e stablished for serum/plasma a nalysis [123, 12 4], c ancer b iomarkers u seful fo r e arly d iagnosis are st ill unknown. Proteomic approaches based on other perspectives might also be needed.

Many r esearchers have h ypothesized t hat t he b est can cer b iomarkers w ere l ikely t o b e secreted p roteins [ 125]. I n a ddition, m any c lassical c ancer bi omarkers, s uch a s C EA a nd CA15.3, are bound to the cell membrane, but their extracellular domains could be found, through shedding, i n the c irculatory s ystem. O n the ba sis of t his h ypothesis, R oeßler e t a l. aimed t o i dentify n ovel s erum b iomarkers o f co lorectal can cer [ 126]. F irst, 16 m atched colorectal cancer and adjacent normal tissue samples were subjected to 2D-DIGE followed by protein i dentification us ing MS. F ive pr oteins w hose e xpression i s i ncreased m arkedly i n colorectal cancer were identified as transforming growth factor-β-induced protein ig-h3 (βIG-H3), ni cotinamide N -methyltransferase ( NNMT), nuc leoside di phosphate ki nase A ( nm23-H1), p urine nuc leoside pho sphorylase ( PNPH), a nd m annose-6-phosphate receptor bi nding protein 1 (M6P1). Validation using ELISA showed that elevated levels of NNMT were found in sera from patients with colorectal cancer. Furthermore, the accuracy of NNMT diagnosis (ROC AUC, 0.84) was higher than that of the established tumor marker CEA (ROC AUC, 0.78). S imilarly, proteomic an alysis o f t umor t issues o ccasionally l eads t o d iscovering serum/plasma biomarkers. Using the s ame s trategy, phosphoglycerate kinase 1 ( PGK1) ha s reportedly been identified as a serum biomarker of pancreatic ductal adenocarcinoma [127].

High abundant proteins are possibly suitable to search for specific cancer biomarkers. As shown in Table 4, high abundant proteins expressed differentially in various cancers appear capable of be ing c ategorized i n t hree gr oups: 1) c omponents of c omplement s ystem a nd regulated pr oteins (C3, C 4, and f actor B etc.), 2) apolipoproteins ( apolipoproteins A1, A 2, and C 1 etc.), 3) iron-metabolic proteins (transferrin, haptoglobin, and hemopexin etc.).


Furthermore, as seen in C3a [75, 78, 85] and fibrinopetide A [81, 82, 90], which are fragment peptides derived from complement C3 and fibrinogen, respectively, protein cleavage products also s erve a s c ancer bi omarkers. C hanges i n t he e xpressions a nd/or pr ocessing pa tterns of highly abundant proteins in cancer may be caused by specific molecules in pathogens. We are focusing on the relationships between the complement system and ATL, a fatal leukemia caused by infection with human T lymphotropic virus type-1 (HTLV-1) [19].

The c omplement s ystem is a major mediator o f innate immune defense [128–130]. T he system consists of a group of over 30 proteins, such as C3, C4, and factor B. When pathogens (e.g., m alignant cells, vi ral-infected c ells, a nd ba cteria) exist in the living bod y, t he s ystem is activated, w ith t he r esult t hat t hey a re c leared b y opsonization a nd/or l ysis. T he c omplement system h as three d istinct p athways (classical, a lternative, a nd lectin pa thways) ( Figure 8 ). Classical, alternative, and lectin pathways are activated by an antibody bound to pathogens, certain non-self m aterials, and m annose-containing carbohydrates e xpressed o n the p athogen s urface, respectively. Whichever pathway is activated, C3 is finally cleaved into C3a, C3b, iC3b, and C3f etc. The increase of C3 cleavage products in cancer patients is a reflection of this activation.

We i dentified C 3f as t he s erum b iomarker o f A TL by d ifferential a nalysis using t he ProteinChip array®–SELDI-TOF-MS system followed by MS/MS ion search using MALDI-TOF-TOF-MS [19]. Further, analysis of markers specific to the activation pathways showed that the serum concentration of the marker of the lectin pathway was significantly higher in ATL pa tients. T hese r esults s uggest t hat t he c omplement s ystem is a ctivated t hrough t he lectin pathway in ATL. Furthermore, this implies the presence of a trigger that activates the lectin pa thway. A s one possibility, i t i s c onsidered t hat a berrant gl ycoproteins a nd glycolipids, including mannose-containing c arbohydrates, a re e xpressed on t he s urface of ATL cells. C3f is not specific to ATL, because its expression is increased in various cancers [19]. The trigger is, however, possibly suitable for diagnosis as an ATL-specific biomarker.

In a ddition t o i dentifying proteins/peptides b y pr oteomic a nalysis of s erum/plasma, clarifying a molecule associated with the signal transduction pathway such as the complement system may be necessary to discovery specific cancer biomarkers.

5.4. From Biomarker Discovery to Clinical Application The most important issue in developing diagnosis methods is reproducibility in c linical

practice as well as in the laboratory. In particular, in the ear ly diagnosis of cancer, because tumors a re microscopic, i nsensible c hanges i n t he l iving body ha ve t o be c aptured. I n addition, pr evious s tudies, including our da ta, i ndicated th at m easurement s ystems u sing multiple biomarkers a re e ffective in early di agnosis [18, 115–117]; t herefore, measurement systems are required that reproducibly quantify multiple biomarkers simultaneously using just one experiment.

The P roteinChip array®–SELDI-TOF-MS s ystem e nables th e d etection o f multiple proteins/peptides i n j ust on e e xperiment. D iagnosis c an be pe rformed by t he decision t ree using m /z a nd pe ak i ntensities w ithout t he i dentification of pr oteins/peptides. W hile m any studies have reported diagnostic methods using decision trees, irreproducibility is a problem [131]. Although our experience a lso shows this, t he i rreproducibility could be a ttributed to analytical s teps f rom th e time th e s erum is c ollected u ntil th e s pectrum is a cquired. Unfortunately, no uni versal standard protocol exists for the collection, storage, and transport


of serum samples. Also, there is no universal quality control of the decision tree and spectra obtained with MS. In particular, the spectral patterns acquired by SELDI-TOF-MS seem to be affected by the aging degradation of the ionization laser. Whether this is true, the cause of the irreproducibility is uncertain. If peaks useful for diagnosis are found, the better choice is to identify proteins/peptides derived from these peaks followed by the development of other measurement systems with good reproducibility and quantitation.

Y Y Y○

□

◎

●

◆

◎ ◆

△

Sample addition

(Patient serum)

○

Y

Y○

YY◇

Y○

YY◇ ☆

★

Antigen

Capture

antibody

Detecting

antibody

Secondary antibody

(enzyme linked)Substrate

addition

Color developing◆

Figure 9. Sandwich ELISA. After the plate is coated with a capture antibody, the sample is added and any antigen present binds to capture the antibody. Next, the detecting antibody is added and binds to the antigen, and then the enzyme-linked secondary antibody is added and binds to the detecting antibody. Finally, the substrate is added and is converted by the enzyme to the detectable form.

ELISA is a biochemical technique used mainly in immunology to detect the presence of an antibody or an antigen i n a s ample [132]. E LISA has been widely used a s a di agnostic system in clinical practice; for example, sandwich ELISA, which has goo d s ensitivity, quantitation and reproducibility, is performed as follows (Figure 9): 1) the plate is coated with a capture antibody; 2) sample is added and any antigen present binds to capture the antibody; 3) t he de tecting a ntibody i s a dded a nd bi nds t o t he a ntigen; 4) e nzyme-linked s econdary antibody is added and binds to the detecting antibody; 5) substrate is added and is converted by the enzyme to the detectable form. Early diagnosis of ovarian cancer has reportedly been performed us ing E LISA f or pr oteomic biomarkers ( transthyretin, he moglobin subunit β, apolipoprotein A 1, a nd t ransferrin) [101]. A lso, a s a hi gh-throughput and multiplex measurement s ystem, an tibody ar ray h as r ecently b ecome o f i nterest [ 133]. T o apply proteomic biomarkers in a clinical setting, the development of measurement systems, such as ELISA and antibody array, may be a key step.

6. CANCER MALDI IMAGING TECHNIQUE AND DIAGNOSIS


CONCLUSION Proteomic analysis based on protein/peptide separation and identification using MS is an

essential technology for biomarker discovery. Although most proteomic biomarkers may be not specific for a particular cancer, they are certainly of help in the early diagnosis of several cancers. There i s c onsiderable di fficulty i n di scovering s pecific bi omarkers; how ever, a variety of strategies for discovering specific biomarkers may make it possible. We hope that proteomic cancer biomarkers will be widely used in the future for early diagnosis in clinical practice.

REFERENCES

Yeo, TP; H ruban, RH; Leach, SD; Wilentz, RE; Sohn, TA; Kern, S E; Iacobuzio-Donahue, CA; Maitra, A; Goggins, M; Canto, MI; Abrams, RA; Laheru, D; Jaffee, EM; Hidalgo, M; Yeo, CJ. Pancreatic cancer. Curr. Probl. Cancer, 2002, 26, 176-275.

Chen, G; Gharib, TG; Wang, H; Huang, CC; Kuick, R; Thomas, DG; Shedden, KA; Misek, DE; Taylor, JM; Giordano, TJ; Kardia, SL; Iannettoni, MD; Yee, J; Hogg, PJ; Orringer, MB; H anash, S M; Beer, D G. Protein pr ofiles a ssociated w ith s urvival i n lung adenocarcinoma. Proc. Natl. Acad. Sci. USA, 2003, 100, 13537-13542.

Etzioni, R; Urban, N; Ramsey, S; McIntosh, M; Schwartz, S; Reid, B; Radich, J; Anderson, G; H artwell, L. T he cas e f or ear ly d etection. Nat. Rev. Cancer, 2003, 3, 243-252.

Yokota, T ; I shiyama, S ; S aito, T ; T eshima, S ; N arushima, Y ; M urata, K; Iwamoto, K ; Yashima, R ; Yamauchi, H; K ikuchi, S . L ymph node m etastasis a s a s ignificant prognostic f actor in gastric c ancer: a m ultiple lo gistic re gression a nalysis. Scand. J. Gastroenterol., 2004, 39, 380-384.

Ludwig, JA; Weinstein JN. Biomarkers in cancer staging, prognosis, and treatment selection. Nat. Rev. Cancer., 2005, 5, 845-856.

Lam, T ; Nabi, G. Potential of ur inary biomarkers in early b ladder cancer d iagnosis. Expert Rev. Anticancer. Ther., 2007, 7, 1105-1115.

Sidransky, D . E merging molecular markers of c ancer. Nat. Rev. Cancer, 2 002, 2 , 210-209.

Menon, U; Jacobs, I. Recent developments in ovarian cancer screening. Curr. Opin. Obstet. Gynecol., 2000, 12, 39-42.

Chatterjee, SK; Zetter, BR. Cancer biomarkers: knowing the present and predicting the future. Future Oncol., 2005, 1, 37-50.


Polanski, M; Anderson, NL. A l ist of candidate cancer biomarkers for targeted proteomics. Biomark. Insights, 2006, 2, 1-48.

Willard, HF; Angrist, M; Ginsburg, GS. Genomic medicine: genetic variation and its impact on the future of health care. Phil. Trans. R. Soc., 2005, 360, 1543-1550.

Stathopoulou, FT. Genome and proteome in oncology. New York: Nova Science Publishers Inc.; 2005.

Kim, Y S; M aruvada, P . Frontiers i n metabolomics f or cancer r esearch: proceedings o f a National Cancer Institute workshop. Metabolomics., 2008, 4, 105-113.

Wulfkuhle, JD; Liotta, LA; Petricoin, EF. Proteomic applications for the early de tection of cancer. Nat. Rev. Cancer, 2003, 3, 267-275.

Diamandis, E P. M ass s pectrometry a s a di agnostic a nd a c ancer bi omarker di scovery t ool: opportunities and potential limitations. Mol. Cell. Proteomics, 2004, 3, 367-378.

Righetti, PG; Castagna, A; Antonioli, P; Cecconi, D; Campostrini, N; Righetti, SC. Proteomic approaches for s tudying chemoresistance in cancer. Expert Rev. Proteomics, 2005, 2, 215-228.

Chen, CH. Review of a current role of mass spectrometry for proteome research. Anal. Chim. Acta, 2008, 624, 16-36.

Kanmura, S ; U to, H ; Kusumoto, K; Ishida, Y ; Hasuike, S ; Nagata, K; Hayashi, K; Ido, A ; Stuver, SO; Tsubouchi, H. Early diagnostic potential for hepatocellular carcinoma using the SELDI ProteinChip system. Hepatology, 2007, 45, 948-956.

Ishida, Y ; Y amashita, K ; S asaki, H ; T akajou, I ; K ubuki, Y ; M orishita, K ; T subouchi, H ; Okayama, A. Activation of complement system in adult T-cell leukemia (ATL) occurs mainly through lectin pathway; a serum proteomic approach using mass spectrometry. Cancer lett., 2008, 271, 167-177.

Shastry, BS. SNP alleles in human disease and evolution. J. Hum. Genet., 2002, 47, 561-566. Wang, XQ; Luk, JM; Garcia-Barcelo, M; Miao, X; Leung, PP; Ho, DW; Cheung, ST; Lam,

BY; Cheung, CK; Wong, AS; Lau, SS; So, MT; Yu, WC; Cai, Q; Liu, KS; Hui, CK; Lau, GK; Poon, RT; Wong, J; Fan, ST. Liver intestine-cadherin (CDH17) haplotype is associated with increased risk of hepatocellular carcinoma. Clin. Cancer Res., 2006, 12, 5248-5252.

Esteller, M; Corn, PG; Baylin, SB; Herman, JG. A gene hypermethylation profile of human cancer. Cancer Res., 2001, 61, 3225-3229.

Tischoff, I ; T annapfe, A. D NA m ethylation i n he patocellular c arcinoma. World J. Gastroenterol., 2008, 14, 1741-1748.

Bowtell, D; Sambrook, J . DNA microarrays: a molecular c loning manual. New York: Cold Spring Harbor Laboratory Press; 2002.

Lee, NH; Saeed, AI. Microarrays: an overview. Methods Mol. Biol., 2007, 353, 265-300. Sasaki, H ; Nishikata, I ; S hiraga, T ; A kamatsu, E ; Fukami, T ; H idaka, T ; K ubuki, Y ;

Okayama, A ; H amada, K; O kabe, H ; M urakami, Y ; T subouchi, H ; M orishita, K . Overexpression of a cell adhesion molecule, TSLC1, as a possible molecular marker for acute-type adult T-cell leukemia. Blood, 2005, 105, 1204-1213.

Simon, R. D iagnostic a nd pr ognostic pr ediction us ing ge ne e xpression pr ofiles i n hi gh-dimensional microarray data. Br. J. Cancer, 2003, 89, 1599-1604.

Brennan, DJ; Kelly, C; Rexhepaj, E; Dervan, PA; Duffy, MJ; Gallagher, WM. Contribution of D NA a nd t issue m icroarray t echnology t o t he i dentification a nd v alidation of


biomarkers and personalised medicine in breast cancer. Cancer Genomics Proteomics, 2007, 4, 121-134.

Sotiriou, C; Piccart, MJ. Taking gene-expression profiling to the clinic: when will molecular signatures become relevant to patient care? Nat. Rev. Cancer, 2007, 7, 545-553.

Harrigan, GG; G oodacre, R . M etabolic pr ofiling: i ts r ole i n bi omarker di scovery a nd ge ne function analysis. Boston: Kluwer Academic Publishers; 2003.

Tomita, M; Nishioka, T . Metabolomics: the f rontier systems biology. New York; Springer; 2005.

Griffin, JL; Nicholls, AW. Metabolomics as a functional genomic tool for understanding lipid dysfunction i n di abetes, ob esity a nd r elated di sorders. Parmacogenomics, 2006, 7, 1095-1107.

Fiehn, O; Kopka, J; Dörmann, P; Altmann, T; Trethewey, RN; Willmitzer, L. Metabolite profiling for plant functional genomics. Nat. Biotechnol., 2000, 18, 1157-1161.

Plumb, R; Granger, J; Stumpf, C; Wilson, ID; Evans, JA; Lenz, EM. Metabonomic analysis of m ouse ur ine by l iquid-chromatography-time o f fl ight m ass sp ectrometry ( LC-TOFMS): detection of strain, diurnal and gender differences. Analyst, 2003, 128, 819-823.

Soga, T; Baran, R; Suematsu, M; Ueno, Y; Ikeda, S; Sakurakawa, T; Kakazu, Y; Ishikawa, T; Robert, M; Nishioka, T; Tomita, M. Differential metabolomics reveals ophthalmic acid as a n o xidative s tress bi omarker i ndicating he patic gl utathione c onsumption. J. Biol. Chem., 2006, 281, 16768-16776.

Spraul, M; Freund, AS; Nast, RE; Withers, RS; Maas, WE; Corcoran, O. Advancing NMR sensitivity f or LC-NMR-MS us ing a c ryoflow pr obe: application t o t he a nalysis of acetaminophen metabolites in urine. Anal. Chem., 2003, 75, 1536-1541.

Yang, Z . O nline hy phenated l iquid c hromatography-nuclear magnetic r esonance spectroscopy-mass sp ectrometry for dr ug metabolite a nd na ture pr oduct a nalysis. J. Pharm. Biomed. Anal., 2006, 40, 516-527.

Schnackenberg, L K; B eger, R D. M onitoring the he alth t o disease continuum with global metabolic profiling and systems biology. Pharmacogenomics, 2006, 7, 1077-1086.

Kaddurah-Daouk, R ; Kristal, BS; Weinshilboum, RM. Metabolomics: a global biochemical approach to drug response and disease. Annu. Rev. Pharmacol. Toxicol., 2008, 48, 653-683.

Kind, T; Tolstikov, V; Fiehn, O; Weiss, RH. A comprehensive urinary metabolomic approach for identifying kidney cancer. Anal. Biochem., 2007, 363, 185-195.

Mitani, Y ; Lezhava, A ; Kawai, Y ; Kikuchi, T ; O guchi-Katayama, A ; Kogo, Y ; Itoh, M ; Miyagi, T; Takakura, H; Hoshi, K; Kato, C; Arakawa, T; Shibata, K; Fukui, K; Masui, R; Kuramitsu, S; Kiyotani, K; Chalk, A; Tsunekawa, K; Murakami, M; Kamataki, T; Oka, T; S himada, H ; C izdziel, P E; H ayashizaki, Y . R apid S NP di agnostics us ing asymmetric isothermal amplification and a new mismatch-suppression technology. Nat. Methods, 2007, 4, 257-262.

Kuzuhara, T ; K ise, D ; S hirakawa, Y; S asada, K ; S uganuma, M ; F ujiki, H . Generation o f mouse m onoclonal a ntibody a gainst ( -)-epigallocatechin g allate. Biol. Pharm. Bull., 2008, 31, 816-819.

Grossklaus, DJ; Smith, JA; Shappell, SB; Coffey, CS; Chang, SS; C ookson, M S. T he free/total p rostate-specific antigen r atio ( %fPSA) is the b est p redictor o f tu mor


involvement in the radical prostatectomy specimen among men with an elevated PSA. Urol. Oncol., 2002, 7, 195-198.

Whitehouse, C; Solomon, E. Current s tatus of the molecular characterization of the ovarian cancer a ntigen CA125 a nd i mplications f or i ts us e i n c linical s creening. Gynecol. Oncol., 2003, 88, S152-S157.

Irmak, S ; T ilki, D ; Heukeshoven, J ; Oliveira-Ferrer, L; Friedrich, M; Huland, H; Ergün, S . Stage-dependent i ncrease o f or osomucoid a nd z inc-alpha2

Tirumalai, R S; C han, K C; Prieto, DA; I ssaq, H J; C onrads, TP; Veenstra, T D. Characterization of the low molecular weight human serum proteome. Mol. Cell. Proteomics, 2003, 2, 1096-1103.

-glycoprotein i n ur inary bladder cancer. Proteomics, 2005, 5, 4296-4304.

Hortin, G L. T he M ALDI-TOF mass s pectrometric view o f the p lasma p roteome an d peptidome. Clin. Chem., 2006, 52, 1223-1237.

Link, AJ. 2-D proteome analysis protocols. Methods Mol. Biol., 112, Humana Press: Totowa; 1999.

Kinter, M; S herman, NE. P rotein sequencing a nd identification using tandem mass spectrometry. New York: Wiley Interscience; 2000.

Tonge, R; Shaw, J; Middleton, B; Rowlinson, R; Rayner, S; Young, J; Pognan, F; Hawkins, E; Currie, I; Davison, M. Validation and development of fluorescence two-dimensional differential gel electrophoresis proteomics technology. Proteomics, 2001, 1, 377-396.

Marouga, R; David, S; Hawkins, E. The development of the DIGE system: 2D fluorescence difference gel analysis technology. Anal. Bioanal. Chem., 2005, 382, 669-678.

Tang, N; Tornatore, P; Weinberger, SR. Current developments in SELDI affinity technology. Mass Spectrom. Rev., 2004, 23, 34-44.

Seibert, V; Ebert, MP; Buschmann, T. Advances in clinical cancer proteomics: SELDI-TOF-mass spectrometry and biomarker discovery. Brief. Funct. Genomic. Proteomic. 2005, 4, 16-26.

Ketterlinus, R; Hsieh, SY; Teng, SH; Lee, H; Pusch, W. Fishing for biomarkers: analyzing mass s pectrometry d ata w ith t he n ew C linProTools so ftware. Biotechniques, 20 05, Suppl, 37-40.

Solassol, J; Jacot, W; L hermitte, L; B oulle, N ; M audelonde, T ; Mangé, A. C linical proteomics a nd mass s pectrometry pr ofiling f or c ancer de tection. Expert Rev. Proteomics., 2006, 3, 311-320.

McDonald, WH; Yates, JR 3rd. Shotgun proteomics and biomarker discovery. Dis. Markers, 2002, 18, 99-105.

Wu, CC; MacCoss, MJ. Shotgun proteomics: tools for the analysis of complex biological systems. Curr. Opin. Mol. Ther., 2002, 4, 242-250.

Motoyama, A; Yates, JR 3rd. Multidimensional LC separations in shotgun proteomics. Anal. Chem., 2008, 80, 7183-7193.

Tanaka, K ; W aki, H ; Ido, Y ; A kita, S ; Y oshida, Y; Y oshida, T; M atsuo, T . Protein a nd polymer a nalyses up t o m/z 100 000 by l aser i onization t ime-of-flight ma ss spectrometry. Rapid Commun. Mass Spectrom., 1988, 2, 151-153.

Karas, M ; H illenkamp, F. Laser desorption i onization of pr oteins with molecular masses exceeding 10,000 daltons. Anal. Chem., 1988, 60, 2299-3201.

Whitehouse, C M; D reyer, RN; Y amashita, M; Fenn, J B. Electrospray i nterface f or l iquid chromatographs and mass spectrometers. Anal. Chem., 1985, 57, 675-679.


Fenn, J B; M ann, M ; M eng, C K; W ong, S F; W hitehouse, C M. E lectrospray ionization f or mass spectrometry of large biomolecules. Science, 1989, 246, 64-71.

Pappin, DJ; Hojrup, P; Bleasby, A J. Rapid i dentification of pr oteins by pe ptide-mass fingerprinting. Curr. Biol., 1993, 3, 327-332.

Henzel, W J; B illeci, T M; S tults, J T; W ong, S C; G rimley, C ; W atanabe, C . I dentifying proteins from two-dimensional gels by molecular mass searching of peptide fragments in protein sequence databases. Proc. Natl. Acad. Sci. USA, 1993, 90, 5011-5015.

Mann, M; Højrup, P; Roepstorff, P. Use of mass spectrometric molecular weight information to identify proteins in sequence databases. Biol. Mass Spectrom., 1993, 22, 338-345.

James, P ; Q uadroni, M; C arafoli, E; G onnet, G . P rotein identification by mass profile fingerprinting. Biochem. Biophys. Res. Commun., 1993, 195, 58-64.

Yates, J R 3r d; Speicher, S ; Griffin, PR; Hunkapiller, T. Peptide mass maps: a highly informative approach to protein identification. Anal. Biochem., 1993, 214, 397-408.

Siuzdak, G. The expanding role of mass spectrometry in biotechnology. MCC Press: San Diego, CA; 2003.

Shevchenko, A ; J ensen, O N; P odtelejnikov, A V; Sagliocco, F ; W ilm, M ; V orm, O ; Mortensen, P; Shevchenko, A; Boucherie, H; Mann, M. Linking genome and proteome by mass spectrometry: large-scale identification of yeast proteins from two dimensional gels. Proc. Natl. Acad. Sci. USA, 1996, 93, 14440-14445.

Johnson, RS; Davis, MT; Taylor, JA; Patterson, SD. Informatics for protein identification by mass spectrometry. Methods, 2005, 35, 223-236.

Shadforth, I; Crowther, D; Bessant, C. Protein and peptide identification algorithms using MS for use in high throughput, automated pipelines. Proteomics, 2005, 5, 4082-4095.

Bast, RC Jr; K lug, TL; St. John, E ; J enison, E; Niloff, JM; Lazarus, H ; Berkowitz, RS; Leavitt, T; Griffiths, CT; Parker, L; Zurawski, VR Jr; Knapp, RC. A radioimmunoassay using a m onoclonal a ntibody t o monitor t he c ourse of e pithelial o varian c ancer. N. Engl. J. Med., 1983, 309, 883-887.

Bast, RC Jr; Urban, N; Shridhar, V; Smith, D; Zhang, Z; Skates, S; Lu, K; Liu, J; Fishman, D; Mills, G . E arly d etection o f o varian can cer: p romise an d r eality. Cancer Treat. Res., 2002, 107, 61-97.

Cohen, LS; Escobar, PF; Scharm, C; Glimco, B; Fishman, DA. Three-dimensional power Doppler ultrasound improves the di agnostic accuracy for ovarian cancer prediction. Gynecol. Oncol., 2001, 82, 40-48.

Li, J; Orlandi, R; White, CN; Rosenzweig, J; Zhao, J; Seregni, E; Morelli, D; Yu, Y; Meng, XY; Zhang, Z; Davidson, NE; Fung, E T; C han, DW. I ndependent v alidation of candidate breast cancer s erum b iomarkers i dentified b y mass spectrometry. Clin. Chem., 2005, 51, 2229-2235.

Gonçalves, A; Esterni, B; Bertucci, F; Sauvan, R; Chabannon, C; M, Cubizolles, M; Bardou, VJ; Houvenaegel, G; Jacquemier, J; Granjeaud, S; Meng, XY; Fung, ET; Birnbaum, D; Maraninchi, D; Viens, P; Borg, JP. Postoperative serum proteomic profiles may predict metastatic r elapse in h igh-risk p rimary b reast can cer p atients r eceiving ad juvant chemotherapy. Oncogene, 2006, 25, 981-989.

Liu, L; Wang, J; Liu, B; Dai, S; Wang, X; Chen, J; Huang, L; Xiao, X; He, D. Serum levels of variants of transthyretin down-regulation in cholangiocarcinoma. J. Cell. Biochem., 2008, 104, 745-755.


Ward, DG; Suggett, N; Cheng, Y; Wei, W; Johnson, H; Billingham, LJ; Ismail, T; Wakelam, MJO; Johnson, PJ; Martin, A. Identification of serum biomarkers for colon cancer by proteomic analysis. Br. J. Cancer, 2006, 94, 1898-1905.

Albrethsen, J ; Bøgebo, R ; Gammeltoft, S ; O lsen, J; Winther, B ; Raskov, H . U pregulated expression of human neutrophil peptides 1, 2, a nd 3 ( HNP 1-3) in colon cancer serum and tumours: a biomarker study. BMC Cancer, 2005, 5, 8.

Engwegen, JYMN; H elgason, H H; Cats, A ; Harris, N; B onfrer, JMG, S chellens, J HM, Beijnen, JH. Identification of serum proteins discriminating colorectal cancer patients and healthy controls us ing surface-enhanced l aser d esorption ionisation-time o f f light mass spectrometry. World J. Gastroenterol., 2006, 12, 1536-1544.

Su, Y; Shen, J; Qian, H; Ma, H; Ji, J; Ma, H; Ma, L, Zhang, W; Meng, L; Li, Z; Wu, J; Jin, G; Zhang, J; Shou, C. Diagnosis of gastric cancer using decision tree classification of mass spectral data. Cancer Sci., 2006, 98, 37-43.

Ebert, M P; N iemeyer, D ; D eininger, S O; W ex, T ; K nippig, C ; H offmann, J ; S auer, J ; Albrecht, W; Malfertheiner, P, Röcken, C. Identification and confirmation of increased fibrinopeptide a s erum p rotein l evels i n g astric can cer sera b y magnet b ead as sisted MALDI-TOF mass spectrometry. J. Proteome Res., 2006, 5, 2152-2158.

Liu, W; Liu, B; Xin, L; Zhang, Y; Chen, X; Zhu, Z; Lin, Y. Down-regulated expression of complement f actor I : a p otential s uppressive p rotein f or g astric c ancer id entified by serum proteome analysis. Clin. Chim. Acta., 2007, 377, 119-126.

Ward, DG; Cheng, Y; N’Kontchou, G; Thar, TT; Barget, N; Wei, W; Martin, A; Beaugrand, M; Johnson, PJ. P reclinical and post-treatment changes in the HCC-associated s erum proteome. Br. J. Cancer, 2006, 95, 1379-1383.

Lee, IN; Chen, CH; Sheu, JC; Lee, HS; Huang, GT; Chen, DS; Yu, CY; Wen, CL; Lu, FJ; Chow, L P. I dentification of c omplement C 3a a s a c andidate bi omarker i n hum an chronic he patitis C a nd H CV-related h epatocellular car cinoma u sing a p roteomics approach. Proteomics, 2006, 6, 2865-2873.

Feng, JT; Liu, YK; Song, HY; Dai, Z; Qin, LX; Almofti, MR; Fang, CY; Lu, HJ; Yang, PY; Tang, ZY. Heat-shock protein 27: a potential biomarker for hepatocellular car cinoma identified by serum proteome analysis. Proteomics, 2005, 5, 4581-4588.

He, M; Qin, J; Zhai, R; Wei, X; Wang, Q; Rong, M; Jiang, Z; Huang, Y; Zhang, Z. Detection and i dentificaiton of NAP-2 a s a bi omarker i n hpa titis B -related h epatocellular carcinoma by proteomic approach. Proteome Sci., 2008, 6, 10.

Zinkin, NT; Grall, F; Bhaskar, K; Otu, HH; Spentzos, D; Kalmowitz, B; Wells, M; Guerrero, M; A sara, JM; L ibermann, T A; Afdhal, N H. Serum pr oteomics a nd bi omarkers i n hepatocellular carcinoma and chronic l iver d isease. Clin. Cancer Res., 2008, 14, 470-477.

He, QY; Zhu, R ; Lei, T ; Ng, MYM; Luk, JM; Sham, P; Lau, GKK; Chiu, JF. Toward the proteomic identification of biomarkers for the prediction of HBV related hepatocellular carcinoma. J. Cell. Biochem., 2008, 103, 740-752.

Orvisky, E; Drake, SK; Martin, BM; Abdel-Hamid, M; Ressom, HW; Varghese, RS; An, Y; Saha, D; Hortin, GL; Loffredo, CA; Goldman, R. Enrichment of low molecular weight fraction of serum for MS analysis of peptides associated with hepatocellular carcinoma. Proteomics, 2006, 6, 2895-2902.

Dowling, P; O’Driscoll, L; Meleady, P; Henry, M; Roy, S; Ballot, J; Moriarty, M; Crown, J; Clynes, M . 2 -D di fference gel e lectrophoresis of t he l ung s quamous c ell c arcinoma


versus n ormal sera d emonstrates co nsistent al terations i n t he l evels o f t en s pecific proteins. Electrophoresis, 2007, 28, 4302-4310.

Okano, T; Kondo T; Kakisaka, T; Fujii, K; Yamada M; Kato, H; Nishimura, T; Gemma, A; Kudoh, S ; H irohashi, S . P lasma pr oteomics of l ung cancer b y a l inkage o f multi-dimensional liquid chromatography and two-dimensional difference gel electrophoresis. Proteomics, 2006, 6, 3938-3948.

Doustjalali, SR; Yusof, R; Govindasamy, GK; Bustam, AZ; Pillay, B; Hashim, OH. Patients with nasopharyngeal carcinoma demonstrate enhanced serum and tissue ceruloplasmin expression. J. Med. Invest., 2006, 53, 20-28.

Liao, Q; Zhao, L; Chen, X; Deng, Y; Ding, Y. Serum proteome analysis for profiling protein markers associated with carcinogenesis and lymph node metastasis in nasopharyngeal carcinoma. Clin. Exp. Metastasis, 2008, 25, 465-476.

Cheng, AJ; Chen LC; Chien, KY; Chen YJ; Chang, JT; Wang, HM; Liao, CT; Chen, IH. Oral cancer plasma tumor marker identified with bead-based affinity-fractionated proteomic technology. Clin. Chem., 2005, 51, 2236-2244.

Jackson, D; Craven, RA; Hutson, RC; Graze, I; Lueth, P; Tonge, RP; Hartley, JL; Nickson, JA; R ayner, S J; J ohnston, C ; D ieplinger, B ; H ubalek, M ; W ilkinson, N ; P erren, TJ; Kehoe, S; Hall, GD; Daxenbichler, G; Dieplinger, H; Selby, PJ; Banks, RE. Proteomic profiling i dentifies a famin a s a pot ential bi omarker f or ovarian c ancer. Clin. Cancer Res., 2007, 13, 7370-7379.

Ahmed, N; Barker, G; Oliva, KT; Hoffmann, P; Riley, C; Reeve, S; Smith, AI; Kemp, BE; Quinn, MA; Rice, GE. Proteomic-based identification of haptoglobin-1 precursor as a novel circulating biomarker of ovarian cancer. Br. J. Cancer, 2004, 91, 129-140.

Ahmed, N; Oliva, KT; Barker, G; Hoffmann, P; Reeve, S; Smith, IA; Quinn, MA; Rice, GE. Proteomic tr acking of s erum pr otein i soforms a s s creening bi omarkers of ovarian cancer. Proteomics, 2005, 5, 4625-4636.

Moshkovskii, SA; Serebryakova, MV; Kuteykin-Teplyakov, KB; Tikhonova, OV; Goufman, EI; Zgoda, VG; Taranets, IN; Makarov, OV; Archakov, AI. Ovarian cancer marker of 11.7 kDa detected by proteomics is a serum amyloid A1. Proteomics, 2005, 3790-3797.

Woong-Shick, A; Sung-Pil, P; Su-Mi, B; Joon-Mo, L; Sung-Eun, N; Gye-Hyun, N; Young-Lae, C ; Ho-Sun, C; Heung-Jae, J ; Chong-Kook, K; Young-Wan, K; Byoung-Don, H; Hyun-Sun, J . I dentification of he moglobin-α and -β subunits as potential serum biomarkers for t he di agnosis and prognosis of ovarian cancer. Cancer Sci., 2005, 96, 197-201.

Kozak, KR; Su, F; Whitelegge, JP; Faull, K; Reddy, S; Farias-Eisner, R. Characterization of serum b iomarkers f or d etection o f early stage o varian cancer. Proteomics, 2005, 5, 4589-4596.

Kakisaka, T; Kondo T; Okano, T; Fujii, K; Honda, K; Endo, M; Tsuchida, A; Aoki, T; Itoi, T; Moriyasu, F; Y amada, T ; K ato, H ; N ishimura, T ; T odo, S ; Hirohashi, S . P lasma proteomics o f p ancreatic cancer p atients b y multi-dimensional l iquid chromatography and t wo-dimensional difference gel electrophoresis (2D-DIGE): up-regulation of leucine-rich a lpha-2-glycoprotein i n p ancreatic can cer. J. Chromatogr. B, 2007, 8 52, 257-267.

Bloomston, M ; Zhou, JX; R osemurgy, AS; Frankel, W; Muro-Cacho, CA; Y eatman, T J. Fibrinogen γ overexpression in pancreatic cancer identified by large-scale p roteomic analysis of serum samples. Cancer Res., 2006, 66, 2592-2599.


Malik, G; Ward, MD; Gupta, SK; Trosset, MW; Grizzle, WE; Adam, BL; Diaz, JI, Semmes, OJ. Serum levels of an isoform of apolipoprotein A2 as a potential marker for prostate cancer. Clin. Cancer Res., 2005, 11, 1073-1085.

Tolson, J; B ogumil, R ; B runst, E ; B eck, H ; Elsner, R ; H umeny, A ; Kratzin, H ; D eeg, M ; Kuczyk, M ; M ueller, G A; Mueller, C A; F lad, T . S erum pr otein pr ofiling b y S ELDI mass s pectrometry: d etection o f m ultiple variants o f s erum a myloid a lpha in r enal cancer patients. Lab. Invest., 2004, 84, 845-856.

Akiba, J ; Yano, H; Ogasawara, S ; H igaki, K ; K ojiro, M . Expression a nd f unction of interleukin-8 in human hepatocellular carcinoma. Int. J. Oncol., 2001, 18, 257-264.

Aivado, M; Spentzos, D; Germing, U; Alterovitz, G; Meng, XY; Grall, F; Giagounidis, AA; Klement, G; Steidl, U; Otu, HH; Czibere, A; Prall, WC; Iking-Konert, C; Shayne, M; Ramoni, M F; G attermann, N; Haas, R ; Mitsiades, C S; F ung, E T; L ibermann, T A. Serum pr oteome pr ofiling detects m yelodysplastic s yndromes a nd i dentifies C XC chemokine ligands 4 and 7 as markers for advanced disease. Proc. Natl. Acad. Sci. USA, 2007, 104, 1307-1312.

Ren, Y; Poon, R T; Tsui, HT; C hen, WH; L i, Z ; Lau, C ; Y u, WC; F an, S T. I nterleukin-8 serum levels in p atients w ith h epatocellular c arcinoma: c orrelations w ith clinicopathological features and prognosis. Clin. Cancer Res., 2003, 9, 5996-6001.

Yang, M; Chen, T; Han, C; Li, N; Wan, T; Cao, X. Rab7b, a novel lysosome-associated small GTPase, i s i nvolved i n monocytic di fferentiation of hum an a cute pr omyelocytic leukemia cells. Biochem. Biophys. Res. Commun., 2004, 318, 792-799.

Benson, MD; Cohen, AS. Serum amyloid A protein in amyloidosis, rheumatic and neoplastic diseases. Arthritis. Rheum., 1979, 22, 36-42.

Uhlar, CM; Whitehead, AS. Serum amyloid A, the major vertebrate acute-phase reactant. Eur. J. Biochem., 265, 501-523.

Cho, WC; Yip, TT; Yip, C; Yip, V; Thulasiraman, V; Ngan, RK; Yip, TT; Lau, WH; Au, JS; Law, SC; Cheng, WW; Ma, VW; Lim, CK. Identification of serum amyloid a protein as a potentially useful biomarker t o monitor r elapse of na sopharyngeal c ancer by serum proteomic profiling. Clin. Cancer Res., 2004, 10, 43-52.

Storm, FK; Gilchrist, KW; Warner, TF; Mahvi, DM. Distribution of HSP-27 and HER-2/neu in in situ and invasive ductal breast carcinomas. Ann. Surg. Oncol., 1995, 2, 43-48.

Assimakopoulou, M ; Sotiropoulou-Bonikou, G ; Maraziotis, T; V arakis, I. P rognostic significance o f H SP-27 i n a strocytic br ain t umors: a n i mmunohistochemical s tudy. Anticancer Res., 1997, 17, 2677-2682.

Yu, Y; Chen, S; Wang, LS; Chen, WL; Guo, WJ; Yan, H; Zhang, WH; Peng, CH; Zhang, SD; L i, H W; C hen, GQ. Prediction o f p ancreatic can cer b y s erum b iomarkers u sing surface-enhanced l aser d esorption/ionization-based d ecision t ree cl assification. Oncology, 2005, 68, 79-86.

Wei, Y S; Z heng, Y H; L iang, W B; Z hang, JZ; Y ang, Z H; Lv, M L; J ia, J ; Z hang, L . Identification of s erum bi omarkers f or na sopharyngeal c arcinoma b y pr oteomic analysis. Cancer, 2008, 112, 544-551.

Wang, J; Zhang, X; Ge, X; Guo, H; Xiong, G; Zhu, Y. Proteomic studies of early-stage and advanced ovarian cancer patients. Gynecol. Oncol., 2008, 111, 111-119.

Yoshida, H; Shiratori, Y; Moriyama, M; Arakawa, Y; Ide, T; Sata, M; Inoue, O; Yano, M; Tanaka, M ; F ujiyama, S ; N ishiguchi, S ; K uroki, T ; I mazeki, F; Y okosuka, O ; Kinoyama, S; Y amada, G; O mata, M . I nterferon t herapy r educes t he r isk f or


hepatocellular c arcinoma: na tional s urveillance pr ogram of c irrhotic a nd nonc irrhotic patients w ith c hronic he patitis C i n J apan. I HIT S tudy G roup. I nhibition of Hepatocarcinogenesis by Interferon Therapy. Ann. Intern. Med., 1999, 131, 174-181.

Heathcote, EJ. Prevention of he patitis C virus-related h epatocellular car cinoma. Gastroenterology, 2004, 127, S294-302.

Uto, H; Hayashi, K; Kusumoto, K; Hasuike, S; Nagata, K; Kodama, M; Ido, A; Kohara, M; Stuver, SO; Tsubouchi, H. Spontaneous elimination of hepatitis C virus RNA in individuals w ith pe rsistent i nfection i n a hyperendemic a rea of Japan. Hepatol. Res., 2006, 34, 28-34.

Kusumoto, K; Uto, H; Hayashi, K; Takahama, Y; Nakao, H; Suruki, R; Stuver, SO; Ido, A; Tsubouchi, H. Interleukin-10 or tumor necrosis factor-α polymorphisms and the natural course of hepatitis C virus infection in a hyperendemic area of Japan. Cytokine, 2006, 34, 24-31.

Uto, H; Kurogi, J; Takahama, Y; Kusumoto, K; Hayashi, K; Ido, A; Kohara, M; Stuver, SO; Moriuchi, A; Hasegawa, S; Oketani, M; Tsubouchi, H. Alanine aminotransferase flare-up in hepatitis C virus carriers with persistently normal alanine aminotransferase levels in a hyperendemic area of Japan. J. Gastroenterol., 2007, 42, 673-680.

Tang, HY; Ali-Khan, N; Echan, LA; Levenkova, N; Rux, JJ; Speicher, DW. A novel four-dimensional s trategy c ombining pr otein a nd pe ptide s eparation m ethods e nables detection o f lo w-abundance pr oteins i n human pl asma a nd s erum pr oteomes. Proteomics, 2005, 5, 3329-3342.

Fujii, K; N akano, T; K awamura, T ; U sui, F ; B ando, Y; W ang, R ; Nishimura, T . Multidimensional pr otein p rofiling t echnology a nd i ts a pplication t o hu man plasma proteome. J. Proteome Res., 2004, 3, 712-718.

Welsh, JB; Sapinoso, LM; Kern, SG; Brown, DA; Liu, T; Bauskin, AR; Ward, RL; Hawkins, NJ; Quinn, DI; Russell, PJ; Sutherland, RL; Breit, SN; Moskaluk, CA; Frierson, HF; Jr; Hampton, GM. Large-scale delineation of secreted protein biomarkers overexpressed in cancer tissue and serum. Proc. Natl. Acad. Sci. USA, 2003, 100, 3410-3415.

Roeßler, M; Rollinger, W; Palme, S; Hagmann, ML; Berndt, P; Engel, AM; Schneidinger, B; Pfeffer, M; Andres, H; K arl, J ; B odenmüller, H ; Rüschoff, J; Henkel, T; R ohr, G; Rossol, S; Rösch, W; Langen, H; Zolg, W; Tacke, M. Identification of nicotinamide N-methyltransferase as a n ovel s erum tumor marker for co lorectal can cer. Clin. Cancer Res., 2005, 11, 6550-6557.

Hwang, T L; Liang, Y ; C hien, K Y; Y u, JS. Overexpression a nd e levated s erum l evels of phosphoglycerate k inase 1 in pancreatic ductal adenocarcinoma. Proteomics, 2006, 6, 2259-2272.

Law, S K; D odds, A W. The i nternal t hioester a nd t he c ovalent bi nding pr operties of t he complement proteins C3 and C4. Protein Sci., 1997, 6, 263-274.

Rother, K ; T ill, G O; Hansch, GM. T he c omplement s ystem, s econd e d., S pringer pr ess: Berlin; 1998.

Sim, RB; Laich, A. Innate immunity: serine proteases of the complement system. Biochem. Soc. Trans., 2000, 28, 545-550.

Veenstra, TD; Prieto, DA; Conrads, TP. Proteomic patterns for early cancer detection. Drug Discov. Today, 2004, 9, 889-997.

Crowther, JR. The ELISA guidebook: methods in molecular biology. Totawa: Humana Press; 2001.


Nielsen, U B; G eierstanger, B H. M ultiplexed s andwich a ssays i n microarray f ormat. J. Immunol. Methods, 2004, 290, 107-120.

www.elsevier.com/locate/ybbrc

Biochemical and Biophysical Research Communications 334 (2005) 1004–1013

BBRC

Demonstration of an in vivo generated sub-picomolar affinityfully human monoclonal antibody to interleukin-8

Palaniswami Rathanaswami a,*, Shelly Roalstad b, Lorin Roskos c, Qiaojuan Jane Su c,Steve Lackie b, John Babcook a

a Abgenix Biopharma Inc., Burnaby, BC, Canadab Sapidyne Instruments Inc., Boise, ID, USA

c Abgenix Inc., Fremont, CA, USA

Received 22 June 2005Available online 13 July 2005

Abstract

The high specificity and affinity of monoclonal antibodies make them attractive as therapeutic agents. In general, the affinities ofantibodies reported to be high affinity are in the high picomolar to low nanomolar range and have been affinity matured in vitro. Ithas been proposed that there is an in vivo affinity ceiling at 100 pM and that B cells producing antibodies with affinities for antigenabove the estimated ceiling would have no selective advantage in antigen-induced affinity maturation during normal immuneresponses. Using a transgenic mouse producing fully human antibodies, we have routinely generated antibodies with sub-nanomolaraffinities, have frequently rescued antibodies with less than 10 pM affinity, and now describe the existence of an in vivo generatedanti-hIL-8 antibody with a sub-picomolar equilibrium dissociation constant. This confirms the prediction that antibodies with affin-ities beyond the proposed affinity ceiling can be generated in vivo. We also describe the technical challenges of determining such highaffinities. To further understand the importance of affinity for therapy, we have constructed a mathematical model to predict therelationship between the affinity of an antibody and its in vivo potency using IL-8 as a model antigen.� 2005 Elsevier Inc. All rights reserved.

Keywords: Human antibodies; B cells; Affinity ceiling; High affinity; Sub-picomolar affinity; Affinity maturation; KinExA; SLAM; Interleukin-8 (IL-8); XenoMouse

Ever since the development of hybridoma technologya quarter century ago, there has been a promise formonoclonal antibodies (mAbs) to be used as humantherapeutic agents. Eighteen therapeutic mAbs arenow on the market in the United States, with 16 of thesehaving been approved in the last decade, and over 100mAbs are currently in clinical development [1].

For the most part, the recent success of mAbs can beattributed to advances in antibody generation technolo-gies. The first products of hybridoma technology weremurine antibodies. The usefulness of these antibodies

0006-291X/$ - see front matter � 2005 Elsevier Inc. All rights reserved.

doi:10.1016/j.bbrc.2005.07.002

* Corresponding author. Fax: +1 604 676 8349.E-mail address: [email protected] (P. Rathana-

swami).

as human therapeutics was limited by their immunoge-nicity; patients treated with murine antibodies rapidlymounted an immune response to these foreign proteins,a response known as a HAMA, for human anti-mouseantibody [2]. To avoid the HAMA response, we havedeveloped XenoMouse strains that functionally recapit-ulate the human antibody response, including a vast rep-ertoire of high affinity, somatically hypermutatedhuman antibodies [3,4].

In addition to immunogenicity, the kinetic propertiesof the antibody often dictate their utility. A higher affin-ity antibody will either be able to bind its ligand faster(determined by the association rate constant, kon),remain bound longer (determined by the dissociationrate constant, koff) or possess both properties. Depend-

mailto:[email protected]

P. Rathanaswami et al. / Biochemical and Biophysical Research Communications 334 (2005) 1004–1013 1005

ing on the tissue distribution, expressed forms (mem-brane-bound or soluble), concentration, and biologicalactivity of a target, there can be a direct correlation be-tween the affinity and potency of an antibody [5]. Theadministration of an antibody of higher affinity andpotency may also translate into better in vivo efficacy[6,7]. Antibodies of higher affinity may be able to beused at lower doses to achieve the desired clinical effects.Lower dosing may allow for more convenient routes ofadministration and decreased injection volumes, whichwould translate into lower cost of goods.

Antibodies reported to be of high affinity are general-ly in the nanomolar range [8,9] and occasionally in thesub-nanomolar range [10]. Extensive in vitro modifica-tions have been employed to increase the affinity ofthe antibodies to the picomolar range [7,11–13]. Theuse of XenoMouse mice, in conjunction with well-estab-lished antibody generation procedures, reproducibly re-sults in high affinity fully human mAbs suitable forrepeated administration to humans [14,15].

There are a number of technologies available to mea-sure the affinity of an antigen–antibody interaction inthe nanomolar range [16–20]. However, to measure theaffinity of an interaction in the low picomolar range, anumber of parameters may need to be modified [21]. Itwas recently shown that for two of these technologies,Biacore and KinExA, similar kinetic rate constants andaffinities could be measured for a number of antibodiesover a range of high affinity interactions [15]. Here, we de-scribe the affinity characterization of very high affinity(low picomolar to sub-picomolar) fully human antibodiesgenerated from XenoMouse animals and the modifica-tions required to measure these high affinities with preci-sion using KinExA. Experiments were performed in twoseparate laboratories: Abgenix Biopharma, Burnaby,Canada and Sapidyne Instruments, Boise, USA.

Although it has been hypothesized that there could bean intrinsic affinity ceiling (100 pM) for the selection ofantibodies generated in vivo [22,23], this report supportsthe existence of higher affinity antibodies derived fromin vivo somatic hypermutation and affinity maturation,with affinities in the sub-picomolar range. This impliesthat beyond the selection-driven affinity ceiling, higheraffinities can still occur by chance and are not selectedagainst. Alternatively, the assumptions underlying thelimits of selection-driven affinity may need to bere-visited.

Materials and methods

Instrumentation and materials. Solution based kinetic exclusionassays were performed using the KinExA 3000 instrument (SapidyneInstruments, Boise, ID). NHS-activated Sepharose 4 fast flow, CNBr-activated Sepharose 4B and protein A–Sepharose were obtained fromAmersham Pharmacia (Amersham Biosciences, Piscataway, NJ). BSAwas obtained from Sigma (St. Louis, MO). Cy5-conjugated (fluores-

cent dye based on indodicarbocyanine) affinity-purified goat anti-human IgG (Fcc fragment-specific), goat anti-human IgG (H + L-specific) and mouse anti-human IgG [F(ab 0)2 fragment-specific] Abs(all three have minimal cross-reactivity to bovine, horse and mouseserum proteins) were purchased from Jackson ImmunoResearchLaboratories (West Grove, PA). The recombinant human interleukin-8(rhIL-8) (72 amino acid form) (CXCL8) (Cat. No. 200-08M) wasobtained from PeproTech, Canada (Ottawa, ON). Titermax goldadjuvant was obtained from Sigma (Oakville, ON).

Generation of fully human anti-human interleukin-8 monoclonal

antibodies. XenoMouse strains were produced as described previously[4]. Cohorts of XenoMouse mice were immunized with rhIL-8. Theinitial BIP (base of the tail by s.c. injection and i.p.) immunization waswith 50 lg of rhIL-8, mixed 1:1 v/v with complete Freund�s adjuvant(CFA), per mouse. Subsequent boosts were made with 50 lg of rhIL-8,mixed 1:1 v/v with incomplete Freund�s adjuvant (IFA), per mouse.The animals were immunized on days 0, 14, 28, and 42. Thenimmunizations were continued for some mice with 50 lg of rhIL-8 inTitermax gold adjuvant i.p. (on days 146, 160, and 181) and then with10 lg of rhIL-8 in PBS i.p. on day 205. A final boost was done with10 lg of rhIL-8 in PBS i.p. on day 226 (for two mice) and on day 234(for two mice). Spleen and lymph nodes from XenoMouse animalswere harvested 4 days after the last boost. The application of SLAMtechnology [24,25] resulted in the rescue of multiple high affinity, fullyhuman anti-hIL-8 mAbs from the hyper-immunized XenoMouse ani-mals. Briefly, B cells from the animals were harvested and cultured inplates [25]. Wells were screened by ELISA to identify B cells producingIL-8-specific antibodies. 1063 wells were identified as being positive forIL-8 binding. The single B cells secreting anti-IL-8 antibodies wereidentified using an antigen-specific hemolytic plaque assay, picked bymicromanipulation, the heavy and light chain variable region cDNAswere amplified by RT-PCR and molecularly cloned into expressionvectors as previously described [25]. The expressed recombinant anti-bodies were purified using protein A–Sepharose affinity chromatog-raphy and analyzed by non-reducing SDS–PAGE to assess purity andyield. Concentration was also confirmed by UV analysis at A280 andELISA.

Preparation of antigen coated beads. Fifty micrograms of rhIL-8was coupled to CNBr-activated Sepharose 4B. Alternatively, 25 lg/mlor 100 lg/ml of rhIL-8 was coupled to NHS-activated Sepharose. Theremaining active groups on the beads were blocked as recommendedby the manufacturer. The beads were then finally blocked with 10 mg/ml BSA in 1 M Tris and stored in the blocking solution.

KinExA equilibrium assays. The experiments with KinExA wereperformed using an automated flow immunoassay system, KinExA3000 [26], in which rhIL-8-coupled beads served as the solid phase.Briefly, a constant amount of antibody between 0.67 and 5000 pMbinding site concentration (two binding sites for an IgG) was incubatedwith titrating concentrations of rhIL-8 antigen starting at 100 nM inPBS with 0.1% BSA (sample buffer) to reduce non-specific binding.Antigen–antibody complexes were incubated at RT for 36–144 h toallow equilibrium to be reached. The mixture was drawn through therhIL-8-coupled beads to accumulate unbound antibody. The capturedanti-hIL-8 mAb is directly proportional to the remaining free bindingsites [26], and was detected using solutions containing Cy5-conjugatedanti-human secondary antibody in sample buffer. The concentrations,volumes, and flow rates of the secondary antibody solutions werevaried to optimize the signal to noise ratio in each experiment. Theschematics of the experiment and the measurement of free mAb havebeen discussed previously [20,26]. The bound signals were convertedinto relative values as a proportion of control in the absence of rhIL-8.Three replicates of each sample (in some experiments two replicates)were measured for all equilibrium experiments. The equilibrium dis-sociation constant (Kd) was obtained from non-linear regressionanalysis of the data using a one-site homogeneous binding modelcontained within the software [20,27]. The software calculates the Kd

and determines the 95% confidence interval by fitting the data points to

Table 1Kd measurements of anti-hIL-8 mAbs by KinExA using standardconditions

Anti-hIL-8 mAb Kd (pM) Kd range (pM) mAb conc. (pM)

33 280 150–420 10142 400 190–680 25203 190 64–340 25215 360 230–450 50469 870 640–1010 200809 2.2 0.36–4.8 23837 11 0.054–31 90861 2.9 0.010–8.2 25928 0.057 <0.010–1.8 20

1064 54 29–72 501080 630 160–980 251093 200 76–300 100

Serial dilutions of rhIL-8 were mixed with the respective binding siteconcentration of anti-hIL-8 mAbs and allowed to reach equilibriumfor 36 h. A proportionate amount of free mAb present in the equi-librium mixture was measured by flowing 0.25–3 ml of the sample onrhIL-8 coupled to CNBr-activated Sepharose beads and detected using1.0 ml of 1.7 lg/ml of Cy5-labeled goat anti-human IgG Fcc fragment-specific secondary antibody. Each sample was run in triplicate. The Kd

was calculated by the KinExA software and the 95% confidenceinterval is given as the Kd range.

1006 P. Rathanaswami et al. / Biochemical and Biophysical Research Communications 334 (2005) 1004–1013

a theoretical Kd curve. The 95% confidence interval is given as Kd lowand Kd high.

KinExA kinetic assays. For measuring the association rate con-stant using KinExA, the same rhIL-8-coupled beads were used as theprobe and the ‘‘Kinetics, Direct’’ method was used. The ‘‘Kinetics,Direct’’ experiments are identical to KinExA equilibrium assays withrespect to bead column height, antibody capture, antibody concen-tration, and antibody detection. However, the mAb was mixed withan amount of antigen that bound approximately 80% of the mAb inthe equilibrium experiments and the free antibody present in thesample was probed repeatedly, pre-equilibrium. Since the bindingsignals are proportional to the concentration of free antibody in thesolution, the signals decrease over time until the solution has equil-ibrated. The volumes and flow rates of the antigen–mAb mixturesand the Cy5-labeled secondary antibody were varied depending uponthe mAb tested. Data were analyzed utilizing the KinExA analysissoftware that comes with the KinExA 3000 instrument. This softwaregraphically represents the decrease in binding signals over time, andfits the collected data points to an exact solution of the kinetic dif-ferential equations for binding. From this curve, an optimal solutionfor the kon is determined. The koff is indirectly calculated fromsolutions for the kon and Kd.

Theoretical effect of affinity on potency. A mathematical modelwas developed to simulate the effect of affinity on the dose of mAbrequired to suppress the serum IL-8 levels in vivo by at least 90% atsteady-state following a multiple dose regimen. The model consistedof four differential equations describing the synthesis and eliminationof endogenous IL-8, the dosing and elimination of mAb, and themonovalent and divalent binding of mAb to the IL-8. The half-life ofthe mAb was assumed to be 3 weeks, and dosing was conductedevery 3 weeks for 15 weeks. The simulation was conducted at 3, 30,300 pM, and 3 nM baseline levels of IL-8 to reflect concentrationsthat may be present in serum and at active sites of inflammation,respectively [28–30]. Simulations were conducted using SAAM II v.1.2 (SAAM Institute, Seattle, WA). Numerical integration was con-ducted using the Rosenbrock method with an adjustable step size.The time course in serum of unbound IL-8 (IL8u), unbound mAb(ABu), mAb monovalently bound to IL8 (ABmv), and mAb biva-lently bound to IL8 (ABbv) was described by the followingequations:

dIL8udt

¼ S0 þ fluxunbindmvþ fluxunbindbv� fluxbindmv

� fluxbindbv� CLag

VIL8u;

dABudt

¼ fluxunbindmv� fluxbindmv� CLab

VABu;

dABmvdt

¼ fluxbindmvþ fluxunbindbv� fluxunbindmv� fluxbindbv

� CLab

VABmv;

dABbvdt

¼ fluxbindbv� fluxunbindbv� CLab

VABbv;

where:

fluxbindmv ¼ 2konABuIL8u

V

� �;

fluxbindbv ¼ konABmvIL8u

V

� �;

fluxunbindmv ¼ koffABmv;

fluxunbindbv ¼ 2koffABbv.

The zero-order production rate of IL-8, S0, was varied to achievedifferent initial conditions for baseline levels of unbound IL-8. In thesimulations, parameter values for CLab, CLag, and V were fixed at0.0025, 0.493, and 0.064 L/kg, respectively. The clearance estimateswere based on unpublished data, and the volume of distribution as-sumed limited distribution of mAb and IL-8 into extra vascular space.Parameter sensitivity analysis indicated that changes in assumptionsregarding clearance and distribution of mAb and IL8 did not influencethe estimate of optimum mAb affinity; however, the dose required toachieve 90% suppression of unbound IL-8 was influenced by theseassumptions. The association rate constant, kon, was fixed and changesin affinity were assumed to be related to changes in the dissociationrate constant, koff. For any given affinity, changes in the values of koffand kon did not affect the simulated steady-state concentration of un-bound IL-8.

Results

Affinity measurements of Anti-hIL-8 mAbs by KinExA

The affinity measurements of anti-hIL-8 mAbs fromXenoMouse determined by KinExA are shown inTable 1. Generally, when the measured Kd of an anti-body was in the range of high picomolar or above, the95% confidence interval of the predicted Kd was withina narrow range. However, when the measured Kd wasapproaching low picomolar to sub-picomolar values,the 95% confidence interval tended to be broader, indi-cating the measured Kd was not as precise. For thelow to sub-picomolar Kd measurements, the binding siteconcentration of mAb used in the equilibrium mixturewas higher than the Kd, causing the precision of these


measurements to be less accurate, thus warrantingKinExA signal optimization experiments.

Signal optimization

In order to make an accurate Kd measurement byKinExA, the active binding site concentration of theantibody must be near or below the actual Kd. To mea-sure antibodies with very high affinities, as described inthis paper, one needs to work at low picomolar bindingsite concentrations of antibody. At these low antibodyconcentrations the sensitivity of the assay needs to befurther optimized. To increase the detection sensitivityof free antibody, bead coating concentrations and sam-pling parameters were optimized and different labelswere evaluated.

Optimization of bead coating and labeled secondary

antibody

Immobilization of certain antigens to beads may af-fect their ability to bind antibodies, due to steric hin-drance by passive coating or destruction of epitopes byside chain reactive coupling. To test whether the signalin the bead pack could be increased, NHS-activatedSepharose beads were coupled with rhIL-8 and the beadsaturation was tested, as previously described [26]. Itwas determined that there was no signal gain above acoupling concentration of 100 lg/ml of antigen. At25 lg/ml of antigen, approximately 70% of the maxi-mum signal was observed. Subsequently, the NHS–Sepharose beads were coupled with 100 lg/ml ofrhIL-8 for the low antibody binding site concentration(Kd controlled) experiments. The Kd measurements withhigh binding site concentration of mAb and on-rateexperiments were performed with 25 lg/ml of rhIL-8to conserve antigen.

Different Cy5-labeled anti-human secondary antibod-ies were checked to identify one that would give thehighest signal-to-noise ratio. When a 100 pM solutionof anti-hIL-8 mAb was tested with various secondary la-bels (goat anti-human IgG (H + L)-specific, Fcc frag-ment-specific or mouse anti-human IgG F(ab 0)2fragment-specific), the highest signal to noise ratio wasgenerated using the Fcc-specific label (9.9 times higherthan background). This Fcc fragment-specific goatanti-human secondary antibody was used for subse-quent equilibrium and on-rate measurements.

Optimization of sample volume and flow rate

Since the initial equilibrium analysis of a few anti-hIL-8mAbs showed low picomolar to sub-picomolarKds withbroader 95% confidence intervals (Table 1;mAb 809, 837,861, 928), the sample volume and flow rate were then opti-mized to obtain a higher signal. When a 100 pM of

anti-hIL-8 mAb 928 was tested either flowing 0.5 ml at0.25 ml/min or 3.0 ml at 1.5 ml/min for 120 s, it was deter-mined that increasing the sample volume and flow rategave about 20% higher signal (data not shown).

The binding site concentration of the anti-hIL-8 mAb928 was then lowered to 2 pM and a signal test was per-formed by flowing 30 ml at 1.5 ml/min for 1200 s. How-ever, when the sample flow volume and the rate wereraised, bubbles were forming in the bead pack, whichaffected the raw data traces. This problem was solvedby degassing the samples and label immediately beforerunning an experiment.

Time to reach equilibrium and sample viability

The antibody binding site concentration was subse-quently lowered to 1 pM, to measure the Kd of low tosub-picomolar mAbs. At this very low binding site con-centration of antibody, the samples require a longer timeto reach equilibrium. Using the Kd curve obtained at alow mAb concentration and the measured kon, it wasestimated in the kinetic theory curve software fromSapidyne Instruments that the minimum time neededto reach equilibrium would be 6 days (data not shown).Otherwise, one could experimentally determine whetherthe antigen–antibody mixture reached equilibrium bymeasuring the free mAb left in solution, which shouldremain constant during triplicate measurements, ifequilibrium has been achieved. The viability of 1 pManti-hIL-8 mAb 928 was then checked at 0, 3, and 6day incubation times and found that the antibodyactivity had not decreased during this period at roomtemperature. Subsequently, all equilibrium reactionswere allowed to take place for 6 days.

Equilibrium and on-rate measurements of mAbs having

low picomolar affinity

The Kd measurements were then repeated, for themAbs (809, 837, 861, and 928) which had Kds below20 pM and broad 95% confidence intervals (Table 1),with optimized parameters for bead coupling, samplevolume, flow rate, and secondary antibody-label. InKinExA, the percent of free mAb left in solution is plot-ted against each concentration of antigen, generating asigmoidal curve, as shown in Fig 1. A theoretical curveis fit to the data and 95% confidence intervals are calcu-lated by the KinExA software. It is known that when thebinding site concentration of mAb is very close to, or be-low the Kd value, the shape of the curve is constant andis referred to as a Kd controlled curve (Fig. 1, curve 1).However, for mAb binding site concentrations abovethe Kd, the curve shifts towards the right and becomessteeper (Fig. 1, curve 2). These steeper curves are re-ferred to as mAb concentration controlled curves, andwill contain little information on the Kd. One rationale

Table 2Kd measurements of the highest affinity anti-hIL-8 mAbs by KinExAusing optimized conditions

Anti-hIL-8 mAb Kd (pM) Kd range (pM) mAb conc. (pM)

809 3.3 1.9–5.2 4.6, 27837 16 9.3–25 18, 120861 3.0 2.0–4.2 1.3, 13928 0.61 0.38–0.94 0.68, 2.0, 14

Serial dilutions of rhIL-8 were mixed with the respective anti-hIL-8mAbs in the binding site concentrations shown above and allowed toreach equilibrium for 36 h for the high binding site concentration ofeach mAb, or for 144 h for the low binding site concentration of eachmAb. A proportionate amount of free mAb present in the equilibriummixture was measured in KinExA under optimized conditions. Forexample, for the mixtures containing the low binding site concentra-tion of mAb, 18–36 ml of the sample was flowed through rhIL-8coupled to NHS-activated Sepharose beads and detected using 1–2 mlof 2.0 lg/ml of Cy5-labeled goat anti-human IgG Fcc fragment-spe-cific secondary antibody. The Kd was calculated using KinExA soft-ware by n-curve analysis and the 95% confidence interval is given as theKd range. Six experiments were performed for mAb 928 and two wereperformed for mAbs 809, 837, and 861.

Kd Distribution for mAb 928

0

200

400

600

800

1000

1200

1400

1600

0 1 2 3 4 5 6 7 8 9

Kd

(fM

)

Experiment no.

Fig. 2. Distribution of Kd measurements for the sub-picomolar affinityanti-hIL-8 mAb 928 by KinExA. Two fold dilutions of rhIL-8 (50 pM–24 fM) were mixed with anti-hIL-8 mAb 928 at an active binding siteconcentration of either 2 pM (Experiment No.1) or 670 fM (Experi-ments 2–4). Experiments 5–8 represent n-curve analysis, done eitherwith 2 pM and 13.5 pM mAb (Experiment No. 5) or with 670 fM and13.5 pM mAb (Experiments 6–8). The diamonds represent the optimalKd determined by KinExA data analysis for each experiment. Theerror bars represent the 95% confidence intervals from KinExA dataanalysis. The error bars are not centered on the optimal Kd valuebecause of the non-linear fit of the theoretical curve to the equilibrium.

0

20

40

60

80

100

120

1.00E-15 1.00E-12 1.00E-09

Concentration of Antigen (M)

% F

ree

mA

b

Curve 1 Curve 2

Fig. 1. Kd measurement of mAb 928 by KinExA. Twofold dilutions(50 pM to 24 fM) of rhIL-8 were mixed with 1 pM binding siteconcentration of anti-hIL-8 mAb 928 and allowed to reach equilibriumfor 144 h. A proportionate amount of free mAb present in theequilibrium mixture was measured by flowing 22.5 ml of the sample at0.25 ml/min on rhIL-8 coupled to NHS-activated Sepharose beads.The bead bound mAb was detected by flowing 2.0 ml of 2.0 lg/ml ofCy5-labeled goat anti-human IgG Fcc fragment-specific secondaryantibody at 0.25 ml/min. Each sample was run in duplicate. The Kd

calculated by the KinExA software by single curve analysis for arepresentative experiment (curve 1) was 870 fM and the 95% confi-dence interval calculated as the Kd high and Kd low was 1.3 pM and500 fM, respectively. A representative mAb controlled experiment(curve 2) was performed using 20 pM binding site concentration ofanti-hIL-8 mAb 928, mixed with two fold dilutions (200 pM–97 fM) ofrhIL-8 and allowed to reach equilibrium for 48 h. A proportionateamount of free mAb present in the equilibrium mixture was measuredby flowing 5 ml of the sample at 0.5 ml/min on rhIL-8 coupled toNHS-activated Sepharose beads. The bead bound mAb was detectedby flowing 1.0 ml of 2.0 lg/ml of Cy5-labeled goat anti-human IgGFcc fragment-specific secondary antibody at 0.25 ml/min. Each samplewas run in triplicate. The Kd was calculated to be 790 fM by usingn-curve analysis, by the KinExA software and the 95% confidenceinterval was calculated to be between 1.1 pM (Kd high) and 570 fM (Kd

low). Analysis of curve 2 also determined the active binding siteconcentration of antibody present in each reaction to be equal to 67%.


for performing a mAb concentration controlled experi-ment is that the KinExA software will be able to calcu-late precisely the active binding site concentration ofmAb present in solution during the equilibriumreaction.

Additionally, one can further increase the accuracy ofvery high affinity measurements by performing one ormore experiments using a mAb binding site concentra-tion near or slightly higher than its Kd value and at leastone or more experiments with a 10- to 100-fold higherbinding site concentration of mAb than the first experi-ment [15]. All these experiments can then be analyzedsimultaneously with the KinExA software by using the‘‘n-curve analysis’’ selection. This ‘‘n-curve analysis’’selection allows one to obtain a precise Kd value (Table2) by fitting all of the given curves to a single Kd valuesimultaneously. We performed a number of experimentsfor n-curve analysis of mAb 928. In these experiments,we used mAb 928 either near the Kd concentration orat a concentration of about 15- to 20-fold higher thanthe Kd. By n-curve analysis, the Kd of mAb 928 was

determined to be 610 fM (Kd high = 940 fM and Kd

low = 380 fM), which is very close to the Kd determinedby low mAb binding site concentration experiments ana-lyzed in single curves (590 ± 220 fM). The Kd distribu-tions of individual measurements of mAb 928 aregiven in Fig. 2.

The on-rate measurement by KinExA is not limitedby the mAb binding site concentration used in the exper-iment. Therefore, the measurements of kon for the veryhigh affinity antibodies were straightforward and did

Table 3On-rate measurements for the highest affinity anti-hIL-8 mAbs inKinExA

Anti-hIL-8 mAb kon (M�1 s�1) kon range (M�1 s�1) koff (s�1)

809 4.8E+6 4.5E+6–5.5E+6 1.6E�5837 1.1E+6 1.1E+6–1.2E+6 1.9E�5861 4.4E+6 4.0E+6–5.0E+6 1.3E�5928 6.0E+6 5.6E+6–6.4E+6 3.7E�6

The binding site concentration of mAb used for on-rate experimentswas the same as those used in equilibrium binding experiments. Anamount of IL8 that bound 80% of the mAb in equilibrium bindingexperiments was mixed with the mAb. The amount of free mAb left inthe mixture was measured in KinExA by flowing 1 ml of the mixturerepeatedly through rhIL-8 coupled to NHS-activated Sepharose beadsand detected using 1 ml of 2.0 lg/ml of Cy5-labeled goat anti-humanIgG Fcc fragment-specific secondary antibody. The kon was measuredusing the KinExA software by the kinetic direct method and the 95%confidence intervals are given as kon range. The koff for the mAb wascalculated from the measured kon and Kd of the respective mAb.


not require further optimizations. The measured kon andthe calculated koff are given in Table 3.

Simulated effect of affinity on potency

The theoretical dose required to suppress IL-8 in vivoby at least 90% at steady-state is shown in Fig. 3 as afunction of affinity for four baseline levels of endoge-nous IL-8 in serum. Improvements in mAb affinity arepredicted to provide proportional improvements inpotency until the Kd falls to 1/10th the endogenousIL-8 level. When the antigen level is at least 10 timesgreater than the Kd, the dose of mAb required to sup-press IL-8 by 90% is determined by antigen level andnot by affinity; thus, further improvements in affinity

Affinity (pM)

0.1 1 10 100 1000

Dos

e (m

g/kg

/3w

eeks

)

0.001

0.01

0.1

1

10

100

3 pM Baseline IL-830 pM Baseline IL-8300 pM Baseline IL-83 nM Baseline IL-8

Fig. 3. Theoretical effect of mAb affinity on potency. The dose (mg/kg/3 weeks) of mAb required to suppress IL-8 levels in vivo by at least90% at steady-state was simulated as a function of mAb affinity (Kd) atfour baseline levels of endogenous IL-8 (3 pM, circles; 30 pM,triangles; 300 pM, squares; 3 nM, diamonds). When the Kd of themAb is less than 1/10th the baseline IL-8 levels, further improvementsin affinity will not provide any improvements in potency.

are not predicted to yield a further improvement inpotency. For example, the mean concentration of IL-8in exudates from psoriatic plaques was reported to be250 pM [30]. Improvement of mAb affinity to approxi-mately 25 pM is predicted to decrease the dose requiredto suppress IL-8 concentrations in the psoriatic plaque;a further increase in affinity is not predicted to improvepotency.

Mutational analysis

The role of somatic mutation in optimizing hIL-8binding affinity was analyzed for mAbs 928, 809, 215,and 1064. These mAbs were derived from the samegermline sequence and thus the backbones should beessentially identical. Table 4 lists the mutations locatedwithin the complementary determining regions (CDRs)and framework regions (FRs) of each of these antibod-ies in comparison with the germline residues. The resi-dues are numbered according to the scheme of Kabatet al. [31]. Mutational analysis indicates that these fourmAbs were derived from an ancestral clone and aValH50 fi Asp mutation occurred within mAbs 215,809, and 928. The light chain of the mAb 215 indepen-dently diverged from the precursor ValH50 fi Asp cloneand generated a SerL27A fi Asn mutation resulting in ahigh affinity Kd of 362 pM. Then, mAbs 809 and 928 di-verged from the ancestral ValH50 fi Asp clone with theaddition of a HisH35 fi Leu mutation. Antibody 809independently further diverged at TyrH79 fi Phe inFR3 finally resulting in a higher affinity Kd of2.23 pM. The heavy chain of mAb 928 independentlycontinued to diverge in FR3 at TyrH90 fi Phe. The lightchain of the mAb 928 further mutated in the CDRs from

Table 4Somatic mutations in the CDR and FR regions of mAb 928, 809, 215and 1064

Kabatnumber

Germline mAb928

mAb809

mAb215

mAb1064

CDR regions

H50 Val Asp Asp Asp LeuH35 His Leu LeuL92 Gly AspL32 Tyr PheL27A Ser AsnL91 Tyr AspH96 Arg HisH31 Ser ThrL52 Ser TyrL53 Ser Arg

FR regions

L39 Lys ArgL72 Thr IleH79 Tyr PheH90 Tyr Phe

Residues are numbered according to the scheme of Kabat et al. [31].

,


GlyL92 fi Asp and TyrL32 fi Phe, as well as in FR3ThrL72 fi Ile, resulting in a final product with a superioraffinity in the sub-picomolar range (Kd = 610 fM). How-ever, mAb 1064 appears to have evolved independentlyfrom a germline ancestral clone and resulted in muta-tions in the CDRs at ValH50 fi Leu, SerH31 fi Thr,ArgH96 fi His, SerL52 fi Tyr, SerL53 fi Arg, andTyrL91 fi Asp, along with a mutation in FR2 at Ly-sL39 fi Arg, which ultimately resulted in a high affinityKd of 54 pM.

Discussion

There are a number of reports describing monoclonalantibodies with sub-nanomolar affinities [10,11,15].However, in vitro affinity maturation by phage, yeastor ribosome display is often required to increase theaffinity of an antibody to the picomolar range[11,13,32–34]. Here, we describe the affinity character-ization of fully human mAbs generated in vivo fromXenoMouse that were directly rescued using the selectedlymphocyte antibody method (SLAM) [24] and were notsubjected to further in vitro affinity maturation.Rigorous affinity analyses of these antibodies wereperformed at two separate laboratories utilizingKinExA technology.

A number of technologies, such as surface plasmonresonance (SPR) (Biacore) [16], resonant mirror (IAsys)[19], solution based kinetic exclusion assay (KinExA)[20], isothermal titration calorimetry [18], and fluores-cence-polarization [17] are used to measure equilibriumconstants. Commonly used Biacore technology utilizessurface-based biophysical methods to measure the konand koff, from which one can calculate the Kd. Whereas,KinExA utilizes solution-based biophysical methods todirectly measure the Kd and kon. The koff can then be cal-culated using the measured Kd and kon. Both Biacoreand KinExA have been successfully used to measureaffinities in the picomolar range [15,21,26,35]. Recently,it has been shown that for some antigen–antibody affin-ity measurements, KinExA and Biacore correlate closely[15]. However, Biacore requires proper experimental de-sign to avoid avidity, mass transport, steric hindrance,aggregation, and crowding effects, as reviewed extensive-ly before [21,36]. Here we have used KinExA technolo-gy, where both binding constituents are present insolution without being modified, to precisely measurelow picomolar affinities. Since the concentration ofIL-8 used in our kinetic equilibrium measurements were100 nM or below and IL-8 exists as a monomer at thisconcentration [37,38], we were able to measure the trueaffinity, rather than the avidity, of the anti-hIL-8 mAbs.We also describe modifications to standard equilibriummeasurement techniques for KinExA that allowed theprecise measurement of sub-picomolar affinities.

The challenge in KinExA for sub-picomolar affinitymeasurements is to precisely measure the proportionateamount of very low concentrations of free mAb left insolution after equilibrium is reached. To achieve this,we modified a number of parameters. As shown foranti-hIL-8 mAb 928, by coupling the NHS-activated Se-pharose beads with a saturating concentration of IL-8,increasing the sample flow volume, and using the opti-mized Cy5-labeled detection antibody, we were able toincrease significantly the signal-to-noise ratio. Thesemodifications together allowed us to measure the pro-portionate amount of free antibody left in solution toa very low concentration range. In addition, we incubat-ed the antibody–antigen mixture for a sufficient periodof time to allow the reactants to reach equilibrium. Byapplying these modified parameters, we performed anumber of Kd controlled experiments, which were alsoused in n-curve analysis along with mAb controlledexperiments and measured the average Kd of mAb 928to be in the sub-picomolar range. Using n-curve analy-sis, one may even be able to extend the system to deter-mine equilibrium dissociation constants in the lowfemtomolar range.

To our knowledge there has only been one publica-tion describing a sub-picomolar affinity antibody [34].This murine-derived, in vitro mutated and yeast displayaffinity matured antibody targets the hapten fluorescein.However, we have not seen any reports describing a sub-picomolar affinity antibody targeting a protein antigen,generated in vivo from any species. The reason for thismight be due to either the inherent difficulties in rescuingvery high affinity mAbs by available antibody genera-tion technologies or in measuring high affinity interac-tions with existing technologies. We have shown herethat if the technologies that measure the affinity, suchas solution-based KinExA, are used properly, high affin-ity measurements between a protein antigen and an anti-body can be measured with precision. The very highaffinity determined for mAb 928 was confirmed to besub-picomolar by two independent laboratories usingKinExA.

Additionally, we have demonstrated for the first timethat fully human mAbs that exhibit an equilibrium dis-sociation constant in the sub-picomolar range can begenerated in vivo. In addition, we are now regularly iso-lating low picomolar (less than 10 pM) fully human anti-bodies from XenoMouse [15]. Using SLAM technology,we have also generated a number of low picomolar affin-ity antibodies directly from human peripheral blood Bcells targeting a variety of antigens from human patho-gens. We have also generated many low picomolar affin-ity antibodies, as well as a sub-picomolar affinity mAbfrom rabbit peripheral blood B cells, using SLAM tech-nology (unpublished results). These observations indi-cate that the generation of exceptionally high affinityantibodies occurs across species in vivo.


It was proposed by Foote and Eisen [22] that, underphysiological conditions, there is a ceiling at 100 pM onantibody affinity maturation in vivo for any antigen. Itwas proposed that above this threshold there would beno selective in vivo advantage for somatic mutationand affinity maturation to happen. This argument wasbased on the limits on both on- and off-rate constantsof an antibody. The on-rate constant of an antibody iscontrolled by the diffusion coefficients of the antigenand antibody [39,40] and the maximum on-rate constantproposed was approximately 105–106 M�1 s�1. The pro-posed limit on the off-rate constant is based on the res-idence time of an antigen complexed to an antibody on aB cell surface before it is taken up and processed by theB cell [23,41]. These studies led to the conclusion that Bcells cannot differentiate between antibodies that haveoff-rate constants less than 10�4 s�1. However, no studydescribes a disadvantage for in vivo affinity maturationbeyond these affinity ceilings. In fact, Foote and Eisen[22] themselves suggested that affinities beyond the ceil-ing could happen randomly. We have generated anti-bodies with on-rates greater than 106 M�1 s�1 [15,42],with off-rates less than 10�4 s�1 [15] and the presentstudy extends those observations. Others have alsoreported on-rates higher than 106 M�1 s�1 for antibod-ies [43] and for Fab fragments [44]. Increased on-ratesbeyond the proposed diffusion limit may be explainedby additional electrostatic forces [45]. For the associa-tion of barnase with barstar, a basal association rateconstant of 105 M�1 s�1 was increased to over5 · 109 M�1 s�1 by electrostatic forces [45]. For off-rates, it is possible that B cell receptor turn-over in vivomay be slower or more heterogeneous than that mea-sured in vitro using cell lines [23]. In this case, B cellsexpressing antibodies with off-rates slower than10�4 s�1 may have a selective advantage. Alternatively,somatic hypermutation could stochastically result inthe generation of antibodies with higher affinities thanthe proposed in vivo ceiling, which are neither selectedfor nor selected against in vivo. Supporting this state-ment, it has even been predicted that a single amino acidsubstitution can alter binding kinetics in protein–proteininteractions by up to six logs [46]. In addition, somatichypermutations in the CDR and frame work regionsof an anti-HEL mAb resulted in a three log increase inaffinity [44].

Among the four high affinity mAbs (215, 809, 928,and 1064) that were derived from the same germlinesequences, it appears that the affinity has evolvedthrough a progressive somatic mutation process. Thesuperior affinity of mAb 928 (Kd = 610 fM) seems tobe the result of a combination of enhanced electrostaticinteractions by introduction of multiple aspartic acidmutations, and enhanced hydrophobic interactions withthe introduction of phenylalanine and leucine muta-tions. Alternatively, it is possible that the parental germ-

line sequence could have existed in multiple preexistingconformational isomers [47,48] and that somatic muta-tions in the antibodies resulted in the stabilization ofthe high affinity binding isomers.

Antibodies with affinities in the picomolar or femtom-olar range may provide significant improvements in mAbpotency, depending on the biological characteristics ofthe antigen, particularly the in vivo concentration(Fig. 3). By using SLAM, amore efficient antibody gener-ation procedure, to screen a larger proportion of the avail-able immune repertoire in hyperimmune animals, it waspossible to identify a number of exceptionally high affinityantibodies including one of sub-picomolar affinity with-out the need for time consuming in vitro manipulations.

We have determined the affinities of a panel of fullyhuman anti-hIL-8 antibodies generated in vivo fromXenoMouse animals to be in the low picomolar range.Furthermore, one of these antibodies was found to be asub-picomolar affinity antibody. As far as we are aware,this is the first such description of an in vivo generatedsub-picomolar affinity antibody. This report not onlyconfirms the prediction that there can be affinity matura-tion in vivo beyond the proposed affinity ceiling [22], butin fact provides precise affinity measurements of an anti-body that does so by over two orders of magnitude.

Acknowledgments

The authors thank Mike Gallo, Ian Foltz, and GeoffDavis for critical reading of the manuscript. We thankKaren Richmond for her technical assistance.

References

[1] J. Reichert, A. Pavolu, Monoclonal antibodies market, Nat. Rev.Drug Discov. 3 (2004) 383–384.

[2] G.G. Klee, Human anti-mouse antibodies, Arch. Pathol. Lab.Med. 124 (2000) 921–923.

[3] L.L. Green, M.C. Hardy, C.E. Maynard-Currie, H. Tsuda, D.M.Louie, M.J. Mendez, H. Abderrahim, M. Noguchi, D.H. Smith,Y. Zeng, et al., Antigen-specific human monoclonal antibodiesfrom mice engineered with human Ig heavy and light chain YACs,Nat. Genet. 7 (1994) 13–21.

[4] M.J. Mendez, L.L. Green, J.R. Corvalan, X.C. Jia, C.E.Maynard-Currie, X.D. Yang, M.L. Gallo, D.M. Louie, D.V.Lee, K.L. Erickson, J. Luna, C.M. Roy, H. Abderrahim, F.Kirschenbaum, M. Noguchi, D.H. Smith, A. Fukushima, J.F.Hales, S. Klapholz, M.H. Finer, C.G. Davis, K.M. Zsebo, A.Jakobovits, Functional transplant of megabase human immuno-globulin loci recapitulates human antibody response in mice, Nat.Genet. 15 (1997) 146–156.

[5] L.S. Zuckier, E.Z. Berkowitz, R.J. Sattenberg, Q.H. Zhao, H.F.Deng, M.D. Scharff, Influence of affinity and antigen density onantibody localization in a modifiable tumor targeting model,Cancer Res. 60 (2000) 7008–7013.

[6] J.S. Li, F. Chu, A. Reilly, G.M. Winslow, Antibodies highlyeffective in SCID mice during infection by the intracellularbacterium Ehrlichia chaffeensis are of picomolar affinity and


exhibit preferential epitope and isotype utilization, J. Immunol.169 (2002) 1419–1425.

[7] Z. Zhu, K. Hattori, H. Zhang, X. Jimenez, D.L. Ludwig, S. Dias,P. Kussie, H. Koo, H.J. Kim, D. Lu, M. Liu, R. Tejada, M.Friedrich, P. Bohlen, L. Witte, S. Rafii, Inhibition of humanleukemia in an animal model with human antibodies directedagainst vascular endothelial growth factor receptor 2. Correlationbetween antibody affinity and biological activity, Leukemia 17(2003) 604–611.

[8] A.D. Griffiths, S.C. Williams, O. Hartley, I.M. Tomlinson, P.Waterhouse, W.L. Crosby, R.E. Kontermann, P.T. Jones, N.M.Low, T.J. Allison, et al., Isolation of high affinity humanantibodies directly from large synthetic repertoires, EMBO J. 13(1994) 3245–3260.

[9] H.J. de Haard, N. van Neer, A. Reurs, S.E. Hufton, R.C.Roovers, P. Henderikx, A.P. de Bruine, J.W. Arends, H.R.Hoogenboom, A large non-immunized human Fab fragmentphage library that permits rapid isolation and kinetic analysis ofhigh affinity antibodies, J. Biol. Chem. 274 (1999) 18218–18230.

[10] T.J. Vaughan, A.J. Williams, K. Pritchard, J.K. Osbourn, A.R.Pope, J.C. Earnshaw, J. McCafferty, R.A. Hodits, J. Wilton, K.S.Johnson, Human antibodies with sub-nanomolar affinities isolat-ed from a large non-immunized phage display library, Nat.Biotechnol. 14 (1996) 309–314.

[11] Y. Chen, C. Wiesmann, G. Fuh, B. Li, H.W. Christinger, P.McKay, A.M. de Vos, H.B. Lowman, Selection and analysis of anoptimized anti-VEGF antibody: crystal structure of an affinity-matured Fab in complex with antigen, J. Mol. Biol. 293 (1999)865–881.

[12] G.P. Adams, R. Schier, Generating improved single-chain Fvmolecules for tumor targeting, J. Immunol. Methods 231 (1999)249–260.

[13] J. Hanes, C. Schaffitzel, A. Knappik, A. Pluckthun, Picomolaraffinity antibodies from a fully synthetic naive library selected andevolved by ribosome display, Nat. Biotechnol. 18 (2000) 1287–1292.

[14] L.L. Green, Antibody engineering via genetic engineering of themouse: XenoMouse strains are a vehicle for the facile generationof therapeutic human monoclonal antibodies, J. Immunol. Meth-ods 231 (1999) 11–23.

[15] A.W. Drake, D.G. Myszka, S.L. Klakamp, Characterizing high-affinity antigen/antibody complexes by kinetic- and equilibrium-based methods, Anal. Biochem. 328 (2004) 35–43.

[16] U. Jonsson, L. Fagerstam, B. Ivarsson, B. Johnsson, R. Karlsson,K. Lundh, S. Lofas, B. Persson, H. Roos, I. Ronnberg, et al.,Real-time biospecific interaction analysis using surface plasmonresonance and a sensor chip technology, Biotechniques 11 (1991)620–627.

[17] W.J. Checovich, R.E. Bolger, T. Burke, Fluorescence polariza-tion—a new tool for cell and molecular biology, Nature 375 (1995)254–256.

[18] L. Joss, T.A. Morton, M.L. Doyle, D.G. Myszka, Interpretingkinetic rate constants from optical biosensor data recorded on adecaying surface, Anal. Biochem. 261 (1998) 203–210.

[19] P.A. Lowe, T.J. Clark, R.J. Davies, P.R. Edwards, T. Kinning, D.Yeung, New approaches for the analysis of molecular recognitionusing the IAsys evanescent wave biosensor, J. Mol. Recognit. 11(1998) 194–199.

[20] R.C. Blake 2nd, A.R. Pavlov, D.A. Blake, Automated kineticexclusion assays to quantify protein binding interactions inhomogeneous solution, Anal. Biochem. 272 (1999) 123–134.

[21] D.G. Myszka, Kinetic analysis of macromolecular interactionsusing surface plasmon resonance biosensors, Curr. Opin. Bio-technol. 8 (1997) 50–57.

[22] J. Foote, H.N. Eisen, Kinetic and affinity limits on antibodiesproduced during immune responses, Proc. Natl. Acad. Sci. USA92 (1995) 1254–1256.

[23] F.D. Batista, M.S. Neuberger, Affinity dependence of the B cellresponse to antigen: a threshold, a ceiling, and the importance ofoff-rate, Immunity 8 (1998) 751–759.

[24] J.S. Babcook, K.B. Leslie, O.A. Olsen, R.A. Salmon, J.W.Schrader, A novel strategy for generating monoclonal antibodiesfrom single, isolated lymphocytes producing antibodies ofdefined specificities, Proc. Natl. Acad. Sci. USA 93 (1996) 7843–7848.

[25] G.R. McLean, O.A. Olsen, I.N. Watt, P. Rathanaswami, K.B.Leslie, J.S. Babcook, J.W. Schrader, Recognition of humancytomegalovirus by human primary immunoglobulins identifiesan innate foundation to an adaptive immune response, J.Immunol. 174 (2005) 4768–4778.

[26] N. Ohmura, S.J. Lackie, H. Saiki, An immunoassay for smallanalytes with theoretical detection limits, Anal. Chem. 73 (2001)3392–3399.

[27] R.M. Jones, H. Yu, J.B. Delehanty, D.A. Blake, Monoclonalantibodies that recognize minimal differences in the three-dimen-sional structures of metal–chelate complexes, Bioconjug. Chem.13 (2002) 408–415.

[28] S.E. Jacob, M. Nassiri, F.A. Kerdel, V. Vincek, Simultaneousmeasurement of multiple Th1 and Th2 serum cytokines inpsoriasis and correlation with disease severity, MediatorsInflamm. 12 (2003) 309–313.

[29] R.E.Nocker,D.F. Schoonbrood, E.A. van deGraaf, C.E.Hack,R.Lutter, H.M. Jansen, T.A. Out, Interleukin-8 in airway inflamma-tion in patients with asthma and chronic obstructive pulmonarydisease, Int. Arch. Allergy Immunol. 109 (1996) 183–191.

[30] D. Biasi, A. Carletto, P. Caramaschi, P. Bellavite, T. Maleknia, C.Scambi, N. Favalli, L.M. Bambara, Neutrophil functions and IL-8 in psoriatic arthritis and in cutaneous psoriasis, Inflammation 22(1998) 533–543.

[31] E.A. Kabat, T.T. Wu, M. Redi-Miller, H.M. Perry, K.S.Gottesman, Sequences of Proteins of Immunological interest,fourth ed., National Institutes of Health, Bethesda, MD,1987.

[32] R. Schier, A. McCall, G.P. Adams, K.W. Marshall, H. Merritt,M. Yim, R.S. Crawford, L.M. Weiner, C. Marks, J.D. Marks,Isolation of picomolar affinity anti-c-erbB-2 single-chain Fv bymolecular evolution of the complementarity determining regionsin the center of the antibody binding site, J. Mol. Biol. 263 (1996)551–567.

[33] C.F. Barbas 3rd, and D.R. Burton, Selection and evolution ofhigh-affinity human anti-viral antibodies, Trends Biotechnol. 14(1996) 230–234.

[34] E.T. Boder, K.S. Midelfort, K.D. Wittrup, Directed evolutionof antibody fragments with monovalent femtomolar antigen-binding affinity, Proc. Natl. Acad. Sci. USA 97 (2000)10701–10705.

[35] N. Ohmura, Y. Tsukidate, H. Shinozaki, S.J. Lackie, H. Saiki,Combinational use of antibody affinities in an immunoassay forextension of dynamic range and detection of multiple analytes,Anal. Chem. 75 (2003) 104–110.

[36] D.G. Myszka, Improving biosensor analysis, J. Mol. Recognit. 12(1999) 279–284.

[37] J.F. Paolini, D. Willard, T. Consler, M. Luther, M.S. Krangel,The chemokines IL-8, monocyte chemoattractant protein-1, andI-309 are monomers at physiologically relevant concentrations, J.Immunol. 153 (1994) 2704–2717.

[38] S.D. Burrows, M.L. Doyle, K.P. Murphy, S.G. Franklin, J.R.White, I. Brooks, D.E. McNulty, M.O. Scott, J.R. Knutson, D.Porter, et al., Determination of the monomer-dimer equilibriumof interleukin-8 reveals it is a monomer at physiological concen-trations, Biochemistry 33 (1994) 12741–12745.

[39] S.H. Northrup, H.P. Erickson, Kinetics of protein-protein asso-ciation explained by Brownian dynamics computer simulation,Proc. Natl. Acad. Sci. USA 89 (1992) 3338–3342.


[40] C.S. Raman, R. Jemmerson, B.T. Nall, M.J. Allen, Diffusion-limited rates for monoclonal antibody binding to cytochrome c,Biochemistry 31 (1992) 10370–10379.

[41] C. Watts, H.W. Davidson, Endocytosis and recycling of specificantigen by human B cell lines, EMBO J. 7 (1988) 1937–1945.

[42] X.D. Yang, X.C. Jia, J.R. Corvalan, P. Wang, C.G. Davis, A.Jakobovits, Eradication of established tumors by a fully humanmonoclonal antibody to the epidermal growth factor receptorwithout concomitant chemotherapy, Cancer Res. 59 (1999)1236–1243.

[43] J.D. Marks, Deciphering antibody properties that lead to potentbotulinum neurotoxin neutralization, Mov. Disord. 19 (Suppl. 8)(2004) S101–S108.

[44] A. Cauerhff, F.A. Goldbaum, B.C. Braden, Structural mechanismfor affinity maturation of an anti-lysozyme antibody, Proc. Natl.Acad. Sci. USA 101 (2004) 3539–3544.

[45] G. Schreiber, A.R. Fersht, Rapid, electrostatically assisted asso-ciation of proteins, Nat. Struct. Biol. 3 (1996) 427–431.

[46] R.F. Balint, J.W. Larrick, Antibody engineering by parsimoniousmutagenesis, Gene 137 (1993) 109–118.

[47] J. Foote, C. Milstein, Conformational isomerism and thediversity of antibodies, Proc. Natl. Acad. Sci. USA 91 (1994)10370–10374.

[48] L.C. James, P. Roversi, D.S. Tawfik, Antibody multispecificitymediated by conformational diversity, Science 299 (2003) 1362–1367.

doi:10.1016/j.jmb.2008.06.051 J. Mol. Biol. (2008) 381, 1238–1252

Available online at www.sciencedirect.com

Picomolar Affinity Fibronectin Domains EngineeredUtilizing Loop Length Diversity, Recursive Mutagenesis,and Loop Shuffling

Benjamin J. Hackel1, Atul Kapila2 and K. Dane Wittrup1,3⁎
1Department of ChemicalEngineering, MassachusettsInstitute of Technology,Cambridge, MA 02139, USA2Department of Biology,Massachusetts Institute ofTechnology, Cambridge,MA 02139, USA3Department of BiologicalEngineering, MassachusettsInstitute of Technology,Cambridge, MA 02139, USA
Received 11 December 2007;received in revised form5 June 2008;accepted 19 June 2008Available online24 June 2008

*Corresponding author. DepartmenEngineering, Massachusetts InstitutCambridge, MA 02139, USA. [email protected] used: CDR, complem

region; DSC, differential scanning cafluorescence-activated cell sorting; Fndomain of human fibronectin; PBSAsaline with bovine serum albumin.

0022-2836/$ - see front matter © 2008 P

The 10th type III domain of human fibronectin (Fn3) has been validated as aneffective scaffold for molecular recognition. In the current work, it wasdesired to improve the robustness of selection of stable, high-affinity Fn3domains. A yeast surface display library of Fn3 was created in which threesolvent-exposed loops were diversified in terms of amino acid compositionand loop length. The library was screened by fluorescence-activated cellsorting to isolate binders to lysozyme. An affinity maturation scheme wasdeveloped to rapidly and broadly diversify populations of clones by randommutagenesis as well as homologous recombination-driven shuffling of mu-tagenized loops. The novel library and affinitymaturation scheme combinedto yield stable, monomeric Fn3 domains with 3 pM affinity for lysozyme. Asecondary affinity maturation identified a stable 1.1 pM binder, the highestaffinity yet reported for an Fn3 domain. In addition to extension of theaffinity limit for this scaffold, the results demonstrate the ability to achievehigh-affinity binding while preserving stability and the monomeric state.This library design and affinity maturation scheme is highly efficient, uti-lizing an initial diversity of 2×107 clones and screening only 1×108 mutants(totaled over all affinity maturation libraries). Analysis of intermediatepopulations revealed that loop length diversity, loop shuffling, and recursivemutagenesis of diverse populations are all critical components.

© 2008 Published by Elsevier Ltd.

Keywords: fibronectin type III domain; Fn3; protein engineering; loopshuffling; affinity maturation
Edited by F. Schmid
Introduction

The 10th type III domain of human fibronectin(Fn3) has been demonstrated as an effective scaffoldfor molecular recognition.1–9 It is a small (10 kDa),stable (Tm=84 °C)7 cysteine-free beta-sandwich pro-tein that maintains its native fold in reducing envi-ronments and can be produced at up to 50 mg/L inEscherichia coli.9 The protein fold resembles an immu-noglobulin variable domain and contains threesolvent-exposed loops, termed BC, DE, and FG, for

t of Chemicale of Technology,l address:

entarity-determininglorimetry; FACS,3, 10th type III, phosphate-buffered

ublished by Elsevier Ltd.

the β strands they connect, which are structurallyanalogous to antibody complementarity-determin-ing regions (CDRs). Modification of the amino acidsequence in these loops can impart novel bindingcapacity on the scaffold. Koide and colleagues havediversified the BC and FG loops and screenedlibraries by phage display to yield a 5 μM binder toubiquitin4 and a 250 nM binder to Src SH3 domain2;FG loop libraries yielded improved integrin binders8

and novel estrogen receptor binders,3 the latter ofwhich was identified by a yeast two-hybrid screen.Phage display and yeast surface displaywere used toscreen a three-loop library with tyrosine/serine di-versity to isolate binders of 5, 7, and 30 nM to yeastsmall ubiquitin-like modifier, human small ubiqui-tin-like modifier 4, and maltose-binding protein,respectively.5 mRNA display was used to screenthree-loop libraries to identify a 20 pM binder totumor necrosis factor α9 and a 340 pM binder tovascular endothelial growth factor receptor 2.7

Lipovsek et al. used yeast surface display to engineera 350 pM binder to lysozyme from a library with

http://dx.doi.org/10.1016/j.jmb.2008.06.051

1239Fibronectin Domains of 1 pM Affinity

diversified BC and FG loops.6 Engineered bindershave been used as intracellular3 and extracellularinhibitors8 as well as labeling reagents in flow cyto-metry8 andWestern blots2 and also have been immo-bilized for affinity purification1 and for use in proteinarrays.9

Although wild-type Fn3 is highly stable andmono-meric, previously reported high-affinity Fn3 clonesare oligomeric or unstable. Two clones with 300 pMand 1 nMaffinity for tumor necrosis factorα exhibitedmidpoints of proteolysis susceptibility at appro-ximately 30 and 42 °C, respectively.9 A clone with340 pM affinity for vascular endothelial growth factorreceptor 2 exhibited Tm values of 32–52 °C; a differentclone was more stable but only 40–50% monomericand was of moderate affinity (13 nM).7 Stabilityengineering yielded a stable,monomeric 2 nMbinder.The ability to engineer Fn3 domainswith high affinityand stability represents a critical need in scaffolddevelopment, since both properties are required forclinical and biotechnological applications.Synthesis errors during our previous library

creation resulted in rare clones with non-wild-typeloop lengths.6 These loop length variants werepreferentially selected during binder isolation, andbinding titrations revealed their critical importanceto binding.Moreover, sequence analysis reveals non-wild-type lengths in tumor necrosis factor α binders9

and carcinoembryonic antigen binders (B.J.H., un-published data). The rarity of these length variants inthe initial library resulted in very low coverage oftheoretical length variant sequence space. These re-sults suggest that more extensive sampling of lengthvariant sequences could improve the binding cap-abilities of the Fn3 scaffold. This hypothesis issupported by the fact that CDR length diversity isalready an acknowledged component of antibodyengineering,10,11 and it has been demonstrated thatantibody affinity maturation is improved with theinclusion of length diversity.12,13 Koide and collea-gues incorporated loop length diversity in a recentFn3 library and successfully obtained binders withlength variation.5 From a structural perspective,

Batori et al. demonstrated that the BC, DE, and FGloops can tolerate insertions of four glycine residueswhile maintaining a native fold.14 In addition, phy-logenetic analysis reveals significant length diversityin each of the three loops across different fibronectintype III domains inmultiple species (Fig. 1). Thus, wehypothesize that loop length diversity will be stablytolerated and improve the functional binding capabi-lities of the Fn3 scaffold and perhaps could improvethe stability–affinity relationship.Yet, loop length diversity increases theoretical

sequence space, which is already immense, increas-ing the demand for efficient protein engineeringmethods in terms of both clonal selection and se-quence diversification. Yeast surface display enablesquantitative selections using fluorescence-activatedcell sorting (FACS) and correspondingly fine affinitydiscrimination.15 The eukaryotic protein processingof yeast also improves the functional stringency ofselections.16 One potential limitation of yeast surfacedisplay is that cellular transformation efficiencyholds library size to 107–109 without excessive effort.In vitro display methods enable larger theoreticallibrary size that correlateswith selection of improvedbinders17,18; however, as has been shown for phagedisplay, nominal library size does not necessarilyequatewith functional diversity.16 The intrinsicmuta-genesis from the requisite PCR step in mRNA andribosome display is a key contributor to the success ofthese display technologies.19We hypothesized that anincrease in the frequency of mutagenesis duringdirected evolution will improve the efficiency ofcellular display methods by increasing the breadthof the sequence space search in the vicinity of manylead clones, rather than only a select few.Another important engineering component is the

manner in which selected sequences are diversifiedthroughout directed evolution. Successful techni-ques include DNA shuffling,20 CDR shuffling,21,22

CDRwalking,23 and error-prone PCRmutagenesis24

either toward the entire gene or toward the sus-pected paratopes. In the current work, we combineerror-prone PCR and an analog to CDR shuffling to

Fig. 1. Loop length variability offibronectin type III domains. Therelative frequency of loop lengths ofthe BC (white), DE (gray), and FG(black) loops of sixteen type IIIdomains of fibronectin, relative tothe 10th type III domain (10Fn3) ofhuman fibronectin, are shown.Values and error bars represent theaverage and standard deviation forfive species: human, cow, rat,mouse,and chicken.

1240 Fibronectin Domains of 1 pM Affinity

yield both mild and significant changes in sequencespace both directed at the expected paratope andthroughout the Fn3 domain.Here we demonstrate that loop length diversity

and a novel affinity maturation scheme enablerobust and efficient selection of stable, high-affinitybinders to lysozyme. The binders are characterizedin terms of binding, stability, and structure includ-ing detailed analysis of the molecular basis ofbinding of a picomolar affinity clone.

Results

Fn3 library construction

A library was created in which 8, 5, and 10 aminoacids of the BC, DE, and FG loops, respectively, werediversified both in amino acid length (Fig. 1) and incomposition (Fig. 2a). Amino acid composition wasrandomized using NNB degenerate codons to yieldall 20 amino acids with reduced stop codon fre-quency. Four different loop lengths were chosen foreach loop based partially on the loop lengths ob-served in fibronectin type III domains in multiplespecies (Fig. 1). The library of Fn3 genes was incor-porated into a yeast surface display system by homo-logous recombination with a vector incorporating anN-terminal Aga2 protein for display on the yeastsurface and a C-terminal c-myc epitope for detectionof full-length Fn3.26 Library transformation yielded6.5×107 yeast transformants. Sixteen (62%) of 26clones sequenced matched library design. Nine(35%) contained frameshifts and 1 (4%)was annealedimproperly or underwent unintentional homologousrecombination in yeast. NNB diversification of theloops yields stop codons in approximately 44% ofclones. Thus, 34% [16/26×(1−0.44)] of transformedcells should display full-length Fn3. This percentagewas verified by flow cytometry analysis (data notshown). The library contains approximately 2.3×107

(6.5×107×0.34) full-length Fn3 clones.

Population mutagenesis design

A negative side effect of loop length diversity is afurther increase in the vastness of possible proteinsequence space to 1034 possible amino acid sequencesfor the three loops. Thus, the 2.3×107 clones in theoriginal library grossly undersample sequence space,so an extensive diversification methodwas desired tobroadly search sequence space during affinitymaturation. Error-prone PCR directed solely at thesolvent-exposed loops was used to focus diversity onthe likely paratope. Additionally, to make substantialchanges in sequence, a loop shuffling approach wasdeveloped. Fn3 genes were constructed with con-served wild-type framework sequence and randomlyshuffled mutated loops from the pool of selectedclones. Yet, as effective clones are selected, such subs-tantial sequence modification is not desired. Thus,error-prone PCR without shuffling was also em-

ployed. This mutagenesis targeted the entire gene toalso enable selection of beneficial framework muta-tions. Both mutagenesis strategies were used inparallel at each mutagenesis step throughout affinitymaturation (Supplementary Fig. 1). Analysis ofselected clones after the fact indicates that both loopshuffling and whole-gene error-prone PCR contrib-uted to improved phenotypes.The Fn3 library was screened using yeast surface

display and FACS. Plasmids from thousands of leadclones in a partially screened yeast population werecollected by yeast zymoprep. Quantitative real-timePCR revealed that full population diversity wasrecovered (data not shown). Both error-prone PCRof the full gene and shuffling of mutagenized loopswere successfully employed to diversify the popula-tion of lead clones (Fig. 2b). Gene mutagenesis wasperformed by error-prone PCR with nucleotide ana-logs.27 Independent error-prone PCRs were conduc-ted for each of the three loops using primers thatoverlap either the adjacent loop primer or the plas-mid vector. This overlap enabled shuffled gene re-construction via homologous recombination duringyeast transformation (Fig. 2c). The number of diver-sified transformants during affinity maturationranged from 2 to 20 million (mean, 8.1 million) forfull gene mutagenesis and 0.1 to 6.5 million (mean,2.5 million) for loop mutagenesis. Sequence analysisexhibited mutagenesis that matched the desired 1 to5 amino acid mutations per gene (data not shown),which was achieved with a single combination ofnucleotide analog concentration and number of PCRcycles determined by mathematical modeling.

High-affinity binder engineering

The efficacy of the new library and affinity matu-ration scheme was tested by engineering binders tolysozyme, a model protein that we have previouslyused as an Fn3 target with different library andmaturation approaches.6

The naive library was sorted three times by FACSusing multivalent biotinylated lysozyme preloadedon streptavidin–fluorophore conjugates. The resul-tant population was diversified by both gene andloop mutagenesis. Transformed mutants as well asthe original clones were sorted twice by FACS, yield-ing enrichment of evident lysozyme-binding clones(Fig. 3). Four additional rounds of isolation andmaturation, each consisting of two to three FACSselections followed by mutagenesis (SupplementaryFig. 1), were conducted usingmonovalent lysozyme,ranging from 1 μM to 20 pM lysozyme. Labeling atlow picomolar concentrations while maintainingstoichiometric excess of target to Fn3 would requireimpractically large volumes; thus, kinetic competi-tion sorting was used for the final three rounds ofisolation and maturation. The final two sublibrarieswere sorted an additional three times and the collect-ed populations were sequenced to identify the high-est affinity clones.Three dominant clones were identified by sequence

analysis (Table 1). Affinity titrations indicate that all

Fig. 2. Library design and affinity maturation scheme. (a) The naive library is randomized in the BC, DE, and FG loops (structure schematic derived from Main et al.25) at theindicated positions to four possible loop lengths each. (b) Error-prone PCR is used to introduce random mutations either into the entire Fn3 gene or into the three loops (separatePCRs). Arrowed lines indicate PCR primers; matching shades correspond to primers in a single PCR reaction. The framework (fw) and loop regions of the gene are indicated.Zigzag lines represent nucleotide mutations. Smaller unlabeled images are used to emphasize that PCR templates and products are part of a repertoire based on many genevariants. (c) Two yeast sublibraries are created by transformation via electroporation. During transformation, the full plasmid is created by homologous recombination of thelinearized vector with either the single mutated gene insert or the three mutated loop inserts.

1241Fibronectin

Dom

ainsof

1pM

Affinity

Fig. 3. Binder isolation and affinity maturation. Yeast libraries displaying Fn3 were labeled with mouse anti-c-mycantibody followed by goat anti-mouse fluorophore as well as biotinylated lysozyme and streptavidin–fluorophore andanalyzed by flow cytometry. (a) Yeast population during second round of isolation and maturation labeled with 50 nMmultivalent lysozyme preloaded in a 3:1 stoichiometry on streptavidin–R-phycoerythrin. (b) Yeast population duringsixth round of isolation and maturation labeled with 0.5 nM monovalent lysozyme followed by streptavidin–AlexaFluor488. (c) Yeast population during eighth round of isolation and maturation labeled with 2 nM monovalentlysozyme for 15 min followed by 35 h of dissociation and labeling with streptavidin–R-phycoerythrin. Polygonal regionsrepresent sort gates for cell selection.


three clones have comparable equilibrium dissocia-tion constants (Kd) for binding to biotinylatedlysozyme: 2.6±0.6 pM (clone L7.5.1), 2.9±1.4 pM(L8.5.2), and 2.8±0.5 pM (L8.5.7) (Fig. 4a and Table 1).The high-affinity binding was validated using pur-ified Fn3 domains in an equilibrium competitiontitration,which indicated an affinity of 6.9±0.3 pM forL7.5.1 (Fig. 4b). Dissociation rates for the three clonesrange from (2.5–5.4)×10−6 s−1, which correspond todissociation half times of 36–78 h (Fig. 4c). Theassociation rate of L7.5.1 was measured experimen-tally as (2.0±0.5)×106 M−1 s−1 (Fig. 4d). The cal-culated value of koff/kon is 1.2±0.3 pM, which isreasonably consistent with the experimentally mea-sured Kd of 2.6±0.6 pM. In addition to high affinity,the clones exhibit target specificity, as they do notshow appreciable binding to an array of other mole-cules (Fig. 5).Since previously reported high-affinity clones were

unstable or oligomeric, it was desired to examine thebehavior of the current Fn3 domains. The cloneswere

Table 1. Characterization of wild-type, L7.5.1, L8.5.2, and L8

Clone

Amino acid sequence

BC loop DE loop FG loop Framework

WT DAPAVTR GSKST GRGDSPASSKL7.5.1 RGYPWAT GVTN RVGRTFDTPG P15S, R33G, T35I

P44L, V50ML8.5.2 RGCPWAI GVTN RVGRMLCAPG R33G, I34V, N42S

P44L, V45A, V50MK63E, K98R

L8.5.7 RDRPWAI GVTN RLSIVPYA D3G, L18I, R33GN42S, P44L, V50MY73H, N91T, S100

Kd, equilibrium dissociation constant at 25 °C; koff, dissociation rate comonomer in analytical size-exclusion chromatography; Tm, midpoint omidpoint of thermal denaturation curve as determined by yeast surfa

produced in bacterial culture and purified forbiophysical characterization. Analytical size-exclu-sion chromatography indicated that all three clonesare predominantly monomeric with only L8.5.2present in a significant oligomeric state (80% mono-meric; Table 1 and Supplementary Fig. 2a). It shouldbe noted that L8.5.2 contains two cysteine residuesand intermolecular disulfide bonding contributes tooligomerization as indicated by nonreducing SDS-PAGE (data not shown). The thermal stabilities of theproteins were analyzed by differential scanning calo-rimetry (DSC) and circular dichroism of purifiedprotein as well as thermal denaturation of proteindisplayed on the yeast surface (Table 1 and Supple-mentary Fig. 3). L7.5.1 is the most stable clone withmidpoints of thermal denaturation ranging from 55.7to 58.8 °C for the three methods. It should be notedthat denaturation of L7.5.1 is not reversible becauseof aggregation at high temperatures. Both otherclones are also stable with Tm values over 50 °C. Inaddition, far-UV circular dichroism analysis reveals

.5.7

Kd(pM)

koff(10−5 s−1)

Monomer(%)

Tm (°C)T1/2(°C)DSC CD

N107 n/d 99 85.7 84.2 n/d, 2.6±0.6 0.25±0.02 99 58.1 58.8±1.6 55.7±0.1

,,

2.9±1.4 0.32±0.01 80 n/d 52.5±0.2 50.6±0.5

,,P

2.8±0.5 0.54±0.06 93 54.5 n/d 53.0±0.4

nstant at 25 °C; Monomer, percent of purified protein present asf thermal denaturation curve as determined by DSC or CD; T1/2,ce display residual activity assay.

Fig. 4. Determination of binding parameters for high-affinity clones. L7.5.1, L8.5.2, or L8.5.7 was displayed on theyeast surface and assayed for binding to biotinylated lysozyme at 25 °C. Different symbols indicate replicate experiments.Continuous lines indicate theoretical values. (a) Fraction of displayed L7.5.1 binding to biotinylated lysozyme atequilibrium. Only L7.5.1 is shown for simplicity. (b) Relative binding of 20 pM biotinylated lysozyme to displayed L7.5.1at equilibrium in the presence of indicated amount of soluble L7.5.1 competitor. ∞ indicates samples without biotinylatedlysozyme. (c) Fraction of displayed L7.5.1 (squares), L8.5.2 (diamonds), or L8.5.7 (triangles) binding to biotinylatedlysozyme after dissociation at 25 °C for the indicated time. (d) Fraction of displayed L7.5.1 binding to biotinylatedlysozyme after association at 25 °C; biotinylated lysozyme was present at 49 pM (crosses), 96 M (triangles), 185 pM(diamonds), or 345 pM (squares). Results are presented from a single experiment, which is representative of triplicateexperiments performed at different sets of concentrations.


no significant differences in secondary structurebetween wild-type Fn3 and L7.5.1, indicating thatdespite loopmutation and length variation the struc-tural integrity of the domain is maintained (Supple-mentary Fig. 2b).

units from binding (PE signal) normalized by the amount of Ferror bars represent mean±standard deviation of duplicate me

Analysis of intermediate populations

Intermediate populationswere investigated to elu-cidate the progress of affinity maturation. Each sub-library (i.e., library of mutagenized clones created

Fig. 5. Binding specificity. Wild-type Fn3 (wt), L7.5.1, L8.5.2, or L8.5.7was displayed on the yeast surface,washed, and incubated with 100 pMbiotinylated lysozyme (black) or100 nM nontarget molecules: bioti-nylated carcinoembryonic antigen(bCEA, dark gray), biotinylated epi-dermal growth factor receptordomain IV (bEGFR, light gray), bio-tinylated huntingtin peptide (bHtt,white), or streptavidin–R-phycoery-thrin (strep, striped). Binding wasdetected with streptavidin–R-phy-coerythrin secondary labeling andflow cytometry. The data are pre-sented as the mean fluorescence

n3 on the yeast surface (AlexaFluor488 signal). Values andasurements. Asterisk (*) indicates a value less than 0.005.

Table 2. Highest affinity Fn3 domains in each sublibrary

Round

Amino acid sequenceKd

(pM)koff

(10−5 s−1) t1/2 (h)BC loop DE loop FG loop Framework

0 RDCPWAT WTPVCF SSQRGCM none ≫100,000 n/d n/d1 SLDNQAN GQSD RCEPSRNSAV none N100,000 n/d n/d2 same clone as round 13 SLDNQAN GVTN RVGRMLDTPG P44S, V50M 7600±1100 460 0.044 SLDNQAK GATN RCKPFRNSAV P44S, V50M, T97I 330±50 n/d n/d5 RDCPWAI GVTN RVGRMSCTSG V1A, S1P, T14A, R33G,

P44L, V50M16±6 4.5±0.3 4.2±0.3

6 RGCPWAI GVTN RVGRMLCTPG P15S, R33G, N42S,P44L, V50M, K63E

6.6±1.3 0.72±0.07 27±3

7 RGYPWAT GVTN RVGRTFDTPG P15S, R33G, T35I, P44L, V50M 2.6±0.6 0.25±0.02 78±18a RGCPWAI GVTN RVGRMLCAPG R33G, I34V, N42S, P44L,

V45A, V50M, K63E2.9±1.4 0.32±0.01 60±1

RDRPWAI GVTN RLSIVPYA D3G, L18I, R33G, N42S,P44L, V50M, Y73H, N91T

2.8±0.5 0.54±0.06 36±4

Round, number of mutagenesis cycles (round 0 indicates naive library); Kd, equilibrium dissociation constant at 25 °C; Koff, dissociationrate constant at 25 °C; t1/2, time for 50% dissociation of the Fn3–lysozyme complex at 25 °C; n/d, not determined.

a Two clones with similar affinities were identified from round 8.


during affinity maturation) was sorted by FACSwithout mutagenesis to identify the highest affinityclone at each stage of affinity maturation. The high-est affinity clone was identified as the dominantclone from sequencing several clones from the exten-sively sorted sublibraries. Sequences, equilibriumdissociation constants and dissociation rate con-stantswere determined (Table 2). The highest affinitybinder in the original library exhibits apparent mid-micromolar affinity. However, affinity maturationprogressively improves binding performance toyield nanomolar and then picomolar binders (Fig.6). It is noteworthy that affinitymaturation exhibits arelatively consistent correlation to the number ofclones analyzed throughout the course of affinitymaturation until the final round of directed evolu-tion. A steady increase characteristic of a consistentlyprogressing affinity maturation scheme is observedrather than one single exceptional increase in affinity,indicative of a fortuitous mutation or recombination.

The impact of loop shuffling is evident in multiplecases. The BC loop from the round 1 clone is recom-bined with new DE and FG loops in round 3 to yieldan 8 nM binder. A mutated version of the FG looppresent in round 1 is then recombined with mutatedBC and DE loops from round 3 to achieve picomolarbinding. The highest affinity clones in rounds 5–8result from apparent shuffling of the BC loop ob-served in round 0 and the DE and FG loops observedin round 3 as well as multiple framework mutations.The appearance of point mutations is also apparent,both within the loops and in the framework region.The impact of these frameworkmutationswas inves-tigated in more detail in the context of L7.5.1.

L7.5.1 analysis

L7.5.1 was selected for more thorough analysis be-cause it has the fewest framework mutations and isthe most stable of the high-affinity clones. The engi-

Fig. 6. Affinity maturation pro-gress. The highest affinity clone ineach affinity maturation sublibrarywas identified by FACS. The equili-brium dissociation constant of thefirst three clones could not be deter-mined accurately by yeast surfacedisplay titration because of the rela-tively low affinity. Thus, an estimatedrange of possible affinities consistentwith equilibrium labeling at highnanomolar concentrations is indi-cated. The equilibrium dissociationconstant of the latter six clones wasdetermined by titration and is repre-sented as the mean±one standarddeviation of replicate measurements.The number of total clones analyzedis the cumulative total of the numberof full-length clones in the naivelibrary and total yeast transformantsin all affinity maturation sublibraries.

Table 3. Affinity and stability of framework mutationreversions of L7.5.1

AA WT L7.5.1 Kd (pM) T1/2 (°C) Location

L7.5.1-parent clone 2.6±0.6 55.7±0.115 Pro Ser 3.4±0.1 59.5±0.7 AB loop33 Arg Gly 74±23 56.2±0.6 C strand35 Thr lle 5.8±1.6 54.5±0.5 C strand44 Pro Leu 14.3±4.0 50.3±1.2 CD loop50 Val Met 25±16 53.7±1.8 D strand

The five framework mutations of L7.5.1 were individually re-verted to wild-type amino acids. Binding affinity and thermalstability were determined by yeast surface display. AA, aminoacid position; WT, wild-type side chain; L7.5.1 clone, 7.5.1 sidechain; Kd, equilibrium dissociation constant at 25 °C; T1/2,midpoint of thermal denaturation as determined by yeast surfacedisplay.


neered elements of the clone were analyzed for theirimpact on affinity and stability. Each randomizedloop and framework mutation was independentlyrestored to wild-type sequence and both affinity andstability were measured using yeast surface display.Reversion of the FG loop had the strongest effect onbinding, aswild-type restoration decreased affinity tothe micromolar level; conversely, DE loop reversionhad a nearly negligible effect on binding, and the BCloop reversion had an intermediate effect with a 4700-fold reduction (Fig. 7a). Both BC and DE loop rever-sion significantly destabilize the domain (Fig. 7b).Although these loop sequences confer high thermalstability in the wild-type context, modification ofadjacent loops apparently adjusts the intramolecularcontacts. Thus, during multiloop diversification,selected sequences must provide not only properintermolecular contacts for high-affinity binding butalso proper intramolecular contacts for structuralintegrity and stability. The stability of the FG loopreversion could not be accurately determined by thismethod because of its weak binding.Three of the five framework mutations in the se-

lected clone (T35I, P44L, and V50M) are relatively

Fig. 7. L7.5.1 loop analysis. Each engineered loop ofL7.5.1 was independently restored to wild-type sequenceand the resultant clonesweredisplayed on the yeast surface.Filled and empty symbols indicate replicate experiments. (a)Binding to biotinylated lysozyme at 25 °Cwas quantified byflow cytometry. Clone and equilibrium dissociation con-stant: L7.5.1 (squares), 2.6±0.6 pM; L7.5.1wtBC (diamonds),12.4 ± 0.7 nM; L7.5.1wtDE (triangles), 3.7 ± 1.4 pM;L7.5.1wtFG (crosses), 1.6±0.2 μM. (b) Yeast displaying Fn3were incubated at the indicated temperature for 30 minfollowed by a flow cytometric assay for biotinylatedlysozyme binding. Clone and midpoint of thermal dena-turation: L7.5.1 (squares), 55.7±0.1 °C; L7.5.1wtBC (dia-monds), 37.4±1.5 °C; L7.5.1wtDE (triangles), 44.5±0.1 °C.

conservative mutations and were beneficial to bothaffinity and stability (Table 3). Conversely, R33Greplaces a large, positively charged side chain with asingle hydrogen; this mutation provides 28-foldstronger binding without a substantial effect on sta-bility. Perhaps the arginine side chain was not accom-modated sterically and/or electrostatically at theFn3–lysozyme interface or the glycine enables a bene-ficial backbone conformation. The P15Smutation hadonly a very minor affinity improvement but substan-tially destabilized the domain. The lack of impact onbinding is reasonable given its distance from theexpected paratope, and the substantial destabilizationis not surprising given the removal of the backbone-constraining proline in the turn between the C and Dstrands. Selection of this mutation was likely coin-cident with another beneficial mutation, as it does notimpart significant benefit; it should be noted thatwhile the most beneficial mutations, R33G, P44L, andV50M, are consistently observed throughout matura-tion (Table 2), P15S and T35I are rare. Overall, thesedata suggest the paratope is focused on the BCandFGloops as well as potential key contacts on the C and Dstrands of the β sheet.

Focused mutagenesis

The conservation of the DE loop sequence through-out much of the affinity maturation (Table 2) despiteits relatively low impact on affinity relative to wildtype prompted further study of this loop. The func-tionally tolerable diversity and the potential forimproved binding were explored through re-rando-mization of the loop sequence. Within the context ofL7.5.1 S15P, the loop was randomized using NNBnucleotide diversity and the same four loop lengthoptions as the original library. Labeling the librarywith 30 pM lysozyme, which yields 90% binding tothe parent clone L7.5.1 S15P, yields many clones thatprovide effective binding, indicating that the loopcan tolerate significant diversity and maintain bind-ing. Specifically, 7% of clones exhibit binding at 75%of maximum, which is characteristic of a 10 pM Kd,or within threefold of the parent clone (Fig. 8). Yet,30% of the full-length clones bind at less than 1% ofmaximum, indicating that many loop sequences can

Fig. 8. L7.5.1 S15P DE loop randomization librarybinding analysis. The DE loop of L7.5.1 S15P was ran-domized using NNB codon diversity with four looplengths. The yeast surface display library was labeled with30 pM biotinylated lysozyme and mouse anti-c-mycantibody followed by streptavidin–R-phycoerythrin andgoat anti-mouse antibody conjugated to AlexaFluor488.Percentages indicate the relative number of c-myc positiveclones with bLys binding: c-myc display signals with 75%of maximum (7%), 1–75% of maximum (63%), or b1% ofmaximum (30%).


greatly hinder binding either through direct interac-tion with the binding partner or through structuralmodification of the Fn3 paratope. In addition, thelibrary was sorted for binding to biotinylated lyso-zyme using FACS. Sequence analysis after four selec-tions identified a single clone, DE0.4.1, with GDLSHRreplacing GVTN in the DE loop. Binding and stabilityanalyses indicate a 3.1× improvement in binding to1.1 pM at the expense of an 8.5 °C decrease in the T1/2(Table 4). The maintenance of glycine at position 52from wild type to L7.5.1 to DE0.4.1 is noteworthy forpossible future library design especially consideringthe adjacent amino acid is proline; the proline–glycinepair provides an effective turn motif to begin the DEloop.Lack of improvement from round 7 to round 8 in

conjunction with sequence similarities during affi-nity maturation prompted investigation into a com-plementary method of affinity maturation. Sinceonly a fraction of protein sequence space is accessiblethrough single nucleotide mutations, a library wasconstructed in which the DNA encoding noncon-served amino acids in the BC and FG loops wasrandomized by degenerate oligonucleotides. Non-

Table 4. Affinity and stability of clones from focused mutage

Clone

Amino acid sequen

BC loop DE loop FG loop

L7.5.1 S15P RGYPWAT GVTN RVGRTFDTPGCons 0.4.1 REDPWAK GVTN RVGWASYTLGDE 0.4.1 RGYPWAT GDLSHR RVGRTFDTPG

Kd, equilibrium dissociation at 25 °C; T1/2, midpoint of thermal denactivity assay.

conserved amino acids were identified as thepositions of L7.5.1-homologous clones at which se-quence diversity was observed in any sequence du-ring affinity maturation or sublibrary analysis. Thenonconserved amino acids were randomized in twomodes: either to all 20 amino acids using NNB co-dons or to amino acids observed during sequenceanalysis using tailored codons at each position. Thelibrary was sorted for binding to biotinylated lyso-zyme using FACS. Sequence analysis after fourselections yielded a single clone, Cons0.4.1. Theaffinity of Cons0.4.1 wasmeasured as 1.1±0.6 pM byyeast surface display titration and 0.33±0.15 pM byequilibrium competition with purified Fn3 domain.The midpoint of thermal denaturation was mea-sured as 52.5±2.1 °C by circular dichroism analysisand 58.8±3.4 °C by yeast surface display thermalresistance assay. Far-UV circular dichroism indicatesthat Cons0.4.1 maintains a secondary structure simi-lar to that of the wild-type Fn3 domain (Supplemen-tary Fig. 2b). Two of the eight codon changes werenot possible with a single nucleotide mutationmaking this clone unlikely to be reached by error-prone PCR. Thus, while error-prone PCR provides ahighly effective means of diversification, an im-provement was observed through more thoroughsearching of a focused region of sequence space oncea consistent binding motif was identified.

Discussion

In this work we explore the impact of multiplecomponents of engineering a high-affinity bindingsite in the Fn3 scaffold. The combination of looplength diversity and recursive mutagenesis includ-ing mutated loop shuffling enabled selection of thehighest affinity Fn3 domains yet reported despite arelatively modest initial library size of 2.3×107 full-length Fn3 clones. The results extend the affinityattainable by this single domain scaffold, furthervalidating its use as amolecular recognition scaffold.In addition, insights were gained for both Fn3 designand engineering and protein engineering in general.Analysis of the highest affinity binders in each

sublibrary (Table 2) demonstrates deviation fromwild-type loop lengths in all three loops implicatingthe importance of this diversity element in Fn3library design. Interestingly, the observed BC loopsare all one amino acid shorter than wild type, whichis observed both in other fibronectin type III domains

nesis

ceKd

(pM)T1/2(°C)Framework

R33G, T35I, PAAL, V50M 3.4±0.1 59.5±0.7R33G, T35I, P44L, V50M 1.1±0.6 59±3R33G, T35I, P44L, V50M 1.1±0.5 51±3

aturation curve as determined by yeast surface display residual


(Fig. 1) and in a previously engineered binder.6 DEloops are observed with either one less or one moreamino acid than wild type. Clone DE0.4.1 and thehighest affinity clone from the naive library (Table 2)justify inclusion of a six-amino acid DE loop despiteits lack of existence in native fibronectin type IIIdomains. FG loops of wild-type length as well asboth two- and three-amino acid reductions wereobserved. The observed length diversity, as well asthe success of length diversity in a restricted diver-sity Fn3 library published during preparation of thismanuscript,5 support the inclusion of length diver-sity in all future Fn3 engineering. Moreover, thecombination of phylogenetic analysis and sequenceanalysis of engineered binders should continue toelucidate the relative preference of each length tofurther improve library design.Loop length diversity increases sequence space to

1034 possible amino acid sequences for the threeloops, although only 2.3×107 clones are present inthe yeast library. The two modes of diversity intro-duced during recursive mutagenesis effectivelysearch this large sequence space despite the sparsesampling. Sequence and binding analyses indicatethat each of the elements of the mutagenesis methodwere beneficial. The loop mutagenesis path of thediversification was effective, as both loop shufflingand loop-focused point mutations are evident (Table2). In addition, the gene mutagenesis path wasadvantageous, as an effective loop combination wasmaintained in rounds 5–8 and beneficial frameworkmutationswere introduced throughout (Tables 2 and3). The importance of diversifying many clones du-ring each round of maturation is evident, sincehomologs of the eventual DE and FG loops were notpresent in 30 sequences from the enriched popula-tions from the naive library. As a result of thesecombined components, affinity maturation rapidlyand efficiently progressed to yield the clones with3 pM affinity without rational intervention. The affi-nity maturation procedure is straightforward andsimple; plasmid recovery,mutagenesis, amplification,and yeast transformation can be achieved in 1–2 days.The increased frequency of mutagenesis is obviouslyapplicable to any protein engineering method and isstrongly recommended to improve both the speed ofbinder isolation and the overall efficacy. Shuffling canbe implemented in any analogous protein scaffoldprovided the regions of interest are relatively prox-imal at the DNA level to enable overlap for homolo-gous recombination. Yeast surface display providesan effective system for shuffling because DNA frag-ments can be recombined with high fidelity duringcellular transformation simplifying the method.The sequence diversity of the highest-affinity

clones is striking. L8.5.2 has two cysteines at loca-tions consistent with formation of an interloop disul-fide bond, as was found previously from a differentFn3 library screened by yeast display.6 However,unlike in that case, the disulfide is dispensable forhigh-affinity binding, since L7.5.1 has highly relatedloop sequences but lacks both cysteines. A thirdclone with similar affinity (L8.5.7, 2.8 pM)was found

with a substantially different FG loop, but conservedDE and BC loops.It is noteworthy that the focused mutagenesis

results indicate the potential for mild improvementof affinity maturation through targeted randomiza-tion atmultiple residues identified as variable throughsequence analysis. It is not yet clear if the success ofthis avenue of affinity maturation resulted from therelatively high number of mutations from the parentclone (eight amino acid changes in the two loops) orthe ability to reach amino acids that would requiremore than one nucleotide mutation. Regardless, thefact that a secondary, semirational approach was onlyable to yield 3.1× enhancement in affinity is indicativeof a relatively effective search of sequence space by theinitial affinity maturation.Diversification of the DE loop is an important

aspect of Fn3 engineering. Because it is the shortestand least flexible wild-type loop, it is unclear if thebinding potential gained by diversification offsets thepossible structural destabilization. Thus, analysis ofthe contribution of the DE loop to binding and sta-bility are valuable to clone maturation and future Fn3library design. The originally selected DE loop,GVTN, stabilizes clone L7.5.1 relative to wild typebut does not significantly aid binding. Although loopreversion analysis suggests that the binding paratopeis dominated by the FG andBC loops, theDE loop canbe engineered for improved binding (DE0.4.1) albeitat the expense of stability. Since selections wereexplicit for affinity, the lack of binding performancefrom the L7.5.1 DE loop likely resulted from a lack ofDE loop diversity after a few rounds of directed evo-lution (Table 2). Thus, a future improvement on loopshuffling will be to incorporate a small percentage ofnaive loops with the engineered loops. A broaderuncertainty regarding the DE loop is the extent towhich it should be diversified in future Fn3 libraries.Previous binders have been engineered with wild-type DE loops with 350 pM affinity for lysozyme,6

30 nM for maltose-binding protein,5 250 nM for SrcSH3domain,2 and approximately 5μMforubiquitin.4

The DE reversion of L7.5.1 represents a 3.7 pM binderwith a wild-type DE loop. Collectively, it is clear thatbinders can be engineered without DE loop diversity.Conversely, engineeredDE loops in the L7.5.1 contextcan either improve the affinity to 1.1 pM or stabilizethe molecule by an 11 °C increase in T1/2, the latter ofwhich is not surprising given its extensive contactwith the BC loop residues as well as the shortened BCloop length. Moreover, engineered vascular endothe-lial growth factor receptor 2 binders require theirengineered DE loop for binding, although it destabi-lizes themolecule by 2.0 kcal/mol or∼30 °C decreasein Tm.

7 Consequently, as expected, DE loop impact iscontext dependent. One possible general approachwould be to introduce mild DE diversity in the ori-ginal library to allow for binding constraint removaland structural complementation of selected BC loopsaswell as the possibility of beneficial binding contacts.After binder selection, affinity or stability maturationcould be employed with more diverse DE loopshuffling in the context of effective BC and FG loops.


Unlike several examples of previous high-affinityFn3 binder engineering, the selected high-affinityclones are relatively stable andmonomeric. L7.5.1 is a2.6 pM binder and N99%monomeric with amidpointof thermal denaturation of 56–59 °C. Moreover,reversion of serine to proline at position 15 improvesthe T1/2 to 60 °C. In addition, Cons0.4.1 is a 1.1 pMbinderwith Tm of 53 °C, and L8.5.7 is a 2.8 pMbinder,is 93% monomeric, and has a Tm of 53–55 °C.Although stability was not an explicit element ofselection, the eukaryotic secretion machinery of yeastprovides some level of quality control against mis-folded proteins.28,29 In general, Fn3 clones of higherstability are displayed at high densities on the yeastcell surface (data not shown); thus, unstable clones areslightly selected against based on the two-color sortregions. It remains to be seen if stable clones willgenerally result from selections by yeast surfacedisplay or if this was a fortuitous result; regardless,yeast surface display provides a means for stabilityengineering either during or after binder selection.Overall, the method can be further improved

through refinement of naive loop length distribution,an increase in magnitude of the initial library,inclusion of naive loops during loop shuffling, andperhaps more expansive mutation than attainable byerror-prone PCR with nucleotide analogs. Never-theless, the combination of yeast surface display, looplength diversity, and dual-mode affinity maturationrapidly yielded stable, high-affinity binders. Themethod should be valuable toward development ofFn3 binders to additional targets as well as transfer-able to engineering of other proteins for any screen-able functionality.

Materials and Methods

Fn3 library construction

Oligonucleotides were purchased from MWG Biotech(High Point, NC) and IntegratedDNATechnologies (Coral-ville, IA) (see Supplementary Data online for sequences).The Fn3 library was constructed to produce wild-typesequence in the framework regions and to randomize theBC,DE, and FG loops of Fn3. TheDNAencoding for aminoacids 23–30 (DAPAVTVR) was replaced by (NNB)x wherex=6, 7, 8, or 9 to yield a loop length that is −2, −1, 0, or +1amino acids relative to wild type. Similarly, the DNA foramino acids 52–56 (GSKST)was replaced by (NNB)ywherey=4, 5, 6, or 7, and the DNA for amino acids 77–86(GRGDSPASSK) was replaced by (NNB)z where z=5, 6, 8,or 10.The library was constructed by sequential annealing and

extension of eight overlapping oligonucleotides.6 Thefollowing components were combined in a 50-μL reaction:two oligonucleotides (0.2 μM a2, b3nmix, c6t, or d7nmix+0.4 μM a1, b4n, c5nmix, or d8n, respectively), 1× polymerasebuffer, 0.2mMdeoxynucleotide triphosphate (dNTP), 1mMMgSO4, 1 U KOD Hot Start DNA Polymerase (Novagen,Madison, WI), 1 M betaine, and 3% dimethyl sulfoxide. Themixture was denatured at 95 °C for 2 min followed by 10cycles of 94 °C for 30 s, 58 °C for 30 s, and 68° for 1min and afinal extension of 68 °C for 10 min. Forty microliters of theproducts (a1+a2, b3nmix+b4n, c5nmix+c6t, or d7nmix+d8n)

were combined and thermally cycled at identical conditions.Two microliters of this product was combined with 0.4 μMprimer (a1–b4n amplified with p1; c5nmix–d8n amplifiedwith p8) in a new 100-μL reaction and thermally cycledunder identical conditions to amplify the appropriatestrand. The products were combined and thermally cycledat identical conditions. The final productswere concentratedwith PelletPaint (Novagen). The plasmid acceptor vectorpCTf1f4 (Ref. 6) was digested with NcoI, NdeI, and SmaI(New England Biolabs, Ipswich, MA). Multiple aliquots of∼10 μg of Fn3 gene and 3 μg plasmid vectorwere combinedwith 50–100 μL of electrocompetent EBY100 and electro-porated at 0.54 kVand 25 μF.Homologous recombination ofthe linearized vector and degenerate insert yielded intactplasmid. Cells were grown in YPD (1% yeast extract, 2%peptone, 2% glucose) for 1 h at 30 °C, 250 rpm. The numberof total transformants was 6.5×107 cells as determined byserial dilutions plated on SD-CAA plates (0.1 M sodiumphosphate, pH 6.0, 182 g/L sorbitol, 6.7 g/L yeast nitrogenbase, 5 g/L casamino acids, 20 g/L glucose). The librarywaspropagated by selective growth in SD-CAA, pH 5.3 (0.07 Msodium citrate, pH 5.3, 6.7 g/L yeast nitrogen base, 5 g/Lcasamino acids, 20 g/L glucose, 0.1 g/L kanamycin,100 kU/L penicillin, and 0.1 g/L streptomycin) at 30 °C,250 rpm.

Fluorescence-activated cell sorting

Yeast were grown in SD-CAA, pH 5.3, at 30 °C, 250 rpmto logarithmic phase, pelleted, and resuspended to 1×107

cells/mL in SG-CAA, pH 6.0 (0.1M sodium phosphate, pH6.0, 6.7 g/L yeast nitrogen base, 5 g/L casamino acids,19 g/L dextrose, 1 g/L glucose, 0.1 g/L kanamycin,100 kU/L penicillin, and 0.1 g/L streptomycin) to induceprotein expression. Induced cells were grown at 30 °C,250 rpm for 12–24h.Round 0 (three FACS selections) and round 1 (two FACS

selections) were conducted with multivalent lysozymeprepared by incubating streptavidin–fluorophore (R-phy-coerythrin, AlexaFluor488, or AlexaFluor633; Invitrogen,Carlsbad, CA) with biotinylated lysozyme (Sigma, St.Louis, MO) in a 1:3 ratio in phosphate-buffered saline withbovine serum albumin (PBSA). Yeast were pelleted,washed in 1 mL PBSA (0.01 M sodium phosphate, pH7.4, 0.137 M sodium chloride, 1 g/L bovine serumalbumin), resuspended in PBSA with 10–40 mg/L mouseanti-c-myc antibody (clone 9E10, Covance), and incubatedon ice. Cells were washed with 1 mL PBSA and resus-pended in PBSAwithmultivalent lysozyme (0.5 μM for thefirst four sorts and 50 nM for the fifth sort) and goat anti-mouse antibody conjugated to R-phycoerythrin, Alexa-Fluor488, or AlexaFluor633.Intermediate FACS selections were conducted with

near-equilibrium labeling with monovalent lysozyme.Three, two, two, and three selections were performed inrounds 2–5, respectively. Yeast were pelleted, washed in1 mL PBSA, resuspended in PBSA with biotinylatedlysozyme (ranging from 1 μM to 20 pM) andmouse anti-c-myc antibody, and incubated on ice. Cells were thenwashed with 1 mL PBSA and resuspended in PBSA withstreptavidin–fluorophore and fluorophore-conjugatedgoat anti-mouse antibody.FACS selections of very high affinity populations were

conducted with kinetic competition. Two, three, and twoselections were performed in rounds 6–8. Yeast werewashed and incubated briefly with 1–2 nM biotinylatedlysozyme. Yeast were then washed and resuspended withPBSA with 140 nM unbiotinylated lysozyme (to preventfurther association of labeled target) and incubated at


room temperature for 2 h to 7 days to enable dissociation ofbiotinylated lysozyme. Cells were washed in PBSA,resuspended in PBSA with mouse anti-c-myc antibody,and incubated on ice. Cells were washed and labeled withsecondary reagents as in equilibrium labeling.In all cases, labeled cells were washed with 1 mL PBSA,

resuspended in 0.5–2.0 mL PBSA and analyzed by flowcytometry using either a MoFlo (Cytomation) or Aria(Becton Dickinson) cytometer. c-myc+ cells with the top0.2–3% of lysozyme binding/c-myc display ratio wereselected. Collected cells were grown in SD-CAA, pH 5.3, at30 °C, 250 rpm and either induced in SG-CAA, pH 6.0, forfurther selection or used for plasmid recovery.

Fn3 mutagenesis

Plasmid DNA from 1×108 cells was isolated using twocolumns of Zymoprep kit II (Zymo Research, Orange, CA)according to the manufacturer's instructions except foradditional centrifugation of neutralized precipitate. Thezymoprep elution was cleaned using the Qiagen PCRPurification kit (Qiagen, Valencia, CA), and eluted in 40 μLof elution buffer. Error-prone PCR of the entire Fn3 genewas performed in a 50-μL reaction containing 1×Taq buffer,2 mMMgCl2, 0.5 μM each of primers W5 and W3, 0.2 mM(each) dNTPs, 5 μL of zymoprepped DNA template, 2 mM8-oxo-deoxyguanosine triphosphate (TriLink, San Diego,CA), 2 mM 2'-deoxy-p-nucleoside-5'-triphosphate (Tri-Link), and 2.5 U of Taq DNA polymerase (Invitrogen).In parallel, error-prone PCR of the loop regions wasperformed via three separate 50 μL reactions with20 mM 8-oxo-deoxyguanosine triphosphate and 20 mM2'-deoxy-p-nucleoside-5'-triphosphate and primers BC5and BC3 for the BC loop, DE5 and DE3 for the DE loop,and FG5 and FG3 for the FG loop. The reaction mixtureswere denatured at 94 °C for 3 min, cycled 15 times at 94 °Cfor 45 s, 60 °C for 30 s, and 72 °C for 90 s, and finallyextended at 72 °C for 10 min. Multiple preliminarymutagenesis reactions of the wild-type plasmid wereconducted at different nucleotide analog concentrations.Sequence analysis and comparison to a theoreticalframework30 indicated the aforementioned conditionsyield one to five amino acid mutations per gene. The PCRproducts were purified by agarose gel electrophoresis andeach amplified in four 100-μL PCR reactions containing 1×Taq buffer, 2mMMgCl2, 1 μMof each primer, 0.2mM (each)dNTPs, 4 μL of error-prone PCR product (of 40 μL from gelextraction), and 2.5UofTaqDNApolymerase. The reactionswere thermally cycled at the same conditions except that 35cycleswereused.Reaction productswere concentratedwithPelletPaint (Novagen) and resuspended in 1 μL of water.Plasmid pCT-Fn3 was digested with PstI, BtgI, and

BamHI to create linearized vector pCT-Fn3-Gene with theentire Fn3 gene removed. Plasmid pCT-Fn3 was digestedwith BclI, BtgI, and PasI to create linearized vector pCT-Fn3-Loop with the wild-type gene removed from the BCloop through the FG loop.Electrocompetent EBY100 (100 μL) was combined with

0.5–2.0 μg of pCT-Fn3-Gene and the gene-based PCRproduct and electroporated at 0.54 kVand 25 μF in a 2 mMelectroporation cuvette. Similarly, 100 μL of electrocompe-tent EBY100 was combined with 0.5–2.0 μg of pCT-Fn3-Loop and the three loop-based PCR products and electro-porated. Homologous recombination of the linearizedvector and mutagenized insert(s) yielded intact plasmid.Cells were grown in YPD for 1 h at 30 °C, 250 rpm. Themedium was switched to SD-CAA to enable selectivepropagation of successful transformants via growth at30 °C, 250 rpm, for 24–48 h.

DNA sequencing

Plasmid DNA was isolated using the Zymoprep kit II,cleaned using the Qiagen PCR Purification kit, and trans-formed into DH5α (Invitrogen) or XL1-Blue E. coli (Stra-tagene, La Jolla, CA). Individual clones were grown,miniprepped, and sequenced using BigDye chemistry onan Applied Biosystems 3730.

Measurement of Kd, kon, and koff

The equilibrium dissociation constant for a clone wasdetermined essentially as described.26 Briefly, yeast contain-ing the plasmid for an Fn3 clone was grown and induced asfor FACS selection. Cells were washed in 1 mL PBSA andresuspended in PBSA containing biotinylated lysozyme inconcentrations generally spanning 4 orders of magnitudesurrounding the equilibrium dissociation constant. Thenumbers of cells and sample volumes were selected toensure excess lysozyme relative to Fn3. For clones of lowpicomolar affinity, this criterion necessitates very low celldensity, which makes cell collection by centrifugationprocedurally difficult. To obviate this difficulty, uninducedcells are added to the sample to enable effective cell pelletingduring centrifugationwith no effect on lysozyme binding ofthe Fn3-displaying induced cells. Cells were incubated at25 °C for sufficient time to ensure that the approach toequilibrium was at least 98% complete. Cells were thenpelleted, washed with 1 mL PBSA, and incubated in PBSAwith 10 mg/L streptavidin–R-phycoerythrin for 10–30 min.Cells were washed and resuspended with PBSA andanalyzed with an Epics XL flow cytometer. The minimumand maximum fluorescence and the Kd value weredetermined by minimizing the sum of squared errors.For determination of the dissociation constant, koff,

clonal cell cultures were grown, induced, and washed asabove. Cells were incubated in PBSA with a saturatingconcentration of biotinylated lysozyme at 25 °C. At varioustimes, an aliquot of cells was washed with PBSA withexcess unbiotinylated lysozyme, resuspended in PBSAwith excess unbiotinylated lysozyme, and incubated at25 °C. Simultaneously, all samples of differing dissociationtimes were washed with PBSA and incubated in 10 mg/Lstreptavidin–R-phycoerythrin for 10–30 min. Cells werewashed and resuspendedwith PBSAand analyzedwith anEpics XL flow cytometer. The minimum and maximumfluorescence and the koff value were determined byminimizing the sum of squared errors.For determination of the association constants, kon,

clonal cell cultures were grown, induced, and washed asabove. At various times, an aliquot of cells was resus-pended in biotinylated lysozyme and incubated at 25 °C.Simultaneously, all samples of differing association timeswere washed with PBSA with excess unbiotinylatedlysozyme and incubated in PBSA with 10 mg/L strepta-vidin–R-phycoerythrin for 10–30 min. Cells were washedand resuspended with PBSA and analyzed with an EpicsXL flow cytometer. The maximum fluorescence and konwere determined byminimizing the sum of squared errorsassuming a 1:1 binding model. The experimentallydetermined value of koff was used to determine theeffective association rate, kon[lysozyme]+koff.The equilibrium dissociation constant was also deter-

mined for the soluble forms of L7.5.1 and Cons0.4.1 byequilibrium competition titration. Varying concentration ofpurified Fn3 domains were incubated with 20 pM biotiny-lated lysozyme in 50 mL of PBSA. Yeast displaying L7.5.1were added and incubated for 7 days to near equilibrium.Cells were then pelleted, washed with 1 mL PBSA, and


incubated in PBSAwith 10 mg/L streptavidin–R-phycoer-ythrin for 15min. Cells werewashed and resuspendedwithPBSA and analyzed with an Epics XL flow cytometer. Atwo-state binding model was assumed and the minimumand maximum fluorescence and equilibrium dissociationconstant were determined by minimizing the sum ofsquared errors.

Fn3 production and biophysical characterization

Fn3 clones were produced as previously described.6

Briefly, BL21(DE3)pLysS E. coli (Invitrogen) containing thepET-24b-based Fn3 plasmid were grown in Luria–Bertanimedium with 50 μg/mL kanamycin and 34 μg/mLchloramphenicol at 37 °C, 250 rpm, to an A600 of 0.1–0.2and induced with 0.5 mM IPTG for 18–24 h at 30 °C,250 rpm. Cells were lysed by sonication and the insolublefraction was removed by centrifugation at 19,000g for40 min. His6-tagged Fn3 was purified from the solublefraction with TALON Superflow Metal Affinity Resin(Clontech), dialyzed against PBS, and concentrated to0.5 mL with an Amicon Ultra centrifugal filter (Millipore).The oligomeric state was analyzed by size-exclusion

chromatography on a Superdex 75 HR10/300 column(Amersham Pharmacia Biotech, Piscataway, NJ). Mono-merwas isolated for biophysical analysis. PBS standards ormonomeric protein in PBS was thermally denatured from25 to 95 °C at a rate of 1 °C/min in a differential scanningcalorimeter (VP-DSC, MicroCal). Irreversible aggregationof L7.5.1 occurs at high temperatures. The midpoint ofthermal denaturation for this clone is identified as thetemperature of maximum heat capacity. Samples weredialyzed in 10 mM sodium phosphate buffer, pH 7.0, anddiluted to 8–10 μM for far-UV circular dichroism analysis.Ellipticity was measured from 250 to 190 nm on an Aviv202 spectrometer (Aviv Biomedical, Lakewood, NJ) with aquartz cuvette with a 1-mm path length (New Era, Vine-land, NJ). Thermal denaturation was conducted by mea-suring ellipticity at 216 nm from 25 to 95 °C and calculatingTm from a standard two-state unfolding curve.

Thermal stabilities were also determined using a yeastsurface display thermal denaturation assay derived fromOrr et al.31 Fn3 was displayed on the yeast surface as formeasurement of kinetic and equilibrium binding con-stants. Cells were washed and resuspended with PBSA,incubated at 20–85 °C for 30 min, and incubated on ice for5 min. Biotinylated lysozyme was added at a saturatingconcentration (e.g., 20 nM for L7.5.1) andmouse anti-c-mycantibody was added at 40 mg/L and incubated on ice for20 min. Cells were washed and incubated in PBSA with10 mg/L streptavidin–R-phycoerythrin and 25 mg/LAlexaFluor488 conjugated goat anti-mouse antibody.Cells were washed and resuspended in PBSA and ana-lyzed on an Epics XL flow cytometer. The minimum andmaximum fluorescence (Fmin and Fmax, respectively), theT1/2, and the enthalpy of unfolding at T1/2 (ΔHm) weredetermined by minimizing the sum of squared errorsbetween experimental data and theoretical values accord-ing to a two-state unfolding equation:

F� Fmin

Fmax � Fmin¼ ffolded ¼ 1þ exp

DHm

R1

T1=2� 1T

� �� 1

L7.5.1 reversion clone construction

Reversion of engineered loops of L7.5.1 to wild-typesequence was accomplished by annealing and extending

two PCR products created with a gene-terminal primerand a primer that annealed adjacent to the loop of interestbut was extended to include wild-type sequence. Specifi-cally, one PCR reaction contained a 5′ gene terminal primerand a primer that annealed to the 25 nucleotides imme-diately 5′ of the loop of interest but included a nonanneal-ing ‘tail’ encoding for the wild-type loop sequence. Inparallel, PCR was performed with a 3′ gene terminalprimer and a primer annealing to the 25 nucleotidesimmediately 3′ of the loop and including a nonannealingtail. The first PCR product encodes from the start of thegene to the loop of interest and the second PCR productencodes from the loop of interest to the end of the gene.These two products are annealed and extended to yield thefull Fn3 gene containing thewild-type sequence in the loopof interest. Framework reversions were introduced bystandard site-directed mutagenesis using the QuikChangeMutagenesis Kit (Stratagene) according to the manufac-turer's instructions. Clone construction was verified byDNA sequencing.

Focused library construction

The DE randomization library was created in a mannersimilar to that of the L7.5.1 loop reversion clones. One PCRamplified the L7.5.1 S15P gene fragment 5′ of the DNAencoding for the DE loop. A second PCR amplified thegene 3′ of the DNA encoding the DE loop using a primerthat included a degenerate (NNB) DE loop sequence and20 nucleotides of overlap with the other PCR product. ThePCR products were annealed, extended to produce the fullgene, and amplified. This process was conducted inde-pendently with four oligonucleotides encoding the fourdifferent DE loop lengths. The gene fragments wereelectroporated into electrocompetent EBY100 along withpCT-Fn3-Loop vector. The resulting library encoded forL7.5.1 S15Pwith a fully randomDE loop of length 4, 5, 6, or7 amino acids.A library randomizing the unconserved residues of clones

similar to L7.5.1was constructed by PCR of L7.5.1 S15P. The5′ primer contained 17 nucleotides 5′ of the BC loop, 21nucleotides to encode the BC loop, and 22 nucleotides toanneal 3′ of the BC loop. The 3′ primer contained 19nucleotides 3′ of the FG loop, 30 nucleotides to encode forthe FG loop, and 10 nucleotides 5′ of the FG loop (note thatthe nucleotides encoding the first three conserved aminoacids of the FG loop also enable annealing during PCR). ThePCR products were amplified with extended primers toincrease the length of the conserved sequence flanking theloops to improve homologous recombination. The genefragments were electroporated into electrocompetentEBY100 along with pCT-Fn3-Loop vector. Two versions ofthe BC and FG loops were included. One oligonucleotidecompletely randomized the unconserved residues usingNNB degeneracy (BC, RXXPWAX; FG, RVGRXXXXXG).The other oligonucleotide restricted diversity to amino acidsobserved during affinity maturation [BC, R(D/G)(C/H/R/Y)PWA(I/T); FG, RVG(R/W)(A/M/T/V)(F/L/P/S)(C/D/G/Y)(A/T)(L/P/S)(G/S)].

Acknowledgements

The research was funded by CA96504, CA101830,a National Defense Science and Engineering Grad-uate Fellowship and a National Science Foundation


Graduate Fellowship. Assistance from the MITFlow Cytometry Core Facility is greatly appre-ciated. The Biophysical Instrumentation Facility forthe Study of Complex Macromolecular Systems(NSF-0070319 and NIH GM68762) is gratefullyacknowledged.

Supplementary Data

Supplementary data associated with this articlecan be found, in the online version, at doi:10.1016/j.jmb.2008.06.051

References

1. Huang, J., Koide, A., Nettle, K. W., Greene, G. L. &Koide, S. (2006). Conformation-specific affinity pur-ification of proteins using engineered binding pro-teins: application to the estrogen receptor. ProteinExpression Purif. 47, 348–354.

2. Karatan, E.,Merguerian,M., Han, Z. H., Scholle,M. D.,Koide, S. & Kay, B. K. (2004). Molecular recognitionproperties of FN3 monobodies that bind the Src SH3domain. Chem. Biol. 11, 835–844.

3. Koide, A., Abbatiello, S., Rothgery, L. & Koide, S.(2002). Probing protein conformational changes inliving cells by using designer binding proteins: appli-cation to the estrogen receptor. Proc. Natl Acad. Sci.USA, 99, 1253–1258.

4. Koide, A., Bailey, C. W., Huang, X. L. & Koide, S.(1998). The fibronectin type III domain as a scaffold fornovel binding proteins. J. Mol. Biol. 284, 1141–1151.

5. Koide, A., Gilbreth, R. N., Esaki, K., Tereshko, V. &Koide, S. (2007). High-affinity single-domain bindingproteins with a binary-code interface. Proc. Natl Acad.Sci. USA, 104, 6632–6637.

6. Lipovsek, D., Lippow, S. M., Hackel, B. J., Gregson,M. W., Cheng, P., Kapila, A. & Wittrup, K. D. (2007).Evolution of an interloop disulfide bond in high-affinity antibody mimics based on fibronectin type IIIdomain and selected by yeast surface display: mole-cular convergence with single-domain camelid andshark antibodies. J. Mol. Biol. 368, 1024–1041.

7. Parker, M. H., Chen, Y., Danehy, F., Dufu, K., Ekstrom,J., Getmanova, E. et al. (2005). Antibody mimics basedon human fibronectin type three domain engineeredfor thermostability and high-affinity binding to vas-cular endothelial growth factor receptor two. ProteinEng. Des. Sel. 18, 435–444.

8. Richards, J., Miller, M., Abend, J., Koide, A., Koide, S.& Dewhurst, S. (2003). Engineered fibronectin type IIIdomain with a RGDWXE sequence binds with en-hanced affinity and specificity to human alpha v beta 3integrin. J. Mol. Biol. 326, 1475–1488.

9. Xu, L. H., Aha, P., Gu, K., Kuimelis, R. G., Kurz, M.,Lam, T. et al. (2002). Directed evolution of high-affinityantibody mimics using mRNA display. Chem. Biol. 9,933–942.

10. Rock, E., Sibbald, P., Davis, M. & Chien, Y. (1994).CDR3 length in antigen-specific immune receptors. J.Exp. Med. 179, 323–328.

11. Ohlin,M.&Borrebaeck, C.A. K. (1996). Characteristicsof human antibody repertoires following activeimmune responses in vivo.Mol. Immunol. 33, 583–592.

12. Lamminmaki, U., Pauperio, S., Westerlund-Karlsson,A., Karvinen, J., Virtanen, P. L., Lovgren, T. & Savi-

ranta, P. (1999). Expanding the conformational diver-sity by random insertions to CDRH2 results inimproved anti-estradiol antibodies. J. Mol. Biol. 291,589–602.

13. Lee, C. V., Liang, W. C., Dennis, M. S., Eigenbrot, C.,Sidhu, S. S. & Fuh, G. (2004). High-affinity humanantibodies from phage-displayed synthetic fab lib-raries with a single framework scaffold. J. Mol. Biol.340, 1073–1093.

14. Batori, V., Koide, A. & Koide, S. (2002). Exploring thepotential of the monobody scaffold: effects of loopelongation on the stability of a fibronectin type IIIdomain. Protein Eng. 15, 1015–1020.

15. VanAntwerp, J. J. &Wittrup, K. D. (2000). Fine affinitydiscrimination by yeast surface display and flowcytometry. Biotechnol. Prog. 16, 31–37.

16. Bowley, D. R., Labrijn, A. F., Zwick, M. B. & Burton,D. R. (2007). Antigen selection from an HIV-1 im-mune antibody library displayed on yeast yieldsmany novel antibodies compared to selection fromthe same library displayed on phage. Protein Eng.Des. Sel. 20, 81–90.

17. Ling, M. M. (2003). Large antibody display librariesfor isolation of high-affinity antibodies. Comb. Chem.High Throughput Screening, 6, 421–432.

18. Sheets, M. D., Amersdorfer, P., Finnern, R., Sargent, P.,Lindqvist, E., Schier, R. et al. (1998). Efficient construc-tion of a large nonimmune phage antibody library: theproduction of high-affinity human single-chain anti-bodies to protein antigens. Proc. Natl Acad. Sci. USA,95, 6157–6162.

19. Hanes, J., Schaffitzel, C., Knappik, A. & Pluckthun,A. (2000). Picomolar affinity antibodies from a fullysynthetic naive library selected and evolved byribosome display. Nat. Biotechnol. 18, 1287–1292.

20. Stemmer, W. P. C. (1994). Rapid evolution of a proteinin-vitro by DNA shuffling. Nature, 370, 389–391.

21. Jirholt, P., Ohlin, M., Borrebaeck, C. A. K. & Soderlind,E. (1998). Exploiting sequence space: shuffling in vivoformed complementarity determining regions into amaster framework. Gene, 215, 471–476.

22. Marks, J. D., Griffiths, A. D., Malmqvist, M.,Clackson, T. P., Bye, J. M. & Winter, G. (1992).Bypassing immunization—building high-affinityhuman-antibodies by chain shuffling. Bio/Technology,10, 779–783.

23. Barbas, C. F., Hu, D., Dunlop, N., Sawyer, L., Cababa,D., Hendry, R. M. et al. (1994). In-vitro evolution of aneutralizing human-antibody to human-immunode-ficiency-virus type-1 to enhance affinity and broadenstrain cross-reactivity. Proc. Natl Acad. Sci. USA, 91,3809–3813.

24. Cadwell, R. C. & Joyce, G. F. (1992). Randomizationof genes by PCR mutagenesis. PCR Methods Appl. 2,28–33.

25. Main, A. L., Harvey, T. S., Baron, M., Boyd, J. &Campbell, I. D. (1992). The three-dimensional structureof the tenth type III module of fibronectin: an insightinto RGD-mediated interactions. Cell, 71, 671–678.

26. Chao, G., Lau, W. L., Hackel, B. J., Sazinsky, S. L.,Lippow, S. M. & Wittrup, K. D. (2006). Isolating andengineering human antibodies using yeast surfacedisplay. Nat. Protocols, 1, 755–768.

27. Zaccolo, M., Williams, D. M., Brown, D. M. &Gherardi, E. (1996). An approach to randommutagen-esis of DNA using mixtures of triphosphate deriva-tives of nucleoside analogues. J.Mol. Biol. 255, 589–603.

28. Kowalski, J. M., Parekh, R. N., Mao, J. &Wittrup, K. D.(1998). Protein folding stability can determine the




efficiency of escape from endoplasmic reticulumquality control. J. Biol. Chem. 273, 19453–19458.

29. Shusta, E. V., Kieke, M. C., Parke, E., Kranz, D. M.& Wittrup, K. D. (1999). Yeast polypeptide fusionsurface display levels predict thermal stability andsoluble secretion efficiency. J. Mol. Biol. 292,949–956.

30. Moore, G. L. & Maranas, C. D. (2000). Modeling DNAmutation and recombination for directed evolutionexperiments. J. Theor. Biol. 205, 483–503.

31. Orr, B. A., Carr, L. M., Wittrup, K. D., Roy, E. J. &Kranz, D. M. (2003). Rapid method for measuringScFv thermal stability by yeast surface display.Biotechnol. Prog. 19, 631–638.

Picomolar sensitivity MRI and photoacoustic imagingof cobalt nanoparticlesLouis-S. Boucharda,1, M. Sabieh Anwarb,1, Gang L. Liuc,2, Byron Hannc, Z. Harry Xied, Joe W. Grayc, Xueding Wange,1,Alexander Pinesf,g,1, and Fanqing Frank Chenc,h,i,1

aDepartment of Chemistry and Biochemistry, University of California, Los Angeles, CA 90095; bSchool of Science and Engineering, Lahore University ofManagement Sciences, Opposite Sector U, D.H.A. Lahore 54792, Pakistan; cComprehensive Cancer Center, University of California, San Francisco, CA 94143;dMinispec Division, Bruker Optics, Inc., 2700 North Crescent Ridge Drive, The Woodlands, TX 77381; eDepartment of Radiology, University of Michigan,Ann Arbor, MI 48109-0553; fCollege of Chemistry, University of California, Berkeley, CA 94720; gMaterials Sciences Division and hLife Sciences Division,Lawrence Berkeley National Laboratory, Berkeley, CA 94720; and iZhejiang California Nanosystems Institute, Zhejiang University, Hangzhou 310029,People’s Republic of China

Contributed by Alexander Pines, December 22, 2008 (sent for review November 6, 2008)

Multimodality imaging based on complementary detection principleshas broad clinical applications and promises to improve the accuracyof medical diagnosis. This means that a tracer particle advantageouslyincorporates multiple functionalities into a single delivery vehicle. Inthe present work, we explore a unique combination of MRI andphotoacoustic tomography (PAT) to detect picomolar concentrationsof nanoparticles. The nanoconstruct consists of ferromagnetic (Co)particles coated with gold (Au) for biocompatibility and a uniqueshape that enables optical absorption over a broad range of frequen-cies. The end result is a dual-modality probe useful for the detectionof trace amounts of nanoparticles in biological tissues, in which MRIprovides volume detection, whereas PAT performs edge detection.

dual-modality imaging � ferromagnetic nanoparticle �molecular imaging � MRI contrast � photoacoustic tomography

We have synthesized nanoparticles for dual-modality (1–7)MRI and photoacoustic tomography (PAT). The incorpo-

ration of MRI and PAT into a single probe offers the uniquepossibility of combining the complementary strategies of contrast-based volume imaging and edge detection. Our nanoconstructconsists of zero-valence ferromagnetic cobalt (Co) particles (8) witha gold (Au) coating for biocompatibility and a unique shaperendering increased optical absorption over a broad range offrequencies (9–19). This research theme follows the rapid devel-opments in nanotechnology, diagnostic radiology, and targetedmolecular imaging (20), whereby nanoparticulate contrast agents,with the desirable properties of high chemical specificity, biocom-patibility, and a reasonable half-life, are administered within aspecific region of interest. In nanoparticle-based imaging studies,higher particle concentrations lead to better signal-to-noise con-trasts, but this also poses a tradeoff with the toxicity. Therefore, oneof the most important parameters when developing particle-basedcontrast is the safest and lowest nanoparticle concentration thatoffers sufficient contrast sensitivity.

In MRI, magnetic materials such as gadolinium chelates andmagnetic nanoparticles are often used (21–23) to enhance imagecontrast. The magnetic nanoparticles are passivated by biocompat-ible coatings such as dextrin, citrate, polystyrene/divinylbenzene,and elemental gold. These coatings also detoxify the particles,resulting in enhanced lifetimes in vivo. Typical examples of mag-netic nanoparticulate core-shell configurations include magnetite–dextrin, magnetite–silica (24) and iron–gold (25).

Laser-based PAT (9–19) is a hybrid imaging modality [seesupporting information (SI) Fig. S1]. It uses a pulsed laser sourceto illuminate a biological sample. Light absorption by the tissueresults in a transient temperature rise on the order of 10 mK. Therapid thermoelastic expansion excites ultrasonic waves that aremeasured by using broadband ultrasonic transducers conformallyarranged around the sample. Finally, a modified back-projectionreconstruction algorithm (26) is used to construct a map of thedistribution of the optical energy deposition within the sample. The

spatial resolution of PAT is not limited by optical diffusion, butinstead by the bandwidth of the acoustic detectors. It has beenshown that PAT can depict subsurface tissue structures and func-tional changes noninvasively with resolution up to 100 �m (14, 15).Like other optical modalities, PAT is highly sensitive in mappingand quantifying the dynamic distribution of optical contrast agentssuch as metallic nanocolloids and organic dyes (16–19, 27).

In the present work, we have fabricated a composite-materialnanoparticle, which we call ‘‘nanowonton.’’ The nanowonton has aCo core and an Au thin-film coating and is a construct similar to theChinese eatable called the wonton (see Fig. 1 and Fig. S2). Thenanowontons have been characterized by scanning and transmis-sion electron microscopies, absorption spectroscopy, and NMRrelaxometry (Fig. 2). The nanowonton is shown to exhibit acombination of ferromagnetic and optical responses (Fig. 2), mak-ing it amenable to dual-modality MRI and PAT studies. NMR T2relaxivity measurements reveal a per-particle relaxivity of 1 � 107

s�1mM�1 (Table 1).Previously, the oxidation-induced instability and toxicity of Co

nanoparticles have prohibited their wide use as MRI contrastagents, but in the present case, the Au coating circumvents thisissue. Furthermore, the shape and thickness of the Au capping layerare designed so that the center of its optical absorption rangematches the near infrared laser excitation wavelength used in PATimaging (700 nm) optimizing the photothermal response. We havealso reported the geometry-dependent optical absorption for sim-ilarly shaped nanostructures such as nanocrescents (28). Thenanowonton design provides wavelength tunability for PAT (Fig.S3) and can be further improved through control of the fabricationprocedure.

The PAT imaging contrast is demonstrated in Fig. 3A, for anporcine gel containing several inclusions of different nanowontonconcentrations. The inclusion with a nanowonton concentration of13 pM can hardly be recognized from the background, leading usto conclude that this PAT system has a detection sensitivity of theorder of 25 pM. Spin echo MRI images of Co nanowonton agarosegel phantoms A and B are shown in Fig. 4 A and B. Spin echo imagesshow that higher concentrations lead to shorter T2 values for thewater protons. The smallest detectable concentration is 2.5 pM and

Author contributions: L.-S.B., M.S.A., G.L.L., X.W., A.P., and F.F.C. designed research; L.-S.B.,M.S.A., G.L.L., B.H., Z.H.X., and X.W. performed research; A.P. contributed new reagents/analytic tools; L.-S.B., M.S.A., G.L.L., and X.W. analyzed data; and L.-S.B., M.S.A., X.W., andF.F.C. wrote the paper.

The authors declare no conflict of interest.

1To whom correspondence may be addressed. E-mail: [email protected],[email protected], [email protected], [email protected], or f�[email protected].

2Present address: Department of Electrical and Computer Engineering, University of Illinois atUrbana-Champaign, 3104 Micro and Nanotechnology Laboratory, 208 North Wright Street,MC-249, Urbana, IL 61801.

This article contains supporting information online at www.pnas.org/cgi/content/full/0813019106/DCSupplemental.

www.pnas.org�cgi�doi�10.1073�pnas.0813019106 PNAS � March 17, 2009 � vol. 106 � no. 11 � 4085–4089

MED

ICA

LSC

IEN

CES

CHEM

ISTR

Y

http://www.pnas.org/cgi/data/0813019106/DCSupplemental/Supplemental_PDF#nameddest=SF1




http://www.pnas.org/cgi/content/full/0813019106/DCSupplemental

http://www.pnas.org/cgi/content/full/0813019106/DCSupplemental

the contrast with respect to 5 pM is also clearly visible in Fig. 4B and D.

A T2-weighted spin echo image from a slice through the mouse’sleg muscles is shown in Fig. 5. The injection of PBS-buffered Conanoparticles at 50 pM concentration results in a substantial dropin the MR signal in this region whereas the control injection with

PBS shows no such contrast enhancement. In Fig. S4, PAT imagesacquired before and after a rat tail injection show the type ofcontrast enhancement which can be expected from a local injectionof 100 pM contrast agent. Because of high-frequency ultrasounddetection, the PAT modality is generally better suited at delineatingedges at the location of the contrast. These nanowonton particlesadvantageously combine the strengths of both MRI and PATmodalities into a single delivery vehicle.

This dual-modality PAT/MRI contrast agent demonstrates, sofar, the most sensitive detection experiment of magnetic nanopar-ticles with particle concentrations in the picomolar and tens ofpicomolars range. The particles may even be used for stand-aloneMRI or PAT. For example, in the MRI studies, the T2 contrast isclearly visible to concentrations as low as 2.5 pM in phantoms and50 pM in tissues. These detection thresholds are 7 orders ofmagnitude better than those demonstrated by Lu et al. (23) formonocrystalline iron oxide particles. Our T2 relaxivity (see Table 1)per-particle concentration is 5 orders of magnitude better than thecited work. The particle relaxivity and T2-weighted MRI detectionthreshold are also better than those demonstrated by Cho et al. (25)for 18 nm-diameter Fe/Au nanoparticles. This degree of sensitivityis, to our knowledge, unprecedented and compares with sensitivitiesapproaching those of radioactive labels. This improved perfor-mance is in large part contributed by our choice of a ferromagneticmaterial, cobalt, which has a saturation magnetization 3.42 timeslarger than magnetite, leading to a per-particle relaxivity that isnearly 12 times larger. Because T2-weighted MRI depends expo-nentially on the relaxation rate, this leads to a substantial differencein contrast observed in our experiments. We refer the reader to theSI section for a more extensive discussion of relaxivity effects,including a comparison with results from other agents reported inthe literature.

The highly stable, thin (10 nm) film (Au) coating providesbiocompatibility, as demonstrated by experimental results (Fig. S5).Furthermore, the Au thin film deposition process can be wellcontrolled to allow tunable absorption spectra, allowing PAT atdifferent optical wavelengths (see Fig. S3). It has been demon-

Fig. 2. Characterization of the cobalt nanowontons.(A) Scanning electron microscopy image ofnanowontons. (B) Transmission electron microscopy im-age of 3 nanowontons in various diameters; notice thatthe lighter regions in the nanoparticle are relatively hol-low, and are responsible for photothermal tuning prop-erties of the nanowonton (38). (C) Particle diameter dis-tribution of 150 nanowontons. Because of theinhomogeneous polysilicon nanopillar diameter, the sizeof the nanowontons varies from 30 to 90 nm, and theaverage diameter is 60 nm. (D) Absorption spectrum ofnanowonton, medium peak wavelength is �700 nm. (E)Spin–spin relaxation time T2 measured at 20 MHz protonfrequency and 37 °C. T2 begins to change when the con-centration exceeds 20 pM. Note that although 20 MHz ismuch less than 300 MHz (used for the MRI), this gives alower bound on relaxivity and shows that the contrastworks even at low fields, such as those from portableNMR devices.

Fig. 1. Fabrication procedure of the nanowontons including 6 steps: (1) Etchingpolysiliconnanopillarsonthesurfaceofsinglecrystallinesiliconwafer (forsimplerpresentation,weomittedfromthe illustrationthepreparatory stepofdepositing5nmofchromiumbeforegoingtostep2, see Methods); (2)depositionofa10-nmgold thin film; (3) deposition of 10-nm cobalt thin film; (4) deposition of 10-nmgold thin film; (5) etching polysilicon nanopillars in KOH batch solution; and (6)completeremovalofpolysiliconandchromiumbyKOHetchingandseparationofnanowontons.

4086 � www.pnas.org�cgi�doi�10.1073�pnas.0813019106 Bouchard et al.




strated that a variety of gold nanocolloids are already entering invivo clinical trials (19, 29–33). Among them, Au nanorods presentparticularly good optical absorption in the near-infrared region,tunable by changing the aspect ratio. It has already been demon-strated that gold nanorod contrast agents can be imaged with PAT,both ex and in vivo (34, 35). Our study has shown that the sensitivityof PAT in imaging the nanowonton is equivalent to that for goldnanorods. In fact, the MRI contrast is also expected to be stronglydependent on the shape of the nanoconstruct. It is envisaged thatnanorods or needle-shaped structures can elicit greater contrastbecause of larger shape-induced susceptibility gradients. Thepresent nanowonton shape is, to first order, a compromise betweenoptical and magnetic responses. However, further work on shapeoptimization will be required to conclusively comment on this.Furthermore, the Au sandwich structure also allows additionaltuning of absorbed wavelengths (28, 36, 37). This can furtherimprove the sensitivity of the PAT technique. Last, the Au coatingsare especially attractive because of the possibility of conjugating theparticles with specific molecules such as antibodies, specific ligands,thiol functional groups and therapeutic drugs, opening up prospectsfor targeted molecular imaging (20). An additional imaging mo-dality built into our nanoconstruct is the optical thermal conversioncapability making these structures highly suited for photothermaltherapy (32, 38–41).

MethodsFor additional information on materials and methods used, see SI Text.

Fabrication of the Nanowontons. The schematic diagram of the fabricationprocedure is illustrated in Fig. 1. First, a batch-fabricated vertical silicon nanopillararray was fabricated on the surface of a 4-inch diameter silicon wafer. Thecoverage of the nanopillar structure was �90% of the total wafer surface area.On the top of each silicon nanopillar, there was a spherical silicon oxide nano-structure. Four metallic layers of 5-nm chromium, 10-nm gold, 10-nm cobalt, and10-nm gold were sequentially deposited on the wafer surface. However, afterdeposition, the sidewalls of all of the nanopillars remained exposed. The siliconwaferwastherefore immersedina10%KOHbathsolutionat80 °C,etchingawaythe nanopillars from the unprotected sidewalls in 10 min, The multilayer metallicnanostructure on the top of the nanopillars was lifted off and suspended in theKOH bath solution. Because silicon oxide and chromium were also etched awayby KOH, only the gold–cobalt–gold sandwich nanostructures, the nanowontons,remained in solution. These were finally separated by centrifugation. The SEM,TEM, and size distribution measurements are shown in Fig. 2 A–C. After fabrica-tion, these samples were chemically analyzed by inductively coupled plasma massspectrometry (ICP-MS),measuringthetotalamountofCoorAu ions.Byassumingbulk parameters of the materials and the size of the nanoparticles, we deducedthe mass of Co and Au per nanoparticle, and calculated the nanoparticleconcentration.

PAT System. The PAT system is schematically shown in Fig. S1. An OPO (VibrantB; Opotek) pumped by an Nd/YAG laser (Brilliant B; Bigsky) was used to providelaser pulses with a repetition rate of 10 Hz and a pulse width of 5 ns. In this study,the wavelength of the laser light was tuned to 700 nm, which was in thenear-infrared region and enabled good penetration in biological tissues. Thelaser beam, after being expanded and homogenized, illuminated the imagedsample with an input energy density of �10 mJ/cm2, well below the AmericanNational Standards Institute safety limit of 22 mJ/cm2 at the applied wavelength.The laser light penetrated into the sample and generated photoacoustic signalsthat were scanned by an ultrasonic transducer (XMS-310; Panametrics) with acenter frequency at 10 MHz and a receiving bandwidth of 100%. To realize 2Dcross-sectional imaging, the sample was rotated axially in the xy plane while thetransducer and the laser beam were kept static. To couple the signals, the sampleand the transducer were immersed in water. After a preamplifier (PR5072;Panametrics), the detected signals were digitized by an oscilloscope (TDS 540B;Tektronics) and then collected by a computer. The current PAT system exhibitsspatial resolutionof200�minthexyplane,whichhasbeenverifiedbymeasuringthe line spread function (LSF) (40).

Photoacoustic Imaging on Phantoms. To demonstrate the PAT imaging contrast,we constructed a phantom made of 5% porcine gel in which 4 inclusions withdifferent concentrations of nanowontons are embedded (Fig. 3). Cylindrical-shaped phantoms with a 20-mm diameter were made from porcine gel. Spher-ical-shaped droplets with a size of 2.8 mm were made with the same gel andcontained different concentrations of the contrast agent. These droplets wereembedded 1 cm deep in the phantoms. For example, the phantom shown in Fig.3 contains 4 such embedded droplets, where the nanowonton concentrationswere 100, 50, 25, and 13 pM. The corresponding PAT image is shown in Fig. 3A,

Fig. 3. Photoacoustic imaging of nanowonton phan-tom gels. (A) PAT image of 4 absorbing objects contain-ing nanowonton contrast agent embedded in a gelphantom (5% agarose). The concentrations ofnanowontons were 100, 50, 25, and 13 pM, respectively,for objects A, B, C, and D. (B) Intensity profiles extractedfrom the image along 4 lines (horizontal and verticaldashed lines indicated on the image) going through theabsorbing centers are plotted to highlight the visibility ofnanowonton inclusions in thereconstructed image.Witha CNR close to 1, the object D, where the nanowontonconcentration is 13 pM, can hardly be recognized fromthe background, showing that the current PAT systemhas detection sensitivity on the order of 25 pM.

Table 1. T2 relaxivity per particle concentration is calculatedto be 1 � 107 s�1mM�1

Sample no. Conc, pM T2, ms Conc, mM 1/T2, s�1

1 1,000 123 0.000001 8.13012 800 124 0.0000008 8.06453 600 163 0.0000006 6.13504 500 204 0.0000005 4.90205 400 239 0.0000004 4.18416 300 333 0.0000003 3.00307 200 451 0.0000002 2.21738 100 828 0.0000001 1.20779 50 1384 0.00000005 0.7225

10 25 1788 0.000000025 0.559311 10 2250 0.00000001 0.444412 5 2258 0.000000005 0.442913 2 2268 0.000000002 0.440914 1 2270 0.000000001 0.4405

The T2 reaches saturation point above 800 pM, so the 1,000-pM data pointwas excluded from the fit to obtain linearity for calculating 1/T2.

Bouchard et al. PNAS � March 17, 2009 � vol. 106 � no. 11 � 4087

MED

ICA

LSC

IEN

CES

CHEM

ISTR

Y

http://www.pnas.org/cgi/data/0813019106/DCSupplemental/Supplemental_PDF#nameddest=STXT


the locations of the objects being marked with dashed circles. In Fig. 3A, we havealso quantified the contrast-to-noise ratio (CNR) � (So � Sb)/�, where So is theaverage intensity within the object, Sb is the average intensity within the back-ground defined by the mean of all of the pixels in the image beyond the bigdashed circle (i.e., the area out of the gel phantom), and � is the standarddeviation. The computed CNRs for the objects A, B, C, and D are 10.6, 5.7, 2.1, and1.0. It is clear that the objects A, B, and C have been imaged with sufficient opticalabsorption contrast.

MRI of the Phantoms. For thephantomMRIstudies,6holesweredrilled intoa (�3cm diameter) Teflon cylinder, azimuthally distributed around the center. Thephantomis showninFig.4.Eachholewas5mmindiameterand�1cmdeep.Twosimilar pieces (A and B) were machined. Nanocolloidal solutions of thenanowonton in (5%) agarose gel were prepared in concentrations of 500, 375,250, 125, 50, 12.5, 5, and 2.5 pM. The agarose gel was heated until it becametransparent, and the nanoparticles were subsequently transferred to the hotagarose, preparing the desired concentrations. Precise volumes of the nanocol-loids were then slowly transferred to the cylindrical recesses and allowed to coolin ambient conditions, ensuring that no air bubbles were formed during thecooling. The solutions were intermittently pried to ensure that the distribution ofthe nanowontons would be kept as uniform as possible. The tops and bottoms ofthe phantoms were sealed with polystyrene to prevent leakage during phantomhandling. The phantom A contained the concentrations 500, 375, 250, 125, and50 pM, whereas B contained 125, 12.5, 5, and 2.5 pM. The latter had 1 cylinderempty, and each of the phantoms was also loaded with 0.6 mM MnCl2 dopedwater to act as the control reference.

The MRI was performed in a 300-MHz NMR spectrometer (Varian Inova)equipped with triple-axis magnetic field gradients. The phantoms were imagedby using a spin-echo pulse sequence, with slice selection along the z axis, phaseencodingalongtheyaxis,andreadoutalongthexaxis.Weusedanechotime(TE)of 50 ms and recycle time (TR) of 1 s. The field of view was 3 � 3 cm, the numberof points was 256 � 128, and the slice thickness along the z direction was 1 mm.T2 measurements were also performed by repeating the spin echo sequence withvarying TEs; the values used were 10, 30, 40, 50, 100, 150, and 200 ms.

Fig. 4. MRI of nanowonton gel phantoms. The nanowonton gels are arranged along the perimeter of a circle. (A and B) Spin-echo images for the phantoms A andB, respectively, with an echo time (TE) of 50 ms and recycle time (TR) of 1 s. The higher-concentration samples appear darker in the images, with doped water used asa control and exhibiting the strongest T2-weighted intensity. The concentrations of the gels are given in the figure along with the T2 values that are deduced from a7-point curve-fitting procedure. The field of view for this image is 3 � 3 cm, the number of points is 256 � 128, and the slice thickness is 1 mm. (C and D) Intensity profilesfor the images in A and B. The relative intensities along a circular contour drawn through the middle of the gels are plotted as a function of the gel azimuthal anglefrom the x axis. The nanoparticles contrast remains detectable down to 2.5 pM. The images for phantom A and B are plotted to different (normalized) scales.

Fig. 5. Transverse (axial) MRI image in mouse leg muscle injected with Conanoparticles in PBS solution. The position of the blue arrows indicates the sitesof injection for the Co nanoparticles (upper right corner) and the PBS control(lower left corner). Two water-carrying test tubes are also visible in the scans forpurposes of MRI slice alignment (red arrows). MRI parameters were: TE � 50 ms;TR � 1 s; field of view is 2.6 cm � 2.6 cm, and slice thickness is 0.5 mm.

4088 � www.pnas.org�cgi�doi�10.1073�pnas.0813019106 Bouchard et al.

MRI on Mouse Muscle. For the animal studies, the mouse was i.p. anesthetizedwith 400 �L of Avertin. After 5 min, 50 �L of 60-nm-sized gold-coated Conanoparticles at 50 pM in a solution of PBS were intramuscularly injected intothe leg. As a control, PBS solution without the nanoparticles was also injectedin the diametrically opposite position to the site of injection of the cobaltnanowonton. After an additional 10 min, the mouse was killed and placed intothe vertical bore of the Varian 300-MHz NMR spectrometer. The mouse tailwas then imaged by using a T2-weighted spin-echo sequence. Multiple trans-verse slices were imaged, the slice thickness (along the z direction) being 0.5mm, the TR was 1 s, the number of points was 256 � 128, and the field of viewwas 2.6 � 2.6 cm.

ACKNOWLEDGMENTS. This work was supported by the Prostate Cancer Foun-dation and University of California, San Francisco (UCSF), Prostate CancerSpecialized Program of Research Excellence (SPORE) ; Camille and HenryDreyfus Foundation (L.-S.B.); National Natural Science Foundation of ChinaGrant NSFC-30828010, the Zhejiang California Nanosystem Institute, Depart-ment of Defense Breast Cancer Research Program Grants BC045345 andBC061995, and University of California San Francisco Prostate Cancer SPOREaward (National Institutes of Health Grant P50 CA89520) (F.F.C.); Director,Office of Science, Office of Basic Energy Sciences, Materials Sciences Division,of the U.S. Department of Energy Contract DE-AC03-76SF00098; NationalInstitutes of Health Grant R01 AR055179; and Department of Defense ProstateCancer Research Program Grant W81X WH-07-1-0231.

1. T Al-Jamal W, Kostarelos, K (2007) Liposome-nanoparticle hybrids for multimodaldiagnostic and therapeutic applications. Nanomedicine-UK 2:85–98.

2. Jaffer FA, et al. (2006) Cellular imaging of inflammation in atherosclerosis usingmagnetofluorescent nanomaterials. Mol Imaging 5:85–92.

3. Mulder WJM, et al. (2005) MR molecular imaging and fluorescence microscopy foridentification of activated tumor endothelium using a bimodal lipidic nanoparticle.FASEB J 19:doi:10.1096/fj.05–4145fje.

4. Tan WB, Zhang Y (2007) Multi-functional chitosan nanoparticles encapsulating quan-tum dots and Gd-DTPA as imaging probes for bio-applications. J Nanosci Nanotechno7:2389–2393.

5. Dosev D, et al. (2007) Magnetic/luminescent core/shell particles synthesized by spraypyrolysis and their application in immunoassays with internal standard. Nanotechnol-ogy 18:055102.

6. Prinzen L, et al. (2007) Optical and magnetic resonance imaging of cell death andplatelet activation using annexin A5-functionalized quantum dots. Nano Lett 7:93–100.

7. Medarova Z, Pham W, Kim Y, Dai GP, Moore A (2006) In vivo imaging of tumor responseto therapy using a dual-modality imaging strategy. Int J Cancer 118:2796–2802.

8. Bala T, Arumugam SK, Pasricha R, Prasad BLV, Sastry M (2004) Foam-based synthesis ofcobalt nanoparticles and their subsequent conversion to Co core Ag shell nanoparticlesby a simple transmetallation reaction. J Mater Chem 14:1057–1061.

9. Hoelen CGA, de Mul FFM, Pongers R, Dekker A (1998) Three-dimensional photoacous-tic imaging of blood vessels in tissue. Opt Lett 23:648–650.

10. Oraevsky AA, Jacques SL, Tittel FK (1997) Measurement of tissue optical properties bytime-resolved detection of laser-induced transient stress. Appl Optics 36:402–415.

11. Kruger RA, Liu PY, Fang YR, Appledorn CR (1995) Photoacoustic Ultrasound (PAUS) -Reconstruction Tomography. Medical Physics 22:1605–1609.

12. Andreev VG, Karabutov AA, Oraevsky AA (2003) Detection of ultrawide-band ultra-sound pulses in optoacoustic tomography. IEEE T. Ultrason. Ferr. 50:1383–1390.

13. Kolkman RGM, Hondebrink E, Steenbergen W, de Mul FFM. (2003) In vivo photoacous-tic imaging of blood vessels using an extreme-narrow aperture sensor. IEEE J. Sel. Top.Quant. 9:343–346.

14. Wang XD, et al. (2004) Noninvasive photoacoustic angiography of animal brains in vivowith near-infrared light and an optical contrast agent. Opt Lett 29:730–732.

15. Wang XD, Pang YJ, Ku G, Stoica G, Wang LHV (2003) Three-dimensional laser-inducedphotoacoustic tomography of mouse brain with the skin and skull intact. Opt Lett28:1739–1741.

16. Ku G, Wang XD, Stoica G, Wang LHV (2004) Multiple-bandwidth photoacoustic to-mography. Physics in Medicine and Biology 49:1329–1338.

17. Ku G, Wang XD, Xie XY, Stoica G, Wang LHV (2005) Imaging of tumor angiogenesis inrat brains in vivo by photoacoustic tomography. Appl Optics 44:770–775.

18. Wang XD, et al. (2003) Noninvasive laser-induced photoacoustic tomography forstructural and functional in vivo imaging of the brain. Nat Biotechnol 21:803–806.

19. Wang YW, et al. (2004) Photoacoustic tomography of a nanoshell contrast agent in thein vivo rat brain. Nano Lett 4:1689–1692.

20. Wickline SA, Lanza GM (2002) Molecular imaging, targeted therapeutics, and nano-science. J Cell Biochem, 90–97.

21. Cunningham CH, et al. (2005) Positive contrast magnetic resonance imaging of cellslabeled with magnetic nanoparticles. Magnet Reson Med 53:999–1005.

22. Kim DK, et al. (2001) Characterization and MRI study of surfactant-coated superpara-magnetic nanoparticles administered into the rat brain. J Magn Magn Mater 225:256–261.

23. Lu J, et al. (2006) Solid-state synthesis of monocrystalline iron oxide nanoparticle basedferrofluid suitable for magnetic resonance imaging contrast application. Nanotech-nology 17:5812–5820.

24. Lu CW, et al. (2007) Bifunctional magnetic silica nanoparticles for highly efficienthuman stem cell labeling. Nano Lett 7:149–154.

25. Cho SJ, Jarrett BR, Louie AY, Kauzlarich SM (2006) Gold-coated iron nanoparticles: anovel magnetic resonance agent for T1 and T2 weighted imaging. Nanotechnology17:640–644.

26. Xu MH, Wang LHV (2005) Universal back-projection algorithm for photoacousticcomputed tomography. Phys Rev E 71:016706.

27. Copland JA, et al. (2004) Bioconjugated gold nanoparticles as a molecular basedcontrast agent: Implications for imaging of deep tumors using optoacoustic tomog-raphy. Mol Imaging Biol 6:341–349.

28. Lu Y, Liu GL, Kim J, Mejia YX, Lee LP (2005) Nanophotonic crescent moon structureswith sharp edge for ultrasensitive biomolecular detection by local electromagneticfield enhancement effect. Nano Lett 5:119–124.

29. Durr NJ, et al. (2007) Two-photon luminescence imaging of cancer cells using molec-ularly targeted gold nanorods. Nano Lett 7:941–945.

30. Huff TB, et al. (2007) Hyperthermic effects of gold nanorods on tumor cells. Nano-medicine-UK 2:125–132.

31. Lee KS, El-Sayed MA (2006) Gold and silver nanoparticles in sensing and imaging:Sensitivity of plasmon response to size, shape, and metal composition. J Phys Chem B110:19220–19225.

32. Loo C, Lowery A, Halas N, West J, Drezek R (2005) Immunotargeted nanoshells forintegrated cancer imaging and therapy. Nano Lett 5:709–711.

33. Kim K, et al. (2007) Photoacoustic imaging of early inflammatory response using goldnanorods. Appl Phys Lett 90:223901.

34. Chamberland DL, et al. (2008) Photoacoustic tomography of joint aided by Etanerceptconjugated gold nanoparticle contrast agent - an ex vivo preliminary rat study.Nanotechnology 19:95–101.

35. Eghtedari M, et al. (2007) High sensitivity of in vivo detection of gold nanorods usinga laser optoacoustic imaging system. Nano Lett 7:1914–1918.

36. Liu GL, et al. (2006) A nanoplasmonic molecular ruler for measuring nuclease activityand DNA footprinting. Nat Nanotechnol 1:47–52.

37. Liu GL, et al. (2007) Peptide-nanoparticle hybrid SERS probes for optical detection ofprotease activity. J Nanosci Nanotechno 7:2323–2330.

38. Liu GL, Kim J, Lu Y, Lee LP (2006) Optofluidic control using photothermal nanoparticles.Nat Mater 5:27–32.

39. Loo C, et al. (2004) Nanoshell-enabled photonics-based imaging and therapy of cancer.Technol Cancer Res T 3:33–40.

40. Chou CH, Chen CD, Wang CRC. (2005) Highly efficient, wavelength-tunable, goldnanoparticle based optothermal nanoconvertors. J Phys Chem B 109:11135–11138.

41. Hirsch LR, et al. (2003) Nanoshell-mediated near-infrared thermal therapy of tumorsunder magnetic resonance guidance. P Natl Acad Sci USA 100:13549–13554.

Bouchard et al. PNAS � March 17, 2009 � vol. 106 � no. 11 � 4089

MED

ICA

LSC

IEN

CES

CHEM

ISTR

Y

Current Nanoscience, 2005, 1, 47-64 47

1573-4137/05 $50.00+.00 © 2005 Bentham Science Publishers Ltd.

Nanosystems in Drug Targeting: Opportunities and Challenges

Jaspreet K. Vasir1, Maram K. Reddy1 and Vinod D. Labhasetwar1,2,*

1Department of Pharmaceutical Sciences, College of Pharmacy, 986025 Nebraska Medical Center, Omaha, NE 68198-6025, USA, 2Department of Biochemistry and Molecular Biology, University of Nebraska Medical Center, Omaha, NE68198, USA

Abstract: The long cherished goal of targeting drugs to specific sites in the body, where the pharmacological action isdesired and sparing other tissues has been actively pursued all these years. The concept of ‘magic bullets’ given byEhrlich has now seen a metamorphosis to ‘magic wands’, in the form of targeted drug delivery systems. The magic, alldue to the specific targeting ligands which guide the drug carriers to the molecular targets be it on cell surface or nuclearmembranes. Nanosystems including the nano-sized (<1000 nm) drug carrier systems, such as polymeric nanoparticles,liposomes, micelles and polymer-drug conjugates are the vanguards of this ever-evolving field. Targeting drugs to specificsites, and maintaining pharmacologically relevant drug levels at the site for a period required for desired therapeutic actionis what makes the nanosystems – the burgeoning magic wands. Substantial challenges still exist in terms of biologicalbarriers. Nevertheless, the approaches like directly reaching the target using catheters, or using exogenous guidingmechanisms (magnetic fields and ultrasound), and exploiting the accessible targets on vascular endothelium are emergingas new and promising trends. It is conceivable, that despite all the formidable challenges, interplay of different disciplinesranging from engineering to biology will make the dream of drug targeting come true!

Keywords: Nanoparticles, nanosystems, targeting, ligands, drug and gene therapy.

1. INTRODUCTION

In the past few decades, rapid advances in cell andmolecular biology have allowed us to develop a betterunderstanding of the pathophysiology of various diseases.This has cast a floodlight on the possible cellular andmolecular targets which can be exploited not only for drugdiscovery but also for imaging and therapy. Further,integration of high throughput techniques and combinatorialchemistry with biology has brought an explosive growth inthe number of new drug molecules (in discovery pipeline)aimed to act on the identified targets. Therapeuticarmamentarium has thus seen an exponential growth andnow ranges from low molecular weight drugs tomacromolecules like proteins and plasmid DNA. However,molecular complexity associated with drugs andinaccessibility of most physiological targets presents theHoly Grail of drug delivery - to deliver these specific drugsto their site of action at therapeutically relevant levels. Thus,drug targeting has evolved as the most desirable but elusivegoal in drug delivery science. It can potentially increaseefficacy and reduce toxicity of new and pre-existing drugs byaltering their pharmacokinetics and biodistribution andrestricting the action of drugs to the treated tissue. Thepurpose of this review is to present an overview of thebarriers to drug targeting, various approaches designed toovercome these barriers and to focus on the development ofnanosystems as potential carriers for drugs to enable sitespecific targeting (Fig. 1). Special mention is made for

*Address correspondence to this author at the 986025 Nebraska MedicalCenter, Omaha, NE 68198-6025, USA; Tel: (402) 559-9021; Fax: (402)559-9543; E-mail: [email protected]

emerging targeting technologies such as transcriptionaltargeting, physical targeting, targeting of vascular andintracellular targets.

2. BARRIERS TO DRUG TARGETING

Targeting of drugs offers enormous advantages but isequally challenging. A better understanding of thephysiological barriers which a drug needs to overcomeshould enable the pharmaceutical scientists to developsuccessful design of targeted drug delivery systems. Mainhurdles to drug targeting include physiological barriers,biochemical challenges to identify and validate the moleculartargets and the pharmaceutical challenges to deviseappropriate techniques of conjugating targeting ligands to thenanosystems. The challenge in drug targeting is not only thetargeting of drug to a specific site but also retaining it for thedesired duration to elicit pharmacological action. For ananosystem administered intravenously, the first andforemost barrier is that of the vascular endothelium and thebasement membrane [1]. Also, plasma proteins have theability to affect the biodistribution of drug carrier systemsintroduced in the blood stream. The in vivo biodistributionand opsonization of nanosystems in blood circulation isgoverned by their size and surface characteristics. For thenanosystem to remain in blood circulation for a long time,the major problem is to avoid its opsonization andsubsequent uptake by the phagocytic cells. The passage ofdrug molecules and drug delivery systems across theendothelium is sensitive to the molecular weight and size ofthe system, respectively. The tight endothelial cells in brainconstitute the blood brain barrier (BBB), which restricts theentry of most drugs and delivery systems. However vascularendothelium is not uniform throughout. The altered

48 Current Nanoscience, 2005, Vol. 1, No. 1 Labhasetwar et al.

endothelia in tumors allow an enhanced permeability tomacromolecules and the particulate drug delivery systems.Another barrier is that of the extracellular matrix, whichshould be crossed to access the target cells in a tissue. If thewhole tissue constitutes a target then the uniform distributionof drug throughout the tissue is another problem [1].Targeting nanosystems to specific receptors or antigens onthe cell surface provides the driving force for diffusion of thesystem to the specific cells.

For drugs whose targets are located in thecytoplasm/nucleus of a cell, further barrier needs to becrossed to allow internalization of nanosystems into specificcells [2]. The barriers not end here, a number of endocyticpathways have been described for the cellular entry ofnanosystems. Drugs/nanosystems need to diffuse through theviscous cytosol to access the particular cytoplasmic targetswhere site of action is located. Nuclear membrane posesanother formidable barrier for drugs such as oligo-nucleotides, plasmid DNA and other low molecular weightdrugs whose site of action is located in the nucleus of a cell.

Although a number of cellular and molecular targets areemerging, the real problem lies with the poor accessibility ofdrugs/ nanosystems to the target tissue. The presence of suchbarriers leads to a poor in vitro/in vivo correlation when thetargeted delivery systems are tested in receptor bearing cellsin vitro and fail in vivo [3]. Thus, to harness the potential ofnew targets in imaging and therapy, one would need todevelop targeted systems which can successfully overcome

the physiological barriers, and for drug therapy, deliver thepharmacological agent to its site of action at therapeuticallyrelevant drug levels for a time sufficient to allow therapeuticaction. Conjugation of targeting ligands to drugs or drugcarrier nanosystems is the most popular way of directingthem to their target sites. To this end, various techniqueshave been devised, including covalent and non-covalentconjugation [4]. The emphasis is that the ligand must beattached stably and accessibly to the drug carrier, so that theligand is presented in its right orientation for binding to thetarget receptors. For example, the monoclonal antibodiesmust bind to the drug/its carrier with their F c part, so thattheir antigen binding site (F ab) is free to interact with theantigenic targets on cells [5]. The coupling reactions mustnot affect the biological activity of ligand and should notadversely affect the structure of drug delivery nanosystems.Further, such coupling reactions must be optimized so thatbinding of ligands takes place in a homogeneous manner onthe surface of the drug carrier nanosystems.

3. APPROACHES TO DRUG TARGETING

An ideal targeted drug delivery approach would not onlyincrease therapeutic efficacy of drugs but also decrease thetoxicity associated with drug to allow lower doses of thedrug to be used in therapy. A vast array of methods, whichcan further be classified into two key approaches – activeand passive have been explored for targeting drugs by meansof designing innovative nanosystems (Table 1). Some of the

Fig. (1). Schematic describing barriers to drug targeting and the role of nanosystems in overcoming these barriers.

Nanosystems in Drug Targeting: Opportunities and Challenges Current Nanoscience, 2005, Vol. 1, No. 1 49

methods like catheterization, direct injection, prodrugs orchemical delivery systems have inherent capability to deliverdrugs or their appropriate modifications to the specific site ofaction. However, others require design of a suitable carriersystem to be able to deliver the drug to its target site.Following is a description of such approaches to drugtargeting.

3.1. Passive Targeting Approaches

Passive targeting refers to the accumulation of drug ordrug-carrier system at a particular site due to physico-chemical or pharmacological factors [1]. Drug or drug carriernanosystems can be passively targeted making use of thepathophysiological and anatomical opportunities.

3.1.1. Pathophysiological Opportunities

The physiology of diseased tissues may be altered in avariety of pathological conditions, and can be exploited forpassively targeting drugs. Release of various chemotacticfactors from the infected/inflamed tissues results in vascularremodeling to enable leukocyte extravasation and hence alsoincreases permeability for particulate drug carriers [6, 7].Such pathophysiological opportunities include increasedvascular permeability in various inflammatory conditionswhich allows extravasation of the nanosystems and theirselective localization in the inflamed tissue [8]. It was shownthat polymeric nanoparticles administered per orally could beselectively targeted to the inflamed colonic mucosa ininflammatory bowel disease [9]. Similarly, the blood brainbarrier can also be crossed to access the target sites for braindelivery in some inflammatory conditions [10].

Solid tumors present much more favorable conditions forpreferential accumulation of macromolecular drugs andcolloidal sized drug delivery systems like polymeric-drugconjugates, liposomes etc. Rapidly growing or metastasizingtumors recruit new blood vessels through the process ofangiogenesis to meet the nutritional demands of increasingnumber of cells. Unlike the tight endothelium of normal

blood vessels, the vascular endothelium of angiogenic bloodvessels has large gaps (600-800 nm) in between the adjacentendothelial cells [8]. This increased vascular permeabilitycoupled with the impaired lymphatic drainage in tumorsallows an enhanced permeability and retention (EPR) effectof the nanosystems in the tumor [11, 12]. For passive tumoraccumulation using the EPR effect, the targeted systemshould have long blood circulation time, should not lose drugactivity while in circulation and the drug should stay with thecarrier until accumulation of drug at the target is attained.Other factors which may influence tumor accumulation bythe EPR effect include the degree of tumorvascularization/angiogenesis and the size of the deliverysystem. Liposomes, polymeric nanoparticles, micellarsystems as well as polymeric-drug conjugates have beensuccessfully used to target drugs to tumor tissues in a passivemanner. The EPR effect has also been explored to targetdrugs to the injured artery following balloon angioplasty toinhibit restenosis. Since the angioplasty disrupts theendothelium of the target artery, it is more permeable tonanosystems than the normal artery [13].

3.1.2. Anatomical Opportunities

Drugs may be introduced into “discrete anatomicalcompartments” for example: lungs, knee joints, respiratorytract, and eye by means of minimally invasive procedures,using catheters or direct injections at local sites. Thesemethods of site-specific localized drug delivery preventunwanted systemic exposure of the drug and thus avoidadverse effects of drugs in the non target tissues. This notonly minimizes the dose but also the cost of therapy. Theabove approach of overcoming biological barriers offers anexcellent method of localizing the cytotoxic action ofantineoplastic agents to tumors and delivering drugs toinaccessible organs like brain.

We have been studying the catheter based local deliveryof poly-D,L-lactide- co-glycolide (PLGA) nanoparticles forinhibiting restenosis. In restenosis or the reobstruction of anartery following an interventional procedure, the pathology

Table 1. Drug Targeting Approaches

PASSIVE TARGETING ACTIVE TARGETING

1. Pathophysiological factors 1. Biochemical targets

• Inflammation/infection • Organs

• EPR effect • Cellular

2. Physicochemical factors • Organelles

• Size • Intracellular

• Molecular weight 2. Physical/External stimuli

3. Anatomical opportunities • Ultrasound

• Catheterization • Magnetic field

• Direct injection 3. Pretargeting/Sandwich targeting

4. Chemical approaches 4. Promoter/Transcriptional targeting

• Prodrugs

• Chemical delivery systems


is limited to the localized area of endothelial injury followingangioplasty procedures. We have demonstrated nanoparticlelocalization in arterial tissue after injecting a suspension atthe site of injured artery using a cardiac infusion catheter(Fig. 2). Nanoparticles, due to their colloidal size, canpenetrate the arterial wall through the disrupted epitheliumand are immobilized in the arterial wall [14]. This allowssustained levels of the pharmacological agent at the localizedsite. Using the above strategy, we have demonstrated 35%inhibition of restenosis in a rat carotid model with a singlelocalized intraluminal delivery of dexamethasone-loadednanoparticles. Dexamethasone levels were sustained formore than one week in the infused section of artery and notin the adjacent carotid artery; presenting nanoparticles aslocalized, sustained drug delivery approach for restenosis.

Blood brain barrier (BBB), which is formed by the tightjunctions within the capillary endothelium of the brain,constitutes a formidable barrier to the CNS delivery oftherapeutic agents [15]. Although selective transportmechanisms are present in the BBB such as diffusion,carrier-mediated transport, receptor-mediated, adsorptive,

and fluid-phase endocytosis, the transport of therapeuticagents via systemic administration to the brain is limited [16,17]. To overcome these challenges, a promising approachcould be the direct injection of therapeutic agent or ananosystem to the brain tissue using microinjectiontechniques. This approach could be more effective intargeted delivery to a certain area of the brain, howeverwould not be as effective if the whole brain is the target fordrug therapy. This is because of poor diffusion of therapeuticagents through the brain tissue. To improve diffusion of theinjected drug, some of the strategies include conjugating thetherapeutic proteins with non-toxic neuronal binding domainof tetanus toxin (tetanus toxin fragment C-TTC). Proteinslinked to TTC not only show a better retention but also asuperior distribution in the extracellular space on directinjection into brain [18]. In our studies, we are exploring thestrategy of directly injecting nanoparticles into the brain(Fig. 3). The drug encapsulated into nanoparticles will beprotected and also will be released slowly, thus can beeffective in providing targeted and sustained drug delivery.In a preliminary study, we have demonstrated the feasibility

Fig. (2). Angiogram demonstrating infusion of nanoparticles in the target artery following angioplasty using a cardiac infusion catheter in aporcine coronary model of restenosis. LAD: Left Anterior Descending coronary artery.

Fig. (3). Fluorescent PLGA nanoparticles were injected stereotactically into the brain through the striatum of the rat and the animals wereallowed to survive for 3 days post injection. Fig 3A is the digital camera view of a frozen section of the brain showing injection site and Fig3B is the fluorescent low-magnification image of the injection site. Outlined area in Fig 3A represents the site of injection. Nanoparticleswere found to be spreading from the injection site to surrounding parenchyma. Scale bar in Fig 3A is 1 mm and in Fig 3B 100 µm.


of the above approach and the injected nanoparticles are seento distribute through the brain parenchyma. It would beinteresting to determine in the future studies, how effectiveare these nanoparticles in drug distribution in the brain andsustaining the drug retention and effect.

3.1.3. Physicochemical Factors

The clearance kinetics and in vivo biodistribution ofnanosystems depend on the physicochemical factors likesize, surface charge and surface hydrophobicity and can bemanipulated to enable passive targeting [19]. A major part(~90%) of the nanosystems injected intravenously generallyis lost to the reticulo-endothelial system (RES), mainly fixedmacrophages in the liver and spleen after opsonization byproteins present in the blood stream [20]. Particles < 100 nmcan pass through the fenestrations in the liver endotheliumand the sieve plates of sinusoids to localize in the spleen andbone marrow. This natural tendency of nanosystems tolocalize in the RES presents an excellent opportunity forpassive targeting of drugs to the macrophages present in theliver and the spleen. It has been utilized for targetingantibiotics and antiviral agents for intracellular infections.Nanoparticles are promising drug carrier systems forantiparasitic drugs used in the treatment of leishmaniasiscaused by an intracellular parasite- Leishmania donovani.The binding of anti-leishmanial drugs to various kinds ofnanoparticles was found to increase their therapeutic index(by 5 times) due to passive targeting of the reticulo-endothelial system [21]. Acute lethal toxicity studies,performed in healthy mice after intravenous injection of freeprimaquine and primaquine-loaded poly-lactic acid (PLA)nanoparticles, reported a two folds increase in the lethal dose(LD50) with the nanoparticle formulation. Rapid clearance ofthe drug from the blood stream due to the RES uptakerestricted the drug accumulation to liver and spleen only andthus was responsible for reducing the lethal toxicity of drug.

Polymeric nanoparticles have also demonstratedsignificant promise for anti-HIV therapy by delivering theanti-HIV drugs with poor selectivity and short plasma half-lives into the phagocytic cells. Polyacrylcyanoacrylate(PACA) nanoparticles loaded with Saquinavir were found tobe effective in HIV- infected human macrophage culture[22]. Large sized liposomes especially multi-lamellarvesicles (MLV) are cleared from the blood circulation by theRES system and have been explored to replace geneticallydeficient enzymes in these cells and to target antimicrobialagents to intracellular pathogens.

Although the RES uptake of the colloidal sizednanosystems holds promise for passive targeting of manydrugs; it becomes a major hurdle for drugs whose site ofaction is located in tissues other than the RES. A variety ofmethods have been explored to overcome the rapidopsonization and the subsequent RES uptake of nanosystemsand to make them long circulating for effective targeting ofdrugs to tissues other than the RES.

3.2. Active Targeting Approaches

Passive targeting approaches are limited in their scopeand thus, tremendous effort has been directed towards thedevelopment of active approaches for drug targeting. Active

targeting employs specific modification of a drug/drug-carrier nanosystems with “active” agents having selectiveaffinity for recognizing and interacting with a specific cell,tissue or organ in the body [23]. Direct coupling of drugs totargeting ligand, restricts the coupling capacity to a few drugmolecules. In contrast, coupling of drug carrier nanosystemsto ligands allows import of thousands of drug molecules bymeans of one receptor targeted ligand.

Drug targeting to specific cells has been exploredutilizing the presence of various receptors, antigens/proteinson the plasma membrane of cells and also by virtue of thelipid components of the cell membranes. The receptors andsurface bound antigens may be expressed uniquely indiseased cells only or may exhibit differentially higherexpression in diseased cells as compared to the normal cells.Active agents, such as ligands for the receptors andantibodies to the surface proteins have been used extensivelyto target specific cells. The following section gives a briefoverview of the targets and the ligands which can be used asactive targeting moieties to target nanosystems to specificcells in the body.

3.2.1. Targets

(A) Receptors

The presence of receptors on cell membranes potentiatesactive targeting by not only allowing specific interaction ofdrug carrier system with cells but also facilitating its uptakevia receptor mediated endocytosis [24].

(a) Folic Acid Receptor

Folic acid - a vitamin essential for de novo nucleotidesynthesis is taken up by cells via receptor mediatedendocytosis using membrane associated folate receptors.Folate receptors are expressed only on certain epithelial cellsin humans, but are differentially over-expressed in cells ofcancers with epithelial origin. It has been primarily used fortumor specific drug delivery in many cancers includingbreast, ovary, brain and lung malignancies [25].

(b) LDL Receptors

Low density lipoprotein (LDL) receptors are a family ofnine endocytic receptors that transport cholesterol richlipoproteins (LDL) into cells via receptor mediatedendocytosis. Many drugs and some lipid-based systems suchas liposomes and cholesterol rich emulsions have beenproposed to interact with these receptors [26]. Anionicliposomes [27] and apolipoproteinE (apoE) enrichedliposomes [28] were found to mimic LDL and provided sitespecific delivery of antitumor agents to cancer cells via theLDL receptors. B16 melanoma cell line showing higherexpression of LDL receptors was used to assess the specificbinding affinity of apoE labeled liposomes in vitro.Moreover, increased uptake of liposomes was observed inliver and adrenals- organs showing relatively higherexpression of LDL receptors in normal rats pretreated with17-α-ethinyl estradiol (used to selectively upregulate LDLreceptors on liver and adrenals) (Fig. 4). This approach,however, is not very specific as LDL receptors are expressedon cells of almost all organs.


(c) Peptide Receptors

A large number of peptide receptors are expressed inlarge quantities in certain tumor cells. Receptors for peptidessuch as somatostatin analogs, vasoactive intestinal peptide,gastrin related peptides, cholecystokinin, leutanising hor-mone releasing hormone have been localized on tumor cells[29]. Peptides/peptide analogs can be conjugated to a drugcarrier system to allow tumor specific targeting of cytotoxicagents, following interaction with peptide receptors.

(B) Lipid Components of Cell Membranes

Lipid components of cellular membranes are emerging asnovel targets for antineoplastic drugs [30]. Interaction ofsynthetic phospholipid analogs with cellular membraneschanges the lipid composition, membrane permeability andfluidity, thereby influencing signal transduction mechanismsand inducing apoptotic cell death. Two such phospholipidanalogs- Edelfosine and Miltefosine can selectively killmalignant cells and thus offer promising approaches incancer chemotherapy.

(C) Surface Antigens/Proteins

Expression of different proteins on the surface of cellsconstitutes the biochemical writing on the cells, which canbe deciphered using monoclonal antibodies against theseproteins. The diseased cells may express either new proteinsor may exhibit differential (under/over) expression of theproteins found on normal cells. The advent of Proteomicstechnology has brought a revolution in the field ofidentification and validation of tumor specific antigens. Anideal tumor specific antigen, which can be used for drug

targeting, would be the one expressed exclusively by tumorcells and similarly by all cells present in the tumor.

Prostate specific antigen, differentially expressed onprostate cancer cells, and erbB2- a growth factor that is overexpressed in 20-30% human breast adenocarcinomas, areexamples of tumor specific antigens. erbB2, expressed on thetumor cell surface is readily accessible, has low expressionon normal cells and shows a homogeneous distributionwithin the tumor. It has been used for immunotherapy withliposomal doxorubicin formulation coupled to anti-erbB2-antibody [31]. A significant increase in the in vivo efficacywas observed with anti-erbB2 immunoliposomes ascompared to the unmodified liposomes and saline control ina human erbB2-overexpressing breast cancer model (Fig. 5).Immunotherapy of a non-uniformly distributed tumorantigen in a tumor mass, may result in emergence of resistantcells- which do not express the specific antigens. Tumorspecific antigens are the emerging and promising targets forsuccessful and exquisitely specific tumor therapy. But,clinical use of such targets requires appropriate validation oftargets in human tumor tissues to ensure that these antigensare specific to tumor cells and are not shed into thecirculation.

3.2.2. Targeting Ligands

Active targeting of drugs can be realized by the use ofactive agents or ligands which interact with some degree ofexclusivity with the specific targets/receptors identified onparticular cell types. Targeting ligands employed for activelytargeted drug delivery include simple molecules like sugars,folic acid, peptides and specifically engineered antibodies.

Fig. (4). Organ distribution of apoE-enriched liposomes in control and 17 α-EE-pretreated rats. [ 3H]CO-labeled liposomes (1.0 mg ofphospholipid) were injected after previous incubation (30 min at 37°) with 100 mg of apoE into anesthetized propylene glycol treated rats (�)versus 17α-EE-treated rats ( ). After 120 min of circulation, the total ( left) and specific (i.e., per gram wet weight) ( right) organ distri-butions were determined. Recoveries of 3H-radioactivity in the rats exceeded 95%. Values are corrected for serum radioactivity and representmean ± standard error from three experiments. Reprinted with permission from ref [28].


(A) Folic Acid

Significant over expression of folate receptors on avariety of tumor cells allows targeting using folic acid as atargeting moiety. Folate has been conjugated to radiolabelleddendrimers and to liposomes to effect preferential uptake andaccumulation of therapeutic agents in tumor cells [32, 33].Induction of folate receptors with retinoid compoundsfollowed by folate conjugated liposomes has been usedsuccessfully for therapy of acute myelogenous leukemia.This treatment modality was also found to bypass the P-glycoprotein (Pgp) mediated drug efflux mechanisms [34].(B) Sugars

Presence of endogenous lectins on the epithelial cells ofdifferent gut regions can be used for drug targeting by usingsugars as the targeting moieties. Galactose shows greaterinteraction in proximal regions of gut while fucose binds tothe distal positions. Asialoglycoprotein receptor expressedon the surface of hepatocytes is another example ofendogenous lectin, useful for drug targeting by galactosebearing drug carrier system [35].

(C) Lectins

Different cell types, particularly diseased cells expressdifferent glycans on their surfaces in comparison to theirnormal counterparts. Thus, selection of suitable lectinsallows targeting of drugs to specific cells [35]. Anoutstanding example of such an interaction is that of wheatgerm agglutinin (WGA) with primary endothelial cells of

porcine brain. High affinity of WGA for the brain endotheliaallows drug targeting by binding to cell surface andsubsequent internalization without disrupting the barrierproperties of brain endothelia [36, 37].

(D) Modified Albumins

Cationized albumin (CBSA) coupled to stericallystabilized liposomes has been demonstrated as a promisingtargeting ligand for brain specific targeting [38]. CBSAcoupled liposomes showed better interaction with brainendothelial cells and higher intracellular accumulation ascompared to bovine serum albumin (BSA). It was concludedthat cationized albumin is taken up into the brain endotheliavia a caveolae mediated endocytic pathway. Sugar modifiedalbumins have been proposed as suitable carriers fortargeting drugs selectively to various cells in liver [39].

(E) Peptides

Peptides like vascular endothelial growth factor (VEGF),vasoactive intestinal peptide (VIP), leutanising hormonereleasing hormone (LHRH), somatostatin etc. are used fortargeting drugs to their specific receptors which areexpressed abundantly in different disease states. Proteindelivery to the brain has been achieved successfullyfollowing conjugation of protein to TAT peptide [40]. Inanother approach, non-toxic binding fragment of tetanustoxin has been used to improve CNS delivery of proteins bysystemic administration [41].

(F) Antibodies

The discovery of hybridoma technology has heralded anextensive use of antibodies as targeting moieties, againstspecific antigens/proteins expressed on the surface of cells.Intact antibodies have been used as highly specific targetingagents with a high affinity towards their targets. These canbe conjugated either directly to the drugs (immuno-therapeutics) or to the nano-sized delivery systems (e.g.,immunoliposomes, immunomicelles, etc.) to effect drugtargeting. However, the use of mouse monoclonal antibodies(mAbs) in humans presents some problems. Their in vivo usein humans is restricted for repeated administration, due to thedevelopment of immune response against murine antibodydomains, producing human anti-mouse antibody (HAMA).Moreover, the success of complete mAbs in solid tumors islimited by their high molecular weight and poor penetrationpower which results in non-uniform distribution in the solidtumors. Solutions to these problems lie in the recombinanttechnology which has revolutionized the field of ‘engineeredantibodies’ for human use (Fig. 6). Newer strategies toovercome these problems include fusion of mouse variableregions to human constant regions (chimeric antibodies),removal of T cell epitopes (de-immunization) and graftingmouse antigen binding regions onto human acceptorantibody frameworks (humanization of antibodies) [42].These three strategies to generate humanized mAbs canavoid the problems due to host anti- antibody reactions.Further engineering of antibodies to generate smaller buteffective antibody fragments has allowed a successful use ofthe immuno-therapies for solid tumors. Novel antibodyformats include the monovalent F ab fragments, single chain

Fig. (5). Efficacy of anti-erbB2 immunoliposomes in a humanerbB2- overexpressing breast cancer model. Anti-erbB2immunoliposome-Doxil containing the F5 scFv (triangles)administered on days 21, 28, and 35 (arrows, same dose as above)are compared to control (PBS) treatment (open circles). Datarepresent mean tumor volumes; mm 3 ± S.E. Reprinted withpermission from ref [31], Copyright (2002), with permission fromElsevier.


Fv fragments (scFv) and the bispecific antibodies. A F abfragment takes one fourth of the time taken by an intactimmunoglobin molecule to move a distance of 1mm into asolid tumor [43]. Thus smaller antibody fragments exhibitbetter pharmacokinetics for tissue penetration with fullbinding specificity but show poor retention time at the targetsite. F ab fragment of anti HER-2 monoclonal antibodyconjugated to liposomes was shown to bind specifically tobreast cancer cell line SK-BR-3 which overexpressed HER-2[44]. A scFv fragment against the human transferrin receptorwas used to target cationic lipoplexes to breast tumor cellsfor p53 gene therapy upon systemic administration. Theseimmunolipoplexes demonstrated increased tumor cellbinding, improved gene delivery and transfection efficiencyin vitro and in vivo [45].

4. NANOSYSTEMS

Better appreciation of biology, advances in polymerchemistry and their interaction with nanotechnology hasbrought a renaissance in the field of drug delivery, wherebynano-sized drug delivery systems are emerging as a panaceafor the global quest of drug targeting. A plethora of highlyoptimized methods for the preparation, versatility of

formulation, availability of polymers and newer charac-terization tools have tremendously expanded the researchinterests in such nanosystems. For the purposes of thisreview, the term “nanosystems” shall include all the nano-sized (<1000 nm) drug carriers systems, such as polymericnanoparticles, liposomes, micelles and polymer-drugconjugates. Some of the newer nanosystems that are beingdeveloped include nanocages, nanogels, nanofibers, nano-shells, nanorods, nanocontainers, etc. As new systems andmolecular complexes are being developed, the existing drugdelivery systems are undergoing transformation to meet theneeds of effective drug targeting. For example, liposomesfrom simple lipid vesicles are modified to stealth liposomesto more complex systems with specific ligands and peptidesto achieve cellular and tissue targeted drug therapy. Also, theintegrated approaches to drug delivery by scientists fromdifferent disciplines have created new opportunities forapplications of nanosystems in medicine and drug therapy.Submicron size range of nanosystems offers excellentopportunities to breach the physiological barriers and accessdifferent tissues followed by an efficient cellular uptake andintracellular internalization. This is not only being exploredfor the purposes of drug delivery but widely accepted andemployed as imaging tools. Moreover, these nanosystems

Fig. (6). Schematic representation of an intact Ig together with Fab and Fv fragments and single V (colored ovals; dots represent antigen-binding sites) and C domains (uncolored). Engineered recombinant antibodies are shown as scFv monomers, dimers (diabodies), trimers(triabodies) and tetramers (tetrabodies), with linkers represented by a black line. Minibodies are shown as two scFv modules joined by two Cdomains. Also shown are Fab dimers (conjugates by adhesive polypeptide or protein domains) and Fab trimers (chemically conjugated).Colors denote different specificities for the bispecific scFv dimers (diabodies) and Fab dimers and trimers. Reprinted with permission fromref [42].


have the potential to effectively carry and target drugs byvirtue of their colloidal size range and ability to becoupled/conjugated with different targeting ligands.

4.1. Liposomes

Liposomes have been extensively investigated as apotential drug delivery system due to the enormous diversityof structure and compositions that can be achieved [5]. Theyare capable of targeting drugs by passive as well as activemeans. Unmodified liposomes undergo rapid clearance fromblood stream due to sequestration by macrophages of theRES. This property and the flexibility of altering size ofliposomes have allowed its use in passive targeting of anumber of drugs. Liposomes can be used to deliverimmunomodulators, cytotoxic and anti-microbial agents tothe macrophages by means of passive targeting. Activatingfactors such as cytokines have been delivered tomacrophages to render them tumoricidal. Such activatedmacrophages selectively kill tumor cells and this approachhas been proposed for treatment of metastasis after surgicalremoval of primary tumors. The EPR effect in tumorsproposed by Maeda has been successfully employed toselectively deliver cytotoxic agents to tumor cells by meansof appropriately sized liposomes. The rapid uptake andaccumulation of liposomes in tumor cells has been proposedto be a reason for reduced plasma levels of the anthracyclinecompounds and thus reduced chances of myocardial toxicity.Two such formulations are currently commercially available-Daunosome™ liposome (NeXstar, Inc.) incorporatesdaunorubicin and Doxil™ (Sequus Pharmaceuticals) basedon doxorubicin. These formulations have an extendedcirculation time by virtue of their small size and stericallystabilized surfaces [46].

Selective uptake of unmodified liposomes by thephagocytic cells has also been exploited for intracellulardelivery of antimicrobial agents, thereby increasing theirtherapeutic index and decreasing toxicity substantially. Thishas been used for intracellular pathogens such as Leishmaniadonovani, systemic fungal infections, HIV virus andmycobacterial infections, particularly in immunocom-promised patients [5]. The natural tendency of liposomes tobe cleared by the RES has widened the scope of theirapplications for imaging as well. Liposomes have been usedas carriers for radioisotopes and contrast agents to be used indiagnostic imaging [47]. Liposomes carrying contrast agentsaccumulate more in surrounding cells than in tumor cells dueto the low phagocytic activity of tumor cells. This can beused for tumor imaging, wherein tumor cells are observed asholes within the surrounding background. Localized sitespecific delivery of drugs is possible by introducing drugcarrying liposomes by means of catheters. Selectivetransfection of a plasmid DNA associated with liposomeswas observed in vivo after injecting into an isolated region ofartery by a catheter [48]. This approach is applicable fordiseases such as atherosclerosis or neointimal hyperplasia totreat particular sections of the tissues. Liposomes can also belocalized to a particular target tissue by external applicationof localized heat and use of temperature sensitive liposomes[49]. Phospholipids, with a transition temperature slightlyhigher than body temperature are used in this design. Suchformulations show increased permeability at temperatures

near to transition temperature and can be induced to deliverdrugs to target tissues, for example, tumors on externalapplication of heat. Relatively inaccessible inflammationsites (e.g. brain) have been accessed by the use of negativelycharged and arginine-glycine-aspartic acid (RGD) sequencecoated magnetic liposomes, which mediate delivery of drugsthrough monocytes/neutrophils [50].

“Stealth” or long-circulating liposomes designed to evadedetection by the RES, have found significant applicationspertinent to ligand mediated targeted drug delivery [51]. Thiscan be attained by producing a hydrophilic surface bygrafting polyethylene glycol (PEG) chains on the liposomesurface or by inclusion of phosphatidylinositol andgangliosides in the formulation. These sterically stabilizedliposomes act as circulating drug reservoirs and enabletargeting of drugs to non-RES target sites [52]. Ligandmediated targeting has been avidly explored with such longcirculating liposomes but still no products are available onmarket. The concept of antibody targeted liposomes orimmunoliposomes has been existing for quite a few decadesnow. Antibodies can be conjugated to the PEG chains onPEG stabilized liposomes to achieve specific tissue targeting.Accumulation of liposomes at the target site doesn’tessentially guarantee therapeutic efficacy of the encapsulateddrug. Cellular uptake of the liposomes/ encapsulated drugcan occur either by endocytotic pathway or by fusion ofliposomal surfaces with the cell membrane. Receptormediated endocytosis has been shown to be responsible forinternalization of liposomes coupled to an anti-transferrinreceptor antibody at the BBB.

Liposomes have also been actively pursued as a potentialtool for targeted gene delivery to specific cells in the body.Cationic liposomes can be complexed with polyanionicplasmid DNA to form highly compact nanostructures with anet positive charge called the “lipoplexes”. This non-viralgene delivery system was designed for gene targeting to liverby conjugating modified galactolipids to the lipoplex.Labeling of lipoplexes with asialofetuin and protaminesulfate has documented improved in vivo efficiency due toefficient targeting to nucleus by the nuclear localizationsignal present in protamine sulfate [53]. The addition ofprotamine sulfate to asialofetuin-lipoplexes increased the invivo gene expression levels by 75 folds as compared to thoseby conventional lipoplexes (Fig. 7). This was attributed tothe smaller size of the lipoplexes produced in combinationwith protamine and the possible nuclear targeting by nuclearlocalization signals present in protamine. Despite thesignificant progress made so far in the formulationtechnology and the new methods of ligand conjugation, thereare still numerous lacunas to be filled before liposomes canbe projected as the ultimate nanosystems for targeted drugdelivery. Major obstacles include gene transfer to the nucleusof a cell and obtaining a sustained long term expression ofthe genes.

4.2. Nanoparticles

Nanoparticles are sub-micron sized polymeric colloidalparticles with a therapeutic agent of interest encapsulatedwithin their polymeric matrix or adsorbed or conjugated ontothe surface. Polymeric nanoparticles constitute a versatile


drug delivery system, which can potentially overcomephysiological barriers, and guide the drugs to specific cellsor intracellular compartments; either by passive or ligandmediated targeting approaches. It also allows controlling therelease pattern of drug and sustaining drug levels for a longtime by appropriately selecting the polymeric carriers.

Like all colloidal drug carrier systems, nanoparticles arerapidly opsonized and cleared by the macrophages of RES.Thus, intravenously injected nanoparticles are mostly (90%)taken up by the liver and the spleen within minutes followingdrug administration [20]. This characteristic biodistributionis of special benefit for targeting drugs to macrophages and

Fig. (7). Histologic evaluation of GFP gene expression in hepatocytes after tail vein injection of PBS and naked DNA (a,b), plain (c,d) andprotamine-AF-lipoplexes (e,f). Cells under phase contrast microscopy (a,c,e) and under fluorescence microscopy (b,d,f) at an originalmagnification of X 200. A representative overview is shown. Reprinted with permission from ref [53].


has been employed for chemotherapy of the RES localizedtumors. Anticancer drug loaded nanoparticles are mainlyconcentrated in Kupffer cells in liver on intravenousinjection. These cells act as reservoirs and allow prolongeddiffusion of anticancer drug into the neighboring tumor cells.This approach of passive targeting has been useful for thetreatment of hepatic metastasis [54].

The RES uptake of nanoparticles depends on the particlesize, surface charge and surface hydrophobicity. Particles<100 nm, with hydrophilic surfaces undergo relatively lessopsonization and clearance by RES uptake. It prolongs thecirculation time of nanoparticles in the blood and allowstargeting of nanoparticles to tissues other than the RES. Suchlong circulating nanoparticles have been designed as‘biomimetic nanoparticles’ or ‘sterically stabilized nano-particles’ [55-57]. The former approach involves preventingthe recognition of surface of nanoparticles by the RES bycoating the nanoparticles with biomimetic substances such assialic acid, albumin or inactivated complement moieties.This approach has not been successful due to rapiddesorption of the coatings in vivo. Steric stabilization ofnanoparticles has been achieved by making the surface ofparticles more hydrophilic so as to prevent opsonization inthe blood stream. This is done either by adsorbinghydrophilic surfactants on nanoparticle surface or usingblock/branched copolymers [58]. Polyethylene oxide (PEO)or poly ethylene glycol (PEG) is the most successful non-ionic hydrophilic polymer used for this purpose [58, 59].

Overcoating of nanoparticles with polysorbates is knownto result in greater transport of nanoparticles across the BBB,allowing brain targeting following an intravenous injection[60]. Polysorbates on the surface of nanoparticles act asanchors for apolipoprotein E (present in plasma), whichadsorbs on the nanoparticles. The particles then mimic LDLparticles and interact with LDL receptors on endothelial cellsin the brain capillaries leading to their cellular uptake.Kreuter’s group has enlisted a number of possiblemechanisms for nanoparticle mediated drug transport acrossthe BBB. These include endocytotic/transcytotic pathwaywith a possibility of inhibition of P-glycoprotein effluxmechanisms [60]. Lymphatic targeting of drugs is possibleby a sub-cutaneous injection of nanoparticles. This can beused as a tool for chemotherapy against lymphatic tumors ormetastasis [23].

Nanoparticles offer numerous opportunities for targetingby surface modifications which would allow specific bio-chemical interactions with the proteins/receptors expressedon target cells. We have recently demonstrated increasedefficacy of paclitaxel loaded nanoparticles on conjugationwith transferrin in a murine model of prostate cancer [61].Transferrin receptors are over expressed in most cancer cellsby 2-10 folds than in normal cells. Conjugation of transferrinligand to nanosystems or the antibodies against transferrinreceptor has been explored for targeting drugs to tumor cells.We have demonstrated that the transferrin conjugation tonanoparticles enhances the therapeutic efficacy of theencapsulated paclitaxel as compared to that withunconjugated nanoparticles. The cellular uptake ofconjugated nanoparticles was about 3 times more than that ofunconjugated nanoparticles in a prostate cancer (PC3) cell

line. IC 50 values for paclitaxel were 5 folds lower withtransferrin conjugated nanoparticles than that with solutionor unconjugated nanoparticles. A complete regression oftumor and a significantly higher survival rate of animals wasobserved with a single-dose intra-tumoral injection oftransferrin conjugated nanoparticles in a murine model ofprostate cancer as compared to that with unconjugatednanoparticles or drug dissolved in Cremophor ® EL (Fig. 8).The mechanism of greater efficacy of transferrin conjugatednanoparticles appears to be due to the greater cellular uptakeof drug. Thus, nanoparticles are emerging as a promisingtool for the intracellular delivery of practically insolubledrugs (such as taxol) and sensitive drugs (such as proteinsand oligonucleotides or DNA) [62]. Nanoparticles not onlyact as an effective carrier for such drugs protecting themfrom undesirable physiological conditions but also allow aselective and controlled release of drugs at their target sites.

They are of specific importance in cancer chemotherapy,where cellular uptake of drugs is limited due to efflux ofdrug by P-glycoprotein and drug resistance is a commonproblem. Anticancer drugs encapsulated in nanoparticlescannot be recognized by these efflux mechanisms and thuscircumvent multidrug resistance [63].

The versatility of formulation, colloidal size, bio-compatibility and sustained release properties of nano-particles have already been accepted with growing interestfor a wide range of applications. However, the recentstrategy gaining widespread interest and attention is that of“functionalized nanoparticles”. Methods are being devised totailor make the surface characteristics of nanoparticles toachieve specific ligand mediated targeting of therapeutic andimaging agents. Optimization of these techniques, coupledwith the considerable progress made in the field ofnanoparticle characterization and a better understanding oftheir in vivo behavior has raised hopes for the successfuldevelopment of commercial products based on targetednanoparticles for use in therapy and imaging.

4.3. Polymeric Micelles

Amphiphilic copolymers self-assemble in an aqueousmilieu to form spherical colloidal nanoparticles calledpolymeric micelles [64]. Polymeric micelles have beenproposed as novel carrier systems in drug targeting due totheir increased loading capacities, stability in physiologicalconditions and the possibility of engineering the core-shellarchitecture for different applications [65, 66]. The size ofpolymeric micelles (usually <100 nm) and their hydrophilicsurface is of crucial importance in overcoming the majorobstacle of drug targeting- the RES uptake. This uniquedisposition characteristic can be used for effective tumortargeting by means of the EPR effect [67]. Functionalizationof polymeric micelles enables ligand-targeted systems to bedeveloped for cell-specific targeting. A pre-requisite for thedevelopment of such actively targeted micelles is aneffective synthetic method for end-functionalization of theamphiphilic polymers and conjugation of ligands to the endsof the hydrophilic segments. A variety of PEG-PLA blockcopolymers having amino/saccharide end functional groupshave been prepared. Such micelles with peptide/sugarmoieties on the surface can be used to target specific peptide


receptors on cell surface for cell specific targeting of drugs[68].

Stability of micelles is another critical factor for effectivedrug delivery in vivo. The micelles must dissociate to releasethe entrapped drug at the target site. Polyion complexes(PIC) micelles were prepared with polyethylene glycol andpoly lysine (PEG-PLys) and oligo-DNA and the cores werereversibly cross linked by oxidation of thiol groups [69].This allowed specific intracellular release of oligo-DNAbecause the disulfide cross-links could be cleaved only in thepresence of strong reducing environment. Immunomicellesprepared by conjugating antitumor antibodies to polymericmicelles were shown to recognize and bind to various cancercells in vitro [70]. Taxol- a water insoluble anticancer drugloaded into these immunomicelles showed higheraccumulation in experimental tumors in mice than non-targeted micelles.

4.4. Polymer-Drug Conjugates

The concept of polymer-drug conjugation was developedas a means of improving cell specificity of low molecularweight anticancer drugs, to effect targeting to tumor cells[71]. Conjugation of a polymeric carrier to low molecularweight drugs brings a drastic change in the pharmacokineticdisposition of drug at both the whole body and the cellularlevels [72]. These conjugates allow passive targeting oftumors via EPR effect and their long circulation time inblood stream. N-(2-hydroxy propyl) methacrylamide(HPMA) copolymer conjugated doxorubicin system was

found to passively target tumors in a rodent model by EPReffect, without showing any signs of immunogenicity orpolymer related toxicity [73]. Ligand directed polymer-drugconjugate can also be prepared to target specific receptors oncellular surfaces. The essential features of such nanosystemsinclude a polymeric carrier, a bio-responsive linker and aspecific ligand. HPMA copolymers have been extensivelyused as the carrier in such systems. Polymer-drug linkersmust be selected such that the conjugate remains intactduring transport to the target site, but allows drug release atappropriate rate on reaching its target. The tetrapeptidyllinker, Gly-Phe-Leu-Gly, which is cleaved by endolysosomalproteases is the most popular linker, which is stable incirculation but cleaves subsequent to its endocytic uptake bycells. pH-sensitive cis-aconityl, hydrazone and acetallinkages are alternatives for polymer-drug conjugation [74].

The only clinically tested polymer-drug conjugatebearing a targeting ligand is the galactosamine directedHPMA-tetrapeptidyl-doxorubicin [75]. The system isdesigned for treatment of liver cancer by targeting theasialoglycoprotein receptor on hepatocytes. Doxorubicinconcentration was found to be 12-50 folds higher (inhepatoma tissue) than could be achieved by free drug.

5. EMERGING TRENDS

5.1. Vascular Targeting

The blood and lymphatic vessels of individual tissues innormal and pathological conditions are biochemically

Fig. (8). Antitumor activity of Tx-NPs-Tf in a murine prostate tumor model. PC3 cells (2 × 106 cells) were implanted s.c. in athymic nudemice. Tumor nodules were allowed to grow to diameter of about 50 mm 3 prior to receiving different formulations as a single-dose treatment.Tx-NPs-Tf (l, 24 mg/kg; ¨, 12 mg/kg), Tx-NPs ( Χ, 24 mg/kg), Tx-Cremophor ® EL formulation ( s, 24 mg/kg), (u) control NPs and ( n)Cremophor® EL formulation. Data are means + s.e.m., n=6. * p <0.005 Tx-NPs-Tf versus Tx-NPs and Tx-Cremophor ® EL groups. Reprintedwith permission from ref [61], Copyright (2004) Wiley-Liss, Inc., A Wiley Company.


distinct from each other. The vasculature, especially theendothelial cells express molecules that are specific to aparticular tissue and can be utilized for targeting drugs.Vascular targeting of drugs is thus highly gaining importancedue to these biochemical differences and the easyaccessibility by all drugs absorbed or injected in the bloodstream [76]. Thus, recently the focus has shifted to identifyand validate such biochemical markers in the normal andpathological tissues, especially in tumor tissues. Tumorvasculature is distinct from normal vasculature due to theprocess of neovascularization which is accompanied byexpression of many new proteins or over expression of manynormal proteins on the cell surfaces. Molecular markers ofangiogenesis include additional peptidases/proteases andintegrins. Cell adhesion receptors, integrins are also overexpressed in tumor vasculatures [77, 78]. One such moleculeis the integrin- αvβ3 which is preferentially expressed inangiogenic endothelium of tumors and is a potential targetfor drugs. αvβ3 has been used as a targeting ligand coupledwith lipid based cationic nanoparticles to affect genedelivery targeted to neo vascularization in tumor bearingmice [79]. A group of tumor bearing mice was injected withnon-targeted nanoparticles, other one with targeted nano-particles and third one received targeted nanoparticles co-injected with 20-fold molar excess of the soluble αvβ3targeting ligand. The specificity of gene delivery using thisapproach was evaluated by quantifying the gene expressionlevels in various tissues in tumor bearing mice (Fig. 9).Tumor specific gene expression was completely blockedwhen mice were co-injected with excess of the targetingligand.

Recently Oh and others have described a novel approachfor identifying and validating distinct molecular signatures in

normal and neoplastic tissues [80]. Using integratedanalytical techniques such as subtractive proteomics, bio-informatics and molecular imaging, they identified newmolecular targets on endothelial cells of lung and lung tumortissues. They validated that aminopeptidase-P and annexin-A1 are distinctly present on the endothelial cells of lungs andlung tumors, respectively. These novel accessible tissuespecific targets can be used for molecular imaging andtargeted delivery of drugs/genes in vivo. Specific molecularmarkers have also been identified in inflammatory lesionssuch as arthritic synovium and atherosclerotic plaques [81].5.2. Intracellular Targeting

Intracellular targeting is desirable for drugs whose site ofaction is located in the intracellular compartments or whichundergo extensive efflux from the cell by the effluxtransporters such as multidrug resistance proteins (MRP) andP-glycoproteins (Pgp) [82]. Drugs such as glucocorticoids(e.g., dexamathasone) have cytoplasmic receptors, whilemost anticancer drugs (DNA intercalators) have their site ofaction located in the nucleus. Macromolecular drugs such asproteins and plasmid DNA usually have their site of action inthe cytoplasm and nucleus, respectively. The ultimate goal ofgene delivery can be fulfilled only if the transgene is able tolocalize and integrate with the nuclear or mitochondrialDNA. Further, steps need to be taken to ensure the protectionof proteins/plasmid DNA from the nucleases or theendolysosomal enzymes while delivering them into thecytoplasm or targeting to nucleus. A number of drug carriersystems (lipid based) have been explored for specificintracellular delivery of proteins and DNA. Although suchsystems result in relatively higher intracellular deliveryefficiency, they are incapable of maintaining therapeutic

Fig. (9). Athymic WEHI mice were subcutaneously injected with M21-L cells (5 X10 6), and tumors were allowed to grow to ~ 100 mm 3.Mice were then injected i.v. with 450 nmoles of NP electrostatically coupled to 25 µg of plasmid expressing firefly luciferase. One group alsoreceived a coinjection of 20-fold molar excess of the soluble αvβ3-targeting ligand. After 24 hours, mice were killed, tissues were surgicallyremoved, and luciferase activity was quantified. ( Inset) Luciferase expression as a function of DNA dose injected. Each bar represents themean ± SD of five replicates. Reprinted with permission from ref [79]. Copyright [2002] American Association for the Advancement ofScience.


drug concentrations for long times [83]. Thus the focus isnow-a-days shifted to the development of drug deliverysystems which not only target the drug to its site of actionbut also maintain the drug concentrations at therapeuticallyrelevant levels for a sustained period of time. We areinvestigating biodegradable nanoparticles formulated frompoly-(D,L-lactide-co-glycolide) for cytoplasmic delivery ofdrugs and plasmid DNA [84, 85].

We have earlier demonstrated that nanoparticles areinternalized into vascular smooth muscle cells (VSMCs) andvascular endothelial cells by means of endocytosis, partlythrough fluid phase pinocytosis and partly through clathrincoated pits [86]. Following cellular uptake, nanoparticles aretransported to primary endosomes and to sorting endosomes.A fraction of nanoparticles can be recycled back to theexterior of cell by means of exocytosis, while the restreaches secondary endosomes. Secondary endosomes fusewith lysosomes to form endo-lysosomes and the efficiencyof cytoplasmic delivery is governed by the ease and rapidityof escape of nanoparticles from endo-lysosome. Weproposed and confirmed that the surface charge reversal ofnanoparticles in the acidic pH of endo-lysosomes wasresponsible for the endo-lysosomal escape of nanoparticles[87]. Thus, surface modification of nanoparticles can beattempted in order to affect targeting of drugs either to endo-lysosomes or to cytoplasm. A rapid escape of nanoparticlesfrom endo-lysosomes would result in efficient delivery ofnanoparticles into the cytoplasm. The therapeutic efficacy ofdrugs with cytoplasmic targets depends on the intracellulardrug levels and their maintenance for sustained period oftime. We have reported efficient cytoplasmic delivery ofdexamethasone and sustained intracellular drug levels byusing biodegradable nanoparticles [84]. VSMCs were treatedwith dexamethasone in solution or entrapped in nanoparticlesfor 3 days, following which no further dose of drug was

added. Nanoparticles could be detected inside the cytoplasmof cells even 14 days after nanoparticle incubation with cells(Fig. 10). Thus, it was hypothesized that nanoparticlesescape endo-lysosomes and then release dexamethasone in asustained manner, while present in the cytoplasm.Intracellular drug levels were maintained for up to 8 days ofstudy and therapeutic efficiency was shown in terms of anti-proliferative activity. Dexamethasone loaded nanoparticlesshowed significantly higher and more sustained inhibition ofcell proliferation as compared to drug solution.

In another study, we have demonstrated sustainedintracellular localization of DNA with nanoparticles ascompared to transient localization of DNA in cellstransfected with naked DNA alone. Breast cancer cells(MDA-MB-435S) transfected with wt-p53 plasmid DNAloaded nanoparticles showed significantly higher and moresustained (7 days) intracellular DNA levels as opposed totransfection with naked DNA and DNA-Lipofectamine™complex [85]. This study has significant implications incancer gene therapy, where sustained gene expression is ofimportance for greater therapeutic benefit. Thus, biode-gradable nanoparticles can not only be used for efficientcytoplasmic delivery of low molecular weight drugs but alsofor macromolecules. Further, ligands can be chemicallycoupled to nanoparticles surface to mediate targeting ofnanoparticles to mitochondria or nucleus of the cell. Variouscell penetrating peptides including synthetic peptides, proteintransduction domains (PTD), and membrane-translocatingsequences (MTS) have been described for the intracellulardelivery of drugs especially macromolecules [88]. In general,these peptides are made up of 30 amino acids, are positivelycharged and have the ability to translocate the cell membraneand deliver drugs into the cytoplasm or the nucleus. HIVprotein derived trans-activating regulatory protein (HIV-TAT) and penetratin are the most extensively investigatedpeptides for intracellular delivery of drugs. Efficient genetransfection has been demonstrated by means of TAT-conjugated liposomes both in vitro and in vivo [89]. Theconjugation of TAT on the surface of liposomes was foundto influence their intracellular uptake. The TAT conjugatedliposomes take a non-endocytic route for cellular entry andthe uptake was observed to be energy independent. However,the mechanism of cell penetration of such peptides is yet tobe elucidated completely and thus the issue of cell specificityby means of these peptides is still unresolved [88].

For drugs with nucleus as the site of action, the nuclearmembrane offers another challenge. Nuclear uptake ofplasmid DNA is the rate limiting step in efficient transfectionand successful gene therapy [90]. Molecules smaller than 40-45kDa and less than 100 nm have been shown to cross thenuclear membrane by passive transport, while othermolecules interact with cytosolic factors to cross the nuclearmembrane through the nuclear pore complexes [91]. Certainpeptide sequences, better known as nuclear localizationsignals (NLS) have been reported to specifically interactwith the cytoplasmic factors which can then target moleculesto the nucleus. NLS are peptides with no general consensussequence, however mostly composed of basic amino acids.NLS present in the SV-40 large T antigen was the first andthe most extensively studied classical NLS [92]. Variousstrategies have been devised to target plasmid DNA to the

Fig. (10). Inhibition of VSMC proliferation with dexamethasone insolution and encapsulated in nanoparticle formulations. Cell growthwas followed by an MTS assay, where absorbance is proportionalto the number of viable cells. Data are means ± the standarddeviation (n= 6). Reprinted with permission from ref [84],Copyright (2004) American Chemical Society.


nucleus of a cell, including conjugation of NLS peptide tothe end of a linear DNA molecule and non covalentmodification of a plasmid DNA with NLS among others[91]. Non-viral vectors for gene delivery (liposomes,nanoparticles) can be surface modified and conjugated to theidentified NLS sequences, to effect nuclear targeting. Inpioneering work, gold nanoparticles coupled with SV-40large T antigen on their surface were shown to efficientlytranslocate into the nucleus, upon microinjection into thecells [93]. A recent study proposed enzymatically digestedlow molecular weight protamine as a non-toxic and efficientgene carrier. Plasmid DNA complexed with low molecularweight protamine was found to efficiently translocate intothe cell and then transferred to the nucleus of cell owing tothe structural similarity of protamine with HIV-TAT peptide[94]. Another evolving concept is that of utilization of theviral proteins for nuclear targeting of molecules or the designof synthetic viruses. However, two or more targetingpeptides may be used to target the nanosystems to specificcells and then translocate the drug to the nucleus. One suchapproach using a combination of cell penetrating and nucleartargeting peptides has been explored with gold nanoparticles[95].

5.3. Promoter/Transcriptional Targeting

It’s a novel approach towards targeted gene therapywhich involves the use of specific tissue, tumor, or inducedpromoters that can limit gene expression to target cells whichexpress a specific transcription factor [96]. This specificallyrestricts the transgene expression in the target tissue.Moreover, it allows the flexibility to regulate the durationand level of expression exogenously by using promoters thatare preferentially activated under certain conditions. An idealtumor specific promoter is the one highly active in tumorcells with little or no activity in normal cells. Promoter fortelomerase gene can be genuinely classified as tumorspecific and is being used to drive transgene expression in avariety of cancer cells [97]. Other examples include α-fetoprotein gene which is transcriptionally silent in adultliver but expressed in fetal and hepatocellular carcinomacells [98]. Furthermore, many aspects of tumormicroenvironment (like specific promoters for endothelialcells and hypoxia) and many exogenous factors (radiation,heat and drugs) can also be used to induce promoters toallow transgene expression specifically in tumor cells [96].This new approach is extensively pursued for cancer genetherapy to target damaging events to tumor cells, whilesparing normal tissues.

5.4. Physical Targeting

In addition to the aforementioned active targetingmethods, targeting and release from a drug delivery systemcan be achieved by applying external stimuli such asultrasound or magnetic fields.

5.4.1. Ultrasound

Ultrasound focused on tumor tissue has been shown totrigger the release of drugs from polymeric micelles only atthe tumor site. This is a new drug delivery modality

providing effective drug targeting by using an externalstimulus. Ultrasound could not only trigger the drug releaseat the target site but also enhanced the cellular uptake ofdrug. Although the precise mechanism of ultrasound affecteddrug targeting is not completely understood, some of theproposed mechanisms include, ultrasonic triggered extra-vasation of polymeric micelles, ultrasonic activated releaseof drug from micelles and ultrasonic enhanced uptake intothe cells [99]. A decrease in tumor volume was demonstratedby focusing ultrasound on the tumors following anintravenous injection of Doxorubicin loaded Pluronic ®

micelles.

5.4.2. Magnetic Field

Magnetic nanoparticles offer exciting new opportunitiestowards developing an effective drug targeting deliverysystem. It is possible to produce, characterize andspecifically tailor the functional properties of magneticnanoparticles for their clinical applications [100]. A potentialbenefit of using magnetic nanoparticles is the use oflocalized magnetic field gradients to attract the particles to achosen site, and the possibility to hold them there until thetherapy is complete and then to remove them. Magneticnanoparticles loaded with a therapeutic agent may beinjected intravenously, and then the blood circulation couldtransport these particles to the region of interest under theinfluence of magnetic field gradient. Alternatively,depending upon the pathological condition (e.g., solidtumors), one could inject these particles directly into theartery leading to the desired site and then use magnetic fieldgradient to localize them into the general area where thetreatment is desired [101]. An effective targeting of magneticnanoparticles via systemic administration would require atargeting magnetic force that would offset the force due tolinear blood-low rates of ~0.05 cm/s in capillaries to 10 cm/sin arteries and 50 cm/s in the aorta [102]. To overcome theabove difficulty, magnetic nanoparticles with highermagnetic moments or magnets that can provide highermagnetic field gradient are being developed.

5.5. Pretargeting or Sandwich Targeting

The basic objective of pretargeting is to decouple thetargeting ligand and the active pharmacological moiety. Itinvolves pretargeting of specific cells with a ligand followedby treatment with a drug linked to another molecule havinghigh affinity for the ligand used for pretargeting. Avidin-biotin is one of the biological pairs employed forpretargeting by Hoya et al. [103]. The proteins of target cells(endothelial cells) were biotinylated by accessing the tissueusing a catheter. The active drug or drug carrier systembound to avidin was then injected intravenously. In vivoexperiments demonstrated that drugs bind to biotinylatedendothelial cells through avidin, without getting cleared byblood flow in the artery. This novel approach for targeteddrug delivery has implications in targeting drugs forintravascular diseases which need continuous presence ofdrug at the target site for a long period of time, for exampleangiogenesis in tumors. Pretargeting has been proposed todeliver targeted radio immunotherapy for treatment ofcancer; so that cell killing high doses of radiation can beselectively delivered to tumor cells with reduced exposure to


normal healthy cells. Mab’s against tumor antigens bound tostreptavidin can be used to target the desired tissue.Radioisotopes conjugated with biotin bind only to the tumorcells through the biotin-streptavidin interaction. This canpotentially reduce the dose limiting toxicities of directlyradio labeled antibodies (such as bone marrow toxicity).However, this strategy is in its infancy and needs furtherevaluation before it can be used in humans.

5.6. Chemical Strategies

The objective of targeting drugs is not just anafterthought but now-a-days it’s implicit in the drugdevelopment process. Chemical approaches towards drugtargeting include the well-known prodrug approach and theemerging chemical delivery systems. Two prodrugtechnologies for targeting cytotoxic compounds selectivelyto tumor cells are undergoing evaluation at present. Theseinclude the antibody directed enzyme prodrug therapy(ADEPT) and the gene directed enzyme prodrug therapy(GDEPT). ADEPT involves antibody targeted against tumorantigen to deliver a non endogenous enzyme specifically totumor cells. The prodrug is administered subsequently andgets converted into an active drug by the enzymatic actiononly in the desired cell population. GDEPT approach issimilar to ADEPT however, here inspite of using enzyme,the cells are transfected with the gene responsible forexpression of the non endogenous enzyme [104]. The gene istargeted to specific cells and expressed intracellularly so thatthe prodrug gets converted into its active form only inselected tissues. One such example is that of suicide genetherapy based on the metabolism of prodrug – ganciclovir bythe transgene herpes simplex thymidine kinase [105]. Thetransgene is used to transfect selected cell populations andsubsequent to its expression in the cells, the prodrug-ganciclovir is administered which gets metabolized to itsactive cytotoxic form upon enzymatic action, thus resultingin killing of only the selected populations. These approacheshave been explored in clinical settings for targeted drugeffect to inhibit cell proliferation in tumors and restenosis.

Chemical delivery systems (CDS) offer novel strategiesto target active drugs to specific sites based on predictableenzymatic transformations. Targeting is attained by means ofchemical design of the system wherein the active drug isconverted into a biologically inert form by attaching atargetor and a modifier moiety to it, via bioremovable links[106]. The targetor is responsible for targeting and providingsite specificity to the system, while the modifier serves as alipophilizer and prevents unwanted metabolism of thesystem. Targeting drugs to brain, anterior segment of eye andlungs has been tried using this approach. CDS undergosequential enzymatic transformations targeting and releasingthe active drug in a controlled manner at the desired siteonly. An interesting application includes targetingneuropeptides to brain using CDS [107]. This is based on theidea that a lipophilic compound can gain access across theBBB and if subsequent to crossing the barrier it getsconverted to a lipid insoluble form, it shall get locked in thebrain, providing effective targeting.

CONCLUDING REMARKS

Nanosystems offer opportunities to achieve drugtargeting with newly discovered disease-specific targets,however challenges remain, primarily because of thecomplexity of the body and the layers of barriers that thesesystems need to overcome to reach to the target. Effectivetargeting would require dual focus, a better understanding ofthe target and a simultaneous development of the targetingsystem. Some of the physical approaches such as by directlyreaching to the accessible targets using a catheter system orusing guiding mechanisms such as a magnetic field gradientcould also be effective in drug targeting. As nanotechnologyfor biomedical applications evolves, the caution is alsoexpressed about their safety and it should be investigatedthoroughly.

ACKNOWLEDGEMENTS

We gratefully acknowledge the financial support fromthe Nebraska Research Initiative and the National Institutesof Health. JKV is supported by a pre-doctoral fellowshipfrom the Department of Defense, US Army MedicalResearch and Materiel Command (DAMD17-02-1-0506).VL acknowledges the contribution of his laboratorymembers and collaborators for some of the data that aredescribed in this review. We would also like to thank Ms.Elaine Payne and Ms. DeAnna Loibl for providingadministrative support.

REFERENCES[1] Garnett, M.C. Adv. Drug Deliv. Rev., 2001, 53, 171-216.[2] Tachibana, R.; Harashima, H.; Shono, M.; Azumano, M.; Niwa,

M.; Futaki, S.; Kiwada, H. Biochem. Biophys. Res. Commun., 1998,251, 538-44.

[3] Illum, L.; Jones, P.D.; Baldwin, R.W.; Davis, S.S. J. Pharmacol.Exp. Ther., 1984, 230, 733-6.

[4] Nobs, L.; Buchegger, F.; Gurny, R.; Allemann, E. J. Pharm. Sci.,2004, 93, 1980-92.

[5] Mathiowitz, E., ed. Encyclopedia of controlled drug delivery. 1999,Wiley: New York.

[6] Edens, H.A.; Levi, B.P.; Jaye, D.L.; Walsh, S.; Reaves, T.A.;Turner, J.R.; Nusrat, A.; Parkos, C.A. J. Immunol., 2002, 169, 476-86.

[7] Schiffelers, R.M.; Storm, G.; Bakker-Woudenberg, I.A.J.M.Pharm. Res., 2001, 18, 780-7.

[8] Allen, T.M.; Cullis, P.R. Science, 2004, 303, 1818-22.[9] Lamprecht, A.; Ubrich, N.; Yamamoto, H.; Schafer, U.; Takeuchi,

H.; Maincent, P.; Kawashima, Y.; Lehr, C.M. J. Pharmacol. Exp.Ther., 2001, 299, 775-81.

[10] Merodio, M.; Irache, J.M.; Eclancher, F.; Mirshahi, M.; Villarroya,H. J. Drug Target, 2000, 8, 289-303.

[11] Matsumura, Y.; Maeda, H. Cancer Res, 1986, 46, 6387-92.[12] Maeda, H.; Wu, J.; Sawa, T.; Matsumura, Y.; Hori, K. J. Control

Release, 2000, 65, 271-84.[13] Uwatoku, T.; Shimokawa, H.; Abe, K.; Matsumoto, Y.; Hattori, T.;

Oi, K.; Matsuda, T.; Kataoka, K.; Takeshita, A. Circ. Res., 2003,92, e62-9.

[14] Guzman, L.A.; Labhasetwar, V.; Song, C.; Jang, Y.; Lincoff, A.M.;Levy, R.; Topol, E.J. Circulation, 1996, 94, 1441-8.

[15] Pardridge, W.M. Adv. Drug Deliv. Rev., 1999, 36, 299-321.[16] Abbott, N.J.; Romero, I.A. Mol. Med. Today, 1996, 2, 106-13.[17] Egleton, R.D.; Davis, T.P. Peptides, 1997, 18, 1431-9.[18] Francis, J.W.; Bastia, E.; Matthews, C.C.; Parks, D.A.;

Schwarzschild, M.A.; Brown, R.H., Jr.; Fishman, P.S. Brain Res.,2004, 1011, 7-13.

[19] Stolnik, S.; Illum, L.; Davis, S.S. Adv. Drug Deliv. Rev., 1995, 16,195-214.

[20] Juliano, R.L. Adv. Drug Deliv. Rev., 1998, 2, 31-54.


[21] Rodrigues Jr., J.M.; Fessi, H.; Bories, C.; Puisieux, F.; Devissaguet,J.-P. Int. J. Pharm., 1995, 126, 253-60.

[22] Bender, A.R.; von Briesen, H.; Kreuter, J.; Duncan, I.B.;Rubsamen-Waigmann, H. Antimicrob. Agents Chemother, 1996,40, 1467-71.

[23] Nishioka, Y.; Yoshino, H. Adv. Drug Deliv. Rev., 2001, 47, 55-64.[24] Rihova, B. Adv. Drug Deliv Rev., 1998, 29, 273-89.[25] Lu, Y.; Low, P.S. Adv. Drug Deliv. Rev., 2002, 54, 675-93.[26] Chung, N.S.; Wasan, K.M. Adv. Drug Deliv. Rev., 2004, 56, 1315-

34.[27] Amin, K.; Ng, K.Y.; Brown, C.S.; Bruno, M.S.; Heath, T.D.

Pharm. Res., 2001, 18, 914-21.[28] Rensen, P.C.; Schiffelers, R.M.; Versluis, A.J.; Bijsterbosch, M.K.;

Van Kuijk-Meuwissen, M.E.; Van Berkel, T.J. Mol. Pharmacol.,1997, 52, 445-55.

[29] Minko, T.; Dharap, S.S.; Pakunlu, R.I.; Wang, Y. Curr. DrugTargets, 2004, 5, 389-406.

[30] Jendrossek, V.; Handrick, R. Curr. Med. Chem. Anti-Canc. Agents,2003, 3, 343-53.

[31] Nielsen, U.B.; Kirpotin, D.B.; Pickering, E.M.; Hong, K.; Park,J.W.; Refaat Shalaby, M.; Shao, Y.; Benz, C.C.; Marks, J.D.Biochim. Biophys. Acta, 2002, 1591, 109-18.

[32] Shukla, S.; Wu, G.; Chatterjee, M.; Yang, W.; Sekido, M.; Diop,L.A.; Muller, R.; Sudimack, J.J.; Lee, R.J.; Barth, R.F.; Tjarks, W.Bioconjug. Chem., 2003, 14, 158-67.

[33] Leamon, C.P.; Cooper, S.R.; Hardee, G.E. Bioconjug. Chem., 2003,14, 738-47.

[34] Ratnam, M.; Hao, H.; Zheng, X.; Wang, H.; Qi, H.; Lee, R.; Pan,X. Expert Opin. Biol. Ther., 2003, 3, 563-74.

[35] Bies, C.; Lehr, C.M.; Woodley, J.F. Adv. Drug Deliv. Rev., 2004,56, 425-35.

[36] Banks, W.A.; Kastin, A.J. J. Neurosci. Res., 1998, 54, 522-9.[37] Fischer, D.; Kissel, T. Eur. J. Pharm. Biopharm., 2001, 52, 1-11.[38] Thole, M.; Nobmanna, S.; Huwyler, J.; Bartmann, A.; Fricker, G. J.

Drug Target, 2002, 10, 337-44.[39] Meijer, D.K.; Beljaars, L.; Molema, G.; Poelstra, K. J. Control

Release, 2001, 72, 157-64.[40] Schwarze, S.R.; Ho, A.; Vocero-Akbani, A.; Dowdy, S.F. Science,

1999, 285, 1569-72.[41] Fishman, P.S.; Savitt, J.M.; Farrand, D.A. J. Neurol. Sci., 1990, 98,

311-25.[42] Hudson, P.J.; Souriau, C. Nat. Med., 2003, 9, 129-34.[43] Jain, R.K.; Baxter, L.T. Cancer Res., 1988, 48, 7022-32.[44] Park, J.W.; Hong, K.; Carter, P.; Asgari, H.; Guo, L.Y.; Keller,

G.A.; Wirth, C.; Shalaby, R.; Kotts, C.; Wood, W.I.; et al. Proc.Natl. Acad. Sci. U S A, 1995, 92, 1327-31.

[45] Xu, L.; Tang, W.H.; Huang, C.C.; Alexander, W.; Xiang, L.M.;Pirollo, K.F.; Rait, A.; Chang, E.H. Mol. Med., 2001, 7, 723-34.

[46] Gregoriadis, G. Trends Biotechnol., 1995, 13, 527-37.[47] Lasic, D.D. Liposomes: from physics to applications, Elsevier:

Amsterdam ; New York, 1993.[48] Nabel, E.G.; Plautz, G.; Nabel, G.J. Science, 1990, 249, 1285-8.[49] Weinstein, J.N.; Magin, R.L.; Cysyk, R.L.; Zaharko, D.S. Cancer

Res., 1980, 40, 1388-95.[50] Jain, S.; Mishra, V.; Singh, P.; Dubey, P.K.; Saraf, D.K.; Vyas, S.P.

Int. J. Pharm., 2003, 261, 43-55.[51] Woodle, M.C.; Lasic, D.D. Biochim. Biophys. Acta, 1992, 1113,

171-99.[52] Lasic, D.D. Trends Biotechnol., 1998, 16, 307-21.[53] Arangoa, M.A.; Duzgunes, N.; Tros de Ilarduya, C. Gene. Ther,

2003, 10, 5-14.[54] Leroux, J.C.; Doelker, E.; Gurny, R. Microencapsulation methods

and industrial applications, Marcel Dekker: New York, 1996.[55] Bazile, D.V.; Ropert, C.; Huve, P.; Verrecchia, T.; Marlard, M.;

Frydman, A.; Veillard, M.; Spenlehauer, G. Biomaterials, 1992, 13,1093-102.

[56] Olivier, J.C.; Huertas, R.; Lee, H.J.; Calon, F.; Pardridge, W.M.Pharm. Res., 2002, 19, 1137-43.

[57] Verrecchia, T.; Spenlehauer, G.; Bazile, D.V.; Murry- Brelier, A.;Archimbaud, Y.; Veillard, M. J. Control Release, 1995, 36, 49-61.

[58] Storm, G.; Belliot, S.O.; Daemen, T.; Lasic, D. Adv. Drug Deliv.Rev., 1995, 17, 31-48.

[59] Lee, J.H.; Kopecek, J.; Andrade, J.D. J. Biomed. Mater. Res., 1989,23, 351-68.

[60] Kreuter, J. Adv. Drug Deliv. Rev., 2001, 47, 65-81.

[61] Sahoo, S.K.; Ma, W.; Labhasetwar, V. Int. J. Cancer, 2004, 112,335-40.

[62] Brigger, I.; Dubernet, C.; Couvreur, P. Adv. Drug Deliv. Rev.,2002, 54, 631-51.

[63] Bennis, S.; Chapey, C.; Couvreur, P.; Robert, J. Eur. J. Cancer,1994, 30A, 89-93.

[64] Kataoka, K.; Harada, A.; Nagasaki, Y. Adv. Drug Deliv. Rev.,2001, 47, 113-31.

[65] Kwon, G.S.; Kataoka, K. Adv. Drug Deliv. Rev., 1995, 16, 295-309.

[66] Kwon, G.S.; Okano, T. Adv. Drug Deliv. Rev., 1996, 21, 107-16.[67] Kwon, G.S.; Suwa, S.; Yokoyama, M.; Okano, T.; Sakurai, Y.;

Kataoka, K. J. Control Release, 1994, 29, 17-23.[68] Yasugi, K.; Nakamura, T.; Nagasaki, Y.; Kato, M.; Kataoka, K.

Macromolecules, 1999, 32, 8024-32.[69] Kakizawa, Y.; Harada, A.; Kataoka, K. J. Am. Chem. Soc., 1999,

121, 11247-8.[70] Torchilin, V.P.; Lukyanov, A.N.; Gao, Z.; Papahadjopoulos-

Sternberg, B. Proc. Natl. Acad. Sci. USA, 2003, 100, 6039-44.[71] Duncan, R. Nat. Rev. Drug Discov., 2003, 2, 347-60.[72] Duncan, R.; Spreafico, F. Clin. Pharmacokinet, 1994, 27, 290-306.[73] Vasey, P.A.; Kaye, S.B.; Morrison, R.; Twelves, C.; Wilson, P.;

Duncan, R.; Thomson, A.H.; Murray, L.S.; Hilditch, T.E.; Murray,T.; Burtles, S.; Fraier, D.; Frigerio, E.; Cassidy, J. Clin. CancerRes., 1999, 5, 83-94.

[74] Etrych, T.; Jelinkova, M.; Rihova, B.; Ulbrich, K. J. ControlRelease, 2001, 73, 89-102.

[75] Seymour, L.W.; Ferry, D.R.; Anderson, D.; Hesslewood, S.;Julyan, P.J.; Poyner, R.; Doran, J.; Young, A.M.; Burtles, S.; Kerr,D.J. J. Clin. Oncol., 2002, 20, 1668-76.

[76] Rouslahti, E. Drug Discov. Today, 2002, 7, 1138-43.[77] Kim, S.; Bell, K.; Mousa, S.A.; Varner, J.A. Am. J. Pathol., 2000,

156, 1345-62.[78] Eliceiri, B.P.; Cheresh, D.A. J. Clin. Invest., 1999, 103, 1227-30.[79] Hood, J.D.; Bednarski, M.; Frausto, R.; Guccione, S.; Reisfeld,

R.A.; Xiang, R.; Cheresh, D.A. Science, 2002, 296, 2404-7.[80] Oh, P.; Li, Y.; Yu, J.; Durr, E.; Krasinska, K.M.; Carver, L.A.;

Testa, J.E.; Schnitzer, J.E. Nature, 2004, 429, 629-35.[81] Houston, P.; Goodman, J.; Lewis, A.; Campbell, C.J.; Braddock,

M. FEBS Lett., 2001, 492, 73-7.[82] Panyam, J.; Labhasetwar, V. Curr. Drug Deliv., 2004, 1, 235-47.[83] Zelphati, O.; Wang, Y.; Kitada, S.; Reed, J.C.; Felgner, P.L.;

Corbeil, J. J. Biol. Chem., 2001, 276, 35103-10.[84] Panyam, J.; Labhasetwar, V. Mol. Pharmaceutics, 2004, 1, 77-84.[85] Prabha, S.; Labhasetwar, V. Mol. Pharmaceutics, 2004, 1, 211-9.[86] Panyam, J.; Labhasetwar, V. Pharm. Res., 2003, 20, 210-8.[87] Panyam, J.; Zhou, W.Z.; Prabha, S.; Sahoo, S.K.; Labhasetwar, V.

Faseb J., 2002, 16, 1217-26.[88] Jarver, P.; Langel, U. Drug Discov. Today, 2004, 9, 395-402.[89] Torchilin, V.P.; Rammohan, R.; Weissig, V.; Levchenko, T.S.

Proc. Natl. Acad. Sci. U S A, 2001, 98, 8786-91.[90] Subramanian, A.; Ranganathan, P.; Diamond, S.L. Nat. Biotechnol,

1999, 17, 873-7.[91] Nakanishi, M.; Akuta, T.; Nagoshi, E.; Eguchi, A.; Mizuguchi, H.;

Senda, T. Eur J. Pharm. Sci., 2001, 13, 17-24.[92] Gorlich, D.; Kutay, U. Annu. Rev. Cell Dev. Biol., 1999, 15, 607-

60.[93] Feldherr, C.M.; Akin, D. J. Cell Biol., 1990, 111, 1-8.[94] Park, Y.J.; Liang, J.F.; Ko, K.S.; Kim, S.W.; Yang, V.C. J. Gene.

Med., 2003, 5, 700-11.[95] Tkachenko, A.G.; Xie, H.; Coleman, D.; Glomm, W.; Ryan, J.;

Anderson, M.F.; Franzen, S.; Feldheim, D.L. J. Am. Chem. Soc.,2003, 125, 4700-1.

[96] Robson, T.; Hirst, D.G. J. Biomed. Biotechnol, 2003, 2003, 110-37.[97] Kim, N.W.; Piatyszek, M.A.; Prowse, K.R.; Harley, C.B.; West,

M.D.; Ho, P.L.; Coviello, G.M.; Wright, W.E.; Weinrich, S.L.;Shay, J.W. Science, 1994, 266, 2011-5.

[98] Ishikawa, H.; Nakata, K.; Mawatari, F.; Ueki, T.; Tsuruta, S.; Ido,A.; Nakao, K.; Kato, Y.; Ishii, N.; Eguchi, K. Gene. Ther., 1999, 6,465-70.

[99] Rapoport, N.; Pitt, W.G.; Sun, H.; Nelson, J.L. J. Control Release,2003, 91, 85-95.

[100] Alexiou, C.; Jurgons, R.; Schmid, R.J.; Bergemann, C.; Henke, J.;Erhardt, W.; Huenges, E.; Parak, F. J. Drug Target, 2003, 11, 139-49.


[101] Leakakos, T.; Ji, C.; Lawson, G.; Peterson, C.; Goodwin, S. CancerChemother Pharmacol., 2003, 51, 445-50.

[102] Häfeli, U., ed. Scientific and clinical applications of magneticcarriers. Plenum Press: New York, 1997, xiii, 628.

[103] Hoya, K.; Guterman, L.R.; Miskolczi, L.; Hopkins, L.N. DrugDeliv., 2001, 8, 215-22.

[104] Springer, C.J.; Niculescu-Duvaz, I. Adv. Drug Deliv. Rev., 1996,22, 351-64.

[105] Engelmann, C.; Panis, Y.; Bolard, J.; Diquet, B.; Fabre, M.; Nagy,H.; Soubrane, O.; Houssin, D.; Klatzmann, D. Hum. Gene. Ther.,1999, 10, 1545-51.

[106] Bodor, N. J. Control Release, 1999, 62, 209-22.[107] Bodor, N.; Buchwald, P. Adv. Drug Deliv. Rev., 1999, 36, 229-54.

Received: 13 September, 2004 Accepted: 29 September, 2004

Tumor Imaging Using a Picomolar Affinity HER2 Binding

Affibody Molecule

Anna Orlova,1,3Mikaela Magnusson,

1Tove L.J. Eriksson,

1Martin Nilsson,

1Barbro Larsson,

1

Ingmarie Hoiden-Guthenberg,1Charles Widstrom,

2Jorgen Carlsson,

3

Vladimir Tolmachev,1,3Stefan Stahl,

4and Fredrik Y. Nilsson

1,3

1Affibody AB, Bromma; 2Department of Hospital Physics, Uppsala University Hospital; 3Department of Oncology, Radiology, andClinical Immunology, Rudbeck Laboratory, Uppsala University, Uppsala; and 4Department of Biotechnology,AlbaNova University Center, Royal Institute of Technology, Stockholm, Sweden

Abstract

The detection of cell-bound proteins that are produced due toaberrant gene expression in malignant tumors can provideimportant diagnostic information influencing patient man-agement. The use of small radiolabeled targeting proteinswould enable high-contrast radionuclide imaging of cancersexpressing such antigens if adequate binding affinity andspecificity could be provided. Here, we describe a HER2-specific 6 kDa Affibody molecule (hereinafter denoted Affi-body molecule) with 22 pmol/L affinity that can be used forthe visualization of HER2 expression in tumors in vivo usinggamma camera. A library for affinity maturation wasconstructed by re-randomization of relevant positions identi-fied after the alignment of first-generation variants of nano-molar affinity (50 nmol/L). One selected Affibody molecule,ZHER2:342 showed a >2,200-fold increase in affinity achievedthrough a single-library affinity maturation step. Whenradioiodinated, the affinity-matured Affibody moleculeshowed clear, high-contrast visualization of HER2-expressingxenografts in mice as early as 6 hours post-injection. Thetumor uptake at 4 hours post-injection was improved 4-fold(due to increased affinity) with 9% of the injected dose pergram of tissue in the tumor. Affibody molecules represent anew class of affinity molecules that can provide small sized,high affinity cancer-specific ligands, which may be well suitedfor tumor imaging. (Cancer Res 2006; 66(8): 4339-48)

Introduction

Molecular phenotyping of gene products that are aberrantlyexpressed in tumors, thus allowing for diagnosis of tumors bytargeted visualization of cancer markers (e.g., HER2), holds promisefor a shift in clinical practice towards personalizing cancertreatment. However, development has been hampered by the lackof specific agents that bind such gene products, which in additionhave properties making them suitable for use as medical imagingagents.A targeting agent for medical imaging should have high affinity

for its target (1) and should be able to specifically target relevantdisease-related structures in the body while avoiding normal tissue(2). In the patient, it should quickly find its target whereas unbound

molecules should be rapidly excreted, thus facilitating high-contrast tumor imaging and reducing the time between injectionand examination. Although antibodies can be selected for goodspecificity, they are too large to allow for rapid kinetics. Therefore,smaller antibody fragments have gained increasing interest andthere are a number of reports on the use of antibody fragments forin vivo imaging (3–5). However, further size reduction of theseaffinity proteins (27-54 kDa) could potentially improve extravasa-tion, increase tissue penetration, and speed up blood clearance (6),thereby increasing imaging contrast. Short peptides, on the otherhand, are small and have very rapid kinetics. Natural peptidereceptor ligand derivatives, for example, show great potential forradionuclide imaging (7, 8), but there is a limited repertoire ofnatural peptides to choose from. Unfortunately, peptides derivedfrom screening using phage display libraries seldom have the highaffinity necessary to be considered a general class of targetingmolecules (9). Therefore, it would be attractive to develop newtargeting agents, which are smaller than antibody fragments inorder to improve tumor penetration but have better bindingspecificity and affinity than short peptides. Efforts towardsimproved affinity proteins have recently been made based ondifferent protein scaffolds (10–12), e.g., ‘‘trinectins’’ (13, 14),‘‘anticalins’’ (15), ‘‘ankyrin repeats’’ (16), engineered T cell receptors(17), and ‘‘Affibody molecules’’ (18).Affibody molecules represent a new class of affinity proteins

based on a 58–amino acid residue protein domain, derived fromone of the IgG-binding domains of staphylococcal protein A. Thisthree-helix bundle domain has been used as a scaffold for theconstruction of combinatorial phagemid libraries, from whichAffibody variants that target the desired molecules can be selectedusing phage display technology (18, 19). The simple, robuststructure of Affibody molecules in combination with their lowmolecular weight (6 kDa), make them suitable for a wide variety ofapplications, for instance, as detection reagents (20) and to inhibitreceptor interactions (21). In this work, an Affibody moleculespecific for HER2 is described.HER2 is a transmembrane protein belonging to the human

epidermal growth factor tyrosine kinase receptor family and isoverexpressed in several cancer types, but is present only to a smallextent or not at all in normal adult tissue (22, 23). In breast andovarian cancers, HER2 is overexpressed in 23% to 30% of all cases(24). In patients with breast cancer, the overexpression of HER2 isassociated with shorter time to disease progression and decreasedoverall survival, and it has been shown that the overexpression ispreserved in metastases (25, 26). Clinical practice guidelines of theAmerican Society of Clinical Oncology recommend detection ofHER2 expression in all newly diagnosed or recurrent breastcarcinomas in order to select patients who will benefit from

Note: A. Orlova and M. Magnusson contributed equally to this study.Requests for reprints: Fredrik Y. Nilsson, Affibody AB, Box 20137, SE-161 02

Bromma. Phone: 46-8-5988-3851; Fax: 46-8-5988-3801; E-mail: [email protected].

I2006 American Association for Cancer Research.doi:10.1158/0008-5472.CAN-05-3521

www.aacrjournals.org 4339 Cancer Res 2006; 66: (8). April 15, 2006

Research Article

treatment with Herceptin and anthracyclines (27). The use ofradionuclide molecular imaging of HER2 expression may help torapidly identify HER2 expression in tumor lesions, it may assist toavoid false-negative results due to sampling errors and hetero-geneity of expression from biopsy probes, and can provideinformation on HER2 expression in cancers resistant to trastuzu-mab treatment. Together, these factors provide strong argumentsfor the development of radionuclide targeting strategies directedagainst HER2.A previously selected Affibody molecule, ZHER2:4 (28), was

shown to bind selectively to the extracellular domain of HER2with an affinity of 50 nmol/L. Literature data shows that anincrease of affinity beyond 10�8 mol/L improves the tumoruptake of targeting proteins (29, 30). To increase the affinity, wetried dimerization of ZHER2:4 (31, 32) and affinity maturation bydirected evolution, as reported here. For antibodies, secondarylibraries for affinity maturation have proven successful ingenerating high-affinity scFv binders to HER2 (e.g., 13 pmol/L;ref. 33). However, to this date, it has not been shown whethersingle domain scaffold proteins, such as the 6 kDa Affibodymolecule, could yield such high-affinity binders. Here, wedescribe a monovalent HER2-binding Affibody molecule with a2,200-fold increased affinity obtained through an affinitymaturation process of the parental binder (28). The binderwas characterized in vitro and in vivo and compared with thenanomolar parental molecule. Tumor imaging efficiency wasshown in mice using gamma camera.

Materials and Methods

Construction of a secondary phagemid library. The secondary librarywas created by PCR amplification from a single 129-nucleotide template

oligonucleotide with certain degenerated codons (5V-ctc gag gta gac aac aaattc aac aaa gaa nnk mrm mmm gcg tat tgg gag atc nnk nnk tta cct aactta aac nnk nnk caa nnk cgc gcc ttc atc cgc agt tta tat gat gac cca agc caa

agc-3V), encoding helices 1 and 2 of the Z domain. The gene fragment

was amplified using the forward primer 5Vcccccccccctcgaggtaga caacaaatt-caa-5V (Xho I site underlined) and the reverse primer 5Vcccccctg-ctagcaagttagcgctttggcttgggtcatc-3V (NheI site underlined), with 100 fmol

template oligonucleotide for each of 95 parallel reactions. The amplifica-

tions were done using AmpliTaq Gold polymerase (Applied Biosystems,Foster City, CA) for 10 cycles (15 seconds at 96jC, 15 seconds at 60jC, and 1minute at 72jC), pooled, purified using the QIAquick PCR purification kit

(Qiagen, Hilden, Germany), XhoI/NheI digested and ligated to XhoI/NheI-

phagemid vector pAffi128 encoding the third nonvariegated a-helix of theZ domain. Electrocompetent Escherichia coli RRIDM15 (34) cells were

transformed with 60 aliquots of ligated material using 0.2-cm gap size

cuvettes in an ECM 630 set (BTX, Genetronics) at 2,500 V, 125 V and 50 AF.Cells were grown in SOC medium [tryptone soy broth (TSB) + yeast extract(YE) supplemented with 1% glucose, 10 mmol/L MgCl2, 10 mmol/L MgSO4,

10 mmol/L NaCl, and 2.5 mmol/L KCl] for 50 minutes and transferred to six

Erlenmayer flasks, each containing 1 L of TSB + YE (30 g/L TSB and 5 g/L

YE; both from Merck, Darmstadt, Germany) supplemented with 2% glucoseand 25 Ag/mL carbencillin and grown overnight at 37jC. The cells werecentrifuged at 2,000 � g , following resuspension in PBS/glycerol solution

to a final approximate concentration of 20% glycerol, aliquoted and storedat �80jC.

Phage selection procedures. Preparation of phage stocks from the

affinity maturation library (a portion of Zlib2004HER2mat) and between

selections was done according to previously described procedures (18)using the helper phage M13K07 (New England Biolabs, Beverly, MA).

Polyethylene glycol/NaCl precipitation yielded phage titers of f1013

colony-forming units/mL. A 100 kDa recombinant extracellular domain of

HER2 (HER2-ECD) comprising 624 amino acids (35), corresponding to

nucleotides 238 to 2,109, was used as target proteins during selection. Theprotein was biotinylated as previously described (25) using EZ-Link Sulfo-

NHS-LC-Biotin (Pierce, Rockford, IL) with a 30 times molar excess of

biotin.

The library was subjected to five rounds of selection in solution onbiotinylated HER2-ECD using a 10-fold decreasing target concentration for

each round. To avoid nonspecific binders, all tubes used in this procedure

were pretreated with PBST supplemented with 0.1% gelatin and the phage

stock/s were preincubated with streptavidin beads (Dynabeads M-280Streptavidin; 112.06, Dynal A.S., Oslo, Norway) for rounds 1 and 2. The phage

library was subjected to biopanning against HER2-ECD for 1 hour and

45 minutes at room temperature under continuous rotation. To capture

phage binding biotinylated target, SA beads were incubated with the phage-target mixtures for 15 minutes. The beads were washed and the bound

phages were subsequently eluted and propagated as previously described

(36, 37).ELISA-based ranking of second generation binders. Single colonies

were inoculated in 1 mL TSB-YE medium supplemented with 100 Amol/Lisopropyl-L-thio-h-D-galactopyranoside and 100 Ag/mL ampicillin in deep

well plates (Nunc, Roskilde, Denmark), and grown on a shaker overnightat 37jC. Cells were pelleted by centrifugation at 3,000 � g for 10 minutes.The pellets were resuspended in 300 AL PBST and frozen for 30 minutes

at �80jC. The samples were thawed and centrifuged at 3,500 � g for

20 minutes. The supernatants (100 AL), containing ABD-tagged (38)Affibody molecules were loaded in microtiter wells, which had been

previously coated with 6 Ag/mL HSA in 15 mmol/L Na2CO3 and 35 mmol/L

NaHCO3 (pH 9.6) overnight at 4jC and blocked with 2% skimmed milkpowder in PBST for 1 hour at room temperature. The plates were washed

four times with PBST prior to the addition of 100 AL of 1 Ag/mL biotinylatedHER2 per well and incubated for 1.5 hours. After washing the wells four

times, 100 AL streptavidin-horseradish peroxidase (1:5,000; DAKO Cytoma-tion, Denmark) per well were added and incubated for 1 hour. The wells

were washed four times and 100 AL developing solution ImmunoPure TMBsubstrate kit (Pierce) was added to each well. After 30 minutes, 100 AL of thestop solution (2 mol/L H2SO4) was added to each well. The absorbance at450 nm was measured in an ELISA reader.

DNA sequencing and sequence clustering. DNA sequencing was done

with ABI PRISM dGTP, BigDye Terminator v3.0 Ready Reaction CycleSequencing Kit (Applied Biosystems) according to the manufacturer’s

recommendations. The sequences were analyzed on an ABI PRISM 3100

Genetic Analyser.

The Affibody molecules were clustered using the so-called average-link

hierarchical clustering method. Initially, each molecule’s sequence was

defined as a separate subset. A complete clustering was obtained by finding

the two subsets that were least dissimilar to each other and selected

subsets were merged to create a new subset, decreasing the number of

subsets by one. This procedure was repeated until all subsets were

clustered. A subset was regarded as a cluster when it showed up as a node

relatively close to the vertical axis in a dendrogram, whereas the length

of the branch to its parent was relatively large. Cluster validation was done

by applying the so-called Nearest-Exterior-Neighbor/Furthest-Interior-

Neighbor cluster validation method to the subset output from the clus-

tering method. First, a cluster quality index was computed for each subset.

A high value of this index indicates a compact and isolated subset. In a

second step, a reference distribution of cluster quality indices was created,

against which the cluster quality indices for the obtained subset were

compared. This comparison also took subset size into account. The

outcome of these comparisons was a number, the so-called significance

value, for each subset.

Protein expression and purification. Gene fragments encoding the

selected second-generation Affibody molecules ZHER2:473, ZHER2:382, ZHER2:470,

ZHER2:487, ZHER2:489, ZHER2:477, ZHER2:492, ZHER2:475, ZHER2:342, and ZHER2:336, aswell as the first-generation binder, ZHER2:4 (28), were subcloned from the

phagemid vector to a pET-derived expression vector pAY442 under the

control of the T7 promoter. The proteins were expressed in E. coli

BL21(DE3) cells and the His6-tagged proteins were IMAC purified underdenatured conditions using QIAfilter 96 Plates Ni-NTA Superflow kit and

Cancer Research

Cancer Res 2006; 66: (8). April 15, 2006 4340 www.aacrjournals.org

QIAsoft 4.1, medium-scale protein/Ni-NTA Superflow 96 Denat program ona Biorobot 3000 (Qiagen). The protein samples refolded when dialyzed in

PBS using Slide-A-Lyzer (Pierce) according to the manufacturer’s recom-

mendations. The protein concentrations were calculated from the measured

absorption values at 280 nm. Extinction coefficients were theoreticallycalculated from the primary amino acid sequence.

Biosensor analyses. A BIAcore 2000 instrument (Biacore AB, Uppsala,

Sweden) was used for real-time biospecific interaction analysis between

selected Affibody molecules and the target protein. HER2-ECD [diluted in10 mmol/L NaAc (pH 4.5)] was immobilized (f3,500 resonance units) on a

CM5 sensor chip (BR-1000-14, Biacore) according to the manufacturer’s

instructions. Another flow-cell surface was activated and deactivated to be

used as a reference surface, and human IgG (Amersham Biosciences) wasimmobilized (f5,000 resonance units) on a separate flow-cell surface on

CM5 sensor chips, to serve as negative controls. Binding analyses were done

at 25jC, and PBS was used as the running buffer. In a first experiment,

50 nmol/L of each Affibody molecule (diluted in PBS) was injected over allsurfaces with a flow rate of 10 AL/min.In a second experiment, the ZHER2:342 and ZHER2:477 Affibody variants

were subjected to a kinetic analysis, in which the proteins were injected

over a HER2-ECD surface (approximate immobilization levels of 1,200,1,600, and 2,000 resonance units, respectively) at different concentrations

(0-6 nmol/L, with 0.19 nmol/L as the lowest protein concentration diluted

in PBS) with a flow rate of 50 AL/min. The dissociation equilibrium constant

Figure 1. A, the strategy for affinitymaturation. Amino acid sequence of thewild-type Z domain aligned to deducedamino acid sequences of different Affibodyvariants selected against HER2-ECD (28).The 13 randomized amino acid residues(Q9, Q10, N11, F13, Y14, L17, H18, E24,E25, R27, N28, Q32, and K35) arepresented. An amber stop codon is includedin the ZHER2:25 variant. Horizontal bars,amino acid identities. Right, the number oftimes each variant was found upon DNAsequencing of 49 colonies. The design ofthe constructed library aimed for affinitymaturation ZLibHER2:mat is presented insingle letter code, with positions selected forvariation in boldface (X = any amino acid).Note, that a bias for arginine or lysineresidues is introduced at position 10, and abias for glutamine or threonine residuesat position 11. The three a-helices in thewild-type Z domain are boxed. B, sequenceclustering. The Affibody molecules wereclustered using an average-link hierarchicalclustering method, and the result isillustrated in a dendrogram. A subset wasregarded as a cluster when it showed up as anode relatively close to the vertical axis inthe dendrogram, whereas the length of thebranch to its parent was relatively large.Cluster validation was done by applyinga Nearest-Exterior-Neighbor/Furthest-Interior-Neighbor cluster validation methodto the subset output from the clusteringmethod. Positions of the 10 second-generation Affibody molecules selected forfurther characterization. C, sequences ofselected second-generation Affibodyvariants. Amino acid sequence of thewild-type Z domain aligned to deducedamino acid sequences of 10 differentsecond-generation Affibody variantsselected in the affinity maturation effortagainst HER2-ECD. The amino acidresidues found in the variegated position inboldface. Amino acid residues other thanR/K at position 10 and Q/T at position11 could appear by alternative combinationsof the included nucleotides. Right, theselection cycle from which the clone wasisolated/the selection principle applied forthis specific clone/the number of times eachvariant was found on DNA sequencing of160 randomly picked colonies from selectioncycles four and five.

Affibody Tumor Imaging


(KD), the association rate constant (ka), and the dissociation rateconstant (kd) were calculated using BIAevaluation 3.2 software (Biacore),

assuming a one-to-one Langmuir binding model taking mass transfer

effects into account. A long dissociation phase was applied (60 minutes),

due to the slow off rate, to get fully reliable results.In a third experiment, the difference in binding to HER2-ECD between

the monovalent ZHER2:4 (28) and the divalent (ZHER2:4)2 (31) first-generation

Affibody variants and the affinity-matured second-generation ZHER2:342Affibody molecule, was evaluated by injection of 5 Amol/L of each proteinover the HER2-ECD surface, with a flow rate of 5 AL/min.

Immunofluorescence. The human ovarian adenocarcinoma SKOV-3

(HTB-77, ATCC), and the human neuroblastoma cell line, SH-SY5Y (CRL-

2266, ATCC), were grown in 5% CO2 incubator at 37jC in 25 cm2 flasks.SKOV-3 cells were cultured in McCoy’s 5a and SH-SY5Y in DMEM medium

(Life Technologies, Invitrogen) supplemented with 2 mmol/L glutamine and

10% fetal calf serum (Life Technologies, Invitrogen). Cells were trypsinated(Trypsin-EDTA, Life Technologies, Invitrogen) the day before analysis,

washed in medium, and resuspended to a concentration of 2.5 � 105 cells/

mL. Approximately 5,000 to 10,000 cells were added per well of an eight-

well, multi-well slide (Histolab, Sweden). The cells were incubated on multi-well slides overnight to obtain a flat morphology. The next day, medium was

gently removed from the slides and cells were stained for 1 hour with the

ZHER2:342 Affibody molecule (5 Ag/mL) or 0.5 Ag/mL of the monoclonal

antibody trastuzumab (Herceptin, Apoteket AB, Sweden), with a volume of20 AL/well. After staining, cells were fixed with 3% formaldehyde in PBS, one

drop per well for 15 minutes. The slides were rinsed in PBS and the ZHER2:342Affibody molecule was detected with mouse anti-Affibody mouse serum(prepared in-house) followed by 5 Ag/mL of goat anti-mouse IgG Alexa

Fluor 488 (Molecular Probes, Eugene, OR). Trastuzumab was detected with

5 Ag/mL of goat anti-human IgG Alexa Fluor 488 (Molecular Probes). The

slides were counterstained with 20 AL of 4V,6-diamidino-2-phenylindole

(Molecular Probes), 1 Ag/mL for 5 seconds, washed, dried for 1 hour, andmounted with antifading reagent (Vector Laboratories, Inc.).

ZHER2:342 and anti-HER2 antibody were used to simultaneously stain

HER2 on SKOV-3 cells. Slides with formaldehyde-fixed cells were stained

with ZHER2:342 (5 Ag/mL) Affibody molecule followed by goat anti-AffibodyIgG and rabbit anti-goat IgG Alexa Fluor 488 (Molecular Probes). Antibody

staining was done on Affibody molecule–stained cells and the antibody was

detected with anti-human IgG Alexa Fluor 555 as described above. The

slides were counterstained with 20 AL of 4V,6-diamidino-2-phenylindole(Molecular Probes), 1 Ag/mL for 5 seconds, washed, dried for 1 hour, and

mounted with antifading reagent (Vector Laboratories).

Membrane fluorescence was analyzed in a DMLA microscope, equipped

with a digital camera (Leica Microsystems AG). Digital images of green, redand blue fluorescence were acquired and saved as overlays using the IM1000

software (Leica Microsystems).

Immunohistochemical staining. SKOV-3 xenograft tumors (describedbelow) were excised and snap-frozen in liquid nitrogen. Six-micron-thick

cryosections of the tumors (Ljung CM3000, Leica Microsystems) were fixed

in 3% formaldehyde in PBS for 15 minutes and blocked in 1% FCS in PBS for

1 hour before staining. The sections were either stained with ZHER2:342 at aconcentration of 5 Ag/mL, followed by polyclonal rabbit anti-Affibody IgG

(prepared in-house) and horseradish peroxidase-conjugated anti-rabbit IgG

(DAKO Cytomation) or with trastuzumab (0.5 Ag/mL) followed by

horseradish peroxidase–conjugated anti-human IgG (DAKO Cytomation).The staining was developed for 7 minutes with 3,3V-diaminobenzidinechromogen substrate (DAKO Cytomation) and then washed for 2 to

3 minutes. The slides were counterstained with Mayers HTX (Histolab) for20 seconds and then rinsed for 10 minutes before mounting with Mount-

quick (Histolab). The staining was analyzed in a DMLA microscope,

equipped with a digital camera (Leica Microsystems). Digital images were

acquired and saved using the IM1000 software (Leica Microsystems).

Figure 2. Biacore ranking of second-generation Affibody variants and comparisonto first generation binders. A, sensorgramsobtained after injection of the 10 differentsecond-generation Affibody moleculesover a sensor chip flow-cell surfacecontaining amine-coupled HER2-ECD.Sensorgrams are: (a ) ZHER2:336, (b )ZHER2:342, (c ) ZHER2:475, (d) ZHER2:477,(e ) ZHER2:492, (f ) ZHER2:487, (g ) ZHER2:489, (h)ZHER2:473, (i) ZHER2:382, and (j) ZHER2:470.B, sensorgrams obtained after injection ofmonovalent ZHER2:4 Affibody molecule,compared with divalent (ZHER2:4)2 Affibodymolecule and second-generation ZHER2:342Affibody molecule over a sensor chipcontaining amine-coupled HER2-ECD.

Cancer Research


Radioiodination. The radiolabeling with 125I of the ZHER2:342 Affibodymolecule was done using N-succinimidyl p-(trimethylstannyl)benzoate as a

precursor, as previously described (25). Labeling conditions were adjusted

to provide attachment for a single prosthetic group per protein molecule.

Radiochemical purity was controlled using ITCL SG eluted with 70%acetone in water and was always better than 95%. The radiolabeled Affibody

molecules were analyzed using Biacore technology to verify that the labeling

procedure had not affected the binding affinity to HER2-ECD.

Biodistribution in tumor-bearing mice. Animal experiments wereapproved by the local ethics committee for animal research. SKOV-3

xenografts were established on female BALB/c nu/nu mice (10-12 weeks old

on arrival), 4 to 8 weeks before the experiment by implanting f5 � 106

cells s.c. on the right hind leg. All mice were injected with f50 AL (0.5 Ag,100 kBq) of the 125I-labeled ZHER2:4, (ZHER2:4)2, or ZHER2:342 Affibody

molecules into the tail vein. Groups of four mice were sacrificed after 1,

4, and 24 hours, and organs were dissected, weighted, and their radioactivitycontent was measured in the automated gamma counter. Tissue uptake was

calculated as the percentage of injected radioactivity per gram of tissue

(% IA/g). To show that tumor uptake was target-mediated, a control group

of four animals was pretreated with 0.5 mg of cold, nonlabeled Affibodymolecules 45 minutes before the injection of 0.5 Ag of the radioiodinatedcounterpart. Four hours later, these animals were sacrificed and tumor

radioactivity uptake was measured.

Gamma camera–based imaging. SKOV-3 xenografts were establishedfor f2 months, as described above. The tumors had an average volume of

f0.5 cm3 at the day of the experiment. In total, 60 Ag of ZHER2:342 was

labeled with 111 MBq 125I, resulting in a specific activity of 1.3 MBq/Ag.Each mouse was injected with 90 AL (2.3 Ag, 2.9 MBq) 125I-ZHER2:342. After6 or 24 hours, respectively, mice were euthanized with a lethal dose of

Ketalar/Rompun i.p. The mice were imaged using an e.CAM gamma

camera (Siemens, Erlangen, Germany) with a low-energy high-resolutioncollimator.

Results

Affinity maturation of first-generation HER2 binders. Wedesigned and constructed the affinity maturation library based on aprimary set of HER2-binding Affibody molecules (28). These werealigned with an allowed bias for the strongest binders, and wefound it reasonable to fix 5 (13, 14, 28, 32, and 35), and allow a biasfor R and K in position 10, and Q and T in position 11, of the 13originally randomized amino acid residues. Thus, positions 9, 17,18, 24, 25, and 27 were targeted for randomization (Fig. 1A) usingNNG/T degenerated codons. Due to the small size of the Affibodymolecule, it was possible to use a single 129 nucleotideoligonucleotide with degenerated codons, encoding helices 1 and2 of the Z-domain, to create the secondary library. Theoligonucleotide was PCR-amplified and subsequently ligated intoa phagemid vector encoding the third a-helix of the Z-domain. Theresulting library consisted of f3 � 108 members, which should beenough to sample a majority of all theoretical variants. Phagestocks were prepared and selections done essentially as previouslydescribed (18, 28), using decreasing concentrations of targetprotein and intensive washing to select for the strongest HER2-binding Affibody variants in the library.Clones obtained after four and five rounds of selection were

cultivated in 96-well plates, freeze-thawed to release periplasmiccontent, and subjected to an ELISA screening procedure for HER2binding activity (data not shown). When subjecting 160 randomlypicked clones to this ELISA screening, using the first-generationAffibody molecule ZHER2:4 as reference, a majority of the clonesgave significantly higher absorbance values than the referenceZHER2:4, thus, indicating stronger HER2 binding (data not shown).The same 160 clones were subjected to DNA sequencing, and upon

clustering of the sequenced clones using an average-link hierar-chical clustering method developed in-house, the relationshipbetween selected clones was visualized (Fig. 1B). Based on thevalues in the ELISA screening and the clustering results, 10 cloneswere selected for further characterization (Fig. 1C). The cycle fromwhich the particular clone was isolated, and the frequency ofoccurrence are indicated (Fig. 1C), as well as their position in theclustering analysis (Fig. 1B).Biosensor screening, stability, and affinity determination.

To obtain an initial ranking of binding affinities, the 10 selectedAffibody variants as well as the ZHER2:4 and dimeric (ZHER2:4)2 wereexpressed and analyzed for their HER2 binding using a Biacoreinstrument. The different Affibody molecules were separatelyinjected over sensor chip flow-cell surfaces containing theimmobilized target protein HER2-ECD and a control protein IgG,respectively. An overlay plot of the obtained sensorgrams suggested

Figure 3. Fluorescence staining of SKOV-3 and SH-SY5Y cells with ZHER2:342and trastuzumab. A, SKOV-3 cells stained for HER2 expression with ZHER2:342Affibody molecules; B, SKOV-3 cells stained for HER2 expression withtrastuzumab; C, the HER2-negative cell line SH-SY5Y stained with ZHER2:342;and D, the HER2-negative cell line SH-SY5Y stained with trastuzumab. E and F,the same cells double-stained for ZHER2:342 (green ) and trastuzumab (red),respectively. G, immunohistochemical staining of HER2 with ZHER2:342 Affibodymolecules was done on a section of SKOV-3 cell xenograft tumor. Thesection was counterstained with hematoxylin (blue ). The transversing mouseconnective tissue remains negative. H, a consecutive section of SKOV-3 cellxenograft tumor was stained with trastuzumab.



that approximately half of the analyzed Affibody variants bound tothe target with significantly higher affinities than ZHER2:4 (Fig. 2A).The slower off-rate kinetics of second-generation variants couldclearly be seen by visual inspection of the sensorgrams. Asexpected, no significant binding could be seen in the controlprotein IgG (data not shown). A group of four Affibody variants,ZHER2:336, ZHER2:342, ZHER2:475, and ZHER2:477, with good bindingprofiles were selected for further studies.Differences in solubility between the Affibody variants was

investigated. The threshold was set at no loss of protein at aconcentration of 4 mg/mL upon freezing and thawing, and withno signs of aggregates in gel filtration (data not shown). ZHER2:336and ZHER:475 did not meet these criteria. The melting temperatures(Tm’s) of the selected Affibody variants varied substantially: 73jCfor ZHER2:336, 67jC for ZHER2:342, 46jC for ZHER2:475, and 49jC forZHER2:477, respectively. The Affibody variants were also testedin vivo for tumor uptake and kidney retention at 4 hours withtumor to blood ratios for ZHER2:336, ZHER2:342, ZHER2:475, andZHER2:477 of 18 F 6, 40 F 3, 18 F 4, and 21 F 6, respectively, andtumor to kidney ratios of 0.6 F 0.2, 1.0 F 0.1, 0.5 F 0.2, and 0.62 F0.06, respectively. Taking these different factors into account, theZHER2:342 followed by the ZHER2:477 variant seemed to be mostpromising for further studies.The ZHER2:342 and ZHER2:477 Affibody molecules were subjected to

a kinetic analysis in order to determine the kinetic bindingconstants. Prior to the kinetic analysis, the protein concentrationwas carefully determined by amino acid analysis, and gel filtrationwas done (data not shown) to confirm a monomeric state. TheAffibody molecules were then injected over a HER2-ECD flow-cellsurface or a control EGFR-ECD flow-cell surface at differentconcentrations, and the dissociation equilibrium constant (KD) for

HER2-ECD was determined to be f22 pmol/L for ZHER2:342 andf32 pmol/L for ZHER2:477. The association rate constant (ka) wascalculated to be f4.8 � 106 M�1 s�1 and the dissociation rateconstant (kd) f1.1 � 10�4 s�1 for ZHER2:342, and for ZHER2:477, theka wasf4.3 � 106 M�1 s�1 and the kd wasf1.4 � 10�4 s�1. Noneof the Affibody molecules bound EGFR-ECD (HER1), a proteinwhich has strong homology with HER2. Based on these data,ZHER2:342 was selected for further characterization.Comparing first- and second-generation binders in vitro .

The affinity-matured ZHER2:342 (KD f22 pmol/L for HER2) wascompared with a monovalent (KD f50 nmol/L for HER2) and adivalent form (KD f3 nmol/L for HER2) of the first-generationAffibody molecule (28, 31), ZHER2:4 using Biacore analysis (Fig. 2B).The monovalent and bivalent first-generation Affibody moleculehad similar association rates (31), whereas the bivalent (ZHER2:4)2had a 13-fold slower dissociation rate. Interestingly, the affinity-matured ZHER2:342 Affibody molecule did not only have a 90-foldslower dissociation rate than the parental ZHER2:4, but also a 27-foldfaster association rate (Fig. 2B).Fluorescence and immunohistochemical analysis. The ability

of the matured Affibody molecules to detect membrane-boundHER2 was analyzed using fluorescence microscope analysis. TheAffibody molecule ZHER2:342 molecule worked very well as adetection reagent and gave a membrane staining on HER2-positiveSKOV-3 cells (Fig. 3A), but not on HER2-negative SH-SY5Y cells(Fig. 3C). Approximately 50% of the SKOV-3 cells were stronglystained with ZHER2:342 with an almost identical staining pattern asthe monoclonal antibody trastuzumab, which was used as apositive control (Fig. 3B and D). Double staining with trastuzumaband ZHER2:342 Affibody molecules showed that both reagentsbind to the same cells and with identical staining morphology

Table 1. Biodistribution of ZHER2:4, (ZHER2:4)2, and ZHER2:342

Organ

(% IA/g)

1 hour 4 hours 24 hours

ZHER2:4 (ZHER2:4)2 ZHER2:342 ZHER2:4 (ZHER2:4)2 ZHER2:342 ZHER2:4 (ZHER2:4)2 ZHER2:342

Blood 2.61 F 0.35 3.83 F 0.29 4.50 F 0.78 0.39 F 0.09 0.56 F 0.14 0.25 F 0.03 0.03 F 0.00 0.02 F 0.01 0.04 F 0.01

Heart 1.00 F 0.16 1.34 F 0.12 2.11 F 0.64 0.13 F 0.06 0.31 F 0.14 0.11 F 0.02 0.01 F 0.00 0.01 F 0.01 0.01 F 0.02

Lung 2.25 F 0.40 2.56 F 0.10 4.56 F 1.02 0.40 F 0.09 0.44 F 0.12 0.37 F 0.03 0.04 F 0.01 0.05 F 0.02 0.07 F 0.02

Liver 4.08 F 0.70 5.10 F 0.25 8.32 F 1.18 0.77 F 0.09 1.00 F 0.38 0.30 F 0.05 0.07 F 0.03 0.02 F 0.00 0.04 F 0.00Spleen 1.81 F 0.40 1.41 F 0.10 2.68 F 0.37 0.58 F 0.17 0.19 F 0.04 0.12 F 0.02 0.06 F 0.03 0.01 F 0.01 0.03 F 0.01

Pancreas 0.93 F 0.10 2.43 F 0.08 2.48 F 0.48 0.16 F 0.06 0.26 F 0.05 0.13 F 0.03 0.01 F 0.00 0.02 F 0.01 0.01 F 0.00

Kidney 24.98 F 3.69 68.90 F 10.62 49.64 F 6.72 6.04 F 0.64 29.85 F 3.16 9.51 F 0.78 0.16 F 0.01 0.87 F 0.13 0.46 F 0.21Stomach 1.18 F 0.10 2.26 F 0.29 2.21 F 0.46 0.59 F 0.30 0.76 F 0.21 0.25 F 0.04 0.02 F 0.00 0.06 F 0.08 0.03 F 0.01

Small intestine 2.35 F 0.70 ND F ND 2.65 F 1.01 0.41 F 0.13 ND F ND 0.61 F 0.27 0.01 F 0.01 ND F ND 0.01 F 0.01

Large intestine 0.90 F 0.15 ND F ND 1.93 F 0.67 0.12 F 0.06 ND F ND 0.66 F 0.50 0.00 F 0.00 ND F ND 0.01 F 0.01

Salivary glands 1.33 F 0.18 2.70 F 0.53 2.33 F 0.27 0.60 F 0.14 1.39 F 0.31 0.33 F 0.09 0.02 F 0.00 0.03 F 0.02 0.02 F 0.02Skin 1.33 F 0.24 3.31 F 1.18 2.58 F 0.75 0.25 F 0.07 0.27 F 0.09 0.26 F 0.10 0.02 F 0.00 0.02 F 0.01 0.03 F 0.01

Muscle 0.36 F 0.06 0.84 F 0.12 0.66 F 0.14 0.05 F 0.02 0.07 F 0.03 0.06 F 0.03 0.00 F 0.00 ND F ND 0.00 F 0.00

Bone 0.54 F 0.08 0.56 F 0.11 1.03 F 0.24 0.06 F 0.05 0.08 F 0.04 0.11 F 0.02 0.00 F 0.00 0.05 F 0.04 0.03 F 0.04

Brain 0.07 F 0.01 ND F ND 0.12 F 0.02 0.00 F 0.00 ND F ND 0.04 F 0.02 0.00 F 0.00 ND F ND 0.00 F 0.00Tumor 4.05 F 1.28 3.23 F 0.12 8.23 F 2.08 2.40 F 0.17 2.27 F 0.55 9.46 F 1.60 0.17 F 0.03 0.34 F 0.11 4.13 F 0.46

Ratios

Tumor/blood 1.6 0.8 1.8 6.2 4.1 37.8 5.7 17.0 103.3Tumor/kidney 0.2 0.0 0.2 0.4 0.1 1.0 1.1 0.4 9.0

Tumor/liver 1.0 0.6 1.0 3.1 2.3 31.1 2.5 17.0 103.3

NOTE: Tumor and normal organ uptakes are expressed as the percentage of injected activity per gram F SD (n = 4).

Cancer Research


(Fig. 3E and F). Importantly, in accordance with earlier studies (28),ZHER2:342 and trastuzumab could bind simultaneously to SKOV-3cells without blocking each other’s binding sites (data not shown)confirming that they bind to different epitopes on HER2 (31).Furthermore, ZHER2:342 and trastuzumab were used for traditionalimmunohistochemical staining of cryosections of SKOV-3 xeno-graft tumor tissues. Both the Affibody molecule and trastuzumabbound to SKOV-3 tumor cells leaving the surrounding andtransversing mouse connective tissue negative (Fig. 3G and H)with no qualitative difference in the staining behavior betweenZHER2:342 and trastuzumab.In vivo biodistribution. To investigate whether improved

affinity would result in improved tumor uptake, the affinity-matured ZHER2:342 Affibody molecule (KD f22 pmol/L) wascompared with the first-generation Affibody molecule in itsmonovalent (ZHER2:4, KD f50 nmol/L for HER2) and divalentforms (KD f3 nmol/L). Mice carrying SKOV-3 xenografts of f3mm in diameter were injected with 125I-labeled Affibody molecules.The radioactivity in excised tumors was measured at 1, 4, and 24hours post-injection. and presented as the percentage of injecteddose per gram of tissue (% ID/g; Table 1). Very little difference wasobserved between the two variants of the first-generation binderwhereas the uptake for ZHER2:342 compared with the parentalZHER2:4 monomer was improved by 2.03, 3.94, and 24.29 times at 1,4, and 24 hours, respectively, with a tumor uptake of f9% ID/g at4 hours (Fig. 4A). To verify that the observed tumor-targeting effect

was indeed target mediated, one group of mice was pretreatedwith a 1,000-fold (0.5 mg) excess of unlabeled ZHER2:342 Affibodymolecule before injection of radiolabeled ZHER2:342 Affibodymolecule. Tumor uptake at 4 hours post-injection was reducedfrom 9.5 F 2.3% to 1.5 F 1.5% ID/g (P < 0.01) by the pretreatment,demonstrating that tumor uptake was target-specific. The sameconfirmation of target specificity was obtained both for themonomeric and dimeric form of first-generation of Affibodymolecules, where tumor uptake was significantly reduced by theinjection of a large excess of unlabeled forms of the same peptide(P < 0.001 and P < 0.01, respectively). The only sites of nonspecificaccumulation were the kidneys, which could be expected forproteins with a size significantly below renal filtration threshold.However, the radioactivity was rapidly washed out from thekidneys.It was apparent that the ratio of uptake in tumor compared

with other organs was improved with increased affinity when theradioactivity concentration in tumor and blood was compared overtime for the different Affibody molecules (Fig. 4B ; Table 1). In spiteof the increased affinity of the dimer, the difference in tumor-to-blood ratio between the monomeric and dimeric variants of thefirst-generation binder was modest. In contrast, a dramaticimprovement was found for the ZHER2:342 Affibody molecule,showing an impressive tumor-to-blood ratio of 38:1 at 4 hoursand 100:1 at 24 hours post-injection. At 4 hours post-injection,tumor-to-organ ratios for most other organs were at least 2-fold

Figure 4. Evaluation of the Affibody molecules in vivo.A, biodistribution of 125I-labeled Affibody molecules 4 hourspost-injection. Percentage of injected dose per gram of tissue isplotted for ZHER2:4 (white ), and ZHER2:342 (gray ), monovalent Affibodymolecules (columns, mean; bars, F SE; n = 4). B, in a secondexperiment, the relation in uptake between tumor and bloodwas investigated at different time points relevant to imaging;tumor (squares ) and blood (circles ) values for ZHER2:4 (dashed line ),and ZHER2:342 (solid line ). Points, mean; bars, F SE (n = 4).



better for ZHER2:342 in comparison with ZHER2:4. Twenty-four hourspost-injection, that difference was >10-fold. Furthermore, at24 hours, the tumor to kidney ratio was improved to 9:1 for theZHER2:342 as compared with the ZHER2:4, where the tumor to kidneyratio was 1:1.Gamma camera–based imaging of HER2 expression in

SKOV-3 xenografts. To evaluate whether the matured ZHER2:342could be used for in vivo imaging of HER2 expression, theAffibody molecule was labeled with 125I and injected into micecarrying SKOV-3 tumors f0.5 cm3 in size. After 6 or 24 hours,images of tumor-bearing mice were acquired using a gammacamera. After 6 hours, the tumor could clearly be visualized withhigh contrast. At this time point, the only site with elevatedradioactivity uptake besides the tumor were the kidneys. However,radioactivity accumulation in the tumor was appreciably higherthan in the kidneys (Fig. 5B). The increase in contrast due toincrease in affinity is striking when compared with the first-generation (ZHER2:4)2 Affibody molecule, in which the kidneyuptake is much more prominent than the tumor after 6 hours(Fig. 5A). After 24 hours, the nonspecific radioactivity accumu-lation was further reduced whereas a high dose remained on thetumor (Fig. 5C).

Discussion

In this study, we describe in vivo tumor targeting by use of a newclass of small affinity proteins called Affibody molecules. Wereasoned that the very high affinity for the HER2 tumor target,combined with the very small size of the targeting molecule, may

be an avenue for improving tumor specificity (39, 40)—and thus,the contrast of a radionuclide image—over existing larger affinityproteins like antibodies and fragments thereof (41, 42). To obtainsmall high affinity binders for HER2, a 6 kDa HER2-specificAffibody molecule was affinity maturated in vitro . The procedureused for affinity maturation resulted in an increased affinity bythree orders of magnitude in one single maturation step. Thematured Affibody molecule with highest affinity (KD of 22 pmol/L)was selected through various in vitro and in vivo characterizationsand showed improved tumor targeting in biodistribution studiesand gamma imaging.It is interesting to compare the in vivo data of the ZHER2:342

Affibody molecule described in this study with published results forother HER2-targeting proteins. We have restricted ourselves to theanalysis of targeting of HER2 using radioiodinated proteins.Moreover, all studies which were selected for comparison usedSKOV-3 ovarian carcinoma xenografts, the same as that used in thisstudy. Together, this should minimize differences in agent-targetinteraction, antigen density, and cellular processing. Comparisonsin the literature show that the uptake of the ZHER2:342 Affibodymolecule at 4 hours post-injection (9.5 F 1.6%, ID/g) was similar tothe uptake in tumor of the C6.5 scFv diabody (10.1% ID/g; ref. 43),and exceeded the uptake of the 741F8-2 sFv (2.9% ID/g; ref. 44), andits homodimer (5.6% ID/g; ref. 45). However, better clearance ofZHER2:342 provided tumor-to-blood ratios, which at this time point,was 7.8 times better than for the C6.5 scFv diabody, 4.5 times betterthan for the 741F8-2 sFv, and 9.3 times better than for the 741F8-2(sFvV)2. Advantages in contrast remained over time even at24 hours.

Figure 5. Gamma camera–based imaging of xenografted mice, at 6 and 24 hours after injection of 125I-labeled Affibody molecules. In a set of experiments, thefirst-generation Affibody molecule, ZHER2:4 (A) was compared with the affinity-matured Affibody molecule ZHER2:342 (B) at 6 hours post-injection. Intensity of colorcorresponds to radioactive accumulation: low (blue ) and high (yellow ) accumulation. Kidney uptake is more prominent for the ZHER2:4 Affibody molecule (A).The tumor uptake of ZHER2:342 was stable up to at least 24 hours post-injection (C ).

Cancer Research


It has been suggested that bivalently binding antibody fragmentshave superior tumor specificity compared with their monovalentcounterparts (46), and in a recent study, Wu and colleagues showedthat even relatively large (f80 kDa) HER2-specific divalentminibodies could be used for tumor imaging in mice (47). However,a bivalent antibody fragment is, by necessity, larger than amonovalent one, and should thus have poorer tissue penetrationproperties (39). It remains an open question if the targeting benefitof divalency could be overcome by increased affinity for amonovalent smaller fragment. In this study, the first-generationbivalent HER2 Affibody molecule (KD 3 nmol/L) did not showimproved tumor-to-blood ratios compared with the monovalentmolecule (KD 50 nmol/L), in spite of higher affinity. In contrast,when improving the affinity, both the total tumor uptake andtumor-to-blood contrast improved dramatically. These results maysuggest that for very small proteins, keeping the size small is moreimportant than gaining functional affinity by dimerization.Taken together, these results suggest that the radioiodinated

ZHER2:342 Affibody molecule may be a suitable agent for in vivoimaging of HER2 expression, and a literature comparison suggeststhat ZHER2:342 may provide better tumor contrast at early timepoints than HER2-specific antibodies and fragments thereof (48). It

is important to choose the optimal time point for the image, asthere is an initial high uptake of activity in the kidneys due toexcretion. The activity is, however, rapidly decreasing, from almost50% ID/g at 1 hour post-injection to <10% ID/g at 4 hours post-injection. Six hours post-injection, the uptake in the tumor is muchmore prominent than the kidneys, as can be seen in Fig. 5B .If the molecule was to be tested in the clinic, obviously a

single-photon emitter 123I (T1/2 = 13.3 hours) or positron-emitter124I (T1/2 = 4.17 days) should be used instead of 125I. Thedevelopment of suitable chemistry to label the ZHER2:342 moleculewith generator-produced 99mTc (T1/2 = 6 hours) or

68Ga (T1/2 = 68minutes) could further facilitate future imaging studies.

Acknowledgments

Received 10/6/2005; revised 12/19/2005; accepted 2/13/2006.Grant support: P25882-1 from the Swedish Agency for Innovation Systems

(VINNOVA) and by two research grants from the Swedish Cancer Society.The costs of publication of this article were defrayed in part by the payment of page

charges. This article must therefore be hereby marked advertisement in accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Per-Ake Nygren and Elin Gunneriusson for valuable advice on the affinitymaturation, Per-Olof Fjallstrom for computing the software for clustering of selectedclones, and Lars Abrahmsen and Birger Jansson for comments on the manuscript.



References1. Behr TM, Gotthardt M, Barth A, Behe M. Imagingtumors with peptide-based radioligands. Q J Nucl Med2001;45:189–200.

2. Britz-Cunningham SH, Adelstein SJ. Molecular target-ing with radionuclides: state of the science. Nucl Med2003;44:1945–61.

3. Santimaria M, Moscatelli G, Viale GL, et al. Immuno-scintigraphic detection of the ED-B domain of fibro-nectin, a marker of angiogenesis, in patients withcancer. Clin Cancer Res2003;9:571–9.

4. Olafsen T, Kenanova VE, Sundaresan G, et al.Optimizing radiolabeled engineered anti-p185HER2antibody fragments for in vivo imaging. Cancer Res2005;65:5907–16.

5. Robinson MK, Doss M, Shaller C, et al. Quantita-tive immuno-positron emission tomography imagingof HER2-positive tumor xenografts with an iodine-124 labeled anti-HER2 diabody. Cancer Res 2005;65:471–8.

6. Michel CC, Curry FE. Microvascular permeability[review]. Physiol Rev 1999;79:703–61.

7. Heppler A, Froidevaux S, Eberle AN, Maecke H.Receptor targeting for tumor localization and therapywith radiopeptides. Curr Med Chem 2000;7:791–4.

8. Reubi JC. Somatostatin and other peptide receptors astools for tumor diagnosis and treatment. Neuroendo-crinology 2004;80:51–6.

9. Landon LA, Deutscher SL. Combinatorial discovery oftumor targeting peptides using phage display. J CellBiochem 2003;90:509–17.

10. Nygren PA, Skerra A. Binding proteins fromalternative scaffolds [review]. J Immunol Methods2004;290:3–28.

11. Skerra A. Engineered protein scaffolds for molecularrecognition. J Mol Recognit 2001;13:167–87. ReviewErratum in: J Mol Recognit 2001;14:141.

12. Binz HK, Amstutz P, Pluckthun A. Engineering novelbinding proteins from nonimmunoglobulin domains.Nat Biotechnol 2005;23:1257–68.

13. Xu L, Aha P, Gu K, et al. Directed evolution of high-affinity antibody mimics using mRNA display. ChemBiol 2002;9:933–42.

14. Koide A, Abbatiello S, Rothgery L, Koide S. Probingprotein conformational changes in living cells by usingdesigner binding proteins: application to the estrogenreceptor. Proc Natl Acad Sci U S A 2002;99:1253–8.

15. Skerra A. Lipocalins as a scaffold [review]. BiochimBiophys Acta 2000;1482:337–50.

16. Binz HK, Amstutz P, Kohl A, et al. High-affinitybinders selected from designed ankyrin repeat proteinlibraries. Nat Biotechnol 2004;22:575–82.

17. Li Y, Moysey R, Molloy PE, et al. Directed evolution ofhuman T-cell receptors with picomolar affinities byphage display. Nat Biotechnol 2005;23:349–54.

18. Nord K, Gunneriusson E, Ringdahl J, Stahl S, UhlenM, Nygren PA. Binding proteins selected from combi-natorial libraries of an a-helical bacterial receptordomain. Nat Biotechnol 1997;15:772–7.

19. Ronnmark J, Gronlund H, Uhlen M, Nygren PA.Human immunoglobulin A (IgA)-specific ligands fromcombinatorial engineering of protein A. Eur J Biochem2002;269:2647–55.

20. Ronnmark J, Hansson M, Nguyen T, et al. Construc-tion and characterization of affibody-Fc chimerasproduced in Escherichia coli . J Immunol Methods2002;261:199–211.

21. Sandstrom K, Xu Z, Forsberg G, Nygren PA. Inhibitionof the CD28–CD80 co-stimulation signal by a CD28-binding Affibody ligand developed by combinatorialprotein engineering. Protein Eng 2003;16:691–7.

22. Natali PG, Nicotra MR, Bigotti A, et al. Expression ofthe p185 encoded by HER2 oncogene in normal andtransformed human tissues. Int J Cancer 1990;45:457–61.

23. Press MF, Cordon-Cardo C, Slamon DJ. Expression ofthe HER2/neu proto-oncogene in normal human adultand fetal tissues. Oncogene 1990;5:953–62.

24. Wang SC, Zhang L, Hortobagyi GN, Hung MC.Targeting HER2: recent developments and futuredirections for breast cancer patients [review]. SeminOncol 2001;28:21–9.

25. Gancberg D, Di Leo A, Cardoso F, et al. Comparisonof HER2 status between primary breast cancer andcorresponding distant metastatic sites. Ann Oncol 2002;13:1036–43.

26. Carlsson J, Nordgren H, Sjostrom J, et al. HER2expression in breast cancer primary tumours andcorresponding metastases. Original data and literaturereview [review]. Br J Cancer 2004;90:2344–8.

27. Bast RC, Jr., Ravdin P, Hayes DF, et al. AmericanSociety of Clinical Oncology Tumor Markers ExpertPanel. 2000 update of recommendations for the use oftumor markers in breast and colorectal cancer: clinicalpractice guidelines of the American Society of ClinicalOncology. J Clin Oncol 2001;19:1865–78. Erratum in:

J. Clin. Oncol. 2001;19:4185–8 and J Clin Oncol 2002;20:2213.

28. Wikman M, Steffen AC, Gunneriusson E, et al.Selection and characterization of HER2/neu-bindingAffibody ligands. Protein Eng Des Sel 2004;17:455–62.

29. Kuan CT, Wikstrand CJ, Archer G, et al. Increasedbinding affinity enhances targeting of glioma xeno-grafts by EGFRvIII-specific scFv. Int J Cancer 2000;88:962–9.

30. Adams GP, Schier R, Marshall K, et al. Increasedaffinity leads to improved selective tumor delivery ofsingle-chain Fv antibodies. Cancer Res 1998;58:485–90.

31. Steffen AC, Wikman M, Tolmachev V, et al. In vitrocharacterization of a bivalent anti-HER2 Affibody withpotential for radionuclide-based diagnostics. CancerBiother Radiopharm 2005;20:239–48.

32. Steffen AC, Orlova A, Wikman M, et al. Affibody-mediated tumour targeting of HER-2–expressing xeno-grafts in mice. Eur J Nucl Med Mol Imaging. 2006 Mar 15[Epub ahead of print].

33. Schier R, McCall A, Adams GP, et al. Isolation ofpicomolar affinity anti-c-erbB-2 single-chain Fv bymolecular evolution of the complementarity determin-ing regions in the center of the antibody binding site.J Mol Biol 1996;263:551–67.

34. Ruther U. pUR 250 allows rapid chemical sequencingof both DNA strands of its inserts. Nucleic Acids Res1982;10:5765–72.

35. Horak EM, Heitner T, Garrison JL, et al. Engineeringbispecific single-chain Fv molecules to alter signaling ofthe EGF receptor family. Proc Am Assoc Cancer Res2001;42:774.

36. Gunneriusson E, Nord K, Uhlen M, Nygren P. Affinitymaturation of a Taq DNA polymerase specific Affibodyby helix shuffling. Protein Eng 1999;12:873–8.

37. Nord K, Nord O, Uhlen M, Kelley B, Ljungqvist C,Nygren PA. Recombinant human factor VIII-specificaffinity ligands selected from phage-displayed combi-natorial libraries of protein A. Eur J Biochem 2001;268:4269–77.

38. Nilsson J, Larsson M, Stahl S, Nygren PA, Uhlen M.Multiple affinity domains for the detection, purificationand immobilization of recombinant proteins. J MolRecognit 1996;9:585–94.

39. Yokota T, Milenic DE, Whitlow M, Schlom J. Rapidtumor penetration of a single-chain Fv and comparisonwith other immunoglobulin forms. Cancer Res 1992;52:3402–8.

40. Batra SK, Jain M, Wittel UA, Chauhan SC, Colcher D.Pharmacokinetics and biodistribution of geneticallyengineered antibodies. Curr Opin Biotechnol 2002;13:603–8.

41. Holt LJ, Herring C, Jespers LS, Woolven BP,Tomlinson IM. Domain antibodies: proteins for therapy[review]. Trends Biotechnol 2003;21:484–90.

42. Cortez-Retamozo V, Lauwereys M, Hassanzadeh GhG, et al. Efficient tumor targeting by single-domainantibody fragments of camels. Int J Cancer 2002;98:456–62.

43. Adams GP, Schier R, McCall AM, et al. Prolongedin vivo tumour retention of a human diabody targetingthe extracellular domain of human HER2/neu. Br JCancer 1998;77:1405–12.

44. Adams GP, McCartney JE, Tai MS, et al. Highlyspecific in vivo tumor targeting by monovalent anddivalent forms of 741F8 anti-c-erbB-2 single-chain Fv.Cancer Res 1993;53:4026–34.

45. Tai MS, McCartney JE, Adams GP, et al. Targetingc-erbB-2 expressing tumors using single-chain Fvmonomers and dimers. Cancer Res 1995;55:5983s–9s.

46. Nielsen UB, Adams GP, Weiner LM, Marks JD.Targeting of bivalent anti-ErbB2 diabody antibodyfragments to tumor cells is independent of the intrinsicantibody affinity. Cancer Res 2000;60:6434–40.

47. Olafsen T, Kenanova VE, Sundaresan G, et al.Optimizing radiolabeled engineered anti-p185�HER2

antibody fragments for in vivo imaging. Cancer Res2005;65:5907–16.

48. Smith-Jones PM, Solit DB, Akhurst T, et al. Imagingthe pharmacodynamics of HER2 degradation in responseto Hsp90 inhibitors. Nat Biotechnol 2004;22:701–6.

Cancer Research


by Gloria J. Kim and Shuming Nie*

Department of Biomedical Engineering,

Emory University and Georgia Institute of Technology,

101 Woodruff Circle Suite 2001,

Atlanta, GA 30322, USA

*E-mail: [email protected]

Significant progress has been made in the

development of new agents against cancer and new

delivery technologies. Proteomics and genomics

continue to uncover molecular signatures that are

unique to cancer. Yet, the major challenge remains in

targeting and selectively killing cancer cells while

affecting as few healthy cells as possible.

Nanometer-sized particles have novel optical,

electronic, and structural properties that are not

available from either individual molecules or bulk

solids. When linked with tumor-targeting moieties,

such as tumor-specific ligands or monoclonal

antibodies, these nanoparticles can be used to target

cancer-specific receptors, tumor antigens

(biomarkers), and tumor vasculatures with high

affinity and precision.

Conventional cancer therapy and diagnostics involves

the application of catheters, surgery, biopsy,

chemotherapy, and radiation. Most current anticancer

agents do not greatly differentiate between

cancerous and normal cells. This leads to systemic

toxicity and adverse effects. Consequently, the

systemic application of these drugs often causes

severe sideeffects in other tissues (e.g. bone marrow

suppression, cardiomyopathy, and neurotoxicity),

which greatly limits the maximal allowable dose of

the drug. In addition, rapid elimination and

widespread distribution into nontargeted organs and

tissues requires the administration of a drug in large

quantities, which is uneconomical and is often

complicated because of nonspecific toxicity.

Nanotechnology could offer a less invasive alternative,

enhancing the life expectancy and quality of life of the

patient. The diameter of human cells spans 10-20 µm. The size

of cell organelles ranges from a few nanometers to a few

hundred nanometers. Nanoscale devices can readily interact

with biomolecules on the cell surface and within the cells in

a noninvasive manner, leaving the behavior and biochemical

properties of those molecules intact. In their ‘mesoscopic’

size range of 10-100 nm in diameter, nanoparticles have

more surface areas and functional groups that can be linked

to multiple optical, radioisotopic, or magnetic diagnostic and

therapeutic agents. When linked with tumor-targeting

ligands such as monoclonal antibodies, these nanoparticles

can be used to target tumor antigens (biomarkers), as well as

tumor vasculatures with high affinity and specificity. In this

cancer nanotherapy

ISSN:1369 7021 © Elsevier Ltd 2005

Targeted

August 200528

nano2_Nie 07/13/2005 12:03 Page 28


RESEARCH REPORT

article, we discuss different targeting strategies for nanoscale

drug delivery systems (see Scheme 1), and offer a perspective

on cancer nanotherapy.

Passive targetingSolid tumors have a diffusion-limited maximal size1,2 of

about 2 mm3 and will remain at this size until angiogenesis

occurs, thus granting them access to the circulation3. Rapid

vascularization to serve fast-growing cancerous tissues

inevitably leads to a leaky, defective architecture and

impaired lymphatic drainage. This structure allows an

enhanced permeation and retention (EPR) effect4,5 (first

described by Matsumura et al.6), as a result of which

nanoparticles accumulate at the tumor site. For such a

passive targeting mechanism to work, the size and surface

properties of drug delivery nanoparticles must be controlled

to avoid uptake by the reticuloendothelial system (RES)7. To

maximize circulation times and targeting ability, the optimal

size should be less than 100 nm in diameter and the surface

should be hydrophilic to circumvent clearance by

macrophages (large phagocytic cells of the RES). A

hydrophilic nanoparticle surface safeguards against plasma

protein adsorption, and can be achieved through hydrophilic

polymer coating (e.g. by polyethylene glycol (PEG),

poloxamines, poloxamers, and polysaccharides) or the use of

branched or block copolymers8,9. The covalent linkage of

amphiphilic copolymers (polylactic acid, polycaprolactone,

and polycyanonacrylate) chemically coupled to PEG9-11 is

generally preferred, as it avoids aggregation and ligand

desorption when in contact with blood components7.

An alternative passive targeting strategy is to use the

unique tumor environment in a scheme called tumor-activated

August 2005 29

Passive Targeting

Active Targeting

EPR Effect

Tumor Environment

Direct Local Delivery

Lectin-carbohydrate Interaction

Ligand-receptor Interaction

Antibody-antigen Interaction

Scheme 1. Targeting strategies for nanoscale drug delivery systems.

Glossary

Agonist A drug that triggers an action from a cell or

another drug.

Angiogenesis The process of vascularization of a

tissue involving the development of new capillary blood

vessels. This is caused by the release of chemicals by

the tumor.

Antagonist A compound that inhibits the effect of a

hormone or drug.

Apoptosis A form of programmed cell death induced

by external or internal signals that trigger the activity

of proteolytic caspases, whose actions dismantle the

cell and result in cell death.

Doxorubicin A type of anticancer drug called an

‘anthracycline glycoside’. It works by impairing DNA

synthesis, and hence can target rapidly dividing cells.

Epitope A part of a molecule that an antibody will

recognize and bind to.

Integrin Superfamily of cell surface proteins that are

involved in binding to extracellular matrix components

in some cases.

Liposome Lipid bilayers that form colloidal particles in

an aqueous medium.

Matrix metalloproteinase A member of a group of

enzymes that can break down proteins, such as

collagen, normally found in the spaces between cells

in tissues (i.e. extracellular matrix proteins). They

need Zn or Ca ions to work properly. Matrix

metalloproteinases are involved in wound healing,

angiogenesis, and tumor cell metastasis.

Reticuloendothelial system A system composed of

monocytes and macrophages located in reticular

connective tissue. These cells are responsible for

engulfing (phagocytosis) and removing cellular debris,

old cells, pathogens, and foreign substances from the

bloodstream.

Xenobiotics Chemicals that are foreign to the body.

Xenograft Cells of one species transplanted to another.

nano2_Nie 07/13/2005 12:03 Page 29

prodrug therapy. The drug is conjugated to a tumor-specific

molecule and remains inactive until it reaches the target12

(Fig. 1). Overexpression of the matrix metalloproteinase

(MMP), MMP-2, in melanoma has been shown in a number of

preclinical as well as clinical investigations. Mansour et al.13

reported a water-soluble maleimide derivative of doxorubicin

(DOX) incorporating an MMP-2-specific peptide sequence

(Gly-Pro-Leu-Gly-Ile-Ala-Gly-Gln) that binds rapidly and

selectively to the cysteine-34 position of circulating albumin.

The albumin-doxorubicin conjugate is cleaved efficiently and

specifically by MMP-2, releasing a doxorubicin tetrapeptide

(Ile-Ala-Gly-Gln-DOX) and subsequently doxorubicin. pH and

redox potential have also been explored as drug-release

triggers at the tumor site14.

Yet another passive targeting method is the direct local

delivery of anticancer agents to tumors. This approach has

the obvious advantage of excluding the drug from the

systemic circulation. However, administration can be highly

invasive, as it involves injections or surgical procedures. For

some tumors that are difficult to access, such as lung cancers,

the technique is nearly impossible to use.

Active targetingActive targeting is usually achieved by conjugating to the

nanoparticle a targeting component that provides preferential

accumulation of nanoparticles in the tumor-bearing organ, in

the tumor itself, individual cancer cells, intracellular

organelles, or specific molecules in cancer cells. This approach

is based on specific interactions such as lectin-carbohydrate,

ligand-receptor, and antibody-antigen.

Lectin-carbohydrate is one of the classic examples for

targeted drug delivery15. Lectins are proteins of

nonimmunological origin that are capable of recognizing and

binding to glycoproteins expressed on cell surfaces. Lectin

interactions with certain carbohydrates are very specific.

Carbohydrate moieties can be used to target drug delivery

systems to lectins (direct lectin targeting), and lectins can be

used as targeting moieties to target cell surface

carbohydrates (reverse lectin targeting). However, drug

delivery systems based on lectin-carbohydrate have been

developed mainly to target whole organs16, which can harm

normal cells. Therefore, in most cases, the targeting moiety is

directed toward specific receptors or antigens expressed on

the plasma membrane or elsewhere at the tumor site.

Multiple drug resistance (MDR), which is a major challenge

in chemotherapy, often stems from the overexpression of the

plasma membrane P-glycoprotein (Pgp)17,18. In general, Pgp

acts as an efflux pump to extrude positively charged

xenobiotics – including some anticancer drugs – out of the

cell. Many tumor cells are resistant to doxorubicin, which is a

Pgp substrate. To overcome the resistance,

poly(cyanocarylate) nanoparticles have been developed19,20.

Adsorption of the nanoparticles onto the plasma membrane

and the subsequent release of doxorubicin leads to saturation

of Pgp. Furthermore, the negatively charged degradation

products of the polymer form an ion pair and neutralize the

positive charge of doxorubicin19, enhancing the diffusion of

the drug across the plasma membrane. Blagosklonny

proposed an approach to selectively kill resistant cancer cells

that is based on a temporary increase in the resistance of

sensitive cells against certain drugs by specific protectors,

such as pharmacological inhibitors of apoptosis21. These

protectors are pumped out by MDR cells, while increasing the

resistance in sensitive cells that do not have active drug

efflux pumps. After applying a cytotoxic drug, sensitive cells

are protected and survive the exposure, while unprotected

MDR counterparts are killed. By abolishing dose-limiting side-

effects of chemotherapy, this strategy might provide a means

to selectively treat aggressive and resistant cancers. Tsuruo

suggested that antibodies to P-glycoprotein overexpressed on

multidrug resistant (MDR) cells could make an attractive

targeting moiety22.

The overexpression of receptors or antigens in many

human cancers lends itself to efficient drug uptake via

receptor-mediated endocytosis (cellular ingestion) – see Fig. 2.

RESEARCH REPORT

nucleusnucleus

Anticancer

agent

Biocompatible polymer

Cleavable linkage

responsive to tumor environment

Anticancer

agent

Biocompatible polymer

Cleavable linkage

responsive to tumor environment

Fig. 1 Tumor-activated prodrug delivery and targeting. The anticancer agent is conjugated

to a biocompatible polymer via an ester bond. The linkage is hydrolyzed by cancer-specific

enzymes, or by high or low pH, at the tumor site, at which time the nanoparticle releases

the drug.

August 200530

nano2_Nie 07/13/2005 12:03 Page 30

RESEARCH REPORT

Since glycoproteins cannot remove polymer-drug conjugates

that have entered the cells via endocytosis23,24, this active

targeting mechanism provides an alternative route for

overcoming MDR.

The cell surface receptor for folate is inaccessible from the

circulation to healthy cells because of its location on the apical

membrane of polarized epithelia, but is overexpressed on the

surface of various cancers, including ovary, brain, kidney,

breast, and lung malignancies. Surface plasmon resonance

studies reveal that folate-conjugated PEGylated cyanoacrylate

nanoparticles have a ten-fold higher affinity for the folate

receptor than free folate11. Folate receptors are often

organized in clusters and bind preferably to the multivalent

forms of the ligand. Furthermore, confocal microscopy

demonstrated selective uptake and endocytosis of folate-

conjugated nanoparticles by tumor cells that bear folate

receptors. Interest in exploiting folate receptor targeting in

cancer therapy and diagnosis has increased rapidly, as attested

by many conjugated systems, including proteins, liposomes,

imaging agents, and neutron activation compounds9,11,25-31.

Tumor targeting by antibodies with engineered properties

(Table 1) is in its infancy, but holds much promise. The

monoclonal antibody (mAb) BR96 (anti-sialyl Lewis Y

antigen), conjugated with doxorubicin, has proved highly

efficacious in tumour xenograft studies32 but, unfortunately,

has shown little or no efficacy in Phase II trials for metastatic

breast cancer33 and advanced gastric adenocarcinoma34,

respectively. Moreover, dose-limiting gastrointestinal

toxicities are observed in the breast cancer trial, because the

immunoconjugate binds to antigen-positive normal cells in

gastric mucosa, small intestine, and pancreas33.

Calicheamicins35-37 and maytansinoids38 are the most

extensively evaluated of many small-molecule toxins that are

used for direct antibody arming, but indirect arming has met

with better success. For example, anti-ERBB2

immunoliposomes loaded with doxorubicin show greater

antitumor activity than the free drug or the drug loaded in

nontargeted liposomes in several tumor xenograft

models39,40. Moreover, the systemic toxicity of the

immunoliposome-targeted doxorubicin was much less than

that of free doxorubicin. Bispecific antibodies, which are non-

natural antibodies with two different epitopes, have been

used most widely for the delivery of immune effector cells

and, to a lesser extent, for the delivery of radionuclides,

drugs, and toxins to tumors41,42.

Alternatively, tumor vasculatures can be targeted to allow

targeted delivery to a wide range of tumor types. A number of

angiogenesis inhibitors are undergoing clinical trials (Table 2).

Antiangiogenic therapy prevents neovascularization (the

proliferation of blood vessels in different tissues) by

inhibiting proliferation, migration, and differentiation of

endothelial cells43. Vascular endothelial growth factor (VEGF)

is expressed in many solid tumors. A potent angiogenesis-

1. Internalization

2. Early Endosomes

6. Membrane Fusion

3. Acidified Endosomes

5. Recycling Endosomes

4. Endosomal Release

Cancer Cell SurfaceH+

H+

H+

H+

H+

H+ H+

H+

H+

H+

H+

H+ H+

H+

H+

Cargo

Ligand

Cleavable Bond

Receptor

Fig. 2 Nanoparticle drug delivery and targeting using receptor-mediated endocytosis.

August 2005 31

Table 1 Antibodies in cancer therapeutics.

Mechanism Antibody target Trade name

Agonist activity CD40, CD137 Various

Antagonist activity CTLA4 MDX-010

Angiogenesis inhibition VEGF Avastin™

Antibody-dependent cell-mediated cytotoxicity CD20 Rituxan®, HuMax-CD20

HER-2/neu Herceptin®

EGF receptor HuMax-EGFr

Toxin-mediated killing CD33 Mylotarg®

Disruption signaling HER-2/neu Pertuzumab (2C4)

Complement-dependent cytotoxicity CD20 Rituxan®, HuMax-CD20

Blockage ligand binding EGF receptor Erbitux™

nano2_Nie 07/13/2005 12:03 Page 31

stimulating protein, it also increases the permeability of

tumor blood vessels. This leads to swelling of the tumor,

ultimately hindering the ability of cancer cells to recruit

blood supply through angiogenesis. VEGF has been used in

liposomes and polymeric nanospheres to deliver angiostatin

and endostatin44,45. The αvβ3 integrin is one of the most

specific biomarkers that can differentiate newly formed

capillaries from their mature counterparts46. Although all

endothelial cells use integrin receptors to attach to the

extraluminal submatrix, one unique receptor (αvβ3 integrin)

is found on the luminal surface of the endothelial cell only

during angiogenesis47. High-affinity αvβ3 selective ligands,

Arg-Gly-Asp (RGD), have been identified by phage display

studies. The cyclic form, which contains a conformationally

constrained RGD, has a higher binding affinity than the linear

form48. Doxorubicin-loaded PEG nanoparticles conjugated to

cyclic RGD49 and paclitaxel-cyclic RGD nanoparticles50 have

been reported recently.

PerspectiveNanotechnology is beginning to change the scale and

methods of drug delivery. For decades, researchers have been

developing new anticancer agents and new formulations for

delivering existing and new agents.

More than 40% of active substances identified through

combinatorial screening programs are poorly soluble in

water51. The conventional, and most current, formulations of

such drugs are frequently plagued with problems such as poor

and inconsistent bioavailability. The most widely used

method for enhancing solubility is to generate a salt. For

nonionizable compounds, micronization, soft-gel technology,

cosolvents, surfactants, or complexing agents have been

used52,34. Since it is faster and more cost effective to

redesign the molecule than to develop a new one, a broadly

based technology applicable to poorly water-soluble drugs

could have a tremendous impact.

Paclitaxel (Taxol™) is a microtubule-stabilizing agent that

promotes tubulin polymerization, disrupting cell division and

causing cell death54. It displays neoplastic activity against

primary epithelial ovarian carcinoma, breast, colon, and lung

cancers. Because it is poorly soluble in aqueous solution, the

formulation that is currently available is Chremophor EL

(polyethoxylated castor oil) and ethanol55. In a new

formulation used in Abraxane (recently approved by the FDA

to treat metastatic breast cancer), paclitaxel was conjugated

RESEARCH REPORT

nucleus

Receptor-mediated

endocytosis

Release of the anticancer agent

Solid tumor

Receptor

nucleus

Solid tumor

Receptor

nucleus

Anticancer

agent

Self-assembled polymerTargeting ligand

100 nm

Tumor specific

enzyme cleavable linkageCovalent bond

Anticancer

agent

Self-assembled polymerTargeting ligand

100 nm

Tumor specific

enzyme cleavable linkageCovalent bond

Fig. 3 Self-assembled polymeric nanoparticles for both tumor targeting and therapeutic

functions. Inset: delivery of the nanoparticle drugs by receptor-mediated endocytosis and

controlled drug release inside the cytoplasm.

August 200532

Table 2 Angiogenesis inhibitors undergoing clinical trial.

Mechanism Antiangiogenic drug Target Stage of clinical development

Monoclonal antibodies targeting VEGF-A Avastin™ VEGF-A Phase I, II, III

VEGF-Trap VEGF-A Phase I

Antibodies targeting VEGFR-2 IMC-ICII VEGFR-2 Phase I

Receptor tyrosine kinase inhibitors SU5416 VEGFR-2 Phase I, II, III

SU6668 VEGFR-2, bFGFR, PDGFR Phase I, II

ZD 6474 VEGFR-2, EGFR Phase I, II

Endothelial cell proliferation inhibitors ABT-510 Endothelial CD-36 Phase I, II

Angiostatin Various Phase I

Endostatin Various Phase I

TNP-470 Methionine aminopeptidase, Phase I

cyclin dependent kinase 2

Integrin activity inhibitors Vitaxin Integrin ανβ3 Phase I, II

Medi-522 Integrin αvβ3 Phase I

Vascular targeting agents ZD 6126 Endothelial tubulin Phase I

DMXAA Endothelial tubulin Phase I

nano2_Nie 07/13/2005 12:03 Page 32

RESEARCH REPORT

to albumin nanoparticles56. In addition to circumventing side-

effects of the highly toxic Chremophor EL, which include

hypersensitivity reactions and toxicity to kidney cells and

nerve tissue (nephrotoxicity and neurotoxicity)55,57, the

system cleverly takes advantage of albumin receptors for

improved drug delivery to cancer cells.

For specific targeting, the differences between cancerous

cells and normal cells, which include uncontrolled

proliferation, insensitivity to negative growth regulation, and

antigrowth signals, angiogenesis and metastasis can be

exploited. Thanks to recent advances in proteomics and

genomics, there is a growing body of knowledge of unique

cancer markers. They form the basis of complex interactions

between bioconjugated nanoparticles and cancer cells. Carrier

design and targeting strategies may vary according to the

type, developmental stage, and location of cancer. There is

much synergy between imaging and nanotechnology in

biomedical applications. Many of the principles used to target

delivery of drugs to cancers may also be applied to target

imaging and diagnostic agents. The full in vivo potential of

cancer nanotechnology in targeted drug delivery and imaging

can be realized by using strategically engineered

multifunctional nanoparticles (Figs. 3 and 4). NT

REFERENCES

1. Grossfeld, G. D., et al., Urology (2002) 59 (1), 97

2. Storm, G., et al., Adv Drug Deliv Rev. (1995) 17 (1), 31

3. Jain, R. K., Annu. Rev Biomed Eng. (1999) 1, 241

4. Teicher, B. A., Drug Resist. Update (2000) 3 (2), 67

5. Duncan, R., Nat. Rev. Drug Discov. (2003) 2 (5), 347

6. Matsumura, Y., and Maeda, H., Cancer Res. (1986) 46 (12), 6387

7. Gref, R., et al., Science (1994) 263, 1600

8. Moghimi, S. M., and Hunter, A. C., Trends Biotechnol. (2000) 18 (10), 412

9. Park, E. K., et al., Biomaterials (2005) 26 (9), 1053

10. Farokhzad, O. C., et al., Cancer Res. (2004) 64 (21), 7668

11. Stella, B., et al., J. Pharm Sci. (2000) 89 (11), 1452

12. Chari, R. V., Adv. Drug Delivery Rev. (1998) 31 (1-2), 89

13. Mansour, A. M., et al., Cancer Res. (2003) 63 (14), 4062

14. Guo, X.and Szoka, F. C., Jr., Acc. Chem. Res. (2003) 36 (5), 335

15. Kannagi, R., et al., Cancer Sci. (2004) 95 (5), 377

16. Yamazaki, N., et al., Adv. Drug Delivery Rev. (2000) 43 (2-3), 225

17. Links, M.and Brown, R., Expert Rev. Mol. Med. (1999) 1999, 1

18. Krishna, R.and Mayer, L. D., Eur. J. Pharm. Sci. (2000) 11 (4), 265

19. Vauthier, C., et al., J. Controlled Release (2003) 93 (2), 151

20. de Verdiere, A. C., et al., Br. J. Cancer (1997) 76 (2), 198

21. Blagosklonny, M. V., Trends Mol. Med. (2003) 9 (7), 307

22. Tsuruo, T., Intern Med. (2003) 42 (3), 237

23. Bennis, S., et al., Eur. J. Cancer (1994) 30A (1), 89

24. Larsen, A. K., et al., Pharmacol. Ther. (2000) 85 (3), 217

25. Xu, L., et al., J. Controlled Release (2001) 74 (1-3), 115

26. Wang, S., et al., Proc. Natl. Acad. Sci. U.S.A. (1995) 92 (8), 3318

27. Shukla, S., et al., Bioconjugate Chem. (2003) 14 (1), 158

28. Moon, W. K., et al., Bioconjugate Chem. (2003) 14 (3), 539

29. Leamon, C. P., and Low, P. S., J. Drug Target. (1994) 2 (2), 101

30. Konda, S. D., et al., MAGMA (2001) 12 (2-3) 104

31. Goren, D., et al., Clin. Cancer Res. (2000) 6 (5), 1949

32. Trail, P. A., et al., Science (1993) 261, 212

33. Tolcher, A. W., et al., J. Clin. Oncol. (1999) 17 (2), 478

34. Ajani, J. A., et al., Cancer J. (2000) 6, 78

35. Lode, H. N., et al., Cancer Res. (1998) 58 (14), 2925

36. Hinman, L. M., et al., Cancer Res. (1993) 53 (14), 3336

37. Sievers, E. L., et al., J. Clin. Oncol. (2001) 19 (13), 3244

38. Liu, C., et al., Proc. Natl. Acad. Sci. U.S.A. (1996) 93 (16), 8618

39. Park, J. W., et al., Adv. Pharmacol. (1997) 40, 399

40. Park, J. W., et al., Cancer Lett. (1997) 118 (2), 153

41. Koelemij, R., et al., J. Immunother. (1999) 22 (6), 514

42. Segal, D. M., et al., Curr. Opin. Immunol. (1999) 11 (5), 558

43. Jain, R. K., Science (2005) 307, 58

44. Chen, Q. R., et al., Mol. Genet. Metab. (2001) 74, 120

45. Reynolds, A. R., et al., Trends Mol. Med. (2003) 9 (1), 2

46. Brooks, P. C., et al., Cell (1994) 79 (7), 1157

47. Cleaver, O., and Melton, D. A., Nat. Med. (2003) 9 (6), 661

48. Cheng, S., et al., J. Med. Chem. (1994) 37 (1), 1

49. Bibby, D. C., et al., Int. J. Pharm. (2005) 293 (1-2) 281

50. Chen, X., et al., J. Med. Chem. (2005) 48 (4), 1098

51. Merisko-Liversidge, E., et al., Eur. J. Pharm. Sci. (2003) 18 (2), 113

52. Fishman, M. L., et al., Biomacromolecules (2004) 5 (2), 334

53. Stella, V. J., and Rajewski, R. A., Pharm. Res. (1997) 14 (5), 556

54. Diaz, J. F., et al., J. Biol. Chem. (2000) 275 (34), 26265

55. Singla, A. K., et al., Int. J. Pharm. (2002) 235 (1-2), 179

56. Garber, K., J. Natl. Cancer Inst. (2004) 96 (2), 90

57. Gelderblom, H., et al., Eur. J. Cancer (2001) 37 (13), 1590

August 2005 33

Imaging agentAnticancer agent

Self-assembled polymer

Targeting ligand100 nm

Tumor specific

enzyme cleavable linkage

Covalent bond

100 nm

Imaging agentAnticancer agent

Self-assembled polymer

Targeting ligand100 nm100 nm

Tumor specific

enzyme cleavable linkage

Covalent bond

100 nm100 nm

Fig. 4 Multifunctional nanoparticles for integrated cancer imaging and therapy. As

nanoparticles are developed for clinical applications, an exciting opportunity is to monitor

drug delivery and treatment efficacy by using embedded imaging agents.

nano2_Nie 07/13/2005 12:03 Page 33

INSTITUTE OF PHYSICS PUBLISHING JOURNAL OF PHYSICS D: APPLIED PHYSICS

J. Phys. D: Appl. Phys. 36 (2003) R167–R181 PII: S0022-3727(03)40035-1

TOPICAL REVIEW

Applications of magnetic nanoparticlesin biomedicineQ A Pankhurst1, J Connolly2, S K Jones3 and J Dobson4

1 Department of Physics and Astronomy, and London Centre for Nanotechnology,University College London, Gower Street, London WC1E 6BT, UK2 Department of Applied Physics, Curtin University of Technology, GPO Box U1987,Perth 6845, Western Australia3 SIRTeX Medical Limited, PO Box 760, North Ryde, New South Wales 2113, Australia4 Department of Biomedical Engineering and Medical Physics, Centre for Science andTechnology in Medicine, Keele University, Stoke-on-Trent ST4 7QB, UK

E-mail: [email protected]

Received 4 April 2002Published 18 June 2003Online at stacks.iop.org/JPhysD/36/R167

AbstractThe physical principles underlying some current biomedical applications ofmagnetic nanoparticles are reviewed. Starting from well-known basicconcepts, and drawing on examples from biology and biomedicine, therelevant physics of magnetic materials and their responses to appliedmagnetic fields are surveyed. The way these properties are controlled andused is illustrated with reference to (i) magnetic separation of labelled cellsand other biological entities; (ii) therapeutic drug, gene and radionuclidedelivery; (iii) radio frequency methods for the catabolism of tumours viahyperthermia; and (iv) contrast enhancement agents for magnetic resonanceimaging applications. Future prospects are also discussed.

1. Introduction

Magnetic nanoparticles offer some attractive possibilities inbiomedicine. First, they have controllable sizes ranging froma few nanometres up to tens of nanometres, which placesthem at dimensions that are smaller than or comparable tothose of a cell (10–100 µm), a virus (20–450 nm), a protein(5–50 nm) or a gene (2 nm wide and 10–100 nm long). Thismeans that they can ‘get close’ to a biological entity of interest.Indeed, they can be coated with biological molecules to makethem interact with or bind to a biological entity, therebyproviding a controllable means of ‘tagging’ or addressing it.Second, the nanoparticles are magnetic, which means thatthey obey Coulomb’s law, and can be manipulated by anexternal magnetic field gradient. This ‘action at a distance’,combined with the intrinsic penetrability of magnetic fieldsinto human tissue, opens up many applications involving thetransport and/or immobilization of magnetic nanoparticles, orof magnetically tagged biological entities. In this way theycan be made to deliver a package, such as an anticancer drug,or a cohort of radionuclide atoms, to a targeted region of the

body, such as a tumour. Third, the magnetic nanoparticlescan be made to resonantly respond to a time-varying magneticfield, with advantageous results related to the transfer of energyfrom the exciting field to the nanoparticle. For example, theparticle can be made to heat up, which leads to their useas hyperthermia agents, delivering toxic amounts of thermalenergy to targeted bodies such as tumours; or as chemotherapyand radiotherapy enhancement agents, where a moderatedegree of tissue warming results in more effective malignantcell destruction. These, and many other potential applications,are made available in biomedicine as a result of the specialphysical properties of magnetic nanoparticles.

In this paper we will address the underlying physics ofthe biomedical applications of magnetic nanoparticles. Afterreviewing some of the relevant basic concepts of magnetism,including the classification of different magnetic materialsand how a magnetic field can exert a force at a distance,we will consider four particular applications: magneticseparation, drug delivery, hyperthermia treatments andmagnetic resonance imaging (MRI) contrast enhancement. Wewill conclude with an outlook on future prospects in this area,

0022-3727/03/130167+15$30.00 © 2003 IOP Publishing Ltd Printed in the UK R167

http://stacks.iop.org/jd/36/R167

Topical Review

especially with regard to the challenges faced in bringingcurrent laboratory-tested technologies into mainstream use.

2. Basic concepts

2.1. M–H curves

Figure 1 shows a schematic diagram of a blood vessel intowhich some magnetic nanoparticles have been injected. Themagnetic properties of both the injected particles and theambient biomolecules in the blood stream are illustrated bytheir different magnetic field response curves. To understandthese curves better, we need to be aware of some of thefundamental concepts of magnetism, which will be recalledbriefly here. Further details can be found in one of the manyexcellent textbooks on magnetism (e.g. [1, 2]).

If a magnetic material is placed in a magnetic field ofstrength H, the individual atomic moments in the materialcontribute to its overall response, the magnetic induction:

B = µ0(H + M), (1)

where µ0 is the permeability of free space, and themagnetization M = m/V is the magnetic moment per unit

H

M

DM(x1000)

H

M

PM(x100)

H

M

FM

H

M

SPM

Figure 1. Magnetic responses associated with different classes ofmagnetic material, illustrated for a hypothetical situation in whichferromagnetic particles of a range of sizes from nanometre up tomicron scale are injected into a blood vessel. M–H curves areshown for diamagnetic (DM) and paramagnetic (PM) biomaterialsin the blood vessel, and for the ferromagnetic (FM) injectedparticles, where the response can be either multi-domain (- - - - inFM diagram), single-domain (—— in FM diagram) orsuperparamagnetic (SPM), depending on the size of the particle.

volume, where m is the magnetic moment on a volume V

of the material. All materials are magnetic to some extent,with their response depending on their atomic structure andtemperature. They may be conveniently classified in terms oftheir volumetric magnetic susceptibility, χ , where

M = χH, (2)

describes the magnetization induced in a material by H.In SI units χ is dimensionless and both M and H areexpressed in A m−1. Most materials display little magnetism,and even then only in the presence of an applied field;these are classified either as paramagnets, for which χ

falls in the range 10−6–10−1, or diamagnets, with χ in therange −10−6 to −10−3. However, some materials exhibitordered magnetic states and are magnetic even without a fieldapplied; these are classified as ferromagnets, ferrimagnetsand antiferromagnets, where the prefix refers to the natureof the coupling interaction between the electrons within thematerial [2]. This coupling can give rise to large spontaneousmagnetizations; in ferromagnets M is typically 104 times largerthan would appear otherwise.

The susceptibility in ordered materials depends not juston temperature, but also on H, which gives rise to thecharacteristic sigmoidal shape of the M–H curve, withM approaching a saturation value at large values of H.Furthermore, in ferromagnetic and ferrimagnetic materialsone often sees hysteresis, which is an irreversibility in themagnetization process that is related to the pinning of magneticdomain walls at impurities or grain boundaries within thematerial, as well as to intrinsic effects such as the magneticanisotropy of the crystalline lattice. This gives rise to openM–H curves, called hysteresis loops. The shape of theseloops are determined in part by particle size: in large particles(of the order micron size or more) there is a multi-domainground state which leads to a narrow hysteresis loop since ittakes relatively little field energy to make the domain wallsmove; while in smaller particles there is a single domainground state which leads to a broad hysteresis loop. At evensmaller sizes (of the order of tens of nanometres or less) onecan see superparamagnetism, where the magnetic momentof the particle as a whole is free to fluctuate in response tothermal energy, while the individual atomic moments maintaintheir ordered state relative to each other. This leads to theanhysteretic, but still sigmoidal, M–H curve shown in figure 1.

The underlying physics of superparamagnetism is foundedon an activation law for the relaxation time τ of the netmagnetization of the particle [3, 4]:

τ = τ0 exp

(�E

kBT

), (3)

where �E is the energy barrier to moment reversal, andkBT is the thermal energy. For non-interacting particlesthe pre-exponential factor τ0 is of the order 10−10–10−12 sand only weakly dependent on temperature [4]. The energybarrier has several origins, including both intrinsic andextrinsic effects such as the magnetocrystalline and shapeanisotropies, respectively; but in the simplest of cases ithas a uniaxial form and is given by �E = KV , whereK is the anisotropy energy density and V is the particle

R168

Topical Review

volume. This direct proportionality between �E and V isthe reason that superparamagnetism—the thermally activatedflipping of the net moment direction—is important for smallparticles, since for them �E is comparable to kBT at, say,room temperature. However, it is important to recognize thatobservations of superparamagnetism are implicitly dependentnot just on temperature, but also on the measurement time τm

of the experimental technique being used (see figure 2). Ifτ � τm the flipping is fast relative to the experimental timewindow and the particles appear to be paramagnetic (PM);while if τ � τm the flipping is slow and quasi-static propertiesare observed—the so-called ‘blocked’ state of the system. A‘blocking temperature’ TB is defined as the mid-point betweenthese two states, where τ = τm. In typical experiments τm

can range from the slow to medium timescales of 102 s for DCmagnetization and 10−1–10−5 s for AC susceptibility, throughto the fast timescales of 10−7–10−9 s for 57Fe Mossbauerspectroscopy.

All of the different magnetic responses discussed aboveare illustrated in figure 1 for the case of ferromagneticor ferrimagnetic nanoparticles injected into a blood vessel.Depending on the particle size, the injected material exhibitseither a multi-domain, single-domain or superparamagnetic(SPM) M–H curve. The magnetic response of the bloodvessel itself includes both a PM response—for example, fromthe iron-containing haemoglobin molecules, and a diamagnetic(DM) response—for example, from those intra-vessel proteinsthat comprise only carbon, hydrogen, nitrogen and oxygenatoms. It should be noted that the magnetic signal from theinjected particles, whatever their size, far exceeds that fromthe blood vessel itself. This heightened selectivity is one of theadvantageous features of biomedical applications of magneticnanoparticles.

Returning to the hysteresis which gives rise to the openM–H curves seen for ferromagnets and antiferromagnets,it is clear that energy is needed to overcome the barrier to

(a) (b)

Figure 2. Illustration of the concept of superparamagnetism, wherethe circles depict three magnetic nanoparticles, and the arrowsrepresent the net magnetization direction in those particles.In case (a), at temperatures well below themeasurement-technique-dependent blocking temperature TB of theparticles, or for relaxation times τ (the time between momentreversals) much longer than the characteristic measurement time τm,the net moments are quasi-static. In case (b), at temperature wellabove TB, or for τ much shorter than τm, the moment reversals are sorapid that in zero external field the time-averaged net moment on theparticles is zero.

domain wall motion imposed by the intrinsic anisotropy andmicrostructural impurities and grain boundaries in the material.This energy is delivered by the applied field, and can becharacterized by the area enclosed by the hysteresis loop.This leads to the concept that if one applies a time-varyingmagnetic field to a ferromagnetic or ferrimagnetic material,one can establish a situation in which there is a constant flowof energy into that material, which will perforce be transferredinto thermal energy. This is the physical basis of hyperthermiatreatments, which are discussed further in section 5. Note thata similar argument regarding energy transfer can be made forSPM materials, where the energy is needed to coherently alignthe particle moments to achieve the saturated state; this also isdiscussed in more detail in section 5.

2.2. Forces on magnetic nanoparticles

To understand how a magnetic field may be used to manipulatemagnetic nanoparticles, we need to recall some elements ofvector field theory. This is not always intuitive, and the readeris directed to recent reviews for further details [5–7]. It is alsoimportant to recognize that a magnetic field gradient is requiredto exert a force at a distance; a uniform field gives rise to atorque, but no translational action. We start from the definitionof the magnetic force acting on a point-like magnetic dipole m:

Fm = (m · ∇)B, (4)

which can be geometrically interpreted as differentiation withrespect to the direction of m. For example, if m = (0, 0, mz)

then m ·∇ = mz(∂/∂z) and a force will be experienced on thedipole provided there is a field gradient in B in the z-direction.In the case of a magnetic nanoparticle suspended in a weaklyDM medium such as water, the total moment on the particlecan be written m = VmM, where Vm is the volume of theparticle and M is its volumetric magnetization, which in turnis given by M = �χH, where �χ = χm − χw is the effectivesusceptibility of the particle relative to the water. For the caseof a dilute suspension of nanoparticles in pure water, we canapproximate the overall response of the particles plus watersystem by B = µ0H, so that equation (4) becomes:

Fm = Vm�χ

µ0(B · ∇)B. (5)

Furthermore, provided there are no time-varying electric fieldsor currents in the medium, we can apply the Maxwell equation∇ × B = 0 to the following mathematical identity:

∇(B · B) = 2B × (∇ × B) + 2(B · ∇)B = 2(B · ∇)B, (6)

to obtain a more intuitive form of equation (5):

Fm = Vm�χ∇(

B2

2µ0

)or

Fm = Vm�χ∇ (12 B · H

),

(7)

in which the magnetic force is related to the differential of themagnetostatic field energy density, 1

2 B · H. Thus, if �χ > 0the magnetic force acts in the direction of steepest ascent of theenergy density scalar field. This explains why, for example,when iron filings are brought near the pole of a permanentbar magnet, they are attracted towards that pole. It is also thebasis for the biomedical applications of magnetic separationand drug delivery, as will be discussed in sections 3 and 4.

R169

Topical Review

3. Magnetic separation

3.1. Cell labelling and magnetic separation

In biomedicine it is often advantageous to separate outspecific biological entities from their native environmentin order that concentrated samples may be prepared forsubsequent analysis or other use. Magnetic separation usingbiocompatible nanoparticles is one way to achieve this. Itis a two-step process, involving (i) the tagging or labellingof the desired biological entity with magnetic material, and(ii) the separating out of these tagged entities via a fluid-basedmagnetic separation device.

Tagging is made possible through chemical modificationof the surface of the magnetic nanoparticles, usually by coatingwith biocompatible molecules such as dextran, polyvinylalcohol (PVA) and phosopholipids—all of which have beenused on iron oxide nanoparticles [8–10]. As well as providinga link between the particle and the target site on a cell ormolecule, coating has the advantage of increasing the colloidalstability of the magnetic fluid. Specific binding sites on thesurface of cells are targeted by antibodies or other biologicalmacromolecules such as hormones or folic acid [11–13]. Asantibodies specifically bind to their matching antigen thisprovides a highly accurate way to label cells. For example,magnetic particles coated with immunospecific agents havebeen successfully bound to red blood cells [8, 14], lung cancercells [15], bacteria [16], urological cancer cells [17] and Golgivesicles [18]. For larger entities such as the cells, bothmagnetic nanoparticles and larger particles can be used: forexample, some applications use magnetic ‘microspheres’—micron sized agglomerations of sub-micron sized magneticparticles incorporated in a polymeric binder [19].

The magnetically labelled material is separated from itsnative solution by passing the fluid mixture through a region inwhich there is a magnetic field gradient which can immobilizethe tagged material via the magnetic force of equation (7). Thisforce needs to overcome the hydrodynamic drag force actingon the magnetic particle in the flowing solution,

Fd = 6πηRm�v, (8)

where η is the viscosity of the medium surrounding the cell(e.g. water), Rm is the radius of the magnetic particle, and�v = vm −vw is the difference in velocities of the cell and thewater [6]. There is also buoyancy force that affects the motion,but this is dependent on the difference between the density ofthe cell and the water, and for most cases of interest in biologyand medicine can be neglected. Equating the hydrodynamicdrag and magnetic forces, and writing Vm = 4

3πR3m, gives the

velocity of the particle relative to the carrier fluid as:

�v = R2m�χ

9µ0η∇(B2) or �v = ξ

µ0∇(B2), (9)

where ξ is the ‘magnetophoretic mobility’ of the particle—aparameter that describes how manipulable a magnetic particleis. For example, the magnetophoretic mobility of magneticmicrospheres can be much greater than that of nanoparticles,due to their larger size. This can be an advantage, for example,in cell separations, where the experimental timeframe for theseparations is correspondingly shorter. On the other hand,

smaller magnetic particle sizes can also be advantageous, forexample, in reducing the likelihood that the magnetic materialwill interfere with further tests on the separated cells [20].

3.2. Separator design

Magnetic separator design can be as simple as the applicationand removal of a permanent magnet to the wall of a test tubeto cause aggregation, followed by removal of the supernatant(figure 3(a)). However, this method can be limited by slowaccumulation rates [21]. It is often preferable to increase theseparator efficiency by producing regions of high magneticfield gradient to capture the magnetic nanoparticles as theyfloat or flow by in their carrier medium. A typical wayto achieve this is to loosely pack a flow column with amagnetizable matrix of wire (e.g. steel wool) or beads [22]and to pump the magnetically tagged fluid through the columnwhile a field is applied (figure 3(b)). This method is fasterthan in the first case, although problems can arise due to thesettling and adsorption of magnetically tagged material on thematrix. An alternative, rapid throughput method which doesnot involve any obstructions being placed in the column is theuse of specifically designed field gradient systems, such asthe quadrupolar arrangement shown in figure 4 which createsa magnetic gradient radially outwards from the centre of theflow column [23].

As well as separating out the magnetically tagged material,the spatially varying magnitude of the field gradient canbe used to achieve ‘fluid flow fractionation’ [24]. Thisis a process in which the fluid is split at the outlet intofractions containing tagged cells or proteins with differing

(a)

(b)

Figure 3. The standard methods of magnetic separation: in (a) amagnet is attached to the container wall of a solution of magneticallytagged (•) and unwanted (◦) biomaterials. The tagged particles aregathered by the magnet, and the unwanted supernatant solution isremoved. In (b) a solution containing tagged and unwantedbiomaterials flows continuously through a region of strong magneticfield gradient, often provided by packing the column with steel wool,which captures the tagged particles. Thereafter the tagged particlesare recovered by removing the field and flushing through with water.

R170

Topical Review

(a) (b)

Figure 4. A rapid throughput method of magnetic separation, inwhich an annular column containing a flowing solution ofmagnetically tagged (•) and unwanted (◦) biomaterials is placedwithin a set of magnets arranged in quadrature: (a) longitudinalcross-section of the annular column; (b) transverse cross-section ofthe four magnets with the resulting magnetic field lines. Under theaction of the magnetic field gradient the tagged particles move to thecolumn walls, where they are held until the field is removed andthey are recovered by flushing through with water. The central coreof the column is made of non-magnetic material to avoidcomplications due to the near-zero field gradients there.

magnetophoretic mobilities. In a variant of this, the fluid isstatic while an applied magnetic field is moved up the container[25]. The particles move up the container in the resultingfield gradient at a velocity dependent on their magnetophoreticmobility. At the top of the container they enter a removablesection and are held here by a permanent magnet. The bottomsection of the container moves to the next section, a magneticfield with different strength to the first is applied and the processrepeats. The result is a fractionation of the sample into aliquotsof differing magnetophoretic mobility.

3.3. Applications

Magnetic separation has been successfully applied to manyaspects of biomedical and biological research. It has provento be a highly sensitive technique for the selection of raretumour cells from blood, and is especially well suited to theseparation of low numbers of target cells [26]. This has, forexample, led to the enhanced detection of malarial parasitesin blood samples either by utilizing the magnetic properties ofthe parasite [27] or through labelling the red blood cells withan immunospecific magnetic fluid [28]. It has been used asa pre-processing technology for polymerase chain reactions,through which the DNA of a sample is amplified and identified[29]. Cell counting techniques have also been developed. Onemethod estimates the location and number of cells tagged bymeasuring the magnetic moment of the microsphere tags [30],while another uses a giant magnetoresistive sensor to measurethe location of microspheres attached to a surface layered witha bound analyte [31].

In another application, magnetic separation has been usedin combination with optical sensing to perform ‘magneticenzyme linked immunosorbent assays’ [32, 33]. These assaysuse fluorescent enzymes to optically determine the numberof cells labelled by the assay enzymes. Typically the targetmaterial must first be bound to a solid matrix. In a modificationof this procedure the magnetic microspheres act as the surfacefor initial immobilization of the target material and magnetic

separation is used to increase the concentration of the material.The mobility of the magnetic nanoparticles allows a shorterreaction time and a greater volume of reagent to be used thanin standard immunoassays where the antibody is bound to aplate. In a variation of this procedure, magnetic separationhas been used to localize labelled cells at known locationsfor cell detection and counting via optical scanning [14]. Thecells are labelled both magnetically and fluorescently and movethrough a magnetic field gradient towards a plate on which linesof ferromagnetic material have been lithographically etched.The cells align along these lines and the fluorescent tag is usedfor optical detection of the cells.

4. Drug delivery

4.1. Motivation and physical principles

The major disadvantage of most chemotherapies is thatthey are relatively non-specific. The therapeutic drugsare administered intravenously leading to general systemicdistribution, resulting in deleterious side-effects as the drugattacks normal, healthy cells in addition to the target tumourcells. For example, the side effects of anti-inflammatorydrugs on patients who have chronic arthritis can lead to thediscontinuation of their use. However, if such treatments couldbe localized, e.g. to the site of a joint, then the continued use ofthese very potent and effective agents could be made possible.

Recognition of this led researchers in the late 1970sto propose the use of magnetic carriers to target specificsites (generally cancerous tumours) within the body [34–36].The objectives are two-fold: (i) to reduce the amount ofsystemic distribution of the cytotoxic drug, thus reducingthe associated side-effects; and (ii) to reduce the dosagerequired by more efficient, localized targeting of thedrug. In magnetically targeted therapy, a cytotoxic drug isattached to a biocompatible magnetic nanoparticle carrier.These drug/carrier complexes—usually in the form of abiocompatible ferrofluid—are injected into the patient viathe circulatory system. When the particles have entered thebloodstream, external, high-gradient magnetic fields are usedto concentrate the complex at a specific target site within thebody (figure 5). Once the drug/carrier is concentrated at thetarget, the drug can be released either via enzymatic activity orchanges in physiological conditions such as pH, osmolality,or temperature [37], and be taken up by the tumour cells.This system, in theory, has major advantages over the normal,non-targeted methods of cytotoxic drug therapy.

The physical principles underlying magnetic targetingtherapy are similar to those used in magnetic separation,and are derived from the magnetic force exerted on a SPMnanoparticle by a magnetic field gradient, as in equation (7).The effectiveness of the therapy is dependent on severalphysical parameters, including the field strength, gradientand volumetric and magnetic properties of the particles.As the carriers (ferrofluids) are normally administeredintravenously or intra-arterially, hydrodynamic parameterssuch as blood flow rate, ferrofluid concentration, infusionroute and circulation time also will play a major role—as willphysiological parameters such as tissue depth to the target site(i.e. distance from the magnetic field source), reversibility andstrength of the drug/carrier binding, and tumour volume [38].

R171

Topical Review

Magnet

Blood VesselMagneticNanoparticles

Tissue

Tissue

Figure 5. A hypothetical magnetic drug delivery system shown incross-section: a magnet is placed outside the body in order that itsmagnetic field gradient might capture magnetic carriers flowing inthe circulatory system.

In general, larger particles (e.g. magnetic microspheres, ca1 µm in diameter, comprising agglomerates of SPM particles)are more effective at withstanding flow dynamics within thecirculatory system—particularly in larger veins and arteries.In most cases the magnetic field gradient is generated bya strong permanent magnet, such as Nd-Fe-B, fixed outsidethe body over the target site. Preliminary investigationsof the hydrodynamics of drug targeting suggest that formost magnetite-based carriers, flux densities at the targetsite must be of the order of 0.2 T with field gradients ofapproximately 8 T m−1 for femoral arteries and greater than100 T m−1 for cartoid arteries [39]. This suggests that targetingis likely to be most effective in regions of slower bloodflow, particularly if the target site is closer to the magnetsource. More recently, Richardson and others have developedmathematical models to determine particle trajectories fora variety of field/particle configurations in two dimensions,including consideration of their motion as they approach thevessel wall [40]. This is significant since near the walls theparticle motion is no longer governed by Stoke’s law for thedrag force, equation (8), and the hydrodynamic parametersare modified. Experimental work is underway to inform thedevelopment of such models [41].

4.2. Magnetic drug carriers

Since the first magnetic polymer carriers of the 1970s, avariety of magnetic nanoparticle and microparticle carriershave been developed to deliver drugs to specific target sitesin vivo. The optimization of these carriers continues today.Generally, the magnetic component of the particle is coatedby a biocompatible polymer such as PVA or dextran, althoughrecently inorganic coatings such as silica have been developed.The coating acts to shield the magnetic particle from thesurrounding environment and can also be functionalized byattaching carboxyl groups, biotin, avidin, carbodi-imide andother molecules [42–44]. As shown in figure 6, thesemolecules then act as attachment points for the coupling ofcytotoxic drugs or target antibodies to the carrier complex.

The carriers typically have one of two structuralconfigurations: (i) a magnetic particle core (usually magnetite,Fe3O4, or maghemite, γ -Fe2O3) coated with a biocompatible

Fe rr iteCor e

SiO2

F

F

F

F

F

F

F

FFerriteCore

SiO2

F

F

F

F

F

F

F

F

Figure 6. Schematic diagram of a functionalized magneticnanoparticle showing a core/shell structure with a shell of silica,SiO2, and functional groups attached to the shell.

polymer or (ii) a porous biocompatible polymer in whichmagnetic nanoparticles are precipitated inside the pores [45].Recent developmental work on carriers has largely focused onnew polymeric or inorganic coatings on magnetite/maghemitenanoparticles [10, 46–52], although noble metal coatings suchas gold are also being considered [53]. Research also continuesinto alternative magnetic particles, such as iron, cobalt or nickel[54–56] or yttrium aluminium iron garnet [57]. Cobalt/silicacarriers are currently being investigated for their use in eyesurgery to repair detached retinas [58, 59].

4.3. Targeting studies

Magnetic carriers were first used to target cytotoxic drugs(doxorubicin) to sarcoma tumours implanted in rat tails [60].The initial results were encouraging, showing a total remissionof the sarcomas compared to no remission in another groupof rats which were administered with ten times the dosebut without magnetic targeting. Since that study, successin cytotoxic drug delivery and tumour remission has beenreported by several groups using animals models includingswine [61, 62], rabbits [37] and rats [38, 63, 64]. Kuboet al recently offered a variation on these techniques. Theyimplanted permanent magnets at solid osteosarcoma sites inhamsters and delivered the cytotoxic compounds via magneticliposomes. They reported a four-fold increase in cytotoxicdrug delivery to the osteosarcoma sites when compared withnormal intravenous (non-magnetic) delivery [65], as well as asignificant increase in anti-tumour activity and the eliminationof weight-loss as a side effect [66].

This technique has also been employed to target cytotoxicdrugs to brain tumours. Brain tumours are particularlydifficult targets due to the fact that the drug must cross theblood–brain barrier. Pulfer and Gallo [63] demonstrated thatparticles as large as 1–2 µm could be concentrated at the siteof intracerebral rat glioma-2 (RG-2) tumours. Though theconcentration of the particles in the tumour was low, it wassignificantly higher than non-magnetic particles. In a later

R172

Topical Review

study the group demonstrated that 10–20 nm magnetic particleswere even more effective at targeting these tumours in rats [64].Electron microscopic analysis of brain tissue samples revealedthe presence of magnetic carriers in the interstitial space intumours but in normal brain tissue, they were only found inthe vasculature. Mykhaylyk and others [67] recently had lesssuccess using magnetite–dextran nanoparticles but were ableto target rat glial tumours by disrupting the blood–brain barrierimmediately prior to particle injection.

Studies of magnetic targeting in humans were rare up tonow. A Phase I clinical trial conducted by Lubbe and others[68–70] demonstrated that the infusion of ferrofluids was welltolerated in most of the 14 patients studied. In addition, theauthors reported that the ferrofluid was successfully directedto the advanced sarcomas without associated organ toxicity.More recently, FeRx Inc. was granted fast-track status toproceed with multi-centre Phases I and II clinical trials oftheir magnetic targeting system for hepatocellular carcinomas(a type of liver tumour). This appears to be the most promisingclinical application at present.

As promising as these results have been, there are severalproblems associated with magnetically targeted drug delivery[38, 71]. These limitations include (i) the possibility ofembolization of the blood vessels in the target region due toaccumulation of the magnetic carriers, (ii) difficulties in scalingup from animal models due to the larger distances between thetarget site and the magnet, (iii) once the drug is released, it is nolonger attracted to the magnetic field, and (iv) toxic responsesto the magnetic carriers. Recent pre-clinical and experimentalresults indicate, however, that it is still possible to overcomethese limitations and use magnetic targeting to improve drugretention and also address safety issues [68, 72].

4.4. Radionuclide and gene delivery

One way to overcome the limitation caused by the release of thedrug from the carrier is to use a system in which the therapeuticagent remains coupled to the magnetic carrier throughoutthe duration of the treatment. Based on this idea, thepossibility of targeting radionuclides via magnetic carriershas been investigated. The advantage these complexes haveover cytotoxic drug/magnetic carrier complexes is that theeffectiveness of the radionuclide does not require the tumourcells to actually take up the agent. If the radionuclide is targetedto a region near the tumour site and held there, the radiation willaffect the surrounding tumour tissue while it is still attached tothe magnetic carrier. This type of system was tested in 1995using in vitro and mouse models. In both cases, targeting ofa magnetic carrier coupled to a β-emitter (Y-90) was effectiveat concentrating radiation to the desired site. In the mousetumour model, there was a significant increase in radioactivityat the tumour site compared to using the same complex withouta magnetic field: 73 ± 32% vs 6 ± 4% [73]. Since this study,Hafeli and others [74–76] have demonstrated the effectivenessof this technique in both animal and cell culture studies usingboth yttrium-90 and rhenium-188.

Investigations also have begun with a view towards usingmagnetic carriers for gene therapy. In this case, a viral vectorcarrying the therapeutic gene is coated onto the magneticcarrier’s surface. By holding the carrier at the target site via

external magnetic fields, the virus is in contact with the tissuefor a longer period of time, increasing the efficiency of genetransfection and expression [77, 78]. New magnetic carriersare being developed specifically for these applications [79]and this is an area which shows great promise.

5. Hyperthermia

5.1. Catabolism of tumours by hyperthermia

The possibility of treating cancer by artificially inducedhyperthermia has led to the development of manydifferent devices designed to heat malignant cells whilesparing surrounding healthy tissue [80–82]. Experimentalinvestigations of the application of magnetic materials forhyperthermia date back to 1957 when Gilchrist et al [83]heated various tissue samples with 20–100 nm size particlesof γ -Fe2O3 exposed to a 1.2 MHz magnetic field. Sincethen there have been numerous publications describing avariety of schemes using different types of magnetic materials,different field strengths and frequencies and different methodsof encapsulation and delivery of the particles [84–102]. Inbroad terms, the procedure involves dispersing magneticparticles throughout the target tissue, and then applying an ACmagnetic field of sufficient strength and frequency to causethe particles to heat. This heat conducts into the immediatelysurrounding diseased tissue whereby, if the temperature canbe maintained above the therapeutic threshold of 42˚C for30 min or more, the cancer is destroyed. Whereas themajority of hyperthermia devices are restricted in their utilitybecause of unacceptable coincidental heating of healthy tissue,magnetic particle hyperthermia is appealing because it offersa way to ensure only the intended target tissue is heated (seefigure 7).

5.2. Operational constraints

A number of studies have demonstrated the therapeutic efficacyof this form of treatment in animal models (see, e.g. the review

30

32

5040

34

36

38

40

42

44

46

0 10 20 30 4

Time (minutes)

Tem

pera

ture

(˚ C

)

Figure 7. Animal trial data on hyperthermia treatments in rabbits,showing preferential heating of a tumour using intra-vascularlyinfused ferromagnetic microspheres; ( ) tumour edge, (�) tumourcentre, (�) normal liver 1–2 cm from tumour, (×) alternative lobe,and (♦) core body temperature.

R173

Topical Review

by Moroz et al [103]). To date, however, there have beenno reports of the successful application of this technology tothe treatment of a human patient. The challenge lies in beingable to deliver an adequate quantity of the magnetic particlesto generate enough heat in the target using AC magneticfield conditions that are clinically acceptable. Most of thelaboratory and animal model based studies reported so far arecharacterized by the use of magnetic field conditions that couldnot be safely used with a human patient. In most instances,reducing the field strength or frequency to safer levels wouldalmost certainly lead to such a reduction in the heat outputfrom the magnetic material as to render it useless in thisapplication.

It is important therefore to understand the underlyingphysical mechanisms by which heat is generated in smallmagnetic particles by alternating magnetic fields. Enoughheat must be generated by the particles to sustain tissuetemperatures of at least 42˚C for 30 min or so. Calculatingthe heat deposition rate required to achieve this is complicatedby the presence of blood flow and tissue perfusion, both ofwhich are dominant sources of tissue cooling and both ofwhich vary actively as tissue is heated. Several authors haveanalysed the heat transfer problem whereby a defined volumeof tissue is heated from within by evenly dispersed sources suchas microscopic magnetic particles [104–106]. The problemposed by the cooling from discrete blood vessels is generallyavoided because of the mathematical complexity and lack ofgenerality of the results. However, an often-used rule of thumbis that a heat deposition rate of 100 mW cm−3 of tissue willsuffice in most circumstances.

The frequency and strength of the externally appliedAC magnetic field used to generate the heating is limitedby deleterious physiological responses to high frequencymagnetic fields [107, 108]. These include stimulation ofperipheral and skeletal muscles, possible cardiac stimulationand arrhythmia, and non-specific inductive heating of tissue.Generally, the useable range of frequencies and amplitudes isconsidered to be f = 0.05–1.2 MHz and H = 0–15 kA m−1.Experimental data on exposure to much higher frequency fieldscomes from groups such as Oleson et al [109] who developed ahyperthermia system based on inductive heating of tissue, andAtkinson et al [110] who developed a treatment system basedon eddy current heating of implantable metal thermoseeds.Atkinson et al concluded that exposure to fields where theproduct H · f does not exceed 4.85 × 108 A m−1 s−1 is safeand tolerable.

The amount of magnetic material required to producethe required temperatures depends to a large extent on themethod of administration. For example, direct injectionallows for substantially greater quantities of material to belocalized in a tumour than do methods employing intravascularadministration or antibody targeting, although the latter twomay have other advantages. A reasonable assumption is thatca 5–10 mg of magnetic material concentrated in each cm3

of tumour tissue is appropriate for magnetic hyperthermia inhuman patients.

Regarding the choice of magnetic particle, the iron oxidesmagnetite (Fe3O4) and maghemite (γ -Fe2O3) are the moststudied to date because of their generally appropriate magneticproperties and biological compatibility, although many others

have been investigated. Particle sizes less than about 10 µm arenormally considered small enough to enable effective deliveryto the site of the cancer, either via encapsulation in a largermoiety or suspension in some sort of carrier fluid. Nanoscaleparticles can be coupled with antibodies to facilitate targetingon an individual cell basis. Candidate materials are divided intotwo main classes; ferromagnetic or ferrimagnetic (FM) singledomain or multi-domain particles, or SPM particles. The heatgenerating mechanisms associated with each class are quitedifferent, each offering unique advantages and disadvantages,as discussed later.

5.3. Heating mechanisms

FM particles possess hysteretic properties when exposed to atime varying magnetic field, which gives rise to magneticallyinduced heating. The amount of heat generated per unitvolume is given by the frequency multiplied by the area ofthe hysteresis loop:

PFM = µ0f

∮H dM. (10)

This formula ignores other possible mechanisms formagnetically induced heating such as eddy current heatingand ferromagnetic resonance, but these are generally irrelevantin the present context. The particles used for magnetichyperthermia are much too small and the AC field frequenciesmuch too low for the generation of any substantial eddycurrents. Ferromagnetic resonance effects may becomerelevant but only at frequencies far in excess of those generallyconsidered appropriate for this type of hyperthermia.

For FM particles well above the SPM size limit thereis no implicit frequency dependence in the integral ofequation (10), so PFM can be readily determined fromquasi-static measurements of the hysteresis loop using, forexample, a VSM or SQUID magnetometer. As discussedin section 2, the re-orientation and growth of spontaneouslymagnetized domains within a given FM particle depends onboth microstructural features such as vacancies, impuritiesor grain boundaries, and intrinsic features such as themagnetocrystalline anisotropy as well as the shape and size ofthe particle. In most cases it is not possible to predict a prioriwhat the hysteresis loop will look like.

In principle, substantial hysteresis heating of the FMparticles could be obtained using strongly anisotropic magnetssuch as Nd-Fe-B or Sm-Co; however, the constraints on theamplitude of H that can be used mean that fully saturatedloops cannot be used. Minor (unsaturated) loops could beused, and would give rise to heating, but only at much reducedlevels. In fact, as is evident in equation (10), the maximumrealizable PFM should involve a rectangular hysteresis loop.However, this could only be achieved with an ensemble ofuniaxial particles perfectly aligned with H , a configurationthat would be difficult if not impossible to achieve in vivo. Fora more realistic ensemble of randomly aligned FM particlesthe most that can be hoped for is around 25% of this idealmaximum.

However, over the last decade the field of magnetic particlehyperthermia has been revitalized by the advent of ‘magneticfluid hyperthermia’, where the magnetic entities are SPM

R174

Topical Review

nanoparticles suspended in water or a hydrocarbon fluid tomake a ‘magnetic fluid’ or ‘ferrofluid’ [98, 111, 112]. When aferrofluid is removed from a magnetic field its magnetizationrelaxes back to zero due to the ambient thermal energy ofits environment. This relaxation can correspond either tothe physical rotation of the particles themselves within thefluid, or rotation of the atomic magnetic moments within eachparticle. Rotation of the particles is referred to as ‘Brownianrotation’ while rotation of the moment within each particleis known as ‘Neel relaxation’. Each of these processesis characterized by a relaxation time: τB for the Brownianprocess depends on the hydrodynamic properties of the fluid;while τN for the Neel process is determined by the magneticanisotropy energy of the SPM particles relative to the thermalenergy. Both Brownian and Neel processes may be present ina ferrofluid, whereas only τN is relevant in fixed SPM particleswhere no physical rotation of the particle is possible. Therelaxation times τB and τN depend differently on particle size;losses due to Brownian rotation are generally maximized ata lower frequency than are those due to Neel relaxation for agiven size.

The physical basis of the heating of SPM particles by ACmagnetic fields has been reviewed by Rosensweig [113]. It isbased on the Debye model, which was originally developed todescribe the dielectric dispersion in polar fluids [114], and therecognition that the finite rate of change of M in a ferrofluidmeans that it will lag behind H . For small field amplitudes, andassuming minimal interactions between the constituent SPMparticles, the response of the magnetization of a ferrofluid to anAC field can be described in terms of its complex susceptibilityχ = χ ′ + iχ ′′, where both χ ′ and χ ′′ are frequency dependent.The out-of-phase χ ′′ component results in heat generationgiven by [113]:

PSPM = µ0πf χ ′′H 2, (11)

which can be interpreted physically as meaning that if M

lags H there is a positive conversion of magnetic energyinto internal energy. This simple theory compares favourablywith experimental results, for example, in predicting a squaredependence of PSPM on H [91], and the dependence of χ ′′ onthe driving frequency [115–117].

Measurements of the heat generation from magneticparticles are usually quoted in terms of the specific absorptionrate (SAR) in units of W g−1. Multiplying the SAR by thedensity of the particle yields PFM and PSPM, so the parameterallows comparison of the efficacies of magnetic particlescovering all the size ranges [88, 111, 118–121]. It is clearfrom such comparisons that most real FM materials requireapplied field strengths of ca 100 kA m−1 or more before theyapproach a fully saturated loop, and therefore only minorhysteresis loops can be utilized given the operational constraintof ca 15 kA m−1, giving rise to low SARs. In contrast, SPMmaterials are capable of generating impressive levels of heatingat lower fields. For example, the best of the ferrofluids reportedby Hergt et al [121] has a SAR of 45 W g−1 at 6.5 kA m−1

and 300 kHz which extrapolates to 209 W g−1 for 14 kA m−1,compared to 75 W g−1 at 14 kA m−1 for the best FM magnetitesample. While all of these samples would be adequate formagnetic particle hyperthermia, importantly, it seems clear thatferrofluids and SPM particles are more likely to offer usefulheating using lower magnetic field strengths.

6. MRI contrast enhancement

6.1. Physical principles

MRI relies on the counterbalance between the exceedinglysmall magnetic moment on a proton, and the exceedinglylarge number of protons present in biological tissue, whichleads to a measurable effect in the presence of large magneticfields [122, 123]. Thus, even though the effect of a steadystate field of B0 = 1 T on a collection of protons, such asthe hydrogen nuclei in a water molecule, is so small that itis equivalent to only three of every million proton momentsm being aligned parallel to B0, there are so many protonsavailable—6.6×1019 in every mm3 of water—that the effectivesignal, 2 × 1014 proton moments per mm3, is observable. Asillustrated in figure 8, this signal can be captured by making useof resonant absorption: applying a time-varying magnetic fieldin a plane perpendicular to B0, tuned to the Larmor precessionfrequency ω0 = γB0 of the protons. For 1H protons thegyromagnetic ratio γ = 2.67 × 108 rad s−1 T−1, so that in afield of B0 = 1 T the Larmor precession frequency correspondsto a radio frequency field with ω0/2π = 42.57 MHz. Inpractice the radio frequency transverse field is applied in apulsed sequence, of duration sufficient to derive a coherentresponse from the net magnetic moment of the protons in theMRI scanner. From the instant that the radio frequency pulse isturned off the relaxation of the coherent response is measuredvia induced currents in pick-up coils in the scanner. Theseresonantly tuned detection coils enhance the signal by a qualityfactor of ca 50–100. As shown in figure 8, for B0 parallel to

B B

m

m

timsig

nal

am

plit

ude

mxy

B0 B0

m

m

timesig

nal

am

plit

ude

mxy

time

sig

nal

am

plit

ud

e

mz

time

sig

nal

am

plit

ud

e

mz

(a) (b)

(c)

(d)

Figure 8. Illustration of magnetic resonance for a large ensemble ofprotons with net magnetic moment m in the presence of a externalmagnetic field B0. In (a) the net moment precesses around B0 at thecharacteristic Larmor frequency, ω0. In (b) a second external field isapplied, perpendicular to B0, oscillating at ω0. Despite being muchweaker than B0, this has the effect of resonantly exciting themoment precession into the plane perpendicular to B0. In (c) and (d)the oscillating field is removed at time zero, and the in-plane (c) andlongitudinal (d) moment amplitudes relax back to their initial values.

R175

Topical Review

the z-axis these relaxation signals are of the form:

mz = m(1 − e−t/T1) (12)

andmx,y = m sin(ω0t + φ)e−t/T2 , (13)

where T1 and T2 are the longitudinal (or spin–lattice) andtransverse (or spin–spin) relaxation times, respectively, andφ is a phase constant. The longitudinal relaxation reflects aloss of energy, as heat, from the system to its surrounding‘lattice’, and is primarily a measure of the dipolar couplingof the proton moments to their surroundings. The relaxationin the xy-plane is relatively rapid, and is driven by theloss of phase coherence in the precessing protons due totheir magnetic interactions with each other and with otherfluctuating moments in the tissue. Dephasing can also beaffected by local inhomogeneities in the applied longitudinalfield, leading to the replacement of T2 in equation (13) by theshorter relaxation time, T ∗

2 :

1

T ∗2

= 1

T2+ γ

�B0

2, (14)

where �B0 is the variation in the field brought about eitherthrough distortions in the homogeneity of the applied fielditself, or by local variations in the magnetic susceptibility ofthe system [124, 125].

6.2. MRI contrast enhancement studies

Both T1 and T ∗2 can be shortened by the use of a magnetic

contrast agent. The most common contrast agents currentlyused are PM gadolinium ion complexes, although agents basedon SPM nanoparticles are commercially available, such as‘Feridex I. V.’, an iron oxide contrast agent marketed byAdvanced Magnetics Inc. for the organ-specific targetingof liver lesions. The SPM particles used are magneticallysaturated in the normal range of magnetic field strengths usedin MRI scanners, thereby establishing a substantial locallyperturbing dipolar field which leads, via equation (14), to amarked shortening of T ∗

2 (see figure 9) along with a less markedreduction of T1 [123].

(a) (b)

time

Figure 9. Effect of magnetic particle internalization in cells on T ∗2

relaxation times: (a) the protons in cells tagged by magnetic particleshave a shorter T ∗

2 relaxation time than those in (b) untagged cells.

Iron oxide nanoparticles are the most commonly usedSPM contrast agents. Dextran coated iron oxides arebiocompatible and are excreted via the liver after the treatment.They are selectively taken up by the reticuloendothelial system,a network of cells lining blood vessels whose function is toremove foreign substances from the bloodstream; MRI contrastrelies on the differential uptake of different tissues [126]. Thereis also a size effect: nanoparticles with diameters of ca 30 nmor more are rapidly collected by the liver and spleen, whileparticles with sizes of ca 10 nm or less are not so easilyrecognized. The smaller particles therefore have a longerhalf-life in the blood stream and are collected by reticulo-endothelial cells throughout the body, including those in thelymph nodes and bone marrow [127, 128]. Similarly, suchagents have been used to visualize the vascular system [129],and to image the central nervous system [130]. It is also notablethat tumour cells do not have the effective reticuloendothelialsystem of healthy cells, so that their relaxation times are notaltered by the contrast agents. This has been used, for example,to assist the identification of malignant lymph nodes [131],liver tumours [132] and brain tumours [133].

Iron oxide nanoparticles also lend themselves toencapsulation into target-specific agents, such as a liposomethat is known to localize in the bone marrow [134]. Dendrimercoatings comprising a highly branched polymer structure thathas a high rate of non-cell-specific binding and intracellularuptake have also been used to good effect [135], as has the useof lipofection agents, normally used to carry DNA into the cellnucleus, to enable intracellular incorporation of the magneticnanoparticles into stem cells [136]. Magnetic nanoparticleshave also been utilized for the in vivo monitoring of geneexpression, a process in which cells are engineered to over-express a given gene. This process leads to the production ofincreased numbers of certain cell wall receptors, which in turncan be targeted using specially coated nanoparticles, therebyallowing a differentiation between the expressing cells andtheir surroundings [137]. Another aspect of the targeting ofcell receptors has been to selectively study cells that are in theprocess of cell death [138].

7. Discussion and future prospects

In this paper we have reviewed some basic concepts regardingthe interactions between magnetic nanoparticles and a staticor time-varying external magnetic field, and shown howthese pertain to current biomedical applications of magneticnanoparticles. We have focused in particular on magneticseparation, drug delivery, hyperthermia and MRI contrastenhancement, although these are only four of the manybiomedical applications of magnetic nanoparticles that arecurrently being explored. For example, research is beingconducted into magnetic twisting cytometry, a process in whichferromagnetic microspheres are bound to specific receptors ona cell wall. Changing the direction of an applied magnetic fieldtwists the microsphere by a measurable amount, which can thenbe related to the mechanical properties of the cell membraneand cytoskeleton [139–143]. Magnetic nanoparticles arealso being tested for tissue engineering applications, forexample, in the mechanical conditioning of cells growing inculture [144–146]. In such systems magnetic particles are

R176

Topical Review

attached to either the cell membrane, or to mechanosensitiveion channels in the membrane, and a magnetic force isapplied which activates the channels and initiates biochemicalreactions within the cell, thereby promoting the growth ofe.g. functional bone and cartilage. A third example ismagnetic biosensing: using magnetic nanoparticles coupled toanalyte-specific molecules to detect target molecules via thecrosslinking between coatings and the resultant aggregationof the magnetic particles, monitored by changes in protonrelaxation times on a bench-top nuclear magnetic resonancesystem [147, 148].

One thing that almost all of these applications have incommon, however, is that they are not yet market-basedtechnologies. The exceptions are magnetic separation viacell and protein labelling, which is found in most biomedicaland biochemistry laboratories today, and MRI contrastenhancement using encapsulated SPM nanoparticles, which isavailable (albeit not routinely used) in most hospital scanningfacilities. Drug delivery via coating of nanoparticles iscurrently undergoing preliminary human trials, after successfultests in animals, with promising results, but it will be sometime before it will be clinically available; and hyperthermiatreatment of tumours is not yet accessible in humans, despitehaving been proven to be effective in animals. This highlightsthe fact that there is a significant step between a provenhypothetical process that has been tested under controlledlaboratory conditions, and a close-to-market technology whichcan cope with the added complexity of in-service use. This isespecially so if the goal is to transfer a procedure that haslargely been the subject of in vitro or animal in vivo testinginto a human in vivo therapy.

In short, one of the biggest challenges in biomedicalapplications of magnetic nanoparticles lies in dealing withthe issue of technology transfer. There are opportunitiesin this respect for more interdisciplinary approaches, forexample, to ensure that the laboratory based experiments canmore explicitly emulate the expected conditions that wouldbe encountered in vivo. There is also scope for significantcontributions via the mathematical modelling of complexsystems, with the objective of understanding more specificallythe full gamut of physical phenomena and effects that togetherdetermine whether, in the final analysis, a given applicationwill be successful.

References

[1] Jiles D 1991 Introduction to Magnetism and MagneticMaterials (London: Chapman and Hall)

[2] Morrish A H 2001 The Physical Principles of Magnetism(New York: IEEE Press)

[3] Neel L 1949 Ann. Geophys. 5 99–136[4] Brown W F Jr 1963 Thermal fluctuations of a single-domain

particle Phys. Rev. 130 1677–86[5] Gerber R 1994 Magnetic separation Applied Magnetism

ed G Asti (Dordrecht: Kluwer) pp 165–220[6] Zborowski M 1997 Physics of magnetic cell sorting Scientific

and Clinical Applications of Magnetic Carriersed M Zborowski (New York: Plenum) pp 205–31

[7] Hatch G P and Stelter R E 2001 Magnetic designconsiderations for devices and particles used for biologicalhigh-gradient magnetic separation (HGMS) systemsJ. Magn. Magn. Mater. 225 262–76

[8] Molday R S and MacKenzie D 1982 Immunospecificferromagnetic iron–dextran reagents for the labeling andmagnetic separation of cells J. Immunol. Methods 52353–67

[9] Sangregorio C, Wiemann J K, O’Connor C J andRosenzweig Z 1999 A new method for the synthesis ofmagnetoliposomes J. Appl. Phys. 85 5699–701

[10] Pardoe H, Chua-anusorn W, St Pierre T G and Dobson J 2001Structural and magnetic properties of nanoscale iron oxideparticles synthesized in the presence of dextran orpolyvinyl alcohol J. Magn. Magn. Mater. 225 41–6

[11] Hogemann D, Josephson L, Weissleder R and Basilion J P2000 Improvement of MRI probes to allow efficientdetection of gene expression Bioconjugate Chem. 11941–6

[12] Levy L, Sahoo Y, Kim K S, Bergey E J and Prasad P N 2002Nanochemistry: synthesis and characterization ofmultifunctional nanoclinics for biological applicationsChem. Mater. 14 3715–21

[13] Zhang Y, Kohler N and Zhang M 2002 Surface modificationof superparamagnetic magnetite nanoparticles and theirintracellular uptake Biomaterials 23 1553–61

[14] Tibbe A, de Grooth B, Greve J, Liberti P, Dolan G andTerstappen L 1999 Optical tracking and detection ofimmunomagnetically selected and aligned cells NatureBiotechnol. 17 1210–13

[15] Kularatne B Y, Lorigan P, Browne S, Suvarna S K,Smith M O and Lawry J 2002 Monitoring tumour cells inthe peripheral blood of small cell lung cancer patientsCytometry 50 160–7

[16] Morisada S, Miyata N and Iwahori K 2002 Immunomagneticseparation of scum-forming bacteria using polyclonalantibody that recognizes mycolic acids J. Microbiol.Methods 51 141–8

[17] Zigeuner R E, Riesenberg R, Pohla H, Hofstetter A andOberneder R 2003 Isolation of circulating cancer cellsfrom whole blood by immunomagnetic cell enrichmentand unenriched immunocytochemistry in vitro J. Urol. 169701–5

[18] Mura C V, Becker M I, Orellana A and Wolff D 2002Immunopurification of Golgi vesicles by magnetic sortingJ. Immunol. Methods 260 263–71

[19] Ugelstad J, Prestvik W S, Stenstad P, Kilaas L andKvalheim G 1998 Selective cell separation withmonosized magnetizable polymer beads Magnetism inMedicine ed H Nowak (Berlin: Wiley-VCH) pp 471–88

[20] Hancock J P and Kemshead J T 1993 A rapid and highlyselective approach to cell separations using animmunomagnetic colloid J. Immunol. Methods 164 51–60

[21] Owen C S 1983 Magnetic cell sorting Cell Separation:Methods and Selected Applications (New York: Academic)

[22] Rheinlander T, Kotitz R, Weitschies W and Semmler W 2000Magnetic fractionation of magnetic fluids J. Magn. Magn.Mater. 219 219–28

[23] Moore L, Rodeiguez A, Williams P, McCloskey B,Nakamura M, Chalmers J and Zborowski M 2001Progenitor cell isolation with a high-capacity quadrapolemagnetic flow sorter J. Magn. Magn. Mater. 225 277–84

[24] Rheinlander T, Roessner D, Weitschies W and Semmler W1999 Comparison of size–selective techniques for thefractionation of magnetic nanospheres Presented atFrontiers in Magnetism (Stockholm, Sweden)

[25] Todd P, Cooper R, Doyle J, Dunn S, Vellinger J andDeuser M 2001 Multistage magnetic particle separatorJ. Magn. Magn. Mater. 225 294–300

[26] Liberti P A, Rao C G and Terstappen L W M M 2001Optimization of ferrofluids and protocols for theenrichment of breast tumor cells in blood J. Magn. Magn.Mater. 225 301–7

[27] Paul F, Melville D, Roath S and Warhurst D 1981 A benchtop magnetic separator for malarial parasite concentrationIEEE Trans. Magn. MAG-17 2822–4

R177

Topical Review

[28] Seesod N, Nopparat P, Hedrum A, Holder A, Thaithong S,Uhlen M and Lundeberg J 1997 An integrated systemusing immunomagnetic separation, polymerase chainreaction, and colorimetric detection for diagnosis ofPlasmodium Falciparum Am. J. Tropical Med. Hygiene 56322–8

[29] Hofmann W-K, de Vos S, Komor M, Hoelzer D,Wachsman W and Koeffler H P 2002 Characterization ofgene expression of CD34+ cells from normal andmyelodysplastic bone marrow Blood 100 3553–60

[30] Delgratta C, Dellapenna S, Battista P, Didonato L, Vitullo P,Romani G and Diluzio S 1995 Detection and counting ofspecific cell populations by means of magnetic markerslinked to monoclonal antibodies Phys. Med. Biol. 40671–81

[31] Edelstein R L, Tamanaha C R, Sheehan P E, Miller M M,Baselt D R, Whitman L J and Colton R J 2000 The BARCbiosensor applied to the detection of biological warfareagents Biosensors Bioelectron. 14 805–13

[32] Kala M, Bajaj K and Sinha S 1997 Magnetic beadenzyme-linked immunosorbent assay (ELISA) detectsantigen-specific binding by phage-displayed scFvantibodies that are not detected with conventional ELISAAnal. Biochem. 254 263–6

[33] Yazdankhah S P, Hellemann A-L, Ronningen K and Olsen E1998 Rapid and sensitive detection of Staphylococcusspecies in milk by ELISA based on monodispersemagnetic particles Veterinary Microbiol. 62 17–26

[34] Widder K J, Senyei A E and Scarpelli D G 1978 Magneticmicrospheres: a model system for site specific drugdelivery in vivo Proc. Soc. Exp. Biol. Med. 58 141–6

[35] Senyei A, Widder K and Czerlinski C 1978 Magneticguidance of drug carrying microspheres J. Appl. Phys. 493578–83

[36] Mosbach K and Schroder U 1979 Preparation and applicationof magnetic polymers for targeting of drugs FEBS Lett.102 112–6

[37] Alexiou C, Arnold W, Klein R J, Parak F G, Hulin P,Bergemann C, Erhardt W, Wagenpfeil S and Lubbe A S2000 Locoregional cancer treatment with magnetic drugtargeting Cancer Res. 60 6641–8

[38] Lubbe A S, Bergemann C, Brock J and McClure D G 1999Physiological aspects in magnetic drug-targeting J. Magn.Magn. Mater. 194 149–55

[39] Voltairas P A, Fotiadis D I and Michalis L K 2002Hydrodynamics of magnetic drug targeting J. Biomech. 35813–21

[40] Cummings L J, Richardson G and Hazelwood L 2000 Drugdelivery by magnetic microspheres Proc. Mathematics inMedicine Study Group (Nottingham, UK: University ofNotingham) pp 69–83

[41] Dobson J, Gross B, Exley C, Mickhailova A, Batich C andPardoe H 2001 Biomedical applications of biogenic andbiocompatible magnetic nanoparticles Modern Problemsof Cellular and Molecular Biophysics ed S Ayrapetyanand A North (Yerevan: Noyan Tapan/UNESCO)pp 121–30

[42] Mehta R V, Upadhyay R V, Charles S W and Ramchand C N1997 Direct binding of protein to magnetic particlesBiotechnol. Techn. 11 493–6

[43] Koneracka M, Kopcansky P, Antalk M, Timko M,Ramchand C N, Lobo D, Mehta R V and Upadhyay R V1999 Immobilization of proteins and enzymes to finemagnetic particles J. Magn. Magn. Mater. 201427–30

[44] Koneracka M, Kopcansky P, Timko M, Ramchand C N,de Sequeira A and Trevan M 2002 Direct bindingprocedure of proteins and enzymes to fine magneticparticles J. Mol. Catalysis B Enzymatic 18 13–8

[45] Hans M L and Lowman A M 2002 Biodegradablenanoparticles for drug delivery and targeting Curr. Opin.Solid State Mater. Sci. 6 319–27

[46] Gruttner C and Teller J 1999 New types of silica-fortifiedmagnetic nanoparticles as tools for molecular biologyapplications J. Magn. Magn. Mater. 194 8–15

[47] Rudge S R, Kurtz T L, Vessely C R, Catterall L G andWilliamson D L 2000 Preparation, characterization, andperformance of magnetic iron–carbon compositemicroparticles for chemotherapy Biomaterials 21 1411–20

[48] Arias J L, Gallardo V, Gomez-Lopera S A, Plaza R C andDelgado A V 2001 Synthesis and characterization ofpoly(ethyl-2-cyanoacrylate) nanoparticles with a magneticcore J. Control. Release 77 309–21

[49] Gomez-Lopera S A, Plaza R C and Delgado A V 2001Synthesis and characterization of sphericalmagnetite/biodegradable polymer composite particlesJ. Colloid Interface Sci. 240 40–7

[50] Mornet S, Grasset F, Portier J and D E 2002Maghemite/silica nanoparticles for biological applicationsEur. Cells Mater. 3 110–3

[51] Deng Y, Wang L, Yang W, Fu S and Elaissari A 2003Preparation of magnetic polymeric particles via inversemicroemulsion polymerization process J. Magn. Magn.Mater. 257 69–78

[52] Santra S, Tapec R, Theodoropoulou N, Dobson J, Hebard Aand Tan W 2001 Synthesis and characterization ofsilica-coated iron oxide nanoparticles in microemulsion:the effect of nonionic surfactants Langmuir 17 2900–6

[53] Carpenter E E 2001 Iron nanoparticles as potential magneticcarriers J. Magn. Magn. Mater. 225 17–20

[54] Stevenson J P, Rutnakornpituk M, Vadala M, Esker A R,Charles S W, Wells S, Dailey J P and Riffle J S 2001Magnetic cobalt dispersions in poly(dimethylsiloxane)fluids J. Magn. Magn. Mater. 225 47–58

[55] Connolly J, St Pierre T G, Rutnakornpituk M and Riffle J S2002 Silica coating of cobal nanoparticles increases theirmagnetic and chemical stability for biomedicalapplications Eur. Cells Mater. 3 106–9

[56] Sun X, Gutierrez A, Yacaman M J, Dong X and Jin S 2000Investigations on magnetic properties and structure forcarbon encapsulated nanoparticles of Fe, Co, Ni Mater.Sci. Eng. 286 157–60

[57] Grasset F, Mornet S, Demourgues A, Portier J, Bonnet J,Vekris A and Duguet E 2001 Synthesis, magneticproperties, surface modification and cytotoxicityevaluation of Y3Fe5−xAlxO12 garnet submicron particlesfor biomedical applications J. Magn. Magn. Mater. 234409–18

[58] Dailey J P, Phillips J P, Li C and Riffle J S 1999 Synthesis ofsilicone magnetic fluid for use in eye surgery J. Magn.Magn. Mater. 194 140–8

[59] Rutnakornpituk M, Baranauskas V, Riffle J S, Connolly J,St Pierre T G and Dailey J P 2002 Polysiloxane fluiddispersions of cobalt nanoparticles in silica spheres for usein ophthalmic applications Eur. Cells Mater. 3 102–5

[60] Widder K J, Morris R M, Poore G A, Howard D P andSenyei A E 1983 Selective targeting of magnetic albuminmicrospheres containing low-dose doxorubicin—totalremission in Yoshida sarcoma-bearing rats Eur. J. CancerClin. Oncol. 19 135–9

[61] Goodwin S, Peterson C, Hob C and Bittner C 1999 Targetingand retention of magnetic targeted carriers (MTCs)enhancing intra-arterial chemotherapy J. Magn. Magn.Mater. 194 132–9

[62] Goodwin S C, Bittner C A, Peterson C L and Wong G 2001Single-dose toxicity study of hepatic intra-arterial infusionof doxorubicin coupled to a novel magnetically targeteddrug carrier Toxicol. Sci. 60 177–83

[63] Pulfer S K and Gallo J M 1999 Enhanced brain tumorselectivity of cationic magnetic polysaccharidemicrospheres J. Drug Targeting 6 215–28

[64] Pulfer S K, Ciccotto S L and Gallo J M 1999 Distribution ofsmall magnetic particles in brain tumor-bearing ratsJ. Neuro-Oncol. 41 99–105

R178

Topical Review

[65] Kubo T, Sugita T, Shimose S, Nitta Y, Ikuta Y andMurakami T 2000 Targeted delivery of anticancer drugswith intravenously administered magnetic liposomes inosteosarcoma-bearing hamsters Int. J. Oncol. 17 309–16

[66] Kubo T, Sugita T, Shimose S, Nitta Y, Ikuta Y andMurakami T 2001 Targeted systemic chemotherapy usingmagnetic liposomes with incorporated adriamycin forosteosarcoma in hamsters Int. J. Oncol. 18 121–6

[67] Mykhaylyk O, Cherchenko A, Ilkin A, Dudchenko N,Ruditsa V, Novoseletz M and Zozulya Y 2001 Glial braintumor targeting of magnetite nanoparticles in rats J. Magn.Magn. Mater. 225 241–7

[68] Lubbe A S, Bergemann C, Huhnt W, Fricke T, Riess H,Brock J W and Huhn D 1996 Preclinical experiences withmagnetic drug targeting: tolerance and efficacy CancerRes. 56 4694–701

[69] Lubbe A S, Bergemann C, Riess H, Schriever F, Reichardt P,Possinger K, Matthias M, Doerken B, Herrmann F andGuertler R 1996 Clinical experiences with magnetic drugtargeting: a phase I study with 4′-epidoxorubicin in14 patients with advanced solid tumors Cancer Res. 564686–93

[70] Lubbe A S and Bergemann C 1996 Selected preclinical andfirst clinical experiences with magnetically targeted4′-epidoxorubicin in patients with advanced solid tumorsPresented at International Conf. on Scientific and ClinicalApplications of Magnetic Carriers (Rostock, Germany)

[71] Hafeli U O and Pauer G J 1999 In vitro and in vivo toxicity ofmagnetic microspheres J. Magn. Magn. Mater. 194 76–82

[72] Gallo J M and Hafeli U 1997 Preclinical experiences withmagnetic drug targeting: tolerance and efficacy andclinical experiences with magnetic drug targeting:a phase I study with 4′-epidoxorubicin in 14 patients withadvanced solid tumors Cancer Res. 57 3063–4

[73] Hafeli U O, Sweeney S M, Beresford B A and Humm J L1995 Effective targeting of magnetic radioactive90Y-microspheres to tumor cells by an externally appliedmagnetic field. Preliminary in vitro and in vivo resultsNucl. Med. Biol. 22 147

[74] Hafeli U O, Pauer G J, Roberts W K and Humm J L 1996Magnetically targeted microspheres for intracavitary andintraspinal Y-90 radiotherapy Presented at InternationalConf. on Scientific and Clinical Applications of MagneticCarriers (Rostock, Germany)

[75] Hafeli U O, Casillas S, Dietz D W, Pauer G J, Rybicki L A,Conzone S D and Day D E 1999 Hepatic tumorradioembolization in a rat model using radioactiverhenium (186Re/188Re) glass microspheres Int. J. Radiat.Oncol. Biol. Phys. 44 189–200

[76] Hafeli U, Pauer G, Failing S and Tapolsky G 2001Radiolabeling of magnetic particles with rhenium-188 forcancer therapy J. Magn. Magn. Mater. 225 73–8

[77] Mah C, Zolotukhin I, Fraites T J, Dobson J, Batich C andByrne B J 2000 Microsphere-mediated delivery ofrecombinant AAV vectors in vitro and in vivo Mol.Therapy 1 S239

[78] Mah C, Fraites T J, Zolotukhin I, Song S, Flotte T R,Dobson J, Batich C and Byrne B J 2002 Improved methodof recombinant AAV2 delivery for systemic targeted genetherapy Mol. Therapy 6 106–12

[79] Hughes C, Galea-Lauri J, Farzaneh F and Darling D 2001Streptavidin paramagnetic particles provide a choice ofthree affinity-based capture and magnetic concentrationstrategies for retroviral vectors Mol. Therapy 3 623–30

[80] Van der Zee J 2002 Heating the patient: a promisingapproach? Ann. Oncol. 13 1173–84

[81] Wust P, Hildebrandt B, Sreenivasa G, Rau B, Gellermann J,Riess H, Felix R and Schlag P M 2002 Hyperthermia incombined treatment of cancer Lancet Oncol. 3 487–97

[82] Moroz P, Jones S K and Gray B N 2001 Status ofhyperthermia in the treatment of advanced liver cancerJ. Surg. Oncol. 77 259–69

[83] Gilchrist R K, Medal R, Shorey W D, Hanselman R C,Parrott J C and Taylor C B 1957 Selective inductiveheating of lymph nodes Ann. Surg. 146 596–606

[84] Mosso J A and Rand R W 1972 Ferromagnetic siliconevascular occlusion Ann. Surg. 663–8

[85] Rand R W, Snyder M, Elliott D G and Snow H D 1976Selective radiofrequency heating of ferrosilicone occludedtissue: a preliminary report Bull. Los Angeles Neurol. Soc.41 154–9

[86] Gordon R T, Hines J R and Gordon D 1979 Intracellularhyperthermia: a biophysical approach to cancer treatmentvia intracellular temperature and biophysical alterationsMed. Hypotheses 5 83–102

[87] Rand R W, Snow H D, Elliott D G and Snyder M 1981Thermomagnetic surgery for cancer Appl. Biochem.Biotechnol. 6 265–72

[88] Borrelli N F, Luderer A A and Panzarino J N 1984 Hysteresisheating for the treatment of tumours Phys. Med. Biol. 29487–94

[89] Hase M, Sako M and Hirota S 1990 Experimental study offerromagnetic induction heating combined with hepaticarterial embolization of liver tumoursNippon-Igaku-Hoshasen-Gakkai-Zasshi 50 1402–14

[90] Suzuki S, Arai K, Koike T and Oguchi K 1990 Studies onliposomal ferromagnetic particles and a technique of highfrequency inductive heating—in vivo studies of rabbitsJ. Japan. Soc. Cancer Therapy 25 2649–58

[91] Chan D C F, Kirpotin D B and Bunn P A Jr 1993 Synthesisand evaluation of colloidal magnetic iron oxides for thesite-specific radiofrequency-induced hyperthermia ofcancer J. Magn. Magn. Mater. 122 374–8

[92] Matsuki H, Yanada T, Sato T, Murakami K and Minakawa S1994 Temperature sensitive amorphous magnetic flakes forintratissue hyperthermia Mater. Sci. Eng. A181/A1821366–8

[93] Mitsumori M et al 1994 Development of intra-arterialhyperthermia using a dextran–magnetite complex Int. J.Hyperthermia 10 785–93

[94] Suzuki M, Shinkai M, Kamihira M and Kobayashi T 1995Preparation and characteristics of magnetite-labelledantibody with the use of poly(ethylene glycol) derivativesBiotechnol. Appl. Biochem. 21 335–45

[95] Mitsumori M, Hiraoka M, Shibata T, Okuno Y, Nagata Y,Nishimura Y, Abe M, Hasegawa M, Nagae H andEbisawa Y 1996 Targeted hyperthermia using dextranmagnetite complex: a new treatment modality for livertumours Hepato-Gastroenterology 43 1431–7

[96] Jordan A, Scholz R, Wust P, Fahling H, Krause J,Wlodarczyk W, Sander B, Vogl T and Felix R 1997 Effectsof magnetic fluid hyperthermia (MFH) on C3H mammarycarcinoma in vivo Int. J. Hyperthermia 13 587–605

[97] Shinkai M, Yanase M, Suzuki M, Honda H, Wakabayashi T,Yoshida J and Kobayashi T 1999 Intracellularhyperthermia for cancer using magnetite cationicliposomes J. Magn. Magn. Mater. 194 176–84

[98] Jordan A, Scholz R, Wust P, Fahling H and Felix R 1999Magnetic fluid hyperthermia (MFH): cancer treatmentwith AC magnetic field induced excitation ofbiocompatible superparamagnetic nanoparticles J. Magn.Magn. Mater. 201 413–19

[99] Minamimura T, Sato H, Kasaoka S, Saito T, Ishizawa S,Takemori S, Tazawa K and Tsukada K 2000 Tumourregression by inductive hyperthermia combined withhepatic embolization using dextran magnetite incorporatedmicrospheres in rats Int. J. Oncol. 16 1153–8

[100] Moroz P, Jones S K, Winter J and Gray B N 2001 Targetingliver tumors with hyperthermia: ferromagneticembolization in a rabbit liver tumor model J. Surg. Oncol.78 22–9

[101] Jones S K, Winter J W and Gray B N 2002 Treatment ofexperimental rabbit liver tumours by selectively targetedhyperthermia Int. J. Hyperthermia 18 117–28

R179

Topical Review

[102] Hilger I, Fruhauf K, Andra W, Hiergeist R, Hergt R andKaiser W A 2002 Heating potential of iron oxides fortherapeutic purposes in interventional radiology Acad.Radiol. 9 198–202

[103] Moroz P, Jones S K and Gray B N 2002 Magneticallymediated hyperthermia: current status and futuredirections Int. J. Hyperthermia 18 267–84

[104] Rabin Y 2001 Is intracellular hyperthermia superior toextracellular hyperthermia in the thermal sense? Int. J.Hyperthermia

[105] Granov A M, Muratov O V and Frolov V F 2002 Problemsin the local hyperthermia of inductively heatedembolized tissues Theor. Foundations Chem. Eng.36 63–6

[106] Craciun V, Calugaru G and Badescu V 2002 Acceleratedsimulation of heat transfer in magnetic fluid hyperthermiaCzechoslovak J. Phys. 52 725–8

[107] Oleson J R, Cetas T C and Corry P M 1983 Hyperthermia bymagnetic induction: experimental and theoretical resultsfor coaxial coil pairs Radiat. Res. 95 175–86

[108] Reilly J P 1992 Principles of nerve and heart excitation bytime-varying magnetic fields Ann. New York Acad. Sci.649 96–117

[109] Oleson J R, Heusinkveld R S and Manning M R 1983Hyperthermia by magnetic induction: II. Clinicalexperience with concentric electrodes Int. J. Radiat.Oncol. Biol. Phys. 9 549–56

[110] Atkinson W J, Brezovich I A and Chakraborty D P 1984Usable frequencies in hyperthermia with thermal seedsIEEE Trans. Biomed. Eng. BME 31 70–5

[111] Jordan A, Wust P, Fahling H, Johns W, Hinz A and Felix R1993 Inductive heating of ferrimagnetic particles andmagnetic fluids: physical evaluation of their potential forhyperthermia Int. J. Hyperthermia 9 51–68

[112] Jordan A et al 2001 Presentation of a new magnetic fieldtherapy system for the treatment of human solid tumourswith magnetic fluid hyperthermia J. Magn. Magn. Mater.225 118–26

[113] Rosensweig R E 2002 Heating magnetic fluid with alternatingmagnetic field J. Magn. Magn. Mater. 252 370–4

[114] Debye P 1929 Polar Molecules (New York: The ChemicalCatalog Company)

[115] Fannin P C and Charles S W 1991 Measurement of theNeel relaxation of magnetic particles in the frequencyrange 1 kHz–160 MHz J. Phys. D: Appl. Phys.24 76–7

[116] Fannin P C, Scaife B K P and Charles S W 1993 Relaxationand resonance in ferrofluids J. Magn. Magn. Mater. 122159–63

[117] Hanson M 1991 The frequency dependence of the complexsusceptibility of magnetic liquids J. Magn. Magn. Mater.96 105–13

[118] Jones S K, Gray B N, Burton M A, Codde J P and Street R1992 Evaluation of ferromagnetic materials for lowfrequency hysteresis heating of tumours Phys. Med. Biol.37 293–9

[119] Maehara T, Konishi K, Kamimori T, Aono H, Naohara T,Kikkawa H, Watanabe Y and Kawachi K 2002 Heating offerrite powder by an AC magnetic field for localhyperthermia Japan. J. Appl. Phys. 41 1620–1

[120] Kirpotin D, Chan D C F and Bunn P A 1995 Magneticmicroparticles US Patents vol 5,411,730 (Tucson, AZ)(USA: Research Corporation Technologies, Inc.)

[121] Hergt R, Andra W, d’Ambly C, Hilger I, Kaiser W, Richter Uand Schmidt H 1998 Physical limits of hyperthermiausing magnetite fine particles IEEE Trans. Magn. 343745–54

[122] Livingston J D 1996 Driving Force: The Natural Magic ofMagnets (Cambridge: Harvard University Press)

[123] Elster A and Burdette J 2001 Questions and Answers inMagnetic Resonance Imaging (St Louis, USA:Mosby)

[124] Leach M O 1988 Spatially localised nuclear magneticresonance The Physics of Medical Imaging ed S Webb(Bristol, UK: Adam Hilger) pp 389–487

[125] Browne M and Semelka R C 1999 MRI: Basic Principles andApplications (New York: Wiley)

[126] Lawaczeck R, Bauer H, Frenzel T, Hasegawa M, Ito Y,Kito K, Miwa N, Tsutsui H, Vogler H and Weinmann H J1997 Magnetic iron oxide particles coated withcarboxydextran for parenteral administration and livercontrasting Acta Radiol. 38 584–97

[127] Weissleder R, Elizondo G, Wittenburg J, Rabito C A,Bengele H H and Josephson L 1990 Ultrasmallsuperparamagnetic iron oxide: characterization of a newclass of contrast agents for MR imaging Radiol. 175489–93

[128] Ruehm S G, Corot C, Vogt P, Kolb S and Debatin J F 2001Magnetic resonance imaging of atherosclerotic plaquewith ultrasmall superparamagnetic particles of ironoxide in hyperlipidemic rabbits Circulation 103415–22

[129] Wacker F K, Reither K, Ebert W, Wendt M, Lewin J S andWolf K J 2003 MR image-guided endovascular procedureswith the ultrasmall superparamagnetic iron oxideSHU555C as an intravascular contrast agent: study in pigsRadiology 226 459–64

[130] Dosset V, Gomez C, Petry K G, Delalande C and Caille J-M1999 Dose and scanning delay using USPIO for centralnervous system macrophage imaging Magn. Res. Mater.Phys., Biol. Med. 8 185–9

[131] Michel S C A, Keller T M, Frohlich J M, Fink D, Caduff R,Seifert B, Marincek B and Kubik-Huch R A 2002Preoperative breast cancer staging: MR imaging of theaxilla with ultrasmall superparamagnetic iron oxideenhancement Radiology 225 527–36

[132] Semelka R C and Helmberger T K G 2001 Contrast agentsfor MR imaging of the liver Radiology 218 27–38

[133] Enochs W S, Harsh G, Hochberg F and Weissleder R 1999Improved delineation of human brain tumors on MRimages using a long-circulating, superparamagnetic ironoxide agent J. Magn. Res. Imag. 9 228–32

[134] Bulte J, de Cuyper M, Despres D and Frank J 1999 Short- vslong-circulating magnetoliposomes as bonemarrow-seeking MR contrast agents J. Magn. Res. Imag. 9329–35

[135] Bulte J et al 2001 Magnetodendrimers allow endosomalmagnetic labeling and in vivo tracking of stem cells NatureBiotechnol. 19 1141–7

[136] Hoehn M et al 2002 Monitoring of implanted stem cellmigration in vivo: a highly resolved in vivo magneticresonance imaging investigation of experimental stroke inrat Proc. Natl Acad. Sci. USA 99 16267–72

[137] Weissleder R, Moore A, Mahmood U, Bhorade R,Benveniste H, Chiocca E A and Basilion J P 2000 In vivomagnetic resonance imaging of transgene expressionNature Med. 6 351–4

[138] Zhao M, Beauregard D, Loizou L, Davletov B and Brindle K2001 Non-invasive detection of apoptosis using magneticresonance imaging and a targeted contrast agent NatureMed. 7 1241–4

[139] Wang N, Butler J P and Ingber D E 1993Mechanotransduction across the cell surface and throughthe cytoskeleton Science 260 1124

[140] Fabry B, Maksym G N, Hubmayr R D, Butler J P andFredberg J J 1999 Implications of heterogeneous beadbehavior on cell mechanical properties measured withmagnetic twisting cytometry J. Magn. Magn. Mater. 194120–5

[141] Fabry B, Maksym G N, Butler J P, Glogauer M, Navajas Dand Fredberg J J 2001 Scaling the microrheology of livingcells Phys. Rev. Lett. 87 148102

[142] Bausch A R, Ziemann F, Boulbitch A A, Jacobson K andSackmann E 1998 Local measurements of viscoelastic

R180

Topical Review

parameters of adherent cell surfaces by magnetic beadmicrorheometry Biophys. J. 75 2038–49

[143] Bausch A R, Moeller W and Sackmann E 1999Measurement of local viscoelasticity and forces inliving cells by magnetic tweezers Biophys. J. 76573–9

[144] Dobson J, Keramane A and El Haj A J 2002 Theory andapplications of a magnetic force bioreactor Eur. CellsMater. 4 130–1

[145] Cartmell S H, Dobson J, Verschueren S, Hughes S andEl Haj A J 2002 Mechanical conditioning of bone cellsin vitro using magnetic microparticle technology Eur.Cells Mater. 4 42–4

[146] Cartmell S H, Dobson J, Verschueren S and El Haj A J 2003Preliminary analysis of magnetic particle techniques foractivating mechanotransduction in bone cells IEEE Trans.NanoBiosci. at press

[147] Perez J M, Josephson L, O Loughlin T, Hogemann D andWeissleder R 2002 Magnetic relaxation switches capableof sensing molecular interactions Nature Biotechnol. 20816–20

[148] Perez J M, O Loughin T, Simeone F J, Weissleder R andJosephson L 2002 DNA-based magnetic nanoparticleassembly acts as a magnetic relaxation nanoswitchallowing screening of DNA-cleaving agents J. Am. Chem.Soc. 124 2856–7

R181

BioMed CentralJournal of Nanobiotechnology

ss
Open AcceReviewApplications of nanoparticles in biology and medicineOV Salata*
Address: Sir William Dunn School of Pathology, University of Oxford, South Parks Road, Oxford OX1 3RE, UK

Email: OV Salata* - [email protected]

* Corresponding author

nanotechnologynanomaterialsnanoparticlesquantum dotsnanotubesmedicinebiologyapplications

AbstractNanomaterials are at the leading edge of the rapidly developing field of nanotechnology. Theirunique size-dependent properties make these materials superior and indispensable in many areasof human activity. This brief review tries to summarise the most recent developments in the fieldof applied nanomaterials, in particular their application in biology and medicine, and discusses theircommercialisation prospects.

IntroductionNanotechnology [1] is enabling technology that dealswith nano-meter sized objects. It is expected that nanote-chnology will be developed at several levels: materials,devices and systems. The nanomaterials level is the mostadvanced at present, both in scientific knowledge and incommercial applications. A decade ago, nanoparticleswere studied because of their size-dependent physical andchemical properties [2]. Now they have entered a com-mercial exploration period [3,4].

Living organisms are built of cells that are typically 10 µmacross. However, the cell parts are much smaller and arein the sub-micron size domain. Even smaller are the pro-teins with a typical size of just 5 nm, which is comparablewith the dimensions of smallest manmade nanoparticles.This simple size comparison gives an idea of using nano-particles as very small probes that would allow us to spyat the cellular machinery without introducing too muchinterference [5]. Understanding of biological processes onthe nanoscale level is a strong driving force behind devel-opment of nanotechnology [6].

Out of plethora of size-dependant physical propertiesavailable to someone who is interested in the practicalside of nanomaterials, optical [7] and magnetic [8] effectsare the most used for biological applications.

The aim of this review is firstly to give reader a historicprospective of nanomaterial application to biology andmedicine, secondly to try to overview the most recentdevelopments in this field, and finally to discuss the hardroad to commercialisation. Hybrid bionanomaterials canalso be applied to build novel electronic, optoelectronicsand memory devices (see for example [9,10]). Neverthe-less, this will not be discussed here and will be a subject ofa separate article.

ApplicationsA list of some of the applications of nanomaterials to biol-ogy or medicine is given below:

- Fluorescent biological labels [11-13]

- Drug and gene delivery [14,15]

Published: 30 April 2004

Journal of Nanobiotechnology 2004, 2:3

Received: 23 December 2003Accepted: 30 April 2004

This article is available from: http://www.jnanobiotechnology.com/content/2/1/3

© 2004 Salata; licensee BioMed Central Ltd. This is an Open Access article: verbatim copying and redistribution of this article are permitted in all media for any purpose, provided this notice is preserved along with the article's original URL.

Page 1 of 6(page number not for citation purposes)

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=15119954

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=10.1186/1477-3155-2-3

http://www.jnanobiotechnology.com/content/2/1/3

http://www.biomedcentral.com/

http://www.biomedcentral.com/info/about/charter/

Journal of Nanobiotechnology 2004, 2 http://www.jnanobiotechnology.com/content/2/1/3

- Bio detection of pathogens [16]

- Detection of proteins [17]

- Probing of DNA structure [18]

- Tissue engineering [19,20]

- Tumour destruction via heating (hyperthermia)[21]

- Separation and purification of biological molecules andcells [22]

- MRI contrast enhancement [23]

- Phagokinetic studies [24]

As mentioned above, the fact that nanoparticles exist inthe same size domain as proteins makes nanomaterialssuitable for bio tagging or labelling. However, size is justone of many characteristics of nanoparticles that itself israrely sufficient if one is to use nanoparticles as biologicaltags. In order to interact with biological target, a biologicalor molecular coating or layer acting as a bioinorganicinterface should be attached to the nanoparticle. Exam-ples of biological coatings may include antibodies,biopolymers like collagen [25], or monolayers of smallmolecules that make the nanoparticles biocompatible[26]. In addition, as optical detection techniques are widespread in biological research, nanoparticles should eitherfluoresce or change their optical properties. Theapproaches used in constructing nano-biomaterials areschematically presented below (see Figure 1).

Nano-particle usually forms the core of nano-biomaterial.It can be used as a convenient surface for molecularassembly, and may be composed of inorganic or poly-meric materials. It can also be in the form of nano-vesiclesurrounded by a membrane or a layer. The shape is moreoften spherical but cylindrical, plate-like and other shapesare possible. The size and size distribution might beimportant in some cases, for example if penetrationthrough a pore structure of a cellular membrane isrequired. The size and size distribution are becomingextremely critical when quantum-sized effects are used tocontrol material properties. A tight control of the averageparticle size and a narrow distribution of sizes allow creat-ing very efficient fluorescent probes that emit narrow lightin a very wide range of wavelengths. This helps with creat-ing biomarkers with many and well distinguished colours.The core itself might have several layers and be multifunc-tional. For example, combining magnetic and lumines-cent layers one can both detect and manipulate theparticles.

The core particle is often protected by several monolayersof inert material, for example silica. Organic moleculesthat are adsorbed or chemisorbed on the surface of theparticle are also used for this purpose. The same layermight act as a biocompatible material. However, moreoften an additional layer of linker molecules is required toproceed with further functionalisation. This linear linkermolecule has reactive groups at both ends. One group isaimed at attaching the linker to the nanoparticle surfaceand the other is used to bind various moieties like bio-compatibles (dextran), antibodies, fluorophores etc.,depending on the function required by the application.

Recent developmentsTissue engineeringNatural bone surface is quite often contains features thatare about 100 nm across. If the surface of an artificial boneimplant were left smooth, the body would try to reject it.Because of that smooth surface is likely to cause produc-tion of a fibrous tissue covering the surface of the implant.This layer reduces the bone-implant contact, which mayresult in loosening of the implant and further inflamma-tion. It was demonstrated that by creating nano-sized fea-tures on the surface of the hip or knee prosthesis onecould reduce the chances of rejection as well as to stimu-late the production of osteoblasts. The osteoblasts are thecells responsible for the growth of the bone matrix and arefound on the advancing surface of the developing bone.

Typical configurations utilised in nano-bio materials applied to medical or biological problemsFigure 1Typical configurations utilised in nano-bio materials applied to medical or biological problems.



The effect was demonstrated with polymeric, ceramic and,more recently, metal materials. More than 90% of thehuman bone cells from suspension adhered to the nanos-tructured metal surface [27], but only 50% in the controlsample. In the end this findings would allow to design amore durable and longer lasting hip or knee replacementsand to reduce the chances of the implant getting loose.

Titanium is a well-known bone repairing material widelyused in orthopaedics and dentistry. It has a high fractureresistance, ductility and weight to strength ratio. Unfortu-nately, it suffers from the lack of bioactivity, as it does notsupport sell adhesion and growth well. Apatite coatingsare known to be bioactive and to bond to the bone.Hence, several techniques were used in the past to pro-duce an apatite coating on titanium. Those coatings sufferfrom thickness non-uniformity, poor adhesion and lowmechanical strength. In addition, a stable porous structureis required to support the nutrients transport through thecell growth.

It was shown that using a biomimetic approach – a slowgrowth of nanostructured apatite film from the simulatedbody fluid – resulted in the formation of a strongly adher-ent, uniform nanoporous layer [19]. The layer was foundto be built of 60 nm crystallites, and possess a stable nan-oporous structure and bioactivity.

A real bone is a nanocomposite material, composed ofhydroxyapatite crystallites in the organic matrix, which ismainly composed of collagen. Thanks to that, the bone ismechanically tough and, at the same time, plastic, so itcan recover from a mechanical damage. The actual nano-scale mechanism leading to this useful combination ofproperties is still debated.

An artificial hybrid material was prepared from 15–18 nmceramic nanoparticles and poly (methyl methacrylate)copolymer [20]. Using tribology approach, a viscoelasticbehaviour (healing) of the human teeth was demon-strated. An investigated hybrid material, deposited as acoating on the tooth surface, improved scratch resistanceas well as possessed a healing behaviour similar to that ofthe tooth.

Cancer therapyPhotodynamic cancer therapy is based on the destructionof the cancer cells by laser generated atomic oxygen,which is cytotoxic. A greater quantity of a special dye thatis used to generate the atomic oxygen is taken in by thecancer cells when compared with a healthy tissue. Hence,only the cancer cells are destroyed then exposed to a laserradiation. Unfortunately, the remaining dye moleculesmigrate to the skin and the eyes and make the patient very

sensitive to the daylight exposure. This effect can last forup to six weeks.

To avoid this side effect, the hydrophobic version of thedye molecule was enclosed inside a porous nanoparticle[28]. The dye stayed trapped inside the Ormosil nanopar-ticle and did not spread to the other parts of the body. Atthe same time, its oxygen generating ability has not beenaffected and the pore size of about 1 nm freely allowed forthe oxygen to diffuse out.

Multicolour optical coding for biological assays [29]The ever increasing research in proteomics and genomicgenerates escalating number of sequence data andrequires development of high throughput screening tech-nologies. Realistically, various array technologies that arecurrently used in parallel analysis are likely to reach satu-ration when a number of array elements exceed severalmillions. A three-dimensional approach, based on optical"bar coding" of polymer particles in solution, is limitedonly by the number of unique tags one can reliably pro-duce and detect.

Single quantum dots of compound semiconductors weresuccessfully used as a replacement of organic dyes in vari-ous bio-tagging applications [7]. This idea has been takenone step further by combining differently sized and hencehaving different fluorescent colours quantum dots, andcombining them in polymeric microbeads [29]. A precisecontrol of quantum dot ratios has been achieved. Theselection of nanoparticles used in those experiments had6 different colours as well as 10 intensities. It is enough toencode over 1 million combinations. The uniformity andreproducibility of beads was high letting for the beadidentification accuracies of 99.99%.

Manipulation of cells and biomolecules [30]Functionalised magnetic nanoparticles have found manyapplications including cell separation and probing; theseand other applications are discussed in a recent review [8].Most of the magnetic particles studied so far are spherical,which somewhat limits the possibilities to make thesenanoparticles multifunctional. Alternative cylindricallyshaped nanoparticles can be created by employing metalelectrodeposition into nanoporous alumina template[30]. Depending on the properties of the template, nano-cylinder radius can be selected in the range of 5 to 500 nmwhile their length can be as big as 60 µm. By sequentiallydepositing various thicknesses of different metals, thestructure and the magnetic properties of individual cylin-ders can be tuned widely.

As surface chemistry for functionalisation of metal sur-faces is well developed, different ligands can be selectivelyattached to different segments. For example, porphyrins



with thiol or carboxyl linkers were simultaneouslyattached to the gold or nickel segments respectively. Thus,it is possible to produce magnetic nanowires with spa-tially segregated fluorescent parts. In addition, because ofthe large aspect ratios, the residual magnetisation of thesenanowires can be high. Hence, weaker magnetic field canbe used to drive them. It has been shown that a self-assem-bly of magnetic nanowires in suspension can be control-led by weak external magnetic fields. This wouldpotentially allow controlling cell assembly in differentshapes and forms. Moreover, an external magnetic fieldcan be combined with a lithographically defined mag-netic pattern ("magnetic trapping").

Protein detection [31]Proteins are the important part of the cell's language,machinery and structure, and understanding their func-tionalities is extremely important for further progress inhuman well being. Gold nanoparticles are widely used inimmunohistochemistry to identify protein-protein inter-action. However, the multiple simultaneous detectioncapabilities of this technique are fairly limited. Surface-enhanced Raman scattering spectroscopy is a well-estab-lished technique for detection and identification of singledye molecules. By combining both methods in a singlenanoparticle probe one can drastically improve the multi-plexing capabilities of protein probes. The group of Prof.Mirkin has designed a sophisticated multifunctional

Table 1: Examples of Companies commercialising nanomaterials for bio- and medical applications.

Company Major area of activity Technology

Advectus Life Sciences Inc. Drug delivery Polymeric nanoparticles engineered to carry anti-tumour drug across the blood-brain barrier

Alnis Biosciences, Inc. Bio-pharmaceutical Biodegradable polymeric nanoparticles for drug delivery

Argonide Membrane filtration Nanoporous ceramic materials for endotoxin filtration, orthopaedic and dental implants, DNA and protein separation

BASF Toothpaste Hydroxyapatite nanoparticles seems to improve dental surface

Biophan Technologies, Inc. MRI shielding Nanomagnetic/carbon composite materials to shield medical devices from RF fields

Capsulution NanoScience AG Pharmaceutical coatings to improve solubility of drugs Layer-by-layer poly-electrolyte coatings, 8–50 nmDynal Biotech Magnetic beadsEiffel Technologies Drug delivery Reducing size of the drug particles to 50–100 nmEnviroSystems, Inc. Surface desinfectsant NanoemulsionsEvident Technologies Luminescent biomarkers Semiconductor quantum dots with amine or carboxyl

groups on the surface, emission from 350 to 2500 nmImmunicon Tarcking and separation of different cell types magnetic core surrounded by a polymeric layer

coated with antibodies for capturing cellsKES Science and Technology, Inc. AiroCide filters Nano-TiO2 to destroy airborne pathogensNanoBio Cortporation Pharmaceutical Antimicrobal nano-emulsionsNanoCarrier Co., Ltd Drug delivery Micellar nanoparticles for encapsulation of drugs,

proteins, DNANanoPharm AG Drug delivery Polybutilcyanoacrylate nanoparticles are coated with

drugs and then with surfactant, can go across the blood-brain barrier

Nanoplex Technologies, Inc Nanobarcodes for bioanalysisNanoprobes, Inc. Gold nanoparticles for biological markers Gold nanoparticles bio-conjugates for TEM and/or

fluorescent microscopyNanoshpere, Inc. Gold biomarkers DNA barcode attached to each nanoprobe for

identification purposes, PCR is used to amplify the signal; also catalytic silver deposition to amplify the signal using surface plasmon resonance

NanoMed Pharmaceutical, Inc. Drug delivery Nanoparticles for drug deliveryOxonica Ltd Sunscreens Doped transparent nanoparticles to effectively

absorb harmful UV and convert it into heatPSiVida Ltd Tissue engineering, implants, drugs and gene delivery,

bio-filtrationExploiting material properties of nanostructured porous silicone

Smith & Nephew Acticoat bandages Nanocrystal silver is highly toxic to pathogenesQuantumDot Corporation Luminescent biomarkers Bioconjugated semiconductor quantum dots



probe that is built around a 13 nm gold nanoparticle. Thenanoparticles are coated with hydrophilic oligonucle-otides containing a Raman dye at one end and terminallycapped with a small molecule recognition element (e.g.biotin). Moreover, this molecule is catalytically active andwill be coated with silver in the solution of Ag(I) and hyd-roquinone. After the probe is attached to a small moleculeor an antigen it is designed to detect, the substrate isexposed to silver and hydroquinone solution. A silver-plating is happening close to the Raman dye, whichallows for dye signature detection with a standard Ramanmicroscope. Apart from being able to recognise smallmolecules this probe can be modified to contain antibod-ies on the surface to recognise proteins. When tested in theprotein array format against both small molecules andproteins, the probe has shown no cross-reactivity.

Commercial explorationSome of the companies that are involved in the develop-ment and commercialisation of nanomaterials in biologi-cal and medical applications are listed below (see Table1). The majority of the companies are small recentspinouts of various research institutions. Although notexhausting, this is a representative selection reflectingcurrent industrial trends. Most of the companies are devel-oping pharmaceutical applications, mainly for drug deliv-ery. Several companies exploit quantum size effects insemiconductor nanocrystals for tagging biomolecules, oruse bio-conjugated gold nanoparticles for labelling vari-ous cellular parts. A number of companies are applyingnano-ceramic materials to tissue engineering andorthopaedics.

Most major and established pharmaceutical companieshave internal research programs on drug delivery that areon formulations or dispersions containing componentsdown to nano sizes. Colloidal silver is widely used in anti-microbial formulations and dressings. The high reactivityof titania nanoparticles, either on their own or then illu-minated with UV light, is also used for bactericidal pur-poses in filters. Enhanced catalytic properties of surfacesof nano-ceramics or those of noble metals like platinumare used to destruct dangerous toxins and other hazardousorganic materials.

Future directionsAs it stands now, the majority of commercial nanoparticleapplications in medicine are geared towards drug delivery.In biosciences, nanoparticles are replacing organic dyes inthe applications that require high photo-stability as wellas high multiplexing capabilities. There are some develop-ments in directing and remotely controlling the functionsof nano-probes, for example driving magnetic nanoparti-cles to the tumour and then making them either to releasethe drug load or just heating them in order to destroy the

surrounding tissue. The major trend in further develop-ment of nanomaterials is to make them multifunctionaland controllable by external signals or by local environ-ment thus essentially turning them into nano-devices.

References1. Feynman R: There's plenty of room at the bottom. Science 1991,

254:1300-1301.2. Murray CB, Kagan CR, Bawendi MG: Synthesis and characterisa-

tion of monodisperse nanocrystals and close-packed nanoc-rystal assemblies. Annu Rev Mater Sci 2000, 30:545-610.

3. Mazzola L: Commercializing nanotechnology. NatureBiotechnology 2003, 21:1137-1143.

4. Paull R, Wolfe J, Hebert P, Sinkula M: Investing innanotechnology. Nature Biotechnology 2003, 21:1134-1147.

5. Taton TA: Nanostructures as tailored biological probes. TrendsBiotechnol 2002, 20:277-279.

6. Whitesides GM: The 'right' size in Nanobiotechnology. NatureBiotechnology 2003, 21:1161-1165.

7. Parak WJ, Gerion D, Pellegrino T, Zanchet D, Micheel C, WilliamsCS, Boudreau R, Le Gros MA, Larabell CA, Alivisatos AP: Biologicalapplications of colloidal nanocrystals. Nanotechnology 2003,14:R15-R27.

8. Pankhurst QA, Connolly J, Jones SK, Dobson J: Applications ofmagnetic nanoparticles in biomedicine. J Phys D: Appl Phys 2003,36:R167-R181.

9. Yan H, Park SH, Finkelstein G, Reif JH, LaBean TH: DNA-templatedself-assembly of protein arrays and highly conductivenanowires. Science 2003, 301:1882-1884.

10. Keren K, Berman RS, Buchstab E, Sivan U, Braun E: DNA-tem-plated carbon nanotube field-effect transistor. Science 2003,302:1380-1382.

11. Bruchez M, Moronne M, Gin P, Weiss S, Alivisatos AP: Semiconduc-tor nanocrystals as fluorescent biological labels. Science 1998,281:2013-2016.

12. Chan WCW, Nie SM: Quantum dot bioconjugates for ultrasen-sitive nonisotopic detection. Science 1998, 281:2016-2018.

13. Wang S, Mamedova N, Kotov NA, Chen W, Studer J: Antigen/anti-body immunocomplex from CdTe nanoparticlebioconjugates. Nano Letters 2002, 2:817-822.

14. Mah C, Zolotukhin I, Fraites TJ, Dobson J, Batich C, Byrne BJ: Micro-sphere-mediated delivery of recombinant AAV vectors invitro and in vivo. Mol Therapy 2000, 1:S239.

15. Panatarotto D, Prtidos CD, Hoebeke J, Brown F, Kramer E, Briand JP,Muller S, Prato M, Bianco A: Immunization with peptide-func-tionalized carbon nanotubes enhances virus-specific neutral-izing antibody responses. Chemistry&Biology 2003, 10:961-966.

16. Edelstein RL, Tamanaha CR, Sheehan PE, Miller MM, Baselt DR, Whit-man LJ, Colton RJ: The BARC biosensor applied to the detec-tion of biological warfare agents. Biosensors Bioelectron 2000,14:805-813.

17. Nam JM, Thaxton CC, Mirkin CA: Nanoparticles-based bio-barcodes for the ultrasensitive detection of proteins. Science2003, 301:1884-1886.

18. Mahtab R, Rogers JP, Murphy CJ: Protein-sized quantum dotluminescence can distinguish between "straight", "bent",and "kinked" oligonucleotides. J Am Chem Soc 1995,117:9099-9100.

19. Ma J, Wong H, Kong LB, Peng KW: Biomimetic processing ofnanocrystallite bioactive apatite coating on titanium. Nanote-chnology 2003, 14:619-623.

20. de la Isla A, Brostow W, Bujard B, Estevez M, Rodriguez JR, Vargas S,Castano VM: Nanohybrid scratch resistant coating for teethand bone viscoelasticity manifested in tribology. Mat ResrInnovat 2003, 7:110-114.

21. Yoshida J, Kobayashi T: Intracellular hyperthermia for cancerusing magnetite cationic liposomes. J Magn Magn Mater 1999,194:176-184.

22. Molday RS, MacKenzie D: Immunospecific ferromagnetic irondextran reagents for the labeling and magnetic separation ofcells. J Immunol Methods 1982, 52:353-367.

23. Weissleder R, Elizondo G, Wittenburg J, Rabito CA, Bengele HH,Josephson L: Ultrasmall superparamagnetic iron oxide: char-


http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=10.1146/annurev.matsci.30.1.545



http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=10.1038/nbt1003-1137




http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=10.1016/S0167-7799(02)01973-X


http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=10.1038/nbt872


http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=10.1088/0957-4484/14/7/201




http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=10.1126/science.1089389







http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=10.1126/science.281.5385.2013






http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=10.1021/nl0255193



http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=10.1006/mthe.2000.0174




http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=10.1016/S0956-5663(99)00054-8







http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=10.1016/0022-1759(82)90007-2






Publish with BioMed Central and every scientist can read your work free of charge

"BioMed Central will be the most significant development for disseminating the results of biomedical research in our lifetime."

Sir Paul Nurse, Cancer Research UK

Your research papers will be:

available free of charge to the entire biomedical community

peer reviewed and published immediately upon acceptance

cited in PubMed and archived on PubMed Central

yours — you keep the copyright

Submit your manuscript here:http://www.biomedcentral.com/info/publishing_adv.asp

BioMedcentral

acterization of a new class of contrast agents for MRimaging. Radiology 1990, 175:489-493.

24. Parak WJ, Boudreau R, Gros ML, Gerion D, Zanchet D, Micheel CM,Williams SC, Alivisatos AP, Larabell CA: Cell motility and meta-static potential studies based on quantum dot imaging ofphagokinetic tracks. Adv Mater 2002, 14:882-885.

25. Sinani VA, Koktysh DS, Yun BG, Matts RL, Pappas TC, Motamedi M,Thomas SN, Kotov NA: Collagen coating promotes biocompat-ibility of semiconductor nanoparticles in stratified LBL films.Nano Letters 2003, 3:1177-1182.

26. Zhang Y, Kohler N, Zhang M: Surface modification of superpar-amagnetic magnetite nanoparticles and their intracellularuptake. Biomaterials 2002, 23:1553-1561.

27. Gutwein LG, Webster TJ: Affects of alumina and titania nano-particulates on bone cell function. American Ceramic Society 26 thAnnual Meeting Conference Proceedings 2003 in press.

28. Roy I, Ohulchanskyy TY, Pudavar HE, Bergey EJ, Oseroff AR, MorganJ, Dougherty TJ, Prasad PN: Ceramic-based nanoparticlesentrapping water-insoluble photosensitizing anticancerdrugs: a novel drug-carrier system for photodynamictherapy. J Am Chem Soc 2003, 125:7860-7865.

29. Han M, Gao X, Su JZ, Nie S: Quantum-dot-tagged microbeadsfor multiplexed optical coding of biomolecules. NatureBiotechnology 2001, 19:631-635.

30. Reich DH, Tanase M, Hultgren A, Bauer LA, Chen CS, Meyer GJ: Bio-logical applications of multifunctional magnetic nanowires. JAppl Phys 2003, 93:7275-7280.

31. Cao YC, Jin R, Nam JM, Thaxton CS, Mirkin CA: Raman dye-labeled nanoparticle probes for proteins. JACS 2003,125:14676-14677.




http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=10.1002/1521-4095(20020618)14:12<882::AID-ADMA882>3.0.CO;2-Y









http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=10.1021/ja0343095




http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=10.1038/90228

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=10.1038/90228


http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=10.1063/1.1558672

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Abstract&list_uids=10.1063/1.1558672




http://www.biomedcentral.com/info/publishing_adv.asp


Chemical synthesis of magnetic nanoparticles

Taeghwan Hyeon National Creative Research Initiative Center for Oxide Nanocrystalline Materials and School of ChemicalEngineering, Seoul National University, Seoul 151-744, Korea. E-mail: [email protected]

Received (in Cambridge, UK) 9th September 2002, Accepted 12th November 2002First published as an Advance Article on the web 3rd December 2002

Recent advances in the synthesis of various magnetic nano-particles using colloidal chemical approaches are reviewed.Typically, these approaches involve either rapid injection ofreagents into hot surfactant solution followed by aging at hightemperature, or the mixing of reagents at a low temperature andslow heating under controlled conditions. Spherical cobaltnanoparticles with various crystal structures have been synthe-sized by thermally decomposing dicobalt octacarbonyl or byreducing cobalt salts. Nanoparticles of Fe–Pt and other relatediron or cobalt containing alloys have been made by simultane-ously reacting their constituent precursors. Many differentferrite nanoparticles have been synthesized by the thermaldecomposition of organometallic precursors followed by oxida-tion or by low-temperature reactions inside reverse micelles.Rod-shaped iron nanoparticles have been synthesized from theoriented growth of spherical nanoparticles, and cobalt nano-disks were synthesized from the thermal decomposition ofdicobalt octacarbonyl in the presence of a mixture of twosurfactants.

IntroductionThe development of uniform nanometer sized particles has beenintensively pursued because of their technological and funda-mental scientific importance.1 These nanoparticulate materialsoften exhibit very interesting electrical, optical, magnetic, andchemical properties, which cannot be achieved by their bulkcounterparts.2 The synthesis of discrete magnetic nanoparticleswith sizes ranging from 2 to 20 nm, is of significant importance,because of their applications in multi-terabit in22 magneticstorage devices.3 Such magnetic nanoparticles could also find

applications in ferrofluids, magnetic refrigeration systems,contrast enhancement in magnetic resonance imaging, magneticcarriers for drug targeting and catalysis.4 In the current article,I will discuss recent advances made in the synthesis of variousmagnetic nanoparticles using colloidal chemical syntheticprocedures. A general review on the synthesis and under-standing of nanostructured magnetic materials up to 1996 isgiven in the article by Leslie-Pelecky and Rieke.5 The materialscovered in the current article are colloidal systems composed ofisolated particles with nanometer-sized dimensions that arestabilized by surfactant molecules and dispersed in solventmedia. In the ideal case, these non-interacting systems derivetheir unique magnetic properties mostly from the reduced sizeof the isolated nanoparticles, and contributions from inter-particle interactions are negligible. The surfactant coating onmagnetic nanoparticles prevents clustering due to steric repul-sion. Dynamic adsorption and desorption of surfactant mole-cules onto particle surfaces during synthesis enables reactivespecies to be added onto the growing particles. Thesenanoparticles can be dispersed in many organic solvents and canbe retrieved as powder forms by removing the solvent.

The followings are several key issues of nanoparticlesynthesis.

4 Particle size distribution (uniformity): Can we synthesizemonodisperse nanocrystals?4 Particle size control: Can we control the particle size ofnanocrystals in a reproducible manner?4 Crystallinity and crystal structure: Can we obtain materialswith satisfactory high crystallinity and the desired crystalstructure?4 Shape-control: Can we synthesize non-spherical and aniso-tropic nanoparticles?4 Alignment: For device applications, we should also considerthe alignment of nanoparticles on substrates.

In the present article, I am especially interested in thesynthesis of monodisperse magnetic nanoparticles. Mono-disperse nanoparticles are generally considered to be sampleswith standard deviations s @ 5% diameter for sphericalparticles. Often nanocrystals are referred to as well-definedcrystalline materials, whereas nanoparticles is a term used moregenerally for particles with diameters of 2–50 nm with variablecrystallinity.

Chemical methods have been widely used to producenanostructured materials due to their straightforward nature andtheir potential to produce large quantities of the final product.Realizable particle sizes range in size from nanometers tomicrometers, by controlling particle size during synthesis byusing competition between nucleation and growth. There areseveral synthetic procedures for synthesizing monodispersenanoparticles. It is well known that a short burst of nucleationfollowed by slow controlled growth is critical to producemonodisperse particles. The following describe two representa-

Taeghwan Hyeon received his B. S. (1987) and M. S. (1989) inChemistry from Seoul National University, Seoul, Korea. Heobtained his Ph.D. from the University of Illinois at Urbana-Champaign (1996) under the supervision of Professor KennethS. Suslick. After conducting postdoctoral research with Pro-fessor Wolfgang M. H. Sachtler at Northwestern University, hejoined the faculty of the School of Chemical Engineering ofSeoul National University in September 1997. He has receivedseveral awards including the T. S. Piper Award (University ofIllinois, 1996), Korean Young Scientist Award (Korean Govern-ment, 2002), KCS–Wiley Young Chemist Award (KoreanChemical Society, 2001), and The Scientist of the Month Award(Korean Science and Engineering Foundation, 2002). In July2002, he became the director of the National Creative ResearchInitiative Center for Oxide Nanocrystalline Materials sup-ported by the Korean Ministry of Science and Technology. Hiscurrent research interests include synthesis of metallic andoxide nanoparticles, synthesis of nanoporous carbon materials,and catalytic applications of nanostructured materials.

Th is journa l i s © The Roya l Soc ie ty of Chemist ry 2003

DO

I: 10

.103

9/b

2077

89b

927CHEM. COMMUN. , 2003, 927–934

tive synthetic procedures. In the first synthetic procedure (Fig.1), the rapid injection of the reagents, often organometallic

compounds, into hot surfactant solution induces the simultane-ous formation of many nuclei.6 Alternatively, reducing agentsare added to metal salt solutions at high temperature. In thesecond procedure, reagents are mixed at low temperature andthe resulting reaction mixtures are slowly heated in a controlledmanner to generate nuclei. Subsequently, particle growth occursby the addition of reactive species. Particle size can be alsoincreased by aging at high temperature by Oswalt ripening, inwhich smaller nanoparticles dissolve and deposit on the biggernanoparticles. The growth of nanoparticles can be stopped byrapidly decreasing the reaction temperature. Nanoparticlesproduced using these synthetic procedures often have particlesize distributions with s ~ 10%. Further size-selection proc-esses can narrow the particle size distribution below s = 5%.The most frequently applied size-selection involves the additionof a poor solvent to precipitate the larger particles. When poorsolvent is added to a mixture containing nanoparticles ofvarious sizes, the biggest particles flocculate first because oftheir greatest van der Waals attraction. This precipitate can beretrieved by centrifugation. The precipitate can be re-dissolvedin solvent, and further size-selection process can be performedto yield particles with narrower particle size distribution. Veryrecently, Klabunde and coworkers reported the fabrication ofextremely monodisperse gold nanoparticles through digestiveripening.7

The particle sizes of nanoparticles can be controlled bysystematically adjusting the reaction parameters, such as time,temperature, and the concentrations of reagents and stabilizingsurfactants. In general, particle size increases with increasingreaction time, because more monomeric species are generated,and with increasing reaction temperature because the rate ofreaction is increased.

Nanoparticles of iron, cobalt and nickelThe Murray group at the IBM Watson research center hasstudied the synthesis of monodisperse metallic magneticnanoparticles intensively.8,9 They have synthesized cobaltnanoparticles with several different crystal structures usingdifferent synthetic procedures. High temperature reduction ofcobalt chloride was employed to synthesize e-phase cobaltnanoparticles.9a The injection of superhydride (LiBEt3H)solution in dioctyl ether into a hot cobalt chloride solution indioctyl ether (200 °C) in the presence of oleic acid andtrialkylphosphine induced the simultaneous formation of manysmall metal clusters, which acted as nuclei for nanoparticleformation. Continued heating at 200 °C induced the growth ofthese clusters to the nanoparticle level. Particle size wascontrolled by the steric bulkiness of the stabilizing surfactants.

Short-chain alkylphosphines allowed faster growth, and re-sulted in the bigger particles, while bulkier surfactants reducedparticle growth and favored production of smaller nano-particles. For example, when bulky trioctylphosphine wasapplied in the synthesis, 2–6 nm sized particles were generated.In contrast, when tributylphosphine was used as a stabilizer7–11 nm sized nanoparticles were produced. A further narrow-ing of particle size distribution was proceeded using a sizeselective precipitation by the gradual addition of alcohol (e.g.ethanol) into a hydrocarbon (e.g. hexane) dispersion containingnanoparticles with wide size distribution. Monodisperse cobaltnanoparticles organize into two- and three-dimensional super-lattices. Fig. 2 shows a 2D assembly of 9 nm cobalt

nanoparticles. A high-resolution transmission electron micro-graph of a single nanoparticle, shown in the inset of Fig. 2,reveals its highly crystalline nature. The X-ray diffractionpattern of these nanoparticles reveals a complex cubic structurerelated to the beta phase of elemental manganese (e-Co).Heating the e-Co nanoparticles at 300 °C generated hcp Conanoparticles. The Bawendi group have reported on thesynthesis of e-Co nanoparticle powders from the thermaldecomposition of Co2(CO)8 in TOPO.10 The Alivisatos grouphave reported on the synthesis of monodisperse e-Co nano-particles by the rapid pyrolysis of dicobalt octacarbonyl in thepresence of a surfactant mixture composed of oleic acid, lauricacid and trioctylphosphine.11 They could synthesize 3–17 nmsized cobalt nanoparticles by controlling the precursor/surfac-tant ratio, the reaction temperature, and the injection time. Blacket al. later demonstrated spin-dependent electron transport usingself-assembled 10 nm e-Co nanoparticles over 100 nm wideelectrodes.12 Single electron tunneling was clearly observed forthe two-dimensional hexagonal arrays of cobalt nanoparticles.Similar single electron tunneling in cobalt nanoparticles wasalso reported by Pileni and workers.13

Hcp cobalt nanoparticles were synthesized by using the so-called polyol process, in which high boiling alcohol is appliedas both a reductant and a solvent.8 In a typical synthesis,1,2-dodecanediol is added into hydrated cobalt acetate solutiondissolved in diphenyl ether containing oleic acid and trioctyl-phosphine at 250 °C. Nanoparticles were isolated by size-selective precipitation, and particle size was controlled bychanging the relative concentration of precursor and stabilizer.For example, when a 1+1 molar ratio of cobalt acetate and oleicacid was used in the synthesis, 6–8 nm sized cobalt nano-particles were produced, while increasing the concentration ofstabilizing surfactants by a factor of two yielded smaller 3–6 nmnanoparticles. Particle size could be also varied by controllingthe steric bulkiness of the phosphine stabilizers, for example,

Fig. 1 Generalized synthesis of monodisperse nanoparticles by the injectionof reagents into hot surfactant solution followed by aging and size-selectiveprocess.

Fig. 2 TEM images of 9 nm e-Co nanoparticles (inset, high-resolution TEMimage).9a Reprinted with permission from reference 9(a). Copyright 1999American Institute of Physics.

928 CHEM. COMMUN. , 2003, 927–934

using tributylphosphine, 10–13 nm Co nanoparticles weregenerated. Using hydrated nickel acetate as a metal precursor,nickel nanoparticles with particle sizes in the range 8–13 nmwere obtained. Co/Ni alloy nanoparticles were also producedusing a mixture of cobalt acetate and nickel acetate through asimilar synthetic procedure.8b Diehl et al. used ferromagneticresonance (FMR) techniques to investigate the detailed mag-netic characteristics of cobalt nanoparticles with differentcrystalline structures.9b The structural characterization of anordered assembly of cobalt nanoparticles using high-resolutiontransmission electron microscopy was reported by Wang et al.9c

They synthesized multiply twinned fcc cobalt nanocrystals fromthe thermal decomposition of dicobalt octacarbonyl in thepresence of a surfactant mixture composed of oleic acid andtributylphosphine. An HRTEM image of the nanoparticlesrevealed complicated interference patterns.8

Reverse micelles, which are water-in-oil droplets stabilizedby a monolayer of surfactant, have been applied as nanoscalereactors for the synthesis of various nanoparticles.14 Chen et al.and later the Pileni group reported on the synthesis of relativelyuniform cobalt nanoparticles by the reduction of cobalt saltinside reverse micelles.15 Chen et al. synthesized cobaltnanoparticles with particle size in the range 1.8–4.4 nm byreducing CoCl2 with NaBH4 in reverse micelles formed usingdidodecyldimethylammonium bromide (DDAB). The Pilenigroup reported the synthesis of cobalt nanoparticles from thereaction between a micellar solution containing NaAOT(sodium bis(2-ethylhexyl)sulfosuccinate) and Co(AOT)2, andNaBH4 dissolved in NaAOT solution.16 Collisions followed byexchange between micellar droplets induce chemical reaction ofcobalt ions and sodium borohydride to generate cobalt nano-particles. The as-synthesized nanoparticles were poorly crystal-line while XRD, after heating at 500 °C, revealed thecharacteristic pattern of fcc cobalt. The average particle sizemeasured using TEM data was 6 nm with a particle sizedistribution s = 9%. A size selection process involving theextraction of nanoparticles from reverse micelles significantlyreduced the polydispersity of the particle size distribution andthese uniform particles self-assembled to generate two-dimen-sionally hexagonally ordered arrays.17

Relatively very little work has been done to synthesizeuniform iron nanoparticles. Suslick et al. reported the synthesisof iron nanoparticles from the sonochemical decomposition ofiron pentacarbonyl in the presence of polyvinylpyrrolidone(PVP) or oleic acid.18 The sonochemical decomposition ofvolatile organometallic compounds has been applied to synthe-size various nanostructured materials.19 Transmission electronmicrographs showed that the iron particles ranged in size from3 to 8 nm. Electron diffraction revealed that the particles as-formed are amorphous, and that after in situ electron beamheating they crystallized to bcc iron. These iron nanoparticleswere readily oxidized to FeO when exposed to air. Gedankenand coworkers reported the synthesis of amorphous cobaltnanoparticles by sonicating Co(CO)3(NO) in decane solution inthe presence of oleic acid.20 Spherical 5–10 nm sized amor-phous nanoparticles were produced. One drawback of thesonochemical process in the synthesis of nanoparticles is itsinability to control particle size.

Very recently, our research group synthesized monodisperseiron nanoparticles from the high temperature (300 °C) aging ofan iron–oleic acid metal complex, which was prepared by thethermal decomposition of iron pentacarbonyl in the presence ofoleic acid at 100 °C.21 We were able to synthesize monodispersenanoparticles with sizes ranging from 4 to 20 nm without usingany size selection process. Initially, the iron oleate complex wasprepared by reacting Fe(CO)5 and oleic acid at 100 °C. Ironnanoparticles were then generated by aging the iron complex at300 °C. A TEM image of the iron nanoparticles revealed that the

nanoparticles were monodisperse and the electron diffractionpattern showed that the nanoparticles were nearly amorphous.The XRD pattern of the sample after being heat-treated in anargon atmosphere at 500 °C revealed a bcc a-iron structure. Theparticle size could be controlled from 4 to 11 nm by usingdifferent molar ratios of iron pentacarbonyl to oleic acid. Fig.3(a) and (b) show TEM images of iron nanoparticles with

particle sizes of 7 and 11 nm, respectively, which were preparedusing 1+2 and 1+3 molar ratios of Fe(CO)5+oleic acid. In orderto produce nanoparticles larger than 11 nm, a reaction mixturewith a 1+4 molar ratio of Fe(CO)5 and oleic acid was usedduring the synthesis. However, the particle size of the resultingnanoparticles was still around 11 nm. We could produce ironnanoparticles with particle sizes bigger than 11 nm by addingmore iron oleate complex to previously prepared 11 nm ironnanoparticles, and then aging at 300 °C. Using this syntheticprocedure, we were able to tune the particle sizes of thenanoparticles from 11 to 20 nm.

Using a similar synthetic procedure, we prepared moderatelymonodisperse cobalt nanoparticles and successfully appliedthem to recyclable catalysts for Pauson–Khand reactions, whichinvolve the cycloaddition of alkynes, alkenes and carbonmonoxide to synthesize cyclopentenones.22 We also usedaqueous colloidal cobalt nanoparticles stabilized by sodiumdodecyl sulfate, which were synthesized by the reverse micellemethod described above, for use as recyclable aqueous-phasePauson–Khand catalysts.23

Nanoparticles of alloys of cobalt and ironSun et al. synthesized monodisperse iron–platinum nano-particles by the simultaneous reduction of platinum acet-ylacetonate (Pt(acac)2) and the thermal decomposition of ironpentacarbonyl (Fe(CO)5) in the presence of oleic acid and oleylamine.24a The composition was adjusted by changing the molarratio of the two precursors. Using a 3+2 molar ratio of Fe(CO)5

to Pt(acac)2, Fe48Pt52 nanoparticles were produced, while using4+1 mixtures Fe70Pt30 nanoparticles were formed. The particlesize could be varied from 3 to 10 nm by adding more precursorsto previously synthesized 3 nm particles, which acted as nuclei.When the solvent was evaporated slowly, three-dimensionalsuperlattices are generated, and by controlling the chain lengthsof the stabilizing surfactants, inter-particle spacing could betuned. For example, 6 nm Fe48Pt52 nanoparticles were arranged

Fig. 3 TEM images of iron nanoparticles: (a) three-dimensional array of 7nm Fe nanoparticles and (b) 11 nm Fe nanoparticles.

CHEM. COMMUN. , 2003, 927–934 929

with an interparticle spacing of 4 nm using oleic acid and oleylamine as stabilizers, while the same sized particles stabilizedwith hexanoic acid and hexyl amine exhibited ~ 1 nm spacing.Fig. 4 shows the TEM images of 6 nm sized Fe50Pt50

nanoparticles stabilized with oleic groups (Fig. 4(a)) andhexanoic groups (Fig. 4(b)), respectively. XRD analysisrevealed that the as-synthesized nanoparticles showed a dis-ordered face centered cubic (fcc) crystal structure and that thiswas transformed to an ordered face centered tetragonal structurewhen heated at 500 °C. Magnetic studies on 4 nm sized Fe52Pt48

nanoparticles showed that the as-synthesized nanoparticleswere superparamagnetic at room temperature and that theybecame ferromagnetic after heating at 500 °C. A write/readexperiment demonstrated that the annealed 120 nm thicksuperlattice of 4 nm Fe48Pt52 nanoparticles supported magneti-zation reversal transitions at moderate linear densities. Dai et al.reported the detailed structural characterization of Fe–Ptnanoparticles during annealing using high-resolution transmis-sion electron microscopy.24b Very recently, Rogach andcoworkers grew micrometer-sized colloidal crystals of Fe–Ptnanoparticles using a three-layer technique based on the slowdiffusion of a non-solvent into a concentrated solution ofnanoparticles.25

Using a similar synthetic procedure, Chen and Niklesreported the synthesis of FePd and CoPt nanoparticles.26 7 nmsized Co48Pt52 nanoparticles were synthesized by the simultane-ous reduction of platinum acetylacetonate and the thermaldecomposition of cobalt tricarbonylnitrosyl (Co(CO)3(NO)) inthe presence of oleic acid and oleyl amine. 11 nm Fe50Pd50

nanoparticles were produced using a similar synthetic methodemploying palladium acetylacetonate and iron pentacarbonyl asprecursors. They also synthesized FexCoyPt1002x2y nano-particles from the simultaneous reduction of cobalt acet-ylacetonate and platinum acetylacetonate and the thermaldecomposition of iron pentacarbonyl in the presence of oleicacid and oleyl amine.27 The average particle diameter was 3.5nm and the particle size distribution was very narrow.

Moreover, these nanoparticles self-assemble to generate super-lattices, and when annealed the nanoparticles were transformedto the tetragonal phase.

Park and Cheon synthesized nanoparticles of Co–Pt alloysand CocorePtshell via transmetalation reactions.28 The injectionof 0.25 mmol of Co2(CO)8 into hot toluene solution containing0.5 mmol of Pt(hfac)2 and oleic acid produced CoPt3 alloynanoparticles with a particle size of 1.8 nm (s = 0.1 nm).Reaction of Pt(hfac)2 with previously prepared 6.33 nm sizedCo nanoparticles produced moderately monodisperse CocorePt-shell nanoparticles with a particle size of 6.27 nm (s = 0.58 nm).The Pt shell thickness and the Co core size were estimated to be1.82 and 4.75 nm, respectively.

Chaudret and coworkers reported the synthesis of 1–2 nmsized bimetallic Co–Pt nanoparticles from the reaction ofCo(h3-C8H13)(h4-C8H12) and Pt2(dba)3 (dba = dibenzylidenea-cetone) under dihydrogen.29 The composition of the nano-particles could be varied by changing the initial molar ratio ofthe two organometallic precursors. XRD and TEM analysisrevealed that the Pt rich particles adopted a fcc crystallinestructure while the cobalt rich particles adopted a poly-tetrahedral arrangement.

Nanoparticles of ferritesTransition metal oxides constitute one of the most fascinatingclasses of inorganic solids, as they exhibit a very wide variety ofstructures, properties, and phenomena.30 Oxide materials havebeen employed in many important advanced technology areasdue to their many interesting physical properties, includingmagnetic, ferroelectric, superconducting, ionic and electricalconducting characteristics. Nanoparticles of magnetic oxides,including most representative ferrites, have been studied formany years for their applications as magnetic storage media andas ferrofluids. However, the synthesis of monodisperse mag-netic oxide nanoparticles was only reported very recently.

The Alivisatos group reported the synthesis of g-Fe2O3

nanoparticles from the thermal decomposition of iron Cupfer-ron complexes (Cup: N-nitrosophenylhydroxylamine,C6H5N(NO)O2).31 Particle sizes were controlled by eithervarying the reaction temperature or by using a varying amountof the complex. Relatively uniform 4–10 nm sized g-Fe2O3

nanoparticles were synthesized. Using different metal Cupfer-ron complexes, nanoparticles of Mn3O4 and Cu2O were alsogenerated.

Very recently, our group reported on the synthesis ofmonodisperse and highly-crystalline maghemite nanoparticles,which did not involve a size selection process.21 The overallsynthetic procedure is depicted in the Fig. 5. Monodisperse ironnanoparticles were first synthesized, and then they weretransformed to monodisperse g-Fe2O3 nanocrystals by con-trolled oxidation using trimethylamine N-oxide ((CH3)3NO), asa mild oxidant. In the current synthetic process, the sizeuniformity was determined during the synthesis of metallicnanoparticles. Transmission electron microscopic images of theparticles showed two- and three-dimensional assembly ofparticles, demonstrating the uniformity of these nanoparticles.Electron diffraction, X-ray diffraction, and high-resolutiontransmission electron microscopy proved the highly crystallinenature of the g-Fe2O3 structures. XPS and Raman spectroscopicresults confirmed the g-Fe2O3 structures. Particle size could bevaried from 4 to 20 nm by altering the experimental parameters.The dominating size controlling factor was the molar ratio ofiron pentacarbonyl to oleic acid. Fig. 6(a) and (b) show the TEMimages of maghemite nanoparticles with sizes of 7 and 11 nmthat were synthesized by using reaction mixtures with [Fe-(CO)5]+[oleic acid] molar ratios of 1+2 and 1+3, respectively.As described in the synthesis of iron nanoparticles, nano-

Fig. 4 TEM images of 6 nm sized Fe50Pt50 nanoparticles stabilized witholeic groups (a) and hexanoic groups (b).24a Reprinted with permission fromreference 24(a). Copyright 2000 American Association for the Advance-ment of Science.

930 CHEM. COMMUN. , 2003, 927–934

particles larger than 11 nm were synthesized by a seeded growthprocess, employing addition of more iron–oleate complex to the11 nm sized particles followed by aging at 300 °C. The resultinglarger iron nanoparticles were then transformed to iron oxidenanocrystals by oxidizing with trimethylamine N-oxide. Wecould also synthesize monodisperse g-Fe2O3 nanocrystals bydirectly injecting Fe(CO)5 into a solution containing both thesurfactant and trimethylamine N-oxide. Fig. 6(c) shows a TEMimage of 13 nm sized g-Fe2O3 nanocrystals synthesized throughthe direct injection method. However, the first method employ-ing the synthesis of monodisperse iron nanoparticles followed

by mild oxidation allowed better control of particle size andreproducibility.

Using a similar procedure, based on the thermal decomposi-tion of a metal–surfactant complex followed by mild oxidation,we synthesized highly crystalline and monodisperse cobaltferrite (CoFe2O4) nanocrystals.32 In this synthesis, uniformiron–cobalt alloy nanoparticles were first generated from thethermal decomposition of a metal–oleate complex, and thenfurther oxidized to yield cobalt ferrite nanocrystals. A singlemolecular precursor, (h5-C5H5)CoFe2(CO)9, was employed inthe synthesis of the mixed-metal–oleate complex. The as-synthesized poorly crystalline metallic nanoparticles were thentransformed into cobalt ferrite nanocrystals by oxidizing themwith the mild oxidant, trimethylamine N-oxide. Cobalt ferritenanoparticles with particle diameters of 6 nm were synthesizedusing starting reaction mixtures with a metal precursor to oleicacid molar ratio of 1+3. The uniformity of the nanocrystals wasfurther demonstrated by the formation of a three-dimensionalclose-packed superlattice assembly. Larger 9 nm sized nano-crystals, as shown in Fig. 7, were produced when lauric acid was

used in combination with oleic acid at a molar ratio of [metalprecursor]+[oleic acid]+[lauric acid] = 1+1+2. Monodispersenanocrystals were obtained without a further size selectionprocess. Nanoscale elemental analysis of ten 9 nm cobalt ferritenanocrystals by energy-dispersive X-ray spectroscopy (EDS)showed that an average molar atomic ratio of Fe to Co was 2.3with a standard deviation of 20%. The EDS data well matchedthe bulk elemental analysis results obtained by inductivelycoupled plasma atomic emission spectrometry (ICP-AES).

Moumen and Pileni reported on a synthesis of cobalt ferritenanoparticles using a microemulsion method,33 which had beenintroduced earlier by Klabunde and coworkers.34 Aqueousmethylamine (CH3NH3OH) was added to sodium dodecylsulfate solution containing cobalt(II) dodecyl sulfate (Co(DS)2)and iron(II) dodecyl sulfate (Fe(DS)2) to generate a precipitate.The size of the nanoparticles was controlled from 2 to 5 nm bychanging the sodium dodecyl sulfate concentration and by

Fig. 5 Synthetic procedure for monodisperse g-Fe2O3 nanoparticles.

Fig. 6 TEM images of g-Fe2O3 nanoparticles: (a) 7 nm, (b) 11 nm, (c) 13nm.21 Reprinted with permission from reference 21. Copyright 2001American Chemical Society.

Fig. 7 TEM images of 9 nm sized cobalt ferrite nanoparticles.32 Reprintedwith permission from reference 32. Copyright 2002 American ChemicalSociety.

CHEM. COMMUN. , 2003, 927–934 931

maintaining constant concentrations of Co(DS)2, Fe(DS)2, andmethylamine. The particles were not well-isolated and aggre-gated into networks, and the particle size was not uniform witha standard deviation of about 30%. Later, detailed structural andmagnetic characterizations of these cobalt ferrite nanoparticleswere reported by Zhang and coworkers.35 From neutrondiffraction studies, they calculated that 78% of tetrahedral sites(the A site in normal spinel structure with the general formula ofAB2O4) were occupied by Fe3+ cations. The 12 nm nano-particles had a blocking temperature of 320 K. The sameresearch group also prepared magnesium ferrite nanoparticlesusing a similar microemulsion method, and the magneticproperties were compared with cobalt ferrite nanoparticles.36

The blocking temperature of cobalt ferrite was found to be atleast 150 K higher than that of the same sized magnesium ferritenanoparticles.

Markovich and coworkers synthesized magnetite (Fe3O4)nanoparticles by reacting aqueous ammonia with an aqueoussolution containing FeCl3 and FeCl2 at a molar ratio of 2+1.37

The aqueous nanoparticles were dispersible in hexane afterbeing coated with oleic acid. After several cycles of size-selective precipitation, uniform nanoparticles were obtained.The Langmuir–Blodgett (LB) technique was used to producetwo-dimensional (2D) arrays of these uniform nanoparticles.

Very recently, Sun and Zeng synthesized monodispersemagnetite nanoparticles from a high temperature reaction ofiron(III) acetylacetonate (Fe(acac)3).38 For example, 4 nm sizedmagnetite nanoparticles were synthesized by refluxing areaction mixture composed of 2 mmol of Fe(acac)3, 20 mL ofdiphenyl ether, 10 mmol of 1,2-hexadecanediol, 6 mmol of oleicacid, and 6 mmol of oleyl amine. A seed-mediated growthmethod was used to synthesize larger nanoparticles. Smallersized nanoparticles ( < 4 nm) were mixed with more precursormaterials and the resulting mixture was refluxed to produce thelarger nanoparticles. By controlling the quantity of the nano-particle seeds, various sized larger nanoparticles were gen-erated. Fig. 8 shows the TEM image of 16 nm Fe3O4

nanoparticles. The as-synthesized magnetite nanoparticles weretransformed to either maghemite (g-Fe2O3) or iron (a-Fe) byannealing in an oxygen or in an argon atmosphere, re-spectively.

Shape control of nanoparticlesRecently, the shape control of nanoparticles to generateanisotropic nanoparticles was recognized as a very importantissue. The Alivisatos group reported the synthesis of rod-shapedCdSe nanocrystals from the high temperature reaction betweendimethylcadmium and selenium in the presence of a mixedsurfactant solution composed of trioctylphosphine oxide andhexylphosphonic acid.39 Later the same group reported on thesynthesis of rod-, arrow-, teardrop- and tetrapod-shaped CdSe

nanocrystals by controlling experimental conditions such as theratio of surfactants, the injection volume, and the time-dependent monomer concentration.40 They explained the for-mation of anisotropic rod-shaped nanoparticles based on therelative differences between the growth rates of different crystalfaces at high monomer concentration. Cheon and coworkersdescribed the synthesis of CdS nanorods and of relatednanostructures in pure hexadecylamine using a single-sourcemolecular precursor, Cd(S2CNEt2)2.41 The Peng group con-ducted detailed mechanistic studies on the shape-controlledsynthesis of CdSe nanoparticles, and explained the growth ofanisotropic CdSe nanocrystals on the basis of diffusioncontrol.42 Wang and coworkers synthesized gold nanorodsusing an electrochemical method.43 In this synthesis, theelectrochemical cell consisted of a gold anode and platinumcathode and the electrolyte solution was a solution of hex-adecyltrimethylammonium bromide and tetradecylammoniumbromide. The aspect ratios of the nanorods were controlled bysurfactant molar ratios.

Anisotropic magnetic nanoparticles are expected to offermany advantages because shape anisotropy would exert atremendous influence on their magnetic properties. The ten-dency for the magnetization to align in a particular direction ina specimen is denoted magnetic anisotropy. Magnetic aniso-tropy generally originates from crystal symmetry, shape, stress,or directed atomic pair ordering.3d,5,44 Of these, crystallineanisotropy and shape anisotropy are common forms of aniso-tropy in magnetic materials. Symmetrically shaped nano-particles, such as spheres, do not have any net shape anisotropy.However, rod-shaped nanoparticles have shape anisotropy inaddition to crystalline anisotropy, which will increase thecoercivity. Submicrometer-sized acicular (rod-shaped) mag-netic particles have been used in commercial particulatemagnetic recording media.

Relatively few anisotropic magnetic nanoparticles have beenreported until 2000. Particle dimensions of the anisotropicmagnetic nanoparticles are too big to exhibit any nanosize effectand these materials were often extensively agglomerated.45

Gibson and Putzer reported on the synthesis of disk-shapedcobalt nanoparticles of width 100 nm and thickness 15 nm bysonicating an aqueous solution of Co2+ and hydrazine.45 Thenanoparticles obtained were single-domain particles. Theparticles were not uniform and agglomerated.

Our group reported the synthesis of nearly uniform rod-shaped iron nanoparticles from the controlled growth ofmonodisperse spherical nanoparticles.46 In the first step of thesynthesis, monodisperse 2 nm spherical iron nanoparticles wereprepared by the thermal decomposition of an organometallicprecursor (Fe(CO)5) in the presence of a stabilizing surfactant,trioctylphosphine oxide (TOPO). An aliquot of Fe(CO)5 intrioctylphosphine (TOP) was added to the resulting sphericalnanoparticles in TOPO, at 320 °C, and the resulting blacksolution was aged for 30 min at 320 °C. The nanoparticles soproduced were retrieved by precipitation. The precipitate wasdissolved in 19 mL of pyridine containing 0.5 g of didodecyldi-methylammonium bromide (DDAB), and the resulting solutionwas refluxed for 12 h. The precipitate formed during the refluxwas removed by centrifugation and the supernatant wasvacuum-dried to yield a black powder. A TEM image of thesample (Fig. 9) revealed nearly monodisperse rod-shapedparticles of dimensions 2 nm (width) 3 11 nm (length) with astandard deviation of 5.7% in the length. The electrondiffraction pattern showed the bcc structure of a-Fe. When theDDAB concentration in pyridine was increased, rod-shapedparticles with higher aspect ratios were obtained. The width ofthese nanorods (2 nm) was almost unchanged, but the lengthincreased to 22 nm (standard deviation of 5.9%) and 27 nm(standard deviation of 5.5%). The transformation of nano-

Fig. 8 TEM images of three-dimensional array of 16 nm Fe3O4

nanoparticles.38 Reprinted with permission from reference 38. Copyright2002 American Chemical Society.

932 CHEM. COMMUN. , 2003, 927–934

spheres to nanorods seems to be due to the so-called orientedattachment mechanism, which is initiated by the relativelystrong binding of DDAB surfactant on the central region of thegrowing nanoparticles. After the fusion of two nanospheres, athird nanosphere will bind on the edge (instead of the centralregion where the DDAB is strongly bound), thus generating acatenated structure. The continued attachment of nanospheresonto the ends of the growing nanoparticles generates theunidirectional nanorods. In some parts of TEM image, catenatednanospheres, which seem to be of the intermediate structure,were observed. Magnetic studies revealed that the blockingtemperatures of 2 nm 3 11 nm rod-shaped nanoparticles and 2nm spherical nanoparticles were 110 and 12 K, respectively.These results are consistent with classical micromagnetictheory, which predicts that the anisotropy energy is proportionalto the volume of a single particle and the anisotropy constant.The magnetic anisotropy constant (K) was deduced from theblocking temperature using the equation K = 25kbTB/V, wherekb is the Boltzman constant and V is the volume of singlenanoparticle.47 The magnetic anisotropy constant of 2 nmnanospheres was calculated to be 9.1 3 106 erg cm23, and thatof the rod-shaped particles as 1.6 3 107 erg cm23. By treatingthe rod-shaped particles as prolate spheroids, the shapeanisotropy constant of 2 nm 3 11 nm rod-shaped nanoparticleswas calculated to be 7.9 3106 erg cm23. When this shapeanisotropy constant was added to the magnetocrystallineanisotropy constant of the spheres, the overall value was foundto agree well with the experimentally determined anisotropyconstant of the rod-shaped particles.48 The longer iron nanorodswith lengths of 22 nm, 27 nm, and 37 nm were found to bereadily oxidized to FeO.

The Alivisatos group extended the synthetic proceduredeveloped for CdSe nanorods to synthesize cobalt nanorods,49a

which are now confirmed to be nanodisks as will be discussedbelow.49b In this synthesis, dicobalt octacarbonyl, Co2(CO)8, ino-dichlorobenzene was injected into hot (185 °C) o-di-chlorobenzene solution containing oleic acid and trioctylphos-phine oxide (TOPO). Quenching of the reaction mixture shortlyafter this injection then generated relatively uniform sizedcobalt nanorods (nanodisks) with a hcp crystal structure. Uponfurther thermal aging of several minutes, the high-energy hcprod (disk)-shaped particles were transformed to monodispersespherical e-Co nanocrystals. The change of crystal structure atthe relatively low temperature of < 200 °C can be explained bythe facile diffusion and phase transitions for nanometer sizedparticles. The transformation to spherical particles was ex-

plained by the high surface tension that reduces the surface-to-volume ratio. The size of nanoparticles could be controlled bythe ratio of surfactant to precursor, which is consistent withobservations in the synthesis of CdSe and other nanoparticles.The use of two different surfactants seemed to be important forthe synthesis of disk-shaped nanoparticles by modulating therelative growth rates of different faces. At a fixed oleic acidconcentration, the length of the nanorods (nanodisks) was foundto be proportional to the concentration of TOPO. Highresolution transmission electron microscopic images showedthat the long direction of the nanorods (nanodisks) is parallel tothe (101) planes, demonstrating that TOPO selectively stabi-lizes the (101) face of Co and decreases its relative growth rate.The size uniformity was demonstrated by the spontaneous self-assembly of the nanocrystals and by the formation of super-structures such as ribbons of nanorods (nanodisks). A TEMimage of self-assembled 4 nm by 25 nm hcp Co nanorods(nanodisks) is shown in Fig. 10.

Very recently, the same research group reported the synthesisof hcp-Co nanodisks, which were thought to be nanorods asdescribed above.49b The disk-shape of the nanocrystals wasverified by tilting experiments in TEM. The same workersdeveloped a much easier procedure to synthesize cobaltnanodisks, employing a linear amine instead of TOPO. Thermaldecomposition of dicobalt octacarbonyl in the presence of oleicacid and a linear amine followed by aging generated disk-shaped nanoparticles along with spherical nanoparticles. A purefraction of hcp Co nanodisks could be obtained after magneticseparation. The length and diameter can be controlled, primarilyby variation of the reaction time following nucleation as well asby varying the precursor to surfactant ratios. The average size ofthe smaller disks was about 2 3 4 nm and about 4 3 90 nm forthe largest.

Chaudret and coworkers reported on the synthesis of nickelnanorods from the reduction of Ni(COD)2 (COD = cycloocta-1,5-diene) in the presence of hexadecylamine.50 They reportedthat increasing concentrations of amine induced the formationof nanorods. Magnetic studies revealed that these nanorodsshowed a saturation magnetization close to the bulk value,demonstrating the absence of the influence of amine coordina-tion on magnetic properties of the nanorods. Gedanken andcoworkers reported the fabrication of the magnetite nanorods bythe sonication of aqueous iron(II) acetate in the presence of b-cyclodextrin.51

Conclusion and outlookIn the past few years, considerable progress has been made inthe synthesis of monodisperse and well-defined structuredmagnetic nanoparticles with sizes ranging from 2 to 20 nm. Theoutlook of such magnetic nanoparticles is very promisingbecause these materials will find many important industrialapplications including ultrahigh-density magnetic storage

Fig. 9 TEM images of rod-shaped 2 nm (width) 3 11 nm (length) ironnanoparticles.46 Reprinted with permission from reference 46. Copyright2000 American Chemical Society.

Fig. 10 TEM images of 4 nm (width) 3 25 nm (length) disk-shaped cobaltnanoparticles.49a Reprinted with permission from reference 49(a). Copy-right 2001 American Association for the Advancement of Science.

CHEM. COMMUN. , 2003, 927–934 933

media and drug-delivery systems. From a synthetic point ofview, there are several interesting research areas worthpursuing. First, we need a generalized synthetic procedure,which can synthesize various monodisperse nanoparticlesdirectly without any further size-selection process. Second,magnetic nanoparticles with various chemical compositions,such as CrO2 and lanthanide-containing materials, should beprepared. Third, more extensive shape-control of magneticnanoparticles should be investigated. Forth, a large-areaalignment of magnetic nanoparticles is critical in magneticstorage media applications. To improve our understanding ofsuch systems and their applications, a collaborative multi-disciplined effort is anticipated.

AcknowledgementsThe work was supported by the National Creative ResearchInitiative Program of the Korean Ministry of Science andTechnology. I thank my students, Dr Sang-Wook Kim,Jongnam Park, Sang-Jae Park and Yunhee Chung for theirimportant contributions to the work discussed herein. I alsothank Dr Shouheng Sun and Dr Chris Murray at IBM T. J.Watson Research Center for very helpful discussion on Fe–Ptalloy nanoparticles.

Notes and references1 (a) A. P. Alivisatos, Science, 1996, 271, 933; (b) H. Weller, Angew.

Chem., Int. Ed. Engl., 1993, 32, 41; (c) Clusters and Colloids, ed. G.Schmid, VCH Press, New York, 1994; (d) Nanoscale Materials inChemistry, ed. K. J. Klabunde, Wiley-Interscience, New York, 2001; (e)Nanoparticles and Nanostructured Films, ed. J. H. Fendler, Wiley-VCH, Weinheim, 1998.

2 (a) S. A. Majetich and Y. Jin, Science, 1999, 284, 470; (b) C. B. Murray,C. R. Kagan and M. G. Bawendi, Science, 1995, 270, 1335; (c) A. J.Zarur and J. Y. Ying, Nature, 2000, 403, 65.

3 (a) D. D. Awschalom and D. P. DiVicenzo, Phys. Today, 1995, 4, 43; (b)K. Raj and R. Moskowitz, J. Magn. Magn. Mater., 1990, 85, 233; (c) I.M. L. Billas, A. Chatelain and W. A. de Heer, Science, 1994, 265, 1682;(d) R. C. O’Handley, Modern Magnetic Materials, Wiley, New York,1999.

4 (a) V. E. Fertman, Magnetic Fluids Guidebook: Properties andApplications, Hemisphere Publishing Co., New York, 1990; (b) B. M.Berkovsky, V. F. Medvedev and M. S. Krakov, Magnetic Fluids:Engineering Applications, Oxford University Press, Oxford, 1993; (c)R. F. Ziolo, E. P. Giannelis, B. A. Weinstein, M. P. O’Horo, B. N.Ganguly, V. Mehrotra, M. W. Russel and D. R. Huffman, Science, 1992,257, 219.

5 D. L. Leslie-Pelecky and R. D. Rieke, Chem. Mater., 1996, 8, 1770.6 C. B. Murray, D. J. Norris and M. G. Bawendi, J. Am. Chem. Soc., 1993,

115, 8706.7 (a) X. M. Lin, H. M. Jaeger, C. M. Sorensen and K. J. Klabunde, J. Phys.

Chem. B, 2001, 105, 3353; (b) S. Stoeva, K. J. Klabunde, C. M.Sorensen and I. Dragieva, J. Am. Chem. Soc., 2002, 124, 2305.

8 (a) C. B. Murray, S. Sun, W. Gaschler, H. Doyle, T. A. Betley and C. R.Kagan, IBM J. Res. & Dev., 2001, 45, 47; (b) C. B. Murray, S. Sun, H.Doyle and T. A. Betley, MRS Bull., 2001, 985.

9 (a) S. Sun and C. B. Murray, J. Appl. Phys., 1999, 85, 4325; (b) M. R.Diehl, J.-Y. Yu, J. R. Heath, G. A. Held, H. Doyle, S. Sun and C. B.Murray, J. Phys. Chem. B, 2001, 105, 7913; (c) Z. L. Wang, Z. Dai andS. Sun, Adv. Mater., 2000, 12, 1944.

10 D. P. Dinega and M. G. Bawendi, Angew. Chem., Int. Ed., 1999, 38,1788.

11 V. F. Puntes, K. M. Krishnan and A. P. Alivisatos, Appl. Phys. Lett.,2001, 78, 2187.

12 C. T. Black, C. B. Murray, R. L. Sandstrom and S. Sun, Science, 2000,290, 1131.

13 C. Petit, T. Cren, D. Roditchev, W. Sacks, J. Klein and M. P. Pileni, Adv.Mater., 1999, 11, 1198.

14 M. P. Pileni, Langmuir, 1997, 13, 3266.15 J. P. Chen, C. M. Sorensen, K. J. Klabunde and G. C. Hadjipanayis, J.

Appl. Phys., 1994, 76, 6316.16 C. Petit, A. Taleb and M. P. Pileni, J. Phys. Chem. B, 1999, 103,

1805.17 C. Petit, A. Taleb and M. P. Pileni, Adv. Mater., 1998, 10, 259.18 K. S. Suslick, M. Fang and T. Hyeon, J. Am. Chem. Soc., 1996, 118,

11960.19 K. S. Suslick, T. Hyeon and M. Fang, Chem. Mater., 1996, 8, 2172.20 K. V. P. M. Shafi, A. Gedanken and R. Prozorov, Adv. Mater., 1998, 10,

590.21 T. Hyeon, S. S. Lee, J. Park, Y. Chung and H. B. Na, J. Am. Chem. Soc.,

2001, 123, 12798.22 S.-W. Kim, S. U. Son, S. S. Lee, T. Hyeon and Y. K. Chung, Chem.

Commun., 2001, 2212.23 S. U. Son, S. I. Lee, Y. K. Chung, S.-W. Kim and T. Hyeon, Org. Lett.,

2002, 4, 277.24 (a) S. Sun, C. B. Murray, D. Weller, L. Folks and A. Moser, Science,

2000, 287, 1989; (b) Z. R. Dai., S. Sun and Z. L. Wang, Surf. Sci., 2002,505, 325.

25 E. Shevchenko, D. Talapin, A. Kornowski, F. Wiehorst, J. Koetzler, M.Haase, A. Rogach and H. Weller, Adv. Mater., 2002, 14, 287.

26 M. Chen and D. E. Nikles, J. Appl. Phys., 2002, 91, 8477.27 M. Chen and D. E. Nikles, NanoLett., 2002, 2, 211.28 J.-I. Park and J. Cheon, J. Am. Chem. Soc., 2001, 123, 5743.29 T. O. Ely, C. Pan, C. Amiens, B. Chaudret, F. Dassenoy, P. Lecante, M.-

J. Casanove, A. Mosset, M. Respaud and J.-M. Broto, J. Phys. Chem. B,2000, 104, 695.

30 (a) C. N. R. Rao and B. Raveau, Transition Metal Oxides, Wiley-VCH,New York, 2nd edn., 1998; (b) R. M. Cornell and U. Schwertmann, Theiron oxides, VCH, Weinheim, 1996.

31 J. Rockenberger, E. C. Scher and A. P. Alivisatos, J. Am. Chem. Soc.,1999, 121, 11595.

32 T. Hyeon, Y. Chung, J. Park, S. S. Lee, Y.-W. Kim and B. H. Park, J.Phys. Chem. B, 2002, 106, 6831.

33 (a) N. Moumen and M. P. Pileni, Chem. Mater., 1996, 8, 1128; (b) N.Moumen and M. P. Pileni, J. Phys. Chem., 1996, 100, 1867.

34 J. P. Chen, K. M. Lee, C. M. Sorensen, K. J. Klabunde and G. C.Hadjipanayis, J. Appl. Phys., 1994, 75, 5876.

35 C. Liu, B. Zou, A. J. Rondinone and Z. J. Zhang, J. Am. Chem. Soc.,2000, 122, 6263.

36 A. J. Rondinone, A. C. S. Samia and Z. J. Zhang, J. Phys. Chem., 1999,103, 6876.

37 T. Fried, G. Shemer and G. Markovich, Adv. Mater., 2001, 13, 1158.38 S. Sun and H. Zeng, J. Am. Chem. Soc., 2002, 124, 8204.39 X. Peng, L. Manna, W. Yang, J. Wickham, E. Scher, A. Kadavanich and

A. P. Alivisatos, Nature, 2000, 404, 59.40 L. Manna, E. Scher and A. P. Alivisatos, J. Am. Chem. Soc., 2000, 122,

12700.41 Y.-W. Jun, S.-M. Lee, N.-J. Kang and J. Cheon, J. Am. Chem. Soc.,

2001, 123, 5150.42 (a) Z. A. Peng and X. Peng, J. Am. Chem. Soc., 2001, 123, 1389; (b) Z.

A. Peng and X. Peng, J. Am. Chem. Soc., 2002, 124, 3343.43 (a) Y. Y. Yu, S. Chang, C. J. Lee and C. R. C. Wang, J. Phys. Chem. B,

1997, 101, 6661; (b) S.-S. Chang, C.-W. Shih, C.-D. Chen, W.-C. Laiand C. R. C. Wang, Langmuir, 1999, 15, 701; (c) S. Link, M. B.Mohamed and M. A. El-Sayed, J. Phys. Chem. B, 1999, 103, 3073.

44 For the general introduction on the magnetic properties of anisotropicmagnetic nanoparticles, please refer to ch. 6 (Magnetism) of ref. 1c.

45 C. P. Gibson and K. J. Putzer, Science, 1995, 267, 1338.46 S.-J. Park, S. Kim, S. Lee, Z. G. Khim, K. Char and T. Hyeon, J. Am.

Chem. Soc., 2000, 122, 8581.47 Q. Chen and Z. J. Zhang, Appl. Phys. Lett., 1998, 73, 3156.48 B. D. Cullity, Introduction to Magnetic Materials, Addison-Wesley,

Reading, MA, 1972.49 (a) V. F. Puntes, K. M. Krishnan and A. P. Alivisatos, Science, 2001,

291, 2115; (b) V. F. Puntes, D. Zanchet, C. K. Erdonmez and A. P.Alivisatos, J. Am. Chem. Soc., 2002, 124, 12874.

50 N. Cordente, M. Respaud, F. Senocq, J.-J. Casanove, C. Amiens and B.Chaudret, NanoLett., 2001, 1, 565.

51 R. V. Kumar, Y. Koltypin, X. N. Xu, Y. Yeshurun, A. Gedanken and I.Felner, J. Appl. Phys., 2001, 89, 6324.

934 CHEM. COMMUN. , 2003, 927–934

Shape-Controlled Synthesis and Shape-Induced Texture of MnFe 2O4Nanoparticles

Hao Zeng,† Philip M. Rice,‡ Shan X. Wang,§ and Shouheng Sun*,†

IBM T. J. Watson Research Center, Yorktown Heights, New York 10598, IBM Almaden Research Center,650 Harry Road, San Jose, California 95120, and Department of Materials Science and Engineering,

Stanford UniVersity, Stanford, California 94305

Received July 8, 2004; E-mail: [email protected]

The shape is one of the most important factors in determiningthe structural, physical, and chemical properties of a nanoparticleand an assembled array of the particles.1 Shape-controlled synthesisof nanoparticles has become a recent focus, as different shapes ofthe particles can introduce electronic, optical, and magneticproperties that are different from those observed in their sphericalcounterparts. Semiconductor nanoparticles have been made invarious nonspherical shapes of triangle, rod, cube, arrow, andtetrapod,2 which are interesting for quantum confinement studiesand important for potential sensor, transistor, and solar cellapplications. As the restoring force on the conduction electrons isextraordinarily sensitive to particle curvature, nonspherical metallicAu or Ag nanoparticles, produced via various approaches,3 havebeen found to shift their surface plasmon frequency drastically,4

making them useful as multicolor diagnostic labels and for otheroptical devices. Controlling the shape of a magnetic nanoparticleis even more appealing as the shape may lead to anisotropicmagnetic properties of the nanoparticles.5 Although recent progressin shape-controlled synthesis of magnetic nanoparticles has led tonanorods,6 nanocubes,7 and other variously shaped Fe2O3 nano-particles,8 there is still no example of using nanoparticles withdifferent shapes as building blocks to control the crystal orientationof the nanoparticles in an assembly. As the magnetic easy axis ofa particle correlates closely with its crystal structure,9 the shape-induced crystal orientation of each nanoparticle in an assemblywould lead to an aligned magnetic easy axis. Such alignment isoften a necessity for a nanoparticle assembly to be suitable forvarious magnetic applications in such as data storage and advancedmagnets.10

Here we report shape-controlled synthesis of MnFe2O4 nano-particles and shape-induced texture of the particles in a self-assembled superlattice. We recently reported that reaction ofFe(acac)3 and M(acac)2 with 1,2-hexadecanediol, oleic acid, andoleylamine in a high-boiling ether solvent could lead to mono-disperse MFe2O4 nanoparticles (M) Fe, Co, Mn, Mg, etc.). Thecomposition of the particles was controlled by the initial molar ratioof Fe(acac)3 and M(acac)2, and the size was tuned by varying thereaction conditions or via seed-meditated growth.11 Our furtherexperiments indicated that the size of the MFe2O4 particles (M)Mn in this work) could be tuned more conveniently through a one-pot reaction by changing the concentration of the reactants. Moreimportantly, this one-pot reaction could also lead to the particleswith controlled shapes. Slow evaporation of the particle dispersionyielded nanoparticle superlattices. The nanoparticles within thesuperlattice exhibited preferred alignment of crystal orientationscorrelated with their shape. With this texture control, the shaped

nanoparticles could become useful building blocks for the construc-tion of functional nanostructures.

The size of the MnFe2O4 particles is tuned by varying theconcentration of the precursors, while the shape is controlled bythe amount of stabilizers added to the reaction mixture. For areaction containing 2 mmol of Fe(acac)3, 1 mmol of Mn(acac)2,10 mmol of 1,2-hexadecandiol, 6 mmol of oleic acid, 6 mmol ofoleylamine, 20 mL of benzyl ether gave 12 nm MnFe2O4, while22 mL or 25 mL of benzyl ether yielded 10- or 8 nm particles,respectively. When the surfactant/Fe(acac)3 ratio is smaller than3:1, the particles are nearly spherical with no well-defined facets.12

Increasing the ratio to 3:1 yields cubelike particles.13 If the particleswere prepared using the seed-mediated growth as reported before,11

polyhedron-shaped particles were obtained. Figure 1 shows therepresentative TEM images of the MnFe2O4 nanoparticles withdifferent morphologies, with 1A being cubelike and 1B polyhedron-shaped particles. The HRTEM image of a part of a single cubelikeMnFe2O4 nanoparticle (Figure 1C) shows lattice fringes with theinterfringe distance measured to be 0.214 nm, and that of apolyhedron-shaped particle (Figure 1D) shows the interfringedistance to be 0.257 nm, close to the lattice spacing of the{400}planes at 0.213 nm and{311} planes at 0.256 nm in the cubicspinel-structured MnFe2O4. HRTEM further reveals that the{400}lattice fringes are parallel to the edges of the cube, while the{220}fringes (spacing 0.301 nm) run face-diagonally in the cube,12

indicating that the cubelike particles are terminated with{100}planes. The TEM tilted angle analyses of the polyhedron-shapedparticles indicated that the hexagon shape shown in Figure 1B likely

† IBM T. J. Watson Research Center.‡ IBM Almaden Research Center.§ Stanford University.

Figure 1. TEM images of the as synthesized (A) 12 nm cubelike and (B)12 nm polyhedron-shaped MnFe2O4 nanoparticles. HRTEM images of apart of a single (C) cubelike and (D) polyhedron-shaped MnFe2O4

nanoparticles.

Published on Web 08/28/2004

11458 9 J. AM. CHEM. SOC. 2004 , 126, 11458-11459 10.1021/ja045911d CCC: $27.50 © 2004 American Chemical Society

comes from a truncated dodecahedron projected along the⟨110⟩direction, an observation that is similar to that of polyhedron-shapedFe2O3 nanoparticles reported previously.8

Controlled evaporation of the carrier solvent from hexanedispersion (∼2 mg/mL) of the particles led to MnFe2O4 nanoparticlesuperlattices. Figure 2A shows the superlattice assembly from thecubelike particles, while Figure 2B is the assembly from thepolyhedron-shaped particles. The fast Fourier transformation (FFT)of these two images reveals four-fold symmetry,12 indicating thatboth assemblies have the cubic packing. Scanning electron micros-copy (SEM) images of the self-assembled nanoparticle superlatticeson Si (100) substrate show large-area superlattice patterns12 similarto those shown in Figure 2A,B. Although the assembly structuresare the same for both cubelike and polyhedron-shaped particles,the different shapes possessed by each group of the particles doaffect the crystalline orientation of individual particles within theself-assembled superlattices. XRD of the self-assembled cubelikeparticles on Si (100) substrate shows the intensified (400) peak(Figure 2C), and that of polyhedron-shaped particles reveals twostrong reflections from (220) (Figure 2D) and (440) planes (notshown here).12 These are markedly different from that of a 3-Drandomly oriented spinel-structured MnFe2O4 nanoparticle as-sembly, which shows a strong (311) peak.12 The fact that the XRDof the cubelike particle assembly shows a strong (400) peakindicates that each of the cubelike particles in the cubic assemblyhas preferred crystal orientation with{100} planes parallel to theSi substrate. Selected area electron diffraction (SAED) on theassembly shown in Figure 2A reveals a single-crystal-like pattern,12

indicating strong preferential alignment. The strongest reflectionsfrom {400} and {440} show four-fold modulation of the ringintensity, which is due to the coherent alignment of the individualparticle’s spinel lattice within the cubic superlattice. This meansthat the incident beam is parallel to the⟨100⟩ orientation, furtherconfirming the XRD observation shown in Figure 2C. Similarly,for the polyhedron-shaped particle assembly, both XRD in Figure2D and SAED12 indicate that the{110} planes are parallel to thesubstrate. These demonstrate clearly that particle shape can indeedinduce texture in a self-assembled nanoparticle superlattice.

The present study on shape-controlled synthesis and shape-induced texture of MnFe2O4 nanoparticles offers a practical example

that chemical process can be used to regulate the growth of a particleand control the crystal orientation of a particle in a superlatticeassembly. Such capability of control will be significant in futurefabrication of high-performance magnetic nanodevices. A robustmagnetic nanoparticle superlattice array with an aligned magneticeasy axis and controlled magnetic interactions between the neigh-boring particles may revolutionize the information storage technol-ogy with unprecedented sensitivity and density.

Acknowledgment. The work was supported in part by DARPA/ONR under Grant No. N00014-01-1-0885.

Supporting Information Available: A typical experimental pro-cedure for making 12 nm cubelike MnFe2O4 nanoparticles, and furthercharacterization data for MnFe2O4 nanoparticles and their assemblies.This material is available free of charge via the Internet at http://pubs.acs.org.

References

(1) (a) Wang, Z. L.AdV. Mater. 1998,10, 13-30. (b) Wang, Z. L.J. Phys.Chem. B2000,104, 1153-1175. (c) Wang, Z. L.; Dai, Z.; Sun, S.AdV.Mater. 2000, 12, 1944-1946. (d) Jana, N. R.Angew. Chem., Int. Ed.2004, 43, 1536-1540. (e) Mohraz, A.; Moler, D. B.; Ziff, R. M.; Solomon,M. J. Phys. ReV. Lett.2004,92, 155503-(1-4). (f) Narayanan, R.; El-Sayed, M. A.Nano Lett.2004,4, 1343-1348.

(2) (a) Peng, X. G.; Manna, L.; Yang, W.; Wickham, J.; Scher, E.; Kadavanich,A.; Alivisatos, A. P.Nature2000,404, 59-61. (b) Manna, L.; Scher, E.C.; Alivisatos, A. P.J. Am. Chem. Soc. 2000, 122, 12700-12706. (c)Chen, C.-C.; Chao, C.-Y.; Lang, Z.-H.Chem. Mater.2000,12, 1516-1518. (d) Li, L.; Hu, J.; Yang, W.; Alivisatos, A. P.Nano Lett.2001,1,349-351. (e) Pinna, N.; Weiss, K.; Urban, J.; Pileni, M.-P.AdV. Mater.2001,13, 261-264. (f) Jun, Y.-W.; Lee, S.-M.; Kang, N.-J.; Cheon, J. J.Am. Chem. Soc.2001, 123, 5150-5151. (g) Jun, Y.-W.; Jung, Y.-Y.;Cheon, J.J. Am. Chem. Soc.2002,124, 615-619. (h) Huynh, W. U.;Dittmer, J. J.; Alivisatos, A. P.Science2002,295, 2425-2427. (i) Watt,A. A. R.; Meredith, P.; Riches, J. D.; Atkinson, S.; Rubinsztein-Dunlop,H. Curr. Appl. Phys.2004,4, 320-322.

(3) (a) Nikoobakht, B.; Wang, Z. L.; El-Sayed, M. A.J. Phys. Chem. B2000,104, 8635-8640. (b) Jin, R.; Cao, Y.; Mirkin, C. A.; Kelly, K. L.; Schatz,G. C.; Zheng, J. G.Science, 2001, 294, 1901-1903. (c) Jana, N. R.;Gearheart, L.; Murphy, C. J.J. Phys. Chem. B2001,105, 4065-4067.(d) Kim, F.; Song, J. H.; Yang, P.J. Am. Chem. Soc.2002,124, 14316-14317. (e) Sun, Y.; Xia, Y.Science2002,298, 2176-2179. (f) Jin, R.;Cao, Y. C.; Hao, E.; Me˘ traux, G. S.; Schatz, G. C.; Mirkin, C. A.Nature2003,425, 487-490. (g) Chen, S.; Wang, Z. L.; Ballto, J.; Foulger, S.H.; Carroll, D. L. J. Am. Chem. Soc.2003, 125, 16186-16187.

(4) (a) van der Zande, B. M. I.; Bohmer, M. R.; Fokkink, L. G. J.;Schonenberger, C.J. Phys. Chem. B1997,101, 852-854. (b) Yu, Y.-Y.;Chang, S.-S.; Lee, C.-L.; Wang, C. R. C.J. Phys. Chem. B1997; 101,6661-6664. (c) Sonnichsen, C.; Franzl, T.; Wilk, T.; von Plessen, G.;Feldmann, J.; Wilson, O.; Mulvaney, P.Phys. ReV. Lett. 2002, 88, 077402-(1-4).

(5) (a) Thurn-Albrecht, T.; Schotter, J.; Kastle, C. A.; Emley, N.; Shibauchi,T.; Krusin-Elbaum, L. Guarini, K.; Black, C. T.; Tuominen, M. T.; Russell,T. P.Science2000,290, 2126-2129. (b) Zeng, H.; Zheng, M.; Skomski,R.; Sellmyer, D. J.; Liu, Y.; Menon, L.; Bandyopadhyay, S.J. Appl. Phys.2000,87, 4718-4720.

(6) (a) Park, S.-J.; Kim, S.; Lee, S.; Khim, Z. G.; Xhar, K.; Hyeon, T.J. Am.Chem. Soc.2000,122, 8581-8582. (b) Puntes, V. F.; Krishnan, K. M.;Alivisatos, A. P.Science2001,291, 2115-2117.

(7) (a) Dumestre, F.; Chaudret, B.; Amiens, C.; Renaud, P.; Fejes, P.Science2004,303, 821-823. (b) Song, Q.; Zhang, Z. J.J. Am. Chem. Soc.2004,126, 6164-6168.

(8) Cheon, J.; Kang, N.-J.; Lee, S.-M.; Lee, J.-H.; Yoon, J.-H.; Oh, S. J.J.Am. Chem. Soc.2004,126, 1950-1951.

(9) Cullity, B. D. Introduction to Magnetic Materials; Addison-Wesley:Reading, Massachusetts, 1972; pp 207-247.

(10) (a) Sun, S.; Murray, C. B.; Weller, D.; Folks, L.; Moser, A.Science2000,287, 1989-1992. (b) Zeng, H.; Li, J.; Wang, Z. L.; Liu, J. P.; Sun, S.Nature2002,420, 395-398.

(11) (a) Sun, S.; Zeng, H.J. Am. Chem. Soc., 2002124, 8204-8205. (b) Sun,S.; Zeng, H.; Robinson, D. B.; Raoux, S.; Rice, P. M.; Wang, S. X.; Li,G. J. Am. Chem. Soc.2004126, 273-279.

(12) See the Supporting Information.(13) The growth conditions seem to be consistent with those from several recent

observations that more surfactant facilitates thermodynamical growth ofthe particles. See, for example, refs 3e, 7b, and 8.

JA045911D

Figure 2. TEM images of 12 nm MnFe2O4 nanoparticle superlattices of(A) cubelike and (B) polyhedron-shaped nanoparticles. XRD (Co KR λ )1.788965 Å) of (C) cubelike and (D) polyhedron-shaped nanoparticlesuperlattice on Si(100) substrates.

C O M M U N I C A T I O N S

J. AM. CHEM. SOC. 9 VOL. 126, NO. 37, 2004 11459

Nanotechnologies for the Life Sciences Vol. 8Nanomaterials for Biosensors. Edited by Challa S. S. R. KumarCopyright 8 2007 WILEY-VCH Verlag GmbH & Co. KGaA, WeinheimISBN: 978-3-527-31388-4

1

Biosensing using Carbon Nanotube Field-effect

Transistors

Padmakar D. Kichambare and Alexander Star

1.1

Overview

This chapter covers recent advances in biodetection using single-walled carbon

nanotube field-effect transistors (NTFETs). In particular, we describe fabrication of

NTFET devices and their application for electronic detection of biomolecules. A

typical NTFET fabrication process consists of combination of chemical vapor depo-

sition (CVD) and complementary metal oxide semiconductor (CMOS) processes.

The NTFET devices have electronic properties comparable to traditional metal

oxide semiconductor field-effect transistors (MOSFETs) and readily respond to

changes in the chemical environment, enabling a direct and reliable pathway for

detection of biomolecules with extreme sensitivity and selectivity. We address the

challenges in effective integration of carbon nanotubes into conventional electron-

ics for biosensor applications. We also discuss in detail recent applications of

NTFETs for label-free electronic detection of antibody–antigen interactions, DNA

hybridization, and enzymatic reactions.

1.2

Introduction

The interplay between nanomaterials and biological systems forms an emerging

research field of broad importance. In particular, novel biosensors based on

nanomaterials have received considerable attention [1–4]. Integration of one-

dimensional (1D) nanomaterials, such as nanowires, into electric devices offers

substantial advantages for the detection of biological species and has significant

advantages over the conventional optical biodetection methods [5]. The first advan-

tage is related to size compatibility: Electronic circuits in which the component

parts are comparable in size to biological entities ensure appropriate size compati-

bility between the detector and the detected biological species. The second advan-

tage to developing nanomaterial based electronic detection is that most biological

processes involve electrostatic interactions and charge transfer, which are directly

1

detected by electronic nanocircuits. Nanowire-based electronic devices, therefore,

eventually integrate the biology and electronics into a common platform suitable

for electronic control and biological sensing as well as bioelectronically driven

nanoassembly [6].

One promising approach for the direct electrical detection of biomolecules uses

nanowires configured as field-effect transistors (FETs). FETs readily change their

conductance upon binding of charged target biomolecules to their receptor linked

to the device surfaces. For example, recent studies by Lieber’s group have demon-

strated the use of silicon nanowire FETs for detecting proteins [7], DNA hybrids

[8], and viruses [9]. This biodetection approach may allow in principle selective de-

tection at a single particle levels [10, 11]. Nanowires hold the possibility of very

high sensitivity detection owing to the depletion or accumulation of charge car-

riers, which are caused by binding of a charged biomolecules at the surface. This

surface binding can affect the entire cross-sectional conduction pathway of these

nanostructures. For some nanowires, such as hollow carbon nanotubes, every

atom is on the surface and exposed to the environment; even small changes in

the charge environment can drastically change their electrical properties. Thus,

among different nanomaterials, carbon nanotubes have a great potential for

biosensing.

Among numerous applications of carbon nanotubes [12–14], carbon nanotube

based sensing technology is rapidly emerging into an independent research field.

As for any new research field, there is no yet consensus in the literature about the

exact sensing mechanism. In this chapter, in addition to selected examples of

carbon nanotube based sensors, we address the controversial carbon nanotube

sensing mechanism.

To date, sensor applications of carbon nanotubes have been summarized and

discussed in several excellent review articles [15–17], which primarily focus on

carbon nanotube based electrochemical sensors. This chapter covers only recent

advances in biodetection using carbon nanotube field-effect transistors (NTFETs).

It is divided into two large sections: NTFET fabrication and their sensor applica-

tions. Section 1.3 gives a detailed description of NTFET device structure, its fabri-

cation method and introduces device characteristics. This section also addresses

technical challenges in effective integration of carbon nanotubes into CMOS elec-

tronics. Section 1.4, which focuses on sensor applications of NTFETs, is divided

into several subsections. Before discussing NTFET application for biological detec-

tion we describe the effect of environmental conditions on NTFET device character-

istics. We give selected examples of NTFET sensitivity for small molecules, mobile

ions, and water (relative humidity). The effect of these factors should be well

understood before NTFET biodetection is reviewed. We also briefly describe the

operation of NTFETs in conducting media, which is particularly important for bio-

sensor applications. Then we briefly summarize interactions of carbon nanotubes

with biomolecules (e.g., polysaccharides, DNA and proteins) to set a stage for the

subsequent subsections that describe in great details recent applications of NTFETs

for label-free electronic detection of proteins, antibody–antigen interactions, DNA

hybridization, and enzymatic reactions.

2 1 Biosensing using Carbon Nanotube Field-effect Transistors

1.3

Carbon Nanotube Field-effect Transistors (NTFETs)

1.3.1

Carbon Nanotubes

Since their discovery by Iijima over a decade ago [18], interest in carbon nanotubes

has grown considerably [19]. Recent advances in the synthesis and purification of

carbon nanotubes have turned them into commercially available materials. Subse-

quently, several experiments have been undertaken to study the physical and elec-

trical properties of carbon nanotubes on the individual and macroscopic scale [20–

23]. On the macroscopic scale, spectroscopic and optical absorption measurements

have been carried out to test the purity of the carbon nanotubes [24, 25]. For elec-

tronic transport measurements it is particularly interesting to perform experiments

on isolated, individual carbon nanotubes. The properties of carbon nanotubes de-

pend strongly on physical aspects such as their diameter, length, and presence of

residual catalyst [12]. The properties measured from a large quantity of nanotubes

could be an average of all nanotubes in the sample, so that the unique character-

istics of individual carbon nanotubes could be shadowed. Experiments on individ-

ual nanotubes are very challenging due to their small size, which prohibits the

application of well-established testing techniques. Moreover, their small size also

makes their manipulation rather difficult. Specialized techniques are needed to

mount or grow an individual carbon nanotube on the electrode with sub-micron

precision.

Carbon nanotubes are hollow cylinders made of sheets of carbon atoms and

can be divided into single-walled carbon nanotubes (SWNTs) and multi-walled car-

bon nanotubes (MWNTs). SWNTs possess a cylindrical nanostructure with a high

aspect ratio, formed by rolling up a single graphite sheet into a tube (Fig. 1.1).

SWNTs are, typically, a few nanometers in diameter and up to several microns

long. MWNTs consist of several layers of graphene cylinders that are concentrically

nested like rings of a tree trunk, with an interlayer spacing of 3.4 A [26]. Because of

their unique properties, carbon nanotubes have become a material that has gener-

ated substantial interest on nanoelectronic devices and nanosensors [27, 28]. These

properties are largely dependent upon physical aspects such as diameter, length,

presence of catalyst and chirality. For example, SWNT can be metallic or semicon-

ducting, depending upon the intrinsic band gap and helicity [29]. Semiconducting

SWNTs can be used to fabricate FET devices, as demonstrated by Dekker and co-

workers [30]. In addition, semiconducting SWNTs exhibit significant conductance

changes in response to the physisorption of different gases [24, 31, 32]. Therefore,

SWNT-based nanosensors can be fabricated based on FET layout, where the solid-

state gate is replaced by adsorbed molecules that modulate the nanotube conduc-

tance [33]. Since semiconducting SWNTs have a very high mobility and, because

all their atoms are located at the surface, they are the perfect nanomaterial for sen-

sors. These sensors offer several advantages for the detection of biological species.

First, carbon nanotubes form the conducting channel in a transistor configura-

1.3 Carbon Nanotube Field-effect Transistors (NTFETs) 3

tion. Second, the nanotubes are typically located on the surface of the supporting

substrate and are in direct contact with the environment. This device geometry

contrasts with traditional metal oxide semiconductor field-effect transistors

(MOSFETs) where the conducting channel is buried in the bulk material in which

the depletion layer is formed. Lastly, all of the electrical current flows at the surface

of nanotubes. All these remarkable characteristics lead to a FET device configura-

tion that is extremely sensitive to minute variations in the surrounding environ-

ment.

1.3.2

Nanotube Synthesis

Several synthesis methods are used to produce carbon nanotubes [34]. The three

most commonly used methods are the arc discharge, laser ablation, and chemical

vapor deposition (CVD) techniques. While the arc and laser methods can produce

large quantities of carbon nanotubes they lead to resilient contaminants, including

pyrolytic and amorphous carbon [35, 36], which are difficult to remove from the

sample. Such impurities result in low recovery yield for the carbon nanotube prod-

uct. However, recent advances in scaling up these methods, as well as development

new fabrication methods such as high pressure carbon monoxide (HiPCO), have

created commercial supplies of carbon nanotubes with more than 90% purity

with competitive prices. In contrast, the less scalable CVD process offers the best

chance of obtaining controllable routes for the selective production of carbon nano-

tubes with defined properties [37]. CVD is catalytically driven, wherein a metal cat-

alyst is used in conjunction with the thermal decomposition of hydrocarbon feed-

stock gases to produce carbon nanotubes. In most cases, the resultant growth of

nanotubes occurs on a fixed substrate within the process. Figure 1.2 illustrates a

typical CVD process for the generation of SWNTs. SWNTs are synthesized by the

Fig. 1.1. A seamlessly rolled-up single

graphite sheet forms a single-walled carbon

nanotube (SWNT). SWNTs are, typically, a few

nanometers in diameter and up to several

micrometers long. They can be either metallic

or semiconducting, depending on their helicity

and diameter. Semiconducting SWNTs are

used for the fabrication of nanotube field-effect

transistors (NTFETs). (Adapted with

permission from Ref. [4], 8 Wiley-VCH Verlag).


reaction of a hydrocarbon (e.g., CH4) vapor over a dispersed Fe catalyst. The syn-

thesis apparatus consists of a quartz tube reactor inside a combined preheater and

furnace set-up. The preheat section is operated at@200 �C. The catalysts are depos-

ited and then hydrocarbon vapors are carried into the reaction zone of the furnace.

An Ar/(10%)-H2 carrier gas is used that controls the partial pressure inside the

quartz tube reactor. Reaction temperatures are typically in the range 900–1000 �C.

The SWNTs grow on the substrates (Fig. 1.3) and form thick mats that are readily

harvested. This process produces highly pure SWNTs at a yield approaching 50%

conversion of all hydrocarbon feedstock into carbon nanotube product. Similarly,

a CVD processes sometimes utilize a feed of hydrocarbon-catalyst liquid for the

production of nanotubes, and for this purpose a syringe pump is used to allow

the continuous injection of this solution into a preheat section. Various gaseous

feed-stocks are used to produce nanotubes, ranging from CH4 to C2H2. A wide

range of transition metals and rare earth promoters have been investigated for the

synthesis of SWNTs by CVD. In general, a transition metal is the major compo-

nent in the catalyst particles used regardless of the catalyst support. The most com-

mon metals found to be successful in the growth of SWNTs are Fe, Co, and Ni [38,

39]. However, bimetallic catalysts consisting of Fe/Ni, Co/Ni, or Co/Pt [40] are re-

Fig. 1.2. Schematic of a chemical vapor deposition (CVD)

reactor that uses a two-zone furnace. Carbon nanotubes grow

on the substrate placed inside the quartz tube. (Reprinted with

permission from Ref. [34], 8 2001, CRC Press).

Fig. 1.3. Transmission electron microscopy (TEM) image of an

SWNT synthesized by chemical vapor deposition (CVD).

(Reprinted with permission from Ref. [37], 8 2001, The

American Chemical Society).


ported to give the best yield of nanotubes, in addition to some rare earth metals

that have also been studied [41].

Despite challenges, the understanding the growth mechanism of SWNTs is cru-

cial for, ultimately, tailoring the production of SWNTs with known lengths, diame-

ters, helical structures and placement of SWNTs at the desired location. Presently,

efforts are underway to understand the mechanism of catalytic growth of SWNTs

on surfaces and the role of impurities and to increase nanotube yield by varying

the substrate, catalyst, and growth conditions [42]. Directional growth of SWNTs

has been achieved by electric fields [43, 44], gas flow [45], lattice directions [46],

and atomic steps [47].

1.3.3

Fabrication of NTFETs

To date carbon nanotubes have been used to fabricate various devices, including

nanotube-based mechanical devices [48] and field emission devices [13]. This sec-

tion focuses specifically on fabrication of carbon nanotube field-effect transistors

(NTFETs). Figure 1.4(a) shows a schematic drawing of NTFET. A semiconducting

carbon nanotube is contacted by source and drain electrodes while the gate elec-

Fig. 1.4. (a) Schematic representation of a

nanotube field-effect transistor (NTFET) device

with a semiconducting SWNT contacted by two

Ti/Au electrodes, representing the source (S)

and the drain (D) with a Si back gate

separated by a SiO2 insulating layer in a

transistor-configured circuit. (b) Atomic force

microscopy (AFM) image of a typical NTFET

device consisting of a single semiconducting

SWNT. (c) Scanning electron microscopy

(SEM) image of a typical NTFET device

consisting of a random array of carbon

nanotubes. (d) Typical NTFET transfer

characteristic – dependence of the source-drain

conductance (GSD) on the gate voltage (VG) –

(i) maximum conductance, (ii) modulation,

(iii) transconductance, (iv) hysteresis, and (v)

threshold voltage.


trode, which is electrically insulated from the nanotube channel, is used to manip-

ulate the nanotube’s conductivity. Depending on the particular method of nano-

tube fabrication, a NTFET can be structured in different ways [49]. However, most

publications on nanotube transistors report the use of a degenerately doped Si-

substrate with a comparatively thick (100–500 nm) thermally grown oxide layer

[30, 50–53]. Silicon substrates are readily available and can be used with both

bulk-produced nanotubes and nanotubes grown directly on the Si-substrate by

CVD. If doped highly, the Si substrate stays conductive even at low temperatures,

making it usable as a so-called back-gate with the SiO2 as a stable, if low-k, gate

dielectric. Bulk produced nanotubes (laser ablation [54] or HiPCO [55]) are usually

purified and deposited onto Si substrates by suspending them in organic solvents

(e.g., chloroform, dichloroethane, etc.) and then spin-coating or drop casting on the

substrates. In this approach the nanotubes create a random network over the sub-

strate surface. Alignment of the nanotubes is possible, if an AC dielectric field is

applied during deposition [56]. Generally, two different configurations of NTFETs,

regarding source and drain contacts, are possible. By patterning the contacts before

nanotube deposition [30] one can contact nanotubes in bulk, whereas by deposit-

ing the contacts onto the nanotubes one contacts only the ends of nanotubes [57],

because depositing contact material on top of the nanotube normally destroys the

nanotubes underneath the contacts. The second configuration usually guarantees

a lower contact resistance than what is achievable in bulk-contacted devices

[58]. Room temperature NTFET manufacturing methods, while compatible with

CMOS, are limited by ability to disperse effectively carbon nanotubes in the solu-

tion and deposit them without further aggregation on device surfaces [59]. Nano-

tube bundle formation may decrease the semiconducting character of NTFET due

to the occasional presence of nanotube bundles containing metallic nanotubes.

For CVD grown carbon nanotubes metal contacts are deposited onto the nano-

tubes [60, 61], because typical contact materials cannot withstand CVD tempera-

tures, thus making it impossible to grow carbon nanotubes on CMOS structures.

Often, the metal contacts are annealed to lower contact resistance [62]. Several

studies have tried to optimize the material used for the contacts, including Cr/Au

[49] and Pt [30], but only the Cr/Au contacts have been used widely. In this type of

contact the chromium layer is a thin (1–3 nm) adhesion layer that facilitates adhe-

sion of the gold to SiO2. An adhesion layer of Ti [63], especially when annealed,

allows deposition of smooth films of many metals onto carbon nanotubes because

Ti forms titanium carbide at the interface with the nanotube. For this reason,

Ti/Au-contacts are another frequently used combination of contact materials. Many

publications investigating Schottky barriers between a nanotube and its contacts

[61, 64] have employed such contacts. Palladium (Pd) is another material investi-

gated that wets nanotubes well and has been used recently to produce NTFETs

with ohmic contacts, i.e., contacts without Schottky barriers [52, 63].

Depending on the number of carbon nanotubes connecting the source and drain

electrodes, there are two different device architectures. In the first device architec-

ture, a single nanotube connects the source and drain (Fig. 1.4b). These devices

have been used for biosensing with excellent sensitivity. However, there is substan-


tial variation between the different devices that are fabricated and this variation is

reflected in the electronic characteristics of individual nanotubes. In addition, the

interface between the nanotube and the metallic contact may vary from device to

device. Specialized techniques are needed either to mount or grow an individual

carbon nanotube at a predetermined location. Placement is difficult and impracti-

cal for mass fabrication of NTFETs. For example, although the process of attaching

a carbon nanotube strand via arc-discharge or contact method to sharp metal probe

is fast, simple and economical it suffers from low yield. Therefore, it is difficult to

determine the quality of carbon nanotube strand attached to metal tip unless exam-

ined under SEM. When checked under SEM a large percentage of the metal probes

have multiple nanotubes attached or clusters of amorphous carbon accompanying

the carbon nanotubes [65]. Hence random networks of SWNTs have been explored

as an alternative [66].

Nanotube networks take up more space than individual SWNTs, but they are

much easier to fabricate and show great promise towards simple mass fabrication

of NTFETs. In this second configuration, the devices contain a random array of

nanotubes between source and drain electrodes (Fig. 1.4c). In this configuration,

current flows along several conducting channels that determine the overall device

resistance. The device characteristics depend on the number of nanotubes and

density of the nanotube network. It is reported that the conductance drops are as-

sociated with junctions formed by crossed semiconducting and metallic nanotubes.

Local conductance is more dependent on the number of connections to the specific

area; clusters of nanotubes with many paths to the electrode have significantly

higher conductance than those parts of the network connected through fewer

paths. Areas with low conductance typically only have two to three connections to

the network, thus it is likely that these connections are dominated by the presence

of highly resistive metallic/semiconducting junctions. When a sufficient back gate

voltage is applied to the sample, current flow through the semiconducting tubes is

suppressed. Using this technique, differences between metallic and semiconduct-

ing SWNT can be distinguished. This type of device configuration, containing a

network of conducting nanotube channels, is less sensitive than devices made of

single nanotubes. In both types of device configurations, the parameter used for

detection is the transfer characteristic – the dependence of either the source-drain

current (ISD) or conductance (GSD) (for a fixed source-drain voltage VSD) on the

gate voltage (VG) (Fig. 1.4d).

NTFETs can operate as p-type or n-type transistors. The mode of operation can

be changed from the pristine p-type to n-type by either adding electron donor mol-

ecules (n-doping) or removing adsorbed oxygen by annealing the contacts under

vacuum [67]. Polymer-gated NTFETs can also tune their modes of operation: a

change in the chemical group of the polymer changes the NTFET from p-type to

n-type [68, 69]. Oxygen doping was attributed to the fact that the oxygen interacts

with the nanotube–metal junction and causes the p-type characteristic for NTFETs

in air by pinning the metal’s Fermi level near the nanotube’s valance band maxi-

mum [33]. However, there is no apparent consensus in the literature about the

exact mechanism of chemical sensitivity of NTFETs.


1.4

Sensor Applications of NTFETs

Before discussing NTFET applications for biological detection we first describe

the effect of small molecules, relative humidity, and conductive liquid media on

NTFET devices characteristics. Effect of these factors should be well understood be-

fore NTFET biodetection is reviewed.

1.4.1

Sensitivity of NTFETs to Chemical Environment

Generally, the molecular species in the ambient environment have a significant

impact on the electrical properties of NTFETs. The conductance of semiconducting

SWNTs can be substantially increased or decreased by exposure to NO2 or NH3

[24]. Exposure to NH3 effectively shifts the valance band of the nanotube away

from the Fermi level, resulting in hole depletion and reduced conductance. In con-

trast, on exposure to NO2 molecules the conductance of nanotubes increases by

three orders of magnitude [70]. Here, exposure of the initially depleted sample to

NO2 resulted in the nanotube Fermi level shifting closer to the valence band. This

caused enriched hole carriers in the nanotube and enhanced sample conductance.

These results show that molecular gating effects can shift the Fermi level of semi-

conducting SWNTs and modulate the resistance of the sample by several orders of

magnitude.

The electronic properties of SWNTs are also extremely sensitive to air or oxygen

exposure [33]. Isolated semiconducting nanotubes can be converted into apparent

metals through room temperature exposure to oxygen. As the surrounding me-

dium was cycled between vacuum and air, a rapid and reversible change in the

SWNT resistance occurred in step with the changing environment. Initially, in a

pure atmospheric pressure oxygen environment, the thermoelectric power (TEP)

S was positive with a magnitude of nearly þ20 mV K�1. This relatively large posi-

tive TEP is consistent with that reported for pristine SWNTs near room tempera-

ture [71]. As oxygen was gradually removed from the chamber, the TEP changed

continuously from positive to negative, with a final equilibrium value of approxi-

mately �10 mV K�1. When oxygen was reintroduced into the chamber, the TEP

reversed sign and once again became positive. These dramatic 10–15% variations

in R and change in sign of the TEP demonstrate that SWNTs are exceptionally sen-

sitive to oxygen.

In the carbon nanotubes sensors mentioned above, chemical sensing experi-

ments have been conducted with devices in which both nanotubes and nanotube–

metal contacts were directly exposed to the environment. The sensing could be

dominated by the interaction of molecules with the metal contacts or the contact

interfaces. Adsorbed molecules would modify the metal work functions and, there-

by, the Schottky barrier [72, 73]. Heinze et al. [64] have assigned the effect of oxy-

gen to the Schottky barrier. Recently, a new device architecture has been studied in

which the interface between the metallic contacts and nanotubes is covered by a

1.4 Sensor Applications of NTFETs 9

passivation layer, referred to as contact-passivated [74]. In this configuration, with

the junction isolated and only the central length of the nanotube channels exposed,

the contacts should be isolated from the effect of chemicals. At the same time, the

section of the device that is open to the environment can be doped via charge trans-

fer. NTFETs with such configuration have been investigated by measuring sensitiv-

ity to NH3, NO2, and poly(ethylene imine) (Fig. 1.5).

The NTFET devices were fabricated using SWNTs grown by CVD on 200 nm of

silicon dioxide on doped silicon from iron nanoparticles as described in Section

1.3.1. These particles were exposed to flowing hydrocarbon to grow carbon nano-

tubes, and after growth optical lithography was used to pattern electrical leads

(35 nm titanium capped with 5 nm gold) on top of the nanotubes. Contact passiva-

Fig. 1.5. (a) AFM image of a contact passivated NTFET device

covered with poly(ethylene imine). (b) ISD–VG dependence for

the device in vacuum (center curve), as well as in NH3 and

NO2 gases. (Adapted with permission from Ref. [74], 8 2003

American Institute of Physics).


tion was achieved by 70 nm silicon monoxide layer. Source and drain electrodes

were separated by nearly a micrometer. The dependence of the source-drain cur-

rent (ISD) as function of the gate voltage (VG) was measured from þ10 to �10 V

using a semiconductor parameter analyzer in air/water/gas mixtures. The low con-

centrations of gas mixtures could be introduced to the devices by mixing different

proportions of air and gases. The contact-passivated devices demonstrated NH3

and NO2 sensitivity similar to regular NTFETs. Poly(ethylene imine) also produced

negative threshold shifts of tens of volts, despite being in contact with only the

center region of devices. Thus, the NTFET sensor character was preserved despite

isolating Schottky barriers.

Several groups have reported that NTFET fabricated on SiO2/Si substrates exhib-

its hysteresis in current versus gate-voltage characteristics and attributed the hyste-

resis to charge traps in bulk SiO2, oxygen-related defect trap sites near nanotubes,

or the traps at the SiO2/Si interface. It is mentioned that thermally grown SiO2 sur-

face consists of Si-OH silanol groups and is hydrated by a network of water mole-

cules that are hydrogen bonded to the silanols. The CVD nanotube growth condi-

tion (900 �C) may dehydrate the surface and condense to form SiaOaSi siloxanes.

When such a surface is exposed to and stored in ambient air, the surface siloxanes

on the substrate react with water and gradually revert to SiaOH, after which the

substrate becomes rehydrated. Heating under dry conditions significantly removes

water and reduces hysteresis in the transistors.

Kim et al. have reported that the hysteresis in electrical characteristics of NTFETs

is due to charge trapping by water molecules around the nanotubes, including

SiO2 surface-bound water proximal to the nanotubes [75]. They have demon-

strated that coating nanotube devices with PMMA can afford nearly hysteresis-

free NTFETs [75]. This passivation is attributed to two factors. First, the ester

groups of poly(methyl methacrylate) (PMMA) can hydrogen bond with silanol

group on SiO2. Baking at 150 �C combined with the polymer–SiO2 interaction

can significantly remove the silanol-bound water. Second, PMMA is hydrophobic

and can keep water in the environment from permeating the PMMA and adsorb-

ing on the nanotube in a significant manner.

Bradley et al. have attributed hysteresis in NTFET devices to cation diffusion

[76], based on the following experiments. First, NTFET devices that exhibit very

small hysteresis were fabricated. Subsequently, these devices were modified by

the addition of an electrolyte coating that created mobile ions on the surface of

the device and resulted in the large hysteresis. Experiments were also conducted

to explore possible mechanisms for cation-induced hysteresis by varying the

humidity that changes the hydration layer around the nanotubes, thus leading to

the increase of the ionic mobility. The hysteresis has been found to be sensitive to

humidity on sub-second time scales, showing promise as a humidity sensor [77].

Sensitivity of NTFETs to charges as well as NTFET operation in conducting

liquid media is important for biosensor design where the sensor should operate

in physiological buffers with complex mixtures of biomolecules. Figure 1.6 shows

a typical transfer characteristic of NTFETmeasured in air and water using the sili-

con and water as the gate electrode, respectively. The change in device characteris-


tics upon exposure to a water/gas mixture is reflected in the transfer characteris-

tics. Saline or electrolytes can also gate NTFETs and give high transconductance

[62, 78].

1.4.2

Bioconjugates of Carbon Nanotubes

Numerous reports demonstrate the ability to chemically functionalize nanotubes

for biological applications [79, 80]. Such chemistry is readily transferable to many

applications, ranging from sensors [81, 82] to electronic devices [83]. SWNTs are

chemically stable and highly hydrophobic. Therefore, they require surface modifi-

cation to establish effective SWNT–biomolecule interaction.

So far, two methods of exohedral functionalization of SWNTs have been devel-

oped – namely covalent and noncovalent. While covalent modifications [84] are

often effective at introducing functionality, they impair the desirable mechanical

and electronic properties of SWNTs. Noncovalent modifications [85], however, not

only improve the solubility of SWNTs in water, but they also constitute non-

destructive processes, which preserve the primary structures of the SWNTs, along

with their unique mechanical and electronic properties.

Previously, it has been shown that polysaccharides such as starch [86, 82, 83],

gum Arabic [84], and the b-1,3-glucans, curdlan and schizophyllan [85], will solu-

bilize SWNTs in water. It has been proposed that at least some of these polymers

achieve their goal by wrapping themselves in helical fashion around SWNTs (Fig.

Fig. 1.6. (a) Detection in liquid with NTFET

devices by using either the back gate or liquid

gate configuration. (b) NTFET transfer

characteristics in air (solid line), using the

back gate, and in water (dashed line), using

the liquid gate. Note the different x-scales for

the back and liquid gates. (Adapted with

permission from Ref. [93], 8 2003, The

American Physical Society.)


1.7). Solubilization of the SWNTs with cyclodextrins (CD), which are macrocyclic

polysaccharides, has been also investigated [86]. The observed aqueous solubility

of SWNTs with g-CD is unlikely due to encapsulation because the inner cavity di-

mensions of this CD are far too small to allow it to thread onto even the smallest

diameter SWNTs. More recently, however, it has been shown [87] that h-CD, which

has 12 a1,4-linked d-glucopyranose residues and therefore is large enough, does

thread onto SWNTs in water, not only solubilizing the NTs but also permitting

some partial separations according to their diameters.

Nucleic acids, such as single-stranded DNA, short double-stranded DNA, and

some total RNA can also disperse SWNTs in water [88, 89]. Molecular modeling

has shown [20] that the non-specific DNA–SWNT interactions in water are from

the nucleic acid–base stacking on the nanotube surface, resulting in the hydro-

philic sugar–phosphate backbone pointing to the exterior to achieve the solubility

in water. The mode of interaction could be helical wrapping or simple surface

adsorption. The charge differences among the DNA–SWNT conjugates, which are

associated with the negatively charged phosphate groups of DNA and the different

electronic properties of SWNTs, have allowed post-production preparation of sam-

ples enriched in metallic and semiconducting SWNTs.

Various proteins can also strongly bind to the nanotube exterior surface via

non-specific adsorption. Proteins such as streptavidin and HupR crystallize in heli-

cal fashion, resulting in ordered arrays of proteins on the nanotube surface [90].

Mechanistically, the non-specific adsorption of proteins onto the nanotube surface

may be more complicated than the widely attributed hydrophobic interactions.

Quite possibly, the observed substantial protein adsorption is, at least in part, asso-

ciated with the amino affinity of carbon nanotubes, as was demonstrated recently

by monitoring the conductance change in the carbon nanotube [91]. Also, inter-

Fig. 1.7. Molecular model of SWNT wrapped in an amylose

coil. (Reprinted from Ref. [79], 8 2002, The American Chemical

Society.)


molecular interactions involving aromatic amino acids, i.e., histidine and trypto-

phan, in the polypeptide chains of the proteins can contribute to the observed affin-

ity of the peptides to carbon nanotubes [92].

1.4.3

Protein Detection

Carbon nanotube interactions with proteins have been explored by NTFET devices

[91]. In NTFETdevices, the ability to measure the electronic properties of the nano-

tube allowed to query the electronic state of the immobilization substrate. In that

work two types of measurements of the device transfer characteristics were per-

formed. In the first measurement, referred to as a substrate-gate transfer character-

istic, the current through the drain contact (at fixed source-drain bias) was moni-

tored while a variable gate voltage was applied through a metallic gate buried

underneath the SiO2 substrate. In the second measurement, referred to as liquid-

gate transfer characteristics, the device was immersed in a buffer solution and a

variable gate voltage was applied through a platinum electrode. The current was

passed through the drain contact and a silver reference electrode in the solution.

During these measurements, the assembly was shaken gently, using a lab rotator

at 3 Hz. The effect of protein adsorption was studied with both measurements. De-

vices were incubated with streptavidin (40 nm) in 15 mm phosphate buffer at 25 �C.

Liquid-gate transfer characteristics were measured continually during the incuba-

tions. After 10 h, the devices were rinsed with distilled water and blown dry, and

the substrate-gated transfer characteristics of the dried devices were measured.

These results were discussed in terms of a simple model in which adsorbed

streptavidin coats the single-walled nanotube (Fig. 1.8). The gradual shift in the

threshold voltage is assumed to result from the slow accumulation of a full mono-

layer of adsorbed protein. This coverage-dependent threshold shift is analogous

Fig. 1.8. (a) Size comparison between a

carbon nanotube and a streptavidin molecule.

(b) Current versus gate voltage for a nanotube

device; VSD ¼ 10 mV. (ii) In phosphate buffer

before streptavidin addition. (i) same

conditions, to measure the uncertainty in the

threshold voltage. (iii) After 10 h of incubation

with streptavidin. Arrows indicate the threshold

voltages for the three curves [the arrow for (i)

is behind that for (ii)]. (Adapted with

permission from Ref. [91], 8 2003, The

American Chemical Society.)


to the concentration-dependent shift observed when such devices are exposed to

aqueous ammonia [93]. The protein adsorbate equilibrates over several hours so

that only the full monolayer can be conclusively determined. Such protein mono-

layers form under various conditions at interfaces that permit protein crystalliza-

tion, including sidewalls of MWNTs [90, 94]. The results support the proposal

that conductance changes are due to charge injection or field effects caused by pro-

teins adsorbed solely along the lengths of the nanotubes.

Protein adsorption on NTFET leads to appreciable changes in the electrical con-

ductance of the devices that can be exploited for label-free detection of biomole-

cules with a high potential for miniaturization. For example, Dai and coworkers

[95] have used a sensor design consisting of an array of four NTFET sensors on

SiO2/Si chips. Each NTFET consists of multiple SWNTs connected roughly in par-

allel across two closely spaced bridging metal electrodes. Three types of devices

with different surface functional groups were prepared for the investigation of the

biosensing: (1) unmodified as-made devices, (2) devices fabricated with mPEG-SH

SAMs formed on, and only on, the metal contact electrodes and, lastly, (3) devices

with mPEG-SH SAMs on the metal contacts and a Tween 20 coating on the carbon

nanotubes. Electrical conductance of these devices upon the addition of various

protein molecules was monitored. While device type 1 showed a significant con-

ductance change with protein adsorption, device type 2 with an mPEG-SH SAM

on the metal electrodes did not give any conductance change, except in the case of

the protein avidin. It was reported that the metal–nanotube interface or contact

region is highly susceptible to modulation by adsorbed species [64]. Modulation

of metal work function can alter the Schottky barrier of the metal–nanotube inter-

face, thus leading to a significant change in the nature of contacts and, conse-

quently, a change in the conductance of the devices.

In situ detection of a small number of proteins by directly measuring the elec-

tron transport properties of a single SWNT has been reported by Nagahara and

coworkers [96]. Cytochrome c (cytc) adsorption onto individual NTFET has been

detected via the changes in the electron transport properties of the transistors.

The adsorption of cytc induces a decrease in the conductance of the NTFET

devices, corresponding to a few tens of molecules. This experiment was carried

out by measuring the conductance versus electrochemical potential of the SWNT

with respect to a reference electrode inserted in the solution, and observed a nega-

tive shift in the conductance versus potential plot upon protein adsorption. The

number of adsorbed proteins has been estimated from this shift.

1.4.4

Detection of Antibody–Antigen Interactions

Specific sensitivity can be achieved by employing recognition layers that induce

chemical reactions and modify the transfer characteristics. In this two-layer archi-

tecture carbon nanotubes function as extremely sensitive transducers while the rec-

ognition layer provides chemical selectivity and prevents non-specific binding that

is common for complex biological samples.


Following this design, nanotubes have been functionalized to be biocompatible

and to be capable of recognizing proteins. This functionalization has involved

noncovalent binding between a bifunctional molecule and a nanotube to anchor a

bioreceptor molecule with a high degree of control and specificity. Star and co-

workers have fabricated [97] NTFET devices sensitive to streptavidin using a bio-

tin-functionalized carbon nanotube bridging two microelectrodes (source and

drain, Fig. 1.9a). The SWNT in the NTFET device was coated with a mixture of

two polymers, poly(ethyleneimine) and poly(ethylene glycol). The former provided

amino groups for the coupling of biotin–N-hydroxysuccinimidyl ester (Fig. 1.9b)

and the latter prevented the nonspecific adsorption of proteins on the functional-

ized carbon nanotube. Figure 1.9(c) shows an AFM image of the device after its

exposure to streptavidin labeled with gold nanoparticles (10 nm). Lighter dots rep-

resent gold nanoparticles and indicate the presence of streptavidin bound to the

Fig. 1.9. (a) Schematic of NTFET coated

with a biotinylated polymer layer for specific

streptavidin binding. (b) Biotinylation reaction

of the polymer layer (PEI/PEG) on the side-wall

of the SWNT. (c) AFM image of the polymer-

coated and biotinylated NTFET device after

exposure to streptavidin labeled with gold

nanoparticles (10 nm in diameter). (d) Source-

drain current dependence on gate voltage of

the NTFET device based on SWNTs functioned

with biotin in both the absence and presence

of streptavidin. (Adapted with permission from

Ref. [97], 8 2003, The American Chemical

Society.)


biotinylated carbon nanotube. The source-drain current dependence on the gate

voltage of the NTFET shows a significant change upon the streptavidin binding

to the biotin-functionalized carbon nanotube (Fig. 1.9d). The experiments reveal

the specific binding of the streptavidin, which occurs only at the biotinylated

interface.

The mechanism of the biodetection was explained in terms of the effect of

the electron doping of the carbon nanotube channel upon the binding of the

charged streptavidin molecules. Dai and coworkers [98] have also analyzed specific

antigen–antibody interactions using NTFET devices. In particular, they have

studied the affinity binding of 10E3 mAbs antibody (a prototype target of the auto-

immune response in patients with systematic lupus erythematosus and mixed con-

nective tissue disease) to human auto antigen U1A.

1.4.5

DNA Detection

DNA biosensors based on nucleic acid recognition processes are quickly being

developed towards the goal of rapid, simple and inexpensive testing of genetic

and infectious diseases. To date, there are several reports on the electrochemical

detection of DNA hybridization using multi-walled carbon nanotube (MWNT) elec-

trodes [99]. Whereas electrochemical methods rely on the electrochemical behavior

of the labels, measurements of the direct electron transfer between SWNTs and

DNA molecules paves the way for label-free DNA detection (Fig. 1.10). To illustrate

the practical utility of this new nanoelectronic detection method, an allele-specific

assay to detect the presence of SNPs using NTFETs has been recently developed

[100]. This DNA assay targeted the H63D polymorphism in the human HFE

gene, which is associated with hereditary hemochromatosis, a common and easily

treated disease of iron metabolism [101, 102].

DNA sensing mechanism using NTFETs has been recently explored by selective

attachment of DNA molecules at different device segments. Tang et al. [103] have

found that DNA hybridization on gold electrodes rather than on SWNT sidewalls is

mainly responsible for NTFET detection due to Schottky barrier modulation. In

another approach, DNA hybridization occurs on the surface at the gate of NTFET

[104]. As a result, the conductance in SWNTs was changed through the gate insu-

lators. In the work, the 5 0 end-amino modified peptide nucleic acid (PNA) oligonu-

cleotides were covalently immobilized onto the Au surfaces of the back gate of

NTFETs. PNA is a synthetic analog of DNA, in which both the phosphate and the

deoxyribose of the DNA backbone are replaced by a polypeptide. PNA mimics the

behavior of DNA and hybridizes with complementary DNA or RNA sequences,

thus enabling PNA chips to be used in biosensors. The micro-flow chip was fabri-

cated by using poly(dimethylsiloxane) (PDMS) prepolymer. The NTFET nano-

sensor array was placed onto the PDMS chip in such a way that the PNA probe-

modified Au side was positioned to face the open chamber for the introduction of

solutions and the electrical measurements. A PNA probe with the base sequence

5 0-NH2-ACC ACC ACT TC-3 0, which was fully complementary to the tumor necro-


sis factor-a (TNF-a) gene sequence, was used as a model system. The base se-

quence for full complementary target DNA was 5 0-GGT TTC GAA GTG GTG

GTC TTG-3 0 while the non-complementary DNA oligonucleotide sequence was

5 0-CCC TAA GCC CCC AA-3 0.

The electrical properties of the NTFET devices were measured at room tempera-

ture in air. First, the blank PBS solution was introduced into the PDMS-based

micro flow chip, revealing that no substantial change in the source-drain current

of NTFET was obtained. The current increased dramatically while monitoring in

real time for about 3 h. The increase in conductance for the p-type NTFET device

was consistent with an increase in negative surface charge density associated with

binding of negatively charged oligonucleotides at the surface. DNA hybridization

can be detected by measuring the electrical characteristics of NTFETs, and SWNT

based FET can be employed for label-free, direct real time electrical detection of

biomolecule binding.

Fig. 1.10. Label-free detection of DNA

hybridization using NTFET devices. (a) G–Vg

curves after incubation with allele-specific wild-

type capture probe and after challenging the

device with wild-type synthetic HFE target

(50 nm). (b) G–Vg curves in the experiment

with mutant capture probe. (c) Graph with

electronic ð1� G=G0Þ and fluorescent

responses in SNP detection assays. (d)

Fluorescence microscopy image of the NTFET

network device, with the electrodes 10 mm

apart, after incubation with Cy5-labeled DNA

molecules. (Adapted with permission from

Ref. [100], 8 2006, The National Academy of

Sciences of the USA.)


1.4.6

Enzymatic Reactions

SWNTs can be made water soluble by wrapping in amylose (linear component of

starch) [86]. These SWNT solutions are stable for weeks, provided nobody spits on

them. Indeed, the addition of saliva, which contains a-amylase, precipitates the

nanotubes as the enzyme breaks amylose down into smaller carbohydrate frag-

ments, finally resulting in the formation of glucose. The enzymatic degradation of

starch has been recently monitored electronically using NTFETs [105]. Figure

1.11(a) shows the experimental setup used for this study. NTFET devices display

transconductance and source-drain current–voltage characteristics typical of the

p-type device behavior. The device characteristics, i.e., the source-drain current ISDas a function of the gate voltage VG, were measured to evaluate the effect of starch

deposition and the subsequent enzymatic degradation of the starch layer on the

carbon nanotubes.

Starch was deposited onto the FET by soaking the silicon wafer in a 5% aqueous

starch solution and the device characteristics were found to be shifted by approxi-

Fig. 1.11. (a) NTFET device for electronic

monitoring of the enzymatic degradation of

starch with amyloglucosidase (AMG) to

glucose. (b) High-resolution transmission

electron microscopy (HRTEM) image of a

SWNT (2.0 nm diameter) after treatment with

a drop of a 1% of an aqueous solution of

starch. The starch had been stained with RuO4

vapor. (c) NTFET device characteristics in the

form of ISD–VG curves measured from þ10 to

�10 V gate voltage with a þ0:6 V bias voltage

before (bare) and after starch deposition, as

well as after hydrolysis with AMG. (Adapted

with permission from Ref. [104], 8 2004, The

American Chemical Society.)


mately 2 V toward more negative gate voltages. The direction of the shift equates

with electron doping of the nanotube channel by the polysaccharide. Quantitatively

similar doping effects have been observed when carbon nanotube FET devices were

exposed to NH3 gas, amines [106], poly(ethylene imine) (PEI) [107], and proteins

[91]. After the enzyme-catalyzed reaction had been performed on the starch-

functionalized devices and washed with buffer, the ISD vs. VG characteristics recov-

ered almost completely to the trace recorded before starch deposition (Fig. 1.11).

This indicates that, during the enzyme-catalyzed reaction, nearly all the starch

deposited on the surface of the nanotube device is hydrolyzed to glucose which is

washed off by the buffer prior to the electronic measurements.

1.4.7

Glucose Detection

The diagnosis and management of diabetes mellitus requires a tight monitoring

of blood glucose levels. Dekker and coworkers have demonstrated the use of indi-

vidual semiconducting SWNT as a versatile biosensor [108]. The redox enzyme

glucose oxidase (GOx) that catalyses the oxidation of b-d-glucose (C6H12O6) to d-

glucono-1,5-lactone (C6H10O6) has been studied. The redox enzymes go through

a catalytic reaction cycle where groups in the enzyme temporarily change their

charge state and conformational changes occur in the enzyme that can be detected

using NTFET devices.

In addition to pH sensitivity, GOx-coated semiconducting SWNTs appeared to be

sensitive to glucose, the substrate of GOx. Figure 1.12 exhibits real-time measure-

Fig. 1.12. Real time electronic response of the

NTFET sensor to glucose, the substrate of

glucose oxidase (GOx). The conductance of a

semiconducting SWNT with immobilized GOx

is measured as a function of time in 5 mL milli-

Q water. The conductance of the GOx-coated

SWNT increases upon addition of glucose to

the liquid. Inset: (a) the same measurement on

a second device where the conductance was a

factor of 10 lower; (b) the same measurement

on a semiconducting SWNT without GOx; no

conductance increase is observed in this case.

(Reprinted with permission from Ref. [107],

8 2003, The American Chemical Society.) (B)

Schematic of GOx immobilized on SWNT for

electronic glucose detection.


ments where the conductance of a GOx-coated semiconducting SWNT in milli-Q

water has been recorded in the liquid (left-hand arrow in each graph in Fig. 1.12).

No significant change in conductance was observed as a result of water addition.

When 0.1 m glucose in milli-Q water was added to the liquid (right-hand arrow in

each graph), however, the conductance of the tube increased by about 10%. A sim-

ilar 10% conductance change was observed for another device (Fig. 1.12a inset),

which had a factor 10 lower conductance. Glucose did not change the conductance

of the bare SWNT but did increase the device conductance after GOx was immobi-

lized. Inset (b) of Fig. 1.12 shows such measurement on a bare semiconducting

SWNT. These measurements clearly indicate that the GOx activity is responsible

for the observed increase in conductance upon glucose addition, thus rendering

such nanodevices as feasible enzymatic-activity sensors.

1.5

Conclusion and Outlook

Recent advances in the rapidly developing area of biomolecule detection using car-

bon nanotube systems have been summarized here. SWNTs appear as structurally

defined components for various electronic devices. The semiconductive properties

of SWNTs are of special interest as these SWNTs have been applied to fabricate

FETs for sensing applications. This area requires further development, particularly

related to the fabrication of FETs based on individual SWNTs. The use of carbon

nanotubes as nanocircuitry elements is particularly interesting. Biomaterials linked

to nanotubes may be used as binding elements for the specific linkage of the nano-

tube to surface in the form of addressable structures.

Important chemical means to functionalize SWNTs with other electronic materi-

als such as conductive polymers or nanoparticles is anticipated to generate materi-

als of new properties and functions. The localized nanoscale contacts of SWNTs

with bio-surfaces will be a major advance in understanding and exploring the new

applications. The use of nanodevices to monitor various biologically significant

reactions is envisioned. In future, it should be possible to connect the living cells

directly to these nanoelectronic devices to measure the electronic responses of liv-

ing systems. The combination of the unique electronic properties of SWNTs and

catalytic features of biological system could provide new opportunities for carbon

nanotubes based bioelectronics.

References

1 Park, S., Taton, T. A., Mirkin, C. A.

Array-based electrical detection of

DNA with nanoparticle probes.,

Science 2002, 295, 1503–1506.2 He, L., Musick, M. D., Nicewarner,

S. R., Salinas, F. G., Benkovic, S. J.,

Natan, M. J., Keating, C. D.

Colloidal Au-enhanced surface

plasmon resonance for ultrasensitive

detection of DNA hybridization.,

J. Am. Chem. Soc. 2000, 122, 9071–9077.

References 21

3 McFarland, A. D., Van Duyne, R. P.

Single silver nanoparticles as real-time

optical sensors with zeptomole

sensitivity., Nano Lett. 2003, 3,1057–1062.

4 Graham, A. P., Duesberg, G. S.,

Seidel, R. V., Liebau, M., Unger, E.,

Pamler, W., Kreupl, F., Hoenlein,

W. Carbon nanotubes for microelec-

tronics?, Small 2005, 1, 382–390.5 Gruner, G. Carbon nanotube

transistors for biosensing applications.,

Anal. & Bioanal. Chem., 2006, 384,322–335.

6 Willner, I. Biomaterials for sensors,

fuel cells, and circuitry., Science 2002,298, 2407–2408.

7 Cui, Y., Wei, Q., Park, H., Lieber,

C. M. Nanowire nanosensors for

highly sensitive and selective detection

of biological and chemical species.,

Science 2001, 293, 1289–1292.8 Hahm, J., Lieber, C. M. Direct

ultrasensitive electrical detection of

DNA and DNA sequence variations

using nanowire nanosensors., NanoLett. 2004, 4, 51–54.

9 Patolsky, F., Zheng, G., Hayden, O.,

Lakadamyali, M., Zhung, X., Lieber,

C. M. Electrical detection of single

viruses, Proc. Natl. Acad. Sci. U.S.A.2004, 101, 14017–14022.

10 Weiss, S. Measuring conformational

dynamics of biomolecules by single

molecule fluorescence spectroscopy.,

Nat. Struct. Biol. 2000, 7, 724–729.11 Zhuang, X. W., Rief, M. Single-

molecule folding., Curr. Opin. Struct.Biol. 2003, 13, 88–97.

12 Ajayan, P. M. Nanotubes from

carbon., Chem. Rev. 1999, 99, 1787–1799.

13 Ajayan, P. M., Zhou, O. Z.

Applications of carbon nanotubes.,

Top. Appl. Phys. 2001, 80, 391–425.14 Special issue on carbon nanotubes,

Acc. Chem. Res. 2002, vol. 35.15 Wang, J. Carbon-nanotube based

electrochemical biosensors: A review.,

Electroanalysis 2005, 17, 7–14.16 Lin, Y., Taylor, S., Li, H., Shiral

Fernando, K. A., Qu, L., Wang, W.,

Gu, L., Zhou, B., Sun, Y. P. Advances

toward bioapplications of carbon

nanotubes., J. Mater. Chem. 2004, 14,527–541.

17 Katz, E., Willner, I. Biomolecule-

functionalized carbon nanotubes:

applications in nanobioelectronics.,

ChemPhysChem 2004, 5, 1084–1104.

18 Iijima, S. Helical microtubules of

graphitic carbon., Nature (London)1991, 354, 56–58.

19 Dresselhaus, M. S., Dresselhaus,

G., Eklund, P. C. Science of Fullerenesand Carbon Nanotubes, Academic

Press, San Diego, 1996.

20 Saito, R., Dresselhaus, G.,

Dresselhaus, M. S. Physical Propertiesof Carbon Nanotubes, Imperial College

Press, London, 1999.

21 de Jonge, N., Lamy, Y., Schoots, K.,

Oosterkamp, T. H. High brightness

electron beam from a multi-walled

carbon nanotube, Nature (London)2002, 420, 393–395.

22 Fransen, M. J., Van Rooy, Th. L.,

Kruit, P. Field emission energy

distributions from individual

multiwalled carbon nanotubes., Appl.Surf. Sci. 1999, 146, 312–327.

23 Wang, Z. L., Poncharal, P., de

Heer, W. A. Measuring physical and

mechanical properties of individual

carbon nanotubes by in situ TEM.,

J. Phys. Chem. Solids 2000, 61,1025–1030.

24 Kong, J., Franklin, N. R., Zhou,

C. W., Chapline, M. G., Peng, S.,

Cho, K., Dai, H. Nanotube molecular

wires as chemical sensors., Science2000, 287, 622–625.

25 Sugie, H., Tanemura, M., Filip, V.,

Iwata, K., Takahashi, K., Okuyama,

F. Carbon nanotubes as electron

source in an x-ray tube., Appl. Phys.Lett. 2001, 78, 2578–2580.

26 Baughman, R. H., Zakhidov, A., de

Heer, W. A. Carbon nanotubes – the

route toward applications., Science2002, 297, 787–792.

27 Dekker, C. Carbon nanotubes as

molecular quantum wires., Phys.Today 1999, 52, 22–28.

28 McEuen, P. L., Fuhrer, M. S., Park,

H. Single-walled carbon nanotube

electronics., IEEE Trans. Nanotechnol.2002, 1, 78–85.


29 Dresselhaus, M. S. Burn and

interrogate., Science 2001, 292,650–651.

30 Tans, S. J., Verschueren, A. R. M.,

Dekker, C. Room-temperature

transistor based on a single carbon

nanotube., Nature 1998, 393, 49–52.31 Valentini, L., Armentano, I.,

Kenny, J. M., Cantalini, C., Lozzi,

L., Santucci, S. Sensor for sub-ppm

NO2 gas detection based on carbon

nanotube thin films., Appl. Phys. Lett.2003, 82, 961–963.

32 Li, J., Lu, Y. J., Ye, Q., Cinke, M.,

Han, J., Meyyappan, M. Carbon

nanotube sensors for gas and organic

vapor detection, Nano Lett. 2003, 3,929–933.

33 Collins, P. G., Bradley, K.,

Ishigami, M., Zettl, A. Extreme

oxygen sensitivity of electronic

properties of carbon nanotubes.,

Science 2000, 287, 1801–1804.34 Sinnott, S. B., Andrews, R. Carbon

nanotubes: Synthesis, properties, and

applications., Crit. Rev. Solid StateMater. Sci. 2001, 26, 145–249.

35 Kokai, F., Takahashi, K., Yudasaka,

M., Yamada, R., Ichihashi, T. and

Iijima, S. Laser ablation of graphite-

Co/Ni and growth of single-wall

carbon nanotubes in vortexes formed

in an Ar atmosphere, J. Phys. Chem. B2000, 104, 6777–6784.

36 Takizawa, M., Bandow, S., Yudasaka,

M., Ando, Y., Shimoyama, H., Iijima,

S. Change of tube diameter

distribution of single-wall carbon

nanotubes induced by changing the

bimetallic ratio of Ni and Y catalysts.,

Chem. Phys. Lett. 2000, 326, 351–357.37 Li, Y., Kim, W., Zhang, Y., Rolandi,

M., Wang, W., Dai, H. Growth of

single-walled carbon nanotubes from

discrete catalytic nanoparticles of

various sizes., J. Phys. Chem. B 2001,

105, 11 424–11 431.

38 Iijima, S., Ichihasi, T. Single-shell

carbon nanotubes of 1-nm diameter.,

Nature (London) 1993, 363, 603–605.39 Bethune, D. S., Kiang, C. H., de

Vires, M. S., Gorman, G., Savoy, R.,

Vazquez, J., Beyers, R. Cobalt-

catalysed growth of carbon nanotubes

with single-atomic-layer walls, Nature(London) 1993, 363, 605–607.

40 Seraphin, S., Zhou, D. Single-walled

carbon nanotubes produced at high

yield by mixed catalyst, Appl. Phys.Lett. 1994, 64, 2087–2089.

41 Willems, I., Konya, Z., Colomer,

J.-F., Van Tendeloo, G., Nagaraju,

N., Fonseca, A. and Nagy, J. B.

Control of the outer diameter of thin

carbon nanotubes synthesized by

catalytic decomposition of hydro-

carbon, Chem. Phys. Lett. 2000, 317,71–76.

42 Hata, K., Futaba, D. N., Mizuno, K.,

Namai, T., Yumura, M., Iijima, S.

Water-assisted highly efficient

synthesis of impurity-free single-

walled carbon nanotubes., Science2004, 306, 1362–1364.

43 Joselevich, E., Lieber, C. M. Vectorial

growth of metallic and semiconducting

single-wall carbon nanotubes, NanoLett. 2002, 2, 1137–1141.

44 Ural, A., Li, Y., Dai, H. Electric-field-

aligned growth of single-walled carbon

nanotubes on surfaces., Appl. Phys.Lett. 2002, 81, 3464–3466.

45 Gao, J., Yu, A., Itkis, M. E.,

Bekyarova, E., Zhao, B., Niyogi, S.,

Haddon, R. C. Large-scale fabrication

of aligned single-walled carbon

nanotube array and hierarchical

single-walled carbon nanotube

assembly., J. Am. Chem. Soc. 2004,126, 16 698–16 699.

46 Su, M., Li, Y., Maynor, B., Bwldum,

A., Lu, J. P., Liu, J. Lattice-oriented

growth of single-walled carbon

nanotubes, J. Phys. Chem. B 2000, 104,

6505–6508.

47 Ismaoh, A., Sergev, L., Wachtel, E.,

Joselevich, E. Atomic-step-templated

formation of single wall carbon

nanotube patterns, Angew. Chem, Int.Ed. 2004, 43, 6140–6143.

48 Qian, D., Wagner, G. J., Liu, W. K.,

Yu, M. F., Ruoff, R. S. Mechanics of

carbon nanotubes., Appl. MechanicsRev. 2002, 55, 495–533.

49 Durkop, T., Electronic Properties of

carbon nanotubes studied in field-

effect transistor geometries, PhD

thesis 2004.

References 23

50 Martel, R., Schmidt, T., Shea, H. R.,

Hertel, T., Avouris, Ph. Single- and

multi-wall carbon nanotube field-effect

transistors., Appl. Phys. Lett. 1998, 73,2447–2449.

51 Fuhrer, M. S., Kim, B. M., Chen,

Y. F., Durkop, T., Brintlinger, T.

High-mobility nanotube transistor

memory., Nano Lett. 2002, 2, 755–759.52 Javey, A., Guo, J., Wang, Q.,

Lundstrom, M., Dai, H. Ballistic

carbon nanotube field-effect

transistors, Nature 2003, 424, 654–657.53 Durkop, T., Getty, S. A., Cobas, E.,

Fuhrer, M. S. Extraordinary mobility

in semiconducting carbon nanotubes.,

Nano Lett. 2004, 4, 35–39.54 Thess, A., Lee, R., Nikolaev, P., Dai,

H., Petit, P., Robert, J., Xu, C., Lee,

Y. H., Kim, S. G., Rinzler, A. G.,

Colbert, D. T., Scuseria, G. E.,

Tomanek, D., Fisher, J. E., Smalley,

R. E. Crystalline ropes of metallic

carbon nanotubes, Science 1996, 273,483–487.

55 Bronikowski, M. J., Willis, P. A.,

Colbert, D. T., Smith, K. A.,

Smalley, R. E. Gas-phase production

of carbon single-walled nanotubes

from carbon monoxide via the HiPco

process: A parametric study., J. Vac.Sci. Technol. A 2001, 19, 1800–1805.

56 Diehl, M. R., Yaliraki, S. N., Beck-

man, R. A., Barahona, M., Heath, J.

R. Self-assembled, deterministic carbon

nanotube wiring networks., Angew.Chem. Int. Ed. 2002, 41, 353–356.

57 Martel, R., Derycke, V., Lavoie, C.,

Appenzeller, J., Chan, K. K., Tersoff,

J., Avouris, Ph. Ambipolar electrical

transport in semiconducting single-

wall carbon nanotubes., Phys. Rev. Lett.2001, 87, 256 805-1–256 805-4.

58 Bockrath, M., Cobden, D. H., Liu,

J., Rinzler, A. G., Smalley, R. E.,

Balents, L., McEuen, P. L. Luttinger-

liquid behaviour in carbon nanotubes.,

Nature 1999, 397, 598–601.59 Hu, L., Hecht, D. S., Gruner, G.

Percolation in transparent and conduct-

ing carbon nanotube networks., NanoLett. 2004, 4, 2513–2517.

60 Radosavljevic, M., Freitag, M.,

Thadani, K. V., Johnson, A. T.

Nonvolatile molecular memory

elements based on ambipolar

nanotube field effect transistors., NanoLett. 2002, 2, 761–764.

61 Appenzeller, J., Knoch, J., Derycke,

V., Martel, R., Wind, S., Avouris,

Ph. Field-modulated carrier transport

in carbon nanotube transistors., Phys.Rev. Lett. 2002, 89, 126 801-1–126 801-4.

62 Rosenblatt, S., Yaish, Y., Park, J.,

Gore, J., Sazonova, V., McEuen, P. L.

High performance electrolyte gated

carbon nanotube transistors., NanoLett. 2002, 2, 869–872.

63 Zhang, Y., Dai, H. Formation of

metal nanowires on suspended single-

walled carbon nanotubes, Appl. Phys.Lett. 2000, 77, 3015–3017.

64 Heinze, S., Tersoff, J., Martel, R.,

Derycke, V., Appenzeller, J.,

Avouris, Ph. Carbon nanotubes as

Schottky barrier transistors., Phys.Rev. Lett. 2002, 89, 106 801-1–106 801-4.

65 de Jonge, N., van Druten, N. J. Field

emission from individual multiwalled

carbon nanotubes prepared in an

electron microscope., Ultramicroscopy2003, 95, 85–91.

66 Stadermann, M., Papadakis, S. J.,

Falvo, M. R., Novak, J., Snow, E., Fu,

Q., Liu, J., Fridman, Y., Boland, J. J.,

Superfine, R., Washburn, S.

Nanoscale study of conduction

through carbon nanotube networks.,

Phys. Rev. B 2004, 69,

201 402-1–201 402-3.

67 Derycke, V., Martel, R.,

Appenzeller, J., Avouris, Ph.

Controlling doping and carrier

injection in carbon nanotube


68 Lu, C., Fu, Q., Huang, S., Liu, J.

Polymer electrolyte-gated carbon

nanotube field-effect transistor., NanoLett. 2004, 4, 623–627.

69 Siddons, G. P., Merchin, D., Back,

J. H., Jeong, J. K., Shim, M. Highly

efficient gating and doping of carbon

nanotubes with polymer electrolytes.,

Nano Lett. 2004, 4, 927–931.70 Martel, R., Schmidt, T., Shea, H. R.,

Hertel, T., Avouris, Ph. Single- and


multi-wall carbon nanotube field-effect


71 Hone, J., Ellwood, I., Muno, M.,

Mizel, A., Cohen, M. L., Zettl, A.,

Rinzler, A. G., Smalley, R. E.

Thermoelectric power of single-walled

carbon nanotubes., Phys. Rev. Lett.1998, 80, 1042–1045.

72 Nakanishi, T., Bachtold, A.,

Dekker, C. Transport through the

interface between a semiconducting

carbon nanotube and a metal

electrode., Phys. Rev. B 2002, 66,

073 307-1–073 307-4.

73 Freitag, M., Johnson, A. T.,

Kalinin, S., Bonnell, D. Role of

single defects in electronic transport

through carbon nanotube field-effect

transistors., Phys. Rev. Lett. 2002, 89,216 801-1–216 801-4.

74 Bradley, K., Gabriel, J.-C. P., Star,

A., Gruner, G. Short-channel effects

in contact-passivated nanotube

chemical sensors., Appl. Phys. Lett.2003, 83, 3821–3823.

75 Kim, W., Javery, A., Vermesh, O.,

Wang, Q., Li, Y., Dai, H. Hysteresis

caused by water molecules in carbon

nanotube field-effect transistors., NanoLett. 2003, 3, 193–198.

76 Bradley, K., Cumings, J., Star, A.,

Gabriel, J.-C. P., Gruner, G. Influence

of mobile ions on nanotube based FET

devices., Nano Lett. 2003, 3, 639–641.77 Star, A., Han, T.-R., Joshi, V.,

Stetter, J. R. Sensing with nafion

coated carbon nanotube field-effect

transistors., Electroanalysis 2004, 16,108–112.

78 Kruger, M., Buitelaar, M. R.,

Nussbaumer, T., Schonenberger, C.,

Forro, L. Electrochemical carbon

nanotube field-effect transistor., Appl.Phys. Lett. 2001, 78, 1291–1293.

79 Dagani, R. DNA matches., Chem.Eng. News 2002, 80, 8.

80 Kim, O.-K., Je, J., Baldwin, J. W.,

Kooi, S., Pehrsson, P. E., Buckley,

L. Solubilization of single-wall carbon

nanotubes by supramolecular

encapsulation of helical amylose.,

J. Am. Chem. Soc. 2003, 125,4426–4427.

81 Bandyopadhyaya, R., Nativ-Roth, E.,

Regev, O., Yerushalmi-Rozen, R.

Stabilization of individual carbon

nanotubes in aqueous solutions., NanoLett. 2002, 2, 25–28.

82 Numata, M., Asai, M., Kaneko, K.,

Hasegawa, T., Fujita, N., Kitada, Y.,

Sakurai, K., Shinkai, S. Curdlan and

schizophyllan (beta-1,3-glucans) can

entrap single-wall carbon nanotubes

in their helical superstructure, Chem.Lett. 2004, 33, 232–233.

83 Chen, J., Dyer, M. J., Yu, M.-F.

Cyclodextrin-mediated soft cutting of

single-walled carbon nanotubes., J. Am.Chem. Soc. 2001, 123, 6201–6202.

84 Chen, J., Hamon, M. A., Hiu, H.,

Chen, Y., Rao, A. M., Eklund, P. C.,

Haddon, R. C. Solution properties of

single-walled carbon nanotubes.,

Science 1998, 282, 95–98.85 Star, A., Stoddart, J. F., Steuerman,

D., Diehl, M., Boukai, A., Wong,

E. W., Yang, X., Chung, S. W.,

Heath, J. R. Preparation and

properties of polymer-wrapped single-

walled carbon nanotubes., AngewChem. Int. Ed. 2001, 40, 1721–1725.

86 Star, A., Steuerman, D. W., Heath,

J. R., Stoddart, J. F. Starched carbon

nanotubes., Angew. Chem. Int. Ed.2002, 41, 2508–2512.

87 Dodziuk, H., Ejchart, A.,

Anczewski, W., Ueda, H.,

Krinichnaya, E., Dolgonos, G.,

Kutner, W. Water solubilization,

determination of the number of

different types of single-wall carbon

nanotubes and their partial separation

with respect to diameters by

complexation with h-cyclodextrin.,

Chem. Commun. 2003, 986–987.88 Zheng, M., Jagota, A., Semke, E. D.,

Diner, B. A., McLean, R. S., Lustig,

S. R., Richardson, R. E., Tassi, N. G.

DNA-assisted dispersion and

separation of carbon nanotubes., Nat.Mater. 2003, 2, 338–342.

89 Zheng, M., Jagota, A., Strano, M. S.,

Santos, A. P., Barone, P., Chou,

S. G., Diner, B. A., Dresselhaus,

M. S., Mclean, R. S., Onoa, G. B.,

Samsonidze, G. G., Semke, E. D.,

Usrey, M., Walls, D. J. Structure-

References 25

based carbon nanotube sorting by

sequence-dependent DNA assembly.,

Science 2003, 302, 1545–1548.90 Balavoine, F., Schultz, P., Richard,

C., Mallouh, V., Ebbesen, T. W.,

Mioskowski, C. Helical crystallization

of proteins on carbon nanotubes:

A first step towards the development

of new biosensors., Angew. Chem. Int.Ed. 1999, 38, 1912–1915.

91 Bradley, K., Briman, M., Star, A.,

Gruner, G. Charge transfer from

adsorbed proteins., Nano Lett. 2004, 4,253–256.

92 Wang, S., Humphreys, E. S., Chung,

S.-Y., Delduco, D. F., Lustig, S. R.,

Wang, H., Parker, K. N., Rizzo,

N. W., Subramoney, S., Chiang,

Y.-M., Jagota, A. Peptides with

selective affinity for carbon nanotubes,

Nat. Mater. 2003, 2, 196–200.93 Bradley, K., Gabriel, J.-C. P.,

Briman, M., Star, A., Gruner, G.

Charge transfer from ammonia phy-

sisorbed on nanotubes, Phys. Rev. Lett.2003, 91, 218 301-1–218 301-4.

94 Wilson-Kubalek, E. M., Brown, R. E.,

Celia, H., Milligan, R. A. Lipid

nanotubes as substrates for helical

crystallization of macromolecules,

Proc. Natl. Acad. Sci. U.S.A. 1998, 95,8040–8045.

95 Chen, R. J., Choi, H. C.,

Bangsaruntip, S., Yenilmez, E.,

Tang, X., Wang, Q., Chang, Y. L.,

Dai, H. An investigation of the

mechanisms of electronic sensing of

protein adsorption on carbon

nanotube devices., J. Am. Chem. Soc.2004, 126, 1563–1568.

96 Boussaad, S., Tao, N. J., Zhang, R.,

Hopson, T., Nagahara, L. A. In situdetection of cytochrome c adsorption

with single walled carbon nanotube

device.,Chem. Comm. 2003, 1502–1503.

97 Star, A., Gabriel, J.-C. P., Bradley,

K., Gruner, G. Electronic detection of

specific protein binding using

nanotube FET devices., Nano Lett.2003, 3, 459–463.

98 Chen, R. J., Bangsaruntip, S.,

Drouvalakis, K. A., Wong Shi Kam,

N., Shim, M., Li, Y., Kim, W., Utz,

P. J., Dai, H. Noncovalent

functionalization of carbon nanotubes

for highly specific electronic

biosensors., Proc. Natl. Acad. Sci.U.S.A. 2003, 100, 4984–4989.

99 Li, H., Ng, H. T., Cassell, A., Fan,

W., Chen, H., Ye, Q., Koehne, J.,

Han, J., Meyyappan, M. Carbon

nanotube nanoelectrode array for

ultrasensitive DNA detection, NanoLett. 2003, 3, 597–602.

100 Star, A., Tu, E., Niemann, J.,

Gabriel, J.-C. P., Joiner, C. S.,

Valcke, C. Label-free detection of

DNA hybridization using carbon

nanotube field-effect transistors, Proc.Natl. Acad. Sci. U.S.A. 2006, 103, 921–926.

101 Limdi, J. K., Crampton, J. R.

Hereditary haemochromatosis.,

Q. J. Med. 2004, 97, 315–324.102 Franchini, M., Veneri, D. Recent

advances in hereditary hemochromato-

sis., Ann. Hematol. 2005, 84, 347–352.103 Tang, X., Bansaruntip, S., Nakayama,

N., Yenilmez, E., Chang, Y.-I., Wang,

Q. Carbon nanotube DNA sensor and

sensing mechanism., Nano Lett. 2006,6, 1632–1636.

104 Maehashi, K., Matsumoto, K.,

Kerman, K., Takamura, Y., Tamiya,

E. Ultrasensitive detection of DNA

hybridization using carbon nanotube

field-effect transistors., Jpn. J. Appl.Phys. 2004, 43, L 1558–L 1560.

105 Star, A., Joshi, V., Han, T. R., Altoe,

V. P., Gruner, G., Stoddart, J. F.

Electronic detection of the enzymatic

degradation of starch., Org. Lett. 2004,6, 2089–2092.

106 Kong, J., Dai, H. Full and modulated

chemical gating of individual carbon

nanotubes by organic amine

compounds., J. Phys. Chem B 2001,

105, 2890–2893.

107 Shim, M., Javey, A., Kam, N. W. S.,

Dai, H. Polymer functionalization for

air-stable n-type carbon nanotube field-

effect transistors., J. Am. Chem. Soc.2001, 123, 11 512–11 513.

108 Besteman, K., Lee, J. O., Wiertz,

F. G. M., Heering, H. A., Dekker,

C. Enzyme-coated carbon nanotubes

as single-molecule biosensors., NanoLett. 2003, 3, 727–730.


Dec 2008

Why are we here ?

We are here to serve the students and

knowledge seekers throughout the world.

Infoshare

Comprehensive

knowledge

Source

A

Contact: [email protected]

Infoshare.silverforum.net

Http://Infoshare.silverforum.net

IMPLEMENTING NANO ROBOTS

IN

VENTILATION

&

IN BLOOD SUGAR DETECTION (mobile phones)

ABSTRACT

Nano-biotechnology is now becoming an emerging field that is

going to bring a lot of changes in the current century of technological revolution. It is a

one part of NANO-TECHNOLOGY. Apart from its participation in all fields, the part of

nanos in human science and medicine is large. Nanomedicine is the process of

diagnosing, treating, preventing disease and traumatic injury, of relieving pain, and of

preserving and improving human health, using molecular tools and molecular knowledge

of the human body.

Most symptoms such as fever and itching have specific biochemical

causes that can also be managed, reduced, and eliminated using the appropriate injected

nanorobots.

Our paper mainly concentrated on implementing Nano robots in

detecting human physiology. This paper mainly concentrates on implementing nano

robots in medical field. In this paper we have two ideas.

One is using nano robots to exhale oxygen and carbon dioxide according

to the human pressure. The nano robots are called as artificial red cells.



The second part of our paper deals with introducing nanosensors and

nanorobots in detecting Human blood sugar level. These nanorobots are Embedded

with mobile phones and the status of the patient can be read from remote places. These

nano particles that reduce the size of microelectronic components will become a major

part in human medicines, which may make this entire world to hide in a single chip.

APPLICATIONS:

* We could then hold our breath for nearly 4 hours if sitting quietly

at the bottom of a swimming pool.

* If we were sprinting at top speed, we could run for at least 15

minutes before we had to take a breath!

INTRODUCTION:

The term “nanotechnology” generally refers to engineering and manufacturing

at the molecular or nanometer length scale. (A nanometer is one-billionth of a meter,

about the width of 6 bonded carbon atoms.) . Nanotechnology will have given us

specially engineered drugs which are nanoscale cancer-seeking missiles, a molecular

technology that specifically targets just the mutant cancer cells in the human body, and

leaves everything else blissfully alone. To do this, these drug molecules will have to be

big enough – thousands of atoms – so that we can code the information into them of

where they should go and what they should kill. They will be examples of an exquisite,

human-made nanotechnology of the future. It is most useful to regard the emerging field

of nanomedicine as a set of three mutually overlapping and progressively more powerful

technologies. First, in the relatively near term, nanomedicine can address many

important medical problems by using nanoscale-structured materials that can be

manufactured today. This includes the interaction of nanostructures materials with

biological systems. Second, over the next 5-10 years, biotechnology will make possible

even more remarkable advances in molecular medicine and biorobotics (microbiological



robots), some of which are already on the drawing boards. . Third, in the longer term,

perhaps 10-20 years from today, the earliest molecular machine systems and nanorobots

may join the medical armamentarium, finally giving physicians the most potent tools

imaginable to conquer human disease, ill-health, and suffering. Our paper concentrates

mainly on our dream system that user nano sensor in mobile phones to detect human

blood sugar level and also nano robots in respiratory process. Most broadly,

nanomedicine is the process of diagnosing, treating, and preventing disease and traumatic

injury, of relieving pain, and of preserving and improving human health, using molecular

tools and molecular knowledge of the human body. “Over the past century we have

learned about the workings of biological nanomachines to an incredible level of detail,

and the benefits of this knowledge are beginning to be felt in medicine. In coming

decades we will learn to modify and adapt this machinery to extend the quality and length

of life.”

MAKING NANO ROBOTS:

The typical medical nanodevice will probably be a micron-scale robot assembled from

nanoscale parts. These parts could range in size from 1-100 nm (1 nm = 10-9

meter), and

might be fitted together to make a working machine measuring perhaps 0.5-3 microns (1

micron = 10-6

meter) in diameter. Three microns is about the maximum size for blood

borne medical nanorobots, due to the capillary passage requirement. Carbon will likely be

the principal element comprising the bulk of a medical nanorobot, probably in the form

of diamond or diamonded/fullerene nanocomposites largely because of the tremendous

strength and chemical inertness of diamond. Many other light elements such as hydrogen,



sulfur, oxygen, nitrogen, fluorine, silicon, etc. will be used for special purposes in

nanoscale gears and other components.

APPEARANCE OF NANO ROBOTS:

It is impossible to say exactly what a generic nanorobot would look like.

Nanorobots intended to travel through the bloodstream to their target will probably be

500-3000 nanometers (1 nanometer = 10-9 meter) in characteristic dimension. Non-blood

borne tissue-traversing nanorobots might be as large as 50-100 microns, and alimentary

or bronchial-traveling nanorobots may be even larger still. Each species of medical

nanorobot will be designed to accomplish a specific task, and many shapes and sizes are

possible.

In most cases a human patient who is undergoing a nanomedical treatment

is going to look just like anyone else who is sick. The typical nanomedical treatment (e.g.

to combat a bacterial or viral infection) will consist of an injection of perhaps a few cubic

centimeters of micron-sized nanorobots suspended in fluid (probably a water/saline

suspension). The typical therapeutic dose may include up to 1-10 trillion (1 trillion =

1012) individual nanorobots, although in some cases treatment may only require a few

million or a few billion individual devices to be injected. Each nanorobot will be on the

order of perhaps 0.5 micron up to perhaps 3 microns in diameter. (The exact size depends

on the design, and on exactly what the nanorobots are intended to do.) The adult human

body has a volume of perhaps 100,000 cm3 and a blood volume of ~5400 cm

3, so adding

a mere ~3 cm3 dose of nanorobots is not particularly invasive. The nanorobots are going

to be doing exactly what the doctor tells them to do, and nothing more (barring

malfunctions). So the only physical change you will see in the patient is that he or she

will very rapidly become well again. Most symptoms such as fever and itching have

specific biochemical causes which can also be managed, reduced, and eliminated using

the appropriate injected nanorobots. Major rashes or lesions such as those that occur



when you have the measles will take a bit longer to reverse, because in this case the

broken skin must also be repaired.

ARTIFICIAL RED CELL:

We named this Nanorobot as ventilons.The ventilons measures about 1

micron in diameter and just floats along in the bloodstream. It is a spherical nanorobot

made of 18 billion atoms. These atoms are mostly carbon atoms arranged as diamond in a

porous lattice structure inside the spherical shell. The ventilons is essentially a tiny

pressure tank that can be pumped full of up to 9 billion oxygen (O2) and carbon dioxide

(CO2) molecules. Later on, these gases can be released from the tiny tank in a controlled

manner. The gases are stored onboard at pressures up to about 1000 atmospheres.

(Ventilons can be rendered completely nonflammable by constructing the device

internally of sapphire, a flame proof material with chemical and mechanical properties

otherwise similar to diamond.). The surface of each ventilons is 37% covered with 29,160

molecular sorting rotors that can load and unload gases into the tanks. There are also gas

concentration sensors on the outside of each device. When the nanorobot passes through

the lung capillaries, O2 partial pressure is high and CO2 partial pressure is low, so the



onboard computer tells the sorting rotors to load the tanks with oxygen and to dump the

CO2. When the device later finds itself in the oxygen-starved peripheral tissues, the

sensor readings are reversed. That is, CO2 partial pressure is relatively high and O2 partial

pressure relatively low, so the onboard computer commands the sorting rotors to release

O2 and to absorb CO2.

Ventilons mimic the action of the natural hemoglobin-filled

red blood cells. But a ventilons can deliver 236 times more oxygen per unit volume than

a natural red cell. This nanorobot is far more efficient than biology, mainly because its

diamonded construction permits a much higher operating pressure. (The operating

pressure of the natural red blood cell is the equivalent of only about 0.51 atm, of which

only about 0.13 atm is deliverable to tissues.) So the injection of a 5 cm3 dose of 50%

ventilons aqueous suspension into the bloodstream can exactly replace the entire O2 and

CO2 carrying capacity of the patient's entire 5,400 cm3 of blood! Ventilons will have

pressure sensors to receive acoustic signals from the doctor, who will use an

ultrasound-like transmitter device to give the ventilons commands to modify their

behavior while they are still inside the patient's body. For example, the doctor might

order all the ventilons to just stop pumping, and become dormant. Later, the doctor might

order them all to turn on again.



APPLICATION:

By adding 1 liter of ventilons into our bloodstream, we could then hold

our breath for nearly 4 hours if sitting quietly at the bottom of a swimming pool. Or if we

were sprinting at top speed, we could run for at least 15 minutes before we had to take a

breath! It is clear that very "simple" medical nanodevices can have extremely useful

abilities, even when applied in relatively small doses. Other more complex devices will

have a broader range of capabilities. Some devices may have mobility the ability to swim

through the blood, or crawl through body tissue or along the walls of arteries. Others will

have different shapes, colors, and surface textures, depending on the functions they must

perform. They will have different types of robotic manipulators, different sensor arrays

and so forth. Each medical nanorobot will be designed to do a particular job extremely

well, and will have a unique shape and behavior.

OUR NANOSYSTEM TO DETECT HUMAN PHYSIOLOGY:

CARBON-DI-

OXIDE

&

OXYGEN

ARROW indicates

high pressure of

1000 atm.

CIRCLE indicates

nano particles.



Currently operate with micron sized active regions and offer the ability to do

thousands of measurements individual gene activities. Such arrays will allow hundreds of

thousands of human genes to be monitored throughout a mission and will allow the

determination of the effects of microgravity on human physiology in ways that are not

imagined at present, as well as providing early warning of cancer or other disease states.

By determining which genes are activated or inhibited, rack-mounted intelligent medical

systems will be able to apply preventative care at the earliest possible point.

Comprehensive cellular protein analysis and enzyme assays are equally feasible and

instructive. Nanotech-based gas chromatograph/mass spectrometer similar technologies,

such as a nanotech-based MS/MS, will allow the characterization and quantification of

multitudes of substances in a single small biological sample. In many cases, sensors will

be integrated with on-chip signal processing and data acquisition along with micro

fluidics and other sample transport and preparation technologies. Systems for sensing

biological and inorganic substances of interest in both aqueous and gaseous phases are

needed. Technologies such as micro-machined ion-mobility spectrometers, ion trap mass

spectrometers, calorimetric spectrometers, micro lasers and optics, on-chip separators,

optical spectrometers (e.g., UV, visible, and infrared), ultra sensitive acoustic wave

detectors, polymerase chain reaction (PCR) gene sequencing instrumentation (including

restriction enzyme digestion and PCR amplification) and many others could potentially

reside on the same chip or in close proximity allowing minute quantities of sample to

provide a wealth of information. The advantages of a laboratory on a chip include device

miniaturization for the space and volume restrictions of space travel, lower power

consumption, nearly instantaneous response times for near-real time results, conservation

of reagents, and ease of operation by non-laboratory personnel, such as astronauts. As

with many advances in nanotechnology, the chief difficulty may be in integrating these

many different units into functioning systems and interfacing them to the macro real

world.



NANOSENSORS IN MOBILEPHONES

System demonstration:

Our mobile system has small pins attached to the mobile

phones.

These pins help in taking samples of glucose.

From these samples the corpuscles are readed using the

small specific nanorobots inside the mobile.

Nano-chromatrons separate the glucose molecules which

cause diabetes.

o The molecules inhibited are read and compared with the other section and

the approximation is made about the sugar level.

o These sugar levels are compared with compressed DB, s and precautions

are displayed.

o By having sound sensors it may possible to calculate heartbeats & pulse

rates there by calculating the BP level.

DISPLAY

MOBILE

COMPONENTS

NANO SENSORS TO

DETECT PULSE RATE

&

CORPUSCLES

NANO ROBOTS TO

EXTRACT GLUCOSE CELLS

IN BLOOD

Pins to inject

robots



Nano robots used in our mobile phones

CONCLUSION:

Nanomedicine will eliminate virtually all common diseases of

the 20th century, virtually all medical pain and suffering, and allow the extension of

human capabilitiesmost especially our mental abilities. Consider that a nanostructured

data storage device measuring ~8,000 micron3, a cubic volume about the size of a

single human liver cell and smaller than a typical neuron, could store an amount of

information equivalent to the entire Library of Congress. If implanted somewhere in

the human brain, together with the appropriate interface mechanisms, such a device

could allow extremely rapid access to this information.

A single nanocomputer CPU, also having the volume of just

one tiny human cell, could compute at the rate of 10 teraflops (1013

floating-point

operations per second), approximately equalling (by many estimates) the

computational output of the entire human brain. Such a nanocomputer might produce



only about 0.001 watt of waste heat, as compared to the ~25 watts of waste heat for

the biological brain in which the nanocomputer might be embedded.

But perhaps the most important long-term benefit to human society as a

whole could be the dawning of a new era of peace. We could hope that people who

are independently well-fed, well-clothed, well-housed, smart, well-educated, healthy

and happy will have little motivation to make war. Human beings who have a

reasonable prospect of living many "normal" lifetimes will learn patience from

experience, and will be extremely unlikely to risk those "many lifetimes" for any but

the most compelling of reasons.



Toxicology Letters 176 (2008) 1–12

Available online at www.sciencedirect.com

Mini review

Nanosilver: A nanoproduct in medical application

X. Chen ∗, H.J. SchluesenerInstitute of Brain Research, University of Tuebingen Calwer Str. 3, D-72076 Tuebingen, Germany

Received 11 April 2007; received in revised form 8 October 2007; accepted 9 October 2007Available online 16 October 2007

Abstract

Nanotechnology is a most promising field for generating new applications in medicine. However, only few nanoproducts arecurrently in use for medical purposes. A most prominent nanoproduct is nanosilver. Nanosilver particles are generally smaller than100 nm and contain 20–15,000 silver atoms. At nanoscale, silver exhibits remarkably unusual physical, chemical and biologicalproperties. Due to its strong antibacterial activity, nanosilver coatings are used on various textiles but as well as coatings on certainimplants. Further, nanosilver is used for treatment of wounds and burns or as a contraceptive and marketed as a water disinfectantand room spray. Thus, use of nanosilver is becoming more and more widespread in medicine and related applications and dueto increasing exposure toxicological and environmental issues need to be raised. In sharp contrast to the attention paid to newapplications of nanosilver, few studies provide only scant insights into the interaction of nanosilver particle with the human bodyafter entering via different portals. Biodistribution, organ accumulation, degradation, possible adverse effects and toxicity are onlyslowly recognized and this review is focusing on major questions associated with the increased medical use of nanosilver and relatednanomaterials.© 2007 Elsevier Ireland Ltd. All rights reserved.

Keywords: Silver; Nanoparticles; Rout of entry; Toxicity; Mechanism

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12. General characteristics of silver nanoparticles and their entry portals into human body . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33. Nanosilver’s interactions with tissues and routes of exposure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

3.1. Respiratory system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33.2. Skin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53.3. Gastrointestinal tract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

4. Impacts of nanosilver on other tissues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75. Possible mechanisms of cytotoxicity of silver nanoparticles . .

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

∗ Corresponding author. Tel.: +49 7071 2984882;fax: +49 7071 294846.

E-mail address: [email protected] (X. Chen).

1

t“

0378-4274/$ – see front matter © 2007 Elsevier Ireland Ltd. All rights reservdoi:10.1016/j.toxlet.2007.10.004

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

. Introduction

Silver is a white and brilliant metallic element, posi-ioned 47th in the periodic chart with Ag, meaningargentum”, as its chemical symbol. Pure silver is ideally

ed.


dx.doi.org/10.1016/j.toxlet.2007.10.004

2 oxicolo

dtcgwarsfanusmdBmaoivpouru

emvatolainm“wp12leaat(Een

tetn(sdotibveevparaTvwimaTisiwelScwcM(iti(ctsati

X. Chen, H.J. Schluesener / T

uctile and malleable and has the highest electrical andhermal conductivity of any metal as well as the lowestontact resistance (Nordberg et al., 1988). Along withold, another rare and precious metal, silver has beenidely utilized for thousands of years in human history,

pplications including jewellery, utensils, monetary cur-ency, dental alloy, photography or explosives. Amongilver’s many applications, those exploiting its disin-ectant property for hygienic and medicinal purposesre time honoured and prominent, though the mecha-ism is not yet fully understood. Silver vessels weresed in ancient times to preserve water and wine andilver powder was believed by Hippocrates, father ofodern medicine, to have beneficial healing and anti-

isease properties and listed as a treatment for ulcers.ut it is mainly silver compounds that actually enterededical practice. Silver compounds were major weapon

gainst wound infection in World War I until the adventf antibiotics. In 1884 German obstetrician C.S.F. Credentroduced l% silver nitrate as an eye solution for pre-ention of Gonococcal ophthalmia neonatorum, which iserhaps the first scientifically documented medical usef silver (Russell and Hugo, 1994). Further, topicallysed silver sulfadiazine cream was standard antibacte-ial treatment for serious burn wounds and is still widelysed in burn units.

Irreversible pigmentation of the skin and/or theye, i.e. argyria or argyrosis, due to silver deposition,ay develop after prolonged exposure to silver or sil-

er compounds (Van de Voorde et al., 2005; Eturskand Obreshkova, 1979; Spencer et al., 1980). Due tohis problem and together with the advent of antibi-tics like penicillins and cephalosporins, silver’s lustreargely faded away as an anti-infection agent. However,dvancement of modern science has helped silver renewts lost lustre. Metallic silver is subjected to new engi-eering technologies with resultant extraordinarily novelorphologies and characteristics. Instead of being made

big”, metallic silver is engineered into ultrafine particleshose size is measured in nanometres (nm). When thesearticles have at least one dimension, which is less than00 nm, they are named nanoparticles (Oberdorster et al.,005b; Warheit et al., 2007). Upon reaching nanoscale,ike other nanomaterials and primarily by virtue ofxtremely small size, silver particles exhibit remark-bly unusual physicochemical properties and biologicalctivities. Great research efforts have been committed tohis respect and yielded exciting and encouraging results
Lee and El-Sayed, 2006; Evanoff and Chumanov, 2005;lechiguerra et al., 2005; Liu et al., 2006; Makaravat al., 2005). As a consequence, applications of engi-eered silver nanoparticles (nanosilver) especially in
rtav

gy Letters 176 (2008) 1–12

he healthcare sector have been and are being heatedlyxplored. “Silver nanoparticles are emerging as one ofhe fastest growing product categories in the nanotech-ology industry”, according to a market research reporthttp://www.bourneresearch.com). Still, the remarkablytrong anti-microbial activity is a major direction forevelopment of nanosilver products. A wide categoryf products in this respect has already been available onhe market. In medical arena, there are wound dress-ngs, contraceptive devices, surgical instruments andone prostheses all coated or embedded with nanosil-er (Cheng et al., 2004; Chen et al., 2006b; Muangmant al., 2006; Cohen et al., 2007; Lansdown, 2006; Zhangt al., 2007c). In daily life, consumers may have nanosil-er containing room spays, laundry detergents, waterurificants and wall paint (Cheng et al., 2004; Zhangnd Sun, 2007b). Silver nanoparticles are also incorpo-ated into textiles for manufacture of clothing, underwearnd socks (Lee et al., 2007; Vigneshwaran et al., 2007).here are washing machines, which employs nanosil-er (http://www.samsung.co.za/silvernano/silvernano/ash faq popup.html). More nanosilver products are

n the pipeline. It is estimated that of all the nano-aterials in medical and healthcare sector, nanosilver

pplication has the highest degree of commercialization.herefore, exposure to nanosilver in the body is becom-

ng increasingly widespread and intimate. Consequently,ilver in the form of nanoparticles has gained an increas-ng access to tissues, cells and biological moleculesithin the human body. The traditional belief is that

xcept for argyria or argyrosis and some minor prob-ems, silver is relatively non-toxic to mammalian cells.ilver poisoning only occurs among workers who havehronic history of silver exposure. Drake and Hazel-ood have reviewed health effects of silver and silver

ompounds from a perspective of occupational exposure.etallic silver was viewed to be a minimal health risk

Drake and Hazelwood, 2005). However, once reach-ng nanoscale, certain materials do exhibit significantoxicity to mammalian cells even if they are biochem-cally inert and biocompatible in bulk size, e.g. carbonMagrez et al., 2006; Shvedova et al., 2005). When itomes back to silver nanoparticles, in sharp contrasto the attention paid to their applications, only a fewtudies have provided very limited insights into suchspects as their entry portals into human body, biodis-ribution, organ accumulation as well as their potentialnteractions with tissues, cells and molecules and their
elevant toxicological implications. It is our opinionhat these are questions that need to be imperativelynswered before people rush to indulge into the nanosil-er boom.
http://www.bourneresearch.com/

http://www.samsung.co.za/silvernano/silvernano/wash_faq_popup.html

oxicolo

2itfnt

3r

3

aaciM22aRma2abcddric1ctcpswpfssssc2


2. General characteristics of silver nanoparticlesand their entry portals into human body

Silver nanoparticles have been synthesized throughan array of methods, e.g. spark discharging, electrochem-ical reduction, solution irradiating and cryochemicalsynthesis, to name a few (Sun and Xia, 2002; Zhanget al., 2002; Bogle et al., 2006; Sergeev et al., 1999;Pyatenko et al., 2004). Particle morphologies includespheres, rods, cubes, wires and multifacets, normallywithin a size range of <100 nm. As is the case withall nanomaterials, the principle characteristic of silvernanoparticles is their ultrasmall size. Ultrasmall parti-cle size leads to ultralarge surface area per mass wherea large proportion of atoms are in immediate contactwith ambiance and readily available for reaction. Uniqueinteractions with bacteria and virus have been demon-strated of silver nanoparticles of certain size ranges andshapes (Elechiguerra et al., 2005; Jose et al., 2005; Loket al., 2006). Small size also confers greater particlemobility both in the environment and in the body. Fur-ther, nanoparticles produced through different processesand for different purposes may vary in surface chargeand agglomeration state. Some silver nanoparticles havecoatings and others are hybridized with other materialsto form nanocomposites (Kobayashi et al., 2005; Lesniaket al., 2005). In addition, nanoparticle colloids may needdifferent stabilizers. All these combined may probablymodify the intrinsic physiochemical properties of silverand may therefore give rise to modify cellular uptake,interaction with biological macromolecules and translo-cation within the human body. Adverse reactions canoccur that would not otherwise be seen with silver inbulk form.

The human body has several semi-open interfacesfor direct substance exchange with the environment,i.e. the respiratory tract, gastrointestinal tract (GIT) andskin. In view of nanosilver’s diverse forms of existence,they are also the principle routes of exposure to silvernanoparticles. Female genital tract is also an entrywayof potential importance since nanosilver has been incor-porated into maternal hygiene products (Cheng et al.,2004; Zhang and Sun, 2007b). At these sites, nanoparti-cles can undergo a series of processes like binding andreacting with proteins, phagocytosis, deposition, clear-ance and translocation. On the other hand nanoparticlescan elicit a spectrum of tissue responses such as cellactivation, generation of reactive oxygen species (ROS),
inflammation and cell death (Chen et al., 2006a; Xiaet al., 2006; Ahn et al., 2005). Additionally, colloidalnoble metal nanoparticles have been exploited for diag-nostic imaging or therapeutic purposes (West and Halas,
paaa

gy Letters 176 (2008) 1–12 3

003; Lee and El-Sayed, 2006). Thus, systemic admin-stration is also a potential route of entry. Nanoparticleshat have been systemically administered or translocatedrom other portals have direct contact with blood compo-ents, and the cardiovascular system, and are distributedhroughout the body.

. Nanosilver’s interactions with tissues andoutes of exposure

.1. Respiratory system

The respiratory system serves as a major portal formbient particulate materials. Pathologies resulting fromirborne particle materials, e.g. quartz, asbestos andarbon, have long been topics thoroughly researchedn occupational and environmental medicine (Alfaro-

oreno et al., 2007; Kanj et al., 2006; Gillissen et al.,006; Lam et al., 2006; Ovrevik et al., 2005; Warheit,001a; Donaldson et al., 2001; Parks et al., 1999). Usu-lly the sizes of these particles are at micron level.ecently, the pathogenic effects and pathology of inhaledanufactured nanoparticles received attention (Nel et

l., 2006; Oberdorster et al., 2005a; Donaldson et al.,006; Lam et al., 2006). Being different than micron andbove level particles that are largely trapped and clearedy upper airway mucocilliary escalator system, parti-les less than 2.5 �m can get down to the alveoli. Theeposition of inhaled ultrafine particles (aerodynamiciameter <100 nm) mainly takes place in the alveolaregion. Healthcare and hygiene spray products contain-ng silver nanoparticles have entered daily use. Mostommercialized silver nanoparticles are usually less than00 nm, way far under the 2.5 �m size. Of great con-ern are silver nanoparticle aerosol directly applied intohe nasal or oral cavity, as concentrated nanoparticlesan be channelled into the lungs. At the alveolar region,articles will first encounter and be submersed into theurfactant lining of the alveoli before having contactith any cell. The submersion process seems to takelace regardless of particle type and nature of the sur-ace (Geiser et al., 2003; Gehr et al., 2000). Results oftudies using different types of particles suggests thaturfactant dipalmitoylphosphatidylcholine (DPPC) andurfactant protein (SP-D) are absorbed to the particleurface, which may be a mediation mechanism for toxi-ity of particulate matters (Liu et al., 1998; Kendall et al.,004; Gerber et al., 2006). The surface structures of silica
articles are found to interact with the lining fluid layernd produce surface radicals and ROS which are associ-ted with silica particle’s specific toxicity (Fubini, 1997)nd particle–surfactant lining interaction influences sub-

4 oxicolo

saTitoaseetlaaaIesc

imptfoTtlw1atptftcmIpc2WltiFoRiB

lf(asa7nratoTbtcdefiim(2I

caaSorbiaaieamuedtshenu


equent particle–cell interaction (Fubini, 1997; Emersonnd Davis, 1983; Bridges et al., 2000; Peters et al., 2006).oxicological data on silver nanoparticles in this respect

s scarcely available. But an important fact must be notedhat silver, a member of the transition metal family, has anxidation state, which makes it useful as a catalyst. Actu-lly high catalysis activity for oxidation has been demon-trated of novel silver nanoparticles based catalysts (Zhait al., 2006; Shen et al., 2004). In some recent studiesxposure of nanoparticles of various chemical composi-ion were found or indicated to induce oxidative stress inung epithelial cells (Kaewamatawong et al., 2006; Koikend Kobayashi, 2006; Limbach et al., 2007; Sharma etl., 2007). Such damaging effects of nanoparticles weressociated with catalytic activity (Limbach et al., 2007).t is sound to assume, that in such a highly pro-oxidativenvironment as the intra-alveolar space, the enormousurface area of silver nanoparticles may serve as an effi-ient facilitator of generation of radicals and ROS.

The lungs have defence mechanisms to eliminatenhaled solid particles so that particle deposition nor-

ally will not achieve a lung burden that is sufficient toroduce respiratory pathogenic effects. This is true whenhe inhaled particles do not interfere with the eliminationunction (Lippmann et al., 1980). Particles in the alve-lar region are eliminated through several major routes.he first is mucocilliary escalator transport along the

racheobronchial tree, the second translocation into theymphatic system and the third is particle dissolutionith subsequent transfer into the blood (Oberdorster,988; Takenaka et al., 2000). Alveolar macrophages arekey component of the clearance mechanisms. They are

he chief cells that are recruited to confront depositedarticles and their capacity for phagocytosis as well asheir response to the phagocytic stimulus determines theate of the particles (Brain, 1992). Deposited nanopar-icles, partly through interaction with lung epithelialells, could induce rapid and increased recruitment ofacrophages (Renwick et al., 2004; Barlow et al., 2005).

n addition, it has been shown that inhaled ultrafinearticles, such as carbon and TiO2, impair the phago-ytic function of alveolar macrophages (Renwick et al.,001, 2004; Lundborg et al., 2001; Moller et al., 2002;infried et al., 2005). Functionally impaired alveo-

ar macrophages could favor retention of particles inhe lungs, resulting in building up of a toxic dose;nflammation could be stimulated (Warheit et al., 1997).urther, phagocytosis of particles can lead to activation
f macrophages and release of chemokines, cytokines,OS, and other mediators which may result in sustained
nflammation (Kang et al., 2005; Hubbard et al., 2002;rown et al., 2004). High inflammogenicity has been

Tati

gy Letters 176 (2008) 1–12

inked to ultrafine particles’ small sizes and large sur-ace area with associated free radical generating systemsXia et al., 2006; Sayes et al., 2006). Both inhalationnd instillation experiments have shown that ultrafineilver particles are taken up by alveolar macrophages andggregated silver particles persist there at least for up todays (Takenaka et al., 2001). And aggregated silver

anoparticles and some other nanomaterials have beeneported to be cytotoxic to alveolar macrophage cellss well as epithelial lung cells (Soto et al., 2007). Yethe impact mechanisms of silver nanoparticles on alve-lar macrophages’ function have yet to be elucidated.he respiratory impacts of inhaled nanoparticles need toe addressed in a more complicated context with regardo a morbid respiratory system with basal inflammatoryonditions, e.g. asthma, chronic obstructive pulmonaryiseases or respiratory infection. Since there is amplevidence that inhalation of particles (engineered ultra-ne particles included) induces/aggravates respiratory

nflammations and epithelial damage, in which alteredacrophage responses could play an important role

Renwick et al., 2004; Dick et al., 2003; Lundborg et al.,006; Schaumann et al., 2004; Ken-Ichiro et al., 2006;noue et al., 2006, 2007).

Significant alveolar-capillary translocation via endo-ytosis and transcytosis has been suggested by animalnd human studies (Takenaka et al., 2001; Kreyling etl., 2002; Oberdorster et al., 2002; Kato et al., 2003;himada et al., 2006) and the impact of inhaled particlesn other organs has received recognition. For instance,espiratory exposure to single-wall carbon nanotubes haseen suggested to provoke not only pulmonary toxic-ty but vascular effects in laboratory mice (Zheng etl., 2007). Cardiovascular events such as coagulationnd cardiac rhythm disturbances were associated withnhalation of ambient ultrafine particles (Nurkiewiczt al., 2006; Yeates and Mauderly, 2001; Brook etl., 2004). Takenaka et al. (2001) examined the pul-onary and systemic distribution of inhaled and instilled

ltrafine silver particles (14.6 ± 1.0 nm) in rats. Theirxperiment showed that the silver content in the lungecreases rapidly with time after inhalation of a rela-ively low concentration of ultrafine silver which wasubsequently detected in blood and other organs such aseart, liver, kidney, and even brain. These results providevidence for penetration and circulation of inhaled silveranoparticles. As circulating nanoparticles can be takenp by other organs, many systemic effects might occur.
he authors especially mentioned that at each checkpointsignificant portion (9–21%) of silver was observed in
he liver. This indicates that liver might play a major rolen clearance of circulatory silver nanoparticles.

oxicolo

w2b2tt2ma

cidafllirbrefuttaldn

sadDd(2hPticciecos


3.2. Skin

Through various techniques, textile fibers can becoated or impregnated with silver nanoparticles (Lee etal., 2007; Vigneshwaran et al., 2007). These textiles arecalled “smart textiles” which are claimed to have advan-tages over normal textiles having the ability to inhibitgrowth of bacteria and mold. These anti-microbiallyactive textiles have been employed for manufacturingof underwear, lingerie, socks as well as hospital andlaboratory gowns and clothes. Nanosilver is gaining intextiles and there is an increase in interest due to itsclose contact with human skin. One of skin’s majorroles is to provide protection to the underlying organs.It consists of an outer epidermis and dermis. The stra-tum corneum layer of the epidermis is a strict barrierallowing limited penetration of particulate materials.This aspect has potential to serve as novel route fordrug delivery and has attracted enormous pharmaceu-tical research interests (Shim et al., 2004; Lopez et al.,2005; Kohli and Alpar, 2004). Transdermal penetrationof fine particles has been documented. For example,TiO2 particles (micron-sized), a common ingredient insunscreen products, are described to get through thehuman stratum corneum to epidermis and even reachdermis (Lademann et al., 1999). It was also reported thatflexing movement of normal skin facilitated the pen-etration of micrometer-size fluorescent beads into thedermis (Tinkle et al., 2003). Yet, data on nanosilverare few to none. For each kind of nanosilver-based tex-tile, the release of nanoparticles from the textile fibersunder various conditions, e.g. sweating, repetitive attri-tion and laundering needs to be investigated. Appropriatemodels should be employed to assess the possibilityof transdermal penetration of silver nanoparticles sinceseveral recent studies have reported transdermal penetra-tion of nanoparticles. Ryman-Rasmussen et al. (2006)demonstrated that quantum dots with diverse physico-chemical properties could penetrate the intact stratumcorneum barrier and localize within the epidermal anddermal layers. Fullerene-based peptides were also shownto be capable of penetrating intact skin and mechanicalstressors could facilitate their traversion into the dermis(Rouse et al., 2007). Intradermal nanoparticles couldenter subcutaneous lymphatics (Gopee et al., 2007).Damaged skin also allows for micron-sized particles togain access to the dermis and regional lymph nodes (Kimet al., 2004). This along with the observation that parti-
cles in the skin can be phagocytized by macrophages andLangerhans cells could possibly lead to perturbations ofthe immune system (Tinkle et al., 2003). A case of acti-vation of mast cells by silver nanoparticles in a patient
ckta

gy Letters 176 (2008) 1–12 5

ith localized argyria has been reported (Kakurai et al.,003). In vitro mast cells were found to be activatedy silver salts (Suzuki et al., 2001; Yoshimaru et al.,006). Epidermal keratinocytes have also been showno be capable of phagocytosing nanoparticles and set-ing off inflammatory responses (Monteiro-Riviere et al.,005). Mechanical irritation and interference with der-al resident microflora by nanosilver-based fibers might

lso pose potential problems.Disposition of transdermally penetrated nanoparti-

les is not clearly understood. Yet there is evidence thatntradermal nanoparticles could gain access to systemicistribution through subcutaneous lymphatics. Gopee etl. (2007) intradermally injected hairless mice with UVuorescent quantum dots which, within minutes fol-

owing injection, could be observed moving from thenjection sites through the lymphatic duct system toegional lymph nodes. While intact skin is shown toe pervious to nanoparticles, skin wounds may giveise to easier translocations. Dressings and bandagesmbedded or coated with silver nanocrystallines are usedor prevention of sepsis of skin wounds like burns andlcers. Close contact may allow nanoparticles to pene-rate through compromised skin barrier and gain accesso the dermal capillaries. One clinical report describedbnormal elevation of blood silver level and argyria-ike symptoms following the use of nanosilver coatedressings for burns (Trop et al., 2006). Absorption ofanosilver into the circulation has been indicated.

Though nanosilver-based dressings and surgicalutures have received approval for clinical applicationnd good control of wound infection is achieved, theirermal toxicity is still a topic of dispute and concern.espite laboratory and clinical studies confirming theermal biocompatibility of nanosilver-based dressingsWright et al., 2002; Supp et al., 2005; Chen et al.,006b; Muangman et al., 2006) several other researchesave demonstrated the cytotoxicity of these materials.addles-Ledinek et al. exposed cultured keratinocytes

o extracts of several types of silver containing dress-ngs. The results showed that extracts of nanocrystallineoated dressings are among those, which are the mostytotoxic. Keratinocyte proliferation was significantlynhibited and cell morphology affected (Paddle-Ledinekt al., 2006). In vitro, keratinocyte viability also pre-ipitously dropped after direct contact with one typef nanosilver dressing (Lam et al., 2004). In a thirdtudy, nanosilver crystallines released from a commer-
ially available dressing were found to be toxic to botheratinocytes and fibroblasts, and fibroblasts appearedo be more sensitive to silver than keratinocytes (Poonnd Burd, 2004). These obvious discrepancies might

6 oxicolo

baatmevtaptoaa

cdhktea2e

3

ganrlamaftcsdfolctpilptc

1a

etmalfilttet2lsaafdsTpiacmiibcabpatlTp

erbspob


e attributable to difference in laboratory conditionsnd techniques employed. Thus, the establishment of

set of integrated and standardized evaluation pro-ocols is necessary. Further, it has to be noted that

ost experiments have been carried out by vitro mod-ls in which cells may behave differently from inivo conditions. In clinical situations, wound exuda-ion might provide as a barrier against nanosilver’sdverse effect on wound healing. The high content ofrotein exudates may probably neutralize nanosilver’sissue toxicity. Still, critical observations through rigor-usly designed clinical studies can provide substantialnd matter-of-fact data for comprehensive toxicologicalssessments.

It is worth noting that some other types of nanoparti-les, i.e. single-/multi-wall carbon nanotubes, quantumots with surface coating, and nanoscale Titania,ave been shown to have toxical effects on epidermaleratinocytes and fibroblasts or be capable of alteringheir gene/protein expression (Ding et al., 2005; Sarkart al., 2007; Ryman-Rasmussen et al., 2007; Zhang etl., 2007a; Tian et al., 2006; Monteiro-Riviere et al.,005; Witzmann and Monteiro-Riviere, 2006; Christiet al., 2006).

.3. Gastrointestinal tract

All materials given orally are in close contact with theastrointestinal tract (GIT) which has an overall surfacerea up to 200 m2 for nutrient exchange. Gastrointesti-al ingestion is probably the most common voluntaryoute of exposure for nanosilver since numerous col-oidal silver nanoparticle products are publicly peddleds so called “health maintainers” or “immuno-boosters”;ost of them are used orally. Silver nanoparticles are

lso employed in products for water disinfection andood stabilisation (Zhang and Sun, 2007b). Besides, par-icles discharged by respiratory mucocilliary escalationan end up in GIT. But despite extensive GIT expo-ure, apart from occasional cases of systemic argyriaue to prolonged ingestion and disturbance of intestinalunction, reports on local or systemic adverse effects ofrally ingested nanosilver are remarkably few. Neverthe-ess, the occurrence of systemic argyria after ingestion ofolloidal nanosilver in itself is an evidence that transloca-ion of silver nanoparticles from the intestinal tract takeslace. The kinetic mechanism of nanosilver translocations unclear, but it has been demonstrated that the intestinal
ymphatic tissue (Peyer’s patches) can take up intestinalarticles (Smith et al., 1995). Uptake can also take placerans-cellularly via normal enterocytes and through para-ellular pathways (Jani et al., 1990; Aprahamian et al.,
nwot

gy Letters 176 (2008) 1–12

987). Particle characteristics like size, surface charge,nd coatings can modify the translocation process.

Particles once in the submucosal region are able tonter both lymphatics and capillaries. Lymphatic absorp-ion may give rise to immune response, for instance the

ucosal secretory immune function may probably beffected. At the same time those particles entering capil-aries become circulatory and will soon encounter theirrst pass, i.e. liver. Based on the extent to which col-

oidal nanosilver is orally used, it is a logical assumptionhat ingested silver nanoparticles might have impact onhe liver since the liver serves as the first checkpoint forverything absorbed through GIT before becoming sys-emic. As has been shown in the study by Takenaka et al.,001 the liver seems to be a major depository of circu-atory ultrafine silver particles. Further, toxic effects ofilver nanoparticles to liver cells have been reported fromn in vitro experiment (Hussain et al., 2005). In addition,t least one clinical report has associated impaired liverunction to silver nanoparticles released from a woundressing (Trop et al., 2006). As stated earlier, however,ystemic toxicity of ingested nanosilver is scarcely seen.his situation may probably be accounted for by theresence in the GIT of a complex mixture of compoundsncluding ingested food, digestive enzymes, electrolytes,nd intestinal microbial flora, etc. Ingested nanoparti-les can have interactions with these compounds, whichight change reactivity and toxicity of the particles. This

s of particular relevance to silver, as it is well known forts high affinity for the thiol groups of proteins. It haseen described that medium high protein concentrationould lessen the in vitro cytotoxicity of nanoparticlesnd nanosilver’s antibacterial activity could be blockedy thiol containing agents. Apart from that, GIT ingestedarticles will undergo sequential pH stress from gastriccid and intestinal fluids. Particle surface characteris-ics may be modified by the shift in ambient pH, whicheads to altered solubility and ion state of the particles.he renewal of the epithelium also hinders nanoparticlesenetration through the intestinal wall.

Considering the rare occurrence of system adverseffects of orally ingested silver nanoparticles, moreesearch is still needed on their kinetics in the GIT for aetter safety assessment of GIT as a major route of expo-ure. It is noteworthy that recently nano- and micro-sizedarticles were found to induce granulomas in differentrgans and tissues. In an histological analysis reportedy Gatti et al., inorganic particles, heterogeneous in
ature but homogeneous in size (nano- to micro-size)ere identified in livers and kidneys with “granulomasf unknown origin” (Gatti and Rivasi, 2002), and colonissues affected by cancer and Crohn’s disease were also

oxicolo

tiacapied

cLifgbslc(stocvtvshtgverimstvib(1

5n


found to contain micro- and nanoparticles (Gatti et al.,2004). Most of the particles are debris of materials tradi-tionally considered to be biocompatible, silver included.But clearly the presence of these particles is linked topathologies. The authors suggest that material whichis accepted in bulk form can become less biocompati-ble when its size is reduced below a certain “critical”threshold.

4. Impacts of nanosilver on other tissues

Haemo-compatibility is a top safety concern for sys-temically administrated materials. Nanosized inorganic,non-biodegradable particles were detected in bloodand were associated with thrombosis and activation ofimmunological reactions (Gatti et al., 2005). Further,exposure to ambient ultrafine particles has been associ-ated with cardiovascular morbidity and mortality (Peterset al., 2004; Pope et al., 2004; Samet et al., 2000). Studieshave provided evidence that exposure to ambient ultra-fine particles elicits inflammatory responses in vascularendothelial cells, blood cells, i.e. leukocytes, platelet andinduces changes in other blood components (Andrea etal., 2007; Mark et al., 2006; Regina et al., 2007). For sil-ver, interactions of metal colloids with erythrocytes werestudied by Garner et al. (1994). It was observed that treat-ment of intact erythrocyte samples with citrate coatedcolloidal silver particles induced a large depletion ofintracellular glutathione through a mechanism differentthan oxidization shown with silver nitrate. In cell lysates,solute species exchanged glutathione with citrate on thenanosilver particle surface. The specific mechanism ofinteraction between haemoglobin and silver nanoparti-cles is not known. But a recent study might providesome clues for further investigation: Gan et al. (2004)revealed that silver nanoparticles could greatly enhancethe electron-transfer reactivity of myoglobin which isclosely related to haemoglobin and often used as a modelfor study on heme proteins. Interactions with other bloodcomponents such as platelets and plasma proteins havealso to be examined before silver nanoparticles wereintravascularly injected, especially for diagnostic imag-ing purposes.

Recently the identification of cytotoxicity of nanopar-ticles towards mammalian germline stem cells hasaroused great concern over the biosafety of nanoma-terials (Braydich-Stolle et al., 2005). In their study,Braydich-Stolle et al. utilized a cell line with spermato-
gonial stem cell characteristics to test the in vitro toxicityof several types of nanoparticles. The results showedthat of all the tested materials (Ag, MoO3 and Al),silver nanoparticles were the most toxic with manifes-
tgrs

gy Letters 176 (2008) 1–12 7

ations like drastic reduction of mitochondrial function,ncreased membrane leakage, necrosis and induction ofpoptosis. The findings are of significant practical impli-ations because silver nanoparticles are now able toccess human sperms via a variety of commercializedroducts like contraceptive devices and maternal hygienetems. Fertility problems may occur. In addition, as a fairxtrapolation, another question emerged: what will theyo to egg cells?

Liver appears to be a major accumulation site of cir-ulatory silver nanoparticles (Takenaka et al., 2001).ike germ line stem cells, similar patterns of cytotox-

city of silver nanoparticles (decrease of mitochondriaunction, LDH leakage and abnormal cell morpholo-ies) were observed with in vitro BRL 3A rat liver cells,ut to a lesser extent (Hussain et al., 2005). In anothertudy by the same researchers, a neuronendocrine celline (PC-12 cells) was exposed to silver nanoparti-les as a control against Mn nanoparticles and Mn2+

Hussain et al., 2006). Experimental results showed thatilver nanoparticles were more toxic to mitochondriahan Mn nanoparticles and Mn2+. These findings aref importance, because considerable amount of silverould be detected in rat brain following inhalation of sil-er nanoparticles (Takenaka et al., 2001). The scope ofhe concerns raised goes beyond the possibility of sil-er nanoparticles penetrating across blood brain barrierince results of experiments using different nanoparticlesave suggested that inhaled nanoparticles are very likelyo deposit in the olfactory mucosa of the nasopharyn-eal region and subsequently be translocated into brainia the olfactory nerve (Oberdorster et al., 2004; Eldert al., 2006). A re-examination of Takenaka et al. (2001)eport revealed that both immediately and 1 day follow-ng inhalation Ag concentration in olfactory region was

uch higher than in the rest of brain. Whether inhaledilver nanoparticles could enter brain through transloca-ion via the olfactory nerve is a topic of interest awaitingalidation. The neurological toxicity of silver is not clin-cally ascertained, however, several seizure cases haveeen related to exposure to silver or silver compoundsMirsattari et al., 2004; Ohbo et al., 1996; Iwasaki et al.,997).

. Possible mechanisms of cytotoxicity of silveranoparticles

After reviewing the literature a fact is noticeable
hat for the limited number of cell lines tested (C18-4erm cell line, BRL 3A liver cell line and PC-12 neu-oendocrine cell line) exposure to silver nanoparticlesignificantly decreased the function of mitochondria.

8 oxicolo

Tape2tagmd3RttAckraCfimb(atota(2

ftctcnsetaiotd

mmdp

itDpmAnbsneiRDTeuwann

R

A

A

A

A

A

A

B

B


his is a shared characteristic of all cellular responses,nd apoptosis or apoptosis-like change of cell mor-hology also occurred in all three cell lines followingxposure (Braydich-Stolle et al., 2005; Hussain et al.,005, 2006). It has been well established that dysfunc-ion of mitochondria is an early and key step towardspoptosis. Thus, mitochondria seem to be sensitive tar-ets of cytotoxicity of silver nanoparticles. However, theechanism of silver nanoparticles’ action on mitochon-

rion is yet to be elucidated. In the study with BRLA liver cell line, depletion of GSH level and increasedOS was found in association with mitochondrial per-

urbation, suggesting that oxidative stress might mediatehe cytotoxicity of silver nanoparticles. Researches ong+ seem to be able to lend some support to this view,

oncerning mitochondrial oxidative stress. As a well-nown thiol group reactive cytotoxin, Ag+ has beeneported to be able to cause mitochondria perturbationnd disrupt mitochondrial function (Kone et al., 1988;happell and Greviller, 1954). Recently, it has been

ound that Ag+ seems to perturb mitochondria throughnteractions with thiol groups of the mitochondrial inner

embrane. As these effects of Ag+ could be completelylocked by sulfhydryl reagents, e.g. reduced glutathioneGSH), the findings clearly suggest that mitochondriare under oxidative stress when the cells are exposedo Ag+ (Almofti et al., 2003). On the other hand, datan silver nanoparticles’ anti-microbial activity suggesthat silver nanoparticles exert their anti-microbial effectslso through the interaction with proteins thiol groupsElechiguerra et al., 2005; Jose et al., 2005; Andre,005).

In conclusion, a preliminary impression can beormed that silver nanoparticles may interact with pro-eins and enzymes with thiol groups within mammalianells. These proteins and enzymes like glutathione,hioredoxin, SOD and thioredoxin peroxidase, are keyomponents of the cell’s antioxidant defense mecha-ism which is responsible to neutralize the oxidativetress of ROS largely generated by mitochondrialnergy metabolism. Silver nanoparticles may depletehe antioxidant defense mechanism, which leads to ROSccumulation. Over accumulation of ROS can initiate annflammatory response and perturbation and destructionf the mitochondria take place. Then apoptogenic fac-ors like cytochrome C are released and programmed celleath is a final result.

Besides mitochondria destruction, damage to cell
embranes appears to be another part of nanosilver’sechanism of cytotoxicity that precedes mitochon-
rial perturbation. Theoretically, nanoparticles need toenetrate across the cell membrane to get access to

B

B

gy Letters 176 (2008) 1–12

ntracellular organelles. It is known that thiol-group con-aining proteins are abundant in the cell membrane.etrimental protein–nanosilver interactions are highlyossible and lipoperoxidation may play some role. Theechanisms mentioned above seem to be shared byg+. Truly, at least Ag+ could be released from silveranoparticles. Yet, silver’s catalysis potential should note underestimated. In view of the tremendous per massurface area of nanoparticles, it is arbitrary to decide thatanosilver would not churned out plethora of ROS. Forxample, the proinflammatory effects of nanoparticlesn the lung are directly related to their surface area andOS-generating capability (Oberdorster et al., 2005b;onaldson and Tran, 2002; Donaldson et al., 2004).aking into account their unique physicochemical prop-rties, it is unlikely that nanoparticles do not possessnique toxicity mechanisms. It remains to be determinedhether silver nanoparticles as well as other nanomateri-

ls will introduce new mechanisms of injury from whichew pathologies may result. Finally for silver, whetherano-sized or not, there is always the problem of argyia.

eferences

hn, M.H., Kang, C.M., Park, C.S., Park, S.J., Rhim, T., Yoon, P.O.,Chang, H.S., Kim, S.H., Kyono, H., Kim, K.C., 2005. Titaniumdioxide particle-induced goblet cell hyperplasia: association withmast cells and IL-13. Respir. Res. 13 (6), 34.

lfaro-Moreno, E., Nawrot, T.S., Nemmar, A., Nemery, B., 2007. Par-ticulate matter in the environment: pulmonary and cardiovasculareffects. Curr. Opin. Pulm. Med. 13, 98–106.

lmofti, M.R., Ichikawa, T., Yamashita, K., Terada, H., Shinohara, Y.,2003. Silver ion induces a cyclosporine a-insensitive permeabil-ity transition in rat liver mitochondria and release of apoptogeniccytochrome C. J. Biochem. 134, 43–49.

ndre, N., 2005. Air pollution-related illness: effects of particles. Sci-ence 6 (308), 804–806.

ndrea, G., Bing, G., Rama, S.K., John, C.R., Ian, M.K., Abdul, I.B.,2007. Induction of inflammation in vascular endothelial cells bymetal oxide nanoparticles: effect of particle composition. Environ.Health Perspect. 115, 403–409.

prahamian, M., Michel, C., Humbert, W., Devissaguet, J.P.,Damge, C., 1987. Transmucosal passage of polyalkylcyanoacry-late nanocapsules as a new drug carrier in the small intestine. Biol.Cell 61, 69–76.

arlow, P.G., Clouter-Baker, A., Donaldson, K., Maccallum, J., Stone,V., 2005. Carbon black nanoparticles induce type II epithelial cellsto release chemotaxins for alveolar macrophages. Part Fibre Toxi-col. 2, 11.

ogle, K.A., Dhole, S.D., Bhoraskar, V.N., 2006. Silver nanoparticles:synthesis and size control by electron irradiation. Nanotechnology17, 3204–3208.

rain, J.D., 1992. Mechanisms, measurement, and significance of lungmacrophage function. Environ. Health Perspect. 97, 5–10.

raydich-Stolle, L., Hussain, S., Schlager, J.J., Hofmann, M.C., 2005.In vitro cytotoxicity of nanoparticles in mammalian germline stemcells. Toxicol. Sci. 88, 412–419.

oxicolo

E

E

E

E

F

G

G

G

G

G

G

G

G

G

G

H

H

H

I


Bridges, J.P., Davis, H.W., Damodarasamy, M., Kuroki, Y., Howles, G.,Hui, D.Y., McCormack, F.X., 2000. Pulmonary surfactant proteinsA and D are potent endogenous inhibitors of lipid peroxidation andoxidative cellular injury. J. Biol. Chem. 275, 38848–38855.

Brook, R.D., Franklin, B., Cascio, W., Hong, Y., Howard, G., Lipsett,M., 2004. Air pollution and cardiovascular disease: a statement forhealthcare professionals from the Expert Panel on Population andPrevention Science of the American Heart Association. Circulation109, 2655–2671.

Brown, D.M., Donaldson, K., Borm, P.J., Schins, R.P., Dehnhardt,M., Gilmour, P., Jimenez, L.A., Stone, V., 2004. Calcium andROS-mediated activation of transcription factors and TNF-alphacytokine gene expression in macrophages exposed to ultrafineparticles. Am. J. Physiol. Lung Cell. Mol. Physiol. 86, L344–L353.

Chappell, J.B., Greviller, G.D., 1954. Effect of silver ions on mito-chondrial adenosine triphosphatase. Nature 174, 930–931.

Chen, H.W., Su, S.F., Chien, C.T., Lin, W.H., Yu, S.L., Chou,C.C., Chen, J.J., Yang, P.C., 2006a. Titanium dioxide nanoparti-cles induce emphysema-like lung injury in mice. FASEB J. 20,2393–2395.

Chen, J., Han, C.M., Lin, X.W., Tang, Z.J., Su, S.J., 2006b. Effect of sil-ver nanoparticle dressing on second degree burn wound. ZhonghuaWai Ke Za Zhi 44, 50–52.

Cheng, D., Yang, J., Zhao, Y., 2004. Antibacterial materials of silvernanoparticles application in medical appliances and appliances fordaily use. Chin. Med. Equip. J. 4, 26–32.

Christie, M.S., Wahi, R., Preetha, A.K., Liu, Y., Jennifer, L.W., Kevin,D.A., David, B.W., Vicki, L.C., 2006. Correlating nanoscale titaniastructure with toxicity: a cytotoxicity and inflammatory responsestudy with human dermal fibroblasts and human lung epithelialcells. Toxicol. Sci. 92, 174–185.

Cohen, M.S., Stern, J.M., Vanni, A.J., Kelley, R.S., Baumgart, E., Field,D., Libertino, J.A., Summerhayes, I.C., 2007. In vitro analysisof a nanocrystalline silver-coated surgical mesh. Surg. Infect. 8,397–403.

Dick, C.A., Brown, D.M., Donaldson, K., Stone, V., 2003. The role offree radicals in the toxic and inflammatory effects of four differentultrafine particle types. Inhal. Toxicol. 15, 39–52.

Ding, L., Stilwell, J., Zhang, T., Elboudwarej, O., Jiang, H., Selegue,J.P., Cooke, P.A., Gray, J.W., Chen, F.F., 2005. Molecular char-acterization of the cytotoxic mechanism of multiwall carbonnanotubes and nano-onions on human skin fibroblast. Nano Lett.5, 2448–2464.

Donaldson, K., Tran, C.L., 2002. Inflammation caused by particles andfibers. Inhal. Toxicol. 14, 5–27.

Donaldson, K., Stone, V., Duffin, R., Clouter, A., Schins, R., Borm, P.J.,2001. The quartz hazard: effects of surface and matrix on inflam-mogenic activity. Environ. Pathol. Toxicol. Oncol. 1 (Suppl.),109–118.

Donaldson, K., Stone, V., Tran, C.L., Kreyling, W., Borm, P.J.A., 2004.Nanotoxicology. Occup. Environ. Med. 61, 727–728.

Donaldson, K., Aitken, R., Tran, L., Stone, V., Duffin, R., Forrest, G.,Alexander, A., 2006. Carbon nanotubes: a review of their propertiesin relation to pulmonary toxicology and workplace safety. Toxicol.Sci. 92, 5–22.

Drake, P.L., Hazelwood, K.J., 2005. Exposure-related health effects
of silver and silver compounds: a review. Ann. Occup. Hyg. 49,575–585.
Elder, A., Gelein, R., Silva, V., Feikert, T., Opanashuk, L., Carter, J.,Potter, R., Maynard, A., Ito, Y., Finkelstein, J., Oberdorster, G.,2006. Translocation of inhaled ultrafine manganese oxide parti-

I

gy Letters 176 (2008) 1–12 9

cles to the central nervous system. Environ. Health Perspect. 114,1172–1178.

lechiguerra, J.L., Burt, J.L., Morones, J.R., Camacho-Bragado, A.,Gao, X., Lara, H.H., Yacaman, M.J., 2005. Interaction of silvernanoparticles with HIV-1. J. Nanobiotechnol. 3, 6.

merson, R.J., Davis, G.S., 1983. Effect of alveolar lining material-coated silica on rat alveolar macrophages. Environ. HealthPerspect. 51, 81–84.

turska, M., Obreshkova, E., 1979. Argyria in the prolonged use ofadsorgan. Vutr. Boles. 18, 121–123.

vanoff Jr., D.D., Chumanov, G., 2005. Synthesis and optical prop-erties of silver nanoparticles and arrays. Chemphyschem 6,1221–1231.

ubini, B., 1997. Surface reactivity in the pathogenic response to par-ticulates. Environ. Health Perspect. 5 (Suppl.), 1013–1020.

an, X., Liu, T., Zhong, J., Liu, X., Li, G., 2004. Effect of silvernanoparticles on the electron transfer reactivity and the catalyticactivity of myoglobin. Chembiochem 5, 1686–1691.

arner, M., Reglinski, J., Smith, W.E., Stewart, M.J., 1994. The inter-action of colloidal metals with erythrocytes. J. Inorg. Biochem. 56,283–290.

atti, A.M., Rivasi, F., 2002. Biocompatibility of micro- and nanopar-ticles. Part I. In liver and kidney. Biomaterials 23, 2381–2387.

atti, A.M., Montanari, S., Monari, E., Gambarelli, A., Capitani, F.,Parisini, B., 2004. Detection of micro- and nano-sized biocompat-ible particles in the blood. J. Mater. Sci. Mater. Med. 15, 469–472.

atti, A.M., Montanari, S., Gambarelli, A., Capitani, F., Salvatori, R.,2005. In vivo short- and long-term evaluation of the interactionmaterial blood. J. Mater. Sci. Mater. Med. 16, 1213–1219.

ehr, P., Geiser, M., Hof, V.I., Schurch, S., 2000. Surfactant ultrafineparticle interactions: what we can learn from PM10 studies. Phil.Trans. R. Soc. Lond. A 358, 2707–2718.

eiser, M., Schurch, S., Gehr, P., 2003. Influence of surface chem-istry and topography of particles on their immersion into the lung’ssurface-lining layer. J. Appl. Physiol. 94, 1793–1801.

erber, P.J., Lehmann, C., Gehr, P., Schurch, S., 2006. Wetting andspreading of a surfactant film on solid particles: influence of sharpedges and surface irregularities. Langmuir 22, 5273–5281.

illissen, A., Gessner, C., Hammerschmidt, S., Hoheisel, G., Wirtz, H.,2006. Health significance of inhaled particles. Dtsch Med Wochen-schr 131, 639–644.

opee, N.V., Roberts, D.W., Webb, P., Cozart, C.R., Siitonen, P.H.,Warbritton, A.R., Yu, W.W., Colvin, V.L., Walker, N.J., Howard,P.C., 2007. Migration of intradermally injected quantum dots tosentinel organs in mice. Toxicol. Sci. 98, 249–257.

ubbard, A.K., Timblin, C.R., Shukla, A., Rincon, M., Mossman, B.T.,2002. Activation of NF-kappa B-dependent gene expression bysilica in lungs of luciferase reporter mice. Am. J. Physiol. LungCell. Mol. Physiol. 282, L968–L975.

ussain, S.M., Hess, K.L., Gearhart, J.M., Geiss, K.T., Schlager, J.J.,2005. In vitro toxicity of nanoparticles in BRL 3A rat liver cells.Toxicol. In Vitro 19, 975–983.

ussain, S.M., Javorina, A.K., Schrand, A.M., Duhart, H.M., Ali, S.F.,Schlager, J.J., 2006. The interaction of manganese nanoparticleswith PC-12 cells induces dopamine depletion. Toxicol. Sci. 92,456–463.

noue, K., Takano, H., Yanagisawa, R., Hirano, S., Sakurai, M., Shi-
mada, A., Yoshikawa, T., 2006. Effects of airway exposure tonanoparticles on lung inflammation induced by bacterial endotoxinin mice. Environ. Health Perspect. 114, 1325–1330.
noue, K., Takano, H., Yanagisawa, R., Hirano, S., Sakurai, M., Shi-mada, A., Yoshikawa, T., 2007. Effects of inhaled nanoparticles on

1 oxicolo

I

J

J

K

K

K

K

K

K

K

K

K

K

K

K

K

L

L

L

L

L

L

L

L

L

L

L

L

L

L

L

M

M

M

0 X. Chen, H.J. Schluesener / T

acute lung injury induced by lipopolysaccharide in mice. Toxicol-ogy 238, 99–110.

wasaki, S., Yoshimura, A., Ideura, T., Koshikawa, S., Sudo, M., 1997.Elimination study of silver in a hemodialyzed burn patient treatedwith silver sulfadiazine cream. Am. J. Kidney Dis. 30, 287–290.

ani, P., Halbert, G.W., Langridge, J., Florence, A.T., 1990. Nanopar-ticle uptake by the rat gastrointestinal mucosa: quantitation andparticle size dependency. J. Pharm. Pharmacol. 42, 821–826.

ose, R.M., Jose, L.E., Alejandra, C., Katherine, H., Juan, B.K., Jose,T.R., Miguel, J.Y., 2005. The bactericidal effect of silver nanopar-ticles. Nanotechnology 16, 2346–2353.

aewamatawong, T., Shimada, A., Okajima, M., Inoue, H., Morita,T., Inoue, K., Takano, H., 2006. Acute and subacute pulmonarytoxicity of low dose of ultrafine colloidal silica particles in miceafter intratracheal instillation. Toxicol. Pathol. 34, 958–965.

akurai, M., Demitsu, T., Umemoto, N., Ohtsuki, M., Nakagawa, H.,2003. Activation of mast cells by silver particles in a patient withlocalized argyria due to implantation of acupuncture needles. Br.J. Dermatol. 148, 822.

ang, C.M., Jang, A.S., Ahn, M.H., Shin, J.A., Kim, J.H., Choi, Y.S.,Rhim, T.Y., Park, C.S., 2005. Interleukin-25 and interleukin-13production by alveolar macrophages in response to particles. Am.J. Respir. Cell. Mol. Biol. 33, 290–296.

anj, R.S., Kang, J.L., Castranova, V., 2006. Interaction between pri-mary alveolar macrophages and primary alveolar type II cells underbasal conditions and after lipopolysaccharide or quartz exposure.J. Toxicol. Environ. Health A 69, 1097–1116.

ato, T., Yashiro, T., Murata, Y., Herbert, D.C., Oshikawa, K., Bando,M., Ohno, S., Sugiyama, Y., 2003. Evidence that exogenoussubstances can be phagocytized by alveolar epithelial cells andtransported into blood capillaries. Cell Tissue Res. 311, 47–51.

endall, M., Brown, L., Trought, K., 2004. Molecular adsorption atparticle surfaces: a PM toxicity mediation mechanism. Inhal. Tox-icol. 1 (Suppl.), 99–105.

en-Ichiro, I., Hirohisa, T., Rie, Y., Seishiro, H., Miho, S., Akinori, S.,Toshikazu, Y., 2006. Effects of airway exposure to nanoparticles onlung inflammation induced by bacterial endotoxin in mice. Environ.Health Perspect. 114, 1325–1330.

im, S., Lim, Y.T., Soltesz, E.G., De Grand, A.M., Lee, J., Nakayama,A., Parker, J.A., Mihaljevic, T., Laurence, R.G., Dor, D.M., Cohn,L.H., Bawendi, M.G., Frangioni, J.V., 2004. Near-infrared fluores-cent type II quantum dots for sentinel lymph node mapping. Nat.Biotechnol. 22, 93–97.

obayashi, Y., Katakami, H., Mine, E., Nagao, D., Konno, M., Liz-Marzan, L.M., 2005. Silica coating of silver nanoparticles using amodified Stober method. J. Colloid Interf. Sci. 283, 392–396.

ohli, A.K., Alpar, H.O., 2004. Potential use of nanoparticles for tran-scutaneous vaccine delivery: effect of particle size and charge. Int.J. Pharm. 275, 13–17.

oike, E., Kobayashi, T., 2006. Chemical and biological oxidativeeffects of carbon black nanoparticles. Chemosphere 65, 946–951.

one, B.C., Kaleta, M., Gullans, S.R., 1988. Silver ion (Ag+) inducedincreases in cell membrane K+ and Na+ permeability in renal prox-imal tubule: reversal by thiol reagents. J. Membr. Biol. 102, 11–19.

reyling, W.G., Semmler, M., Erbe, F., Mayer, P., Takenaka, S., Schulz,H., Oberdorster, G., Ziesenis, A., 2002. Translocation of ultrafineinsoluble iridium particles from lung epithelium to extrapulmonary
organs is size dependent but very low. J. Toxicol. Environ. HealthA 65, 1513–1530.
ademann, J., Weigmann, H., Rickmeyer, C., Barthelmes, H., Schaefer,H., Mueller, G., Sterry, W., 1999. Penetration of titanium diox-ide microparticles in a sunscreen formulation into the horny layer

M

gy Letters 176 (2008) 1–12

and the follicular orifice. Skin Pharmacol. Appl. Skin Physiol. 12,247–256.

am, P.K., Chan, E.S., Ho, W.S., Liew, C.T., 2004. In vitro cytotoxicitytesting of a nanocrystalline silver dressing (Acticoat) on culturedkeratinocytes. Br. J. Biomed. Sci. 61, 125–127.

am, C.W., James, J.T., McCluskey, R., Arepalli, S., Hunter, R.L.,2006. A review of carbon nanotube toxicity and assessment ofpotential occupational and environmental health risks. Crit. Rev.Toxicol. 36, 189–217.

ansdown, A.B., 2006. Silver in health care: antimicrobial effects andsafety in use. Curr. Probl. Dermatol. 33, 17–34.

ee, K.S., El-Sayed, M.A., 2006. Gold and silver nanoparticles insensing and imaging: sensitivity of plasmon response to size, shape,and metal composition. J. Phys. Chem. 110, 19220–19225.

ee, H.Y., Park, H.K., Lee, Y.M., Kim, K., Park, S.B., 2007. A practicalprocedure for producing silver nanocoated fabric and its antibac-terial evaluation for biomedical applications. Chem. Commun.(Camb.) 28, 2959–2961.

esniak, W., Bielinska, A.U., Sun, K., Janczak, K.W., Shi, X., BakerJr., J.R., Balogh, L.P., 2005. Silver/dendrimer nanocompositesas biomarkers: fabrication, characterization, in vitro toxicity, andintracellular detection. Nano Lett. 5, 2123–2130.

imbach, L.K., Wick, P., Manser, P., Grass, R.N., Bruinink, A., Stark,W.J., 2007. Exposure of engineered nanoparticles to human lungepithelial cells: influence of chemical composition and catalyticactivity on oxidative stress. Environ. Sci. Technol. 41, 4158–4163.

ippmann, M., Yeates, D.B., Albert, R.E., 1980. Deposition, retention,and clearance of inhaled particles. Br. J. Ind. Med. 37, 337–362.

iu, X., Keane, M.J., Harrison, J.C., Cilento, E.V., Ong, T., Wal-lace, W.E., 1998. Phospholipid surfactant adsorption by respirablequartz and in vitro expression of cytotoxicity and DNA damage.Toxicol. Lett. 96–97, 77–84.

iu, H.H., Cao, X., Yang, Y., Liu, M.G., Wang, Y.F., 2006. Array-basednano-amplification technique was applied in detection of hepatitisE virus. J. Biochem. Mol. Biol. 39, 247–252.

ok, C.N., Ho, C.M., Chen, R., He, Q.Y., Yu, W.Y., Sun, H., Tam,P.K., Chiu, J.F., Che, C.M., 2006. Proteomic analysis of the modeof antibacterial action of silver nanoparticles. J. Proteome Res. 5,916–924.

opez, V.C., Hadgraft, J., Snowden, M.J., 2005. The use of colloidalmicrogels as a (trans)dermal drug delivery system. Int. J. Pharm.292, 137–147.

undborg, M., Johard, U., Lastbom, L., Gerde, P., Camner, P., 2001.Human alveolar macrophage phagocytic function is impaired byaggregates of ultrafine carbon particles. Environ. Res. 86, 244–253.

undborg, M., Dahlen, S.E., Johard, U., Gerde, P., Jarstrand, C., Cam-ner, P., Lastbom, L., 2006. Aggregates of ultrafine particles impairphagocytosis of microorganisms by human alveolar macrophages.Environ. Res. 100, 197–204.

agrez, A., Kasas, S., Salicio, V., Pasquier, N., Seo, J.W., Celio, M.,Catsicas, S., Schwaller, B., Forro, L., 2006. Cellular toxicity ofcarbon-based nanomaterials. Nano Lett. 6, 1121–1125.

akarava, N., Parfenov, A., Baskakov, I.V., 2005. Water-soluble hybridnanoclusters with extra bright and photostable emissions: a newtool for biological imaging. Biophys. J. 89, 572–580.

ark, W.F., Judith, C.S., Gunter, O., Paul, E.M., David, C., Anthony,P.P., Lauren, M.F., Donna, M.S., Christopher, C., Li-Shan, H.,
Mark, J.U., 2006. Inhalation of ultrafine particles alters bloodleukocyte expression of adhesion molecules in humans. Environ.Health Perspect. 114, 51–58.
irsattari, S.M., Hammond, R.R., Sharpe, M.D., Leung, F.Y.,Young, G.B., 2004. Myoclonic status epilepticus following

oxicolo

P

P

P

R

R

R

R

R

R

R

S

S

S

S

S

S


repeated oral ingestion of colloidal silver. Neurology 62, 1408–1410.

Moller, W., Hofer, T., Ziesenis, A., Karg, E., Heyder, J., 2002. Ultrafineparticles cause cytoskeletal dysfunctions in macrophages. Toxicol.Appl. Pharmacol. 182, 197–207.

Monteiro-Riviere, N.A., Nemanich, R.J., Inman, A.O., Wang, Y.Y.,Riviere, J.E., 2005. Multi-walled carbon nanotube interactions withhuman epidermal keratinocytes. Toxicol. Lett. 155, 377–384.

Muangman, P., Chuntrasakul, C., Silthram, S., Suvanchote, S., Ben-jathanung, R., Kittidacha, S., Rueksomtawin, S., 2006. Comparisonof efficacy of 1% silver sulfadiazine and Acticoat for treatmentof partial-thickness burn wounds. J. Med. Assoc. Thai. 89, 953–958.

Nel, A., Xia, T., Madler, L., Li, N., 2006. Toxic potential of materialsat the nanolevel. Science 311, 622–627.

Nordberg, G., Gerhardsson, L., Silver, 1988. In: Seiler, H.G., Sigel, H.,Sigel, A. (Eds.), Handbook on Toxicity of Inorganic Compounds.Marcel Dekker, New York, pp. 619–624.

Nurkiewicz, T.R., Porter, D.W., Barger, M., Millecchia, L., Rao, K.M.,Marvar, P.J., Hubbs, A.F., Castranova, V., Boegehold, M.A., 2006.Systemic microvascular dysfunction and inflammation after pul-monary particulate matter exposure. Environ. Health Perspect. 114,412–419.

Oberdorster, G., 1988. Lung clearance of inhaled insoluble and solubleparticles. J. Aerosol Med. 1, 289–329.

Oberdorster, G., Sharp, Z., Atudorei, V., Elder, A., Gelein, R., Lunts,A., Kreyling, W., Cox, C., 2002. Extrapulmonary translocation ofultrafine carbon particles following whole-body inhalation expo-sure of rats. J. Toxicol. Environ. Health A 65, 1531–1543.

Oberdorster, G., Sharp, Z., Atudorei, V., Elder, A., Gelein, R., Kreyling,W., Cox, C., 2004. Translocation of inhaled ultrafine particles tothe brain. Inhal. Toxicol. 16, 437–445.

Oberdorster, G., Oberdorster, E., Oberdorster, J., 2005a. Nanotoxi-cology: an emerging discipline evolving from studies of ultrafineparticles. Environ. Health Perspect. 113, 823–839.

Oberdorster, G., Maynard, A., Donaldson, K., Castranova, V., Fitz-patrick, J., Ausman, K., Carter, J., Karn, B., Kreyling, W., Lai, D.,Olin, S., Monteiro-Riviere, N., Warheit, D., Yang, H., 2005b. PartFibre Toxicol. 2, 8–43.

Ohbo, Y., Fukuzako, H., Takeuchi, K., Takigawa, M., 1996. Argyriaand convulsive seizures caused by ingestion of silver in a patientwith schizophrenia. Psychiatry Clin. Neurosci. 50, 89–90.

Ovrevik, J., Myran, T., Refsnes, M., Lag, M., Becher, R., Hetland, R.B.,Schwarze, P.E., 2005. Mineral particles of varying compositioninduce differential chemokine release from epithelial lung cells:importance of physico-chemical characteristics. Ann. Occup. Hyg.49, 219–231.

Paddle-Ledinek, J.E., Nasa, Z., Cleland, H.J., 2006. Effect of differentwound dressings on cell viability and proliferation. Plast. Reconstr.Surg. 117, 110S–118S.

Parks, C.G., Conrad, K., Cooper, G.S., 1999. Occupational exposureto crystalline silica and autoimmune disease. Environ. Health Per-spect. 5 (Suppl.), 793–802.

Peters, K., Unger, R.E., Kirkpatrick, C.J., Gatti, A.M., Monari, E.,2004. Effects of nano-scaled particles on endothelial cell func-tion in vitro: studies on viability, proliferation and inflammation.J. Mater. Sci. Mater. Med. 15, 321–325.

Peters, A., Veronesi, B., Calderon-Garciduenas, L., Gehr, P., Chen,L.C., Geiser, M., Reed, W., Rothen-Rutishauser, B., Schurch, S.,Schulz, H., 2006. Translocation and potential neurological effectsof fine and ultrafine particles a critical update. Part Fibre Toxicol.3, 13.

S

gy Letters 176 (2008) 1–12 11

oon, V.K., Burd, A., 2004. In vitro cytotoxity of silver: implicationfor clinical wound care. Burns 30, 140–147.

ope III, C.A., Burnett, R.T., Thurston, G.D., Thun, M.J., Calle,E.E., Krewski, D., 2004. Cardiovascular mortality and longtermexposure to particulate air pollution: epidemiological evidence ofgeneral pathophysiological pathways of disease. Circulation 109,71–77.

yatenko, A., Shimokawa, K., Yamaguchi, M., 2004. Synthesis of sil-ver nanoparticles by laser ablation in pure water. J. Appl. Phys. A:Mater. Sci. Proces. A79, 803–806.

egina, R., Richard, P.P., Alexandra, S., Mark, F., Josef, C., Gunther,O., Erich, W., Annette, P., 2007. Ultrafine particles and plateletactivation in patients with coronary heart disease—results from aprospective panel study. Part Fibre Toxicol. 4, 1.

enwick, L.C., Donaldson, K., Clouter, A., 2001. Impairment of alveo-lar macrophage phagocytosis by ultrafine particles. Toxicol. Appl.Pharmacol. 172, 119–127.

enwick, L.C., Brown, D., Clouter, A., Donaldson, K., 2004. Increasedinflammation and altered macrophage chemotactic responsescaused by two ultrafine particle types. Occup. Environ. Med. 61,442–447.

ouse, J.G., Yang, J., Ryman-Rasmussen, J.P., Barron, A.R., Monteiro-Riviere, N.A., 2007. Effects of mechanical flexion on thepenetration of fullerene amino acid-derivatized peptide nanopar-ticles through skin. Nano Lett. 7, 155–160.

ussell, A.D., Hugo, W.B., 1994. Antimicrobial activity and action ofsilver. Prog. Med. Chem. 31, 351–370.

yman-Rasmussen, J.P., Riviere, J.E., Monteiro-Riviere, N.A., 2006.Penetration of intact skin by quantum dots with diverse physico-chemical properties. Tox. Sci. 91, 159–165.

yman-Rasmussen, J.P., Riviere, J.E., Monteiro-Riviere, N.A., 2007.Surface coatings determine cytotoxicity and irritation potential ofquantum dot nanoparticles in epidermal keratinocytes. J. Invest.Dermatol. 127, 143–153.

amet, J.M., Dominici, F., Curriero, F.C., Coursac, I., Zeger, S.L.,2000. Fine particulate air pollution and mortality in 20 U.S. cities,1987–1994. N. Engl. J. Med. 343, 1742–1749.

arkar, S., Sharma, C., Yog, R., Periakaruppan, A., Jejelowo, O.,Thomas, R., Barrera, E.V., Rice-Ficht, A.C., Wilson, B.L., Ramesh,G.T., 2007. Analysis of stress responsive genes induced bysingle-walled carbon nanotubes in BJ foreskin cells. J. Nanosci.Nanotechnol. 7, 584–592.

ayes, C.M., Wahi, R., Kurian, P.A., Liu, Y., West, J.L., Ausman, K.D.,Warheit, D.B., Colvin, V.L., 2006. Correlating nanoscale titaniastructure with toxicity: a cytotoxicity and inflammatory responsestudy with human dermal fibroblasts and human lung epithelialcells. Toxicol. Sci. 92, 174–185.

chaumann, F., Borm, P.J., Herbrich, A., Knoch, J., Pitz, M., Schins,R.P., Luettig, B., Hohlfeld, J.M., Heinrich, J., Krug, N., 2004.Metal-rich ambient particles (particulate matter 2.5) cause airwayinflammation in healthy subjects. Am. J. Respir. Crit. Care Med.170, 898–903.

ergeev, M.B., Kasaikin, A.V., Litmanovich, A.E., 1999. Cry-ochemical synthesis and properties of silver nanoparticledispersions stabilised by poly(2-dimethylaminoethyl methacry-late). Mendeleev Commun. 9, 130–132.

harma, C.S., Sarkar, S., Periyakaruppan, A., Barr, J., Wise, K.,
Thomas, R., Wilson, B.L., Ramesh, G.T., 2007. Single-walled car-bon nanotubes induces oxidative stress in rat lung epithelial cells.J. Nanosci. Nanotechnol. 7, 2466–2472.
hen, J., Shan, W., Zhang, Y., Du, J., Xu, H., Fan, K., Shen, W., Tang,Y., 2004. A novel catalyst with high activity for polyhydric alcohol

1 oxicolo

S

S

S

S

S

S

S

S

S

T

T

T

T

T

V

V

W

W

W

W

W

W

W

X

Y

Y

Z

Z

Z

Z

Z

2 X. Chen, H.J. Schluesener / T

oxidation: nanosilver/zeolite film. Chem. Commun. (Camb.) 21,2880–2881.

him, J., Seok, K.H., Park, W.S., Han, S.H., Kim, J., Chang, I.S.,2004. Transdermal delivery of mixnoxidil with block copolymernanoparticles. J. Control Release 97, 477–484.

himada, A., Kawamura, N., Okajima, M., Kaewamatawong, T., Inoue,H., Morita, T., 2006. Translocation pathway of the intratracheallyinstilled ultrafine particles from the lung into the blood circulationin the mouse. Toxicol. Pathol. 34, 949–957.

hvedova, A.A., Kisin, E.R., Mercer, R., Murray, A.R., Johnson,V.J., Potapovich, A.I., Tyurina, Y.Y., Gorelik, O., Arepalli, S.,Schwegler-Berry, D., Hubbs, A.F., Antonini, J., Evans, D.E., Ku,B.K., Ramsey, D., Maynard, A., Kagan, V.E., Castranova, V.,Baron, P., 2005. Unusual inflammatory and fibrogenic pulmonaryresponses to single-walled carbon nanotubes in mice. Am. J. Phys-iol. Lung Cell. Mol. Physiol. 289, L698–L708.

mith, M.W., Thomas, N.W., Jenkins, P.G., Miller, N.G., Cremaschi,D., Porta, C., 1995. Selective transport of microparticles acrossPeyer’s patch follicle-associated M cells from mice and rats. Exp.Physiol. 80, 735–743.

oto, K., Garza, K.M., Murr, L.E., 2007. Cytotoxic effects of aggre-gated nanomaterials. Acta Biomater. 3, 351–358.

pencer, W.H., Garron, L.K., Contreras, F., Hayes, T.L., Lai, C., 1980.Endogenous and exogenous ocular and systemic silver deposition.Trans. Ophthalmol. Soc. U.K. 100, 171–178.

un, Y., Xia, Y., 2002. Shape-controlled synthesis of gold and silvernanoparticles. Science 298, 2176–2179.

upp, A.P., Neely, A.N., Supp, D.M., Warden, G.D., Boyce, S.T.,2005. Evaluation of cytotoxicity and antimicrobial activity of Acti-coat burn dressing for management of microbial contamination incultured skin substitutes grafted to athymic mice. J. Burn CareRehabil. 26, 238–246.

uzuki, Y., Yoshimaru, T., Yamashita, K., Matsui, T., Yamaki, M.,Shimizu, K., 2001. Exposure of RBL-2H3 mast cells to Ag(+)induces cell degranulation and mediator release. Biochem. Bio-phys. Res. Commun. 283, 707–714.

akenaka, S., Karg, E., Moller, W., Roth, C., Ziesenis, A., Heinzmann,U., Schramel, P., Heyder, J., 2000. A morphologic study on the fateof ultrafine silver particles: distribution pattern of phagocytizedmetallic silver in vitro and in vivo. Inhal. Toxicol. 12, 291–299.

akenaka, S., Karg, E., Roth, C., Schulz, H., Ziesenis, A., Heinzmann,U., Schramel, P., Heyder, J., 2001. Pulmonary and systemic distri-bution of inhaled ultrafine silver particles in rats. Environ. HealthPerspect. 4 (Suppl.), 547–551.

ian, F., Cui, D., Schwarz, H., Estrada, G.G., Kobayashi, H., 2006.Cytotoxicity of single-wall carbon nanotubes on human fibroblasts.Toxicol. In Vitro 20, 1202–1212.

inkle, S.S., Antonini, J.M., Rich, B.A., Roberts, J.R., Salmen, R.,DePree, K., Adkins, E.J., 2003. Skin as a route of exposure and sen-sitization in chronic beryllium disease. Environ. Health Perspect.111, 1202–1208.

rop, M., Novak, M., Rodl, S., Hellbom, B., Kroell, W., Goessler, W.,2006. Silver-coated dressing Acticoat caused raised liver enzymesand argyria-like symptoms in burn patient. J. Trauma 60, 648–652.

an de Voorde, K., Nijsten, T., Schelfhout, K., Moorkens, G., Lambert,J., 2005. Long-term use of silver containing nose-drops resultingin systemic argyria. Acta Clin. Belg. 60, 33–35.

igneshwaran, N., Kathe, A.A., Varadarajan, P.V., Nachane, R.P.,Balasubramanya, R.H.J., 2007. Functional finishing of cotton fab-

Z

gy Letters 176 (2008) 1–12

rics using silver nanoparticles. Nanosci. Nanotechnol. 7, 1893–1897.

arheit, D.B., 2001a. Inhaled amorphous silica particulates: what dowe know about their toxicological profiles? J. Environ. Pathol.Toxicol. Oncol. 1 (Suppl.), 133–141.

arheit, D.B., Hansen, J.F., Yuen, I.S., Kelly, D.P., Snajdr, S.I., Hart-sky, M.A., 1997. Inhalation of high concentrations of low toxicitydusts in rats results in impaired pulmonary clearance mecha-nisms and persistent inflammation. Toxicol. Appl. Pharmacol. 145,10–22.

arheit, D.B., Borm, P.J., Hennes, C., Lademann, J., 2007. Testingstrategies to establish the safety of nanomaterials: conclusions ofan ECETOC workshop. Inhal. Toxicol. 19, 631–643.

est, J.L., Halas, N.J., 2003. Engineered nanomaterials for biopho-tonics applications: improving sensing, imaging, and therapeutics.Annu. Rev. Biomed. Eng. 5, 285–292.

infried, M., David, M.B., Wolfgang, G.K., Vicki, S., 2005. Ultrafineparticles cause cytoskeletal dysfunctions in macrophages:role ofintracellular calcium. Part Fibre Toxicol. 2, 7–18.

itzmann, F.A., Monteiro-Riviere, N.A., 2006. Multi-walled carbonnanotube exposure alters protein expression in human ker-atinocytes. Nanomedicine 2, 158–168.

right, J.B., Lam, K., Buret, A.G., Olson, M.E., Burrell, R.E., 2002.Early healing events in a porcine model of contaminated wounds:effects of nanocrystalline silver on matrix metalloproteinases, cellapoptosis, and healing. Wound Repair Regen. 10, 141–151.

ia, T., Kovochich, M., Brant, J., Hotze, M., Sempf, J., Oberley, T.,Sioutas, C., Yeh, J.I., Wiesner, M.R., Nel, A.E., 2006. Comparisonof the abilities of ambient and manufactured nanoparticles to inducecellular toxicity according to an oxidative stress paradigm. NanoLett. 6, 1794–1807.

eates, D.B., Mauderly, J.L., 2001. Inhaled environmen-tal/occupational irritants and allergens: mechanisms ofcardiovascular and systemic responses. Introduction. Environ.Health Perspect. 4 (Suppl.), 479–481.

oshimaru, T., Suzuki, Y., Inoue, T., Niide, O., Ra, C., 2006. Silveractivates mast cells through reactive oxygen species production anda thiol-sensitive store-independent Ca2+ influx. Free Radic. Biol.Med. 40, 1949–1959.

hai, H.J., Sun, D.W., Wang, H.S., 2006. Catalytic properties ofsilica/silver nanocomposites. J. Nanosci. Nanotechnol. 6, 1968–1972.

hang, Y., Sun, J., 2007b. A Study on the bio-safety for nano-silver asanti-bacterial materials. Chin. J. Med. Instrumen. 31, 35–38.

hang, Y., Chen, F., Zhuang, J., 2002. Synthesis of silver nanoparticlesvia electrochemical reduction on compact zeolite film modifiedelectrodes. Chem. Commun. (Camb.) 7, 2814–2815.

hang, L.W., Zeng, L., Barron, A.R., Monteiro-Riviere, N.A., 2007a.Biological interactions of functionalized single-wall carbon nan-otubes in human epidermal keratinocytes. Int. J. Toxicol. 26,103–113.

hang, Z., Yang, M., Huang, M., Hu, Y., Xie, J., 2007c. Study ongermicidal efficacy and toxicity of compound disinfectant gel ofnanometer silver and chlorhexidine acelate. Chin. J. Health Lab.
Technol. 17, 1403–1406.
heng, L., Tracy, H., Rebecca, S., Rebecca, C., Stephen, S.L.,Shih-Houng, Y., Anna, S., Michael, I.L., Petia, P.S., 2007. Car-diovascular effects of pulmonary exposure to single-wall carbonnanotubes. Environ. Health Perspect. 115, 377–382.

Drug Transport to Brain with Targeted Nanoparticles

Jean-Christophe Olivier

Faculty of Medicine and Pharmacy, University of Poitiers, 86000 Poitiers, France

Summary: Nanoparticle drug carriers consist of solid biode-gradable particles in size ranging from 10 to 1000 nm (50–300nm generally). They cannot freely diffuse through the blood-brain barrier (BBB) and require receptor-mediated transportthrough brain capillary endothelium to deliver their content intothe brain parenchyma. Polysorbate 80-coated polybutylcyano-acrylate nanoparticles can deliver drugs to the brain by a stilldebated mechanism. Despite interesting results these nanopar-ticles have limitations, discussed in this review, that may pre-clude, or at least limit, their potential clinical applications.Long-circulating nanoparticles made of methoxypoly(ethyleneglycol)- polylactide or poly(lactide-co-glycolide) (mPEG-PLA/

PLGA) have a good safety profiles and provide drug-sustainedrelease. The availability of functionalized PEG-PLA permits toprepare target-specific nanoparticles by conjugation of cell sur-face ligand. Using peptidomimetic antibodies to BBB transcy-tosis receptor, brain-targeted pegylated immunonanoparticlescan now be synthesized that should make possible the deliveryof entrapped actives into the brain parenchyma without induc-ing BBB permeability alteration. This review presents theirgeneral properties (structure, loading capacity, pharmacokinet-ics) and currently available methods for immunonanoparticlepreparation. Key Words: Nanoparticle, immunonanoparticle,brain targeting, blood brain barrier, transcytosis, PEG.

INTRODUCTION

Nanoparticles are solid colloidal matrix-like particlesmade of polymers1 or lipids.2 Generally administered bythe intravenous route like liposomes, they have beendeveloped for the targeted delivery of therapeutic orimaging agents. Their main advantages over liposomesare the low number of excipients used in their formula-tions, the simple procedures for preparation, a high phys-ical stability, and the possibility of sustained drug releasethat may be suitable in the treatment of chronic diseases.Until the mid 1990s, their development as drug carrierswas seriously limited by the lack of long-circulatingproperties.3 Therefore, in contrast to liposomes and de-spite the abundance of experimental works and achieve-ments in the field of nanoparticle technology, no nano-particle-based drug formulation has been marketed sofar. Due to their size ranging from 10 to 1000 nm (gen-erally 50–300 nm), and like liposomes, they are unableto diffuse through the blood-brain barrier (BBB) to reachthe brain parenchyma. Based on general parenteral for-mulation considerations and specific BBB features, Ta-

ble 1 summarizes the ideal nanoparticle properties re-quired for drug brain delivery.4 One particularlyinteresting application of nanoparticule could be the drugbrain delivery, accompanied with the local sustained re-lease, of the new large molecule therapeutics now avail-able to treat the CNS: peptides, proteins, genes, antisensedrugs. Due to their poor stability in biological fluids,rapid enzymatic degradation, unfavorable pharmacoki-netic properties, and lack of diffusion toward the CNS,they may be advantageously formulated in brain-targetedprotective nanocontainers.5 Compared with conventionaldrugs, they possess a high intrinsic pharmacological ac-tivity. The small dose requested for therapeutic effi-ciency could easily fit the loading capacity of nanopar-ticles and would not require the administration of largeamount of potentially toxic nanoparticle excipient. Be-cause of the large variety of the nanoparticles developedso far, this review will focus on nanoparticles investi-gated for brain delivery. Nanoparticles made of polybu-tylcyanoacrylate (PBCA, FIG. 1) have been intenselyinvestigated since the first papers in 1995 showing thatwhen coated with the nonionic surfactant polysorbate 80they permitted to deliver drugs to the brain.6,7 Despiteinteresting results, PBCA nanoparticles have limitations,discussed in this review, that may preclude, or at leastlimit, their potential clinical applications. Nanoparticlesmade of polylactide homopolymers (PLA) or poly(lac-

Address correspondence and reprint requests to Jean-ChristopheOlivier, Ph.D., Faculty of Medicine and Pharmacy, University of Poitiers,34rueduJardindesPlantes,86000Poitiers,France.E-mail:[email protected].

NeuroRx�: The Journal of the American Society for Experimental NeuroTherapeutics

Vol. 2, 108–119, January 2005 © The American Society for Experimental NeuroTherapeutics, Inc.108

tide-co-glycolide) heteropolymers (PLGA) may be apromising alternative. In the mid 1990s, long-circulatingpegylated PLA or PLGA nanoparticles have been madeavailable that opened great opportunities for drug target-ing.3 Pegylated nanoparticles are made of methoxypoly-(ethylene glycol)-PLA/PLGA (mPEG-PLA/PLGA, FIG.1), i.e., esters of PLA or PLGA with PEG of variousmolecular weights. More recently, the synthesis of func-tionalized pegylated PLA/PLGA nanoparticles openednew perspectives for targeted drug delivery in general,and for drug brain targeting in particular. This reviewwill present their general properties and will proposepreparation methods of brain-targeted pegylated nano-particles.

PBCA NANOPARTICLES

General considerationsNanoparticles made of poly(alkylcyanoacrylate) poly-

mers (FIG. 1) were first described in 19778 and wererecently the subject of a comprehensive review of theirproperties, preparation methods and potential therapeuticapplications.9 They are generally prepared from (iso)bu-tylcyanoacrylate or (iso)hexylcyanoacrylate monomersby emulsion anionic polymerization in an acidic aqueoussolution of a colloidal stabilizer such as dextran 70,polysorbates, and poloxamers. Inclusion of drug can bemade during the polymerization process or by adsorptiononto preformed nanoparticles. Using the first method,chemical reactions may occur between drugs and mono-mers.10 Alternatively, the interaction between adsorbeddrugs and the nanoparticle may lack stability, especiallywhen a surfactant is subsequently added to the prepara-tion11 or, once the nanoparticles are dispersed in blood,by a combined effect of serum protein competition andpolymer degradation.12 The length of the alkyl pendantgoverns degradation rates13,14 and toxicity15,16 of poly-(alkylcyanoacrylate) nanoparticles, which decrease in theorder methyl�ethyl�butyl/isobutyl�hexyl/isohexyl.

Phase I trials were therefore carried out with poly(iso-hexyl cyanoacrylate) nanoparticles that have the bestsafety profile and an appropriate degradation rate.17 Lack-ing stealth properties, poly(alkylcyanoacrylate) nanopar-ticles administered intravenously are rapidly cleared fromthe blood stream by the monuclear phagocyte system(MPS) and mainly accumulate in liver and spleen,18–20

together with the entrapped compounds.21–23 Only pegy-lated polyalkylcyanoacrylate nanoparticles have lowerMPS uptake and prolonged blood circulation in vivo.24

Brain delivery with PBCA nanoparticlesAdsorbed onto polysorbate 80-coated PBCA nanopar-

ticles administered intravenously compounds with poorbrain diffusion as diverse as doxorubicin,25,26 loperam-ide,27 tubocurarine,28 the hexapeptide dalargin6,7 weresuccessfully delivered to the brain, where they induced apharmacological effect (for review, see Kreuter29). Thechemical nature of the overcoating surfactant is of im-portance, because only polysorbates, not poloxamers(184, 188, 388, or 407), poloxamine 908, Cremophors(EZ or RH40) or polyoxyethylene(23)-laurylether, led toa CNS pharmacological effect of dalargin.30 As themechanism of action, it was hypothesized that polysor-bate-coated nanoparticles were transported across theBBB via endocytosis by the brain capillary endothelialcells.29 This endocytosis would be triggered by a serumprotein, apolipoprotein E, reported to adsorb on polysor-bate 20, 40, 60, or 80-coated nanoparticles after a 5-min

FIG. 1. Structure of poly(alkylcyanoacrylate), methoxypoly(eth-ylene glycol)-polylactide [or poly(lactic acid)] (mPEG-PLA) andmethoxypoly(ethylene glycol)-poly(lactide-co-glycolide) [or poly-(lactic-co-glycolic acid)] (mPEG-PLGA).

TABLE 1. Ideal Properties of Nanoparticles for DrugBrain Delivery

● Nontoxic, biodegradable, and biocompatible● Particle diameter � 100 nm● Physical stability in blood (no aggregation)● Avoidance of the MPS (no opsonization), prolonged

blood circulation time● BBB-targeted and brain delivery (receptor-mediated

transcytosis across brain capillary endothelial cells)● Scalable and cost-effective manufacturing process● Amenable to small molecules, peptides, proteins, or

nucleic acids● Minimal nanoparticle excipient-induced drug alteration

(chemical degradation/alteration, protein denaturation)● Possible modulation of drug release profiles

BRAIN-TARGETED NANOPARTICLES 109

NeuroRx�, Vol. 2, No. 1, 2005

incubation in citrate-stabilized plasma at 37°C, but noton nanoparticles coated with poloxamers 338, 407, Cre-mophor EL, or RH 40.29 Despite numerous argumentslisted by Kreuter,29 this hypothesis raises questions,based on the following observations. 1) Apolipoprotein Eadsorption is not specific of polysorbate 80-coated sur-faces because it was shown to adsorb onto pegylatedPLA nanoparticles.31,32 2) Polysorbate 80-coated poly-(methylmethacrylate) nanoparticles are not distributedinto the brain after IV administration.33 3) Replacingpolysorbate 80-coated PBCA nanoparticles with polysor-bate 80-coated polystyrene nanoparticles completelyabolished dalargin brain delivery.11 4) The pharmacoki-netic profile of polysorbate 80-coated nanoparticles isnot favorable to brain distribution, due to a massiveuptake by the MPS resulting in liver and spleen accu-mulation.33 5) Polysorbate 80 and serum protein compe-tition, as well as the rapid nanoparticle degradation inserum/plasma, were shown to induce desorption of com-pounds adsorbed onto PBCA nanoparticles within a fewminutes.11,12 As an evidence of this desorption, bloodpharmacokinetic profiles of drugs adsorbed onto poly-sorbate 80-coated PBCA nanoparticles administered in-travenously were actually similar to free solutions,25,34,35

and not at all typical of drugs associated to nonstealthcolloidal drug carriers.21,22,23,36 Therefore, as an alterna-tive to the brain uptake of nanoparticles, we hypothe-sized a nanoparticle-induced nonspecific BBB permeabi-lization.11 It has been known for a long time thatpolysorbate 80 causes BBB disturbance at intravenoussystemic doses as low as 3 mg/kg37 (25–100 mg/kgpolysorbate 80 doses were used in brain targeting exper-iments7,25,27). Recently, Calvo et al.36 showed that apolysorbate 80 intravenous dose of 20 mg/kg in ratsdramatically increased BBB permeability to sucrose. Inrats treated with polysorbate 80-coated PBCA nanopar-ticles (polysorbate 80: 25 mg/kg, nanoparticles: 50 mg/kg) inulin spaces increased by 10% (not significant) after10 min and by 99% (significant) after 45 min comparedwith control.38 Because apparently no brain uptake wasobserved with control drug-polysorbate 80 solutions, thetoxicity of PBCA nanoparticles was proposed as a syn-ergistic factor for BBB permeabilization.11 The nanopar-ticle doses permitting brain delivery (100–166 mg/kggenerally) were close to the lethal dose 50% of PBCAnanoparticles (230 mg/kg in mice16). Polysorbate 80-coated or uncoated PBCA nanoparticles (unloaded withdrug) induced a dramatic decrease in mice locomotoractivity (associated with obvious signs of distress) at ananoparticle dose of 135 mg/kg and the permeabilizationof an in vitro BBB model at a concentration of 10 �g/ml(to be compared to the 1.5 mg/ml theoretical concentra-tion reached in mice blood after dosing animals with a135 mg/kg nanoparticle dose).11 In contrast, the nontoxicpolysorbate 80-coated polystyrene nanoparticles were in-

effective at delivering dalargin to the brain.11 In a con-text of general toxicity induced by the high dose ofPBCA nanoparticles and associated to the synergisticBBB permeabilization effect of polysorbate 80, majordamage to the BBB cannot be excluded. Beyond theongoing controversy about their mechanism of action,polysorbate 80-coated PBCA nanoparticles should beevaluated in term of benefit/risk ratio and of innovativetherapeutics. In addition to the toxicity issue, the shortduration of the pharmacological effect observed afteradministration of drugs formulated with this carrier (210min at the best39) would probably necessitate daily in-travenous administrations, a perspective not suitable forthe treatment of chronic brain diseases.

PEGYLATED PLA OR PLGANANOPARTICLES

General considerationsAmong the few biodegradable polymers, polymers de-

rived from glycolic acid and from D,L-lactic acid enanti-omers are presently the most attractive compounds be-cause of their biocompatibility and their resorbabilitythrough natural pathways.40,41 They are widely used forthe preparation of biodegradable medical devices and ofdrug-sustained release microspheres or implants mar-keted in Europe, Japan, and the U.S.42 Degradation ofPLA or PLGA occurs by autocatalytic cleavage of theester bonds through spontaneous hydrolysis into oli-gomers and D,L-lactic and glycolic acid monomers.43

Lactate converted into pyruvate and glycolate enter theKrebs’ cycle to be degraded into CO2 and H2O. Afterintravenous administration of 14C-PLA18000 radiolabelednanoparticles to rats, 90% of the recovered 14C waseliminated within 25 days, among which 80% was asCO2.44 Degradation rate depends on four basic parame-ters: hydrolysis rate constant (depending on the molec-ular weight, the lactic/glycolic ratio, and the morphol-ogy), amount of water absorbed, diffusion coefficient ofthe polymer fragments through the polymer matrix, andsolubility of the degradation products in the surroundingaqueous medium.40,41 All of these parameters are influ-enced by temperature, additives (including drug mole-cules), pH, ionic strength, buffering capacity, size andprocessing history, steric hindrance etc. Despite a higherwater uptake the PLA or PLGA blocks of mPEG-PLA/PLGA block copolymers have similar degradation be-haviors.45,46 mPEG blocks are released (10–25% within3 days and 30–50% within 20 days at pH 7.4, 37°C) aftercleavage of the ester bonds,47–49 and, in the range ofmolecular weights of 1000–20,000, are mainly excretedvia the kidney.50 Up to an extensive PLA/PLGA polymerdegradation, nanoparticle morphology and size are gen-erally preserved.48,51 Generally considered as biocom-patible,41 PLA or PLGA microspheres have also a good

JEAN-CHRISTOPHE OLIVIER110


CNS biocompatibility.52,53 No mortality was reportedwith albumin-coated nanoparticles in mice with up to a2000 mg/kg dose.44 However, PLA60000 nanoparticlesstabilized with sodium cholate were much more toxicwith two of five deaths at a 220 mg/kg dose and five offive at a 440 mg/kg dose associated with marked clinicalsigns (dyspnea, reduced locomotor activity), alteration ofhematological and biochemical parameters and lunghemorrhage.54 This toxicity was attributed to a dissem-inated intravascular coagulation and associated eventsrelated to the physical surface properties of the nanopar-ticles rather than to the chemical toxicity of cholate orPLA. In contrast, mPEG2000-PLA30000 nanoparticleswere shown to have a good safety profile, with no ap-parent signs of toxicity at the highest studied dose of 440mg/kg in mice.54

Nanoparticle preparationNanoparticles made of mPEG-PLA/PLGA copolymers

are mainly prepared using the emulsion/solvent evapo-ration technique or the precipitation solvent diffusiontechnique.1 In the first method, copolymers are dissolvedin an organic solvent immiscible to water (such as di-chloromethane, chloroform, ethylacetate) and emulsifiedin an aqueous phase generally containing an emulsifyingagent (mainly polyvinylalcohol and sodium cholate).Then the solvent is evaporated off under normal or lowpressure to form nanoparticles. Hydrophobic compounds(drug or else) to be incorporated are dissolved in theorganic phase. Hydrosoluble compounds are first dis-solved in water and emulsified in the polymer-dissolvingorganic phase. The primary water-in-oil emulsion thusformed is then processed like the organic polymer phasedescribed above. This variant of the first method is called[(water-in-oil) in water] (or multiple emulsion) solventevaporation technique. In the second method, polymersare dissolved in an organic solvent miscible to water(such as acetone or ethanol) and dispersed in an aqueousphase generally containing a colloid stabilizer. The al-most instantaneous diffusion of the organic solvent intothe aqueous phase results in the precipitation of the co-polymers as nanoparticles. Finally, the solvent is evap-orated off as above or extracted by dialysis against wa-ter.55 In principle, only compounds soluble in the organicsolvent can be incorporated using the second method.Both basic methods require formulation optimization de-pending on the type of polymers/copolymers used, theirmolecular weights, the compound to be incorporated, thenanoparticle size to be achieved, etc.56–59 Other lessfrequently used methods include the emulsion solventdiffusion in an oil phase60,61 and the salting out pro-cess.62,63 Because of their different water solubility, thehydrophobic PLA/PLGA and hydrophilic PEG blocks ofthe mPEG-PLA/PLGA copolymer tend to phase-separatein the presence of water. Therefore, during the organic

solvent evaporation or diffusion, the PEG moieties mi-grate toward the aqueous phase, whereas the hydrophobicPLA/PLGA moieties aggregate as the nanoparticle core.mPEG-PLA copolymers with relatively high PEG to PLAweight ratio (e.g., mPEG5000-PLA2000-3000) may self-as-semble as polymeric micelles.59,64–66 Depending on thecopolymer solubility in water, polymeric micelles may beprepared either by self-dispersion in water (mPEG5000-PLA1500-2000

65,67) or by the precipitation/solvent evapora-tion technique using a classical solvent extraction procedure(mPEG5000-PLA3000-110900

67) or by dialysis.55 Self-dispers-ing mPEG-PLA copolymers are also used as emulsion sta-bilizers in the preparation of PLA nanoparticles.57 Thesize of mPEG-PLA nanoparticles prepared with constantPEG5000 was found to increase with the PLA block mo-lecular weight.59 With mPEG5000-PLA2000-30000 nano-particle diameters (from 26 to 64 nm in diameter) wereshown to be independent of the copolymer concentrationin the organic phase, whereas with higher PLA blockmolecular mass (45,000 Da) nanoparticle size was de-pendent on the copolymer concentration in the organicphase.59 After preparation, nanoparticles can be freeze-dried in the presence of appropriate cryoprotector for long-term preservation.51,62,63

Pegylated nanoparticle structureNanoparticles prepared from mPEG-PLA/PLGA co-

polymers are constituted of a PLA/PLGA hydrophobiccore surrounded by a hydrophilic PEG corona or outershell. In mPEG5000-PLA2000-75000 nanoparticles, negligi-ble penetration of the PEG into the solid-like PLA corewas reported, whereas as much of 25% PEG is entrappedwithin the nanoparticle core in the case of mPEG5000-PLA110000.59 It is likely that the [(water-in-oil) in water]solvent evaporation technique increases PEG entrapment,compared with the precipitation/solvent evaporation tech-nique.46 The water content of mPEG5000PLA45000 nanopar-ticles (200 nm diameter) is around 30% compared witharound 10% for PLA nanoparticles.46 At room tempera-ture and 37°C, a solid-like central core and more mobileinterfacial region coexist within the PLA core of nano-particles made of mPEG5000-PLA [glass transition tem-perature of around 333K], whereas the PEG corona layersituated on the nanoparticle surface is in the liquidphase.67 The PLA chain packing density increases withthe PLA molecular weights due to an increase in thenumber of attractive hydrophobic interactions betweenlactic acid units.59 In nanoparticles made of mPEG5000-PLA2000–3000, PLA chains possess some mobility.59 Be-cause of the relatively high critical micellar concentra-tion, these nanoparticles may dissociate upon dilution inblood.65 PEG conformation at the PLA-PEG nanopar-ticle surface is of utmost importance for the opsonin-repelling function of the PEG layer and has been exten-sively studied.57–59,66,68 The PEG layer thickness



depends on the PEG molecular weight and surface den-sity.57 Depending on their surface density, PEG blockshave brush-like (elongated coil, high density) or mush-room-like (random coil, low density) conformations.66,68

PEG surfaces in brush-like and intermediate configura-tions reduced phagocytosis and complement activation,whereas PEG surfaces in mushroom-like configurationwere potent complement activators and favored phago-cytosis.32,47,69,70 Based on the Alexander-de Gennesmodel, the distance between PEG chains should bearound 1 nm to repel small globular proteins (approxi-mately 2 nm radius) and 1.5 nm to repel large ones (6-8nm).32 Due to the large choice in the PLA or PEGmolecular weights available, the conformation of PEGblocks at the PEG-PLA nanoparticle surface is a com-plex issue to be addressed. At nanoparticle surface, thearea available per PEG chain at the outer boundary of theshell is dependent on PEG to PLA molecular weight ratiothat governs the PLA packing density and the surfacecurvature (linked to the nanoparticle size) of the assem-bly.57,58,71 As an example, an increase in the diameter ofnanoparticles made of mPEG5000-PLA45000 results in alower surface curvature, thus in an apparent increasein PEG surface coverage59 and in an improved colloidalstability.58

PharmacokineticsLike any colloidal drug carrier not especially designed

to escape from MPS uptake, PLA or PLGA nanoparticlesare rapidly removed from the blood stream after vascularadministration and preferentially accumulate in liver andspleen.44,72 Blood half-lives are generally around 2-3min.44,73–75 After intravenous administration, the firststep of the process that leads to the nanoparticle uptakeby the MPS is the opsonization phenomenon. Opsonins,including complement proteins, apolipoproteins, fi-bronectin, and Igs,31 interact with specific membranereceptors of monocytes and tissue macrophages, result-ing in recognition and phagocytosis. It is generally ad-mitted that hydrophobic surfaces promote protein ad-sorption and that negative surfaces are activators of thecomplement system.76 Following the rule hydrophobicand negative PLA or PLGA nanoparticle surfaces57,58

activate the complement system32 and coagulation fac-tors77 in vitro. In contrast, hydrophilic coating with PEGsterically stabilizes PLA or PLGA nanoparticles and re-duces opsonization and phagocytosis in vitro32 or exvivo,78 and uptake by neutrophilic granulocytes in vivo.79

Compared with nonpegylated PLA nanoparticles, pegy-lated nanoparticle surfaces have lower negative � poten-tial values, due to the surface shielding by the PEGcorona.3,57,58 mPEG2000-PLA nanoparticles did not acti-vate the complement47 and the coagulation77 systems invitro and did not alter coagulation parameters in vivo.54

Gref et al.32 showed a maximum antiopsonic effect with

PEG molecular weights of 5000 and above. Covalentlinkage of the PEG coating and sufficient PLA blockmolecular weight is essential to ensure a sufficient sta-bility and to avoid loss of the coating benefit bydesorption and/or displacement in vivo.57,72,73,80 Inmice, blood circulation times of 111In-labeled mPEG-PLGA5000-20000 nanoparticles (140 � 10 nm diameter)increased compared to PLGA ones with an advantage tothe higher PEG molecular weight.81 Within 5 min, how-ever, �50% (PEG20000) to 75% (PEG5000) of injectednanoparticles (estimated from the blood clearancecurves) had been cleared from the blood compartment(compared with 95% with control PLGA nanoparticles).In another study performed in rats, the blood half-lives of[14C]PLA-labeled mPEG-PLA30,000 nanoparticles withPEG molecular weight of 200073 (205 nm diameter) or500078 (140 � 60 nm) were markedly higher (6 h) andindependent of the PEG molecular weights. Less prolongedblood circulation times were observed with PLGA nano-particles coated with PLA3000-PEG4000 (147 � 3.6nm) orPLA3000-PEG5000 (161 � 3.7nm) (T[1/2] � 15 min andT[1/2] � 1 h, respectively, estimations from the bloodclearance curves).72 With nanoparticles made ofmPEG5000-PLA7000-125I (150 � 2nm diameter) or ofmPEG14000-PLA6000-125I (35.8 � 0.5nm) blood half-lives determined in rats were 29.9 � 12.4 and 42.3 �16.2 min respectively (no statistical difference).82 In rats,a blood half-life of 270.9 min was determined for 125I-BSA loaded in mPEG5000-PLGA45000 nanoparticles(around 200 nm diameter), compared with 13.6 minwhen formulated in PLGA nanoparticles.75 The largevariability in blood half-lives determined in those works,even with the same PEG block molecular weight of5000, may be ascribed to the above discussed density-related PEG conformation in the coating layer. The poly-dispersity of the PLA block molecular weights should bealso considered, which renders the pegylated nanopar-ticle system more complex than liposomes (the molecu-lar weight of the hydrophobic moieties of the pegylatedphospholipids are constant and the fluidity of the lipidicmembrane permits a statistically homogeneous distribu-tion of pegylated phospholipids) and could lead to asurface heterogeneity pointed out by Gbadamosi et al.69

Such a surface heterogeneity may explain the rapid clear-ance of a significant fraction of intravenously injectedlong-circulating nanoparticles by the MPS.72,81,83 Be-cause of this polydispersity, space available for PEGblock expansion is likely to be variable on nanoparticlesurface. Mushroom-like and brush-like conformationsmay coexist within a single nanoparticle or among apopulation of polydispersed nanoparticles (the size ofmicellar-like mPEG-PLA nanoparticles and therefore thePEG conformation in the corona are dependent on thePLA molecular weight, see above), thus explaining vari-ability observed in blood half-lives. Therefore, molecular



weights of PEG and PLA block, as well as polydispersityof copolymers, should be carefully selected in designinglong-circulating pegylated nanoparticles.

Drug loadingConventional drugs and general principles. Various

kinds of conventional drugs were formulated as PLA, PLGA,or mPEG-PLA nanoparticles. Examples are savoxepine,84

doxorubicin,85 irinotecan,86 paclitaxel,87,88antiestrogenRU58668,89 tyrphostin AG-1295,90 lidocaine,91 propranololhydrochloride,92 heparin,93 and enalaprilat.94 Basically, drugentrapment efficiency depends on the solid-state drugsolubility in PLA/PLGA polymer (solid dissolution ordispersion), which is related to the polymer composition(lactic/glycolic ratio), the molecular weight, the drug-polymer interaction and the presence of end-functionalgroups (ester or carboxyl).95–99 The PEG moiety has noor little effect on drug loading.91 Because PLA andPLGA are hydrophobic polymers, lipophilic drugs areeasier to formulate (in dissolved state) in PLA/mPEG-PLA nanoparticles, than hydrosoluble ones (segregationin separate domains). Despite the [(water-in-oil) in wa-ter] solvent evaporation technique, the entrapment ofhydrophilic drugs may be a challenge due to the drugdiffusion from the inner to the outer aqueous phasespromoted by the large surface area developed. Nanopar-ticle formulators have nevertheless several means to op-timize drug encapsulation: the selection of the prepara-tion procedure,61,84,87,100 the use of additives,96,97 the pHoptimization of the aqueous phases,92 the use of union-ized base or acid form of drugs,84,86,96,97 the PLA/PLGAblock polymer molecular weight. The incorporation ofcarboxylic groups to mPEG-PLA55 or the drug chemicalconjugation via cleavable linkage101 may be interestingalternatives to improve drug loading efficiency and ad-just release rates. Early drug release during storage maybe solved by freeze-drying. Drug entrapment efficiencycan reach more than 80%84,92,91 and drug content up to50%.91 In most cases, however, drug contents are 5-10%(wt/wt) of nanoparticle weights86,88,94,96,97,102 or evenless.87 Therefore, when formulating drug nanoparticles,it should always be kept in mind that generally as high as90% of the material to be administered will likely benanoparticle excipients with their potential toxicity.Drugs with high intrinsic pharmacological activitiesshould be preferred to avoid the administration of mas-sive dose of nanoparticle material. Drug release frombiodegradable polymeric nanoparticles depends on theFickian diffusion through the polymer matrix and on thedegradation rate of the polymer. The prediction of therelease profile is complex because it results from a com-bined effect of various parameters: solid-state drug poly-mer solubility98 and drug-polymer interactions,55,91,92,100

polymer degradation rate,61 block copolymer molecularweight and polydispersity,103 PEG content and molecular

weight,89,91,103 water uptake by nanoparticles48 and drugsolubility in the biological medium. In most studies, invitro release profiles are characterized by an initial fastrelease (burst) of drug close to or at the surface followedby a sustained release.91,92,103 Removing the low molec-ular weight fraction from the polymer was shown toreduce the initial burst of drug release.103 Depending onformulations in vitro, drug releases last from a fewhours84,91,92 or a few days87 to several weeks.61,84,88,90

Administered locally, betamethasone sodium phosphate-loaded PLGA nanoparticles were efficient at controllinginflammation over at least 3 weeks in a rabbit model ofarthritis, compared with one day for the solution.61

Peptides, polypeptides, and protein drugs. Certainlyone of the most promising, and challenging, applicationsof nanoparticles in brain delivery are the sustained re-lease of therapeutic peptides and proteins. Due to theirhydrosolubility the preparation method is generallybased on the [(water-in-oil)-in water] solvent evapora-tion technique.75,104,105 Entrapment efficiencies gener-ally range from 10% to 90%,75,104,106 and nanoparticlecontents from 1% or less99,107,108 to more than 15%.106

Apart from formulation issues inherent to peptide chem-ical instability or chemical reaction between peptides andpolymer degradation products,109 the formulation of pep-tide-loaded nanoparticles is similar to conventionaldrugs.99,107,108 Proteins, however, are highly organized,complex structures that have to be preserved to maintainbiological activity (receptor binding, antigenicity, enzy-matic activity, etc.). The general issues of the proteinstability and assessment and stabilization methods inPLA or PLGA delivery systems have been extensivelyreviewed.110–112 Structural and chemical integrity arelost during nanoparticle preparation and storage by pro-tein exposure to damaging conditions, such as interfaces(aqueous/organic in emulsions, hydrophobic surfaces ofpolymers), elevated temperatures (e.g., by sonication),shear force (e.g., sonication, vigorous stirring, extrusion,high pressure homogenization process), surfactants,(freeze-) drying etc.110 Moreover, upon administration,proteins are exposed to physiological temperature andacidic by-products of PLA/PLGA polymer degradationwithin nanoparticles for long time periods that can alsoaffect their stability.110,112 The study of the physical andchemical structure of the entrapped protein accompaniedwith an appropriate evaluation of the biological activityof the released material is the only way to confirm themaintenance of the protein integrity and activity.112 Eachnanoparticle formulation of protein is unique and re-quires specific adaptation and evaluation. Improved pro-tein stability was achieved by altering preparationprocesses,113 by changing polymer/copolymer,105,114

by changing or mixing solvents,49,113 by adding protec-tive additives110,111 such as hydrophilic polymers(PEG115,116), surfactants (poloxamer 188104,117), pro-



teins (serum albumin,118 gelatin105), cyclodextrins118,119

to the inner aqueous phase. Such formulation optimiza-tions permitted sustained release of active protein overseveral weeks in vitro.105,114

Plasmid DNA, oligonucleotides. Plasmid DNA-loaded nanoparticles are generally prepared using the[(water-in-oil)-in water] solvent evaporation tech-nique.120,121 The plasmid DNA loading, release rates,and transfection efficiency were shown to be dependentof the nature and the molecular weight of the polymer,122

the nanoparticle size121 and the colloid stabilizer.122 Anin vitro plasmid gene sustained release over severalweeks was achieved with PLGA123 or mPEG-PLA nano-particles.120 PLGA nanoparticles were shown to be en-docytosed by cells in vitro124. After endocytosis, PLGAnanoparticles escape from the endolysosomal compart-ment to the cytoplasm and gradually release their con-tent, resulting in sustained gene expression.125,126 In a ratosteotomy model, PLGA nanoparticles administered inthe bone-gap tissue permitted a plasmid gene expressionfor at least 5 weeks demonstrating their sustained releaseproperties.123 Like PLGA nanoparticles with an impor-tant poly(vinyl alcohol) coating,127 pegylated nanopar-ticles may interact poorly with cells, which may result inlow, or even no gene expression. Such a problem may beovercome with appropriate targeting ligands able to trig-ger endocytosis.128

Oligonucleotides were successfully encapsulatedwithin PLA129–131 or mPEG-PLA132 nanoparticles. Invitro-sustained release and intracellular delivery weredemonstrated.131,133

Perspectives in brain targeting. The most achievedwork in the field of brain targeting with colloidal drugcarriers has been carried out with pegylated immunoli-posomes that access the brain from blood via receptor-mediated transcytosis and deliver their content (smalldrug molecules, plasmid) into the brain parenchyma,without damaging the BBB.134–137 This requires thepresence of receptor-specific targeting ligands at the tipof 1-2% of the PEG2000 strands. Targeting ligands arepeptidomimetic monoclonal antibodies, i.e., able to trig-ger the activation of receptors (transferrin or insulin re-ceptors) that are highly expressed on the brain capillaryendothelium.134,136,137 These antibodies directed againstexternal receptor epitopes do not interfere with the nat-ural ligand binding sites, thus avoiding competition. Col-loidal carriers should have diameter less than 100 nm tofit the loading capacity of these transport systems. Be-cause immunoliposomes are not able of sustained releaseof transported compounds, as shown by the relativelyshort-lasting plasmid expression in brain,136 they requirefrequent administrations to sustain a pharmacologicaleffect.138 Pegylated PLA immunonanoparticles with sus-tained release properties may offer an interesting alter-native. Because of the presence of unreactive methoxy

terminal groups, mPEG-PLA copolymers do not permitligands to be tethered to the PEG chain. The covalentconjugation of protein ligands to pegylated nanoparticlesrequires chemically reactive functions at the free end of1-2% of the PEG strands of the PEG corona. Severalfunctionalized copolymers have been recently synthe-sized: the biotinylated,139 the amine-reactive64,140 andthe thiol-reactive copolymers141,142 that permit proteinchemical conjugation in nondenaturing conditions143

(FIG. 2). They are generally synthesized by ring openingpolymerization starting from heterobifunctional PEG andlactide and/or glycolide.64,140–142 Polymer block conju-gation is an alternative method.144 Biotinylated PEG-PLA nanoparticles may link biotinylated antibodiesthrough an avidin spacer145 (FIG. 3, panel 1a), or avidin-antibody conjugates146 (FIG. 3, panel 1b). Amine-reac-tive PEG-PLA (succinimide and aldehyde derivatives)can directly react with �-amino groups of the lysine

FIG. 2. Structure of functionalized PEG-PLA. Biotin-PEG-PLA(a); succinimidyl tartrate PEG-PLA (b), succinimidyl succinatePEG-PLA (c), aldehyde-PEG-PLA (d), maleinimido propionatePEG-PLA (e), and maleimide-PEG-PLA (f).



residues of antibodies in mild conditions (FIG. 3, panel2). An �-acetal-PEG-PLA block copolymer is requiredto prepare aldehyde-functionalized PEG-PLA nanopar-ticles64,147,148 (FIG. 3, panel 2b). After nanoparticlepreparation, the acetal groups are converted by mild acidtreatment (pH 2) into aldehyde functions that are reactivewith amine of peptidyl ligand at pH 7.147,149 Antibodiesmay be chemically linked through Schiff base formationand successive reductive amination using NaBH3CN.147

Due to the lack of free thiol, antibody conjugation tothiol-reactive functions (maleimide) requires the intro-duction of thiol residues by reacting 2-iminothiolane(Traut’s reagent) with �-amino groups of the lysine res-idues. The thiolation was shown not to interfere withtarget recognition.150 In mild conditions that preserveantibody reactivity, a stable thioether bond can be estab-lished between maleimide and thiol (FIG. 3, panel 3).Such a method was successfully applied to the prepara-tion of brain-targeted immunoliposomes.134 In a recentwork, we used the same procedure to design brain-tar-geted pegylated immunonanoparticles.142 Maleimide-functionalized pegylated nanoparticles were preparedwith maleimide-PEG3500-PLA40000 and mPEG2600-PLA40000 (according to a 1:40 molar ratio) using the[(water-in-oil) in water] solvent evaporation technique.Thiolated mouse OX26 anti-rat transferrin receptormonoclonal antibodies were then successfully conju-gated to the functionalized nanoparticles. The meannumber of antibodies per nanoparticles was determinedto be 67 and visualized at the nanoparticle surface bytransmission electron microscopy after labeling with ananti-mouse IgG antibody gold conjugate (FIG. 4).

CONCLUSION

Even though being effective at delivering drug to thebrain by a still-debated mechanism, polysorbate 80-coated PBCA nanoparticles may have limited clinicalapplications due to a potential toxicity, BBB permeabi-lization, and short lasting delivery. Technology now ex-

FIG. 4. Transmission electron micrograph of pegylated immu-nonanoparticles negatively stained with phosphotungstic acidsolution. Antibodies conjugated to the nanoparticle are revealedby binding with a 10-nm gold-labeled secondary antibody. Themagnification bar is 15 nm. Reprinted with permission fromOlivier et al. Synthesis of pegylated immunonanoparticles.Pharm Res 19:1137–1143. Copyright � 2002, Kluwer AcademicPublishers, with kind permission of Springer Science and Busi-ness Media. All rights reserved.142

FIG. 3. Currently available conjugation techniques to preparepegylated PLA immunonanoparticles. For comments, see text.



ists to prepare safe brain-targeted long-circulating nano-particles, the pegylated PLA immunonanoparticles,capable of sustained drug release. Their physicochemicaland biological properties and methods of preparationhave been extensively described. Various drug mole-cules, including proteins, plasmid DNA, and oligonucle-otides, were formulated and preservation of activity wasdemonstrated. A long way of optimization and evalua-tion is still, however, needed before potential clinicalapplication. Providing PLA nanoparticles with stealthproperties is a complex issue that involves the optimiza-tion of combined parameters, such as PEG molecularweight, PEG/PLA molecular weight ratio, and nanopar-ticle size. Stealth properties and BBB transportation ofimmunonanoparticles, as well as effective drug release inthe brain parenchyma, remain to be investigated.

REFERENCES

1. Soppimath KS, Aminabhavi TM, Kulkarni AR, Rudzinski WE.Biodegradable polymeric nanoparticles as drug delivery devices.J Control Release 70:1–20, 2001.

2. Wissing SA, Kayser O, Muller RH. Solid lipid nanoparticles forparenteral drug delivery. Adv Drug Deliv Rev 56:1257–1272,2004.

3. Gref R, Domb A, Quellec P, Blunk T, Muller RH, Verbavatz JM,et al. The controlled intravenous delivery of drugs using PEG-coated sterically stabilized nanospheres. Adv Drug Deliv Rev16:215–233, 1995.

4. Lockman PR, Mumper RJ, Khan MA, Allen DD. Nanoparticletechnology for drug delivery across the blood-brain barrier. DrugDev Ind Pharm 28:1–13, 2002.

5. Pardridge WM. Brain drug targeting: the future of brain drugdevelopment. Cambridge, UK: Cambridge University Press, Inc.,2001.

6. Alyautdin R, Gothier D, Petrov V, Kharkevich D, Kreuter JAnalgesic activity of the hexapeptide dalargin adsorbed on thesurface of polysorbate 80-coated poly(butyl cyanoacrylate) nano-particles. Eur J Pharm Biopharm 41:44–48, 1995.

7. Kreuter J, Alyautdin RN, Kharkevich DA, Ivanov AA. Passage ofpeptides through the blood-brain barrier with colloidal polymerparticles (nanoparticles). Brain Res 674:171–174, 1995.

8. Couvreur P, Tulkens P, Roland M, Trouet A, Speiser P. Nano-capsules: a new type of lysosomotropic carrier. FEBS Lett 84:323–326, 1977.

9. Vauthier C, Dubernet C, Fattal E, Pinto-Alphandary H, CouvreurP. Poly(alkylcyanoacrylates) as biodegradable materials for bio-medical applications. Adv Drug Deliv Rev 55:519–548, 2003.

10. Guise V, Drouin JY, Benoit J, Mahuteau J, Dumont P, CouvreurP. Vidarabine-loaded nanoparticles: a physicochemical study.Pharm Res 7:736–741, 1990.

11. Olivier JC, Fenart L, Chauvet R, Pariat C, Cecchelli R, Couet W.Indirect evidence that drug brain targeting using polysorbate 80-coated polybutylcyanoacrylate nanoparticles is related to toxicity.Pharm Res 16:1836–1842, 1999.

12. Olivier JC, Vauthier C, Taverna M, Puisieux F, Ferrier D Cou-vreur P. Stability of orosomucoid-coated polyisobutylcyanoacry-late nanoparticles in the presence of serum. J Control Release40:157–168, 1996.

13. Muller RH, Lherm C, Herbort J, Couvreur P. In vitro model forthe degradation of alkylcyanoacrylate nanoparticles. Biomaterials11:590–595, 1990.

14. Muller R H, Lherm C, Herbort J, Blunk T, Couvreur P. Alkyl-cyanoacrylate drug carriers: I. Physicochemical characterizationof nanoparticles with different alkyl chain length. Int J Pharm84:1–11, 1992.

15. Lherm C, Muller RH, Puisieux F, Couvreur P. Alkylcyanoacry-late drug carriers: II. Cytotoxicity of cyanoacrylate nanoparticleswith different alkyl chain length. Int J Pharm 84:13–22, 1992.

16. Kante B, Couvreur P, Dubois-Krack G, De Meester C, Guiot P,Roland M, et al. Toxicity of polyalkylcyanoacrylate nanoparticlesI: free nanoparticles. J Pharm Sci 71:786–790, 1982.

17. Kattan J, Droz JP, Couvreur P, Marino JP, Boutan-Laroze A,Rougier P, et al. Phase I clinical trial and pharmacokinetic eval-uation of doxorubicin carried by polyisohexylcyanoacrylatenanoparticles. Invest New Drugs 10:191–199, 1992.

18. Grislain L, Couvreur P, Lenaerts V, Roland M, Deprez-De-campeneere D, Speiser P. Pharmacokinetics and distribution of abiodegradable drug-carrier. Int J Pharm 15:335–345, 1983.

19. Douglas S J, Davis SS, Illum L. Biodistribution of poly(butyl2-cyanoacrylate) nanoparticles in rabbits. Int J Pharm 34:145–152, 1986.

20. Waser PG, Muller U, Kreuter J, Berger S, Munz K, Kaiser E, etal. Localization of colloidal particles (liposomes, hexylcyanoac-rylate nanoparticles and albumin nanoparticles) by histology andautoradiography in mice. Int J Pharm 39:213–227, 1987.

21. Simeonova M, Ivanova T, Raikova E, Georgieva M, Raikov Z.Tissue distribution of polybutylcyanoacrylate nanoparticles car-rying spin-labelled nitrosourea. Int J Pharm 43:267–271, 1988.

22. Verdun C, Brasseur F, Vranckx H, Couvreur P, Roland M. Tissuedistribution of doxorubicin associated with polyisohexylcyanoac-rylate nanoparticles. Cancer Chemother Pharmacol 26:13–18,1990.

23. Lobenberg R, Araujo L, von Briesen H, Rodgers E, Kreuter JBody distribution of azidothymidine bound to hexyl-cyanoacry-late nanoparticles after i.v. injection to rats. J Control Release50:21–30, 1998.

24. Peracchia MT, Fattal E, Desmaele D, Besnard M, Noel JP, GomisJM, et al. Stealth PEGylated polycyanoacrylate nanoparticles forintravenous administration and splenic targeting. J Control Re-lease 60:121–128, 1999.

25. Gulyaev AE, Gelperina SE, Skidan IN, Antropov AS, KivmanGY, Kreuter J Significant transport of doxorubicin into the brainwith polysorbate 80-coated nanoparticles. Pharm Res 16:1564–1569, 1999.

26. Steiniger SC, Kreuter J, Khalansky AS, Skidan IN, Bobruskin AI,Smirnova ZS, et al. Chemotherapy of glioblastoma in rats usingdoxorubicin-loaded nanoparticles. Int J Cancer 109:759–767,2004.

27. Alyautdin RN, Petrov VE, Langer K, Berthold A, KharkevichDA, Kreuter J. Delivery of loperamide across the blood-brainbarrier with polysorbate 80-coated polybutylcyanoacrylate nano-particles. Pharm Res 14:325–328, 1997.

28. Alyautdin RN, Tezikov EB, Ramge P, Kharkevich DA, BegleyDJ, Kreuter J Significant entry of tubocurarine into the brain ofrats by adsorption to polysorbate study. J Microencapsul 15:67–74, 1998.

29. Kreuter J. Nanoparticulate systems for brain delivery of drugs.Adv Drug Deliv Rev 47:65–81, 2001.

30. Kreuter J, Petrov VE, Kharkevich DA, Alyautdin RN. Influenceof the type of surfactant on the analgesic effects induced by thepeptide dalargin after its delivery across the blood-brain barrierusing surfactant-coated nanoparticles. J Control Release 49:81–87, 1997.

31. Allemann E, Gravel P, Leroux JC, Balant L, Gurny R. Kinetics ofblood component adsorption on poly(D,L-lactic acid) nanopar-ticles: evidence of complement C3 component involvement.J Biomed Mater Res 37:229–234, 1997.

32. Gref R, Luck M, Quellec P, Marchand M, Dellacherie E, Har-nisch S. ‘Stealth’ corona-core nanoparticles surface modified bypolyethylene glycol (PEG): influences of the corona (PEG chainlength and surface density) and of the core composition on phago-cytic uptake and plasma protein adsorption. Colloids Surfaces B:Biointerfaces 18:301–313, 2000.

33. Lode J, Fichtner I, Kreuter J, Berndt A, Diederichs JE, Reszka R.Influence of surface-modifying surfactants on the pharmacoki-netic behavior of 14C-poly (methylmethacrylate) nanoparticles inexperimental tumor models. Pharm Res 18:1613–1619, 2001.

34. Darius J, Meyer FP, Sabel BA, Schroeder U. Influence of nano-



particles on the brain-to-serum distribution and the metabolism ofvalproic acid in mice. J Pharm Pharmacol 52:1043–1047, 2000.

35. Schroeder U, Schroeder H, Sabel BA. Body distribution of 3H-labelled dalargin bound to poly(butyl cyanoacrylate) nanopar-ticles after i.v. injections to mice. Life Sci 66:495–502, 2000.

36. Calvo P, Gouritin B, Chacun H, Desmaele D, D’Angelo J, NoelJP, Georgin D, Fattal E, Andreux JP, Couvreur P. Long-circulat-ing PEGylated polycyanoacrylate nanoparticles as new drug car-rier for brain delivery. Pharm Res 18:1157–1166, 2001.

37. Azmin MN, Stuart JF, Florence AT. The distribution and elimi-nation of methotrexate in mouse blood and brain after concurrentadministration of polysorbate 80. Cancer Chemother Pharmacol14:238–242, 1985.

38. Alyaudtin RN, Reichel A, Lobenberg R, Ramge P, Kreuter J,Begley DJ. Interaction of poly(butylcyanoacrylate) nanoparticleswith the blood-brain barrier in vivo and in vitro. J Drug Target9:209–221, 2001.

39. Friese A, Seiller E, Quack G, Lorenz B, Kreuter J. Increase of theduration of the anticonvulsive activity of a novel NMDA receptorantagonist using poly(butylcyanoacrylate) nanoparticles as a par-enteral controlled release system. Eur J Pharm Biopharm 49:103–109, 2000.

40. Vert M, Schwach G, Engel R, Coudane J. Something new in thefield of PLA/GA bioresorbable polymers? J Control Release 53:85–92, 1998.

41. Anderson JM, Shive MS. Biodegradation and biocompatibility ofPLA and PLGA microspheres. Adv Drug Deliv Rev 28:5–24,1997.

42. Ueda H, Tabata Y. Polyhydroxyalkanonate derivatives in currentclinical applications and trials. Adv Drug Deliv Rev 55:501–518,2003.

43. Li S. Hydrolytic degradation characteristics of aliphatic polyes-ters derived from lactic and glycolic acids. J Biomed Mater Res48:342–353, 1999.

44. Bazile DV, Ropert C, Huve P, Verrecchia T, Marlard M, Fryd-man A, et al. Body distribution of fully biodegradable [14 C]-poly(lactic acid) nanoparticles coated with albumin after paren-teral administration to rats. Biomaterials 13:1093–1102, 1992.

45. von Burkersroda F, Gref R, Gopferich A. Erosion of biodegrad-able block copolymers made of poly(D,L-lactic acid) and poly-(ethylene glycol). Biomaterials 18:1599–1607, 1997.

46. Quellec P, Gref R, Perrin L, Dellacherie E, Sommer F, VerbavatzJM, Alonso MJ. Protein encapsulation within polyethylene gly-col-coated nanospheres. I. Physicochemical characterization.J Biomed Mater Res 42:45–54, 1998.

47. Vittaz M, Bazile D, Spenlehauer G, Verrecchia T, Veillard M,Puisieux F, et al. Effect of PEO surface density on long-circulat-ing PLA-PEO nanoparticles which are very low complementactivators. Biomaterials 17:1575–1581, 1996.

48. Quellec P, Gref R, Dellacherie E, Sommer F, Tran MD, AlonsoMJ. Protein encapsulation within polyethylene glycol-coatednanospheres. II. Controlled release properties. J Biomed MaterRes 47:388–395, 1999.

49. Zambaux MF, Bonneaux F, Gref R, Dellacherie E, Vigneron C.Protein C-loaded monomethoxypoly (ethylene oxide)-poly(lacticacid) nanoparticles. Int J Pharm 212:1–9, 2001.

50. Working PK, Newman MS, Johnson J, Cornacoff JB. Safety ofpoly(ethylene glycol) and poly(ethylene glycol) derivatives. In:Poly(ethylene glycol): chemistry and biological applications(Harris JM, Zalipsky S, eds.) ACS Symposium Series, No 680, pp45-57. Washington, DC: American Chemical Society, 1997.

51. Lemoine D, Francois C, Kedzierewicz F, Preat V, Hoffman M,Maincent P. Stability study of nanoparticles of poly(epsilon-ca-prolactone), poly(D,L-lactide) and poly(D,L-lactide-co-glycol-ide). Biomaterials 17:2191–2197, 1996.

52. Emerich DF, Tracy MA, Ward KL, Figueiredo M, Qian R, Hen-schel C, et al. Biocompatibility of poly (DL-lactide-co-glycolide)microspheres implanted into the brain. Cell Transplant 8:47–58,1999.

53. Menei P, Daniel V, Montero-Menei C, Brouillard M, Pouplard-Barthelaix A, Benoit JP. Biodegradation and brain tissue reactionto poly(D,L-lactide-co-glycolide) microspheres. Biomaterials 14:470–478, 1993.

54. Plard JP, Didier B. Comparison of the safety profiles of PLA50

and Me.PEG-PLA50 nanoparticles after single dose intravenousadministration to rat. Colloids Surfaces B: Biointerfaces 16:173–183, 1999.

55. Lee J, Cho EC, Cho K. Incorporation and release behavior ofhydrophobic drug in functionalized poly(D,L-lactide)-block-poly(ethylene oxide) micelles. J Control Release 94:323–335,2004.

56. Zambaux MF, Bonneaux F, Gref R, Maincent P, Dellacherie E,Alonso MJ, Labrude P, Vigneron C. Influence of experimentalparameters on the characteristics of poly(lactic acid) nanopar-ticles prepared by a double emulsion method. J Control Release50:31–40, 1998.

57. Chognot D, Six JL, Leonard M, Bonneaux F, Vigneron C, Della-cherie E. Physicochemical evaluation of PLA nanoparticles sta-bilized by water-soluble MPEO-PLA block copolymers. J Col-loid Interface Sci 268:441–447, 2003.

58. Riley T, Govender T, Stolnik S, Xiong CD, Garnett MC, Illum L,et al. Colloidal stability and drug incorporation aspects of micel-lar-like PLA-PEG nanoparticles. Colloids Surfaces B: Biointer-faces 16:147–159, 1999.

59. Riley T, Stolnik S, Heald CR, Xiong CD, Garnett MC, Illum L,et al. Physicochemical evaluation of nanoparticles assembledfrom poly(lactic acid)-poly(ethylene glycol) (PLA-PEG) blockcopolymers as drug delivery vehicles. Langmuir 17:3168–3174,2001.

60. Niwa T, Takeuchi H, Hino T, Nohara M, Kawashima Y. Biode-gradable submicron carriers for peptide drugs: preparation of-lactide/glycolide copolymer (PLGA) nanospheres with nafarelinacetate by a novel emulsion-phase separation method in an oilsystem. Int J Pharm 121:45–54, 1995.

61. Horisawa E, Hirota T, Kawazoe S, Yamada J, Yamamoto H,Takeuchi H, Kawashima Y. Prolonged anti-inflammatory actionof DL-lactide/glycolide copolymer nanospheres containing beta-methasone sodium phosphate for an intra-articular delivery sys-tem in antigen-induced arthritic rabbit. Pharm Res 19:403–410,2002.

62. Konan YN, Gurny R, Allemann E. Preparation and characteriza-tion of sterile and freeze-dried sub-200 nm nanoparticles. IntJ Pharm 233:239–252, 2002.

63. De Jaeghere F, Allemann E, Leroux JC, Stevels W, Feijen J,Doelker, E, et al. Formulation and lyoprotection of poly(lacticacid-co-ethylene oxide) nanoparticles: influence on physical sta-bility and in vitro cell uptake. Pharm Res 16:859–866, 1999.

64. Scholz C, Iijima M, Nagasaki Y, Kataoka K. A novel reactivepolymeric micelle with aldehyde groups on its surface. Macro-molecules 28:7295–7297, 1995.

65. Hagan SA, Coombes AGA, Garnett MC, Dunn SE, Davies MC,Illum L, et al. Polylactide-poly(ethylene glycol) copolymers asdrug delivery systems. 1. Characterization of water dispersiblemicelle-forming systems. Langmuir 12:2153–2161, 1996.

66. Stolnik S, Heald CR, Neal J, Garnett MC, Davis SS, Illum L, etal. Polylactide-poly(ethylene glycol) micellar-like particles as po-tential drug carriers: production, colloidal properties and biolog-ical performance. J Drug Target 9:361–378, 2001.

67. Heald CR, Stolnik S, Kujawinski KS, De Matteis C, Garnett MC,Illum L. Poly(lactic acid)-poly(ethylene oxide) (PLA-PEG) nano-particles: NMR studies of the central solidlike PLA core and theliquid PEG corona. Langmuir 18:3669–3675, 2002.

68. Gref R, Babak B, Bouillot P, Lukina I, Bodorev M, Dellacherie E.Interfacial and emulsion stabilising properties of amphiphilic wa-ter-soluble poly(ethylene glycol)-poly(lactic acid) copolymers forthe fabrication of biocompatible nanoparticles. Colloids SurfacesA 143:413–420, 1998.

69. Gbadamosi JK, Hunter AC, Moghimi SM. PEGylation of micro-spheres generates a heterogeneous population of particles withdifferential surface characteristics and biological performance.FEBS Lett 532:338–344, 2002.

70. Moghimi SM, Szebeni J. Stealth liposomes and long circulatingnanoparticles: critical issues in pharmacokinetics, opsonizationand protein-binding properties. Prog Lipid Res 42:463–478,2003.

71. Heald CR, Stolnik S, De Matteis C, Garnett MC, Illum L, Davis



SS, et al. Characterisation of poly(lactic acid):poly(ethyleneox-ide) (PLA:PEG) nanoparticles using the self-consistent theorymodelling approach. Colloids Surfaces A 212:57–64, 2003.

72. Stolnik S, Dunn SE, Garnett MC, Davies MC, Coombes AG,Taylor DC, et al. Surface modification of poly(lactide-co-glycol-ide) nanospheres by biodegradable poly(lactide)-poly(ethyleneglycol) copolymers. Pharm Res 11:1800–1808, 1994.

73. Verrecchia T, Spenlehauer G, Bazile DV, Murry-Brelier A,Archimbaud Y, Veillard M. Non-stealth (poly(lactic acid/albu-min)) and stealth (poly(lactic acid-polyethylene glycol)) nanopar-ticles as injectable drug carriers. J Control Release 36:49–61,1995.

74. Le Ray AM, Vert M, Gautier JC, Benoıt JP. Fate of [14C]poly(-lactide-co-glycolide) nanoparticles after intravenous and oral ad-ministration to mice. Int J Pharm 106:201–211, 1994.

75. Li Y, Pei Y, Zhang X, Gu Z, Zhou Z, Yuan W, et al. PEGylatedPLGA nanoparticles as protein carriers: synthesis, preparationand biodistribution in rats. J Control Release 71:203–211, 2001.

76. Moghimi SM, Hunter AC, Murray JC. Long-circulating and tar-get-specific nanoparticles: theory to practice. Pharmacol Rev 53:283–318, 2001.

77. Sahli H, Tapon-Bretaudiere J, Fischer AM, Sternberg C, Spenle-hauer G, Verrecchia T, et al. Interactions of poly(lactic acid) andpoly(lactic acid-co-ethylene oxide) nanoparticles with the plasmafactors of the coagulation system. Biomaterials 18:281–288,1997.

78. Bazile D, Prud’homme C, Bassoullet MT, Marlard M, Spenle-hauer G, Veillard M. Stealth Me.PEG-PLA nanoparticles avoiduptake by the mononuclear phagocytes system. J Pharm Sci84:493–498, 1995.

79. Zambaux MF, Faivre-Fiorina B, Bonneau F, Marchal S, MerlinJL, Dellacherie E, et al. Involvement of neutrophilic granulocytesin the uptake of biodegradable non-stealth and stealth nanopar-ticles in guinea pig. Biomaterials 21:975–980, 2000.

80. Mosqueira VC, Legrand P, Gulik A, Bourdon O, Gref R, LabarreD, et al. Relationship between complement activation, cellularuptake and surface physicochemical aspects of novel PEG-mod-ified nanocapsules. Biomaterials 22:2967–2979, 2001.

81. Gref R, Minamitake Y, Peracchia MT, Trubetskoy V, TorchilinV, Langer R. Biodegradable long-circulating polymeric nano-spheres. Science 263:1600–1603, 1994.

82. Novakova K, Laznicek M, Rypacek F, Machova L. Pharmacoki-netics and distribution 125 I-PLA-b-PEO block copolymers in rats.Pharm Dev Technol 8:153–161, 2003.

83. Moghimi SM. Chemical camouflage of nanospheres with a poorlyreactive surface: towards development of stealth and target-spe-cific nanocarriers. Biochim Biophys Acta 1590:131–139, 2002.

84. Allemann E, Leroux JC, Gurny R, Doelker E. In vitro extended-release properties of drug-loaded poly(DL-lactic acid) nanopar-ticles produced by a salting-out procedure. Pharm Res 10:1732–1737, 1993.

85. Yoo HS, Park TG. Biodegradable polymeric micelles composedof doxorubicin conjugated PLGA-PEG block copolymer. J Con-trol Release 70:63–70, 2001.

86. Onishi H, Machida Y, Machida Y. Antitumor properties of iri-notecan-containing nanoparticles prepared using poly(DL-lacticacid) and poly(ethylene glycol)-block-poly(propylene glycol)-block-poly(ethylene glycol). Biol Pharm Bull 26:116–119, 2003.

87. Dong Y, Feng SS. Methoxy poly(ethylene glycol)-poly(lactide)(MPEG-PLA) nanoparticles for controlled delivery of anticancerdrugs. Biomaterials 25:2843–2849, 2004.

88. Feng SS, Mu L, Win KY, Huang G. Nanoparticles of biodegrad-able polymers for clinical administration of paclitaxel. Curr MedChem 11:413–424, 2004.

89. Ameller T, Marsaud V, Legrand P, Gref R, Renoir JM. Pureantiestrogen RU 58668-loaded nanospheres: morphology, cell ac-tivity and toxicity studies. Eur J Pharm Sci 21:361–370, 2004.

90. Fishbein I, Chorny M, Rabinovich L, Banai S, Gati I, Golomb G.Nanoparticulate delivery system of a tyrphostin for the treatmentof restenosis. J Control Release 65:221–229, 2000.

91. Peracchia MT, Gref R, Minamitake Y, Domb A, Lotan N, LangerR. PEG-coated nanospheres from amphiphilic diblock and

multiblock copolymers: investigation of their drug encapsulationand release characteristics. J Control Release 46:223–231, 1997.

92. Ubrich N, Bouillot P, Pellerin P, Hoffman M, Maincent P. Prep-aration and characterization of propranolol hydrochloride nano-particles: a comparative study. J Control Release 97:291–300,2004.

93. Hoffart V, Ubrich N, Simonin C, Babak V, Vigneron C, HoffmanM, et al. Low molecular weight heparin-loaded polymeric nano-particles: formulation, characterization, and release characteris-tics. Drug Dev Ind Pharm 28:1091–1099, 2002.

94. Ahlin P, Kristl J, Kristl A, Vrecer F. Investigation of polymericnanoparticles as carriers of enalaprilat for oral administration. IntJ Pharm 239:113–120, 2002.

95. Nicoli S, Santi P, Couvreur P, Couarraze G, Colombo P, Fattal E.Design of triptorelin loaded nanospheres for transdermal ionto-phoretic administration. Int J Pharm 214:31–35, 2001.

96. Govender T, Stolnik S, Garnett MC, Illum L, Davis SS. PLGAnanoparticles prepared by nanoprecipitation: drug loading andrelease studies of a water soluble drug. J Control Release 57:171–185, 1999.

97. Govender T, Riley T, Ehtezazi T, Garnett MC, Stolnik S, Illum L,et al. Defining the drug incorporation properties of PLA-PEGnanoparticles. Int J Pharm 199:95–110, 2000.

98. Panyam J, Williams D, Dash A, Leslie-Pelecky D, LabhasetwarV. Solid-state solubility influences encapsulation and release ofhydrophobic drugs from PLGA/PLA nanoparticles. J Pharm Sci93:1804–1814, 2004.

99. Hiroaki O, Yamamoto M, Heya T, Inoue Y, Kamei S, Ogawa Y,et al. Drug delivery using biodegradable microspheres. J ControlRelease 28:121–129, 1994.

100. Niwa T, Takeuchi H, Hino T, Kunou N, Kawashima Y. In vitrodrug release behavior of D,L-lactide/glycolide copolymer(PLGA) nanospheres with nafarelin acetate prepared by a novelspontaneous emulsification solvent diffusion method. J PharmSci 83:727–732, 1994.

101. Yoo HS, Lee EA, Park TG. Doxorubicin-conjugated biodegrad-able polymeric micelles having acid-cleavable linkages. J ControlRelease 82:17–27, 2002.

102. Leo E, Brina B, Forni F, Vandelli MA. In vitro evaluation of PLAnanoparticles containing a lipophilic drug in water-soluble orinsoluble form. Int J Pharm 278:133–141, 2004.

103. Matsumoto J, Nakada Y, Sakurai K, Nakamura T, Takahashi Y.Preparation of nanoparticles consisted of poly(L-lactide)-poly-(ethyleneglycol)-poly(L-lactide) and their evaluation in vitro. IntJ Pharm 185:93–101, 1999.

104. Blanco MD, Alonso MJ. Development and characterization ofprotein-loaded poly(lactide-co-glycolide) nanospheres. EurJ Pharm Biopharm 43:287–294, 1997.

105. Tobio M, Gref R, Sanchez A, Langer R, Alonso MJ. StealthPLA-PEG nanoparticles as protein carriers for nasal administra-tion. Pharm Res 15:270–275, 1998.

106. Gref R, Quellec P, Sanchez A, Calvo P, Dellacherie E, AlonsoMJ. Development and characterization of CyA-loaded poly(lacticacid)-poly(ethylene glycol)PEG micro- and nanoparticles. Com-parison with conventional PLA particulate carriers. Eur J PharmBiopharm 51:111–118, 2001.

107. Kawashima Y, Yamamoto H, Takeuchi H, Hino T, Niwa T.Properties of a peptide containing DL-lactide/glycolide copoly-mer nanospheres prepared by novel emulsion solvent diffusionmethods. Eur J Pharm Biopharm 45:41–48, 1998.

108. Elamanchili P, Diwan M, Cao M, Samuel J. Characterization ofpoly(d,l-lactic-co-glycolic acid) based nanoparticulate system forenhanced delivery of antigens to dendritic cells. Vaccine 22:2406–2412, 2004.

109. Lucke A, Fustella E, Tessmar J, Gazzaniga A, Gopferich A. Theeffect of poly(ethylene glycol)-poly(D,L-lactic acid) diblock co-polymers on peptide acylation. J Control Release 80:157–168,2002.

110. van de Weert M, Hennink WE, Jiskoot W. Protein instability inpoly(lactic-co-glycolic acid) microparticles. Pharm Res 17:1159–1167, 2000.



111. Schwendeman SP. Recent advances in the stabilization of pro-teins encapsulated in injectable PLGA delivery systems. Crit RevTher Drug Carrier Syst 19:73–98, 2002.

112. Cleland JL. Protein delivery from biodegradable microspheres.Pharm Biotechnol 10:1–43, 1997.

113. Zambaux MF, Bonneaux F, Gref R, Dellacherie E, Vigneron C.Preparation and characterization of protein C-loaded PLA nano-particles. J Control Release 60:179–188, 1999.

114. Gaspar MM, Blanco D, Cruz MEM, Alonso MJ. Formulation of-asparaginase-loaded poly(lactide-co-glycolide) nanoparticles:influence of polymer properties on enzyme loading, activity andin vitro release. J Control Release 52:53–62, 1998.

115. Pean JM, Boury F, Venier-Julienne MC, Menei P, Proust JE,Benoit JP. Why does PEG 400 co-encapsulation improve NGFstability and release from PLGA biodegradable microspheres?Pharm Res 16:1294–1299, 1999.

116. Pean JM, Menei P, Morel O, Montero-Menei CN, Benoit JP.Intraseptal implantation of NGF-releasing microspheres promotethe survival of axotomized cholinergic neurons. Biomaterials 21:2097–2101, 2000.

117. Sanchez A, Tobio M, Gonzalez L, Fabra A, Alonso MJ. Biode-gradable micro- and nanoparticles as long-term delivery vehiclesfor interferon-�. Eur J Pharm Sci 18:221–229, 2003.

118. Sah H. Stabilization of proteins against methylene chloride/waterinterface-induced denaturation and aggregation. J Control Re-lease 58:143–151, 1999.

119. Morlock M, Koll H, Winter G, Kissel T. Microencapsulation ofrh-erythropoietin, using biodegradable poly(D,L-lactide-co-gly-colide): protein stability and the effects of stabilizing excipients.Eur J Pharm Biopharm 43:29–36, 1997.

120. Perez C, Sanchez A, Putnam D, Ting D, Langer R, Alonso MJ.Poly(lactic acid)-poly(ethylene glycol) nanoparticles as new car-riers for the delivery of plasmid DNA. J Control Release 75:211–224, 2001.

121. Prabha S, Zhou WZ, Panyam J, Labhasetwar V. Size-dependencyof nanoparticle-mediated gene transfection: studies with fraction-ated nanoparticles. Int J Pharm 244:105–115, 2002.

122. Prabha S, Labhasetwar V. Critical determinants in PLGA/PLAnanoparticle-mediated gene expression. Pharm Res 21:354–364,2004.

123. Labhasetwar V, Bonadio J, Goldstein SA, Levy RJ. Gene trans-fection using biodegradable nanospheres: results in tissue cultureand a rat osteotomy model. Colloids Surfaces B: Biointerfaces16:281–290, 1999.

124. Davda J, Labhasetwar V. Characterization of nanoparticle uptakeby endothelial cells. Int J Pharm 233:51–59, 2002.

125. Panyam J, Labhasetwar V. Biodegradable nanoparticles for drugand gene delivery to cells and tissue. Adv Drug Deliv Rev 55:329–347, 2003.

126. Panyam J, Zhou WZ, Prabha S, Sahoo SK, Labhasetwar V. Rapidendo-lysosomal escape of poly(DL-lactide-co-glycolide) nano-particles: implications for drug and gene delivery. FASEB J 16:1217–1226, 2002.

127. Sahoo SK, Panyam J, Prabha S, Labhasetwar V. Residual poly-vinyl alcohol associated with poly (D,L-lactide-co-glycolide)nanoparticles affects their physical properties and cellular uptake.J Control Release 82:105–114, 2004.

128. Zhang Y, Jeong Lee H, Boado RJ, Pardridge WM. Receptor-mediated delivery of an antisense gene to human brain cancercells. J Gene Med 4:183–194, 2002.

129. Delie F, Berton M, Allemann E, Gurny R. Comparison of twomethods of encapsulation of an oligonucleotide into poly(D,L-lactic acid) particles. Int J Pharm 214:25–30, 2001.

130. Berton M, Allemann E, Stein CA, Gurny R. Highly loaded nano-particulate carrier using an hydrophobic antisense oligonucleotidecomplex. Eur J Pharm Sci 9:163–170, 1999.

131. Berton M, Benimetskaya L, Allemann E, Stein CA, Gurny R.Uptake of oligonucleotide-loaded nanoparticles in prostatic can-cer cells and their intracellular localization. Eur J Pharm Bio-pharm 47:119–123, 1999.

132. Emile C, Bazile D, Herman F, Helene C, Vieillard M. Encapsu-lation of nucleotides in stealth Me.PEG-PLA50 nanoparticles bycomplexation with structures oligopeptides. Drug Deliv 3:187–195, 1996.

133. Berton M, Benimetskaya L, Allemann E, Stein CA, Gurny R.Uptake of oligonucleotide-loaded nanoparticles in prostatic can-cer cells and their intracellular localization. Eur J Pharm Bio-pharm 47:119-123.

134. Huwyler J, Wu D, Pardridge WM. Brain drug delivery of smallmolecules using immunoliposomes. Proc Natl Acad Sci USA93:14164–14169, 1996.

135. Zhang Y, Schlachetzki F, Pardridge WM. Global non-viral genetransfer to the primate brain following intravenous administra-tion. Mol Ther 7:11–18, 2003.

136. Shi N, Zhang Y, Zhu C, Boado RJ, Pardridge WM. Brain-specificexpression of an exogenous gene after i.v. administration. ProcNatl Acad Sci USA 98:12754–12759, 2001.

137. Shi N, Pardridge WM. Noninvasive gene targeting to the brain.Proc Natl Acad Sci USA 97:7567–7572, 2000.

138. Zhang Y, Boado RJ, Pardridge WM. Absence of toxicity ofchronic weekly intravenous gene therapy with pegylated immu-noliposomes. Pharm Res 20:1779–1785, 2003.

139. Salem AK, Cannizzaro SM, Davies MC, Tendler SJ, Roberts CJ,Williams PM, et al. Synthesis and characterisation of a degrad-able poly(lactic acid)-poly(ethylene glycol) copolymer with bio-tinylated end groups. Biomacromolecules 2:575–580, 2001.

140. Tessmar J, Mikos A, Gopferich A. Amine-reactive biodegradablediblock copolymers. Biomacromolecules 3:194–200, 2002.

141. Tessmar J, Mikos A, Gopferich A. The use of poly(ethyleneglycol)-block-poly(lactic acid) derived copolymers for the rapidcreation of biomimetic surfaces. Biomaterials 24:4475–4486,2003.

142. Olivier JC, Huertas R, Lee HJ, Calon F, Pardridge WM. Synthesisof pegylated immunonanoparticles. Pharm Res 19:1137–1143,2002.

143. Roberts MJ, Bentley MD, Harris JM. Chemistry for peptide andprotein PEGylation. Adv Drug Deliv Rev 54:459–476, 2002.

144. Yoo HS, Park TG. Folate receptor targeted biodegradable poly-meric doxorubicin micelles. J Control Release 96:273–283, 2004.

145. Gref R, Couvreur P, Barratt G, Mysiakine E. Surface-engineerednanoparticles for multiple ligand coupling. Biomaterials24:4529–4537, 2003.

146. Kang YS, Saito Y, Pardridge WM. Pharmacokinetics of [3H]bi-otin bound to different avidin analogues. J Drug Target 3:159–165, 1995.

147. Yamamoto Y, Nagasaki Y, Kato M, Kataoka K. Surface chargemodulation of poly(ethylene glycol)poly(L-lactide) block copol-ymer micelles: conjugation of charged peptides. Colloids Sur-faces B: Biointerfaces 16:135–146, 1999.

148. Emoto K, Nagasaki Y, Iijima M, Kato M, Kataoka K. Preparationof non-fouling surface through the coating with core-polymerizedblock copolymer micelles having aldehyde-ended PEG shell. Col-loids Surfaces B: Biointerfaces 18:337–346, 2000.

149. Yamamoto Y, Nagasaki Y, Kato Y, Sugiyama Y, Kataoka K.Long-circulating poly(ethylene glycol)-poly(-lactide) block co-polymer micelles with modulated surface charge. J Control Re-lease 77:27–38, 2001.

150. Pardridge WM, Boado RJ, Kang YS. Vector-mediated delivery ofa polyamide (“peptide”) nucleic acid analogue through the blood-brain barrier in vivo. Proc Natl Acad Sci USA 92:5592–5596,1995.



a new emerging picoscale biotechnology

Documents