aptamer-conjugated nanoparticles for the collection and detection of multiple cancer cells

Aptamer-Conjugated Nanoparticles for theCollection and Detection of Multiple Cancer Cells

Joshua E. Smith, Colin D. Medley, Zhiwen Tang, Dihua Shangguan, Charles Lofton, and Weihong Tan*

Center for Research at the Bio/Nano Interface, Department of Chemistry and Shands Cancer Center, UF Genetics Instituteand McKnight Brain Institute, University of Florida, Gainesville, Florida 32611

We have extended the use the aptamer-conjugated nano-particles for the collection and detection of multiple cancercells. The aptamers were selected using a cell-basedSELEX strategy in our laboratory for cancer cells that,when utilized in this method, allow for the selectiverecognition of the cells from complex mixtures includingfetal bovine serum samples. Aptamer-conjugated magneticnanoparticles were used for the selective targeting cellextraction, and aptamer-conjugated fluorescent nanopar-ticles were employed for sensitive cellular detection.Employing both types of nanoparticles allows for selectiveand sensitive detection not possible by using the particlesseparately. Fluorescent nanoparticles amplify the signalintensity versus a single fluorophore label resulting inimproved sensitivity. In addition, aptamer-conjugatedmagnetic nanoparticles allow for extraction and enrich-ment of target cells not possible with other separationmethods. Fluorescent imaging and a microplate readerwere used for cellular detection to demonstrate the wideapplicability of this methodology for medical diagnosticsand cell enrichment and separation.

INTRODUCTIONThe essential need for accurate, sensitive diagnosis and

understanding of human diseases at the molecular level has beenlimited by the lack of molecular probes available to recognize thedistinct molecular features of diseases. Diseases like canceroriginate due to mutations and alterations at the genetic level.These changes cause the affected cells to behave differently atthe molecular level, which can facilitate the effective treatmentprograms implemented by clinicians. Determining the molecularcharacteristics of cancers, particularly knowing the characteristicproteins associated with a specific cancer, can be of great benefit.These differences contain significant potential for aiding theunderstanding of diseases on the basis of the biological processesand mechanisms and could be vital for disease diagnosis, preven-tion, and treatment. In addition, techniques for sensitive minimalresidual disease detection are essential for monitoring diseasedevelopment and determining those who might be more suscep-tible to relapse. Current diagnosis methods for leukemia combinethe analysis of bone marrow and peripheral blood cytochemical

by karyotyping,1 immunophenotyping by flow cytometry2 ormicroarray,3 and amplification of malignant cell mutations by PCR.4

Immunophenotypic analyses of leukemia cells use antibodyrecognition elements to exploit the variation of surface antigensto differentiate disease cells from healthy cells. The limitation ofthis method is that typically the target antigens required forcellular recognition are not expressed exclusively on any one celltype, which dramatically influences sensitivity and results in falsepositive signals. In addition, antibody integrity is continually inquestion due to the lifetime of the sources needed to obtain areliable antibody. Due to this, immunophenotypic analyses requiremultiple antibody recognition elements for accurate cell detectionresulting in the increased complexity and cost of the method,including the time-consuming nature of the techniques.

PCR-based methods have proven to be highly sensitivediagnostic techniques for cellular recognition,4-6 but they indi-rectly detect cells by monitoring RNA expression. PCR techniquesrequire prolonged RNA isolation steps before analysis can beperformed, and the variable sensitivity of PCR can limit itseffectiveness as a diagnostic technique. These limitation can leadto false-negative results particularly with occult tumor cells wherelow-level signals are expected.4 Therefore, the need to developnew technologies for rapid, economical cell recognition still exists.

Previously we have described aptamer-conjugated nanopar-ticles (ACNPs) for the rapid detection of acute leukemia cells,which used high-affinity DNA aptamers for signal recognition. An88-base oligonucleotide sequence with specific binding properties(Kd ) 5 nM) for CCRF-CEM acute leukemia cells was attachedto both magnetic nanoparticles (MNPs) and fluorescent nanopar-ticles (FNPs) to develop a specific platform for collecting andimaging intact leukemia cells from mixed cell and whole bloodsamples.7

Here we have further developed the ACNP protocols toperform the extraction of multiple cancer cell targets usingadditional high-affinity DNA aptamers for recognition. The 88-

* To whom correspondence should be addressed. Phone: 352-846-2410.Fax: 352-846-2410. E-mail: [email protected].

(1) Faderl, S.; Kantarjian, H. M.; Talpaz, M.; Estrov, Z. Blood 1998, 91 (11),3995-4019.

(2) Paredes-Aguilera, R.; et al. Am. J. Hematol. 2001, 68, 69-74.(3) Belov, L.; de la Vega, O.; dos Remedios, C. G.; Mulligan, S. P.; Christo-

pherson, R. I. Cancer Res. 2001, 61, 4483-4489.(4) Ghossein, R. A.; Bhattacharya, S. Eur. J. Cancer 2000, 36, 1681-1694.(5) Iinuma, H.; Okimaga, K.; Adachi, M.; Suda, K.; Sekine, T.; Sakagawa, K.;

Baba, Y.; Tamura, J.; Kumagai, H.; Ida, A. Int. J. Cancer 2000, 89, 337-344.

(6) Liu Yin, J. A.; Grimwade, D. Lancet 2002, 360, 160-162.(7) Herr, J. K.; Smith, J. E.; Medley, C. D.; Shangguan, D.; Tan, W. Anal. Chem.

2006, 78, 2918-2924.

Anal. Chem. 2007, 79, 3075-3082

10.1021/ac062151b CCC: $37.00 © 2007 American Chemical Society Analytical Chemistry, Vol. 79, No. 8, April 15, 2007 3075Published on Web 03/10/2007

base oligonucleotide sequence for CCRF-CEM acute leukemiacells was used in addition to oligonucleotide sequences withspecific binding properties for cell lines of Burkitt’s lymphoma(Ramos) and non-Hodgkin’s B cell lymphoma (Toledo). Again,the sequences were attached to MNPs and FNPs to complete theplatform for collecting and imaging the respective intact cancercells from buffer and fetal bovine serum (FBS) samples.

Aptamers are selected by a process referred to as SELEX, orsystematic evolution of ligands by exponential enrichment,8,9 froma pool of DNA, and even RNA, candidates that bind highlyspecifically with molecular or cellular targets. More recentlyaptamers have been recognized as reliable affinity ligands rivalingantibodies for diagnostic potential.10 The major difference beingthat antibodies are generally extracted and purified from animals,where the problem of variation from animal-to-animal of the sameantibody can affect binding. As compared to aptamers, once asequence is determined, it can be easily synthesized over-and-over without concern for animal destruction.11 Aptamers are ableto fold into complex three-dimensional structures with distinctmolecular binding motifs and have been used for protein detectionby sensor array12 and affinity capillary electrophoresis13 and fortargeted therapeutic applications.11,14,15

An in vitro process identifying DNA sequences with strongaffinities toward intact tumor cells referred to as tumor cellSELEX16 was used to select aptamers with high specificity towardour target cancer cell lines. Aptamers selected by cell SELEX havethe ability to distinguish one cell type from numerous other celltypes. This discrimination ability was revealed during the selectionprocess.17

We have demonstrated that combining the selectivity ofaptamers with the power of MNP-based separation has produceda selective and sensitive method for collecting, enriching, andsubsequently detecting various targets. In this work, the aptamerswere attached to three spectrally different FNPs to provideenhanced signaling. Fluorophore-doped silica NPs have been usedto replace fluorescent dyes for their increased signal and theirease of being functionalized with biomolecules.18-20 Exploiting theavailability of functional groups on the particle surface has provenuseful for oligonucleotide detection,19,21 protein, and antigen

detection.21-26 Each silica NP contains many dye molecules andbinds to the cell surface via the aptamer conjugated to the NP.The ACNP produce a fluorescent signal that is significantlybrighter than that of any individual dye probe upon excitation.

Aptamer-conjugated MNP based cell sorting was employed forselective malignant cell collection. Magnetic activated cell sorting(MACS) has been used extensively for extraction and enrichmentof epithelial cells,27 endothelial cells,28 bacteria,29 and circulatingtumor cells.5,30-32 However, these methods typically use micrometer-sized magnetic polymer beads and we are utilizing 65 nm silica-coated MNPs. MNPs have been used for gene collection,33 peptideisolation for MS analysis,34 and leukemia cell extraction.7 The smallsize and relatively high surface area of NPs provide enhancedextraction capabilities versus those of the micrometer-sizedparticles.33 In addition to enabling selective target extraction, MNP-based sorting removes the need for presample cleanup, sinceunwanted NP aggregates and unbound materials are easilyseparated from the target cells.

Fluorescence imaging and a microplate reader spectrometerwere used to confirm the selectivity and to expand the ACNPtechnique to more commonly available clinical detection equip-ment. Using three different FNPs and aptamers allowed for thedemonstration of this method for both inorganic and organic dye-doped silica NPs and the multiple extraction of three differentrelated cell lines from the same sample. This method alsodemonstrates the capacity to reproducibly extract target cells fromcomplex mixtures and fetal bovine serum, further establishing afoundation for the relevance of this method for clinical applications.

MATERIALS AND METHODSMaterials. All materials were purchased from Sigma-Aldrich

(St. Louis, MO) unless other noted. Fetal bovine serum (FBS)was obtained from Invitrogen (Carlsbad, CA). Carboxylethyl-silanetriol sodium salt was purchased from Gelest, Inc. (Morris-ville, PA). 1-Ethyl-3-(3-dimethylaminopropyl)carbodiimide hydro-chloride (EDC) was purchased from Pierce Biotechnology, Inc.(Rockford, IL), and ammonium hydroxide was obtained fromFisher, Inc. Cy5-NHS was purchased from Amersham Biosciences,and tetramethylrhodamine succinimidyl ester (TMR SE) (mixed

(8) Ellington, A. D.; Szostak, J. W. Nature 1990, 346, 816-820.(9) Tuerk, C.; Gold, L. Science 1990, 249, 505-510.

(10) Brody, E. N.; Gold, L. Rev. Mol. Biotechnol. 2000, 74, 5-13.(11) Tombelli, S.; Minunni, M.; Mascini, M. Biosensens. Bioelectron. 2005, 20,

2424-2434.(12) Kirby, R.; Cho, E. J.; Gehrke, B.; Bayer, T.; Park, Y. S.; Neikirk, D. P.;

McDevitt, J. T.; Ellington, A. D. Anal. Chem. 2004, 76, 4066-4075.(13) German, I.; Buchanan, D. D.; Kennedy R. T. Anal. Chem. 1998, 70, 4540-

4545.(14) Osborne, S. E.; Matsumura, I.; Ellington, A. D. Curr. Opin. Chem. Biol. 1997,

1, 5-9.(15) Farokhzad, O. C.; Jon, S.; Khademhosseini, A.; Tran, T. T.; LaVan, D. A.;

Langer, R. Cancer Res. 2004, 64, 7668-7672.(16) Daniels, D. A.; Chen, H.; Hicke, B. J.; Swiderek, K. M.; Gold, L. Proc. Natl.

Acad. Sci. U.S.A. 2003, 100, 15416-15421.(17) Shangguan, D.; Li, Y.; Tang, Z.; Cao, Z. C.; Chen, H. W.; Mallikaratchy, P.;

Sefah, K.; Yang, C. J.; Tan, W. Proc. Natl. Acad. Sci. U.S.A. 2006, 103,11838-11843.

(18) Bagwe, R.; Zhao, X.; Tan, W. J. Dispersion Sci. Technol. 2003, 24, 453-464.

(19) Zhao, X.; Tapec-Dytioco, R.; Tan, W. J. Am. Chem. Soc. 2003, 125, 11474-11475.

(20) Zhao, X.; Bagwe, R.; Tan, W. Adv. Mater. 2004, 16, 173.(21) Lian, W.; Litherland, S.; Badrane, H.; Tan, W.; Wu, D.; Baker, H.; Gulig, P.;

Lim, D.; Jin, S. Anal. Biochem. 2004, 334, 135-144.

(22) Yang, W.; Zhang, C. G.; Qu, H. Y.; Yang, H. H.; Xu, J. G. Anal. Chim. Acta2004, 503, 163-169.

(23) Santra, S.; Wang, K.; Tapec-Dytioco, R.; and Tan, W. J. Biomed. Opt. 2001,6.

(24) Santra, S.; Zhang, P.; Wang, K.; Tapec, R.; Tan, W. Anal. Chem. 2001, 73,4988-4993.

(25) Yang, H.; Qu, H.; Lin, P.; Li, S.; Ding, M.; Xu, J. Analyst 2003, 128, 462-466.

(26) Ye, Z.; Tan, M.; Wang, G.; Yuan, J. Anal. Chem. 2004, 76, 513-518.(27) Griwatz, C.; Brandt, B.; Assmann, G.; Zanker, K. S. J. Immunol. Methods

1995, 183, 251-265.(28) Marelli-Berg, F. M.; Peek, E.; Lidington, E. A.; Stauss, H. J.; Lechler, R. I.

J. Immunol. Methods 2000, 244, 205-215.(29) Porter, J.; Robinson, J.; Pickup, R.; Edwards, C. J. Appl. Microbiol. 1998,

84, 722-732.(30) Stanciu, L. A.; Shute, J.; Holgate, S. T.; Djukanovic, R. J. Immunol. Methods

1998, 189, 107-115.(31) Hu, X. C.; Wang, Y.; Shi, D. R.; Loo, T. Y.; Chow, L. W. C. Oncology 2003,

64, 160-165.(32) Benez, A.; Geiselhart, A.; Handgretinger, R.; Schiebel, U.; Fierlbeck, G. J.

Clin. Lab. Anal. 1999, 13, 229-233.(33) Zhao, X.; Tapec-Dytioco, R.; Wang, K.; Tan, W. Anal. Chem. 2003, 75, 3476-

3483.(34) Turney, K.; Drake, T. J.; Smith, J. E.; Tan, W.; Harrison, W. W. Rapid

Commun. Mass Spectrom. 2004, 18, 1-8.

3076 Analytical Chemistry, Vol. 79, No. 8, April 15, 2007

isomers) was purchased from Molecular Probes. Deoxyribonucle-otides, 5′-amino-modifiers, and biotin phosphoramidite werepurchased from Glen Research (Sterling, Va).

Fluorescent Nanoparticle Synthesis. Rubpy dye-dopednanoparticles (NPs) were synthesized by the reverse microemul-sion method.7 Briefly, the NPs were synthesized by adding 1.77mL of Triton X-100, 7.5 mL of cyclohexane, and 1.6 mL ofn-hexanol to a 20 mL glass vial with continuous magnetic stirring.Next, 400 µL of H2O and 80 µL of 0.1 M tris(2,2′-bipyridyl)-dichlororuthenium(II) hexahydrate (Rubpy) dye (MW ) 748.63)were added. Followed by the addition of 100 µL of tetraethylorthosilicate (TEOS), the materials were stirred for 30 min. Toinitiate silica polymerization, 60 µL of NH4OH was added. After18 h, a postcoating of carboxyl-modified silica was performed byadding 50 µL of TEOS and 50 µL of carboxylethylsilanetriolsodium salt. This polymerization was allowed to proceed for 18h. The particles were centrifuged, sonicated, and vortexed fourtimes with 10 mL aliquots of fresh 95% ethanol, followed by a washwith a 10 mL aliquot of H2O. Each wash step was performed fromthe addition of fresh ethanol or H2O with sonication and vortexingto the next centrifugation step typically within 3-5 min. The DNAmodification was carried out by adding 1.2 mg of EDC, 0.5 nmolof DNA, and 2 mg of particles to 1.5 mL of 10 mM MES buffer(pH ) 5.5). The solution was vortexed for 3.5 h. Particles werewashed by centrifuging at 14 000 rpm and dispersing in 200 µLof 0.1 M phosphate-buffered saline (PBS) (pH ) 7.2) three times.Rubpy NPs were stored at 4 °C and dispersed in cell media bufferat a final concentration of ∼10 mg/mL.

Tetramethylrhodamine (TMR SE) and Cy5 doped NPs weresynthesized according to the following method: TMR SE and Cy5-NHS were each dissolved in DMSO at a concentration of 5 mg/mL, and (3-aminopropyl)triethoxysilane (APTS) was added at amolar ratio of 1.2:1 APTS:dye. The APTS was allowed to conjugateto the amine reactive dye for 24 h in the dark with shaking priorto synthesis of the particles. Glass reaction vessels and Teflon-coated magnetic stir rods were washed with 1 M NaOH solutionfor 30 min, rinsed with DI water and ethanol, and allowed to dry.This wash step was performed to clean the glass vessel and stirrods and smooth the inside surface of the glass vessel, whichprevents unwanted seeding and NP formation. After conjugation,4.19 mL of ethanol was mixed with 239 µL of ammoniumhydroxide solution in the reaction vessel. A 36 µL volume ofTMR-APTS conjugate or 54 µL of Cy5-APTS conjugate wasadded to the reaction vessels, yielding 3.44 × 10-7 mol of dye/reaction (ratio of 2300 mol of silica/mole of dye). A 177 µL volumeof TEOS was added rapidly to the reaction mixture, and the vesselswere sealed. The reaction was allowed to proceed for 48 h in thedark before the particles were recovered by centrifugation at14 000 rpm. The particles were washed three times with phosphatebuffer to remove any dye molecules that are weakly bound. Thesynthesis method was found to reproducibly produce a numberaverage particle size of 50 nm ( 5 nm with a monomodaldistribution when measured with a Honeywell UPA 150 dynamiclight scattering instrument.

To modify with carboxy groups these NPs, a 2 mg solution ofNPs was diluted in 200 µL of 10 mM PBS, pH 7.4, and combinedwith 40 µL of the carboxyl silane with the solution beingcontinuously mixed for 4 h. The NPs were washed three times,

resuspended in 200 µL aliquots of 10 mM PBS by centrifuging at14 000 RPM for 15 min, and stored at room temperature until use.DNA modification was completed as described above.

Magnetic Nanoparticle Synthesis. Iron oxide core MNPs34

were synthesized by coprecipitating iron salts. A mechanical stirrerwas used to mix ammonia hydroxide (2.5%) with an iron chloridesolution at 350 rpm for 10 min. The iron chloride solutioncontained ferric chloride hexahydrate (0.5 M), ferrous chloridetetrahydrate (0.25 M), and HCl (0.33 M). The iron oxide NPs werewashed three times with 5 mL aliquots of H2O and once with a 5mL aliquot of ethanol. Each wash was performed from decantingthe supernatant, adding fresh wash solution, and redispersing inthe fresh solution typically within 3-5 min. Next the iron oxideNPs were dispersed in an ethanol solution containing ∼1.2%ammonium hydroxide at a final concentration of ∼7.5 mg/mL.

The magnetite core particles were coated with silica by addingtetraethoxyorthosilicate (200 µL), and the mixture was sonicatedfor 90 min to complete the hydrolysis process. An additionalaliquot of TEOS (10 µL) was added and again sonicated for 90min to add a postcoating to the NPs. The sample was washedthree times with ethanol to remove excess reactants.

A solution of 0.1 mg/mL Fe3O4-SiO2 (silica-coated MNPs)solution in 10 mM PBS, pH 7.4, and a 5 mg/mL avidin solutionin 10 mM PBS, pH 7.4, was vortexed for 5-10 min to initiate anavidin coating. The resulting sample was incubated at 4 °C for12-14 h. Next the particles were washed three times anddispersed at 1.2 mg/mL with 100 mM PBS. The avidin coatingwas stabilized by cross-linking the coated NPs with 1% glutar-aldehyde (1 h at 25 °C). Again the particles were magneticallyseparated, washed three times, and dispersed in 1 M Tris-HClbuffer. The samples was incubated in the 1 M Tris-HCl buffer (3h at 4 °C), followed by three additional washes with 20 mM Tris-HCl, 5 mM MgCl2, pH 8.0, at a concentration of ∼0.2 mg/mL.

Finally, the DNA was attached to the particles by addingbiotinylated DNA (3 pmol) to a solution of 500 µL at 0.2 mg/mLin 20 mM Tris-HCl, 5 mM MgCl2, pH 8.0, avidin-coated MNPs.The attachment was performed at 4 °C for 12 h, and three finalwashes were performed using 20 mM Tris-HCl and 5 mM MgCl2

at pH 8.0. The MNPs were stored at 4 °C and used at a finalconcentration of ∼0.5 mg/mL.

Magnetic Extraction. Magnetic extraction was performed byadding the specified amounts of MNPs to each sample asdescribed in the experimental sections for the respective extrac-tion procedures. The aptamer-conjugated MNPs were incubatedwith the cell samples for 15 min, a magnetic field was introducedto the sample container, and after 2-5 min the nonmagneticmaterials were decanted using a Pasteur pipet. To complete thewash process, the magnetic field was removed and the sampleswere redispersed in 200 µL of fresh media buffer, and this processwas repeated three times.

Cells. CCRF-CEM cells (CCL-119 T-cell, human acute lym-phoblastic leukemia), Ramos cells (CRL-1596, B-cell, humanBurkitt’s lymphoma), and Toledo cells (CRL-2631, non-Hodgkin’sB cell lymphoma) were obtained from ATCC (American TypeCulture Association). The cells were cultured in RPMI mediumsupplemented with 10% fetal bovine serum (FBS) and 100 IU/mL penicillin-Streptomycin. The cell density was determinedusing a hemocytometer, and this was performed prior to any

Analytical Chemistry, Vol. 79, No. 8, April 15, 2007 3077

experiments. After this approximately one million cells dispersedin RPMI cell media buffer were centrifuged at 920 rpm for 5 minand redispersed in cell media three times and were then redis-persed in 1 mL cell media buffer. During all experiments, the cellswere kept in an ice bath at 4 °C.

DNA Aptamer Synthesis. The following aptamers have beenselected for the CCRF-CEM, Ramos, and Toledo cells respec-tively: 5′-TTT AAA ATA CCA GCT TAT TCA ATT AGT CAC ACTTAG AGT TCT AGC TGC TGC GCC GCC GGG AAA ATA CTGTAC GGA TAG ATA GTA AGT GCA ATC T-3′; 5′-AAC ACC GGGAGG ATA GTT CGG TGG CTG TTC AGG GTC TCC TCC CGGTG-3′; 5′-ATA CCA GCT TAT TCA ATT ATC GTG GGT CAC AGCAGC GGT TGT GAG GAA GAA AGG CGG ATA ACA GAT AATAAG ATA GTA AGT GCA ATC T-3′. Both the amine andbiotinylated versions of the aptamer sequencers were synthesizedin-house. An ABI3400 DNA/RNA synthesizer (Applied Biosys-tems, Foster City, CA) was used for the synthesis of all DNAsequences. A ProStar HPLC (Varian, Walnut Creek, CA) with aC18 column (Econosil, 5u, 250 × 4.6 mm) from Alltech (Deerfield,IL) was used to purify all fabricated DNA. A Cary Bio-300 UVspectrometer (Varian, Walnut Creek, CA) was used to measureabsorbances to quantify the manufactured sequences. All oligo-nucleotides were synthesized by solid-state phosphoramiditechemistry at a 1 µmol scale. The completed sequences were thendeprotected in concentrated ammonia hydroxide at 65 °C over-night and further purified twice with reverse phase high-pressureliquid chromatography (HPLC) on a C-18 column.

Cell Imaging. Fluorescence imaging was conducted with aconfocal microscope setup consisting of an Olympus IX-81 invertedmicroscope with an Olympus Fluoview 500 confocal scanningsystem and three lasers, a tunable argon ion laser (458, 488, 514nm), a green HeNe laser (543 nm), and a red HeNe laser (633nm) with three separate photomultiplier tubes (PMT) for detec-tion. The cellular images were taken with a 10× objective. TheRubpy NPs were excited with 488 nm line of the argon ion laser,and emission was detected using a 610 nm long pass filter. TheTMR NPs were excited with the 543 nm laser line and weredetected with a 560-600 nm band-pass filter. The Cy5 NPs wereexcited with the 633 nm laser line, and the emission was detectedwith a 660 nm long pass filter.

Microplate Reader. All plate reader experiments were con-ducted with a Tecan Safire microplate reader with 384 wellCorning small volume plates. The excitation and emission wave-length used were the same as those in the fluorescent imaging.For each experiment 20 µL of the extracted cell solution wasplaced in the well and the fluorescence of the sample wasmeasured immediately.

Magnetic Extraction and Labeling. To establish the extrac-tion and detection capabilities of the method, equal amounts ofcells in media were tested using the two NP approach. When theCEM cells were used as the target, the Ramos and Toledo celltypes were used as the control (nontarget) cells, when the Ramoscells were used as the target, the CEM and Toledo cell types wereused as the control cells, and finally when the Toledo cells wereused as the target, the CEM and Ramos cell types were used forthe control experiments.

The extraction of the multiple individual cell types (CEM,Ramos, and Toledo) using their respective aptamer-conjugated

NPs was performed by the following procedure. Approximately105 of each cell type were obtained in individual test tubes. Tothe cell samples, 5 µL of the MNP solution was added, and themixture was incubated for 15 min. After incubation, the cells werewashed by magnetic extraction with 200 µL of fresh cell mediathree times and resuspended in 200 µL of the media buffer. Thewash was performed by removal of the supernatant and additionof fresh buffer, and the sample was resuspended in the fresh buffertypically within 3-5 min. To complete the stepwise process, 2 µLof FNPs was added and incubated for 5 min. The concentrationof MNPs to FNPs in the samples was 2:1. Again, the sample waswashed three times with 200 µL of cell media as describedpreviously and then dispersed in 20 µL of media for imaging andmicroplate reader analysis. The FNPs used to label the Ramoscells were doped with cy5, and the fluorescent dyes doped in theNPs for Toledo and CEM cells were Rubpy and TMR, respectively.All pure cell samples contained 1.0 × 105-5.0 × 105 cells beforeNP incubation. The multiple cell type extraction procedure willbe described in a later section.

For determining the detection limit, the extraction wasperformed by first determining the number of cells in 1 µL ofstock cell solution. The total number of cells was counted. Thisprocess was repeated five times; the determined values wereaveraged and extrapolated to obtain the number of cells/microliter. The cell samples were diluted to 200 µL with cell mediaaccordingly. Next, cell samples were subjected to MNP incuba-tions for 15 min followed by magnetic extraction and washing asdescribed previously and FNP incubations for 5 min with magneticextraction and washing as described above. All cell samples weretreated with MNPs and FNPs at a ratio of 2:1 using the stepwiseformat, and extracted samples were analyzed by the microplatereader.

RESULTS AND DISCUSSIONMultiple Cell Extraction. To expand the concept of the two

particle-based magnetic collection and detection technique toinclude three different leukemia cell lines by using three differentaptamer molecules for three different cell lines, CEM, Ramos, andToledo cell samples were extracted using ACNPs followed byfluorescent imaging and analysis by the microplate reader. Eachpure cell sample extraction was repeated 10 times. As wasmentioned in Materials and Methods, the control cells used forthe CEM experiments were the Ramos and Toledo cell types,those for Ramos were CEM and Toledo, and those for the Toledowere CEM and Ramos. Figure 1 shows representative confocalimages of 2 µL aliquots of the CEM target cells (left) and Ramosnontarget cells (right) using CEM ACNPs (red) (A), Toledo targetcells (left) and CEM nontarget cells (right) using Toledo ACNPs(green) (B), and Ramos target cells (left) and CEM nontargetcells (right) using Ramos ACNPs (blue) (C) after NP incubationsand magnetic washes. The figure indicates Ramos, CEM, andCEM as the respective controls for those experiments. The othercontrol cell type experiments were performed and resulted in thesame responses as the one presented (data not shown). Inaddition, some fluorescence spots were observed in the images.However when the samples were analyzed with the microplatereader and compared to sample blanks treated with the MNPsand FNPs, the levels of fluorescence signal were the same (imagesnot shown). Table 1 provides the fluorescence data obtained from


the microplate reader. The first column represents the cell samplethat was analyzed, the second column represents the signalproduced by the CEM NPs, the third column represents the signalproduced by the Toledo NPs, and the fourth column representsthe signal produced by the Ramos NPs. The rows in the tabledisplay the cell samples that were investigated using the ACNPs.

On the basis of the fluorescence images, a significant differenceis evident in both the amount of cells extracted and fluorescentsignal present between the target and control cells in all samples.However, some control cells were inadvertently collected and evenlabeled with some FNPs but no significant signal was indicatedby the microplate reader data for those samples producing signalsin the same realm as sample blanks (images not shown). On thecontrary, the target cells subjected to this procedure had very

intense fluorescent signals that made them easily discernible fromthe control cells. With a closer look at the characterization ofexpanding the ACNP technique to multiple cell types, themicroplate reader data (Table 1) demonstrated that when using100 000 cells in each of the pure cell samples at a collectionefficiency of 85% (determined in a previous publication7), all targetcell samples produced signals upward of 24-fold enhancementsabove the background and as high as 50-fold. The target samplesindicated in Table 1 for this experiment were the CEM cells (row2) for the CEM NPs (column 1), the Toledo cells (row 3) for theToledo NPs (column 2), and the Ramos cells (column 3) for theRamos NPs (column 3). The control samples for these experi-ments were represented in the remainder of the table for each ofthe NPs and cell types. The signals for the control samples at theconditions mentioned above and the collection efficiency for theMNP amounts used in these experiments for the control cellswere determined to be no greater than 5% (determined in aprevious publication7), resulting in fluorescence signals at thesame level as a buffer blank sample treated with the ACNPs. Thesedata indicate that the MNPs were both selective for the targetcells by discriminating against the control cells and reproduciblein all sample types investigated.

Detection Limit. The limit of detection (LOD) was determinedusing pure cell samples, and the extractions were performed asdescribed previously. The limit of detection threshold was takento be three standard deviations above the blank, and because ofthis, any residual fluorescence in the blank was accounted for.The LOD was performed using CEM target cells. Each of thesamples was then analyzed with NPs using the previouslymentioned protocols with the fluorescence intensity being deter-mined on the microplate reader following completion of the ACNPtechnique. The detection limit was computed by plotting thefluorescence intensity versus the cell number present in thesample. Consequently, the plotted data produced a linear responseas seen in Figure 2A, and Figure 2B displays the data zoomed inon the lower sample concentrations. From this plot the detectionlimit was determined to be approximately 250 cells with a dynamicrange covering more than 2 orders of magnitude. This indicatesthat the ACNP system has the ability to sensitively detect lowamounts of intact targets cells from a given sample and that a

Figure 1. Fluorescence images of pure cell samples in buffer aftermagnetic extraction and washes: (A) image of CEM target cells (left)and Ramos nontarget cells (right) using CEM ACNPs; (B) image ofToledo target cells (left) and CEM nontarget cells (right) using ToledoACNPs; (C) image of Ramos target cells (left) and CEM nontargetcells (right) using Ramos ACNPs.

Table 1. Single Cell Type Extraction

samplecells

CEMNP signal

ToledoNP signal

RamosNP signal

Ramos 945 965 7 574CEM 48 967 1 056 314Toledo 1 075 36 728 438

Figure 2. Limit of detection for the stepwise addition of the MNPand FNP using the microplate reader for detection: (A) full calibrationcurve; (B) enlarged depiction of the lower concentration regime.


wide range of cell concentrations can be analyzed by this methodwith little to no sample preparation depending on the amount ofACNPs used.

Multiple Cell Type Extraction Method. Determining themultiple extraction and detection capability of the ACNP wasperformed by creating artificial complex samples of CEM, Ramos,and Toledo cells. Figure 3 displays the schematic diagram of themultiple cell extraction procedure that was employed. The sampleswith one, two, and three cell types were analyzed using theACNPs. The samples were prepared by obtaining approximately105 cells of each type for the respective sample type. The stepwiseextraction protocol was performed by adding the specifiedamounts of MNPs for Ramos cells, followed by CEM aptamer-conjugated MNPs, and finally with Toledo specific MNPs. Eachset of MNPs was incubated with the cell samples separately for15 min. After the Ramos MNPs were incubated with the cellsamples, magnetic extraction was performed and the supernatantkept to be treated with the CEM-specific MNPs. The remainder

of the magnetic extractions was carried out as described in theMagnetic Extraction section. The sample was redispersed in 200µL of cell media, followed by addition of the Ramos aptamer-conjugated FNPs with 5 min of incubation, and the magneticextraction procedure was performed. Similarly, the respectiveCEM and Toledo aptamer-conjugated FNPs were subsequentlyintroduced to their samples. After the final wash, the cell samplewas dispersed in 20 µL of media buffer. The samples wereanalyzed by confocal imaging with 2 µL of aliquots and a platereader spectrometer with 20 µL of aliquots.

Mixed Cell Samples. The power of the multiple extractionprocedure needed to be evaluated using complex sample mixtures.Figure 4 A-C demonstrates the results from artificial complexsamples by mixing equal amounts of the appropriate cell typesfor the three different multiple extraction samples diluted in cellmedia buffer, where CEM, Ramos, and Toledo cells were mixedand the ACNP process applied as described above. A total of100 000 cells in all samples was used; cell and buffer volumes were

Figure 3. Schematic representation of the multiple extraction procedure with the MNPs being added and extracted stepwise and thecorresponding FNPs being added post magnetic extraction of cell samples.

Figure 4. Fluorescence images of buffer extracted mixed cell samples using the multiple extraction procedure: (A) contains only Ramoscells; (B) contains CEM and Toledo cells; (C) contains all three CEM, Toledo, and Ramos cells. Flourescence images D-F displays the mixedcell samples extracted from FBS using the multiple extraction procedure, where (D) contains only Ramos cells, (E) contains CEM and Ramoscells, and (F) contains all three CEM, Toledo, and Ramos cells.


adjusted accordingly. To exhibit that the MNPs indeed have theability to selectively differentiate the cells from one another in amultiple cell mixture format, single, double, and triple cell mixedsamples were evaluated. Figure 4A illustrates the selective natureof the technique by performing the ACNP steps with a single cellsample, Ramos cells. The single cell sample was first treated withCEM ACNPs followed by Toledo ACNPs and finally RamosACNPs. The samples were incubated at 4 °C with the MNPs andFNPs as expressed in the previous section. On the basis of thefluorescence images, this method was able to selectively collectthe Ramos cells (blue) only when the Ramos ACNP was intro-duced to the cell sample Figure 4A. The Toledo and Ramos cellswere used in single cell sample extractions as well (image notshown). This method was further tested by performing the ACNPsteps with a mixture of two different cell types, CEM and Toledo.Figure 4B displays the selective nature of this technique for thecells indicated. The fluorescence images again demonstrateselective isolation of the CEM (red) and Toledo (green) cells.Other CEM, Toledo, and Ramos two cell type mixed samples wereperformed as well (image not shown). The final test was toperform this technique with a mixture of all three cell types inthe same sample using the CEM, Toledo, and Ramos cells at thesame time. Figure 4C reveals the selective nature of the methodfor each of the cells indicated. Fluorescence images again depictthe selective isolation of the CEM, Toledo, and Ramos cells, Figure4C left (red), middle (green), and right (blue) images, respectively.

Microplate reader data were in complete agreement with theconfocal image data and presented in Table 2. The first columnrepresents the cell samples that were analyzed, the second columnrepresents the signal produced by the CEM NPs, the third columnrepresents the signal produced by the Toledo NPs, and the fourthcolumn represents the signal produced by the Ramos NPs. Therows in the table display the cell samples that were investigatedusing the ACNPs. With 100 000 total cells present in all samples,samples containing target cells produced signals in upward of 24-fold enhancements above the background and as high as 47-fold.The signals for the control samples at the conditions mentionedabove resulted in fluorescence signals at the same level as a bufferblank sample treated with the ACNPs with the exception of theToledo nontarget sample in sample 4. Standard deviations deter-mined for all these samples were determined to be 8-12%. Thesedata indicate that the MNPs were both selective for the targetcells by discriminating against the control cells and reproduciblein all sample types investigated.

Serum Samples. To show applicability of the stepwise processin real biological samples, fetal bovine serum (FBS) was used.FBS was spiked with each of the respective cell types for thecorresponding one cell, two cell, and three cell extraction experi-

ments (500 µL). The process was performed as described above.Confocal imaging and a fluorescence microplate reader were usedto characterize cell extractions. Figure 4D-F illustrates the resultsof the FBS-spiked complex samples by mixing equal amounts ofthe indicated cells at a total cell concentration of approximately100 000 cells. Figure 4D-F presents the selective nature of thetechnique for the single, double, and triple mixed cell typesamples. The samples were treated with CEM ACNPs followedby Toledo ACNPs and finally Ramos ACNPs. For the single cellsample experiment, the sample was first treated with CEM ACNPsfollowed by Toledo ACNPs and finally Ramos ACNPs. Thesamples were incubated at 4 °C with the MNPs and FNPs asexpressed previously. The Figure 4D fluorescence image showsthe sample contains Ramos cells extracted and labeled only afterbeing treated with Ramos ACNP (blue). Extractions with the CEMand Toledo cell types were completed as well (images not shown).The Figure 4E left image (red) and right image (blue) show theCEM cells and Ramos cells extracted when treated with CEMand Ramos ACNP for the two cell type extraction experiment.Other two cell type extractions with the CEM, Toledo, and Ramoscell types were performed as well (images not shown). Figure 4Fleft (red), middle (green), and right (blue) images show theextraction of all three cells treated with all the ACNP.

The fluorescence imaging data were confirmed by collectingfluorescence data using the microplate reader, Table 3. The tablelayout was the same as in the previous table: first column wascell samples, second column was the CEM NP signals, thirdcolumn was Toledo NP signals, and fourth column was the RamosNP signal. The rows in the table display the cells mixed to makethe samples that were analyzed. The standard deviations weredetermined to be 8-12% for all samples measured in the FBS.With 100 000 total cells present in each sample dispersed in FBS,the signal enhancements determined above the backgroundranged from 10 to about 24. In all cases, the signals for all targetsamples were lower than those for the cell media buffer, and thebackground signals were all higher. The Toledo samples producedthe lowest enhancement of all the extracted samples, whichproduced the highest background signal of the three ACNPs pairs.The Toledo aptamer is less selective then the other aptamers thatwere used, which would explain the higher background producedin this particular sample. The fluorescence images and microplatereader data demonstrated that the MNPs were both selective forthe target cells by discriminating against the control cells andreproducible even in spiked-FBS samples.

Discussion. In this work, ACNPs were demonstrated toselectively extract and detect multiple intact target cells fromcomplex samples with limited sample preparation. Aptamer-conjugated FNPs were used for their high signal intensity andsignal stability, and aptamer-conjugated MNPs were utilized fortheir ability to selectively extract analytes. The target cells had a

Table 2. Multiple Cell Type Extraction

sample cellsCEM

NP signalToledo

NP signalRamos

NP signal

Ramos 1 281 1 040 7 862CEM 44 972 920 375Toledo 1 025 34 972 320CEM, Ramos 46 874 1 505 7 385CEM, Toledo 43 890 37 896 414CEM, Toledo, Ramos 42 145 32 945 7 524

Table 3. Multiple Cell Type Extraction in Serum

sample cellsCEM

NP signalToledo

NP signalRamos

NP signal

Ramos 1 845 3 241 6 776CEM, Ramos 43 835 3 554 6 980CEM, Toledo 40 767 31 240 452CEM, Toledo, Ramos 42 973 33 112 7 078


multitude of available aptamer binding sites that allowed theACNPs to sufficiently interact with the cell surface. Magneticextraction allows for easy sample cleanup by removing excessFNPs and other unbound fluorescent materials resulting in a lowerbackground. Employing magnetic extractions to such a systemprovides a separation and scavenging capability that is unlike anyother method available in that the MNPs can be introduced tothe sample of interest, locate the target of interest, and removethe bound substance from the remainder of the sample. Thisisolation step can also be used to enrich or concentrate the sampleafter removal of the unwanted sample components with providedease of washing due to the ability to manipulate the bound sampleswith a magnetic field. The MNPs have also demonstrated highcollection efficiencies for specific biological species.

When the dynamic range and detection limit studies wereperformed, this method resulted in a wide-ranged linear responsewith a good detection limit. We should mention that, above thenumber of cells indicated in the plot, the fluorescent response ofthis technique began to plateau. In addition, nonspecific bindingof nontarget cells affecting the LOD of this technique is a concern,but we feel that there will be little effect on the LOD due to theidentical signals seen in both the sample blanks and nontargetsamples. Also the determined value becomes a good starting pointfor a protocol that is designed to be rapid with minimal expertiserequired to perform the analysis using commercially availableinstrumentation. Some of the options we are considering forreaching lower detection limits are using multiple aptamersequences on each of the NP types, using different sized NP, andusing multiple NPs with different aptamer sequences on them.

The ACNP detection process has been performed relativelyquickly in as little as 30 min in complex samples of FBS containingmultiple target cells, as opposed to the hours needed for immu-

nophenotyping or PCR-based methods. Therefore, the ACNPprocedure has been used as a rapid detection method for cancerdetection. This method has shown the benefit of efficient, rapid,and selective potential determinations of clinical diagnostics.

We want to explore expanding this technique to other diseasecell types, specifically cancers and other disease states that arecurrently difficult to monitor. Having spectrally resolvable fluo-rescent dye doped NPs available allows for simple and easyidentification of the various target cells present, which couldpotentially be used for multiplexed detection of target cells bythis system in future applications. In the case of the dye-dopedNPs that were used, the cy5-doped particles consistently producedlower fluorescent signals than both the Rubpy- and TMR-dopedparticles. This was due to the lower concentration of the cy5-dopedwithin these NPs. The ACNP method has the potential to be usedin profiling applications because as panels of aptamers areproduced during the SELEX process, a variety of ACNPs couldbe used to not only identify the disease type but also provide addedinformation to the health care providers that could aid intreatment. This technique lends itself easily to possible automationwith additional investigations involving the magnetic and fluores-cent NPs.

ACKNOWLEDGMENTWe acknowledge Ms. Hui Lin for her expertise in DNA

synthesis. This work was partially supported by a NSF-NIRTgrant and NIH grants.

Received for review November 14, 2006. AcceptedJanuary 16, 2007.

AC062151B


aptamer-conjugated nanoparticles for the collection and detection of multiple cancer cells

Documents