P

Ma

b

a

ARRAA

KRCPMNP

1

wpam(aofats

prdpo

Uf

0d

Talanta 80 (2009) 593–599

Contents lists available at ScienceDirect

Talanta

journa l homepage: www.e lsev ier .com/ locate / ta lanta

rion protein detection in serum using micromechanical resonator arrays

adhukar Varshney a,∗, Philip S. Waggoner a, Richard A. Montagna b, Harold G. Craighead a

School of Applied and Engineering Physics, Cornell University, Ithaca, NY 14853, United StatesRheonix Inc., Ithaca, NY 14850, United States

r t i c l e i n f o

rticle history:eceived 25 May 2009eceived in revised form 15 July 2009ccepted 16 July 2009vailable online 23 July 2009

eywords:esonatorsantilevers

a b s t r a c t

Prion proteins that have transformed from their normal cellular counterparts (PrPc) into infectious form(PrPres) are responsible for causing progressive neurodegenerative diseases in numerous species, such asbovine spongiform encephalopathy (BSE) in cattle (also known as mad cow disease), scrapie in sheep,and Creutzfeldt–Jakob disease (CJD) in humans. Due to a possible link between BSE and CJD it is highlydesirable to develop non-invasive and ante mortem tests for the detection of prion proteins in bovinesamples. Such ante mortem tests of all cows prior to slaughter will help to prevent the introductionof PrPres into the human food supply. Furthermore, detection of PrPres in donated blood will also helpto prevent the transmission of CJD among humans through blood transfusion. In this study, we have

c

rionass amplificationanoparticlesrion protein

continued development of a micromechanical resonator array that is capable of detecting PrP in bovineblood serum. The sensitivity of the resonators for the detection of PrPc is further enhanced by the useof secondary mass labels. A pair of antibodies is used in a sandwich immunoassay format to immobilizePrPc on the surface of resonators and attach nanoparticles as secondary mass labels to PrPc. Secondarymass labeling is optimized in terms of incubation time to maximize the frequency shifts that correspond

c the −1

e det

to the presence of PrP onPrPc in blood serum can b

. Introduction

Prions, discovered by Stanley Prusiner in 1982 [1], are proteins,hich in mis-folded form are believed to be responsible for causing

rogressive neurodegenerative diseases in numerous species, suchs bovine spongiform encephalopathy (BSE) in cattle (also known asad cow disease), scrapie in sheep, and Creutzfeldt–Jakob disease

CJD) in humans. These are fatal diseases and currently untreat-ble. It is believed that these diseases are caused by the conversionf normal cellular prion proteins (PrPc) into a mis-folded infectiousorm (PrPres). Although the infectious form has the same aminocid sequence as the normal form, it is rich in beta-sheet struc-ures as compared to the normal form which is rich in alpha helicaltructures [2].

The current state of technology to detect BSE relies upon theost-mortem detection of prions in homogenates of brain tissue

emoved from slaughtered cows. The numerous methods used toetect PrPres in the brain homogenate are described below. Suchost-mortem testing therefore is not suitable for routine screeningf all animals, regardless of initial symptoms of BSE when animals

∗ Corresponding author at: School of Applied and Engineering Physics, Cornellniversity, 212 Clark Hall, Ithaca, NY 14853, United States. Tel.: +1 607 255 6286;

ax: +1 607 255 7658.E-mail address: mv265@cornell.edu (M. Varshney).

039-9140/$ – see front matter © 2009 Elsevier B.V. All rights reserved.oi:10.1016/j.talanta.2009.07.032

surface of resonators. Our results show that a minimum of 200 pg mL ofected using micromechanical resonator arrays.

© 2009 Elsevier B.V. All rights reserved.

cannot stand or walk. But, it has been shown that the animals notshowing any symptoms of the disease could potentially be carryingthe disease [3,4]. If these animals are not tested for PrPres, theycould conceivably make their way into the human food supply.Thus in order to improve the safety of human food, new tech-nologies must be developed to allow ante mortem tests for PrPres

proteins in the body fluids of the cows. Blood is considered as themost appropriate body fluids, and is known to contain infectivityeven before the onset of the clinical symptoms [5–7]. An impor-tant aspect of these ante mortem tests is the short detection time,so that all animals can be routinely screened prior to slaughter.Contraction of CJD in humans is not only attributed to the con-sumption of contaminated meat, but also to blood transfusions froman infected donor [5,8]. This is a serious problem, especially if anasymptomatic infected individual was not prevented from donat-ing blood because of the lack of sufficiently sensitive screening tests[9–11]. Such presumed healthy humans could trigger the transmis-sion of the disease to other humans through blood transfusion.Therefore the lack of routine testing of cattle is compounded bythe lack of routine testing of human blood for transfusion, estab-lishing a dangerous scenario whereby transmission of infectious

prions from cattle to humans could lead to additional transmissionof those infectious prions from one human to another via bloodtransmission.

There are several methods available for the detection of prionproteins in animals, namely, protein misfolding cyclic amplification,

5 alanta

clflsibdnan(Ptss(sorestnbt

dpmabtcidissdnmatcfibitn

dWtnpuslrlf(rbas

94 M. Varshney et al. / T

onformation-dependent immunoassay, dissociation enhancedanthanide fluorescent immunoassay, capillary gel electrophoresis,uorescence correlation spectroscopy, flow microbead immunoas-ay, etc. [12]. Most of these techniques are intended to detect PrPres

n the brain homogenate and thus will not be suitable for routinelood screening based tests. Saa et al. [13] developed a method toetect PrPres proteins in the blood of presymptomatic animals. Theumber of PrPres proteins in the blood of infected hamsters wasmplified by mixing them with normal brain homogenate and run-ing protein misfolding cyclic amplification (PMCA) to convert PrPc

in the normal brain homogenate) into PrPres. Seven rounds of 144MCA cycles were performed in order to sufficiently amplify the ini-ial low numbers of PrPres to the detection level. The technique wasensitive to detect as low as 20–50 molecules of monomeric ham-ter PrP, however, the time of cyclic amplification alone was 525 h7× 75 h). Another study based on cytometric methods demon-trated analytical sensitivity in the range of 10 aM or 0.24 fg mL−1

f PrPres for serum samples [14]. The cytometric test, however,equires at least a 20 h incubation time, followed by flow cytom-try. Both of these methods are highly sensitive to detect extremelymall amounts of infectious prions in blood, but until either ofhese methods can be automated and performed in a faster man-er, they will prove to be too complicated and time consuming toe practical for routine rapid screening of all cattles prior to slaugh-er.

Micromechanical resonators are considered as suitable candi-ate to develop sensitive and rapid blood based tests to detectrion proteins. Resonators have been used to detect attograms ofasses [15]. They are suitable to detect variety of biomolecules,

s the resonators can be functionalized with a number of differentiorecognition molecules (antibody, nucleic acid, and other pro-ein). Resonators can be fabricated with a variety of materials oroated with a particular material as suitable for the functional-zation chemistry. These devices resonate at a frequency whichecreases as mass is added to the surface. This frequency shift

s measured to detect the presence of mass on the surface. Theensitivity of these resonators to mass detection depends on size,hape, thickness and the fabrication material [16–22]. In order toetect biomolecules of smaller masses, mass amplification tech-iques are used. Amplification is achieved by using secondaryass labeling with massive particles binding to the target analyte

lready present on the surface of resonators [23]. These amplifica-ion techniques are not time consuming as compared to PMCA orytometry used for prion protein detection, although the ampli-cation is limited by the background noise due to non-specificinding. In addition to this, the array format of the resonators makes

t possible to run multiple tests on the same sample to improvehe statistics of detection as well as to reduce false positives andegatives.

In the current work, we demonstrate the potential of theseevices for the direct detection of prion proteins in blood serum.e have fabricated an array of paddlever shaped resonators for

he detection PrPc in blood samples. Secondary antibodies andanoparticles were used for the mass amplification to detect theresence of small amounts of PrPc. A sandwich immunoassay wassed to immobilize PrP to the surface of resonators and sub-equently add nanoparticles. Kinetic studies of secondary massabeling have also been performed to improve the sensitivity of theesonators for detection of PrPc. Due to the limitation of biosafetyevel of our research lab, we were able to work on non-infectiousull length prion protein (PrPc) and not infectious prion protein

PrPres). We project this work to be successfully used with PrPres ineal world based on the affinity of the antibodies used, which haveeen claimed to be equally sensitive for the detection of both PrPc

nd PrPres (based on the information provided by the antibodiesupplier).

80 (2009) 593–599

2. Materials and methods

2.1. Reagents

Two antibodies were used for the sandwich assay. One is con-jugated to the surface of the resonators (capture antibodies) whileanother is used to attach nanoparticles as additional mass labels(detection antibodies). Both antibodies are monoclonal and are pro-duced in mice against different epitopes of bovine prion protein.Prion proteins used in this work are histidine tagged recombinantproteins and are full-length mature part of bovine PrP (25–244)expressed in E. coli BL21 (Millipore Inc.). Capture antibodies areagainst amino acids 23–237 of bovine prion protein (MilliporeInc.) and detection antibodies are against amino acids 123–136and 140–160 of bovine prion protein (Abcam Inc.). Bovine serumalbumin was used to block the surface of the resonators andwas purchased from Sigma. Streptavidin conjugated nanoparticles(R&D Systems, Minneapolis, MN) of diameter 150 nm were usedas secondary mass labels. These particles were irregularly shapedparticles with diameters ranging from 80 nm to 200 nm (as ana-lyzed with SEM). Deprionized fetal bovine serum was obtainedfrom BioRad Laboratories. BioRad obtained fetal bovine serum fromBiowest, Rau. Della Caille, Nuaille, France and origination of theproduct is from Canada. Deprionization was performed by BioRadLaboratories. Aliquots were made from the stock and were storedfrozen until used. Before use the aliquots were thawed and vor-texed. 3-Aminopropyl triethoxysilane (APTES, Sigma) was used tofunctionalize the surface of resonators with primary amine groups.Glutaraldehyde was used as a homo-bifunctional cross-linker toattach antibodies on the silanized surfaces. Glycine solution pre-pared in deionized (DI) water was used for quenching. Both of thesechemicals were purchased from Sigma. Detection antibodies weremodified with biotin to attach 5–7 biotin molecules per antibodymolecule by using NHS-PEO4-biotin (Pierce Chemicals). The biotinper antibody was determined by using EZ biotin quantification kit(Pierce Chemicals).

2.2. Resonator fabrication and functionalization

Resonators were fabricated as described in our previous work[24]. Briefly, a 150 nm thick layer of low-stress nitride was depositedon a thermally oxidized silicon wafer. The silicon nitride devicelayer was then patterned using an anisotropic reactive ion etch.Chips were then dipped in hydrofluoric acid in order to removethe 1.5 �m thick sacrificial silicon dioxide layer under the res-onators and release them from the substrate. Release allows themto move freely and resonate. Resonators in this work were can-tilevered structures with a 3 �m × 10 �m paddle at the free end,also called paddlevers, with typical resonant frequencies of approx-imately 4.6 MHz and quality factors of about 7000 for bare devices.

In this study, APTES (10% solution in dry toluene) was used tosilanize the surface of the resonators. Glutaraldehyde (5% solu-tion in 10 mM borate buffer pH 8.0) was used as a cross-linkerbetween amine groups on capture antibodies (50 �g mL−1 in PBSbuffer) and the surface of the resonators. The conjugation chem-istry to modify the surface of resonator with capture antibodies isshown in Fig. 1. After a wash with DI water (95 rpm for 4 min, twotimes), the devices were quenched with glycine (50 mM solutionin DI water, 30 min) to block unreacted amine group binding siteson the surface of the resonators followed by a blocking with 1%BSA (w/v) in PBS buffer (30 min). Blood serum obtained from Bio-

Rad laboratories was spiked with different concentrations of prionprotein. Excess of prion protein was removed by washing with DIwater (95 rpm for 4 min, two times). This was followed by a 10 minblocking step. Detection antibodies (50 �g mL−1 in PBS buffer) wereused to label prion protein on the surface of resonators (80 min

M. Varshney et al. / Talanta 80 (2009) 593–599 595

F rs. Am3 aldeha

iicaww

2

cmiclrTiqrttiimwd

butsaoAttra

ig. 1. Schematic diagram of the capture of antibodies on the surface of resonato-aminopropyl-triethoxy silane and silanol groups on silicon nitride surface. Glutarntibodies.

ncubation time). After the removal of excess of detection antibod-es (DI water wash, 95 rpm for 4 min, two times), the streptavidinonjugated nanoparticles were attached to biotin on the detectionntibodies. Finally, excess nanoparticles were washed (DI waterash, 95 rpm for 4 min, two times). An orbital shaker was used forashing.

.3. Frequency measurement

Resonators were both excited and sensed in vacuum using opti-al techniques as shown in Fig. 2a [18]. A 405 nm diode laser wasodulated in intensity and used to excite the resonators by focus-

ng the beam near its clamped end. Due to the higher thermalonductivity of the silicon nitride layer as compared to the under-ying silicon dioxide layer, a temperature gradient is induced that isesponsible for the inhomogenous thermal expansion of the device.hermal expansion mismatch between the silicon nitride and oxide

s ultimately responsible for actuation of the sensor. Resonant fre-uencies were determined interferometrically by measuring theeflectance variation from an incident HeNe red laser focused athe free end of the resonator. As the sensor moves in and out ofhe plane, the gap between the wafer and device layers changesn thickness. This dynamic film stack changes the degree of opticalnterference and modulates the intensity of reflected light at the

echanical resonant frequency of the device. The reflected signalas collected by a photodetector, and the resonant spectrum of theevice extracted using a spectrum analyzer.

Resonators chips were dried under the stream of nitrogen gasefore frequency measurements. They were loaded into a small vac-um chamber that was mounted to a motorized stepper stage andhe resonant frequencies were measured. Computer control of thetage and spectrum analyzer allowed easy measurement of largerrays of resonators in a short amount of time. The chip consistedf blocks of arrays of the resonators, each containing 20 devices.

ll devices on the chip were alike and they gave similar response in

erms of the resonant frequency and quality factor. For each concen-ration of PrPc and control experiments (without PrPc), 2 arrays ofesonators (40 devices) per chip were chosen randomly on the chipnd they were used for frequency measurement in a single test. The

ine groups were formed on the resonator surface by covalent reaction betweenyde was used as a cross linker between amine groups on the resonator surface and

average values and errors were calculated based on frequency mea-surements from 40 resonators. These experiments were repeatedon other day as well.

The paddlever shaped cantilevers were used for mass detectionbecause the tip of the cantilever is sensitive to bound mass [22].Paddlever design increases the area for binding in the tip region andin turn improves its sensitivity. Fig. 2b and c shows the conjugationof prion proteins and nanoparticles mass labels on the surface ofresonators. The addition of mass decreases the resonant frequenciesof resonators (Fig. 2d).

3. Results and discussion

3.1. Reaction kinetics of secondary mass labeling

Secondary mass labeling was optimized by varying the incuba-tion time during the antibody and nanoparticles conjugation step.Separate experiments were performed for antibodies and nanopar-ticles to see the effect of each mass label on frequency shifts. Theconcentrations of PrPc used for this purpose for antibodies andnanoparticles were 2 �g mL−1 and 20 ng mL−1, respectively. Theincubation times for antibodies and nanoparticles were varied from20 to 120 min and 10 to 60 min, respectively. The results showedthat the frequency shifts (measured before and after secondarymass labeling) for control and sample increased with the increasein incubation time (data not shown). The optimum incubation timewas determined based on the normalized frequency shift calculatedby subtracting the frequency shift of the control sample from thefrequency shift of the specimen being tested. The control deviceshad no PrPc added to the blood serum. The results indicated that theincubation times of 80 min and 30 min were sufficient to maximizethe effect (normalized frequency shifts) of secondary mass labelingfor the detection of PrPc in blood serum.

The first order Langmuir model is used to study adsorption

kinetics of secondary antibodies and nanoparticles labeling. Thismodel is based on the following assumptions—adsorption occursonly at specific localized sites on the surface and coverage of thesurface takes place in the form of a monolayer of adsorbate. Ourexperimental conditions will sufficiently satisfy these conditions.

596 M. Varshney et al. / Talanta 80 (2009) 593–599

Fig. 2. Schematic diagrams of mechanism of actuation and detection of resonant frequencies and secondary mass labeling. (a) shows an array of bare resonators actuatedand detected by blue and red laser, respectively. Dotted rectangle shows the area of the resonator shown in figures b and c; (b) shows the conjugation of prion protein on thes article( proter

Aatboctotmaoofi

�

wtm

urface using capture antibodies; (c) shows the secondary mass labeling with nanopd) shows the change in resonant frequencies with mass addition. Antibodies, prionespectively. Diagram is not drawn to scale.

s far as the localization of binding sites is concerned, secondaryntibodies are developed against PrP and they will bind only tohe location where PrPc are present. Self-binding and non-specificinding of antibodies were minimum as shown by the presencef very small number of nanoparticles in the SEM image of theontrol (Fig. 3). The possibility of multiple layer formation duringhe addition of nanoparticles is also avoided by sequential additionf secondary antibodies and nanoparticles. Keeping in mind thathere are multiple biotin molecules per antibody and streptavidin

olecules per nanoparticles, cluster formation is minimized by theddition of nanoparticles after conjugation of secondary antibodiesn the surface of the resonator. Assuming an irreversible adsorptionf antibodies and nanoparticles on the surface of the resonator, therst order Langmuir model can be written as:

= 1 − exp−kobs� (1)

here � is the fraction of sites occupied by antibodies or nanopar-icles at time �, kobs is the absolute rate constant observed for the

olecules adsorbed on the surface. � can also be represented in

s. Detection antibodies were used in between nanoparticles and prion protein; andins and nanoaprticles are represented by Y-shape, intertwined ribbon, and spheres,

terms of n, the number of binding sites at time �, and N, the maxi-mum number of sites required to form a monolayer:

� = n

N(2)

Since the frequency shift can be related to the mass change on thesurface of the resonator by:

�f

f= − 1

2�m

m(3)

We can write

�f˛ n and �f∞˛ n (4)

where �f and �f∞ are the frequency shifts when surface is partially(with n number) and completely (with N number) covered with

molecules. Rearranging Eq. (3) and putting the ratio of above twoequations into n/N, we get:

ln(

�f∞ − �f

�f∞

)= −kobs� (5)

M. Varshney et al. / Talanta 80 (2009) 593–599 597

F ng for (a) control, and (b) 200 pg mL−1, (c) 20 ng mL−1 of prion proteins in blood serum.S

ytpcyont

rbsaacntataowtbo

3

wdi

ig. 3. Scanning electron micrographs of resonators with nanoparticle mass labelicale bar is 1 �m in length.

The plot of this equation with left hand side of the equation on-axis and time (�) on x-axis will determine the rate constant forhe adsorption of antibodies and nanoparticles. Fig. 4 shows thelot of Eq. (5) for binding of secondary antibodies and nanoparti-les. A linear curve fit results in a straight line with the equation= −mx where m represents the rate constant of adsorption. Basedn the curve fit, we observed that the rate constant of adsorption ofanoparticles is twice as large as that of antibodies, which revealshat nanoparticles bind faster as compared to antibodies.

The total time of secondary mass labeling could have beeneduced by preparing detection antibody–nanoparticle-conjugatesefore adding to the surface of the resonators. However, previoustudies have shown that these conjugates form clusters and theyre not uniform in size [25]. There are several biotin molecules perntibody and hundreds of streptavidin molecules per nanoparti-le. As a result, several bonds are formed between antibodies andanoparticles that give rise to the formation of clusters. The size ofhe cluster will depend on the relative concentrations of antibodiesnd nanoparticles and incubation time. Therefore, cluster forma-ion will give non-uniform mass labeling and could also result in

higher variation in frequency shifts from one resonator to thether. Additionally, antibodies would not be available for bindingith antigen due to physical entrapment into the clusters. Keeping

hese possibilities in mind, in the current study, we have added anti-odies and nanoparticles in separate steps and have compromisedn the total time for secondary mass labeling.

.2. Prion protein detection in blood serum

The optimized conditions determined in the previous sectionere used to maximize the effect of secondary mass labeling toetect of PrPc in blood serum. PrPc was spiked in blood serum

n a range of concentrations from 20 pg mL−1 to 200 ng mL−1.

Fig. 4. Fitting of the Langmuir model to the change in the values of frequency shiftsfor the (a) secondary antibodies and (b) nanoparticles mass labeling with respect to

immunoreaction time. Y-axis shows the left hand side of the equation ln(

�f∞−�f�f∞

)=

−kobs�

598 M. Varshney et al. / Talanta

Fi

Tpeiietftpasfsmmwosttfcatlcwwabf

m(htmrcaiicr

n[

ig. 5. Calibration curve for detection of unknown concentrations of prion proteinsn blood serum by measuring the frequency shifts.

he frequency shift due to non-specific binding of non-targetedroteins in the blood serum was measured by using a controlxperiment with no PrPc. This control experiment will also providenformation about the presence of normal prion protein presentn the fetal bovine serum. The frequency shift of the controlxperiments were −5519 ± 1201 Hz which matches clearly withhe control experiments that were performed with buffers not withetal bovine serum (shown in our previous work). It seems thathe deprionization performed by BioRad Laboratories removed allrion proteins from fetal bovine serum. No information was madevailable to us about the deprionization protocol to make somepecific comments about removal of PrPc or PrPres or both. Therequency shift for each concentration of PrPc was determined byubtracting frequencies measured after PrPc addition from thateasured after nanoparticle addition. Such frequency shifts wereeasured for 40 resonators per chip and an average of these valuesas used to calculate the mean frequency shift for a concentration

f PrPc. Based on the comparison of mean values of frequencyhifts for different concentrations of PrPc and control sample,he minimum detection limit of PrPc in blood serum was showno be 200 pg mL−1. Fig. 5 shows the calibration curve suitableor the measurement of an unknown concentration of PrPc. Thealibration curve is plotted with the frequency shifts on the x-axisnd concentrations on a logarithmic scale on y-axis. It is observedhat the frequency shifts and concentrations follow an exponentialinear relationship with a function of y = 3.464 exp−7E−0x and aoefficient of relationship (R2) value of 0.963. In an applicationhere we have a sample with unknown concentration of the PrP,e will experimentally determine two frequency shifts—control

nd sample. The normalized frequency shifts for the sample wille located on (x-axis) the calibration curve and tracked (on y-axis)

or the unknown concentration of the PrP in the sample.In order to show the conjugation chemistry of secondary

ass labeling, SEM pictures are shown for two concentrations20 pg mL−1 and 2 ng mL−1) of PrP and the control sample (Fig. 3). Itas been shown in our previous work that the number of nanopar-icles increases with the increase in concentration of the PrPc. The

ass calculated from the number of particles counted on eachesonator was not significantly different (P > 0.05) from the massalculated using experimentally observed frequency shifts. Fig. 3lso shows that the numbers of nanoparticles increase with thencrease in the concentrations of the PrP and the non-specific bind-ng of antibodies and nanoparticles on the control is negligible as

ompared to samples. The number of nanoparticles on control rep-esents the non-specific binding of antibodies and nanoparticles.

The current detection limit of PrP in blood is an order of mag-itude lower than our previous work of detecting PrP in PBS buffer24]. This can be attributed to the optimized protocol, thinner

80 (2009) 593–599

resonators and the blood serum proteins acting as blocking agents.Thinner resonators with smaller mass show large frequency shiftsfor the same change in mass on the surface, thus are more sensitivefor the detection of small masses [16]. The albumin present inthe bovine serum is most commonly used as blocking agent inimmunoassays. Other blood serum proteins also act as block-ing agent which may have reduced the non-specific binding ofnanoparticles on the surface of resonators. The average frequencyshift of the control sample in the current study was reduced by25% as compared to the average frequency shift for the control inour previous work. Though these frequency shift values were notsignificantly different (P > 0.05).

The concentration of PrPres in the buffy coat (a componentof the blood) extracted from animals in their symptomatic andpresymtomatic phase of the disease is expected to be 1 pg mL−1 and0.1 pg mL−1, respectively [6,11]. The resonators used in the presentstudy could be used for the detection of such low concentrationsof the PrPres in the blood samples. Our current detection limit isnot the fundamental limit of the system, rather is dependent onthe capture efficiency of antibodies and uniform coating of anti-bodies on the surface. In one of our unpublished work, we haveused cantilever array technology for the detection of fg/ml of simi-lar mass of protein (prostate specific antigen). Thus antibodies andother reagents also play critical role in the detection low amounts ofprotein with this detection system. Moreover, thickness and shapesof the device can also play role in improving the detection limits.Concentration of PrPres prior to detection could also be performed todetect low levels of PrPres in the buffy coat of presymptomatic ani-mals. Thus combining all these possibilities we are very confidentthat in the future current detection system could be used for thedetection PrP in preclinical animals. The specificity of the systemis a function of the antibodies. Although, our current system wasdeveloped for the detection of PrPc, the same antibodies are alsoclaimed to be sensitive for the detection of PrPres (Product specifi-cation sheet, Chemicon Inc. and Abcam Inc.), and thus the currentsystem could also be optimized for the detection of PrPres. The com-bination of a concentration step, use of high density nanoparticles,thinner devices and different shapes of resonators could be instru-mental in the detection of extremely low levels of infectious prionproteins in the blood of animals in the early stages of the disease. Ifso, resonators based system could serve as an industry standard toscreen all animals going into the human food supply.

4. Conclusions

A detection system based on arrayed resonant sensors was suc-cessfully used to detect a minimum of 200 pg mL−1 of PrPc in bloodserum, and the response of the system was a linear exponential for arange of concentration from 20 pg mL−1 to 200 pg mL−1. Additionaleffort will be required to improve the current detection limit andrun appropriate field trials to detect infectious prions in the bloodof animals at an early stage of the disease.

Acknowledgments

This material is based upon work supported by the CooperativeState Research, Education, and Extension Service, U.S. Departmentof Agriculture, under Award No. 2007-35603-17746. Any opinions,findings, conclusions, or recommendations expressed in this pub-lication are those of the author(s) and do not necessarily reflect the

view of the U.S. Department of Agriculture.

References

[1] S.B. Prusiner, Science 216 (1982) 136.

alanta

[

[[[

[

[

[

[

[

[[

M. Varshney et al. / T

[2] K.M. Pan, M. Baldwin, J. Nguyen, M. Gasset, A. Serban, D. Groth, I. Mehlhorn, Z.Huang, R.J. Fletterick, F.E. Cohen, S.B. Prusiner, Proc. Natl. Acad. Sci. 90 (1993)10962.

[3] P. Brown, Vox Sang. 89 (2005) 63.[4] N. Hunter, J. Foster, A. Chong, S. McCutcheon, D. Parnham, S. Eaton, C. MacKenzie,

F. Houston, J. Gen. Virol. 83 (2002) 2897.[5] C.A. Llewelyn, P.E. Hewitt, R.S. Knight, K. Amar, S. Cousens, J. Mackenzie, R.G.

Will, Lancet 363 (2004) 417.[6] P. Brown, L. Cervenakova, H. Diringer, J. Lab. Clin. Med. 137 (2001) 5.[7] F. Houston, J.D. Foster, A. Chong, N. Hunter, C.J. Bostock, Lancet 356 (2000) 999.[8] A.H. Peden, M.W. Head, D.L. Ritchie, J.E. Bell, J.W. Ironside, Lancet 364 (2004)

527.

[9] Q. Schiermeier, Nature 409 (2001) 658.10] L. Ingrosso, V. Vetrugno, F. Cardone, M. Pocchiari, Trends Mol. Med. 8 (2002)

273.11] C. Soto, Nat. Rev. Microbiol. 2 (2004) 809.12] A. Sakuda, I. Nakamura, K. Ikuta, T. Onodera, J. Vet. Med. Sci. 69 (2007) 329.13] P. Saa, J. Castilla, C. Soto, Science 313 (2006) 92.

[[[

[

80 (2009) 593–599 599

14] L. Trieschmann, A.N. Santos, K. Kaschig, S. Torkler, E. Maas, H. Schatzl, G. Bohm,BMC Biotechnol. 5 (2005) 26.

15] B. Ilic, H.G. Craighead, S. Krylov, W. Senaratne, C. Ober, P. Neuzil, J. Appl. Phys.95 (2004) 3694.

16] P.S. Waggoner, H.G. Craighead, Lab on a Chip, 7, 2007, p. 1238.[17] A. Gupta, D. Akin, R. Bashir, J. Vac. Sci. Technol. B 22 (2004) 2785.18] B. Ilic, Y. Yang, K. Aubin, R. Reichenbach, S. Krylov, H.G. Craighead, Nano Lett. 5

(2005) 925.19] A.K. Gupta, P.R. Nair, D. Akin, M.R. Ladisch, S. Broyles, M.A. Alam, R. Bashir, Proc.

Natl. Acad. Sci. 103 (2006) 13362.20] J. Tamayo, D. Ramos, J. Mertens, M. Calleja, Appl. Phys. Lett. 89 (2006) 224104.21] J. Fritz, M.K. Baller, H.P. Lang, H. Rothuizen, P. Vettiger, E. Meyer, H.-J. Güntherodt,

C. Gerber, J.K. Gimzewski, Science 288 (2000) 316.22] P.S. Waggoner, H.G. Craighead, J. Appl. Phys. 105 (2009) 054306.23] M. Su, S. Li, V.P. Dravid, Appl. Phys. Lett. 82 (2003) 3562.24] M. Varshney, P.S. Waggoner, C.P. Tan, R.A. Montagna, H.G. Craighead, Anal. Chem.

80 (2008) 2141.25] M. Varshney, L. Yang, X. Su, Y. Li, J. Food Protect. 68 (2005) 1804.