Ecc

Ja

Cb

a

ARRAA

KCFBSSP

1

ptinemeAMia�Smcs

0d

Biosensors and Bioelectronics 24 (2009) 3372–3378

Contents lists available at ScienceDirect

Biosensors and Bioelectronics

journa l homepage: www.e lsev ier .com/ locate /b ios

nhancement of sensitivity and specificity by surface modification ofarbon nanotubes in diagnosis of prostate cancer based onarbon nanotube field effect transistors

un Pyo Kima, Byung Yang Leeb, Joohyung Leeb, Seunghun Hongb, Sang Jun Sima,∗

Nano-optics and Biomolecular Engineering National Laboratory, Department of Chemical Engineering, Sungkyunkwan University, 300 Chunchun-dong,hangan-gu, Suwon 440-746, South KoreaDepartment of Physics and Astronomy, Seoul National University, San 56-1, Sillim-dong, Gwanak-gu, Seoul 151-742, South Korea

r t i c l e i n f o

rticle history:eceived 13 March 2009eceived in revised form 29 April 2009ccepted 30 April 2009vailable online 7 May 2009

eywords:arbon nanotube

a b s t r a c t

This paper presents a simple and sensitive method for the real-time detection of a prostate cancer marker(PSA-ACT complex) through label-free protein biosensors based on a carbon nanotube field effect transis-tor (CNT-FET). Herein, the CNT-FET was functionalized with a solution containing various linker-to-spacerratios, the binding event of the target PSA-ACT complex onto the receptor detected by monitoring thegating effect caused by charges in the target PSA-ACT complex. Since the biosensors were used in a buffersolution, it was crucial to control the distance between the receptors through introduction of linkers andspacers so that the charged target PSA-ACT complex could easily approach the CNT surface within the

ield effect transistoriosensorurface modificationpacerrostate cancer

Debye length to give a large gating effect. The results show that CNT-FET biosensors modified with onlylinkers could not detect target proteins unless a very high concentration of the PSA-ACT complex solution(∼500 ng/ml) was injected, while those modified with a 1:3 ratio of linker-to-spacer could detect 1.0 ng/mlwithout any pretreatment. Moreover, our linker and spacer-modified CNT-FET could successfully blocknon-target proteins and selectively detect the target protein in human serum. Significantly, this strategy

antibnsors

can be applied to generalsensitive electronic biose

. Introduction

Prostate-specific antigen (PSA), a glycoprotein consisting of 93%eptide and 7% sugar, produced exclusively by prostatic tissue, ishe best serum marker currently available for diagnosing and mon-toring prostate cancer (Loeb and Catalona, 2007). Since there iso curative therapy available for prostate cancer, early stage dis-ase detection is the best hope for an increasing mortality rate. Theajor forms of PSA found in serum are complexes with two major

xtracellular serine protease inhibitors, �1-antichymotrypsin (PSA-CT, MW 90 kDa) and �2-macroglobulin, and a free form (f-PSA,W 34 kDa). PSA-ACT is the predominant form of PSA complex;

t is immunoreactive, whereas PSA-AMG is not. The minor formsre constituted by a combination of PSA and protein C inhibitor,1-antitrypsin, and �-trypsin (Armbruster, 1993; Lilja et al., 1991;

avage and Waxman, 1996). Therefore, PSA-ACT and f-PSA are twoolecules that, if measured, can be used to determine prostate can-

er. In cases of normal human, the concentration of PSA in humanerum will be smaller than 4.0 ng/ml, while the cancer is sup-

∗ Corresponding author. Tel.: +82 31 290 7341; fax: +82 31 290 7272.E-mail address: simsj@skku.edu (S.J. Sim).

956-5663/$ – see front matter © 2009 Elsevier B.V. All rights reserved.oi:10.1016/j.bios.2009.04.048

ody-based detection schemes and enables production of very simple andto detect clinically important biomarkers for disease diagnosis.

© 2009 Elsevier B.V. All rights reserved.

posed to be present if it is higher than 20 ng/ml. The range from4.0 to 20 ng/ml is considered a “gray scale” where further medicalexamination should be performed before the disease is identified.Therefore, PSA quantifying assays are highly recommended by med-ical doctors to screen, diagnose, or monitor prostate cancer.

Conventional assays for PSA detection mostly involve a mon-oclonal or polyclonal antibody of PSA tagged with an enzyme,fluorophore, or radioactive isotope (Armbruster, 1993). Whilethese methods are sensitive and specific, they nevertheless sufferfrom several disadvantages such as their inherent complexity andrequirement for multiple reagents and steps, signal amplification,relatively large sample size, complex data analysis, and high cost.Therefore, it is highly desirable to develop instrumentation withfeatures of (i) high sensitivity and specificity; (ii) real-time detec-tion; (iii) non-labeling method; (iv) rapid, flexible, multiplexingassaying; (v) portable, disposable, and low cost.

Recently, a biosensor satisfying all the aforementioned require-ments was discovered using a carbon nanotube field effect

transistor (CNT-FET) whose conductance changes by the chargetransfer from molecules adsorbed onto the CNT surface (Allen etal., 2007; Bradley et al., 2004; Chen et al., 2003; Guo et al., 2005;Liu, 2008; Merkoci et al., 2005; Qi et al., 2003; Star et al., 2003;Robertson, 2004; Trojanowicz, 2006). In one of the bio-detection

Bioelec

sttcFoaeddam

tpelto2sfoObpstm

F((

J.P. Kim et al. / Biosensors and

chemes, the CNT-FET surface is functionalized with specific recep-or molecules that bind to desired target biomolecules. When thearget molecules bind to the receptor molecules in solution, theharges of the target molecules affect the conductance of the CNT-ETs. In buffer solution, the double layer is formed within the rangef Debye length (approximately 3 nm in 10 mM buffer solution)round the CNT-FETs (Schasfoort and Bergveld, 1990; Schasfoortt al., 1990). The target and receptor biomolecules should fit in theouble layer to change the conductance of CNT-FETs. However, theimension of the antibodies (∼10–12 nm) used as receptors is usu-lly much larger than the Debye length, implying that the targetolecules cannot approach the CNT-FETs within the double layer.The strategy to overcome this problem is utilizing small recep-

ors such as aptamers (Maehashi et al., 2007; So et al., 2005),eptides, and antibody-binding fragments [Fab and F(ab)2] (Kimt al., 2008). Although aptamers, which are artificial nucleic acidigands, are very small (approximately 2 nm) and show high selec-ivity, specificity, and affinity for amino acids, drugs, proteins, andther small molecules (Lee et al., 2006a,b; James, 2001; Song et al.,008), they have not yet been developed for many diseases and canuffer from pleomorphism. Moreover, to make antibody-bindingragments as potential receptors, antibodies must be cut by vari-us enzymes such as papain and pepsin (Harris et al., 1992; Ng andsawa, 1997). This method has very high sensitivity on the CNT-FET

iosensor. However, it must pass through a complex pretreatmentrocess involving enzymatic digestion, reaction incubation, columneparation, etc. In order to solve the above-mentioned disadvan-ages, a CNT surface was specially designed by introducing spacer

olecules (1-pyrenbutanol).

ig. 1. (A) Schematic diagram of the CNT surface modified with various ratios of linker toSurface C); (d) 1:9 PASE to PB (Surface D); (e) only PB (Surface E). (B) Photographs of the Cb) optical micrograph of a CNT-FET device with gold electrodes passivated with PR; (c) sc

tronics 24 (2009) 3372–3378 3373

In this paper, we report the first successful demonstration ofa CNT-FET biosensor using 1-pyrenebutanoic acid succinimidylester as linkers and 1-pyrenbutanol as spacers. CNT-FETs werefunctionalized with five kinds of solutions containing variouslinker-to-spacer ratios: (a) Surface A [only linker]; (b) Surface B[linker:spacer = 1:1]; (c) Surface C [linker:spacer = 1:3]; (d) SurfaceD [linker:spacer = 1:9]; (e) Surface E [only spacer] (Fig. 1A). Theresponse of each CNT-FET was then monitored in real-time afterintroduction of the PSA-ACT complex at various concentrations(0.1–500 ng/ml).

2. Materials and methods

2.1. Materials

Purified single-walled nanotubes (SWNTs) were purchased fromCarbon Nanotechnologies Inc. (USA), and 1-pyrenebutanoic acidsuccinimidyl ester (PASE) was obtained from Molecular ProbesInc. (USA). HS(CH2)11(OCH2CH2)6OCH2NH2 (HS-OEG6-NH2) andHS(CH2)11(OCH2CH2)3OH (HS-OEG3-OH) were purchased from CosBiotech (Korea). PSA-ACT complex and PSA-ACT complex mon-oclonal antibody (PSA-ACT mAb) were supplied by FitzgeraldIndustries International Inc. (USA). 1-Pyrenbutanol (PB), humanserum, bovine serum albumin (BSA) and other chemical reagents

were supplied by Sigma–Aldrich (USA). Platinum wire (diame-ter 0.5 mm) was purchased from Tae Won Scientific Corporation(Korea). Silicon Isolators (diameter 2.5 mm, height 2.0 mm) forthe reaction chamber were also purchased from Sigma–Aldrich.Deionized water (DI, 18.3 M� resistance), obtained from a water

spacer: (a) only linker (Surface A); (b) 1:1 PASE to PB (Surface B); (c) 1:3 PASE to PBNT-FET device: (a) optical image of a CNT-FET chip with patterned gold electrodes;anning electron microscopy image of the channel of the CNT-FET device.

3374 J.P. Kim et al. / Biosensors and Bioelectronics 24 (2009) 3372–3378

F PSA-A2 nker av

pp

2

tfipfTi

F(

ig. 2. (a) Cross-section schematic diagram of the experimental setup for real-time0 mm. (b) Electrical properties of the CNT-FET devices before modification of the lis. bias voltage (VDS) at various gate voltages (VG).

urification system (Human Corporation, Korea), was used for thereparation of the washing and buffer solutions.

.2. Fabrication of CNT-FET devices

Self-assembled monolayer of methyl-terminated octadecyl-richlorosilane patterns on silicon oxide wafer were generated by

rst patterning AZ 5214 photoresist via standard photolithogra-hy, dipping the wafer in OTS solution (1:500, v/v in hexane)or 3 min, and finally removing the PR patterns using acetone.he SWNT solution was prepared by dispersing purified SWNTsn 1,2-dichlorobenzene with ultrasonication for 1 h (concentra-

ig. 3. SEM images of the channel area after 12 nm gold nanoparticles were introduced ob) Surface B, (c) Surface C, (d) Surface D, (e) Surface E and (f) schematic diagram of gold n

CT complex sensing. The respective diameter and height of the chamber are 25 andnd spacer molecules in PBS buffer solution (10 mM PBS, pH 7.4). Drain current (IDS)

tion ∼0.1 mg/ml). Then, the patterned silicon oxide wafer wasdipped in the SWNT solution for 10 s, rinsed thoroughly with 1,2-dichlorobenzene, and dried with nitrogen gas (Lee et al., 2006a,b).This step allowed SWNTs to be adsorbed selectively onto bareSiO2 regions on the wafer, while the methyl-terminated OTS SAMsblocked non-specific adsorption of the CNTs. After assembly of theSWNTs, electrodes (30-nm Au layer on 10-nm Pd) were fabricated

via standard photolithography and a lift-off process (Lee et al.,2006a,b). Finally, additional photolithography processing to patternPR (AZ 5214) was performed to cover up the electrodes and avoidleakage current from electrodes in the buffer solution. To make thereaction chamber, a Silicone Isolator (silicon well) was attached

n the CNT surfaces modified with various ratios of linkers to spacers. (a) Surface A,anoparticles attached to the linkers and spacers modified-CNT surface.

Bioelec

oF

2

(opPnwaeteoww

2

rettbspasO1ytat1(te

2

dcstbee(tct1cawuicc

erence electrode (gate) bias (relative to the source and drain). Thecharacteristics of the CNT-FET in the buffer solution were inves-tigated as shown in Fig. 2(b). The CNT-FET exhibited increasingsource–drain current with negative gate bias, but decreasing as the

J.P. Kim et al. / Biosensors and

n the channel area of CNT-FET at the black square as shown inig. 1B.

.3. Preparation of the CNT surface for sensing in solution

The CNT surface was designed by a linker (PASE) and spacerPB). The 1.0-mM mixed solution consisting of a 1:1, 1:3, 1:9 ratiof PASE to PB and 1.0 mM homogeneous PASE solution were pre-ared in absolute methanol (Surface A, only linker; Surface B, 1:1ASE to PB; Surface C, 1:3; Surface D, 1:9; Surface E, only PB). Foroncovalent functionalization of the CNT surface, CNT-FET devicesere incubated with prepared solutions for 2 h at room temper-

ture followed by rinsing with pure methanol to wash away anyxcess reagent (Chen et al., 2001). For covalent immobilization ofhe receptor proteins on the CNT surface, each CNT-FET device wasxposed to 20 �g/ml PSA-ACT mAb in 10 mM PBS buffer (pH 7.4)vernight at room temperature. The un-reacted functional groupsere blocked by 100 mM ethanolamine and washed thoroughlyith DI water for 6 h, and then dried with nitrogen gas.

.4. Preparation of gold nanoparticles

Gold nanoparticles (AuNPs) were synthesized by sodium cit-ate reduction of an aqueous HAuCl4 solution as described by Jit al. (2007). A volume of 0.5 ml of 50 mM HAuCl4 was heatedo a boil and 2.5 ml of 1% trisodium citrate solution was addedo the boiling solution under vigorous stirring. The solution wasoiled another 15 min to complete reduction of the gold ions. Theolution was then stirred another 15 min and cooled to room tem-erature. This method yielded spherical, ruby red particles withn average diameter of ∼12 nm. After synthesis of the colloidalolution, the AuNPs were capped with a 1:10 molar ratio of HS-EG6-NH2:HS-OEG3-OH by mixing 9.0 ml of the gold solution and.0 ml of an ethanolic solution of HS-OEG6-NH2:HS-OEG3-OH toield an effective HS-OEG6-NH2:HS-OEG3-OH capping concentra-ion of 0.5 mM (Cao and Sim, 2007). After gentle stirring for 6 ht room temperature, the mixed solution was washed to separatehe unbound chemicals from AuNPs by centrifugation for 15 min at4,000 rpm. Then, the pellet was resuspended in 10 mM PBS bufferpH 7.4). The washing step was repeated three times and the func-ionalized AuNPs stored in 10 mM PBS buffer at 4 ◦C for furtherxperiments.

.5. Electrical measurement

The electrical properties of the CNT-FET devices during the intro-uction of the target proteins were measured by a semiconductorharacterization system (Keithley, 4200, USA) connected to a probetation that makes electrical contact to the source and drain elec-rodes of the CNT-FET. The devices were stabilized in a 10-mM PBSuffer solution (pH 7.4), and a platinum wire was inserted as a gatelectrode (VG). When biased at VG = 0, the gate electrode eliminatedlectrical noise caused by the addition of analyte. A source–drainSD) bias of 50 mV and VG = 0 were maintained throughout the elec-rical measurements. For IDS–VG measurements before and afterhemical and protein adsorption on the CNT surface, the gate elec-rode potential was swept from −0.3 to +0.3 V with an SD bias of0 mV. First, a PBS buffer (10 mM, pH 7.4) solution of 5.0 �l for theontrol of the electrical signal was introduced into the silicon well

ttached to channel regions of the CNT-FET devices. Then, an analyteith increasing concentrations was introduced into the chambersing sample volumes of 5.0 �l. At that time, to examine sensitiv-

ty, specificity, and selectivity of the CNT-FET device, the PSA-ACTomplex was diluted with 10 mg/ml of human serum to yield con-entrations of 0.1, 1, 10, 50, 100, and 500 ng/ml.

tronics 24 (2009) 3372–3378 3375

3. Results and discussion

3.1. Device layout and electrical characterization of CNT-FET

The CNT-FET devices fabricated via the linker-free directedassembly method (Lee et al., 2006a,b) and the detailed images ofCNT-FET device are shown in Fig. 1A (additional information). In thismethod, the CNT network patterns were formed directly on a baresilicon oxide surface without any linker molecules and were usedas the channel for FETs. The on–off ratio was low (∼3), as with othernetwork transistors, because the SWNT network was comprised ofboth semiconducting and metallic CNTs. However, our CNT-FET hadsignificant advantages for sensor applications. First, since the fab-rication method did not use any linker molecules, possible signalcontamination was minimized by the linker molecules. Further-more, in this process, as the SWNTs were adsorbed onto the baresilicon oxide surface, the surface became non-polar and blocked for-mation of the multiple SWNT layers. This ‘self-limiting mechanism’warranted the reproducibility of the CNT-FET fabrication process(Lee et al., 2006a,b).

The electrical properties of the CNT-FETs were measured in real-time at room temperature. The experimental configuration shownin Fig. 2(a) indicates the electric field applied to the CNT via the ref-

Fig. 4. Change in device characteristic IDS (VG) upon modification of the CNT surface.(a) Current vs. gate voltage characteristics before and after the CNT-FET was modifiedwith a 1:3 PASE to PB and PSA-ACT mAb on bare CNT surface. (b) Current vs. gatevoltage characteristics before and after the CNT-FET was modified with only PB andPSA-ACT mAb on a bare CNT surface.

3 Bioelec

gTi

3l

ntboilraoairteprCft

Fitoba

376 J.P. Kim et al. / Biosensors and

ate bias increased positively at the source–drain bias of 50 mV.his result indicated that this device showed p-type characteristicsn the blank PBS buffer.

.2. Characterization of carbon nanotube surface modified withinkers and spacers

In the general biosensor system, sensitivity of the specific sig-al is closely related to the immobilized pattern of the ligand onhe sensor surface. That is, sensitivity of the CNT-FET device cane enhanced by optimizing ligand surface density. To obtain anptimal surface density of the ligand, the space between ligandss important for the target proteins to easily access the immobi-ized receptors. Thus, we introduced the surface design by a molaratio of linker:spacer (PASE:PB) on the CNT surface to improveccessibility of the target. To confirm the immobilization patternf the receptors depending on molar ratio of linker and spacer,bout 12 nm AuNPs modified with amine and hydroxyl groups weremmobilized on the CNT surface treated with chemicals instead ofeal PSA-ACT mAb receptors (Fig. 3). SEM images of Fig. 3 showhat AuNPs as receptors bind strongly to the immobilized link-rs on the CNT surface, and that the distance between receptors,

er same area, is also different with a molar ratio of linker toeceptor. When only linker was treated without a spacer on theNT surface, Au nanoparticles were compactly immobilized (Sur-

ace A). Moreover, by increasing the molar ratio of the spacer,he distance between AuNPs widened because the molar ratio of

ig. 5. Electric response of the CNT-FET following molecular recognition with a PSA-ACT cmmune reaction of PSA-ACT mAb and the PSA-ACT complex. (b) Electrical signal of CNT-he PSA-ACT complex. (c) Electrical signal of CNT-FET modified with Surface C, following thf CNT-FET modified with Surface D, following the immune reaction of PSA-ACT mAb anduffer (10 mM, pH 7.4) was introduced in the chamber of the CNT-FET device. Electrical sigfter the control curve was obtained in PBS buffer. The arrow indicates the point of analyt

tronics 24 (2009) 3372–3378

the spacer gradually increased and the molar ratio of the linkerwas reduced (Surface B, C, and D). The SEM image of the con-trol experiment, in which the CNT surface was without any priortreatment, showed a lack of AuNPs binding, like the case of Sur-face E, in which the CNT surface was treated with only PB (SurfaceE in Fig. 3). Therefore, as shown in the above results, controlof the distance between receptors by using the spacer could beasserted.

Fig. 4 shows the drain current versus gate voltage characteris-tics of the CNT-FET device at a fixed bias of 0.01 V, before and afterchemical and antibody modifications on the bare CNT surface. Wealso observed from the curve of bare CNT that the current (IDS) wassuppressed with the positive increase of VG, indicating a typicalcharacter of p-type FETs. After PASE:PB mixture solution and onlyspacer solution were treated on the bare CNT surface for 2 h, con-ductance of the CNT-FET device changed in both cases, indicatingthat the selected linker and spacer were successfully immobilizedon the CNT surface. Then, the modified CNT surface was exposedto 20 �g/ml of PSA-ACT mAb in PBS buffer overnight. Conductancein the case of the CNT-FET device modified with PASE/PB mixturesolution was changed. However, in case of CNT-FET device modi-fied with only spacer, a change in conductance was not observed,

indicating that the selected PASE linker effectively immobilizedreceptors for the detection of target proteins. Moreover, selectedPB spacer was successfully blocked non-specific binding on the CNTsurface. Therefore, PASE and PB are of utility for stabilization of theCNT surface.

omplex. (a) Electrical signal of the CNT-FET modified with Surface A, following theFET modified with Surface B, following the immune reaction of PSA-ACT mAb ande immune reaction of PSA-ACT mAb and the PSA-ACT complex. (d) Electrical signal

the PSA-ACT complex. The control indicates electrical conductance where only PBSnal curves were determined through injection of analyte at various concentrationse injection.

Bioelectronics 24 (2009) 3372–3378 3377

3C

raw1m(mimhssd

lps(wPinZicefbrtlcst

itideotowoaambicCa1d

CaPaui

Fig. 6. Calibration curve of IgG detection on the CNT-FETs modified with (a) 1:3

J.P. Kim et al. / Biosensors and

.3. Detection of the PSA-ACT complex in human serum onNT-FET modified by a linker and spacer

Using the prepared devices, we performed a systematic studyegarding the effects of surface design, by various ratios of linkersnd spacers on the sensitivity of CNT-FET biosensors. First, CNT-FETsere functionalized with four solutions (only linker, 1:1, 1:3, and:9 ratio of PASE to PB). Then, after the formation of a self-assembledonolayer on the CNT surface, change in electrical conductance

G/G0) of the devices upon the addition of only human serum andixtures containing human serum and PSA-ACT complex with var-

ous concentrations (0.1–500 ng/ml) was monitored. In all cases ofodified CNT-FET devices, when the PBS buffer solution and only

uman serum were separately injected into chamber, the electricalignal remained unchanged in either case. This indicated that non-pecific binding was effectively suppressed in the modified CNT-FETevice.

Fig. 5(a) shows the response of the CNT-FETs modified with onlyinker when analyte was injected on the devices. The PSA-ACT com-lex bore negative charges because its isoelectric point (prostatepecific antigen pI = 6.8, �1-antichymotrypsin pI = 6.8 and 4.1–4.3)Armbruster, 1993; Lilja et al., 1991; Savage and Waxman, 1996)as lower than the pH of the PBS buffer solution (pH 7.4). Thus, the

SA-ACT complex adsorbed onto modified the CNT surface shouldncrease the CNT-FET conductance as it is equivalent to applyingegative gate voltages (Li et al., 2005; Patolsky and Lieber, 2005;heng et al., 2005). However, upon addition of all solutions contain-ng 0.1–500 ng/ml of the PSA-ACT complex, the devices exhibitedonstant electrical conductance-like control. This behavior can bexplained by the PSA-ACT complex not approaching the CNT sur-ace within the distance of the Debye length to give a gating effectecause the size of PSA-ACT mAb was large and a distance betweeneceptors were immobilized so closely on the CNT surface. Thus,arget proteins were not accessible to PSA-ACT mAb within Debyeength due to dense formation between receptors. As a result, theonductance of the CNT-FET was unchanged upon addition of theolution unless a very high concentration PSA-ACT complex solu-ion was injected (∼500 ng/ml).

To resolve this problem, various CNT surfaces were designed,ntroducing 1-pyrenebutanol as a spacer (Fig. 1). Fig. 5(b) showshe response of the CNT-FETs modified with a solution contain-ng a 1:1 ratio of PASE to PB when the analyte was injected on theevices. Upon addition of 0.1–100 ng/ml of the PSA-ACT complex,lectrical signal went unchanged, however, addition of 500 ng/mlf the PSA-ACT complex increased device conductance. As the dis-ance between the receptor widened due to addition of spacersn the CNT surface, the negatively charged proteins can approachithin the distance of the Debye length and affect the conductance

f the CNT-FETs with greater ease. Thus, the CNT-FET modified with1:1 ratio of linker to spacer could detect the PSA-ACT complex

t a concentration of 500 ng/ml. However, the detectable mini-um concentration of the CNT-FET biosensor was still very high

ecause the space that target proteins could bind to the receptorsmmobilized onto the CNT surface within Debye length was insuffi-ient. Therefore, a distance between receptors immobilized on theNT surface should be wider than the CNT surface modified with1:1 ratio of PASE to PB; the CNT-FET biosensor modified with a:1 ratio of linker to spacer was not applicable in the diagnosis ofisease.

To obtain better sensitivity for the CNT-FET-based detection, theNT surface was functionalized with a 1:3 ratio of linker to spacer,

nd their change in conductance monitored upon the addition of aSA-ACT complex solution at various concentrations, between 0.1nd 500 ng/ml. In this case, the conductance increased stepwisepon exposure to the PSA-ACT complex at concentrations increas-

ng from 1.0 to 100 ng/ml (Fig. 5(c)). However, the analyte could

and (b) 1:9 PASE to PB upon the addition of the PSA-ACT complex under variousconcentrations (0.1–500 ng/ml). The error bars illustrate the relative standard devi-ation (R.S.D.) for the three replicates. The inset describes a logarithmic linear rangebetween the electrical signal and the PSA-ACT complex concentrations.

not be determined at lower concentrations of 0.1 ng/ml. At a con-centration of 500 ng/ml, the change in the electrical signal was notmuch higher than that of the 100 ng/ml PSA-ACT complex sample(Fig. 6(a)). Thus, as shown in the inlet of Fig. 6(a), the linear dynamicrange was shown to be from 1 to 100 ng/ml and the detectable min-imum concentration (1 ng/ml) in the case of the CNT-FETs modifiedwith 1:3 ratio of linker to spacer was ameliorated, it was lower thanthat of the CNT-FETs modified with only linker or 1:1 ratio of linkerand spacer.

On the other hand, in cases of the CNT surface modified withsolutions containing a 1:9 ratio of PASE to PB, the conductanceincreased stepwise upon exposure to the PSA-ACT complex atconcentrations increasing from 1.0 to 50 ng/ml, except for concen-trations of 0.1 and 500 ng/ml of the PSA-ACT complex (Fig. 5(d)).In this case, the detectable minimum concentration was 1.0 ng/ml,the same concentration as the CNT-FET modified with a 1:3 ratio ofPASE to PB. However, the dynamic range of CNT-FET modified witha 1:9 ratio of PASE to PB was narrower than the CNT-FET modifiedwith a 1:3 ratio of PASE to PB (inset of Fig. 6(b)). This result indicates

that the binding sites of CNT-FET modified with a 1:9 ratio of PASEto PB were reduced more than that of CNT-FET modified with 1:3ratio of PASE to PB because the ratio of the spacer was increasedrelatively more than the ratio of the linker in CNT surface modifiedwith PASE and PB. Therefore, the sensitivity and the dynamic range

3 Bioelec

ow

4

spvtlaotamnassaeatiFiutm

A

NMRg

378 J.P. Kim et al. / Biosensors and

f CNT-FET biosensor can improve accordingly, as the CNT surfaceas properly modified with a linker and spacer.

. Conclusions

The important factors for the development of effective biosen-ors were high sensitivity and simplification of the pretreatmentrocess. Herein, we devised CNT-FETs that were modified underarious molar ratios of linkers to spacers so that the charged pro-eins could more easily approach within the distance of the Debyeength as the distance between the receptor widened due to theddition of spacers on the CNT surface and affect the conductancef the CNT-FETs. In this way, we were able to lower the detec-ion limit to a protein concentration of 1.0 ng/ml, without cuttingntibody or labeling the target proteins. Moreover, our CNT-FETodified with the linker to spacer could successfully suppress

on-target proteins and selectively detect the target protein inmixed environment containing high concentrations of human

erum. This strategy enabled the construction of CNT-FET biosen-or systems based on the well-established immune reactions ofntigen–antibody. The CNT surface modification by spacers has sev-ral advantages in biosensors based on CNT-FET. First, all of thentibodies developed for disease diagnosis can be used to buildhe CNT-FET biosensors. In addition, CNT surface modification byntroducing spacers significantly improves sensitivity of the CNT-ET biosensors and the pretreatment process for use of the antibodys simplified. Thus, the strategy of the CNT surface modificationsing a spacer in the CNT-FET biosensors can serve as a major break-hrough, enabling various important applications in proteomics and

edical diagnostics.

cknowledgments

This work was supported by the grant (No. 10017190) fromext Generation New Technology Development Program from theinistry of Commerce, Industry and Energy (MOCIE) and National

esearch Laboratory (NRL) Program grant funded by the Koreaovernment (MEST) (grant no. R0A-2008-000-20078-0) of the

tronics 24 (2009) 3372–3378

Republic of Korea. SH acknowledges the support from the NRL pro-gram (no. R0A-2004-000-10438-0).

References

Allen, B.L., Kichambare, P.D., Star, A., 2007. Adv. Mater. 19, 1439–1451.Armbruster, D.A., 1993. Clin. Chem. 39, 181–195.Bradley, K., Briman, M., Star, A., Gru1ner, G., 2004. Nano Lett. 4, 253–256.Cao, C., Sim, S.J., 2007. Biosens. Bioelectron. 22, 1874–1880.Chen, R.J., Zhang, Y., Wang, D., Dai, H., 2001. J. Am. Chem. Soc. 123, 3838–3839.Chen, R.J., Bangsaruntip, S., Drouvalakis, K.A., Kam, N.W.S., Shim, M., Li, Y., Kim, W.,

Utz, P.J., Dai, H., 2003. PNAS 100, 4984–4989.Guo, X., Huang, L., O’Brien, S., Kim, P., Nuckolls, C., 2005. J. Am. Chem. Soc. 127,

15045–15047.Harris, L.J., Larson, S.B., Hasel, K.W., Day, J., Greenwod, A., McPherson, A., 1992. Nature

360, 369–372.James, W., 2001. Curr. Opin. Pharm. 1, 540–546.Ji, X., Song, X., Li, J., Bai, Y., Yang, W., Peng, X., 2007. J. Am. Chem. Soc. 129,

13939–13948.Kim, J.P., Lee, B.Y., Hong, S., Sim, S.J., 2008. Anal. Biochem. 381, 193–198.Lee, J.F., Stovall, G.M., Ellington, A.D., 2006a. Curr. Opin. Chem. Biol. 10, 282–289.Lee, M., Im, J., Lee, B.Y., Myung, S., Kang, J., Huang, L., Kwon, Y.K., Hong, S., 2006b.

Nat. Nanotechnol. 1, 66–71.Li, C., Curreli, M., Lin, H., Lei, B., Ishikawa, F.N., Datar, R., Cote, R.J., Thompson, M.E.,

Zhou, C., 2005. J. Am. Chem. Soc. 127, 12484–12485.Lilja, H.C.A., Dahlı̌en, U., Matikainen, M., Nilsson, O., Petterson, K., Lovgren, T., 1991.

Clin. Chem. 37, 1618–1625.Liu, A., 2008. Biosens. Bioelectron. 24, 167–177.Loeb, S., Catalona, W.J., 2007. Cancer Lett. 249, 30–39.Maehashi, K., Katsura, T., Kerman, K., Takamura, Y., Matsumoto, K., Tamiya, E., 2007.

Anal. Chem. 79, 782–787.Merkoci, A., Pumera, M., Llopis, X., Perez, B., Valle, M.D., Alegret, S., 2005. Trends

Anal. Chem. 24, 826–838.Ng, P.C., Osawa, Y., 1997. Steroids 62, 776–781.Patolsky, F., Lieber, C.M., 2005. Mater. Today 8, 20–28.Qi, P., Vermesh, O., Grecu, M., Javey, A., Wang, Q., Dai, H., Peng, S., Cho, K.J., 2003.

Nano Lett. 3, 347–351.Robertson, J., 2004. Mater. Today 7, 46–52.Savage, P., Waxman, J., 1996. Eur. J. Cancer 32A, 1097–1099.Schasfoort, R.B.M., Bergveld, P., 1990. Anal. Chim. Acta 238, 323–329.Schasfoort, R.B.M., Kooyman, R.P.H., Bergveld, P., Greve, J., 1990. Biosens. Bioelectron.

5, 103–124.So, H., Won, K., Kim, Y.H., Kim, B., Ryu, B.H., Na, P.S., Kim, H., Lee, J., 2005. J. Am. Chem.

Soc. 127, 11906–11907.Song, S., Wang, L., Li, J., Zhao, J., Fan, C., 2008. Trends Anal. Chem. 27, 108–117.Star, A., Gabriel, J.P., Bradley, K., Grulner, G., 2003. Nano Lett. 3, 459–463.Trojanowicz, M., 2006. Trends Anal. Chem. 25, 480–489.Zheng, G., Patolsky, F., Cui, Y., Wang, W.U., Lieber, C.M., 2005. Nat. Biotechnol. 23,

1294–1301.