ultrasensitive carbon nanotube-based biosensors using antibody-binding fragments
TRANSCRIPT
Analytical Biochemistry 381 (2008) 193–198
Contents lists available at ScienceDirect
Analytical Biochemistry
journal homepage: www.elsevier.com/ locate /yabio
Ultrasensitive carbon nanotube-based biosensors using antibody-binding
fragments
Jun Pyo Kim a, Byung Yang Lee b, Seunghun Hong b, Sang Jun Sim a,*
a Department of Chemical Engineering, Sungkyunkwan University, Changan-gu, Suwon 440-746, South Koreab School of Physics and Astronomy, Seoul National University, Gwanak-gu, Seoul 151-742, South Korea
a r t i c l e i n f o a b s t r a c t
Article history:
Received 10 March 2008
Available online 9 July 2008
We report a method to build ultrasensitive carbon nanotube-based biosensors using immune binding
reaction. Here carbon nanotube–field effect transistors (CNT–FETs) were functionalized with antibody-
binding fragments as a receptor, and the binding event of target immunoglobulin G (IgG) onto the frag-
ments was detected by monitoring the gating effect caused by the charges of the target IgG. Because the
biosensors were used in buffer solution, it was crucial to use small-size receptors so that the charged
target IgG could approach the CNT surface within the Debye length distance to give a large gating effect.
The results show that CNT–FET biosensors using whole antibody had very low sensitivity (detection limit
»1000 ng/ml), whereas those based on small Fab fragments could detect 1 pg/ml (»7 fM level). Moreover,
our Fab-modified CNT–FET could successfully block the nontarget proteins and could selectively detect
the target protein in an environment similar to that of human serum electrolyte. Significantly, this strat-
egy can be applied to general antibody-based detection schemes, and it should enable the production
of label-free ultrasensitive electronic biosensors to detect clinically important biomarkers for disease
diagnosis.
© 2008 Elsevier Inc. All rights reserved.
Keywords:
Carbon nanotube
Field effect transistor
Biosensor
Antibody-binding fragments [F(ab9)2, Fab]
Immune reaction
00
do
de
ci
in
st
pl
ie
[4
th
tw
co
to
tr
ti
ge
m
th
ge
im
DI
m
Most biological sensing techniques depend mainly on optical
tection principles. These methods are highly sensitive and spe-
fic, although they suffer from several disadvantages such as their
herent complexity and requirement for multiple reagents and
eps, signal amplification, a relatively large sample size, and com-
ex data analysis [1–3]. Recently, several researchers have stud-
d electronic biodetection methods to overcome these problems
–6]. One example is carbon nanotube (CNT)1-based biosensors
at take advantage of the exotic properties of CNTs [7–9]. Because
o-thirds of as-synthesized single-walled CNTs exhibited semi-
nducting properties, one can build CNT-based field effect transis-
rs (FETs) whose conductance changes sensitively by the charge
ansfer from molecules adsorbed onto the CNT surface [10–12].
In one of the biodetection schemes, the CNT–FET surface is func-
onalized with specific receptor molecules that bind to desired tar-
t biomolecules. When the target molecules bind to the receptor
olecules in solution, the charges of the target molecules affect
e conductance of the CNT–FETs. Thus, one can detect specific tar-
t molecules electrically in real time by monitoring the conduc-
03-2697/$ - see front matter © 2008 Elsevier Inc. All rights reserved.
i:10.1016/j.ab.2008.06.040
* Corresponding author. Fax: +82 31 290 7272.
E-mail address: [email protected] (S.J. Sim).1 Abbreviations used: CNT, carbon nanotube; FET, field effect transistor; IgG,
munoglobulin G; SWNT, single-walled nanotube; BSA, bovine serum albumin;
, deionized; OTS, octadecyltrichlorosilane; SA, streptavidin; SAM, self-assembled
onolayer; PR, photoresist; PBS, phosphate-buffered saline.
tance of CNT–FETs. In buffer solution, the double layer is formed
within the range of Debye length (»3 nm in 10 mM buffer solution)
[13,14] around the CNT–FETs, and the target and receptor biomol-
ecules should fit in the double layer to change the conductance of
CNT–FETs. However, the dimension of the antibodies (»10–12 nm)
used as receptors is usually much larger than the Debye length,
implying that the target molecules cannot approach the CNT–FETs
within the double layer. One strategy to overcome this problem is
using small receptors such as aptamers [15,16]. Although aptamers,
which are artificial nucleic acid ligands, are very small (»2 nm)
and show high selectivity, specificity, and affinity for amino acids,
drugs, proteins, and other small molecules [17,18], they have not
yet been developed for many diseases and they can also suffer from
pleomorphism. Here we suggest a novel method of reducing the
receptor size on CNT–FET biosensors using antibody-binding frag-
ments [F(ab9)2, Fab] for the antigen–antibody immune reaction.
Antibodies are immune system-related proteins called immuno-
globulins [19]. Each antibody monomer has a molecular weight of
approximately 150,000 Da (150 kDa) and is composed of two identi-
cal heavy polypeptide chains and two identical light chains joined
to form a Y-shaped molecule covalently bonded via interchain
disulfide (S–S) linkages between cysteine residues. Moreover,
these monomers are arranged in three discrete domains: two Fab
fragments at the tips of the “Y” and one Fc at its pole. The enzyme
papain can be used to cleave an immunoglobulin monomer into
two Fab fragments and an Fc fragment [20,21]. The enzyme pepsin
194 Ultrasensitive carbon nanotube-based biosensors / J.P. Kim et al. / Anal. Biochem. 381 (2008) 193–198
cleaves below the hinge region, so an F(ab9)2 fragment and an Fc
fragment are formed. Because it is the Fab fragments that bind to
antigens, the size of an antibody can be reduced substantially by
removing the Fc fragment.
In this article, we report the first successful demonstration of a
CNT–FET biosensor using Fab fragments as receptors. We function-
alized CNT–FETs with three types of anti-human immunoglobulin
G (IgG): (i) whole antibody (Fig. 1A), (ii) F(ab9)2 (Fig. 1B), and (iii)
Fab (Fig. 1C). Next, the response of each CNT–FET was monitored
in real time after the introduction of human IgG at various concen-
trations (100 fg/ml–1000 ng/ml).
Materials and methods
Materials
Purified single-walled nanotubes (SWNTs) were purchased from
Carbon Nanotechnologies (USA), and 1-pyrenebutanoic acid succ-
inimidyl ester was obtained from Molecular Probes (USA). Whole
antibody, F(ab9)2 fragments and Fab fragments of goat anti-human
IgG as receptors and human IgG as target biomolecules were pur-
chased from Jackson Immuno Research Laboratories (USA). Bovine
serum albumin (BSA), fibrinogen from human plasma, streptavidin
(SA), and other chemical reagents were supplied by Sigma. Deion-
ized (DI) water, obtained from a water purification system (Human
Corporation, Korea), was used for the preparation of the washing
and buffer solutions.
Fabrication of CNT–FET devices
Methyl-terminated octadecyltrichlorosilane (OTS) self-assem-
bled monolayer (SAM) patterns on silicon oxide wafer were gen-
erated by first patterning AZ 5214 photoresist (PR) via standard
photolithography, dipping the wafer in OTS solution (1:500 [v/v] in
hexane) for 3 min, and finally removing the PR patterns using ace-
tone. SWNT solution was prepared by dispersing purified SWNTs
in 1,2-dichlorobenzene with ultrasonication for 1 h (concentration
»0.1 mg/ml). Next, the patterned silicon oxide wafer was dipped
in the SWNT solution for 10 s, rinsed thoroughly with 1,2-dichloro-
benzene, and then dried with nitrogen gas [22]. This step allowed
SWNTs to be adsorbed selectively onto bare SiO2 regions on the
wafer, while methyl-terminated OTS SAMs blocked nonspecific
adsorption of CNTs. After the assembly of the SWNTs, electrodes
(30-nm Au layer on 10 nm Pd) were fabricated via a standard pho-
tolithography and lift-off process [23]. Finally, we also performed
an additional photolithography process to pattern PR (AZ 5214) to
cover up the electrodes and avoid leakage current from electrodes
in buffer solution.
Preparation of the CNT surface for sensing in solution
For the noncovalent functionalization of the CNT surface, CNT–
FET devices were incubated with 1 mM 1-pyrenebutanoic acid
succinimidyl ester in pure methanol for 1 h at room temperature
followed by rinsing with pure methanol to wash away any excess
reagent [24]. For the covalent immobilization of the receptor pro-
teins on the CNT surface, each CNT–FET device was exposed to
20 nM anti-human IgG [whole antibody, F(ab9)2 and Fab fragments]
in phosphate-buffered saline (PBS, pH 7.4) overnight at room tem-
perature, rinsed thoroughly in DI water for 6 h, and then dried with
nitrogen gas. To deactivate and block the excess reactive groups
remaining on the CNT surface, 100 mM ethanolamine was added
onto the channel region of the CNT–FET device and incubated for
30 min [16]. Then the CNT–FET device was rinsed with PBS.
Electrical measurement
The electrical properties of the CNT–FET devices during the
introduction of the target proteins were measured by a source
meter (Keithley 2400, USA) after the devices were immobilized by
a probe station that was able to connect with each source and drain
electrode. A source drain bias of 10 mV was maintained through-
out the measurements of the electrical signal, and the pulse width
was 1 s. The sample solutions containing the target proteins with
increasing concentrations were introduced sequentially into the
channel regions of the CNT–FET devices using sample volumes
of 10 ll. After the measurement of the electrical signal was com-
pleted for each sample, the devices were washed thoroughly with
DI water and dried with nitrogen gas. Then they were reloaded in
the source meter.
Results and discussion
Fig. 2 shows the CNT–FET devices fabricated via the linker-free
directed assembly method [22]. In this method, the CNT network
patterns were formed directly on a bare silicon oxide surface with-
out any linker molecules, and they were used as the channel for
FETs. These devices exhibited typical “p-type” characteristics, with
decreased source drain current with positive gate bias. The on/off
ratio was low (»3), just like other network transistors, because the
SWNT network was composed of both semiconducting and metal-
lic CNTs. However, our CNT–FET had significant advantages for sen-
sor applications. First, because our fabrication method did not use
any linker molecules, we could minimize possible signal contam-
ination by linker molecules. Furthermore, during this process, as
SWNTs were adsorbed onto bare silicon oxide surface, the surface
became nonpolar and blocked the formation of multiple SWNT
a b c
Whole antibody
F(ab’)2 Fab
Debyelength
Ethanolamine1-pyrenebutanoic acid succinimidyl ester
Fig. 1. Schematic diagram of CNT–FETs modified with three types of receptors on CNT surface: (A) immobilization of whole antibody; (B) immobilization of F(ab9)2; (C)
immobilization of Fab.
Ultrasensitive carbon nanotube-based biosensors / J.P. Kim et al. / Anal. Biochem. 381 (2008) 193–198 195
Fig. 2. Photographs of CNT–FET device: (A) optical image of a CNT–FET chip with patterned gold electrodes; (B) optical micrograph of a CNT–FET device with gold electrodes
passivated with PR; (C) atomic force microscopy image of channel of CNT–FET device.
layers. This self-limiting mechanism warranted the reproducibility
of our CNT–FET fabrication process [23].
Using the prepared devices, we performed the systematic
study regarding the effect of the receptor size on the sensitivity
of CNT–FET biosensors. First, CNT–FETs were functionalized with
three types of receptors [the whole antibody, F(ab9)2, and Fab from
anti-human IgG] after the formation of an SAM using 1-pyreneb-
utanoic acid succinimidyl ester on the CNT surface (Fig. 1). Next,
the electrical conductance change (G/G0) of the devices on the addi-
tion of human IgG solution with various concentrations (1 fg/ml
to 1000 ng/ml) was monitored. The experiments in each condition
were repeated three times.
In any biosensor for detection of target proteins, interaction
between nontarget and target proteins could be a source of noise
leading to aberrances in the final result. Therefore, to investigate
the specificity of the device between the surface materials, nontar-
get proteins such as BSA, fibrinogen, and SA were used on Fab-mod-
ified CNT–FETs. Fig. 3 shows the electrical responses of CNT–FETs
after the introduction of PBS solution, nontarget proteins (BSA,
fibrinogen, SA, each 10 mg/ml), and target protein (10 pg/ml IgG)
onto the IgG Fab-modified CNT–FET. When target protein was intro-
duced on the CNT channel, the electrical signal rapidly decreased.
In contrast, upon addition of PBS buffer as control or nontarget pro-
teins, the devices showed a slight increase in electrical conductance.
The addition of PBS liquid droplets created an additional electrical
path around the CNT network channels and could increase conduc-
tance. Moreover, after the introduction of nontarget proteins into
the CNT–FET channel, electrical conduction increased slightly as
in the case of PBS. These phenomena were also observed in cases
of whole antibody and F(ab)2-modified CNT–FETs. Therefore, these
results indicate that the nonspecific binding of nontarget proteins
was successfully suppressed in the CNT–FET device with the treat-
ment of 100 mM ethanolamine.
Fig. 4 shows the response of the CNT–FETs modified with
whole anti-human IgG when droplets of human IgG (10 ll) at var-
ious concentrations (10–1000 ng/ml) were placed on the devices.
On the addition of 10 to 100 ng/ml human IgG, the devices exhib-
ited increased electrical conductance as in the case of PBS. This
result means that the CNT–FETs modified with whole anti-human
IgG could not detect concentrations below 100 ng/ml IgG. How-
ever, the addition of a high-concentration solution (1000 ng/ml
human IgG in PBS) reduced the device conductance. These results
were quite consistent throughout three replicates. Human IgG had
positive charges because its isoelectric point (pI = 8.6 ± 04) [25,26]
was higher than the pH of the PBS solution (pH 7.4). Thus, human
IgG adsorbed onto the CNT surface should decrease the CNT–FET
conductance because it is equivalent to applying positive gate volt-
ages [27–30]. When the CNT–FETs were functionalized with whole
Sample injection
Time (s)0 50 100 150 200
Ele
ctri
cal s
ign
al G
/G0
0.6
0.8
1.0
1.2
BlankPBS bufferBSAFibrinogenStreptavidinIgG
Fig. 3. Electrical conductance of CNT–FETs after the introduction of PBS buffer solu-
tion, nontarget proteins (BSA, fibrinogen, streptavidin), and target protein onto the
IgG Fab-modified CNT–FET. “Blank” means electrical conductance where nothing
was introduced on the CNT–FET device. (For interpretation of the references to
color in the legend to this figure, the reader is referred to the Web version of this
article.)
Time (s)
0 50 100 150 200 250 300
Ele
ctri
cal s
ign
al G
/G0
0.8
1.0
1.2
1.4 Blankhuman IgG 10 ng/mlhuman IgG 100 ng/mlhuman IgG 1000 ng/ml
Sampleinjection
Fig. 4. Electrical conductance change of the CNT–FETs modified with whole antibod-
ies on the addition of human IgG solution with various concentrations (10–1000 ng/
ml). The arrow indicates the point of IgG injection. (For interpretation of the refer-
ences to color in the legend on this figure, the reader is referred to the Web version
of this article.)
196 Ultrasensitive carbon nanotube-based biosensors / J.P. Kim et al. / Anal. Biochem. 381 (2008) 193–198
anti-human IgG that was much larger than the Debye length, most
of the human IgG molecules could not approach the CNT surface
within the distance of the Debye length to give a gating effect. As
a result, the conductance of CNT–FET increased on the addition of
the solution due to the additional current path unless very high-
concentration human IgG solution (»100 ng/ml) was added. There-
fore, whole antibodies with the size of 10–12 nm are not suitable as
receptors for CNT–FET-based biosensors.
To resolve this problem, we introduced a method using Fab
fragments of the antibody. The immune reactions on CNT–FETs
modified with F(ab9)2 were performed at various concentration of
human IgG ranging from 1 to 5000 ng/ml. Fig. 5A shows the changes
in the conductance due to the immune reaction when F(ab9)2 was
immobilized on the CNT surface. The conductance decreased step-
wise with exposure to human IgG at concentrations increasing
from 10 to 1000 ng/ml. However, the analyte could not be deter-
mined at low concentration of 1 ng/ml IgG. At the concentration of
5000 ng/ml IgG, the conductance signal was not much higher than
that of the 1000 ng/ml IgG sample. This could be explained because
almost all binding sites of the anti-human IgG F(ab9)2 on CNT sur-
face were already occupied by human IgG molecules at the concen-
tration of 1000 ng/ml human IgG. Thus, as shown in the inset to
Fig. 5B, the linear dynamic range was from 10 to 1000 ng/ml and
the detectable minimum concentration (10 ng/ml) in the case of
the CNT–FETs modified with F(ab9)2 was ameliorated; it was lower
than that of the CNT–FETs modified with whole antibodies.
The hinge region between the Fc and two Fab domains in
antibodies is extremely flexible [31]. As a result, it is difficult
to predict the exact conformation of the IgG and its size when
the antibody is adsorbed on a surface. The approximate size of
human IgG was determined by computational simulation [32]
and small-angle X-ray scattering in solution [33]. The F(ab9)2 of
human IgG consisted of two Fabs, with the unavailable Fc region
(elliptical cylinder shape, height » 7.0 nm) [33] in the antibody
being cut off by pepsin; the vertical length of F(ab9)2 was approx-
imately 5 nm, which is half the size of the whole antibody. As the
size of the receptors is reduced, the positively charged proteins
can more easily approach the CNT surface within the distance of
the Debye length and affect the conductance of the CNT–FETs.
Thus, the sensitivity in the CNT–FETs modified with F(ab9)2 was
100 times greater than that in the CNT–FETs modified with whole
antibodies. Although the height of F(ab9)2 was only half that of
the antibody, the detectable minimum concentration was still
very high because the horizontal length of F(ab9)2, which con-
[Human IgG] (pg/ml)
0 200 400 600 800 1000 1200
Ele
ctri
cal s
ign
al (
G/G
0)
0.6
0.8
1.0
1.2
B
Log10 [Human IgG concentration] (pg/ml)
1 10 100
Ele
ctri
cal s
ign
al (
G/G
0)
0.5
0.6
0.7
0.8
0.9
1.0
A
Time (s)0 50 100 150 200 250 300
Ele
ctri
cal s
ign
al (
G/G
0)
0.6
0.8
1.0
1.2
1.4 Blankhuman IgG 100 fg/mlhuman IgG 1 pg/mlhuman IgG 10 pg/mlhuman IgG 100 pg/ml
Sampleinjection
Fig. 6. (A) Electrical conductance change of the CNT–FETs modified with Fab on
the addition of human IgG solution with various concentrations (100 fg/ml–100 pg/
ml). (B) The calibration curve of IgG detection on the CNT–FETs modified with Fab.
The error bars illustrate the relative standard deviation (RSD) of three replicates.
The inset shows the logarithmic linear range between the electrical signal and the
human IgG concentrations. The arrow indicates the point of IgG injection. (For
interpretation of the references to color in the legend on this figure, the reader is
referred to the Web version of this article.)
A
Time (s)0 50 100 150 200 250 300
Ele
ctri
cal s
ign
al (
G/G
0)
0.8
0.9
1.0
1.1
1.2
1.3
Blankhuman IgG 1 ng/mlhuman IgG 10 ng/mlhuman IgG 100 ng/mlhuman IgG 1000 ng/mlSample
injection
[Human IgG] (ng/ml)0 1000 2000 3000 4000 5000 6000
Ele
ctri
cal s
ign
al (
G/G
0)
0.8
0.9
1.0
1.1
1.2
B
Log10 [Human IgG concentration] (ng/ml)10 100 1000
Ele
ctri
cal s
ign
al (
G/G
0)
0.7
0.8
0.9
1.0
1.1
Fig. 5. (A) Electrical conductance change of the CNT–FETs modified with F(ab9)2
upon addition of human IgG solution at various concentrations (1–1000 ng/ml).
(B) The calibration curve of IgG detection on the CNT–FETs modified with F(ab9)2.
The error bars illustrate the relative standard deviation (RSD) of three replicates.
The inset shows the logarithmic linear range between the electrical signal and the
human IgG concentrations. The arrow indicates the point of IgG injection. (For
interpretation of the references to color in the legend on this figure, the reader is
referred to the Web version of this article.)
Ultrasensitive carbon nanotube-based biosensors / J.P. Kim et al. / Anal. Biochem. 381 (2008) 193–198 197
tained a disulfide bond forming a uniform angle (115–130°) [32]
between the Fabs, was very large (»10–11 nm). Therefore, the
size of the receptors should be smaller than F(ab9)2; the CNT–FET
biosensor modified with F(ab9)2 was not applicable in the diagno-
sis of disease.
To obtain better sensitivity for the CNT–FET-based detection,
CNT surface was functionalized with Fab, the smallest unit con-
taining an epitope in the antibody, and their conductance change
was monitored on the addition of human IgG solution with var-
ious concentrations in the range of 100 fg/ml–1000 pg/ml. As
shown in (Fig. 6)A, the conductance value decreased with increas-
ing IgG concentrations. However, the target analyte could not be
detected at concentrations lower than 100 fg/ml or higher than
1000 pg/ml. Therefore, the linear dynamic range of the Fab-mod-
ified CNT–FET biosensor was shown to be from 1 to 100 pg/ml
and the detectable minimum concentration was as low as 1 pg/
ml (7 fM level) of human IgG (Fig. 6B). The improvement in sensi-
tivity can be explained by the small size of Fab. Because the size
of Fab (height »3–5 nm) [33,34] was much smaller than that of
whole antibody or F(ab9)2, it is more likely that the binding event
between the Fab fragments and the target proteins occurred
inside an electrical double layer close to the CNT surface so that
the charges of human IgG protein had a large effect on the con-
ductance of CNT–FETs.
Finally, the CNT–FET device was examined in a mixed environ-
ment containing BSA, fibrinogen, and SA to investigate nonspe-
cific absorption and selectivity of the biosensor. Aliquots of these
undesired proteins (1 mg/ml) and PBS buffer (pH 7.4) were mixed
together, and then 10 ll of this mixture was introduced into the
channel regions to check the nonspecific binding on the CNT–FET
devices. To investigate the selectivity of the assay, the experiment
was performed by replacing the volume of PBS buffer with an
equal volume of the target protein (10 pg/ml) in the preparation
described above. As shown in Fig. 7, when the mixture solution
without target proteins was injected into the channel of the device,
the electrical signal increased slightly as with only PBS buffer. This
shows that nothing was detected in the mixture without target pro-
tein, and the nonspecific adsorption of these proteins was negli-
gible even at very high concentration. Interestingly, the electrical
signal was rapidly reduced in the solution with human IgG. This
result indicates that target protein can be selectively detected at
low levels in the presence of high concentrations of nontarget pro-
teins by the CNT–FET devices.
Conclusions
The Debye length is known to be one of the most important
factors determining the detection limits of CNT–FET biosensors
in solution. Here we devised CNT–FETs that were modified with
Fab fragments as receptors so that the immune-binding reaction
occurred within the Debye length from the CNT surface. In this
way, we were able to lower the detection limit to a protein concen-
tration of 1 pg/ml (»7 fM level) with no labeling of target proteins.
Moreover, our Fab-modified CNT–FET could successfully suppress
the nontarget proteins and could selectively detect the target pro-
tein in a mixed environment containing BSA, fibrinogen, SA at high
concentration. This strategy allowed us to build a CNT–FET biosen-
sor system based on the well-established immune reaction of anti-
gen–antibody. The use of small receptors such as Fab fragments
has several advantages in biosensors based on CNT–FET. First, all of
the antibodies developed for disease diagnosis can be used to build
CNT–FET biosensors. In addition, the reduced size of receptors sig-
nificantly improves the detection limit of CNT–FET biosensors by
binding target protein close to the CNT surface. Thus, the strategy
of using Fab fragments in CNT–FET biosensors can be a major break-
through that enables various important applications, for example,
in proteomics and medical diagnostics.
Acknowledgments
This work was supported by Grant 10017190 from the Next Gen-
eration New Technology Development Program of the Ministry of
Knowledge Economy (MKE).
References
[1] P.B. Luppa, L.J. Sokoll, D.W. Chan, Immunosensors: principles and applications to clinical chemistry, Clin. Chim. Acta 314 (2001) 1–26.
[2] M. Wells, Advances in optical detection strategies for reporter signal measure-ments, Curr. Opin. Biotechnol. 17 (2006) 28–33.
[3] M. Seydack, Nanoparticle labels in immunosensing using optical detection methods, Biosens. Bioelectron. 20 (2005) 2454–2469.
[4] Y. Wang, Z. Tang, N.A. Kotov, Bioapplication of nanosemiconductors, Mater. Today 8 (2005) 20–31.
[5] G. Gruner, Carbon nanotube transistors for biosensing applications, Anal. Bio-anal. Chem. 384 (2006) 322–335.
[6] A. Star, J.C.P. Gabriel, K. Bradley, G. Gru1ner, Electronic detection of specific protein binding using nanotube FET devices, Nano Lett. 3 (2003) 459–463.
[7] A. Merkoci, M. Pumera, X. Llopis, B. Perez, M.D. Valle, S. Alegret, New materi-als for electrochemical sensing: VI. Carbon nanotubes, Trends Anal. Chem. 24 (2005) 826–838.
[8] P. Avouris, J. Chen, Nanotube electronics and optoelectronics, Mater. Today 9 (2006) 46–54.
[9] H. Dai, Carbon nanotubes: synthesis, integration, and properties, Acc. Chem. Res. 35 (2002) 1035–1044.
[10] J. Robertson, Realistic applications of CNTs, Mater. Today 7 (2004) 46–52.[11] M. Trojanowicz, Analytical applications of carbon nanotubes: a review, Trends
Anal. Chem. 25 (2006) 480–489.[12] P. Qi, O. Vermesh, M. Grecu, A. Javey, Q. Wang, H. Dai, S. Peng, K.J. Cho, Toward
large arrays of multiplex functionalized carbon nanotube sensors for highly sensitive and selective molecular detection, Nano Lett. 3 (2003) 347–351.
[13] R.B.M. Schasfoort, P. Bergveld, Possibilities and limitations of direct detection of protein charges by means of an immunological field-effect transistor, Anal. Chim. Acta 238 (1990) 323–329.
[14] R.B.M. Schasfoort, R.P.H. Kooyman, P. Bergveld, J. Greve, A new approach to immunoFET operation, Biosens. Bioelectron. 5 (1990) 103–124.
[15] H. So, K. Won, Y.H. Kim, B. Kim, B.H. Ryu, P.S. Na, H. Kim, J. Lee, Single-walled carbon nanotube biosensors using aptamers as molecular recognition ele-ments, J. Am. Chem. Soc. 127 (2005) 11906–11907.
[16] K. Maehashi, T. Katsura, K. Kerman, Y. Takamura, K. Matsumoto, E. Tamiya, Lable-free protein biosensor based on aptamer-modified carbon nanotube field-effect transistors, Anal. Chem. 79 (2007) 782–787.
[17] J.F. Lee, G.M. Stovall, A.D. Ellington, Aptamer therapeutics advance, Curr. Opin. Chem. Biol. 10 (2006) 282–289.
[18] W. James, Nucleic acid and polypeptide aptamers: a powerful approach to ligand discovery, Curr. Opin. Pharm. 1 (2001) 540–546.
[19] R.G. Hamilton, D. Abmli, The Human IgG Subclasses, CalBioChem–NovaBio-Chem, Darmstadt, Germany, 2001.
Time (sec)
0 50 100 150 200 250
Ele
ctri
cal s
ign
al G
/G0
0.6
0.8
1.0
1.2
BlankMixtureIgG + Mixture
Sampleinjection
Fig. 7. The selectivity of target proteins (human IgG) in mixture solution (BSA, fibrin-
ogen, streptavidin) on CNT–FETs modified with Fab.
198 Ultrasensitive carbon nanotube-based biosensors / J.P. Kim et al. / Anal. Biochem. 381 (2008) 193–198
[20] P.C. Ng, Y. Osawa, Preparation and characterization of the Fab and F(ab9)2 frag-ments of an aromatase activity-suppressing monoclonal antibody, Steroids 62 (1997) 776–781.
[21] L.J. Harris, S.B. Larson, K.W. Hasel, J. Day, A. Greenwod, A. McPherson, The three-dimensional structure of an intact monoclonal antibody for canine lym-phoma, Nature 360 (1992) 369–372.
[22] S.G. Rao, L. Huang, W. Setyawan, S. Hong, Large-scale assembly of carbon nano-tubes, Nature 425 (2003) 36–37.
[23] M. Lee, J. Im, B.Y. Lee, S. Myung, J. Kang, L. Huang, Y.K. Kwon, S. Hong, Linker-free directed assembly of high-performance integrated devices based on nano-tubes and nanowires, Nat. Nanotechnol. 1 (2006) 66–71.
[24] R.J. Chen, Y. Zhang, D. Wang, H. Dai, Noncovalent sidewall functionalization of single-walled carbon nanotubes for protein immobilization, J. Am. Chem. Soc. 123 (2001) 3838–3839.
[25] P.A.J. Rosa, A.M. Azevedo, M.R. Aires-Barros, Application of central composite design to the optimisation of aqueous two-phase extraction of human antibod-ies, J. Chromatogr. 1141 (2007) 50–60.
[26] R.P. Tracy, R.M. Currie, R.A. Kyle, D.S. Young, Two-dimensional gel electropho-resis of serum specimens from patients with monoclonal gammopathies, Clin. Chem. 28 (1982) 900–907.
[27] R.J. Chen, S. Bangsaruntip, K.A. Drouvalakis, N.W.S. Kam, M. Shim, Y. Li, W. Kim, P.J. Utz, H. Dai, Noncovalent functionalization of carbon nanotubes for highly specific electronic biosensors, Proc. Natl. Acad. Sci. USA 100 (2003) 4984–4989.
[28] C. Li, M. Curreli, H. Lin, B. Lei, F.N. Ishikawa, R. Datar, R.J. Cote, M.E. Thomp-son, C. Zhou, Complementary detection of prostate-specific antigen using In2O3 nanowires and carbon nanotubes, J. Am. Chem. Soc. 127 (2005) 12484–12485.
[29] G. Zheng, F. Patolsky, Y. Cui, W.U. Wang, C.M. Lieber, Multiplexed electrical detection of cancer markers with nanowire sensor arrays, Nat. Biotechnol. 23 (2005) 1294–1301.
[30] F. Patolsky, C.M. Lieber, Nanowire nanosensors, Mater. Today 8 (2005) 20–28.[31] C.J. Roberts, P.M. Williams, J. Davies, A.C. Dawkes, J. Sefton, J.C. Edwards, A.G.
Haymes, C. Bestwick, M.C. Davies, S.J.B. Tendler, Real-space differentiation of IgG and IgM antibodies deposited on microtiter wells by scanning force micros-copy, Langmuir 11 (1995) 1822–1826.
[32] E.O. Saphire, R.L. Stanfield, M.D.M. Crispin, W.H.I. Parren, P.M. Rudd, R.A. Dwek, D.R. Burton, I.A. Wilson, Contrasting IgG structures reveal extreme asymmetry and flexibility, J. Mol. Biol. 319 (2002) 9–18.
[33] I. Pilz, E. Schwarz, W. Palm, Small-angle X-ray studies of the Fab and Fc frag-ments from the human immunoglobulin molecule K01, Eur. J. Biochern. 71 (1976) 239–247.
[34] A.F. Labrijn, P. Poignard, A. Raja, M.B. Zwick, K. Delgado, M. Franti, J. Binley, V. Vivona, C. Grundner, C. Huang, M. Venturi, C.J. Petropoulos, T. Wrin, D.S. Dim-itrov, J. Robinson, P.D. Kwong, R.T. Wyatt, J. Sodroski, D.R. Burton, Access of antibody molecules to the conserved coreceptor binding site on glycoprotein gp120 is sterically restricted on primary human immunodeficiency virus type 1, J. Virol. 77 (2003) 10557–10565.