Functionalized Carbon Nanotubes for
Detecting Viral Proteins
Yian-Biao Zhang,†,§Mandakini Kanungo,‡,§Alexander J. Ho,‡,|Paul Freimuth,†
Daniel van der Lelie,*,†Michelle Chen,XSamuel M. Khamis,⊥Sujit S. Datta,⊥
A. T. Charlie Johnson,*,⊥James A. Misewich,*,‡and Stanislaus S. Wong*,‡,#
Biology Department, BrookhaVen National Laboratory, Building 463, Upton,
New York 11973, Condensed Matter Physics and Materials Science Department,
BrookhaVen National Laboratory, Building 480, Upton, New York 11973, Biomedical
Engineering Department, State UniVersity of New York at Stony Brook, Stony Brook,
New York 11794, Department of Materials Science and Engineering, UniVersity of
PennsylVania, Philadelphia, PennsylVania 19104, Department of Physics and
Astronomy, UniVersity of PennsylVania, Philadelphia, PennsylVania 19104, and
Department of Chemistry, State UniVersity of New York at Stony Brook, Stony Brook,
New York 11794
Received July 1, 2007; Revised Manuscript Received August 26, 2007
We investigated the biocompatibility, specificity, and activity of a ligand−receptor-protein system covalently bound to oxidized single-walled
carbon nanotubes (SWNTs) as a model proof-of-concept for employing such SWNTs as biosensors. SWNTs were functionalized under ambient
conditions with either the Knob protein domain from adenovirus serotype 12 (Ad 12 Knob) or its human cellular receptor, the CAR protein,
via diimide-activated amidation. We confirmed the biological activity of Knob protein immobilized on the nanotube surfaces by using its
labeled conjugate antibody and evaluated the activity and specificity of bound CAR on SWNTs, first, in the presence of fluorescently labeled
Knob, which interacts specifically with CAR, and second, with a negative control protein, YieF, which is not recognized by biologically active
CAR proteins. In addition, current−gate voltage (I−Vg) measurements on a dozen nanotube devices explored the effect of protein binding on
the intrinsic electronic properties of the SWNTs, and also demonstrated the devices’ high sensitivity in detecting protein activity. All data
showed that both Knob and CAR immobilized on SWNT surfaces fully retained their biological activities, suggesting that SWNT−CAR complexes
can serve as biosensors for detecting environmental adenoviruses.
In recent years, there has been growing interest in forming
viable strategies for the controlled functionalization of single-
walled nanotubes (SWNTs) with biological systems.1-3Such
bio-nano integrated systems, combining the conducting and
semiconducting properties of carbon nanotubes with the
recognitive and catalytic properties of biomaterials, offer
particular promise for developing novel biosensor systems.
Specific recognition of target molecules is the essential
feature for biological sensing. Accordingly, we have been
interested in addressing the following key issues:
1. Do ligand-receptor proteins, bound onto SWNTs, retain
their active configuration and conformation so as to remain
amenable to biological interactions? We answered this
question by demonstrating that functionally active Knob and
CAR can be bound onto SWNT surfaces through an amide
linkage generated via a diimide reagent.
2. Can this system subsequently be utilized for biological
sensing? Our electrical measurements herein demonstrated
the applicability of single biofunctionalized carbon nanotubes
as field effect transistor (FET) biosensors wherein the
biological moiety maintains its activity, proper binding
conformation, and biospecificity.
Several different biomolecular systems have been previ-
ously affixed to the external surfaces of SWNTs with the
goal of creating functional devices. For example, enzyme-
coated SWNTs were used as sensors that either modulate
their optical properties upon adsorption or alter their
conductance upon variations in pH.4,5Viruses were employed
to assemble SWNTs and other materials into organized
networks.6Proteins such as ferritin, avidin, bovine serum
albumin (BSA), and streptavidin,7-9as well as metallopro-
*To whom correspondence should be addressed. E-mail: email@example.com
(D.v.d.L.); firstname.lastname@example.org (A.T.C.J.); email@example.com
(J.A.M.); firstname.lastname@example.org (S.S.W.).
†Biology Department, Brookhaven National Laboratory.
‡Condensed Matter Physics and Materials Science Department, Brookhaven
§These authors contributed equally to this work.
|Biomedical Engineering Department, State University of New York at
XDepartment of Materials Science and Engineering, University of
⊥Department of Physics and Astronomy, University of Pennsylvania.
#Department of Chemistry, State University of New York at Stony
Vol. 7, No. 10
10.1021/nl071572l CCC: $37.00
Published on Web 09/26/2007
© 2007 American Chemical Society
teins and enzymes10have been noncovalently bound onto
SWNT surfaces so as to generate highly specific electronic
biosensors. Other groups have coated peptides11with selec-
tive affinity for SWNTs onto their surfaces while different
laboratories have bound proteins12,13such as either ferritin
or BSA onto oxidized SWNTs via an amide linkage in
aqueous solution. In these latter studies, the biological activ-
ities of the attached moieties were confirmed but the electrical
activity of the functionalized nanotubes was not measured.
Although this is a relatively unexplored area of research,
what is abundantly evident is that the covalent functional-
ization of biologically active ligand-receptor proteins onto
single SWNTs and SWNT bundles clearly affords a viable
strategy toward developing specific, quantitative biosensors.
Precedence for such a strategy lies in a few unrelated studies.
For instance, complementary detection of prostate-specific
antigen was demonstrated by using In2O3 nanowire and
SWNT devices.14SWNT-FET-based biosensors composed
of either DNA aptamers15or single-stranded DNA16as
molecular recognition elements were also reported, although
most of this DNA work relied on noncovalent interactions
with the SWNTs.
In the present study, we demonstrate a simple, fast-
response, highly sensitive, real-time biosensor composed of
a ligand-receptor protein complex covalently attached by a
diimide linker to oxidized SWNTs via a mild, ambient,
straightforward, and economical protocol. That is, we not
only retained the intrinsic biological activity and specificity
of the attached complex but also conserved the highly
favorable electronic properties of SWNTs in these biofunc-
tionalized single-tube devices. The proteins we used were
the adenovirus protein, Ad12 Knob, and its complementary
human “Coxsackie virus and adenovirus receptor”, CAR.
Adenoviruses are one of many subclasses of viruses that
cause infections such as the common cold and mild ailments
of the upper respiratory and gastrointestinal tracts. Unlike
viruses such as HIV, Ebola, and poliovirus, adenoviruses do
not use either envelope proteins or capsid domains to infect
cells. Rather, infection is initiated by the formation of a high
affinity complex between the Knob trimer and its comple-
mentary adenovirus CAR receptor present in human cells.
Upon binding CAR, the Knob-coated virus replicates within
the cell nucleus, triggering infections.17,18Currently, aden-
oviruses are the leading candidates as vectors for gene
therapy.19In our work, we used 6 mg/mL of purified Knob
and 2.5 mg/mL of CAR protein, as verified by using a BCA
Raw HiPco (high-pressure carbon monoxide decomposi-
tion process) SWNT bundles as well as individual SWNTs
(prepared on surfaces by in situ catalytic chemical vapor
deposition) were purified and air oxidized by using a modifi-
cation of the gasification-dissolution method described
earlier by Chiang et al.20This process generates surface
functionalities on the nanotubes, particularly carboxylic acids
at their ends and sidewalls. Air-oxidized SWNTs were then
suspended in a 50 mM phosphate buffer (pH 8) solution at
a concentration of 1 mg/mL. Proteins were attached to the
processed SWNTs via a two-step process of carbodiimide
(EDAC)-mediated activation previously described.14We
confirmed that the proteins had, indeed, bound to the SWNTs
by atomic force microscopy (AFM) height analysis. Further
experimental details, including protein labeling, are given
in the Supporting Information (Figures S1-S3).
We obtained AFM measurements before and after protein
attachment on four samples. Representative AFM height
images and statistical analysis of selected cross sections
(shown in Figure 1) conclusively showed that this function-
alization procedure attaches protein complexes (CAR +
Knob) along the length of the SWNT (on average, ∼1 µm
for individual tubes). The observed protein density in Figure
1 was approximately 1 per 200 nm, but in other samples, it
approached the limit of the AFM resolution (∼1 per 20 nm).
Figure 1. (a) AFM height image of single-walled carbon nanotubes
on a Si/SiO2substrate after oxidation. The z-axis color scale is
10 nm. (b) AFM height image of the same nanotube after the
attachment of the CAR protein and subsequent exposure to Knob
protein. Arrows indicate seven main sites along the nanotube’s
length where the CAR + Knob complex is bound to the nanotube’s
surface. The vertical z-axis color scale is 10 nm. The images were
low-pass filtered for clarity.
Nano Lett., Vol. 7, No. 10, 2007 3087
Height analysis (Figure 2) demonstrated that the diameter
of an individual SWNT bundle after oxidation was 1.9 (
0.2 nm and that the additional height observed, associated
with the attached protein complex (CAR + Knob), was 2.5
( 0.2 nm. AFM measurements on five other samples
exposed to CAR alone yielded a height increase of 0.5 nm
(data not shown) so that by extension, the intrinsic height
increase due to Knob itself was approximately 2 nm. These
values are somewhat smaller than the accepted molecular
sizes of these proteins, consistent with our expectation that
(i) pressure from the AFM tip may have distorted the protein
and that (ii) proteins attached to the mostly hydrophobic
nanotube surface in air can be slightly different from their
intrinsic morphology in aqueous solution. As demonstrated
later and further corroborated in the Supporting Informa-
tion (data on SWNT bundles, Figure S5), CAR proteins
effectively attached to individual SWNTs as a single layer
and retained their critical molecular recognition function-
After confirming the formation of our protein-SWNT
constructs, we explored their interaction with both labeled
complementary (Ad 12 Knob) and noncomplementary (YieF)
proteins, thereby enabling us to assess the activity and
specificity of the bound, attached proteins. Both Ad 12 Knob
and YieF were labeled by using Alexa Fluor (Molecular
Probes). All optical and fluorescence images of the labeled
proteins were recorded by using a Zeiss Axiovert 200
The biological activity of bound Ad 12 Knob was
investigated by targeting rhodamine-labeled anti-Knob an-
tibodies to the oxidized SWNT-Knob constructs. These
antibodies were purified by using an affinity column of
immobilized Ad 12 Knob, and hence, we targeted only those
Knob proteins folded in specifically active conformations.
Nonspecific binding of rhodamine-labeled anti-Ad 12 Knob
protein attached to carbon nanotubes was prevented by
blocking this reaction with 4% milk, which contains a
number of unrelated, nonspecific proteins in high concentra-
tions. Figure 3a shows the optical image (left) and the
corresponding fluorescence image (right) of fluorescently
labeled anti-Knob antibodies targeting Ad 12 Knob protein
bound to the carbon nanotubes. The fluorescence of the
functionalized carbon nanotubes confirms that Ad 12 Knob
bound to SWNTs indeed retains its biologically active
conformation. As control experiments, labeled anti-Knob
antibodies lacking protein were targeted to SWNTs either
in the presence (blocked) or absence of milk (not blocked).
In the latter case, we observed that the sample fluoresced,
suggesting that the labeled antibodies bound to the SWNTs.
Conversely, blocked SWNTs exhibited little or no fluores-
cence (Supporting Information Figure S4), demonstrating the
efficacy of milk as a blocking agent. Hence, it is evident
that labeled anti-Knob antibodies could specifically target
bound Knob proteins on carbon nanotube surfaces.
By analogy, to investigate the biological activity and
specificity of bound CAR, we used fluorescently labeled
Knob, which shows high specific binding to CAR. In a
separate control experiment, we attached YieF, an unrelated
22.4 kDa protein isolated from E. coli bacteria, to the
SWNTs; YieF is nonspecific for CAR. Hence, the presence
of bound CAR proteins in their biologically active conforma-
tion will show specificity to Knob proteins. SWNT-CAR
constructs were blocked by 4% milk to prevent nonspecific
binding of the labeled proteins to the nanotubes. Figure 3b
shows an optical image (left) and the corresponding fluo-
rescent image (right) of SWNT-CAR constructs targeted
by fluorescently labeled Ad 12 Knob. The sample fluoresces,
indicating that Knob is bound to the CAR proteins. On the
other hand, after replacing the labeled Ad 12 Knob with
labeled YieF, the samples did not fluoresce (Figure 3c). This
observation afforded the following evidence: (1) CAR is
bound to the carbon nanotubes, (2) bound CAR is biologi-
cally active, and (3) CAR-functionalized nanotubes will
specifically bind to Ad 12 Knob. Thus, this construct
provides us with the basis for a biological sensor to detect
the presence of the Ad 12 Knob viral protein.
We measured current-gate voltage (I-Vg) data on a dozen
nanotube devices to explore the effect of the attachment
process and of protein binding on the SWNTs’ electronic
properties. Typical data are displayed in Figure 4. Nanotube
FET devices were of high quality; they consisted of
individual SWNTs with ON/OFF ratios exceeding 1000 and
possessed on-state resistance values of 100-500 kΩ. Find-
ings discussed below were reproduced in all the devices,
although there was some scatter, as noted, in individual
All devices showed a hysteretic I-Vg response, as is
typical of nanotube FETs on untreated silica substrates. This
response results from charge injection from the nanotube
into nearby regions due to the substantial electric field
(∼10 V/nm) existing at the SWNT surface associated with
a large gate voltage (Vg).21,22The electric field of this injected
Figure 2. Histogram of SWNT diameters based on 50 equally
spaced height sections from each AFM image with Gaussian fits.
Before applying any proteins, the data (red, striped bars and red
fit) show a SWNT diameter of 1.9 ( 0.2 nm. After applying CAR
+ Knob proteins, the distribution (gray bars and blue fits) exhibits
two peaks, corresponding to heights of 2.0 ( 0.5 nm and 4.4 ( 4
nm, ascribed to regions of the nanotube without and with attached
protein complexes, respectively.
Nano Lett., Vol. 7, No. 10, 2007
charge and of other charge traps near the SWNT is partially
screened when the FET is in its ON state, while almost no
screening occurs when the FET is in the OFF state. Hence,
in the following discussion, we assume that the leftmost
(ON-to-OFF) transition of the I-Vg characteristic is more
reproducible than the OFF-to-ON transition in the presence
of unavoidable charge switching and is, therefore, more
amenable to physical interpretation.
The oxidation process typically either increased the ON
state current of the device (by 10-25%) or left it unchanged
(Figure 4). We concluded that mild oxidation created a low
density of defect sites that did not degrade electron transport
in the device; we attribute the small increase in the ON state
current to contact annealing. Oxidation also generated a
reproducible increase of 0.5-3 V in the ON-OFF threshold
voltage, consistent with the notion that defect sites created
by oxidation are functionalized with oxygenated moieties
(such as predominantly carboxyl groups) that become depro-
tonated in the presence of adsorbed water. This change leaves
the groups negatively charged, so that a more positive value
of Vgis needed to turn the FET OFF. Assuming a typical
backgate capacitance of 25 aF/µm for this geometry, this
shift in Vgcorresponds to an increase in the carrier density
of 80-400 holes/µm.
Figure 3. (a) Optical (left) and corresponding fluorescence (right) images of rhodamine-labeled anti-Knob antibodies targeting Ad12
Knob functionalized air-oxidized SWNTs. (b) and (c) show, respectively, the optical and corresponding fluorescence images of fluorescently
labeled Ad 12 Knob and YieF targeting CAR-functionalized air-oxidized SWNTs. The sample in (b) fluoresces because the Ad 12 Knob
is bound to CAR. On the other hand, there is no observable fluorescence in the sample in (c), where labeled YieF targets functionalized
CAR air-oxidized SWNTs. All samples were blocked with milk to prevent the nonspecific binding of proteins onto the SWNT surfaces.
The scale bar is 2.5 µm.
Nano Lett., Vol. 7, No. 10, 20073089
Subsequently, incubating the device in EDAC/NHS solu-
tion engendered a negative shift of the threshold voltage to
its original value or to an even more negative voltage, in
agreement with the expectation that the proton had been
replaced by a stable active ester.12CAR protein attachment
led to a 1-2 V decrease in the ON-OFF threshold voltage
and a corresponding 20-40% decrease in the ON-state
current. These observations are consistent with the FET
experiencing a positive charge and enhanced carrier scattering
due to the presence of the protein, as other groups have
proposed.7,23,24Finally, exposing the CAR-SWNT hybrid
to the complementary Knob protein further suppresses the
ON state current; the molecular recognition event was thus
fully detectable in this system (Figure 4). To quantify the
extent of protein binding, we can reasonably assume that
the nanotube is 1 µm long, with a corresponding protein
density of 1 per 20 nm. Hence, on average, 50 proteins coat
the nanotube and the corresponding decrease in current we
observed is approximately 1 nA/protein, a sizable value
enabling the detection of single protein molecules. Because
the applied voltage is 100 mV, the resistance of CAR and
of Knob bound to CAR is approximately 667 kΩ and 1 MΩ,
respectively, implying a resistance of about 5 kΩ/protein.
Using the Landauer formulation in the incoherent transport
regime, we thereby obtain a reflection coefficient of ap-
proximately 40% per protein, implying that the protein
complex is closely bound to the nanotube surface.
In a separate control experiment (Figure 5), CAR-
functionalized devices showed no evident change in I-Vg
response, as expected, after exposure to (noncomplementary)
YieF, implying that the in vivo chemical specificity of the
CAR protein is retained even when it is immobilized on the
SWNT surface. In another experiment, we noted that the
electrical profile of SWNTs, which had been noncovalently
functionalized with CAR proteins, reverted to its original
signal upon extended washing with phosphate buffer and
water; these data highlighted the importance of covalent
protein binding in our experiments and implied that weakly
bound, physically absorbed proteins were lost upon washing
of the SWNTs (data not shown).
The present study provides proof-of-concept for develop-
ing a simple, efficient, sensitive, fast-response, and real-time
miniaturized nanotube FET biosensor for detecting the Ad
12 Knob virus using CAR-Knob specificity. Moreover, this
methodology can be extended to uncover the presence of
serotype12 and all other possible CAR-binding adenoviruses
(about 30 serotypes, including Ad2 and Ad5), as well as
subgroup B Coxsackie viruses. This is the first evidence of
straightforward, ambient covalent immobilization of a viral
ligand-receptor-protein system onto individual SWNTs and
SWNT bundles and of our subsequent confirmation of the
bound proteins’ retention of biological activity and specific-
ity, as revealed by systematic electrical measurements. Our
future goal will be to develop a single-molecule biosensor
based on the conductivity change of a single SWNT by
adding a discrete CAR domain to the nanotube.
acknowledge support of this work through the U.S. Depart-
ment of Energy Office of Basic Energy Sciences under
contract DE-AC02-98CH10886. S.S.W. also thanks the
National Science Foundation (CAREER DMR-0348239) as
well as the Alfred P. Sloan Foundation for supplies and
financial support. Work conducted in the laboratory of
A.T.C.J. was supported by the JSTO DTRA as well as the
Army Research Office grant no. W911NF-06-1-0462; re-
search at UPenn was also partially supported by the Nano/
Bio Interface Center through the National Science Foundation
under contract NSEC DMR-0425780. We thank A. Wood-
head for helpful comments.
Y.Z., D.v.d.L., J.M., and S.S.W.
Figure 4. Measured source-drain current as a function of gate
voltage for a SWNT FET demonstrates that covalently bound CAR
protein retains its molecular recognition functionality. Data are
shown for as-grown SWNT (black), SWNTs after oxidation
(purple), after exposure to EDAC/NHS (green), upon CAR attach-
ment (blue), and after exposure to the complementary Knob protein
(red). The data indicate that Knob specifically binds to the CAR,
leading to a significant decrease (∼33%) in the ON-state current
of the FET. The bias voltage is 100 mV for all measurements.
Figure 5. Measured source-drain current vs gate voltage for a
SWNT sensor functionalized with CAR protein (blue data) shows
no response when exposed to the nonspecific YieF protein (red).
Nano Lett., Vol. 7, No. 10, 2007
Supporting Information Available: Expression and Download full-text
purification of AD 12 Knob; CAR protein purification;
YieF purification; fluorescent labeling of proteins; pre-
paration of SWNT-protein hybrids; microscopy char-
acterization of SWNTs and of SWNT-protein hybrids;
attachment of labeled proteins onto SWNT-protein hy-
brids; growth, fabrication, and electrical measurements of
SWNTs; additional AFM height measurements on SWNT
(1) Katz, E.; Willner, I. ChemPhysChem 2004, 5, 1084-1104.
(2) Bianco, A.; Prato, M. AdV. Mater. 2003, 15, 1765-1768.
(3) Wong, S. S.; Joselevich, E.; Woolley, A. T.; Cheung, C. L.; Lieber,
C. M. Nature 1998, 394, 52-55.
(4) Baron, P. W.; Baik, S.; Heller, D. A.; Strano, M. S. Nat. Mater.
2005, 4, 86-92.
(5) Besteman, K.; Lee, J.-O.; Wiertz, F. G. M.; Heering, H. A.; Dekker,
C. Nano Lett. 2003, 3, 727-730.
(6) Portney, N. G.; Singh, K.; Chaudhary, S.; Destito, G.; Schneemann,
A.; Manchester, M.; Ozkan, M. Langmuir 2005, 21, 2098-2103.
(7) Chen, R. J.; Bangsarnutip, S.; Drouvalakis, K. A.; Kam, N. W. S.;
Shim, M.; Kim, W.; Utz, P. J.; Dai, H. Proc. Natl. Acad. Sci. U.S.A.
2003, 100, 4984-4989.
(8) Shim, M.; Kam, N. W. S.; Chen, R. J.; Li, Y.; Dai, H. Nano Lett.
2002, 2, 285-288.
(9) Chen, R. J.; Zhang, Y.; Wang, D.; Dai, H. J. Am. Chem. Soc. 2001,
(10) Azamian, B. R.; Davis, J. J.; Coleman, K. S.; Bagshaw, C. B.; Green,
M. L. H. J. Am. Chem. Soc. 2002, 124, 12664-12665.
(11) Wang, S.; Humphreys, E. S.; Chung, S.-Y.; Delduco, D. F.; Lustig,
S. R.; Wang, H.; Parker, K. N.; Rizzo, N. W.; Subramoney, S.;
Chiang, Y.-M.; Jagota, A. Nat. Mater. 2003, 2, 196-200.
(12) Jiang, K.; Schadler, L. S.; Seigel, R. W.; Zhang, X.; Zhang, H.;
Terrones, M. J. Mater. Chem. 2004, 14, 37-39.
(13) Huang, W.; Taylor, S.; Fu, K.; Lin, Y.; Zhang, D.; Hanks, T. W.;
Rao, A. M.; Sun, Y.-P. Nano Lett. 2002, 2, 311-314.
(14) Li, C.; Curreli, M.; Lin, H.; Lei, B.; Ishikawa, F. N.; Datar, R.; Cote,
R. J.; Thomson, M. E.; Zhou, C. J. Am. Chem. Soc. 2005, 127,
(15) So, H.-M.; Won, K.; Kim, Y. H.; Ryu, B. H.; Na, P. S.; Kim, H.;
Lee, J.-O. J. Am. Chem. Soc. 2005, 127, 11906-11907.
(16) Staii, C.; Johnson, A. T., Jr.; Chen, M.; Gelperin, A. Nano Lett. 2005,
(17) Freimuth, P.; Springer, K.; Berard, C.; Hainfeld, J.; Bewley, M.;
Flanagan, J. M. J. Virol. 1999, 73, 1392-1398.
(18) Bewley, M. C.; Springer, K.; Zhang, Y.-B.; Freimuth, P.; Flanagan,
J. M. Science 1999, 286, 1579-1583.
(19) Nabel, G. J. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 324-326.
(20) Chiang, I. W.; Brinson, B. E.; Huang, A. Y.; Willis, P. A.;
Bronikowski, M. J.; Margrave, J. L.; Smalley, R. E.; Hauge, R. H.
J. Phys. Chem. B 2001, 105, 8297-8301.
(21) Radosavljevic, M.; Freitag, M.; Thadani, K. V.; Johnson, A. T. Nano
Lett. 2002, 2, 761-764.
(22) Fu ¨hrer, M. S.; Kim, B. M.; Durkop, T.; Brintlinger, T. Nano Lett.
2002, 2, 755-759.
(23) Chen, R. J.; Choi, H. C.; Bansaruntip, S.; Yenilmez, E.; Tang, X.
W.; Wang, Q.; Chang, Y. L.; Dai, H. J. Am. Chem. Soc. 2004, 126,
(24) Star, A.; Gabriel, J. C. P.; Bradley, K.; Gru ¨ner, G. Nano Lett. 2003,
Nano Lett., Vol. 7, No. 10, 20073091