Aptamer-Based Sensor Arrays for the Detection
and Quantitation of Proteins
Romy Kirby,†Eun Jeong Cho,†Brian Gehrke,†Travis Bayer,†Yoon Sok Park,‡Dean P. Neikirk,‡
John T. McDevitt,†and Andrew D. Ellington*,†
Department of Chemistry and Biochemistry, Department of Electrical & Computer Engineering, and Institute for Cell and
Molecular Biology, University of Texas at Austin, Austin, Texas 78712
Aptamer biosensors have been immobilized on beads,
introduced into micromachined chips on the electronic
tongue sensor array, and used for the detection and
quantitation of proteins. Aptamer chips could detect
proteins in both capture and sandwich assay formats.
Unlike most protein-based arrays, the aptamer chips
could be stripped and reused multiple times. The aptamer
chips proved to be useful for screening aptamers from in
vitro selection experiments and for sensitively quantitating
the biothreat agent ricin.
Aptamers are functional binding species that have been
selected from combinatorial oligonucleotide libraries by in vitro
selection.1,2To date, numerous aptamers with high affinity and
selectivity have been created against a variety of targets, such as
small organics, peptides, proteins, and even whole cells.3,4Aptam-
ers can compete with antibodies in a number of analytical
applications, such as ELISA-like assays,5flow cytometry,6,7affinity
probe capillary electrophoresis,8capillary electrochromatogra-
phy,9,10affinity chromatography,11,12in high-throughput screening
assays,13and more generally as biosensors, including in arrays.14-20
Aptamers have several advantages over traditional antibody-based
reagents. Unlike antibodies, which are generally produced in
organisms, aptamers can be chemically synthesized, and therefore
specifically labeled with radioscopic, fluorescent, or other report-
ers.21Moreover, while many antibodies are temperature-sensitive
and can denature upon contact with surfaces, leading to limited
shelf lives and possible compromise of assay integrity, aptamers
are stable to long-term storage, can be transported at ambient
temperature, and undergo reversible denaturation.
Because of these advantages, it would be extremely useful to
make aptamer arrays. This is especially true because in recent
years, high-density DNA microarrays have proved to be powerful
tools for genetic analyses and diagnostic assays.22-24Since
aptamers are nucleic acids, experience with DNA arrays should
be applicable to the development of aptamer arrays. In turn,
aptamer arrays can potentially expand the scope of DNA micro-
arrays to recognize expressed proteins as well as expressed
mRNAs. In this regard, numerous aptamers have already been
selected against a wide array of proteins, and the possibility of
acquiring aptamers against proteomes has been advanced by
automation of the in vitro selection procedure.25
The nature of the platform that will be used for the generation
of aptamer arrays is open to question. Simply printing aptamers
on polylysine-coated slides is unlikely to be successful, as the
aptamers will denature upon electrostatic capture. Walt and co-
workers have adapted aptamers to high-density fiber-optic arrays.18
Stanton and researchers at the company Archemix have generated
a small aptamer array that relies on scanning fluorescence
We wanted to integrate aptamer arrays with a device that could
also deliver samples and perform complex assay procedures. In
recent years, a chip-based microsphere array (or “electronic taste
chip”)26has been developed for the digital analysis of complex
fluids. This platform relies on micromachined wells that contain
microspheres with immobilized receptors and thus should allow
* Corresponding author. Phone: 512-232-3424. Fax: 512-471-7014. E-mail:
†Department of Chemistry and Biochemistry and Institute for Cell and
‡Department of Electrical & Computer Engineering.
(1) Ellington, A. D.; Szostak, J. W. Nature 1990, 346, 818-822.
(2) Tuerk, C.; Gold, L. Science 1990, 249, 505-510.
(3) Wilson, D. S.; Szostak, J. W. Annu. Rev. Biochem. 1999, 68, 611-647.
(4) Famulok, M.; Mayer, G.; Blind, M. Acc. Chem. Res. 2000, 33, 591-599.
(5) Drolet, D. W.; Moon-McDermott, L.; Romig, T. S. Nat. Biotechnol. 1996,
(6) Davis, K. A.; Abrams, B.; Lin, Y.; Jayasena, S. D. Nucleic Acids Res. 1996,
(7) Davis, K. A.; Lin, Y.; Abrams, B.; Jayasena, S. D. Nucleic Acids Res. 1998,
(8) German, I.; Buchanan, D. D.; Kennedy, R. T. Anal. Chem. 1998, 70, 4540-
(9) Kotia, R. B.; Li, L.; McGown, L. B. Anal. Chem. 2000, 72, 827-831.
(10) Rehder, M. A.; McGown, L. B. Electrophoresis 2001, 22, 3759-3764.
(11) Romig, T. S.; Bell, C.; Drolet, D. W. J. Chromatogr., B: Biomed. Sci. Appl.
1999, 731, 275-284.
(12) Deng, Q.; German, I.; Buchanan, D.; Kennedy, R. T. Anal. Chem. 2001,
(13) Green, L. S.; Bell, C.; Janjic, N. Biotechniques 2001, 30, 1094-1096, 1098,
(14) Kleinjung, F.; Klussmann, S.; Erdmann, V. A.; Scheller, F. W.; Fuerste, J.
P.; Bier, F. F. Anal. Chem. 1998, 70, 328-331.
(15) Potyrailo, R. A.; Conrad, R. C.; Ellington, A. D.; Hieftje, G. M. Anal. Chem.
1998, 70, 3419-3425.
(16) Liss, M.; Petersen, B.; Wolf, H.; Prohaska, E. Anal. Chem 2002, 74, 4488-
(17) Spiridonova, V. A.; Kopylov, A. M. Biochemistry (Moscow) 2002, 67, 706-
(18) Lee, M.; Walt, D. R. Anal. Biochem. 2000, 282, 142-146.
(19) McCauley, T. G.; Hamaguchi, N.; Stanton, M. Anal. Biochem. 2003, 319,
(20) Rajendran, M.; Ellington, A. D. Opt. Biosens. 2002, 369-396.
(21) Jayasena, S. D. Clin. Chem. 1999, 45, 1628-1650.
(22) Lockhart, D. J.; Winzeler, E. A. Nature 2000, 405, 827-836.
(23) Brown, P. O.; Botstein, D. Nat. Genet. 1999, 21, 33-37.
(24) Walt, D. R. Science 2000, 287, 451-452.
(25) Cox, J. C.; Rudolph, P.; Ellington, A. D. Biotechnol. Prog. 1998, 14, 845-
Anal. Chem. 2004, 76, 4066-4075
Analytical Chemistry, Vol. 76, No. 14, July 15, 2004
10.1021/ac049858n CCC: $27.50© 2004 American Chemical Society
Published on Web 05/25/2004
the development of sandwich assays. This so-called “electronic
taste chip” has demonstrated its analytical potential in detecting
a variety of analyte classes (acids, metal cations, metabolic
cofactors, sugars, and proteins) in complex, homogeneous
solution.26-31More importantly, the chip-based microsphere array
has proven useful for the rapid and specific detection of nucleic
acid sequences.32Therefore, we attempted to adapt the micro-
sphere array to aptamer receptors. A unique method for the facile
immobilization of aptamers was developed and proved to be
immediately useful for screening aptamer libraries. Aptamer
receptors could be used for the quantitation of protein targets and
for the development of sandwich assays, including the quantitation
of unlabeled ricin. Surprisingly, unlike most protein biosensors,
the aptamer biosensors could be denatured and reused many
times on the same chip, without loss of function.
Detection System. The “electronic tongue” (ET)26developed
by the McDevitt lab at the University of Texas at Austin consists
of a flow cell connected to an FPLC (fast performance liquid
chromatograph) pump for sample introduction and a fluorescence
microscope for observation (Figure 1A). In more detail, the
individual components included an Amersham Pharmacia Biotech
P900 pump with carbon dioxide and sample injection ports added
downstream of the pump (three-way stopcock; Baxter, Deerfield,
IL); a Zeiss Axioplan2 compound microscope (5× objective) with
a fluorescence unit containing band-pass filters for Alexa-Fluor488
(λex480/40 nm, λem535/50 nm) and Cy3 (λex545/30 nm, λem610/
75 nm) and dichroic filters (Alexa-Fluor488λ 505 nm LP and Cy3
λ 570 nm LP), and a Hamamatsu C4742-95 b/w digital camera.
The heart of the system was the flow cell, which consisted of a
circular metal housing (3.5 cm diameter), top and bottom lenses
(3 cm diameter), and an etched silicon chip that could be inserted
into the housing (1.2 cm × 1.2 cm). Captured images were
analyzed using image analysis software (IP Lab 3.6.3, Scanalytics,
Inc., Fairfax, VA).
Aptamers were biotinylated at their 5′ ends during transcription
by introducing a biotin-GpG dinucleotide into the reaction mixture.
Biotin-GpG was synthesized utilizing standard RNA phosphor-
amidite chemistry (dmf-G-RNA-CPG, i-Pr-Pac-G-CE phosphor-
amidite, and 5′-biotin phosphoramidite were purchased from Glen
Research, Sterling, VA). Deprotected biotin-GpG was purified on
a Poly-Pak column (Glen Research, Sterling, VA) following the
manufacturer’s protocol. A 20 µL transcription reaction typically
contained 800 ng of template DNA, 30 mM DTT, 5 mM ATP, 5
mM CTP, 5 mM UTP, 2 mM GTP, 3 mM biotin-GpG, and 2 units
of T7 RNA polymerase (Epicentre, Madison, WI) in transcription
buffer (60 mM Tris‚Cl pH 8.0, 10 mM NaCl, 40 mM MgCl2, 6.8
mM spermidine). Transcription reactions were incubated at 37
°C for 4 h. The transcription reactions were then treated with 1
µL of RNase-free DNase (Epicentre, Madison, WI) for 15 min at
37 °C, purified on an 8% denaturing acrylamide gel, and ethanol-
precipitated. The purified, biotinylated aptamers were resuspended
in water and stored at -20 °C.
All of the anti-lysozyme, anti-ricin aptamers, and the anti-Rev
aptamer used in the current studies have previously been
described33-35with three exceptions. The anti-lysozyme aptamers
E, J, and M were selected from a random sequence library by
(26) Goodey, A.; Lavigne, J. J.; Savoy, S. M.; Rodriguez, M. D.; Curey, T.; Tsao,
A.; Simmons, G.; Wright, J.; Yoo, S. J.; Sohn, Y.; Anslyn, E. V.; Shear, J. B.;
Neikirk, D. P.; McDevitt, J. T. J. Am. Chem. Soc. 2001, 123, 2559-2570.
(27) Christodoulides, N.; Tran, M.; Floriano, P. N.; Rodriguez, M.; Goodey, A.;
Ali, M.; Neikirk, D.; McDevitt, J. T. Anal. Chem. 2002, 74, 3030-3036.
(28) Curey, T. E.; Goodey, A.; Tsao, A.; Lavigne, J.; Sohn, Y.; McDevitt, J. T.;
Anslyn, E. V.; Neikirk, D.; Shear, J. B. Anal. Biochem. 2001, 293, 178-
(29) Curey, T. E.; Salazar, M. A.; Oliveira, P.; Javier, J.; Dennis, P. J.; Rao, P.;
Shear, J. B. Anal. Biochem. 2002, 303, 42-48.
(30) Goodey, A. P.; McDevitt, J. T. J. Am. Chem. Soc. 2003, 125, 2870-2871.
(31) McCleskey, S. C.; Griffin, M. J.; Schneider, S. E.; McDevitt, J. T.; Anslyn,
E. V. J. Am. Chem. Soc. 2003, 125, 1114-1115.
(32) Ali, M. F.; Kirby, R.; Goodey, A. P.; Rodriguez, M. D.; Ellington, A. D.;
Neikirk, D. P.; McDevitt, J. T. Anal. Chem. 2003, 75, 4732-4739.
Figure 1. Detection system. (A) The electronic tongue setup contains a fluid delivery system, fluorescence microscope, digital camera, flow
cell in which the aptamer chip will be loaded, and computer for data analysis. (B) Close-up look at a bead in a rectangular-shaped micromachined
well of the aptamer chip.
Analytical Chemistry, Vol. 76, No. 14, July 15, 20044067
methods similar to those described by Cox et.al.34The aptamer
sequences are summarized in Table 1.
Relative Affinity and Reactivity Studies of Anti-Lysozyme
Aptamer Clones. Nine different anti-lysozyme aptamer clones,
from selections done in our lab, were biotinylated and immobilized
onto streptavidin agarose beads (Sigma, St. Louis, MO). For each
of the clones, 10 µL of the streptavidin agarose bead suspension
was washed with 500 µL of Tris buffer (20 mM Tris pH 7.6, 100
mM NaCl, 5 mM MgCl2) three times and resuspended in 40 µL
of Tris buffer. Biotinylated anti-lysozyme aptamer (600 pmol) was
resuspended in 40 µL of Tris buffer, heat-denatured for 3 min at
70 °C, and allowed to cool to room temperature. The heat-
denatured biotinylated anti-lysozyme aptamer was mixed with the
washed streptavidin agarose beads and allowed to rotate for 30
min. The newly conjugated anti-lysozyme aptamer beads were
washed with 200 µL of Tris buffer three times, resuspended in
600 µL of Tris buffer, and stored at 4 °C. Lysozyme (Sigma, St.
Louis, MO) was labeled with Cy3 following the protocol from a
Cy3 monoreactive dye pack (Amersham Biosciences, Piscataway,
Initial experiments were done to gauge the relative affinities
of the anti-lysozyme aptamer clones. A small aliquot of each of
the anti-lysozyme aptamer clone beads was removed, and the
beads were loaded into the wells of a 4 × 3 silicon chip. Negative
control beads (anti-Rev aptamer, pool RNA, and unmodified
streptavidin agarose beads) were also loaded on the chip. The
loaded chip was enclosed in the flow cell device.
The loaded flow cell was washed with Tris buffer supplied
through the fluidic pump. All additions of elution buffer (100 mM
sodium citrate, 10 mM EDTA, 7 M urea, pH 5.0) and Cy3-labeled
lysozyme were made through an injection port inserted directly
upstream of the inlet to the flow cell. The assay was broken into
four steps: (1) The beads were washed with 1 mL of elution buffer,
followed by a subsequent reintroduction of Tris buffer. (2) Cy3-
labeled lysozyme (1.8 µg/mL, 100 µL) was injected into the system
and allowed to incubate for 2 min with the anti-lysozyme aptamer
beads. (3) The beads were washed with Tris buffer at a flow rate
of 0.1 mL/min for 4 min to wash away any unbound Cy3-labeled
lysozyme. (4) An image was captured showing the beads’
response. Fluorescence intensity data were collected for each anti-
lysozyme aptamer clone, corrected by subtraction of the intensity
recorded from the negative control bead, and graphed.
Subsequent experiments were performed with anti-lysozyme
aptamer clones 1, 3, and 6. A small aliquot of beads from each
clone type was removed and loaded into the wells of a 4 × 3 silicon
chip. Negative control beads (RNA pool) were also loaded on the
chip. The loaded chip was enclosed in the flow cell device.
The loaded flow cell was washed with Tris buffer supplied
through the fluidic pump. All additions of elution buffer and Cy3-
labeled lysozyme were made through an injection port inserted
directly upstream of the inlet to the flow cell. The assay was
broken into five steps: (1) The beads were washed with 1 mL of
elution buffer, followed by a subsequent reintroduction of Tris
buffer. (2) Cy3-labeled lysozyme (1.44 µg/mL, 100 µL) was
injected into the system and allowed to incubate for 2 min with
the anti-lysozyme aptamer beads. (3) The beads were washed with
Tris buffer at a flow rate of 0.4 mL/min for 28 min to wash away
any unbound Cy3-labeled lysozyme. Images were captured at six
different intervals. (4) The beads were washed again with 1 mL
of elution buffer to remove the bound Cy3-labeled lysozyme and
regenerate the anti-lysozyme aptamers, followed by a reintroduc-
tion of Tris buffer. An image was captured. (5) Cy3-labeled
lysozyme (1.44 µg/mL, 100 µL) was reintroduced into the system
and allowed to incubate for 2 min with the anti-lysozyme aptamer
beads before they were washed with Tris buffer. A final image
was captured. Fluorescence intensity data were collected for each
anti-lysozyme aptamer clone, corrected by subtraction of the
intensity of the negative control bead, and graphed as a function
Anti-Ricin Aptamer Detection. Biotinylated anti-ricin aptam-
ers were immobilized onto streptavidin agarose beads. A 10 µL
volume of a streptavidin agarose bead suspension was washed
(33) Giver, L.; Bartel, D. P.; Zapp, M. L.; Green, M. R.; Ellington, A. D. Gene
1993, 137, 19-24.
(34) Cox, J. C.; Ellington, A. D. Bioorg. Med. Chem. 2001, 9, 2525-2531.
(35) Hesselberth, J. R.; Miller, D.; Robertus, J.; Ellington, A. D. J. Biol. Chem.
2000, 275, 4937-4942.
Table 1. Sequences of Anti-Protein Aptamers
anti-lysozyme aptamersclone 15′GGGAATGGATCCACATCTACGAATTCATCAGGGCTAAAGAG
Analytical Chemistry, Vol. 76, No. 14, July 15, 2004
three times with 500 µL of PBSM buffer (PBS, 5 mM MgCl2) and
resuspended in 40 µL of PBSM buffer. Biotinylated anti-ricin
aptamer (600 pmol) was resuspended in 40 µL of PBSM buffer,
heat-denatured for 3 min at 70 °C, and allowed to cool to room
temperature. The heat-denatured, biotinylated anti-ricin aptamer
was mixed with the washed streptavidin agarose beads and gently
rotated for 30 min. The conjugated anti-ricin aptamer beads were
washed with 200 µL of PBSM buffer three times, resuspended in
600 µL of PBSM buffer, and stored at 4 °C. Ricin A chain (Sigma,
St. Louis, MO) was labeled with Alexa-Fluor488according to the
protocol supplied with Alexa-Fluor488maleimide (Molecular Probes,
Eugene, OR). Alexa-Fluor488-labeled ricin was stored at 4 °C.
A small aliquot of anti-ricin aptamer beads was removed from
the stock solution, and three individual beads were loaded into
the wells of a 3 × 2 silicon chip. Three negative control beads
(streptavidin agarose beads) were also loaded on the chip. The
loaded chip was enclosed in the flow cell device. Carbon dioxide
was used to displace the air in the loaded flow cell prior to the
initial introduction of PBSM buffer via the FPLC pump. The beads
were then washed with 1 mL of elution buffer, followed by a
reintroduction of PBSM buffer. All additions of elution buffer and
Alexa-Fluor488-labeled ricin were made through an injection port
which was inserted directly upstream of the inlet to the flow cell.
Binding assays were typically broken into three steps: (1) Ricin
in PBSM was injected (8 µg/mL, 400 µL Alexa-Fluor488-labeled
ricin) and allowed to flow over the anti-ricin aptamer beads. (2)
The beads were washed with PBSM buffer at a flow rate of 0.1
mL/min for 4 min, followed by a flow rate of 0.5 mL/min for 30
s, to wash away any unbound Alexa-Fluor488-labeled ricin. (3) An
image was captured showing the aptamer beads’ response to the
addition of the ricin in PBSM. Steps 1-3 were repeated for nine
more additions of ricin in PBSM (each sample was 8 µg/mL, 400
µL of Alexa-Fluor488-labeled ricin solution) in order to explore the
cumulative effects of serial injections of target. Fluorescence
intensity data were collected from the images, corrected by
subtraction of the intensity of the negative control bead, and
graphed against the concentration of Alexa-Fluor488-labeled ricin.
Aptamer-Antibody Sandwich Assay. An anti-ricin antibody
(anti-lectin, Ricinus communis (RCAII), Sigma, St. Louis, MO) was
labeled with Alexa-Fluor488following the protocol provided with
an Alexa-Fluor488 NHS protein-labeling kit (Molecular Probes,
Eugene, OR). A small aliquot of anti-ricin aptamer beads was
removed from the stock solution, and the beads were loaded into
the wells of five 3 × 2 silicon chips. Negative control beads
(containing biotinylated RNA pool) were also loaded on the chip.
The loaded chip was enclosed in the flow cell device and washed
with PBSM buffer supplied through the FPLC pump. All subse-
quent additions of solutions were made through an injection port
inserted directly upstream of the inlet to the flow cell. The assay
was broken into five steps: (1) The beads were washed with 1
mL of elution buffer, followed by a reintroduction of PBSM buffer
(1 mL). (2) Ricin in PBSM (3.5 µg/mL, 100 µL) was injected into
the system and incubated for 5 min with the anti-ricin aptamer
beads. (3) The beads were washed with 1 mL of PBSM buffer to
wash away any unbound ricin. (4) Alexa-Fluor488-labeled anti-ricin
antibody in PBSM (53 µg/mL, 100 µL) was injected into the
system and incubated for 10 min with the anti-ricin aptamer beads.
(5) The beads were washed with 1 mL of PBSM buffer to wash
away any unbound Alexa-Fluor488-labeled anti-ricin antibody. An
image was captured showing the beads’ response. The chip was
removed from the flow cell, and a new one was inserted. Steps
1-5 were repeated using progressively lower concentrations of
ricin and Alexa-Fluor488-labeled anti-ricin antibody in PBSM: chip
2, ricin (2.24 µg/mL, 100 µL) and Alexa-Fluor488-labeled anti-ricin
antibody (33.6 µg/mL, 100 µL); chip 3, ricin (0.96 µg/mL, 100
µL) and Alexa-Fluor488-labeled anti-ricin antibody (14.4 µg/mL, 100
µL); chip 4, ricin (0.32 µg/mL, 100 µL) and Alexa-Fluor488-labeled
anti-ricin antibody (4.8 µg/mL, 100 µL). Fluorescence intensity
data were collected from the images, corrected by subtraction of
the intensity of the negative control bead, and graphed against
the concentration of injected ricin.
Assaying Aptamer Selectivity. Anti-ricin aptamer on beads
or anti-lysozyme aptamer on beads (clone 1) was denatured with
the elution buffer. The aptamer beads were washed three times
with 500 µL of elution buffer, followed by a reintroduction of the
appropriate buffer (500 µL of PBSM three times for anti-ricin
aptamer beads and 500 µL of Tris buffer three times for anti-
lysozyme aptamer beads; each individual wash step took 2 min).
The beads (three per each aptamer) were loaded into the wells
of the 3 × 2 silicon chip. The loaded chip was enclosed in the
flow cell device. Carbon dioxide was used to displace the air in
the loaded flow cell prior to the initial buffer introduction. PBSM
buffer (supplied by an FPLC pump) was then introduced into the
loaded flow cell.
Two stock solutions of Alexa-Fluor488-labeled proteins were
made. Ricin (Sigma, St. Louis, MO) and lysozyme (Sigma, St.
Louis, MO) were labeled with Alexa-Fluor following the protocols
from Alexa-Fluor488-labeling kits. The Alexa-Fluor488-labeled pro-
teins were stored at 4 °C.
The loaded flow cell was washed with 1 mL of elution buffer,
followed by a reintroduction of PBSM buffer. A background image
was captured after washing with 1 mL of PBSM (0.1 mL/min).
Wash cycles with PBSM were introduced through the FPLC
pump. All additions of elution buffer and Alexa-Fluor488-labeled
ricin and lysozyme were made through an injection port inserted
directly upstream of the inlet to the flow cell. The assay was
broken into seven steps: (1) Alexa-Fluor488-labeled ricin was
injected (24 µg/mL, 400 µL) and allowed to flow over the anti-
ricin and anti-lysozyme aptamer beads for 1 min. (2) The beads
were washed with PBSM buffer at a flow rate of 0.1 mL/min for
10 min to wash away any unbound Alexa-Fluor488-labeled ricin.
Images were captured every 2 min during the wash cycle to show
the beads’ response to the addition of Alexa-Fluor488-labeled ricin.
(3) Elution buffer (5 mL) was injected into the flow cell to denature
the aptamers and elute all of the bound Alexa-Fluor488-labeled ricin.
An image was captured to show that the Alexa-Fluor488-labeled
ricin had been removed and that the aptamer beads exhibited a
signal comparable to the initial background signal. (4) PBSM
buffer was reintroduced into the flow cell. (5) Alexa-Fluor488-
labeled lysozyme was injected (10.8 µg/mL, 400 µL) and allowed
to flow over the anti-ricin and anti-lysozyme aptamer beads for 1
min. (6) The beads were washed with PBSM buffer at a flow rate
of 0.1 mL/minute for 10 min to wash away any unbound Alexa-
Fluor488-labeled lysozyme. Images were captured every 2 min
during the wash cycle to show the beads’ response to the addition
of Alexa-Fluor488-labeled lysozyme. (7) Elution buffer (5 mL) was
injected into the flow cell to denature the aptamers and elute all
of the bound Alexa-Fluor488-labeled lysozyme. An image was
Analytical Chemistry, Vol. 76, No. 14, July 15, 20044069
captured to show that the Alexa-Fluor488-labeled lysozyme had
been removed and that the aptamer beads exhibited a signal
comparable to the initial background signal. Fluorescence intensity
data were collected from the images and graphed against time.
Aptamer Regeneration. In initial regeneration assays, anti-
ricin aptamer beads (400 pmol) were separated into four tubes.
The anti-ricin aptamer beads were then either used as is, heat-
denatured (3 min at 70 °C), washed with an elution buffer (100
mM sodium citrate, 10 mM EDTA, 7 M urea, pH 5.0), or heat-
denatured and washed with an elution buffer. All beads were then
incubated with Alexa-Fluor488-labeled ricin (128 µg/mL, 25 µL)
and washed with 200 µL of PBSM three times to remove any
unbound Alexa-Fluor488-labeled ricin. Fluorescence intensity data
were collected by fluorescence microscopy.
Regeneration experiments were also carried out on the ET with
both anti-ricin aptamer beads and pool RNA beads. Anti-ricin
aptamer beads (400 µL) and pool RNA beads (100 µL) were
washed with 500 µL of elution buffer two times (rotating 2 min
for the first wash), followed by a reintroduction of PBSM buffer
(500 µL two times; rotating 2 min for the first wash), and a final
resuspension in 100 µL of PBSM buffer. A total of 5 µL was
removed from each tube for the background fluorescent measure-
ments of unexposed anti-ricin aptamer beads and pool RNA beads.
Alexa-Fluor488-labeled ricin (128 µg/mL, 50 µL) was incubated with
each tube of beads for 2 min (first protein incubation). Unbound
Alexa-Fluor488-labeled ricin was washed away with 500 µL of PBSM
buffer two times. The washed beads were denatured with 500 µL
of elution buffer two times and reintroduced to PBSM buffer (500
µL two times, rotating 2 min for the first wash). The anti-ricin
aptamer beads were put through five more cycles of protein
introduction, followed by elution buffer washes. Pool RNA beads
were subject to two more cycles of protein introduction and elution
wash. Images were captured after each protein incubation, as well
as each elution wash. The fluorescence intensity of anti-ricin
aptamer beads and pool RNA beads was graphed for each step.
RESULTS AND DISCUSSION
Description of the Electronic Tongue Sensor Array. The
ET sensor array has previously been used to monitor the
performance of a variety of chemical and biological sensors,
including chemical sensors and antibodies.26-32The heart of the
ET is a flow cell that contains a silicon chip with multiple,
micromachined, pyramidal wells through which fluid can pass;
the flow cell is integrated with a fluid delivery system for sample
introduction (Figure 1A). A 4 × 3 or a 3 × 2 array was typically
used in our experiments; however, somewhat larger arrays (10
× 10) have also been manufactured. Beads containing sensor
elements are nestled into each well. The flow cell delivers solutions
and analytes to the “top” of the chip, where the apertures of the
wells are widest. The solute must then pass by each bead before
exiting through the smaller apertures on the “bottom” of the chip.
The ease with which aptamers can be generated and modified
for immobilization made their adaptation to the device very
In order to better accommodate aptamer biosensors, modifica-
tions were introduced into the basic configuration of the ET. A
carbon dioxide port (3-way valve) was added to the inlet tube
(Figure 1A). Carbon dioxide was used to displace the air in the
loaded flow cell prior to the initial buffer introduction. This
innovation greatly reduced bubble formation and capture in the
flow cell, a problem that had sometimes hampered data acquisition
and that had previously limited the utility of the device for high-
throughput applications. Carbon dioxide was chosen because of
its high solubility in aqueous solutions.
In addition, the path length that was to be traversed by sample
solutions introduced into the device via the FPLC pump was
decreased by introducing a sample injection port (3-way valve)
onto the inlet tube, immediately upstream of the flow cell (Figure
1A). This innovation substantially reduced the dilution of samples
during their transit to the flow cell.
Engineering Aptamer Bead Sensor Elements for the
Electronic Tongue. While chemical and biological sensors other
than aptamers have previously been adapted to the ET, intensive
engineering efforts were required for the introduction of each new
sensor element. For example, in order to utilize chemical sensors
the initial synthetic reactions had to plan for linkers that could
be conjugated to beads, or else the sensor had to be further
modified following synthesis. Biosensors, such as antibodies, could
be generated more quickly and in parallel, but their immobilization
on beads relied on relatively nondirected conjugation chemistries
to any of a number of surface amines, frequently resulting in loss
of function and requiring the optimization of both immobilization
and analytical protocols.
Because aptamers are generated in vitro, their synthesis can
potentially be modified to accommodate both immobilization and
sensing. To this end, we have site-specifically introduced a single,
5′ terminal biotin during transcription. The compound biotin-GpG
was used as an initiator for T7 RNA polymerase36-38transcription
(but cannot be internally incorporated during transcription). While
previous reports have shown that a relatively small number of
RNA species can be initiated with biotinylated nucleotides,39we
have now used this technique with literally hundreds of different
constructs and have typically observed biotinylation yields of 40%
(data not shown). Alternatively, biotinylated DNA aptamers could
be chemically synthesized with a terminal biotin residue, or
amplified using a primer that contained a biotin conjugate.
Commercially available streptavidin agarose beads (approxi-
mately 60-100 µm in diameter) were incubated with the biotin-
ylated aptamers to produce aptamer-bead sensor elements. The
use of streptavidin agarose beads obviated an avidin conjugation
step that was used in previous bead-preparation protocols26-31and
further decreased the time required to prepare individual chips
for assay. The beads bound 50 pmol of biotin per µL of bead
suspension and were incubated with a 20% excess of biotinylated
aptamer for 30 min to fully saturate the binding sites (and thus
minimize any variance in signal due to beads size). Any collected
signals were eventually normalized relative to the size of the beads,
and in fact, the standard deviation of signal was less than 12.5%
in all experiments, irrespective of bead size. However, since the
streptavidin agarose beads were much smaller than the beads that
were previously used (120-350 µm), the dimensions of the silicon
chip had to be altered so that the beads would not fall through
the wells. The well dimensions that ultimately proved to be optimal
(36) Cheetham, G. M.; Jeruzalmi, D.; Steitz, T. A. Nature 1999, 399, 80-83.
(37) Bandwar, R. P.; Jia, Y.; Stano, N. M.; Patel, S. S. Biochemistry 2002, 41,
(38) Kuzmine, I.; Gottlieb, P. A.; Martin, C. T. J. Biol. Chem. 2003, 278, 2819-
(39) Pitulle, C.; Kleineidam, R. G.; Sproat, B.; Krupp, G. Gene 1992, 112, 101-
4070Analytical Chemistry, Vol. 76, No. 14, July 15, 2004
were 70 µm × 140 µm (Figure 1B).
The use of rectangular rather than square wells was also an
innovation. Square wells tended to be completely blocked by some
beads, leading to a reduction in flow and a concomitant buildup
of pressure within the system, sometimes resulting in cracking
of the silicon chip. Rectangular wells always allowed solution to
flow around individual beads. Nonetheless, fluid flow through the
porous beads and past the aptamer biosensors in rectangular wells
was more than sufficient for analyte detection, as will be seen
Overall, one of the advantages of loading biotinylated aptamers
directly onto preconjugated streptavidin beads was that multiple
different biosensors could be processed in parallel and chips could
be quickly prepared. The total time for the construction of a 4 ×
3 aptamer array was typically 8 h (6 h for aptamer transcription
and purification, 45 min for bead conjugation, and 20-40 min for
bead loading). Multiple chips were prepared in parallel, and only
the time required for bead loading was extended.
Using the Electronic Tongue To Assess in Vitro Selection
Experiments. The facile loading of aptamer biosensors onto the
electronic tongue immediately suggested a practical application
for this platform. An in vitro selection experiment typically yields
a number of aptamers or aptamer families (a series of related
aptamers). While these aptamers as a whole represent the tightest-
binding species in a nucleic acid population, each selected
sequence must nonetheless still be screened for its relative ability
to bind to a target ligand. Current methods for screening aptamers
generally require that multiple filter-binding assays be carried out
in series and are slow and cumbersome. Bead-based immobiliza-
tion of aptamers in the context of the electronic tongue may
therefore provide an opportunity to screen large numbers of
variants in parallel for relative binding to target (and nontarget)
To evaluate the use of the electronic tongue for aptamer
screening, we utilized nine different anti-lysozyme aptamers
(denoted as 1-6, E, J, and M in Figure 2A) that had previously
been selected from several different random sequence popula-
tions.34The anti-lysozyme aptamers were transcribed in parallel
with biotin-GpG, purified, and immobilized on streptavidin agarose
beads. As a negative control, unselected pool RNA similarly
conjugated to beads (denoted as “0” in Figure 2A) shows little
background binding to lysozyme. Unmodified streptavidin agarose
beads (denoted as SA in Figure 2A), as well as beads labeled with
an anti-Rev aptamer (denoted as Rev in Figure 2A), served as
additional negative controls. All nine anti-lysozyme aptamer and
negative control beads were loaded onto a single 4 × 3 chip for
screening (Figure 2A). The chip was incubated with fluorescently
labeled lysozyme, and the relative amounts of captured fluores-
cence were determined (Figure 2B, C). Clones 1, 3, and 6
appeared to bind most tightly to the labeled lysozyme. Looking
at the image carefully (Figure 2B), there are four additional images
surrounding the bead itself (especially for beads denoted as “1”
and “6”). As previously discussed,26each well is a pyramidal pit
whose side is angled inward at 54.7°. This pyramidal shape causes
the observed reflections when the bead is observed from the top
(as opposed to configurations26,29,30in which the flow cell with the
chip sits atop an optical/fluorescence microscope). Fluorescent
signals were collected only from the center circle, which corre-
sponds to the actual bead, and signals were normalized based on
the bead’s diameter.
The anti-lysozyme aptamers were then split into three groups,
based on their apparent affinities. Three new 4 × 3 chips were
loaded with anti-lysozyme aptamers. The three clones which
showed the best responsivities (clones 1, 3, and 6) were loaded
in triplicate on one chip, while the other six clones were divided
into two groups of three (clones 2, 4, 5 and clones M, J, E) and
loaded in triplicate on the other two chips. In addition, a negative
control (the nascent RNA pool) was loaded in triplicate on each
chip. Following the introduction of fluorescent lysozyme, the
fluorescence of individual wells was monitored, and the data for
each aptamer was averaged.
The binding abilities of the two best anti-lysozyme aptamers
(clones 1 and 6) and one aptamer (clone 3) that gave a relatively
small signal were further assessed as a function of time (Figure
3). After an initial incubation of 2 min, the beads were washed
with 0.4 mL/min of buffer for 28 min (11.2 mL of buffer). Right
after the sample incubation, all beads including negative control
beads exhibited very bright fluorescence intensity due to non-
specific binding. This nonspecific signal dramatically decreased
with washing (compare left and right columns, Figure 3A). The
fluorescence intensity of the negative control beads fell to a basal
line during the first 10 min of washing (data not shown), while
the fluorescence intensity of the aptamer-labeled beads decreased
slowly depending on each clone’s binding affinity as summarized
in Figure 3B. Clone 1 bound approximately 1.1-fold more fluo-
rescently labeled lysozyme than clone 6 and 7-fold more than clone
3 after 10 min of washing. At the end of the buffer wash, clones
1 and 6 exhibited a slight decrease (1.2-fold) in bound fluores-
cently labeled lysozyme, while clone 3 showed a greater decrease
(3.3-fold). This qualitative comparison of dissociation rate con-
Figure 2. Parallel screening of anti-lysozyme aptamer clones. (A) Position of anti-lysozyme aptamer clones (1-6, E, J, and M) and three
negative control beads in in a 3 × 4 chip. The negative controls were an anti-Rev aptamer (Rev), pool RNA (0), and unmodified streptavidin
agarose beads (SA). (B) Fluorescent images of the anti-lysozyme aptamers and controls responding to Cy3-labeled lysozyme. (C) Fluorescence
intensities of Cy3-labeled lysozyme captured by different anti-lysozyme aptamer clones.
Analytical Chemistry, Vol. 76, No. 14, July 15, 20044071
stants confirmed the original rank-ordering of the clones. How-
ever, much less aptamer was lost over time based on calculated
off-rates than was expected. The retention of fluorescently labeled
lysozyme on the beads was likely due to analyte-trapping by the
highly concentrated biosensor; once the protein was released from
one aptamer, it was quickly bound by an adjacent aptamer.
However, while proteins may be partially trapped by interactions
with concentrated biosensors, they are not held irreversibly. After
stripping the chips with a chaotrope (urea), the fluorescent signal
fell to background levels and all three anti-lysozyme aptamer
clones once again responded to the reintroduction of fluorescently
Based on these results, we predicted that clones 1 and 6 were
the best binding species, with clone 3 having a lesser affinity for
lysozyme. To confirm the results obtained with immobilized
aptamers on the ET, filter-binding curves were obtained for all of
the anti-lysozyme aptamer clones by incubating a constant
concentration of radiolabeled anti-lysozyme aptamer with increas-
ing concentrations of lysozyme. The solution-phase binding curves
were plotted and used to derive Kdvalues for each aptamer-
lysozyme complex. Clones 1, 3, and 6 had calculated Kdvalues of
29 ( 5 nM, 83 ( 6 nM, and 128 ( 9 nM, respectively, while the
other six clones were well above 230 nM (data not shown).
As can be discerned from the solution-phase Kdvalues, the
results with the immobilized aptamers were not identical. For the
solution-phase aptamers, the apparent order of affinity was clone
1 > clone 3 > clone 6, while for the immobilized aptamers the
order was clone 1 > clone 6 > clone 3. The differences in the
relative affinities of the anti-lysozyme aptamer clones could
potentially be due to the fact that solution-phase aptamers were
probed with unlabeled protein, while the immobilized aptamers
were probed with fluorescently labeled protein.
The effect of aptamer and protein derivatization on measured
binding affinities were further explored through a series of
additional solution-phase binding assays with clones 1, 3, and 6.
In these assays, RNA was either biotinylated or not, and lysozyme
was either labeled with Cy3 or not. Biotinylation was found to
slightly decrease the aptamers’ ability to bind unlabeled lysozyme
(data not shown). However, labeling the protein resulted in far
less lysozyme being bound by the aptamers, and the relative order
of affinities changed, with clone 3 being more sensitive to
fluorescent labeling of lysozyme than clone 6. As a negative
control, naı ¨ve pool RNA (labeled and unlabeled) showed no
significant binding to either labeled or unlabeled lysozyme.
Although there are subtle differences between the affinities
of suboptimal binders measured on the electronic tongue and
those measured in solution assays, the electronic tongue nonethe-
less provided a quick method for screening the best clones from
a selection experiment. Moreover, to the extent that aptamers
selected against unmodified proteins are eventually to be used
with labeled proteins (for example, in the context of the electronic
tongue), the screen provides for the immediate identification of
those variants that will have the best analytical performance.
Assay Formats for Arrayed Aptamers. Beyond using the
electronic tongue to functionally screen aptamers, it should be
possible to use this biosensor array to quantify protein concentra-
tions. In particular, we will demonstrate both the quantitation of
labeled proteins, using aptamers as capture reagents (Figure 4A),
and the quantitation of unlabeled proteins, in a sandwich assay
format with antibodies (Figure 4B).
Quantitation assays were carried out with an aptamer originally
selected to bind to the biothreat agent ricin.35As before, the
aptamer was biotinylated, immobilized, and probed with fluores-
cently labeled protein (Figure 5). Images of beads with no aptamer
(negative control, SA) and beads loaded with anti-ricin aptamer
(Aptamer) are shown before (Figure 5A) and after (Figure 5B)
injection of the fluorescently labeled ricin. As with lysozyme, the
ability of the immobilized aptamer to capture its protein target
The fluorescence intensities of the captured proteins could be
extracted from these images, and the dose-response curve was
generated by plotting the mean fluorescence intensity ( standard
deviation (SD) for three independent beads (Figure 5C). One of
the advantages of the array format is that data could be collected
in parallel, as well as serially. A four-parameter Gompertz equation
(SigmaPlot, SPSS, Chicago, IL) was used to plot the best line
Figure 3. Reactivity of anti-lysozyme aptamers. (A) Fluorescent image of three anti-lysozyme aptamer clones (clones 1, 3, and 6; left column,
denoted as “Aptamer”) and three negative control beads (right column, denoted as SA) after exposure to Cy3-labeled lysozyme. The image was
taken after the clones had been washed for 28 min with Tris buffer. (B) Fluorescence data showing the reactivity of the three anti-lysozyme
aptamer clones after 10 min of washing. Images were captured at 10, 12, 16, 20, 24, and 28 min during the washing cycle. Intensity data for
individual beads was first corrected by subtracting the fluorescence intensity of the mean of several control beads. Each data point represents
the corrected mean fluorescence intensity for three experimental beads ( the standard deviation.
Analytical Chemistry, Vol. 76, No. 14, July 15, 2004
through the data points. The calculated detection limit (concentra-
tion corresponding to 3 SD above the mean of the background)
was slightly lower than 100 pmol of applied sample (8 µg/mL in
a volume of 400 µL). The estimated Kdof the aptamer-protein
complex was 1.24 µM, which was considerably higher than the
reported Kd(7.4 nM).35There are several possible reasons for
this observed difference. In addition to the disruption of binding
due to labeling, immobilization of the aptamers onto beads may
have affected the binding affinity by causing steric hindrance or
other surface effects.18,19The addition of a linker between the
aptamer and biotin may alleviate these effects, as McCauley et al.
have previously demonstrated.19Slow sample penetration into the
porous beads could also have led to a decrease in the observed
affinity. According to recent confocal analyses,32sample access
was restricted in the lower half of beads due to interactions with
the walls of the pyramidal pits, and this was thought to operation-
ally decrease the detection of molecular interactions. However,
since the sample was pumped through the beads under pressure
over a period of several minutes, it seems likely that the receptors
would have had adequate time to equilibrate.
In order to examine whether unlabeled ricin could also be
detected, a sandwich assay format as shown in Figure 4B was
adopted. In this system, anti-ricin aptamer acted as a capture
reagent and unlabeled ricin served as a bridge to a fluorophore-
labeled antibody that served as a reporter. Although it was not
known in advance whether the aptamer and antibody would
compete for the same epitope on ricin, a fluorescent signal was
readily obtained when all of the components of the sandwich were
in place. A negative control that lacked ricin yielded no fluorescent
signal (Figure 5D). By employing this sandwich format we could
detect as little as 1 pmol of unlabeled ricin (320 ng/mL in a volume
of 100 µL).
The general problem of using modified proteins with reagents
selected to bind to unmodified proteins could potentially be solved
through the use of sandwich assays, as described above. In
addition, it should be relatively simple to adapt in vitro selection
procedures to select for aptamers that will bind to protein targets
that have previously been modified with the same dye or other
conjugate that will be used during detection.
Because of the unexpected influence of analyte modification
on affinity, experiments were also performed to ensure that the
solution-phase selectivity of the aptamers remained intact following
immobilization. Two different aptamers, the anti-ricin aptamer and
Figure 4. Aptamer assays. (A) Schematic representation of an anti-
protein aptamer (immobilized on a streptavidin agarose bead) captur-
ing a fluorescently labeled protein. (B) Schematic representation of
an anti-protein aptamer/protein/anti-protein antibody sandwich assay.
Figure 5. Response of anti-ricin aptamer. (A) Background of control streptavidin agarose beads (denoted as SA) and anti-ricin aptamer
beads (denoted as Aptamer). (B) Aptamer response after the addition of Alexa-Fluor488-labeled ricin. (C) Quantitation of Alexa-Fluor488-labeled
ricin with anti-ricin aptamer beads. (D) Quantitation of unlabeled ricin via an anti-ricin aptamer/ricin/anti-ricin antibody sandwich assay. Intensity
data for individual beads was first corrected by subtracting the mean fluorescence intensity of several negative control beads. Each data point
represents the corrected mean fluorescence intensity for five experimental beads ( the standard deviation.
Analytical Chemistry, Vol. 76, No. 14, July 15, 20044073
anti-lysozyme aptamer (clone 1) were assayed on a single 3 × 2
chip. Initial background images were captured prior to protein
incubation (Figure 6A, and t ) 0 in Figure 6F). The beads were
incubated with fluorescently labeled ricin in PBSM for 1 min (t )
1 min in Figure 6F), and then washed with PBSM for 10 min in
order to remove any nonspecifically bound protein. Upon the
addition of Alexa-Fluor488-labeled ricin, an increase in fluorescence
intensity was seen for both sets of aptamer beads. However,
following the wash step the anti-lysozyme aptamer beads (clone
1) returned to background levels of fluorescence (Figure 6F).
Images showing the aptamer beads’ response to the addition of
fluorescently labeled protein (captured after the 10 min PBSM
buffer wash) were acquired (Figure 6B). The aptamer beads were
washed with an elution buffer after each protein incubation to
verify that the fluorescently labeled proteins could be removed.
Signals equivalent to the initial background level were generated
(Figure 6C, and t ) 0 in Figure 6G). The process was then
repeated with the addition of fluorescently labeled lysozyme,
rather than fluorescently labeled ricin (Figure 6D, E, and G).
Fluorescence data were again collected as a function of time.
Conversely, the anti-ricin aptamer beads showed no appreciable
increase in fluorescence intensity upon the addition of the Alexa-
Fluor488-labeled lysozyme (Figure 6G). Lysozyme is a particularly
good control in this regard, as it is very basic (pI ) 9.1) and is
known to bind nonspecifically to nucleic acids. Overall, while there
may be some nonspecific binding of labeled proteins to aptamers,
specificity is ultimately obtained by simply washing the beads.
Aptamers can be directly compared with antibodies as capture
reagents. Ligler and co-workers40-4340-43could detect ricin con-
centrations as low as 8 ng/mL40by employing an antibody-based
immunosensor array, in which biotinylated capture antibodies
were immobilized onto avidin-coated microscope slides, and
sandwich assays were carried out with ricin and a fluorescently
labeled antibody. It is instructive that aptamers have detection
limits and performance characteristics that are similar to those
previously exhibited by antibodies. The initial success of the
sandwich assay and the fact that beads are contained within
microwells on the electronic tongue format also raises the
possibility that enzyme-linked immunoassays can be carried out
to boost signals and lower detection limits.
Reusable Biosensors. The development of a chip array that
could be used for repetitive or continuous analyte monitoring
would be extremely useful. However, antibodies and other protein-
based biosensors cannot generally be used more than once. Once
a protein has captured or interacted with an analyte, the analyte
must be stripped from the protein under conditions that frequently
also denature the protein. Because of this, most protein-based
biosensors or chip arrays are used only once, then discarded.
(40) Ligler, F. S.; Taitt, C. R.; Shriver-Lake, L. C.; Sapsford, K. E.; Shubin, Y.;
Golden, J. P. Anal. Bioanal. Chem. 2003, 377, 469-477.
(41) Delehanty, J. B.; Ligler, F. S. Anal. Chem. 2002, 74, 5681-5687.
(42) Taitt, C. R.; Anderson, G. P.; Lingerfelt, B. M.; Feldstein, M. J.; Ligler, F. S.
Anal. Chem. 2002, 74, 6114-6120.
(43) Rowe-Taitt, C. A.; Golden, J. P.; Feldstein, M. J.; Cras, J. J.; Hoffman, K. E.;
Ligler, F. S. Biosens. Bioelectron. 2000, 14, 785-794.
Figure 6. Selective responses of anti-ricin aptamer and anti-lysozyme aptamer. (A) Background of anti-ricin aptamer beads (denoted as Ric)
and anti-lysozyme aptamer beads (denoted as Lys). (B) Aptamer response after the addition of Alexa-Fluor488-labeled ricin. (C) Removal of
bound protein from aptamer beads (aptamer reactivation). (D) Aptamer response after the addition of Alexa-Fluor488-labeled lysozyme. (E)
Removal of bound protein from aptamer beads (aptamer reactivation). (F) Fluorescence intensity of the aptamer beads after exposure to Alexa-
Fluor488-labeled ricin. (G) Fluorescence intensity of the aptamer beads after exposure to Alexa-Fluor488-labeled lysozyme. Each data point
represents the mean fluorescence intensity ( standard deviation for three beads.
Analytical Chemistry, Vol. 76, No. 14, July 15, 2004
In contrast, aptamers can frequently be denatured and refolded
multiple times without loss of activity, in part because the aptamer
binding function is largely dependent upon simple, stable second-
ary structural interactions, rather than more complex, weaker
tertiary structural interactions, as in proteins. The primacy of
secondary structure in aptamer binding function also means that
aptamers can be readily engineered for additional stability by
adding base-pairs to stem structures. Moreover, since aptamers
are typically denatured and refolded during each successive round
of in vitro selection, they have to some extent been “pre-evolved”
All of these considerations suggested that the aptamer chip
arrays that we had produced might be reusable. In order to test
the structural robustness of aptamers, aptamer-labeled beads were
treated by either heating to 3 min at 70 °C, by washing with a
buffer containing 7 M urea, or by both heating and washing. After
each treatment, beads were incubated with Alexa-Fluor488-labeled
ricin, washed, and the amount of captured protein was determined
by fluorescence microscopy. There was no drop in the efficiency
of protein capture compared with no treatment.
In order to demonstrate that aptamer chips could in fact be
reused multiple times, we carried out an assay similar to the one
shown in Figure 4A, but between each assay captured ricin was
eluted from the aptamer with a buffer containing a high concen-
tration of urea. This buffer was exchanged for the normal binding
buffer, the aptamers were allowed to refold, and ricin was
reintroduced into the buffer stream (Figure 7) In greater detail,
initial background fluorescent images were taken of anti-ricin
aptamer beads and pool RNA beads (Figure 7, interval 0). The
anti-ricin aptamer beads were then incubated with Alexa-Fluor488-
labeled ricin (Figure 7, interval P1; circles), followed by an elution
buffer wash (Figure 7, interval E1; circles). In parallel, pool RNA
beads were incubated with Alexa-Fluor488-labeled ricin (Figure 7,
interval P1; triangles), followed by an elution buffer wash (Figure
7, interval E1; triangles). Five more incubation/elution cycles were
performed with the anti-ricin aptamer beads (Figure 7, intervals
P2-P6 and E2-E6; circles), and two more cycles were performed
with the pool RNA beads (Figure 7, intervals P2-P3 and E2-E3;
It is interesting to note that the fluorescence intensity signal
from the initial protein exposure (P1) for the anti-ricin aptamer
beads is slightly lower than the signal generated from all five
subsequent protein incubations. These results are consistent with
the notion that aptamers must undergo denaturation and refolding
steps similar to those that were present during their initial
selection in order to attain full activity. As a further confirmation
of this observation, immobilized aptamers that were initially heat-
denatured were able to capture more analyte than immobilized
aptamers that did not see such a heat denaturation step (data not
Aptamers are potentially useful biosensor reagents that can
both substitute for antibodies and that can be adapted in novel
ways to sensor platforms. However, methods must be developed
to adapt aptamers to sensor arrays. To this end, a facile method
for aptamer conjugation to a bead-based sensor array, the
electronic tongue, was developed. The immobilized aptamers
could specifically capture protein analytes, and aptamer chips were
used to identify the best aptamers from in vitro selection
Assays for protein identification and quantitation were devel-
oped and applied to the detection of the biothreat agent ricin. The
limits of detection (320 ng/mL) in a sandwich assay format were
comparable to those previously observed with antibody assays.
However, unlike antibody assays, the aptamer arrays could be
completely stripped of protein analyte and reused without loss of
function. These results raise the prospect that aptamers could be
adapted to function as reusable, long-term biosensors in arrays.
This research was supported by the Arnold and Mabel
Beckman Foundation, Office of Naval Research (N00014-00-1-
0930), Department of the Army Research, MURI (DAAD19-99-1-
0207), National Science Foundation (EIA-0219447), and the
Countermeasures to Biological and Chemical Threats Program
at the Institute for Advanced Technology (DAAD13-02-C-0079;
Received for review January 23, 2004. Accepted April 22,
Figure 7. Reactivation of anti-ricin aptamer beads. Fluorescence
measurements of anti-ricin aptamer beads (circles) were taken at 13
different intervals. Fluorescence measurements of pool RNA beads
(triangles) were taken at seven different intervals. Interval 0 records
the background fluorescence of unexposed anti-ricin aptamer and
pool RNA beads. Intervals P1-P6 show the fluorescence intensity
of the anti-ricin aptamer beads upon binding an Alexa-Fluor488-labeled
ricin. Intervals P1-P3 also show the fluorescence intensity of the pool
RNA beads upon binding Alexa-Fluor488-labeled ricin. Intervals E1-
E6 show a drop in fluorescence (down to the background level)
consistent with the denaturing and reactivation of the anti-ricin
aptamer beads. Intervals E1-E3 also show the fluorescence intensity
of the denatured pool RNA beads. Each data point (for every interval
in the sequence) represents the mean fluorescent intensity ( standard
deviation for five beads.
Analytical Chemistry, Vol. 76, No. 14, July 15, 20044075