4830–4837
Nucleic Acids Research, 1999, Vol. 27, No. 24
© 1999 Oxford University Press
Single-base mismatch detection based on charge
transduction through DNA
Shana O. Kelley, Elizabeth M. Boon, Jacqueline K. Barton*, Nicole M. Jackson
1
and
Michael G. Hill
1
Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, CA 91125, USA and
1
Department of Chemistry, Occidental College, Los Angeles, CA 90041, USA
Received April 7, 1999; Revised September 15, 1999; Accepted October 1, 1999
ABSTRACT
High-throughput DNA sensors capable of detecting
single-base mismatches are required for the routine
screening of genetic mutations and disease. A new
strategy for the electrochemical detection of single-
base mismatches in DNA has been developed based
upon charge transport through DNA films. Double-
helical DNA films on gold surfaces have been
prepared and used to detect DNA mismatches elec-
trochemically. The signals obtained from redox-
active intercalators bound to DNA-modified gold
surfaces display a marked sensitivity to the presence
of
base
mismatches
within
the
immobilized
duplexes. Differential mismatch detection was
accomplished irrespective of DNA sequence compo-
sition and mismatch identity. Single-base changes in
sequences hybridized at the electrode surface are
also detected accurately. Coupling the redox reac-
tions of intercalated species to electrocatalytic proc-
esses in solution considerably increases the
sensitivity of this assay. Reporting on the electronic
structure of DNA, as opposed to the hybridization
energetics of single-stranded oligonucleotides, elec-
trochemical sensors based on charge transport may
offer fundamental advantages in both scope and
sensitivity.
INTRODUCTION
DNA-based sensors have potential applications that range
from genomic sequencing to mutation detection and pathogen
identification (1–4). Biochemical assays (5–10), traditional
separation methods (11), gravimetric analyses (12–15) and
spectroscopic probes (16–23) have all been employed in the
construction of DNA biosensors. Indeed, sophisticated analyt-
ical schemes employing high-resolution microscopy to assay
the hybridization of DNA target sequences with arrays of
immobilized single-stranded oligonucleotides have been
developed for highly parallel genomic sequencing and the
detection of mutations (16,17). Simpler electrochemical
schemes have also been explored (24–35). In a typical assay,
single-stranded probe sequences are immobilized on an
electrode, then treated with test DNA samples. If hybridization
occurs, the electrochemistry of a positively charged redox-
active reporter molecule [e.g. Co(phen)
3
2+
] added to the solu-
tion shows an enhanced response owing to its increased attrac-
tion to the more negatively charged duplex-modified surface.
All of these assays rely ultimately on molecular recognition
events associated with DNA hybridization to catalog sequence
information. Applied to base-mismatch detection, hybridiza-
tion assays are inherently limited in sensitivity: detection of a
point mutation in the test sequence (e.g. a small segment of
genomic DNA) requires a distinguishable difference in pairing
energies between the probe sequence and a completely
complementary versus mutated target strand. With only a
single mutation in an extended oligonucleotide, these differ-
ences can be very small. Moreover, duplex stabilities for oligo-
nucleotides of a fixed length can vary considerably as a
function of base content, with GC-rich sequences significantly
more stable than AT-rich analogs. As a consequence, detection
of point mutations within libraries of immobilized oligo-
nucleotides (where duplex-binding energies for adjacent probe
sequences may vary significantly more than the differential
binding energies of a particular probe with its complementary
versus mutated test sequences) requires extensive manipula-
tion of hybridization conditions as well as sophisticated decon-
volution algorithms.
Monitoring charge transport through double-stranded DNA
offers an alternative approach to the detection of point muta-
tions. Photoinduced electron transfer through donor/acceptor-
labeled duplexes has been observed in a variety of systems
(36–39), and efficient electrochemical reduction of redox-
active molecules intercalated into the individual helices of
double-stranded DNA films has been reported (40,41). Signif-
icantly, DNA-mediated reactions show a weak dependence on
distance but are exceptionally sensitive to perturbations in the
base stack: intervening bulges inhibit long-range photochem-
ical guanine oxidation (42), and single-base mismatches mark-
edly reduce photoinduced electron-transfer yields (37). Thus,
while single-base mismatches may cause only subtle changes
in duplex stability and structure (43,44), they appear to induce
significant perturbations in the electronic structure of the base-
pair stack. DNA-mediated charge transfer may therefore
provide a complementary signaling mechanism for DNA-
based sensors.
*To whom correspondence should be addressed. Tel: +1 626 395 6075; Fax: +1 626 577 4976; Email: jkbarton@its.caltech.edu
Nucleic Acids Research, 1999, Vol. 27, No. 24
4831
The properties of DNA films and the electron-transfer reac-
tions of intercalators bound to these monolayers have been
explored previously (41–46). In studies of daunomycin (DM, a
redox-active antitumor agent) site-specifically bound to immo-
bilized DNA duplexes, efficient electron transfer was observed
over DM/electrode separations up to 35 Å (41). Indeed, while
not sensitive to distance, this reaction was dramatically attenu-
ated by the presence of an intervening CA mismatch. This
observation demonstrated that mismatches could be detected at
a DNA-modified electrode, using the efficiency of DNA-
mediated charge transport as a signaling device.
In these first studies, the intercalator binding site was
controlled by site-specifically crosslinking DM to guanine
residues in the duplex. Both to eliminate the time-consuming
chemical crosslinking, and to avoid multiple labeling at GC
base steps that are possible at any site along a potential
genomic test sequence, a practical adaptation of this system
would require (i) probes that were non-covalently bound to the
DNA films; (ii)
in situ
hybridization of test sequences at the
electrode surface; and (iii) large differences in the electro-
chemical responses of intercalators at electrodes featuring fully
base-paired versus mismatched duplexes.
Here, we report the electrochemical detection of DNA
mismatches using different redox-active intercalators non-
covalently bound to DNA-modified surfaces. The response of
intercalating probes associated with the films effectively
reports the presence of a wide variety of mismatches. More-
over, single-base changes in sequences of varied base compo-
sition are detected with equal sensitivity, demonstrating an
advantage of this detection approach over hybridization-based
methods. Mismatch detection can also be performed at films
formed by the reversible
in situ
hybridization of oligonucle-
otides to probe sequences immobilized on the surfaces, an
important feature for the implementation of a practical assay.
Furthermore, we have coupled direct electrode-intercalator
electron transfer to an electrocatalytic cycle involving a non-
intercalating substrate in solution. The resulting assay exhibits
greatly enhanced differentiation between complementary
versus mismatched duplexes, and allows the ready detection of
point mutations in DNA oligonucleotides.
MATERIALS AND METHODS
Materials
All DNA synthesis reagents were obtained from Glen
Research. DM was obtained from Fluka; methylene blue,
potassium ferricyanide and ruthenium pentamine chloride
were purchased from Aldrich and used as received.
Preparation of DNA-modified surfaces
Thiol-modified oligonucleotides were prepared as previously
described (40,41,45); thiol-terminated linkers were attached to
single-stranded oligonucleotides (47), which after stringent
purification were hybridized to unmodified complements. The
resultant duplexes were deposited on polycrystalline gold elec-
trodes for 24 h. Before electrochemical measurements, the
electrodes were rinsed thoroughly with 5 mM phosphate, 50
mM NaCl buffer (pH 7). As electrodes containing a high
surface coverage of DNA were most useful for our experi-
ments, surfaces were routinely assayed for coverage by
monitoring the attenuation of the oxidation of ferrocyanide
(40). Comparable results were obtained with commercial poly-
crystalline electrodes (BAS) or Au(III) films vapor deposited
on mica substrates (Molecular Imaging).
Electrochemical measurements
Cyclic voltammetry and chronocoulometry were carried out on
0.02 cm
2
gold electrodes using a Bioanalytical Systems (BAS)
Model CV-50W electrochemical analyzer. A normal three-
electrode configuration consisting of a modified gold-disk
working electrode, a saturated calomel reference electrode
(SCE, Fisher Scientific) and a platinum wire auxiliary elec-
trode was used. The working compartment of the electro-
chemical cell was separated from the reference compartment
by a modified Luggin capillary. Potentials are reported versus
SCE. Volumes of 2.5 ml were typically employed. Unless
specifically noted, all measurements were recorded at 20
±
2
°
C
in 5 mM sodium phosphate buffer containing 50 mM NaCl, pH
7 that had been thoroughly degassed with Ar.
RESULTS AND DISCUSSION
Electrochemical detection of single-base mismatches
As previously described (40,41,45), DNA-modified surfaces
are prepared by the self-assembly of 15-bp duplexes deriva-
tized at the 5
′
-end with a thiol-terminated aliphatic linker.
Atomic-force microscopy studies have shown that the
duplexes form densely-packed monolayers with the individual
helices in an upright orientation with respect to the gold
surface (45) (Fig. 1). Redox-active cations [e.g. Ru(NH
3
)
6
3+
]
and DNA intercalators bind strongly to the modified surfaces
and yield well-behaved electrochemical signals; anions [e.g.
Fe(CN)
6
4–
] and non-binding neutral species (e.g. dimethyl-
aminoferrocene) do not associate with the electrodes and are
electrochemically silent (40).
Detection of mismatches using non-covalently bound DM.
The
electrochemistry of DM at electrodes modified with the duplex
5
′
-AGTACAGT
C
ATCGCG [where
C
indicates the location of
a CA mismatch; all sequences are labeled at the 5
′
end of
sequence with the linker of the formula SH-(CH
2
)
2
CO-
NH(CH
2
)
6
NHCO-] is shown in Figure 2. The integrated charge
due to the reduction of DM at fully base-paired films indicates
an intercalator-to-DNA binding stoichiometry of ~1:1 (40,45).
Because this ratio is far smaller than that predicted by neighbor
exclusion, it is likely that DM binds predominantly near the
solvent-exposed terminus of the film, with diffusion into the
monolayer inhibited by the tight packing of the DNA helices.
Similar results were obtained in a previous study where meth-
ylene blue was non-covalently bound to DNA monolayers
(40). Here methylene blue showed the same binding affinity
but lower stoichiometry at saturation when compared with
neighbor excluded binding to the same DNA duplexes in solu-
tion. When films with lower surface coverage are used, a larger
signal is obtained for the redox-active DNA intercalator, which
is likely due to increased access to the interior of the film.
Thus, even without covalent crosslinking, intercalators appear
to be somewhat constrained to the top of densely-packed
monolayers.
4832
Nucleic Acids Research, 1999, Vol. 27, No. 24
The presence of a single mismatch in the DNA duplexes
caused a striking decrease in the electrochemical response
(Fig. 2). Based on the observation of an almost quantitative
decrease in the electrochemical signal with an intervening
mismatch using crosslinked DM, it would be expected in the
non-covalent system that as long as the base mismatch inter-
venes between the DM-binding site and the electrode, the
charge-transfer event should be inhibited. It is likely that the
residual DM response results from a small percentage of inter-
calator that binds beneath the mismatch. Additionally, some
intercalators may adsorb directly to the electrode surface and
therefore may be unaffected by perturbations in the DNA
monolayer. Notably, the signals at these films correlate with
the location of the mismatch along the sequence. Films
containing mismatches closest to the electrode surface show
the largest attenuation in signal (Table 1). Moreover, the effect
of mismatches is most pronounced in densely packed films
where larger populations of the intercalating probes appear to
bind at the periphery of the monolayer.
Variation of redox-active probe.
A range of intercalators and
groove-binders (Fig. 1) were examined as probes for the detec-
tion of mismatches within DNA films; the results are summa-
rized in Table 2. The efficiencies of mismatch detection using
the various reporter molecules reveal several important charac-
teristics of this assay.
A redox probe that intercalates into the DNA base stack
appears to be a necessary component for mismatch detection.
Probes that associate with DNA through purely electrostatic
interactions (48) do not yield measurable differences in the
electrochemical response in the presence of base mismatches.
Thus, while the electrochemical signals obtained from the
intercalators
DM
(49),
methylene
blue
(50)
and
Ir(bpy)(phen)(phi)
3+
(C.S. Stinner, S.O. Kelley, M.G. Hill and
J.K. Barton, unpublished results), are all affected by the pres-
ence of the mismatch, the response of a groove-binding agent,
Ru(NH
3
)
5
Cl
2+
[or Ru(NH
3
)
6
3+
], is essentially identical at fully-
base paired or mismatched films. The reduction of
Ru(NH
3
)
5
Cl
2+
likely proceeds through the facilitated diffusion
Figure 1.
(
A
) Schematic illustration of DNA duplexes (cyan and white) immobilized on a gold surface. A variety of species are depicted in the vicinity of the
monolayer, including the intercalators Ir(bpy)(phen)(phi)
3+
(orange), DM (red) and methylene blue (blue), the groove-binder Ru(NH
3
)
5
Cl
2+
(purple) and Fe(CN)
6
4–
(green), which does not associate with the immobilized helices due to its negative charge. (
B
) Chemical structures of Ir(bpy)(phen)(phi)
3+
, daunomycin (DM),
methylene blue (MB), Ru(NH
3
)
5
Cl
2+
and Fe(CN)
6
4–
.
Nucleic Acids Research, 1999, Vol. 27, No. 24
4833
of the ruthenium complex along the grooves of the immobi-
lized helices, while the intercalated species may participate in
electron transfer mediated by the stacked bases. Therefore,
because single-base mismatches do not affect the overall struc-
ture of the DNA helix, but rather subtly perturb the electronics
of the DNA base stack, intercalated probes are needed for
detection schemes based on DNA-mediated charge transport.
Among the intercalators, the bulkier probes exhibit smaller
CA/TA charge ratios. This observation suggests that the
smaller intercalators more readily diffuse into the monolayer
and bind beneath the mismatch. All of the intercalators exhibit
linear current versus scan rate relationships (51), however,
indicating that the reactive species are strongly bound to the
surface and thus diffusion based reactivity is minimal. Never-
theless, for the detection of base mismatches using the direct
electrochemistry of molecules bound to DNA films, bulky
intercalators occupying sites close to the periphery of the well-
packed monolayer provide the most sensitive differentiation.
By eliminating a step in the preparation of samples, non-
covalently bound probes provide an important advance over
fluorescence-based sensors currently in use that require the
covalent attachment of a reporter group.
Variation in sequence composition.
The detection of base
mismatches using a charge transport-based assay appears to be
independent of DNA sequence context and composition. As
shown in Figure 3, the characteristic drop in coulometric
signals for DM bound to DNA films containing a single CA
mismatch compared to fully-paired films was essentially invar-
iant across AT-rich to GC-rich sequences tested under iden-
tical conditions. This sequence-independent response is not
achievable based upon differential hybridization, where the
detection of mismatches within these different oligonucle-
otides would require drastically different conditions. The
ability to detect a base mispair within a hybridized duplex
would therefore allow sequences of different base content to be
assayed simultaneously without the need for any manipulation
of conditions or readings. Indeed, this feature highlights a
unique and valuable aspect of the charge transport method-
ology.
Variation of mismatch.
This assay allows the detection of a
series of different mutations as demonstrated by monitoring
the electrochemical signals obtained from DM non-covalently
bound to films containing a variety of mismatches (Table 3). In
general,
pyrimidine–pyrimidine
and
purine–pyrimidine
mismatches caused marked decreases in the electrochemical
signals; the one purine–purine pair studied, a GA mismatch,
did not show a measurable diminution. Interestingly, photo-
physical studies of the effects of base mismatches on long-
range electron transfer through DNA also revealed insensi-
tivity to GA mispairs (37). This purine–purine pair may be
sufficiently well-stacked within the DNA helix to support effi-
cient electron transfer (44).
Surprisingly, a GT pair caused a substantial decrease in
current, although it is not highly disruptive to the helix (52).
T
a
bl
e
1
.
E
l
ectroc
h
em
i
ca
l
response o
b
ta
i
ne
df
rom DM non-cova
l
ent
l
y
bound to DNA-modified electrodes: dependence on mismatch location
a
a
Based on cyclic voltammograms measured for 1.0
μ
MDMnon-
covalently bound to duplex-modified electrodes. Values are based on
more than five trials each, and results were comparable for experiments
run side-by side, or from different sample preparations.
b
Integrated background-subtracted cathodic charge.
c
Determined by thermal denaturation measurements obtained by moni-
toring hypochromicity at 260 nm in duplex solutions containing 10
μ
M
duplex, 5 mM sodium phosphate, 50 mM NaCl and 100 mM MgCl
2
in
the absence of DM.
Table 2.
Electrochemical mismatch detection: dependence on
DNA-binding probe
a
a
Sequence: SH-5
′
-AGTACAGT
C
ATCGCG (CA mismatch located at
bold C). Based on cyclic voltammograms measured for various probes
non-covalently bound to duplex-modified electrodes. Values are based
on more than three trials each, and results were comparable for experi-
ments run side-by-side, or from different sample preparation.
b
See (49).
c
See (50).
d
See (Stinner
et al
., unpublished results).
e
See (48).
Figure 2.
Cyclic voltammetry of 1.0
μ
M DM at a gold electrode modified with
SH-5
′
-AGT
A
CAGTCATCGCG
hybridized
to
its
fully
base-paired
complement (TA) and a complement that features a C opposite the bold
A
(CA). Voltammograms were obtained with scan rate (
ν
) = 100 mV/s, electrode
area (A) = 0.02 cm
2
.
4834
Nucleic Acids Research, 1999, Vol. 27, No. 24
This wobble base pair, although thermodynamically stable,
appears to mediate electron transfer poorly. This effect could
result from increased base dynamics for the GT pair, or a
poorly stacked conformation assumed by this pair that is
unfavorable to electron transfer through DNA. Irrespective of
the mechanism by which these mismatches attenuate the
resultant electrochemical signals, the observation of significant
changes in the efficiency of charge transport in the presence of
these different pairs indicates that this approach can be gener-
ally applied for mismatch detection.
Mismatch detection via
in situ
hybridization
As would be required in an oligonucleotide array, mismatch
detection can also be achieved with sequences hybridized
reversibly
in situ
at the electrode surface. Thiol-modified
duplexes can be deposited on the gold surface, heat denatured,
thoroughly rinsed and then rehybridized with the desired target
by incubation with
≥
50 pmol of single-stranded oligonucle-
otide. This reversible assay is illustrated in Figure 4. Here, two
separate duplex-modified electrodes were prepared, each
containing the same 15-bp oligonucleotide derivatized with a
thiol-terminated linker. One electrode featured this oligonucle-
otide hybridized to its native complement, while the other was
modified with a duplex containing a CA mismatch. Once
immersed in DM solution, the electrodes exhibited electro-
chemical responses characteristic of fully base-paired and CA-
mutated films, respectively. The DNA films were stripped of
their complements by heat denaturation, yielding single-
stranded monolayers of identical sequence. New duplexes
were then formed by incubation of the electrodes with the
swapped complements (TA
→
CA, CA
→
TA). The electro-
chemistry of DM at the new films again showed the character-
istic behavior expected for fully base-paired and CA-mutated
duplexes. Electrodes can be cycled through this sequence of
events repeatedly.
It is noteworthy that the cyclic voltammetry of DM at single-
stranded films shows an irreversible and broad reduction,
which becomes smaller upon subsequent scans. These signals
are smaller than those obtained at surfaces modified with fully
base-paired duplexes, but larger than those observed at
Figure 3.
Charge obtained for DNA-modified electrodes in the presence o
f
1.0
μ
M DM. These duplexes, featuring varying percentages of GC content,
were either fully base paired, or contained a single CA mismatch. Regardless
of the sequence composition, and therefore over a wide range of duplex
stabilities (
∆
T
m
=50
°
C), the CA mismatches contained within these duplexes
were accurately detected. The melting temperatures for these duplexes, as
determined by thermal denaturation measurements obtained by monitoring
hypochromicity at 260 nm in duplex solutions containing 10
μ
M duplex, 5 mM
sodium phosphate, 50 mM NaCl and 100 mM MgCl
2
were (SH-5
′
-
ATATAATATATGGAT): TA = 47
°
C, CA = 32
°
C; (SH-5
′
-AGTACAGT-
CATCGCG): TA = 68
°
C, CA = 56
°
C; (SH-5
′
-GGCGCCCGGCGCCGG):
GC = 82
°
C, CA = 69
°
C. Charge was quantitated from integrating background-
subtracted cyclic voltammograms obtained at
ν
= 100 mV/s with A = 0.02 cm
2
.
Table 3.
Electrochemical response obtained from DM non-covalently
bound to DNA-modified electrodes: dependence on mismatch composition
a
a
Based on cyclic voltammograms measured for 1.0
μ
M DM non-cova-
lently bound to duplex-modified electrodes. Values are based on more
than five trials each, and results were comparable for experiments run
side-by-side, or from different sample preparation as long as electrodes
exhibited high surface coverages. Electrodes with lower surface cover-
ages yielded higher charges (>1 DM/duplex), and decreased attenua-
tions in the presence of mismatches.
b
Integrated background-subtracted cathodic charge.
c
Determined by thermal denaturation measurements obtained by moni-
toring hypochromicity at 260 nm in duplex solutions containing 10
μ
M
duplex, 5 mM sodium phosphate, 50 mM NaCl and 100 mM MgCl
2
in
the absence of DM.
Figure 4.
Charges (Q
c
) measured during the
in situ
detection of a CA mis-
match. Electrodes were derivatized with sequence SH-5
′
-AGTACAGT-
C
A
TCGCG, where either a C or T was incorporated into the complement
across from the italicized A. Using cyclic voltammetry, the electrochemical
response of 1.0
μ
M DM non-covalently bound to duplex-modified electrodes
was first measured for the intact TA- or CA-containing duplexes (
TA
/
CA
), then
the electrodes were immersed in 90
°
C pure buffer for 2 min, rinsed and the
charge was remeasured (
ss
). The electrodes were then incubated with 100 pmol
of the opposite complement in the presence of 5 mM phosphate buffer contain-
ing50mMNaCland100mMMgCl
2
. Upon completion of this hybridization,
the electrodes were rinsed and the charge was again remeasured. Finally,
electrodes were again heated and the response was quantitated.
Nucleic Acids Research, 1999, Vol. 27, No. 24
4835
analogous mismatched duplexes; electrochemistry can there-
fore be used to confirm
in situ
hybridization, but mutations are
ultimately identified at duplexes that feature single-base
mismatches. We believe these broad signals reflect the less
dense monolayer of single stranded oligonucleotides, which
allows DM access to the gold surface. Assays of genomic
DNA with small single stranded regions or multiple
mismatches likely would not hinder full monolayer formation
and thus should provide full coverage of the gold surface.
Mismatch detection using electrocatalysis
Although duplexes containing mismatches can be distin-
guished by direct voltammetry of redox-active intercalators,
the absolute electrochemical signals are limited by the surface
concentration of the intercalator (~50 pmol/cm
2
). In order to
increase the inherent sensitivity of this assay, we have coupled
the direct electron transfer to an electrocatalytic process
involving a species freely diffusing in solution (Fig. 5). This
effectively amplifies the intercalator signal and improves the
discrimination between signals obtained for mismatched
versus base-paired duplex films.
Methylene blue was chosen as the intercalated catalyst, with
potassium ferricyanide as the solution substrate. Possessing a
large negative charge, Fe(CN)
6
3–
is electroinactive at the DNA-
modified surface even at overpotentials as high as ~1 V (Fig.
6), yet its chemical reduction by reduced MB is thermodynam-
icallyfavoredby~0.6eV.Giventhelowreorganizationenergy
expected for this process (53), the cross reaction between the
electrochemically generated catalyst and substrate should be
very rapid. Depending on the rates of the various steps in the
overall reaction, the signals may now be limited by the concen-
tration of substrate in solution. Because the presence of
mismatches effectively decreases the amount of reduced inter-
calator bound to the film, the presence of mispairs should also
decrease signals obtained from electrocatalytic reactions.
Electrocatalytic reduction of methylene blue.
Addition
of
micromolar MB to a 2.0 mM ferricyanide solution causes a
pronounced electrochemical signal at the DNA-modified elec-
trode (Fig. 6). Notably, this signal comes at the reduction
potential of MB and is completely irreversible: electrons flow
from the Au electrode to intercalated MB and then are accepted
by Fe(CN)
6
3–
in solution (thus no electrochemical oxidation
peak is observed). Chemically oxidized MB is again available
for electrochemical reduction and the catalytic cycle continues
as long as the potential of the gold electrode is sufficiently
negative to reduce MB.
Although there is no requirement for the redox intercalator to
dissociate fully from the monolayer, electrocatalysis involving
intercalators bound to DNA-modified electrodes appears to
require a catalyst that can dynamically shuttle electrons to
solution-borne acceptors. DM is a very poor electrocatalyst
(data not shown) and exhibits only very low levels of catalytic
reduction in the presence of Fe(CN)
6
3–
.DMhasastronger
affinity for DNA (54) than does methylene blue (55) and may
have slower exchange dynamics that would not allow the
passage of electrons out to the acceptor.
Electrocatalytic detection of base mismatches.
Incorporation
of a mismatch into the duplex significantly attenuates the elec-
trocatalytic response obtained with methylene blue (Fig. 7).
Fewer MB molecules are reduced at the mismatched-DNA
electrode: the steady-state concentration of active catalyst is
lower, and a diminished overall catalytic rate results. A range
of catalyst and substrate concentrations was investigated to
maximize the difference in electrocatalytic response at the
fully base-paired (TA) and mismatched (CA) duplexes. Under
optimized conditions, the presence of the mismatch causes a
6-fold decrease in the electrocatalytic current, compared to a
2-fold decrease in the peak current obtained when monitoring
the direct electrochemistry of methylene blue (at scan rates of
100 mV/s). This enhancement is consistent with the idea that
electrocatalysis amplifies only the DNA-mediated charge
transport to the catalyst bound near the periphery of the film;
MB bound in the interior would not be effective as a catalyst.
Hence, coupling direct electrochemistry to a catalytic event
both increases the sensitivity of mismatch detection and
provides larger absolute signals.
In contrast, Ru(NH
3
)
5
Cl
2+
(a groove binder with approxi-
mately the same potential as MB) is an effective electrocatalyst
for the reduction of ferricyanide at DNA-modified surfaces,
but the catalytic signal obtained with this cation is not sensitive
to mismatches in the film (Fig. 7). These data are consistent
with the results of direct voltammetry of intercalated versus
non-intercalated probes (Table 2), and indicate that the
Figure 5.
Schematic representation of electrocatalytic reduction of Fe(CN)
6
3–
by MB at a DNA-modified electrode. LB
+
is leucomethylene blue, the product
of the electrochemical reduction.
Figure 6.
Cyclic voltammetry (
ν
= 100 mV/s, A = 0.02 cm
2
) at a gold electrode
modified with DNA (sequence: SH-5
′
-AGTACAGTCATCGCG) of 2.0 mM
Fe(CN)
6
3–
(black), 2.0
μ
M MB (blue), 2.0 mM Fe(CN)
6
3–
and 2.0
μ
MMB
(red).
4836
Nucleic Acids Research, 1999, Vol. 27, No. 24
detection of mismatches based on electrocatalysis also requires
a probe that interacts with the DNA through intercalation.
Because the charge transport-based assay features a catalytic
reaction whose rate depends on the degree of complementarity
within the individual duplexes, the measured charge resulting
from the reduction of methylene blue at TA versus CA-
containing films increases disproportionately with longer inte-
gration times (Fig. 8). Using 0.5
μ
MMBand2.0mMferri-
cyanide, 10 s potential steps to –350 mV gave faradaic charges
of 36 and 6
μ
C respectively. Increased sampling times
continue to increase the differentiation of signals obtained with
mismatched versus paired complements. These results high-
light the versatility of electrochemical detection methods that
are also more amenable to the portability required for a prac-
tical device.
Advantages and disadvantages of sensing based on
DNA-mediated electron transfer
We have presented results indicating that the sensitivity of
electron transfer through DNA films to intervening base
mismatches provides a suitable means to probe DNA
sequences for the presence of mutations. Several features of
this new approach provide important advances over existing
technology, including (i) an insensitivity of the detection of
single-base mismatches to AT or GC sequence content, (ii) the
detection of variety of mismatches (including some with
comparable thermodynamic stability to Watson–Crick pairs),
and (iii) the use of non-covalently bound probes which elimi-
nate the need for chemical modification of biological samples.
The detection of mutations based on DNA-mediated electron
transfer therefore provides a complementary method to hybrid-
ization-based assays. The ability to detect mutations within
intact duplexes would greatly simplify the analysis of multiple
test sequences at an addressable array. Analyses could be
performed under strongly hybridizing conditions, allowing
both native and mutated test strands to bind to the probe
sequences, regardless of the overall base composition of the
individual addresses in the array. In addition, electrochemical
detection methods are better suited for the development of
inexpensive, portable devices than the sensors currently avail-
able employing high-resolution confocal microscopy (17).
However, electrochemical detection presents new challenges
for the fabrication of functional DNA ‘chips’ on electrode
surfaces. The reproducibility and effectiveness of the assay
presented here requires tightly packed films that inhibit the
diffusion of intercalators into the monolayer. Furthermore, the
applicability of our potentiometric and amperometric methods
have not been proven with biological samples. While the
charge transport-based method described here holds great
promise for the detection of known mutations within defined
sequences, hybridization based methods combined with algo-
rithmic analyses may be better employed for sequencing
assays. We have not yet systematically varied the DNA length
on the electrode surface, but an 8
×
8 array of 15mers should be
sufficient to assay for a typical gene with no redundancies.
Overall, charge transport and hybridization-based assays may
then ultimately provide complementary methods for the anal-
ysis of DNA sequence composition and abnormalities.
The efficiency of charge transport through DNA films offers
a new approach to DNA-based sensors. Using this method-
ology, a broad range of point mutations can be detected within
heterogeneous DNA sequences, irrespective of base composi-
tion. Monitoring electrochemical signals at addressable elec-
trodes, as opposed to detecting fluorescence by high-resolution
microscopy or radioactive labelling, may provide a practical
detection system for inexpensive devices to search for known
Figure 7.
Cyclic voltammograms (
ν
= 100 mV/s, A = 0.02 cm
2
)of 2.0 mM
Fe(CN)
6
3–
plus 2.0
μ
M MB (top) or 28
μ
M Ru(NH
3
)
5
Cl
2+
(bottom) at a gold
electrode modified with the thiol-terminated sequence SH-5
′
-AGTACAGT-
C
ATCGCG hybridized to a fully base-paired complement (red) and a comple-
ment that features an A opposite the bold
C
(black).
Figure 8.
Chronocoulometry at –350 mV of 2.0 mM Fe(CN)
6
3–
plus 0.5
μ
M
MB (pH 7) at a gold electrode modified with the thiol-terminated sequence
SH-5
′
-AGTACAGT
C
ATCGCG hybridized to a fully base-paired complement
(red) and a complement that features an A opposite the bold
C
(black). The
discrimination between base-paired and mismatched sequences increases with
increased sampling time.
Nucleic Acids Research, 1999, Vol. 27, No. 24
4837
mutations on targeted genes. While others have explored elec-
trochemical schemes for the development of DNA biosensors,
the reliance of these schemes on hybridization assays does not
offer the same advantages as a charge transport-based
approach. The discovery that DNA-mediated electron-transfer
reactions are exquisitely sensitive to the stacking of the inter-
vening bases has provided insight into the role of the DNA
base stack in modulating this reactivity. As a result, we can
now exploit this sensitivity to stacking in the development of a
practical assay for single-base changes in DNA sequence.
ACKNOWLEDGEMENTS
We wish to thank Professor F. A. Anson for several helpful
discussions. This work was supported by the Camile and
Henry Dreyfus Foundation (Faculty Start-up Grant to M.G.H.),
the Research Corporation (M.G.H.), and the National Institutes
of Health (GM49216 to J.K.B.).
REFERENCES
1. Skogerboe,K.J. (1995)
Anal. Chem.
,
67
, 449R–454R.
2. Fodor,S.P.A. (1997)
Science
,
277
, 393.
3. Southern,E.M. (1996)
Trends Genet.
,
12
, 110–115.
4. Eng,C. and Vijg,J. (1997)
Nature Biotechnol.
,
15
, 422–426.
5. Golz,S., Birkenkampdemtroder,K. and Kemper,B. (1998)
Nucleic Acids
Res.
,
26
, 1132–1133.
6. Bernard,P.S., Lay,M.J. and Wittwer,C.T. (1998)
Anal. Biochem.
,
255
,
101–107.
7. Osborne,R.J., Merlo,G.R., Mitsudomi,T., Venesio,T., Liscia,D.S.,
Cappa,A.P.M. and Chiba,I. (1991)
Cancer Res.
,
51
, 6194–6198.
8. Jackson,B.A. and Barton,J.K. (1997)
J. Am. Chem. Soc.
,
119
,
12986–12987.
9. Yamaguchi,M., Dao,V. and Modrich,P. (1998)
J. Biol. Chem.
,
273
,
9197–9201.
10. Ehsani,A., Low,J., Wallace,R.B. and Wu,A.M. (1993)
Genomics
,
15
,
426–429.
11. Orita,M., Suzuki,Y., Sekiya,T. and Hayashi,I. (1989)
Genomics
,
5
,
874–879.
12. Okahata,Y., Matsunobu,Y., Ijiro,K., Mukae,M., Murakami,A. and
Makino,K. (1992)
J. Am. Chem. Soc.
,
114
, 8299–8300.
13. Wang,J., Nielsen,P.E., Jiang,M., Cai,X.H., Fernandes,J.R., Grant,D.H.
and Ozsoz,M. (1997)
Anal. Chem.
,
69
, 5200–5202.
14. Bardea,A., Dagan,A., Bendov,I., Amit,B. and Willner,I. (1998)
Chem.
Commun.
, 839–840.
15. Chen,Y.H., Song,J.D. and Li,D.W. (1997)
Sci. China Ser. C
,
40
, 463–469.
16. Drmanac,R., Drmanac,S., Strezoska,Z., Paunesku,T., Labat,I.,
Zeremski,M., Snoddy,J., Funkhouser,W.K., Koop,B., Hood,L. and
Crkvenjakov,R. (1993)
Science
,
260
, 1649–1653.
17. Chee,M., Yang,R., Hubbell,E., Berno,A., Huang,X.C., Stern,D.,
Winkler,J., Lockhart,D.J., Morris,M.S. and Fodor,S.P.A. (1996)
Science
,
274
, 610–614.
18. Theil,A.J., Frutos,A.G., Jordan,C.E., Corn,R.M. and Smith,L.M. (1997)
Anal. Chem.
,
69
, 4948–4956.
19. Isola,N.R., Stokes,D.L. and Vodinh,T. (1998)
Anal. Chem.
,
70
,
1352–1356.
20. Lin,V.S.Y., Motesharei,K., Dancil,K.P.S., Sailor,M.J. and Ghadiri,M.R.
(1997)
Science
,
278
, 840–843.
21. Gotoh,M., Hasebe,M., Ohira,T., Hasegawa,Y., Shinohara,Y., Sota,H.,
Nakao,J. and Tosu,M. (1997)
Genet. Anal. Biomol. E
,
14
, 47–50.
22. Elghanian,R., Storhoff,J.J., Mucic,R.C., Letsinger,R.L. and Mirkin,C.A.
(1997)
Science
,
277
, 1078–1081.
23. Healey,B.G., Matson,R.S. and Walt,D.R. (1997)
Anal. Biochem.
,
251
,
270–279.
24. Millan,K.M. and Mikkelsen,S.R. (1993)
Anal. Chem.
,
65
, 2317–2323.
25. Millan,K.M., Saraullo,A. and Mikkelsen,S.R. (1994)
Anal. Chem.
,
66
,
2943–2948.
26. Xu,X.H. and Bard,A.J. (1995)
J. Am. Chem. Soc.
,
117
, 2627–2631.
27. Xu,X.H., Yang,H.C., Mallouk,T.E. and Bard,A.J. (1994)
J. Am. Chem.
Soc.
,
116
, 8386–8387.
28. Johnston,D.H., Glasgow,K.C. and Thorp,H.H. (1996)
J. Am. Chem. Soc.
,
117
, 8933–8938.
29. Napier,M.E., Loomis,C.R., Sistare,M.F., Kim,J., Eckhardt,A.E. and
Thorp,H.H. (1997)
Bioconjugate Chem.
,
8
, 906–913.
30. Wang,J., Palecek,E., Nielsen,P.E., Rivas,G., Cai,X.H., Shiraishi,H.,
Dontha,N., Luo,D.B. and Farias,P.A.M. (1996)
J. Am. Chem. Soc.
,
118
,
7667–7670.
31. Wang,J., Cai,X.H., Rivas,G., Shiraishi,H., Farias,P.A.M. and Dontha,N.
(1996)
Anal. Chem.
,
68
, 2629–2634.
32. Palanti,S., Marrazza,G. and Mascini,M. (1996)
Anal. Lett.
,
29
,
2309–2331.
33. Liu,S.H., Ye,J.N., He,P.G. and Fang,Y.H. (1996)
Anal. Chim. Acta
,
335
,
239–243.
34. DeLumley-Woodyear,T., Campbell,C.N. and Heller,A. (1996)
J. Am.
Chem. Soc.
,
118
, 5504–5505.
35. Hashimoto,K., Ito,K. and Ishimori,Y. (1994)
Anal. Chim. Acta
,
286
,
219–224.
36. Holmlin,R.E., Dandliker,P.J. and Barton,J.K. (1998)
Angew. Chem. Int.
Ed. Eng.
,
36
, 2714–2730.
37. Kelley,S.O., Holmlin,R.E., Stemp,E.D.A. and Barton,J.K. (1997)
J. Am.
Chem. Soc.
,
119
, 9861–9870.
38. Kelley,S.O. and Barton,J.K. (1998)
Chem. Biol.
,
5
, 413–425.
39. Kelley,S.O. and Barton,J.K. (1999)
Science
,
283
, 375–381.
40. Kelley,S.O., Jackson,N.M., Barton,J.K. and Hill,M.G. (1997)
Bioconjugate Chem.
,
8
, 31–37.
41. Kelley,S.O., Jackson,N.M., Hill,M.G. and Barton,J.K. (1999)
Angew.
Chem.
,
38
, 941.
42. Hall,D.B. and Barton,J.K. (1997)
J. Am. Chem. Soc.
,
119
, 5045–5046.
43. Hunter,W.N., Brown,T. and Kennard,O. (1987)
Nucleic Acids Res.
,
15
,
6589–6606.
44. Brown,T., Hunter,W.N., Kneale,G. and Kennard,O. (1986)
Proc. Natl
Acad. Sci. USA
,
83
, 2402–2406.
45. Kelley,S.O., Barton,J.K., Jackson,N.M., McPherson,L.D., Potter,A.B.,
Spain,E.M., Allen,M.J. and Hill,M.G. (1998)
Langmuir
,
14
, 6781–6784.
46. Ihara,T., Takata,J. and Takagi,M. (1998)
Anal. Chim. Acta
,
365
, 49–54.
47. Harrison,J.G. and Balasubramanian,S.A. (1997)
Bioorg. Med. Chem.
Lett.
,
7
, 1041–1046.
48. Karthe,P. and Gautham,N. (1998)
Acta Crystallogr.
,
54
, 501–509.
49. Wang,A.H.J., Ughetto,G., Quigley,G.J. and Rich,A. (1987)
Biochemistry
,
26
, 1152–1163.
50. Becker,H.C. and Norden,B. (1997)
J. Am. Chem. Soc.
,
119
, 5798–5803.
51. Bard,A.J. and Faulkner,L.R. (1980)
Electrochemical Methods
.Wiley&
Sons, NY.
52. Patel,D.J., Kozlowski,S.A., Ikuta,S. and Itakura,K. (1984)
FASEB
,
11
,
2663–2670.
53. Marcus,R.A. and Sutin,N. (1985)
Biochim. Biophys. Acta
,
811
, 265–322.
54. Chaires,J.B., Dattagupta,N. and Crothers,D.M. (1982)
Biochemistry
,
21
,
3933–3940.
55. Tuite,E. and Kelly,J.M. (1995)
Biopolymers
,
35
, 419.