Supplementary Information for
Single-Molecule Electrical Random Resequencing
of DNA and RNA
Takahito Ohshiro, Kazuki Matsubara, MakusuTsutsui, Masayuki Furuhashi,
Masateru Taniguchi & Tomoji Kawai
The Supplementary Information includes:
1. Supplementary Text.
2. Supplementary Figures (Figs. S1–S14).
3. Supplementary Table (Tables S1-S3).
4. Supplementary References.
1. Supplementary Text.
Calibration of gap distance
In the nucleotide and oligonucleotide electrical measurements, we used an optimal
gap distance for the nanogap electrodes. This gap distance was experimentally found
to be 0.8 nm, which is comparable to the size of the nucleotide molecules. A gap
distance of 0.8 nm was reproducibly set according to the following procedure.
First, a lithographically fabricated gold nanowire on a thin polyimide-coated
phosphorous bronze substrate was broken by mechanically bending the substrate at
300 K in air. A scanning electron microscope (SEM) image of a typical fabricated
MCBJ is shown in Fig. S1. The fabrication procedures for the lithographically
defined MCBJs are described in detail in our previous reports.S1
Second, after reconnecting the gold nanowire, a constant dc bias voltage of 0.1 V
was applied, and the nanowire substrate was gradually bent using a piezoactuator.
Throughout the junction breaking process, the junction conductance (G) was
monitored using a picoammeter (Keithley 6487). A series of conductance jumps on
the order of G0 = 2e2/h (e and h are the electron mass and Planck’s constant,
respectively) was observed, and the final conductance value was 1 G0. (Fig. S2)
Third, several seconds after the 1 G0 state was obtained for the gold nanowire, the
one gold atom contact was naturally ruptured, resulting in a pair of electrodes. The
gap size (dG) was found to be 0.5 nm in our previous reports.S1 In this study, the gap
size (dG) was estimated as follows: First, we measured the incremental values of the
piezovoltage for the piezoactuator by closing the junction at 1 G0 (ΔVp), which was
linearly proportional to the change in the gap distance controlled by the
piezoactuator: ΔdG (ΔdG = αVp with the conversion efficient ). Next, we measured
the tunneling current, I, in the gap when the junction was closed on each MCBJ, and
calibrated the displacement using Vp by examining the Vp dependence of the
tunneling current I. In theory, I is described as an exponential function of the gap
distance dG as I exp (βdG) with the decay constant 4 2m h , where m, φ, and
h are electron mass (9.11 × 10−32 kg), the work function of gold (5.3 eV), and
Planck’s constant, respectively. From this Vp log I plot result, we found that the
piezovoltage (Vp) was linearly proportional to log I: log I ≈ γVp, where γ is a
proportional constant. From the experimentally determined γ, the formed gap
distance dG was found to be ~0.5 nm.
Finally, upon increasing the gap distance (ΔdG) controlled by the piezovoltage,
(ΔdG = αVp), the electrode gap distance was increased to 0.8 nm, which was used for
the sample nucleotide measurements. The gap distance formed was found to be 0.79
± 0.05 nm, as analyzed by at least 50 cycles of dG measurements of ten MCBJ
devices (Fig. S3). The fact that the gap distance of 0.83 ± 0.08 nm after the
experiments was comparable to the formed gap distance (Fig. S3) indicated that the
nanogap electrode was stable during the electrical measurements. Although electrode
gaps far from the value of 0.8 nm were sometimes formed, such extraordinary gaps
can be readily excluded because the gaps showed abnormal conductance profiles
during the self-breaking process.
Control experiment for single nucleotide and oligonucleotide solutions
To investigate the influence of water (Milli-Q) and buffer molecules on the current
profiles in DNA and RNA solutions, we performed electrical measurements in air,
Milli-Q, and buffer solutions. These control experiments were performed at 300 K
under conditions same as those for measuring the DNA and RNA solutions. The data
acquisition time was 15 min per experiment. The typical I–t profiles in air, Milli-Q, and
buffer solutions are shown in Fig. S4. The observed current values were stable during
the measurements in these control solutions. However, there were no spike-like signals
as observed in the DNA and RNA sample solutions.
We found that the background current values increased with the bias voltage between
the nanogap electrodes (Fig. S5) and the salt concentration (Fig. S6). The voltage
dependence and salt concentration dependence of the currents indicate that the
background currents originate from ionic currents.
Control experiment for hetero-base oligonucleotides
We performed electrical measurements of homo-base polymers (dAAA, dCCC,
dGGG, and dTTT) and control experiments for the hetero-base oligonucleotides by
using the nanogap electrodes. All four homo-base oligonucleotides tended to give
single-level current steps, whereas hetero-base nucleotide signals often exhibited a
multi-level step-like feature (see the left-side histogram of the I–t profiles in Figs. S9
and S11). These results suggest that the number of conductance levels during one
current spike signal somehow reflects the number of base types in the oligonucleotides.
Nevertheless, this does not mean that we can completely exclude the possible influence
of molecular conformation change on the I–t profiles, which could contribute to the
step-like feature in the I–t curves obtained for the hetero-base oligonucleotides.
However, as far as we have measured, multi-level current steps were rarely obtained in
the homo-base oligonucleotides. Therefore, it is expected that the influence of
molecular conformations on the current signals may be causing an error in determining
the base sequence; however, these effects are only marginal.
Detailed base assignment procedure
To clarify the detailed base assignment procedure, we demonstrated the procedure
for GTG (Fig. S13). First, we measured the GTG molecular signals in a 0.1 μM
solution with 10 MCBJ samples. The average number of signals was around 200 for
each MCBJ samples. Second, we constructed current histograms using all the data
obtained from the I–t profiles. Third, we determined the current baseline on the basis
of the lowest conductance peak (not shown here). Fourth, signals with long td > 2 ms
(yellow colored signals in Fig. S13) were extracted. Fifth, conductance histograms
were constructed from all data points of the corresponding signals. Sixth, the raw
signals were smoothed. Seventh, we calculated the single-molecule conductance on
the basis of the current baseline. Eighth, the relative single-molecule conductance
was calculated. Finally, we determined the base molecule type by using the relative
single-molecule conductance of the base molecule database (Table 1).
The resequenced base species for each signal region represents a fragment of the
nucleotide base molecules just passing through the electrode gap. In this way, we can
read fragments of sample nucleotide sequences. For example, for GTG I–t profiles
containing 758 molecular signals for an experiment, we applied this base assignment
procedure and read the nucleotide-fragment information. Among 758 molecular
signals, 698 signals were assigned: 140 to G , 71 to T, 92 to GT, 142 to GTG, 48 to
TGT, 23 to GTGT, 23 to TGTG , 34 to GTGTG, 8 to GTGTGT, 6 to TGTGTG, 6 to
GTGTGTG, and 2 to GTGTGTGTG (Fig. S13 and Table S3). However, 59 signals
could not be assigned in this analysis because of the large current fluctuations within
the 2 ms period, which was used for the nucleotide retention time. The “correct” read
(GTG) rate was 18.7%, and the other read signals reflect the short “partial transit”
reads (G, T, GT, TG) and duplication reads (GTGT, TGTG, GTGTG, GTGTGT,
TGTGTG, GTGTGTG, and GTGTGTGTG), which are caused by the stochastic
translocation of the nucleotide molecules. Compared with the GTG I–t profile,
92.7% of the signals in a GGG I–t profile could be assigned to the trinucleotide
molecules containing only G (Fig. S12 and Table S2).
2. Supplementary Figures.
Figure S1. Formation of electrode gaps using a nanofabricated MCBJ. (a) A
schematic illustration of the experimental set up (left). An MCBJ sample was
loaded on the two counter supports. The right panel shows a scanning electron
microscope image of a nanofabricated MCBJ, which consists of a free-standing
Au junction fabricated on a polyimide surface. (b) A magnified view of the Au
bridge before mechanical breakage. The narrowest constriction has a
cross-section of 100 nm × 100 nm. (c) Three-point bending of the sample
substrate leads to mechanical breakdown of the Au microbridge. The electrode
gap thus formed was finely adjusted to 0.8 nm and used for transverse current
measurements of single-molecule DNA.
Figure S2. Piezovoltage–conductance (Vp–G) curve during the junction
breaking process. These experiments were performed by applying a bias
voltage of 0.1 V to the nanogap electrodes at 300 K. In the Vp–G curve, ΔVp was
found to be 11 V. A decrease in the piezovoltage corresponding to pushing the
piezoactuator up increases the nanogap, whereas an increase in the
piezovoltage corresponding to pushing the piezoactuator down decreases the
nanogap. At the breaking point (25 V), the metal wire is fractured spontaneously.
On the other hand, a metal wire is reconnected at the reconnecting point (36 V).
The nanogap can be calculated from the piezovoltage.
Figure S3. Gap distance (dG) histograms before (red) and after (black) the
electrical measurements. The electrical measurements were performed for 15
min. These gap formation processes were performed in Milli-Q. The gap
distance peaks in the histograms were found to be 0.8 nm. The histograms were
constructed using at least 50 dG measurements obtained from 10 MCBJ
Figure S4. Current–time profiles in several measuring environments.
Electrical measurements were performed in air (black), vacuum (red), 100 mM
Tris-HCl (green), 1 x PBS (purple), and Milli-Q (dark blue). These experiments
were performed at a bias voltage of 0.7.
Figure S5. Voltage dependence of the current–time profiles. Electrical
measurements were performed in 100 mM of Tris-HCl (pH 7.4).
Figure S6. Salt concentration dependence of the current–time profiles.
Electrical measurements were performed in Milli-Q containing NaCl at a bias
voltage of 0.1 V.
Figure S7. Conductance (G)–time profiles of 3 deoxyribonucleic acids of
DNA and 4 ribonucleic acids of RNA. G–t data of (a) dAMP, (b) dCMP, (c)
dTMP, (d) rAMP, (e) rCMP, (f) rUMP and (g) rGMP.
Figure S8. Typical duration of the current (td) histograms of dG, dGG, dGGG,
and dGGGG. (a) All histograms were constructed from 1000 td data obtained
using 10 MCBJ samples. (b) Typical dGGG signals showed short td (gray) and
long td (orange).
Figure S9. Current–time raw and processed data for oligonucleotides.
Current–time profiles of (a) ATA, (b) GTG, (c) CAC, and (d) GAG. The blue, red,
purple, and green bands represent the relative single-molecule conductance
bands for dGMP, dAMP, dCMP, and dTMP, respectively. Raw data were
processed using the resequencing procedure.
Figure S10. Conductance histograms of 10 dGGG samples. Each
conductance histogram was constructed from the corresponding current–time
profiles. #n represents the sample number. Although the conductance
histograms vary depending on the type of sample used, we could obtain
statistically significant conductance histograms by using more than 10 samples.
Figure S11. Conductance histograms and current–time (I–t) profiles of
homo-base trinucleotides. The left and right panels show the conductance
histograms and I–t profiles of (a) dGGG, (b) dAAA, (c) dCCC, and (d) dTTT. The
conductance histograms were constructed from full data points obtained from
the corresponding right panels (I–t profiles). Each low- and high-conductance
peak observed in each histgogram corresponds to the base current and
single-molecule conductance of the base molecules.
Figure S12. Detailed resequencing procedure for dGGG. We measured
dGGG molecular signals in a 0.1-μM solution using 10 MCBJ samples. The
average number of signals was about 100 for each MCBJ sample. (a) Typical
raw I–t profile data. (b) Enlarged I–t profile in (a). We constructed current
histograms using all data obtained from the I–t profiles. Next, we determined the
current baseline on the basis of the lowest conductance peak. Signals with long
td > 2 ms (yellow colored signals) were extracted. (c) Enlarged extracted signals
in (b) and the conductance histogram constructed from all data points of the
corresponding signals. (d) Raw signals (black) and smoothed signals (red). (e)
Base-assigned smoothed signals. We calculated single-molecule conductance
on the basis of the current baseline. Next, the relative single-molecule
conductance was calculated. Finally, we determined the base molecule type
using the relative single-molecule conductance from the base molecule
database (Table 1). (f) Final data to be processed by the resequencing
Figure S13. Detailed resequencing procedure for dGTG. We measured GTG
molecular signals in a 0.1 μM solution using 10 MCBJ samples. The average
number of signals was about 200 for each MCBJ sample. (a) Typical raw I–t
profile data. (b) Enlarged I–t profile in (a). We constructed current histograms
using all the data obtained from the I–t profiles. Next, we determined the current
baseline on the basis of the lowest conductance peak (not shown here). Signals
with long td > 2 ms (yellow colored signals) were extracted. (c) Enlarged
extracted signals in (b) and the conductance histogram constructed from all data
points of the corresponding signals. (d) Raw signals (black) and smoothed
signals (red). (e) Base-assigned smoothed signals. We calculated
single-molecule conductance on the basis of the current baseline. Next, the
relative single-molecule conductance was calculated. Finally, we determined the
base molecule type using the relative single-molecule conductance of the base
molecule database (Table 1). (f) Final data to be processed using the
Figure S14. Current histograms, current–time raw and processed data for
random fragments of 5′-UGAGGUA-3′. Conductance histograms, current–time
raw data, and processed data for (a) UGA, (b) UAGA, (c) UGAGG, and (d)
UAGAGG. Raw data were processed using the resequencing procedure.
3. Supplementary Table.
Table S1. Electronic structures of base molecules. In an attempt to
determine the electronic states of base molecules, their molecular structures
were completely optimized by density functional theory (B3LYP) using 6–31+G
(d, p) basis sets. All calculations were performed using the Gaussian 03
HOMO (eV) LUMO (eV) Gap (eV)
A −5.90 −0.44 5.46
C −6.14 −0.78 5.36
G −5.66 −0.16 5.50
T −6.57 −1.03 5.54
U −6.88 −1.19 5.69
Table S2. Base assignment of a GGG I–t profile. We performed electrical
measurements of a GGG aqueous solution using 10 MCBJ samples. The
number of observed signals was approximately 100 per MCBJ sample in a 0.1
μM GGG aqueous solution. We performed a base-assignment procedure for
each GGG I–t profile, which contained 339 signals for a MCBJ sample. Among
339 signals, 312 signals were assigned as “GGG,” comprising signals that have
only a single-current level. However, 27 signals could not be assigned to base
molecules because of the multi-current fluctuations shown in the signals
surrounded by brown lines in Fig. S12a. These signals were classified as
“Others” in this table.
Number of Pulse Percentage (%)
GGG 312 92.0
Others 27 8.0
Table S3. Base assignment for a GTG I–t profile. We performed electrical
measurements of a GTG aqueous solution using 10 MCBJ samples. The
number of observed signals was approximately 200 per MCBJ sample in a 0.1
μM GTG aqueous solution. To show an example of GTG signal information and
its distribution in the I–t profiles, we performed a base assignment procedure for
a typical GTG I–t profile, which contained 758 signals for a MCBJ sample.
Among 758 signals, 698 signals were assigned to “G,” “T,” “GT,” “GTG,” “TGT,”
“GTGT” “TGTG,” “GTGTG,” “GTGTGT,” “TGTGTG,” “GTGTGTG,” and
“GTGTGTGTG. However, 59 signals could not be assigned to the base
molecules because of the large current fluctuations. These signals were
classified as “Others” in this table.
Number of Pulse Percentage (%)
others 59 7.8
4. Supplementary Reference.
S1. Tsutsui, M., Taniguchi, M. & Kawai, T. Fabrication of 0.5 nm electrode gaps using
self-breaking technique. Appl. Phys. Lett. 93, 163115 (2008).
S2. Gaussian03, revisionC.02; M. J. Frisch et al., Gaussian, Inc., Pittsburgh PA, 2003.