of 35
DNA Charge Transport: From Chemical Principles to the Cell
Anna R. Arnold
1
,
Michael A. Grodick
1
, and
Jacqueline K. Barton
*
Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena
CA 91125
Abstract
The DNA double helix has captured the imagination of many, bringing it to the forefront of
biological research. DNA has unique features that extend our interest into areas of chemistry,
physics, material science and engineering. Our laboratory has focused on studies of DNA charge
transport (CT), wherein charges can efficiently travel long molecular distances through the DNA
helix while maintaining an exquisite sensitivity to base pair
π
-stacking. Because DNA CT
chemistry reports on the integrity of the DNA duplex, this property may be exploited to develop
electrochemical devices to detect DNA lesions and DNA-binding proteins. Furthermore, studies
now indicate that DNA CT may also be used in the cell by, for example, DNA repair proteins, as a
cellular diagnostic, in order to scan the genome to localize efficiently to damage sites. In this
review, we describe this evolution of DNA CT chemistry from the discovery of fundamental
chemical principles to applications in diagnostic strategies and possible roles in biology.
Introduction
DNA is considered as the repository for genetic information in the cell. Structurally,
individual strands of DNA consist of a phosphate-deoxyribose backbone connecting
nitrogenous bases, either purines (adenine and guanine) or pyrimidines (cytosine and
thymine). The nitrogenous bases of DNA are composed of aromatic rings with base pairs
spaced 3.4 Å in the double helix. This structure allows the electron density of adjacent bases
to overlap, resulting in
π
-stacking and a large measure of stabilization to the double helix.
Significant structural similarity exists between stacked base pairs in DNA and the
z
-
direction of graphite, a known conductive material. Specifically, adjacent graphene sheets
are spaced at 3.35 Å (
Chaban et al., 2014
). The similar spacing of aromatic moieties in
graphite and DNA led researchers to hypothesize that DNA could also be conductive
through the overlap of adjacent
π
-orbitals (Figure 1). Decades of research have now
established that DNA can indeed conduct charge efficiently through the
π
-stack of the
nitrogenous bases (
Grodick et al., 2015
;
Elias et al., 2008
). Thus in addition to its role as the
*
to whom correspondence should be addressed at jkbarton@caltech.edu.
1
present address: Intel Corporation, Hillsboro OR 97124
Publisher's Disclaimer:
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our
customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of
the resulting proof before it is published in its final citable form. Please note that during the production process errors may be
discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
HHS Public Access
Author manuscript
Cell Chem Biol
. Author manuscript; available in PMC 2017 January 21.
Published in final edited form as:
Cell Chem Biol
. 2016 January 21; 23(1): 183–197. doi:10.1016/j.chembiol.2015.11.010.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
repository for genetic information, electrons as well as electron holes are readily transported
through the DNA
π
-stack (
Genereux and Barton, 2010
).
As a macromolecular assembly in solution, however, DNA differs from graphite and other
π
-stacked solids. The bases of DNA are constantly undergoing dynamic motion on
timescales from picoseconds to milliseconds, and these motions, though subtle, are
sufficient to facilitate or interrupt CT. DNA CT is attenuated by large and small
perturbations in
π
-stacking. Single base mismatches, base modifications and lesions, and
even protein binding events that kink the DNA duplex or flip out a base are sufficient to
interrupt DNA CT (
Boal and Barton, 2005
;
Gorodetsky et al., 2008a
;
Hall and Barton, 1997
;
Kelley et al., 1997a
,
1999
). Interestingly, nicks in the phosphate-deoxyribose backbone are
tolerated as long as the sugar-backbone modifications do not interfere with base-base
stacking (
Liu and Barton, 2005
). DNA CT thus reports on the integrity of the base pair
stack.
Our laboratory has focused on studies of DNA charge transport (CT), beginning with
understanding the basic chemistry and the parameters that govern charge transport, and
moving towards understanding how this chemistry may be harnessed within the cell. Here,
we describe that evolution. DNA CT represents powerful chemistry that permits redox
reactions to be activated over long molecular distances, enabling sensing of small
perturbations to the DNA base pair stack with high sensitivity, and potentially providing a
means to communicate across the genome.
Platforms for Exploring DNA Charge Transport
Various platforms for investigating DNA CT have been developed. Two of the most
effective platforms consist of photoinduced reactions using DNA assemblies in solution with
tethered donors and/or acceptors (Figure 2) and ground state electrochemistry on DNA
monolayers (Figure 3). With these very different platforms, the important aspects of DNA
CT chemistry are evident: (i) that DNA CT can proceed over long molecular distances; (ii)
that DNA CT is sensitive to intervening perturbations in
π
-stacking.
Photoinduced DNA CT with tethered intercalators
For photoinduced CT experiments in solution, we have utilized DNA assemblies with
tethered photooxidants containing an intercalating ligand that allows for
π
-stacking and thus
electronic coupling between the photooxidant and the DNA bases (Figure 2). Irradiation of
the photooxidant typically produces an excited state that is sufficiently oxidizing and long-
lived to withdraw an electron from DNA. Examples of metallointercalators that have been
used to probe the redox properties of DNA include [Rh(phi)
2
(bpy
)]
3+
, [Ru(phen)(bpy
)
(dppz)]
2+
, [Re(CO)
3
(dppz)(py
)]
+
, and [Ir(ppy)
2
(dppz
)]
+
, where phi = 9,10-
phenanthrenequinone diimmine, bpy
= 4-methyl-4
-(butyric acid)-2,2
-bipyridine, phen =
1,10-phenanthroline, dppz = dipyrido[2,3-a:2
,3
-c]phenazine, py
= 3-(pyridin-4-yl)-
propanoic acid, ppy = 2-phenylpyridine, and dppz
= 6-(dipyrido[3,2-a:2
,3
-c]phenazin-11-
yl)hex-5-ynoic acid) (
Olmon et al., 2011
;
Shao and Barton, 2007
;
Williams et al., 2004
).
The yield of oxidative DNA damage produced by metallointercalators has been found to
depend primarily on the thermodynamic driving force for CT, the efficiency of back electron
Arnold et al.
Page 2
Cell Chem Biol
. Author manuscript; available in PMC 2017 January 21.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
transfer (ET) processes, and, importantly, the degree of electronic coupling to the DNA
π
-
stack (
Olmon et al., 2011
). These complexes can be covalently tethered to DNA by utilizing
modified ligands (dppz
, bpy
, phen
) in order to localize the complex to one end of the DNA
(
Holmlin et al., 1999
).
An early example of the application of metallointercalators to the study of DNA CT
employed an intercalating donor, a dppz complex of Ru(II), and as acceptor, a phi complex
of Rh(III), both of which were tethered to either ends of the DNA duplex (
Murphy et al.,
1993
). The dppz complex is a DNA light-switch: whereas its luminescence is quenched in
aqueous solution, upon the addition of DNA, the ruthenium complex luminesces brightly
(
Friedman et al., 1990
). However, when [Rh(phi)
2
(phen
)]
3+
is appended to the opposite
strand, the DNA-bound [Ru(phen
)
2
(dppz)]
2+
luminescence is completely quenched
(
Murphy et al., 1993
). Nanosecond time-resolved luminescence experiments were too slow
to observe this quenching process. Extensive control experiments, including those that ruled
out intermolecular events, coupled with the improbability of energy transfer, confirmed that
this quenching is due to rapid intramolecular DNA-mediated electron transfer between the
metal complexes, from the *Ru
2+
excited state to the rhodium complex. This experiment
provided the first insights into long-range DNA charge transport.
Oxidative DNA damage through DNA CT
Besides examining DNA CT spectroscopically, we also used DNA-bound photooxidants to
explore oxidative damage to the DNA from a distance using biochemical methods. Here the
electron donor was the DNA itself, specifically the 5
-G of guanine doublets and triplets.
Because guanine is the most easily oxidized of the DNA bases (
Fukuzumi et al., 2005
) and
the presence of adjacent guanines further lowers the guanine oxidation potential (
Sugiyama
and Saito, 1996
), guanine doublets and triplets are electron hole sinks within DNA. Work in
our laboratory demonstrated that oxidative damage can be generated from a distance via
DNA CT, and that this damage is localized to these low potential guanine multiplets (
Arkin
et al., 1997
;
Hall et al., 1996
). Here metallointercalating photooxidants were covalently
tethered to one end of the DNA duplex, ensuring spatial separation between the
photooxidant and the guanine doublet. After irradiation and piperidine treatment of
radiolabeled DNA, damage was observed predominantly at the 5
-G of a guanine doublet
located far from the site of metallointercalation. Because the timescale of DNA CT is much
faster than the formation of permanent oxidative lesions (
vide infra
), the injected electron
hole can equilibrate along the base-pair
π
-stack and localize to the low potential guanine
doublet. Thus the pattern of oxidative damage to guanine multiplets is a characteristic of
damage from a distance via DNA CT. In these studies we also observed the sensitivity of
DNA CT to intervening perturbations in base pair stacking; introduction of an intervening
DNA bulge could turn off the long-range damage (
Hall and Barton, 1997
). These studies
further illustrated the long distances over which DNA CT could proceed: long range
oxidative damage to DNA was demonstrated over 200 Å away from the tethered
photooxidant, a remarkable molecular distance (Núñez et al., 1999).
The ability to carry out long range oxidative damage is not special to the
metallointercalating photooxidants; indeed long range photoxidation of DNA was also
Arnold et al.
Page 3
Cell Chem Biol
. Author manuscript; available in PMC 2017 January 21.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
demonstrated using a tethered and helix-capping anthraquinone (
Schuster, 2000
). Studies
with various organic photooxidants underscored how long range oxidative damage depended
not on the probe but on the DNA duplex.
Kinetics of DNA CT
Experiments using picosecond spectroscopy explored electron transfer between non-
covalently bound, intercalated ruthenium and rhodium metal complexes to learn more about
the timescale for DNA CT (
Arkin et al., 1996
). While even on the picosecond timescale,
DNA-mediated CT between the metal complexes was faster than the instrumental
resolution; these studies established a lower limit for the electron transfer rate of 3 × 10
10
s
−1
(
Arkin et al., 1996
).
The rates of DNA CT along with effects of driving force were then extensively studied on a
fast timescale using synthetic DNA hairpins (
Lewis et al., 2000
). Here mechanistic studies
were carried out (see below) and the reorganization energy within the DNA duplex could be
determined, but given the rigidity of the crosslinked hairpins, dynamical motions within the
duplex are limited, and new tethered probes were required to probe the importance of DNA
dynamics to DNA CT.
Femtosecond transient absorption experiments were then able to observe DNA-mediated
electron transfer between photoexcited tethered ethidium and a modified base in DNA (
Wan
et al., 1999
). Here, the modified base 7-deazaguanine was positioned at varying distances
from the tethered ethidium, which is sufficiently oxidizing in the excited state to oxidize 7-
deazaguanine but not the other DNA bases. After femtosecond-resolved laser excitation of
the ethidium, 5 ps and 75 ps decay components were observed that can be ascribed to
electron transfer from 7-deazaguanine to the ethidium excited state. These two electron
transfer rates correspond to different ethidium orientations: the 5 ps rate corresponds to an
ethidium orientation that is favorable for CT, whereas the slower 75 ps rate corresponds to
an initially unfavorable conformation that requires reorientation or rotation of the ethidium
in order for CT to occur (
Wan et al., 1999
). Importantly, the rates of electron transfer are
unaffected by donor-acceptor distance from 10 to 17 Å. However, the efficiency of CT was
observed to decrease with increasing distance. Hence it appears that the CT is gated by the
motions of the base pairs. Overall, this study demonstrated the ultrafast nature of favorable
electron transfer through DNA, as well as the shallow distance dependence of the CT rate
and the importance of DNA dynamics. Interestingly, in another study where the ethidium
was incorporated into the DNA in a more constrained conformation, no CT could be
observed; these subtle changes reflect the sensitivity of DNA CT to stacking dynamics and
electronic coupling (
Valis et al., 2006
).
DNA Electrochemistry
In addition to DNA photooxidant assemblies in solution, the other major platform that has
been effectively used to study DNA CT in our laboratory has been electrochemistry on DNA
monolayers (Figure 3). Here we can explore DNA CT in the ground state. Typically, a
single strand of DNA is modified with a terminal alkanethiol moiety and annealed to its
complementary strand. These thiol-modified DNA duplexes can then be self-assembled into
Arnold et al.
Page 4
Cell Chem Biol
. Author manuscript; available in PMC 2017 January 21.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
DNA monolayers on gold electrode surfaces, forming covalent gold-thiol bonds (
Kelley et
al., 1997b
). Monolayers can be assembled in the absence and presence of MgCl
2
to create
low and high density monolayers, respectively (
Pheeney and Barton, 2012
). More recently,
click chemistry methods have been utilized to enable controlled variation in the amount of
DNA on the surface, while concurrently producing more evenly spaced monolayers (
Furst et
al., 2013
). Finally, DNA can also be functionalized with pyrene to form DNA monolayers
on highly oriented pyrolytic graphite (HOPG), allowing for a wider potential window than
gold electrodes (
Gorodetsky and Barton, 2006
).
After DNA monolayer formation, a DNA-bound redox probe can be exploited to investigate
DNA CT on a surface. Organic dyes such as Nile blue and methylene blue have been
commonly used as redox probes (
Kelley et al., 1997b
;
Pheeney and Barton, 2012
;
Slinker et
al., 2011
). In these systems, charge can flow directly from the electrode to the redox probe,
or charge can be conducted in a DNA-mediated fashion, flowing from the electrode through
the alkane-thiol tether and the DNA
π
-stack to reach the redox probe. Here again, stacking
of the redox probe with the DNA base-pair
π
-stack to facilitate electronic coupling is vital
(
Boon et al., 2003a
). The intercalated redox probe may be covalently tethered to the distal
end of the DNA relative to the electrode surface; in this case, the nature of the linkage to the
DNA is also important (
Gorodetsky et al., 2007
). Long, saturated linkages do not maintain
electronic coupling to the DNA
π
-stack and thus do not facilitate DNA CT, in contrast to
short, unsaturated linkages which preserve this coupling (
Gorodetsky et al., 2007
).
A DNA-mediated pathway can be demonstrated by comparing well-matched DNA with
DNA that contains a mismatch or abasic site; the disruption to the base pair
π
-stacking
attenuates the amount of charge that reaches the redox probe (Figure 3). Electrochemical
charge transport through the DNA
π
-stack can occur well below the potential of individual
DNA bases, likely due to charge delocalization (
Genereux and Barton, 2010
). Therefore,
unlike experiments with DNA photooxidants, these experiments do not involve potentials
that damage the DNA. Instead, at these potentials CT is much like that through graphite (in
the perpendicular direction to the graphene sheets), where charge migration depends upon a
delocalized interaction among the
π
-stacked sheets. Indeed, using single molecule
electrochemical studies, where we measured DNA CT across a carbon nanotube gap, we
found the resistivity of the DNA duplex to be quite close to that calculated for a segment of
graphite of similar size arranged in the perpendicular direction in the nanotube gap (
Guo et
al., 2008
). Again in this experiment, the presence of a single base mismatch interfered with
CT and increased the resistivity of the DNA duplex 100-fold.
DNA electrochemistry experiments have directly demonstrated the shallow distance
dependence of DNA CT (
Slinker et al., 2011
). Multiplexed chips were developed that allow
for simultaneous investigation of up to four different types of DNA on a single gold surface
with four-fold redundancy (
Slinker et al., 2010
). An application of these chips directly
compared DNA CT between 17-mer and 100-mer duplexes covalently modified on the distal
end with a Nile blue redox probe (
Slinker et al., 2011
). In quadrants of the multiplexed chip
where a single C:A mismatch was incorporated into the 100-mer duplex, DNA CT was
significantly attenuated even over this long molecular distance, indeed to the same extent as
for the 17-mer. Moreover, the estimated ET rates of 25 – 40 s
−1
were indistinguishable
Arnold et al.
Page 5
Cell Chem Biol
. Author manuscript; available in PMC 2017 January 21.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
between the 17-mer and 100-mer duplex constructs, indicating that tunneling through the
alkanethiol tether is still rate-limiting, consistent with earlier work (
Drummond et al., 2004
),
while CT through the DNA 100-mer is rapid. Overall, this DNA electrochemistry
experiment established that charge is efficiently transported through DNA in the ground
state over long molecular distances, at least up to 100 base pairs or 34 nm, the longest
documented distance for this type of molecular wire.
Mechanism of DNA CT
Many experiments taken together suggest a model for the mechanism of DNA CT, although
there is much we still need to understand. Superexchange involves coherent orbital-mediated
tunneling along the entire DNA bridge between the electron donor and acceptor and displays
an exponential dependence of the rate of electron transfer on distance. This substantial
distance dependence with a superexchange mechanism is inconsistent with the fast electron
transfer rates that have been measured over relatively long molecular distances (
Genereux
and Barton, 2010
). For example, the DNA-mediated electron transfer from excited [Ru(phen
)(dppz)]
2+
to [Rh(phi)
2
(phen
)]
3+
occurs over a distance of 41 Å within just 3 nanoseconds
(
Murphy et al., 1993
).
Instead, an incoherent hopping mechanism, where there is some intermediate state of charge
localization on the bridge, is more likely for long-range charge transport through DNA
because of its more shallow distance dependence. One hopping proposal consists of
thermally induced localized hopping on individual DNA bases (
Giese et al., 2001
).
Localized hole hopping is typically envisioned as occurring through guanine hopping, given
that guanine is the most easily oxidized base (
Fukuzumi et al., 2005
,
Berlin et al., 2001
), but
this mechanism does not explain several experimental observations regarding DNA CT,
such the coherent transmission of the energy of the injected charge to the final distal
acceptor or the ground state electrochemical studies (
Genereux and Barton, 2010
).
Alternatively, a delocalized hopping mechanism is possible wherein charge is delocalized
over multiple bases. Based on studies with the photooxidant anthraquinone, Schuster and
coworkers proposed that such delocalization could occur through a thermally assisted
polaron hopping mechanism (
Schuster, 2000
). However, thermal activation mechanisms still
do not explain ground state electrochemistry studies where transport occurs significantly
below the potentials of the DNA bases (
Kelley et al., 1997b
). Overall, many researchers
have now found evidence for charge delocalization along multiple DNA bases during DNA
CT (
Kawai and Majima, 2013
;
Renaud et al., 2013
;
Genereux et al., 2011
).
Importantly, a periodic oscillation with distance has been observed in different systems
measuring DNA CT. Work by Tao and coworkers investigated the resistance of duplexed
DNA between two electrodes (
Xiang et al., 2015
). They found that the resistance of the
DNA circuit, which is inversely proportional to CT rate, increases linearly with distance,
and detected an oscillation with a period of 2 to 3 bases in sequences with stacked GC base-
pairs (
Xiang et al., 2015
). Similarly, in photooxidation experiments, our group has observed
a periodic oscillation of the yield of DNA CT with distance with a period of 3 to 4 bases
(
O’Neill and Barton, 2004
;
Genereux et al., 2008
). The source of this oscillation is
Arnold et al.
Page 6
Cell Chem Biol
. Author manuscript; available in PMC 2017 January 21.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
considered as the delocalization of electrons or electron holes along the number of DNA
bases in the period and depends sensitively on the sequence-dependent dynamics of DNA.
Taken together, current experimental evidence suggests that charge transport through DNA
is best described as a partially coherent hopping mechanism, consisting of multiple hopping
steps between delocalized (approximately 3 base pair) domains of well-coupled stacked
bases (
O’Neill and Barton, 2004
;
Xiang et al., 2015
). These CT-active domains of
π
-stacked
bases are created through conformational dynamics of the bases that modulate their
electronic coupling. Coherent superexchange would occur along these
π
-stacked domains of
approximately 3 base pairs. Thus DNA CT can essentially be envisioned as hopping
between adjacent stacked regions in the DNA. This mechanism also explains why a
mismatch or lesion would attenuate DNA CT: it perturbs the formation of these well-
coupled 3 base pair delocalized domains, and does so similarly in long and short DNA
duplexes.
Detection and diagnostic applications of DNA CT
Given the sensitivity of DNA CT to structural modifications that disrupt
π
-stacking of the
bases, DNA-modified electrochemistry can be harnessed for the detection of DNA damage,
base modifications, and DNA-binding proteins that either attenuate CT or contain redox-
active moieties that couple into the DNA
π
-stack. DNA-modified electrodes that employ
DNA CT as a reporting mechanism have been used to detect single base mismatches (
Boon
et al., 2000
;
Kelley et al., 1999
;
Slinker et al., 2010
) and a variety of DNA lesions such as
oxidized bases (
Boal and Barton, 2005
), all of which disrupt the dynamics of base-pair
stacking within DNA (
Genereux and Barton, 2010
). Additionally, proteins that disrupt the
structure of DNA can be effectively detected. These proteins include enzymes that flip bases
out of the helix during enzymatic activity such as the HhaI methylase or uracil DNA
glycosylase (
Boon et al., 2002
). The TATA-binding protein can also be detected
electrochemically using DNA CT because it significantly kinks the DNA when bound (
Boon
et al., 2002
;
Furst et al., 2013
;
Gorodetsky et al., 2008a
).
DNA CT methods can measure not only the presence but also the activity of DNA-
processing enzymes. One such example is detection of human methyltransferases activity. It
is well established that hyper-and hypo-methylation within cells is an early marker of
numerous cancerous phenotypes (
Baylin and Herman, 2000
;
Das and Singal, 2004
) and that
the methylation levels are correlated with the levels of methyltransferase activity.
Unfortunately, currently available methods to detect methyltransferase activity in the clinic
are comparatively expensive, time-consuming, and are not as sensitive. In contrast, the
activity of human methyltransferases can be sensitively detected using DNA-modified
electrodes (Figure 4). Within this assay, DNA-modified electrodes containing a hemi-
methylated DNA substrate are incubated with Dnmt1, a methyltransferase that prefers to
methylate hemi-methylated DNA. Next, the electrodes are treated with a restriction enzyme
that preferentially cuts unmethylated or hemi-methylated DNA. Fully methylated DNA,
which arises from methylation by Dnmt1, is protected from restriction enzyme cutting
(
Muren and Barton, 2013
). Given that the electrochemical readout is provided by a redox-
active probe intercalated into DNA on the surface of the electrode, signal attenuation occurs
Arnold et al.
Page 7
Cell Chem Biol
. Author manuscript; available in PMC 2017 January 21.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
if the unmethylated or hemi-methylated DNA is cut by the restriction enzyme and the probe
is released into solution. Thus, the retention of the electrochemical current following
treatment with the restriction enzyme represents a signal-on reporter of the activity of the
methyltransferase. The methylation activity of human Dnmt1 has been detected using this
electrochemical strategy in a variety of contexts and coupled to electrocatalysis in a two-
electrode platform for high sensitivity in detection (
Furst and Barton, 2015
;
Furst et al.,
2014
;
Muren and Barton, 2013
).
Most recently, DNA-modified electrodes have been used to detect Dnmt1 from the crude
lysates of tumor samples, eliminating the need to purify the tumor sample prior to analysis
(
Furst et al., 2014
). By electrochemically measuring the activity of Dnmt1 using these DNA-
modified electrodes, hyperactivity of Dnmt1 in colorectal cancer tumors but not adjacent
tissue is observed. Notably, this hyperactivity cannot be observed by assaying for Dnmt1
expression using qPCR or western blotting, which measure protein amount rather than
activity, nor by assaying activity using a much less sensitive radiometric labeling assay
(
Furst and Barton, 2015
). This strategy can be expanded to encompass different families of
DNA-binding proteins and thus offers a completely new platform for rapid and sensitive
detection and diagnosis in cell lysates.
We can also use these DNA electrodes to monitor the redox cofactors inherently present in
many DNA processing enzymes. MutY, a base excision repair (BER) protein containing a
[4Fe4S] cluster, was the first protein to be thus investigated (
Boon et al., 2003b
).
Importantly, these studies were able to determine the
DNA-bound
redox potential of MutY.
The observed redox potential of 90 mV versus NHE was assigned to the [4Fe4S]
3+/2+
couple, and is consistent with potential ranges for high potential iron-sulfur proteins
(HiPIPs). The DNA-mediated nature of the electrochemical signal was confirmed through
introduction of an intervening abasic site in the DNA that attenuated the signal. Finally, a
MutY mutant was assayed wherein one of the ligating cysteine residues of the [4Fe4S]
cluster was changed to a histidine; C199H MutY displayed a redox potential of 65 mV
versus NHE, a negative shift relative to the WT protein as expected for histidine ligation
(
Boon et al., 2003b
). This mutant experiment confirmed the iron-sulfur cluster as the origin
of the observed electrochemical signal. Indeed, an array of DNA-processing enzymes
containing iron-sulfur clusters have now been detected using DNA-modified
electrochemistry (
Grodick et al., 2015
).
DNA CT by Repair Proteins containing [4Fe4S] Clusters
How might DNA charge transport be used by proteins inside the cell? Following the work
with
E. coli
MutY (
Boon et al., 2003b
), other BER proteins with [4Fe4S] clusters were
discovered to have similar DNA-bound [4Fe4S]
3+/2+
reduction potentials, including
E. coli
Endonuclease III (EndoIII) and
Archeoglobus fulgidus
uracil DNA glycosylase (UDG)
(
Boal et al., 2005
). As the first step in the BER pathway, these glycosylase enzymes prevent
mutagenesis by, for example, removing adenine mispaired with 8-oxoguanine in the case of
MutY, or oxidized pyrimidines in the case of EndoIII, followed by the insertion of the
correct base by a polymerase (
Kim and Wilson, 2012
). While the mechanistic details of
glycosylase enzymes once they have found their substrates are relatively well understood
Arnold et al.
Page 8
Cell Chem Biol
. Author manuscript; available in PMC 2017 January 21.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
(
David et al., 2007
), how the search is efficiently coordinated is less clear given their low
copy numbers in the cell. In
E. coli
, there are an estimated 500 copies of EndoIII per cell,
and only approximately 30 copies of MutY (
Demple and Harrison, 1994
). Given their low
copy numbers and the vast quantity of DNA that must be searched, diffusion-only search
models simply are too slow to permit scanning of the entire genome within the doubling
time of
E. coli
(
Boal et al., 2009
).
Could the [4Fe4S] cluster of these glycosylase enzymes be involved in the search process?
The [4Fe4S]
2+
cluster of EndoIII is relatively insensitive to reduction and oxidation in
solution (
Cunningham et al., 1989
), leading originally to a proposed structural role for the
cluster, although it is not required for folding or stability in the homologous enzyme MutY
(
Porello et al., 1998
). However early redox studies of these proteins were performed in the
absence of the DNA polyanion, which surely could be expected to alter the cluster potential.
Using DNA-modified electrodes, we measured the DNA-bound potential of these proteins
and found that the redox potential of the EndoIII cluster shifts upon DNA binding
(
Gorodetsky et al., 2006
). On DNA-modified HOPG electrodes, the observed midpoint
potential of 20 mV versus NHE was assigned to the [4Fe4S]
3+/2+
couple, similarly to what
had been previously reported on DNA-modified gold electrodes. On bare HOPG, without
DNA, two signals were observed: an irreversible anodic peak at 250 mV and a cathodic
peak at −300 mV, assigned to the [4Fe4S]
3+/2+
and [4Fe4S]
2+/1+
couples, respectively. Thus
DNA binding negatively shifts the [4Fe4S]
3+/2+
redox potential of the cluster by at least 200
mV, activating the cluster towards oxidation (
Gorodetsky et al., 2006
). Because significant
conformational changes do not occur upon DNA binding, a thermodynamic consequence of
this shift in redox potential is that the oxidized [4Fe4S]
3+
form of EndoIII has a much higher
affinity for DNA (3 orders of magnitude) than the reduced [4Fe4S]
2+
form. Lower DNA
binding affinity for the reduced enzyme was also observed qualitatively with bulk
electrolysis experiments on EndoIII, MutY, and UDG (
Boal et al., 2005
). Therefore while
these proteins are relatively insensitive to oxidation in solution, the redox potential of the
[4Fe4S] cluster shifts into the physiologically relevant range when bound to DNA.
The combination of (
i
) the negative shift in redox potential of the [4Fe4S] cluster of
glycosylase enzymes upon DNA binding that entails higher binding affinity in the oxidized
[4Fe4S]
3+
state compared to the reduced [4Fe4S]
2+
state, (
ii
) similar DNA-bound
[4Fe4S]
3+/2+
redox potentials of approximately 80 mV versus NHE for all of the enzymes
studied, and (
iii
) the rapid kinetics of DNA CT (ps), suggested a model whereby these
proteins could use interprotein DNA-mediated CT, in a kind of electron transfer self-
exchange reaction, to cooperate in order to find lesions inside the cell (
Boal et al., 2009
)
(Figure 5). In this model, a repair enzyme containing a [4Fe4S] cluster, bound to DNA in the
reduced form, could initially become oxidized through DNA CT by guanine radicals in
DNA or reactive oxygen species. A second enzyme, also containing a [4Fe4S] cluster, binds
DNA within (at a minimum) 100 bases of the first enzyme, becoming activated towards
oxidation and releasing an electron into the
π
-stack of the DNA. This electron can then
reduce the first distally bound repair enzyme via DNA CT if the intervening DNA is
undamaged, resulting in dissociation of this reduced enzyme. The repair proteins have thus
scanned this intervening region of the genome and found it to be free of damage. Because
Arnold et al.
Page 9
Cell Chem Biol
. Author manuscript; available in PMC 2017 January 21.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
the DNA-bound redox potentials of the clusters are very similar, this interprotein DNA-
mediated CT can be viewed as an activationless self-exchange CT reaction. However, if
there is a lesion in the DNA between the proteins, DNA CT will be attenuated and the
electron will not efficiently reach the distal protein. Instead, both proteins will remain bound
to the DNA and can processively diffuse to the location of damage. In this manner, the range
over which the slower process of diffusion must occur is significantly reduced. Thus we
propose that DNA-mediated signaling would be an efficient way to localize BER proteins to
the vicinity of DNA lesions within the cell.
Observing protein redox activation via DNA CT spectroscopically
Experiments first focused on gathering
in vitro
evidence for the feasibility of [4Fe4S] cluster
oxidation in proteins by guanine radicals through DNA CT. MutY oxidation via the flash-
quench technique was monitored by EPR and transient absorption spectroscopies (
Yavin et
al., 2005
). After the “flash” of photoinduced excitation, the Ru
2+
photooxidant excited state
is oxidatively “quenched” by a diffusing molecule in order to yield a highly oxidizing
intercalated ground state Ru
3+
species
in situ
. With DNA, [Ru(phen)
2
(dppz)]
2+
, a diffusing
quencher, and MutY,
g
-values consistent with the oxidation of the cluster to [4Fe4S]
3+
and
its decomposition product, the [3Fe4S]
+
cluster, are observed upon irradiation using low
temperature EPR. With similarly composed transient absorption experiments utilizing
alternating poly(dG-dC) DNA, a very long-lived positive transient is observed with a fast
phase corresponding to guanine radical and a slow phase with characteristics consistent with
[4Fe4S]
3+
, also indicating oxidized MutY. Importantly, the long-lived positive transient is
not observed with poly(dA-dT), suggesting the importance of guanine radical as an
intermediate in DNA CT to yield MutY oxidation. Furthermore, the yield of guanine
oxidation was monitored biochemically in radiolabeled mixed sequence DNA containing a
guanine doublet (
Yavin et al., 2005
). In the absence of MutY, damage is localized
specifically to the 5
-G of the guanine doublet as expected for oxidative damage generated
through DNA CT; this damage is inhibited upon titration with MutY. Overall, these data
indicate that while MutY can be oxidized without guanine radical as an intermediate, the
thermodynamically favorable oxidation of the MutY [4Fe4S]
2+
cluster by guanine radical
enables more efficient MutY oxidation.
Subsequently, EPR experiments with a nitroxide spin label conjugated to uracil within the
DNA sequence demonstrated the feasibility of CT from both EndoIII and MutY to the
oxidized spin label (
Yavin et al., 2006
). The EPR-active,
S
= ½ nitroxide species can be
oxidized with a mild Ir
4+
oxidant to yield an EPR-silent species with a reduction potential
sufficient to oxidize the [4Fe4S]
2+
cluster of BER proteins. The reappearance of the
nitroxide signal upon the addition of protein indicates iron-sulfur cluster oxidation to reduce
the spin label. A DNA-mediated mechanism is suggested because of the dependence of spin
probe reduction on the electronic coupling of the nitroxide spin label: when the spin label is
well-coupled to the DNA via an unsaturated linkage, the nitroxide is efficiently reduced,
whereas the reduction yield is significantly attenuated with a poorly coupled saturated
linkage. The spin label thus acts as a trap for DNA-mediated CT from the proteins.
Arnold et al.
Page 10
Cell Chem Biol
. Author manuscript; available in PMC 2017 January 21.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript