DNA charge transport as a first step in coordinating
the detection of lesions by repair proteins
Pamela A. Sontz
a
, Timothy P. Mui
a
, Jill O. Fuss
b
, John A. Tainer
b,c
, and Jacqueline K. Barton
a,1
a
Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, CA 91125;
b
Life Sciences Division, Lawrence Berkeley
National Laboratory, Berkeley, CA 94720; and
c
Department of Molecular Biology, Skaggs Institute of Chemical Biology, The Scripps Research Institute,
La Jolla, CA 92037
Contributed by Jacqueline K. Barton, December 13, 2011 (sent for review November 1, 2011)
Damaged bases in DNA are known to lead to errors in replication
and transcription, compromising the integrity of the genome. We
have proposed a model where repair proteins containing redox-
active [4Fe-4S] clusters utilize DNA charge transport (CT) as a first
step in finding lesions. In this model, the population of sites to
search is reduced by a localization of protein in the vicinity of
lesions. Here, we examine this model using single-molecule atomic
force microscopy (AFM). XPD, a 5
′
-3
′
helicase involved in nucleotide
excision repair, contains a [4Fe-4S] cluster and exhibits a DNA-
bound redox potential that is physiologically relevant. In AFM
studies, we observe the redistribution of XPD onto kilobase DNA
strands containing a single base mismatch, which is not a specific
substrate for XPD but, like a lesion, inhibits CT. We further provide
evidence for DNA-mediated signaling between XPD and Endo-
nuclease III (EndoIII), a base excision repair glycosylase that also
contains a [4Fe-4S] cluster. When XPD and EndoIII are mixed to-
gether, they coordinate in relocalizing onto the mismatched strand.
However, when a CT-deficient mutant of either repair protein is
combined with the CT-proficient repair partner, no relocalization
occurs. These data not only indicate a general link between the
ability of a repair protein to carry out DNA CT and its ability to re-
distribute onto DNA strands near lesions but also provide evidence
for coordinated DNA CT between different repair proteins in their
search for damage in the genome.
DNA electron transfer
∣
iron-sulfur clusters
∣
oxidative damage
P
roteins in cellular repair pathways are charged with finding
and correcting genomic lesions arising from a variety of
sources including oxidative stress, UV radiation, and environ-
mental mutagens (1). Specific repair proteins are allocated to
target different types of DNA damage for a concerted attack. For
example, the nucleotide excision repair (NER) pathway involves
proteins that repair lesions that largely distort the helical struc-
ture of DNA in a variety of ways; dipyrimidine adducts and
cis-platin-crosslinks are two structurally distinct examples. In con-
trast, different base excision repair (BER) glycosylases remove
specific base lesions; in
Escherichia coli
, for example, Endonu-
clease III (EndoIII) targets oxidized pyrimidines while MutY
repairs oxo-guanine-adenine mismatches (1
–
4).
Given that DNA facilitates charge transport (CT) over long
molecular distances (5, 6) and this CT chemistry is sensitive to
the wide variety of lesions that perturb DNA base stacking (7, 8),
our laboratory has proposed that repair proteins exploit this
unique property of DNA to search for damage (9
–
11). Several
DNA repair proteins contain redox-active [4Fe-4S] clusters that
are not required for folding (12, 13). Examples of these proteins
from the BER pathway, EndoIII, and MutY, are activated toward
oxidation as they bind DNA (14
–
16). While EndoIII effectively
removes oxidized pyrimidines, the repair protein is found in very
low copy number in
E. coli
(approximately
500
copies per cell)
(17, 18). Surprisingly, MutY, which removes adenine from
8-oxo-G:A mismatches, is found in even lower copy number with
only approximately
30
copies per cell (19). These low copy num-
bers, along with the low selectivity of these proteins for their sub-
strates relative to unmodified DNA, begs the question of how
they can so effectively find and repair their target lesions in the
genome. Importantly, mutations in the human MutY homolog
(MUTYH) directly correlate to the development of colorectal
cancer (20, 21).
We have recently examined whether DNA CT may play some
role in how these BER proteins find their site (22, 23). Using DNA
electrochemistry and atomic force microscopy (AFM) experi-
ments, we found that the ability of EndoIII mutants to localize in
the vicinity of a base mismatch correlates with their ability to carry
out DNA CT. Moreover, using a genetics assay, we found that
EndoIII cooperates with MutY in vivo in finding MutY lesions, but
that a CT-deficient mutant of EndoIII cannot similarly aid MutY
(22). Interestingly, more proteins involved in genome maintenance
that contain [4Fe-4S] clusters are emerging, many with no clear
structural or enzymatic role for their clusters. One such protein
is XPD, which is part of the transcription factor IIH (TFIIH)
machinery and is involved in NER (24, 25). Here we consider if
XPD, which is not involved in BER, similarly utilizes DNA CT to
relocalize in the vicinity of lesions, and whether signaling between
different DNA-binding proteins can occur so as to coordinate the
search for damage (26, 27).
XPD, an ATP-dependent 5
′
-3
′
helicase from the NER pathway,
is a major component of TFIIH, the transcriptional and repair
machinery that unwinds damaged DNA for lesion repair in NER
(28, 29). Mutations in the human XPD helicase gene (ERCC2)
lead to the genetic disorders trichothiodystrophy (TTD), Cockayne
syndrome (CS), and xeroderma pigmentosum (XP) (28
–
31). There
is now evidence that mutations specifically in the iron-sulfur cluster
domain of XPD result in TFIIH instability, thus leading to TTD
(28, 30). Similarly, mutations in the analogous region on the re-
lated
Fa ncJ
gene (which also encodes a helicase with an iron-sulfur
cluster) causes predisposition to early onset breast cancer (25, 32,
33). We have demonstrated that the [4Fe-4S] cluster in
Sulfolobus
acidocaldarius
(Sa)XPD is redox-active, with a redox potential
equivalent to that of BER proteins EndoIII and MutY [approxi-
mately
80
mV vs. normal hydrogen electrode (NHE)] (34).
Here we explore, using AFM, whether XPD can participate
in DNA-mediated signaling to localize in the vicinity of a DNA
mismatch. Similar to base lesions, a single base mismatch inhibits
DNA CT but importantly, the base mismatch is not a substrate
for repair by XPD (35). In our model for facilitated search by
DNA CT (Fig. 1), the search for DNA damage is initiated, under
oxidative stress, by the one electron oxidation of a DNA-bound
protein by a nearby guanine radical; the oxidation takes the clus-
ter from a
2
þ
oxidation state to a
3
þ
oxidation state. But it is
Author contributions: P.A.S., T.P.M., and J.K.B. designed research; P.A.S. and T.P.M.
performed research; J.O.F. and J.A.T. contributed new reagents/analytic tools; P.A.S.,
T.P.M., J.O.F., J.A.T., and J.K.B. analyzed data; and P.A.S., T.P.M., J.O.F., J.A.T., and J.K.B.
wrote the paper.
The authors declare no conflict of interest.
1
To whom correspondence should be addressed. E-mail: jkbarton@caltech.edu.
This article contains supporting information online at
www.pnas.org/lookup/suppl/
doi:10.1073/pnas.1120063109/-/DCSupplemental
.
1856
–
1861
∣
PNAS
∣
February 7, 2012
∣
vol. 109
∣
no. 6
www.pnas.org/cgi/doi/10.1073/pnas.1120063109
DNA CT between two repair proteins that leads to the redistri-
bution in the vicinity of a lesion. With a DNA-bound potential
shifted negative vs. free protein (14), the proteins are expected to
have a lower affinity for DNA in their reduced form. Thus, in the
model, DNA-mediated reduction of one protein by the other
leads to the dissociation of the reduced protein from DNA,
effectively giving overall dissociation of repair proteins away from
a region of the genome without lesions. Our model relies on the
sensitivity of CT to proper
π
stacking of the bases between the
donor protein and the distant acceptor protein (22, 23); if instead
there is an intervening lesion, DNA-mediated CT does not occur,
and the repair protein remains bound and can processively move
to the lesion and carry out enzymatic repair. In this process, the
repair proteins eventually relocalize in the vicinity of lesions or
any modification that inhibits DNA CT. As a result the overall
search regime for the repair proteins is made smaller than the
full genome: they preferentially localize where needed for repair.
Despite various studies of EndoIII and MutY independently, it
still remains to be established whether these proteins from differ-
ent repair pathways (NER and BER) cooperate with one another
in a coordinated search. Because a C
∶
A mismatch, for example,
does not distort the DNA helix, it would be expected to evade the
biological NER pathway. Alternatively, if enzymes rely on DNA/
protein CTas a first step in finding damage, then any protein that
can carry out DNA CT can participate in the search, integrating
repair pathways that were previously thought to be separate. To
examine this coordinated search, AFM is used to visualize mix-
tures containing two types of proteins: (
i
) those that are able to
carry out DNA CT and (
ii
) those that are defective at DNA CT.
AFM, a technique that allows for visualization of single molecule
protein/DNA interactions, can be used to observe proteins bound
to long strands (3.8 kbp) of DNA containing a site-specific mis-
match (C
∶
A) (23, 36
–
39). Binding density ratios calculated from
counting strands and proteins in AFM images can then give us
clues to the first step in lesion detection. Here, not only do we
examine whether proteins from the NER pathway redistribute
in the vicinity of lesions as a function of their ability to perform
DNA CT but also whether they can cooperate with other repair
proteins in the DNA-mediated search for damage.
Results
Experimental Strategy Using AFM to Probe Protein Distribution.
We
have previously demonstrated with BER proteins that the ability
of a protein to perform DNA CT directly correlates to its redis-
tribution in the vicinity of base lesions or mismatches that inhibit
CT (23). We tested XPD redistribution promoted through DNA
CT by preparing DNA strands containing a single C
∶
A mismatch,
a modification that we know to inhibit CT (40), alongside DNA
strands containing no mismatches. To distinguish the strands in
microscopy experiments, we made mismatched strands
>
1
;
200
base pairs longer than the matched strands. AFM was utilized to
gather images of DNA and bound protein that could be further
analyzed to examine the propensity of XPD to redistribute. ATP
was not incorporated in AFM experiments, as XPD helicase ac-
tivity should not be required for a DNA-mediated redistribution.
Long (3.8 kilobase) DNA duplexes containing a site-specific
C
∶
A mismatch and short (2.2 and 1.6 kilobases) well matched
duplexes of the same total sequence were constructed. To prepare
these DNA samples, pUC19 plasmid was amplified with primers
incorporating a 2
′
O
-methyl residue to yield two short duplexes
containing 14-nucleotide single-strand overhangs (23, 41). Upon
ligation of these two duplexes, 3.8 kbp strands were formed that
contained only one C
∶
A mismatch in the middle of the strand.
The ligation reaction was not taken to completion so as to have
a mixture of well matched short and mismatched long strands
for protein distribution studies. Although a C
∶
A mismatch effec-
tively inhibits DNA CT (40), it is not a lesion that is preferentially
bound by XPD. However, the 14-nucleotide overhangs generated
with PCR are specific substrates for XPD helicase (
K
d
approxi-
mately
1
μ
M) (25, 42). We directly observed this preference in
initial AFM experiments. Protein assignments were verified
through analysis of their 3
–
4 nm heights in the images; without
protein, features of this dimension were not observed and larger
heights (
>
7
nm) indicated salt precipitates or protein aggrega-
tion. Only clearly identifiable long or short strands and bound
proteins were counted. XPD protein position was determined
based on the distance of the protein from the end of the strand.
We examined distribution with and without blocking the ends
of short strands, given that XPD has some preferential binding
affinity for single-strand overhangs. Comparing raw protein posi-
tion (middle vs. ends) on long and short strands, XPD exhibited a
large preference for the ends (approximately
300
bp range) of
short duplexes containing overhangs (
>
50%
of bound proteins).
To block XPD from binding to the 14 bp segment, overhang com-
plements were added to DNA/protein solutions in excess. We
found, when we blocked the overhangs, no preferential binding
to the short strands was observed (
vide infra
).
Detection of XPD Complexes.
AFM images of DNA-bound protein
complexes are represented in Fig. 2. XPD protein bound to
matched DNA (Fig. 2 top) can be easily distinguished based on
strand height profile of tapping mode images. Zoomed in images
display clearly identifiable long and short strands with protein
bound (Fig. 2 bottom). Images of DNA and protein complexes
were acquired with a scan size of
2
×
2
μ
m
2
or
3
×
3
μ
m
2
at a scan
rate of 3.05 Hz. Because AFM images vary with surface cleavage,
sample wetness, deposition time, and volume,
>
200
long or short
strands were counted for at least three independent samples. The
uncertainty was based on the total number of proteins counted.
Fig. 1.
Model for damage detection and redistribution by repair proteins.
(Top) Given two repair proteins, containing [4Fe-4S] clusters, that are able to
carry out DNA CT, such as XPD or EndoIII (blue), CT can proceed through well
matched unmodified DNA. Driven by protein oxidation by guanine radicals
(yellow) formed under oxidative stress, DNA CT occurs between the DNA-
bound proteins, promoting the dissociation of reduced protein (light blue).
DNA CT does not occur, however, in the presence of an intervening mismatch
or lesion (red X); with the intervening mismatch or lesion, repair proteins
are not reduced from a distance, and therefore do not dissociate but instead
remain bound in the vicinity of the lesion. (Bottom) If repair proteins are
mutants such as XPD L325V or EndoIII Y82A (red) and are unable to perform
DNA CT (dashed line), they cannot send or receive signals to locate damage.
In the presence of a lesion, these proteins then do not preferentially redis-
tribute in the vicinity of the mismatch.
Sontz et al.
PNAS
∣
February 7, 2012
∣
vol. 109
∣
no. 6
∣
1857
BIOCHEMISTRY
CHEMISTRY
Importantly, we measured the relative binding affinity of mutants
and wild type (WT) proteins, and in all of the samples described,
the number of proteins bound per base pair remained constant
(
Table S1
). Thus, any changes we see in distribution are not due to
differences in binding affinities of proteins.
In order to probe the DNA CT properties of a redox-active
DNA helicase, XPD and DNA complexes were examined with
AFM. XPD shows redistribution onto long mismatched strands
with a ratio of protein binding densities,
r
(long/short) of
1
.
54
0
.
08
(Fig. 3). We observed 0.23 proteins per kilobase of long
strand and 0.15 proteins per kilobase of short strand (see
Table S1
for further details). If, instead, we examine XPD distri-
bution when long and short strands are fully matched, we see a
binding density ratio of
0
.
94
0
.
05
. Previously, for EndoIII we
found
r
, for (long/short) was
1
.
6
0
.
09
(22). We expect a ratio
of 1 if there is an equal distribution of proteins on matched
and mismatched strands (22). Thus, similar to EndoIII distribu-
tion, XPD, an NER protein with DNA-bound redox activity,
redistributes to localize in the vicinity of a C
∶
A mismatch.
We also tested the possible redistribution of a CT-deficient
mutant of SaXPD, L325V (L461 in human XPD). L325V, aligns
with mutated residues in human XPD and
Schizosaccharomyces
pombe
Rad 15 that are associated with TTD, XP, and XP/CS (25,
31, 33). It should be noted, however, that the structural scaffold
required to form the TFIIH machinery is not disrupted in the
L325V mutant. As evident in Fig. 3, the L325V XPD mutant ex-
hibits an electrochemical signal that is less than half that of WT
XPD, indicating it is deficient at performing DNA CT. Interest-
ingly, L325V is 30 Å away from the [4Fe-4S] cluster, yet affects
protein/DNA CT. Significantly, when we examine whether this
electrochemically deficient mutant redistributes onto the mis-
matched strand in our AFM assay, we find that there is no pre-
ference for the mismatched strands; L325V exhibits a protein
binding density ratio,
r
(long/short) of
1
.
14
0
.
06
in mismatched
samples, which is within error of the fully matched binding
density ratio
r
(long/short) of
1
.
05
0
.
07
(Fig. 3). Thus, as with
the CT-deficient EndoIII mutant Y82A (22), we find a correla-
tion between the inability of L325V to perform CT and its lack
of redistribution onto mismatched strands. Even though the
[4Fe-4S] cluster is intact, this mutant protein cannot cooperate
with other proteins using DNA CT.
Atomic Force Microscopy of Protein Mixtures.
We have established
that DNA-mediated CT occurs on a much faster time scale than
protein diffusion (43, 44), and we have characterized several
mutants of EndoIII, from the BER repair pathway that exhibit
similar midpoint potentials of approximately
80
ð
30
Þ
mV vs.
NHE yet differ in their ability to carry out DNA CT (14, 23, 26).
We have not, however, yet provided evidence that different pro-
teins can signal one another through DNA-mediated CT. AFM
provides an opportunity to examine this issue by assaying protein
mixtures.
To test this model, WT XPD and EndoIII, both proficient at
DNA CT, were mixed in a
1
∶
1
equimolar ratio with DNA. Short
duplex overhangs were blocked with excess complementary single
strand. Fig. 4 shows representative images of fully matched DNA
strands or mismatched DNA strands incubated overnight with
XPD/EndoIII
1
∶
1
protein (WT
XPD
∕
WT
EndoIII
). XPD is twice
as large as EndoIII (65 and 32 kDa respectively) (PDB IDs:
3CRV, 1P59). The average height of the protein in AFM studies
can be estimated using the equation:
R
¼
0
.
717
ð
M
Þ
1
∕
3
where
R
is
the radius of the protein globule in Å and
M
is the mass of the
protein in Da (45). Thus, XPD is expected to have a height of
5.6 nm, which is within error of EndoIII (4.6 nm). While in gen-
eral we cannot distinguish them, select images (Fig. 4 inset) are
suggestive of both EndoIII and XPD binding. XPD (0.3
μ
M)/
EndoIII (0.3
μ
M)
1
∶
1
protein mixture shows that XPD and
EndoIII redistribute onto long mismatched strands, with a bind-
ing density ratio
r
(long/short) of
1
.
75
0
.
13
. Fully matched con-
trols yield a binding density ratio
r
(long/short) of
1
.
02
0
.
07
.
Importantly, controls were also performed with only XPD
(0.3
μ
M) under the same conditions as mixed protein experi-
ments. Consistent with XPD at a higher concentration (0.6
μ
M),
XPD at half the concentration (0.3
μ
M) redistributes onto long
mismatched strands (
r
¼
1
.
38
0
.
07
) with no redistribution in
Fig. 2.
AFM to visualize DNA-bound proteins. Tapping mode AFM images
of DNA and XPD protein on mica imaged in air. The matched strands and
protein (dots) are visible on the surface (top). Because the single strand over-
hangs on the short duplexes have been blocked by annealing short oligo-
mers, XPD is bound in random positions on the DNA. Zoomed-in images
of long (3.8 kbp) and short (1.6 and 2.2 kbp) DNA strands with bound proteins
(black arrows) are shown below. Images were acquired with a scan size of
2
×
2
μ
m
2
or
3
×
3
μ
m
2
, at a rate of 3.05 Hz with a data scale of
10
nm
∕
0
.
5
V.
Fig. 3.
DNA CT with the L325V mutant. (Left) Cyclic Voltammogram (CV) of
WT SaXPD [8
μ
M] (green) and L325V SaXPD [8
μ
M] (red) on DNA-modified
electrodes after 90 min. (Right) Quantitation of XPD and mutant L325V
protein density ratios (<
10%
uncertainty) where C
∶
A indicates a mismatch
is contained in long strands. The unmarked bars show the control measure-
ments for fully matched long and short strands. XPD redistributes onto
mismatched strands. L325V, CT deficient, does not show redistribution.
1858
∣
www.pnas.org/cgi/doi/10.1073/pnas.1120063109
Sontz et al.
fully matched samples (
r
¼
1
.
08
0
.
07
). Significantly, mixing
XPD and EndoIII protein results in a binding density ratio of
1.75, which is higher than that of each protein separately. This
result suggests that these two proteins signal one another to
localize in the vicinity of the lesion.
In order to explore cooperativity between XPD and EndoIII,
we then replaced WT EndoIII with mutant Y82A EndoIII, a
protein that binds to DNA but cannot perform DNA-mediated
CTefficiently (22). In previous studies, Y82A did not redistribute
in mismatched samples (
r
¼
0
.
9
0
.
1
) (22). When XPD is mixed
with Y82A
1
∶
1
(WT
XPD
∕
Y
82
A
EndoIII
), there is no redistribution
onto mismatched strands (
r
¼
0
.
98
0
.
05
). WT
XPD
∕
Y
82
A
EndoIII
matched controls are within error of mismatched results with a
binding density ratio of
1
.
11
0
.
07
. The number of proteins
bound per kilobasepair remains the same between WT
XPD
∕
Y
82
A
EndoIII
and WT
XPD
∕
WT
EndoIII
mixtures (
0
.
13
proteins
∕
kbp). When WT
XPD
is titrated into Y
82
A
EndoIII
at a ratio of
3
∶
1
,
increasing the probability that two proteins bound on the same
strand are both electrochemically active, the proteins once again
redistribute onto mismatched strands.
Complementing WT
XPD
∕
WT
EndoIII
and WT
XPD
∕
Y
82
A
EndoIII
mixture studies, we also investigated signaling between EndoIII
and XPD mutant L325V. Similar to EndoIII Y82A, XPD L325V
cannot redistribute onto mismatched strands. WT
XPD
∕
L
325
V
XPD
1
∶
1
mixtures were examined initially to determine whether
L325V, deficient in DNA CT, could signal XPD. WT
XPD
∕
L
325
V
XPD
mixtures showed no redistribution (
r
¼
0
.
88
0
.
04
).
This result was within error of matched controls, with a binding
density ratio revealing even a slight preference for short strands
(
r
¼
0
.
90
0
.
05
). L325V inhibits XPD from relocalizing in the
vicinity of lesions, but does L325V have any influence on the
search for damage by EndoIII BER protein? WT
EndoIII
∕
L
325
V
XPD
mixtures were examined and exhibited no difference
in mismatched vs. matched samples, where
r
¼
1
.
02
0
.
05
and
0
.
94
0
.
04
respectively. As with the WT
XPD
∕
Y
82
A
EndoIII
mix-
ture, here we see that L325V binding alters EndoIII redistribu-
tion. Fig. 5 shows binding density ratio comparisons for
WT
XPD
∕
WT
EndoIII
,WT
XPD
∕
Y
82
A
EndoIII
,WT
EndoIII
∕
L
325
V
XPD
,
and WT
XPD
∕
L
325
V
XPD
1
∶
1
protein mixtures.
In the case of the mixtures of active proteins and CT-deficient
mutants, we might have expected binding density ratios to be
equal to that of pure samples of active repair protein. However,
we are testing these redistributions at protein loadings on the
DNA strands of approximately
2
per strand. Thus if one of the
proteins is CT defective, there is no protein to which the CT-pro-
ficient protein may signal. Certainly the striking differences
seen between CT-active protein mixtures and those containing
CT-deficient mutants support the idea that the proteins can carry
out DNA CT to one another. Having established that this ability
to redistribute in the vicinity of mismatches depends on the ability
of the protein to carry out DNA CT (23), these results thus pro-
vide evidence for DNA CT as a means of interprotein signaling.
Discussion
The data presented here indicate that XPD, an archaeal protein
from the NER pathway, may cooperate with other proteins that
are proficient at DNA CT to localize in the vicinity of damage.
XPD, a superfamily 2 DNA helicase with 5
′
-3
′
polarity, is a
component of TFIIH that is essential for repair of bulky lesions
generated by exogenous sources such as UV light and chemical
carcinogens (28, 46, 47). XPD contains a conserved [4Fe-4S] clus-
ter suggested to be conformationally controlled by ATP binding
and hydrolysis (25). Mutations in the iron-sulfur domain of XPD
can lead to diseases including TTD and XP, yet the function of
the [4Fe-4S] cluster appears to be unknown (23, 28, 29). Electro-
chemical studies have shown that when BER proteins MutY and
EndoIII bind to DNA, their [4Fe-4S] clusters are activated to-
ward one electron oxidation (14, 26). XPD exhibits a DNA-bound
midpoint potential similar to that of EndoIII and MutY when
bound to DNA (approximately
80
mV vs. NHE), indicative of a
possible role for the [4Fe-4S] cluster in DNA-mediated CT (34).
For EndoIII we have also already determined a direct correlation
between the ability of proteins to redistribute in the vicinity of
mismatches as measured by AFM, and the CT proficiency of the
proteins measured electrochemically (23). Thus, we may utilize
single-molecule AFM as a tool to probe the redistribution of
proteins in the vicinity of base lesions and in so doing, the pro-
ficiency of the protein to carry out DNA CT.
Here we show that, like the BER protein EndoIII, XPD, in-
volved both in transcription and NER, redistributes in the vicinity
of a lesion. Importantly, this ability to relocalize is associated with
the ability of XPD to carry out DNA CT. The mutant L325V is
defective in its ability to carry out DNA CTand this XPD mutant
also does not redistribute effectively onto the mismatched strand.
Moreover, these data provide evidence that two different re-
pair proteins, each containing a [4Fe-4S] cluster at similar DNA-
bound potential, can communicate with one another through
DNA-mediated CT. This result becomes more interesting still
given that in the experiments conducted here, the proteins are
from completely different organisms. Nonetheless, what is critical
Fig. 4.
Representative tapping mode AFM images. Fully matched long and
short DNA strands (left) or mismatched long DNA and matched short DNA
strands (right) are incubated overnight with XPD/EndoIII
1
∶
1
protein mixture.
Inset shows zoomed-in view of a long DNA strand with both large (approxi-
mately
6
nm) and small (approximately
3
.
5
nm) proteins bound.
Fig. 5.
Summary of binding density ratios for XPD (red text)/EndoIII (blue
text) and mutant mixtures. Quantitation of protein density ratios (<
10%
un-
certainty) where C
∶
A indicates a mismatch is contained in long strands.
WT
XPD
∕
WT
EndoIII
mixtures
1
∶
1
(purple), both proteins CT proficient, redistri-
bute onto mismatched strands. WT
XPD
∕
Y
82
A
EndoIII
and WT
EndoIII
∕
L
325
V
XPD
1
∶
1
mixtures (green and orange respectively), Y82A and L325V both mutants
deficient in CT, do not show redistribution. WT
XPD
∕
L
325
V
XPD
1
∶
1
mixtures
(blue) do not show redistribution, as XPD cannot perform DNA-mediated
signaling if the L325V mutant is present.
Sontz et al.
PNAS
∣
February 7, 2012
∣
vol. 109
∣
no. 6
∣
1859
BIOCHEMISTRY
CHEMISTRY
is that the protein clusters have similar DNA-bound potentials,
facilitate many electron exchanges, and have the ability to carry
out DNA-mediated CT. Furthermore, no signaling to effect the
redistribution in the vicinity of lesions occurs when one partner is
CT deficient.
Lesion detection by repair proteins, based on our model, de-
pends on the maximum distance over which DNA-mediated CT
can occur and the percentage of proteins oxidized (6, 22, 48).
While we have documented that DNA CT occurs over at least
100 bp, we do not yet know the maximum distance for this process
(5, 49). If CTcould proceed only over short distances (<
500
bp),
more than six proteins would be required for signaling across
strands. Instead, only 1
–
3 proteins are bound to the long matched
or mismatched strands in AFM experiments, suggesting that
DNA CT is occurring over distances much greater than those
we have measured electrochemically.
How might this signaling be utilized inside the cell? Surely, this
ability to redistribute in the vicinity of lesions reduces the search
process required to find lesions across the genome. The higher
the concentration of total proteins involved in signaling that are
at similar potentials, the more efficient the search process becomes
(22). Indeed, XPD may utilize DNA-mediated CT to signal its
presence and perhaps to
“
call off
”
other proteins from other repair
pathways. Various DNA-binding proteins, such as those involved
in repair and other DNA transactions, have been found to contain
iron-sulfur domains and other redox cofactors (50
–
53). DNA-
mediated signaling among DNA-binding proteins that are involved
in maintaining the integrity of the genome allows a coordination of
repair, transcription, and replication processes.
Materials and Methods
All chemicals were purchased from Sigma Aldrich. All enzymes were pur-
chased from New England Biolabs unless otherwise specified. Mica surfaces
were purchased from SPI supplies. Silicon AFM probes were purchased from
Budget Sensors. Oligonucleotides were purchased from IDTor synthesized on
a 3400 DNA synthesizer (Applied Biosystems).
Mismatched (C
∶
A) Strand Synthesis.
Four primers with the following se-
quences were synthesized using standard phosphoramidite chemistry:
1. 5
′
-GTACAGAGTTCAGT
CGGCATCCGCTTACAGACAAGC-3
′
(forward),
2. 5
′
-CCGGTAACTATCGTCTTGAGTCC-3
′
(reverse),
3. 5
′
-GACTGAACTCTGTA
CCTGGCACGACAGGTTTCCCG-3
′
(forward),
4. 5
′
-GACTGAACTCTATA
CCTGGCACGACAGGTTTCCCG-3
′
(forward)
The underlined bases highlight the location of a 2
′
-O-methyl residue (Glen
Research). Primers were phosphorylated using 5 U PNK, 10% PNK buffer,
0.5 mM ATP for 5 h at 37 °C. Primers were purified using phenol-chloroform
extraction followed by ethanol precipitation (54). After being dried in vacuo,
primers were redissolved in 20
μ
L water and used in separate PCR reactions
(41) using pUC19 as a template to generate two duplexes 1,610 bp and
2,157 bp (matched), each containing one 14-nucleotide single-strand over-
hang. Each 100
μ
L PCR reaction contained 50 pmol of each of two primers,
1X
Taq
buffer (100 mM Tris-HCl, 15 mM MgCl
2
, 500 mM KCl, pH 8.3), 0.2 mM
each dNTP, 1 ng plasmid template pUC19 and 3 U
Taq
polymerase (Roche). A
typical step program for PCR was as follows: After incubation at 94 °C for
10 min, 34 cycles were performed as follows: 94 °C for 1 min, (52 °C for pri-
mers
1
þ
2
, 54 °C for primers
2
þ
3
,or
2
þ
4
), for 1 min, and 72 °C for 3 min.
The PCR product was then suspended in 50 mM NaCl/ 5 mM phosphate buffer
and quantitated. Separate duplexes were annealed at 65 °C for 8 min in
10 mM Tris buffer, then cooled to 20 °C during 2 h. A total of 15 units of
T4 DNA ligase and 10% T4 ligase buffer were added (total reaction volume
∼
20
μ
L) and incubated overnight at 16 °C, followed by deactivation for
10 min at 65 °C to yield the 3,767 bp (mismatched) long strand. We did
not bring the ligation reaction to completion, so as to obtain a mixture of
DNA samples that were equivalent other than the presence of the mismatch
at the ligation site. The DNA duplexes (ligated and unligated) were then
analyzed by 0.6% agarose gel electrophoresis. Single DNA strands comple-
mentary to short duplex overhangs were ordered from IDT: 5
′
-GACT-
GAACTCTGTAC-3
′
Tm
¼
41
.
6
°C (1.6 kbp duplex overhang), 5
′
-GTACAGAG-
TTCAGTC-3
′
Tm
¼
41
.
6
°C (2.2 kbp matched duplex overhang), and 5
′
-GTATA-
GAGTTCAGTC-3
′
Tm
¼
37
.
5
°C (2.2 kbp mismatched duplex overhang).
Single-strand DNA was purified using reversed-phased HPLC and verified
with MALDI-TOF mass spectrometry.
Protein Purification and Expression.
EndoIII and Y82A EndoIII were expressed
from the pNTH10 expression vector and purified as described previously (22).
EndoIII and Y82A were stored in 20 mM sodium phosphate pH 7.5, 100 mM
NaCl, 20% glycerol, and 0.5 mM EDTA buffer. Protein concentrations were
determined using the UV-visible absorbance of the [4Fe-4S] cluster (410 nm,
ε
¼
17
;
000
) (55). XPD was purified as previously described (34).
AFM Experiments.
AFM experiments were performed using the protocol simi-
lar to that reported previously (23). Stock DNA solution contained 50
–
200 ng
of total DNA (approximately
6
μ
M) composed of the mixture of ligated 3.8-
kbp duplexes and the two unligated duplexes (1.6 and 2.2 kb) in 6 mM MgCl
2
/
Tris-EDTA buffer. The 1.6 kb overhang complement (60
μ
M) was added to
the DNA solution to block the 14 bp single-strand overhangs generated
by PCR. This sample was then incubated overnight at 4 °C. XPD protein
was dialyzed against the protein buffer (20 mM phosphate, 100 mM NaCl,
1 mM EDTA, 5% glycerol, pH 7.5 and filtered prior to use) to remove residual
DTT. The concentration of individual proteins were determined by UV/visible
spectrophotometry (Beckman DU 7400) using
ε
¼
17
;
000
M
−
1
at 410 nM for
the [4Fe-4S] cluster. After addition of excess 2.2 kb duplex complement
(60
μ
M), XPD (0.6
μ
M) was added to the stock DNA solution. This protein/
DNA solution was incubated at 4 °C overnight. Sample was then deposited
(5
–
10
μ
L) onto a freshly cleaved mica surface for 1
–
2 min, rinsed with
2 mL of water, and dried under argon. Mutant XPD protein (L325V) was
added to a stock solution of 50 ng DNA for a final protein concentration
of 0.6
μ
M. Deposition conditions were identical to that for WT XPD- DNA
samples after incubation at 4 °C overnight.
For mixed protein experiments (WT
XPD
∕
WT
EndoIII
,WT
XPD
∕
Y
82
A
EndoIII
,
WT
EndoIII
∕
L
325
V
XPD
, and WT
XPD
∕
L
325
V
XPD
), XPD, EndoIII, L325V, or Y82A were
added to the prepared DNA solution described previously at equimolar (
1
∶
1
)
concentrations (0.3
μ
M each) prior to incubation at 4 °C overnight. Protein/
DNA complexes were formed with DNA solution containing approximately
200
ng of the mixture of PCR products and overhang complements (approxi-
mately
6
μ
M DNA) in 6 mM MgCl
2
/Tris-EDTA buffer at 4 °C overnight. The
reaction mixture was then deposited (5
–
10
μ
L) on the mica surface for 1
–
2 min, rinsed with 2 mL water and dried under argon.
AFM Instrumentation.
Silicon AFM Probes purchased from Budget Sensors,
with a spring constant of
3
N
∕
m and a resonance frequency of 75 kHz, were
used in a Digital Instruments Multimode SPM. Images were captured in air
with scan areas of
2
×
2
μ
m
2
or
3
×
3
μ
m
2
in tapping mode, at an amplitude
of 0.54
–
2.00 V and at a scan rate of 3.05 Hz. Scan rates of 3.05 Hz were used in
order to obtain images of higher quality.
Binding Density Ratio Calculations.
WSxM software was used to measure gen-
eral DNA contour lengths and height profiles of the proteins as described
previously (23, 56). For each dataset, images from at least three independent
samples were analyzed, compared, and pooled (
>
200
long or short strands).
Distinguishable strands and protein positions were counted by hand. The
binding density ratio,
r
, is defined as the ratio of the proteins bound on long
strands divided by proteins bound on short strands. The ratio is normalized
for length by dividing by 1.9 kbp, which is the average length of the short
strands. Binding affinities were found by determining the number of proteins
bound per kilobasepair strand. The uncertainty was determined through the
total number of proteins observed.
Protein Electrochemistry.
Protein electrochemistry was performed as pre-
viously described (34). Briefly, individual proteins samples were dialyzed to
remove residual DTT and quantified based on 410 nm absorbance. Protein
was then added to a DNA modified electrode containing a 9 nucleotide 5
′
single-strand overhang. Cyclic Voltammograms were then obtained using
Ag/AgCl reference electrodes, Pt auxiliary electrode at
50
mV
∕
s scan rate
on a CH Instruments 620C electrochemical analyzer.
ACKNOWLEDGMENTS.
We thank Alison Parisian for technical assistance and
Eric Olmon for preparation and purification of WT and Y82A EndoIII protein.
We are also grateful to the Beckman Institute MMRC for AFM instrumenta-
tion. We also thank the National Institutes of Health (NIH) (GM49216 to
J.K.B.; CA112093 to J.A.T.), and the Department of Energy (DOE) (ENIGMA
program under Contract No. DE-AC02-05CH11231 to J.A.T.) for funding.
We also thank the National Science Foundation (NSF) for a graduate fellow-
ship to T.P.M.
1860
∣
www.pnas.org/cgi/doi/10.1073/pnas.1120063109
Sontz et al.
1. Kuraoka I, et al. (2000) Removal of oxygen free-radical-induced 5
′
, 8-purine cyclodeox-
ynucleosides from DNA by the nucleotide excision-repair pathway in human cells.
Proc
Natl Acad Sci USA
97:3832
–
3837.
2. Sancar A, Lindsey-Boltz A, Ünsal-Kaçmaz K, Linn S (2004) Molecular mechanisms of
mammalian dna repair and the DNA damage checkpoints.
Annu Rev Biochem
73:39
–
85.
3. David SS, Williams SD (1998) Chemistry of glycosylases and endonucleases involved in
base-excision repair.
Chem Rev
98:1221
–
1262.
4. David SS, O
’
Shea VL, Kundu S (2007) Base-excision repair of oxidative DNA damage.
Nature
447:941
–
950.
5. Slinker JD, Muren NB, Renfrew SE, Barton JK (2011) DNA charge transport over 34 nm.
Nature Chem
3:230
–
235.
6. Nunez ME, Hall DB, Barton JK (1999) Long range oxidative damage to DNA: effects of
distance and sequence.
Chem Biol
6:85
–
97.
7. Kelley SO, Boon EM, Barton JK, Jackson NM, Hill MG (1999) Single-base mismatch
detection based on charge transduction through DNA.
Nucleic Acids Res
27:4830
–
4837.
8. Boal AK, Barton JK (2005) Electrochemical detection of lesions in DNA.
Bioconjugate
Chem
16:312
–
321.
9. Genereux JC, Boal AK, Barton JK (2010) DNA-mediated charge transport in redox
sensing and signaling.
J Am Chem Soc
132:891
–
905.
10. Boon EM, Livingston AL, Chmiel NH, David SS, Barton JK (2003) DNA-mediated charge
transport for DNA repair.
Proc Natl Acad Sci USA
100:12543
–
12547.
11. Barton JK, Olmon ED, Sontz PA (2011) Metal complexes for DNA-mediated charge
transport.
Coord Chem Rev
255:619
–
634.
12. Kuo CF, et al. (1992) Atomic structure of the DNA repair [4Fe-4S] enzyme endonuclease
III.
Science
258:434
–
440.
13. Thayer MM, Ahern H, Xing D, Cunningham RP, Tainer JA (1995) Novel DNA binding
motifs in the DNA repair enzyme endonuclease III crystal structure.
EMBO J
14:4108
–
4120.
14. Gorodetsky AA, Boal AK, Barton JK (2006) Direct electrochemistry of endonuclease III
in the presence and absence of DNA.
J Am Chem Soc
128:12082
–
12083.
15. Porello SL, Cannon MJ, David SS (1998) A substrate recognition role for the
½
4
Fe-
4
S
2
þ
cluster of the DNA repair glycosylase MutY.
Biochemistry
37:6465
–
6475.
16. Lukianova OA, David SS (2005) A role for iron-sulfur clusters in DNA repair.
Curr Opin
Chem Biol
9:145
–
151.
17. Fromme JC, Verdine GL (2003) Structure of a trapped endonuclease III- DNA covalent
intermediate.
EMBO J
22:3461
–
3471.
18. Genereux JC, Boal AK, Barton JK (2010) DNA-mediated charge transport in redox
sensing and signaling.
J Am Chem Soc
132:891
–
905.
19. Fromme JC, Banerjee A, Huang SJ, Verdine GL (2004) Structural basis for removal of
adenine mispaired with 8- oxoguanine by MutY adenine DNA glycosylase.
Nature
427:652
–
656.
20. Cheadle JP, Sampson JR (2007) MUTYH-associated polyposis-from defect in base
excision repair to clinical genetic testing.
DNA Repair
6:274
–
279.
21. Fodde R (2002) The APC gene in colorectal cancer.
Eur J Cancer
38:867
–
871.
22. Boal AK, et al. (2009) Redox signaling between DNA Repair Proteins for efficient lesion
detection.
Proc Natl Acad Sci USA
106:15237
–
15242.
23. Romano CA, Sontz PA, Barton JK (2011) Mutants of the base excision repair glycosy-
lase, endonuclease III: DNA charge transport as a first step in lesion detection.
Bio-
chemistry
50:6133
–
6145.
24. Gorbalenya AE, Koonin EV (1993) Helicases: amino acid sequence comparisons and
structure-function relationships.
Curr Opin Struct Biol
3:419
–
429.
25. Fan L, et al. (2008) XPD helicase structures and activities: insights into the cancer and
aging phenotypes from XPD mutations.
Cell
133:789
–
800.
26. Boal AK, et al. (2005) DNA-bound redox activity of DNA repair glycosylases containing
[4Fe4S] clusters.
Biochemistry
44:8397
–
8407.
27. Yavin E, et al. (2005) Protein-DNA charge transport: redox activation of a DNA repair
protein by guanine radical.
Proc Natl Acad Sci USA
102:3546
–
3551.
28. Lehmann AR (2008) XPD structure reveals its secrets.
DNA Repair
7:1912
–
1915.
29. Fuss JO, Tainer JA (2011) XPB and XPD helicases in TFIIH orchestrate DNA duplex open-
ing and damage verification to coordinate repair with transcription and cell cycle via
CAK kinase DNA repair.
DNA Repair
10:697
–
713.
30. Takayama K, et al. (1996) Defects in the DNA repair and transcription gene
ERCC2
(XPD)
in trichothiodystrophy.
Am J Hum Genet
58:263
–
270.
31. Taylor EM, et al. (1997) Xeroderma pigmentosum and trichothiodystrophy are
associated with different mutations in the XPD (ERCC2) repair/transcription gene.
Proc
Natl Acad Sci USA
94:8658
–
8663.
32. Rudolf J, Makrantoni V, Ingledew WJ, Stark MJR, White MF (2006) The DNA repair
helicases XPD and FancJ have essential iron-sulfur domains.
Mol Cell
23:801
–
808.
33. Wolski SC, et al. (2008) Crystal structure of the FeS cluster-containing nucleotide ex-
cision repair helicase XPD.
PLos Biol
6:1332
–
1341.
34. Mui TP, Fuss JO, Ishida JP, Tainer JA, Barton JK (2011) ATP-stimulated, DNA-mediated
redox signaling by XPD, a DNA repair and transcription helicase.
J Am Chem Soc
133:16378
–
16381.
35. Rudolf J, Rouillon C, Schwarz-Linek U, White MF (2010) The helicase XPD unwinds bub-
ble structures and is not stalled by DNA lesions removed by the nucleotide excision
repair pathway.
Nucleic Acids Res
38:931
–
941.
36. Sun HB, Yokota H (2000) MutS-mediated detection of DNA mismatches using atomic
force microscopy.
Anal Chem
72:3138
–
3141.
37. Sorel I, et al. (2006) The EcoRI-DNA complex as a model for investigating protein-DNA
interactions by atomic force microscopy.
Biochemistry
45:14675
–
14682.
38. Pastré D, et al. (2010) Specific DNA-protein interactions on mica investigated by atomic
force microscopy.
Langmuir
26:2618
–
2623.
39. Jiang Y, Marszalek PE (2011) Atomic force microscopy captures MutS tetramers initi-
ating DNA mismatch repair.
EMBO J
30:2881
–
2893.
40. Boon EM, Ceres DM, Drummond TG, Hill MG, Barton JK (2000) Mutation detection by
electrocatalysis at DNA-modified electrodes.
Nat Biotechnol
18:1096
–
1100.
41. Donahue WF, Turczyk BM, Jarrell KA (2002) Rapid gene cloning using terminator
primers and modular vectors.
Nucl Acid Res
30:e95.
42. Rudolf J (2006) Characterization of XPD from
Sulfolobus Acidocaldarius
: an iron-sul-
phur cluster containing DNA repair helicase. Thesis submitted for the degree of Doctor
of Philosophy (University of St. Andrews, Scotland, United Kingdom).
43. Wagenknecht HA, ed. (2005)
Charge Transfer in DNA: From Mechanism to Application
(Wiley-VCH, Weinheim).
44. Berg OG, Winter RB, von Hippel PH (1981) Diffusion-driven mechanisms of protein
translocation on nucleic acids, 1 Models and theory.
Biochemistry
20:6929
–
6948.
45. Braga PC, Ricci D, eds. (2004)
Atomic Force Microscopy Biomedical Methods and
Applications
(Humana Press, Totowa, NJ).
46. Liu H, et al. (2008) Structure of the DNA repair helicase XPD.
Cell
133:801
–
812.
47. Broughton BC, et al. (2001) Two individuals with features of both xeroderma pigmen-
tosum and trichothiodystrophy highlight the complexity of the clinical outcomes of
mutations in the XPD gene.
Hum Mol Genet
19:2539
–
2547.
48. Yavin E, Stemp EDA, O
’
Shea VL, David SS, Barton JK (2006) Electron trap for DNA-
bound repair enzymes: a strategy for DNA-mediated signaling.
Proc Natl Acad Sci
USA
103:3610
–
3614.
49. Nunez ME, Noyes KT, Barton JK (2002) Oxidative charge transport through DNA in
nucleosome core particles.
Chem Biol
9:403
–
415.
50. Lee PE, Demple B, Barton JK (2009) DNA-mediated redox signaling for transcriptional
activation of SoxR.
Proc Natl Acad Sci USA
106:13164
–
13168.
51. Augustyn KE, Merino EJ, Barton JK (2007) A role for DNA-mediated charge transport
in regulating p53: Oxidation of the DNA-bound protein from a distance.
Proc Natl
Acad Sci USA
104:18907
–
18912.
52. Merino EJ, Barton JK (2008) DNA oxidation by charge transport in mitochondria.
Biochemistry
47:1511
–
1517.
53. Klinge S, Hirst J, Maman JD, Krude T, Pellegrini L (2007) An iron-sulfur domain of the
eukaryotic primase is essential for initiation of RNA primer synthesis.
Nat Struct Mol
Biol
14:875
–
877.
54. Davis RW, Botstein D, Roth JR (1980)
Advanced Bacterial Genetics: A Manual for
Genetic Engineering
(Cold Spring Harbor Laboratory Press, Cold Spring Harbor).
55. Cunningham RP, et al. (1989) Endonuclease III is an iron-sulfur protein.
Biochemistry
28:4450
–
4455.
56. Horcas I, et al. (2007) WSXM: a software for scanning probe microscopy and a tool for
nanotechnology.
Rev Sci Instrum
78:013705.
Sontz et al.
PNAS
∣
February 7, 2012
∣
vol. 109
∣
no. 6
∣
1861
BIOCHEMISTRY
CHEMISTRY