Surprising repair activities of nonpolar analogs of 8-oxoG expose features of recognition and catalysis by base excision repair glycosylases

Article abstract of DOI:10.1021/ja208510m

Repair glycosylases locate and excise damaged bases from DNA, playing central roles in preservation of the genome and prevention of disease. Two key glycosylases, Fpg and hOGG1, function to remove the mutagenic oxidized base 8-oxoG (OG) from DNA. To inves

Full text of DOI:10.1021/ja208510m

Article
pubs.acs.org/JACS
Surprising Repair Activities of Nonpolar Analogs of 8-oxoG Expose
Features of Recognition and Catalysis by Base Excision Repair
Glycosylases
Paige L. McKibbin,†,§ Akio Kobori,‡,§ Yosuke Taniguchi,‡ Eric T. Kool,*,‡ and Sheila S. David*,†
†Department of Chemistry, University of California, Davis, Davis, California 95616, United States
‡Department of Chemistry, Stanford University, Stanford, California 94305, United States
S
* Supporting Information
ABSTRACT: Repair glycosylases locate and excise damaged bases from DNA, playing central roles in preservation of the
genome and prevention of disease. Two key glycosylases, Fpg and hOGG1, function to remove the mutagenic oxidized base 8-
oxoG (OG) from DNA. To investigate the relative contributions of conformational preferences, leaving group ability, enzyme-
base hydrogen bonding, and nucleobase shape on damage recognition by these glycosylases, a series of four substituted indole
nucleosides, based on the parent OG nonpolar isostere 2Cl-4F-indole, were tested as possible direct substrates of these enzymes
in the context of 30 base pair duplexes paired with C. Surprisingly, single-turnover experiments revealed that Fpg-catalyzed base
removal activity of two of the nonpolar analogs was superior to the native OG substrate. The hOGG1 glycosylase was also found
to catalyze removal of three of the nonpolar analogs, albeit considerably less efficiently than removal of OG. Of note, the analog
that was completely resistant to hOGG1-catalyzed excision has a chloro-substituent at the position of NH7 of OG, implicating
the importance of recognition of this position in catalysis. Both hOGG1 and Fpg retained high affinity for the duplexes
containing the nonpolar isosteres. These studies show that hydrogen bonds between base and enzyme are not needed for
efficient damage recognition and repair by Fpg and underscore the importance of facile extrusion from the helix in its damaged
base selection. In contrast, damage removal by hOGG1 is sensitive to both hydrogen bonding groups and nucleobase shape. The
relative rates of excision of the analogs with the two glycosylases highlight key differences in their mechanisms of damaged base
recognition and removal.
OG by catalyzing removal of OG bases from OG:C base pairs
within DNA.3,4 Subsequent trimming of the sugar fragment,
nucleotide insertion, and ligation serve to restore the original
G:C base pair.5
INTRODUCTION
■
DNA damage can threaten the integrity of the genome by
increasing the frequency of mutations and exacerbating
replication errors.1 Organisms possess a cornucopia of
damage-specific repair pathways that prevent the deleterious
consequences of DNA damage. One of these, the base excision
repair (BER) pathway, is primarily responsible for the
recognition and initiation of repair of modified nucleobases
arising from oxidation, alkylation, methylation, and deami-
nation reactions. The low redox potential of guanine makes it
particularly vulnerable to oxidative damage, a major product of
which is 8-oxo-7,8-dihydroguanine (8-OxoG or OG). The
incorporation of the oxo-substituent at the C-8 position of the
purine ring causes the nucleotide to preferentially adopt the syn
orientation (Figure 1A).2 In this conformation in the template
strand during DNA replication, OG codes like thymine, causing
adenine to be preferentially incorporated over cytosine. The
stable OG:A base pair, if allowed to persist in DNA, will
eventually produce a G→T transversion mutation (Figure 1A).
Base excision repair glycosylases, Fpg in Escherichia coli and its
human homologue, hOGG1, prevent mutations associated with
The mechanisms by which BER glycosylases locate subtle
alterations in DNA bases and catalyze N-glycosidic bond
cleavage with high efficiency and accuracy within the context of
a large excess of normal DNA bases are questions central to
their roles in the prevention of mutations and disease.1
Synthetically derived nucleotide analogs are particularly useful
in measuring how specific modifications to an enzyme substrate
impact recognition, excision, and overall repair. Nonpolar
isosteres have been used to investigate the importance of
hydrogen bonding within varied enzyme DNA complexes while
preserving the nucleobase shape.2,6−11 Previous work estab-
lished that such nonpolar nucleotide analogs can effectively
mimic the natural nucleotides when used to evaluate DNA
polymerase activity.8−10,12 The hydrophobic adenine analogs, 4-
Received: September 8, 2011
Published: December 14, 2011
© 2011 American Chemical Society
1653
dx.doi.org/10.1021/ja208510m | J. Am. Chem.Soc. 2012, 134, 1653−1661

Journal of the American Chemical Society
Article
Figure 1. Structures of OG(anti):C(anti) pair and OG(syn):A(anti) mispair in DNA (A). Structures of OG deoxynucleoside and of OG analogs 1−
4 (B). Nucleosides are drawn with their preferred glycosidic orientation (i.e., OG,1,2 are syn; 3,4 are anti). The standard numbering scheme for
purines and indoles is shown for key positions analyzed in this work.
methylbenzimidazole (B) and 9-methyl-1H-imidazo-[4,5-b]-
pyridine (Q) were also shown in previous studies with Fpg to
be useful at revealing features of base excision and recognition.7
Fpg was found to cleave OG faster when paired opposite these
A analogs, which suggested that the lack of hydrogen bonds
between the base pair facilitates OG flipping into the Fpg active
site.7
mirrors the structure of OG closely, it lacks basic atoms and
canonical hydrogen bonding groups, and so is ideal for testing
the importance of such groups in N-glycosidic bond hydrolysis.
To further test mechanisms of recognition, we synthesized and
studied a series of three additional related nonpolar analogs
(2−4) that alter the electron withdrawing abilities and/or the
positioning of indole substituents. This would also allow for
discernment of the relative importance of base shape, hydrogen
bonding capacity and glycosidic bond conformation in substrate
recognition and catalysis by Fpg and hOGG1.
Here we report the surprising finding that these two BER
glycosylases can recognize and cleave nonpolar OG mimics,
despite their lack of hydrogen bonding groups and basic atoms.
Most remarkably, we observed for Fpg an approximate 2-fold
increase in the rate of excision of analogs 2 and 3 relative to OG.
In contrast, hOGG1 was able to recognize and initiate cleavage
of three of the nonpolar analogs, but at levels that are
significantly reduced compared to those for OG. The results
suggest that Fpg does not require activation of the leaving
group via hydrogen-bonding or base protonation in catalysis of
base excision, whereas hOGG1 does gain an advantage from
polar/hydrogen bonding atoms. Neither enzyme was strongly
influenced by the syn/anti base conformational preference nor
by the leaving group ability of the base. Finally, one analog (4)
was resistant to hOGG1 enzymatic cleavage but was still found
to bind tightly to DNA containing it paired opposite C. The
results provide useful new insights into the mechanisms of
damage repair by these two central enzymes.
Although considerable structural and mechanistic study has
focused on these enzymes, the salient features allowing for
efficient damaged base recognition and excision by Fpg and
hOGG1 remain unclear. A structure of Fpg bound to DNA
shows the OG base oriented in the syn conformation, with
hydrogen bonds to NH7 and to amino and imino groups along
the Watson−Crick pairing edge.13 In addition to OG, Fpg
cleaves a wide variety of oxidized guanine and thymine bases
with a common feature being the presence of a carbonyl group
adjacent to the C−N glycosidic bond. This common substrate
feature has prompted the hypothesis that this carbonyl group
acts as an electron acceptor, allowing the base to leave the
sugar.14 In contrast, structures of hOGG1 bound to DNA show
the OG base in the anti conformation, with hydrogen bonds
from the enzyme using amino acid residues as well as tightly
bound water molecules to the amino and imino groups and the
N7H of OG.15 Despite the structural studies with these two
enzymes, it is not known how the syn preference of OG relates
to the efficiency of cleavage by either of these enzymes, and it is
not known to what degree the ability to form hydrogen bonds
affects the recognition of the damaged base. Similarly, it is not
clear how critical the presence of the 8-oxo carbonyl group of
the substrate base is in promoting leaving of the base, nor is it
known the extent and site(s) at which the bases are protonated
during the excision process.
To address these questions, we have designed a set of
nonpolar OG analogs with varied conformational and leaving
group abilities. Recently, a synthetic nonpolar isostere of OG,
2-chloro-4-fluoroindole (1), was shown to adopt the syn
glycosidic bond conformation and code like T in polymer-
ization reactions with Klenow fragment of E. coli DNA
polymerase Pol 1 (KF exo-) (Figure 1B).2,16 Because it was
shown to effectively behave like OG in polymerase reactions,
we sought to determine if this nonpolar isostere would mimic
OG with the BER glycosylases that are involved in repair of
OG, specifically Fpg and hOGG1. Although the compound
RESULTS
■
Design, Synthesis, and Structure of OG Analogs 1−4.
The nonpolar OG analog nucleotides 1−4 (Figure 1B) were
designed to test a number of questions regarding the
mechanisms by which the glycosylases Fpg and hOGG1
recognize the damaged OG base and effectively release it from
the C-1′ of deoxyribose in DNA. First is the question of the
contribution of hydrogen bonding to enzyme recognition and
base excision catalysis. The four analogs lack all the imino,
amino, and carbonyl groups of OG; thus if hydrogen bonds are
mechanistically important, the compounds would be expected
to be very poor substrates. Second is the issue of whether the
inherent syn preference of OG affects its ability to be processed
by the enzymes; thus we included syn- (1, 2) and anti-oriented
1654
dx.doi.org/10.1021/ja208510m | J. Am. Chem.Soc. 2012, 134, 1653−1661

Journal of the American Chemical Society
Article
Figure 2. (A) Representative plot of extent of Fpg catalyzed removal of OG and analogs 1−4 as a function of time under conditions of single-
◆
■
●
▼
▲
turnover from 30 base pair duplexes containing OG:C ( ), 1:C ( ), 2:C ( ), 3:C ( ), and 4:C ( ). The inset shows region of Fpg experiments
expanded to 1 min time scale. Experiments were performed using 200 nM Fpg and 20 nM duplex DNA at 37 °C. (B) Representative plot for extent
of hOGG1 catalyzed removal of OG and nonpolar OG analogs 1−4 as a function of time under conditions of single-turnover from 30 base pair
◆
■
●
▼
▲
●
duplexes containing OG:C ( ), 1:C ( ), 2:C ( ), 3:C ( ), and 4:C ( ). The inset shows region of OG:C ( ) fit expanded to 1 min time scale.
Experiments were performed using 200 nM hOGG1 and 20 nM duplex DNA at 37 °C. (C) Measurement of Fpg affinity for a DNA duplex
containing compound 1:C ( ) pair. The dissociation constant Kd was determined from kobs measurements as a function of [Fpg] performed under
◆
STO conditions with 5 pM duplex DNA.
(3, 4) analogs to test this. The third question being addressed is
how the base leaves with negative charge; 8-oxoG and related
damaged bases (e.g Fapy and hydantoin lesions) might require
the oxo group to accept electrons and stabilize charge buildup.
Because the analogs 1−4 lack any basic atoms to accept the
developing charge, they would be expected to be resistant to
enzyme-mediated cleavage if such an acceptor were required. In
a related issue, compound 1 has an electronegative chlorine
adjacent to the nitrogen leaving atom of the indole, which
might enhance leaving group ability by decreasing basicity of
the anion; to control for this effect we can compare 1 and 4
(which also has a chlorine in the ring) to 2 (which has a
similarly sized, but nonelectronegative, methyl group replacing
chlorine).
exhibited first-order kinetics and proceeded to completion. The
data were fit to a single-exponential equation to determine kobs
,
which under these pseudo-first-order conditions allows for kobs
to be simplified to the rate constant, kg, which includes all steps
involving base excision (see Scheme 3 in the Supporting
Information). Representative reaction progress curves of the
reaction of Fpg with DNA duplexes containing OG:C, 1:C,
2:C, 3:C, and 4:C base pairs are shown in Figure 2A, and rate
constants are listed in Table 1. Surprisingly, the rate constants
Table 1. Summary of Rate Constants for Base Excision (kg)
and Equilibrium Dissociation Constants (Kd) Determined
for Fpg and hOGGl with Nonpolar Analogs
equilibrium dissoctation
constants (Kd) (nM)
Compounds 1 and 3 were known previously but neither has
been studied as a BER glycosylase substrate. These were
synthesized following published methods.16,17 Compounds 2
and 4 are unknown in the literature; we developed synthetic
approaches to them from commercial 4-fluoroindole. Schemes
and methods are given in detail in the Supporting Information
file. We established the glycosidic conformations of all four
nucleosides by 2-D NOESY methods; strong nuclear Over-
hauser enhancements from the indole C-6 proton to the 5′
protons on deoxyribose signaled syn conformation, as expected
from the 2-substitution, whereas similarly strong enhancements
to the 1′ proton were consistent with a normal anti orientation
of 3 and 4, which lack this substitution. Details of the
conformational studies are given in the Supporting Information.
Glycosylase Activity of Fpg with DNA Substrates
Containing OG or Analogs 1−4. The glycosylase activity of
Fpg was evaluated in vitro using a 30 base pair duplex
containing a central X:C base pair (X = OG, 1−4) (see
sequence in the Experimental Section). The general method
involved 5′-end-labeling the X-containing strand with [γ-32P]-
ATP and monitoring by denaturing PAGE the extent of strand
cleavage at the X-nucleotide upon quenching with 0.5 M
NaOH. The base treatment ensured that cleavage occurred at
all abasic sites and that the ability of Fpg to remove the
modified bases was evaluated, rather than an alteration in the
associated β or δ-lyase activities. Under conditions of single-
turnover (STO) where [Fpg] > [DNA], the reaction profiles
−1
rate constants (kg) (min )
a
a
d
e
substrate
Fpg
14
16 0.2
hOGG1
20
Fpg
hOGGl
f
OG
1
3
3
0.46 0.03
0.04 0.01
0.56 0.04
0.8 0.2
ND
b
b
g
g
g
0.09 0.01
0.20 0.02
<0.003
<0.003
<0.003
2
23
31
1
3
b
3
1.0 0.2
c
4
1.3 0.1
NC
0.05 0.01
0.04 0.02
a
Rate constants determined at 37 °C under single-turnover conditions
with OG and nonpolar analogs base paired to cytosine in 30 bp duplex.
These reactions did not go to completion, and these rates are based
on the fits; see Figure 2B. NC = no detectable cleavage (<5% after
background correction at all time points). Kd values determined from
glycosylase reaction at 37 °C (see methods). Kd values determined
using EMSA at 25 °C (see methods). ND = not determined. Upper
limit estimation: too tight to measure accurately, Kd is estimated to be
lower than the [enzyme] used. Errors reported are the standard
deviations of the average of at least three independent trials.
b
c
d
e
f
g
for Fpg-catalyzed removal of 2 and 3 are approximately 2-fold
greater (kg = 23 1 and 31 3 min−1, respectively) than the
corresponding rate constant measured for OG removal (14
3
min−1), establishing these analogs as better substrates for Fpg
than the natural substrate OG. The rate constants measured for
the duplexes containing 1:C and 4:C (kg = 1.6 0.2 and kg =
1.3 0.1 min−1) indicate that both chloro-containing analogs
are good substrates for Fpg, albeit approximately 10-fold slower
in this duplex context than OG (Table 1).
1655
dx.doi.org/10.1021/ja208510m | J. Am. Chem.Soc. 2012, 134, 1653−1661

Journal of the American Chemical Society
Article
Glycosylase Activity of hOGG1 with DNA Substrates
Containing OG or Analogs 1−4. The glycosylase activity of
hOGG1 with the nonpolar OG analogs was carried out in a
manner similar to that with Fpg. The hOGG1-catalyzed OG
removal from an OG:C base pair in this duplex context is
efficient (kg = 20 3 min−1) and is similar to previous reports
of the reaction of hOGG1 with this base pair in different duplex
sequence contexts.18,19 However, the reactions in this duplex
under the analogous conditions with the nonpolar isosteres
paired with C were considerably less efficient (Figure 2B and
Table 1) and did not go to completion. The incomplete
reactions were unanticipated because the same duplexes were
converted completely to product by Fpg (Figure 1A). Rate
constants listed in Table 1 were obtained from fitting of the
observed progress curves. Clearly all of the nonpolar isosteres
are removed less efficiently than OG. In the case of the duplex
containing analog 3, the measured rate constant (kg = 1.0 0.2
min−1) is approximately 20-fold reduced compared to the
corresponding rate constant with the OG:C substrate. The
rates of excision of the corresponding duplexes with analogs 1
and 2 when base paired to C by hOGG1 were found to be
approximately 200-fold reduced compared to those for OG (kg
= 0.09 0.01 min−1 and kg = 0.20 0.02 min−1, respectively).
Albeit reduced compared to that for OG, the ability to observe
any cleavage of these nonpolar isosteres was surprising due to
the narrow substrate preference of hOGG1 (e.g., primarily OG
and FapyG).18−20 This suggests that the correct size and
conformation of 1, 2, and 3 is enough to at least partially fulfill
the OG-like lesion requirements in the active site to allow for
observation of glycosidic bond cleavage. In contrast, no
detectable cleavage was measured with the duplex containing
analog 4. This shows that 3-chloro-substitution on the indole
ring is particularly deleterious to hOGG1-catalyzed cleavage.
Equilibrium Dissociation Constants of Fpg with
Nonpolar Analog-Containing DNA Measured with
Glycosylase Activity. Under single turnover (STO) con-
ditions the dissociation constant (Kd) is equal to the enzyme
concentration when the kobs equals half of the kg.21 Therefore, it
is possible to evaluate kobs as a function of varied enzyme
concentration under these conditions and determine the
substrate Kd. We used this approach to determine the Kd of
Fpg for DNA duplexes containing 1−4 or OG paired opposite
C, at varied [Fpg] ranging from 12.5 pM to 20 nM with 5 pM
duplex. Low DNA concentrations were used to ensure a
[DNA] below the Kd value. Concentrations of Fpg greater than
lead to high affinity enzyme recognition are not necessarily the
same as those that lead to facile glycosidic bond cleavage, such
as inherent lability of the base or base analog (vide infra).
Determination of Equilibrium Dissociation Constants
(Kd) for hOGG1 with Nonpolar Analog-Containing
Duplexes. Due to the poor or absent activity of hOGG1
with the nonpolar isostere duplexes, we were unable to measure
dissociation constants using the glycosylase activity. However,
the lack of activity hOGG1 with the duplex containing the
nonpolar analog 4, permitted determination of the relative
equilibrium dissociation constants (Kd) using electrophoretic
mobility shift assays (EMSA). Previously, we have shown that
hOGG1 has a strong affinity for duplexes containing the
noncleavable nucleotide FOG opposite C and therefore this
duplex provides an excellent benchmark for high affinity
hOGG1 binding.22 EMSA studies were performed under
conditions where [DNA-duplex] < Kd, to detect the
[hOGG1-DNA complex] as a function of [hOGG1] and the
resulting data were fit to a one-site binding isotherm. The Kd
values for duplexes containing FOG:C (2 1 nM) relative to
4:C (0.04
0.02 nM) indicate a higher affinity for the
nonpolar analog 4 than the 2′F version of the natural substrate
(Table 1). Indeed, the extremely high affinity of the 4:C-
containing duplex clearly shows that reduced binding is not the
origin of the low activity with this particular analog. In the case
of duplexes containing 1:C, 2:C, and 3:C, the extent of
conversion to product during the incubation period for the Kd
experiments is relatively small (15−25%) such that we used
EMSA to estimate the relative dissociation constants for these
analog-containing duplexes as well. In these experiments, the
duplexes containing analogs 1:C, 2:C, and 3:C were completely
bound to hOGG1, even at the lowest enzyme concentration
tested (3 pM). This suggests an extremely high affinity for
hOGG1 with Kd value that is lower than 3 pM. The observation
of complete and tight binding of hOGG1 to these duplexes also
shows that incomplete processing of these duplexes cannot be
due to lack of binding to the analog duplex. An experiment with
the OG:C substrate under the same conditions showed similar
behavior; however, in this case the high affinity would be due to
the binding to the product (∼85% product under these
conditions). Importantly, these results reveal that reduced
excision for analogs 1, 2, and 3 and lack of excision for 4 is not
due to an inability of hOGG1 to recognize the nonpolar base
isosteres within DNA.
Acid-Catalyzed Depurination of Nonpolar Analog 1−
4 Containing DNA. The base excision mechanism for DNA
glycosylases has been suggested to be a highly dissociative SN1
mechanism involving the formation of an oxacarbenium ion
intermediate based on kinetic isotope studies.23,24 The ability of
the base to be a leaving group, either in the protonated state or
as an anion, would be an important factor in the efficiency of
the reaction. To examine the susceptibility of the nonpolar
analogs to acid-catalyzed depurination, the 5′-32P-end-labeled ss
DNA containing the analogs 1−4 were subjected to a modified
Maxam−Gilbert G+A sequencing reaction utilizing piperidine-
formate to initiate depurination, followed by standard
piperidine treatment to reveal abasic sites.25 The extent of
depurination along the DNA was resolved with gel electro-
phoresis and quantified using storage phosphor autoradiog-
raphy (see the Supporting Information). The relative extents of
depurination of 1−4 as well as G and OG located at position 16
of the oligonucleotide sequence were normalized to the
depurination of guanine at position 12 in each sequence
20 nM did not result in significant increases in the kobs.
A
representative plot of kobs for the 1:C duplex as a function of
Fpg enzyme concentration is shown in Figure 2C, and binding
data are given in Table 1. The data show that Fpg binds tightly
to all four nonpolar analogs in this study. The Kd of Fpg for
OG:C, 2:C, and 3:C is respectively 460 30, 560 40, and
800
200 pM, with the enzyme showing similar affinity to
DNA containing the analogs or OG. Significantly, the Kd with
chlorinated analogs 1:C and 4:C was 41 8 and 50 10 pM
(Table 1); thus Fpg binds over 10-fold more tightly to the
DNA containing these latter analogs than to the natural
substrate in this context. It is noteworthy that the two nonpolar
isosteres that exhibit the highest affinity for Fpg are not the
ones that are removed with highest efficiency. These results
show that interactions within the Fpg base binding site are
extremely sensitive to the electronic characteristics of the
nonpolar base analog, and presumably also the natural damaged
bases. Moreover, these results illustrate that characteristics that
1656
dx.doi.org/10.1021/ja208510m | J. Am. Chem.Soc. 2012, 134, 1653−1661

Journal of the American Chemical Society
Article
(Figure 3). The results reveal that OG is the most easily
depurinated, yielding 2-fold more depurination than G under
interactions reduced the kinetic rate of base excision but did
not obliterate substrate recognition.7 In addition, Fpg was
found to cleave OG faster when paired with nonpolar adenine
analogs than from an OG:A base pair. This was attributed to
the lack of hydrogen bonds between the bases in the target base
pair leading to more facile flipping of the OG base into the Fpg
active site.7 Indeed, disruption of proper base pairing leading to
increased base excision activity has been observed with several
glycosylases.26−28 The present studies are distinct in that the
designed nonpolar analogs are mimics of the damaged base.
The nonpolar isostere of OG, analog 1, was earlier shown to be
an effective mimic of OG in its ability both to adopt the syn
orientation in DNA and to miscode as thymine in polymerase
studies.2 The preferred syn orientation of OG has long been
suggested to be an important recognition feature utilized by
glycosylases in lesion detection.14 Here we find for the first
time that a number of nonpolar OG analogs act as surprisingly
good substrates for enzymatic deglycosylation, despite their lack
of polar and basic atoms.
The activity of hOGG1 on this series of nonpolar analogs
provides insight into how this enzyme accurately selects OG for
excision. The enzyme has been shown to be highly selective for
OG and formamidopyrimidine lesions within specific base pair
contexts, preferably excising the lesions when base paired to
cytosine.19 Here we find that the nonpolar analogs 1, 2, and 3
can in fact be excised by hOGG1, but at a reduced efficiency
compared to that of OG. Notably, the 3-chlorinated analog 4
was not cleaved by hOGG1 in DNA. The substantial reduction
in cleavage rate that results with these analogs lacking polar
groups suggests a substantial requirement of the enzyme for
hydrogen-bonded interactions in its cleavage mechanism. A
wealth of crystallographic studies of hOGG1 bound to an
OG:C containing DNA duplex have revealed many of the
intricacies of OG-specific recognition mediated by this enzyme
(Figure 4A).1 There is a critical OG-specific contact between
the Gly42 carbonyl oxygen of hOGG1 and the NH7 of
OG.29,30 With G, this results in an unfavorable interaction
between the lone pair on N7G and the backbone carbonyl O of
Gly42. In the case of the nonpolar analogs 1−3, a hydrophobic
C−H group replaces NH7 of OG and therefore lacks the
Figure 3. Histogram illustrating the extent of depurination of G, OG,
and analogs 1−4 at position 16 of a 30 base pair duplex normalized to
extent of cleavage of G at position 12. Quantitation of data from
several experiments provided the relative % extents of cleavage: G, 1.1
0.1; OG, 2.3 0.6; 1, 0.5 0.1; 2, 1.4 0.2; 3, 0.9 0.3; 4, 0.6
0.1.
these conditions (Figure 3). The extents of acid-catalyzed
depurination of analogs 2 and 3 were within error of that of G,
suggesting that 2 and 3 have comparable sensitivity to G
despite the lack of basic proton acceptors. Lastly, the
chlorinated analogs 1 and 4 exhibited approximately half the
susceptibility of G to acid-catalyzed depurination, suggesting
that they are relatively resistant to depurination (Figure 3)
despite containing electronegative substituents.
DISCUSSION
■
The study of the “GO” repair glycosylases hOGG1 and Fpg
through the use of synthetically derived nonpolar nucleoside
analogs of the OG damaged base provides a new approach to
examine the recognition and catalytic properties of these two
enzymes.9,10 Previously, nonpolar adenine analogs were used
only as a partner of OG to study the adenine glycosylase, MutY,
revealing that the lack of possible hydrogen bonding
Figure 4. Hydrogen bonding interactions of OG observed in X-ray crystal structure of K249Q hOGG1 and E3Q Fpg bound to an OG:C containing
duplex. DNA is shown in yellow with oxygen (red), nitrogen (blue), and phosphorus (purple). Important residues are labeled and shown in gray
with oxygen (red), nitrogen (blue), and sulfur (orange). Hydrogen bonds are shown as dotted lines. (A) The view of the base-specific pocket of
hOGG1, showing residues involved in hydrogen bonding interactions in recognition of OG. (B) The view of the base-specific pocket of Fpg,
showing residues involved in hydrogen bonding interactions in recognition of OG. Images generated from the pdb files 1EBM (hOGG1) and 1R2Y
(Fpg) from the Worldwide Protein Data Bank based on data from refs 13 and 15.
1657
dx.doi.org/10.1021/ja208510m | J. Am. Chem.Soc. 2012, 134, 1653−1661

Journal of the American Chemical Society
Article
favorable hydrogen-bonding interaction; however, it also lacks
the unfavorable interaction that would be present for G, thus
providing a rationale for the higher activity relative to G.
Another feature of hOGG1 leading to OG specific recognition
is a complementary dipole−dipole interaction for OG created
and provides an additional checkpoint to prevent inappropriate
bases from accessing the OG binding site.
In contrast to hOGG1, Fpg was found to efficiently remove
all of the analogs examined in this study when paired opposite
C. Remarkably, Fpg cleaved the analogs 2 and 3 with an
approximately 2-fold increase relative to OG. Analogs 1 and 4
were also good substrates for Fpg with cleavage rates only ∼10-
fold reduced relative to OG:C. The ability of Fpg to catalyze
the removal of all of the nonpolar analogs suggests that neither
the presence of the OG polar groups nor any syn/anti
conformational preference are critical for activity. We find that
Fpg functions extremely well without making hydrogen bonds
to the base and can flip base conformation with little cost.
Indeed, Fpg seems to be more sensitive to the types of
substituents on the indole ring rather than their orientation,
because the chlorinated compounds were slower substrates
than the nonchlorinated ones. In addition to efficient catalysis,
the glycosylase was also able to capably bind to all four
nonpolar analogs in DNA, again confirming that aberrant base
recognition by this enzyme can occur regardless of hydrogen
bonding groups or base conformation. Again the types of
substituents rather than their positition influenced Fpg binding
the most with the chlorinated analogs binding more tightly than
the nonchlorinated version. However, high affinity did not
translate to the most efficient removal, likely due to the
additional complication of altering base lability by the addition
of chloro substituents. The lack of influence of the nonpolar
isostere conformation on activity with Fpg is also consistent
with the observation in structural studies of OG and a
carbacyclic-FapyG lesion located within the same binding
pocket in syn versus anti conformations, respectively.13,35
The presence of atoms corresponding to the purine NH7
and O6 of OG are common features of many Fpg substrates,
such as the formamidopyrimidines, and the hydantoin lesions,
guanidinohydantoin and spiroiminodihydantoin.5,18,36 In the
structure of Fpg with OG, there is an OG-specific hydrogen
bond between the main chain amide of Ser220 and the NH7
whereas G and OG-specific contacts are provided by several
hydrogen bonds to both the 2-amino and 1-imino positions of
OG (Figure 4B). Moreover, an important feature of lesion-
containing structures of Fpg is ordering of the βFα10 peptide
loop to position the main chain amides for hydrogen-bonding
to the O6 atom of OG (Figure 4B). Despite this network of
hydrogen-bonding and polar interactions, the current results
show that they apparently add little, if anything, to stabilization
of the transition state needed for facile base cleavage. Indeed,
such interactions may be more important for OG base
recognition rather than glycosidic bond cleavage and therefore
may only be needed for processing OG and similar lesions.
The overall process leading to lesion excision mediated by
DNA glycosylases requires multiple steps, including initial
nonspecific binding to DNA, encounter with the lesion base
pair, nucleotide flipping, and glycosidic bond cleavage.1,37
Under single-turnover conditions, the rate constant kg describes
the process of ES to E-abasic site product, including all steps
subsequent to the initial encounter of the enzyme with the
lesion-containing base pair. The overall efficiency of excision of
the nonpolar isosteres by Fpg will be related to the
consequences of the modified substrate on all of these various
steps in the process. Because nonpolar base analogs generally
exhibit low pairing affinity for polar natural bases,2,38 the
current nonpolar analogs are expected to be more easily
extruded from the helix than OG, and this would provide for an
by Lys-249-NH3 and Cys-253-S− within the lesion-binding
+
site.30 This dipole−dipole interaction would be lacking in the
active site with the nonpolar analogs, contributing to less
efficient cleavage activity. Several other OG interactions
observed in hOGG1 structures include stacking interactions
with Phe319 and hydrogen-bonding to the Watson−Crick face
(N1, 2-NH2) by Gln315 (Figure 4A).15 Consistent with the
importance of Gln315-mediated hydrogen bonding, Q315A
hOGG1 exhibits no OG glycosylase activity;31 however, this
apparent critical interaction is absent using the nonpolar
analogs, establishing that this contact is not required to support
base cleavage and may be more critical for base recognition
(e.g., damaged G over A) and proper orientation within the
active site pocket. In addition, our analog studies suggest that
syn versus anti geometry does not have a discernible influence
on OG recognition or catalysis by hOGG1, because syn and
anti analogs were cleaved at a similar rate.
A particularly revealing result with hOGG1 was that one
analog (4) was completely resistant to base cleavage by the
enzyme. On the basis of the structural studies, the presence of a
3-chloro substituent on the indole at the position equivalent to
purine NH7 likely provides an unfavorable interaction.
Surprisingly, however, the measured dissociation constant
with the 4:C-containing DNA, showing considerably greater
affinity than with substrate analog FOG, indicates that this
steric clash, if present, does not prevent high-affinity binding to
hOGG1. We surmise that the lack of cleavage may be due to
incomplete or inappropriate engagement of the base within the
OG-specific pocket compared to the case of OG and the other
OG nonpolar isosteres (e.g., 1, which is shaped very similarly to
OG). Indeed, duplexes containing 1:C, 2:C, and 3:C exhibited
higher affinity for hOGG1 than 4:C (Table 1). In addition, the
incomplete processing of these three duplexes, despite high
affinity, suggests considerable nonproductive binding of
hOGG1 to these nonpolar analogs as well. These results are
reminiscent of the structure of Q315F hOGG1 bound to an
OG:C duplex, where the OG residue was found to be almost
but not completely inserted in the active site pocket.32 Notably,
Q315F hOGG1 was unable to cleave OG from OG:C base
pairs.32 Similarly, structures of hOGG1 bound to a product
analog show that in the absence of OG, key catalytic residues
are improperly positioned, indicating a strong coupling between
substrate recognition and catalysis.33 Furthermore, recent
studies from our laboratory have shown that hOGG1 exhibits
a preference for a pyrrolidine transition state analog harboring a
benzyl substituent to mimic the base over the analogous analog
lacking the substituent.34 The results herein further illustrate
the high level of quality control utilized by hOGG1 to ensure
proper base excision by requiring proper alignment of the target
base within the active site to attain the transition state needed
for cleavage. The presence of the bulky chloro substituent in 4
at the key recognition site may alter the base docking within the
base specific pocket and positioning of active site residues,
thereby thwarting catalysis. Alternatively, the nonpolar analogs
may be tightly bound in an alternative “exo” site that has been
shown in structural studies of hOGG1 with a nonspecific
duplex to bind G.30 Binding at this site would be nonproductive
1658
dx.doi.org/10.1021/ja208510m | J. Am. Chem.Soc. 2012, 134, 1653−1661

Journal of the American Chemical Society
Article
increased excision rate compared to that for OG if all other
steps in the base excision process remained unaltered. In
addition, the facile extrusion of the nonpolar analogs 1−4 may
explain the lack of observable dependence of the rates of base
excision due to a substituent at the position corresponding to
the 8-oxo group in OG. Indeed, with Fpg, analog 3, which lacks
a substituent that would favor the syn orientation, is the best
substrate in this series, and is superior to OG. The presence of
steric bulk at C8 (e.g., oxo or halogen) in purines destabilizes
the duplex relative to a normal base,39,40 suggesting that the 8-
oxo substituent may aid in lesion recognition by facilitating
lesion base-pair opening. Moreover, recent structural studies of
a mutated version of Fpg lacking a portion of βFα10 peptide
loop covalently trapped interrogating an OG:C base pair
provided compelling evidence that the 8-oxo-substituent of OG
provides the means for its interhelical detection.41 Specifically,
these studies illustrated that upon Fpg interrogation of OG:C
base pairs, a steric clash of the 8-oxo group with the DNA
backbone occurs that helps propel the OG out the helix into
the base specific pocket. In the case of the nonpolar analogs,
the lack of base pair hydrogen bonding likely removes the
reliance on this enzyme-mediated steric destabilization
provided by the 8-substituent of the syn nucleotide in the
initial recognition event. Once lodged within the base specific
pocket, there are no direct contacts of Fpg (or hOGG1) to the
8-oxo group of OG, which is also consistent with a lack of an
effect of substituents in the indole at this position on glycosidic
bond cleavage.
Another important aspect of base excision is the ability of the
enzyme to stabilize the departing base, in either anionic or
neutral protonated form. Indeed, kinetic isotope effect studies
on glycosylases to date indicate highly dissociative transition
states consistent with an SN1-type mechanism.24 In such a
mechanism, the nature and stability of the leaving nucleobase
anion would be expected to be an important factor. Indeed,
such a direct correlation has been documented for thymine−
DNA glycosylase.42 Our data measuring the leaving group
ability of the nonpolar analogs under acid-catalyzed depurina-
tion conditions revealed that the nonpolar isosteres are less
easily depurinated than OG, suggesting that the analogs are not
highly labile. However, they are surprisingly more easily
removed than might be expected; for example, analogs 2 and
3 exhibit susceptibility to depurination that is comparable to
that of G. Moreover, the nonpolar compounds are remarkably
well excised by Fpg. Most notably, analogs 1 and 4 are the least
susceptible to acid catalyzed depurination and are also the least
efficiently removed of the nonpolar analogs by Fpg. However,
the removal of these nonpolar analogs to any extent is
surprising because at first glance an indole is a much poorer
leaving group than OG. The carbonyl (at position 8 or 6) that
can accept negative charge in OG is missing in indole. Indeed,
the positioning of an array of amides in the βFα10 peptide loop
of Fpg toward the O6 of OG suggests a role in stabilizing the
developing negative charge at this position as the base is
excised.13 How, then, does indole leave in an SN1-like
mechanism? Clearly it cannot leave with full negative charge,
as the indole anion is highly basic. We propose that, instead, the
indoles are protonated on carbon, at C3, analogous to the well-
known nucleophilicity of indoles at this position. If an acidic
proton from an amino acid side chain or from water were
nearby, this protonation could occur simultaneously with
leaving of the base, thus stabilizing the transition state. Also
consistent with this are our acid-catalyzed depurination data,
which show that all the indoles can be depurinated, but that the
less basic chlorinated cases are less well depurinated. In the
enzyme active site several different sources for the protonation
of OG have been previously proposed including the backbone
amides of the βFα10 loop, Glu 3, Glu 174, Lys 57, and Lys
155.13,43,44 Analysis of the consequences of mutations of the
βFα10 loop and other amino acids within the active site on
cleavage of OG versus the nonpolar analogs would be
enlightening. Our expectation is that mutations of amino
acids that participate in intrinsic aspects of catalysis of bond
hydrolysis will reduce excision of both OG and the nonpolar
analogs, whereas alterations of residues that participate in lesion
recognition and extrusion will only reduce activity with OG.
CONCLUSION
■
The superior activity of Fpg relative to hOGG1 with the
nonpolar substrates reveals distinct differences in mechanisms
that are intimately related to their distinct biological functions
requiring balance between accuracy and efficiency.20 Clearly,
the presence of polar groups, involved either in hydrogen
bonding or in stabilizing the anion, are much more important
for hOGG1 than for Fpg. This suggests that hOGG1 relies on
hydrogen bonds within the active site to aid in leaving group
departure and to couple correct base recognition with proper
positioning of catalytic residues. In contrast, Fpg utilizes a
flexible lesion recognition loop and active site to accommodate
a wide variety of different types of damaged bases while still
attaining the proper transition state for catalysis. To ensure
accuracy of base excision, a glycosylase may use an initial
recognition step to only allow specific bases to be extruded
from the helix and additional catalytic checkpoints to prevent
cleavage of inappropriate N-glycosidic linkages. The distinct
differences revealed herein and taken together with the wealth
of structural and biochemical information on these two
enzymes suggest that hOGG1 uses both extrusion and catalytic
checkpoints to ensure proper substrate base excision. In
contrast, Fpg appears to rely more on the extrusion step to
select the appropriate bases for excision. This illustrates that
fine-tuning of overall similar catalytic strategies allows for broad
versus specific substrate processing and the exquisite control
needed to preserve the genome.
EXPERIMENTAL SECTION
■
General Synthetic Methods. 2′-Deoxynucleoside compounds 1
and 3 were synthesized following published methods.16,17 New
nucleoside analogs 2 and 4 were synthesized from 4-fluoroindole
(see the Supporting Information for detailed methods and character-
ization data). 5′-DMT and 3′ phosphoramidite derivatives of all four
analogs were prepared using standard methods (see the Supporting
Information for details). 8-Oxo-7,8-dihydro-2′-deoxyguanosine (OG)
phosphoramidite was purchased from Glen Research whereas 8-oxo-
7,8-dihydro-9-(2′-deoxy-2′-fluoro-β-D-ribofuranosyl) guanine (FOG)
phosphoramidite was synthesized as reported previously.22 All
1
compounds were characterized by H and 13C NMR and by high-
1
resolution mass spectrometry. Conformations were measured by H
NOESY methods; details are given in the Supporting Information file.
Oligodeoxynucleotides containing analogs 1−4 were synthesized on
CPG solid support by standard automated methods on an ABI 392
synthesizer. They were purified by preparative polyacrylamide gel
electrophoresis and characterized by MALDI-mass spectrometry (see
the Supporting Information for details and data). DNA oligonucleo-
tides containing standard phosphoramidites were purchased from
Integrated DNA technologies. Oligodeoxynucleotides containing the
OG and FOG phosphoramidites were synthesized at the University of
Utah core facility. The oligonucleotides containing natural bases, FOG,
1659
dx.doi.org/10.1021/ja208510m | J. Am. Chem.Soc. 2012, 134, 1653−1661

Journal of the American Chemical Society
Article
and OG were purified by ion-exchange HPLC. The 5′ radiolabeling
was done using [γ-32P]-ATP and T4 polynucleotide kinase following
standard protocols. Gels were imaged using a Typhoon Trio scanner.
Substrate Preparation for Glycosylase Assays. The following
duplex sequence was used in all glycosylase assays: 5′-CGA TCA TGG
AGG CTA XCG CTC CCG TTA CAG-3′:3′-GCT AGT ACC TCC
GAT YGC GAG GGC AAT GTC-5′ where X = 1−4, G, OG, or FOG,
and Y = C. Glycosylase assays were conducted using this duplex with
the 5′-hydroxyl of the X-strand labeled with γ-32P-[ATP] (see the
Supporting Information for details). For glycosylase assays, an
additional nonlabeled X-containing strand was added to the labeled
strand to produce a 5% labeled X-oligonucleotide. The was was then
annealed with a 20% excess of the complement by heating to 90 °C for
10 min in annealing buffer (20 mM Tris−HCl, pH 7.6, 10 mM EDTA,
and 150 mM NaCl) and then allowed to cool slowly overnight.
Enzyme Purification and Glycosylase Assays. Recombinant
Fpg and hOGG1 were purified and active enzyme concentrations
determined as described previously.19,36 All enzyme concentrations
listed are active enzyme concentrations. Single-turnover experiments,
where [enzyme] > [DNA], were performed using the 30 base pair
duplex sequence (vide supra) to evaluate the glycosylase activity of
Fpg and hOGG1.19,36,45 For reactions under single-turnover
conditions in which the glycosylase reaction was too rapid to measure
manually, a Rapid Quench Flow instrument (RQF-3) from Kintek was
used. The enzyme was mixed with 20 nM final DNA duplex for time
points ranging from 0.2 s to 1 min and quenched with 0.5 M NaOH.
Denaturing polyacrylamide gel analysis provided separation of the 15-
nucleotide DNA fragment arising from the product and the 30-
nucleotide fragment originating from the substrate. Gels were imaged
using storage phosphor autoradiography and band intensities
quantitated to provide binding plots. For kinetic determination of
the Kd values for Fpg with the lesion and analog containing DNA the
substrate concentration was 5 pM of 100% label X-strand and the
[Fpg] was varied between 20 nM and 12.5 pM. Values for Kd were
ACKNOWLEDGMENTS
■
We thank the National Institutes of Health (CA090689 to
S.S.D. and GM072705 to E.T.K.) for support. A.K. acknowl-
edges the KIT Faculty Research Abroad Fellowship Program.
REFERENCES
■
(1) David, S. S.; O’Shea, V. L.; Kundu, S. Nature 2007, 447, 941.
(2) Taniguchi, Y.; Kool, E. T. J. Am. Chem. Soc. 2007, 129, 8836.
(3) Michaels, M. L.; Miller, J. H. J. Bacteriol. 1992, 174, 6321.
(4) Michaels, M. L.; Cruz, C.; Grollman, A. P.; Miller, J. H. Proc. Natl.
Acad. Sci. U. S. A. 1992, 89, 7022.
(5) David, S. S.; Williams, S. D. Chem. Rev. 1998, 98, 1221.
(6) Dzantiev, L.; Alekseyev, Y. O.; Morales, J. C.; Kool, E. T.;
Romano, L. J. Biochemistry 2001, 40, 3215.
(7) Francis, A. W.; Helquist, S. A.; Kool, E. T.; David, S. S. J. Am.
Chem. Soc. 2003, 125, 16235.
(8) Kim, T. W.; Delaney, J. C.; Essigmann, J. M.; Kool, E. T. Proc.
Natl. Acad. Sci. U. S. A. 2005, 102, 15803.
(9) Mizukami, S.; Kim, T. W.; Helquist, S. A.; Kool, E. T.
Biochemistry 2006, 45, 2772.
(10) Sintim, H. O.; Kool, E. T. J. Am. Chem. Soc. 2006, 128, 396.
(11) Krueger, A. T.; Kool, E. T. Curr. Opin. Chem. Biol. 2007, 11, 588.
(12) Sintim, H. O.; Kool, E. T. Angew. Chem., Int. Ed. Engl. 2006, 45,
1974.
(13) Fromme, J. C.; Verdine, G. L. J. Biol. Chem. 2003, 278, 51543.
(14) Tchou, J.; Bodepudi, V.; Shibutani, S.; Antoscheckin, I.; Miller,
J.; Grollman, A. P.; Johnson, F. P. J. Biol. Chem. 1994, 269, 15318.
(15) Bruner, S. D.; Norman, D. P.; Verdine, G. L. Nature 2000, 403,
859.
(16) Taniguchi, Y.; Kool, E. T. Nucleic Acids Symp. Ser. (Oxford)
2007, 51, 217.
(17) Bozilovic, J.; Bats, J. W.; Engels, J. W. Can. J. Chem. 2007, 85,
283.
determined by fitting the observed rate of glycosylase activity (kobs
)
versus log [Fpg] to the one-site binding isotherm (GraFit 5.0).
(18) Krishnamurthy, N.; Haraguchi, K.; Greenberg, M. M.; David, S.
Equilibrium Dissociation Constant (Kd) Measurements.
Electrophoretic mobility shift assays (EMSA) were performed to
determine the Kd values of hOGG1 for lesion and analog containing
DNA similarly to previously reported.7 Reaction volumes of 30 μL
contained duplex DNA concentrations estimated at 1 pM. The
reaction contained DNA 32P 5′-end-labeled X-strand, 20 mM Tris−
HCl pH 7.6, 10 mM NaCl, 1 mM EDTA, 1 mM DTT, 10% glycerol,
0.1 mg/mL BSA, and enzyme concentrations ranging from 2.3 μM to
3 pM.
S. Biochemistry 2008, 47, 1043.
(19) Leipold, M. D.; Workman, H.; Muller, J. G.; Burrows, C. J.;
David, S. S. Biochemistry 2003, 42, 11373.
(20) O’Brien, P. J. Chem. Rev. 2006, 106, 720.
(21) Kundu, S.; Brinkmeyer, M. K.; Livingston, A. L.; David, S. S.
DNA Repair (Amsterdam) 2009, 8, 1400.
(22) Cao, S. Ph.D. Dissertation, University of Utah, 2010.
(23) McCann, J. A.; Berti, P. J. J. Am. Chem. Soc. 2008, 130, 5789.
(24) Berti, P. J.; McCann, J. A. Chem. Rev. 2006, 106, 506.
(25) Sambrook, J.; Russell, D. W. Molecular Cloning, A Laboratory
Manual, 3rd ed.; Cold Spring Harbor Laboratory Press: Plainview, NY,
2001; Vol. 1.
Acid-Catalyzed Depurination Assays. A modified Maxam−
Gilbert G+A sequencing reaction was performed as previously
described.7,25 Details are given in the Supporting Information.
(26) Lyons, D. M.; O’Brien, P. J. J. Am. Chem. Soc. 2009, 131, 17742.
(27) Krosky, D. J.; Song, F.; Stivers, J. T. Biochemistry 2005, 44, 5949.
(28) Krosky, D. J.; Schwarz, F. P.; Stivers, J. T. Biochemistry 2004, 43,
4188.
(29) Banerjee, A.; Verdine, G. L. Proc. Natl. Acad. Sci. U. S. A. 2006,
103, 15020.
(30) Banerjee, A.; Yang, W.; Karplus, M.; Verdine, G. L. Nature 2005,
434, 612.
(31) van der Kemp, P. A.; Charbonnier, J. B.; Audebert, M.; Boiteux,
S. Nucleic Acids Res. 2004, 32, 570.
(32) Radom, C. T.; Banerjee, A.; Verdine, G. L. J. Biol. Chem. 2006,
9182.
ASSOCIATED CONTENT
■
S
* Supporting Information
General synthetic methods and materials, synthetic methods for
synthesis of nonpolar analogs (2 and 4), mass spectrometry
characterization of oligonucleotides, enzyme purification and
assay methods, PAGE analysis of acid-catalyzed depurinations,
and NMR spectra. This information is available free of charge
via the Internet at http://pubs.acs.org/.
(33) Norman, D. P.; Bruner, S. D.; Verdine, G. L. J. Am. Chem. Soc.
2001, 123, 359.
(34) Chu, A. M.; Fettinger, J. C.; David, S. S. Bioorg. Med. Chem. Lett.
2011, 21, 4969.
(35) Coste, F.; Ober, M.; Carell, T.; Boiteux, S.; Zelwer, C.; Castaing,
B. J. Biol. Chem. 2004, 279, 44074.
(36) Leipold, M. D.; Muller, J. G.; Burrows, C. J.; David, S. S.
Biochemistry 2000, 39, 14984.
AUTHOR INFORMATION
■
Corresponding Author
E-mail: E.T.K., kool@stanford.edu; S.S.D., ssdavid@ucdavis.
edu
Author Contributions
§These authors contributed equally to the work, and should be
considered cofirst authors.
(37) Friedman, J. I.; Stivers, J. T. Biochemistry 2010, 49, 4957.
1660
dx.doi.org/10.1021/ja208510m | J. Am. Chem.Soc. 2012, 134, 1653−1661

Journal of the American Chemical Society
Article
(38) Kool, E. T.; Morales, J. C.; Guckian, K. M. Angew. Chem., Int. Ed.
2000, 39, 990.
(39) Hamm, M. L.; Rajguru, S.; Downs, A. M.; Chlorera, R. J. Am.
Chem. Soc. 2005, 127, 12220.
(40) Oka, N.; Greenberg, M. M. Nucleic Acids Res. 2005, 33, 1637.
(41) Qi, Y.; Spong, M. C.; Nam, K.; Banerjee, A.; Jiralspong, S.;
Karplus, M.; Verdine, G. L. Nature 2009, 462, 762.
(42) Bennett, M. T.; Rodgers, M. T.; Hebert, A. S.; Ruslander, L. E.;
Eisele, L.; Drohat, A. C. J. Am. Chem. Soc. 2006, 128, 12510.
(43) Rabow, L. E.; Kow, Y. W. Biochemistry 1997, 36, 5084.
(44) Lavurkhin, O. V.; Lloyd, R. S. Biochemistry 2000, 39, 15266.
(45) Porello, S. L.; Leyes, A. E.; David, S. S. Biochemistry 1998, 37,
14756.
1661
dx.doi.org/10.1021/ja208510m | J. Am. Chem.Soc. 2012, 134, 1653−1661