ABSTRACT
The mutagenic and lethal effects of abasic sites in DNA are averted by repair
initiated by `class II' apurinic (AP) endonucleases, which cleave immediately 5
'
to abasic sites. We examined substrate binding by the human AP endonuclease,
Ape protein (also called Hap1, Apex or Ref-1). In electrophoretic mobility-shift experiments, Ape bound synthetic DNA substrates containing single AP sites or tetrahydrofuran (F) residues. No complexes were detected with single-stranded substrates or unmodified duplex DNA. In EDTA, the concentration of Ape
required to shift 50% of duplex F-DNA was
~
50 nM, while the addition of 10 mM MgCl
2
nearly eliminated detectable F-DNA
@
Ape complexes. Filter-binding studies demonstrated a half-life of
~
50 s at 0
o
C for F-DNA
@
Ape complexes in the presence of EDTA, and <15 s after the addition of Mg
2+
. The DNA recovered from F-DNA
@
Ape complexes was intact but was rapidly cleaved upon addition of Mg
2+
, which suggests that these protein-DNA complexes are on the catalytic pathway for incision. Methylation and ethylation interference experiments identified DNA
contacts critical for Ape binding, and Cu-1,10-phenanthroline footprinting suggested an Ape-induced structural distortion at the abasic site prior to
cleavage.
Abasic sites in DNA arise via spontaneous or mutagen-induced hydrolysis of the N-glycosylic bond, or through the repair activity of DNA glycosylases (
1
,
2
). If left unrepaired, apurinic/apyrimidinic (AP) sites are potentially lethal or mutagenic (
3
). To cope with the deleterious consequences of AP sites, organisms possess AP
endonucleases that initiate the repair of these DNA lesions (
2
). Most notably, bacteria and yeast strains deficient in AP endonuclease activity display increased spontaneous mutation rates driven by AP site
formation (
4
-
7
).
`Class II' AP endonucleases initiate repair by catalyzing the hydrolysis of the
5'-phosphodiester of an abasic site to generate a 3'-OH group and a 5'-abasic residue (
2
). These enzymes also generate 3'-OH groups by removing fragmentary 3'-termini that arise from free radical attack on DNA, or
from spontaneous or protein-catalyzed [beta]-elimination reactions at AP sites. Class II AP endonucleases
form two protein families based on homology to
Escherichia coli
exonuclease III or endonuclease IV (
2
).
Using synthetic DNA substrates containing different AP site analogs, we have
found that exonuclease III and the homologous human AP endonuclease, Ape (
8
; also known as Hap1, Apex and Ref1;
9
-
11
), have near-identical substrate specificities (
12
,
13
) that differ in key ways from the specificities of endonuclease IV and the
related Apn1 protein of
Saccharomyces cerevisiae
(
14
). These findings corroborate other studies indicating differences in substrate
preference between exonuclease III or mammalian AP endonuclease and
endonuclease IV (
15
-
18
). Collectively, the data indicate (i) branching or oxidation at the 4-carbon inhibit exonuclease III and Ape strongly, but inhibit endonuclease
IV and Apn1 weakly or not at all; (ii) stereospecific effects of different
phosphorothioate diastereomers positioned on the 5' side of a tetrahydrofuran (F) residue. For all the enzymes, incision is
reduced significantly by a mismatch immediately 5' to F and only slightly affected by the base opposite F. Whether these
effects result from altered DNA binding or from changes in the rate of the
incision step remains unknown.
A key question of long standing has been the way in which repair proteins such
as Ape engage damaged DNA. Here we present methods for detecting complexes of
Ape protein with abasic DNA, and analyze the DNA contact sites in complexes of
Ape with an abasic residue. Together with results from enzymatic analysis (
13
) and structural studies (
19
,
20
), this new work helps establish a framework for unraveling the mechanism of
damage recognition and incision by these critical DNA repair enzymes.
[[gamma]-
32
P]ATP (3000 Ci/mmol) was purchased from DuPont NEN (Boston, MA), T4
polynucleotide kinase from New England Biolabs (Beverly, MA), and uracil-DNA glycosylase (UDG) from Gibco-BRL (Gaithersburg, MD). An oligonucleotide containing a single F (tetrahydrofuran, an AP site analog) residue was synthesized as described previously (
12
,
13
,
21
). A uracil-containing 18mer (U at position 10) was obtained from Operon Technologies, Inc. (Emeryville, CA).
Highly purified (>95% purity; ref.
22
) native Ape protein from HeLa cells was used unless otherwise specified. HeLa
Ape of >80% purity (
13
) was used for some initial experiments and is so indicated in the Figure
1
legend. Recombinant (glutathione
S
-transferase)-Ape (GST-Ape) fusion protein (>90% purity) was isolated from
E.coli
bearing plasmid pGEX-Ape essentially as described (
13
), but without Factor Xa cleavage. Recombinant native-size Ape (>95% purity) was isolated using chromatography on DEAE-Bio-Gel agarose (BioRad) and phosphocellulose (Whatman)
essentially as described (
13
).
Incision activity of Ape at the synthetic AP sites was determined in 10 [mu]l reactions in 50 mM HEPES-KOH (pH 7.5), 50 mM KCl, 100 [mu]g/ml BSA, 10 mM MgCl
2
, 0.05% Triton X-100 performed as described in (
13
).
Ape protein was incubated with 5'-
32
P-labeled duplex DNA substrates (
13
) for 5 min at 0oC in 10 [mu]l SBC buffer (50 mM HEPES-KOH, pH 7.5, 50 mM KCl, and 10% glycerol)
unless otherwise noted. The amount of residual single-stranded oligonucleotide present in these reactions was ~5% of the total substrate. Typically, 10 ng Ape protein (28 nM final concentration) was mixed with 0.05 pmol duplex DNA (5 nM final concentration). Binding reactions were resolved in non-denaturing polyacrylamide gels (8% acrylamide, 0.1% bis-acrylamide, 2.5% glycerol; 14 cm * 16 cm * 0.8 mm) in 20 mM Tris-HCl, pH 7.5, 10 mM
sodium acetate, pH 7.5, 0.5 mM EDTA, and electrophoresis was performed at 4oC for ~2 h at 20 mA. The gels were dried and autoradiographed to identify the
location of bound and unbound DNA. The percent of complexed DNA substrate was
determined by excising bound and free DNA, and assaying the gel slices for
32
P content using a Beckman LS1801 scintillation counter [percent bound = 100% * (c.p.m. of protein-DNA complex)/ (c.p.m. of protein-DNA complex + c.p.m. of unbound duplex DNA)].
For antibody `supershift' experiments, Ape-specific rabbit antiserum (
8
,
22
) was added 2 min after the addition of Ape to DNA binding mixtures, and the
incubation was continued at 0oC for an additional 20 min prior to electrophoresis. For competition
reactions, a 10-fold excess of competitor DNA was mixed with end-labeled DNA substrates prior to the addition of Ape protein for DNA
binding analysis.
Duplex DNAs with a centrally located uracil base in one strand were treated with
1 U UDG for 5 min at 37oC to generate a hydrolytic AP site. The resulting DNA was used directly in
binding experiments; UDG did not detectably bind this DNA substrate under our
conditions.
Nitrocellulose filters (0.45 mm; Schleicher & Schuell) were pre-treated with 0.4 M KOH for 40 min at room temperature, washed thoroughly
with distilled water and stored at 4oC in SBC buffer. Binding reactions in 10 [mu]l SBC supplemented with 1 mM EDTA contained 0.01 pmol duplex DNA and
various amounts of Ape protein. After a 2 min incubation at 0oC, the samples were applied rapidly to filters (
23
) and washed twice with 500 [mu]l SBC containing 1 mM EDTA. The filters were then dried and assayed for
bound
32
P as described above. To determine the half-life of the protein-DNA complexes, a 100-fold excess of unlabeled competitor DNA was added with or without 10 mM MgCl
2
and the percentage of bound DNA measured at the time points indicated.
Following non-denaturing gel electrophoresis, the gels were immediately autoradiographed
at -80oC to identify the location of free DNA and protein-DNA complexes. The DNA containing regions were then excised from the gel and submerged into 500
[mu]l elution buffer (0.3 M sodium acetate, pH 7.0, 1 mM EDTA) pre-heated to 65oC to inactivate Ape, followed by incubation at 65oC for 15 min and at room temperature overnight to elute the
DNA. In some experiments, the gel slices were incubated in SBC supplemented
with 10 mM MgCl
2
for 1 min at 37oC, then returned to the sodium acetate/EDTA buffer and processed as
described above. After the overnight incubation, residual gel pieces were
removed by centrifugation at 14 000
g
for 5 min and the supernatants retained. The DNA was precipitated by the
addition of 10 [mu]g yeast tRNA and 2.5 vol 100% ethanol, and incubation at -80oC for 15 min. The precipitated DNA was collected by
centrifugation at 14 000
g
for 30 min, resuspended in formamide gel loading buffer (
23
), and analyzed under denaturing conditions in 20% polyacrylamide gel containing 7 M urea (
13
).
The oligonucleotide strand to be analyzed was 5'-
32
P-labeled and annealed to unlabeled complementary DNA (
13
). In methylation interference studies, 1.25 pmol of this duplex DNA was
methylated by dimethylsulfate treatment (
23
), incubated with Ape protein, and the binding reactions resolved by non-denaturing gel electrophoresis. Bound and unbound DNA was extracted from
the respective gel slices, treated with 1 M piperidine (
23
), and analyzed on denaturing gels. In ethylation interference experiments, 1.25
pmol substrate DNA was treated with ethylnitrosourea (
24
), and the free and bound DNA analyzed for ethylation content by heat and alkali
treatment (
25
).
In situ
footprinting with 1,10-phenanthroline-copper ion was performed with 5'-
32
P-labeled DNA as described by Kuwabara and Sigman (
26
).
The initial studies determined whether Ape could bind abasic sites in DNA sufficiently stably to allow detection of DNA-protein complexes using conventional assays. Ape protein was isolated from HeLa
cells and incubated with a synthetic DNA substrate containing a tetrahydrofuran
residue (F, an AP site analog not sensitive to [beta]-elimination; ref.
21
). After incubation of Ape and DNA in SBC buffer (containing <= 0.5 mM residual MgCl
2
from the kinase reaction), EMSA revealed that Ape protein formed detectable
complexes with duplex oligonucleotides bearing F (Fig.
1
A). Ape never formed complexes detectable by EMSA with unmodified duplex DNA
(data not shown). Positioning a guanine or cytosine opposite F, which may have
different structural effects (
27
-
29
), did not dramatically affect (<2-fold) the amount of F-DNA@Ape complexes formed (data not shown). Under these conditions (SBC
buffer), 5-10% of the available F-DNA substrate was shifted upon incubation with 10
ng of two different Ape preparations from HeLa cells (Fig.
1
A, lanes 2 and 3). This determination omits the `smear' (8-27% of the total DNA substrate; see Fig.
1
A), which likely represents protein-DNA complexes that dissociated during electrophoresis. Thus, this method
probably underestimates the amount of bound DNA.
A sample of recombinant GST-Ape fusion protein (~67 kDa) isolated from
E.coli
also showed binding to F-containing DNA, but yielded a complex of significantly slower mobility
(lane 4 in Fig.
1
A), consistent with the larger size of the fusion protein. GST protein (not
fused to Ape) purified in the same manner (see Materials and Methods) did not
form detectable complexes with F-DNA (data not shown). Purified recombinant Ape
protein (not fused to GST) isolated from
E.coli
formed complexes with F-DNA that had the same mobility as those formed by Ape
from HeLa cells (Fig.
1
B, lanes 1 and 2), and Ape-specific antibodies (
8
,
22
) supershifted the complexes formed by recombinant Ape (Fig.
1
B). Thus, the observed protein@F-DNA complexes are formed by Ape and not by a contaminating protein.
Competition experiments were used to determine the selectivity of Ape binding to F-DNA. A 10-fold molar excess of unlabeled duplex competitor DNA containing F reduced
the amount of EMSA complexes with the F substrate by 8-fold, while unmodified duplex DNA competitor caused only a slight decrease (11%) in the amount of
Ape-DNA complexes detected. Single-stranded F-DNA was without effect in competition experiments.
Because Mg
2+
stimulates Ape endonuclease activity (
30
) and low amounts of Mg
2+
were present in SBC (see above), we determined the effect of this metal on DNA
binding. Addition of 10 mM MgCl
2
reduced detectable Ape binding to F-DNA, an effective substrate for cleavage (
13
), to <5% of that detected in SBC (Fig.
1
C). In contrast, the addition of 1 mM EDTA increased the formation of F-DNA@Ape complexes >3-fold (at 10 ng of Ape) over SBC conditions (Fig.
1
D, compare lanes 2 and 4). EDTA also stabilized Ape binding to DNA with a
regular AP site generated by UDG (Fig.
1
D, lanes 6 and 7), with binding efficiency similar to that for F in the same duplex (compare Fig.
1
D, lane 4). However, F was used for interference and footprinting experiments (see below), because AP sites are prone to spontaneous cleavage via [beta]-elimination (
1
,
3
,
21
).
Filter binding could also be used to detect Ape-DNA complexes in the presence of EDTA. An increasing fraction of F-containing DNA was trapped on nitrocellulose filters as the amount
of Ape was increased from 0.1 to 100 ng per assay, while control DNA without
abasic sites was not bound detectably (Fig.
2
A). Filter- binding allowed us to determine the half-life of F-DNA@Ape complexes. In one set of experiments, after binding
incubations, unlabeled F-DNA was added in a 100-fold excess over the labeled substrate with simultaneous addition of EDTA
to 1 mM or MgCl
2
to 10 mM, and the amount of complex remaining was measured by filtration at
specific times thereafter. These studies indicate half-lives of ~50 s for the F-DNA@Ape complexes in EDTA and <15 s in MgCl
2
(Fig.
2
B). The results also indicate that a minor fraction of more stable complexes may
persist in the presence of Mg
2+
(Fig.
2
B).
Pre-incision of duplex F-DNA with a catalytic amount of endonuclease IV
virtually eliminated ( >= 95%) F-DNA@Ape complex formation even in the presence of EDTA (data not shown), which
indicates that the complexes observed in EMSA are not due to product binding.
When the DNA in F-DNA@Ape complexes (accounting for >70% of the total DNA; Fig.
3
A) was isolated, >95% was present as the uncleaved 18mer in reactions with EDTA,
and >83% in reactions in SBC buffer (Fig.
3
B). For the latter, some incision may have occurred during DNA isolation. Intact
DNA was also found within the `smear' region of EMSA gels (Fig.
3
A and B), consistent with the relative instability of F-DNA@Ape complexes even in the presence of EDTA (Fig.
2
B). When gel slices containing F-DNA@Ape complexes were soaked briefly in Mg
2+
containing buffer at 37oC, the 18mer substrate was rapidly converted to the 9mer product (Fig.
3
B). When F-DNA@Ape binding was performed in buffer containing 1 mM EDTA and MgCl
2
then added (10 mM) together with a 100-fold excess of unlabeled competitor F-DNA for 1 min at 0oC, the initial fraction of substrate DNA bound (37.3%) was close to
the fraction cleaved (42.6% of the substrate converted to product). Thus, the complexed material is on the catalytic pathway for incision by Ape.
For incision of F-DNA, Ape requires >4 bp of duplex DNA on the 5' side of the lesion and >3 bp on the 3' side (
13
). To identify potential base contacts in F-containing DNA, methylation interference experiments were conducted using
a pair of 23mer substrates (Fig.
5
A). For the strand containing the abasic site, methylation of guanines located 1
or 3 bp 5' of the F residue prevented complex formation (Fig.
5
B). Unexpectedly, methylation of an adenine residue 2 bp 3' of the abasic site stimulated Ape binding (Fig.
5
B). Analysis of the strand opposite the AP lesion showed that methylations at 2
bp 5', or 1 or 3 bp 3' of the abasic site (Fig.
5
B) interfered with Ape binding.
A previous study revealed several structural features of DNA substrates that
affect Ape endonuclease activity (
13
). Meanwhile, a crystal structure was determined for exonuclease III (
19
) and used to model a proposed structure for the homologous Ape protein (
20
). However, structures for these enzymes bound to substrate DNA molecules have
not been reported. We have addressed the issue of how AP endonucleases engage
their substrate by defining contact sites of Ape protein with a stable abasic
site in DNA.
Our findings show that binding and incision by Ape protein can be separated by
relatively simple procedures. Two independent methods (EMSA and filter binding)
demonstrated a clear preference of Ape for binding double-stranded DNA substrates with an F residue or a regular AP site compared to
undamaged DNA. These complexes are evidently on the catalytic pathway, since
F-DNA bound by Ape in the presence of EDTA is rapidly incised upon addition of
MgCl
2
. Complex formation is not merely due to the sequence context, because EMSA
revealed Ape that bound a 51mer duplex (
32
) containing a central, UDG- generated AP site, or an F-site, with the same affinity as seen for the F substrate examined here (R.A.O. Bennett, D.M.W., D. Wong and B.D., submitted). Thus, the complexes observed in this work are
appropriate for studying the details of the interaction of Ape with its
substrate.
We are grateful to Drs Arthur P.Grollman and Richard A.O.Bennett for their valuable comments on the manuscript. We thank Drs Bennett, Elena
Hidalgo, Ziyi Li and Lynn Harrison, and Mr Edy Kim for discussions. We are
indebted to Drs Jim Carney and Mark Kelley for providing pGEXApe. These studies
were supported by grants from the National Institutes of Health to B.D.
(GM40000) and to A.P.Grollman (CA47995 & CA17395), and by a National Research Service Award to D.M.W.III (CA62845).
*To whom correspondence should be addressed. Tel: +1 617 432 3462; Fax: +1 617
432 2590; Email: demple@mbcrr.harvard.edu
REFERENCES
Return