ABSTRACT
Base excision repair is one of the major mechanisms by which cells correct damaged DNA. We have developed an
in vitro
assay for base excision repair which is dependent on a uracil-containing DNA template. In this report, we demonstrate the fractionation of a human cell extract into two required components. One fraction was extensively purified and by several criteria shown to be identical to
DNA polymerase
[beta]
(Pol
[beta]
). Purified, recombinant Pol
[beta]
efficiently substituted for this fraction.
Escherichia coli
PolI, mammalian Pol
[delta]
and to a lesser extent Pol
[alpha]
and
[epsilon]
also functioned in this assay. We provide evidence that multiple polymerases function in base excision repair in human cell extracts. A neutralizing antibody to Pol
[beta]
, which inhibited repair synthesis catalyzed by pure Pol
[beta]
by
~
90%, only suppressed repair in crude extracts by a maximum of
~
70%. An inhibitor of Pol
[beta]
, ddCTP, decreased base excision repair in crude extracts by
~
50%, whereas the Pol
[alpha]
/
[delta]
/
[epsilon]
inhibitor, aphidicolin, reduced the reaction by
~
20%. A combination of these chemical inhibitors almost completely abolished
repair synthesis. These data suggest that Pol
[beta]
is the major base excursion repair polymerase in human cells, but that other
polymerases also contribute to a significant extent.
Cells have multiple strategies for repairing the various types of damage that
constantly alter their DNA. These DNA repair mechanisms are crucial in preventing mutagenesis and carcinogenesis. Certain modified or improper bases are repaired by a base excision repair
mechanism in which the offending base is removed, the sugar-phosphate backbone is nicked, and a short repair patch is formed (
1
-
4
). For example, the deamination of a dCMP residue in the DNA leads to a U-G mispair which, if not corrected, can result in a C to T mutation (
4
). Additionally, DNA polymerases can incorporate dUMP into DNA during
replication, repair and recombination (
2
). Base excision repair of uracil-containing DNA is believed to be initiated by a uracil-DNA glycosylase which cleaves the bond between the uracil base and the deoxyribose sugar to yield an abasic site with the sugar-phosphate backbone intact. The phosphodiester bond 5' to the abasic site is then nicked by an AP
(apurinic/apyrimidinic) endonuclease, and the abasic sugar-phosphate residue is removed by the action of a
deoxyribophosphodiesterase (or a 5' -> 3' exonuclease). A DNA polymerase fills in the resulting
single nucleotide gap and a DNA ligase seals the repair patch. Some reports
suggest that this model for base excision repair may be overly simplistic or that alternative pathways may exist (
5
-
9
). Furthermore, the identity of the enzymes involved in base excision repair remains speculative.
For example, previous studies aimed at determining the identity of the DNA
polymerase involved have been inconclusive or conflicting. Three reports have
used polymerase inhibitors to conclude that base excision repair in mammalian
cell extracts is carried out exclusively by Pol[beta] (
3
,
10
,
11
). Singhal
et al.
(
11
) also show that reconstitution of the repair reaction with partially purified
components can be accomplished using Pol[beta] but not Pol[alpha], [delta] or [epsilon]. Additionally, Sobol
et al.
(
12
) provide genetic evidence for the involvement of Pol[beta] in base excision repair. A cell line lacking the Pol[beta] gene was found to be hypersensitive to alkylating agents, and
extracts from these cells did not support base excision repair unless Pol[beta] was provided exogenously. This body of evidence has led to the conclusion
that Pol[beta] is essential for base excision repair in mammalian cells (
3
,
10
-
12
).
In contrast, experiments in other eukaryotic systems have suggested that Pol[delta] and/or [epsilon] play an important role in base excision repair. Using fractions
derived from a
Xenopus
ovarian extract, Matsumoto
et al.
(
13
) demonstrated that repair of a synthetic tetrahydrofuran AP site was dependent
on Pol[delta] and its accessory protein, proliferating cell nuclear antigen (PCNA).
Henderson
et al.
(
14
,
15
) have shown that mutations in the
Drosophila
gene encoding the Pol[delta]/[epsilon] auxiliary factor, PCNA, cause a hypersensitivity to alkylating agents and ionizing
radiation.
Inhibitor studies using intact or permeable mammalian cells have not
conclusively determined the polymerase requirement for base excision repair.
Early experiments indicated that Pol[beta] is involved in repair of bleomycin-induced DNA damage while an aphidicolin-sensitive polymerase (probably Pol[delta] or [epsilon]) carries out repair of alkylated DNA (
16
,
17
). More recent reports suggest that both Pol[beta] and Pol[delta]/[epsilon] may play a role in DNA repair induced by both bleomycin and
alkylating agents (
18
-
21
). One limitation of these studies is their reliance on inhibitors with limited
specificity (
22
). Another problem is that bleomycin and alkylating agents produce DNA damage
that is repaired not only by base excision repair but by other mechanisms as
well.
Thus, uncertainty clearly remains over which DNA polymerase(s) and other proteins are involved in base excision repair. To address this
question, we are using a DNA molecule containing a single, defined uracil
residue as a substrate for base excision repair in human cell extracts. In this
report we initiate the fractionation and purification of the components
required for human base excision repair. We demonstrate that although Pol[beta] is responsible for the majority of uracil-dependent repair synthesis in HeLa cell extracts, other DNA polymerases contribute to a significant degree.
We suggest reasons why previous studies using mammalian cell extracts led to the conclusion that only Pol[beta] could fulfill this role.
The following materials were obtained from commercial sources: HeLa S3 cells
(National Cell Culture Center); [
3
H]dTTP (Amersham); [
32
P]dNTPs (DuPont-NEN); ATP, creatine phosphate, creatine phosphokinase, [beta]-lactoglobulin A, aphidicolin, bovine serum albumin and single-stranded (ss) DNA-cellulose (Sigma); dideoxyTTP (ddTTP), SP
Sepharose and Mono S (Pharmacia LKB); unlabelled dNTPs, phenylmethyl sulfonylfluoride (PMSF) and
E.coli
PolI-large fragment (Gibco-BRL); leupeptin and pepstatin (Boehringer Mannheim Biochemicals); DEAE-cellulose DE52 and phosphocellulose P-11 (Whatman); chromatography columns (Kontes). Pol[alpha] was immunoaffinity purified from HeLa cell
extract (
23
). PCNA was purified from
E.coli
which contained a plasmid encoding the human PCNA cDNA under the control of an
inducible bacteriophage T7 promoter (
24
). Pol[epsilon] and RF-C were purified from calf thymus and were generously provided by
Vladimir Podust and Ulrich Hübscher. Purified Pol[delta] from HeLa cells was a gift from Hernan Flores-Rozas and Jerard Hurwitz.
Lyophilized oligonucleotides (Oligos Etc. Inc.) were resuspended in TE (pH 8.0)
and concentrations were adjusted according to spectrophotometric readings at OD
260
. Equal concentrations (500 [mu]g/ml each) of complementary stands were annealed in the presence of 200 mM
NaCl by heating at 90oC for 5 min and cooling slowly to room temperature. Annealing was checked
by native polyacrylamide gel electrophoresis. The experiments shown in this paper used a 30 base pair (bp) UG oligonucleotide of the following
sequence:
5'-GAGCCGGCACTGG
U
GCCCAGCTGATATCGC-3'
3'-CTCGGCCGTGACC
G
CGGGTCGACTATAGCG-5'
We have also used a 30 bp UA oligonucleotide containing a single UA base pair. As controls, the corresponding normal homoduplexes containing CG or TA base pairs were utilized.
This assay measures DNA synthesis using radioactive nucleotides, a uracil-containing duplex oligonucleotide, and a human cell extract or fractions
derived therefrom. Reactions (25 [mu]l) contained 40 mM creatine phosphate-diTris salt (pH 7.7), 5 mM MgCl
2
, 1 mM dithiothreitol, 2 mM ATP, 20 [mu]M each of dATP, dGTP, dTTP and dCTP, 1 [mu]Ci [[alpha]-
32
P]dCTP (for UG and CG substrates) or dTTP (for UA and TA substrates), 2.5 [mu]g creatine phosphokinase, 50 mM NaCl, 0.5 [mu]g (25 pmol) 30 bp duplex oligonucleotide, and HeLa crude extract (~10 [mu]g), fractions or purified proteins as indicated. The experiments shown in this paper used the UG
oligonucleotide. Reaction mixtures were incubated at 37oC for 30 min. Reactions were stopped by addition of 5 [mu]l of 6* gel loading dye (20% Ficoll, 100 mM EDTA, 2% SDS, 0.2%
bromophenol blue, 0.2% xylene cyanol). Samples were directly loaded onto a 15%
polyacrylamide gel (29:1/acrylamide:bis) and electrophoresed at 150 V for ~90 min. The dried gel was exposed to film and the radioactive bands were
excised from the gel for quantitation in a scintillation counter.
HeLa cell S-100 extract
. An S-100 extract from 50 l of HeLa S3 cells (5.1 * 10
5
cells/ml) was prepared as previously described (
25
,
26
).
DEAE-cellulose and SP Sepharose chromatography.
DEAE-cellulose and SP Sepharose columns (120 c.c. each) were equilibrated with
Buffer AN (20 mM Tris-HCl pH 7.5, 0.1 mM EDTA, 10% glycerol, 0.01% NP-40, 1 mM dithiothreitol) containing 50 mM NaCl and connected in
series. HeLa cell extract (1024 mg of protein) was loaded onto the DEAE-cellulose column and flowed through onto the SP Sepharose column. While connected
in series, the columns were washed with 600 ml of Buffer AN-50 mM NaCl. The columns were disconnected and the SP Sepharose column was
eluted with Buffer AN-250 mM NaCl (360 ml) followed by Buffer AN-1.0 M NaCl (360 ml). Protein peaks were pooled for the flow through
(DSP1, 585 mg), 250 mM elution (DSP2, 49 mg) and 1 M elution (DSP3, 14 mg). DNA
repair activity was recovered by combining the DSP2 and DSP3 fractions.
In the experiment presented in Table
1
, the DEAE-cellulose column was not connected to the SP Sepharose column. The
extract was loaded onto the DEAE-cellulose column and the column was washed as above. Bound proteins were
eluted with Buffer AN-1.0 M NaCl. The fractions containing the peak protein concentrations were
pooled and designated fractions DE1 (flow through) and DE2 (bound).
Single stranded DNA-cellulose chromatography.
DSP3 (14 mg) was dialyzed overnight against Buffer AN-250 mM NaCl and loaded onto a ssDNA-cellulose column (9 c.c.) which had been equilibrated in Buffer AN-250 mM NaCl. The column was washed with 50 ml of Buffer AN-250 mM NaCl. A linear gradient elution of 80 ml from
0.25 to 1.0 M NaCl in Buffer AN was performed. Fractions supporting repair
synthesis when combined with the DSP2 fraction eluted at ~700 mM NaCl and were pooled.
Mono S chromatography.
The ssDNA-cellulose pool (192 [mu]g) was dialyzed against Buffer AN-250 mM NaCl, and lactoglobulin was added to 100 [mu]g/ml to the dialysate as a carrier. A Mono S FPLC column (8
c.c.) was equilibrated with Buffer AN-250 mM NaCl. The sample was loaded and the column was washed with 20 ml of
Buffer AN-250 mM NaCl containing 30 [mu]g/ml of lactoglobulin. A 20 ml linear gradient to 1.0 M NaCl in Buffer
AN containing 30 [mu]g/ml lactoglobulin was used for elution. Repair complementation activity and DNA polymerase activity co-eluted at ~750 mM NaCl and peak fractions were pooled.
Standard DNA polymerase reactions (50 [mu]l) contained 50 mM Tris-HCl (pH 8.0), 8 mM MgCl
2
, 4 mM dithiothreitol, 10 [mu]g bovine serum albumin, 40 [mu]M each of dATP, dGTP and dCTP, 10 [mu]M [
3
H]dTTP (~500 c.p.m./pmol), and 10 [mu]g activated DNA (DNase I-treated calf thymus DNA). Where indicated, polymerase activity was measured under repair assay conditions. These
conditions were identical to those of base excision repair reactions except for the
addition of 10 [mu]g of bovine serum albumin, the use of 10 [mu]g of activated DNA as the DNA template, and the use of 10 [mu]M [
3
H]dTTP as the radioactive nucleotide. All DNA polymerase reactions were
incubated at 37oC for 60 min and acid-insoluble radioactivity was determined.
This method was performed essentially as described by Karawya
et al.
(
27
). Briefly, protein samples were heated to only 37oC prior to electrophoresis on a 10% SDS-polyacrylamide gel containing 100 [mu]g/ml activated DNA. Following electrophoresis, the gel was
incubated in several changes of wash buffer to ensure removal of the SDS and
renaturation of proteins. The gel was then incubated in buffer containing
radioactive dNTPs overnight at room temperature. This allows incorporation of labelled nucleotides at the site of any DNA polymerase within the gel. Unincorporated nucleotides
were then removed by several washes in 5% trichloroacetic acid, 1% sodium
pyrophosphate. The gel was dried and autoradiographed.
Immunoblot (Western blot) assays were performed by standard methods (
28
). Protein samples were subjected to SDS-PAGE and then transferred to nitrocellulose. Pol[beta] was detected using rabbit polyclonal serum (1:500 dilution), horseradish peroxidase-conjugated donkey anti-rabbit immunoglobulin and enhanced chemiluminescence detection (ECL, Amersham). For
quantitative immunoblotting, several amounts of each protein fraction were analyzed and the
relative band strengths were compared.
Escherichia coli
strain BL21 (DE3) pLysE (Novagen, Inc.), transformed with the rat Pol[beta] expression plasmid pRSET (generously supplied by Dr Samuel Wilson), was
cultured in LB broth (800 ml) at 37oC until the OD
600
reached 0.6. IPTG was added to a concentration of 1 mM, and incubation was
continued at 37oC for 4 h. The cells were pelleted and frozen at -20oC. The bacterial pellet was thawed and resuspended in 80 ml of
50 mM Tris-HCl (pH 7.5), 1 mM EDTA, 0.1 mM PMSF, 0.5 [mu]g/ml leupeptin and 0.5 [mu]g/ml pepstatin A. The suspension was incubated for 15 min at room
temperature allowing endogenous lysozyme to act. The suspension was vortexed
and disrupted by sonication (while cooling on ice) until the viscosity was
minimal. All subsequent procedures were carried out at 4oC. The mixture was then centrifuged at 12 000
g
for 20 min, and the supernatant decanted. To the supernatant, 50 mM Tris-HCl pH 7.5, 1 mM EDTA and 4 M NaCl was added to give a final NaCl concentration of 200 mM. The supernatant was dialyzed overnight against Buffer A (20 mM Tris-HCl pH 7.5, 0.1 mM EDTA, 10% glycerol, 1 mM dithiothreitol)
containing 200 mM NaCl. The dialysate was clarified by centrifugation at 12 000
g
for 20 min. Pol[beta] was purified by sequential chromatography on DEAE-cellulose, SP Sepharose, ssDNA-cellulose and phosphocellulose, essentially as described above for purification from HeLa extract.
Approximately 6 mg of purified Pol[beta] were obtained starting from an 800 ml culture.
Rabbits were immunized with purified Pol[beta] (Cocalico Biologicals). Polyclonal antibodies against Pol[beta] were purified from rabbit antisera using a Pol[beta] affinity column (
28
). The affinity resin was prepared by coupling purified recombinant rat Pol[beta] (1 mg in PBS) to cyanogen bromide-activated Sepharose 4B (Sigma). Pol[beta]-specific antibodies were eluted from this column with
100 mM glycine-HCl pH 2.5 or 100 mM triethylamine pH 11.5. Purified antibodies were
dialyzed against PBS, concentrated using a Centriprep 10 apparatus (Amicon),
and stored in aliquots at -80oC. Affinity-purified anti-PCNA antibodies were produced in a similar manner using
purified recombinant human PCNA as the immunogen and in preparation of the
affinity resin.
We established a base excision repair assay based on those reported by Dianov
et al.
(
3
) and Wang
et al.
(
29
). Our
in vitro
assay measures uracil-dependent repair synthesis using a human HeLa cell extract and a duplex
oligonucleotide containing a single uracil residue. A 30 bp UG oligonucleotide
was identical to a CG oligonucleotide except for a single dUMP in place of a
dCMP at position 14 (see Materials and Methods). Likewise, a 30 bp UA
oligonucleotide differed from a TA oligonucleotide by virtue of a single uracil
residue in place of a thymine. A UG base pair represents the result of cytosine
deamination in normal duplex DNA, whereas a UA base pair would result from the incorporation of a dUMP during the course of DNA replication. The oligonucleotides were
incubated in HeLa cell extract with radiolabelled nucleotides and the reaction
products were electrophoresed on a polyacrylamide gel. We typically observed >= 10-fold more synthesis on the uracil-containing oligonucleotides compared with the normal duplex oligonucleotides. A duplex
oligonucleotide containing a GT mismatch (which would result from deamination
of a 5-methyl cytosine residue) did not support significant levels of repair
synthesis. Likewise, single-stranded oligonucleotides were not efficient substrates in this reaction.
The vast majority of repair events consisted of a single nucleotide repair
patch (
3
,
11
; K.N. and M.K.K., unpublished observations).
Because we intended to use this assay for fractionation and purification, it was
essential for it to be rapid, sensitive and quantitative. Importantly, we found
that purification of the reaction products was neither necessary nor desirable.
Addition of gel loading dye (containing SDS and EDTA) to stop the reaction
followed by direct loading of the mixture onto the gel led to more rapid and
quantitatively more reproducible results. Quantitation of repair synthesis was
achieved by excising radioactive bands from the dried gel and measuring
radioactivity in a scintillation counter. Using ~10 [mu]g of extract, the reaction remained linear for at least 30 min at which
point ~20% of the template molecules are typically repaired. This compares favorably with other complex mammalian
in vitro
DNA synthesis assays such as SV40 DNA replication and nucleotide excision
repair.
A potential role for poly ADP-ribose polymerase was investigated because of previous evidence suggesting the involvement of this enzyme in
DNA repair processes (
2
). The addition of the poly ADP-ribose polymerase substrate (NAD) or inhibitor (3-aminobenzamide) had no effect on repair synthesis. Surprisingly, a 1.5-3-fold stimulation of repair synthesis by ATP was
consistently observed, and thus ATP and an ATP-regenerating system were included in all assays. Neutralizing antibodies
to human SSB (RP-A) and Pol[alpha] had no significant effect on the repair reaction.
To investigate the protein requirements for base excision repair of a uracil-containing template further, we have begun fractionating the crude
extract. Preliminary experimentation with various fractionation steps indicated
that base excision repair activity flowed through a DEAE-cellulose column when loaded at 50 mM NaCl. This procedure removed ~50% of the protein and the majority of the nucleic acids in the crude extract which contributed to some non-specific DNA synthesis. It was also noted that the vast
majority of DNA polymerase activity bound to the DEAE-cellulose.
The material that flowed through the DEAE-cellulose column was directly loaded onto an SP Sepharose column. The
column was washed with buffer containing 50 mM NaCl and protein was eluted with
two steps of 250 mM and 1 M NaCl. The protein peaks of the flow through and the
step elutions were collected and these fractions were designated DSP 1, 2 and
3, respectively (Fig.
1
A). These fractions were assayed individually and in combination for repair synthesis on the UG oligonucleotide. As shown in Figure
1
B the fractions were individually incapable of supporting repair synthesis. However, the combination of the fractions DSP 2 and 3 led to high levels of repair. DSP 1 had no effect on the synthesis by the other
two fractions. Repair synthesis by DSP 2 and 3 exhibited even greater
dependence on a uracil-containing DNA template than the crude extract presumably due to removal
of non-specific nucleases. The reconstituted system was able to utilize both the
UG and UA oligonucleotides, and the size of the repair patch synthesized was a
single nucleotide (data not shown).
Because DSP3 contained <= 2% of the protein of the crude extract, and because initial fractionation
attempts suggested that it may contain a single required component, attention
was focused on purifying that component from DSP3. Since virtually all of the DNA polymerase activity that flowed through the DEAE-cellulose column was found in fraction DSP3, fractions were assayed for polymerase
activity as well as for the ability to perform repair synthesis in combination
with DSP2. DSP3 was loaded onto a ssDNA-cellulose column and eluted with buffer containing a linear gradient from
0.25 to 1.0 M NaCl. The majority of the repair synthesis and DNA polymerase
activity bound to the resin and co-eluted in a peak at ~0.7 M NaCl, resulting in a 26-fold purification (data not shown).
This peak was pooled and loaded onto a Mono S FPLC column. Both repair and
polymerase activities bound to this column and were eluted with a linear salt
gradient from 0.25 to 1.0 M NaCl. Figure
2
A shows the co-elution of repair and polymerase activities around the peak fractions at ~0.75 M NaCl. Initial characterization of this co-eluting polymerase suggested that it was Pol[beta]. To test this possibility further, a DNA polymerase
activity gel assay was performed. This assay measures DNA polymerase activity
in situ
in an SDS-polyacrylamide gel and thus provides information on the molecular weight
of the polymerase. Figure
2
B shows an activity gel analysis of the Mono S fractions containing the peak of
repair activity. A DNA polymerase activity which migrated identically to
purified Pol[beta] at 39 kDa, co-eluted with repair activity. Figure
2
C shows the result of an immunoblot analysis of these same peak fractions using
antibodies against Pol[beta]. A 39 kDa band, immunologically related to Pol[beta], also co-eluted with repair activity. Although it is difficult to
determine the precise degree of purification achieved by Mono S chromatography
due to the presence of carrier protein in the elution buffer, we estimate that
the overall purification of this factor from crude extract is >1000-fold.
Although Pol[beta] was purified as a factor required for base excision repair, it is
possible that other polymerases might also function in this type of DNA repair.
Various DNA polymerases were tested for their ability to complement DSP2 in the
DNA repair assay (Fig.
3
). Purified recombinant Pol[beta] was able to fully substitute for DSP3 in its ability to complement DSP2
in the repair reaction. This further confirms that there is only one essential
factor in DSP3 and that this factor is Pol[beta].
Escherichia coli
PolI also supported efficient repair synthesis in combination with DSP2. Of the
other mammalian DNA polymerases tested, Pol[delta] yielded the highest level of repair synthesis. Although the
incorporation seen with Pol[alpha] and [epsilon] was low, it was still above background. The addition of the
polymerase accessory proteins, PCNA and RF-C, did not significantly influence the level of repair synthesis by Pol[delta] or [epsilon] (data not shown). The synthesis by all of the polymerases
was dependent on DSP2 and a uracil-containing oligonucleotide.
In this report, we have initiated the identification of enzymes involved in the
repair of uracil-containing DNA by fractionating and purifying factors required to
reconstitute repair synthesis. Ion exchange chromatography was used to separate
the HeLa cell crude extract into two required fractions. One of these fractions
was extensively purified and appears to be identical to Pol[beta] based on several criteria: (i) co-purification of repair and polymerase activities; (ii) the co-purifying polymerase is the same size as Pol[beta]; (iii) is immunologically-related to Pol[beta]; and (iv) has the same inhibitor
sensitivity as Pol[beta]; (v) purified, recombinant Pol[beta] can substitute for this fraction. This represents the first time
Pol[beta] has been purified as a factor required for DNA repair. Additionally, inhibition experiments indicate that Pol[beta] is required for the majority of the repair synthesis in crude
extracts. This important role for Pol[beta] in base excision repair is consistent with what is known of this enzyme. For example, Pol[beta] has been shown previously to efficiently repair DNA templates with short gaps
in vitro
(
17
,
30
,
31
). Also, Pol[beta] is present in most cells regardless of their proliferative state, and Pol[beta] mRNA levels are increased in response to alkylating agents (
32
,
33
).
Although Pol[beta] appears to be the major base excision repair polymerase in HeLa cell
extracts, our results indicate that other polymerases can and do function in
this process. Other DNA polymerases, including
E.coli
PolI and human Pol[delta], substituted for Pol[beta] in reconstituting repair in a partially purified system. Secondly,
the initial fractionation of the crude extract yielded one fraction enriched
for Pol[beta] and another fraction enriched for Pol[alpha]/[delta]/[epsilon] (containing only 25% of the total Pol[beta]). Each fraction contributed equally to
satisfying the polymerase requirement in the reconstituted repair reaction.
Thirdly, a neutralizing antibody against Pol[beta] was unable to completely abolish repair synthesis in crude extracts.
Lastly, repair in extracts exhibited a strong but incomplete sensitivity to
dideoxynucleotides and a partial sensitivity to aphidicolin. A combination of
these inhibitors led to a greater effect than either alone.
Studies with other organisms indicate that Pol[beta], [delta] and [epsilon] may each participate in base excision repair. In
combination with fractions from a
Xenopus
ovarian extract, both Pol[beta] and Pol[delta] are able to support repair of DNA containing natural AP sites (
13
). In
Saccharomyces cerevisiae
, base excision repair is dependent on Pol[delta] and/or Pol[epsilon], and Pol[beta] does not appear to be required (
11
,
29
,
34
). This polymerase dependency may reflect the fact that yeast Pol[beta] differs significantly from the mammalian enzyme and mitotically-growing yeast cells contain little if any Pol[beta] compared to meiotic cells (
35
). Genetic evidence suggesting a role for Pol[delta]/[epsilon] in base excision repair comes from the study of the
Drosophila
mutant mus209, which is extremely sensitive to alkylating agents and ionizing
radiation. The mus209 gene encodes the Pol[delta]/[epsilon] accessory factor, PCNA (
14
,
15
).
However, other studies in mammalian cells have concluded that Pol[beta] alone is responsible for catalyzing base excision repair. Wiebauer and Jiricny (
10
) reported that base excision repair of DNA containing a G-T mismatch by human cell extracts was inhibited by a polyclonal antibody
against Pol[beta]. However, reversal of this inhibition by exogenously-added Pol[beta] was not demonstrated and the specificity of this inhibition
has been questioned (
11
). Dianov
et al.
(
3
) used DNA polymerase inhibitors to address the question of which polymerase was
involved in base excision repair of uracil-containing DNA in human cell extracts. They found that 100 [mu]M ddTTP, an inhibitor of Pol[beta], reduced repair synthesis by 67%. On the other hand, 100 [mu]g/ml aphidicolin, an inhibitor of Pol [alpha], [delta] and [epsilon], diminished synthesis by ~40%. These data were used to
suggest that Pol[beta] carried out repair synthesis in this system. Alternatively, it could be
argued that a combination of dideoxynucleotide-sensitive and aphidicolin-sensitive polymerases were responsible for the repair synthesis.
Wilson and co-workers (
11
,
12
) have presented evidence that Pol[beta] is essential for base excision repair of uracil-containing DNA in mammalian cell extracts and that this polymerase
requirement cannot be satisfied by Pol[alpha], [delta] or [epsilon]. Differences in the method of extract preparation may
explain the disparity with the results presented here. The 0-40% ammonium sulfate precipitation step used for nuclear extract preparation by the Wilson laboratory (
11
,
12
) may have resulted in the removal of Pol[alpha], [delta], [epsilon] and accessory factors from the extract. Reconstitution of
repair by the addition of pure Pol[alpha], [delta] or [epsilon] may have failed because of the loss of accessory proteins such as
deoxyribophosphodiesterase or exonuclease during extract preparation. Pol[beta], on the other hand, contains an intrinsic deoxyribophosphodiesterase activity and thus might not require this factor to be present in the extract (
36
).
Sobol
et al.
(
12
) also present genetic evidence that Pol[beta] is involved in base excision repair
in vivo
. They found that a mouse cell line lacking the Pol[beta] gene exhibited normal viability and growth characteristics but was
moderately hypersensitive to alkylating agents. They concluded that Pol[beta] was essential for base excision repair since most alkylation damage is
believed to be repaired by this process. However, it has been estimated that
mammalian cells acquire at least 10 000 mutagenic and cytotoxic `base lesions'
per day per genome. It seems unlikely that cells lacking Pol[beta] would grow normally if their base excision repair was completely shut
down. Additionally, it would be surprising if Pol[delta] can participate in base excision repair in
Xenopus
,
Drosophila
and yeast but not at all in mammals.
It appears, therefore, that multiple polymerases can catalyze base excision
repair in mammals as well as lower eukaryotes. The abundance and availability of the various polymerases in a given cell may in part determine their relative contribution to base excision repair in
that particular cell. In the HeLa cell extracts used here, Pol[beta] appears to catalyze ~75% of the base excision repair synthesis while other polymerases (in particular Pol[delta]) are responsible for the remaining 25%.
Additionally, the nature of the base damage may also influence which DNA
polymerase is used. In yeast, for example, Pol[delta] has been reported to repair methylated DNA while Pol[epsilon] was shown to repair thymine glycol-containing DNA (
29
,
34
). In the
Xenopus
system, both Pol[delta] and Pol[beta] could participate in the repair of natural AP sites, but only Pol[delta] efficiently repaired synthetic AP sites (
13
). It is possible that the proteins and mechanisms involved in repairing these various types of DNA damage may differ significantly.
We believe that there may be two base excision repair pathways which differ in
the polymerases used, the mechanisms of removal of the deoxyribose phosphate
residue, and the ligases used. Recently, Frosina
et al.
(
9
) provided evidence for two pathways in the repair of AP sites in mammalian
extracts. One pathway resulted in a repair patch of ~7 nt and was sensitive to inhibition by a PCNA antibody. The other pathway
was PCNA-independent and produced a single nucleotide repair patch. Our findings
are consistent with this model and provide the first direct evidence that both
Pol[delta] and Pol[beta] can function in base excision repair in mammals. The fact that Pol[delta] can load onto short linear templates in the absence of PCNA
and RF-C explains the lack of a requirement for these proteins in our system. In
contrast, Frosina
et al.
(
9
) used large circular duplex DNA containing a single AP site. The finding that Pol[beta] exhibits intrinsic deoxyribophosphodiesterase activity (
36
) also suggests that Pol[delta] will require different proteins to carry out the complete repair
reaction.
Whether Pol[beta] is the only base excision repair polymerase or simply the major one is an
important distinction. There is currently interest in inhibiting DNA repair pathways in conjunction with standard chemotherapy protocols (
37
,
38
). It is crucial to know whether there is a single pathway or multiple sub-pathways when choosing targets to inhibit. This report indicates that
there may be multiple pathways to achieve base excision repair and represents
an important first step in reconstituting these pathways with purified
proteins.
We thank Samuel Wilson for the generous gift of Pol[beta] expression constructs. We greatly appreciate the gifts of Pol[delta] from Hernan Flores-Rozas and Jerard Hurwitz, and Pol[epsilon] and RF-C from Vladimir Podust and Ulrich Hübscher. We thank Birgit Woelker for
critical reading of this manuscript. This work was supported by funds from The
Picower Institute for Medical Research.
REFERENCES
Return