ABSTRACT
More than 20 polypeptides are required for the process of nucleotide excision
repair (NER) in both human and yeast cells. This pathway of excision repair has
most often been viewed as an ordered multi-step process involving steps of damage recognition, incision/excision and finally repair DNA synthesis. Here we present evidence for
the existence of a complex of human NER proteins pre-assembled in the absence of damaged DNA. This multi-protein complex was initially isolated from HeLa cell extracts by
affinity chromatography on a matrix containing the damage recognition protein
XPA. Subsequent co-immunoprecipitation and gel filtration experiments demonstrated that a significant portion of the human NER proteins was present in the form of a high
molecular weight complex and that these complexes, or repairosomes, were
capable of performing all steps of NER
in vitro
. Consistent with studies indicating that DNA polymerases
[delta]
and
[epsilon]
can both function in NER, these two polymerases are found in these repairosome
complexes.
Previous reports have documented direct protein-protein interactions amongst many of the proteins involved in nucleotide excision repair
(NER) (
1
-
3
). Since these direct interactions could lead to the sequential recruitment of
the different NER proteins to sites of DNA lesions, the process of NER has most
often been viewed as a stepwise series of reactions (
3
-
6
). Accordingly, DNA lesions would first be bound by two proteins, XPA and RPA,
known to interact directly (
7
-
9
). Then, perhaps by virtue of an interaction between XPA and TFIIH (
10
), the DNA helicases XPB and XPD, as part of TFIIH, may be recruited to unwind DNA adjacent to the lesion. Interactions between the endonuclease ERCC1-XPF and XPA (
11
-
14
) and between the other endonuclease XPG and RPA (
7
) and/or TFIIH (
15
) may then lead to recruitment of these two enzymatic activities to the growing
NER complex and help position them to make the 5' and 3' incisions on the damaged strand flanking the lesion (
16
). After displacement of the excised oligomer, the gapped DNA, possibly still
complexed with one or more of these incision proteins (
17
), would become a substrate for the activity of PCNA-dependent DNA polymerase [delta] or [epsilon] (
18
,
19
). Since each of the documented protein-protein interactions has been detected
in vitro
in the absence of a damaged DNA substrate, it is conceivable that some or all
of these NER proteins could exist in pre-assembled sub-complexes or even a single complex or repairosome. Indeed, such a
repairosome complex containing TFIIH and a number of other NER proteins was
partially purified from
Saccharomyces cerevisiae
cell extracts, although an ability of this yeast complex to perform NER was not
reported (
20
). In this report we show that a complex of repair proteins can be isolated from
human cell extracts by steps of XPA affinity chromatography followed by gel
filtration and that these human complexes are capable of performing NER
in vitro
.
Frozen HeLa S3 cells were purchased from Cellex Biosciences (Minneapolis, MN). The XP-C cell lymphoblast line (GM02246C) was purchased from Coriell Cell Repositories and was grown in RPMI 1640
medium with 15% serum by Cellex Biosciences. The whole cell extracts were
prepared according to Sopta
et al.
(
21
). Histidine-tagged XPA protein was expressed in and purified from
Escherichia coli
as described by Jones and Wood (
22
). Affinity chromatography was carried out virtually as described previously (
7
,
23
). XPA protein was coupled to Affigel 10 (BioRad) at 2 mg/ml. Aliquots of 1 ml HeLa cell extract in ACB buffer (
7
) containing 0.1 M NaCl were loaded onto 50 [mu]l affinity columns. After washing with 200 [mu]l ACB buffer containing 0.1 M NaCl, the columns were eluted with 200 [mu]l ACB buffer containing 1.0 M NaCl. For the gel filtration
chromatography,
in vitro
NER assays and immunoprecipitation experiments the bound fractions eluted from
control or XPA columns were dialyzed against NER buffer (45 mM HEPES-KOH, pH 7.8, 70 mM KCl, 7.4 mM MgCl
2
, 0.9 mM DTT, 0.4 mM EDTA, 10% glycerol) and in some cases further concentrated 3-fold by centrifugation through Centricon-10 concentrators (Amicon).
The antibodies against RPA and XPA have been described previously (
7
). For making anti-XPG antibodies, a 519 bp fragment (encoding amino acids 455-627) was released by
Pst
I digestion from pRAD2synthetic (a gift from Dr S.Clarkson) and subcloned into
the vector QE-11 (Qiagen). Following the manufacturer's instructions, histidine-tagged XPG polypeptide was expressed in and purified from
E.coli
and was used to generate rabbit anti-XPG serum. Additional antibodies used in these studies were obtained as follows: rabbit anti-p62, anti-RAP30 (
24
) and anti-TFIIB, J.Greenblatt; anti-XPB and XPD (
25
), R.Drapkin and D.Reinberg; anti-ERCC1 (
16
), A.Sancar; anti-PCNA mAb (PC10), Santa Cruz Biotechnology Inc.; anti-RFC(140) mAb (
26
), B.Stillman; anti-ligase I mAb 7A12, G.Daly and T.Lindahl; rabbit anti-pol [delta] (C20) (
27
), H.Ochs and U.Hübscher; anti-pol [epsilon] mAb 93G1A (
28
), J.Syvaoja; rabbit anti-hMSH2(N20), Santa Cruz Biotechnology Inc. For Western blotting, aliquots of column
fractions or whole cell extract were subjected to SDS-PAGE. After electrophoretic transfer onto nitrocellulose, the filters were probed with different antibodies. The immunoblots were then developed using an enhanced chemiluminescence procedure
(Pierce).
The
in vitro
NER assay was performed essentially as previously described (
29
). DNA from plasmids pUC18 (2.7 kb) and pGEM-3Zf(+) (3.2 kb) was isolated by alkaline lysis and CsCl-ethidium bromide centrifugation. The pUC18 DNA was treated with
N
-acetoxy-2-acetylaminofluorene (AAAF) and purified on a 5-20% sucrose gradient as described by Wang
et al.
(
30
). Reaction mixtures (50 [mu]l) contained 300 ng AAAF-treated pUC18 DNA and 300 ng control pGEM-3Zf(+) DNA, 45 mM HEPES-KOH, pH 7.8, 70 mM KCl, 7.4 mM MgCl
2
, 0.9 mM DTT, 0.4 mM EDTA, 20 [mu]M each dGTP, dATP, TTP, 8 [mu]M dCTP, 2 [mu]Ci [[alpha]-
32
P]dCTP (3000 Ci/mmol), 2 mM ATP, 40 mM disodium phosphocreatine, 2.5 [mu]g creatine kinase, 3.4% glycerol, 18 [mu]g bovine serum albumin and the indicated amount of column fraction.
Reactions were incubated at 28oC for 2 h. Plasmid DNA was then purified from the reaction mixtures,
linearized by digestion with
Hin
dIII, electrophoresed on a 1% agarose gel and autoradiographed as described by
Wood
et al.
(
29
). For the antibody inhibition experiments shown in Figure
2
B, the indicated amount of pre-immune or anti-RPA antiserum was preincubated with the cell extracts for 15 min at
28oC before addition of reaction buffer and plasmid DNAs.
Pre-immune or anti-XPA serum (500 [mu]l) was incubated with 500 [mu]l of a 1:1 mixture of Sepharose 4B-protein A beads (Sigma) in phosphate-buffered saline (PBS) at room temperature for 2
h. The beads were sedimented, washed three times with 10 ml PBS and four times
with 0.2 M sodium borate, pH 9.0, and re-suspended in 10 ml 0.2 M sodium borate, pH 9.0. Antibodies were then
covalently coupled to protein A-Sepharose by addition of solid dimethylpimelimidate at a final
concentration of 20 mM and a further incubation at room temperature for 2 h.
The reactions were then terminated by incubation with 0.5 M ethanolamine, pH
7.6. The beads (30 [mu]l) were incubated at 4oC for 4 h with the dialyzed column fractions (30 [mu]l), washed three times with 100 [mu]l NER buffer and then bound proteins were eluted by boiling for
3 min with 50 [mu]l SDS sample buffer.
A Sepharose CL-2B (Pharmacia) column (0.7 * 15 cm, total volume 6 ml) was packed and equilibrated with NER
buffer according to the manufacturer's instructions. Samples of 250 [mu]l dialyzed and concentrated bound fraction eluted from XPA columns were
applied to the column and 300 [mu]l fractions were collected and analyzed by Western blotting or by an
in vitro
NER assay after being concentrated 5-fold on Centricon-10 concentrators (Amicon). Aliquots of
E.coli
70S ribosomes (a gift of M.Kiel) or high molecular weight markers (BioRad) were
also chromatographed on the same column under similar conditions.
In this study we have used protein affinity chromatography as a means to isolate
a complex of human NER proteins. Since the damage recognition protein XPA has
been reported to bind directly to several other NER proteins, namely RPA (
7
-
9
), ERCC1 (
11
-
14
) and TFIIH (
10
), as well as to itself (
14
), we chose to make affinity columns using purified recombinant histidine-tagged XPA covalently coupled to an Affigel-10 matrix. Aliquots of HeLa cell extracts were chromatographed on
the XPA affinity columns and the high salt eluates from both control and XPA
columns were initially assayed for the presence of repair proteins by Western
blotting (Fig.
1
). Consistent with previous reports, XPA, RPA, ERCC1 and the TFIIH components
XPB, XPD and p62 were each detected in the bound fraction eluted from the XPA
column, but not in control column eluates. The endonuclease XPG, which does not
bind directly to XPA (
7
), was also retained on the XPA column. The NER proteins binding to the XPA
matrix, however, were not limited to just those involved in the early steps of NER, damage recognition and incision/excision. We also found that proteins involved in the DNA synthesis step of repair,
namely the p140 subunit of RFC, PCNA and DNA ligase I, were in the XPA column
eluate. By comparing the intensity of individual polypeptides in the input HeLa
cell extracts, we estimate that ~20-40% of the input RPA, ERCC1 and TFIIH subunits, each of which has
been reported capable of binding XPA directly, was retained by the XPA column,
whereas 10-20% of the input XPG, RFC or PCNA bound to the XPA column. The recovery
of XPA appeared artificially high, since some of the column-bound XPA dissociated from the column in the high salt elution step. We
have not determined the precise stoichiometry of the components in this XPA
eluate. Since a number of these NER proteins have been previously shown capable
of direct interaction with XPA, both free and repairosome complexed forms of
these polypeptides were recovered by this affinity column procedure.
The observation that many proteins required for NER were present in the bound
fraction eluted from the XPA column prompted us to ask if this XPA eluate
fraction could perform NER
in vitro
. As shown in Figure
2
A, the XPA column eluates (lanes 3 and 4), but not the eluates from control columns (lanes 1 and 2),
preferentially incorporated radiolabeled deoxynucleotides into AAAF-damaged DNA during repair synthesis. Several lines of evidence suggest
that the radiolabel incorporation in this assay reflects bona fide NER
activity. First, when a cell extract made from a NER-deficient XP-C cell line was chromatographed on similar XPA affinity columns,
although many of these same NER proteins were detected by Western blotting
(data not shown) the XPA-bound fraction from this extract failed to perform repair synthesis (Fig.
2
A, lanes 7 and 8). Secondly, the preferential incorporation of radiolabeled
deoxynucleotides into AAAF-treated DNA was inhibited by anti-RPA antibodies (Fig.
2
B, compare lane 3 with 1) and this inhibition was reversed by addition of
recombinant human RPA (lane 4). Inhibition of nucleotide incorporation into
AAAF-treated DNA was also seen with anti-XPG antibodies (data not shown). Quantitation by phosphorimaging of
assays employing 15 [mu]l of the 3-fold concentrated eluate from the XPA column indicated that typically
40-50 fmol dCMP were incorporated preferentially into the AAAF-treated DNA. As observed by Aboussekha
et al.
(
4
), biochemical fractionation of HeLa cell extracts leads to a reduction in the efficiency of this NER reaction.
To test whether these repair proteins present in the XPA column eluates are
complexed with each other, we used anti-XPA antibodies to immunoprecipitate XPA and any associated proteins within the bound fraction eluted from the XPA column. The majority of XPA
protein was precipitated and other repair proteins representive of activities
required at early as well as later steps of excision repair, namely RPA, XPG,
the TFIIH subunit p62, PCNA and the p140 subunit of RFC, were also co-immunoprecipitated by anti-XPA antiserum as well (Fig.
3
). Use of pre-immune serum in the same precipitation experiments failed to bring down these proteins (Fig.
3
). The co-immunoprecipitation of these proteins is unlikely to be mediated by contaminating DNA, since
the inclusion in the reaction of ethidium bromide, which has been shown
previously to efficiently inhibit DNA- dependent protein interactions (
34
), did not affect precipitation of repair proteins by anti-XPA antibodies (data not shown). Similar co-immunoprecipitation of NER proteins was obtained when the bound
fraction eluted from the XPA column was precipitated with either anti-RPA or anti-XPG antibodies (data not shown).
To provide additional evidence for the existence of a high molecular weight repairosome complex in the XPA column eluate, the eluate from an XPA column was subjected to gel filtration on Sepharose CL-2B. Fractions across the elution profile were assayed for the presence of
a number of repair proteins by immunoblotting and for their ability to perform
NER
in vitro
. The XPA, RPA, TFIIH (p62) and PCNA polypeptides each co-migrated on this sizing step, as evidenced by the Western blots shown in Figure
4
A. A peak of NER activity was also detected in the same fractions in which these
NER proteins appeared (Fig.
4
B). These data provide further evidence that the excision repair proteins which
co-eluted from the XPA column exist in the form of a high molecular weight
complex, a human repairosome. We did not detect peaks of any free NER proteins
in other fractions. Since the elution position of these NER proteins and of the
in vitro
repair activity is similar to that for purified 70S ribosomal particles on this Sepharose CL-2B column (data not shown), which fractionates proteins up to 40 MDa in size, we suggest that this human repairosome
complex could be as large as 2.7 MDa. The predicted size of a complex
containing all the human NER proteins needed for reconstituted repair reactions
(
4
) is 2.1 MDa.
Although mammalian DNA polymerase [epsilon] has been suggested to be most suitable for repair synthesis during NER (
35
), studies in
Saccharomyces cerevisiae
with mutations in each of four different DNA polymerases have indicated that either DNA polymerase [delta] or [epsilon] can function to repair the DNA damage induced by UV irradiation (
36
). We have examined which of these DNA polymerases are within the human
complexes isolated by XPA affinity chromatography. As indicated in Figure
5
, both DNA polymerase [delta] and DNA polymerase [epsilon] were detected in the bound fraction eluted from an XPA column,
but not from a control column. Moreover, both DNA polymerases could be co-immunoprecipitated by anti-XPA, anti-RPA and anti-XPG antibodies (data not shown). Although we have not
determined which DNA polymerase functions within this isolated repairosome
complex, our results suggest that human cells, like yeast cells, may be able to
use either DNA polymerase [delta] or [epsilon] for the gap filling step of DNA repair synthesis.
The results in this study suggest that a significant portion of each of the
proteins required for NER in human cells can be isolated in assembled
repairosome complexes in the absence of DNA damage. Unlike the complex of yeast
NER proteins isolated after chromatography steps on Bio-Rex 70, phosphocellulose and Ni
2+
-NTA-agarose columns (
20
), these human NER complexes, isolated by XPA affinity and gel filtration
chromatography, contain all of the components required to perform the reactions
of NER
in vitro
. Proteins involved in the incision/excision reactions of NER as well as those
involved in the later steps of repair synthesis are present in these human
complexes. Details of the specific protein-protein contacts which interconnect the incision/excision and DNA
synthesis groups of proteins are not yet clear. RPA has been reported to
interact with DNA polymerase [delta] (
37
) and may also interact with DNA polymerase [epsilon] (
38
). Since RPA can make direct contacts with both XPA and XPG (
7
), this multi-functional protein could be serving as a central link within an NER
complex.
The repairosome complexes we have characterized appear to exist in the absence
of DNA lesions. Thus, during DNA repair an entire NER complex may be recruited
in a single step to sites of DNA lesions. Since a significant portion of some
of these NER proteins was not retained on our XPA columns, certain of the NER
proteins almost certainly also exist free within the cell and perhaps in
partial sub-complexes. It is difficult, therefore, to determine the contribution of
each of these different pools of NER proteins to the NER process
in vivo
. Reports on the amount of free versus complexed forms of NER proteins purified
in vitro
may also reflect differences in individual procedures. Using cell extracts from
a yeast strain carrying a tagged
TFB1
gene, Guzder
et al.
(
39
) recently failed to demonstrate a significant portion of the Rad1, Rad10 and
Rad14 proteins associated with TFIIH and on that basis argued against the
existence of assembled repairosome complexes in yeast. However, the
chromatographic procedures used to partially purify the NER proteins used in
this particular study differed significantly from those used previously by
Svejstrup
et al.
(
20
) and may have led to the loss of important proteins such as Rad4 and RPA, which
could be indispensable for the stability of any repairosome complex. Others
have reported that certain of these NER proteins, as well as proteins
implicated in double-stranded break repair, can be isolated together with a high molecular
weight form of RNA polymerase II (
40
). The complexes described in our study do not contain significant quantities of
RNA polymerase II (data not shown) and its general initiation factors (Fig.
1
). Furthermore, the presence of NER proteins, apart from subunits of TFIIH, has
not been documented in other preparations of the RNA polymerase II holoenzyme (
31
,
32
,
41
). Therefore, the relationship between the RNA polymerase II complexes reported
by Maldonado
et al.
(
40
) and those reported here is not obvious.
Recent studies of the incision/excision steps of DNA repair using both purified
human and yeast NER proteins have indicated that the steps of this portion of
the NER process are tightly coupled (
5
,
17
,
42
). Moreover, there is an
in vitro
requirement for the simultaneous presence of many of the NER proteins. Thus, to
make the 3' incision there is a requirement for XPG, as well as XPA, RPA, XPC,
TFIIH, ERCC1-XPF and IF-7, although it is the endonuclease XPG that makes the actual
cleavage (
42
). An attractive explanation for these observations would be the existence of a
multi-protein repairosome complex. Moreover, that certain mutations in yeast
PCNA are specifically defective in NER but not in DNA replication (
43
) could be an indication that PCNA is required to make a unique set of direct
interactions with NER-specific proteins. The isolation of assembled human and yeast RNA polymerase II holoenzyme complexes has necessitated an altered view of the steps of initiation of transcription. This
demonstration of human repair complexes assembled in the absence of damaged DNA
may similarly prompt a different view of the process of nucleotide excision
repair.
We thank Michael Shales for technical assistance and the investigators indicated
in Materials and Methods that provided us with various antibodies. This work
was supported by a grant to C.J.I. from the National Cancer Institute of
Canada.
*To whom correspondence should be addressed. Tel: +1 416 978 7400; Fax: +1 416
978 8528; Email: cj.ingles@utoronto.ca
REFERENCES
Return