Binding and repair of mismatched DNA mediated by Rhp14, the fission yeast homologue of human XPA

: Rhp14 of Schizosaccharomyces pombe is homologous to human XPA and Saccharomyces cerevisiae Rad14, which act in nucleotide excision repair of DNA damages induced by ultraviolet light and chemical agents. Cells with disrupted rhp14 were highly sensitive to ultraviolet light, and epistasis analysis with swi10 (nucleotide excision repair) and rad2 (Uve1-dependent ultraviolet light damage repair pathway) revealed that Rhp14 is an important component of nucleotide excision repair for ultraviolet light-induced damages. Moreover, defective rhp14 caused instability of a GT repeat, similar to swi10 and synergistically with msh2 and exo1. Recombinant Rhp14 with an N-terminal hexahistidine tag was purified from Escherichia coli. Complementation studies with a rhp14 mutant demonstrated that the tagged Rhp14 is functional in repair of ultraviolet radiation-induced damages and in mitotic mutation avoidance. In bandshift assays, Rhp14 showed a preference to substrates with mismatched and unpaired nucleotides. Similarly, XPA bound more efficiently to C/C, A/C, and T/C mismatches than to homoduplex DNA. Our data show that mismatches and loops in DNA are substrates of nucleotide excision repair. Rhp14 is likely part of the recognition complex but alone is


INTRODUCTION
In the course of nucleotide excision repair (NER) 1

of damaged bases, a preincision
complex is assembled at the lesion. This complex contains XPA, RPA, XPC, hHR23B and TFIIH in humans, and homologous proteins in other eukaryotes (1)(2)(3).
Subsequently, the damaged base is released from DNA in a 24-32 nucleotide long oligomer after dual incision by XPG, incising 3' to the lesion, and ERCC1-XPF, incising 5' to the lesion. Finally, repair is completed by resynthesis of the gap and ligation. It is not exactly known, whether one of the components of the preincision complex is the first damage recognition factor, or whether rather the entire preincision complex is required.
However, the single proteins, as well as the complexes XPA-RPA and XPC-hHR23B exhibit only a moderate preference for damaged DNA. Thus, these factors likely contribute to recognition of damaged DNA, but alone might be not sufficient for a high discrimination.
NER is able to repair a variety of bulky DNA adducts, like  photoproducts and cyclobutane pyrimidine dimers induced by ultraviolet (UV) radiation, intrastrand crosslinks produced by cis-diamine-dichloroplatinum(II), and adducts formed by carcinogens such as benzo[a]pyrene diol epoxide (1,(12)(13)(14). In addition, nonbulky lesions such as methylated bases, and apurinic/apyrimidinic sites and to a low level, even G/A and G/G base-base mismatches are processed by NER (1,12). In general, DNA that contains a lesion and some degree of helical distortion is more efficiently processed by NER than lesions without helical distortions or distortions alone (15)(16)(17)(18). Although NER incises mismatch-containing DNA rather poorly in vitro (12,(15)(16)(17), there is some evidence that NER factors have a function in correction of mismatched bases. Several mutations in the S. cerevisiae gene RAD3, encoding a homologue of human XPD, cause increased spontaneous mutation rates (19,20). A mutated mei-9 of Drosophila melanogaster, encoding a homologue of human XPF, results in increased postmeiotic segregation of genetic markers (21,22). Elevated postmeiotic segregation frequencies are the consequence of non-repaired mismatches formed in heteroduplex DNA during meiotic recombination. In vitro mismatch correction is reduced in protein extracts of a Drosophila mei-9 mutant (23). Mutations in the Schizosaccharomyces pombe NER genes swi10 (ERCC1 homologue), rad16 (XPF homologue), and rhp14 (XPA homologue) cause a defect in repair of base-base mismatches, arising during vegetative growth and meiotic recombination (24).
This study aimed to extend the analysis on mismatch correction by NER factors. We report the characterization of the fission yeast S. pombe Rhp14 with respect to its function in DNA repair. A rhp14 gene disruption mutant was constructed and tested for sensitivity to UV light and for stability of a GT repeat. In addition, recombinant Rhp14 was purified and analyzed for its capacity to bind to base-base mismatches and small loops. 6 (minimal medium) were prepared as described (34,35). EMM (Edinburgh minimal medium) without thiamine (36) was used for expression of rhp14, inserted in derivatives of pREP42 or pREP82 (see below). In these vectors rhp14 is under the control of the nmt promoter, which is transcribed in the absence of thiamine (37,38). Reactions including S. pombe crude extracts were incubated for 20 min at 4 ºC and subsequently loaded on 6% non-denaturing polyacrylamide gels. Reactions with purified [His] 6 -Rhp14 or XPA were incubated for 30 min at 4 ºC and loaded on 5% nondenaturing polyacrylamide gels. Electrophoresis was performed in 40 mM Tris-acetate (pH 7.5) at 90 V and 4 ºC. XPA with an N-terminal [His] 6 -tag was purified as described previously (5).
For quantification of DNA binding, gels were exposed to a phosphoimager and subsequently analyzed using ImageQuant software (Molecular Dynamics). The percentage of bound substrate was calculated from the intensity of shifted bands relative to the sum of bound and free radioactively labeled oligonucleotides. A reaction mix without protein was loaded on the gels to normalize the background level of radioactivity at the position of the shifted bands in protein containing reactions as well as to serve as a control for estimation of the total amount of substrates (see Fig. 6A, lane 1). A smear in the gels between free and bound substrates was usually detected when reactions contained either Rhp14 or XPA. The smear likely reflects loss of protein-DNA interaction during electrophoresis and was not included in the calculation.  Determination of mitotic mutation rates-Reversion rates per cell division were calculated from the median number of Ade + per total cell number of cultures (40). For fluctuation tests, either ade6-(GT) 8 or ade6-485 was used. The ade6-(GT) 8 allele represents a (GT) 8 repeat, which was constructed by insertion of seven GT units at an existing GT site (32). Nine colonies grown on YEA were each inoculated in 5-ml YEL and grown to stationary phase. Appropriate amounts were plated on MMA for selection of revertants and on MMA supplemented with adenine for determination of cell titers.

Construction of plasmids for expression of Rhp14 in S. pombe-In
Plates were incubated for 7 days at 30 ºC.
Cells were harvested by centrifugation, suspended in 600 µl 0.85% NaCl and plated on EMM. Appropriate dilutions were plated on EMM (+ adenine) for cell titer determination.
Plates were incubated for 12 days at 30 ºC. For each strain background, experiments were carried out at least three times. 8

Rhp14 is involved in the NER pathway for UV induced damages-The S. pombe
rhp14 gene was identified by the S. pombe Genome Sequencing Project (http://www.sanger.ac.uk/Projects/S_pombe/). The deduced amino acid sequence shows 33% identity to human XPA and 39% identity to Rad14 of S. cerevisiae, which are involved in the damage recognition step of NER (1-3). A rhp14 gene disruption strain was constructed as described in "Experimental Procedures" and was tested for cell survival after irradiation with different doses of UV light (Fig. 2). rhp14 cells were highly sensitivity to UV, to a similar extent as swi10, defective in the NER 5' endonuclease (26,41,42), and somewhat more affected than rad2 cells, defective in the second, Uve1-dependent pathway for repair of UV damages in S. pombe (43,44).
The rhp14 swi10 double mutant showed similar sensitivity to UV as either single mutant, while the rhp14 rad2 double mutant was clearly more sensitive (Fig. 2). Thus, Rhp14 is an important factor of NER of UV induced damages, but not component of the Uve1dependent pathway.
GT repeat stability is affected in the rhp14 mutant-The long-patch mismatch repair (MMR) system of S. pombe efficiently corrects base-base mismatches, except C/C, as well as one to several unpaired nucleotides (24, 25, 30-32, 35, 45). A genetic test system was developed that allows measuring instability of GT repeats (32). Loss of MMR by a mutation in msh2, msh6, or pms1 resulted in dramatically increased instability of GT repeats. Exo1 is likely involved in MMR-dependent repair of base-base mismatches (29,30), but has only a minor and rather MMR independent function in GT repeat stability (32). It has been previously shown that NER factors of S. pombe are involved in MMR independent short-patch repair of base-base mismatches arising during meiotic recombination and vegetative growth (24). Here we tested the consequences of mutations in the NER factors rhp14 and swi10 on Ade + reversions of a (GT) 8 repeat introduced into the ade6 gene. In addition, epistasis analysis, including msh2 and exo1, was performed (Table I). ade6-(GT) 8 originated from an insertion of seven GT units at an existing GT site, and thus represents a frame shift mutation causing adenine auxotrophy (32). Reversions to Ade + , which restore the reading frame, can either occur by deletions of 2 or 8-bp (1

or 4 GT units) or by insertions of 4-bp (two GT units).
In repair-proficient wild type a reversion rate of 3 × 10 -9 was measured (Table I).
With a 1.8 × 10 4 -fold increased rate, the (GT) 8 repeat was highly destabilized in the msh2 mutant, as expected from the previous study (32). Reversion to Ade + occurred 12times more frequently in exo1 cells than in wild type. A similar increase was found with the NER mutants rhp14 and swi10. Compared to respective single mutants, a further increase of reversion rates was found in the double mutants msh2 rhp14, msh2 swi10, rhp14 exo1, and swi10 exo1, but not with rhp14 swi10. Thus, Rhp14 and Swi10 act in the same pathway for maintaining GT repeat stability, and independent of Msh2 and  (Table II). In contrast, almost all of the revertants in msh2 background produced white colonies (Table II), and sequencing of 11 of them exclusively identified a (GT) 7 repeat and thus a 2-bp deletion (32). In rhp14 and swi10 mutants, about 70% of revertants formed pink colonies. Thus, (GT) 8 mainly reverted to Ade + by insertions of 4-bp (Table II). Among the white revertants, both (GT) 4 and (GT) 7 repeats were identified by sequencing (data not shown). Thus, the mutation spectra of rhp14 and swi10 are similar to that of wild type.

C/C mismatch binding by Cmb2 is not affected in rhp14 cells-Since Rhp14 of S.
pombe is a component of MMR-independent repair of mismatched DNA (24, and this study), it is conceivable that Rhp14 specifically recognizes mismatched or unpaired nucleotides. The defect in correction of base-base mismatches in NER mutants is most pronounced for C/C mismatches, which are not substrate of MMR (24,30,31). The Cmb1 protein was recently identified as a recognition factor for C/C and other cytosine containing mismatches (28). In crude protein extracts of a cmb1 disruption strain, binding to cytosine-containing mismatches was abolished, with the exception of C/C, where still some binding was detected. The gene encoding the second C/C binding activity, Cmb2, was not yet identified. Therefore, we started our mismatch-binding studies on Rhp14 with the analysis of the C/C binding capacity of crude extracts of rhp14 mutants, either additionally mutated in cmb1 or not (Fig. 3). In crude extracts of wild-type cells, specific binding to C/C and T/C was detectable. In cmb1 cells, binding to C/C by Cmb2 remained. Binding by either Cmb1 or Cmb2 was not affected in rhp14 extracts, and C/C binding by Cmb2 was still present in extracts of the rhp14 cmb1 double mutant. Thus, Cmb2 is not encoded by the rhp14 gene.  (Fig. 4A). The E. coli proteins were detected in the flow through fractions, while most of Rhp14 bound to the column and eluted in the presence of ~100-160 mM NaCl. That the purified protein was indeed [His] 6 -Rhp14 was proved by a Western using a His-tag specific antibody as probe (Fig. 4B).
[His] 6 -Rhp14 is active in DNA repair in vivo-To test whether the [His] 6 -tag interferes with the function of Rhp14, complementation of DNA repair defects caused by mutated rhp14 was studied. Therefore, an S. pombe rhp14 mutant strain was transformed with plasmids derived from pREP42L either containing rhp14 (pRhp14) or tagged rhp14 (p[His] 6 Rhp14). As controls wild-type and rhp14 strains were transformed with the empty vector pREP42L. Both, complementation of UV sensitivity and of increased mitotic mutation rates were tested.

Binding of Rhp14 to base-base mismatches and DNA loops-Rhp14 is based on its
homology to XPA and Rad14, likely involved in the recognition step of NER, and might therefore have some preference in binding to modified over homoduplex DNA. In addition, Rhp14 is required for mismatch correction (24; and Fig. 5B) and to some degree for GT repeat stability (Table I). Therefore, we tested whether purified Rhp14 can specifically bind to substrates containing base-base mismatches or unpaired nucleotides by a band shift assay. The oligonucleotides were in the sequence context of ade6-485 (see Fig. 1), which is known to be substrate of NER in vivo (24). Substrates containing C/T, C/C or C/∆ mismatches were about two-times better bound than homoduplex DNA (Fig. 6). A similar affinity was found for T/G and T/T (data not shown), while binding to C/A was only slightly stronger and not significantly different to homoduplex binding.
Rhp14 also showed increased affinity to substrates containing loops with one, two or four nucleotides (Fig. 7). Tendentiously, the loops with four unpaired nucleotides,

DISCUSSION
This study reports the characterization of Rhp14 with respect to its function in repair of UV damages and mismatched DNA, as well as its ability to specifically bind to basebase mismatches and small insertion/deletion loops. Rhp14 is, based on its homology to human XPA and S. cerevisiae Rad14, likely involved in the recognition step of NER.
In the initial experiment we found that the rhp14 mutant was as sensitivity to UV light as the swi10 mutant, defective in the NER 5' endonuclease, and UV sensitivity was not further increased in the rhp14 swi10 double mutant (Fig. 2). Thus, Rhp14 plays indeed an important role in NER. Consistently, the rhp14 rad2 mutant was more sensitive to UV than either single mutant, showing that Rhp14 is not involved in the Uve1-dependent pathway.
Since NER factors in S. pombe have an MMR-independent role in repair of basebase mismatches (24), we analyzed the effects of mutated rhp14 and swi10 on stability of a GT repeat, a common type of microsatellite in eukaryotic genomes. Reversions of the (GT) 8 repeat occurred with high frequency in msh2 cells, defective in MMR (Table I).
Compared to wild type, about 12-14 fold increased reversion rates were observed with exo1, rhp14 and swi10 mutants. Epistasis analysis with double mutants revealed that Rhp14 and Swi10 act in the same pathway for maintaining GT repeat stability and distinct from Msh2 and Exo1. The distribution of insertions and deletions of GT units was similar to that of wild type (Table II), while in the msh2 mutant, most reversions occurred by deletion of 2-bp (32; and Table II). The data confirm previous studies, which showed that MMR plays a dominant role in stability of GT repeats and other microsatellites (32,(46)(47)(48)(49) and further suggest that NER contributes to maintaining the tract length of a (GT) 8 repeat in S. pombe. Thus, NER might serve as a backup system for stabilization of microsatellites. Consistently with the genetic analysis, Rhp14 showed an about 2-fold preference to substrates with either a GT, (GT) 2 , AC or (AC) 2 loop (Fig.   7). Such loops can be formed in GT repeats by strand slippage during replication.
DNA-binding capacity was studied with Rhp14 containing an N-terminal [His] 6 -tag.
To ensure that the tag did not interfere with the function of the protein, we tested complementation of DNA repair defects caused by mutated rhp14. Both UV sensitivity and increased ade6-485 reversion rates were complemented to the same degree by [His] 6 -tagged Rhp14 and by untagged Rhp14 expressed on a plasmid (Fig. 5). Thus, [His] 6 -Rhp14 is functional in vivo.
In band shift assays the Rhp14 protein showed preferential binding to substrates containing base-base mismatches or loops with unpaired nucleotides (Fig. 6 and 7). Our recent study revealed that mutated rhp14 and swi10 caused elevated mitotic mutation rates of the ade6-485 allele (a C to G transversion), likely due to the failure to repair C/C mismatches (24). In addition, short-patch repair of C/C mismatches formed during meiotic recombination was strongly affected in the NER mutants. We found no significant difference in binding of Rhp14 to C/C and other types of mismatches (Fig. 6).
Thus, the observation that NER predominantly repairs C/C is rather due to the fact that C/C mismatches are not processed by MMR, and likely not a consequence of preferential recognition of C/C. Consistently, in the absence of MMR, NER can also process other types of mismatches (24). However, it should be noted that S. pombe contains two activities, Cmb1 and Cmb2, which recognize C/C (28). Cmb1 also binds to other types of cytosine-containing mismatches, while Cmb2 exclusively binds to C/C. Generally, only weak effects on DNA repair were found in a cmb1 mutant 3 , which might be due to redundant functions with Cmb2, whose gene was not yet identified. Since binding to C/C remained in the protein extract of a rhp14 cmb1 mutant, the rhp14 gene does not encode for Cmb2 (Fig. 3). The role of Cmb1 and Cmb2 in DNA repair is not yet understood. One possibility is that they act as accessory factors in DNA repair.
Binding of Rhp14 to mismatches and loops was about 2-fold stronger than to homoduplex. A similar specific affinity was found for binding to C/C, A/C and T/C mismatches by XPA (Fig. 8). A recent study revealed that XPA shows an about 2-fold higher affinity to  photoproducts than to unmodified homoduplex (10). The preference for damaged and mismatched DNA might be due to local melting of the double helix. In fact, XPA binds with similar affinities to substrates containing either a bubble with three mispaired nucleotides or a benzo[a]pyrene adduct (14). It was suggested that XPA recognition requires sites in DNA with disrupted base-pairing. To serve as recognition signal for XPA, the helical distortion should be in the context of duplex DNA, since single-stranded DNA is not better bound than double-stranded unmodified DNA (14). The observation that NER is able to correct mismatches in S. pombe suggests that mismatches are actively recognized by NER factors. Rhp14 likely contributes to discrimination between modified and unmodified DNA, but for efficient recognition additional factors are likely required. An interesting question is whether correction of mismatches and loops by NER is a special situation in S. pombe and may be in a few other organisms, or whether it is a general feature of eukaryotes. Only a limited set of This activity is also present in rhp14 cmb1 extracts, demonstrating that Cmb2 is not encoded by rhp14. Oligonucleotides were in the M13mp9 context (39). Conditions of the band shift assay are described in "Experimental Procedures".   8

repeat
Reversion rates of the (GT) 8 repeat in the ade6 gene (32) were determined as described in "Experimental Procedures". Reversions can occur by deletion of 2-or 8bp or by insertion of 4-bp (see Table II