Characterization of the alternative excision repair pathway of UV-damaged DNA in Schizosaccharomyces pombe
Characterization of the alternative excision repair pathway of UV-damaged DNA in Schizosaccharomyces pombe Rie Yonemasu1,2, Shirley J. McCready3, Johanne M. Murray4, Fikret Osman3, Masashi Takao1, Kazuo Yamamoto2, Alan R. Lehmann5 and Akira Yasui1,*
1Institute of Development, Aging and Cancer and 2Biological Institute, Graduate School of Science, Tohoku University, Seiryomachi 4-1, Sendai 980-77, Japan, 3Department of Biochemistry, University of Oxford, Oxford OX1 3QU, UK, 4School of Biological Sciences, Sussex University, Falmer, Brighton BN1 9QG, UK and 5MRC Cell Mutation Unit, Sussex University, Falmer, Brighton BN1 9RR, UK
Received December 31, 1996;Accepted February 17, 1997
ABSTRACT
Schizosaccharomyces pombe cells deficient in nucleotide excision repair (NER) are still able to remove photoproducts from cellular DNA, showing that there is a second pathway for repair of UV damage in this organism. We have characterized this repair pathway by cloning and disruption of the genomic gene encoding UV damage endonuclease (UVDE). Although uvde gene disruptant cells are only mildly UV sensitive, a double disruptant of uvde and rad13 (a S.pombe mutant defective in NER) was synergistically more sensitive than either single disruptant and was unable to remove any photoproducts from cellular DNA. Analysis of the kinetics of photoproduct removal in different mutants showed that the UVDE-mediated pathway operates much more rapidly than NER. In contrast to a previous report, our genetic analysis showed that rad12 and uvde are not the same gene. Disruption of the rad2 gene encoding a structure- specific flap endonuclease makes cells UV sensitive, but much of this sensitivity is not observed if the uvde gene is also disrupted. Further genetic and immunochemical analyses suggest that DNA incised by UVDE is processed by two separate mechanisms, one dependent and one independent of flap endonuclease.
INTRODUCTION
Various repair mechanisms have evolved to counteract the deleterious effects of UV irradiation on cellular DNA. One of the most widely distributed repair enzymes is photolyase, which utilizes visible light to monomerize UV-induced DNA damage (1 ). In the bacterium Micrococcus luteus and T4 phage-infected Escherichia coli, specific excision repair enzymes for UV-induced cyclobutane pyrimidine dimers (CPDs) were found which introduce nicks at CPDs in DNA and have been designated UV endonucleases. These enzymes are, however, pyrimidine dimer DNA glycosylases with a DNA lyase activity which acts 3' of the abasic site created by the glycosylase activity (for a review see 2 ). They do not produce nicks at (6-4) photoproducts (6-4 PPs).
The most widely distributed DNA repair process is nucleotide excision repair (NER), in which an oligonucleotide containing damaged DNA is removed and repair is completed by DNA synthesis and subsequent rejoining of the gaps. In NER, many proteins are necessary for the early steps of damage recognition and introduction of nicks flanking the UV-induced DNA damage (3 ). NER can repair not only UV-induced DNA damage, but also many other types of DNA damage, including chemically induced DNA monoadducts and DNA-DNA crosslinks. NER has been thought to be the only mechanism by which cells can excise CPDs as well as 6-4 PPs from their DNA. In Saccharomyces cerevisiae, for example, null mutants in NER are completely unable to remove such damage. In the fission yeast Schizosaccharomyces pombe, however, strains carrying deletion mutations in NER genes retain a substantial ability to remove both CPDs and 6-4 PPs from their DNA. This has provided evidence for the presence of another excision repair pathway in this organism (4 ,5 ). In previous work we used epistasis analysis to identify a series of candidate genes that were likely to be involved in this process (6 ,7 ), but the gene encoding the enzyme which could carry out the incision step was not identified and all the double mutants that we analyzed retained the ability to remove photoproducts.
An enzymatic activity has been found in extracts of S.pombe cells that introduces nicks immediately 5' of both CPDs and 6-4 PPs (8 ). We recently isolated a cDNA from a S.pombe library which corrected the UV sensitivity of a totally repair deficient (uvrA recA phr) strain of E.coli (9 ). This gene encodes a protein of 69 kDa with significant sequence similarity to an UV endonuclease which we previously isolated from Neurospora crassa (10 ). Since the partially purified recombinant protein prepared from E.coli cells harboring the cloned S.pombe gene showed the same cleavage activity as the Neurospora protein and the reported extract of S.pombe, we concluded that the gene encodes a homolog of the N.crassa UV endonuclease, UV damage endonuclease (UVDE). To understand the mechanism of the putative alternative excision repair, possibly initiated by nicks introduced by UVDE at the damaged sites, we have constructed various S.pombe strains with single and double disruptions in different repair genes. We report here that UVDE initiates the alternative excision repair process, which is distinct from NER. UVDE-mediated excision repair removes UV damage much more rapidly than NER. Furthermore, we show that the nicked sites are further processed by two separate mechanisms, one dependent and one independent of Rad2, a structure-specific flap endonuclease.
MATERIALS AND METHODS
Cell strains
Escherichia coli strain SY2 (uvrA recA phr) (11 ) was used for screening of UV-resistant transformants. For gene disruptions in S.pombe, the S.cerevisiaeLEU2 marker gene was introduced into the cloned genomic uvde gene. Double and triple disruption mutants were constructed by crossing the uvde disruptant with existing repair-defective rad disruptants (or by disruption of the uvde gene in the rad disruptants). Table 1 lists the S.pombe strains used in this paper.
Cell survival
For UV irradiation, 254 nm UVC light was used. Overnight cultures of cells were washed and resuspended in water. Cells were irradiated at ~1 * 106 cells/ml, after which appropriate cell dilutions were plated on YPD (10 g/l yeast extract, 20 g/l polypeptone, 20 g/l glucose, 15 g/l agar) plates. Alternatively, cells were grown for 4 h in liquid medium, resuspended in H2O and plated on YPD prior to UV irradiation.
Cloning of the uvde gene from a genomic library
A S.pombe genomic library (a generous gift of Dr M.Yanagida) was introduced into E.coli SY2 cells. Aliquots of 100 [mu]l of an overnight culture of SY2 cells transformed with the genomic library were irradiated with a UV dose of 0.1 J/m2 on a LB plate with four antibiotics (ampicillin, kanamycin, chloramphenicol and tetracycline) and incubated overnight. Surviving colonies were collected and cultured before the next round of UV irradiation. After three rounds of selection, a number of surviving colonies were examined to test UV resistance. One clone contained a 6.7 kb insert which conferred UV resistance to the recipient cells. By comparison of the restriction maps and partial sequence between the cloned genomic gene and the previously isolated S.pombe cDNA (9 ) encoding the uvde homolog of N.crassa (10 ), we found that the cloned genomic DNA contains the uvde gene of S.pombe.
Immunoassay for repair of photoproducts
Samples of 180 ml cells were irradiated with a dose of 100 J/m2 UVC in water at a density of between 1 and 2 * 107 cells/ml. To these were added 18 ml 10* yeast extract medium and cells were incubated at 30oC. Aliquots of 30 ml were harvested at various times into an equal volume of ethanol to stop repair. DNA was extracted, using glass beads and prolonged vortex mixing of the disrupted cells, followed by phenol/chloroform extraction, ethanol precipitation and RNase treatment. The amounts of DNA in the samples were equalized by running aliquots on agarose gels, scanning the gels and readjusting as necessary. The dot blot immunoassay for CPDs and 6-4 PPs has been described in detail previously (5 ). The two classes of photoproducts are readily distinguished because of their differential sensitivities to hot alkali (6-4 PPs) and E.coli photolyase (CPDs). Lesions were detected using a polyclonal rabbit antiserum, a biotin/ExtrAvidin (Sigma) detection system and a BioRad model GS-670 imaging densitometer.
Cloning and disruption of the uvde gene in wild-type cells
We previously isolated a uvde cDNA clone of S.pombe that showed significant sequence similarity to the uvde gene from N.crassa and we designated the S.pombe gene uvde (9 ). In order to construct a deletion in the uvde gene we have isolated a S.pombe genomic DNA fragment containing the uvde gene by complementation of UV sensitivity in repair-deficient E.coli SY2 cells with a genomic library of S.pombe DNA (see Materials and Methods). By comparison of the restriction maps of the cloned genomic and the cDNA fragments, as well as by partial sequencing of the genomic DNA, we found that the cloned genomic DNA contains the uvde gene and that the gene does not contain any introns. A part of the cloned uvde gene was replaced with the S.cerevisiaeLEU2 gene (Fig. 1 ) and introduced into a wild-type S.pombe strain, resulting in a uvde gene disruptant (uvded). This disruption was confirmed by analysis of the genomic DNA of the disruptant using gene-specific oligonucleotides and PCR (not shown).
UVDE acts in a rapid repair process distinct from NER
Figure 2 shows the UV sensitivities of two uvde disruptants (uvde3 and uvde10) independently constructed from a wild-type strain. The uvde gene disruption made cells only mildly sensitive to UV, even after high doses. A uvde gene disruption was also introduced into a rad13d NER-defective strain. rad13 is the S.pombe homolog of the human XPG and S.cerevisiaeRAD2 genes, which encode an endonuclease which cleaves DNA 3' of the damage during NER. Figure 3 shows that rad13d cells were more sensitive to UV than uvded and the double disruption was much more sensitive than rad13d. These data indicate that the UVDE-mediated repair pathway is distinct from NER in this organism.
Post-incision steps
Schizosaccharomyces pomberad2 encodes a homolog of mouse FEN-1, which is a `flap' structure-specific endonuclease (12 ,13 ). A rad2d mutant is sensitive to UV and was shown to be deficient in a repair pathway distinct from NER (7 ). We found that the double mutant uvded rad2d was more resistant than a rad2d single mutant (Fig. 3 ). This implies that the Rad2 protein is very important for processing nicks introduced by UVDE. If, however, UVDE is not present in the cell, the role of Rad2 in repair of UV damage is less important. The influence of rad2 was further tested by measuring the UV sensitivities of other double and triple mutants. In contrast to the results in NER-proficient cells, in the absence of NER uvded rad13d cells are much more UV sensitive than rad2d rad13d. Furthermore, a rad13d uvded rad2d triple disruptant had the same UV sensitivity as the rad13d uvded double disruptant. These results show that the Rad2 protein is involved only in the UVDE-mediated second pathway and that there are both rad2-dependent and rad2-independent components of the UVDE-mediated repair pathway.
Relation of Rad12 to UVDE
A UV-sensitive S.pombe mutant, rad12, has been reported to be severely deficient in UV endonuclease activity (14 ). Furthermore, a rad13 rad12 double mutant was more sensitive than either single mutant. These results were interpreted as suggesting that rad12 defined a second repair pathway and they raised the possibility that rad12 might encode UVDE (14 ). However, in a cross between a uvded and a rad12-502 point mutant (rad12 has not yet been cloned), the two genes segregated as unlinked loci. Furthermore, as shown in Figure 5 , the rad12-502 uvded double mutant was more sensitive to UV than either single mutant. These results rule out the possibility that rad12 encodes UVDE and demonstrate that the UV sensitivity of the rad12 mutant is not caused by a defect in the uvde pathway.
Effects of the introduction of the uvde gene in a multi-copy plasmid into repair-deficient cells
To determine the influence of increased uvde expression on UV resistance in various repair-deficient strains, we introduced the genomic uvde gene with a 3.9 kb 5' upstream sequence in a multi-copy plasmid harboring ade6 as the selection marker into different mutants. The plasmid (pSPuvde-g) rendered uvded (Fig. 6 a), uvded rad2d rad13d (Fig. 6 b) and uvded rad12-502 (Fig. 6 c) cells more resistant to UV than the corresponding uvde+ strains, namely wild-type, rad2d rad13d and rad12-502 respectively. Interestingly, introduction of the plasmid into uvded rad2d cells had no influence on their UV response (Fig. 6 d). This shows that in NER+ cells high level expression of UVDE increases the biological effectiveness of photoproduct repair, but only if Rad2 is also present.
DISCUSSION
In an earlier report we showed that S.pombe has a second repair pathway, distinct from NER, for removing UV photoproducts (5 ). Further work implicated several genes in this second repair pathway (6 ), but we were not able to identify the protein which initiates the pathway. In the current work we have shown conclusively that UVDE is this key enzyme. The double disruptant of uvde and the NER rad13 gene is synergistically more sensitive to UV than either single disruptant (Fig. 3 ), demonstrating that UVDE-mediated repair is distinct from NER. Furthermore, this double disruptant is completely unable to repair UV photoproducts (Fig. 4 ). Since partially purified UVDE of S.pombe has an endonuclease activity for CPDs and 6-4 PPs in in vitro analyses (9 ) and since both types of damage disappeared rapidly in rad13d cells, we infer that the second excision repair pathway of S.pombe is a rapid process initiated by the nicking activity of UVDE at both types of UV damage.
By using cells disrupted in uvde and other repair genes, we have carried out a genetic analysis of this pathway. The single uvde disruptant is mildly sensitive to UV, as was also found with uvde-deficient mus-18 mutants of N.crassa (15 ,10 ). Several other genes have been proposed to be involved in the second repair pathway. We suggested that the rad2, rad18 and rhp51 genes, which belong to the same epistasis group, distinct from NER, were involved in the later stages of the second pathway (6 ). A rad2 disruptant is more sensitive to UV than a rad2uvde double disruptant (Fig. 3 ). This indicates that Rad2 has a crucial role in processing of UVDE-nicked sites, resulting from the action of UVDE. We infer that following cleavage of the damaged DNA by UVDE, strand displacement synthesis occurs from the nicked sites with the formation of a flap structure, known to be an excellent substrate for Rad2-like nucleases (12 ,13 ). In the absence of Rad2, incisions by UVDE at damaged sites results in the accumulation of potentially lethal intermediates, thereby conferring UV sensitivity upon rad2 mutants. If UVDE is also absent, these intermediates are not produced and there is a reduced requirement for Rad2. Thus the rad2 uvde double mutant is less sensitive than the rad2 single mutant.
Expression experiments further showed the importance of Rad2 in the UVDE-dependent repair pathway in NER-proficient cells. Whereas introduction of a genomic uvde fragment made uvded cells more UV resistant than wild-type cells, it had no influence on rad2d cells. This confirms that Rad2 is most important for processing UVDE-induced nicks in NER-proficient cells. As discussed below, UVDE-induced nicks are also substrates for a rapid Rad2-independent repair process, only a part of which, however, leads to complete repair of DNA damage and cell survival. Therefore, in rad2d cells, UV-induced damage processing is greater by the incomplete Rad2-independent repair than by the complete NER. Reasonably, high expression of UVDE in rad2d cells does not increase cell survival. In the absence of NER, the rad2 rad13uvde triple disruptant was more sensitive than the rad2 rad13 double disruptant, indicating that some nicks produced by UVDE can also be processed and repaired by a Rad2-independent pathway(s). Introduction of a high copy plasmid harboring a genomic uvde gene increased the resistance of the triple disruptant to a level slightly higher than that of the double rad2 rad13 disruptants. These results are quite different from those for NER+ cells, because the Rad2-independent repair process is the only repair mechanism available in the rad13 rad2 double mutant.
Another gene, rad12, has been reported to be related to UVDE-mediated repair (14 ). Endonuclease activity against a synthetic oligonucleotide containing a thymine-thymine dimer was not detected in extracts of rad12-502 cells. Genetic crosses showed, however, that uvde and rad12 are unlinked loci and, as shown in Figure 5 , uvde rad12 double mutants were more UV sensitive than either single mutant, indicating that rad12 is not the structural gene for UVDE and that the UV sensitivity of rad12 is not related to a deficiency in UVDE-mediated repair. These data are consistent with the report that rad12 and rad2 belong to different epistasis groups for UV sensitivity (14 ). rad12 may be involved in regulation of the expression of uvde, so that it can be expressed at the right moment in the cell cycle or in response to DNA damages. However, mutation in the rad12 gene had no influence on the increase in UV resistance due to introduction of a plasmid containing the genomic uvde gene with 3.9 kb upstream sequence (Fig. 6 c). Although UVDE activity in extracts of S.pombe appears to be UV inducible (14 ; J.M.Murray and A.M.Carr, in preparation), Northern analysis of the uvde gene did not show any UV (200 J/m2 for wild-type and 20 J/m2 for rad13d) inducibility at the transcription level in the first hour after UV (not shown).
ACKNOWLEDGEMENTS
We are grateful to Dr M.Yanagida, Kyoto University, and Dr H.Murakami, Tokyo University, for providing us with the genomic library and the expression vector pCLX respectively. We thank Drs S.Yasuhira and Y.Kubota for their experimental help. This work was supported in part by a Grant-in-Aid for Scientific Research on Priority Areas from the Japanese Government (no. 08280101) to A.Y., a Welcome Trust Project Grant (no. 043822/Z/95) to S.M. and a MRC Project Grant to J.M.M.
