HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 276, Issue 34, 31551-31560, August 24, 2001

This Article Abstract Full Text (PDF) All Versions of this Article: 276/34/31551    most recent M103588200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Greenberg, R. B. Articles by Saffran, W. A. Search for Related Content PubMed PubMed Citation Articles by Greenberg, R. B. Articles by Saffran, W. A.

Recombinational and Mutagenic Repair of Psoralen Interstrand Cross-links in Saccharomyces cerevisiae^*

Ross B. Greenberg, Marie Alberti^§, John E. Hearst^§, Mark A. Chua, and Wilma A. Saffran^¶

From the  Department of Chemistry and Biochemistry, Queens College, City University of New York, Flushing, New York 11367 and the ^§ Department of Chemistry, Lawrence Berkeley National Laboratory, University of California, Berkeley, California 94720

Received for publication, April 23, 2001

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Psoralen photoreacts with DNA to form interstrand cross-links, which can be repaired by both nonmutagenic nucleotide excision repair and recombinational repair pathways and by mutagenic pathways. In the yeast Saccharomyces cerevisiae, psoralen cross-links are processed by nucleotide excision repair to form double-strand breaks (DSBs). In yeast, DSBs are repaired primarily by homologous recombination, predicting that cross-link and DSB repair should induce similar recombination end points. We compared psoralen cross-link, psoralen monoadduct, and DSB repair using plasmid substrates with site-specific lesions and measured the patterns of gene conversion, crossing over, and targeted mutation. Psoralen cross-links induced both recombination and mutations, whereas DSBs induced only recombination, and monoadducts were neither recombinogenic nor mutagenic. Although the cross-link- and DSB-induced patterns of plasmid integration and gene conversion were similar in most respects, they showed opposite asymmetries in their unidirectional conversion tracts: primarily upstream from the damage site for cross-links but downstream for DSBs. Cross-links induced targeted mutations in 5% of the repaired plasmids; all were base substitutions, primarily T  C transitions. The major pathway of psoralen cross-link repair in yeast is error-free and involves the formation of DSB intermediates followed by homologous recombination. A fraction of the cross-links enter an error-prone pathway, resulting in mutations at the damage site.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

DNA interstrand cross-linkers are used widely in cancer chemotherapy because of their high cytotoxicity in replicating cells (1). These lesions are complex, and their repair involves several different DNA repair pathways. As with other forms of chemical damage, excision repair systems incise the damaged DNA strands; however, there is no undamaged strand to act as a template, and full repair requires the participation of additional pathways. Recombinational repair pathways are involved in restoring the intact duplex structure after excision (2-7). Additionally, cross-links efficiently induce mutations, implicating error-prone pathways in their repair (8, 9).

Psoralens are photoreactive DNA cross-linking agents that react with pyrimidine bases on opposite DNA strands in the presence of near ultraviolet light; 5'-TpA-3' sequences are preferred cross-linking sites (10, 11). There are two photoreactive positions in the psoralen molecule, the 4',5' furan and the 3,4 pyrone double bonds, which can undergo sequential photoreactions to form cross-links. The major products of the first photoreaction step are furan-side monoadducts; these can undergo a second photoreaction at the pyrone side to generate interstrand cross-links. Psoralen plus ultraviolet A (PUVA) therapy is used to treat the skin disorders psoriasis and vitiligo; although effective, this treatment has been found to induce nonmelanoma skin cancers in a dose-dependent manner (12).

Both nucleotide excision repair (NER)¹ and recombinational repair pathways participate in the error-free repair of psoralen cross-links in Escherichia coli (2, 5, 13-15). The Uvr (A)BC complex makes single-strand incisions on the 5' and 3' sides of the cross-link. This generates an intermediate with a gap opposite to a lesion consisting of the excised oligonucleotide fragment cross-linked, through psoralen, to the uncut strand. Repair of the gap is accomplished by RecA-mediated recombination with an undamaged homologous copy of the affected DNA sequence. This yields an intact strand opposite the lesion, which can then be removed by a second round of NER.

Psoralen cross-link repair in eukaryotes is not as well understood but is likely to differ from prokaryotes. In the yeast Saccharomyces cerevisiae, NER of psoralen monoadducts generates short-lived single-strand breaks that are efficiently rejoined. However, psoralen cross-links induce the formation of double-strand breaks (DSBs) in chromosomal DNA; these breaks are long-lived, and rejoining depends on homologous recombination (3, 4, 16). Production of the DSBs depends on the NER genes RAD2 and RAD3. Nitrogen mustard cross-links also induce DSBs in yeast (7) and mammalian (17) cells; however, in contrast to psoralen cross-links, the formation of nitrogen mustard-induced DSBs does not depend on NER.

Rejoining of psoralen cross-link-induced DSBs and regeneration of intact chromosomal DNA requires the recombinational repair genes RAD51 and RAD52 (18). These observations suggest that error-free repair of psoralen cross-links in yeast can be carried out by a two-phase process: 1) NER acts on an interstrand cross-link to produce a DSB at the lesion site, and 2) the DSB is repaired by homologous recombination.

This model predicts that, because the recombinogenic repair intermediate of a psoralen interstrand cross-link is a DSB, both psoralen cross-links and DSBs will induce similar levels and patterns of recombination. Psoralen monoadducts, which do not generate DSBs, should not be recombinogenic. We have tested these predictions by comparing the repair of plasmid molecules carrying a single site-specifically placed psoralen monoadduct, psoralen cross-link, or DSB at the same position. Induced recombination, measured as both gene conversion and crossing over, between the damaged plasmid and homologous chromosomal sequences was similar for both forms of double-strand damage. Psoralen monoadducts did not induce recombination. Psoralen cross-links, but not DSBs, also enter an alternate error-prone pathway that produces mutations at the damage site.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Plasmid Construction-- The phis3 plasmids are yeast shuttle plasmids, derived from the phagemid vector pIBI25, carrying the yeast TRP1 and HIS3 genes. The TRP1-ARS1 EcoRI fragment from YRP12 (19) was inserted into the EcoRI site of the polylinker sequence of pIBI25, and a modified his3 gene carrying an 8-bp XbaI linker insertion mutation was subcloned into the BamHI polylinker site. The plasmids phis3-75X, phis3-207X, phis3-304X, and phis3-622X have XbaI linker insertions at positions 75, 207, 304, and 622 of the HIS3 gene, respectively. To construct the his3X alleles carrying the XbaI site insertions, the XbaI site within the polylinker region of pUC18-HIS3 was first removed by digesting with XbaI, filling in the cohesive ends with Klenow fragment, and recircularizing the plasmid. XbaI linker insertions at positions 75, 207, 304, and 622 were carried out by complete or partial digestion with AvaII, MscI, HindIII, or KpnI, respectively, followed by the addition of XbaI linkers and recircularization. XbaI linker insertion sites were confirmed by restriction mapping and DNA sequencing.

Yeast Strains-- Yeast strains are derivatives of W303, MAT leu2-3,112 trp1-1 ade2-1 ura3-1 can1-100 his3-11,15 rad5-535. The his3-11,15 allele was replaced with a his3X allele by two-step gene replacement (20). WS101, WS102, WS103, and WS104 carry the his3 alleles his3-75X, his3-207X, his3-304X, and his3-622X, respectively. RY1, RY2, RY3, and RY4 are RAD5 derivatives of WS101, WS102, WS103, and WS104, respectively.

Psoralen-monoadducted Oligonucleotide-- The 14-bp oligonucleotide 5'-CAGGCCGTACGCAG-3' was used for the preparation of uniquely psoralen-adducted plasmid DNA molecules. This oligonucleotide spans positions 411-424 within the coding strand of the HIS3 gene; the sequence 5'-CGTACG-3' is a BsiWI site and contains the preferred psoralen target sequence, 5'-TA-3'. The 14-mer was annealed to the complementary 8-mer 5'-GCGTACGG-3' and photoreacted at 366 nm with 4'-hydroxymethyl-4,5',8-trimethylpsoralen. Cross-linked oligonucleotides were purified by reverse-phase HPLC. The psoralen cross-links were photoreversed by irradiation at 254 nm, and 14-mers with furan-side monoadducts were resolved from 8-mers and unmodified or pyrone-side monoadducted 14-mers by reverse-phase HPLC (21, 22).

Preparation of Plasmids with Site-specific Damage-- Plasmids with site-specifically placed psoralen monoadducts or cross-links were prepared by the extension of psoralen-modified primers annealed to single-stranded circular DNA (23). Single-stranded phagemid DNA was isolated according to Sambrook et al. (24). Modified oligonucleotide primers were phosphorylated by treatment with T4 polynucleotide kinase and annealed to single-stranded template DNA in several 500-µl annealing reactions, each containing 5 pmol of template DNA and 160 pmol of kinased primer, in 20 mM Tris-HCl, pH 7.4, 2 mM MgCl₂, and 50 mM NaCl. Interstrand psoralen cross-links were formed by irradiating the annealed preparations on ice in drops of 500 µl at 365 nm for 45 min with a Schleicher and Schüell RAD-FREE long wave UV lamp with a BLE-760B Spectronics Corp. bulb. The total dose was 130 J·m². The irradiated samples were pooled, and noncross-linked primers were removed by ultrafiltration through Centricon-30 filters in 0.1× TEN7.4 (1 mM Tris-HCl, 1 mM EDTA, pH 8.0) followed by the addition of urea to 8 M and column chromatography on Sepharose 4B in 0.1× TEN7.4.

Double-stranded circular DNA was formed by primer extension of 5 pmol of primer template complex in a 650-µl reaction volume with 50 units of T4 DNA polymerase and 200 units of T4 DNA ligase in 25 mM Tris, pH 7.5, 5 mM MgCl₂, 40 mM NaCl, 4 mM dithiothreitol, 7.7% glycerol, 1.5 mM ATP, and 0.8 mM each dATP, dCTP, dGTP, and dTTP for 5 min at room temperature followed by 90 min at 37 °C. Closed circular plasmid DNA molecules were prepared by centrifugation through CsCl gradients with 0.4 mg/ml ethidium bromide. The preparation of monoadduct-containing plasmid molecules was similar, but the irradiation, urea treatment, and Sepharose chromatography steps were omitted, and the annealing step was followed immediately by primer extension. Unincorporated primers remaining after the CsCl gradient were removed by urea treatment.

The position of 4'-hydroxymethyl-4,5',8-trimethylpsoralen modification within the plasmid was confirmed by BsiWI digestion. Cross-links completely inhibited and monoadducts partially inhibited BsiWI cleavage at the target site (Fig. 1A). The presence of interstrand cross-links was verified by testing for the ability of cross-linked DNA to rapidly renature after denaturation (Fig. 1B). Plasmid DNA (0.2 µg) was digested with BamHI, which produces a 1.8-kilobase fragment containing the HIS3 gene. The digested DNA was desalted on Sepharose spin columns to remove Mg²⁺, which interferes with denaturation. The samples were divided in half; one portion was denatured by incubation for 10 min with 0.2 N NaOH. Both portions were analyzed by electrophoresis on nondenaturing agarose gels in Tris acetate-EDTA buffer. The 1.8-kilobase fragment ran as double-stranded DNA, indicating that the HIS3 gene was cross-linked, whereas the larger fragment was fully denatured and ran as single strands. The plasmid preparations with furan-side monoadducts were completely denatured by this treatment. However, a 15-min irradiation of the BamHI-digested plasmid converted the 1.8-kilobase HIS3 fragment to the rapidly renaturing form, confirming that these preparations contained cross-linkable monoadducts within HIS3. A control preparation synthesized with an unmodified 14-mer primer was tested similarly for the presence of cross-links or cross-linkable adducts at the target site. BsiWI cut this preparation completely, and the BamHI fragments were fully denatured.

View larger version (44K): [in this window] [in a new window]   Fig. 1.   Psoralen damage position. A, the inhibition of BsiWI digestion by psoralen modification. Lanes 1-3, undigested plasmid; lanes 4-6, BsiWI-digested plasmid; lanes 1 and 4, unmodified; lanes 2 and 5, psoralen monoadducts; lanes 3 and 6, psoralen interstrand cross-links. MA, monoadduct; XL, cross-link. B, the rapid renaturation of psoralen cross-linked DNA fragments. Odd lanes, no treatment; even lanes, alkali denaturation followed by electrophoresis at a neutral pH. Cross-linked fragments run as double-stranded (ds) DNA, and noncross-linked fragments run as single-stranded (ss) DNA. Lanes 1 and 2, unmodified plasmid; lanes 3 and 4, psoralen monoadducts; lanes 5 and 6, psoralen monoadducts with UV irradiation; lanes 7 and 8, psoralen cross-links.

Plasmid Repair and Genetic Analysis-- Yeast cells were transformed with unmodified or damaged plasmid DNA as described previously (25) using 0.1 µg of plasmid and 5 µg of single-stranded carrier DNA per sample. Transformed cells were selected on tryptophan omission medium, and colonies were scored after 4 days. Repair efficiency was calculated as the ratio of Trp⁺ transformants with damaged plasmid to Trp⁺ transformants with undamaged plasmid. Trp⁺ colonies were replica-plated to histidine omission medium to determine the histidine phenotype. Plasmid integration was measured by determining the stability of the Trp⁺ and His⁺ phenotypes; Trp⁺ colonies were serially replicated to three yeast extract-peptone-dextrose plates to dilute out extrachromosomally replicating plasmid, then replicated back to synthetic dextrose-Trp and synthetic dextrose-His plates (26).

Plasmid Rescue-- Plasmids were transferred from yeast to E. coli by a modification of the method of Strathern and Higgins (27). Yeast cultures were grown 2 days in 2 ml of yeast extract-peptone-dextrose medium at 30 °C. The cells were harvested and resuspended in 0.1 ml of lysis buffer (2.5 M LiCl, 50 mM Tris-HCl, pH 8.0, 20 mM EDTA, and 4% Triton X-100). An equal volume of phenol/chloroform/isoamyl alcohol plus one-third volume of acid-washed glass beads (0.45-0.50 mm) were added, and the mixture was vortexed vigorously for 10 min. The mixture was incubated at 65 °C for 5 min and then centrifuged; the aqueous phase was recovered and re-extracted with phenol/chloroform/isoamyl alcohol. The preparation was purified over 0.5 ml of Wizard miniprep resin (Promega) according to manufacturer directions using column wash buffer consisting of 55% ethanol, 200 mM NaCl, 20 mM Tris-HCl, and 5 mM EDTA, pH 7.5. The DNA was eluted from the resin in 50 µl of TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 8.0). Competent cells prepared by the method of Inoue et al. (28) were transformed with 10 µl of the DNA preparation, and minipreps were isolated from bacterial cells by alkaline lysis (24). All plasmid analyses were done on duplicate colonies from the E. coli transformation step.

Plasmid DNA was digested with XbaI to monitor gene conversion between plasmid and chromosome his3X alleles. All the plasmids used contain an invariant 3046-bp XbaI fragment. The remaining 3061-bp fragment is intact in plasmids with a wild-type HIS3 gene and is cleaved into two fragments of 1228 and 1833 bp in phis3-75X, 1096 and 1965 bp in phis3-207X, 999 and 2062 bp in phis3-304X, and 678 and 2383 bp in phis3-622X. Mutations at the damage target site were detected by BsiWI digestion. The plasmids that were not cut by BsiWI were subjected to DNA sequencing.

Southern Analysis-- Southern analysis of samples with integrated plasmids was performed as described previously (29). Genomic DNA was digested with EcoRI to distinguish single and multiple plasmid integrations and with XbaI to characterize gene conversion. The his3-75X, his3-207X, his3-304X, and his3-622X alleles generated XbaI digestion fragments of 1804, 1672, 1575, and 1257 bp, respectively, from the downstream flanking chromosomal XbaI site; in addition, the integrated plasmids produced the same XbaI fragments as the extrachromosomal plasmids as described above.

Gene Conversion Patterns-- A combination of genetic and physical analysis was used to construct gene conversion spectra. Colonies were determined to carry extrachromosomal or integrated plasmids by genetic characterization of Trp⁺ stability, and the two categories were analyzed separately. Direct assignments to some classes were made for colonies with unambiguous phenotypes, i.e. His⁺ or His^± extrachromosomal plasmids. Physical analysis by Southern hybridization or plasmid rescue was carried out on samples with ambiguous phenotypes, i.e. integrated plasmids and His extrachromosomal plasmids. The frequency of each conversion class, determined by physical analysis, was applied to the total fraction of colonies with a given phenotype to calculate the final conversion spectrum.

Polymerase Chain Reaction-- Total genomic DNA was isolated according to Sherman et al. (30). The primers used to amplify the HIS3 gene were 5'-TCCACCTAGCGGATGACTCT-3' and 5'-CACTTGCCACCTATCACCAC-3'. Polymerase chain reaction was carried out for 25 cycles of 1 min at 94 °C and 2 min at 60 °C in 2.5 mM MgCl₂, 0.25 µM of each primer, and 0.2 mM each dATP, dCTP, dGTP, and dTTP. The amplified DNA was digested with BsiWI to detect targeted mutations.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Repair of Damaged Plasmid DNA in Yeast Cells-- Plasmid DNA molecules carrying single site-specifically placed lesions were prepared in vitro and introduced into yeast cells for in vivo repair. The lesion site was the unique BsiWI restriction site within the HIS3 gene at position 416. This site was chosen because it contains a preferred 5'-TA-3' sequence for psoralen photoaddition. DNA repair substrates with targeted psoralen adducts were prepared by annealing 14-base oligonucleotides modified with 4'-hydroxymethyl-4,5',8-trimethylpsoralen furan-side monoadducts to single-stranded circular DNA. Irradiation with long wave UV light converted the monoadducts to cross-links. Primer extension of the cross-linked primer produced double-stranded circular plasmid molecules with cross-links at the BsiWI site. The plasmids with targeted psoralen monoadducts were formed similarly except the UV irradiation step was omitted. DSBs were produced by BsiWI digestion.

The modified plasmid molecules were transfected into yeast cells, and the transformation efficiency relative to that of undamaged plasmid was used as a measure of repair. The plasmids contain, in addition to the his3 target gene, a copy of TRP1; we followed transformation as the appearance of Trp⁺ colonies. The repair efficiencies for DSBs, psoralen monoadducts, and psoralen cross-links are presented in Table I. Single psoralen cross-links were repaired less efficiently, by a factor of 2-3, than either single psoralen monoadducts or DSBs placed at the same site, confirming that these complex double-strand lesions are more difficult for yeast cells to repair than single-strand lesions or clean breaks.

Damage-induced Gene Conversion-- The major pathway for repair of double-strand DNA breaks in yeast is through homologous recombination (31). Psoralen interstrand cross-links are processed to DSBs by the nucleotide excision repair pathway (3, 4) and also stimulate homologous recombination (25, 32). Psoralen monoadducts, in contrast, are processed by nucleotide excision repair to form single-strand breaks and are less efficient in stimulating recombination (33). We compared the abilities of double-strand breaks and both forms of psoralen photodamage to stimulate crossing over and gene conversion (31).

Gene conversion, defined as the transfer of genetic information from a donor DNA molecule to a recipient, is a consequence of recombinational repair. Repair replication, initiated from an invading strand of the damaged DNA molecule on the undamaged molecule, can form a segment of asymmetric heteroduplex DNA. Branch migration of the Holliday junction joining the two molecules may further extend the region of heteroduplex DNA symmetrically on both damaged and undamaged DNA. Mismatch repair processes mismatches within the heteroduplex DNA to generate conversions.

Genetic exchanges between damaged plasmid and undamaged chromosomal DNA molecules were followed by analyzing the retention of markers placed at distances of 100-300 bp from the damage site within the HIS3 gene. In all experiments the damage consisting of a double-strand break, psoralen monoadduct, or psoralen cross-link,was located at the BsiWI site at position 416 of the plasmid HIS3 gene. Genetic markers consisting of XbaI linker insertions were placed at positions 75, 207, 304, or 622, producing the his3 alleles his3-75X, his3-207X, his3-304X, and his3-622X, respectively. Each his3 allele contained only one XbaI linker insertion mutation; we thus were able to detect recombination events between pairs of his3 alleles that led to changes in the His phenotype.

We were able to measure gene conversion at two positions in each experiment, corresponding to the sites of the XbaI markers in the plasmid and chromosome HIS3 alleles. Some of the conversion events gave rise to phenotypic changes that could be detected by genetic analysis. The plasmid repair substrate carrying a marker at position 622 of the HIS3 gene was most informative in this respect (Fig. 2). The coding region of the HIS3 gene is only 663 nucleotides in length, and the insertion at position 622 places the mutation near the C terminus of the gene product. Cells with the multicopy plasmid phis3-622X show a slow growth phenotype in histidine omission medium, indicating that there is residual enzymatic activity in these mutants; we denote this phenotype by His^±. Cells with the corresponding chromosomal his3-622X allele in single copy do not grow in the absence of histidine, suggesting that overexpression of the mutant his3-622X gene is necessary for the residual growth that we observed.

View larger version (12K): [in this window] [in a new window]   Fig. 2.   Gene conversion of extrachromosomally replicating plasmid phis3-622X. The plasmid HIS3 gene (thin line) has a mutation at the 3' end; the gene product has partial activity, leading to slow growth on histidine omission medium (His±). The chromosomal HIS3 locus (thick line) has a mutation in the 5' half of the gene at position 75, 207, or 304; all these alleles are phenotypically His. Recombinational repair can generate gene conversion tracts (arrows); conversion tracts are assumed to initiate at the damage site (//). The plasmid DNA molecule is shown as the recipient. A, no conversion or short conversion tracts that reach neither marker. The phenotype remains His±. B, upstream conversion. The plasmid acquires an additional marker and becomes His. C, downstream conversion. The plasmid receives wild-type sequences at the 3' end of the HIS3 gene and becomes His+. D, bidirectional conversion, transferring the entire chromosomal allele to the plasmid. The phenotype is His.

Fig. 2 illustrates the possible outcomes of gene conversion in extrachromosomal phis3-622X. Conversion tracts are assumed to initiate at the damage site at position 416 and to extend continuously in one or both directions from this point, and the damaged plasmid is assumed to be the recipient (chromosome-to-plasmid conversion). No conversion, or short tracts that do not reach either marker, will produce an unchanged phis3-622X with a His^± phenotype (Fig. 2A). Unidirectional conversion in the upstream direction transfers the chromosome marker to the plasmid, generating a double mutant plasmid with a His phenotype (Fig. 2B). Unidirectional conversion in the downstream direction replaces the plasmid marker with the corresponding wild-type sequence, generating HIS3 with a His⁺ phenotype (Fig. 2C). Finally, bidirectional gene conversion replaces the plasmid his3-622X allele with the chromosomal allele, which has a His phenotype (Fig. 2D). This outcome cannot be distinguished from the product of upstream conversion by its phenotype, because both are His.

Asymmetries in Conversion Tracts-- The observation of phenotypic changes depends on the length of the conversion tract; the conversion frequency is expected to be higher close to the damage site if most tracts are short. For extrachromosomal plasmids, His colonies can be produced by upstream or bidirectional conversion tracts that transfer the chromosomal marker to the plasmid (Fig. 2, B and D); markers closer to the initiation site should be transferred at higher frequencies than more distant markers. The strain with a marker at position 304, about 100 bp away from the damage site, had a slightly higher frequency of cross-link-induced His colonies than the strains with markers at positions 75 or 207 (Fig. 3), but the difference was not significant (² = 3.3, p > 0.05). Psoralen monoadducts did not induce conversion to His⁺ or His. DSB-induced conversion showed a similar slight dependence on distance from the damage site, suggesting that the conversion gradient is shallow over this range, and the average tract length is more than 300 nucleotides. However, there was a striking difference in the type of phenotypic change induced by DSBs and cross-links. Cross-links were more efficient than DSBs in the induction of His colonies, but this pattern was reversed for the induction of His⁺ colonies, with a higher incidence seen for DSBs than for cross-links (data not shown). Downstream conversion seemed preferentially induced by DSBs, whereas cross-links induced upstream and/or bidirectional conversion. We carried out physical analysis of the repaired extrachromosomal plasmids to distinguish between these possibilities.

View larger version (12K): [in this window] [in a new window]   Fig. 3.   Damage-induced conversion of extrachromosomal plasmids to His phenotype. Yeast strains with XbaI insertion mutations at the positions shown were transformed with cross-linked (filled squares) or DSB-containing (open circles) plasmid phis3-622X, and the His phenotype of transformed colonies with extrachromosomally replicating plasmids was scored.

The direction of gene conversion tracts was characterized further by determining the XbaI restriction patterns of repaired extrachromosomal plasmids. Fig. 4A shows the conversion tract distribution for phis3-622X repair in cells with the his3-75X, his3-207X, or his3-304X chromosomal alleles. Most gene conversion tracts were unidirectional, although both cross-links and DSBs induced bidirectional tracts, at similar frequencies (Table II). The direction of gene conversion was markedly different for cross-links and DSBs. Cross-links induced nearly twice as many upstream as downstream conversions, whereas DSBs preferentially induced downstream conversions. Although in most cases the XbaI digestion patterns found corresponded to those predicted by the phenotype, some of the His cross-linked samples showed no evidence of gene conversion; some of these had rearrangements (insertions or deletions), and others proved to have mutations at the damage site as discussed below. Repair experiments with the opposite configuration of markers showed the same asymmetry in damage-induced gene conversion tracts (Fig. 4B). Plasmid phis3-75X, repaired in his3-622X cells, also underwent upstream conversion preferentially with psoralen cross-links and downstream conversion with DSBs.

View larger version (20K): [in this window] [in a new window]   Fig. 4.   Asymmetric gene conversion tracts induced by cross-links and DSBs. Left, a schematic diagram of damaged plasmid (thin line) and undamaged chromosome (thick line) with the indicated XbaI linker insertion sites is shown on top. The XbaI sites in repaired plasmids are shown below along with conversion tracts (arrows). Right, the percentage of extrachromosomal plasmids with indicated XbaI sites induced by cross-links (open bars) or DSBs (filled bars). A, plasmid phis3-622X repaired in his3-75X, his3-207X, and his3-304X yeast cells. The His phenotype was determined for 162 cross-link-damaged and 280 DSB-damaged colonies; 38 cross-link and 42 DSB His colonies underwent physical analysis. B, plasmid phis3-75X repaired in his3-622X yeast cells. The His phenotype was determined for 146 cross-link-damaged and 131 DSB-damaged colonies; 31 cross-link and 29 DSB colonies underwent physical analysis.

Discontinuous Conversion Tracts-- Gene conversion tracts are usually continuous; that is, adjacent markers are generally converted using the same strand of the heteroduplex as the donor. However, a minor fraction of tracts are discontinuous (34). We were able to detect discontinuous conversion tracts in experiments with plasmid and chromosome markers on the same side of the damage site. When plasmid phis3-75X is repaired in his3-304X, only discontinuous tracts using the plasmid allele at position 304 but chromosomal DNA as the donor at position 75 will generate the wild-type HIS3 allele on the plasmid, assuming that the conversion tracts initiate at the damage site (Fig. 5). Conversion to His⁺ was seen in 5% of the repaired extrachromosomal plasmids.

View larger version (11K): [in this window] [in a new window]   Fig. 5.   Discontinuous gene conversion tracts. Plasmid phis3-75X repaired in his3-304X yeast cells.

With both markers on the same side of the damage site, we were able to characterize gene conversion tract lengths in more detail. Although the majority of the conversion events extended at least 300 bp from the damage site, the detected tract length was between 100 and 300 bp in 40% of the conversion tracts in cross-linked samples and 30% of the DSB-containing samples. We attempted to further examine conversion tracts in the region of 200-300 bp from the damage site using plasmid phis3-75X repaired in strain his3-207X. However, overall gene conversion levels were very low with this combination of markers; only 4% of cross-linked and 14% of DSB-containing plasmids showed evidence of conversion (data not shown). This combination of plasmid and chromosome alleles had heterologies only 132 bp apart compared with a distance of over 200 bp between positions 75 and 304. Closely spaced mismatches have been reported to inhibit conversion and decrease conversion tract length; it has been proposed that the mismatch repair system disrupts heteroduplexes containing high levels of mismatches (heteroduplex DNA rejection) (35). We therefore examined DSB-induced conversion of plasmid phis3-207X in strain his3-304X; this pair of markers is also ~100 bp apart but is closer to the damage site. The level of gene conversion in this cross was 25%, a value similar to that in the his3-75X-his3-304X cross. Close spacing between mismatches alone does not inhibit damage-induced gene conversion in this system.

Damage-induced Plasmid Integration-- We assayed plasmid integration as a measure of crossing over. The plasmids used lack centromeres and segregate unevenly; plasmid integration leads to stable Trp⁺ colonies. The levels of damage-induced plasmid integration are shown in Table III. Both DSBs and psoralen cross-links induced integration at mean frequencies of 40-50%. Although DSBs were slightly more efficient on average than psoralen cross-links at stimulating integration, the difference was not significant. In contrast to these two forms of double-strand damage, psoralen monoadducts were quite inefficient at inducing integration.

We characterized gene conversion patterns in integrated plasmids by Southern hybridization, scoring changes in XbaI markers as for extrachromosomal plasmids. In these samples we were able to analyze conversion events affecting the chromosomal allele as well as the plasmid. In a simple crossover without conversion, one copy each of the plasmid and chromosome markers is retained. In most conversion events the chromosomal allele was the donor, leading to gain of the chromosome marker, loss of the plasmid marker, or both depending on the direction and length of the conversion tract (Fig. 6). There was, in addition, a minor class of events in which gain of the plasmid marker or loss of the chromosome marker indicated that the plasmid allele was the donor (plasmid-to-chromosome conversion). Similar levels of plasmid-to-chromosome conversion were induced by DSBs and cross-links. The observation of plasmid-to-chromosome conversion events confirms that the damaged plasmid and undamaged chromosome participate in reciprocal homologous recombination. We also observed more complex products including deletions and rearrangements among the integrated plasmids; the frequency of these events was similar for DSBs and cross-links. These products were not characterized further.

View larger version (20K): [in this window] [in a new window]   Fig. 6.   Gene conversion tracts in integrated plasmids. Left, a schematic diagram of the starting plasmid and chromosome alleles (top) and the integrated plasmids (below) with XbaI sites as indicated. The XbaI sites flanking or between the HIS3 genes are not shown. Right, the percentage of integrants with the indicated XbaI pattern. open bars, cross-links; filled bars, DSBs. A, the plasmid phis3-622X repaired in his3-75X, his3-207X, and his3-304X yeast cells. Southern analysis was performed on 90 cross-link-damaged and 176 DSB-damaged colonies B, the plasmid phis3-75X repaired in his3-622X yeast cells. Southern analysis was performed on 27 cross-link-damaged and 46 DSB-damaged colonies.

As for noncrossover recombination, although the overall levels of conversion were similar for DSBs and cross-links, there was a pronounced disparity in the directions of the conversion tracts, with more upstream conversion for cross-links and more downstream conversion for DSBs. This pattern is shown for both phis3-622X repaired in strains with the XbaI marker on the upstream side of the damage site (Fig. 6A) and for plasmid and chromosomal alleles with the opposite arrangement of markers (Fig. 6B). Although a majority of the conversion tracts scored were unidirectional, bidirectional conversion tracts made up a higher proportion of total conversion in crossover than in noncrossover events (Table II); the frequency was similar for cross-links and DSBs.

The overall levels of gene conversion were higher for integrated than for extrachromosomal plasmids (Table IV). The difference was more pronounced for cross-links than for DSBs, with only half as much gene conversion in the nonintegrated plasmids. Although DSBs induced significantly more conversion than cross-links in extrachromosomal plasmids, there was no difference between the two forms of damage in integrated plasmids. The difference in conversion levels between DSBs and cross-links for extrachromosomal plasmids suggests that cross-links may be repaired by an additional nonrecombinational pathway that does not lead to gene conversion or crossing over. We examined repaired plasmids for evidence of error-prone repair.

Psoralen Cross-link-induced Mutations-- We screened repaired extrachromosomal plasmids for mutations at the damage site by digestion with BsiWI; resistance to cleavage by BsiWI indicates an alteration in the recognition sequence. About 9% of the extrachromosomal plasmids carrying psoralen cross-links were found to have mutations at the BsiWI site (Table V). In addition, of the 183 plasmids analyzed, three had insertions and two had deletions of several hundred base pairs. In contrast to psoralen cross-links, none of the plasmids carrying DSBs had mutations at the damage site or rearrangements. Thus, although both DSBs and psoralen cross-links are repaired by homologous recombination, only cross-links are, in addition, acted upon by a mutagenic repair pathway.

None of the cross-linked plasmids we analyzed had undergone simultaneous mutation at the damage site and gene conversion at the marker positions. This suggests that recombination and mutation are produced by alternate repair pathways acting on psoralen cross-links rather than proceeding from a common repair intermediate. We looked for targeted mutations in integrated plasmids, all of which had presumably undergone recombinational repair. None of these samples had mutations at the BsiWI site, confirming that there was no overlap between the repair products with evidence of mutation and recombination.

The sequence changes are shown in Fig. 7. Because psoralen reacts with pyrimidines, the mutations are presented as being derived from the thymine residues at the damage site. Of the 16 mutants sequenced, there were 10 T·A  C·G transitions, 5 T·A  G·C transversions, and one T·A  A·T transversion. One of the mutations was a double transversion involving the position immediately upstream of the central 5'-TpA site; this is also a potential cross-link site between a thymine on one strand and a cytosine on the opposite strand.

View larger version (11K): [in this window] [in a new window]   Fig. 7.   Psoralen cross-link-induced mutations. Mutations are shown as resulting from changes at thymidine residues. The box indicates a double mutation at adjacent residues.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

We have characterized the repair products of site-specifically placed psoralen cross-links and compared the end points with those of psoralen monoadducts and double-strand breaks. Psoralen cross-links induce gene conversion and crossing over at frequencies similar to DSBs, confirming homologous recombination as the major repair pathway. Cross-links also induce targeted mutations, implicating error-prone post-replication repair as an alternate pathway for cross-link repair. A model for cross-link repair is shown in Fig. 8.

View larger version (13K): [in this window] [in a new window]   Fig. 8.   Model for the repair of psoralen cross-links.

Cross-link- and DSB-induced Recombination-- Psoralen cross-linking induces the NER-dependent formation of double-strand breaks in S. cerevisiae. In this organism, the major pathway of DSB repair is homologous recombination. A cross-link repair model based on these two observations predicts that cross-links that enter the NER-homologous recombination pathway should produce end products similar to those induced by restriction endonuclease-generated DSBs. Psoralen monoadducts, which do not generate DSBs, should not be recombinogenic.

We tested this prediction by characterizing gene conversion and reciprocal exchange induced by single DSBs, psoralen cross-links, or psoralen monoadducts, formed at the same site. Psoralen monoadducts, as expected, did not induce recombination, whereas DSBs and psoralen cross-links were highly recombinogenic. In most respects the predictions from the model were fulfilled: the levels of crossing over, total gene conversion frequencies of integrated plasmids, conversion tract lengths, the fraction of bidirectional conversion tracts, incidence of discontinuous conversion tracts, and proportion of plasmid-to-chromosome conversion events were similar for DSBs and cross-links. However, DSB- and cross-link-induced recombination differed significantly in two aspects: 1) DSBs and cross-links induced opposite asymmetries in conversion-tract direction, and 2) cross-links induced lower levels of gene conversion than DSBs in extrachromosomally replicating plasmids.

Psoralen Cross-links Reverse the DSB Conversion Tract Asymmetry-- Both DSBs and psoralen cross-links induced asymmetric gene conversion tracts, but in opposite directions. DSBs generated more downstream (3') conversion in the HIS3 gene, whereas cross-link-induced conversion was biased in the upstream (5') direction. Asymmetries in gene conversion have been reported previously in other systems. DSB-induced plasmid-chromosome recombination in Ustilago maydis produced more frequent upstream conversion in the LEU1 and PYR6 genes (36). Studies of DSB-induced conversion within the S. cerevisiae URA3 gene found a bias toward upstream conversion tracts in both allelic and plasmid-chromosome recombination (34, 37, 38), although this bias was reversed in URA3 substrates with reduced 5'-side homology.

The observed asymmetry in gene conversion may result from bias at several stages in recombination (31). The initiating lesion may have inherent asymmetry or may be processed to an asymmetric intermediate. The initiating DSB is processed by exonuclease activity to expose single-stranded tails with 3' ends; the invasion of a homologous duplex sequence initiates formation of a junction between the two homologues and generates heteroduplex regions. Junction migration can further extend or decrease the length of heteroduplex. Mismatch repair produces conversion by acting on mismatches within the heteroduplex region; the mismatch repair system may also regulate the extent of heteroduplex by limiting junction migration within heterologous regions. Conversion tract asymmetries may be generated at one or more of these steps by sequence-specific biases, by chromatin structure, or by directional processes such as replication or transcription.

Psoralen interstrand cross-links reverse the conversion tract bias seen with DSBs. Because the two lesions were placed at the same position, it is likely that differences between the initiating lesions or early repair intermediates govern the differences in the direction of gene conversion tracts and, therefore, that the recombinogenic intermediates of cross-link repair differ from clean DSBs such as those produced by restriction endonucleases. It should be noted here that although all the DSBs in this study were produced by BsiWI, which generates 5' overhangs, a similar preponderance of downstream conversion in HIS3 was also induced by restriction endonucleases that generate 3' overhangs or blunt ends (39).

DSBs have been detected as intermediates of psoralen cross-link repair in yeast by centrifugation through neutral sucrose gradients (3, 4) and pulsed field gel electrophoresis (18, 40); formation of these DSBs depends on NER genes, but the structure of these intermediates has not been characterized. Studies in mammalian systems have shown that NER generally acts on single-strand DNA damage, including psoralen monoadducts, to produce incisions on both the 5' and 3' sides of the lesion; the 5' incision is 22-24 bp from the damage site and the 3' incision is ~5 bp away from the lesion, producing an excised fragment of 27-29 nucleotides (41). In this system NER also produced dual incisions at psoralen interstrand cross-links, but both incisions were formed on the 5' side of the cross-link rather than flanking it (42). This processing produced a single-strand gap of 22-28 nucleotides but left the cross-link in place. Although incision did occur on both faces of the psoralen cross-link, the pyrone side was incised more frequently than the furan side (43). Asymmetric repair intermediates like these could introduce a bias into the initiation of conversion tracts, because only one free 3' end would be available for invasion into a homologous duplex.

A very different incision pattern was observed in a system using a human chromatin-associated DNA endonuclease complex (44, 45). In this system there were dual incisions on both the 5' and 3' sides of the psoralen cross-link instead of being placed to one side, and the furan side was preferentially incised. This pattern is similar to psoralen cross-link repair in E. coli (46, 47). It is not clear how to resolve these contradictory observations, but Kumaresan et al. (44) suggested that multiple pathways of cross-link incision may function in mammalian cells. Although it is not known whether either incision pattern might apply to yeast cells, a bias in the initial processing of psoralen interstrand cross-links may serve to reverse the normal preference for DSB-induced downstream conversion within the HIS3 gene.

Psoralen Cross-links Are Repaired by an Alternate Error-prone Pathway-- Although the gene conversion frequencies of integrated plasmids were similar for both forms of damage, DSBs induced significantly higher levels of conversion among extrachromosomally replicating plasmids than did psoralen cross-links. A comparison of extrachromosomal and integrated plasmids reveals that for both DSB- and cross-link-induced repair, conversion levels were higher among integrated plasmids. The difference was small, but still significant, for DSBs and more pronounced for cross-links.

One possible interpretation is that the same recombinational repair pathway acting on damaged plasmids yields both crossover (integrated) and noncrossover (extrachromosomal) repair products, but the level of detectable gene conversion is higher among crossover events. Gene conversion tracts associated with crossing over have been found previously to be longer than noncrossover conversion tracts (48, 49). In this system, however, tract lengths do not seem limiting for markers at the distances we have scored.

Another interpretation is that a separate repair pathway not yielding gene conversion or crossing over produces a portion of the extrachromosomal plasmids. For psoralen cross-links this second pathway would seem to be error-prone DNA repair. A substantial fraction of the extrachromosomal plasmids had targeted mutations at the cross-link site, indicating that a distinct alternate pathway is acting on this form of damage. We found no repaired plasmids with evidence of both recombination and mutation, indicating that the recombinational and error-prone cross-link repair pathways do not overlap. This observation suggests that a commitment to either recombinational or mutagenic repair is made early in the repair process. Of the two, recombinational repair seems to be the major pathway, judging from the relative levels of the end products of the two pathways. In this study ~43% of cross-link repair was accompanied by crossing over; among the 57% of repair products that remained extrachromosomal, 40% had conversions (23% of the total), yielding a total of ~66% with some evidence of recombination. In contrast, only ~5% of the total (9% of the extrachromosomal plasmids) had mutations. Both figures are minimal estimates, because some plasmids were repaired without detectable mutation or recombinational end points.

DSBs also may be repaired by alternate pathways. Although we saw no evidence of mutagenic DSB repair, DSBs are also subject to repair by nonhomologous end joining, which does not yield conversions or crossovers (31). Nonhomologous end joining is the major pathway of DSB repair in mammalian cells but is a minor pathway in yeast cells with functioning homologous recombination. The lower level of DSB-induced gene conversion seen among the extrachromosomal than among the integrated plasmids suggests that DSBs, similar to psoralen cross-links, are subject to repair by a second minor pathway. Further support for the involvement of an additional pathway comes from the observation that DSB repair of these plasmids is still substantial in yeast cells deficient in the homologous recombination genes RAD51, RAD54, RAD55, and RAD57 (39).

Psoralen cross-links that enter the error-prone pathway are highly mutagenic. We can estimate a lower limit for the mutation frequency from the fraction of extrachromosomal plasmids with mutations (9%) and the fraction that lacked gene conversions and therefore may have gone through error-prone repair (60%). Therefore, at least 15% of the cross-links processed by error-prone repair yielded mutations; this fraction may actually be higher, because some of the plasmids without gene conversions may have been repaired by recombination. The strong mutagenic potential of psoralen cross-links is in contrast to psoralen monoadducts, which did not induce mutations. Psoralen monoadducts may be repaired by error-free mechanisms such as NER or damage avoidance by strand switching; alternatively, the damaged plasmid DNA strand may be lost, leaving only the undamaged strand to be replicated (50).

Psoralen cross-links induced base substitutions, primarily TA to CG transitions, at the two modified thymidine residues. In studies of psoralen cross-links targeted to a specific site by triplex-forming oligonucleotides, Barre et al. (51, 52) found the major classes of mutations in yeast to be transversions and insertions. However, a psoralen cross-link targeted by a triplex-forming oligonucleotide but subsequently cleaved from it produced a mutation spectrum similar to that reported here.

Removal of DNA lesions by NER is coupled to transcription (53), and the transcribed strand is repaired more quickly than the nontranscribed strand (54). The result is a higher mutation frequency for the nontranscribed strand, because error-prone repair acts on the remaining lesions. We have found that the majority of the psoralen cross-link-induced mutations were in the nontranscribed strand, in agreement with previous reports (55-57).

Multiple Repair Pathways for Psoralen Interstrand Cross-links-- The model depicts entry into error-prone repair both before and after nucleotide excision repair processing of cross-links. Barre et al. (51, 52) found NER-dependent and -independent components of mutagenesis, with different mutation spectra, induced by psoralen cross-links in yeast cells. We have also observed cross-link-induced mutagenesis in NER-deficient yeast strains, indicating that cross-links are able to enter an error-prone repair pathway independently of NER.² It is possible that the NER-processed intermediates that enter error-prone repair are not DSBs. The higher incidence of mutations on the nontranscribed strand suggests that incision may occur on the transcribed strand only; this would leave a gap that is filled in by trans-lesion synthesis by an error-prone DNA polymerase such as Pol (58). There may be an additional pathway for interstrand cross-link removal distinct from NER. Nitrogen mustard and cisplatin cross-links induce DSBs in yeast and mammalian cells; this DSB formation does not depend on NER or on base excision repair (7, 17).

In summary, at least three DNA repair pathways participate in the cellular response to psoralen interstrand cross-links. These complex lesions are inefficiently repaired in comparison with single-strand lesions such as psoralen monoadducts or less complex double-strand lesions such as DSBs. Although the major pathway is error-free, a substantial minority of psoralen cross-links are repaired by an error-prone mechanism.

	ACKNOWLEDGEMENTS

We thank Rodney Rothstein for providing yeast strains and plasmids, Donna Luisi and Louis Roccanova for plasmid and strain construction, and Eduardo Sanz Navares for assistance with the plasmid rescue analyses.

	FOOTNOTES

^* This work was supported in part by grants from the City University of New York PSC-CUNY Research Award Program (to W. A. S.), the National Institutes of Health Grants CA42377 (to W. A. S.) and GM47945 (to J. E. H.), and the Office of Basic Energy Sciences in the Biological Energy division of the Department of Energy under contract number DE-AC03-76SF00098 (to J. E. H.).The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^¶ To whom correspondence should be addressed: Dept. of Chemistry and Biochemistry, Queens College, City University of New York, 65-30 Kissena Blvd., Flushing, NY 11367. Tel.: 718-997-4195; Fax: 718-997-5531; E-mail: Wilma_Saffran@qc.edu.

Published, JBC Papers in Press, June 4, 2001, DOI 10.1074/jbc.M103588200

² S. Thomas and W. Saffran, unpublished observations.

	ABBREVIATIONS

The abbreviations used are: NER, nucleotide excision repair; DSB, double-strand breaks; HPLC, high pressure liquid chromatography; bp, base pair(s).

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

This article has been cited by other articles:

				A. Pamidi, R. Cardoso, A. Hakem, E. Matysiak-Zablocki, A. Poonepalli, L. Tamblyn, B. Perez-Ordonez, M. P. Hande, O. Sanchez, and R. Hakem Functional Interplay of p53 and Mus81 in DNA Damage Responses and Cancer Cancer Res., September 15, 2007; 67(18): 8527 - 8535. [Abstract] [Full Text] [PDF]

				X. Shen, S. Jun, L. E. O'Neal, E. Sonoda, M. Bemark, J. E. Sale, and L. Li REV3 and REV1 Play Major Roles in Recombination-independent Repair of DNA Interstrand Cross-links Mediated by Monoubiquitinated Proliferating Cell Nuclear Antigen (PCNA) J. Biol. Chem., May 19, 2006; 281(20): 13869 - 13872. [Abstract] [Full Text] [PDF]