Curr Genet (1999) 35: 499–505

© Springer-Verlag 1999

O R I G I N A L PA P E R

Yingying Yang · Xiaolin Kang · Lester Kohalmi Ramachandran Karthikeyan · Bernard A. Kunz

Strand interruptions confer strand preference during intracellular correction of a plasmid-borne mismatch in Saccharomyces cerevisiae

Received: 4 June 1998 / 18 January 1999

Abstract Site-directed mutagenesis was used to construct yeast centromere plasmids in which a strand nick or gap could be placed 5′ or 3′, on either strand, to a reporter gene (SUP4-o) carrying defined base mismatches. The plasmids were then transformed into yeast cells and the direction and efficiency of mismatch repair were assayed by scoring colouring of the transformant colonies. Strands that were nicked were consistently corrected more often than intact strands, but the effect was very small. However, placement of a small gap at the same positions as the nicks resulted in a marked increase in selection for the gapped strand and an enhanced efficiency of mismatch repair. Both the preference for the gapped strand and correction of the mismatch were offset by deletion of the mismatch repair gene PMS1. Together, the results suggest that strand interruptions can direct intracellular mismatch correction of plasmid-borne base mispairs in yeast. Key words Heteroduplex repair · Strand discrimination · Strand interruptions · Saccharomyces cerevisiae

Y. Yang Department of Pathology, Vancouver Hospital at the Health Sciences Centre, 855 West 12th Avenue, Vancouver, British Columbia, V5Z 1M9, Canada L. Kohalmi Department of Plant Sciences, University of Western Ontario, 1151 Richmond Street North, London, Ontario, N6A 5B7, Canada X. Kang Ligand Pharmaceuticals, 10275 Science Center Drive, San Diego, California 92121, USA R. Karthikeyan · B. A. Kunz (½) School of Biological and Chemical Sciences, Deakin University, Geelong, Victoria 3216, Australia e-mail: [email protected] Fax: +61-3-5227-3330 Communicated by S. W. Liebman

Introduction

Accurate DNA replication plays a critical role in maintaining the genetic integrity of an organism. High levels of replication fidelity are achieved via the sequential operation of three important processes. First, selectivity of base insertion by DNA polymerases ensures that most often the correct nucleotide is placed opposite a template base resulting in the preferred Watson-Crick base pair (Echols and Goodman 1991; Kunkel 1992). Second, exonucleolytic proofreading by DNA polymerases removes incorrectly inserted nucleotides from the end of a growing chain (Echols and Goodman 1991; Kunkel 1992; Goodman et al. 1993). Whether or not replication errors that evade proofreading give rise to mutations depends on the efficiency of the third mechanism, post-replicative mismatch repair (MMR) (Modrich and Lahue 1996; Modrich 1997). This process extracts mispaired nucleotides that were incorporated by extension of the chain. Clearly, the effectiveness of MMR must rely on recognition of the mispair (different mispairs may be recognised with different efficiencies) and identification of the strand containing the incorrect base. In Escherichia coli, mismatches are recognised by a homodimer of the MutS protein (Modrich and Lahue 1996). Strand discrimination is provided by a delay in post-replicative methylation of 5′-GATC-3′ sequences in the nascent strand (Modrich and Lahue 1996). Repair is then initiated by mismatch-dependent incision of the daughter strand at the hemi-methylated 5′-GATC-3′ sequence. It is, therefore, the nick that ultimately directs repair to the nascent strand in this organism. Genetic and in vitro evidence suggests that eukaryotic MutS homologs are responsible for mismatch recognition in the yeast Saccharomyces cerevisiae and human cells (Modrich 1997). Furthermore, studies of MMR by HeLa and Drosophila melanogaster cell extracts indicate that in vitro heteroduplex repair is biased to the strand containing an interruption (Holmes et al. 1990; Thomas et al. 1991). Whether strand breaks can direct correction of a mismatch within eukaryotic cells is uncertain. Repair was found to

500

favour the nicked strand of a simian virus 40 shuttle vector following transfection into mammalian cells (Hare and Taylor 1985), but contradictory results were subsequently reported (Heywood and Burke 1990). In addition, the systems used did not allow potential effects on the directionality of MMR to be determined in the cells in which repair actually occurred. In the present study, we have assessed the ability of strand interruptions to direct repair of base mismatches on heteroduplex plasmids transformed into yeast cells. Our results suggest that nicks can bias correction to the strand containing the break, but the effect is obscured possibly by rapid ligation of the nick upon entry of the plasmid into a cell. A strand gap on either side of a defined mismatch, however, can markedly increase repair of the gapped plasmid strand and enhance the efficiency of correcting the mismatch on the vector.

Fig. 1 Structure of YCpLK1E and YCpLK3EB. The 1.0-kb BamHI-HindIII and 1.1-kb SalI-HindIII fragments that carry SUP4-o in YCpLK1E and YCpLK3EB, respectively, are not drawn to scale. The direction of SUP4-o transcription is marked by the arrow

Materials and methods

YCpLK3EB-E were constructed by inserting the same fragments into the single BamHI or EcoRI site of YCpLK3EB.

Yeast strains. Construction of the S. cerevisiae strain MKP-o (MATα, can1-100, ade2-1, lys2-1, ura3-52, leu2-3,112, his3-∆200, trp1-∆901) was described previously (Pierce et al. 1987). RKpms1, an isogenic, MMR-deficient derivative of MKP-o deleted for PMS1, was constructed by transforming MKP-o with BstXI-linearised pWBK4 (Kramer et al. 1989 b) to target integration of pWBK4 to the BstXI site 3′ to PMS1 via homologous recombination (Rothstein 1991). Integration at this site leaves the URA3 gene flanked 5′ by the PMS1 locus and 3′ by a duplication of the 1.3-kb KpnI-MluI fragment immediately 5′ to PMS1. Ura+ transformants were selected and integration of pWBK4 as described was confirmed by DNA hybridisation analysis. Ura+ isolates were propagated in uracil-containing medium and plated on appropriately supplemented minimal medium containing 5-fluoro-orotic acid (Boeke et al. 1984) to select for simultaneous loss of PMS1 and URA3 via crossing-over between the repeated 1.3-kb KpnI-MluI fragments. Loss of URA3 and PMS1 in an appropriate isolate (RKpms1) was confirmed by DNA hybridisation analysis. The MMR deficiency of this strain was confirmed phenotypically by assessing heteroduplex correction as described in the Results. Plasmids. Assembly of the yeast centromere plasmids YCpMP2 and YCpJA1 was described previously (Pierce et al. 1987; Armstrong and Kunz 1995). These plasmids carry an ochre suppressor allele (SUP4-o) of a yeast transfer RNA gene and components that allow autonomous replication and selection in yeast (ARS1, URA3) and bacterial (pBR322 origin, ampicillin resistance) cells. In addition, they have the M13 phage origin of DNA replication, and so can be induced to form single-stranded plasmid DNA (Vieira and Messing 1987), and they bear a yeast centromere sequence (CEN4). YCpLK1E (Fig. 1) was derived from YCpMP2 by using site-directed mutagenesis (Kohalmi and Kunz 1992) to eliminate the EcoRI site immediately adjacent to the HindIII site in YCpMP2. YCpLK3EB (Fig. 1) was derived from YCpJA1 by eliminating the same EcoRI site as for YCpLK1E and removing a BamHI site and an EcoRI site in linker sequences flanking the yeast chromosomal DNA fragment carrying SUP4-o. The SUP4-o sequence features a G · C pair and an A · T pair at sites 32 and 35, respectively (Pierce et al. 1987). Derivatives (YCpLK1E 32T, YCpLK1E 35G, YCpLK3EB 32T, YCpLK3EB 35G) of YCpLK1E and YCpLK3EB carrying mutated sup4-o alleles with A · T or G · C base pairs at sites 32 and 35, respectively, were constructed via site-directed mutagenesis. Subsequently, YCpLK1E 32T-B and YCpLK1E 32T-E were constructed by inserting the 24-bp BamHI or the 42-bp EcoRI pUC7 (Messing et al. 1981) linker fragment into the single BamHI or EcoRI site, respectively, of YCpLK1E 32T. Similarly, YCpLK3EB-B and

Media. Media used for growth of yeast strains and scoring of mismatch correction were described previously (Pierce et al. 1987; Kunz et al. 1991). Plasmid DNA and heteroduplex preparation, and transformation. Single- and double-stranded circular plasmid DNAs were prepared, and yeast cells transformed as described (Kunz et al. 1991). Heteroduplex plasmid DNA was prepared from linearised double-stranded DNA and circular single-stranded DNA (ssDNA) (see Fig. 2) using a thermal denaturation/renaturation method (Kunz et al. 1991; Kang et al. 1992). The preparations were (detectably) free of circular ssDNA and were made free of double-stranded linear DNA by treatment with Micrococcus luteus endonuclease V (U.S. Biochemical) (Holmes et al. 1990). Statistical analysis. Differences in some parameters were assessed with chi-square contingency tests employing Yates’ correction for continuity (Sokal and Rohlf 1969). Values of P < 0.05 were considered significant.

Results and discussion

Placement of strand nicks or gaps on heteroduplex molecules To assess the effects of strand interruptions on MMR, we employed yeast centromere plasmids carrying defined base mismatches in the ochre suppressor tRNA allele SUP4-o. Nicked heteroduplex molecules were constructed by annealing circular ssDNA (derived from YCpLK1E or YCpLK3EB or their derivatives having site-specific base-pair substitutions in SUP4-o at site 32 or 35) to complementary DNA strands obtained from linearised double-stranded molecules (prepared from the derivatives or parental plasmids as appropriate) (Fig. 2). The circular ssDNA strand encompasses the nontranscribed strand of SUP4-o for YCpLK1E but the transcribed strand of SUP4-o for YCpLK3EB. The linearised plasmids were obtained by digestion with BamHI, EcoRI,

501 Fig. 2 Construction of heteroduplex plasmids. The diagram shows the preparation of heteroduplex molecules having a nick at the XhoI recognition sequence

or XhoI. For each of these enzymes there is a unique recognition sequence on the plasmids. BamHI is situated 119 bp or 122 bp 5′ to SUP4-o site 32 or 35, respectively, whereas EcoRI is 116 bp or 113 bp 3′ to site 32 or 35, respectively. XhoI is located approximately 4 kb in either direction from SUP4-o (Fig. 1). Thus, by digestion with the appropriate restriction enzyme, the nick can be positioned 5′ or 3′ to SUP4-o, or at the opposite end of the plasmid from SUP4-o. Furthermore, by annealing circular and linear DNA strands prepared from different combinations of plasmids, the nick can be placed on the plasmid DNA strand that encompasses the transcribed or nontranscribed strand of SUP4-o. Similarly, by using plasmids containing the 24-bp BamHI and 48-bp EcoRI pUC7 linkers inserted at the BamHI or EcoRI sites, we can position gaps 5′ or 3′ to SUP4-o on the transcribed or nontranscribed strands (Fig. 3). Simply put, if the doublestranded gene is viewed as a rectangle, the approach de-

scribed allowed us to construct heteroduplex plasmids having a nick or gap in any one of the four corners of the rectangle. In order to place the nicks or gaps on the transcribed strand, SUP4-o must be orientated within the resulting heteroduplex molecule as shown for plasmid YCpLK1E in Fig. 1. However, it must be in the other orientation (as shown for YCpLK3EB in Fig. 1) to place the nicks or gaps on the nontranscribed strand. This results in two orientations for each mismatch used (Table 1). In one orientation, the correct base is on the nicked or gapped strand. The other orientation features the incorrect base on the nicked or gapped strand. Note, however, that a specific mismatch at a particular site within SUP4-o resides in the same sequence context regardless of SUP4-o orientation.

502 Fig. 3 Construction of gapped heteroduplex plasmids. The diagram shows the assembly of heteroduplex molecules having a gap at the EcoRI or BamHI recognition sequence 3′ or 5′ to SUP4-o on the plasmid strand encompassing the transcribed strand of SUP4-o. Similarly, by using YCpLK3EB-B and YCpLK3EB-E to generate the circular ssDNA, and YCpLK3EB 32T digested with BamHI or EcoRI to produce the linear duplex DNA, the gaps can be positioned at the EcoRI or BamHI recognition sequence 3′ or 5′ to SUP4-o on the plasmid strand encompassing the nontranscribed strand of SUP4-o

Assay for heteroduplex repair To assess intracellular correction of plasmid-borne mismatches, the heteroduplex molecules are transformed into yeast cells. Prior to the first episode of plasmid DNA replication in the transformed cells, the mismatch can be repaired to generate a functional or non-functional SUP4-o allele, depending on the direction of correction. Alternatively, it might not be repaired. The yeast strains used in this study each carry the ochre ade2-1 mutation which causes red colouring. However, the phenotypic effects of ade2-1 can be suppressed by SUP4-o so that cells carrying ade2-1 and a functional SUP4-o allele form white colonies. Since yeast centromere plasmids are maintained in haploid cells predominantly as single copies (Newlon 1988), three types of transformant colony can emerge on medium selective for the plasmid: (1) white (mismatch restored to normal base-pair); (2) red (mismatch converted to incorrect base-pair); (3) sectored red/white (mismatch

not repaired). The sectored colony appears in the absence of MMR because one of the two daughter cells produced after the first round of DNA replication following transformation receives a plasmid with a functional SUP4-o allele while the other acquires a plasmid with a defective copy. Thus, alterations in the direction or efficiency of MMR should modify the red : white colony ratio or the fraction of sectored colonies, respectively. Previously, we demonstrated that colony colouring is associated with mismatch resolution and less than 1% of the sectored transformants are due to cotransformation (Kunz et al. 1991). Effects of strand nicks on heteroduplex repair Nicked heteroduplex plasmids carrying C-A or T-G mispairs at sites 32 or 35 in SUP4-o were transformed into MKP-o, and the numbers of red, white or sectored colonies that emerged were tabulated. The results were then

503 Table 1 Heteroduplex DNA molecules constructed for this study Mismatch and site a

DNA sequence flanking the mismatch b

SUP4-o allele c on strands

C-A: 32

5′-TAA ATT 5′-GCG CGC 5′-TAA ATT 5′-GCG CGC 5′-CAA GTT 5′-AAT TTA 5′-CAA GTT 5′-AAT TTA

+ – – + – + + – – + + – + – – +

A-C: 32 T-G: 32 G-T: 32 C-A: 35 A-C: 35 T-G: 35 G-T: 35

AGT TCA CAA GTT AGT TCA CAA GTT GAC CTG TAA AAT GAC CTG TAA AAT

C A A C T G G T C A A C T G G T

TTG AAC ACT TGA TTG AAC ACT TGA TTA AAT GTC CAG TTA AAT GTC CAG

CGC-3′ GCG TTA-3′ AAT CGC-3′ GCG TTA-3′ AAT ATT-3′ TAA TTG-3′ AAC ATT-3′ TAA TTG-3′ AAC

Table 2 Influence of strand nicks on heteroduplex correction Mismatch and site a

Nick location b

Allele Total on colonies nicked scored d strand c

Ratio of red colonies to white colonies e

Per cent sectored colonies e

C-A: 32

X-T 5′-T 3′-T X-NT 5′-NT 3′-NT X-T 5′-T 3′-T X-NT 5′-NT 3′-NT X-T 5′-T 3′-T X-NT 5′-NT 3′-NT X-NT 5′-NT 3′-NT X-T 5′-T 3′-T

+ + + – – – – – – + + + – – – + + + + + + – – –

0.68 1.27 *** 1.15 *** 0.62 0.57 0.60 0.58 0.41 *** 0.46 0.16 0.37 *** 0.32 *** 1.09 1.41 *** 1.18 1.46 1.10 *** 1.37 1.50 1.76 ** 1.68 1.83 1.34 *** 1.62 *

35 28 29 25 23 23 10 10 11 5 4 5 25 23 17 *** 21 22 23 9 9 9 12 11 16

A-C: 32

T-G: 32

G-T: 32

C-A: 35

A-C: 35

a

The first base is on the nicked strand, the second base is on the continuous strand. Site 1 corresponds to the first base of the tRNA b Each mismatch (in bold) is present in two orientations with respect to the nicked plasmid strand. The orientations depend on the plasmids used to construct the heteroduplexes (see text). The upper strand of each duplex is nicked c +: a functional SUP4-o allele; –: an inactive sup4-o allele

T-G: 35

G-T: 35

2839 1067 1560 3233 3243 3417 1413 1748 1723 2901 1090 1519 1909 2982 2165 2802 1657 2243 3461 3248 1352 3774 3890 2325

a

compared for plasmids carrying a specific mismatch and a nick at the BamHI or EcoRI sites flanking SUP4-o, or at the XhoI sequence approximately 4 kb from the mismatch positions (Table 2). If a strand break can direct repair to the interrupted strand, then the ratio of red : white colonies should increase if the nick is on the strand carrying the functional SUP4-o allele but decrease if it is on the strand containing the defective sup4-o allele. With respect to plasmids nicked at the XhoI site, 65–95% of the mismatches were repaired as determined by the proportion of non-sectored colonies. However, T-G mispairs were consistently corrected more often than C-A mismatches. Since the immediate sequence context for different mispairs at the same site in SUP4-o is identical (Table 1), differences in the repair of such mismatches most likely reflect differences in mismatch recognition. Consequently, the data argue that the efficiency of MMR was mismatch-dependent (also see below, Table 3). The ratio of red to white colonies did vary somewhat from mismatch to mismatch, but not in a manner consistent with the nature of the allele on the nicked strand. For example, with a functional SUP4-o allele on the nicked strand the ratio was 0.68 for a C-A mismatch at site 32 but 1.50 for a T-G mismatch at site 35. Furthermore, the average of the red : white ratios was 0.95 for a functional SUP4-o allele on the nicked strand and 1.03 for a defective sup4-o allele on the interrupted strand. Consequently, we detected no general bias in favour of the nicked strand when the break was at the XhoI site. This observation supports the results of an earlier investigation of heteroduplex correction in transformed yeast cells (Kramer et al. 1989 a). In that study,

The first base is on the nicked strand, the second base is on the continuous strand X: nicked at the XhoI site approximately 4 kb in either direction from the site of the mismatch; T/NT: the strand encompassing the transcribed (T) or nontranscribed (NT) segment of SUP4-o; 5′: nicked 119 bp (site 32) or 122 bp (site 35) 5′ to the mismatch; 3′: nicked 116 bp (site 32) or 113 bp (site 35) 3′ to the mismatch c +: a functional SUP4-o allele; –: an inactive sup4-o allele d The values are the totals for 2–4 independent transformations e Asterisks indicate statistically significant differences. *: P < 0.05; **: P < 0.005; ***: P < 0.001 b

a nick 3.5-kb from the mismatch site did not bias correction to the nicked strand. It may be that breaks ≥3.5-kb from the mismatch simply are too distant to serve as strand discrimination signals, a possibility consistent with a reported in vivo mismatch repair tract length of approximately 1 kb (Bishop and Kolodner 1986). Relative to the fractions of sectored colonies for a nick at the XhoI site, proportionally fewer such colonies were recovered in some instances where the break was at the BamHI or EcoRI sequence, within 125 bp of the mismatches in SUP4-o. Although this suggests that the presence of a nearby strand interruption might have slightly enhanced the efficiency of MMR, the decrease in the fraction of sectored colonies was significant in only one case (C-A mismatch at site 35 with a 3′ nick on the transcribed strand). On the other hand, for all of the mismatches tested, there did appear to be a minor preference for correction of the nicked strand when the break was at the BamHI or EcoRI sequence. If a functional SUP4-o allele was on the interrupted strand, the red : white ratio increased, whereas it decreased if a defective sup4-o allele was on the nicked

504 Table 3 Influence of strand gaps on correction of a mismatch at site 32 Strain

Mismatch

Nick/ gap location a

Allele Total on colonies nicked scored strand

Ratio of red colonies to white colonies

Per cent sectored colonies

MKP-o

C-A

MKP-o

A-C

RKpms1

C-A

X-T 5′-T 3′-T X-NT 5′-NT 3′-NT X-T 5′-T 3′-T

+ + + – – – + + +

0.69 7.75 3.57 0.62 0.32 0.24 0.71 0.91 1.13

41 11 24 36 31 28 69 68 72

2216 2743 3164 3805 3243 1283 3110 3204 3491

a

Plasmids were nicked at the XhoI site or had gaps located 122-bp 5′ to the mismatch or 113-bp 3′ to the mismatch. The remainder of the legend is as for Table 2

strand. Although the effects were very small, they were significant in 10/16 instances. In addition, the magnitudes of the ratio-changes varied somewhat for C-A and T-G mismatches that were at the same position (and so were within the same sequence context), were nicked on the same side of the mismatch, and had the same type of SUP4-o allele on the nicked strand. The apparent mismatch dependency of the variations suggests, therefore, that the ratio changes may have been due to repair provoked by the mismatch (also see below). One explanation for the minor effects of the strand nicks might be that the nick is quickly ligated when the heteroduplex enters the cell. This might leave only a small fraction of the intracellular population of heteroduplex molecules with a nick at the time the mismatch is repaired. To assess this possibility, linearised plasmid molecules were treated with alkaline phosphatase prior to heteroduplex formation to remove the 5′ terminal phosphates and so inhibit ligation. For a C-A mismatch at site 32 (wildtype SUP4-o allele on the nicked strand), the red : white ratio increased from 1.27 to 1.4 for a 5′ nick and from 1.15 to 1.25 for a 3′ nick (data not shown). Thus, the effect of the nick appeared to be marginally enhanced by alkaline phosphatase treatment. It may be that even nicks lacking a 5′ terminal phosphate can be efficiently sealed within a cell. Strand gaps influence heteroduplex repair To eliminate the possibility of rapid sealing of the nick, four heteroduplex plasmids having a C-A mismatch at site 32 and gaps on the transcribed or nontranscribed strands 5′ or 3′ to the mismatch were constructed (Fig. 3). Analysis of the colonies that emerged following yeast transformation with these plasmids revealed that the strand containing the gap was repaired preferentially (by as much as 11-fold) and that the presence of the gap was associated

with an increase in the efficiency of mismatch correction (Table 3). For a gap located on the transcribed strand 5′ to the mismatch, the percentages of total repair events (indicated by the fraction of nonsectored colonies) and repair events involving this strand (indicated by the proportion of red colonies) increased from 59 to 89% and 24 (535/2216) to 79% (2162/2743), respectively. The magnitude of these effects varied with the location of the gap but there was no clear correlation. For example, the gaps that provoked the most pronounced effects were located 5′ to the mismatch on the transcribed strand but 3′ to the mismatch on the nontranscribed strand (both gaps were generated by removal of the 42-bp EcoRI pUC7 linker). Elimination of the MMR gene PMS1 dramatically reduced the preference for the gapped strand and the correction efficiency for the mismatch tested (Table 3). These results argue that the effects of the gaps were dependent on MMR. Despite the MMR deficiency, however, the red : white ratio remained slightly greater than that for the heteroduplex nicked at the XhoI site and, approximately 70% of the colonies recovered were sectored. This fraction of uncorrected mismatches is close to the average value (76%) reported previously for heteroduplex repair in a strain having PMS1 eliminated (Kramer et al. 1989 a). It is not possible that partial activity of the Pms1 protein was responsible for the apparent correction of some mismatches in either study since PMS1 was completely deleted. Recently, evidence was presented that the MMR protein Msh2 interacts with yeast nucleotide excision repair (NER) proteins including Rad1 and Rad3, and msh2 mutations increase the UV-sensitivity and are epistatic to the mutator phenotypes of NER-deficient strains (Bertrand et al. 1998). This suggests that Msh2 and NER proteins might function in common processes. However, rendering NER inoperative by deletion of RAD1, or the presence of a point mutation in RAD3, does not alter the directionality or decrease the efficiency of heteroduplex correction in MKP-o (Kang and Kunz 1992; Yang et al. 1996). Thus, NER does not appear to play a significant role in the correction of plasmid-borne mismatches. Perhaps the non-sectored colonies that emerged in the MMR-deficient mutant were due to random strand loss (Wagner and Meselson 1976) prior to or during replication of the heteroduplex plasmids, or to mismatch-independent processing (e.g., nick translation). Nonetheless, our findings indicate that the effects of the strand gaps on the directionality of correction are mediated predominantly through MMR. Collectively, our data are consistent with strand interruptions targeting mismatch correction during intracellular repair of heteroduplex plasmids. We caution, however, that neither these results nor data for nick-directed heteroduplex correction in cell extracts (Holmes et al. 1990; Thomas et al. 1991) can be readily extrapolated to the repair of mismatches formed during the replication of eukaryotic chromosomes. Recent findings indicate that proliferating cell nuclear antigen, which is required for DNA replication, can interact with yeast mismatch repair proteins (Johnson et al. 1996; Umar et al. 1996). Thus, mismatch correction on chromosomes may be linked to DNA

505

replication. The systems so far used to study the influence of strand interruptions on mismatch repair require that correction of the mismatch on a heteroduplex substrate occur prior to replication of the heteroduplex molecule. Acknowledgments This work was supported by the Australian Research Council and the Natural Sciences and Engineering Research Council of Canada.

References Armstrong JD, Kunz BA (1995) Nucleotide excision repair and gene orientation modulate the strand specificity of ultraviolet mutagenesis in yeast. Environ Mol Mutagenesis 25: 12–22 Bertrand P, Tshkoff DX, Filosi N, Dasgupta R, Kolodner RD (1998) Physical interaction between components of DNA mismatch repair and nucleotide excision repair. Proc Natl Acad Sci USA 95: 14278–14283 Bishop DK, Kolodner RD (1986) Repair of heteroduplex plasmid DNA after transformation into Saccharomyces cerevisiae. Mol Cell Biol 6: 3401–3409 Boeke JD, LaCroute F, Fink GR (1984) A positive selection for mutants lacking orotidine-5′-phosphate decarboxylase activity in yeast: 5-fluoro-orotic acid resistance. Mol Gen Genet 197: 345– 346 Echols H, Goodman MF (1991) Fidelity mechanisms in DNA replication. Annu Rev Biochem 60: 477–511 Goodman MF, Creighton S, Bloom LB, Petruska J (1993) Biochemical basis of DNA replication fidelity, CRC Crit Rev Biochem Mol Biol 28: 83–126 Hare JT, Taylor JH (1985) One role for DNA methylation in vertebrate cells is strand discrimination in mismatch repair. Proc Natl Acad Sci USA 82: 7350–7354 Heywood LA, Burke JF (1990) Repair of single nucleotide DNA mismatches transfected into mammalian cells can occur by shortpatch excision. Mutat Res 236: 59–66 Holmes J Jr, Clark S, Modrich P (1990) Strand-specific mismatch correction in nuclear extracst of human and Drosophila melanogaster cell lines. Proc Natl Acad Sci USA 87: 5837–5841 Johnson RE, Kovali G, Guzder SN, Amin NS, Holm C, Habraken, Sung P, Prakash, Prakash S (1996) Evidence for involvement of yeast proliferating cell nuclear antigen in DNA mismatch repair. J Biol Chem 271: 27987–27990 Kang X, Kunz BA (1992) Inactivation of the RAD1 excision repair gene does not affect correction of mismatches on heteroduplex plasmid DNA in yeast. Curr Genet 21: 261–263 Kang X, Yadao F, Gietz RD, Kunz BA (1992) Elimination of the yeast RAD6 ubiquitin conjugase enhances base-pair transitions and G · C → T · A transversions as well as transposition of the Ty element: implications for the control of spontaneous mutation. Genetics 130: 285–194

Kohalmi L, Kunz BA (1992) In vitro mutagenesis of the yeast SUP4o gene to identify all substitutions that can be detected in vivo with the SUP4-o system. Environ Mol Mutagenesis 19: 282–287 Kramer, B, Kramer W, Williamson MS, Fogel S (1989 a) Heteroduplex DNA correction in Saccharomyces cerevisiae is mismatch specific and requires functional PMS genes. Mol Cell Biol 9: 4432–4440 Kramer W, Kramer B, Williamson MS, Fogel S (1989 b) Cloning and nucleotide sequence of DNA mismatch repair gene PMS1 from Saccharomyces cerevisiae: homology of PMS1 to procaryotic MutL and HexB. J Bacteriol 171: 5339–5346 Kunkel TA (1992) DNA replication fidelity. J Biol Chem 267: 18251– 18254 Kunz BA, Kang X, Kohalmi L (1991) The yeast rad18 mutator specifically increases G · C → T · A transversions without reducing correction of G-A or C-T mismatches to G · C pairs. Mol Cell Biol 11: 218–225 Messing J, Crea R, Seeburg PH (1981) A system for shotgun sequencing. Nucleic Acids Res 9: 309–321 Modrich P (1997) Strand-specific mismatch repair in mammalian cells. J Biol Chem 272: 24727–24730 Modrich P, Lahue R (1996) Mismatch repair in replication fidelity, genetic recombination, and cancer biology. Annu Rev Biochem 65: 101–133 Newlon CS (1988) Yeast chromosome replication and segregation. Microbiol Rev 52: 568–601 Pierce MK, Giroux CN, Kunz BA (1987) Development of a yeast system to assay mutational specificity. Mutat Res 182: 65–74 Rothstein R (1991) Targeting disruption, replacement and allele rescue: integrative DNA transformation in yeast. Methods Enzymol 194: 281–301 Sokal RR, Rohlf FJ (1969) Biometry. WH Freeman, San Francisco, California Thomas DC, Roberts JD, Kunkel TA (1991) Heteroduplex repair in extracts of human HeLa cells. J Biol Chem 266: 3744–3751 Umar A, Buermeyer AB, Simon JA, Thomas DC, Clark AB, Liskay RM, Kunkel TA (1996) Requirement for PCNA in DNA mismatch repair at a step preceding DNA resynthesis. Cell 87: 65–73 Vieira J, Messing J (1987) Production of single-stranded plasmid DNA. Methods Enzymol 153: 3–11 Wagner R Jr, Meselson M (1976) Repair tracts in mismatched DNA heteroduplexes. Proc Natl Acad Sci USA 71: 1544–1548 Yang Y, Johnson AJ, Johnston LH, Siede W, Friedberg EC, Ramachandran K, Kunz BA (1996) A mutation in a Saccharomyces cerevisiae gene (RAD3) required for nucleotide excision repair and transcription increases the efficiency of mismatch correction. Genetics 144: 459–466