Role of Proliferating Cell Nuclear Antigen Interactions in the Mismatch Repair-Dependent Processing of Mitotic and Meiotic Recombination Intermediates in Yeast

Abstract

The mismatch repair (MMR) system is critical not only for the repair of DNA replication errors, but also for the regulation of mitotic and meiotic recombination processes. In a manner analogous to its ability to remove replication errors, the MMR system can remove mismatches in heteroduplex recombination intermediates to generate gene conversion events. Alternatively, such mismatches can trigger an MMR-dependent antirecombination activity that blocks the completion of recombination, thereby limiting interactions between diverged sequences. In Saccharomyces cerevisiae, the MMR proteins Msh3, Msh6, and Mlh1 interact with proliferating cell nuclear antigen (PCNA), and mutations that disrupt these interactions result in a mutator phenotype. In addition, some mutations in the PCNA-encoding POL30 gene increase mutation rates in an MMR-dependent manner. In the current study, pol30, mlh1, and msh6 mutants were used to examine whether MMR–PCNA interactions are similarly important during mitotic and meiotic recombination. We find that MMR–PCNA interactions are important for repairing mismatches formed during meiotic recombination, but play only a relatively minor role in regulating the fidelity of mitotic recombination.

THE failure to accurately replicate and repair genomic DNA leads to a wide variety of somatic and germ-line mutations, most of which are deleterious. Organisms thus have evolved multiple mechanisms to promote the stability of DNA and ensure faithful genome propagation. One of these mechanisms is the mismatch repair (MMR) system, best known for its role in correcting errors made by DNA polymerases during DNA replication (reviewed by Harfe and Jinks-Robertson 2000; Schofield and Hsieh 2003; Kunkel and Erie 2005). In addition to this replication-associated “spellchecker” function, the MMR system promotes genome stability via an antirecombination activity that prevents recombination between diverged sequences (reviewed by Surtees et al. 2004). Because such sequences are not identical, there is the potential for mismatches to be present within heteroduplex recombination intermediates. The MMR system recognizes such mismatches and prevents recombination from going to completion, thereby limiting detrimental genome rearrangements between dispersed repeated sequences. In some cases, a single mismatch within heteroduplex DNA is sufficient to reduce mitotic recombination in an MMR-dependent manner (Datta et al. 1997).

In addition to the spellchecker and antirecombination functions that promote mitotic genome stability, the MMR system is important during meiosis, specifically in meiotic recombination (reviewed by Hoffmann and Borts 2004; Surtees et al. 2004). As in mitosis, sequence divergence can trigger meiotic antirecombination activity of the MMR machinery, an activity thought to be important for enforcing homolog–homolog interactions and species barriers. Mismatches formed in meiotic recombination intermediates between different alleles, however, are more often simply repaired by the MMR system. Depending on which strand is used as the template for repair, either Mendelian segregation of the alleles will be restored or a gene conversion event will occur. In the nomenclature of eight-spored asci, restoration-type repair is manifested as 4:4 events while gene conversion results in 6:2 or 2:6 allele segregation. If MMR fails to correct a mismatch, the resulting meiotic product will have both alleles, resulting in a 5:3 or 3:5 ratio, referred to as postmeiotic segregation (PMS). Loss of MMR function typically results in an increase in PMS events and a concomitant reduction in gene conversion. Finally, in addition to mismatch correction, certain MMR proteins have also been shown to be involved in the processing of meiotic recombination intermediates to produce crossovers (Hunter and Borts 1997; Wang et al. 1999).

The mechanism of MMR associated with DNA replication is best understood in prokaryotes, where it involves the proteins MutS, MutL, and MutH (reviewed by Iyer et al. 2006; Joseph et al. 2006). MutS homodimers recognize errors made by DNA polymerase, binding to both base substitution and frameshift intermediates. MutH homodimers bind to nearby hemimethylated Dam sites, which mark the region as newly replicated and provide a mechanism for distinguishing the template and nascent strands. MutL homodimers promote an interaction between MutS and MutH, which activates MutH to nick the unmethylated, nascent strand. This nick provides an entry point for a helicase to unwind and exonucleases to remove the mismatch-containing region. Thus, MMR is nick directed, and importantly, repair is targeted specifically to the nascent strand.

In eukaryotes, the bacterial MutS homodimer is replaced by Msh (MutS homolog) protein heterodimers (reviewed by Kunkel and Erie 2005). MutSα is composed of Msh2 and Msh6 and broadly recognizes base/base mismatches and small-loop frameshift intermediates. MutSβ, which is a heterodimer of Msh2 and Msh3, recognizes small- and large-loop frameshift intermediates and may recognize specific base/base mismatches (Harrington and Kolodner 2007). MutSα and MutSβ interact primarily with the MutL-like heterodimer MutLα (composed of Mlh1 and Pms1) to correct mismatches. No MutH homolog has been found in eukaryotes, and the method of strand discrimination during the repair of replication errors is not fully understood. As in Escherichia coli, however, experiments using purified proteins or cell extracts have shown that a nick is sufficient to direct eukaryotic MMR (reviewed by Iyer et al. 2006). In addition to the nicks that are naturally present in nascent DNA, recent studies have demonstrated that human MutLα has the ability to introduce additional nicks into DNA in vitro (Kadyrov et al. 2006) and that disruption of the yeast MutLα endonuclease domain results in a mutator phenotype in vivo (Erdeniz et al. 2007; Kadyrov et al. 2007).

In addition to the classical MMR proteins, the proliferating cell nuclear antigen (PCNA) sliding clamp is important for the repair of replication errors. During replication, the PCNA homotrimer encircles duplex DNA and tethers the replication machinery to the template, thereby increasing DNA polymerase processivity. PCNA also acts as a landing pad for many other proteins involved in DNA metabolism (reviewed by Moldovan et al. 2007). The first indication of a role for PCNA in MMR was obtained in yeast, where some alleles of the PCNA-encoding POL30 gene produce a mutator phenotype and cause instability of microsatellite repeats (Johnson et al. 1996; Umar et al. 1996). PCNA has since been shown to interact directly with the MutS homologs Msh3 and Msh6 (Umar et al. 1996; Flores-Rozas et al. 2000), as well as with the MutL homolog Mlh1 (Umar et al. 1996; Dzantiev et al. 2004; Lee and Alani 2006). Each of these three proteins contains a putative PCNA interaction domain called the PIP box (Warbrick 1998) which, when mutated, disrupts interaction with PCNA in vitro and confers a mutator phenotype in vivo (Kokoska et al. 1999; Clark et al. 2000; Flores-Rozas et al. 2000; Lee and Alani 2006). Additionally, in vitro experiments have shown that the inclusion of PCNA enhances the ability of MutSα to discriminate between DNAs containing mismatches and those that lack mismatches (Flores-Rozas et al. 2000). Finally, work done with human cell extracts has shown that a nick positioned 5′ of a mismatch requires only hMutSα, RPA, and hExo1 to direct subsequent strand removal, but that a nick 3′ to a mismatch additionally requires MutLα, PCNA, and replication factor C (RFC) for strand removal (Constantin et al. 2005).

While PCNA is clearly important in the MMR-directed repair of replication errors, it is not known if PCNA plays an equivalent role in the recombination-related functions of MMR. In spite of the basic similarities between mismatch recognition during replication and recombination, there are data suggesting fundamental differences between how the MMR system removes replication errors and how it affects recombination processes in Saccharomyces cerevisiae. For example, it appears that the MutL homologs, while essential for the repair of replication-generated mismatches, are not as important as the MutS homologs in some antirecombination processes (Chen and Jinks-Robertson 1999; Nicholson et al. 2000; Sugawara et al. 2004). In addition, separation-of-function alleles of PMS1 have been identified that reduce the mitotic antirecombination activity of MutLα and the repair of mismatches formed during meiotic recombination, but have little or no effect on the repair of replication errors (Welz-Voegele et al. 2002). Mutant alleles of MLH1 have also been identified that are deficient in the repair of mismatches formed during either replication or meiotic recombination, but not both (Argueso et al. 2002, 2003). Finally, a member of the RecQ family of DNA helicases, Sgs1, has been shown to be involved in the mitotic antirecombination function of MMR, but plays no known role in the spellchecker or meiotic recombination functions (Myung et al. 2001; Spell and Jinks-Robertson 2004b; Sugawara et al. 2004). Given these basic differences, it is important to determine if the mitotic role of PCNA is limited only to the spellchecker function of MMR, or if it is also required for the antirecombination activity. Similarly, it is important to determine if the role of PCNA in MMR is confined to mitotic MMR processes, or if it is necessary for the repair of mismatches formed during meiotic recombination as well. In this study, we examine the effect of altering the MMR–PCNA interaction on MMR functions during both mitotic and meiotic recombination in yeast. Our results demonstrate that PCNA is important for the repair of mismatches in meiotic recombination intermediates but plays only a minor role in regulating mitotic recombination fidelity.

MATERIALS AND METHODS

Media and growth conditions:

For strain constructions, cells were grown in YEP medium (2% yeast extract and 4% peptone) supplemented with 2% dextrose and 0.25 g/liter adenine (YPD; 2.5% agar for plates). Drug selections were done on YPD containing either 300 μg/ml hygromycin or 200 μg/ml G418. For mitotic recombination and mutation-rate determinations in the SJR328-derived strains with mutant chromosomal alleles, cultures were grown in YEP supplemented with 4% galactose/2% glycerol (YEPGG) and 0.25 g/liter adenine; mutation rates in AS4- and AS13-derived strains were measured in YPD-grown cultures. Selection for His⁺ recombinants was performed on synthetic medium supplemented with 2% galactose/2% glycerol/2% ethanol and a complete amino acid mix deficient in histidine (GGE −His). For mutation-rate analysis, cells were plated on arginine-deficient synthetic dextrose medium (SD −Arg) supplemented with 60 μg/ml canavanine. For SJR328-derived strains with plasmid-encoded msh6 alleles, cells were grown in synthetic medium supplemented with 4% galactose/2% glycerol and deficient in leucine (SGG −Leu). His⁺ and canavanine-resistant cells were selected as above except that leucine was omitted from the media. For the plasmid-containing strains, total cell numbers were determined on SD −Leu medium. For meiotic recombination studies, standard growth and sporulation conditions were used (Sherman 1991) except as noted.

Plasmids:

Plasmid pSR559 was constructed by subcloning a KpnI/NotI fragment containing the pol30-52 allele from pBL241-52 (Ayyagari et al. 1995) into KpnI/NotI-digested pRS306 (Sikorski and Hieter 1989). pSR873, which contains the pol30-201,204 allele, was constructed by site-directed mutagenesis (QuickChange, Stratagene, La Jolla, CA) using the pol30-201 plasmid pRDK925 (Lau et al. 2002) as a template and introducing the pol30-204 mutation using primers pol30-204Fmut (5′-CCTCACTAAGTAAAATCCTACGTCGTGGTAACAACACCGATACATTAACACTAATTGC-3′) and pol30-204Rmut (5′-GCAATTAGTGTTAATGTATCGGTGTTGTTACCACGACGTAGGATTTTACTTAGTGAGG -3′). The sequence change is underlined and creates an Hpy99I site.

Strain constructions:

A complete list of yeast strains is given in Table 1. The mitotic antirecombination experiments were done in the SJR328 background (MATα ade2-101 his3Δ200 ura3-Nhe lys2ΔRV∷hisG leu2-R) with the cβ2-100% or cβ2-4ns constructs integrated at LEU2 (Nicholson et al. 2000). The meiotic experiments were done using diploids obtained by mating appropriate derivatives of haploids AS4 (MATα arg4-17 trp1-1 tyr7-1 ade6 stp22 ura3; Stapleton and Petes 1991) and AS13 (MATa leu2-Bst ade6 ura3 rme1; Stapleton and Petes 1991). In preparation for the introduction of chromosomal pol30 or plasmid-encoded msh6 alleles, Leu⁻ derivatives of Leu⁺ strains were constructed by replacing LEU2 with a leu2Δ∷kanMX4 allele generated using pFA6-kanMX4 (Wach et al. 1994) as a template (forward primer 5′-ATGTCTGCCCCTAAGAAGATCGTCGTTTTGCCAGGTGACCACGTTGTCAAGcagctgaagcttcgtacg and reverse primer 5′-TTAAGCAAGGATTTTCTTAACTTCTTCGGCGACAGCATCACCGACTTCGGTGGaggccactagtggatctg; LEU2-homologous sequences are capitalized). The leu2 derivative of HMY134 (SJR2705) was constructed by transforming HMY134 with a PCR-generated leu2Δ∷hyg fragment. This fragment was produced using the same oligonucleotides described above to amplify hphMX4 from plasmid pAG32 (Goldstein and McCusker 1999).

TABLE 1.

Yeast strains

Strain	Relevant genotypea	Sourceb
SJR328-derived haploids
GCY313	cβ2-100% substrates	Nicholson et al. (2000)
GCY420	cβ2-100%, msh3Δ∷hisG msh6Δ∷hisG	Nicholson et al. (2000)
SJR1906	cβ2-100%, pol30-201:LEU2	This study
SJR2173	cβ2-100%, msh3Δ∷hisG msh6Δ∷hisG	This study
	pol30-201:LEU2
SJR1896	cβ2-100%, pol30-204:LEU2	This study
SJR2127	cβ2-100%, pol30-201,204:LEU2	This study
SJR2084	cβ2-100%, pol30-104:LEU2	This study
SJR2083	cβ2-100%, pol30-52	This study
SJR2573	cβ2-100%, mlh1Δ∷hygMX	This study
SJR2575	cβ2-100%, mlh1-QLF:kanMX	This study
SJR2690	cβ2-100%, msh6Δ∷hisG	SJR1856 transformed with pRS315
SJR2691	cβ2-100%, msh6Δ∷hisG/MSH6	SJR1856 transformed with pRDK3572
SJR2692	cβ2-100%, msh6Δ∷hisG/msh6Δ2-251	SJR1856 transformed with pRDK4650
SJR2693	cβ2-100%, msh6Δ∷hisG/msh6Δ51-251	SJR1856 transformed with pRDK4715
SJR2694	cβ2-100%, msh6Δ∷hisG/msh6-FFAA,Δ51-251	SJR1856 transformed with pRDK4758
GCY562/615	cβ2-4ns substrates	Nicholson et al. (2000)
GCY834	cβ2-4ns, msh3Δ∷hisG msh6Δ∷hisG	Nicholson et al. (2000)
SJR1889	cβ2-4ns, pol30-204:LEU2	This study
SJR1899	cβ2-4ns, pol30-201:LEU2	This study
SJR2174	cβ2-4ns, msh3Δ∷hisG msh6Δ∷hisG pol30-201:LEU2	This study
SJR2126	cβ2-4ns, pol30-201,204:LEU2	This study
SJR2082	cβ2-4ns, pol30-104:LEU2	This study
SJR2081	cβ2-4ns, pol30-52	This study
SJR2574	cβ2-4ns, mlh1Δ∷hygMX	This study
SJR2576	cβ2-4ns, mlh1-QLF:kanMX	This study
SJR2695	cβ2-4ns, msh6Δ∷hisG	SJR1850 transformed with pRS315
SJR2696	cβ2-4ns, msh6Δ∷hisG/MSH6	SJR1850 transformed with pRDK3572
SJR2697	cβ2-4ns, msh6Δ∷hisG/msh6Δ2-251	SJR1850 transformed with pRDK4650
SJR2698	cβ2-4ns, msh6Δ∷hisG/msh6Δ51-251	SJR1850 transformed with pRDK4715
SJR2699	cβ2-4ns, msh6Δ∷hisG/msh6-FFAA,Δ51-251	SJR1850 transformed with pRDK4758
AS4-derived haploids
RKY1721	msh2∷Tn10LUK7-7	Alani et al. (1994)
JSY203	msh2∷Tn10LUK7-7 ARG4	This study
JSY208	ARG4	This study
JSY173	pol30-52	This study
JSY209	pol30-52 ARG4	This study
SJR2183	leu2Δ∷kanMX	This study
SJR2202	leu2Δ∷kanMX pol30-201:LEU2	This study
JSY345	pol30-201:LEU2	Spore derived from SJR2202 × JSY125
JSY125c	MATa + pDJ173	Stone and Petes (2006)
HMY104	msh2Δ∷kanMX4	Kearney et al. (2001)
JSY218	MATamsh2Δ∷kanMX4 + pDJ173	Spore derived from JSY125 × HMY104
JSY352	msh2Δ∷kanMX4 pol30-201:LEU2	Spore derived from JSY218 × JSY345
SJR2577	mlh1-QLF:kanMX	This study
HMY134	msh6Δ∷kanMX4	Kearney et al. (2001)
SJR2705	msh6Δ∷kanMX4 leu2Δ∷hyg	This study
AS13-derived haploids
PD73	his4-AAG	Detloff et al. (1991)
PD98	his4-3133	Detloff et al. (1992)
RKY1452	his4-AAG msh2∷Tn10LUK7-7	Alani et al. (1994)
JSY162	his4-AAG pol30-52	This study
SJR2184	his4-AAG leu2Δ∷kanMX	This study
SJR2203	his4-AAG leu2Δ∷kanMX pol30-201:LEU2	This study
JSY222	his4-AAG msh2Δ∷kanMX pol30-52	Spore derived from JSY214 × JSY162
JSY332	his4-AAG leu2Δ∷kanMX pol30-201:LEU2 msh2Δ∷hygB	msh2Δ∷hygB derivative of SJR2203
HMY101	msh2Δ∷kanMX his4∷U1.1a	Kearney et al. (2001)
HMY131	MATα	Kearney et al. (2001)
JSY214	MATα his4-AAG msh2Δ∷kanMX	Spore derived from HMY101 × JSY127
JSY350	his4-AAG msh2Δ∷hygB pol30-201:LEU2	Spore derived from JSY332 × HMY131
JSY351	MATα pol30-201:LEU2	Spore derived from JSY332 × HMY131
JSY344	his4-AAG pol30-201:LEU2	Spore derived from JSY127 × SJR2203
JSY127	MATα his4-AAG	Stone and Petes (2006)
JSY355	his4-3133 pol30-201:LEU2	Spore derived from JSY351 X PD98
SJR2578	his4-AAG mlh1-QLF:kanMX	This study
JSY144	his4-AAG msh6Δ∷kanMX4	Stone and Petes (2006)
SJR2506	his4-AAG msh6Δ∷kanMX4/MSH6	JSY144 transformed with pRDK3572
SJR2507	his4-AAG msh6Δ∷kanMX4/msh6-FFAA,Δ51-251	JSY144 transformed with pRDK4758
Diploid strains
PD83	his4-AAG/HIS4	Detloff et al. (1991)
PD99	his4-3133/HIS4	Detloff et al. (1992)
MW103	his4-Sal/HIS4	Stone and Petes (2006)
JSY338	his4-51/his4-51 his4-AAG/HIS4	Stone and Petes (2006)
No. 5	his4-AAG/HIS4 msh2∷Tn10LUK7-7/ msh2∷Tn10LUK7-7	Alani et al. (1994)
JSY240	his4-Sal/HIS4 msh2Δ∷kanMX/msh2Δ∷kanMX	Stone and Petes (2006)
JSY230	his4-Sal/HIS4 mlh1Δ∷kanMX/mlh1Δ∷hygB	Stone and Petes (2006)
JSY343	his4-51/his4-51 his4-AAG/HIS4 msh2Δ∷kanMX/msh2Δ∷kanMX	Stone and Petes (2006)
JSY175	his4-AAG/HIS4 pol30-52/pol30-52	JSY173 × JSY162
JSY346	his4-AAG/HIS4 pol30-201:LEU2/pol30-201:LEU2	JSY345 × JSY344
JSY354	his4-AAG/HIS4 msh2Δ∷kanMX/mshΔ2∷hygB pol30-201:LEU2/pol30-201:LEU2	JSY352 × JSY350
JSY356	his4-3133/HIS4 pol30-201:LEU/pol30-201:LEU2	JSY345 × JSY355
HMY95	his4-AAG/HIS4 mlh1Δ∷kanMX4/mlh1Δ∷URA3	Welz-Voegele et al. (2002)
DB101	his4-AAG/HIS4 mlh1Δ∷URA3/mlh1Δ∷URA3	Welz-Voegele et al. (2002)
SJR2577 × SJR2578	his4-AAG/HIS4 mlh1-QLF:kanMX/mlh1-QLF:kanMX	This study
SJR2705 × SJR2706	HIS4/his4-AAG msh6Δ∷kanMX4/msh6Δ∷kanMX4 + plasmid-borne MSH6	This study
SJR2705 × SJR2707	HIS4/his4-AAG msh6Δ∷kanMX4/msh6Δ∷kanMX4 + plasmid-borne msh6-FFAA,Δ51-251	This study

^aAll strains were derived from haploid strains SJR328 (MATα ade2-101 his3Δ200 ura3ΔNhe lys2ΔRV∷hisG leu2-R; Nicholson et al. 2000), AS4 (MATα arg4-17 trp1-1 tyr7-1 stp22 ade6 ura3; Stapleton and Petes 1991), or AS13 (MATa leu2-Bst ade6 ura3 rme1; Stapleton and Petes 1991) by transformation or by crosses with isogenic strains. Only those markers that are different from the haploid progenitor strains are shown.

^bIn crosses, AS4- and AS13-derived strains are shown on the left and the right of the ×, respectively.

^cThe plasmid pDJ173 (URA3 STP22) allows diploid strains derived from AS4 to sporulate (Stone and Petes 2006).

Strains containing the chromosomal pol30-104, pol30-201, pol30-204, or pol30-201,204 allele were constructed by one-step allele replacement using a SacI fragment derived from the appropriate plasmid and containing both the pol30 allele and a LEU2 marker [pCH1577 (Johnson et al. 1996), pRDK925, pRDK926 (Lau et al. 2002), or pSR873, respectively]. pol30-52 strains were constructed via two-step allele replacement using either EcoRI-digested pSR559 or pBL241-52 (Ayyagari et al. 1995). The mlh1-QLF allele was introduced by transformation with an mlh1-QLF:kanMX4 fragment derived from KpnI-digested pEAA282 (Lee and Alani 2006), and the mlh1Δ∷hygMX2 allele was introduced using a PCR-derived fragment (forward primer 5′-TTTTGATACGATAGTGATAGTAAATGGAAGGTAAAAATAACATAGACCTATCAATAAGCAcagctgaagcttcgtacg and reverse primer 5′-CACAATCACACTCAGGAAATAAACAAAAAACTTTGGTATTACAGCCAAACGTTTTAAAGaggccactagtggatctg; MLH1-homologous sequences are capitalized). The presence of the mlh1-QLF and pol30 alleles was inferred by phenotype (elevated mutation rates and cold sensitivity as appropriate) and confirmed by sequencing. For complementation analysis, low copy-number plasmids derived from the CEN-LEU2 vector pRS315 (Sikorski and Hieter 1989) and containing wild-type (WT) or mutant MSH6 alleles were transformed into various msh6Δ leu2Δ strains (Table 1). MSH6 was contained on pRDK3572, msh6Δ2-251 on pRDK4650, msh6Δ51-251 on pRDK4715, and msh6-FFAA,Δ51-251 on pRDK4758 (Shell et al. 2007). Transformants used for complementation analysis were selected and maintained on leucine-deficient medium.

The progenitor strains (indicated in parentheses) used in various constructions in the AS4- and AS13-derived strains were as follows: JSY162 (PD73), SJR2184 (PD73), SJR2203 (SJR2184), JSY332 (SJR2203), SJR2577 (AS4), and SJR2578 (PD73). The msh2Δ∷hygB allele in JSY332 was introduced by transforming SJR2203 with a PCR fragment generated by amplification of the plasmid pAG32 (Goldstein and McCusker 1999) with the primers f msh2-Δ and r msh2-Δ (Kearney et al. 2001). ARG4 derivatives were constructed of the strains AS4, RKY1721 (Alani et al. 1994), and JSY173 (JSY208, JSY203, and JSY209, respectively) by one-step transplacement with the 2.4-kb SalI fragment derived from the plasmid pMW52 (White et al. 1993).

Measuring mitotic recombination and mutation rates:

For strains in the SJR328 background, individual colonies were grown in 5 ml YEPGG or 10 ml SGG −Leu medium at 30° for 2 or 3 days, respectively; strains of the AS4 or AS13 background were grown for 1 day under the same conditions in YPD. The cells were harvested by centrifugation and washed with water. Appropriate dilutions were plated to select for His⁺ recombinants or canavanine-resistant (Can-R) mutants, as well as on the appropriate nonselective medium to determine the total number of cells per culture. Colonies were counted after 2 days on YPD and SD or after 4 days on GGE media.

To calculate recombination and mutation rates, the median number of His⁺ or Can-R colonies was determined using at least 12 independent cultures. The method of the median (Lea and Coulson 1949) was used to calculate recombination rates (number of events per cell per generation). Ninety-five percent confidence intervals (C.I.'s) were obtained by ranking the selective medium counts in ascending order (Spell and Jinks-Robertson 2004a), and then using Table B11 from Altman (1990) to determine which cultures provided the upper and lower limits of each C.I. Confidence limits on the ratio of homeologous to homologous recombination (HER:HR) were calculated using the 95% C.I.'s on individual homeologous and homologous recombination rates. To calculate the lower confidence limit for the ratio, we divided the lower value of the 95% C.I. for the homeologous rate by the upper value of the 95% C.I. for the homologous rate. The upper confidence limit for the ratio was calculated by dividing the upper limit for the homeologous rate by the lower limit for the homologous rate.

Measuring meiotic recombination:

Diploid strains were sporulated at either 18° or 30° as indicated in the tables. Because some of the mutants confer a mitotic mutator phenotype that causes high levels of spore inviability, haploid strains were mated overnight at 30° and the resulting diploids were transferred directly to sporulation medium without prior purification. Following tetrad dissection onto YPD medium, spore colonies were replica plated to various synthetic media lacking the appropriate amino acid to check the segregation of markers. Spore colonies on SD −His or SD −Arg medium were examined microscopically to detect PMS.

In the experiments involving the plasmid-borne MSH6 and msh6-FFAA,Δ51-251 alleles, the MATa haploid strains with the plasmids (SJR2706 and SJR2707) were grown in SD −Leu medium at 30° overnight. The MATα strain without the plasmid (SJR2705) was grown overnight on YPD at 30°. Strains were mated and transferred to sporulation medium as described above.

RESULTS

Previous studies have shown that mutations within either the PCNA-binding (PIP box) domain of MMR proteins (Clark et al. 2000; Flores-Rozas et al. 2000; Lee and Alani 2006; Shell et al. 2007) or the PCNA-encoding POL30 gene (Johnson et al. 1996; Umar et al. 1996; Kokoska et al. 1999; Lau et al. 2002) reduce the efficiency of replication-error removal. It has not been determined, however, if these mutations similarly affect the role of the MMR machinery in recombination-related processes. We thus examined the effects of mutations in POL30, MLH1, or MSH6 on the MMR-dependent regulation of mitotic recombination fidelity and compared these effects to those seen in mutation-rate assays. Some alleles were additionally examined for their effects on the repair of mismatches in meiotic recombination intermediates.

The fidelity of mitotic recombination was examined using an inverted-repeat (IR) assay system (Nicholson et al. 2000). This system utilizes two cassettes oriented as IRs, each of which contains a 350-bp recombination substrate, a portion of the HIS3 gene, and a segment of an artificial intron (the “IR-intron” system). The substrates were either 100% identical to one another (cβ2-100% “homologous” substrates) or differed by four evenly-spaced base substitutions (cβ2-4ns “homeologous” substrates; 99% identical). A recombination event between the substrates reorients the region between them and generates a functional HIS3 gene, allowing cells to grow on medium lacking histidine (Figure 1A). While mutations in genes that have a general effect on recombination should affect homologous and homeologous recombination (HR and HER, respectively) to the same extent, those in genes that specifically regulate the fidelity of recombination will have a greater effect on the rate of HER than on the rate of HR, resulting in a ratio of HER:HR higher than that in a WT strain.

Figure 1.—

Mitotic and meiotic recombination assays. (A) The mitotic antirecombination assay utilizes inverted repeats, each of which contains a recombination substrate composed of 350 bp of the chicken β-tubulin (cβ2) cDNA sequence (shaded boxes), the 5′ (or 3′) portion of an artificial intron (solid boxes), and the 5′ (or 3′) portion of the HIS3 gene (open boxes). The recombination substrates used here were either identical (cβ2-100%) or contained four single-nucleotide differences (cβ2-4ns); heteroduplex recombination intermediates formed between the cβ2-4ns substrates contain up to four A/G or C/T mismatches. Recombination between the substrates results in reorientation of the 3′ and 5′ portions of HIS3 gene, generating a full-length gene interrupted by a substrate-containing, functional intron. Although a crossover event is illustrated, reorientation can also occur via gene conversion between sister chromatids (Chen and Jinks-Robertson 1998). (B) The creation and repair of heteroduplex DNA during meiotic recombination. Open and solid circles represent WT and mutant alleles, respectively, and recombination is initiated by a double-strand break (DSB) on the WT chromosome. Following resection of the 5′ ends, DNA synthesis (dashed lines) initiated from an invading 3′ end displaces a D-loop. The heteroduplex DNA formed when the D-loop pairs with the unbroken chromosome contains a mismatch, as indicated by the pairing of open and solid circles. Failure to repair the mismatch results in a 3:5 PMS event; repair of the mismatch results in 2:6 gene conversion event or restoration of Mendelian segregation (4:4). Reciprocal 5:3 PMS and 6:2 gene conversion events can occur if the initiating DSB occurs on the mutant chromosome.

To examine the relevance of PCNA–MMR interactions to the repair of mismatches created during meiotic recombination (meiotic MMR), we sporulated diploid strains heterozygous for the his4-AAG and arg4-17 alleles. The his4-AAG mutation is a T-to-A transversion at the second position of the HIS4 start codon (Detloff et al. 1991) and the arg4-17 mutation is a T-to-A transversion at position +127 of ARG4 (White et al. 1985). In both cases, heteroduplexes formed between the WT and mutant alleles create an A/A or T/T mismatch. These mismatches can be repaired to produce either a detectable gene conversion event (2 WT:6 mutant or 6 WT:2 mutant) or an undetectable restoration event (4 WT:4 mutant). If a mismatch fails to be repaired, the resulting spore colony will exhibit a sectored His⁺/His⁻ or Arg⁺/Arg⁻ phenotype. Such PMS events are manifested as either a 3 WT:5 mutant or a 5 mutant:3 WT segregation pattern in tetrads (Figure 1B). Other patterns of segregation in which more than one conversion and/or PMS event occur within a single tetrad are also observed, although such tetrads are much less frequent than tetrads with a single event (Detloff et al. 1991). The abundance of PMS tetrads relative to all classes of aberrant segregants reflects meiotic MMR efficiency.

Regulation of mitotic recombination fidelity in MMR-defective pol30 mutants:

Four of the pol30 alleles used previously by others to demonstrate a role for PCNA in the repair of replication errors were used in our experiments to determine whether PCNA is important in the MMR-dependent regulation of mitotic recombination fidelity. Two of the alleles, pol30-52 and pol30-104, confer microsatellite instability in an MMR-dependent manner (Ayyagari et al. 1995; Johnson et al. 1996; Kokoska et al. 1999), but also have been shown to cause phenotypes not associated with MMR, such as cell–cycle and replication defects (Ayyagari et al. 1995; Amin and Holm 1996; Johnson et al. 1996; Merrill and Holm 1998; Chen et al. 1999; Kokoska et al. 1999). The other two pol30 alleles, pol30-201 and pol30-204, were identified in a screen for PCNA mutations that increase mutation rate without causing the additional defects associated with the pol30-52 and pol30-104 alleles (Lau et al. 2002). In addition to measuring HR and HER rates, we examined the forward mutation rate at the CAN1 locus (canavanine resistance, Can-R) in each pol30 mutant to compare the associated mutator phenotypes in our strain backgrounds to those reported previously.

The effects of the pol30 alleles in the forward mutation and mitotic recombination assays are presented in Tables 2 and 3, respectively, along with those conferred by a complete MMR deficiency (msh3Δ msh6Δ mutant). In the SJR328 background, elimination of MMR resulted in a 55-fold elevation in the Can-R rate. Although a slightly stronger mutator phenotype for the pol30-104 than for the pol30-52 allele was reported previously (Chen et al. 1999), in our experiments the pol3-52 allele conferred a significantly stronger mutator phenotype than did the pol30-104 allele (123- and 29-fold increases in Can-R relative to WT, respectively). Relative to the pol30-52 and pol30-104 alleles, the pol30-201 and pol30-204 alleles conferred only a modest mutator phenotype, with Can-R rates being elevated 8- and 6-fold, respectively. These pol30-201 and pol30-204 mutants appear to retain 85–90% of their MMR-related spellchecker activity in our strain background, a level >25–30% reported previously (Lau et al. 2002). The variability in relative Can-R rates in different strain backgrounds suggests that inherent genetic differences can modulate the magnitude of the mutator effects associated with specific PCNA mutations (see below as well).

TABLE 2.

CAN1 forward mutation rates in pol30 haploid strains

Background	Relevant genotype	Rate × 108	(C.I.)	Relative to WT
SJR328a	WT	8.8	(7.9–9.3)	1.0
	msh3Δ msh6Δ	481	(422–513)	54.6
	pol30-52	1080	(437–1400)	123
	pol30-104	258	(196–298)	29.3
	pol30-201	68.8	(62–121)	7.8
	pol30-204	50.6	(39.2–61.7)	5.7
	pol30-201 msh3Δ msh6Δ	499	(295–624)	56.6
	pol30-201, 204	728	(626–801)	82.6
AS13b	WT	14.1	(11.6–16.2)	1.0
	msh2Δ	252	(234–282)	17.9
	pol30-52	908	(707–1390)	64.4
	msh2Δ pol30-52	4120	(2400–5260)	292
	pol30-201	173	(149–195)	12.3
	pol30-201 msh2Δ	342	(325–394)	24.3
AS4b	WT	11.7	(11.0–16.5)	1.0
	msh2Δ	193	(179–209)	16.5
	pol30-52	1290	(857–1830)	110

WT, wild type; C.I. = 95% confidence interval.

^aRates determined by combining data from the cβ2-100% and cβ2-4ns strains (see Table 1 for strain names).

^bStrain names for the AS13 derivatives were PD73 (WT), RKY1452 (msh2Δ), JSY162 (pol30-52), JSY222 (msh2Δ pol30-52), SJR2203 (pol30-201), and JSY332 (pol30-52 msh2Δ); names for the AS4 derivatives were JSY208 (WT), JSY203 (msh2Δ), and JSY209 (pol30-52).

TABLE 3.

Mitotic recombination rates between the HR (cβ2-100%) and HER (cβ2-4ns) substrates

	Recombination rate × 108
Genotype	HR	(C.I.)	HER	(C.I.)	HER:HR ratea
Wild type	135	(119–158)	8.2	(7.1–8.7)	0.061	(0.045–0.073)	1.0
msh3Δ msh6Δ	251	(206–283)	266	(227–290)	1.1	(0.8–1.4)	17
pol30-52	1020	(758–1490)	131	(115–305)	0.13	(0.077–0.40)	2.1
pol30-104	432	(306–680)	25	(15–30)	0.058	(0.02–0.098)	0.95
pol30-201	225	(174–364)	42.3	(36.3–52.2)	0.19	(0.10–0.30)	3.1
pol30-204	535	(389–578)	42.2	(35.8–53.2)	0.079	(0.062–0.14)	1.3
pol30-201 msh3Δ msh6Δ	613	(437–1000)	382	(348–473)	0.62	(0.35–1.1)	10
pol30-201, 204	705	(483–886)	91.8	(70.7–119)	0.13	(0.08–0.25)	2.1

C.I. = 95% confidence interval.

^a99% confidence limits of the HER:HR ratios are shown in parentheses and were calculated as described in materials and methods. The underlined numbers are the HER:HR rates normalized to wild type.

The effects of the pol30 alleles on mitotic recombination fidelity are best seen by dividing the HER rate by the HR rate and comparing this ratio to the HER:HR rate ratio in the WT control strain (Table 3). In the msh3Δ msh6Δ mutant, the HER:HR ratio was elevated 17-fold relative to that in a WT strain. Although both the pol30-52 and pol30-104 alleles conferred a significant increase in the HR rate (8- and 3-fold, respectively), consistent with a general replication defect, only the pol30-52 allele conferred a slightly elevated HER:HR ratio (2-fold increase). The HR rates in both the pol30-201 and pol30-204 mutant were also significantly elevated relative to that in the WT strain (2- and 4-fold, respectively). We suggest that these mild hyperrecombination phenotypes may be indicative of a subtle, perhaps background-related, replication defect that was not apparent in the initial examination of these alleles (Lau et al. 2002). The 5-fold increase in HER associated with the pol30-204 allele was similar to the 4-fold increase in HR, indicating no significant impairment in the efficiency of MMR-associated antirecombination. In contrast, in the pol30-201 strain, recombination between the HER substrates was elevated significantly more than was recombination between the HR substrates (5- and 2-fold, respectively). In the complete absence of MMR (msh3Δ msh6Δ mutant), the presence of the pol30-201 allele conferred no additional increase in the HER:HR ratio, consistent with a weak MMR-associated antirecombination activity for PCNA.

To determine the joint effect of the pol30-201 and pol30-204 mutations on the MMR-related roles of PCNA, we constructed the pol30-201,204 double mutant allele. The Can-R rate of the pol30-201,204 double mutant was 10-fold greater than that of either of the single-mutant strains and was similar to that observed in the MMR-defective (msh3Δ msh6Δ) background (Table 2). Thus, in terms of mutator phenotype, there appears to be a synergistic interaction between the two mutations when both are combined in the same protein. This synergism could reflect a complete loss of MMR or combined MMR and replication defects. In contrast, the 5-fold elevation in homologous recombination in the double mutant was very similar to the 4-fold elevation in the pol30-204 single mutant. The HER:HR ratio in the double mutant was elevated 2-fold over that of the WT strain, a value slightly less than that obtained in the pol30-201 single mutant and much less than that obtained in the msh3Δ msh6Δ background. The double-mutant data indicate that PCNA alterations can differentially affect the MMR-associated editing of replication errors and recombination intermediates.

Repair of mismatches in meiotic heteroduplex DNA in pol30 mutant strains:

The slightly elevated HER:HR ratios in the pol30-52 and pol30-201 mutants indicate that PCNA may play a minor role in the MMR-dependent regulation of mitotic recombination fidelity. These mutant alleles were introduced into diploid strains heterozygous for the his4-AAG and arg4-17 alleles to examine PMS frequencies, the elevation of which would indicate a role for PCNA in meiosis-specific MMR. As in the SJR328 background used to monitor mitotic recombination fidelity, the mutator phenotypes conferred by the pol30-52 and pol30-201 alleles were determined in the haploid parents of the diploids used in the meiotic studies. In both the AS4- and AS13-derived pol30-52 haploids, Can-R mutation rates were 3- to 6-fold higher than those observed in the corresponding MMR-defective (msh2Δ) mutants (Table 2). In addition, the pol30-52 msh2Δ AS13-derived strain had a Can-R mutation rate that was 4-fold higher than that of the pol30-52 mutant (Table 2). Thus, as reported previously, the pol30-52 allele elevates mutation rates in this genetic background in two ways: by reducing the efficiency of MMR and by elevating mutation rates by an MMR-independent mechanism (Chen et al. 1999; Kokoska et al. 1999). In the AS13-derived haploid, pol30-201 conferred a mutation rate that was only slightly less than that of an msh2Δ mutant, a phenotype much stronger than that observed in the SJR328 background, where 85% of MMR spellchecker activity was retained in the pol30-201 mutant (Table 2). In the AS13 msh2Δ pol30-201 background, the mutation rate was slightly higher than that observed in the single mutants, suggesting that the pol30-201 allele may also generate mutations in an MMR-independent manner.

In AS4 × AS13-derived diploids, sporulation at 18° results in elevated frequencies of meiotic recombination at the HIS4 locus (Nag and Petes 1993; Fan et al. 1995). pol30-52 mutants are cold sensitive (Ayyagari et al. 1995), however, and pol30-52 diploids did not sporulate efficiently at either 18° or 25°. Because sporulation of the pol30-52 diploid was most efficient at 30°, MMR-proficient (WT) and MMR-deficient (msh2Δ) strains were also sporulated at 30° for comparison. Sporulation of the WT strain at 30° resulted in a 3-fold decrease in the percentage of tetrads with aberrant segregation of his4-AAG relative to sporulation at 18° (Table 4), but did not significantly reduce the frequency of tetrads with aberrant segregation of arg4-17 (Table 5). Compared to the WT strain, the pol30-52 allele resulted in a significant decrease in the frequency of tetrads with aberrant segregation of arg4-17 and a slight but not statistically significant decrease in tetrads with aberrant segregation of his4-AAG.

TABLE 4.

Meiotic segregation of the his4-AAG marker

Sporulation temperature	Strain	Relevant genotype	No. tetrads	Ab tetrads (%)	PMS tetrads (%)	PMS/Ab events (%)
30°	PD83	WT	335	19	1.2	6.1
	No. 5	msh2Δ	79	20	14	72
	JSY175	pol30-52	74	14	4.1	30d
18°	PD83a	WT	482	58	10	18
	No. 5b	msh2Δ	111	64	55	90
	JSY346	pol30-201	336	34	16	52e
	JSY354	pol30-201 msh2Δ	184	35	27	79e
	DB101/HMY95a	mlh1Δ	465	57	49	89
	SJR2577 × SJR2578	mlh1-QLF	226	51	38	78f
	SJR2705 × SJR2706c	WT	165	63	21	30
	SJR2705 × SJR2707c	msh6-FFAA,Δ51-251	305	60	38	62g

The columns of % events are as follows: % Ab tetrads, percentage of total tetrads with aberrant (non-4:4) segregation; % PMS tetrads, percentage of total tetrads with one or more PMS events, excluding tetrads with one gene conversion event and one PMS event (7:1, 1:7); % PMS/Ab events, percentage of aberrant events that are PMS events, calculated as described by Stone and Petes (2006). Two-tailed Fisher's exact tests were used to determine P-values and P-values were considered significant when <0.05. WT, wild type.

^aData from Welz-Voegele et al. (2002).

^bData from Kirkpatrick and Petes (1997).

^cAlleles were encoded by the following plasmids: MSH6 by pRDK3572 and msh6-FFAA,Δ51-251 by pRDK4758 (Shell et al. 2007). To ensure that the plasmid was present during meiosis, only those tetrads that had at least one Leu⁺ spore were scored.

^dSignificant increase from PD83 (P = 0.04) and significant decrease from JSY200/No. 5 (P = 0.05) in the number of PMS vs. the number of non-PMS aberrant events.

^eSignificant increase from PD83 (P < 0.001) and significant decrease from JSY200/No. 5 (P < 0.001) in the number of PMS vs. the number of non-PMS aberrant events.

^fSignificant increase from PD83 (P < 0.001) and significant decrease from DBY101/HMY95 (P = 0.003) in the number of PMS vs. the number of non-PMS aberrant events.

^gSignificant increase from the WT SJR2705 × SJR2706 control (P < 0.001).

TABLE 5.

Meiotic segregation patterns of the arg4-17 marker

Sporulation temperature	Strain	Relevant genotype	No. tetrads	Ab tetrads (%)	PMS tetradsb (%)	PMS/Ab events (%)
30°	PD83	WT	335	9.6	0	0
	No. 5	msh2Δ	79	8.9	5	57
	JSY175	pol30-52	74	1.4	1.4	100c
18°	MW103/JSY338a	WT	742	7.5	0.1	1.7
	JSY240/JSY343a	msh2Δ	502	16	11	67
	JSY346	pol30-201	336	8.6	1.5	16d
	JSY354	pol30-201 msh2Δ	184	11	3.8	35e
	JSY230	mlh1Δ	192	13	6.8	50
	SJR2577 × SJR2578	mlh1-QLF	226	20	10	50f
	SJR2705 × SJR2706b	WT	165	13	1.2	9.1
	SJR2705 × SJR2707b	msh6-FFAA,Δ51-251	305	14	1.6	11g

The columns of % events are as follows: % Ab tetrads, percentage of total tetrads with aberrant (non-4:4) segregation; % PMS tetrads, percentage of total tetrads with one or more PMS events, excluding tetrads with one gene conversion event and one PMS event (7:1, 1:7); % PMS/Ab events, percentage of aberrant events that are PMS events, calculated as described by Stone and Petes (2006). Two-tailed Fisher's exact tests were used to determine P-values and P-values were considered significant when <0.05. WT, wild type.

^aData from Stone and Petes (2006). MW103/JSY338 and [JSY240/JSY343, JSY230] are isogenic to PD83 and No. 5, respectively, except for mutations at the HIS4 locus, which have no significant effect on segregation of the arg4-17 marker. Data from MW103/JSY338, JSY240/JSY343, and JSY230 are shown because spore colonies from these strains were examined microscopically for small sectors on medium lacking arginine, while spore colonies from 18° sporulations of PD83 and No. 5 were not.

^bAlleles were encoded by the following plasmids: MSH6 by pRDK3572 and msh6-FFAA,Δ51-251 by pRDK4758 (Shell et al. 2007). To ensure that the plasmid was present during meiosis, only those tetrads that had at least one Leu⁺ spore were scored.

^cSignificant increase from PD83 (P = 0.03) but no significant difference from No. 5 (P = 1) in the number of PMS vs. the number of non-PMS aberrant events.

^dSignificant increase from MW103/JSY338 (P = 0.02) and significant decrease from JSY240/JSY343 (P < 0.001) in the number of PMS vs. the number of non-PMS aberrant events.

^eSignificant increase from MW103/JSY338 (P < 0.001), significant decrease from JSY240/JSY343 (P = 0.01), and no significant change from JSY346 (P = 0.18) in the number of PMS vs. the number of non-PMS aberrant events.

^fSignificant increase from MW103/JSY338 (P < 0.001) and no significant change from JSY230 (P = 1) in the number of PMS vs. the number of non-PMS aberrant events.

^gNo significant difference from SJR2705 × SJR2706 (P = 1).

The efficiency of meiotic MMR is commonly expressed as the ratio of the number of PMS tetrads to the total number of aberrant tetrads (PMS + gene conversion tetrads). Because of the high level of meiotic recombination in the AS4 × AS13 diploids, however, a single tetrad may contain multiple aberrant events. To more accurately reflect MMR efficiency in these diploids, we calculated the ratio of total PMS events to total aberrant events (see Stone and Petes 2006 for an explanation of the calculation). We previously reported that there was a low level of PMS among total aberrant events for the his4-AAG mismatch in the WT strain sporulated at 18° (Detloff et al. 1991; Stone and Petes 2006); the percentage of PMS/total aberrant events was 18% (Table 4). At 30°, this ratio was even lower (6%; Table 4). In the pol30-52 mutant strain sporulated at 30°, 30% of the aberrant events were PMS events (Table 4). This significant increase in PMS events relative to that observed in WT demonstrates that the pol30-52 mutation reduces the efficiency of meiotic repair of mismatches in recombination intermediates. The ratio of PMS to total aberrant events observed for the pol30-52 strain was only about half of the ratio found in MMR-deficient msh2Δ strain, however, indicating that the pol30-52 mutation reduces but does not eliminate meiotic MMR. For the arg4-17 marker, a significant reduction in the percentage of aberrant tetrads in the pol30-52 diploid (from 9.6% in WT to 1.4%) made it difficult to determine the effect of the pol30-52 mutation on PMS. Of the 74 complete tetrads examined, however, the one that was aberrant was also a PMS tetrad, whereas no PMS events were observed in the 32 aberrant tetrads from the WT strain.

In contrast to the pol30-52 allele, the pol30-201 allele does not confer cold sensitivity, thus allowing examination of the pol30-201 diploid after sporulation at 18°. The pol30-201 diploid had a lower level of aberrant segregation events at HIS4 relative to the WT strain (Table 4) but the frequency of aberrant segregation at the ARG4 locus was not affected by pol30-201 (Table 5). The efficiency of MMR, as measured by the elevated PMS/aberrant events ratio, for both the his4-AAG and arg4-17 mismatches was significantly reduced by pol30-201, with the effect on the his4-AAG mismatch being more severe. For the his4-AAG mismatch, the PMS/aberrant events ratio was ∼60% of that observed in the msh2Δ strain; for arg4-17 the ratio was only 25% of that in the msh2Δ strain.

The decrease in aberrant segregation of the his4-AAG marker in the pol30-201 diploid could be due to a reduction in either the initiation of recombination at HIS4 or the subsequent extension of heteroduplex DNA. Shortening of meiotic heteroduplexes can be detected genetically by examining the frequency of aberrant segregation as a function of distance from an initiating double-strand break (DSB). The aberrant segregation of markers far from the DSB will be greatly reduced compared to WT, while there will be less of an effect on segregation of markers located close to the DSB site. We thus examined the effect of the pol30-201 mutation on segregation of the his4-3133 marker, a poorly repaired marker that is located >2 kb farther from the HIS4 DSB site than the his4-AAG marker (Detloff et al. 1992). There was 37% aberrant segregation (126/344 tetrads) for the his4-3133 marker in the WT strain PD99 (Detloff et al. 1992), but only 9.4% aberrant segregation (27/288 tetrads) for this marker in the pol30-201 mutant strain JSY356. This difference corresponds to a 75% reduction relative to the WT level of aberrant segregation for his4-3133. Because the level of aberrant segregation of the his4-AAG marker was reduced by only 40% in the pol30-201 strain (Table 4), these data are consistent with a shorter average length of meiotic heteroduplexes in the pol30-201 strain. Also consistent with shortened meiotic heteroduplexes in pol30-201 strains was a 20–40% reduction in crossovers in each of three genetic intervals examined on chromosome III (data not shown). An association between shortened meiotic heteroduplexes and reduced crossing over has previously been documented in strains with a mutant DNA polymerase δ (pol3-ct; Maloisel et al. 2004).

Effect of perturbing the Mlh1–PCNA interaction on MMR functions during recombination:

Mlh1 interacts physically with PCNA and contains an amino acid sequence related to the PIP box that is required for this interaction in in vitro assays (Lee and Alani 2006). Because mutation of this region of Mlh1 (the mlh1-Q572A,L575A,F578A or mlh1-QLF allele) confers a complete MMR defect in a frameshift reversion assay (Lee and Alani 2006), we examined its effect on the MMR-related spellchecker and antirecombination functions in the SJR328 background. As reported for the repair of frameshift intermediates, the mlh1-QLF allele resulted in a Can-R mutation rate indistinguishable from that of an mlh1Δ strain (Table 6). With diverged substrates in the IR-intron system, we consistently have observed that ∼50% of the MMR-associated antirecombination activity persists in pms1Δ or mlh1Δ strains, indicating that the yeast MutS homologs can function independently of the major MutL heterodimer in this assay (Chen and Jinks-Robertson 1999; Nicholson et al. 2000; Spell and Jinks-Robertson 2003). In agreement with this observation, there was a 6-fold increase in the HER:HR ratio in the mlh1Δ relative to the WT strain (Table 6), compared to a 17-fold increase in this ratio in the msh3Δ msh6Δ mutant (Table 3). In the mlh1-QLF mutant, the HER:HR ratio was elevated 4-fold relative to the WT ratio. Importantly, the HER rates did not differ statistically in the mlh1Δ and mlh1-QLF mutants.

TABLE 6.

Mutagenesis and mitotic recombination in strains with PCNA interaction-defective alleles of MLH1 and MSH6

	CAN1 forward mutationa			Recombination rate × 108
Relevant genotype	Rate × 108	(C.I.)	Relative to WT	HR	(C.I.)	HER	(C.I.)	HER:HR rateb
WT	11.0	(7.2–13.6)	1.0	213	(198–232)	16.6	(14.4–19.3)	0.078 (0.062–0.98)	1.0
mlh1Δ	1110	(768–1480)	101	222	(169–306)	102	(56.2–159)	0.46 (0.18–0.94)	5.9
mlh1-QLF	508	(444–818)	57.7	192	(144–208)	54.4	(43.9–64.5)	0.28 (0.21–0.45)	3.6
Plasmid-encoded allele in msh6Δ strainsc
MSH6	17.3	(13.4–18.6)	1.0	91.3	(78.8–99.3)	7.67	(6.31–9.48)	0.084 (0.064–0.12)	1.0
None	973	(765–1160)	56.2	108	(92.8–131)	75.9	(66.3–96.1)	0.70 (0.51–1.0)	8.3
msh6Δ2-251	402	(272–466)	23.2	92.2	(73.5–106)	13.5	(11.9–15.7)	0.15 (0.11–0.21)	1.8
msh6Δ51-251	37.2	(21.5–40.0)	2.15	103	(68.0–135)	9.50	(8.00–14.2)	0.092 (0.059–0.21)	1.1
msh6-FFAA,Δ51-251	236	(151–279)	13.6	135	(96.3–151)	11.5	(9.30–15.6)	0.085 (0.062–0.16)	1.0

C.I. = 95% confidence interval; WT, wild type.

^aThe Can-R rates were calculated by pooling data from the HR and HER strains of a given genotype.

^bNinety-nine percent confidence limits of the HER:HR ratios are shown in parentheses and were calculated as described in materials and methods. The underlined numbers are the HER:HR rates normalized to WT.

^cAlleles were encoded by the following plasmids: MSH6 by pRDK3572, msh6Δ2-251 by pRDK4650, msh6Δ51-251 by pRDK4715, and msh6-FFAA,Δ51-251 by pRDK4758 (Shell et al. 2007). HR and HER rates were measured in the SJR1856 and SJR1850 strain backgrounds, respectively.

As with the pol30 alleles that confer mitotic MMR defects, we also examined the effects of the mlh1-QLF allele on meiotic segregation of the his4-AAG and arg4-17 alleles (Tables 4 and 5, respectively). In the mlh1Δ strain, 89% of the his4-AAG aberrant segregation events were PMS events compared to only 18% PMS events in the WT strain. There also was a strong elevation in PMS events in the mlh1-QLF mutant, with 78% of the aberrant events being PMS events. With the arg4-17 allele, 50% of the aberrant events were PMS events in both the mlh1Δ and mlh1-QLF strains, compared to only 1.7% PMS events in the WT control strain.

Recombination-related effects of MSH6 alleles that disrupt interaction with PCNA:

Although the only defect of Mlh1-QLF protein detected in vitro is its interaction with PCNA (Lee and Alani 2006), it is possible that this property may not accurately reflect the molecular defect in vivo. First, it is unclear whether the region of Mlh1 containing the QLF residues is available for potential interaction with PCNA. Although the crystal structure of a bacterial MutL fragment suggests that this region would be on the surface of the eukaryotic MutLα heterodimer (Guarne et al. 2004), there is a report that this region is buried at the heterodimer interface (Kosinski et al. 2005). Second, the interaction of Mlh1 with PCNA described in vitro is not only very weak relative to the Msh6–PCNA interaction, but it involves electrostatic rather than the typical hydrophobic interactions (Lee and Alani 2006). Because of uncertainty of the molecular defect associated with the Mlh1-QLF protein, we also examined alleles of MSH6 that perturb its interaction with PCNA.

The Msh6 protein has a consensus PIP box that is required for interaction with PCNA in in vitro assays, but mutation of only the PIP box (msh6-F33AF34A allele, abbreviated here as msh6-FFAA) has only subtle effects on mutation rates (Flores-Rozas et al. 2000) and did not detectably affect antirecombination activity in the IR-intron assay (data not shown). A recent analysis of the Msh6–PCNA interaction indicates that the unstructured N terminus of Msh6 forms an extended tether to PCNA, and msh6 alleles that are additionally missing this region have a much stronger phenotype than those that simply have a mutated PIP box (Shell et al. 2007). We thus examined the ability of three plasmid-encoded msh6 alleles to complement the mitotic phenotypes of an msh6Δ strain: msh6Δ2-251, msh6Δ51-251, and msh6-FFAA,Δ51-251 (Shell et al. 2007). Whereas the msh6Δ51-251 allele produced a weak mutator phenotype similar to that of an msh6-FFAA allele (data not shown), the msh6-FFAA,Δ51-251 allele with both mutations as well as the msh6Δ2-251 allele produced strong mutator phenotypes in the CAN1 assay (Table 6). Our mutator results are completely consistent with those reported by Shell et al. (2007). In the msh6Δ control strain, the HER:HR rate ratio was elevated 8.3-fold (Table 6), which, as reported previously, is 2-fold less than that typically seen in msh3Δ msh6Δ or msh2Δ mutants (Nicholson et al. 2000; Spell and Jinks-Robertson 2003). Only the msh6Δ2-251 allele elevated the HER:HR ratio a very modest 1.8-fold. Of particular significance, the Msh6-FFAA,Δ51-251 protein, which had lost substantial MMR function in the CAN1 forward mutation assay (14-fold elevation in Can-R rate relative to a 56-fold elevation in the absence of Msh6), retained much, if not all, antirecombination activity.

The mitotic behavior of the msh6-FFAA,Δ51-251 allele suggests that the antirecombination and spellchecker functions of the encoded protein may be separable. To explore this further, we analyzed the repair of meiotic heteroduplexes in an msh6Δ/msh6Δ diploid containing either an MSH6 or an msh6-FFAA,Δ51-251 allele on a complementing LEU2/CEN plasmid (Table 4). Because of plasmid stability issues, only those tetrads that produced at least one Leu⁺ spore were included in the analyses of heteroduplex repair. Even so, PMS of the his4-AAG allele was elevated in the Msh6-complemented control strain relative to a diploid with a chromosomal MSH6 allele (30% vs. 18% PMS/aberrant events). In the strain with the complementing msh6-FFAA,Δ51-251 allele, the level of PMS was further elevated to 62%, demonstrating a clear and substantial loss of the meiotic mismatch repair activity of the encoded protein.

DISCUSSION

The yeast MMR system regulates genome stability by editing the fidelity of DNA replication and homologous recombination. In addition to the canonical MutS and MutL homologs, PCNA is important for MMR-associated spellchecker activity, as mutations that disrupt physical interaction with Msh3, Msh6, or Mlh1 result in a mutator phenotype (Clark et al. 2000; Flores-Rozas et al. 2000; Lee and Alani 2006; Shell et al. 2007). Although the precise role of PCNA during replication-associated MMR is not clear, there are three general, nonmutually exclusive ways in which it may be important. First, the localization of PCNA to the replication fork provides a mechanism to concentrate MMR proteins to newly replicated DNA and thereby enhance the efficiency of error recognition. In addition, because of the directional orientation of the PCNA ring with respect to the template and nascent strands, PCNA may also provide a signal for strand discrimination (Umar et al. 1996). Second, in vitro binding assays have shown that “free” PCNA enhances the ability of MMR complexes to distinguish between DNA molecules with mismatches and those that lack mismatches (Flores-Rozas et al. 2000). Although the relevance of this observation to in vivo MMR is uncertain, it should be noted that PCNA–MMR interactions are important in the repair of mismatches engineered into transforming plasmids (Lau et al. 2002). Whether this repair occurs prior to the initiation of plasmid replication is not known. Third, because proteins that are sequentially involved in MMR (e.g., Msh3/6, Mlh1, and Exo1) interact with PCNA, PCNA may coordinate the overall repair process by a protein hand-off mechanism (Lee and Alani 2006).

In all current models of recombination strand invasion is followed by replicative extension of the invading 3′ end (for a review see Krogh and Symington 2004), and hence one might expect PCNA to be present at an early stage. Because most mismatches in recombination intermediates would be formed via strand invasion (and possibly branch migration) rather than as a result of replication, however, it is not obvious whether PCNA might be similarly required for mismatch repair and/or antirecombination. It is possible, for example, that mismatch recognition and processing occurs at the initial strand-invasion step, before any DNA synthesis occurs. In addition, although repair directed to the newly synthesized DNA strand is essential for the spellchecker function of MMR, unbiased repair of mismatches generated during recombination would have relatively minor biological consequences. In the current study, we have specifically addressed the role of PCNA in the recombination-related processing of mismatches in both mitotic and meiotic recombination intermediates.

The fate of mismatches in meiotic heteroduplex DNA was assessed by tetrad dissection, where a decrease in the efficiency of mismatch repair leads to an increase in the ratio of PMS to total aberrant events. It should be noted that, as in the case of replication errors, the repair of meiotic heteroduplex DNA presumably involves mismatch excision followed by DNA synthesis to fill in the resulting gap. In the case of spontaneous mitotic recombination, it is not possible to directly study mismatch correction in an analogous manner. We thus used a mitotic antirecombination assay in which the production of recombinants between diverged sequences is strongly inhibited by the yeast MutS and MutL complexes. Because only antirecombination is dependent on the helicase activity of Sgs1 (Myung et al. 2001; Spell and Jinks-Robertson 2004b), its mechanism may be distinct from the nucleolytic destruction that characterizes the removal of replication errors. One can imagine scenarios in which MMR–PCNA interactions might be relevant to all, some, or none of the recombination processes involving mismatched heteroduplex DNA. As discussed in detail below, our data demonstrate that PCNA interactions are required for efficient MMR during meiotic recombination processes, but play a relatively minor role during mitotic antirecombination.

Some mutator alleles of PCNA are weakly defective in the processing of recombination intermediates:

PCNA was first implicated in MMR through the analysis of the pol30-104 (Johnson et al. 1996) and pol30-52 (Umar et al. 1996) alleles, each of which additionally confers cold and mutagen sensitivity. Although teasing apart the contributions of general replication problems and specific MMR defects to mutagenesis has been difficult in the corresponding mutants, they are generally assumed to be at least partially MMR defective. In our mitotic analyses, both alleles caused a significant hyperrecombination phenotype between identical substrates, underscoring the need to include homologous substrates as a control when examining HER. Whereas the pol30-104 allele did not elevate HER to a greater extent than HR, the HER:HR ratio was elevated a modest twofold in the pol30-52 mutant. The effect of the pol30-52 allele on the repair of meiotic mismatches was much more striking, with the percentage of his4-AAG PMS events among aberrant events (hereafter referred to as % PMS) increasing fivefold, from 6.1% in the WT strain to 30% in the mutant. Even so, this increase in PMS was still significantly below the 72% observed in an msh2Δ strain.

To clarify the role of PCNA in MMR-related recombination processes, we also examined two pol30 alleles whose only known defect is in MMR: pol30-201 and pol30-204 (Lau et al. 2002). Both alleles caused significant increases in mitotic HR (1.7- and 4.0-fold increases for pol30-201 and pol30-204, respectively), however, and we suggest that each likely confers a subtle replication defect. The large hyperrec phenotype of the pol30-204 strain relative to the pol30-201 strain is consistent with the observation that pol3-204, but not pol30-201, mutants are slightly sensitive to MMS and UV (Lau et al. 2002). With respect to possible non-MMR defects in the pol30-201 mutant, we found that this allele also resulted in a significant reduction in aberrant segregation of the his4-AAG marker (58% for WT, 64% for msh2Δ, and 34% for pol30-201). The reduction in aberrant segregation was even more severe for the his4-3133 allele, which is located 2 kb farther from the site of recombination initiation than the his4-AAG marker, suggesting that heteroduplex extension is reduced in the pol30-201 mutant. Together our mitotic and meiotic recombination analyses indicate a subtle impairment of DNA synthesis in the pol30-201 mutant, at least in the strain backgrounds used in this study. On the basis of the large number of processes in which PCNA participates and the fact that many of the relevant proteins interact with the interdomain connector loop of PCNA (Moldovan et al. 2007), we suggest that it may not be possible to specifically perturb one DNA metabolic process through changes in PCNA without having collateral effects on other processes.

In the mitotic antirecombination assay, the pol30-201, but not the pol30-204, allele significantly elevated the HER:HR ratio (3-fold), but to a much lesser extent than did the complete elimination of MMR (17-fold). The pol30-201 allele also reduced, but did not eliminate, the repair of mismatches in meiotic recombination intermediates (the pol30-204 allele was not analyzed), with the % PMS being elevated for both the his4-AAG and arg4-17 alleles (Tables 4 and 5). As with the pol30-52 allele, it should be noted that the meiotic defect associated with the pol30-201 allele was more pronounced than the mitotic defect. The observation that effects of the pol30 alleles on meiotic and especially mitotic recombination processes were less than those associated with the complete loss of MMR may reflect some degree of continued function of the mutant PCNAs. Alternatively, PCNA might simply increase the efficiency of, but not be absolutely required for, the MMR-dependent regulation of recombination fidelity and meiotic heteroduplex repair. Although the data obtained with the pol30-52 and pol30-201 mutants are suggestive of a minor role for PCNA in the MMR-dependent regulation of recombination, the additional DNA metabolic defects conferred by all available pol30 alleles temper the conclusions that can be drawn. For this reason, we additionally examined mutations in MMR proteins that are thought to specifically disrupt interactions with PCNA.

MMR-dependent processing of recombination intermediates is abolished in mlh1-QLF mutants:

Alteration of the canonical Msh6 PIP box sequence (msh6-FFAA allele) causes only a weak mutator phenotype (Flores-Rozas et al. 2000) while that of the putative Mlh1 PIP-like domain (the mlh1-QLF allele) completely eliminates the spellchecker activity of the MMR machinery (Lee and Alani 2006; Table 2). Consistent with the relative mutator phenotypes of msh6-FFAA and mlh1-QLF strains, a PIP box-defective msh6 allele did not detectably impair the antirecombination activity of the MMR machinery (data not shown) while the mlh1-QLF allele was indistinguishable from an mlh1 null allele in terms of antirecombination activity (Table 6) and meiotic heteroduplex repair (Tables 4 and 5).

The interaction of MutLα and MutSα with PCNA was recently examined using surface plasmon resonance. Not only was the MutSα interaction with PCNA much stronger, but the salt resistances of the PCNA-containing complexes indicated that the MutLα interaction is largely ionic whereas the MutSα interaction is hydrophobic (Lee and Alani 2006). These different strengths and modes of interaction would not be expected if Mlh1, like Msh6, interacts with PCNA via its PIP-related domain. Finally, there is debate in the literature as to whether the region of Mlh1 defined by the mlh1-QLF allele would even be available for interaction with PCNA (Kosinski et al. 2005, but also see Clark et al. 2007). Given these considerations, it is not clear whether the phenotype conferred by mutating the PIP-related region of Mlh1 indeed reflects a specific PCNA-interaction defect, or whether the mlh1-QLF allele is MMR defective for an unrelated, uncharacterized reason. Additional data obtained with defined msh6 alleles and discussed below suggest that the latter may be more likely.

Relevance of the PCNA interaction to the recombination-related roles of Msh6:

Recent data suggest that many PIP-domain-containing proteins are tethered to PCNA via an unstructured, flexible linker region (Shell et al. 2007). In the case of Msh6, this linker corresponds roughly to the region between the PIP box at the amino terminus and the DNA-binding domain, which is the first region of high homology between MutS proteins. Combining a PIP-box mutation with an internal deletion that shortens the Msh6 linker (msh6-FFAA,Δ51-251 allele) confers a strong mutator phenotype, whereas the individual mutations produce only a weak phenotype (Shell et al. 2007). Given the caveats associated with MMR-defective pol30 alleles and with the mlh1-QLF allele, the double mutant msh6-FFAA,Δ51-251 allele may provide the most direct test for an involvement of PCNA in the recombination-related functions of the yeast MMR machinery. While the clear elevation in PMS frequency indicates defective repair of meiotic heteroduplex DNA in the msh6-FFAA,Δ51-251 mutant, there was no detectable impairment of MutSα antirecombination activity. Given the small, 10-fold range over which antirecombination can be measured in the IR assay, however, we cannot rule out a minor role of PCNA in regulating recombination fidelity. The relatively weak antirecombination defects conferred by the pol30-201 and pol30-52 alleles would be consistent with such a minor role, but they also could reflect the collateral effects of these alleles on DNA metabolism. We suggest that the msh6-FFAA,Δ51-251 allele is a separation-of-function allele that distinguishes PCNA-dependent repair activities of the yeast MMR machinery (i.e., the repair of mismatches in replication and recombination intermediates) from the antirecombination activity of the yeast MMR machinery, which our data suggest is largely, if not completely, PCNA independent. While this interpretation assumes that the functional redundancy of the Msh6 PIP domain with the region between residues 51 and 251 reflects PCNA-specific interactions, it is possible that redundancy might instead reflect alternative modes of localizing MMR proteins to regions of active DNA synthesis (see Shell et al. 2007 for further discussion and also Clark et al. 2007).

Implications of the PCNA requirement for the MMR-dependent processing of recombination intermediates:

The MMR machinery detects and removes replication-generated mismatches in a reaction that is facilitated by PCNA and hence likely takes place in close proximity to the replication fork. In current models of recombination, invasion of a duplex DNA is followed by DNA synthesis primed from the invading 3′ end, a reaction that likewise requires a PCNA-tethered DNA polymerase. If the MMR-directed removal of mismatches in heteroduplex recombination intermediates occurs after extension of the invading 3′ end begins, one might expect an involvement of PCNA that is similar to that seen during replication. The meiotic data presented here is consistent with such a model, with disruption of the Msh6–PCNA interaction resulting in elevated PMS among non-Mendelian tetrads. In contrast, the interaction of the MMR machinery with PCNA appears to be of relatively little importance during antirecombination. This suggests that there may be a very early recombination-associated fidelity check that occurs during the initial strand-invasion process, before DNA synthesis is initiated from the invading 3′ end. Antirecombination data obtained using a transformation-based gap-repair assay indeed suggest that the blockage of recombination is separable from the repair of mismatches in recombination intermediates. In the gap-repair assay, recombinants produced in the presence of MMR show no evidence of persistent heteroduplexes (C. Welz-Voegele and S. Jinks-Robertson, unpublished data), indicating that intermediates can escape antirecombination activity of the MMR machinery and still be subject to mismatch correction. An early editing step during recombination might involve an Sgs1-mediated unwinding of the invading 3′ end, as Sgs1 is involved in antirecombination but not in mismatch correction (Myung et al. 2001; Spell and Jinks-Robertson 2004b; Sugawara et al. 2004). Once DNA synthesis initiates, however, we speculate that the MMR machinery switches to a repair mode, thus making the repair of mismatches in recombination intermediates mechanistically very similar to the repair of replication errors. We note that a two-stage model also has been proposed to explain antirecombination in E. coli (Stambuk and Radman 1998). Whether the late repair process we propose can also become a mechanism of antirecombination may depend on the density of mismatches in recombination intermediates.