Conserved Properties between Functionally Distinct MutS ... - CiteSeerX
THE JOURNAL OF BIOLOGICAL CHEMISTRY © 1997 by The American Society for Biochemistry and Molecular Biology, Inc.
Vol. 272, No. 48, Issue of November 28, pp. 30345–30349, 1997 Printed in U.S.A.
Conserved Properties between Functionally Distinct MutS Homologs in Yeast* (Received for publication, July 25, 1997, and in revised form, September 17, 1997)
Pascale Pochart, Dana Woltering, and Nancy M. Hollingsworth‡ From the Department of Biochemistry and Cell Biology, Institute for Cell and Developmental Biology, State University of New York at Stony Brook, Stony Brook, New York 11794-5215
In the yeast Saccharomyces cerevisiae there are five nuclear MutS homologs that act in two distinct processes. MSH2, 3, and 6 function in mismatch repair in both vegetative and meiotic cells, whereas MSH4 and MSH5 act specifically to facilitate crossovers between homologs during meiosis. Coimmunoprecipitation as well as two-hybrid experiments indicate that the Msh4 and Msh5 proteins form a hetero-oligomeric structure similar to what is observed for the Msh proteins involved in mismatch repair. Mutation of conserved amino acids in the NTP binding and putative helix-turnhelix domains of Msh5p abolish function but are still capable of interaction with Msh4p, suggesting that NTP binding plays a role downstream of hetero-oligomer formation. No hetero-oligomers are observed between the mismatch repair MutS proteins (Msh2p and Msh6p) and either Msh4p or Msh5p. These results indicate that one level of functional specificity between the mismatch repair and meiotic crossover MutS homologs in yeast is provided by the ability to form distinct hetero-oligomers.
Much attention has been focused in recent years on the MutS family of proteins whose role in mismatch repair has been implicated in preventing the development of human cancers (for review, see Ref. 1). Mutations in the genes encoding these proteins from bacteria to humans result in higher mutation rates and increased genomic instability. In addition, mismatch repair proteins play a role in preventing recombination from occurring between highly diverged sequences. MutS family members typically exhibit 20 –30% amino acid identity throughout the length of the protein with two more highly conserved subdomains, a putative NTP binding site and helixturn-helix structural domain, in the COOH-terminal half. Evidence from both in vitro and in vivo analyses has demonstrated that in eukaryotes, MutS proteins function as heterodimers to recognize and bind mismatched DNA base pairs that result from errors in DNA replication or from recombination, for example (2– 8). Once the MutS homolog heterodimer is bound to the DNA, a second heterodimer comprised of MutL homologs is recruited to the mismatch (9). In analogy to the MutS-MutL system in bacteria, an endonuclease is presumably activated to allow repair of the mismatch. The budding yeast Saccharomyces cerevisiae contains six MutS homologs (for review, see Ref. 10). Five of these genes * This work was supported by a grant from the Pew Charitable Trusts as well as National Institutes of Health Grant GM50717. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ‡ To whom correspondence should be addressed: Dept. of Biochemistry and Cell Biology, 314 Life Sciences Bldg., SUNY Stony Brook, Stony Brook, NY 11794-5215. Tel.: 516-632-8581; Fax: 516-632-8575; E-mail: [email protected]. This paper is available on line at http://www.jbc.org
(MSH2– 6) encode proteins that act on nuclear DNA, whereas one, MSH1, is required for stability of the mitochondrial genome (11, 12). Of the nuclear MutS homologs, three function to repair mismatches in both vegetatively growing and meiotic cells (10). Specificity of mismatch recognition has been shown to occur by the formation of distinct heterodimeric complexes. Msh2p interacts with Msh6p to recognize both single base pair mismatches as well as small loops formed by insertions or deletions in the DNA (2, 5, 13); in addition, Msh2p can complex with Msh3p to recognize insertion/deletion loops (4, 13). The Msh2p-Msh6p protein-protein interaction and mismatch binding specificity are independent of the ability to bind and hydrolyze ATP, suggesting that ATP may be required for downstream events such as recruitment of MutL homologs (14, 15). Despite being highly homologous to other members of the MutS protein family, the two remaining MutS homologs in yeast, MSH4 and MSH5, play no role in mismatch repair in either vegetative or meiotic cells. Instead their role is to facilitate crossovers between homologs to ensure that chromosomes segregate to opposite poles at the first meiotic division (16, 17). This second function for MutS homologs appears to be conserved through evolution as homologs of Msh4p and Msh5p have been found in both nematodes and mammals.1,2 Mutants in MSH4 and/or MSH5 display levels of meiotic crossing over and spore viability which are 30 –50% of wild type. These phenotypes are observed even in homothallic diploids where no mismatches are present, underscoring the idea that Msh4p and Msh5p, unlike the other MutS family members, do not need to recognize base pair mismatches to perform their function. Instead it has been proposed that the ability to recognize aberrant DNA structures has been modulated in these proteins through evolution to recognize recombination intermediates such as Holliday junctions (16, 17). Because MSH4 and MSH5 function in a separate process from MSH2, 3, and 6, the question exists as to whether Msh4p and Msh5p have any features in common with the mismatch repair MutS homologs. The finding that msh4 msh5 diploids exhibit the same level of spore viability as either single mutant (16) suggested that these two proteins might also function as heterodimers. The work presented here confirms that the Msh4 and Msh5 proteins interact, and it demonstrates as well that, similar to Msh2p-Msh6p, hetero-oligomerization between Msh4p and Msh5p can occur in the presence of mutations that presumably destroy the ability of Msh5p to bind and/or hydrolyze ATP. Finally, data are presented which show that proteinprotein interactions between the mismatch repair and meiotic crossover classes of MutS proteins do not occur, indicating that at least one level of functional specificity between the two classes is regulated by who can form heterodimers with whom.
1 2
30345
N. M. Hollingsworth, unpublished observations. R. Kolodner, personal communication.
30346
Yeast MutS Protein Interactions EXPERIMENTAL PROCEDURES
Yeast Strains—The following S. cerevisiae strains were used: NH141 MATa/MATa leu2/leu2 HIS4/his4 ura3/ura3 his3/HIS3 trp1/trp1 CAN1/can1 CYH2/cyh2 SPO13/spo13 ADE2/ade2; NH129, NH141 containing msh5::URA3/msh5::URA3; YPP10, NH141 containing msh5::LEU2/msh5::LEU2; YPP10-66-1, YPP10 with TPIp-MSH5-3HA; L40 (18). Isogenic diploids were derived by transformation (19) of the same haploid parents with the appropriate knockout fragment (msh5::URA3 (16); msh5::LEU2, pCH2– 6) followed by mating. YPP10 was transformed with pPAS66 linearized with XhoI to integrate TPIpMSH5-3HA at URA3. Yeast media have been described (20). Plasmids—pCH2– 6 contains LEU2 inserted into the BglII site of MSH5 in pNH181 (16). The msh5::LEU2 fragment is released by cutting with HindIII and SacI. The MSH5-3HA fusion gene under control of the TPI promoter was constructed in several steps. First, a vector containing the TPI promoter and a multiple cloning site (pJAY-2) was created by digestion of pTiaO (21) (provided by N. Dean) with SacI and religation. Then the MSH5 gene alone was placed downstream of the TPI promoter by cloning the 4.2-kilobase HpaI/HindIII MSH5 fragment from pNH189 (16) into pJAY-2 cut first with EcoRI and blunt ended by T4 DNA polymerase and then HindIII to make pPAS29. To create a tagged allele of MSH5, a NotI site was first substituted for the stop codon in MSH5 by polymerase chain reaction. A polymerase chain reaction fragment containing the triple HA3 tag flanked by NotI sites was amplified using pGTEP (22) as a template, cut with NotI, and ligated in-frame to the end of MSH5. A 200-base pair EcoRI/HindIII fragment containing this fusion was then substituted for the equivalent fragment in pPAS29 to generate pPAS66. pPAS65 (lexA-MSH4), pPAS58 (GAD-MSH4), pPAS64 (GADMSH5), and pPAS111 (GAD-MSH6) were made by polymerase chain reaction using Vent polymerase (New England Biolabs) to amplify the various genes. The resulting fragments were then ligated into either pBTM116 or pGAD424 (23). Site-directed mutagenesis (24) was used to introduce specific nucleotide changes in the MSH5 gene carried in pNH189. 1.2-kilobase BglII/EcoRI fragments containing the mutations were exchanged with the equivalent fragment in pPAS64 to generate pDW10 (GAD-msh5-G643D, A644V), pPAS122 (GAD-msh5-G821D), and pPAS127 (GAD-msh5-G648R). For complementation, the 1.2-kilobase BglII/EcoRI fragments were substituted in pPAS79, which carries the MSH5 gene under its own promoter, to create pPAS135, pPAS136, and pPAS137, respectively. All three 1.2-kilobase fragments containing the mutations were sequenced to ensure that no additional mutations were introduced during the mutagenesis (data not shown). pNH108 is described in Ref. 25. 2mMSH4, pRDK371, and pORC15–268 were provided by P. Ross-MacDonald, D. Tishkoff, and T. Triolo, respectively. Immunoprecipitation—Immunoprecipitation was performed essentially as described in Ref. 26 using 100 ml of YPP10-66-1 carrying either pBTM116 or pPAS65 grown in SD-Trp to an A600 of 0.8. The proteins were fractionated by SDS-polyacrylamide gel electrophoresis, and immunoblots were probed using anti-lexA antibodies (CLONTECH) or the anti-HA 12CA5 ascites fluid (Babco). The presence of the antibody was detected using ECL (Amersham). Two-hybrid Assay—Relevant plasmids were cotransformed into L40 selecting on SD-Trp,-Leu medium. The presence of the fusion proteins was confirmed by immunoblot analysis using anti-lexA or anti-GAD antibodies (CLONTECH) (data not shown). To rule out potential artifacts created by polymerase chain reaction, several independent clones for each plasmid were tested. In addition, only fusion proteins that were known to give a positive signal in the two-hybrid system in combination with at least one other protein were used. Histidine prototrophy was assayed on SD-His plates, and b-galactosidase activity was monitored as described (25). RESULTS
Msh4p and Msh5p Physically Interact in Vivo—Protein-protein interactions between Msh4p and Msh5p were tested by two different methods. The first assay involved coimmunoprecipitation of Msh4p and Msh5p after ectopic expression in vegetative yeast cells. The MSH4 gene was fused to the 39-end of lexA under control of the constitutive ADH1 promoter, thereby allowing detection of lexA-Msh4p using anti-lexA antibodies. The MSH5 gene was tagged by the addition of three 3 The abbreviations used are: HA, hemagglutinin; GAD, Gal4p activation domain.
FIG. 1. Coimmunoprecipitation of MsH5-3HAp and lexAMsh4p. Crude extracts from YPP10-66-1 (MSH5-3HA) transformed with either lexA (pBTM116) or lexA-MSH4 (pPAS65) were incubated with anti-HA antibodies (12CA5) and the antibodies precipitated with protein A-Sepharose. The precipitated complexes were collected by centrifugation, and the proteins were fractionated by SDS-polyacrylamide gel electrophoresis. After transfer to nitrocellulose, the proteins were probed with either 12CA5 antibodies to detect MsH5-3HAp (panel A) or with anti-lexA antibodies to detect lexA-Msh4p (panel B). The asterisk (*) indicates an anti-HA cross-reacting band.
HA epitopes at the end of the MSH5 coding sequence and integrated under control of the TPI promoter in a msh5::LEU2/msh5::LEU2 diploid to create YPP10-66-1 (see “Experimental Procedures”). The Msh5-3HAp was observed using the 12CA5 antibody directed against the HA epitope. Both the lexA-MSH4 and the MSH5-3HA fusion genes under control of the heterologous promoters were able to complement the spore viability defect of msh4 and msh5 diploids, respectively (data not shown). Plasmids containing lexA alone (pBTM116) or lexA-MSH4 (pPAS65-2) were transformed into YPP10-66-1. Cell extracts derived from these strains were incubated with the 12CA5 antibody. The antibodies were precipitated by the addition of protein A-Sepharose, and the proteins were fractionated by SDS-polyacrylamide gel electrophoresis and transferred to nitrocellulose. When the immunoblots were probed with the anti-HA antibodies, the Msh5-3HA protein was found to precipitate with the beads (Fig. 1A, lanes 1 and 2). When the immunoprecipitates were probed with lexA antibodies, a band migrating with the expected molecular mass for lexA-Msh4p (123 kDa) was observed to coimmunoprecipitate (Fig. 1B, lane 6). This band is not seen when lexA alone is present (Fig. 1B, lane 4), nor when the 12CA5 antibodies are omitted (Fig. 1B, lanes 3 and 5), suggesting a specific physical interaction between Msh4p and Msh5p. The second assay utilized the two-hybrid system, a genetic method for detecting protein-protein interactions that occur in vegetatively growing yeast cells (27). This method utilizes a strain, L40, containing two reporter genes, lacZ and HIS3, which have lexA operator sites present in their promoters (18). When L40 is transformed with pPAS65-2, lexA-Msh4p is produced and presumably bound upstream of each reporter by virtue of the interaction between the lexA DNA binding domain and its binding sites. Introduction of the Gal4p activation domain (GAD) can result in activation of the reporter constructs when the Gadp is fused to a protein that interacts with lexAMsh4p. Activation of HIS3 transcription makes the cells prototrophic for histidine, whereas activation of the lacZ gene can
Yeast MutS Protein Interactions
FIG. 2. Two-hybrid interactions between lexA-MSH4 and various GAD fusions. L40 was cotransformed with pPAS65 (lexA-MSH4) and either pGAD424 (GAD), pPAS58 (GAD-MSH4), pPAS64 (GADMSH5), GAD-RED1537– 827, pNH108 (GAD-HOP1), or GAD-ORC15–268. Transformants were streaked for single colonies on SD2his, 2trp, 2leu medium to assess prototrophy for histidine.
be observed using an enzymatic assay for b-galactosidase activity. The combination of lexA-MSH4 and GAD-MSH5 produces both histidine prototrophy (Fig. 2) as well as substantial amounts of b-galactosidase activity (data not shown), indicating that lexA-Msh4p and Gad-Msh5p interact. This interaction is specific as no signal is observed when lex-MSH4 is combined either with GAD or with GAD fusions to HOP1, RED1537– 827, or ORC15–268 (Fig. 2). Like MSH4, HOP1 and RED1 encode meiosis-specific proteins that are localized on meiotic chromosomes in S. cerevisiae (17, 28, 29). ORC1 encodes part of a protein complex bound at origins of replication in yeast (30). These fusions have all been observed to interact with other proteins in the two-hybrid system and therefore are known to function in this assay (25, 31, 32). The interaction between lexA-Msh4p and Gad-Msh5p is not being mediated by lexA, as no interaction is detected with the lexA and GAD-MSH5 combination (data not shown). A positive signal was obtained with the combination of lexAMSH4 and GAD-MSH4, indicating that Msh4p can form homooligomers in vivo (Fig. 2). Whether the homo-oligomer has a function in the cell is unclear, however. The presence of MSH4 on a high copy number plasmid is unable to suppress the msh5 spore viability defect (data not shown). Furthermore, the msh4 phenotype is no more severe than msh5 (16), contrary to what one might expect if Msh4p homo-oligomers effect some function distinct from the Msh4p-Msh5p hetero-oligomer. Mutation of the Putative NTP Binding Domain of MSH5 Abolishes Function but Still Allows Interaction with MSH4 — Mutations in the NTP binding domain of both yeast MSH2 and bacterial MutS reduce ATPase activity and create noncomplementing alleles that are dominant negative when overexpressed in a wild type background (14, 33, 34). Alani et al. (14) further showed that changing a completely conserved glycine in the p-loop of the ATP binding domain of Msh2p to aspartic acid (position 693) does not affect the ability of Msh2p to interact with Msh6p, nor does it affect DNA mismatch binding specificity. To test whether mutations in the putative NTP binding domain of MSH5 have similar properties, two sitedirected mutant alleles of MSH5 were generated. msh5-G648R substitutes an arginine for the glycine at position 648, which corresponds to glycine 693 in the p-loop of Msh2p (Fig. 3). The second allele, msh5-G643D, A644V, changes the absolutely conserved glycine at position 643 of the p-loop to aspartic acid as well as making a conservative change of alanine to valine at
30347
FIG. 3. Alignment of two conserved domains in MutS and the yeast nuclear MutS homologs. Numbers represent the positions of the first amino acid in the line. Bold letters indicate amino acids conserved in all six proteins. Bold letters below the Msh5p sequence indicate amino acids substitutions in Msh5p. Panel A, p-loop of the nucleotide binding domain. Panel B, putative helix-turn-helix domain.
amino acid 644. Wu and Marinus (34) have shown previously that mutation of the corresponding glycine in MutS (position 614) creates a nonfunctional allele. The mutant alleles were then either placed under the MSH5 promoter and assayed for complementation, or they were fused to GAD and assayed for their ability to interact with lexA-MSH4. The msh5-G643D, A644V, and msh5-G648R alleles were tested for complementation of the msh5 spore inviability defect by first transforming the msh5::URA3/msh5::URA3 diploid NH129 with the plasmids pPAS135 and pPAS137, respectively. The transformants were then sporulated and asci dissected to determine the fraction of viable spores. Both mutant alleles failed to complement and appear to be null (Table I). The msh5-G643D, A644V allele was also tested for a dominant negative effect in a MSH5/msh5 diploid, and no phenotype was observed (data not shown). This finding is most likely due to a failure to overexpress the mutant protein to a sufficient level. Both GAD-msh5-G643D, A644V and GAD-msh5-G648R interact with lexA-MSH4 in the two-hybrid assay. For GADmsh5-G643D, A644V, the amount of both HIS3 (as measured by colony size on SD2His medium) and lacZ activity observed was less than that found for GAD-MSH5 (Fig. 4A; data not shown), even though more of the mutant protein is made compared with Gad-Msh5p (Fig. 4B, compare lanes 1 and 3). The Gad-msh5-G648R protein interacted slightly less well than Gad-Msh5p, but in this case the difference may be because slightly less of the mutant protein was made (Fig. 4B, lane 2). Alani et al. (14) discovered that mutation of some of the amino acids in the helix-turn-helix domain of MSH2 gives phenotypes similar to those in the G693D mutation in the ATP binding domain. For example, substituting aspartic acid for the glycine at position 855 of Msh2p (Fig. 3B) results in a protein that fails to complement a deletion of MSH2 and has reduced ATPase activity yet still interacts with Msh6p. This glycine is conserved in all of the yeast nuclear MutS homologs, as well as in MutS itself (Fig. 3B). The corresponding glycine codon in MSH5 was therefore mutated to aspartic acid thereby creating msh5-G821D. This mutation creates a null allele of MSH5 (Table I). In combination with lexA-MSH4, the GAD-msh5G821D gene shows a wild type level of histidine prototrophy (Fig. 4A) and b-galactosidase activity (data not shown) in the two-hybrid assay, indicating that these proteins interact. It therefore appears that, similar to the mismatch repair MutS proteins, ATP binding and hydrolysis are essential steps that occur downstream of hetero-oligomer formation in the meiotic crossover MutS homologs. Hetero-oligomers Are Not Observed between the Mismatch
30348
Yeast MutS Protein Interactions
TABLE I Complementation analysis of site-directed msh5 alleles Transformants of NH129 with the appropriate plasmids were selected on SD2leu medium. Several transformants were patched onto SD2leu medium, grown overnight, and replica plated to spo plates to induce sporulation. Incubation was at 30 °C for 3 days. The resulting asci were dissected and incubated at 30 °C for several days to determine the fraction of spores able to form colonies. Strain/plasmid a
NH129/YCplac111 NIH129/pPAS79 NH129/pPAS135 NH129/pPAS137 NH129/pPAS36 a
Relevant genotype
% Spore viability (asci)
msh5::URA3 msh5::URA3 msh5::URA3/MSH5 msh5::URA3 msh5::URA3/msh5-G643D,A644V msh5::URA3 msh5::URA3/msh5-G648R msh5::URA3 msh5::URA3/msh5-G821D msh5::URA3
40.1 (159) 80.8 (52) 44.3 (75) 49.3 (68) 48.8 (71)
Includes data from NH129 dissected without vector.
TABLE II Two hybrid assays using mismatch repair and meiotic crossover MutS homologs in yeast L40 was transformed with the appropriate plasmids, and filter assays were performed as described under “Experimental Procedures.” —, no signal after 4 h at 30 °C; 11, blue color observed after 90 min at 30 °C; 111, blue color observed after 30 min at 30 °C. Plasmids
pPAS65/pGAD424 pPAS65/pPAS64 pPAS65/pPAS111 pRDK371/pPAS111 pRDK371/pPAS58 pRDK371/pPAS58
lex-A
GAD-
b-Galactosidase activity
MSH4 MSH4 MSH4 MSH2 MSH2 MSH2
— MSH5 MSH6 MSH6 MSH4 MSH4
— 111 — 11 — —
using a lexA-MSH2 plasmid and GAD-MSH6 (see “Experimental Procedures”). Because the selectable marker on the lexAMSH2 plasmid is HIS3, only activation of the lacZ reporter was assayed (Table II). Cross-combinations of lexA-MSH4 and GAD-MSH6, lexA-MSH2 and GAD-MSH4, and lexA-MSH2 and GAD-MSH5 gave no activity even under conditions (long reaction time) where very weak interactions would be detected. These results indicate that at least one level of functional specificity of these proteins is mediated by hetero-oligomer formation. DISCUSSION
FIG. 4. Two-hybrid interactions between lexA-MSH4 and various alleles of GAD-MSH5. L40 was cotransformed with pPAS65 (lexA-MSH4) and either pPAS64 (GAD-MSH5), pPAS127 (GAD-msh5G648R), pDW10 (GAD-msh5-G643D, A644V), pPAS122 (GAD-msh5G821D) or pGAD424 (GAD). Panel A, transformants were streaked for single colonies on SD2his, 2trp, 2leu medium to assess prototrophy for histidine. Panel B, 10 mg of crude extract from each set of transformants was subjected to immunoblot analysis using anti-GAD antibodies. Lane 1, GAD-MSH5; lane 2, GAD-msh5-G648R; lane 3, GAD-msh5-G643D, A644V; lane 4, GAD-msh5-G821D.
Repair and the Meiotic Crossover MutS Homologs in Yeast— Given the fact that all five nuclear MSH genes in S. cerevisiae are known to function during meiosis and that both sets of proteins function as hetero-oligomers, it was of interest to determine whether hetero-oligomers can form between the two functionally distinct classes. Msh2p and Msh6p have been shown previously to interact by coimmunoprecipitation (2, 13). We reconstituted this interaction in the two-hybrid system
Given the fact that the MutS-MutL system of mismatch repair is highly conserved between prokaryotes and eukaryotes, it was initially surprising to discover two MutS homologs in yeast, MSH4 and MSH5, which have no role in mismatch repair. Instead, these genes are required to facilitate reciprocal crossovers between homologous chromosomes (16, 17), thereby providing the physical connection needed for proper meiosis I segregation (35). To resolve the apparent paradox between homology and function, it was proposed that Msh4p and Msh5p have evolved different DNA binding specificities so as to recognize recombination intermediates such as Holliday junctions instead of mismatches (16, 17). Alternatively, given the recent evidence that the mismatch repair protein Msh2p can itself bind to Holliday junctions (36), Msh4p and Msh5p may recognize a similar set of substrates but act to recruit different downstream components. The assumption underlying both of these hypotheses is that the general features characteristic of eukaryotic mismatch repair MutS proteins are conserved with the meiotic crossover family members. The work presented here validates this assumption. The first of three general features of the mismatch repair MutS homologs is that the functional units are comprised of heterodimers. Genetic epistasis data comparing mutants of MSH2, MSH3, and MSH6 suggested the hypothesis that the Msh2 protein forms heterodimers with either Msh3p or Msh6p to recognize different types of mismatches (6, 13). Subsequently, heterodimerization between Msh2p and either Msh3p or Msh6p (as well as their mammalian counterparts) has been shown either by coimmunoprecipitation or copurification (2–5, 7, 8, 13). One function of heterodimerization appears to be providing mismatch specificity to the DNA binding activity. This idea comes from the fact that the binding specificities of Msh2p alone compared with Msh2p complexed with Msh6p are different (2, 26), and it is the Msh2p-Msh6p complex that exhibits high affinity for the substrates expected from in vivo studies. In addition, genetic and biochemical analyses of mutant msh2 proteins complexed with Msh6p provide strong evidence that the complex is required to determine substrate specificity (14). Two independent assays presented here, coimmunoprecipitation and two-hybrid analysis, indicate that Msh4p and Msh5p physically interact. Although the possibility exists that this interaction is being bridged by a third protein,
Yeast MutS Protein Interactions it seems unlikely given that both assays used vegetatively growing cells, and MSH4 and MSH5 are usually only transcribed during meiosis (17).4 It has been shown recently that the Msh4 and Msh5 proteins colocalize at discrete foci on meiotic chromosomes,5 providing further evidence that these proteins function together in vivo. Although both the coimmunoprecipitation and two-hybrid assays indicate that Msh4p and Msh5p interact with each other in vivo, neither assay addresses the stoichiometry with which they interact. Based on analogy to the mismatch repair MutS proteins we assume that the hetero-oligomeric structure formed by Msh4p and Msh5p is a heterodimer, but this fact has not yet been demonstrated. The second general feature of the mismatch repair MutS homologs is that the ATP binding domain is essential for function. Mutagenesis of the p-loop of the ATP binding domain creates null alleles of the MutS genes in Escherichia coli and Salmonella typhimurium, as well as the MSH2 gene of yeast (14, 33, 34). When similar residues are changed in MSH5, function is lost without affecting the amount of msh5 protein in the cell. The p-loop mutations in S. typhimurium and MSH2 were shown in vitro to decrease ATPase activity, and the assumption is that the same activity is what is affected in the msh5 mutants, but this remains to be tested. Interestingly, Alani et al. (14) showed that certain changes in the putative helix-turn-helix domain of MSH2 have in vivo and in vitro phenotypes similar to the p-loop mutations. Making the identical change in MSH5 (G821D) as was made in MSH2 (G855D) creates a null allele of MSH5 as well. The third and final general feature is that mutations in the ATP binding domain do not affect hetero-oligomerization. Alani et al. (14) used coimmunoprecipitation and copurification to assay protein-protein interactions between Msh6p and various mutant Msh2 proteins. This work found that mutations in the p-loop and putative helix-turn-helix domains which reduced ATPase activity in vitro and created null alleles in vivo still allowed interaction with Msh6p. Using the two-hybrid system to analyze msh5 mutants in these domains (G643D, A644V; G648R; G821D) a similar finding was obtained; that is, although these mutations created null alleles of MSH5, there was little or no effect on the the ability of the mutant msh5 proteins to interact with Msh4p. This result suggests that, as with the Msh2p-Msh6p protein complex, ATP binding and hydrolysis are necessary for steps downstream of hetero-oligomerization such as creating a conformational change in the protein complex which allows other proteins to bind to the Msh4p-Msh5p complex (14, 33, 34). Recent evidence suggests that MLH1, an MutL homolog in yeast, is a good candidate for such a component (36). The meiotic crossover MutS proteins therefore share many attributes in common with the mismatch repair proteins. Previous studies using msh2 mutants suggested that the mismatch repair machinery may affect the length of conversion tracts perhaps by preventing the extension of heteroduplex formation after a mismatch is recognized. Consistent with this idea, Msh2p has recently been shown to bind to Holliday junctions in vitro (37). Because Holliday junctions are a potential substrate proposed for Msh4 and Msh5, the authors speculated that Msh2 may interact with Msh4 and/or Msh5 “resulting in the formation of complexes that have different abilities to recognize structural features of recombination intermediates and/or have altered abilities to interact with different recombination pathways.” The experiments presented here indicate
4 5
J. Engebrecht, personal communication. J. Novak and G. S. Roeder, personal communication.
30349
that such cross-hetero-oligomerization does not occur. The Msh2p-Msh6p interaction observed by others using biochemical techniques was reproduced using the two-hybrid system. It was therefore possible to make different combinations of mismatch repair and meiotic crossover MutS homologs and ask whether proteins from different functional classes could form a complex. No hetero-oligomerization was observed between Msh2p and either Msh4p or Msh5p; nor were any proteinprotein interactions detected between Msh6p and Msh4p. Therefore, the functional specificity observed between the two classes of MutS homologs is due in part to the specificity of hetero-oligomerization. An interesting question that remains to be answered is how the different types of hetero-oligomers mediate their functions. Does the Msh4p-Msh5p complex recognize different substrates compared with the mismatch repair MutS proteins (which seems likely), and/or does it recruit a different set of proteins to the DNA once bound? Understanding these differences may provide important insights into how similar DNA-binding proteins can evolve to perform different types of functions in a cell. Acknowledgments—We thank Loic Giot, Aaron Neiman, Neta Dean, and Eric Alani for valuable discussions. Aaron Neiman, Neta Dean, and Rolf Sternglanz provided helpful comments on the manuscript. Neta Dean, Petra Ross-MacDonald, Daniel Tishkoff, and Tom Triolo provided plasmids. REFERENCES 1. Modrich, P., and Lahue, R. (1996) Annu. Rev. Biochem. 65, 101–133 2. Alani, E. (1996) Mol. Cell. Biol. 16, 5604 –5615 3. Drummond, J. T., Li, G.-M., Longley, M. J., and Modrich, P. (1995) Science 268, 1909 –1911 4. Habraken, Y., Sung, P., Prakash, L., and Prakash, S. (1996) Curr. Biol. 6, 1185–1187 5. Iaccarino, I., Palombo, F., Drummond, J., Totty, N. F., Hsuan, J. J., Modrich, P., and Jiricny, J. (1996) Curr. Biol. 6, 484 – 486 6. Johnson, R. E., Kovvali, G. K., Prakash, L., and Prakash, S. (1996) J. Biol. Chem. 271, 7285–7288 7. Palombo, F., Gallinari, P., Iaccarino, I., Lettieri, T., Hughes, M., D’Arrigo, A., Truong, O., Hsuan, J. J., and Jiricny, J. (1995) Science 268, 1912–1914 8. Palombo, F., Iaccarino, I., Jakajima, E., Ikejima, M., Shimada, T., and Jiricny, J. (1996) Curr. Biol. 6, 1181–1184 9. Prolla, T. A., Pang, Q., Alani, E., Kolodner, R. D., and Liskay, R. M. (1994) Nature 265, 1091–1093 10. Crouse, G. F. (1997) in Mismatch Repair Systems in S. cerevisiae, Hoekstra, M. F., and Nickoloff, J. A., eds, Humana Press, Totowa, NJ 11. Chi, N.-W., and Kolodner, R. D. (1994) J. Biol. Chem. 269, 29984 –29992 12. Reenan, R. A. G., and Kolodner, R. D. (1992) Genetics 132, 975–985 13. Marsischky, G. T., Filosi, N., Kane, M. F., and Kolodner, R. (1996) Genes Dev. 10, 407– 420 14. Alani, E., Sokolsky, T., Studamire, B., Miret, J. J., and Lahue, R. S. (1997) Mol. Cell. Biol. 17, 2436 –2447 15. Grilley, M., Welsh, K. M., Su, S.-S., and Modrich, P. (1989) J. Biol. Chem. 264, 1000 –1004 16. Hollingsworth, N. M., Ponte, L., and Halsey, C. (1995) Genes Dev. 9, 1728 –1739 17. Ross-Macdonald, P., and Roeder, G. S. (1994) Cell 79, 1069 –1080 18. Hollenberg, S. M., Sternglanz, R., Cheng, P. F., and Weintraub, H. (1995) Mol. Cell. Biol. 15, 3813–3822 19. Ito, H., Fukada, Y., Murata, K., and Kimura, A. (1983) J. Bacteriol. 153, 163–168 20. Hollingsworth, N. M., and Johnson, A. D. (1993) Genetics 133, 785–797 21. Dean, N., and Pelham, H. R. B. (1990) J. Cell Biol. 111, 369 –377 22. Tyers, M., Tokiwa, G., Nash, R., and Futcher, B. (1992) EMBO J. 11, 1773–1784 23. Bartel, P. L., Chien, R., Sternglanz, R., and Fields, S. (1993) Using the Two-hybrid System to Detect Protein-Protein Interactions, 153–179, IRL Press, Oxford 24. Smith, M. (1985) Annu. Rev. Genet. 19, 423– 462 25. Hollingsworth, N. M., and Ponte, L. (1997) Genetics 147, 33– 42 26. Alani, E., Chi, N.-W., and Kolodner, R. (1995) Genes Dev. 9, 234 –247 27. Fields, S., and Sternglanz, R. (1994) Trends Genet. 10, 286 –291 28. Hollingsworth, N. M., Goetsch, L., and Byers, B. (1990) Cell 61, 73– 84 29. Smith, A. V., and Roeder, G. S. (1997) J. Cell Biol. 136, 957–967 30. Bell, S. P., Mitchell, J., Leber, J., Kobayshi, R., and Stillman, B. (1995) Cell 83, 563–568 31. Tu, J., Song, W., and Carlson, M. (1996) Mol. Cell. Biol. 16, 4199 – 4206 32. Triolo, T., and Sternglanz, R. (1996) Nature 381, 251–253 33. Haber, L. T., and Walker, G. C. (1991) EMBO J. 10, 2707–2715 34. Wu, T.-H., and Marinus, G. (1994) J. Bacteriol. 176, 5393–5400 35. Hawley, R. S. (1987) in Exchange and Chromosomal Segregation in Eukaryotes (Moens, P. B., ed) pp. 497–527, Academic Press, New York 36. Hunter, N., and Borts, R. H. (1997) Genes Dev. 11, 1573–1582 37. Alani, E., Lee, S., Kane, M. F., Griffith, J., and Kolodner, R. D. (1997) J. Mol. Biol. 265, 289 –301
Vol. 272, No. 48, Issue of November 28, pp. 30345–30349, 1997 Printed in U.S.A.
Conserved Properties between Functionally Distinct MutS Homologs in Yeast* (Received for publication, July 25, 1997, and in revised form, September 17, 1997)
Pascale Pochart, Dana Woltering, and Nancy M. Hollingsworth‡ From the Department of Biochemistry and Cell Biology, Institute for Cell and Developmental Biology, State University of New York at Stony Brook, Stony Brook, New York 11794-5215
In the yeast Saccharomyces cerevisiae there are five nuclear MutS homologs that act in two distinct processes. MSH2, 3, and 6 function in mismatch repair in both vegetative and meiotic cells, whereas MSH4 and MSH5 act specifically to facilitate crossovers between homologs during meiosis. Coimmunoprecipitation as well as two-hybrid experiments indicate that the Msh4 and Msh5 proteins form a hetero-oligomeric structure similar to what is observed for the Msh proteins involved in mismatch repair. Mutation of conserved amino acids in the NTP binding and putative helix-turnhelix domains of Msh5p abolish function but are still capable of interaction with Msh4p, suggesting that NTP binding plays a role downstream of hetero-oligomer formation. No hetero-oligomers are observed between the mismatch repair MutS proteins (Msh2p and Msh6p) and either Msh4p or Msh5p. These results indicate that one level of functional specificity between the mismatch repair and meiotic crossover MutS homologs in yeast is provided by the ability to form distinct hetero-oligomers.
Much attention has been focused in recent years on the MutS family of proteins whose role in mismatch repair has been implicated in preventing the development of human cancers (for review, see Ref. 1). Mutations in the genes encoding these proteins from bacteria to humans result in higher mutation rates and increased genomic instability. In addition, mismatch repair proteins play a role in preventing recombination from occurring between highly diverged sequences. MutS family members typically exhibit 20 –30% amino acid identity throughout the length of the protein with two more highly conserved subdomains, a putative NTP binding site and helixturn-helix structural domain, in the COOH-terminal half. Evidence from both in vitro and in vivo analyses has demonstrated that in eukaryotes, MutS proteins function as heterodimers to recognize and bind mismatched DNA base pairs that result from errors in DNA replication or from recombination, for example (2– 8). Once the MutS homolog heterodimer is bound to the DNA, a second heterodimer comprised of MutL homologs is recruited to the mismatch (9). In analogy to the MutS-MutL system in bacteria, an endonuclease is presumably activated to allow repair of the mismatch. The budding yeast Saccharomyces cerevisiae contains six MutS homologs (for review, see Ref. 10). Five of these genes * This work was supported by a grant from the Pew Charitable Trusts as well as National Institutes of Health Grant GM50717. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ‡ To whom correspondence should be addressed: Dept. of Biochemistry and Cell Biology, 314 Life Sciences Bldg., SUNY Stony Brook, Stony Brook, NY 11794-5215. Tel.: 516-632-8581; Fax: 516-632-8575; E-mail: [email protected]. This paper is available on line at http://www.jbc.org
(MSH2– 6) encode proteins that act on nuclear DNA, whereas one, MSH1, is required for stability of the mitochondrial genome (11, 12). Of the nuclear MutS homologs, three function to repair mismatches in both vegetatively growing and meiotic cells (10). Specificity of mismatch recognition has been shown to occur by the formation of distinct heterodimeric complexes. Msh2p interacts with Msh6p to recognize both single base pair mismatches as well as small loops formed by insertions or deletions in the DNA (2, 5, 13); in addition, Msh2p can complex with Msh3p to recognize insertion/deletion loops (4, 13). The Msh2p-Msh6p protein-protein interaction and mismatch binding specificity are independent of the ability to bind and hydrolyze ATP, suggesting that ATP may be required for downstream events such as recruitment of MutL homologs (14, 15). Despite being highly homologous to other members of the MutS protein family, the two remaining MutS homologs in yeast, MSH4 and MSH5, play no role in mismatch repair in either vegetative or meiotic cells. Instead their role is to facilitate crossovers between homologs to ensure that chromosomes segregate to opposite poles at the first meiotic division (16, 17). This second function for MutS homologs appears to be conserved through evolution as homologs of Msh4p and Msh5p have been found in both nematodes and mammals.1,2 Mutants in MSH4 and/or MSH5 display levels of meiotic crossing over and spore viability which are 30 –50% of wild type. These phenotypes are observed even in homothallic diploids where no mismatches are present, underscoring the idea that Msh4p and Msh5p, unlike the other MutS family members, do not need to recognize base pair mismatches to perform their function. Instead it has been proposed that the ability to recognize aberrant DNA structures has been modulated in these proteins through evolution to recognize recombination intermediates such as Holliday junctions (16, 17). Because MSH4 and MSH5 function in a separate process from MSH2, 3, and 6, the question exists as to whether Msh4p and Msh5p have any features in common with the mismatch repair MutS homologs. The finding that msh4 msh5 diploids exhibit the same level of spore viability as either single mutant (16) suggested that these two proteins might also function as heterodimers. The work presented here confirms that the Msh4 and Msh5 proteins interact, and it demonstrates as well that, similar to Msh2p-Msh6p, hetero-oligomerization between Msh4p and Msh5p can occur in the presence of mutations that presumably destroy the ability of Msh5p to bind and/or hydrolyze ATP. Finally, data are presented which show that proteinprotein interactions between the mismatch repair and meiotic crossover classes of MutS proteins do not occur, indicating that at least one level of functional specificity between the two classes is regulated by who can form heterodimers with whom.
1 2
30345
N. M. Hollingsworth, unpublished observations. R. Kolodner, personal communication.
30346
Yeast MutS Protein Interactions EXPERIMENTAL PROCEDURES
Yeast Strains—The following S. cerevisiae strains were used: NH141 MATa/MATa leu2/leu2 HIS4/his4 ura3/ura3 his3/HIS3 trp1/trp1 CAN1/can1 CYH2/cyh2 SPO13/spo13 ADE2/ade2; NH129, NH141 containing msh5::URA3/msh5::URA3; YPP10, NH141 containing msh5::LEU2/msh5::LEU2; YPP10-66-1, YPP10 with TPIp-MSH5-3HA; L40 (18). Isogenic diploids were derived by transformation (19) of the same haploid parents with the appropriate knockout fragment (msh5::URA3 (16); msh5::LEU2, pCH2– 6) followed by mating. YPP10 was transformed with pPAS66 linearized with XhoI to integrate TPIpMSH5-3HA at URA3. Yeast media have been described (20). Plasmids—pCH2– 6 contains LEU2 inserted into the BglII site of MSH5 in pNH181 (16). The msh5::LEU2 fragment is released by cutting with HindIII and SacI. The MSH5-3HA fusion gene under control of the TPI promoter was constructed in several steps. First, a vector containing the TPI promoter and a multiple cloning site (pJAY-2) was created by digestion of pTiaO (21) (provided by N. Dean) with SacI and religation. Then the MSH5 gene alone was placed downstream of the TPI promoter by cloning the 4.2-kilobase HpaI/HindIII MSH5 fragment from pNH189 (16) into pJAY-2 cut first with EcoRI and blunt ended by T4 DNA polymerase and then HindIII to make pPAS29. To create a tagged allele of MSH5, a NotI site was first substituted for the stop codon in MSH5 by polymerase chain reaction. A polymerase chain reaction fragment containing the triple HA3 tag flanked by NotI sites was amplified using pGTEP (22) as a template, cut with NotI, and ligated in-frame to the end of MSH5. A 200-base pair EcoRI/HindIII fragment containing this fusion was then substituted for the equivalent fragment in pPAS29 to generate pPAS66. pPAS65 (lexA-MSH4), pPAS58 (GAD-MSH4), pPAS64 (GADMSH5), and pPAS111 (GAD-MSH6) were made by polymerase chain reaction using Vent polymerase (New England Biolabs) to amplify the various genes. The resulting fragments were then ligated into either pBTM116 or pGAD424 (23). Site-directed mutagenesis (24) was used to introduce specific nucleotide changes in the MSH5 gene carried in pNH189. 1.2-kilobase BglII/EcoRI fragments containing the mutations were exchanged with the equivalent fragment in pPAS64 to generate pDW10 (GAD-msh5-G643D, A644V), pPAS122 (GAD-msh5-G821D), and pPAS127 (GAD-msh5-G648R). For complementation, the 1.2-kilobase BglII/EcoRI fragments were substituted in pPAS79, which carries the MSH5 gene under its own promoter, to create pPAS135, pPAS136, and pPAS137, respectively. All three 1.2-kilobase fragments containing the mutations were sequenced to ensure that no additional mutations were introduced during the mutagenesis (data not shown). pNH108 is described in Ref. 25. 2mMSH4, pRDK371, and pORC15–268 were provided by P. Ross-MacDonald, D. Tishkoff, and T. Triolo, respectively. Immunoprecipitation—Immunoprecipitation was performed essentially as described in Ref. 26 using 100 ml of YPP10-66-1 carrying either pBTM116 or pPAS65 grown in SD-Trp to an A600 of 0.8. The proteins were fractionated by SDS-polyacrylamide gel electrophoresis, and immunoblots were probed using anti-lexA antibodies (CLONTECH) or the anti-HA 12CA5 ascites fluid (Babco). The presence of the antibody was detected using ECL (Amersham). Two-hybrid Assay—Relevant plasmids were cotransformed into L40 selecting on SD-Trp,-Leu medium. The presence of the fusion proteins was confirmed by immunoblot analysis using anti-lexA or anti-GAD antibodies (CLONTECH) (data not shown). To rule out potential artifacts created by polymerase chain reaction, several independent clones for each plasmid were tested. In addition, only fusion proteins that were known to give a positive signal in the two-hybrid system in combination with at least one other protein were used. Histidine prototrophy was assayed on SD-His plates, and b-galactosidase activity was monitored as described (25). RESULTS
Msh4p and Msh5p Physically Interact in Vivo—Protein-protein interactions between Msh4p and Msh5p were tested by two different methods. The first assay involved coimmunoprecipitation of Msh4p and Msh5p after ectopic expression in vegetative yeast cells. The MSH4 gene was fused to the 39-end of lexA under control of the constitutive ADH1 promoter, thereby allowing detection of lexA-Msh4p using anti-lexA antibodies. The MSH5 gene was tagged by the addition of three 3 The abbreviations used are: HA, hemagglutinin; GAD, Gal4p activation domain.
FIG. 1. Coimmunoprecipitation of MsH5-3HAp and lexAMsh4p. Crude extracts from YPP10-66-1 (MSH5-3HA) transformed with either lexA (pBTM116) or lexA-MSH4 (pPAS65) were incubated with anti-HA antibodies (12CA5) and the antibodies precipitated with protein A-Sepharose. The precipitated complexes were collected by centrifugation, and the proteins were fractionated by SDS-polyacrylamide gel electrophoresis. After transfer to nitrocellulose, the proteins were probed with either 12CA5 antibodies to detect MsH5-3HAp (panel A) or with anti-lexA antibodies to detect lexA-Msh4p (panel B). The asterisk (*) indicates an anti-HA cross-reacting band.
HA epitopes at the end of the MSH5 coding sequence and integrated under control of the TPI promoter in a msh5::LEU2/msh5::LEU2 diploid to create YPP10-66-1 (see “Experimental Procedures”). The Msh5-3HAp was observed using the 12CA5 antibody directed against the HA epitope. Both the lexA-MSH4 and the MSH5-3HA fusion genes under control of the heterologous promoters were able to complement the spore viability defect of msh4 and msh5 diploids, respectively (data not shown). Plasmids containing lexA alone (pBTM116) or lexA-MSH4 (pPAS65-2) were transformed into YPP10-66-1. Cell extracts derived from these strains were incubated with the 12CA5 antibody. The antibodies were precipitated by the addition of protein A-Sepharose, and the proteins were fractionated by SDS-polyacrylamide gel electrophoresis and transferred to nitrocellulose. When the immunoblots were probed with the anti-HA antibodies, the Msh5-3HA protein was found to precipitate with the beads (Fig. 1A, lanes 1 and 2). When the immunoprecipitates were probed with lexA antibodies, a band migrating with the expected molecular mass for lexA-Msh4p (123 kDa) was observed to coimmunoprecipitate (Fig. 1B, lane 6). This band is not seen when lexA alone is present (Fig. 1B, lane 4), nor when the 12CA5 antibodies are omitted (Fig. 1B, lanes 3 and 5), suggesting a specific physical interaction between Msh4p and Msh5p. The second assay utilized the two-hybrid system, a genetic method for detecting protein-protein interactions that occur in vegetatively growing yeast cells (27). This method utilizes a strain, L40, containing two reporter genes, lacZ and HIS3, which have lexA operator sites present in their promoters (18). When L40 is transformed with pPAS65-2, lexA-Msh4p is produced and presumably bound upstream of each reporter by virtue of the interaction between the lexA DNA binding domain and its binding sites. Introduction of the Gal4p activation domain (GAD) can result in activation of the reporter constructs when the Gadp is fused to a protein that interacts with lexAMsh4p. Activation of HIS3 transcription makes the cells prototrophic for histidine, whereas activation of the lacZ gene can
Yeast MutS Protein Interactions
FIG. 2. Two-hybrid interactions between lexA-MSH4 and various GAD fusions. L40 was cotransformed with pPAS65 (lexA-MSH4) and either pGAD424 (GAD), pPAS58 (GAD-MSH4), pPAS64 (GADMSH5), GAD-RED1537– 827, pNH108 (GAD-HOP1), or GAD-ORC15–268. Transformants were streaked for single colonies on SD2his, 2trp, 2leu medium to assess prototrophy for histidine.
be observed using an enzymatic assay for b-galactosidase activity. The combination of lexA-MSH4 and GAD-MSH5 produces both histidine prototrophy (Fig. 2) as well as substantial amounts of b-galactosidase activity (data not shown), indicating that lexA-Msh4p and Gad-Msh5p interact. This interaction is specific as no signal is observed when lex-MSH4 is combined either with GAD or with GAD fusions to HOP1, RED1537– 827, or ORC15–268 (Fig. 2). Like MSH4, HOP1 and RED1 encode meiosis-specific proteins that are localized on meiotic chromosomes in S. cerevisiae (17, 28, 29). ORC1 encodes part of a protein complex bound at origins of replication in yeast (30). These fusions have all been observed to interact with other proteins in the two-hybrid system and therefore are known to function in this assay (25, 31, 32). The interaction between lexA-Msh4p and Gad-Msh5p is not being mediated by lexA, as no interaction is detected with the lexA and GAD-MSH5 combination (data not shown). A positive signal was obtained with the combination of lexAMSH4 and GAD-MSH4, indicating that Msh4p can form homooligomers in vivo (Fig. 2). Whether the homo-oligomer has a function in the cell is unclear, however. The presence of MSH4 on a high copy number plasmid is unable to suppress the msh5 spore viability defect (data not shown). Furthermore, the msh4 phenotype is no more severe than msh5 (16), contrary to what one might expect if Msh4p homo-oligomers effect some function distinct from the Msh4p-Msh5p hetero-oligomer. Mutation of the Putative NTP Binding Domain of MSH5 Abolishes Function but Still Allows Interaction with MSH4 — Mutations in the NTP binding domain of both yeast MSH2 and bacterial MutS reduce ATPase activity and create noncomplementing alleles that are dominant negative when overexpressed in a wild type background (14, 33, 34). Alani et al. (14) further showed that changing a completely conserved glycine in the p-loop of the ATP binding domain of Msh2p to aspartic acid (position 693) does not affect the ability of Msh2p to interact with Msh6p, nor does it affect DNA mismatch binding specificity. To test whether mutations in the putative NTP binding domain of MSH5 have similar properties, two sitedirected mutant alleles of MSH5 were generated. msh5-G648R substitutes an arginine for the glycine at position 648, which corresponds to glycine 693 in the p-loop of Msh2p (Fig. 3). The second allele, msh5-G643D, A644V, changes the absolutely conserved glycine at position 643 of the p-loop to aspartic acid as well as making a conservative change of alanine to valine at
30347
FIG. 3. Alignment of two conserved domains in MutS and the yeast nuclear MutS homologs. Numbers represent the positions of the first amino acid in the line. Bold letters indicate amino acids conserved in all six proteins. Bold letters below the Msh5p sequence indicate amino acids substitutions in Msh5p. Panel A, p-loop of the nucleotide binding domain. Panel B, putative helix-turn-helix domain.
amino acid 644. Wu and Marinus (34) have shown previously that mutation of the corresponding glycine in MutS (position 614) creates a nonfunctional allele. The mutant alleles were then either placed under the MSH5 promoter and assayed for complementation, or they were fused to GAD and assayed for their ability to interact with lexA-MSH4. The msh5-G643D, A644V, and msh5-G648R alleles were tested for complementation of the msh5 spore inviability defect by first transforming the msh5::URA3/msh5::URA3 diploid NH129 with the plasmids pPAS135 and pPAS137, respectively. The transformants were then sporulated and asci dissected to determine the fraction of viable spores. Both mutant alleles failed to complement and appear to be null (Table I). The msh5-G643D, A644V allele was also tested for a dominant negative effect in a MSH5/msh5 diploid, and no phenotype was observed (data not shown). This finding is most likely due to a failure to overexpress the mutant protein to a sufficient level. Both GAD-msh5-G643D, A644V and GAD-msh5-G648R interact with lexA-MSH4 in the two-hybrid assay. For GADmsh5-G643D, A644V, the amount of both HIS3 (as measured by colony size on SD2His medium) and lacZ activity observed was less than that found for GAD-MSH5 (Fig. 4A; data not shown), even though more of the mutant protein is made compared with Gad-Msh5p (Fig. 4B, compare lanes 1 and 3). The Gad-msh5-G648R protein interacted slightly less well than Gad-Msh5p, but in this case the difference may be because slightly less of the mutant protein was made (Fig. 4B, lane 2). Alani et al. (14) discovered that mutation of some of the amino acids in the helix-turn-helix domain of MSH2 gives phenotypes similar to those in the G693D mutation in the ATP binding domain. For example, substituting aspartic acid for the glycine at position 855 of Msh2p (Fig. 3B) results in a protein that fails to complement a deletion of MSH2 and has reduced ATPase activity yet still interacts with Msh6p. This glycine is conserved in all of the yeast nuclear MutS homologs, as well as in MutS itself (Fig. 3B). The corresponding glycine codon in MSH5 was therefore mutated to aspartic acid thereby creating msh5-G821D. This mutation creates a null allele of MSH5 (Table I). In combination with lexA-MSH4, the GAD-msh5G821D gene shows a wild type level of histidine prototrophy (Fig. 4A) and b-galactosidase activity (data not shown) in the two-hybrid assay, indicating that these proteins interact. It therefore appears that, similar to the mismatch repair MutS proteins, ATP binding and hydrolysis are essential steps that occur downstream of hetero-oligomer formation in the meiotic crossover MutS homologs. Hetero-oligomers Are Not Observed between the Mismatch
30348
Yeast MutS Protein Interactions
TABLE I Complementation analysis of site-directed msh5 alleles Transformants of NH129 with the appropriate plasmids were selected on SD2leu medium. Several transformants were patched onto SD2leu medium, grown overnight, and replica plated to spo plates to induce sporulation. Incubation was at 30 °C for 3 days. The resulting asci were dissected and incubated at 30 °C for several days to determine the fraction of spores able to form colonies. Strain/plasmid a
NH129/YCplac111 NIH129/pPAS79 NH129/pPAS135 NH129/pPAS137 NH129/pPAS36 a
Relevant genotype
% Spore viability (asci)
msh5::URA3 msh5::URA3 msh5::URA3/MSH5 msh5::URA3 msh5::URA3/msh5-G643D,A644V msh5::URA3 msh5::URA3/msh5-G648R msh5::URA3 msh5::URA3/msh5-G821D msh5::URA3
40.1 (159) 80.8 (52) 44.3 (75) 49.3 (68) 48.8 (71)
Includes data from NH129 dissected without vector.
TABLE II Two hybrid assays using mismatch repair and meiotic crossover MutS homologs in yeast L40 was transformed with the appropriate plasmids, and filter assays were performed as described under “Experimental Procedures.” —, no signal after 4 h at 30 °C; 11, blue color observed after 90 min at 30 °C; 111, blue color observed after 30 min at 30 °C. Plasmids
pPAS65/pGAD424 pPAS65/pPAS64 pPAS65/pPAS111 pRDK371/pPAS111 pRDK371/pPAS58 pRDK371/pPAS58
lex-A
GAD-
b-Galactosidase activity
MSH4 MSH4 MSH4 MSH2 MSH2 MSH2
— MSH5 MSH6 MSH6 MSH4 MSH4
— 111 — 11 — —
using a lexA-MSH2 plasmid and GAD-MSH6 (see “Experimental Procedures”). Because the selectable marker on the lexAMSH2 plasmid is HIS3, only activation of the lacZ reporter was assayed (Table II). Cross-combinations of lexA-MSH4 and GAD-MSH6, lexA-MSH2 and GAD-MSH4, and lexA-MSH2 and GAD-MSH5 gave no activity even under conditions (long reaction time) where very weak interactions would be detected. These results indicate that at least one level of functional specificity of these proteins is mediated by hetero-oligomer formation. DISCUSSION
FIG. 4. Two-hybrid interactions between lexA-MSH4 and various alleles of GAD-MSH5. L40 was cotransformed with pPAS65 (lexA-MSH4) and either pPAS64 (GAD-MSH5), pPAS127 (GAD-msh5G648R), pDW10 (GAD-msh5-G643D, A644V), pPAS122 (GAD-msh5G821D) or pGAD424 (GAD). Panel A, transformants were streaked for single colonies on SD2his, 2trp, 2leu medium to assess prototrophy for histidine. Panel B, 10 mg of crude extract from each set of transformants was subjected to immunoblot analysis using anti-GAD antibodies. Lane 1, GAD-MSH5; lane 2, GAD-msh5-G648R; lane 3, GAD-msh5-G643D, A644V; lane 4, GAD-msh5-G821D.
Repair and the Meiotic Crossover MutS Homologs in Yeast— Given the fact that all five nuclear MSH genes in S. cerevisiae are known to function during meiosis and that both sets of proteins function as hetero-oligomers, it was of interest to determine whether hetero-oligomers can form between the two functionally distinct classes. Msh2p and Msh6p have been shown previously to interact by coimmunoprecipitation (2, 13). We reconstituted this interaction in the two-hybrid system
Given the fact that the MutS-MutL system of mismatch repair is highly conserved between prokaryotes and eukaryotes, it was initially surprising to discover two MutS homologs in yeast, MSH4 and MSH5, which have no role in mismatch repair. Instead, these genes are required to facilitate reciprocal crossovers between homologous chromosomes (16, 17), thereby providing the physical connection needed for proper meiosis I segregation (35). To resolve the apparent paradox between homology and function, it was proposed that Msh4p and Msh5p have evolved different DNA binding specificities so as to recognize recombination intermediates such as Holliday junctions instead of mismatches (16, 17). Alternatively, given the recent evidence that the mismatch repair protein Msh2p can itself bind to Holliday junctions (36), Msh4p and Msh5p may recognize a similar set of substrates but act to recruit different downstream components. The assumption underlying both of these hypotheses is that the general features characteristic of eukaryotic mismatch repair MutS proteins are conserved with the meiotic crossover family members. The work presented here validates this assumption. The first of three general features of the mismatch repair MutS homologs is that the functional units are comprised of heterodimers. Genetic epistasis data comparing mutants of MSH2, MSH3, and MSH6 suggested the hypothesis that the Msh2 protein forms heterodimers with either Msh3p or Msh6p to recognize different types of mismatches (6, 13). Subsequently, heterodimerization between Msh2p and either Msh3p or Msh6p (as well as their mammalian counterparts) has been shown either by coimmunoprecipitation or copurification (2–5, 7, 8, 13). One function of heterodimerization appears to be providing mismatch specificity to the DNA binding activity. This idea comes from the fact that the binding specificities of Msh2p alone compared with Msh2p complexed with Msh6p are different (2, 26), and it is the Msh2p-Msh6p complex that exhibits high affinity for the substrates expected from in vivo studies. In addition, genetic and biochemical analyses of mutant msh2 proteins complexed with Msh6p provide strong evidence that the complex is required to determine substrate specificity (14). Two independent assays presented here, coimmunoprecipitation and two-hybrid analysis, indicate that Msh4p and Msh5p physically interact. Although the possibility exists that this interaction is being bridged by a third protein,
Yeast MutS Protein Interactions it seems unlikely given that both assays used vegetatively growing cells, and MSH4 and MSH5 are usually only transcribed during meiosis (17).4 It has been shown recently that the Msh4 and Msh5 proteins colocalize at discrete foci on meiotic chromosomes,5 providing further evidence that these proteins function together in vivo. Although both the coimmunoprecipitation and two-hybrid assays indicate that Msh4p and Msh5p interact with each other in vivo, neither assay addresses the stoichiometry with which they interact. Based on analogy to the mismatch repair MutS proteins we assume that the hetero-oligomeric structure formed by Msh4p and Msh5p is a heterodimer, but this fact has not yet been demonstrated. The second general feature of the mismatch repair MutS homologs is that the ATP binding domain is essential for function. Mutagenesis of the p-loop of the ATP binding domain creates null alleles of the MutS genes in Escherichia coli and Salmonella typhimurium, as well as the MSH2 gene of yeast (14, 33, 34). When similar residues are changed in MSH5, function is lost without affecting the amount of msh5 protein in the cell. The p-loop mutations in S. typhimurium and MSH2 were shown in vitro to decrease ATPase activity, and the assumption is that the same activity is what is affected in the msh5 mutants, but this remains to be tested. Interestingly, Alani et al. (14) showed that certain changes in the putative helix-turn-helix domain of MSH2 have in vivo and in vitro phenotypes similar to the p-loop mutations. Making the identical change in MSH5 (G821D) as was made in MSH2 (G855D) creates a null allele of MSH5 as well. The third and final general feature is that mutations in the ATP binding domain do not affect hetero-oligomerization. Alani et al. (14) used coimmunoprecipitation and copurification to assay protein-protein interactions between Msh6p and various mutant Msh2 proteins. This work found that mutations in the p-loop and putative helix-turn-helix domains which reduced ATPase activity in vitro and created null alleles in vivo still allowed interaction with Msh6p. Using the two-hybrid system to analyze msh5 mutants in these domains (G643D, A644V; G648R; G821D) a similar finding was obtained; that is, although these mutations created null alleles of MSH5, there was little or no effect on the the ability of the mutant msh5 proteins to interact with Msh4p. This result suggests that, as with the Msh2p-Msh6p protein complex, ATP binding and hydrolysis are necessary for steps downstream of hetero-oligomerization such as creating a conformational change in the protein complex which allows other proteins to bind to the Msh4p-Msh5p complex (14, 33, 34). Recent evidence suggests that MLH1, an MutL homolog in yeast, is a good candidate for such a component (36). The meiotic crossover MutS proteins therefore share many attributes in common with the mismatch repair proteins. Previous studies using msh2 mutants suggested that the mismatch repair machinery may affect the length of conversion tracts perhaps by preventing the extension of heteroduplex formation after a mismatch is recognized. Consistent with this idea, Msh2p has recently been shown to bind to Holliday junctions in vitro (37). Because Holliday junctions are a potential substrate proposed for Msh4 and Msh5, the authors speculated that Msh2 may interact with Msh4 and/or Msh5 “resulting in the formation of complexes that have different abilities to recognize structural features of recombination intermediates and/or have altered abilities to interact with different recombination pathways.” The experiments presented here indicate
4 5
J. Engebrecht, personal communication. J. Novak and G. S. Roeder, personal communication.
30349
that such cross-hetero-oligomerization does not occur. The Msh2p-Msh6p interaction observed by others using biochemical techniques was reproduced using the two-hybrid system. It was therefore possible to make different combinations of mismatch repair and meiotic crossover MutS homologs and ask whether proteins from different functional classes could form a complex. No hetero-oligomerization was observed between Msh2p and either Msh4p or Msh5p; nor were any proteinprotein interactions detected between Msh6p and Msh4p. Therefore, the functional specificity observed between the two classes of MutS homologs is due in part to the specificity of hetero-oligomerization. An interesting question that remains to be answered is how the different types of hetero-oligomers mediate their functions. Does the Msh4p-Msh5p complex recognize different substrates compared with the mismatch repair MutS proteins (which seems likely), and/or does it recruit a different set of proteins to the DNA once bound? Understanding these differences may provide important insights into how similar DNA-binding proteins can evolve to perform different types of functions in a cell. Acknowledgments—We thank Loic Giot, Aaron Neiman, Neta Dean, and Eric Alani for valuable discussions. Aaron Neiman, Neta Dean, and Rolf Sternglanz provided helpful comments on the manuscript. Neta Dean, Petra Ross-MacDonald, Daniel Tishkoff, and Tom Triolo provided plasmids. REFERENCES 1. Modrich, P., and Lahue, R. (1996) Annu. Rev. Biochem. 65, 101–133 2. Alani, E. (1996) Mol. Cell. Biol. 16, 5604 –5615 3. Drummond, J. T., Li, G.-M., Longley, M. J., and Modrich, P. (1995) Science 268, 1909 –1911 4. Habraken, Y., Sung, P., Prakash, L., and Prakash, S. (1996) Curr. Biol. 6, 1185–1187 5. Iaccarino, I., Palombo, F., Drummond, J., Totty, N. F., Hsuan, J. J., Modrich, P., and Jiricny, J. (1996) Curr. Biol. 6, 484 – 486 6. Johnson, R. E., Kovvali, G. K., Prakash, L., and Prakash, S. (1996) J. Biol. Chem. 271, 7285–7288 7. Palombo, F., Gallinari, P., Iaccarino, I., Lettieri, T., Hughes, M., D’Arrigo, A., Truong, O., Hsuan, J. J., and Jiricny, J. (1995) Science 268, 1912–1914 8. Palombo, F., Iaccarino, I., Jakajima, E., Ikejima, M., Shimada, T., and Jiricny, J. (1996) Curr. Biol. 6, 1181–1184 9. Prolla, T. A., Pang, Q., Alani, E., Kolodner, R. D., and Liskay, R. M. (1994) Nature 265, 1091–1093 10. Crouse, G. F. (1997) in Mismatch Repair Systems in S. cerevisiae, Hoekstra, M. F., and Nickoloff, J. A., eds, Humana Press, Totowa, NJ 11. Chi, N.-W., and Kolodner, R. D. (1994) J. Biol. Chem. 269, 29984 –29992 12. Reenan, R. A. G., and Kolodner, R. D. (1992) Genetics 132, 975–985 13. Marsischky, G. T., Filosi, N., Kane, M. F., and Kolodner, R. (1996) Genes Dev. 10, 407– 420 14. Alani, E., Sokolsky, T., Studamire, B., Miret, J. J., and Lahue, R. S. (1997) Mol. Cell. Biol. 17, 2436 –2447 15. Grilley, M., Welsh, K. M., Su, S.-S., and Modrich, P. (1989) J. Biol. Chem. 264, 1000 –1004 16. Hollingsworth, N. M., Ponte, L., and Halsey, C. (1995) Genes Dev. 9, 1728 –1739 17. Ross-Macdonald, P., and Roeder, G. S. (1994) Cell 79, 1069 –1080 18. Hollenberg, S. M., Sternglanz, R., Cheng, P. F., and Weintraub, H. (1995) Mol. Cell. Biol. 15, 3813–3822 19. Ito, H., Fukada, Y., Murata, K., and Kimura, A. (1983) J. Bacteriol. 153, 163–168 20. Hollingsworth, N. M., and Johnson, A. D. (1993) Genetics 133, 785–797 21. Dean, N., and Pelham, H. R. B. (1990) J. Cell Biol. 111, 369 –377 22. Tyers, M., Tokiwa, G., Nash, R., and Futcher, B. (1992) EMBO J. 11, 1773–1784 23. Bartel, P. L., Chien, R., Sternglanz, R., and Fields, S. (1993) Using the Two-hybrid System to Detect Protein-Protein Interactions, 153–179, IRL Press, Oxford 24. Smith, M. (1985) Annu. Rev. Genet. 19, 423– 462 25. Hollingsworth, N. M., and Ponte, L. (1997) Genetics 147, 33– 42 26. Alani, E., Chi, N.-W., and Kolodner, R. (1995) Genes Dev. 9, 234 –247 27. Fields, S., and Sternglanz, R. (1994) Trends Genet. 10, 286 –291 28. Hollingsworth, N. M., Goetsch, L., and Byers, B. (1990) Cell 61, 73– 84 29. Smith, A. V., and Roeder, G. S. (1997) J. Cell Biol. 136, 957–967 30. Bell, S. P., Mitchell, J., Leber, J., Kobayshi, R., and Stillman, B. (1995) Cell 83, 563–568 31. Tu, J., Song, W., and Carlson, M. (1996) Mol. Cell. Biol. 16, 4199 – 4206 32. Triolo, T., and Sternglanz, R. (1996) Nature 381, 251–253 33. Haber, L. T., and Walker, G. C. (1991) EMBO J. 10, 2707–2715 34. Wu, T.-H., and Marinus, G. (1994) J. Bacteriol. 176, 5393–5400 35. Hawley, R. S. (1987) in Exchange and Chromosomal Segregation in Eukaryotes (Moens, P. B., ed) pp. 497–527, Academic Press, New York 36. Hunter, N., and Borts, R. H. (1997) Genes Dev. 11, 1573–1582 37. Alani, E., Lee, S., Kane, M. F., Griffith, J., and Kolodner, R. D. (1997) J. Mol. Biol. 265, 289 –301
