INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The process of DNA mismatch repair (MMR) has been the focus of intense study since human MMR gene mutations were implicated in hereditary and sporadic forms of human cancer (13, 21, 39, 52). A primary role of MMR is to correct base-base mismatches and insertion-deletion loops (IDLs) resulting from DNA replication, endogenous or exogenous sources of DNA damage, and recombination (14, 36, 38, 43). In Escherichia coli, where MMR has been reconstituted in vitro by using purified proteins, a homodimer of MutS binds the mismatch, followed by the formation of an ATP-binding-dependent ternary complex with a homodimer of MutL. The latent endonuclease activity of MutH is stimulated by the ATP-dependent MutS-MutL ternary complex to nick the undermethylated strand at the nearest hemimethylated GATC site. After incision, UvrD helicase and four exonucleases excise the nascent strand some distance past the mismatch (68). The resultant single-strand gap (up to 1 kb) is filled in by DNA polymerase III, and the nick is sealed by DNA ligase (43).

In the yeast Saccharomyces cerevisiae, the mutation avoidance functions of MMR involve multiple MutS and MutL heterodimers with partially overlapping functions (36, 38). For example, a heterodimer of Msh2p and Msh6p, MutS, appears to be involved primarily in correcting mismatches and +1 IDL heterologies (2, 3, 12, 19, 23, 32, 33, 37, 41, 42, 48, 64), whereas, an Msh2p-Msh3p heterodimer, MutS, functions in correction of IDLs with 1 to 14 bases (23, 37, 41, 49). The major MMR MutL activity, MutL, in yeast is a heterodimer of Mlh1p and Pms1p (27, 51, 58, 59). An additional yeast MutL activity, MutL, comprised of Mlh1p and Mlh3p, appears to act in conjunction with MutS to correct a small fraction of IDLs (22).

An important clue to a possible biochemical activity of the MutL homologues was the appreciation of sequence similarity between the highly conserved NH₂ termini of the MutLs and a new family of ATPases (11, 44). The so-called GHL ATPase family is comprised of E. coli gyrase b subunit, the Hsp90 homologues, and the MutL homologues (10, 20). The supercoiling activity of E. coli DNA gyrase is dependent on the ATPase activity of the homodimeric gyrase b subunits (65). Recently, the homodimer Hsp90 has been demonstrated to have a weak intrinsic ATPase activity required for Hsp90 function (47, 50). The crystal structures of the NH₂ termini of Hsp90 and gyrase b revealed strong structural similarity within their ATPase motifs (55, 56, 71). In addition, Hsp90 and gyrase b appear to have similar ATPase cycles, which include functionally important NH₂-terminal conformational changes (4, 25, 26, 55, 56, 71). The NH₂-terminal conformational changes for gyrase b have been associated with dimerization of the NH₂-terminal domains in the ATP-bound form (4, 55, 56, 71). Recently, the crystal structure of an NH₂-terminal fragment of MutL was solved and demonstrated that MutL possesses an ATP-binding pocket homologous to the gyrase b and Hsp90 proteins. In addition, MutL appears to have the ATPase-dependent NH₂-terminal dimerization cycle found in the other GHL family member. Interestingly, Ban et al. reported that the NH₂-terminal-dimerized, ATP-bound form of MutL could activate the MutH endonuclease in a MutS-independent manner (9, 10).

Our previous studies have shown the importance of the NH₂ terminus of yeast Mlh1p and Pms1p in MMR (51). The abovementioned findings for the GHL family of proteins now present a working paradigm for detailed studies of the ATPase motifs found in the eukaryotic MutL homologues. In this report, we investigate the function of predicted ATPase motifs in S. cerevisiae MutL (Mlh1p-Pms1p). Our results suggest that yeast MutL has structural and functional properties consistent with other members of the GHL family of ATPases. Specifically, genetic results suggest that the ATPase motifs of both Mlh1p and Pms1p are absolutely required for MMR in vivo. In addition, biochemical and in vivo findings suggest that ATP binding induces conformational changes in MutL that are associated with heterodimerization between the NH₂ termini of Mlh1p and Pms1p. Surprisingly, our genetic results suggest differential requirements for Mlh1p and Pms1p ATPase motifs during MMR.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Strains and media. E. coli strains DH5 and DH-10B were used for plasmid construction and amplification. E. coli MAX Efficiency DH-10Bac [F mcrA D(mrr-hsdRMS-mcrBC) 80dlacZDM15 lacX74 deoR recA1 endA1 araD139 (ara, leu)7697 galU galK rpsL nupG/bMON14272/pMON7124] was used to produce recombinant baculoviruses as described below under "Expression and purification of yeast MutL." The S. cerevisiae strains used in this study are described in Table 1. Bacterial and yeast strains were grown under conditions described previously (51). Yeast transformations were performed by the polyethylene glycol-lithium acetate method (24).

mlh3

Plasmid construction. All DNA manipulations were performed by standard molecular biology procedures (40). Automated DNA sequencing was done at the Vollum Institute core sequencing facilities with an ABI automated sequencer.

(i) Targeting vectors. pYI-mlh1-31 was created as follows. The MLH1 open reading frame (ORF) and approximately 800 bp of upstream sequence were cloned into pYI-lacZ. The E31A mutation was generated in the resultant construct by using the Quikchange Site-Directed Mutagenesis kit (Stratagene) and the following oligonucleotides: ymlh1-31.S and ymlh1-31.AS (5'-GGAATTCTCCATCATAGCTTTGAGAGCATTTACGGG-3'). The desired mutations were detected by sequencing with oligonucleotide ymlh1-98.AS. An approximately 400-bp KpnI fragment containing the E31A mutation was cloned back into the parental construct to erase the potential for second site mutations elsewhere in the construct. pYI-mlh1-98 was created in a similar fashion, except the following oligonucleotides were used instead: ymlh1-98.S (5'-CAGATTCAAACGTATGGATTCCGAGCTGAAGCTTTAGCC-3') and ymlh1-98.AS for mutagenesis and ymlh1-31.S for identification of the point mutation.

(ii) Two-hybrid vectors. All of the following constructs were sequenced to confirm that point mutations were present in the desired plasmids. pNBTM116 was a generous gift of Stanley Hollenberg (Oregon Health Sciences University, Department of Cell and Developmental Biology) and allows construction of two-hybrid "bait" fusions with the lexA DNA binding domain fused at the carboxy terminus of the bait protein. pNBTM-mlh1 N-354 was engineered with the oligonucleotides ymlh1 N-anchor.S (5'-CGGGATCCATGTCTCTCAGAATAAAAGCAC-3') and ymlh1 N-354.AS (5'-AGCCTCGAGCTCTGGCTTGTTTGTTGAAATTG-3') to generate a PCR amplicon that was cloned into pNBTM116 at the BamHI and XhoI sites. Plasmid pNBTM-mlh1-31 N-354 was generated in an identical fashion, but the PCR was performed on template DNA that contained the E31A codon mutation.

(iii) Baculovirus plasmids. The 6×His-MLH1 recombinant baculovirus was constructed as follows. A PCR product was generated that engineered a 6×His affinity tag in frame with the MLH1 ORF after the initiator methionine. This 6×His-encoding PCR product was cloned into the BamHI and NdeI restriction sites of pBTM-MLH1, replacing approximately 360 bp of the native gene. Automated sequencing of the construct confirmed that the 6×His tag was in frame with the MLH1 ORF and that no PCR-generated mutations arose. The 6×His-MLH1 ORF was then cloned into pFastBac DUAL (pFBD) (Life Technologies) by using polylinker sites BamHI and SalI. Based upon mutator assays, the 6×His epitope-tagged Mlh1p functionally complemented an mlh1 strain (data not shown).

Two-hybrid analysis and -galactosidase assays. Protein-protein interactions were assessed by the two-hybrid technique. Bait and prey plasmids were transformed into L40 and AMR70 yeast, respectively (69). L40 bait strains were mated with AMR70 prey strains as described previously (67). Growth on Ura Trp Leu (UTL) plates indicated efficiency of mating, while growth on Trp His Ura Leu Lys (THULL) plates indicated bait-prey interaction. Expression of a subset of constructs was confirmed by Western blotting of L40 strains with the indicated bait or prey. Extracts were made from 10-ml saturated cultures by glass bead lysis for 30 min at 4°C in a mixture of 25 mM Tris (pH 7.5), 1 mM EDTA, 10 mM -mercaptoethanol (-ME), 1 mM phenylmethylsulfonyl fluoride (PMSF), and Complete Proteolysis Inhibitor (Roche Molecular Biochemicals) and centrifuged at 14,000 × g for 5 min; concentrations of soluble protein fractions were determined by Bradford (Bio-Rad). Ten to 15 µg of each extract was separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) on 10% polyacrylamide gels, transferred to nylon membranes (Ambion), probed with the either anti-GAL4-TA (1:200 dilution) or anti-lexA-DB (1:200 dilution) (Santa Cruz Bio. Inc.) followed by the appropriate secondary antibody, and detected by chemiluminescence.

Measurement of mutation rates and CAN1 mutational spectrum analysis. Briefly, strains were streak purified before the mutation rate assay, individual colonies were grown to saturation in YPD, and then various dilutions were plated onto complete synthetic medium (CSM), Thr and +canavanine (+CAN [60 µg/ml]) plates and colonies were counted after 2 to 3 days of growth at 30°C. Rates were determined as previously described (51). Statistical analyses were performed by using a two-tailed Mann-Whitney test with Prism 2.0a software (GraphPad Software, Inc.); P values of <0.05 were considered statistically significant.

Expression and purification of yeast MutL. The Bac-to-Bac Baculovirus (Life Technologies) expression system was used to express MutL in Spodoptera frugiperda (Sf9) cells infected with recombinant baculovirus. Recombinant baculoviruses that express 6×His-Mlh1p and Pms1p were created as described in the manufacturer's instructions (Bac-to-Bac Baculovirus expression system; Life Technologies). A 200-ml culture of Sf9 cells (typically 1 × 10⁶ to 2 × 10⁶ cells/ml) was coinfected with recombinant baculoviruses at multiplicities of infection of 2 to 2.5 and 11 to 15 for 6×His-Mlh1p- and Pms1p-expressing baculoviruses, respectively. Cells were harvested at 44 to 48 h of coinfection, frozen as cell pellets with liquid nitrogen, and stored at 80°C.

Limited proteolysis assays. Limited proteolysis reaction mixtures (20 µl) consisted of 150 ng of MutL, 30 mM Tris (pH 7.6), 150 mM NaCl, 5 mM MgCl₂, and 0.5 mM dithioerythritol with or without 5 mM ATP, adenosine 5'-(,-imino)triphosphate (AMP-PNP), ATPS, or ADP. Reaction mixtures were incubated for 15 min at 30°C, followed by the addition of 50 ng of modified trypsin (Promega Corp.), incubation at 30°C for a specified interval, addition of SDS-sample buffer, and boiling for 7 min. Processed reactions were separated on an SDS-PAGE (10% polyacrylamide) gel and transferred onto polyvinylidene difluoride membranes (Ambion), and Western blotting was performed with specified antibodies. Anti-4×His Western blots were performed as described above. The anti-Mlh1p polyclonal antibody was a kind gift of T. Kunkel (National Institute of Environmental Health Sciences) and was generated against a COOH-terminal peptide of yeast Mlh1p. The anti-Mlh1p polyclonal antibody was used to probe limited proteolysis blots at a 1:10,000 dilution and detected as described above.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Predicted ATPase residues of MutL are necessary for MMR in vivo. To examine the importance of putative ATPase domains of S. cerevisiae Mlh1p and Pms1p, we examined the effects of mutations introduced at two of the ATPase motifs, I and III, conserved in the GHL family (Fig. 1). We chose to examine residues E31 and E61 of Mlh1p and Pms1p, respectively, because mutations at the homologous glutamate of MutL, gyrase B, and Hsp90 have been shown to eliminate ATP hydrolysis with little or no effect on ATP binding (9, 10, 34, 47, 50). In motif III, we focused on residues G98 and G128 of Mlh1p and Pms1p, respectively, which are modeled to affect ATP binding and/or an associated conformational change induced upon ATP binding (9, 10, 26, 47, 50). For brevity, we will refer to alanine substitution mutations at E31 and E61 of Mlh1p and Pms1p, respectively, as hydrolysis mutations and the mutations G98A and G128A in Mlh1p and Pms1p, respectively, as ATP-binding mutations.

View larger version (15K): [in this window] [in a new window]   FIG. 1.   NH2-terminal ATPase domains of GHL ATPases. ATPase motifs I to IV are designated by black boxes, and sequences are shown above motif boxes. Numbers correspond to the number of amino acids preceding or following sequence alignments. Boldface letters are the absolutely conserved residues that were substituted for with alanine in Mlh1p and Pms1p.

View this table: [in this window] [in a new window]   TABLE 2.   Mutation rate effects of genomic mutations altering the NH2 termini of MutL and MutL

Mutator effects of single alterations in the putative ATPase domains of S. cerevisiae MutL. To examine the individual contributions of Mlh1p and Pms1p ATPase motifs to MutL function, we examined the effect of single mutations on mutation avoidance. As shown in Table 2, the single mutations affecting the ATPase motifs had effects on mutation avoidance that were significantly less than the corresponding mlh1 and pms1 strains, with the exception of the mlh1 binding mutant (PTY300), the effect of which was only slightly less than the mlh1 strain (PTY100). Interestingly, the homologous ATPase mutations made in MLH1 and PMS1 had different effects on mutation avoidance. The mlh1 hydrolysis mutant (PTY200) displayed 16-fold and 7-fold higher rates of mutation at hom3-10 and CAN1, respectively, than the corresponding pms1 hydrolysis mutant strain PTY201 (P < 0.0286 for both loci). Likewise, the mlh1 binding mutant (PTY300) showed 9- and 5.5-fold higher rates of mutation at hom3-10 and CAN1, respectively, than the corresponding pms1 binding mutant strain PTY301 (P < 0.0286 for both loci). One trivial explanation for the differential effects of homologous mlh1 and pms1 mutations on mutation avoidance was that MLH3, which is involved in a minor mutation avoidance pathway (22), compensates for the pms1 mutations. However, as shown in Table 2, pms1 point mutant strains deleted for MLH3, PTY202 (pms1-E61A mlh3), and PTY302 (pms1-G128A mlh3) still demonstrated mutation rates significantly smaller than that observed for the respective homologous mlh1 ATPase point mutant strains PTY200 (mlh1-E31A) and PTY300 (mlh1-G98A) (P < 0.0159 for PTY200 versus PTY202 and P < 0.0286 for PTY300 versus PTY302). These results suggest that the differences seen between homologous mlh1 and pms1 ATPase point mutations with respect to mutation rate are not due to the redundant functions of MLH3, but rather infer an intrinsic asymmetry within the MutL complex.

pms1 mlh3

msh2

MutL undergoes an ATP-dependent conformational change. To investigate further the role of candidate ATP-binding or hydrolysis motifs in MutL function, we used limited proteolysis to examine the effects of adenine nucleotides on the conformation of recombinant MutL purified from insect cells (data not shown). We used an antibody directed against the 6×His tag at the NH₂ terminus of Mlh1p to detect NH₂-terminal fragments following limited proteolysis. As depicted in Fig. 2a, the presence of ATP led to the protection of distinct NH₂-terminal fragments of Mlh1p from trypsin proteolysis. The protected NH₂-terminal fragments of approximately 42 and 38 kDa coincide with the E. coli MutL LN40 thrombin proteolytic fragment that possessed the core ATPase domain (9, 10). We did not observe any differences between MutL in the presence or absence of ATP when using a polyclonal antibody directed against the COOH terminus of Mlh1p (data not shown), suggesting that the COOH terminus of Mlh1p does not undergo an ATP-dependent conformational change. However, we did detect an approximately 30-kDa band that was resistant to proteolysis in the presence or absence of ATP even after a 30-min incubation with 750 ng of trypsin (data not shown). This highly trypsin-resistant Mlh1p COOH-terminal fragment may represent the COOH-terminal heterodimerization domain of Mlh1p. As shown in Fig. 2b, lanes 3 to 5 demonstrate that nonhydrolyzable ATP analogs AMP-PNP and ATPS, as well as ADP, also protect the NH₂ terminus of Mlh1p from trypsin proteolysis. Qualitatively, the relative levels of protection from proteolysis in the presence of nucleotide are as follows: ATP  AMP-PNP > ATPS > ADP. As demonstrated for ATP, we saw no differential protection of the COOH terminus of Mlh1p in the presence of ADP, AMP-PNP, or ATPS by reprobing with the antibody directed against the COOH terminus of Mlh1p (data not shown). The limited proteolysis results indicate that at least the Mlh1p NH₂ terminus of MutL undergoes an ATP-binding-dependent conformational change. We were unable to address whether the NH₂ terminus of Pms1p undergoes a similar ATP-dependent conformational change, because an antibody specific for the NH₂ terminus of Pms1p antibody was not available. However, as described below, yeast two-hybrid results suggest that Pms1p also undergoes an ATP-dependent conformational change.

View larger version (23K): [in this window] [in a new window]   FIG. 2.   Adenine nucleotides alter trypsin sensitivity of MutL. (a) One hundred fifty nanograms of MutL was subjected to proteolysis with modified trypsin as described in Materials and Methods in the presence or absence of 5 mM ATP for the indicated time at 30°C. Products were treated with SDS-sample buffer, boiled, separated on an SDS-PAGE (10% polyacrylamide) gel, and detected by immunoblotting with anti-4×His antibody. Arrows denote full-length 6×His-Mlh1p, and asterisks designate NH2-terminal (term.) fragments of 6×His-Mlh1p that are protected from proteolysis in the presence of ATP. Equal loading of samples and even transfer of the blot were demonstrated by using a polyclonal antibody raised against the COOH terminus of Mlh1p (data not shown). (b) The same analysis was performed as described for panel a, but the effects of 5 mM adenine nucleotides ADP, AMP-PNP, and ATPS were examined.

ATP binding promotes heterodimerization of the NH₂ termini of MutL in vivo. ATP-binding-dependent conformational changes in other GHL family members are associated with dimerization of their NH₂-terminal ATP binding domains (9, 71). To inquire into the functional significance of the ATP-dependent conformational change data described above, we used the yeast two-hybrid system to assay interactions between wild-type and mutant NH₂-terminal fragments of Mlh1p and Pms1p. Based upon sequence alignments with GHL family members, the fragments 1 to 354 and 1 to 401 of Mlh1p and Pms1p, respectively, should each contain the structural elements necessary for ATP binding and hydrolysis (9). These NH₂-terminal fragments of Mlh1p and Pms1p were fused at their COOH termini to either the lexA DNA binding domain or the GAL4 activation domain (Fig. 3a). Consistent with our previous studies (51), no interaction was seen between wild-type Mlh1p and Pms1p NH₂-terminal fragments (Fig. 3a, group I). As stated before, homologous ATP hydrolysis mutations in other GHL ATPases have been shown to abolish ATP hydrolysis activity with little or no effect on ATP binding (9, 10, 34, 47, 50). Interestingly, for an NH₂-terminal fragment of gyrase b, ATP binding was only observed in vitro for a hydrolysis-deficient form (34). Therefore, we reasoned that ATP hydrolysis mutations in the NH₂ termini of both Mlh1p and Pms1p might prolong a double ATP-bound state in vivo and allow interaction to be detected by the yeast two hybrid assay. Indeed, as shown in Fig. 3a (group III), a robust interaction was seen when both Mlh1p and Pms1p NH₂-terminal fragments possessed ATP hydrolysis mutations. However, the interaction was not observed when only the Mlh1p or Pms1p fragment possessed the ATP hydrolysis mutation E31A or E61A, respectively (Fig. 3a, group II). To demonstrate that this novel interaction was dependent on the putative ATP-binding activities of Mlh1p and Pms1p, we superimposed either the mlh1-G98A or pms1-G128A ATP-binding mutation onto the hydrolysis-defective NH₂-terminal fragments of mlh1p-E31A or pms1p-E61A, respectively. Supporting our hypothesis, we observed that superimposing a mutation designed to prevent the putative ATP-binding or conformational change in one fragment ablated the two-hybrid interaction (Fig. 3a, compare groups III and IV). Interestingly, the two-hybrid interaction seen in Fig. 3a, group III, was specific only for the mlh1p-E31A and pms1p-E61A NH₂-terminal fusion pairs, because the NH₂ terminus of mlh1p-E31A did not interact with itself (data not shown). The same observation was seen with the NH₂ terminus of pms1p-E61A (data not shown), suggesting that, similar to their respective COOH-terminal domains (51), the NH₂ termini of Mlh1p and Pms1p do not homodimerize. The two-hybrid results of Fig. 3a are not due to ATP hydrolysis or ATP-binding mutations grossly affecting expression or stability of the fusion proteins, because Western analysis demonstrates that all fusion proteins are expressed at similar levels (Fig. 3b). Taken together, the two-hybrid results suggest that ATP binding, but not hydrolysis, by both Mlh1p and Pms1p is necessary for MutL NH₂-terminal heterodimerization.

View larger version (48K): [in this window] [in a new window]   FIG. 3.   Two-hybrid analysis detects NH2-terminal Mlh1p and Pms1p interaction. (a) Boxes correspond to bait and prey constructs tested for interaction. The residues included in the fusions are indicated below the group I and V constructs, respectively. Amino acid substitutions designated above each construct are indicated by black bars within the construct boxes. Interaction is scored as growth on HIS media and blue color development with the substrate X-Gal (5-bromo-4-chloro-3-indolyl--D-galactopyranoside) as described in Materials and Methods. Group I, wild-type NH2-terminal fusion fragments; group II, one NH2-terminal fusion fragment contains a hydrolysis point mutation; group III, both NH2-terminal fusion fragments contain hydrolysis point mutations; group IV, one NH2-terminal fusion fragment has the indicated compound mutations; and group V, positive control reaction with full-length Pms1p and Mlh1p. (b) Western analysis of L40 strains with two-hybrid constructs from panel a using anti-GAL4-TA or anti-lexA-DB monoclonal antibody as described in Materials and Methods. Lanes: 1, pCAD3 (empty vector); 2, pCAD-mlh1 N-354; 3, pCAD-mlh1-E31A N-354; 4, pCAD-mlh1-E31A, -G98A N-354; 5, pNBTM (lexA); 6, pNBTM-pms1 N-401; 7, pNBTM-pms1-E61A N-401; 8, pNBTM-pms1-E61A, -G128A N-401. Fusion products and lexAp are indicated by arrowheads. The approximately 90-kDa band in lanes 1 to 4 may be endogenous Gal4p. The other bands present in control lanes 1 and 5 and in lanes 2 to 4 and 6 to 8, respectively, represent nonspecificity by the primary and secondary antibodies. In lanes 6 to 8, the faster-migrating specific anti-lexA-DB reacting species is unknown, but may be a pms1p(1-401)-lexAp degradation product.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Although clearly crucial for MMR, little information exists on the function of the major MutL activity in yeast, MutL, composed of Mlh1p and Pms1p. Previous studies with yeast have defined COOH-terminal domains as important for Mlh1p and Pms1p interaction (51) and conserved NH₂-terminal residues as necessary for MMR activity (51, 63). Recent investigations of MutL and other members of the GHL family of ATPases have suggested guidelines for more detailed studies of MutL function. Here, we present evidence defining yeast MutL as a functional member of the GHL ATPase superfamily. First, residues critical for the ATPase function of GHL family members, when substituted for alanine in both Mlh1p and Pms1p, disrupt MMR. Second, adenine nucleotide binding protects the NH₂ terminus of Mlh1p from trypsin proteolysis, suggesting that MutL undergoes ATP-dependent conformational changes. Third, results from the two-hybrid system suggest that one consequence of the ATP-induced conformational changes is an interaction between the NH₂ termini of Mlh1p and Pms1p. Finally, analysis of single mlh1 and pms1 ATPase motif mutants indicates a functional asymmetry within yeast MutL.

GHL family members appear to share an ATPase cycle that is highlighted by an NH₂-terminal dimerized intermediate in the ATP-bound form (4, 9, 10, 55-57, 71). In the case of MutL, this ATP-binding-induced NH₂-terminal dimerization activated MutH endonuclease in vitro (10). The limited proteolysis and two-hybrid analyses presented here support an ATPase cycle for MutL, composed of at least four intermediates, that is similar to those of other GHL family members (Fig. 4). Limited proteolysis suggests that MutL undergoes a conformational change in vitro that is dependent on ATP binding, as AMP-PNP (and to a lesser extent ATPS) produced the same effect as ATP (Fig. 4, intermediate 2). Furthermore, we observed a specific two-hybrid interaction between Mlh1p and Pms1p NH₂-terminal fragments, each containing ATP hydrolysis mutations (Fig. 4, intermediate 3). Interaction was not detected by a yeast two-hybrid assay with wild-type NH₂-terminal fragments of Mlh1p and Pms1p, presumably because ATP hydrolysis renders the interaction transient. In further support of the existence of intermediate 3, our recent observations suggest that the double hydrolysis mutant form of MutL is highly resistant to trypsin-limited proteolysis even without added ATP (unpublished observations). Also, that ADP provided some protection from limited proteolysis suggests the existence of an ADP-bound intermediate (Fig. 4, intermediate 4). Double mutant mlh1 pms1 strains with alanine substitutions at ATP hydrolysis or ATP-binding residues showed increased spontaneous mutation rates indistinguishable from those of completely MMR-defective cells. This double mutant analysis suggests that the candidate ATPase domains of both Mlh1p and Pm1p and the ATPase cycle described above are required for MutL function in yeast MMR for mutation avoidance. It is intriguing to speculate on the function of the NH₂-terminal ATP-bound MutL intermediate, because similar findings from other GHL ATPases (25, 55-57), namely MutL (10), suggest that this MutL intermediate may play a significant role in coordinating downstream steps with known and perhaps unidentified MMR proteins. Although our data are consistent with the ATPase cycle represented in Fig. 4, further biochemical work with MutL is required to confirm and characterize the contribution of Mlh1p and Pms1p ATP binding and hydrolysis activities to MutL function. Similar to earlier work from the Hsp90 field (35, 61), we have not been able to specifically assign an intrinsic ATPase activity to MutL with our current protein preparations (unpublished observations). However, similar to what is currently known for GHL ATPases (9, 10, 34, 47, 50), our double mlh1 pms1 hydrolysis mutant phenotype and our two-hybrid results suggest a crucial role for ATP hydrolysis during mutation avoidance in yeast MMR (Table 2).

View larger version (18K): [in this window] [in a new window]   FIG. 4.   A model for the yeast MutL ATPase cycle. Briefly, intermediate 1 is the nucleotide-free state. ATP binding induces conformational changes in the NH2 termini of Mlh1p and Pms1p, represented by a change in shape from rectangular to oval that occurs in the step(s) between intermediates 2 and 3. Intermediate 3 is heterodimerization of the NH2 termini of Mlh1p and Pms1p in the ATP-bound state. Intermediate 4 is the ADP-bound form following ATP hydrolysis. The mlh1-G98A and pms1-G128A ATP-binding mutants were constructed to affect the transition(s) from intermediate 1 to intermediate 2 and/or intermediate 2 to intermediate 3. In contrast, the ATP hydrolysis mutations, mlh1-E31A and pms1-E61A, were modeled to prevent the transition from intermediate 3 to intermediate 4. M, the NH2 terminus of Mlh1p; P, the NH2 terminus of Pms1p; C, COOH termini of Mlh1p and Pms1p. Each arrow may represent multiple distinct steps. This model, which is consistent with the studies reported here, was adapted from a model for MutL proposed by Ban and Yang (9).

Interestingly, genetic analysis revealed a functional asymmetry with respect to the two ATPase domains of the MutL heterodimer. Specifically, alanine substitution mutations affecting the predicted ATPase motifs of Mlh1p had a greater impact on mutation avoidance than the corresponding mutations in Pms1p. Formally, our genetic results argue that Mlh1p can compensate better for ATPase mutations in Pms1p for mutation avoidance than can Pms1p for the corresponding ATPase mutations in Mlh1p. The apparent genetic asymmetry detected for MutL may reflect at the mechanistic level a kinetic asymmetry similar to that observed in the homodimeric ATPases, topoisomerase II from S. cerevisiae, and the  complex from E. coli (6, 28, 29). Biologically, the genetic asymmetry observed with MutL may represent distinct but overlapping roles of Mlh1p and Pms1p during mutation avoidance, e.g., excision tracts originating 5' versus 3' from the mismatch or IDL (18, 46) or differential roles during strand discrimination.

MutL has been referred to as a "molecular matchmaker," coupling the mismatch binding activity of MutS to the latent endonuclease MutH (60). One criterion of a molecular matchmaker that MutL has always appeared to lack was an intrinsic ATPase activity. Recent work has now identified this "missing" activity, and as suggested previously, it appears to be critical for MutL activity in MMR (5, 9, 10). Moreover, the MutL ATPase activity was responsible for coordinated interaction and activation of MutH in vitro (10). In this report, we have shown that the conserved ATPase motifs of MutL are necessary for mutation avoidance by MMR in yeast. The role of the ATPase motifs of Mlh1p and Pms1p in other MMR-related functions, such as meiotic (7, 8, 30, 70) and homeologous recombination (16, 17, 31, 62), remains to be determined. Finally, as for MutL (9, 10), the ATP-dependent conformational changes in yeast MutL are likely to facilitate interaction with downstream proteins in MMR. Our ability to produce a stable Mlh1p-Pms1p NH₂-terminal interaction via the yeast two-hybrid system may provide a means to identify these proteins.