The MutL ATPase is required for mismatch repair.

Members of the MutL family contain a novel nucleotide binding motif near their amino terminus, and the Escherichia coli protein has been found to be a weak ATPase (Ban, C., and Yang, W. (1998) Cell 95, 541-552). Genetic analysis has indicated that substitution of Lys for Glu-32 within this motif of bacterial MutL results in a strong dominant negative phenotype (Aronshtam, A., and Marinus, M. G. (1996) Nucleic Acids Res. 24, 2498-2504). By in vitro comparison of MutL-E32K with the wild type protein, we show the mutant protein to be defective in DNA-activated ATP hydrolysis, as well as MutS- and MutL-dependent activation of the MutH d(GATC) endonuclease and the mismatch repair excision system. MutL-E32K also acts in dominant negative manner in the presence of wild type MutL in vitro, inhibiting the overall mismatch repair reaction, as well as MutH activation. As judged by protein affinity chromatography, MutL and MutL-E32K both support formation of ternary complexes that also contain MutS and MutH or MutS and DNA helicase II. These findings imply that the MutL nucleotide binding center is required for mismatch repair and suggest that the dominant negative behavior of the MutL-E32K mutation is due to the formation of dead-end complexes in which the MutL-E32K protein is unable to transduce a signal from MutS that otherwise results in mismatch-dependent activation of the MutH d(GATC) endonuclease or the unwinding activity of helicase II.

Mismatch repair stabilizes the bacterial genome by correcting DNA biosynthetic errors and by ensuring the fidelity of homologous genetic recombination (1)(2)(3)(4). The pathway displays a broad specificity for different mispairs, with repair of DNA biosynthetic errors targeted to the daughter strand by virtue of the transient absence of d(GATC) methylation on newly synthesized sequences (5). Repair is initiated by the binding of MutS to the mismatch, a reaction that can occur with the dimeric form of the protein (6 -9). MutL, which also exists as a dimer in solution, binds to heteroduplex DNA in a MutS-dependent manner (9 -11). Assembly of this ternary complex is sufficient to activate the d(GATC) endonuclease activity of MutH, which cleaves the newly synthesized, unmethylated strand (12), and to activate unwinding by DNA helicase II, which enters the helix at the incised d(GATC) sequence and unwinds toward the mismatch (13). That portion of the incised strand unwound in this manner is subject to degradation by one of several single strand exonucleases (14,15). Repair synthesis of the ensuing gap is mediated by DNA polymerase III holoenzyme in the presence of single strand DNA-binding protein, and DNA ligase restores covalent continuity to the repaired strand (16).
MutL plays a critical role in mismatch repair, but the molecular functions of the protein in the reaction are only partially understood. It has been suggested that MutL serves to interface mismatch recognition by MutS to other activities involved in repair (17, 18), and evidence supporting this idea is available. As noted above, MutL binds to heteroduplex DNA in the presence of MutS (9 -11). MutL also activates unwinding by DNA helicase II on conventional helicase substrates (19) and activates the d(GATC) endonuclease of MutH in the absence of MutS under certain conditions (20). Both of these effects are attributable at least in part to physical interaction of MutL with the latter two activities (19,21,22).
MutL homologs have been identified in yeast, mouse, and human cells (3,4,23). In contrast to the homodimeric structure of bacterial MutL, eukaryotic MutL function is provided by heterodimeric complexes of homologs of the bacterial protein, e.g. the MLH1⅐PMS1 complex in yeast and the MLH1⅐PMS2 heterodimer in mammalian cells (24,25). These bacterial and eukaryotic polypeptides display sequence homology within their amino-terminal regions, which contains a novel nucleotide binding motif that was originally identified in type II topoisomerases, HSP90, and histidine kinase families (26,27). A number of dominant negative mutations have been localized to this motif in the bacterial protein (28), and about 50% of missense mutations found in human MLH1 in nonpolyposis colon cancers are also within the conserved region (29). The structure of a 40-kDa amino-terminal segment of Escherichia coli MutL has been determined as the apoprotein (20) and as a complex with AMPPNP 1 or ADP, work that has demonstrated a large conformational transition associated with nucleotide occupancy (30). In addition, the bacterial protein has been reported to be a weak ATPase (20, 30). However, others have failed to detect this activity (19,22), perhaps because of the high MutL K m for ATP and the fact that purified MutL preparations are commonly contaminated by trace levels of DNA helicase II, which also hydrolyzes this nucleotide.
To further clarify the role of the MutL nucleotide binding site in mismatch repair, we have studied the biochemical properties of a mutant form of the protein that harbors a Glu to Lys substitution at position 32 within motif N of the nucleotide binding center. The E32K amino acid substitution results in a strong dominant negative mutL phenotype, and unlike other dominant negative mutL mutations that have been characterized genetically, the phenotype of the E32K substitution mutation is not suppressed by increased gene dosage of the wild type allele (28). * This work was supported by Grant GM23719 from NIGMS, National Institutes of Health. 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.

MATERIALS AND METHODS
Bacterial Strains-E. coli BL21(DE3) mutL103::Tn5 uvrD::tet was constructed by P1 transduction of mutL103::Tn5 from NK7510 (a gift from Nancy Kleckner, Harvard University) into BL21(DE3) uvrD::tet (provided by Stephen Matson, University of North Carolina), with selection for resistance to tetracycline and kanamycin. Deficiency of both MutL and helicase II in the double mutant was confirmed by Western blotting and by in vitro complementation to restore mismatch repair (not shown).
A MutL-E32K mutation was constructed by converting the Glu-32 GAA codon to a Lys AAA codon using the megaprimer polymerase chain reaction method (31). The mutant gene was inserted into pET3A after digestion with BamHI and NdeI as described previously (19) to yield plasmid pCS1. The sequence of the MutL-E32K gene was determined in its entirety to confirm the presence of the desired single mutation.
Protein Expression and Purification-Wild type MutL was prepared using the overproducing plasmid L1-pET3a as described previously (19) and was isolated after expression in a BL21(DE3) uvrD::tet host to eliminate trace contamination by DNA helicase II that was observed previously (19). MutL-E32K mutant protein was expressed from plasmid pCS1 in E. coli BL21(DE3) mutL103::Tn5 uvrD::tet. Cells were grown in Luria broth to an optical density (600 nm) of 0.8 at 37°C, induced with 1 mM isopropyl-␤-D-thiogalactopyranoside for 2 h at 37°C, and harvested by centrifugation. The mutant protein was isolated by the procedure of Grilley et al. (10) during which it fractionated like wild type MutL. Preparations had a purity greater than 98% as judged by electrophoresis in the presence of SDS. MutS, MutH, and DNA helicase II were isolated by published methods (6,32,33). Protein concentrations were determined by Bradford assay using bovine serum albumin (BSA) as standard (34). When cited in molar terms, protein concentrations are expressed as monomer equivalents.
Determination of Native Aggregation State-The aggregation state of MutL-E32K was compared with that of the wild type protein by gel filtration through a Superose Protein cross-linking was performed in 50 mM KPO 4 , pH 7.4, in the presence of 4.8 mM glutaraldehyde. After 5 min at room temperature, sodium borohydride (2 mol/mol of glutaraldehyde) was added. After 20 min at 0°C, reactions were supplemented with 8 molar equivalents of sodium pyruvate to quench the excess sodium borohydride, and incubation at 0°C continued for 60 min. Cross-linked samples were supplemented with BSA to 1.6 mg/ml and precipitated with ice-cold trichloroacetic acid (10%), and the pellet dissolved in 50 mM Tris-HCl, pH 6.8, 2% SDS, 0.6 M 2-mercaptoethanol, 10% glycerol, 0.1% bromphenol blue and heated at 100°C for 2 min. Samples were analyzed on a 4 -15% gradient polyacrylamide gel in the presence of 0.1% SDS and blotted to a nitrocellulose membrane. MutL was determined immunologically using ECL chemiluminescence assay according to the manufacturer's (Amersham Pharmacia Biotech) instructions. Standard proteins with M r from 32,500 to 175,000 were included for estimation of the molecular mass.
ATPase and ATP Binding Assays-ATP hydrolysis was determined at 37°C in reactions (20 l) containing 20 mM Tris-HCl, pH 8.0, 90 mM KCl, 1 mM dithiothreitol, and 0.8 M MutL or MutL-E32K. The effect of single-stranded DNA (30) on ATP hydrolytic activity was determined by inclusion of a 58-residue synthetic oligonucleotide (2 M molecules unless noted otherwise). Hydrolysis was initiated by the addition of [␣-32 P]ATP⅐(Mg 2ϩ ) 2 to prewarmed reactions to a final concentration of 1 mM. At 10 min intervals 2-l samples were taken and quenched with 50 l of 0.5 M EDTA, pH 8.0. The extent of ATP hydrolysis was determined by chromatography of 1 l of quenched samples on polyethyleneiminecellulose plates (EM Science, Gibbstown NJ) that were developed in 0.3 M KPO 4 , pH 7.0. Dried plates were phophorimaged overnight and optically scanned using a STORM imager system, and data were analyzed using Imagequant software (Molecular Dynamics, Sunnyvale CA). Initial steady-state rates of ATP hydrolysis were determined by least squares analysis of the linear portion of the progress curve.
ATP binding was determined in reactions (20 l) containing 20 mM Tris-HCl, pH 7.6, 25 mM KCl, 75 g/ml BSA, 1 mM dithiothreitol, and 3.7-18.4 M MutL or 8 M MutL-E32K. Binding was initiated by the addition of indicated concentrations of [␣-32 P]ATP⅐(Mg 2ϩ ) 2 or [␥-32 P]ATP⅐(Mg 2ϩ ) 2 (0.2 Ci/mmol). After incubation at 0°C as indicated, 1 ml of reaction buffer was mixed with each sample, and the entire volume was passed through a 0.45-m nitrocellulose membrane prewetted with reaction buffer. Filters were washed three times with 1 ml of reaction buffer lacking BSA and dried, and radioactivity was determined by liquid scintillation counting. ATP binding to membranes in the absence of MutL was typically less than 3% of that observed in complete reactions, and background in the presence of heat denatured MutL was less than 6%. The latter values were used as blanks and were subtracted from the data shown.
The MutL-bound nucleotide was quantitatively released from filters by soaking for 2 h at room temperature in 0.4 ml of 20 mM Tris-HCl, pH 8.0, containing 0.2% SDS or 0.3 g/ml proteinase K. The two methods yielded identical results. Samples were spotted onto polyethyleneiminecellulose plates that were developed and analyzed as described above.
Mismatch Repair and Partial Reaction Assays-MutL activity in mismatch repair was determined by complementation of extracts prepared from E. coli MG102 (mutL::Tn10) as described previously (10) using an f1 heteroduplex containing a G-T mismatch at position 5632 and a single d(GATC) site at position 216 about 1000 base pairs from the mismatch (7). d(GATC) methylation was on the complementary DNA strand. Repair was scored by restriction endonuclease assay, and reaction products were visualized by ethidium stain after agarose gel electrophoresis (7). Repair products were quantitated by use of a cooled, photometric grade charge-coupled device imager (Photometrics, Inc.). d(GATC) cleavage by activated MutH endonuclease was determined by a modification of the method of Au et al. (12) using a hemimethylated 32 P-end-labeled G-T heteroduplex. This DNA was prepared by linearization of the circular, hemimethylated f1 heteroduplex mentioned above with Bsp106 endonuclease, 3Ј-end labeling of the product with exonuclease-free Klenow DNA polymerase I (Amersham Pharmacia Biotech) in the presence of dATP, dGTP, dTTP, and [␣-32 P]dCTP (NEN Life Science Products, 3000 Ci/mmol), termination of the reaction by heating to 75°C for 20 min, and removal of unincorporated nucleotide by Sephadex G-50 chromatography in 10 mM Tris-HCl, pH 8.0, 1 mM EDTA. MutH endonuclease reactions (10 l) contained 20 mM Tris-HCl, pH 8.0, 20 mM KCl, 5 mM MgCl 2 , 50 g/ml BSA, 1 mM dithiothreitol, 100 ng of 32 P-end-labeled G-T heteroduplex, and MutS, MutL, and MutH as indicated. After preincubation for 5 min at 37°C, reactions were initiated by the addition of ATP to 3 mM. Reactions were terminated as indicated by the addition of one-third volume of 0.2 N NaOH, 0.04 M EDTA, 10% Ficoll 400, 0.1% Bromcresol green, and samples electrophoresed through 1% agarose gels in 0.03 N NaOH, 1 mM EDTA. Gels were dried, quantitated using a Molecular Dynamics PhosphorImager and autoradiographed.
Mismatch-provoked gap formation was performed with purified proteins as described (15,16). Reactions (10 l) contained 35 ng of MutS, 30 ng of MutL, 10 ng of DNA helicase II, 200 ng of single strand-binding protein, 3 ng exonuclease I, and 24 fmol of nicked circular G-T heteroduplex containing a single strand break at the gpII cleavage site. This viral strand nick is located 181 base pairs 3Ј to the mismatch as measured along the shorter path between the two sites in the circular molecule (35). Presence of a single-stranded gap in the DNA was determined by virtue of resistance of the gapped intermediate to NheI endonuclease, the recognition site for which is located 5 base pairs 5Ј to the mismatch as viewed on the viral strand (35).
MutS Affinity Columns-Purified MutS (850 g) was coupled to 1, 1Ј-carbonyldiimidazole activated agarose (1 ml) according to the supplier (Pierce) in 0.1 M NaHCO 3 , pH 8.9. Residual active groups on the resin were then blocked by incubating the gel for an additional 2 h at room temperature with 1 M ethanolamine, pH 8.9. After transfer to a chromatography column, excess soluble protein was removed by washing the gel with buffer A (30 mM potassium phosphate, pH 7.4, 50 mM NaCl, 0.5 mM dithiothreitol, 10% glycerol) containing a final concentration of 1 M NaCl, and the column was then equilibrated with buffer A. A BSA-coupled control column material was prepared in an identical manner.
Indicated proteins (34 g in 2 ml of buffer A) were applied to MutS or control columns, which were washed three times with 1 ml of buffer A, with each wash collected as an individual fraction. Columns were eluted with 2 ml of buffer A containing NaCl at a final concentration of 250 mM, and 0.4 ml fractions were collected. Fractions were analyzed by electrophoresis on SDS-polyacrylamide gels, which were either stained with Coomassie Blue (for BSA control) or blotted to a nitrocellulose membrane. MutL, MutH, and helicase II were quantitated by immunological chemiluminescence assay as described above. When indicated, 2 mM ATP and 5 mM MgCl 2 were added to buffer A.

RESULTS
Aronshtam and Marinus (28) have identified 72 mutations within the E. coli mutL gene that behave in a dominant negative fashion when expressed from a multicopy plasmid in the presence of a single wild type copy of the gene on the bacterial chromosome. With one exception, the dominant negative phenotype of these mutations was reversed when the wild type protein was also expressed from a multicopy plasmid, suggesting that dominant negative behavior in most cases was because of production of mixed mutant wild type MutL heterodimers or sequestration of other repair activities by the mutant protein (28). The latter possibility is consistent with the finding that the phenotype of a substantial fraction of the MutL mutations was also partially compensated for by overproduction of MutS or MutH. The single exception was a mutation resulting in substitution of Glu-32 by Lys (mutL705 allele), which retained dominant negative behavior even in the presence of elevated levels of wild type MutL, MutS, or MutH (28). Glu-32 is located within motif N of the MutL ATP hydrolytic center (20, 26,27,30). Screen of GenBank TM data bases (not shown) indicate that this residue is highly conserved among bacteria, fission yeast, insect, plant, rodent, and human species, although it is replaced by Asp in the Saccharomyces cerevisiae PMS1 polypeptide.
To clarify the role of the MutL ATP hydrolytic center in mismatch repair, we have isolated the MutL-E32K mutant protein in essentially homogeneous form and have compared its biochemical properties with those of the wild type protein.
Because MutL is known to activate and interact with DNA helicase II (19,21), the mutant protein was isolated from a strain with insertion mutations in chromosomal mutL and uvrD loci ("Materials and Methods").
ATP Binding and Hydrolysis-MutL has been reported to catalyze the hydrolysis of ATP in a slow reaction that is enhanced substantially by single-stranded DNA (20, 30). We have confirmed this observation and have compared the wild type protein and MutL-E32K in this regard. MutL hydrolyzed ATP with a K m of 0.41 Ϯ 0.11 mM and a k cat of 0.86 Ϯ 0.08 min Ϫ1 / monomer equivalent (not shown). This k cat value is similar to that reported by Ban and Yang (20), although the K m we have determined is somewhat higher than that of the previous study, perhaps reflecting their use of His-tagged MutL. MutL-E32K also hydrolyzed ATP with a similar K m of 0.35 Ϯ 0.08 mM and a slightly reduced k cat of 0.38 Ϯ 0.03 min Ϫ1 . The hydrolytic activity of the wild type protein was increased 10-fold in the presence of a 58-residue single strand oligonucleotide, which increased the k cat for hydrolysis to 8.7 Ϯ 0.7 min Ϫ1 without altering the K m for ATP, which remained at 0.40 Ϯ 0.10 mM. However, as shown in Fig. 1 (upper panel), the presence of the single-stranded DNA had a minimal effect on the hydrolytic activity of the E32K mutant protein.
Binding of the magnesium chelate of [␣-32 P]ATP to wild type MutL was demonstrable by nitrocellulose membrane assay (Fig. 1, lower panel). Binding was hyperbolic with a dissociation constant of 0.48 Ϯ 0.04 mM at 0°C. Apparent stoichiometry where V e is the elution volume, V 0 is the void volume, and V t is the volume of the packed bed) for both proteins were 0.50, consistent with a dimer structure. Lower, wild type and E32K MutL (48 nM) were subjected to glutaraldehyde cross-linking, and polypeptide species were analyzed by polyacrylamide gel electrophoresis in the presence of SDS followed by Western immunoblot assay using rabbit polyclonal serum raised against MutL. with [␣-32 P]ATP was 0.54 Ϯ 0.05 mol nucleotide/mol MutL monomer, and a similar value was obtained with [␥-32 P]ATP (0.52 Ϯ 0.03 mol/monomer). Because MutL is a dimer in solution (10), this may indicate that only one of the two nucleotide binding sites is active in the native oligomer. However, we cannot rule out the possibility that the reduced stoichiometry is because of a filter retention efficiency of less than unity. Despite the similar hydrolytic activities of wild type MutL and the E32K mutant protein in the absence of DNA, ATP binding by the mutant protein was reduced by more than an order of magnitude at a nucleotide concentration that is near saturating for the wild type protein (Fig. 1). Because the membranes used to trap MutL⅐ATP complexes were subjected to buffer wash prior to quantitation, this difference may be indicative of an increased rate of dissociation of the nucleotide from the mutant protein.
The similar binding stoichiometries observed for wild type MutL with [␣-32 P]ATP and [␥-32 P]ATP suggested that the bound nucleotide observed in the experiments above is predominantly the triphosphate. This was confirmed by elution of MutL-bound nucleotide from filtered complexes prepared with [␣-32 P]ATP ("Materials and Methods"). Analysis by thin layer chromatography showed 80% of the MutL-associated nucleotide to be ATP (not shown), suggesting that the rate-limiting step for hydrolysis occurs at or prior to the chemical step.
Aggregation State of MutL-E32K-Sedimentation and gel filtration analysis has previously shown that native MutL exists in solution as a homodimer with a significant degree of nonspherical asymmetry (10). Comparison of wild type and E32K proteins demonstrated identical behavior during gel filtration (Fig. 2). Furthermore, glutaraldehyde treatment yielded identical cross-linking patterns for both proteins. As judged by these criteria, the E32K amino acid substitution does not alter the oligomeric state of MutL.
MutL-E32K Behaves as a Dominant Negative Inhibitor of Mismatch Repair in Vitro-As shown previously, extracts of E. coli MG102 (mutL::Tn10) are defective in methyl-directed repair, but mismatch correction can be restored to normal levels by the addition of MutL (Fig. 3, lanes a-c). MutL-E32K not only failed to complement extracts of the MutL-deficient strain (lane d), but inhibited repair when present together with the wild type protein (lanes e-h). Quantitation of the repair products ("Materials and Methods") from several experiments like those shown in Fig. 3 demonstrated that repair was reduced by 28 -38% (n ϭ 2) at a ratio of mutant:wild type protein 1:2 and by 71 Ϯ 16% (Ϯ 1 standard deviation, n ϭ 11) when the two proteins were present at equimolar concentration. This dominant negative inhibitory effect was independent of the order of addition of the two proteins to cell extract (Fig. 3, lanes f-h). These observations are in accord with the biological behavior of the E32K mutation (28).
MutL-E32K Is Defective in MutH Activation and Mismatchprovoked Excision Reactions-Methyl-directed repair initiates via the mismatch-dependent activation of the MutH d(GATC) endonuclease activity, a reaction that requires MutS, MutL, and ATP (12). Although MutL can activate the MutH endonuclease in an ATP-dependent manner in the absence of MutS or a mismatched base pair (20, 22), this effect is not detectable at the DNA and protein concentrations used in our experiments (Fig. 4, upper panel). As observed previously (12), efficient activation of the MutH endonuclease does occur with heteroduplex DNA in the presence of MutS, MutL, and ATP (Fig. 4,  middle panel). By contrast, the MutL-E32K protein does not support this reaction, and in fact inhibits MutH activation that occurs in the presence of wild type MutL (Fig. 4, lower panel). The second step in the bacterial mismatch repair reaction is excision of that portion of the unmethylated strand spanning the incised d(GATC) sequence and the mispair (15). This reaction involves mismatch-, MutS-, and MutL-dependent activation of DNA helicase II (19), with unwinding initiating at the strand break and proceeding in an orientation-dependent manner toward the mispair (13). The single strand displaced in this manner is subject to degradation by a single strand-specific exonuclease (14). Because MutL activates and interacts with DNA helicase II (19,21), we have compared the ability of wild type and E32K MutL with respect to their ability to support mismatch-provoked excision on a G-T heteroduplex containing a site-specific, strand-specific nick located 3Ј to the mismatch as viewed along the shorter path joining the two sites in the circular substrate. As shown in Fig. 5, wild type MutL supported efficient excision in a purified system that also contained MutS, DNA helicase II, single strand-binding protein, and exonuclease I (46% of the heteroduplex converted to the gapped form in 15 min). However, MutL-E32K failed to support the excision reaction. The mutant protein is therefore defective with respect to activation of both the MutH endonuclease and the mismatch repair excision system.
MutL-E32K Retains Its Ability to Interact with MutS-The in vitro inhibitory properties of MutL-E32K described above are consistent with the dominant negative phenotype associated with this amino acid substitution in vivo (28). Whereas such effects might be due to the formation of mixed dimers of wild type and mutant polypeptides, we regard this possibility as unlikely on biological and biochemical grounds. In contrast to other characterized, dominant mutL mutations, the dominant negative behavior of the mutL705 (E32K) allele was not suppressed by a gene dosage increase in the wild type allele (28). Furthermore, experiments described above demonstrate that MutL-E32K forms a stable homodimer and inhibits the overall mismatch repair reaction in cell extracts at low molar ratios relative to the wild type protein.
An alternate explanation for MutL-E32K dominant negative effects invokes interaction of the E32K homodimer with other repair activities, leading to dead-end complexes. Because a MutS-MutL interaction has been demonstrated by protein affinity chromatography (36), we have used this method to assess potential interactions between MutL-E32K and other components of the methyl-directed system. MutL and MutL-E32K were retained to a similar degree on a MutS column (Fig. 6A). Serum albumin was not retained by the MutS column, nor did MutL bind to a serum albumin column (Fig. 6, B and C), confirming the specificity of this method. g of bovine serum albumin was applied to a MutS affinity column as in A. C, 34 g of MutL was applied to a column otherwise identical to those above except that the protein linked to the support was BSA rather than MutS. Individual fractions were subjected to electrophoresis in the presence of SDS, and gels were visualized by Western blot (A and C) or by staining with Coomassie Blue (B). The curves show the relative masses of the eluted proteins as determined by densitometric analysis, with the left ordinate corresponding to wash through material and the right ordinate corresponding column bound protein that was eluted by high salt (note scale differences between the two ordinates). Inset figures show typical gel results. and is required for mismatch-and MutS-dependent activation of MutH and helicase II on heteroduplex substrates (12,13). MutL also activates DNA helicase II and MutH on model substrates and interacts with both proteins (19 -22). As shown in Fig. 7 (A and C) neither MutH nor DNA helicase II (the uvrD product) bound detectably to the MutS affinity support. However, both proteins were bound by the column, when loaded in the presence of either MutL or MutL-E32K, and the presence of ATP did not significantly alter the outcome observed in the presence of wild type MutL (Fig. 7, B and D). Thus, MutS can form ternary complexes with MutL and MutH or MutL and DNA helicase II in the absence of DNA, and the MutL-E32K mutation does not interfere with formation of these assemblies as judged by protein affinity chromatography. It therefore seems likely that the in vivo and in vitro dominant negative effects associated with this mutation are because of the formation of dead-end complexes in which the MutL-E32K protein is unable to transduce a signal from MutS that otherwise results in mismatch-dependent activation of the MutH d(GATC) endonuclease or the unwinding activity of helicase II. DISCUSSION MutL function has been previously implicated in the initiation and excision steps of methyl-directed mismatch repair. MutL is required along with MutS for the mismatch-dependent activation of the MutH d(GATC) endonuclease (12), and both proteins are necessary for the activation of DNA helicase II that occurs at a strand break in an incised heteroduplex (13,19). These observations have led to the suggestion that MutL serves to interface mismatch recognition by MutS to activation of downstream repair functions (17, 18). Indeed, MutL dramatically enhances the activity of DNA helicase II on conventional helicase substrates in a MutS-independent manner, an effect that is evident when the two proteins are present at comparable concentrations (19), and also activates MutH in the absence of MutS under some conditions (20,22). The finding that the latter effect is supported by nonhydrolyzable ATP analogues implies that nucleotide hydrolysis is not necessary for MutH activation scored by this assay.
The analysis of the dominant negative MutL-E32K mutant protein described here confirms and extends these observations. The finding that this nucleotide binding site mutation renders the protein nonfunctional in the MutS-and MutL-dependent activation of MutH and the mismatch repair excision system implicates function of the MutL ATP hydrolytic function in multiple steps of the reaction. Although it could be argued that these effects are a consequence of secondary effects of the mutation on the structure of the protein, we regard this possibility as unlikely for several reasons. The mutant protein maintains the native dimeric structure characteristic of MutL in solution (10), and as in the case of wild type MutL, the mutant protein is able to form ternary complexes with MutS and MutH and with MutS and DNA helicase II. In fact, the finding that MutL-E32K behaves as a dominant negative inhibitor of MutH activation and the overall mismatch repair reaction strongly argues for retention of structural integrity. The simplest explanation for the inhibitory effects of the mutant protein, in the context of the other observations above, is that it supports repair complex assembly up to a point but is unable to couple ATP binding or hydrolysis to activation of required downstream functions. Although effects of other repair proteins on the weak MutL ATPase have not been addressed, the hydrolytic activity of the protein is enhanced substantially by single-stranded DNA (Ref. 30 and Fig. 1). Whereas the k cat for ATP hydrolysis by MutL-E32K is reduced somewhat relative to that observed with wild type protein, the most obvious defect associated with the mutant protein is its failure to respond significantly to a single strand DNA cofactor. Although significance of this effect is uncertain at present, this finding may bear on the mechanism by which MutL activates DNA helicase II. Although it is clear that MutL functions with MutS to activate helicase II initiation at a strand break (13), it has also been suggested that MutL serves as a helicase activator during the subsequent course of unwinding by increasing the processive behavior of the protein (19). Because MutL and helicase II interact physically (19,21), the MutL DNA binding site may function in this respect via interaction with a single strand product of helicase action. for advice and suggestions during the course of this work. We also thank Keith Bjornson and Leonard Blackwell for comments on the manuscript.