Hydrolytically deficient MutS E694A is defective in the MutL-dependent activation of MutH and in the mismatch-dependent assembly of the MutS.MutL.heteroduplex complex.

The roles of ATP binding and hydrolysis by MutS in mismatch repair are poorly understood. MutS E694A, in which Glu-694 of the Walker B motif is substituted with alanine, is defective in hydrolysis of bound ATP and has been reported to support MutL-dependent activation of the MutH d(GATC) endonuclease in a trans DNA activation assay (Junop, M. S., Obmolova, G., Rausch, K., Hsieh, P., and Yang, W. (2001) Mol. Cell 7, 1-12). Because the MutH trans activation assay used in these previous studies was characterized by high background and low efficiency, we have re-evaluated the activities of MutS E694A. In contrast to native MutS, which can be isolated in a nucleotide-free form, purified MutS E694A contains 1.0 mol of bound ATP per dimer equivalent, and substoichiometric levels of bound ADP (0.08-0.58 mol/dimer), consistent with the suggestion that the ADP.MutS.ATP complex comprises a significant fraction of the protein in solution (Bjornson, K. P. and Modrich, P. (2003) J. Biol. Chem. 278, 18557-18562). In the presence of Mg2+, endogenous ATP is hydrolyzed with a rate constant of 0.12 min-1 at 30 degrees C, and hydrolysis yields a protein that displays increased specificity for heteroduplex DNA. As observed with wild type MutS, ATP can promote release of MutS E694A from a mismatch. However, the mutant protein is defective in the methyl-directed, mismatch- and MutL-dependent cis activation of MutH endonuclease on a 6.4-kilobase pair heteroduplex, displaying only 1 to 2% of the activity of wild type MutS. The mutant protein also fails to support normal assembly of the MutS.MutL.DNA ternary complex. Although a putative ternary complex can be observed in the presence of MutS E694A, assembly of this structure displays little if any dependence on a mismatched base pair.

Mismatch recognition by MutS initiates mismatch repair in bacterial cells (1)(2)(3). MutS recruits MutL to the heteroduplex in an ATP-dependent fashion (4 -6), and assembly of the MutL⅐MutS⅐heteroduplex ternary complex is sufficient to activate the MutH endonuclease, which incises the unmethylated strand at a hemimethylated d(GATC) strand signal (7). The ensuing strand break, which may reside either 3Ј or 5Ј to the mismatch, serves as the site for initiation of excision by a system comprised of DNA helicase II and an appropriate 3Ј to 5Ј or 5Ј to 3Ј single-strand specific exonuclease (8 -13). This bidirectional excision capability implies that the mismatch repair system must establish the relative orientation of the mismatch and strand break that directs the reaction. This is necessary to ensure loading of a 3Ј to 5Ј excision system when the nick resides 3Ј to the mispair and a 5Ј to 3Ј system when it is located 5Ј to the mismatch.
In addition to its mismatch binding site, MutS has a carboxyl-terminal ATPase that is required for function of the protein in mismatch repair (7,14,15), and this element is conserved in MutS homologs in higher cells (16 -20). Structural studies have shown that MutS is a member of the ABC (adenine nucleotide binding cassette) family (21)(22)(23), which is largely comprised of proteins that couple the energy of ATP hydrolysis to transport of molecules across biological membranes (24,25).
Several models for ATPase function in MutS homolog action have been suggested. One class of mechanism is based on a variety of observations indicating that in the presence of ATP, MutS homologs can leave a mismatch by movement along the helix (26 -29). This movement is postulated to link mismatch recognition to activation of downstream events at the strand break that directs excision, and can in principle account for the orientation-dependent loading of the excision system at the strand break that is necessitated by the bidirectional nature of the repair system. Two types of mechanisms have been proposed to explain ATP-dependent movement of MutS homologs along the DNA contour. One model posits that movement depends on ATP hydrolysis by the DNA-bound protein. The alternate molecular switch model postulates a G-protein like mechanism whereby binding of a MutS⅐ADP complex to a mismatch promotes ATP exchange for ADP, with the resulting MutS⅐ATP complex diffusing freely along the helix. Evidence consistent with both the hydrolytic model (26, 27, 29 -31) and the molecular switch model (28,32) is available.
A distinct role for ATP binding and hydrolysis has been proposed based on use of a trans assay for MutH activation (23). In this work, a mismatch on one oligonucleotide duplex was shown to lead to MutH activation and d(GATC) incision on a second synthetic duplex in a reaction dependent on MutL and MutS. These observations have led to the suggestion that once mismatch recognition occurs, MutS remains bound to the mispair during the course of repair, with mismatch-strand signal interaction mediated by DNA bending (23). In this proposal, ATP binding and hydrolysis by DNA-bound MutS functions in a kinetic proofreading mechanism that serves to verify mismatch recognition. Once bound to a putative mismatch, MutS is postulated to bind ATP. If the complex involves a misrecognition event, MutS is released and ATP is hydrolyzed to ADP. However, if MutS resides at a bona fide mismatch, ATP binding serves to verify mismatch recognition and is sufficient for activation of downstream repair activities in the absence of hydrolysis (23). The latter feature of the model is based on the experimental observation that MutS E694A, which is defective in ATP hydrolysis, supports MutH activation in the trans assay. However, the trans activation reactions on which this conclusion is based are several hundredfold less efficient than cis activation that occurs on 6,400-base pair heteroduplexes that have been used to study methyl-directed repair in vivo and in vitro (2,7,33). When normalized to input MutH, the rate of the cis reaction is about 0.1 min Ϫ1 per mol of MutH (7), whereas the corresponding value for the trans reaction calculated from the data of Junop et al. (23) is about 0.0002 min Ϫ1 per mol. Furthermore, in contrast to the cis reaction (7), the trans activation reaction employed in these earlier studies (23,34) is subject to high background d(GATC) cleavage in the absence of MutS or a mismatched base pair (Ϸ45% of that observed when these required repair components are present). Corresponding values for the cis assay are Ͻ3 and Յ10%, respectively (7). For these reasons we have re-evaluated the ability of MutS E694A to support MutH activation using the cis assay. In contrast to results obtained with the trans assay, we find MutS E694A to be defective in MutH activation, a defect that is probably due to its inability to support mismatch-dependent assembly of a MutL⅐MutS⅐heteroduplex complex.

EXPERIMENTAL PROCEDURES
Proteins and DNAs-The expression plasmid MutSE694A/pET15b for a His 6 -tagged version of Escherichia coli MutS E694A (23) was generously provided by Peggy Hsieh (NIDDK, National Institutes of Health). An expression plasmid for a non-tagged version of the mutant protein was constructed by insertion of a NcoI-BamHI fragment containing COOH-terminal MutS E694A sequences into NcoI-and BamHIcleaved pVB661. The latter plasmid is a derivative of pET3a containing a 934-bp NdeI fragment that corresponds to NH 2 -terminal mutS sequences beginning with the initiating ATG located within the NdeI recognition site. Sequence analysis of the entire mutS gene of the resulting plasmid, SpET3a-E694A, confirmed the presence of the E694A mutation within an otherwise native sequence that lacks the His 6 tag. E. coli strain BL21(DE3) mutS::Tn10 pLysS containing SpET3a-E694A was grown at 37°C in a BioFlow 4500 fermentor (New Brunswick Scientific) to an A 590 of 5.6 in 22 liters of supplemented L broth (0.1 M KPO 4 , pH 7.4, 1% (v/v) glycerol, 22 g/liter yeast extract, 11 g/liter tryptone, 10 g/liter NaCl, 4 mg/liter thymine, and 10 mg/liter thiamine) with O 2 aeration. After addition of isopropyl-1-thio-␤-D-galactopyranoside to 0.5 mM, growth was continued for 2 h at 37°C (final A 590 ϭ 13.8) and cells were harvested by centrifugation. Use of BL21(DE3) mutS::Tn10 as host for expression of the mutant protein precludes contamination of MutS E694A isolates with wild type MutS expressed by the cellular gene.
E. coli MutS and MutS E694A were purified as described (30) except that chromatography on a heparin affinity column was added as a last step. Pooled fractions eluting from the hydroxylapatite column (75-150 mg, ϳ0.5 mg/ml) were dialyzed against 20 mM KPO 4 , pH 7.4, 100 mM KCl, 1 mM EDTA, 1 mM dithiothreitol and loaded onto a Heparin Hi-Trap column (Amersham Biosciences, 5 ml) equilibrated with 20 mM KPO 4 , pH 7.4, 100 mM KCl, 0.1 mM EDTA, 1 mM dithiothreitol. After a wash with 20 ml of starting buffer at 1 ml/min, the column was eluted with a 40-min gradient of KCl (100 -640 mM) buffered as above. MutS E694A eluted from the column as a single peak at about 280 mM KCl. Wild type MutS eluted in two peaks at 230 and 330 mM KCl. The earlier eluting material, which was the minor component, displayed reduced activity and was discarded. MutS E694A fractions, or those corresponding to the later eluting peak of wild type protein, were pooled, dialyzed, and stored as concentrated stocks (100 -400 M as monomer) in 20 mM KPO 4 , pH 7.4, 150 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, 50% (v/v) glycerol at Ϫ20°C. Preparations obtained in this manner were estimated to be greater than 98% pure as judged by electrophoresis in the presence of sodium dodecyl sulfate. MutS and MutS E694A concentra-tions were determined from the absorbance at 280 nM using a monomer extinction coefficient of 69,240 calculated from primary amino acid sequence (35). Unless specified otherwise, MutS, MutL, and MutH concentrations are expressed as monomer equivalents.
As described under "Results," preparations of MutS E694A contain 1 mol of bound ATP per dimer equivalent and substoichiometric quantities of bound ADP. Analysis of individual fractions indicated that ADP content of the protein decreased across the hydroxylapatite elution profile. To obtain MutSE E694A isolates of differing ADP content, the hydroxylapatite peak was in some cases split into two pools corresponding to early and late eluting fractions. These two pools were separately subjected to chromatography on Heparin Hi-Trap columns as described above. Forty-one and 201-bp G-T heteroduplexes and otherwise identical A⅐T homoduplexes tagged with 5Ј-terminal biotin at one end were prepared as described previously (27,30).
Adenine Nucleotide Content of Isolated Proteins-Adenine nucleotide content of MutS and MutS E694A was determined using a bioluminescent assay kit (Sigma). Proteins were diluted to 400 nM (as monomer) with 5 mM Tris-HCl, pH 8.0, 1 mM EDTA, heated to 92°C for 5 min, and transferred to an ice bath for 15 min. After centrifugation through Nanosep MF 0.45-m filters (Pall Corporation, Ann Arbor, MI), samples of the filtrate were assayed directly for ATP content. ADP was determined after conversion to ATP by incubation of filtrate samples for 45 min at room temperature in reactions containing 67 mM Tris acetate, pH 7.75, 10 mM MgCl 2 , 1.5 mM KCl, 3.2 mM EDTA, 0.18 mM phosphoenolpyruvate, 5 mM 2-mercaptoethanol, and 2 units of pyruvate kinase (U. S. Biochemical Corp.; desalted prior to use by passage through Sephadex G-25 equilibrated in 88 mM Tris acetate, pH 7.8). ADP content was calculated as the difference between the ATP content of samples treated with pyruvate kinase and those not treated. Nucleotide content was determined at MutS concentrations of 12-200 nM (as monomer). Assay response was linear over ADP and ATP concentration ranges of 1.5 to 200 nM and 3 to 400 nM, respectively.
Nucleotide Removal from MutS E694A-Two methods were used to deplete MutS E694A isolates of endogenous nucleotide. In Method A the protein was incubated at room temperature for 5 min in 1.5-ml reactions containing 20 mM Tris-HCl, pH 7.6, 100 mM KCl, 1 mM dithiothreitol, 0.2 mM EDTA, 5 mM MgCl 2 , 2 M MutS E694A, followed by dialysis at 4°C against 1 liter of 20 mM KPO 4 , pH 7.4, 150 mM KCl, 1 mM dithiothreitol, 1 mM EDTA for 6 h and against a second 1.5-liter change for an additional 26 h. MutS E694A preparations treated in this manner contained about 0.02 mol of residual ATP and Ͻ0.01 mol of ADP per mol of monomer. In Method B, MutS E694A (20 M) was incubated in 2.5-ml reactions containing 20 mM bis-Tris 1 HCl, pH 7.6, 150 mM KCl, 1 mM MgCl 2 , 0.1 mM ZnCl 2 , 1 mM dithiothreitol, 5% glycerol, and 31 units of biotinylated calf intestinal alkaline phosphatase (Pierce) for 2 h at 4°C. The reaction mixture was passed over a streptavidin-agarose column (1 ml, Pierce) equilibrated with reaction buffer. Material passing through the column was dialyzed at 4°C for 3 h against 2 liters of 20 mM KPO 4 , pH 7.4, 100 mM KCl, 1 mM dithiothreitol, 1 mM EDTA, and loaded onto a 1-ml Heparin Hi-Trap column, which was washed and eluted as described above. Residual nucleotide content was similar to that obtained with Method A (Ϸ0.02 mol of ATP and Ͻ0.01 mol of ADP per mol monomer).
Surface plasmon resonance spectroscopy (SPRS) was performed on a BIAcore 2000. Streptavidin SA sensor chips were derivatized with about 150 resonance units of biotin-derivatized 41-or 201-bp heteroduplex and homoduplex as indicated. MutS or MutS E694A in 25 mM Tris-HCl, pH 7.6, 1 mM dithiothreitol, 100 mM KCl, 0.005% surfactant P20 (Buffer B) containing 5 mM MgCl 2 and nucleotides as indicated were flowed across the SA chip at 20 l/min. Measurements were performed at 25°C, and samples were maintained at 4°C prior to injection. Chips were regenerated by a 20-l injection of 0.5% sodium dodecyl sulfate. DNA binding isotherms for MutS and MutS E694A were determined by titration of chip-bound DNA with increasing concentrations of MutS or MutS E694A. Maximum binding values obtained in this manner were analyzed as a function of protein concentration as described previously (30). We have previously found that isolated MutS is free of ATP but can contain variable, substoichiometric quantities of bound ADP (Ͻ0.01-0.15 mol/mol of monomer) (36). 2 Recently modified purification methods (Ref. 30 and "Experimental Procedures") typically yield MutS isolates that are free of detectable nucleotide. Preparations isolated as described under "Experimental Procedures" contained Ͻ0.01 mol of adenine nucleotide per mol of MutS monomer. By contrast, near homogeneous preparations of MutS E694A contained 0.50 Ϯ 0.03 mol of ATP and 0.04 -0.29 mol of ADP/mol of monomer (four independent preparations; as noted under "Experimental Procedures," chromatography on hydroxylapatite leads to partial resolution of E694A species according to their ADP content). The presence of the ATP in near homogenous isolates of the mutant protein presumably reflects the hydrolytic defect conferred by the E694A mutation. Because MutS exists as dimers and tetramers in solution and these oligomers harbor two classes of nucleotide binding sites that display specificity for ADP or ATP (22,31,38,39), the simplest interpretation of the composition of bound nucleotide in MutS E694A preparations is that the triphosphate site is fully occupied in these isolates, but the diphosphate site is only partially populated. ATP and ADP in these preparations are tightly bound as 70 -80% of both nucleotides were retained upon 7 h dialysis against buffers containing 1 mM EDTA and 150 mM KCl or 1 M NaCl (not shown). However, incubation of the protein in the presence of 5 mM MgCl 2 followed by dialysis ("Experimental Procedures") resulted in the loss of 90 -95% of bound nucleotide.

MutS E694A Is Defective in ATP Hydrolysis and
The latter observation suggested that release of endogenous nucleotides may depend on hydrolysis of bound ATP. Indeed, incubation of MutS E694A in the presence of Mg 2ϩ resulted in the hydrolysis of the bound ATP with the concomitant appearance of ADP. As shown in Fig. 1, this event is well described by a single exponential with a rate constant of 0.12 min Ϫ1 at 30°C. The simplest interpretation of this value is that it corresponds to the intrinsic ATP hydrolytic rate constant for the mutant protein. This parameter is almost identical to the k cat for steady state ATP hydrolysis by MutS E694A at 30°C (see above), which is 0.08 min Ϫ1 when expressed per dimer, the minimum MutS functional unit (22,26,37,38). The similarity of these values indicates that hydrolytic chemistry is largely rate-limiting for turnover by MutS E694A. This differs from wild type MutS, for which ATPase turnover in the absence of DNA is governed by product release (36). The magnitude of the hydrolytic defect conferred by the E694A substitution can be estimated by comparing the apparent hydrolytic rate constant of 0.12 min Ϫ1 at 30°C for the mutant protein with the rate constant of 80 min Ϫ1 at 4°C that governs the pre-steady-state burst of ATP hydrolysis by wild type MutS (36).
MutS E694A Is Defective in the MutL-dependent Activation of the MutH Endonuclease-Initiation of E. coli mismatch repair depends on the mismatch-, MutS-, and MutL-dependent activation of the MutH d(GATC) endonuclease (7). This activity has been scored in vitro using heteroduplexes in which the mismatch and d(GATC) site reside on the same DNA molecule (6,7,37), and by a trans activation assay in which the two DNA sites reside on separate synthetic oligonucleotide duplexes (23,34,40). The finding that a His-tagged version of MutS E694A supports MutH activation in the trans assay has led to the suggestion that ATP binding by MutS is sufficient for activation of downstream activities (23). In view of the low efficiency and high background of the trans assay on which this conclusion is based (23, 34), we have re-examined the ability of MutS E694A to support MutS-and MutL-dependent cis-activation of MutH on a 6,400-bp G-T heteroduplex containing a single hemi-modified d(GATC) sequence (7). As shown in Fig. 2, MutS E694A is defective in the cis activation of MutH on this DNA, displaying a specific activity of only 1-2% that of the wild type protein. The magnitude of this defect is similar to the ATP hydrolytic defect of the mutant protein mentioned above (k cat of 0.06 min Ϫ1 at 37°C for MutS E694A as compared with 2.9 min Ϫ1 for wild type MutS expressed per monomer).
Mismatch Recognition by MutS E694A and Nucleotide Effects on MutS E694-DNA Interaction-MutS E694A has been shown to recognize mismatched base pairs with modest specificity (23), but affinities of the protein for heteroduplex and homoduplex DNAs have not been reported (23,34). As shown in Table I about 10-fold less than that of the wild type protein, and this is associated with limited heteroduplex specificity, particularly in isolates with low ADP content. However, analysis of binding in the presence of Mg 2ϩ resulted in a significant increase in heteroduplex affinity, and in the case of low ADP preparations, enhanced specificity as well.
The differential modulation of heteroduplex affinity and specificity of wild type MutS by adenine nucleotides suggests an explanation for these observations. Whereas the wild type protein exhibits high mismatch affinity and specificity in the absence of nucleotide or in the presence of ADP, affinity and specificity are reduced substantially when binding is scored in the presence of nonhyrolyzable AMP-PNP⅐Mg 2ϩ or ATP (no Mg 2ϩ ) (30). As described above, isolated MutS E694A contains 1 mol of bound ATP per dimer and variable amounts of bound ADP. We think it likely that the low heteroduplex affinity of MutS E694A that is observed in the absence of Mg 2ϩ reflects the presence of bound ATP and that specificity differences observed with different isolates are because of differential occupancy of the MutS ADP site in accord with the different levels of the diphosphate in different preparations. Because MutS E694A slowly hydrolyzes endogenous ATP when Mg 2ϩ is present (Fig. 1), the increased heteroduplex affinity and specificity observed in the presence of this divalent cation is probably because of hydrolysis of endogenous ATP. This is a reasonable possibility because the average MutS E694A sample was incubated in the presence of Mg 2ϩ for about 1 h at 4°C prior to automated injection into the BIAcore used for binding constant determination ("Experimental Procedures"). Indeed, nucleotide-depleted samples of MutS E694A (Ͻ0.05 mol of adenine nucleotide/mol of dimer), prepared by hydrolysis of endogenous nucleotide or by phosphatase treatment ("Experimental Procedures") display heteroduplex affinities and specificities that differ from those of wild type protein by only a factor of two, irrespective of the presence of Mg 2ϩ (Table I).
Previous studies have demonstrated that the interaction of MutS with DNA is modulated by adenine nucleotides (4,26,30), and as expected, the converse is also true (36,41). We have therefore employed SPRS analysis to compare adenine nucleo-tide effects on MutS E694A-and MutS-DNA interaction. MutS⅐heteroduplex complexes formed in the absence of nucleotide or in the presence of ADP⅐Mg 2ϩ undergo rapid and complete dissociation upon subsequent challenge with ATP⅐Mg 2ϩ (30). This effect is reproduced in Fig. 3 (panels A and B), and it can be seen that wild type MutS and MutS E694A respond in a similar manner to triphosphate challenge. MutS⅐heteroduplex complexes formed in the presence of ATP (ϮMg 2ϩ ) behave differently. The presence of the triphosphate reduces the specificity of MutS-DNA interaction because of a reduction in mismatch affinity, and in contrast to MutS⅐heteroduplex complexes that assemble in the presence of ADP or absence of nucleotide, those formed in the presence of ATP are resistant to dissociation upon buffer wash and subsequent challenge with ATP⅐Mg 2ϩ (30) . Fig. 3 (panels C and D) demonstrates that complexes of MutS E694A with heteroduplex DNA behave like wild type MutS in this respect. These observations confirm a previous study (30) that has shown that the lifetime of MutS⅐heteroduplex complexes upon ATP challenge depends upon the nature of the nucleotide present during the binding phase of the reaction. This effect requires the presence of a physical barrier on at least one end of the heteroduplex, as is the case in the BIAcore experiments of Fig. 3 in which the heteroduplex is attached to the sensor chip via biotin-streptavidin linkage.
Thus, although slightly compromised with respect to heteroduplex affinity, MutS E694A is clearly capable of mismatch recognition, and assembly of MutS E694A⅐heteroduplex complexes responds to adenine nucleotides in a manner similar to that of wild type MutS. Consequently, the causative basis for the failure of the mutant protein to support MutH activation is not evident at the level of its interaction with a mismatch.
MutS E694A Fails to Support Mismatch-dependent Assembly of a MutL⅐MutS⅐DNA Ternary Complex-A ternary complex comprised of heteroduplex DNA, MutS, and MutL is believed to be a key intermediate in the initiation of mismatch repair. This ternary complex has been visualized in the bacterial and human systems by DNase footprinting, SPRS, magnetic bead assay, and gel shift analysis (4,5,42). In the SPRS assay, this complex is manifested as an enhanced mass of DNA-bound protein when chip-bound heteroduplex is simultaneously exposed to MutS and MutL (5) or MutS␣ and MutL␣ (42) in the presence of ATP. The behavior of wild type MutS and MutL in this assay is illustrated in panels A and B of Fig. 4. In the presence of ATP, the mass of G-T heteroduplex-bound protein was increased dramatically when MutS and MutL are both present as compared with that observed in the presence of MutS alone (panel A). In the absence of other proteins, MutL does not bind at detectable levels to heteroduplex under these conditions (5). 3 These results are identical to those reported by Galio et al. (5), who showed that this mass enhancement is because of MutS-dependent assembly of a MutS⅐MutL complex on heteroduplex DNA. Control experiments with an A⅐T homoduplex demonstrated a strong dependence of ternary complex assembly on a mismatched base pair (compare panels A and B). The substantial increase in heteroduplex-bound mass observed in the presence of both MutS and MutL, as compared with that observed with MutS alone (Fig. 4A), might suggest that each MutS molecule recruits multiple copies of MutL to the heteroduplex. This is not necessarily the case, however, because MutL may enhance the steady-state level of heteroduplexbound MutS. The nature of the BIAcore assay does not permit distinction between these possibilities. Although enhanced protein binding to DNA was also observed in the presence of MutS E694A and MutL, an effect that was ATP-dependent, the extent of binding to heteroduplex and homoduplex DNAs did not differ significantly (Fig. 4, panels C and D). Thus, whereas the mutant protein supports the apparent formation of a ternary MutL⅐MutS E694A⅐DNA complex, assembly of this structure does not depend on the presence of a mismatched base pair. The ternary complexes that are assembled on heteroduplex and homoduplex DNA in the presence of MutS E694A also differ from those obtained with wild type MutS with respect to behavior upon subsequent buffer wash. As can be seen, MutS E694A ternary complexes dissociate from DNA more slowly than those assembled in the presence of the wild type protein (compare panel A of Fig. 4 with panel C, and panel B with panel D). The inability of MutS E694A to support normal assembly of the MutL⅐MutS⅐DNA ternary complex provides a simple explanation for the failure of the mutant protein to support MutL-dependent activation of the MutH endonuclease as described above.

DISCUSSION
Previous work using a trans assay for MutH activation has led to the conclusion that hydrolytically defective MutS E694A supports MutL-dependent activation of MutH (23,34). However, use of a cis assay for MutH activation indicates that MutS E694A is defective in this regard, with the magnitude of the defect comparable with the observed reduction in the k cat for ATP hydrolysis. Differences in the experimental conditions used in these two cases suggest a possible explanation for this discrepancy. The cis assay for MutH activation is typically performed in the presence of 2.4 nM heteroduplex, 1 nM MutH, 24 nM MutL, and 30 -300 nM MutS (7, 37) (Fig. 2). By contrast, the trans assays that led to the conclusion that MutS E694A is active in MutH activation employed much higher DNA and protein concentrations: 50 nM DNA, 1,000 nM MutH, 500 nM MutL, and 250 -500 nM MutS or MutS E694A (23). The high MutH and MutL concentrations used in the trans reactions   (Table I) may permit assembly of a MutH activation complex by a nonbiological pathway that bypasses key assembly step(s) that may otherwise depend on a functional MutS ATP hydrolytic center. That this may be the case is suggested by the low efficiency and high background characteristic of the trans assays employed in the previous study. When normalized to input MutH, the efficiency of trans activation is less than 1% of that observed with the cis reaction (7,23). Whereas conditions for more efficient trans activation have been described (40), analysis of MutS E694A for its ability to support this reaction has not been reported. The finding that MutS E694A is defective in cis MutH activation thus raises questions concerning the validity of the conclusion based on the trans assay that ATP binding by MutS is sufficient for activation of downstream activities as postulated by the mismatch verification model (23).
As discussed above, the two other models for MutS homolog ATPase function invoke nucleotide-dependent movement of the protein from the mismatch along the DNA contour (26 -29). These two models differ in that one invokes ATP binding and hydrolysis by DNA-bound MutS/MutS␣ during the course of movement (26,27), whereas the alternate molecular switch model postulates mismatch recognition by the MutS/ MutS␣⅐ADP complex, exchange of ATP for ADP, and diffusion of the MutS/MutS␣⅐ATP complex along the helix (28). These models derive from the direct visualization of the temporal evolution of MutS⅐DNA complexes (26) and study of the dynamics of MutS interaction with linear heteroduplexes (27)(28)(29)(30). Use of the latter approach has shown that ATP⅐Mg 2ϩ greatly reduces the steady-state level of heteroduplex-bound MutS (or MutS␣) and promotes dissociation of preformed complexes from a linear heteroduplex. However, placement of streptavidin blocks at the heteroduplex termini prevents ATP-promoted dissociation resulting in elevated steady-state levels of MutS(MutS␣)⅐heteroduplex complexes.
Results of the MutS E694A experiments described here can be considered in the context of the hydrolysis-dependent translocation and molecular switch models for MutS ATPase function. The finding that the mutant protein is defective in MutLdependent activation of MutH and mismatch-dependent assembly of the MutS⅐MutL⅐DNA ternary complex suggests that ATP hydrolysis by DNA-bound MutS may be necessary for its function in mismatch repair. Analysis of the interaction of human MutS␣ with linear heteroduplexes derivatized with avidin end blocks has led to similar conclusions. Whereas ATP⅐Mg 2ϩ challenge of pre-formed complexes of MutS␣ with an end-blocked heteroduplex leads to formation of a long lived complex, challenge with AMP-PNP⅐Mg 2ϩ or ATP␥S⅐Mg 2ϩ results in dissociation (27). A similar observation has been made by Iaccarino et al. (29) using a hydrolytically defective form of human MutS␣ (Val substitution for Asp-1213 in MSH6) that retains the ability to bind ATP. In contrast to wild type MutS␣, pre-formed complexes of MutS␣(MSH6 D1213V) with an endblocked heteroduplex dissociate upon ATP⅐Mg 2ϩ challenge. Whereas results obtained with nonhydrolyzable ATP analogues, and the abnormal behavior of MutS E694A and MutS␣(MSH6 D1213V) might be attributed to conformational effects unrelated to an inability to hydrolyze triphosphate, the simplest explanation for this set of observations is that ATP hydrolysis by DNA-bound MutS/MutS␣ is required for function of the proteins in mismatch repair.
The MutS dimer displays intrinsic asymmetry (21,22), and the nucleotide binding sites within E. coli MutS dimers and tetramers are of two types that display differential specificities for ADP and nonhydrolyzable ATP analogues (22,31,37,39). Our finding that the triphosphate binding site is fully occupied (1 mol of ATP per dimer equivalent) and the diphosphate site partially populated (0.1 to 0.6 mol/dimer) in isolated MutS E694A is consistent with the previous demonstration that the two classes of nucleotide binding site within wild type MutS can be simultaneously occupied by ADP and a nonhydrolyzable ATP analogue (31). In particular, the presence of bound ATP and ADP in the isolates of the mutant protein is in agreement with the suggestion that the ternary ADP⅐MutS⅐ATP complex comprises a substantial fraction of the protein in solution. The existence of this ternary complex is incompatible with twostate models for MutS action that restrict the protein to only the ADP or ATP forms (28).