HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 276, Issue 30, 28291-28299, July 27, 2001

This Article Abstract Full Text (PDF) All Versions of this Article: 276/30/28291    most recent M103148200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Schofield, M. J. Articles by Hsieh, P. Search for Related Content PubMed PubMed Citation Articles by Schofield, M. J. Articles by Hsieh, P.

Interaction of Escherichia coli MutS and MutL at a DNA Mismatch^*

Mark J. Schofield, Sunil Nayak, Thomas H. Scott, Chunwei Du, and Peggy Hsieh

From the Genetics and Biochemistry Branch, NIDDKD, National Institutes of Health, Bethesda, Maryland 20892

Received for publication, April 9, 2001, and in revised form, May 18, 2001

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

MutS and MutL are both required to activate downstream events in DNA mismatch repair. We examined the rate of dissociation of MutS from a mismatch using linear heteroduplex DNAs or heteroduplexes blocked at one or both ends by four-way DNA junctions in the presence and absence of MutL. In the presence of ATP, dissociation of MutS from linear heteroduplexes or heteroduplexes blocked at only one end occurs within 15 s. When both duplex ends are blocked, MutS remains associated with the DNA in complexes with half-lives of 30 min. DNase I footprinting of MutS complexes is consistent with migration of MutS throughout the DNA duplex region. When MutL is present, it associates with MutS and prevents ATP-dependent migration away from the mismatch in a manner that is dependent on the length of the heteroduplex. The rate and extent of mismatch-provoked cleavage at hemimethylated GATC sites by MutH in the presence of MutS, MutL, and ATP are the same whether the mismatch and GATC sites are in cis or in trans. These results suggest that a MutS-MutL complex in the vicinity of a mismatch is involved in activating MutH.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Misincorporation of bases during DNA replication that escapes proofreading results in the formation of mismatches, either unpaired or mispaired bases. The DNA mismatch repair pathway targets these mismatches for correction and contributes as much as 1000-fold to the overall fidelity of DNA replication (reviewed in Refs. 1 and 2). The importance of this repair pathway is highlighted by the finding that mutations in mismatch repair genes segregate with the cancer predisposition syndromes hereditary nonpolyposis colon cancer and familial colorectal cancer (reviewed in Refs. 3 and 4). In addition, inactivation of mismatch repair genes by promoter methylation has been documented in some sporadic tumors (reviewed in Ref. 1).

Much of our current understanding of mismatch repair stems from studies of the Escherichia coli methyl-directed mismatch repair pathway where repair has been reconstituted in vitro using purified proteins (1, 2). MutS recognizes and binds to seven of eight possible base pair mismatches as well as one to four unpaired bases. These mismatches arise by misincorporation of deoxynucleotides or template slippage, respectively. In addition to binding DNA, all MutS proteins have an intrinsic ATPase activity and are members of the ATP binding cassette transporter superfamily (5). Recognition of the mismatch by MutS triggers subsequent steps of mismatch repair in which MutS together with MutL protein and ATP activates an endonuclease, MutH, that cleaves a transiently unmethylated daughter strand at hemimethylated GATC sites, thereby conferring strand specificity on the repair process. The mismatch-provoked strand scission constitutes a point of entry for helicase II (UvrD) and exonucleases that excise the daughter strand in the region spanning the GATC site and the mismatch. Gap repair by polymerase III, single-strand binding protein, and DNA ligase complete the process.

ATP binding and hydrolysis are essential for the function of MutS in MMR (reviewed in Ref. 1). Mutations in the ATP-binding site of E. coli MutS give rise to a dominant negative mutator phenotype (6). The interaction of MutS with MutL requires ATP (7-10), and the activation of MutH by MutS and MutL requires a functional MutS ATP-binding site (11). The mismatch-binding site and the ATP-binding site of MutS are separated by some 70 Å as revealed in the three-dimensional structures of Taq and E. coli MutS proteins (11-13). Nevertheless, mismatch binding and nucleotide binding and hydrolysis are tightly linked. Nucleotide cofactor is involved in the allosteric regulation of DNA binding such that the affinity of some MutS proteins for mismatches is reduced in the presence of ATP, and the rate of ATP hydrolysis by MutS can be modulated by DNA binding (reviewed in Refs. 1 and 3).

A critical problem in mismatch repair is understanding how binding to a mismatch by MutS triggers downstream repair events such as the activation of MutH endonuclease. In two models, migration of MutS away from the mismatch has been invoked in the search by MutS for enzymes involved in the excision step. In one model, the energy of ATP hydrolysis is used to fuel a unidirectional translocation of MutS away from the mismatch (8, 14). Support for this model derives, in part, from electron microscopic studies in which MutS is present at the base of -loop structures. In an alternate model, binding of MutS to mismatched DNA provokes the exchange of ADP for ATP in the protein resulting in the transformation of MutS from a mismatch-binding protein to a sliding clamp that diffuses away from the mismatch in search of other repair proteins (15, 16). Both models are consistent with experiments demonstrating that in the presence of ATP MutS dissociates from the ends of a linear heteroduplex but remains associated with the DNA if the ends are physically blocked by biotin-streptavidin complexes. A weakness of these models is that neither model assigns a role for MutL in signaling downstream repair events. More recently, it has been shown that E. coli MutS together with MutL and ATP can activate MutH for cleavage at hemimethylated GATC sites in trans, i.e. when the mismatch and the hemimethylated GATC sites reside on different DNA molecules (11). This finding suggests that MutS does not need to migrate from a mismatch to signal downstream repair steps.

Here, we characterize the ATP-dependent migration of MutS protein away from a mismatch, and we investigate the effect of MutL on the rate of dissociation of MutS from a heteroduplex DNA. Whereas MutS dissociates rapidly from a mismatch in the presence of ATP unless DNA ends are blocked, in the presence of MutL, MutS remains stably bound to the mismatch DNA. Examination of MutH cleavage at hemimethylated GATC sites in the presence of MutS, MutL, and ATP reveals that intramolecular reactions (in cis) occur with the same efficiency as reactions in trans. These findings suggest that a MutS-MutL complex remains bound to DNA in the vicinity of the mismatch to activate downstream repair proteins.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Protein Purification-- Taq MutS protein was purified from E. coli BL21(DE3) harboring pETMutS as described previously (17). E. coli His₆-tagged MutS, MutL, and MutH proteins were purified from E. coli HMS174(DE3)pLysS (Novagen) transformed with pTX412 (His₆-MutS), pTX417 (His₆-MutH), or pTX418 (His₆-MutL) (18). Transformants were grown at 37 °C in LB containing 100 µg/ml ampicillin to an A₆₀₀ of 0.5 and induced at 37 °C by the addition of isopropyl-1-thio--D-galactopyranoside to 0.1 mM. Cells were harvested after 3 h and resuspended in 35 ml/liter of culture of lysis buffer (20 mM HEPES, pH 7.8, 500 mM NaCl, 10% glycerol, 1 mM -mercaptoethanol) supplemented with Complete EDTA-free protease inhibitor (Roche Molecular Biochemicals) and stored at 80 °C. Thawed cells were lysed by sonication. Lysates were cleared by centrifugation and applied to a 5-ml metal ion chelating column (Amersham Pharmacia Biotech) pre-equilibrated with nickel ions. The column was washed with lysis buffer containing 60 mM imidazole and eluted with lysis buffer containing 300 mM imidazole. The buffer was exchanged by applying the eluted protein to a HiPrep 26/10 Desalting column (Amersham Pharmacia Biotech) pre-equilibrated with buffer A (20 mM HEPES, pH 7.8, 100 mM NaCl, 1 mM dithiothreitol, 1 mM EDTA, 20% glycerol, 1 µg/ml each leupeptin and pepstatin, and 1:1000 dilution of saturated phenylmethylsulfonyl fluoride). Protein was then applied to a MonoQ HR 10/10 column (Amersham Pharmacia Biotech) equilibrated in buffer A and eluted in buffer A with increasing NaCl concentration. His₆-tagged MutS eluted at 250-300 mM NaCl; MutL and MutH eluted at 120-150 mM NaCl. Protein fractions were brought to a 30% glycerol final concentration. Aliquots were stored at 4 °C for MutS and 20 °C for MutL and MutH. Proteins were judged to be 95% pure or greater by Coomassie staining after SDS-gel electrophoresis. Protein concentration was determined using the theoretical extinction coefficient as follows: for MutS, 76,874 M¹ cm¹; for MutL, 57,981 M¹ cm¹; and for MutH, 38,845 M¹ cm¹. Concentrations of proteins are those of dimers for MutS and MutL and monomers for MutH.

DNA Substrates-- Deoxyoligonucleotides were synthesized on a 394 DNA synthesizer (Applied Biosystems) and purified by gel electrophoresis on denaturing polyacrylamide gels or by reversed-phase high pressure liquid chromatography using an Oligo R3 column (Perspective Biosystems) followed by detritylation and chromatography on a C4 preparative scale column (Vydac). Oligonucleotides were 5'-end-labeled using [-³²P]ATP (PerkinElmer Life Sciences) and T4 polynucleotide kinase (Amersham Pharmacia Biotech). N⁶-Methyladenine phorphoramidites were purchased from Glen Research.

Deoxyoligonucleotides for doubly-blocked substrate plus and minus mismatch (see Fig. 1A) are as follows (T denotes the unpaired thymidine): a, 5'GCT ACA GGG ACC TAG CAA GGG GCT GCT ACC CTT CTG CCG T; b +1T bulge, 5'ACG GCA GAA GGG TAG CAG CCT GAG CGG TGG TTC CTT ATG GCA GCG TGC GCG TGA CGT TGC ATG CTA GCT AAG CTC CAA CCG TAC AAG TAT T; b no mismatch, 5'ACG GCA GAA GGG TAG CAG CCT GAG CGG TGG TTC CTT ATG GCA GCG GCG CGT GAC GTT GCA TGC TAG CTA AGC TCC AAC CGT ACA AGT ATT; c, 5'AAT ACT TGT ACG GTT GGA GCC TTG CTG GAT CCA ACG TAG G; d, 5'CCT ACG TTG GAT CCA GCA AGC ATG GCC ACT TAA GAA GTC C; e, 5'GGA CTT CTT AAG TGG CCA TGT TAG CTA GCA TGC AAC GTC ACG CGC CGC TGC CAT AAG GAA CCA CCG CTC AAC TCA ACT GCA GAC TCT TAC; f, 5'GTA AGA GTC TGC AGT TGA GTC CTT GCT AGG TCC CTG TAG C.

Deoxyoligonucleotides for MutH activation assays are listed below (see Fig. 5).

The linear substrate containing both the mismatch and a GATC site (Fig. 5A) consisting of a top strand with +1 bulge T is as follows: 5'ACG GCA GAA GGG TAG CAG CAC TGA GCG TGT GGT TCC TTA TGG CAA AGA AAC GTG ACG TTG CAT GCT AGC TAA GCT CGA TCC GTA CAA GTA TT; the bottom strand containing the N⁶-methyladenine denoted by 6 is as follows: 5'AAT ACT TGT ACG G6T CGA GCT TAG CTA GCA TGC AAC GTC ACG TTT CTT TGC CAT AAG GAA CCA CCG CTC AGT GCT GCT ACC CTT CTG CCG T.

Two different four-way junction substrates containing a mismatch and a GATC site were constructed from the following oligonucleotides (Fig. 5C): a (where 6 denotes N⁶-methyladenine (Glen Research)), 5'GGT CAT CGG CCA TG6 TCG ATT ACA TGC GAT TAG CAA GCC GCT GCT AAC CTG TGT TCG TTA CAA GGC ATA CGT AGA C; b, 5'GTC TAC GTA TGC CTT GTA ACG AAC ACA GGT TAG CAG CGA GAG CGG TCG TCC GAT TCC GAA CCG TCA GTT GCC AAC G; c, 5'CGT TGG CAA CTG ACG TGT TCG GAA TCG GAC GAC CGC TCT TCT CTT CTC GTA GTA GCT ATG CAC CCA GGA TCG AGC TT; c', 5'CGT TGG CAA CTG ACG GTT CGG AAT CGG ACG ACC GCT CTT CTC TTC TCG TAG TAG CTA TGC ACC CAG GAT CGA GCT T; d, 5'AAG CTC GAT CCT GGG TGC ATA GCT ACT ACG AGA AGA GAG CTT GCT AAT CGC ATG TAA TCG ATC ATG GCC GAT GAC C; and d', 5' AAG CTC GAT CCT GGT GTG CAT AGC TAC TAC GAG AAG AGA GCT TGC TAA TCG CAT GTA ATC GAT CAT GGC CGA TGA CC.

The doubly-blocked GATC substrate (Fig. 5B) uses oligonucleotides a, c, d and f as above plus the following: b, 5'ACG GCA GAA GGG TAG CAG CCT GAG CGG TGG TTC CTT ATG GCA AAG 6TC CGT GAC GTT GCA TGC TAG CTA AGC TCC AAC CGT ACA AGT ATT, where 6 denotes N⁶-methyladenine; and e, 5'GGA CTT CTT AAG TGG CCA TTA GCT AGC ATG CAA CGT CAC GGA TCT TTG CCA TAA GGA ACC ACC GCT CAG AAC TCA ACT GCA GAC TCT TAC.

Measurement of Dissociation Binding Constants-- MutS was incubated with 1 nM 5'-³²P-labeled duplex in binding buffer (20 mM HEPES, pH 7.8, 50 mM NaCl, 5 mM MgCl₂, 1 mM dithiothreitol, and 0.1 mg/ml bovine serum albumin) in 10 µl total volume. After 2 min at 37 °C for E. coli MutS and 65 °C for Taq MutS, 2 µl of 15% Ficoll (without dye) was added, and samples were loaded under voltage onto 5% native polyacrylamide gels (29:1 acrylamide:bisacrylamide) in 25 mM Tris-HCl, pH 8.3, 192 mM glycine, and 5 mM MgCl₂. After electrophoresis in the same buffer, gels were dried and quantitated on a Fuji BAS-1500 PhosphorImager. Data were analyzed as fraction bound versus protein concentration and were fitted by nonlinear regression analysis to an equation for a simple two-state binding process (19).

Measurement of the Half-life of MutS-DNA Complexes-- 50 nM MutS dimer was incubated with 1 nM 5'-³²P-labeled DNA in binding buffer (as described above) in a total volume of 30 µl. Under these conditions the dissociation constant for heteroduplex DNA ranged from 1 to 16 nM, so the majority of DNA was bound. Reactions were incubated between 1 and 5 min on ice and then for 1 min at 37 °C for E. coli MutS and at 65 °C for Taq MutS. Immediately before t = 0, an aliquot was removed, brought to 3% w/v Ficoll, and loaded onto a native 5% polyacrylamide gel under voltage as described above. At t = 0, 500 nM unlabeled heteroduplex DNA was added, and the reaction was brought to 1 mM ATP. After varying times aliquots were removed and electrophoresed on a native polyacrylamide gel as described above. Half-lives of complexes were calculated from rate constants determined from the slope of ln (fraction bound) versus time. The fit to simple exponential decay of MutS-DNA complexes was supported by the linearity of semi-log plots. For reactions performed in the presence of purified IHF¹ protein, kindly provided by Steve Goodman (20), it was included at a concentration of 65 nM. Where indicated, 100 nM MutL was present in the initial reaction or added at t = 0 along with the cold heteroduplex competitor.

DNase I Footprinting-- DNase I footprinting was performed in binding buffer with 1 nM of +1T bulge mismatch duplex, 5'-³²P-labeled on the strand containing the unpaired T. MutS (either Taq or E. coli) was present at 50 nM, MutL at 100 nM, and ATP at 1 mM in a reaction volume of 20 µl. Complexes were preformed by incubation for 1-5 min at 21 °C followed by the addition of 1 unit of DNase I (Life Technologies, Inc.). Reactions were terminated after 30 s by the addition of an equal volume of 50% formamide containing 50 mM EDTA. Samples were heated at 95 °C for 2 min and analyzed on 15% denaturing polyacrylamide gels.

MutH Endonuclease Activity Assays-- In cis reactions, the DNA substrate containing both a hemimethylated GATC site (5'-³²P-labeled on the unmethylated strand) and the mismatch was incubated at a concentration of 1 nM with 10 nM MutS, 10 nM MutL, 10 nM MutH, 1 mM ATP, and 70 nM unlabeled 36-bp homoduplex DNA in binding buffer (as described above) in a final volume of 30 µl. Trans-reactions were carried out under identical conditions except that they contained 1 nM ³²P-labeled GATC duplex and 1 nM mismatch duplex in addition to the 70 nM homoduplex DNA. Aliquots were removed at various times and added to an equal volume of formamide and brought to 1% w/v SDS and 50 mM EDTA. Samples were heated to 90 °C and analyzed on 8% denaturing urea polyacrylamide gels. After electrophoresis, gels were dried and quantitated on a Fuji BAS-1500 PhosphorImager.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

ATP-dependent Dissociation of MutS from Mismatched DNA Is Blocked by Four-way DNA Junctions-- The ATP-dependent translocation of MutS from a mismatch was studied using mismatch duplexes containing blocked ends. DNA duplexes, either perfectly base-paired (homoduplex) or containing a mismatch (heteroduplex), were engineered with four-way DNA junctions at both ends (see Fig. 1A). The distance between junctions was 50 bp. Unlike duplexes blocked with streptavidin-biotin complexes used in previous studies, duplexes blocked with four-way junctions could be reproducibly generated in high yield. Linear heteroduplex or homoduplex DNAs without blocked ends were used for comparison. The rate of dissociation of E. coli MutS from mismatched DNAs in the presence and absence of ATP was determined in gel electrophoretic mobility assays. MutS was preincubated with ³²P-labeled duplexes followed by the addition at time 0 of 1 mM ATP and a 500-fold molar excess of unlabeled heteroduplex DNA to prevent rebinding by MutS. Aliquots were removed after varying lengths of time and electrophoresed on native polyacrylamide gels (see "Materials and Methods"). Half-lives were calculated from the slopes of semi-log plots of the fraction bound versus time.

View larger version (30K): [in this window] [in a new window]   Fig. 1.   Four-way DNA junctions block ATP-dependent dissociation of MutS from a mismatch duplex. A, schematic of heteroduplex DNA substrates either unblocked or blocked at both ends by four-way junctions. Heteroduplexes contained an unpaired thymidine residue, 1T bulge. Sequences of strands labeled alphabetically are given under "Materials and Methods." * denotes strand containing 32P label. B, the dissociation of E. coli MutS from heteroduplex DNA was assessed in a polyacrylamide gel electrophoresis assay as described under "Materials and Methods." MutS was preincubated with the mismatch substrate followed by the addition at time 0 of a large excess of cold competitor DNA and 1 mM ATP as indicated. Data points are means of triplicate measurements with error bars showing standard deviations. C, quantitation of gel mobility shift assays. Data are plotted as percent bound versus time (left) or replotted as the logarithm of the fraction bound versus time (right) from which rate constants and half-lives were determined as described under "Materials and Methods."

The presence of four-way junctions at both ends of the duplex greatly retarded the ATP-dependent dissociation of MutS from the DNA (Fig. 1, B and C, and Table I). Whereas ATP induced a rapid dissociation of MutS from linear DNA resulting in the loss of 80-90% of complexes within 15 s, MutS dissociated much more slowly when ends were blocked with four-way junctions. The half-lives of MutS-DNA complexes from the blocked versus unblocked substrate was 1800 s versus an upper limit of 8 s, respectively (Table I). Similar results were obtained with Thermus aquaticus (Taq) MutS protein (Table I). Blocking both ends of mismatch DNAs with three-way junctions yielded the same result (data not shown). Thus, branched DNA structures blocked the ATP-dependent dissociation of MutS from DNA.

In the absence of ATP, MutS complexes were quite stable, and for E. coli MutS, the half-lives were 570 versus 410 s for the doubly-blocked and unblocked heteroduplexes, respectively. In the case of Taq MutS, half-lives for the doubly-blocked and unblocked heteroduplexes were 170 and 120 s, respectively. The drop in the number of bound complexes observed upon the addition of competitor mismatch DNA, evident in the difference at the 0 versus 15-s time points (Fig. 1C), represented the loss of nonspecifically bound MutS protein. The observation that MutS dissociated rapidly from linear duplexes while remaining associated with DNA when ends were blocked supports the notion that MutS migrates along the DNA helix in an ATP-dependent fashion.

An examination of the relative binding affinities of MutS for blocked and unblocked duplexes established that E. coli and Taq MutS proteins have no specific affinity for the four-way DNA junction. Thus, the presence of four-way DNA junctions had no effect on the apparent K_D values measured by polyacrylamide gel retardation assays in the case of either heteroduplex or homoduplex DNA substrates (Table II). In the case of Taq MutS, mismatch duplexes containing an unpaired thymidine were recognized about 1000 times better than homoduplexes. In the case of E. coli MutS, discrimination between heteroduplex and homoduplex DNA was substantially less, amounting to a 10-fold difference in relative affinities. Surprisingly, the presence of 1 mM ATP had little if any effect on the affinity of MutS for mismatch DNA, despite its huge effect on the stability of complexes reflected in the large increase in the rate of dissociation. Given the rate of ATP hydrolysis (17), <5% of the ATP would be expected to be cleaved over the duration of the experiment making it unlikely that ADP was having any significant effect on observed binding. One explanation for this result is that binding of ATP greatly accelerates the rate of formation of stable MutS-mismatch complexes such that it is very rapid compared with ATP-dependent translocation (see also Ref. 8). Additional experimentation will be required to understand the molecular basis for the apparent high affinity of MutS for mispairs in the presence of ATP given that ATP destabilizes MutS-mispair complexes.

Migration of MutS on Singly-blocked Duplexes-- Complexes of Taq MutS bound to doubly-blocked heteroduplexes were analyzed by DNase I footprinting in the absence and presence of ATP (Fig. 2A). In the absence of ATP, MutS protected about 10 bp on either side of the mismatch, as shown previously (21). In the presence of ATP, the footprint was no longer confined to the mismatch but now extended throughout the region of the duplex between the two junctions. This result was consistent with movement being bi-directional, i.e. any given MutS dimer was capable of moving in both directions. Unidirectional translocation would predict that MutS would migrate to one end of the duplex and remain there. Since only 50 bp separated the two junctions and the DNase I footprint of MutS bound to a mismatch extends roughly 20 bp, the directionality of movement was further addressed using heteroduplexes blocked at only one end.

View larger version (32K): [in this window] [in a new window]   Fig. 2.   Translocation of MutS is not unidirectional. A, DNase I footprinting of MutS-mismatch complexes. DNase I footprinting was performed on the 32P-labeled, doubly-blocked substrate shown in Fig. 1 in the presence and absence of Taq MutS and 1 mM ATP as described under "Materials and Methods." B, the dissociation of Taq MutS from mismatch substrates blocked at only one end with a DNA four-way junction. Half-lives were calculated as described above. C, schematic showing the effect of ATP on the DNA-binding of MutS. ATP causes MutS to diffuse away from the mismatch. Encounter with a junction results in the reversal of direction. * denotes position of 32P label.

The dissociation of MutS from heteroduplexes blocked by four-way junctions at either the left or the right end was monitored in polyacrylamide gel retardation assays as described above (Fig. 2B). If translocation were unidirectional and were equally likely to occur in either direction, then each substrate would remain half-bound. If translocation were unidirectional and occurred in only one direction, then one substrate would be fully bound, and the other substrate would be completely unbound. These predicted outcomes were derived from the enormous stability of MutS complexes on doubly-blocked duplexes with half-lives of 30 min as described above. Neither of these outcomes was observed. Instead, dissociation of MutS from singly-blocked duplexes was rapid and extensive and indistinguishable from that of unblocked duplexes. In the presence of ATP, 90% of bound complexes dissociated by 15 s (Table I). The simplest interpretation of these results is that, in the presence of ATP, MutS migrates away from the mismatch and can change direction. Similar results were obtained for the E. coli MutS protein (Table I), i.e. the rapid and extensive dissociation from singly-blocked duplexes was nearly identical to that observed for unblocked duplexes.

These experiments also served as controls demonstrating that MutS is not interacting specifically with four-way DNA junctions. DNase I footprinting failed to detect any significant interaction of MutS with four-way junctions in the absence or presence of ATP. Additionally, the extremely rapid dissociation of MutS from singly-blocked duplexes in the presence of ATP demonstrated that the block to dissociation seen with double-blocked substrates was not due to irreversible binding of MutS to a four-way junction.

Translocation Is Blocked by the Chromosomal Protein IHF-- To test whether a DNA-bound protein residing in the path of MutS would block the migration of MutS, we used integration host factor (IHF), an abundant protein involved in the condensation of the bacterial chromosome and in the initiation of DNA replication and integration of bacteriophage  (reviewed in Ref. 22). In addition to binding at sequence-specific, high affinity sites, much of the protein is bound to the genome nonspecifically due to the very high intracellular concentration of IHF, between 8 and 50 µM (23, 24).

The rate of dissociation of MutS from a heteroduplex DNA to which IHF was bound was assessed using three 110-bp DNA substrates containing the specific IHF-binding site H' from bacteriophage  at both ends, the left end, or neither end of the duplex (Fig. 3). The kinetics of ATP-dependent MutS dissociation from mismatch DNAs was carried out as described above except that in all cases 65 nM IHF was included in the reaction. Based on DNA binding assays carried out with IHF protein (data not shown), ~50% of the substrate with 2 IHF-binding sites was expected to be fully bound under these conditions. Higher concentrations of IHF were not used as they resulted in nonspecific binding. In the absence of ATP, dissociation from the three substrates was indistinguishable resulting in half-lives of several hundred seconds. In the presence of ATP, the bulk of MutS protein dissociated from the substrate with one or no IHF-binding sites by 15 s. In contrast, with 2 IHF sites more than 40% of the doubly-blocked substrate was bound by MutS, and the half-life of these complexes was increased at least 6-fold. Thus IHF bound to DNA posed a barrier to migration of MutS. Since a single IHF site posed no apparent block at all, E. coli MutS was not specifically interacting with IHF.

View larger version (28K): [in this window] [in a new window]   Fig. 3.   Migration of E. coli MutS is blocked by IHF. Top, the sequence of the IHF-binding site from  H'. Mismatch substrates containing an unpaired thymidine were created with an IHF-binding site at both ends, one end or neither end. Half-lives of complexes of MutS bound to each of these three DNA complexes were measured in the presence of 65 nM IHF as described under "Materials and Methods." The species denoted as "free" represents DNA substrates free of MutS but bound to zero, one, or two IHF molecules as these could not be distinguished under the conditions of electrophoresis.

MutL Blocks the Translocation of MutS-- As both MutS and MutL are required in the downstream activation of MutH, we investigated the effect of MutL on the ATP-dependent migration of MutS from a mismatch. As expected, in the presence of ATP, MutS alone dissociated very rapidly from a 110-bp duplex containing an unpaired thymidine, t_1/2 8 s (Fig. 4, A and B, Table III). Remarkably, the inclusion of MutL in the reaction increased the half-life of MutS-DNA complexes 25-fold to 170 s. MutL also induced a supershift of the MutS complex bound to DNA consistent with a physical interaction of the two proteins in a ternary MutS-MutL-DNA complex. This complex was only detected when ATP was present throughout the electrophoresis step suggesting that ATP was critical for the formation and stability of ternary complexes. When experiments were repeated with ATP present only in the reaction buffer but not during electrophoresis, again MutL greatly extended the half-life of MutS-DNA complexes from less than 8 s in its absence to 180 s (Table III); however, only the smaller MutS complex was observed on native gels (data not shown).

View larger version (50K): [in this window] [in a new window]   Fig. 4.   MutL blocks the migration of MutS. A, the dissociation of 50 nM E. coli MutS from a 32P-labeled mismatch DNA in the presence of 1 mM ATP was determined in polyacrylamide gel electrophoresis assays in the presence or absence of 100 nM MutL. ATP at a concentration of 1 mM was present during electrophoresis. B, semi-logarithmic plot of the data shown above. Half-lives of MutS complexes were calculated as described under "Materials and Methods." C, the length dependence of the block to MutS migration imposed by MutL. Dissociation of MutS from a mismatch DNA was assessed as described above using mismatch duplexes 110, 60, or 37 bp in length containing a 1T bulge. D, DNase I footprinting of mismatch duplexes in the presence of E. coli MutS, MutL, and 1 mM ATP. DNase I footprinting was performed as described under "Materials and Methods" on a 32P-labeled duplex 110 bp in length containing an unpaired thymidine. Protein concentrations were 50 nM MutS and 100 nM MutL.

To ensure that the stabilization of MutS-DNA complexes by MutL was not confined to the 1T bulge mismatch, we repeated the experiment with a G:T mismatch and observed a 13-fold increase, from less than 8 to 93 s, in the half-life of MutS-DNA complexes in the presence of ATP. Based on these results, we concluded that MutL significantly retards the ATP-dependent translocation of MutS from a mismatch. The interaction of MutL with MutS required that MutS reside in its ATP-induced translocation mode since, in the absence of ATP, MutL had no significant effect on the half-life of MutS-DNA complexes (Table III).

The modulation of MutS migration by MutL protein exhibited a dependence on the length of the mismatch DNA. Whereas MutL significantly stabilized binding of MutS to a 110-bp duplex, it failed to do so in the case of a 37-bp duplex (Fig. 4C and Table III). In the latter case, almost 90% of MutS complexes had dissociated within 15 s. In the case of a 60-bp duplex, stabilization of MutS-DNA complexes by MutL, while apparent, was not as extensive resulting in a half-life of 53 s. Similar results were obtained for the 110- and 60-bp duplex when the electrophoresis was carried out in the presence of ATP to visualize the MutL-induced supershift (data not shown). In the absence of MutL, the rate of dissociation of MutS from DNA either in the presence or absence of ATP did not exhibit a dependence on duplex length in the range of 37-110 bp (Table III). The observed dependence on duplex length might reflect a steric requirement for a minimal DNA length to accommodate the MutS-MutL complex. Alternatively, MutL interacts with MutS only after the latter has bound ATP and started to translocate. In that case, the longer the DNA, the better the odds that MutL will productively associate with MutS.

The possibility that MutL was impeding migration by MutS by binding nonspecifically to DNA was ruled out in control experiments. The relative affinity of MutL for DNA binding in the presence of ATP, ATPS, or no nucleotide cofactor was assessed in gel-mobility shift assays (data not shown) and established that MutL does not bind appreciably to DNA at the concentration of 100 nM used in translocation experiments. The absence of nonspecific DNA binding by MutL was confirmed by DNase I probing (see below, Fig. 4D). In a more direct test, the half-life of MutS-DNA complexes in the presence of MutL and ATP was measured in the presence of 10 nM plasmid DNA. This represented more than a 5000-fold nucleotide excess of plasmid over ³²P-labeled DNA. The competitor plasmid DNA was added after the MutS-³²P-labeled mismatch complexes were formed but prior to the addition of MutL. The inclusion of plasmid DNA had no effect on the half-life of the complex (data not shown) making it very unlikely that MutL was blocking MutS translocation by nonspecific DNA binding.

The interaction of MutS with mismatched DNA in the presence of MutL and ATP was probed in a DNase I footprinting experiment. ³²P-Labeled 110-bp duplexes containing an unpaired thymidine were subjected to DNase I footprinting in the presence of MutS, MutL, and ATP (Fig. 4D). E. coli MutS protected ~10 bp on either side of the mismatch. In the presence of ATP, the footprint of MutS was diminished slightly, probably because ATP binding caused MutS to migrate off the mismatch followed by rapid rebinding by the pool of free MutS, restoring the footprint. In the presence of MutL and ATP, the footprint of MutS was enlarged extending the length of the 110-bp duplex. MutL by itself failed to yield a footprint in the absence or presence of ATP indicating that it did not bind appreciably to the mismatch DNA. Such an extended footprint afforded by the presence of both MutS and MutL was previously reported for a 143-bp duplex DNA containing a G:T mismatch (7).

MutS and MutL Activate MutH without Migration Along a Duplex-- If MutS must leave the mismatch and translocate along the DNA to find MutL and activate MutH, then intramolecular reactions in which the mismatch and the hemimethylated GATC site reside on the same duplex (reactions in cis) should proceed much more efficiently than reactions in which the two target sites reside on different duplexes (reactions in trans). We compared the rate and extent of MutH cleavage reactions using cis and trans DNA substrates. The "cis" DNA substrate contained a 1T bulge mismatch and a hemimethylated GATC site on a single, linear 110-bp DNA fragment. The "trans" substrate consisted of a hemimethylated GATC site on one ³²P-labeled duplex having four-way junctions at both ends and a mismatch, unpaired T, on a second, unlabeled linear duplex (see Fig. 5B).

View larger version (16K): [in this window] [in a new window]   Fig. 5.   MutH is activated by MutS and MutL in the absence of translocation by MutS. A, MutH cleavage assays. Assays were performed as described under "Materials and Methods" in the presence of 10 nM each of MutS, MutL, and MutH, 1 mM ATP, and 1 nM 32P-labeled cis DNA substrate containing a single, hemimethylated GATC sequence and a 1T bulge in a linear duplex 110 bp long and visualized on denaturing polyacrylamide gels. B, activation of MutH by MutS and MutL in cis and in trans. MutH cleavage was measured when the mismatch and hemimethylated GATC resided on the same DNA molecule (cis, squares), or on two separate DNA molecules (trans, circles), or in the absence of a mismatch DNA (diamonds). Only DNA substrates containing the GATC site were 32P-labeled. Reactions were performed as above in the presence (filled symbols) or absence (open symbols) of MutS protein. C, MutH cleavage assays with junction substrates. MutH cleavage was monitored using two different 32P-labeled DNA junctions containing a hemimethylated GATC site and a 1T bulge mismatch on two different arms of a four-way junction. The two junction substrates differed with respect to the identity of the arms containing the mismatch (see "Results"). Reactions were performed in the presence (open symbols) or absence (closed symbols) of MutS. Jxn1, squares; Jxn 2, diamonds.

The ³²P-labeled DNA substrates were incubated at 37 °C in the presence or absence of MutS as well as in the presence of MutL and MutH proteins and ATP, and the extent of cleavage at the hemimethylated GATC site was monitored on native polyacrylamide gels (see Fig. 5A). Quantitation of MutH cleavage at the hemimethylated GATC site in the absence of MutS revealed that 20-30% of the substrates were cleaved after 1 h reflecting low level activation of MutH by MutL alone as reported previously (11, 25, 26). In the presence of MutS, cleavage at the hemimethylated site by MutH approached 65-70% after 1 h (Fig. 5B). Notably, the rates of cleavage of the cis and trans substrates were indistinguishable. Control experiments established that cleavage by MutH was mismatch-provoked. When cleavage experiments were performed with the ³²P-labeled GATC substrate but no mismatch DNA, activation of MutH in the presence of MutS and MutL was no greater than that observed with MutL alone (Fig. 5B, closed diamonds).

To test whether translocation of MutS from a mismatch site to the MutH cleavage site was required for activation of MutH by MutS, we used two four-way DNA junctions in which a 1T bulge mismatch and a hemimethylated GATC site resided on different arms of the four-way junction (see Fig. 5C). Given preferred stacking conformers of four-way junctions (reviewed in Ref. 27), the mismatch and the GATC site are oriented differently with respect to each other in these two substrates. Nonetheless, in both four-way junctions, translocation along the DNA by MutS from the mismatch to the site of MutH cleavage was prevented. In the presence of MutS, the rates of MutH cleavage for the two junction substrates were virtually identical to the rate of the linear cis substrate described above resulting in ~70% of duplexes cleaved after 1 h (Fig. 5C). Again, 20-30% of the substrates were cleaved after 1 h in the absence of MutS due to low level activation of MutH by MutL alone. The cleavage of the linear cis substrate was <3% after 1 h when MutH, MutL, or ATP was omitted (data not shown). Although these junction reactions could, in principle, occur intramolecularly, it is likely that they too occurred in trans.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The data presented here are relevant to the molecular mechanisms underlying signaling of downstream events by MutS during mismatch repair. Mismatch repair is initiated when MutS targets mismatches. Genetic and biochemical studies implicate MutS and MutL in the recruitment of proteins involved in subsequent steps such as the activation of MutH for cleavage of newly synthesized strands at hemimethylated GATC sites and excision of the daughter strand involving helicase II and exonucleases (reviewed in Ref. 1). Two models have been proposed to explain how the recognition of MutS triggers downstream repair events in an ATP-dependent fashion. Both models propose that MutS leaves the mismatch to signal subsequent repair events. The translocation model invokes ATP binding and hydrolysis in the migration of MutS along the DNA helix after binding to a mismatch (8, 14, 28). This unidirectional translocation is postulated to be involved in coupling mismatch recognition with downstream repair events. In the molecular switch model, hMSH2-MSH6 (hMutS) is likened to GTP-binding proteins (15, 16, 29). A hMutS-ADP complex is proficient for mismatch binding. Nucleotide exchange results in a hMutS-ATP complex that then diffuses from the mismatch as a hydrolysis-independent sliding clamp in search of proteins that function downstream.

Recently, we showed that it is possible for MutH to be activated by MutS and MutL in trans, i.e. when the mismatch and GATC site reside on two separate duplexes, and we suggested that MutS does not leave the mismatch to signal downstream events (11). In an extension of that study, we show here that under one set of reaction conditions, both the cis and trans reactions proceed at the same rate and to the same extent (Fig. 5B). The observations that MutS no longer moves off a mismatch in the presence of MutL and that MutH activation can occur in trans do not support models invoking migration of MutS away from the mismatch as being a trigger for repair. If MutS must move along the DNA helix from a mismatch to a hemi-methylated GATC site to stimulate MutH, then it would be expected that the intramolecular reaction should be much more efficient than any reaction in trans, a finding that is contrary to what we observe.

The data presented here also reconcile previous findings regarding the rapid ATP-induced migration of MutS away from a mismatch with the idea that MutS remains at a mismatch to signal downstream repair events. Both E. coli and Taq MutS diffuse away from the mismatch in the presence of ATP. Examination of the interaction of MutS and MutL with a mismatch DNA in the presence of ATP reveals that MutL blocks the migration of MutS and results in a supershift on polyacrylamide gels consistent with a ternary complex composed of MutS, MutL, and heteroduplex DNA. Such a ternary complex of E. coli MMR proteins is consistent with experiments employing co-immunoprecipitation (8), surface plasmon resonance (10), affinity chromatography (25), gel electrophoresis (30), and DNase I footprinting (7).

A model for the activation of MutH by MutS and MutL in a mismatch- and ATP-dependent manner that takes into account the findings presented here as well as previously published data is shown in Fig. 6. MutS recognizes and binds to a base pair or insertion/deletion mismatch as a homodimer undergoing a large conformational change in the protein that constrains otherwise mobile domains I and IV of the protein that now clamp MutS to the mismatch (12, 13). In the ATP-bound form, MutS undergoes another conformational change resulting in the loss of strong contacts in the mismatch binding site and diffusion away from the mismatch in a configuration in which the DNA duplex is now threaded through a single, large channel (11, 16). Such a configuration would explain why ATP-induced dissociation of MutS from DNA occurs at the ends of short duplexes and is blocked by branched DNA structures and by bound IHF protein.

View larger version (17K): [in this window] [in a new window]   Fig. 6.   The interaction of MutS and MutL at a mismatch. A model for how recognition of a mismatch by MutS signals downstream repair events (see "Discussion"). MutS in solution (A) binds to a mismatch in an induced fit model involving extensive conformational changes in both MutS and the mismatch DNA (B). Binding of ATP induces an additional conformation change (C) resulting in diffusion of MutS away from the mismatch (D). The signal to initiate downstream events is provided when MutL associates with MutS that has bound both a mismatch and ATP and prevents MutS from leaving the immediate vicinity of the mismatch (E). This MutS-MutL complex is postulated to be a critical intermediate in the activation of MutH for incision of the daughter strand (F).

A distinguishing feature of our model is that a ternary complex of MutS and MutL bound to a mismatch is the intermediate that signals downstream repair events in the presence of ATP. MutS that has targeted a mismatch and bound ATP interacts with MutL before MutS can wander any significant distance, forcing it to remain in the vicinity of the mismatch. The increase in the size of the DNase I footprint obtained in the presence of both MutS and MutL may reflect limited migration of MutS in the presence of ATP before MutL associates with MutS. Alternatively, it may reflect multiple MutS-MutL complexes clustered in the vicinity of the mismatch. In either case, MutS remains near the mismatch and can continue to specify the location of the mismatch to other participants. Activation of MutH for cleavage at a hemimethylated GATC site can occur via protein-protein interactions mediated by MutL (26, 31). Implicit in this model is the formation of a looped DNA intermediate during activation of MutH by MutS and MutL that may be related to the -loop structures previously observed by electron microscopy (8).

An attractive feature of the model is that MutS-MutL complexes bound to a mismatch may provide the signal for termination of the excision step involving helicase II and exonucleases. Mismatch-provoked DNA synthesis in both E. coli and human cells is localized to the region between the strand break and the mismatch with gap repair typically extending no more than 100 bp past the mismatch (32, 33). Limiting the extent of gap repair in a mismatch-dependent manner is hard to envision if MutS has left the mismatch. In this regard, it is intriguing that interactions between MutS and/or MutL proteins and their homologs and enzymes that function in the excision phase of repair, e.g. E. coli helicase II, eukaryotic EXO1, and proliferating cell nuclear antigen, have been observed (reviewed in Refs. 1, 3, and 4).

Although our studies are confined to prokaryotic mismatch repair proteins, it is quite possible that eukaryotic MutL homologs exert a similar effect on MSH2-MSH6 and MSH2-MSH3. First, there is a high degree of conservation among MutS proteins, and although conservation of MutL homologs is less extensive, the general features of mismatch repair appear to have been retained throughout evolution. Second, as discussed above, human MutS has also been observed to migrate from the mismatch in the presence of ATP (14, 16). Third, ternary complexes involving yeast and human MSH2-MSH6 and MLH1-PMS1 and a mismatch have been detected in the presence of ATP, although how these complexes direct subsequent repair events remains unclear (34-38).

A number of questions remain to be answered, among them how does MutL tether MutS to a mismatch and what triggers dissociation of MutS from the DNA? The observed dependence on heteroduplex length of MutS-MutL interactions at the mismatch and the enlarged DNase I footprint suggest that MutL may also bind to DNA and tether MutS to the heteroduplex. MutL has been shown to bind DNA by itself under some conditions (39, 40) and possesses an ATPase activity that is stimulated by DNA (25, 39). Further studies of MutS and MutL interactions are required to address these questions.

	ACKNOWLEDGEMENTS

We are grateful to Steve Goodman for providing purified IHF and for advice and to Malcolm Winkler for providing overexpressing plasmids. We thank Dan Camerini-Otero and Wei Yang for stimulating discussions and comments on the manuscript and an anonymous reviewer for thoughtful suggestions. We are grateful to George Poy for oligonucleotide syntheses and Linda Robinson for assistance.

	FOOTNOTES

^* The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

To whom correspondence should be addressed. Tel.: 301-496-0306; Fax: 301-496-9878; E-mail: hsieh@ncifcrf.gov.

Published, JBC Papers in Press, May 22, 2001, DOI 10.1074/jbc.M103148200

	ABBREVIATIONS

The abbreviations used are: IHF, integration host factor; bp, base pair; ATPS, adenosine 5'-O-(thiotriphosphate); Taq, T. aquaticus.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

This article has been cited by other articles:

				J. M. Harrington and R. D. Kolodner Saccharomyces cerevisiae Msh2-Msh3 Acts in Repair of Base-Base Mispairs Mol. Cell. Biol., September 15, 2007; 27(18): 6546 - 6554. [Abstract] [Full Text] [PDF]

				R. D. Kolodner, M. L. Mendillo, and C. D. Putnam Coupling distant sites in DNA during DNA mismatch repair PNAS, August 7, 2007; 104(32): 12953 - 12954. [Full Text] [PDF]

				A. Pluciennik and P. Modrich From the Cover: Protein roadblocks and helix discontinuities are barriers to the initiation of mismatch repair PNAS, July 31, 2007; 104(31): 12709 - 12713. [Abstract] [Full Text] [PDF]

				M. Elez, M. Radman, and I. Matic The frequency and structure of recombinant products is determined by the cellular level of MutL PNAS, May 22, 2007; 104(21): 8935 - 8940. [Abstract] [Full Text] [PDF]

				S. W. Matson and A. B. Robertson The UvrD helicase and its modulation by the mismatch repair protein MutL Nucleic Acids Res., September 10, 2006; 34(15): 4089 - 4097. [Abstract] [Full Text] [PDF]

				A. B. Robertson, S. R. Pattishall, E. A. Gibbons, and S. W. Matson MutL-catalyzed ATP Hydrolysis Is Required at a Post-UvrD Loading Step in Methyl-directed Mismatch Repair J. Biol. Chem., July 21, 2006; 281(29): 19949 - 19959. [Abstract] [Full Text] [PDF]

				M. Cho, S. Lee, S.-Y. Han, J.-Y. Park, M. A. Rahman, Y.-B. Shim, and C. Ban Electrochemical detection of mismatched DNA using a MutS probe Nucleic Acids Res., June 21, 2006; 34(10): e75 - e75. [Abstract] [Full Text] [PDF]

				R. Ahrends, J. Kosinski, D. Kirsch, L. Manelyte, L. Giron-Monzon, L. Hummerich, O. Schulz, B. Spengler, and P. Friedhoff Identifying an interaction site between MutH and the C-terminal domain of MutL by crosslinking, affinity purification, chemical coding and mass spectrometry Nucleic Acids Res., June 13, 2006; 34(10): 3169 - 3180. [Abstract] [Full Text] [PDF]

				A. Robertson, S. R. Pattishall, and S. W. Matson The DNA Binding Activity of MutL Is Required for Methyl-directed Mismatch Repair in Escherichia coli J. Biol. Chem., March 31, 2006; 281(13): 8399 - 8408. [Abstract] [Full Text] [PDF]

				Y. Yang, L. E. Sass, C. Du, P. Hsieh, and D. A. Erie Determination of protein-DNA binding constants and specificities from statistical analyses of single molecules: MutS-DNA interactions Nucleic Acids Res., August 1, 2005; 33(13): 4322 - 4334. [Abstract] [Full Text] [PDF]

				M. L. Mendillo, D. J. Mazur, and R. D. Kolodner Analysis of the Interaction between the Saccharomyces cerevisiae MSH2-MSH6 and MLH1-PMS1 Complexes with DNA Using a Reversible DNA End-blocking System J. Biol. Chem., June 10, 2005; 280(23): 22245 - 22257. [Abstract] [Full Text] [PDF]