MutS and MutL Activate DNA Helicase II in a Mismatch-dependent Manner*

MutS, MutL, and DNA helicase II are required for the mismatch-provoked excision step that occurs duringEscherichia coli methyl-directed mismatch repair. In this study MutL is shown to enhance the unwinding activity of DNA helicase II more than 10-fold on a conventional helicase substrate in which a 35-residue oligonucleotide is annealed to a M13 circular single-stranded phage DNA under conditions where the two proteins are present at approximately molar stoichiometry with respect to the substrate. MutS- and MutL-dependent activation of DNA helicase II has also been demonstrated with a model substrate in which a 138-residue oligonucleotide was hybridized to a 138-nucleotide gap in an otherwise duplex 7,100-base pair circular DNA. Displacement of the oligonucleotide requires MutS, MutL, DNA helicase II, and ATP and is dependent on the presence of a mismatch within the hybrid region. Although DNA helicase II and Rep helicase share substantial sequence homology and features of mechanism, Rep helicase is inactive in this reaction.

Escherichia coli methyl-directed mismatch repair initiates via the mismatch-provoked incision of the unmethylated strand at a hemimethylated d(GATC) sequence in a reaction that involves the MutS-and MutL-dependent activation of the MutH d(GATC) endonuclease activity (1). The single-strand break thus produced may occur either 3Ј or 5Ј to the mismatch on the unmethylated strand and directs the exonucleolytic excision of that portion of the unmodified strand spanning the incised d(GATC) sequence and the mispair (2,3). Excision requires MutS, MutL, DNA helicase II, and an appropriate exonuclease. When the strand break that directs repair occurs 5Ј to the mismatch, excision requires RecJ exonuclease or exonuclease VII (3,4), both of which support 5Ј 3 3Ј hydrolysis (5,6). For repair directed by a strand break 3Ј to the mismatch, the 3Ј 3 5Ј hydrolytic activity of exonuclease I (7) is sufficient to meet the exonuclease requirement (2). 1 Since helicase II is required for excision from either side of the mismatch and because each of these exonucleases is specific for single-stranded DNA (5)(6)(7), the action of helicase II presumably serves to unwind the incised strand so as to render it exonuclease sensitive. According to this interpretation, the exonuclease functions in excision are secondary to those of DNA helicase II. We have therefore sought partial reactions in which MutS, MutL, and a mismatch might enhance the activity of helicase II. We show here that MutL stimulates helicase II activity on a conventional substrate and that helicase activity on incised duplex DNA is enhanced by MutS and MutL in a mismatch-dependent manner. The accompanying paper (8) demonstrates that MutS, MutL, and mismatch-dependent entry of helicase II into an incised heteroduplex occurs at the strand break with helicase entry biased so that translocation occurs toward the mispair.

EXPERIMENTAL PROCEDURES
Proteins, DNA, and Nucleotides-MutS (9) and DNA helicase II (10) were purified as described. Rep helicase (11) was kindly provided by Timothy Lohman (Washington University, St. Louis, MO). Exonuclease was from Life Technologies, Inc., restriction enzymes and VENT polymerase from New England BioLabs, and T4 polynucleotide kinase from Amersham Pharmacia Biotech. E. coli MutL was purified by a modification of the method described previously (12) after subcloning the mutL gene from pAL51 (13) into a pET3a expression vector that had been cleaved with NdeI and BamHI to yield L1-pET3a. Briefly, a 180-liter culture of E. coli BL21(DE3)/pLysS/L1-pET3a was grown at 37°C in L broth supplemented with thymine (4 g/ml), thiamine (10 g/ml), glucose (10 mg/ml), and 10 mM KPO 4 , pH 7.4. The culture was induced at A 590 ϭ 1.0 by addition of isopropyl-1-thio-␤-galactopyranoside (Amersham) to 0.4 mM. The culture was chilled to 10°C 2.75 h after induction and harvested by centrifugation. Cell paste (780 g) was stored at Ϫ70°C. MutL was purified from 175 g of cell paste as described (12), except that the Bio-Rex 70 column was increased in size to 20-cm ϫ 30-cm 2 , and the concentration step and the Sephadex G-150 column were eliminated. Fractions from the Bio-Rex 70 column were pooled and dialyzed against 0.05 M KPO 4 (pH 7.4), 0.05 M KCl, 0.1 mM EDTA, 0.5 mM dithiothreitol (DTT), 2 1 mM phenylmethylsulfonyl fluoride, frozen in small aliquots in liquid N 2 , and stored at Ϫ70°C. The concentration of the MutL preparation used in this work, which was about 98% pure as judged by Coomassie-stained sodium dodecyl sulfate gels, was determined by Bradford assay (14). MutL preparations isolated by the previous procedure contain low levels of a DNA-dependent ATPase (12), which immunological analysis has shown to be DNA helicase II (not shown). MutL isolated as described above also contains helicase II, but the trace levels of the activity present had no effect on the experiments described here. When cited in molar terms, protein concentrations are expressed as monomer equivalents assuming 100% activity.
Wild type M13mp2 and C103 M13mp2 phages, which differ at nucleotide 103 (T 3 C) in the lacZ␣ gene (15), were gifts from Thomas Kunkel (National Institute of Environmental Health Sciences, Research Triangle Park, NC). [␥-32 P]ATP was purchased from New England Nuclear. Oligonucleotides were synthesized by Oligos Etc.
Conventional Helicase Assays-The 35-residue oligonucleotide d(ATCGTCGCTATTAATTAATTTTCCCTTAGAATCCT) was 5Ј-end-labeled using T4 polynucleotide kinase and [␥-32 P]ATP (3000 Ci/mmol). After phenol extraction, phenol-chloroform-isoamyl alcohol extraction, and chloroform-isoamyl alcohol extraction, the labeled oligonucleotide * Supported in part by Grant GM23719 from the National Institute of General Medical Sciences. 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.
A 138-residue single-strand oligomer was prepared for hybridization to the gapped circular duplex by PCR amplification of a segment of M13mp2 wild type RF DNA or C103 M13mp2 RF DNA using 5Ј-Pd(CACTGGCCGTCGTTTTACAACGTC) and 5Ј-OH-d(GCAACTGTT-GGGAAGGGCGATCGG) as primers. Reactions (0.1 ml) contained 375-ng template DNA (wild-type or C103 M13mp2 RF DNA depending on whether no mismatch or a G-T mismatch was to be generated in the final substrate), 100 pmol of each primer, and 4 units of Vent DNA polymerase, which was used according to the recommendations of the manufacturer. Amplification used a Perkin Elmer Gene Amp 9600 with an initial incubation at 94°C for 2 min, followed by 12 cycles of 30 s at 94°C, 30 s at 70°C, and 30 s at 72°C. Protein was extracted as described above and excess primers, dNTPs, and templates were removed by filtration through Sephacryl-400 (Amersham Pharmacia Biotech, 25-cm ϫ 0.785-cm 2 ) equilibrated with TNE. Pooled fractions containing the 138-bp PCR product were concentrated by ethanol precipitation, and the washed pellet was dissolved in 200 l of 67 mM glycine-NaOH (pH 9.4), 2.5 mM MgCl 2 , 50 g/ml bovine serum albumin. The 5Ј-phosphorylated viral strand component of the PCR product was hydrolyzed by digestion with 5 units of exonuclease at 37°C for 30 min (18), and protein was extracted as described above. Undigested DNA was collected by ethanol precipitation, and 5Ј-end-labeled using T4 polynucleotide kinase. After deproteinization and ethanol precipitation, unincorporated label was removed by Sephadex G-50 filtration.
The end-labeled 138-nucleotide complementary strand PCR product (10 pmol, corresponding to M13mp2 nucleotides 6236 -6373) was hybridized to the gapped circular DNA (1 pmol) in 250 l of 20 mM Tris-HCl, pH 7.6, 10 mM MgCl 2 , 100 mM NaCl, 1 mM DTT. Incubation was at 95°C for 5 min, 65°C for 15 min, and the solution was then cooled to room temperature. In an alternate procedure, a 3-ml reaction containing 10 mM KPO 4 , pH 7.4, 10 mM NaCl, 1 mM EDTA, 2.1 nmol of the 138-nucleotide strand, and 210 pmol of gapped circular duplex was supplemented with 90 l of 10 N NaOH and incubated at room temperature for 5 min. After addition of 300 l of 2.9 N acetic acid, 135 l of 3 M KCl, and 372 l of 1 M KPO 4 , pH 7.4 (final pH 6.9), DNA was annealed at 65°C for 30 min, slow cooled to 37°C, and chilled on ice. Hybridized products were purified by heating to 70°C for 5 min and filtration through Sephacryl-400 columns equilibrated with TNE at 70°C. Quantitation was by ethidium fluorescence using M13mp2 RF DNA as a standard.
Helicase activity on the duplex substrate was initiated by addition of repair proteins in a total volume of 3 l of the diluent described above to 7 l of a solution containing 71 mM HEPES-KOH, pH 8.0, 29 mM KCl, 8.6 mM MgCl 2 , 1.4 mM DTT, 71 g/ml bovine serum albumin, 2.9 mM ATP, and 23 fmol of duplex substrate. Unless specified otherwise, proteins were added in the following order and amounts: 370 fmol of MutS, 250 fmol of MutL, 120 fmol of DNA helicase II. After incubation at 30°C as indicated, reactions were terminated as described above and supplemented with 2 g/ml proteinase K. After a further 10 min of incubation at 30°C, samples were loaded on a 3% Meta-Phor-agarose (FMC Corp.) in 44.5 mM Tris, 44.5 mM boric acid, pH 7.6, 1 mM EDTA and run at 4.5 V/cm for 3.5 h. The agarose gel was dried, and radioactivity was quantitated using a PhosphorImager.

RESULTS
MutL Stimulates DNA Helicase II-During initial attempts to identify a partial reaction dependent on MutS, MutL, and DNA helicase II, we found that near homogeneous preparations of MutL markedly stimulated helicase activity on a conventional substrate constructed by hybridization of a 35-residue oligonucleotide to M13mp2 viral strand DNA (see "Experimental Procedures"). Under conditions where helicase II was present at one monomer equivalent per mol of substrate (as molecules), the addition of two monomer equivalents of MutL enhanced the initial rate of oligonucleotide displacement 11-fold (Fig. 1, upper panel). MutS and/or the presence of a G-T mismatch within the hybrid region had no effect on the rate of oligonucleotide displacement in the presence or absence of MutL (not shown). E. coli Rep helicase and DNA helicase II are about 40% homologous at the sequence level (19), the two proteins can form heterodimers (20), and the duplex unwinding activity of both is stimulated by a 3Ј-single-stranded DNA tail. Low levels of MutL also stimulated Rep unwinding about 5-fold, although higher concentrations of the Rep activity were required to support significant unwinding with the substrate used here (Fig. 1, lower panel).
MutS and MutL Enhance Helicase Unwinding at a Strand Break in a Mismatch-dependent Manner-The failure to observe MutS or mismatch effects with the conventional helicase substrate might be due to the presence of mismatches within the secondary structure assumed by the 7 kilobases of singlestranded DNA in the molecule. To circumvent this potential problem, we constructed a duplex helicase substrate similar to that described by Washburn and Kushner (21) by hybridizing a 138-residue oligonucleotide to M13mp2 circular duplex DNA that contained a 138-nucleotide gap (Fig. 2). The 138-residue oligonucleotide was produced by PCR amplification using two different templates so that duplex helicase substrates could be constructed with or without a G-T mismatch between the two strand breaks (see "Experimental Procedures").
As shown in Fig. 3, neither MutS nor MutL detectably altered the integrity of either homoduplex or G-T heteroduplex DNAs prepared in this manner. Furthermore, both DNAs supported only trace levels of unwinding by helicase II in the presence or absence of MutL or MutS. However, extensive displacement of the 138-residue oligonucleotide of the G-T heteroduplex occurred when all three proteins were present. The MutS, MutL, and helicase II-dependent unwinding reaction required ATP and Mg 2ϩ and was not further enhanced by the addition of single-stranded DNA binding protein (not shown). It is noteworthy that the DNA and protein concentrations used in this experiment are those that have been determined to be optimal for the reconstituted mismatch repair reaction (2). Significant unwinding of the homoduplex control also occurred in the presence of MutS, MutL, and helicase II, but the degree of oligonucleotide displacement was considerably less than that observed with the G-T heteroduplex (Fig. 3). The kinetics of oligonucleotide displacement with these two substrates is illustrated in Fig. 4.
In view of the number of DNA manipulations involved in construction of the duplex helicase substrates, the background activity observed with the homoduplex control molecules might reflect polymerase errors occurring during the PCR amplification step used to prepare the 138-residue oligonucleotide (22)(23)(24)(25) or damage incurred during denaturation and annealing steps. Given the random nature of production of such lesions, this hypothesis predicts that significant differences in mismatch dependence would be observed between independent sets of heteroduplex/homoduplex constructs. As summarized in Table I, this was in fact the case. With seven independent sets of heteroduplex/homoduplex constructs, the rate of unwinding of the G-T substrate ranged from 2.4 to 8.3 times that observed with the A⅐T control with a mean value of 4.4. Nevertheless, other explanations for the homoduplex background cannot be excluded. For example, it is possible that the MutS and MutL pair might simply activate helicase II with duplex substrates. However, as shown in the accompanying paper, we have failed to observe detectable MutS, MutL, and helicase II-dependent unwinding of a similar circular homoduplex that contained a single nick and was prepared without using PCR methodology (8).
Activation of Unwinding by MutS and MutL Is Specific for DNA Helicase II-In contrast to mutU (uvrD) mutations that inactivate helicase II function in mismatch repair (17,26), rep mutants are not mutators (27), and despite the significant MutL stimulation of Rep in conventional helicase assays (Fig. 1), the protein has no known function in mismatch repair. As shown in Fig. 5, activation of unwinding of the nicked duplex G-T substrate by MutS and MutL is specific for helicase II. Rep activity with this DNA was limited, and the presence of MutS and MutL had no significant effect on unwinding even at Rep concentrations 50 times that of the highest helicase II level tested. DISCUSSION Previous work has demonstrated that in addition to their roles in the initiation stage of methyl-directed repair, MutS and MutL are required for one or more subsequent steps (2), including the excision stage of the reaction (3). The experiments described here provide further support for involvement of these two proteins in the excision stage of the reaction. Results obtained with duplex substrates of the type shown in Fig. 2 indicate that the two proteins activate unwinding by helicase II in a mismatch-dependent manner. We have not directly addressed the mechanism of this effect, but it is possible that activation may involve several steps in the unwinding reaction.
The accompanying paper (8) demonstrates that mismatchdependent unwinding by MutS-and MutL-activated helicase II initiates at a strand break. Whereas helicase II is capable of unwinding from a nick (28), this reaction requires very high helicase II concentrations with initiation being rate-limiting (29). Nevertheless, the finding that a genetically altered form of helicase II is selectively defective in unwinding from a nick suggests that this reaction may be biologically significant (30). In view of these findings, our observations with the duplex helicase substrate indicate that MutS and MutL act in mismatch-dependent manner to enhance the rate of helicase II initiation at a strand break. As noted above, this effect is specific for helicase II and does not occur with 40% homologous Rep helicase, consistent with known requirements for helicase II in mismatch repair (17,26).
In contrast to our failure to observe significant unwinding of a 138-residue oligonucleotide in the duplex substrate at low helicase II concentrations (Figs. 2-4), a 12-residue oligonucleotide in this type of molecule was previously shown to be efficiently displaced at low concentrations of the protein (21). We think it likely that these disparate results are due to the smaller size of the oligonucleotide in the latter substrate. Since helicase II binds at nicks (29), it is possible that proximity of the two strand breaks in the latter DNA results in a concerted effect on initiation of unwinding by the enzyme. Differences due to chain length effects may also reflect the relatively modest processive behavior of helicase II (31), and it has been shown that the number of base pairs unwound by the activity depends on the amount of the protein (32).
In addition to the MutS-and MutL-dependent activation of helicase II at a strand break, we have also found that MutL markedly enhances the activity of helicase II, and to a lesser extent the Rep enzyme, on conventional substrates. This effect is evident at very low protein concentrations (in the case of helicase II, about one monomer equivalent and 2 MutL monomer equivalents per 7-kb DNA molecule), indicating high specificity. In fact, preliminary experiments indicate that MutL and helicase II interact in the absence of DNA as judged by native electrophoretic analysis (33). Thus, in addition to its role with  MutS in activating helicase II initiation at a nick, it is likely that MutL also functions as a helicase activator subsequent to initiation of unwinding. This activation could occur by increasing the processive behavior of the helicase by a direct interaction as is the case for X174 gene-A protein and Rep helicase (34). The differential effects observed with the Rep helicase in the two assays are also consistent with the idea that MutS and/or MutL modulate helicase action at several levels of the unwinding reaction. MutL stimulation of both Rep and helicase II in conventional assays can be understood in terms of the 40% sequence homology shared by the two helicases (19). However, the failure of MutS and MutL to activate Rep unwinding at a single-strand break must be due to amino acid sequence elements that distinguish Rep from helicase II. Although MutL is required for several steps in methyl-directed mismatch repair, precise molecular roles for the protein in the reaction have not been assigned. MutL binds to MutSheteroduplex complexes (12), is required along with MutS and a mismatch for activation of the latent d(GATC) endonuclease of MutH (1), and stimulates the ATP-dependent MutS-catalyzed formation of heteroduplex DNA loops (35). Such observations have led to the suggestion that MutL serves to interface mismatch recognition by MutS to other components of the repair system (36,37). The observations that MutL physically interacts with and activates helicase II are consistent with this view (33). Coupled with the finding that MutL is also required for the mismatch, MutS, and helicase II-dependent unwinding observed with nicked circular heteroduplex substrate, we think it likely that the protein has an important role in the mismatchdependent loading of helicase II into the nicked heteroduplex, an event that is shown in the accompanying paper (8) to occur at the strand break.
A single hemimodified d(GATC) sequence is sufficient to provide strand specificity to heteroduplex repair by the E. coli methyl-directed pathway (4,38), and mismatch-provoked incision of the unmethylated strand of this d(GATC) site by activated MutH (1) provides a strand break that is thought to serve as the initiation site for exonucleolytic removal of that portion of the new strand spanning the nick and the mispair (3). The latter conclusion is based on the finding that excision on a single d(GATC) site heteroduplex depends not only on MutS, MutL, and DNA helicase II (2, 3), but also requires an appropriate 3Ј 3 5Ј or 5Ј 3 3Ј single-strand exonuclease, depending on whether the MutH strand break is produced 3Ј or 5Ј to the mispair (4). By contrast, we have shown here that MutS, MutL, and helicase II are sufficient to displace a mismatched oligonucleotide from a heteroduplex in which a strand break is present on either side of the mismatch, raising questions concerning the biological significance of this exonuclease-independent reaction. Unfortunately, it is difficult to compare results obtained with the two types of heteroduplex. In the single d(GATC) site substrates examined previously, the distance between the MutH-produced strand break and the mispair is 1,000 bp (39), whereas nicks in the heteroduplex described here were located 45-and 92-bp to either side of the mismatch. Since activated helicase II can clearly unwind the latter substrate, concurrent exonucleolytic hydrolysis of the displaced strand product may be necessary for continued unwinding of the 1000 base pairs that separate the mismatch and the strand break in the former heteroduplexes. The distinct properties of these two types of substrates may be clarified by study of two d(GATC) site molecules and by examination of doubly incised heteroduplexes as a function of strand break-mismatch separation distance.
Note Added in Proof-Using two-hybrid analysis and protein affinity chromatography, Steve Matson and colleagues (Hall, M. C., Jordon, J. R., and Matson, S. W. (1998) EMBO J. 17, 1535-1541) have also demonstrated a physical interaction between MutL and DNA helicase II.