THE JOURNAL
OF
BIOLOGICAL CHEMISTRY
Vol. 277, No. 5, Issue of February 1, pp. 3673–3679, 2002
Printed in U.S.A.
Differential ATP Binding and Intrinsic ATP Hydrolysis by
Amino-terminal Domains of the Yeast Mlh1 and Pms1 Proteins*
Received for publication, July 1, 2001, and in revised form, November 15, 2001
Published, JBC Papers in Press, November 20, 2001, DOI 10.1074/jbc.M106120200
Mark C. Hall‡, Polina V. Shcherbakova‡, and Thomas A. Kunkel‡§
From the ‡Laboratory of Molecular Genetics, NIEHS, National Institutes of Health, Research Triangle Park,
North Carolina 27709
MutL homologs belong to a family of proteins that share
a conserved ATP binding site. We demonstrate that amino-terminal domains of the yeast MutL homologs Mlh1
and Pms1 required for DNA mismatch repair both possess
independent, intrinsic ATPase activities. Amino acid substitutions in the conserved ATP binding sites concomitantly reduce ATP binding, ATP hydrolysis, and DNA mismatch repair in vivo. The ATPase activities are weak,
consistent with the hypothesis that ATP binding is primarily responsible for modulating interactions with
other MMR components. Three approaches, ATP hydrolysis assays, limited proteolysis protection, and equilibrium dialysis, provide evidence that the amino-terminal
domain of Mlh1 binds ATP with >10-fold higher affinity
than does the amino-terminal domain of Pms1. This is
consistent with a model wherein ATP may first bind to
Mlh1, resulting in events that permit ATP binding to
Pms1 and later steps in DNA mismatch repair.
Correction of DNA replication errors by the general DNA
mismatch repair (MMR)1 system is one of several cellular processes contributing to mutation avoidance and genomic stability. MMR defects result in increased spontaneous mutation
rates and susceptibility to cancer in humans and mice (reviewed recently in Refs. 1 and 2). Several proteins are essential
for DNA mismatch repair in prokaryotes and eukaryotes, including bacterial MutL and its eukaryotic homologs. In Escherichia coli, homodimeric MutL protein coordinates the MMR
pathway by interacting with and modulating the activity of
other proteins, including MutS, MutH, and UvrD (1, 2). In
eukaryotes, matchmaking functions are thought to be performed by heterodimeric complexes of MutL homologs.
MutL and its homologs belong to the GHKL family of proteins, which includes type II topoisomerases, the hsp90 chaperone proteins, and histidine kinases (3–5). GHKL family members all share conserved amino acid motifs implicated in ATP
binding and hydrolysis. Consistent with this fact, E. coli MutL
binds ATP, resulting in a large conformational change in the
protein (6). MutL also catalyzes a weak ATPase reaction (7)
that is thought to be required for DNA mismatch repair (8).
* 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.
§ To whom correspondence should be addressed: Laboratory of Structural Biology, NIEHS, NIH, 111 T. W. Alexander Dr., Research Triangle Park, NC 27709. Tel.: 919-541-2644; Fax: 919-541-7613; E-mail:
[email protected].
1
The abbreviations used are: MMR, mismatch repair; NTD, aminoterminal domain; ADPNP, adenosine 5⬘-(␤,␥-imino)triphosphate; BisTris, 2-[bis(2-hydroxyethyl)amino]-2-(hydroxymethyl)propane-1,3-diol;
EK, enterokinase.
This paper is available on line at http://www.jbc.org
Based on their homology to MutL and the other GHKL family
members, eukaryotic MutL homologs are also expected to bind
and hydrolyze ATP. In yeast, mutation rate measurements
indicate that amino acids in the conserved ATP binding motifs
of Mlh1 and Pms1 are essential for mismatch repair, and
partial proteolysis experiments indicate that Mlh1 undergoes
an ATP-induced conformational change (9). Here, we provide
direct evidence that yeast Mlh1 and Pms1 both bind and hydrolyze ATP independently of each other and that they both
change conformation upon ATP binding. Moreover, we provide
three lines of evidence that Mlh1 binds ATP with substantially
higher affinity than does Pms1. The results suggest that proper
coordination of the MMR pathway may involve sequential
binding of ATP, first to one then the other ATP binding site in
the Mlh1-Pms1 heterodimer.
EXPERIMENTAL PROCEDURES
Plasmids—Regions of yeast MLH1 encoding amino acids 1–343 and
PMS1 encoding amino acids 32–396 were amplified by PCR using
pMH1 (10) and pMH8 (11) as templates, respectively. NcoI and NotI
restriction enzyme sites introduced by the primers were used to clone
the PCR products into pET30a (Novagen). The constructs allowed isopropyl-␤-D-thiogalactopyranoside-induced production of the two proteins, designated Mlh1(343) and Pms1(396), as fusions with a 6-His tag
for nickel affinity purification. Missense mutations in Mlh1(343) and
Pms1(396) were made directly in these constructs using the
QuikChange site-directed mutagenesis kit (Stratagene). MLH1 and
PMS1 alleles were sequenced using an ABI PRISM 377 DNA sequencer
(PerkinElmer Life Sciences) to confirm the correct sequence. The
URA3-based yeast integrative plasmid YIpMLH1 has been described
previously (12). The 2 ␮-based plasmid pMMR82 containing the yeast
PMS1 gene with 5⬘ upstream and 3⬘ downstream regions cloned in
YEplac195 (13) was kindly provided by L. Prakash (University of Texas). To create plasmid YIpPMS1, a 4.4-kb SalI fragment with the PMS1
gene from pMMR82 was cloned into the SalI site of yeast integrative
URA3 plasmid pRS306 (14). Missense mutations in the MLH1 and
PMS1 genes were created in YIpMLH1 and YIpPMS1 by site-directed
PCR mutagenesis and confirmed by partial sequencing of MLH1 and
PMS1 as described previously (12). For construction of mutant yeast
strains, at least three independently obtained mutant derivatives of
YIpMLH1 or YIpPMS1 were used.
Yeast Strains—Saccharomyces cerevisiae haploid strains E134
(MAT␣ ade5 lys2::InsEA14 trp1-289 his7-2 leu2-3,112 ura3-52) and E68
(MAT␣ ade2-1 arg4-8 leu2-3,112 thr1-4 trp1-1 ura3-52 lys2⌬ cup1-1)
and mlh1⌬ and pms1⌬ derivatives of E134 have been described previously (11, 12). To create haploid strains with missense mutations in the
MLH1 gene, the chromosomal MLH1 gene in strain E134 was replaced
with the mutant alleles using mutant derivatives of plasmid YIpMLH1
as described previously (12). Haploid strains with mutations in the
PMS1 gene were constructed in a similar way using mutant derivatives
of plasmid YIpPMS1 cut with HpaI. Diploid strains heterozygous for
mlh1 or pms1 mutations and a control wild-type diploid were constructed by crossing E68 to E134 or its mlh1 or pms1 mutants.
Mutation Rate Measurements—Rates of lys2::InsEA14 and his7-2 reversion and forward mutation to canavanine resistance were measured
by fluctuation analysis as described previously (12).
Protein Purification—Proteins were overproduced from pET30a in E.
coli BL21(DE3)pLysS. A single transformant was grown up to 5 liters in
3673
3674
ATP Binding and Hydrolysis by Mlh1 and Pms1
2⫻YT medium under antibiotic selection at 30 °C. At an optical density
(600 nm) of 0.75, 0.5 mM isopropyl-␤-D-thiogalactopyranoside (Invitrogen) was added to the 5-liter cultures, and growth was continued for 4 h.
Cells were harvested by centrifugation and frozen at ⫺20 °C. All subsequent steps were performed at 4 °C. The protocols for purification of
Mlh1(343), Pms1(396), and all mutants were identical. Thawed cells
were resuspended in 100 ml of buffer A (25 mM NaPO4, pH 7.6 at 23 °C,
10% glycerol, 0.1 mM EDTA, 500 mM NaCl, and 5 mM 2-mercaptoethanol) containing 10 mM imidazole, Complete EDTA-free protease inhibitors (Roche Molecular Biochemicals), and 0.5 mM phenylmethylsulfonyl fluoride. The cell suspension was sonicated using a model W140 Cell
Disrupter (Branson Sonic Power Co.) to achieve a water-like viscosity
and cleared by centrifugation at 30,000 ⫻ g for 30 min.
The cleared lysate was loaded onto a 5-ml nickel-nitrilotriacetic acid
superflow (Qiagen) column (1.5 cm inner diameter) equilibrated with
buffer A ⫹ 10 mM imidazole at 1.0 ml/min. The column was washed with
50 ml of buffer A ⫹ 20 mM imidazole and developed with a linear
gradient of imidazole (20 –250 mM) in buffer A while collecting 1.0-ml
fractions. Fractions containing the desired protein were pooled and
dialyzed overnight against 1 liter of buffer B (25 mM Tris-HCl, pH 8.0 at
23 °C, 10% glycerol, 10 mM dithiothreitol, and 0.1 mM EDTA) containing
100 mM NaCl. The dialysate was loaded onto a 5-ml HiTrap heparinSepharose column (Amersham Biosciences, Inc.) equilibrated with
buffer B ⫹ 100 mM NaCl. This column was washed with 25 ml of the
same buffer and developed with a linear gradient of NaCl (100 –750 mM)
in buffer B while collecting 1.0-ml fractions. Fractions containing the
desired protein (elution peak at ⬃500 mM NaCl) were pooled and
dialyzed overnight against 1 liter of buffer C (25 mM Tris-HCl, pH 8.0
at 23 °C, 10% glycerol, 50 mM NaCl, 4 mM MgCl2, and 10 mM
dithiothreitol).
To remove the 5-kDa affinity tag, the purified fusion proteins were
treated with recombinant enterokinase (Novagen) as directed by the
supplier, leaving Ala-Met-Ala attached to the amino terminus. The
cleaved proteins were loaded by gravity flow onto a 1-ml singlestranded DNA cellulose (Amersham Biosciences, Inc.) column equilibrated with buffer C, washed with 3 ml of buffer C, and eluted with
buffer C containing 500 mM NaCl. Proteins were dialyzed against either
buffer C or buffer D (25 mM NaPO4, pH 7.6 at 23 °C, 50 mM NaOAc, 10
mM dithiothreitol, and 4 mM MgCl2), flash-frozen in liquid nitrogen, and
stored in small aliquots at ⫺80 °C. Protein concentrations were determined by the method of Bradford (15) using bovine serum albumin as a
standard.
ATPase Assays—All ATPase assays were performed in buffer D at
30 °C. For rate measurements, reactions (30 ␮l) contained 1 mM
[␣-32P]ATP (Amersham Biosciences, Inc.) for Mlh1(343) and its mutant
derivatives or 10 mM [␣-32P]ATP for Pms1(396) and its mutant derivatives. For Km measurements, reactions (15 ␮l) contained varying concentrations of [␣-32P]ATP. Reaction products were separated on TLC
plates (Mallinckrodt Baker Inc.) developed with 0.8 M LiCl and 1 M
formic acid. Results were extracted using a Storm 860 PhosphorImager
and ImageQuant software (Molecular Dynamics).
Partial Proteolysis—Reactions (20 ␮l) were performed in buffer C
and contained ⬃2.5 ␮g of protein and the indicated concentration of a
1:1 ATP/MgCl2 mixture. After 5 min to allow ATP binding, reactions
were initiated with 125 ng of trypsin and incubated at room temperature for 45 min (Mlh1(343) and mutants) or 15 min (Pms1(396) and
mutants). Reactions were stopped by addition of SDS loading dye,
subjected to SDS-PAGE, and products visualized by Coomassie Blue
staining. Digital image analysis was used to calculate the percentage of
protein protected at each ATP concentration using the formula (Pint ⫺
P0)/(Ptot ⫺ P0), where Pint is the band intensity of the intact protein at
a given ATP concentration, Ptot is the band intensity of the intact
protein in a mock reaction lacking trypsin, and P0 is the band intensity
of intact protein in the absence of ATP.
Equilibrium Dialysis—Equilibrium microdialyzers (Harvard Apparatus Inc.) containing two 25-␮l chambers separated by a 10,000-dalton
molecular mass cutoff dialysis membrane were used. One chamber
contained protein in buffer D, and the other chamber contained
[␥-32P]ATP (Amersham Biosciences, Inc.) in buffer D. A 48 h incubation
at 4 °C was required to achieve equilibrium. Three 5-␮l samples were
drawn from each side of microdialyzers and counted with a Beckman LS
7800 liquid scintillation counter. Control experiments lacking protein
were included to ensure that equilibrium was achieved. In all experiments, more than 90% of the protein remained soluble after the 48-h
incubation.
RESULTS
A 43-kilodalton NH2-terminal domain (NTD) of E. coli DNA
gyrase B, a GHKL family member, was previously used to
study ATP binding and hydrolysis (16, 17), and a similar NTD
of E. coli MutL also binds and hydrolyzes ATP (6). Based on
these precedents, we began this study using NTDs of yeast
Mlh1 and Pms1 proteins that were identified by proteolysis and
mass spectrometry.2 These proteins (Fig. 1A) contain the first
343 amino acids of Mlh1 and amino acids 32–396 of Pms1. The
latter choice was prompted by structure-based sequence alignments (7) and by in vivo complementation studies with plasmid-borne yeast PMS1 genes containing a mutation in either
codon 1 or 32 and studies mapping the 5⬘ end of PMS1 mRNA
(10).3 These approaches indicate that functional Pms1 can be
expressed using methionine codon 32 of the published PMS1
sequence as a start codon for translation. To avoid confusion
relative to the published amino acid numbering for yeast Pms1,
the NTD of Pms1 is designated here as Pms1(396), while the
corresponding Mlh1 NTD is designated Mlh1(343). Both proteins were overproduced in E. coli and purified to apparent
homogeneity (Fig. 2).
ATP Hydrolysis—When the purified Mlh1(343) and
Pms1(396) preparations were examined for the ability to hydrolyze ATP, ATPase activities were readily detected (compare
first and last three lanes in Fig. 3, A and B). To determine
whether these ATPase activities were intrinsic to Mlh1(343)
and Pms1(396), we purified two different mutant derivatives
predicted to lack ATPase activity. In the crystal structure of the
MutL NTD, a conserved Asn (residue 35 in Mlh1 and 65 in
Pms1, Fig. 1, B and C) participates in coordinating a Mg2⫹ ion
thought to be required for ATP binding (6). Based on crystal
structures of NTDs of gyrase B and MutL (6, 18) and biochemical studies of gyrase B (19), Glu-31 in Mlh1 and Glu-61 in
Pms1 (Fig. 1, B and C) may serve as general bases for catalyzing ATP hydrolysis. Mlh1(343) and Pms1(396) proteins with
homologous N35A and N65A substitutions have levels of
ATPase activity that are reduced by 84 and 95%, respectively
(Fig. 3, A and B, and Table I). The E31A and E61A mutant
preparations also had decreased ATPase activities (Table I).
These data indicate that the majority of the ATPase activities
in the wild-type protein preparations are intrinsic to Mlh1(343)
and Pms1(396). The weak ATPase activities detected in the
mutant protein preparations (Table I) may reflect contaminating ATPase that is too faint to be seen by SDS-PAGE. For
example, the N35A mutant of Mlh1(343), which does not interact significantly with ATP (see below), still exhibited an
ATPase level 18% that of wild-type Mlh1(343). This residual
activity is more apparent in Mlh1(343) mutant preparations,
because the kcat for ATPase activity of Mlh1(343) is weaker
than that of Pms1(396) (Table I).
Functional Defects in MMR Due to Mutations in ATPase
Active Site Residues—To determine whether loss of ATPase
activity in the mutant proteins correlates with loss of MMR in
vivo, we examined mutation rates at three different loci in
haploid yeast strains harboring these same mutations. The
mlh1-N35A and pms1-N65A mutations both resulted in a
strong mutator phenotype indistinguishable from that of the
mlh1 deletion or pms1 deletion strain. The mlh1-E31A and
pms1-E61A mutations were also mutators (Table II), consistent
with a previous report (9). Immunoblot analysis indicated that
the level of expression of wild-type, E31A, and N35A Mlh1p in
the three haploid yeast strains was similar (data not shown).
2
M. C. Hall, P. V. Shcherbakova, C. Borchers, J. M. Dial, K. B.
Tomer, and T. A. Kunkel, manuscript in preparation.
3
W. Kramer, unpublished observations.
ATP Binding and Hydrolysis by Mlh1 and Pms1
3675
FIG. 1. E. coli MutL and S. cerevisiae
Mlh1 and Pms1 proteins. A, schematic
representation showing the location of
MutL ATP binding domain (pink) and the
location of the putative Mlh1 and Pms1
ATP binding domains by homology to
MutL. Purple bars indicate the conserved
ATP binding motifs I, II, III, and IV (from
left to right, see Refs. 4 and 7). The Pms1
interaction region in Mlh1 and the Mlh1
interaction region in Pms1 (blue) were described previously (27). B, alignment of
amino acids in ATP binding motif I. Numbers above the sequences correspond to
amino acids in E. coli MutL. Amino acid
substitutions made in this study are
shown below the alignment. C, ATP binding site of E. coli MutL. The diagram is
drawn using coordinates of the NTD of
MutL complexed with ADPNP (6). ATP
(orange) and side chains of Glu-29 and
Asn-33 (yellow) are shown as sticks, other
amino acid residues contributing to ATP
binding are drawn in the space-filling
mode. The phosphate binding loop (Ploop) is shown in dark blue. Oxygen atoms
of the three phosphates of the ATP, two
water molecules (light blue balls), and the
oxygen atom of Asn-33 coordinate the
Mg2⫹ ion (green ball). The white dotted
line represents Mg2⫹ coordination by
Asn33.
Parallel attempts with the pms1 strains were unsuccessful,
because there was insufficient Pms1 expressed from the natural promoter on the chromosome to detect using anti-Pms1.
Note that the pms1-E61A mutation only conferred a weak
mutator phenotype (60-fold, as compared with 11,000-fold for
the mlh1 deletion mutant), despite a 50-fold reduction in
ATPase activity by the mutant protein (Table I). In contrast,
the decrease in ATPase activity of Mlh1-E31A resulted in a
severe reduction of mismatch repair in vivo. The reduced
ATPase activity of Pms1-E61A may be sufficient for completion
of the majority of mismatch repair events, whereas completion
of repair is apparently more sensitive to perturbation of the
Mlh1 ATPase activity.
Semidominant Mutator Effect of N35A Mutation—To determine whether the mlh1 or pms1 mutations confer a mutator
phenotype in the presence of the wild-type MLH1 or PMS1
alleles, we constructed diploid yeast strains heterozygous for
each of the four missense mutations. The MLH1/mlh1⌬ strain
heterozygous for mlh1 deletion has a Lys⫹ reversion rate ele-
vated 3–5-fold (Table III). We showed previously that this
mutator effect results from loss of the wild-type MLH1 allele in
a small fraction of cells rather than reduced MMR capacity in
the heterozygous cells (12). Heterozygosity for the mlh1-E31A
or pms1-E61A mutations did not increase the mutation rate
above the increase conferred by heterozygosity for mlh1 deletion (Table III). However, heterozygosity for the mlh1-N35A
mutation led to a 480-fold elevation of mutation rate. This
suggests that the mutant Mlh1p inhibits MMR by wild-type
Mlh1p, albeit not completely, since deletion of mlh1 results in
an 11,000-fold mutator effect. In contrast to the results with
the mlh1-N35A mutation, the analogous pms1-N65A mutation
did not confer a dominant mutator effect. The semidominant
effect of mlh1-N35A, but not pms1-N65A, and the differential
mutator effects of mlh1-E31A and pms1-E61A mutations in
haploid strains, are both consistent with the idea of functional
asymmetry within the Mlh1-Pms1 heterodimer. This interpretation is qualified by the fact that the N65A Pms1 expression
level from the natural promoter is unknown, although we do
3676
ATP Binding and Hydrolysis by Mlh1 and Pms1
note the N65A substitution did not adversely affect solubility or
the stability of Pms1(396) overproduced in E. coli (Fig. 2).
Kinetic Analysis of ATP Binding and Hydrolysis—Next, the
ATPase activities of the wild-type Mlh1(343) and Pms1(396)
proteins were measured as a function of increasing ATP concentration (Fig. 3, C and D) to obtain steady state kinetic
values (Table I). The Km for Mlh1(343) (69 ␮M) is 22-fold lower
than the Km for Pms1(396) (1.5 mM), suggesting that Mlh1(343)
has a greater ATP binding affinity. The turnover number (kcat)
FIG. 2. Purified proteins used in this study. Approximately 2.5
␮g of purified proteins were run on 4 –12% NuPAGE BisTris SDSpolyacrylamide gels and stained with Coomassie Brilliant Blue R-250.
A, lanes 1, 3, and 5 are purified Pms1(396), E61A, and N65A prior to
enterokinase (EK) cleavage, respectively; lanes 2, 4, and 6 are purified
Pms1(396), E61A, and N65A following EK cleavage, respectively. Faint
bands near the 31-kDa marker in lanes 2, 4, and 6 result from nonspecific enterokinase cleavage. B, lane 1 is purified Mlh1(343) prior to EK
cleavage; lanes 2– 4 are purified Mlh1(343), E31A, and N35A following
EK cleavage, respectively.
FIG. 3. ATP hydrolysis catalyzed by
Mlh1(343) and Pms1(396). A, a typical
TLC plate showing ADP generated at various time points during reactions containing 1 mM [␣-32P]ATP and 20 ␮M
Mlh1(343) or 25 ␮M Mlh1(343) N35A. B, a
typical TLC plate showing ADP generated at various time points during reactions containing 10 mM [␣-32P]ATP and 12
␮M Pms1(396) or 12 ␮M Pms1(396) N65A.
C, the rate of ATP hydrolysis to ADP catalyzed by 20 ␮M Mlh1(343) was plotted as
a function of ATP concentration. D, the
rate of ATP hydrolysis to ADP catalyzed
by 22 ␮M Pms1(396) was plotted as a function of ATP concentration. In both C and
D, data represent the average of three
independent experiments. Nonlinear
least squares curve fits using the Michaelis-Menten equation were generated by
Kaleidagraph (Synergy Software, Reading, PA) to obtain the solid lines in C and
D and the Km values reported in Table I.
of Mlh1(343) is also lower (by 6.5-fold) than that of Pms1(396)
(Table I), such that the catalytic efficiencies of the two proteins
differ by only 3-fold (kcat/Km of 3 ⫻ 10⫺4 for Mlh1(343) and 1 ⫻
10⫺4 for Pms1(396)). The large difference in Km between
Mlh1(343) and Pms1(396) illustrates a biochemical asymmetry
in the Mlh1-Pms1 heterodimer that may be related to the
functional asymmetry observed in vivo.
ATPase Activity of Intact Mlh1-Pms1 Heterodimer—Because
the NTD of E. coli MutL hydrolyzes ATP at a 10-fold lower rate
than does full-length MutL (6), we tested whether intact yeast
Mlh1-Pms1 heterodimer possesses higher ATPase activity than
Mlh1(343) and Pms1(396). When purified as described previously (10) and analyzed under conditions similar to those used
in Fig. 3A, the Mlh1-Pms1 preparation hydrolyzes ATP at a
rate of 0.3– 0.4 min⫺1. This rate is only slightly higher than the
combined rates for the individual NTDs. Difficulties in purifying the N35A/N65A double mutant heterodimer preclude the
interpretation that all this ATPase activity is intrinsic to Mlh1Pms1. Nonetheless, we conclude that the ATPase activity of the
heterodimer is weak and not substantially greater than that of
the individual NTDs.
Limited Proteolysis Asssays—Given the large difference in
affinity for ATP suggested by ATPase assays, we examined
ATP binding by Mlh1(343) and Pms1(396) in more detail. Ligand binding often induces conformational changes in proteins
that can be monitored by limited proteolysis. For example, the
addition of ATP to the Mlh1-Pms1 heterodimer was shown by
immunoblotting to protect the amino terminus of Mlh1 from
trypsin proteolysis (9). We used the individual NTDs of Mlh1
and Pms1 here, rather than the Mlh1-Pms1 heterodimer, to
examine the effect of ATP on proteolysis, because the resulting
fragment patterns are much simpler to analyze. Use of the
ATP Binding and Hydrolysis by Mlh1 and Pms1
full-length Mlh1-Pms1 heterodimer yields essentially the same
results due to rapid cleavage of a very sensitive proteolytic
cleavage site, presumably in an exposed interdomain region
separating the NTDs from COOH-terminal heterodimerization
regions.2
Both Mlh1(343) and Pms1(396) were protected from limited
trypsin proteolysis when ATP was present (Fig. 4). However,
there was a large difference in the ATP concentration required
to protect Mlh1(343) and Pms1(396) (Fig. 4, A and C versus B
and D). The concentration of ATP required to achieve halfmaximal protection (KATP in Table I) was 14-fold lower for
Mlh1(343) (130 ␮M) than for Pms1(396) (1.9 mM). This suggests
that Mlh1(343) binds ATP with higher affinity than Pms1(396).
The KATP values for Mlh1(343) and Pms1(396) are remarkably
similar to the Km values obtained in ATPase assays, providing
further evidence that the observed ATPase activities are intrinsic to Mlh1(343) and Pms1(396).
We then examined the mutant derivatives of Mlh1(343) and
Pms1(396) in proteolysis protection assays. Consistent with the
inferred role of the conserved asparagine residue in ATP binding, the N35A mutant of Mlh1(343) and the N65A mutant of
Pms1(396) were protected from trypsin proteolysis by ATP to a
lesser extent than the wild-type proteins (Fig. 4). We also
observed lower ATP-dependent protection of the E61A mutant
TABLE I
Parameters for ATPase and proteolysis protection of
Mlh1(343) and Pms1(396)
ATPase activity
Protein
Relative ratec
KATPd
␮M
pmol/min
␮M
0.023 ⫾ 0.007
69 ⫾ 5
1
0.38
0.16
130 ⫾ 19
660 ⫾ 75
N
0.15 ⫾ 0.006
1500 ⫾ 320
1
0.02
0.05
1900 ⫾ 360
N
N
min
Mlh1(343)
Wild-type
E31A
N35A
Proteolysis
Km b
kcata
Pms1(396)
Wild-type
E61A
N65A
⫺1
3677
(Fig. 4, B and D), suggesting that its ATP binding capacity may
also be somewhat lower. However, the degree of the ATP binding defects of E61A and N65A is difficult to quantitate, because
the affinity of Pms1(396) for ATP is already very low. Thus, it
is possible that these two mutants possess ATP binding defects
of different severity that could not be distinguished in this
assay. Our in vivo results on mutation rates (Table II) support
this possibility. The E31A mutant was protected by ATP (Fig.
4, A and C) but yielded a KATP value (660 ␮M) that was 5-fold
higher than for wild-type Mlh1(343). This suggests that the
E31A mutant can still bind ATP but has a lower affinity, which
is consistent with the lower ATP binding affinity reported for
the homologous E29A mutant of E. coli MutL (6). The proteolysis patterns of all mutants (not shown) were not significantly
different from their respective wild-type proteins. In multiple
experiments, the extent of proteolysis in the absence of ATP
was not significantly different among the mutant and wild-type
proteins. However, the sensitivity of this approach does not
rule out the possibility of slight conformational differences
caused by the amino acid changes.
Equilibrium Dialysis—To obtain direct evidence for differential ATP binding by the two MutL homologs, we measured
binding of ATP to Mlh1(343), Pms1(396), and the N35A mutant
of Mlh1(343) by equilibrium dialysis (Fig. 5). Consistent with
the proteolysis results, Mlh1(343) clearly bound ATP at concentrations between 6 and 100 ␮M (closed circles). In contrast,
Pms1(396) did not bind ATP in this same concentration range.
These data provide direct confirmation that the large differences between Mlh1(343) and Pms1(396) in Km and KATP (Table I) are due to differences in their ATP binding affinities. As
predicted for the role of Asn-35 in coordinating Mg2⫹, no ATP
binding was detected with the Mlh1(343) N35A mutant
protein.
TABLE III
Mutator phenotypes of diploids heterozygous for
mlh1 and pms1 mutations
a
Results represent the average of three independent experiments
with S.D. Protein concentrations were 22 ␮M for Mlh1(343) and 28 ␮M
for Pms1(396).
b
Three separate trials were averaged together and fit with the
Michaelis-Menten equation using Kaleidagraph to give the Km and
standard error.
c
Relative ATPase rates per pmol of protein were measured at 1 mM
ATP for Mlh1(343) mutants and at 10 mM ATP for Pms1(396) mutants.
The kcat values for wild-type proteins were set to 1 for the comparisons.
The values for mutants do not represent true kcat values, because the
ATP concentrations are not saturating. Protein concentrations were 9
␮M for E31A, 28 ␮M for N35A, 21 ␮M for E61A, and 12 ␮M for N65A.
Results represent the average of three independent experiments.
d
Three separate experiments were combined to obtain an average
data set that was fit with a rectangular hyperbola using Kaleidagraph
to yield KATP ⫾ S.E. values. N ⫽ not measurable.
Lys⫹ reversiona
Genotype
Reversion rate
(⫻10⫺8)b
Fold
increasec
Wild-type
MLH1/mlh1⌬
MLH1/mlh1-E31A
PMS1/pms1-E61A
MLH1/mlh1-N35A
PMS1/pms1-N65A
17 (12–32)
56 (42–80)
33 (29–45)
21 (16–28)
8300 (7200–9500)
78 (72–130)
1
3.2
1.9
1.2
480
4.6
a
Reversion rates were measured for diploid strains heterozygous for
mlh1 or pms1 mutations and an isogenic wild-type diploid strain.
b
Rates are given as medians for nine independent cultures, with 95%
confidence limits in parentheses.
c
Relative to the wild-type mutation rate.
TABLE II
Effect of amino changes in ATP binding domains of Mlh1p and Pms1p on spontaneous mutagenesis
Lys⫹ reversiona
Genotype
Wild-type
mlh1⌬
pms1⌬
mlh1-N35A
pms1-N65A
mlh1-E31A
pms1-E61A
a
Reversion rate (⫻10⫺6)b
0.15 (0.12–0.18)
1600 (1200–2600)
1400 (1200–1700)
1400 (1100–2300)
1700 (1300–2700)
490 (400–580)
8.7 (6.3–12)
His⫹ reversiona
Canr mutationa
Fold
increasec
Reversion rate (⫻10⫺8)b
Fold
increasec
Mutation rate
(⫻10⫺8)b
Fold
increasec
1
11,000
9400
9800
11,000
3300
60
0.63 (0.41–1.1)
79 (67–96)
89 (75–97)
110 (74–140)
78 (72–96)
23 (18–27)
3.5 (2.6–5.2)
1
130
140
170
120
36
6
19 (16–29)
370 (340–440)
540 (350–1300)
540 (440–780)
400 (330–620)
160 (120–180)
27 (17–41)
1
19
28
28
21
8
1
Mutation rates for haploid strains with mutations in MLH1 or PMS1 and isogenic wild-type strain E134.
Rates are given as medians for at least nine independent cultures, with 95% confidence limits in parentheses.
c
Relative to the wild-type mutation rate.
b
3678
ATP Binding and Hydrolysis by Mlh1 and Pms1
FIG. 4. Effect of ATP on limited
trypsin proteolysis. A and B, products
of limited proteolysis reactions with 2.5
␮g of protein were separated on 4 –12%
NuPAGE SDS-polyacrylamide gels and
stained with Coomassie Brilliant Blue
R-250. For mutants, only the full-length
protein band is shown to conserve space.
Neither of the mutations in Mlh1(343) or
Pms1(396) drastically altered the pattern
of proteolytic fragments (not shown). NP,
control reactions without protease. C, the
percent protection at each ATP concentration for wild-type Mlh1(343) (●) and the
E31A (Œ) and N35A (⽧) mutants was
quantitated by digital image analysis of
proteolysis gels (such as those in A) as
described under “Experimental Procedures.” The average results from at least
three independent experiments were plotted as a function of ATP concentration.
Solid lines are nonlinear least squares
curve fits using the equation for a rectangular hyperbola. KATP values obtained
from these curve fits are reported in Table
I. D, the same analysis was performed for
proteolysis experiments such as those
represented in B with Pms1(396) (●) and
the E61A (Œ) and N65A (⽧) mutants.
FIG. 5. ATP binding measured by equilibrium dialysis. Equilibrium dialysis experiments were performed as described under “Experimental Procedures” with 26 ␮M Mlh1(343) (●), 28 ␮M Pms1(396) (⫻),
and 32 ␮M Mlh1(343) N35A (䡺). Data represent the average of three
independent experiments. Error bars for Mlh1(343) are S.D. values
about the mean. We were unable to evaluate binding at higher ATP
concentrations due to the solubility limit (⬃30 ␮M) of the protein.
DISCUSSION
We have demonstrated that NTDs of yeast Mlh1 and Pms1
proteins bind and hydrolyze ATP, consistent with the presence
of conserved GHKL family ATP binding motifs. The rates of
ATP hydrolysis by both proteins and the intact Mlh1-Pms1
heterodimer are similar to those of other family members,
including yeast Hsp90 and MutL, whose kcat values are ⬃0.4
min⫺1 (7, 20, 21). Such weak ATPase activity is probably not
used for a motor function but rather to drive conformational
changes within the protein itself that affect its association with
other DNA repair components (22). This idea has been previously discussed by Tran and Liskay (9) and is supported by the
present study. MutL homologs may use the energy of ATP
hydrolysis only to return the enzyme to its initial conformation
to participate in another round of repair. Such a mechanism
was previously proposed for gyrase B (23). In this scenario, it
may be ATP binding that is key to promoting protein conformational changes that facilitate downstream events in MMR.
For example, ATP binding is sufficient for E. coli MutL to
activate the MutH endonuclease (7, 24). The similar strong
mutator effects of mlh1-N35A and pms1-N65A mutations suggest that ATP binding by both Mlh1 and Pms1 is essential for
MMR, while weaker mutator effects of mlh1-E31A and pms1E61A suggest that a significant fraction of repair events can be
completed in the absence or reduction of Mlh1 or Pms1 ATPase
activity.
Amino acids that are important for ATPase activity in the
NTD of E. coli MutL are disordered in the crystal structure of
the monomer (7) but become properly ordered upon homodimerization (6). Current evidence suggests that dimerization of the NTDs of other GHKL proteins such as MutL and
gyrase B is essential for ATPase activity (6, 16). It is worth
noting that all previously studied GHKL family ATPases normally function as homodimers. However, Mlh1 and Pms1 are
thought to normally function as a heterodimer during mis-
ATP Binding and Hydrolysis by Mlh1 and Pms1
match repair. In this study, the Mlh1(343) and Pms1(396)
proteins catalyze ATP hydrolysis in the absence of their partner in the normal heterodimer. Thus, it is formally possible
that the NTDs of Mlh1(343) and Pms1(396) can homodimerize,
as has been detected with intact Mlh1 in the absence of intact
Pms1 (11). However, we have been unable to detect homodimerization by glutaraldehyde crosslinking experiments2
in the presence and absence of ATP or the nonhydrolyzable
analog ADPNP, suggesting that these NTDs are primarily monomeric in solution. This suggests that, unlike the NTD of E.
coli MutL and other homodimeric GHKL family members, the
weak ATPase active sites in the NTDs of Mlh1 and Pms1 may
function as monomers. In support of this, recent studies of an
NTD of human PMS2 indicate that it too hydrolyzes ATP, yet
dimerization is not detected by gel filtration, protein crosslinking, or equilibrium ultracentrifugation in the presence of
nonhydrolyzable ATP analogues (25).
Kinetic analysis of ATP hydrolysis (Table I), ATP-dependent
protection against proteolysis (Fig. 4), and direct equilibrium
binding data (Fig. 5) all suggest that Mlh1(343) has a higher
ATP binding affinity than does Pms1(396). This is consistent
with a model wherein ATP binds to Mlh1 first, then to Pms1. If
so, then ATP-dependent conformational changes in Mlh1 and
Pms1 may occur sequentially and modulate different events
required to search for the strand discrimination signal, excise
the mismatch, and resynthesize the nascent DNA strand. This
is further suggested by the fact that mutations in conserved
residues in the putative ATP binding pockets of Mlh1 and
Pms1 have different effects on mutation rates in vivo. Thus,
Tran and Liskay first demonstrated (9) and we now confirm
(Table II) that the E31A mutation in yeast Mlh1 confers a
much stronger mutator phenotype than does the homologous
E61A mutation in Pms1. We further show here that the N35A
mutation in Mlh1 produces a semidominant effect while the
N65A mutation in Pms1 does not.
Although the exact nature and ordering of the many interactions that occur during mismatch repair remain to be determined, the semidominant effect of mlh1-N35A mutation but
not the pms1-N65A mutation may be instructive. Apparently,
the inability of Mlh1-N35A to bind ATP and change conformation not only disrupts MMR, but also partially inhibits MMR by
wild-type proteins present in heterozygous diploid cells (Table
III). Thus, it is possible that Mlh1-N35A might still interact
with MutS homologs or other proteins (e.g. PCNA) but may not
be able to recruit the downstream factors necessary for mismatch processing. This inactive complex may block access to
the mismatch by wild-type proteins to prevent repair of a
fraction of mismatches. However, the corresponding N65A mutation in Pms1, which also results in a complete MMR defect in
a haploid strain, does not produce a mutator phenotype in a
heterozygous diploid strain. It is possible that an ATP-induced
3679
conformational change in Pms1 may occur at a later stage
when the mismatch is no longer occupied by a protein complex.
In this case, abortive repair attempts by mutant proteins would
not prevent repair by wild-type proteins. In this model, the
semidominant effect of the mlh1-N35A mutation, but not the
pms1-N65A mutation, is consistent with sequential binding of
ATP, first to Mlh1, then to Pms1.
Other models are of course possible. The enormous complexity of the DNA mismatch repair pathway is illustrated by the
fact that, like MutL proteins, MutS proteins also form asymmetric dimers that contain two ATPase active sites thought to
sequentially bind and hydrolyze ATP (reviewed in Ref. 26).
MutS and MutL dimers interact with a variety of different
DNA substrates and with a number of other proteins to fulfill
their roles in MMR, transcription-coupled repair, checkpoint
control, apoptosis, and recombination. For some of these roles,
it will also be interesting to determine whether other MutL
homologs that partner with Mlh1, such as Mlh2 and Mlh3,
have low ATP binding affinities like that of Pms1.
Acknowledgments—We are grateful to Leroy Worth and Karin
Drotschmann for helpful comments on the manuscript.
REFERENCES
1. Buermeyer, A. B., Deschenes, S. M., Baker, S. M., and Liskay, R. M. (1999)
Annu. Rev. Genet. 33, 533–564
2. Harfe, B. D., and Jinks-Robertson, S. (2000) Annu. Rev. Genet. 34, 359 –399
3. Bergerat, A., de Massy, B., Gadelle, D., Varoutas, P.-C., Nicolas, A., and
Forterre, P. (1997) Nature 386, 414 – 417
4. Mushegian, A. R., Bassett, D. E. J., Boguski, M. S., Bork, P., and Koonin, E. V.
(1997) Proc. Natl. Acad. Sci. U. S. A. 94, 5831–5836
5. Dutta, R., and Inouye, M. (2000) Trends Biochem. Sci. 25, 24 –28
6. Ban, C., Junop, M., and Yang, W. (1999) Cell 97, 85–97
7. Ban, C., and Yang, W. (1998) Cell 95, 541–552
8. Spampinato, C., and Modrich, P. (2000) J. Biol. Chem. 275, 9863–9869
9. Tran, P., and Liskay, R. M. (2000) Mol. Cell. Biol. 20, 6390 – 6398
10. Hall, M. C., and Kunkel, T. A. (2001) Protein Expression Purif. 21, 333–342
11. Shcherbakova, P. V., Hall, M. C., Lewis, M. S., Bennett, S. E., Martin, K. J.,
Bushel, P. R., Afshari, C. A., and Kunkel, T. A. (2001) Mol. Cell. Biol. 21,
940 –951
12. Shcherbakova, P. V., and Kunkel, T. A. (1999) Mol. Cell. Biol. 19, 3177–3183
13. Gietz, R. D., and Sugino, A. (1988) Gene (Amst.) 74, 527–534
14. Sikorski, R. S., and Heiter, P. (1989) Genetics 122, 19 –27
15. Bradford, M. M. (1976) Anal. Biochem. 72, 248 –254
16. Ali, J. A., Jackson, A. P., Howells, A. J., and Maxwell, A. (1993) Biochemistry
32, 2717–2724
17. Ali, J. A., Orphanides, G., and Maxwell, A. (1995) Biochemistry 34, 9801–9808
18. Wigley, D. B., Davies, G. J., Dodson, E. J., Maxwell, A., and Dodson, G. (1991)
Nature 351, 624 – 629
19. Jackson, A. P., and Maxwell, A. (1993) Proc. Natl. Acad. Sci. U. S. A. 90,
11232–11236
20. Obermann, W. M. J., Sondermann, H., Russo, A. A., Pavletich, N. P., and
Hartl, F. U. (1998) J. Cell Biol. 143, 901–910
21. Panaretou, B., Prodromou, C., Roe, S. M., O’Brien, R., Ladbury, J. E., Piper,
P. W., and Pearl, L. H. (1998) EMBO J. 17, 4829 – 4836
22. Yang, W. (2000) Mutat. Res. 460, 245–256
23. Sugino, A., Higgins, N. P., Brown, P. O., Peebles, C. L., and Cozzarelli, N. R.
(1978) Proc. Natl. Acad. Sci. U. S. A. 75, 4838 – 4842
24. Hall, M. C., and Matson, S. W. (1999) J. Biol. Chem. 274, 1306 –1312
25. Guarné, A., Junop, M. S., and Yang, W. (2001) EMBO J. 20, 5521–5531
26. Sixma, T. K. (2001) Curr. Opin. Struct. Biol. 11, 47–52
27. Pang, Q., Prolla, T. A., and Liskay, R. M. (1997) Mol. Cell. Biol. 17, 4465– 4473