 |
INTRODUCTION |
DNA mismatch repair
(MMR)1 is a highly conserved
process that removes nucleotides misincorporated into the newly
synthesized DNA strand during replication or recombination (1, 2).
Unlike other DNA repair pathways, correction of biosynthetic errors
requires not only that noncanonically paired bases be recognized but
also that the repair system be able to differentiate between the
template and the newly synthesized DNA strand, such that it can direct the correction process to the latter. In Escherichia coli,
mismatched base pairs and short insertion/deletion loops are recognized
by the MutS homodimer, which initiates the repair process by recruiting the remaining members of the repairosome: the MutL homodimer, the
strand discrimination factor MutH, and DNA helicase II (UvrD) as well
as one of several exonucleases (both 5' 
3' and 3' 5'), DNA
polymerase III holoenzyme, and DNA ligase. The MutL protein plays a key
role in coordinating this process. It activates the cryptic
endonuclease activity of MutH (3-5), which then nicks the newly
synthesized DNA strand. This reaction is enabled by the transient
undermethylation of GATC sequences in newly synthesized DNA and by the
ability of MutH to incise only the unmethylated strand. MutL also
appears to be responsible for directional loading of DNA helicase II at
the site of the nick (6, 7). The displaced DNA is exonucleolytically
degraded, and the removed stretch is then resynthesized by the
replicative DNA polymerase (for reviews, see Refs. 8 and 9). Despite
the fact that the MMR pathway of E. coli could be
reconstituted from the purified individual components (10), our
knowledge of the molecular mechanism of the process is limited. This is
especially true for the MutL protein, the role of which is particularly
enigmatic. What is clear is that its transactions, like those of MutS,
are governed by ATP binding and/or hydrolysis. This was first implied
by experiments of Grilley et al. (11), who showed that MutL
and MutS interact on a heteroduplex substrate in an
ATP-dependent manner, and from in vivo studies
of Aronshtam and Marinus (12), who showed that MutL alleles carrying
single amino acid substitutions in the vicinity of the ATP binding
domain of MutL were unable to complement the MMR defect of a
mutL
strain. More recently, the N-terminal
domain of MutL could be shown to contain four short sequence motifs
that are found also in the type II topoisomerases, the HSP90 heat shock
proteins and histidine kinases (Fig. 1a) (13, 14).
Structural analyses of members of all four families revealed that these
proteins share a novel nucleotide-binding fold and suggested an
important role for the invariant residues in binding and/or hydrolysis
of ATP (for a review, see Ref. 15). Thus, this superfamily of proteins, known as GHKL ATPases, is likely to have evolved from a common ancestor. However, as illustrated by the divergent roles of these proteins, their common fold is a poor predictor of biological function.
The recently described crystal structures of LN40, the N-terminal 349 amino acid residues of MutL (16, 17), revealed that the polypeptide
folds into two 
/
domains (Fig. 1a). The first (amino
acids 20-200) harbors all four conserved motifs that form the
nucleotide binding site. In the second domain (amino acids 224-331), a
loop containing an invariant lysine residue (Lys-307), is likely
to act as a sensor that triggers a conformational change in the protein
in response to nucleotide binding within the first domain. Such changes
have been noted with both LN40 and the entire MutL protein (16).
Analysis of gel filtration and sedimentation properties of the
full-length MutL suggested that its elongated structure became globular
upon binding of ADPPNP. More dramatically, binding of
ADPPNP triggered dimerization of LN40 in solution as well as in
crystallization experiments. The structure of LN40·ADPPNP revealed that more than 60 amino acid residues within this domain became ordered and contributed to nucleotide binding or to the dimerization of the NBDs. Assuming that the carboxyl-terminal ends of
MutL remain stably associated during this time, the amino-terminal part
of MutL could function as an ATP-driven molecular gate. Type II
topoisomerases make use of such a gate to capture a segment of duplex
DNA in order to pass it through a transient double strand break (18),
and ATP-induced dimerization of HSP90 (19) has been suggested to play a
role in clamping the substrate polypeptides during the folding reaction
(20). In addition, ATP binding/hydrolysis is important in the
coordination of HSP90 interactions with other proteins. For example,
HSP90 interacts with the co-chaperone p23 only in its ATP-bound form
(21, 22). Given that MutL needs to interact with different partners, a
similar role of ATP binding might be anticipated. Based on the
observation that nonhydrolyzable ATP analogues promoted the binding of
single-stranded DNA by MutL (17, 23), it was hypothesized that the
molecular gate might be used to clamp MutL onto DNA. In agreement with
this, single-stranded DNA was found to stimulate the very weak ATPase
activity of MutL (17, 24). Importantly, substitution of an arginine
residue located within the gate (R266E) abolished both DNA binding and the stimulation of the ATPase activity (17).
The molecular mechanism of MMR appears to be highly conserved, as
witnessed by the identification of MutS and MutL (but not MutH)
homologues in most organisms studied to date. In human cells, the MutS
homologues hMSH2, hMSH3, and hMSH6 form two heterodimers, hMutS
(hMSH2/hMSH6) and hMutS (hMSH2/hMSH3), which are involved in
mismatch recognition (25-28). Of the four mutL homologues
(hMLH1, hPMS1, hPMS2, and
hMLH3) identified to date in the human genome (29),
hMLH1 and hPMS2 play a predominant role in MMR
(30-35). The polypeptides encoded by these genes interact via their
carboxyl-terminal halves to form a stable heterodimer, hMutL, which
is indispensable for MMR (30, 36-38).
Germ line mutations in MMR genes, such as those found in the hereditary
nonpolyposis colon cancer (HNPCC) kindreds, predispose to cancers of
the colon, endometrium, and ovary (39). Of the nearly 300 families
entered in the HNPCC data base (available on the World Wide Web at
www.nfdht.nl), more than 160 carry mutations in the hMLH1
locus, of which about 30% cause single amino acid changes. Many of
these missense mutations are found in the highly conserved
amino-terminal region, suggesting that they might affect the ATPase
function of the hMLH1 protein (Fig. 1b). Interestingly, no
corresponding mutations have been identified in the hPMS2
gene, with only four families with germ line mutations in this locus having been described to date (33, 40, 41). This implies either that
the hPMS2 gene is less prone to mutagenesis than the other
HNPCC loci, that the function of hPMS2 is not essential for MMR due to
a possible functional redundancy with another member of the MutL
homologue family, such as for example hMLH3 (42), or that missense
mutations in hPMS2 escape detection, because they are phenotypically
silent. We set out to address the latter hypothesis by studying the
relative importance of the ATP binding domains of the two subunits of
hMutL in the MMR process. We could show previously that recombinant
hMutL, purified from baculovirus-infected insect cells, was able to
complement in vitro the MMR defect of human cell lines
carrying mutations in either the hMLH1 or hPMS2 genes (37). In the present study, we made use of this system by
introducing defined amino acid substitutions into the ATPase active
sites of hMLH1 and hPMS2 and by investigating the properties of the
recombinant variants in several biochemical assays.
| |
EXPERIMENTAL PROCEDURES |
Site-directed Mutagenesis and Production of Recombinant
Baculoviruses
The Bac-To-Bac baculovirus expression system (Invitrogen)
was used according to the instructions of the manufacturer. Vectors for
the expression of the wild type proteins have been described in Ref.
37. Site-directed mutagenesis was used to generate vectors encoding
MutL homologues that carry single amino substitutions in the ATP
binding domain.
pFastBacI-hMLH1[E34A]--
Two PCR fragments amplified from
pFastBacI-hMLH1 using the primer pairs mr_34/mr_19 and
mr_18/mr_2 were used as templates in a second PCR step, using the
mr_34/mr_2 primers. The final PCR product was cloned between the
BamHI/PvuII sites of
pFastBacI-hMLH1.
pFastBacI-His6-hMLH1[E34A]--
Two PCR fragments
amplified from pFastBacI-His6-hMLH1 using
the primer pairs mr_4/mr_19 and mr_18/mr_2 were used as templates in a
second PCR step, using the mr_4/mr_2 primers. The final PCR product was
cloned between the BamHI/PvuII sites of
pFastBacI-His6-hMLH1.
pFastBacI-hMLH1[N38A]--
Two PCR fragments amplified from
pFastBacI-hMLH1 using the primer pairs mr_34/mr_21 and
mr_20/mr_2 were used as templates in a second PCR step, using the
mr_34/mr_2 primers. The final PCR product was cloned between the
BamHI/PvuII sites of
pFastBacI-hMLH1.
pFastBacI-His6-hMLH1[N38A]--
Two PCR fragments
amplified from pFastBacI-His6-hMLH1 using
the primer pairs mr_4/mr_21 and mr_20/mr_2 were used as templates in a
second PCR step, using the mr_4/mr_2 primers. The final PCR product was
cloned between the BamHI/PvuII sites of
pFastBacI-His6-hMLH1.
pFastBacI-hMLH1[D63N]--
Two PCR fragments amplified from
pFastBacI-hMLH1 using the primer pairs mr_34/mr_39 and
mr_38/mr_2 were used as templates in a second PCR step, using the
mr_34/mr_2 primers. The final PCR product was cloned between the
BamHI/PvuII sites of
pFastBacI-His6-hMLH1.
pFastBacI-His6-hMLH1[D63N]--
Two PCR fragments
amplified from pFastBacI-His6-hMLH1 using
the primer pairs mr_4/mr_39 and mr_38/mr_2 were used as templates in a
second PCR step, using the mr_4/mr_2 primers. The final PCR product was
cloned between the BamHI/PvuII sites of
pFastBacI-His6-hMLH1.
pFastBacI-His6-hPMS2--
A fragment containing the
5'-end of the hPMS2 cDNA was amplified by PCR from
pFastBacI-hPMS2 using primer mr55 and primer mr41 and cleaved with BamHI/AflII. The
resulting fragment was cloned between the BamHI and
AflII sites of pFastBacI-hPMS2.
pFastBacI-hPMS2[E41A]--
Two PCR fragments amplified from
pFastBacI-hPMS2 using the primer pairs mr_34/mr_23 and
mr_22/mr_26 were used as templates in a second PCR step, using the
mr_34/mr_26 primers. The final PCR product was cloned between the
BamHI/AflII sites of
pFastBacI-hPMS2.
pFastBacI-His6-hPMS2[E41A]--
Two PCR fragments
amplified from pFastBacI-His6-hPMS2 using
the primer pairs mr_34/mr_23 and mr_22/mr_26 were used as templates in
a second PCR step, using the mr_34/mr_26 primers. The final PCR product
was cloned between the BamHI/AflII sites of
pFastBacI-hPMS2.
pFastBacI-hPMS2[N45A](#139)--
Two PCR fragments amplified
from pFastBacI-hPMS2 using the primer pairs mr_34/mr_25 and
mr_24/mr_26 were used as templates in a second PCR step, using the
mr_34/mr_26 primers. The final PCR product was cloned between the
BamHI/AflII sites of
pFastBacI-hPMS2.
pFastBacI-His6-hPMS2[N45A]--
Two PCR fragments
amplified from pFastBacI-His6-hPMS2 using
the primer pairs mr_34/mr_25 and mr_24/mr_26 were used as templates in
a second PCR step, using the mr_34/mr_26 primers. The final PCR product
was cloned between the BamHI/AflII sites of
pFastBacI-hPMS2.
pFastBacI-hPMS2[D70N]--
Two PCR fragments amplified from
pFastBacI-hPMS2 using the primer pairs mr_34/mr_41 and
mr_40/mr_42 were used as templates in a second PCR step, using the
mr_34/mr_42 primers. The final PCR product was cloned between the
BamHI/PvuII sites of
pFastBacI-PMS2.
pFastBacI-His6-hPMS2 [D70N]--
An
AflII/XbaI fragment of vector
pFastBacI-hPMS2[D70N] was cloned between the
AflII/XbaI sites of
pFastBacI-His6-hPMS2.
All constructs were sequenced over the entire length of the sub-cloned
PCR fragments and checked for the presence of the desired base substitutions.
Oligonucleotides
Oligonucleotides used were as follows: mr_2
(5'-CGGAATTCTATCTGTATGCACACTTTCCAT-3'), mr_4
(5'-CGGGATCCAAGCTCGTTCGTGGCAGGGGTT-3'), mr_18
(5'-ATGCTATCAAAGCGATGATTGAGAAC-3'), mr_19
(5'-CAGTTCTCAATCATCGCTTTGATAGC-3'), mr_20
(5'-GAGATGATTGAGGCCTGTTTAGATGC-3'), mr_21
(5'-GCATCTAAACAGGCCTCAATCATCTC-3'), mr_22
(5'-GCGGTAAAGGCGTTAGTAGAAAACAG-3'), mr_23
(5'-CTGTTTTCTACTAACGCCTTTACCGC-3'), mr_24
(5'-GGAGTTAGTAGAAGCCAGTCTGGATGC-3'), mr_25
(5'-GCATCCAGACTGGCTTCTACTAACTCC-3'), mr_26 (5'-GCTGCACGCTGACTGTGG-3'),
mr_34 (5'-GATTATTCATACCGTCCCACC-3'), mr_38
(5'-CAGATCCAAAACAATGGCACC-3'), mr_39 (5'-GGTGCCATTGTTTTGGATCTG-3'), mr_40 (5'-GAAGTTTCAAACAATGGATGTGG-3'), mr_41
(5'-CCACATCCATTGTTTGAAACTTC-3'), mr_42 (5'-CACACACGGAGTCACTAGGG-3'),
mr_55
(5'-GCGGATCCACCATGTCGTACTACCATCACCATCACCATCACGATTACGATATCCCAACGACCGAAAACCTNGTATTTTCAGGGCGAATTCGAGCGAGCTGAG-3'), 48_Homo, (5'-CCCAGTTGCACCCGTTTCTGGAGCCACGTTCGGTCTTAACCCGTCCAG-3'), 48_
1
(5'-CCCAGTTGCACCCGTTTCTGGAGCACACGTTCGGTCTTAACCCGTCCAG-3'), 48_Bottom (5'-CTGGACGGGTTAAGACCGAACGTGGCTCCAGAAACGGGTGCAACTGGG-3'), and 81BG-GTGT-
(5'-GACACAGGAAACAGCTATGACCATGATTACGCCAAGCTTGGCTGCAGGTTGTGCGACGGATCCCACTAGCCCAACTCATCC-3').
Overexpression of the hMutL Variants
Typically, 1.8 × 108 Spodoptera
frugiperda (Sf9) cells were infected with a combination of
two viruses encoding the hMLH1 and hPMS2 variants
at a multiplicity of infection of 10. 72 h after infection, cells
were harvested by centrifugation (1500 rpm, 10 min). After washing the
cells with 10 pellet volumes of cold phosphate-buffered saline
containing 0.5 mM PMSF, the cells were resuspended in 3 pellet volumes of cold buffer A (25 mM HEPES, pH 8.0, 2 mM 2-mercaptoethanol, 0.5 mM spermidine,
0.15 mM spermine, 0.5 mM PMSF, and 2×
CompleteEDTA-free (Roche Molecular Biochemicals). After
incubation (20 min, 4 °C), the cells were lysed in a Dounce
homogenizer (15 strokes). 2 pellet volumes of a cold 50% glycerol
solution and NaCl (to a final concentration of 300 mM) were
added. After incubation (30 min, 4 °C, mixing), extracts were
centrifuged (36,000 rpm, 30 min, 4 °C in a Sorvall TH-641 rotor) and
stored in aliquots at 
80 °C.
Purification of the hMutL Variants
Extracts from infected Sf9 cells were diluted with an
equal volume of buffer H (25 mM HEPES, pH 7.6, 2 mM -ME) and centrifuged (36,000 rpm, 30 min, 4 °C in
a Sorvall TH-641 rotor). Filtered extracts (0.45 µm) were applied to
a 5-ml heparin HiTrap column (AP Biotech) equilibrated with buffer H
containing 150 mM NaCl, and proteins were eluted with a
rising 2%/ml salt gradient. Pooled fractions containing the hMutL
variants were adjusted to 5 mM imidazol. An appropriate
volume of Ni2+-nitrilotriacetic acid-agarose (Stratagene)
(~7 ml of a 50% slurry per 100 mg of extracts from Sf9
cells expressing wild-type hMutL) was added. After incubation on an
end-over-end shaker for 60 min at 4 °C, extracts were transferred
into dispensable plastic columns (Bio-Rad). The columns were washed
five times with 5 column volumes of cold buffer WB (300 mM
NaCl, 10% glycerol, 20 mM HEPES, pH 8, 2 mM
2-mercaptoethanol, 200 mM PMSF) containing 10 mM imidazol. Bound proteins were eluted with 5 column
volumes of buffer WB containing 100 mM imidazol and with 5 column volumes of buffer WB containing 200 mM imidazol.
Fractions containing the highly purified MutL variants were pooled,
adjusted to 150 mM NaCl, and concentrated on a 1-ml
Resource-Q (AP Biotech) column. Prior to the injection into the
Resource-Q column, proteins were passed through a 6-ml Resource-S
column to remove possible contaminants. After injection, the Resource-S
column was removed, and proteins were eluted with a rising 4%/ml salt
gradient. The fractions containing the pure proteins were pooled,
dialyzed extensively against storage buffer (20 mM HEPES,
pH 7.6, 0.1 mM EDTA, 110 mM NaCl, 10% sucrose, 2 mM 2-mercaptoethanol, and 0.5 mM PMSF), and
stored in aliquots at 80 °C.
Other Proteins
The purified hMutS was described previously (43, 44).
UV Cross-linking
An aliquot of purified protein (1 µg) was incubated for 30 min
at 4 °C in 10 µl of cross-linking buffer (50 mM
Tris-HCl (pH 7.5), 6 mM MgCl2, 10% glycerol,
10% sucrose, 5 mM DTT, 0.2 mg/ml bovine serum albumin, and
0.5 µM [-32P]ATP). The samples were
placed on ice, irradiated for 5 min in an UV-Stratalinker (Stratagene),
and separated by 7.5% SDS-PAGE. Gels were extensively washed in fixing
solution (30% methanol, 10% acetic acid) and analyzed by autoradiography.
Partial Proteolysis Experiments
In a final volume of 20 µl, 3 µg (16.5 pmol) of the
recombinant proteins were incubated with 25 mM
Tris-Cl (pH 8), 120 mM NaCl, 1 mM DTT in the
presence or absence of 1 mM MgCl2. The ATP concentration was between 0 and 10 mM, as shown in Fig. 6.
After a 15-min preincubation at 20 °C, 10 ng of trypsin (Promega)
were added, and the reaction was allowed to proceed for 8 min at
20 °C. The reactions were stopped by the addition of 4 µl of 6×
SDS loading dye and boiling for 5 min at 95 °C. The samples were
loaded onto 10% SDS-polyacrylamide gels, and the protein bands were
visualized with Coomassie Blue.
ATPase Assays
The ATPase assays were carried out at 37 °C in a 15-µl
reaction mix containing 20 mM Tris-Cl (pH 8), 120 mM NaCl, 5 mM MgCl2, 1 mM DTT, 400 µM cold ATP, 222 nM
[
-32P]ATP, and 60 pmol of purified protein (4 µM). To test the DNA-dependent stimulation of
the hMutL heterodimer, either an 81-mer single-stranded oligonucleotide (81BG-GTGT-; final concentration 12 µM) or a 3193-nucleotide-long single-stranded and
covalently closed phagemid DNA (pGEM-T; final concentration
2, 20, or 200 nM) were added to the reactions (data not
shown). At selected time points, 2-µl aliquots were removed and mixed
with 5 µl of formamide-loading dye. 2-µl aliquots were loaded onto
20% denaturing polyacrylamide gels, which were then exposed to Biomax
MR films (Eastman Kodak Co.). The bands were quantified using
ImageQuant Software version 1.2 (Molecular Dynamics, Inc., Sunnyvale,
CA). For the calculation of the Michaelis-Menten constants, the ATPase
activity was measured in the presence of varying concentrations of cold
ATP (200-1000 µM).
In Vitro MMR Assays
The MMR assays were carried out as described previously (45).
Briefly, cytoplasmic protein extracts were prepared from TK6 and HCT116
cell lines, using 5 × 108 cells harvested in the
exponential growth phase. After resuspension in ice-cold hypotonic
buffer (20 mM HEPES, pH 7.9, 5 mM KCl, 1.5 mM MgCl2, 0.1 mM PMSF, 1 mM DTT) at a density of 1 × 108 cells/ml,
the cells were allowed to swell for 10 min in a glass Dounce
homogenizer on ice and then lysed mechanically by applying four or more
strokes with a tight pestle. When more than 80% of cells were lysed,
the nuclei were pelleted, and the supernatant was centrifuged
(12,000 × g for 10 min at 4 °C) and stored in small
aliquots at 80 °C.
Wild type and mutant M13mp2 phage used to generate the
mismatch-containing heteroduplexes were kindly provided by Tom Kunkel. The heteroduplex DNA contained the indicated base/base mispair within
the coding sequence of the lacZ -complementation gene and a nick
either 5' or 3' from the mispair. 1 fmol of substrate was incubated
with 50 µg of cytoplasmic extract, supplemented where necessary with
200 ng of the purified recombinant hMutL variants. The repair
reaction (25 µl) contained 30 mM HEPES, pH 7.8; 7 mM MgCl2; 4 mM ATP; 200 µM each CTP, GTP, and UTP; 100 µM each
dATP, dCTP, dGTP, and dTTP; 40 mM creatine phosphate; 100 fmol of creatine phosphokinase; and 15 mM sodium phosphate, pH 7.5. After 20 min of incubation at 37 °C, the heteroduplex DNA
was purified and electroporated into E. coli NR9162 (mutS), plated on minimal medium in a soft agar layer containing 0.5 ml of a
log culture of E. coli CSH50, 0.5 mg of
isopropyl-1-thio--D-galactopyranoside, and 2 mg of
5-bromo-4-chloro-3-indolyl--D-galactopyranoside (X-gal). After incubation for 16 h at 37 °C, repair efficiency could be determined by analyzing the color of the plaques.
Bandshift Assays--
The bandshift experiments were performed
with either a 48-mer heteroduplex oligonucleotide containing a 1-base
insertion loop (1) or a homoduplex G/C. The substrates (80 fmol, 4 nM), hMutS (75 nM), and hMutL variants
(150 nM) were incubated for 20 min on ice in 20 µl of
binding buffer (20 mM HEPES, pH 7.6, 1 mM DTT, 50 µg/ml bovine serum albumin, 120 mM NaCl, 1.75 ng/µl
poly(dI-dC)·poly(dI-dC) (AP Biotech), 12.5% glycerol). If indicated,
2 mM MgCl2 and/or 0.5 mM ATP was
added prior to the addition of the proteins. 2 mM EDTA was
added to reactions in which MgCl2 was omitted. If indicated, monoclonal antibodies against hMSH6 (Serotec; 66H6, 0.2 µg/µl) or hMLH1 (Pharmingen; catalog no. 13291A, 0.05 µg/µl) were added 10 min after the start of the incubation. 5 µl of each sample was carefully loaded onto a 4% nondenaturing polyacrylamide gel
(acrylamide/bisacrylamide, 37.5:1). Gels were run in TAE buffer (pH
7.5) at 4 °C, and 32P-labeled protein-DNA complexes were
visualized using a STORM PhosphorImager (Molecular Dynamics). The
homoduplex and 1 substrates were constructed by annealing the
oligonucleotide 48_Homo or 48_1 with the complementary strand
(48_Bottom) that was labeled at its 5'-end with 32P using
the polynucleotide kinase.
| |
RESULTS |
Rationale for Mutational Analysis of the ATP Binding Domains of
hMutL--
Comparison of the primary amino acid sequences and the
available crystal structures of the nucleotide binding domains of MutL, NgyrB, and HSP90 revealed that the ATP binding sites, including all
known catalytic residues, are highly conserved (16, 19, 46). According
to the GHKL consensus motifs (13, 14), three invariant amino acids were
selected for mutagenesis (Fig.
1a, red
stars). Their choice was based on the recently determined structure of the amino-terminal fragment of MutL in complex with the
nonhydrolyzable ATP analogue ADPPNP, which revealed the distinct contributions of these residues toward the ATPase function (16) (Fig.
2). In MutL, the base of the ATP
nucleotide is bound in a deep hydrophobic pocket. The first residue, a
polar Asp-58 lies at the bottom of this pocket and forms a direct
hydrogen bond with the exocyclic amino group of the adenine base. In
HSP90, substitution of the equivalent aspartate by alanine or
asparagine completely abolished nucleotide binding and the in
vivo function of the protein (21, 47). The second residue, Asn-33
plays a crucial role in the binding of the Mg2+ ion,
because its side chain contributes both directly and via a water
molecule to the coordination of the metal ion. In GHLK ATPases, the
nucleotide adopts a kinked conformation and oxygen atoms from all three
phosphates contribute to the coordination of the metal. Accordingly,
MutL and HSP90 are unable to bind ATP in the absence of
Mg2+ (16, 19), and substitution of the equivalent
asparagine residue in HSP90 completely abolished its ATPase activity
(21). The third residue, Glu-29, is most likely involved in the
activation of the catalytic water molecule. In agreement with its
proposed role as a general base, substitution of this residue with
alanine abolished the ATPase activity of MutL (16, 17), HSP90 (47, 48),
and type II topoisomerases (18, 49) with only little or no effect on
the binding of the nucleotide.
View larger version (58K):
[in this window]
[in a new window]
|
Fig. 1.
a, conserved sequence motifs that
constitute the ATP binding pockets of GHKL ATPases. Left,
structure-based alignment of the amino-terminal sequences of
MutL (E. coli), hMLH1, hPMS2, gyrase B (E. coli),
and human HSP90 reveals the presence of four highly conserved motifs
(I-IV, red). Red stars
indicate amino acid residues that were substituted in this study. The
secondary structures of the nucleotide-bound proteins are schematically
drawn above the sequence alignment. Right, ribbon
diagram of the MutL-(1-349) crystal structure in complex with
ADPPNP. The polypeptide folds into two globular domains (amino
acids 20-220 and 224-331). Motifs I-IV (red) mark the
site of ATP binding within the first domain (see also Fig. 2). Note
that only one subunit of the dimeric protein complex is shown. (The
figure was prepared using file 1b63 from the Brookhaven Protein Data
Bank (17).) b, clustering of HNPCC germ line mutations in
the hMLH1 gene. All hMLH1 single amino acid
substitutions reported in the HNPCC data base were aligned along the
primary sequence of hMLH1. Amino acid substitutions shown in
red gave rise to low expression in other functional assays
(50, 52, 55). ?, the pathogenicity of the substitution is uncertain,
due to incomplete segregation data or because the variant was
proficient in a functional assay. Green boxes
indicate regions conserved in all MutL homologues. Red
boxes indicate the four conserved motifs that constitute the
ATP binding pocket of hMLH1. Blue boxes indicate
regions that are conserved among eukaryotic MLH1 homologues.
These regions have been implicated in the interaction with the
respective PMS homologues (36, 38).
|
|
View larger version (16K):
[in this window]
[in a new window]
|
Fig. 2.
ATP binding pocket of MutL. Amino acids
from the four conserved motifs (I-IV, red) are shown in
ribbon representation. Left, structure of the
nucleotide-free protein. Helix A (motif I) harbors the catalytic base
(Glu-29) and the asparagine residue (Asn-33) that is involved in the
coordination of the Mg2+ ion. Motifs II and IV are part of
an extended -sheet. Asp-58 sticks out from this -sheet to form a
hydrogen bond with the bound nucleotide. Note that helix D (motif III)
occupies the ATP binding pocket and has to unfold in order for the
nucleotide to gain access to its binding site. Center,
structure of the protein in complex with ADPPNP. In the presence
of the nucleotide, the amino-terminal part of helix D unfolds to form a
flexible loop involved in the binding of the phosphate groups (P-Loop).
ADPPNP and Mg2+ are shown in gray and
green, respectively, and water molecules are shown as
blue spheres. Right, amino acid
substitutions introduced into hMLH1 and hPMS2, respectively. (The
figure was prepared using the files 1BKN (left) and 1b63
(center) from the Brookhaven Protein Data Bank.)
|
|
To study the role of ATP binding and/or hydrolysis in the function of
hMutL, we used site-directed mutagenesis to generate variants of
hMLH1 or hPMS2 that carried single amino acid substitutions in one of
the three above sites (Fig. 2). The hMutL variants studied here
are the following: the ATP binding-deficient double mutant
hMutL (hMLH1(D63N)/hPMS2(D70N)), designated LDN/DN,
the Mg2+ binding-deficient double mutant hMutL
(hMLH1(N38A)/hPMS2(N45A)), designated LNA/NA, and the
catalytic site double mutant hMutL (hMLH1(E34A)/hPMS2(E41A)), designated LEA/EA. LDN/WT and
LWT/DN indicate mixed heterodimers where either the
hMLH1 or the hPMS2 subunit contains the relevant substitution, respectively.
Expression of hMutL Can Be Severely Affected by Single Amino
Acid Substitutions in the ATP Binding Domains of Its
Subunits--
Detailed analysis of the crystal structure of the
N-terminal MutL fragment or of modeled structures of the hMLH1
ATP binding domain suggested that none of the planned
substitutions should affect the overall protein fold. We were therefore
surprised to find that some of the amino acid changes, such as the
substitution of Asp-63 in the ATP binding pocket of hMLH1(D63N),
resulted in a dramatically reduced expression of the hMutL variants
containing this subunit (Fig. 3,
lanes 2 and 4). This effect was caused
presumably through destabilization of the mutated hMLH1 and the
consequent degradation of hPMS2, which is labile in the absence of the
hMLH1 subunit (37). Interestingly, the equivalent substitution (D70N) in the hPMS2 subunit had no apparent effect on protein expression (lane 3). Sequencing of the expression constructs
and detection of the full-length hMLH1 protein in these extracts by
Western blotting ruled out the possibility that the reduced expression was due to additional mutations, which might have caused the premature truncation of the proteins (data not shown). To rule out an artifact inherent to the baculovirus system, we also analyzed this substitution in a homologous expression system, based on human 293T cells, which do
not express hMLH1 (50). Co-transfection of these cells with vectors
encoding either wild type or mutant hMLH1 and hPMS2 proteins also led
to a drastic reduction of the expression levels of hMutL when the
hMLH1 subunit carried the D63N substitution (data not shown). Mutation
of the asparagine involved in coordination of the Mg2+ ion
had a deleterious effect on the expression of hMutL only if both
subunits carried the alanine substitution in their ATP binding pockets
(Fig. 3, lane 7). Because identical titers of the
baculovirus encoding hMLH1(N38A) were used for the parallel production
of LNA/WT and LNA/NA, the difference in
expression levels cannot be attributed to differences in the
multiplicity of infection or in the quality of the insect cells used in
the experiment. Our results suggest, therefore, that ATP binding by
hMLH1 and, to a lower extent, also by hPMS2 is critical for the stable
expression of the hMutL heterodimer.
View larger version (38K):
[in this window]
[in a new window]
|
Fig. 3.
Expression of hMutL
variants in baculovirus-infected Sf9 cells.
Sf9 cells were co-infected with baculovirus vectors encoding
either wild type or mutant forms of the untagged hMLH1 and hPMS2
proteins as indicated above the panel. Total cell
extracts (20 µg) were the separated on a 7.5% SDS-polyacrylamide gel
and stained with Coomassie Blue. Note that a very abundant band
migrated with the same size as hMLH1 (compare extracts of infected
versus noninfected cells shown in lane
10).
|
|
In contrast to the mutants described above, substitution of the
catalytic residues in hMLH1 (E34A) or in hPMS2 (E41A) affected neither
the expression nor the stability of the proteins (Fig. 3,
lanes 8 and 9). Given that these
latter mutants still bind ATP (Ref. 51; see also Fig. 6), our results
suggest that ATP binding, but not hydrolysis, is required for the
stabilization of hMutL.
Purification of the hMutL Variants--
In order to facilitate
the purification of the hMutL variants, all hPMS2 expression vectors
were modified to include an amino-terminal His6 tag. These
vectors were co-expressed with the untagged hMLH1 in Sf9 cells,
and the resulting heterodimers were purified by a three-step procedure
including chromatography on Ni2+-nitrilotriacetic acid
(Fig. 4; see "Experimental
Procedures"). The fact that the untagged hMLH1 variants co-purified
with the His6-hPMS2 variants in an equimolar ratio implies
that the ATP binding domains are unlikely to be involved in the
regulation of the stable dimerization of the carboxyl-terminal domains
of hMLH1 and hPMS2 (36, 38). However, the poor expression of the
hMLH1(D63N) variant precluded the purification of the
LDN/WT and LDN/DN heterodimers. In
addition, the LNA/NA variant was expressed in low
amounts and could not be purified to the same extent as the others.
This protein preparation contained a contaminant, which migrated
slightly slower than hMLH1 in SDS-PAGE (Fig. 4). Although the
contaminating protein did not appear to interfere with our assays, the
results of experiments in which it was used were interpreted with
caution.
View larger version (23K):
[in this window]
[in a new window]
|
Fig. 4.
Purity of the hMutL
variants used in this study. Purified hMutL variants (2 µg) were separated on a 12% SDS-polyacrylamide gel and stained with
Coomassie Blue. Note that hMLH1 runs as a doublet on these gels, as
previously reported (37). Due to the low expression levels,
LNA/NA was obtained only in a partially purified form in
low amounts (~0.5 µg loaded; see "Experimental
Procedures" for details). LDN/WT and
LDN/DN could not be purified.
|
|
ATP Cross-linking Reveals an Inherent Asymmetry of the Two ATP
Binding Sites of hMutL--
We used UV cross-linking to study the
ATP binding properties of the hMutL variants. Surprisingly, whereas
the hMLH1 subunit of the wild type hMutL complex was efficiently
cross-linked to [-32P]ATP, only little radioactivity
was found associated with the hPMS2 subunit (Fig.
5). In agreement with the extensive
interaction between the three phosphate groups of the nucleotide and
the Mg2+ ion observed in all structures of the GHLK
ATPases, UV cross-linking of ATP to the hMLH1 subunit showed an
absolute requirement for Mg2+. Correspondingly,
substitution of the Mg2+ coordinating asparagine with
alanine greatly reduced ATP cross-linking efficiency to the hMLH1
(N38A) mutant (Fig. 5). These results indicate that hMLH1 and hPMS2
bind the nucleotide with substantially different affinities and could
be explained in two ways, either by the binding of the nucleotide in
the hMLH1 subunit and not in hPMS2 or by the substantially longer
residence times of the bound ATP molecule within the nucleotide binding
pocket of hMLH1, which results in more efficient cross-linking. In an
attempt to understand which of these two possibilities applies, we
carried out partial proteolysis experiments and ATP hydrolysis
assays.
View larger version (31K):
[in this window]
[in a new window]
|
Fig. 5.
UV cross-linking of
[-32P]ATP to
hMutL variants. Purified hMutL
variants (1 µg) were incubated with [-32P]ATP. After
short UV irradiation, the proteins were separated on a 7.5%
SDS-polyacrylamide gel. Left, gel stained with Coomassie
Blue. Right, autoradiograph of the same gel. A faint band
was detected at the height of hPMS2 after very long exposures (data not
shown).
|
|
ATP Binding Brings About a Conformational Change in hMLH1 That Is
Detectable by Partial Proteolysis--
We incubated the wild type
hMutL heterodimer with or without ATP in the presence of trypsin. In
the absence of the nucleotide, the two polypeptides were degraded to
numerous peptide fragments, ranging from ~60 to ~33 kDa, and this
pattern was retained also in the presence of low (0.1 and 1 µM) ATP concentrations (Fig. 6a). However, in the presence
of 10 µM ATP, a novel band of about 39 kDa appeared (Fig.
6, arrow) and persisted up to an ATP concentration of 10 mM. Interestingly, no other major differences were detected in the peptide pattern. This suggests that one of the hMutL subunits bound ATP at a concentration of ~10 µM and above and
that this binding event resulted in a conformational change, which
altered its proteolytic pattern. The results of the ATP cross-linking experiments suggested that the polypeptide in question should be hMLH1.
This was confirmed through the partial proteolysis of the
LNA/WT and LWT/NA variants. The latter
heterodimer produced a tryptic peptide pattern that was identical to
the wild type hMutL. In contrast, the LNA/WT variant,
in which the ATP binding site of MLH1 was mutated, failed to give rise
to the 39-kDa proteolytic band (Fig. 6b). We were able to
confirm the identity of the band also in liquid chromatography/MS/MS experiments (data not shown), where it was unambiguously assigned to
the N terminus of hMLH1, since one of its tryptic peptides corresponded
to amino acid residues 10-18 (RLDETVVN) of this protein.
View larger version (40K):
[in this window]
[in a new window]
|
Fig. 6.
ATP and magnesium binding induce a
conformational change in the hMLH1 subunit. a, 3 µg
of purified wild type hMutL were incubated with increasing
concentrations of ATP and 1 mM Mg2+ for 15 min
at 20 °C. The proteins were then partially digested with 10 ng of
trypsin for further 8 min at 20 °C, and the tryptic peptides were
separated on 10% SDS-polyacrylamide gels and visualized with Coomassie
Blue. The proteolytic fragment appearing in the presence of >10
µM ATP is marked with an arrow. b,
the appearance of the 39-kDa proteolytic fragment (arrow) is
dependent on the presence of ATP and magnesium. The fragment was absent
from the partial tryptic digest of LNA/WT, which is
unable to bind ATP in the active site of the hMLH1 subunit. This
suggests that the fragment originates from the latter
polypeptide (see "Results" for details).
|
|
As can be seen in Fig. 6b, the 39-kDa proteolytic fragment
of hMLH1 only appears in the presence of both ATP and Mg2+.
This finding further confirms that the binding of the nucleotide is
absolutely dependent on the presence of the divalent metal cation, as
was shown for MutL (16) and HSP90 (19). The partial proteolysis
experiments further showed that the EA mutation in the catalytic site
of the hMLH1 subunit does not affect nucleotide binding, since the
ATP-dependent 39-kDa tryptic band was present in all three
variants, LWT/WT, LEA/WT, and
LEA/EA (Fig. 6c).
The ATPase Activity of hMutL Is Very Low and Is Not Stimulated
by Single-stranded DNA--
The ATPase activity of the highly purified
wild type hMutL is very low. Our semiquantitative estimates put the
ATPase activity of wild type hMutL in a similar range as that of
MutL (i.e. ~0.5 min1; Refs. 16 and 24 and
data not shown). As anticipated, the ATPase activity of the
LEA/EA heterodimer was very close to background.
Interestingly, mutations in the ATP catalytic site of hMLH1 and hPMS2
had different effects. Thus, while the hMLH1 mutant
LEA/WT displayed an ATPase activity that was
intermediate between the wild type and the double mutant factors, the
activity of the LWT/EA variant was severely attenuated,
being close to that of LEA/EA (Fig.
7).
View larger version (17K):
[in this window]
[in a new window]
|
Fig. 7.
ATPase activity of
hMutL. Purified LWT/WT or
the catalytic single or double mutants LEA/WT,
LWT/EA, and LEA/EA (4 µM)
were incubated with 400 µM [-32P]ATP.
After the indicated time, the samples were separated on 20% denaturing
polyacrylamide gels and exposed to Biomax MR film, and the bands were
quantitated using ImageQuant software (Molecular Dynamics). The assay
was performed in triplicate using three independent preparations of the
recombinant proteins (error bars show the
S.D.).
|
|
In contrast to its prokaryotic homologue (17), the ATPase activity of
the wild type hMutL complex was not stimulated by the addition of
single-stranded DNA (81-mer oligonucleotide or 3.2-kb phagemid; data
not shown).
Substitutions in the ATP Binding Domains of hMLH1 and hPMS2 Affect
the Biological Activity of hMutL in in Vitro MMR Assay to Similar
Extents--
We tested the biological activity of the hMutL
variants in an in vitro MMR assay, in which the recombinant
proteins were used to complement the MMR-deficient extract of HCT116
cells that lack hMutL (37). As shown in Fig.
8, no repair of heteroduplex plasmid
substrates containing a nick located either 5' or 3' from a single GT
mismatch was observed after 20 min of incubation with the HCT116
extracts. In contrast, more than 80% of the substrate were repaired if
wild type hMutL (200 ng) was added to these extracts. Titration
experiments revealed that the MMR complementation efficiency of the
His-tagged hMutL was quantitatively indistinguishable from that of
an untagged hMutL complex (data not shown; Ref. 37). Moreover, since
removal of the His tag by the TEV protease did not change the
complementation efficiency of the heterodimer (data not shown), we
concluded that the presence of the His tag on the hPMS2 subunit did not
have any effect on the biological activity of hMutL.
View larger version (22K):
[in this window]
[in a new window]
|
Fig. 8.
ATPase activity of hMutL
is required for mismatch correction. Purified hMutL
variants (200 ng) were used to complement extracts of mismatch
repair-deficient HCT116 cells. The substrates were M13 heteroduplexes
carrying a G/T mispair and a strand discrimination signal (a nick)
either 3' or 5' from the mismatch (see "Experimental Procedures").
Similar results were obtained when the HCT116 extracts were
supplemented with extracts from Sf9 cells overexpressing the
hMutL variants, ruling out the possibility that the mismatch repair
activity of the mutant heterodimers was lost during the purification
procedure (data not shown). Because the repair activity of various
MMR-deficient cell lines fluctuates between 0 and 25% in this assay
(data not shown), levels below 25% are scored as MMR-deficient.
|
|
Substitution of the catalytic glutamate residue affected the in
vitro MMR activity of hMutL to different extents. Thus, while the catalytic double mutant was inactive in the complementation assay,
the variants carrying the substitution in only one subunit displayed
intermediate levels of MMR activity. Importantly, repair from both the
5' or 3' direction was affected to a similar extent, regardless of
which subunit carried the substitution. These results clearly
demonstrate that although the ATPase activities of both subunits are
required for efficient MMR, the process can still function, albeit less
efficiently, when only one subunit can hydrolyze the nucleotide. This
was not the case when the ATP binding ability of the hMutL subunits
was affected; in this case, already a single subunit mutation resulted
in the loss of complementation activity, which implies that ATP binding
by both subunits is required for MMR.
ATP Stimulates the Formation of a Ternary Complex between hMutL,
hMutS, and DNA--
The interaction between purified hMutL and
hMutS on mismatched DNA was investigated in bandshift assays using a
48-mer homoduplex (G/C) or a similar 48/49-mer heteroduplex substrate
containing a single nucleotide insertion in one strand (Fig.
9, left panel). In the
presence of hMutS, a specific protein-DNA complex (*) was formed,
which dissociated upon the addition of 0.5 mM ATP or
ATP-Mg2+ (see also Ref. 43). However, while ATP
brought about the dissociation of hMutS from the oligonucleotide
substrate, it promoted the formation of a ternary complex (**) between
the oligonucleotide, hMutS, and hMutL. This species appears to
contain both heterodimers, since the addition of anti-hMSH6,
anti-hMLH1, or anti-hPMS2 antibodies resulted in a further retardation
of the complex (***). Interestingly, formation of the ternary complex
was dependent on the presence of ATP but not magnesium. Since hMutL
does not undergo an ATP-driven conformational change in the absence of
the metal ion (Fig. 6b), we concluded from these experiments
that interaction of the hMutS and hMutL heterodimers requires ATP
binding, but not hydrolysis, in the hMutS factor. These results
support and extend the findings reported for the S. cerevisiae MutS and MutL homologues (52-54).
View larger version (43K):
[in this window]
[in a new window]
|
Fig. 9.
Interaction of hMutL
variants with hMutS on heteroduplex
DNA. Left, bandshift analysis performed with wild type
hMutS (75 nM) and hMutL (150 nM), using
32P-labeled 48-mer homoduplex G/C or a 48/49-mer
heteroduplex oligonucleotide containing a one-nucleotide
insertion/deletion loop (4 nM). Where indicated, monoclonal
antibodies against hMSH6 (+1)) or against hMLH1
(+2)) were added to the reactions. The free substrate
was separated from protein-bound complexes on 4% native polyacrylamide
gels, which were then analyzed on a STORM PhosphorImager using
ImageQuant software (Molecular Dynamics). Where indicated, reactions
contained 2 mM MgCl2 and/or 0.5 mM
ATP. Right, bandshift analysis carried out with hMutL
variants. All reactions contained wild-type hMutS (75 nM), the indicated hMutL variants (150 nM),
and 0.5 mM ATP. MgCl2 was omitted from all
reactions. Note that due to the low concentration of the purified
protein, the reaction in lane 7 contained only 75 nM LNA/NA. However, the presence of the
band, albeit only a weak one, at the height of the supershift (**)
suggests that also this mutant was able to interact with DNA-bound
hMutS.
|
|
Mutations in the ATP Binding Domains of hMutL Do Not Affect the
Formation of the Ternary Complex with hMutS on DNA--
The
ATP-dependent interaction between hMutL and hMutS
observed in the bandshift experiments might reflect the recruitment of
hMutL to the hMutS-bound mispair at a very early step during the
repair reaction. We were therefore interested to see whether the
in vitro complementation defect of the hMutL variants was caused by their inability to interact with the hMutS complex. As
shown in Fig. 9 (right panel), all of the hMutL variants
that we were able to purify in sufficient amounts were able to form the
ternary complexes (**), which suggests that the nucleotide dependence
of the ternary complex formation is not dictated by the hMutL
heterodimer. (Note that the supershifted band in lane 7 of the right panel of Fig. 9 due to
LNA/NA is weaker than that observed in the other lanes.
This is not due to the reduced ability of the protein to form a ternary
complex but rather to the lower amounts of the heterodimer used in this experiment. See also the legend to Fig. 9.) Taken together, our results
suggest that the deleterious effect of the substitutions in the ATP
binding sites of hMutL is linked to a step downstream from the
recruitment of the heterodimer to the DNA-bound hMutS complex.
| |
DISCUSSION |
Despite the identification of several MutL homologues in the human
genome, the hMLH1/hPMS2 heterodimer hMutL plays the predominant role
in postreplicative MMR. This is witnessed primarily by the fact that
human tumor cell lines carrying mutations in either gene have mutator
phenotypes of similar magnitude (32, 34). Since this shows that both
subunits of hMutL are equally important for MMR, the finding that
the hPMS2 gene was only extremely rarely mutated in HNPCC
families was unexpected and implied either that the two genetic loci
are differentially susceptible to mutagenesis or that mutations in
hPMS2 are phenotypically silent. The results of our present
investigation tend to support the former hypothesis, since they
revealed major differences in the effects that the seemingly equivalent
mutations had on the two polypeptides. Thus, the conservative
substitution of an aspartate by an asparagine within the NBDs of hMLH1
and hPMS2 was anticipated to have no effect on the folding of the
polypeptides. This expectation was based on the finding that the
HSP90(D93N) mutant could be shown to retain its structural fold (47)
and was substantiated in the case of the hMutL variant carrying the
D70N mutation in the ATP binding domain of hPMS2, which was expressed
with an efficiency similar to wild type. Unexpectedly, the D63N
mutation that abolished ATP binding in the hMLH1 subunit deleteriously
affected the expression efficiency of the heterodimer (Fig. 3),
similarly to observations made with hMutL variants carrying missense
mutations identified in HNPCC families that mapped close to the ATP
binding site of hMLH1 (Fig. 1b; Refs. 50, 52, and 55). This
effect may be linked with an increased susceptibility of the mutated
protein to degradation, as suggested by experiments where ATP binding was shown to protect the amino terminus of the S. cerevisiae
MLH1 against proteolysis (51). These results demonstrated that the ability to stably bind ATP is essential for the stability of hMLH1 but
not hPMS2 and implied that the two polypeptides may have different affinities for the nucleotide, despite the conservation of their NBDs.
This latter prediction could be substantiated by the greater efficiency
of [-32P]ATP cross-linking to hMLH1 (Fig. 5). We
postulate that the ATP binding site of hMLH1 needs to be occupied at
physiological ATP concentrations and thus that it is adversely affected
by the absence of the nucleotide, whereas the ATP binding-deficient
variant of hPMS2 would not be destabilized, since this protein does not
appear to require bound ATP for correct folding or stabilization
against proteolysis. Given the importance of ATP binding to hMLH1
stability, it was surprising to see that the LNA/WT
variant was expressed at levels similar to the wild type protein. This
phenomenon could be explained if we assume that the affinity of this
variant for ATP was strongly reduced but not eliminated. In this
scenario, the low level of ATP binding observed in the UV cross-linking
experiments (Fig. 5) might be sufficient to stabilize the protein but
too low to support its biological activity (see also below).
Interestingly, while the ATP cross-linking (Fig. 5) and partial
proteolysis (Fig. 6) experiments implied that the hMLH1 subunit of
hMutL bound the nucleotide with higher affinity, attenuation of ATP
binding in the hPMS2 subunit affected the ATPase activity of the
heterodimer to a greater extent than did the same mutation in the hMLH1
subunit. These findings suggest that hPMS2 is catalytically more active
than hMLH1 and that the ATP residence times of the nucleotide in the
ATP binding pocket of hPMS2 might be too short for efficient
cross-linking or for effective protection from trypsin. These findings
are supported by those of Yang and colleagues (56), who demonstrated
that the isolated N-terminal domain of hPMS2 has ATPase activity.
To test how the observed differences in ATPase activities of the two
hMutL subunits affected the biological activity of the hMutL
variants under study, we tested the recombinant factors in in
vitro MMR assays. As anticipated, the LEA/EA and
LNA/NA double mutants that were unable to hydrolyze or
bind ATP, respectively, were unable to complement the
hMutL-deficient extracts of HCT116 cells (Fig. 8). The same was
true for the single subunit mutants that were deficient in nucleotide
binding. These results show that ATP binding in both subunits is
essential for MMR. Surprisingly, the LWT/EA and
LEA/WT hMutL variants, in which the hydrolytic
capacity was attenuated in one or the other subunit, displayed
significant but similar reductions in MMR efficiency. This result was
unexpected, given that the ATPase activities of the two heterodimers
were affected by the respective mutations to different extents, and
implied that hMutL must be able to hydrolyze ATP in order to be
active in MMR but that the activity can be extremely low (Fig. 8).
Interestingly, we made a similar observation when we mutated the ATP
binding sites of the hMSH2 and hMSH6 subunits of hMutS (43). Whereas the latter subunit contributed more toward the ATPase activity of the
heterodimer, the effect of mutations in hMSH2 and hMSH6 on mismatch
repair efficiency in vitro was similar.
While this study was in progress, the mutator phenotypes of knock-in
S. cerevisiae strains carrying mutations within the ATPase domains of MLH1 or PMS1 (the functional homologue of hPMS2) were described (51). The substitutions altered amino acids corresponding to
the catalytic glutamates or to invariant glycine residues located in
the respective phosphate binding loops of the proteins (motif III in
Figs. 1 and 2), and were predicted, similarly to our study, to abolish
nucleotide hydrolysis or nucleotide binding, respectively. In agreement
with our results, the strain in which both genes carried a substitution
of the catalytic glutamate residue displayed a strong mutator
phenotype, while mutation of only a single MutL subunit produced a
weaker mutator effect. However, in the yeast system, the
mlh1 single mutant was more affected than the corresponding pms1 variant, and this functional asymmetry became even more
pronounced for mutants that carried the glycine to alanine
substitutions; the mlh1-G98A strain showed a full mutator
phenotype comparable with an MMR-deficient strain, whereas the
pms1-G128A strain was only a very weak mutator. This latter
finding cannot be easily explained. Our results showed that ATP binding
is more important than ATP hydrolysis, and similar data were reported
for "heterodimeric" topoisomerase II variants (57-59) mentioned
above. Thus, assuming that the G128A mutation in the P-loop motif of
the S. cerevisiae protein abolished ATP binding, the mutator
phenotype of the pms1-G128A strain would have been expected
to be stronger than that of the PMS1 mutant deficient in ATP
hydrolysis. The discrepancy between our results and those generated by
Liskay and colleagues (51) suggests that mutation of the
conserved glycine within the P-loop of the yeast PMS1 protein gave rise
to a complex phenotype, which cannot be explained simply by the loss of
nucleotide binding.
The functional asymmetry observed with the yeast MutL was suggested
to reflect an unequal contribution of the two subunits toward the
repair events with a 5' 3' or a 3' 5' directionality (51). Such
an interpretation would require the two MutL subunits to interact
with different downstream factors and the two ATP binding domains to
act independently of each other. We addressed this question by
comparing the repair efficiencies of the single mutants on substrates
that contained the nick either on the 5' or the 3' side of the
mismatch. The finding that these substitutions affected MMR from both
directions to a similar extent implies that the above hypothesis does
not apply to the human system and that the two ATP binding domains of
hMutL act in an interdependent way. Evidence for such a coordinated
function is provided by genetic analysis (60) and by elegant two-hybrid
experiments carried out with the N-terminal fragments of yeast MutL
(51), which clearly showed that ATP binding at both subunits is
required in order for the nucleotide binding domains to dimerize.
Assuming that the carboxyl-termini of the full-length proteins remain
stably associated, the function of the amino termini could be to form an ATP-driven molecular clamp, which then interacts with the downstream MMR factors.
Taken together, the data presented in this study allowed us to propose
a model for hMutL function, in which ATP binding operates a
molecular clamp formed by the NBDs of its two subunits. The ATP binding
site of the hMLH1 subunit might be mostly occupied, whereas that of the
hPMS2 subunit might oscillate between the bound and the unbound form
and thus ensure that the clamp does not close stably. Interaction with
DNA and/or other MMR proteins (e.g. mismatch-bound hMutS)
might alter the nucleotide binding properties of the ATP binding pocket
of the hPMS2 subunit, which could result in closing of the clamp and in
immobilization of hMutL at its site of action. The heterodimer might
then be recycled by the hydrolysis of the nucleotides, which would lead
to opening and release of the clamp. This mode of action would explain
the MMR phenotype of all of our mutants. While variants unable to bind
ATP at one or both of their subunits would fail to close the clamp, the
catalytic double mutants might be MMR-deficient, because their clamp
would remain closed. The partial defect in MMR of the single catalytic
mutants could also be explained if we assume that ATP hydrolysis
catalyzed by only one subunit might be sufficient to induce the
dissociation of the amino termini and thus to recycle the repair
factor, but more slowly.
The biological role of the MutL homologues has been frequently
described as that of "molecular matchmakers." Studies carried out
with the prokaryotic MutL suggest that it functions as a bridging molecule that transmits signals from the mismatch-bound MutS to the
downstream factors. MutL could be shown to interact with the DNA
helicase II and to load it at the site of the nick, where the process
of exonucleolytic degradation of DNA is initiated. This nick was
introduced into the DNA by MutH, the endonucleolytic activity of which
could be correspondingly shown to be activated by MutL in a reaction
that requires ATP binding but not hydrolysis (4, 16). This activation
is even more efficient in the presence of MutS and a mismatched
substrate, but in this case it requires ATP hydrolysis. That the ATPase
activity of MutL is involved was shown by the inability of the MutL
mutant (E29A), which binds ATP but fails to hydrolyze it, to bring
about the full, MutS-dependent activation of MutH (4, 5,
16, 61). Although the mechanism of strand discrimination in eukaryotes
differs from that of E. coli, the role of the eukaryotic
MutL homologues lies most likely in controlling protein/protein
interactions, as in the prokaryotic system. In eukaryotic cells,
biochemical studies suggested that MutS and MutL interact on DNA
in an ATP-dependent manner (53, 62, 63). The human factors
are capable of similar interactions (Fig. 9 and Ref. 64).
Interestingly, our study shows that because the ternary complexes were
formed between hMutS and all of the hMutL variants, their
formation could not be driven by ATP binding within the hMutL
component. This prediction was further substantiated by the formation
of the MutS-MutL-DNA complexes in the absence of magnesium
(i.e. under experimental conditions where hMutL does not
bind ATP). It would therefore appear that the association of hMutS,
hMutL, and mismatched DNA is driven by ATP binding within hMutS,
since this heterodimer was shown previously to undergo an ATP-driven
conformational change that was independent of nucleotide hydrolysis
(44). Although the supershift experiments leave many questions open,
they show that the hMutL variants studied here retained their
overall structures, such that they were still able to interact with
DNA-bound hMutS. We therefore conclude that the observed lack of
activity of these proteins in our in vitro MMR assay must be
due to their failure to interact with components of the MMR machinery
that are downstream from mismatch recogni
Mismatch Repair Factor Affect Its ATPase Activity, but Not Its
Ability to Interact with hMutS
TOP
ABSTRACT
INTRODUCTION
