 |
INTRODUCTION |
The DNA mismatch repair system acts to recognize and repair
mispaired nucleotides and plays an important role in maintaining genomic integrity (for reviews see Refs. 1-3). The importance of this
mutation avoidance system can be seen in
HNPCC1 kindreds, where
germ-line mutations in the genes associated with mismatch repair lead
to microsatellite instability, elevated mutation rates, and
predisposition to cancer (4, 5). Similar defects in MMR have also been
found in spontaneous colon cancers and other solid tumors (6). Although
the precise functions of the human MMR proteins are just beginning to
be understood, the human MMR genes are homologous to those in the
relatively well characterized Escherichia coli MutHLS
strand-specific mismatch repair pathway (for review see Ref. 7).
Indeed, the degree of conservation between such divergent species
reflects the essential role of MMR. In bacteria, the MutS protein
recognizes mismatches and initiates the repair process. Next, MutS
interacts with MutL and translocates along the DNA strand, forming an
-loop structure (8). This process requires ATP hydrolysis and MutL
interaction, although the exact role MutL plays is unclear. Following
MutH incision of the unmethylated strand, a segment of DNA is excised
by Rec I, exonuclease I, or exonuclease VII, and the DNA is
re-synthesized by the polymerase III holoenzyme.
In humans, the MMR pathway is more complex than that of E. coli. At least six genes have now been demonstrated to be involved in eukaryotic MMR. These include the MutS homologs
hMSH2, hMSH6, and hMSH3 and the MutL
homologs hMLH1, hPMS2, and hPMS1 (for reviews see
Refs. 9 and 10). hMSH2 and hMSH6 have been shown to interact with each
other in the hMutS heterodimer that binds to heteroduplex DNA
(11-13). hMSH2 can also pair with hMSH3 to form the hMutS
heterodimer that preferentially binds to small insertion/deletion loops
(14, 15). Both hMSH2 and hMSH6 have intrinsic ATPase activity (16-18).
In the presence of mismatched DNA the hMutS heterodimer binds to
ATP, resulting in its release from or translocation along
oligonucleotide substrates (18-20). As with bacteria, the exact roles
of the MutL homologs are still unclear in human cells. It has been
shown that a heterodimer of hPMS2 and hMLH1 is capable of restoring
mismatch repair activity in human cancer cells that are hMLH1-deficient
(21), suggesting the existence of a hMutL heterodimer. In addition,
a heterodimer of hPMS1 and hMLH1 (hMutL) has recently been reported
(22).
Based on the E. coli model, it is hypothesized that the MutS
and MutL homologs interact in the MMR process. In yeast purified yMSH2,
yMLH1, and yPMS1 (homologous to human PMS2) proteins have been
demonstrated by gel-shift analysis to interact with heteroduplex DNA
(23). In humans the hPMS2 and hMSH2 proteins have been shown to
co-immunoprecipitate with hMLH1 (24). The role of ATP in these
eukaryotic MutS/MutL homolog interactions is unclear. In yeast the
MutS-MutL homologs were reported to interact in the presence and
absence of additional ATP (23), whereas in other studies using purified
proteins ATP was necessary to observe the interaction (25). In humans,
the co-precipitation of hPMS2 and hMSH2 with hMLH1 was observed to
occur only in the presence of ATP (24). It has been suggested that the
MutL homologs could act to modulate the ATPase activity of the hMutS
heterodimer or that they act as "molecular matchmakers," recruiting
other proteins that function in steps of DNA mispair correction
subsequent to mismatch recognition.
To determine the exact interactions within nuclear extracts between MMR
proteins and oligonucleotide substrates before and after addition of
ATP we performed immunoprecipitation and Western blotting assays
together with UV cross-linking to homoduplex and heteroduplex
oligonucleotide substrates. We find that hMSH2, hMSH6, hMLH1, and
either hPMS2 or hPMS1 (but not both together) co-immunoprecipitate in
the absence of added ATP, indicating the presence of MMR protein complexes consisting of at least four MMR proteins. The protein-protein interactions occur in the presence of both homoduplex and heteroduplex DNA; however, we find that the complex specifically binds only to the
mismatched substrate. This binding occurs primarily through hMSH6,
although we also observe hMSH2 cross-linked to heteroduplex DNA when
not in the larger MMR protein complexes. Addition of ATP results in a
marked decrease in the amount of binding of the complexes to the
oligonucleotide. Furthermore, co-immunoprecipitation followed by
Western blot analysis demonstrates that the protein-protein interaction
between hMSH2 and hMSH6 (hMutS) remains intact but that hMLH1,
hPMS2, and hPMS1 no longer associate either with each other or with the
hMutS complex. Interestingly, after the addition of ATP, hMLH1 has
an increased ability to co-immunoprecipitate heteroduplex DNA, although
it does not become UV cross-linked to it. This suggests that hMLH1
participates in a second protein complex involved in mismatch repair
that may be involved in subsequent steps in the repair process.
| |
EXPERIMENTAL PROCEDURES |
Nuclear Extracts, Antibodies, and Reagents--
HeLa nuclear
extracts were prepared essentially as described by Dignum et
al. (26) with final resuspension in a buffer consisting of 20 mM HEPES (pH 7.9), 20% glycerol, 0.1 M KCl,
0.2 mM EDTA, 0.2 mM phenylmethylsulfonyl
fluoride, and 0.5 mM dithiothreitol. Antibodies against
hMSH2 (polyclonal) were purchased from Oncogene (Cambridge, MA).
Antibodies against hMSH6 (polyclonal), hMLH1 (polyclonal), hPMS2
(polyclonal), and hPMS1 (polyclonal) were purchased from Santa Cruz
Biotechnology (Santa Cruz, CA). All other reagents were purchased from
Sigma unless otherwise noted.
Preparation of Oligonucleotides--
A 32-mer oligonucleotide
having the sequence of
5'-GGTGGGCGCCGGCGGTGTGGGCAAGAGTGCGC-3' (Operon;
Alameda, CA) was used for these assays with the bold G specifying the
location of the mismatch. The oligonucleotide was 5'-end-labeled using
[
-32P]ATP (6,000 Ci/mmol) (NEN Life Science Products)
and T4 polynucleotide kinase (Promega; Madison, WI) and purified using
a Sephadex G-25 spin column. The 32P-labeled
oligonucleotide was annealed to a complementary oligonucleotide containing a C (homoduplex) or a T (heteroduplex) opposite the bold G,
thereby producing double-stranded oligonucleotides with similar
specific radioactivities. The annealing was performed at a 1:3 molar
ratio of labeled to unlabeled oligonucleotide in a final volume of 100 µl in 10 mM Tris-HCl (pH 8.0), 10 mM
MgCl2. The solution was heated to 95 °C for 5 min and
then allowed to cool to room temperature over 2 h. We found no
detectable labeled single-stranded oligonucleotides present when these
substrates were examined by native polyacrylamide gel electrophoresis.
UV Cross-linking and Immunoprecipitation--
Nuclear
protein-DNA binding reactions were carried out in a buffer containing
4% glycerol, 1 mM MgCl2, 0.5 mM
EDTA, 0.5 mM dithiothreitol, 50 mM NaCl, 10 mM Tris-HCl (pH 7.5), 0.05 µg/ml poly(dI-dC)·poly(dI-dC), and 165 µg of nuclear extract in a total volume of 200 µl for 30 min at room temperature. A DNA mixture containing 1.05 pmol of 32P-labeled double-stranded
oligonucleotide (either heteroduplex or homoduplex) and 27 pmol of
unlabeled double-stranded homoduplex competitor (~1:25 molar ratio
32P-labeled to unlabeled oligonucleotide) was added, and
the mixtures were placed on ice for 40 min. All subsequent steps were
performed at 0-4 °C. The samples were cross-linked by a 5-min
exposure in an UV Stratalinker (Stratagene; La Jolla, CA) and then
immunoprecipitated overnight with 3 µg of the indicated antibody.
After incubation for 3 h with protein G-Sepharose (Amersham
Pharmacia Biotech), the immunoprecipitates were recovered by
centrifugation (200 × g, 1 min), washed 2 times with a
buffer consisting of 10 mM Tris-HCl (pH 8.0), 150 mM NaCl, 0.1% Triton X-100, 0.025% NaN3, and
0.1% bovine serum albumin, washed 1 time with a buffer consisting of 10 mM Tris-HCl (pH 8.0), 150 mM NaCl, and
0.025% NaN3, and then washed 1 time with a buffer
consisting of 50 mM Tris-HCl (pH 6.8). Samples were
resuspended in an SDS sample buffer, heated for 5 min at 100 °C, and
loaded on a 7.5% SDS-polyacrylamide gel. Following electrophoresis at
10 mA for 2 h, gels were dried and exposed to Biomax film (Eastman
Kodak Co.). Assays involving ATP were performed as above except that
ATP at 0.1 mM was added prior to the UV cross-linking step.
Western Blotting--
Nuclear protein was bound to the DNA
substrates, UV cross-linked, immunoprecipitated, and electrophoresed as
above. The resulting gel was then transferred onto a nitrocellulose
membrane for Western blot analysis. Membranes were blotted with a
primary antibody against either hMSH2, hMSH6, hMLH1, hPMS2, or hPMS1.
Bound antibodies were detected by ECL detection using a
biotin-conjugated secondary antibody and streptavidin-horseradish
peroxidase as described by the manufacturer (Amersham Pharmacia
Biotech). Between each probe the membrane was stripped by 30 min
incubation at 55 °C in 100 mM 2-mercaptoethanol, 2%
SDS, and 62.5 mM Tris-HCl (pH 6.7).
| |
RESULTS |
hMSH6, hMSH2, hMLH1, and Either hPMS2 or hPMS1 Form a MMR Protein
Complex That Interacts Specifically with Mismatched DNA--
To
determine the exact nature of the interactions between the MMR
proteins, immunoprecipitation experiments were performed followed by
Western blotting. HeLa nuclear extracts, which are mismatch repair
proficient (27), were incubated with a mixture of
32P-labeled DNA substrate (either homoduplex or
heteroduplex), unlabeled homoduplex competitor to control for
non-mismatch-specific interactions, and poly(dI-dC)·poly(dI-dC) (not
recognized as a mismatch) to control for nonspecific DNA binding. Use
of whole nuclear extracts instead of purified proteins allowed us to
search for novel protein interactions. The sequence of the
oligonucleotide substrate used was based on that of Ha-ras
with the mismatch located at the middle nucleotide of codon 10. Mutations at this location have not been found in any human tumors (28,
29), and we have previously demonstrated 96% correct repair of GT
mismatches at codon 10 in mammalian cells (30), lending further support
to the mismatch repair proficiency of our system.
Nuclear extracts were incubated with the DNA, the bound proteins were
UV cross-linked to the oligonucleotides, and immunoprecipitation was
performed followed by Western blotting. In Fig.
1, antibody against hMSH6 was used for
immunoprecipitation, and the resulting membrane was sequentially probed
for hMSH6, hMSH2, hMLH1, hPMS2, and hPMS1, with stripping of bound
antibody between each probe. As shown in lanes 1 and
2 of Fig. 1B, hMSH6 is capable of
co-immunoprecipitating hMSH2, hMLH1, hPMS2, and hPMS1, indicating the
existence of an MMR complex(es) consisting of at least these proteins.
These protein-protein interactions occur in the absence of additional
ATP and occur to an approximately equal extent in the presence of
homoduplex or heteroduplex DNA. The protein-protein interactions within
nuclear extracts can also be observed when the 32-mer oligonucleotides, both mismatch-containing and homoduplex, were omitted from the reaction
(results not shown). As poly(dI-dC)·poly(dI-dC) was included in all
reactions, we cannot conclusively state whether the complexes are
pre-formed in the absence of any DNA; however, this is not a relevant
in vivo condition for nuclear proteins.
View larger version (39K):
[in this window]
[in a new window]
|
Fig. 1.
hMSH6 immunoprecipitation showing UV
cross-linking to heteroduplex DNA and co-precipitation of MMR
proteins. A, HeLa nuclear extracts were incubated with
32P-labeled homoduplex DNA (GC, lane
1) or heteroduplex DNA (GT, lane 2). Bound
proteins were UV cross-linked to the DNA and immunoprecipitated with
antibody against hMSH6. An autoradiogram of the SDS-polyacrylamide gel
is shown. Arrows indicate mismatch-specific binding
activities (hMSH6 and proteins that run to positions of approximately
115 (A), 95 (B), and 50 kDa (C)).
B, HeLa nuclear extracts were incubated with DNA as above in
either the absence (lanes 1 and 2) or presence
(lanes 3) of 0.1 mM ATP. After hMSH6
immunoprecipitation and SDS-polyacrylamide electrophoresis, parallel
lanes from the same gel as in A were transferred to a
nitrocellulose membrane. Western blotting was performed with the
indicated antibody probes (Western antibody (ab)).
Chemiluminescence was used for detection with sequential stripping of
the membrane between probes. The large dark band at the
bottom of all gels results from detection of immunoglobulins
used in immunoprecipitation.
|
|
In these same experiments, autoradiography was performed on parallel
lanes to determine which of the proteins identified above were also
capable of becoming UV cross-linked to the 32P-labeled DNA.
As seen in Fig. 1A four bands are observed that bind
specifically to heteroduplex DNA (lane 2) and not to the similar homoduplex DNA (lane 1). By comparing the bands seen
in Fig. 1A to those identified in the Western blots in Fig.
1B, it can be seen that the highest molecular weight band
runs to approximately the same position on the gel as hMSH6. We believe
this to be hMSH6 bound to the mismatched DNA because 1) the size of
this band correlates with the predicted molecular mass of 160 kDa for
hMSH6, 2) binding to the mismatched DNA is disrupted upon addition of
ATP (see Fig. 5), and 3) a number of previous studies have indicated
the mismatch-specific DNA binding activity of hMSH6 (13, 17). We
therefore conclude that the predominant mismatch-specific DNA binding
activity that we observe within nuclear extracts in these studies is
hMSH6. This binding to the DNA occurs only with the mismatched
oligonucleotide and is not seen when the 32P-labeled
homoduplex substrate is used. We also observe mismatch-specific DNA-binding proteins that run to positions of approximately 115 (A), 95 (B), and 50 kDa (C) (indicated
with arrows). These proteins may be components of the
complexes involving hMSH2, hMSH6, hMLH1, and hPMS2/hPMS1.
Alternatively, as HeLa nuclear extracts are mismatch repair-proficient,
these bands may represent proteins that become bound to mismatched DNA
in subsequent steps in the repair process. The 115-kDa (A as
marked by arrows) activity correlates with the observed
positions of both hPMS2 and hPMS1. Further studies are necessary to
determine if this band represents a previously undetected DNA binding
activity of either the hPMS2 or hPMS1 protein.
To confirm that the MMR proteins exist as a protein complex within HeLa
nuclear extracts, the above experiments were repeated using antibody
against hMLH1 for the immunoprecipitation. Western blotting was then
performed with the membrane probed with antibodies against hMSH6,
hMSH2, hMLH1, hPMS2, and hPMS1. As seen in lanes 1 and
2 of Fig. 2 these experiments
confirm the interaction between hMLH1 and hMSH2, hMSH6, hPMS2, and
hPMS1 in the absence of added ATP. hMLH1 was able to
co-immunoprecipitate each of the other proteins in the presence of both
homoduplex and heteroduplex DNA in the absence of ATP. Similar
co-immunoprecipitation of all the MMR proteins was observed when
antibody against hMSH2 was used (results not shown). Interestingly,
when antibody against hPMS2 is used for immunoprecipitation, hPMS1 does
not co-immunoprecipitate (see Fig. 3). As
well, antibody against hPMS1 does not co-immunoprecipitate hPMS2.
As we observe co-precipitation of both hPMS2 and hPMS1 when antibodies
against hMSH6 (Fig. 1), hMLH1 (Fig. 2) or hMSH2 (results not shown) are
used for the initial immunoprecipitation, we conclude that either hPMS2
or hPMS1 exists in the MMR complex but not both together.
View larger version (43K):
[in this window]
[in a new window]
|
Fig. 2.
hMLH1 immunoprecipitation showing
co-precipitation of MMR proteins. HeLa nuclear extracts were
incubated with homoduplex (GC, lane 1) or
heteroduplex (GT, lanes 2 and 3) DNA
in the absence or presence of ATP. Bound proteins were UV cross-linked
to the DNA, and immunoprecipitation was performed with antibody against
hMLH1. Following SDS-polyacrylamide electrophoresis and transfer to
nitrocellulose, Western blotting was performed with the indicated
antibody probes (Western antibody (ab)), with stripping of
bound antibody between probes. Chemiluminescence was used for
detection.
|
|
View larger version (43K):
[in this window]
[in a new window]
|
Fig. 3.
hPMS1 and hPMS2 do not co-precipitate each
other. HeLa nuclear extracts were incubated with heteroduplex DNA
in the absence of ATP, and immunoprecipitation was performed using
antibody against hPMS1 (IP ab in lanes 1 and
4) or hPMS2 (IP ab in lanes 2 and
3). After SDS-polyacrylamide electrophoresis and transfer to
nitrocellulose, Western blotting was performed with the indicated
antibody probes (Western antibody (ab)). Chemiluminescence was used for
detection.
|
|
In Fig. 4 nuclear extracts were incubated
with 32P-labeled heteroduplex or homoduplex
oligonucleotides; the proteins were UV cross-linked to the DNA, and
then antibodies against four of the MMR proteins were used for separate
immunoprecipitations. The proteins that became UV cross-linked to the
labeled oligonucleotides were observed by autoradiography of the
resulting gel. The same four proteins (hMSH6, approximately 115 (A), 95 (B), and 50 kDa (C)) became
bound to the mismatched DNA and not to the homoduplex DNA when hMSH6,
hMSH2, hMLH1, hPMS2, or hPMS1 were used for immunoprecipitation. The
predominant mismatch-specific DNA binding activity is that of hMSH6 as
identified in Fig. 1, regardless of the antibody used for
immunoprecipitation. The degree of binding by hMSH6 is greater when
antibody against hMSH6 is used in a direct immunoprecipitation than
when hMSH6 is co-immunoprecipitated using antibodies against hMSH2,
hMLH1, hPMS2, or hPMS1. This is the expected result because during the
co-immunoprecipitations the protein-protein interactions can be
disrupted and thus further confirms that this band represents hMSH6.
Alternatively, it is possible that the intensity of this band results
from hMSH6 homomultimers bound to the oligonucleotides as well as
binding by the larger mismatch repair complexes. However, we consider
this to be unlikely as studies using purified hMSH6 have shown it to be
incapable of binding to DNA when not interacting with hMSH2 in the
hMutS heterodimer (13). The amount of hMSH6 UV cross-linked to the
heteroduplex when antibody against hMSH2 is used is greater than that
seen when antibody against hPMS1, hPMS2, or hMLH1 is used. This could
be because the hMutS complex alone binds to the mismatched DNA, in
addition to the larger MMR complexes, or could reflect the strength of
the protein-protein interaction between hMSH2 and hMSH6. When antibody
against hMSH2 is used to immunoprecipitate, a unique band is observed
of approximately 105 kDa. This corresponds to the known molecular mass
of hMSH2 (102 kDa) and runs to approximately the same position as hMSH2 as indicated on the Western blots in Fig. 1B and Fig. 2.
Other investigators have demonstrated purified hMSH2 to be capable of specifically binding to mismatched DNA (31). As we cannot
co-immunoprecipitate bound hMSH2 with antibodies against any of the
other MMR proteins, it appears that hMSH2 in HeLa nuclear extracts does
specifically bind to mismatched DNA but only when not associated with
hMSH6 or the larger MMR complexes.
View larger version (49K):
[in this window]
[in a new window]
|
Fig. 4.
UV cross-linking of nuclear proteins to
32P-labeled DNA and immunoprecipitation of the protein-DNA
complexes with antibodies against MMR proteins. The
32P-labeled 32-mer oligonucleotide duplex containing GC
(homoduplex) or GT (heteroduplex) was incubated with HeLa nuclear
extracts, and bound proteins were UV cross-linked to the DNA.
Immunoprecipitation was performed with the indicated antibodies
(IP ab) and the immunoprecipitates were electrophoresed on
an SDS-polyacrylamide gel. The figure is an autoradiogram of the
resulting gel.
|
|
Protein Interactions after Addition of ATP--
Nuclear extracts
were incubated with 32P-labeled heteroduplex or homoduplex
oligonucleotides, and the proteins were UV cross-linked to the DNA as
above, except that 0.1 mM ATP was added after incubation of
the nuclear extract with the 32P-labeled DNA but prior to
UV cross-linking. In Fig. 5, antibodies against hMSH6 and hMLH1 were used to immunoprecipitate proteins UV
cross-linked to 32P-labeled heteroduplex DNA with and
without addition of ATP. The amount of hMSH6 that is bound to the
mismatched substrate clearly decreases upon addition of ATP, regardless
of the MMR antibody used for immunoprecipitation. This correlates with
previous studies that have shown hMutS to disassociate from
oligonucleotide substrates upon binding of ATP (13, 16, 18, 19). As
well, when antibody against hMSH6 is used for immunoprecipitation after
the addition of ATP, the amount of binding to the mismatched DNA by the
other three proteins (approximately 115 (A), 95 (B), and 50 kDa (C)) also decreases. Binding by
hMSH6 does not completely disappear when antibody against hMSH6 is used
for immunoprecipitation in these experiments using 0.1 mM
ATP. ATP concentrations of 10 mM do, however, completely
obliterate hMSH6 binding (results not shown).
View larger version (58K):
[in this window]
[in a new window]
|
Fig. 5.
Autoradiogram of immunoprecipitates after UV
cross-linking in the absence or presence of ATP. The
32P-labeled oligonucleotides containing GC (homoduplex) or
GT (heteroduplex) were incubated with HeLa nuclear extracts. 0.1 mM ATP was added to samples in lanes 2, 4, and
6, and bound proteins were UV cross-linked to the DNA.
Immunoprecipitation was then performed using antibodies against the
indicated proteins (IP ab). After SDS-polyacrylamide
electrophoresis, the resulting gels were dried and exposed to film.
Arrows indicate hMSH6 and hMSH2 bound to
32P-labeled oligonucleotides and free oligonucleotides that
have co-immunoprecipitated.
|
|
Interestingly, we detect a novel interaction between hMLH1 and the
heteroduplex DNA when ATP is added to the nuclear extract-DNA incubation. As seen by comparing Fig. 5, lanes 5 and
6, when ATP is added to an hMLH1 immunoprecipitation, the
amount of mismatched oligonucleotide that co-precipitates increases, as
evidenced by the increased amount of unbound co-immunoprecipitated
oligonucleotide. This occurs specifically with the heteroduplex and not
the homoduplex oligonucleotide. This phenomenon does not occur when
antibody against any of the other MMR proteins is used. hMLH1 itself
(70 kDa) does not become directly UV cross-linked to the DNA as a band
at the expected site is not observed. However, the amount of labeled
DNA bound to the proteins that run to approximately 95 (B)
and 50 kDa (C) does increase. hMLH1, therefore, participates in an interaction with the heteroduplex DNA after addition of ATP. This
interaction may include the B and C proteins but does not appear to
involve a direct interaction between hMLH1 and the mismatched oligonucleotide.
To compare the protein-protein interactions in the MMR complexes after
addition of ATP, we also performed immunoprecipitation followed by
Western blot analysis. As seen in each lane 3 of Fig. 1B, when antibody against hMSH6 is used to immunoprecipitate
in the presence of ATP, hPMS2, hPMS1, and hMLH1 no longer
co-precipitate. The only protein-protein interaction that remains after
addition of ATP is that between hMSH2 and hMSH6. Each lane 3 of Fig. 2 demonstrates that hMLH1 no longer co-precipitates any of the
MMR proteins after addition of ATP. Similar results were observed when
antibodies against hPMS2, hPMS1, and hMSH2 were used for immunoprecipitation with all interactions disrupted after addition of
ATP except that between hMSH2 and hMSH6 (results not shown). The lack
of binding seen after addition of ATP in Figs. 1 and 2 also confirms
that the observed co-immunoprecipitations are the result of specific
MMR protein-protein interactions as these are known to be modulated by ATP.
| |
DISCUSSION |
To determine MMR protein-protein and protein-DNA interactions, we
performed UV-cross-linking, immunoprecipitation, and Western blotting
using HeLa nuclear extracts. In these studies we have identified
protein interactions between the human MutS homologs hMSH2 and hMSH6
and the human MutL homologs hMLH1, hPMS1, and hPMS2. We observe these
interactions to occur in the absence of ATP; thus ATP binding or
hydrolysis is not a prerequisite for complex formation in HeLa nuclear
extracts in these experiments. Although we observe both hPMS2 and hPMS1
in the complex, they do not co-precipitate each other. Thus the MMR
complex involves either hPMS2 or hPMS1 but not both together. This is
in agreement with a recent publication that describes the hMutL
complex of hMLH1 and hPMS1 as distinct from the hMutL complex of
hMLH1 and hPMS2 (22). Our results indicate that either hMutL or
hMutL is capable of interacting with hMutS and binding to
mismatched DNA. If these complexes are at least partially functionally
redundant, this would explain the rarity of HNPCC kindreds with
mutations in hPMS2 or hPMS1 (33). We theorize that these different
complexes have distinct, as yet unknown functions in the repair process.
The protein-protein interactions in nuclear extracts occur in the
presence of both homoduplex and heteroduplex DNA. However, the protein
complex binds specifically only to heteroduplex substrates. As
previously reported (17), hMSH6 appears to be the predominant DNA-binding protein in this complex, although we also observe binding
by proteins that run to approximately 115 (A), 95 (B), and 50 kDa (C). Interestingly, the position
of the 115-kDa (A) mismatch binding activity is similar to
that of hPMS2 and hPMS1 as identified by Western blotting. Current
studies are in progress to determine if either hPMS2 or hPMS1 has an as
yet unidentified DNA binding activity that allows it to become UV
cross-linked to the heteroduplex DNA when in the MMR protein complex,
or perhaps to function in subsequent steps in the repair process.
However, we do not see a stronger 115-kDa (A) band when either hPMS2 or hPMS1 is used for direct precipitation as compared with the
co-immunoprecipitations. If the 115-kDa (A) band is either hPMS2 or
hPMS1, a stronger band would be expected when using antibody directed
specifically against the protein, as during co-immunoprecipitation the
non-covalent protein-protein interactions are easily disrupted during
the immunoprecipitation procedure. This suggests that the 115-kDa (A)
band may represent a distinct protein. Interestingly, the position of
the B band correlates with the reported 94-kDa molecular mass of human
exonuclease 1 (34). This 5' 
3'-exonuclease has previously been
shown to interact physically with hMSH2 in human nuclear extracts (34), and genetic studies in yeast have shown the yeast homolog to function in the same epistatic pathway as hMSH2 (35, 36). The size of the C band
could correlate with one or more of the RPC II, III, IV, and V subunits
(37, 36, 40, and 38 kDa, respectively), which are suspected to play a
role in MMR through loading of proliferating cell nuclear antigen onto
DNA (2). It is also very probable that there are more proteins yet to
be identified as necessary for mismatch repair in humans, such as a DNA
helicase and a 3' 5'-exonuclease (2). The identification of the
three unknown mismatch binding proteins in our studies is therefore of interest.
We also observe hMSH2 to become UV cross-linked to the heteroduplex DNA
but only when antibody against hMSH2 is used for direct immunoprecipitation (Fig. 4). As this mismatch-specific DNA binding activity of hMSH2 does not co-immunoprecipitate with hMSH6, hMLH1, hPMS2, or hPMS1, we conclude that hMSH2 is able to bind to mismatched DNA but only when it is not participating in the larger MMR complexes identified here. One possibility could be that the hMSH2-protein interaction occurs when hMSH2 is bound to hMSH3 in the hMutS heterodimer. Alternatively, as studies with purified hMSH2 have shown
it to bind to mismatched oligonucleotides as a homomultimer (31), the
band we observe could be due to an interaction between hMSH2
homomultimers and the mismatched substrate in nuclear extracts. Further
studies are necessary to identify the proteins involved in this
interaction and their possible function.
After addition of ATP to the mixtures we observed that the
protein-protein interaction between hMLH1 and hPMS1 or hPMS2 is disrupted as well as the interaction between these proteins and the
hMutS heterodimer. The only protein-protein interaction that we
continue to observe after addition of ATP is that between hMSH2 and
hMSH6 (hMutS). This is in conflict with yeast studies that have
reported that the MSH2, MSH6, and PMS1 complex either remains intact
upon addition of ATP (23) or that the protein interactions increase in
the presence of ATP (25). We believe the most probable explanation for
this apparent discrepancy is that the previous studies used purified
proteins, whereas our experiments were performed using whole nuclear
extracts. In the presence of ATP, the MMR-proficient HeLa nuclear
extracts can proceed with the repair process resulting in the eventual
disruption of the protein-protein interactions. The excess of ATP then
prevents formation of the protein complex. Although the reactions were
maintained at 0-4 °C throughout, the immunoprecipitations were
performed overnight, allowing time for the mismatch repair processes to
occur, even at these sub-optimal temperatures. Alternatively, it has
been shown in E. coli that the heteroduplex substrate is
necessary for MutS and MutL complex formation (37). In our experiments
the size of the oligonucleotide substrate used (32-mer) could affect
the ability of the MutS and MutL homologs to form a complex with each
other and with the DNA after the addition of ATP. Addition of ATP also
decreases the amount of hMSH6 that becomes UV cross-linked to the DNA.
Numerous studies involving purified hMSH2 and hMSH6 have confirmed that the purified hMutS heterodimer disassociates from mismatch
containing oligonucleotides upon binding of ATP (13, 16, 18, 19).
In Fig. 5 we observe a novel interaction between hMLH1 and the
mismatched DNA after the addition of ATP, with hMLH1 capable of
co-immunoprecipitating the labeled heteroduplex and not homoduplex DNA.
This suggests that hMLH1 participates in a complex that interacts specifically with mismatched DNA after addition of ATP. This complex does not appear to involve hMSH6. hMLH1 does not interact directly with
the DNA as it does not become UV cross-linked to the oligonucleotide. Instead, hMLH1 may participate in a novel protein complex in which other proteins bind to the mismatched oligonucleotide. Other
investigators (21) have failed to observe direct interactions between
mismatch-containing DNA and the purified hMutL heterodimer. This is
understandable if the complex involves additional nuclear proteins that
directly interact with the heteroduplex DNA and that have not as yet
been identified. We do continue to observe mismatch-specific binding by
the proteins that run to approximately 95 (B) and 50 kDa (C), suggesting that they participate in the novel hMLH1 containing complex.
Our findings suggest that hMLH1 plays a role in steps in the repair
process subsequent to mismatch recognition. Although little is known in
regard to the exact biochemical role of hMLH1 in humans, a number of
studies in E. coli have shed some light on the subject. One
suggested role for MutL is to facilitate the interactions between MutS
and the proteins necessary for subsequent steps in the repair process
such as MutH and DNA helicase II (7, 38-41). The ability of hMLH1 to
participate in the initial complexes involving hMutS as well as in
the novel complex after addition of ATP suggests that it may have an
analogous role in human mismatch repair. As well, recent studies have
suggested that MutL acts as a molecular chaperone. This hypothesis
arises from the observed homology of MutL to the molecular chaperone
molecule Hsp90 (42) and recent x-ray crystallography studies that have
shown it to similarly bind ATP and have a weak ATPase activity (32,
43). Indeed, upon ATP hydrolysis the E. coli MutL protein
was observed to undergo conformational changes that are likely to
modulate the interactions between MutL and the other components of the
repair machinery. In hMLH1, conformational changes upon ATP hydrolysis
could result in the different protein-protein and protein-DNA
interactions that we observe after addition of ATP to the reaction.
Genetic analysis of HNPCC kindreds has demonstrated the importance of
the hMLH1 gene product in mismatch repair (33) with approximately 61%
of HNPCC kindreds reported as of May, 1999, carrying mutations in the
hMLH1 gene. The central role of hMLH1 in both the hMutL
and hMutL heterodimers (with hPMS2 and hPMS1, respectively) (21, 22)
and in both the MMR complexes described here helps to explain the
genetic data. As well, the novel interaction we observe between hMLH1
and heteroduplex DNA suggests that hMLH1 may have a further unique role
in steps in MMR subsequent to mismatch recognition.
Our results suggest a model for human MMR where the hMutS
heterodimer, in a complex with either hMutL or hMutL, recognizes and binds to mispairs. The formation of these protein-protein complexes
occurs in nuclear extracts in the absence of additional ATP. The
complex can then stably bind to mismatched DNA, again in the absence of
ATP. This protein-DNA complex contains either hPMS2 (hMutL) or hPMS1
(hMutL), suggesting possible distinct functions for each. In the
presence of ATP, conformational changes occur resulting in the eventual
release of the protein complex from the heteroduplex oligonucleotides
and the dissociation of hMLH1 from hMutS and hPMS2 or hPMS1. hMLH1
is then capable of further interaction with the mismatched DNA in a
novel manner, presumably playing a role in subsequent steps in the
repair process.
TOP
ABSTRACT
INTRODUCTION
To whom correspondence and requests for reprints should be
addressed: Biomedical Program, University of Alaska, 3211 Providence Dr., Anchorage, AK 99508. Tel.: 907-786-4859; Fax: 907-786-1946; E-mail: afkjw1@uaa.alaska.edu.
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
