 1997 Oxford University Press
Human Molecular Genetics, 1997, Vol. 6, No. 7
1117–1123
Human MSH2 binds to trinucleotide repeat DNA
structures associated with neurodegenerative
diseases
Christopher E. Pearson1,*, Amy Ewel2, Samir Acharya2, Richard A. Fishel2,* and
Richard R. Sinden1
1Center
for Genome Research, Institute of Biosciences and Technology in the Texas Medical Center, Texas A&M
University, 2121 West Holcombe, Houston, TX, 77030, USA and 2DNA Repair and Molecular Carcinogenesis
Program, Kimmel Cancer Center, Thomas Jefferson University, Philadelphia, PA 19107, USA
Received February 14, 1997; Revised and Accepted April 9, 1997
The expansion of trinucleotide repeat sequences is
associated with several neurodegenerative diseases.
The mechanism of this expansion is unknown but may
involve slipped-strand structures where adjacent
rather than perfect complementary sequences of a
trinucleotide repeat become paired. Here, we have
studied the interaction of the human mismatch repair
protein MSH2 with slipped-strand structures formed
from a triplet repeat sequence in order to address the
possible role of MSH2 in trinucleotide expansion.
Genomic clones of the myotonic dystrophy locus containing disease-relevant lengths of (CTG)n·(CAG)n
triplet repeats were examined. We have constructed
two types of slipped-strand structures by annealing
complementary strands of DNA containing: (i) equal
numbers of trinucleotide repeats (homoduplex slipped
structures or S-DNA) or (ii) different numbers of repeats (heteroduplex slipped intermediates or SI-DNA).
SI-DNAs having an excess of either CTG or CAG repeats were structurally distinct and could be separated
electrophoretically and studied individually. Using a
band-shift assay, the MSH2 was shown to bind to both
S-DNA and SI-DNA in a structure-specific manner. The
affinity of MSH2 increased with the length of the repeat
sequence. Furthermore, MSH2 bound preferentially to
looped-out CAG repeat sequences, implicating a
strand asymmetry in MSH2 recognition. Our results are
consistent with the idea that MSH2 may participate in
trinucleotide repeat expansion via its role in repair and/
or recombination.
INTRODUCTION
The etiology of nine human genetic diseases, including
Huntington’s disease, spinocerebellar ataxia type 1 (SCA-1),
spino-bulbar muscular atrophy, myotonic dystrophy (DM) and
fragile X-syndrome (FRAXA) have been traced to genetic
variation in the lengths of (CTG)n·(CAG)n or (CGG)n·(CCG)n
trinucleotide repeats within the affected locus (for a review see
ref. 1). In normal individuals, these genes contain 5–37
(CTG)n·(CAG)n repeats, while increases in the number of repeats
to 50–200 are associated with unstable alleles and disease
symptoms (1,2). Explosive expansion to >3000 repeats can occur
at the DM and FRAXA loci. The probability of expansion
strongly depends upon the number of existing trinucleotide
repeats; the longer the repeat tract, the greater the possibility for
expansion. This length-sensitive probability of mutation has led
to the term ‘dynamic mutation’ to describe this process (3). In
SCA-1 and FRAXA, the repeat tracts are normally interrupted
with the non-repeat units CAT and AGG respectively (4–7). In
each case, the non-repeat interruptions confer increased genetic
stability to the repeat tract, such that interrupted tracts are less
likely to undergo expansions. Loss of non-repeat interruptions,
and hence an enhanced length of the pure repeat tracts, are
correlated with instability and disease. In the case of FRAXA,
changes in the repeat length are polar and occur primarily in the
3′ end of the (CGG)n tract, implicating replication direction in the
process (6). For some of the diseases, there also appear to be
specific tissues as well as ‘windows’ for triplet expansion which
are believed to occur during gametogenesis and/or early postzygotic development (1).
The mechanism(s) of repeat expansion is unknown. Since both
the length and the sequence of the genetically unstable tandem
repeats are limited to only certain triplet repeats
[(CTG)n·(CAG)n, (CGG)n·(CCG)n, (AAG)n·(CTT)n], it is
thought that some type of DNA secondary structure may
contribute to trinucleotide instability in these neurodegenerative
diseases (1–13). Spontaneous mutations in prokaryotic and
eukaryotic (including human) cells frequently are associated with
simple DNA repeat sequences that can form alternative non-B
DNA structures (reviewed in 10 and 11). The formation of
slipped-strand structures between repeated trinucleotide sequences
is one possible mechanism that may contribute to expansion (Fig.
1). Such slipped-strand structures may be formed during DNA
*To whom correspondence should be addressed. C.E.P. Tel: +1 713 677 7676; Fax: +1 713 677 7689; Email: [email protected], or R.F. Tel: +1 215 503
1345; Fax: +1 215 923 1098; Email: [email protected]
1118 Human Molecular Genetics, 1997, Vol. 6, No. 7
via recognition and binding to DNA at the site of a mispair
(17–20). The yeast (21) and human (22,23) MSH2 protein also
bind specifically to duplex oligonucleotides containing single
base pair mismatched nucleotides as well as short (5–16
nucleotide) insertions/deletions loops (IDLs), suggesting that
these proteins play a role similar to that of the bacterial MutS.
Human tumor cell lines that are deficient in MSH2 display a
generalized increase in spontaneous mutation rates, instability in
simple repeat (microsatellite) sequences, and yield extracts that
are defective for single base pair and IDL mismatch repair in vitro
(24–26). Mutations in MSH2 appear to account for ∼50% of
hereditary nonpolyposis colon cancers (HNPCCs), suggesting
that MSH2 plays a central role in carcinogenesis (27). Furthermore, hMSH2 forms specific mispair binding complexes with
two other MutS homologs, MSH3 (28) and MSH6 (GTBP/p160)
(28,29). Although the contribution of mismatch repair to
trinucleotide repeat stability in human cells remains controversial
(30), we have demonstrated that mutations in the bacterial mutS,
mutL or mutH genes decrease the frequency of large deletions of
plasmid-borne trinucleotide repeats in E.coli (9), implicating a
functional mismatch repair system in the mechanism of large
deletions in bacteria. To address further a possible role for human
mismatch repair in trinucleotide repeat expansion, we have
investigated the binding of MSH2 to the slipped-strand structures
expected to be intermediates in trinucleotide expansions as a
model for the fundamental recognition process that is required in
mismatch repair.
Figure 1. Replication slippage models. (A) Slippage of the DNA polymerase
and the nascent strand backwards on the template strand would result in the
formation of structures containing an excess of repeats on the nascent strand,
resulting in expansion products. (B) Slippage of the DNA polymerase and the
nascent strand forward on the template strand would result in the formation of
structures containing an excess of repeats on the template strand, giving rise to
deletion products. Both expansion and deletion processes will give rise to
replicated duplex DNAs having a stretch of mismatched repeats (a heteroduplex region). The expected products (in the absence of repair) of two rounds of
replication are shown. For simplicity, we have assumed that structure formation
occurs only in the lagging strand, although it may occur in either strand. (C)
Alternative homoduplex repeat structures may form in non-replicating DNA
(12). Thin lines represent repeat tracts, broad lines are non-repetitive flanks.
replication, and may result in both expansions and deletions of the
trinucleotide repeated sequences (Fig. 1A and B) (12,13). In
addition, slipped-strand structures may also be formed between
complementary strands with identical numbers of repeats (Fig.
1C) (12), as may occur intrachromasomally or during meiotic
chromosome synapsis. All of these slipped-strand structures
contain unpaired nucleotides and may be substrates for mismatch
repair (12,13). Using genomic clones of the DM, SCA-1 and
FRAXA loci having normal, pre-mutant and full mutant lengths
of repeats, we previously have demonstrated the formation of an
apparent slipped-strand structure (12,14, C.E. Pearson unpublished results). In each case, the probability of slipped-strand
structure formation increased with increased repeat length. In the
case of SCA-1 (CAG)n repeats or FRAXA (CGG)n repeats the
presence of CAT or AGG interruptions, respectively, inhibited
structure formation (14, C.E. Pearson unpublished results). These
results support the participation of a slipped-strand structure in
the process of trinucleotide repeat instability.
The human homolog of the Escherichia coli MutS protein,
MSH2, is a member of the highly conserved family of mismatch
repair proteins (15,16). In E.coli, MutS initiates mismatch repair
RESULTS
Slipped structures
Two types of trinucleotide DNA structures were used—homoduplex
slipped strand DNA structures (S-DNA) (12), formed between
two complementary strands having equal numbers of repeats, and
heteroduplex slipped intermediate DNA structures (SI-DNA), in
which the strands have different numbers of repeats. Two
genomic clones derived from the human DM locus that have
identical non-repetitive flanking sequences and 30 or 50 triplet
repeats (Fig. 2A) (12), corresponding to the normal and expanded
alleles (1), respectively, were used. Plasmids were linearized by
HindIII digestion, 32P labeled (*) on the 3′ ends to identical
specific activity then treated by denaturation–renaturation (12).
For the production of S-DNA, only individual plasmids were
reduplexed and for SI-DNA equimolar amounts of the 30 and 50
triplet repeat-containing plasmids were mixed and heteroduplexed. Following renaturation, DNAs were digested with
EcoRI, resulting in DM inserts uniquely labeled on the 3′ end of
the CTG strand (12). Reduplexing resulted in a reproducible
pattern of slow migrating products not present in the non-treated
samples (12). The ethidium-stained patterns of the reduplexed n
= 30 or n = 50 S-DNAs were identical to their respective
autoradiographic patterns (Fig. 2, compare B with C). In the case
of reduplexed (CTG)30·(CAG)30-containing plasmid, 39% of the
repeat-containing fragment migrated as a series of closely spaced
distinct products between 232 and 410 bp, with a single region of
greatest intensity migrating as 287 bp (Fig. 2B). For reduplexed
(CTG)50·(CAG)50-containing plasmid, 70% of the repeatcontaining fragment migrated as a series of closely spaced distinct
products between 295 and 696 bp, with a single region of greatest
intensity migrating as 370 bp (Fig. 2B). A larger percentage of the
1119
Human
Genetics,
1997,
6, No.
NucleicMolecular
Acids Research,
1994,
Vol. Vol.
22, No.
1 7
1119
tures of the SI-DNAs containing looped-out CTG or CAG strands
are sufficient to confer different electrophoretic mobilities to the
sister heteroduplexes, with limited overlap. We previously have
demonstrated the electrophoretic resolution of two sister threeway junctions of identical sequence composition (31,32). For
both reduplexing and heteroduplexing reactions, the pattern and
relative proportion of products was highly reproducible.
Each of the linear non-treated DNAs, the major S-DNAs and
the major SI-DNAs were gel purified (12). We found that the
gel-purified products were stable as judged by a minimal
interconversion back to linear or other forms (compare Fig. 2C
with D) as previously observed (12). However, longer autoradiographic exposures (data not shown) revealed that both the isolated
SI-DNAs demonstrated some spontaneous self-association to
very slow migrating products (positions indicated by arrowheads
in Fig. 2D), whose intensity varied from one preparation to
another. It is likely that these represent intermolecular multimeric
associations similar to those observed for (CA)n·(TG)n and
(GA)n·(TC)n tracts (33,34).
Figure 2. Reduplexing/heteroduplexing DM DNA fragments results in slowly
migrating DNAs. (A) Map of the HindIII–EcoRI (CTG)n·(CAG)n-containing
fragments from DM genomic clones, where n = 30 and n = 50 (12). Human
non-repetitive flanking sequences (positions 357–375 and 391–433, as in ref.
2) are in bold, while the thin lines represent plasmid vector sequences. Positions
of the unique 3′ CTG radiolabels are indicated (*). LNT, linear non-treated
DNAs of n = 30 and n = 50 plasmids (lanes 1 and 4, respectively), reduplexed
DNAs (S) from n = 30 and n = 50 plasmids (lanes 2 and 5, respectively),
heteroduplexed DNAs (SI) of n = 30 and n = 50 plasmids (lanes 3 and 6,
respectively) and the *HindIII–EcoRI pUC19 vector band are indicated.
Products were separated on a 4% polyacrylamide gel, stained with ethidium
bromide (B) or dried and exposed for autoradiography (C and D). For (B) and
(C), samples and lane numbers are identical, and the upper and lower arrows
indicate the *(CTG)30·(CAG)50 and *(CTG)50·(CAG)30 SI-DNAs, respectively. Samples in (D) are gel-purified products from their corresponding lanes in
(C). These gel-purified products were loaded in equimolar amounts. Positions
of the 123 bp ladder are shown. As previously observed (12), a small amount
of conversion to the linear form is evident in both gel-purified S-DNA and
SI-DNA samples. The three filled arrowheads indicate post-elution multimeric
associations of SI-DNAs, which are more evident with longer exposures.
DNA containing expanded repeat length (n = 50) formed S-DNA
structures than DNA containing repeat lengths found in normal
individuals (n = 30) (Fig. 2B and C, compare lanes 2 and 5),
confirming a previously observed relationship that increasing
length of trinucleotide repeats results in increased slipped-strand
DNA products (12).
The ethidium-stained products of the *30×50 or *50×30
heteroduplexing reactions revealed that both n = 30 and n = 50
linear and S-DNA products were formed. The two major slower
migrating products and the fainter closely spaced products, that
only formed in the reactions that contained both the n = 30 and n
= 50 parent DNAs, are the heteroduplex SI-DNAs (Fig. 2B, see
arrows). The ethidium-stained pattern was identical, regardless of
whether n = 30 or the n = 50 was radiolabeled (Fig. 2B). By
selectively labeling the (CTG)30- or (CTG)50-containing strands,
it is apparent that the uppermost major band represents the
*(CTG)30·(CAG)50 heteroduplex (Fig. 2C, lane 3, see arrow),
while the faster migrating major band represents the
*(CTG)50·(CAG)30 heteroduplex (Fig. 2C, lane 6, see arrow). As
with the S-DNA structures formed from reannealing single parent
DNAs, there appear to be several structural isomers, with the
major product presumably representing the most energetically
favorable conformation. Apparently, the difference in the struc-
MSH2 binding
A band-shift assay was used to detect MSH2 binding to each of
the gel-purified substrate DNAs similar to that of Fishel et al.
(1994) (22,23). Under these conditions, MSH2 did not bind
significantly to the linear *(CTG)30·(CAG)30 or *(CTG)50·
(CAG)50 fragments, although, some low level binding was
observed (Fig. 3A, lanes 1–3 and 10–12, see open arrowheads).
Binding of hMSH2 to either of the S-DNAs resulted in a single
and distinct shifted complex (Fig. 3A, lanes 4–6 and 13–15, see
arrow). MSH2 also bound to the *(CTG)30·(CAG)50 SI-DNA,
resulting in a similarly shifted complex (Fig. 3A, lanes 7–9, see
arrow). MSH2 bound poorly to the *(CTG)50·(CAG)30 SI-DNA
(Fig. 3A, lanes 16–18; Fig. 3B). These results suggest that
purified hMSH2 binds to the slipped-strand S-DNAs and
SI-DNAs. Moreover, MSH2 appears to bind the SI-DNAs in a
structure- or sequence-specific manner in which there is preferential binding to looped-out CAG repeats (Fig. 3B and C).
Binding specificity
To test the specificity of MSH2 binding, competition assays were
performed (22,23,31). In these studies, the amount of MSH2
bound to *(CTG)30·(CAG)30 S-DNA, *(CTG)50·(CAG)50 S-DNA,
or *(CTG)30·(CAG)50 SI-DNA was examined, when excess
gel-purified linear non-treated DNAs, the S-DNAs or the
SI-DNAs was introduced into the binding reaction. Such a
competition reaction, where MSH2 binding to the
*(CTG)30·(CAG)50 SI-DNA is examined, is shown in Figure 4A
and B. We found that the linear (CTG)30·(CAG)30 DNA was a
poor competitor, while the S-DNA and both SI-DNAs were
effective competitors of MSH2 binding. At low concentrations
(50-fold molar excess), the (CTG)30·(CAG)50 SI-DNA was more
effective than the (CTG)50·(CAG)30 SI-DNA as competitor (Fig.
4B–D). The efficiency of each competitor for MSH2 binding to
*(CTG)30·(CAG)30 S-DNA (Fig. 4C) or *(CTG)50·(CAG)50
S-DNA (Fig. 4D) was similar to that observed for binding to
*(CTG)30·(CAG)50 SI-DNA (Fig. 4B). These results suggest that
the binding of MSH2 to S-DNA and SI-DNAs is specific and is
similar to the binding of hMSH2 to single base pair and IDL
mismatched nucleotides.
1120 Human Molecular Genetics, 1997, Vol. 6, No. 7
*(CTG)30·(CAG)50 SI-DNA (Fig. 3B), which has an excess of 20
CAG repeats, and suggests that there is an intrinsic difference in
the recognition of CAG trinucleotide repeat sequences over CTG
trinucleotide repeat sequences. These results also suggest that on
either of the homoduplex S-DNAs it is the looped-out CAG, and
not the looped-out CTG strand that is being bound preferentially
by MSH2.
DISCUSSION
Figure 3. Binding of MSH2 protein to trinucleotide repeat structures.
(A) Band-shift assays were performed as described in Materials and Methods.
Gel-purified 32P-labeled trinucleotide repeat DNAs in the linear non-treated
(LNT), major S-DNA (S) and major SI-DNA (SI) forms, as indicated. The
arrows indicate the band-shifted slipped DNA–hMSH2 complexes, while open
arrowheads indicate the linear DNA–MSH2 complexes. (B) Binding titration
of MSH2 to *(CTG)30·(CAG)50 SI-DNA or *(CTG)50·(CAG)30 SI-DNA, as
indicated, using 0, 485 ng (4.6 pmol), 727.5 ng (6.7 pmol) and 970 ng (9 pmol)
of MSH2. (C) Quantification of 3–5 binding assays as shown in (A) and (B),
error bars represent plus and minus one standard deviation.
To address further the possibility that there is an intrinsic
difference in the ability of MSH2 to recognize CTG and CAG
intrastranded structures, we tested MSH2 binding to synthetic
(CTG)15 and (CAG)15 oligonucleotides (Fig. 5). We found that
MSH2 bound to the (CAG)15 oligonucleotide 4- to 5-fold more
efficiently than to the (CTG)15 oligonucleotide. This result is
similar to the observed preferential binding of the
The yeast and human MSH2 proteins are required for the
recognition of single base pair mismatched nucleotides, duplex
oligonucleotides containing short IDLs, palindromic DNA IDL
structures and Holliday junctions (21–23,35). Both purified proteins
demonstrated a binding preference for the IDL substrates, and
binding appeared to increase with increased loop size. Here, we
observed significant and specific MSH2 binding to both slippedstrand S-DNAs and SI-DNAs and minimal binding to linear duplex
DNAs. There appears to be a binding hierarchy with (CTG)50·
(CAG)50 S-DNA>(CTG)30·(CAG)30 S-DNA ∼(CTG)30·(CAG)50
SI-DNA>(CTG)50·(CAG)30 SI-DNA>>(CTG)50·(CAG)50 linear =
(CTG)30·(CAG)30 linear. In addition, we found a direct association
between MSH2 binding and the number of trinucleotide repeats.
This relationship parallels the effect of repeat length on instability in
vivo. The KD of hMSH2 binding to these trinucleotide repeat
structures was calculated to be in the high nanomolar range
(50–200 nM), which is similar to the mispair binding affinity
exhibited by the bacterial MutS protein and our previously reported
MSH2 mispair binding affinity (36).
The structure(s) within the S-DNA and SI-DNA substrates that
are recognized by MSH2 are unknown. However, these structures
are likely to be somewhat different from the IDL substrates that
previously were shown to be recognized efficiently by the human
MSH2 protein (22,23). This is because the slipped-strand DNA
structures formed within both S-DNA and SI-DNA substrates are
capable of branch migration, which can contribute to the
production of a wide variety of structural isomers. Additionally,
the looped-out region(s) formed by SI-DNAs are significantly
larger than that used in our previous experiments (21–23).
Several studies have shown that the efficiency and specificity of
substrate recognition and correction varies with the mismatch,
loop, DNA structure (hairpin, palindrome or chemical modification) and the sequence context surrounding the mismatch
(17–23,26,35–38). These results suggest that MSH2 may recognize a variety of mispair and DNA structures with wide-ranging
affinity. It is interesting to note that all of the slipped-strand
structures tested here gave rise to a single similarly shifted
complex. There are a number of possible DNA structures that
might be recognized by MSH2 within these slipped-strand
substrates and include: (i) the junction of the interstrand region
with the looped-out strand; (ii) a CAG or CTG hairpin/loop; (iii)
an A·A, T·T, C·C, C·T, A·C, G·G A·G or G·T mismatch(es)
contained in perfectly or imperfectly annealed CTG or CAG
hairpins; or (iv) the tip of an intrastrand hairpin. The binding to
the (CAG)15 and (CTG)15 oligos appears to diminish a loop
junction or three-way junction as the only recognition site since
these oligonucleotides are not large enough to form such
structures.
Our studies show a clear MSH2 binding preference for the
CAG strand. Presumably some, as yet undefined, difference in the
structure of the two looped-out CTG and CAG strands accounts
1121
Human
Genetics,
1997,
6, No.
NucleicMolecular
Acids Research,
1994,
Vol. Vol.
22, No.
1 7
1121
Figure 4. Structure-specific binding by human MSH2 protein. DNA competition assays were performed (23, 24) using gel-purified linear non-treated DNAs (LNT),
S-DNAs (S) and SI-DNAs (SI) as competitors. Binding was done as in Figure 3 using 0 and 727.5 ng (6.7 pmol) of MSH2. After binding, the quantity of competitor
indicated was added and the mixture was incubated for an additional 15 min. Reaction products were separated by gel electrophoresis, the gels were dried and
autoradiographed. (A) Autoradiograph of *(CTG)30·(CAG)50 SI-DNA–MSH2 complex competition assay with 100 and 200 molar-fold amount of competitor; no
protein (NP), no competitor (NC). The arrow indicates the band-shifted MSH2–DNA complex. (B) Quantification of *(CTG)30·(CAG)50 SI-DNA–MSH2 complex
competition assay, with 50, 100 and 200 molar-fold amount of competitor. (C) Quantification of *(CTG)30·(CAG)30 S-DNA–hMSH2 complex competition assay.
(D) Quantification of *(CTG)50·(CAG)50 S-DNA–MSH2 complex competition assay. Quantification of 3–5 competition assays as shown; error bars represent plus
and minus one standard deviation.
Figure 5. Human MSH2 protein binding to (CAG)15 and (CTG)15. (A) Individual
oligonucleotides were twice purified by TLC (SurePure, USB) and radiolabeled
to identical specific activity. Binding assays were performed as in Figure 3.
DNAs were incubated with 0, 485 ng (4.6 pmol) and 727.5 ng (6.7 pmol) of
MSH2. Products were separated on a 4% polyacrylamide gel, dried and
exposed for autoradiography. (B) Quantification of 3–5 binding assays as
shown in (A); error bars represent plus and minus one standard deviation.
for the difference in the recognition by MSH2. While both CTG
and CAG DNA strands will form hairpins, the stability of the
CTG hairpin is greater than that of the CAG hairpin (39–41). We
have shown previously that the CAG strand in the S-DNAs is
preferentially sensitive to the single-strand-specific mung bean
nuclease (12), suggesting a greater single-stranded character of
the looped-out/hairpin CAG relative to the CTG strand. Perfect
hairpins formed by the CAG and CTG strands would contain an
A·A and T·T mispair respectively, flanked by two C·G base pairs.
In simian cells, A·A and T·T mismatches are corrected with 64
and 39% efficiency, respectively (37). Thus, if only these
mispairs in their respective looped-out stems were recognized,
the preferential binding of MSH2 to the slipped-strand intermediate with an excess of CAG correlates with the efficiency of
mispair correction in vivo.
Several studies have shown that trinucleotide repeat sequences
are unstable to a degree greater than that observed for mono- and
dinucleotide repeat tracts in tumors or tumor cell lines that display
microsatellite instability (42,43). Furthermore, the relative frequency of insertions versus deletions differed depending on the
sequence of the repeat tract (43). Interestingly, neither the DM nor
the FRAXA repeat loci displayed large repeat expansions in the
MSH2-deficient human tumor cell line Lovo, although shorter
instabilities of these repeats was observed (30). These results
suggested that an MSH2 deficiency alone does not cause the large
changes in trinucleotide repeat lengths typical of DM and
FRAXA individuals. However, it is possible for the post-replication mismatch repair process to be unrelated to large trinucleotide expansions, while the mismatch repair genes (proteins) may
play a fundamental role. For example, the mismatch repair genes
have been found to participate in several other DNA metabolic
processes, such as transcription-coupled repair (TCR) (44,45)
and the elimination of genetic recombination between the
divergent DNA sequences (termed: homoeologous recombination)
1122 Human Molecular Genetics, 1997, Vol. 6, No. 7
(46,47). These latter studies have suggested that the mismatch
repair proteins may be part of a system that alters or aborts
recombination events initiated between divergent repeated
sequences, thereby maintaining the genetic distinction of these
sequences as well as the global integrity of chromosome
containing such repeats. Furthermore, genetic studies suggest that
Saccharomyces cerevisiae MSH2-dependent mismatch repair
controls the length of a mitotic genetic recombination tract
(48,49). Thus, the observed lack of large trinucleotide expansion
in mismatch repair-deficient human tumor cell lines may not
address directly the possibility of tissue specificity, altered
expression and or mutation of MSH2 during development, the
contribution(s) of transcription/TCR, or cells undergoing high
rates of recombination as a causes of trinucleotide expansion.
Furthermore, while HNPCC patients do not appear to be
predisposed to the neurodegenerative diseases associated with
expansion of trinucleotide repeat sequences, it should be noted
that they are heterozygous carriers, while only homozygous or
hemizygous mutation(s) of the mismatch repair genes result in a
mismatch repair deficiency.
Mismatch repair in human cells appears to be far more
complicated than in bacteria. It is now clear that MSH2 forms at
least two heteromeric protein complexes that enhance the
specificity of MSH2-dependent mispair recognition (28,29,50).
These two complexes have overlapping and redundant mispair
binding activities (28) that appear to account for the lack of
observed mutations in MSH6 and, by analogy MSH3, in HNPCC.
Furthermore, while the binding of MSH2 alone appears to be
somewhat lower than MSH2 complexed with either MSH3 or
MSH6, its fundamental mispair binding activity appears to
completely recapitulate the types of binding observed when it is
complexed with either MSH3 or MSH6. The enhancement of
mispair binding observed for the MSH2–MSH6 and
MSH2–MSH3 protein complexes over MSH2 alone has suggested that MSH3 and MSH6 are ‘specificity factors’ that
increase the stability of the MSH2 mispair binding activity (28).
These observations suggest that MSH2 may provide a model
system for mispair binding in the absence of clear knowledge of
the appropriate protein specificity partner and that the association
of MSH3 and/or MSH6 with MSH2 may broaden the number of
mismatch repair proteins (genes) that might contribute to
trinucleotide repeat expansion.
Our results suggest that slipped intermediates having an excess
of CTG repeats may escape detection more frequently than those
with CAG repeats. This asymmetry may indicate that it is the
CTG hairpin, rather than the CAG hairpin, which determines
trinucleotide instability at neurodegenerative disease loci. These
results appear to parallel the observation that certain palindromic
sequences can escape repair in yeast and mammalian cells
(51,52). An alternative possibility is that MSH2 either alone or in
a complex form may stabilize or protect slipped-strand CAG
loops from repair, thus allowing replication to finalize an
expansion. It is also possible that MSH2 binding to the
slipped-strand CAG loops exposes the looped-out CTG strand to
resolvases. Such asymmetric recognition may account for
strand-specific differences in the replication fidelity (8,53),
inequality in mutation rates between complementary DNA
strands (54,55) and perhaps the polar mutability of trinucleotide
repeat tracts observed in vivo (5–8).
Finally, our observation that MSH2 binding increases with
increasing length of the slipped-strand structure suggests that the
role of MSH2 in trinucleotide expansion may become more
important with longer repeat tracts. Whether MSH2 or other
mismatch repair factors are intimately involved in the biology of
trinucleotide expansion clearly requires further study.
MATERIALS AND METHODS
Plasmids and DNA treatments
The DM clones have been described previously (12) (see also Fig.
2). Isolation, radiolabeling and gel purification were as described
(12,31,32). Denaturation and renaturation for reduplexing and
heteroduplexing DNAs was by alkali treatment (pH 13) and
neutralization (pH 8) with T.E. at 68C, as described in detail
(12).
hMSH2 expression and purification
MSH2 protein was expressed in a baculovirus system and
purified by a modification of the procedure described by Fishel
et al. (28).
Band-shift assay
Band-shift assays were performed (23,24) using 1.9 pmol/reaction and 2.4 pmol/reaction of gel-purified 32P-labeled 30×30 or
50×50 DNAs, or 1.4 pmol/reaction of oligonucleotide. DNAs
were incubated with 0–970 ng (0–9 pmol) of purified MSH2 (97
ng/µl) in binding buffer [20 mM potassium phosphate (pH 7.5),
50 mM KCl, 0.1 mM dithiothreitol (DTT), 1 mM EDTA, and
23% glycerol], plus 150 pmol of double-stranded poly(dI–dC)
(Pharmacia) in 20 µl reactions, and incubated at 25C for 15 min.
Reaction products were separated on a 4% polyacrylamide gel,
dried and exposed for autoradiography as described (23,24). All
quantification was by PhosphorImage analysis using ImageQuant software (Molecular Dynamics) and represent 3–5 experiments.
ACKNOWLEDGEMENTS
We thank Tetsuo Ashizawa, Fred Gimble, David L. Nelson,
Vladimir Potaman, Stephen T. Warren and Samuel Wilson for
critical comments on the manuscript. This work was funded in
part by funds from the Institute of Biosciences and Technology
and from NIH grant GM52982 (R.R.S.) and CA67007 (R.F.).
REFERENCES
1. Ashley, C.T.,Jr and Warren, S.T. (1995) Trinucleotide repeat expansion and
human disease. Annu. Rev. Genet., 29, 703–728.
2. Mahadevan, M., Tsilfidis, C., Sabourin, L., Shutler, G., Amemiya, C., Jansen,
G., Neville, C., Narang, M., Barceló, J., O’Hoy, K., LeBlond, S., Earle-MacDonald, J., DeJong, P.J., Wieringa, B. and Korneluk, R.G. (1992) Myotonic
dystrophy mutation: an unstable CTG repeat in the 3′ untranslated region of
the gene. Science, 255, 1253–1255.
3. Richards, R.I. and Sutherland, G.R. (1992) Heritable unstable DNA
sequences. Nature Genet., 1, 7–9.
4. Chung, M.-y., Ranum, L.P.W., Duvick, L.A., Servadio, A., Zoghbi, H.Y. and
Orr, H.T. (1993) Evidence for a mechanism predisposing to intergenerational
CAG repeat instability in spinocerebellar ataxia type I. Nature Genet., 5,
254–258.
1123
Human
Genetics,
1997,
6, No.
NucleicMolecular
Acids Research,
1994,
Vol. Vol.
22, No.
1 7
5. Eichler, E.E., Holden, J.J.A., Popovich, B.W., Reiss, A.L., Snow, K.,
Thibodeau, S.N., Richards, C.S., Ward, P.A. and Nelson, D.L. (1994) Length
of uninterrupted CGG repeats determines instability in the FMR1 gene.
Nature Genet., 8, 88–94.
6. Kunst, C.B. and Warren, S.T. (1994) Cryptic and polar variation in the fragile
X repeat could result in predisposing normal alleles. Cell, 77, 853–861.
7. Snow, K., Tester, D.J., Kruckeberg, K.E., Schaid, D.J. and Thibodeau, S.N.
(1994) Sequence analysis of the fragile X trinucleotide repeat: implications
for the origin of the fragile X mutation. Hum. Mol. Genet., 9, 1543–1551.
8. Kang, S., Jaworski, A., Oshima, K. and Wells, R.D. (1995) Expansion and
deletion of CTG repeats from human disease genes are determined by the
direction of replication in E.coli. Nature Genet., 10, 213–218.
9. Jaworski, A., Rosche, W.A., Gallibolian, R., Kang, S., Shimizu, M., Bowater,
R.P., Sinden, R.R. and Wells, R.D. (1995) Mismatch repair in Escherichia
coli enhances instability of (CTG)n triplet repeats from human hereditary
diseases. Proc. Natl Acad. Sci. USA, 92, 11019–11023.
10. Sinden, R.R. and Wells, R.D. (1992) DNA structure, mutations, and human
genetic disease. Curr. Opin. Biotechnol., 3, 612–622.
11. Wells, R.D. and Sinden, R.R. (1993) In Davies, K.E. and Warren, S.T. (eds),
Genome Analysis, Genome Rearrangement and Stability. Cold Spring
Harbor Laboratory Press, Cold Spring Harbor, NY, vol. 7, pp. 107–138.
12. Pearson, C.E. and Sinden, R.R. (1996) Alternative structures in duplex DNA
formed within the trinucleotide repeats of the myotonic dystrophy and fragile
X loci. Biochemistry, 35, 5041–5053.
13. Richards, R.I. and Sutherland, G.R. (1994) Simple repeat DNA is not
replicated simply. Nature Genet., 6, 114–116.
14. Pearson, C.E. Eichler, E.E., Lorenzetti, D., Acharya, S., Kramer, P.R.,
Kramer, S.F., Nelson, D.L., Zoghbi, H.Y. and Sinden, R.R. (1996) Slipped
strand structures in neurodegenerative disease-associated trinucleotide
repeats: a role for human mismatch repair and cryptic interruptions. Am. J.
Hum. Genet., 59 (suppl.), A48, 244.
15. Fishel, R., Lescoe, M.K., Rao, M.R.S., Copeland, N.G., Jenkins, N.A.,
Garber, J., Kane, M. and Kolodner, R. (1993) The human mutator gene
homolog MSH2 and its association with hereditary nonpolyposis colon
cancer. Cell, 75, 1027–1038.
16. Leach, F.S. et al., (1993) Mutations of a mutS homolog in hereditary
nonpolyposis colorectal cancer. Cell, 75, 1215–1225.
17. Su, S.-S. and Modrich, P. (1986) Escherichia coli mutS-encoded protein binds
to mismatched DNA basepairs. Proc. Natl Acad. Sci. USA, 83, 5057–5061.
18. Su, S.-S., Lahue, R.S., Au, K.G. and Modrich, P. (1988) Mispair specificity of
methyl-directed DNA mismatch correction in vitro. J. Biol. Chem., 263,
6829–6835.
19. Jiricny, J., Su, S.-S., Wood, S.G. and Modrich, P. (1988) Mismatch-containing
oligonucleotide duplexes bound by the E.coli mutS-encoded protein. Nucleic
Acids Res., 16, 7843–7853.
20. Parker, B.O. and Marinus, M.G. (1992) Repair of DNA heteroduplexes
containing small heterologous sequences in Escherichia coli. Proc. Natl
Acad. Sci. USA, 89, 1730–1734.
21. Alani, E., Chi, N.-W. and Kolodner, R.D. (1995)The Saccharomyces cerevisiae
Msh2 protein specifically binds to duplex oligonucleotides containing mismatched DNA base pairs and insertions. Genes Dev., 9, 234–247.
22. Fishel, R., Ewel, A. and Lescoe, M.K. (1994) Purified human MSH2 protein
binds to DNA containing mismatched nucleotides. Cancer Res., 54,
5539–5542.
23. Fishel, R., Ewel, A., Lee, S., Lescoe, M.K. and Griffith, J. (1994) Binding of
mismatched microsatellite DNA sequences by the human MSH2 protein.
Science, 266, 1403–1405.
24. Bhattacharyya, N.P., Skandalis, A., Ganesh, A., Groden, J. and Meuth, M.
(1994) Mutator phenotype in human colorectal carcinoma cell lines. Proc.
Natl Acad. Sci. USA, 91, 6319–6323.
25. Parsons, R., Li, G.-M., Longely, M.J., Fang, W.-H., Papadopoulos, N., Jen, J.,
de la Chapelle, A., Kinzler, K.W., Vogelstein, B. and Modrich, P. (1993)
Hypermutability and mismatch repair deficiency in RER+ tumor cells. Cell,
75, 1227–1236
26. Umar, A., Boyer, J.C. and Kunkel, T.A. (1994) DNA loop repair by human
cell extracts. Science, 266, 814–816.
27. Fishel, R. and Kolodner, R.D. (1995) Identification of mismatch repair genes and
their role in the development of cancer. Curr. Opin. Genet. Dev., 5, 382–395.
28. Acharya, S., Wilson, T., Gradia, S., Kane, M.F., Guerrette, S., Marsischky,
G.T., Kolodner, R. and Fishel, R. (1996) hMSH2 forms specific mispair-binding complexes with hMSH3 and hMSH6. Proc. Natl Acad. Sci. USA, 93,
13629–13634.
29. Drummond, J.T., Li, G.-M., Longley, M.J. and Modrich, P.(1995) Isolation of
an hMSH2–p160 heterodimer that restores DNA mismatch repair to tumor
cells. Science, 268, 1909–1912.
1123
30. Kramer, P.R., Pearson, C.E. and Sinden, R.R. (1996) Stability of triplet
repeats of myotonic dystrophy and fragile X loci in human mutator mismatch
repair cell lines. Hum. Genet., 98, 151–157.
31. Pearson, C.E., Ruiz, M.T., Price, G.B. and Zannis-Hadjopoulos, M. (1994)
Cruciform DNA binding protein in HeLa cell extracts. Biochemistry, 33,
14185–14196.
32. Pearson, C.E., Zannis-Hadjopoulos, M., Price, G.B. and Zorbas, H. (1994) A
novel type of interaction between a cruciform DNA structure and a cruciform
binding protein from HeLa cells. EMBO J., 14, 1571–1580.
33. Gaillard, C. and Strauss, F. (1994) Association of poly(CA)·(poly(TG) DNA
fragments into four-stranded complexes bound by HMG1 and 2. Science,
264, 433–436.
34. Belotserkovskii, B.P. and Johnston, B.H. (1996) Polypropylene tube surfaces
may induce denaturation and multimerization of DNA. Science, 271, 222–223.
35. Alani, E., Lee, S., Kane, M.F., Griffith, J. and Kolodner, R. (1997)
Saccharomyces cerevisiae MSH2, a mispair recognition protein, also
recognizes Holliday junctions. J. Mol. Biol., 265, 289–301
36. Mello, J.A., Acharya, S. Fishel, R. and Essigman, J.M. (1996) The mismatch
repair protein hMSH2 binds selectively to DNA adducts of the anticancer
drug cisplatin. Chem. Biol., 3, 579–589.
37. Brown, T.C. and Jiricny, J. (1988) Different base/base mispairs are corrected
with different efficiencies and specificities in monkey kidney cells. Cell, 54,
705–711.
38. Jones, M., Wagner, R. and Radman, M. (1987) Repair of a mismatch is
influenced by the base composition of the surrounding nucleotide sequence.
Genetics, 115, 605–610.
39. Gacy, A.M., Goellner, G., Juranic, N., Macura, S. and McMurray, C.T. (1995)
Trinucleotide repeats that expand in human disease form hairpin structures in
vitro. Cell, 81, 533–540.
40. Mitas, M., Yu, A., Dill, J., Kamp, T.J., Chambers, E.J. and Haworth, I.S.
(1995) Hairpin properties of single-stranded DNA containing a GC-rich
triplet repeat: (CTG)15. Nucleic Acids Res., 23, 1050–1059.
41. Smith, G.K., Jie, J., Fox, G.E. and Gao, X. (1995) DNA CTG triplet repeats
involved in dynamic mutations of neurobiologically related gene sequences
form stable duplexes. Nucleic Acids Res., 23, 4303–4311.
42. Wooster, R., Cleton-Jansen, A.-M., Collins, N., Mangion, J., Cornelis, R.S.,
Cooper, C.S., Gusterson, B.A., Ponder, B.A.J., von Demling, A., Wiestler,
O.D., Cornelisse, C.J., Devilee, P. and Stratton, M.R. (1994) Instability of short
tandem repeats (microsatellites) in human cancers. Nature Genet., 6, 152–156.
43. Shibata, D., Peinado, M.A., Ionov, Y., Malkhosyan, S. and Perucho, M. (1994)
Genomic instability in repeated sequences is an early somatic event in colorectal
tumorigenesis that persists after transformation. Nature Genet., 6, 273–281.
44. Mellon, I. and Champe, G.N. (1995) Products of DNA mismatch repair genes
mutS and mutL are required for transcription-coupled nucleotide excision
repair of the lactose operon in Escherichia coli. Proc. Natl Acad. Sci. USA, 93,
1292–1297.
45. Mellon, I., Deepak, K.R., Koi, M.C., Boland, R.C. and Champe, G.N. (1996)
Transcription-coupled repair deficiency and mutations in human mismatch
repair genes. Science, 272, 557–560.
46. Rayssiguier, C., Thaler, D.S. and Radman, M. (1989) The barrier to
recombination between Escherichia coli and Salmonella typhimurium is
disrupted in mismatch-repair mutants. Nature, 342, 396–401.
47. Bailis, A.M. and Rothstein, R. (1990) A defect in mismatch repair in
Saccharomyces cerevisiae stimulates ectopic recombination between homeologous genes by an excision repair dependent process. Genetics, 126, 535–547.
48. Alani, E., Reenan, R.A.G. and Kolodner, R. (1994) Interaction between
mismatch repair and genetic recombination in Saccharomyces cerevisiae.
Genetics, 137, 19–39.
49. Datta, A., Adjiri, A., New, L., Crouse, G.F. and Jinks-Robertson, S. (1996)
Meiotic crossovers between divergent sequences are regulated by mismatch
repair proteins in Saccharomyces cerevisiae. Mol. Cell. Biol., 16, 1085–1093.
50. Palombo, F., Gallinari, P., Iaccarino, I., Lettieri, T., Hughes, M., D’Arrigo, A.,
Truong, O., Hsuan, J.J. and Jiricny, J. (1995) GTBP, a 160-kilodalton protein
essential for mismatch-binding activity in human cells. Science, 268,
1912–1914.
51. Nag, D.K., White, M.A. and Petes, T.D. (1989) Palindromic sequences in
heteroduplex DNA inhibit mismatch repair in yeast. Nature, 340, 318–320.
52. Bollag, R.J., Elwood, D.R., Tobin, E.D., Godwin, A.R. and Liskay, R.M.
(1992) Formation of heteroduplex DNA during mammalian intrachromasomal gene conversion. Mol. Cell. Biol., 12, 1546–1552.
53. Bohr, V.A. (1991) Gene specific DNA repair. Carcinogenesis, 12, 1983–1992.
54. Trinh, T.Q. and Sinden, R.R. (1991) Preferential DNA secondary structure
mutagenesis in the lagging strand of replication in E.coli. Nature, 352, 544–547.
55. Wu, C.-I. and Maeda, N. (1987) Inequality in mutation rates of the two strands
of DNA. Nature, 327, 169–170.