The MLH1 variants p.Arg265Cys and p.Lys618Ala affect protein stability while p.Leu749Gln affects heterodimer formation.

Abstract

Germline mutations in the mismatch repair genes lead to the Lynch syndrome/ HNPCC, and genetic testing is offered to identify individuals at risk of developing this disease. About a third of all genetic alterations detected in the mismatch repair gene MutL homolog 1 (MLH1) are missense variants. The majority of these variants are of equivocal clinical significance. In this report, we investigate the molecular basis of the pathogenicity of three missense variants localized to distinct functional domains. Our results demonstrate that the MLH1 variants p.Arg265Cys (c.793C>T) and p.Lys618Ala (c.1852_1853AA>GC) significantly decrease the stability of the protein, while another missense variant p.Leu749Gln (c.2246T>A) is able to compromise heterodimerization of the MLH1-PMS2 complex.

Full text

Page 1

HUMAN MUTATION Mutation in Brief #____ (19XX) Online
MUTATION IN BRIEF
© 2007 WILEY-LISS, INC.
Received <date>; accepted <date>.
The MLH1 variants, Arg265Cys and Lys618Ala affect
protein stability, while Leu749Gln affects
heterodimer formation
Sheron Perera and Bharati Bapat
Department of Laboratory Medicine and Pathobiology, University of Toronto, Toronto, Canada; Samuel Lunenfeld
Research Institute, and Department of Pathology and Laboratory Medicine, Mt. Sinai Hospital, Toronto, Canada.

*Correspondence to Bharati Bapat, PhD, Samuel Lunenfeld Research Institute, The Joseph and Wolf Lebovic
Centre, Mt. Sinai Hospital, 60 Murray Street, Box 30, Toronto, Ontario M5T 3L9, Canada. E-mail:
bapat@mshri.on.ca

Contract grant sponsor: N/A; Contract grant number: N/A
Short Title: MLH1 variants affect stability and heterodimerization
Communicated by <Please don’t enter>
Germline mutations in the mismatch repair genes lead to the Lynch syndrome/ HNPCC, and
genetic testing is offered to identify individuals at risk of developing this disease. About a
third of all genetic alterations detected in the mismatch repair gene MutL homolog 1 (MLH1)
are missense variants. The majority of these variants are of equivocal clinical significance. In
this report, we investigate the molecular basis of the pathogenicity of three missense variants
localized to distinct functional domains. Our results demonstrate that the MLH1 variants
p.Arg265Cys (c.793C>T) and p.Lys618Ala (c.1852_1853AA>GC) significantly decrease the
stability of the protein, while another missense variant p.Leu749Gln (c.2246T>A) is able to
compromise heterodimerization of the MLH1-PMS2 complex. © 2007 Wiley-Liss, Inc.
KEY WORDS: MLH1, MMR, HNPCC, Lynch Syndrome, MutLα, Missense Variants, Protein Stability
INTRODUCTION
Lynch syndrome or hereditary nonpolyposis colorectal cancer (HNPCC [MIM# 120435]), is a common cancer
predisposition syndrome that is inherited in an autosomal dominant manner. HNPCC is associated with germline
mutations in the mismatch repair (MMR) genes (Reviewed in Lynch et al., 2006). Individuals in HNPCC families
are at an increased risk of developing colorectal cancer, while women in these kindreds are predisposed to
endometrial cancer. In addition, malignancies of the stomach, ovaries, urinary tract, small intestine, hepatobilliary
tract and brain frequently occur as part of the HNPCC tumor spectrum. In the clinical setting, diagnosis of HNPCC
is complicated by the lack of precise characteristic features. While clinical guidelines including Amsterdam
Criteria 1 and 2, and Bethesda Criteria have proven to be useful in diagnosis, these criteria are limited in their
ability to accurately identify patients with HNPCC. Thus, the diagnosis of HNPCC relies primarily on the
detection of germline defects in the MMR genes of these patients (Casey et al., 2005).
The MMR pathway plays a central role in maintaining genomic stability by repairing both mismatches and

Page 2

2 <FIRST AUTHOR NAME>
insertion/deletion loops that arise during DNA synthesis (Reviewed in Modrich, 2006). MMR improves the fidelity
of DNA biosynthesis several hundred-fold and reduces the error rate to about one error per 1010 bases (Modrich
and Lahue, 1996). Defective MMR results in the mutator phenotype, which increases the likelihood of mutations in
a battery of genes that regulate growth, differentiation or apoptosis. Deficiencies in MMR lead to microsatellite
instability (MSI). Tumors from the majority of HNPCC patients (>85%) display MSI, making it a hallmark feature
of this syndrome (Aaltonen et al., 1993; Aaltonen et al., 1994). In addition to the error repair function, MMR
proteins act on DNA damage caused by several factors. Recognition and perhaps processing of these lesions by
MMR proteins leads to activation of DNA damage signaling pathways, which can in turn result in cell-cycle arrest
or apoptosis. Nevertheless, the mechanism by which this occurs is not fully understood. In addition to inherited
cancers, MMR deficiency also gives rise to ~15% of sporadic colorectal, endometrial and gastric cancers.
Advances in the understanding of the genetic basis of HNPCC has facilitated predictive genetic testing programs
to identify individuals with an inherited MMR mutation who are at an increased risk of developing colorectal and
other HNPCC-associated cancers. By predicting risk prior to the progression of disease, these programs play an
integral role in the effort to improve surveillance, treatment and prevention of cancer. A recent database
catalogues over 1500 MMR gene variants [http://www.med.mun.ca/MMRvariants (Woods et al., 2007)]; while the
International Society for Gastrointestinal Hereditary Tumors (InSiGHT, http://www.insight-group.org/) lists over
400 pathogenic mutations that have been identified in the MMR genes. Among these, mutations in Mut L homolog
1 (MLH1; MIM# 120436) and Mut S homolog 2 (MSH2; MIM# 609309) are responsible for over 90% of known
cases of HNPCC (Chung and Rustgi, 2003). About 25% of MLH1 alterations are missense alterations of equivocal
clinical significance (Jacob and Praz, 2002). Detection of such an unclassified variant affects the accuracy of
predictive genetic testing and counseling of patients carrying these variants, as well as their family members at risk
of developing HNPCC. It is likely that missense alterations of MLH1 affect its expression and function(s) to
varying degrees and thereby contribute to the variable disease phenotype observed in HNPCC.
In this study we investigate the effects of three representative missense variants in MLH1 on its expression and
function, in order to clarify their pathogenic significance. We functionally characterized the MLH1 alterations
c.793C>T leading to the substitution p.Arg265Cys, located within the ribosomal S5 domain 2-type fold and
c.1852_1853AA>GC leading to p.Lys618Ala, which is located in the MutL homolog binding domain. The
pathogenicity of these variants remain debatable even after several analyses due to the existence of conflicting data
(Blasi et al., 2006; Casey et al., 2005; Ellison et al., 2001; Fearnhead et al., 2004; Guerrette et al., 1999; Kondo et
al., 2003; Mohd et al., 2006; Trojan et al., 2002). We also hypothesized that missense alterations that occur in the
C-terminal region, adjacent to the Mut L homolog binding domain likely affects MLH1 heterodimerization, and to
investigate this we constructed an alteration c.2246T>A which leads to the missense substitution p.Leu749Gln and
is located within the c-terminal homology domain.
MATERIALS AND METHODS
Computational methods
We used the computational programs Polymorphism Phenotyping (PolyPhen)
(http://genetics.bwh.harvard.edu/pph/), and Sorting Intolerant From Tolerant (SIFT)
(http://blocks.fhcrc.org/sift/SIFT.html) to predict the pathogenicity of these variants. PolyPhen uses structural
information, along with multiple sequence alignments to classify the pathogenicity an amino acid substitution with
an accuracy of about 80% (Ramensky et al., 2002). Based on these criteria it predicts a substitution to be benign,
possibly pathogenic or probably pathogenic respectively, in ascending order of its predicted impact. SIFT uses an
algorithm that aligns closely related sequences, and calculates normalized probabilities for all possible
substitutions (Ng and Henikoff, 2001). It has been reported that SIFT is able to predict the impact of non-
synonymous substitutions with an accuracy ranging from 65-92% in a range of genes associated with human
disease (Chan et al., 2007). It predicts positions with normalized probabilities of less that .05 as deleterious, while
those with probabilities greater than or equal to .05 are classified as tolerated. Default settings were used for both
programs.

Page 3

<RUNNING TITLE> 3
Site-directed mutagenesis and construction of MLH1 mutants
The pcDNA3.1+ vector containing the full-length wild-type MLH1 cDNA (GenBank accession no.
NM_000249.2) was kindly provided to us by Dr. G. Plotz and Dr. R. Kolodner. Of the variants chosen for this
study, R265C and K618A both occur naturally, while L749Q was artificially generated by us. We used site-
directed mutagenesis to create the expression constructs according to the specifications of the manufacturer (Quick
Change Site-directed Mutagenesis, Stratagene, LaJolla, CA). MLH1 cDNA coding for the variants were generated
using primer pairs that contained the specific nucleotide alteration. To confirm the integrity and the presence of the
respective missense substitutions, all cDNA inserts were directly sequenced. We sub-cloned these constructs into
myc-His tagged expression vectors (Invitrogen, Carlsbad, CA), to use them for a wider range of experimental
procedures. All variant numbering is based on cDNA sequence, with codon 1 corresponding to the major initiation
codon Primer sequences and PCR conditions for all protocols listed above are available on request.
Cell culture, production of recombinant proteins and western blotting
HCT116 cells were cultured in McCoy’s 5A media supplemented with 10% fetal bovine serum and
maintained at 37°c in a humified atmosphere with 5% CO2. HCT116 cells were transfected with the appropriate
vectors using the Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer’s instructions, and
harvested at the indicated time points. For total protein extraction, cells were washed twice with cold phosphate
buffered saline (PBS). Cold radioimmunoprecipitation assay (RIPA) lysis buffer (0.5 % Nonidet P-40, 0.5%
sodium deoxycholate, 0.1% sodium dodecyl sulfate, 1x PBS) buffer containing Complete protease inhibitor tablets
(Roche, Mannheim, Germany) was used to lyse the cells on ice for approximately 20 minutes. The lysates were
then collected by centrifugation at 14,000 rpm for approximately 15 minutes. The expression levels for the proteins
were examined by sodium dodecyl sulphate/ polyacrylamide gel electrophoresis, followed by western blot analysis
with anti-MLH1 (clone 168-728; BD Biosciences) and anti-postmeiotic segregation increased 2 (PMS2; MIM#
600259: clone A16-4; BD Biosciences) antibodies. β-Actin antibody (Sigma) was used to assess equal loading in
all lanes. To validate our results, we co-transfected cells with 0.15μg of pCMV β-galactosidase vector, and
measured its enzymatic activity as described previously (Esufali and Bapat, 2004).
Combined coimmunoprecipitation and western blot analysis
For coimmunoprecipitation 1000 μg of whole cell lysate transfected with each of the constructs were incubated
with 2.5 μg of anti-MLH1 antibody. The lysates were then incubated for a further 1.5 hours after the addition of
40 μl of protein A/G agarose beads (Santa Cruz Biotechnology, Santa Cruz, CA). The precipitates were then
centrifuged at 1000 rpm for 2 minutes, and washed 3 to 4 times with 500μl of RIPA buffer and resolved on a 10%
SDS-PAGE gel. Interactions between MLH1 and PMS2 proteins were detected by western blot analysis with anti-
MLH1 and anti-PMS2 antibodies.
Pulse chase assays
HCT116 cells were plated in 6 cm dishes, and then transiently transfected with wild-type or variant MLH1
cDNA. 24 hours post transfection, the media was aspirated from the plates and the cells were washed twice in
warm PBS. 1 ml Methionine (Met) -, and Cysteine (Cys)-free DMEM (Sigma - supplemented with Glutamine)
was added to the cells, and incubated for 30 minutes at 37°c to starve the cells of endogenous pools of Met and
Cys. Thereafter 150μCi of EasyTag Express 35S (Perkin-Elmer, Waltham, MA) was added to each plate, and the
cells were incubated for an additional 40 minutes at 37°c. The radioactive media was then removed and the cells
were washed twice with warm PBS. For the chase, 5 ml of warm DMEM supplemented with 2mM Met and 2mM
Cys was added to all plates except the ‘0’ time point. The plates were then incubated at 37°c and harvested at the
specified time points from 0-30h. The harvested lysates were pre-cleared for one hour with protein A/G agarose
beads. Thereafter, first total MLH1, and then total β-Actin was immunoprecipitated, and analyzed by SDS-PAGE
followed by autoradiography.
Cycloheximide experiments
HCT116 cells were transfected with either wild-type or variant MLH1 constructs. 24 hours post-transfection,
cells were treated with 25μg/ml of cycloheximide (Sigma) or vehicle (DMSO), and lysed at indicated time points
as described before. The levels of MLH1 were assessed by western blotting.

Page 4

4 <FIRST AUTHOR NAME>
RESULTS
In silico analysis of MLH1 variants
For this study we chose three variants located in distinct domains of MLH1. Nucleotide alterations that lead to
different amino acids substitutions at codon 265 have been reported in the literature including the polymorphism
p.R265H (Trojan et al., 2002) and the unclassified variant, p.R265C. Codon 618 incurs a point mutation leading to
a ‘bona fide’ mutation p.K618T (Brieger et al., 2002), and a two base-pair substitution leading to the variant
p.K618A. Similarly, codon 749 is shown to result in a substitution leading to p.L749P (Colombino et al., 2003;
Losi et al., 2005). Thus, nucleotide alterations at all three codons have been detected in HNPCC probands.
We first employed the computational tools, SIFT and PolyPhen, to predict the pathogenicity of the missense
substitutions chosen for this study (Table 1). SIFT has been used to predict the pathogenicity of the variants R265C
and K618A in previous reports (Raevaara et al., 2005; Takahashi et al., 2007); PolyPhen however is yet to be
applied to the analysis of these variants. All three of the missense variants chosen were predicted to be pathogenic
by either of these two programs. However, we observed that these computational databases did not always provide
concordant predictions. For example, while SIFT predicts K618A as a “tolerated” alteration, PolyPhen predicts it
to be possibly damaging.
To gain a better understanding of how these substitutions affect the structure and consequently the function of
the protein, we mapped it onto the crystal structure of the MLH1 homolog in Escherichia coli, MutL (PDB
accession no.: 1B63, Ban and Yang, 1998). A high degree of sequence conservation is seen only within the N-
terminal domain of E. coli MutL (Ban and Yang, 1998; Guarne et al., 2004). The C-terminal region of E. coli
MutL shares little sequence similarity to other MutL homologs except human Mut L homolog 3 (MLH3) and
PMS2 (Guarne et al., 2004). In addition, these authors were unable to detect sequence conservation between E.
coli MutL and human MLH1 (Guarne et al., 2004). Therefore, we decided to use this approach to assess the R265C
variant that occurs in the N-terminal region of MLH1. The amino acid R265 corresponds to R261 in E.coli, and it
is located between an alpha helix and a beta sheet (Fig.1B, C). R261 interacts with the carbonyl groups of Ser91
and Leu92; in addition it also forms an internal water-mediated salt bridge with Glu32. However, the level of this
effect may depend on the amino acid it is substituted with, and its electrochemical properties These interactions
indicate the potential role of this residue in stabilizing the folding of the N-terminal domain. R261 has also been
shown to destabilize the P-loop, consequently decreasing ATP binding (Ban et al., 1999)..


Table 1. Summary of predictions generated by computational tools for the MLH1 variants


Nucleotide
Alteration change
Domain

Amino acid

Exon

Functional

SIFT

PolyPhen

c.793C>T


c.1852_1853
AA>GC

c.2246T>A

p.Arg265Cys


p.Lys618Ala


p.Leu749Gln



10


16


19

Ribosomal S5 domain 2-
type fold*


MutL homolog
interaction

C-terminal homology


Intolerant


Tolerant


Intolerant


Probably Damaging


Possibly Damaging


Probably Damaging


Sequence variant numbering is based on MLH1 cDNA sequence (GenBank NM_000249.2). Nucleotide numbering uses the A
of the ATG translation initiation site as nucleotide +1. The location and nature of the missense substitutions, as well as the
output generated by the computational tools are listed. The classification of the output generated is as follows: SIFT, intolerant
or tolerant, and PolyPhen, probably damaging > possibly damaging> benign. * MLH1 gene card (http://www.genecards.org/cgi-
bin/carddisp.pl?gene=MLH1#domain).

Page 5

<RUNNING TITLE> 5


A.
B.
R 261
R 261


Figure 1. A: Ribbon diagram of the approximate position of R261 on the crystal structure of E. coli MutL: R265, which is
equivalent to R261 in Escherichia coli was mapped onto the crystal structure of the N-terminus of MutL (PDB#:1B63) using
sequence homology. B: A more magnified view of the amino acid R261 indicates its likely involvement in stabilizing the
folding and the structure of the protein.
R265C and K618A decrease expression of MLH1
The colorectal cancer cell line, HCT116, was chosen as a model system for this study, as it is deficient for
endogenous MLH1 due to a hemizygous nonsense mutation in codon 252 (Boyer et al., 1995). Upon transient
transfection, we confirmed that the expression of wild-type MLH1 was transiently restored from 12-96 hours post-
transfection. We also observed very low levels of PMS2 in untransfected HCT116 cells. However, transfection of
ectopic MLH1 restored endogenous levels of PMS2. This is consistent with previous studies that indicate that
PMS2 is unstable in the absence of MLH1 (Brieger et al., 2002; Li and Modrich, 1995; Mohd et al., 2006; Pang et
al., 1997).
We then transiently transfected the variant constructs into HCT116 cells and assayed for their expression up to 72
hours post-transfection. Results from both tagged and non-tagged expression vectors indicate that all variant
constructs are capable of producing variant MLH1 protein of the expected size (~85kDa and ~ 90kDa
respectively). The R265C and K618A variants however, appear to either produce less protein or produce protein
that is less stable than the wild-type MLH1 construct (Fig. 2A, B). On the other hand, the expression profile for the
variant L749Q was comparable to the wild-type protein (Fig. 2B). No significant differences in β-galactosidase
transfection efficiency were seen between the wild-type and variant constructs upon co-transfection.
Representative data for K618A is depicted in Fig. 2C. This provides additional evidence that the differences in
protein expression likely arise due to the variant substitution introduced into the MLH1 cDNA.