|Home | About | Journals | Submit | Contact Us | Français|
This work is licensed under the Creative Commons Attribution-NonCommercial-Share Alike 3.0 License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-sa/3.0/.
Germline mutations in the mismatch repair (MMR) genes are associated with Lynch syndrome, also known as hereditary non-polyposis colorectal cancer (HNPCC) syndrome. Here, we characterise a variant of hMLH1 that confers a loss-of-function MMR phenotype. The mutation changes the highly conserved Gly67 residue to a glutamate (G67E) and is reminiscent of the hMLH1-p.Gly67Arg mutation, which is present in several Lynch syndrome cohorts. hMLH1-Gly67Arg has previously been shown to confer loss-of-function (Shimodaira et al, 1998), and two functional assays suggest that the hMLH1-Gly67Glu protein fails to sustain normal MMR functions. In the first assay, hMLH1-Gly67Glu abolishes the protein's ability to interfere with MMR in yeast. In the second assay, mutation of the analogous residue in yMLH1 (yMLH1-Gly64Glu) causes a loss-of-function mutator phenotype similar to yMLH1-Gly64Arg. Despite these molecular similarities, an unusual spectrum of tumours is associated with hMLH1-Gly67Glu, which is not typical of those associated with Lynch syndrome and differs from those found in families carrying the hMLH1-Gly67Arg allele. This suggests that hMLH1 may have different functions in certain tissues and/or that additional factors may modify the influence of hMLH1 mutations in causing Lynch syndrome.
Lynch syndrome, also known as hereditary non-polyposis colorectal cancer (HNPCC), is an autosomal dominant cancer predisposition disorder. Apart from colorectal cancers, affected individuals are also predisposed to an array or other cancers, such as endometrial cancer, transitional cell carcinomas of the upper urothelial tract, carcinoma of the stomach, hepatobiliary and small intestinal cancers (reviewed in Lynch and de la Chapelle, 2003). Sets of criteria that trigger investigation of the possibility of a Lynch syndrome diagnosis include the revised Bethesda and the Amsterdam I and II criteria, which consider the combinations of tumours in individuals and families, the age of tumour onset and pathological features of tumours (Vasen et al, 1999; Umar et al, 2004).
In the 1990s, a number of studies linked Lynch syndrome to mutations in the mismatch repair (MMR) genes (Fishel et al, 1993; Bronner et al, 1994; Nicolaides et al, 1994; Papadopoulos et al, 1994; Li, 2003). The MMR proteins prevent the accumulation of DNA replication errors by detecting and effecting the removal of base–base mismatches as well as small insertion-deletion loops from the newly synthesised strand. Tissue samples from tumours and human cell lines in which MMR has been inactivated show variation in the length of repetitive DNA sequences (microsatellites), a phenomenon known as microsatellite instability (MSI). These include mononucleotide repeats of adenine, which are often mutated in MMR-defective cells, and are present in DNA damage repair genes MRE11, RAD50 and BRCA1, checkpoint component CHK1 (Yoshikawa et al, 2000; Menoyo et al, 2001; Giannini et al, 2002; Kim et al, 2007) as well as TGFβ-R2, PTEN, APC and MBD4 (Markowitz et al, 1995; Huang et al, 1996; Kong et al, 1997; Tashiro et al, 1997; Riccio et al, 1999).
Components of the MMR pathway are highly conserved from bacteria to human (Modrich, 2006). In eukaryotes, heterodimers of MSH2-MSH6 (MutSα) and MSH2-MSH3 (MutSβ) detect mismatches and insertion–deletion loops and form complexes with heterodimers of MLH1-PMS2 (MutLα), and to a lesser extent with MLH1-MLH3 (MutLβ). hPMS2 contains endonuclease activity that can initiate the removal of the newly synthesised strand, which contains the incorrectly inserted base (Kadyrov et al, 2006). In vitro reconstitution of MMR shows that MutLα is required for 3′-directed excision of the mismatch, in addition to factors such as PCNA, RPA and EXOI (Dzantiev et al, 2004).
More than 450 germline abnormalities of MMR genes have been described (InSiGHT, http://www.insight-group.org). Whether a specific missense variant in hMSH2 or hMLH1 confers pathogenicity can be difficult to determine, as it requires the evaluation of multiple parameters, including statistical analyses of the segregation of the variant with the phenotype and functional assessment of the variant protein. Segregation of the variant within a family may provide some evidence, but this could in theory be caused by another closely linked mutation. Small family size, incomplete penetrance of mutations and lack of confirmatory records on tumours from older generations can often limit the power of such analyses. Collation of data on variants from different families is helpful, but is not available if variants are rare or novel.
There are several functional tests available to assess variant protein activity, although they each have their limitations. Mismatch repair-deficient tumour-derived cell lines can be transfected with plasmid containing variant MLH1 genes and assessed for mutator phenotype and/or drug-resistance phenotypes (Brieger et al, 2002; Trojan et al, 2002; Blasi et al, 2006). For C-terminal mutations in hMLH1 and hPMS2, testing the variant proteins for interactions with their partner can be informative (Nystrom-Lahti et al, 2002). There are also in vitro tests for MMR activity, which rely upon the complementation of MMR-deficient extracts by the variant protein. However, these assays require expression and purification of recombinant MMR proteins (Nystrom-Lahti et al, 2002; Heinen et al, 2002), and although MMR repair in vitro may be deemed normal when sufficient quantities of the variant protein is used, some variants are poorly expressed, suggesting that the cellular levels of the variant may be lower (Kariola et al, 2002). All of these assays require tissue culture facilities and/or recombinant protein expression, significant skill, and can be costly. To overcome these limitations, several yeast and bacteria-based systems for the functional analysis of variant MMR proteins have been developed. In both yeast and bacteria, hMLH1 interferes with normal, endogenous yMMR/bMMR, resulting in enhanced mutation rates (Shimodaira et al, 1998; Quaresima et al, 2003). Several mutations in hMLH1 abolish the capacity of the hMLH1-variant to interfere with yMMR/bMMR, thus restoring normal mutation rates (Shimodaira et al, 1998; Quaresima et al, 2003; Takahashi et al, 2007). Generally, in vitro assays concur with findings in such studies (Takahashi et al, 2007). Furthermore, there are alternative/complementary strategies for assessing hMLH1 function in yeast. The N-terminus of MLH1 contains the highly conserved ATPase domain, allowing analogous variant hMLH1 mutations to be introduced into yMLH1 (Shcherbakova and Kunkel, 1999; Hoffmann et al, 2003). As this method is restricted to evolutionarily conserved N-terminus residues, hybrid mammal–yeast MLH1 have been generated to assess MMR function of hMLH1 variants in budding yeast (Polaczek et al, 1998).
Here, we describe the functional analysis of a variant hMLH1 protein following our description of a Lynch syndrome kindred with a unique collection of tumours rarely seen in this condition. Functional analysis of hMLH1 suggests that the mutation of p.Gly67Glu perturbs the protein's ability to prevent the accumulation of mutations. This loss-of-function is similar to a previously analysed allele, hMLH1-Gly67Arg, which has been described in Lynch syndrome families. Our observations suggest that the loss-of-function of hMLH1 is not sufficient to predict the type of tumours. We suggest that other genetic modifiers may be present to explain the phenotype and/or that the function of hMLH1 may differ from one cell type to another.
Plasmids pRS315 containing hMLH1 and hMLH1-G67R were kindly provided by Dr Shimodaira. The hMLH1-G67E and yMLH1-G64R mutations were generated using the Quick Change Site-directed Mutagenesis kit from Stratagene (La Jolla, CA, USA) using primers hMLHG67E for 5′-AGGGAGGCCTGAAGTTGATTCAGATCCAAGACAATGGCA CCGAGATCAGG-3′ and hMLHG67Erev for 5′-GGGGTTTGCTCAGAGGCTGCAGAAATGCATCAAGC-3′. Plasmids were introduced into the haploid yeast strain Y39 (Y55 background, MATa, his4-R1, leu2-R1, ade1-1, lys2InsE-A14 CAN1, ura3-N). Plasmid yMLH1-G64E was generated as above using plasmid yMLH1-G64R as a template and primers yMLHG64E for 5′-CAAATAACAGATAACGGATCTGAAATTAATAAAGCAGACCTGCCA-3′ and yMLHG64Erev for 5′-TGGCAGGTCTGCTTTATTAATTTCAGATCCGTTATCTGTTATTTG-3′. The yMLH1-G64E mutation was integrated into the yeast genome by two-step gene replacement, as described previously (Hoffmann et al, 2003).
Yeast media were generated and used according to (Hoffmann et al, 2003).
Total protein was extracted from the transformants expressing the hMLH1variants. A total of 3.5 × 107 cells were disrupted by glass beads in 200μl 20% trichloroacetic acid (TCA) in a FastPrep cell disrupter (Thermo Savant, Breda, Netherlands). The glass beads were removed and the precipitated protein was resuspended in 100μl sample loading buffer (250mM Tris-HCl, pH 8.0, 5% glycerol, 0.4% SDS, 1% β-mercaptoethanol, 0.02% Bromophenol Blue). Equal amounts of protein, as determined by the Bradford assay, were separated by SDS polyacrylamide gel electrophoresis (6%). Proteins were electrophoretically transferred onto a nitrocellulose membrane (Amersham Biosciences, Amersham, UK), which was probed first with a mouse hMLH1 antibody (1:500, G168-15, BD Biosciences, Erembodegem, Belgium), and secondly with a horseradish peroxidase-conjugated antibody raised against mouse IgG (1:3000, Dako, Glostrup, Denmark). Expressed protein was detected by chemiluminescence (Western Lightning Chemiluminescence Reagent, PerkinElmer, Waltham, MA, USA).
A Caucasian male was seen in clinic with a history of male breast cancer and leiomyosarcoma of the thigh (presenting concurrently in his third decade), colon cancer (presented in his fourth decade) and prostate cancer (presented in his fifth decade). There were colorectal cancers with the age of onset in the fourth decade in his paternal grandfather and two paternal uncles. This family history is consistent with the revised Amsterdam criteria for the clinical diagnosis of Lynch syndrome (Vasen et al, 1999). His father had oesophageal cancer at the age of 47 years. Other unusual tumours in carriers of the variant included cervical adenosquamous carcinoma, oligodendroglioma and prostate cancer.
Initial genetic screening for mutations (by full sequencing) in the TP53 gene responsible for the Li-Fraumeni syndrome was negative. The breast cancer susceptibility genes BRCA1 and BRCA2 also did not harbour any mutations as screened by dHPLC (denaturing high-performance liquid chromatography) and MLPA (multiplex ligation-dependent probe amplification).
Several of the tumours in this family including the proband's sarcoma showed microsatellite instability. Molecular analysis of the proband's germline DNA identified a missense variant in the hMLH1 gene in exon 2 at nucleotide position 200 (c.200 G>A). This results in an amino-acid substitution from glycine to glutamic acid (Gly67Glu/G67E). The Gly67Glu mutation segregates with cancer predisposition across the pedigree.
To determine whether hMLH1-p.Gly67Glu (denoted hMLH1-G67E henceforth) is functional in MMR, we used two different approaches.
We took the advantage of the findings that the expression of functional human MLH1 (hMLH1) interferes with normal MMR in the budding yeast, whereas several point mutants associated with Lynch syndrome fail to do so. For example, hMLH1-p.Gly67Arg (G67R), known to occur in HNPCC families, shows reduced ability to interfere with yeast MMR in this assay (Figure 1; Shimodaira et al, 1998). We transformed budding yeast (Y39) with vector only (pRS315), vector containing hMLH1, hMLH1-G67R or hMLH1-G67E and assessed the forward mutation rate of CAN1 as well as the reversion rate of lys2::InsE-A14. CAN1 encodes an arginine permease that renders cells sensitive to the arginine analogue, canavanine. Mutations in CAN1 include base substitutions as well as insertions and deletions, which cause cells to become resistant to canavanine. The lys2InsE-A14 contains a homopolymeric run of 14 adenines, which causes a +1 frameshift mutation in LYS2 (Tran et al, 1997). Reversion reconstitutes LYS2, an α-aminoadipate reductase that allows cells to synthesise lysine. Qualitative assessment of four independent transformants shows that the ‘vector only' has no discernible effect on mutation rates of lys2InsE-A14 (Figure 1A). In contrast, introduction of ‘vector+hMLH1' causes an increase in the mutation rate of lys2InsE-A14, consistent with aberrant yMMR. Cells transformed with the previously characterized mutation, hMLH1-G67R, do not display an increased mutator phenotype, although some of the transformants do show a stochastic increase (data not shown). hMLH1-G67E behaved very similarly to hMLH1-G67R in that the majority of transformants did not show a mutator phenotype (Figure 1A and data not shown). These observations show that hMLH1-G67E displays a strongly reduced capacity for interfering with normal MMR function in budding yeast. We also quantified the mutation rates using the method of the median (see Materials and Methods) for both CAN1 and lys2InsE-A14. Introduction of hMLH1 causes a 18700-fold and 5.1-fold increase in the mutation rate estimates of lys2InsE-A14 and CAN1, respectively, whereas hMLH1-G67R did not enhance the mutation rates (1.2-fold for lys2InsE-A14 and 0.6-fold for CAN1 change in mutation rates as compared with ‘vector only'; Table 1). hMLH1-G67E shows a similar phenotype to hMLH1-G67R, with 1.9-fold and 1.1-fold difference compared with ‘vector only' (Table 1), suggesting that hMLH1-G67E has a similar loss-of-function phenotype as hMLH1-G67R in this assay.
The hMLH1 and hMLH1-G67R proteins are expressed stably in yeast cells (Shimodaira et al, 1998). To determine whether this is also the case for hMLH1-G67E, we carried out western blot analyses on the four transformants (Figure 1B). Protein extracts from cells containing hMLH1, hMLH1-G67R and the four independent transformants of hMLH1-G67E, but not vector only, showed the diagnostic 80kD band. Thus, hMLH1-G67E is stably expressed in budding yeast cells, but shows reduced ability to interfere with normal MMR functions.
The second yeast assay we employed is based on the observation that eukaryotic MLH1 homologues are highly conserved in the N-terminal ATPase domain. Several Lynch syndrome mutations, including hMLH1-G67R, confer a loss-of-function phenotype for mutator activity when introduced into the analogous residue in yeast MLH1 (yMLH1-G64R; Shcherbakova and Kunkel, 1999). We introduced the G67E change directly into the budding yeast locus (yMLH1-G64E) and assessed the mutation rates (Table 2). Deletion of yMLH1 causes a 30-fold and 19000-fold increase in the mutation rates for CAN1 and lys2InsE-A14, respectively, compared with the wild-type strain. The yMLH1-G64R mutant protein confers an intermediate phenotype, but nevertheless shows highly elevated mutation rates. The yMLH1-G64E mutation is similar to yMLh1-G64R, suggesting that changing the G64 residue in yMLH1 to either a positively charged arginine or a negatively charged glutamate confers similar inhibition of protein activity during MMR.
Characterising MMR gene alterations in Lynch syndrome families is important in clarifying risk and advising on management issues, such as cancer surveillance, treatment and prevention. Detailed study of a family with a tumour spectrum satisfying the Amsterdam criteria for Lynch syndrome has allowed us to characterise a variant in hMLH1. Interestingly, in addition to classic tumours, this family also exhibits a collection of atypical tumours. Functional analysis of this variant shows that it confers a loss-of-function mutator phenotype.
The p.Gly67Glu (p.Gly64Glu in yeast) mutation alters an amino-acid residue in the highly conserved N-terminus of hMLH1 (Ban and Yang, 1998). By analogy to the crystal structure of bacterial MutL, G67 is positioned on a loop, which is part of the ATP-binding pocket (Ban and Yang, 1998; Ban et al, 1999). ATP-binding induces conformational changes in the N-terminus that allows MutL monomers to dimerise (Ban et al, 1999), thus facilitating the interaction between hMLH1 and hPMS2. Moreover, cycles of ATP-binding and hydrolysis have been suggested to regulate MutLα activity (Tran and Liskay, 2000). Structure predictions would suggest that the hMLH1-p.Gly67Glu substitution alters the local conformation of the hMLH1 protein, analogous to that predicted for hMLH1-pGly67Arg (Ban and Yang, 1998). We tested this directly by comparing the G67E/G64E mutation to a previously characterised mutant in the same residue (G67R/G64R) in two functional assays. It is clear that expressing hMLH1-G67E does not appear to interfere with yeast MMR, which is not the case when normal hMLH1 is present. Furthermore, when the analogous substitution is made in the yeast protein, yMLH1-G64E, this protein does not sustain MMR fully and is as compromised as yMLH1-G64R. Thus, we conclude that hMLH1-G67E is a loss-of-function allele, similar to hMLH1-G67R.
We have identified that hMLH1-G67E segregates with disease in the family reported and is associated with atypical cancers. This is in sharp contrast to reported kindreds harbouring the hMLH1-G67R mutation who only showed classical elements of Lynch syndrome (Herfarth et al, 1997; Heinimann et al, 1999; Plevová et al, 2004). Furthermore, the hMLH1-G67E mutation has not been reported in 1688 individuals studied in the general population (Barnetson et al, 2008), consistent with the mutation being pathogenic. Our observations suggest that alleles that modify the effect of the G67E mutation (Scott, 2008) may be present in this family to explain the unusual spectrum. Alternatively, there may be functions of hMLH1 that are not assessed simply by determining whether the resulting change causes a mutator phenotype.
Because of the rarity of tumours in our Lynch syndrome family, this poses significant clinical management issues regarding recommendations on screening and treatment. Identification of more families with this pathogenic mutation could shed light on the mechanism for carcinogenesis in our kindred.
We thank Dr Shimodaira for providing us the plasmids containing hMLH1 and hMLH1-G67R. We are grateful to Drs Stewart J Payne, Sam Fisher, Ian Frayling, Rebecca Barnetson and Ann Barrett for helpful discussion during the preparation of the manuscript.
Conflict of interest
We declare the authors do not have any competing interests.
Funding: MC is funded by an MRC studentship and EH is funded by a Royal Society Dorothy Hodgkin fellowship. Work in the Hoffmann Laboratory was funded by a project grant from the Cancer Research, UK.
Work in the Yu Laboratory was funded by the MRC.
RE is supported by the Institute of Cancer Research and by the NHS funding to the NIHR Biomedical Research Centre.