Association of low-risk MSH3 and MSH2 variant alleles with Lynch syndrome: Probability of synergistic effects



Mutations in the MLH1 and MSH2 genes account for a majority of cases of families with Lynch Syndrome. Germ-line mutations in MSH6, PMS2 and MLH3 are responsible for disease in a minority of cases, usually associated with milder and variable phenotypes. No germ-line mutations in MSH3 have so far been associated with Lynch Syndrome, although it is known that impaired MSH3 activity leads to a partial defect in mismatch repair (MMR), with low levels of microsatellite instability at the loci with dinucleotide repeats in colorectal cancer (CRC), thus suggesting a role for MSH3 in carcinogenesis. To determine a possible role of MSH3 as predisposing to CRC in Lynch syndrome, we screened MSH3 for germ-line mutations in 79 unrelated Lynch patients who were negative for pathogenetic mutations in MLH1, MSH2 and MSH6. We found 13 mutant alleles, including silent, missense and intronic variants. These variants were identified through denaturing high performance liquid chromatography and subsequent DNA sequencing. In one Lynch family, the index case with early-onset colon cancer was a carrier of a polymorphism in the MSH2 gene and two variants in the MSH3 gene. These variants were associated with the disease in the family, thus suggesting the involvement of MSH3 in colon tumour progression. We hypothesise a model in which variants of the MSH3 gene behave as low-risk alleles that contribute to the risk of colon cancer in Lynch families, mostly with other low-risk alleles of MMR genes.


Hereditary nonpolyposis colorectal cancer (HNPCC), also known as Lynch Syndrome, is an autosomal dominantly inherited cancer syndrome that accounts for about 3–5% of all colorectal cancer (CRC) cases.1 HNPCC is commonly associated with DNA mismatch repair (MMR) genes.2 Mutations in the MLH1 and MSH2 genes account for the majority of the families with Lynch Syndrome.3 Germ-line mutations in MSH6,4PMS25 and MLH36 are responsible for the disease in a minority of cases. So far, no germ-line mutations in MSH3 have been found to be associated with HNPCC. MSH3 is located at chromosome 5q11-q12 and encodes for the MMR protein MSH3.7MSH3 forms a heterodimer with MSH2 (MutSb), which detects insertion-deletion loops and targets them for MMR. MSH3 carries out a partially redundant function with MSH68, 9 that is likely to increase the overall efficiency of MMR.10 A mouse model of MSH3 deficiency11 demonstrated that the MSH3 defect can cause late-onset microsatellite instability (MSI) positive gastrointestinal cancer, suggesting that MSH3 deficiency could contribute to tumour initiation.12MSH3 deficiency has also been associated with the progression of the disease; Dukes' stages C and D being more frequent in primary tumours with loss of MSH3 expression.13 Recently, studies examining the association between common polymorphisms in MMR genes and the risk of CRC indicate that common polymorphisms in the MSH3 gene could increase the risk of CRC.14, 15 Therefore, to investigate a possible role of MSH3 as predisposing to CRC in Lynch syndrome, we screened MSH3 for germ-line mutations in index patients from 79 Lynch families with no germ-line pathogenetic mutations in MLH1, MSH2 or MSH6. Our results suggest a likely involvement of the MSH3 gene in colon tumour progression.

Materials and Methods


Seventy-nine families of Italian origin, 51 families classified according to the Amsterdam criteria and 28 atypical Lynch families selected according to MSI high status (MSI-H), without germ-line point mutations or large rearrangements in the MLH1, MSH2 or MSH6 genes, were recruited from several health centres in Campania (Southern Italy).

All patients received genetic counselling and gave their written informed consent to participate in this study.

Isolation of genomic DNA

Total genomic DNA was extracted from 4-mL peripheral blood lymphocytes using a BACC2 Nucleon Kit (Amersham Life Science) and from tumour tissues and surgical margins by standard methods.16

DNA amplification

All MSH3 exons, including intron-exon boundaries, were amplified with oligonucleotide primers developed in this study (see Table 1). PCR reactions were performed in a total volume of 50 μL containing 5 μL of 10× PCR buffer (Roche, Basle, Switzerland), 200 μM of each dNTP, 25 pM of each primer, 1.5 mM of MgCl2, 2 U of FastStart Taq DNA polymerase (Roche) and 100 ng of genomic DNA. PCR conditions were as follows: initial denaturation at 95°C for 4 min, followed by 35 cycles with denaturation at 95°C for 30 sec annealing for 30 sec at the melting temperature of each primer, extension at 72°C for 45 sec, followed by a final extension step at 72°C for 7 min. Before dHPLC analysis, the PCR products were run on 1–2% agarose gel to check for unspecific amplicons.

Table 1. Primers, amplicon sizes, annealing PCR, and recommended dHPLC temperatures
inline image

dHPLC analysis

A Transgenomic Wave DNA Fragment Analysis System was used to perform dHPLC analysis (Transgenomic Inc., Omaha, Nebraska, USA). Before dHPLC analysis, the PCR products were denatured at 95°C for 5 min and gradually cooled to 20°C using a temperature ramp of 1°C/min on a PCR machine to enable efficient formation of heteroduplex. The mobile phase gradient and running column temperature selected for optimal heteroduplex separation were determined for each amplicon using the Wave Marker 4.4 software provided with the instrument. Aliquots of 8 μL of the PCR products were loaded onto a preheated DNASep® HT Cartridge column (Transgenomic Inc., Omaha, NB, USA). DNA was eluted at a flow rate of 0.9 mL/min using a linear acetonitrile gradient that consisted of buffer A (0.1 M triethylammonium acetate -TEAA-) and buffer B (0.1 M TEAA, 25% acetonitrile). Abnormal elution profiles were identified by visual inspection of the chromatogram on the basis of the appearance of one or more additional earlier eluting peaks. In this study, we set the optimum temperatures for each amplicon used in the dHPLC analysis, as shown in Table 1. For all samples exhibiting abnormal dHPLC profiles, genomic DNA was reamplified.

Sequencing analysis

The PCR products were sequenced in both the forward and reverse directions using an ABI 3100 Genetic Analyser (Applied Biosystems, Foster City, Ca., USA).

In silico analysis

Structure analysis of missense point mutations is very important to understand the functional activity of the mutated protein. For this purpose, we used the server PolyPhen,17 which is available at, for missense variants identified in this study. Predictions are based on a combination of phylogenetic, structural and sequence annotation information characterising a substitution with its position in the protein. In addition, we analysed the silent and intronic variants detected in this study by the Human Splicing Finder18 (HSF) software (available at, a new tool used to predict the effects of mutations on splicing signals or to identify splicing motifs in human sequences. It contains all available matrices for auxiliary sequence prediction and also presents a new position weight matrix to assess the strength of 5′ and 3′ splice sites and branch points.

DNA amplification and microsatellite analysis

The MSI status was confirmed with a fluorescent multiplex system comprising five quasimonomorphic mononucleotide repeats (BAT-25, BAT-26, NR-21, NR-24 and NR-27)19 using the CC-MSI Kit (AbAnalitica, Padova, Italy) and subsequent capillary electrophoresis analysis using an ABI 3110 Prism (Applied Biosystems).


Mutation analysis

All MSH3 exons were analysed in 79 Lynch families. We identified 13 variants, listed in Table 2, of which four are missense mutations, one a silent mutation and eight intronic variants. To date, 11 of these have not been described in the literature. The presence of all variants was tested in the 52 healthy controls. To make the pathogenicity more plausible of the variants, we performed in silico analysis using PolyPhen and HSF softwares. The results are shown in Tables 3 and 4, respectively.

Table 2. Sequence variations of hMSH31 evaluated for DHPLC analysis in familial colorectal cancer
inline image
Table 3. In silico analysis of the exonic variants in MSH3 gene
inline image
Table 4. Intronic mutation in MSH3 analyzed by HSF matrix
inline image

The missense mutation, c.1693 G>A in exon 12 (p. D564N), was found in the index cases of two unrelated families. This mutation determines at the protein level the substitution of an aspartatic acid residue with an asparagine residue, both of which are polar amino acids. The PolyPhen in silico analysis predicted that this variant would preserve the functional integrity of the protein. Analysis by HSF indicated that this variant might activate a cryptic splice site. This variant was not identified in the 52 healthy controls analysed. Unfortunately, it was not possible to test whether the mutation was associated with the disease in the two families.

The missense variant, c.2732 T> G in exon 20 (p. L911W), determines at the protein level the substitution of a leucine residue with a tryptophan residue. This variant falls in an evolutionarily highly conserved region. The PolyPhen in silico analysis predicted that it might affect the functionality of the protein (Table 3). It was identified in the index cases of two families (602 and 504) and was not found in the 52 healthy controls analysed. The proband of family 602 developed endometrial cancer at the age of 63 years and colon cancer at the age of 68 years. No other family members were available for the analysis. The index case of family 504 (also a carrier of a silent mutation in the MSH2 gene, c.984 C>T in exon 6) developed a tubulo-villous adenoma of the colon at the age of 34 years and an adenocarcinoma of the right colon at the age of 42 years. To elucidate whether the mutation was associated with the disease, we analysed affected and unaffected family members. The variant was also found in a brother and a son of the index case, both with CRC (Fig. 1). The index case of this family also showed a second variant in the MSH3 gene, c.693 G>A, a silent variant in exon 4 [p.= (Pro)] that was identified only in this family. The in silico analysis carried out using the HSF software showed that this variant occurs in a region involved in the splicing process (Table 3). It was also identified in the daughter of our index case and in another two family members, all three with CRC.

Figure 1.

(a) Pedegree of 504 family. Symbols and abbreviations used are denoted as fellow: Arrows, analysed members of family; black symbol, colorectal cancer or cancer associate with HNPCC; gray symbols, adenomas or cancer not associated with HNPCC; CRC, colorectal cancer; Br, brain cancer; GU, gastric ulcer; BL, bladder cancer; Bre, breast cancer; TA, tubular adenoma. Number next to diagnosis denote age at oneset; • not detected; (b) genotypes of analysed patients; the patients are identified with number of pedigree.

The other sequence variants identified are all intronic, and the results of the HSF analysis are shown in Table 4. The variant c.358+40 t>c in intron 2, detected at the homozygous level, was identified in one family. This variant was found in the index case, a woman with right colon cancer diagnosed at the age of 58 years, and it was not found among the healthy controls. The variant c.910−64 c>t, in intron 5, was identified in 33 Lynch families, and it was found in one of the 52 healthy controls analysed. The variants c.1028–9 t>c in intron 6 and c.1454–62 a>g in intron 9 were identified in three Lynch families and in none of the healthy controls. The probands of these families had developed right colon cancer at about 40 years of age. The variant c.1763+71 c>a in intron 12 was identified in two Lynch Families and in one healthy control. Both probands of these families had also shown the missense variant c.1693 G>A, in exon 12, as described above. The variant c.3001−22 t>a in intron 21 was identified in nine Lynch families and in seven healthy controls. The HSF analysis showed a significant ΔCV value for the creation of new potential cryptic splice sites (Table 4). The variant c.3001−33_36del tgaa in intron 21 was identified in forty-four Lynch families and with a high frequency even in the healthy control subjects.

Finally, in this study, we identified the known MSH3 sequence variant, c.2846 A>g and c.3133 G>a. These polymorphisms were identified in 31 and 14 families, respectively (see Table 2).

MSI analysis

The MSI analysis was performed on DNA extracted from tumour tissues and surgical margins of the index case of family 504 and of his son. In accordance with the literature data,18 the patients analysed showed an MSI-H status. The following nucleotide repeats resulted unstable: Bat 25, Bat 26 and D2S123 for the index case and Bat 25, Bat 26 and D17S250 for his son.


In this study, we identified 13 germ-line genetic variants within the MSH3 gene in 79 unrelated patients with Lynch syndrome.

The MSH3 gene has long been known as a member of the DNA MMR genes. The human protein MSH3 is functionally redundant8, 9 with MSH6. Specifically, MSH3 recognises and repairs insertion/deletion loops of two or more nucleotides in combination with MSH2 to form the heteroduplex MutSβ.10MSH3 knockout mice showed a low susceptibility to a cancer phenotype development causing late-onset gastrointestinal cancer,11 whereas double mutant MSH3-MSH6 mice showed a very similar phenotype to that found in mice lacking MSH2.20 No germ-line mutations in MSH3 have yet been associated with HNPCC, although it is known that mutations in MSH3 lead to a partial defect in MMR and to some MSI,12 suggesting a role for MSH3 in the carcinogenesis. Therefore, we conducted this study to clarify the role of the MSH3 gene in the carcinogenesis of Lynch syndrome. We identified several variants that were silent, intronic and missense variants. No frameshift or nonsense mutations were identified. Because the variants identified may or may not lead to an altered protein function, and for most of these it was not possible to determine an association of the variants with the disease, we also screened 52 healthy controls (Table 2). Furthermore, we used an in silico analysis to predict the pathogenetic effects of the variants detected. The computational assays help to analyse the putative functional effects of the sequence variants. We used the PolyPhen17 software for missense variants and the HSF18 software for all variants identified. These softwares have often been used to study unclassified variants (UVs) present in patients with Lynch syndrome.21, 22

In this study, we focused our attention on two of the 13 genetic variants identified, the missense c.2732 T>G and the silent c.693 G>A. For these variants, we analysed the association with the disease in the family.

The missense variant c.2732 T>G in exon 20 was found in the index case of family 504 (II-5 in Fig. 1), who had developed an adenocarcinoma of the left colon at the age of 34 years, an adenocarcinoma of the right colon at the age of 53 years and a new malignancy of the colon at 59 years of age. The PolyPhen in silico analysis showed that this missense variant might alter the function of the protein, because it falls into a highly conserved region in different species. Besides the missense c.2732 T>G variant, this patient was a carrier of a second variant of the MSH3 gene, c.693 G>A, in exon 4. This is a silent variant for which the HSF analysis showed that it might affect the splicing process (Table 3). Previously, the same patient had been identified as a carrier also of a polymorphism23 or UV24 within the MSH2 gene, c.984 C>T, in exon 6. According to literature data, no aberrant splicing of mRNA had been found in this patient (data not shown). In this family we analysed eight other members. Of these, one brother (II-6 in Fig. 1) showed the same genotype as our proband; in fact, he had all three mutations found in our index case. Interestingly, this patient also showed the same phenotype (Fig. 1). Instead, another brother (II-8 in Fig. 1) showed only a variant in the MSH2 gene and no genetic variants in the MSH3 gene. This patient had developed a polyp of the colon at 47 years of age. Today he is 59 years old, undergoes regular colonoscopy and so far has not presented other polyps. In the third generation (Fig. 1), we analysed four affected family members. Subjects III-1 and III-2, in Figure 1, showed a silent variant in MSH3 and a variant in MSH2; both subjects showed an early-onset right colon tumour. Subjects III-3 and III-4, in Figure 1, the sons of our proband, developed colon cancer at 36 years of age and a tubular adenoma of the colon at 34 years of age, respectively. Both subjects showed a silent variant in MSH2 and a missense variant and a silent variant in MSH3. The MSI analysis performed on DNA extracted from tumour tissues of patients II-5 and III-3 showed an MSI-H status. Thus, both subjects presented a strong mutator phenotype, probably due to an additive effect by several variants that leads to inefficiency of the MMR complex. The other family members analysed showed only one mutation in the MSH3 gene and did not present a typical phenotype of Lynch syndrome. Therefore, in this family, it is clear that all subjects with the Lynch phenotype showed the c.984T allele of MSH2 and a germ-line variant in the MSH3 gene (a missense and/or silent variant). We, therefore, speculate that the presence of only one mutation in the MSH3 gene is not enough to determine the onset of the tumour, whereas the association with a weak mutation in the major MMR genes (MSH2 or MLH1) could determine the onset of the tumour. Recently, the effect of polymorphisms and missense mutations in human MMR genes was studied in a Saccharomyces cerevisiae-based system. A number of weak alleles of MMR genes and MMR gene polymorphisms have been identified that are capable of interacting with other weak alleles of MMR genes to produce strong polygenic MMR defects.25 The similar situation found for our family might support the hypothesis that weak MMR gene alleles are capable of polygenic interaction with other MMR gene alleles that might lead to tumour progression in Lynch syndrome. The literature data indicate that MSI-H tumours show frequent somatic alterations in the MSH3 gene and its simultaneous deactivation could potentiate the effects of other mutations in other MMR genes.12, 26 Thus, these germ-line variants in the MSH3 gene might also influence the penetrance of mutations in other MMR genes. Further support to this hypothesis is given by recent studies showing that the MSH2-MSH3 complex also plays an important role in triplet repeat expansions.27, 28 It has long been known that the phenomena of anticipation, the severity of the phenotypic expression and the penetrance appear to be related to the extent of the triplet expansion. This could explain why the phenotype of cancers with the same germ-line mutation can vary in different families. Our results indicate that it is very important to analyse the MSH3 gene in patients with mutations in the MLH1 or MSH2 genes, to evaluate whether the presence of germ-line variants in the MSH3 gene might modify the penetrance and phenotypic expression of the disease.

In conclusion, several germ-line variants were identified in the MSH3 gene using a dHPLC procedure; this method, robust, automated and highly sensitive, is fast, feasible and particularly useful for high-throughput analyses. The effect of the variants identified is not always clear, and further studies are certainly needed to define their pathogenic role. On the basis of this study, it is conceivable to hypothesise a model in which the variants of the MSH3 gene behave as low-risk alleles that contribute to the risk of colon cancer in Lynch families, mostly together with other low-risk alleles of other MMR genes.

Therefore, if our assumptions are correct, the study carried out for the MSH3 gene may indicate a novel inheritance model in the Lynch syndrome, and might suggest that the risk alleles identified to date represent just the tip of an iceberg of risk variants likely to include hundreds of modest effects and possibly thousands of very small effects. This could pave the way toward new diagnostic perspectives.


The authors acknowledge the contribution of Doctor Antonio Paolucci to this study.