Introduction

The clinical variability observed in autosomal dominant cancer predisposition syndromes may be explained by modifying genes, environmental factors and stochastic effects. In conditions characterised by a variable penetrance, inconsistent manifestation and often sporadic presentation, combination of low-penetrant germline mutations in two or more functionally related genes may aggravate the phenotype and affect the tumour spectrum, exceeding the threshold level to medical attention.

Germline mutations in the MMR genes MLH1, MSH2, MSH6 and PMS2 are known to cause hereditary non-polyposis colorectal cancer (HNPCC/Lynch syndrome), an autosomal dominantly inherited tumour predisposition syndrome associated with colorectal and endometrial cancer and several other extracolonic malignancies.1 Mutations in the MSH6 gene account for about 10–15% of all HNPCC germline mutations. The gene products of MSH6 and MSH2 form the MutSα heterodimeric protein complex, which recognises base–base mismatches and small insertion–deletion loops and initiates the repair by excising the mispaired base.2 Thus, mutations in MSH6 decrease the capacity of the MMR system, resulting in augmentation of somatic mutations in tumour suppressor genes and oncogenes.3 However, the penetrance of MSH6 mutations is lower compared to germline mutations of the MMR genes MLH1 and MSH2. MSH6 mutation carriers tend to have a later age of onset and a lower incidence of colorectal cancer; the families of these patients often neither fulfil the Amsterdam criteria for HNPCC nor follow the typical autosomal dominant pattern of inheritance.4 This raises the question whether the MSH6 mutation alone or in combination with additional inherited susceptibility factors is responsible for the increased tumour risk observed in these patients.

Some aspects of the phenotype of HNPCC have already been associated with variants in other modifying genes, in particular apoptosis-related genes.5, 6, 7 In 2002, Gu et al2 reported an interaction between MSH6 and the base excision repair (BER) protein MUTYH, a DNA glycosylase involved in the repair of oxidative DNA damage. The authors showed that the MSH2/MSH6 complex enhances the binding affinity of MUTYH to the mismatched DNA substrate and the glycosylase activity of MUTYH. Biallelic mutations in MUTYH are known to cause MUTYH-associated polyposis, an autosomal recessive precancerous condition of the colorectum.8, 9, 10, 11, 12, 13 The heterozygote frequency in the general population is estimated to be 1–2%.9, 12, 14 The risk for colorectal cancer in heterozygote MUTYH mutation carriers has been discussed controversially but is apparently low.15, 16, 17, 18

Due to the functional interaction of the MUTYH protein with the MMR system, it was hypothesised that monoallelic MUTYH germline mutations affect the phenotype of HNPCC patients and, in particular, contribute to cancer susceptibility in carriers of an MSH6 mutation. Recently, Niessen et al19 reported a significantly higher rate of monoallelic MUTYH mutations in a group of 20 MSH6 missense mutation carriers compared to carriers of either MLH1 or MSH2 mutations or to healthy controls and concluded that the interaction of mutations in MSH6 and MUTYH may lead to an increased risk for colorectal cancer.

To further examine the suspected interaction of mutations in both genes, we sequenced the complete coding region of the MUTYH gene in 64 MSH6 mutation carriers recruited by the German HNPCC Consortium and compared the frequency of MUTYH mutations among these patients with the frequency of that among healthy controls.

Materials and methods

Patients

The patients included in the study were recruited from six German university hospitals as described.20 Briefly, in all centres, patient ascertainment, data analysis and documentation were carried out in accordance with the common study protocol. Patients were referred to the study from other hospitals, institutes of human genetics, private practice physicians and private practice human geneticists or came by self-referral. Patients who were enrolled in the study had to fulfil either the Amsterdam criteria II,21 the original Bethesda guidelines22 or the revised Bethesda guidelines.23 Histopathological analysis was controlled at the central reference pathology unit (Department of Pathology, University of Bonn).

We included 64 unrelated index patients from the German HNPCC Consortium who had been examined for germline MMR gene mutations according to the study protocol (see below) and had been found to harbour an MSH6 germline mutation. Patients with pathogenic mutations in MLH1 or MSH2 were not included, but 11 patients were found to have an additional mutation of unknown pathogenic relevance in MLH1 (nine probands) or MSH2 (two probands). (For details see Supplementary Table 1). Of the 64 patients, 42 had a pathogenic MSH6 mutation (frameshift mutation, nonsense mutation or splice-site mutation) and 22 patients had a mutation of unknown pathogenic relevance in MSH6 (19 missense mutations and 3 silent mutations). Ten patients fulfilled the Amsterdam II criteria, 45 patients met the original Bethesda guidelines and 9 patients the even less stringent revised Bethesda guidelines.21, 22, 23 All patients gave their written informed consent authorising data documentation, examination of tumour tissue for HNPCC characteristics and molecular genetic analysis of genes associated with HNPCC. The study was approved by the ethical committees of all participating clinical centres.

Immunohistochemical staining of MMR proteins

Immunohistochemistry (IHC) for the MMR proteins MLH1, MSH2 and MSH6 was performed according to the study protocol as described previously.20, 24 The level of protein staining in the tumour cells was compared to the protein expression in normal tissue. Tumours were scored to exhibit loss of expression of a repair protein if the nuclei showed no or only very weak immunostaining in comparison to normal tissue.

DNA extraction

Genomic DNA was extracted from peripheral EDTA-anticoagulated blood samples, according to a standard salting-out procedure. With small DNA samples, whole genome amplification was performed using the GenomiPhi DNA amplification kit by GE Healthcare (Chalfont St Giles, Great Britain, UK), according to the recommendations of the manufacturer. The coding sequence of MSH6 was amplified and sequenced from genomic DNA as described previously.24, 25

Microsatellite analysis

Analysis for microsatellite instability (MSI) had been performed on matched pairs of tumour DNA and normal DNA using the National Cancer Institute/International Collaborative Group on HNPCC (NCI/ICG-HNPCC) reference marker panel for the evaluation of MSI in colorectal cancer (BAT25, BAT26, D2S123, D5S346 and D17S250). Tumours were scored as highly unstable (MSI-H) if two or more of these five markers exhibited additional alleles, and as stable (MSS) if none of the five markers showed instability. If only one marker showed instability, an additional panel of five markers (BAT40, D10S197, D13S153, MYCL1 and D18S58) was examined. In these cases, the tumour was classified as MSI-H if 3 or more of the 10 markers showed instability, and as low unstable (MSI-L) if only one or two markers showed additional alleles. MSI-H tumours were found in 6 of the Amsterdam-positive patients, 35 of the patients meeting the Bethesda criteria and 6 of the patients fulfilling the revised Bethesda criteria.

Mutation analysis of the MSH6 gene

In general, mutation analysis of the MSH6 gene was performed according to the study protocol if the patient's tumour tissue showed an isolated loss of MSH6 in the immunohistochemical staining or – when no tumour tissue was available – if the patient's family met the Amsterdam criteria and no pathogenic germline mutation in MLH1 or MSH2 was found. Individual patients were examined for other reasons at the discretion of the respective centres. Mutations were either categorized as pathogenic (frameshift mutations, nonsense mutations, splice site mutations and one deletion of two exons), unclassified variants (missense mutations and silent mutations, that have not been functionally tested so far and are not described in the literature as polymorphisms) or polymorphisms (if described in the databases and literature; known polymorphisms are not shown in the Supplementary Table 1). The mutations in our 64 patients are distributed over the whole MSH6 gene. An apparent clustering of mutations in exon 4 can be explained by the size of this exon containing more than one-third of the MSH6 cDNA.

Screening for germline mutations in the MUTYH gene

Screening for MUTYH mutations was performed by amplifying and sequencing the whole coding region and the flanking exon–intron boundaries of the MUTYH gene as described previously.10 We applied the description of the coding sequence used by Al-Tassan et al8 (GenBank: U63329.1) for mutation description and not the actual reference sequence of the coding MUTYH sequence (GenBank: NM_012222.1). All mutations were confirmed by a second independent PCR.

MUTYH mutation frequency in Caucasian controls

Data on the MUTYH mutation frequency in the general population were taken from three published studies in which large groups of normal Caucasian controls had been screened for mutations in the whole coding region of the MUTYH gene.9, 12, 14 One of the three control groups encompasses 116 German probands (healthy blood donors).14 In all 577 control individuals, 9 monoallelic MUTYH mutations were identified (some of them of unknown functional relevance), indicating a frequency of around 1.6% in the general population. In the German control group, monoallelic MUTYH mutations were found in two patients (1.7%).

Statistical analysis

The statistical comparison (frequency of monoallelic MUTYH mutations in the different subgroups of our patients) was performed using Fisher's exact test for categorical variables. A P-value of <0.05 was considered to be statistically significant.

Results

We sequenced the whole coding region of the MUTYH gene in 64 carriers of an MSH6 germline mutation and identified two monoallelic MUTYH mutations (3.1%). No patient harboured a biallelic MUTYH mutation. The frequencies of the previously reported MUTYH polymorphisms were consistent with published data.9, 11, 12 The inclusion criteria, results of the tumour tissue examination and mutation analysis are summarised in Table 1.

Table 1 Mutation status and results of MSA and IHC in 64 MSH6 mutation carriers according to clinical HNPCC criteria
Full size table

The patient with the pathogenic MSH6 mutation, c.3324_3325insT;p.Ile1109TyrfsX3, harboured the truncating MUTYH mutation c.247C>T;p.Arg83X that had been described earlier.9 The patient had been diagnosed with ovarian cancer at the age of 54 and with colorectal cancer at the age of 55 years and, therefore, met the Bethesda criteria. Her tumour tissue showed MSI-H and a nuclear loss of the MSH6 protein. No other family members could be investigated.

The second patient with the silent MSH6 mutation, c.1770C>T;p.Pro590, of unknown functional relevance carried the MUTYH missense mutation c.502C>T;p.Arg168Cys that leads to an amino-acid exchange in an evolutionarily conserved residue, suggesting pathogenicity.11 She had been diagnosed with an MSS rectal cancer at the age of 51 years, and her family history met the Amsterdam criteria. Her tumour tissue was positive for all tested MMR proteins in immunohistochemical staining. In addition, the patient harboured the MLH1 missense mutation c.2146G>A;p.Val716Met, which is suspected of being a rare polymorphism.26, 27, 28 Unfortunately, no other family members were available for further investigation.

Thus, one of the two MUTYH mutations was found in the 42 patients harbouring a pathogenic (truncating) MSH6 mutation (2.4%) and the other in one of the three patients with a silent MSH6 mutation. No MUTYH mutation occurred in the 19 patients with an MSH6 missense mutation. (For clinical and molecular genetic details, see Supplementary Table 1). Regarding the clinical inclusion criteria, one MUTYH mutation was found among the 10 patients whose family history met the Amsterdam criteria (10.0%), and one in the group of 45 patients fulfilling the Bethesda criteria (2.2%). No MUTYH mutation was found in the nine patients meeting the revised Bethesda criteria.

The heterozygote frequency in the general population was estimated to be around 1.6%, based on the published results of a complete MUTYH screening among 577 normal Caucasian controls.9, 12, 14 Hence, the frequency of heterozygous MUTYH mutations among our MSH6 mutation carriers did not differ significantly from the frequency observed in healthy individuals, neither in the whole cohort of 64 patients (P=0.30) nor in different subgroups (pathogenic MSH6 mutations, P=0.51; MSH6 missense mutations, P=0.75). In contrast to Niessen et al,19 we found no increased frequency of MUTYH mutations among our MSH6 missense mutation carriers, although the difference between both studies was not significant (P=0.106). Based on our sample sizes, the statistical power of our study to detect the same difference in the frequencies of MUTYH mutations between MSH6 missense mutation carriers and normal controls as seen by Niessen et al19 (ie 20 vs 1.5%) was 79%. If all MSH6 mutation carriers were considered (64 patients), the power was >99%.

Discussion

It is a reasonable hypothesis to assume that the combination of low-penetrant germline mutations in two or more functionally related genes explains the phenotype variability in hereditary tumour predisposition syndromes characterised by an incomplete penetrance (and sporadic appearance) like that in MSH6 related HNPCC families.

Interactions between MSH6 and the BER protein, MUTYH, were first described by Gu et al.2 These authors could demonstrate that the MSH2/MSH6 heterodimer stimulates the DNA binding and glycosylase activities of MUTYH to misincorporated adenines opposite 8-OxoG. Interestingly, they found a physical interaction between MUTYH and MSH6 that was substantially decreased in MT1 cell lines with compound heterozygous MSH6 missense mutations (Asp1213Val and Val1260Ile) in the C-terminal region of MSH6. These cells are deficient for base–base mismatch repair.

Since the interaction of the MMR system and the BER system had been reported, two studies were undertaken to investigate a possible modifying effect of MUTYH mutations on the course of disease in HNPCC patients. Ashton et al29 could not find a higher frequency of MUTYH mutations among MLH1 and MSH2 mutation carriers (the type of mutation was not reported) and mutation-negative HNPCC patients, compared to each other and to normal controls: they screened 442 HNPCC patients for the presence of the two common MUTYH mutations, Y165C and G382D, and identified monoallelic MUTYH mutations in two of the 209 MLH1 or MSH2 mutation carriers (all fulfilled the Amsterdam or the Bethesda criteria). In the group of 233 mutation-negative HNPCC patients, 3 harboured a monoallelic and 2 biallelic MUTYH mutations.

Niessen et al19 examined 76 carriers of MMR germline mutations (25 with mutations in MLH1, 26 in MSH2 and 25 in MSH6) for MUTYH variants. Of 20 patients carrying a MSH6 missense mutation, 4 (20%) were found to have a monoallelic MUTYH mutation. The proportion was significantly increased compared to normal controls (1.5%; P=0.001) and to a group of 134 CRC patients without an MMR mutation (0.7%; P=0.002). The frequency of MUTYH mutations among the MLH1 and MSH2 missense mutation carriers and among the carriers of truncating mutations in the MMR genes was not significantly different from the frequency in the healthy controls. In summary, the combined results of the two studies indicate that an interaction of monoallelic MUTYH and MMR germline mutations might be relevant only in MSH6 missense mutation carriers, but not in the more penetrant MLH1, MSH2 and truncating MSH6 mutations.

To identify a potential interaction between mutations of the two genes, we screened 64 MSH6 mutation carriers of the German HNPCC Consortium for MUTYH mutations and found two monoallelic mutations among all patients. In contrast to the findings of Niessen et al,19 no MUTYH mutation was identified in the 19 carriers of an MSH6 missense mutation. Compared to the general population, the frequency of MUTYH mutations was not increased in our sample, neither in the whole group of 64 patients nor in different subgroups regarding mutation type.

One explanation for the divergent findings might be a different composition of the two patient groups. In fact, both samples seem to vary in certain aspects, but are similar in others. The clinical inclusion criteria for HNPCC diagnostics mentioned by Niessen et al19 are comparable to those of the German HNPCC consortium (Amsterdam or Bethesda), suggesting no relevant ascertainment bias between the groups, at least in the majority of patients. Since neither the 20 MSH6 missense mutations nor the IHC and MSI results are described in detail by Niessen et al,19 the degree of correlation between the two groups cannot be determined exactly. The diverse proportion of MSH6-truncating and missense mutations in both studies (5/20 vs 42/19) may indicate a difference in the spectrum of MSH6 missense variants. However, two of the three different MSH6 missense mutations published by Niessen et al19 (all had a MUTYH mutation) were also identified in our patients (c.1186C>G;p.Leu396Val; c.431G>T;p.Ser144Ile), indicating a substantial overlap.

Notwithstanding, in the majority of our MSH6 missense mutation carriers (13/19), mutation analysis in MSH6 was performed only if the patient's tumour tissue was MSI-H or showed a loss of MSH6 in IHC whereas all tumours of the four MSH6/MUTYH mutation carriers examined by Niessen et al19 showed normal IHC staining for MSH6 and only one tumour tissue was MSI-H. According to these results of Niessen et al,19 a combined MSH6/MUTYH mutation might be relevant only in case of low-penetrant MSH6 missense mutations, which neither alter IHC nor microsatellite stability. As a consequence, the standard diagnostic criteria used for MSH6 mutation analysis (MSI-H, IHC) would be inappropriate to detect the patients identified by Niessen et al.19

It cannot be ruled out that the difference in the frequency of MUTYH mutations between the German and the Dutch patients occurred by chance due to the limited number of MSH6 missense mutation carriers. It will be difficult to recruit much larger cohorts of MSH6 missense mutation carriers in the near future; however, further studies are needed to clarify this issue.

In conclusion, according to our results, the frequency of MUTYH mutations that was identified in a large cohort of MSH6 mutation carriers who fulfilled the clinical criteria for HNPCC is not increased compared to normal controls. The over-representation of heterozygous MUTYH mutations in MSH6 missense mutation carriers reported by Niessen et al19 remains to be explained.