Editor—Colorectal carcinoma (CRC), among the most common neoplasms in humans, has a moderately large hereditary component. The two most frequent hereditary CRC syndromes are hereditary non-polyposis colorectal cancer (HNPCC) and familial adenomatous polyposis (FAP), both dominantly inherited disorders. HNPCC is caused by inherited defects in the DNA mismatch repair (MMR) geneshMLH1, hMSH2,PMS1, and PMS2; FAP is caused by mutations in the APCgene.1 Between 1 and 4% of all CRC fully satisfy the international criteria for HNPCC.2-4 In addition to CRC, these families also have an excess incidence of adenocarcinoma of the endometrium and to a lesser degree, cancer of the stomach, ovary, and other sites.5 Alterations inhMSH2 or hMLH1account for ∼90% of the germline mutations detected in HNPCC.1 Mutations in hMSH6 have been described in CRC kindreds that resemble HNPCC but do not always fulfil the above mentioned criteria.6 ,7

Familial adenomatous polyposis (FAP) accounts for less than 1% of CRC; it is characterised by the development of a large number (>100) of colorectal polyps, which, if untreated, will inevitably lead to CRC.8 The gene underlying FAP,APC, was identified in 1991; germline mutations truncating the APC gene product are usually found in FAP.9 ,10 An interesting missense variant in APC, I1307K, was discovered in a 39 year old man with eight colorectal polyps11 with a family history of colorectal polyps and CRC but not FAP. The T to A polymorphism at nt 3920 in APC was postulated to predispose to cancer, not as a direct effect of the protein, but by rendering its region of APChypermutable. The mutation does not seem to be common in populations other than Ashkenazi Jews, where it is found at a frequency of 6.1%11 (recently we have found the allele in a French Canadian kindred12). The relative risk of CRC and other neoplasms associated with I1307K APC has been investigated.13-17 Woodage et al 15 concluded that the cumulative risk of CRC among first degree relatives of CRC survivors among carriers compared to non-carriers was 1.9, although this was not statistically significant. By examining a cohort of unselected Ashkenazi Jewish patients with colorectal carcinoma or adenomatous polyps, Gryfeet al 16 found statistically significant (p=0.01) relative risks of colorectal neoplasia of 1.72 (95% confidence interval 1.13-2.61) and 1.48 (95% confidence interval 1.08-2.04) compared to controls from New York and Washington, DC, respectively. Recent investigations have indicated that I1307KAPC is a candidate low penetrance breast cancer susceptibility allele, particularly in heterozygotes for germline mutations in BRCA1 orBRCA2.15 ,17 It is hoped that the identification of mutations in CRC susceptibility genes will result in decreased morbidity and mortality by the application of appropriate screening and intervention strategies. To this end, in an Ashkenazi Jewish kindred with HNPCC, we used single strand conformation analysis (SSCA) to search for mutations in MMR genes and an allele specific oligonucleotide hybridisation (ASOH) assay to identify I1307KAPC. The proband (fig 1, II.11) was seen by both the medical genetics and colorectal surgery services and a family history was taken, identifying the family as an Amsterdam criteria positive HNPCC kindred (MON702). There are three relatives with CRC, two of the affected subjects are mother and son, and two cases were diagnosed at less than 50 years of age (II.13 and III.2). Informed consent was received for blood donation as approved by the institutional ethics committee. To assess the occurrence of MSI in CRCs from two affected subjects, six high sensitivity MSI markers (dinucleotide repeats D2S123, D18S58, and LNS and mononucleotide repeats BAT25, BAT26, and BAT40)18 ,19 were used to evaluate one CRC sample from each patient. The proband’s tumour DNA exhibited instability with three of the six microsatellite markers (data not shown) and her son’s (III.2, fig 1) tumour DNA showed alterations in allele length compared to matched normal DNA with BAT25, BAT26, and D18S58 (fig 2), defining both neoplasms as MSI⁺according to established criteria.18 ,20 We then used PCR-SSCA to analyse hMLH1 21 andhMSH2 22 for heterozygous mutations. A mobility shift was identified in exon 12 ofhMSH2 of genomic DNA amplified from two members of MON702, II.3 and III.2. A germline heterozygous G to C transversion at nucleotide 1905, resulting in a predicted protein sequence change of alanine to proline (A636P), was identified in both subjects showing the mobility shifts. Three unaffected sibs of the affected son (III.1, III.3, and III.4; fig 1) do not carry this mutation. We sequenced all 16 hMSH2 and 19hMLH1 exons in the two affected subjects and did not find another mutation elsewhere in eitherhMSH2 or hMSH1that might otherwise account for the MSI⁺observed.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Pedigree of the MON702 kindred. The age at death or the current age of the subjects are shown. +/A636P or +/+=presence or absence of exon 12 mutation in hMSH2; +/I1307K or +/+=presence or absence of I1307K variant in APC.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Microsatellite analysis of CRC removed from III.2 in the MON702 kindred. For each locus, the left lane shows the results from the neoplastic tissue; the right lane was loaded with amplification products from adjacent normal tissue. This sample was scored as MSI for BAT25, BAT26, and D18S58; thus, the neoplasm was considered MSI positive.

The A636P mutation appears to be associated with MSI⁺. To exclude the possibility that it represents a rare polymorphism over-represented in the Ashkenazim, but not associated with HNPCC, we therefore screened 100 Ashkenazi Jews who donated blood anonymously for DNA studies but were not known to have CRC. No carriers of A636P were found by ASOH analysis. Another important question is whether this is a common mutation in Ashkenazi Jewish CRC patients. In total, DNA samples from 196 unselected Jewish CRC patients were successfully amplified by PCR and tested for the presence of the A636P allele. BAT-26 MSI⁺ was present in the tumour DNA extracted from the neoplasms of 15 (7.6%) of these patients. No carriers of A636PhMSH2 were identified; these findings further indicate that A636P is not a common sequence variant in this population.

Following the report11 that I1307KAPC is common in Ashkenazi Jewish CRC families, we investigated whether this mutation was present in the MON702 family. CRCs from subjects carrying this variant are usually MSI negative.11 ,23 It was, therefore, not surprising that I1307 APC was not present in the affected subjects in MON702, whose CRCs were MSI⁺. However, unaffected subjects III.3 and III.4 were found to carry I1307KAPC; these positive ASOH results were confirmed by sequencing. We also observed I1307KAPC in eight out of 153 samples from Ashkenazi Jewish controls (5.2%), a similar figure to those reported previously.11 ,15

Most of the hMSH2 mutations described so far in HNPCC result in a truncated protein and allow an immediate inference of pathogenicity. In the absence of large kindreds with numerous affected subjects, it is difficult to evaluate the pathogenicity of non-truncating mutations. Ideally, previously conducted functional studies can shed light on the aetiological significance of missense mutations in small kindreds such as MON702; inactivating missense mutations at the ATP binding site of hMSH2 have been described.24 On the other hand, six missense mutations found in HNPCC patients were found not to affect protein-protein interactions of hMSH2 and hMSH6 or hMSH3.25 Unfortunately, however, amino acid 636 of the hMSH2 protein has not been directly examined in any of the published studies on the function ofhMSH2.24-27 TheAPC gene has been examined inE coli using a colorimetric assay based on an in frame fusion with the β-galactosidase gene.28Andreutti-Zaugg et al 29designed a fusion protein assay in S cerevisiae in which hMSH2 is inserted upstream of the ADE2 gene, in the same reading frame. When this plasmid is expressed in S cerevisiae, high levels of active Ade2 protein are formed, resulting in white colonies in a colorimetric assay. Truncating mutations in the hMSH2 gene lead to truncated fusion protein, hence red colonies, reflecting lack of Ade2 activity. Surprisingly, some missense changes, including the HNPCC mutation G674D at the ATP binding site, resulted in pink colonies, reflecting partial activity of the ADE2 reporter gene. Since protein expression was similar, the reduced enzymatic activity of the fusion protein was inferred to be the result of incorrect protein folding, leading to a denatured protein.29 We introduced A636PhMSH2 into plasmid pCA57 by site directed mutagenesis (QuikChange kit, Stratagene), creating an A636PhMSH2::ADE2 fusion protein. A mutant protein of the correct relative molecular mass was observed to be expressed at comparable levels to its wild type counterpart by immunoblotting analysis of cellular extracts. Next, the phenotype of the A636P fusion protein was determined in the colorimetric in vivo assay.29 Colonies from cells containing the pCA57-hMSH2 A636P mutant plasmid were pink. This indicated a partial loss of function of the fusion protein, and suggested that the mutation does affect hMSH2 protein folding inS cerevisiae and by implication may affect structure and function in human cells. This finding further validates the hMSH2::ADE2 assay as capable of detecting certain non-truncating mutations inhMSH2.

In MON702, an Amsterdam criteria positive HNPCC kindred, we found that the CRCs from two affected subjects (II.3 and III.2) showed MSI⁺. A missense G to C (nucleotide 1905) transversion mutation in hMSH2 was identified at codon 636 in both these subjects, leading to a predicted protein change of alanine to proline in the hMSH2 protein. This mutation has not been previously published, but was recently posted on the HNPCC ICG database (http://www.nfdht.nl, Bressac-de Paillerets, unpublished data). However, these investigators were uncertain about the pathogenicity of the A636P mutation. We consider the A636P missense mutation to be pathogenic for the following reasons: (1) the mutation segregates with the disease, and three colonoscopy negative relatives do not carry the mutation; (2) the mutation was not found in 100 Ashkenazi Jewish controls, who were not known to have CRC; (3) no other truncating or missense mutation was identified on complete screening ofhMLH1 or hMSH2 by SSCA and direct sequencing in affected subjects; (4) the mutation substitutes an amino acid belonging to a different polarity group and can be inferred to affect protein folding as shown in thehMSH2::ADE2 fusion assay.

The I1307K missense mutation in APC has been reported to predispose to CRC. Virtually restricted to the Ashkenazim, it is not associated with classical FAP. We wished to study MON702 for this mutation, because it has been suggested that I1307KAPC may be more frequent in CRC kindreds carrying MMR gene mutations.19 On testing the living members of the kindred for this mutation, we found that I1307KAPC was indeed present in the family, but was carried by those who have neither CRC nor polyps. The proband and her affected son do not carry the mutation. As well, analysis of the sole available archival pathology specimen from the father (fig 1, II.10) suggested that the mutation is coming from his side of the family. There are no confirmed cases of CRC in his 10 sibs. In the MON702 HNPCC kindred, affected A636P hMSH2carriers do not carry the I1307K APCvariant; therefore, our present study does not suggest a biological relationship between mutations in MMR genes and I1307KAPC. Previously, we reported12a French Canadian kindred (MON1061) in which I1307KAPC was segregating with a novel truncating mutation in hMLH1 (1784delT). There was no consistent relationship of the presence or absence of I1307KAPC with CRC, but all tested subjects with early onset CRC were heterozygous for thehMLH1 mutation.

Clearly, further work is required to resolve this issue, including follow up of these families and ascertainment of subjects carrying both a germline mutation in the hMSH2 gene and I1307K APC. Until that time caution should be exercised in deciding to discontinue cancer surveillance of a person in an Ashkenazi Jewish CRC family when MMR gene mutation analysis is negative, even in a family where a germline MMR mutation has been identified, unless status at APC codon 1307 has been determined. Recent studies of Ashkenazi Jewish populations have resulted in odds ratios of under 2 for the risk of CRC associated with the I1307K allele ofAPC.11 ,13-16 Ashkenazi Jews carrying the I1307K allele but who have no family history of CRC have a risk of colon cancer that is only marginally increased, and more evidence that knowledge of I1307K status can prevent CRC is required to support general screening efforts. However, in an Ashkenazi Jewish family with multiple cases of CRC, APCI1307K determination of all subjects who request mutation analysis may be justified.