Germline epimutations of APC are not associated with inherited colorectal polyposis

We have recently reported that germline epimutations of the mismatch repair gene, MLH1, produce a clinical phenotype consistent with hereditary non‐polyposis colorectal cancer (HNPCC).¹MLH1 germline epimutations have now been reported in eight individuals clinically classified as having HNPCC and in whom no sequence mutations of the mismatch repair genes were identified.¹,²,³,⁴ These manifest as soma‐wide monoallelic hypermethylation of the promoter (in the absence of a sequence mutation within the gene) and cause transcriptional silencing of the epimutant allele.¹ These cases provide evidence that germline epimutations represent a novel mechanism for disease, which can mimic that of a genetic mutation by conferring a similar phenotype. We hypothesised that germline epimutations may occur in genes other than MLH1 to cause other cancer prone syndromes apart from HNPCC. The APC gene represented a likely candidate as heterozygous inactivating mutations of APC are strongly associated with colorectal polyposis and cancer, but not all individuals harbour germline sequence mutations. Autosomal dominant familial adenomatous polyposis (FAP) is caused by heterozygous sequence mutations of the APC gene,⁵ and a recessive form is associated with biallelic mutations of MYH.⁶ However, no mutations of either gene are identifiable in approximately 20% of cases. Furthermore, somatic methylation of the CpG island encompassing the APC‐1A promoter is observed in colorectal carcinomas that do not express APC, indicating that this promoter is susceptible to epigenetic inactivation.⁷,⁸,⁹

To determine whether germline epimutation of the APC gene contributes to the development of the polyposis syndromes, we screened probands with a clinical diagnosis of FAP, attenuated FAP, or hyperplastic polyposis (HPS), for methylation of the APC‐1A promoter. Methylation of the major 1A promoter is observed in up to 20% of colorectal carcinomas, but no methylation changes have been detected at the 1B promoter from which a minor transcript is derived, and so this latter promoter was not analysed.⁷,⁸,⁹ With the approval of the Human Research and Ethics Committee of St Vincent's Hospital, this study enrolled 140 individuals with colonic polyps from family cancer clinics in Australia and the UK. None of the individuals harboured pathogenic germline sequence mutations in the APC gene, and biallelic MYH mutations had been excluded in all but 10 cases. Of the 140 individuals, 22 were diagnosed with HPS (mean age 60.5 (9.6) years), 29 with FAP (mean age 27.8 (10.5) years), and 89 with attenuated FAP (mean age 59 (18.6) years).

Probands were screened for hypermethylation of the APC‐1A promoter in peripheral blood DNA using the method of combined bisulfite and restriction analysis. Genomic DNA was converted by standard sodium bisulfite treatment and subject to polymerase chain reaction (PCR) amplification using primers 5′‐GGG TTA GGG TTA GGT AGG TTG‐3′ and 5′‐ACA CCT CCA TTC TAT CTC CAA TAA C‐3′, which are specific to bisulfite converted DNA, and co‐amplify both methylated and unmethylated alleles with equal efficiency (fig 1A). PCR amplification was performed using FastStart Taq Polymerase (Roche Diagnostics, New South Wales, Australia) with a final concentration of 3 mM MgCl₂, and annealing at 55°C. The amplification products were digested separately with the restriction enzymes TaqI and RsaI, which specifically cleave amplicons derived from templates that are methylated at the corresponding CpG dinucleotides, whereas the unmethylated promoter remains undigested (fig 1). To serve as a positive control, genomic DNA was methylated in vitro using CpG M.SssI methylase (New England Biolabs, Ipswich, Massachusetts, USA). Genomic DNA from a normal control subject was used as a negative control.

Figure 1 Methylation analysis of the APC promoter. (A) Map of the APC‐1A promoter. The transcription start site of APC exon 1 (grey box) is at nucleotide +1. The position of the primers used to amplify a 229 bp fragment of the promoter upstream of exon 1 is shown by arrows. Individual CpG dinucleotides are denoted by vertical black bars. The diagnostic restriction enzyme sites that cleave specific CpG sites when methylated are indicated. (B) Gel showing an example of combined bisulfite and restriction analysis of APC‐1A promoter. L, pUC19/MspI DNA ladder; C+, in vitro methylated genomic DNA shows cleavage of the methylated amplicons with RsaI to give a digestion pattern of 162 and 67 bp (and 141, 60, and 28 bp for TaqI, not shown); C−, unmethylated genomic DNA from a normal control individual remains undigested. Wells labelled 1–9 contain peripheral blood DNA from nine familial adenomatous polyposis subjects, each negative for APC‐1A methylation.

No methylation of the APC‐1A promoter was identified in the constitutional DNA of any of the probands studied, indicating germline epimutations of APC do not underlie the phenotype in these individuals. Our results suggest that germline epimutation of APC is not a significant factor in the polyposis syndromes studied, and does not account for cases with no detectable sequence mutation of the APC or MYH genes. This study adds further support to the concept that germline changes in APC are not responsible for HPS. However, germline epimutations remain a possibility in polyposis cases where there is a marked allelic imbalance in APC transcription. Yan et al reported an FAP family in which there was considerably reduced expression of one specific allele in each of the family members carrying this haplotype.¹⁰ No sequence mutation within the APC gene, including the promoter and 3′ untranslated region, was identified despite intensive screening. However, the methylation status of the APC‐1A promoter was not investigated in this family. It remains possible that some families and isolated cases such as this may have germline epimutations of APC, causing the affected allele to be silenced.

Although somatic APC promoter methylation occurs in a spectrum of cancers, unlike the association of microsatellite instability with MLH1 methylation, it does not give rise to an obvious phenotype. Furthermore, while germline APC mutations are well known as causing multiple adenomatous polyps and increased cancer risk, it is likely that germline defects in other genes can produce a similar phenotype. In APC mutation negative cases that produce a florid syndrome in the absence of a strong family history, it may still be worth investigating epigenetic mechanisms of gene inactivation as well as genetic causes.