Investigation of the pituitary epigenome: a genome-wide analysis of changes associated with sporadic tumours

Pituitary tumours harbour epigenetic aberrations; however, characterisation of these aberrations, on a genome-wide basis, is hampered by their infrequent occurrence and their small size. To overcome the constraint of limited tissue, whole genome amplification of sodium bisulphite converted DNA was employed and provided a consistent 25-fold amplification from individual samples. This material was used in a genome-wide analysis the DNA methylation of 27,578 CpG sites for each of the major adenoma subtypes. In a discovery cohort, on the basis of stringent criteria, pyrosequencing validated 12 of 16 hypermethylated genes. Overall, the criteria identified 40 genes in non-functional, 21 in growth hormone, six in prolactin and two in corticotrophinoma. In an independent cohort, different frequencies of hypermethylation were apparent for each of these genes; however, association between methylation and reduced transcript expression was infrequent.\ud \ud For the EFEMP1 gene, following its initial identification, studies of an independent cohort of tumours showed frequent reduced EFEMP1 expression, irrespective of adenoma subtype. However, reduced expression was not invariantly associated CpG island methylation. Conversely, chromatin immunoprecipitation assays (ChIP) showed histone modifications that were consonant with expression status. The causal relationship between gene silencing and epigenetic change was established by observing that epidrug challenges induced re-expression of EFEMP1 in pituitary cells that was concomitant with histone modification associated with expressed genes.\ud \ud Enforced expression of EFEMP1 was without effect on cell proliferation or apoptotic end-points but was responsible for decreased expression of the MMP2 transcript. This association was not apparent in primary adenomas, however, MMP7, showed a positive correlation with EFEMP1 and this may reflect cell or species specific differences, suggesting that the relationship between EFEMP1 and MMP7 requires more detailed investigation. This study is the first whole genome identification of a potential biomarker signature and their functional characterisation will provide insight of tumour aetiology and identification of new therapeutic targets