Reduced nuclear DNA methylation and mitochondrial transcript changes in adenomas do not associate with mtDNA methylation.

BACKGROUND: Altered mitochondrial function and large-scale changes to DNA methylation patterns in the nuclear genome are both hallmarks of colorectal cancer (CRC). Mitochondria have multiple copies of a 16 kb circular genome that contains genes that are vital for their function. While DNA methylation is known to alter the nuclear genome in CRC, it is not clear whether it could have a similar influence in mtDNA; indeed, currently, the issue of whether mitochondrial genome (mtDNA) methylation occurs is controversial. Thus our goal here was to determine whether the methylation state of mtDNA is linked to mitochondrial gene transcription in colorectal adenomas, and to assess its suitability as a biomarker in CRC. METHODS: To investigate the relationship between DNA methylation and mitochondrial transcripts in adenomas, we performed RNA-sequencing and Whole Genome Bisulphite Sequencing (WGBS) of mtDNA-enriched DNA from normal mucosa and paired adenoma patient samples. RESULTS: Transcriptional profiling indicated that adenomas had reduced mitochondrial proton transport versus normal mucosa, consistent with altered mitochondrial function. The expression of 3 tRNAs that are transcribed from mtDNA were also decreased in adenoma. Overall methylation of CG dinucleotides in the nuclear genome was reduced in adenomas (68%) compared to normal mucosa (75%, P < 0.01). Methylation in mtDNA was low (1%) in both normal and adenoma tissue but we observed clusters of higher methylation at the ribosomal RNA genes. Levels of methylation within these regions did not differ between normal and adenoma tissue. CONCLUSIONS: We provide evidence that low-level methylation of specific sites does exist in the mitochondrial genome but that it is not associated with mitochondrial gene transcription changes in adenomas. Furthermore, as no large scale changes to mtDNA methylation were observed it is unlikely to be a suitable biomarker for early-stage CRC.

AZA-MS: a novel multiparameter mass spectrometry method to determine the intracellular dynamics of azacitidine therapy in vivo.

The cytidine analogue, 5-azacytidine (AZA; 5-AZA-cR), is the primary treatment for myelodysplastic syndrome and chronic myelomonocytic leukaemia. However, only ~50% of treated patients will respond to AZA and the drivers of AZA resistance in vivo are poorly understood. To better understand the intracellular dynamics of AZA upon therapy and decipher the molecular basis for AZA resistance, we have developed a novel, multiparameter, quantitative mass spectrometry method (AZA-MS). Using AZA-MS, we have accurately quantified the abundance of the ribonucleoside (5-AZA-cR) and deoxyribonucleoside (5-AZA-CdR) forms of AZA in RNA, DNA and the cytoplasm within the same sample using nanogram quantities of input material. We report that although AZA induces DNA demethylation in a dose-dependent manner, it has no corresponding effect on RNA methylation. By applying AZA-MS to primary bone marrow samples from patients undergoing AZA therapy, we have identified that responders accumulate more 5-AZA-CdR in their DNA compared with nonresponders. AZA resistance was not a result of impaired AZA metabolism or intracellular accumulation. Furthermore, AZA-MS has helped to uncover different modes of AZA resistance. Whereas some nonresponders fail to incorporate sufficient 5-AZA-CdR into DNA, others incorporate 5-AZA-CdR and effect DNA demethylation like AZA responders, but show no clinical benefit.

RASSF2 associates with and stabilizes the proapoptotic kinase MST2.

RASSF2 is a tumour suppressor that in common with the rest of the RASSF family contains Ras association and SARAH domains. We identified the proapoptotic kinases, MST1 and MST2, as the most significant binding partners of RASSF2, confirmed the interactions at endogenous levels and showed that RASSF2 immunoprecipitates active MST1/2. We then showed that RASSF2 can be phosphorylated by a co-immunoprecipitating kinase that is likely to be MST1/2. Furthermore, we showed that RASSF2 and MST2 do indeed colocalize, but whereas RASSF2 alone is nuclear, the presence of MST1 or MST2 results in colocalization in the cytoplasm. Expression of RASSF2 (stably in MCF7 or transiently in HEK-293) increases MST2 levels and knockdown of RASSF2 in HEK-293 cells reduces MST2 levels, in addition colorectal tumour cell lines and primary tumours with low RASSF2 levels show decreased MST2 protein levels. This is likely to be mediated by RASSF2-dependent protection of MST2 against proteolytic degradation. Our findings suggest that MST2 and RASSF2 form an active complex in vivo, in which RASSF2 is maintained in a phosphorylated state and protects MST2 from degradation and turnover. Thus, we propose that the frequent loss of RASSF2 in tumours results in the destablization of MST2 and thus decreased apoptotic potential.

Epigenetic regulation of the ras effector/tumour suppressor RASSF2 in breast and lung cancer.

RASSF2 is a recently identified member of a class of novel tumour suppressor genes, all containing a ras-association domain. RASSF2 resides at 20p13, a region frequently lost in human cancers. In this report we investigated methylation status of the RASSF2 promoter CpG island in a series of breast, ovarian and non-small cell lung cancers (NSCLC). RASSF2 was frequently methylated in breast tumour cell lines (65%, 13/20) and in primary breast tumours (38%, 15/40). RASSF2 expression could be switched back on in methylated breast tumour cell lines after treatment with 5'-aza-2'deoxycytidine. RASSF2 was also frequently methylated in NSCLC tumours (44%, (22/50). The small number of corresponding normal breast and lung tissue DNA samples analysed were unmethylated. We also did not detect RASSF2 methylation in ovarian tumours (0/17). Furthermore no mutations were found in the coding region of RASSF2 in these ovarian tumours. We identified a highly conserved putative bipartite nuclear localization signal (NLS) and demonstrated that endogenous RASSF2 localized to the nucleus. Mutation of the putative NLS abolished the nuclear localization. RASSF2 suppressed breast tumour cell growth in vitro and in vivo, while the ability of NLS-mutant RASSF2 to suppress growth was much diminished. Hence we demonstrate that RASSF2 has a functional NLS that is important for its tumour suppressor gene function. Our data from this and a previous report indicate that RASSF2 is frequently methylated in colorectal, breast and NSCLC tumours. We have identified RASSF2 as a novel methylation marker for multiple malignancies and it has the potential to be developed into a valuable marker for screening several cancers in parallel using promoter hypermethylation profiles.

Evaluation of the 3p21.3 tumour-suppressor gene cluster.

Deletions of the 3p21.3 region are a frequent and early event in the formation of lung, breast, kidney and other cancers. Intense investigation of allelic losses and the discovery of overlapping homozygous deletions in lung and breast tumour-cell lines have defined a minimal critical 120 kb deletion region containing eight genes and likely to harbor one or more tumour-suppressor genes (TSGs). The candidate genes are HYAL2, FUS1, Ras-associated factor 1 (RASSF1), BLU/ZMYND10, NPR2L, 101F6, PL6 and CACNA2D2. Recent research indicates that several of these genes can suppress the growth of lung and other tumour cells. Furthermore, some genes (RASSF1A and BLU/ZMYND10) are very frequently inactivated by non-classical mechanisms such as promoter hypermethylation resulting in loss of expression. These data indicate that the 120 kb critical deletion region at 3p21.3 may represent a TSG cluster with preferential inactivation of particular genes depending on tumour type. The eight genes within this region and their potential role in cancer will be the focus of this review.

Analysis of the Birt-Hogg-Dubé (BHD) tumour suppressor gene in sporadic renal cell carcinoma and colorectal cancer.

Germline mutations in the BHD gene cause the dominantly inherited cancer susceptibility disorder, Birt-Hogg-Dubé (BHD) syndrome. Individuals with BHD are reported to have an increased risk of renal cell carcinoma (RCC) and of colorectal polyps and cancer. The BHD gene maps to 17p11.2, and to investigate whether somatic inactivation of the BHD gene region is implicated in the pathogenesis of sporadic RCC and colorectal cancer (CRC), we performed mutation analysis in 30 RCC primary tumours and cell lines, and 35 CRCs and cell lines. A somatic missense mutation (Ala444Ser) with loss of the wild type allele (consistent with a two hit mechanism of tumorigenesis) was detected in a primary clear cell RCC, and a further missense mutation (Ala238Val) was identified in a clear cell RCC cell line for which matched normal DNA was not available. A somatic missense substitution (Arg392Gly) was identified in a primary CRC, and the same change was detected in three RCCs (all oncocytomas) for which matched normal DNA was not available. A germline Arg320Gln missense variant detected in a primary CRC was not detected in 40 control individuals or in a further 159 familial and sporadic CRC cases. However, AA homozygotes for an intronic single nucleotide polymorphism (c.1517+6 G-->A) were under-represented in familial cases compared with controls (p = 0.03). For some tumour suppressor genes, epigenetic silencing is a more common mechanism of inactivation than somatic mutations. However, we did not detect evidence of epigenetic silencing of BHD in 19 CRC and RCC cell lines, and BHD promoter region hypermethylation was not detected in 20 primary RCCs. These findings suggest that BHD inactivation occurs in a subset of clear cell RCC and CRC.