Identifying primary and secondary MLH1 epimutation carriers displaying low-level constitutional MLH1 methylation using droplet digital PCR and genome-wide DNA methylation profiling of colorectal cancers.

BACKGROUND: MLH1 epimutation is characterised by constitutional monoallelic MLH1 promoter hypermethylation, which can cause colorectal cancer (CRC). Tumour molecular profiles of MLH1 epimutation CRCs were used to classify germline MLH1 promoter variants of uncertain significance and MLH1 methylated early-onset CRCs (EOCRCs). Genome-wide DNA methylation and somatic mutational profiles of tumours from two germline MLH1: c.-11C > T and one MLH1: c.-[28A > G; 7C > T] carriers and three MLH1 methylated EOCRCs (< 45 years) were compared with 38 reference CRCs. Methylation-sensitive droplet digital PCR (ddPCR) was used to detect mosaic MLH1 methylation in blood, normal mucosa and buccal DNA. RESULTS: Genome-wide methylation-based Consensus Clustering identified four clusters where the tumour methylation profiles of germline MLH1: c.-11C > T carriers and MLH1 methylated EOCRCs clustered with the constitutional MLH1 epimutation CRCs but not with the sporadic MLH1 methylated CRCs. Furthermore, monoallelic MLH1 methylation and APC promoter hypermethylation in tumour were observed in both MLH1 epimutation and germline MLH1: c.-11C > T carriers and MLH1 methylated EOCRCs. Mosaic constitutional MLH1 methylation in MLH1: c.-11C > T carriers and 1 of 3 MLH1 methylated EOCRCs was identified by methylation-sensitive ddPCR. CONCLUSIONS: Mosaic MLH1 epimutation underlies the CRC aetiology in MLH1: c.-11C > T germline carriers and a subset of MLH1 methylated EOCRCs. Tumour profiling and ultra-sensitive ddPCR methylation testing can be used to identify mosaic MLH1 epimutation carriers.

TERT structural rearrangements in metastatic pheochromocytomas.

Pheochromocytomas (PC) and paragangliomas (PGL) are endocrine tumors for which the genetic and clinicopathological features of metastatic progression remain incompletely understood. As a result, the risk of metastasis from a primary tumor cannot be predicted. Early diagnosis of individuals at high risk of developing metastases is clinically important and the identification of new biomarkers that are predictive of metastatic potential is of high value. Activation of TERT has been associated with a number of malignant tumors, including PC/PGL. However, the mechanism of TERT activation in the majority of PC/PGL remains unclear. As TERT promoter mutations occur rarely in PC/PGL, we hypothesized that other mechanisms - such as structural variations - may underlie TERT activation in these tumors. From 35 PC and four PGL, we identified three primary PCs that developed metastases with elevated TERT expression, each of which lacked TERT promoter mutations and promoter DNA methylation. Using whole genome sequencing, we identified somatic structural alterations proximal to the TERT locus in two of these tumors. In both tumors, the genomic rearrangements led to the positioning of super-enhancers proximal to the TERT promoter, that are likely responsible for the activation of the normally tightly repressed TERT expression in chromaffin cells.

Assessing alternative base substitutions at primer CpG sites to optimise unbiased PCR amplification of methylated sequences.

BACKGROUND: Determining the role of DNA methylation in various biological processes is dependent on the accurate representation of often highly complex patterns. Accurate representation is dependent on unbiased PCR amplification post bisulfite modification, regardless of methylation status of any given epiallele. This is highly dependent on primer design. Particular difficulties are raised by the analysis of CpG-rich regions, which are the usual regions of interest. Here, it is often difficult or impossible to avoid placing primers in CpG-free regions, particularly if one wants to target a specific part of a CpG-rich region. This can cause biased amplification of methylated sequences if the C is placed at those positions or to unmethylated sequences if a T is placed at those positions. METHODS: We examined the effect of various base substitutions at the cytosine position of primer CpGs on the representational amplification of templates and also examined the role of the annealing temperature during PCR. These were evaluated using methylation-sensitive high-resolution melting and Pyrosequencing. RESULTS: For a mixture of fully methylated and unmethylated templates, amplification using the C-, C/T (Y-) and inosine-containing primers was biased towards amplification of methylated DNA. The bias towards methylated sequences increased with annealing temperature. Amplification using primers with an A/C/G/T (N) degeneracy at the cytosine positions was not biased at the lowest temperature used but became increasingly biased towards methylated DNA with increased annealing temperature. Using primers matching neither C nor T was in the main unbiased but at the cost of poor PCR amplification efficiency. Primers with abasic sites were also unbiased but could only amplify DNA for one out of the two assays tested. However, with heterogeneous methylation, it appeared that both the primer type and stringency used have a minimal influence on PCR bias. CONCLUSIONS: This is the first comprehensive analysis of base substitutions at CpG sites in primers and their effect on PCR bias for the analysis of DNA methylation. Our findings are relevant to the appropriate design of a wide range of assays, including amplicon-based next-generation sequencing approaches that need to measure DNA methylation.

Evaluation of Different Oligonucleotide Base Substitutions at CpG Binding sites in Multiplex Bisulfite-PCR sequencing.

Multiplex bisulfite-PCR sequencing is a convenient and scalable method for the quantitative determination of the methylation state of target DNA regions. A challenge of this application is the presence of CpGs in the same region where primers are being placed. A common solution to the presence of CpGs within a primer-binding region is to substitute a base degeneracy at the cytosine position. However, the efficacy of different substitutions and the extent to which bias towards methylated or unmethylated templates may occur has never been evaluated in bisulfite multiplex sequencing applications. In response, we examined the performance of four different primer substitutions at the cytosine position of CpG's contained within the PCR primers. In this study, deoxyinosine-, 5-nitroindole-, mixed-base primers and primers with an abasic site were evaluated across a series of methylated controls. Primers that contained mixed- or deoxyinosine- base modifications performed most robustly. Mixed-base primers were further selected to determine the conditions that induce bias towards methylated templates. This identified an optimized set of conditions where the methylated state of bisulfite DNA templates can be accurately assessed using mixed-base primers, and expands the scope of bisulfite resequencing assays when working with challenging templates.

MethPat: a tool for the analysis and visualisation of complex methylation patterns obtained by massively parallel sequencing.

BACKGROUND: DNA methylation at a gene promoter region has the potential to regulate gene transcription. Patterns of methylation over multiple CpG sites in a region are often complex and cell type specific, with the region showing multiple allelic patterns in a sample. This complexity is commonly obscured when DNA methylation data is summarised as an average percentage value for each CpG site (or aggregated across CpG sites). True representation of methylation patterns can only be fully characterised by clonal analysis. Deep sequencing provides the ability to investigate clonal DNA methylation patterns in unprecedented detail and scale, enabling the proper characterisation of the heterogeneity of methylation patterns. However, the sheer amount and complexity of sequencing data requires new synoptic approaches to visualise the distribution of allelic patterns. RESULTS: We have developed a new analysis and visualisation software tool "Methpat", that extracts and displays clonal DNA methylation patterns from massively parallel sequencing data aligned using Bismark. Methpat was used to analyse multiplex bisulfite amplicon sequencing on a range of CpG island targets across a panel of human cell lines and primary tissues. Methpat was able to represent the clonal diversity of epialleles analysed at specific gene promoter regions. We also used Methpat to describe epiallelic DNA methylation within the mitochondrial genome. CONCLUSIONS: Methpat can summarise and visualise epiallelic DNA methylation results from targeted amplicon, massively parallel sequencing of bisulfite converted DNA in a compact and interpretable format. Unlike currently available tools, Methpat can visualise the diversity of epiallelic DNA methylation patterns in a sample.

Exemplary multiplex bisulfite amplicon data used to demonstrate the utility of Methpat.

BACKGROUND: DNA methylation is a complex epigenetic marker that can be analyzed using a wide variety of methods. Interpretation and visualization of DNA methylation data can mask complexity in terms of methylation status at each CpG site, cellular heterogeneity of samples and allelic DNA methylation patterns within a given DNA strand. Bisulfite sequencing is considered the gold standard, but visualization of massively parallel sequencing results remains a significant challenge. FINDINGS: We created a program called Methpat that facilitates visualization and interpretation of bisulfite sequencing data generated by massively parallel sequencing. To demonstrate this, we performed multiplex PCR that targeted 48 regions of interest across 86 human samples. The regions selected included known gene promoters associated with cancer, repetitive elements, known imprinted regions and mitochondrial genomic sequences. We interrogated a range of samples including human cell lines, primary tumours and primary tissue samples. Methpat generates two forms of output: a tab-delimited text file for each sample that summarizes DNA methylation patterns and their read counts for each amplicon, and a HTML file that summarizes this data visually. Methpat can be used with publicly available whole genome bisulfite sequencing and reduced representation bisulfite sequencing datasets with sufficient read depths. CONCLUSIONS: Using Methpat, complex DNA methylation data derived from massively parallel sequencing can be summarized and visualized for biological interpretation. By accounting for allelic DNA methylation states and their abundance in a sample, Methpat can unmask the complexity of DNA methylation and yield further biological insight in existing datasets.

Quantitative methodology is critical for assessing DNA methylation and impacts on correlation with patient outcome.

BACKGROUND: DNA hypermethylation is reported as a frequent event and prognostic marker in head and neck squamous cell carcinomas (HNSCC). Methylation has been commonly assessed with non-quantitative methodologies, such as methylation-specific PCR (MSP). We investigated previously reported hypermethylated genes with quantitative methodology in oral tongue squamous cell carcinomas (OTSCC). RESULTS: The methylation status of 12 genes in 115 OTSCC samples was assessed by one or more of three quantitative analyses: methylation sensitive high resolution melting (MS-HRM), sensitive-melting analysis after real time-methylation specific PCR (SMART-MSP), and bisulfite pyrosequencing. In contrast to much of the literature, either no or infrequent locus-specific methylation was identified by MS-HRM for DAPK1, RASSF1A, MGMT, MLH1, APC, CDH1, CDH13, BRCA1, ERCC1, and ATM. The most frequently methylated loci were RUNX3 (18/108 methylated) and ABO (22/107 methylated). Interrogation of the Cancer Genome Atlas (TCGA) HNSCC cohort confirmed the frequency of significant methylation for the loci investigated. Heterogeneous methylation of RUNX3 (18/108) and ABO (22/107) detected by MS-HRM, conferred significantly worse survival (P = 0.01, and P = 0.03). However, following quantification of methylation levels using pyrosequencing, only four tumors had significant quantities (>15%) of RUNX3 methylation which correlated with a worse patient outcome (P <0.001), while the prognostic significance of ABO hypermethylation was lost. RUNX3 methylation was not prognostic for the TCGA cohort (P = 0.76). CONCLUSIONS: We demonstrated the critical need for quantification of methylation levels and its impact on correlative analyses. In OTSCC, we found little evidence of significant or frequent hypermethylation of many loci reported to be commonly methylated. It is likely that previous reports have overestimated the frequency of significant methylation events as a consequence of the use of non-quantitative methodology.

Methylation profiling of normal individuals reveals mosaic promoter methylation of cancer-associated genes.

Epigenetic silencing by promoter methylation of genes associated with cancer initiation and progression is a hallmark of tumour cells. As a consequence, testing for DNA methylation biomarkers in plasma or other body fluids shows great promise for detection of malignancies at early stages and/or for monitoring response to treatment. However, DNA from normal leukocytes may contribute to the DNA in plasma and will affect biomarker specificity if there is any methylation in the leukocytes. DNA from 48 samples of normal peripheral blood mononuclear cells was evaluated for the presence of methylation of a panel of DNA methylation biomarkers that have been implicated in cancer. SMART-MSP, a methylation specific PCR (MSP) methodology based on real time PCR amplification, high-resolution melting and strategic primer design, enabled quantitative detection of low levels of methylated DNA. Methylation was observed in all tested mononuclear cell DNA samples for the CDH1 and HIC1 promoters and in majority of DNA samples for the TWIST1 and DAPK1 promoters. APC and RARB promoter methylation, at a lower average level, was also detected in a substantial proportion of DNA samples. We found no BRCA1, CDKN2A, GSTP1 and RASSF1A promoter methylation in this sample set. Several individuals had higher levels of methylation at several loci suggestive of a methylator phenotype. In conclusion, methylation of many potential DNA methylation biomarkers can be detected in normal peripheral blood mononuclear cells, and is likely to affect their specificity for detecting low level disease. However, we found no evidence of promoter methylation for other genes indicating that panels of analytically sensitive and specific methylation biomarkers in body fluids can be obtained.

Closed-tube PCR methods for locus-specific DNA methylation analysis.

Closed-tube PCR methods (sometimes referred to as in-tube PCR methods) for locus-specific DNA -methylation analysis are methodologies in which the amplification and analysis of bisulphite-modified DNA take place in one tube without the need to remove the PCR products for further analysis. Closed-tube methodologies lend themselves to high-throughput applications and molecular diagnostics but are also applicable as a research tool. We review three closed-tube methodologies, methylation-sensitive high-resolution melting (MS-HRM), MethyLight, and sensitive melting after real-time analysis - methylation-specific PCR (SMART-MSP). Closed-tube detection can be performed by simultaneously amplifying both methylated and unmethylated templates and subsequent melting curve analysis (MS-HRM). Alternatively, methylation-specific primers are used in real-time quantitative PCR and monitored either by a fluorescent hydrolysis probe (MethyLight) or using a double-stranded DNA binding fluorescent dye with a subsequent quality control step by melting curve analysis (SMART-MSP).

Assessing combined methylation-sensitive high resolution melting and pyrosequencing for the analysis of heterogeneous DNA methylation.

Heterogeneous DNA methylation leads to difficulties in accurate detection and quantification of methylation. Methylation-sensitive high resolution melting (MS-HRM) is unique among regularly used methods for DNA methylation analysis in that heterogeneous methylation can be readily identified, although not quantified, by inspection of the melting curves. Bisulfite pyrosequencing has been used to estimate the level of heterogeneous methylation by quantifying methylation levels present at individual CpG dinucleotides. Sequentially combining the two methodologies using MS-HRM to screen the amplification products prior to bisulfite pyrosequencing would be advantageous. This would not only replace the quality control step using agarose gel analysis prior to the pyrosequencing step but would also provide important qualitative information in its own right. We chose to analyze DAPK1 as it is an important tumor suppressor gene frequently heterogeneously methylated in a number of malignancies, including chronic lymphocytic leukemia (CLL). A region of the DAPK1 promoter was analyzed in ten CLL samples by MS-HRM. By using a biotinylated primer, bisulfite pyrosequencing could be used to directly analyze the samples. MS-HRM revealed the presence of various extents of heterogeneous DAPK1 methylation in all CLL samples. Further analysis of the biotinylated MS-HRM products by bisulfite pyrosequencing provided quantitative information for each CpG dinucleotide analyzed, and confirmed the presence of heterogeneous DNA methylation. Whereas each method could be used individually, MS-HRM and bisulfite pyrosequencing provided complementary information for the assessment of heterogeneous methylation.

The implications of heterogeneous DNA methylation for the accurate quantification of methylation.

DNA methylation based biomarkers have considerable potential for molecular diagnostics, both as tumor specific biomarkers for the early detection or post-therapeutic monitoring of cancer as well as prognostic and predictive biomarkers for therapeutic stratification. Particularly in the former, the accurate estimation of DNA methylation is of compelling importance. However, quantification of DNA methylation has many traps for the unwary, especially when heterogeneous methylation comprising multiple alleles with varied DNA methylation patterns (epialleles) is present. The frequent occurrence of heterogeneous methylation as distinct from a simple mixture of fully methylated and unmethylated alleles is generally not taken into account when DNA methylation is considered as a cancer biomarker. When heterogeneous DNA methylation is present, the proportion of methylated molecules is difficult to quantify without a method that allows the measurement of individual epialleles. In this article, we critically assess the methodologies frequently used to investigate DNA methylation, with an emphasis on the detection and measurement of heterogeneous DNA methylation. The adoption of digital approaches will enable the effective use of heterogeneous DNA methylation as a cancer biomarker.

Detection of MGMT promoter methylation in normal individuals is strongly associated with the T allele of the rs16906252 MGMT promoter single nucleotide polymorphism.

Methylation of the CpG island in the MGMT promoter region is a frequent event in several cancer types including colorectal cancer, lung cancer, lymphoma, and glioblastoma. A correlation between methylation and the T allele of the rs16906252 single nucleotide polymorphism (SNP) in colorectal carcinomas has previously been reported. As aberrant MGMT methylation can be an early event in tumor development, we tested the hypothesis that normal individuals possessing the T allele may be predisposed to somatic methylation at the MGMT promoter. Peripheral blood monononuclear cell DNA from 89 normal, healthy individuals was genotyped at rs1690625 and assessed for the methylation status of the MGMT promoter region using independent quantitative methodologies capable of detecting low-level methylation: MethyLight and Sensitive Melting Analysis after Real-time Methylation-Specific PCR (SMART-MSP). There was a strong association between presence of the T allele and detectable methylation (P = 0.00005) in the peripheral blood DNA. Furthermore, when a MSP assay flanking the SNP was used to amplify methylated sequences in heterozygotes, only the T allele was methylated. Thus, detectable somatic methylation of the MGMT promoter in normal individuals is strongly associated with the T allele of the rs16906252 MGMT promoter SNP.

Rapid analysis of heterogeneously methylated DNA using digital methylation-sensitive high resolution melting: application to the CDKN2B (p15) gene.

BACKGROUND: Methylation-sensitive high resolution melting (MS-HRM) methodology is able to recognise heterogeneously methylated sequences by their characteristic melting profiles. To further analyse heterogeneously methylated sequences, we adopted a digital approach to MS-HRM (dMS-HRM) that involves the amplification of single templates after limiting dilution to quantify and to determine the degree of methylation. We used this approach to study methylation of the CDKN2B (p15) cell cycle progression inhibitor gene which is inactivated by DNA methylation in haematological malignancies of the myeloid lineage. Its promoter region usually shows heterogeneous methylation and is only rarely fully methylated. The methylation status of CDKN2B can be used as a biomarker of response to treatment. Therefore the accurate characterisation of its methylation is desirable. RESULTS: MS-HRM was used to assess CDKN2B methylation in acute myeloid leukaemia (AML) samples. All the AML samples that were methylated at the CDKN2B promoter (40/93) showed varying degrees of heterogeneous methylation. Six representative samples were selected for further study. dMS-HRM was used to simultaneously count the methylated alleles and assess the degree of methylation. Direct sequencing of selected dMS-HRM products was used to determine the exact DNA methylation pattern and confirmed the degree of methylation estimated by dMS-HRM. CONCLUSION: dMS-HRM is a powerful technique for the analysis of methylation in CDKN2B and other heterogeneously methylated genes. It eliminates both PCR and cloning bias towards either methylated or unmethylated DNA. Potentially complex information is simplified into a digital output, allowing counting of methylated and unmethylated alleles and providing an overall picture of methylation at the given locus. Downstream sequencing is minimised as dMS-HRM acts as a screen to select only methylated clones for further analysis.