Effective DNA/RNA co-extraction for analysis of microRNAs, mRNAs, and genomic DNA from formalin-fixed paraffin-embedded specimens

Background

Retrospective studies of archived human specimens, with known clinical follow-up, are used to identify predictive and prognostic molecular markers of disease. Due to biochemical differences, however, formalin-fixed paraffin-embedded (FFPE) DNA and RNA have generally been extracted separately from either different tissue sections or from the same section by dividing the digested tissue. The former limits accurate correlation whilst the latter is impractical when utilizing rare or limited archived specimens.

Principal Findings

For effective recovery of genomic DNA and total RNA from a single FFPE specimen, without splitting the proteinase-K digested tissue solution, we optimized a co-extraction method by using TRIzol and purifying DNA from the lower aqueous and RNA from the upper organic phases. Using a series of seven different archived specimens, we evaluated the total amounts of genomic DNA and total RNA recovered by our TRIzol-based co-extraction method and compared our results with those from two commercial kits, the Qiagen AllPrep DNA/RNA FFPE kit, for co-extraction, and the Ambion RecoverAll™ Total Nucleic Acid Isolation kit, for separate extraction of FFPE-DNA and -RNA. Then, to accurately assess the quality of DNA and RNA co-extracted from a single FFPE specimen, we used qRT-PCR, gene expression profiling and methylation assays to analyze microRNAs, mRNAs, and genomic DNA recovered from matched fresh and FFPE MCF10A cells. These experiments show that the TRIzol-based co-extraction method provides larger amounts of FFPE-DNA and –RNA than the two other methods, and particularly provides higher quality microRNAs and genomic DNA for subsequent molecular analyses.

Significance

We determined that co-extraction of genomic DNA and total RNA from a single FFPE specimen is an effective recovery approach to obtain high-quality material for parallel molecular and high-throughput analyses. Our optimized approach provides the option of collecting DNA, which would otherwise be discarded or degraded, for additional or subsequent studies.     

Methylation-capture and Next-Generation Sequencing of free circulating DNA from human plasma

Background

Free circulating DNA (fcDNA) has many potential clinical applications, due to the non-invasive way in which it is collected. However, because of the low concentration of fcDNA in blood, genome-wide analysis carries many technical challenges that must be overcome before fcDNA studies can reach their full potential. There are currently no definitive standards for fcDNA collection, processing and whole-genome sequencing. We report novel detailed methodology for the capture of high-quality methylated fcDNA, library preparation and downstream genome-wide Next-Generation Sequencing. We also describe the effects of sample storage, processing and scaling on fcDNA recovery and quality.

Results

Use of serum versus plasma, and storage of blood prior to separation resulted in genomic DNA contamination, likely due to leukocyte lysis. Methylated fcDNA fragments were isolated from 5 donors using a methyl-binding protein-based protocol and appear as a discrete band of ~180 bases. This discrete band allows minimal sample loss at the size restriction step in library preparation for Next-Generation Sequencing, allowing for high-quality sequencing from minimal amounts of fcDNA. Following sequencing, we obtained 37×10⁶-86×10⁶ unique mappable reads, representing more than 50% of total mappable reads. The methylation status of 9 genomic regions as determined by DNA capture and sequencing was independently validated by clonal bisulphite sequencing.

Conclusions

Our optimized methods provide high-quality methylated fcDNA suitable for whole-genome sequencing, and allow good library complexity and accurate sequencing, despite using less than half of the recommended minimum input DNA.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2164-15-476) contains supplementary material, which is available to authorized users.     

Genome-Wide Massively Parallel Sequencing of Formaldehyde Fixed-Paraffin Embedded (FFPE) Tumor Tissues for Copy-Number- and Mutation-Analysis

Background

Cancer re-sequencing programs rely on DNA isolated from fresh snap frozen tissues, the preparation of which is combined with additional preservation efforts. Tissue samples at pathology departments are routinely stored as formalin-fixed and paraffin-embedded (FFPE) samples and their use would open up access to a variety of clinical trials. However, FFPE preparation is incompatible with many down-stream molecular biology techniques such as PCR based amplification methods and gene expression studies.

Methodology/Principal Findings

Here we investigated the sample quality requirements of FFPE tissues for massively parallel short-read sequencing approaches. We evaluated key variables of pre-fixation, fixation related and post-fixation processes that occur in routine medical service (e.g. degree of autolysis, duration of fixation and of storage). We also investigated the influence of tissue storage time on sequencing quality by using material that was up to 18 years old. Finally, we analyzed normal and tumor breast tissues using the Sequencing by Synthesis technique (Illumina Genome Analyzer, Solexa) to simultaneously localize genome-wide copy number alterations and to detect genomic variations such as substitutions and point-deletions and/or insertions in FFPE tissue samples.

Conclusions/Significance

The application of second generation sequencing techniques on small amounts of FFPE material opens up the possibility to analyze tissue samples which have been collected during routine clinical work as well as in the context of clinical trials. This is in particular important since FFPE samples are amply available from surgical tumor resections and histopathological diagnosis, and comprise tissue from precursor lesions, primary tumors, lymphogenic and/or hematogenic metastases. Large-scale studies using this tissue material will result in a better prediction of the prognosis of cancer patients and the early identification of patients which will respond to therapy.     

Critical comparison of sample preparation strategies for shotgun proteomic analysis of formalin-fixed,paraffin-embedded samples: insights from liver tissue

Background

The growing field of formalin-fixed paraffin-embedded (FFPE) tissue proteomics holds promise for improving translational research. Direct tissue trypsinization (DT) and protein extraction followed by in solution digestion (ISD) or filter-aided sample preparation (FASP) are the most common workflows for shotgun analysis of FFPE samples, but a critical comparison of the different methods is currently lacking.

Experimental design

DT, FASP and ISD workflows were compared by subjecting to the same label-free quantitative approach three independent technical replicates of each method applied to FFPE liver tissue. Data were evaluated in terms of method reproducibility and protein/peptide distribution according to localization, MW, pI and hydrophobicity.

Results

DT showed lower reproducibility, good preservation of high-MW proteins, a general bias towards hydrophilic and acidic proteins, much lower keratin contamination, as well as higher abundance of non-tryptic peptides. Conversely, FASP and ISD proteomes were depleted in high-MW proteins and enriched in hydrophobic and membrane proteins; FASP provided higher identification yields, while ISD exhibited higher reproducibility.

Conclusions

These results highlight that diverse sample preparation strategies provide significantly different proteomic information, and present typical biases that should be taken into account when dealing with FFPE samples. When a sufficient amount of tissue is available, the complementary use of different methods is suggested to increase proteome coverage and depth.     

Elevated pressure improves the extraction and identification of proteins recovered from formalin-fixed, paraffin-embedded tissue surrogates

Background

Proteomic studies of formalin-fixed paraffin-embedded (FFPE) tissues are frustrated by the inability to extract proteins from archival tissue in a form suitable for analysis by 2-D gel electrophoresis or mass spectrometry. This inability arises from the difficulty of reversing formaldehyde-induced protein adducts and cross-links within FFPE tissues. We previously reported the use of elevated hydrostatic pressure as a method for efficient protein recovery from a hen egg-white lysozyme tissue surrogate, a model system developed to study formalin fixation and histochemical processing.

Principal Findings

In this study, we demonstrate the utility of elevated hydrostatic pressure as a method for efficient protein recovery from FFPE mouse liver tissue and a complex multi-protein FFPE tissue surrogate comprised of hen egg-white lysozyme, bovine carbonic anhydrase, bovine ribonuclease A, bovine serum albumin, and equine myoglobin (55∶15∶15∶10∶5 wt%). Mass spectrometry of the FFPE tissue surrogates retrieved under elevated pressure showed that both the low and high-abundance proteins were identified with sequence coverage comparable to that of the surrogate mixture prior to formaldehyde treatment. In contrast, non-pressure-extracted tissue surrogate samples yielded few positive and many false peptide identifications. Studies with soluble formalin-treated bovine ribonuclease A demonstrated that pressure modestly inhibited the rate of reversal (hydrolysis) of formaldehyde-induced protein cross-links. Dynamic light scattering studies suggest that elevated hydrostatic pressure and heat facilitate the recovery of proteins free of formaldehyde adducts and cross-links by promoting protein unfolding and hydration with a concomitant reduction in the average size of the protein aggregates.

Conclusions

These studies demonstrate that elevated hydrostatic pressure treatment is a promising approach for improving the recovery of proteins from FFPE tissues in a form suitable for proteomic analysis.     

Pre-Diagnostic Leukocyte Genomic DNA Methylation and the Risk of Colorectal Cancer in Women

Background

Abnormal one-carbon metabolism may lead to general genomic (global) hypomethylation, which may predispose an individual to the development of colorectal neoplasia.

Methods

We evaluated the association between pre-diagnostic leukocyte genomic DNA methylation level and the risk of colorectal cancer in a nested case-control study of 358 colorectal cancer cases and 661 matched controls within the all-female cohort of the Nurses’ Health Study (NHS). Among control subjects, we further examined major plasma components in the one-carbon metabolism pathway in relation to genomic DNA methylation level. Liquid chromatography/tandem mass spectrometry was used to examine leukocyte genomic DNA methylation level. We calculated odds ratios (ORs) and 95% confidence intervals (95% CIs) using logistic regression.

Results

Overall genomic DNA methylation level was not associated with the risk of colorectal cancer (p for trend, 0.45). Compared with women in the lowest quintile of methylation, the multivariate OR of colorectal cancer risk was 1.32 (95% CI, 0.82–2.13) for those in the highest quintile. We did not find significant associations between major plasma components of one-carbon metabolism or risk factors for colorectal cancer and genomic DNA methylation level (all p for trend >0.05). Also, neither one-carbon metabolism-related plasma components nor well-known risk factors for colorectal cancer modified the association between genomic DNA methylation level and the risk of colorectal cancer (all p for interaction >0.05).

Conclusions

We found no evidence that hypomethylation of leukocyte genomic DNA increases risk of colorectal cancer among women. Additional studies are needed to investigate the association between pre-diagnostic genomic DNA methylation level and colorectal cancer risk among diverse populations.     

Deciphering the heterogeneity in DNA methylation patterns during stem cell differentiation and reprogramming

Background

Human induced pluripotent stem cells (iPSCs) have a wide range of applications throughout the fields of basic research, disease modeling and drug screening. Epigenetic instable iPSCs with aberrant DNA methylation may divide and differentiate into cancer cells. Unfortunately, little effort has been taken to compare the epigenetic variation in iPSCs with that in differentiated cells. Here, we developed an analytical procedure to decipher the DNA methylation heterogeneity of mixed cells and further exploited it to quantitatively assess the DNA methylation variation in the methylomes of adipose-derived stem cells (ADS), mature adipocytes differentiated from ADS cells (ADS-adipose) and iPSCs reprogrammed from ADS cells (ADS-iPSCs).

Results

We observed that the degree of DNA methylation variation varies across distinct genomic regions with promoter and 5’UTR regions exhibiting low methylation variation and Satellite showing high methylation variation. Compared with differentiated cells, ADS-iPSCs possess globally decreased methylation variation, in particular in repetitive elements. Interestingly, DNA methylation variation decreases in promoter regions during differentiation but increases during reprogramming. Methylation variation in promoter regions is negatively correlated with gene expression. In addition, genes showing a bipolar methylation pattern, with both completely methylated and completely unmethylated reads, are related to the carbohydrate metabolic process, cellular development, cellular growth, proliferation, etc.

Conclusions

This study delivers a way to detect cell-subset specific methylation genes in a mixed cell population and provides a better understanding of methylation dynamics during stem cell differentiation and reprogramming.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2164-15-978) contains supplementary material, which is available to authorized users.     

HPV genotypes in high grade cervical lesions and invasive cervical carcinoma as detected by two commercial DNA assays, North Carolina, 2001-2006

Background

HPV typing using formalin fixed paraffin embedded (FFPE) cervical tissue is used to evaluate HPV vaccine impact, but DNA yield and quality in FFPE specimens can negatively affect test results. This study aimed to evaluate 2 commercial assays for HPV detection and typing using FFPE cervical specimens.

Methods

Four large North Carolina pathology laboratories provided FFPE specimens from 299 women ages18 and older diagnosed with cervical disease from 2001 to 2006. For each woman, one diagnostic block was selected and unstained serial sections were prepared for DNA typing. Extracts from samples with residual lesion were used to detect and type HPV using parallel and serial testing algorithms with the Linear Array and LiPA HPV genotyping assays.

Findings

LA and LiPA concordance was 0.61 for detecting any high-risk (HR) and 0.20 for detecting any low-risk (LR) types, with significant differences in marginal proportions for HPV16, 51, 52, and any HR types. Discordant results were most often LiPA-positive, LA-negative. The parallel algorithm yielded the highest prevalence of any HPV type (95.7%). HR type prevalence was similar using parallel (93.1%) and serial (92.1%) approaches. HPV16, 33, and 52 prevalence was slightly lower using the serial algorithm, but the median number of HR types per woman (1) did not differ by algorithm. Using the serial algorithm, HPV DNA was detected in >85% of invasive and >95% of pre-invasive lesions. The most common type was HPV16, followed by 52, 18, 31, 33, and 35; HPV16/18 was detected in 56.5% of specimens. Multiple HPV types were more common in lower grade lesions.

Conclusions

We developed an efficient algorithm for testing and reporting results of two commercial assays for HPV detection and typing in FFPE specimens, and describe HPV type distribution in pre-invasive and invasive cervical lesions in a state-based sample prior to HPV vaccine introduction.     

High throughput interrogation of somatic mutations in high grade serous cancer of the ovary

Background

Epithelial ovarian cancer is the most lethal of all gynecologic malignancies, and high grade serous ovarian cancer (HGSC) is the most common subtype of ovarian cancer. The objective of this study was to determine the frequency and types of point somatic mutations in HGSC using a mutation detection protocol called OncoMap that employs mass spectrometric-based genotyping technology.

Methodology/Principal Findings

The Center for Cancer Genome Discovery (CCGD) Program at the Dana-Farber Cancer Institute (DFCI) has adapted a high-throughput genotyping platform to determine the mutation status of a large panel of known cancer genes. The mutation detection protocol, termed OncoMap has been expanded to detect more than 1000 mutations in 112 oncogenes in formalin-fixed paraffin-embedded (FFPE) tissue samples. We performed OncoMap on a set of 203 FFPE advanced staged HGSC specimens. We isolated genomic DNA from these samples, and after a battery of quality assurance tests, ran each of these samples on the OncoMap v3 platform. 56% (113/203) tumor samples harbored candidate mutations. Sixty-five samples had single mutations (32%) while the remaining samples had ≥2 mutations (24%). 196 candidate mutation calls were made in 50 genes. The most common somatic oncogene mutations were found in EGFR, KRAS, PDGRFα, KIT, and PIK3CA. Other mutations found in additional genes were found at lower frequencies (<3%).

Conclusions/Significance

Sequenom analysis using OncoMap on DNA extracted from FFPE ovarian cancer samples is feasible and leads to the detection of potentially druggable mutations. Screening HGSC for somatic mutations in oncogenes may lead to additional therapies for this patient population.     

5-Hydroxymethylcytosine is an essential intermediate of active DNA demethylation processes in primary human monocytes

Background

Cytosine methylation is a frequent epigenetic modification restricting the activity of gene regulatory elements. Whereas DNA methylation patterns are generally inherited during replication, both embryonic and somatic differentiation processes require the removal of cytosine methylation at specific gene loci to activate lineage-restricted elements. However, the exact mechanisms facilitating the erasure of DNA methylation remain unclear in many cases.

Results

We previously established human post-proliferative monocytes as a model to study active DNA demethylation. We now show, for several previously identified genomic sites, that the loss of DNA methylation during the differentiation of primary, post-proliferative human monocytes into dendritic cells is preceded by the local appearance of 5-hydroxymethylcytosine. Monocytes were found to express the methylcytosine dioxygenase Ten-Eleven Translocation (TET) 2, which is frequently mutated in myeloid malignancies. The siRNA-mediated knockdown of this enzyme in primary monocytes prevented active DNA demethylation, suggesting that TET2 is essential for the proper execution of this process in human monocytes.

Conclusions

The work described here provides definite evidence that TET2-mediated conversion of 5-methylcytosine to 5-hydroxymethylcytosine initiates targeted, active DNA demethylation in a mature postmitotic myeloid cell type.     

High Resolution Size Analysis of Fetal DNA in the Urine of Pregnant Women by Paired-End Massively Parallel Sequencing

Background

Fetal DNA in maternal urine, if present, would be a valuable source of fetal genetic material for noninvasive prenatal diagnosis. However, the existence of fetal DNA in maternal urine has remained controversial. The issue is due to the lack of appropriate technology to robustly detect the potentially highly degraded fetal DNA in maternal urine.

Methodology

We have used massively parallel paired-end sequencing to investigate cell-free DNA molecules in maternal urine. Catheterized urine samples were collected from seven pregnant women during the third trimester of pregnancies. We detected fetal DNA by identifying sequenced reads that contained fetal-specific alleles of the single nucleotide polymorphisms. The sizes of individual urinary DNA fragments were deduced from the alignment positions of the paired reads. We measured the fractional fetal DNA concentration as well as the size distributions of fetal and maternal DNA in maternal urine.

Principal Findings

Cell-free fetal DNA was detected in five of the seven maternal urine samples, with the fractional fetal DNA concentrations ranged from 1.92% to 4.73%. Fetal DNA became undetectable in maternal urine after delivery. The total urinary cell-free DNA molecules were less intact when compared with plasma DNA. Urinary fetal DNA fragments were very short, and the most dominant fetal sequences were between 29 bp and 45 bp in length.

Conclusions

With the use of massively parallel sequencing, we have confirmed the existence of transrenal fetal DNA in maternal urine, and have shown that urinary fetal DNA was heavily degraded.     

Longitudinal,genome-scale analysis of DNA methylation in twins from birth to 18 months of age reveals rapid epigenetic change in early life and pair-specific effects of discordance

Background

The extent to which development- and age-associated epigenetic changes are influenced by genetic, environmental and stochastic factors remains to be discovered. Twins provide an ideal model with which to investigate these influences but previous cross-sectional twin studies provide contradictory evidence of within-pair epigenetic drift over time. Longitudinal twin studies can potentially address this discrepancy.

Results

In a pilot, genome-scale study of DNA from buccal epithelium, a relatively homogeneous tissue, we show that one-third of the CpGs assayed show dynamic methylation between birth and 18 months. Although all classes of annotated genomic regions assessed show an increase in DNA methylation over time, probes located in intragenic regions, enhancers and low-density CpG promoters are significantly over-represented, while CpG islands and high-CpG density promoters are depleted among the most dynamic probes. Comparison of co-twins demonstrated that within-pair drift in DNA methylation in our cohort is specific to a subset of pairs, who show more differences at 18 months. The rest of the pairs show either minimal change in methylation discordance, or more similar, converging methylation profiles at 18 months. As with age-associated regions, sites that change in their level of within-pair discordance between birth and 18 months are enriched in genes involved in development, but the average magnitude of change is smaller than for longitudinal change.

Conclusions

Our findings suggest that DNA methylation in buccal epithelium is influenced by non-shared stochastic and environmental factors that could reflect a degree of epigenetic plasticity within an otherwise constrained developmental program.     

Bis-class: a new classification tool of methylation status using bayes classifier and local methylation information

Background

Whole genome sequencing of bisulfite converted DNA (‘methylC-seq’) method provides comprehensive information of DNA methylation. An important application of these whole genome methylation maps is classifying each position as a methylated versus non-methylated nucleotide. A widely used current method for this purpose, the so-called binomial method, is intuitive and straightforward, but lacks power when the sequence coverage and the genome-wide methylation level are low. These problems present a particular challenge when analyzing sparsely methylated genomes, such as those of many invertebrates and plants.

Results

We demonstrate that the number of sequence reads per position from methylC-seq data displays a large variance and can be modeled as a shifted negative binomial distribution. We also show that DNA methylation levels of adjacent CpG sites are correlated, and this similarity in local DNA methylation levels extends several kilobases. Taking these observations into account, we propose a new method based on Bayesian classification to infer DNA methylation status while considering the neighborhood DNA methylation levels of a specific site. We show that our approach has higher sensitivity and better classification performance than the binomial method via multiple analyses, including computational simulations, Area Under Curve (AUC) analyses, and improved consistencies across biological replicates. This method is especially advantageous in the analyses of sparsely methylated genomes with low coverage.

Conclusions

Our method improves the existing binomial method for binary methylation calls by utilizing a posterior odds framework and incorporating local methylation information. This method should be widely applicable to the analyses of methylC-seq data from diverse sparsely methylated genomes. Bis-Class and example data are provided at a dedicated website (http://bibs.snu.ac.kr/software/Bisclass).

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2164-15-608) contains supplementary material, which is available to authorized users.     

Quantification of epigenetic biomarkers: an evaluation of established and emerging methods for DNA methylation analysis

Background

DNA methylation is an important epigenetic mechanism in several human diseases, most notably cancer. The quantitative analysis of DNA methylation patterns has the potential to serve as diagnostic and prognostic biomarkers, however, there is currently a lack of consensus regarding the optimal methodologies to quantify methylation status. To address this issue we compared five analytical methods: (i) MethyLight qPCR, (ii) MethyLight digital PCR (dPCR), methylation-sensitive and -dependent restriction enzyme (MSRE/MDRE) digestion followed by (iii) qPCR or (iv) dPCR, and (v) bisulfite amplicon next generation sequencing (NGS). The techniques were evaluated for linearity, accuracy and precision.

Results

MethyLight qPCR displayed the best linearity across the range of tested samples. Observed methylation measured by MethyLight- and MSRE/MDRE-qPCR and -dPCR were not significantly different to expected values whilst bisulfite amplicon NGS analysis over-estimated methylation content. Bisulfite amplicon NGS showed good precision, whilst the lower precision of qPCR and dPCR analysis precluded discrimination of differences of < 25% in methylation status. A novel dPCR MethyLight assay is also described as a potential method for absolute quantification that simultaneously measures both sense and antisense DNA strands following bisulfite treatment.

Conclusions

Our findings comprise a comprehensive benchmark for the quantitative accuracy of key methods for methylation analysis and demonstrate their applicability to the quantification of circulating tumour DNA biomarkers by using sample concentrations that are representative of typical clinical isolates.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2164-15-1174) contains supplementary material, which is available to authorized users.     

Hypermethylation Leads to Bone Morphogenetic Protein 6 Downregulation in Hepatocellular Carcinoma

Background

In the liver, bone morphogenetic protein 6 (BMP-6) maintains balanced iron metabolism. However, the mechanism that underlies greater BMP-6 expression in hepatocellular carcinoma (HCC) tissue than adjacent non-cancerous tissue is unclear. This study sought to investigate the epigenetic mechanisms of BMP-6 expression by analysing the relationship between the DNA methylation status of BMP-6 and the expression of BMP-6.

Methods

Methylation-specific polymerase chain reaction (PCR), bisulphite sequencing PCR, the MethyLight assay, and quantitative real-time PCR were performed to examine BMP-6 methylation and mRNA expression levels. Immunohistochemistry (IHC) was performed on tissue arrays to evaluate the BMP-6 protein level.

Results

BMP-6 mRNA expression was approximately 84.09% lower in HCC tissues than in adjacent non-cancerous tissues, and this low level of expression was associated with a poor prognosis. Moreover, the hypermethylation observed in HCC cell lines and HCC tissues was correlated with the BMP-6 mRNA expression level, and this correlation was validated following treatment with 5-aza-CdR, a demethylation agent. In addition, BMP-6 DNA methylation was upregulated by 68.42% in 114 clinical HCC tissue samples compared to adjacent normal tissues, whereas the BMP-6 staining intensity was downregulated by 77.03% in 75 clinical HCC tissue samples in comparison to adjacent normal tissues. Furthermore, elevated expression of BMP-6 in HCC cell lines inhibited cell colony formation.

Conclusions

Our results suggest that BMP-6 CpG island hypermethylation leads to decreased BMP-6 expression in HCC tissues.     

Genome-wide signatures of differential DNA methylation in pediatric acute lymphoblastic leukemia

Background

Although aberrant DNA methylation has been observed previously in acute lymphoblastic leukemia (ALL), the patterns of differential methylation have not been comprehensively determined in all subtypes of ALL on a genome-wide scale. The relationship between DNA methylation, cytogenetic background, drug resistance and relapse in ALL is poorly understood.

Results

We surveyed the DNA methylation levels of 435,941 CpG sites in samples from 764 children at diagnosis of ALL and from 27 children at relapse. This survey uncovered four characteristic methylation signatures. First, compared with control blood cells, the methylomes of ALL cells shared 9,406 predominantly hypermethylated CpG sites, independent of cytogenetic background. Second, each cytogenetic subtype of ALL displayed a unique set of hyper- and hypomethylated CpG sites. The CpG sites that constituted these two signatures differed in their functional genomic enrichment to regions with marks of active or repressed chromatin. Third, we identified subtype-specific differential methylation in promoter and enhancer regions that were strongly correlated with gene expression. Fourth, a set of 6,612 CpG sites was predominantly hypermethylated in ALL cells at relapse, compared with matched samples at diagnosis. Analysis of relapse-free survival identified CpG sites with subtype-specific differential methylation that divided the patients into different risk groups, depending on their methylation status.

Conclusions

Our results suggest an important biological role for DNA methylation in the differences between ALL subtypes and in their clinical outcome after treatment.