FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2179220709> ?p ?o ?g. }

• W2179220709 endingPage "2146" @default.
• W2179220709 startingPage "2135" @default.
• W2179220709 abstract "•Pervasive, focal hypermethylation distinguishes low- from high-grade prostate cancer•High-grade-specific changes are most commonly outside promoters and CpG islands•Intergenic regions are enriched for functional elements including EZH2 binding sites•High-grade versus cancer-specific changes have different potential functions A critical need in understanding the biology of prostate cancer is characterizing the molecular differences between indolent and aggressive cases. Because DNA methylation can capture the regulatory state of tumors, we analyzed differential methylation patterns genome-wide among benign prostatic tissue and low-grade and high-grade prostate cancer and found extensive, focal hypermethylation regions unique to high-grade disease. These hypermethylation regions occurred not only in the promoters of genes but also in gene bodies and at intergenic regions that are enriched for DNA-protein binding sites. Integration with existing RNA-sequencing (RNA-seq) and survival data revealed regions where DNA methylation correlates with reduced gene expression associated with poor outcome. Regions specific to aggressive disease are proximal to genes with distinct functions from regions shared by indolent and aggressive disease. Our compendium of methylation changes reveals crucial molecular distinctions between indolent and aggressive prostate cancer. A critical need in understanding the biology of prostate cancer is characterizing the molecular differences between indolent and aggressive cases. Because DNA methylation can capture the regulatory state of tumors, we analyzed differential methylation patterns genome-wide among benign prostatic tissue and low-grade and high-grade prostate cancer and found extensive, focal hypermethylation regions unique to high-grade disease. These hypermethylation regions occurred not only in the promoters of genes but also in gene bodies and at intergenic regions that are enriched for DNA-protein binding sites. Integration with existing RNA-sequencing (RNA-seq) and survival data revealed regions where DNA methylation correlates with reduced gene expression associated with poor outcome. Regions specific to aggressive disease are proximal to genes with distinct functions from regions shared by indolent and aggressive disease. Our compendium of methylation changes reveals crucial molecular distinctions between indolent and aggressive prostate cancer. Prostate cancer remains the second most common cause of cancer deaths in American men, with 220,800 estimated new cases and 27,540 estimated deaths for 2015 (Siegel et al., 2015Siegel R.L. Miller K.D. Jemal A. Cancer statistics, 2015.CA Cancer J. Clin. 2015; 65: 5-29Crossref PubMed Scopus (11642) Google Scholar). Accurately differentiating between indolent and aggressive disease is of utmost importance to reduce overtreatment so that men with indolent cancers are spared the morbidities of radical therapy while those with potentially life-threatening cancers will undergo treatment with curative intent. However, triggers for radical therapy still require refinement for the 14% to 41% of low-risk patients who actually have more aggressive disease (Cooperberg et al., 2011Cooperberg M.R. Carroll P.R. Klotz L. Active surveillance for prostate cancer: progress and promise.J. Clin. Oncol. 2011; 29: 3669-3676Crossref PubMed Scopus (241) Google Scholar). Recently, a gene-expression-based assay using biopsy tissues has provided a genomically derived aggressiveness measure that clinicians can use in conjunction with clinicopathologic risk factors for supporting clinical decision-making (Klein et al., 2014Klein E.A. Cooperberg M.R. Magi-Galluzzi C. Simko J.P. Falzarano S.M. Maddala T. Chan J.M. Li J. Cowan J.E. Tsiatis A.C. et al.A 17-gene assay to predict prostate cancer aggressiveness in the context of Gleason grade heterogeneity, tumor multifocality, and biopsy undersampling.Eur. Urol. 2014; 66: 550-560Abstract Full Text Full Text PDF PubMed Scopus (463) Google Scholar). Still, knowledge about the molecular and biological differences between indolent and aggressive prostate cancer can be increased further, as most genome-wide studies of copy-number variation, somatic mutations, DNA methylation (Kim et al., 2011Kim J.H. Dhanasekaran S.M. Prensner J.R. Cao X. Robinson D. Kalyana-Sundaram S. Huang C. Shankar S. Jing X. Iyer M. et al.Deep sequencing reveals distinct patterns of DNA methylation in prostate cancer.Genome Res. 2011; 21: 1028-1041Crossref PubMed Scopus (159) Google Scholar, Yu et al., 2013Yu Y.P. Ding Y. Chen R. Liao S.G. Ren B.-G. Michalopoulos A. Michalopoulos G. Nelson J. Tseng G.C. Luo J.-H. Whole-genome methylation sequencing reveals distinct impact of differential methylations on gene transcription in prostate cancer.Am. J. Pathol. 2013; 183: 1960-1970Abstract Full Text Full Text PDF PubMed Scopus (36) Google Scholar), and gene expression have focused on tumor versus normal comparisons rather than stratifying by aggressiveness subtypes. We sought to uncover molecular mechanisms behind aggressive prostate cancer by comparing DNA methylation profiles between benign prostate, indolent prostate cancer, and aggressive prostate cancer. DNA methylation is a compelling candidate for involvement in prostate cancer aggressiveness due to the comparatively low somatic mutation rate in prostate tumors (Taylor et al., 2010Taylor B.S. Schultz N. Hieronymus H. Gopalan A. Xiao Y. Carver B.S. Arora V.K. Kaushik P. Cerami E. Reva B. et al.Integrative genomic profiling of human prostate cancer.Cancer Cell. 2010; 18: 11-22Abstract Full Text Full Text PDF PubMed Scopus (2783) Google Scholar) and the reported overexpression of DNA methyltransferases in prostate cancer (Kobayashi et al., 2011Kobayashi Y. Absher D.M. Gulzar Z.G. Young S.R. McKenney J.K. Peehl D.M. Brooks J.D. Myers R.M. Sherlock G. DNA methylation profiling reveals novel biomarkers and important roles for DNA methyltransferases in prostate cancer.Genome Res. 2011; 21: 1017-1027Crossref PubMed Scopus (192) Google Scholar). Two alternative ways to divide prostate cancers into aggressiveness groups for research are by outcomes (recurrence and/or prostate cancer specific mortality) or by histopathological grading. We have elected to stratify by histology, because our goal is to find molecular differences between these subtypes, and outcomes can be confounded by stage at presentation and case management decisions. Histologic Gleason scores (6–10 used in practice) derived from radical prostatectomy sections highly correlate with tumor aggressiveness when considered in terms of biochemical recurrence, development of metastatic disease, and mortality due to prostate cancer (Albertsen et al., 1998Albertsen P.C. Hanley J.A. Gleason D.F. Barry M.J. Competing risk analysis of men aged 55 to 74 years at diagnosis managed conservatively for clinically localized prostate cancer.JAMA. 1998; 280: 975-980Crossref PubMed Scopus (623) Google Scholar, Epstein et al., 1996Epstein J.I. Partin A.W. Sauvageot J. Walsh P.C. Prediction of progression following radical prostatectomy. A multivariate analysis of 721 men with long-term follow-up.Am. J. Surg. Pathol. 1996; 20: 286-292Crossref PubMed Scopus (500) Google Scholar, Pan et al., 2000Pan C.C. Potter S.R. Partin A.W. Epstein J.I. The prognostic significance of tertiary Gleason patterns of higher grade in radical prostatectomy specimens: a proposal to modify the Gleason grading system.Am. J. Surg. Pathol. 2000; 24: 563-569Crossref PubMed Scopus (181) Google Scholar). Moreover, studies have established that Gleason score 6 tumors almost entirely lack the capacity to metastasize to lymph nodes, as metastatic disease is nearly entirely confined to tumors with Gleason scores 7 and above (Ross et al., 2012Ross H.M. Kryvenko O.N. Cowan J.E. Simko J.P. Wheeler T.M. Epstein J.I. Do adenocarcinomas of the prostate with Gleason score (GS) ≤6 have the potential to metastasize to lymph nodes?.Am. J. Surg. Pathol. 2012; 36: 1346-1352Crossref PubMed Scopus (263) Google Scholar). To detect DNA methylation changes specific to either indolent or aggressive cancers, we divided samples into three groups: benign prostatic tissue, low-grade (indolent, Gleason score 6) prostate cancer, and high-grade (aggressive, Gleason score 8–10) prostate cancer. We used benign tissue from men who underwent cystoprostatectomy for non-prostate pathology that did not have any evidence of prostate cancer or pre-cancerous lesions, because tumor-adjacent benign tissue from men with prostate cancer is known to harbor DNA methylation changes associated with prostate cancer (Mehrotra et al., 2008Mehrotra J. Varde S. Wang H. Chiu H. Vargo J. Gray K. Nagle R.B. Neri J.R. Mazumder A. Quantitative, spatial resolution of the epigenetic field effect in prostate cancer.Prostate. 2008; 68: 152-160Crossref PubMed Scopus (60) Google Scholar). Of the patterns of differentially methylated regions (DMRs) we detected, hypermethylation specific to high-grade (high grade DMRs) and hypermethylation shared by low and high grades (shared DMRs) were the strongest and most statistically robust. High-grade DMRs occurred more frequently at intergenic regions and gene bodies than other genomic contexts. These DMRs, including the intergenic, are enriched for putative functional elements (DNaseI hypersensitive sites, chromatin immunoprecipitation sequencing (ChIP-seq) peaks, and enhancer-like chromatin states). Several high-grade DMRs inversely correlate with the expression of genes that associate with poorer outcomes in meta-analysis. A network analysis of genes proximal to high-grade DMRs revealed distinct functional enrichments from genes proximal to shared DMRs. These results provide valuable insight into aggressive prostate cancer biology, and future work will address the function of specific loci and the possibility of a DNA-methylation-based diagnostic test distinguishing indolent from aggressive cancers. We performed MBD (methyl-CpG binding domain)-isolated genome sequencing (MiGS) (Serre et al., 2010Serre D. Lee B.H. Ting A.H. MBD-isolated Genome Sequencing provides a high-throughput and comprehensive survey of DNA methylation in the human genome.Nucleic Acids Res. 2010; 38: 391-399Crossref PubMed Scopus (311) Google Scholar) on benign prostatic tissue obtained from cystoprostatectomy specimens and low-grade and high-grade prostate cancer specimens obtained from radical prostatectomies (Table S1). A principal component analysis (PCA) was performed to investigate whether variations in DNA methylation genome-wide can distinguish among the groups (Figures 1A and S1A). The first principal component partly separates low-grade from high-grade cases, and benign samples separate when considering the second and third principal components with one outlier (excluded in Figure 1A and included in Figure S1A). Excluding the outlier, bootstrap false discovery rates (FDRs) for the distances between group centroids are below 10%: 3.2% for benign versus low grade, 7.0% for benign versus high grade, and 3.2% for low grade versus high grade. Although the samples are from a mixture of patients from European American and African American ancestry, a MANOVA indicates that sample group significantly (p < 0.05) associates with the first three principal components (p = 4.9 × 10−7), whereas ancestry (p = 0.16) and the interaction of ancestry and group (p = 0.05) do not. In spite of some heterogeneity within groups, the dimensionality reduction and clustering indicate that the DNA methylation sequencing data contain variability that can distinguish these three groups. We identified all statistically significant quantitative differences in read counts (at adjusted p value < 0.01) produced by MiGS in regions ≥150 bp for all possible patterns of methylation differences among the three groups (Figure 1B; Table S2). The two most robust patterns of DMRs were hypermethylation shared by low and high grades (shared DMRs: 8,944 sites, ranging from 150 bp to 3,450 bp, covering 2,972.4 kb in total) and hypermethylation specific to high grade (high-grade DMRs: 4,932 sites, ranging from 150 bp to 900 bp, covering 919.15 kb in total). To establish significance of observing DMRs from each pattern, we called DMRs in 1,000 bootstrap resamples of the group assignments, and these two patterns had the lowest FDRs (19.3% for shared DMRs and 22.2% for high-grade DMRs). We also stratified DMRs into a “frequent” subset where the qualitative methylation status of two-thirds of samples in each group was in agreement (Figure 1C). In this subset, FDRs were even lower for the high-grade (6.5%) and shared (4.9%) DMRs yet remained high (17.9%) for the third most significant pattern (hypomethylation specific to low grade). At a p value cutoff of 0.001 (Figures S1B–S1D), FDRs for the high-grade and shared DMRs are even lower (13% and 6%), yet both patterns retain a substantial number of DMRs (1,712 and 4,018). One thousand bootstraps were sufficient to provide stable estimates of the random populations, as a plot of permutation numbers versus FDRs shows stabilization for all DMRs (Figures S1E and S1F). We also used the DMR lists from these bootstraps to compute FDRs for each observed DMR and found that nearly all DMRs, regardless of pattern (all but 4 of 33,788), had site-specific FDRs below 5% (Figure S1G). When considering fold change, differences were higher in the high-grade (log2 high/benign of 2.00 and high/low of 1.61) and shared (log2 high/benign of 2.79 and low/benign of 2.39) DMRs when compared to others, particularly the patterns of hypomethylation (Figure S1H). Thus, high-grade and shared DMRs were the most statistically robust and quantitatively strong patterns, and we focused our characterization on these two sets. Here, we highlight examples of high-grade and shared DMRs to illustrate the regional and contextual nature of these differences. The first example occurs in the gene body of CD14, which contains both high-grade DMRs and a shared DMR (Figure 1D). The regional signal plots for two neighboring high-grade DMRs reveal that high-grade samples have high methylation signals compared to both low-grade (log2 high/low of 1.90 and 2.77) and benign (log2 high/benign of 2.66 and 4.49) samples. Notably, only the African American low-grade samples show hypermethylation at the shared DMR near the gene promoter, and this DMR is 166 bp away from a SNP that is associated with aggressive prostate cancer in African Americans (Mason et al., 2010Mason T.E. Ricks-Santi L. Chen W. Apprey V. Joykutty J. Ahaghotu C. Kittles R. Bonney G. Dunston G.M. Association of CD14 variant with prostate cancer in African American men.Prostate. 2010; 70: 262-269PubMed Google Scholar). This is interesting in light of prior research suggesting that patients with African ancestry have a higher risk of aggressive disease (Yamoah et al., 2015Yamoah K. Deville C. Vapiwala N. Spangler E. Zeigler-Johnson C.M. Malkowicz B. Lee D.I. Kattan M. Dicker A.P. Rebbeck T.R. African American men with low-grade prostate cancer have increased disease recurrence after prostatectomy compared with Caucasian men.Urol. Oncol. 2015; 33: 70.e15-70.e22Abstract Full Text Full Text PDF PubMed Scopus (33) Google Scholar). The second example is a shared DMR within the first exon of protocadherin gamma subfamily A, 11 (PCDHGA11) (Figure 1E). Robust DNA methylation is detected in both the low-grade and high-grade groups, but not in the benign specimens (log2 high/benign of 2.54 and log2 low/benign of 2.54). This DMR is located in the gene bodies of overlapping genes from the protocadherin family. Differential DNA methylation in this region has been previously detected in a chemical screen for initiating DNA methylation events in prostate cancer (Severson et al., 2012Severson P.L. Tokar E.J. Vrba L. Waalkes M.P. Futscher B.W. Agglomerates of aberrant DNA methylation are associated with toxicant-induced malignant transformation.Epigenetics. 2012; 7: 1238-1248Crossref PubMed Scopus (26) Google Scholar). In summary, the high-grade and shared DMR sets of focal differences in DNA methylation robustly distinguish indolent and aggressive prostate cancers from each other and from benign prostatic tissue in our cohort. To begin understanding the potential functions of DMRs, we examined the spatial relationship between DMRs and RefSeq coding gene models (promoters, exons, introns, and gene 3′ ends) by overlapping the DMR coordinates with annotated genes (Figure 2A). The analysis revealed that DMRs are enriched for promoters (17.5% of the shared DMRs and 8.7% of high-grade DMRs versus 1.4% in the background set) and gene 3′ ends (5.1% of the shared DMRs and 4.7% of high-grade DMRs versus 1.6% in the background set). Notably, the promoter context was more common for shared DMRs than for high-grade DMRs (odds ratio [OR], 2.23; 95% confidence interval [CI], 1.99–2.51). Both sets of DMRs also frequently occur in intergenic contexts (36.0% of the shared DMRs and 31.6% of the high-grade DMRs). This enrichment of DMRs in regions beyond gene promoters raises the possibility of varied functions for DNA methylation in different genomic contexts in prostate cancer, and these non-promoter functions may be especially pervasive in high-grade cancers. We next examined the DMRs with respect to putative functional elements and CpG islands (Figures 2B and 2C) and observed two striking features. First, the intergenic DMRs appear highly enriched for DNaseI hypersensitivity sites (72% of high-grade DMRs and 89% of the shared DMRs versus 25% in the background set) and ChIP-seq peaks from the ENCODE project (57% of high-grade DMRs and 78% of the shared DMRs versus 17%% in the background set). Second, shared DMRs were more commonly located over CpG islands and shores than high-grade DMRs. These observations prompted us to further investigate the overlap of DMRs with specific putative regulatory elements in more detail. We next explored if binding sites for any particular DNA-binding proteins are enriched in the DMRs using reference data. An enrichment analysis of high-grade and shared DMRs with consensus ChIP-seq peaks from the ENCODE project revealed a strong enrichment for enhancer of zeste 2 (EZH2) and retinoblastoma binding protein 5 (RBBP5) in the intergenic context (Figures 2D and S2A). EZH2 is recruited to chromatin via interactions with nascent RNA, indicating transcriptional activity (Davidovich et al., 2013Davidovich C. Zheng L. Goodrich K.J. Cech T.R. Promiscuous RNA binding by Polycomb repressive complex 2.Nat. Struct. Mol. Biol. 2013; 20: 1250-1257Crossref PubMed Scopus (335) Google Scholar, Kaneko et al., 2013Kaneko S. Son J. Shen S.S. Reinberg D. Bonasio R. PRC2 binds active promoters and contacts nascent RNAs in embryonic stem cells.Nat. Struct. Mol. Biol. 2013; 20: 1258-1264Crossref PubMed Scopus (223) Google Scholar). Potential for binding of this factor in the intergenic DMRs suggests we may have identified unannotated transcription units that could impact phenotype. These sites may also be associated with distal enhancer and repressor elements, which can produce regulatory RNAs (Wang et al., 2011Wang D. Garcia-Bassets I. Benner C. Li W. Su X. Zhou Y. Qiu J. Liu W. Kaikkonen M.U. Ohgi K.A. et al.Reprogramming transcription by distinct classes of enhancers functionally defined by eRNA.Nature. 2011; 474: 390-394Crossref PubMed Scopus (654) Google Scholar). To complement the experimentally derived transcription factor binding data, we also performed a de novo motif analysis (Figure S2B). Notably, an EGR1-like motif was found in high-grade DMRs, and a MAFB-like motif was found in shared DMRs. Analysis using the entirety of ENCODE data indicated the potential for regulatory activity at the intergenic DMRs, but not necessarily in prostate cells specifically. Therefore, we leveraged datasets specific to prostate and prostate cancer cells to corroborate the findings above. Consistently, high-grade and shared DMRs across all genomic contexts are enriched for DNaseI hypersensitive sites and ChIP-seq peaks in prostate cancer cell lines including Polycomb repressive complex 2 components EZH2 and SUZ12 (Figure 2E). High-grade and shared DMRs also overlap with nucleosome depleted regions (NDRs) that are present in benign prostate cells and lost in PC3 prostate cancer cells, which may represent enhancers decommissioned in carcinogenesis (Taberlay et al., 2014Taberlay P.C. Statham A.L. Kelly T.K. Clark S.J. Jones P.A. Reconfiguration of nucleosome-depleted regions at distal regulatory elements accompanies DNA methylation of enhancers and insulators in cancer.Genome Res. 2014; 24: 1421-1432Crossref PubMed Scopus (142) Google Scholar). Furthermore, non-promoter high-grade (OR, 5.6; 95% CI, 5.3–6.0) and shared (OR, 5.0; 95% CI, 4.8–5.3) DMRs overlap substantially with annotated enhancers in prostatic cells and tissues than expected of genomic background (Figure 2F). Additionally, >70% of non-promoter high-grade and shared DMRs are annotated as putative active enhancers across cells and tissues. Taken together, the results of the enrichment analyses raise many functional possibilities for the DMRs and suggest that the impact of DMRs residing in intergenic sequences may be just as strong as those proximal to known genes. We sought to verify our DMRs by leveraging the DNA methylation microarray data from The Cancer Genome Atlas (TCGA) (Table S3). However, this comparison is complicated by the fact that the array only assays single CpG sites and covers less than 2% of CpG sites across the genome (Clark et al., 2012Clark C. Palta P. Joyce C.J. Scott C. Grundberg E. Deloukas P. Palotie A. Coffey A.J. A comparison of the whole genome approach of MeDIP-seq to the targeted approach of the Infinium HumanMethylation450 BeadChip(®) for methylome profiling.PLoS ONE. 2012; 7: e50233Crossref PubMed Scopus (76) Google Scholar), while our data from MiGS detects regions ≥150 bp and covered 80.7% of 50-bp windows genome-wide that contain three or more CpG sites. Shared DMRs were covered by array sites more frequently than high-grade DMRs, and 77% of our high-grade DMRs were not able to be measured by TCGA (Figure 3A). Focusing on DMRs that do contain array sites, 74.0% of shared DMRs contain hypermethylation shared by low- and high-grade TCGA samples, as stratified using the available clinical data (Figures 3B and S3B). A smaller percentage (8.9%, 101 DMRs) of high-grade DMRs contained statistically significant TCGA high-grade hypermethylation, and this proportion is greater than the null expectation of 1.7% from the entire array (OR, 11.3; 95% CI, 9.0–14.0) (Figures 3B and S3A). A small number of shared DMRs had high-grade hypermethylation in TCGA, and a large number of high-grade DMRs tested significant for shared hypermethylation in TCGA. Also, 4.1% of high-grade DMRs overlapped with differentially variable sites where variance was uniquely elevated in high grade when compared to low grade and the adjacent normal tissues (Figure 3C), suggesting that some high-grade DMRs may concentrate in subtypes and not be sufficient to test as statistically significant differences in means. The array sites for the CD14 DMR and PCDHGA11 DMR show significant differential hypermethylation (Figures 3D and 3E), and an example of a site with differential variability occurs near the gene EYA1 (Figures S3C and S3F). Because many of our high-grade DMRs overlap with cancer-specific hypermethylation in TCGA, we surmise that the limited concordance of total high-grade DMRs could stem from differences in sample classification due to inter-observer variability in Gleason grade assignments between our current study and TCGA. Such inter-observer variability, while minimally affecting cancer versus non-cancer categorization, is well known (McKenney et al., 2011McKenney J.K. Simko J. Bonham M. True L.D. Troyer D. Hawley S. Newcomb L.F. Fazli L. Kunju L.P. Nicolas M.M. et al.Canary/Early Detection Research Network Prostate Active Surveillance Study InvestigatorsThe potential impact of reproducibility of Gleason grading in men with early stage prostate cancer managed by active surveillance: a multi-institutional study.J. Urol. 2011; 186: 465-469Abstract Full Text Full Text PDF PubMed Scopus (79) Google Scholar) and can contribute to discrepancies in the detection of high-grade-specific DMRs. Regardless of exact histological assignments and purity in TCGA, large fractions of both high-grade and shared DMRs overlap with differential methylation and differential variance sites in TCGA. Having established hypermethylation and hypervariability at TCGA array sites within our DMRs, we leveraged TCGA’s RNA-seq data to explore correlations with gene expression. DMRs were divided into promoter and gene body with respect to a reference-guided de novo transcriptome assembly (Pertea et al., 2015Pertea M. Pertea G.M. Antonescu C.M. Chang T.-C. Mendell J.T. Salzberg S.L. StringTie enables improved reconstruction of a transcriptome from RNA-seq reads.Nat. Biotechnol. 2015; 33: 290-295Crossref PubMed Scopus (4957) Google Scholar) we performed on 537 TCGA RNA-seq samples that also had DNA methylation microarray data. For each site within a DMR, we computed Pearson’s correlation coefficients between the site’s fractional methylation level (beta-value) and the fragments per kilobase of transcript per million mapped reads (FPKM) expression level of the overlapping gene. Against a null expectation established by correlating each gene against all quality control (QC)-passed 440,706 sites on the array, both promoter and gene body show a large population of inverse correlations, whereas gene bodies uniquely have a large population of positive correlations (Figure 4A). Requiring an FDR of <5% and a correlation magnitude of greater than 0.4, nine high-grade genes and 75 shared genes had strong promoter correlations with expression, while 26 high-grade genes and 147 shared genes had strong gene body correlations (Tables 1 and S4). For example, the high-grade DMR in the promoter of CCDC8 showed a strong negative correlation with gene expression (Figures 4B and 4C). In contrast, the high-grade DMR in the gene body of HOXC4 showed a strong positive correlation with gene expression (Figures 4D and 4E). These possibilities demonstrate that DNA methylation may have different functions depending on the genomic context and that correlations of DNA methylation to gene expression are not limited to gene promoters and inverse relationships. Indeed, the potential activating effect of gene body methylation has been reported by others (Jjingo et al., 2012Jjingo D. Conley A.B. Yi S.V. Lunyak V.V. Jordan I.K. On the presence and role of human gene-body DNA methylation.Oncotarget. 2012; 3: 462-474Crossref PubMed Scopus (331) Google Scholar), and effects of gene body DMRs on gene expression may be linked to the presence of gene body enhancers (Agirre et al., 2015Agirre X. Castellano G. Pascual M. Heath S. Kulis M. Segura V. Bergmann A. Esteve A. Merkel A. Raineri E. et al.Whole-epigenome analysis in multiple myeloma reveals DNA hypermethylation of B cell-specific enhancers.Genome Res. 2015; 25: 478-487Crossref PubMed Scopus (100) Google Scholar).Table 1Strong and Significant Correlations between TCGA Sites that Overlap with DMRs and Gene ExpressionPromoter (−1,000 to +500 of a Putative txStart)Gene Body (Non-promoter Introns and Exons)High GradeSharedHigh GradeSharedDMR to TCGA Site OverlapsNumber of DMRs5051,6263,5755,657DMRs with site on 450K2391,2517632,839Sites on 450K with DMR4983,9721,1246,867Sites on 450K with DMR and IQR > 0.054553,9138916,671Total Expression Correlations (FDR < 5% and |r| > 0.4)DMRs87433175Sites1323246406Genes97526147IQR, interquartile range. Open table in a new tab IQR, interquartile range. Next, we evaluated whether the genes correlated with DMRs have implications for prostate cancer aggressiveness by examining the survival z-scores generated by the PRECOG meta-analysis of prostate cancer outcomes (Gentles et al., 2015Gentles A.J. Newman A.M. Liu C.L. Bratman S.V. Feng W. Kim D. Nair V.S. Xu Y. Khuong A. Hoang C.D. et al.The prognostic landscape of genes and infiltrating immune cells across human cancers.Nat. Med. 2015; 21: 938-945Crossref PubMed Scopus (1782) Google Scholar). The majority of high-grade genes with strong and significant inverse correlations between methylation and expression also showed a negative PRECOG z-score, which indicates that decreased expression is associated with poorer outcomes (Figure 4F). For example, reduced expression of myosin light chain kinase (MYLK) is associated with poorer outcomes (PRECOG z-score −3.2) and has a gene body DMR (Figure 4G; log2 high/low of 1.04) inversely correlating with gene expression (Figure 4H; r = −0.54, FDR = 0.19%). While our analysis of putative functional elements and integration with RNA-seq data provided a view of the DMRs from a sequence context perspective, we have also analyzed how DMRs relate to known genes to provide a cellular context. 249 genes overlapping with 376 high-grade DMRs have strong links to either prostate cancer or cancers in general based on a gene panel assembled from the literature (Table S5). These genes span multiple pathways, including androgen receptor (AR) signaling (NCOR2, SRD5A2), DNA damage response (PRKDC), and growth factor receptors (FGFR1, FGFR2, IGF1R, EGFR). Some genes have specific and established links to aggressiveness, such as protein kinase, DNA-activated, catalytic polypeptide (PRKDC) that is involved with induction of metastasis and independently predicts for recurrence and poor survival (Goodwin et al., 2015Goodwin J.F. Kothari V. Drake J.M. Zhao S. Dylgjeri E. Dean J.L. Schiewer M.J. McNair C. Jones J.K. Aytes A. et al.DNA-PKcs-Mediated Transcriptional" @default.
• W2179220709 created "2016-06-24" @default.
• W2179220709 creator A5001069772 @default.
• W2179220709 creator A5016306089 @default.
• W2179220709 creator A5016502031 @default.
• W2179220709 creator A5024488464 @default.
• W2179220709 creator A5025347286 @default.
• W2179220709 creator A5034716746 @default.
• W2179220709 creator A5047160101 @default.
• W2179220709 creator A5070471092 @default.
• W2179220709 creator A5071600865 @default.
• W2179220709 date "2015-12-01" @default.
• W2179220709 modified "2023-10-16" @default.
• W2179220709 title "Methylome-wide Sequencing Detects DNA Hypermethylation Distinguishing Indolent from Aggressive Prostate Cancer" @default.
• W2179220709 cites W1107917109 @default.
• W2179220709 cites W1263857580 @default.
• W2179220709 cites W1558158633 @default.
• W2179220709 cites W1767798659 @default.
• W2179220709 cites W1970092233 @default.
• W2179220709 cites W1981045578 @default.
• W2179220709 cites W1986560938 @default.
• W2179220709 cites W1989062076 @default.
• W2179220709 cites W1996060670 @default.
• W2179220709 cites W1996487270 @default.
• W2179220709 cites W2015915171 @default.
• W2179220709 cites W2020764301 @default.
• W2179220709 cites W2039289317 @default.
• W2179220709 cites W2045062938 @default.
• W2179220709 cites W2052014424 @default.
• W2179220709 cites W2065498897 @default.
• W2179220709 cites W2076399666 @default.
• W2179220709 cites W2084990854 @default.
• W2179220709 cites W2092992881 @default.
• W2179220709 cites W2101436815 @default.
• W2179220709 cites W2102983187 @default.
• W2179220709 cites W2104005232 @default.
• W2179220709 cites W2107633226 @default.
• W2179220709 cites W2110128489 @default.
• W2179220709 cites W2111465765 @default.
• W2179220709 cites W2114884175 @default.
• W2179220709 cites W2118397540 @default.
• W2179220709 cites W2120411240 @default.
• W2179220709 cites W2123918450 @default.
• W2179220709 cites W2125114687 @default.
• W2179220709 cites W2129259763 @default.
• W2179220709 cites W2139199979 @default.
• W2179220709 cites W2151494814 @default.
• W2179220709 cites W2153737103 @default.
• W2179220709 cites W2156776019 @default.
• W2179220709 cites W2162178715 @default.
• W2179220709 cites W2164218164 @default.
• W2179220709 cites W2170551349 @default.
• W2179220709 cites W2170552969 @default.
• W2179220709 cites W2329106677 @default.
• W2179220709 doi "https://doi.org/10.1016/j.celrep.2015.10.078" @default.
• W2179220709 hasPubMedCentralId "https://www.ncbi.nlm.nih.gov/pmc/articles/4684962" @default.
• W2179220709 hasPubMedId "https://pubmed.ncbi.nlm.nih.gov/26628371" @default.
• W2179220709 hasPublicationYear "2015" @default.
• W2179220709 type Work @default.
• W2179220709 sameAs 2179220709 @default.
• W2179220709 citedByCount "36" @default.
• W2179220709 countsByYear W21792207092015 @default.
• W2179220709 countsByYear W21792207092016 @default.
• W2179220709 countsByYear W21792207092017 @default.
• W2179220709 countsByYear W21792207092018 @default.
• W2179220709 countsByYear W21792207092019 @default.
• W2179220709 countsByYear W21792207092020 @default.
• W2179220709 countsByYear W21792207092021 @default.
• W2179220709 countsByYear W21792207092022 @default.
• W2179220709 countsByYear W21792207092023 @default.
• W2179220709 crossrefType "journal-article" @default.
• W2179220709 hasAuthorship W2179220709A5001069772 @default.
• W2179220709 hasAuthorship W2179220709A5016306089 @default.
• W2179220709 hasAuthorship W2179220709A5016502031 @default.
• W2179220709 hasAuthorship W2179220709A5024488464 @default.
• W2179220709 hasAuthorship W2179220709A5025347286 @default.
• W2179220709 hasAuthorship W2179220709A5034716746 @default.
• W2179220709 hasAuthorship W2179220709A5047160101 @default.
• W2179220709 hasAuthorship W2179220709A5070471092 @default.
• W2179220709 hasAuthorship W2179220709A5071600865 @default.
• W2179220709 hasBestOaLocation W21792207091 @default.
• W2179220709 hasConcept C104317684 @default.
• W2179220709 hasConcept C121608353 @default.
• W2179220709 hasConcept C140173407 @default.
• W2179220709 hasConcept C150194340 @default.
• W2179220709 hasConcept C190727270 @default.
• W2179220709 hasConcept C2780192828 @default.
• W2179220709 hasConcept C502942594 @default.
• W2179220709 hasConcept C51679486 @default.
• W2179220709 hasConcept C54355233 @default.
• W2179220709 hasConcept C552990157 @default.
• W2179220709 hasConcept C70721500 @default.
• W2179220709 hasConcept C86803240 @default.
• W2179220709 hasConceptScore W2179220709C104317684 @default.
• W2179220709 hasConceptScore W2179220709C121608353 @default.
• W2179220709 hasConceptScore W2179220709C140173407 @default.
• W2179220709 hasConceptScore W2179220709C150194340 @default.
• W2179220709 hasConceptScore W2179220709C190727270 @default.

• next