Petr Volkov,1 Karl Bacos,1 Jones K. Ofori,2 Jonathan Lou S. Esguerra,2

Lena Eliasson,2 Tina Rönn,1 and Charlotte Ling1

Whole-Genome Bisulfite Sequencing ofHuman Pancreatic Islets Reveals NovelDifferentially Methylated Regions inType 2 Diabetes PathogenesisDiabetes 2017;66:1074–1085 | DOI: 10.2337/db16-0996

Current knowledge about the role of epigenetics in type 2diabetes (T2D) remains limited. Only a few studies haveinvestigated DNAmethylation of selected candidate genesor a very small fraction of genomic CpG sites in humanpancreatic islets, the tissue of primary pathogenic impor-tance for diabetes. Our aimwas to characterize the whole-genome DNA methylation landscape in human pancreaticislets, to identify differentially methylated regions (DMRs)in diabetic islets, and to investigate the function of DMRsin islet biology. Here, we performed whole-genome bi-sulfite sequencing, which is a comprehensive and un-biased method to study DNA methylation throughout thegenome at a single nucleotide resolution, in pancreaticislets from donors with T2D and control subjects withoutdiabetes. We identified 25,820 DMRs in islets from indi-viduals with T2D. These DMRs cover loci with known isletfunction, e.g., PDX1, TCF7L2, and ADCY5. Importantly,binding sites previously identified by ChIP-seq for islet-specific transcription factors, enhancer regions, and dif-ferent histone marks were enriched in the T2D-associatedDMRs. We also identified 457 genes, including NR4A3,PARK2, PID1, SLC2A2, and SOCS2, that had both DMRsand significant expression changes in T2D islets. To mimicthe situation in T2D islets, candidate genes were overex-pressed or silenced in cultured b-cells. This resulted inimpaired insulin secretion, thereby connecting differen-tial methylation to islet dysfunction. We further exploredthe islet methylome and found a strong link betweenmethylation levels and histone marks. Additionally, DNAmethylation in different genomic regions and of differenttranscript types (i.e., protein coding, noncoding, and pseu-dogenes) was associated with islet expression levels. Our

study provides a comprehensive picture of the islet DNAmethylome in individuals with and without diabetes andhighlights the importance of epigenetic dysregulation inpancreatic islets and T2D pathogenesis.

Impaired insulin secretion is a key feature of type 2 diabetes(T2D). However, the molecular mechanisms underlyingpancreatic islet dysfunction in T2D are largely unknown.Although genetic risk factors are known to contribute toT2D, ,20% of the estimated T2D heritability can beexplained by single nucleotide polymorphisms (SNPs) iden-tified by genome-wide association studies (1). Hence, onemust look elsewhere to find disease-causing mechanisms.Given the important role of environmental factors in T2Dpathogenesis, mechanisms mediating gene-environment in-teractions, such as epigenetics, may be of particular signifi-cance. Studying epigenetic processes in the tissue of primarypathogenetic importance, the pancreatic islets, may revealcentral mechanisms for T2D. Indeed, we and others haveidentified altered DNA methylation in islets from humanT2D donors (2–4). However, the methods used in previousstudies only covered up to ;1.5% of genomic CpG sites (5).To obtain a more complete picture of the human islet meth-ylome and to further dissect the impact of epigenetics in T2D,analyses covering the majority of CpG sites are needed.

Whole-genome bisulfite sequencing (WGBS) is the mostcomprehensive method to study DNA methylation on asingle nucleotide resolution. This method is costly and moststudies have so far been based on few samples in selectedsets of human tissues (6–8). The largest effort to describe

1Epigenetics and Diabetes Unit, Department of Clinical Sciences, Lund UniversityDiabetes Centre, Scania University Hospital, Malmö, Sweden2Islet Cell Exocytosis Unit, Department of Clinical Sciences, Lund University Di-abetes Centre, Scania University Hospital, Malmö, Sweden

Corresponding author: Charlotte Ling, charlotte.ling@med.lu.se.

Received 15 August 2016 and 28 December 2016.

This article contains Supplementary Data online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db16-0996/-/DC1.

T.R. and C.L. contributed equally to this study.

© 2017 by the American Diabetes Association. Readers may use this article aslong as the work is properly cited, the use is educational and not for profit, and thework is not altered. More information is available at http://www.diabetesjournals.org/content/license.

1074 Diabetes Volume 66, April 2017

GENETIC

S/G

ENOMES/P

ROTEOMIC

S/M

ETABOLOMIC

S

the human epigenome was performed by the National In-stitutes of Health Roadmap Epigenomics Consortium (9).However, previous WGBS studies did not include humanislets or T2D case-control cohorts. To address this knowl-edge gap and to dissect epigenetic alterations in T2D, weperformed WGBS of human islets from control subjectswithout diabetes and T2D donors.

RESEARCH DESIGN AND METHODS

Human IsletsHuman islets were obtained from the Nordic Network forIslet Transplantation. The islet cohort is described in theSupplementary Data.

WGBSA total of 300 ng of islet DNA was bisulfite treated withthe EZ DNA Methylation-Gold D5005 kit (Zymo ResearchCorp., Irvine, CA) according to the manufacturer’s pro-tocol (Supplementary Data).

WGBS Data AnalysisWGBS data in FASTQ format generated using the IlluminaHiSeq platform (Illumina, San Diego, CA) were used forfurther analyses (Supplementary Data).

Infinium 450K ArrayDNA methylation of human islets (Table 1) was analyzedwith the Infinium HumanMethylation450K BeadChip (5)(Supplementary Data).

RNA-seq DataThe RNA library was prepared with the TruSeq kit (Illumina)and then sequenced with HiSeq2000 as previously described(10) (Supplementary Data).

Genomic AnnotationGenomic elements, such as transcription start sites (TSS),transcription end sites (TES), and exons and introns for198,442 transcripts corresponding to 60,483 genes, wereextracted from GENCODE version 22 (GRCh38). Eachdifferentially methylated region (DMR) was annotatedbased on its position in relation to all transcripts above;hence, one DMR can have multiple annotations.

PyrosequencingDNA methylation for biological replication of a genomicregion in the most significant PDX1 DMR (chr13:27921804:27925104) and regions in DMRs annotated to ARX, CADM1,

PARK2, PID1, SLC2A2, and SOCS2 was analyzed by pyro-sequencing and the PyroMark Q96 ID (Qiagen, Hilden,Germany). Primers were designed using PyroMark AssayDesign 2.0 (Supplementary Table 1). Procedures were per-formed according to recommended protocols.

Luciferase AssaysLuciferase assays were performed as previously described(4) (Supplementary Data).

b-Cell LinesINS-1 832/13 rat b-cells were used in functional experi-ments as previously described (2) (Supplementary Data).

Statistical AnalysisDonor characteristics, pyrosequencing data, and luciferaseexperiments were analyzed using unpaired t tests (Sup-plementary Data).

RESULTS

WGBS in Human IsletsTo characterize the methylome in human islets, we generatedWGBS data from eight control subjects and six T2D donors(Table 1). An average of 74% of the resulting reads per samplewas uniquely mapped to the human reference genome (hg38).Sequencing information and alignment statistics are reportedin Supplementary Table 2. The samples were sequenced withan average coverage of 213 per base and methylation levelsof;2.43 107 CpG sites (;83% of all genomic CpG sites) onthe forward strand were obtained for all samples.

We compared the methylation data obtained by WGBSwith data generated with the Infinium 450K array for thesame 14 samples. The replicates of each islet sample analyzedby both WGBS and microarray showed high reproducibility(Supplementary Table 2).

One islet sample was analyzed by WGBS using bothIllumina type 3 and 4 chemistry and by two librarypreparation kits; the TruSeq (EpiGnome) DNA MethylationKit and the NEXTflex Bisulfite Library Prep Kit. The highcorrelation between the WGBS data generated with the twodifferent sequencing chemistries (r = 0.995) and with dif-ferent library preparations (r = 0.987) further confirmedthe data quality (Supplementary Fig. 1A).

Human Islet MethylomeWe first sought to characterize the overall variability of thehuman islet methylome. The degree of DNA methylation

Table 1—Characteristics for donors of pancreatic islets included in the WGBS analysis

Donors without diabetes (n = 8) Donors with T2D (n = 6) P value

Sex (M/F) 4/4 3/3

Age (years) 52.5 6 3.2 (40–67) 58.2 6 3.6 (45–66) 0.26

BMI (kg/m2) 24.9 6 0.3 (23.9–26.6) 28.0 6 2.0 (22.9–34.6) 0.10

HbA1c (%) 5.47 6 0.10 (n = 7) 7.12 6 0.21 ,0.0001

HbA1c (mmol/mol) 36.4 6 1.1 (n = 7) 54.2 6 2.3 ,0.0001

Stimulatory index 3.7 6 0.47 6.08 6 1.94 0.28

Data are presented as mean 6 SEM (range). Two-tailed P values and t tests were used to detect differences between groups.

diabetes.diabetesjournals.org Volkov and Associates 1075

throughout the genome was highly correlated among theanalyzed islets samples (r = 0.973–0.989) and the averagemethylation level was 75.9%. The distribution of the meth-ylation level in human islets is bimodal, with the highestpeak at 90.2%, showing that most CpG sites are highlymethylated, and the second peak at 1.4%, representingCpG sites with very low levels of methylation (Fig. 1A).

Next, we computed the average genome-wide methyl-ation level in relation to different genomic regions in humanislets. While introns and exons had the overall highestdegree of methylation (78.5% and 77.4%), regions close tothe TSS, such as the 1st exon and promoter regions (TSS200 and TSS 1500), had the lowest degree of methylation(34.7%, 25.4%, and 44.4%, respectively). Regions moredistant to the TSS had a rather high methylation level(TSS 50 kb 73.0% and TES 10 kb 71.2%).

We also studied methylation based on annotation todifferent types of transcripts; protein-coding genes, non-coding RNAs, and pseudogenes (Fig. 1B–D). There was astriking difference in average DNA methylation betweenthe different transcript types, where the typical drop inmethylation seen close to the TSS for protein-coding genes(TSS 1500, TSS 200, and 1st exon) was less pronounced innoncoding RNAs and almost completely absent in pseudo-genes. Additionally, there were no differences in averagemethylation for either genomic regions or transcript typesbetween islets from T2D and control donors (P = 0.4–1.0).

To gain further insight into how the human isletepigenome is coordinated, we studied the relationshipbetween DNA methylation and histone modificationsacross the genome. Here, we integrated our WGBS datawith histone modifications generated in human islets by theRoadmap Epigenomics Consortium (9). Regions occupied byhistone modifications associated with active chromatin hadthe lowest degree of methylation (11.8% for H3K9ac and7.8% for H3K4me3), whereas regions occupied by modifi-cations associated with repressive chromatin had highermethylation levels (49.6% for H3K27me3 and 80.2% forH3K9me3) (Fig. 1E). Additionally, regions occupied by his-tone modifications considered to be enriched at enhancerregions had either quite low (21.4% for H3K27ac) or high(69.8% for H3K4me1) degree of methylation (Fig. 1E). Wealso examined the relationship between DNA methylationlevels and genomic binding of islet-specific transcriptionfactors (PDX1, FOXA2, MAFB, NKX6.1, and NKX2.2) bycombining our WGBS data with published ChIP-seq data(11). Regions occupied by transcription factors had a lowerdegree of methylation (range 20.9%–43.9%) compared withthe whole genome (Fig. 1F). There were no differences inthe methylation levels in any of the regions occupied by thestudied histone marks or transcription factors in controlversus T2D islets (Fig. 1E and F) (P = 0.4–1.0).

We proceeded to explore the relationship between DNAmethylation and gene expression levels using WGBS andRNA-seq data from the same 14 human islet donors.Here, we categorized 60,483 transcripts into not expressed(TPM,0.1, 38,261 transcripts) and equally sized groups of

low- (7,407), medium- (7,407), and high-expressed (7,408)transcripts. We also studied protein-coding genes, noncod-ing RNAs, and pseudogenes separately. We found an as-sociation between DNA methylation and the expressionlevel in all genomic regions and for all types of transcripts(P , 0.0001) (Fig. 1G–I). Importantly, protein-codinggenes not expressed were hypermethylated in regions closeto the TSS (TSS 1500, TSS 200, and 1st exon) (Fig. 1G).However, protein-coding nontranscribed genes had signif-icantly lower methylation levels in exon and intron regionscompared with the transcribed genes (Fig. 1G), supportingthe hypothesis that increased methylation in the gene bodyis associated with a higher gene transcription (12). Moremodest differences in methylation were seen when com-paring the different expression levels of transcribed genes,i.e., dividing transcribed genes in low-, medium-, and high-expressed genes (Fig. 1G–I). Some similarities in the meth-ylation pattern were seen for both noncoding RNAs andpseudogenes compared with protein-coding genes (Fig. 1Hand I). However, regions close to the TSS were hyperme-thylated in nontranscribed genes and the methylation levelin these regions was much higher in transcribed noncodingRNAs and pseudogenes (Fig. 1H and I).

Principal Component Analysis of the Islet WGBS DataIn an unsupervised principal component analysis of the isletWGBS data, the methylation data was segregated accordingto sex (Supplementary Fig. 1B), which is in agreement withour published 450K array data (13). Next, we correlated thetop five principal components of the WGBS data with T2D,age, sex, and BMI. Here, T2D and sex correlated with one ofthe top five principal components (P , 0.002) (Table 2).

DMRs in Human T2D IsletsTo address the epigenetic basis of T2D, we used BSmooth(14) to analyze the islet WGBS data. BSmooth determinesmethylation in WGBS data and identifies DMRs thataccount for biological variability. DMRs were defined as re-gions of three or more consecutive differentially methylatedsites with an average absolute methylation difference $5%between groups. This analysis identified 25,820 DMRs inT2D islets (Supplementary Table 3). These were includedin an unsupervised hierarchical clustering analysis presentedas a heatmap in Fig. 2A, which shows distinction in meth-ylation between donors with and without diabetes. A totalof 13,696 DMRs showed average increased and 12,124 de-creased levels of methylation in T2D islets. The mean DMRsize was 414 bp (range 6–3,411 bp), and the mean CpG sitecount in the DMRs was 8.7 (range 3–164). The maximummethylation difference for a DMR was 27.5% when compar-ing diabetic versus control islets, and 692 DMRs had amethylation difference $10% (Supplementary Table 3).Among the DMRs found to have the largest absolute differ-ences in methylation were regions annotated to ARX andTFAM (Fig. 2B and C), both important in islet function(15,16).

Interestingly, two of the most significant DMRs covered164 and 105 CpG sites that span 3,301 and 2,676 bp

1076 WGBS of Human Islets Diabetes Volume 66, April 2017

regions of PDX1 (Fig. 2D and E), a key islet transcriptionfactor (17). In total, seven DMRs were annotated toPDX1 (Supplementary Table 3). Using pyrosequencing, we

biologically replicated differential methylation of sites in themost significant DMR, located in PDX1 (chr13:27921804-27925104) (Fig. 2D and F), in an independent cohort

Figure 1—The DNA methylome of human pancreatic islets. A: The distribution of DNA methylation in human islets. B–D: Average DNAmethylation in human pancreatic islets, separated by different Gencode transcript types (protein-coding, noncoding [long and small combined],and pseudogenes) and gene regions. Here, TSS 50 kb represents 1,501–50,000 bp upstream from the TSS, TSS 1500 represents 201–1,500 bpupstream from the TSS, TSS 200 represents 1–200 bp upstream from the TSS, and TES 10 kb represents 1–10,000 bp downstream of the TES.Donors with T2D and normoglycemic control subjects display no differences in genome-wide average methylation for any genomic region ortranscript type (P = 0.4–1.0). E and F: Average DNA methylation for regions overlapping with different histone marks or transcription factorbinding sites. G–I: Average methylation levels are significantly different between transcripts of different expression levels. Here, we identified thenot expressed transcripts based on transcript per million<0.1 and then divided the expressed transcripts into three equally sized groups that wecategorize into low-, medium-, and high-expressed transcripts. Additionally, different genomic regions display specific methylation patterns thatare also dependent on transcript type. Data presented as mean 6 SEM (***P < 0.0001, as analyzed by ANOVA).

diabetes.diabetesjournals.org Volkov and Associates 1077

(Supplementary Table 4). We previously reported increasedmethylation of 10 sites and decreased expression of PDX1in human T2D islets (4). Importantly, a DMR from thecurrent study (chr13:27918705-27919232) confirmed ourprevious finding, and comparisons of seven PDX1 sites cov-ered by both studies validated the significant associationwith T2D (Fig. 2G).

We next examined the overlap between T2D-associatedislet DMRs and 65 T2D candidate genes identified bythe DIAGRAM (DIAbetes Genetics Replication And Meta-analysis) consortium (18). Among our islet DMRs, 159 wereannotated to 43 known T2D candidate genes (Supplemen-tary Table 5). Out of these, the DMR with the largestdifference in methylation between donors with diabetesand donors without diabetes was located in ADCY5. T2Dcandidate genes with the highest number of DMRs wereGLIS3, THADA, KCNQ1, and TCF7L2 (16, 14, 10, and 9,respectively).

We further investigated if our islet DMRs covered anyof the SNPs reported to be associated with T2D (www.genome.gov/gwastudies, accessed 2 February 2016). Among theseSNPs, rs163184 was covered by a DMR annotated toKCNQ1 (chr11:2825382:2826548) and rs11257655 by aDMR annotated to RP11 (chr10:12265391:12266540).

Distribution of T2D-Associated Islet DMRsWe then examined the genomic distribution of T2D-associated islet DMRs. We found ;55% of the DMRslocated within TSS 50 kb (1,501–50,000 bp upstreamfrom the TSS), ;1.5% located within TSS 1500 (201–1,500 bp upstream from the TSS), and ;1.0% locatedwithin TSS 200 (1–200 bp upstream from the TSS) (Fig.3A). Furthermore, ;1.4% of all DMRs were located in the1st exon, ;3.0% in subsequent exons, and ;20% in in-trons, whereas ;12.5% of the DMRs were located withinTES 10 kb (1–10,000 bp downstream of the TES). Finally,;5%–6% of all DMRs were located .50 kb from thenearest transcript and considered intergenic.

To explore the relationship between chromatin stateand DNA methylation, we integrated our DMRs withpublished maps of histone modifications and enhancerregions in human islets (9,11). When comparing ourT2D-associated islet DMRs with epigenetic marks gener-ated in human islets by the Roadmap Epigenomics

Consortium (9), 12.4% of our islet DMRs were occupiedby modifications associated with active chromatin (3,203by H3K4me3 and 3,194 by H3K9ac) (Supplementary Table6 and Fig. 3B), which are significant enrichments for both(P , 1 3 1026). Here, DMRs with higher methylation indonors with diabetes occupied 1,872 H3K4me3 and 1,851H3K9ac marks. Moreover, 14.5% and 19.1% of islet DMRswere occupied by histone modifications enriched at enhancerregions (3,744 by H3K27ac and 4,935 by H3K4me1, respec-tively) (P , 1 3 1026). Here, DMRs with higher methyl-ation in donors with diabetes occupied 2,358 H3K27ac and2,478 H3K4me1 marks. A smaller fraction of the DMRswere occupied by modifications associated with repressedchromatin (1,364 DMRs [5.3%] by H3K27me3 and120 DMRs [0.5%] by H3K9me3). H3K27me3 was over-represented, whereas H3K9me3 was underrepresented(P , 1 3 1026) in this overlap. Moreover, DMRs withhigher methylation in donors with diabetes occupied583 H3K27me3 and 51 H3K9me3 marks. There was alsoan underrepresentation of DMRs (581, or 2.3%) overlap-ping with H3K36me3 (P , 1 3 1026), and out of those,381 showed higher methylation in donors with diabetes.Additionally, 618 islet DMRs overlapped with active en-hancer regions identified in islets (Supplementary Table 6and Fig. 3B) (11), which is more than expected by chance(P , 1 3 1026), and out of those, 381 showed highermethylation in donors with diabetes.

Next, Hypergeometric Optimization of Motif EnRichment(HOMER) analysis (19) showed that motifs specific to keytranscription factors in islets, including FOXA2, NeuroD1,MAFA, RFX, PDX1, and HNF1, and binding sites for theinsulator CCCTC binding factor were significantly enrichedin T2D-associated islet DMRs (Fig. 3C and SupplementaryTable 7). We also identified novel motifs with statisticalenrichment (Supplementary Table 7).

To gain insight into the relationship between thegenomic binding of islet-specific transcription factors anddifferential methylation in islets from T2D donors, we usedChIP-sequencing data for five key transcription factors(FOXA2, MAFB, NKX2.2, NKX6.1, and PDX1) in humanislets (11). We found an overrepresentation of DMRs over-lapping with the binding sites of each of these transcrip-tion factors (P , 1 3 1026) (Supplementary Table 8 andFig. 3B). Interestingly, we found an overlap between T2D-associated DMRs annotated to SLC2A2, KCNJ11, andPDX1 and sites bound by PDX1 (Supplementary Table 8).These data suggest that differential methylation in tran-scription factor binding sites may be of importance in T2D.

Tissue-Specific DMRsTo identify DMRs in human islets that are also alteredbetween other tissues or cell types, we analyzed the overlapbetween our T2D-associated islet DMRs and a set of716,087 cross-tissue dynamic DMRs (20). Of our 25,820DMRs, 12,911 (49.8%) overlapped with a dynamic DMRidentified in other tissues (P , 1 3 1026) (SupplementaryTable 9).

Table 2—P values for correlations of the top five principalcomponents for the WGBS data in human pancreatic isletswith T2D, age, sex, and BMI

Principalcomponent T2D

Age(years)

Sex(M/F)

BMI(kg/m2)

1 0.70 0.54 0.65 0.26

2 0.90 0.13 0.0017 0.47

3 0.25 0.19 0.41 0.62

4 0.07 0.58 0.75 0.11

5 0.0019 0.44 0.95 0.22

Data in boldface type are significant correlations (P , 0.05).

1078 WGBS of Human Islets Diabetes Volume 66, April 2017

Figure 2—DMRs in human pancreatic islets from donors with T2D. A: Heatmap of the 25,820 T2D-associated DMRs. Among the DMRswith the largest absolute difference were regions upstream of ARX (B) and TFAM (C). Among the most significant DMRs were two largeintergenic regions of PDX1, chr13:27921804-27925104 (D) and chr13:27926170-27928845 (E). F: Biological replication of the PDX1 regionlocated in intronic regions and exon two and presented in panel D in human pancreatic islets from an independent cohort of 56 normo-glycemic control and 19 T2D donors. G and H: Validation of DNA methylation in the PDX1 distal promoter in human pancreatic islets fromT2D and control donors. G: Data from Yang et al. (4), produced using Sequenom’s EpiTYPER technology. H: Data from current study,based on WGBS DMR chr13:27918705-27919232. Data presented as mean 6 SEM (*P < 0.05).

diabetes.diabetesjournals.org Volkov and Associates 1079

Altered Expression of Genes Annotated toT2D-Associated DMRsTo determine whether genes annotated to T2D-associatedDMRs also show altered expression in T2D versus controlislets, we combined the 25,820 DMRs presented in Supple-mentary Table 3 with islet RNA-seq data of a previouspublication (10). We identified 457 genes that had bothsignificantly altered expression (false discovery rate ,5%;q,0.05) and DMRs in T2D islets (Supplementary Table 10).These include genes important for islet function and

metabolism, such as CACNA1D, CHL1, GLP1R, IGF1R, IL6,NR4A3, PARK2, PDX1, PID1, SEPT9, SIK2, SLC2A2(also known as GLUT2), SOCS2, and SOX6 (Fig. 4A)(2,13,21–32). Moreover, 26 genes exhibited differential ex-pression and DMRs with a methylation difference $10%.

Next, we performed a KEGG (Kyoto Encyclopedia ofGenes and Genomes) pathway analysis using WebGestalt(33) to identify biological pathways with enrichment ofgenes that had both significantly altered expression andDMRs (n = 457) in islets from T2D donors. These genesare significantly enriched for categories including Ribo-some, Jak-STAT signaling pathway, Pathways in cancer,Type II diabetes mellitus, and Metabolic pathways (ad-justed P , 0.006) (Fig. 4B and Supplementary Table 11),further supporting the hypothesis that epigenetic and tran-scriptional changes in islets may contribute to altered me-tabolism and T2D.

We then functionally studied the impact of alteredmethylation on the transcriptional activity using luciferaseassays. We selected TMED6 and KIF3A, two genes where theT2D-associated DMRs cover promoter regions and with aninverse relation between methylation and expression (Sup-plementary Fig. 1C and Supplementary Table 10). TMED6has also been linked to insulin secretion (10). Luciferaseconstructs containing respective promoter sequence wereeither methylated or mock-methylated and transfectedinto clonal b-cells. In line with our islet data, complete meth-ylation resulted in reduced transcriptional activity of bothpromoters, whereas partial methylation only reduced KIF3Atranscriptional activity (Supplementary Fig. 1D and E).

We then asked if genes with both DMRs and alteredexpression in human T2D islets have a functional role inb-cells. Genes were selected for functional follow-up basedon having multiple and/or large DMRs and exhibiting dif-ferential expression (q ,0.05) in diabetic islets (Supple-mentary Table 10 and Fig. 4A and C) together with apotential role in b-cell function (22,25,27,31). To modelthe situation in T2D islets, we overexpressed Nr4a3,Pid1, and Socs2 and silenced Park2 in rat clonal b-cells (Fig.4D–G and J). We then measured insulin secretion at basal(2.8 mmol/L) and stimulatory (16.7 mmol/L) glucose lev-els. Overexpression of Pid1 and Socs2 resulted in reducedglucose-stimulated insulin secretion (GSIS), and all threeoverexpressed genes caused a slight increase in basal in-sulin secretion (Fig. 4H). These changes resulted in de-creased fold change of insulin secretion (secretion atstimulatory divided by the secretion at basal glucose levels)in b-cells overexpressing either gene (Fig. 4I), which is inline with what is seen in donors with diabetes (2). Park2deficiency resulted in reduced GSIS (Fig. 4J and K) but didnot affect fold change of insulin secretion (Fig. 4L). Thesesecretory defects were not due to altered insulin content(data not shown).

Replication of Differential DNA Methylation in DMRsWe finally used pyrosequencing to replicate differentialDNA methylation of some CpG sites in DMRs annotated

Figure 3—Genomic distribution of T2D-associated DMRs. A:Genomic distribution of T2D-associated islet DMRs, separatedbased on increased (n = 13,696) or decreased (n = 12,124) av-erage DNA methylation. Each DMR can be annotated to severaltranscripts and will then be counted in all gene regions, exceptfrom intergenic DMRs that only include DMRs located more than50 kb from a transcript and thereby are not annotated to any ofthe TSS, intragenic, or TES regions. B: Overlap between ourT2D-associated DMRs and chromatin state, transcription factorbinding, and active enhancer regions. Arrows represent signifi-cant over- or underrepresentation (P < 1 3 1026). C: Enrichment oftranscription factor recognition sequences in the T2D-associatedislet DMRs based on HOMER (19). CTCF, CCCTC binding factor.

1080 WGBS of Human Islets Diabetes Volume 66, April 2017

Altered metabolism and islet func�on

CACNA1D, CHL1, GLP1R, IGF1R, IL6, NR4A3, PARK2, PDX1, PID1, SEPT9,

SIK2, SLC2A2, SOCS2, SOX6and

Altered DNA methyla�on

Altered mRNA levels

A B

Metabolic pathways (Padj=2.5x10-3)ADPGK, AHCY, ARG2, ATP6V0A1, B3GNT3, B4GALT6, CHSY1, CYP2U1,

GALNT2, GFPT2, GNS, HADH, HIBADH, ITPA, MGAT2, MGLL, NME1, NME2, NNT, PLCD3, PTGDS, PTGES, PYCRL, SRM, ST3GAL5

Ribosome (Padj=1.1x10-10)FAU, RPLP2, RPL3L, RPL10A, RPL18, RPL18A, RPL28, RPL29, RPS2,

RPS16, RPS18, RPS19, RPS21, RPS27

Type II diabetes mellitus (Padj=1.0x10-4)CACNA1D, PDX1, PIK3R3, SLC2A2, SOCS2, SOCS3

Pathways in cancer (Padj=1.56x10-8)ARNT2, BCL2L1, CBLC, CDKN1A, CTNNA2, FGF7, FZD7, IGF1R, IL6, JUP ITGA2, MMP1, MSH6, MYC, PIAS1, PIK3R3, PPARD, RUNX1, TCF7, TGFA

Jak-STAT signaling pathway (Padj=7.0x10-9)BCL2L1, CBLC, CLCF1, CSF3, IL11, IL15RA, IL2RA, IL6, IL23A, LIF, MYC,

PIAS1, PIK3R3, SOCS2, SOCS3

D E F

J

H

G

K

C

I

L

Figure 4—Altered expression, T2D-associated DMRs, and functional consequences in b-cells. In total, 457 genes had both significantlyaltered expression in pancreatic islets from donors with T2D (q <0.05) (10) and significant DMRs (Supplementary Table 10). These includegenes of importance in islet function and metabolism as depicted in panel A. B: Selected KEGG pathways based on enrichment of genesannotated to T2D-associated DMRs that also show altered RNA expression in human pancreatic islets from donors with T2D (10). A total of457 genes were included in the pathway analysis and the boxes include genes contributing to the enrichment score for each pathway. A fulllist of significantly enriched KEGG pathways is found in Supplementary Table 11. C: DMRs from the four genes selected for functional

diabetes.diabetesjournals.org Volkov and Associates 1081

to ARX, CADM1, PARK2, PID1, SLC2A2, and SOCS2 (Supple-mentary Table 3) in an independent cohort (SupplementaryTable 4). Here, we could replicate differential methylation inT2D versus control islets (Supplementary Table 12).

Additionally, as a technical replication, we correlatedmethylation data for CpG sites in islet DMRs (Supplemen-tary Table 3) overlapping with sites analyzed by the Infin-ium 450K array for the same 14 islet samples. Here, theWGBS and 450K array data showed high reproducibility(Supplementary Table 13).

DISCUSSION

Alternative approaches are needed to end the rapidincrease in T2D incidence. This study provides valuableinsights into T2D pathology and human biology, includingthe first comprehensive DNA methylation analysis of24 million sites in a case-control cohort of human islets.We identified 25,820 DMRs in islets from T2D versuscontrol donors. These cover loci annotated to genesimportant for islets and T2D pathogenesis as well as novelcandidates. The identified DMRs were enriched in bothexperimentally identified and putative binding motifs forislet-specific transcription factors. Integrating our WGBSdata with RNA-seq data further identified novel candidategenes that contribute to impaired insulin secretion.

Intriguingly, seven of the identified DMRs were anno-tated to PDX1, which encodes a transcription factor of keyimportance during pancreatic development and in matureb-cells where it regulates insulin expression (34). Addition-ally, mutations in PDX1 cause maturity-onset diabetes ofthe young 4 (35), knockout of Pdx1 in b-cells causes di-abetes (21), and epigenetic changes of Pdx1 in islets of ratsexposed to an impaired intrauterine environment predis-pose to future diabetes (36). Using a candidate gene ap-proach, we previously found altered methylation andexpression of PDX1 in T2D islets (4). One PDX1 DMR inthe current study validated these previous data. We alsofound that glucose directly increases Pdx1 methylation incultured b-cells (4), and a SNP in PDX1 associated withhyperglycemia alters PDX1 methylation in human islets(37). These data support the hypothesis that epigeneticmodifications of PDX1 may contribute to diabetes.

The array-based methods used in previous islet studiesonly covered;450,000 and;27,000 CpG sites, respectively(2,3), whereas the current study covered ;24 million sites.Additionally, we studied DMRs, whereas previous studies

analyzed individual CpG sites, making it difficult to comparethe results among the different studies. However, we founda strong correlation between DNA methylation analyzedwith the 450K array and WGBS from the same islet samples.Additionally, regions/sites annotated to 47 genes werefound to have differential methylation by all three studies.

The majority of SNPs associated with T2D impact insulinsecretion, supporting their importance in islet function(38). However, these SNPs only explain a modest propor-tion of the estimated heritability of T2D, and it is possiblethat combinations of genetic and epigenetic variationcontribute to disease susceptibility (37,39). This hypothesisis supported by the fact that a large number of our T2D-associated DMRs are located in the same regions as T2Dcandidate genes identified by genome-wide associationstudies, such as TCF7L2, ADCY5, KCNQ1, and GLIS3. Func-tional studies further show the importance of some of thesecandidate genes, e.g., TCF7L2 and ADCY5, in b-cells (40,41).

DMRs overlapping with enhancer regions and transcrip-tion factor binding sites may have important roles in generegulation and disease development (42). Indeed, these re-gions were overrepresented in the DMRs we detected. Arecent study proposed competition between DNA methyl-ation and binding of transcription factors to DNA (43). Inrelation to their data, it is worth mentioning that we found20.9%–43.9% methylation in regions occupied by transcrip-tion factors such as PDX1, FOXA2, MAFB, NKX6.1, andNKX2.2. It is possible that these transcription factors bindto DNA in a subset of islet cells with hypomethylated DNA,whereas they may not bind to DNA in the cells with meth-ylated DNA. Methylation may thereby regulate gene expres-sion differentially in different islet cells. The fact that motifsfor several islet-specific transcription factors were enrichedin T2D-associated islet DMRs further supports an impor-tant role of methylation in these regulatory regions.

When combining the identified DMRs with RNA-seqdata from islets of T2D and control donors, we identified457 genes with both altered methylation and expression.For example, SLC2A2 had increased methylation and de-creased expression in T2D islets. SLC2A2 encodes GLUT2,which is a major glucose transporter in rodent islets,whereas other glucose transporters have been suggestedto be more important in human b-cells (44). Nevertheless,Sansbury et al. (45) reported a loss of function of SLC2A2as a cause of neonatal diabetes, suggesting a role forGLUT2 in human b-cells. Additionally, we could replicate

studies; NR4A3, PID1, SOCS2, and PARK2. Overexpression of Nr4a3, Pid1, and Socs2 in INS-1 832/13 cells was verified by qPCR (D–F)and Western blot (G). ***P < 0.001 and **P < 0.01, as analyzed by one-tailed paired t tests (n = 7). H: Overexpression of Nr4a3, Pid1, andSocs2 resulted in perturbed insulin secretion (*P < 0.05 compared with pcDNA3.1 at 16.7 mmol/L glucose; #P < 0.01 compared withpcDNA3.1 at 2.8 mmol/L glucose, as analyzed by two-tailed paired t tests; n = 7). I: Altered fold change of insulin secretion (calculated asinsulin secretion at 16.7 mmol/L glucose divided by insulin secretion at 2.8 mmol/L glucose) for each of the three genes (*P< 0.05). J: Smallinterfering RNA–mediated knockdown of Park2 was verified by qPCR (**P < 0.01, as analyzed by a one-tailed paired t test; n = 6). K: Park2deficiency resulted in reduced insulin secretion at stimulatory glucose levels (*P < 0.05, as analyzed by a two-tailed paired t test; n = 6) butdid not affect fold change of insulin secretion (L). siNC, negative control siRNA; siPark2, siRNA targeting Park2.

1082 WGBS of Human Islets Diabetes Volume 66, April 2017

increased methylation of SLC2A2 in T2D islets in an in-dependent cohort.

To further model human T2D, we performed functionalfollow-up experiments of additional genes exhibiting DMRsand altered expression in diabetic islets. We overexpressedNr4a3, which reduces insulin expression by modulatingPdx1 and NeuroD1 expression (46); Socs2, which regulatesproinsulin processing and insulin secretion in mice (27); andPid1, which has been implicated in mitochondrial dysfunc-tion (22). Overexpression of all genes decreased the foldchange of insulin secretion, which is in line with what isseen in human diabetic islets. We also silenced Park2 expres-sion, which regulates the mitochondrial control system inb-cells (25). Again, this impaired GSIS. These experimentssupport the hypothesis that genes identified by WGBS andRNA-seq may contribute to islet dysfunction in T2D.

DNA methylation was initially thought to be a silencingmark; however, recent data show that its function mayvary with genomic context and is more complex thaninitially thought (12). Here, we integrated the islet meth-ylome with histone modifications analyzed by the Road-map Epigenomics Consortium (9). Notably, Kundaje et al.(9) generated WGBS data in several tissues but only re-duced representation bisulfite sequencing methylationdata for human islets. Our study is hence the first to pro-vide WGBS coverage in human islets. As may be expected,histone marks enriched around the TSS of actively tran-scribed genes (H3K9ac, H3K27ac, and H3K4me3) and ac-tive enhancers (H3K27ac) had a relatively low methylationlevel, whereas histone marks enriched around inactive TSS(H3K27me3) or inactive regions (H3K9me3) had a mediumor high methylation level. We also found a relatively highmethylation level in regions enriched with H3K4me1,which is found at enhancer regions and gene bodies ofactively transcribed genes. Thus, the high methylationlevel in regions enriched for H3K4me1 is in line withthe high methylation level that is often found in genebodies (47). We also integrated the islet methylome andtranscriptome and examined the degree of methylation indifferent genomic regions of protein-coding genes, non-coding RNAs, and pseudogenes. Interestingly, meth-ylation of different transcript types shows differentpatterns. The relation to gene expression was also distinctwhen comparing nonexpressed and expressed transcripts,with reduced impact with increasing distance from TSS.Our data clearly show that the relationship between DNAmethylation and gene expression depends on transcripttype, expression level, and distance from the TSS.

As DMRs may be cell-type specific (20), it is importantto consider the impact of cellular heterogeneity. Here, allsamples had a high purity, i.e., a high endocrine content,without differences between the groups. Furthermore, wehave previously shown that there is no difference in b-cellcontent in the pancreatic islets from donors with T2Dversus control subjects without diabetes (2). These datasuggest that the identified T2D-associated DMRs are notdue to altered cell composition.

It has been suggested that minimal sequencing require-ments in WGBS experiments starts from 53 (48). How-ever, it was also stated that higher coverage is required todetect shorter DMRs with smaller methylation differences.On the basis of our previous findings, we expected someCpG sites to show absolute differences in methylation lessthan 10% between islets from T2D and control donors (2);hence, we implemented a higher coverage (213) in ourstudy design and used an average difference in methylationof 5% as a threshold for the DMRs. Moreover, T2D is apolygenic disease where modest effect sizes of a large num-ber of genes are expected to contribute to the disease. In-deed, we discovered numerous DMRs that show absolutemethylation differences between 5% and 10%. It shouldalso be noted that some individual CpG sites in a DMRwith an average methylation difference of 5% will showbigger differences in methylation.

This comprehensive study identified novel diabetes-related changes in DNA methylation that support animportant role for epigenetics in T2D. We need tocombine multiple layers of biological information tounderstand the pathogenesis and progression of T2D.Here, we combined WGBS and RNA-seq data fromhuman islets with known regulatory elements such ashistone marks, transcription factor binding sites, andenhancer regions. These integrated data advance ourunderstanding of the etiology of T2D and highlight theimportance of epigenetic dysregulation in pancreaticislets and T2D pathogenesis.

Acknowledgments. Sequencing was performed by the SNP&SEQ Tech-nology Platform in Uppsala, Sweden. The facility is part of the National GenomicsInfrastructure Sweden and Science for Life Laboratory. The SNP&SEQ Platform isalso supported by the Swedish Research Council and the Knut and Alice WallenbergFoundation. Human pancreatic islets were obtained through collaboration with OlleKorsgren at the Nordic Network for Islet Transplantation (Uppsala University,Sweden) and the Human Tissue Laboratory at Lund University Diabetes Centre,coordinated by Ulrika Krus. The authors thank Anna-Maria Veljanovska-Ramsay,of Lund University, for technical assistance and Manolis Kellis, of MIT, Boston, forsupport with the computational analysis.Funding. This work was supported by grants from the Swedish ResearchCouncil, Region Skåne (ALF), Knut and Alice Wallenberg Foundation, Novo NordiskFoundation, European Foundation for the Study of Diabetes/Lilly, SöderbergFoundation, Royal Physiographic Society in Lund, Swedish Diabetes Foundation,Påhlsson Foundation, EXODIAB, and Linné grant (B31 5631/2006). Part of thework described in this article was undertaken as part of the 2013-2014 BLUEScY educational exchange program, which was supported by the Faculty ofMedicine at Umeå University and EpiHealth (Lund University).Duality of Interest. No potential conflicts of interest relevant to this articlewere reported.Author Contributions. P.V. performed computational analyses. P.V. andC.L. designed the study. P.V., K.B., T.R., and C.L. wrote the manuscript. K.B., J.K.O.,J.L.S.E., and T.R. performed experiments. P.V., K.B., J.K.O., J.L.S.E., and T.R.performed statistical analyses. P.V., K.B., J.K.O., J.L.S.E., L.E., T.R., and C.L.designed experiments and edited the manuscript. P.V., T.R., and C.L. areguarantors of this work and, as such, had full access to all the data in the studyand take responsibility for the integrity of the data and the accuracy of the dataanalysis.

diabetes.diabetesjournals.org Volkov and Associates 1083

References1. Groop L, Pociot F. Genetics of diabetes–are we missing the genes or thedisease? Mol Cell Endocrinol 2014;382:726–7392. Dayeh T, Volkov P, Salö S, et al. Genome-wide DNA methylation analysis ofhuman pancreatic islets from type 2 diabetic and non-diabetic donors identifiescandidate genes that influence insulin secretion. PLoS Genet 2014;10:e10041603. Volkmar M, Dedeurwaerder S, Cunha DA, et al. DNA methylation profilingidentifies epigenetic dysregulation in pancreatic islets from type 2 diabetic pa-tients. EMBO J 2012;31:1405–14264. Yang BT, Dayeh TA, Volkov PA, et al. Increased DNA methylation and de-creased expression of PDX-1 in pancreatic islets from patients with type 2 di-abetes. Mol Endocrinol 2012;26:1203–12125. Bibikova M, Barnes B, Tsan C, et al. High density DNA methylation arraywith single CpG site resolution. Genomics 2011;98:288–2956. Agirre X, Castellano G, Pascual M, et al. Whole-epigenome analysis inmultiple myeloma reveals DNA hypermethylation of B cell-specific enhancers.Genome Res 2015;25:478–4877. Busche S, Shao X, Caron M, et al.; Multiple Tissue Human ExpressionResource. Population whole-genome bisulfite sequencing across two tissueshighlights the environment as the principal source of human methylome var-iation. Genome Biol 2015;16:2908. Court F, Tayama C, Romanelli V, et al. Genome-wide parent-of-origin DNAmethylation analysis reveals the intricacies of human imprinting and suggests agermline methylation-independent mechanism of establishment. Genome Res2014;24:554–5699. Kundaje A, Meuleman W, Ernst J, et al.; Roadmap Epigenomics Consortium.Integrative analysis of 111 reference human epigenomes. Nature 2015;518:317–33010. Fadista J, Vikman P, Laakso EO, et al. Global genomic and transcriptomicanalysis of human pancreatic islets reveals novel genes influencing glucosemetabolism. Proc Natl Acad Sci U S A 2014;111:13924–1392911. Pasquali L, Gaulton KJ, Rodríguez-Seguí SA, et al. Pancreatic islet enhancerclusters enriched in type 2 diabetes risk-associated variants. Nat Genet 2014;46:136–14312. Jones PA. Functions of DNA methylation: islands, start sites, gene bodiesand beyond. Nat Rev Genet 2012;13:484–49213. Hall E, Volkov P, Dayeh T, et al. Sex differences in the genome-wide DNAmethylation pattern and impact on gene expression, microRNA levels and insulinsecretion in human pancreatic islets. Genome Biol 2014;15:52214. Hansen KD, Langmead B, Irizarry RA. BSmooth: from whole genome bi-sulfite sequencing reads to differentially methylated regions. Genome Biol 2012;13:R8315. Collombat P, Hecksher-Sørensen J, Krull J, et al. Embryonic endocrinepancreas and mature beta cells acquire alpha and PP cell phenotypes upon Arxmisexpression. J Clin Invest 2007;117:961–97016. Gauthier BR, Wiederkehr A, Baquié M, et al. PDX1 deficiency causes mi-tochondrial dysfunction and defective insulin secretion through TFAM suppres-sion. Cell Metab 2009;10:110–11817. Kaneto H, Miyatsuka T, Kawamori D, et al. PDX-1 and MafA play a crucialrole in pancreatic beta-cell differentiation and maintenance of mature beta-cellfunction. Endocr J 2008;55:235–25218. Morris AP, Voight BF, Teslovich TM, et al.; Wellcome Trust Case ControlConsortium; Meta-Analyses of Glucose and Insulin-related traits Consortium(MAGIC) Investigators; Genetic Investigation of ANthropometric Traits (GIANT)Consortium; Asian Genetic Epidemiology Network–Type 2 Diabetes (AGEN-T2D)Consortium; South Asian Type 2 Diabetes (SAT2D) Consortium; DIAbetes Ge-netics Replication And Meta-analysis (DIAGRAM) Consortium. Large-scale as-sociation analysis provides insights into the genetic architecture andpathophysiology of type 2 diabetes. Nat Genet 2012;44:981–99019. Heinz S, Benner C, Spann N, et al. Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required formacrophage and B cell identities. Mol Cell 2010;38:576–589

20. Ziller MJ, Gu H, Müller F, et al. Charting a dynamic DNA methylationlandscape of the human genome. Nature 2013;500:477–48121. Ahlgren U, Jonsson J, Jonsson L, Simu K, Edlund H. Beta-cell-specificinactivation of the mouse Ipf1/Pdx1 gene results in loss of the beta-cell phe-notype and maturity onset diabetes. Genes Dev 1998;12:1763–176822. Bonala S, McFarlane C, Ang J, et al. Pid1 induces insulin resistance in bothhuman and mouse skeletal muscle during obesity [retracted in: Mol Endocrinol.2016;30:949]. Mol Endocrinol 2013;27:1518–153523. Ellingsgaard H, Hauselmann I, Schuler B, et al. Interleukin-6 enhances in-

sulin secretion by increasing glucagon-like peptide-1 secretion from L cells andalpha cells. Nat Med 2011;17:1481–148924. Iguchi H, Urashima Y, Inagaki Y, et al. SOX6 suppresses cyclin D1 promoteractivity by interacting with beta-catenin and histone deacetylase 1, and its down-regulation induces pancreatic beta-cell proliferation. J Biol Chem 2007;282:

19052–1906125. Jin HS, Kim J, Lee SJ, et al. The PARK2 gene is involved in the maintenanceof pancreatic b-cell functions related to insulin production and secretion. Mol CellEndocrinol 2014;382:178–18926. Kulkarni RN, Holzenberger M, Shih DQ, et al. Beta-cell-specific deletion of

the Igf1 receptor leads to hyperinsulinemia and glucose intolerance but does notalter beta-cell mass. Nat Genet 2002;31:111–11527. Lebrun P, Cognard E, Gontard P, et al. The suppressor of cytokine signalling2 (SOCS2) is a key repressor of insulin secretion. Diabetologia 2010;53:1935–194628. Reinbothe TM, Alkayyali S, Ahlqvist E, et al. The human L-type calciumchannel Cav1.3 regulates insulin release and polymorphisms in CACNA1D as-sociate with type 2 diabetes. Diabetologia 2013;56:340–34929. Sakamaki J, Fu A, Reeks C, et al. Role of the SIK2-p35-PJA2 complex in

pancreatic b-cell functional compensation. Nat Cell Biol 2014;16:234–24430. Taneera J, Lang S, Sharma A, et al. A systems genetics approach identifiesgenes and pathways for type 2 diabetes in human islets. Cell Metab 2012;16:122–13431. Tessem JS, Moss LG, Chao LC, et al. Nkx6.1 regulates islet b-cell pro-liferation via Nr4a1 and Nr4a3 nuclear receptors. Proc Natl Acad Sci U S A 2014;111:5242–524732. van de Bunt M, Gloyn AL. A tale of two glucose transporters: how GLUT2re-emerged as a contender for glucose transport into the human beta cell. Di-

abetologia 2012;55:2312–231533. Zhang B, Kirov S, Snoddy J. WebGestalt: an integrated system for exploringgene sets in various biological contexts. Nucleic Acids Res 2005;33(Web Serverissue):W741–74834. Stoffers DA, Thomas MK, Habener JF. Homeodomain protein IDX-1: a

master regulator of pancreas development and insulin gene expression. TrendsEndocrinol Metab 1997;8:145–15135. Stoffers DA, Ferrer J, Clarke WL, Habener JF. Early-onset type-II diabetesmellitus (MODY4) linked to IPF1. Nat Genet 1997;17:138–13936. Park JH, Stoffers DA, Nicholls RD, Simmons RA. Development of type 2diabetes following intrauterine growth retardation in rats is associated withprogressive epigenetic silencing of Pdx1. J Clin Invest 2008;118:2316–232437. Olsson AH, Volkov P, Bacos K, et al. Genome-wide associations betweengenetic and epigenetic variation influence mRNA expression and insulin secretion

in human pancreatic islets. PLoS Genet 2014;10:e100473538. Rosengren AH, Braun M, Mahdi T, et al. Reduced insulin exocytosis inhuman pancreatic b-cells with gene variants linked to type 2 diabetes. Diabetes2012;61:1726–173339. Dayeh TA, Olsson AH, Volkov P, Almgren P, Rönn T, Ling C. Identification ofCpG-SNPs associated with type 2 diabetes and differential DNA methylation inhuman pancreatic islets. Diabetologia 2013;56:1036–104640. Lyssenko V, Lupi R, Marchetti P, et al. Mechanisms by which commonvariants in the TCF7L2 gene increase risk of type 2 diabetes. J Clin Invest 2007;

117:2155–2163

1084 WGBS of Human Islets Diabetes Volume 66, April 2017

41. Hodson DJ, Mitchell RK, Marselli L, et al. ADCY5 couples glucose to insulinsecretion in human islets. Diabetes 2014;63:3009–302142. Bell RE, Golan T, Sheinboim D, et al. Enhancer methylation dynamics con-tribute to cancer plasticity and patient mortality. Genome Res 2016;26:601–61143. Domcke S, Bardet AF, Adrian Ginno P, Hartl D, Burger L, Schübeler D.Competition between DNA methylation and transcription factors determinesbinding of NRF1. Nature 2015;528:575–57944. De Vos A, Heimberg H, Quartier E, et al. Human and rat beta cells differ inglucose transporter but not in glucokinase gene expression. J Clin Invest 1995;96:2489–2495

45. Sansbury FH, Flanagan SE, Houghton JA, et al. SLC2A2 mutations cancause neonatal diabetes, suggesting GLUT2 may have a role in human insulinsecretion. Diabetologia 2012;55:2381–238546. Gao W, Fu Y, Yu C, et al. Elevation of NR4A3 expression and its possible role inmodulating insulin expression in the pancreatic beta cell. PLoS One 2014;9:e9146247. Siggens L, Ekwall K. Epigenetics, chromatin and genome organization:recent advances from the ENCODE project. J Intern Med 2014;276:201–21448. Ziller MJ, Hansen KD, Meissner A, Aryee MJ. Coverage recommendationsfor methylation analysis by whole-genome bisulfite sequencing. Nat Methods2015;12:230–232

diabetes.diabetesjournals.org Volkov and Associates 1085