The FASEB Journal article fj.14-264093. Published online December 9, 2014.

The FASEB Journal • Research Communication

Dynamic changes in the cardiac methylome during postnatal development Choon Boon Sim,*,1 Mark Ziemann,†,1 Antony Kaspi,† K. N. Harikrishnan,† Jenny Ooi,† Ishant Khurana,† Lisa Chang,† James E. Hudson,* Assam El-Osta,†,2 and Enzo R. Porrello*,2 *School of Biomedical Sciences, University of Queensland, Brisbane, Queensland, Australia; and † Epigenetics in Human Health and Disease Laboratory, Baker IDI Heart and Diabetes Institute, Melbourne, Victoria, Australia Relatively little is known about the epigenetic control mechanisms that guide postnatal organ maturation. The goal of this study was to determine whether DNA methylation plays an important role in guiding transcriptional changes during the first 2 wk of mouse heart development, which is an important period for cardiomyocyte maturation, loss of proliferative capacity and loss of regenerative potential. Gene expression profiling (RNA-seq) and genome-wide sequencing of methylated DNA (MBD-seq) identified dynamic changes in the cardiac methylome during postnatal development [2545 differentially methylated regions (DMRs) from P1 to P14 in the mouse]. The vast majority (~80%) of DMRs were hypermethylated between P1 and P14, and these hypermethylated regions were associated with transcriptional shut down of important developmental signaling pathways, including Hedgehog, bone morphogenetic protein, TGF-b, fibroblast growth factor, and Wnt/b-catenin signaling. Postnatal inhibition of DNA methylation with 5aza-29-deoxycytidine induced a marked increase (~3-fold) in cardiomyocyte proliferation and ~50% reduction in the percentage of binucleated cardiomyocytes compared with saline-treated controls. This study provides novel evidence for widespread alterations in DNA methylation during postnatal heart maturation and suggests that cardiomyocyte cell cycle arrest during the neonatal period is subject to regulation by DNA methylation.—Sim, C. B., Ziemann, M., Kaspi, A., Harikrishnan, K. N., Ooi, J., Khurana, I., Chang, L., Hudson, J. E., El-Osta, A., and Porrello, E. R. Dynamic changes in the cardiac methylome during postnatal development. FASEB J. 29, 000–000 (2015). www.fasebj.org ABSTRACT

Key Words: DNA methylation • cardiomyocyte proliferation binucleation • neonatal heart • epigenetics

•

Abbreviations: 5aza-dC, 5-aza-29-deoxycytidine; CM, cardiomyocyte; CpG, cytosine-guanine dinucleotide; DMR, differentially methylated region; Dnmts, DNA methyltransferases; ENCODE, The Encyclopedia of DNA Elements; FDR, false discovery rate; FGF, fibroblast growth factor; GO, Gene Ontology; GSEA, Gene Set Enrichment Analysis; MACS, modelbased analysis for ChIP-Seq; MBD-seq, methyl binding domain enrichment and sequencing; P, postnatal d; pH3, phosphorylatedhistone H3; qPCR, quantitative PCR; RNA-seq, next-generation mRNA sequencing; TET, Ten-eleven Translocase; TSS, transcription start site

0892-6638/15/0029-0001 © FASEB

THE MAMMALIAN HEART undergoes several important physiologic transitions during neonatal life, which allow the heart to adapt to the postnatal environment. In rodents, the first 2 wk of postnatal heart development coincide with rapid cardiac growth, cardiomyocyte cell cycle withdrawal, a marked increase in the proportion of binucleated cardiomyocytes, and cardiomyocyte hypertrophy. Additionally, there is a metabolic transition from glycolysis to fatty acid oxidation, increased extracellular matrix production, and changes in the expression and organization of sarcomeric proteins (1). It is noteworthy that it is also during this period that the robust regenerative potential of the neonatal rodent heart diminishes (2, 3). These neonatal developmental transitions are associated with alterations in the expression of thousands of genes embedded within tightly controlled transcriptional networks. However, the transcriptional regulatory mechanisms that guide neonatal heart development remain poorly understood. Epigenetic modifications have emerged as critical regulators of gene expression changes during heart development and disease. The control of gene expression is subject to diverse regulatory actions that involve DNA bound transcription factors including ATP-dependent chromatin remodeling, covalent histone modifications such as acetylation and methylation, and DNA methylation. Genetic deletion of chromatin modifying proteins (including histone methyltransferases, acetylases, deacetylases, and ATP-dependent remodeling factors) are implicated in cardiac developmental processes, including defects in cardiomyocyte proliferation (4). Recent studies have further interrogated the relationship between chromatin modifications (histone acetylation and methylation) and gene transcription on a genome-wide scale, which has identified important relationships between chromatin transitions and transcription during cardiac lineage commitment (5). It has also become apparent that the cardiac 1

These authors contributed equally to this study. Correspondence: A.E.-O., Epigenetics in Human Health and Disease Laboratory, Baker IDI Heart and Diabetes Institute, Melbourne, VIC 3004, Australia. E-mail: assam.el-osta@ bakeridi.edu.au; or E.R.P., School of Biomedical Sciences, University of Queensland, Brisbane, QLD 4072, Australia. E-mail: [email protected] doi: 10.1096/fj.14-264093 This article includes supplemental data. Please visit http:// www.fasebj.org to obtain this information. 2

1

cell cycle is under epigenetic control, as comparison of embryonic and adult rodent cardiomyocytes revealed an enrichment of histone H3 trimethylated lysine 9 (associated with transcriptional repression) at the promoters of cell cycle genes during cardiac maturation (6). A recent study has also identified numerous epigenetic modifications at Notch responsive promoters that appear to be important determinants of myocyte proliferative responses to Notch ligands (7). Curiously, despite intensive interrogation of the roles of chromatin modifiers during cardiac development and disease, there have been few systematic studies on the impact of DNA methylation. DNA methylation occurs by covalent modification at carbon 5 of cytosine, predominantly in the context of a cytosineguanine dinucleotide (CpG) (8). The mechanisms for establishing and maintaining DNA methylation are well established and involve 3 methyltransferases (Dnmts) in mammals: Dnmt1, 3a, and 3b (9). Germ-line deletion studies in mice have established essential roles for Dnmts during embryonic heart development (10). More recent genomewide DNA methylation profiling studies have identified widespread alterations in DNA methylation patterns during early embryonic heart development and have revealed potentially important associations with transcriptional changes in cardiogenic genes (11). Moreover, dynamic changes in DNA methylation patterns correlate with gene expression changes in heart failure (12) and dilated cardiomyopathy (13) in humans, and inhibition of DNA methylation may be cardioprotective in settings of norepinephrine-induced cardiac hypertrophy (14) and ischemic heart disease (15). These findings suggest that DNA methylation is important for cardiac gene expression during embryonic heart development and in various adult disease settings. However, the impact of DNA methylation during cardiac maturation in the neonatal period has not been investigated. In this study, we hypothesized that DNA methylation plays a key role in the repression of transcriptional networks governing cardiomyocyte proliferation during the neonatal period. The current study presents the first systematic analysis of DNA methylation during postnatal heart development. We identify two waves of DNA methylation occurring during the neonatal period and demonstrate that inhibition of DNA methylation during neonatal heart development prolongs the proliferative capacity of cardiomyocytes and is associated with a profound reduction in cardiomyocyte binucleation. Moreover, we have undertaken a genome-wide approach to analyze neonatal changes in DNA methylation and transcription, which shows differential methylation events across thousands of loci, some of which are associated with transcriptional changes in important developmental signaling pathways for muscle growth and differentiation. These results provide new insight and information on the regulation of cardiac gene expression subject to changes by DNA methylation in the neonatal period. MATERIALS AND METHODS Experimental animals and tissue collection All protocols were approved by The University of Queensland Animal Ethics Committee. Male C57BL/6 mice (The Jackson

2

Vol. 29

April 2015

Laboratory, Bar Harbor, ME, USA) were used for genome-wide DNA methylation and mRNA expression profiling during postnatal development (P1 and P14). ICR/CD-1 mice (Charles River Laboratories, Wilmington, MA, USA) were used for neonatal 5-aza-29-deoxycytidine (5aza-dC) studies (described below). For DNA and RNA analysis, atrial tissues were removed from ventricles (septum intact), and ventricles were blotted and weighed before snap-freezing in liquid nitrogen for storage at 280°C prior to processing.

DNA extraction and global analysis of DNA methylation DNA was isolated from mouse cardiac ventricles using the DNeasy Blood & Tissue Kit (Qiagen, Venlo, Limburg, The Netherlands) according to the manufacturer’s instructions. For analysis of global DNA methylation, 5-methylcytosine levels were measured in DNA samples (100 ng/sample) using the MethylFlash Methylated DNA Quantification Kit (Colorimetric; Epigentek, Farmingdale, NY, USA), in which DNA is bound to strip wells that are specifically treated to have a high DNA affinity, and the methylated fraction of DNA is detected using capture and detection antibodies. The fraction of methylated to unmethylated DNA is then quantified through an ELISA-like reaction by reading the absorbance in a microplate spectrophotometer at 450 nm. RNA extraction, cDNA synthesis, and real-time quantitative PCR Total RNA was extracted from mouse cardiac ventricles using TRIzol (Ambion, Carlsbad, CA, USA) using the manufacturer’s recommended protocol. Genomic DNA contamination was removed using the TURBO DNA-free kit (Ambion). cDNA templates were synthesized using the SuperScript III First-Strand Synthesis System (Invitrogen, Carlsbad, CA, USA) and random hexamers as primers. Quantitative PCR (qPCR) was performed on the StepOnePlus Real-Time PCR System (Applied Biosystems, Carlsbad, CA, USA) using SYBR Green PCR Master Mix (Invitrogen), as previously described (16). PCR primers for qPCR are listed in Table 1. 18S ribosomal RNA was used as a housekeeping control for all PCR reactions. Histologic analysis Hearts were briefly rinsed in PBS (Gibco, Carlsbad, CA, USA), fixed in 4% paraformaldehyde overnight, and then stored in PBS at 4°C until paraffin embedding. Sections (5 mm thickness) were processed for hematoxylin and eosin staining according to standard procedures, as described previously (17). Immunofluorescence Sections were de-paraffinized and washed with PBS prior to immunostaining as described previously (2). For Dnmt1, Dnmt3a, and phosphorylated-histone H3 (pH3) immunostaining, sections underwent antigen retrieval by boiling in Tris-EDTA buffer (pH 9.0) at 110°C for 20 min. Sections were then recovered in PBS for 20 min at room temperature, followed by permeabilization in 0.3% Triton X-100 for 5 min, washing in PBS (3 3 3 min), and then blocking in 10% goat serum (in PBS) at room temperature for 20 min. Primary antibodies were diluted in 2% goat serum and incubated overnight at 4°C. The following primary antibodies were used for immunofluorescence staining: Dnmt1 (D63A6, 1:100, rabbit monoclonal #5032S; Cell Signaling, Danvers, MA, USA), Dnmt3a (D23G1, 1:100, rabbit monoclonal

The FASEB Journal x www.fasebj.org

SIM ET AL.

TABLE 1. List of primer sequences for qPCR analysis and bisulfite sequencing Function

SYBR Green qPCR

Bisulfite sequencing

Targeted gene

Primer sequence

18s R9 18s L9 Dnmt1 R9 Dnmt1 L9 Dnmt2 F9 Dnmt2 R9 Dnmt3a R9 Dnmt3a L9 Dnmt3b F9 Dnmt3b R9 Dnmt3l R9 Dnmt3l L9 Tet1 F9 Tet1 R9 Tet2 F9 Tet2 R9 Tet3 R9 Tet3 L9 Gadd45a F9 Gadd45a R9 Gadd45b R9 Gadd45b L9 Gadd45g R9 Gadd45g L9 Igf2bp3 R9 Igf2bp3 L9 Neat1 F9 Neat1 R9 Igf2bp3 F9 Igf2bp3 R9 Neat1 F9 Neat1 R9 Topo M13 Reverse

CCCTCCAATGGATCCTCGTT TCGAGGCCCTGTAATTGGAA GCCATCTCTTTCCAAGTCTTT TGTTCTGTCGTCTGCAACCT AGAGGATGGAACCTCTGCGT CCACATGTGCAGGGATATGA GCTTTCTTCTCAGCCTCCCT CCATGCCAAGACTCACCTTC ATCCATAGTGCCTTGGGACC CTGGCACCCTCTTCTTCATT GCTTGCTCCTGCTTCTGACT GGTGTGGAGCAACATTCCAG GGGAGCTCATGGAGACTAGG AGAGCTCTTCCCTTCCTTCC CTTCTCTGCTCATTCCCACA AGCTCCGACTTCTCGATTGT CAGCGATTGTCTTCCTTGGT GCCTGCATGGACTTCTGTG AAGACCGAAAGGATGGACAC ACCACGTTATCGGGGTCTAC TCTGCAGAGCGATATCATCC CGGCCAAACTGATGAATGT AAGTTCGTGCAGTGCTTTCC CGCACAATGACTCTGGAAGA GGGCGGGATATTTCGTATCT GCGCTTTCAGGTAAAATGGA TAGGTTCCGTGCTTCCTCTT ACATCCTCCACAGGCTTACC TAGGTAAAATGGAATTATATGGGAAAT ACAACAATAACCCAAAAACAAAAAC GGTTGTGAATGTTTTAGATGAATGTT CAAACTACAAAAACAAAAAAAATAAAC CAGGAAACAGCTATGAC

#5398S; Cell Signaling), cardiac troponin-T (1:100, mouse monoclonal MS-295-P1; Thermo Fisher Scientific, Waltham, MA, USA), and pH 3 (Ser10, 1:100, rabbit polyclonal 06-570; Merck Millipore, Darmstadt, Germany). Following primary antibody incubation, sections were washed in PBS (3 3 5 min) and then incubated with secondary antibodies conjugated with Alexa Fluor 488 or 555 (1:400; Molecular Probes, Carlsbad, CA, USA) and Hoechst dye (1:1000, H21492; Invitrogen) at room temperature for 1 h in the dark. Sections were washed in PBS (3 3 3 min) before being mounted with Fluoromount-G (Southern Biotech, Birmingham, AL, USA). Wheat germ agglutinin (50 mg/ml, W11261; Molecular Probes) staining was performed as described previously (2). Quantitative analyses of cardiomyocyte mitosis and cross-sectional area were performed using ImageJ as described previously (2).

mounted on slides. Bright field and fluorescence images were acquired for 100 cells per sample and overlayed for postanalysis in ImageJ. Quantification of cardiomyocyte nucleation, cell width, and cell length were performed under blinded conditions. In vivo administration of 5aza-dC to neonatal mice The hypomethylating agent 5aza-dC (Sigma-Aldrich, St. Louis, MO, USA) was dissolved in saline and administered to neonatal ICR/CD-1 mice through daily subcutaneous injections (1 mg/kg) from postnatal d 2 (P2) to P12. Control mice received a daily subcutaneous injection of an equivalent volume of saline. All animals were euthanized at P7 or P12 by asphyxiation with CO2 followed by cervical dislocation. Hearts were harvested for DNA/ RNA extraction, histology, and cell isolations as described above.

Cardiomyocyte isolation and quantification of binucleation Hearts were harvested and fixed in 4% paraformaldehyde for 1 h at room temperature. Cardiomyocytes were isolated by enzymatic digestion of fixed cardiac tissues using a combination of collagenase D (2.4 mg/ml; Roche, Basel, Switzerland) and collagenase B (1.8 mg/ml; Roche) with overnight shaking at 37°C until the tissue was fully digested. Cell suspensions were then pelleted at 300 g for 3 min at room temperature and then resuspended in PBS. For Hoechst dye staining, cell suspensions were stained with Hoechst dye (1:500; Invitrogen) for 10 min at room temperature. Cells were then pelleted at 300 g for 3 min and washed in PBS before being resuspended in Fluoromount-G and

DNA METHYLATION REGULATES CARDIAC DEVELOPMENT

Transcriptional profiling with RNA-seq Cardiac ventricles were harvested at P1 and P14 (n = 9 per group) from C57BL6/J mice as described above and were homogenized in PBS. Three individual hearts were pooled per replicate. This homogenate was divided between transcriptome and methylome assays. RNA was isolated by TRIzol extraction (Ambion), followed by QiaQuick MinElute cleanup including on-column DNase digestion. RNA quality was verified with MultiNA bioanalyzer (Shimadzu, Kyoto, Japan). Purified mRNA was prepared using Dynabeads Oligo(dT) 25 (Ambion) enrichment. Libraries were

3

generated using the NEBNext mRNA Library Prep Reagent Set for Illumina (New England Biolabs, Ipswich, MA, USA). Libraries were validated on MultiNA bioanalyzer and sequenced at a concentration of 10 pM (Illumina, San Diego, CA, USA). Base-calling was performed off-line with OLB version 1.8.

Bisulfite sequencing

Genomic DNA for CpG methylation profiling was isolated using the DNeasy Blood & Tissue Kit (Qiagen). DNA was fragmented by BioRuptor sonication (Diagenode, Denville, NJ, USA). Fragmented DNA was analyzed with bioanalyzer, and 1 mg sonicated DNA underwent methyl-CpG enrichment using the MethylMiner Methylated DNA Enrichment Kit (Applied Biosystems). Sequencing libraries were generated using the NEBNext DNA Library Prep Reagent Set for Illumina (New England Biolabs). Libraries were validated on MultiNA bioanalyzer and sequenced as above. Methyl binding domain enrichment and sequencing (MBD-seq) measures average methyl-cytosine levels in the sequenced fragments (150–250 bp in length) (18).

Bisulfite sequencing was performed on genomic DNA from cardiac ventricles obtained at P1, P7, and P28 (n = 3 per time point). DNA samples underwent bisulfite conversion using the EZ DNA Methylation kit (Zymo Research, Irvine, CA, USA). Bisulfite PCR primers are listed in Table 1. Bisulfite-converted DNA samples were then used as a template for PCR reactions using Platinum Taq DNA polymerase (Invitrogen) with the following cycling conditions: 95°C for 3 min, 40 cycles of 95°C for 30 s, 58°C for 30 s, and 72°C for 45 s, followed by 72°C for 5 min. The PCR product was purified using the QIAquick Gel Extraction Kit (Qiagen) and cloned into the TOPO plasmid (Invitrogen) for downstream sequencing. Following transformation and blue-white Xgal selection in competent Escherichia coli (New England Biolabs), 6 positive clones per heart sample were picked and processed for Sanger sequencing (Australian Genome Research Facility, Brisbane, QLD, Australia). Bisulfite sequencing primers are listed in Table 1. Bisulfite sequencing results were then analyzed using BISMA software, as previously described (26).

Bioinformatic analyses

Data access

Sequence reads underwent 39 base quality trimming using the Fastx toolkit with a minimum Phred quality score (Q) of 30 and a minimum read length of 20 nt. RNA-seq reads were aligned to the mouse genome sequence (GRCm38.70/mm10) using Olego with default settings (19). Reads aligning to Ensembl-annotated exons with a strict mapping quality (Q $ 20) were counted with BedTools (20) and used to construct a data matrix comprising genes with an average of 10 reads or more per sample across the experiment. Differential gene expression analysis was performed using edgeR (v3.2.4) (21). Heat maps were generated from normalized and scaled gene expression values in R (22). MBD-seq reads were aligned to the mouse genome (GRCm38.70/mm10) using Burrows Wheeler Aligner with default settings (23). Reads from each developmental stage were pooled, and a single large alignment file (bam format) was prepared for P1 and P14. Model-based analysis for ChIP-Seq (MACS) (24) was then used to directly compare P1 with P14 and identify regions of interest. MACS peak-calling was performed with a shift size of 75 bp and otherwise default parameters. Differentially methylated regions (DMRs) were defined as coordinates initially identified using MACS peak calling determined to have a statistically significant signal difference [P1 vs. P14, false discovery rate (FDR) # 0.05], and all DMRs exhibited .30% fold change. Intersection between our DMRs and Ensembl genomic features was performed with BedTools. Likewise, intersection between DMRs and transcription factor binding sites or DNaseI hypersensitivity sites defined by the Encyclopedia of DNA Elements (ENCODE) project (http://genome.ucsc.edu/ENCODE/downloadsMouse.html) was determined. Peak files from Mouse ENCODE project (mm9) were converted to mm10 coordinates using liftOver tool (http://hgdownload.soe.ucsc.edu/admin/exe/) with the mm9 to mm10 chain file (http://hgdownload-test.cse.ucsc.edu/goldenPath/ mm9/liftOver/mm9ToMm10.over.chain.gz) using the default settings. Enrichment of transcription factor binding motifs between P14 and P1 peaks was computed using Homer (http:// homer.salk.edu/homer/ngs/index.html). Gene expression, methylation, and gene set enrichments with an FDR P # 0.05 were considered significant. Gene Ontology (GO) analysis (pathway/ process/network) was performed using MetaCore from Thomson Reuters (https://portal.genego.com/cgi/data_manager.cgi) on significant gene sets. All network relationship analyses were performed using the direct interaction analysis function. Gene Set Enrichment Analysis (GSEA) was performed on these data as described previously (25) using publicly available gene sets (www. broadinstitute.org/gsea/msigdb).

RNA-seq and MBD-seq data from this project are available from Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/) under accession number GSE59971.

Methyl binding domain enrichment and sequencing

4

Vol. 29

April 2015

Statistical analysis All nonbioinformatic statistical analyses were performed using GraphPad Prism 6.0. For comparisons between 2 groups, a 2-tailed unpaired Student t test was used. For multiple group comparisons, a 1-way ANOVA followed by Bonferroni posttest was used. For comparison between multiple groups with multiple variables, a 2way ANOVA followed by Bonferroni posttest was used. All data are presented as mean 6 SEM, except for the scatter plot of quantified cardiomyocyte cross-sectional surface area (Fig. 2I), which is shown as a median. Detailed statistical analysis methods and sample sizes are indicated in the figure legends. For all statistical analyses, *P , 0.05 and **P , 0.001. For bioinformatics analysis, including RNAseq, MBD-seq, and gene set enrichment analyses, an FDR P # 0.05 using the correction procedure of Benjamini and Hochberg was used (27). GO analysis (pathway/process/network) was performed using only statistically significant gene sets.

RESULTS Dynamic regulation of DNA methylation machinery during postnatal cardiac maturation Total DNA methylation was measured in mouse hearts at P1, P7, P14, P28, and P84. Genomic DNA methylation levels remained steady from P1 to P14 but were markedly decreased thereafter (Fig. 1A). We next measured the expression levels of Dnmts by real-time qPCR in mouse hearts from P1 to P84 (Fig. 1B). Consistent with an overall decrease in genomic DNA methylation levels after P14, qPCR profiling of Dnmt1, Dnmt3a, and Dnmt3b confirmed decreased mRNA expression levels in postnatal maturation of cardiac tissue. This result was verified at the protein level by immunohistochemistry staining (Fig. 1C). Dnmt1 and Dnmt3a were predominantly localized to the nucleus of cardiomyocytes (CMs) with high expression levels at P1 and P7 (Fig. 1C). Although Dnmt1 exhibited a dramatic decrease in expression levels after P7, such that it became

The FASEB Journal x www.fasebj.org

SIM ET AL.

Figure 1. DNA methylation profiling during postnatal cardiac development. A) Genomic DNA methylation levels in mouse cardiac ventricles from P1 to P84 (n = 3 per time point; *P , 0.05). B) qPCR gene expression profiling of Dnmt1, 3a, and 3b in mouse cardiac ventricles from P1 to P84. Expression levels of each Dnmt are normalized to 18S and represented as a fold change relative to P1 ventricles (n = 3 per time point; *P , 0.05, **P , 0.001). C) Immunohistochemistry staining for Dnmt1 and Dnmt3a in mouse hearts from P1 to P28. Red, cardiac troponin T; green, Dnmt1 or Dnmt3a; blue, nuclei. Dnmt1/3a positive cardiomyocytes are also shown in the high magnification inset in the top right corner of each panel. Scale bar = 50 mm in main figure and 5 mm in high-magnification inset.

hardly detectable in CMs by P14 and P28, Dnmt3a maintained its expression level until P14 but was undetectable in P28 hearts (Fig. 1C). Dnmt1 and Dnmt3a were also expressed in non-myocytes at early developmental time points but were undetectable in all cell types by P28 (Fig. 1C). We also checked the expression levels of the noncanonical Dnmts, Dnmt2 and Dnmt3l, by qPCR analysis. Although Dnmt2 expression levels did not dramatically change throughout life, Dnmt3l was significantly upregulated after P14 (Supplemental Fig. S1A). Given the decrease in global DNA methylation levels after P14, we profiled the expression of Ten-eleven Translocase (TET) and Growth Arrest and DNA Damage 45 (Gadd45) proteins because they are implicated in active DNA demethylation (9). Real-time qPCR analysis showed that Gadd45a and Gadd45g were up-regulated at P14 and Gadd45a expression continued to increase until P28 (Supplemental Fig. S1B, C). However, the mRNA expression levels of Tet1-3 and Gadd45b all decreased during postnatal heart maturation (Supplemental Fig. S1B, C). DNA methylation is required for cardiomyocyte cell cycle arrest and binucleation in vivo To better understand the role of DNA methylation in regulating postnatal cardiac development, we pharmacologically inhibited DNA methylation using 5aza-dC to neonatal mice DNA METHYLATION REGULATES CARDIAC DEVELOPMENT

from P2 (Fig. 2A). 5aza-dC–treated mice grew normally for the first 5 d of exposure but stopped gaining weight thereafter, and the experiment was terminated at P12 (Fig. 2A). Systemic administration of 5aza-dC induced an ;50% reduction of genomic DNA methylation levels in P12 hearts (Fig. 2B). Inhibition of DNA methylation was associated with a significant increase in heart size, as indicated by a pronounced increase in the heart weight-to-body weight ratio at P7 and P12 (Fig. 2C, D). To understand whether the increased heart weight following inhibition of neonatal DNA methylation was caused by cell proliferation, heart sections were stained with antibodies recognizing pH3 and cardiac troponin T to quantify the incidence of CM mitosis. At P7, an increase in the number of pH3+ cells by ;50% in CMs and ;25% in nonmyocytes was observed in 5aza-dC–treated hearts (Fig. 2E, F) and was associated with an ;50% reduction in the percentage of binucleated CMs (Fig. 2G). A reduced percentage of binucleated CMs in the 5aza-dC–treated group was also noted at P12 (Fig. 2G). Although we did not detect any effect of 5aza-dC on cell size at P7, a modest increase in cell size was observed at P12 (Fig. 2H, I). Interestingly, CM hypertrophy at P12 was associated with an increase in cell length in both mono- and binucleated CM populations (Fig. 2J, K, L). Together, these findings suggest that inhibition of DNA methylation during neonatal life prevents CM cell cycle arrest and binucleation. 5

Figure 2. Pharmacological inhibition of DNA methylation during neonatal life promotes cardiac cell proliferation and blocks cardiomyocyte binucleation. A) Growth curve for CD-1/ICR mice treated with 5aza-dC from P2 to P12 (n = 14–15 per group). *P , 0.05, **P , 0.001. B) Hematoxylin and eosin staining of saline and 5aza-dC–treated hearts at P12. C) Genomic DNA methylation levels of saline and 5aza-dC–treated mice at P12 (n = 4 per group; *P , 0.05). D) Heart weight to body weight ratios for saline and 5aza-dC–treated mice at P7 (n = 11–17 per group) and P12 (n = 7–14 per group). **P , 0.001. E) Immunohistochemistry staining of saline and 5aza-dC-treated hearts at P7. Green, pH3; red, cardiac troponin T; blue, nuclei. pH3+ cardiomyocytes are also shown in the high-magnification inset in the top right corner of each panel. Scale bar = 50 mm in the main figure and 10 mm in the high magnification inset. F) Quantification of pH3+ CMs and non-CMs in saline and 5aza-dC– treated hearts at P7. Data are presented as cells per field for n = 4–6 independent samples per group (10 fields of cells assessed per heart). **P , 0.001. G) Quantification of the percentage of binucleated cardiomyocytes at P7 (n = 4–6 per group; N = 100 cells per sample) and P12 (n = 3–5 per group; N = 100 cells per sample). **P , 0.001. H) Wheat germ agglutinin staining of saline and 5aza-dC-treated hearts at P12. Scale bar = 50 mm. I) Scatter plot of quantified cardiomyocyte cross-sectional surface area in saline and 5aza-dC-treated hearts at P7 (n = 4–5 per group) and P12 (n = 3–5 per group) (10 fields of cells assessed per heart). Blue line represents median. **P , 0.001. J) Hoechst staining of mononucleated and binucleated cardiomyocytes. Scale bar = 10 mm. K, L) Quantification of cardiomyocyte length (K) and width (L) in saline and 5aza-dC-treated hearts at P12 (n = 3–5 per group; N = 100 cells per heart). Data are presented as average cell length or width per group. All values are presented as mean 6 SEM. *P , 0.05, **P , 0.001.

6

Vol. 29

April 2015

The FASEB Journal x www.fasebj.org

SIM ET AL.

Figure 3. RNA-seq analysis of gene expression changes during neonatal heart development. A) Smear plot of RNA-seq data showing average signal intensity (x axis) vs. log2 fold change in gene expression (P14/P1). Differentially expressed genes (FDR # 0.05, n = 6476) are shown in red, and nonsignificant changes are shown in black. B) Heat map of differentially expressed genes from P1 to P14. Genes with higher expression levels are shown in yellow, whereas genes with lower expression levels are shown in red. C) GO pathway analysis (MetaCore) for differentially regulated genes in P1 vs. P14 hearts. Genes that have higher expression levels at P1 (blue) or higher expression levels at P14 (red).

Neonatal cardiac maturation is associated with profound alterations in gene transcription To understand the transcriptional changes during neonatal heart maturation, mRNA sequencing (RNA-seq) was performed in P1 and P14 hearts. A total of 6476 genes were differentially expressed (FDR # 0.05) between P1 and P14 cardiac ventricles (Fig. 3A and Table 2). Among these DNA METHYLATION REGULATES CARDIAC DEVELOPMENT

differentially expressed genes, 3233 genes were expressed at significantly higher levels at P14, whereas 3243 genes were expressed at higher levels at P1 (Fig. 3A, B, and Table 2). GO pathway analysis revealed an increase in the expression of genes at P14 that were predominantly involved in metabolism (oxidative phosphorylation) and cytoskeletal remodeling, which is consistent with the well-known changes in metabolic substrate utilization and cytoskeletal 7

TABLE 2. Summary of the total number of differentially expressed genes and differentially methylated regions between P1 and P14 from RNA-seq and MBD-seq

Profiling method

Number of regions higher at P1

Number of regions higher at P14

Total

3243

3233

6476

540

2005

2545

RNA-seq (FDR # 0.05) MBD-seq (FDR # 0.05)

rearrangement (1) that occur after birth in mammals (Fig. 3C). On the other hand, the majority of genes that decreased in expression levels at P14 were associated with cell cycle checkpoints, DNA replication, the DNA damage response, and cytoskeletal remodeling, consistent with the well-established phenomenon of CM cell cycle arrest and ultrastructural maturation following birth (Fig. 3C) (1). Differential methylated regions characterize neonatal cardiac maturation Given the importance of DNA methylation for postnatal CM cell cycle arrest and binucleation, we next assessed whether there are alterations in DNA methylation across the genome during neonatal cardiac maturation. DNA samples from P1 and P14 cardiac ventricles were processed for whole-genome DNA methylation sequencing using a methyl binding protein enrichment sequencing approach (MBD-seq), known as MethylMiner, which we have previously validated (28). Analysis of the total number of reads processed indicated that the numbers of aligned reads used in peak calling were similar for P1 and P14 sample groups (Table 3). The MBD-seq analysis identified 2545 DMRs between P1 and P14 (Fig. 4A and Table 2). Among these DMRs, 2005 regions (;80%) were hypermethylated and 540 regions (;20%) were hypomethylated at P14 relative to P1 (Fig. 4A, B, and Table 2). Notably, P14 DMRs had a higher CpG dinucleotide content (28.95 6 0.006/kpb at P14 vs. 13.02 6 0.012/kpb at P1) and higher G/C nucleotide content (553.0 6 0.021/kpb at P14 vs. 396.8 6 0.116/ kpb at P1) than the P1 DMRs. Interestingly, DNA hypermethylation was noted across all mouse chromosomes at P14 (Fig. 4C). Intersection of differentially methylated peaks with the mouse genome revealed that ;4.8% of the genome was differentially methylated between P1 and P14 (Fig. 4D). The majority of DNA methylation peaks at P14 occurred across protein coding DNA sequences and exons, with almost one-third of coding regions displaying an increase in methylation levels from P1 to P14 (Fig. 4D). Gene expression changes implicated in cardiac maturation are subject to DNA methylation Because neonatal cardiac development is subject to broad changes in DNA methylation, we examined associations with gene expression by intersecting RNA-seq and MBDseq data sets. In total, 564 DMRs overlapping promoters [63 kb transcription start site (TSS)], and gene bodies were associated with significant changes in mRNA expression 8

Vol. 29

April 2015

levels between P1 and P14 (Fig. 5A, upper; and Table 4). The majority of DMRs associated with gene expression changes at P14 overlapped protein coding sequences and exons (Table 4). In total, 143 hypermethylated regions were associated with transcriptional repression at P14, whereas a much larger number of hypermethylated regions (n = 385) were associated with transcriptional activation during neonatal cardiac maturation (Fig. 5A, lower, and Table 4). A smaller number of hypomethylated regions were similarly associated with either transcriptional repression or activation at P14 (Fig. 5A, lower, and Table 4). To explore relationships between DNA methylation and transcription in greater detail, we performed statistical correlation analyses between P14 methylation peaks across different genomic features and mRNA expression changes. At P14, DNA methylation across exons was highly positively correlated with gene expression (Supplemental Fig. S2). Similarly, DNA methylation across promoter regions at P14 was also typically associated with increased transcription at P14 (Supplemental Fig. S2). In contrast, DNA methylation across exons and promoter regions at P1 negatively correlated with gene expression levels (Supplemental Fig. S2). These observations are consistent with genome-wide methylation profiling studies in other cell types, which suggest the impact of DNA methylation on gene transcription is highly context dependent (29). To validate these novel associations between methylation and transcription during neonatal heart development, 2 of the most highly differentially methylated genes between P1 and P14 were analyzed by qPCR and bisulfite sequencing. Insulin-like growth factor 2 binding protein 3 (Igf2bp3) represented the top candidate for genes displaying increased methylation associated with decreased mRNA expression at P14, whereas the long noncoding RNA, nuclear paraspeckle assembly transcript 1 (Neat1), represented the top candidate for genes displaying decreased methylation and increased expression at P14 (Fig. 5A, lower). Real-time qPCR analysis confirmed that Igf2bp3 expression levels decreased from P1 to P84, whereas the expression level of Neat1 significantly increased after birth (Fig. 5B), consistent with the RNA-seq data. Bisulfite sequencing was also performed on these 2 gene loci (Fig. 5D). Bisulfite sequencing results strongly validated our MBD-seq data and confirmed increased methylation across the Igf2bp3 locus and decreased methylation across the Neat1 locus from P1 to P28 (Fig. 5D). To further confirm that the expression level of Igf2bp3 and Neat1 was regulated by DNA methylation, their mRNA expression level was also examined in heart samples from mice that were treated with 5aza-dC. Both Igf2bp3 and Neat1 TABLE 3. MBD-seq quality control summary Data set

P1-1 P1-2 P1-3 P14-1 P14-2 P14-3 P1 Total P14 Total

The FASEB Journal x www.fasebj.org

Total reads

QC passed reads

Aligned reads

33,903,541 34,214,266 33,554,267 30,736,962 34,335,375 32,694,036 101,672,074 97,766,373

33,675,705 33,751,319 33,361,709 30,597,908 34,131,579 32,526,986 100,788,733 97,256,473

32,911,534 33,036,386 32,690,433 30,167,706 33,532,376 31,806,532 98,638,353 95,506,614

SIM ET AL.

Figure 4. MBD-seq analysis of DNA methylation changes occurring during neonatal heart development. A) Smear plot of MBDseq data showing average signal intensity (x axis) vs. log2 fold change in MBD enrichment (P14/P1). Differentially methylated regions (FDR # 0.05, n = 2545) are shown in red, and nonsignificant changes are shown in black. B) Heat map of differentially methylated regions identified from MBD-seq. Regions with higher methylation levels are shown in yellow, whereas regions with lower methylation levels are shown in red. C) Distribution of differentially methylated regions across each chromosome at P1 (black) and P14 (red) for mouse (mm10) genome normalized by chromosome length. D) Distribution of differentially methylated regions shown by genomic features. Hypermethylated regions at P1 are shown in yellow, hypermethylated regions at P14 are shown in red, and regions without significant differential methylation are shown in blue.

exhibited higher expression levels in 5aza-dC–treated groups (Fig. 5C), which was consistent with their methylation/ expression patterns during neonatal cardiac development. DNA methylation regulates transcription of critical developmental signaling networks during neonatal life We next used GO analysis to explore the potential mechanisms underlying relationships between DNA methylation and gene transcription during neonatal heart development. GO was used for process network analysis of various gene sets representing different relationships between DNA methylation and gene transcription (Fig. 6A and Supplemental Figs. S3A and 4A). Process network analysis DNA METHYLATION REGULATES CARDIAC DEVELOPMENT

for genes with increased DNA methylation and reduced expression at P14 identified a highly significant enrichment for genes associated with several important developmental signaling pathways including Hedgehog, TGF-b, fibroblast growth factor (FGF), vascular endothelial growth factor, Notch, and the Wnt/b-catenin pathway (Fig. 6A). In addition, genes with increased methylation and decreased expression during neonatal cardiac maturation were also highly enriched for genes associated with cell adhesion and extracellular matrix interactions (Fig. 6A). Further analysis of this gene set revealed an interaction network centered on Smad3, a critical downstream transcription factor for the TGF-b signaling pathway (Fig. 6B). For genes displaying increased methylation and increased transcription at P14, Gene GO 9

Figure 5. Integration of RNA-seq and MBD-seq identifies complex relationships between DNA methylation and transcription during neonatal heart development. A) (Upper) Venn diagram of differentially regulated genes from RNA-seq and MBD-seq data sets; 564 genes with differential methylation between P1 and P14 were also differentially expressed during this period. (Lower) The scatter plot shows log2 fold changes in mRNA expression levels (y axis) against log2 fold changes in MBD enrichment (x axis). B) qPCR gene expression profiling of Igf2bp3 and Neat1 in mouse cardiac ventricles from P1 to P84. Expression levels of Igf2bp3 and Neat1 are normalized to 18S and represented as a fold change relative to P1 ventricles (n = 3 per group). Data are presented as mean 6 SEM. *P , 0.05, **P , 0.001. C) qPCR gene expression profiling of Igf2bp3 and Neat1 in saline and 5aza-dCtreated hearts at P7. Expression levels of Igf2bp3 and Neat1 are normalized to 18S and represented as a fold change relative to P1 ventricles (n = 4–6 per group). Data are presented as mean 6 SEM. *P , 0.05, **P , 0.001. D) Bisulfite sequencing of Igf2bp3 and Neat1 in cardiac ventricles at P1, P14, and P28 (n = 3 per group). A schematic of the genomic region selected for bisulfite sequencing is shown above with the sequenced region denoted by red arrows. Black dot, methylated CpG site; white dot, unmethylated CpG site. Quantification of the percentage of CpG methylation of target genes at P1, P14, and P28 is also shown. Data are presented as mean 6 SEM. *P , 0.05, **P , 0.001.

analysis also revealed a significant enrichment for genes associated with developmental signaling pathways such as Notch, but, in addition, this gene set was also highly enriched for genes associated with muscle contraction, 10

Vol. 29 April 2015

inflammation, and the immune response (Supplemental Fig. S3A, B). Analysis of this gene set identified multiple interaction nodes converging on Notch1, NF-kB, and the CREB binding protein (Supplemental Fig. S3B). Although

The FASEB Journal x www.fasebj.org

SIM ET AL.

TABLE 4. Summary of the total number of genes and distribution of DNA methylation marks with differential methylation levels and mRNA expression changes between P1 and P14 % Correlation of mRNA expression levels and DNA methylation levels

↑CpG ↓mRNA ↓CpG ↑mRNA ↑CpG ↑mRNA ↓CpG ↓mRNA Total number of genes

Number of genes

143 36 355 30 564

TSS 1 kbp

TSS 3 kbp

Exon

Coding DNA sequence

25.9 19.4 36.1 16.7

37.1 41.7 55.5 30.0

69.9 22.2 90.1 43.3

65.0 16.7 87.0 30.0

CGI

CGS

Intergenic

Intron

16.1 0.0 34.9 0.0

25.2 16.7 45.6 13.3

4.9 2.8 8.2 10.0

90.9 88.9 94.1 86.7

Please note that many DMRs overlapped multiple regions of a gene (e.g., a DMR can span exons and introns of the same gene). Therefore, the percentage of the distribution of DNA methylation marks does not add up to 100%. CGI, CpG island; CGS, CpG shore.

we did not identify any significant interaction networks for the small subset of genes displaying decreased methylation and decreased expression at P14, analysis of the gene set associated with decreased methylation and increased expression identified a significant interaction network involving multiple components of the epidermal growth factor receptor and the phosphoinositide 3-kinase/Akt signaling pathway (Supplemental Fig. S4A, B). These findings suggest that DNA methylation regulates the expression of multiple components of key developmental signaling pathways for cardiac development during neonatal life. Analysis of DNase hypersensitivity regions and transcription factor binding sites associated with DNA methylation during neonatal cardiac maturation DNA methylation peaks at P14 were intersected with DNase hypersensitivity sites, which are indicative of open chromatin and active transcription, using GSEA with reference to the mouse ENCODE project. Intersection of DNA methylation peaks at P14 with ENCODE data identified DNase hypersensitivity regions associated with undifferentiated mouse embryonic stem cells (Fig. 7A). GSEA was also used to intersect P14 methylation peaks with ChIPseq data from the mouse ENCODE project for identification of transcription factor binding sites. Intersection of DNA methylation peaks at P14 with ENCODE revealed a significant overlap with ChIP-seq data sets for multiple transcription factors associated with muscle differentiation in C2C12 cells (Fig. 7B). Notably, P14 methylation peaks displayed significant intersection with myogenic transcription factor binding sites (e.g., myogenin and Myod), as well as transcriptional repressors (neuron-restrictive silencer factor and CCTC binding factor) and transcriptional effectors of Wnt signaling (transcription factor 3) (Fig. 7B). We additionally performed a de novo search for enriched transcription factor binding sites that overlapped with DNA methylation peaks at P1 and P14. Intersection of DNA methylation peaks at P14 with known transcription factor binding motifs using Homer revealed a highly significant enrichment for several transcription factor binding sites associated with myogenic differentiation including MADSbox, Tbx5, Smad2/3/4, and Foxh1 (Fig. 7C). These findings point to a potential interaction between myogenic and Smad transcription factors with the DNA methylation machinery during neonatal heart maturation. DNA METHYLATION REGULATES CARDIAC DEVELOPMENT

DISCUSSION In this study, we demonstrate for the first time that DNA methylation events are associated with changes in cardiac gene expression during neonatal heart development. Our findings indicate that DNA methylation levels are not static during postnatal heart development but rather undergo dynamic alterations during neonatal life. Specifically, we identified 2 postnatal waves of DNA methylation in the rodent heart involving increased site-specific methylation from P1 to P14, followed by a global decrease in genomic methylation levels after P14. Importantly, global inhibition of DNA methylation during the neonatal period promotes cardiac cell proliferation and inhibits cardiomyocyte binucleation, suggesting that DNA methylation is required for neonatal cardiac maturation. Bioinformatic analyses suggest that this process might be regulated through methylation of several canonical developmental and differentiation-associated signaling pathways including Hedgehog, TGF-b, FGF, Notch, and Wnt. Notably, DMRs were highly enriched for MyoG, Tbox, and Smad 2/3/4 transcription factor binding sites, suggesting that interplay between myogenic transcription factors and the epigenetic machinery might contribute to the regulation of gene expression networks during neonatal cardiac differentiation. The current study identifies a novel role for DNA methylation during neonatal heart development and is consistent with an epigenetic mechanism for cell cycle arrest in the postnatal heart. One of the most surprising observations of the current study was the identification of 2 postnatal waves of DNA methylation: one coinciding predominantly with increases in site-specific methylation from P1 to P14 and another involving global hypomethylation after P14. This phenomenon is highly reminiscent of previously reported patterns of DNA methylation associated with aging and cancer (30). Specifically, methylation drift occurs during aging and cancer development, with numerous studies identifying an increase in site-specific DNA methylation (e.g., at promoter regions), occurring coincident with global hypomethylation across intergenic regions and repeat sequences (30). This DNA methylation pattern is conserved between species and across multiple organs during aging. Notably, a prominent methylation shift is observed in more proliferative tissues, including colon, stomach, and liver (30). Interestingly, multiple studies suggest that agerelated changes in DNA methylation predominantly occur 11

Figure 6. GO and biologic network analysis for hypermethylated genes associated with transcriptional repression during neonatal cardiac maturation. A) GO process network analysis for the 143 genes with increased methylation and decreased mRNA expression from P1 to P14. B) Interaction network created by GO using genes with increased DNA methylation and decreased mRNA expression levels from P1 to P14.

at genes associated with developmental signaling pathways and organogenesis (30), which is similar to the findings of the current study in neonatal hearts. Although the functional significance and specific genomic regions affected by the global hypomethylation after P14 are currently unclear, this process may reflect an agedependent loss of DNA integrity during postnatal cardiac maturation. A recent report provided evidence for a progressive increase in oxidative DNA damage during postnatal cardiac maturation (31), and several studies have suggested that the presence of unrepaired lesions in DNA substantially alters the methylation capacity of 12

Vol. 29 April 2015

DNA methyltransferases, leading to DNA hypomethylation (32, 33). Interestingly, Gadd45a, which is induced by DNA damage and has been implicated in active DNA demethylation (34), increased at P14. However, transcription of other genes implicated in DNA demethylation, such as Tet1-3, decreased during postnatal heart development. It is currently unclear whether these DNA demethylating enzymes play a major role in the hypomethylation of the cardiac genome after P14, and further studies will be required to determine whether this is a cardiac-specific phenomenon or whether it has broader implications for postnatal maturation of other tissues.

The FASEB Journal x www.fasebj.org

SIM ET AL.

Figure 7. Analysis of differentially methylated regions for enriched DNase hypersensitivity sites and enriched transcription factor binding sites at P14. A) Relative fold enrichment of mouse ENCODE DNase HS between P14:P1 peak sets. Peak intersections with a positive value show enrichment in P14 peak regions (hypermethylation), whereas negative values depict enrichment in the P1 peak regions (hypomethylation). B) Relative fold enrichment of Mouse ENCODE transcription factor binding sites between P14:P1 peak sets. Peak intersections with a positive value show enrichment in P14 peak regions (hypermethylation), whereas negative values depict enrichment in the P1 peak regions (hypomethylation). C) De novo analysis of enriched transcription factor binding sites between P14 and P1 using Homer. Transcription factor binding motifs for Foxh1, Smad3, and MADS box are displayed on the right.

Our findings suggest that DNA methylation events during the first 2 wk of life in rodents are required for CM cell cycle arrest and binucleation in vivo. Although our study design does not permit exclusion of a possible contribution of systemic effects of 5-azacytidine to the observed cardiac phenotype, our in vivo findings are consistent with 2 recent in vitro studies in cultured neonatal rat CMs, which suggested that 5-azacytidine induces CM proliferation and DNA METHYLATION REGULATES CARDIAC DEVELOPMENT

inhibits endothelin-induced CM binucleation (7, 35, 36). These striking effects of 5-azacytidine on CM binucleation are particularly interesting given that hypomethylating agents such as 5-azacytidine are commonly used as anticancer drugs and have antiproliferative effects in other cell types. However, it is also known that DNA methylation can have either positive or negative effects on cell proliferation depending on the epigenetic modification of different 13

subsets of genes (e.g., oncogenes or tumor suppressor genes). In the present study, we have identified complex relationships between DNA methylation during neonatal life and multiple components of key developmental signaling pathways. Neonatal cardiac maturation is associated with hypermethylation of multiple genes associated with critical cardiac developmental signaling pathways such as Hedgehog, FGF, TGF-b, Notch, and Wnt signaling. Interestingly, many of these signaling pathways play key roles in the maintenance of CM proliferation during zebrafish heart regeneration (37) and activation of these pathways drives precocious proliferation of neonatal rodent CMs (1). However, reactivation of developmental signaling in adult CMs does not robustly reinduce cell cycle activity in adult binucleated CMs (1, 7). These findings suggest that epigenetic regulation of key developmental signaling pathways during neonatal life contributes to CM terminal differentiation. It is noteworthy that differentially methylated regions at P14 were highly enriched for transcription factor binding motifs for myogenic transcription factors (e.g., MADS-box and Tbx5) and Smad-related transcription factors (e.g., Smad2/3/4 and Foxh1). The over-representation of these transcription factor binding sites is particularly interesting given the critical role of TGF-b and myogenic transcription factors during muscle differentiation. For example, Smad transcription factors interact with Mef2 to target critical components of the myogenic transcription machinery during C2C12 differentiation (38). Although the precise mechanism by which the DNA methylation machinery is recruited to specific DNA sequences during neonatal heart development is currently unclear, this process may involve interactions between de novo methyltransferases (Dnmt3a/b) and myogenic transcription factors (39). Moreover, the role of TGF-b signaling warrants further investigation, as a recent study suggests that TGF-binduced antiproliferative responses involve the recruitment of DNA methylation/de-methylation enzymes to Smad2/3 target genes (40). A recent study also suggests that interactions between Foxh1/Smad2/3 transcriptional complexes and switch-enhancer elements for pluripotency genes underlie context-specific responses to TGF-b mesoendoderm specification (41). Interestingly, we observed significant enrichment for pluripotency-related genes at DNase hypersensitivity regions, as well as enrichment for Smad2/3/4 and Foxh1 binding sites, suggesting that a similar mechanism may be engaged during cardiomyocyte terminal differentiation. Future studies will be required to dissect the potential involvement of TGF-b signaling, as well as interactions between the DNA methylation machinery and myogenic transcription factors, during cardiomyocyte terminal differentiation. Overall, the present study provides novel insights into the regulatory role of DNA methylation during neonatal cardiac maturation. Although further work is required to elucidate the mechanisms that induce specific methylation changes in the postnatal heart, the current study provides support for an epigenetic mechanism for cardiomyocyte terminal differentiation. Additional interrogation of the role of DNA methylation during neonatal heart development may provide an enhanced understanding of the mechanisms that drive cardiac cell cycle arrest and binucleation. 14

Vol. 29 April 2015

The authors thank Prof. Eric Olson (University of Texas Southwestern Medical Center) for provision of tissue samples and for helpful discussions on the manuscript. The authors thank Ms. Sindhu Igoor (University of Queensland) for technical assistance with animal studies and Mr. Gregory Quaife-Ryan (University of Queensland) for advice on bioinformatics analysis methods. The authors also acknowledge Queensland Facility for Advanced Bioinformatics (QFAB) Bioinformatics and funding provided by Australian Research Council (ARC) Linkage Infrastructure, Equipment and Facilities Funding (LIEF) LE120100071 for providing access to the MetaCore program, as well as support, in part, by the Victorian Government’s Operational Infrastructure Support program. E.R.P., A.E.-O., and J.E.H. are supported by fellowships and grants from the National Health and Medical Research Council (Australia) and the National Heart Foundation of Australia. E.R.P. is also supported by a University of Queensland (UQ) postdoctoral fellowship. C.B.S. is supported by a UQ International (UQI) PhD scholarship.

REFERENCES 1. Porrello, E. R., and Olson, E. N. (2014) A neonatal blueprint for cardiac regeneration [Epub ahead of print]. Stem Cell Res. (Amst.) 10.1016/j.scr.2014.06.003 2. Porrello, E. R., Mahmoud, A. I., Simpson, E., Hill, J. A., Richardson, J. A., Olson, E. N., and Sadek, H. A. (2011) Transient regenerative potential of the neonatal mouse heart. Science (New York, N.Y 331, 1078–1080 3. Porrello, E. R., Mahmoud, A. I., Simpson, E., Johnson, B. A., Grinsfelder, D., Canseco, D., Mammen, P. P., Rothermel, B. A., Olson, E. N., and Sadek, H. A. (2013) Regulation of neonatal and adult mammalian heart regeneration by the miR-15 family. Proc. Natl. Acad. Sci. USA 110, 187–192 4. Chang, C. P., and Bruneau, B. G. (2012) Epigenetics and cardiovascular development. Annu. Rev. Physiol. 74, 41–68 5. Wamstad, J. A., Alexander, J. M., Truty, R. M., Shrikumar, A., Li, F., Eilertson, K. E., Ding, H., Wylie, J. N., Pico, A. R., Capra, J. A., Erwin, G., Kattman, S. J., Keller, G. M., Srivastava, D., Levine, S. S., Pollard, K. S., Holloway, A. K., Boyer, L. A., and Bruneau, B. G. (2012) Dynamic and coordinated epigenetic regulation of developmental transitions in the cardiac lineage. Cell 151, 206–220 6. Sdek, P., Zhao, P., Wang, Y., Huang, C. J., Ko, C. Y., Butler, P. C., Weiss, J. N., and Maclellan, W. R. (2011) Rb and p130 control cell cycle gene silencing to maintain the postmitotic phenotype in cardiac myocytes. J. Cell Biol. 194, 407–423 7. Felician, G., Collesi, C., Lusic, M., Martinelli, V., Ferro, M. D., Zentilin, L., Zacchigna, S., and Giacca, M. (2014) Epigenetic modification at Notch responsive promoters blunts efficacy of inducing notch pathway reactivation after myocardial infarction. Circ. Res. 115, 636–649 8. Suzuki, M. M., and Bird, A. (2008) DNA methylation landscapes: provocative insights from epigenomics. Nat. Rev. Genet. 9, 465–476 9. Wu, H., and Zhang, Y. (2014) Reversing DNA methylation: mechanisms, genomics, and biological functions. Cell 156, 45–68 10. Li, E., Bestor, T. H., and Jaenisch, R. (1992) Targeted mutation of the DNA methyltransferase gene results in embryonic lethality. Cell 69, 915–926 11. Chamberlain, A. A., Lin, M., Lister, R. L., Maslov, A. A., Wang, Y., Suzuki, M., Wu, B., Greally, J. M., Zheng, D., and Zhou, B. (2014) DNA methylation is developmentally regulated for genes essential for cardiogenesis. J Am Heart Assoc 3, e000976 12. Movassagh, M., Choy, M. K., Knowles, D. A., Cordeddu, L., Haider, S., Down, T., Siggens, L., Vujic, A., Simeoni, I., Penkett, C., Goddard, M., Lio, P., Bennett, M. R., and Foo, R. S. (2011) Distinct epigenomic features in end-stage failing human hearts. Circulation 124, 2411–2422 13. Haas, J., Frese, K. S., Park, Y. J., Keller, A., Vogel, B., Lindroth, A. M., Weichenhan, D., Franke, J., Fischer, S., Bauer, A., Marquart, S., Sedaghat-Hamedani, F., Kayvanpour, E., K¨ohler, D., Wolf, N. M., Hassel, S., Nietsch, R., Wieland, T., Ehlermann, P., Schultz, J. H., D¨osch, A., Mereles, D., Hardt, S., Backs, J., Hoheisel, J. D., Plass, C., Katus, H. A., and Meder, B. (2013) Alterations in cardiac DNA methylation in human dilated cardiomyopathy. EMBO Mol. Med. 5, 413–429

The FASEB Journal x www.fasebj.org

SIM ET AL.

14. Xiao, D., Dasgupta, C., Chen, M., Zhang, K., Buchholz, J., Xu, Z., and Zhang, L. (2013) Inhibition of DNA methylation reverses norepinephrine-induced cardiac hypertrophy in rats. Cardiovasc. Res. 101, 373–382 15. Watson, C. J., Collier, P., Tea, I., Neary, R., Watson, J. A., Robinson, C., Phelan, D., Ledwidge, M. T., McDonald, K. M., McCann, A., Sharaf, O., and Baugh, J. A. (2013) Hypoxiainduced epigenetic modifications are associated with cardiac tissue fibrosis and the development of a myofibroblast-like phenotype. Hum. Mol. Genet. 23, 2176–2188 16. Porrello, E. R., D’Amore, A., Curl, C. L., Allen, A. M., Harrap, S. B., Thomas, W. G., and Delbridge, L. M. (2009) Angiotensin II type 2 receptor antagonizes angiotensin II type 1 receptormediated cardiomyocyte autophagy. Hypertension 53, 1032–1040 17. Porrello, E. R., Johnson, B. A., Aurora, A. B., Simpson, E., Nam, Y. J., Matkovich, S. J., Dorn II, G. W., van Rooij, E., and Olson, E. N. (2011) MiR-15 family regulates postnatal mitotic arrest of cardiomyocytes. Circ. Res. 109, 670–679 18. Stevens, M., Cheng, J. B., Li, D., Xie, M., Hong, C., Maire, C. L., Ligon, K. L., Hirst, M., Marra, M. A., Costello, J. F., and Wang, T. (2013) Estimating absolute methylation levels at single-CpG resolution from methylation enrichment and restriction enzyme sequencing methods. Genome Res. 23, 1541–1553 19. Wu, J., Anczuk´ow, O., Krainer, A. R., Zhang, M. Q., and Zhang, C. (2013) OLego: fast and sensitive mapping of spliced mRNASeq reads using small seeds. Nucleic Acids Res. 41, 5149–5163 20. Quinlan, A. R., and Hall, I. M. (2010) BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics 26, 841–842 21. Robinson, M. D., McCarthy, D. J., and Smyth, G. K. (2010) edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26, 139–140 22. R Core Team (2013) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria 23. Li, H., and Durbin, R. (2009) Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 25, 1754–1760 24. Zhang, Y., Liu, T., Meyer, C. A., Eeckhoute, J., Johnson, D. S., Bernstein, B. E., Nusbaum, C., Myers, R. M., Brown, M., Li, W., and Liu, X. S. (2008) Model-based analysis of ChIP-Seq (MACS). Genome Biol. 9, R137 25. Subramanian, A., Tamayo, P., Mootha, V. K., Mukherjee, S., Ebert, B. L., Gillette, M. A., Paulovich, A., Pomeroy, S. L., Golub, T. R., Lander, E. S., and Mesirov, J. P. (2005) Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. USA 102, 15545–15550 26. Rohde, C., Zhang, Y., Reinhardt, R., and Jeltsch, A. (2010) BISMA—fast and accurate bisulfite sequencing data analysis of individual clones from unique and repetitive sequences. BMC Bioinformatics 11, 230 27. Benjamini, Y., and Hochberg, Y. (1995) Controlling the false discovery rate: a practical and powerful approach to multiple testing. J Roy Stat Soc B Met 57, 289–300 28. Pirola, L., Balcerczyk, A., Tothill, R. W., Haviv, I., Kaspi, A., Lunke, S., Ziemann, M., Karagiannis, T., Tonna, S., Kowalczyk, A., Beresford-Smith, B., Macintyre, G., Kelong,

DNA METHYLATION REGULATES CARDIAC DEVELOPMENT

29. 30. 31.

32.

33. 34. 35.

36.

37.

38. 39. 40.

41.

M., Hongyu, Z., Zhu, J., and El-Osta, A. (2011) Genome-wide analysis distinguishes hyperglycemia regulated epigenetic signatures of primary vascular cells. Genome Res. 21, 1601–1615 Jjingo, D., Conley, A. B., Yi, S. V., Lunyak, V. V., and Jordan, I. K. (2012) On the presence and role of human gene-body DNA methylation. Oncotarget 3, 462–474 Issa, J. P. (2014) Aging and epigenetic drift: a vicious cycle. J. Clin. Invest. 124, 24–29 Puente, B. N., Kimura, W., Muralidhar, S. A., Moon, J., Amatruda, J. F., Phelps, K. L., Grinsfelder, D., Rothermel, B. A., Chen, R., Garcia, J. A., Santos, C. X., Thet, S., Mori, E., Kinter, M. T., Rindler, P. M., Zacchigna, S., Mukherjee, S., Chen, D. J., Mahmoud, A. I., Giacca, M., Rabinovitch, P. S., Aroumougame, A., Shah, A. M., Szweda, L. I., and Sadek, H. A. (2014) The oxygen-rich postnatal environment induces cardiomyocyte cell-cycle arrest through DNA damage response. Cell 157, 565–579 James, S. J., Pogribny, I. P., Pogribna, M., Miller, B. J., Jernigan, S., and Melnyk, S. (2003) Mechanisms of DNA damage, DNA hypomethylation, and tumor progression in the folate/methyl-deficient rat model of hepatocarcinogenesis. J. Nutr. 133(11, Suppl 1)3740S–3747S Cerda, S., and Weitzman, S. A. (1997) Influence of oxygen radical injury on DNA methylation. Mutat. Res. 386, 141–152 Niehrs, C., and Sch¨afer, A. (2012) Active DNA demethylation by Gadd45 and DNA repair. Trends Cell Biol. 22, 220–227 Paradis, A., Xiao, D., Zhou, J., and Zhang, L. (2014) Endothelin-1 promotes cardiomyocyte terminal differentiation in the developing heart via heightened DNA methylation. Int. J. Med. Sci. 11, 373–380 Kou, C. Y., Lau, S. L., Au, K. W., Leung, P. Y., Chim, S. S., Fung, K. P., Waye, M. M., and Tsui, S. K. (2010) Epigenetic regulation of neonatal cardiomyocytes differentiation. Biochem. Biophys. Res. Commun. 400, 278–283 Choi, W. Y., Gemberling, M., Wang, J., Holdway, J. E., Shen, M. C., Karlstrom, R. O., and Poss, K. D. (2013) In vivo monitoring of cardiomyocyte proliferation to identify chemical modifiers of heart regeneration. Development 140, 660–666 Liu, D., Kang, J. S., and Derynck, R. (2004) TGF-beta-activated Smad3 represses MEF2-dependent transcription in myogenic differentiation. EMBO J. 23, 1557–1566 Hervouet, E., Vallette, F. M., and Cartron, P. F. (2009) Dnmt3/ transcription factor interactions as crucial players in targeted DNA methylation. Epigenetics 4, 487–499 Thillainadesan, G., Chitilian, J. M., Isovic, M., Ablack, J. N., Mymryk, J. S., Tini, M., and Torchia, J. (2012) TGF-b-dependent active demethylation and expression of the p15ink4b tumor suppressor are impaired by the ZNF217/CoREST complex. Mol. Cell 46, 636–649 Beyer, T. A., Weiss, A., Khomchuk, Y., Huang, K., Ogunjimi, A. A., Varelas, X., and Wrana, J. L. (2013) Switch enhancers interpret TGF-b and Hippo signaling to control cell fate in human embryonic stem cells. Cell Reports 5, 1611–1624 Received for publication September 9, 2014. Accepted for publication November 18, 2014.

15