Polycomb Repressive Complex 2 Regulates Normal Development of the Mouse HeartNovelty and Significance
Rationale: Epigenetic marks are crucial for organogenesis, but their role in heart development is poorly understood. Polycomb repressive complex 2 (PRC2) trimethylates histone H3 at lysine 27, which establishes H3K27me3 repressive epigenetic marks that promote tissue-specific differentiation by silencing ectopic gene programs.
Objective: We studied the function of PRC2 in murine heart development using a tissue-restricted conditional inactivation strategy.
Methods and Results: Inactivation of the PRC2 subunit Ezh2 by Nkx2–5Cre (Ezh2NK) caused lethal congenital heart malformations, namely, compact myocardial hypoplasia, hypertrabeculation, and ventricular septal defect. Candidate and genome-wide RNA expression profiling and chromatin immunoprecipitation analyses of Ezh2NK heart identified genes directly repressed by EZH2. Among these were the potent cell cycle inhibitors Ink4a/b (inhibitors of cyclin-dependent kinase 4 A and B), the upregulation of which was associated with decreased cardiomyocyte proliferation in Ezh2NK. EZH2-repressed genes were enriched for transcriptional regulators of noncardiomyocyte expression programs such as Pax6, Isl1, and Six1. EZH2 was also required for proper spatiotemporal regulation of cardiac gene expression, because Hcn4, Mlc2a, and Bmp10 were inappropriately upregulated in ventricular RNA. PRC2 was also required later in heart development, as indicated by cardiomyocyte-restricted TNT-Cre inactivation of the PRC2 subunit Eed. However, Ezh2 inactivation by TNT-Cre did not cause an overt phenotype, likely because of functional redundancy with Ezh1. Thus, early Ezh2 inactivation by Nk2–5Cre caused later disruption of cardiomyocyte gene expression and heart development.
Conclusions: Our study reveals a previously undescribed role of EZH2 in regulating heart formation and shows that perturbation of the epigenetic landscape early in cardiogenesis has sustained disruptive effects at later developmental stages.
Congenital heart disease is among the most frequent major birth defects. Intensive studies have revealed numerous genes that control the intricate process of heart development in humans and mice.1 Prominent among congenital heart disease genes are transcription factors, which points to the important role of transcriptional regulation in orchestrating heart development.2 Recent studies on the SWI/SNF chromatin remodeling complex have also highlighted the important role of chromatin structure in regulating heart development.3 In Drosophila, Polycomb group genes genetically antagonize these chromatin remodeling genes.4 The Polycomb genes Ezh2 (enhancer of Zeste 2), Eed (embryonic ectoderm development), and Suz12 (suppressor of Zeste 12) are major components of the Polycomb repressive complex 2 (PRC2).5 This highly conserved complex trimethylates histone H3 on lysine 27 (H3K27), which establishes the H3K27me3 epigenetic mark that characterizes transcriptionally repressed chromatin.4 Through this mechanism, PRC2 regulates tissue-specific differentiation by orchestrating the repression of inappropriate tissue- and stage-specific transcriptional programs.6–8 The role of histone methylation or PRC2 in heart development has not been reported.
Here, we used Cre-LoxP technology to conditionally inactivate Ezh2, encoding the catalytic subunit of PRC2,5,9 in the developing heart. Cardiac Ezh2 deficiency during a narrow window early in cardiogenesis caused abnormal cardiac gene expression and lethal heart malformations that included hypertrabeculation, thinning of the compact myocardium, and ventricular and atrial septal defects. Our results indicate that establishment of the epigenetic landscape early in cardiogenesis is necessary for normal heart development. Perturbation of this process may contribute to congenital heart disease.
An expanded Methods section is available in the Online Data Supplement at http://circres.ahajournals.org.
Ezh2fl, Nkx2–5Lz, TNT-Cre, and Nkx2–5Cre alleles were described previously.7,10–12 Construction of the Eedfl allele will be described elsewhere. All animal experiments were performed according to protocols approved by the Institutional Animal Care and Use Committee of Children's Hospital Boston.
Histological sections were processed as described previously13 and imaged on an Olympus FV1000 confocal microscope. Antibodies are listed in Online Table I. For morphometric measurements, we measured transverse sections through the atrioventricular valves. Wall thickness was calculated as the ventricular compact myocardial thickness divided by its outer circumference. Trabecular and myocardial area were measured in Adobe Photoshop after selection of image areas with myocardial color range.
Quantitative reverse-transcription–polymerase chain reaction (qRTPCR) was performed with SYBR green chemistry and normalized to GAPDH. Chromatin immunoprecipitation (ChIP) was performed with antibodies listed in Online Table I. Primers for qRTPCR and quantitative ChIP-PCR (ChIP-qPCR) are provided in Online Table II.
RNA analysis by next-generation sequencing (RNA-seq) was performed as described previously, with omission of library normalization steps.14 Genes were analyzed for differential expression by use of exact tests between 2 negative binomial groups,15 and the false-discovery rate was calculated by the Benjamini and Hochberg method. ChIP followed by next-generation sequencing (ChIP-seq) was performed as described previously.16 Libraries were sequenced on an Illumina HiSeq 2000. ChIP-seq peaks were identified by Sole-Search17 with default parameters. Genomic intervals were annotated, combined, and analyzed for gene ontology term enrichment by use of the completeMOTIFs pipeline.18 Data were deposited to the Gene Expression Omnibus, accession number GSE29997 and are available at the Cardiovascular Development Consortium Server, http://b2b.hci.utah.edu.
qRTPCR results are expressed as mean±SEM, whereas ChIP-qPCR results are displayed as mean±SD. Two-group comparisons were performed with Student t test, with P<0.05 taken as statistically significant.
Early Cardiac Inactivation of Ezh2 by Nkx2–5Cre Disrupts Heart Development
We investigated expression of the PRC2 subunits Eed, Ezh2, and Suz12 in fetal and adult heart. In fetal heart, all 3 subunits were robustly expressed at the mRNA and protein levels (Figures 1A and 1B). In adult heart, these transcripts were downregulated, and EZH2 and EED proteins were not detected. Despite reduced Suz12 mRNA, SUZ12 protein was upregulated in adult heart (Figure 1B), which suggests posttranscriptional regulation. Immunohistochemistry of E9.5 and E12.5 heart confirmed EZH2 expression in cardiomyocytes and nonmyocytes (Figure 1C).
To study the role of PRC2 in heart development, we inactivated an Ezh2 floxed allele (Ezh2fl; Shen et al7) in cardiomyocytes using Nkx2–5Cre.12 By embryonic day 9.5 (E9.5), Ezh2fl/fl ::Nkx2–5Cre/+ (abbreviated Ezh2NK) cardiomyocytes exhibited markedly reduced EZH2 immunoreactivity (Figure 1C). EZH2 expression appeared unchanged in epicardial and endocardial cells, although we cannot exclude Ezh2 inactivation in a subset of these cells. Quantitative PCR confirmed marked Ezh2 depletion in Ezh2NK mutant hearts (Figure 1D). Ezh1, which shows partial functional redundancy with Ezh2,7 was upregulated in Ezh2NK mutants (Figure 1D). Increased Ezh1 expression may mitigate the severity of the Ezh2NK phenotype (Figure 1D). The major described product of EZH2 methyltransferase activity, H3K27me3, was markedly reduced in Ezh2NK mutants at E9.5 and E16.5 (Figure 1E), consistent with an essential role of EZH2 in H3K27 trimethylation.
Ezh2NK embryos were present at the expected Mendelian ratio at E16.5, but the vast majority died perinatally, so that only 2% of pups at weaning had the mutant genotype (Figure 1F). The mutant hearts had 4 chambers, 4 valves, and normal inflow and outflow tract alignment. The ventricular septa were abnormal and exhibited both muscular and membranous septal defects (Figure 1G). The compact myocardium (free wall) of both ventricles was hypoplastic, with the right ventricle more severely affected than the left ventricle. Trabeculation of both ventricles was increased. Morphometric measurements confirmed these observations (Figures 1H and 1I). The right atria were markedly dilated, and some mutant hearts had atrial septal defects.
Rare Ezh2NK mutant mice survived to adulthood. These had persistent hypertrabeculation, right ventricular hypoplasia, atrial and ventricular septal defects, myocardial fibrosis, and moderately impaired left ventricular systolic function (Figure 2).
Decreased Cardiomyocyte Proliferation in Ezh2NK Mutant Heart
Hypoplasia of Ezh2NK compact myocardium suggested reduced cardiomyocyte proliferation. To test this hypothesis, we stained histological sections of E16.5 control and mutant heart for phosphorylated histone H3 (pH3), a cell cycle M-phase marker (Figure 3A). Loss of EZH2 caused more than 2-fold reduction in the fraction of pH3+ cardiomyocytes (P<0.05; Figures 3A and 3B). In contrast, we did not detect significant cardiomyocyte apoptosis by terminal dUTP nick end-labeling staining in control or mutant heart (Figure 3C).
In skin, cultured fibroblasts, and hematopoietic cells, PRC2 directly repressed inhibitors of cyclin-dependent kinase 4 (Ink4) A and B (Ink4a and Ink4b; Mouse Genome Informatics: Cdkn2a/b), powerful negative cell cycle regulators that inhibit the G1-to-S transition.19,20 Using ChIP followed by quantitative PCR (ChIP-qPCR), we found that EZH2 and H3K27me3 were enriched at Ink4a and Ink4b promoters of E12.5 heart ventricle (Figure 3D). Ink4a and Ink4b were dramatically upregulated in EZH2-deficient ventricle at both E12.5 and E16.5 (Figure 3E). Thus, Ezh2 promotes cardiomyocyte proliferation in fetal heart by repressing the negative cell cycle regulators Ink4a and Ink4b.
Abnormal Gene Expression in Ezh2NK Mutant Heart
PRC2 promotes lineage-specific differentiation by repressing inappropriate transcriptional programs. To investigate this function of EZH2 in heart, we measured the expression of selected master transcriptional regulators of noncardiac programs by qRTPCR (Figure 4A). Pax6, a transcriptional regulator of eye, brain, spinal cord, and pancreas development,21 was upregulated 49-fold in Ezh2NK mutant heart. Six1, a homeodomain transcription factor expressed in key progenitor populations, including second heart field, and downregulated in most differentiated tissues, including cardiomyocytes,22 was upregulated 17-fold in Ezh2NK mutant heart. Isl1, an essential regulator of cardiac progenitor differentiation, the expression of which is normally downregulated in differentiated cardiomyocytes,23 was upregulated 7.5-fold. We further confirmed upregulation of ISL1 protein in differentiated cardiomyocytes by immunostaining (Figure 4B). Consistent with qRTPCR results, ISL1 immunoreactivity was present in E9.5 Ezh2NK ventricular cardiomyocytes but not in littermate control cardiomyocytes. Both EZH2 and H3K27me3 were highly enriched at Pax6, Six1, and Isl1 promoters by ChIP-qPCR (Figure 4C), which indicates that EZH2 directly represses these genes by establishing H3K27me3 repressive marks.
Cardiac genes undergo an orderly change of expression during cardiomyocyte maturation. Bmp10, which promotes cardiomyocyte growth and ventricular trabeculation,24,25 is expressed in compact and trabecular cardiomyocytes at E10.5, becomes restricted to trabecular cardiomyocytes by E12.5, and is no longer expressed in ventricles by E15.5.24,25 In Ezh2NK mutant embryos, Bmp10 expression was upregulated by >2-fold at both E12.5 and E16.5 (Figure 4D).
During cardiac morphogenesis, ventricular cardiomyocytes undergo lineage restriction, repressing genes normally expressed in atrial or conduction system cardiomyocytes. We measured the expression of conduction system– or atrium-specific markers in E16.5 ventricular myocardium. Hcn4, which encodes a cardiac pacemaker channel normally restricted to conduction system myocardium,26 was upregulated 2.6-fold (Figure 4D). The atrium-specific myosin light chain isoform Mlc2a (Mouse Genome Informatics: Myl7)27 was upregulated 5.3-fold in ventricular myocardium (Figure 4D). Consistent with qRTPCR results, MLC2A immunoreactivity was present in E12.5 Ezh2NK ventricles but not in littermate control ventricles (Figure 4E). These results indicate that EZH2 is required to establish normal patterns of gene expression in the developing heart.
EZH2 and H3K27me3 were moderately enriched at the Hcn4 promoter of wild-type ventricles (Figure 4F); however, neither was enriched at Mlc2a or Bmp10 promoters (Figure 4F). Thus, EZH2 directly repressed some markers of regionalized cardiac gene expression, whereas its regulation of other such markers was either indirect or occurred through a mechanism other than H3K27me3 deposition.
Genome-Wide Identification of EZH2-Repressed Cardiac Genes
To obtain a global view of EZH2-regulated gene expression in the heart, we measured gene expression and EZH2 and H3K27me3 chromatin occupancy in E12.5 ventricle apex by RNA-seq and ChIP-seq, respectively (Table). In the RNA-seq experiment, we obtained 32.2 and 31.5 million uniquely aligned 50-nucleotide paired end reads from Ezh2 heterozygous control (Ezh2fl/+ ::Nkx2–5Cre/+) and mutant (Ezh2fl/fl ::Nkx2–5Cre/+) hearts, respectively. Expression levels in control and mutant samples were compared on the basis of the density of reads that mapped to each gene. A total of 511 genes were differentially expressed (false-discovery rate <0.01 and fold change >50%), including the previously validated genes Ink4a, Six1, Isl1, Pax6, Myh6, Hcn4, and Mlc2a (Figure 5A; Online Table III). These 511 genes were enriched for functional terms related to gland, appendage, ear, and muscle development (Online Table III).
To identify genes directly occupied by EZH2 and its H3K27me3 repressive mark, we performed ChIP-seq using wild-type E12.5 heart ventricle apex. We obtained 16 to 23 million uniquely aligned 50-nucleotide single end reads in input, EZH2, and H3K27me3 ChIP samples, which led to the identification of 819 and 1052 genes whose promoters were occupied by EZH2 and H3K27me3, respectively (Table; Online Table IV). Consistent with EZH2 deposition of H3K27me3 epigenetic marks, in most (697) cases, promoters were co-occupied by both marks (Figure 5A). The ChIP-qPCR data from Figure 4 independently validated the ChIP-seq data: Of the 7 ChIP-qPCR–positive genes, 5 were recovered by ChIP-seq, whereas 2 (Hcn4 and Ink4b) were not. EZH2- and H3K27me3-bound genes had functional annotations related to cell fate commitment, organ morphogenesis, and development of the limbs and skeletal, nervous, and endocrine systems (Online Table IV).
Next, we looked for the genes that were likely to be directly and functionally regulated by EZH2 through deposition of H3K27me3. There were 52 genes that were both differentially expressed downstream of EZH2 loss of function and bound by H3K27me3 and EZH2. All 52 of these genes were upregulated in EZH2 loss of function (Fisher exact test P<0.0001), as expected for genes directly regulated by EZH2 through establishment of repressive H3K27me3 marks (Figure 5A; Online Table V). These 52 EZH2-repressed genes were enriched 14-fold for transcriptional regulators (P=3.9×10−9; green arrowheads in Figure 5B) and functional annotations related to gland, limb, forebrain, ear, and heart development (Online Table V). Among them were Ink4a, Six1, Isl1, and Pax6, identified by the candidate gene approach. The large majority of these showed reduced expression in heart compared with other tissues, such as brain (Figure 5B; Online Figure I).
By analyzing correlations between gene expression in multiple tissues and conditions, in silico we identified tissue-specific gene expression modules and their transcriptional regulators (P.C. and G.Q.D., unpublished data). Components of 3 modules, 1 related to neuronal development and 2 related to fibroblast and mesenchymal-type cells, were overrepresented among genes upregulated in Ezh2NK (P<0.01; Online Table VI). Key transcriptional regulators of these 3 modules were among the 52 EZH2-repressed genes (Online Table VI). These data support an essential role of EZH2 in repressing these mesenchymal and neuronal gene programs in cardiomyocytes.
Later Cardiac Inactivation of Ezh2 by TNT-Cre Is Compatible With Normal Heart Development
Gene inactivation by TNT-Cre occurs in differentiated cardiomyocytes slightly later than Nkx2–5Cre.11 Unlike Ezh2NK mutants, Ezh2fl/fl ::Tg(TNT-Cre) mutant mice (Ezh2TNT) were born at the expected mendelian frequency (Figure 6A) and had normal heart structure (Figure 6B). Echocardiographic analysis showed no difference in heart function (control fractional shortening 50.5±2.3% versus Ezh2TNT fractional shortening 50.7±0.9%; n=7). qRTPCR validated efficient Ezh2 inactivation by E9.5 (Figure 6C). The Nkx2–5Cre knock-in allele is Nkx2–5 haploinsufficient. To exclude Nkx2–5 haploinsufficiency as the cause of the differences observed between Ezh2NK and Ezh2TNT mutants, we generated Ezh2fl/fl ::Tg(TNT-Cre) ::Nkx2–5Lz/+ mice. Survival of these mice to weaning was not distinguishable from Ezh2TNT mice, which indicates that Nkx2–5 haploinsufficiency did not account for the different phenotypes obtained with Nkx2–5Cre versus TNT-Cre.
Although the extent of EZH2 inactivation was similar between Ezh2NK and Ezh2TNT by E9.5, Ezh2NK mutants developed lethal heart defects, whereas Ezh2TNT mutants did not. We compared the effect of Ezh2 inactivation by Nkx2–5Cre versus TNT-Cre on gene expression in E12.5 ventricle (Figure 6D). The cell cycle inhibitors Ink4a and Ink4b were weakly upregulated in Ezh2TNT, but the degree of upregulation was substantially and significantly greater in Ezh2NK. The cardiac progenitor gene Isl1 and the noncardiac transcriptional regulator Pax6 were also more strongly upregulated in Ezh2NK than in Ezh2TNT. The gene Mlc2a, normally expressed in atrial myocardium, was upregulated in Ezh2NK but not Ezh2TNT ventricular RNA. However, the cardiac progenitor gene Six1 was upregulated to the same degree in Ezh2NK and Ezh2TNT, which suggests that Six1 upregulation by itself did not cause lethal heart malformation in combination with Ezh2 inactivation. Collectively, these data indicate that slightly earlier inactivation of Ezh2 by Nkx2–5Cre compared with TNT-Cre causes sustained effects on gene expression and heart development.
Ezh2 and the related Ezh1 show partial functional redundancy in some cell types.7,28 Functional redundancy between Ezh2 and Ezh1 in the fetal heart may account for the lack of phenotype in Ezh2TNT mutants. To test this hypothesis, we inactivated the essential, nonredundant PRC2 component Eed with TNT-Cre in fetal cardiomyocytes. In crosses between Eedfl/+ :: Tg(TNT-Cre) and Eedfl/fl parents, only 11% of progeny had the Eedfl/fl ::Tg(TNT-Cre) mutant genotype (abbreviated EedTNT) at birth, and most mutants died by postnatal day 3, so that only 2% survived to weaning (Figure 7A). These data indicated that EedTNT mutants had perinatal lethal heart defects. Histological sections of late-gestation embryo hearts showed that the mutant compact myocardium was dramatically thinned (Figure 7B). By morphometric analysis, the mutant compact myocardium thickness was reduced to 32% that of control myocardium (P<0.005, n=4; Figure 7C). Unlike Ezh2NK mutants, EedTNT mutants had intact atrial and ventricular septae. Consistent with myocardial thinning, cardiomyocyte proliferation, as assessed by pH3 staining, was strongly reduced (Figures 7D and 7E). As in Ezh2NK, EedTNT mutants exhibited strong upregulation of Ink4a, Ink4b, Isl1, Pax6, and Six1 (Figure 7F). Cardiac genes Mlc2a and Hcn4 were also upregulated in EedTNT, as in Ezh2NK; however, Bmp10 was downregulated in EedTNT, whereas it was upregulated in Ezh2NK. These data indicate an ongoing requirement for PRC2 in fetal cardiomyocytes for normal heart growth. This requirement was not reflected in Ezh2TNT mutants, likely because of functional redundancy with Ezh1.
Organ formation requires the establishment and maintenance of tissue-specific gene expression programs in response to developmental cues. This involves both activation of tissue-specific genes and repression of inappropriate genes. By repressing inappropriate gene expression, Polycomb group genes play pivotal roles in cellular differentiation and organ formation. The present results establish the essential role of PRC2 and its subunits EZH2 and EED in orchestrating heart development and cardiac gene expression. Loss of EZH2 in cardiac progenitors and in cardiomyocytes mediated by Nkx2–5Cre resulted in lethal heart abnormalities and disrupted cardiomyocyte gene expression. The ongoing requirement for PRC2 activity in cardiomyocytes was reinforced by TNT-Cre–mediated EED loss of function. Thus, our studies of PRC2 function in the developing heart highlight the key role of histone modifications and the chromatin landscape in regulating cardiac gene expression and heart development.
We identified 52 genes that were directly repressed by EZH2-mediated deposition of H3K27me3. These were highly enriched for transcriptional regulators of noncardiomyocyte gene expression programs, particularly those of neuronal and mesenchymal cells. Expression of these ectopic programs likely contributed to abnormal heart development and function in Ezh2NK.
Interestingly, only a small fraction of genes occupied by EZH2 and H3K27me3 were upregulated by Ezh2 inactivation (52 of 697 genes). This observation may be explained in part by functional redundancy with Ezh1, which showed compensatory upregulation. However, overall H3K27me3 levels were strongly reduced in the Ezh2NK mutant. This suggests that additional mechanisms besides PRC2-mediated deposition of H3K27me3 maintain repression of the majority of Ezh2/H3K27me3 repressed genes. These additional mechanisms may include establishment of other epigenetic repressive marks, such as DNA methylation.29 EZH2 and H3K27me3 may participate in initiation of these additional repressive marks but may be dispensable for their maintenance.
One of the major effects of PRC2 loss of function (Ezh2NK and EedTNT) was dramatic thinning of the compact myocardium and decreased cardiomyocyte proliferation. PRC2 inactivation derepressed the potent cell cycle inhibitors Ink4a/b. The strength of Ink4a/b upregulation correlated with myocardial hypoplasia, with Ezh2NK and EedTNT mutants exhibiting strong upregulation and Ezh2TNT mutants showing less substantial upregulation. PRC2 has been shown to repress Ink4a/b and promote cellular proliferation in skin maturation, fibroblasts, and hematopoietic cells,19,20 which suggests that this is a common mechanism by which PRC2 regulates proliferation in development and disease.
The normal spatiotemporal regulation of cardiac gene expression was also disrupted in Ezh2NK mutants. Mlc2a and Hcn4, genes normally confined to the atria and conduction system, respectively,26,27 were inappropriately expressed in the mutant ventricle. We also detected upregulation of a third cardiac gene, Bmp10, a secreted regulator of myocardial growth and trabeculation, in Ezh2NK mutant heart. Mlc2a upregulation was also observed in EedTNT mutant heart, which suggests that PRC2 acts genetically upstream of this gene to repress its ventricular expression. On the other hand, Bmp10 was downregulated in EedTNT, which suggests that its regulation by PRC2 is complex. In addition, expression of Bmp10 did not track with myocardial growth phenotypes, because it was also upregulated in Ezh2TNT (phenotypically normal) and downregulated in EedTNT (phenotypically similar to Ezh2NK). Thus, the pathogenic role of Bmp10 upregulation in Ezh2NK is uncertain.
PRC2 represses gene expression by establishing H3K27me3 epigenetic marks. EZH2 and H3K27me3 were highly enriched at wild-type Ink4a and Ink4b, consistent with an important role for PRC2 in repressing these cell cycle genes. Hcn4 was moderately enriched for H3K27me3 and EZH2 in wild-type ventricle, which suggests that its regulation by EZH2 is also direct. Mlc2a and Bmp10, however, were not substantially enriched for either H3K27me3 or EZH2, which suggests that their regulation by EZH2 was through a mechanism other than direct EZH2 occupancy and H3K27me3 deposition. In fact, the large majority of genes (432 of 511) differentially expressed in EZH2 mutants were not occupied by either EZH2 or H3K27me3, which suggests that either secondary effects lead to their differential expression or that PRC2 regulates them through additional noncanonical mechanisms.
H3K27me3 epigenetic marks serve as docking sites for repressive complexes including the Polycomb repressive complex 1 (PRC1). The PRC1 complex component Polyhomeotic-like 1 (also known as Rae28) was shown to be essential for heart development, in part by sustaining expression of the Nkx2-5.30 More recently, sumoylation of another PRC1 component, Pc2, was shown to enhance PRC1 recruitment to H3K27me3, and excessive Pc2 sumoylation caused by inactivation of Senp2, which encodes a desumoylating enzyme, was linked to abnormal cardiac development and reduced cardiomyocyte proliferation.31 Indeed, heterozygous loss of SUMO-1 caused perinatal lethality and cardiac septal defects, and SUMO-1 was required for normal expression of cell cycle genes in the fetal heart.32 The interplay between sumoylation and Polycomb-mediated gene silencing may be complex, because both EZH2 and SUZ12 have been shown to be sumoylated.33 The regulation and biological significance of these modifications remain unknown.
Although Ezh2 inactivation by Nkx2–5Cre caused lethal congenital heart malformations, unexpectedly, its inactivation by TNT-Cre did not. These 2 Cre alleles differ in several ways: (1) timing, with Nkx2–5Cre being active in cardiac progenitors and TNT-Cre restricted to differentiated cardiomyocytes; (2) Nkx2–5 gene dosage, because of haploinsufficiency in Nkx2–5Cre; and (3) recombination domain, with TNT-Cre constrained to differentiated cardiomyocytes and Nkx2–5Cre potentially additionally including pharyngeal endoderm, endocardium, and epicardium.12,34 Nkx2–5 haploinsufficiency did not unmask a phenotype in Ezh2TNT ::Nkx2–5Lz/+ heart, which excludes this possibility. Nkx2–5Cre activity in pharyngeal endoderm was not central to causing the Ezh2NK phenotype, because Ezh2NK hearts had normal outflow tracts, and their abnormalities primarily resided in the ventricles. By immunohistochemistry, EZH2 persisted in the majority of nonmyocytes in the Ezh2NK ventricle, although it may have been inactivated in a minor subset of nonmyocytes. This observation suggests that Ezh2 inactivation by Nkx2–5Cre and TNT-Cre occurred in similar spatial domains, and thus, differences in spatial recombination domains are unlikely to account for the dramatic difference in phenotype between Ezh2NK and Ezh2TNT. Conditional inactivation of the nonredundant PRC2 component Eed by TNT-Cre indicates that PRC2 is required in the TNT-Cre recombination domain, which provides further strong support for this conclusion (see below). Ezh1 has been shown to functionally substitute for Ezh2 in postnatal skin homeostasis28 and in execution of embryonic stem cell pluripotency.7 Collectively, these data suggest that Ezh1 and Ezh2 are functionally redundant in cardiomyocytes but not in cardiac progenitor cells (see summary model, Online Figure II). As a result, there is a small window during which Ezh2 is indispensable in cardiac progenitor cells, and this window is probed by Nkx2–5Cre but not TNT-Cre.
TNT-Cre inactivation of EED recapitulated many aspects of the phenotype observed with Nkx2–5Cre inactivation of EZH2, including thin compact myocardium and hypertrabeculation; however, EedTNT mutants had intact cardiac septae, whereas EZH2NK had both atrial and ventricular septal defects. We believe this difference is caused by the timing of cardiomyocyte inactivation by Nkx2–5Cre compared with TNT-Cre. First, the majority of nonmyocytes did not undergo Ezh2 inactivation in Ezh2NK mutants. Second, Eed inactivation with Nkx2–5Cre did fully recapitulate the EZH2NK phenotype, including septal defects (A.H. and W.T.P., unpublished data), strongly arguing that both proteins contribute to cardiac septation through the PRC2 complex. Third, inactivation of Eed in endocardium by Tie2Cre did not cause defects in the muscular septae or in myocardial growth (B.Z. and W.T.P., unpublished data). The membranous ventricular septum also constricts normally in these embryos; however, fetal death due to hematologic defects before the time that closure of the membranous septum is normally completed prevented us from fully excluding a role of endocardial PRC2 in formation of the membranous ventricular septum. Together, the data strongly support the model that earlier PRC2 inactivation in cardiomyocytes by Nkx2–5Cre than by TNT-Cre causes septal defects during later heart development.
Interestingly, abnormalities in gene expression and chromatin landscape initiated in cardiac progenitor cells as a result of EZH2 loss of function were sustained in cardiomyocytes despite the fact that EZH2 was no longer essential in cardiomyocytes. Later inactivation of EZH2 by TNT-Cre either did not change the expression of these genes or changed them to a lesser degree. This observation likely reflects a form of “epigenetic memory,” in which features of the chromatin landscape established under the direction of transient developmental cues are maintained in the absence of those cues. For PRC2-established H3K27me3 marks, this appears to be implemented at a molecular level by mechanisms that recruit PRC2 to preexisting H3K27me3 marks, which leads to maintenance of these marks and their propagation after cell division.35,36 Thus, transient loss of PRC2 activity in Ezh2NK mutant cells would disrupt maintenance of existing H3K27me3 marks and establishment of new marks. Partial restoration of PRC2 activity as a result of Ezh1 expression would then be incapable of reestablishing the proper H3K27me3 modifications. This may account for the severe global decrease of H3K27me3 and abnormalities of gene expression in E12.5 Ezh2NK cardiomyocytes, at a stage when the Ezh2TNT mutants indicate there is sufficient PRC2 activity to sustain normal heart development (Online Figure II). Thus, the present results show that transient insults to the chromatin landscape early in embryo development can have delayed and sustained effects on cardiac morphogenesis and gene regulation. This result has strong implications for understanding the pathogenesis of congenital heart disease and may provide new insights into how transient environmental exposures can disrupt organogenesis.
Sources of Funding
This work was supported by funding from the National Institutes of Health (W.T.P., U01HL098166 and R01HL095712) and the American Heart Association (A.H., postdoctoral fellowship 10POST4290028), and by charitable donations from Edward Marram, Karen Carpenter, and Gail Federici Smith.
In October 2011, the average time from submission to first decision for all original research papers submitted to Circulation Research was 15 days.
This manuscript was sent to Elizabeth McNally, Consulting Editor, for review by expert referees, editorial decision, and final disposition.
Non-standard Abbreviations and Acronyms
- quantitative chromatin immunoprecipitation
- chromatin immunoprecipitation followed by next-generation sequencing
- mutant Eed ::Tg(TNT-Cre) genotype
- mutant Ezh2fl/fl ::Nkx2–5Cre/+ genotype
- mutant Ezh2fl/fl ::Tg(TNT-Cre) genotype
- histone H3 trimethylated on lysine 27
- histone H3 phosphorylated on serine 10, a marker of M phase of the cell cycle
- quantitative reverse-transcription–polymerase chain reaction
- RNA analysis by next-generation sequencing
- Received July 8, 2011.
- Revision received November 24, 2011.
- Accepted November 30, 2011.
- © 2011 American Heart Association, Inc.
- Han P,
- Hang CT,
- Yang J,
- Chang CP
- Kuzmichev A,
- Nishioka K,
- Erdjument-Bromage H,
- Tempst P,
- Reinberg D
- Pasini D,
- Bracken AP,
- Hansen JB,
- Capillo M,
- Helin K
- Boyer LA,
- Plath K,
- Zeitlinger J,
- Brambrink T,
- Medeiros LA,
- Lee TI,
- Levine SS,
- Wernig M,
- Tajonar A,
- Ray MK,
- Bell GW,
- Otte AP,
- Vidal M,
- Gifford DK,
- Young RA,
- Jaenisch R
- Cao R,
- Wang L,
- Wang H,
- Xia L,
- Erdjument-Bromage H,
- Tempst P,
- Jones RS,
- Zhang Y
- Tanaka M,
- Chen Z,
- Bartunkova S,
- Yamasaki N,
- Izumo S
- Jiao K,
- Langworthy M,
- Batts L,
- Brown CB,
- Moses HL,
- Baldwin HS
- Christodoulou DC,
- Gorham JM,
- Herman DS,
- Seidman JG
- Robinson MD,
- Smyth GK
- He A,
- Kong SW,
- Ma Q,
- Pu WT
- Blahnik KR,
- Dou L,
- O'Geen H,
- McPhillips T,
- Xu X,
- Cao AR,
- Iyengar S,
- Nicolet CM,
- Ludascher B,
- Korf I,
- Farnham PJ
- Bracken AP,
- Kleine-Kohlbrecher D,
- Dietrich N,
- Pasini D,
- Gargiulo G,
- Beekman C,
- Theilgaard-Monch K,
- Minucci S,
- Porse BT,
- Marine JC,
- Hansen KH,
- Helin K
- Pashmforoush M,
- Lu JT,
- Chen H,
- Amand TS,
- Kondo R,
- Pradervand S,
- Evans SM,
- Clark B,
- Feramisco JR,
- Giles W,
- Ho SY,
- Benson DW,
- Silberbach M,
- Shou W,
- Chien KR
- Chen H,
- Shi S,
- Acosta L,
- Li W,
- Lu J,
- Bao S,
- Chen Z,
- Yang Z,
- Schneider MD,
- Chien KR,
- Conway SJ,
- Yoder MC,
- Haneline LS,
- Franco D,
- Shou W
- Ezhkova E,
- Lien WH,
- Stokes N,
- Pasolli HA,
- Silva JM,
- Fuchs E
Novelty and Significance
What Is Known?
Development of the heart requires tightly choreographed changes in gene transcription.
Epigenetic histone modifications are essential regulators of gene transcription.
Polycomb repressive complex 2 (PRC2) catalyzes trimethylation of histone H3 lysine 27 (H3K27me3), a major epigenetic repressive mark.
What New Information Does This Article Contribute?
PRC2 is essential for normal heart development.
PRC2 represses noncardiac transcriptional programs and contributes to normal spatiotemporal regulation of cardiac gene expression.
Perturbation of the epigenetic landscape early in cardiac morphogenesis has sustained disruptive effects at later developmental stages.
Congenital heart disease is one of the most frequent birth defects, and transcription factors are prominent among congenital heart disease genes. Covalent modification of histones is being increasingly recognized as a critical transcriptional regulatory mechanism, but little is known about the role of covalent histone modifications in heart development. In this article, we report on cardiac inactivation of 2 essential subunits of PRC2, the only enzyme known to establish repressive H3K27me3 marks. Our study shows that PRC2 is required for normal heart development. PRC2 was required to repress noncardiac transcriptional programs, as well as potent cell cycle inhibitors and spatiotemporally regulated cardiac genes. Early Nkx2–5Cre–mediated inactivation of EZH2, the PRC2 catalytic subunit, caused perinatal lethal heart malformations. Interestingly, later TNT-Cre–mediated inactivation of EZH2 was compatible with normal survival. This was likely because of functional redundancy with EZH1, because TNT-Cre–mediated inactivation of the nonredundant PRC2 subunit Eed also caused lethal heart malformations. Thus, our results demonstrate an essential role for PRC2 and epigenetic transcriptional regulation in heart development. Our data suggest that perturbation of the epigenetic landscape early in cardiogenesis has sustained, disruptive effects at later stages. These results provide new insights into the transcriptional regulation of heart development and the pathogenesis of congenital heart disease.