Deciphering the Epigenetic Code of Cardiac Myocyte TranscriptionNovelty and Significance
Rationale: Epigenetic mechanisms are crucial for cell identity and transcriptional control. The heart consists of different cell types, including cardiac myocytes, endothelial cells, fibroblasts, and others. Therefore, cell type–specific analysis is needed to gain mechanistic insight into the regulation of gene expression in cardiac myocytes. Although cytosolic mRNA represents steady-state levels, nuclear mRNA more closely reflects transcriptional activity. To unravel epigenetic mechanisms of transcriptional control, cell type–specific analysis of nuclear mRNA and epigenetic modifications is crucial.
Objective: The aim was to purify cardiac myocyte nuclei from hearts of different species by magnetic- or fluorescent-assisted sorting and to determine the nuclear and cellular RNA expression profiles and epigenetic marks in a cardiac myocyte–specific manner.
Methods and Results: Frozen cardiac tissue samples were used to isolate cardiac myocyte nuclei. High sorting purity was confirmed for cardiac myocyte nuclei isolated from mice, rats, and humans. Deep sequencing of nuclear RNA revealed a major fraction of nascent, unspliced RNA in contrast to results obtained from purified cardiac myocytes. Cardiac myocyte nuclear and cellular RNA expression profiles showed differences, especially for metabolic genes. Genome-wide maps of the transcriptional elongation mark H3K36me3 were generated by chromatin-immunoprecipitation. Transcriptome and epigenetic data confirmed the high degree of cardiac myocyte–specificity of our protocol. An integrative analysis of nuclear mRNA and histone mark occurrence indicated a major impact of the chromatin state on transcriptional activity in cardiac myocytes.
Conclusions: This study establishes cardiac myocyte–specific sorting of nuclei as a universal method to investigate epigenetic and transcriptional processes in cardiac myocytes of different origins. These data sets provide novel insight into cardiac myocyte transcription.
Precise spatiotemporal regulation of gene expression in cardiac myocytes is essential to meet the requirements during cardiac development, as well as under pathophysiological conditions of the heart.1 This requires adaptation of chromatin density by epigenetic processes like DNA methylation, histone modifications, and ATP-dependent chromatin remodeling complexes.2 Trimethylation of H3K4, as well as acetylation of H3K27, at promoter regions are linked to transcriptional activation, whereas DNA methylation and trimethylation of H3K27 at promoters and in gene bodies are associated with gene repression.3 In addition, trimethylation of H3K36 is enriched at gene bodies of expressed genes linked to cotranscriptional splicing.4 Cellular steady-state mRNA levels are affected by post-transcriptional splicing, polyadenylation, or mRNA degradation.5 All of these processes control the amount of mRNA available for protein translation. Therefore, analysis of nuclear unprocessed RNA is advisable to get direct insight into transcriptional activity6 and its correlation with chromatin states. Most epigenetic studies in the heart focused on the analysis of cardiac tissue. Because cardiac myocyte nuclei represent only one third of cardiac nuclei in adult hearts and cell type composition changes during development of the heart and in disease,7 cardiac myocyte–specific maps of epigenetic modifications and nuclear transcriptome data are needed. Cardiac myocytes can be isolated by enzymatic digestion of cardiac tissue at physiological temperatures.8 As this might alter chromatin states and transcriptional activity, we developed protocols that do not involve enzymatic steps and are performed under cold conditions to maintain the in vivo situation.
Editorial, see p 392
In This Issue, see p 389
In the present study, we demonstrate purification of cardiac myocyte nuclei from several species, including human. This enabled us to generate cardiac myocyte–specific genome-wide maps of several histone modifications and nuclear transcriptomes. The high proportion of detected nascent unspliced RNA allows novel insight into transcriptional activity of cardiac myocytes. The nuclear transcript abundancies greatly differ from cellular levels in cardiac myocytes. In addition, gene transcription could be predicted by a model incorporating data of several key histone modifications. Our results open new possibilities to analyze epigenetic signatures and transcriptional activity of cardiac myocytes in development and disease.
All animal procedures were approved by the responsible animal care committee (Regierungspräsidium Freiburg, Germany and Landesamt für Natur, Umwelt und Verbraucherschutz, Recklinghausen, Germany); these conformed to the Guide for the Care and Use of Laboratory Animals published by the National Academy of Sciences 2011. Twelve-week-old male C57BL/6 N mice, 12-week-old female Myh6-H2B-mCherry transgenic mice9 and Wistar rats were used. Experimental infarction by ligation of the left anterior descending coronary artery was performed as described.9
Human Cardiac Biopsies
Left ventricular biopsies from hearts were used for cardiac myocyte nuclei isolation. These investigations were approved by the ethics committee of the University of Freiburg.
Cryosections were stained with an antibody against pericentriolar material 1 (PCM1)10 and a labeled secondary antirabbit antibody. Griffonia simplicifolia lectin I staining was used to visualize the glycocalyx of endothelial cells. Nuclei were stained with DAPI (4′,6-diamidino-2-phenylindole) and Hoechst 33342 (Life technologies).
Isolation of Cardiac Nuclei
Cardiac nuclei were isolated as described.7 For the isolation of nuclear RNA, RNAsin (80 U/mL, Promega) was added to all buffers.
Magnetic-Assisted Sorting of Cardiac Myocyte Nuclei
Flow Cytometric Analysis and Sorting of Cardiac Myocyte Nuclei
Cardiac myocyte nuclei were identified by an anti-PCM1 antibody. Nuclei were analyzed (CyFlow Space, Partec) or sorted (Bio-Rad S3, Bio-Rad) by flow cytometry.
ChIP-seq was performed as described previously with modifications.7,11 Chromatin was precipitated with an anti-H3K36me3 antibody (1 μg, ab9050, Abcam) overnight. Data were deposited to the National Center for Biotechnology Information BioSample database (SRP033385).
Gene Expression Analysis
Total RNA was isolated from cardiac myocyte nuclei or intact cardiac myocytes7 derived from individual mouse hearts using RNeasy Micro Kit (Qiagen). RNA-seq libraries were generated from ribosomal depleted RNA and were sequenced on a HiSeq 2500 (50 bp, Illumina). Data were deposited to the National Center for Biotechnology Information BioSample database (SRP033386).
Sequencing reads were mapped to the Mus musculus genome (mm9) and duplicate reads were removed. Transcript abundance was calculated as fragments per kilobase of exon per million mapped reads (FPKM) and enrichment of histone modifications determined as reads per kilobase per million mapped reads. The major isoform of each gene was used for all analysis steps. Quantified ChIP-seq reads were used to predict transcript levels. Reads per kilobase per million mapped reads values of different histones were combined to increase the goodness of fit. Genomic regions enriched for histone modifications were identified by magnetic-assisted nuclei sorting 2.12
External Data Sets
Previously published MethylC-seq data and ChIP-seq data (H3K27ac, H3K4me3, H3K4me1, and H3K27me3) from adult mouse cardiac myocyte nuclei were analyzed as described.7 ChIP-seq data, MethylC-seq data, and RNA-seq data were reanalyzed.13–16
Gene Ontology Analysis
Enriched GO terms of the category biological process were identified by ClueGO.17
We developed a protocol for the isolation of highly pure cardiac myocyte nuclei from small tissue biopsies (Figure 1). In contrast to traditional approaches involving enzymatic tissue dissociation to obtain cell suspensions,8 the described procedures do not involve enzymatic steps and are performed under cold conditions to maintain the in vivo epigenetic state. Furthermore, our protocol enables characterization of frozen cardiac samples from experimental animal models, as well as cardiac tissue biopsies from patients. Sorting relies either on fluorescence-assisted nuclei sorting (FACS) or magnetic-assisted nuclei sorting. As we have previously shown, these sorted nuclei were suitable for epigenetic analysis.7 These procedures were optimized for genome-wide analysis of epigenetic marks in cardiac myocyte nuclei derived from small amounts of tissue (Online Table I). Furthermore, we established RNA sequencing of nuclear mRNA (nucRNA) to gain insight into transcriptional activity. To assess cellular steady-state RNA levels, we performed RNA sequencing from intact cardiac myocytes purified by FACS (Figure 1).
Staining of Cardiac Myocyte Nuclei
The basis of the presented method is the detection of PCM1, a protein which in the heart is specifically bound to cardiac myocyte nuclei.7,10 To further prove the specificity of the peri-nuclear PCM1 staining in cardiac myocytes, we took advantage of a transgenic mouse model, in which a fusion protein of human histone H2B and mCherry (H2B-mCh) under control of the cardiac myocyte–specific Myh6 promoter labels exclusively all cardiac myocyte nuclei.9 We stained cryosections of left ventricles from these transgenic mice with an antibody against PCM1 (Figure 2A). Analysis of 10 167 nuclei (n=2 hearts) revealed a high overlap of PCM1 and H2B-mCh signals (Figure 2A). PCM1 staining had a high specificity (>99.9%, ie, only 1 false-positive PCM1+ nucleus out of 6499 H2B-mCh− nuclei) and sensitivity (99.9%, ie, only 5 false-negative PCM1− nuclei out of 3667 Hb2-mCh+ nuclei; Figure 2A). We also explored the cardiac myocyte–specific expression pattern in the scar area of infarcted hearts from H2B-mCh transgenic mice. Analysis of 4392 nuclei showed for 99.5% of the nuclei PCM1 and H2B-mCh signals (Figure 2B). Only negligible fractions of nuclei single positive for PCM1 (0.21%±0.1% of nuclei) or for H2B-mCh (0.34%±0.2%) were found (Figure 2B). Furthermore, we analyzed PCM1 staining in endothelial and smooth muscle cells. The glycocalyx of endothelial membranes was visualized by fluorescently labeled Griffonia simplicifolia lectin I18 (Figure 2A and 2B), whereas smooth muscle cells were visualized by α-smooth muscle actin (ACTA2) staining (Online Figure II). Analysis of 309 α-smooth muscle actin–positive cells did not reveal a single cell with perinuclear PCM1 staining (Online Figure IIA and IIB). In addition, endothelial cells as identified by presence of Griffonia simplicifolia lectin I and absence of H2B-mCh signals did neither show PCM1 staining in native hearts nor in hearts after myocardial infarction (Online Figure IIC and IID). Three out of 1658 Griffonia simplicifolia lectin I–positive cells were both H2B-mCh-positive and PCM1-positive and thus could not be classified as cardiac myocytes or endothelial cells (Online Figure IIC and IID). We also tested PCM1 specificity in single cardiac cells after Langendorff cell isolation (Figure 2C). Similarly, we did not detect PCM1 staining in nonmyocytes (H2B-mCh-negative and nonrod-shaped, 0 out of 173; Figure 2C). All rod-shaped cardiac myocytes were PCM1- and H2B-mCh-positive. Some nonrod-shaped cardiac myocytes (H2B-mCh-positive) showed loss of the PCM1 signal, possibly because of cell death (Figure 2C).
Analysis and Sorting of Cardiac Myocyte Nuclei
Flow cytometry experiments showed that PCM1 staining also labeled isolated nuclei (Figure 2D–2F). To evaluate whether the protocol was applicable to different species, we also analyzed human and rat hearts in addition to mouse hearts. In all 3 species, PCM1 staining of cardiac myocyte nuclei was evident (Figure 3A). Remarkably, the proportion of PCM1-positive nuclei detected in frozen and fresh hearts was 99%±1.1% (mean±SEM, n=8) identical. Purification of cardiac myocyte nuclei by magnetic-assisted nuclei sorting was highly reproducible as assessed by flow cytometry (>94% purity; Figure 3B). These results highlight that PCM1 enables isolation of cardiac myocyte nuclei from hearts of different species, including frozen clinical samples.
Nuclear RNA-Seq of Cardiac Myocytes
We chose mouse cardiac myocyte nuclei to generate transcriptomes by paired-end sequencing using a random priming strategy. From 2 independent experiments, we generated sequencing libraries with mean insert sizes of 253 and 343 bp. We obtained 80 million uniquely mapped reads. For comparison, we assessed gene expression in intact cardiac myocytes by RNA-seq (cellular RNA-seq). Langendorff digestion was used to isolate cardiac cells. The obtained cell suspension contained 19%±4% nonmyocytes (n=4, mean±SEM; Online Figure IA). Therefore, we further purified viable cardiac myocytes by FACS (Online Figure IB) and generated a library with a mean insert size of 312 bp and obtained 38 million uniquely mapped RNA-seq reads.
In contrast to cellular RNA-seq, nuclear RNA-seq showed high read coverage of intronic regions as illustrated in Figure 4A for the representative gene, phospholamban (Pln; Figure 4A). Reads of cellular RNA-seq mapped predominantly into exonic regions of Pln (Figure 4A). However, nuclear RNA-seq data showed a large fraction of unspliced nascent Pln transcripts (Figure 4A). Based on intronic reads, we calculated the fraction of unspliced mRNAs for all genes with high read coverage (>10 FPKM) and exon–exon junctions. This global analysis revealed that 60.8% of nuclear mRNAs were unspliced. In intact cardiac myocytes, the unspliced fraction was only 2.6% (Figure 4B). These findings confirmed our hypothesis that nuclear mRNA analysis is less affected by posttranscriptional modifications and thus enables insight into transcriptional activity. Remarkably, nuclear RNA-seq showed a >4-fold higher fraction of reads uniquely mapping to noncoding RNA transcripts as compared with cellular RNA-seq (Figure 4C; Online Table II).
A drawback of nuclear RNA preparation is the likelihood of genomic DNA contamination. To quantify remaining DNA contamination, we included 2 megabases of random sequences from the repeat-masked noncoding part of the genome in our bioinformatics analysis. In average, we detected 0.062 FPKM (0.049–0.074 FPKM, 95% confidence interval) in these regions. This value was regarded as the lower detection limit of our nuclear RNA-seq. 12 653 of 20 174 annotated coding genes showed transcriptional rates above this background (Figure 4D). Genes with the highest nuclear transcript levels were the cardiac myocyte genes troponin T type 2, α-myosin heavy chain, titin, and myosin light-chain 2.
Next we asked whether nuclear and cellular mRNA profiles differ in cardiac myocytes. By visual inspection, we identified natriuretic peptide A (Nppa) mRNA predominantly in the cellular RNA-seq (Figure 5A), whereas myosin heavy chain 7b (Myh7b) mRNA was mainly detected in nuclei (Figure 5B). To identify transcripts with the most differential abundance, we calculated log2 ratios of FPKM values between cellular and nuclear mRNAs for transcripts with expression >1 FPKM in cardiac myocyte nuclei (Figure 5C). Cellular to nuclear RNA expression ratios differed over several orders of magnitude (Figure 5C). Gene ontology analysis of RNAs that were primarily expressed in cells as compared with nuclei (>95th percentile; Figure 5C) showed significant enrichment for genes involved in energy metabolism (Figure 5D). On the other hand, mRNAs with higher abundance in nuclei (<5th percentile; Figure 5C) showed an overrepresentation of genes associated with transcriptional regulation (Figure 5E).
Cell-Type Specificity of Epigenetic and Transcriptome Data Obtained From PCM1-Sorted Nuclei
The heart consists of many different cell types, including cardiac myocytes, fibroblasts, endothelial cells, smooth muscle cells, and others.19 Analysis of cardiac tissue thus shows the overlay of epigenetic and transcriptome signatures of different cardiac cell-types. We postulated that data obtained from PCM1–sorted nuclei were highly cardiac myocyte–specific. We compared our data from PCM1-positive cardiac myocyte nuclei with previously published data sets from cardiac tissue biopsies (Figure 5A–5D; Online Figure III).14–16 Figure 5 shows the results of the representative cardiac myocyte–specific gene myosin light chain 3 (Myl3) and the mesenchymal nonmyocyte gene vimentin (Vim).20 In cardiac myocyte nuclei, the promoter region of Myl3 was enriched for acetylation of lysine 27 of histone 3 (H3K27ac; Figure 6A). Acetylation of H3K27 is an epigenetic hallmark of active genomic regions.21 In contrast, no H3K27ac signal could be detected in the promoter region of Vim in cardiac myocyte nuclei (Figure 6B). On the other hand, the Vim promoter was enriched for trimethylated H3K27, a mark that is considered to be an indicator of gene silencing22 (Figure 6B). This repressive mark was absent in the promoter region of Myl3 in cardiac myocyte nuclei (Figure 6A). These results were in line with RNA-seq data revealing that Myl3 but not Vim was expressed in cardiac myocyte nuclei (Figure 6A and 6B). Analysis of heart tissue showed gene expression of both genes and a mixture of active and inactive histone marks (Figure 6C and 6D). In addition, CpG methylation profiles of cardiac myocytes showed a clear topology as compared with profiles obtained from cardiac tissue biopsies (Figure 6A–6D). Previously reported cardiac myocyte–specific demethylation of the genic region of Myl37 was only visible in cardiac myocyte nuclei but not in total heart samples (Figure 6A and 6C). The obtained results were further validated by analysis of the endothelial cell gene angiotensin-converting enzyme (Ace)23 and the fibroblast gene collagen α-2(I) chain (Col1a224; Online Figure III). We compared our ChIP-seq data with data derived from in vitro differentiated cardiac myocytes13 to confirm validity. Similar enrichment profiles of histone modifications were observed for representative cardiac myocyte genes, as well as for averages of genes grouped according to quantiles of expression in cardiac myocyte nuclei (Online Figure IV). Differences were observed regarding nonmyocyte marker genes and histone peaks adjacent to developmental genes (Online Figure IV).
We asked which gene promoters were active according to acetylation of H3K27 in cardiac myocyte nuclei and in cardiac tissue. Therefore, we identified all promoters overlapping with an H3K27ac peak in hearts and in cardiac myocyte nuclei, respectively. We could identify 10 123 active promoters in hearts (Figure 7A and 7C). Nine hundred thirty-six of these lacked H3K27ac in cardiac myocyte nuclei, indicating that they were not active in cardiac myocytes but in other cardiac cell types (Figure 7A and 7C, black section). Gene ontology classification of H3K27ac ChIP-seq peaks detected in heart tissue but not in cardiac myocyte nuclei (Figure 7C) showed association with extracellular matrix organization, vascular development, regulation of immune system processes, and cell migration (P<10−10; Figure 7B). This suggests that these H3K27ac signals originate from cardiac fibroblasts, endothelial cells, or immune cells but not from cardiac myocytes. Analysis of these promoter regions showed enrichment for H3K27me3 in hearts and in cardiac myocyte nuclei (Figure 7D). Promoters active in hearts and cardiac myocytes (Figure 7D; gray box) were depleted for H3K27me3 in cardiac myocytes. In contrast, results obtained from hearts showed enrichment for H3K27me3, probably as a result of silencing of these genes in nonmyocytes (Figure 7D; gray box).
Histone Marks Are Predictive for Transcription in Cardiac Myocytes
We extended the analysis to further histone marks to get a more detailed picture. Previous studies showed that H3K36me3 is tightly linked to active transcription and peaks in gene bodies of active genes.11,25 We performed ChIP-seq experiments for the histone mark H3K36me3 in purified cardiac myocyte nuclei. We generated 22 million uniquely mapped reads. In addition, we reanalyzed our previously published ChIP-seq data and asked which epigenetic marks were most predictive for cardiac myocyte transcription.
The tropomyosin 1 gene (Tpm1) encodes for a protein of the contractile apparatus.26 It showed the typical epigenetic signature of a highly expressed gene (Figure 8). Several isoforms of Tpm1 are annotated (Online Figure V). Previous studies have shown that Tpm1 isoforms are expressed in different cell-types.26 For a proper analysis of the interplay of chromatin features and RNA transcription, it is crucial to identify the expressed isoform(s). We used nuclear RNA-seq reads from the spliced fraction to determine genome-wide exon–exon junctions and selected the main cardiac myocyte isoform of each gene for further analysis steps. In cardiac myocyte nuclei, only one Tpm1 isoform was detected (transcript 8, Online Figure V). This cardiac myocyte isoform of Tpm1 was concordant with epigenetic patterns of the Tpm1 gene. The region around the annotated transcription start site was decorated with H3K27ac. We observed strong enrichment of H3K36me3 in exonic regions of Tpm1 (Figure 7A; Online Figure V). During transcription, RNA polymerase II interacts with SETD2 (SET domain containing 2),27 resulting in trimethylation of H3K36. Previous studies linked exon marking by H3K36me3 to splicing activity during active transcription.4,28 In contrast, H3K4me3 is enriched at active promoters and promoters with high CpG density.29 For Tpm1, 2 different transcription start sites are annotated (Online Figure V). Both overlap with regions of high CpG density (CpG islands) and thus were marked by H3K4me3 (Online Figure IV). Only the cardiac myocyte–specific transcription start region was enriched for H3K27ac, too (Figure 8A; Online Figure V). The H3K4me3 mark was strongly enriched at the 5′ region of the Tpm1 gene, whereas the remainder of the gene body was decorated by H3K4me1 (Figure 8A; Online Figure V). This showed the mutually exclusive nature of H3K4me1 and H3K4me327 in purified nuclei. The repressive histone mark H3K27me3 was absent at the entire Tpm1 locus (Figure 8A; Online Figure V). In a previous study, we showed that low levels of gene body CpG methylation were characteristic for highly expressed cardiac myocyte genes.7 In agreement with these results, large domains of the gene body of Tpm1 showed strong demethylation of CpGs with lowest levels of DNA methylation in the 5′-region (Online Figure V). These data demonstrate that specific histone patterns mark expression of gene isoforms in cardiac myocytes.
To correlate transcript levels with histone marks, the genomic regions decorated by specific histone marks in cardiac myocyte nuclei were identified. Coding genes showed a clear pattern with marking of promoters of expressed genes by H3K27ac and H3K4me3 and of gene bodies by H3K4me1 and H3K36me3 (Online Figure VI). Furthermore, genes lacking these marks were marked by H3K27me3 in genic regions (Online Figure VI). These genomic locations were selected for quantitative analysis.
To test whether histone marks are predictive for nuclear mRNA expression, we ranked all coding genes according to expression level. Genes not expressed in cardiac myocyte nuclei (<0.062 FPKM, gene expression rank >12,653) showed high H3K27me3 enrichment (Figure 8B). These genes were depleted for the active marks H3K4me3, H3K4me1, H3K36me3, and H3K27ac. Expressed genes gained active histone marks concordant with expression levels and lost H3K27me3 (Figure 8B).
These data prompted us to analyze the predictive information of the individual marks for transcriptional activity. Linear modeling of transcription from individual histone marks showed a significant positive correlation with measured nucRNA levels in case of H3K4me1, H3K4me3, H3K27ac, and H3K36me3 (P<0.001; Figure 8C). A negative correlation with gene expression was observed for H3K27me3 (P<0.001). H3K27ac contained the highest predictive information for a single histone mark. 70% of nucRNA expression differences could be explained by enrichment of H3K27ac at promoter regions (Figure 8C). The predictive power of the model was further improved to 78% by addition of results from assessed histone marks (Figure 8C and 8D). A scatter plot of predicted and measured nuclear RNA showed a large overlap for expressed and not-expressed genes. A linear correlation of gene expression values obtained from this prediction with cellular and heart RNA-seq data (Online Figure VII) yielded 74% and 71% (Online Figure VII), respectively. Remarkably, the prediction was especially weaker in case of expressed genes. This was even more evident in case of noncardiac data sets,15 further reducing the strength of correlation to 39% to 42% (Online Figure VII).
We provide detailed methods for the in-depth analysis of epigenetic modifications in cardiac myocytes. Furthermore, we present—for the first time—a technique to study nascent RNA transcripts of cardiac myocytes. Fresh and frozen cardiac tissue from different species, including human, can be purified using our workflow.
To date, most epigenetic and transcriptome studies in cardiovascular research have analyzed cardiac tissue biopsies.30 Epigenetic and gene expression signatures are highly cell type–specific.31 Therefore, analysis of tissues consisting of complex cell type mixtures may lead to ambiguous results. Indeed, changes in cellular composition itself may lead to varying results.31 This point should not be neglected in heart research because several conditions are known to affect the cellular composition of the heart. For example, pathophysiological conditions like cardiac pressure overload or myocardial infarction induce cellular changes as part of cardiac remodeling.32,33 These changes involve apoptosis of cardiac myocytes, proliferation of fibroblasts and endothelial cells, as well as invasion of inflammatory cells.
The importance to consider these alterations is highlighted by a recent manuscript reanalyzing previously published data sets of age-associated epigenetic changes in the blood. Most of the observed variability could be explained simply by cellular composition.34 Our recent findings support the significance of altered cellular composition for cardiac epigenetic studies.7 Significant differences of CpG methylation levels in biopsies of hypertrophic and healthy hearts were caused by disease-associated loss of cardiac myocytes and increased abundance of fibroblasts and other nonmyocyte cells.7 Analysis of complex tissues like hearts may also lead to misinterpretation of chromatin signatures. Coexistence of repressive and active marks on the same allele has been observed in cell types with restricted differentiation potency or pluripotency.35 Remarkably, this bivalency could not be observed in our data sets of adult cardiac myocytes. ChIP-seq data from hearts resemble the picture of bivalency, but it is likely that the signals of the active and inactive marks originate from different cell-types.
These facts underscore the need for cell type–specific studies. To meet this demand, we developed methods for the epigenetic and transcriptome analysis of purified cardiac myocyte nuclei. We and others reported PCM1 as a marker of cardiac myocyte nuclei in different species.7,10 In the present article, different immunohistochemistry quantification strategies in healthy, transgenic, and diseased hearts show that >98% of cardiac myocyte nuclei are positive for PCM1. These results are in good agreement with our previously reported FACS data.7 Furthermore, the specificity of PCM1 has recently been reported for human cardiac myocytes.36 Our whole genome data of nuclear mRNA expression and epigenetic marks further support the validity of PCM1 as a cardiac myocyte–specific marker.
In contrast to traditional approaches involving enzymatic tissue dissociation to obtain cell suspensions,8 the described procedures do not involve enzymatic steps and are performed under cold conditions to maintain the in vivo epigenetic state. Our method is applicable to different developmental and disease stages of several species, including frozen clinical samples without protocol modifications.
The initial step of the workflow is disruption of cardiac biopsies and purification of nuclei. The described tissue disruption and purification procedures ensure efficient and reliable isolation of nuclei from frozen and fresh cardiac biopsies. Isolated nuclei can be sorted by FACS. In addition, we established magnetic-assisted sorting of antibody-labeled nuclei, which enables processing of samples in parallel. Both methods yielded highly pure preparations.
To date, several studies performed deep sequencing of the cardiac transcriptome.37,38 In the present article, we studied for the first time the nuclear transcriptome of cardiac myocytes. In comparison to cellular RNA, nucRNA represents freshly transcribed (nascent) mRNA and therefore correlates more closely with transcriptional activity. The validity of this assumption was proven by the large fraction of detected unspliced mRNAs and is in good agreement with previous reports assessing nucRNA in other cell-types.6,39 Our data sets enabled us for the first time to identify the entire set of genes transcribed in cardiac myocytes. Epigenetic data confirmed these results because transcribed genes showed an epigenetic signature distinct from nonexpressed genes.
After transcription, several mechanisms modulate mRNA abundance. Comparing nuclear and cellular mRNA levels in cardiac myocytes revealed distinct patterns of expression. Remarkably, a large fraction of metabolic genes seems to accumulate in the cytosol. Several mechanisms could account for these differences, including post-transcriptional processing, mRNA decay, and miRNA-mediated degradation.5,11,40
Next, we proposed a model to deduce quantitatively nuclear transcript levels from epigenetic data. The highest informative value among the assessed histone modifications had H3K27ac. To improve the accuracy of the model, we combined the results of 5 key histone marks. This model explained 78% of gene expression alteration by epigenetic marks. Comparable results have been obtained in different cell lines.11,41,42 A cross-comparison of the predictive model of cardiac myocyte gene expression with measured RNA levels in noncardiac tissues showed a weak correlation (Online Figure VII). This underscores that the chromatin state of promoter and genic regions is a major determinant of cell-type–specific transcriptional activity. Factors explaining the remaining transcriptional variability include varying activity of enhancers43 and transcription factors.44 This tight link of chromatin state and transcriptional activity can only be uncovered in cell-type–specific data sets.
Nuclei purified with the proposed protocol are principally suitable to study other nuclear processes in cardiac myocytes. The proposed protocol opens new possibilities to study gene transcription and epigenetic mechanisms in cardiac myocytes from different species and developmental stages, including clinical samples. This will be of great value to pave the way to a better understanding of cardiac myocyte transcription in development and disease.
We thank the Deep Sequencing Facility, MPI of Immunobiology and Epigenetics (Freiburg, Germany) for sequencing. R. Gilsbach, L. Hein, and S. Preissl conceived the study, R. Gilsbach and S. Preissl designed, performed, and analyzed experiments, A. Raulf, B.K. Fleischmann, M. Hesse, C. Köbele, and M. Schwaderer performed experiments and data analysis, B.A. Grüning and R. Backofen provided bioinformatics tools, R. Gilsbach and S. Preissl performed computational analysis. B.K. Fleischmann edited the article. R. Gilsbach, L. Hein, and S. Preissl wrote the article.
Sources of Funding
This study was supported by the Deutsche Forschungsgemeinschaft SFB 992 project B03 (L. Hein) and FOR 1352 (B.K. Fleischmann) and the BIOSS Centre for Biological Signalling Studies (L. Hein).
In May 2015, the average time from submission to first decision for all original research papers submitted to Circulation Research was 15.49 days.
The online-only Data Supplement is available with this article at http://circres.ahajournals.org/lookup/suppl/doi:10.1161/CIRCRESAHA.115.306337/-/DC1.
- Nonstandard Abbreviations and Acronyms
- fluorescence-assisted nuclei sorting
- fragments per kilobase of exon per million mapped reads
- nuclear mRNA
- pericentriolar material 1
- Received February 27, 2015.
- Revision received June 22, 2015.
- Accepted June 23, 2015.
- © 2015 American Heart Association, Inc.
- de Almeida SF,
- Grosso AR,
- Koch F,
- Fenouil R,
- Carvalho S,
- Andrade J,
- Levezinho H,
- Gut M,
- Eick D,
- Gut I,
- Andrau JC,
- Ferrier P,
- Carmo-Fonseca M
- Gilsbach R,
- Preissl S,
- Grüning BA,
- Schnick T,
- Burger L,
- Benes V,
- Würch A,
- Bönisch U,
- Günther S,
- Backofen R,
- Fleischmann BK,
- Schübeler D,
- Hein L
- Raulf A,
- Horder H,
- Tarnawski L,
- Geisen C,
- Ottersbach A,
- Röll W,
- Jovinge S,
- Fleischmann BK,
- Hesse M
- Bindea G,
- Mlecnik B,
- Hackl H,
- Charoentong P,
- Tosolini M,
- Kirilovsky A,
- Fridman WH,
- Pagès F,
- Trajanoski Z,
- Galon J
- Di Salvo TG,
- Haldar SM
- Michels KB,
- Binder AM,
- Dedeurwaerder S,
- Epstein CB,
- Greally JM,
- Gut I,
- Houseman EA,
- Izzi B,
- Kelsey KT,
- Meissner A,
- Milosavljevic A,
- Siegmund KD,
- Bock C,
- Irizarry RA
- Kehat I,
- Molkentin JD
- Voigt P,
- Tee WW,
- Reinberg D
- Lee JH,
- Gao C,
- Peng G,
- Greer C,
- Ren S,
- Wang Y,
- Xiao X
- Yang KC,
- Yamada KA,
- Patel AY,
- Topkara VK,
- George I,
- Cheema FH,
- Ewald GA,
- Mann DL,
- Nerbonne JM
- Karlić R,
- Chung HR,
- Lasserre J,
- Vlahovicek K,
- Vingron M
- Paige SL,
- Plonowska K,
- Xu A,
- Wu SM
Novelty and Significance
What Is Known?
The cellular composition of the heart is changing during development and disease.
Control of gene expression is essential for proper function of cardiac myocytes.
Epigenetic control of transcription is highly cell-type specific.
What New Information Does This Article Contribute?
This study provides a method for purification of cardiac myocyte nuclei from frozen heart tissue of several species.
We have identified and quantified mRNAs transcribed in cardiac myocytes on the nuclear and cellular level.
Chromatin state is a main predictor of transcriptional activity in cardiac myocytes.
Regulation of gene expression involves several processes, including epigenetic control of transcriptional activity. However, the precise molecular mechanisms in cardiac myocytes in vivo are unknown. Thus, the aim of the present study was to establish epigenetic and transcriptomic profiling of purified cardiac myocyte nuclei from heart tissue. Cardiac myocyte nuclei were identified on staining with an antibody against pericentriolar material 1. Extensive validation demonstrated high degrees of specificity and sensitivity of pericentriolar material 1 staining for cardiac myocyte nuclei. Using this technique, we identified RNA transcripts expressed in cardiac myocyte nuclei in vivo. A quantitative analysis of 5 histone marks was performed to define the chromatin state of cardiac myocyte genes. Chromatin status correlated significantly better with nuclear RNA than cellular RNA profiles. This integrative analysis indicated a major impact of the chromatin state on transcriptional activity in cardiac myocytes. Thus, cardiac myocyte–specific sorting of nuclei is a reliable method to investigate epigenetic and transcriptional processes in cardiac myocytes. The findings may open new mechanistic insight into the interplay of epigenetic mechanisms and transcription in cardiac myocytes during development and under pathophysiological conditions.