Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2018 Feb 27.
Published in final edited form as: Dev Cell. 2017 Feb 27;40(4):392–404.e5. doi: 10.1016/j.devcel.2017.01.013

Resolving Heart Regeneration by Replacement Histone Profiling

Joseph Aaron Goldman 1,2,5, Guray Kuzu 3,5, Nutishia Lee 1,2, Jaclyn Karasik 1,2, Matthew Gemberling 1,2, Matthew J Foglia 1,2, Ravi Karra 2,4, Amy L Dickson 1,2, Fei Sun 1,2, Michael Y Tolstorukov 3,*, Kenneth D Poss 1,2,6,*
PMCID: PMC5367476  NIHMSID: NIHMS848174  PMID: 28245924

SUMMARY

Chromatin regulation is a principal mechanism governing animal development, yet it is unclear to what extent structural changes in chromatin underlie tissue regeneration. Non-mammalian vertebrates like zebrafish activate cardiomyocyte (CM) division after tissue damage to regenerate lost heart muscle. Here, we generated transgenic zebrafish expressing a biotinylatable H3.3 histone variant in CMs and derived cell type-specific profiles of histone replacement. We identified an emerging program of putative enhancers that revise H3.3 occupancy during regeneration, overlaid upon a genome-wide reduction of H3.3 from promoters. In transgenic reporter lines, H3.3-enriched elements directed gene expression in subpopulations of CMs. Other elements increased H3.3 enrichment and displayed enhancer activity in settings of injury- and/or Neuregulin1-elicited CM proliferation. Dozens of consensus sequence motifs containing predicted transcription factor binding sites were enriched in genomic regions with regeneration-responsive H3.3 occupancy. Thus, cell-type specific regulatory programs of tissue regeneration can be revealed by genome-wide H3.3 profiling.

eTOC

Cell type-specific chromatin profiling can shed light on intrinsic genetic programs but such analysis in regenerating tissues has technical challenges. Goldman, Kuzu et al. develop transgenic zebrafish enabling cardiomyocyte-specific histone H3.3 profiling to capture sites of nucleosome turnover. They identify regulatory elements preferential for heart regeneration during the dynamic process.

graphic file with name nihms848174u1.jpg

INTRODUCTION

Teleost fish and salamanders have a profound ability to regenerate new tissue after major damage or loss. For example, unlike adult mammals, zebrafish can regenerate as much as 60% of lost heart muscle cells, leaving little or no scar tissue (Poss et al., 2002; Wang et al., 2011). Whereas CM division was historically considered to be absent in adult mammals, recent studies indicate that CM division underlies the regenerative response of the neonatal mouse heart (Porrello et al., 2011). Moreover, limited CM division can be detected in adult mice subjected to ischemic myocardial injury (Bersell et al., 2009; Senyo et al., 2013). Thus, an emerging belief is that mammals, including humans, retain a latent capacity for heart regeneration that can potentially be unlocked.

Prevailing models of regeneration suggest that adult cells revert to a proliferative state in part through reversal of adult features constructed during development. For CM division to occur in a regenerating heart, contractile programs must give way to kinetochore formation, karyokinesis, and cytokinesis, a process generally referred to as dedifferentiation. Gene regulatory mechanisms underlie establishment of cellular attributes, yet such events enabling heart regeneration are less clear. Several transcription factors have been implicated during heart regeneration and postnatal cardiac growth in zebrafish and mice (Fang et al., 2013; Gupta et al., 2013; Heallen et al., 2011; Jopling et al., 2012; Karra et al., 2015; Mahmoud et al., 2013). Yet, there is no established regulatory hierarchy that merges known factors and mechanisms, and fundamental questions remain as to how regenerative CMs retain or establish competency for division upon injury.

Intrinsic genetic programs, such as those underlying a stage- or tissue-specific phenotype, can be determined by comprehensive chromatin profiling, which can include a genomic map of DNA accessibility to transcription factors (Ho et al., 2014). Genome-wide signatures of open chromatin regions can be derived by a variety of profiling methods, including ChIP-based profiling of the enhancer complex p300 and the mark it deposits, acetylation of lysine 27 on the N-terminal tail of histone H3 (H3K27Ac) (Rada-Iglesias et al., 2011), as well as assays of DNAse hypersensitivity sites (DHS), density of chromatin fibers (FAIRE), and transposase-accessible chromatin (ATACseq) (Boyle et al., 2008; Buenrostro et al., 2013; Giresi et al., 2007). These methods have the potential to systematically detail the genome-wide alterations that enable CM dedifferentiation and proliferation after cardiac injury. The cellular complexity of cardiac tissue is predicted to generate noise if subjected in toto to chromatin analysis. Changes in cell-type composition that occur after injury, like muscle loss and inflammatory cell infiltration, would likely amplify this noise. Cell sorting might control for compositional changes; however, the harsh process of the isolation of cells like CMs itself creates an injury-like state (Sander et al., 2013).

Here, we adapted a transgenic strategy capitalizing on the known specificity of histone H3.3, the replacement histone H3 isoform deposited at sites of nucleosome turnover (Mito et al., 2007). By restricting expression of a tagged H3.3 to CMs in zebrafish, we show it is possible to non-invasively obtain cell type-specific profiles of nucleosome replacement within a complex organ during a dynamic developmental process. Furthermore, we show that H3.3 profiling can be used to find previously unknown cis-regulatory elements, many of which we validate in transgenic reporter assays as bona fide transcriptional enhancers. Cardiac injury and regeneration elicited widespread changes in open chromatin, and from our bioinformatical analyses we validate additional DNA elements that could activate expression specifically in contexts of CM proliferation. Finally, we bioinformatically derived a panel of predicted sequence regulatory motifs that are enriched in regions with high H3.3 occupancy during heart regeneration. Thus, by profiling a marker of nucleosome replacement, we provide a high-resolution resource of gene regulatory changes in CMs during the process of heart regeneration. This approach and resource are generalizable to other cell types, injury contexts, and species.

RESULTS AND DISCUSSION

Transgenic System for Profiling Histone Replacement in CMs

To probe changes in the replacement histone H3.3 during zebrafish heart regeneration, we generated stable transgenic zebrafish with the CM-restricted promoter, cardiac myosin light chain2 (cmlc2), driving expression of the E. coli biotin ligase BirA. Tg(cmlc2:BirA-EGFP-2A-FLAG-BRLP-H3.3)pd185 animals (hereafter referred to as cmlc2:H3.3-bio) co-cistronically express a biotinylateable version of FLAG-tagged histone H3.3 (Figures 1A and 1B). Immunostaining of adult cardiac tissue revealed localization of the tagged H3.3 specifically to CM nuclei (Figures 1C, 1D, and S1A). To isolate H3.3-bio-enriched chromatin from CMs within whole cmlc2:H3.3-bio ventricles, we used streptavidin-based enrichment. We then generated libraries for high-throughput DNA sequencing of H3.3-bio-associated chromatin. As predicted, our profiles indicated extensive H3.3 enrichment within gene promoter regions or transcribed regions (hereafter referred to as gene bodies) and in intergenic sites.

Figure 1. Profiling of H3.3 Occupancy in Adult Zebrafish CMs.

Figure 1

Open in a new tab

(A) The structure of the cmlc2:H3.3-bio transgene. EGFP is fused to a biotin ligase (BirA) followed by an in-frame 2A peptide sequence for polycistronic expression with the zebrafish histone H3.3 gene H3f3d. Fused in frame to H3.3 are sequences coding an N-terminal FLAG and biotin ligase recognition peptide (BLRP).

(B) Schematic of H3.3-bio profiling.

(C) 2 days post-fertilization cmlc2:H3.3-bio larva, showing cardiac EGFP expression. Scale bar, 20 μm.

(D) CMs (green) from adult uninjured cmlc2:H3.3-bio hearts contain nuclear H3.3-bio, detected with an anti-FLAG antibody (red). Scale bar, 20 μm.

(E) Genome browser snapshots indicating sparse H3.3 enrichment in and around genes with little or no expression in CMs (top track, dark blue). H3K27Ac from whole hearts (bottom track, light blue) is enriched around all promoters (blue arrows) and some gene bodies (dashed blue lines). y-axis = enrichment of H3.3 = (Nc+1) / (Nc) / (Ni+1) / (Ni), Nc = H3.3 ChIP, Ni = Input (top left). Top right, x-axis (kb).

(F) Cardiac transcription factors are enriched with H3.3 throughout the gene body and in nearby intergenic regions. Many H3.3 peaks overlap with enrichment of H3K27Ac (pink arrows). Many other peaks are only enriched for H3.3 (purple stars).

(G) Sarcomeric protein genes are enriched for both H3.3 and H3K27Ac, however there are many more H3.3 occupancy peaks.

To validate the cell type-specificity of the ChIP, we examined several cardiac genes for H3.3-bio occupancy. We observed little or no tagged H3.3 within or near genes with known preferential expression in endocardial or epicardial tissue, such as fli1a, raldh2, or cav1 (Figure 1E). By contrast, H3.3-bio was enriched in peaks throughout loci containing genes known to be expressed in multiple cardiac cell types, like the cardiogenic transcription factors gata4, hand2 and meis1b (Figure 1F). We also observed conspicuous H3.3-bio occupancy in loci containing muscle-restricted genes, such as those encoding sarcomeric proteins tnnt2a, vmhcl, and ttna (Figure 1G).

To gauge the efficacy of H3.3 profiling, we examined these same genes for their signatures of an alternative indicator of open chromatin, H3K27Ac, obtained by whole-genome profiling of adult ventricles. Unlike H3.3-bio profiles, H3K27Ac marks were enriched in the loci containing fli1a, raldh2, and cav1 (Figure 1E, bottom tracks), highlighting the cell type-specificity of the H3.3-bio strategy. Genomic regions containing cardiogenic transcription factor genes or sarcomeric protein genes had H3K27Ac enrichment at many of the same sequences represented by H3.3-bio profiling. Interestingly, we also identified many sequences within these genes with elevated H3.3-bio occupancy that were not also enriched with H3K27Ac marks, suggesting that histone H3.3 profiling identifies a broader range of elements (Figures 1F and 1G). Thus, cmlc2:H3.3-bio transgenic zebrafish enable profiling of open chromatin and possible regulatory elements active in CMs.

Identification of New Enhancers that Direct Expression in CMs

H3K27Ac marks are commonly used to distinguish enhancer elements (Rada-Iglesias et al., 2011). The presence of H3.3 at enhancer regulatory elements has been previously described (Jin et al., 2009). To test H3.3-bio as a genome-wide indicator of CM enhancers, we systematically compared profiles of H3.3 occupancy and H3K27Ac marks. To focus on putative enhancers in these comparisons, we considered intergenic and gene body peaks, and we excluded peaks within 2 kb of transcriptional start sites that likely represent promoter sequences. Cumulative analysis of the remaining 30,176 H3K27Ac-enriched peaks and 35,127 H3.3-bio occupancy peaks revealed a positive Pearson correlation of 0.45 (Figure 2A), with almost half of the identified peaks shared between the two datasets (Figure 2B). In addition to these shared peaks, we found 20,541 H3.3 occupancy peaks that showed no evidence of enrichment with H3K27Ac marks. This possibly indicates chromatin regulation that is independent of the H3K27Ac modification, although we cannot exclude H3.3-bio enrichment of rare cell populations. Additionally, there were 16,005 peaks enriched for H3K27Ac marks but not H3.3-bio, from which we suspected a major proportion of peaks represent chromatin regulation in non-muscle cells (Figure 2B). A search for homology to mammalian transcription factor binding sites corroborated the cell-type specificity of the two profiles. Sequences with H3.3-only peaks uniquely contained homology to the binding sites for GATA4, MEIS1 and MEF2D, factors associated with CMs. By contrast, sequences with H3K27Ac-only peaks contained homology to binding sites for ETS1, FLI1 and ELF3, factors associated with non-CM cells (Table S1). Thus, our datasets indicate that in addition to evidence for novel regulatory targets, sequence enrichment of H3.3 correlates well with that of H3K27Ac, a known indicator of enhancers.

Figure 2. H3.3 Profiling Reveals Enhancer Elements that Direct Expression in CMs.

Figure 2

Open in a new tab

(A) H3.3 enrichment (x-axis) and H3K27Ac enrichment (y-axis) correlate (Pearson test, p value = 0.454) in uninjured ventricles. The actual trend line between H3.3 and H3K27Ac is shown in red, and a hypothetical perfect correlation trend line is shown in dashed blue.

(B) Overlap of H3.3 (dark blue) and H3K27Ac (light blue) peaks from samples containing uninjured adult CMs. Top, the total peak numbers in each group. Predicted binding sites for transcription factors active in CMs (GATA4, MEIS1, MEF2C) were found in H3.3-only peaks. H3K27Ac-only peaks contained sequences homologous to binding sites of factors not recognized to be present in CMs (See Table S1).

(C–H) H3.3 peaks near cardiac transcription factor genes identify transcriptional enhancers directing expression in different CM subtypes. Transgenic fish contained stable enhancer reporter constructs with the H3.3 peak subcloned upstream of the minimal c-fos promoter driving EGFP. The dashed grey box shows the cloned H3.3 peak from within the browser tracks. A DNA region 16 kb upstream from tbx5a (16tbx5aENfos:EGFP) directs EGFP in trabecular CMs (C). A region 66 kb upstream of tbx20 (66tbx20ENfos:EGFP) directs EGFP in CMs in the single CM-thick primordial layer (D). A region within the first intron of ltbp3 (IN1ltbp3ENfos:EGFP) directs expression sporadically in CMs adjacent to the endocardial cell layer (E). A region 2 kb upstream of nkx2.5 (2nkx2.5ENfos:EGFP) directs expression broadly in ventricular CMs and weakly in smooth muscle of the outflow tract (F). A region 5 kb downstream of the nkx2.5 transcription start site (4.8nkx2.5ENfos:EGFP) directs expression in all CMs except those adjacent to the valve (G). Control transgenic fish containing only the c-fos minimal promoter express little or no detectable cardiac EGFP (H). Scale bars, 200 μm.

Nucleosome replacement with H3.3 variant has been observed at a range of genomic loci, and thus to this point it was unclear whether H3.3 profiling could effectively distinguish cis-regulatory enhancer elements. To functionally evaluate putative enhancers from H3.3-bio profiles, we subcloned eight different one- to 3-kb DNA regions indicated by H3.3 peaks into reporter gene constructs containing a c-fos minimal promoter and EGFP cassette. These regions have not been characterized in any species and were chosen based on their proximities to genes active in CMs. We generated stable transgenic zebrafish lines using these constructs and examined adult hearts for EGFP fluorescence (Figures 2C–2G). Whereas transgenic animals harboring the c-fos:EGFP construct alone showed little or no detectable cardiac EGFP, 6 of the 8 small elements directed EGFP expression in CMs of transgenic animals (Figure 2H). These elements were located within 2 to 66 kb of the transcription start sites of CM transcription factor genes nkx2.5, ltbp3, tbx5a and tbx20. Enhancer validation was irrespective of overlap between H3.3 and H3K27Ac peaks; in both overlapping and non-overlapping cases, 3 of 4 sequences representing H3.3 peaks displayed enhancer activity. These experiments indicate that enrichment of H3.3 alone is a reliable marker of enhancer activity.

Interestingly, the validated enhancers revealed by H3.3 profiling had distinct associated myocardial EGFP expression patterns (Figures 2C–2G), in some cases clearly representing distinct CM subpopulations. For example, the enhancer 16 kb upstream from the transcription factor gene tbx5a directed EGFP mainly in inner trabecular CMs, with little or no expression in CMs within the ventricular wall (Figure 2C). Deletion of 16tbx5aEN from the genome resulted in no detectable effects on embryonic development or gene expression, suggesting redundancy of this enhancer (data not shown). The enhancer linked to the transcription factor gene tbx20 preferentially directed expression in the single-cell thick primordial layer of CMs, located between the outermost cortical layer and inner trabeculae, and for which there is no known marker (Figure 2D) (Gupta and Poss, 2012). Detection of this domain by CM H3.3 profiling was unexpected, as the primordial layer occupies a small proportion of total ventricular muscle. In summary, transgenic H3.3 profiling can identify enhancers that direct expression in specific tissue domains, in this case within CM subpopulations of the adult zebrafish ventricle.

H3.3 Profiling Uncovers an Emerging Regenerative Program

Chromatin structure is a marker of developmental change in many systems (Romanoski et al., 2015), but it has not been extensively surveyed in cell lineages within a regenerating tissue. As a replacement histone, H3.3 is deposited de novo after nucleosomes are lost from mobilized genomic loci such as transcribed genes and their enhancers (Goldberg et al., 2010; Jin et al., 2009). Therefore, an increase in H3.3 enrichment during regeneration is expected to mark genomic regions that undergo nucleosome replacement as part of regenerative programs. To map changes in H3.3 enrichment during regeneration, we crossed the cmlc2:H3.3-bio transgene into a strain that enables tamoxifen-inducible genetic ablation of ~half of all CMs. Unlike a local injury like apical resection or cryoinjury, this model of injury depletes CMs diffusely throughout the ventricle and thus boosts the proportion of CMs that are actively regenerating (Wang et al., 2011). We compared H3.3 occupancy in uninjured hearts to those 14 days post injury (14 dpi) by high-throughput sequencing of purified H3.3-associated DNA fragments. Using a conservative false discovery rate (FDR) threshold of 0.01 (see Experimental Procedures), we detected 45,782 H3.3-enriched loci (peaks) in uninjured adult CMs, 94% of which are retained in regenerating hearts (Figure 3A). Moreover, we identified a genome-wide near doubling of H3.3 peaks in regenerating hearts, with 41,660 novel H3.3 loci, in total covering 2.4% of the genome. These data indicate that widespread changes in CM chromatin structure occur during regeneration, superimposed upon a baseline adult CM signature.

Figure 3. H3.3 Profiling Uncovers an Emerging Regulatory Program during Heart Regeneration.

Figure 3

Open in a new tab

(A) Most of the total H3.3 peaks recovered in samples from uninjured hearts overlapped with peaks from samples of regenerating hearts (purple), with a relatively small number of H3.3 peaks unique to the uninjured profile (blue). A large number of regeneration-specific H3.3 peaks emerged (red). See Figure S1A for peak distributions relative to genes.

(B) H3.3 peaks were used to identify genes in their proximity. Most of identified H3.3 peaks were found within genes, either at the promoter (blue) or in gene bodies (light blue). Promoters here are defined as within 2 kb of a transcription start site (TSS). Intergenic peaks (dark blue) do not overlap within any known transcriptional units. Percentages are calculated based upon the total number of peaks.

(C) Genes were identified by their proximity to H3.3 peaks. Peaks were separated into promoter (TSS), gene body (GB) and intergenic (IG) categories. See Figure S1B. 40% and 52% of the genes identified from samples of uninjured and regenerating hearts, respectively, were identified in at least two categories.

(D) The cartoons show examples of each of the four classes of dynamic peaks.

(E) Promoters are characterized mainly by H3.3 peaks decreasing in magnitude from the uninjured (82%, light blue / total). An addition 2,629 promoters have emerging H3.3 peaks during regeneration (red).

(F) Most emerging H3.3 peaks (4,592 / 4,752) are putative enhancers (gene body and intergenic peaks). Many genes were also identified with decreasing H3.3 at putative enhancers (1,432), but not to the extent of that in promoters.

H3.3 occupancy has been described as a general feature of promoters and bodies of active genes, in addition to potential enhancers (Jin et al., 2009). We scored H3.3-enriched DNA regions from uninjured and regenerating hearts with respect to their position within and near genes. We found that ~75% of H3.3 peaks are found in promoters or gene bodies, with the remainder in intergenic regions as far as 530 kb from a promoter (Figure 3B). A total of 14,148 genes contained or were assigned H3.3 peaks in samples of CMs from uninjured hearts, and 17,216 genes had associated H3.3 peaks in samples of regenerating CMs (Figures 3C and 3D). More than 40% of these genes had multiple assigned H3.3 peaks; for example, one at the promoter and another upstream. Such an arrangement suggests open chromatin, gene expression, and the location of a possible cis-regulatory element.

Of the genes with significantly increasing enrichment of H3.3 during regeneration, 93% include peaks appearing de novo, whereas the other 7% contain H3.3 peaks increasing significantly in magnitude from a preexisting enrichment in uninjured CMs (Figures 3E and 3F). Interestingly, 84% of genes identified from H3.3 at promoters in the uninjured profile have decreasing levels of H3.3 during regeneration (Figure 3F; Table S2, third column). Although it is unclear whether this reflects underlying changes in gene expression (see below), widespread change in H3.3 enrichment is indicative of regulation of chromatin architecture at basal CM promoters that might be permissive for CM dedifferentiation and proliferation.

To further understand structural features of chromatin in regenerating CMs, we assessed the sequence lengths of H3.3-enriched loci. The size (length) of profiled features present in promoter chromatin has been shown to mirror underlying changes in modes of transcriptional regulation (Benayoun et al., 2014; Duncan et al., 2015). We observed a shift from ~800 bp to ~600 bp in the average lengths of H3.3-associated regions during regeneration (Figure S1C; Table S2). Narrow occupancy peaks (< 710 bp) are more frequent in regenerating samples compared to uninjured samples (40.4% vs. 26.5%, respectively; percentage of the total number of peaks found for each group in this size range). Emerging H3.3 peaks are mostly narrow peaks, as are peaks that disappear (Figure S1C, 62 and 72% respectively). This group of narrow H3.3 loci is enriched mainly for regions associated with putative enhancers (see Table S2, uninjured: enhancers (29%), promoters (22%); regeneration: enhancers (43%), promoters (32%)). Thus, most highly dynamic peaks are narrow, and they are more likely to be enhancers rather than promoters. Peaks that are decreasing but still present after injury are mostly wide peaks and are located at promoters (76%). In summary, changes in the size of H3.3 loci mostly reflect the large-scale activation of cis-regulatory elements during regeneration.

We identified differential H3.3 peaks in and around many genes known to increase cardiac transcript levels during heart regeneration, including gata4, tbx5a, tbx20, stat3, nppa, and nfkbiab (Figure S2A) (Fang et al., 2013; Gupta et al., 2013; Karra et al., 2015; Lepilina et al., 2006). Previously, we identified 145 genes up-regulated during regeneration by profiling ribosome-associated RNA specifically in the CMs of the cmlc2:TRAP zebrafish line (Fang et al., 2013). Although different forms of cardiac injury were administered, increasing H3.3 was found near 129 (or 89%) of these identified genes. To determine if dynamics of H3.3 broadly reflect changes in gene expression during regeneration, we generated RNAseq profiles from uninjured and regenerating zebrafish hearts. We found that the presence of H3.3 at transcription start sites correlated positively with mRNA abundance in both uninjured and regenerating samples, consistent with the reported properties of H3.3 (Figure 4A). Correlation was observed for those genes identified with >2 fold changes in proximal H3.3 and those identified as dynamic by RNAseq (Figure 4B). This association is particularly meaningful as the RNAseq and H3.3-bio profiles were obtained from different cellular complexities, and as H3.3 proteins that insert into open chromatin after injury might persist in nucleosomes for a much longer time period than that of labile mRNAs assayed by RNAseq. To further assess the relationship between H3.3 and gene expression changes, we used in situ hybridization (ISH) to visually inspect a sampling of transcripts from genes that increase in H3.3 occupancy during regeneration. Assessment of genes with H3.3 peaks that increase throughout the gene body (including the promoter) indicated induction of visible expression during regeneration in each case (Figure 4C). Similarly, genes identified with H3.3 increasing only in putative enhancers within the gene body and/or nearby intergenic regions demonstrated an increase in expression (Figure 4D and S3C). The increased ISH signal in regenerating tissue samples indicates that dynamic H3.3 enrichment can be used to identify gene expression changes during regeneration.

Figure 4. H3.3 Profiles Identify Genes Changing Expression during Regeneration.

Figure 4

Open in a new tab

(A) Promoters that contain H3.3 have an increased frequency of RNA expression. RNAseq fragment frequencies in uninjured (blue) and regenerating (red) hearts correlate with score/height of the highest H3.3 peak at transcription start sites (TSS, +/− 2 kb). Kolmogorov-Smirnov (KS) (p value < 2.2 x 10−16 for each comparison (uninjured and regeneration).

(B) Changes in H3.3 occupancy correlate with changes in mRNA abundance (RNAseq). Left, Fold-change in enrichment of H3.3 in promoter regions for genes with increased (UP) or decreased (DOWN) expression during regeneration. Only fold changes with log2 (FC) > 0.5 were considered. Right, Fold-change of gene expression for genes either enriched or depleted of H3.3 at their promoter regions. False discovery rate <1. Pearson score left to right : 0.064, −0.019, 0.043, 0.072.

(C) In situ hybridization (ISH) evidence that emerging/increasing H3.3 peaks indicate changes in gene expression. Top, Uninjured ventricular apices (left) show little detectable signal, whereas hearts 7 days after induction of ablation (7 dpi) indicate gene expression by a violet signal. Bottom, Probes were designed toward transcripts highlighted in blue at the bottom of the genome browser shot. The tracks show enrichment of H3.3 (top two) from uninjured (blue) and regenerating (red) profiles (see Methods). Peaks assayed for enhancer activity in transgenic reporters are indicated by a dashed grey box (see Figure S3). Genes with H3.3 increasing throughout the locus also mark genes that increase mRNA levels as gauged by ISH. Zebrafish smooth muscle actin (tagln), sarcomeric stress protein cardiac muscle ankyrin repeat domain 1a (ankrd1a) and hormone endothelin2 have H3.3 increasing throughout gene bodies. Scale bar, 125 μm.

(D) Transcription factors runx1 and bach1b, and cytokinesis factor anillin each have static H3.3 at transcription start sites (blue arrow) but increasing H3.3 at nearby putative enhancers, and each show regeneration-responsive ISH signals. Top right, x-axis. See Figure S2 for other examples.

(E) CM H3.3 and whole ventricle H3K27Ac peak overlap during regeneration is similar to that from uninjured hearts (see Figure 2A). 38,905 peaks are detected only by H3.3 enrichment. Total numbers of peaks are displayed at the top.

(F) H3.3 enrichment (x-axis) and H3K27Ac enrichment (y-axis) correlate in regenerating ventricles (Pearson test, p value = 0.404). The actual trend line between H3.3 and H3K27Ac is shown in red, and a hypothetical perfect correlation trend line is shown in dashed blue.

Several new regeneration-responsive markers were revealed by H3.3 profiling and ISH sampling. In mammals, smooth muscle actin is induced in failing hearts and considered to be a marker of dedifferentiation (Kubin et al., 2011). transgelin (tagln), the zebrafish ortholog of smooth muscle actin, is also upregulated in CMs during zebrafish heart regeneration (Figure 4C). Also induced in CMs during regeneration were factors expected to facilitate CM division like anillin, required for cleavage furrow formation during cytokinesis, and ankrd1a, encoding a muscle ankyrin repeat protein that is upregulated during cell stress (Zhang and Maddox, 2010; Zolk et al., 2002). Genes encoding transcriptional regulators such as bach1b, csrp1a and runx1 were upregulated throughout the heart (Figure 4D and S3C). Runx1 is a transcription factor important in blood development that was also shown to be a marker of dedifferentiation in mammalian CMs (Kubin et al., 2011). The translational regulator eif4e1c was an induced marker, as was the vasoconstriction factor endothelin2 (Figure 4C and 4D) and angptl2, encoding an extracellular matrix protein involved in endothelial cell sprouting (Figure S3C). In summary, our data indicate that a new genetic program emerges during heart regeneration, representing a near doubling in the number of H3.3 loci and occupancy of 2.4% of the zebrafish genome. Concurrently, H3.3 occupancies that comprise the adult CM signature generally undergo signal diminution, a possible indicator of dedifferentiation. Furthermore, genes induced during regeneration can be identified by their proximity to de novo H3.3 peaks associated with regeneration.

CM Regeneration Enhancer Elements Are Marked by H3.3

Interestingly, overlap between total H3.3 and total H3K27Ac enrichments was less pronounced during heart regeneration than among uninjured samples (see Figures 2A and 2B). During regeneration, the calculated Pearson correlation of 0.40 was similar to that from uninjured samples (0.45), yet 78% of total H3.3 peaks were not represented in the regenerating H3K27Ac data (Figures 4E and 4F). Thus, the majority of H3.3 peaks emerging during regeneration are not identified with the known enhancer mark H3K27Ac. Furthermore, subsets of loci that contained either H3.3 or H3K27Ac peaks in uninjured samples often displayed distinct responses during regeneration. For example, 5,940 loci were marked with H3K27Ac in uninjured samples but have an emerging H3.3 peak only during regeneration. In kind, 1,716 other H3.3-containing uninjured loci acquired new H3K27Ac marks during regeneration. In these instances, either H3.3 or H3K27Ac would identify these loci as putative enhancers, but comparison of the two signatures has the potential to uncover further information about their regulation. It is formally possible that some H3.3-only regions may represent detection of rare CM subtypes that are cloaked in whole-ventricle H3K27Ac profiles. Yet we note that in this study, using H3.3 enrichment, we isolated and validated an enhancer that directed broad cardiac expression during regeneration but was not identifiable by H3K27Ac enrichment (see 22sema3aaEN below).

To identify enhancer elements that may be responsible for changes in CM gene expression during regeneration, we examined intergenic DNA regions that increase in H3.3 occupancy during regeneration. We found 11,964 intergenic H3.3 peaks (28% of total) arising de novo in the vicinity of 5,232 genes (Figure S2C; Table S2). From these genes, transcriptome sequencing identified 1,352 genes with a nearby emerging H3.3 peak at a putative enhancer that also changed in expression. Many (59.4%) of these genes also had increasing enrichment of H3.3 at their promoters or in gene bodies, suggesting possible cis-regulation from these nearby intergenic sequences. We next created transgenic reporter lines using 28 H3.3 peaks that increase during regeneration (for examples see dashed grey boxes in Figures 4C and 4D). From this pool of new lines, 23 of 28 (82%) of the H3.3 loci could direct EGFP expression in tissue domains of larval zebrafish (Figure S3), a similar success rate as we observed with H3.3 peaks near cardiac transcription factor genes. Only 3 of these 23 elements directed larval cardiac expression. Sixteen elements directed EGFP fluorescence in the developing central nervous system (Figure S4). Thus, many elements activated in regenerating adult CMs ostensibly can engage with transcription factors in developing larval tissues.

The concordance between H3.3 occupancy and enhancer activity suggested that emerging H3.3 peaks represent cis-regulatory sequences that differentially regulate gene expression during regeneration. Therefore, we assayed EGFP fluorescence induced after cardiac injury in a subset of stable transgenic lines generated using these sequences. Transgenic lines with the minimal fos promoter alone showed negligible cardiac EGFP fluorescence (Figures 5A and 5B). An enhancer located 103 kb upstream from the runx1 gene directed minimal EGFP fluorescence in uninjured adult 103runx1ENfos:EGFP ventricles, limited to CMs near valves (Figure 5C). By contrast, ablation or resection injuries induced 103runx1ENfos:EGFP in CMs and other cardiac cell types (Figures 5D). In EdU labeling experiments, EGFP-positive CMs adjacent to the wound incorporated EdU, indicative of proliferation and participation in regeneration (Figure 5E). zgc:136858, a gene of unknown function, acquires a large intragenic H3.3 peak during regeneration (Figure S5A). H3.3 enrichment and RNAseq data suggest that this peak might regulate genes for voltage-gated potassium channels, kcna1 and/or kcna6a, located ~50 kb upstream. IN13zgc:136858ENfos:EGFP fluorescence is weak in uninjured CMs and strongly stimulated in regenerating CMs (Figures 5F and 5G). The endothelin2-linked element directed expression in adult CMs near the ventricular lumen in uninjured fish and throughout the ventricle after injury (Figures 5H and 5I). Homozygous deletion of 3endothelin2EN from the genome resulted in no defect in regeneration or endothelin2 expression (data not shown). A sequence ~5 kb upstream of anillin directed sporadic EGFP fluorescence in uninjured hearts, possibly reflecting cell cycle-dependent regulation of the anillin gene product (Figure 5J) (Zhang and Maddox, 2010). Consistent with this notion, the anillin-linked enhancer enabled injury-induced EGFP fluorescence in multiple cardiac cell types (Figure 5K). An enhancer 22 kb upstream of the sema3aa gene was induced throughout the ventricle after genetic ablation injury (Figures 5L and 5M). As only a subset of the stable transgenic lines activated EGFP during regeneration, we suspect this reflects that the 1–2 kb regions are not sufficient for regeneration-stimulated activation without an additional element or their endogenous promoters.

Figure 5. Regeneration Enhancers Direct Gene Expression in CMs after Injury.

Figure 5

Open in a new tab

(A and B) Uninjured adult hearts from transgenic reporter lines containing the c-fos minimal promoter alone do not express EGFP whether uninjured (Un) or after induced ablation of CMs (7 dpi). Scale bar, 200 μm.

(C and D) Uninjured 103runx1ENfos:EGFP animals express EGFP in CMs adjacent to valves. Ablation injury induced EGFP throughout regenerating CMs, mainly in the compact layer.

(E) EGFP fluorescence is induced at 103runx1ENfos:EGFP injury sites at 7 days after apical resection (7 dpa). EGFP positive CM from the site of injury in 103runx1ENfos:EGFP hearts incorporate EdU (red). (E1), high-magnification view of box 1 in (E).

(F and G) IN13unkENfos:EGFP hearts have low myocardial EGFP fluorescence that is enhanced by ablation injury. (G2), high-magnification view of box 2 in (G). (H and I) 3endothelin2ENfos:EGFP animals express EGFP in CMs near the ventricular lumen and occasionally in the compact layer. Expression increases in regenerating CMs at 7 dpi.

(J and K) 5anillinENfos:EGFP animals display EGFP sporadically in uninjured hearts. Expression increases in regenerating CMs after induced ablation of CMs.

(L and M) 22sema3aaENfos:EGFP zebrafish show induced expression throughout CMs only after injury.

An antibody against Myosin Heavy Chain (MHC, red) was used to stain cardiac muscle. Boxed areas are shown at high magnification in E1 and G2. See also Figure S5.

We recently reported identification of a short DNA element linked to the zebrafish leptin b gene that is sufficient to activate gene expression upon injury to fins or hearts and during their regeneration (Kang et al., 2016). This study provided evidence for the existence of tissue regeneration enhancer elements (TREEs) that can be applied as reagents to modulate regenerative capacity. Here, we have identified several DNA elements that are marked by increased CM H3.3 occupancy during heart regeneration. These sequences activate reporter gene expression in regenerating hearts of transgenic fish, and thus represent new TREEs (Kang et al., 2016). Because of their origin and domains we refer to them hereafter as CM regeneration enhancer elements, or CREEs. Interestingly, three of the five CREEs also induced EGFP in caudal fins after amputation (Figure S5B), indicating that a subset of CREEs can activate expression in other regenerating tissues.

A Potential Core Regulatory Signature for CM Proliferation

Tissue regeneration has 3 or more recognized phases, including: 1) early injury responses and wound healing; 2) source cell activation and proliferation; and 3) patterning and morphogenesis (Poss, 2010). In our experiments, we profiled H3.3 occupancy a full 14 days after the initiation of injury, in attempts to focus on activation and maintenance of CM proliferation. Yet, it remains possible that a proportion of sequences marked by H3.3 peaks in our regeneration samples reflect an active or past injury response rather than cardiomyogenesis.

To assess the adult cardiomyogenic program in a second context, we profiled H3.3 occupancy in transgenic zebrafish that enable tamoxifen-inducible cardiac expression of Nrg1. Induced Nrg1 expression stimulates overt CM proliferation in the absence of injury, which to our knowledge is a unique effect of a single expressed growth factor (Gemberling et al., 2015). CM H3.3 profiling after 7 days of induced Nrg1 stimulation revealed greater overlap with the profile of regenerating CMs (89%) than that of uninjured CMs (80%) (Figure 6A). Interestingly, the total number of emerging H3.3 peaks from Nrg1-stimulated samples was much smaller than that from regenerating CMs, lacking 46% of the peaks that emerge during regeneration. This difference may reflect observations that, in this system, Nrg1 mainly enhances CM proliferation in the ventricular wall and not trabeculae (Gemberling et al., 2015). Even small genomic regions could contain elements with distinct dynamisms in these contexts. For instance, we identified an element 5 kb upstream of the start site of the gene tnnc1b that sharply increases H3.3 occupancy during either regeneration or Nrg1 stimulation, as well as a second element a further 5 kb upstream that boosts H3.3 during regeneration but not during Nrg1 stimulation (Figure 6C). This suggests the presence within the same regulatory space of both an injury-responsive enhancer and a separate enhancer that tracks CM proliferation. At the RNA level, tnnc1b is induced in both contexts (Figure 6D).

Figure 6. Shared Histone Dynamism Signatures Between Injury- and Nrg1- induced CM Proliferation.

Figure 6

Open in a new tab

(A) The Nrg1 H3.3 profile (green) overlaps more with the regeneration profile (red) than the uninjured profile (blue). The percentages represent the fraction of the total for each group and are labeled with color-codes. For example, 12% of Nrg1 peaks overlap with 5% of regeneration peaks, neither of which overlap with the uninjured. However, the Nrg1 H3.3 profile only overlaps with 10% of the total peaks emerging during regeneration (5% / 5% + 46%).

(B) Left, H3.3 peaks near 4,336 genes that increase during regeneration with respect to uninjured CMs, also increase during Nrg1-stimulated hyperplasia (yellow). See Table S2. Many peaks increase H3.3 occupancy only during regeneration (red) or Nrg1 stimulation (green). Right, Genes identified near H3.3 peaks that decrease from the uninjured profile highly overlap between regenerating and Nrg1-stimulated CM samples (13,340 total or 99% of Regeneration and 96% of Nrg1 peaks). See Table S5.

(C) The tnnc1b gene is located just downstream of a regeneration-specific H3.3 enrichment site (red box) and an site enriched with H3.3 during regeneration and Nrg1 stimulation (yellow box).

(D) In situ hybridization shows visual increases of the tnnc1b transcript (violet signals, arrows) in ventricles regenerating from ablation (7 dpi) and resection (7 dpa), and in ventricles overexpressing Nrg1 (7 dpt). Scale bar, 125 μm.

(E) Genome browser tracks indicating H3.3 occupancy responses during Nrg1 stimulation of 3 sequences used to generate CREE reporter lines (dashed boxes). The top two tracks show enrichment of H3.3 in uninjured (blue) and Nrg1-stimulated (green) ventricles. Genes are cartooned at the bottom in light blue. Top right, x-axis.

(F) CREE reporters 5anillinENfos:EGFP (left), IN13zgc:136858ENfos:EGFP (middle), and103runx1ENfos:EGFP (right) induce EGFP in CMs in the ventricular wall during transgenic Nrg1 stimulation. An antibody against Myosin Heavy Chain (MHC, red) was used to stain cardiac muscle. Scale bar, 50 μm.

We identified 4,336 genes that contain or are near emerging H3.3 peaks that are shared in both regeneration and Nrg1 overexpression settings (Figures 6B and 6C, left). These genes were enriched for dozens of gene ontology categories including those related to heart muscle development and morphogenesis (Table S4). The bulk of genes (79%) were identified from H3.3 peaks appearing in gene bodies and intergenic regions, rather than promoters (Table S2). Many of the genetic elements described earlier in this study that increase H3.3 occupancy during regeneration, including those within or near bach1b, eif4e1c and transgelin, also become enriched with H3.3 during Nrg1 stimulation (Figures 6E and S6). To further assess H3.3 occupancy, we examined 4 CREE reporter lines in the context of induced cardiac Nrg1 overexpression. Each of the elevated H3.3 peaks that identified the CREE during regeneration were also present in the Nrg1 H3.3 profile. Of these, CREEs linked to runx1, anillin and zgc:136858 genes showed robust stimulation of EGFP throughout the compact muscle layer - where most CM proliferation occurs (Figures 6E and 6F) (Gemberling al., 2015). The 3endothelinEN reporter was not activated (Figure S6B). Of note, we observed a genome-wide decrease in H3.3 occupancy levels from the uninjured CM profile during Nrg1 stimulation, supporting this reduction as a general feature of CM proliferation (Figure 6B, right). Overall 13,430 genes were identified with decreasing H3.3, including 90% of the total promoters identified as containing H3.3 in the uninjured profile (Table S5).

In summary, by comparing H3.3 occupancy between CMs undergoing injury-induced or Nrg1-stimulated hyperplasia, our datasets suggest a core regulatory signature of the regeneration program with separable components representing injury and proliferation.

Identification of Regulatory Motifs Enriched in Regenerating CMs

Enhancer elements are bound by transcription factors and can harbor one or many binding sites for these factors. To determine whether H3.3 enrichment regions from regenerating CMs contained sequence motifs potentially recognized by transcriptional regulators, we used the MEME software suite (Bailey et al., 2009). We examined H3.3 peaks located in gene bodies or intergenic regions and that were emerging during regeneration. Overall, we identified 325 motifs that were comprised of simple repeats as well as others with more complex sequence organization up to 41 bp long (Figures 7A and 7B). Although these motifs were identified in regeneration-specific peaks, it is possible they represent general CM cis-regulatory sequences. A ‘regeneration specificity score’ was calculated for the relative affinity of each motif for regeneration-only or uninjured-only H3.3 peaks (Table S5, see Methods). Using a conservative cutoff of ‘regeneration specificity score’ < 0.05, we filtered these to 118 of those motifs that are enriched in regeneration-specific H3.3 peaks (Figure S7A). There were multiple similar motifs represented within several of the more complex sequences, which enabled consolidation into 38 groups (see STAR Methods for details). While the complete set of sequence motifs important in CMs for regeneration is likely to include those sequences also enriched for high H3.3 occupancy in uninjured CMs, these variations on 38 different motifs enriched only during regeneration are hereafter referred to as CM regeneration motifs (CRMs).

Figure 7. Enriched Transcription Factor Binding Motifs Within Regeneration-Associated H3.3 Peaks.

Figure 7

Open in a new tab

(A, B) Multiple DNA sequence motifs are enriched genome-wide within H3.3 peaks that emerge during regeneration, called CM regeneration motifs (CRMs). Shown are position weight matrices of the longest motif (A) and a simple repeat-containing motif (B). E-values represent the significance score for the motif as calculated by MEME. It is the log likelihood ratio for a randomly generated set of sequences with the same width and frequency.

(C) Five different identified CRMs are shown on a map of the 5 CREE elements that were validated in transgenic reporter strains (grey rectangles, see Figures 5, S5 and Table S4). Highlighted are sequences homologous to the human transcription factors (labeled from JASPAR). P-values for binding sites: MEIS1 = 0.004, Myod = 0.0001, NFKB = 0.005, NKX2.5 = 0.003, SPI1 = 0.005, RFX2 = 4.1 x 10−5, RUNX2B = 0.005, STAT1 = 0.017.

(D) Histogram shows the number of human p300 peaks (y-axis) with homology to identified CRM as ranked by increasing ‘regeneration specificity score’ or -log10 (p-value of Kolmogorov-Smirnov test, p < 1 x 106, x-axis). The motif with similarity to the binding site of the MYOD family of transcription factors is indicated in red.

(E) Histogram shows all human p300 peaks (y-axis) ranked by percentage identity to validated CREE.

Each CRM was present in tens to hundreds of other putative enhancers and promoters distributed throughout the genome, including within the CREEs we validated with transgenic reporter lines (Figure 7C). To uncover potential regulators that engage with these CREE sequences, we used the FIMO software package to compare homology between the embedded CRMs and predicted binding sites of transcription factors from the JASPAR database (Grant et al., 2011). In addition to containing single-site submotifs predicted to be recognized by transcription factors like RUNX2B, RFX2, and SPI1, these CRMs possessed sequences homologous to composite transcription factor binding sites. For example, a CRM found in 3 of 5 validated CREEs contained predicted binding sites for MYOD and NFkB, the latter of which is essential in CMs for zebrafish heart regeneration (Karra et al., 2015). A different CRM also found in 4 of 5 validated CREEs contained predicted sites for MEIS1, a transcription factor that regulates CM proliferation (Mahmoud et al., 2013), as well as those for NKX2.5 and the signaling effector STAT1. Multiple predicted RUNX1/2B sites were contained within the runx1-linked CREE, suggesting the possibility of auto-regulation. Notably, H3.3 occupancy increased during regeneration in sequence regions linked to zebrafish rfx2, runx2b and spi1a genes themselves, suggesting that the factors they encode are expressed in CMs and regulated during regeneration (Figure S7B). Thus, the H3.3 datasets we report here reveal potential upstream regulators, downstream target genes, and regulatory interactions that underlie programmatic changes within a single central cell type in regenerating cardiac tissue.

To assess evolutionary conservation of CRMs, we examined whether the sequences are present in the human genome. We focused on loci identified by p300/CBP occupancy from whole human heart tissue likely to be accessible in either fetal or postnatal contexts (May et al., 2012). CRMs that were the most statistically enriched in regenerating samples had no homology with the human genome (86% of CRMs with a ‘regeneration specificity score’ > 3), with a few notable exceptions (Figure 7D). CRM18 was almost identical to the binding site of the myogenic family of transcription factors characterized by MYOD and could be found in 416 different human p300 cardiac peaks. Three other CRMs with ‘regeneration specificity score’ > 3 shared homology to the human genome (CRM2, CRM17 and CRM27 with 93, 99 and 68 human p300 peaks using a p value < 1 x 10−6, respectively; Table S6). Interestingly we also found multiple sequences enriched for p300 near the human ENDOTHELIN2 gene with homology to CRM17, possibly indicative of similarity to the CREE linked to zebrafish endothelin2 (see Figure 5). We extended this analysis to see if CREEs were more broadly conserved beyond the CRM motifs and searched 10-mer fragments comprising each CREE for identity to human p300 peaks. We found 27 putative human enhancers with 10-mer-based homology to the described CREEs: 5anillinEN (1), 22sema3aa (1), IN13unkEN (7), and 103runx1EN (18) (Figure 7E and Table S6). The same p300 peak ~37 kb upstream from the FGF18 transcription start site was identified from 10-mer identity to both CREEs, 22sema3aaEN and 103runx1EN, and by homology to motif CRM27. Using restrictive cut-offs, from the 7,261 total p300 peaks detected in human cardiac tissue (May et al., 2012), 286 (3.9%) shared some homology with either regeneration enriched motifs or validated CREEs. We speculate that a portion of these shared putative regulatory elements are activated during CM proliferation across species.

Conclusions

Transgenic H3.3 profiling was originally developed by the Henikoff group, revealing that highly expressed genes also contain the highest H3.3 enrichment in whole C. elegans embryos (Ooi et al., 2010). Here we show that H3.3-bio profiling can also uncover gene expression associated with a specific cell type within a whole organ. Further, we show evidence that H3.3-enriched loci have a high probability of marking transcriptional enhancers, whether located distal to or within gene bodies. By transgenic profiling of the histone variant H3.3 in CMs, we have constructed context-specific atlases of the activated genome. We compare adult zebrafish CMs in their native state, during injury-induced regeneration, and under conditions of growth factor-stimulated hyperplasia. This resource provides genome-wide signatures, new markers, validated regulatory sequences useful for tool generation, and new transgenic strains to facilitate investigations of heart regeneration.

H3.3-profiling requires a new transgenic strain for cell type-specific applications and is in this way similar to FAIRE or ATACseq, which typically involves cell sorting from a transgenic line. Unlike FAIRE or ATACseq, H3.3-bio profiling does not require cell isolation methods known to stimulate a regeneration-like response and raise the possibility of in vitro artifacts. As compared with whole-ventricle H3K27Ac assays, we detected thousands more peaks with H3.3-occupancy in CM genomes as compared to H3K27Ac in cardiac genomes, several that we validated as enhancers in transgenic animals. Interestingly, we found that 78% of regions in CMs with dynamic acquisition of H3.3 did not show evidence for a concurrent measurable change in H3K27Ac occupancy, suggesting that H3.3 enrichment need not occur through p300 acetylation during regeneration. The histone variant H2A.Z has also been found at enhancers where it has been linked with H3.3 (Jin et al., 2009). Whether H2A.Z profiling can similarly function as a tool for genetic program discovery is unclear, as similar transgenic lines we generated did not thrive (data not shown). It will be interesting and important to examine other key chromatin features in a cell-type specific manner to finely tune the data presented here and possibly uncover other specific mechanisms of chromatin regulation.

We expect that the histone replacement profiles and tools we report here will have generalizable value for studies of tissue development and regeneration. For instance, tbx20-linked regulatory sequences marking primordial layer CMs should prove useful for lineage tracing or genetic manipulation of this subtype of CM. Databases of candidate and validated CREEs that preferentially activate during regeneration will provoke comparative studies in other species, including how orthologous sequences respond to cardiac injuries. Comparison to profiles from other cardiac cell types like epicardial and endothelial cells, and cells from other tissues, will also be informative in this regard. Future discoveries in mammalian heart repair might be enabled with new markers of proliferative CMs, or sequences that can be engineered to naturally deliver pro-regenerative factors upon injury (Kang et al., 2016). Targeted studies that dissect the functions of the regeneration enhancers and motifs we have identified, and the factors that engage with them, will help construct a gene regulatory network of tissue regeneration that has therapeutic implications.

STAR METHODS

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact Kenneth Poss (kenneth.poss@duke.edu).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Wild-type or transgenic male and female zebrafish of the outbred Ekkwill (EK) strain ranging in age from 3 to 12 months were used for all experiments with adult animals. Approximately equal sex ratios were used for experiments. Fish were housed at approximately 4 fish per liter in Pentair aquarium racks and fed three times daily, and the water temperature was maintained at 26°C.

METHOD DETAILS

Construction of cmlc2:H3.3-bio Zebrafish

Zebrafish histone H3.3 (H3f3d) was amplified by RT-PCR and subcloned into pCS2 using BamHI and XhoI restriction enzymes. Oligos containing the FLAG peptide sequence followed by the biotin ligase recognition peptide sequence were inserted into the upstream EcoRI site in frame. The BirAGFP fusion protein gene was amplified with oligos containing a forward primer with a BamHI site and a reverse oligo containing the 2A co-cistronic expression peptide followed by EcoRI. A plasmid containing the 5.1 kb cmlc2 promoter (Kikuchi et al., 2010) was digested with BamHI and XhoI to remove the EGFP insert. Finally a tripartite ligation was performed to fuse the BirAGFP to the cmlc2 promoter, followed by the Flag-BLRP-H3f3d fusion protein gene fused to the 2A peptide sequences, followed by an SV40 polyA sequence. The entire construct contained flanking I-Sce meganuclease restriction sites that facilitate transgenesis (Babaryka et al., 2009). Briefly, ~1 ug of constructs were digested for 30 minutes at 37C with 10 U enzyme. Injected embryos were screened for cardiac fluorescence, and one of the stable cmlc2:H3.3-bio lines showing strong expression was selected for this study. The full name of the line is Tg(cmlc2:BirAGFP-2A-FLAG-BLRP-H3f3d)pd185. We confirmed expression of the transgenic H3.3 using a western blot against the FLAG epitope on whole heart nuclear extract (10 ug). H3.3-bio was not detectable using an H3.3 antibody, however endogenous H3.3 was detectable. Loading of increasing amounts of total protein resulted in the endogenous signal overwhelming the region of H3.3-bio as determined by the blot against the FLAG epitope. We also immunopurified H3.3-bio using anti-FLAG beads (M2) and were then able to detect both transgenic and endogenous H3.3 in similar amounts (Figure S1B).

Construction of Transgenic Reporter Strains

Test loci were amplified from EK zebrafish strain genomic DNA using PCR oligos listed in Table S3 and inserted as described into pCR8-GW-topoTA (Invitrogen K2500-20). Vectors were subsequently recombined with PMP6, a Gateway vector produced in the Poss lab containing LR recombination sites upstream of the 95 bp minimal mouse fos promoter driving EGFP (Fivaz et al., 2000). The entire region was flanked by the GAB and HS4 insulators with distal I-Sce restriction sites. When necessary, the test regions were recombined in an orientation such that expression of EGFP would be the result of enhancer rather than possible alternative promoter activity. After digestion with the I-SceI meganuclease, constructs were injected into fertilized zebrafish embryos using standard transgenesis techniques. EGFP-positive embryos were examined at 1, 2 or 3 dpf and expression domains were noted if discernible. For constructs in which less than 20% of embryos showed EGFP fluorescence, both EGFP+ and EGFP− F0 embryos were raised to adulthood for screening.

Enhancers were considered ‘positive’ when two or more separate stable lines were recovered with similar EGFP expression domains in larvae. These domains were compared for consistency to those found in the injected F0 animals. We selected stable lines using PCR on pools of F1 embryos for two of the constructs with no detectable larval EGFP, (12bach1bENfos:EGFP)pd200 and (110tfebENfos:EGFP)pd224, as it was possible that these elements would activate EGFP in the adult or after injury. We also established several lines comprising minimal promoters without additional elements, to be used as negative controls for injury responsiveness (Figure 2H). The 5.5hand2-ENfos:EGFP, 72luzp1fos:EGFP and 70zeb2aENfos:EGFP constructs did not yield larval EGFP fluorescence, and we recovered only one transgenic line each. Thus, it is possible that further injections and screening would indicate these lines as enhancers. We selected 5 independent lines for the 116foxp4ENfos:EGFP construct, each of which had a unique larval expression domain; this likely reflects local cis-regulatory effects dependent on the area of insertion rather than activity of the tested enhancer candidate.

None of the negative control lines (fos:EGFP)pd210 showed cardiac EGFP expression after injuries to larvae or adults. Additionally, 16tbx5aENfos:EGFP and IN1ltbp3+IN1ENfos:EGFP transgenic lines, generated using elements with H3.3 levels that remain stable during regeneration, showed no detectable changes in EGFP fluorescence upon injury.

Native ChIP

For each chromatin immunoprecipitation experiment, ~30 uninjured or regenerating cmlc2:H3.3-bio hearts were pooled. Hearts were dissected from 4–6 month old adult zebrafish, the outflow track and atrium were removed, and ventricles were placed in Hanks Buffered Salt Solution containing Mg2+ and Ca2+ and 20 U/ml heparin. Hearts were resuspended in Hypotonic buffer (10 mM Tris-HCl pH 7.5, 10 mM KCl, 0.15 mM spermine and 0.5 mM spermidine), blended twice for 20 s at setting 5 on a tissue grinder from Fisher Scientific (PowerGen125) and incubated on ice for 15 minutes. Lysis buffer (10 mM TrisHCl pH 7.5, 10 mM KCl, 1% digitonin, 0.34 M sucrose, 1% albumin, 1 mM DTT, 0.15 mM spermine and 0.5 mM spermidine) was added and the homogenate was dounced with seven strokes of a type B pestle. Nuclei were spun down and treated with Mnase for 30 min at room temperature, before reactions were stopped with 5 mM EGTA. The homogenate was spun down one further time, and the supernatant was kept as the input fraction of chromatin. Supernatant (10–20 μg total chromatin) was incubated overnight with 10 μL of streptavidin beads and washed for several hours in the morning using a native ChIP wash buffer (50 mM Tris pH 7.5, 0.2 M NaCl, 5 mM EDTA pH 8.0) and subsequently eluted with 1% SDS in wash buffer. DNA was isolated and Illumina sequencing libraries were prepared for single-end sequencing as described (Bowman et al., 2013).

For ChIP against the histone H3K27Ac epitope, 1 μg antibody (Active Motif - 39133) was incubated with native chromatin over night in place of streptavidin beads. In the morning, 10 μl Protein-A beads were incubated with the mixture for three hours. Washes, elutions and library preparation were then performed identically as described above.

Zebrafish and Cardiac Injuries

Resection of ~20% of the ventricular apex was performed on zebrafish anesthetized in Tricaine and placed ventral side up on a sponge. Iridectomy scissors were used to make an incision through the skin and pericardial sac. Gentle abdominal pressure exposed the heart and ~20% of the apex was removed with scissors, penetrating the chamber lumen. A Kimwipe was used to slow bleeding and animals were revived in aquarium water (Poss et al., 2002). To genetically ablate CMs, cmlc2:CreERpd10; bactin2:loxp-mCherry-STOP-loxp-DTApd36 (Z-CAT) fish were incubated in or 0.4 μM tamoxifen for 18 hrs (Wang et al., 2011). Hearts were harvested 7 or 14 days after induction depending on the experiment. All procedures with animals were approved by the animal care and use committee at Duke University.

Histological Assays

To generate probes, we amplified anillin, ankrd1a, bach1b, endothelin2, eif4e1c and transgelin cDNA from 7 dpi regenerating hearts. cDNA was blunt ligated into a pGEMt vector, and digoxigenin-labeled RNA probes were generated using either SP6 or T7 RNA Polymerase. In situ hybridization on cryosections of 4% paraformaldehyde-fixed hearts was performed as described previously (Poss et al., 2002).

For immunofluorescence, we imaged 6 to 13 adult hearts from two stable transgenic lines for the images shown here (see Table S3 for details on transgenics). Primary and secondary antibodies used in this study were: anti-Myosin heavy chain (mouse, F59, Developmental Studies Hybridoma Bank), anti-Raldh2 (Kikuchi et al., 2011), anti-EGFP (rabbit, A11122, Life Technologies), Alexa Fluor 488 (mouse and rabbit; Life Technologies), Alexa Fluor 594 (mouse and rabbit; Life Technologies).

QUANTIFICATION AND STATISTICAL ANALYSIS

ChIPseq Data Analysis

H3.3 and H3K27Ac ChIPseq reads were aligned to the GRCz10 zebrafish genome assembly using Bowtie (Langmead et al., 2009), with parameters “-M 5 -k 1 -I 50 -X 500 --solexa-quals –best”. The numbers of aligned reads from each sample are shown in Table S7. The positions with anomalously high read counts are potential amplification artifacts and were removed from consideration (Kharchenko et al., 2008). The remaining paired-end reads were used to calculate positional coverage and fragment frequencies in 100 bp bins. These values were normalized by the corresponding library sizes. Fragment frequencies were further used to compute enrichments or ratios of normalized fragment frequencies in ChIP and input samples. Peaks in the ChIPSeq fragment distribution were identified using the SPP software package (Kharchenko et al., 2008) with default parameters (for H3.3 - sliding window of 300 bp, and Z-score cutoff of 3; for H3K27Ac - sliding window of 500 bp, and Z-score cut-off of 5). False Discovery Rate (FDR < 0.01) was used as a significance threshold in peak calling and for intersection replicate sets.

RNAseq Data Analysis

RNAseq reads were aligned to the transcriptome of the zebrafish genome (GRCz10, gene model annotation release 80) using TopHat version 2.0.13 with default parameters (Trapnell et al., 2009). The numbers of aligned reads from each sample are shown in Table S9. Reads mapping to genes were counted using HTseq-count version 0.5.4p3 with parameters ‘-s no -a 10’ (Anders et al., 2015). Differential expression analysis of the RNAseq was done using the DESeq tool (Anders and Huber, 2010).

Motif Analysis

Motif analysis was performed using MEME version 4.9.0 with parameters ‘-dna -revcomp -mod zoops –nmotifs 10’ (Bailey et al., 2009). Genomic sequences on the H3.3 enriched loci (H3.3 ChIPseq peaks) were used in the motif search. Since MEME has a limit of 100,000 characters (nucleotides) in input, the sequence set was randomly separated into subsets with maximum 100,000 nucleotides, and MEME was run for each subset. To merge the results of subsets, we categorized motifs as poly-A loci, AT-repeats, CA-repeats GA-repeats and more complex sequence motifs; motif occurrence sites in each category were subjected to MEME search. E-values and number of sites in each motif were output of the final MEME run. TOMTOM package was used to find similarity between motifs (Gupta et al., 2007). Two motifs were grouped together if the p-value was smaller than 10−7 for the shorter motifs (up to 8 bp) and 10−10 for longer motifs. FIMO package was used to find occurrences (hits) of the motifs within genomic sequences (Grant et al., 2011).

Affinity Score Comparison

Affinity scores characterizing protein affinity to enhancer sequences based on the protein motif match were calculated using Bioconductor package PWMEnrich (Stojnic et al 2015). Relative affinity for emerging and disappearing H3.3 peaks at putative enhancers was compared using the Kolmogorov–Smirnov test. The ‘regeneration specificity score’ represents the −log10 (p-value).

Comparison to Human Putative Cardiac Enhancers

To identify similarity between CREE and human p300 peaks, we aligned 10-mers of the query sequences (CREE) onto the target sequence (p300 peaks), and calculated the percent of each p300 peak covered by the 10-mers. We used 15% of coverage as a restrictive cut-off to call homology. The p-value was computed by the Mann-Whitney-Wilcoxon test or 10-mer coverage distributions vs the Poisson distribution (10,000 points with the same mean values).

DATA AND SOFTWARE AVAILABILITY

Data Resources

The RNAseq and ChIP data have been deposited in the GEO database under ID code GSE81893 (GSE81865: RNAseq, GSE81862: H3.3 ChIPseq, and GSE81863: H3K27ac ChIPseq).

Supplementary Material

1
2
3
4
5
6
7
8

Highlights.

  • A transgenic reagent for profiling histone H3.3 occupancy in cardiomyocytes

  • Identification of regulatory elements directing gene expression in cardiomyocytes

  • Elucidation of enhancer elements preferential to heart regeneration

  • Discovery of sequence motifs in H3.3-enriched genomic regions during regeneration

Acknowledgments

We thank J. Savage for preliminary bioinformatics analysis, Y.-S. Ang for comments on the manuscript, B. Bruneau and J. Kang for discussions, and J. Burris, T. Thoren, B. Thomas, and T. Loffredo for zebrafish care. This work was supported by postdoctoral fellowships to J.A.G. from NIH (F32 HL120494) and AHA (13POST16820036), a travel award to J.A.G. from the Company of Biologists, a Clinical Investigator Award to R.K. from NIH (K08 HL116485), a subaward from NIH (R01 GM098461) to M.Y.T., an AHA Merit Award to K.D.P., and a grant from NIH (R01 HL081674) to K.D.P.

Footnotes

SUPPLEMENTAL INFORMATION

Supplemental Information includes 7 figures and 7 tables.

Author Contributions Conceptualization, J.A.G. and K.D.P. Methodology, J.A.G., K.D.P., G.K. and M.Y.T. Formal Analysis, G.K. and M.Y.T. Investigation, J.A.G., N.L., J.K., and A.D. Writing – Original Draft, J.A.G.; Writing – Review & Editing, J.A.G., G.K., M.Y.T., and K.D.P. Funding Acquisition, K.D.P. Resources, M.G., M.F., R.K., and F.S.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Anders S, Huber W. Differential expression analysis for sequence count data. Genome Biol. 2010;11:R106. doi: 10.1186/gb-2010-11-10-r106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Anders S, Pyl PT, Huber W. HTSeq--a Python framework to work with high-throughput sequencing data. Bioinformatics. 2015;31:166–169. doi: 10.1093/bioinformatics/btu638. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Babaryka A, Kuhn E, Koster RW. In vivo synthesis of meganuclease for generating transgenic zebrafish Danio rerio. J Fish Biol. 2009;74:452–457. doi: 10.1111/j.1095-8649.2008.02075.x. [DOI] [PubMed] [Google Scholar]
  4. Bailey TL, Boden M, Buske FA, Frith M, Grant CE, Clementi L, Ren J, Li WW, Noble WS. MEME SUITE: tools for motif discovery and searching. Nucleic Acids Res. 2009;37:W202–208. doi: 10.1093/nar/gkp335. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Benayoun BA, Pollina EA, Ucar D, Mahmoudi S, Karra K, Wong ED, Devarajan K, Daugherty AC, Kundaje AB, Mancini E, et al. H3K4me3 breadth is linked to cell identity and transcriptional consistency. Cell. 2014;158:673–688. doi: 10.1016/j.cell.2014.06.027. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bersell K, Arab S, Haring B, Kuhn B. Neuregulin1/ErbB4 signaling induces cardiomyocyte proliferation and repair of heart injury. Cell. 2009;138:257–270. doi: 10.1016/j.cell.2009.04.060. [DOI] [PubMed] [Google Scholar]
  7. Bowman SK, Simon MD, Deaton AM, Tolstorukov M, Borowsky ML, Kingston RE. Multiplexed Illumina sequencing libraries from picogram quantities of DNA. BMC Genomics. 2013;14:466. doi: 10.1186/1471-2164-14-466. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Boyle AP, Davis S, Shulha HP, Meltzer P, Margulies EH, Weng Z, Furey TS, Crawford GE. High-resolution mapping and characterization of open chromatin across the genome. Cell. 2008;132:311–322. doi: 10.1016/j.cell.2007.12.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Buenrostro JD, Giresi PG, Zaba LC, Chang HY, Greenleaf WJ. Transposition of native chromatin for fast and sensitive epigenomic profiling of open chromatin, DNA-binding proteins and nucleosome position. Nat Methods. 2013;10:1213–1218. doi: 10.1038/nmeth.2688. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Duncan EM, Chitsazan AD, Seidel CW, Sanchez Alvarado A. Set1 and MLL1/2 Target Distinct Sets of Functionally Different Genomic Loci In Vivo. Cell Rep. 2015;13:2741–2755. doi: 10.1016/j.celrep.2015.11.059. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Fang Y, Gupta V, Karra R, Holdway JE, Kikuchi K, Poss KD. Translational profiling of cardiomyocytes identifies an early Jak1/Stat3 injury response required for zebrafish heart regeneration. Proc Natl Acad Sci U S A. 2013;110:13416–13421. doi: 10.1073/pnas.1309810110. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Fivaz J, Bassi MC, Pinaud S, Mirkovitch J. RNA polymerase II promoter-proximal pausing upregulates c-fos gene expression. Gene. 2000;255:185–194. doi: 10.1016/s0378-1119(00)00340-1. [DOI] [PubMed] [Google Scholar]
  13. Gemberling M, Karra R, Dickson AL, Poss KD. Nrg1 is an injury-induced cardiomyocyte mitogen for the endogenous heart regeneration program in zebrafish. Elife. 2015;4 doi: 10.7554/eLife.05871. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Giresi PG, Kim J, McDaniell RM, Iyer VR, Lieb JD. FAIRE (Formaldehyde-Assisted Isolation of Regulatory Elements) isolates active regulatory elements from human chromatin. Genome Res. 2007;17:877–885. doi: 10.1101/gr.5533506. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Goldberg AD, Banaszynski LA, Noh KM, Lewis PW, Elsaesser SJ, Stadler S, Dewell S, Law M, Guo X, Li X, et al. Distinct factors control histone variant H3.3 localization at specific genomic regions. Cell. 2010;140:678–691. doi: 10.1016/j.cell.2010.01.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Grant CE, Bailey TL, Noble WS. FIMO: scanning for occurrences of a given motif. Bioinformatics. 2011;27:1017–1018. doi: 10.1093/bioinformatics/btr064. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Gupta S, Stamatoyannopoulos JA, Bailey TL, Noble WS. Quantifying similarity between motifs. Genome Biol. 2007;8:R24. doi: 10.1186/gb-2007-8-2-r24. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Gupta V, Gemberling M, Karra R, Rosenfeld GE, Evans T, Poss KD. An injury-responsive gata4 program shapes the zebrafish cardiac ventricle. Curr Biol. 2013;23:1221–1227. doi: 10.1016/j.cub.2013.05.028. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Gupta V, Poss KD. Clonally dominant cardiomyocytes direct heart morphogenesis. Nature. 2012;484:479–484. doi: 10.1038/nature11045. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Heallen T, Zhang M, Wang J, Bonilla-Claudio M, Klysik E, Johnson RL, Martin JF. Hippo pathway inhibits Wnt signaling to restrain cardiomyocyte proliferation and heart size. Science. 2011;332:458–461. doi: 10.1126/science.1199010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Ho JW, Jung YL, Liu T, Alver BH, Lee S, Ikegami K, Sohn KA, Minoda A, Tolstorukov MY, Appert A, et al. Comparative analysis of metazoan chromatin organization. Nature. 2014;512:449–452. doi: 10.1038/nature13415. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Jin C, Zang C, Wei G, Cui K, Peng W, Zhao K, Felsenfeld G. H3.3/H2A.Z double variant-containing nucleosomes mark 'nucleosome-free regions' of active promoters and other regulatory regions. Nat Genet. 2009;41:941–945. doi: 10.1038/ng.409. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Jopling C, Sune G, Faucherre A, Fabregat C, Izpisua Belmonte JC. Hypoxia induces myocardial regeneration in zebrafish. Circulation. 2012;126:3017–3027. doi: 10.1161/CIRCULATIONAHA.112.107888. [DOI] [PubMed] [Google Scholar]
  24. Kang J, Hu J, Karra R, Dickson AL, Tornini VA, Nachtrab G, Gemberling M, Goldman JA, Black BL, Poss KD. Modulation of tissue repair by regeneration enhancer elements. Nature. 2016;532:201–206. doi: 10.1038/nature17644. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Karra R, Knecht AK, Kikuchi K, Poss KD. Myocardial NF-kappaB activation is essential for zebrafish heart regeneration. Proc Natl Acad Sci U S A. 2015;112:13255–13260. doi: 10.1073/pnas.1511209112. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Kharchenko PV, Tolstorukov MY, Park PJ. Design and analysis of ChIP-seq experiments for DNA-binding proteins. Nat Biotechnol. 2008;26:1351–1359. doi: 10.1038/nbt.1508. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Kikuchi K, Holdway JE, Major RJ, Blum N, Dahn RD, Begemann G, Poss KD. Retinoic acid production by endocardium and epicardium is an injury response essential for zebrafish heart regeneration. Dev Cell. 2011;20:397–404. doi: 10.1016/j.devcel.2011.01.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Kikuchi K, Holdway JE, Werdich AA, Anderson RM, Fang Y, Egnaczyk GF, Evans T, Macrae CA, Stainier DY, Poss KD. Primary contribution to zebrafish heart regeneration by gata4(+) cardiomyocytes. Nature. 2010;464:601–605. doi: 10.1038/nature08804. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Kubin T, Poling J, Kostin S, Gajawada P, Hein S, Rees W, Wietelmann A, Tanaka M, Lorchner H, Schimanski S, et al. Oncostatin M is a major mediator of cardiomyocyte dedifferentiation and remodeling. Cell Stem Cell. 2011;9:420–432. doi: 10.1016/j.stem.2011.08.013. [DOI] [PubMed] [Google Scholar]
  30. Langmead B, Trapnell C, Pop M, Salzberg SL. Ultrafast and memory-efficient alignment of short DNA sequences to the human genome. Genome Biol. 2009;10:R25. doi: 10.1186/gb-2009-10-3-r25. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Lepilina A, Coon AN, Kikuchi K, Holdway JE, Roberts RW, Burns CG, Poss KD. A dynamic epicardial injury response supports progenitor cell activity during zebrafish heart regeneration. Cell. 2006;127:607–619. doi: 10.1016/j.cell.2006.08.052. [DOI] [PubMed] [Google Scholar]
  32. Mahmoud AI, Kocabas F, Muralidhar SA, Kimura W, Koura AS, Thet S, Porrello ER, Sadek HA. Meis1 regulates postnatal cardiomyocyte cell cycle arrest. Nature. 2013;497:249–253. doi: 10.1038/nature12054. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. May D, Blow MJ, Kaplan T, McCulley DJ, Jensen BC, Akiyama JA, Holt A, Plajzer-Frick I, Shoukry M, Wright C, et al. Large-scale discovery of enhancers from human heart tissue. Nat Genet. 2012;44:89–93. doi: 10.1038/ng.1006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Mito Y, Henikoff JG, Henikoff S. Histone replacement marks the boundaries of cis-regulatory domains. Science. 2007;315:1408–1411. doi: 10.1126/science.1134004. [DOI] [PubMed] [Google Scholar]
  35. Porrello ER, Mahmoud AI, Simpson E, Hill JA, Richardson JA, Olson EN, Sadek HA. Transient regenerative potential of the neonatal mouse heart. Science. 2011;331:1078–1080. doi: 10.1126/science.1200708. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Poss KD, Wilson LG, Keating MT. Heart regeneration in zebrafish. Science. 2002;298:2188–2190. doi: 10.1126/science.1077857. [DOI] [PubMed] [Google Scholar]
  37. Rada-Iglesias A, Bajpai R, Swigut T, Brugmann SA, Flynn RA, Wysocka J. A unique chromatin signature uncovers early developmental enhancers in humans. Nature. 2011;470:279–283. doi: 10.1038/nature09692. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Romanoski CE, Glass CK, Stunnenberg HG, Wilson L, Almouzni G. Epigenomics: Roadmap for regulation. Nature. 2015;518:314–316. doi: 10.1038/518314a. [DOI] [PubMed] [Google Scholar]
  39. Sander V, Sune G, Jopling C, Morera C, Izpisua Belmonte JC. Isolation and in vitro culture of primary cardiomyocytes from adult zebrafish hearts. Nat Protoc. 2013;8:800–809. doi: 10.1038/nprot.2013.041. [DOI] [PubMed] [Google Scholar]
  40. Senyo SE, Steinhauser ML, Pizzimenti CL, Yang VK, Cai L, Wang M, Wu TD, Guerquin-Kern JL, Lechene CP, Lee RT. Mammalian heart renewal by pre-existing cardiomyocytes. Nature. 2013;493:433–436. doi: 10.1038/nature11682. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Stojnic R, Diaz D. PWMEnrich: PWM enrichment analysis. 2015. R package version 4.10.0. [Google Scholar]
  42. Trapnell C, Pachter L, Salzberg SL. TopHat: discovering splice junctions with RNA-Seq. Bioinformatics. 2009;25:1105–1111. doi: 10.1093/bioinformatics/btp120. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Wang J, Panakova D, Kikuchi K, Holdway JE, Gemberling M, Burris JS, Singh SP, Dickson AL, Lin YF, Sabeh MK, et al. The regenerative capacity of zebrafish reverses cardiac failure caused by genetic cardiomyocyte depletion. Development. 2011;138:3421–3430. doi: 10.1242/dev.068601. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Zhang L, Maddox AS. Anillin. Curr Biol. 2010;20:R135–136. doi: 10.1016/j.cub.2009.12.017. [DOI] [PubMed] [Google Scholar]
  45. Zolk O, Frohme M, Maurer A, Kluxen FW, Hentsch B, Zubakov D, Hoheisel JD, Zucker IH, Pepe S, Eschenhagen T. Cardiac ankyrin repeat protein, a negative regulator of cardiac gene expression, is augmented in human heart failure. Biochem Biophys Res Commun. 2002;293:1377–1382. doi: 10.1016/S0006-291X(02)00387-X. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS848174-supplement-1.pdf (4.6MB, pdf)
2
NIHMS848174-supplement-2.xlsx (53.4KB, xlsx)
3
NIHMS848174-supplement-3.xlsx (31.7KB, xlsx)
4
NIHMS848174-supplement-4.xlsx (34KB, xlsx)
5
NIHMS848174-supplement-5.xlsx (1.4MB, xlsx)
6
NIHMS848174-supplement-6.xlsx (44.6KB, xlsx)
7
NIHMS848174-supplement-7.xlsx (146.4KB, xlsx)
8
NIHMS848174-supplement-8.xlsx (37.6KB, xlsx)

RESOURCES