AP-1 Contributes to Chromatin Accessibility to Promote Sarcomere Disassembly and Cardiomyocyte Protrusion During Zebrafish Heart Regeneration

Abstract

Rationale:

The adult human heart is an organ with low regenerative potential. Heart failure following acute myocardial infarction is a leading cause of death due to the inability of cardiomyocytes to proliferate and replenish lost cardiac muscle. While the zebrafish has emerged as a powerful model to study endogenous cardiac regeneration, the molecular mechanisms by which cardiomyocytes respond to damage by disassembling sarcomeres, proliferating, and repopulating the injured area remain unclear. Furthermore, we are far from understanding the regulation of the chromatin landscape and epigenetic barriers that must be overcome for cardiac regeneration to occur.

Objective:

To identify transcription factor regulators of the chromatin landscape which promote cardiomyocyte regeneration in zebrafish, and investigate their function.

Methods and Results:

Using the Assay for Transposase-Accessible Chromatin coupled to high-throughput sequencing (ATAC-Seq), we first find that the regenerating cardiomyocyte chromatin accessibility landscape undergoes extensive changes following cryoinjury, and that AP-1 binding sites are the most highly enriched motifs in regions that gain accessibility during cardiac regeneration. Furthermore, using bioinformatic and gene expression analyses, we find that the AP-1 response in regenerating adult zebrafish cardiomyocytes is largely different from the response in adult mammalian cardiomyocytes. Using a cardiomyocyte-specific dominant negative approach, we show that blocking AP-1 function leads to defects in cardiomyocyte proliferation as well as decreased chromatin accessibility at the fbxl22 and ilk loci, which regulate sarcomere disassembly and cardiomyocyte protrusion into the injured area, respectively. We further show that overexpression of the AP-1 family members Junb and Fosl1 can promote changes in mammalian cardiomyocyte behavior in vitro.

Conclusions:

AP-1 transcription factors play an essential role in the cardiomyocyte response to injury by regulating chromatin accessibility changes, thereby allowing the activation of gene expression programs that promote cardiomyocyte dedifferentiation, proliferation, and protrusion into the injured area.

Graphical Abstract

Heart failure following acute myocardial infarction is a leading cause of death due to the inability of mammalian CMs to proliferate and replenish lost cardiac muscle. There remains an urgent need to develop therapies that facilitate survival and regeneration of healthy heart tissue after ischemia. Here, we study the adult zebrafish whose heart has the endogenous ability to regenerate following cryoinjury. We provide the first genome-wide chromatin accessibility analysis of zebrafish border zone CMs and find that their chromatin accessibility landscape undergoes extensive changes in response to injury. Furthermore, we find that AP-1 transcription factors contribute to regulation of chromatin remodeling, and that AP-1 function is essential for multiple aspects of CM regeneration, including dedifferentiation, sarcomere disassembly, proliferation, and protrusion into the injured area. We also find that the AP-1 response in regenerating zebrafish cardiomyocytes is largely different from the response in adult mammalian cardiomyocytes following injury. However, one can promote proliferation and protrusion of mammalian cardiomyocytes in culture by overexpressing the AP-1 family members Junb and Fosl1. These results suggest that manipulating the AP-1 response in mammalian CMs may help to promote cardiac repair and regeneration.

INTRODUCTION

Acute myocardial infarction is a leading cause of morbidity and mortality in industrialized nations worldwide due to the inability of the adult mammalian heart to replenish lost cardiac muscle following injury.^1,2 The murine heart has been shown to regenerate following injury, but only during embryogenesis and in neonatal stages.^3,4 Following postnatal day 7 (P7), the regenerative capacity of the murine heart drops drastically. However, there is evidence that adult human cardiomyocytes undergo low levels of proliferation throughout life,⁵ and manipulation of signaling pathways, cell cycle regulators, microRNA expression, and extracellular matrix protein levels, can improve myocardial function following ischemia in the adult mouse heart.^6–12 These results have generated hope that endogenous regeneration and repair pathways in the adult human heart can be stimulated following myocardial infarction to improve functional outcome.

The zebrafish has emerged as an attractive model system to study cardiac regeneration due to its high regenerative capacity following injury.^13–17 While significant progress has been made in understanding the process of heart regeneration in zebrafish, the molecular mechanisms by which cardiomyocytes (CMs) sense injury and respond by disassembling sarcomeres, proliferating, and repopulating the injured area remain unclear. Furthermore, we are only starting to understand the role of chromatin accessibility and epigenetic barriers that must be overcome in zebrafish CMs for successful heart regeneration. Through histone H3 lysine (K) 27 acetylation (H3K27ac) profiling in whole ventricles and histone H3.3 profiling specifically in CMs, tissue-specific enhancer elements have been discovered that are able to drive the expression of GFP and other genes in the regenerating heart.^18,19 Moreover, through dominant negative approaches, it was shown that brahma-related gene 1 (Brg1, or Smarca4), a central catalytic subunit of the SWItch/sucrose non-fermentable (SWI/SNF) ATP-dependent chromatin-remodeling complex, is essential for CM proliferation and subsequent heart regeneration.²⁰ Recently, deposition of H3K27 methylation at cytoskeletal genes has been shown to regulate sarcomere disassembly and invasion of wound border CMs into the injured area.²¹ While these studies indicate that CM-specific changes in chromatin accessibility are necessary for cardiac regeneration, it remains unclear which transcription factors are responding to injury and recruiting chromatin-remodeling complexes to regulate the chromatin landscape.

The activator protein 1 (AP-1) transcription factor genes constitute a family of immediate-early response genes whose transcription is activated in response to a number of extracellular stimuli, including cytokines, growth factors, pathogens, and cellular stress.²² The activation of Jun/Fos (AP-1) genes has been shown to occur rapidly in response to stimuli and AP-1 transcription factors have been reported to regulate a number of targets involved in cell survival, apoptosis, cell differentiation, and migration in a context-dependent manner.^23,24 In more recent years, it has been shown that in addition to their role as transcriptional activators, AP-1 transcription factors regulate multiple aspects of chromatin biology. For example, in a murine mammary epithelial cell line, AP-1 binding to DNA was necessary to increase chromatin accessibility in order for glucocorticoid receptor to bind to its target genomic regions upon dexamethasone treatment.²⁵ Further evidence that AP-1 transcription factors regulate chromatin accessibility comes from studies in primary mouse embryonic fibroblasts. It was shown that all FOS family member proteins physically interact with members of the BAF ATP-dependent chromatin-remodeling complex, and that FOS, FOSB, and JUNB contribute to recruitment of BAF complex members to growth factor-inducible enhancer regions to promote chromatin accessibility.²⁶ These data indicate that AP-1 transcription factors regulate the chromatin landscape and promote gene expression programs in response to multiple extracellular stressors.

A very recent study has shown that AP-1 motifs are present in chromatin regions that gain accessibility in adult murine border zone CMs in response to myocardial infarction (MI).²⁷ However, the functional role of AP-1 transcription factors in the response to MI remains unclear. Here, we show that AP-1 transcription factors play an essential role in cardiomyocytes during adult zebrafish heart regeneration. We find the significant presence of AP-1 binding motifs in genomic regions that gain chromatin accessibility in regenerating CMs. Through the use of a dominant negative approach, we show that AP-1 transcription factor function is necessary for CM proliferation, and furthermore, that AP-1 function contributes to chromatin remodeling at genes that play a role in sarcomere disassembly and CM protrusion into the injured area. Altogether, our data indicate that AP-1 transcription factors respond to tissue injury, contribute to chromatin accessibility, and regulate gene expression programs to promote multiple aspects of CM regeneration.

METHODS

Materials and data that support the findings of this study are available from the corresponding authors upon reasonable request.

Detailed materials and methods can be found in the Online Supplement and the Major Resources Table.

Zebrafish lines and constructs.

All zebrafish husbandry was performed under standard conditions in accordance with institutional (MPG) and national ethical and animal welfare guidelines.

The dominant negative A-Fos construct was generated by fusion PCR of a FLAG tag (DYKDDDK) in frame with the acidic extension (DLEQRAEELARENEELEKEAEELEQELAELE) and the bZIP dimerization domain of zebrafish Fosab (LQAETDQLEDEKSALQNDIANLLKEKERLEFILAAHKPICKIPAD).²⁸ This fragment was then fused with a P2A-tagBFP coding sequence and cloned into a Tol2 transgenesis vector downstream of the 3.5 kb ubb promoter and a loxp-flanked EGFP coding sequence. The junba and junbb overexpression constructs were generated by PCR amplification of the junba or junbb open reading frame from zebrafish cDNA and cloning into a Tol2 transgenesis vector downstream of the 3.5 kb ubb promoter and a loxp-flanked EGFP coding sequence. Transgenic lines were generated by injecting 20 pg of plasmid and 75 pg of Tol2 mRNA into one-cell stage zebrafish embryos. Founder fish were identified and propagated to establish Tg(ubb:loxp-EGFP-loxp-P2A-A-Fos-P2A-tagBFP-HA)^bns335, Tg(ubb:loxp-EGFP-loxp-P2A-junba-P2A-tagBFP-HA)^bns336, and Tg(ubb:loxp-EGFP-loxp-P2A-junbb-P2A-tagBFP-HA)^bns337 stable lines.

Data availability.

ATAC-Seq data from this study have been deposited in the Gene Expression Omnibus (GEO) database under the accession number GSE130940.

RESULTS

The cardiomyocyte chromatin accessibility landscape undergoes substantial changes during regeneration.

In order to determine the CM chromatin accessibility changes that occur following cardiac injury, we performed the Assay for Transposase-Accessible Chromatin coupled to high-throughput sequencing (ATAC-Seq)²⁹ in regenerating versus uninjured CMs. To isolate regenerating CMs, we used the Tg(gata4:EGFP) line, whose expression is commonly used as a CM dedifferentiation marker and gata4:EGFP+ CMs have been shown by lineage tracing experiments to give rise to the majority of regenerated myocardium following apical resection in the adult zebrafish heart (Figure 1a).^30,31 gata4:GFP+ CMs were sorted by FACS at 4 days post cryoinjury (dpci), a timepoint when one first observes Tg(gata4:EGFP) expression in wound border CMs. DsRed+ CMs from uninjured Tg(myl7:nucDsRed)³² ventricles were used as a control. Comparison of chromatin accessibility between uninjured and regenerating CMs revealed a large number of genomic regions (n=1,507) that became more accessible in regenerating CMs and a smaller number of genomic regions (n=739) that were more accessible in uninjured CMs (fold change ≥ 2, Figure 1b and Online Table I). Gene ontology analysis of genes near regions whose chromatin was more accessible in uninjured CMs revealed an enrichment in biological processes such as myofibril assembly (Online Figure Ia). This observation is in line with previous reports that CMs near the wound border undergo a process of dedifferentiation, accompanied by sarcomere disorganization.^31,33 On the other hand, gene ontology analysis of genes near regions whose chromatin was more accessible in regenerating CMs revealed an enrichment in biological processes such as regulation of cell migration, regulation of response to stimulus, and cytoskeleton organization. Pathway analysis of genes near regions whose chromatin was more accessible in regenerating CMs revealed an enrichment of genes involved in PDGF, Wnt, and Integrin signaling pathways (Online Figure Ia). Altogether, these data provide insight into the CM chromatin accessibility landscape during the response to cryoinjury. It is likely that these chromatin accessibility changes are a prerequisite for the activation of multiple genes and signaling pathways, some of which have been previously shown to be involved in zebrafish heart regeneration.^34–36

Figure 1. ATAC-Seq analysis reveals a putative role for AP-1 transcription factors in regulating chromatin dynamics in regenerating cardiomyocytes.

(a) Mef2/GFP immunostaining of a Tg(gata4:EGFP) adult ventricle section at 4 dpci. White dashed line marks the wound border. (b) Volcano plot comparing chromatin accessibility peaks in Tg(gata4:EGFP)+ CMs at 4 dpci to Tg(myl7:nucDsRed)+ uninjured CMs. Blue points are more accessible in Tg(gata4:EGFP)+ CMs and red points are more accessible in Tg(myl7:nucDsRed)+ uninjured CMs (fold change ≥ 2, padj ≤ 0.05). p-values were calculated using the Wald significance test, and adjustment for multiple comparisons of 20,540 total ATAC-Seq peaks was performed using the Benjamini-Hochberg method. (c) Analysis of motif enrichment in genomic regions with increased accessibility in Tg(gata4:EGFP)+ CMs reveals the significant enrichment of AP-1 motifs, found in 56% of regions with increased chromatin accessibility. (d) Genomic distribution of ATAC-Seq peaks that are more accessible in Tg(gata4:EGFP)+ CMs and contain an AP-1 motif. (e) RT-qPCR analysis of jun/fos mRNA levels in CMs isolated by density gradient centrifugation in uninjured, 1 dpci, 3 dpci, 1 dps, and 3 dps ventricles (n=3–5 pools of 6 hearts each). Data are represented as mean ± std. dev.; p-values were calculated using ordinary one-way ANOVA and adjusted for comparison between 5 groups with the Bonferroni multiple comparisons test. (f) Volcano plot depicting results from TOBIAS (Transcription factor Occupancy prediction By Investigating ATAC-Seq Signal). Blue points depict motifs that are enriched and have a TF footprint in Tg(gata4:EGFP)+ CMs, while red points depict motifs that are enriched and have a TF footprint in Tg(myl7:nucDsRed)+ uninjured CMs. (g) Aggregate ATAC-Seq footprints for 14,394 occurrences of the JUNB motif in Tg(myl7:nucDsRed)+ uninjured CMs (red) and Tg(gata4:EGFP)+ CMs at 4 dpci (blue). Dashed lines represent the borders of the 11 base pair JUNB motif. Negative Tn5 insertions represent protection due to putative protein binding. padj: adjusted p-value; BZ: border zone; dps: days post sham; dpci: days post cryoinjury; Uninj: uninjured. Scale bar: 100 μm.

Motif analysis suggests a role for AP-1 transcription factors in regulating the cardiomyocyte chromatin accessibility landscape.

To identify the transcription factors that may recruit chromatin remodeling enzymes to control the chromatin landscape in regenerating CMs, we performed transcription factor motif analysis in the 1,507 regions that were more accessible in Tg(gata4:EGFP)+ CMs. The most highly enriched motifs included those bound by multiple members of the AP-1 transcription factor family (Online Figure Ib and Online Table II). AP-1 motifs were found in 56% of the regions that became more accessible in regenerating CMs (Figure 1c). When we examined the genomic distribution of these AP-1 motifs, we found that a small proportion of them were found in promoter regions of genes (Figure 1d, between −5 kb and +500 bp from the transcriptional start site (TSS)). The majority of these AP-1 motifs were found in gene bodies or intergenic regions that became more accessible in regenerating CMs (Figure 1d). These observations suggest that AP-1 transcription factors regulate the chromatin accessibility landscape in regenerating CMs, and that this regulation occurs largely outside of the promoter region of genes.

The AP-1 family of transcription factors is comprised of multiple Jun and Fos members. In order to determine which family members may be involved in regulating chromatin accessibility during zebrafish cardiac regeneration, we examined previously published RNA-Seq data from zebrafish ventricles at multiple time points following cryoinjury.³⁷ First, we excluded genes from further analysis that were either lowly expressed in the regenerating zebrafish heart (i.e., fosaa and fosl1b) or downregulated upon cryoinjury (i.e., jun and jund). The upregulation of remaining jun/fos family members was confirmed by RT-qPCR from whole ventricles (Online Figure Ic). Furthermore, RT-qPCR analysis from isolated CMs revealed the upregulation of junba, junbb, fosab, and fosl1a at 1 and 3 dpci (Figure 1e). in situ hybridization on sections from cryoinjured adult ventricles revealed that junba and junbb genes are expressed in wound border CMs at 1 and 3 dpci (Online Figure Id). These data indicate that multiple jun/fos genes are upregulated in CMs in response to cryoinjury of the adult heart.

In order to determine whether AP-1 transcription factors were directly bound to AP-1 motifs that arise in regenerating CMs, we attempted to find antibodies suitable for chromatin immunoprecipitation. We tested multiple JUNB antibodies and were unable to immunoprecipitate zebrafish Junba/bb (Online Figures IIa–IIc). Therefore, we performed further bioinformatics analysis to predict transcription factor (TF) occupancy in our ATAC-Seq dataset (TOBIAS, https://github.com/loosolab/TOBIAS). This analysis corrects ATAC-Seq reads for Tn5 transposase sequence bias to estimate the presence of footprints that arise when DNA is protected from transposase insertion due to the putative binding of a TF to DNA. The analysis then combines differences in chromatin accessibility and footprints between two datasets to estimate differentially bound TF motifs. When we compared chromatin accessibility data from regenerating and uninjured CMs, we found a significant gain of putative TF binding at AP-1 motifs in accessible chromatin from regenerating CMs (Figure 1f). The high differential binding score indicates that the AP-1 binding sites, which became more accessible in regenerating CMs, have increased footprint scores due to the putative binding of an AP-1 TF. Transcription factor motifs that had differentially higher footprint scores in uninjured CMs included CUX1, whose role in cardiomyocytes is not well-studied, and mesodermal transcription factors, such as MIXL1 (Figure 1f). When we looked specifically at JUNB motifs across all ATAC-Seq peaks, we found the significant presence of a TF footprint in gata4:EGFP+ regenerating CMs when compared to myl7:nucDsRed uninjured CMs (Figure 1g). While we cannot rule out that another AP-1 factor may be binding, our data highly suggest that Junb binds to AP-1 motifs that become more accessible in regenerating CMs in adult zebrafish.

A recent study has shown that AP-1 motifs are also enriched in border zone (BZ) CMs of the non-regenerative adult mouse heart following MI at 4 and 7 days post MI.²⁷ We were interested in determining whether the AP-1 response was similar between regenerative adult zebrafish CMs and non-regenerative adult mouse CMs. Therefore, we subjected the raw ATAC-Seq data from mouse to a bioinformatics pipeline similar to the one we used for our zebrafish CM ATAC-Seq dataset. We first performed a liftover of mouse AP-1-containing peaks to the zebrafish genome, but lost most information due to the poor sequence conservation between genomes. Instead, we assigned each AP-1-containing peak to a proximal promoter or gene body and excluded intergenic regions from our analysis. Notably, of the 573 zebrafish peaks that became more accessible in regenerating CMs (fold change ≥ 2 when compared to uninjured) and contained an AP-1 motif, only 22 (3.8%) had a corresponding AP-1-containing peak in an orthologous gene in adult mouse CMs (fold change ≥ 2 when compared to sham, Online Figures IId and IIe). We found that this overlap was statistically significant with a ranked p-value of 0.0073 after 10,000 iterations of the same number of randomly sampled genes that had at least one associated peak in zebrafish (total of 10,219, Online Figure IId). Notably, the overlapping peaks were found near genes involved in ER stress and autophagy (Hspa5, Ralgds, Vmp1, Sbf2, and Ppp1r13b),^38–42 anti-apoptotic genes (Bcl2),⁴³ and genes mediating myofibril repair (Fmnl2).⁴⁴ This observation suggests that although the AP-1 response may help to promote CM survival following injury in both species, there are still many AP-1-containing peaks in regenerative adult zebrafish CMs that largely differ from those in adult mouse CMs. Of note, the AP-1 family members that are expressed in zebrafish CMs (i.e., junba/bb and fosl1a, Figure 1e) are not highly expressed (Junb) or upregulated (Fosl1) in mouse BZ CMs following myocardial infarction when compared to sham.²⁷

AP-1 transcription factor function in cardiomyocytes is necessary for zebrafish heart regeneration.

We next sought to determine whether AP-1 transcription factors function during CM regeneration in zebrafish. Due to the expression of multiple jun/fos genes in CMs during regeneration, we generated a dominant negative transgenic line to inhibit all AP-1 transcription factor function in a tissue-specific manner. We utilized a previously published dominant negative, A-Fos, which has been shown in several settings to inhibit AP-1 transcription factor function.^25,28,45,46 A-Fos was placed downstream of a loxP-flanked stop cassette and its expression was driven by Cre-mediated recombination specifically in CMs through the use of a tamoxifen-inducible Tg(myl7:CreERT2) line³¹ (Figures 2a and 2b). 4-hydroxytamoxifen (4-HT) was administered prior to cryoinjury and expression of the dominant negative A-Fos cassette was confirmed by immunostaining (Online Figure IIIa). We performed Acid Fuchsin Orange G (AFOG) staining and found that at 45 and 90 dpci, a large fibrotic scar was still present in Tg(ubb:loxP-EGFP-loxP-A-Fos-P2A-tagBFP-HA); Tg(myl7:CreERT2) ventricles (hereafter referred to as CM:A-Fos (+)) when compared to control Tg(ubb:loxP-EGFP-loxP-A-Fos-P2A-tagBFP-HA) (hereafter referred to as CM:A-Fos (−)) ventricles (Figure 2c). This observation indicates that AP-1 function is necessary for cardiac regeneration. We investigated early processes that occur following cryoinjury, including immune cell recruitment and revascularization, which have been shown to be necessary for heart regeneration,^37,47,48 and found them to be unaffected in CM:A-Fos (+) ventricles when compared to control (Online Figures IIIb and IIIc). We then proceeded to assess multiple hallmarks of CM regeneration to understand the specific role of AP-1 transcription factors in CMs following cryoinjury. We first assessed CM dedifferentiation at 7 dpci by immunostaining for embCMHC^49,50 and found a dramatic reduction in CM:A-Fos (+) ventricles when compared to control (Figure 2d). Furthermore, we quantified CM proliferation at the wound border through the use of PCNA/Mef2 co-immunostaining and observed a significant decrease in CM proliferation in CM:A-Fos (+) ventricles when compared to control (Figures 2d and 2e). Altogether, our data indicate that AP-1 transcription factor function is necessary for CM dedifferentiation and proliferation, as well as for subsequent scar resolution.

Figure 2. AP-1 transcription factor function is necessary for zebrafish heart regeneration.

(a) Schematic illustrating the Tg(ubb:loxP-EGFP-loxP-AFos-P2A-tagBFP-HA) and Tg(myl7:CreERT2) lines. (b) Schematic illustrating the 4-HT injection and cryoinjury scheme to assess CM dedifferentiation/proliferation and scar resolution. (c) AFOG staining of sections from CM:A-Fos (−) and CM:A-Fos (+) ventricles at 45 (n=4–5) and 90 (n=4) dpci. Bar graph on the right depicts quantification of scar area relative to the size of the ventricle. Data are represented as mean ± std. dev.; p-values were calculated using the unpaired t-test. (d) F-actin/embCMHC and PCNA/Mef2 staining of sections from CM:A-Fos (−) and CM:A-Fos (+) ventricles at 7 dpci (n=6). White dashed lines mark the wound border. Yellow boxes denote the areas shown in zoomed images. (e) Quantification of PCNA+ CMs compared to total CM number within 100 μm of the wound border in CM:A-Fos (−) and CM:A-Fos (+) ventricles at 7 dpci (n=6). Data are represented as mean ± std. dev.; p-value was calculated using the unpaired t-test. 4-HT: 4-hydroxytamoxifen; CI: cryoinjury; Dediff: dedifferentiation; Prolif: proliferation. Scale bars: 100 μm in whole ventricle images, 20 μm in zoomed images.

AP-1 transcription factor function is necessary for sarcomere disassembly and cardiomyocyte protrusion into the injured area.

Previous studies have implicated sarcomere disassembly²¹ and CM protrusion/migration into the injured area^12,51 during heart regeneration. We assessed sarcomere disassembly by immunostaining for α-Actinin at 7 dpci and found that in control hearts, the intensity of α-Actinin staining decreased in the distal portion of the CMs protruding into the injured area as compared to the proximal portion (Figures 3a and 3b). Blocking AP-1 function in CMs led to a defect in sarcomere disassembly as evidenced by an increase in α-Actinin intensity in the distal portion of CMs from CM:A-Fos (+) ventricles when compared to CM:A-Fos (−) ventricles (Figures 3a and 3b). Furthermore, when we quantified CM protrusion into the injured area, we found a decrease in the total number and length of CM protrusions in CM:A-Fos (+) ventricles when compared to control (Figures 3c and 3d). Together, these data indicate that AP-1 transcription factors play a crucial role in CM regeneration and regulate key CM processes, including sarcomere disassembly and CM protrusion into the injured area.

Figure 3. Sarcomere disassembly and cardiomyocyte protrusion into the injured area are disrupted upon cardiomyocyte-specific A-Fos overexpression.

(a) α-Actinin and F-actin localization at the wound border of 7 dpci CM:A-Fos (+) versus control CM:A-Fos (−) ventricles from staining of sections. Yellow dashed lines indicate axis of measurement in (b). (b) Normalized α-Actinin intensity along the length of protruding CMs (along the proximal (P) to distal (D) axis) in CM:A-Fos (+) and CM:A-Fos (−) ventricles at 7 dpci; n=32 and 59 CMs from a total of 11 CM:A-Fos (−) and 12 CM:A-Fos (+) ventricles, respectively. Data are represented as mean ± s.e.m.; p-values were calculated using the Mann-Whitney test and adjusted for multiple comparisons between 20 groups using the Bonferroni correction. (c) Maximum intensity projections of F-actin staining in thick sections from CM:A-Fos (+) versus control CM:A-Fos (−) ventricles at 7 dpci. Yellow arrowheads point to CM protrusions and yellow boxes denote the areas shown in zoomed images. (d) Quantification of the number of CM protrusions per 100 μm of wound border (left) and length of CM protrusions (right) in CM:A-Fos (+) (n=10) versus control CM:A-Fos (−) (n=6) ventricles at 7 dpci. Data are represented as mean ± std. dev.; p-values were calculated using the unpaired t-test (number of CM protrusions) or the Mann-Whitney test (length of CM protrusions). P: proximal; D: distal. Scale bars: 100 μm in whole ventricle images, 20 μm in zoomed images.

AP-1 transcription factors contribute to remodeling of the cardiomyocyte chromatin landscape after injury.

In order to identify target genes that contribute to the inability of CM:A-Fos (+) hearts to regenerate, we performed additional ATAC-Seq analyses. Due to the defects in CM dedifferentiation in CM:A-Fos (+) hearts (Figure 2d), and thus a likely reduction in Tg(gata4:EGFP) expression²¹, we manually dissected the wound border area and purified CMs by density gradient centrifugation at 4 dpci (Figure 4a). Comparison of chromatin accessibility and transcription factor occupancy prediction in CM:A-Fos (+) and CM:A-Fos (−) CMs revealed a decrease in the number of genomic regions that are accessible and contain AP-1 motif footprints that are protected from transposase insertion in CM:A-Fos (+) CMs (Figure 4b and Online Table I). This observation suggests that blocking AP-1 function through the dominant negative A-Fos disrupts the ability of these TFs to regulate chromatin accessibility and, furthermore, points to the specificity of the dominant-negative A-Fos. However, comparison of chromatin accessibility at regions that became more accessible in Tg(gata4:EGFP)+ regenerating CMs (n=1,507) revealed no noticeable difference in average accessibility in CM:A-Fos (+) CMs compared to control (Online Figure IVa), which may be due to our inability to sort Tg(gata4:EGFP)+ regenerating CMs from these hearts. Therefore, we took a second approach to find putative AP-1 target genes and to test whether CM-specific expression of AP-1 transcription factors was sufficient to promote chromatin accessibility in uninjured CMs. We generated a transgenic line to overexpress junba specifically in CMs (Tg(ubb:loxP-EGFP-loxP-junba-P2A-tagBFP-HA); Tg(myl7:CreERT2), hereafter referred to as CM:junbaOE). 4-HT or vehicle was injected into adult transgenic zebrafish, ventricles were isolated 7 days post injection (dpi), and CMs were isolated by density gradient centrifugation and subjected to ATAC-Seq analysis (Figure 4c). Overexpression of junba was confirmed by RT-qPCR from whole ventricles (Figure 4d). Comparison of chromatin accessibility in uninjured CM:junbaOE CMs and uninjured Tg(myl7:nucDsRed)+ CMs revealed a number of differentially accessible regions (Online Table I). Furthermore, comparison of chromatin accessibility in the regions that became more accessible in Tg(gata4:EGFP)+ regenerating CMs (n=1,507), revealed an increase in average accessibility in uninjured CM:junbaOE CMs (Figure 4e). Of the 1,507 sites that were upregulated in gata4:EGFP+ CMs, we found that 503 were upregulated (log₂FC>0.5) in junbaOE CMs versus uninjured CMs (Online Table I). When we used more stringent criteria (log₂FC ≥ 1, mean count > 20, p_adj ≤ 0.05), we found 46 peaks that were upregulated in junbaOE CMs versus uninjured CMs (p=0.0001, Figure 4e). These observations indicate that AP-1 transcription factors are sufficient to promote chromatin accessibility changes in uninjured CMs that resemble those in regenerating CMs. Interestingly, while transcription factor occupancy prediction from the ATAC-Seq dataset revealed the presence of protected AP-1 motifs enriched in CM:junbaOE CMs, there was a much higher enrichment of ETS TF motifs (Online Figure IVb), which was also observed in CM:A-Fos (−) compared to CM:A-Fos (+) CMs. These data may indicate an interaction between AP-1 and ETS TFs in zebrafish CMs.

Figure 4. AP-1 transcription factors contribute to remodeling of the cardiomyocyte chromatin accessibility landscape after injury.

(a) Schematic illustrating the 4-HT/EtOH injections, cryoinjury, and microdissection scheme to assess chromatin accessibility by ATAC-Seq. Black dashed line indicates the plane of dissection. (b) Volcano plot depicting results from TOBIAS. Green points depict motifs that are enriched and have a TF footprint in CM:A-Fos (−) CMs, while purple points depict motifs that are enriched and have a TF footprint in CM:A-Fos (+) CMs at 4 dpci. (c) Schematic illustrating the Tg(ubb:loxP-EGFP-loxP-junba-P2A-tagBFP-HA) and Tg(myl7:CreERT2) lines and 4-HT/EtOH injection scheme to promote junba overexpression and assess chromatin accessibility by ATAC-Seq. (d) RT-qPCR of junba mRNA levels in control (Tg(ubb:loxP-EGFP-loxP-junba-P2A-tagBFP-HA); Tg(myl7:CreERT2) injected with EtOH) and CM:junbaOE (Tg(ubb:loxP-EGFP-loxP-junba-P2A-tagBFP-HA); Tg(myl7:CreERT2) injected with 4-HT) whole ventricles (n=9). Data represented as mean ± std. dev.; p-value was calculated using the unpaired t-test. (e) Average plot of ATAC-Seq signal ±1.5 kb around peaks enriched in gata4:EGFP(+) regenerating CMs (n=1,507 regions) from gata4:EGFP(+) CMs at 4 dpci, myl7:nucDsRed(+) uninjured CMs, and uninjured CM:junbaOE CMs at 7 dpi. The Venn diagram on the right depicts regions that are more accessible in both gata4:EGFP(+) CMs at 4 dpci compared to myl7:nucDsRed(+) uninjured CMs and CM:junbaOE uninjured CMs compared to myl7:nucDsRed(+) uninjured CMs (log₂FC ≥ 1, p_adj ≤ 0.05, mean count > 20). This overlap has a ranked p-value=1E-4 after 10,000 iterations of the same number of randomly sampled peaks (1,507) in zebrafish (total of 10,219). kb: kilo base pairs; dpi: days post injection; EtOH: ethanol; Uninj: uninjured.

Based on the ability of junba OE to promote chromatin accessibility changes in uninjured CMs, we investigated whether CM:junbaOE can promote precocious heart regeneration. We checked scar resolution at 21 dpci, a timepoint when scar tissue is still present in control hearts, and found a slight, but not significant, decrease in scar size in CM:junbaOE ventricles compared to control (Online Figure IVc). We also observed a slight increase in CM proliferation and no difference in CM protrusion formation (Online Figures IVd and IVe). From these data, we conclude that while junba OE in uninjured CMs can promote chromatin accessibility changes, junba OE in injured CMs does not further promote heart regeneration, likely due to the presence of junba in BZ CMs in control ventricles.

Putative AP-1 target gene fbxl22 can promote cardiomyocyte sarcomere disassembly.

Closer inspection of chromatin accessibility at specific genomic regions revealed a number of putative AP-1 target loci that may contribute to the lack of regeneration in CM:A-Fos (+) hearts. Specifically, we filtered genomic regions based on a number of criteria: (i) regions that became more accessible in Tg(gata4:EGFP)+ regenerating CMs when compared to uninjured Tg(myl7:nucDsRed)+ CMs, suggesting that nearby genes would be upregulated in regenerating CMs, and (ii) regions that contained an AP-1 motif, had increased accessibility in uninjured CM:junbaOE CMs when compared to control uninjured Tg(myl7:nucDsRed)+ CMs, and decreased accessibility in CM:A-Fos (+) when compared to CM:A-Fos (−) CMs, which together would suggest that nearby genes are bona fide AP-1 targets. One example that meets the abovementioned criteria is within the proximal promoter of the fbxl22 gene, which encodes an F-box and leucine rich repeat protein previously shown to be cardiac-enriched and regulate sarcomeric protein degradation through interaction with Skp1/Cul1/F-box (SCF) E3-ligases (Figure 5a).⁵² We confirmed the upregulation of fbxl22, in addition to AP-1 family members junba, junbb, and fosl1a, in Tg(gata4:EGFP)+ regenerating CMs compared to uninjured Tg(myl7:nucDsRed)+ CMs by RT-qPCR of FACS sorted cells (Figure 5b). To further determine whether fbxl22 is an AP-1 target, we examined genome browser tracks of transcription factor footprinting analysis. We found increased protection of DNA from transposase insertion at AP-1 motifs, which correlates with AP-1 binding, in gata4:EGFP+ regenerating CMs and uninjured CM:junbaOE CMs compared to uninjured myl7:nucDsRed+ CMs (Figure 5c). In contrast, we found decreased protection of DNA from transposase insertion at AP-1 motifs, which correlates with a loss of AP-1 binding, in CM:A-Fos (+) CMs when compared to CM:A-Fos (−) CMs (Figure 5c). As a second approach, we injected junba and fosl1a mRNA into one-cell stage zebrafish embryos and assessed fbxl22 expression at 9 hours post injection (hpi). We found that co-injection of junba and fosl1a was sufficient to activate fbxl22 expression (Online Figure Va). Together, these results suggest that fbxl22 is a target of AP-1 transcription factors.

Figure 5. Putative AP-1 target fbxl22 can promote cardiomyocyte sarcomere disassembly.

(a) Integrated Genome Viewer (IGV) tracks of average profiles (n=2–3 replicates each) of normalized ATAC-Seq data at the fbxl22 locus. Color code at the top of the panel denotes CM datasets. Yellow shaded box indicates the region of interest. (b) RT-qPCR analysis of gata4:EGFP+ CMs (n=3) at 7 dpci and uninjured myl7:nucDsRed+ CMs (n=3) sorted by FACS. Data are represented as mean ± std. dev.; p-values were calculated using the unpaired t-test and corrected for multiple comparisons between 5 groups (including those in Figure 6b) using the Holm-Sidak method. (c) IGV tracks of TF footprinting analysis. Blue peaks (positive values) indicate ATAC-Seq signal, while red peaks (negative values) indicate regions of ATAC-Seq signal that are protected from transposase insertion by putative TF binding. Yellow shaded box indicates the region of interest. (d) RT-qPCR analysis of fbxl22 mRNA levels in microdissected border zone from CM:A-Fos (+) versus control CM:A-Fos (−) at 7 dpci (n=3 pools of 3 ventricles each). Data are represented as mean ± std. dev.; p-value was calculated using the unpaired t-test. (e) Tol2 transgenesis constructs for mosaic analysis of overexpression of fbxl22 and mScarlet in zebrafish larvae. (f) Illustration of injection, heat shock, and imaging scheme. (g) Single-plane confocal images of Tg(myl7:BFP-CAAX); Tg(myl7:actn3b-EGFP) ventricles from 82–84 hpf larvae injected with hsp70l:fbxl22-P2A-mScarlet and hsp70l:mScarlet plasmids. Two representative ventricles are shown and yellow arrowheads point to CMs that have lost Actn3b-EGFP+ sarcomeric structures. Quantification of mScarlet (+) CMs that have lost Actn3b-EGFP is shown on the right. n=125 and 71 CMs from hsp70l:fbxl22-P2A-mScarlet and hsp70l:mScarlet injected embryos, respectively. Data are represented as mean ± std. dev.; p-value was calculated using the unpaired t-test. bp: base pairs; hpf: hours post fertilization; HS: heat shock. Scale bars: 20 μm.

We hypothesized that dominant-negative A-Fos disrupts sarcomere disassembly in border zone CMs through the regulation of chromatin accessibility at the fbxl22 locus, as murine FBXL22 was previously shown to target sarcomere components for proteasomal degradation.⁵² RT-qPCR analysis revealed that fbxl22 expression was decreased in the border zone of CM:A-Fos (+) ventricles when compared to control (Figure 5d). To test whether fbxl22 expression in zebrafish leads to degradation of sarcomeric proteins similar to observations in mouse, we performed mosaic analysis in larval zebrafish hearts using Tol2-mediated transgenesis of constructs encoding Fbxl22-P2A-mScarlet, or mScarlet alone, downstream of a heat-shock protein 70-like (hsp70l) promoter (Figure 5e). We injected these constructs into one-cell stage Tg(myl7:BFP-CAAX); Tg(myl7:actn3b-EGFP) embryos, with BFP marking CM membranes⁵³ and EGFP fused to sarcomeric Actn3b⁵⁴ in CMs (Figure 5f). The injected animals were heat shocked at 72 hours post fertilization (hpf), a timepoint when sarcomeres have already been established in the larval heart, as well as at 78 hpf, and then imaged between 82–84 hpf (Figure 5f). CMs expressing fbxl22 exhibited a clear loss of sarcomeric Actn3b-EGFP fusion protein when compared to controls, suggesting that overexpression of fbxl22 in zebrafish CMs is sufficient to promote sarcomeric protein degradation (Figure 5g). Altogether, our data suggest that AP-1 transcription factors regulate the chromatin accessibility landscape and subsequent expression of fbxl22, which can promote sarcomere degradation and disassembly in zebrafish CMs.

Putative AP-1 target gene ilk can promote cardiomyocyte protrusion into the injured area.

Based on the criteria described above (Figure 6a), we found another putative AP-1 target gene, integrin linked kinase (ilk), that we hypothesized might contribute to the decrease in CM protrusion into the injured area of CM:A-Fos (+) ventricles. ILK has been shown to bind the cytoplasmic domain of several integrins and thereby link the extracellular matrix to the actin cytoskeleton through its interaction with various proteins, including PINCH and Parvin.^55–59 Furthermore, ILK has been shown to play critical roles in several cellular processes, including actin cytoskeleton rearrangement, cell migration, and cell polarization.⁶⁰ We confirmed the upregulation of ilk in Tg(gata4:EGFP)+ regenerating CMs compared to uninjured Tg(myl7:nucDsRed)+ CMs by RT-qPCR of FACS sorted cells (Figure 6b). We also obtained further evidence that ilk is an AP-1 target through TF footprinting analysis (Figure 6c) and junba/fosl1a mRNA injections, which revealed that co-injection of junba and fosl1a into one-cell stage zebrafish embryos activates ilk expression (Online Figure Va). RT-qPCR analysis revealed that ilk expression was decreased in CM:A-Fos (+) ventricles when compared to control (Figure 6d). We examined the localization of Ilk in cryoinjured zebrafish ventricles and found high levels of Ilk within the injured area, but also at the leading edge of CMs protruding into the injured area at 7 dpci (Figure 6e). We further observed that at 7 dpci, Ilk levels at the leading edge of CMs was reduced in CM:A-Fos (+) ventricles when compared to control (Figure 6e), and that the overall level of Ilk protein within the ventricle was reduced (Online Figure Vb). Treatment of wild-type adult fish with an Ilk-specific inhibitor, Cpd-22⁶¹, from 3–6 dpci resulted in a decrease in the number of CMs protruding into the injured area, while the length of CM protrusions appeared unchanged (Figure 6f and Online Figure Vc). These data suggest that ilk is an AP-1 target that promotes CM protrusion into the injured area.

Figure 6. Putative AP-1 target ilk can promote cardiomyocyte protrusion into the injured area.

(a) IGV tracks of average profiles (n=2–3 replicates each) of normalized ATAC-Seq data at the ilk locus. Color code at the top of the panel denotes CM datasets. Yellow shaded box indicates the region of interest. (b) RT-qPCR analysis of ilk mRNA levels in gata4:EGFP+ CMs at 7 dpci (n=3) and uninjured myl7:nucDsRed+ CMs (n=3) sorted by FACS. Data are represented as mean ± std. dev.; p-value calculated using the unpaired t-test and corrected for multiple comparisons between 5 groups (including those in Figure 5b) using the Holm-Sidak method. (c) IGV tracks of TF footprinting analysis. Blue peaks (positive values) indicate ATAC-Seq signal, while red peaks (negative values) indicate regions of ATAC-Seq signal that are protected from transposase insertion by putative TF binding. Yellow shaded box indicates the region of interest. (d) RT-qPCR analysis of ilk mRNA levels in whole ventricles from CM:A-Fos (+) versus control CM:A-Fos (−) at 7 dpci (n=3–5). Data are represented as mean ± std. dev.; p-value was calculated using the unpaired t-test. (e) Ilk and F-actin staining in sections from CM:A-Fos (+) versus control CM:A-Fos (−) ventricles at 7 dpci. (f) Quantification of the number of CM protrusions per 100 μm of wound border (left) and length of CM protrusions (right) in ILK inhibitor (ILKi, n=4) versus DMSO control (n=5) treated ventricles at 7 dpci. Data are represented as mean ± std. dev.; p-values were calculated using the unpaired t-test (number of CM protrusions) or the Mann-Whitney test (length of CM protrusions). Scale bars: 100 μm in whole ventricle images, 20 μm in zoomed images.

JUNB and FOSL1 can modulate mammalian cardiomyocyte behavior.

To determine whether AP-1 transcription factors can also modulate mammalian CM behavior, we utilized primary rat neonatal CM (rNCM) cultures. We chose to overexpress Junb and Fosl1, AP-1 family members that are expressed in regenerating zebrafish CMs (Figure 1e), but not highly expressed or upregulated in BZ CMs of the adult mouse heart following MI.²⁷ We transfected rNCMs with GFP, Junb, Fosl1, or Junb+Fosl1 mRNA (Figure 7a), and observed an increase in Junb and Fosl1 mRNA levels by RT-qPCR (Figure 7b) and JUNB and FOSL1 protein levels by immunoblotting and immunostaining (Figure 7c and Online Figure VIa). We determined that the rate of transfection efficiency was approximately 65% (Online Figure VIb). We found that expression of the putative zebrafish AP-1 target Ilk (Figure 6 and Online Figure Va) was also upregulated upon Junb+Fosl1 overexpression in nRCMs by RT-qPCR (Figure 7b) and immunoblotting (Figure 7c). Furthermore, we observed an increase in the expression of the CM dedifferentiation marker, Runx1, and cardiac precursor marker, Tbx5. Interestingly, overexpression of both Junb+Fosl1 led to the dramatic stabilization of JUNB and FOSL1 proteins (Figure 7c), which may explain why co-overexpression led to the increased expression of genes such as Ilk, Runx1, and Tbx5.

Figure 7. JUNB and FOSL1 can modulate mammalian cardiomyocyte behavior.

(a) Schematic illustrating primary neonatal rat CM (nRCM) culture and transfection of Junb, Fosl1, and GFP mRNA. (b) RT-qPCR analysis of mRNA levels in Junb, Fosl1, Junb+Fosl1, and GFP transfected nRCMs at 24 hours post transfection (hpt). n=4 biological replicates. Data are represented as mean ± std. dev.; p-values were calculated using ordinary one-way ANOVA and adjusted for multiple comparisons between 4 groups using the Bonferroni multiple comparisons test. (c) Representative immunoblot of nRCM extracts from Junb, Fosl1, Junb+Fosl1, and GFP transfected nRCMs at 24 hpt using antibodies for JUNB, FOSL1, ILK, and LAMINB1. n=3 biological replicates. (d) Quantification of CM protrusion number and length from a nRCM scratch assay in Junb, Fosl1, Junb+Fosl1, and GFP transfected nRCMs at 24 hpt. n=3 biological replicates. Data are represented as mean ± std. dev.; p-values were calculated using ordinary one-way ANOVA and adjusted for multiple comparisons between 4 groups using the Bonferroni multiple comparisons test (number of CM protrusions) or the Kruskal-Wallis test with Dunn’s multiple comparisons test (length of CM protrusions). (e) Immunostaining of Junb, Fosl1, Junb+Fosl1, and GFP transfected nRCMs at 24 hpt for cardiac Troponin I (CTNI) and KI67. Yellow arrowheads point to KI67+ CMs. The quantification on the right depicts the percentage of CMs that are KI67+. n=3,455 GFP transfected, n=3,157 Junb transfected, n=3,256 Fosl1 transfected, and n=2,844 Junb+Fosl1 transfected CMs counted in total from 4 biological replicates. Data are represented as mean ± std. dev.; p-values were calculated using ordinary one-way ANOVA and adjusted for multiple comparisons between 4 groups using the Bonferroni multiple comparisons test. Scale bars: 20 μm. kDa: kilodaltons.

We further investigated CM protrusion formation in response to Junb and Fosl1 overexpression using a nRCM scratch assay. We found that Fosl1 and Junb+Fosl1 overexpression led to a slight increase in the number of CM protrusions formed and a significant increase in the length of CM protrusions (Figure 7d and Online Figure VIc). Treatment of Junb+Fosl1 transfected nRCMs with the small molecule ILK inhibitor, Cpd-22, resulted in a decrease in the number of CM protrusions, without significantly affecting CM protrusion length (Online Figures VId and VIe), similar to our observations in zebrafish (Figure 6f). Lastly, upon Junb, Fosl1, and Junb+Fosl1 overexpression, we observed an increase in KI67+ CMs (Figure 7e). Altogether, these data suggest that overexpression of Junb+Fosl1, AP-1 family member genes that are not normally upregulated or highly expressed in mammalian CMs following MI, leads to changes in CM behavior that are remniscent of those occuring in zebrafish CMs during regeneration, including CM proliferation and CM protrusion.

DISCUSSION

Our data indicate that AP-1 transcription factors are key regulators of the cardiomyocyte response to injury through their ability to promote chromatin accessibility changes, allowing for the activation of gene expression programs that support regeneration. Using ATAC-Seq, we provide a genome-wide overview of chromatin accessibility changes in regenerating cardiomyocytes compared to uninjured and show that the chromatin landscape is highly dynamic following damage to the heart. Similar to previous studies that have assessed histone modification dynamics during zebrafish heart regeneration,^18,19 and chromatin accessibility during imaginal disc regeneration in Drosophila melanogaster⁶² and whole body regeneration in the acoel worm Hofstenia miamia⁶³, we find that the majority of chromatin accessibility changes occur outside of promoter regions. These observations suggest that a rewiring of the enhancer accessibility landscape is a broad response to tissue injury and may be a general prerequisite to activate gene expression programs to promote regeneration in a number of model systems.

Through bioinformatic analyses, we find that AP-1 motifs are significantly enriched and present in more than 50% of the regions that gain accessibility in regenerating cardiomyocytes. Although we cannot assess direct AP-1 binding by ChIP-Seq in our system because of a lack of antibodies for zebrafish AP-1, we find a significant enrichment of AP-1 motifs within regions of accessible chromatin that are protected from transposase insertion. Transcription factor footprinting from chromatin accessibility data has been shown in multiple publications to correlate with TF binding.^64–68 Furthermore, the overexpression of a dominant negative AP-1 variant leads to decreased chromatin accessibility at a subset of these regions. The overexpression of junba is also sufficient to promote chromatin accessibility changes in the absence of injury, suggesting that AP-1 transcription factors are important mediators of chromatin accessibility in the cardiomyocyte response to damage. This function is in line with previous studies that have shown that AP-1 is a mediator of chromatin accessibility in response to hormone or growth factor stimulation of cells in vitro.^25,26 Furthermore, in a recent study of chromatin accessibility dynamics in whole-body regeneration of Hofstenia miamia, 18% of the top 50 motifs enriched in chromatin regions that show highly variable accessibility during the regenerative process were AP-1 transcription factor family motifs.⁶³ AP-1 motifs are also present in chromatin regions exhibiting highly variable accessibility during regeneration in the planarian worm Schmidtea mediterranea 6 hours post amputation.⁶³ These authors further show that egr, another immediate early response gene, is necessary for regulating chromatin accessibility and promoting a transcriptional cascade to support whole-body regeneration. Altogether, these findings suggest an elegant mechanism whereby the activation of immediate early genes, including AP-1, in response to tissue damage leads to genome-wide changes in chromatin accessibility. This response enables cells to activate gene regulatory networks that function to promote various aspects of organ regeneration. It appears that this mechanism is widespread based on studies showing that AP-1 is necessary for the regeneration of multiple organ systems, including the zebrafish heart as shown here, whole body in planarians,⁶⁹ zebrafish larval fin fold⁷⁰ and adult heart⁷¹, mouse axons,⁷² and mouse liver.⁷³

During heart regeneration, cardiomyocytes undergo dedifferentiation, proliferation, and subsequent repopulation of the injured area to replenish lost myocardium. While the adult mammalian heart is a largely non-regenerative organ, several molecules and signaling pathways have been shown to promote adult heart regeneration by regulating these processes.^6–12 Our data indicate that AP-1 is a key regulator of zebrafish cardiomyocyte proliferation, sarcomere disassembly, and protrusion into the injured area, in response to damage. While the regulation of cardiomyocyte proliferation during heart regeneration has been the subject of intense research efforts, the mechanisms regulating sarcomere disassembly and cardiomyocyte repopulation of the injured area are still unclear. Our data indicate that AP-1 promotes chromatin accessibility at genomic loci, including fbxl22 and ilk, whose expression regulates sarcomere disassembly and cardiomyocyte protrusion into the injured area, respectively. Studies of ILK in the mammalian heart have shown that adenoviral expression of Ilk in rat hearts following myocardial infarction leads to improved cardiac function and remodeling, due to increased angiogenesis, cardiomyocyte survival, and proliferation.⁷⁴ In addition, ILK has been implicated in cardiomyocyte migration, in conjunction with Thymosin beta-4,⁷⁵ consistent with the results of our studies here.

AP-1 motifs are also found in chromatin regions that gain accessibility in border zone CMs from the non-regenerative adult mouse.²⁷ Our analysis suggests that the AP-1 response largely differs between mouse and zebrafish CMs in response to cardiac injury. We find differences in both the AP-1 family member genes that are expressed and the genes containing AP-1 peaks that arise in BZ CMs from zebrafish and mouse. Furthermore, modulation of AP-1 gene expression in mammalian CMs leads to changes in CM behavior, including increased CM proliferation and CM protrusion. It will be interesting to understand how and why these differences arise between regenerative and non-regenerative species and whether manipulation of the AP-1 response can promote regeneration in the mammalian heart in vivo.

The function of AP-1 in zebrafish heart regeneration exhibits many parallels to that of Hippo signaling and its effectors YAP/TAZ in promoting mammalian heart regeneration. For instance, it was shown that the expression of constitutively active YAP (YAP5SA) in adult murine cardiomyocytes promotes chromatin accessibility changes to induce a more fetal-like transcriptional program.⁷⁶ These transcriptional changes were consistent with a more primitive and proliferative state and ultimately led to cardiomyocyte hyperplasia and heart failure. Interestingly, AP-1 binding motifs were found, in addition to TEAD motifs, in regions that gained chromatin accessibility upon YAP5SA overexpression,⁷⁶ suggesting that these transcription factors may function combinatorially to regulate chromatin accessibility changes in cardiomyocytes. Further, the same group had previously shown that in Hippo-deficient adult mouse hearts, YAP directly regulates cell cycle progression genes, along with genes regulating the actin cytoskeleton, to promote cardiomyocyte proliferation and migration following injury.¹² Using Mdx mutant mice, it was shown that disruption of the dystrophin glycoprotein complex compromises neonatal heart regeneration and results in a lack of cardiomyocyte protrusion and migration into the injured area.¹² Our data in the zebrafish heart show that blocking AP-1 function leads to a reduction in sarcomere disassembly and cardiomyocyte protrusion into the injured area and, subsequently, defects in scar resolution. Previous studies in zebrafish have also implicated Cxcr4 in cardiomyocyte migration and proposed that this process is required for cardiac regeneration following apical resection.⁵¹ Altogether, these data indicate that cardiomyocyte migration is an important process during cardiac regeneration and that promotion of cardiomyocyte proliferation alone is not sufficient for successful regeneration. These results have important implications in the design of therapeutic interventions to promote cardiac regeneration in the adult mammalian heart.

NOVELTY AND SIGNIFICANCE.

What Is Known?

• Expression of Activator protein 1 (AP-1) genes is upregulated in response to multiple stimuli and stressors, including various models of tissue injury.

• AP-1 transcription factors have been shown to regulate chromatin accessibility through their interaction with the SWI/SNF chromatin-remodeling complex.

• AP-1 transcription factor motifs are observed in regions of chromatin that gain accessibility in border zone cardiomyocytes (CMs) following myocardial infarction in adult mice.

What New Information Does This Article Contribute?

• We provide the first genome-wide chromatin accessibility analysis of border zone CMs following cardiac injury in adult zebrafish.

• AP-1 function in zebrafish CMs is essential to promote their dedifferentiation and proliferation during regeneration; AP-1 contributes to chromatin remodeling at loci involved in sarcomere disassembly and CM protrusion into the injured area.

• While the AP-1 response in border zone CMs largely differs between regenerative zebrafish and non-regenerative adult mice, one can promote the proliferation and protrusion of mammalian CMs in culture by overexpressing the AP-1 family members Junb and Fosl1.

Nonstandard Abbreviations and Acronyms:

4-HT

4-hydroxytamoxifen

AFOG

Acid Fuchsin Orange G

BZ

border zone

CI

cryoinjury

CM

cardiomyocyte

D

distal

Dediff.

dedifferentiation

dpci

days post cryoinjury

dpi

days post injection

dps

days post sham

embCMHC

embryonic cardiac myosin heavy chain

ER

endoplasmic reticulum

EtOH

ethanol

hpf

hours post fertilization

hpi

hours post injection

hpt

hours post transfection

HS

heat shock

kb

kilo base pairs

MI

myocardial infarction

P

proximal

pg

picograms

Prolif.

proliferation

rNCM

rat neonatal cardiomyocyte

TF

transcription factor

TSS

transcriptional start site

uninj.

uninjured