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 activator protein-1 (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.

Keywords: cardiomyocyte protrusion; chromatin; regeneration; transcription factors; zebrafish.

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.

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.

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.

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.

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.

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.

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.