leptin b and its regeneration enhancer illustrate the regenerative features of zebrafish hearts
Abstract
Background
Zebrafish possess a remarkable regenerative capacity, which is mediated by the induction of various genes upon injury. Injury-dependent transcription is governed by the tissue regeneration enhancer elements (TREEs). Here, we utilized leptin b (lepb), an injury-specific factor, and its TREE to dissect heterogeneity of noncardiomyocytes (CMs) in regenerating hearts.
Results
Our single-cell RNA sequencing (scRNA-seq) analysis demonstrated that the endothelium/endocardium(EC) is activated to induce distinct subpopulations upon injury. We demonstrated that lepb can be utilized as a regeneration-specific marker to subset injury-activated ECs. lepb+ ECs robustly induce pro-regenerative factors, implicating lepb+ ECs as a signaling center to interact with other cardiac cells. Our scRNA-seq analysis identified that lepb is also produced by subpopulation of epicardium (Epi) and epicardium-derived cells (EPDCs). To determine whether lepb labels injury-emerging non-CM cells, we tested the activity of lepb-linked regeneration enhancer (LEN) with chromatin accessibility profiles and transgenic lines. While nondetectable in uninjured hearts, LEN directs EC and Epi/EPDC expression upon injury. The endogenous LEN activity was assessed using LEN deletion lines, demonstrating that LEN deletion abolished injury-dependent expression of lepb, but not other nearby genes.
Conclusions
Our integrative analyses identify regeneration-emerging cell-types and factors, leading to the discovery of regenerative features of hearts.
1 INTRODUCTION
Adult mammals poorly regenerate damaged hearts, resulting in high morbidity and mortality from cardiac diseases. In contrast, zebrafish possess a remarkable ability to regenerate injured hearts. Heart regeneration studies across animal species identified multiple regeneration-driving genes and signaling pathways, which are highly conserved between mammals and zebrafish.1 These results suggest that a key difference between mammals and zebrafish is not the presence or absence of regeneration-driving genes but in the mechanisms controlling expression of these genes after injury.2-4 Dissecting the regulatory mechanism governing zebrafish heart regeneration will provide insights into understanding the molecular basis of cardiac regeneration.
Heart regeneration is a complex process, in which cardiomyocytes (CMs) and non-CMs cooperatively play roles to trigger regenerative programs. As CMs are muscular cells that ensure cardiac functions to circulate blood throughout the body, an essential event for heart regeneration is to activate CM proliferation. Non-CMs, such as endocardium, epicardium (Epi), and immune cells, are also crucial cardiac tissues that respond to cardiac injury to initiate regenerative process. Endocardium and Epi in zebrafish are rapidly activated within several hours and 1 day postinjury, respectively, contributing to heart regeneration in multiple aspects.5, 6 For instance, cardiac injury-activated endocardium and Epi produces paracrine factors to stimulate CM proliferation5, 7-9 and extracellular matrix proteins (ECMs) to construct the regenerative niche/environment.10, 11 Activated endocardium and Epi are thought to be heterogenous,12-15 but molecular identity representing diverse subgroups of these cardiac tissues in regenerating hearts is relatively unexplored.
Tissue regeneration enhancer elements (TREEs) are key regulatory elements that relay injury signals to direct gene expression.3, 16-20 We previously identified the first cardiac TREE in zebrafish, the leptin b (lepb)-linked regeneration enhancer (LEN).17 lepb, the target gene of LEN, are not active during development but are robustly activated upon injury,21 highlighting their regeneration-specific characteristics. Here, we dissect lepb-expressing cell populations at the single-cell level to infer dynamic changes of non-CM populations in injured hearts. Further studies of epigenome profiles, enhancer assays, and LEN deletion mutant analysis determine the regeneration-dependent specificity of LEN in non-CMs. Overall, our comprehensive analyses of multiple transcriptomic and epigenomic profiles of injured hearts identify novel molecular and cellular targets for heart regeneration and provide insights into the regeneration-specific features of the hearts.
2 RESULTS
2.1 scRNA-seq analysis combined with lepb, a regeneration-specific marker, enhances classification of heterogenous non-CM populations into subgroups
Injury-induced genes can be utilized for representing cell types emerging upon injury. Our previous work demonstrated that lepb exhibits regeneration-specific expression in endocardial cells,21 indicating the potential advantage of lepb as a regeneration marker. To investigate transcriptomic changes at the single cell level in adult hearts, we analyzed available single cell sequencing (scRNA-seq) datasets generated with ventricular cells sorted for either runx1P2:Citrine or kdrl:mCherry.22 runx1P2:Citrine is expressed in a wide range of cells in injured hearts, including CMs, Epi, endocardium, and blood cells and kdrl:mCherry labelled endothelium/endocardium in the hearts.22 Datasets obtained from wild-type uninjured and 3 days postcryo-injury (dpi) hearts were used for unsupervised clustering, identifying 23 different clusters. To determine cell types composing these clusters, we annotated each cluster with known marker genes (Figure 1A–C): CMs with tnnt2a and myl7;23, 24 epicardial cells (Epi) and cardiac fibroblasts (cFB) with tcf21, fn1b, col1a1a, tagln, and col5a1;12, 25-27 endocardial/endothelial cells (ECs) with cdh5, kdrl, and flt1;28, 29 coronary endothelial cells (cEC) composing blood vessels in hearts with cdh5, kdrl, flt1, and aplnra;12 mesenchyme-like cells (Mes) with mgp, angptl7, and rspo1;30, 31 thrombocytes (throm) with itga2b and gp1bb;32, 33 neutrophils (Neu) with mpx and lyz;34, 35 macrophages (MC) with mfap4, c1qa, and mpeg1.1;36-38 leukocytes (leu) with mhc2a, coro1a, and cxcr4b39-41 (Table 1).

Cluster | Injured (3dpi) | Uninjured | Total cell number | Marker genes |
---|---|---|---|---|
EC (uninjured) | 225 | 1940 | 2165 | cdh5, kdrl, flt1 |
lepb− activated EC | 315 | 8 | 323 | cdh5, kdrl, flt1 |
lepb+ activated EC | 740 | 128 | 868 | cdh5, kdrl, flt1, lepb |
Activated/uninjured EC | 119 | 69 | 188 | cdh5, kdrl, flt1 |
coronary EC (cEC) | 41 | 19 | 60 | cdh5, kdrl, flt1, aplnra |
tcf21low activated Epi/cFB | 177 | 4 | 181 | tcf21, fn1b, col1a1a, tagln, col5a1 |
tcf21High activated Epi/cFB | 381 | 44 | 425 | tcf21, fn1b, col1a1a, tagln, col5a1 |
CM/EC | 57 | 149 | 206 | tnnt2a and myl7 |
Mes | 24 | 10 | 34 | mgp, angptl7, rspo1 |
Uninjured thrombocytes | 180 | 792 | 972 | itga2b, gp1bb |
Activated thrombocytes | 636 | 10 | 646 | itga2b, gp1bb |
Activated Macrophage | 540 | 32 | 572 | mfap4, c1qa, mpeg1.1 |
Neutrophil | 176 | 41 | 217 | mpx, lyz |
Macrophage/Leukocytes | 91 | 41 | 132 | mhc2a, coro1a, cxcr4b, mpeg1.1 |
Leukocytes | 118 | 370 | 488 | mhc2a, coro1a, cxcr4b |
Total | 3820 | 3657 | 7477 |
To determine clusters emerging upon injury, we assess the enriched cell composition for the injury. EC, Epi/cFB, MC, and throm were clearly distinguished by the injury status, identifying injury-induced subgroups (Figure 1B and Table 1). We next focused on clusters enriched with lepb expression. lepb is detectable in the cells of injured but nearly undetectable in the uninjured hearts, indicating a regeneration-specific feature of lepb (Figure 1D,E). The major clusters expressing lepb are the activated/injured EC. Interestingly, we found that lepb expression is high in some, but not all, activated EC clusters, revealing the heterogeneity of ECs in injured hearts. Although lepb expression in ECs was previously reported,17, 21, 42 our analysis identified additional lepb expressing cell type that one of the activated Epi/cFB clusters have the notable number of lepb expressing cells. This lepb+ Epi/cFB cluster displayed higher expression of tcf21, a well-defined Epi and epicardial-derived cell (EPDC) marker,43 compared to another activated Epi/cFB cluster. In addition, some of the less well-defined clusters, including ECs containing uninjured cells and leukocytes, are combined. Collectively, our scRNA-seq analysis identified 15 distinct clusters representing various cardiac cell types and the injury-induced status (Figure 1A).
2.2 Identification of lepb+ EC subgroup that directs injury-induced expression of secreted factors
The lepb expression level can separate the activated ECs into subgroups (Figure 2A), prompting us to dissect the molecular basis of distinctly activated ECs. To this end, we analyzed differentially expressed genes between lepb+ and lepb− activated ECs (Supplementary Data S1). Our analysis identified 80 genes with significant changes in expression levels (P-value <0.05 and fold change >2), including 47 and 33 genes with increased and decreased expression, respectively (Figure 2B). Gene Ontology (GO) analysis of these downregulated genes (representing lepb− activated ECs) indicated enrichment in regulation of transcription by RNA polymerase II and cell differentiation, while upregulated genes (representing lepb+ activated ECs) were enriched for responses to stress, response to biotic stimulus, response to stimulus, and response to chemical (Figure 2C). Consistent with GO analysis, gene set enrichment analysis (GSEA) demonstrated a significant increase of components for stress response (Figure 2D). A major category of highly expressed genes in lepb− activated ECs is transcription factors (TFs), including the injury-responsive AP-1 complex (fos and jun family TFs)44 and endothelial/endocardial TFs (klf2a and klf6a).45, 46 However, further analysis to visualize cells expressing these factors demonstrated that a significant number of lepb+ activated ECs expresses these TFs (Figure 2E). A possible explanation is that these TFs are commonly expressed in both clusters, but the limited sequencing capacity causes biased reading. Another possible explanation is that the highly induced genes likely dampen the relative expression levels of these TFs in lepb+ activated EC cluster.

A sizable portion (18 of 47) of enriched genes (P-value <0.05 and fold change >2) in lepb+ activated ECs are secreted factors or genes related to synthesize secreted factors. lepb+ activated ECs are also characterized by enrichment of regenerative factors or injury-inducible cytokines/chemokines, including serpine1,8 inhbaa,47 fn1a,26 hbegfa,48 lgals2a,49 timp2b,50 tnfaip2b, cxcl8a,51 and cxcl18b52 (Figure 2B,F). Notably, two natriuretic peptides, nppb and nppc, are highly enriched in lepb+ activated ECs; in fact, nppc was the most highly enriched gene in lepb+ activated ECs. nppb and nppc are robustly induced in diseased and injured hearts,53-55 providing additional evidence for lepb+ ECs being an injury-activated cell population. Moreover, recent comprehensive scRNA-seq analysis also identified that nppc is a representative marker for endocardium-derived cells.12, 56 Potent neutrophil recruitment chemokines, including cxcl8a and cxcl18b, were highly enriched,52, 57 indicating that lepb+ activated ECs produce chemokines to direct migration of immune cells to the wound site. Our results accord to the recent finding that cxcl18b is enriched in one of EC clusters in regenerating zebrafish hearts.12 We noticed that two key prostaglandin E2 (PGE2) synthesis enzymes, including prostaglandin-endoperoxide synthase 2b (ptgs2b, also known as cox2b) and prostaglandin E synthase 3b (ptges3b), are enriched in lepb+ activated ECs. PGE2 is an acute inflammatory signaling molecules that promote heart regeneration,58, 59 implicating lepb+ activated ECs as a source producing pro-regenerative factors. Therefore, our analysis highlights lepb+ activated ECs as a signaling center that senses cardiac injury signals, produces signaling molecules to interact with other cardiac cells, such as immune cells, and secretes pro-regenerative factors to promote heart regeneration.
2.3 Heterogenous Epi/EPDC lineages in regenerating hearts
Our scRNA-seq analysis defined two Epi/cFB clusters displaying signatures of cardiac fibroblasts, such as col1a1a, col1a1b, col5a1, fn1a, and tagln (Figure 3A). As the number of cells derived from the uninjured hearts was considerably low (10% and 2%, Table 1), these two clusters appear to emerge upon injury. In zebrafish, cFBs arise from tcf21+ Epi upon injury,25, 60 indicating that these two cFB clusters are derived from Epi. We analyzed major TFs governing central epicardial events, such as Epi formation, epicardial epithelial-to-mesenchymal transition (EMT), and EPDC lineage specification61 and found evident differences in the number of cells expressing tcf21, tbx18a, snai2, twist1a, twist1b, and hand2. Cells expressing these Epi-related TFs are high in the tcf21high Epi/cFB cluster (Figure 3B). Notably, tcf21high Epi/cFB cluster also contains more lepb expressing cells than the tcf21low cluster (Figure 3B). We next compared tcf21high and tcf21low clusters and identified 131 genes with significant changes in expression levels (P-value <0.05 and fold change >2), including 63 and 68 genes with increased and decreased expression, respectively (Figure 3C and Supplementary Data S2). GO analysis of the significantly downregulated genes (enriched in the tcf21low activated Epi/cFBs) indicated enrichment in regulation of localization, ion transmembrane transport, and cell morphogenesis involved in differentiation, whereas upregulated genes (enriched in the tcf21high activated Epi/cFBs) were associated with cell adhesion, defense response, wound healing, ECM, response to hormone (Figure 3D). Consistent with GO analysis, GSE analysis indicated a significant increasement of components for cell adhesion, ECM organization and wound healing in tcf21high activated Epi/cFBs (Figure 3E). Thus, our data suggested that two molecularly distinct subpopulations are present in the Epi/EPDC lineage upon injury.

The majority of enriched genes (P-value <0.05 and fold change >2) in tcf21high activated Epi/cFBs consists of ECM-related genes, including fn1b,26 dpt,62 dcn,63 hapln1a,64 postnb,10 pcolce2b,65 lum,66 loxa,67 mxra8a, vim,68 timp2a, and timp2b69 (Figure 3F–H). Higher expression of ECM genes indicates that tcf21high activated Epi/cFBs are the major cell-type producing ECM and ECM-modifying enzymes upon injury, implying that this cluster potentially remodels extracellular environments in a manner favorable for heart regeneration. tcf21high activated Epi/cFBs are also characterized by several signaling factors, including dkk3b (wnt antagonists),7, 70 cfd (adipsin),71 and ccn2a,72 and angiogenic factors, including f3b (coagulation factor III, cd142b)73 and angptl2a.74 These factors are considered to be pro-regenerative factors in injured tissues, implicating the paracrine signaling roles of tcf21high activated Epi/cFBs. In contrast, tcf21low activated Epi/cFBs are enriched with smooth muscle markers, such as myh11a, mylka and acta2. We observed that tcf21+ epicardial cells (red) are barely colocalized with Acta2 positive cells (green) in 3 days post amputation (dpa) hearts, indicating that these cells are likely vascular smooth muscle cells derived from epicardium (Figure 3I). Thus, our analysis indicates that Epi/cFBs exhibit heterogeneity upon injury with a unique molecular feature of key TF and lepb expressing subpopulation.
Sun and colleagues recently published a report to analyze tcf21+ sorted epicardial cells at the single cell level, demonstrating that Epi/EPDCs are the heterogenous populations in regenerating zebrafish hearts.15 As their scRNA-seq analysis used substantial numbers of epicardial cells (around 4000 cells for each uninjured and regenerating hearts), we utilized this high-quality scRNA-seq data to further dissect lepb enriched tcf21+ Epi/EPDCs. lepb+ tcf21+ clusters are expected to be unique based on the fact of low numbers of lepb+ tcf21+ cells, and thus we increased UMAP resolution to separate Epi/EPDC clusters explicitly. As more cells tcf21+ Epi/EPDCs are analyzed, compared to Figure 1 (~600 Epi/cFB cells), our analysis identified a total of 16 clusters (Figure 4A and Table 2). Similar to Sun and colleagues report, we found cell cycle gene-enriched cluster (Cluster 13 characterized with mki67, fen1, mcm2, PCNA, and rpa2), defense-responsive cluster (Cluster 6 characterized with mxb, rsad2 and saa), crabp1a+ and Frzb+ cluster (Clusters 8 and 14) and cxcl12b+ cluster (Cluster 12) (Figure 5). Our analysis subdivided the largest cell-contained and immune responsive clusters in the original report into multiple clusters. Interestingly, lepb expression is higher in regenerating samples of the Cluster 10 (Figure 4B,C). This lepb+ Epi/EPDC cluster is characterized with chemokine and inflammation-related genes, such as cxcl8b.1,75 c3a.3,76 and steap477 (Figure 4C), postulating their roles in the immune response. Overall, our approach demonstrates that scRNA-seq analysis combined with regeneration-specific gene profiles can subset the heterogenous clusters to identify unique subpopulations.

Cluster | Uninjured | Injured | Total |
---|---|---|---|
0 | 366 | 515 | 881 |
1 | 370 | 437 | 807 |
2 | 424 | 382 | 806 |
3 | 601 | 201 | 802 |
4 | 374 | 313 | 687 |
5 | 266 | 412 | 678 |
6 | 407 | 258 | 665 |
7 | 526 | 40 | 566 |
8 | 125 | 332 | 457 |
9 | 252 | 190 | 442 |
10 | 206 | 176 | 382 |
11 | 126 | 167 | 293 |
12 | 57 | 98 | 155 |
13 | 138 | 11 | 149 |
14 | 40 | 106 | 146 |
15 | 10 | 8 | 18 |
Total | 4288 | 3646 | 7934 |

2.4 Epi/cFB-specific epigenomic profiles indicate LEN is active in a subset of Epi/cFBs upon injury
LEN was identified as an enhancer directing regeneration-specific expression in hearts, which is regulated by a ~ 300 bp sequence at the proximal end of LEN (cardiac LEN or cLEN).17 Our previous work demonstrated that cLEN can direct injury-induced expression in a subset of EC,21 which is likely the lepb+ EC cluster defined by our scRNA-seq analysis as lepb is the target gene of LEN. Although our previous work did not focus on epicardial activity of LEN, scRNA-seq analysis demonstrated that some lepb expressing cells are found in tcf21high Epi/cFBs and the Cluster 10 of tcf21+ Epi/EPDCs. Thus, we assessed lepb expression and LEN activity in Epi/EPDCs. We first examined lepb expression levels with RNA-seq profiles of Epi/EPDCs generated by tcf21:EGFP+ cells sorted from 0, 3 and 7 dpa hearts.20 Compared to the 0 dpa sample, lepb RNA levels were sharply elevated at 3 and 7 dpa with a 2.69 and 3.86-fold change (Figure 6A), respectively, indicating injury-inducible lepb expression. To determine whether LEN is accessible in Epi/EPDCs upon injury, we used Assay for Transposase-Accessible Chromatin using sequencing (ATAC-seq) of tcf21:EGFP+ cells sorted from 0, 3 and 7 dpa hearts. ATAC-seq profiles of tcf21+ cells showed LEN accessibility is increased by 1.62-fold in 3 dpa samples (FDR = 0.001008207), compared to 0 dpa (Figure 6A), indicating that LEN is activated in Epi/EPDC cells upon injury.

We next asked whether LEN can direct injury-dependent expression in Epi/ EPDC in vivo. LENP2:EGFP transgenic fish carry enhancer reporter constructs, in which LEN is coupled with the 2 kb upstream minimal promoter of lepb (P2). To assess injury-dependent activity, we amputated the apex of ventricles, collected hearts at 3 dpa, and immunostained cardiac section samples with a Raldh2 antibody. While Raldh2 was detected only in the outermost Epi in the uninjured hearts, cardiac injury expanded Raldh2 expression to the EC, Epi and EPDCs.5, 78 LENP2:EGFP has no EGFP expression in uninjured adult hearts but are robustly induced EGFP at 3 dpa (Figure 6B). As previously reported,17, 21 we observed strong EGFP signals in Raldh2+ EC cells near the amputation site and a subset of EC cells inside of the ventricle at 3 dpa (Figure 6B). In uninjured hearts, tcf21+ Epi cells are restricted to the outermost ventricular layer, but a subset of tcf21+ Epi emerges in the cortical myocardial layers to generate cFBs and other EPDC lineages.25 We identified that LENP2:EGFP are also highly detectable in Raldh2+ and tcf21:mCherry+ Epi/EPDCs in the cortical muscle of ventricles, confirming LEN activity in Epi/EPDC (Figure 6B–D). Thus, our results demonstrate that LEN drives injury-dependent expression in EC and Epi/EPDCs, two major non-CMs.
2.5 LEN deletion abolishes injury-dependent lepb induction in the hearts, but not other surrounding genes
Recent advent of genome editing allows to delete enhancers to assess their functional significance. To determine whether LEN is required for injury-induced lepb expression in hearts, we established LEN deletion mutants (LENΔ), in which 3.7 kb, including 1.3 kb of LEN and surrounding 1.5 kb of 5′ and 0.6 kb of 3′ regions, is removed by CRISPR/Cas9 (Figure 7A).79 LEN deletion animals are viable, and we did not detect any noticeable overt phenotype, such as obesity, under the standard husbandry condition. To examine whether LEN deletion causes defects in heart regeneration, we quantified CM proliferation at 7 dpa and fibrotic scar resolution at 30 dpa after partial ventricular resection. LENΔ/Δ mutants exhibited normal injury-induced CM proliferation and were able to resolve fibrotic scar similar to that of controls (Figure 7B–D). These results are in agreement with our previously discovery that lepb mutants regenerate heart normally.17

To examine the requirement of LEN in control of endogenous lepb expression upon injury, we performed RNA-seq analysis with uninjured and 3 dpa hearts of control and LENΔ/Δ mutants. lepb transcript level is sharply upregulated with a 36-fold increase at 3 dpa control hearts compared to uninjured control hearts (Figure 7E–G). Notably, transcriptome analyses revealed no detectable elevation of lepb at 3 dpa of LENΔ/Δ hearts (Figure 7F,G). These results suggest that injury-induced lepb expression is governed by LEN and there is no alternative enhancer element to redundantly regulate lepb expression at this locus. To determine whether LEN can control injury-dependent expression of multiple genes, we surveyed expression of neighboring genes, including si:dkey-31f5.8 (85 kb downstream), and snd1 (359 kb upstream). Injury-dependent gene expression of these genes was hardly affected by LEN deletion in 3 dpa hearts (Figure 7E–G). RNA-seq analysis demonstrated that si:dkey-31f5.8 and snd1 are induced upon injury in control hearts with 7.77 and 1.83-fold change, respectively. In LENΔ/Δ hearts, these genes are also induced upon injury with 9.96 and 2.10-fold change, respectively (Figure 7F,G). si:dkey31f5.11 and impdh1b, other two genes in close proximity to lepb, are not differentially expressed upon injury, indicating they are not cardiac injury-responsible genes (Figure 7F). Next, we validated our RNA-seq results using quantitative reverse transcription PCR (RT-qPCR) analysis. Our RT-qPCR analysis also demonstrated that si:dkey-31f5.8 was upregulated at 3 dpa of WT and LENΔ/Δ hearts and that robust lepb induction was abrogated in LENΔ/Δ, but not WT, hearts (Figure 7H). snd1 was not upregulated in both WT and LENΔ/Δ in our RT-qPCR analysis (Figure 7H). Collectively, our results demonstrate that LEN specifically governs injury-induced lepb expression, but not other neighboring genes.
3 DISCUSSION
The heart is a complex tissue comprising of multiple cell types, of which intercellular interaction is crucial for heart repair. Here, we analyzed two scRNA-seq data of uninjured and injured hearts and utilized lepb, a regeneration-specific gene, to enhance analysis power to identify unique cell subpopulations. These datasets were obtained by distinct injury models, such as cryoinjury and amputation, potentially leading to slightly different expression patterns.
However, interestingly, lepb specifies EC and Epi/EPDC linages into distinct populations in the injured hearts. The prominent feature of lepb+ ECs is the over-representation of secreted factors, which are known to act as regenerative factors. In zebrafish hearts, ECs are the most rapidly responding cell types to injury cues by changing their morphology and activating expression of secreted factors within hours of injury. As ECs are in direct contact with CMs, ECs are considered to be the most effective cell types to interact with CMs.5, 80 LEN activity is restricted to the wound area,21 suggesting that lepb+ ECs emerge at the wound area upon injury and serve as a paracrine signaling center to trigger CM proliferation. lepb+ ECs are enriched with cxcl8a and cxcl18b (immune cell attractant chemokines),52, 57 and lepb-enriched Epi/EPDCs are characterized with high expression of cxcl8b.1, c3a.3, and steap4 (immune-related genes),75-77 highlighting their roles for immunomodulation at the wound area. Similarly, recent scRNA-seq studies identified similar expression features of EC and Epi/EPDC activated by injury.12, 15, 56 Our approach utilizing a regeneration-specific gene can enable to subdivide these clusters to mine a group of interesting regeneration genes.
Leptin, encoded by the obese (ob) gene, is a well-characterized adipocytokine controlling feeding and energy balance regulation.81 In addition to these obesity-related effects, multiple studies demonstrated regeneration-associated roles of Leptin. In mammals, Leptin levels are elevated in cardiovascular disease like myocardial infarction (MI)82, 83 and in skin upon injury,84 revealing Leptin as an injury-inducible factor across vertebrates. Leptin also exerts pro-regenerative functions in multiple tissues, including skin and hearts.83-87 For instance, Leptin mutant mice (ob) showed higher mortality after cardiac injury, whereas administration of Leptin in ob mice yielded improved cardiac function and survival rate.83, 85, 86 Zebrafish have two leptin homologs: lepa and lepb. Although roles for zebrafish leptin signaling in the regulation of the feeding and obesity are unclear due to contradicting observation of the obese phenotype,88, 89 their involvement in tissue regeneration has been suggested. In eyes, lepa and lepb are robustly induced upon injury and administration of Leptin can stimulate eye regeneration through the Jak/Stat pathway.90 The same study also revealed that il6 family cytokines, including il11 and cntf, were able to stimulate eye regeneration via the Jak/Stat pathway. While LEN deletion and lepb mutant17 fish can regenerate their hearts, lepb regenerative roles may be compensated by il11 signaling as both il11 and leptin signaling share downstream effectors, such as Jak/Stat. Our transcriptomic analysis indicates that lif/m17, il11 family gene, is upregulated upon injury in both control and LENΔ/Δ hearts (Figure 7I,J). Our RNA-seq result indicated that lepa is not highly up-regulated in injured LENΔ/Δ hearts, suggesting there is no genetic compensation between leptin homologs (Figure 7K). Recent study reported that il11 receptor a (il11ra) mutant fish displayed impaired heart regeneration by failed scar resolution and decreased CM proliferation at the later phase of hear regeneration.91 However, this work indicated that CM proliferation at 7 dpa is likely normal, suggesting that il11 signaling on CM proliferation at the early phase of heart regeneration may be compensated by Leptin signaling.
TREEs are crucial for triggering injury-dependent expression in a tissue-specific manner and directing gene expression stably during regeneration. Much research in recent years have been performed to identify TREE or regeneration enhancer candidates using genome-wide analysis. in vivo activity of several candidates, including LEN, have been confirmed via transgenic assays.4, 19, 20, 79, 92 In addition to the typical transgenic assay, we validated the LEN activity directing regeneration-dependent gene expression using the enhancer deletion line. Importantly, we demonstrated that LEN deletion completely abrogated injury-inducible expression of lepb in hearts without affecting other nearby genes. This implies that one class of TREEs selectively regulates expression of a single gene within a short range. Exploring 3D chromatin conformational change of this short-ranged TREE and nearby regions upon injury will be interesting future work to understand how 3D genome architecture change affects regeneration-dependent transcription.
Heart regeneration is a highly complicated process governed by diverse cell populations with various transcriptional programs. Our integrative analyses of multiple sequencing data and genetic animal models characterize regeneration-emerging cell types and regulatory elements, leading to discovery of novel regeneration features, such as cells, factors, and cis-regulatory elements.
4 EXPERIMENTAL PROCEDURES
4.1 Zebrafish maintenance and procedures
Wild-type or transgenic male and female zebrafish of the outbred Ekkwill (EK) strain ranging up to 18 months of age were used for all zebrafish experiments. The water temperature was maintained at 26°C for animals unless otherwise indicated. Partial ventricular resection surgery was performed as described previously,93 in which ∼20% of the cardiac ventricle was removed at the apex. For expression patterns to determine enhancer activity, at least four hearts were examined per experiment. To define LEN activity, Tg(LENP2:EGFP)pd130 and Tg(tcf21:mCherry-NTR)pd108 were used. LENΔ/Δpd281 was created using CRISPR/Cas9 as described in79 using acgATTTAGGTGACACTATAGAatgtatccgtataccata GTTTTAGAGCTAGAAAtagc and acgATTTAGGTGACACTATAGAgaacccaattaggattta GTTTTAGAGCTAGAAAtagc oligos. Work with zebrafish species was performed in accordance with University of Wisconsin-Madison guidelines.
4.2 Histology and imaging
Hearts were fixed with 4% paraformaldehyde overnight at 4°C or for 1 h at room temperature. Cryosectioning and immunohistochemistry were performed as described previously.21 Hearts were cryosectioned at 10 μm thickness. Heart sections were equally distributed onto four or five serial slides such that each slide contained sections representing all areas of the ventricle. A solution comprising 5% goat serum, 1% bovine serum albumin, 1% dimethyl sulfoxide and 0.1% Tween-20 was used for blocking and antibody staining. The primary and secondary antibodies used in this study were as follows: anti-myosin heavy chain (mouse; F59; Developmental Studies Hybridoma Bank; 1:50), anti-EGFP (rabbit; A11122; Life Technologies; 1:200), anti-EGFP (chicken; GFP-1020; Aves Labs; 1:2000), anti-mCherry (chicken; MC87977980; Aves Labs; 1:200), anti-smooth muscle alpha-2 actin (ACTA2) (rabbit; GTX124505, Genetex; 1:200), anti-Ds-Red (rabbit; 632 496; Clontech; 1:500), anti-Raldh2 (rabbit; GTX124302; Genetex; 1:200), anti-MHC (mouse; F59; Developmental Studies Hybridoma Bank), Alexa Fluor 488 (mouse, rabbit and chicken; A11029, A11034 and A11039; Life Technologies; 1:500) and Alexa Fluor 594 (mouse and rabbit; A11032 and A11037; Life Technologies; 1:500). anti-MEF2 (rabbit, sc-313, Santa Cruz Biotechnology), anti-PCNA (mouse, P8825, Sigma), Alexa Fluor 488 (mouse and rabbit; Life Technologies), Alexa Fluor 594 (mouse and rabbit; Life Technologies). Cardiac section images were acquired using BZ-X810 fluorescence microscope (Keyence), LSM 700 confocal microscope (Zeiss), A1R-s confocal microscope (Nikon). Image stitching was automatically processed using BZ-X800 analyzer. Further image processing was carried out manually using Photoshop or FIJI/ImageJ software. For AFOG staining, cardiac cryosection slides were fixed with Bouin's solution for 2 h at 60°C and stained as described previously.93 Imaging was performed using Eclipse Ti-U inverted compound microscope (Nikon) and processed by Photoshop.
4.3 RNA isolation and qPCR
RNA was isolated from uninjured and partly resected hearts using TriReagent (ThermoFisher). Complementary DNA (cDNA) was synthesized using a NEB ProtoScript II first strand cDNA synthesis kit (NEB, E6560). Quantitative PCR was performed using the qPCRBIO SyGreen Blue Mix Separate-ROX (Genesee Scientific, 17-507) and a Bio-Rad CFX Connect system. All samples were analyzed in at least biological quadruplicate with two technical repeats. The sequences of the primers used are listed in Table S1. Transcript levels were normalized to actb2 levels in all experiments.
4.4 RNA-seq and ATAC-seq analyses
For RNA-sequencing, total RNA was prepared from uninjured and 3 dpa resected hearts of wild-type siblings and LENΔ/Δ. Generation of mRNA libraries and sequencing were performed at the Duke Center for Genomic and Computational Biology using Illumina HiSeq4000 with 50 bp single read runs. Adapter sequences were trimmed by Cutadapt. Sequences were aligned to the zebrafish genome (genome assembly GRCz11, Ensembl gene annotation release 104) using HISAT2.94 Gene counts were obtained by featureCounts and Transcripts Per Kilobase Million (TPM) was used to calculate fold-change. For epicardium, we used RNA-seq and ATAC-seq datasets of GSE89444, which were aligned to the GRCz11 using HISAT2. IGV genome browser was used to browser track images. Accession numbers for transcriptome data sets are GSE199697.
4.5 scRNA-seq analysis
For scRNA-seq analysis of uninjured and injured hearts, we obtained original sequencing files from GSE13818122 and reanalyzed this profile using 10× Genomics cloud service (https://cloud.10xgenomics.com/cloud-analysis). The Danio_rerio.GRCz11 (release 104) version of the zebrafish reference genome and annotation files were downloaded from Ensemble database (ensembl.org). Raw counts of wild-type uninjured and injured hearts were used for scRNA-seq analysis with the Seurat package.95 Low quality cells (nUMI ≤500, nGene ≤250, mitoRatio >0.15, log10GenesPerUMI < 1.7) were filtered out. After careful inspection, the 40 principal components (PCs) of the PCA with resolution 1 were used for clustering. Differential expression (DE) analysis was performed using the FindMarkers function of the Seurat package. GO-term and GSEA analyses were done by the enrichGo and gseGO functions of clusterProfiler.96 Volcano plot was generated by the EnhancedVolcano package.97
For scRNA-seq analysis of Epi/EPDC cells, we obtained raw count files from GSE17251115 and reanalyzed it. Low quality cells (nUMI ≤500, nGene ≤250, mitoRatio >0.15, log10GenesPerUMI < 1.0) were filtered out. Nonepicardial cells (myl7, fli1a, or lcp1 positive cells) and tcf21− cells were filtered out. After careful inspection, the 40 principal components (PCs) of the PCA with resolution 1.4 and integration were used for clustering using the Seurat package.95
AUTHOR CONTRIBUTIONS
Kwangdeok Shin: Conceptualization (equal); data curation (equal); formal analysis (equal); methodology (equal); resources (equal); validation (equal); writing – review and editing (equal). Ian J. Begeman: Data curation (equal); writing – review and editing (equal). Jingli Cao: Resources (equal); writing – review and editing (equal). Junsu Kang: Conceptualization (equal); data curation (equal); funding acquisition (equal); investigation (equal); methodology (equal); project administration (equal); supervision (equal); visualization (equal); writing – original draft (equal); writing – review and editing (equal).
ACKNOWLEDGMENTS
We thank the UW-Madison SMPH BRMS staff for zebrafish care; Nutishia Lee for contributions to experiments; Shuyang Chen for comments on bioinformatic analysis; Kang lab members for comments on the manuscript.
CONFLICT OF INTEREST
The authors declare no competing interests.
Open Research
DATA AVAILABILITY STATEMENT
Sequencing data have been deposited in GEO under accession code GSE199697.