Cross-species single-cell RNA-seq analysis reveals disparate and conserved cardiac and extracardiac inflammatory responses upon heart injury Authors: Eric Cortada, Jun Yao, Yu Xia, Friederike Dündar, Paul Zumbo, Boris Yang, Alfonso Rubio-Navarro, Björn Perder, Miaoyan Qiu, Anthony M Pettinato, Edwin A Homan, Lisa Stoll, Doron Betel, Jingli Cao, James C Lo Published: 2024 ABSTRACT AbstractThe immune system coordinates the response to cardiac injury and controls regenerative and fibrotic scar outcomes in the heart and subsequent chronic low-grade inflammation associated with heart failure. Adult mice and humans lack the ability to fully recover while adult zebrafish spontaneously regenerate after heart injury. Here we profile the inflammatory response to heart cryoinjury in zebrafish and coronary artery ligation in mouse using single cell transcriptomics. We interrogate the extracardiac reaction to cardiomyocyte necrosis to assess the specific peripheral tissue and immune cell reaction to chronic stress. Cardiac macrophages play a critical role in determining tissue homeostasis by healing versus scarring. We identify distinct transcriptional clusters of monocytes/macrophages (mono/Mϕ) in each species and find analogous pairs in zebrafish and mice. However, the reaction to myocardial injury is largely disparate between mice and zebrafish. The dichotomous response to heart damage between the murine and zebrafish mono/Mϕ and/or the presence of distinct zebrafish mono/Mϕ subtypes may underlie the impaired regenerative process in adult mammals and humans. Our study furnishes a direct cross-species comparison of immune responses between regenerative and profibrotic myocardial injury models, providing a useful resource to the fields of regenerative biology and cardiovascular research.Subject terms: Cardiac regeneration, Myocardial infarction Introduction Cardiovascular disease is the leading cause of death worldwide with myocardial infarction (MI) and its complications including congestive heart failure (CHF) accounting for the lion’s share of the burden1,2. It has long been known that local tissue injury induces distal organ or even whole-body responses3. Despite the local nature of the cardiac injury, the adverse effects of MI are not restricted to the cardiovascular system. Survivors of MI often experience liver and kidney injury, fever, inflammation and have increased risk of certain cancers and ischemic stroke4–6. In humans and other mammals, MI resulting from an acute disruption in coronary artery blood flow to the heart leads to cardiomyocyte necrosis. Due to the incapacity of the adult mammalian heart to regenerate, the heart is ultimately repaired through a predominantly fibrotic process with many patients suffering from CHF. It is well known that MI and heart failure with reduced ejection fraction are associated with chronic low-grade inflammation7. That chronic inflammation after MI can drive adverse cardiovascular events has been elegantly shown in clinical trials with blockade of some inflammatory pathways8,9. Post-MI inflammation may also contribute to the pathogenesis of breast and lung cancer, and chronic liver and kidney disease10–13. The dynamic inflammatory process post-MI evolves from an acute phase characterized by the elaboration of proinflammatory cytokines and chemokines, immune cell infiltration to the site of injury, and systemic leukocytosis. This is followed by a resolution or reparative phase punctuated by a tissue clearing and fibrotic cell and gene program in the heart with macrophages playing a large role and unresolved systemic inflammation in some patients14–20. Unlike mammals, adult zebrafish possess a remarkable capacity for cardiac regeneration with minimal scarring. This is achieved through proliferation of spared cardiomyocytes with cellular and molecular supports from non-muscle cells including immune cells21,22. Cryoinjury of the zebrafish heart triggers a sequence of dynamic pro- and anti-inflammatory programs that govern successful myocardial regeneration23,24. Swift neutrophil infiltration to the injury site in the acute inflammation stage (within 1 day of injury) is followed by macrophage recruitment that spans the inflammatory and regenerative stages (until 7 days after injury), while natural killer (NK) or T cells actively participate in the regeneration stage23,25–28. This process employs various, functionally diverse populations of macrophages that play vital roles for zebrafish heart regeneration23,27–30. Early macrophage infiltration of the wound is critical for successful regeneration and results in transient fibrosis through direct and indirect contributions to the extracellular matrix (ECM)27,31. Subsequently, the microenvironment undergoes a transition to an anti-inflammatory stage that involves distinct macrophages in scar resolution23,24. Moreover, macrophages are reported to stimulate cardiomyocyte proliferation during heart development through activation of the epicardium and promote angiogenesis during wound healing32,33. This leaves open the possibility that the different immune responses to cardiac injury may underly the diametrically opposed cardiac and extra-cardiac outcomes between adult mammals and zebrafish. Here we used unbiased single-cell transcriptomics to profile the dynamic immune response to cardiac injury in the heart, blood, liver, kidney, and pancreatic islets of mice and zebrafish. This allowed us to identify analogous immune cell subtypes between the two species along with conserved and disparate responses to cardiac injury within the heart and peripheral organs. We identified similar cardiac monocyte/macrophage (mono/Mϕ) subclusters present in both mice and zebrafish, although the responses of the shared mono/Mϕ subtypes were dramatically different after myocardial injury. Additionally, there were unique mono/Mϕ subclusters for each species. This study provides support for both differences in mono/Mϕ subtypes between species and the largely disparate reactions to heart injury among these analogous subtypes that may determine healing versus heart failure with systemic inflammation. Furthermore, the datasets and hypotheses generating analyses supplied here will serve as a resource to the field of regenerative biology to enable follow-up studies. Results Single cell transcriptomic dissection of the multiorgan response to cardiac injury in mice and zebrafish To interrogate the multiorgan response to heart injury between fibrotic and regenerative models, we performed single cell RNA-Seq (scRNA-Seq) analyses on the heart, blood, and peripheral organs of adult mice and zebrafish (Fig. 1a, b). In a model of cardiac scarring, 10–12 week old mice underwent permanent left anterior descending (LAD) coronary artery ligation or sham surgery with scRNA-Seq analyses at 1, 7, and 30 days post injury (dpi) (Fig. 1a). Sham controls were chosen to specifically assess the effect of MI rather than acute inflammation from open chest surgery34. 1 dpi is representative of an acute MI with the initial proinflammatory phase, while day 7 is part of the subacute phase when the inflammatory response shifts toward a reparative phase35. Finally, 30 dpi represents chronic MI with left ventricular (LV) systolic heart failure (HF) with an average left ventricular ejection fraction (LVEF) of 27% (Supplementary Fig. 1a–c)36. For comparisons between regenerative and non-regenerative models, we chose the cryoinjury model in adult zebrafish with the hope to enrich for immune cells in order to detect potential rare subtypes. This is because cryoinjury leaves necrotic tissues in the heart and causes stronger inflammatory responses with more infiltrating immune cells than the wildely-used apex resection model (e.g., reviewed in ref. 21). We performed scRNA-Seq of zebrafish cells from the heart, liver, pancreas, and whole kidney marrow (WKM) at 1 and 7 days post heart cryoinjury (dpci) together with the uninjured control (Ctrl) (Fig. 1b). These 2 timepoints represent acute inflammatory (1 dpci) and regenerative stages (7 dpci), respectively25. Because cardiac immune responses upon heart injury in zebrafish are predominantly enriched in the first 2 weeks, we did not include a 30 dpci sample, which is likely similar to an uninjured sample23. Approximately 196,000 murine and 70,783 zebrafish cells were included in the analyses (Fig. 1c, d). The zebrafish scRNA-Seq analyses were performed with two biological replicates, while samples from two tissue donors were pooled for most of mouse experiments. Cell type labeling was performed using SingleR (see Methods) and validated by the expression of known marker genes (Fig. 1c and Supplementary Fig. 1d; Fig. 1d and Supplementary Fig. 2a–d and Supplementary Data 1)37. Multiorgan response to cardiac injury in mouse and zebrafish by scRNA-Seq. Experimental outline for mouse (a) or zebrafish (b) heart injury and sampling for scRNA-seq analysis. A scRNA-seq analysis was done per sample, for a total of 28 murine samples (5 tissues, 3 timepoints except for the kidney which were analyzed at 7 and 30 dpi, and 2 conditions, sham and MI. Each sample included cells from one or two biological replicates combined 1:1). 24 zebrafish samples (4 tissues, 3 conditions, and 2 biological replicates) were included. a was created with BioRender. c UMAPs (Uniform manifold approximation and projection) of the 194,315 cells (from 41 mice; 22 sham and 19 MI) colored by timepoint (left), tissue of origin (center) and cell type (right). Dpi, days post infarction. d UMAP of 70,783 zebrafish cells showing treatments timepoints (left), tissue of origin (center) and cell type (right). Dpci, days post cardiac injury; WKM, whole kidney marrow; Ctrl, uninjured. Bar plot of DEGs (MI compared to sham procedure) in mice (e) and zebrafish (f) by tissue and cell type. Bars are colored by timepoint. Genes with FDR < 0.05 and absolute log2(FC) > 0.25 were considered differentially expressed. Chronic systemic inflammation in mice following MI Across the mouse tissues analyzed, immune cells represented the majority (56%) of the cells surveyed by scRNA-Seq. Granulocytes represented the majority of white blood cells (WBCs) at 1 dpi and were abundant in the heart of the MI group at 1 dpi (35%) but dropped to almost zero at 7 dpi (Supplementary Fig. 1e). The WBC composition was independently confirmed by flow cytometry in separate cohorts of mice (Supplementary Fig. 1f, g). Excluding WBCs, immune cells in peripheral tissues accounted for 18% of the murine cells sequenced, with elevated numbers of immune cells in tissues from MI compared to sham (heart 43% vs. 27%; liver 20% vs. 9%), which is consistent with MI inciting an inflammatory response (Supplementary Fig. 1e). In the heart, the major immune cell type was mono/Mϕ at all timepoints. Small numbers (<200) of cardiomyocytes were sequenced as we did not employ single nuclei RNA-Seq (snRNA-Seq) and were excluded from the analysis since this was not the focus of our study. Beta cells were the dominant cell type within pancreatic islets and no major changes in islet cell composition were detected with MI (Supplementary Fig. 1e). We found that MI induced a strong transcriptional response in all of the major immune cell types across multiple organs during all phases of MI (Fig. 1e). Mono/Mϕ cells were the most abundant immune cells in all of the organs we sequenced at all timepoints. While the ratio of sequenced parenchymal cells minimally changed, immune cell ratios and absolute counts displayed dynamic changes in response to MI (Supplementary Fig. 1e–g). Blood B and T cells increased dramatically over time in both sham and MI groups, from close to 0% at 1 and 7 dpi to a majority of the WBCs at 30 dpi, consistent with previous reports that major surgeries or trauma can cause transient lymphopenia in blood38 (Supplementary Fig. 1e, g). The frequency of blood monocytes peaked at 1 dpi in the MI group and at 7 dpi in the sham group. By day 30, mono/Mϕ remained slightly higher in the MI group compared to the sham (7 vs 2%). This trend was confirmed by flow cytometry (Supplementary Fig. 1g). In the heart at 1 dpi, the MI group had increased frequency of granulocytes compared to sham. By day 30, both sham and MI groups had similar number of mono/Mϕ in the heart (Supplementary Fig. 1e). In contrast, liver mono/Mϕ accounted for similar proportions of the liver immune cells at 1 and 7 dpi; however, at 30 dpi, the sham was exclusively composed of Kupffer cells (i.e., resident macrophages) whereas the MI group consisted of a more diverse set of immune cells such as mono/Mϕ, T, B, and NK cells, suggesting chronic liver inflammation post MI (Supplementary Fig. 1e). Similarly, in the kidney we observed a trend of ~2-fold increase in the percentage of mono/Mϕ at 30 dpi in the MI compared to sham. To identify which organs and cells were most strongly perturbed in each of the two diametrically opposed heart injury models, we assessed the number of differentially expressed genes (DEGs). In mouse, MI elicited substantial numbers of DEGs (>700) in the heart, liver, and leukocytes (Fig. 1e and Supplementary Data 2). By contrast, the pancreatic islets were largely unaffected by MI with <25 DEGs. In the heart, most of the gene expression changes with MI in non-immune cells were found at 1 dpi in fibroblasts and endothelial cells. This transcriptional perturbation decreased substantially at later timepoints, consistent with resolution of local injury by 30 dpi (Fig. 1e). In the liver, hepatocytes showed more transcriptional perturbation at 1 dpi with MI, while 30 dpi was highest for endothelial cells (Fig. 1e), indicating major transcriptional responses that vary by cell type in acute and chronic MI. The DEG count for kidney epithelial cells was similar at 7 and 30 dpi with MI, suggesting peripheral organ gene dysregulation with chronic MI. The temporal effects of MI were also dependent on the specific immune cell type as granulocytes displayed high DEG counts early (1 and 7 dpi) whereas T cells were affected later (7 and 30 dpi), which could reflect early activation of the innate immune system with acute injury followed by chronic T cell activation. Overall, our results suggest that acute MI in mice triggered an intense local response that may resolve by day 7 in the heart but remain activated at day 30 in some extracardiac organs. Cardiac and systemic responses after heart injury in zebrafish In zebrafish, heart cells had the most significant changes in cell type fractions with cryoinjury (Supplementary Fig. 2e, f). The uninjured sample contained about 11% cardiomyocytes (myl7+tnnt2a+)39, 7% mono/Mϕ (c1qa+)40, 5% NK or T cells (zbtb32+il2rb+ or lck+zap70+)41–43, 3% thrombocytes (itga2b+)44, 2% B cells (cd37+)45, and 1% neutrophils (mpx+)46. The remaining cell types, including epicardial cells and fibroblasts (tcf21+)47, endocardial and endothelial cells (kdrl+), and mural cells (pdgfrb+)48, contributed about 14% of all cells, which are labeled as Mixed cluster 5 (Fig. 1d and Supplementary Fig. 2). The mono/Mϕ and neutrophil populations sharply expanded to 26% and 6% of all cells at 1 dpci, respectively, indicating an acute inflammatory response. By 7 dpci, when mono/Mϕ were down to 13% and neutrophils to 1%, whereas the number of NK/T cells and B cells increased to 8% and 3%, respectively (Supplementary Fig. 2e, f). These observations match previous reports that macrophages and neutrophils dominate the acute inflammatory stage while NK/T cells take on a larger role during regeneration23,25. For the liver, cell type fractions were largely unchanged across samples with hepatocytes constituting ~70% of cells (fabp10+cp+, Supplementary Fig. 2c, e, f)49. The zebrafish pancreatic islets and exocrine lobules tightly adhere to intestinal and hepatic tissues and our dissection method may not be ideal for isolating islet cells (<2% of the total, included in the Mixed cluster 5). The primary cell type fractions in the Ctrl sample are acinar cells (prss1+ela2+)50, mono/Mϕ, neutrophils, NK/T cells, and B cells (Supplementary Fig. 2e, f). The mono/Mϕ ratio decreased from 27% in ctrl to 19% at 1 dpci and 17% at 7 dpci. The main WKM cell fractions in the Ctrl consisted of neutrophils, mono/Mϕ, erythrocytes, hematopoietic stem or progenitor cells (HPC, runx1t1+meis1b+csf1rb+)51, B cells, and NK/T cells (Supplementary Fig. 2e, f). The neutrophil population transiently expanded from 21% in the Ctrl to 32% at 1 dpci and reduced back to ctrl levels of 23% at 7 dpci. By contrast, the mono/Mϕ fraction shrunk from 20% in the Ctrl to 14–15% at 1 and 7 dpci. We further analyzed the kinetic response of the 29,432 immune cells that consisted of 36% mono/Mϕ, 29% neutrophils, 20% NK/T cells, and 15% B cells (Supplementary Fig. 3a). The number of mono/Mϕ sharply increased in the heart at 1 dpci, making it the most abundant cells, whereas in other organs mono/Mϕ decreased synchronously (Supplementary Fig. 3b). By contrast, the numbers of neutrophils in the heart and WKM increased concurrently at 1 dpci. Interestingly, minor changes in the abundance of immune cells in the liver were detected (Supplementary Fig. 3b), suggesting that the liver likely has minimal immune responses upon heart injury. In summary, compared to the heart samples, cell fractions in the non-cardiac organs were relatively stable upon heart injury. We next assessed the number of DEGs in zebrafish organs by cell type (Fig. 1f). Similar to the mouse results, the most dramatic changes in zebrafish were detected in hepatocytes and immune cells such as blood and tissue macrophages and neutrophils. Hepatocytes showed 504 and 371 DEGs at 1 and 7 dpci, respectively. In heart mono/Mϕ, we detected 666 and 253 DEGs at 1 and 7 dpci, respectively (Fig. 1f). Liver, pancreas, and WKM macrophages and neutrophils showed similar results with generally higher number of DEGs at 1 dpci but continued transcriptional perturbation at 7 dpci (Fig. 1f and Supplementary Data 2). Thus, we observed systemic immune cell activation, especially in macrophages, in both the injured heart and distal organs. While cardiac inflammation resolved by 7 dpci, peripheral inflammation was present. Taken together, these results suggest that both species employ local and peripheral cellular responses to heart injury, and the extracardiac response continues at 30 dpi in the scar-forming mouse model. Dynamic cardiac and extracardiac mono/Mϕ subcluster response to myocardial injury in mice Our data suggest that MI may trigger a dynamic gene program in blood monocytes from 1 to 30 dpi (Fig. 2a, b and Supplementary Data 3). At 7 dpi there was enrichment of EIF2 Signaling, mTOR signaling, Fc gamma receptor-mediated phagocytosis, and integrin signaling pathways with many showing inhibition (Fig. 2a, b and Supplementary Data 3). While at 30 dpi, the complement, antigen presentation, and alternative activation signaling pathways were enriched in monocytes with MI. Collectively, this pattern suggests a potentially transient inhibition of many monocyte activation pathways in the subacute stage that later transitions to activation with chronic MI. Intriguingly, some of the inflammatory pathways active on day 1 were active a month later in WBCs even after resolution of inflammation in the heart, indicating persistent monocyte activation with chronic MI. Mono/Mϕ comprise a heterogenous group of cell types, with significant differences and functions depending on their origin and localization52. Moreover, the simultaneous detection of pro- and anti-inflammatory pathways in monocytes suggests the coexistence of different monocyte subpopulations that may reflect the complex and multifaceted response of these cells to MI. Dynamic cardiac and extracardiac mono/Mϕ subcluster response to myocardial injury in mice. a Dot plot depicting enriched pathways in monocytes at 1, 7, and 30 days post infarction (dpi). b Dot plot shows gene expression in blood monocytes by experimental group (MI or SH) and timepoint. c UMAPs of all 37,215 sequenced mono/Mϕ reclustered, colored by timepoint (left), cluster (center) and tissue of origin (right). d Dot plot shows normalized gene expression (dot color intensity) across all analyzed mono/Mϕ clusters, regardless of tissue, timepoint and sham or MI procedure. Common mono/Mϕ marker genes and MHC-II and TLF (Timd4, Lyve1 and Folr2) genes are indicated. e Dot plot depicting enriched pathways in mono/Mϕ cell subclusters, regardless of tissue of origin, timepoint or sham/MI. f Stacked column plot shows percentage of the 6 murine subclusters (C1-C6) of total mono/Mϕ by condition, timepoint and tissue. In (a) and (e) Dot size represents statistical significance of enrichment, dot color represents z-score. Positive z-score predicts activation of the pathway in the MI compared to the sham group and negative z-score predicts inhibition of the pathway in the MI compared to the sham group. Gray dots denote undetermined activation status. In (b) and (d): Color intensity in dot plots represents normalized mean expression, dot size represents fraction of cells expressing the gene. To determine if MI could induce different responses in specific mono/Mϕ subtypes, we performed a more focused analysis restricted to these cells, from the different organs, timepoints, and sham/MI experimental groups. Cells were reclustered to identify six mono/Mϕ subsets. We performed pathway analyses on the top marker genes for each mono/Mϕ subcluster to ascertain their function, and assessed the expression of macrophage markers (Fig. 2c–e Supplementary Fig. 4a and Supplementary Data 3). Cluster 1 mono/Mϕ were mostly found in the blood and heart at 1 and 7 dpi, expressed the highest levels of Arg1, Spp1, and Ccl2, and were enriched in glycolysis, gluconeogenesis, cytokine storm, and adhesion and diapedesis pathways, suggesting a classical proinflammatory phenotype (Fig. 2d–f and Supplementary Fig. 4a, b). Cluster 2 mono/Mϕ increased in numbers in the blood following 1 dpi in both sham and MI groups, expressed high levels of MHC class II genes and were highly enriched in antigen presentation, EIF2, PD-1, and mTOR signaling pathways, suggesting immune modulation and increased RNA translation and metabolic activity in these cells (Fig. 2d–f and Supplementary Fig. 4a, b). Cluster 3 cells were present in the blood of both experimental groups but found at elevated numbers in the heart at 1 dpi and kidneys and liver at 30 dpi in the MI group (Fig. 2f and Supplementary Fig. 4b). The cluster 3 mono/Mϕ expressed the highest levels of Ly6c2 and were enriched in Fcγ receptor-mediated phagocytosis, production of nitric oxide and reactive oxygen species, EIF2, and regulation of EIF4 and p70S6K signaling, and metabolic pathways (oxidative phosphorylation, mitochondrial dysfunction, and mTOR Signaling) with an activation pattern suggesting high metabolic and phagocytic activity (Fig. 2d–f and Supplementary Fig. 4a, b). By contrast, Cluster 4 cells were predominantly found in tissues but not detected in the blood, consistent with a tissue macrophage phenotype (Fig. 2c, f and Supplementary Fig. 4b). Supporting this, cluster 4 macrophages in the liver had high expression of Clec4f, a Kupffer cell marker (Supplementary Fig. 4c). Cluster 4 macrophages expressed the highest levels of Mrc1, Timd4, Lyve1, and Folr2 and were enriched in LXR/RXR activation, endocytosis, complement, and phagocytic pathways53 (Fig. 2d, e). Mono/Mϕ cluster 5 was temporarily present in the blood of the MI group only at 7 dpi. DEGs for Cluster 5 suggested a role for these cells in iron homeostasis and hypoxia with heme and tetrapyrrole biosynthesis and hypoxia pathways enriched (Fig. 2e, f and Supplementary Fig. 4a, b). Lastly, Cluster 6 cells represented over half (57%) of the blood monocytes in the sham group at 7 dpi and were overrepresented (25%) in the MI group at 30 dpi while these cells were present at trace amounts (4%) in the sham group (Fig. 2f and Supplementary Fig. 4b). Complement, leukocyte extravasation, antigen presentation, and hepatic fibrosis were among the top pathways overrepresented in Cluster 6 cells (Fig. 2e). Altogether our data suggest that MI may regulate both the mono/Mϕ composition and their activation status in blood and peripheral tissues. Dynamic cardiac and extracardiac mono/Mϕ subcluster response to myocardial injury in zebrafish Zebrafish monocytes and macrophages are highly heterogeneous28,30,54,55. Our initial clustering result indicated two macrophage subpopulations in zebrafish (clusters 9 and 15, Supplementary Fig. 2a). To further interrogate the heterogeneity and dynamics of zebrafish macrophages after heart injury, we clustered all the zebrafish macrophages and identified seven clusters (Fig. 3a). Clusters 1 and 2 were predominantly derived from the WKM, constituting 21% and 14% of all macrophages, respectively (Fig. 3b–d). Cluster 3 (8% of the total) was primarily pancreatic and cluster 4 (9% of the total) was primarily from the heart (Fig. 3b–d). Cluster 5 was the smallest mono/Mϕ subpopulation (2%). Cluster 6 (31%) was the most abundant, constituting 31% of all macrophages and present in all 4 organs. Similarly, cluster 7 had no organ specificity and contributed 15% of all macrophages. Interestingly, four subclusters (1, 3, 6, and 7) were enriched in the uninjured control, with their numbers synchronously decreasing at 1 dpci before rebounding at 7 dpci (Fig. 3e). By contrast, the fractions of clusters 4 and 5 (enriched in the heart) drastically increased at 1 dpci, but quickly dropped to the uninjured level at 7 dpci, indicating the transiently expanded mono/Mϕ subsets in the heart. Cluster 2 also increased at 1 dpci and stayed high at 7 dpci (Fig. 3e). Thus, the zebrafish mono/Mϕ population was highly heterogeneous and dynamic with heart injury. Mono/Mϕ response to cryoinjury in zebrafish. a UMAPs of 10,709 reclustered mono/Mϕ, colored by treatment group (Ctrl, 1 dpci, or 7 dpci; left), cluster (center) and tissue of origin (right). WKM, whole kidney marrow; Ctrl, uninjured; dpci, days post heart cryoinjury. b Fractions of each subcluster over total macrophages and monocytes. Tissue distribution of each mono/Mϕ subcluster grouped by cluster (c) or tissue and treatment group (d). e Dynamics of subcluster fractions over total mono/Mϕ in each treatment group (Ctrl, 1 dpci, or 7 dpci) across three groups. f UMAPs showing expression patterns of selected marker genes. g Enriched biological process terms (top) and KEGG pathway (bottom) for the subcluster markers (mean AUC > 0.6). The gene ratio is indicated by the dot size and the significance by the color of the dot (P < 0.05). h, i, Images of heart cryosection showing HCR staining signals of gda in green and moxd1 in magenta in Ctrl (uninjured), 1 dpci, and 7 dpci samples. Anti-lcp1 staining is shown in blue. DAPI staining is shown in white. The framed regions are enlarged to show details on the right with different channel combinations. Arrowheads and arrows indicate representative gda+lcp1+ and moxd1+lcp1+ cells, respectively. Scale bar, 100 μm. j Quantification of gda+lcp1+ cells in the wound in Ctrl (uninjured), 1 dpci, and 7 dpci samples. n = 3 (Ctrl), 7 (1 dpci), and 8 (7 dpci). Mean ± S.D. Student’s t-test versus the control. k Images of heart cryosection showing HCR staining signals of havcr1 and moxd1 at 1 dpci in green and magenta, respectively. Anti-lcp1 staining is shown in blue. DAPI staining is shown in white. The framed regions are enlarged to show details on the right with different channel combinations. Arrowheads and arrows indicate representative havcr1+lcp1+ and moxd1+lcp1+ cells, respectively. Scale bar, 100 μm. We next looked for cluster markers to characterize these mono/Mϕ subsets. The WKM-dominating cluster 1 was enriched for expression of translation and ribosome biogenesis-associated genes such as eef1g, rpl15, and eif2s3 (Supplementary Fig. 3c, Fig. 3f, g, and Supplementary Data 4, 5). This cluster also preferentially expressed the M2 macrophage marker mrc1b (Cd206 in mice), copper ion binding protein gene moxd1, and calcium ion binding protein genes icn and s100a10b. Cluster 2 shared most of the cluster 1 marker genes at a relatively lower expression level, in addition to the enriched expression of histone modification and chromatin regulatory genes such as hmgn2, h2az2b, and mki67 and nucleotide metabolism gene dut (Fig. 3f and Supplementary Fig. 3c). The pancreatic cluster 3 showed high expression of antigen component cd7al, glycoprotein gene gpnmb, and lysosomal activity-related genes such as asah1b, lgmn, and ctsd (Fig. 3f, Supplementary Fig. 3c, and Supplementary Data 4). Notable enriched GO terms in this cluster were negative regulation of cell proliferation, leukocyte activation, and regulation of intrinsic apoptotic signaling pathway (Fig. 3g and Supplementary Data 5). The heart-dominated cluster 4 had enriched expression of cd9b, fn1a, lipf, gpc1b, gda, and hsp70.1, which are associated with cell adhesion, response to oxygen levels, protein or vesicle transport, protein folding, or endomembrane system organization (Fig. 3f, g, Supplementary Fig. 3c, and Supplementary Data 5). Enriched KEGG pathways in cluster 4 were phagosome, lysosome, and endocytosis, suggesting active phagocytosis (Fig. 3g and Supplementary Data 5). Notably, fn1a encodes the pro-regenerative ECM protein fibronectin56. Cluster 5 had enriched expression of oxygen transport-associated gene hbaa1, ribosomal genes such as rpl17, and oxidative stress-responsive gene lgals2a (Fig. 3f, g, and Supplementary Fig. 3d). Cluster 6 had high expression of standard macrophage markers such as mpeg1.1, cd74a, mhc2a, marco, mfap4, and havcr1. The top enriched GO terms were mostly regulators of the immune system process: antigen processing and presentation, leukocyte activations, and KEGG pathways were lysosome, phagosome, and apoptosis (Fig. 3f, g, Supplementary Fig. 3c, and Supplementary Data 4, 5). Cluster 7 exhibited high expression of defense response-related genes lygl1 and cybb and shared marker genes enriched in translation with cluster 1, such as rpl15 and copper ion binding protein gene moxd1 (Fig. 3f, g, Supplementary Fig. 3c, and Supplementary Data 4, 5). Although clusters 6 and 7 were present in all 4 organs, cluster 7 was similar to the WKM clusters 1 and 2, while cluster 6 displayed a pro-inflammatory and phagocytic gene program. This suggests that cluster 6 is the resident macrophage population within each organ, and cluster 7 consists of circulating mono/Mϕ. To aid in assessing the identities of these macrophage subtypes, we analyzed the expression of known zebrafish macrophage subset markers. mpeg1.1 was widely used as a pan-macrophage marker; however, it does not label the entire mono/Mϕ population. Broad expression in B cells was also observed (Supplementary Fig. 3d, e). These observations match recent reports of mpeg1.1- macrophages and mpeg1.1+ B cells in zebrafish32,57. By contrast, c1q genes were mostly restricted to the mono/Mϕ population and broadly expressed in all mono/Mϕ 7 clusters (Supplementary Figs. 2c and 3e), indicating higher specificity than mpeg1.1. Macrophages are traditionally classified as pro-inflammatory M1 macrophages and anti-inflammatory M2 populations. Notable zebrafish M1 markers include tnfa, tnfb, il1b, il6 and cxcl11.154,55. As shown in Supplementary Fig. 3d–f, tnfa, tnfb and cxcl11.1 were preferentially expressed in clusters 3 and 6, in addition to having lower expression in small fractions of clusters 4 and 7. il1b was expressed by <20% of cells in clusters 3, 5, 6, and 7, and il6 was barely detected in clusters 4, 5, 6, and 7. These expression patterns suggest an M1 identity for clusters 3, 6, and 7. However, zebrafish M2 markers mrc1b and cxcr4b were broadly expressed in all clusters except cluster 5, overlapping with the expression of M1 markers (Supplementary Fig. 3d, e)54,55,58. Other M2 genes tgfb1a and ccr2 were enriched in some cluster 4 cells in addition to a lower expression in clusters 3 and 654,55. These observations suggest that cells in clusters 3, 4, and 6 may have both pro- and anti-inflammatory function. Previous studies found that the Tgfβ signaling is required for zebrafish heart regeneration56,59. This finding and our results suggest an anti-inflammatory M2 feature of cluster 4. To better understand the functions of these subclusters in the injured heart, we performed hybridization chain reaction (HCR) staining for a few cluster markers on heart cryosections at 1 and 7 dpci, together with antibody staining against the pan-leukocyte marker L-plastin (i.e., lymphocyte cytosolic protein 1, lcp1)60,61. As shown in Fig. 3h, expression of the cluster 4 marker gda was enriched in macrophages in the boundary zone of the wound at 1 dpci but not detected in the uninjured control. By contrast, the cluster 1, 2, and 7 enriched moxd1 was mostly on the ventricular surface of the wound (Fig. 3h). By 7 dpci, the number of gda+lcp1+ cells in the wound was largely reduced (Fig. 3i, j), which match the scRNA-seq analysis result (Fig. 3e). In addition, the cluster 3, 4, and 6 enriched havcr1 was expressed in mono/Mϕ both in the boundary zone and ventricular wall at 1 dpci, with some overlap with moxd1 expression on the ventricular surface (Fig. 3k). These expression patterns suggest that cluster 4 contains boundary zone macrophages while clusters 1, 2, 3, 6, and 7 are mostly on the ventricular surface. In summary, our results revealed the dynamic heterogeneity of zebrafish mono/Mϕ during heart regeneration and that the boundary zone, phagocytotic, fn1a+tgfb1a+ cluster 4 macrophages may support regeneration29,56,59. Interspecies similarities and differences in mono/Mϕ subsets Given that the immune system plays a critical role in determining the outcome of the cardiac and systemic reaction to myocardial injury, we sought to compare and contrast the immune response between mouse and zebrafish. We focused our analyses on the monocytes and macrophages, which were amongst the cells with the greatest number of DEGs in response to heart injury in both species (Fig. 1e, f). Integration of 6 macrophages subtypes from mice (mm1–mm6) and 7 from zebrafish (dr1–dr7) was performed by sequence homology analysis to identify the commonly expressed genes and generate a joint manifold representation using the SAMap algorithm (Fig. 4a–c). We found 3 macrophage cluster pairs with high inter-species alignment scores: mm1-dr4, mm4-dr6, and mm6-dr3 (Fig. 4d). mm1-dr4 are injury-responsive mono/Mϕ that are robustly increased in the heart following myocardial injury. The mm4-dr6 pair represents tissue resident macrophages that are present in the heart and other organs examined while the mm6-dr3 mono/Mϕ pair was mostly found in the blood in mice and the whole pancreas in zebrafish. Two zebrafish clusters (dr5 and dr7) did not show significant similarity to any mouse mono/Mϕ cluster while mouse cluster mm5 did not align with zebrafish. To understand the similarities between these analogous clusters, we assessed the marker genes for each cluster and performed GO biological process (BP) enrichment analysis. Leukocyte migration, pyruvate metabolism, vesicle organization, hypoxia, and autophagy were among the top shared overrepresented pathways for mm1 and dr4 mono/Mϕ subclusters (Fig. 4e and Supplementary Data 6). The mm4-dr6 pair had the highest number of top shared overrepresented pathways, which include the complement pathway, humoral immunity, cell junction disassembly, and synapse formation (Fig. 4f). mm6-dr3 had the fewest shared pathways that contained neutrophil degranulation, immune regulation, and antigen presentation (Fig. 4g). Collectively, these analyses suggest that there are analogous mono/Mϕ clusters between mouse and zebrafish that may share similar functions. In addition to the transcriptional differences in the analogous mono/Mϕ clusters, there were mono/Mϕ clusters distinct to each species. Interspecies similarities and differences in mono/Mϕ subsets. UMAPs of mono/Mϕ from the mouse (37,215 cells) and zebrafish (10,709 cells) plotted as total (a), zebrafish (b) and murine (c) clusters. d SAMap alignment of the mouse (mm) and zebrafish (dr) mono/Mϕ subclusters. Edges with alignment scores less than 0.2 were omitted. The connections between the subclusters are colored by alignment score. Shared enriched biological process terms (GO-BP) between mm1-dr4 (e), mm4-dr6 (f), and mm6-dr3 (g) clusters based on subcluster markers for each species (mean AUC > 0.6). Only terms with an adjusted p < 0.05 in both species were included. Differential transcriptional response to cardiac injury in analogous cardiac mono/Mϕ subsets between fibrotic and regenerative models To examine shared and divergent species transcriptional responses to myocardial injury in the mono/Mϕ cluster pairs, we next focused on the heart where macrophages and infiltrating monocytes are abundant and important in tissue repair as well as the blood, which is an indicator of the systemic inflammatory response and an important source of infiltrating monocytes. We assessed transcriptional changes induced by myocardial injury (fold change compared to control) in mono/Mϕ pairs in tissues and timepoints where they are jointly found: mm1-dr4 and mm4-dr6 at days 1 and 7 in the heart (Fig. 5 and Supplementary Fig. 5a, b). Focusing solely on the ortholog genes between the two species, the number of genes that were concordantly upregulated or downregulated in the mono/Mϕ cluster pairs was very small compared to those genes that were discordantly transcribed in response to myocardial injury (Fig. 5a, c, e, g, and Supplementary Data 7). Differential transcriptional response to cardiac injury in mono/Mϕ subsets between species. a Scatterplot of the fold changes in murine and zebrafish genes with cardiac injury compared to control in mm1 and dr4 cells at 1 dpi (FDR < 0.10 and |log2FC| > 0.25). b Enriched biological process terms of the differentially expressed genes (DEGs) that are discordant between mm1 and dr4 cells in (a). c Scatterplot of the fold changes in murine and zebrafish genes with cardiac injury compared to control in mm1 and dr4 cells at 7 dpi (FDR < 0.10 and |log2FC| > 0.25). d Enriched biological process terms of the DEGs that are discordant between mm1 and dr4 cells in (c). e Scatterplot of the fold changes in murine and zebrafish genes with cardiac injury compared to control in mm4 and dr6 cells at 1 dpi (FDR < 0.10 and |log2FC| >0.25). f Enriched biological process terms of the DEGs that are discordant between mm1 and dr4 cells in (e). g Scatterplot of the fold changes in murine and zebrafish genes with cardiac injury compared to control in mm4 and dr6 cells at 7 dpi (FDR < 0.10 and |log2FC| >0.25). h Enriched biological process terms of the DEGs discordant between mm1 and dr4 cells in g. In (a), (c), (e), and (g), concordant genes with FDR < 0.10 and |log2FC| > 0.25 in both species were colored blue when downregulated and green when upregulated. Discordant genes with FDR < 0.10 and |log2FC| > 0.25 in one specie were colored red when upregulated in zebrafish and yellow when upregulated in mouse. In (b), (d), (f), and (h), pathways with a -log10(p-value) >1.3 were considered significantly enriched. In the heart, both mm1-dr4 and mm4-dr6 had >3 times more DEGs at day 1 compared to 7. GO terms enrichment analysis of DEGs in the heart at day 1 for mm1-dr4 revealed that genes upregulated in both species were in the BP of negative regulation of cell adhesion (Supplementary Fig. 5a, b). In contrast, BP terms overrepresented in zebrafish but not in mouse were related to plasma membrane homeostasis, lysosome and lipid catabolism, and cell-matrix adhesion (Fig. 5b and Supplementary Data 8). Gene pathways that were overrepresented in mm1 but not in dr4 with injury at day 1 were ribosome biogenesis and ATP/aerobic metabolism. At day 7, the dr4 mono/Mϕ displayed high transcriptional response to stress, wound healing, and phosphatidylinositol-mediated signaling whereas apoptosis and response to toxic substances were found upregulated in mm1 cells (Fig. 5d). Comparing the cells in the mm4-dr6 cluster in the heart, there was a large discordant transcriptional response induced with injury at day 1 (336 genes in mouse; 208 genes in zebrafish; 15 genes concordant) that was dramatically reduced by day 7 (35 genes in mouse; 57 genes in zebrafish; 0 genes concordant) (Fig. 5e, g). At 1 day of injury, the concordant genes signaled BP involved in ATP metabolism, nucleotide biosynthesis, and regulation of actin filament (Supplementary Fig. 5b). The overrepresented BP in mm4 but not dr6 were categorized by translation, protein localization, and immune cell cytotoxicity, all of which suggest robust cellular activation (Fig. 5f). In contrast, dr6 cells were enriched in chromatin modification, B cell differentiation, and cellular insulin signaling processes. At day 7, mm4 injury response genes were enriched in integrin-mediated signaling, protein processing, and cytokine processes (Fig. 5h), while dr6 cells showed enrichment in angiogenesis, type 2 immune response and fibroblast growth factor production. Our analysis suggests that the initial transcriptional response to injury in analogous heart macrophages was likely highly discrepant with minimal conserved pathways between mice and zebrafish. Cardiac macrophage response to heart injury by mono/Mϕ clusters in mice and zebrafish The inter-species differences in the monocyte and macrophage response to heart injury observed by scRNA-Seq was striking. We sought to use an orthogonal system for validation that does not require lysing the cells for analysis and that can be assessed by other labs in their models. In the mouse mono/Mϕ subclusters, we identified and plotted marker genes that are commonly used for flow cytometry analysis to characterize macrophages. We plotted mono/Mϕ lineage and subtype genes for cardiac macrophages at 1 dpi for the mouse clusters (1–4) present in the heart (Supplementary Fig. 5c). H2-Aa was highest in mm2 cells while Ly6c2 and Mrc1 were most abundantly expressed in mm3 and mm4 cells, respectively. Flow cytometry of digested cardiac cells allowed us to distinguish macrophage subsets based on cell surface protein expression of these macrophage genes (Supplementary Fig. 5d). Mice with sham and MI at 1 dpi had different proportions and numbers of cardiac macrophage subsets, corroborating the scRNA-Seq analyses (Figs. 2f and 6a, Supplementary Figs. 4b and 5e). At 1 dpi, there was ~5-fold increase in the number of cardiac mono/Mϕ in the MI group compared to controls (Supplementary Fig. 5e). CD206−Ly6chi cells (predicted to be mm3) and CD206−Ly6clo (predicted mm1) were up >10 and ~8-fold, respectively with MI (Fig. 6a). CD206+ macrophages (predicted mm4) were modestly increased two fold with MI compared to sham (Fig. 6a). This further permitted us to use fluorescent activated cell sorting (FACS) to enrich for mm1 cells (Supplementary Fig. 5d, f, and Fig. 6a) and assess transcriptional responses to MI within the different cardiac macrophage subsets (Fig. 6b, c). In CD206−Ly6clo macrophages, Fth1 was increased >5 fold in the MI group compared to sham (Fig. 6b). However, genes such as Ahnak, Gda, and Psap that were induced at 1 dpci in the zebrafish dr4 macrophage subset were not changed or downregulated in the case of Psap with MI in CD206−Ly6clo mouse cardiac macrophages (Fig. 6b, d). Meanwhile, CD206+ heart mono/Mϕ (mm4 cluster) showed threefold upregulation in Ccl2, validating data from scRNA-Seq analyses (Fig. 6c). We performed HCR staining of cd9b and fn1a on zebrafish heart sections (Fig. 6e). These two genes were induced in mono/Mϕ around the injury site at 1 dpci but were not detectable in the absence of heart injury. Similarly, the zebrafish cluster dr4 specific gda was also highly induced by injury at 1 dpci (Fig. 3h). Thus, while cd9b (Cd9 in mice) was likely upregulated in both species, inductions of gda and fn1a in macrophages were presumably zebrafish specific. Overall, these results suggest both shared and distinct immune responses between regenerative and profibrotic myocardial injury models. Cardiac macrophage response to heart injury by mono/Mϕ clusters in mouse and zebrafish. a Enumeration of LV macrophages expressing CD206 and Ly6c as determined by flow cytometry in mice undergoing MI or sham at 1 dpi. n = 14SH/11MI (males), 4SH/4MI (females). b Gene expression analyses of Ly6cloCD206− FAC-sorted cardiac macrophages from sham or MI at 1 dpi. n = 3SH/6MI (males), 2SH/1MI (females). c Gene expression analysis of CD206+ FAC-sorted murine cardiac macrophages from sham or MI at 1 dpi. n = 5SH/6MI (males), 4SH/2MI (females). a–c Mean ± S.E.M. Two-tailed Student’s t-test. d Violin plots of indicated genes in zebrafish cluster dr4 across timepoints. e Images of zebrafish heart cryosection showing HCR staining signals of cd9b and fn1a at 1 dpci in green and magenta, respectively. No expression was detected in the uninjured sample (Ctrl, left panel). Anti-lcp1 staining is shown in blue. DAPI staining is shown in white. The framed region is enlarged to show details on the right with different channel combinations. Arrowheads indicate representative cd9b+fn1a+lcp1+ cells. Scale bar, 100 μm. Chronic systemic and peripheral inflammation in response to heart injury in mouse but not zebrafish Dissecting the long-term inflammatory impacts of MI systemically and in extracardiac organs was a focus of this study. The murine blood mono/Mϕ response compared to the zebrafish WKM mono/Mϕ response was strikingly different with heart injury (Fig. 7a and Supplementary Data 9). In mice, there were some shared GO-BP terms such as leukocyte adhesion and mono/Mϕ activation that spanned across the timepoints. Carbohydrate catabolic process at day 7 was the only shared enriched GO-BP term across species, revealing largely distinct activation with heart injury in mice and zebrafish. At 30 dpi in mice, the blood mono/Mϕ continued to evolve with a transcriptional program depicting immune activation (migration, antigen processing, and presentation, and lymphocyte immunity). Similar immune activation patterns were observed in mouse T and B cells that were distinct from zebrafish (Supplementary Fig. 6a, b and Supplementary Data 9). Within the mono/Mϕ subpopulations in mice, there was a larger representation of cluster 2 in blood (23% sham vs 40% MI), suggesting these cells could be driving the MI-derived persistent inflammation (Fig. 2f). Cluster 2 blood monocytes in the MI group were enriched in complement, ID1 signaling, neutrophil activation, and macrophage alternative activation pathways compared to sham controls, suggesting immune activation of this subset with chronic MI (Supplementary Fig. 6c and Supplementary Data 3). Chronic systemic and peripheral inflammation in response to heart injury in mouse but not zebrafish. Enriched biological process terms for mouse blood monocytes and zebrafish WKM (whole kidney marrow) monocytes (a) and liver macrophages (b) (mean AUC > 0.6, numbers in parenthesis below each column indicate number of upregulated genes). The gene ratio is indicated by the dot size and the significance by the color of the dot (P < 0.05). mm, mouse; dr, zebrafish. c qPCR expression of liver samples from mice with MI or sham (SH) at 1 and 30 dpi. 1 dpi: n = 5SH/7MI, 30 dpi: n = 18SH/10MI. Dots represent individual mice. Mean ± S.E.M. Two-way ANOVA. d Representative liver sections stained by H&E (n = 3/group). Scale bar is 50 µm. Arrowhead indicates liver damage. e Mac2 immunohistochemistry in the liver (n = 5SH/4MI). Representative pictures are shown. Scale bar is 50 µm. f Quantification of Mac2+ liver macrophages in (e). n = 5SH/4MI male mice. Dots represent individual mice. Mean ± S.E.M. Two-tailed Student’s t-test. g Section images of zebrafish liver showing staining against IB4 (green). DAPI staining is shown in blue. Livers were collected at 30 dpci of heart injury together with the uninjured fish. Arrowheads indicate IB4+ cells. Scale bar, 100 μm. h Quantification of IB4+ macrophages in (g). n = 6 for each group. Mean ± S.D. Two-tailed Student’s t-test. Hepatic mono/Mϕ in both mouse and zebrafish likely underwent a dynamic and drastic transcriptional shift with evolution of heart injury (Fig. 7b). In mice, metabolic and death pathways were upregulated early while complement, lipid, and cytokine pathways were enriched at 30 dpi (Fig. 7b and Supplementary Data 9). In contrast, zebrafish liver mono/Mϕ cells displayed robust immune cell activation and cardiac differentiation pathways at 1 dpci before shifting to ERK1/ERK2, cholesterol, and estrogen pathways at 7 dpci with cardiac injury. The dynamic change in cytokines from day 1 to 30 in the mouse liver was independently validated by qPCR (Fig. 7c). The mice with chronic MI also displayed histological signs of liver damage consistent with chronic heart failure (Fig. 7d) and importantly a >2-fold increase in the number of macrophages with MI compared to sham, validating the scRNA-Seq data (Fig. 7e, f). By contrast, the zebrafish liver showed comparable number of isolectin B4 (IB4)+ macrophages27,62,63 between the 30 dpci and uninjured samples (Fig. 7g, h). Cluster 3 and 4 mono/Mϕ cells were the predominant mono/Mϕ found in the mouse liver at 30 dpi with cluster 3 virtually only present in the MI group (Fig. 2f and Supplementary Fig. 4b). In cluster 4, MI induced a gene expression program enriched in inflammation (acute phase response, production of NO and ROS, IL-12 signaling, coronavirus pathogenesis and PD-1) while the LXR/RXR pathway showed inhibition (Supplementary Fig. 7a, b and Supplementary Data 3). Altogether, our data suggest altered liver macrophage composition and gene expression changes signaling increased inflammation and tissue damage in mice with MI are not observed in zebrafish after cardiac injury. Transcriptional perturbations in the liver after cardiac injury in mice but not zebrafish Macrophage accumulation in the mouse liver and the high number of DEGs in hepatocytes after chronic MI suggested hepatic dysfunction. We used SAMap to delineate the hepatocyte response to cardiac injury between species. The mouse hepatocytes that underwent sham surgery at 1 and 7 dpi were most similar to zebrafish hepatocytes at 1 and 7 dpci while those from mouse sham at 30 dpi aligned with uninjured zebrafish hepatocytes (Fig. 8a). In contrast, the mouse hepatocytes from MI at 7 dpi weakly corresponded to zebrafish 1 and 7 dpci (Fig. 8a). At 1 dpi, enriched pathways in murine hepatocytes with MI included wound healing, fat cell differentiation, and triglyceride and ribonucleotide metabolism while at 7 dpi ribosome biogenesis, fibrinolysis and hemostasis were present (Fig. 8b, Supplementary Fig. 7c, and Supplementary Data 9). On the other hand, the zebrafish response was more homogeneous over time, with enriched terms relating to humoral immune response and cholesterol transport at 1 and 7 dpci (Fig. 8b, Supplementary Fig. 7c and Supplementary Data 9). Between mouse and zebrafish there were relatively few shared pathways such as ER stress, negative regulation of hydrolase activity, and humoral immune response (Fig. 8b). Collectively, these data suggest that the hepatic response to heart injury in zebrafish is a systemic stress response resembling that of the control mice. Hepatic outcomes after cardiac injury in mice and zebrafish. a SAMap alignment of the mouse (mm) and zebrafish (dr) hepatocyte scRNA-seq samples. Edges with alignment scores less than 0.2 were omitted. The connections between the samples are colored by alignment score. b Enriched biological process terms for mouse and zebrafish hepatocytes (mean AUC > 0.6). The gene ratio is indicated by the dot size and the significance by the color of the dot (p < 0.05). c Scatterplot of the fold changes in murine and zebrafish genes with cardiac injury compared to control in hepatocytes at 1 dpi (FDR < 0.10 and |log2FC| > 0.25). d Enriched biological process terms of the DEGs that are discordant between mouse and zebrafish hepatocytes in (c). e Scatterplot of the fold changes in murine and zebrafish genes with cardiac injury compared to control in hepatocytes at 7 dpi (FDR < 0.10 and |log2FC| > 0.25). f Enriched biological process terms of the DEGs that are discordant between mouse and zebrafish hepatocytes in (e). In (c) and (e), concordant genes (FDR < 0.10 and |log2FC| > 0.25) in both species were colored blue when downregulated and green when upregulated. Discordant genes (FDR < 0.10 and |log2FC| > 0.25) between species were colored red when upregulated in zebrafish and yellow when upregulated in mouse. In (d) and (f), GO terms with a -log10(p-value) >1.3 were considered significantly enriched. mm=mouse; dr=zebrafish. To gain further molecular insight into how hepatocytes between the two species respond to cardiac injury, we assessed the ortholog gene response and pathways to heart injury (Fig. 8c–f). At 1 and 7 days, there were 293 and 201 genes, respectively that were discordant between mouse and zebrafish while only 27 and 19 genes were concordant at the corresponding timepoints (Fig. 8c, e and Supplementary Data 10). At 1 dpci, zebrafish hepatocytes had preferential enrichment in cholesterol biosynthetic process whereas murine hepatocytes were enriched in energetic pathways at 1 dpi (Fig. 8d, Supplementary Fig. 7d, and Supplementary Data 11). At 7 dpci, zebrafish hepatocytes were largely enriched in fatty acid metabolism and glycolytic pathways while murine hepatocytes showed upregulation of antigen processing and presentation and ribosome/translation pathways (Fig. 8f, Supplementary Fig. 7e, and Supplementary Data 11). With chronic heart failure in mice at 30 dpi, the hepatocyte response to MI continued to evolve with none of the top pathways overlapping with prior mice or zebrafish at 1 or 7 dpi. The top terms upregulated in hepatocytes related to tissue remodeling (wound healing, cell migration, epithelial cell proliferation), response to metal ion, and receptor-mediated endocytosis (Fig. 8b and Supplementary Fig. 7c). Downregulated terms at 30 dpi included metabolic changes in small molecule, fatty acid, amino acid, and steroid metabolism, which were orthogonally validated by qPCR in separate cohorts of mice (Supplementary Fig. 8a–c and Supplementary Data 9). In summary, these discrepant responses between mouse and zebrafish hepatocytes in metabolic activities and immune activation are associated with cardiac regenerative and fibrotic effects on the liver, respectively. Similarly, mouse kidney epithelial cells in the MI group also showed potential signs of organ stress (Supplementary Fig. 9a and Supplementary Data 3), highlighting the long-term effects of MI on extracardiac organs. Discussion Our study provides a direct cross-species comparison of immune responses between regenerative and profibrotic myocardial injury models across several tissues that can serve as a useful resource to the fields of regenerative biology and cardiovascular research. We profiled immune cells in multiple extracardiac organs by scRNA-Seq after cardiac necrosis in mice and zebrafish. We observe dramatic plasticity in the dynamic immune response that was not limited to the heart. Persistent inflammatory changes, especially in macrophages, are noted in extracardiac organs and blood monocytes. Macrophages are critical for heart repair after injury32,64 and we indeed find multiple subtypes of macrophages in mice and zebrafish. To our surprise, most of the mono/Mϕ subclusters from zebrafish align to a subcluster in mouse. This permitted us to study analogous pairs of macrophages across species. The cross-species comparison is based on sequence homology between genes with projection of expression profile to a common lower dimension embedding65. Although functional similarity implied by the sequence homology is a reasonable assumption, it is possible that not all functional similarities are captured in this approach; and conversely that the aligned zebrafish-mouse macrophage subtypes may not function in a similar way. Notwithstanding these limitations in both species there are noticeable changes in macrophage populations in response to MI. Our results show that at 1 dpi, the transcriptional response in the analogous macrophage pairs was dramatically different. The vast majority of DEGs were discordant and generally <10 were concordant. We also find striking differences in metabolic pathways (ATP, oxidative phosphorylation, glycolysis, cholesterol, fatty acids) induced by heart injury between species in the macrophages. Tian and colleagues previously demonstrated that macrophage loss of mitochondrial complex I results in impaired cardiac healing66. Our data provides additional context for macrophage metabolism playing a role in cardiac regeneration versus fibrosis with metabolic pathways that will need to be tested67. It is interesting to note that while zebrafish and mouse share significant macrophage signatures, their disparate responses to MI may be key to the different fibrotic and healing outcomes between the species. What fundamentally determines the dichotomous macrophage responses to cardiac injury remains to be determined but could include factors extrinsic to the macrophage. Our data also reveal that zebrafish possess a dr7 macrophage subtype not present in mouse, which may play an important role in regeneration. Currently, it is unknown if the different reactions to myocardial injury, the presence of unique regenerative macrophage subclusters in zebrafish or inhibitory macrophage clusters in mouse, or a combination thereof might be responsible for differences in cardiac healing. The answers to these questions will require gain and loss of function studies to specifically modulate macrophages in vivo and test for their ability to impact heart regeneration. How the adult mouse and zebrafish macrophage subclusters compare to neonatal mouse macrophages is another open question fertile for future research. We identify 6 subsets of murine macrophages in our study across tissues that are dynamically regulated with MI. mm4 macrophages are tissue resident macrophages that express Timd4, Lyve1, and Folr2 (TLF+) that closely resemble the subset previously characterized by Epelman and colleagues53. Runx1 and Trem2 are 2 macrophage markers that have been linked to cardiac recovery17,68. Across our 6 mouse macrophage clusters, we observe similar expression of these 2 factors at the mRNA level without one or two clear clusters having higher levels of Runx1 or Trem2 in the heart before or after MI. Cochain and colleagues integrated scRNA-Seq datasets of cardiac macrophages in experimental murine MI and found increases in Isg15hi, Trem2hiSpp1hi, and Trem2hiGdf15hi monocytes days after MI69. Our results also indicate a rise in Isg15, predominantly within clusters 2 and 3 heart macrophages at 7 dpi whereas Spp1 is highly expressed in cluster 1 heart macrophages. We do not detect significant expression of Gdf15 within any of the cardiac macrophage clusters at any timepoint post-MI, which may reflect sensitivity or differences in thresholds for determination. The ontogeny of cardiac macrophages is known to be different between neonatal and adult mice at baseline and with injury70. Olson and colleagues identified 5 subpopulations of cardiac monocytes/macrophages in neonatal mice with regenerative capacity by scRNA-Seq71. Based on select markers published in that study, the “M2”, “DC-like”, and “M1 Mo” neonatal subpopulations most closely resemble our C1, C2, and C3 clusters in adult mice, respectively. There are limitations to using a few marker genes for comparison and functional studies would ultimately be needed to ascribe the role of macrophage subsets in cardiac regeneration. Other groups have previously studied macrophages in non-cardiac tissues after MI. Nahrendorf and colleagues profiled macrophage subtypes in multiple organs mostly by flow cytometry and targeted gene expression72. Their general finding was that tissue macrophages in distal organs were affected by the primary injury, similar to what we find with scRNA-Seq analyses. The Dimmeler lab used scRNA-Seq to study the effects of post-MI heart failure on the bone marrow and found increased inflammation with MI73. Similar to our mouse analyses, scRNA-Seq of peripheral blood mononuclear cells from humans suffering an acute ST-segment–elevated myocardial infarction (STEMI) reveal large transcriptional perturbations early on that decrease over time, though blood monocytes have the largest number of DEGs 6–8 weeks post STEMI74. Of note, human STEMI is similar to the mouse model of permanent coronary artery ligation we use here. Likewise, scRNA-Seq of blood monocytes in humans with chronic heart failure also demonstrate higher counts of monocytes in a small group of patients with heart failure75. Overall, there is now clear evidence that MI and CHF perturb mono/Mϕ function that can have deleterious effects on extracardiac tissues in mammals. This study cannot determine the extent to which acute heart injury or chronic LV systolic heart failure contributes to the peripheral inflammation seen in mice but not zebrafish15. The datasets provided here are designed to serve as a resource for the field and hypothesis generating. Future studies to examine how macrophages post-MI might contribute to chronic liver and kidney injury will be valuable and can inform of new therapeutic targets for extracardiac complications of MI. Moreover, specific macrophage reporters in zebrafish have yet to be developed, and c1q genes are ideal candidates because of their high specificity and broad expression in all mono/Mϕ subtypes. Our results suggest mixed M1 and M2 features in zebrafish mono/Mϕ clusters. It is possible that there is further heterogeneity within our defined mono/Mϕ clusters that are associated with pro-inflammatory, anti-inflammatory, or other functions. It is also worth noting that macrophages expressing the classic M1 marker tnfa were reported to be pro-regenerative in zebrafish spinal cord regeneration76. Thus, the traditional M1/M2 classification may not be applicable in determining regenerative capacity. Here we identified an acute inflammatory stage (i.e., 1 dpci) enriched macrophage cluster 4 expressing phagocytic and pro-regenerative genes (e.g., fn1a and tgfb1a), suggesting a pro-regenerative potential in zebrafish29,56,59. These dr4 cluster macrophages represent a new source of Tgfb1a and Fn1a, factors reported to contribute to heart regeneration56,59,77. This further supports the notion that a pro-regenerative decision is made early in the acute stage upon heart injury. How this transient macrophage population interacts with other cardiac cell types and may trigger a potential regenerative program warrants further investigation. One important limitation of our study is that our mouse scRNA-Seq samples came from one sample that was mostly pooled from two tissue donors while our zebrafish scRNA-Seq came from two biological replicates. The effect this has is to bias towards the mean for the group. There thus may be heterogeneity in the heart injury response between different mice or zebrafish that is not captured in the scRNA-Seq analyses. We validated key findings such as monocyte/macrophage findings using orthogonal approaches in the heart and liver in separate cohorts of mice. However, this validation was not possible for all of the different tissues and timepoints analyzed in our study. Future studies are needed to more precisely determine the extent of heterogeneity in the immune response to cardiac injury in certain organs such as the kidney and liver. As this project focused on inflammation, cell isolation and scRNA-Seq protocols were adapted for immune cells. Many hepatocytes, kidney epithelial cells, and pancreatic islet cells were sequenced. However, there were relatively few cardiomyocytes sequenced due to their size. In general, we find evidence of peripheral organ dysfunction in the liver and kidney but not in the pancreatic islets in mice. Another limitation of our study is that this does not permit robust in silico assessment of ligand-receptor or cell-cell interactions. Although we observe many of the mono/Mϕ clusters across different timepoints and organs, we do not formally know if they represent definitive fixed cell subtypes or activation states. It is possible that there is some degree of plasticity observed within a cell cluster or between clusters. Lineage tracing experiments will be needed in the future to answer this. Moreover, functional validation of the shared and distinct mono/Mϕ clusters between species needs to be rigorously performed to demonstrate their potential contributions to regeneration. Future experiments to determine how the macrophage subclusters respond to injury at later timepoints such as 7 and 30 dpi in mice and zebrafish will be of considerable interest. In summary, our results provide new insights into manipulating immunomodulators for cardiac repair. Materials and methods Mouse maintenance Animal procedures were performed according to approved protocols by the Institutional Animal Care and Use Committee (IACUC) at Weill Cornell Medical College. We have complied with all relevant ethical regulations for animal use. C57BL/6J (Stock #000664) male and female mice were purchased from Jackson Laboratories and maintained in Thoren cages under a 12/12-h light/dark cycle at constant temperature (22 °C) with free access to water and food. Mice were bred in our animal facility to maintain a colony. Mouse LAD ligation procedure 10–12 weeks old male and female C57BL/6J mice were subjected to LAD ligation or sham surgery78. scRNA-Seq analysis was performed on male mice with other experimental analyses consisting of mice of both sexes. Briefly, mice were anesthetized with isoflurane and then orally intubated and connected to a ventilator (Hugo Sachs MiniVent type 845) for mechanical ventilation. The left pectoralis major muscle was bluntly dissociated until the ribs were exposed. The muscle layers were pulled aside and fixed with an eyelid-retractor. Left thoracotomy was performed between the third and fourth ribs to visualize the anterior surface of the heart and left lung. The pericardium was removed and the proximal segment of the left coronary artery ligated with a 7-0 Ethilon nylon suture (Ethicon). In sham surgeries, the suture step was skipped. The rib cage was closed with Vicryl (Ethicon) stitches on the 3rd and 4th ribs and the skin closed with wound clips. Postoperative analgesia consisted of meloxicam (2 mg/kg) and buprenorphine (0.5 mg/kg). The mice were extubated when spontaneous breathing was observed. The animals were kept in a warm cage until recovery. Echocardiography Cardiac morphology and function were evaluated using an ultrasound imaging system Vevo 3100 with the MX400 transducer (VisualSonics). Mice were scanned 7 or 30 dpi under light isoflurane (1.5%) anesthesia within 20–30 min. Mouse body temperature was maintained using a heated platform. LVEF was measured. Venipuncture mouse blood collection For blood cytometry analyses, mouse blood was collected from the tail vein as follows. A small puncture was introduced into the vein with a scalpel. Droplets of blood were collected with a lithium heparin Microvette CB300 (Sarstedt). Mouse tissues single cell isolation Whole blood was collected by cardiac puncture. Red blood cells were lysed with ACK (Ammonium–chloride–potassium) lysis buffer and removed by density centrifugation on a 5% BSA cushion (230 × g for 5 min at 4 °C). Cells were strained with a 40 µm sieve. Kidney cells were isolated as described in Barry et al.79 with modifications: animals were perfused with ice-cold PBS-5mM EDTA through the left ventricle, then kidneys were collected, decapsulated, and minced with a razor blade. Tissue fragments were digested in HBBS 0.15 mg/ml Liberase TM (Sigma) for 25 min at 37 °C with vigorous agitation. Digestion was stopped with ice-cold FBS. The digestion mix was passed through a 70 µm cell strainer and spun down for 5 min at 300 g (4 °C). The pellet was spun down on a 7.5% BSA cushion, for 4 min at 230 g (4 °C) twice to clear debris. For heart, non-myocyte cells, hearts were perfused by the standard Langendorff method and cells obtained by enzymatic dissociation, as described in Cerrone et al.80 with minor changes. Briefly, mice were injected with 0.1 ml heparin (500 IU/ml i.p.) 20 min. before heart excision, and anesthetized with ketamine/xylazine (100 mg of ketamine/kg of body weight and 10 mg of xylazine/kg). Hearts were excised and placed in a Langendorff column. The digestion buffer consisted of 0.15 mg/ml Liberase TM (Sigma), flowing at 3 ml/min and 37 °C, for 10 min. Then the LV of sham-operated animals or the infarcted area of the LV of infarcted hearts was dissected and triturated with a plastic Pasteur pipette. Myocytes were then separated by centrifugation at 30 × g for 4 min (4 °C) and discarded. The supernatant with the rest of the cells was pelleted at 300 × g for 5 min (4 °C). Cells were then passed through a 70 µm cell strainer and debris removed by centrifugation over a 7.5% BSA-PBS cushion at 230 × g for 4 min (4 °C) twice. Liver cells were isolated as described in Kawano et al.81. Briefly, mice were anesthetized with ketamine/xylazine (100 mg of ketamine/kg of body weight and 10 mg of xylazine/kg). Livers were perfused in situ with liver perfusion medium (Invitrogen) and then digested with liver digestion medium (Invitrogen). Livers were then dissected, placed in ice-cold hepatocyte wash medium (Invitrogen), and the capsule of the liver was then opened to release the cell suspension. Hepatocytes were separated by centrifugation at 30 × g for 4 min (4 °C). The rest of the cells (supernatant) were pelleted by centrifugating 5 min at 300 × g (4 °C). Non-hepatocyte liver cells were passed through a 70 µm cell strainer. Hepatocytes and the rest of the liver cells were combined 1:1. Pancreatic islets were isolated as described in Gómez-Banoy et al.82. Briefly, mouse pancreases were perfused with CiZyme (Vitacyte) through the common hepatic duct. Pancreases were removed and digested at 37 °C for 17 min. After two washes with RPMI medium with 3% FBS, islets were separated into a gradient using RPMI medium and Histopaque-1077 (Sigma). Islets were then hand-picked and dispersed into a single-cell suspension with 0.05% trypsin. Mouse cardiac macrophages isolation Mice were injected with 0.1 ml heparin (500 IU/ml i.p.) 20 min. before heart excision and anesthetized with ketamine/xylazine (100 mg of ketamine/kg of body weight and 10 mg of xylazine/kg). Hearts were excised and the blood flushed with ice-cold PBS 5 mM EDTA with a syringe. The LV in sham-operated animals or the infarcted area in MI-operated animals were dissected and minced with a razor blade. The minced tissue was kept in ice-cold PBS 5 mM EDTA. Tissue fragments were spun down at 30 × g for 3 min (4 °C) to further remove blood cells83. Tissue fragments were then digested in HBBS 0.15 mg/ml Liberase TM (Sigma) for 25 min at 37 °C with vigorous agitation. Digestion was stopped with FBS. The digestion mix was passed through a 100 µm sieve and cardiomyocytes were separated by centrifugation at 30 × g for 4 min (4 °C). The rest of the cells (supernatant) were passed through a 70 µm cell strainer and pelleted by centrifugating 5 min at 300 × g (4 °C). The pellet was spun down on a 7.5% BSA cushion, for 4 min at 230 × g to clear debris (4 °C). Cells were then stained for FACS. Mouse single cell-RNA sequencing All cell suspensions were stained with trypan blue and counted using a hemocytometer. Cells from one to two mice were pooled 1:1 per sample (refer to https://cellxgene.cziscience.com/collections/24e324a1-b42d-4438-9bb5-cf5233fa90b0 for detailed information). Cell suspensions were diluted to target a recovery of 10,000 cells per sample at 50,000 reads per cell. The isolated cells were sent to the Epigenomics Core Facility of Weill Cornell Medicine for single-cell RNA-seq library preparation using the 10x Genomics Chromium Single Cell 3’ GEM, Library & Gel Bead Kit v3, and Chromium Single Cell B Chip Kit. The libraries were sequenced on a pair-end flow cell with a 2 × 50 cycles kit on Illumina NovaSeq6000. Mouse RNA extraction and qPCR analysis RNA isolation was performed with RNeasy Micro kits (Qiagen). cDNA was synthesized through reverse transcription using High-Capacity cDNA Reverse Transcription Kit (Thermo). cDNA was analyzed by real-time PCR using specific gene primers (See Supplementary Table in Supplementary Information) and PerfeCTa SYBR Green FastMix (Quanta) with a QuantaStudio6 Real-Time PCR System (Applied Biosystems). Rps-18 was used as reference housekeeping gene. Histology and immunostaining for mouse samples Freshly collected mouse tissues were fixed with 10% neutral buffered formalin (Sigma) overnight at 4 °C. Tissues were then transferred to 70% ethanol and subsequently embedded in paraffin and sectioned at 5 μm thickness. Tissue sections were sent to Weill Cornell’s Laboratory of Comparative Pathology for hematoxylin and eosin (H&E) staining. For immunohistochemistry (IHC) staining, liver sections were dewaxed, and antigen retrieval was performed using 10 mM sodium citrate buffer (pH 6.0) at boiling temperature for 14 min. Sections were incubated overnight at 4 °C with anti-Mac2 antibody (Cedarlane cat. CL8942AP, 1:100 dilution), followed by incubation with a biotinylated secondary antibody (Invitrogen cat. 31830, 1:500 dilution). All antibodies are commercially available and are validated by suppliers. The biotinylated 2ary antibody was detected using the HRP-conjugated avidin–biotin complex reagent VECTASTAIN ABC-HRP Kit, Peroxidase (Standard) (Vector laboratories) following the manufacturer’s protocol. Slides were developed using DAB Substrate Kit, Peroxidase (HRP), with Nickel, 3,3′-diaminobenzidine (DAB) (Vector laboratories) and counterstained with Mayer’s hematoxylin solution (Sigma). Microphotographs were taken on an optical microscope and positive cells were counted manually. White blood cell counts Whole blood was diluted 1:30 with Türk’s solution (1% crystal violet, 1.5% glacial acetic acid in distilled water) and stained cells were counted with a hemocytometer. Flow cytometry and cell sorting for mouse samples Single cells were stained in FACS buffer (2% FBS, 0.05% sodium azide in PBS) at 4 °C with fluorescent antibodies and analyzed and FAC-sorted on a Sony MA900 cell sorter. The following antibodies were used for WBC analysis: Anti-CD16/32 (Biolegend cat. 101302, 0.25 μg/M cells) (to block non-specific binding of immunoglobulins to Fc receptors), FITC Anti-CD45R/B220 (Biolegend cat. 103206, 1 μg/M cells), APC anti-CD3 (BD Pharmigen cat. 553066, 0.4 μg/M cells), Pacific Blue™ anti-CD4 (Biolegend cat. 100427, 0.2 μg/M cells), PE anti-CD8 (BD Pharmigen cat. 553032, 0.25 μg/M cells), FITC anti-CD11b (Biolegend cat. 101206, 0.25 μg/M cells), and Pacific Blue™ anti- Ly-6C (Biolegend cat. 128014, 0.2 μg/M cells). The gating strategy can be found in Supplementary Information. The following antibodies were used for cardiac macrophages: Anti-CD16/32 (Biolegend cat. 101302, 0.25 μg/M cells) (to block non-specific binding of immunoglobulins to Fc receptors), FITC anti-CD11b (Biolegend cat. 101206, 0.25 μg/M cells), PE anti-F4/80 (Biolegend cat. 123110, 0.25 μg/M cells), Pacific Blue™ anti-Ly-6C (Biolegend cat. 128014, 0.2 μg/M cells), and Alexa Fluor® 647 anti-CD206 (MMR) (Biolegend cat. 141712, 0.5 μg/M cells). All antibodies are commercially available and are validated by suppliers. IPA analysis Ingenuity Pathway Analysis (IPA) version 01-20-04 (Qiagen) was utilized for mouse pathway analyses. Redundant pathways and pathways sharing the same genes and p value of enrichment were excluded from the plots. Zebrafish maintenance Animal procedures were performed according to approved protocols by the Institutional Animal Care and Use Committee (IACUC) at Weill Cornell Medical College. We have complied with all relevant ethical regulations for animal use. Adult zebrafish of the Ekkwill (EK) strains were maintained at 28 °C under 14 h/10 h light/dark cycles84,85. Adult zebrafish (8-month-old) of both sexes were used for the experiments. Heart cryosection injury was done as following a standard protocol86. The Tg(tcf21:nucEGFP)pd41 line47 was used. All reporters were analyzed as hemizygotes. Zebrafish cell isolation and single cell-RNA sequencing Two biological replicates were included for all collected samples. Heart ventricles were collected from adult hearts either with cryoinjury or uninjured control fish. Briefly, ventricles were dissected using scissors and forceps59,87. Ventricles were placed in the ice-cold PBS buffer with 1% BSA. After several washes removing blood cells, ventricles were gently cut into small pieces. Tissues were incubated in digestion buffer (0.5 ml HBSS plus 0.26 μ/ml Liberase DH [Roche]) for 45 min at 37 °C while agitating at 750 rpm on an Eppendorf ThermoMixer. Supernatants were collected every 15 min and neutralized with 10% sheep serum, and cells were completely disaggregated by gently pipetting up and down. The dissociated cells were centrifuged at 200 × g for 5 min at 4 °C and re-suspended in HBSS with 0.05% BSA. The cell suspension (2 ml) was then gently added on top of 2 ml HBSS buffer with 7.5% BSA. After spinning down at 200 × g for 5 min at 4 °C, the pellet was resuspended in 500 ml and filtered through a 35 μm strainer. Whole kidney marrow cells were isolated using an established protocol88. Briefly, kidneys were collected on ice and homogenized immediately in PBS plus 0.05% BSA by using a 10 ml syringe with 18.5 G. The dissociated cells were washed twice in PBS plus 0.05% BSA and spun down at 300 × g for 5 min at 4 °C. Pellets were re-suspended in PBS plus 0.05% BSA and filtered through a 35 μm strainer. Livers were collected on ice and incubated in digestion buffer (0.5 ml HBSS plus 0.13 U/ml Liberase DH [Roche]) at room temperature. The cell lyses was gently stirred with a Spinbar® magnetic stirring bar (Bel-Art Products), and the supernatant were collected every 5 min. The cell lysate was filtered through a 70 μm strainer and centrifuged at 150 × g for 3 min at 4 °C. Pellet was resuspended in PBS plus 0.05% BSA and then gently added on top of HBSS buffer with 7.5% BSA. After spinning down at 300 × g for 5 min, pellets were washed one more time with PBS plus 0.05% BSA. After centrifuging at 300 × g for 5 min, the resuspended cell lyses was filtered through a 35 μm strainer. Pancreatic cells were isolated using the same protocol as liver cells except for the initial spinning down at 300 × g for 5 min. All suspended cells were stained with trypan blue and counted using a hemocytometer. The isolated cells were sent to the Epigenomics Core Facility of Weill Cornell Medicine for single-cell RNA-seq library preparation using the 10x Genomics Chromium Single Cell 3’ GEM, Library & Gel Bead Kit v3, and Chromium Single Cell B Chip Kit. The libraries were sequenced on a pair-end flow cell with a 2 × 50 cycles kit on Illumina NovaSeq6000. We sequenced 2 replicates of all samples (24 total). scRNA-seq analysis The raw reads were aligned and processed with the CellRanger pipeline (v6.0.0) using the mouse (mm10) or zebrafish genomes (GRCz11). Subsequent analyses were performed in R following the recommendations of Amezquita et al. (https://osca.bioconductor.org/) using numerous functions provided in the R packages scater89 (v1.24.0) and scran90 (v1.22.1). Briefly, quality control was carried out for each sample separately with functions from the scuttle89 package; cells with low gene content and high mitochondrial gene content were removed from further analyses. For each species, count matrices across all samples were scaled for sequencing depth differences, log-transformed, and integrated using multiBatchNorm and fastMNN functions from the batchelor91 package (v1.10.0). Dimensionality reductions were done with scater (runUMAP) using the batch-corrected values. Cell types were annotated with SingleR37 (v1.8.1) using numerous datasets and manual inspection of marker genes. For zebrafish annotation with SingleR, the following datasets were used as references: pancreas (GSM3509161), WKM (GSE100911)92, and heart (GSE158919)93. For mouse annotation with SingleR, the following datasets were used as references: pancreas (GSE84133)94, kidney (GSM2967051, GSE119531)95, liver (GSE125688, GSE171904)96,97, and heart (GSM2967050, E-MTAB-6173)98. The following referecnes were also used for the mouse cell type markers: endothelial cells and pericytes99, epithelial cells100, fibroblasts101, lymphocytes102, myeloid cells as a whole, and macrophages, dendritic cells103, granulocytes104, hepatocytes105, natural killer cells106, pancreatic cells94,107. References for specific maker genes are listed in the text. Data was also explored and annotated with Cellxgene version v0.16.8. scRNA-seq data UMAPs, SAMaps, dot plots, and violin plots were generated with Cellxgene VIP (10.1101/2020.08.28.270652). Cell proportions were calculated by dividing the number of cells of a specific type by the total number of cells in the sample. Gene ontology analysis was performed on marker genes identified between clusters of the same cell type. Cross-species comparisons SAMap65 (v1.0.2) was used to determine cell homology between mouse and zebrafish mono/Mϕ clusters and hepatocytes. Briefly, a reciprocal BLAST map between mouse and zebrafish transcriptomes was generated using the map_genes.sh script from SAMap, which ran tblastx (nucleotide-nucleotide BLAST, NCBI) in both directions with respect to Ensembl zebrafish and mouse transcriptomes (Danio_rerio.GRCz11.cds.all.fa and Mus_musculus.GRCm38.cds.all.fa). Raw scRNA-seq count matrices of zebrafish and mouse were extracted from SingleCellExperiment objects with cluster annotations and converted to h5ad format using the sceasy R package (v0.0.7). Raw data were then processed and integrated by SAMap. Mapping scores between the mono/Mϕ clusters and hepatocytes of each species were calculated with SAMap get_mapping_scores (n top=0) and visualized in R using ggalluvial; cluster pairs with alignment scores less than 0.2 were excluded. For hepatocytes and mono/Mϕ cluster pairs with high inter-species alignment scores (mm1-dr4, mm4-dr6, and mm6-dr3), marker genes for each cell population were assessed using scran::scoreMarkers (mean AUC > 0.6), and over-representation analyses were performed using clusterProfiler108 (v4.2.2) with respect to gene ontology terms. Gene ontology terms with semantic similarity were reduced using the rrvgo (v1.6.0) R package (with similarity threshold set to 0.5). Sayols, S (2023). rrvgo: a Bioconductor package for interpreting lists of Gene Ontology terms. microPublication Biology. 10.17912/micropub.biology.000811. To compare the transcriptional response to MI (mouse) or cardiac cryoinjury (zebrafish) between the hepatocytes and the mono/Mϕ cluster pairs, differential expression analysis was performed using scran::findMarkers(lfc = 0.25) at days 1 and 7 in the heart and in the liver. Genes were considered statistically significant at an FDR cutoff of 0.10. Zebrafish histology and microscopy Freshly collected zebrafish tissues were fixed with 4% paraformaldehyde (PFA) overnight at 4 C and applied to cryosection at a 10 μm thickness. Hybridization Chain Reaction (HCR 3.0) staining of whole-mounted hearts or cryosections was done following the published protocols61. HCR probes for havcr1, moxd1, cd9b, fn1a, and gda were synthesized by Molecular Instruments Inc. For immunostaining of whole-mounted hearts or heart sections109,110, samples were blocked with 2% bovine serum albumin (BSA, VWR, cat#97061), 1% DMSO, 0.5% goat serum (ThermoFisher, cat#16210) and 0.5% Triton X-100 in PBS for 1 h at room temperature (RT). Primary antibodies were diluted in the blocking buffer and incubated with hearts overnight at 4 C. Hearts were then washed with PBS plus 0.1% Tween 20 and incubated with the secondary antibody diluted in the blocking buffer for 1.5 h at room temperature. Hearts were stained with DAPI (ThermoFisher, D3571) to visualize nuclei. The primary antibody used in this study was rabbit anti-lcp1 (GeneTex GTX124420, 1:200), the secondary antibody used in this study was Alexa Fluor 488 goat anti-rabbit (ThermoFisher, 1:200). IB4 (Vector Laboratories, DL-1207, 1:100) staining was perfomed by standard procedures63. All antibodies are commercially available and are validated by suppliers. Fluorescent images of tissue sections were captured used a Zeiss 800 confocal microscope (Zen 2.6 blue edition software). Statistics and reproducibility Unless otherwise stated, data are presented as mean ± s.e.m. Data are derived from multiple experiments, and measurements were taken from distinct samples (animals) unless stated otherwise. If not mentioned otherwise in the figure legend, statistical significance is indicated by *P < 0.05, **P < 0.01 and ***P < 0.001. Statistical analysis was carried out assuming normal distribution after visualizing the data in histograms and confirming a normal distribution. Data was plotted in a QQ plot when histograms were not clear. Non-parametric analyses were used when normal data distribution could not be assumed. Statistical significance was tested with unpaired, two-tailed t-test, and one-way or two-way ANOVA with Tukey’s honest significant difference (HSD) post hoc test for multiple comparisons when one or two independent variables, respectively, were compared. GraphPad Prism 7 was used for statistical analysis. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary information Description of Additional Supplementary Files Supplementary Data 1 Supplementary Data 2 Supplementary Data 3 Supplementary Data 4 Supplementary Data 5 Supplementary Data 6 Supplementary Data 7 Supplementary Data 8 Supplementary Data 9 Supplementary Data 10 Supplementary Data 11 Supplementary Data 12 Reporting summary Acknowledgements We thank Adedeji A. Afolalu, Chaim Shapiro, Soji Hosten, and Chelsea Quaies for fish care. We thank the Epigenomics Core of Weill Cornell Medicine for their technical support. This work was supported by Rudin Foundation fellowships to Y.X. and J.Y., a predoctoral training grant position in Stem Cell Biology and Regenerative Medicine from New York State Stem Cell Science program (NYSTEM) to B.P. The research was supported by Chan Zuckerberg Initiative grant (DAF2020-217734) to D.B., J.C, and J.C.L., Weill Cornell Start-up fund and NIH grants (R01HL155607 and R01HL166518) to J.C. J.C.L. is supported by NIH grants R01 DK121844, R01 DK121140, and R01 DK132879. We thank G.S.P. and M.H. for constructive feedback on the manuscript. Author contributions E.C., J.Y., D.B., J.C., and J.C.L. designed the study and wrote the manuscript with input from all authors. E.C., J.Y., Y.X., A.R.-N., B.Y., B.P., M.Q., E.A.H., L.S., performed and analyzed the animal experiments. P.Z., F.D., E.C., and A.M.P. analyzed the scRNA-Seq experiments. J.C.L, J.C., and D.B. conceived and supervised the study and acquired funding for the work. Peer review Peer review information Communications Biology thanks the anonymous reviewers for their contribution to the peer review of this work. Primary Handling Editor: Dario Ummarino. Data availability The scRNA-seq datasets generated in this study have been deposited at the Cellxgene platform (https://cellxgene.cziscience.com/collections/24e324a1-b42d-4438-9bb5-cf5233fa90b0) as well as the NCBI’s Gene Expression Omnibus under accession number GSE227191. Numerical data used for plots can be found in Supplementary Datas 1-12. Code availability Scripts and code used for processing and annotating the scRNA-seq are deposited at Github (https://github.com/abcwcm/CZI_MI_mouseZebrafish) and Zenodo111. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. These authors contributed equally: Eric Cortada, Jun Yao, Yu Xia. Contributor Information Doron Betel, Email: dob2014@med.cornell.edu. Jingli Cao, Email: jic4001@med.cornell.edu. James C. Lo, Email: jlo@med.cornell.edu Supplementary information The online version contains supplementary material available at 10.1038/s42003-024-07315-x. References Associated Data Supplementary Materials Supplementary information Description of Additional Supplementary Files Supplementary Data 1 Supplementary Data 2 Supplementary Data 3 Supplementary Data 4 Supplementary Data 5 Supplementary Data 6 Supplementary Data 7 Supplementary Data 8 Supplementary Data 9 Supplementary Data 10 Supplementary Data 11 Supplementary Data 12 Reporting summary Data Availability Statement The scRNA-seq datasets generated in this study have been deposited at the Cellxgene platform (https://cellxgene.cziscience.com/collections/24e324a1-b42d-4438-9bb5-cf5233fa90b0) as well as the NCBI’s Gene Expression Omnibus under accession number GSE227191. Numerical data used for plots can be found in Supplementary Datas 1-12. Scripts and code used for processing and annotating the scRNA-seq are deposited at Github (https://github.com/abcwcm/CZI_MI_mouseZebrafish) and Zenodo111.