hapln1a+ cells guide coronary growth during heart morphogenesis and regeneration

Although several tissues and chemokines orchestrate coronary formation, the guidance cues for coronary growth remain unclear. Here, we profile the juvenile zebrafish epicardium during coronary vascularization and identify hapln1a+ cells enriched with vascular-regulating genes. hapln1a+ cells not only envelop vessels but also form linear structures ahead of coronary sprouts. Live-imaging demonstrates that coronary growth occurs along these pre-formed structures, with depletion of hapln1a+ cells blocking this growth. hapln1a+ cells also pre-lead coronary sprouts during regeneration and hapln1a+ cell loss inhibits revascularization. Further, we identify serpine1 expression in hapln1a+ cells adjacent to coronary sprouts, and serpine1 inhibition blocks vascularization and revascularization. Moreover, we observe the hapln1a substrate, hyaluronan, forming linear structures along and preceding coronary vessels. Depletion of hapln1a+ cells or serpine1 activity inhibition disrupts hyaluronan structure. Our studies reveal that hapln1a+ cells and serpine1 are required for coronary production by establishing a microenvironment to facilitate guided coronary growth.

2) The major issue by far is the lack of evidence that there is anything functionally specific about the hapln1a+ cells. The data showing association with tips is not very compelling in the images shown, and the authors could do a more rigorous job to quantify this.
3) More importantly, no experiments are carried out to show that there is anything special about the hapln1+ cells beyond an association. The ablation of tcf21+ cells has been shown to impede coronary vessel growth and regeneration in previous publications. Here the authors show similar effects with ablation of hapln1a+ cells. However, the hapln1a+ cells comprise a significant part of the epicardium (33% in homeostasis, 50% in regenerating hearts??). The proper control is to ablate a similar proportion of hapln1a-cells, to determine if the role in coronary growth is really specific to the hapln1a+ population. This is not a small ask, but is really essential for interpreting the results especially for a journal with impact of Nature Comm. At the very least, one could compare a partial 50% random ablation using tcf21:NTR with the full ablation of hapln1a:NTR. Does the latter show a much stronger phenotype? 4) Details for inhibition of Serpine1 were missing. Is the concentration used equivalent to just what is needed to inhibit activity? Can inhibition of activity be measured? A complementary genetic approach would be highly desirable to exclude chemical off-targets.

Minor issues
1) The term "shear structures" needs explanation as it is not a commonly used descriptor. Is there some evidence for causation of shear stresses?
2) What does it mean that vessel number is increased by hapln1a cell ablations? Are these different vessels or fragments of previous vessels?
Reviewer #3 (Remarks to the Author): In this study Sun et al identify a subset of epicardial cells that express Hapln1a which are associated with sprouting coronary vessels in the juvenile and regenerating zebrafish hearts, and use ablation experiments to show that Hapln1a is required for coronary growth in both scenarios. They further identify that serpine1, which encodes for a negative regulator of plasminogen activator, is expressed in hapln1a+ epicardial cells, is found at coronary tips, and is required for vascularisation both in coronary development and regeneration.
Previous studies have identified that epicardial subtypes exist, that the epicardium supports coronary development, that hapln1a is expressed in the epicardium, that hapln1a is required for heart regeneration, and that serpine1 also promotes heart regeneration. However, little is understood about heterogeneity in the epicardium, how the epicardium supports coronary growth, and which epicardial cells are required. In this context, the study by Sun et al contributes new knowledge to the field of coronary development, and unifies and extends previous studies on the role of epicardium in vascularisation and revascularisation in development and injury. The experiments are well designed and executed, and the data is well presented. However, the depth of mechanistic analysis of the roles of hapln1a and serpine1 in vascularisation, particularly in terms of their functions and relationship, is a little unsatisfactory. What is the impact of Hapln1a and Serpine1 LOF on the ECM around the vasculature for example? How/why do these ECM modifications support vessel formation? And what is the relationship between them? While further experiments would help to provide better insights into the specific role of the Hapln1a+ Serpine1+ epicardium in supporting coronary (re)development, the authors also do not really dive into these questions in the discussion, which is an omission. Taking steps to address the mechanism would significantly improve the impact of the manuscript.
Major comments: Relating to the comment above, the authors do not investigate changes in ECM in either the Hapln1a or Serpine1 models, or discuss this mechanism of action of each gene. The authors could assess for example HA presence/organisation (as in their previous Circulation Hapln1a epicardium paper) around the vessels in the juvenile stage in wild type and upon ablation of Hapln1a+ cells or Tiplaxtinin treatment. In Line 339 -the authors also suggest that hapln1a regulates coronary vessel development through Serpine1 and that they will test this -of course this will not be happening at a transcriptional level, and since the authors also don't look in detail at the mechanisms of action of either protein, or the impact on the extracellular environment (showing only that they both promote coronary development), the experiments that they perform subsequently don't really address this hypothesis.
It appears that not all vessel tips are associated with Hapln1a (or even Serpine1) for example the 6 hour timepoint shown in Figure 2 -there appear to be sprouts forming without any Hapln1a. Related to this, some revascularisation still occurs in the hapln1a KO. Does this mean Hapln1a is only required by a subset of branching vessels? To investigate this, the authors should quantify the number of tips with/without Hapln1a and Serpine1, and comment on this possible heterogeneity.
There are some assertions or observations in the paper that could be better supported through the extensive single cell sequencing data already included in the paper. For example: Figure 1 -there is a lack of consistency in the colour coding used in the heatmaps -sometimes blue represents the highest value, sometimes the lowest -this could be consistent within the figure to simplify. Figure 4 -The authors show that the vessel density and junction number (graphs D and E) are reduced, which is in line with the observation that vascular growth is less in the hapln1a ablation model. However, vessel number per area is increased (F), which seems contradictory to the previous quantifications. Can the authors explain the difference between these two quantifications? Figure 4 and 5 -the graphs are distorted along the x axis -aspect ratio should be maintained. Figure 6 -The serpine1 expression shown in the images doesn't seem to colocalise with Hapln1a at terminal branches (rather it is localised more centrally within the Hapln1a network) -is Serpine1 expressed broadly within the Hapln1a network as well as around tip cells?
Below is a point-by-point response to the reviewers' comments.

Reviewer 1
This paper describes a close correlation between growing coronary vessels during zebrafish heart development and regeneration with a subpopulation of epicardial cells that are positive for hapln1a. In a series of convincing, high quality imaging in vivo and ex vivo the authors in particular show that hapln1a+ cells appear to guide coronary vessel growth. Genetic ablation of hapln1 shows that these cells are indeed important for coronary vessel growth. Several single cell sequencing profiles are presented (for epicardial cells and hapln1a+ cells) which are a nice resource for further studies. While these results are very interesting, since they add to the author's previous findings that the hapln1a+ epicardium is also important for other events during heart regeneration, in particular for cardiomyocyte proliferation, I find the molecular mechanistic insight into how hapln1a+ epicardial cells regulate coronary vessel growth not substantial enough for Nat. Comm. A: We appreciate the reviewer's comments and per the reviewer's suggestions we have performed additional experiments to further analyze the effect of hapln1a + cells and serpine1 on the ECM component hyaluronan (HA) for further understanding of the underlying mechanism. With our new analyses, we observed HA lining coronary vessel extensions and also preceding these vessels. Further, we discovered that depletion of hapln1a + cells disrupted the HA organization and inhibiting serpine1 activity led to a disorganized HA structure. Our new experiments indicate that HA deposition by hapln1a + cells and then organization by serpine1-expressing cells is required for coronary growth during morphogenesis and regeneration. Although previous tissue engineering studies have demonstrated that HA-based materials are sufficient to stimulate angiogenesis and vascular growth and HA is required for zebrafish coronary re-vascularization (Brandes et al., JCI 1991;Peattie et al., Europe PMC 2004;Turner et al., Biomaterials 2004;Lepidi et al., Eur J Vasc Endovasc Surg 2006;Pardue et al., Organogenesis 2008;Burdick and Prestwich, Adv Mater 2011;Munch et al., Development 2017;Ghose et al., BMC Cancer 2018;Luo et al., Front Bioeng Biotechnol 2022), there is no direct observation of organized HA structures along and preceding coronary growth in vivo. Further, no reports exist regarding the regulation mechanism of HA by hapln1a + cells and serpine1. We have updated the manuscript to include these new results on pages 14-16, and in Fig. 9 of the revised manuscript: "To elucidate the mechanism by which hapln1a + cells and serpine1 activity impact coronary growth, we assessed the distribution of the hapln1a + cell substrate, hyaluronic acid (HA). We recently reported that hapln1a + epicardial cells regulate HA deposition to facilitate cardiomyocyte proliferation during heart morphogenesis and regeneration [48]. HA is required for coronary revascularization during zebrafish heart regeneration [79][80][81]. Previous studies also demonstrated that endothelial cells can attach and grow along HA fibers in vitro, and a scaffold in tissue-engineered vascular grafts formed from HA-based biomaterials can completely generate a new vascular tube. These studies indicate that the administration of HA can enhance cell proliferation, adhesion, tubular sprout formation, and the migration of endothelial cells [79][80][81][82][83][84][85]. We speculate that HA is involved in guided coronary growth. To examine this possibility, we first assessed HA localization with coronary vessels on the ventricular surface in juvenile deltaC:EGFP animals. We detected strong association of linear HA signals with EGFP + signals (Fig. 9a). Next, we examined the HA deposition after ablating hapln1a + cells at 7 wpf. Juvenile hapln1a:mCherry-NTR;deltaC:EGFP animals and deltaC:EGFP siblings were treated with 10 mM Mtz and their hearts were collected for histological analysis. We assessed the size of HA aggregates by quantifying the area of each focus of HA and found that hapln1a + cell-depleted hearts displayed 35% lower HA intensity per HA aggregate, when compared with control siblings (Fig. 9b), indicating that the ablation of hapln1a + cells disrupted the linear structure of HA in the coronary growth area. Next, we examined HA signals in the regeneration area and found HA signals were closely associated with deltaC:EGFP + cells (Fig. 9c). We then assessed the HA deposition within the regenerating area after depleting hapln1a + cells and observed that 7 dpa injury sites of hapln1a + cell-depleted animals displayed ~47% smaller puncta of HA, compared with wild-type siblings (Fig. 9d). Next, we assessed whether serpine1 functions through regulating the HA structure during morphogenesis and regeneration, as previous reports indicated serpine1 protects the ECM to maintain a matrix necessary for endothelial cells to migrate and form capillaries [86]. We examined HA deposition in the ventricle surface after treating juvenile deltaC:EGFP fish with Tiplaxtinin for 12 hours and continuously for 7 days and found that inhibitor treated hearts displayed a 64% lower HA intensity per HA aggregate, when compared with control siblings ( Fig. 9e and f), revealing a disorganized HA linear structure within the coronary growth area. Lastly, we treated adult fish with Tiplaxtinin from 3 dpa for 12 hours and continuously for 4 days and observed 53% lower HA intensity per HA aggregate at 7 dpa, compared with vehicle controls ( Fig. 9g and h).
Our results indicate a mechanism for hapln1a + cell function during morphogenesis and regeneration, in which HA is deposited by hapln1a + cells and then organized by serpine1expressing cells for coronary growth during vascularization and revascularization." While the authors present serpin1 as a candidate factor produced by the hapln1a+ cells that guides vessel growth, this hinges on drug inhibitor experiments only. As far as I can tell, whether the drug actually inhibits zebrafish serpin and how specific it is, has not been shown. Thus, additional evidence, ideally by genetic means should be provided. I'm aware that this is technically very challenging (e.g. if non-conditional mutants are embryonic lethal), but nevertheless think that the paper in its current form is somewhat too light on molecular mechanisms for Nat. Comm. A: We agree with the reviewer that a genetic approach is an excellent means to exclude possible chemical off-target effects and drug non-specific effects/efficiency. As the reviewer mentioned, genetic means of manipulating adult zebrafish genes are timeconsuming and challenging/risky: 1) more than 1 year to develop a CRISPR/Cas9 serpine1 mutant generation and several generation outcrosses; and 2) a serpine1 mutant also does not guarantee that there will be no compensation effects. For these reasons, we chose a pharmacological inhibition approach to perform functional analysis of serpine1 during heart regeneration. We chose the drug Tiplaxtinin as its serpine1 inhibition activity has been previously measured and was utilized for the pharmacological inhibition of serpine1 activity and published by different research groups (Gorlatova et al., JBC 2007;Daniel et al., PLos One 2015;Kang et al., Oncotarget 2016;Munch et al., Development 2017;Lin et al., Cell Rep. 2020;Yamagami et al., BBRC 2020). Natalia V. Gorlatova demonstrated the capacity for Tiplaxtinin to block serpine1 activity, and IC50 values for serpine1 inactivation by Tiplaxtinin are from 9 to 12 μM. The inhibitory effect of 20 µM Tiplaxtinin for serpine1, the concentration which we used in this MS, is about 95%. As the means to examine possible downstream effectors of serpine1, like urokinase plasminogen activator (uPA) and tissue plasminogen activator (tPA) is currently still lacking in zebrafish, we are unable to examine the effect of serpine1 inhibition. However, we performed experiments to exclude the possible toxic effects of Tiplaxtinin by examining coronary growth after removing the drug. We found that inhibited coronary growth restarted after removing the inhibitor. We have now included these data on pages 13-14, and in Supplementary Fig. 8 of the revised manuscript.
"…by treatment with its antagonist Tiplaxtinin, which blocks Serpine1 protease and has been used to inhibit serpine1 activity in different systems, including during zebrafish heart regeneration [70,[73][74][75][76][77]. Previous studies measured and determined that the inhibitory effect of 20 µM of Tiplaxtinin on serpine1 is around 95% [73,78]." "Moreover, these blocked coronary extensions can re-start after drug removal, indicating that the coronary growth restriction by serpine1 inhibition does not occur by coronary cell death ( Supplementary Fig. 8)." Additional major issues: 1) since development and growth of fish is heavily influenced by environmental factors, all stage descriptions should not be given in time after fertilization, but in the "standardized standard length" scheme established by David Parichy. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3030279/ A: We have now added the length information of fish we used at both juvenile and adult stages in the revised manuscript, on page 19, paragraph 2: "… transgenic zebrafish were used at the age of 6-7 weeks with a length between 1.5 -1.8 cm (juvenile), or 4-12 months with a length of at least 3 cm (adult)." and page 24, paragraph 1.
"Juvenile hearts were collected at 7 wpf (1.6 -1.8 cm length). The hearts of adult fish at 6 months old (3.5 -4 cm length)..." 2) Methods: transgenic lines must be identified by their unique allele designation given by the Zebrafish nomenclature committee (ZFIN) A: The information for all transgenic lines according to their unique allele designation given by the ZFIN are now included in the revised manuscript, on page 19, paragraph 2: "Transgenic strains described elsewhere include: Tg(gata4:EGFP) ae1Tg and Tg(tcf21:nucEGFP) pd41Tg [100]; Tg(hapln1a:EGFP) pd338Tg and Tg(hapln1a:mCherry-NTR) em14Tg [48]; and Tg(deltaC:EGFP) em11Tg and Tg(deltaC:mCherry) em11Tg [47]." 3) Methods are generally too sparse and won't allow repetition of experiments since essential information is missing. E.g. the description of the FACS for single-cells preparation should include information on how debris/dead cells were excluded, gates should be published in the supplement.

A:
The updated manuscript now provides a revised but concise description in the methods section that is sufficient for experimental repetition. The information for FACS for scRNA-seq is now included in the Methods and supplementary data. The updated Methods information is now available on pages 19-24, and Supplementary Fig. 1 in the revised manuscript. 4) Where is the data supporting this statement? "...notable difference in cluster 3 with high levels of cell proliferation related genes". A: In the revised manuscript, we have now included feature plots for frzb, mustn1a and tgfbi expression in tcf21 + clusters of the hearts at 7 wpf. These genes were previously indicated to be associated with the cell proliferation process. We have included this information in page 6, paragraph 1, and in Supplementary Fig. 2 in the revised manuscript.
"…we observed a notable difference in cluster 3 with high levels of expression of genes related to cell proliferation, such as frzb, mustn1a, and tgfbi [31][32][33]." 4) Line 204: "The expression of ciarta and dbpb in cluster 5 cells suggests an involvement in the rhythmic process". Please elaborate; which "rhythmic process"? A: Both ciarta and dbpb are suggested to have a role in the circadian rhythm. This information is now included in the Results section on page 6, paragraph 1: "The expression of ciartα and dbpβ in cluster 5 suggests an involvement in the circadian rhythm process [38,39]." 5) "we observed significant hapln1a expression in clusters 1 and 3". Fig. 1D shows hapln1a also in clusters 4 and 5; what's the basis for ignoring these and for focusing on 1 and 3? A: We performed Violin plot analysis comparing the expression of hapln1a in tcf21 + cell clusters. hapln1a shows high expression in cluster 1, slight expression in clusters 3, 4, 5, and little expression occurs in clusters 2 and 6. We have now revised the description on page 7, paragraph 1: "As cluster 1 is the most heavily represented of these 3 cell states, we first focused on cluster 1 and observed high expression of hapln1a in this cluster [23], with slight expression in clusters 3, 4 and 5 ( Fig. 1d and Supplementary Fig. 3)." 6) Fig. 2F: text and figure legends mention "AcT-Hu+" cells, methods mention antiacetylated tubulin and anti-HuC/HuD antibodies. 2F shows the "AcT-Hu+" cells in the red channel. Authors must clarify how this was done. Were hearts co-stained with both antibodies, which were both detected in the red channel? How do authors then know these cells are double positive for acT and Hu? A: We apologize for the confusion and have provided an explanation in the revised manuscript and here. This assay used a combination of antibodies against AcT and HuC/D to identify intracardiac neuronal somata and axons. The utility of this label combination to detect neuronal somata has been established in previous neuroanatomical studies such as in the zebrafish intestine (Bisgrove et al., J Neurobiol 1997; Olsson, Acta Histochemica 2009) and in the goldfish heart (Newton et al., JCN 2014). In order to label all neuronal somata, for regional and total counts of neurons, we employed this AcT-Hu immunohistochemistry in this study. With this method, both AcT and Hu primary antibodies are recognized through the utilization of an Alexa-Fluor 594 goat anti-mouse secondary antibody and then visualized together with the Zeiss LSM800 confocal microscope. As we sought to examine the possible relationship of coronary vessels with all nerves, we only examined the nerves as combined AcT+ and Hu+ signals instead of attempting to examine AcT+ or Hu+ signals in isolation. We included the description in page 9, paragraph 1: "… an established label combination which has previously been used to detect neuronal somas in neuroanatomical studies in the zebrafish intestine and goldfish heart [60][61][62]."

Reviewer 2
In this manuscript, Sun et al. describe a role for hapln1a+ cells for guiding coronary artery growth during juvenile heart development and adult heart regeneration in the zebrafish model system. They make use of nice reporter and conditional cell ablation tools to show that hapln1a+ cells of the epicardium are associated as "shear structures" at the tip ahead of coronary endothelium, in contrast to macrophages, neurons, or cardiomyocytes. Using extracted cultured hearts they provide evidence for tracking of vessels along a hapln1a pathway, and that following ablation of hapln1a+ cells (via NTR-MTZ killing) the vessel growth fails. Similar results are shown for regenerating adult hearts. Serpine1 is identified as a gene expressed in the hapln1a+ cells, and a chemical blocker of Serpine1 also blocked artery growth and regeneration. The manuscript is well written and the topic is important.
Major Issues.
1) The point of scRNA-seq described in Fig. 1 is unclear. The authors published the same analysis previously earlier this year, although somehow a different number of clusters were identified. Other scRNA-seq analyses have been published by the Poss and Cao groups using presumably the exact same cell populations (tcf21+ epicardium). Why are different clusters being shown than was shown previously and for what purpose? A: We apologize for the confusion and provide an explanation here, as well as in the revised manuscript. Previously, the scRNA-seq analyses of the epicardium from the Poss Lab, Cao Lab, Riley Lab, and our lab (Wang Lab) were performed with the epicardium of the embryonic heart, adult uninjured heart, or adult regenerating hearts. The scRNA-seq analysis performed in this manuscript utilized the juvenile zebrafish heart, a developmental stage at which scRNA-seq of the epicardium has not been reported. As cardiac regeneration recapitulates many aspects of heart morphogenesis and critical biological events like coronary vascularization and myocardial compaction occur during this juvenile stage, our scRNA-seq analysis provided in this manuscript will be informative to thoroughly understanding epicardial function at different life stages and during the regeneration process. The information regarding the use of juvenile zebrafish hearts for tcf21 + scRNA-seq is provided on page 6, paragraph 1 of revised manuscript: "Zebrafish coronary growth establishes a dense vasculature network from 5 to 6 weeks post-fertilization (wpf) until the adult stage, a period during which latent epicardial clusters are also undergoing development [23]. As cardiac regeneration recapitulates many aspects of heart morphogenesis and detection of epicardial cell clusters at the juvenile stage has not been reported, we performed scRNA-seq analysis of tcf21 + cells [25] in zebrafish hearts at 7 wpf (Supplementary Fig. 1) to explore juvenile epicardial clusters and their potential effects on coronary growth." More importantly, the data shown does not meet the description. For example, the authors claim enrichment of hapln1a cells in clusters 1 and 3. There is no obvious enrichment in cluster 3. Violin plots were done for many genes but strangely not for hapln1a, but anyway it is unclear why this data is shown and what value it adds. A: We carefully checked our data and the corresponding description in the revised manuscript to make sure the data shown meet the description. We agree with the reviewer that the manuscript requires a Violin plot for hapln1a. We performed Violin plot analysis for hapln1a and observed high expression of hapln1a in cluster 1, slight expression in clusters 3, 4, 5, and little expression in clusters 2 and 6. We have revised the description on page 7, paragraph 1 of the updated manuscript: "As cluster 1 is the most heavily represented of these 3 cell states, we first focused on cluster 1 and observed high expression of hapln1a in this cluster [23], with slight expression in clusters 3, 4 and 5 ( Fig. 1d and Supplementary Fig. 3)." 2) The major issue by far is the lack of evidence that there is anything functionally specific about the hapln1a+ cells. The data showing association with tips is not very compelling in the images shown, and the authors could do a more rigorous job to quantify this. A: We appreciate the reviewer's comments and per the reviewer's suggestions we have performed additional experiments to further analyze the effect of hapln1a + cells and serpine1 on the ECM component hyaluronan (HA) for further understanding of the underlying mechanism. With our new analyses, we observed HA lining coronary vessel extensions and also preceding these vessels. Further, we discovered that depletion of hapln1a + cells disrupted the HA organization and inhibiting serpine1 activity led to a disorganized HA structure. Our new experiments indicate that HA deposition by hapln1a + cells and then organization by serpine1-expressing cells is required for coronary growth during morphogenesis and regeneration. Although previous tissue engineering studies have demonstrated that HA-based materials are sufficient to stimulate angiogenesis and vascular growth and HA is required for zebrafish coronary re-vascularization ( Further, no reports exist regarding the regulation mechanism of HA by hapln1a + cells and serpine1. We also conducted a more rigorous analysis on the position of hapln1a + cells and coronary vessel sprouts and these additional quantifications are now included in Fig. 2d of the revised manuscript. We have updated the manuscript to include these new experiments on pages 14-16, and in Fig. 9 of the revised manuscript: "To elucidate the mechanism by which hapln1a + cells and serpine1 activity impact coronary growth, we assessed the distribution of the hapln1a + cell substrate, hyaluronic acid (HA). We recently reported that hapln1a + epicardial cells regulate HA deposition to facilitate cardiomyocyte proliferation during heart morphogenesis and regeneration [48]. HA is required for coronary revascularization during zebrafish heart regeneration [79][80][81]. Previous studies also demonstrated that endothelial cells can attach and grow along HA fibers in vitro, and a scaffold in tissue-engineered vascular grafts formed from HA-based biomaterials can completely generate a new vascular tube. These studies indicate that the administration of HA can enhance cell proliferation, adhesion, tubular sprout formation, and the migration of endothelial cells [79][80][81][82][83][84][85]. We speculate that HA is involved in guided coronary growth. To examine this possibility, we first assessed HA localization with coronary vessels on the ventricular surface in juvenile deltaC:EGFP animals. We detected strong association of linear HA signals with EGFP + signals (Fig. 9a). Next, we examined the HA deposition after ablating hapln1a + cells at 7 wpf. Juvenile hapln1a:mCherry-NTR;deltaC:EGFP animals and deltaC:EGFP siblings were treated with 10 mM Mtz and their hearts were collected for histological analysis. We assessed the size of HA aggregates by quantifying the area of each focus of HA and found that hapln1a + cell-depleted hearts displayed 35% lower HA intensity per HA aggregate, when compared with control siblings (Fig. 9b), indicating that the ablation of hapln1a + cells disrupted the linear structure of HA in the coronary growth area. Next, we examined HA signals in the regeneration area and found HA signals were closely associated with deltaC:EGFP + cells (Fig. 9c). We then assessed the HA deposition within the regenerating area after depleting hapln1a + cells and observed that 7 dpa injury sites of hapln1a + cell-depleted animals displayed ~47% smaller puncta of HA, compared with wild-type siblings (Fig. 9d). Next, we assessed whether serpine1 functions through regulating the HA structure during morphogenesis and regeneration, as previous reports indicated serpine1 protects the ECM to maintain a matrix necessary for endothelial cells to migrate and form capillaries [86]. We examined HA deposition in the ventricle surface after treating juvenile deltaC:EGFP fish with Tiplaxtinin for 12 hours and continuously for 7 days and found that inhibitor treated hearts displayed a 64% lower HA intensity per HA aggregate, when compared with control siblings (Fig. 9e and f), revealing a disorganized HA linear structure within the coronary growth area. Lastly, we treated adult fish with Tiplaxtinin from 3 dpa for 12 hours and continuously for 4 days and observed 53% lower HA intensity per HA aggregate at 7 dpa, compared with vehicle controls (Fig. 9g and h).
Our results indicate a mechanism for hapln1a + cell function during morphogenesis and regeneration, in which HA is deposited by hapln1a + cells and then organized by serpine1expressing cells for coronary growth during vascularization and revascularization." 3) More importantly, no experiments are carried out to show that there is anything special about the hapln1+ cells beyond an association. The ablation of tcf21+ cells has been shown to impede coronary vessel growth and regeneration in previous publications. Here the authors show similar effects with ablation of hapln1a+ cells. However, the hapln1a+ cells comprise a significant part of the epicardium (33% in homeostasis, 50% in regenerating hearts??). The proper control is to ablate a similar proportion of hapln1acells, to determine if the role in coronary growth is really specific to the hapln1a+ population. This is not a small ask, but is really essential for interpreting the results especially for a journal with impact of Nature Comm. At the very least, one could compare a partial 50% random ablation using tcf21:NTR with the full ablation of hapln1a:NTR. Does the latter show a much stronger phenotype? A: The reviewer is insightful to point out this possibility. Currently, we have no method available to specifically ablate hapln1acells. Therefore, we performed a partial (50%), random ablation of epicardial cells as suggested by the reviewer. After depleting around 50% of tcf21 + cells, we did not observe a significant difference in coronary vascularization during morphogenesis and revascularization after regeneration, compared with controls. These data indicate that ablating most hapln1a + cells has a stronger phenotype on coronary growth, compared with spatial ablation of tcf21 + cells. We have included these new experiments on page 12, paragraph 1 and in Supplementary Fig. 5 of the revised manuscript: "As hapln1a + cells form around 30% of tcf21 + cells during morphogenesis (Fig. 1d) and around 50% during regeneration [23], we also examined the effect of randomly ablating around 50% of tcf21 + cells ( Supplementary Fig. 5a and b). With this method of tcf21 + cell ablation, there was no significant difference in coronary vascularization and revascularization when compared with control animals (Supplementary Fig. 5c-f), indicating that ablating hapln1a + cells has a more severe effect on coronary growth." 4) Details for inhibition of Serpine1 were missing. Is the concentration used equivalent to just what is needed to inhibit activity? Can inhibition of activity be measured? A complementary genetic approach would be highly desirable to exclude chemical offtargets. A: We agree with the reviewer that a genetic approach is an excellent means to exclude possible chemical off-target effects and drug non-specific effects/efficiency. As the reviewer mentioned, genetic means of manipulating adult zebrafish genes are timeconsuming and challenging/risky: 1) more than 1 year to develop a CRISPR/Cas9 serpine1 mutant generation and several generation outcrosses; and 2) a serpine1 mutant also does not guarantee that there will be no compensation effects. For these reasons, we chose a pharmacological inhibition approach to perform functional analysis of serpine1 during heart regeneration. We chose the drug Tiplaxtinin as its serpine1 inhibition activity has been previously measured and was utilized for the pharmacological inhibition of Gorlatova demonstrated the capacity for Tiplaxtinin to block serpine1 activity, and IC50 values for serpine1 inactivation by Tiplaxtinin are from 9 to 12 μM. The inhibitory effect of 20 µM Tiplaxtinin for serpine1, the concentration which we used in this MS, is about 95%. As the means to examine possible downstream effectors of serpine1, like urokinase plasminogen activator (uPA) and tissue plasminogen activator (tPA) is currently still lacking in zebrafish, we are unable to examine the effect of serpine1 inhibition. However, we performed experiments to exclude the possible toxic effects of Tiplaxtinin by examining coronary growth after removing the drug. We found that inhibited coronary growth restarted after removing the inhibitor. We have now included these data on pages 13-14, and in Supplementary Fig. 8 of the revised manuscript.
"…by treatment with its antagonist Tiplaxtinin, which blocks Serpine1 protease and has been used to inhibit serpine1 activity in different systems, including during zebrafish heart regeneration [70,[73][74][75][76][77]. Previous studies measured and determined that the inhibitory effect of 20 µM of Tiplaxtinin on serpine1 is around 95% [73,78]." "Moreover, these blocked coronary extensions can re-start after drug removal, indicating that the coronary growth restriction by serpine1 inhibition does not occur by coronary cell death ( Supplementary Fig. 8)." Minor issues: 1) The term "shear structures" needs explanation as it is not a commonly used descriptor. Is there some evidence for causation of shear stresses? A: We have updated the revised manuscript to include the phrase "linear structure" instead of "shear structure".
2) What does it mean that vessel number is increased by hapln1a cell ablations? Are these different vessels or fragments of previous vessels? A: As hapln1a + cells also wrap around existing coronary vessels and behave as perivascular cells, we speculate that hapln1a + cells play roles in vessel stabilization and the increased vessel numbers result from the fragmentation of newly formed vessels with hapln1a + cell loss. We have included this information in page 11 of the revised manuscript: "…while coronary vessel numbers in hapln1a:NTR;deltaC:EGFP animals were more than in deltaC:EGFP sibling controls (Fig. 4f). As hapln1a + cells also envelop existing coronary vessels and behave as perivascular cells, these results indicate that coronary growth is not only blocked but coronary vessels also become unstable without hapln1a + cells. This instability caused the fragmentation of existing vessels and resulted in the increased vessel numbers in the condition of hapln1a + cell loss."

Reviewer 3
In this study Sun et al identify a subset of epicardial cells that express Hapln1a which are associated with sprouting coronary vessels in the juvenile and regenerating zebrafish hearts, and use ablation experiments to show that Hapln1a is required for coronary growth in both scenarios. They further identify that serpine1, which encodes for a negative regulator of plasminogen activator, is expressed in hapln1a+ epicardial cells, is found at coronary tips, and is required for vascularisation both in coronary development and regeneration. Previous studies have identified that epicardial subtypes exist, that the epicardium supports coronary development, that hapln1a is expressed in the epicardium, that hapln1a is required for heart regeneration, and that serpine1 also promotes heart regeneration. However, little is understood about heterogeneity in the epicardium, how the epicardium supports coronary growth, and which epicardial cells are required. In this context, the study by Sun et al contributes new knowledge to the field of coronary development, and unifies and extends previous studies on the role of epicardium in vascularisation and revascularisation in development and injury. The experiments are well designed and executed, and the data is well presented. However, the depth of mechanistic analysis of the roles of hapln1a and serpine1 in vascularisation, particularly in terms of their functions and relationship, is a little unsatisfactory. What is the impact of Hapln1a and Serpine1 LOF on the ECM around the vasculature for example? A: We appreciate the reviewer's comments and per the reviewer's suggestions we have performed additional experiments to further analyze the effect of hapln1a + cells and serpine1 on the ECM component hyaluronan (HA) for further understanding of the underlying mechanism. With our new analyses, we observed HA lining coronary vessel extensions and also preceding these vessels. Further, we discovered that depletion of hapln1a + cells disrupted the HA organization and inhibiting serpine1 activity led to a disorganized HA structure. Our new experiments indicate that HA deposition by hapln1a + cells and then organization by serpine1-expressing cells is required for coronary growth during morphogenesis and regeneration. Although previous tissue engineering studies have demonstrated that HA-based materials are sufficient to stimulate angiogenesis and vascular growth and HA is required for zebrafish coronary re-vascularization ( "To elucidate the mechanism by which hapln1a + cells and serpine1 activity impact coronary growth, we assessed the distribution of the hapln1a + cell substrate, hyaluronic acid (HA). We recently reported that hapln1a + epicardial cells regulate HA deposition to facilitate cardiomyocyte proliferation during heart morphogenesis and regeneration [48]. HA is required for coronary revascularization during zebrafish heart regeneration [79][80][81]. Previous studies also demonstrated that endothelial cells can attach and grow along HA fibers in vitro, and a scaffold in tissue-engineered vascular grafts formed from HA-based biomaterials can completely generate a new vascular tube. These studies indicate that the administration of HA can enhance cell proliferation, adhesion, tubular sprout formation, and the migration of endothelial cells [79][80][81][82][83][84][85]. We speculate that HA is involved in guided coronary growth. To examine this possibility, we first assessed HA localization with coronary vessels on the ventricular surface in juvenile deltaC:EGFP animals. We detected strong association of linear HA signals with EGFP + signals (Fig. 9a). Next, we examined the HA deposition after ablating hapln1a + cells at 7 wpf. Juvenile hapln1a:mCherry-NTR;deltaC:EGFP animals and deltaC:EGFP siblings were treated with 10 mM Mtz and their hearts were collected for histological analysis. We assessed the size of HA aggregates by quantifying the area of each focus of HA and found that hapln1a + cell-depleted hearts displayed 35% lower HA intensity per HA aggregate, when compared with control siblings (Fig. 9b), indicating that the ablation of hapln1a + cells disrupted the linear structure of HA in the coronary growth area. Next, we examined HA signals in the regeneration area and found HA signals were closely associated with deltaC:EGFP + cells (Fig. 9c). We then assessed the HA deposition within the regenerating area after depleting hapln1a + cells and observed that 7 dpa injury sites of hapln1a + cell-depleted animals displayed ~47% smaller puncta of HA, compared with wild-type siblings (Fig. 9d). Next, we assessed whether serpine1 functions through regulating the HA structure during morphogenesis and regeneration, as previous reports indicated serpine1 protects the ECM to maintain a matrix necessary for endothelial cells to migrate and form capillaries [86]. We examined HA deposition in the ventricle surface after treating juvenile deltaC:EGFP fish with Tiplaxtinin for 12 hours and continuously for 7 days and found that inhibitor treated hearts displayed a 64% lower HA intensity per HA aggregate, when compared with control siblings (Fig. 9e and f), revealing a disorganized HA linear structure within the coronary growth area. Lastly, we treated adult fish with Tiplaxtinin from 3 dpa for 12 hours and continuously for 4 days and observed 53% lower HA intensity per HA aggregate at 7 dpa, compared with vehicle controls (Fig. 9g and h).
Our results indicate a mechanism for hapln1a + cell function during morphogenesis and regeneration, in which HA is deposited by hapln1a + cells and then organized by serpine1expressing cells for coronary growth during vascularization and revascularization." How/why do these ECM modifications support vessel formation? And what is the relationship between them? A: Previous tissue engineering studies have demonstrated that HA-based materials are sufficient to stimulate angiogenesis and vascular growth and HA is required for zebrafish coronary re-vascularization (Brandes et al., JCI 1991;Peattie et al., Europe PMC 2004;Turner et al., Biomaterials 2004;Lepidi et al., Eur J Vasc Endovasc Surg 2006;Pardue et al., Organogenesis 2008;Burdick and Prestwich, Adv Mater 2011;Munch et al., Development 2017;Ghose et al., BMC Cancer 2018;Luo et al., Front Bioeng Biotechnol 2022). However, there is no direct observation of organized HA structures along and preceding coronary growth in vivo. Further, no reports exist regarding the regulation mechanism of HA by hapln1a + cells and serpine1. Our results indicate that HA lines the coronary vessel extensions and also precedes these vessels. These HA cables may function as scaffolds, which could prevent loss of ECM components during tissue remodeling, act as a template for matrix regeneration, and support interactions with other cell types (Evanko et al., J Histochem Cytochem 2009;Sun and Keller, Exp Eye Res 2015;Motte et al., JBC 1999;Wang and Hascall, JBC 2004;Jokela et al., Connect Tissue Res. 2008). We have included these experiments on page 14-16, in Fig. 9 (Please see the response to the first Comment above); and on page 17, paragraph 1 of the revised manuscript: "Our work indicates that hapln1a + cells regulate HA organization during heart morphogenesis and regeneration. Although HA has previously been implicated in vessel growth stimulation and been applied for vessel tube formation during bioengineering [70,[79][80][81][82][83][84][85], no studies have reported that HA lines coronary vessels and paves the road for coronary growth during morphogenesis and endogenous regeneration. These hapln1a + cell derived HA cables may function as scaffolds, which could prevent loss of ECM components during tissue remodeling, act as a template for matrix regeneration, and support interactions with other cell types [94][95][96][97][98]." While further experiments would help to provide better insights into the specific role of the Hapln1a+ Serpine1+ epicardium in supporting coronary (re)development, the authors also do not really dive into these questions in the discussion, which is an omission. Taking steps to address the mechanism would significantly improve the impact of the manuscript. A: We appreciate these comments and have now performed additional experiments to further explore the underlying mechanisms of hapln1a + cells and serpine1 on coronary growth. We uncovered that hapln1a + cells/serpine1 control coronary growth by regulating the deposition/organization of HA, which was previously indicated to be sufficient to stimulate vessel growth by bioengineering (Brandes et al., JCI 1991;Peattie et al., Europe PMC 2004;Turner et al., Biomaterials 2004;Lepidi et al., Eur J Vasc Endovasc Surg 2006;Pardue et al., Organogenesis 2008;Burdick and Prestwich, Adv Mater 2011;Munch et al., Development 2017;Ghose et al., BMC Cancer 2018;Luo et al., Front Bioeng Biotechnol 2022). We have included these results and information on pages 14-16 and Fig. 9 of the revised manuscript. We also included further discussion of hapln1a + cells and serpine1 function in coronary sprouting/growth on page 17 of the revised manuscript. Please see the responses to the first and second Comments above.
Major comments: 1. What is the impact of Hapln1a and Serpine1 LOF on the ECM around the vasculature for example? A: We have now performed HA analyses after depleting hapln1a + cells and inhibiting serpine1 activity. We observed that the HA structure is disorganized around the vessels with hapln1a + cell or serpine1 activity disruption, when compared with controls. We included these results on pages 15-16 and Fig. 9 in the revised manuscript.
"…To examine this possibility, we first assessed HA localization with coronary vessels on the ventricular surface in juvenile deltaC:EGFP animals. We detected strong association of linear HA signals with EGFP + signals (Fig. 9a). Next, we examined the HA deposition after ablating hapln1a + cells at 7 wpf. Juvenile hapln1a:mCherry-NTR;deltaC:EGFP animals and deltaC:EGFP siblings were treated with 10 mM Mtz and their hearts were collected for histological analysis. We assessed the size of HA aggregates by quantifying the area of each focus of HA and found that hapln1a + cell-depleted hearts displayed 35% lower HA intensity per HA aggregate, when compared with control siblings (Fig. 9b), indicating that the ablation of hapln1a + cells disrupted the linear structure of HA in the coronary growth area. Next, we examined HA signals in the regeneration area and found HA signals were closely associated with deltaC:EGFP + cells (Fig. 9c). We then assessed the HA deposition within the regenerating area after depleting hapln1a + cells and observed that 7 dpa injury sites of hapln1a + cell-depleted animals displayed ~47% smaller puncta of HA, compared with wild-type siblings (Fig. 9d). Next, we assessed whether serpine1 functions through regulating the HA structure during morphogenesis and regeneration, as previous report indicated serpine1 protects the ECM to maintain a matrix necessary for endothelial cells to migrate and form capillaries [86]. We examined HA deposition in the ventricle surface after treating juvenile deltaC:EGFP fish with Tiplaxtinin for 12 hours and continuously for 7 days and found that inhibitor treated hearts displayed a 64% lower HA intensity per HA aggregate, when compared with control siblings (Fig.  9e and f), revealing a disorganized HA linear structure within the coronary growth area. Lastly, we treated adult fish with Tiplaxtinin from 3 dpa for 12 hours and continuously for 4 days and observed 53% lower HA intensity per HA aggregate at 7 dpa, compared with vehicle controls (Fig. 9g and h). . However, there are no reports on the organized HA structures along and preceding coronary growth and the regulation mechanism of HA in vivo. Our discoveries indicate that hapln1a + cells/serpine1 control coronary growth by regulating the deposition/organization of HA. We included these results on pages 15-16 in the revised manuscript: "HA has been indicated to be required for coronary revascularization during zebrafish heart regeneration [79][80][81]. Previous studies also demonstrated that endothelial cells can attach and grow along HA fibers in vitro, and a scaffold in tissue-engineered vascular grafts formed with HA-based biomaterials can completely generate a newly formed vascular tube. These studies indicate the administration of HA can enhance cell proliferation, adhesion, tubular sprout formation, and the migration of endothelial cells [79][80][81][82][83][84][85]." "Our results indicate a mechanism for hapln1a + cell function during morphogenesis and regeneration, in which HA is deposited by hapln1a + cells and then organized by serpine1expressing cells for coronary growth during vascularization and revascularization." 3. Further experiments would help to provide better insights into the specific role of the Hapln1a+ Serpine1+ epicardium in supporting coronary (re)development, the authors also do not really dive into these questions in the discussion, which is an omission. A: Our previous study demonstrated that hapln1a + cells are required for the synthesis and organization of the ECM component hyaluronic acid (HA) (Sun et al., Circulation 2022). Additionally, other reports revealed that Serpine1 protects the ECM to maintain a matrix necessary for endothelial cells to migrate and form capillaries (Ismail et al., Int J Mol Sci. 2022). To address this question, with our new analyses, we observed HA lining coronary vessel extensions and also preceding these vessels. Further, we discovered that depletion of hapln1a + cells disrupted the HA organization and inhibiting serpine1 activity led to a disorganized HA structure. Our new experiments indicate that HA deposition by hapln1a + cells and then organization by serpine1-expressing cells is required for coronary growth during morphogenesis and regeneration. These results indicate that hapln1a + cells and serpine1 activity are involved in the establishment of an organized HA structure that is required for continuous coronary growth. We have included in the results on pages 14-16, Fig. 9, and in the discussion on page 17: "Our work indicates that hapln1a + cells regulate HA organization during heart morphogenesis and regeneration. Although HA has previously been implicated in vessel growth stimulation and been applied for vessel tube formation during bioengineering [70,[79][80][81][82][83][84][85], no studies have reported that HA lines coronary vessels and paves the road for coronary growth during morphogenesis and natural regeneration. These hapln1a + cell derived HA cables may function as scaffolds, which could prevent loss of ECM components during tissue remodeling, act as a template for matrix regeneration, and support interactions with other cell types [94][95][96][97][98]." 4. Relating to the comment above, the authors do not investigate changes in ECM in either the Hapln1a or Serpine1 models, or discuss this mechanism of action of each gene. The authors could assess for example HA presence/organisation (as in their previous Circulation Hapln1a epicardium paper) around the vessels in the juvenile stage in wild type and upon ablation of Hapln1a+ cells or Tiplaxtinin treatment. A: We appreciate these comments and have performed HA analyses. Please see our response to the first Comment above.
5. In Line 339 -the authors also suggest that hapln1a regulates coronary vessel development through Serpine1 and that they will test this -of course this will not be happening at a transcriptional level, and since the authors also don't look in detail at the mechanisms of action of either protein, or the impact on the extracellular environment (showing only that they both promote coronary development), the experiments that they perform subsequently don't really address this hypothesis. A: We have examined our language in the manuscript based on this suggestion. We feel it is accurate to state that: "hapln1a + cells precede coronary growth extensions and express serpine1, and these hapln1a + cells and the serpine1 gene are required for an ECM environment through HA organization." 6. It appears that not all vessel tips are associated with Hapln1a (or even Serpine1) for example the 6 hour timepoint shown in Figure 2 -there appear to be sprouts forming without any Hapln1a. Related to this, some revascularization still occurs in the hapln1a KO. Does this mean Hapln1a is only required by a subset of branching vessels? To investigate this, the authors should quantify the number of tips with/without Hapln1a and Serpine1, and comment on this possible heterogeneity.

A:
The reviewer is insightful to point out this possibility. To be accurate, we have changed the word "coronary tip" to "coronary sprout". We agree with the reviewer that coronary sprouting can form without any hapln1a + cells. Our analyses revealed that there is a small percentage (around 18%) of coronary extensions that occur without any halpn1a + cells during heart morphogenesis. Although we also observed the withdrawal of the extended coronary vessels without hapln1a + cells, we cannot exclude other mechanisms that provide guidance cues that may affect coronary growth during vascularization and revascularization. There are other possibilities such as: macrophages affect vessel growth, as we also observed macrophages attached to vessels in some areas. We have updated the revised manuscript to include discussion of coronary vessel tips without hapln1a + cells and possible other guidance cues they depend on for growth on page 18, paragraph 1: "In this study, we discovered that hapln1a + cells and serpine1 activity are responsible for major coronary growth in zebrafish: hapln1a + cells deposit HA and serpine1 regulates the HA organization to form a linear structure, which is required for coronary extension and continuous growth. Further, our analyses also revealed that there is a small percentage (around 18%) of coronary growth extensions that occur without any halpn1a + cells during heart morphogenesis. Although we also observed the withdrawal of the extended coronary vessels without hapln1a + cells, we cannot exclude other mechanisms that provide guidance cues that may affect coronary growth during vascularization and revascularization." 7. The authors state that the location of hapln1a cells indicates they function as perivascular cells (Line 221) -are there perivascular markers in the hapln1a+ clusters? If so, which cluster? A: We examined the scRNA-seq data of tcf21 + cells, and found that the perivascular marker, pdgfrb, is expressed in the tcf21 + cell clusters. This information is now included on page 7, paragraph 2, and Supplementary Fig. 4 of the revised manuscript: "We then analyzed the scRNA-seq data for perivascular cell markers and observed pdgfrβ expression in clusters 1 and 6 [50] ( Supplementary Fig. 4)." 8. The authors also identify 3 different kinds of hapln1a+ cell -2 of which behave more like perivascular cells, surrounding the vessels, and 1 of which (the most prevalent kind, which 'precedes' the growing sprout and is Serpine1 + in juveniles) exhibits behaviour unexpected of a perivascular cell. Do the authors think these cell types are from different Hapln1+ clusters, or are they all the same subtype? In the single cell seq data presented in figure S2 (Hapln1+ sorted), serpine1 is found in a relatively small proportion of those cells -do these form a distinct cluster separate from other Hapln1a+ cells? Does clustering of the Hapln1a+ sc-RNA-seq data give more insights into this population? A: We appreciate these thoughtful comments and indeed performed further analyses with our scRNA-seq data. Unfortunately, we couldn't separate serpine1-expressing cells from other hapln1a + cells. We speculate that this occurred as hapln1a-expressing cells are a subtype of epicardial cells and the hapln1a + cells are likely a relatively homogenous cell population, with different transitional states leading to difficulties with cluster separation. We are in the process of generating BAC and knock-in serpine1 fluorescence reporters and will isolate these cells for deep sequencing to perform comparisons between hapln1a + /serpine1and hapln1a + /serpine1 + cells during morphogenesis and regeneration in future studies. Fig S2, serpine1 is expressed at very low levels in Hapln1a+ cells of juvenile hearts compared to regenerating hearts -the authors should comment on this difference. A: It was previously reported that serpine1 is also expressed in endocardial cells in the inner injured area during heart regeneration (Munch et al., Development 2017). As coronary growth occurs in the ventricular wall and revascularization occurs in the lateral injury area at early stages of regeneration, we speculate that the inner endocardiumderived serpine1 is not involved in coronary growth guidance. We have included this information on page 14, paragraph 1: "Similar to a previous report that the endocardium expresses serpine1 after heart injury [70], we also detected serpine1 signals in the inner injury area. However, we speculate that these serpine1-expressing cells are not directly correlated with coronary extension as coronary vessels mainly extended from the lateral area of the ventricular wall to the middle of the injury site. Together, our results indicate that hapln1a + cell-derived serpine1 controls coronary growth during heart morphogenesis and regeneration." and page 18, paragraph 2 of the revised manuscript:

In
"Further, we noticed that the percentage of serpine1 expression in hapln1a + cells is different during heart morphogenesis and regeneration. We speculate that this difference arose because coronary vascularization occurs for around two months (from 5 wpf -12 wpf) and builds the coronary vasculature from very few coronary cells, while coronary revascularization occurs within 2-3 weeks and generates new vessels from existing vessels. Another possibility is that the endocardium functions through serpine1 to regulate myocardial regeneration while such endocardium-derived serpine1 doesn't exist during heart morphogenesis. We speculate that these differences led us to detect low expression levels of serpine1 in morphogenesis hearts in comparison with serpine1 expression in the injury site." Related to this, the phenotype in the Serpine1-inhibited hearts at 7wpf (Fig. 7C) is very profound given the small number of cells that express Serpine1 (Fig S2). In the explanted hearts (Fig 7A), there even appears to be a reduction in vessels -is this the impact not of defective growth, but rather death/regression of the existing vascular network? The authors should comment on these issues. A: Our previous published work indicated that coronary growth occurs in juvenile fish hearts (Sun et al., Dev Bio. 2022), which indicates that there are newly formed coronary vessels that exist in the ventricular wall. These new vessels (immature vessels) need to be stabilized by ECM and perivascular cells. We speculate that although serpine1 is expressed in a small number of hapln1a + cells, these serpine1 + cells are important for the formation of correct HA structures that stabilize newly formed vessels and for continuous growth of extending coronary sprouts. If serpine1 activity is inhibited, newly formed coronary vessels will become unstable and growing coronary vessels will withdraw. Further, we performed ex vivo culture of juvenile hearts with a serpine1 inhibitor and observed blocked coronary growth, and then detected coronary growth after removal of this drug. This experiment indicates that the regression is not from death of existing vessels. This information has been included in our revised manuscript on page 14 and Supplementary Fig. 8: "Moreover, these blocked coronary extensions can re-start after drug removal, indicating that the coronary growth restriction by serpine1 inhibition does not occur by coronary cell death ( Supplementary Fig. 8)." Minor comments: The authors state there is significant enrichment of Hapln1a in the angiogenic epicardial subclusters of 1 and 3 (Line 208). In Figure 1 it is clear there is also significant expression in clusters 4 and 5, and so this statement is a little misleading (arguably, cluster 4 actually has more hapln1a+ expressing cells than cluster 3). A: We agree with the reviewer and have now clarified hapln1a expression in the revised manuscript. The Violin plot analysis of hapln1a indicated high expression of hapln1a in cluster 1, slight expression in cluster 3, 4, 5, and little expression in clusters 2 and 6. We have revised the description on page 7, paragraph 1: "As cluster 1 is the most heavily represented of these 3 cell states, we first focused on cluster 1 and observed high expression of hapln1a in this cluster [23], with slight expression in clusters 3, 4 and 5 ( Fig. 1d and Supplementary Fig. 3)." Did the authors confirm that the Tiplaxtinin treatment is blocking serpine1 activity/function? For example by assaying conversion of plasminogen to plasmin. A: We chose the drug Tiplaxtinin as its serpine1 inhibition activity has been previously measured and was utilized for the pharmacological inhibition of serpine1 activity and published by different research groups (Gorlatova et al., JBC 2007;Daniel et al., PLos One 2015;Kang et al., Oncotarget 2016;Munch et al., Development 2017;Lin et al., Cell Rep. 2020;Yamagami et al., BBRC 2020). Natalia V. Gorlatova demonstrated the capacity for Tiplaxtinin to block serpine1 activity, and IC50 values for serpine1 inactivation by Tiplaxtinin are from 9 to 12 μM. The inhibitory effect of 20 µM Tiplaxtinin for serpine1, the concentration which we used in this MS, is about 95%. As the means to examine possible downstream effectors of serpine1, like urokinase plasminogen activator (uPA) and tissue plasminogen activator (tPA) is currently still lacking in zebrafish, we are unable to examine the effect of serpine1 inhibition. However, we performed experiments to exclude possible toxic effects of Tiplaxtinin by examining coronary growth after removing the drug. We found that inhibited coronary growth restarted after removing the inhibitor. We have now included these data on pages 13-14, and in Supplementary Fig. 8 of the revised manuscript.
"…by treatment with its antagonist Tiplaxtinin, which blocks Serpine1 protease and has been used to inhibit serpine1 activity in different systems, including during zebrafish heart regeneration [70,[73][74][75][76][77]. Previous studies measured and determined that the inhibitory effect of 20 µM of Tiplaxtinin on serpine1 is around 95% [73,78]." "Moreover, these blocked coronary extensions can re-start after drug removal, indicating that the coronary growth restriction by serpine1 inhibition does not occur by coronary cell death ( Supplementary Fig. 8)." Figure 1 -there is a lack of consistency in the colour coding used in the heatmaps -sometimes blue represents the highest value, sometimes the lowest -this could be consistent within the figure to simplify. A: We have changed the color to be consistent as recommended. Figure 4 -The authors show that the vessel density and junction number (graphs D and E) are reduced, which is in line with the observation that vascular growth is less in the hapln1a ablation model. However, vessel number per area is increased (F), which seems contradictory to the previous quantifications. Can the authors explain the difference between these two quantifications? A: As hapln1a + cells also wrap around existing coronary vessels and behave as perivascular cells, we speculate that hapln1a + cells play roles in vessel stabilization and the increased vessel numbers result from the fragmentation of newly formed vessels with hapln1a + cell loss. We have included these information in page 11 of the revised manuscript: "…while coronary vessel numbers in hapln1a:NTR;deltaC:EGFP animals were more than in deltaC:EGFP sibling controls (Fig. 4f). As hapln1a + cells also envelop existing coronary vessels and behave as perivascular cells, these results indicate that coronary growth is not only blocked but coronary vessels also become unstable without hapln1a + cells. This instability caused the fragmentation of existing vessels and resulted in the increased vessel numbers in the condition of hapln1a + cell loss." Figure 4 and 5 -the graphs are distorted along the x axis -aspect ratio should be maintained. A. We have corrected these graphs as recommended. Figure 6 -The serpine1 expression shown in the images doesn't seem to colocalise with Hapln1a at terminal branches (rather it is localised more centrally within the Hapln1a network) -is Serpine1 expressed broadly within the Hapln1a network as well as around tip cells? A: We observed that hapln1a + cells (which deposit HA) and HA deposition can precede coronary growth sprouts over a long distance. The serpine1 is expressed in hapln1a + cells close to the coronary sprouts and involved in organizing the HA structure, leading to the appearance of serpine1 localizing more centrally within hapln1a + cell linear structures.