Comparative single-cell profiling reveals distinct cardiac resident macrophages essential for zebrafish heart regeneration

Article activity feed

1. 

   Version published to 10.7554/elife.84679 on eLife

   Jul 27, 2023
2. 

   eLife

   Jun 9, 2023

   Author Response

   Reviewer #1 (Public Review):

   In the manuscript, titled "Comparative single-cell profiling reveals distinct cardiac resident macrophages essential for zebrafish heart regeneration," Wei et al. perform bulk and single-cell RNA-sequencing on uninjured and injured zebrafish hearts with or without prior macrophage depletion by clodronate. For the single-cell RNA sequencing, the authors sort macrophages and neutrophils prior to sequencing by using fluorescent reporters for each of the two lineages. The authors characterize the differential gene expression between injured and uninjured hearts with and without prior macrophage depletion. The single-cell analyses allow the characterization of nine discrete subpopulations of macrophages and two distinct neutrophil types. The manuscript is largely descriptive with lots of …

   Author Response

   Reviewer #1 (Public Review):

   In the manuscript, titled "Comparative single-cell profiling reveals distinct cardiac resident macrophages essential for zebrafish heart regeneration," Wei et al. perform bulk and single-cell RNA-sequencing on uninjured and injured zebrafish hearts with or without prior macrophage depletion by clodronate. For the single-cell RNA sequencing, the authors sort macrophages and neutrophils prior to sequencing by using fluorescent reporters for each of the two lineages. The authors characterize the differential gene expression between injured and uninjured hearts with and without prior macrophage depletion. The single-cell analyses allow the characterization of nine discrete subpopulations of macrophages and two distinct neutrophil types. The manuscript is largely descriptive with lots of discussion of specific differentially expressed genes. The authors conclude that tissue-resident macrophages are important for heart regeneration through the remodeling of the microenvironment and by promoting revascularization. Circulating monocyte-derived macrophages cannot adequately replace the resident macrophages even after recovery from clodronate depletion.

   The manuscript presents a very large catalog of useful gene expression data and further characterizes the diversity of macrophages and neutrophils in the heart following injury. Although the conclusions that resident macrophages are important for regeneration and that circulating macrophages cannot adequately substitute for them are not particularly novel, this manuscript provides additional support for those ideas and extends that work by providing a wealth of gene expression data from the different macrophage sub-populations in the zebrafish and how they respond to and promote regeneration. The authors also present a nice analysis supporting the interactions of macrophages with neutrophils via comparing receptors and ligands (from gene expression data) on the two populations - this should be a useful resource.

   We appreciate how reviewer #1 recognizes the work we have put into sample preparation, data collection, and all the bioinformatic analyses to delineate and characterize the inflammatory cells during zebrafish heart regeneration.

   Reviewer #2 (Public Review):

   Wei et al. analysed the composition of immune cells, mostly macrophages, and neutrophils, in the context of zebrafish cardiac injury while utilizing clodronate liposomes (CL) to inhibit regeneration via alteration of the immune response. This work is a direct continuation of Shih-Lei et al. which compared the regenerative outcomes of zebrafish vs the non-cardiac regenerative medaka. In that work, the authors used CL to pre-deplete macrophages and showed significant effects on neutrophil clearance, revascularization, and cardiomyocyte proliferation. In this work, the authors used the same pre-depletion method to study the dynamics, composition, and transcriptomic state of macrophages and neutrophils, to overall assess the effect on cardiac regeneration. Using bulk RNA-seq at CL vs PBS treated hearts 7 and 21 days post cryo injury (dpci) a delayed\altered immune response was evident. Single-cell analysis at 1,3 and 7 dpci showed a wide range of immune populations in which most diverse are the macrophage populations. Pre-depletion using CL, altered the composition of immune cells resulting in the complete removal of a single resident macrophage population (M2) or dramatically reducing the overall numbers of other resident populations, while other populations were retained. Looking at the injury time course and distribution of macrophage populations, the authors identified several macrophage populations and neutrophil population 1 as pro-regenerative as their presence compared to CL-treated hearts correlates with regeneration. CL-treated hearts also show a marked sustained neutrophil retention suggesting that interaction with depleted macrophage populations is required for neutrophil clearance. As the marked reduction in populations 2 and 3 occurs after CL treatment, the authors tested whether early CL treatment (8 days or 1 month prior to injury) could reduce the non-recoverable populations and affect regenerative outcomes and indeed they observed a reduction in key genes characterizing M2 and M3 which caused marked reduction in revascularization, CM proliferation, neutrophil retention, and overall higher scaring of the heart.

   The findings of this paper could be broadly separated into the characterization of myeloid cells after injury and in non-regenerating animals and assessing the effects of early pre-depletion of macrophages on various cardiac functions involved in regeneration. Both parts draw conclusions that are supported by the facts however several questions remain to be clarified.

   We thank the reviewer for recognizing that the conclusions we drew were supported by the data we presented and further replied to the specific suggestions below.

   1. In figures 2 and 3 the main claim is that the main resident macrophage populations, M2 and M3 are depleted and are largely unable to replenish after injury, similar to resident macrophages in mice 1. However, as the identification of this population is made solely using scRNA-seq, an alternative explanation would be that these cell populations do replenish but are sufficiently changed due to CL treatment (directly or indirectly) and thus would be a part of another cluster. To address this, we suggest:

   A. Run trajectory analysis to ascertain whether the different cell clusters are due to differentiating states of the cells

   B. Create a reporter line for M2 and M3 macrophages and assess whether they are indeed depleted or changing.

   We followed the reviewer’s suggestion and performed trajectory analyses (Figure 6). The results suggest that Mac 2 and Mac 3 form unique trajectory, which was not shifted by -1d_CL treatment but only diminished in number. Conditionally-enriched gene ontology analysis (Figure 4) also suggests that Mac 2 and 3 do not change property under -1d_CL condition (unlike monocyte-derived Mac 1 and some other clusters). When we examine homx1a expression (Mac2) and timp4.3 expression (Mac3) in -8d_CL treated hearts, we again observed diminished cell numbers (Figure 8C and Figure 7-figure supplement 1D). These results support the resident macrophages Mac 2 and Mac 3 are more likely to be non-recoverable than changing their property so much thus grouped into other subsets.

   We also agree with the reviewer that the specific reporter and CreER driver lines for the lineage tracing experiment will provide the most concrete answer to this question. We have now generated an endogenous Tg(mpeg1-2A-CreERT2) line in the lab (collaborative work with McGrail lab) and reporter lines using Mac2/3 enriched genes. Unfortunately, this work will take much longer time and might not fit into the scope of the current study.

   1. One of the major findings of this paper is that some macrophage populations can persist throughout injury and promote the regenerative response. Considering that macrophages have a half-life of less than a day in tissue 2 (although could be different in zebrafish and in this population), we estimate that the resident populations should be proliferative. As there is only a single proliferating macrophage population (M5) we speculate that it is a combination of several populations which are clustered together due to the high expression of cell cycle genes. To verify whether the resident populations are proliferating we suggest:

   A. Perform cell-cycle scoring and regression (found in Seurat package) and assess whether after regressing out cell cycle genes there are contributions of M5 to other clusters.

   B. Perform EDU labelling experiments with cell cycle identifiers (staining for hbaa1, Timp4.3) and assess their proliferative dynamics.

   We followed the reviewer’s suggestion and performed cell-cycle scoring and regression (Figure 2-figure supplement 4). Cell cycle scoring suggests there are cells in both Mac 2 and 3 in the G2/M phase and presumably proliferative. Cell-cycle regression results suggest that most macrophage subsets, including Mac 5, still stand as unique clusters after regression (Figure 2-figure supplement 4). These results suggest that Mac 5 might not be constitute of proliferating cells from other clusters.

   On the other hand, we also tried to double-stain the proliferating resident macrophages by EdU and ISH of hbaa1 and timp4.3. Unfortunately, these methods were not comparable in our hands, and we failed to confirm their proliferative dynamics. We did show proliferating macrophages residing in the untouched hearts and will further check their identity once we have the cluster-specific reporter lines ready.

   Last but not least, using the Tg(mpeg1-2A-CreERT2) line to label embryonic macrophages under the Tg(ubi:loxP-EGFP-loxP-mCherry)cz1701 background before 7 dpf, we observed mCherry+ macrophages in juvenile fish at 50 dpf, suggesting some embryonically derived macrophages can last more than a week in the system presumably by self-renewing. As replied previously, these results might not be included in this study.

   1. In connection to the previous point if indeed these resident macrophage populations are proliferative, even a smaller portion of remaining cells should be sufficient to partly replenish given sufficient time after CL 1. However as seen in Fig. 3B, the M2 population has a similar proportion of cells on days 1 and 3 after CL treatment and by day 7 it declines in numbers. Given that CL should not be present anymore, we expect this population to increase in numbers over time.

   We thank the reviewer for pointing out that Figure 3B might be misleading as the proportion of the macrophage subsets was calculated. The persistence of Mac 2 proportion at 1 and 3 dpci might be due to the overall depletion of both resident and recruited macrophages after CL treatment. 2 days after CL treatment still have profound effects on total macrophage numbers (Figure 7-figure supplement 1A and Lai et al., 2017) and the overall macrophage numbers only recovered to the same level as those in untouched or PBS-treated injured hearts by 7 days (Figure 7-figure supplement 1A and Lai et al., 2017). We have also confirmed that Mac 2 diminished in CL-treated hearts by both qPCR and ISH/IHC of homx1a in Figure 7-figure supplement 1C and Figure 8B.

   1. In Figure 6 the authors show a reduction in mpeg+ population however a persistent, large population ({plus minus}70% of the original mpeg+) is retained. The authors suggest that this population is comprised of other, non-macrophage, cell types however as this method is the very core of the paper and the persistence of macrophages could alter our understanding of the results, it must be verified.

   Dick, S. A. et al. Self-renewing resident cardiac macrophages limit adverse remodeling following myocardial infarction. Nature Immunology 20, 29-39, doi:10.1038/s41590-018-0272-2 (2019).

   Leuschner, F. et al. Rapid monocyte kinetics in acute myocardial infarction are sustained by extramedullary monocytopoiesis. J Exp Med 209, 123-137, doi:10.1084/jem.20111009 (2012).

   We acknowledge that mpeg1 might not be the perfect marker for pan-macrophage labeling shown by the work published by Ferrero et al., J Leukoc Biol. 2020, when our profiling work had been undergone. Fortunately, scRNAseq profiling is an unbiased method to reveal gene expression/cell identity, and our results indeed identified non-macrophage/non-neutrophil populations out of the clustering and found mpeg1+ B-cells consistent with the literature. Thus, the mixed input from the mpeg1 reporter does not affect the property of Mac 2 and 3 being both mpeg1-positive macrophages, which diminished after both -1d_CL and -8d_CL treatment. Following the reviewer’s suggestion, we further verified this point by both qPCR of hbaa1 and timp4.3 and ISH/IHC of homx1a and timp4.3 in the CL-treated hearts in Figure 7-figure supplement 1C and D and Figure 8B and C.

   Reviewer #3 (Public Review):

   Macrophages play an important role during heart regeneration. This has been shown in the mouse and zebrafish for example by treating the animals with clodronate liposomes to eliminate phagocytic cells.

   The manuscript follows up on a previous observation by the authors performing these experiments in the zebrafish (Lai et al eLife 2017). When comparing regenerative vs non-regenerative teleosts zebrafish resp Medaka they found that macrophages and neutrophils were the cell types more differentially responding in these two species to a cardiac injury.

   Here the authors analyze in extenso neutrophil and macrophage populations using single-cell RNA-seq at different stages of regeneration. They perform FAC sorting of the two populations using specific reporter lines. They also assess the change in these populations upon clodronate treatment. They find that clodronate treatment affects the gene expression profiles of different subsets of macrophages and neutrophils as well as their abundance.

   They also show that chlodronate treatment performed several days before cryoinjury depleted macrophages from the heart but after injury overall macrophage number recovers. However, heart regeneration does not. Cardiomyocyte is the only parameter that is not affected, but vasculogenesis and scar resolution is impaired.

   The authors conclude that (1) there are different subsets of macrophages and neutrophils, (2) that they interact with each other during regeneration through specific ligand and receptor pairs, and (3) that a cardiac resident population rather than a circulating macrophage population is important for heart regeneration.

   The transcriptomic characterization of the two immune cell populations is very exhaustive and rigorous. No functional validation of subpopulation marker genes was performed, but the data as it stands will already be of great value to the community. The figure quality is outstanding.

   We thank the reviewer for recognizing the value of our study and the quality of the data presented. We further examined the subpopulation markers and their functional relevance in the revised manuscript, as suggested.

   Read the original source
3. 

   eLife

   Jun 3, 2023

   eLife assessment

   The authors analyze changes in the gene expression of different immune cells during heart regeneration using single-cell RNA-sequencing and assess changes upon drug treatment that depletes macrophages. They find that drug treatment affects the gene expression profiles of different and abundance of immune cells. The work provides a wealth of gene expression data and a nice analysis supporting immune cell interactions during heart regeneration, so will be a useful resource.

   Read the original source
4. 

   eLife

   Jun 3, 2023

   Reviewer #1 (Public Review):

   In the manuscript, titled "Comparative single-cell profiling reveals distinct cardiac resident macrophages essential for zebrafish heart regeneration," Wei et al. perform bulk and single-cell RNA-sequencing on uninjured and injured zebrafish hearts with or without prior macrophage depletion by clodronate. For the single-cell RNA sequencing, the authors sort macrophages and neutrophils prior to sequencing by using fluorescent reporters for each of the two lineages. The authors characterize the differential gene expression between injured and uninjured hearts with and without prior macrophage depletion. The single-cell analyses allow the characterization of nine discrete subpopulations of macrophages and two distinct neutrophil types. The manuscript is largely descriptive with lots of discussion of specific …

   Reviewer #1 (Public Review):

   In the manuscript, titled "Comparative single-cell profiling reveals distinct cardiac resident macrophages essential for zebrafish heart regeneration," Wei et al. perform bulk and single-cell RNA-sequencing on uninjured and injured zebrafish hearts with or without prior macrophage depletion by clodronate. For the single-cell RNA sequencing, the authors sort macrophages and neutrophils prior to sequencing by using fluorescent reporters for each of the two lineages. The authors characterize the differential gene expression between injured and uninjured hearts with and without prior macrophage depletion. The single-cell analyses allow the characterization of nine discrete subpopulations of macrophages and two distinct neutrophil types. The manuscript is largely descriptive with lots of discussion of specific differentially expressed genes. The authors conclude that tissue-resident macrophages are important for heart regeneration through the remodeling of the microenvironment and by promoting revascularization. Circulating monocyte-derived macrophages cannot adequately replace the resident macrophages even after recovery from clodronate depletion.

   The manuscript presents a very large catalog of useful gene expression data and further characterizes the diversity of macrophages and neutrophils in the heart following injury. Although the conclusions that resident macrophages are important for regeneration and that circulating macrophages cannot adequately substitute for them are not particularly novel, this manuscript provides additional support for those ideas and extends that work by providing a wealth of gene expression data from the different macrophage sub-populations in the zebrafish and how they respond to and promote regeneration. The authors also present a nice analysis supporting the interactions of macrophages with neutrophils via comparing receptors and ligands (from gene expression data) on the two populations - this should be a useful resource.

   Read the original source
5. 

   eLife

   Jun 3, 2023

   Reviewer #2 (Public Review):

   Wei et al. analysed the composition of immune cells, mostly macrophages, and neutrophils, in the context of zebrafish cardiac injury while utilizing clodronate liposomes (CL) to inhibit regeneration via alteration of the immune response. This work is a direct continuation of Shih-Lei et al. which compared the regenerative outcomes of zebrafish vs the non-cardiac regenerative medaka. In that work, the authors used CL to pre-deplete macrophages and showed significant effects on neutrophil clearance, revascularization, and cardiomyocyte proliferation. In this work, the authors used the same pre-depletion method to study the dynamics, composition, and transcriptomic state of macrophages and neutrophils, to overall assess the effect on cardiac regeneration. Using bulk RNA-seq at CL vs PBS treated hearts 7 and 21 …

   Reviewer #2 (Public Review):

   Wei et al. analysed the composition of immune cells, mostly macrophages, and neutrophils, in the context of zebrafish cardiac injury while utilizing clodronate liposomes (CL) to inhibit regeneration via alteration of the immune response. This work is a direct continuation of Shih-Lei et al. which compared the regenerative outcomes of zebrafish vs the non-cardiac regenerative medaka. In that work, the authors used CL to pre-deplete macrophages and showed significant effects on neutrophil clearance, revascularization, and cardiomyocyte proliferation. In this work, the authors used the same pre-depletion method to study the dynamics, composition, and transcriptomic state of macrophages and neutrophils, to overall assess the effect on cardiac regeneration. Using bulk RNA-seq at CL vs PBS treated hearts 7 and 21 days post cryo injury (dpci) a delayed\altered immune response was evident. Single-cell analysis at 1,3 and 7 dpci showed a wide range of immune populations in which most diverse are the macrophage populations. Pre-depletion using CL, altered the composition of immune cells resulting in the complete removal of a single resident macrophage population (M2) or dramatically reducing the overall numbers of other resident populations, while other populations were retained. Looking at the injury time course and distribution of macrophage populations, the authors identified several macrophage populations and neutrophil population 1 as pro-regenerative as their presence compared to CL-treated hearts correlates with regeneration. CL-treated hearts also show a marked sustained neutrophil retention suggesting that interaction with depleted macrophage populations is required for neutrophil clearance. As the marked reduction in populations 2 and 3 occurs after CL treatment, the authors tested whether early CL treatment (8 days or 1 month prior to injury) could reduce the non-recoverable populations and affect regenerative outcomes and indeed they observed a reduction in key genes characterizing M2 and M3 which caused marked reduction in revascularization, CM proliferation, neutrophil retention, and overall higher scaring of the heart.

   The findings of this paper could be broadly separated into the characterization of myeloid cells after injury and in non-regenerating animals and assessing the effects of early pre-depletion of macrophages on various cardiac functions involved in regeneration. Both parts draw conclusions that are supported by the facts however several questions remain to be clarified.

   1. In figures 2 and 3 the main claim is that the main resident macrophage populations, M2 and M3 are depleted and are largely unable to replenish after injury, similar to resident macrophages in mice 1. However, as the identification of this population is made solely using scRNA-seq, an alternative explanation would be that these cell populations do replenish but are sufficiently changed due to CL treatment (directly or indirectly) and thus would be a part of another cluster. To address this, we suggest:
   A. Run trajectory analysis to ascertain whether the different cell clusters are due to differentiating states of the cells
   B. Create a reporter line for M2 and M3 macrophages and assess whether they are indeed depleted or changing.

   2. One of the major findings of this paper is that some macrophage populations can persist throughout injury and promote the regenerative response. Considering that macrophages have a half-life of less than a day in tissue 2 (although could be different in zebrafish and in this population), we estimate that the resident populations should be proliferative. As there is only a single proliferating macrophage population (M5) we speculate that it is a combination of several populations which are clustered together due to the high expression of cell cycle genes. To verify whether the resident populations are proliferating we suggest:
   A. Perform cell-cycle scoring and regression (found in Seurat package) and assess whether after regressing out cell cycle genes there are contributions of M5 to other clusters.
   B. Perform EDU labelling experiments with cell cycle identifiers (staining for hbaa1, Timp4.3) and assess their proliferative dynamics.

   3. In connection to the previous point if indeed these resident macrophage populations are proliferative, even a smaller portion of remaining cells should be sufficient to partly replenish given sufficient time after CL 1. However as seen in Fig. 3B, the M2 population has a similar proportion of cells on days 1 and 3 after CL treatment and by day 7 it declines in numbers. Given that CL should not be present anymore, we expect this population to increase in numbers over time.

   4. In Figure 6 the authors show a reduction in mpeg+ population however a persistent, large population ({plus minus}70% of the original mpeg+) is retained. The authors suggest that this population is comprised of other, non-macrophage, cell types however as this method is the very core of the paper and the persistence of macrophages could alter our understanding of the results, it must be verified.

   Dick, S. A. et al. Self-renewing resident cardiac macrophages limit adverse remodeling following myocardial infarction. Nature Immunology 20, 29-39, doi:10.1038/s41590-018-0272-2 (2019).
   2 Leuschner, F. et al. Rapid monocyte kinetics in acute myocardial infarction are sustained by extramedullary monocytopoiesis. J Exp Med 209, 123-137, doi:10.1084/jem.20111009 (2012).

   Read the original source
6. 

   eLife

   Jun 3, 2023

   Reviewer #3 (Public Review):

   Macrophages play an important role during heart regeneration. This has been shown in the mouse and zebrafish for example by treating the animals with clodronate liposomes to eliminate phagocytic cells.
   The manuscript follows up on a previous observation by the authors performing these experiments in the zebrafish (Lai et al eLife 2017). When comparing regenerative vs non-regenerative teleosts zebrafish resp Medaka they found that macrophages and neutrophils were the cell types more differentially responding in these two species to a cardiac injury.

   Here the authors anaylse in extenso neutrophil and macrophage populations using single-cell RNA-seq at different stages of regeneration. They perform FAC sorting of the two populations using specific reporter lines. They also assess the change in these populations …

   Reviewer #3 (Public Review):

   Macrophages play an important role during heart regeneration. This has been shown in the mouse and zebrafish for example by treating the animals with clodronate liposomes to eliminate phagocytic cells.
   The manuscript follows up on a previous observation by the authors performing these experiments in the zebrafish (Lai et al eLife 2017). When comparing regenerative vs non-regenerative teleosts zebrafish resp Medaka they found that macrophages and neutrophils were the cell types more differentially responding in these two species to a cardiac injury.

   Here the authors anaylse in extenso neutrophil and macrophage populations using single-cell RNA-seq at different stages of regeneration. They perform FAC sorting of the two populations using specific reporter lines. They also assess the change in these populations upon clodronate treatment. They find that clodronate treatment affects the gene expression profiles of different subsets of macrophages and neutrophils as well as their abundance.

   They also show that chlodronate treatment performed several days before cryoinjury depleted macrophages from the heart but after injury overall macrophage number recovers. However, heart regeneration does not. Cardiomyocyte is the only parameter that is not affected, but vasculogenesis and scar resolution is impaired.

   The authors conclude that (1) there are different subsets of macrophages and neutrophils, (2) that they interact with each other during regeneration through specific ligand and receptor pairs, and (3) that a cardiac resident population rather than a circulating macrophage population is important for heart regeneration.

   The transcriptomic characterization of the two immune cell populations is very exhaustive and rigorous. No functional validation of subpopulation marker genes was performed, but the data as it stands will already be of great value to the community. The figure quality is outstanding.

   Read the original source
7. 

   Version published to 10.1101/2022.11.22.517520v1 on bioRxiv

   Nov 22, 2022