Pigment cell progenitor heterogeneity and reiteration of developmental signaling underlie melanocyte regeneration in zebrafish

  1. William Tyler Frantz
  2. Sharanya Iyengar
  3. James Neiswender
  4. Alyssa Cousineau
  5. René Maehr
  6. Craig J Ceol

  • Curated by eLife

    eLife logo

    eLife assessment

    In this work Frantz et al. assess transcriptomic states of melanocyte stem cells that are recruited to differentiate during the process of melanocyte regeneration in zebrafish and they analyze roles for Kit signaling in this process. The analyses are nicely done, and the paper requires only relatively minor modifications and clarifications. The study will provide new insights into melanocyte stem cell biology that should be of interest to those studying pigmentation, regeneration, and melanoma biology using zebrafish and other systems.

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Tissue-resident stem and progenitor cells are present in many adult organs, where they are important for organ homeostasis and repair in response to injury. However, the signals that activate these cells and the mechanisms governing how these cells renew or differentiate are highly context-dependent and incompletely understood, particularly in non-hematopoietic tissues. In the skin, melanocyte stem and progenitor cells are responsible for replenishing mature pigmented melanocytes. In mammals, these cells reside in the hair follicle bulge and bulb niches where they are activated during homeostatic hair follicle turnover and following melanocyte destruction, as occurs in vitiligo and other skin hypopigmentation disorders. Recently, we identified melanocyte progenitors in adult zebrafish skin. To elucidate mechanisms governing melanocyte progenitor renewal and differentiation we analyzed individual transcriptomes from thousands of melanocyte lineage cells during the regeneration process. We identified transcriptional signatures for progenitors, deciphered transcriptional changes and intermediate cell states during regeneration, and analyzed cell–cell signaling changes to discover mechanisms governing melanocyte regeneration. We identified KIT signaling via the RAS/MAPK pathway as a regulator of melanocyte progenitor direct differentiation and asymmetric division. Our findings show how activation of different subpopulations of mitfa -positive cells underlies cellular transitions required to properly reconstitute the melanocyte pigmentary system following injury.

Article activity feed

  1. Version published to 10.7554/elife.78942 on eLife
  2. eLife

    Author Response

    Reviewer #1 (Public Review):

    In this study, the authors aim to identify the cell state dynamics and molecular mechanisms underlying melanocyte regeneration in zebrafish. By analyzing thousands of single-cell transcriptomes over regeneration in both wild-type and Kit mutant animals, they provide thorough and convincing evidence of (1) two paths to melanocyte regeneration and (2) that Kit signaling, via the RAS/MAPK pathway, is a key regulator of this process. Finally, the authors suggest that another proliferative subpopulation cells, expressing markers of a separate pigment cell type, constitute an additional population of progenitors with the ability to contribute to melanocytes. The data supporting this claim are not as convincing, and the authors failed to show that these cells did indeed differentiate into melanocytes. Despite the challenges of describing this third cell state, this study offers compelling new findings on the mechanisms of melanocyte regeneration and provides paths forward to understanding why some animals lack this capacity.

    The majority of the main conclusions are well supported by the data, but one claim, in particular, should be revisited by the authors.

    (1) Provided evidence that the aox5(hi)mitfa(lo) population of cells contributes to melanocyte regeneration is inconclusive and somewhat circumstantial. First, the transcriptional profiles of these cells are much more consistent with the xanthophore lineage. Indeed, xanthophores have been shown to express mitfa (in embryos in Parichy, et al. 2003 (PMID: 10862741), and in post-embryonic cells in Saunders, et al. 2019). Second, while the authors address this possibility in Supplemental figure 7 by showing that interstripe xanthophores fail to divide following melanocyte ablation, they fail to account for the stripe-resident xanthophores/xanthoblasts. The presence and dynamics of aox5+ stripe-resident xanthophores/xanthoblasts are detailed in McMenamin, et al., 2014 (PMID: 25170046) and Eom, et al., 2015 (PMID: 26701906). Without direct evidence that the symmetrically-dividing, aox5+ cells measured in this study do indeed differentiate into melanocytes, it is more likely that these cells are a dividing population of xanthophores/xanthoblasts. The authors should revise their claims accordingly.

    We agree with the editor and reviewers that the identities of the mitfa+aox5hi cells and the interplay between these cells and the mitfa+aox5lo cells is a fascinating, and originally unexpected, aspect of this manuscript. The issue, as we see it, is whether mitfa+aox5hi cells that arise via cell division during regeneration are multipotent pigment cell progenitors or ‘cryptic’ xanthophores. The experiments we have performed to address this ambiguity have not worked for technical reasons, so we have tempered text in the relevant Results and Discussion sections to leave both options open. We have backed off from calling these cells progenitors but have included additional data showing that they (i.e. the mitfa+aox5hi subpopulation of cells that we believe are daughters of mitfa+aox5hi cycling cells) express multiple markers associated with multipotent pigment cell progenitors that have been characterized in developing zebrafish. Our expanded Discussion is as follows:

    “Heterogeneity may also be evident by the additional mitfa+aox5hi G2/M adj subpopulation that likely arises via cell divisions during regeneration. There are reasons to think that this could be a progenitor subpopulation. Firstly, these cells arose in response to specific ablation of melanocytes. Secondly, this subpopulation expresses markers that are associated with multipotent pigment progenitors cells found during development (Budi, et al., 2011; Saunders, et al., 2019). Thirdly, although this subpopulation expresses aox5 and some other markers associated with xanthophores, we showed that differentiated xanthophores are not ablated by the melanocyte-ablating drug neocuproine and this mitfa+aox5hi subpopulation does not make new pigmented xanthophores following neocuproine treatment. However, current observations cannot definitively determine the potency and fates adopted by these cells. One possibility is that these cells are indeed progenitors that arise through cell divisions, are in an as yet undefined way lineally related to MP-0 and MP-1 subpopulations, and ultimately give rise to new melanocytes during additional rounds of regeneration. Given their expression of markers associated with multipotent pigment cell progenitors, these cells could be multipotent but fated toward the melanocyte lineage following melanocyte-specific ablation. However, we cannot exclude the possibility that these cells are another cell type. For example, there is a type of partially differentiated xanthophores that populate adult melanocyte stripes (McMenamin, et al., 2014). At least some of these cells arise from embryonic xanthophores that transitioned through a cryptic and proliferative state (McMenamin, et al., 2014). That the descendants remain partially differentiated could indicate that they are in more of a xanthoblast state and maintain proliferative capacity (Eom, et al., 2015). It is possible that some or all of the cells in question are melanocyte stripe-resident, partially-differentiated xanthophores that arise: a) from cell divisions that are triggered by loss of interactions with melanocytes or, b) simply to fill space that is vacated due to melanocyte death. Such causes for partially-differentiated xanthophore divisions have not been documented, but nonetheless this possibility must be considered given the mitfa and aox5 expression and proliferative potential of these cells. Transcriptional profiles of ‘cryptic’ xanthophores are not available to help clarify the nature of these cells. Lastly, the relationship between adult progenitor populations – MP-0, MP-1 and, potentially, mitfa+aox5hi G2/M adj – and other progenitors present at earlier developmental stages is unclear and could be defined through additional long-term lineage tracing studies. In particular, previous examinations of pigment cell progenitors in developing zebrafish have identified dorsal root ganglion-associated pigment cell progenitors in larvae that contribute to adult pigmentation patterns (Singh, et al., 2016; Dooley, et al., 2013; Budi, et al., 2011). It is possible that these cells give rise to the adult progenitors we have identified. The further alignment of cell types that have been observed in vivo and cell subpopulations defined through expression profiling is a necessary route for understanding the complex relationship between stem and progenitor cells in development, homeostasis, and regeneration.”

    (1) At line 140, it is noted that Xanthophores are pteridine-producing, but they also get their yellow color from carotenoids (especially in adults). This should be noted as well, especially since the authors display the xanthophore marker, scarb1, which plays a key role in xanthophore carotenoid coloration.

    [Mapping expression levels onto UMAP space for scarb1 and perhaps other markers of xan, irid, or proliferation would be helpful as a supplement to the dot plot in Fig 1 and could help to clarify the transcriptomic signature of mitfa+ aox5-hi cells and plausibility of the model that they are an McSC population. -Parichy]

    We thank the reviewer for the suggestion, and we have changed the text to include the carotenoid coloration facts of xanthophores as follows:

    “aox5 is expressed in differentiated xanthophores, a pteridine- and carotenoid-producing pigment cell type of zebrafish, and in some undifferentiated pigment progenitor cells”

    Additionally, we have also added a new Figure Supplement to Figure 1 (Figure 1 – figure supplement 3) with feature plots demonstrating the expression of xanthophore markers scarb1 and bco2b, iridophore markers lypc and cdh11, and proliferation markers pcna and mki67. As noted above, there is some heterogeneity within the large grouping of mitfa+aox5hi cells. Whereas some markers associated with xanthophores are broadly expressed in this grouping (e.g. scarb1), others have more restricted expression (e.g. bco2b). The heterogeneity could reflect multiple differentiation states of xanthophores, multiple types of differentiated xanthophores, xanthophore progenitors and/or less fate-restricted pigment cell progenitors that cluster in this grouping.

    (2) The authors should provide the list of genes that comprise their cluster signatures (line 252) as part of the supplementary tables.

    We have now included a table of genes in the cluster signatures. The Supplementary Table is called “Supplementary File 2.”

    (3) The authors should more clearly describe how they performed lineage tracing (line 339). Additionally, for the corresponding figure 4E, the authors should list the number of cells traced. The source data only contains calculated percentages rather than counts for each type of differentiation. My understanding is that the number listed in the figure legend is the number of fish (i.e. n = 4), but this should be clarified as well.

    [A supplementary figure of labeled cells is important here with enough context to show that cells can be re-identified unambiguously. Additionally note that "lineage tracing" will typically be assumed to mean single-cell labeling and tracking, so if that is not the case for these experiments it would be preferable to use an alternative descriptor. -Parichy]

    We have included additional detail in our revised manuscript. In Figure 4E we now include the number of cells imaged and have included a breakdown of the raw numbers in the Source Data. We have also included Supplementary Animations as examples of the single-cell tracing that we perform through serial imaging.

    Additionally, the point about using ‘lineage tracing’ is well taken. We have replaced this with ‘serial imaging’ through the text.

    (4) Line 321, the authors list the mean regeneration percentages for the kita and kitlga(lf) mutants, but these differences are not significantly different according to Figure 4B. By listing the means (which should be noted), the authors seem to be highlighting the differences but then do not comment on them. The description and integration of this result into the main text should be clarified.

    We have changed the wording in the text to clarify that the mean percentage is being listed. We have also reworded the text to de-emphasize the mean percentage difference between kita(lf) and kitlga(lf) mutants, instead highlighting that their defects are similar. In the figure legend we have clarified that the mean percentage regeneration is being shown.

    (5) In Figure 6E, the RNA-velocity result is not particularly consistent with the authors' claims. Visually, the arrows seem fairly randomly directed. The data in 6B, showing gene expression associated with the S phase and G2/M phase much more clearly convey the directionality of the loop (S phase, followed by G2/M). I suggest that the authors weaken their claim about the RNA-velocity result or remove it altogether and focus on the cell cycle-related gene expression signatures.

    We thank the reviewer for their careful eye here. We have decided to remove the RNA-velocity result previously displayed in Figure 6E. As the reviewer points out the results are more clearly demonstrated by Figure 6B.

  3. eLife

    eLife assessment

    In this work Frantz et al. assess transcriptomic states of melanocyte stem cells that are recruited to differentiate during the process of melanocyte regeneration in zebrafish and they analyze roles for Kit signaling in this process. The analyses are nicely done, and the paper requires only relatively minor modifications and clarifications. The study will provide new insights into melanocyte stem cell biology that should be of interest to those studying pigmentation, regeneration, and melanoma biology using zebrafish and other systems.

  4. eLife

    Reviewer #1 (Public Review):

    In this study, the authors aim to identify the cell state dynamics and molecular mechanisms underlying melanocyte regeneration in zebrafish. By analyzing thousands of single-cell transcriptomes over regeneration in both wild-type and Kit mutant animals, they provide thorough and convincing evidence of (1) two paths to melanocyte regeneration and (2) that Kit signaling, via the RAS/MAPK pathway, is a key regulator of this process. Finally, the authors suggest that another proliferative subpopulation cells, expressing markers of a separate pigment cell type, constitute an additional population of progenitors with the ability to contribute to melanocytes. The data supporting this claim are not as convincing, and the authors failed to show that these cells did indeed differentiate into melanocytes. Despite the challenges of describing this third cell state, this study offers compelling new findings on the mechanisms of melanocyte regeneration and provides paths forward to understanding why some animals lack this capacity.

    The majority of the main conclusions are well supported by the data, but one claim, in particular, should be revisited by the authors.

    (1) Provided evidence that the aox5(hi)mitfa(lo) population of cells contributes to melanocyte regeneration is inconclusive and somewhat circumstantial. First, the transcriptional profiles of these cells are much more consistent with the xanthophore lineage. Indeed, xanthophores have been shown to express mitfa (in embryos in Parichy, et al. 2003 (PMID: 10862741), and in post-embryonic cells in Saunders, et al. 2019). Second, while the authors address this possibility in Supplemental figure 7 by showing that interstripe xanthophores fail to divide following melanocyte ablation, they fail to account for the stripe-resident xanthophores/xanthoblasts. The presence and dynamics of aox5+ stripe-resident xanthophores/xanthoblasts are detailed in McMenamin, et al., 2014 (PMID: 25170046) and Eom, et al., 2015 (PMID: 26701906). Without direct evidence that the symmetrically-dividing, aox5+ cells measured in this study do indeed differentiate into melanocytes, it is more likely that these cells are a dividing population of xanthophores/xanthoblasts. The authors should revise their claims accordingly.

    Minor revisions

    (1) At line 140, it is noted that Xanthophores are pteridine-producing, but they also get their yellow color from carotenoids (especially in adults). This should be noted as well, especially since the authors display the xanthophore marker, scarb1, which plays a key role in xanthophore carotenoid coloration.
    [Mapping expression levels onto UMAP space for scarb1 and perhaps other markers of xan, irid, or proliferation would be helpful as a supplement to the dot plot in Fig 1 and could help to clarify the transcriptomic signature of mitfa+ aox5-hi cells and plausibility of the model that they are an McSC population. -Parichy]

    (2) The authors should provide the list of genes that comprise their cluster signatures (line 252) as part of the supplementary tables.

    (3) The authors should more clearly describe how they performed lineage tracing (line 339). Additionally, for the corresponding figure 4E, the authors should list the number of cells traced. The source data only contains calculated percentages rather than counts for each type of differentiation. My understanding is that the number listed in the figure legend is the number of fish (i.e. n = 4), but this should be clarified as well.
    [A supplementary figure of labeled cells is important here with enough context to show that cells can be re-identified unambiguously. Additionally note that "lineage tracing" will typically be assumed to mean single-cell labeling and tracking, so if that is not the case for these experiments it would be preferable to use an alternative descriptor. -Parichy]

    (4) Line 321, the authors list the mean regeneration percentages for the kita and kitlga(lf) mutants, but these differences are not significantly different according to Figure 4B. By listing the means (which should be noted), the authors seem to be highlighting the differences but then do not comment on them. The description and integration of this result into the main text should be clarified.

    (5) In Figure 6E, the RNA-velocity result is not particularly consistent with the authors' claims. Visually, the arrows seem fairly randomly directed. The data in 6B, showing gene expression associated with the S phase and G2/M phase much more clearly convey the directionality of the loop (S phase, followed by G2/M). I suggest that the authors weaken their claim about the RNA-velocity result or remove it altogether and focus on the cell cycle-related gene expression signatures.

  5. eLife

    Reviewer #2 (Public Review):

    Franz and colleagues set out to understand the mechanisms and cell types that contribute to melanocyte regeneration in the adult skin. Previously, they used genetics and imaging to identify cell populations (progenitors) in the adult skin that they believe contribute to melanocyte regeneration in adult zebrafish (Iyengar et al., 2015). Here, they use scRNA-seq to understand the molecular nature of these cells following melanocyte ablation with the copper chelator, neocuproine. From these studies, they claim to identify three types of progenitors (called melanocyte stem cells, McSCs): cells that give rise directly to differentiated melanocytes and depend on kit signaling; cells that undergo division before becoming fully differentiated; and cells that express high levels of a xanothophore marker (a yellow pigment cell) that also undergo cell division.

    Strengths:
    The main strength of this work is the generation of scRNA-seq datasets of cells that express a melanocyte marker (mitfa) at multiple time points in adult skin during regeneration. This is an exciting dataset, and unique. The work gives an idea of the complexity of the regeneration process and paves the road for future studies on how McSC lineages contribute to melanoma. It is interesting to see how many of the processes and zebrafish cell types are conserved during evolution. By studying skin-associated melanocyte progenitors in adults, the authors provide insight into mechanisms poorly understood about melanocyte regeneration.

    Weaknesses:
    (1) Data Interpretation in context: We have concerns regarding the labelling of the cells of interest "stem cells"; we prefer the term the authors themselves use "progenitors" (Iyengar et al., 2015). The authors do not place their work in the context of the wider field, especially with regards to the work on xanthophores and on regenerating melanocytes and adult McSCs in the embryo that contribute to the adult stripe.

    (2) Cell type identity: Zebrafish contain another cell type called xanthophores that can also express mitfa and aox5 (Saunders et al., 2019). Indeed, in their supplementary tables, the authors call many of the mitfa+ aox5+ cells "xanthophores" based on their gene expression. There is no evidence here that these cells give rise to melanocytes. In their studies in Figure 7, we think that based on the shape of the cells, they may be looking at dividing xanthophores or unpigmented xanthophore precursors (McMenamin et al., 2014), rather than melanocyte stem cells. We don't know why these cells are dividing, but perhaps the loss of melanocytes in the adult stripe leads to an expansion of xanthophores.

    (3) Analysis: The statistical approaches are not always correct, and some choices in the scRNA-seq analysis should be explained and/or revisited.

  6. eLife

    Reviewer #3 (Public Review):

    This manuscript describes McSC states and McSC function during regeneration in zebrafish using both a scRNAseq timecourse and classic zebrafish experimentation, including lineage tracing and mutant lines. Altogether this study provides a more holistic look at pigment regeneration following injury and helps to validate the role of signaling pathways implicated in McSC biology by previous studies. The major question addressed by this manuscript is whether McSC heterogeneity can explain the highly regenerative nature of the zebrafish pigmentary system. The observations reported in this manuscript confirm this view, eloquently using a time course of single-cell transcriptomics for predictive purposes followed up by mechanistic studies to confirm the fate of different McSC subclusters. This study very nicely complements and extends our current understanding of how McSCs function during regeneration and provides novel datasets for further interrogation. Perhaps the most exciting aspect of the data is the identification of a novel marker (aox5) to identify self-renewing McSCs; this tool could be employed to identify these cells and address their potential in the context of expanding these cells for therapeutic purposes or address their contribution as melanoma stem cells. This study will be of general interest to researchers interested in pigment regeneration, stem cell-based therapeutics for pigment disorders, and the basic biology of stem cells and their heterogeneity.

    While this paper certainly extends previous observations of McSCs, the idea of McSC heterogeneity is not necessarily novel. In mouse, KIT-dependent and KIT-independent McSC populations have been identified (Ueno 2015) as well as other McSC subpopulations with different potentials (CD34+/-, Joshi 2019). While this manuscript does a much more comprehensive job of describing this heterogeneity, which is fantastic, some of the previous literature on the topic could be better acknowledged and integrated. Despite this criticism, this manuscript provides the most comprehensive look to date at McSC dynamics across the regenerative period and provides ample datasets for secondary analyses to generate/confirm additional hypotheses.

  7. Version published to 10.1101/2022.04.27.489712v1 on bioRxiv