Single-cell profiling coupled with lineage analysis reveals vagal and sacral neural crest contributions to the developing enteric nervous system

  1. Jessica Jacobs-Li
  2. Weiyi Tang
  3. Can Li
  4. Marianne E Bronner

  • Curated by eLife

    eLife logo

    eLife assessment

    This is an elegant study combining virally-delivered lineage-tracing with single cell RNA-sequencing of the developing chicken enteric nervous system, showing potentially differential contribution of cell identities from the sacral and vagal neural crest. Addressing this important issue is pivotal to understanding basic enteric nervous system development as well as to devise therapeutic approaches to enteric neuropathies. The study is therefore generally interesting and in particular to researchers in the fields of enteric neuroscience and peripheral nervous system development. Lack of a basic classification scheme of neuronal cell types in the chicken, limited computational and functional analysis on a relatively immature stage and makes the conclusions of this work preliminary in its current state.

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

During development, much of the enteric nervous system (ENS) arises from the vagal neural crest that emerges from the caudal hindbrain and colonizes the entire gastrointestinal tract. However, a second ENS contribution comes from the sacral neural crest that arises in the caudal neural tube and populates the post-umbilical gut. By coupling single-cell transcriptomics with axial-level-specific lineage tracing in avian embryos, we compared the contributions of embryonic vagal and sacral neural crest cells to the chick ENS and the associated peripheral ganglia (Nerve of Remak and pelvic plexuses). At embryonic day (E) 10, the two neural crest populations form overlapping subsets of neuronal and glia cell types. Surprisingly, the post-umbilical vagal neural crest much more closely resembles the sacral neural crest than the pre-umbilical vagal neural crest. However, some differences in cluster types were noted between vagal and sacral derived cells. Notably, RNA trajectory analysis suggests that the vagal neural crest maintains a neuronal/glial progenitor pool, whereas this cluster is depleted in the E10 sacral neural crest which instead has numerous enteric glia. The present findings reveal sacral neural crest contributions to the hindgut and associated peripheral ganglia and highlight the potential influence of the local environment and/or developmental timing in differentiation of neural crest-derived cells in the developing ENS.

Article activity feed

  1. Version published to 10.7554/elife.79156 on eLife
  2. eLife

    Author Response

    Reviewer #2 (Public Review):

    1. Analytical approaches are in the current form preliminary and not enough to draw firm biological conclusions. While the datasets are large (which is highly appreciated), they represent a relatively early stage of ENS development and possible differences between vagal and sacral-derived populations could partially be attributed to difference in maturity. Maturity will surely not explain the whole difference observed but needs to be factored into the interpretation. As scRNA-seq datasets from the mature chicken ENS are lacking (as well as detailed IHC-based neural classification system) the inference made in the paper between molecular classes and functional types are premature.

    We appreciate this comment and think it is an excellent suggestion that we definitely plan to do. This made us realize that we failed to clarify in the text why we chose this particular time point for our study, which is two-fold.

    First, we are particularly interested in how neural crest cells choose their prospective fates. E10 is a time when the post-umbilical gut has been completely populated by both vagal and sacral neural crest cells for 2 days so cells are in the process of differentiation but there still exists a large precursor pool. For this reason, we can capture both precursors and some differentiated neuronal subtypes. We have clarified this point in the revised manuscript and now focus much more on the precursor population to identify both genes that are common to vagal and sacral neural crest cells as well as those that are distinct. This enables us to formulate testable hypotheses for the role of potential role of particular transcription factors is allocation of cell fate. Of particular interest, we find that at E10, the sacral neuronal precursor pool is largely depleted whereas the vagal crest has a substantial neuronal precursor pool. Thus, we believe this is the perfect time point for initial analysis.

    Second and perhaps even more important, in the US, chick embryos are not considered vertebrates until after E10. Thus, E10 represents the last timepoint we can raise embryos without animal approvals which are not currently in hand. We completely agree that performing experiments at later timepoints will be incredibly valuable and therefore are now applying for approvals. But realistically, these take several months and thus would delay publication of our datasets (already delayed due to Covid restrictions) for at least another year. Therefore, we propose to publish the mature dataset as a Research Advance that would focus on differences between mature neuronal subtypes between preumbilical vagal, post-umbilical vagal and sacral datasets that would nicely complement the current work. Instead, we have refocused this paper on the precursor to differentiated neuron transition.

    I should mention that this refocusing seems particularly important given that our original aim was to explore differences between vagal and sacral neural crest contributions to the gut. However, the single cell data reveals strong overlap between sacral and vagal neural crest contributions to the postumbilical gut, suggesting a strong environmental influence on cell fate decisions.

    Specific concerns:

    1. Analysis of scRNA-sequenced sacral- versus vagal-derived ENS reveals clusters consistent with a non-ENS identity (endothelial, muscle, vascular and more). Previous studies in mouse using the neural crest tracing line Wnt1-Cre has not demonstrated such diverse progenies of neural crest from any region. An exception being a small population of mesenchymal-like cells (Ling and Sauka-Spengler, Nat Cell Biol. 2019; Zeisel et al., Cell 2018; Morarach et al., 2021; Soldatov et al., Science 2019). Therefore, the claimed broad potential of 6 of 13 neural crest giving rise to diverse gut cell populations warrants more validating experiments.

    We thank the reviewer for this comment. We clarify that hematopoetic clusters have dropped out upon reanalysis. The other clusters we believe are real based on gene markers used in previous studies to identify cell types such as neural crest-derived melanocytes like Mlana, Dct, and Mitf.

    1. Several earlier studies have revealed that parts of the ENS is derived from neural crest that attach to nerve bundles, obtain a schwann cell precursor-like identity and thereafter migrate into the gut (Uesaka et al. J Neurosci 2015 and Espinosa-Medina et al, PNAS 2017). The current work in chicken needs to be interpretated in the light of these findings and the publications should be discussed in relevant sections of the introduction and discussion.

    Thank you for this suggestion. We agree and indeed our data cannot differentiate between SCPs, which are neural crest-derived, versus early migrating neural crest cells. We have added this point to the discussion and also discuss these papers in more detail.

    1. The analysis indicates the presence of melanocytes. It is not clear why they are part of the GI-tract preparations. Could they correspond to another cell type, with partially overlapping gene expression profile as melanocytes?

    We have assigned these as melanocytes based on expression of Mlana, Mitf, and Dct as highly upregulated genes. These have been used in previous studies to identify neural crest derived melanocytes in the heart (Chen et al., 2021)

    1. As evident, the sacral- and vagal-derived ENS are not clonally related. To decipher differentiation paths and relations between clusters, individual analysis of the different datasets are needed. With only one UMAP representing the merged datasets combined with little information on markers, it is hard to evaluate the soundness of the conclusions regarding cell-identities of clusters and lineage differentiation.

    This is an excellent suggestion and we apologize for not including this previously. We have now added individual pre-umbilical vagal, post-umbilical vagal and sacral neural crest datasets as well as trajectory analysis for each.

    1. E10 is a relatively early stage in chicken ENS development. Around E7, the intestines do not contain differentiated neurons even. The relative high expression of Hes5 (marking mature enteric glia in the mouse; Morarach et al., 2021) in the vagal neural crest population might be explained by the more mature state of vagal versus sacral ENS. As also outlined below, Th/Dbh are known to be transiently expressed in the developing ENS why they could indicate the relative immaturity of sacral neural crest rather than differential neural identities. These issues need to be taken into account when interpreting biology from scRNA-seq data.

    We completely agree. We now clarify that we are particularly interested in how neural crest cells choose their prospective fates. We chose the E10 time point because this reflects a time point when the post-umbilical gut has been completely populated by both vagal and sacral neural crest cells for 2 days so cells are in the process of differentiation but there still exists a large precursor pool. For this reason, we can capture both precursors and some differentiated neuronal subtypes. Notably, the sacral derived precursors seem to be glial in flavor whereas neuronal precursors appear to be absent. We have clarified this point in the revised manuscript.

    1. Unlike the guineapig, and to some extent pig and murine ENS, the physiology of chicken enteric neurons has not been well characterized yet. Therefore, it is highly advisable to refrain from a nomenclature of clusters designating functions. Several key molecular markers are known to differ between murine, guineapig, rat and human systems. IPANs are a good example where differential expression is seen (SST in human but not mice; CGRP labels some IPANS in mouse, but not in guineapig, where Tac1 instead is expressed). IPANs are not defined in the chicken very well, and molecular markers found in other species may not be valid. Adrenergic and noradrenergic neurons have not been validated in the ENS (although, TH and Dbh have been observed in the especially in the submucosal ENS). Cholinergic neurons are also mentioned in the text, but do not appear in the figures as a defined group.

    Another reason to refrain from functional nomenclature is that a rather early stage is analysed in the present study, without possibilities to compare with scRNA-seq data from the mature chicken ENS (which was performed in Morarach et al, 2021 for the mouse). Recent data suggest that considerable differentiation may occur even in postmitotic neurons, and several markers are known to display a transient expression pattern (TH, DBH and NOS1; Baetge and Gershon 1990; Bergner et al., 2014; Morarach et al., 2021) why caution should be taken to infer neuronal identities to clusters.

    This is an excellent point and we thank the reviewer for this valuable input. Accordingly, we have now renamed the clusters based on prominent gene expression rather than neuronal or precursor subtype. Indeed we struggled with finding appropriate names making this comment all the more useful.

    1. The immunohistochemical analysis (Figure 5,6) is an essential complementary addition and validation of scRNA-seq. However, it is very difficult to discern staining when magenda and red are combined to display coexpression.

    Good point. This has been changed to be more readily discernible and higher magnification views have been added.

    1. To give more information to the field and body of evidence for claims made, quantifications relating to the analysis in Figures 5 and 6 are warranted as well as an expanded set of marker genes that align with the scRNA-seq results.

    Good point. We have added additional markers as suggested. In terms of quantitation, we can include numbers of labeled cells in a particular region but this may give a false impression of degree of contribution since we are using different viruses for vagal vs sacral that may have different titers making it a bit like comparing apples and oranges. We now emphasize that our labeling approach does not mark the entire population and that the degree of labeling can be variable.

    1. Correlations between genes and functions/neuron class are in many cases wrong (including Grm3, Gad1, Nts, Gfra3, Myo9d, Cck and more).

    Good point. We have toned this down.

    1. Attempts to subcluster neuronal populations are needed (Figure 7). However, to understand the biology, it is important to address which cells are sacral versus vagal-derived. Additionally, related to previous comment, as the vagal and sacral neurons are not clonally related, it would be important to make separate analysis of neurons relating to each region.

    Good point. We have added additional analysis to address this important point in what is now Fig 6 and in particular validated sacral contributions to glial cells (new Fig 8).

  3. eLife

    eLife assessment

    This is an elegant study combining virally-delivered lineage-tracing with single cell RNA-sequencing of the developing chicken enteric nervous system, showing potentially differential contribution of cell identities from the sacral and vagal neural crest. Addressing this important issue is pivotal to understanding basic enteric nervous system development as well as to devise therapeutic approaches to enteric neuropathies. The study is therefore generally interesting and in particular to researchers in the fields of enteric neuroscience and peripheral nervous system development. Lack of a basic classification scheme of neuronal cell types in the chicken, limited computational and functional analysis on a relatively immature stage and makes the conclusions of this work preliminary in its current state.

  4. eLife

    Reviewer #1 (Public Review):

    The authors of this manuscript address the question of whether vagal and sacral neural crest make distinct contributions to the enteric nervous system (ENS). The ENS regulates intestinal motility and many intestinal homeostatic functions; mutations in genes involved in ENS development lead to defects that can range from mild to catastrophic. The best studied of the ENS neuropathies is Hirschsprung disease, which is thought to result from failure of vagal neural crest cells to migrate properly into the distal intestine to differentiate into ENS neurons and glia. However, sacral neural crest cells are known to contribute to the distal ENS and have to migrate a considerably shorter distance. Thus, understanding whether there are distinct vagal and sacral contributions to the ENS provides insights into basic ENS biology as well as the basis of human disease. Previous transplantation and ablation studies have already revealed that vagal and sacral neural crest have differing ENS developmental potentials, although this has not been directly correlated with discrete cell types. Here the authors combine single cell RNA sequencing and a viral lineage tracing technique that is new to avians to gain insight into the different ENS cell types generated by vagal and sacral neural crest along the length of the intestine. They find that vagal and sacral neural crest exhibit distinct transcriptional profiles and contribute both similar and different progeny to the ENS. For example, both vagal and sacral crest contribute to progenitor cells, connective tissue and neurons, but most secretomotor neurons are vagal crest-derived whereas most adrenergic neurons and melanocytes in the distal intestine are sacral-crest derived. The authors also suggest a role of the local environment in determining the fate of vagal and sacral derivatives. The data presented in this manuscript provide a multitude of hypotheses about similarities and differences between vagal and sacral derived ENS cells. However, a shortcoming of the manuscript is that all of these hypotheses remain untested.

  5. eLife

    Reviewer #2 (Public Review):

    The manuscript by Tang et al investigates the potential difference between the enteric nervous system derived from different axial regions of chicken embryos. By applying single cell RNA-sequencing (scRNA-seq) analysis of virally traced enteric cell populations, the authors conclude that vagal and sacral neural crest may contribute to different neural subtypes and non-neural cells in the sub-umbilical ENS. Confirming previous studies, their method also demonstrates the exact axial levels of the GI-tract populated by sacral neural crest. The analysis suggests that NPY/VIP+ neurons mainly arise from vagal neural crest in both the pre- and postumbilical ENS, while sacral neural crest mainly contribute with Th/Dbh/Ddc+ neurons. Sacral neural crest also appears to generate a greater proportion of schwann cell-like cells and melanocytes to the gut.

    While early studies in the chicken model (combined with quail) founded many of the key principles underlying the emergence of the ENS from different neural crest sources, the chicken model currently lags behind in the implementation of modern transcriptomic and neurophysiological approaches. This paper provides a long-saught comprehensive scRNA-seq datasets of the chicken ENS which is clearly lacking in the ENS field. The elegant viral delivery allows targeting of both vagal and sacral neural crest in the same embryo offering clear advantages to other commonly used model systems (including the mouse). However, analytical approaches are in the current form preliminary and not enough to draw firm biological conclusions. While the datasets are large (which is highly appreciated), they represent a relatively early stage of ENS development and possible differences between vagal and sacral-derived populations could partially be attributed to difference in maturity. Maturity will surely not explain the whole difference observed but needs to be factored into the interpretation. As scRNA-seq datasets from the mature chicken ENS are lacking (as well as detailed IHC-based neural classification system) the inference made in the paper between molecular classes and functional types are premature.

    Specific concerns:

    1. Analysis of scRNA-sequenced sacral- versus vagal-derived ENS reveals clusters consistent with a non-ENS identity (endothelial, muscle, vascular and more). Previous studies in mouse using the neural crest tracing line Wnt1-Cre has not demonstrated such diverse progenies of neural crest from any region. An exception being a small population of mesenchymal-like cells (Ling and Sauka-Spengler, Nat Cell Biol. 2019; Zeisel et al., Cell 2018; Morarach et al., 2021; Soldatov et al., Science 2019). Therefore, the claimed broad potential of neural crest giving rise to diverse gut cell populations warrants more validating experiments.

    2. Several earlier studies have revealed that parts of the ENS is derived from neural crest that attach to nerve bundles, obtain a schwann cell precursor-like identity and thereafter migrate into the gut (Uesaka et al. J Neurosci 2015 and Espinosa-Medina et al, PNAS 2017). The current work in chicken needs to be interpretated in the light of these findings and the publications should be discussed in relevant sections of the introduction and discussion.

    3. The analysis indicates the presence of melanocytes. It is not clear why they are part of the GI-tract preparations. Could they correspond to another cell type, with partially overlapping gene expression profile as melanocytes?

    4. As evident, the sacral- and vagal-derived ENS are not clonally related. To decipher differentiation paths and relations between clusters, individual analysis of the different datasets are needed. With only one UMAP representing the merged datasets combined with little information on markers, it is hard to evaluate the soundness of the conclusions regarding cell-identities of clusters and lineage differentiation.

    5. E10 is a relatively early stage in chicken ENS development. Around E7, the intestines do not contain differentiated neurons even. The relative high expression of Hes5 (marking mature enteric glia in the mouse; Morarach et al., 2021) in the vagal neural crest population might be explained by the more mature state of vagal versus sacral ENS. As also outlined below, Th/Dbh are known to be transiently expressed in the developing ENS why they could indicate the relative immaturity of sacral neural crest rather than differential neural identities. These issues need to be taken into account when interpreting biology from scRNA-seq data.

    6. Unlike the guineapig, and to some extent pig and murine ENS, the physiology of chicken enteric neurons has not been well characterized yet. Therefore, it is highly advisable to refrain from a nomenclature of clusters designating functions. Several key molecular markers are known to differ between murine, guineapig, rat and human systems. IPANs are a good example where differential expression is seen (SST in human but not mice; CGRP labels some IPANS in mouse, but not in guineapig, where Tac1 instead is expressed). IPANs are not defined in the chicken very well, and molecular markers found in other species may not be valid. Adrenergic and noradrenergic neurons have not been validated in the ENS (although, TH and Dbh have been observed in the especially in the submucosal ENS). Cholinergic neurons are also mentioned in the text, but do not appear in the figures as a defined group. Another reason to refrain from functional nomenclature is that a rather early stage is analysed in the present study, without possibilities to compare with scRNA-seq data from the mature chicken ENS (which was performed in Morarach et al, 2021 for the mouse). Recent data suggest that considerable differentiation may occur even in postmitotic neurons, and several markers are known to display a transient expression pattern (TH, DBH and NOS1; Baetge and Gershon 1990; Bergner et al., 2014; Morarach et al., 2021) why caution should be taken to infer neuronal identities to clusters.

    7. The immunohistochemical analysis (Figure 5,6) is an essential complementary addition and validation of scRNA-seq. However, it is very difficult to discern staining when magenda and red are combined to display co-expression.

    8. To give more information to the field and body of evidence for claims made, quantifications relating to the analysis in Figures 5 and 6 are warranted as well as an expanded set of marker genes that align with the scRNA-seq results.

    9. Correlations between genes and functions/neuron class are in many cases wrong (including Grm3, Gad1, Nts, Gfra3, Myo9d, Cck and more).

    10. Attempts to subcluster neuronal populations are needed (Figure 7). However, to understand the biology, it is important to address which cells are sacral versus vagal-derived. Additionally, related to previous comment, as the vagal and sacral neurons are not clonally related, it would be important to make separate analysis of neurons relating to each region.

  6. Version published to 10.1101/2022.05.09.491197v2 on bioRxiv
  7. Version published to 10.1101/2022.05.09.491197v1 on bioRxiv