scMultiome analysis identifies embryonic hindbrain progenitors with mixed rhombomere identities

  1. Yong-Il Kim
  2. Rebecca O'Rourke
  3. Charles G Sagerström

  • Curated by eLife

    eLife logo

    eLife assessment

    This study transcriptomically profiles the developing zebrafish hindbrain from gastrulation through stages of rhombomere formation. The strength is that the transcriptomic data will be a valuable resource to the field. The paper would profit from a deeper analysis of functional aspects of hindbrain development during its segmentation into rhombomeres.

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Rhombomeres serve to position neural progenitors in the embryonic hindbrain, thereby ensuring appropriate neural circuit formation, but the molecular identities of individual rhombomeres and the mechanism whereby they form has not been fully established. Here, we apply scMultiome analysis in zebrafish to molecularly resolve all rhombomeres for the first time. We find that rhombomeres become molecularly distinct between 10hpf (end of gastrulation) and 13hpf (early segmentation). While the embryonic hindbrain transiently contains alternating odd- versus even-type rhombomeres, our scMultiome analyses do not detect extensive odd versus even molecular characteristics in the early hindbrain. Instead, we find that each rhombomere displays a unique gene expression and chromatin profile. Prior to the appearance of distinct rhombomeres, we detect three hindbrain progenitor clusters (PHPDs) that correlate with the earliest visually observed segments in the hindbrain primordium that represent prospective rhombomere r2/r3 (possibly including r1), r4, and r5/r6, respectively. We further find that the PHPDs form in response to Fgf and RA morphogens and that individual PHPD cells co-express markers of multiple mature rhombomeres. We propose that the PHPDs contain mixed-identity progenitors and that their subdivision into individual rhombomeres requires the resolution of mixed transcription and chromatin states.

Article activity feed

  1. Version published to 10.7554/elife.87772 on eLife
  2. eLife

    eLife assessment

    This study transcriptomically profiles the developing zebrafish hindbrain from gastrulation through stages of rhombomere formation. The strength is that the transcriptomic data will be a valuable resource to the field. The paper would profit from a deeper analysis of functional aspects of hindbrain development during its segmentation into rhombomeres.

  3. eLife

    Reviewer #1 (Public Review):

    In this manuscript, Kim et al. investigate the molecular basis for hindbrain segmentation by performing combined single cell nucleus RNAseq and ATACseq (scMultiome) on zebrafish embryonic hindbrain tissue. Hindbrain segmentation is fundamental to head development in vertebrate species. Decades of research have provided many insights into the gene regulatory cascades that control the progressive subdivision of the hindbrain territory into segments (rhombomeres). These studies have enabled the formulation of gene regulatory network (GRN) models that depict these regulatory interactions. However, many aspects of the GRN need further clarification, including the early steps of pre-rhombomeric patterning, and the factors that respond to axial signaling pathways such as RA and FGF. The dataset in this study provides a comprehensive view of gene expression and chromatin states during hindbrain segmentation, thus it is a valuable resource for characterizing the underlying GRN. The authors demonstrate the utility of this data by comparing the molecular profiles between different rhombomeres and tracing when and how these profiles arise during development.

    Four main findings are presented:

    1. Each rhombomere has a unique molecular profile.
    2. There is no clear molecular signature for odd versus even rhombomeres, nor any overt repeating two-segment molecular identities.
    3. The mature rhombomeres emerge through the subdivision of three mixed-identity 'primary hindbrain progenitor domains' (PHPDs) that correspond to r2/r3, r4, and r5/r6, respectively.
    4. RA and FGF signaling control formation of the primary hindbrain progenitor domains.

    These findings are well supported by the data but in my opinion they mainly confirm what was already known and do not significantly advance our mechanistic understanding of rhombomere formation, which is the aim of the paper.

    Strengths:
    This comprehensive dataset will be very valuable to researchers in the field. The authors successfully demonstrate its utility by resolving unique molecular profiles for each rhombomere and identifying some novel markers.

    The authors make excellent use of HCR to validate their findings, such as the co-expression of vgll3 and egr2b in r2/r3 cells at 10hpf, which implies mixed identities of PHPD cells.

    The performance of scMultiome analysis on tissue from DEAB-treated embryos (depleted RA signaling) is exciting and holds much promise for identifying RA-dependent gene regulatory cascades that govern caudal hindbrain patterning. Assessing the contribution of control versus DEAB-treated cells to the various UMAP clusters is a very nice way to identify the altered cell states in the RA-depleted hindbrain. This confirms a complete absence of r5 and r6 in the DEAB-treated embryos at this developmental stage, as was inferred from in-situ approaches in earlier studies.

    Weaknesses:
    The major weakness of this work is that it only provides an incremental mechanistic advance to our current understanding of the molecular basis for rhombomere formation. The descriptions of gene expression are useful but for the most part they are rather shallow lines of enquiry that confirm what was already known from previous, less comprehensive studies of gene expression. For example, regarding the identification of PHPDs, it has long been known that r5/r6 share a progenitor domain that is demarcated by mafba expression. Similarly, RA and Fgf signaling have already been shown to be required for anterior-posterior patterning in the pre-rhombomeric hindbrain. The identification of mixed-identity progenitors in PHPDs, and the characterisation of the changes in transcription and chromatin state in response to RA signaling perturbation are really exciting starting points for deeper analysis of the underlying GRN. However, it is a shame that no effort is made to glean mechanistic insights from this dataset by computational GRN inference.

  4. eLife

    Reviewer #2 (Public Review):

    The hindbrain is one of three primary anatomical domains of the developing brain, and is thought to be important for motor activity, respiratory rhythm, and sleep and wake behavior. The purpose of this study was to analyse spatiotemporal changes in gene expression during early development of the hindbrain. The authors used single cell RNA sequencing and ATAC sequencing at three developmental stages of zebrafish embryo development to characterize the transcriptomic changes that occur as the hindbrain neuroepithelium resolves into rhombomeres and the expression of a small number of genes was validated by in situ hybridization. The bulk of the "omic" dataset potentially provides a resource for the field to functionally analyze, but otherwise only incrementally advances our understanding of hindbrain rhombomere development and patterning. The primary conclusion from the work is that hindbrain progenitor domains contain mixed identity progenitors that eventually resolve into individual mature rhombomeres. This concept has been known historically for quite some time based on the expression of many genes of the Hox and other gene families, despite the authors describing this at higher resolution through analyses of whole genome expression. Unfortunately, the paper is largely a descriptive resource of transcriptomic data which in the absence of functional experimentation tells us very little that's new about the fundamental development or function of rhombomeres.

  5. eLife

    Reviewer #3 (Public Review):

    Rhombomeres are key organizational structures for building cell type and even functional diversity in the brainstem. How these rhombmeres ultimately arise from a broad neuro-epithilium remains unclear. While genetic, cellular, tissue, and morphogen manipulations have revealed key processes in rhombomere development the hierarchical organization of neuron-epithelium into individual rhombomeres was less well understood. For example it is thought that rhombomeres are organized in an even odd fashion where two base identities i.e. even or odd where laminated with paired identifies i.e. rhombomeres 1 and 2 being paired and so on. However, there are many exceptions to these organizing constructs at the gene expression levels.

    To further interrogate early development of the hindbrain neuro-epithelium and gain insight as to how rhombomere identities emerge at the earliest stages, Kim et al used ATACseq and RNAseq to query chromatin landscapes and gene expression for single nuclei at different developmental stages of zebrafish hindbrain development. The goal of the two pronged approach termed scMultiome analysis was to gain additional insight beyond either method individually for characterizing early events in rhombomere differentiation.

    Using scMultiome, three stages of zebrafish hindbrain development were examined at 10hpf(whole embryos), 13hpf, and 16hpf. In the early hindbrain, the data shows that at 13hpf early rhombomere identities can be resolved but that the typical markers seen later are not fully expressed or resolved. At 10hpf clear rhombomere identities are not present. Rather at very early stages, the analysis suggests that three domains for pre-rhombomeres encompassing HB1 - r2+r3 (possibly r1, but this remains to be resolved); HB2 - r5+r6; and HB3 - 4 are present. These clusters or PHPDs are mixed populations that presumably resolve later as the embryo matures. They are shown to be responsive to developmental signals that pattern the neuroepithelium supporting the premise that these are rhombomeric organization structures.

    Altogether the use of two methods of transcriptional interrogation i.e. ATACseq and RNA seq are strengths for the presented work to offer increased resolution of cell type characterization. The data analysis is reasonably supported by expression studies using in situ Hybridization Chain Reaction (HCR) to show mixed markers in the early stages. the PHPDs are also responsive to perturbation in retinoid acid, supporting the overall premise.

    Overall, the work is well executed and analyzed. The impact in the field largely resides in bringing increasing resolution to earlier stages of rhombomere development and re-examining long held paradigms about when and potentially how rhombomere periodicity and pairing are established at the earliest stages. The premise that pre-rhombomeres may first establish large domains that sort or otherwise resolve themselves into rhombomeres is the most notable outcome from the work and will be seen as impactful in the field.

  6. Version published to 10.1101/2023.01.27.525932v1 on bioRxiv