1. Developmental Biology
  2. Neuroscience
Download icon

Enteric glia as a source of neural progenitors in adult zebrafish

  1. Sarah McCallum
  2. Yuuki Obata
  3. Evangelia Fourli
  4. Stefan Boeing
  5. Christopher J Peddie
  6. Qiling Xu
  7. Stuart Horswell
  8. Robert N Kelsh
  9. Lucy Collinson
  10. David Wilkinson
  11. Carmen Pin
  12. Vassilis Pachnis  Is a corresponding author
  13. Tiffany A Heanue  Is a corresponding author
  1. Development and Homeostasis of the Nervous System Laboratory, The Francis Crick Institute, United Kingdom
  2. Bionformatics & Biostatistics Science Technology Platform, The Francis Crick Institute, United Kingdom
  3. Electron Microscopy Science Technology Platform, The Francis Crick Institute, United Kingdom
  4. Neural Development Laboratory, The Francis Crick Institute, United Kingdom
  5. Department of Biology and Biochemistry, University of Bath, United Kingdom
  6. Clinical Pharmacology and Quantitative Pharmacology, Clinical Pharmacology and Safety Sciences, R&D, AstraZeneca, United Kingdom
Research Article
  • Cited 0
  • Views 1,127
  • Annotations
Cite this article as: eLife 2020;9:e56086 doi: 10.7554/eLife.56086

Abstract

The presence and identity of neural progenitors in the enteric nervous system (ENS) of vertebrates is a matter of intense debate. Here, we demonstrate that the non-neuronal ENS cell compartment of teleosts shares molecular and morphological characteristics with mammalian enteric glia but cannot be identified by the expression of canonical glial markers. However, unlike their mammalian counterparts, which are generally quiescent and do not undergo neuronal differentiation during homeostasis, we show that a relatively high proportion of zebrafish enteric glia proliferate under physiological conditions giving rise to progeny that differentiate into enteric neurons. We also provide evidence that, similar to brain neural stem cells, the activation and neuronal differentiation of enteric glia are regulated by Notch signalling. Our experiments reveal remarkable similarities between enteric glia and brain neural stem cells in teleosts and open new possibilities for use of mammalian enteric glia as a potential source of neurons to restore the activity of intestinal neural circuits compromised by injury or disease.

Introduction

Tissue integrity and repair depend on the regulated dynamics of adult stem cells, which share the capacity to replenish cellular compartments depleted by physiological turnover or disease. Studies on neural stem cells (NSCs) have advanced fundamental brain research and opened new and exciting opportunities for regenerative neuroscience (Morales and Mira, 2019). However, as NSC research has focused primarily on the central nervous system (CNS), our understanding of the homeostasis and regenerative potential of peripheral neural networks, and particularly the enteric nervous system (ENS), is minimal and at best phenomenological. This gap in knowledge impedes progress in fundamental gastrointestinal biology and stymies the development of potential therapeutic strategies for repairing intestinal neural circuits with congenital deficits or damaged by injury or disease.

The ENS encompasses the intrinsic neuroglia networks of the gastrointestinal (GI) tract that are essential for digestive function and gut homeostasis (Furness, 2006). In vertebrates, assembly of the ENS begins during embryogenesis with invasion of the foregut by a small founder population of neural crest (NC) cells that proliferate and colonise the entire GI tract, generating diverse types of enteric neurons and glial cells organised into networks of interconnected ganglia (Heanue and Pachnis, 2007). ENS development depends on the integrated activity of NC cell lineage-intrinsic programmes and signals from surrounding non-neuroectodermal gut tissues, which ultimately determine the organisation and physiological properties of intestinal neuroglial networks (Avetisyan et al., 2015; Rao and Gershon, 2018). Despite considerable progress in understanding the developmental mechanisms underpinning the assembly of intestinal neural circuits, much less is known about the dynamics of ENS cell lineages in adult animals, during homeostasis or in response to gut pathology. The predominant view holds that the vast majority of enteric neurons in the mammalian ENS are born during embryogenesis and early postnatal stages and remain functionally integrated into the intestinal circuitry throughout life (Bergner et al., 2014; Joseph et al., 2011; Laranjeira et al., 2011; Pham et al., 1991). Likewise, enteric glial cells (EGCs) are generally quiescent, with only a small fraction proliferating at any given time (Joseph et al., 2011; Kabouridis et al., 2015). Despite this static view of the ENS at homeostasis, lineage tracing experiments in mice have provided evidence that under experimental conditions, such as chemical injury of the ganglionic plexus and bacterial infection, a small fraction of Sox10+ and Sox2+ EGCs can differentiate into neurons (Belkind-Gerson et al., 2017; Belkind-Gerson et al., 2015; Laranjeira et al., 2011). However, a recent study has argued that a population of Sox10-Nestin+ ENS cells undergo extensive proliferation and neuronal differentiation even under physiological conditions, replenishing enteric neurons continuously lost to apoptosis (Kulkarni et al., 2017). Although fundamental tenets of this proposition are not supported by available experimental evidence (Joseph et al., 2011; Laranjeira et al., 2011; White et al., 2018), it highlights critical but unresolved questions regarding the cellular and molecular mechanisms underpinning the maintenance and regenerative potential of the ENS in vertebrates.

To address these questions, we investigated the ENS of zebrafish, an excellent model organism for studies on NSCs and neural regeneration in vertebrates. Using genetic lineage tracing, gene expression profiling, correlative light and electron microscopy (CLEM), live imaging, and computational modelling, we demonstrate that the non-neuronal compartment of the zebrafish ENS expresses the transgenic reporter Tg(her4.3:EGFP) and shares properties with mammalian EGCs and brain NSCs. Tg(her4.3:EGFP)+ ENS cells exhibit morphological features and express genes characteristic of mammalian enteric glia, but canonical glial markers are undetectable. More akin to functional properties of radial glial cells (RGCs) of the zebrafish brain, EGFP+ ENS cells proliferate and undergo constitutive neuronal differentiation which is under the control of Notch signalling. Together, our studies demonstrate the in vivo neurogenic potential of enteric glia in vertebrates and reveal previously unanticipated similarities to NSCs in the brain.

Results

Expression of canonical glial markers is undetectable in the zebrafish ENS

To pave the way for a systematic search for cells harbouring neurogenic potential in the ENS of non-amniotic vertebrates, we first set out to characterise the non-neuronal compartment of the zebrafish ENS, the most likely source of enteric neural progenitors. Initially, we combined the SAGFF234A Gal4 transcriptional activator gene trap with the UAS:GFP transgene in order to generate SAGFF234A;UAS:GFP animals, in which ENS progenitors and enteric neurons were labelled with GFP (Heanue et al., 2016a; Kawakami et al., 2010). In 7 day post fertilisation (dpf) larvae the majority of GFP+ cells (93.76% ± 2.99) co-expressed the pan-neuronal marker HuC/D (Figure 1—figure supplement 1A,D), suggesting that in comparison to mammals, in which EGCs outnumber enteric neurons (Gabella, 1981; Rühl, 2005), the non-neuronal ENS cell population of zebrafish is considerably smaller. To support this supposition, we also quantified the proportion of neurons within the ENS of Tg(−4725sox10:Cre;βactin-LoxP-STOP-LoxP-hmgb1-mCherry) transgenic fish (hereafter abbreviated as Tg(sox10:Cre;Cherry)) in which sox10-driven Cre recombinase activates a nuclear Cherry reporter in early NC cells and all derivative lineages, including the ENS (Rodrigues et al., 2012; Wang et al., 2011b). Although less efficient than a previously published sox10Cre/reporter combination, Tg(sox10:Cre;ef1a:loxP-GFP-loxP-DsRed2) (Rodrigues et al., 2012; Figure 1—figure supplement 1C), Tg(sox10:Cre;Cherry) labels equivalent proportions of neurons and non-neuronal cells (Figure 1—figure supplement 1D). Consistent with the analysis of SAGFF234A;UAS:GFP animals, the majority of Cherry+ cells (84.79 ± 7.70%) in the gut of 7 dpf Tg(sox10:Cre;Cherry) larvae were positive for HuC/D (Figure 1A,C). Similar analysis in adult (≥3 months old) Tg(sox10:Cre;Cherry) zebrafish showed that, although the fraction of non-neuronal Cherry+ cells was higher relative to 7 dpf larvae, even at this stage the majority of ENS+ cells (65.49 ± 4.8%) were neurons (Figure 1B,C). Therefore, the non-neuronal compartment in the zebrafish ENS is notably smaller relative to its mammalian counterpart.

Figure 1 with 1 supplement see all
The non-neuronal compartment of the zebrafish ENS is relatively small and is not identified using canonical glial markers.

(A) Confocal images of the gut of 7 dpf Tg(sox10:Cre;Cherry) larvae immunostained for Cherry (red, top) and HuC/D (cyan, middle) (n = 13). The bottom panel is a merge of the Cherry and HuC/D signals. Inset shows a high magnification of the boxed area. Arrows point to Cherry+HuC/D+ cells and an arrowhead points to a Cherry+HuC/D- cell. Dotted line delineates the gut. Open arrowhead indicates a Cherry+ NC-derived melanocyte (M), which is present outside the intestine. (B) Confocal images of the ENS in adult zebrafish intestine immunostained for Cherry (red, top) and HuC/D (cyan, middle) (n = 13). The bottom panel is a merge of the Cherry and HuC/D signals. Inset shows a high magnification of the boxed area. Arrowheads point to Cherry+HuC/D+ cells and arrows point to Cherry+HuC/D- cells. (C) Quantification of the neuronal (Cherry+HuC/D+) and non-neuronal (Cherry+HuC/D-) cellular compartments within the sox10-lineage at 7 dpf and adult zebrafish, n = 13 biological replicates, data are given as mean ± SD. (D) Confocal images of the gut of 7 dpf zebrafish larvae immunostained for S100β (green) and HuC/D (red). No S100β signal was detected in the ENS, despite abundant neurons throughout the intestine (n = 30). (E) Confocal images of the gut of 7 dpf Tg(gfap:GFP) larvae immunostained for GFP (green) and HuC/D (red). No GFP signal was visible within the intestine despite abundant HuC/D+ neurons (n = 50). GFP+ fibres associated with spinal nerves are observed descending towards the gut but never enter the intestine (open arrowheads). Dotted lines in D and E delineate the gut. (F) Immunostaining of the ENS of adult zebrafish with S100β (green) and HuC/D (red) (n = 5). (G) Immunostaining of the ENS of adult Tg(gfap:GFP) zebrafish with GFP (green) and HuC/D (red) (n = 13). S100β (F) and GFP (G) signal was absent despite the presence of HuC/D+ neurons. All confocal images are max projections of short confocal stacks. 50 µm scale bars shown in merge panels.

All non-neuronal cells of the mammalian ENS are identified as enteric glia expressing combinations of the canonical glial markers S100β, GFAP and BFABP (Hao et al., 2016; Young et al., 2003). To determine whether these marker proteins are also expressed in the zebrafish ENS, we used antibodies raised against them to immunostain 7 dpf larvae, a stage when organised intestinal motility patterns controlled by gut-intrinsic neural networks are clearly evident (Heanue et al., 2016a; Holmberg et al., 2007; Kuhlman and Eisen, 2007). Surprisingly, no signal was detected in the ENS of zebrafish at this stage (Figure 1D and Figure 1—figure supplement 1E–F). Immunostaining signal detected with two antibodies specific for zebrafish GFAP (Baker et al., 2019; Trevarrow et al., 1990) was likely to represent cross-reactivity with non-neuroectodermal gut tissues, as it persisted in ret mutant larvae, which lack enteric neuroglia networks (Figure 1—figure supplement 1G–J; Heanue et al., 2016a). Immunostaining signal for GFAP has previously been reported in the ENS (Baker et al., 2019; Kelsh and Eisen, 2000), however in our experiments the expression is not apparently within the NC-derived lineages. Consistent with the immunostaining, expression of the Tg(gfap:GFP) transgene (Bernardos and Raymond, 2006) was also undetectable in the gut of 7 dpf larvae (Figure 1E). In contrast to the ENS, these immunostaining and transgenic reagents identified the expected signal in the spinal cord (Figure 1—figure supplement 1K–P). To ascertain that the lack of glia marker expression was not due to delayed maturation of enteric glia, we also immunostained adult zebrafish gut for GFAP, S100β, BFABP and (in the case of gfap:GFP transgenics) GFP. Similar to 7 dpf animals, no apparent ENS-specific expression of these markers or the gfap:GFP transgene was detected in the adult gut (Figure 1F–G, Figure 1—figure supplement 1Q–R). Finally, contrary to reports indicating expression of Nestin in non-neuronal cells of mammalian enteric ganglia (Kulkarni et al., 2017), no expression of the nestin:GFP transgene was detected in the ENS of adult zebrafish (Figure 1—figure supplement 1S). Taken together, our studies demonstrate that the non-neuronal compartment of the zebrafish ENS is considerably smaller relative to its mammalian counterpart and cannot be labelled by immunohistochemical reagents commonly used for the identification of enteric glia.

Non-neuronal cells of the zebrafish ENS share with mammalian EGCs early NC cell and ENS progenitor markers

To explore further the gene expression profile of the non-neuronal ENS cell compartment in zebrafish, we carried out bulk RNA sequencing of fluorescent-labelled nuclei (nRNAseq) isolated from Tg(sox10:Cre;Cherry) adult gut muscularis externa. This strategy, which we described recently (Obata et al., 2020), avoids lengthy protocols of tissue dissociation and cell isolation that are often associated with considerable cell damage. Since the available transgenic tools did not allow us to label specifically the non-neuronal ENS cell compartment, bulk nRNAseq was performed on nuclei purified by FACS (fluorescent-activated cell sorting) representing both the Cherry+ (entire ENS) and Cherry- (non-ENS) muscularis externa cell populations of Tg(sox10:Cre;Cherry) zebrafish gut (Figure 2A and Figure 2—figure supplement 1A; see also Materials and Methods). Principal component analysis (PCA) demonstrated a clear separation of the Cherry+ and Cherry- nuclear transcriptomes along PC1 (Figure 2—figure supplement 1B), indicating that variability along this axis is determined predominantly by the lineage origin (NC vs non-NC) of the two cell populations. As expected, genes associated with non-NC tissues, such as smooth muscle cells (myh11a, cald1a, srfa, gata6), interstitial cells of Cajal (ano1, kita, kitb) and immune cells (lcp1, lck, lyz), were upregulated in the Cherry- nuclear transcriptome (Figure 2B). Conversely, genes associated with the NC-derived ENS lineages (such as elavl3, elavl4, ret, vip, chata, sox10) were upregulated in the Cherry+ nuclear population (Figure 2B) (GEO database GSE145885; Supplementary file 1; an interactive data viewer to explore the analysed data can be found here: https://biologic.crick.ac.uk/ENS). Furthermore, gene ontology (GO) terms enriched in the Cherry+ nuclear population were associated with nervous system development and function (Figure 2—figure supplement 1C–E). Finally, direct comparison of the Cherry+ dataset to the transcriptional profile of enteric neurons from 7 dpf larvae expressing the Tg(phox2b:EGFP)w37 transgene (Roy-Carson et al., 2017), identified a large cohort of shared genes (including phox2bb, ret, elavl3, elavl4, vip, nmu) that presumably reflect the neural component of the mixed Cherry+ nuclear population (Figure 2C, yellow dots, Figure 2—figure supplement 1F and Supplementary file 2).

Figure 2 with 3 supplements see all
Transcriptomic profiling of the adult zebrafish ENS.

(A) Experimental strategy for the isolation of ENS nuclei from adult Tg(sox10:Cre;Cherry) guts and nuclear RNAseq. Five biological replicates were performed per condition. (B) Volcano plot shows mean log2 fold-change (x axis) and significance (-log10 adjusted p-value) (y axis) of genes differentially expressed in Cherry+ relative to Cherry- nuclei. Genes characteristic of the ENS are highlighted in red and are more abundant in Cherry+ nuclei, whereas genes characteristic of non-neuroectodermal lineages, such as smooth muscle (purple), interstitial cells of Cajal (green) and immune associated (blue), are more abundant in Cherry- nuclei. (C) Volcano plot (as in B) in which genes previously identified in a transcriptional characterization of larval ENS neurons (Roy-Carson et al., 2017) are shown in yellow. These include established neuronal markers, such as phox2bb, ret, elavl3, elavl4, vip, and nmu. Genes enriched in the Cherry+ nuclear population but absent from the larval ENS neuron transcriptome are shown in black. These include sox10, foxd3, sox2, plp1, the mammalian orthologues of which are expressed by mouse EGCs, tfap2a, a gene required for early NC development, col28a1b, whose mammalian orthologue is a peripheral glial marker, as well as ptprz1a, and ptprz1b, which have been identified in glioblastoma stem cells. Genes with padj <0.05 (Log10p-value<1.3) and/or log2FC < 0 are shown in grey. (D,E) Confocal images of fluorescent in situ hybridization (RNAscope) using probes for sox10 (D) and foxd3 (E) on adult Tg(sox10:Cre;Cherry) gut muscularis externa preparations immunostained for Cherry (ENS lineage) and HuC/D (ENS neurons). Signal for both sox10 and foxd3 (white arrows) corresponds to non-neuronal cells (Cherry+HuC/D-, arrows) but was absent from enteric neurons (Cherry+HuC/D+, arrowheads). (F) Immunostaining of adult Tg(sox10:Cre;Cherry) gut for Sox2 (blue), Cherry (red) and HuC/D (green). Sox2 is expressed specifically by non-neuronal ENS cells. Biological replicates: D, n = 4; E, n = 6; F, n = 5. All confocal images are max projections of short confocal stacks. 10 µm scale bars shown in merge panels.

To identify genes expressed by the non-neuronal compartment of the zebrafish ENS, we next compared the Cherry+ dataset to a recently reported transcriptome of mouse EGCs, which presented a list of the 25 most highly expressed genes in PLP1+ enteric glia (Rao et al., 2015). Zebrafish orthologues for several genes in this list were enriched in the Cherry+ transcriptome (Figure 2—figure supplement 1G), suggesting that they are expressed by the non-neuronal cells of the zebrafish ENS. Among these genes were sox10, foxd3 and plp1, which in mammals are expressed by early NC cells and ENS progenitors and maintained in enteric glia (Dyachuk et al., 2014; Hari et al., 2012; Mundell and Labosky, 2011; Mundell et al., 2012; Weider and Wegner, 2017), as well as genes with established association to glial cells, such as col28a1 (Grimal et al., 2010), ptprz1a and ptprz1b (Fujikawa et al., 2017). In a similar strategy, we have also compared the Cherry+ datatset to a single cell transcriptomic dataset of mouse ENS neurons and glia (Zeisel et al., 2018). We have identified the genes from this mouse dataset that are differentially expressed between mouse ENS glia and neurons and determined their zebrafish orthologues (Supplementary file 3). We show that 366 mouse ENS neuron-enriched genes have orthologues present in our zebrafish Cherry+ transcriptome dataset, including elavl3, elavl4, prph, and phox2bb, and likely reflect the neuronal component of our bulk dataset (Figure 2—figure supplement 2A,B, Supplementary file 4). We also show that 63 mouse ENS glia-enriched genes have orthologues present in the zebrafish Cherry+ dataset, suggesting that these glial expressed genes are detected in the non-neuronal component of the zebrafish ENS (Figure 2—figure supplement 2A,C, Supplementary file 5), including sox10, foxd3, plp1b, and the additional neural crest marker zeb2b (Delalande et al., 2008), sox2, which is expressed by mouse ENS progenitors and adult EGCs (Belkind-Gerson et al., 2017; Heanue and Pachnis, 2011), and the CNS glia associated gene vim (Deng et al., 2013). Significantly, we do not observe canonical glial markers gfap, s100b and fabp7a/b amongst these genes, consistent with the failure to detect expression of these markers by immunostaining analysis (Figure 1F,G and Figure 1—figure supplement 1Q,R), though we cannot exclude the possibility that such markers may be revealed by in depth sequencing of single cells. In a final strategy to identify genes associated with the non-neuronal component of the zebrafish ENS, we applied a strategy that was not reliant on cross-species comparisons. Having delineated the neural component of the Cherry+ transcriptome (Figure 2C, yellow dots, and Figure 2—figure supplement 1F), we removed this cohort of genes in order to enrich for transcripts of the non-neuronal ENS cell compartment (Figure 2C, black dots, Figure 2—figure supplement 1H and Suppl. File 6). This strategy highlighted several genes that were identified by our previous analysis, including sox10 and foxd3. Numerous additional genes were identified, including tfap2a, a gene required in early NC cells (Knight et al., 2003; Wang et al., 2011a), and sox2. Expression of sox10, foxd3 and sox2 in the non-neuronal compartment of the zebrafish ENS was validated by combining multiplex fluorescence in situ hybridisation (RNAscope) with immunostaining for HuC/D and the Cherry reporter on muscularis externa preparations from the gut of adult Tg(sox10:Cre;Cherry) zebrafish (Figure 2D–F). Together, these experiments indicate that, despite our failure to detect expression of commonly used EGC markers, the transcriptomes of the non-neuronal compartment of the zebrafish ENS and mammalian enteric glia have considerable overlap, including genes associated with early NC cells and ENS progenitors.

Non-neuronal cells in the adult zebrafish ENS express the Notch activity reporter Tg(her4.3:EGFP)

In mammals, Notch signalling promotes enteric gliogenesis by attenuating a cell-autonomous neurogenic programme of ENS progenitors (Okamura and Saga, 2008), but the expression of Notch target genes in adult EGCs is unclear. Moreover, the transgenic Notch activity reporter Tg(her4.3:EGFP) (see Materials and Methods for the nomenclature of this transgene) marks NSCs and neural progenitors in the zebrafish brain (Alunni and Bally-Cuif, 2016; Yeo et al., 2007). Given the fact that the non-neuronal component of the ENS appears to be enriched for progenitor markers, and our desire to find a suitable transgenic tool to facilitate further study, we have examined whether the Notch activity reporter Tg(her4.3:EGFP) is also observed in non-neuronal cells of the zebrafish ENS. We examined the adult gut for expression of Tg(her4.3:EGFP) expressing cells. This analysis identified a network of GFP+ cells in the muscularis externa of the gut that was closely associated with enteric neurons and their projections (Figure 3A and Figure 3—figure supplement 1A). To provide direct evidence that Tg(her4.3:EGFP) expressing cells are integral to the ENS, we introduced the her4.3:EGFP transgene into the Tg(sox10:Cre;Cherry) genetic background and immunostained gut preparations from adult Tg(her4.3:EGFP;sox10:Cre;Cherry) animals for GFP, HuC/D and Cherry. As expected, GFP+ cells were negative for HuC/D but expressed the Cherry reporter (Figure 3C), indicating that they belong to the non-neuronal compartment of the ENS. Consistent with this idea, GFP+ cells co-expressed sox2 and sox10 (Figure 3D,E), which were identified by our transcriptomic analysis as genes expressed by the non-neuronal compartment of the zebrafish ENS. We observed heterogeneity within the non-neuronal component: whereas Sox2 was widely expressed in the GFP+HuC/D- cell population in Tg(her4.3:EGFP;sox10:Cre;Cherry) tissue, foxd3 was expressed in only a proportion of the GFP+HuC/D-/Sox2+ cells (Figure 3F). The GFP+HuC/D- cell population in Tg(her4.3:EGFP;sox10:Cre;Cherry) represented approximately a quarter (24.20 ± 5.18%) of all Cherry+ ENS cells, but 12.93 ± 5.33% of Cherry+ cells were negative for both GFP and HuC/D (Cherry+GFP-HuC/D-) (Figure 3B). Therefore, the majority of non-neuronal ENS cells in adult zebrafish gut can be identified by the expression of the Notch activity reporter Tg(her4.3:EGFP).

Figure 3 with 1 supplement see all
The her4.3:EGFP transgene is a novel marker of the non-neuronal cell population in the adult zebrafish ENS.

(A) Confocal images of adult Tg(her4.3:EGFP) zebrafish gut immunostained for GFP (green) and HuC/D (red). Inset is a high magnification of boxed area showing that GFP+ cells (arrow) are closely associated with HuC/D+ neurons (arrowhead) (n = 70). (B) Quantification of neuronal (Cherry+ HuC/D+GFP-, blue) and non-neuronal cell populations (Cherry+HuC/D-GFP+ and Cherry+HuC/D-GFP-, green and red, respectively) in the ENS of adult Tg(her4.3:EGFP;sox10:Cre;Cherry) zebrafish (n = 3). Data are given as mean ± SD. (C) Confocal images of the ENS from adult Tg(her4.3:EGFP;sox10:Cre;Cherry) zebrafish immunostained for Cherry (red), GFP (green) and HuC/D (cyan). Note the presence of Cherry+HuC/D-GFP+ (arrows) and Cherry+ HuC/D- GFP- (grey arrowheads) cells as well as the presence of Cherry+HuC/D+GFP- neurons (white arrowheads) (n = 3). Note that Cherry+ nuclei are of equivalent size in Cherry+HuC/D-GFP+ (arrows), Cherry+ HuC/D- GFP- (grey arrowheads) cells, and Cherry+HuC/D+GFP- neurons (white arrowheads). (D) Immunostaining of adult Tg(her4.3:EGFP;sox10:Cre;Cherry) gut with antibodies for Cherry (red), GFP (green) and Sox2 (blue). Arrows point to cells expressing all three markers (n = 3). (E) RNAscope analysis for ret (red) and sox10 (white) on ENS preparations from adult Tg(her4.3:EGFP) zebrafish guts immunostained for GFP (green). Note that GFP+ cells (arrows) express sox10 and are found in close proximity to ret+GFP- enteric neurons (grey arrowheads) (n = 4). (F) Combined RNA scope for foxd3 and immunostaining for GFP, Cherry and Sox2 on adult Tg(her4.3:EGFP;sox10:Cre;Cherry) gut shows that foxd3 and Sox2 are co-expressed in some ENS cells (white arrows), other GFP+ cells express only Sox2 (grey arrows). All confocal images are max projections of short confocal stacks. Scale bars in merge panels: (A) 50 µm (C–E) 10 µm.

GFP+ cells in the ENS of adult Tg(her4.3:EGFP) zebrafish have morphological characteristics of mammalian EGCs

To provide evidence that Tg(her4.3:EGFP) expressing cells in the zebrafish ENS are equivalent to mammalian EGCs, we characterised the morphology of GFP+ cells in the gut of Tg(her4.3:EGFP) transgenics. At the light microscopy level GFP+ cells were highly branched and fell into four morphological groups that generally corresponded to the four morphological subtypes of mouse EGCs (Types I-IV) (Figure 3—figure supplement 1B–E; Boesmans et al., 2015; Gulbransen and Sharkey, 2012). Although the zebrafish ENS lacks distinct ganglia found in mammalian systems, the presence of GFP+ cells in the myenteric layer, in close association with HuC/D+ cells, with multiple processes wrapping around the HuC/D+ cell bodies, is reminiscent of Type I mammalian EGCs (Figure 3—figure supplement 1B; Hanani and Reichenbach, 1994; Boesmans et al., 2015). Moreover, elongated GFP+ cells in the myenteric layer with cell bodies and processes that follow along AcTu+ neuronal cell processes show clear parallels with Type II mammalian EGCs (Figure 3—figure supplement 1C; Hanani and Reichenbach, 1994; Boesmans et al., 2015). GFP+ cells were also found within the mucosa in close proximity to the intestinal epithelium (Figure 3—figure supplement 1D), similar to Type III mucosal EGCs located within the lamina propria of the mammalian gut (Boesmans et al., 2015; Kabouridis et al., 2015). And finally, bipolar GFP+ cells found within the smooth muscle layers and associated with AcTu+ neuronal fibres are reminiscent of Type IV mammalian glia (Figure 3—figure supplement 1E; Boesmans et al., 2015).

Mammalian EGCs have unique ultrastructural features and establish characteristic contacts with enteric neurons and their projections (Gabella, 1972; Gabella, 1981). To determine whether similar features are exhibited by the GFP+ ENS cell population in Tg(her4.3:EGFP) zebrafish, we analysed EGFP+ cells in Tg(her4.3:EGFP;SAGFF217B;UAS:mmCherry) transgenics using CLEM (Müller-Reichert and Verkade, 2012). In these animals, EGFP marks non-neuronal ENS cells while Cherry, which is driven by the binary reporter Tg(SAGFF217B;UAS:mmCherry) (Kawakami et al., 2010), labels a subset of enteric neurons (Figure 4—figure supplement 1A). CLEM confirmed the close association of EGFP+ cells with enteric neurons and their projections (Figure 4, Figure 4—figure supplement 1B,C and Video 1). Processes emanating from EGFP+ cells could extend to 18 µm and directly contacted enteric neurons (Figure 4B,D and Figure 4—figure supplement 1C), but similar to mammalian EGCs (Gabella, 1981) they did not form complete ‘capsules’ around neuronal somata, allowing large parts of enteric neurons to be in direct contact with adjacent cells (Figure 4A,B, Figure 4—figure supplement 1C and Video 1). EGFP+ cells also extended complex sheet-like extensions, which frequently enclosed and/or subdivided the tightly packed bundles of neural projections into sectors (Figure 4D, Figure 4—figure supplement 1C and Video 1). Scarcity of cytoplasm and deep nuclear crenations, characteristic features of mammalian EGCs and other populations of peripheral glial cells (Gabella, 1981), were also found in the nuclei of EGFP+ cells (Figure 4B,D and Figure 4—figure supplement 1C).

Figure 4 with 1 supplement see all
her4.3:EGFP expressing cells in the adult zebrafish ENS share with mammalian enteric glia characteristic ultrastructural features.

(A and C) Electron micrographs (z-stack # 903 in A and #1039 in C) from a 3D region of interest from the midgut of adult Tg(her4.3:EGFP;SAGFF217;UAS:mmCherry) zebrafish. Insets shows super-resolution light microscopy images of EGFP+ non-neuronal cells and mmCherry+ neurons that correspond to the boxed areas of the electron micrograph. The EGFP+ cells have a cell soma size of ~79.6 µm3 (A) and ~79.1 µm3 (C) with projection lengths that range from sheet-like processes of 4 µm to longer extensions of up to 18 µm. For comparison, the mmCherry+ neurons have cell soma size of ~398.8 µm3 (A, left) and 229.7 µm3 (A, right) with projection lengths that range from 16 µm to 55 µm. (B and D) High-resolution images of the boxed areas shown in A (B) and C (D). The EGFP+ cells are pseudocoloured in green and enteric neurons in red. Black arrowheads indicate points of contact between EGFP+ processes and mmCherry+ neurons. Yellow arrowheads indicate GFP+ sheet-like extensions that compartmentalise axon bundles (white asterisks). Nuclear crenelations in nuclei of EGFP+ cells are indicated with black arrows. Representative images of six regions of interest scanned from two adults. All images are a single z plane. Scale bars: 10 µm (A, C and insets A,C) and 1 µm (B,D).

Video 1
Correlative light and electron microscopy (CLEM) analysis of the adult Tg(her4.3:EGFP;SAGFF234A;UASmmCherry) gut.

Mapping of the super-resolution light microscopy volume into the cropped SBF SEM volume using Bigwarp confirmed the identification and localisation of EGFP+ non-neuronal cells and mmCherry+ neurons within a 3D region of interest from the midgut of Tg(her4.3:EGFP;SAGFF217;UAS:mmCherry) zebrafish. The EGFP+ cells and mmCherry+ neurons that were false coloured in Figure 4 and Figure 4—figure supplement 1 are indicated with green and red arrows, respectively, showing that each forms numerous complex extensions through the volume. Data are shown at 10 frames per second, with 100 nm pixels in XY (cropped to represent a horizontal frame width of 80.5 µm) and 50 nm pixels in Z (representing a depth of 64.8 µm).

Taken together we show that like mammalian EGCs, the her4.3:EGFP+ population share a lineage with ENS neurons, are found within the myenteric layer in close association with ENS neurons, have distinctive morphologies reminiscent of the four types of mammalian EGCs, have ultrastructural features of mammalian ENS glia, and express multiple well-established mammalian EGC markers. This weight of evidence leads us to conclude that her4.3:EGFP+ cells constitute the zebrafish EGC population, and therefore now define them as such. Henceforth, we will be referring to Tg(her4.3:EGFP) expressing cells in the adult zebrafish ENS as EGCs.

Developmental profile of zebrafish EGCs

To examine the developmental profile of zebrafish EGCs, we immunostained Tg(her4.3:EGFP;SAGFF234A;UAS:mmCherry) transgenics for GFP and Cherry at different developmental stages. At 54 hr post fertilisation (hpf), a stage at which NC cell-derived Cherry+ cells are restricted to two distinct migratory columns along the gut (Heanue et al., 2016a), no double positive (Cherry+GFP+) cells were identified (Figure 5—figure supplement 1A). However, at 60 hpf a small number of GFP+ cells were discernible within the Cherry+ streams of NC cells (Figure 5—figure supplement 1B) and became more abundant in 4 dpf larvae (Figure 5—figure supplement 1C). To further examine the developmental dynamics of the GFP+ cell lineage, we performed time-lapse confocal microscopy of live Tg(her4.3:EGFP;SAGFF234A;UAS:mmcherry) embryos at similar stages. Imaging commenced at 56 hpf with the migratory front of mmCherry+ NC cell columns positioned at the rostral end of the field of view (Heanue et al., 2016a) and continued for 40 hr (1 image every 10 min). Consistent with the analysis performed on fixed embryos, no EGFP+ cells were identified within the mmCherry+ population during the first hours of imaging (Figure 5A). However, EGFP+ cells appeared within the columns of mmCherry+ cells at around 62 hpf (Figure 5B), more than 90 µm behind the front of migrating mmCherry+ NC cells, and the number of EGFP+ cells increased over the remaining imaging period (Figure 5C,D; Video 2). On several occasions, we identified individual mmCherry+ cells inducing de novo expression of EGFP (Video 3). EGFP+ cells emerged in a rostro-caudal sequence mirroring the wave of ENS neuron maturation (Heanue et al., 2016b) but they were almost always located behind the front of migrating enteric NC cells. Relative to the tip of the mmCherry+ migratory column, which was displaced caudally at a constant rate until it reached the caudal end of the FOV, EGFP+ cells on average exhibited minimal rostrocaudal displacement (Figure 5E; 132 EGFP+ cells analysed from four fish), suggesting that during ENS development the her4.3:EGFP transgene is expressed in post-migratory cells.

Figure 5 with 1 supplement see all
Live imaging of Tg(her4.3:EGFP)+ cell ontogenesis in the developing zebrafish ENS.

(A–D) Still images from time-lapse recording of a Tg(her4.3:EGFP;SAGFF234A;UAS:mmCherry) embryo imaged from 56 hpf (denoted as 00:00) until 96 hpf (40:00), a representative example of n = 18 biological replicates. At 00:00 (A) the mmCherry+ wavefront of NC cells (red, red arrowhead) is at the rostral side of the field of view (FOV) and no EGFP+ cells (grey) are present. At 05:30 (B), the first EGFP+ cells (grey, arrow) appear within the mmCherry+ NC cell column (red), behind the migratory wavefront. Bright GFP+ melanocytes are designated (grey arrowheads). (C) At 19:50 the NC cell column extends throughout the FOV and the number of EGFP+ cells (grey, arrows) has increased. White arrowhead points to an EGFP+ cell exhibiting a rounded morphology, which can be seen to divide in subsequent time lapse images. An increasing number of bright GFP+ melanocytes appear (grey arrowheads), and are relatively static in the time lapse recordings. (D) At the end of the recording (40:00), EGFP+ cells (grey) can be found throughout the gut (white arrowheads). Abundant brightly GFP+ melanocytes are present in the gut region (grey arrowheads), whose characteristic morphology is apparent. (E) Quantification of cell displacement (normalised distance from reference point/time) of the mmCherry+ wavefront (red) and EGFP+ cells (green), data describing 132 cells from four fish. Data are given as mean ± SD. All confocal images are max projections of short confocal stacks. 50 µm scale bar in A.

Video 2
Representative time-lapse image from a Tg(her4.3:EGFP;SAGFF234A;UASmmCherry) embryo.

Time-lapse imaging revealed that Tg(her4.3:EGFP)+ cells (grey, white arrowheads) are found within the mmCherry+neural crest cells (red) that are colonising the developing gut, but the EGFP+ cells appear behind the wavefront of migration (red arrowheads). Time given is shown as hh:mm from the start of recording. See methods for details.

Video 3
Representative recording of de novo EGFP expression in time-lapse recording from Tg(her4.3:EGFP;SAGFF234A;UASmmCherry) embryos.

De-novo her4.3:EGFP transgene expression (grey) within the enteric nervous system (red) is observed during time lapse recordings of developing Tg(her4.1:EGFP;SAGFF234A;UAS:mCherry) embryos (arrow).

Next, we characterised the cell division patterns of the 79 EGFP+ cells that migrated into the field of view or arose de novo during the live imaging period. Of these, 37 cells gave rise to at least one generation of GFP+ progeny. 26 cells (~33%) underwent a single cell division generating two daughters, many of which lost EGFP expression over the course of imaging. In these cases the EGFP expressing cells were not migratory and the EGFP expression diminished and then extinguished. In a proportion of cells (8 cells; ~10%), after a first division event, one or both of the daughter cells underwent a further cell division, generating EGFP+ granddaughters, some of which lost expression of the reporter. For 3 cells (~4%), following two division events, one granddaughter cell underwent a further division to generate a third generation of EGFP+ progeny. Altogether, 53 EGFP+ cells were seen to undergo a cell division event during the imaging period. Therefore, during development Tg(her4.3:EGFP) expressing cells are capable of entering the cell cycle but those that do so undergo only a limited number of cell divisions and many of their progeny eventually lose expression of EGFP. Loss of EGFP signal could be associated with neuronal differentiation, since we occasionally identified in the gut of 7 dpf her4.3:EGFP transgenic larvae cells that were weakly immunostained for both HuC/D and GFP (Figure 5—figure supplement 1D). Taken together, our analysis of Tg(her4.3:EGFP) expression during zebrafish development suggests that nascent EGCs are postmigratory NC-derived cells which maintain proliferative and neurogenic potential.

Proliferation and neuronal differentiation of adult zebrafish EGCs during homeostasis

Enteric glia in adult mammals are generally quiescent with only a small fraction of cells undergoing cell division at any given time (Joseph et al., 2011). To examine the proliferative potential of EGCs in adult zebrafish, we immunostained whole-mount gut preparations from adult Tg(her4.3:EGFP) transgenics for the proliferation marker mini-chromosome maintenance 5 (MCM5) (Ryu et al., 2005). 10.8 ± 4.2% of GFP+ cells were positive for MCM5 (Figure 6—figure supplement 1), indicating that in contrast to their mammalian counterparts, a considerable proportion of zebrafish EGCs are cycling during homeostasis.

Our earlier observation that EGFP expressing cells in the ENS of Tg(her4.3:EGFP) zebrafish embryos undergo only a limited number of cell divisions suggested that EdU (5-ethynyl-2’-deoxyuridine) labelling of EGCs in adult animals could be used to trace the progeny of proliferating cells and determine their fate. Consistent with the MCM5 immunostaining, we found that at the end of a 3 day EdU labelling pulse (t0), 8.0 ± 4.3% of GFP+ cells in the gut of 3 month old her4.3:EGFP transgenic zebrafish were co-labelled with EdU (Figure 6A,B and D). The similar percentage of MCM5+ and EdU+ cells suggests that Tg(her4.3:EGFP) cells represent a largely quiescent cell population and the that dividing cells have long cell cycles.

Figure 6 with 3 supplements see all
Proliferation and neurogenic differentiation of adult her4.3:EGFP+ ENS cells during homeostasis.

(A) Schematic representation of experimental design. Adult Tg(her4.3:EGFP) zebrafish were immersed in 1 mM EdU for three days and analysed at 0 (t0), 4 (t4) or 11 (t11) days after EdU pulse. (B–C) GFP (green) and HuC/D (blue) immunostaining of intestines from EdU (red) pulsed animals harvested at t0 (B) and t4 (C). Arrowheads (in B and C) point to GFP+HuC/D-EdU+ cells. Arrow (in C) indicates a GFP-HuC/D+EdU+ neuron. 10 µm scale bars in B-C merge panels. All confocal images are max projections of short confocal stacks. (D) Quantification of the percentage of GFP+ cells labelled with EdU at t0, t4 and t11 (mean ± SD). (E) Quantification of the percentage of EdU-labelled enteric neurons at t0, t4 and t11, with biological replicates t0 n = 6, t4 n = 5, t11 n = 5 (mean ± SD.) (F) Strategy for computational analyses of the density of EdU-labelled HuC/D+ and EGFP+ cells. EdU+GFP+ cells were positioned at the centre of concentric circles of increasing radius and the density of EdU+GFP+ and EdU+HuC/D+ cells within each circle was calculated. An example of a 40 μm radius circle (yellow) is shown in higher magnification. (G) Recorded (red graph) and simulated (blue graph) densities of EdU+HuC/D+ and EdU+GFP+ cells (y axis) in concentric circles of increasing radius (x axis) around EdU+GFP+ cells. Monte Carlo simulation of random distribution of EdU+HuC/D+ or EdU+GFP+ cells were performed >2000 times for each dataset in order to establish baseline densities arising in randomly mixed populations. Error bars represent mean ±90% confidence intervals. At all time-points analysed, recorded densities of EdU+HuC/D+ and EdU+GFP+ cells were above the confidence interval (bars) of the simulated densities in 20-60 μm circles (indicated by asterisk). (H, I) Quantification of the percentage of EdU-labelled Cherry+HuC/D+ neurons (H) and Cherry+GFP-HuC/D- cells (I) at t0, t4 and t11 in the intestine of her4.3:gfp;sox10:Cre;Cherry transgenics pulse-labelled with EdU according to the protocol shown in panel A, with biological replicates: t0 n = 6; t4, n = 5; t11 n = 6 (mean ± SD). *p<0.05, **p<0.01, ***p<0.001.

In these experiments, the majority of (GFP+EdU+) cells formed doublets composed of cells with similar morphology and GFP signal intensity (Figure 6B and Figure 6—figure supplement 2B). Occasionally, one or both cells in the doublets exhibited reduced GFP signal (Figure 6—figure supplement 2C,D), suggesting that, similar to larval stages, the daughters of dividing EGCs in adult her4.3:EGFP transgenic zebrafish differentiate into GFP- enteric neurons. This idea was supported by the identification 4 days post-labelling (t4 chase) of EdU+ doublets that included GFP+HuC/D- and GFP-HuC/D+ cells (Figure 6A,C). The loss of GFP signal from the daughters of proliferating EGCs cells was also supported by cell population analysis which demonstrated a reduction in the percentage of EdU+GFP+ cells (t4: 3.6 ± 3.4%, p=6.01×10−7; t11: 3.9 ± 3.8%, p=7.61×10−6) (Figure 6D). Interestingly, the reduced percentage of EdU+GFP+ cells during the EdU chase period was associated with a concomitant increase in the representation of EdU+ enteric neurons at t4 (0.71 ± 0.80%, p=6.0×10−7) and t11 (0.70 ± 0.82%, p=1.5×10−6) relative to t0 (0.068 ± 0.13%) (Figure 6E). Together, these experiments suggest that the progeny of proliferating EGCs in the zebrafish ENS can differentiate into neurons under physiological conditions.

To provide further evidence in support of the lineage relationship between GFP+EdU+ cells and newborn enteric neurons (HuC/D+EdU+), we used confocal microscopy and mathematical modelling to estimate the densities of these cell types within circles of increasing radius centred on EdU+ cells (Figure 6F; Tay et al., 2017). We reasoned that closer proximity of HuC/D+EdU+ and GFP+EdU+ cells relative to that expected from random distribution of lineally unrelated cells would indicate origin from common progenitors undergoing cell division. The densities observed at t0, t4 and t11 were compared to values of uniformly distributed cell types generated randomly by Monte Carlo simulations (>2×103 per sampling time). This analysis revealed that the actual densities of GFP+EdU+ and HuC/D+ EdU+ cells were significantly higher within the smaller radius circles (<60 µm from the cell of interest) in comparison to those expected by chance, suggesting that the observed homotypic (GFP+EdU+/GFP+EdU+) and heterotypic (GFP+EdU+/HuC/D+EdU+) clusters of EdU+ ENS cells were descendants of a common proliferating progenitor (Figure 6G). EdU- cells exhibited densities similar to those expected in randomly mixed populations (data not shown). This analysis provides further support to the idea that descendants of proliferating Tg(her4.3:EGFP) expressing ENS cells are capable of undergoing neuronal differentiation in the gut of adult zebrafish.

Next, we considered the possibility that the GFP- non-neuronal ENS cell population (Figure 3B) is also derived from GFP+ progenitors and represents an intermediate stage of neurogenic commitment, in a process analogous to the differentiation of GFP+ RGCs in the pallium of her4.3:EGFP transgenic zebrafish. To examine this, we pulse-labelled 3 month old Tg(her4.3:EGFP;sox10:Cre;Cherry) transgenics with EdU (Figure 6A) and followed the descendants of proliferating EGCs in the context of the entire ENS lineage. Consistent with our previous analysis (Figure 6E), the percentage of enteric neurons labelled by EdU (Cherry+HuC/D+EdU+) at t4 and t11 was higher relative to t0 (t0: 0.021 ± 0.15%; t4: 0.28 ± 1.2%, p=0.06; t11: 0.37 ± 0.95%, p=0.0014) (Figure 6H). Interestingly, this increase was paralleled by an increased percentage of EdU-labelled GFP- non-neuronal ENS cells (Cherry+GFP-HuC/D-EdU+) at t4 and t11, relative to t0 (t0: 0.12 ± 0.5%; t4: 3.7 ± 12.5%, p=1.84×10−6; t11: 4.1 ± 15.5%, p=0.0024) (Figure 6I). Together these studies suggest that loss of Tg(her4.3:EGFP) expression in the daughters of proliferating EGCs is likely to reflect neurogenic commitment preceding overt neuronal differentiation.

Notch signalling regulates the dynamics of EGCs in the gut of adult zebrafish

Inhibition of Notch signalling promotes the proliferation and neurogenic differentiation of Tg(her4.3:EGFP) expressing RGCs in the telencephalon of zebrafish (Alunni et al., 2013; Chapouton et al., 2010). This, together with the observed downregulation of the her4.3:EGFP transgene upon neuronal differentiation of GFP+ cells (Figure 6C), suggested that canonical Notch activity regulates the proliferation and differentiation dynamics of EGCs in zebrafish. To examine this possibility, we blocked Notch signalling in adult zebrafish by treating them with the γ-secretase inhibitor LY411575 (referred to as LY) (Alunni et al., 2013; Rothenaigner et al., 2011) for 7 days. To assess the proliferative and neurogenic response of ENS cells, animals were also exposed to EdU during the last 3 days of LY treatment (Figure 7A). As expected, LY treatment of Tg(her4.3:EGFP) zebrafish resulted in rapid loss of GFP signal from the gut (Figure 7—figure supplement 1). Although this experiment confirmed that Tg(her4.3:EGFP) is a bona fide target of canonical Notch signalling in the ENS, it precluded the use of this transgene as a marker and lineage tracer of the EGC response to LY treatment. Therefore, we applied LY and EdU to Tg(sox10:Cre;Cherry) animals and analysed the entire population of non-neuronal ENS cells at the end of the LY/EdU treatment period (t0). As shown in Figure 7B, Notch inhibition in 3–4 month old Tg(sox10:Cre;Cherry) zebrafish resulted in a dramatic increase in the percentage of non-neuronal ENS cells incorporating EdU (Cherry+HuC/D-EdU+) (control: 0.0387 ± 0.21%; LY: 15.6 ± 17.0%, p=2.67×10−7). A robust proliferative response of non-neuronal ENS cells was also observed in 6 month old Tg(sox10:Cre;Cherry) animals (control: 0.832 ± 1.87%; LY: 6.95 ± 8.2%, p=1.98×10−5) (Figure 7D). Interestingly, LY treatment also resulted in increased enteric neurogenesis (Cherry+HuC/D+EdU+ cells) in both 3 month old (control: 0.0330 ± 0.18%; LY: 2.12 ± 7.8%, p=3.70×10−4) and 6 month old (control: 0.0652 ± 0.22%; LY: 1.56 ± 3.8%, 3.81 × 10−4) animals (Figure 7C,E). It remains unclear whether the apparent increase in neurogenesis following LY treatment indicates a direct role of Notch signaling on neuronal differentiation of EGCs or an indirect consequence of their enhanced proliferation. Irrespective of the exact mechanisms, our experiments demonstrate that, similar to pallial RGCs (Alunni et al., 2013), Notch signalling regulates the dynamics of EGCs in the vertebrate gut throughout life.

Figure 7 with 1 supplement see all
Notch signalling regulates the activation and differentiation of adult zebrafish EGCs.

(A) Schematic representation of experimental protocol for LY/EdU treatment of adult zebrafish. (B–E) Quantification of the effect of Notch inhibition on the proliferation (B and D) and neuronal differentiation (C and E) of EGCs in 3–4 month old (B and C) and 6–7 month old (D and E) animals. N = 4 biological replicates per condition. Data are given as mean ± SD. ***p<0.001.

Discussion

Here, we characterise the non-neuronal compartment of the zebrafish ENS and identify both familiar and unexpected properties of EGCs in teleosts. Specifically, we demonstrate that markers commonly used for the identification of peripheral glial cells in higher vertebrates are not detected in zebrafish EGCs, but that EGCs share morphological features and gene expression programmes with their mammalian counterparts. However, in contrast to mammalian enteric glia, but in accordance with the properties of brain RGCs, the population of zebrafish EGCs is dynamic, undergoing self-renewing proliferation and neuronal differentiation during homeostasis, which are regulated by Notch signalling. Our findings highlight the neural precursor potential of vertebrate enteric glia in vivo and reveal previously unanticipated similarities to brain NSCs.

Earlier histological studies demonstrated that mammalian enteric glia are remarkably similar to protoplasmic astrocytes and express the intermediate filament GFAP, a characteristic astrocytic marker (Jessen and Mirsky, 1980; Rühl, 2005). Further EM analysis revealed diagnostic ultrastructural characteristics of intestinal neuroglia networks in rodents (Gabella, 1981). Extending these early reports, we and others have identified four morphological subtypes of mammalian enteric glia, which correlate with their position in the gut and relative to the ganglionic network in the gut wall (Boesmans et al., 2015; Gulbransen and Sharkey, 2012; Hanani and Reichenbach, 1994). Our current experiments demonstrate that all cardinal morphological and ultrastructural features ascribed to mammalian enteric glia are also found in the Tg(her4.3:EGFP)+ non-neuronal compartment of the zebrafish ENS, thus providing strong evidence that it represents the EGC lineage of the teleost ENS. Our failure to detect glial markers commonly used to identify mammalian enteric glia (such as GFAP and S100b) indicates that the expression of these genes may not be integral to the genetic programmes operating in the vertebrate ENS, but rather signifies dynamic physiological states of EGCs adopted in response to specialised local cues. In support of this idea, GFAP is dynamic and is normally detected in a subpopulation of mammalian EGCs in vivo (Boesmans et al., 2015) and expression of GFAP and S100b is enhanced in primary cultures of human enteric glia challenged with pro-inflammatory stimuli (Cirillo et al., 2011). It would be interesting to determine whether these glial markers are also upregulated in zebrafish EGCs following inflammatory pathology, infection or injury.

Despite the failure to detect canonical glia marker expression, our transcriptomic analysis of zebrafish EGCs revealed a considerable overlap in the gene expression profile of teleost and mammalian enteric glia. Among the genes expressed by both lineages are those encoding the early NC cell markers sox10, foxd3, tfap2a, zeb2b and plp1 (Hari et al., 2012; Knight et al., 2003; Mundell and Labosky, 2011; Mundell et al., 2012; Wang et al., 2011a; Weider and Wegner, 2017), PNS glia-specific marker col28a1b (Grimal et al., 2010), as well as the stem cell regulators sox2 and ptprz1a/b (Belkind-Gerson et al., 2017; Fujikawa et al., 2017; Heanue and Pachnis, 2011). The roles of these genes have been studied extensively in the context of neural development (sox10, foxd3, tfap2a, zeb2b, vim, sox2) and stem cell dynamics (sox2, ptprz1a/b), but their potential contribution to the homeostasis and function of enteric glia in adult animals remains unknown. We suggest that the shared gene expression modules we have identified between teleost and mammalian enteric glia represent evolutionary conserved regulatory programmes that are critical for intestinal physiology and ENS homeostasis and highlight the potential of vertebrate EGCs to serve as neurogenic precursors.

One of the unexpected findings of our work is the relatively small size of the non-neuronal compartment in the zebrafish ENS relative to its mammalian counterpart. A series of studies demonstrating that glial cells regulate synaptic activity of CNS neural circuits have led to the suggestion that the enhanced capacity for neural processing of the brain in higher vertebrates has been fuelled during evolution by the increased number, size and complexity of astrocytes (Han et al., 2013; Oberheim et al., 2006). Perhaps the higher number of enteric glia in mammals, relative to teleosts, may also reflect an increase in the functional complexity of intestinal neural circuits during vertebrate evolution and an enhanced scope of EGCs in the regulation of the complex gut tissue circuitry that maintains epithelial cell homeostasis, host defence and healthy microbiota (Grubišić and Gulbransen, 2017).

Several reports have documented that peripheral glial cells can acquire properties of neural crest stem cells (NCSCs) and give rise to diverse cell types. For example, Schwann cell precursors (SCPs) associated with growing nerves in mammalian embryos, in addition to generating the Schwann cell lineage of adult animals, also function as multipotent progenitors giving rise to diverse cell types, including mesenchymal and neuroendocrine cells, parasympathetic neurons and melanocytes (Parfejevs et al., 2018; Petersen and Adameyko, 2017). Echoing the developmental potential of SCPs, ENS progenitors already expressing molecular markers attributed to EGCs are also capable of generating enteric neurons and mature enteric glia (Cooper et al., 2016; Cooper et al., 2017; Lasrado et al., 2017). In addition to these studies, a growing body of evidence indicates that NCSC properties can be acquired by peripheral glia cell lineages from adult animals, including Schwann cells, glia of the carotid body and EGCs (Jessen et al., 2015; Pardal et al., 2007). However, it is generally thought that the reprogramming of differentiated glial cells into a NCSC-state is induced by injury, infection or other types of stress, including tissue dissociation and culture. Thus in mammals, EGCs can undergo limited neurogenesis in response to chemical injury to the myenteric plexus, pharmacological activation of serotonin signalling or bacterial gut infection (Belkind-Gerson et al., 2017; Joseph et al., 2011; Laranjeira et al., 2011; Liu et al., 2009). By providing evidence that zebrafish EGCs, in addition to their bona fide role as glial cells, also serve as constitutive ENS progenitors in vivo, our studies argue that the neurogenic potential of mammalian enteric glia disclosed under conditions of injury and stress, reflects an earlier evolutionary state of anamniote vertebrates, in which the same cell type exhibited properties of neural progenitors and mature glia. Although it is currently unclear whether neurogenic potential is a unique property of teleost EGCs, we speculate that peripheral glia lineages in lower vertebrates represent NCSCs that retain their developmental options but adjust to the cellular environment they reside in by acquiring additional specialised functions that contribute to local tissue function and homeostasis. Understanding the transcriptional and epigenetic mechanisms that underpin retention of the NCSC character and simultaneously allow novel functional adaptations during ontogenesis represents an exciting challenge of fundamental biology with practical implications in biomedical research. For example, identification of the molecular mechanisms that drive neuronal differentiation of enteric glia in vivo will facilitate strategies to harness the intrinsic neurogenic potential of mammalian EGCs and restore congenital or acquired deficits of intestinal neural circuits.

By subsuming features of both neural progenitors and glial cells, zebrafish EGCs show remarkable and unexpected parallels to RGCs, NSCs that are distributed widely in teleost brain, reflecting its pronounced neurogenic and regenerative potential (Alunni and Bally-Cuif, 2016; Than-Trong and Bally-Cuif, 2015), and take on functions normally attributed to astrocytes (Lyons and Talbot, 2015). The parallels of RGCs and EGCs are likely to extend beyond a cursory parity imposed by the demands of the resident organs (brain and gut) for continuous growth and specialised glia function, and apply to specific cellular and molecular mechanisms controlling their homeostasis and differentiation. Our proposed model (Figure 6—figure supplement 3) depicts key stages of the neurogenic trajectory available to adult zebrafish EGCs, which mirrors the stepwise differentiation of RGCs to pallial neurons (Than-Trong and Bally-Cuif, 2015; Than-Trong et al., 2018). Thus, similar to RGCs, the majority of EGCs remain quiescent at steady state (qEGCs), but in response to as yet unknown signals, a proportion of them enters the cell cycle giving rise to active EGCs (aEGCs). Whether the ability to enter the cell cycle is a property restricted to a subpopulation of qEGCs is currently unknown. More generally, to what extent qEGCs can be subdivided molecularly into subsets with distinct properties and function is an interesting question for future studies. Both qEGCs and aEGCs are currently identified by the Notch activity reporter Tg(her4.3:EGFP) and represent reversible cellular states distinguished by cell cycle marker expression and thymidine analogue incorporation (aEGCs). Extinction of Tg(her4.3:EGFP) expression is associated with irreversible commitment of aEGCs to enteric neural progenitors (eNPs), which eventually differentiate to mature enteric neurons capable of integration into functional intestinal neural circuits. We suggest that this transient population of eNPs correspond to the HuC/D-GFP- cells identified in the ENS of Tg(her4.3:EGFP;sox10:Cre;Cherry) zebrafish (Figure 3B). The proposed scheme ensures the long-term maintenance of the original population of EGCs and the generation of new enteric neurons to cater for the physical growth of the gut and the plasticity of its intrinsic neural networks.

Previous studies have established the central role of Notch signalling and its target genes in controlling the dynamics of NSCs in vertebrates (Chapouton et al., 2010; Imayoshi et al., 2010) and uncovered the differential contributions of distinct Notch receptors in regulating RGC proliferation and differentiation in the germinal zones of the zebrafish brain (Alunni et al., 2013; Than-Trong et al., 2018). Although the relevant Notch signalling components remain to be identified, our experiments provide evidence that the activation and differentiation of EGCs in adult zebrafish gut is also under the control of Notch signalling, pointing to further fundamental similarities in the mechanisms controlling the homeostasis of the CNS and ENS in vertebrates. Notch signalling has also been implicated in the development of the mammalian ENS by inhibiting the intrinsic neurogenic programme of ENS progenitors (Okamura and Saga, 2008). Our demonstration that the Notch activity reporter Tg(her4.3:EGFP) is activated in ENS progenitors shortly after they invade the gut and initiate neurogenic differentiation suggests a similar role of Notch signalling in the development of the zebrafish ENS, namely attenuation of the strong neurogenic bias of early ENS progenitors acquired as they enter the foregut and induce strong neurogenic transcription factors, such as Phox2B and Ascl1 (Charrier and Pilon, 2017).

The detailed hierarchical relationships of the identified cell types in the non-neuronal compartment of the zebrafish ENS and the potential regional differences in the dynamics of EGCs in zebrafish gut remain to be characterised. In addition, to what extent EGC-driven adult enteric neurogenesis in zebrafish depends on regulatory genes that control the differentiation of enteric neurons during vertebrate development (such as ret, ascl1, phox2b) is currently unclear. Nevertheless, the systematic characterization of the molecular programs underpinning the neuronal differentiation of EGCs in adult zebrafish is likely to inform strategies for the activation of the intrinsic neurogenic potential of mammalian EGCs and the repair of gastrointestinal neural networks damaged by disease or aging.

Materials and methods

Key resources table
Reagent type
(species) or
resource
DesignationSource or
reference
IdentifiersAdditional
information
Genetic reagent (Danio rerio)Tg(SAGFF234A)Asakawa et al., 2008; Kawakami et al., 2010SAGFF(LF)234AzTrap Resource from Koichi Kawakami Lab
Genetic reagent (Danio rerio)Tg(UAS:GFP)Kawakami et al., 2010Resource from Koichi Kawakami Lab
Genetic reagent (Danio rerio)Tg(−4.7sox10:Cre)Rodrigues et al., 2012Tg(−4.7sox10:Cre)ba74
Genetic reagent (Danio rerio)Tg(βactin-LoxP-STOP-LoxP-hmgb1-mCherry)Wang et al., 2011bTg(bactin2:loxP-STOP-loxP-hmgb1-mCherry)jh15
Genetic reagent (Danio rerio)rethu2846ZIRC; Knight et al., 2011ZL3218
Genetic reagent (Danio rerio)Tg(gfap:GFP)ZIRC; Bernardos and Raymond, 2006ZL1070
Genetic reagent (Danio rerio)Tg (−3.9nestin:GFP)EZRC; Lam et al., 200915206
Genetic reagent (Danio rerio)Tg(her4.3:EGFP)Yeo et al., 2007ZDB-ALT-070612–3
Genetic reagent (Danio rerio)Tg(SAGFF217B)Kawakami et al., 2010zTrap Resource from Koichi Kawakami Lab
Genetic reagent (Danio rerio)Tg(UAS:mmCherry)this paper
Antibodyanti-HuC/D (Mouse monoclonal)ThermofisherA21272; RRID:AB_25358221:200
Antibodyanti-Cherry (Goat polyclonal)Antibodies onlineABIN14400571:500
Antibodyanti-GFP (Chick polycloonal)Abcamab13970; RRID:AB_3007981:500
Antibodyanti-S100ß (Rabbit polyclonal)DakoZ0311; RRID:AB_100133831:500
Antibodyanti-mouse GFAP (Rabbit polyclonal)SigmaG9269; RRID:AB_4770351:500
Antibodyanti-zebrafish GFAP (Rabbit polyclonal)GenetexGTX128741; RRID:AB_28148771:500
Antibodyzrf-1 anti-zebrafish GFAP (Mouse mononclonal)Abcamab154474; RRID:AB_100138061:200
Antibodyanti-BFABP (Rabbit polyclonal)MerckABN14; RRID:AB_100003251:500
Antibodyanti-AcTu (Mouse monoclonal)SigmaT6793; RRID:AB_4775851:1000
Antibodyanti-MCM5gift from Soojin Ryu1:500
Commercial assay or kitRNAscope Flourescent Multiplex KitAdvanced Cell Diagnostics320850
Commercial assay or kitRNAscope Probe-Dr-sox10Advanced Cell Diagnostics444691-C3
Commercial assay or kitRNAscope Probe-Dr-foxd3Advanced Cell Diagnostics444681-C3
Commercial assay or kitRNAscope Probe-Dr-retAdvanced Cell Diagnostics579531
Chemical compound, drugNotch inhibitor LY411575Cambridge Bioscience16162

Animals

All animal experiments were carried out in compliance with the Animals (Scientific Procedures) Act 1986 (UK) and in accordance with the regulatory standards of the UK Home Office (Project Licence PCBBB9ABB). Experimental protocols were approved by the local Animal Welfare and Ethical Review Body (AWERB) of the Francis Crick Institute. Zebrafish stocks were maintained as described (Heanue et al., 2016a; Westerfield, 2000). Embryos and larvae were maintained and staged as described (Heanue et al., 2016a), while embryos used for time lapse were reared in 0.2 mM PTU from 24 hpf to inhibit melanisation, as described (Westerfield, 2000). Transgenic and mutant lines used were as follows: Tg(SAGFF234A) (Asakawa et al., 2008; Kawakami et al., 2010); Tg(UAS:GFP) (Kawakami et al., 2010), Tg(−4.7sox10:Cre) (Rodrigues et al., 2012), Tg(βactin-LoxP-STOP-LoxP-hmgb1-mCherry) (Wang et al., 2011b), rethu2846 (Knight et al., 2011), Tg(gfap:GFP) (Bernardos and Raymond, 2006), Tg (−3.9nestin:GFP) (Lam et al., 2009), Tg(her4.3:EGFP) (Yeo et al., 2007), Tg(SAGFF217B) (Kawakami et al., 2010). Note that the Tg(her4.3:EGFP) designation is the current ZFIN reference for this transgene, however it is also variously referred to as Tg(her4:EGFP) (Yeo et al., 2007) or Tg(her4.1GFP) (Kizil et al., 2012). her4.3 is one of 6 (of 9) mammalian orthologues of mammalian Hes5 found in tandem duplication on chromosome 23 of the zebrafish genome (Zhou et al., 2012). The stable Tg(UAS:mmCherry) line was generated by Tol2 transgenesis: co-microinjection of TOL2 transposase with a construct containing membrane-mCherry (mmCherry) downstream of two copies of the Gal4 recognition sequence UAS, with bicistronic α crystalinP:RFP cassette enabling red eye selection of carriers, as described previously (Gerety et al., 2013). Genotyping was done based on the lines’ previously described distinct fluorescent patterns, or by PCR in the case of Tg(ret hu2846/+), as described (Knight et al., 2011).

Immunohistochemistry

Request a detailed protocol

Immunohistochemistry was performed as previously described on whole-mount embryos/larvae or whole-mount adult intestines and brains (Heanue et al., 2016b). Primary antibodies used were as follows: HuC/D (mouse, ThermoFisher A21272, 1:200), Cherry (goat, Antibodies online ABIN1440057, 1:500), GFP (chick, Abcam ab13970, 1:500), S100β (rabbit, Dako Z0311, 1:500), mGFAP (rabbit, Sigma G9269, 1:500), zGFAP (rabbit, Genetex GTX128741, 1:500), zrf-1 (mouse, Abcam ab154474, 1:200), BFABP (Merck ABN14, 1:500), AcTu (mouse, Sigma T6793, 1:1000), MCM5 antibody (1:500, kindly provided by Soojin Ryu, Max Planck Institute for Medical Research, Heidelberg, Germany) and appropriate secondary antibodies conjugated to AlexaFluor 405, 488, 568 and 647 were used for visualisation (Molecular Probes). EdU was developed using the EdU Click-it kit following the manufacturer’s instructions and combined with fluorophores Alexa555 or Alexa647 (C10337 and C10339). For MCM5 labelling, antigen retrieval was required to expose the epitope. Briefly, after immunostaining for GFP, antigen retrieval with Citrate buffer (pH 6.0) was performed. All tissues were mounted on Superfrost Plus slides with Vectashield Mounting Media with/without DAPI (H1200/H1000, respectively). In all experiments, the CNS regions (larval brain and spinal cord or adult brain) provide a positive control (i.e Figure 1—figure supplement 1J–O) and negative controls are provided by immunostaining without primary antibody. Immunohistochemistry images were captured on a Leica CM6000 confocal microscope or an Olympus FV3000 confocal microscope, with standard excitation and emission filters for visualising DAPI, Alexa Flour 405, Alexa Flour 488, Alexa Flour 568 and Alexa Flour 647. Orthogonal views are used to clearly identify cells as expressing a marker of interest. Images processed with Adobe Photoshop 8.

Purification of ENS nuclei from adult gut muscularis externa

Request a detailed protocol

Adult Tg(sox10:Cre;Cherry) zebrafish intestines were first dissected, then cut along their length and immersed in HBSS (no calcium, no magnesium, (ThermoFisher 14170088)) containing 20 mM EDTA and 1% Penicillin/Streptomycin (ThermoFisher, 15140122) for 20–25 min at 37 °C until the epithelia cell layer was seen to begin detaching from the overlying muscularis externa, evident by clouding of the HBSS solution. After several washes in PBS (ThermoFisher 14190094), the tissue was placed under a dissecting microscope and the muscularis externa was grasped in forceps and agitated briefly to detach any remaining associated epithelial cells. Muscularis externa was tranfered to a fresh tube and purification of nuclei was performed essentially as described (Obata et al., 2020). Briefly, dounce homogenization was performed in lysis buffer (250 mM sucrose, 25 mM KCl, 5 mM MgCl2, 10 mM Tris buffer with pH8.0, 0.1 mM DTT) containing 0.1% Triton-X, cOmplete EDTA-free protease inhibitor (Sigma-Aldrich) and DAPI. The homogenate was filtered to remove debris and centrifuged to obtain a pellet containing the muscularis externa nuclei. For flow cytometric analysis, doublet discrimination gating was applied to exclude aggregated nuclei, and intact nuclei were determined by subsequent gating on the area and height of DAPI intensity. Both mCherry+ and mCherry- nuclear populations (termed Cherry+ and Cherry- in text and figures) were collected directly into 1.5 mL tube containing Trizol LS reagent (Invitrogen) using the Aria Fusion cell sorter (BD Biosciences). The obtained FCS data were further analysed using FlowJo software version 10.6.1. For each replicate, sorted cells from an average of 30 adult guts were pooled, containing approximately 30,000 mCherry+ or mCherry- nuclear populations.

RNA-sequencing and bioinformatic analysis

Request a detailed protocol

RNA was isolated from nuclei populations using the PureLink RNA Micro Kit (Invitrogen #12183016), according to the manufacturer’s instructions. Double stranded full-length cDNA was generated using the Ovation RNA-Seq System V2 (NuGen Technologies, Inc). cDNA was quantified on a Qubit 3.0 fluorometer (Thermo Fisher Scientific, Inc), and then fragmented to 200 bp by acoustic shearing using Covaris E220 instrument (Covaris, Inc) at standard settings. The fragmented cDNA was then normalized to 100 ng, which was used for sequencing library preparation using the Ovation Ultralow System V2 1–96 protocol (NuGen Technologies, Inc). A total of 8 PCR cycles were used for library amplification. The quality and quantity of the final libraries were assessed with TapeStation D1000 Assay (Agilent Technologies, Inc). The libraries were then normalized to 4 nM, pooled and loaded onto a HiSeq4000 (Illumina, Inc) to generate 100 bp paired-end reads.

Bioinformatics method summary RNA-sequencing-analysis

Request a detailed protocol

Sequencing was performed on an Illumina HiSeq 4000 machine. The ‘Trim Galore!’ utility version 0.4.2 was used to remove sequencing adaptors and to quality trim individual reads with the q-parameter set to 20 (https://www.bioinformatics.babraham.ac.uk/projects/trim_galore/). Then sequencing reads were aligned to the zebrafish genome and transcriptome (Ensembl GRCz10 release-89) using RSEM version 1.3.0 (Li and Dewey, 2011) in conjunction with the STAR aligner version 2.5.2 (Dobin et al., 2013). Sequencing quality of individual samples was assessed using FASTQC version 0.11.5 (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and RNA-SeQC version 1.1.8 (DeLuca et al., 2012). Differential gene expression was determined using the R-bioconductor package DESeq2 version 1.14.1 (Love et al., 2014R Development Core Team, 2008). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL (http://www.R-project.org). Data deposited at NCBI Geo (GSE145885). For differential gene expression analyses, the Wald-test and log-fold shrinkage was used in the context of the DESeq2 R-package (Parameters ‘test’ of the DESeq2-function was set to ‘Wald’ and the parameter ‘betaPrior’ was set to ‘TRUE’) (Love et al., 2014). Gene set enrichment analysis (GSEA) was conducted as described in Subramanian et al., 2005. For conversion from mouse to zebrafish gene names we used the Ensembl biomart tool (http://www.ensembl.org/biomart/martview).

RNA-Seq literature data gene list integration

Request a detailed protocol

For comparison of our transcriptomic data to published zebrafish data (Roy-Carson et al., 2017), we utilised the gene list presented in this paper (‘upregulated genes’ from Supplementary file 1) which represents the zebrafish larval ENS neuron transcriptome. For comparison of our data to genes previously described as characterising mammalian ENS glia (Rao et al., 2015), we used the list of genes identified in Rao et al., 2015 Table 1 (‘Table 1 Top 25 genes enriched in PLP1+ enteric glia’), and manually curated the zebrafish orthologues (see Supplementary file 7). To identify mouse ENS neuron and ENS glia signature genes we obtained mouse single-cell data from Zeisel et al., 2018. Specifically, we downloaded the single-cell read count data file for the enteric cells in the above project from https://storage.googleapis.com/linnarsson-lab-loom/l1_enteric.loom on the 12th Nov 2019. The data were processed using the Seurat package using the standard workflow (Stuart et al., 2019). The resolution parameter in the FindClusters function was set to 0.3. Neuronal and glia clusters were identified on the basis of the signature genes Elavl3, Elavl4, Prph (neuron) and Sox10, S100b, and Gfap (glia) (See Figure 2—figure supplement 3). The summarized neuron and glia clusters were subjected to a differential gene expression analysis using the FindMarkers Seurat function using the MAST algorithm.

Fluorescence in situ hybridization

Request a detailed protocol

Adult zebrafish intestines from Tg(sox10:Cre;Cherry) or Tg(her4.3:EGFP) were first dissected, then cut along their length, pinned to a silguard plate and immersed in HBSS (ThermoFisher 14170088) containing 20 mM EDTA and 1% penicillin/streptomycin (ThermoFisher, 15140122) for 20–25 min at room temperature to detach the epithelia layer. After several washes in PBS (ThermoFisher 14190094), the epithelia was manually teased away with forceps. After washing in PBS, 4% PFA was added to the plate with pinned tissue to fix overnight at 4 °C. Fluorescence in situ hybridization was then performed using the Advanced Cell Diagnostics RNAscope Fluorescent Multiplex Kit (ACD #320850), according to manufacturer’s specification and essentially as described (Obata et al., 2020). Briefly, tissue was washed in PBS, dehydrated through an ethanol series and then incubated with RNAscope Protease III for 25 min. Tissue was incubated overnight at 40°C in a HybeOven with customized probes (sox10, foxd3, ret). The next day, the tissue was washed twice with Wash Buffer before hybridization the with pre-amplifier, the appropriate amplifier DNA (Amp 1-FL, Amp 2-FL and Amp 3-FL) and appropriate fluorophores (Amp4 Alt A-FL/AltC-FL) at 40°C for 15–30 min, as per the manufacturer’s instructions. Tissues were then processed for immunohistochemistry and mounted directly onto Superfrost Plus slides (ThermoFisher Scientific #10149870) Vectashield Mounting Media without DAPI (VectorLabs H1000). Image were captured on a Leica CM6000 confocal microscope or an Olympus FV3000 confocal microscope, with standard excitation and emission filters for visualising DAPI, Alexa Flour 405, Alexa Flour 488, Alexa Flour 568 and Alexa Flour 647 and images processed with Adobe Photoshop 8.

Correlative light and electron microscopy

Request a detailed protocol

Intestines were dissected from Tg(her4.3:EGFP;SAGFF217B;UAS:mmCherry) adult animals and fixed in 4% formaldehyde 0.1% glutaraldehyde in phosphate buffer (PB) overnight at 4°C. Subsequently, the intestines were sectioned to 150 µm on a Leica vibratome, and stored in 2% formaldehyde in PB. Mid-gut sections were mounted in PB on SuperFrost Plus slides and imaged with an inverted Zeiss 880 confocal microscope with AiryScan, using standard emission and excitation filters for EGFP and mmCherry. A low magnification overview image was acquired using a 20x objective before 2–3 regions of interest (ROI) were identified per section that contained at least one EGFP+ cell of interest. The Airyscan was aligned for EGFP and mmCherry using an area outside of the ROIs where both fluorophores were identified. After Airyscan alignment, the ROIs were captured using a x63 glycerol objective and pixel size, z-depth and zoom (>1.8 x) were defined by Nyquist’s theorem. For super-resolution images, two adult midguts were scanned at low magnification to identify six regions regions of interest (ROIs) containing EGFP+ or Cherry+ cells, and six super resolution images taken. Once fluorescence microscopy was completed, the vibratome slices were further fixed in 2.5% glutaraldehyde and 4% formaldehyde in 0.1 M phosphate buffer (pH 7.4), and processed according to the method of the National Centre for Microscopy and Imaging Research (Deerinck et al., 2010) NCMIR methods for 3D EM: a new protocol for preparation of biological specimens for serial block face scanning electron microscopy (https://ncmir.ucsd.edu/sbem-protocol) before flat embedding between sheets of Aclar plastic. CLEM analysis included herein is taken from a single gut slice, from one of two original ROIs for this slice. This ROI contained six GFP+ cell bodies and two Cherry+ cell bodies.

SBF SEM data collection and image processing

Request a detailed protocol

Serial blockface scanning electron microscopy (SBF SEM) data were collected using a 3View2XP (Gatan, Pleasanton, CA) attached to a Sigma VP SEM (Zeiss, Cambridge). Flat embed vibratome slices were cut out and mounted on pins using conductive epoxy resin (Circuitworks CW2400). Each slice was trimmed using a glass knife to the smallest dimension in X and Y, and the surface polished to reveal the tissue before coating with a 2 nm layer of platinum. Backscattered electron images were acquired using the 3VBSED detector at 8,192*8,192 pixels with a dwell time of 6 µs (10 nm reported pixel size, horizontal frame width of 81.685 µm) and 50 nm slice thickness. The SEM was operated at a chamber pressure of 5 pascals, with high current mode inactive. The 30 µm aperture was used, with an accelerating voltage of 2.5 kV. A total of 1,296 images were collected, representing a depth of 64.8 µm, and volume of 432,374 µm3. Downstream image processing was carried out using Fiji (Schindelin et al., 2012). The images were first batch converted to 8-bit tiff format, then denoised using Gaussian blur (0.75 pixel radius), and resharpened using two passes of unsharp mask (10 pixel radius 0.2 strength, 2 pixel radius 0.4 strength), tailored to suit the resolution and image characteristics of the dataset. Image registration was carried out using the ‘align virtual stack slices’ plugin, with a translation model used for feature extraction and registration. The aligned image stacks were calibrated for pixel dimensions, and cropped to individual regions of interest as required. To generate a composite of the two volumes, Bigwarp (Bogovic JA et al., 2015; Russell et al., 2017) was used to map the fluorescence microscopy volume into the electron microscopy volume which was reduced in resolution to isotropic 50 nm voxels to reduce computational load. The multi-layered cellular composition of the tissue was noted to have caused substantial non-linear deformation during processing of the sample for electron microscopy when compared to prior fluorescence microscopy. After exporting the transformed light microscopy volume from Bigwarp, a two pixel Gaussian blur was applied, the datasets were combined, and the brightness/contrast adjusted for on-screen presentation. False coloured images were composed by annotating separate semi-transparent layers in Adobe Photoshop CC 2015.5 with reference to prior fluorescence microscopy and 3-dimensional context within the image stack. Only processes that could be clearly tracked through the volume from definitively marked cell bodies were coloured. Cell soma dimensions were determined by finding the largest X and Y extent of the cell body when scrolling through the CLEM volume, determining the entire extent of the cell body in Z, and calculating volume assuming an ellipsoid shape. Cell processes were tracked as far as possible through the CLEM volume. If a direct connection to the cell body was still visible at this farthest distance, the process length was traced on the image back to the middle of the cell soma. If no connection was apparent as was particularly the case for longer distance extensions, the distance in a straight line to the cell body was calculated in XYZ.

Time lapse imaging of zebrafish larvae

Request a detailed protocol

Embryos were raised in 0.2 mM PTU, lightly anaesthetised with 0.15 mg/ml Tricaine, and mounted into embryo arrays and overlayed with 0.6% low melt temperature agarose in embryo media essentially as described (Heanue et al., 2016a; Megason, 2009). Once set, the mould was overlaid with embryo media containing 0.15 mg/ml Tricane and 0.15 M PTU, and was replaced at least every 24 hr. Larvae were imaged using a Leica CM6000 confocal microscope, with a 20X water dipping objective. Standard excitation and emission filters were used to visualise EGFP and mmCherry expression. For each individual embryo, 33 z-stacks (z thickness 2.014 µm) were collected at a frame rate of 602 s, for 40.333 hr. Cells from the time-lapse recordings were tracked manually using the MTrack2 plugin on Fiji. To correct for growth or movement during the imaging process a reference point was taken, for each animal, as the point the anterior most spinal nerve, visible in the field of view, touched the gut. All calculated distances were given relative to this reference point.

EdU labelling

Request a detailed protocol

To label proliferative cells, adult zebrafish were kept in system water with 1 mM of EdU (0.05% DMSO) for 72 hr at a density of 4 zebrafish/litre. During chase periods adult zebrafish were kept in system water, which was changed every 2–3 days.

Mathematical modelling

Request a detailed protocol

Since the zebrafish ENS is largely confined to the myenteric plexus, and hence the zebrafish ENS resides within a two dimensional plane, therefore, only X and Y coordinates were used for subsequent analysis. Each image covered a 450 μm-450 μm area and XY coordinates of individual cells were taken as the centre of the nucleus and obtained from the CellCounter plugin for Fiji. We first estimated the density of specifically labelled cells at several distances around every cell type of interest using confocal images with an area 450 μm x 450 μm. Cell density was estimated in circles of increasing radius, r ∈ (20, 30, 40,. ., 100, 150,. .,500 μm), by dividing the number of cells within the circle by the surface area of the circle included within the image. When the radius was larger than the distance of the cell to the image edge, the area of the circle section overlapping with the image was numerically estimated by Monte Carlo simulation methods. We performed 50 Monte Carlo simulations for each confocal image with the observed number of cells of each phenotype in rearranged locations, according to a uniform distribution, on the 450 μm x 450 μm square area. Cell densities were estimated for each simulation as described above. To compare the recorded and simulated densities, we estimated the 90% confidence interval for simulated cell density under the assumption of cell homogeneity by fitting the gamma distribution function to the simulated values. When the average of the measured cell densities lied outside the 90% confidence interval, the observed spatial location was considered to be a non-chance event in a homogenous mixture of cells.

Notch inhibition

Request a detailed protocol

Notch signalling was inhibited by immersion with 10 µM LY411575 (Cambridge bioscience, 16162) (0.04% DMSO) in the system water, and was changed every 2–3 days, control zebrafish were incubated with the equivalent concentration of DMSO (0.04%).

Counting and statistics

Request a detailed protocol

In all experiments, the number of biological replicates (n: individuals, processed independently) is indicated in the figures. For quantifications of cells in embryos and larvae, the entire gut length was quantified. For quantification of cells in adults, nine random regions were counted: three from each of the main gut regions (intestinal bulb, midgut, hindgut). Orthogonal views are used to clearly identify cells as expressing a marker of interest. Cell counting analysis was carried out using the Cell Counter plugin. Statistical analysis was performed using R 3.3.1. Due to the non-normality of most of the data, all comparisons were carried out using a two-sided Mann-Whitney non-parametric test. Resultant p-values were corrected for multiple testing using the Benjamini-Hochberg method as implemented by the p.adjust function. A Pvalue of ≤0.05 was deemed to be significant and in figures designation of graded significance was as follows: p>0.05 (ns = non significant), p≤0.05 (*), p≤0.01 (**), p≤0.01 (***).

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
    The Enteric Nervous System
    1. JT Furness
    (2006)
    Wiley-Blackwell.
  25. 25
    Fine structure of the myenteric plexus in the guinea-pig ileum
    1. G Gabella
    (1972)
    Journal of Anatomy 111:69–97.
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
    The zebrafish colourless gene regulates development of non-ectomesenchymal neural crest derivatives
    1. RN Kelsh
    2. JS Eisen
    (2000)
    Development 127:515–525.
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63
  64. 64
  65. 65
  66. 66
  67. 67
  68. 68
  69. 69
  70. 70
  71. 71
    R: A language and environment for statistical computing, version 2.6.2
    1. R Development Core Team
    (2008)
    R Foundation for Statistical Computing, Vienna, Austria.
  72. 72
  73. 73
  74. 74
  75. 75
  76. 76
  77. 77
  78. 78
  79. 79
  80. 80
  81. 81
  82. 82
  83. 83
  84. 84
  85. 85
  86. 86
  87. 87
  88. 88
  89. 89
  90. 90
    The Zebrafish Book (Fourth edition)
    1. M Westerfield
    (2000)
    Eugene: University of Oregon Press.
  91. 91
  92. 92
  93. 93
  94. 94
  95. 95

Decision letter

  1. Julia Ganz
    Reviewing Editor
  2. Didier YR Stainier
    Senior Editor; Max Planck Institute for Heart and Lung Research, Germany

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This work investigates the cellular and molecular characteristics of the non-neuronal compartment of the enteric nervous system and identifies both expected and surprising properties of enteric glia cells in teleost fish. Importantly, this study finds that enteric glia cells in teleosts – in contrast to their mammalian counterpart – are neural progenitor cells during development into adulthood. These findings are of great importance for understanding the basic biology of how the enteric nervous system develops and is maintained in the adult, its regenerative potential, and the potential for therapeutic applications using enteric progenitor cells to treat human gut diseases.

Decision letter after peer review:

Thank you for submitting your article "Enteric glia as a source of neural progenitors in adult zebrafish" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Didier Stainier as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

As the editors have judged that your manuscript is of interest, but as described below that additional experiments are required before it is published, we would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). First, because many researchers have temporarily lost access to the labs, we will give authors as much time as they need to submit revised manuscripts. We are also offering, if you choose, to post the manuscript to bioRxiv (if it is not already there) along with this decision letter and a formal designation that the manuscript is 'in revision at eLife'. Please let us know if you would like to pursue this option. (If your work is more suitable for medRxiv, you will need to post the preprint yourself, as the mechanisms for us to do so are still in development.)

Summary:

Enteric neurogenesis is an area of intense interest and ongoing debate. At least some of this debate has been caused by discrepancies among data generated from animal models designed to identify neural precursor cells in mice. This question is of great importance both for understanding the basic biology of how the enteric nervous system develops and is maintained in the adult and for potential therapeutic applications, which aim to use progenitor cells to treat ENS diseases. The study is done in the zebrafish model system, in which the non-neuronal compartment of the ENS has not been characterized. Zebrafish are an attractive model system to tackle this question as they have widespread adult neurogenesis in other parts of the nervous system. The authors utilize molecular marker analysis, transcriptional profiling, and live imaging to probe the biology of glia in the zebrafish ENS. This work is a substantial and important contribution because there is little known about enteric glia in zebrafish, and the mechanisms of adult ENS neurogenesis are also poorly understood. The manuscript is well-written and well-illustrated. However, all three reviewers felt that there are major concerns with the conclusions and interpretation of the data with the main concern being that the evidence suggesting that the neural progenitor cells in zebrafish are enteric glia is insufficient and that this conclusion is premature based on the data.

Essential revisions:

1) The characterization of the non-neuronal compartment is done in two ways: first using an enhancer trap Gal4 driver line crossed to UAS:GFP. At 7 dpf, GFP+ cells are colocalized with Elavl. The interpretation that the GFP labels ENS progenitors and their descendants is not completely convincing. GFP will label the offspring for only a short time, except if the SAGFF234A transgene continues to be expressed in non-neuronal offspring, which has not been shown. Therefore, this experiment does not convey a permanent lineage label – the non-neuronal compartment could be much larger if the non-neuronal cells that are not progenitor cells do not express UAS:GFP.

Secondly, the lineage tracing using the Cre/loxP system performed relies on the fact that the sox10:Cre line and the reporter line labels the entire lineage of the ENS. Even though the sox10:Cre line has been published as labeling all neural-crest derivatives, the reporter line that was used in the original paper uses a different minimal promoter than the reporter line used in this study. Besides, Figure 1A shows quite a few Elavl positive cells that are not positive for mCherry and even in the adult, not all Elavl cells are mCherry positive. This shows that this approach does not even label all neuronal cells, which raises the possibility that also the non-neuronal compartment is not completely labeled.

To show that the mCherry+ cells fully encompass the entire ENS lineage, the authors have to explain why there are Elavl+/mCherry- cells. If this cannot be explained, a different Cre driver line needs to be used to label the entire ENS lineage.

2) The evidence provided is not sufficient to conclude that the neural progenitor cells in zebrafish are enteric glia. The data do provide strong evidence that these cells are neural stem cells, but the evidence that they are glia is weak. There is no functional assessment of the cells and the conclusion that they are glia is based solely on morphology and limited overlap in gene expression with mammalian glia. The non-neuronal compartment of the zebrafish ENS does seem to share some properties with enteric glia and stem cells, but the same is true for many cells including fibroblasts and neurons. It may be more accurate to say that mammalian enteric glia share some genomic overlap with neural stem cells in zebrafish than vice versa. The cells identified in this report do not express typical glial markers and are very different in terms of their morphology (see points b and c below). So far, the reviewer's opinion is that the simplest conclusion based on the data provided is that the cells identified here are neural stem cells, but there is little definitive evidence that they are glia. Concluding that the cells are glia in the Results section and referring to the cells as glia throughout the manuscript seems premature and inappropriate. These interpretations could be discussed in relation to the other potential possibilities in the Discussion section. The reviewers have brought up several points listed below that need to be addressed to provide more convincing data to show that the cells identified in the study are enteric glia.

a) There is no functional assessment of the cells and the conclusion that they are glia is based solely on morphology and limited overlap in gene expression with mammalian glia. The non-neuronal compartment of the zebrafish ENS does seem to share some properties with enteric glia and stem cells, but the same is true for many cells including fibroblasts and neurons. It may be more accurate to say that mammalian enteric glia share some genomic overlap with neural stem cells in zebrafish than vice versa. The cells identified in this report do not express typical glial markers and are very different in terms of their morphology (see points b).

b) The morphology of the cells identified in this study does not seem entirely consistent with that of enteric glia in mammals. As shown in Figure 3, GFP+ cells are very large (even larger than the neurons) and have extensive long branches. Data in Figure 7—figure supplement 1 show that GFP+ cells have processes that extend several hundred microns and this is not consistent with the morphology of mammalian enteric glia. The authors suggest that the GFP+ cells identified here correspond to the four morphological subtypes of enteric glia identified in mammals (see work by Hannani for original identification). How this can be true is not clear since two of the four subtypes of mammalian enteric glia are identified based on their localization within ganglia or interganglionic nerve bundles and the zebrafish ENS lacks ganglia. These points need to be addressed in the manuscript.

c) One of the major strengths of the nRNAseq experiments is that this strategy avoids some of the potential issues of cellular damage associated with tissue dissociation and cell isolation for RNA sequencing. Based on this, it is not clear why the authors chose to compare their transcriptional data with the mouse dataset generated by Rao et al. that used the same lengthy and damaging cellular isolation protocols that the authors were trying to avoid. More comparable datasets for mice are available such as the dataset generated by Delvalle et al., 2018, that used the RiboTag method to isolate glial specific translating RNA for sequencing. Similarly, single-cell transcriptional data by Zeisel et al., 2018, is available for mouse enteric glia. Another concern with using the Rao et al. dataset is that it contains a significant amount of contamination from neurons. These issues make the Rao et al. dataset less appropriate for comparisons in this study than other available datasets. In addition, it is not clear why the authors compare their results from RNA-seq of mCherry+ cells just with the 25 most highly expressed genes in the data set of Rao et al. in PLP1+ enteric glia and not with their entire data set. Also, even though some of the genes of the PLP1 data set are enriched, a portion of them are enteric progenitor markers (e.g. foxd3, sox10). Also, these genes could be expressed in different cell subpopulations, which has not been addressed by the authors – is the her4.3:EGFP cell population maybe an enteric progenitor cell population and an enteric glia cell population? It is important to provide much more extensive confirmation of gene expression in her4.3EGFP+ cells, particularly with more "bona fide" glial markers, such as PLP1, instead of only in situs for foxd3 and sox10 (which are also markers for enteric progenitor cells) and sox2 (which in the zebrafish CNS is expressed in progenitor cells and neurons). Also, colocalization between those markers would be important to test if the her4.3:gfp cell population consists of one cell type or different cell populations.

d) In lieu of using any of these empirically identified markers, the authors utilize a Notch activity-driven GFP transgenic reporter Tg(her4.3:EGFP) to identify putative enteric glia in all further experiments. They go on to conclude that all GFP+ cells in this transgenic line are enteric glia. This would be more convincing if: (1) the authors showed what proportion of GFP+ cells in the Tg(her4.3:EGFP)Sox10Cre:Cherry fish are Cherry+ and (2) to what extent Notch pathway targets were enriched in the non-neuronal Cherry+ cells from the RNA sequencing experiment.

3) A major claim is that "Notch signaling regulates neuronal differentiation". This is based on qualitative observations of occasional cells that were GFP+/Hu+ in the larval fish ENS and distinct studies in adult fish (EdU pulse chase in Figure 6C and LY411575 administration to sox10Cre;Cherry animals, Figure 7). Their results show that there is a marked increase in the number of non-neuronal cells incorporating EdU upon Notch pathway inhibition, suggesting glial proliferation, and a much smaller expansion in the number of Hu+ cells that are EdU+. In the absence of lineage tracing, this is insufficient evidence to conclude that glia differentiate into neurons upon inhibition of Notch. There may be a small pool of her4.3:EGFP+ progenitors that are not glia but can give rise to neurons, glia or both. Alternatively, a small number of neurons might re-enter the cell cyle upon Notch inhibition. In either case, the effect seems to be on proliferation rather than differentiation unless the authors have additional experimental data not presented. Live imaging of fish transgenic for both a neuronal reporter and her4.3:EGFP, that could visualize her4.3:EGFP+ cells turning off GFP as they upregulate a neuronal marker in a Notch inhibitor-dependent manner would be stronger evidence.

4) The conclusions from the MCM5 labeling and BrdU data in Figures 6, Figure 6—figure supplement 1 and 2 are not clear. If approximately 10% of the GFP+ cells are proliferative (positive for MCM5), how are only 10% positive for BrdU? Proliferative cells and their progeny should all be positive for BrdU so one would expect the percentage of BrdU+ GFP cells to be much higher. Some of these cells differentiate and lose their GFP signal, but one wouldn't expect this to happen for all cells if these are actually enteric glia. These data would suggest that all proliferative GFP+ cells eventually develop into neurons and there is no ongoing gliogenesis since the same number of GFP+ cells are labeled by both BrdU and MCM5.

5) Appropriate controls need to be included to validate the specificity of the antibodies used and the reporter lines.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Enteric glia as a source of neural progenitors in adult zebrafish" for further consideration by eLife. Your revised article has been evaluated by Didier Stainier (Senior Editor) and a Reviewing Editor.

The reviewers agree that the paper is excellent and has very strong evidence showing that the cells in question are progenitor cells that share characteristics with mammalian enteric glia. The reviewers have identified a remaining issue that needs to be addressed before acceptance, as outlined below:

The reviewers have discussed at length if they think it is appropriate to call the identified cells of the non-neuronal compartment as enteric glia within the Results section. The reviewers think the authors make a good case for calling the cells enteric glia, but feel that the finding that these cells show characteristics of enteric glia cells should be included as the author's interpretation in the Discussion section instead of calling the cells "enteric glia" already in the Results section.

The reviewers suggest either adapting the term "glial-like neural progenitors" or glial-like neural stem cells", as the authors themselves suggested or calling them "putative enteric glia" or "putative glial cells" when referencing the cells in the Results section.

https://doi.org/10.7554/eLife.56086.sa1

Author response

Essential revisions:

1) The characterization of the non-neuronal compartment is done in two ways: first using an enhancer trap Gal4 driver line crossed to UAS:GFP. At 7 dpf, GFP+ cells are colocalized with Elavl. The interpretation that the GFP labels ENS progenitors and their descendants is not completely convincing. GFP will label the offspring for only a short time, except if the SAGFF234A transgene continues to be expressed in non-neuronal offspring, which has not been shown. Therefore, this experiment does not convey a permanent lineage label – the non-neuronal compartment could be much larger if the non-neuronal cells that are not progenitor cells do not express UAS:GFP.

We agree with the reviewer’s view that the SAGFF234A;UAS:GFP line is not a lineage tracer of ENS progenitors and it was never meant to be used as such. Given the shortage of pan-ENS genetic labelling tools in zebrafish, it only provided an initial estimate of the relative size of the neuronal and non-neuronal components of the intestinal neural networks. To avoid potential confusion we have edited the original text to: “SAGFF234A;UAS:GFP animals in which ENS progenitors and enteric neurons were labelled with GFP”. The initial observations with this line were subsequently validated using the sox10Cre;Cherry transgenic combination, which permanently labels the nuclei of early neural crest cells and their descendants.

Secondly, the lineage tracing using the Cre/loxP system performed relies on the fact that the sox10:Cre line and the reporter line labels the entire lineage of the ENS. Even though the sox10:Cre line has been published as labeling all neural-crest derivatives, the reporter line that was used in the original paper uses a different minimal promoter than the reporter line used in this study. Besides, Figure 1A shows quite a few Elavl positive cells that are not positive for mCherry and even in the adult, not all Elavl cells are mCherry positive. This shows that this approach does not even label all neuronal cells, which raises the possibility that also the non-neuronal compartment is not completely labeled.

To show that the mCherry+ cells fully encompass the entire ENS lineage, the authors have to explain why there are Elavl+/mCherry- cells. If this cannot be explained, a different Cre driver line needs to be used to label the entire ENS lineage.

We agree that Tg(sox10Cre;Cherry) does not mark all ENS cells, and that the labelling efficiency is variable from animal to animal (as is the case for most Cre/LoxP-based genetic labelling systems). In the revised manuscript we provide new immunostaining images for Figure 1A and B, which reflect more accurately the proportion of HuC/D+Cherry- cells observed in the corresponding quantifications of Figure 1C.

Please note that even the Cre/reporter combination (Tg(sox10:Cre;ef1a:loxP-GFP-loxP-DsRed2)) used in the original paper does not mark all ENS cells. We specifically describe Cre/reporter efficiencies in the revised text. We provide a new data figure which shows that the labelling of HuC/D+ neurons in the gut of 7 dpf larvae from Tg(sox10Cre;Cherry) transgenics is 47.1% ± 19.9, whereas the corresponding figure from Tg(sox10:Cre;ef1a:loxP-GFP-loxP-DsRed2) animals is 80.8% ± 7.8 (Figure 1—figure supplement 1C). Importantly, these reporters show no differential bias towards neuronal vs. non-neuronal lineages. We support this statement by providing a new data figure (Figure 1—figure supplement 1D), which shows that the proportion of HuC/D+ and HuC/D- cells in the ENS of Tg(sox10Cre;Cherry) (84.8% ± 7.7 and 15.2% ± 7.7, respectively) and Tg(sox10:Cre;ef1a:loxP-GFP-loxP-DsRed2) (86.8% ± 6.4 and 13.2% ± 6.4%) lines are equivalent (p = 0.78). This new data provides independent support to the view that the non-neuronal compartment of zebrafish ENS is considerably smaller relative to its mammalian counterpart.

Our choice of the Tg(sox10Cre;Cherry) line was dictated by the fact that it provides clear nuclear labelling and allows combination with other transgenic reporters, such as Tg(her4.3:EGFP). In contrast, the Tg(sox10:Cre;ef1a:loxP-GFP-loxP-DsRed2) line renders the entire embryo GFP+, making combination with other GFP transgenics impossible. Taken together, we believe that the experiments described in our manuscript, the clarifications provided in this response and new supporting figures strongly support our conclusion that the nonneuronal compartment of the zebrafish ENS is considerably smaller relative to its counterpart in mammals. These findings parallel the glia to neuron ratio in the zebrafish pallium and in the Discussion we provide an evolutionary perspective of this observation.

2) The evidence provided is not sufficient to conclude that the neural progenitor cells in zebrafish are enteric glia. The data do provide strong evidence that these cells are neural stem cells, but the evidence that they are glia is weak. There is no functional assessment of the cells and the conclusion that they are glia is based solely on morphology and limited overlap in gene expression with mammalian glia. The non-neuronal compartment of the zebrafish ENS does seem to share some properties with enteric glia and stem cells, but the same is true for many cells including fibroblasts and neurons. It may be more accurate to say that mammalian enteric glia share some genomic overlap with neural stem cells in zebrafish than vice versa. The cells identified in this report do not express typical glial markers and are very different in terms of their morphology (see points b and c below). So far, the reviewer's opinion is that the simplest conclusion based on the data provided is that the cells identified here are neural stem cells, but there is little definitive evidence that they are glia. Concluding that the cells are glia in the Results section and referring to the cells as glia throughout the manuscript seems premature and inappropriate. These interpretations could be discussed in relation to the other potential possibilities in the Discussion section. The reviewers have brought up several points listed below that need to be addressed to provide more convincing data to show that the cells identified in the study are enteric glia.

a) There is no functional assessment of the cells and the conclusion that they are glia is based solely on morphology and limited overlap in gene expression with mammalian glia. The non-neuronal compartment of the zebrafish ENS does seem to share some properties with enteric glia and stem cells, but the same is true for many cells including fibroblasts and neurons. It may be more accurate to say that mammalian enteric glia share some genomic overlap with neural stem cells in zebrafish than vice versa. The cells identified in this report do not express typical glial markers and are very different in terms of their morphology (see points b).

b) The morphology of the cells identified in this study does not seem entirely consistent with that of enteric glia in mammals. As shown in Figure 3, GFP+ cells are very large (even larger than the neurons) and have extensive long branches. Data in Figure 7—figure supplement 1 show that GFP+ cells have processes that extend several hundred microns and this is not consistent with the morphology of mammalian enteric glia. The authors suggest that the GFP+ cells identified here correspond to the four morphological subtypes of enteric glia identified in mammals (see work by Hannani for original identification). How this can be true is not clear since two of the four subtypes of mammalian enteric glia are identified based on their localization within ganglia or interganglionic nerve bundles and the zebrafish ENS lacks ganglia. These points need to be addressed in the manuscript.

We are pleased that the reviewers are satisfied with our conclusion that her4.3:EGFP+ cells have properties of neural stem cells. However, they argue that we do not provide enough evidence to support the idea that these cells represent the elusive enteric glia of zebrafish. We disagree with their view and the statement in the review summary that the her4.3:EGFP+ cells “…are very different in terms of their morphology.” relative to canonical mammalian enteric glia. On the contrary, considering the evolutionary distance between teleosts and mammals, the distinct organization of zebrafish and mouse ENS and the documented plasticity of glial cells (including enteric glia), we would argue that in terms of morphology, the her4.3:EGFP+ cells are remarkably similar to mammalian enteric glia. Given the lack of a diagnostic morphological criterion(a) for enteric glia, such an argument is based inescapably on a list of criteria described in our manuscript. Specifically:

Our CLEM analysis (Figure 4 and Figure 4—figure supplement 1) demonstrates that the her4.3:EGFP+ cells share the majority of the ultrastructural characteristics of enteric glia, as they have been described in the seminal publications of G. Gabella. Among them are: (1) the close association and specialised contact points between her4.3:EGFP+ cells with enteric neurons, (2) the partial enveloping of enteric neurons by her4.3:EGFP+ processes, (3) the partial enveloping of axonal bundles and their subdivision into sectors by her4.3:EGFP+ cellular processes, (4) the scarcity of cytoplasm, and (5) the presence of nuclear crenations, which are observed almost universally in all glial cells in invertebrates and vertebrates.

Regarding the size of her4.3:EGFP+ cells, Figure 3C shows clearly that the nuclear size of her4.3:EGFP+ cells is comparable (and often smaller) to that of nearby neuronal nuclei. For example, compare the GFP+Cherry+ nuclei (arrows) to the GFP-Cherry+ nuclei (arrowheads). In the revised manuscript we make a note of this feature in the Figure 3C legend. The similar (or smaller) nuclear size of her4.3:EGFP+ cells in combination with the sparsity of their cytoplasm (revealed by the CLEM analysis, Figure 4) clearly indicates that their cell bodies are not larger in comparison to neurons. In our view, light microscopy of cellular networks (such as glia networks) is not the best approach to assess the overall size of individual cells, due to intermingling of cellular processes (see for example the network of GFP+ cells in Figure 3A and in higher magnification in Figure 3C, arrows). For this reason, both figures mentioned in your summary (Figure 3 and Figure 7—figure supplement 1), do not address the question of size of her4.3:EGFP+ cells. We have included a single z image as an inset into Figure 7—figure supplement 1B that clarifies that the GFP immunostainings interpreted by the reviewers as revealing cell processes extending for several hundred microns are actually groups of adjacent cells, and amended the figure legend accordingly.

However, our CLEM analysis has enabled us to compare the size of her4.3:EGFP+ cells and enteric neurons and measure the length of her4.3:EGFP+ cell processes (and thus estimate their size/volume) more accurately. Figure 4 and Video 1, which show her4.3:EGFP+ cells and Cherry+ neurons in close association, indicate that neurons are larger than the nearby GFP+ cells. We support this observation with specific measurements through the CLEM dataset. Specifically, the dimensions of the somata of the GFP+ cells shown in Figure 4A/B, Figure 4C/D and Figure 4—figure supplement 1B,C are: 8x7.6x2.5 µm (=79.6 µm3), 10x6x3 µm (=94.3 µm3) and 7.75x6.5x3 µm (=79.1 µm3), respectively (see also Materials and methods). Individual projections were also measured and are either shorter sheets that extend to approximately 4 µm from the middle of the cell body or they are longer extensions up to 18 µm. Specifically, the GFP+ cell in 4A/B has two processes of 4 µm and a longer process of 18 µm while the GFP+ cell in Figure 4—figure supplement 1B,C has two longer extensions of 8.5 and 9 µm. Cell sizes of mmCherry+ neurons have also been measured for reference. The Cherry+ cell somata in Figure 4A (left) and Figure 4A (right) are approximately 8.5x11.2x8 µm (=398.8 µm3) and 10x4.5x9.75 µm (=229.7 µm3), respectively, with projection lengths ranging from 16 to 55 µm. Taken together, these measurements support the view that, similar to mammalian enteric glia, her4.3:EGFP+ cells in zebrafish gut are highly branched, with shorter sheet like processes that extend 5 µm and longer projections that extend up to 18 µm. We confirm that these cells do not extend several hundred microns, as suggested, and by comparison are significantly smaller to neurons. Our revised manuscript includes these measurements in the text and the corresponding figure legends, with measurement methods described in the Materials and methods section.

These measurements allow us also to compare the size of her4.3:EGFP+ cells to that of mammalian enteric glia. The study of (Hanani and Reichenbach, 1994) indicates that the volume of guinea-pig enteric glia is approximately 330 µm3 and the average process length is ~29 µm (Type I “protoplasmic” glia) and ~83 µm (Type II “fibrous” glia), with the longest process 115 µm long. Boesmans et al. describe 4 types of mouse glia (Types I-IV), with cell bodies ranging from ~241 µm3 – ~335 µm3, and average process length of 30 µm to 124 µm, with the longest process length found in Type IV glia (138 µm). This allows us to conclude that her4.3:EGFP+ cells are smaller in comparison to mammalian enteric glia, which is consistent with the fact that zebrafish cells are typically smaller relative to mammalian cells.

The reviewers also argue that because of the lack of typical enteric ganglia in zebrafish, one would not have expected to detect Type I and Type II enteric glia, which are “defined by their localization within ganglia or interganglionic nerve bundles”. We disagree with this view. First, our qualitative light microscopy analysis demonstrates that her4.3:EGFP+ cells with Type I and Type II morphologies are observed (Figure 4—figure supplement 1B and C, respectively). Secondly, mammalian Type I enteric glia are associated with neuronal cell bodies (within ganglia), and we demonstrate in Figure 3—figure supplement 1B that a proportion of her4.3:EGFP+ cells are similarly associated with enteric neuronal cell bodies, with multiple processes wrapping around neuronal somata. On the other hand, mammalian Type II enteric glia are associated with neuronal projections (in interganglionic regions), and we demonstrate in Figure 3—figure supplement 1C that a proportion of her4.3:EGFP+ cells are similarly associated with enteric neuron processes. We argue that these clear morphological and contact specialisations are not surprising, but rather expected from zebrafish enteric glia, which, in a manner analogous to their mammalian counterparts, must establish close anatomical and functional interactions with individual (or small clusters of) enteric neurons (Type I) and their projections (Type II). In the revised manuscript, we have extended our description of her4.3EGFP+ cell morphologies, to further support the parallels we draw to the 4 types of mouse enteric glia and to address the reviewer’s comments. We have also provided new images for Figure 3—figure supplement 1, which we believe make our point of view clearer.

The blending of glia and neural stem cell characteristics within the same cell type is well established in neuroscience. An extensive literature is available on the glia character of neural stem cells and the neural stem cell character of glia cells, in vertebrates and invertebrates, during development and in adult stages. Several investigators have analysed the stem cell properties of Schwann cell precursors associated with growing nerves (Dyachuk et al., 2014; Kaukua et al., 2014), the acquisition of embryonic stem cell properties by Schwann cells (Clements et al., 2017), (Masaki et al., 2013), the neurogenic properties of carotid body glia cells (Pardal et al., 2007), the neural stem cell character of mammalian enteric glia revealed in vivo (Belkind-Gerson et al., 2017), (Laranjeira et al., 2011), in vitro reprogramming of adult enteric glia from the mouse intestine ((Joseph et al., 2011), (Laranjeira et al., 2011) and our own unpublished observations), and the neural stem cell properties of glia in the fly brain (https://www.biorxiv.org/content/10.1101/721498v1). Highly relevant to our studies is the mixed character of radial glial cells in the zebrafish pallium, which serve as the prototypical neural stem cells of the CNS and at the same time are “standing in” for the missing astrocytes in order to support pallial neurons and circuits. Therefore, given the similarities of her4.3:EGFP+ cells with mammalian enteric glia (in terms of morphology and gene expression profile) and the absence of a cell population that expresses canonical glial markers, we argue that these cells, in addition to their role as enteric neural progenitors, provide also the necessary glia function to intestinal neural circuits. Our Discussion section describes extensively the evidence from the literature that supports the co-existence of glia and neural stem cell characteristics.

We agree with the reviewers that it would have been wonderful to use a diagnostic functional assay for enteric glia. However, it is unclear whether any of the available assays for mammalian enteric glia (such as genetically encoded fluorescent indicators to trace changes in intracellular Ca2+ concentration) would be directly transferable to zebrafish and, more importantly, whether they would be diagnostic of enteric glia cell identity in this species. In any case, such an assay is currently unavailable for in vivo analysis of enteric glia in adult zebrafish and it is of interest to determine whether certain confounding issues can be overcome (such the presence of the liver impeding visibility, even in transparent adult zebrafish). However, we believe that such an undertaking is beyond the scope of the current manuscript.

The topic of constitutive neurogenesis in the vertebrate ENS is hotly debated. We believe that our work makes important contributions to this debate. First, it provides evidence that enteric neurogenesis does take place in vivo and second, that one does not have to look to a distinct population of cells that supports neurogenesis in the adult ENS. According to our data, it can be provided by the all-familiar enteric glia.

We have had longstanding debates amongst ourselves as to whether we should be calling the her4.3:EGFP+ cells of the zebrafish ENS “enteric glia” or not. Ultimately, the accumulated evidence described here of the shared lineage of her4.3:EGFP+ cells with ENS neurons, their cellular locations and close association with enteric neurons, their distinctive morphologies, their ultrastructural features, their expression of multiple well established enteric glial markers and their neural progenitor properties outweighed the lack of expression of GFAP/gfap (not requisite for mammalian glia in any case (Boesmans et al., 2015)), leading us to find “enteric glia” as the best descriptor of this population. We note that the original description of enteric glia from (Gabella, 1981) and (Hanani and Reichenbach, 1994) were based on the criteria of position, morphology and ultrastructural features alone, and that the identity of these cells has stood the test of time. Only subsequently have these cells been found to have progenitor cell properties, transcriptionally studied or analysed with electrophysiological and imaging tools to explore functional properties, such as activity. Moreover, we note that a corollary to not calling these cells “enteric glia” is to declare that the zebrafish ENS does not contain enteric glia, an idea which is undoubtedly more heretical than a glial cell that does not express GFAP. We also considered whether to propose a different terminology that would also capture their undisputed function as ENS stem cells. However, we do not know whether all her4.3:EGFP+ cells function as progenitors or whether it is a subpopulation of her4.3:EGFP+ cells with this property, and thus the general naming of the entire population as “enteric glia” is more accurate and future-proof. So, the fact that we could have chosen to term her4.3:EGFP+ cells “glial-like neural progenitors” or “glial-like neural stem cells” (GNPs? GNSC?) is clear to us, and ultimately would not detract from our findings, but is ambiguous. We argue, for all the reasons laid out in this rebuttal (in points 2a,b,c, above and below), to retain the name “enteric glia”.

c) One of the major strengths of the nRNAseq experiments is that this strategy avoids some of the potential issues of cellular damage associated with tissue dissociation and cell isolation for RNA sequencing. Based on this, it is not clear why the authors chose to compare their transcriptional data with the mouse dataset generated by Rao et al. that used the same lengthy and damaging cellular isolation protocols that the authors were trying to avoid. More comparable datasets for mice are available such as the dataset generated by Delvalle et al., 2018, that used the RiboTag method to isolate glial specific translating RNA for sequencing. Similarly, single-cell transcriptional data by Zeisel et al., 2018, is available for mouse enteric glia. Another concern with using the Rao et al. dataset is that it contains a significant amount of contamination from neurons. These issues make the Rao et al. dataset less appropriate for comparisons in this study than other available datasets. In addition, It is not clear why the authors compare their results from RNA-seq of mCherry+ cells just with the 25 most highly expressed genes in the data set of Rao et al. in PLP1+ enteric glia and not with their entire data set. Also, even though some of the genes of the PLP1 data set are enriched, a portion of them are enteric progenitor markers (e.g. foxd3, sox10). Also, these genes could be expressed in different cell subpopulations, which has not been addressed by the authors – is the her4.3:EGFP cell population maybe an enteric progenitor cell population and an enteric glia cell population? It is important to provide much more extensive confirmation of gene expression in her4.3EGFP+ cells, particularly with more "bona fide" glia markers, such as PLP1, instead of only in situs for foxd3 and sox10 (which are also markers for enteric progenitor cells) and Sox2 (which in the zebrafish CNS is expressed in progenitor cells and neurons). Also, colocalization between those markers would be important to test if the her4.3:GFP cell population consists of one cell type or different cell populations.

We thank the reviewers for their interesting comments and useful suggestions. We do recognise the limitation of the Rao study for statistically robust analysis, including the lack of replicates (1 replicate per condition), which is not mentioned by the reviewers. It was for this reason that we found more in depth use of their data was fraught and we instead relied on the top 25 gene list that was presented by the authors, reasoning that it was the part of their data they felt the most confidence in. Regarding the very interesting work of Delvalle and colleagues, please note that this dataset does not identify glia-specific genes, but rather genes whose expression changes in enteric glia in response to specific experimental manipulation (colitis). Although this analysis is very well placed to identify the “inflammatory” character of mammalian enteric glia, it does not help us explore the similarities between zebrafish and mammalian enteric glia.

We found more fortune however, with the (Zeisel et al., 2018) publication mentioned by the reviewers, despite requiring a comparison between bulk (our) and single cell (Zeisel et al.) data. This analysis has identified an extended list of genes in the zebrafish ENS transcriptome with orthologues expressed in mammalian glia. The new data is described in the Results section of the revised manuscript and presented in a new Figure 2—figure supplement 2 and new Supplementary files 3, 4 and 5. The methodology of the new analysis is described in the Materials and methods section.

Specifically, we first identified those genes from the Zeisel dataset (raw data downloaded from mousebrain.org and analysed as described in the Materials and methods section) that are differentially expressed between mouse ENS neurons and glia and then identified their orthologues in our zebrafish ENS transcriptome (Supplementary file 3). We show that 366 mouse ENS neuron enriched genes have orthologues in our zebrafish transcriptome dataset, including ret, phox2bb, elavl3, elavl4, and likely reflect the neuronal component of our bulk dataset (Figure 2—figure supplement 2A,B). We also show that 63 mouse ENS glia enriched genes have orthologues present in our zebrafish transcriptomic data, suggesting that they are expressed in the non-neuronal component of the zebrafish ENS (Figure 2—figure supplement 2A,C), including sox10, foxd3, plp1b, zeb2b, and Sox2. Significantly, this analysis did not detect canonical glia markers (gfap, s100b and fabp7a/b) in the zebrafish ENS transcriptome.

The reviewers suggest that we use more “bona fide” glia markers, such as PLP1, to characterise the her4.3:EGFP+ cells, rather than the “progenitor markers” sox10 and foxd3. However, out of the “traditional” markers used widely in the literature for the identification of enteric glia (Gfap, S100b, Plp1, BFABP, Sox10, Sox2, Foxd3), none of them defines enteric glia identity. Some may be upregulated in a subset of enteric glia of adult mammals, but all also mark neural progenitors and neural stem cells. This is supported by the literature and our own single cell RNA sequencing of mammalian ENS cells at various developmental stages (unpublished). Since PLP1 was specifically mentioned in the review summary, we note that this gene has been used by the Adameyko lab to drive expression of Cre recombinase in early neural crest cells, and therefore this gene is not materially different from the other genes we have examined. Further future analysis of emerging transcriptomic datasets may perhaps identify novel marker genes that uniquely define enteric (and other peripheral) glia, but such marker(s) do not exist at present.

With regards to whether the her4.3:EGFP+ cell population consists of one cell type or different populations, we now include new data in Figure 3F of our revised manuscript to show that all her4.3:EGFP+ cells express Sox2. In this figure we also show that foxd3 is expressed in only a proportion of the her4.3:EGFP+/Sox2+ cells, giving first indication of molecular differences in her4.3:EGFP+ cells. This new data is described in our revised manuscript. Heterogeneity is also indicated by the fact that only a proportion of her4.3:EGFP+ cells are proliferative in our experiments. We now make a specific mention to potential heterogeneity of EGCs in the Discussion of our revised manuscript. Further robust exploration of the question of cell type heterogeneity would be best approached using a comprehensive single-cell transcriptomic strategy, an endeavour that we feel is outside of the scope of this current manuscript.

d) In lieu of using any of these empirically identified markers, the authors utilize a Notch activity-driven GFP transgenic reporter Tg(her4.3:EGFP) to identify putative enteric glia in all further experiments. They go on to conclude that all GFP+ cells in this transgenic line are enteric glia. This would be more convincing if: (1) the authors showed what proportion of GFP+ cells in the Tg(her4.3:EGFP)Sox10Cre:Cherry fish are Cherry+ and (2) to what extent Notch pathway targets were enriched in the non-neuronal Cherry+ cells from the RNA sequencing experiment.

We used the Tg(her4.3:EGFP) line not “in lieu of…empirically identified markers”, but because our search for molecular markers and transgenic reporters that are traditionally used in the field of mammalian enteric glia was unsuccessful, a finding corroborated by our initial and subsequent transcriptomic analysis. This, and evidence from the literature, led us to the hypothesis that a Notch activity transgenic reporter could identify enteric glia in zebrafish, an idea that was subsequently validated experimentally. Of course, having in hand a transgenic line that marks enteric glia facilitates the study of this cell population, as it can be combined with other immunohistochemical and transgenic tools and allows the application of techniques such as CLEM. We have modified and expanded the text to clarify this logic.

We clarify in response to point 1 (above) that the lineage tools we have used do not label all ENS cells (new Figure 1—figure supplement 1C) and do not show differential bias towards neural vs. non-neural lineages (new Figure 1—figure supplement 1D). As such, no particular significance can be drawn to the presence of her4.3:EGFP+ cells that are Cherry+ or Cherry-, a point made clear in the text. In particular, with regards to the Cherryher4.3:EGFP+ cells, we are confident that they are part of the non-neuronal lineage of the ENS, as allher4.3:EGFP+ cells express Sox2, a gene which in the tunica muscularis of mammalian gut marks exclusively glial cells (this additional data is now presented in Figure 3F). Furthermore, we often find in the gut of Tg(sox10Cre;Cherry) fish large domains in which all her4.3:EGFP+ cells (and HuC/D+ cells) are Cherry+ (see for example Figure 3C, D), suggesting stochastic expression of the Tg(sox10Cre;Cherry) reporter.

We are confident that the Tg(her4.3:EGFP) transgene reflects Notch activity, since inhibition of the Notch signalling using LY411575 silences expression of the transgene (Figure 7—figure supplement 1). In our revised manuscript we have included a transcriptomic results table as a new Supplementary file 1, which highlights differential expression in the Cherry+ and Cherry+ population, without using any database comparisons. However, the Notch receptors (notch1a, notch1b, Notch2, notch3) are not significantly upregulated in either the Cherry+ or Cherry- populations, with similar results for multiple Notch targets (all her and hey genes with the exception of her2, her4.5 and her4.4). This result is not entirely surprising. Given widespread expression of Notch signalling components in multiple tissues of the mammalian adult gut (i.e. lamina propria, blood vessels, epithelia and enteric nervous system; Sander and Powell, 2004, Okamura and Saga, 2008), we expect notch receptors and their downstream targets to be expressed in both Cherry+ and Cherry- populations. In such case, our bulk differential transcriptomic strategy is not best placed to identify the Notch targets acting within adult glia, and we will not use this experiment to make statements about expression of Notch targets. This question would be best addressed using a single-cell transcriptomic strategy, an undertaking we feel is outside the scope of this current manuscript.

3) A major claim is that "Notch signaling regulates neuronal differentiation". This is based on qualitative observations of occasional cells that were GFP+/Hu+ in the larval fish ENS and distinct studies in adult fish (EdU pulse chase in Figure 6C and LY411575 administration to sox10Cre;Cherry animals, Figure 7). Their results show that there is a marked increase in the number of non-neuronal cells incorporating EdU upon Notch pathway inhibition, suggesting glial proliferation, and a much smaller expansion in the number of Hu+ cells that are EdU+. In the absence of lineage tracing, this is insufficient evidence to conclude that glia differentiate into neurons upon inhibition of Notch. There may be a small pool of her4.3:EGFP+ progenitors that are not glia but can give rise to neurons, glia or both. Alternatively, a small number of neurons might re-enter the cell cyle upon Notch inhibition. In either case, the effect seems to be on proliferation rather than differentiation unless the authors have additional experimental data not presented. Live imaging of fish transgenic for both a neuronal reporter and her4.3:EGFP, that could visualize her4.3:EGFP+ cells turning off GFP as they upregulate a neuronal marker in a Notch inhibitor-dependent manner would be stronger evidence.

Thank you for the opportunity to clarify the issues related to this comment.

First, our evidence that Notch signalling regulates neuronal differentiation is based exclusively on the data presented in Figure 7, which shows that upon exposure to the LY411575 inhibitor, both the percentage of nonneuronal ENS cells that incorporate EdU (Cherry+Hu- EdU+) and the percentage of EdU-labelled enteric neurons (Cherry+Hu+EdU+) increases at two adult stages (3 mo and 6 mo). Please note that in these experiments we do not follow specifically her4.3:EGFP+ cells because Notch inhibition silences the transgene (Figure 7—figure supplement 1) and therefore we are unable to use it to follow her4.3:EGFP+ cells. We agree with the reviewers that we do not provide evidence that Notch regulates directly neuronal differentiation in the adult zebrafish ENS and that the increase in the fraction of EdU-lableled Hu+Cherry+ enteric neurons following treatment with LY411575 could result from the increase in the size of the neurogenic progenitor pool. We have now revised our manuscript to reflect more accurately our experimental results and indicate that Notch signalling regulates the dynamics of non-neuronal cells.

Second, our conclusion that her4.3:EGFP+ cells give rise to neurons during homeostasis has no bearing on the identified role of Notch signalling in neuronal differentiation. We simply aim to show that there is constitutive neurogenesis in the ENS of adult zebrafish and that the GFP+ non-neuronal cell population is a source of newborn enteric neurons. Suggestive evidence for neuronal differentiation of her4.3:EGFP+ cells was first provided by the occasional double labelling of individual cells by Hu and GFP (Figure 5—figure supplement 1D) and the live imaging presented in Figure 5. Further direct evidence was provided by the pulse-chase experiments (shown in Figure 6A-E) in conjunction with the computational model presented in Figure 6F, G. Thymidine analogue incorporation has been used widely in neuroscience to demonstrate de novo neurogenesis in different parts of the adult nervous system. Taking advantage of the postmitotic nature of enteric neurons, we used EdU pulse-labelling to identify EdU+ neurons in the gut, which provided unequivocal evidence for constitutive neurogenesis in the adult zebrafish ENS. Furthermore, spatial analysis of EdU+her4.3:EGFP+ and EdU+Hu+ cells established that her4.3:EGFP+ cells serve as precursors of enteric neurons. We recognise that the use of a Cre/LoxP-based lineage tracing system (which we have used extensively in our laboratory) would have been an alternative way to demonstrate neuronal differentiation of her4.3:EGFP+ cells, but so far we have been unable to identify a suitable transgenic line that would allow us to conduct this experiment in adult zebrafish. While such a line would be useful in the future to explore the kinetics of neuronal differentiation of her4.3:EGFP+ cells in the adult ENS, we believe that the current studies provide strong evidence for constitutive neurogenesis and identifies these cells as a source of newborn enteric neurons throughout life. Indeed, in the review summary the reviewers are supportive of the neural stem cell properties our studies demonstrate.

4) The conclusions from the MCM5 labeling and BrdU data in Figures 6, Figure 6—figure supplement 1 and 2 are not clear. If approximately 10% of the GFP+ cells are proliferative (positive for MCM5), how are only 10% positive for BrdU? Proliferative cells and their progeny should all be positive for BrdU so one would expect the percentage of BrdU+ GFP cells to be much higher. Some of these cells differentiate and lose their GFP signal, but one wouldn't expect this to happen for all cells if these are actually enteric glia. These data would suggest that all proliferative GFP+ cells eventually develop into neurons and there is no ongoing gliogenesis since the same number of GFP+ cells are labeled by both BrdU and MCM5.

We thank the reviewers for the opportunity to clarify these points. We believe that the similar size of the MCM5+ and EdU+ subsets of her4.3:EGFP+ cells is explained by the cell cycle dynamics of this cell population. Mcm5 is expressed at varying levels throughout the cell cycle while an antibody that binds to the MCM5 protein (and used widely in the field and in our study (Chapouton et al., 2010; Ryu et al., 2005)) labels the G1/S transition phase (whose length depends on how rapidly cells proliferate) and certain stages of G2. On the other hand, EdU marks cells that go through S phase. In quiescent tissues, one would expect that over a 3-day EdU pulse proliferating cells would have gone through the various phases of the long cell cycles only once, resulting in similar fractions of MCM5+ and EdU+ cells. Indeed, given that MCM5 labels several stages of the cell cycle, including G1/S (which in quiescent tissues can be exceptionally long), it could even label a greater population of cells relative to EdU. We note that examples of similar numbers of cells labelled with MCM5+ and EdU+ in proliferating tissues have also been reported in the literature (Lange et al., 2020).

Our initial attempts to EdU label the her4.3:EGFP+ cells using 1-hour pulses (a protocol used to identify highly proliferative tissues) failed to label enough cells to study effectively their proliferation/neurogenesis and suggested that we are dealing with a cell population which, similar to RGCs in the pallium, is generally quiescent and any dividing cells have long cell cycles. Based on these observations, we reasoned that longer time frames would allow us to capture the her4.3:EGFP+ cells that progress through and complete the cell cycle and therefore we adopted the 3-day EdU pulse protocol. Our finding that the majority of her4.3:EGFP+EdU+ cells at the end of the 3-day EdU pulse (t0) are in doublets supports the view that during this period any cycling cells have undergone only one cell division and explains why we detect similar fractions of MCM5+ and EdU+ cells. In our revised manuscript we compare our MCM5 and EdU results and state our conclusion that Tg(her4.3:EGFP)+ cells represent a largely quiescent cell population and the that dividing cells have long cell cycles.

Despite the neuronal differentiation observed over the 11-day chase period, a significant fraction of her4.3:EGFP+ cells in 3 month old zebrafish retain their EdU label (Figure 6D), suggesting that proliferating glial cells return to quiescence and their glial role (=gliogenesis). Similar observations were made also for 6-month-old animals.

In future studies we plan to characterize in greater detail the proliferation and differentiation dynamics of her4.3:EGFP+ cells.

5) Appropriate controls need to be included to validate the specificity of the antibodies used and the reporter lines.

The antibodies and reporter lines applied in this study are all widely used, and we provide appropriate references and/or catalogue numbers. For many antibodies and reporter lines, particularly for those that we show negative data in the ENS (i.e S100B/BFABP/GFAP/Tg(gfap:GFP)) we show larval spinal cord staining as internal positive controls in Figure 1—figure supplement 1 (J-O). We clarify this point in the Material and methods section to indicate that larval CNS and adult brain are used as positive controls in our experiments, and explicitly refer to this figure. We have also clarified in the Materials and methods section that “no primary antibody” negative control experiments have been performed for immunostaining, and yielded negative results.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

The reviewers agree that the paper is excellent and has very strong evidence showing that the cells in question are progenitor cells that share characteristics with mammalian enteric glia. The reviewers have identified a remaining issue that needs to be addressed before acceptance, as outlined below:

The reviewers have discussed at length if they think it is appropriate to call the identified cells of the non-neuronal compartment as enteric glia within the Results section. The reviewers think the authors make a good case for calling the cells enteric glia, but feel that the finding that these cells show characteristics of enteric glia cells should be included as the author's interpretation in the Discussion section instead of calling the cells "enteric glia" already in the Results section.

The reviewers suggest either adapting the term "glial-like neural progenitors" or glial-like neural stem cells", as the authors themselves suggested or calling them "putative enteric glia" or "putative glial cells" when referencing the cells in the Results section.

We are pleased that our revised manuscript now makes “a good case for calling the cells enteric glia”. The only issue remaining is deciding the appropriate place in the manuscript to present the conclusion that the her4.3:EGFP+ non-neuronal population can be called “enteric glia”.

We have modified our manuscript in accordance with the proposed naming of these cells as “putative enteric glia” in the Results section and “enteric glia” in the Discussion section, after a paragraph outlining our case. On reading this version, we find that calling them “putative enteric glia” in the Results after having presented all the evidence in support of their “enteric glia” character, and then calling them “enteric glia” in the Discussion (without in the meantime presenting any additional evidence), affects the readability of the manuscript, is a source of confusion and blurs the clarity in the overall paper.

We are acutely aware of the need to clearly lay out our case for calling these cells “enteric glia” and appreciate the attention the reviewers have brought to this important point. But at the same time, it’s our wish that the final manuscript’s clarity and readability not be impinged.

We would like to propose an alternative approach to modifying the manuscript, which adopts the reviewers’ suggestion for a summary paragraph, but that we believe would be more effective. We propose that we add a small but clear and evidence-based paragraph at the end of the relevant part of the Results section that would lay out our entire case for calling the her4.3:EGFP+ cells “enteric glia”. This paragraph would follow the last piece of data that led us to this conclusion, the conventional place for summarizing all relevant evidence. Please note that up to this point we have not used the term “enteric glia” to identify the her4.3:EGFP+ cells and the term is only used in the second part of the Results section, after all supporting evidence has been presented. We have now modified our manuscript according to this approach as well.

https://doi.org/10.7554/eLife.56086.sa2

Article and author information

Author details

  1. Sarah McCallum

    Development and Homeostasis of the Nervous System Laboratory, The Francis Crick Institute, London, United Kingdom
    Contribution
    Conceptualization, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2156-3219
  2. Yuuki Obata

    Development and Homeostasis of the Nervous System Laboratory, The Francis Crick Institute, London, United Kingdom
    Contribution
    Conceptualization, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5461-3521
  3. Evangelia Fourli

    Development and Homeostasis of the Nervous System Laboratory, The Francis Crick Institute, London, United Kingdom
    Contribution
    Formal analysis, Validation, Investigation, Visualization, Writing - original draft, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6549-342X
  4. Stefan Boeing

    Bionformatics & Biostatistics Science Technology Platform, The Francis Crick Institute, London, United Kingdom
    Contribution
    Resources, Data curation, Software, Formal analysis, Validation, Visualization, Methodology, Writing - original draft, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0495-5659
  5. Christopher J Peddie

    Electron Microscopy Science Technology Platform, The Francis Crick Institute, London, United Kingdom
    Contribution
    Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing
    Competing interests
    No competing interests declared
  6. Qiling Xu

    Neural Development Laboratory, The Francis Crick Institute, London, United Kingdom
    Contribution
    Resources, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
  7. Stuart Horswell

    Bionformatics & Biostatistics Science Technology Platform, The Francis Crick Institute, London, United Kingdom
    Contribution
    Data curation, Software, Formal analysis, Visualization, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2787-1933
  8. Robert N Kelsh

    Department of Biology and Biochemistry, University of Bath, Bath, United Kingdom
    Contribution
    Resources, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9381-0066
  9. Lucy Collinson

    Electron Microscopy Science Technology Platform, The Francis Crick Institute, London, United Kingdom
    Contribution
    Formal analysis, Supervision, Methodology, Project administration, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0260-613X
  10. David Wilkinson

    Neural Development Laboratory, The Francis Crick Institute, London, United Kingdom
    Contribution
    Resources, Supervision, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6757-7080
  11. Carmen Pin

    Clinical Pharmacology and Quantitative Pharmacology, Clinical Pharmacology and Safety Sciences, R&D, AstraZeneca, Cambridge, United Kingdom
    Contribution
    Resources, Software, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing
    Competing interests
    Carmen Pin is affiliated with AstraZeneca. The author has no financial interests to declare.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-8734-6167
  12. Vassilis Pachnis

    Development and Homeostasis of the Nervous System Laboratory, The Francis Crick Institute, London, United Kingdom
    Contribution
    Conceptualization, Supervision, Funding acquisition, Writing - original draft, Project administration, Writing - review and editing
    For correspondence
    vassilis.pachnis@crick.ac.uk
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-9733-7686
  13. Tiffany A Heanue

    Development and Homeostasis of the Nervous System Laboratory, The Francis Crick Institute, London, United Kingdom
    Contribution
    Conceptualization, Formal analysis, Supervision, Validation, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing
    For correspondence
    tiffany.heanue@crick.ac.uk
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6678-8246

Funding

Francis Crick Institute (core funding)

  • Vassilis Pachnis

BBSRC (BB/L022974/1)

  • Vassilis Pachnis

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We are grateful to Laure Bally-Cuif, Alessandro Alluni, Emmanuel Than-Trong for providing Tg(her4.3:EGFP) transgenic fish and specialist knowledge, to Donald Bell for expert advice in time-lapse imaging experiments, to the Aquatics BRF staff for fish husbandry, to the Flow Cytometry STP for FACS cell sorting, to the Advanced Sequencing Facility STP for library prep and sequencing, and to Joe Brock for scientific illustrations.

Ethics

Animal experimentation: All animal experiments were carried out in compliance with the Animals (Scientific Procedures) Act 1986 (UK) and in accordance with the regulatory standards of the UK Home Office (Project Licence PCBBB9ABB). Experimental protocols were approved by the local Animal Welfare and Ethical Review Body (AWERB) of the Francis Crick Institute.

Senior Editor

  1. Didier YR Stainier, Max Planck Institute for Heart and Lung Research, Germany

Reviewing Editor

  1. Julia Ganz

Publication history

  1. Received: February 17, 2020
  2. Accepted: August 26, 2020
  3. Accepted Manuscript published: August 27, 2020 (version 1)
  4. Version of Record published: September 28, 2020 (version 2)

Copyright

© 2020, McCallum et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,127
    Page views
  • 185
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Chromosomes and Gene Expression
    2. Developmental Biology
    Rozenn Riou et al.
    Research Article

    Erythropoietin (EPO) is a key regulator of erythropoiesis. The embryonic liver is the main site of erythropoietin synthesis, after which the kidney takes over. The adult liver retains the ability to express EPO, and we discovered here new players of this transcription, distinct from the classical hypoxia-inducible factor pathway. In mice genetically-invalidated in hepatocytes for the chromatin remodeler Arid1a, and for Apc, the major silencer of Wnt pathway, chromatin was more accessible and histone marks turned into active ones at the Epo downstream enhancer. Activating β-catenin signaling increased binding of Tcf4/β-catenin complex and upregulated its enhancer function. The loss of Arid1a together with β-catenin signaling, resulted in cell-autonomous EPO transcription in mouse and human hepatocytes. In mice with Apc-Arid1a gene invalidations in single hepatocytes, Epo de novo synthesis led to its secretion, to splenic erythropoiesis and to dramatic erythrocytosis. Thus, we identified new hepatic EPO regulation mechanism stimulating erythropoiesis.

    1. Developmental Biology
    2. Neuroscience
    Amir Rattner et al.
    Tools and Resources

    In the hippocampus, a widely accepted model posits that the dentate gyrus improves learning and memory by enhancing discrimination between inputs. To test this model, we studied conditional knockout mice in which the vast majority of dentate granule cells (DGCs) fail to develop – including nearly all DGCs in the dorsal hippocampus – secondary to eliminating Wntless (Wls) in a subset of cortical progenitors with Gfap-Cre. Other cells in the Wlsfl/-;Gfap-Cre hippocampus were minimally affected, as determined by single nucleus RNA sequencing. CA3 pyramidal cells, the targets of DGC-derived mossy fibers, exhibited normal morphologies with a small reduction in the numbers of synaptic spines. Wlsfl/-;Gfap-Cre mice have a modest performance decrement in several complex spatial tasks, including active place avoidance. They were also modestly impaired in one simpler spatial task, finding a visible platform in the Morris water maze. These experiments support a role for DGCs in enhancing spatial learning and memory.