In toto imaging of early enteric nervous system development reveals that gut colonization is tied to proliferation downstream of Ret

The vertebrate enteric nervous system (ENS) consists of a series of interconnected neurons and glia that form nerve plexuses spanning circumferentially within the muscle walls of the entire gastrointestinal (GI) tract. As the largest component of the peripheral nervous system, the ENS enables the GI tract to perform essential life functions such as peristalsis, gut hormone secretions and water balance during gut homeostasis (Furness, 2006). The ENS consists of various enteric neuron and glial cell types, which are classified based on molecular, electrophysiological and morphological means. In humans, along the entire length of the ENS, spanning from the esophagus to the anus, the ENS contains upwards of 600 million neurons, which together are capable of autonomous reflex activity (Furness, 2006), separate from the central nervous system (CNS).

In jawed vertebrates, the ENS is largely derived from neural crest cells (NCCs), a migratory and multipotent stem cell population that arises along the dorsal neural tube during neurulation within the vertebrate embryo. Specifically, the majority of the ENS is derived from NCCs that originate from a post-otic anatomical area along the neuraxis known as the vagal region (Harris and Erickson, 2007; Hutchins et al., 2018; Kuo and Erickson, 2011; Le Douarin and Teillet, 1973; Shepherd and Eisen, 2011). In zebrafish, vagal NCCs emigrate from the vagal region into anterior foregut mesenchyme tissue and migrate posteriorly down the gut length between 36 and 72 h postfertilization (hpf) (Ganz, 2018; Harrison et al., 2014; Nikaido et al., 2018; Shepherd and Eisen, 2011; Uribe and Bronner, 2015). In all jawed vertebrates, once resident in the gut the NCCs are known as enteric neural crest cells (ENCCs) and are characterized by combinatorial expression of the genes Sox10, Phox2b, Ret and Gfra1 (Harrison et al., 2014; Hockley et al., 2019; Howard et al., 2021; Nagy and Goldstein, 2017; Shepherd et al., 2004; Southard-Smith et al., 1998; Taylor et al., 2016; Theveneau and Mayor, 2011). Following their gut colonization, zebrafish ENCCs differentiate into enteric neurons as early as 54 hpf, continuing between 72 and 120 hpf along the foregut, midgut and hindgut (Olden et al., 2008; Olsson et al., 2008). Although the general phases of ENS development and gut colonization have been mapped out, to date we lack comprehensive understanding of the cellular and intercellular mechanisms that orchestrate early ENS pattern manifestation.

Several crucial steps are known to be required for ENS development along the gut – including migration, proliferation and differentiation of ENCCs. Failure of ENCCs to colonize and differentiate into ENS leads to the congenital condition Hirschsprung disease (HSCR), in which variable regions of the gut lack neurons and glia in humans (Amiel et al., 2008; Heanue and Pachnis, 2007). Previous research focusing on the factors that may guide ENCC migration and their gut colonization have made recent progress. For example, different classes of receptor tyrosine kinases (RTKs), and their corresponding ligands, have been identified as crucial components. Among these, the ENCC co-receptor Ret and its mesenchymal-derived ligand, glial-derived neurotrophic factor (GDNF), are necessary for proper ENCC migration and survival along the gut (Landman et al., 2007; Natarajan et al., 2002; Shepherd et al., 2004; Taraviras et al., 1999; Young et al., 2001). Mutations in RET are the leading known cause of HSCR in humans (Amiel et al., 2008). Accordingly, due to its highly prevalent involvement in HSCR manifestation, Ret has been the primary gene of interest for many studies, mutations in which have previously been shown to induce HSCR-like hypoganglionosis and aganglionosis, in both zebrafish and mice models (Heanue et al., 2016; Uesaka et al., 2008). However, how Ret regulates ENCC mechanisms such as proliferation, differentiation and cell type specification in vivo remains poorly understood due to the highly dynamic development of ENS, GI inaccessibility of traditional amniote ENS models, as well as the ethical and technical limitations of investigating human prenatal development.

Despite advancements in our understanding of how early ENS develops, we still do not fully understand the tissue-wide strategies that ENCCs use to sculpt the ENS. Previous work in zebrafish (Harrison et al., 2014; Kuwata et al., 2019) has contributed to a growing body of work that implicate ENCC proliferation as key components of cell migration during colonization (Barlow et al., 2008; Peters-Van Der Sanden et al., 1993; Simpson et al., 2007; Young et al., 2004). However, we still do not fully understand the temporal dynamics of ENCC proliferation and migration during the entire process of colonization, and how those mechanisms are regulated by Ret. Here, using zebrafish larvae, the goal of this study was to resolve and quantify the cellular mechanisms of ENCC migration and proliferation in the context of Ret loss of function, in order to inform our understanding of how the ENS is constructed during early development.

To capture the entire process of gut colonization in vivo, time-lapse confocal microscopy was used to image larval zebrafish guts in toto and track −8.3phox2bb:Kaede⁺ cells, which faithfully labels ENCCs (Harrison et al., 2014; Howard et al., 2022). All ENCCs were tracked, including new ENCCs generated from cell division. ENCC migration was followed from the foregut to hindgut for 48 h, between 48 and 96 hpf. (Fig. 1A-C; Movie 1). At 48 hpf, −8.3phox2bb:Kaede⁺ ENCCs emigrated into the foregut, and migrated posteriorly until reaching the hindgut boundary at 72 hpf (Fig. 1A,B; arrowheads). Between 72 and 96 hpf, ENCCs increased in number and began patterning into nascent ENS along the gut tube (Fig. 1B,C).

Imaris image analysis software allowed us to manually track the generations of every ENCC during the 48 h time period (Fig. 1D; Movie 2). Tracking including newly generated ENCCs produced by cell division, as shown in spatial cell tracks along the gut tube (Fig. 1D; Movies 3,4) and in an enteric generation tree over time (Fig. S1). Imaris tracking produced unique cellular generation labels that denoted ENCC hierarchical order within their respective generation tree, as schematized in Fig. 1E-G. As a cell divided along the anterior-posterior gut axis, the anterior-most daughter cell retained the unique ID and generation label of the original parent cell, whereas the posterior daughter cell was given a unique ID with a generation label corresponding to +1 to that of the parent cell that gave rise to the daughter (Fig. 1F,G). This imaging and generation analysis pipeline showed that zebrafish ENCC populations can be followed at single cell resolution over time and sets the stage to scrutinize gut-scale events during normal and abnormal early ENS development.

In order to study the cellular mechanisms underlying ENS formation, and inform studies aimed at understanding cellular basis for neurocristopathy manifestation, the evolutionarily conserved gene ret was targeted for CRISPR-Cas9 mutagenesis. To accomplish this, a single guide RNA (sgRNA) targeting exon 8 of zebrafish ret (Fig. 2A) was generated and injected along with cas9 mRNA into −8.3phox2bb:Kaede embryos (Fig. 2B). The mutation was isolated and identified by outcrossing injected F0 −8.3phox2bb:Kaede fish to AB wild type (WT) to produce F1 families (Fig. 2B). F1 fish were fin clipped for genomic DNA (gDNA) isolation that was then sequenced; this identified an 11 base pair (bp) deletion resulting in a predicted premature stop codon within the coding sequence (CDS) (Fig. 2C), which is designated hereafter as ret^wmr1. Following in-cross of ret^wmr1/+ F1s, F2 were seen to display either: WT (ret^+/+) (Fig. 2D), hypoganglionosis (HSCR-like) (ret^wmr1/+) (Fig. 2E) or total aganglionosis (ret^wmr1/wmr1) (Fig. 2F) phenotypes. HSCR-like phenotypic fish were found in less than 50% of F2 progeny because of incomplete penetrance and considerable variability in extent of aganglionosis (Table S1), an observation that is congruent with human HSCR manifestation, mammalian HSCR models and other ret heterozygous zebrafish models (Heanue and Pachnis, 2007; Heanue et al., 2016; Lake and Heuckeroth, 2013; Stanchina et al., 2006). To assess whether ret^wmr1 retained any of its function, a rescue experiment was performed by injecting mRNA coding for ret^wmr1 or ret^WT into F2 in-crossed −8.3phox2bb:Kaede;ret^wmr1/+ embryos (Fig. S2). Injection of ret^wmr1 mRNA failed to rescue the number of larvae exhibiting total aganglionosis phenotypes by 96 hpf, whereas ret^WT mRNA partially rescued diseased phenotypes, suggesting functional loss of ret^wmr1 mutant. The successful generation of the ret^wmr1/+ fish line provides a valuable zebrafish model for studying fundamental ENCC dynamics and ENS development.

To examine how the enteric defect phenotypes manifest in the ret^wmr1/+ model, we used the in toto time-lapse microscopy paradigm, as described in Fig. 1. Time-lapse datasets of ret^+/+ (Fig. 3A; Movie 1) and ret^wmr1/+ (Fig. 3B; Movie 5) fish were used to quantify total ENCC numbers over time, between 48 and 96 hpf (Fig. 3C). Although both conditions showed a steady temporal increase in ENCC number, ret^wmr1/+ fish displayed a significant reduction in overall ENCC number; whereas controls displayed an average of 172 ENCCs by 96 hpf, heterozygotes only had 101 (Fig. 3C). In each condition, ENCCs were observed migrating in two parallel chains (left and right) on lateral sides of the gut tube (Fig. 3D,E,H,I; Movies 2 and 6). The distance traveled by the leading-most ENCCs (vanguard cells) on the left and right migratory chain was averaged across replicates and mapped in relation to the foregut region throughout the time course for both ret^+/+ and ret^wmr1/+ conditions (Fig. 3F). In both conditions, vanguard cells in the left and right migratory chains progressed posteriorly at similar rates until they reached their final distance at 96 hpf (Fig. 3F,G). ret^+/+ vanguard ENCCs displaced in a near linear fashion over time until they reached the distal boundary of the gut tube at ∼78 hpf, ∼1000 µm from the foregut, at which point they could migrate no further (Fig. 3D,F-H). In contrast, although ret^wmr1/+ ENCCs migrated down the gut length, they failed to reach the hindgut and migrated only ∼650 µm from the foregut, a significantly shorter migration distance than ret^+/+ vanguard cells (Fig. 3E-G,I). Collectively, these data demonstrate a reduction in overall ENCC number and migratory extent between 48 and 96 hpf in ret^wmr1/+ fish. These results suggest that the collective cell migration and displacement of leading-edge ENCCs may be directly correlated to overall ENCC number.

To investigate the hypothesis that ENCC number is correlated to cell displacement, we aimed to determine how the production of ENCC generations via cell proliferation was related to the overall displacement and spatiotemporal patterning of cells along the gut length. Using previously described generation labels (Fig. 1F,G), the total number of ENCC generations, as well as distances from the foregut of individual ENCC generations, were averaged and plotted versus hpf in ret^+/+ (Fig. 4A) and ret^wmr1/+ (Fig. 4C) conditions. In ret^+/+ fish, ENCCs produced 18 generations between 48 and 96 hpf, where each subsequent generation migrated further posterior than their predecessors in an ordered fashion until the gut tube was colonized (Fig. 4A,B). Although ret^wmr1/+ ENCCs displayed similar posterior patterning, with subsequent generations migrating further than their predecessors, these ENCCs produced only 13 generations, and these failed to colonize the full gut length (Fig. 4C,D). Plotting the average number of cells per given generation revealed that ret^wmr1/+ ENCC generations contained fewer cells, when compared with ret^+/+ (Fig. 4E). These data support a model of proliferation-driven migration, where continual cell division produces daughter cells that migrate further than their predecessors, and suggests that reducing the number of cell divisions reduces their migratory extent.

Based on our observation that cell division was associated with cell displacement over time, we expected to see a reduction in cell migration speed as a function of cell number during this same window of ENCC migration. To quantify advancement rates of the leading-most ENCCs between 48 and 96 hpf, the migratory speed of vanguard ENCCs was averaged across conditional replicates (Fig. 5A,B). A sliding mean curve (every 5 h) was used to visualize average vanguard speed in ret^+/+ (green) and ret^wmr1/+ (blue) conditions (Fig. 5A). Although ENCCs in both conditions displayed stable speeds between 48 and 70 hpf, ret^wmr1/+ ENCCs showed lower speeds (∼21 µm/h) when compared with ret^+/+ (∼34 µm/h) (Fig. 5A). Indeed, box plots comparing the average vanguard speed at intervals between 48 and 96 hpf revealed that ret^wmr1/+ ENCCs migrated significantly slower than ret^+/+ specifically during the 48-60 hpf interval (Fig. 5B).

In addition to slower speeds, in time-lapse movies we noted that ENCCs localized along the gut tube of ret^wmr1/+ mutants looked qualitatively less densely spaced than ret^+/+ control fish. To measure distances between ENCC along the gut during their migration, the average distance between three nearest ENCC neighbors was calculated and averaged across conditional replicates (Fig. 5C). While the ret^wmr1/+ ENCCs eventually reached a similar average inter-cell distance as ret^+/+ ENCCs by ∼72 hpf (∼24 µm), ret^wmr1/+ ENCCs were found to have greater inter-cell distances between ∼48-72 hpf (∼49 µm versus ∼38 µm), congruent with the migratory window in which ret^wmr1/+ ENCCs displayed slower migration speeds (Fig. 5B). Confirming this observation, box plots comparing the average distance between three nearest neighbors during windows at 48-60 hpf and 61-72 hpf showed that ret^wmr1/+ ENCCs were spaced significantly further apart than their ret^+/+ counterparts (Fig. 5D).

To further investigate the mechanism underlying lower vanguard displacement speed and overall ENCC inter-cell distances, we next sought to quantify ENCC proliferation in situ. Towards this end, a phospho-histone H3 (pHH3) antibody was used on fixed tissue at key time points of 60, 72 and 80 hpf to determine the percentage of ENCCs that were actively proliferating (Fig. 5E) (Gurley et al., 1978; Hendzel et al., 1997; Pérez-Cadahía et al., 2009). From these data, a lower percentage of ENCCs were pHH3⁺ in ret^wmr1/+ fish at 60 hpf compared with ret^+/+ ENCCs, but there was no significant difference at 72 and 80 hpf (Fig. 5E).

Collectively, these above-described data regarding cell speed, inter-cell distances and proliferation highlight an important developmental window between 48 and 72 hpf in which ret^wmr1/+ ENCCs exhibit their most significant deficiencies. These results suggest a model of proliferation-coupled migration, and/or inter-cell distance-dependent migration, in which proliferation-deficient ENCCs fail to reach a cell number threshold needed to sustain posterior displacement of ENCCs along the length of the gut (Fig. 5F).

Single-cell sequencing data from our previous work revealed that ENCCs displayed lineage-specific neuronal gene expression patterns at 69 hpf (Howard et al., 2021). Specifically, in addition to pan-neuronal markers, such as elavl3, ENCCs were beginning to differentially express unique combinations of neurochemical gene markers such as vipb nos1 and slc18a3a, as well as the intrinsic primary afferent neuron (IPAN)-specific transcription factor pbx3b (Howard et al., 2021). Notably, vipb was found to be expressed in nearly all early differentiating ENCCs, whereas the IPAN marker pbx3b was found in a very small subset of ENCCs, denoting a more specific enteric neuron subtype that arises later during ENS development. Due to the spatiotemporal-specific emergence of these enteric markers, we hypothesized that the timing and levels of their expression would be altered within ret^wmr1/+ fish, indicative of aberrant differentiation timing.

In order to investigate the above hypothesis, hybridization chain reaction (HCR) was used to investigate gene expression patterns of ret^wmr1/+ ENCCs, in comparison with ret^+/+, at the developmental time-points between 60 and 120 hpf (Fig. 6A-J). Owing to its near ubiquitous expression, probes against vipb transcript were used to label differentiating enteric neurons, and pbx3b was used to mark enteric neuron specification into differentiating IPANs. From these analyses, ret^wmr1/+ fish exhibited a higher percent of ENCCs displaying an IPAN gene signature compared with ret^+/+ cells, with the peak discrepancy appearing at 96 hpf (Fig. 6E-G). However, by 120 hpf ret^wmr1/+ and ret^+/+ fish exhibited roughly an equal percentage (∼40%) of ENCCs with IPAN signature within the total population of vipb⁺ enteric neurons, suggesting premature enteric neuron specification rather than altered subtype-specification (Fig. 6H-J). To determine whether pbx3b transcript levels were differentially expressed between the conditions, total mean pbx3b pixel intensity was calculated throughout the entire gut tube tissue to approximate its total expression within the ENS, irrespective of its co-expression with vipb (Fig. 6K). Corroborating our previous pbx3b⁺ and vipb⁺ ENCC cell counts, the data demonstrated that ret^wmr1/+ fish exhibited significantly higher levels of pbx3b expression than ret^+/+ tissue as early as 72 hpf, with the greatest discrepancy occurring at 96 hpf (Fig. 6K). Moreover, although the fraction of pbx3b⁺ ENCCs was higher in ret^wmr1/+ fish at 96 hpf, no correlation was found between the fraction of pbx3b⁺ ENCCs and the extent of aganglionosis in HSCR-like larvae (Fig. S3). These findings suggest that Ret signaling serves to regulate the timing at which ENCCs progress through their differentiation program, and consequentially their proliferative potential.

To determine whether enteric neuron differentiation timing was indeed altered, as suggested by the HCR in situ gene expression data, whole-mount immunohistochemistry was performed (Fig. 7) using antibodies targeting the pan-neuronal marker Elavl3, to label all enteric neurons, and choline acetyltransferase (ChAT), to label cholinergic enteric neurons, including IPANs (Furness et al., 2004; Morarach et al., 2021). Experiments were performed on ret^+/+ and ret^wmr1/+ fish at the developmental time points 96 hpf and 120 hpf to compare fractions of ChAT⁺ neuron protein reactivity among Elavl3⁺ enteric neurons (Fig. 7D,G). Comparable with pbx3b expression fractions (Fig. 6), the percentage of ChAT⁺ cells was found to be significantly higher in ret^wmr1/+ larvae, compared with ret^+/+ larvae at 96 hpf (Fig. 7D); however, they were found to be at near equal levels by 120 hpf (Fig. 7G). Together, these data provide supporting evidence that a reduction in Ret function results in accelerated enteric neuron differentiation.

We have used a zebrafish model to show, for the first time, in toto colonization of the gut by ENCC at single-cell resolution in a vertebrate. Over the course of 48 h of development, we found that several cellular and intercellular events are essential during colonization, including ENCC number maintenance, migratory speed, ENCC spacing along the gut and proliferative expansion. Specifically, we mapped out the generations of all ENCC cells during their migration and found that proliferative expansion is dampened during precise temporal phases along the gut in ret^wmr1/+ mutants, greatly stalling enteric colonization and leading to severe colonic aganglionosis. Moreover, we discovered premature neuronal differentiation signatures in the context of aberrant ret heterozygous background, suggesting that Ret is required for proper timing of neurogenesis of ENCCs during their migration.

Previous studies performed in amniote models have demonstrated the necessity of proper ENCC numbers for gut colonization (Barlow et al., 2008; Peters-Van Der Sanden et al., 1993; Young et al., 2004). For example, in the avian ENS, manual ablation of NCCs at E1.5, before their entrance into gut, was sufficient to reduce the extent of ENCC colonization within the gut, highlighting the importance of proper cell numbers during the early stages of NCC migration (Barlow et al., 2008). Collectively, these studies provide supporting evidence for ‘population pressure’-driven migration, whereby sufficient ENCC numbers are required for sustained migration and subsequent ENS development. Our present work in the anamniote zebrafish corroborates and extends these previous findings in amniotes by demonstrating the association between total ENCC number and displacement of vanguard cells along the gut completely in toto – providing evidence of a conserved mechanism of ENCC migration present among vertebrates. Although the findings of our study appear congruent with studies performed in amniote models, it should be noted that the zebrafish ENS model has limitations when making comparisons with mammalian ENS development. Notably, the contribution of sacral-NCCs to the ENS, which has been extensively shown during mammalian ENS development (Rao and Gershon, 2018), has not been shown in zebrafish to date. In addition, the zebrafish gut anatomy is less complex, lacking a submucosal ENS plexus, which is a crucial component of the mammalian ENS (Wallace et al., 2005).

Our data revealed a developmental window of time during which a critical number of ENCCs are required during gut colonization. Specifically, at 60 hpf, when ENCCs were approximately halfway through the zebrafish gut tube (midgut), we saw a significant reduction in ENCC proliferation in ret^wmr1/+ mutants compared with wild-type, the timing of which coincided with significantly increased inter-cell distances and reduced cell migration speeds. A complementary study performed in E12.5 mouse explants, a stage later in ENS development and comparable with 72 hpf in zebrafish, found that, following ablation of anterior localized ENCCs, isolated vanguard ENCCs posterior to the lesion site migrated at slower speeds, leading to the conclusion that cell number and cell-cell contact were required for proper migratory wavefront speed (Young et al., 2004). Very recently, a study using fixed mouse embryos found that colonization was drastically reduced/delayed in a Ret homozygous background and, upon examination of proliferation, discovered that ENCCs displayed reduced proliferation at E10.5 (Natarajan et al., 2022), which is roughly equivalent to zebrafish 54-60 hpf and in general alignment with our current findings. Our work at present also corroborates the conclusions of Young et al. (2004) by demonstrating a lower cell count correlated with shorter migration displacement and slower speeds. It is possible, in addition to the gut-length observation of decreased proliferation and speeds of ENCCs we saw following loss of ret, that early reduced ENCC numbers before gut entry could contribute to the observed phenotypes, which could be tested in a future study.

A recent study performed using a ret^+/− zebrafish model, that also presents an ENS HSCR-like phenotype, found that reduced ENCC migration speed was observed when a subset of ENCCs were imaged during a window of time during colonization; however, proliferation was not noted to play a role in ENS phenotypes (Heanue et al., 2016). It should be noted that Heanue et al. used BrdU staining to assay proliferation, which labels S-phase of the cell cycle rather than M-phase labeling with pHH3. In addition, Heanue et al. assayed the single 48 hpf time point, potentially missing changes in ENCC proliferation present within their ret^+/− zebrafish model that we observed at 60 hpf. As such, incongruences between our similar HSCR-like ret^+/− zebrafish models are likely attributed to differences in assays performed, as well as time points assayed.

Imaging in toto gut colonization enabled us to spatiotemporally map the location of each ENCC generation throughout the full migratory phase of ENS formation in vivo, an undertaking not achieved in previous studies. From our study, we observed that vast generations of ENCCs colonize the gut. Specifically, we noticed that cell divisions with an axis perpendicular to the anterior-posterior (A-P) axis of the gut tube, produced a growing number of ENCC generations over time, which would colonize increasingly distant domains within the gut with each subsequent generation. Our findings on global ENCC tracing adds to previous focal lineage tracing experiments performed in zebrafish between 3.5 and 4 days postfertilization (dpf) (Kuwata et al., 2019). At those specific times, Kuwata and researchers found that ENCC proliferation occurred along all regions of the gut and corresponded with cell separation and displacement. In addition, Harrison et al. found that between 50 and 57 hpf, ENCCs in zebrafish med24 morphants migrated along the gut tube much slower than controls, a finding that was associated with decreased mitotic rate (Harrison et al., 2014).

Collectively with previous work, our study supports the model of cell number-dependent migration, whereby proliferation is required to maintain optimal inter-cell distances among the population of migrating ENCCs in order to sustain the posterior expansion, an observation congruent to the ‘frontal expansion’ model of NCC migration that was previous postulated following mathematical modeling (Simpson et al., 2007).

Previous work carried out by others, as well as ourselves, found the TALE homeodomain transcription factor encoding transcript, Pbx3, to demarcate the transition of enteric neuroblast to differentiated enteric sensory neuron subtype, IPAN (Howard et al., 2021; Fleming et al., 2020; Morarach et al., 2021). Here, we demonstrate that ret heterozygous mutant background is capable of inducing premature expression of pbx3b during ENCC migration, providing evidence that Ret may serve to regulate the timing of neuronal differentiation among enteric neuroblasts. Although we observed higher levels of pbx3b expression in ret^wmr1/+ ENCCs compared with ret^+/+ controls between 72 and 96 hpf, the levels of pbx3b expression were found to be the same between ret^wmr1/+ and ret^+/+ fish by 120 hpf, suggesting premature neuronal differentiation rather than a change in enteric subtype specification. Collectively, these results argue that the reduction in proliferative expansion of ret^wmr1/+ ENCCs is correlated with the timing of their neuronal differentiation.

Our findings suggest that Ret signaling may function to regulate maintenance of a stem-state in enteric neural progenitors, a function that is likely conserved within the closely related vagal-NCC-derived sympathoadrenal precursors that have been shown to form neuroblastoma in the presence of constitutively active Ret mutations (Li et al., 2019). Ret inhibition has been shown to significantly reduce tumor growth in a neuroblastoma mouse model (Cazes et al., 2014), further supporting evidence that Ret signaling serves to regulate NCC-derived, neural progenitor proliferation. Mutations in RET have also been linked to additional human cancers including multiple endocrine neoplasia 2 (MEN2), papillary thyroid carcinoma (PTC) and non-small cell lung cancer (NSCLC) (Cazes et al., 2014; Li et al., 2019). In the case of these cancers, the RET mutations are found to be gain-of-function in nature, while RET missense and nonsense mutations found to cause HSCR are distinctly loss-of-function (Takahashi, 2001). In cancer, we see RET associated with tumorigenesis, whereas in HSCR we see a reduction in overall neuroblast number, collectively suggesting RET functions to regulate cell number.

Recently, a functional link between Pbx3 and Meis1 has been discovered during leukemia progression, where Pbx3 and Meis1 directly interact to promote the stability of Meis1, together co-driving tumor formation from transduced primary cells (Garcia-Cuellar et al., 2015). Furthermore, Garcia-Cueller et al. noted that overexpression of Pbx3 led to a dramatic increase in Meis1 transcription, suggesting genetic interactions. In the developing nervous system, Pbx and Meis factors are well-known regulators of neurogenesis and, thus, their prolonged/enhanced expression could tip the balance between proliferation and differentiation, which remains to be tested in the ENS. The findings of this study implicate Pbx3 signaling downstream of Ret and highlight an area for continued investigation.

All work was performed under protocols approved by, and in accordance with, the Rice University Institutional Animal Care and Use Committee (IACUC). Zebrafish (Danio rerio) Tg(−8.3phox2bb:Kaede)^em2Tg (Harrison et al., 2014) adults were maintained as outcrosses with AB WT in our in-house zebrafish facility at 28.5°C on a 13 h light/11 h dark cycle. Tg(−8.3phox2bb:Kaede);ret^wmr1/+ fish line was maintained as AB outcrossed F1 adults and F2 embryos were collected following in-cross and processed identically to WT embryos. F2 adults were in-crossed. In the resulting embryos the hpf was carefully recorded, and the embryos were sorted for Kaede⁺ individuals. All embryos were cultured in standard E3 media until 24 hpf, then transferred to a 1×1-phenyl 2-thiourea (PTU)/E3 solution (Sigma-Aldrich, P7629) (Karlsson et al., 2001) to prevent the formation of melanin pigment.

A sgRNA targeting exon 8 of ret (GAGACAGCGAGGAGACTGTG) was designed by manually searching the ret CDS (ENSDARG00000055305.7) for protospacer adjacent motifs (PAM) within CDS domains. The generation of sgRNA and generation of cas9 mRNA were based on previously described work (Clements et al., 2017; Gagnon et al., 2014). Briefly, these methods included the use of MEGAscript in vitro transcription kit (Invitrogen, AM1330) to generate cas9 mRNA from a pCS2-nls-cas9 vector (Jao et al., 2013) and the sgRNA from a custom DNA oligo containing an SP6 promoter (Gagnon et al., 2014).

Zebrafish embryos obtained from in-crossing Tg(−8.3phox2bb:Kaede)^+/− adults were injected through the chorion at the one-cell stage with a cocktail containing phenol red dye (1:8 µl), 150 pg cas9 mRNA and 40 pg sgRNA targeting exon 8 of ret (GAGACAGCGAGGAGACTGTG) (as described above). Injected F0 embryos were raised to 5 dpf, and screened for ENS defects in which regions of the intestine lacked Kaede⁺ cells in order to validate successful CRISPR-Cas9 activity (Fig. S4). Injected embryos exhibiting WT phenotype were collected and raised in our in-house zebrafish facility, and when reaching sexual maturity were outcrossed to AB WT. Resulting F1 embryos were screened for WT phenotype at 5 dpf and raised to sexual maturity. Resulting F1 adults were fin clipped and genotyped (Meeker et al., 2007) using forward (GCTATGCGGAATGCAATAGC) and reverse (AATCCTGAGGACAGATGGAG) primers to produce 561 bp amplicon from the ret locus. F1 heterozygous adults harboring identical 11 bp deletions (as described in results) in ret were designated ret^wmr1/+ fish line. Functional loss of ret^wmr1 was validated by injecting 30 pg of ret9^wmr1 and ret9^WT mRNA into F2 embryos obtained from an in-cross of the ret^wmr1/+ fish line. The ret9 isoform has previously been shown to be sufficient to drive ENS formation in zebrafish (Heanue and Pachnis, 2008). Injected F2 embryos were screened at 96 hpf and scored for total number of embryos exhibiting either WT (ret^+/+), hypoganglionosis (HSCR-like) (ret^wmr1/+) or total aganglionosis (ret^wmr1/wmr1) phenotypes (Fig. S2). In order to generate ret9^wmr1 and ret9^WT mRNA, ret^+/+ and ret^wmr1/wmr1 embryos were separately processed for total RNA extraction using standard Trizol/chloroform RNA precipitation. Total RNA extractions were used to generate ret^+/+ and ret^wmr1/wmr1 cDNA libraries using the SuperScript IV kit (Invitrogen, 18091050). Both ret9 (ENSDART00000139237.3) and ret51 (ENSDART00000077627.7) splice variants were amplified from cDNA libraries using forward (GGCTCCTTTCGCTCGAATCA) and reverse (ACACTCAGCTTAATGTAGTTATTGTTGCAC) primers and Phusion PCR (Invitrogen). ret9 and ret51 splice variant amplicons were cloned into pCR-Blunt II-TOPO vector using the manufacturer's instructions (Invitrogen, 45-0245). Zero Blunt TOPO reactions were transformed into Max Efficiency DH5α bacteria (Invitrogen, 18258-012). Bacteria were mini-prepped and screened using restriction digest cloning. Presumptive plasmids containing ret9^wmr1 and ret9^WT cDNA were confirmed following whole-plasmid sequencing by Plasmidsaurus. PolyA sequences were amplified from pCS2+ vector and fastcloned (Li et al., 2011) into TOPO-ret9^wmr1 and TOPO-ret9^WT plasmids using forward (PolyA: CCTTCCATAGGAAGAGCTGTGATCCAGACATGATAAGATAC; TOPO: CAAGCTTGATGCATAGCTTGAG) and reverse (PolyA: CAAGCTATGCATCAAGCTTGGTTAACTTGTTTATTGCAGC; TOPO: ACAGCTCTTCCTATGGAAGG) primers. Fastcloning reactions were transformed, prepped and validated using previously mentioned methods. Sequenced validated TOPO-ret9^wmr1-PolyA and TOPO-ret9^WT-PolyA plasmids were linearized with HpaI restriction enzyme and used to generate ret9^wmr1 and ret9^WT mRNA using mMESSAGE mMACHINE T7 in vitro mRNA synthesis (Invitrogen, AM1344). mRNA synthesis reactions were cleaned using Monarch RNA Cleanup Kit (New England Biolabs, T2040L) and visualized using agarose gel-electrophoresis. Pure ret9^wmr1 and ret9^WT mRNA was titrated and quantified using nanodrop to inject 30 pg using same methods described for CRISPR-Cas9 injections.

F2 embryos from an in-cross of F1 Tg(−8.3phox2bb:Kaede);ret^wmr1/+ fish were manually dechorionated using fine forceps, then mounted in 15 µ-Slide 4 Well imaging chambers (Ibidi, 80427) using 1.0% low melt temperature agarose dissolved in 1× PTU/E3 media. Embedded embryos were then covered in E3 media supplemented with 0.4% Tricane (Sigma-Aldrich, A5040) and 1× PTU/E3 media. Confocal time-lapse microscopy was performed using an Olympus FV3000 and FluoView software (2.4.1.198), using a long working distance 20.0× objective (UCPLFLN20X) at a constant temperature of 28°C using OKOLAB Uno-controller imaging incubator. Four-dimensional time-lapse was performed to capture full gut (in toto) volumes along the x-y-z planes using a tiling method to capture the full-length of the gut in two regions of interest, at ∼524-1314 s intervals for 48 consecutive hours, covering 48-96 hpf of development. ret^+/+ (n=3), ret^wmr1/+ (n=3). Time-lapse datasets were stitched in Cellsens software and were exported and saved as .oir files, then processed further using Imaris, as described below. Fidelity of cell tracking to discriminate individual ENCCs was ensured using an H2A-mCherry marker in alternative fish line, Tg(−8.3phox2bb:Kaede;H2A-mCherry; P.A.B., unpublished) (Fig. S5).

Cell tracking was performed using spots lineage-tracking function in Imaris (version 9.7.2). The position of every individual ENCC body (spot) was tracked across each time point along the x-y-z planes using manual spot detection in the Kaede channel to determine the migratory track and generation of the entire ENCC population per fish (Fig. S1). Every ENCC was labeled manually within the spot lineage tracks by applying a unique ID to denote cell generation as a decimal. Generation IDs were empirically assigned based on assumed asymmetrical division along the A-P axis. That is, as a tracked cell gave rise to two cells by dividing along an axis perpendicular to the A-P axis of the gut tube, the anterior daughter cell retained the ID of the parent cell, while the posterior daughter obtained a new generation ID with a decimal denoted +1 the parent cell. Generations were viewed in the Vantage viewer of Imaris for global visualizations. Vanguard cells were defined as the leading-edge migratory cells along the gut. Therefore, the leader identity may change when the current leader is overtaken by a new cell born by cell division during migration. Total raw data of spots tracking was exported from Imaris and further analyzed for graphical depictions using R studio (version 1.3.959; package ggplot2 version 3.3.2]. For data depicting cell ‘distance from foregut’, distance was defined based on origin reference frames placed in the anterior foregut, at the level of 13-14th somite anterior to the cloaca.

Hybridization chain reaction experiments were performed in accordance with previously described methods (Howard et al., 2021; Ibarra-García-Padilla et al., 2021) (RefSeq IDs: vipb, NM_001114555.1; pbx3b, BC131865.1). Whole-mount immunofluorescence experiments were conducted according to methods previously described (Baker et al., 2019). All F2 ret^wmr1/+ embryos used for fixed tissue experiments were screened for the HSCR-like phenotype in which Kaede⁺ ENCCs had failed to reach the hindgut at 72 hpf, 80 hpf, 96 hpf and 120 hpf. Experiments using 60 hpf fixed tissue were not able to be sorted based on phenotype. For these experiments, individual F2 embryos obtained from F1 ret^wmr1/+ in-cross were fixed in individual wells within a 96-well plate following careful dissection of the head. Head tissue was processed for gDNA in separate 96-well plates with 30 µl NaOH. Heads were boiled at 95°C for 10 min, briefly vortexed, then pH balanced with 3 µl 1 M Tris HCL (pH 8.0) and stored at −20°C. gDNA was used as template for PCR using forward (GCCACGATTTCCAGCTTGTGC) and reverse (ACTGGCATCTCCCTGTTGG) primers that amplify a small, 89 bp segment surrounding the lesion site. PCR product was processed using T7 endonuclease assay (Fig. S6) per the manufacturer's instruction (New England Biolabs, E3321) to identify ret^wmr1/+ embryos, which were then pooled and processed via previously described HCR and whole-mount immunohistochemical preparations. The following primary antibodies were used: mouse monoclonal anti-phospho-Histone H3 IgG1 (Abcam, ab14955, 1:1500), rabbit polyclonal anti-Kaede IgG (MBL international, PM102M, 1:250) and mouse monoclonal anti-HuC/HuD (Elavl3/4) IgG2b (Invitrogen Molecular Probes, A21271, 1:200). The following secondary antibodies were ordered from Invitrogen: Alexa Fluor 488 goat anti-rabbit IgG (A11008, 1:500), Alexa Fluor 647 goat anti-mouse IgG2b (A21242, 1:500) and Alexa Fluor 568 goat anti-mouse IgG1b (A21124, 1:500). Confocal microscopy was performed using an Olympus FV3000 and FluoView software (2.4.1.198), using a long working distance 20.0× objective (UCPLFLN20X).

Statistical analysis was performed in R studio software (version 1.3.959; Package ggsignif version 0.6.0 and Package stats version 4.0.2]. Shapiro-Wilk was used to test normality. For comparisons, normally distributed data were tested using two-tailed unpaired t-test and non-normally distributed using Wilcoxon signed-rank test, *P<0.05, **P<0.01 and ***P<0.001. NS, non-significant (P<0.05). Post hoc power analysis was performed using G*Power 3 and found sufficient statistical power (>0.8) for all tests in which we report statistical significance (Faul et al., 2007).

Imaris image analysis was performed using the Rice University Shared Equipment Authority (SEA) Imaris workstation and we thank Alloysius Budi Utama for his insights and training with this software. We thank Dan Wagner (Rice University) for helpful advice throughout this project. We also thank Thom Yorke and Simon Green for technical assistance.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.200668.reviewer-comments.pdf.