Sox11 genes affect neuronal differentiation in the developing zebrafish enteric nervous system

Abstract

The vertebrate enteric nervous system (ENS) is derived from vagal neural crest cells, which enter the foregut as progenitors that migrate from rostral to caudal to populate the entire length of the gut. Here, we show that transcription factors sox11a and sox11b, zebrafish orthologs of the human sox11 gene, are highly expressed in neural crest cells transitioning from progenitors to differentiating neuronal subtypes. Accordingly, CRISPR-Cas9 depletion shows that loss of sox11 paralogs reduces the number of neurons that express the inhibitory motor neuron marker adcyap1b without affecting cell proliferation or death. Transcription factor footprinting analysis of open chromatin regions identified by ATAC-seq reveals sox11 binding sites in the adcyap1b enhancer. Furthermore, mutational analysis shows these binding sites are required for mediating enhancer-driven reporter expression. Taken together, our results demonstrate an important and previously unrecognized role for sox11a and sox11b in neuronal subtype specification in the developing zebrafish ENS.

Introduction

The vertebrate enteric nervous system (ENS), the largest and most complicated portion of the peripheral nervous system, provides intrinsic innervation to the gastrointestinal (GI) tract and can function independently of the central nervous system. Comprised of a complex network of diverse neurons, the ENS is responsible for regulating gut motility, epithelial secretion, and blood flow. The ENS of jawed vertebrates is entirely derived from neural crest cells, most of which arise from the caudal hindbrain(1). These vagal neural crest cells migrate to and invade the foregut and then migrate along the entire rostrocaudal extent of the gut, differentiating into diverse neuronal subtypes.

Defects in ENS development lead to severe congenital gut motility disorders, such as Hirschsprung’s (HSCR) disease, characterized by the loss of enteric neurons in the terminal regions of the gut. This results in intestinal obstruction and massive distension that causes megacolon, which can be fatal to children without surgery(2–4).

While overall ENS organization is similar between zebrafish and mammals, the zebrafish ENS has a simplified structure, comprised of myenteric neurons but lacking a submucosal layer present in amniotes. Enteric neurons are only observed between the circular and longitudinal smooth muscle layers but not within the adjacent connective tissue(5). In zebrafish embryos, vagal neural crest cells delaminate from the neural tube at 16 hours post fertilization (hpf) and enter the foregut around 32hpf. They then migrate along the rostrocaudal extent of the gut, in two parallel streams along the left and right sides, and reach the most distal regions by the end of 3dpf(6–8). Sequential differentiation into various types of neurons occurs during their rostrocaudal migration. Like amniotes, zebrafish enteric neural crest cells differentiate into various types of neurons, ranging from serotonergic, cholinergic, and dopaminergic to vasoactive intestinal peptide (VIP), substance P (SP), and nitric oxide synthase (NOS)-containing neurons(9). Some factors that regulate enteric neural crest differentiation into these diverse neuronal subtypes have been uncovered, whereas many others remain to be determined. The relative simplicity of the gut structure, accessibility to genetic manipulation, and the clarity for imaging make zebrafish an excellent model system for studying ENS development.

Despite some structural differences, most of the known molecular mechanisms underlying ENS development appear to be conserved across vertebrates. For example, critical factors that regulate ENS development in mice, including the transcription factor sox10, signaling receptor ret, and the homeodomain transcription factor phox2b (paired-like homeobox 2b), are essential for ENS development in both mice(10–13) and zebrafish(7, 14–17). sox10 is required to maintain the multipotency of progenitor cells and is also important for glial cells, pigment cells, and neuronal differentiation in the neural crest lineage. Both ret and phox2b are crucial for the enteric lineage choice, and phox2b is a marker for both neural crest-derived enteric precursors and ENS neurons. Mutations of all three factors result in severe intestinal aganglionosis(6, 10, 13, 18, 19).

Previous studies have revealed important roles for multiple transcription factors in ENS development(20), in processes ranging from regulation of neurogenesis to neuronal lineage specification. Knockdown of transcription factors like hand2 and hoxb5 causes loss of ENS neurons in both zebrafish(21, 22) and mouse(23, 24), whereas pbx3 appears to function as a cell fate switch in the ENS, since its depletion decreases the number of Calbindin-expressing neurons while expanding vip+/nos1+/gal (galanin)+ neurons(25). Moreover, our recent findings suggest roles for gata3, ebf1a, and satb2 in excitatory, inhibitory and serotonergic ENS lineages, respectively(9). These studies have not only increased our knowledge of transcriptional regulation of ENS neurogenesis, but also demonstrated interesting conservation in transcriptional control of ENS neuronal development across vertebrates.

To glean information about additional transcription factors involved in the differentiation of ENS neuronal subtypes, we have delved deeply into a comprehensive single-cell RNA-seq dataset of 25,000 zebrafish ENS precursors and differentiated neurons from 2 to 6 days post-fertilization (dpf)(9). Focusing on factors that are highly enriched in precursors and neuroblasts from this single-cell RNA-seq dataset of the forming zebrafish ENS, here we identified sox11a and sox11b as highly upregulated during early neurogenesis, consistent with a potential role in neuronal differentiation. sox11a and sox11b are zebrafish orthologs of the human Sox11 gene that belongs to the SRY-related high-mobility group (HMG) box (SOX) transcription factors family. The sox11a and sox11b paralogs share 75% protein sequence identity(26, 27). Here, we show that sox11a/b are expressed in both progenitors and developing inhibitory motor neurons of the enteric nervous system. Moreover, CRISPR-Cas9 mediated F0 knock-out crispants reveal a critical role for sox11 in regulating adcyap1b expression, the zebrafish ortholog of peptide PACAP (pituitary adenylate cyclase-activating polypeptide), by directly regulating its enhancer. These results further our knowledge of the transcription factors involved in specification and neuronal differentiation in the developing ENS and identify sox11 as a key regulator of the inhibitory motor neuron lineage.

Result

Single-cell RNA sequencing reveals sox11a and sox11b transcripts in ENS neuronal progenitors and a neuronal cluster

To identify transcription factors with possible roles in ENS development in zebrafish, we took advantage of a single-cell (sc) RNA-seq of thousands of FAC-sorted phox2bb+ cells, expressed in both precursors and differentiated ENS neurons, from the dissected gut of zebrafish embryos from 2 to 6dpf(9). The UMAP from the scRNA-seq data reveals a combination of enteric progenitor cells and differentiated neurons (Fig.1A). Enteric progenitors are identified based on the co-expression of pan-neural crest marker sox10 and phox2bb, whereas post-mitotic neurons are identified by expression of elavl3 and phox2bb but not sox10. In total, 24 cell clusters were identified based on the gene expression profile, covering the transition process from enteric progenitor to mature neuron that further subdivided into three main subtypes, inhibitory motor neurons, excitatory motor neurons, and serotonergic neurons, based on neuronal marker expression (Fig.S1)(9, 28).

Fig 1. Sox11a and sox11b expression in both zebrafish and chicken developing ENS.

(A) UMAP of single cell (sc) RNA seq of zebrafish ENS development. It covers the transition from progenitor cells to mature neurons in ENS development.

(B) UMAP showing the expression pattern of sox11a and sox11b in zebrafish ENS development, suggesting their roles in affecting neuronal differentiation.

(C) UMAP of scRNA seq experiment on chicken ENS development

(D) UMAP showing the expression pattern of sox11a and sox11b in chicken ENS development

Focusing on transcription factors with a possible role in regulating neurogenesis and/or neuronal differentiation, we found that zebrafish orthologs of the human sox11, sox11a, and sox11b, were highly expressed in cell clusters that appear to represent cells transitioning from progenitors to mature neurons at around 3 and 4dpf (Fig.1B). In addition to overlapping expression at the transition region, sox11a was also expressed in the cluster 13 of the inhibitory motor neuron clusters (cluster 0, 6, 11, 13, 16) at 6dpf, suggesting a possible role in ENS cell fate decisions (Fig.1B, Fig.S1). Moreover, in the developing ENS of chicken embryos at E10(29), sox11 expression was also found in both progenitor and neuronal clusters, suggesting a possibly evolutionarily conserved role (Fig.1C, D).

Sox11a and sox11b are expressed during early neuronal differentiation in ENS

To examine the expression pattern of sox11 in ENS, we performed in situ hybridization chain reaction (HCR)(30) with probes to sox11a and sox11b on isolated zebrafish guts as a function of time. We noted that sox11a and sox11b were expressed transiently in the ENS at ~3 and 4dpf (Fig.2A). At 3dpf, both sox11a and sox11b transcripts overlapped with phox2bb and the proliferation marker, PH3 (Fig.2B), indicating that they marked the progenitor population. Interestingly, the expression of both transcription factors decreased by 5dpf. While sox11a expression remained detectable at low levels, sox11b was undetectable by 5dpf (Fig.2A).

Fig 2. Sox11a and sox11b expression overlap with inhibitory motor neuron clusters.

All staining images were after ‘Max Intensity’ z-stack projection.

(A) Left: Illustration of zebrafish developing ENS separated into foregut, midgut and hindgut. Right: Transient expression of both sox11a and sox11b in gut at 3dpf, and only sox11a expression at 5dpf. Scale bar = 100um.

(B) Overlapping expression of sox11a, sox11b, PHOX2B, and PH3 in the ENS at 3dpf. The white dash box indicates the zoomed region on the right. Arrows indicate cells co-expressing sox11a/b and proliferation marker PH3. Scale bar = 100um, and 50um for zoomed region.

(C) Co-expression pattern between sox11a, sox11b, and inhibitory motor neuron markers, adcyap1b, vip, and nos1 at 3dpf. The white dash box indicates the zoomed region on the right. Arrows point to cells co-expressing sox11a/b and neuronal markers: hollow arrows-sox11a+/sox11b+/adcyap1b+/nos1+/vip- cells, white arrows-sox11a+/sox11b+/adcyap1b-/nos1/vip+ cells, orange arrows-sox11a-/sox11b+/adcyap1b+/nos1+/vip- cells, purple arrows-sox11a+/sox11b+/adcyap1b+/nos1+/vip+ cells, blue arrows-sox11a+/sox11b-/adcyap1b+/nos1+/vip+ cells. Scale bar = 100um, and 50um for zoomed region.

(D) Co-expression pattern between sox11a and only vip at 5dpf. The white dash box indicates the zoomed region on the right. Arrows indicate cells co-expressing sox11a and vip. Scale bar = 100um, and 50um for zoomed region.

To probe the relationship between sox11a/b and specific neuronal subtypes, multiplex HCR enabled examination of the co-expression of sox11a and sox11b with other neuronal markers. The scRNA-seq analysis of zebrafish developing ENS revealed three neuropeptide markers, adcyap1b, nos1, and vip, in inhibitory motor neuron clusters. While adcyap1b was expressed in all five clusters, nos1 was expressed in cluster 0, 6, 11, 16, and vip was expressed in cluster 11, 13, 16 (Fig.S1). Since sox11a expression was only found in cluster 13 at 6dpf, we co-stained guts with both vip and nos1 to examine their cell type-specific expression. At 3dpf, when the ENS is mainly populated by progenitor cells, sox11a, and sox11b were co-expressed with both nos1 and vip (Fig.2C). However, at 5dpf, when many cells in the ENS have already differentiated into mature neurons, the expression of sox11a overlapped only with vip (Fig.2D). This correlated with the inhibitory motor neuron cluster observed from our scRNA-seq data and indicated that sox11a and sox11b were expressed differentially among inhibitory motor neuron clusters only at a later stage. In addition, no overlapping expression of sox11a and nmu, a neuronal marker of excitatory motor neurons(9), was found at 5dpf (Fig.2D). Taken together, the HCR analysis suggests specific expression of sox11a and sox11b in both precursors and a subset of inhibitory motor neurons. In addition to ENS expression, sox11a and sox11b are expressed in the brain as seen with whole mount staining (Fig.S2).

Loss of sox11a and sox11b results in a significant reduction in inhibitory motor neurons and gut motility

To interrogate the function of sox11a and sox11b in ENS development, we used CRISPR-Cas9 mediated gene knock-out. Given their overlapping expression, we concomitantly knocked out both sox11a and sox11b by injecting six guide RNAs (gRNAs), three for each paralog(31). Sibling embryos were injected with gRNAs to the tyrosinase gene as controls. Nanopore sequencing validated the CRISPR efficiency for each gRNA (Fig. S3, Document. S1)(32). At 4.5 and 6.5dpf, we dissected the gut from both the experimental and control groups. In these crispants, we observed no noticeable morphological differences between the two groups, with the exception of a slight delay in overall development, as indicated by slow formation of the swim bladder (Fig.S4). The phenotype of sox11 knockout fish was examined using various ENS markers. During ENS development, the down-regulation of sox10 is coupled with the upregulation of elavl3 as an indicator of neuronal maturation. Therefore, to examine the neuronal maturation process, we used antibodies to Phox2b (for neuronal precursors and all ENS neurons), HuC/D (for mature neurons), together with HCR probes to sox10 (to mark ENS precursors) in 4.5dpf and 6.5dpf gut. This enabled investigation of progressive gut development at an early and later stage.

By examining the expression of each marker in both the experimental and control groups followed by cell counting, we quantitatively characterized the effect of sox11a-sox11b double knockout on overall neurogenesis during ENS development. At 4.5dpf, we noted a reduction in both HuC/D and Phox2b expression in the experimental group. There was a significant decrease in the hindgut region for both markers (mean of the relative count of experimental group/control group: Phox2b in hindgut = 0.72/1, with p= 0.019; HuC/D in hindgut = 0.62/1, with p= 0.010; N_sox11a/b = 11, N_tyr = 14, replicates = 3) (Fig.3A, C), while no obvious difference was observed in the sox10 signal (Fig.S5A). This suggests that the decreasing number of mature neurons was not due to ENS cells being arrested at the progenitor cell stage. Instead, the reduction in cells expressing HuC/D and Phox2b in the hindgut region suggested a possible delay in ENS development in the double-knockout embryos.

Fig 3. Quantification analysis reveals differences between sox11 knock-out and control embryos.

All staining images were after ‘Max Intensity’ z-stack projection.

(A) Antibody and HCR staining comparing the expression levels of Phox2b, HuC/D and adcyap1b in sox11 knock-out embryos (Top) with the control embryos (Bottom) at 4.5dpf. On the right, hindgut regions that showed a significant difference in cells expressing Phox2b and HuC/D were zoomed in. A significant difference in adcyap1b expression was observed throughout the whole gut. Scale bar = 100um, and 50um for zoomed region.

(B) Antibody and HCR staining comparing the expression levels of adcyap1b, nos1, and vip in sox11 knock-out embryos with the control embryos at 6.5dpf. On the right, corresponding regions that showed a significant difference in cells expressing nos1 and vip were zoomed in. A significant difference in adcyap1b expression was observed throughout the whole gut. Scale bar = 100um, and 50um for zoomed region.

(C) At 4.5dpf, an overall developmental delay was observed in sox11 knock-out embryos. (N_sox11a/b = 11, N_tyr = 14, replicates = 3)

(D) At 6.5dpf, a specific reduction of the inhibitory motor neuron cluster was observed in sox11 knockout embryos, particularly in adcyap1b expression. (For PHOX2B and HuC/D, N_sox11a/b = 7, N_tyr = 7, replicates = 3; For other neuronal markers, N_sox11a/b = 8, N_tyr = 12, replicates = 3)

(E) A significant increase of incoherent gut movement was found in sox11 knockout embryos. (N_sox11a/b = 7, N_tyr = 8)

(F) Kymographs showing the passage of coherent contraction from proximal to distal in tyr knockout embryos, and little movement at the length of the gut in sox11 knockout embryos. Arrows indicate gut contraction.

Interestingly, at 6.5dpf, we noted no significant difference in the numbers of Phox2b and HuC/D expressing cells (N_sox11a/b = 7, N_tyr = 7, replicates = 3) (Fig.3B, D), suggesting that sox11 knockout specifically affected neuronal differentiation, rather than altering migration of vagal neural crest cells into the hindgut. By staining with antibodies to Caspase 3 and PH3, we found that the knockout of sox11 did not alter cell apoptosis or proliferation (Fig.3D, Fig.S5B, C).

Next, we performed HCR for enteric neuron subtype markers, including adcyap1b, nos1, and vip for inhibitory motor neurons, nmu and slc18a3a for excitatory motor neurons, and tph1b for serotoninergic neuron (Fig.2C, Fig.S1). At 4.5dpf, we noted a significant reduction in adcyap1b expression throughout the length of the gut (mean of the relative count of experimental group/control group: adcyap1b in foregut=0.69/1, with p=0.017, midgut=0.58/1, with p=6.5E-4, hindgut=0.43/1, with p=7.8E-5; N_sox11a/b = 11, N_tyr = 14, replicates = 3) (Fig.3A, C). This suggests that the effect was not due to a general delay in development but rather specifically due to the loss of sox11.

At 6.5dpf, we found that only inhibitory motor neuron markers were reduced (Fig. 3B, D). Quantitative analysis revealed a significant reduction of neurons expressing adcyap1b, nos1, and vip, with no apparent differences in other neuronal markers (Fig.3D). Interestingly, we noted some region-specific differences, with significant loss of nos1 expression in the foregut and midgut regions, and vip expression in the midgut and hindgut regions, while adcyap1b was reduced throughout the gut (mean of the relative count of experimental group/control group: nos1 in foregut=0.81/1, with p=0.039, midgut=0.81/1, with p=0.038; vip in midgut=0.68/1, with p=2.52E-3, hindgut= 0.45/1, with p=1.84E-4; adcyap1b in foregut=0.29/1, with p=8.71E-7, midgut=0.58, with p=1.47E-4, hindgut=0.66/1, with p=1.72E-3; N_sox11a/b = 8, N_tyr = 12, replicates = 3). Taken together, the expression level of neuronal markers belonging to the inhibitory motor neuron clusters showed the most apparent phenotype (p<0.001), suggesting that sox11 has a specific effect on inhibitory motor neuron differentiation, particularly on adcyap1b expression. The effects of loss of sox11 appeared to begin during neuronal differentiation and last until a fully functional ENS should have formed.

To determine whether sox11a and sox11b function independently or redundantly, we separately performed loss of function of sox11a or sox11b, using three guide RNAs (gRNAs) per paralog. As controls, embryos from the same pair of parents were injected with gRNAs to the tyrosinase gene. The guts of F0 crispants larvae were dissected at 6.5 dpf and stained for adcyap1b, vip and nos1, three neuronal markers for the inhibitory motor neuron cluster (Fig S6). In sox11b single knock-outs, we noted a significant difference in adcyap1b expression throughout the gut. In contrast, no significant difference was found in sox11a single knock-out compared with controls. The finding that sox11b single knock-out results in a less severe phenotype than the double knock-out (Fig.S6B) suggests that sox11a/b may function redundantly or compensate for one another, with sox11b being the dominant factor.

To test physiological consequences in our crispants, we examined the effects of loss of sox11a-sox11b on gut movement. Interestingly, double knockout embryos showed a significant increase in the percentage of incoherent gut movement at 10dpf (N_sox11a/b = 7, N_tyr = 8) (Fig.3E, Fig.S7B). Coherent transmission of gut contraction was observed along the length of the whole gut of tyr knockout embryos, whereas most gut motility was lost in the sox11a-sox11b double knockout embryos in the most extreme cases (Fig.3F, Video S1,S2).

While making stable transgenic lines would be optimal, 95% of sox11a and sox11b double knockout embryos died at ~ 12dpf due to the critical nature of the genes in ENS development. However, crispants are frequently used as an alternative to mutant lines(33–35), given that they have a permanent loss of the gene of interest and phenocopy the expected mutant phenotype. Nevertheless, the mosaic nature of crispants resulted in a broad variety of phenotypes, which could be resolved with a stable knockout line.

An adcyap1b enhancer contains potential sox11a and sox11b binding sites

The phenotype of sox11a-sox11b knockout suggests a role for sox11a/b in ENS development, specifically in neuronal differentiation. To understand the potential regulatory mechanisms underlying this effect, we next performed bulk ATAC-seq on phox2bb:kaede fish at 3dpf with the goal of identifying enhancers expressed in the ENS. To this end, separating our samples into phox2bb+ and phox2bb− cells and comparing the differences between groups enabled the identification of open chromatin regions that reflect putative enhancers in ENS cells.

To identify possible enhancers that mediated expression of genes characteristic of inhibitory motor neurons, we first looked for open chromatin in the region of the adcyap1b gene body. We identified a positive peak around 32kb downstream of the transcription starting site (Fig.4A). This putative enhancing region was cloned into a GFP-reporter plasmid to test its ability to mediate GFP expression after injection into one-cell-stage zebrafish embryos(36) (Fig.4B). The result revealed that this regulatory region drives GFP expression in the hindbrain starting at 3 dpf (Fig.S8A), reminiscent of adcyap1b transcript expression(37), suggesting it is a putative adcyap1b enhancer element. In addition to the hindbrain, embryos injected with the putative enhancer exhibited mosaic expression in the ENS that overlapped well with adcyap1b at 4.5 dpf (Fig.4C). Enhancer-driven expression also overlapped with that of sox11a and sox11b (Fig. S2).

Fig 4. Putative sox11 binding site in an adcyap1b enhancer.

All staining images were after ‘Max Intensity’ z-stack projection.

(A) TOBIAS provides Tn5 bias correction of bulk ATAC seq results and sox11 footprinting.

(B) The adcyap1b enhancer with either the intact or mutated sox11 binding site was cloned into the GFP reporter plasmid.

(C) At 4.5dpf, embryos showed adcyap1b enhancer mediating GFP reporter expression in the gut. The white dash box indicates the zoomed region on the right. Scale bar = 100um.

(D) At 4.5dpf, embryos showed a reduction in adcyap1b enhancer-mediated reporter expression after mutation of sox11 binding sites. The white dash box indicates the zoomed region on the right. Scale bar =100um

(E) Enhancers with mutated sox11 binding site showed a significant reduction in the percentage of positive GFP signal in ENS. (ENS, N_adc = 60, N_mut = 51, replicates = 3; overlapping, N_adc = 17, N_mut = 16, replicates = 3)

(F) Antibody and HCR staining comparing co-exprssion of GFP signals, HuC/D and adcyap1b in ENS of Tg(+32kb-adcyap1b:GFP) at 4.5 dpf. The white dash box indicates the zoomed region on the right. Hollow arrows point to GFP+/HuCD+/adcyap1b+ cells, white arrows point to GFP+/HuCD+/adcyap1b- cells, and blue arrows point to GFP+/HuCD-/adcyap1b- cells. Scale bar = 100um, and 50um for zoomed region.

Next, we applied TOBIAS (Transcription factor Occupancy prediction By Investigation of ATAC-seq Signal) to correct Tn5 bias and conduct transcription factor footprinting(38) (Fig.S9). We looked for sox11 binding sites within open chromatin regions in the vicinity of the gene body. In this way, we identified a putative sox11 binding site in the enhancer of adcyap1b (Fig.4A). The site was mutated by replacing the entire site with a random sequence, checking to make sure the random sequence did not generate a new transcription factor binding site (Fig. 4B). After mutating the predicted sox11 binding site we examined the pattern of GFP expression in ENS. We found that the GFP signal was significantly reduced in ENS of embryos injected with the plasmid containing the mutated binding site relative to the control intact enhancer (N_adc = 60, N_mut = 51, replicates = 3; p < 0.001) (Fig.4D, E). Further examining the overlapping pattern of adcyap1b and GFP signal via co-staining confirmed this significant reduction (N_adc = 17, N_mut = 16, replicates = 3; p < 0.05) (Fig.4D, E). This suggests that sox11 is likely to regulate adcyap1b expression in ENS via direct binding to its enhancer.

Next we established a Tg(+32kb-adcyap1b:GFP) in which we examine enhancer driven expression relative to adcyap1b and HuC/D staining. We noted not all GFP+ cells are adcyap1b+, and not all adcyap1b+ cells are GFP+, whereas most GFP+/adcyap1b− cells are HuC/D+ (Fig.4F), suggesting that the enhancer only regulates a portion of adcyap1b-expressing cells, and is also active in neuroblast cells that have not yet differentiated. This is consistent with our observation of incomplete loss of adcyap1b signal in sox11a-sox11b double knockout embryos (Fig.3). However, a few GFP+ cells show no co-expression of either adcyap1b or HuC/D, which could be due to some small leakiness or background noise of the enhancer reporter line. We also identified three putative enhancers in the vicinity of the gene bodies of vip and nos1 that contain sox11 binding sites, two for nos1 and one for vip. Although GFP signal was observed, no signal in ENS was observed in either (Fig. S10).

ascl1 is co-expressed with sox11 during neuronal differentiation

CryoEM structural analysis of neuclesome-SOX11 complex has suggested that it may act as a ‘pioneer’ factor to open chromatin and initiate the recruitment of other factors(39). To probe potential transcription factors that may bind cooperatively with sox11, we returned to our scRNA-seq data to examine other transcription factors expressed in similar patterns in the ENS that also had binding sites in the same enhancer as sox11 from our bulk ATAC data. We then calculated the percentage of overlapping enhancers containing binding sites for these genes and sox11 throughout the genome. We found sox4 sites, that have highly similar binding motifs to sox11, in more than 70% of enhancers with sox11 binding sites. In addition, the proneural gene ascl1 (mash1) also shared more than 30% of binding enhancers with sox11 (Fig.5A).

Fig 5. Ascl1 has a binding site adjacent to the putative binding site of sox11.

All staining images were after ‘Max Intensity’ z-stack projection.

(A) Sox4 and ascl1 show the top percentage of overlapping enhancers that contain putative binding sites with sox11.

(B) Sc-RNA seq analysis suggested that sox4 was expressed in progenitor clusters, while sox11a, sox11b, and ascl1a were expressed in ENS neuroblastoma.

(C) At 3dpf, ascl1a had an overlapping expression with sox11a/b, especially with sox11b. The white dash box indicates the zoomed region on the right. Arrows indicate overlapping expression between ascl1a and sox11a and sox11b. Scale bar = 100um, and 50um for the zoomed region.

(D) Representative image of embryo with near complete loss of HuC/D after ascl1a knockout. The dotted line indicates the outline of gut. (N_sox4a = 6, N_tyr = 6, replicates = 2) Scale bar = 100um.

(E) Representative image of embryo with near complete loss of adcyap1b after ascl1a knockout. The dotted line indicates the outline of gut. (N_sox4a = 6, N_tyr = 6, replicates = 2) Scale bar = 100um.

(F) Ascl1 has putative binding sites in putative enhancer of adcyap1b. A plasmid was generated with a mutated ascl1 binding site.

(G) Enhancers with mutated ascl1 binding site showed a significant reduction in the percentage of positive GFP signal in ENS. (ENS, N_adc = 60, N_mut = 56, replicates = 4)

With our scRNA-seq data capturing the gene expression profile during the transition from ENS progenitor cells to mature neurons in zebrafish developing ENS, sox4a, the zebrafish ortholog of human sox4, was mainly expressed in the progenitor cells cluster, while ascl1a, the zebrafish ortholog of human ascl1, was expressed together with sox11a and sox11b in the neuroblast clusters (Fig.5B). Since both sox4a and ascl1a are expressed at early stage of ENS development, we used CRISPR-Cas9 to knock out these genes and examined ENS development in F0 crispants at 3.5 dpf, respectively (Document S2, Fig. S11). No significant difference was observed in ENS neurons in the sox4a crispants knockout (Fig. S11A). In contrast, the ascl1a knockout showed a near complete loss of HuC/D signal (N_ascl1a = 6, N_tyr = 6, replicates = 2) (Fig. 5C) together with other specific neuronal markers, including adcyap1b (Fig.5D, Fig. S11C) at 3.5 dpf. In contrast, Phox2b was unchanged (Fig. S11B). This indicates that ascl1a regulates neurogenesis in ENS. This finding agrees with a previous study, where ascl1^−/− caused a significant delay in neurogenesis in ENS development in mice (40), although no significant difference in the Phox2b expression pattern was observed.

To further probe the relationship between ascl1a and sox11a/b in ENS development, we performed HCR co-staining, which revealed overlapping expression of ascl1a, sox11a and sox11b at 3 dpf (Fig. 5E). Furthermore, TOBIAS footprinting analysis reveals a putative binding site for ascl1 in the enhancer of adcyap1b that lies adjacent to the sox11 binding site identified above (Fig. 5F). As ascl1a does not affect sox11a/b expression at 3.5 dpf (Fig. S12), these findings suggest that ascl1a likely cooperates with sox11a/b in regulating adcyap1b expression in the ENS. When we mutated the predicted ascl1a binding site, we noted a significant decrease in GFP signal in the ENS compared to embryos injected with an intact adcyap1b enhancer (N_mut = 56, replicates = 4; p < 0.001) (Fig. 5G). We observed a significant difference in the GFP signal in the hindbrain as well (N_mut = 182, replicates = 4; p < 0.05) (Fig.S8B). Altogether, our findings suggest that ascl1a may function cooperatively with sox11a/b via direct binding to the adcyap1b enhancer in the ENS.

Discussion

In zebrafish, sox11a and sox11b have been shown to be transiently expressed in the central nervous system (CNS) and somites(27). Similarly, in mouse and chick embryos, sox11 is widely expressed in the CNS and peripheral nervous system, including the olfactory epithelium, dorsal root, and trigeminal ganglia, suggesting a possible role in neural development of amniote embryos(41–43). Functional studies of mouse sox11 have demonstrated that it affects neurogenesis in the CNS, including oligodendrocyte differentiation, dendritic morphogenesis, and proliferation(41, 43–46). Sox11 has also been shown to affect neuronal specification in the sympathetic nervous system(47). However, the function of sox11 was not previously examined in the ENS, though it appears in recent scRNA-seq data of chicken and mouse ENS(25, 29).

By probing our zebrafish ENS scRNA-seq dataset, we found that sox11a and sox11b are expressed in the developing enteric nervous system, specifically in the progenitors and early differentiating inhibitory neurons that are characterized by adcyap1b expression (Figure 6A). This is consistent with single-cell RNA-seq data from other organisms. For example, sox11 expression is found in both progenitor clusters and neuronal clusters in scRNA-seq data of developing chicken ENS and in neuroblasts clusters in scRNA-seq data of developing murine ENS, suggesting that it may have a conserved role in vertebrates(25, 29). Using multiplex HCR in situ hybridization, we validated the expression of sox11a/b in the zebrafish ENS and found that it was transiently expressed with the highest expression at 3dpf. At 5dpf, we observed a low level of sox11a expression in vip-expressing neurons, a subcluster of adcyap1b-expressing neurons.

Fig 6. Schematic summary of the results.

(A) Identification of sox11 expression in progenitors and inhibitory neurons via single cell RNA seq, confirmed by HCR. (B) ATAC-seq enables identification of a putative adcyap1b enhancer that contains critical sox11 binding sites.

Our loss of function analysis further suggests a critical role for sox11a and sox11b during ENS development. After CRISPR-Cas9 knockout, we noted an initial delay in the development of ENS, with a significant reduction of Phox2b and HuC/D expression in sox11a-sox11b crispants at 4.5dpf. This is consistent with previous findings in the chicken spinal cord, where sox11 affects neuron maturation via the establishment of pan-neuronal protein expression(44). However, by 6.5dpf, when a mature and functional ENS has developed, we saw no significant difference in Phox2b or HuC/D expression levels in the sox11a-sox11b knockout embryos. Thus, development appeared to have recovered by this time point. Instead, we noted a significant reduction of adcyap1b expression throughout the gut of knockout embryos at 4.5 and 6.5dpf and regional-specific reduction in nos1 and vip expression at 6.5dpf. This suggests that sox11 specifically regulates neuronal differentiation of the inhibitory motor neurons subpopulation of the ENS. The regional-specific phenotype of adcyap1b, nos1, and vip signal could be related to their different spatial distribution in ENS, and suggests that specific subtypes of neuronal clusters are affected by loss of sox11a/b.

Individual knockout of sox11b showed a similar but less severe phenotype compared to the double knockout, suggesting that both genes function redundantly, with sox11b being the dominant factor. Although our loss of function results utilize crispants rather than mutant lines due to lethality after loss of sox11 paralogs, our results reveal greater than 90% gene loss after CRISPR mutagenesis, suggesting that these crispants closely phenocopy the putative mutant phenotype. Moreover, individual sox11a knock-out had no noticeable effect on ENS development, suggesting that stress response or off-target effects are unlikely. Crisper knock-outs enable efficient screening of potential co-factors that may work together with sox11 (e.g. ascl1a or sox4a experiment). Nevertheless, compared to transgenic knockout lines, crispants show larger individual variation and more intra-group inconsistency, as relevant to our studies of quantitation of neuronal quantification and examination of gut motility.

Although sox11 is crucial for multiple aspects of neurodevelopment, our findings differ from previous studies in the developing sympathetic nervous system. PH3 and Caspase 3 immunostaining results showed no difference in cell apoptosis or proliferation between control and sox11a-sox11b knockout embryos. Previous studies have found that sox11 is required for the proliferation of tyrosine hydroxylase-positive cells and that its loss results in a dramatic increase in the rate of cell death in the developing mouse spinal cord(44). In contrast, our results reveal an essential role for sox11 in ENS neurogenesis and differentiation that appears neuronal subtype-specific and independent of cell death.

We identified sox11, sox4 and ascl1 binding sites within the adcyap1b enhancer. Mutational analysis suggests that ascl1 may act cooperatively with sox11 in the developing ENS (Figure 6B). CRISPR knockout demonstrated that the loss of ascl1a resulted in a delay in neurogenesis, consistent with previous reports(40). The proneural transcription factor, ascl1, has been found to play crucial roles in the specification of neuronal progenitor cells and in regulating specific neuronal differentiation(48). For example, in mouse ENS development, in addition to its impact on overall neurogenesis, ascl1 has been shown to selectively affect neuronal subtypes that express Calbindin, Vip, and TH(40). In addition to the previous finding, we noticed a decrease in adcyap1b expression with ascl1a knock-out. Since vip and adcyap1b are mainly co-expressed in the inhibitory motor neuron clusters (Fig. S1), the loss of Vip observed in ENS development in mice could be due to the fate change of the inhibitory motor neuron cluster.

Previous studies have suggested a close connection between sox11 and ascl1 during neurogenesis. sox11 and ascl1 have been suggested to function synergistically in activating neuronal enhancers in cultured inhibitory interneuron-like progenitors(49). In chicken spinal cord development, the expression of sox11 is induced by proneural bHLH proteins, like Ngn2 and Ascl1(50). Interestingly, in some cases, a reverse order has been observed; e.g. in both adrenergic neuroblastoma and zebrafish spinal cord regeneration, sox11 appears to act as an upstream regulator of ascl1(51, 52). Different from both findings, our ascl1a knockout resulted in no significant difference in sox11a/b expression level. Instead, a potential ascl1 binding site was found adjacent to the sox11 binding site in the putative adcyap1b enhancer. Mutating this predicted binding site also resulted in a reduction of GFP reporter expression in ENS. Therefore, we speculate that ascl1 may regulate adcyap1b expression cooperatively with sox11. However, we cannot rule out the possibility that other transcription factors may also be involved in adcyap1b regulation. Taken together, our study demonstrates an essential role for sox11 in regulating neuronal differentiation of inhibitory motor neurons in ENS development and suggests its interaction with ascl1 may be important for regulating the differentiation of inhibitory neuron subtypes.

Methods

Zebrafish lines:

Adult zebrafish (Danio rerio) and larvae were maintained at 28°C, with a 13-hour light/ 11-hour dark cycle. All experiments comply with the regulations of the California Institute of Technology Institutional Animal Care and Use Committee. Transgenic fishlines used in the study were the ABWT, Tg(Phox2bb: mNeonGreen)(9) (Generated in Bronner Lab), Tg(−8.3Phox2bb:Kaede)(53) and Tg(+32kb-adcyap1b:GFP). Tg(+32kb-adcyap1b:GFP) was generated by raising embryos injected with adcyap1b-enhancer reporter plasmid constructed in the study to adult, and then crossed with ABWT. F1 embryos were then used for co-expression staining.

Zebrafish Embryo Fixation for Staining:

Zebrafish embryos at selected stages were euthanized with tricaine (MS222) and placed in Ringer’s solution for deyolking. Deyolked embryos were fixed with 4% paraformaldehyde (PFA) overnight at 4°C. The next day, samples were washed with 1 X PBS-0.5% Tween solution (3 × 5 min) and then with methanol (1 × 5 min). Samples were then stored in fresh methanol at −20 °C for later usage.

scRNA-seq Data Processing

The transcriptomic data of the developing zebrafish ENS was obtained from Gene Expression Omnibus (GEO) database accession number GSE 274407(9). The reads of 5 libraries generated from ENS cells collected at 2dpf, 3dpf, 4dpf, 5dpf, and 6dpf were mapped using Kallisto and bustools programs 81,82 to the reference transcriptome based on zebrafish genome assembly GRCz10, and the gene-cell matrices were generated separately. Scanpy package83 was then used for analysis. We concatenated these 5 datasets and performed quality control (removing cells with fewer than 400 genes captured, with more than 15,000 counts detected and a ratio of mitochondrial genes to total genes ≥ 0.05 and eliminating genes expressed in fewer than 50 cells). We saved the preprocessed data, and with the remaining 24,677 cells, we normalized the data to 10,000 total-counts per cell and log-transformed the data. Then, we identified 1768 highly variable genes by setting the threshold with minimum and maximum means equal to 0.0125 and 3, respectively, and with minimum dispersion equal to 0.5. After regressing out the source of variation from the expression of mitochondrial genes, we performed PCA and visualized the data in two dimensions by running the UMAP algorithm with the parameter settings as n_neighbors = 20 and n_pcs = 40. Cell clusters are further revealed using Leiden algorithm with resolution = 0.8, and 25 clusters in total are discovered. Transcriptomics data of developing chicken ENS was obtained from Gene Expression Omnibus (GEO) database, accession number GSE 242228, and was processed as previously described(29). Sox11 expression in both zebrafish and chicken ENS development was visualized with UMAP.

HCR Multiplex Staining and Immunostaining

Deyolked zebrafish embryos stored in methanol were rehydrated in 50% Methanol/PBS-0.5% Tween solution for gut dissection. Dissected guts were then transferred back to fresh methanol solution. Superfrost slides were prepared with a tape boundary and a rectangle window in the middle. Guts were placed inside the boundary and automatically got stuck on the slide after methanol evaporation. The slide was then ready for multiplex staining.

Each round of HCR staining was performed with HCR v3.0 probes following the zebrafish protocol provided by Molecular Technologies. Additional stripping steps were added after each round of HCR staining to strip off the previous probes before starting the next round. For the stripping step, we washed the slide with in situ hybridization buffer without tRNA (50% formamide, 5 X SSCT, 50 μg/mL heparin, 0.1% Tween20) at 50 °C (4 × 20 min), followed by washes with 1XPBS-0.5% Tween solution (3 × 5 min). Slides were then examined with a confocal microscope to validate that no residue probes were detected before starting the next round of staining. HCR probes for sox11a and sox11b were produced based on publicized protocol(54) and all other HCR probes, hairpins, and buffers were designed and ordered through Molecular technologies (https://www.molecularinstruments.com).

Whole mount HCR staining of ABWT was conducted following the zebrafish protocol provided by Molecular Technologies, with a modified PK treatment time based on the stage of the embryo. (ex. 3min on shaker for 3dpf embryos).

Immunostaining was performed with either HCR-stained samples or samples after the stripping step. Embryos were transferred to 1 X PBS-TritonX-100 (0.5% TritonX-100) solution with washing (3 × 5 min), and blocked with the blocking solution (10% Donkey Serum in 1 X PBS- 0.5%TritonX-100) for at least 1 hour. Samples were then incubated with the primary antibody in the blocking solution (1:100) overnight at room temperature. The next day, samples were washed with 1 X PBS-TritonX-100 (0.5% TritonX-100) solution (8 × 20 min), and incubated in the corresponding secondary antibody (1:1000) in the blocking buffer overnight at room temperature. After washing with 1 X PBS-TritonX-100 (0.5% TritonX-100) solution (8 × 20 min) on the third day, samples are ready to be imaged. Primary antibodies used in the study include mouse anti-Phox2b IgG1 (Santa Cruz Biotechnology, sc-376997), mouse anti-HuC/D IgG2b (Thermo Fisher Scientific, A21271), rabbit anti-Phospho Histone H3 (PH3) IgG (Millipore, 06-570), rabbit anti-active Caspase3 IgG (Biotechne, AF835) antibodies and goat anti-GFP IgG (Rockland, 600-101-215).

Perturbation with CRISPR-Cas9:

Guide RNAs of zebrafish sox11a, sox11b, sox4a, ascl1a and tyr were designed using CHOP-CHOP (https://chopchop.cbu.uib.no/) and ordered from IDT. Assembly of CRISPR-Cas9 complex and F0 injection followed the F0 knockout-single gene V.3 protocol(31). Embryos showing significant developmental deficiency were filtered out from later analysis. Sequences of guide RNAS used in the experiment are: sox11a- 5’AGACGTGGAAACGGACCTAG3’, 5’GGGAGCAAGTCGTCGTCTCA3’, 5’CGAGCTCAAAGTTGGGCTGG3’; sox11b- 5’GGAACTCGAAGTGAGAACCG3’, 5’GCGAGAGACCACAGACACCG3’, 5’GCATGGCGCGAGCATGTACG3’; sox4a- 5’AGGGCGTCGGAGACGGCGCG3’, 5’CTGGCCGGCACCGCGACAGA3’, 5’CCTCTTGATGTGTCCCGATG3’; ascl1a- 5’ATTGGGAACGTGTTCGCGGA3’, 5’GTAAGCGCGGCATTCCAGTC3’, 5’CCGCGTCAAAGCAGGCAAAG3’; tyr- 5’CGTTGGGAAGGTCGGACACC3’, 5’TAACTTCACCATCCCGTACT3’, 5’GATGCATTATTACGTGTCCC3’.

DNA extraction and crRNA efficiency analysis

Ten embryos injected with either experimental or control crRNAs were randmoly selected and collected in one 1.5 ml eppendorf tube at 3 dpf and freezed at −20 °C for at least one day before DNA extraction. For DNA extraction, 50ul of 50mM NaOH was added to the sample, and the sample was vortexed and incubated at 95°C for 15 min. The sample was then cooled to 4 °C, and neutralized with 5ul of 1M Tris-HCl, pH = 8.0. After centrifugation, the supernatant was used for the PCR reaction. For crRNA efficiency detection of sox11a and sox11b, the predicted cutting region was amplified with forward and reverse primers that bind to ~ 150 bp upstream and downstream of the gRNA binding site, respectively. Amplicons were then sequenced with Nanopore sequencing by Plasmidsaurus. Raw fastq data was obtained, and indel mutation efficiency was calculated with CRISPResso2(32). For crRNA efficiency detection of ascl1a and sox4a, we counducted qPCR analysis. We added 80 ul of FastStart Universal SYBR Green Master (ROX) (MilliporeSigma, Catalog number 4913850001) into each sample, mixed well, and then distributed it into a 96-well plate with 12.5 ul of the reagent in each well. Primers designed to amplify actb1 (reference) and three targeted regions of each gene were then added, respectively. Three replicates for each PCR product were conducted. The qPCR program for detecting the efficiency was set as: Step 1, 50 °C for 2 min and 95 °C for 10 min; Step 2, 40 cycles at 95 °C for 15 sec and 57 °C for 1 min; Step 3, melting curve step for detecting the amplification specificity. We used QuantStudio 3 Real-time PCR System to perform the qPCR experiment.

Gut motility analysis

Starting from 6 dpf, both tyr knockout and sox11a-sox11b knockout zebrafish larvae were fed once per day in a petri dish. Their gut movements were recorded as a 80-frame video with a 4s interval between each frame, using Zeiss LSM 980 brightfield channel. The coherent gut contraction were then examined for each embryo using the video, and kymographs were made for better visualization of the process. Kymographs were produced with FIJI-Analysis-Multikymograph.

BULK-ATAC analysis

Zebrafish embryos of the Tg(Phox2bb:Kaede) line are raised to 3dpf, transferred to the Ringer’s solution, and anesthetized with Tricaine (MS222). During tissue dissection, the embryos were deyolked and the head part that had a high level of phox2bb expression was removed. The remaining trunk parts were collected. About 150 embryos were dissected and separately stored in 6 low-binding microcentrifuge tubes on ice. After dissection, the supernatant was removed and 400ul Accumax (Invitrogen, Catalog number: 00-4666-56) solution was added into each tube. The tubes were incubated in the Eppendorf ThermoMixer at 30°C and shaken at 1000rpm for about 25mins for cell dissociation. After the dissociation step, the tubes were chilled on ice, and 1ml of cold HBSS-BSA solution was added to quench the reaction. The solution was passed through a 70um cell strainer into a 50ml conical tube, and an additional 500ul HBSS-BSA solution was used to rinse the tube and subsequently collected into the conical tube. Six tubes of solution were then combined and divided into two conical tubes. The solution was then spun down at 4°C at 1200rpm for 11 minutes. The supernatant was carefully removed and the pellet in each conical tube was resuspended in 1ml HBSS-BSA solution. We then performed fluorescence-activated cell sorting and sorted out Kaede-positive and Kaede-negative cells.

The samples were used for the bulk ATAC-seq experiment, following the previously published protocol(55). The adapter sequence was first removed with Cutdapt(56). Alignment, post-alignment processing and quality control, and peak-calling were performed with Galaxy EU, following its ‘ATAC-seq data analysis tutorial’(57). Over 95% overall alignment rate was achieved for both Kaede-positive and Kaede-negative groups after the alignment step.

TOBIAS Analysis

After obtaining the .bw file for both the phox2bb+ (Kaede-positive) and phox2bb− (Kaede-negative) groups, TOBIAS (Transcription factor Occupancy prediction By Investigation of ATAC-seq Signal) analysis was conducted according to the protocol described on the GitHub page(38). ATACorrect, ScoreBIgwig, and BINDetect analysis were conducted with Vertebrate PFMs (redundant) from JASPAR as the ‘motifs’ parameter. Information on the binding sites was obtained after BINDetect analysis and was used for overlapping binding enhancers search. The putative binding sites of transciprition factor and bulk ATAC result were then visualized with UCSC genome browser.

Overlapping Enhancer Analysis

TOBIAS output was scanned to find transcription factors co-expressed with sox11 in our scRNA-seq data that have binding sites in the putative adcyap1b enhancer. Before the comparison, TOBIAS outputs for all binding motifs of each transcription factor were concatenated. The percentage of overlapping enhancers was calculated by comparing the percentage of shared ‘peak_id’ between the two transcription factors. The analysis was accomplished in Python with the help of ChatGPT 3.0.

Reporter plasmid construction and injection

The identified enhancer of adcyap1b that is 32 kb downstream of the transcription start site of adcyap1b contains the intact binding sites of sox11 (ACAAAG) and ascl1 (GCATCTGCTC), and it was cloned into the E1b-GFP-Tol2 plasmid (Addgene plasmid # 37845)(36) with BglII digestion and Gibson reaction. Primers for cloning this enhancer are: adcyap1b-enhancer 5’TGCAGGAGCATCTTTTAACCCTT3’, 5’TGGATGTGTGTTGGCTCGAT3’. To mutate the binding site of either sox11 or ascl1, random sequence was chosen to replace them, and JASPAR vertebrate database was used afterwards to check that no new binding site was generated with a threshold value equal to that of sox11 or ascl1, respectively. The plasmid was mutated with Gibson reaction: the binding site of sox11 was mutated to CGGTCT, and that of ascl1 was mutated to TGCATATATT. 20ng/ul of constructed reporter plasmid and 25ng/ul of Tol2 transposase mRNA were then co-injected into ABWT embryos at the single-cell stage. Three more putative enhancer adjacent to either the vip or nos1 gene body were also examined, with primers: vip-enhancer 5’gatgggccctcgagaCTCACTCACTTAAAGTTTGAAAGGTAGC3’, 5’tataccctctagagtcgagaGCATTTGAACTGCCGCTCACTG3’, nos1-enhancer1 5’GGCCCTCGAGAGATCGGGGCGGTGTATATGTGTCA3’, 5’TAGAGTCGAGAGATCGGCTATCGAGGCACACTCT3’, nos1-enhancer2 5’gatgggccctcgagaCGTCCAATCACAATCAAGCATTCAAG3’, 5’tataccctctagagtcgagaGCTGTAAACAACAGCTATCTGCTAAC3’.

Reporter GFP assay

Quantification of reporter GFP was conducted at 4.5 dpf with injected wild type embryos. The percentage of embryos showing GFP signal in the hindbrain was first calculated, and these GFP+ embryos were then deyolked and fixed with 4% PFA in PB. The embryos injected with the plasmid containing mutated ascl1a binding site were fixed at room temperature for 3 hours. After washing with 1XPBS-0.5% Tween solution (3 × 5 min), the gut was dissected out and scanned for GFP signal under the confocal microscope. The embryos injected with either the plasmid containing intact adcyap1b enhancer, or the plasmid containing mutated sox11 binding site were fixed at 4°C overnight. The next day, these embryos were washed with 1XPBS-0.5% Tween solution (3 × 5 min), followed by 1 X MeOH (1 × 5min), and stored in fresh MeOH at −20 °C. GFP signal in the ENS was examined after staining with GFP antibody, and overlapping expression of GFP and adcyap1b was examined after co-staining with GFP antibody and adcyap1b HCR probe.

Imaging and image processing

All images of the whole gut of zebrafish embryos were taken with confocal microscopes. Staining images in Fig.2 taken with Zeiss LSM 980. Staining images of loss of function studies were taken with Zeiss LSM 900, and staining images in Fig.4, and 5 were taken with Leica STELLARIS 8. Widefield images of whole tyr knockout, sox11a-sox11b knockout, and GFP enhancer zebrafish embryos were taken with Leica M165 FC, and wholemount staining of sox11a and sox11b was taken with Leica STELLARIS 8. All images were processed with FIJI. Maximum Intensity Projection was conducted, and the brightness and contrast of the image were set to the same for comparison between experimental and control groups. Image alignment from multiplex staining was performed with FIJI plugin ‘HWada-CoordinateShift’. Quantitative cell counting was conducted with FIJI plugin ‘Trainable Weka Segmentation’, followed by ‘Analyze particles’. Quantitative signal area analysis was conducted with FIJI, ‘Analyze’-’Measure’.

Statistical Analysis

Quantification of signal expression pattern in loss of function studies was analyzed by two-tailed unpaired t-test or the Mann-Whitney test, respectively, depending on whether or not the data followed a normal distribution. Quantification of the percentage of GFP positive embryos in the enhancer assay was analyzed with χ² test.

Significance Statement.

We demonstrate the importance of zebrafish sox11 genes in neuronal differentiation in the developing enteric nervous system. Multiplex spatial transcript analysis reveals sox11a/b expression in both progenitors and developing inhibitory motor neurons. Consistent with this, CRISPR Cas9 mediated knockout reveals a critical role for sox11 in regulating genes characteristic of these motor neurons. These results not only further our knowledge of the transcription factors involved in specification and neuronal differentiation within the developing ENS, but also identify sox11 as a key regulator of the inhibitory motor neuron lineage.

Data and Materials Availability:

All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials.