Summary
Gastrointestinal complaints in autism are common and impact the quality of life of affected individuals, yet the underlying mechanisms are understudied. We have found that individuals with mutations in CHD8 present with gastrointestinal disturbances. We have shown that loss of chd8, the sole ortholog of CHD8 in zebrafish, leads to reduced number of enteric neurons and decreased intestinal mobility. However, it remains unclear how chd8 acts during the development of the enteric nervous system and whether CHD8-associated gastrointestinal complaints are solely due to impaired neuronal function in the intestine. Here, utilizing a stable chd8 mutant zebrafish model, we found that the loss of chd8 leads to reduced number of vagal neural crest cells (NCCs), enteric neural progenitors, emigrating from the neural tube and their early migration capability was altered. At later stages, although the intestinal colonization by the NCCs was complete, we found decreased numbers of both NCC-derived serotonergic neurons and serotonin-producing enterochromaffin cells, suggesting an intestinal hyposerotonemia in absence of chd8. Moreover, transcriptomic analyses revealed altered expression of key receptors and enzymes in serotonin and acetylcholine signaling pathways. Next, tissue examination of chd8 mutants revealed thinner intestinal epithelium accompanied by accumulation of neutrophils and decreased numbers of goblet cells and eosinophils. Last, single-cell sequencing of whole mid- and posterior intestines showed a global disruption of the immune balance with perturbed expression of inflammatory interleukins and changes in immune cell clusters. Our findings propose a causal developmental link between chd8, NCC development, intestinal homeostasis, and autism-associated gastrointestinal complaints.
Introduction
Autism Spectrum Disorders (ASD) are a group of heterogeneous diseases, characterized by two core symptoms: difficulties in social communication and interactions, and restricted, repetitive and stereotyped behavior and interests. In more than 80 % of cases, ASD is associated with one or several comorbidities including intellectual disability, head circumference defects (i.e. micro/macrocephaly), facial phenotype, attention deficit hyperactivity disorder, marked sleep dysfunction, and increased rates of gastrointestinal (GI) complaints (constipation, diarrhea, abdominal pain, and/or bloating) 1.The prevalence of the GI symptoms in autism varies greatly depending on data collection and methodological approaches: reports indicate rates ranging from 4.2% to 96.8% 2–4. Despite the increasing awareness of the gastrointestinal complaints in ASD and their impact on the quality of life of the patients and their family, the etiology of these ASD-associated endophenotypes has not been thoroughly studied.
Here, to tackle this challenge, we took advantage of the strong association between mutations in the autism candidate CHD8 (chromodomain helicase DNA binding protein 8) and GI complaints. CHD8 is one of the most frequently found mutated gene in ASD cases (0.21 % of individuals presenting with ASD) 5–11. CHD8 mutations define an ASD subtype (MIM#615032) with 80% of CHD8 cases presenting with GI complaints, of which a total of 60% have recurring periods of considerable constipation followed by loose stool or diarrhea 6,12. We have previously shown that transient knockdown of chd8, the sole ortholog of CHD8 in zebrafish, leads to reduced number of enteric neurons and perturbed GI motility, which is consistent with the constipation periods reported by individuals carrying CHD8 truncating mutations 6. However, it remains unclear how chd8 acts during the development of the enteric nervous system (ENS) and whether CHD8-associated GI complaints are solely due to impaired neuronal function in the intestine.
All enteric neurons and glia are neural crest cell (NCC) derivatives 13. The development of the ENS is conserved between human and zebrafish, although it is simplified in the latter 14,15. In human, the ENS derives from the vagal and sacral NCCs 14. Vagal NCCs provide the majority of enteric progenitors that colonize the entire length of the digestive tract, whereas sacral NCCs generate a small number of enteric progenitors that colonize exclusively the posterior intestine 16. In zebrafish, the sacral neural crest has never been described and the ENS derives solely from the vagal neural crest 15. After leaving the dorsal part of the neural tube, vagal NCCs undergo migration, proliferation, and differentiation to form a functional ENS 15. Here, we combined zebrafish phenotypic analyses and transcriptomic approaches to examine these key developmental processes.
In addition to a fully functional ENS, a healthy gut possesses an efficient intestinal mucosal barrier that ensures an adequate containment of undesirable non-sterile contents present within the intestinal lumen. When the mucosal barrier is compromised, micro-organisms and dietary antigens trigger the innate immune response. In inflammatory bowel diseases (IBD) such as ulcerative colitis and Crohn’s disease, the immune system responds inappropriately to environmental triggers, which causes chronic intestinal inflammation 17–19. Individuals with IBD suffer from abdominal pain and impaired GI transit 20, which are reminiscent of CHD8-associated GI complaints. We thus sought to determine whether the intestinal homeostasis could be affected by chd8 loss.
In this study, we determined the contribution of chd8 to the development of the ENS and to the intestinal homeostasis. First, we found that the loss of chd8 leads to reduced number of vagal NCCs emigrating from the neural tube at 24 hours post-fertilization (hpf). Their early migration capability was altered at 48 hpf. At 5 days post-fertilization (dpf), the intestine colonization is complete in chd8 mutants but the NCC differentiation is perturbed with a decreased number of NCC-derived serotonergic neurons. In addition, we found that the number of serotonin-producing enterochromaffin cells are reduced, suggesting an hyposerotonemia in the intestine of chd8 mutants. These observations are further confirmed by transcriptomic analyses of NCC-derived neurons that showed altered expression of key receptors and enzymes in serotonin and acetylcholine signaling pathways. Second, we determined that the intestinal architecture, itself, is compromised in absence of chd8. We observed thinner intestinal epithelium accompanied by an accumulation of neutrophils and decreased numbers of goblet cells and eosinophils in the intestine, suggesting that the mucosal barrier is compromised when chd8 is absent. Last, single-cell sequencing of whole intestine showed a global disruption of the immune balance in chd8 mutants with perturbed expression of inflammatory interleukins, changes in immune cell clusters, and active pro-inflammatory immune response. Taking our data together, we propose a causal developmental link between chd8, impairment of NCC development, dysregulation of serotonergic pathway, alterations of the intestinal and immune homeostasis, and autism-associated gastrointestinal complaints.
Results
Phenotypic characterization of stable zebrafish mutant line chd8sa19827
We obtained a zebrafish mutant line carrying a truncating mutation in chd8, the sole ortholog of CHD8 in zebrafish. The chd8 sa19827 mutant line carries a truncating mutation in the first coding exon at position c.C667T (p.Glu223*). First, we determined whether the obtained chd8 mutant line recapitulates the morphant phenotypes we have previously observed in zebrafish transient knockdown experiments i.e. macrocephaly and decreased number of enteric neurons 6,29. Utilizing our established readouts 30–32, we confirmed the presence of macrocephaly by measuring the distance between the eyes of wild-type and mutant zebrafish larvae at 5 days post-fertilization (dpf) (Supplementary Fig. 1a). We observed a significant increase of head size in heterozygous chd8 sa19827/+ (mean = 141.7 µm), compared to control chd8+/+ larvae (mean = 133.1 µm) (t-test, p<0.0001) (Supplementary Fig. 1b). In addition to macrocephaly, we also confirmed that the number of enteric neurons is reduced in the chd8 sa19827 mutant line. HuC/HuD immunostaining on wild-type and mutant larvae at 5 dpf showed a significant decrease in the number of enteric neurons in the heterozygous chd8 sa19827/+ (mean = 184.6 cells) and homozygous chd8 sa19827/sa19827 larvae (mean = 156 cells), compared to control chd8+/+ larvae (mean = 242.3 cells) (t-test, p< 0.0001) (Supplementary Fig. 1c, d).
Fewer vagal NCCs emigrate from the neural tube in absence of chd8
In zebrafish, the enteric nervous system (ENS), composed of neurons and glial cells, derives exclusively from the vagal neural crest 15. The observation of decreased number of mature enteric neurons prompted us to ask whether the initial pool of vagal NCC was affected in absence of chd8. We utilized the Tg2(phox2bb:EGFP) reporter line that marks all vagal NCCs including migrating enteric NCCs, and immature and differentiated enteric neurons 33,34.
We scored the number of vagal NCCs emigrating from the neural tube in both chd8 mutant and control conditions at 24 hours post-fertilization (hpf) (Fig. 1a). We observed a significant decrease of the number of NCCs released from the neural tube in chd8 sa19827/+ embryos (mean = 3.458 phox2bb+ cells) compared to chd8+/+ embryos (mean = 9.3 phox2bb+ cells) (Mann-Whitney test, p< 0.0001) (Fig. 1b). We then followed the migration of the enteric NCCs at several time points. At 48 hpf, we determined the position of the front of migration utilizing the somites as morphological landmarks (Fig. 1a). We observed that the position of the front of migration in chd8sa19827/+ embryos was more rostral (between the 2nd and 6th somite), compared to chd8+/+ embryos (between the 4th and 8th somite) (Fisher Exact Test, p-value = 0.01705) (Fig. 1c). To monitor the migration speed of enteric NCCs at later stages, we took time-lapse images of Tg2(phox2bb:EGFP); chd8+/+ and Tg2(phox2bb:EGFP); chd8 sa19827/+ embryos, every 10 minutes, between 50 hpf and 54 hpf. We did not observe any significant difference in the migration speed of vagal NCCs between chd8 mutant (mean = 28.70 µm/h) and control conditions (mean = 30.85 µm/h) (t-test, p= 0.5248) (Fig. 1d). Finally, NCCs from both chd8 mutant and control conditions reached the distal end of the posterior intestine at 72 hpf (Fisher Exact Test, p= 0.1515) (Fig. 1e), which indicated that the migration capability of the vagal NCCs at later stages is not affected when chd8 is absent.
Taken together, our results suggested that key steps of NCC development, specifically induction and early migration, are affected in absence of chd8. Our transcriptomic data confirmed this possibility. We observed a significant downregulation of msx1a, necessary for NCC induction 35, in enteric NCCs from chd8 mutant larvae (Supplementary data 1, log2FC = −6.87, p= 3,31.10−09). We also observed a downregulation of phox2ba, one of the two zebrafish orthologs for PHOX2B, a gene involved in the migration and survival of enteric NCCs 36 (Supplementary data 1, log2FC = −5.02, p= 0,00019). Our data also suggested that the reduced pool of vagal NCCs emigrating from the neural tube is likely the cause of the reduced number of mature enteric neurons observed at later stages.
Transcriptional consequences of chd8 suppression in enteric neurons
Differentiation of the NCC progenitors into neurons is accompanied by gene expression changes. To assess the role of chd8 during neuronal differentiation, we sorted phox2bb-positive neurons from the intestines of chd8 heterozygous mutant larvae and controls at 4 dpf and we generated approximately 344 million reads by RNA-sequencing to monitor changes in genome-wide expression. We performed an analysis of differential expression. Overall, 279 genes were differentially expressed (DE) as a consequence of chd8 suppression (|log2(FC)|> 1 and FDR= 0.05). More genes were upregulated than downregulated (186 vs. 93) (Fig. 2a and Supplementary data 1).
Gene ontology (GO) term enrichment analysis revealed that the GO term “excitatory extracellular ligand-gated ion channel activity” was significantly enriched among the downregulated genes (p= 2.50.10−03) (Supplementary data 1). Moreover, DAVID functional annotation tool showed significant enrichment of genes involved in the “acetylcholine-gated channel complex” and in “acetylcholine binding” (adjusted p= 0.011 and adjusted p= 0.041, respectively) among the downregulated genes. Although not significantly enriched, we also noted that 80 DE genes encode “integral component of membrane” and that 13 DE genes are part of the KEGG signaling pathway “neuroactive ligand-receptor interaction” (Supplementary data 1). We did not observe any significant enrichment among the upregulated genes (Supplementary data 1).
Our transcriptomic data indicated that expression of several genes directly involved in serotonin metabolism (downregulated genes: slc6a19a.2, tph2, htr1d, htr3a; upregulated genes: htr6, aox5) are altered in absence of chd8 (Fig. 2b, 2d and Supplementary data 1). We performed STRING analysis on the full list of DE genes and we generated a full network of the query proteins. The resulting Protein-Protein Interaction Network (PPI) had significantly more nodes than expected (p= 1.36.10−7), which indicated that chd8-regulated genes are biologically connected (Fig. 2b-c and Supplementary Fig. 2). We therefore clustered the genes involved in the PPI network. We found a cluster of 14 genes (downregulated genes: opn4.1, npy2r, gpr37l1b, dennd2da, ptgir, gng13b, tph2, htr1d, htr3a; upregulated genes: pdyn, sstr2a, pyyb, adora2aa, htr6), including four components of the serotonin signaling pathway (tph2, htr1bd, htr3a and htr6) (Fig. 2b), and a cluster of four genes, which included three acetylcholine nicotine receptors (downregulated genes: chrna1, chrna2b and chrna6) (Fig. 2c).
Although the enrichment was not significant, we found 14 DE genes whose human orthologues are referenced in the SFARI database and 74 genes whose human orthologues are associated with an OMIM entry (Supplementary data 1).
We then evaluated whether these transcriptomic findings translate into a possible loss or gain of serotonergic cells in the intestine. To visualize the serotonergic neurons and the serotonin-secreting cells, we performed a double immunostaining against HuC/D and serotonin (5-HT) on chd8 sa19827/+ and control chd8+/+ larvae at 5 dpf (Fig. 2e). We observed a significantly decreased number of serotonergic cells in chd8 sa19827/+ larvae compared to controls (mean = 15.55 vs. 37.79 5-HT-positive cells) (Mann-Whitney test, p-value > 0.0001) (Fig. 2f). Since the number of HuC/D-positive neurons are different between chd8 mutants and controls (mean= 165.3 cells vs. 230.3 cells; t-test, p> 0.0001) (Fig. 2g), we determined the percentage of neurons expressing 5-HT by dividing the number of HuC/D-positive/5-HT-positive cells by the total number of HuC/D-positive cells in both mutant and control conditions. In the controls, the serotonergic neurons represented 4.7 % of the total number of neurons whereas in the chd8 heterozygous mutants, we found only 0.7465 % of serotonergic neurons (Mann-Whitney test, p< 0.0001) (Fig. 2h). Moreover, the number of 5-HT-positive cells that are not neurons (HuC/D negative cells) was also reduced in chd8 mutants compared to controls (mean = 14.27 vs. 26.96 HuC/D-negative/5-HT-positive cells), indicating that the number of serotonin-producing enterochromaffin cells was also reduced in absence of chd8 (Mann-Whitney test, p< 0.0001) (Fig. 2i).
The loss of chd8 alters morphology of the mid- and posterior intestines
To investigate further the consequences of chd8 absence, we evaluated the integrity of the intestine in heterozygous and homozygous chd8 adult mutants. To this aim, we performed histological stainings (i.e. Masson’s trichrome and Alcian Blue/periodic acid Schiff’s base reagent (AB-PAS)) on intestinal cross sections. We focused on the mid- and posterior zebrafish intestines that resemble the mammalian ileum and colon respectively 28,37 (Fig. 3a).
We first measured the thickness of the intestinal epithelium and muscle layers. We performed Masson’s trichrome stain and we observed a significant reduction of the epithelium thickness in chd8 sa19827/+ and chd8 sa19827/sa19827 zebrafish compared to controls in the mid-intestine (Mann-Whitney test, p< 0.0001 and p< 0.0001, respectively) and in the posterior intestine (Mann-Whitney test, p= 0.0018 and p< 0.0001, respectively) (Fig. 3b). The width of the muscle layers was also reduced in heterozygous and homozygous mutants in the mid-intestine (Mann-Whitney test, p< 0.0001 and p< 0.0001, respectively) (Fig. 3c).
Then, we performed AB-PAS staining and we scored the number of mature goblet cells (AB positive cells, indicated by black asterisks on Fig. 3a) and the number of eosinophils (PAS positive cells, indicated by black arrows on Fig. 3a) in the mid- and posterior intestines (Fig. 3d and 3e). First, we did not observe any significant difference in the number of AB positive cells per villus in the mid-intestine for both heterozygous and homozygous conditions, compared to controls. In contrast, the homozygous mutants exhibited a significant decrease of the number of AB positive cells in the posterior intestine (Welch’s t-test, p< 0.0001) (Fig. 3d). We then scored the presence of mucus on intestinal sections and defined four classes: absence of mucus (type 1), presence of mucus on the villi (type 2), presence of mucus in the lumen (type 3), presence of mucus on the villi and in the lumen (type 4) (Fig. 3f). Strikingly, we observed a significant decrease of the presence of the mucus on the villi and in the lumen in heterozygous and homozygous mutants in the mid-intestine (Fisher’s exact test, p= 0.014 and p= 0.00013, respectively) and in homozygous mutants in the posterior intestine (Fisher’s exact test, p= 0.00047) (Fig. 3g).
The eosinophils reside in the intestine and exert homeostatic functions including the maintenance of the protective mucosal barrier that contributes to gut-associated immunity 38. The number of PAS-positive eosinophils was significantly reduced for both heterozygous and homozygous mutant conditions, compared to controls, in the mid-intestine (Welch’s t-test, p< 0.0001 and p< 0.0001, respectively). The number of eosinophils was also reduced for the homozygous mutant condition in the posterior intestine (Mann-Whitney test, p=0.0026) (Fig. 3e).
We hypothesized that intestine architecture changes, including thinning of the epithelium and muscle layers, decreased numbers of goblet cells and eosinophils and decreased amount of produced mucus, could be accompanied by perturbed immune balance in the intestine. To test this possibility, we utilized Sudan Black B (SB) which is a lipophilic dye that integrates into granule membranes and therefore marks mature, granulated neutrophils. We observed a significant increase of the number of neutrophils, indicated by black arrowheads, in the intestinal tissue (t-test, p< 0.0001) in mutant larvae at 15 dpf (Fig. 3h, 3i). We also noticed the presence of SB-positive cell bundles, indicated by red arrowheads, that abut the caudal artery dorsally and the somite muscle limit ventrally, consistently with previous reports 39. Although these SB-positive cell bundles in the caudal hematopoietic tissue (CHT) normally disappear between 7 dpf and 13 dpf in wild-type larvae 39, we still observed a significantly high number of these SB-positive cell bundles (Fig. 3h) in heterozygous mutants compared to controls at 15 dpf (Mann-Whitney test, p< 0.0001) (Fig. 3j). A modest but significant increase of the number of neutrophils is also observed in heterozygous juvenile mutants compared to juvenile controls in the anal region of the posterior intestine at 35 dpf (Mann-Whitney test, p= 0.0033) (Fig. 3k and 3l).
Single-cell sequencing revealed perturbed immune balance in the intestine
Our data indicated that the numbers of eosinophils and neutrophils are changed in absence of chd8. To investigate further the impact of chd8 loss on intestinal immune homeostasis, we collected the mid- and posterior intestines of controls and homozygous mutant adult males and we performed single-cell transcriptomic analyses using 10X Genomics technology. We analyzed a total of 6,339 cells: 3,865 cells for control and 2,474 for homozygous mutant conditions.
Utilizing Seurat R package, 14 cell clusters were identified (Fig. 4a). To determine cell cluster identity, we used known sets of markers from published transcriptomic studies 40–42. For instance, we used enterocyte markers such as fabp2, pck1, and cdh17. T-cells were identified by lck, cd3eap, cd4-1 and cd8a. The expression of tnfsf14 and il2rb defined the NK-like cells cluster and cd79a and cd37 are expressed in B-cells. We used ccr9a, ccr9b and il1b as leucocytic markers. Last, macrophages were identified by spi1b, mpeg1.1, and ncf4 (Fig. 4b and Supplementary Fig. 3).
We first analyzed the clusters by comparing the repartition of the cells in the clusters in mutant and control conditions. We observed a significant difference in the overall repartition of cells in the clusters between chd8sa19827/sa19827 homozygous mutants and chd8+/+ controls (Fisher’s exact test, p= 5.52.10−54). Strikingly, we found that the population of T-regulatory lymphocytes expressing foxp3a in cluster 3 is almost absent in the homozygous mutant condition (Fig. 4a and Supplementary Fig. 3).
Gene ontology (GO) term enrichment analysis on DE genes between chd8 mutants and controls in each cluster revealed that several GO terms associated with innate immune response and inflammation were significantly enriched in chd8 mutants (Fig. 4c and Supplementary data 2). In particular, the GO terms “lymphocyte chemotaxis” (GO:0048247), “lymphocyte migration” (GO:0072676), “mononuclear cell migration” (GO:0071674), “monocyte chemotaxis” (GO:0002548), “cellular response to interferon-gamma” (GO:0071346), and “cellular response to interleukin-1” (GO:0071347) were significantly enriched among the upregulated genes in T-cells and NK-like cells clusters (clusters 2, 3 and 6). Furthermore, the GO terms “chemokine-mediated signaling pathway” (GO:0070098), “cellular response to tumor necrosis factor” (GO:0071356), “cellular response to chemokine” (GO:1990869) were also enriched among the upregulated genes in T-cells and NK-like cells clusters (clusters 2 and 6).
Mitochondria play a part in the regulation of inflammation 43,44. Consistently, we observed that the GO terms “electron transport chain” (GO:0022900) and “aerobic respiration” (GO:0009060) are enriched among the upregulated genes in enterocytes and T-cells clusters (clusters 0, 1, 2, 3, 4, 5, 6, 8). Furthermore, the GO term “mitochondrion” (GO:0005739) is enriched among the upregulated genes in T-cells clusters (clusters 3, 4, 5, 6, 8), and among the downregulated genes in enterocytes cluster (cluster 0).
Interleukins and interferons signaling pathways are instrumental in the activation of the immune response 45. Thus, we asked whether interleukins, interleukin receptors and interferons are differentially expressed between homozygous mutants and controls (Fig. 4d). Strikingly, we found that three interleukins were significantly downregulated among the T-cells clusters: the pro-inflammatory il16 was downregulated in a T-cells cluster (cluster 3), whereas il21 and the pro-inflammatory il34 were significantly downregulated in T-cells cluster 1 and in T-cells clusters 1 and 6, respectively. In addition, the expression of three interleukin receptors was altered in several T-cells clusters. The receptor for the pro-inflammatory cytokine il17a, il17ra1a, was upregulated in the clusters 1 and 2, whereas the receptors for il2, il2rb and il2rgb, were downregulated in clusters 1 and 4, respectively. The interferon signaling pathway was also affected in the homozygous mutants. Specifically, irf1b was upregulated in enterocytes (cluster 0), irf2 was upregulated in the T-cells (cluster 2), and the interferon gamma orthologue ifng1r was upregulated in the T-cells (cluster 3).
Taken together, our data strongly suggested that the innate immunity is activated, possibly due to mucosal barrier breakdown, which ultimately leads to intestinal inflammation when chd8 is absent.
Discussion
Gastrointestinal (GI) problems in ASD-associated neurodevelopmental syndromes are common, however their etiology remains largely unknown. Here, we investigated the role of autism-associated chd8 during enteric NCC development and in the maintenance of gut homeostasis. Utilizing zebrafish, we showed that chd8 acts quite early during the NCC development and that its loss affects the number of enteric NCCs emigrating from the neural tube and their early migration. In mature enteric neurons, chd8 regulates serotonin and acetylcholine signaling pathways. Moreover, we found that numbers of both serotonergic neurons and enterochromaffin cells were reduced in the intestine, indicating that chd8 is essential during differentiation of enteric NCCs into serotonergic neurons and that its loss likely leads to hyposerotonemia in the intestine. Finally, we identified a direct role of chd8 in the maintenance of gut homeostasis. In both juvenile and adult zebrafish mutants, tissue examination revealed compromised intestinal architecture accompanied by accumulation of neutrophils and decreased numbers of goblet cells and eosinophils in the intestine. Single-cell sequencing of whole intestine confirmed a global disruption of the immune balance in the intestine, with exacerbated immune response and drastic reduction of the anti-inflammatory regulatory T-cells.
ASD-associated gastrointestinal complaints: are they neurocristopathies?
Although CHD8 disruption is associated with GI complaints 6, its function during vagal NCC development has never been examined. Here, we showed that chd8 loss affects several steps of vagal NCC development including induction, early migration, and differentiation into enteric neurons. We found decreased number of vagal NCCs emerging from the neural tube at 24 hpf, suggesting a perturbed induction when chd8 is inactivated. This possibility is further supported by our transcriptomic data showing that msx1a, necessary for NCC induction 35, was downregulated in enteric NCCs from mutant larvae. We thus propose that chd8 plays a role in the induction of the vagal neural crest, by regulating, directly or indirectly, the factors of induction. Early intervention of chd8 may be important for the newly delaminated vagal NCC progenitors to proceed to migratory stages. This possibility is in line with recent transcriptomic work on cranial NCCs in mice showing that the complex Chd8/Twist1 controls delaminatory and early migratory markers 46. Contrary to Hirschsprung’s disease (HSCR; MIM#142623), a congenital condition associated with a failure of vagal NCCs to colonize the intestine 47,48, we found that chd8 loss do not prevent the completion of the rostro-caudal colonization of the gastrointestinal tract by the vagal NCCs. Absence of aganglionic segments in the posterior intestine of chd8 mutants further suggests that chd8 loss do not affect drastically the initial NCC-progenitor pool. Our work shows that the etiology of motility disturbances in patients with CHD8 mutations is, in part, due to impaired NCC development but is rather different from neurocristopathies affecting the GI tract such as HSCR.
Loss of chd8 leads to hyposerotonemia in the intestine
NCC differentiation is governed by precise sequence of fate decisions at the right time and place 49. We and others have shown that chd8 regulates gene expression in pathways involved in neurodevelopment, supporting a role for chromatin remodelers in neuronal differentiation 29,46,50,51. However, chd8 function in enteric neurons has never been reported. Therefore, we examined the role of chd8 by establishing the functional genomic effects in enteric mature neurons after reducing its expression to a level comparable to that expected from the heterozygous inactivating mutations found in ASD. Hence, in heterozygous mutant condition, we observed fewer enteric neurons that exhibited dysregulated cholinergic and serotonergic signaling pathways in mid- and posterior intestines.
Acetylcholine is the most common neurotransmitter to induce gastrointestinal smooth muscle contractions 52. We found that three genes coding subunits for nicotinic acetylcholine receptors, chrna1, chrna2b and chrna6, are downregulated in enteric neurons in absence of chd8. Mutations in CHRNA1, and CHRNA6 have been implicated in fast-channel congenital myasthenic syndrome (MIM#608930) characterized by early-onset progressive muscle weakness, and chronic pain 53,54. Of note, decrease of cholinergic signaling in individuals with duplication of CHRFAM7A, that encodes a dominant negative α7-nAChR inhibitor, is associated with inflammatory bowel disease (IBD) 55,56. We propose that absence of chd8 might reduce cholinergic signaling in the intestine which could, in turn, affect contraction capability and alter intestinal transit.
In the nervous system, serotonin (5-HT) is produced either by 2-3% of enteric neurons by the tryptophan hydroxylase TPH2 or by the enterochromaffin cells via TPH1 57–59. Conventional functions of serotonin in the gut involve intrinsic reflexes including stimulation of propulsive motility patterns, epithelial secretion, and vasodilation 60. We found altered expression of several receptors for serotonin in neurons lacking chd8 including htr3a, htr6, and htr1d. The 5-HT3 receptor is known to be involved in intestinal motility 60 whereas the 5-HT6 and 5-HT1 receptors regulate the adenylyl cyclase signaling pathway, which, in turn, regulate the hyper-excitability of neurons 61,62. We also found that slc6a19a.2, coding a carrier involved in the absorption of tryptophan, the precursor of serotonin, and the enzyme tph2 are both downregulated in mutant larvae which indicates that serotonin is likely underproduced by the enteric neurons in absence of chd8. In addition, the numbers of both 5-HT-positive neurons and 5-HT-producing enterochromaffin cells were decreased in the mutant intestines. Our work suggests that chd8 tightly controls the serotonin pathway in both neuronal and non-neuronal 5-HT positive cells. Notably, changes in the number of intestinal enterochromaffin cells and in serotonin production have been observed in patients with IBD as well as in animal models of colitis 63,64. Moreover, people with IBD who experience constipation often have lower plasmatic levels of serotonin 65. Recent work utilizing D. melanogaster indicates that loss of CHD8/CHD7 ortholog, kismet, leads to increased levels of serotonin in the brain and in the proventriculus and the anterior midgut which can be zebrafish equivalents of the intestinal bulb and anterior part of the mid-intestine respectively 66. Our work is in contradiction with this study regarding observed levels of serotonin in the mid-intestine. Here, utilizing a vertebrate model, our data suggested that loss of chd8 likely leads to hyposerotonemia in the mid- and posterior intestines.
Consequence of chd8 loss on mucosal barrier maintenance
The chd8 adult mutants exhibited compromised intestinal architecture. Notably, we observed thinning of the intestinal epithelium and muscle layers, reduced number of goblet cells accompanied by reduced presence of mucus in the intestinal lumen, and decreased levels of eosinophils. Altogether these perturbations likely alter the structure and protective functions of the mucosal barrier. This possibility is further supported by the observed increased number of neutrophils in the intestine of mutant larvae as early as 15 dpf. It is known that in the case of mucosal injury, inflammatory monocytes are recruited into the mucosal wound site after neutrophil infiltration to facilitate recovery of the mucosal barrier 67. Mucosal barrier is constituted by antimicrobial peptides and mucus layer constructed by intestinal epithelial cells. Recently, it has been shown that intestinal mucus layer maintenance depends on eosinophil presence in the lamina propria since eosinophil-deficient mice had significantly decreased numbers of mucus-secreting goblet cells in the small intestine 38. Moreover, muc2-deficient mice, in which the mucus layer is defective, develop spontaneous colitis 68. Decreased mucosal barrier function and neutrophil infiltration are observed in the intestines of patients with IBD 69. Although further research is needed to determine whether chd8 is necessary for the establishment and/or the maintenance of the mucosal barrier, we speculate that patients with chd8 mutations are more prone to bacterial infection and/or colitis due to altered mucosal barrier.
Immune balance is perturbed in absence of chd8
To combat bacterial antigens, intestinal epithelial cells indirectly or directly interact with innate and adaptive immune cells by presenting antigens to dentritic cells or T cells, or by expressing cytokines, chemokines, hormones and enzymes 70,71. Our single-cell transcriptomic data revealed a strong impact on immune cell clusters when chd8 is absent. Strikingly, we found that the population of foxp3a-positive regulatory T-cells (Treg) is reduced in the intestine of adult chd8 mutants. In addition, we observed a significant enrichment for GO terms related to innate immune response such as response to interferon gamma, cellular response to chemokines, lymphocyte and monocyte chemotaxis, cellular response to tumor necrosis factor in T-cell clusters, suggesting an overly active immune response in the intestine when chd8 is absent. Furthermore, we found that the expression of il17ra1a, the receptor for IL-17, is increased in mutants compared to controls. IL-17-producing Th17 lymphocytes and Treg cells represent two arms of an immune response (reviewed in 72). The balance of Th17 and Treg cells is critical for the health of the host. Th17 cells participate in the defense against extracellular bacterial and fungal infections. On the other hand, Treg cells regulate the immune response and maintain immune homeostasis. Excessive activation of Th17 leads to inflammation and autoimmune disease. Of note, increased Th17/Treg ratio is associated with a higher severity of the autistic traits in children with ASD 73. Our findings strongly suggest that chd8 loss leads to perturbed Th17/Treg balance which provokes excessive inflammatory response in the intestine.
Taken together, we propose a model in which chd8 loss induces breakdown of the mucosal barrier which, in turn, drives intestinal vulnerability to infection. As a consequence, the intestine is challenged by bacterial antigens, and innate immune response is activated. Inflammation is subsequently maintained in challenged chd8-mutant intestines due to reduced number of Treg cells and increased IL-17 signaling through its receptor IL-17RA.
Several limitations exist in the present study. First, since we utilized a constitutive knockout chd8 zebrafish line, it is rather difficult to establish cause-effect relationships. However, some of our findings are in favor of co-occurring developmental defects due to pleiotropic effects of chd8. Second, several studies report that individuals with ASD harbor altered gut microbiota 74,75. Although unlikely a disease driver, it will be of interest to investigate whether the microbiota is affected in absence of chd8. Third, we postulated that depleted pool of Treg cells might be unable to restrain IL-17 signaling which leads to persistent and uncontrolled inflammation. However, further studies are necessary to examine, specifically, activity of the Th17 lymphocytes and whether downstream effectors of IL-17RA are activated when chd8 is absent.
Our work aimed to unveil the intricacies of GI complaints in autism. Although some mechanisms remain to be elucidated, our work provide several lines of evidence suggesting that GI complaints in individuals with CHD8 mutations are due to complex interplay between neuronal, epithelial, and immune cells. In the future, it will be essential to pursue the unravelling of the links between ENS development, mucosal barrier, and immune balance and to characterize precisely the etiology of the GI complaints in specific ASD population to determine therapeutic actions.
Methods
Zebrafish husbandry
Zebrafish (Danio rerio) were raised and maintained as described in 21. Adult zebrafish were raised in 15 L tanks containing a maximum of 24 individuals, and under a 14 h-10 h light-dark cycle. The water had a temperature of 28.5 °C and a conductivity of 200 µS and was continuously renewed. The fish were fed three times a day, with dry food and Artemia salina larvae. Embryos were raised in E3 medium, at 28.5 °C, under constant darkness. AB strain was used as wild-type for this study. The mutant line chd8 sa19827, carrying the mutation c.C667T (p.Glu223*), was obtained from the European Zebrafish Resource Center (EZRC #24433), and the w37Tg transgenic line, carrying the construct Tg2(phox2bb:EGFP) was obtained from the International Resource Centre for Zebrafish (ZIRC #ZL1748). Experiments on adult zebrafish were performed utilizing 1-year-old males. Developmental stages of zebrafish embryos and larvae are indicated in the text and figures. For zebrafish embryos and larvae, both males and females were used since the sex can only be determined at 2 months of age. All animal experiments were carried out according to the guidelines of the Ethics Committee of IGBMC and ethical approval was obtained from the French Ministry of Higher Education and Research under the number APAFIS#15025-2018041616344504.
Genotyping of the chd8sa19827 mutant line
Adult fish were anesthetized in 80 µg/mL tricaine. Fin clips were digested in 50 µL of 50mM NaOH for 15 minutes at 95 °C, and the reaction was neutralized by adding 5 µL of 1M Tris-HCl pH7. The genomic region encompassing the sa19827 mutation was amplified by PCR reaction, using the following primers: 5’-GTCAGACTCAAGTGCTGCAG-3’ and 5’-GACACTTTGGTCGGAT-3’. The PCR product was digested by the RsaI enzyme, a restriction enzyme whose restriction site is disrupted by the sa19827 mutation. We ran the digestion product on a 2% agarose gel for 30 minutes at 135 V. For control chd8+/+, two bands are detected (250 base pairs and 180 base pairs); for heterozygous chd8 sa19827/+, three bands are detected (428 base pairs, 250 base pairs, and 182 base pairs); and for homozygous chd8 sa19827/ sa19827 a single 428 base pair-band is detected. In figures, chd8+/- refers to heterozygous chd8 sa19827/+ and chd8-/- refers to homozygous chd8 sa19827/ sa19827.
Flow cytometry and RNA sequencing
chd8+/+ and chd8 sa19827/sa19827 males were crossed with Tg2(phox2bb:EGFP) females, and the eggs were incubated at 28.5 °C. At 4 days post-fertilization (dpf), the larvae were euthanized in 2 mg/mL tricaine diluted in RPMI and the heads of the larvae were discarded. The rest of the larval bodies were collected in a 2 mL Eppendorf tube, all RPMI was removed and replaced with 1 mL of Trypsin-EDTA 1X (Sigma, ref 59417C-100ML). The digestion was stopped after 10 minutes by adding 50 µL of inactivated fetal calf serum. The tubes were centrifuged at 2000 g, during 2 minutes at room temperature, the supernatant was removed and 100 µL of FACS Max medium were added (AMSBIO, ref T200100). The larval bodies were then placed on a cell filter (diameter 40 µm, Dutscher, ref 141378C), previously moistened with 100 µL of FACS Max medium, and the cells were filtered, using a 1 mL syringe plunger. The filter was rinsed with 400 µL of FACS Max medium, the cells were collected and placed in a 1.5 mL Eppendorf tube. The GFP-positive cells were immediately sorted, using an ARIA Fusion cell sorter and an excitation wavelength of 488 nm. We stored the GFP-positive cells at −80 °C, in 10 µl of PBS-RNAsine 1 U/µL. Each biological replicate consists of 950 to 1,300 cells harvested from 80 larvae. Harvesting of the GFP-positive cells was conducted on four different days, we thus controlled for batch differences when performing the subsequent differential gene expression analysis. Full length cDNAs were generated using Clontech SMART-Seq v4 Ultra Low Input RNA kit for Sequencing (Takara Bio Europe, Saint Germain en Laye, France), according to manufacturer’s instructions with 12 cycles of PCR for cDNA amplification by Seq-Amp polymerase. 600 pg of pre-amplified cDNA were then used as input for Tn5 transposon tagmentation by the Nextera XT DNA Library Preparation Kit (96 samples) (Illumina, San Diego, CA) followed by 12 cycles of library amplification. Following purification with Agencourt AMPure XP beads (Beckman-Coulter, Villepinte, France), the size and concentration of libraries were assessed by capillary electrophoresis. Libraries were then sequenced on an Illumina Hiseq4000 sequencer as single-end 50bp reads. The reads were pre-processed with cutadapt version 1.10 22 and mapped on the zebrafish genome (GRCz11 assembly), using the STAR software version 2.5.3a 23. For each sample, more than 85 % of the preprocessed reads were uniquely mapped and could be used to quantify gene expression using htseq-count version 0.6.1p1 24, with annotations from Ensembl version 98. One of the chd8+/+ samples was excluded from the analysis because the number of reads aligned on chd8 locus was very low, unlike in the other chd8+/+ samples. The differential gene expression analysis between enteric neurons of chd8+/+ and chd8 sa19827/+ larvae, controlling for batch differences, was conducted using the DESeq2 Bioconductor package version 1.16.1 25(Wald test and p-value adjustment using Benjamini and Hochberg method 26).
We conducted a Gene Ontology analysis on the list of upregulated and downregulated genes, as well as on the full list of differentially expressed genes, using a PANTHER overrepresentation test (using the website geneontology.org). We also used the DAVID functional annotation tool (version 6.8) on the same lists of genes. Finally, we performed STRING analysis on the full list of DE genes and we generated a full network of the query proteins, using all active interaction sources and a minimum interaction score of 0.4. We then clustered the genes involved in the PPI network, using the MCL clustering method and an inflation parameter of 3.1. We generated the heatmap using the Galaxy tool heatmap.2: toolshed.g2.bx.psu.edu/repos/iuc/ggplot2_heatmap2/ggplot2_heatmap2/3.0.1. The data was neither transformed nor clustered, and it was scaled by row.
Single cell RNA sequencing
chd8+/+ and homozygous chd8 sa19827/sa19827 male adult zebrafish were euthanized in 800 µg/mL tricaine solution. The fish were dissected, the guts were harvested and placed in RPMI at room temperature. The guts were rolled on paper moistened with RPMI to remove the fat residue, then placed in RPMI with 10 % fetal calf serum and cut into small pieces that were placed in 1 mL of digestion medium (1 mL of RPMI - 12 µL of activated fetal calf serum - 10 mg of dispase collagenase) for 15 minutes, at 37 °C, under agitation at 500 rpm. The cells were then filtered on a cell filter (diameter 40 µm, Dutscher, ref 141378C), using the plunger of 1 mL syringe. The cell concentration and viability were assessed with Trypan blue. Samples consisted of > 90 % viable cells and were processed on the Chromium Controller from 10X Genomics (Leiden, The Netherlands). 10,000 total cells were loaded per well. Single cell 3’ mRNA seq library were generated according to 10X Genomics User Guide for Chromium Single Cell 3’ Reagent Kits (v3 Chemistry). Briefly, Gel Beads-in-Emulsion (GEMs) were generated by combining barcoded gel beads, a RT master mix containing cells, and partitioning oil onto Chromium Chip B. Following full length cDNA synthesis and barcoding from poly-adenylated mRNA, GEMs were broken and pooled before cDNA amplification by PCR using 11 cycles. After enzymatic fragmentation and size selection, sequencing libraries were constructed by adding Illumina (Paris, France) P5 and P7 primers as well as sample index via end repair, A tailing, adaptor ligation and PCR with 14 cycles. Library quantification and quality control were performed using Bioanalyzer 2100 (Agilent Technologies, Santa Clara CA). Libraries were then sequenced on an Illumina NextSeq550 sequencer (2 runs: 28 + 96 and 101 + 101). Alignment, barcode and UMI filtering and counting were performed with Cell Ranger 3.1.0 count, using GRCz11 assembly and Ensembl release 98 annotations. Filtered gene-barcode matrix obtained with Cell Ranger count was further analyzed using R 4.0.2 and Seurat 3.2.0 27. Cells with at least 200 and less than 2,000 expressed genes and with less than 5% of mitochondrial reads and genes expressed in at least 3 cells were retained for further analysis. After normalization (NormalizeData with LogNormalize method), the two datasets were integrated (finding anchors using FindIntegrationAnchors and using these anchors to integrate the two datasets with IntegrateData using dimensions 1:50). After scaling the integrated data (ScaleData), we performed a Principal Component Analysis with 50 principal components (RunPCA). We use this PCA as input to perform a Uniform Manifold Approximation and Projection (UMAP) dimensional reduction in order to visualize the datasets (RunUMAP). Cell clustering was performed using FindNeighbors (with the first 50 principal components) and FindClusters (with a resolution of 0.3). To identify marker genes that are conserved between conditions for each cluster we used FindConservedMarkers. Differentially expressed genes between homozygous mutants and controls were identified using FindMarkers in each cluster. We conducted a Gene Ontology analysis on the list of upregulated and downregulated genes in each cluster, using a PANTHER overrepresentation test (using the website geneontology.org). Graphical representations were performed using DimPlot (UMAP), DotPlot (dot plots) and FeaturePlot (feature plots, where cells were represented in order of expression).
Imaging of the enteric NCCs in the intestine
Transgenic Tg2(phox2bb:EGFP) larvae were imaged at 24 hpf, 48 hpf and 72 hpf, on a lateral view, in PBS-Tween 0.1%, using a MacroFluo ORCA Flash macroscope (Leica). At least 15 larvae were imaged per condition and z-stacks were acquired. We used the ImageJ software to create a “Maximum Intensity” projection. To monitor the migration speed of enteric NCCs, we took time-lapse pictures of Tg2(phox2bb:EGFP); chd8+/+ and Tg2(phox2bb:EGFP); chd8 sa19827/+ embryos, every 10 minutes, between 50 hpf and 54 hpf, using a TimeLapse videomicroscope (Zeiss). The migration speed was assessed by measuring the distance travelled by the front of migration for one hour, and two measurements were taken per embryos, on two consecutive hours.
Immunostainings on zebrafish larvae
Zebrafish larvae were fixed in 4 % PFA for 1 to 3 hours, then incubated for 10 minutes in PBS-Triton 0.5 % and washed three times in PBS-Triton 0.1 % for 30 minutes, at room temperature. The larvae were then incubated in blocking solution (PBS-Triton 1 % - DMSO 1 % - BSA 1 % - FBS 1 %) for 1 hour at room temperature, then incubated in primary antibody diluted in the blocking solution, overnight, at room temperature. The next day, the larvae were rinsed three times in PBS-Triton 0.1 % for 30 minutes at room temperature and incubated in secondary antibody diluted in the blocking solution, for 2 hours at room temperature, in the dark. The larvae were stored in PBS, at 4 °C, in the dark. Complete list of primary and secondary antibodies is available in the Key Resources table. Larvae were imaged, on a lateral view, in PBS-Tween 0.1 %, using a MacroFluo ORCA Flash macroscope (Leica). At least 15 larvae were imaged per condition and z-stacks were acquired. We used the ImageJ software to create a “Maximum Intensity” projection and scored the number of fluorescent cells using the ICTN plugin.
Sudan black B staining
chd8+/+ and chd8 sa19827/+ larvae at 14 dpf and juveniles at 35 dpf were fixed in 4 % PFA for 4 hours at room temperature. They were washed 3 times for 5 minutes in 1 mL of 1X PBS, under agitation. They were then incubated in 1 mL of filtered Sudan Black B working solution (0,036 % (w/v) Sudan Black B (Merck, 15928), 0.1 % phenol, 94 % ethanol), in tubes covered in aluminum foil at room temperature for 1 hour, under agitation. They were then washed 3 times for 5 minutes in 70 % ethanol under agitation and washed in PBS-Tween 0.1 %. They were bleached in 1 mL of depigmentation solution (0.1 % KOH, 1% H2O2) for 5 minutes under agitation. Finally, they were washed twice in 1 mL PBS-Tween 0.1 % for 5 minutes at room temperature under agitation. The larvae and juveniles were imaged, on a lateral view, using a stereo microscope Leica MZ 125. A total of five or more SB-positive cells defines a bundle.
Paraffin sections and histological stainings
chd8+/+, chd8 sa19827/+ and chd8 sa19827/sa19827 male adult zebrafish were euthanized in 800 µg/ml tricaine solution. Six to ten mid- and posterior intestines per condition were collected 28 and then fixed in 10 % neutral buffered formalin (NBF) for 3 hours at room temperature. They were rinsed twice in 1X PBS and twice in 70 % ethanol. The intestines were paraffin-embedded according to standard procedure. Paraffin blocks were cut at a thickness of 5 μm with a Leica RM 2235 Manual Rotary Microtome. Masson’s Trichrome stain was performed as follows: tissues were post-fixed in Bouin’s solution during 1 hour at 56 °C and rinsed abundantly in running water for 7 minutes. Sections were stained in Weigert Hematoxylin (Sigma-Aldrich, C.I.75290) for 10 minutes. After a wash in water, sections were stained in Biebrich scarlet-acid fuchsin solution for 2 minutes. After another wash in water, slides were differentiated in a phosphotungstic acid solution for 15 minutes and directly transferred in Aniline blue solution (Sigma-Aldrich, C.I.42755) for 30 minutes. Alcian Blue/periodic acid-Schiff (PAS) stain was conducted according to standard procedure with a Harris hematoxylin (Sigma-Aldrich, C.I.75290) counterstain. All the stained tissue sections were cleared with an Histosol clearing agent, mounted with Eukitt medium, and imaged with a motorized Leica DM 4000B microscope equipped with a CoolSnap CF Color camera (Photometrics), 10x/0,30 (objective), 100x/1,30 OIL (objective). Illumination was done with a halogen lamp 100W. The images were merged with the Navigator interface driven by LasX software. The counts and measurements were made manually with the Fiji software, on 3 to 5 consecutive sections for each intestine. For epithelium and muscle layer measurements, a total of 5 measurements per section were taken randomly. Epithelium width measurements were done in the lower one-third of the villus as indicated by the double-headed arrow on Fig. 3a.
Quantification and statistical analyses
We used GraphPad Prism v8.0.2.263 (GraphPad Software, San Diego, CA) to visualize data. Statistical analyses were performed using either GraphPad Prism v8.0.2.263 or R v4.1.0. All experiments from this study were performed at least on three biological replicates with at least 15 larvae per clutch, from three independent clutches, or at least three adult zebrafish per group. When two groups were compared, normality of the distribution was assessed by performing a Shapiro-Wilk test. If the distribution was not normal, a Mann-Whitney test was conducted between pairs of conditions. If the distribution was normal, a F-test was conducted between pairs of conditions to assess whether the variances could be considered equal. If the variances were not statistically different, a Student’s t-test was conducted between pairs of conditions. If the variances were statistically different, a Welch’s t-test was conducted between pairs of conditions. On dot plots, the individual measurements are plotted, and the mean and standard deviation are represented. For qualitative data (e.g. classes based on the presence of mucus), a Fisher’s exact test was conducted between pairs of conditions to assess whether the distribution of samples in the different categories was significantly different. Two groups were considered statistically different if p-value < 0.05. No data were excluded from analyses, unless otherwise specified in the results.
Data availability
Lead contact
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Christelle Golzio, PhD (christelle.golzio{at}igbmc.fr).
Materials availability
This study did not generate new unique reagents.
Data and code availability
Single-cell RNA-sequencing data and bulk RNA-sequencing data have been deposited at GEO and are publicly available as of the date of publication. Accession numbers are listed in the key resources table. Microscopy data reported in this paper will be shared by the lead contact upon request.
This paper does not report original code.
Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.
Key resources table
Author contribution
G.H. conducted the zebrafish and transcriptomic experiments. M.M. and E.F. conducted the histological studies and Sudan Black staining and quantification. C.K. conducted the bioinformatic analyses. C.G. conceived and supervised all the experiments. G.H., M.M. and C.G. wrote the manuscript. All the authors discussed the results and commented on the manuscript.
Declaration of interests
The authors have no conflict of interest to declare.
Supplementary figures and data
Supplementary Figure 2: High resolution image of the Protein-Protein-Interaction network of the differentially expressed genes in chd8 sa19827/+. Nodes with no interactions with other proteins of the PPI network are not shown. MCL clustering was performed, using a 3.1 inflation parameter. Red line: fusion evidence; blue line: co-occurrence evidence; yellow line: text mining evidence; green line: neighborhood evidence; purple line: experimental evidence; light blue line: database evidence; and black line: co-expression evidence.
Supplementary data 1: Differential expression, DAVID and PANTHER analyses for all detected differentially expressed genes from RNA sequencing data. Differential expression analysis was carried out using DESeq2.
Supplementary data 2: PANTHER analyses on differentially expressed genes from single-cell RNA-seq data. For each cluster, salmon color shows the PANTHER analysis performed on up-regulated genes and blue color shows the PANTHER analysis performed on down-regulated genes.
Acknowledgements
This work was funded by Agence Nationale de la Recherche under the project JCJC ANR-17-CE12-0006 CNV (C.G.) and by the grant ANR-10-LABX-0030-INRT, a French State fund managed by the Agence Nationale de la Recherche under the frame program Investissements d’Avenir ANR-10-IDEX-0002-02. C.G. is a permanent INSERM investigator. G.H. is supported by a PhD fellowship (ANR-10-LABX-0030-INRT) and M.M. is supported by a PhD fellowship from Fondation ARD/Fondation de France. We thank the Imaging Center of IGBMC, in particular Didier Hentsch, Jean-Luc Vonesch, Yves Lutz, Elvire Guiot, and Erwan Grandgirard for their assistance in the imaging experiments. We are grateful to the staff of the IGBMC Flow Cytometry Facility, the Histopathology and Embryology Facility at Institut Clinique de la Souris, in particular Hugues Jacobs and Olivia Wendling, the GenomEast Platform, a member of the “France Génomique” consortium, ANR-10-INBS-0009. We thank the IGBMC Zebrafish Facility, in particular Sandrine Geschier. We are also grateful to Chantal Weber, member of C.G. laboratory for technical assistance.