Immunoglobulin superfamily member 3 is required for the vagal neural crest cell migration and enteric neuronal network organization

The immunoglobulin (Ig) superfamily members are involved in cell adhesion and migration, complex multistep processes that play critical roles in embryogenesis, wound healing, tissue formation, and many other processes, but their specific functions during embryonic development remain unclear. Here, we have studied the function of the immunoglobulin superfamily member 3 (IGSF3) by generating an Igsf3 knockout (KO) mouse model with CRISPR/Cas9-mediated genome engineering. By combining RNA and protein detection methodology, we show that during development, IGSF3 localizes to the neural crest and a subset of its derivatives, suggesting a role in normal embryonic and early postnatal development. Indeed, inactivation of Igsf3 impairs the ability of the vagal neural crest cells to migrate and normally innervate the intestine. The small intestine of Igsf3 KO mice shows reduced thickness of the muscularis externa and diminished number of enteric neurons. Also, misalignment of neurons and smooth muscle cells in the developing intestinal villi is detected. Taken together, our results suggest that IGSF3 functions contribute to the formation of the enteric nervous system. Given the essential role of the enteric nervous system in maintaining normal gastrointestinal function, our study adds to the pool of information required for further understanding the mechanisms of gut innervation and etiology behind bowel motility disorders.

and surrounding mesenchyme are heavily reliant on both cell-cell and cell-matrix interactions 5,6 .The regulation of the adhesion properties is strict as both too weak and too strong adhesion leads to migration defects 7 .Thus, cell adhesion molecules play a vital role in NC cell migration and aggregation at their destinations, which is a concerted action regulated by the immunoglobulin superfamily (IgSF) members of cell adhesion molecules (IgCAM), cadherins, integrins, and the proteolytic enzymes 8 .Importantly, the involvement of the EWI subfamily of IgSF molecules (containing the conserved ectodomain Glu-Trp-Ile (EWI) motifs) in NC cell biology is currently unknown.
The IgSFs form a large family of cell surface proteins that are involved, in addition to the signaling and immune associated functions, in adhesion-mediated cellular processes such as recognition, formation of cellto-cell and cell-to-ECM binding complexes and motility 9,10 .In our study, we have focused on the function of the IgSF member 3 (IGSF3) by generating and analyzing a CRISPR/Cas9 knockout (KO) mouse model.The gene encoding Igsf3, an EWI-subfamily member of IgSF, is located in chromosome one, and based on its sequence, it is a putative membrane protein with predicted functions in cell adhesion and cell surface receptor signaling with no predicted role in immune system regulation 11 .However, the physiological role of IGSF3 is largely unknown.Recently, Igsf3 was reported to be involved in axonal growth and branching of granule cells in the developing cerebellum 12 .
The lining of the digestive tract is exposed to a broad range of chemicals and organisms, and the gut is comprised of numerous sensors that in addition to sensing nutrients, also serves as a first-line defense to detect micro-organisms and toxins.As part of the peripheral nervous system, the enteric nervous system (ENS) plays a pivotal part in communicating the sensory information into extensive endocrine, neural, immune, and nonimmune responses 13 .Due to the direct connection between the ENS and the central nervous system (CNS) and the ability of these neuronal connections to serve as gateways for disease transmission, dysfunctions of the ENS are not only associated with digestive disorders but also neurological disorders 14 .The ENS development is necessary to warrant the regulation of gut function 15 .Four general concentric layers make up the gastrointestinal tract: from the innermost lumen facing mucosa, followed by submucosa, muscularis externa (muscularis propria), and the outermost protective serosa.Additionally, each layer consists of structural and functional subcomponents [16][17][18][19] .The epithelium, lamina propria, and muscularis mucosae make up the mucosa; the submucosal plexus constitutes the submucosa.Furthermore, there are three segments of the muscularis externa: a circular smooth muscle layer, a myenteric plexus, and a longitudinal smooth muscle layer [18][19][20] .Thus, two plexuses compose the ENS: the myenteric plexus sits between the outer longitudinal smooth muscle cell layer and the inner circular smooth muscle layer; the submucosal plexus sits between the inner circular smooth muscle cell layer and the epithelial/mucosal layers of the gut 16,17,21,22 .
Here, we demonstrate that during development IGSF3 is expressed in the neural crest and its derivatives, the enteric nervous system neurons.By generating and analyzing the Igsf3 KO mice we show that Igsf3 is required for a faithful NC cell migration and proper development of the small intestine.

IGSF3 is expressed in the developing mouse nervous system and neural crest derivatives
To examine the IGSF3 expression during fetal development, mouse embryos (E13.5 and E17.5) were immunostained with an anti-IGSF3 antibody (Fig. 1A,G).The highest levels of IGSF3 protein expression were detected in the NC-derived tissues and parts of the CNS, in agreement with the previously reported mRNA expression pattern in the brain 12 .Specifically, IGSF3 protein expression was found in the brain cortex (Fig. 1B,C,H), cerebellum (Fig. 1I), and optic nerve (Fig. 1J), and in multiple NC-derived tissues including teeth (Fig. 1K), the craniofacial skeleton (Fig. 1D,L), and developing enteric ganglionic plexi of the small intestine and in the intestinal villi (Fig. 1E,M, also note the higher magnification insert in M).Similarly, the dorsal root ganglia (Fig. 1F,N) and the spinal cord (Fig. 1F,O) expressed IGSF3.

Single-cell RNAseq analysis identifies Igsf3 expression in the neural crest
The high expression of IGSF3 in the NC lineage intrigued us to explore the published single-cell RNA (scRNA) sequencing datasets of mouse NC (E8.5 and E9,5; Wnt1-Cre; R26 TOM background) 23 .At E8.5, cells separate into NC (identified by Sox10) and the CNS forming neural tube (NT, identified by the high expression of the neural progenitor marker Sox2) clusters.The NC cell cluster showed high expression levels of Igsf3 (Fig. 2A-C).Furthermore, the expression pattern continued at E9.5, with the highest Igsf3 expression mostly detected in the Sox10 + NC cells with some expression also in the cells positive for Sox2 (Fig. 2D-F).Part of the Sox2-positive cells may represent NC cells as lower Sox2 expression is typical for the premigratory NC cells 24 .At this point, the cranial NC in the Sox10-positive population has already reached the late migratory stage, whereas the vagal and anterior trunk NC is still emigrating or early migratory 23 .To verify these findings at the protein level, we stained the E9.5 embryos with anti-IGSF3 and anti-Sox10 antibodies.Indeed, high IGSF3 expression was detected in the Sox10-positive premigratory and migratory neural crest region when analyzed in subsequent sections (Fig. 2G and H).
Next, we analyzed whether other IgSF members were expressed in the NC similarly to IGSF3.At E8.5, only Igsf3 showed high exclusive expression in most of the cells in the NC cluster, while the other family members were expressed predominantly by the NT cluster (Igsf8, Igsf9, Igsf21) or equally in some cells in both the NC and NT clusters (Igsf6, Igsf9b, Igsf10).Additionally, Igsf11 expression was found predominantly in the NC cell population but the overall percentage of the Igsf11-positive cells in the NC cell cluster was low (Fig. 3A).No expression of Igsf5 and Igsf23 was detectable at E8.5 in the used dataset.At E9.5, of the expressed Igsf genes, Igsf8 and Igsf9 were predominantly expressed in the NT population, while Igsf9b and Igsf10 were detected about equally in the NC and NT cells, and Igsf11 showed a slight preferential expression in the NC cell cluster (Fig. 3B) while Igsf3 was mainly expressed by the NC cell cluster.At E10.5 25 , Igsf3 was exclusively expressed in the choroid plexus and the NC cells, as no expression of other IgSF members was detected in these cells (Fig. 3C).Regarding the other IgSF members, high Igsf6 expression was observed in immune cells, notable levels of Igsf8 were detected in neuroblasts, ependymal, immune, and Schwann cells, while Igsf9 was restricted to the ectoderm and endoderm, and Igsf9b expression was detected in ependymal cells (Fig. 3C).Combined, the scRNAseq data indicate that Igsf3 is the only IGSF member that is highly expressed by the NC cells.

Inactivation of Igsf3 in mouse by using CRISPR/Cas9
To understand the physiological role of Igsf3, we generated an Igsf3 gene knockout mouse model by inactivating it using the CRISPR-Cas9 mediated genome editing system.Briefly, the in vitro transcribed small guide RNA (sgRNA) targeting the exon 3 of Igsf3 together with Cas9 mRNA was injected into the pronuclei of FVB/NRj zygotes to generate potential founders that were screened by PCR-based analyses followed by Sanger sequencing.Genotyping revealed a 146-nucleotide deletion in the targeted exon 3, which was confirmed by Sanger sequencing.The in-silico translation of the KO DNA sequence by using the Serial Cloner software (http:// seria lbasi cs.free.fr/ Serial_ Cloner.html) showed that this deletion causes a frameshift mutation resulting in a premature stop codon (Fig. 4A,B).We first analyzed the Igsf3 mRNA levels by using qPCR in the postnatal day 2.5 (P2.5) cerebrum and intestinal samples.No Igsf3 mRNA was detected in the KO animals while no difference was observed in the mRNA levels between the heterozygous (HET) and wild type (WT) littermates (Fig. 4C,D).To validate the loss of IGSF3 expression at protein level, we first verified that our anti-human IGSF3 antibody also recognizes the mouse IGSF3 protein.To this end, we transfected the human embryonic kidney (HEK293FT) cells that express low levels of IGSF3 with plasmid encoding the murine Igsf3 gene.Forty-eight hours post-transfection we prepared extracts from the WT, mock-transfected, and Igsf3-transfected HEK293FT cells and analyzed the expression by using Western blot.Only the Igsf3-transfected cells showed IGSF3 protein expression (arrow in Fig. 4E, Supplementary Information) confirming that the antibody also recognizes the murine protein.Next, we performed the analysis of IGSF3 protein expression in the P2.5 cerebrum extracts of WT, heterozygous, and  KO pups.Our results show nearly a complete loss of protein in KOs, whereas the heterozygous animals showed an average reduction of 46% (Fig. 4F,G, Supplementary Information).In addition, Igsf3 loss in the KO animals was confirmed at the protein level using whole mount immunofluorescence staining of the P12 intestine.In the P12 WT animals, IGSF3 expression was detected in the myenteric plexus of the small intestine where it colocalized with the neural marker NCAM1 (Fig. 4H-J).No IGSF3 expression was detected in the KO intestinal wall (Fig. 4K-M).

Igsf3 KO impairs neural crest cell migration and development of intestinal muscularis propria
The Igsf3 KO pups were born in Mendelian ratios and so far, have not shown deviations from the normal lifespan (data not shown).The high IGSF3 expression in NC cells in early embryonic development suggests its involvement in neural crest and peripheral nervous system development.To address this, we first performed an in vitro migration assay using the vagal and trunk neural crest explants isolated at E9.0 from the WT and KO embryos.The explants were cultured on fibronectin-coated plates and migration of NC cells was followed for 48 h after which the explants were imaged (Fig. 5A-H) and the distance of the migrated cells from the explants was quantified (Fig. 5I,J).The migration length of the KO derived vagal NC cells was significantly reduced as compared to the WT derived vagal NC cells (Fig. 5A-D,I).Interestingly, no difference was observed in the migration of the trunk NC cells between the KO and WT (Fig. 5E-J).In line with this, we immunostained the neurons of the lumbar dorsal ganglion, which are derivatives of the trunk NC cells, and no difference in the intensity of the Tuj1 expression was detected between the KO and WT (Fig. 5K-M).As the vagal neural crest contributes to the enteric neurons and glia, we focused our investigation on the effects of the loss of Igsf3 in the small intestine.At P9 the cross sections of the KO small intestine showed disorganized structure with rudimentary outer muscle layer compared to the WT counterpart (Fig. 6A,B).The staining using anti-alpha smooth muscle actin (αSMA) antibodies confirmed the observation at P9 (Fig. 6C-F).More detailed analysis at P12.5 showed reduced thickness of the muscularis externa of the proximal small intestine in the KO (Fig. 6H) compared with the WT littermate controls (Fig. 6G).The quantification revealed a statistically significant reduction in the thickness of muscularis externa in the KO intestines (Fig. 6I).
Next, we investigated the detailed localization of IGSF3 in the intestinal villi by using immunostaining.Our results show that the expression of IGSF3 in the villi co-localizes with the NCAM1-positive neurons (Fig. 7A) similar to what we detected in the myenteric plexus (Fig. 4H-J).These results, together with the observed thinner muscularis externa in KO intestine, prompted us to hypothesize that IGSF3 in the villi is required for the neurons to support the proper development of the smooth muscle cells.To this end, we visualized the smooth muscle cells and neurons of the villi in the whole-mount small intestine at P12 by using immunostaining with antibodies against Tuj1 and αSMA.Our results show reduced expression of both markers in the KO villi as compared to the WT (Fig. 7B).Consequently, we utilized neuronal cell adhesion molecule 1 (NCAM1) as an additional neuronal marker to visualize neurons in the P9 intestine.The samples were co-stained with an antibody against the epithelial adhesion molecule E-cadherin (E-Cad), which marks the lining of the small intestinal villi.While the transverse sections of the WT villi readily show NCAM-positive neurons inside each villus, they are markedly reduced within the KO intestine (Fig. 7C,D).Since IGSF3 is known to be expressed in the brain 12 we performed Western blot analysis of P2.5 cerebrum extracts, which also showed reduced NCAM1 expression in KOs compared to heterozygous and WT animals (Fig. 7E, Supplementary Information) further supporting our results on IGSF3 impacting neuronal development.Of note, the NCAM1 expression was also reduced in the myenteric plexus of the KO mice shown in Fig. 4H and K, but the NCAM1 expression is enhanced to visualize that although defected, the neuronal network in the KO mice is not lost.
Next, we performed several additional quantifications to get a more in depth understanding of the intestinal villus phenotype.First, measurement of the integrated intensity of the fluorescent signal confirmed a significant reduction for both Tuj1 and αSMA in the KO intestines as compared to the WT (Fig. 7F-G).Moreover, The effect of Igsf3 deletion at the mRNA level was evaluated by using qPCR of the P2.5 cerebrum (C) and intestine (D).In heterozygous tissues, no significant differences in the Igsf3 levels were detected at the mRNA level whereas no Igsf3 transcript was detected in the KO tissue samples.(E) HEK293FT cells were transfected with plasmid encoding the murine Igsf3 gene.Cell extracts were prepared 48 h posttransfection and IGSF3 expression was analyzed using Western blot.WT and mock-transfected HEK293FT were used as controls.Arrow points to the band with correct molecular weight.(F) IGSF3 gene deletion was validated at the protein level from P2.5 cerebrum extracts by using Western blot analysis.The KO showed a nearly complete loss of protein expression while reduced expression was detected in the heterozygous brain samples.(G) The IGSF3 protein expression of the cerebrum samples was quantified by normalizing it to the house keeping protein GAPDH.(H-M) The successful loss of the Igsf3 in the KO animals was further validated by immunostaining the myenteric plexus of the intestinal wall on whole mount samples from P12.5 mice using anti-NCAM1 (H,K) and anti-IGSF3 (I,L) antibodies (N = 3 for each genotype).No IGSF3 signal was detected in the KO intestines (L,M) while expression was clearly visible in the WT (I,J) and it colocalized with the NCAM1.Scale bars: (G-L)100 µm.NCAM levels in (H) and (K) have been enhanced to visualize the neuronal network since NCAM1 expression is reduced in the KO compared to the WT.quantitative measurement of the ratio between the smooth muscle fibers (αSMA) and the neurons (Tuj1) in the villi revealed an increased ratio of αSMA positive staining over Tuj1 in the KO villi (Fig. 7H).This suggests that the knockout villi contained either fewer neurons per smooth muscle fiber as what is present in the WT villi, or alternatively, that the Tuj1-levels per neuron are more reduced than the respective reduction in the αSMA expression measured in the smooth muscle cells.To understand this, we quantified the numbers of nerve and muscle fibers in the villi and observed a significantly reduced number of nerve fibers per KO villi as compared to the WT villi (Fig. 7I).No difference in the muscle fiber counts per villi was observed (Fig. 7J).Lastly, we measured whether the alignment of the neurons with the smooth muscle was altered by quantifying co-localization of the two markers.Indeed, the results show impaired alignment in the KO intestine as compared to the WT (Fig. 7K).In sum, while the expression of aSMA in the muscle fibers was reduced (Fig. 7G), the number of the fibers was not changed (Fig. 7J) indicating that the individual fibers expressed lower levels of SMA.On the other hand, we find that while the expression of Tuj1 was reduced (Fig. 7F), the amount of the nerve fibers was also reduced (Fig. 7I).These results suggest that IGSF3 is required for the proper development and organization of the innervation process of the intestinal villi.

Discussion
Several cell adhesion molecules of the immunoglobulin superfamily (IgSF), including the NCAM 26 and members of the L1 family of neuronal cell adhesion molecules 27 , play crucial roles in the development of the nervous system.However, despite extensive study of selected IgSF proteins, the IgS family remains poorly characterized 28 .Here, by using immunohistochemistry and scRNAseq analyses we show that, in addition to the expression of the immunoglobulin super family member 3 (IGSF3) in the central nervous system, as reported by a previous study 12 , it is strongly expressed in the early, premigratory, and migratory NC cells, and subsequently in a broad  selection of tissues that fully or partially originate from the NC such as teeth, the craniofacial skeleton, peripheral ganglia as well as the enteric nervous system 3 .
The neural crest is a transient embryonic cell population with extraordinary migratory abilities, enabling NC cells to migrate across the embryo and colonize different tissues 6 .The NC cells of the vagal axial level give rise to the enteric nervous system as they enter the developing gut from the anterior end, and as it extends, continue to further populate posterior parts of the intestine 29 .To understand the role of Igsf3 in migrating neural crest and its derivatives, we generated an Igsf3 KO mouse model using the CRISPR-Cas9 genome editing system 23 that resulted in a 146-nucleotide deletion in exon 3, which led to a complete loss of protein expression.Collectively, our phenotypic analysis showed reduced migration of the vagal NC cells and impaired development of its derivative, the enteric nervous system.
Immunostaining confirmed that it is the enteric neurons that express IGSF3 during normal embryonic and postnatal development.Analysis of the proximal small intestine at P12 revealed defects in formation of the muscularis externa and the ability of smooth muscle cells to align with their intestinal neuronal counterparts, indicating an overall disorganized innervation of the enteric villi due to loss of Igsf3.The intestinal neurons and smooth muscle are intimately connected and the ENS is the key-regulator of intestinal motility 30 .The age-related loss of enteric neurons associates with the increased incidence of intestinal motility problems including delays in gastric emptying and longer intestinal transit time with associated fecal stasis 31 .Diverticular disease (DD) is an intestinal neuropathy, pathogenesis of which involves intestinal motor disturbances and an underlying enteric neuromuscular pathology.Patients with DD show significantly reduced neuronal density, decreased intramuscular nerve fibers, and abnormal intestinal motility pattern 32,33 .As the changes in ENS affect intestinal muscles and vice versa, we suggest that the effects we detected in the Igsf3 KO animals are due to the loss of function of IGSF3 in enteric neurons.Consequently, we hypothesize that the reduced thickness of the muscularis externa is a secondary effect due to the defects in the myenteric plexus and disorganized neurons in the intestinal villi.However, revealing the molecular mechanism and distinguishing between these functions will require further experimentation outside the focus of this work.
We also observed, using an in vitro preparation of neural tube explants, impaired migration of the vagal NC cells.This may be caused by a defect in the ability to form sufficient amounts of NC cells, a delay in emigration, or an impaired migration function as epithelial-to-mesenchymal transition and migration are both enabled by turnover of cell adhesion molecules, which facilitates rapid cellular rearrangements [34][35][36] .As cells migrate, adhesion to the substrate induces cytoskeletal remodeling to drive membrane protrusion and cell spreading.The actin cytoskeleton and cell-to-cell contacts in collective migration allow cells to couple mechanically and chemically 37 .Consequently, this earlier phenotype of impaired NC cell migration may result in decreased numbers of NC cells reaching the gut to form the enteric nervous system, and defects in co-localization may well be caused by a combination of this and the impaired IGSF3-mediated adhesion of the neurons to the smooth muscle fibers in the KO villi leading to disorganized neuronal network.Our results are in line with previous reports showing that the loss of function of IgSF cell adhesion proteins (CAMs) lead to defective neuronal migration and affects axonal growth 38 .Moreover, IGSF3 was found to regulate axonal growth and branching and accumulate in the axon terminals 12 .In addition, IGSF3 was recently reported to drive glioma progression via synaptic remodeling and brain network hyperactivity 39 .For future studies, using our newly generated Igsf3 knockout mouse model in combination with additional molecular approaches, it will be interesting to investigate the mechanistic details of how IGSF3 regulates adhesion in the neural crest and the enteric nervous system neurons.
Aside from controlling bowel motility and epithelial secretion, the ENS also interacts with enteroendocrine cells to modulate the intestinal immune response, and transmission of extrinsic nerve impulses, epithelial proliferation, and repair.While neurons of the ENS are the primary controllers of gastrointestinal functions, intercellular communication with enteric glia and non-ENS cell types such as macrophages and the smooth muscle cells is an essential aspect of their function 40,41 .Gut motility, including peristalsis waves that move the digested material through the gut, and segmentation movement, which mixes food with digestive enzymes and bile, is regulated through crosstalk with the smooth muscle cells.Human bowel motility disorders include life threatening conditions like Hirschsprung disease and chronic intestinal pseudo-obstruction (CIPO) syndrome 41 , highlighting the clinical relevance of research on neuro-muscular interactions within ENS development.Taken together, we demonstrate that the developing tissues of the neural crest express IGSF3, which is essential for vagal neural crest migration and formation of proper connections between intestinal neurons and smooth muscle cells during intestinal development.

Immunohistochemistry
The tissue sections were processed using the TSA kit (Perkin Elmer) according to the manufacturer's instructions.The deparaffinization of fixed tissues were performed by immersing the slides in tissue clear and rehydrated with graded ethanol (100-70%) and water.The heat-induced antigen retrieval was performed using citrate buffer (1.8 mM citric acid, 8.2mM sodium citrate, pH 5.0).The slides were treated with 3% H 2 O 2 -MetOH for 10 min to inactivate endogenous peroxidase activity.The tissues were blocked for 30 min with TNB buffer (0.1M Tris-HCL (pH 7.5), 0.15M NaCl, 0.5% blocking reagent from TSA kit).The primary antibodies were incubated overnight at 4 °C and biotinylated secondary antibody for 60 min at room temperature.The primary antibodies include polyclonal rabbit IGSF3 (1:750 HPA036305, Sigma-Aldrich), polyclonal mouse SOX10 (1:500 SC-365692, Santa Cruz).The reddish brown color was the stain generated by secondary biotinylated antibodies which includes goat-anti rabbit IgG (E0432, Dako) and goat-anti mouse IgG (E0433, Dako).The washes were performed with TNT (Tris/NaCl/Tween 20) buffer.The Mayer's hematoxylin (S3309, Dako) was used to counterstain the tissue Vol:.( 1234567890

Generation of the Igsf3 KO mouse model
The FVB/NRj mice were obtained from the Janvier labs (France) and used in the study for genome editing.
We confirm that all experiments were performed in accordance with relevant guidelines and regulations. .We confirm that this study does not involve human participants, personal data, human cells, or tissues.Igsf3 inactivation in mouse was performed according to the published in vivo CRISPR/Cas9 methods 42 .The Igsf3 gRNA (25 ng/µl, 5'-GTG TAG ATG GCG TAC GGG AAGG-3'), targeting the exon 3 (Exon ID-ENS-MUSE00000456959) immediately after the signal peptide was obtained from mouse genome-wide arrayed lentiviral CRISPR/Cas9 gRNA libraries (Sigma-Aldrich/Merck).The gRNAs were in vitro transcribed and dissolved in Tris-EDTA buffer.The Cas9 mRNA (25 ng/µl Sigma Aldrich) and gRNA were co-injected into the pronuclei of FVB/Nj (Janvier Labs, France) zygotes.Transplantation of injected zygotes to oviduct of pseudopregnant recipient females gave rise to two founders.Sanger sequencing was performed on all the founders (F0) and F1 offspring of two founders that were used to generate Igsf3 knockout mice.The selected founder lines had a 146and 180-nucleotide deletion in the Igsf3 gene that causes a frameshift resulting in a premature stop codon right after the signal peptide (Serial Cloner software).

Genotyping
The genomic DNA from ear punctures was isolated using the NucleoSpin® Tissue according to manufacturer's protocol.The genotyping primers were designed to span 128 nucleotides upstream and 264 nucleotides downstream of the targeted Igsf3 exon 3. The targeted region was amplified by PCR using the Phusion DNA polymerase using the following primers (5′-CTT GGA AAG GGC CAG TAA CC-3′ and 5′-CCT GGA TCT GGC CTT GAT AA-3′) resulting in 770 nucleotide long products.The cycle conditions were 98 °C 1 min, followed by 98 °C 10 s, 57.6 °C 30 s, 72 °C 30 s (30 cycles), then 72 °C for 5 min and stored at 4 °C.Sanger sequencing of PCR amplicons was performed to identify the editing events such as insertion, deletion, and frameshift mutation.The sequence analyses were performed using the SnapGene Viewer and Serial Cloner.The genotype of the animals was analyzed using Sanger sequencing until the F3 generation.After F3 generation, all the samples were analyzed by gel electrophoresis.

Plasmid construction
The plasmids were generated using the Genome Biology Unit core facility cloning service (ORFeome Library; Genome Biology Unit supported by HiLIFE and the Faculty of Medicine, University of Helsinki and Biocenter Finland, https:// www.helsi nki.fi/ en/ resea rchgr oups/ genome-biolo gy-unit/ orf-libra ries).The mouse Igsf3 ORF (CAT#: MR211786, Origene) sequence was amplified using the Gateway-tailed primers.Briefly, the gatewaytailed entry clone was transferred into the pLenti7.3/V5-DEST(ampicillin) using the standard LR reaction (gateway cloning) protocol.

Expression of murine IGSF3 in human HEK293FT cells
Human embryonic kidney (HEK293FT) cells (ATCC) were cultured in DMEM high glucose (4.5 g/L) with L-glutamine (Gibco) supplemented with 10% FBS (Gibco), 100 U/mL penicillin, and 100 μg/mL streptomycin (both from BioNordika).HEK293FT cells were plated one day before transfection and were approximately 60% confluent on the day of transfection.The murine Igsf3 plasmid (ORFeome Library) was transfected into HEK293FT cells by using the FuGENE6 transfection reagent (Promega) at a 3:1 ratio of FuGENE6 transfection reagent to DNA according to the manufacturer's protocol.Cells were incubated for 48 h, lysed, and the extracts were analyzed for IGSF3 expression by Western blotting.Mock-transfected cells (plasmid without Igsf3) and WT HEK293FT cells were used as controls.

Immunoblotting
The tissue samples were snap frozen using liquid nitrogen and stored at -80 °C.The frozen samples were transferred for homogenization to tubes containing zirconium beads (MB2Z015, Bioline) with RIPA buffer (50mM Tris HCl pH 7.4, 150mM NaCl (Sigma-Aldrich), 1% Sodium Deoxycholate (Sigma-Aldrich), 1% SDS (Sigma-Aldrich), 2% Octyl-β-D-Glucopyranoside (Sigma-Aldrich) with protease and phosphatase inhibitors (both from Roche) and homogenized using the following program: 2*3500 rpm for 15 s with 10 s' delay.The homogenized samples were incubated in ice for 20 min and centrifuged at 4 °C for 20 min at 13,000 rpm.The Pierce™ BCA Protein Assay Kit (Thermo Fisher Scientific) was utilized for determining the protein concentration.The supernatant was collected and diluted in the SDS Laemmli loading buffer (supplemented with 5% β-mercaptoethanol), boiled for 5 min at 95 °C and electrophoresed using NuPAGE Bis-Tris Gels (Invitrogen, Thermo Fisher Scientific), and then transferred onto PVDF membranes (Bio-Rad) using the Transblot Turbo transfer system (Bio-Rad).The blocking was performed for 60 min at RT using non-fat dry milk (5%) in TBS-T buffer.The blocking buffer was used as a diluent for the primary anti-IGSF3 (AF4788, R&D Systems, 1:1000),

Figure 1 .
Figure 1.IGSF3 expression in the developing mouse organs.(A-F) Sagittal sections of mouse embryo at E13.5 stained with an anti-IGSF3 antibody and visualized in brown color.(A) Overall IGSF3 protein expression in the embryo.(B-F) Higher magnification panels show expression (marked by arrows) in the forebrain (B), the hindbrain (C), the craniofacium (D), small intestine (E), and dorsal root ganglia (arrow) and spinal cord (arrowhead) (F).(G-O) Sagittal sections of mouse embryo at E17.5 stained with anti-IGSF3 antibody and visualized in brown color.(G) Overall expression in the embryo.(H-O) Higher magnification panels show expression (marked by arrows) in the cerebral cortex of the brain (H), the cerebellum (arrow) and the choroid plexus (arrow head) (I), optic nerve (J), teeth (K), whiskers (arrow) and the craniofacium (L), developing enteric ganglionic plexi of the small intestine, insert shows higher magnification of the boxed area (M), dorsal root ganglia (N), and the spinal cord (O).Scale bars: (A) 1000 µm; (B-F) 200 µm; (G) 2000 µm; (H-O) 100 µm.

Figure 2 .
Figure 2.Both IGSF3 mRNA and protein are expressed in the neural crest cells.(A-F) Single cell RNA sequencing analysis at E8.5 and E9.5 represented as t-SNE plots reveals the neural crest (NC) and neural tube (NT) cell clusters defined by Sox10 (B,E) and Sox2 (C,F) expression, respectively.The subclusters are assigned according to the annotations from the original analysis of the source data23 .Igsf3 is expression is restricted to the neural crest cells at E8.5 (A).At E9.5, Igsf3 is expressed predominantly by the delaminating and migrating neural crest which is identified by Sox10 with some expression in the neural tube cells (F).(G,H) Serial sections of E9.5 embryo were stained using the anti-IGSF3 and anti-SOX10 antibodies (red color).IGSF3 expression (G) is detected in the Sox10-positive (H) neural crest cells as indicated by arrows pointing to the neural crest cells that are newly migrating out of the dorsal neural tube on both sides.Scale bars: (G,H) 50 µm.

Figure 3 .
Figure 3. Single-cell RNAseq analysis shows that Igsf3 is the only IGSF member that is expressed by the neural crest cells.(A) Large panel in the left upper corner displays the t-SNE embedding showing the cell clusters of E8.5 Wnt1-Cre; R26Tomato embryo.The image is adopted from the original analysis and annotations of the source data, of which our analysis is based on23 .The early neural crest comprises of three spatially distinct clusters including a Hox-negative (-) corresponding to the anterior cranial neural crest, Hoxb2positive ( +) corresponding to the mandibular level, and HoxD3-positive ( +) corresponding to the post-otic (including cardiac and vagal) streams of the neural crest23 .Single cell RNA sequence analysis reveals the expression of different Igsf members in the distinct cell clusters defined respectively by Sox10 (NC cells) and Sox2 expression (NT cells).Igsf3 is the only family member expressed solely by the neural crest cells at E8.5.The other IGSF members are expressed about equally by both cell clusters or preferentially by the neural tube cell cluster.Expression of Igsf5 and Igsf23 was not detected at E8.5.(B) Large panel in the left upper corner displays the t-SNE embedding showing the cell clusters of mouse E9.5 Wnt1Cre/R26R Tomato embryo and reflects the spatiotemporal properties of neural crest (NC) development, as adopted from the original analysis of the source data23 .These neural crest cell clusters are classified into the following major subpopulations such as pre-EMT, delaminating and migrating neural crest, sensory neurons, autonomic neurons, and mesenchyme23 .Single cell RNA sequence analysis at E9.5 represented as t-SNE plot reveals the expression of various IgSF members in different neural tube (NT) and neural crest (NC) cell clusters defined by Sox10 and Sox2 expression, respectively.Igsf3 is expressed in the NC and its derivative cell clusters also at this stage.Igsf8 and Igsf9 are predominantly expressed in the NT population, while Igsf10 and Igsf9b are detected about equally in the NC and NT cells, and Igsf11 shows a slight preferential expression in the NC cell cluster.(C) At E10.5, the choroid plexus and NC cells specifically express Igsf3, but other Igsf members are not co-expressed in these cells.

Figure 4 .
Figure 4. Deletion in the exon 3 of Igsf3 gene leads near to complete loss of the IGSF3 protein.The CRISPR-Cas9 mediated genome engineering targeted exon 3 of the Igsf3 gene.(A) Schematic presentation of the Igsf3 inactivation strategy.Deleted nucleotides span across the sgRNA binding sites.The red star at the end of the nucleotide sequence indicates the pre-mature stop codon.(B) Gene analysis reveals a 146-nucleotide deletion in the KO allele, which resulted in a premature stop codon right after the signal peptide of Igsf3.PAM protospacer adjacent motif.(C,D) The effect of Igsf3 deletion at the mRNA level was evaluated by using qPCR of the P2.5 cerebrum (C) and intestine (D).In heterozygous tissues, no significant differences in the Igsf3 levels were detected at the mRNA level whereas no Igsf3 transcript was detected in the KO tissue samples.(E) HEK293FT cells were transfected with plasmid encoding the murine Igsf3 gene.Cell extracts were prepared 48 h posttransfection and IGSF3 expression was analyzed using Western blot.WT and mock-transfected HEK293FT were used as controls.Arrow points to the band with correct molecular weight.(F) IGSF3 gene deletion was validated at the protein level from P2.5 cerebrum extracts by using Western blot analysis.The KO showed a nearly complete loss of protein expression while reduced expression was detected in the heterozygous brain samples.(G) The IGSF3 protein expression of the cerebrum samples was quantified by normalizing it to the house keeping protein GAPDH.(H-M) The successful loss of the Igsf3 in the KO animals was further validated by immunostaining the myenteric plexus of the intestinal wall on whole mount samples from P12.5 mice using anti-NCAM1 (H,K) and anti-IGSF3 (I,L) antibodies (N = 3 for each genotype).No IGSF3 signal was detected in the KO intestines (L,M) while expression was clearly visible in the WT (I,J) and it colocalized with the NCAM1.Scale bars: (G-L)100 µm.NCAM levels in (H) and (K) have been enhanced to visualize the neuronal network since NCAM1 expression is reduced in the KO compared to the WT.

Figure 5 .
Figure 5. Loss of Igsf3 impairs neural crest cell migration.(A-J) E9.0 WT and KO vagal (collected at somite level 1-7) and trunk (collected at somite level 9-12) neural tube explants were cultured on fibronectin-coated cover slips for 48 h after which they were imaged.(A,B) Vagal neural crest explants isolated from WT (A) and KO (B) embryos.(C,D) Panels show the zoomed-in view of the boxed areas in (A) and (B).(E,F) Trunk neural crest explants isolated from WT (E) and KO (F) embryos.(G,H) Panels show the zoomed-in view to the boxed areas in (E) and (F).(I,J) Quantification of the neural crest cell migration.The area containing the migrating cells is marked with a green line, and the migration distance between the outer edges of the halo and the explant (black) was measured.The vagal neural crest cells of the KO explants migrated significantly less than the WT neural crest cells (I) while no difference was detected in the trunk neural crest cell migration between the WTs and KOs (J) (N = 4 for each genotype).(K,L) Representative images of the dorsal root ganglia in WT (K) and KO (L) pups at P12.5 (N = 3 for each genotype).Neurons are visualized by Tuj1-staining (green) and nuclei with DAPI (blue).(M) Quantification of Tuj1-positive neurons in the dorsal root ganglia showed no difference between the WT and KO samples.Scale bars: (A,B) and (E,F) 1000 µm, (K,L) 100 µm.

Figure 6 .
Figure 6.Loss of Igsf3 results in impaired development of intestinal muscularis propria.(A,B) Small intestines collected from the P9 WT and KO pups (N = 1 for each genotype) and stained with hematoxylin and eosin (H&E).(C-F) To visualize the developing muscle layers of the small intestine the sections were stained with antibodies against αSMA (blue in (C,D), red in (E,F).(C,D) show the whole intestinal section while the panels (E,F) show the higher magnification images.(G,H) Higher magnification H&E staining from a representative sample of the small intestine from WT (G) and KO (H) P12.5 pups (N = 3 for each genotype).The arrows indicate the thickness of the muscularis externa (muscularis propria).(I) Quantification of the thickness of the developing enteric muscularis externa from P12.5 pups shows a significant reduction in the KO intestine as compared to samples from WT littermate controls.N = 3 for each genotype.Data was measured from two sections/intestine, four images/section, and 10 measurements points per image.Scale bars; (C,D) 100 µm, (G,H) 20 µm.

Figure 7 .
Figure 7. Igsf3 KO results in downregulation of neuronal and smooth muscle markers and disorganized innervation of the small intestinal villi.(A) Confocal images of the WT P12.5 small intestinal villi after wholemount staining using antibodies against IGSF3 and NCAM1.IGSF3 (green) colocalizes with NCAM1 (red), a neuronal marker, in the neurons of the small intestinal villi (Merged).(B) Representative images of the intestinal villi from P12.5 WT (upper panels) and Igsf3 KO (lower panels) pups stained with antibodies against the neuron marker Tuj1 (green) and smooth muscle marker αSMA (red) on whole mount samples.(C,D) Representative images of the intestine from P12.5 WT (C) and Igsf3 KO (D) pups stained with antibodies against the E-cadherin (Ecad, green) and NCAM1 (red) on whole mount samples.Cell nuclei are visualized by Hoechst (blue).While no difference was seen in the E-cadherin expression between the WT and KO intestine, NCAM1 expression was substantially reduced in the KO compared to the WT.(E) Western blot analysis of P2.5 cerebrum extracts shows reduced expression of NCAM1 in KO samples also in the brain.Housekeeping protein GAPDH was used as a loading control.(F-K) Quantification from P12.5 intestinal immunostained images.(F) Expression of both Tuj1 and (G) αSMA was significantly reduced.(H) The αSMA:Tuj1 ratio was significantly increased in the KO villi as compared to WTs. (I) Counts of nerves per villus was significantly decreased in the KO villi as compared to the WT intestine, (J) but no difference was observed in the muscle fiber counts.(K) The colocalization of αSMA-Tuj1 was reduced in the KO villi as compared to WT intestines.Scale bars: (A-D) 100 µm.
. For tissue morphological assessment, hematoxylin and eosin staining was performed.Whole stained slides were digitally scanned using a slide scanner (3Dhistec).