Quantitative Spatial Analysis of Neuroligin-3 mRNA Expression in the Enteric Nervous System Reveals a Potential Role in Neuronal–Glial Synapses and Reduced Expression in Nlgn3R451C Mice

Mutations in the Neuroligin-3 (Nlgn3) gene are implicated in autism spectrum disorder (ASD) and gastrointestinal (GI) dysfunction, but cellular Nlgn3 expression in the enteric nervous system remains to be characterised. We combined RNAScope in situ hybridization and immunofluorescence to measure Nlgn3 mRNA expression in cholinergic and VIP-expressing submucosal neurons, nitrergic and calretinin-containing myenteric neurons and glial cells in both WT and Nlgn3R451C mutant mice. We measured Nlgn3 mRNA neuronal and glial expression via quantitative three-dimensional image analysis. To validate dual RNAScope/immunofluorescence data, we interrogated available single-cell RNA sequencing (scRNASeq) data to assess for Nlgn3, Nlgn1, Nlgn2 and their binding partners, Nrxn1-3, MGDA1 and MGDA2, in enteric neural subsets. Most submucosal and myenteric neurons expressed Nlgn3 mRNA. In contrast to other Nlgns and binding partners, Nlgn3 was strongly expressed in enteric glia, suggesting a role for neuroligin-3 in mediating enteric neuron–glia interactions. The autism-associated R451C mutation reduces Nlgn3 mRNA expression in cholinergic but not in VIPergic submucosal neurons. In the myenteric plexus, Nlgn3 mRNA levels are reduced in calretinin, nNOS-labelled neurons and S100 β -labelled glia. We provide a comprehensive cellular profile for neuroligin-3 expression in ileal neuronal subpopulations of mice expressing the R451C autism-associated mutation in Nlgn3, which may contribute to the understanding of the pathophysiology of GI dysfunction in ASD.


Introduction
Neuroligin-3 (NLGN3) is an adhesion molecule expressed at neuronal synapses in the central nervous system (CNS) and is well established as a regulator of brain function [1][2][3][4][5]. Although some studies suggest that NLGN3 is important for proper GI function [6,7], its cellular expression profile in the enteric nervous system has not been characterised. In cultured rat hippocampal neurons, NLGN3 is expressed at the postsynaptic membrane of both excitatory and inhibitory neuronal synapses [8]. NLGN3 is also expressed in nonneuronal cells. For example, Venkatesh and colleagues showed that neurally secreted NLGN3 invades the microenvironment of tumours and induces NLGN3 expression in glioma cells to promote tumour growth [9]. In addition, NLGN3 is expressed in many types of glia during rodent development, including olfactory ensheathing glia, retinal astrocytes, Schwann cells and spinal cord astrocytes [10]. These expression patterns suggest that NLGN3 could also be expressed in enteric glia.
Mutations in the NLGN3 gene, including the missense R451C point mutation as well as NLGN3 deletion, are implicated in autism [11][12][13]. Patients expressing the NLGN3 R451C mutation show GI dysfunction, including diarrhoea, faecal incontinence, post-meal regurgitation, oesophageal inflammation, chronic intestinal pain as well as delayed bladder and bowel control [7]. Modification of NLGN3 expression induces GI dysfunction in mice, as shown in both Nlgn3 knockout (KO) and Nlgn3 R451C mice [6,7]. Nlgn3 KO mice exhibit distended colons and more rapidly propagating colonic muscle contractions [6]. Nlgn3 R451C mice display faster intestinal transit and increased numbers of myenteric neurons in the small intestine as well as GABA A receptor-mediated colonic dysmotility [7]. These findings suggest a role for NLGN3 in gut function.
The R451C mutation impacts Nlgn3 mRNA and protein expression in the brain. In mouse whole-brain samples, Nlgn3 mRNA expression levels are unchanged by the Nlgn3 R451C mutation, but NLGN3 protein levels are reduced [1]. Although initial studies of NLGN3 expression in the GI tract have been reported in rodents [7,14,15] and humans [14], the distribution of NLGN3 expression in different cell populations of the mouse enteric nervous system has not been profiled. Therefore, the objective of this study was to provide comprehensive cellular profiling for neuroligin-3 expression in subpopulations of mouse myenteric and submucosal neurons using RNAScope, RNASeq data mining and immunofluorescence analyses. We also assessed for changes in neurochemical markers and Nlgn3 gene expression in the enteric nervous system of mice expressing the R451C autism-associated mutation in Nlgn3. Characterising Nlgn3 expression in enteric cell subtypes is essential for identifying underlying mechanisms contributing to GI activity and to functional GI changes observed in patients and the Nlgn3 R451C mouse model of autism.

Animals
B6;129-Nlgn3tm1Sud/J mice were obtained from The Jackson Laboratory [1] and bred for over 10 generations on a C57BL/6 background at the Howard Florey Institute, Melbourne, Australia. Mice were subsequently housed in the Biomedical Science Animal Facility at The University of Melbourne. Male Nlgn3 R451C and wild-type (WT) mice (age 12-13 weeks) were sacrificed by cervical dislocation in accordance with The University of Melbourne Animal Experimentation Ethics Committee (#1914843). For each set of neuronal markers analysed, a different group of WT and mutant mice were used. In total, 68 mice were used in the study. These numbers are noted for each experimental component within the results section and supplementary information Table S1 (63 mice in total for RNAscope; 34 WT and 29 Nlgn3 R451C mutant mice). In addition, 5 mice were included in the RNAseq study reported by Morarach et al. in 2021 [16].

Tissue Preparation
Mice were sacrificed, and the abdomen was opened using coarse dissection scissors. A 2 cm segment of the distal ileum (1 cm proximal to the caecum) was isolated and immediately placed in oxygenated ice-cold physiological saline solution (composition in mM: 118 NaCl, 4.6 KCl, 2.5 CaCl 2 , 1.2 MgSO 4 , 1 NaH 2 PO 4 , 25 NaHCO 3 , 11 D-glucose). The intestinal content was gently flushed clean using a Pasteur pipette. Using small dissecting scissors, the distal ileal tissue sample was cut along the mesenteric border and then stretched and pinned flat on a chilled silicon elastomer-lined dish (Sylgard 184, Dow Corning, Pennant Hills, NSW, Australia). The tissue was fixed in 4% formaldehyde for 24 h at 4 • C. After fixation, the tissues were washed three times in 1M phosphate buffer saline (PBS) (composition in mM: 137 NaCl, 2.7 KCl, 8 Na 2 HPO 4 , 2 KH 2 PO 4 ). The wholemount submucosal plexus preparation was isolated by peeling the mucosa-submucosal layer and then by removing the epithelium using fine forceps (Fine Science Tools, North Vancouver, CO, Canada). Myenteric plexus preparations including the longitudinal muscle layer (LMMP) were obtained by microdissection to remove the circular muscle layer.

Dual RNAScope In Situ Hybridization and Immunofluorescence
Localizing the NLGN3 protein in the mouse GI tract is challenging due to non-specific labelling of commercially available antibodies which commonly yield false-positive results in this tissue (Leembruggen et al., unpublished). Therefore, we used the enhanced in situ hybridization technique, RNAScope, to label Nlgn3 mRNA in enteric neurons. We combined RNAScope and immunofluorescence with high-resolution microscopy and 3D image analysis software to compare Nlgn3 mRNA expression in the wild-type (WT) and Nlgn3 R451C mouse gut.
Using RNAScope, individual RNA molecules are visualized using a novel probe design and amplification system to simultaneously magnify the signal and suppress background noise [17]. RNAScope in situ hybridization in mouse ileal tissue was performed using the ACDBio multiplex RNAScope assay kit (ACD, Newark, CA, USA) according to manufacturer's instructions. Briefly, wholemount mouse ileal submucosal and LMMP preparations were pre-treated and hybridized with the RNAScope probe of interest. The hybridization signal was subsequently amplified via a sequence of amplifiers and fluorescently labelled probes.

Tissue Pre-Treatment
Wholemount preparations of submucosal and LMMP preparations were permeabilized via pre-treatment with protease IV prior to RNAScope in situ hybridisation. The tissue samples were incubated in 30 µL of protease IV for 30 min at room temperature (RT) in a sealed humid chamber and washed twice (1 min in each) in 1 M PBS.

RNAscope In Situ Hybridization
Following pre-treatment, ileal tissue preparations were incubated in 20 µL of Nlgn3 RNAScope probe and incubated for 2 h at 40 • C in a humid chamber. The ileal tissues were washed twice in 1× wash buffer for 2 min at RT with agitation. Amplification and detection steps were performed using the RNAScope detection kit reagents. For the signal amplification step, tissues were incubated in 50 µL of amplifier 1 solution (AMP 1) for 30 min at 40 • C and washed twice in 1× wash buffer for 2 min at RT with occasional agitation. Subsequently, the tissue was incubated in 50 µL of AMP 2 solution for 15 min at 40 • C and then washed twice in 1× wash buffer for 2 min at RT with occasional agitation. After AMP 2 amplification, 50 µL of AMP 3 solution was added to the tissue preparations and incubated for 30 min at 40 • C. The tissue was washed twice in 1× wash buffer for 2 min at RT with occasional agitation. Lastly, the tissue was incubated in 50 µL of AMP 4 solution and incubated for 15 min at 40 • C prior to washing the tissue twice in 1× wash buffer for 2 min at RT with occasional agitation. The same protocol was followed for control experiments using the RNAScope universal negative control dapB probe (Supplementary Figure S1).

Immunofluorescence
After washing three times in PBS, the wholemount preparations were permeabilised with 1% Triton X-100 (ProSciTech, Kirwan, Australia) for 30 min at RT and washed three times in PBS. The tissues were then double labelled with various combinations of primary antibodies at different incubation conditions. Excess primary antibodies were removed by washing in PBS (3 × 10 min), and the tissue samples were subsequently incubated in secondary antibodies for 2 h at 4 • C. After removing excess antisera by washing with 0.1 M PBS (3 × 10 min), tissue preparations were mounted using Dakocytomation fluorescence mounting medium (DAKO; Carpinteria, CA, USA). The primary and secondary antibodies that were used to label these neuronal subpopulations and glial cells are listed in Tables 1 and 2, respectively. The number of animals used in each experiment and for different markers are included in the results section and supplementary information Table S1.

Submucosal Plexus Antibodies
Given that the submucosal plexus is predominantly involved in regulating secretory function in the GI tract, we endeavoured to study functionally relevant cell types in this region. It is well established that the two main groups of submucosal secretomotor neurons are cholinergic (i.e., expressing ChAT) and non-cholinergic (and expressing VIP) secretomotor neurons (e.g., Foong et al., 2014 [18]). Therefore, we utilised ChAT and VIP antisera to identify these prominent functional subtypes using immunofluorescence, in combination with RNAScope for Nlgn3 expression.

Myenteric Plexus Antibodies
Similarly, in the myenteric plexus, we aimed to study major functional neuronal subtypes. For example, inhibitory motor neurons and a subpopulation of descending interneurons in the myenteric plexus express nitric oxide; therefore, we used neuronal NOS (nNOS) antiserum to investigate the distribution of Nlgn3 in these neurons. In contrast, only 3% of neurons in the submucosal plexus are labelled with nNOS [19]; thus, the nNOS antiserum was not utilized in the submucosal plexus in the current study. In the myenteric plexus, calretinin is expressed in many different neuronal subtypes, including excitatory motor neurons, ascending interneurons, intrinsic sensory neurons [20]. In addition, calretinin is a useful marker in the myenteric plexus given it has been demonstrated that there is minimal co-expression between nNOS and calretinin in mouse small intestinal myenteric neurons (Sang and Young, 1996). Although calretinin is also expressed in non-cholinergic submucosal plexus neurons, we used VIP staining in the submucosal plexus to identify this neuronal group.
In the small intestine, Sang and Young [21] also found that almost all NOS neuronal cell bodies also contained VIP. For this reason, we utilized NOS to identify a major proportion of neurons in the myenteric plexus, but not VIP staining. In addition, VIP labelling of neuronal cell bodies in the myenteric plexus is unreliable. The ChAT antiserum was not utilized in the myenteric plexus given that this neuronal group corresponds to cells that do not express NOS (i.e., more than 90% of non-NOS labelled myenteric neurons in the mouse small intestine are known to contain ChAT [21]); therefore, the use of ChAT in the myenteric plexus would not have yielded additional information regarding neuron types.

Image Acquisition and Analysis
Multi-channel image acquisition was performed with a laser scanning confocal microscope 800 (Carl Zeiss Microscopy, North Ryde, NSW, Australia) using a 40× oil immersion objective lens. Tissue samples were excited with diode lasers at 488, 561 and 647 wavelengths. Pinhole diameter, detector gain and laser power parameters were optimised to obtain the highest pixel intensity at the same time eliminating pixel saturation. Consecutive Z stacks on the horizontal plane with a frame size of 1024 × 1024 pixels and a bit-depth of 16 bits were captured at 1 µm intervals. Z-stacks together with tile scans were performed where necessary.
Quantification of labelling for RNAscope puncta and antisera was performed using 3D reconstruction tools and statistical annotation for quantification purposes in Imaris 9.0 image analysis software (Bitplane, Zurich, Switzerland). Three-dimensional cell rendering was undertaken using high-resolution Z-stack fluorescent images generated using confocal microscopy. To create a detailed rendered surface, the "Surface" tool in the Imaris software package was used. The minimum diameter of the surface was determined based on the average cell diameter estimated using the "Slice view". The interactive software histogram within the "Create surface" window was used to set a threshold (the minimum diameter of the neuronal surface) to exclusively include neuronal surfaces and to exclude smaller surfaces generated by the background noise. Aggregated neurons within images were separated using the 'Split surfaces" tool in Imaris. The Imaris software is additionally equipped with "Filter options" to label specific populations of surfaces either manually or automatically, and these were utilized in the current study to identify specific cell populations in any given image. Since the RNAScope signal representing Nlgn3 mRNA appears as puncta, the "spot" analysis tool in Imaris was used to create the 3D structure of individual mRNA molecules. The "average diameter of spots" detected was determined using "slice view" in Imaris. Once spots were generated, they were assigned as being located within the neurons using the "Split spots on to surface" tool. The Imaris 9.0 software additionally generates statistical data yielding the number of spots per surface. This option was used to identify the number of mRNA punctate signals within the neurons for the images analysed in the current study.

Statistical Analysis
Data were analysed using GraphPad Prism 8.4.1 (GraphPad software, San Diego, CA, USA). A different group of animals was used to investigate labelling for each set of enteric cellular subpopulations. Frequency distribution analysis was performed to analyse the distribution of Nlgn3 mRNA expression in enteric neurons and glia, and a two-sample Kolmogorov-Smirnov test was conducted to determine statistical significance. In general, to estimate the extent of overlap between two distributions for Nlgn3 mRNA expression between genotypes, the magnitude of the Kolmogorov-Smirnov D statistic (distance; D) relative to the critical value is obtained. D is a measure of the maximum distance between the empirical cumulative distribution function of the sample and the cumulative distribution function of the reference distribution (if the value of D is larger than the critical value, the null hypothesis is rejected). Analysis using the GraphPad Prism algorithm provides a p value calculated based on the maximum distance between the cumulative frequency distributions, considering both D and the D critical value.

Investigation of Nlgn, Nrxn and MDGA Gene Family Expression in Single-Cell RNA Sequencing Datasets
To further validate the RNAScope findings in the current study, we analysed published scRNA sequencing data identifying molecular and functional subtypes of enteric neurons in mice [16] (http://ncbi.nlm.nih.gov/sra/SRP258962; URL accessed on 4 September 2022) Biomolecules 2023, 13, 1063 6 of 23 and Zeisel et al. [22] (https://www.ncbi.nlm.nih.gov/sra/SRP135960; URL accessed on 4 September 2022). ScRNASeq was performed on isolated neurons from myenteric peels of mice aged postnatal day (P)21, and marker genes for neuron types were validated using immunofluorescence in the ENS of P21-90 mice [16]. The transcriptome of single enteric glia cells isolated from P21 and P23 myenteric plexus was assessed using mousebrain.org (web interface for browsing data presented in Zeisel et al., 2018 [22]).

Results
Here, we characterized the cellular localization of Nlgn3 mRNA in the enteric nervous system of the mouse distal ileum. Using RNAscope and immunofluorescence in WT mice, we first determined the frequency distribution and the average number of Nlgn3 mRNA copies per cell in neuronal subpopulations and glia. Using scRNASeq datasets, we then identified enteric neuronal subsets expressing Nlgn3 and their binding partners, the neurexins (Nrxns1, 2 and 3) and MAM domain-containing glycosylphosphatidylinositol anchor genes (MDGA1 and 2). Subsequently, we investigated whether the R451C missense mutation in Nlgn3 impacts cellular expression profiles in the enteric nervous system of the mouse ileum using RNAscope and immunofluorescence.

Nlgn3 mRNA Is Expressed in Most Distal Ileal Submucosal Neurons
All submucosal neurons in distal ileal preparations were labelled with the panneuronal marker Hu. We then analysed the distribution of Nlgn3 mRNA expression in somata of individual submucosal neurons using RNAscope (Figure 1(A1-A6)). Of 1658 submucosal neurons assessed in 14 WT mice, 1555 (94%) expressed Nlgn3 mRNA (Figure 1(A7)). In addition to Nlgn3 mRNA localized to the cell soma, Nlgn3 mRNA expression was observed in neuronal and glial fibres (Supplementary Figure S2).
To assess for differential distribution patterns of Nlgn3 mRNA within the submucosal plexus, the expression profile of  To assess for differential distribution patterns of Nlgn3 mRNA within the submucosal plexus, the expression profile of Nlgn3 mRNA in cholinergic neurons (labelled using

ScRNASeq Confirmation of Nlgn3 Expression in Enteric Neurons and Glia
Our dual RNAScope/immunocytochemistry findings show that most enteric neurons express Nlgn3 mRNA in mice. To confirm this finding, we investigated both the distribution and levels of Nlgn3 expression in a published scRNASeq dataset describing molecular cellular subtypes in the juvenile mouse small intestine [16]. In line with our observations using dual RNAScope/immunohistochemistry, scRNASeq data analysis shows Nlgn3 expression in multiple enteric neuron classes (ENC), including putative intrinsic primary afferent neurons, excitatory and inhibitory motor neurons and interneurons. Given that NLGN3 is thought to form heterodimers, predominantly with NLGN1 in the brain [23], expression patterns for Nlgn1 and Nlgn2 in the scRNASeq atlas were also analysed. Unlike in the brain, Nlgn2 and Nlgn3 have largely overlapping expression profiles in the enteric nervous system ( Figure 4A). Nlgn1 is expressed in most enteric neuron types, but this expression is less prominent in the ENC4, 11, 12 cell classes. Analysis of the scRNASeq atlas showed that although Nlgn3 is detected across all neuronal clusters, Nlgn3 expression levels are highest in ENC6 cells ( Figure 4B). Similarly, Nlgn3 is expressed in the highest proportion of cells in ENC6 (approximately 50% of ENC6 cells express Nlgn3) compared to other ENCs ( Figure 4C). Based on morphological attributions and projection patterns, ENC6 was validated to correspond to IPANs (intrinsic sensory neurons) of the myenteric plexus [16]. These findings validate our dual RNAScope/immunohistochemistry results to confirm wide-spread expression of Nlgn3 mRNA in the enteric nervous system and suggest that Nlgn3 is highly expressed in some IPANs. crine) cells in the intestine [15] but whether Nlgn3 is expressed in enteric glia has not been assessed. We therefore labelled submucosal glia using an antiserum against the calciumbinding protein, S100β, and assessed for the co-presence of Nlgn3 mRNA (Figure 3(A1-A4)). Overall, we found that 31% (70 of a total of 226 cells in 6 WT mice) of submucosal glial cells in the mouse ileum express Nlgn3 mRNA (Figure 3(A1-A9)). Similarly, Nlgn3 mRNA expression in ileal myenteric glial cells (1203 glial cells in 4 WT mice) was assessed ( Figure 3(B1-B8)). In the myenteric plexus, 55% of glial cells (666 of 1203 cells) expressed Nlgn3 mRNA (Figure 3(B9)).  The expression patterns of neuroligin binding partners provide insight into their function in the enteric nervous system. Neurexins are major binding partners of the neuroligin gene family in the brain [23]. Within the enteric nervous system, Nrxn1 and Nrxn2 are almost ubiquitously expressed across all ENCs, including robust expression in ENC6 (IPANs). Specifically, Nrxn1 expression levels are highest in ENC11 (interneurons; INs) whereas Nrxn2 expression is highest in ENC6 (IPANs), closely followed by strong expression levels in ENCs 1,2 and 4 (putative motor neurons; MNs), 5 (INs) and 7 (IPANs/INs). Nrxn3 expression levels are highest in ENC 6, 7 (IPANs/INs) and 10 (INs) but are less prominent in other enteric neuronal subtypes ( Figure 4C). As lower affinity binding partners of neuroligins, MAM domain-containing glycosylphosphatidylinositol anchor genes (Mgda1 and Mdga2) are expressed at varying levels in myenteric neuron populations (Supplementary Figure S3).
The distribution and expression of Nlgn3 alongside Nlgn1 and -2 was also examined in myenteric glia based on scRNASeq data [22]. Nlgn1 was sparsely expressed in myenteric glial cells ( Figure 5A). Although Nlgn2 was present across all myenteric glia (in a greater proportion of cells than Nlgn1), fewer than 50% of cells assayed contained Nlgn2 ( Figure 5B). In stark contrast, Nlgn3 was strongly expressed across each myenteric glial cluster (Figure 5C,D). However, of the 10,535 total myenteric glial cells counted, only 4054 cells (38%) express Nlgn3 mRNA according to the scRNASeq analyses compared to 65% of glial cells as detected using RNAScope. These differences may be due to the lower sensitivity of the scRNASeq approach utilized [22] compared to RNAScope. None of the Nrxns (Figure 5E-G) nor Mgda1 or Mdga2 (Supplementary Figure S4) were notably expressed in myenteric glia. sion levels are highest in ENC6 cells ( Figure 4B). Similarly, Nlgn3 is expressed in the highest proportion of cells in ENC6 (approximately 50% of ENC6 cells express Nlgn3) compared to other ENCs ( Figure 4C). Based on morphological attributions and projection patterns, ENC6 was validated to correspond to IPANs (intrinsic sensory neurons) of the myenteric plexus [16]. These findings validate our dual RNAScope/immunohistochemistry results to confirm wide-spread expression of Nlgn3 mRNA in the enteric nervous system and suggest that Nlgn3 is highly expressed in some IPANs.  The expression patterns of neuroligin binding partners provide insight into their function in the enteric nervous system. Neurexins are major binding partners of the neuroligin gene family in the brain [23]. Within the enteric nervous system, Nrxn1 and Nrxn2 are almost ubiquitously expressed across all ENCs, including robust expression in ENC6 (IPANs). Specifically, Nrxn1 expression levels are highest in ENC11 (interneurons; INs) whereas Nrxn2 expression is highest in ENC6 (IPANs), closely followed by strong expression levels in ENCs 1,2 and 4 (putative motor neurons; MNs), 5 (INs) and 7 (IPANs/INs). Nrxn3 expression levels are highest in ENC 6, 7 (IPANs/INs) and 10 (INs) but are less prominent in other enteric neuronal subtypes ( Figure 4C). As lower affinity binding partners of neuroligins, MAM domain-containing glycosylphosphatidylinositol anchor genes (Mgda1 and Mdga2) are expressed at varying levels in myenteric neuron populations (Supplementary Figure S3).
The distribution and expression of Nlgn3 alongside Nlgn1 and -2 was also examined in myenteric glia based on scRNASeq data [22]. Nlgn1 was sparsely expressed in myenteric glial cells ( Figure 5A). Although Nlgn2 was present across all myenteric glia (in a greater proportion of cells than Nlgn1), fewer than 50% of cells assayed contained Nlgn2 ( Figure 5B). In stark contrast, Nlgn3 was strongly expressed across each myenteric glial cluster ( Figure 5C,D). However, of the 10,535 total myenteric glial cells counted, only 4054 cells (38%) express Nlgn3 mRNA according to the scRNASeq analyses compared to 65% of glial cells as detected using RNAScope. These differences may be due to the lower sensitivity of the scRNASeq approach utilized [22] compared to RNAScope. None of the Nrxns ( Figure 5E-G) nor Mgda1 or Mdga2 (Supplementary Figure S4) were notably expressed in myenteric glia.

The R451C Mutation Selectively Reduces Nlgn3 mRNA Expression
The R451C mutation reduces Nlgn3 mRNA expression in cholinergic submucosal neurons but not in VIPergic neurons.
We next investigated Nlgn3 mRNA expression in wild-type and Nlgn3 R451C mutant mouse ileum using dual RNAScope and immunocytochemistry plus Imaris-based quantification ( Figure 6(A1-A6)). Frequency distribution analysis of Nlgn3 mRNA expression

The R451C Mutation Selectively Reduces Nlgn3 mRNA Expression
The R451C mutation reduces Nlgn3 mRNA expression in cholinergic submucosal neurons but not in VIPergic neurons.
We next investigated Nlgn3 mRNA expression in wild-type and Nlgn3 R451C mutant mouse ileum using dual RNAScope and immunocytochemistry plus Imaris-based quantification ( Figure 6(A1-A6)). Frequency distribution analysis of Nlgn3 mRNA expression in the submucosal neurons labelled with Hu showed that Nlgn3 mRNA expression is significantly different and that the distribution was skewed to the left in Nlgn3 R451C mutant mice compared to WT (1985 neurons in 14 WT and 1300 neurons in 10 Nlgn3 R451C mice were analysed; p < 0.0001; Figure 6(A7)). In Nlgn3 R451C mutant mice, submucosal neuronal cell bodies (n = 1300 neurons) express 21.6 ± 0.8 copies of Nlgn3 mRNA, indicating that significantly less Nlgn3 mRNA is present in this neuronal subset in mutants than in WT mice (24.8 ± 1.0 copies, n = 1985 neurons), p = 0.02; t = 2.2, df = 3283 Figure 6(A8)).
The impact of the R451C mutation on Nlgn3 mRNA production in VIP-expressing neurons was also evaluated ( Figure 6(C1-C8)). The frequency distribution of the Nlgn3 mRNA copy numbers in VIP neurons in ileal tissue from Nlgn3 R451C mice (n = 623 neurons in seven WT mice were analysed as well as n = 449 neurons in 5 Nlgn3 R451C mice) is skewed to the left and differs significantly from WT (p < 0.0001; Figure 6
Overall, we found that Nlgn3 mRNA is expressed in the mouse enteric nervous system (including enteric glia) and that Nlgn3 R451C mice have reduced Nlgn3 mRNA expression in subpopulations of submucosal and myenteric plexus neurons ( Table 3). The molecular identities of enteric neuronal populations have now been well described in mice based on scRNASeq combined with immunostaining verification [16,22]. These studies show the presence of 12 enteric neuronal populations with distinct neurochemical signatures and enteric glia. We therefore investigated the cellular expression of Nlgn3 in the mouse enteric nervous system based on previously reported scRNASeq datasets and found that all enteric neuronal subtypes and glia express Nlgn3 at varying levels. In mice, the R451C mutation reduces Nlgn3 mRNA expression in submucosal cholinergic as well as myenteric nitrergic and calretinin neurons. Nlgn3 R451C mice also had reduced mRNA expression levels in myenteric (but not submucosal) glia. In contrast, Nlgn3 mRNA expression was unchanged in VIPergic submucosal neurons in Nlgn3 R451C mice. Taken together, these findings suggest that NLGN3 is an important component of neuronal and neuron-glial communication in the gut and that mutations in NLGN3 might play a role in GI dysfunction in individuals with autism.

Discussion
In this study, we showed that most submucosal and myenteric neurons in the mouse ileum express Nlgn3 mRNA. For the first time, we revealed that Nlgn3 mRNA is expressed in ileal submucosal and myenteric glia. In addition, we report that the autism-associated Nlgn3 R451C mutation reduces Nlgn3 mRNA expression in both submucosal and myenteric neurons as well as myenteric glia in the mouse ileum.

Most Enteric Neurons Express Nlgn3 mRNA
Although there is ample evidence for the existence of prominent postsynaptic densities within the ENS [25][26][27][28], the molecular characterization remains to be fully characterized. In the rodent brain, neuroligin-3 is expressed at both excitatory and inhibitory synapses [8] and is closely associated with other PDZ-associated molecules to mediate postsynaptic signal transduction. There is evidence that PDZ domain proteins such as postsynaptic density protein 95 (PSD95) [29,30], postsynaptic density protein 93 (PSD93) [30], as well as the cell adhesion molecules L1 [31] and neuroligin-1 [32] are expressed in the ENS.

Cholinergic Transmission
It is well established that NLGN3 is involved in both excitatory and inhibitory synaptic transmission in the brain. However, relatively little is known about the distribution of postsynaptic proteins associated with excitatory (predominantly mediated via nicotinic acetylcholine receptors (nAChRs)) and inhibitory synapses in the ENS. We reveal that most cholinergic submucosal neurons also contain Nlgn3 mRNA. Within the enteric neural circuitry, acetylcholine is the primary excitatory neurotransmitter utilised by cholinergic secretomotor neurons, excitatory muscle motor neurons, ascending interneurons, descending interneurons and intrinsic sensory neurons [33][34][35][36]. Since virtually all submucosal neurons receive synaptic inputs via cholinergic synapses [18], Nlgn3 is likely present in cholinergic synapses in the submucosal plexus. All submucosal neurons receive cholinergic fast excitatory postsynaptic potentials and hence can be expected to have postsynaptic densities with clusters of nicotinic receptors and NLGNs as a part of the synaptic structure. Although there is no available evidence of NLGN3 expression in the cholinergic system of the central or peripheral nervous systems, expression of other NLGN subtypes in cholinergic synapses has been reported. For example, NLGN2 is expressed at the postsynaptic membrane of cholinergic synapses in the mouse brain [37], and NLGN1 is present in cholinergic synapses in the chick ciliary ganglion [38]. Since NLGN3 and other NLGN isoforms can be colocalized in the same synapse [8], NLGN3 might be co-expressed with NLGN1 and NLGN2 in cholinergic synapses in the submucosal plexus.

VIP-Expressing Neurons
We also revealed that Nlgn3 mRNA is expressed in VIP-expressing submucosal neurons in the mouse ileum. In the submucosal plexus, VIP is a primary neurotransmitter of secretomotor neurons and stimulates intestinal secretion [18,39]. In the myenteric plexus, VIP is expressed in inhibitory muscle motor neurons and interneurons [21,40]. VIP also acts as a co-transmitter alongside ACh and NO in a subset of descending interneurons in the mouse and guinea pig ileum [20,21,41]. Moreover, an excitatory role for the vasoactive intestinal peptide 1 (VPAC1) receptor on cholinergic neurons has been identified in the myenteric plexus [42]. These findings open a novel avenue for research in Nlgn3 R451C mice since the expression profiles of NLGN3 or other NLGN isoforms in VIP-containing synapses have not been reported to date in either the CNS or the peripheral nervous system.

nNOS-Expressing Neurons
One of the major findings of this study is that most NO-containing enteric neurons express Nlgn3 mRNA in the cell soma. Nitric oxide (NO) is the predominant inhibitory neurotransmitter to the smooth muscle in the enteric nervous system [20,21] and is also expressed in descending interneurons together with other neurotransmitters [43]. Given that NLGN complexes anchor postsynaptic densities and PSD95 has at least one nNOS binding PDZ domain, Nlgn3 may contribute to synaptic specialization of enteric NO neurons. However, the distributions of PSD95 and PSD93 have not been well characterized in ENS. It has been shown that PSD93 interacts with nNOS via a PDZ-PDZ domain interaction [30]. Some nNOS expressing myenteric neurons express PSD93 [44]. In agreement with this finding, our data indicating Nlgn3 expression in this neuronal population suggest that Nlgn3 might be expressed at the postsynaptic membrane of nNOS immunoreactive myenteric neurons.

Calretinin-Expressing Neurons
We also show Nlgn3 mRNA expression in calretinin-expressing myenteric neurons in the mouse ileum. In the enteric nervous system, calretinin is expressed in intrinsic primary afferent neurons, interneurons and excitatory motor neurons innervating the smooth muscle layer [20,21]. In line with these findings, scRNASeq data analysis confirmed expression of Nlgn3 at varying levels in mouse myenteric neurons, including putative motor neurons, interneurons and intrinsic primary afferent neurons in the mouse small intestine. It has been reported that the PSD93 protein is localized to calretinin neurons in mouse myenteric plexus [44]. Therefore, it is possible that Nlgn3 is associated with postsynaptic protein complexes within calretinin neurons in the myenteric plexus.

Other Signalling Systems
Given that glutamate and GABA have active roles in the ENS [45,46] alongside the presence of postsynaptic proteins including PSD93 [44], the wide distribution of NLGN3 we observed in the ENS suggests that NLGN3 could also be expressed at glutamatergic and GABAergic synapses within the ENS.

Differential Expression of Neuroligin Binding Partners in the Enteric Nervous System
When examining scRNASeq data, Nlgn2 and 3 expression profiles are almost identical in all classes of enteric neurons. In contrast with findings in the central nervous system [8] reviewed by [47], this suggests that NLGN3 and NLGN2 co-express in enteric neurons subtypes, and we speculate that they may form heterodimers in the enteric nervous system. Expression patterns of the neurexin family of neuroligin binding partners, however, show that Nrxn2 and Nrxn3 are more strongly expressed within the enteric neuronal cell clusters compared with Nrxn1. These findings suggest that Nlgn3 may preferentially bind with Nrxn2 and 3 at synapses in IPANs of the myenteric plexus. In the CNS, NLGNs are present at classical synapses receiving fast excitatory postsynaptic potentials (EPSPs), but the most functional evidence suggests that IPANs do not exhibit fast EPSPs. For example, findings from Hibberd and colleagues [48] indicate that IPANs receive fast EPSPs, but in contrast, both [49,50] reported that AH neurons do not exhibit fast EPSPs. Nevertheless, it is unclear whether junctions producing slow EPSPs might express neuroligin complexes; therefore, further research is required to identify the role of NLGN3 in IPANs in the ENS.

A Role for Nlgn3 in Enteric Neuronal-Glial Synapses
For the first time, we reveal that Nlgn3 mRNA is expressed in enteric glia. The higher proportion of myenteric glia represents approximately a three-fold enrichment of Nlgn3 expression compared to submucosal Nlgn3-expressing glia and may provide a useful way to chemically identify and assist in identifying the function of glial subtypes in the gut. scRNAseq analysis showed that most enteric glia express Nlgn3 mRNA but negligible/low levels of Nlgn1 and Nlgn2. Therefore, we propose that NLGN3 acts as a main adhesion protein in glia for modulating glial-neuron synaptic activity in the enteric nervous system. Enteric glia play a major role in enteric nervous system-mediated GI functions, including mucosal secretion, intestinal permeability, mucosal sensation, GI motility and immune responses [51][52][53] in concert with enteric neurons. Enteric neurons synapse onto enteric glia to regulate enteric nervous system-coordinated GI responses [51,54,55]. Enteric glial-neuronal associations revealed by immunocytochemistry highlight that enteric glia are in close contact with nerve fibres and varicose release sites [56]. Evidence for neuronglia communication also comes from live imaging experiments reporting the presence of several signalling pathways. Ca 2+ imaging studies indicate that neuron-glia communication is regulated by neurally released purines which activate purinergic receptors on enteric glia [55][56][57][58]. In the submucosal plexus, neuron-glial transmission occurs via P2Y 1 and P2Y 4 receptors [59]. Furthermore, neuronal ATP release via the pannexin-1 channel also influences neuron-glia interactions [60,61]. Based on this evidence, we propose that NLGN3 expressed by both enteric neurons and glia modulate neuron-glia signalling pathways. This hypothesis is well established in the mouse brain where RNA sequencing of the transcriptome revealed that Nlgn3 transcripts are enriched in glial cell types, including astrocytes and oligodendrocytes [62]. NLGNs expressed in astrocytes aid in neuron-astrocyte communication via bi-and tripartite synapses in mice in C. elegans [63], suggesting that a neuronal-glial communication role may not be specific to mice. Similarly, NLGN3 expressed in enteric glia could also be involved in establishing bi-partite and tripartite synapses in the enteric nervous system. Given this evidence, this study strongly suggests that NLGN3 might play a role in synapse formation and modifying synaptic function during neuron-glia communication in the enteric nervous system.

The R451C Mutation Reduces Nlgn3 mRNA Expression in the Enteric Nervous System
We previously demonstrated that mice expressing the Nlgn3 R451C mutation have faster small intestinal transit and increased numbers of myenteric neurons in the small intestine [7]. How this mutation affects NLGN3 expression in the enteric nervous system, however, is not understood. Here, we show that the R451C mutation decreases Nlgn3 mRNA expression in both neurons and glia in the mouse ileal submucosal plexus. For example, cholinergic submucosal neurons in mutant mice contain fewer Nlgn3 mRNA copies compared to WT. These changes could potentially alter cholinergic signalling in the submucosal plexus, which could affect functions mediated by cholinergic neurons such as secretion, absorption and mucosal barrier functions. Although there are no reports available on changes to the cholinergic system of the enteric nervous system in ASD, CNS studies showed that altered cholinergic neurons are associated with the pathophysiology of autism [64,65]. Specifically, ASD patient basal forebrain tissues show altered cholinergic neuronal numbers, size and structure [66]. In addition, a decreased plasma concentration of choline, a precursor for acetylcholine, has been reported in ASD patients [67,68], and reduced levels of hippocampal cytosolic choline have been correlated with autism severity [67]. In the human GI tract, NLGN3-mediated alterations to the cholinergic system could induce GI dysfunction; however, further studies are required to determine effects of the R451C mutation on cholinergic signalling in the enteric nervous system.
In the myenteric plexus of the distal ileum, the R451C mutation reduces Nlgn3 mRNA expression levels in neurons. Specifically, Nlgn3 mRNA expression is substantially reduced in both calretinin and nNOS-immunoreactive neuronal populations. Although the contribution of calretinin neurons to ASD or ASD-related GI pathophysiology is unclear, the significance of calretinin neurons in other ENS-related diseases has been highlighted [69]. As mentioned, an increased proportion of NOS1 expressing neurons has previously been observed in Nlgn3R451C mice [7], and our current findings showing reduced Nlgn3 expression in these cells might contribute to altered NO neuronal signalling in the enteric nervous system in these mice.
Finally, the impact of the R451C mutation on Nlgn3 expression in glia has not previously been reported. Here, we show that in Nlgn3 R451C mice, Nlgn3 mRNA expression is dramatically reduced in myenteric glia but is unchanged in glia located within the submucosal plexus. Since glia play an important role in mediating intestinal functions such as mucosal barrier regulation, gut motility, immune responses and neurotransmission, such a reduction in NLGN3 levels could contribute to GI dysfunction in Nlgn3 R451C mutant mice. Therefore, in addition to neuronal dysfunction, glial dysfunction could also contribute to GI pathology in individuals diagnosed with ASD.

Conclusions
Here, we characterised Nlgn3 mRNA expression and identified effects of the autismassociated R451C mutation in this gene in the enteric nervous system of the mouse ileum. We show that Nlgn3 mRNA is expressed in situ in most enteric neuronal and glial populations of the myenteric and submucosal plexuses in the mouse ileum. Our RNAseq analysis confirms the presence of Nlgn3 in myenteric neuronal subtypes and glia in mice. We found that Nlgn3, but not Nlgn1 or 2, is expressed at high levels in enteric glia, whereas glial ex-pression of major binding partners for neuroligins, such as neurexins 1-3, is negligible. We therefore propose that neuroligin-3 is an important glial adhesion protein in neuronal-glial synapses in the mouse enteric nervous system. We demonstrate that the R451C mutation differentially reduces Nlgn3 mRNA expression in the enteric nervous system, particularly in cholinergic submucosal neurons, myenteric calretinin and NOS1-expressing neurons and myenteric glia in the mouse ileum. Taken together, these findings suggest that neuroligin-3 plays a role in regulating enteric nervous system-mediated GI physiology and that changes to Nlgn3 mRNA expression could contribute to GI dysfunction in the Nlgn3 R451C mouse model of autism.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/biom13071063/s1, Figure S1: Negative control experiments for the myenteric plexus and the submucosal plexus; Figure S2: Nlgn3 mRNA expression in submucosal neurons, glia and glial fibers; Figure S3: Expression of NLGN3 potential binding partners MDGA1 and 2 in mouse myenteric neurons; Figure S4: Expression of NLGN3 potential binding partners MDGA1 and 2 in mouse enteric glia; Table S1: Number of mice used for identifying enteric nervous system cellular subtypes.

Informed Consent Statement: Not applicable.
Data Availability Statement: Data supporting the findings of this study are available from the corresponding author upon request.