Characterizing enteric neurons in Dopamine Transporter (DAT)-Cre reporter mice reveals dopaminergic subtypes with dual-transmitter content

The enteric nervous system (ENS) comprises a complex network of neurons whereby a subset appears to be dopaminergic, although the characteristics, roles, and implications in disease are less understood. Most investigations relating to enteric dopamine (DA) neurons rely on immunoreactivity to tyrosine hydroxylase (TH) - a rate-limiting enzyme in the production of DA. However, TH immunoreactivity is likely to provide an incomplete picture given previous work has showed that some DA neurons contain little if any TH and its levels tend to be decreased in response to cellular stress. This study herein provides a comprehensive characterization of DA neurons in the gut using a well-accepted reporter mouse line, expressing a fluorescent protein (tdTomato) under control of the DA transporter (DAT) promoter. Our findings confirm a unique localization of DA neurons in the gut and unveil the discrete subtypes of DA neurons in this organ, which we characterized using both immunofluorescence and single-cell transcriptomics, as well as validated using in situ hybridization. We observed distinct subtypes of DAT-tdTomato neurons expressing co-transmitters and modulators across both plexuses; some of them likely co-releasing acetylcholine, while others were positive for a slew of canonical DA markers (TH, VMAT2 and GIRK2). Interestingly, we uncovered a seemingly novel population of DA neurons unique to the ENS which were ChAT/DAT-tdTomato-immunoreactive neurons and were characterised by the expression of Grp, Calcb and Sst. Given the clear heterogeneity of DAergic gut neurons, further investigation is warranted to define their functional signatures and discover any inherent vulnerabilities in disease. Using a reporter mouse line, expressing a fluorescent protein under control of the dopamine transporter (DAT) promoter, discrete subtypes of dopaminergic neurons were unveiled across the ganglionated plexuses of the gut. A novel subpopulations of enteric DA neurons, expressing genes previously reported involved in dopamine signaling in the brain, exhibit a cholinergic phenotype.


Funding information
Aligning Science Across Parkinson's, Grant/Award Number: 000525 Edited by: Yoland Smith and Sst.Given the clear heterogeneity of DAergic gut neurons, further investigation is warranted to define their functional signatures and decipher their implication in disease.
K E Y W O R D S dopamine, enteric nervous system, gut, Parkinson's disease, single-cell RNA sequencing

| INTRODUCTION
Recent single-cell transcriptomics studies in both human and mice have started to explore the complexity of the cellular landscape within the GI tract, including defining unique gene expression signatures within the neuronal lineage (Drokhlyansky et al., 2020;May-Zhang et al., 2021;Memic et al., 2018;Morarach et al., 2021;Wright et al., 2021).However, very few of these studies have included the identification of genes involved in DA synthesis, such as tyrosine hydroxylase (TH) (Memic et al., 2018;Morarach et al., 2021;Wright et al., 2021).Amongst these studies, only one has explored this population, where they describe novel markers in the developing ENS of foetal mice, but it is not clear if this is translatable to the expression patterns of mature DA neurons (Memic et al., 2018).As such, a rigorous study describing the diversity of DA neuronal populations in the GI tract has not yet been carried out.Historically, a neuronal subtype is defined by the primary neurotransmitter it releases and its functional properties.With the gradual appreciation that many if not most neuronal types in the CNS and PNS utilize more than one type of neurotransmitter (El Mestikawy et al., 2011;Trudeau, 2004;Trudeau & El Mestikawy, 2018;Trudeau & Gutiérrez, 2007), and with the advent of singlecell transcriptomic approaches, assigning a molecular gene signature is progressively being used to precisely define neuronal subclasses.In the case of midbrain DA neurons, previous microarray and recent single-cell RNA sequencing (RNAseq) work have revealed extensive heterogeneity including in genes specifying their developmental origin, as well as a battery of genes involved in the synthesis and vesicular packaging of DA, but also of other neurotransmitters including glutamate and various neuropeptides (Gaertner et al., 2022;Kouwenhoven et al., 2020;Poulin et al., 2014Poulin et al., , 2018Poulin et al., , 2020)).
The enteric nervous system (ENS) is a network of neurons situated in two separate regions of the gastrointestinal wall: the submucosal plexus (SMP) found underneath the mucosa and the myenteric plexus (MP) located within the muscularis propria.The two plexuses are responsible for many functions of the gut; the SMP plays a crucial role in secretion and luminal sensing, whereas the MP regulates gastrointestinal motility through peristaltic waves.These complex intestinal functions require a diverse population of neurons including TH-positive DA neurons as well as serotonin, cholinergic and nitrergic neurons (Furness et al., 2014).Although there are several reports of the presence of DA neurons in the gut (Li et al., 2004;Walker et al., 2000), such studies typically rely solely on immunostaining for TH-the rate-limiting enzyme in the synthesis of catecholamines, to define a DAergic phenotype.Further, DA neuron subtyping analysis has only been characterized, to some extent, in one part of the GI tract, the murine ileum, demonstrating that some TH-expressing neurons are immunoreactive for the DA transporter (DAT) (and others are not), and vice versa (Li et al., 2004).Taken together, there is a clear need for a rigourous study describing the heterogeneity of enteric DA neurons.
The use of transgenic reporter mouse lines such as those in which DAT expression drives Cre recombinase to induce the expression of fluorescent molecules, such as tdTomato, has the potential to shed new light on enteric DA neuron diversity.By taking advantage of DAT-Cre-tdTomato mice, we sought to deepen our understanding of DA neurons and their putative subtypes in the ENS.We conducted immunofluorescence staining in colonic tissues along with single-cell RNAseq and in situ hybridization to unravel the heterogeneity of DAergic neuronal populations.Providing an in-depth characterization of these classes of enteric neurons will be conducive to future endeavours aimed at better understanding their functional signatures and any potential selective vulnerabilities of enteric neurons in disease.

| Mice and tissue processing
DAT-Ires-Cre (RRID:IMSR_JAX:006660) mice were bred with Ai9 mice (RRID:IMSR_JAX:007909) to induce the expression of tdTomato in DAT-expressing cells (DAT-Cre-tdTomato mice).Adult DAT-Cre-tdTomato mice (both sexes) of age 2-3 months were utilized in the present study.Mice were housed under pathogen-free conditions, given water and food ad libitum.Studies were approved by the Université de Montréal Animal Care and Use Committee (CDEA).CDEA guidelines on the ethical use and care of animals were followed.The mice were euthanized by isoflurane followed by CO 2 asphyxiation.Age-matched wild type C57Bl6/J mice (RRID: IMSR_JAX:000664) were also used as control for tdTomato expression.We collected gut tissue and focused on colons where fewer investigations of the ENS have been made.After isolation, faecal material was flushed with an 18G needle and adipose tissue was removed.Colons were cut open longitudinally and coiled around themselves, resembling a swiss-roll (as depicted in Figure 1a), then fixed in 4% paraformaldehyde (PFA) for 24 h and dehydrated with a 30% sucrose solution.The tissue was then embedded and cryosectioned serially to obtain coronal sections at 15 μm, and then stored at À80 C. Of note, we used a swiss-roll preparation to easily compare DAergic neurons between ganglionated plexuses within the same tissue section.This is of importance as previous studies have underlined the varying density of THimmunoreactive neurons within the ENS (Chalazonitis et al., 2022;Li et al., 2004;Qu et al., 2008), which is pertinent given that specific locations of neuronal populations may influence their functions.It is also noted in disease, such as PD, that submucosal neurons appear to be selectively vulnerable (Francesca et al., 2023;Lebouvier et al., 2010;Stokholm et al., 2016).

| In situ hybridization
Following thawing, tissue sections were heated for 30 min at 60 C, post-fixed in prechilled 4% PFA for 15 min at 4 C, and dehydrated by subsequent immersion in 50%, 70%, and 100% ethanol for 5 min at RT.In situ staining of fixed-frozen samples was performed according to the RNAscope Multiplex Fluorescence manufacturer's instructions (Advanced Cell Diagnostics, ACD).Briefly, tissue sections were incubated for 10 min at 95 C for antigen retrieval, and Protease III was applied for 30 min at 40 C.After multiple washes, RNAscope multiplex assay was carried out as described in ACD's protocol followed by the aforementioned immunofluorescence protocol to co-stain with tdTomato and ChAT antibodies.All incubations were at 40 C and used a humidity-controlled chamber.Mouse probes used were Grp-C1 (317861), Calcb-C1 (425511), and Sst-C1 (404631), and Opal dye 520 was used at 1:1500 concentration.Hoechst 33258 was used for nuclear staining (1:5000, ThermoFisher H3569).Slides were mounted in Prolong Gold Antifade Mounting media (ThermoFisher P36930).

| Single-cell suspension preparation
Given protocols are established for longitudinal muscle and myenteric plexus (LMMP) isolations, we prepared single-cell suspensions from this intestinal layer.Isolated colons from adult DAT-Cre-tdTomato mice (n = 4, equal sexes) were washed, cut open longitudinally as described above, and pinned with the intestinal lumen facing down on a 4% agarose pad.The LMMP was then microdissected by peeling it off using fine-tip tweezers and was then collected and cut into 1-to 2-cm pieces.The tissues were then enzymatically dissociated to release myenteric neurons and other cell types using 1 mg/mL collagenase IV (Sigma, C5138) and 0.05% trypsin (Sigma, T4049), adapted from Smith et al. (2013).Dissociated cells were filtered through a 40-μm cell strainer and stained with eFluor506 fixable viability dye (ThermoFisher 65-0866-14).Using fluorescence-activated cell sorting (FACS Aria), we detected $60% of events were live single cells and of that $20% were DAT-Cre-tdTomato-positive.

| Single-cell library preparation and analysis
Twenty thousand live cells were FACS collected and then prepared for single-cell sequencing.All cells were processed according to 10Â Genomics Chromium Single Cell 3 0 Reagent Guidelines (https://support.10xgenomics.com/single-cell-gene-expression).Briefly, cells were partitioned into nanoliter-scale Gel Bead-In-EMulsions (GEMs) using 10Â GemCode Technology.Primers containing (i) an Illumina R1 sequence, (ii) a 16 bp 10Â barcode, (iii) a 10 bp Unique Molecular Identifier (UMI), and (iv) a poly-dT primer sequence were incubated with partitioned cells resulting in barcoded, full-length cDNA from poly-adenylated mRNA.Silane magnetic beads were used to remove leftover biochemical reagents/ primers, and then cDNA was amplified by PCR.Enzymatic fragmentation and size selection was used to F I G U R E 1 Detection of enteric neurons in DAT-tdTomato transgenic mice.(a) Schematic diagram of immunofluorescence staining protocol for mouse colon isolated from transgenic mice, where tdTomato (Ai9) is controlled under the promoter for dopamine transporter (Slc6a3, referred to as DAT-Cre).The coronal plane of the colon mounted in a swiss-roll depicts the three subdivision of intestinal layers and the two ganglionated plexuses.The submucosal plexus (SMP) is shown between the mucosa (M) and the muscularis propria (mp), while the myenteric plexus (MP) is within the mp.(b) Representative immunofluorescence staining with cross-sectional view of colonic tissue from DAT-Cre mice.Enteric neurons within the two ganglionated plexuses are detectable with several pan-neuronal markers that are commonly used to identify neurons.Neurons (green) are immunoreactive for RNA-binding human antigen (HuD), tubulin β3 (Tubb3) and ubiquitin carboxyl-terminal hydrolase-1 (Uch-l1/Pgp9.5).Ganglia are outlined by white dotted box.The inset is a magnified view of the ganglion designated with * and an individual cell is delineated by white dotted lines.The density of neurons normalized to ganglionic area (in mm 2 ) is quantified in both plexuses.Mean ± SEM, n = 4 mice per group with three ganglia quantified per mouse.(c) Crosssectional view of colonic tissue from DAT-Cre mice immunoreactive for tdTomato (red).Agrin is used as a marker for blood-myenteric-barrier (white) to depict ganglia.Note the lack of tdTomato signal in wild type C57Bl/6J colonic tissue, as expected.Nuclei are presented in blue.Scale bars: 50 μm.

| Tissue quantification and statistical analysis
Immunostained tissue slices were imaged using a Leica SP8 Confocal microscope and quantified with the Las X Leica software.Images were taken at 40Â magnification with Z-stacks (step size of 0.9).Each ganglion was identified in either the submucosal or myenteric plexus based on morphology as shown in earlier publications (Kulkarni et al., 2018;Li et al., 2004).The quantification was unbiasedly performed using Z-stacks where the assessor defined a nucleus within the ganglia and then scrolled through the Z-stack to ensure that the stains were consistently expressed in and around the cell body surrounding the nuclei.Tubb3 and tdTomato (as well as most of the other stains used) have clear cytoplasmic signal abutting the nuclei, which made this approach ideal for ensuring accurate quantification.All quantitative analysis at the ganglia was normalized to an area of 900 μm 2 (calculated as the average size of a ganglion).
The number of animals used with three to four ganglia analysed per colon at each plexus can be found in the figure legends.All statistical analysis was performed using GraphPad Prism 5 (RRID:SCR_002798, http://www.graphpad.com/).For comparisons across multiple groups, a two-way analysis of variance (ANOVA) followed by Tukey's post-hoc comparison was used.For comparisons of across two groups, an unpaired Student's t test was used.P values <0.05 were considered significant.
For quantification of single-cell RNAseq data, unsupervised clustering of cells was performed.A selection of highly variable genes (2000 genes) was obtained and used for principal component (PC) analysis.Significant principal components were determined using JackStraw analysis.PCs 1 to 20 were used for graph-based clustering (at resolution = 1) to identify distinct groups of cells.These groups were projected onto UMAP analysis using previously computed PCs 1 to 20.Expression of selected marker genes (Elavl4, Tubb3, Uchl1 and Map 2) was used to classify neurons, and these initial clusters were further divided into seven cell clusters identified by graph-based clustering.

| RESULTS
3.1 | Enteric pan-neuronal markers and tdTomato expression are detectable in both colonic plexuses in DAT-Cre-tdTomato transgenic mice DAT-Cre transgenic lines are considered more reliable in the selective identification of midbrain DA neurons compared with other reporter mouse lines such as TH-Cre (Bäckman et al., 2006;Ekstrand et al., 2007;Engblom et al., 2008;Lindeberg et al., 2004;Zhuang et al., 2005).As such, we used DAT-Cre mice crossed with Ai9 Credependent tdTomato mice and immunohistochemistry (Figure 1a) to confirm the existence of submucosal and myenteric DA neurons, as well as characterize their heterogeneity.Using cross-sectional analysis in swiss-roll format, we detected three subdivisions of the intestinal layers and two ganglionated plexuses, as expected.We consistently detected the submucosal plexus (SMP) between the mucosa and the muscularis propria, as well as the myenteric plexus (MP) within the muscularis propria, using a battery of pan-neuronal markers (Figure 1b) (RNA-binding human antigen, HuD; tubulin β3, Tubb3; and ubiquitin carboxyl-terminal hydrolase-1, Uch-l1/ Pgp9.5).We explored the efficacy of these commonly used pan-neuronal markers given, in most cases, they are utilized interchangeably, with little evidence to support which markers are most robust.Tubb3 is a cytoskeletal protein involved in microtubule stability and is often used to label both neuronal soma and axonal projections in the CNS or PNS, whereas HuD and Pgp9.5 are expressed predominantly in peripheral neurons (Bolognani et al., 2010;Hibi et al., 1999;Liu et al., 2007).A direct comparison of the expression of these neuronal markers revealed that there was no statistical difference within or between plexuses across all markers (Figure 1b).We opted to focus largely on Tubb3 as a colabel with tdTomato for consistency and due to compatibility considerations with other antibodies.Of note, we identified nonneuronal tdTomato-positive signal in the vicinity of the plexuses (data not shown), which are likely immune cells (Gopinath et al., 2022;Mackie et al., 2018), and thus we were rigorous in performing all analysis with pan-neuronal marker co-labelling.We also determined that tdTomato expression was robust in colonic tissue in both the SMP and MP of DAT-Cre-tdTomato mice but that the signal was not detected in wild type C57Bl/6J colonic tissue, as expected (Figure 1c).We used a bloodmyenteric-barrier marker to define MP borders, termed Agrin (which is not expressed in the SMP) (Dora et al., 2021), important for defining ganglia localization in the MP (Figure 1c).

| TdTomato-immunoreactive DA neurons vary across plexuses
In keeping with previous descriptions of DA neurons in the gut demonstrating that the density of enteric THexpressing neurons gradually decreases from rostral to caudal regions of the GI tract and from the SMP to MP (Chalazonitis et al., 2022;Li et al., 2004;Qu et al., 2008), we found a reduced density of tdTomato/Tubb3-positive DA neurons in the MP compared with the SMP using DAT-Cre-tdTomato mice (Figure 2a).To ascertain the DAergic phenotype of tdTomato-positive neurons, we also co-labelled these cells with TH, the rate-limiting DA synthesis enzyme.We indeed observed that most of these neurons were also immunoreactive for TH in ganglionated plexuses.Interestingly, a small proportion of neurons were positive for TH but not for tdTomato, consistent to what has been described in the ventral midbrain (Lammel et al., 2015;Poulin et al., 2018;Tiklov a et al., 2019).By contrast, we noted few submucosal and myenteric ganglia containing tdTomato-immunoreactive neurons but negative for TH (Figure 2b).

| A subpopulation of tdTomatoimmunoreactive neurons in the ENS exhibit dual-transmitter content
The complex neural networks in the ENS can be largely categorized as either excitatory or inhibitory motor neurons (Brehmer, 2021).The former mostly express choline acetyltransferase (ChAT), generally classified as cholinergic, whereas the latter is commonly referred to as nitrergic due to their release of nitric oxide, an inhibitory molecule synthesized by neuronal-specific nitric oxide synthase (nNOS) (Dharshika & Gulbransen, 2023).The striking abundance of tdTomato/TH-positive neurons in the colon wall in the present study, especially in the SMP (Figure 2a), led us to speculate that a subset of enteric cholinergic or nitrergic neurons may also have a DAergic phenotype.This conclusion would coincide with a growing literature demonstrating that DA neurons as well as other CNS and PNS neurons use a combination of neurotransmitters (Brunet Avalos & Sprecher, 2021;Dal Bo et al., 2004;El Mestikawy et al., 2011;Trudeau & El Mestikawy, 2018).To examine this, we performed coimmunolabelling for HuD and ChAT, Tubb3 and nNOS, tdTomato and ChAT, and tdTomato and nNOS.First, we confirmed the presence of both cholinergic and nitrergic neurons in both plexuses at similar levels (Figure 3a).Then we assessed ChAT-and nNOS-positive neurons for DAT-tdTomato (Figure 3b).Indeed, we observed the existence of a double-phenotype DA neuron population in both the SMP and MP, which accounts to $20% and $50% of tdTomato-immunoreactive neurons, respectively.Of note, we also found that the majority of tdTomato-immunoreactive neurons do not express both ChAT and nNOS in the SMP but nearly half in the MP do.

| Single-cell transcriptomics corroborates the existence of ChATexpressing dopaminergic neuron subtype
To further elucidate the molecular signature defining dopaminergic subtypes with dual-transmitter content, we used single-cell transcriptomics.Using four mice of equal sexes, we microdissected the LMMP from the colon followed by enzymatic digestion and FACS to obtain live single cells and generate a gene expression library (Figure 4a).Following the application of computational quality control measures, we performed unsupervised clustering and annotated seven enteric clusters based on literature-curated cell type markers (Figure 4b).Out of these cells, 247 are classified as neurons based on the expression of Elavl4 (HuD), Tubb3, Uchl1 (Pgp9.5)and Map 2, which we further subdivided into seven distinct neuronal clusters (Figure 4c; data are also available at http://singlocell.openscience.mcgill.ca/EntericNeurons).Consistent with our immunostaining, we identified Chat-(Cluster 0, 2, 3 and 6) and Nos1 (Cluster 1)-expressing neuronal cell clusters as the most prevalent cell types (Figure 4c).We then evaluated Th expression and found highest expression in Clusters 2 and 5 (Figure 5a).Notably, Cluster 5 showed higher expression of other canonical DA neuron markers, important in the vesicular packaging of DA (i.e., vesicular monoamine transporter 2, Vmat2), in the regulation of SNc neurons excitability (i.e., G-protein-regulated inward-rectifier potassium 2, Girk2), as well as previously reported genes expressed in midbrain DA neurons, such as vesicular glutamate transporter 2, Vglut2, cocaine-and amphetamineregulated transcript protein, Cartpt and 5-hydroxytryptamine/serotonin receptor 2B, Htr2b ( Ásgrímsd ottir & Arenas, 2020; Carpenter et al., 2020; F I G U R E 2 TdTomato-immunoreactivity varies across plexuses and colocalizes with tubulin β 3 (Tubb3) and tyrosine hydroxylase (Th)-positive cell.(a) Representative coronal slice of colonic tissue illustrating that DAT-tdTomato-expressing cells (red) are immunoreactive for a pan-neuronal marker, Tubb3 (green) in the submucosal and myenteric plexuses (SMP, MP).The density of Tubb3 +ve tdTom +ve neurons normalized to ganglionic area (in mm 2 ) was quantified in both plexuses, which revealed significantly less DAT-tdTomato +ve neurons in the MP compared with the SMP.Unpaired, two-tailed t test, P < 0.05.Mean ± SEM, n = 6 mice per group with three ganglia per mouse quantified.(b) Representative immunofluorescence staining for tyrosine hydroxylase (Th) (cyan), which is a rate limiting enzyme for generating L-3,4-dihydroxyphenylalanine (L-DOPA), the precursor for dopamine.Th +ve cells co-localized with tdTomato (red) and Tubb3 (green) in both ganglionated plexuses.The density of Tubb3 +ve enteric neurons expressing either (1) DAT-tdTomato (tdTom +ve Tubb3 +ve ), (2) DAT-tdTomato and Th (tdTom +ve Th + ve Tubb3 +ve ), (3) Thalone (tdTom Àve Th +ve Tubb3 +ve ), or (4) no dopamine neuron markers (tdTom Àve Th Àve Tubb3 +ve ) is presented (normalized to ganglionic area (in mm 2 )).This revealed that most Th-expressing neurons in the colon co-localize with DAT-tdTomato in both plexuses, but a smaller subset does not express DAT-tdTomato.Two-way ANOVA followed by Tukey's multiple comparison's test, P < 0.05.Mean ± SEM, n = 4 mice per group with three ganglia per mouse quantified.Ganglia are outlined by the white dotted box.The inset is a magnified view of the ganglion designated with * and an individual cell is delineated by white dotted lines.Nuclei are presented in blue.Scale bars: 50 μm.Doly et al., 2017;El Mestikawy et al., 2011;Trudeau, 2004;Trudeau & El Mestikawy, 2018;Trudeau & Gutiérrez, 2007) (Figure 5b).Validating our transcriptome observations, we detected immunoreactivity for VMAT2 and GIRK2 in DAT-Cre-tdTomato-expressing neurons within the ganglionated plexuses (Figure 5c).As for Cluster 2, these Th-expressing neurons had very little to no expression of the aforementioned traditional DA neuron markers (Figure 5b); however high Chat expression was noted in this cluster compared with Thexpressing Cluster 5 (Figure 4c).We thus further explored the transcriptomic signature of this unique Th/ Chat-expressing, DAergic neuronal subtype (Figure 6a), and amongst them, we detected a range of genes previously reported in ENS (Grp, Calcb, and Sst) but not yet shown to be present in enteric DA neurons (Figure 6b).
Interestingly, these genes have also been associated with DA signalling in the brain (Charbit et al., 2009;Elina et al., 2023;Kramer et al., 2018;Salesse et al., 2020).Validation experiment using in situ hybridization indicated the expression of these novel markers in a small fraction of cells immunoreactive to both tdTomato and ChAT in the colonic myenteric plexus (Figure 6c).

| DISCUSSION
Enteric neurons are numerous and exhibit diverse neuronal phenotypes analogous to the CNS, yet they are largely understudied.DA plays pivotal roles in the brain and dysfunction of DA signalling underpins a plethora of neuropsychiatric disorders and neurodegenerative diseases, including PD (Björklund & Dunnett, 2007;Hamamah et al., 2022).However, the functions of DA neurons in the gut are much less understood with some studies implicating them in gastrointestinal motility (Chalazonitis et al., 2020;Li et al., 2011Li et al., , 2006;;Walker et al., 2000).Through advances in DA cell classification in the CNS, striking functions have been found to underscore the heterogeneity of these cell populations in healthy and diseased brains.A better understanding of enteric DA neurons and their vulnerability in disease is critical in the face of mounting evidence suggesting roles of the gut in a myriad of brain disorders (Dinan & Cryan, 2017).
The heterogeneity of DAergic neuronal populations across the plexuses in the gut may account for disparate roles of DA within the ENS.In particular, our discovery of DA neurons with dual-transmitter properties in the gut suggests diversity within the enteric plexus that may contribute to functional differences.Previous work has indeed suggested that enteric DA neurons have multiple roles.For example, deletion of Dat in mice suggested the involvement of DAergic neurons in modulating the propagation of colonic migrating motor complexes (CMMC) to promote peristalsis-a central function of the MP (Walker et al., 2000).Meanwhile, intraperitoneal administration of neurotoxins targeting the DA system resulted not only in the ablation of enteric DA neurons but also in the reduction of colonic mucus content due to decreased signalling through the DA receptor D5 expressed in goblet cells (Li et al., 2019).Contrary to myenteric DA neurons, the implication of DA in mucosal secretion may be attributed to a subset of cells primarily in submucosal ganglia.Submucosal DAergic neurons were believed to also participate in gastrointestinal motility whereby mice with DA neuronal hypoplasia in the SMP manifested longer gastrointestinal transit time and slower colonic expulsion time (Chalazonitis et al., 2020).These submucosal DAergic neuronal subtypes may differ from those F I G U R E 4 Single-cell RNA sequencing of colonic LMMP cells from healthy adult mice.(a) Schematic diagram of the colonic longitudinal muscle and myenteric plexus (LMMP) isolation and dissociation for single-cell RNA sequencing (scRNAseq).Live cells were sorted via fluorescence-activated cell sorting (FACS) then gene expression libraries were prepared using 10Â chromium genomics.Quality control, normalization, clustering, and expression analyses of data were done in R using Seurat with default settings.Distinct myenteric cells were identified and displayed as a UMAP plot.(b) Literature-curated genes shown in the violin plot were used to annotate each clusters in the LMMP single-cell dataset.(c) UMAP plot depicting seven neuronal cell subpopulations identified by confirming the expression of 4 panneuronal marker genes in these populations (Elavl4/HuD, Tubb3, Uchl1/Pgp9.5, Map 2).The two main enteric subtypes were also identified: Chat/ChAT-expressing cholinergic (Clusters 0, 2, 3, 6) and Nos1/nNOS-expressing nitrergic (Cluster 1) neurons.
responsible for mucosal secretions.Optogenetic stimulation studies in transgenic mice revealed that CMMC generation is mediated in part by the activation of cholinergic neurons while nitrergic neurons are active during its tonic inhibition (Gould et al., 2019).As such, DA release from DAT-Cre-tdTomato-positive neurons coexpressing either ChAT or nNOS could be involved in governing this complex process.This is in accordance with previous speculation that repeated patterns of interconnected neural networks in the ENS are activated during peristalsis (Chalazonitis et al., 2022).To add complexity, studies have shown that DA receptor blockade conveys a suppressive action on colonic motility (Li et al., 2006(Li et al., , 2011)), in contrast to previous work showing inhibition of DA reuptake elicits fewer smooth muscle contractions (Walker et al., 2000).Clearly, there is a need for a new classification of enteric DA neuron subtypes not only to define which cells control intestinal motility but also to unveil other intricate DA-related mechanisms.
Given advances in single-cell transcriptomics for profiling DA neurons in the CNS, we applied this approach in the ENS to unravel novel cellular markers that can demarcate specific DA neuronal subtypes in the GI tract.We found that traditional markers previously shown at high levels in midbrain DA neurons are also expressed in enteric neuron clusters.Interestingly, we also identified a distinct Th-expressing neuronal cluster that may share similar properties with cholinergic neurons as evidenced by its high Chat expression.This subtype also expressed genes found in disparate classes of CNS DA neurons, such as Grp, Sst, and Calcb (Kramer et al., 2018;La Manno et al., 2016;Poulin et al., 2014;Viereckel et al., 2016).In contrast to previous reports of widespread expression of GRP-immunoreactive DA neurons in the midbrain (Kramer et al., 2018), our study suggests rare expression, which underscores the heterogeneity of DA subtypes across the body.As for Sst and Calcb, DA neurons in the ventral tegmental area (VTA) that co-release somatostatin (SST) are implicated in the response to stress (Elina et al., 2023), while calcitonin gene-related peptide (CALCB), expressed in hypothalamic DA neurons, regulates pain transmission (Charbit et al., 2009).The manifestation of a dual-transmitter phenotype in this neuronal population likely contributes to the multifaceted features of the DAergic system in the colon, but further investigation will be critical to fully characterise the extent of heterogeneity along the entire GI tract.
It is noteworthy that despite the wealth of information single-cell transcriptomics can provide, classification of neurons based solely on molecular signatures is likely not sufficient, and thus their morphology, physiological properties and functional connectivity must also be assessed (Zeng & Sanes, 2017).This is especially the case with respect to our observations.We identify a small subset of DAT-Cre-tdTomato-positive neurons that do not express TH, and thus the DAergic-related functional role, if any, of this population is unclear.Between embryonic days 10 to 13, it is known that a population of neuronal precursor cells, termed transiently catecholaminergic (TC) cells, display high levels of TH, which progressively diminishes during terminal differentiation until TH is no longer expressed (Baetge & Gershon, 1989;Gershon et al., 1984).Li and colleagues have also identified a population of neurons that are TH negative, but they maintain DAT expression.Consistent with our observations, this was observed in the GI tract (in the mouse ileum and colon)l (Li et al., 2004).It is tempting to speculate that these divergent neuronal subtypes have corresponding functional differences whereby TH neurons may represent a more mature counterpart able to synthesize and release DA, while DAT neurons that are TH-negative (and presumably VMAT2 negative) represent cells that are non-traditional DAergic in the sense that they are not (yet) able to synthesize and release DA.
It remains possible that the DAergic phenotype of tdTomato neurons not expressing TH could be "awakened" later in life in response to signals associated with cellular stress.Indeed, in previous work, denervation of axonal projections to the gut culminated in elevated proportions of TH-immunoreactive neurons without affecting the overall number of enteric neurons (Li et al., 2004), implying that existing pools of neurons may upregulate TH expression as a compensatory mechanism in response to injury.Hence, we postulate that a subset of molecularly defined neurons may be acting as reserves with submaximal DAergic function.This further begs the question of whether the localization of these subtypes (being mostly in SMP) relates to the potential vulnerability of submucosal neurons (given the proximity to the intestinal lumen) and thus presumably necessitates the need to have a pool of reserve DA neurons in this region.In agreement with a study depicting the bloodganglia barrier only at the myenteric plexus (Dora et al., 2021), we also noted the absence of immunoreactivity for agrin in the SMP, thereby suggesting a lack of protection from the blood-ganglia barrier in the SMP which could render submucosal neurons more susceptible to immune cell and bacteria-related damage.Collectively, these hypotheses can be addressed by marrying detailed knowledge on gene expression patterns in specific DA subtypes with disease or injury models.
An altered gut-brain axis is increasingly recognized as central to many neurological conditions including Parkinson's disease (PD).PD is the second-most prevalent neurodegenerative disease, affecting more than 10 million people globally (Ou et al., 2021).Its pathological characteristics are commonly defined by the accumulation of Lewy body inclusions and by the gradual degeneration of dopamine (DA) neurons in the substantia nigra pars compacta (SNc) of the midbrain; and the selective loss of DA neurons is thought to underlie some of the canonical motor symptoms of PD (Greffard et al., 2006).There is growing recognition that non-motor abnormalities, including perturbed sleep, anosmia, depression/anxiety and gut dysmotility, precede motor deficits by up to 20 years (Gaenslen et al., 2011;Pont-Sunyer et al., 2015;Schrag et al., 2015).This is likely to result from the dysfunction or loss of multiple types of neurons, albeit quantitative evidence for the extent of neuronal loss in regions other than the SNc is presently fragmentary (Giguère et al., 2018).Although it is likely that most PD patients suffer from some level of dysfunction of the peripheral nervous system (PNS) (Abbott et al., 2001;Borghammer, 2023;Fedorova et al., 2017;Raeder et al., 2023;Savica et al., 2009;Skjaerbaek et al., 2021), whether gut dysmotility and associated constipation are driven in part by dysfunction or loss of enteric DA neurons remains unclear (Anderson et al., 2007;Francesca et al., 2023;Liu et al., 2021;Wang et al., 2012).Various histological studies have demonstrated the presence of Lewy body inclusions in the gastrointestinal (GI) tract (Del Tredici & Braak, 2012;Dickson et al., 2009), including in neurons, that positively correlate with constipation in PD patients (Lebouvier et al., 2010).Pathological features of neurons in the colon are even observed at least 8 years prior to PD diagnosis (Hilton et al., 2014).Further reports also revealneuronal loss in PD colons as well as in the colons of rodent models of PD (Lebouvier et al., 2010;McQuade et al., 2021;Ohlsson & Englund, 2019).Despite all these reports, and clear dysfunction of the gastrointestinal tract in PD, there are inconsistencies in the extent of neuronal loss and pathology reported (O'Day et al., 2022).This may be due to the heterogeneity in clinical presentation and data collection; as well as, a lack of balanced criteria for patient recruitment, a lack of consistency in neuronal quantification methods and issues with study design.Furthermore, adding dedicated analysis for studying neuronal subtypes, such as DA neurons, affected in the GI tract of PD patients will also be critical for determining the heterogeneity of neuronal involvement.
A perplexing idea recently dominating the field of PD is the question of the origin of selective vulnerability of DA neuron subtypes and how specific gene signatures may help us infer the underlying mechanisms (Giguère et al., 2018;Poulin et al., 2020).Although enteric neurodegeneration has been examined in PD, there still exists discrepancies as to the extent and specificity of neuronal damage involved (Lebouvier et al., 2010;McQuade et al., 2021;O'Day et al., 2022;Ohlsson & Englund, 2019).In this regard, the ion channel GIRK2 implicated in modulating neuronal excitability has been hypothesized as a relevant factor that may influence sensitivity of specific DA neurons to degenerate.Strong GIRK2 expression is evident in most SNc DAergic cell types but fewer (with lower levels) in less affected areas, such as the VTA, as described in post-mortem tissues and recent single-cell RNAseq dataset from a murine model of PD (Behzad Yaghmaeian et al., 2023;Reyes et al., 2012).We observed that Th-expressing Clusters 2 and 5 express Girk2, so it will be interesting to assess whether selective vulnerability may be present in these gut DA subtypes.
Taken together, our study provides a steppingstone to further garner interest in exploring the diverse molecular and cellular properties of enteric DA neurons.These data suggest that the DAergic neuronal populations in the gut are more heterogeneous than previously thought, supporting the notion that DA neurons play diverse functions in the gut.In particular, we identified a novel ChAT-positive dopaminergic neuronal subtype in the gut that co-expresses midbrain DA-related genes (Grp, Calcb, and Sst), exhibiting a dual-transmitter neuronal phenotype.

F
I G U R E 2 Legend on next page.

F
I G U R E 3 A subpopulation of tdTomato-immunoreactive neurons in the ENS exhibit dual-transmitter content.(a) Representative cross-sectional view of colonic tissue labelling cholinergic (choline acetyl transferase, ChAT-positive) or nitrergic (neuronal nitric oxide synthase, nNOS-positive) (red) neurons (green, HuD or Tubb3) in submucosal and myenteric plexuses (SMP and MP).The density of ChAT +ve or nNOS +ve neuronal populations (colocalized with pan-neuronal markers HuD or Tubb3, respectively) are normalized to ganglionic area (in mm 2 ).Mean ± SEM, n = 4 mice per group with three ganglia per mouse quantified.(b) Representative immunofluorescence staining for DAT-tdTomato-expressing cells (red) co-localizing with either ChAT or nNOS (green) in both ganglionated plexuses (SMP and MP).The densities of DAT-tdTomato neurons co-expressing either ChAT (tdTom +ve ChAT +ve ) or nNOS (tdTom +ve nNOS +ve ) are compared with that of DAT-tdTomato alone (tdTom +ve ChAT Àve nNOS Àve ) in both plexuses normalized to ganglionic area (in mm 2 ).The proportions of each subpopulations are shown in the inset as % relative to the total tdTomato +ve cells.Twoway ANOVA followed by Tukey's multiple comparison's test, P < 0.05.Mean ± SEM, n = 3 mice per group with three ganglia per mouse quantified.Ganglia are outlined by white dotted box.The inset is a magnified view of the ganglion designated with * and an individual cell is delineated by white dotted lines.Nuclei are presented in blue.Scale bars: 50 μm.

F
I G U R E 5 Two dopaminergic neuronal subclusters expressed traditional dopaminergic neuronal markers at varying degrees.(a) The expression of a canonical DAergic neuronal marker, tyrosine hydroxylase (Th), in the neuronal populations are highest in Cluster 2 and 5. (b) Other previously identified markers of DAergic neurons in the CNS are lower expressed, if any, in Cluster 2 than in 5, suggestive of a potentially distinct Th-expressing neuronal populations.(c) Representative images of colonic tissue demonstrating co-labelling of tdTomato (red) with known dopaminergic neuronal markers (cyan; Vmat2 and Girk2), indicated by arrowheads.Tubb3 is used as pan-neuronal marker (green) and nuclei are in blue.Scale bars: 50 μm.

F
I G U R E 6 Identification of novel enteric dopaminergic neuron subtype with cholinergic phenotype.(a) Heatmap indicating upregulated genes (row) in each neuronal cell populations (column).Highlighted in green is DAergic neuronal subtype co-expressing Th and Chat.(b) UMAP plots showing novel ENS DA neuron marker genes highly expressed in Cluster 2. (c) Representative images of colonic tissue demonstrating expression of putative novel DA markers (green; Grp, Calcb, and Sst) in the myenteric plexus co-labelled with tdTomato (red) and ChAT (green).Nuclei are presented in blue.Scale bars: 50 μm.
Primary antibodies used in immunofluorescence staining of colonic tissue.
T A B L E 1