Profiling of G protein-coupled receptors in vagal afferents reveals novel gut-to-brain sensing mechanisms.

OBJECTIVES: G protein-coupled receptors (GPCRs) act as transmembrane molecular sensors of neurotransmitters, hormones, nutrients, and metabolites. Because unmyelinated vagal afferents richly innervate the gastrointestinal mucosa, gut-derived molecules may directly modulate the activity of vagal afferents through GPCRs. However, the types of GPCRs expressed in vagal afferents are largely unknown. Here, we determined the expression profile of all GPCRs expressed in vagal afferents of the mouse, with a special emphasis on those innervating the gastrointestinal tract.
METHODS: Using a combination of high-throughput quantitative PCR, RNA sequencing, and in situ hybridization, we systematically quantified GPCRs expressed in vagal unmyelinated Nav1.8-expressing afferents.
RESULTS: GPCRs for gut hormones that were the most enriched in Nav1.8-expressing vagal unmyelinated afferents included NTSR1, NPY2R, CCK1R, and to a lesser extent, GLP1R, but not GHSR and GIPR. Interestingly, both GLP1R and NPY2R were coexpressed with CCK1R. In contrast, NTSR1 was coexpressed with GPR65, a marker preferentially enriched in intestinal mucosal afferents. Only few microbiome-derived metabolite sensors such as GPR35 and, to a lesser extent, GPR119 and CaSR were identified in the Nav1.8-expressing vagal afferents. GPCRs involved in lipid sensing and inflammation (e.g. CB1R, CYSLTR2, PTGER4), and neurotransmitters signaling (CHRM4, DRD2, CRHR2) were also highly enriched in Nav1.8-expressing neurons. Finally, we identified 21 orphan GPCRs with unknown functions in vagal afferents.
CONCLUSION: Overall, this study provides a comprehensive description of GPCR-dependent sensing mechanisms in vagal afferents, including novel coexpression patterns, and conceivably coaction of key receptors for gut-derived molecules involved in gut-brain communication.

Introduction

The gastrointestinal (GI) mucosa has long been known to be richly innervated by unmyelinated axons [1], [2]. These axons typically wander within the lamina propria, often approaching its epithelium without penetrating the basal lamina. While many axons present in the GI mucosa are of enteric origin [3], anterograde tracing studies revealed that vagal afferents also abundantly innervate the GI mucosa [4], [5], [6], [7], [8]. Some vagal unmyelinated afferents supplying the GI tract are polymodal [9], and electrophysiological studies have demonstrated that mucosal vagal afferents respond to a wide range of molecules absorbed across the epithelium or locally released, in addition to the distension of the alimentary canal, as well as noxious, thermic and osmotic stimuli [10], [11], [12], [13], [14], [15], [16], [17]. Thus, it is not surprising that a variety of sensations and autonomic reflexes can be elicited by the mechanical or chemical stimulation of the GI mucosa [18], [19], [20], [21].

Many of the molecules that modulate the activity of vagal afferents, including neurotransmitters, gut peptides, lipids, and metabolites, are G protein-coupled receptors (GPCRs) ligands [22]. However, due to a lack of understanding of GPCRs distribution in vagal afferents, it is unknown whether the aforementioned molecules directly act on these neurons. For instance, a prevalent view suggests that vagal GI mucosal afferents directly respond to peptides released from enteroendocrine cells including, cholecystokinin (CCK) and glucagon-like peptide 1 (GLP-1) [23], [24], [25], [26], [27]. In support of this view, the receptors for CCK and GLP-1 are abundantly expressed in vagal afferents [7], [28], [29], [30], and CCK and GLP-1 acutely enhance the activity of vagal afferents [10], [31]. Moreover, the anorectic effects of peripheral CCK and GLP-1 are attenuated in deafferented animals [32], [33], [34], [35]. Despite these findings, several other studies have suggested that gut-derived CCK and GLP-1 act through potentiating vagal responses to distending loads and/or directly on the brain [36], [37], [38], [39], [40]. A recent study further challenged the common view that CCK and GLP-1 can be directly detected by mucosal afferents by demonstrating that neurons expressing receptors for these gut hormones do not innervate the mucosa [7]. Lastly, considering that at least 10 different peptides are known to be secreted by different types of enteroendocrine cells [41], it is unknown which gut peptides, other than CCK and GLP-1, can be detected by vagal afferents.

Here, we determine the expression profile of all GPCRs in the vagal afferents of the mouse. A special emphasis was given to vagal afferents expressing Nav1.8, a voltage-gated sodium channel involved in the generation of action potentials in vagal unmyelinated afferents [16], [42], [43]. Nav1.8-expressing vagal afferents richly innervate the entire GI mucosa [5], making these neurons ideally positioned to detect nutrients and molecules locally released in the GI tract. To better understand how the vast assortment of GPCRs contribute to vagal sensing mechanisms, we fully characterized GPCRs expression in Nav1.8-expressing vagal afferents using a combination of quantitative PCR (qPCR), RNA sequencing, and double in situ hybridization studies.

Materials and methods

Mice

Male C57BL/6JRj and C57BL/6J wild-type mice were purchased from Janvier Breeding Centre (France) and the ARC at UT Southwestern Medical Center (Dallas), respectively. These mice were used for qPCR assays, chromogenic Brown, Red, and Duplex in situ hybridization.

Nav1.8-Cre-DTA mice were generated and genotyped as previously described by us [44]. Briefly, we crossed Nav1.8 knock-in Cre-recombinase mice (gift from Dr. John Wood, University College London) on a pure C57Bl/6J background with ROSA26-eGFP-DTA mice (stock #006331; The Jackson Laboratory, USA) to generate litters with a 1:1 ratio of littermate controls (Nav1.8-Cre) and ablated mice (Nav1.8-Cre-DTA). Young male ablated and control mice were used for RNA-Seq studies.

Nav1.8-Cre-ChR2-YFP male mice were also used for the purpose of ISH combined with immunohistochemistry. Mice carrying one Nav1.8-Cre and one ChR2-YFP alleles were generated and genotyped exactly as previously described by us [45]. In these mice, YFP clearly delineates the membrane of Nav1.8 neurons, hence facilitating double labeling analysis.

All mice used were young males (6–10 weeks old) housed in a barrier facility with temperature controlled environment (∼23 °C). Mice were fed ad libitum with standard chow. The Danish Animal Experiments Inspectorate or Institutional Animal Care and Use Committee at the University of Texas Southwestern Medical Center at Dallas approved all animal procedures.

qPCR array of nodose ganglia

Wild-type mice (C57/Bl6JRj, Janvier) were euthanized using cervical dislocation and the nodose ganglia from 2 mice were dissected into RNAlater at room temperature and kept at 4°C overnight prior of RNA extraction using the NucleoSpin® RNA XS (MACHEREY-NAGEL) and RT-PCR was performed using SuperScript III Reverse Transcriptase (Invitrogen). Custom-designed RT2 Profiler PCR Arrays (Qiagen) was used to analyze 377 GPCRs and 3 Receptor-Activity Modifying Proteins (RAMPS) on a LightCycler480 (Roche). Relative expression was calculated using the formula:where CqTarget = the quantification cycle (Cq) for the target gene; CqRef = the geometrical mean of the quantification cycle for the reference genes (Ywhas, Actb and Gapdh); C is an arbitrary constant dependent on CqRef used to shift the relative expression so Cq values of 35 on average are equal to 1, here C = 24221. Undetectable targets were assigned a Cq value of 40. In total 8 mice was used to generate the 4 samples analyzed. Information of genes on the qPCR array is given in Table S1.

RNA-sequencing of nodose ganglia

RNA sequencing was performed on the nodose ganglia of ablated and control mice (n = 3/group). The two nodose ganglia (left and right) from one mouse were pooled and RNA was extracted as described above for qPCR. Library construction (Truseq, Illumina), sequencing (Hiseq 100 PE, Illumina) and read quality control filtering was performed at BGI Tech solutions (Hong Kong). The sequencing data was processed using the “Tuxedo tools” (Tophat2, Cufflinks2, cummeRbund) as outlined in [46] using the mm10 reference annotation only. Briefly, normalized expression quantification and differential expression tests for Ensembl genes were obtained using Cuffdiff2 with “--frag-bias-correct” [47] and “--multi-read-correct” enabled. The output from Cuffdiff2 [46] was analyzed with the R package cummeRbund (version 2.12.1).

In situ hybridization

RNAScope chromogenic “brown” assay

Wild-type mice received an overdose of chloral hydrate (500 mg/kg, i.p.) and their nodose ganglia were rapidly dissected and frozen on dry ice. Ganglia were cut using a cryostat at 14 μm and stored at −80 °C. Following the manufacturer's protocol (Advanced Cell Diagnostic), ganglia were hybridized with double-Z oligo probes for the genes listed in Table S2. Signal detection was achieved using DAB and tissue was counterstained with Nuclear Fast Red (Sigma).

RNAScope chromogenic “red” assay combined with immunohistochemistry for YFP

Nav1.8-Cre-ChR2-YFP mice were anesthetized with chloral hydrate (500 mg/kg, i.p.) and transcardiacally perfused with 10% formalin (Sigma). Fixed nodose ganglia were kept in 10% formalin for 48 h at 4 °C before being transferred into 20% sucrose in PBS for an additional 24 h. Ganglia were next frozen and cut at 14 μm onto SuperFrost Plus slides (1:5 series). Tissue was baked at 60 °C for 30 min, washed in 1X PBS, treated in H2O2 for 10 min. Pretreatment consisted of hot 1X Target Retrieval solution for 1–2min. Next, slides were rinsed in distilled water and dehydrated in fresh 100% ethanol. Tissue was incubated with Pretreatment 3 at 40 °C for 15mins. After rinsing the slides in distilled water, hybridization was performed using the standard ACD procedure and reagents from the RNAscope® 2.5 HD Detection Kit (RED). The probes listed in Table S2 were applied at 40 °C for 2hrs. Amplification steps were done using the following the manufacturer's instructions. Lastly, signal detection was achieved using a mix of Fast RED-B and Fast RED-A in a ratio of 1:60 at room temperature for about 10 min. Slides were washed in distilled water. The native tdTomato fluorescence was greatly diminished following in situ hybridization. Nodose ganglia were next labeled using a GFP chicken antibody known to detect YFP (1:1,000). On the next day, tissue was rinsed and incubated in a solution of biotinylated anti-chicken (1:1,000; Cat#703065155; Jackson ImmunoResearch) for 1hr, followed by Streptavidin, AlexaFluor 488 (1:1,000; Cat#S32354; Invitrogen) for 1 h. EcoMount mounting medium was applied to the slides and coverslip over the tissue section. Pilot experiments in adjacent tissue sections were also performed to ensure that the combination of in situ hybridization with immunohistochemistry did not alter the staining obtained with each individual procedure.

RNAScope chromogenic “duplex” assay

Fixed nodose ganglia were obtained from C57Bl/6J mice as described for the “red” assay. Ganglia were next frozen and cut at 8 μm onto SuperFrost Plus slides (1:5 series). Tissue was pretreated with boiling 1X Target Retrieval followed by Protease III at 40 °C for 15mins. After pretreatment, the slides were hybridized with the probes listed in Table S2. Amplification steps were done using the reagents from the RNAscope® 2.5 HD Duplex Assay. Signal detection was achieved using two different chromogenic substrates (HRP-C1-Green and AP-C2-Red). Finally, slides were counterstained with hematoxylin and covered with EcoMount mounting medium.

As a negative control, we used a probe recognizing the prokaryotic gene dapB. Ganglia were completely devoid of a signal. As a positive control, we detected the expression of cyclophilin mRNA across the nodose ganglion (data available upon request).

Microscopy

Images of DAB- and Duplex-labeled ganglia were captured using the standard brightfield optics of a Zeiss Axioplan light microscope attached to a digital camera. Images of FastRed-labeled ganglia were acquired using the 63x oil objective of the Leica Sp5 confocal laser scanning microscope (UT Southwestern Live Imaging Core Facility). Estimates were performed using ImageJ. Specifically, we manually counted YFP-positive cells expressing select receptors and expressed the data as percentage of Nav1.8 coexpression. This was done in 2–4 left ganglia from different mice. For each gene, we counted a total of 262–536 cell profiles. Data were expressed as the percentage of YFP-positive cell profiles expressing red dots. In Duplex-labeled ganglia, we counted the number of neuronal profiles unequivocally containing green and/or red dots in 2–3 left ganglia from different mice. For each combination of gene, we counted a total of 215–572 cell profiles. Data were included in pie charts representing the percentage of cell profiles expressing green, red, or both green and red dots. Lastly, Adobe Photoshop CS5 was used to annotate, crop and adjust the contrast of our digital images.

Patch clamp on ex vivo cultures of vagal afferent neurons

Mice were euthanized by cervical dislocation, and the ganglia was dissected into DMEM (Dulbecco's Modified Eagle Medium) at 4 °C. For a typical experiment, ganglia from 4 mice were pooled prior of incubation at 37 °C in DMEM supplemented with 1.3 Wünsch units/ml Liberese TM (Roche) for ∼1 h. During the first half hour, the tubes were shaken every 10 min. The next half hour, the samples were resuspended using a pipette every 10 min. After 1 h, the samples were resuspended using a pipette every 5 min until the sample was clearly digested and the majority of cells in single cell suspension. The cells were passed through a cell strainer (70 μm) and resuspended in DMEM. The cell suspension was distributed on poly-D-lysin coated coverslips and cultured during 4–6 days in Neurobasal Medium (Gibco) containing 100 units/mL penicillin/streptomycin, 10 mmol/L HEPES, 2 mmol/L Glutamax and B27 supplement (Gibco). On day two, the cultures were treated with 0.5 μM Ara-C in the culture medium for 48 h to eliminate glial cells. For electrophysiological measurements, the neurons on coverslips were transferred to a 1 ml recording chamber and superfused at room temperature with an artificial ganglion solution (AGS) containing (in mM): 152 NaCl, 5 KCl, 10 HEPES, 10 D-Glucose, 1 MgCl2, 2 CaCl2 with a pH of 7.4. Whole-cell patch clamp recordings were performed in current clamp using an AxoClamp 2B amplifier (Molecular Devices, Sunnyvale, CA, USA). Data were digitally acquired at a sampling rate of 10 kHz. Glass micropipettes were pulled from filament capillary glass (O.D. 1.5 mm, I.D. 0.86 mm, Harvard Apparatus, Holliston, MA, USA) using a PUL-100 micropipette puller (World Precision Instruments, Sarasota, FL, USA) to a tip resistance of 4–6 MΩ. Patch pipettes were filled with a solution containing (in mM): 130 HCH3SO3, 130 KOH, 10 HEPES, 0.4 NaGTP, 4 Na2ATP, 5 Na2- phosphocreatine, 4 MgCl2, whose osmolarity measured 310 mOsm with a pH of 7.3. Patch pipettes were visually guided to target tdTomato-positive neurons under visual control using ROE-200 micromanipulators (Sutter Instruments, Novato, CA, USA) on a fixed-stage upright microscope (modified Olympus BX51, Olympus Corporation, Tokyo, Japan) under 40X magnification. Local application of CCK (50 nM dissolved in 50% water, 50% AGS) was done by placing a separate patch pipette close to the cell and applying 5–25 psi pressure for 5 – ∼60 s. Electrophysiological data was acquired using pClamp 10.0 (Molecular Devices, Sunnyvale, CA, USA) and subsequently analyzed using Igor Pro 7 (Wavemetrics, Tigard, OR, USA).

Results

NaV1.8 neurons ablated mice lack unmyelinated vagal afferents

Receptors enriched in Nav1.8-expressing vagal afferents were identified by comparing the transcriptional profile of nodose ganglia from mice lacking Nav1.8 neurons (ablated) to that of control littermates. Successful ablation of Nav1.8 neurons in Nav1.8-Cre-DTA mice was previously demonstrated by us and others [44], [48]. RNAseq comparison of the nodose ganglion of ablated and control mice further revealed very little expression of several markers selective of unmyelinated peptidergic and non-peptidergic sensory neurons, including Na1.8, Trpa1, Trpv1, Sst, and Plxnc1 (Figure S1) [49] in the ablated mice. With the exception of Cacna1h, all genes enriched in myelinated neurofilament-containing afferents including Nefh, Trkb, and Spp1, in ablated animals (Figure S1) [49]. Several genes such as Na1.9, Calca, and Tac1 showed reduced expression in ablated mice (Figure S1). Their residual expression in ablated mice could be attributed to their expression in small subtypes of myelinated neurons [49]. The expression of several neuropeptides appeared downregulated in the nodose ganglion of ablated mice including Sst, Calca, Calcb, Nmb and Pcart (Figure S1). Overall, we confirmed that ablated mice showed a selective lack of unmyelinated vagal afferents including many peptidergic neurons. In the following sections, we characterized the change in expression of GPCRs in afferent vagal neurons upon NaV1.8 ablation.

A group of selected gut peptide receptors are expressed in vagal afferents

It is well established that a major part of the effects of certain gut hormones such as CCK and GLP-1 is mediated through afferent vagal signaling [10], [26], [27], [31]. However, the expression of peptide receptors and their subtypes has not been systematically studied. As shown in Figure 1A, qPCR of the whole nodose ganglion revealed that the expression profile was dominated by Cck1r, Glp1r, Npy2r, and Ntsr1. These receptors are conceivably sensing gut-derived CCK, GLP-1, PYY, and neurotensin. Other receptor subtypes for these ligands including Cck2r, Npy1,4-6r, and Ntsr2 were only expressed at a low level. The Sstr4 somatostatin receptor was expressed above noise level. Interestingly, the GIP receptor Gipr was expressed below the detection limit, thus suggesting that GIP does not exert its function through afferent vagal neurons. Likewise, the expression of the ghrelin receptor, Ghsr was also below detection limit in the nodose ganglion, thereby challenging previous reports that certain effects of these hormones are dependent upon vagal sensory signaling [50], [51], [52]. In support of our data, accumulating evidence indicates that only hypothalamic neurons are required for the orexigenic actions of ghrelin [53], [54], [55].

Figure 1

Gut hormone receptors expression and enrichment in Na1.8-expressing vagal afferents. (A) Expression level of gut hormone receptors in vagal afferent (whole nodose ganglion) of wild-type (WT) mice, the dotted line indicates the median for GPCRs expression. (B) Volcano plot for the fold change of expression level for 408 GPCRs (gray dots) in NaV1.8 neurons ablated vs. control mice. Gut hormone receptors are shown in red with corresponding names, the size of the dots relate to the expression level prior of ablation. (C) In situ hybridization of selected gut hormone receptors (i; Cck1r, ii; Glp1r, iii; Npyr2 and iv; Ntsr1) in red vs. immunohistochemical detection of YFP in the nodose ganglion of NaV1.8 reporter mice (Nav1.8-Cre-ChR2-YFP) in green. Asterisks indicate representative NaV1.8 positive cells. Triangles and arrows indicate examples of gut hormone receptor positive cells and double positive cells, respectively.

As shown above, Nav1.8 ablated mice show a selective loss of unmyelinated vagal afferents. Below, we characterized the changes in expression of GPCRs in the nodose ganglion upon Nav1.8-expressing cells ablation. Ntsr1, Npy2r, and Cck1r were highly enriched in Nav1.8 neurons as judged by their large reduction in expression upon ablation (Figure 1B). Npy2r and Ntsr1 were the most downregulated in ablated animals, indicating their nearly exclusive expression in Nav1.8 cells (Figure 1B). The enrichment was further validated using ISH combined with immunohistochemistry in reporter mice for Nav1.8 neurons (Nav1.8-Cre-ChR2-YFP) [45]. For instance, Npy2r and Ntsr1 were expressed in 71% and 40% of Nav1.8 cells, respectively (Figure 1C). In comparison, Cck1r was enriched in a large proportion of Nav1.8 cells (42%) but was not restricted to Nav1.8 cells (Figure 1C). In contrast, Glp1r was only slightly and only barely significantly affected by the ablation (Figure 1B). Double ISH demonstrated that Glp1r was expressed in only 18% of Nav1.8 afferents (Figure 1C). Moreover, about half of Glp1r-expressing afferents were Nav1.8-negative (52%) and appeared to express higher levels of Glp1r mRNA (Figure 1C).

Functional CCK1 receptor was further validated through whole-cell patch clamp performed on isolated neurons from the nodose. The neurons showed typical electroresponsive properties of nodose ganglion neurons, i.e. single or dual spikes in response to a prolonged depolarizing current pulse (Figure S2). Pressure application of CCK (50 nM) from a separate patch pipette positioned close to the neuron depolarized a subset of neurons 9–38 mV (4 responders, 15 non-responders), accompanied by an increase in the membrane noise and shunting of the membrane. Repeated pressure application to responders resulted in desensitization of the depolarizing response (n = 3).

Overall, vagal unmyelinated afferents were found to be enriched in the receptors for only a few select gut peptides.

NTSR1 is the only gut peptide receptor preferentially expressed in mucosal afferents

Because Nav1.8-expressing vagal afferents do not exclusively innervate the GI mucosa, but also the GI muscularis and several other peritoneal and thoracic organs [5], we judged it necessary to further narrow down the identity of vagal afferents enriched with gut peptide receptors. Specifically, we used double in situ hybridization studies with Cck1r and Gpr65, which identify non-overlapping population of vagal afferents preferentially innervating the GI muscularis and mucosa, respectively [56]. Here, we confirmed that very few neurons coexpressed Cck1r and Gpr65 in the nodose ganglion (Figure 2A). Importantly, about half of the Cck1r-expressing neurons also expressed Glp1r and almost all Glp1r neurons were positive for Cck1r (Figure 2B), whereas Glp1r like Cck1r almost never colocalized with Gpr65 (Figure 2C). These data are in agreement with the view that Cck1r- and Glp1r-enriched vagal afferents represent a mixed population of unmyelinated and myelinated fibers innervating the GI muscularis rather than its mucosa [56]. More than half of the Cck1r-expressing neurons also expressed Npy2r (Figure 2D). This is surprising considering that Npy2r neurons were previously described as unmyelinated neurons innervating the lungs [57]. Nonetheless, we also found that many Npy2r neurons neither express Cck1r nor Gpr65 (Figure 2E), presumably corresponding to pulmonary vagal afferents. Interestingly, nearly all Gpr65-expressing neurons expressed Ntsr1 (Figure 2G), which also was expressed in ∼25% of the Cck1r-neurons (Figure 2F). Neurotensin is normally co-secreted from enteroendocrine cells with GLP1 and PYY in response to metabolic stimuli [58]. Our data imply that neurotensin may act in a paracrine manner on vagal unmyelinated afferents innervating the intestinal mucosa.

Figure 2

Double chromogenic . A, C, E, and G in situ hybridization for Gpr65 in red vs. Cck1r, Glp1r, Npy2r, or Ntsr1 in blue. B, D, F in situ hybridization for Cck1r in red vs. Glp1r, Npy2r, orNtsr1 in blue. Asterisks indicate representative double-labeled cell profiles. Pie charts give the percentage of single positive cells (red or blue) and double positive cells (purple). Bright-field images were collected from the nodose ganglion of wild-type mice. Tissue was counterstained with hematoxylin.

Only few receptors for microbial metabolites are expressed in vagal afferents

A number of GPCRs function as sensors of diet-derived, nutrient metabolites, and metabolites generated by the gut microbiota [22]. Whereas these metabolite receptors are known to be expressed on enteroendocrine cells, their neuronal expression has not been systematically studied. As shown in Figure 3, only 4 metabolite receptors were expressed above background in the nodose ganglion. Specifically, Gpr119, Ffar3 (Gpr41), and CaSR were detectable in the whole nodose ganglion but at relatively low levels (Figure 3A). To our knowledge, only FFAR3 was previously described in vagal afferents [59], [60]. In agreement with those studies, both qPCR and chromogenic ISH revealed that the short chain fatty acid receptor Ffar3 was expressed at low level in very few cells (Figure S3) Interestingly, Ffar3 was not down-regulated in Nav1.8 ablated mice (Figure 3B). The calcium-sensing receptor (CaSR) and Gpr119 were both expressed only a little higher than the median (Figure 3A) and were only moderately decreased upon Nav1.8 ablation (Figure 3B). ISH for Gpr119 mRNA confirmed very low levels of expression in the nodose ganglion compared to other receptors (Figure S3). GPR35, a receptor for aromatic, acidic metabolites such as the Trp metabolite kynurenic acid was the most highly expressed metabolite receptor and one of the most highly enriched GPCRs in NaV1.8 neurons according to the more than 100-fold downregulation upon ablation (Figure 3A). Double chromogenic in situ hybridization further demonstrated that Gpr35 was often coexpressed with Cck1r and almost all of the gastrointestinal mucosa innervating GPR65 expressing neurons (Figure 3C).

Figure 3

Metabolite receptors expression and enrichment in Na1.8-expressing vagal afferents. (A) Expression level of metabolite receptors in vagal afferents (whole nodose ganglion) of wild-type (WT) mice. The dotted line indicates the median for GPCRs expression. (B) Volcano plot for the fold change of expression level, for 408 GPCRs (gray dots) in NaV1.8 neurons ablated vs. control mice. Metabolite receptors are shown in red with corresponding names, the size of the dots relate to the expression level prior of ablation. (C) in situ hybridization for Gpr35 in red vs Cck1r and Gpr65 in blue. Pie charts give the percentage of single positive cells (red or blue) and double positive cells (purple).

CB1R and several inflammatory lipid receptors are highly expressed in vagal afferents

GPCRs also act as molecular sensors of lipophilic hormones and lipids [22]. Dietary lipids, prostaglandins, and endocannabinoids were reported to acutely enhance the electrophysiological activity of vagal afferents [61], [62], [63]. However, little is known about the mechanisms underlying lipids sensing in vagal afferents. Here, we found that a large number of lipid receptors were highly expressed in the whole nodose ganglion (Figure 4A) and, furthermore, enriched in many Nav1.8-expressing cells (Figure 4B). These receptors included Cb1r, S1pr3, Lpar5, and several prostaglandin receptors such as Ptger4, Ptger1, Ptgir, Ptgdr, Ptger1, Ptgir, and Prgdr were moderately enriched in Nav1.8 cells (Figure 4B). In situ hybridization revealed that for instance the cannabinoid receptor Cb1r was detected in ∼65% of Nav1.8 cells, but also in other neurons (Figure 4C). Thus, these receptors were not limited to the Nav1.8 positive vagal afferents. In contrast, other lipids receptors were almost exclusively expressed in Nav1.8 cells. These included Ptge4, Cysltr2, and Lpar3. For instance, Cylstr2 and Ptger4 were detected in 41% and 88% of Nav1.8 cells (Figure 4C), respectively, and no other cells.

Figure 4

Lipid receptors expression and enrichment in Na1.8-expressing vagal afferents. (A) Expression level of lipid receptors in vagal afferent (whole nodose ganglion) of WT mice, the dotted line indicates the median for GPCRs expression. (B) Volcano plot for the fold change of expression level for 408 GPCRs (gray dots) in NaV1.8 neurons ablated vs. control mice. Lipid receptors are shown in red with corresponding names, the size of the dots relate to the expression level prior of ablation. (C) In situ hybridization of selected lipid receptors (i; Cb1r, ii; Cysltr2, iii; Ptger4) in red vs. immunohistochemical detection of YFP in the nodose ganglion of NaV1.8 reporter mice (Nav1.8-Cre-ChR2-YFP) in green. Asterisks indicate representative NaV1.8 positive cells. Triangles and arrows indicate examples of lipid receptor positive cells and double positive cells, respectively.

We verified the expression of select highly enriched lipid receptors in Cck1r and Gpr65-expressing neurons using double chromogenic in situ hybridization. As a result, we found that Cysltr2, and Ptger4 were coexpressed with the vast majority of Gpr65 neurons and a large subset of Cck1r neurons (Figure 5A–D). Cb1r was even more broadly expressed in almost all Cckar- and Gpr65-expressing neurons (Figure 5E,F). In summary, vagal unmyelinated afferents supplying the GI tract were found to be highly enriched in many receptors for immunomodulatory lipids and endogenous cannabinoids.

Figure 5

Double chromogenic . A, C, and E in situ for Cck1r in red vs. Cysltr2, Ptger4, Cb1r in blue. B, D, and F in situ for Gpr65 (a marker for gastrointestinal mucosa) in red vs. Cysltr2, Ptger4, Cb1r in blue. Asterisks indicate representative double-labeled cell profiles. Pie charts give the percentage of single positive cells (red or blue) and double positive cells (purple). Bright-field images were collected from the nodose ganglion of wild-type mice. Tissue was counterstained with hematoxylin.

Multiple orphan GPCRs are broadly expressed in vagal afferents

Unexpectedly, we identified at least 21 orphan receptors that were highly enriched in the whole nodose ganglion and enriched in Nav1.8 neurons (Figure 6A,B). For the majority of these receptors, their function(s) in neuronal signaling is unknown and never described before in vagal afferents. The most highly expressed orphan receptor was Gpr149 (Figure 6A). In agreement with a recent study [56], we observed a high expression of Gpr65 in the nodose ganglion. Importantly Gpr65 was the second most down regulated GPCR upon Nav1.8 ablation (Figure 6A,B), indicating that this receptor normally is highly enriched in Nav1.8-expressing vagal afferents. Here, we further showed using in situ hybridization that Gpr65 is present in about 35% of Nav1.8 cells and is not detectable in Nav1.8-negative cells (Figure 6C). As discussed previously, Gpr65 is a marker for GI mucosal afferents of unknown functions [56].

Figure 6

Orphan receptors expression and enrichment in Na1.8-expressing vagal afferents. (A) Expression level of selected orphan receptors in vagal afferent (whole nodose ganglion) of wild-type (WT) mice. The dotted line indicates the median for GPCRs expression. (B) Volcano plot for the fold change of expression level for 408 GPCRs (gray dots) in NaV1.8 neurons ablated vs. control mice. Orphan receptors are shown in red with corresponding names, the size of the dots relate to the expression level prior of ablation. (C) In situ hybridization of selected orphan receptors (i; Gpr65 and ii; Gpr161) in red vs. immunohistochemical detection of YFP in the nodose ganglion of NaV1.8 reporter mice (Nav1.8-Cre-ChR2-YFP) in green. Asterisks indicate representative NaV1.8 positive cells. Triangles and arrows indicate examples of orphan receptor positive cells and double positive cells, respectively.

Many other orphan receptors (Gpr174, Gpr161, Gpr158, Gpr160, Mrgprb4, Mrgprx1, Gpr22) appeared highly expressed in the nodose ganglion and significantly down regulated in ablated mice (Figure 6A,B). In situ hybridization validated that, among these receptors, the orphan receptor Gpr161 is expressed in almost all (97%) of the Nav1.8 neurons (Figure 6C).

Receptors for classic neurotransmitters and neuropeptides

Numerous neurotransmitters of enteric or epithelial origins are released in the GI mucosa [64]. Our studies revealed that Gabbr1 encoding the non ligand-binding protomer of the GABA dimeric receptor was very highly expressed in the nodose ganglion (Figure 7A,B). In situ hybridization confirmed its high expression level and nearly ubiquitous distribution across vagal afferents (Figure S3). The Grm7 glutamate receptor was also ubiquitously expressed at relatively high level in vagal afferents (Figure 7A,B). Interestingly, the Drd2 dopamine and Chrm4 muscarinic acetylcholine receptor were both enriched in Nav1.8 neurons (Figure 7B). For instance, Drd2 was detected in about 43% of Nav1.8 cells (Figure 7C). Interestingly, both Drd2 and Chrm4 were often coexpressed with Cck1r, but rarely with Gpr65 (Figure 7D). Several serotonin receptors were also identified in Nav1.8 cells (Figure 7A,B), which is interesting as 5-HT is produced from enterochromaffin cells, which constitute the largest population of enteroendocrine cells of the GI tract. Serotonin signaling via 5-HT3 in vagal afferents has been studied in the past [11], [56], but the action of other neurotransmitters is poorly documented. Our data demonstrate that the sensitivity of vagal afferents to classic neurotransmitters is not limited to serotonin but likely extends to acetylcholine, glutamate, GABA, and dopamine.

Figure 7

Neurotransmitter receptors expression and enrichment in Na1.8-expressing vagal afferents. (A) Expression level of selected neurotransmitter receptors in vagal afferent (whole nodose ganglion) of WT mice, the dotted line indicates the median for GPCRs expression. (B) Volcano plot for the fold change of expression level for 408 GPCRs (gray dots) in NaV1.8 neurons ablated vs. control mice. Neurotransmitter receptors are shown in red with corresponding names, the size of the dots relate to the expression level prior of ablation. (C) In situ hybridization of Drd2 in red vs. immunohistochemical detection of YFP in the nodose ganglion of NaV1.8 reporter mice (Nav1.8-Cre-ChR2-YFP) in green. Asterisks indicate NaV1.8 positive cells triangles indicate neurotransmitter receptor positive cells and arrows indicate double positive cells. (D) Double in situ hybridizations for-for Drd2 in blue vs. Gpr65 and Cck1r in red (i; iii) or for Chrm4 in blue vs. Gpr65 and Cck1r in red (ii; iv). Pie charts give the percentage of single positive cells (red and blue) and double positive (purple).

Among receptors for neuropeptides, Crhr2, a receptor for corticotropin-releasing factor, was the most highly expressed in the whole nodose ganglion and highly enriched in Nav1.8 cells (Table S1 and Figure S4). Roughly 75% of Nav1.8 neurons were found to express Crhr2 mRNA (Figure S4). Using double in situ hybridization, it was further shown that Crhr2 colocalized with roughly one third of Cck1r- and half of Gpr65-expressing neurons (Figure S4). Since Crhr2 binding sites are prominent in the nucleus of solitary tract [65], it is likely that Crhr2 may serve as a presynaptic receptor on vagal terminals. Hence, it is tempting to speculate that Crhr2 signaling in gastrointestinal vagal afferents may link stress to the regulation of satiety. Other receptors for neuropeptides were also enriched in vagal afferents including Kiss1r and Opmr1 (Table S1). However, these receptors were moderately enriched in the nodose ganglion compared to Crhr2.

Additional receptors

The expression profile for GPCRs measured by qPCR and RNA-seq are given in Table S1. A few other noteworthy receptors that were highly enriched in vagal Nav1.8 cells included receptors not known to be involved in gut-to-brain communication such as F2r, Ackr4 and Ednra (Table S1 and Figure S5). The function of these receptors in neurons remains largely undocumented.

Conclusions and perspectives

Our current understanding of the molecular mechanisms through which microbial, neuronal, metabolic, and immune signals released in the gut influence vagal functions and thereby signaling to the brain and how these signals interact with each other remains incomplete. In the past, transcriptional profiling studies for GPCRs through which such signals act have been proven useful to our understanding of enteroendocrine cell function [66], [67]. In the present study, we provide a comprehensive picture of the expression and coexpression of GPCRs in the nodose ganglion focusing on afferent vagal fibers innervating the GI tract.

The complete RNA expression profile for vagal afferent in Nav1.8 ablated mice and control mice found through RNA-sequencing is given in Supplementary Table 3. It is implied that genes expressed highly in control mice and lowly after ablation are enriched in the NaV1.8 fibers. However, there are potential caveats in this assumption as incomplete ablation will hinder detection for some genes, and ablation induced gene regulation could result in false positive and/or negative results. The ablation procedure has previously been shown to be very effective [44], and here we also show a strong decrease of genes specific for unmyelinated vagal afferents (Figure S1). However, we cannot rule out that some Nav1.8 fibers are still present; likewise, we cannot rule out potential ablation induced gene regulation. To minimize this potential caveat, we also examined gene expression in the ganglia of wild-type mice and performed in situ hybridization for select receptors. Importantly, the distribution and expression levels for receptors we examined by qPCR and ISH matched very well with the data deriving from our RNAseq dataset after ablation, but these caveats should be kept in mind when interpreting our RNA sequencing dataset.

It should also be noted that transcriptional data do not always correspond to protein levels. However, it is notoriously difficult to generate specific antibodies towards GPCRs, thus hindering immunohistochemical detection. This may explain that our data are not always consistent with the literature. For instance, the ghrelin receptor has been reported to be expressed in vagal afferent using immunostaining [68]. However, our study failed to detect the ghrelin receptor mRNA. Alternatively, past studies have successfully characterized the localization and density of binding sites for several gut peptides in vagal cells using radioligands and binding assays [69], [70]. However, binding assays do not allow the study of orphan receptors and are not suitable for the high-throughput screening of a large number of receptors. Hence, there are few methods available to systematically assess the protein localization and level of the receptors we identified in this study.

Many physiological studies have established that vagal afferents provide the postprandial feedback necessary for meal termination [71], [72], [73]. However, there is ongoing controversy as to the physiological importance of gut peptides signaling in vagal afferents [74]. Here we find that vagal afferents expressed relatively few, selected gut peptide receptors, which are typically low or absent in other types of peripheral afferents [75], yet abundant in GI tissues [66], [67]. For instance, CCK1R and NTSR1 were enriched in vagal unmyelinated afferents, but not CCK2R, or NTS2R. Interestingly, the receptor for the other incretin hormone GIP was not expressed in afferent vagal nerves, indicating that GIP functions mainly in a classical endocrine manner. The receptor for ghrelin, a gastric hormone, which exerts its many different functions though the CNS was not expressed in vagal afferents. This supports the notion that ghrelin does not function through the afferent vagus nerve but mainly through endocrine mechanisms getting access to ghrelin receptors in the CNS [53], [54], [55]. Thus, only four specific receptor subtypes for gut peptides were found to be enriched in vagal Nav1.8 cells and, importantly, to be coexpressed in a rather clear pattern, i.e. CCK1R together with GLP1R and the PYY Y2 receptor. In contrast, NTSR1 was expressed in another population of afferent neurons together with the orphan receptor GPR65 (Figure 8). Many of the GPR65 expressing neurons have previously been demonstrated to also express the 5-HT3 receptor, a ligand gated ion-channel activated by serotonin (5-HT) released from the enterochromaffin cells [76]. The enterochromaffin cells have further been described to potentially form “synaptic-like” contacts with fibers from the enteric nervous system and being electrically excitable with depolarization leading to serotonin release [76]. Similar “synaptic-like” contacts between enteroendocrine cells and enteric nerves have been described [77]. However, our anatomical data indicates that similar “synapse-like” contacts may not occur frequently between the axons of vagal afferents and enteroendocrine cells (Figure S6).

Figure 8

Schematic overview of GPCRs enriched in gastrointestinal vagal afferents. This study identified several GPCRs preferentially enriched in unmyelinated vagal afferents (Nav1.8). Identified receptors encompassed a wide range of GPCRs families including receptors for neurotransmitters, metabolites, lipids, and gut peptides. We further inferred their localization in neurons projecting preferentially to either the gastrointestinal muscularis CCK1 and GLP-1 receptor expressing fibers or the mucosa GPR65 expressing fibers based on Williams and coworkers [7] and our results. Strikingly, most receptors showed a relatively widespread distribution in both categories of neurons. Receptors in green (Gαs-coupled) and orange (Gαq-coupled) are anticipated to stimulate neuronal activity. Those in red (Gαi-coupled) are anticipated to inhibit neuronal activity. Nonetheless, the downstream signaling pathways of these receptors are not known with certainty in vagal afferents. Of note, only receptors found to be mostly highly enriched in unmyelinated vagal afferents are listed here. For the purpose of simplification, orphan receptors were not indicated. NTS, nucleus of the solitary tract.

Consequently, neurotensin emerges as an important gut hormone with a corresponding receptor (NTSR1) preferentially expressed in the GI mucosa (paracrine function), while for the other gut hormones CCK, GLP1, and PYY the hormone has to travel further to reach a nerve with a corresponding receptor (paracrine/hormonal function). It is noteworthy that neurotensin immunoreactivity and binding sites are also very abundant in the nucleus of the solitary tract in the brain stem [70], [78]. Thus, more studies are warranted to establish whether neurotensin acts on vagal terminals located either in the GI tract or the nucleus of the solitary tract. In the case of ghrelin and GIP, it appears that these must reach the CNS (classical hormone function).

The ability of unmyelinated afferents to directly respond to proinflammatory lipids has been reported in isolated vagal afferents [79], [80], and it is plausible that this mechanism has been preserved during evolution with the aim of allowing peripheral sensory neurons to detect infections and tissue injuries. Here, we found that vagal afferents including those innervating the GI tract are particularly highly enriched in receptors for leukotrienes, sphingolipids, and prostaglandins (Figure 8). These receptors have been described in spinal nociceptors [75], suggesting that they play a key role in responding to noxious stimuli. However, unlike their spinal counterparts, vagal afferents, including those innervating the GI mucosa, do not convey pain [18]. Instead, we speculated that the stimulation of vagal unmyelinated afferents by inflammatory lipids would trigger an efferent response aimed at containing inflammation and maintaining tissues integrity, a biological process known as “inflammatory reflex” [81]. This hypothesis is in agreement with our data showing that unmyelinated axons in the GI mucosa are tightly associated with macrophages rather than enteroendocrine cells (Figure S6).

Receptors for several classic neurotransmitters were found to be enriched in vagal afferents including both unmyelinated and myelinated fibers. As mentioned before, mucosal 5-HT can stimulate the activity of vagal afferents [76]. However, it is unclear what could be the source(s) of the other identified neurotransmitters and neuropeptides. Hypothetically, neurotransmitters deriving from the enteric nervous system including acetylcholine and glutamate could act on vagal terminals [82]. The aforementioned neurotransmitters and peptides may also be released within the brainstem and modulate the activity of vagal afferents at the presynaptic level [65]. Hence, more studies are warranted to determine whether the receptors for neurotransmitters and neuropeptides may serve as presynaptic receptors in addition to or rather than peripheral receptors.

Vagal afferents have been proposed to serve as a relay between the GI microbiome and brain functions [83], [84], presumably in response to the activation of vagal afferents by microbial metabolites. The fact that peripheral sensory neurons reside outside the blood–brain barrier makes them a particularly relevant target for pharmacological agents mimicking metabolites. Considering the well-documented interactions between microbial metabolites and GPCRs [22], we sought to examine the expression of known or proposed microbial metabolite receptors in vagal afferents. The most abundant metabolite receptor found in vagal neurons was GPR35, thus raising the possibility that vagal GPR35 signaling may be involved in the detection of aromatic acids such as the tryptophan metabolite kynureic acid (Figure 8). Interestingly, GPR35 was extensively coexpressed both with CCK1R and GPR65, two markers of GI afferents. However, surprisingly few other metabolites receptors were identified in vagal afferents, and usually at very low levels. This included FFAR3, a receptor previously described in a small subset of vagal afferents [59], [60]. Importantly, intact vagal afferents were shown to be required for FFAR3 effects on metabolism [60]. It is therefore conceivable that a small number of microbiome-derived molecules may directly act on GPCRs expressed by vagal afferents. Nonetheless, most microbial metabolites are unlikely to directly target vagal afferents. Instead, they may affect afferent vagal nerves through enteroendocrine cells such as enterochomaffin cells, which are highly enriched in multiple types of receptors for microbial metabolites [85].

In summary, the unexpected identification of many GPCRs with unknown ligands and functions in vagal afferents highlights how little is known about vagal signaling. Many of the genes identified in this study may be used to generate novel transgenic mice to investigate the functional neuroanatomy of vagal afferents, for instance, using diphtheria toxin-assisted ablation tools [86] to determine their exact functions. Thus, we are hopeful that our data will guide future studies aimed at addressing the role of GPCRs signaling in visceral afferents in relation to metabolic and immune sensing, as well as metabolic diseases and their complications.