Authors: Kouki K. Touhara (1Department of Physiology, University of California, San Francisco, United States), Nathan D. Rossen (1Department of Physiology, University of California, San Francisco, United States; 2Tetrad Graduate Program, University of California, San Francisco, United States), Fei Deng (3State Key Laboratory of Membrane Biology, New Cornerstone Science Laboratory, School of Life Sciences, Peking University, China; 4PKU-IDG/McGovern Institute for Brain Research, China), Joel Castro (6Visceral Pain Research Group, Hopwood Centre for Neurobiology, Lifelong Health Theme, South Australian Health and Medical Research Institute (SAHMRI), North Terrace, Adelaide, South Australia 5000, Australia; 7Faculty of Health and Medical Sciences, University of Adelaide, North Terrace, Adelaide, South Australia 5000, Australia), Andrea M. Harrington (6Visceral Pain Research Group, Hopwood Centre for Neurobiology, Lifelong Health Theme, South Australian Health and Medical Research Institute (SAHMRI), North Terrace, Adelaide, South Australia 5000, Australia), Tifany Chu (5Departmen of Pediatrics, University of California, San Francisco, United States), Sonia Garcia Caraballo (6Visceral Pain Research Group, Hopwood Centre for Neurobiology, Lifelong Health Theme, South Australian Health and Medical Research Institute (SAHMRI), North Terrace, Adelaide, South Australia 5000, Australia), Mariana Brizuela (6Visceral Pain Research Group, Hopwood Centre for Neurobiology, Lifelong Health Theme, South Australian Health and Medical Research Institute (SAHMRI), North Terrace, Adelaide, South Australia 5000, Australia), Tracey O’Donnell (6Visceral Pain Research Group, Hopwood Centre for Neurobiology, Lifelong Health Theme, South Australian Health and Medical Research Institute (SAHMRI), North Terrace, Adelaide, South Australia 5000, Australia), Jinhao Xu (1Department of Physiology, University of California, San Francisco, United States), Onur Cil (5Departmen of Pediatrics, University of California, San Francisco, United States), Stuart M. Brierley (6Visceral Pain Research Group, Hopwood Centre for Neurobiology, Lifelong Health Theme, South Australian Health and Medical Research Institute (SAHMRI), North Terrace, Adelaide, South Australia 5000, Australia; 7Faculty of Health and Medical Sciences, University of Adelaide, North Terrace, Adelaide, South Australia 5000, Australia), Yulong Li (3State Key Laboratory of Membrane Biology, New Cornerstone Science Laboratory, School of Life Sciences, Peking University, China; 4PKU-IDG/McGovern Institute for Brain Research, China), David Julius (1Department of Physiology, University of California, San Francisco, United States)
Categories: Article
Source: Nature
Topological Segregation of Stress Sensors Along the Gut Crypt-Villus Axis
Authors: Kouki K. Touhara, Nathan D. Rossen, Fei Deng, Joel Castro, Andrea M. Harrington, Tifany Chu, Sonia Garcia Caraballo, Mariana Brizuela, Tracey O’Donnell, Jinhao Xu, Onur Cil, Stuart M. Brierley, Yulong Li, David Julius
Abstract
The crypt-villus structure of the small intestine serves as an essential protective barrier whose integrity is monitored by the gut’s complex sensory system, in which serotonergic enterochromaffin (EC) cells play an important part^1,2^. These rare sensory epithelial cells surveil the mucosal environment for luminal stimuli and transmit signals both within and outside the gut^3–6^. However, whether EC cells in crypts and villi detect different stimuli or elicit distinct physiological responses remains unknown. Here, we address these questions by exploiting a new reporter mouse model to quantitatively measure release and propagation of serotonin from EC cells in live intestines. We showed that crypt EC cells exhibit a tonic low-level mode that activates epithelial 5-HT4 receptors to modulate basal ion secretion, as well as a stimulus-evoked high-level mode that activates 5-HT3 receptors on sensory nerve fibers. Both modes can be initiated by the irritant receptor, TRPA1, which is confined to crypt EC cells, and whose activation by luminal irritants is enhanced when the protective mucus layer is compromised. Villus EC cells also signal damage through a distinct mechanism whereby oxidative stress activates TRPM2 channels, leading to corelease of serotonin and ATP and consequent excitation of sensory nerve fibers. This topological segregation of EC cell functionality along the complex mucosal architecture constitutes a mechanism for surveilling, maintaining, and protecting gut integrity under diverse physiological conditions.
Introduction
The gastrointestinal (GI) tract is equipped with a complex sensory system that detects the state of the gut mucosa and transmits signals within and outside this visceral organ. The first line of stimulus detection is mediated by enteroendocrine cells, which are rare specialized sensory cells within the gut epithelium that release hormones and neurotransmitters in response to endogenous and exogenous stimuli^7^.
Enterochromaffin (EC) cells are a subclass of excitable enteroendocrine cells that release serotonin (5-HT) in response to bacterial metabolites, neurotransmitters, peptide hormones, interleukins, and ingested or endogenous irritants^1–6,8^. Interestingly, EC cells also show spontaneous (basal) activity for which an underlying mechanism or physiologic role remains unknown^4,9^.
Several serotonin receptor subtypes are present in the gut^2^, including ionotropic 5-HT3 and metabotropic 5-HT4 receptors, which have been pharmacologically targeted to treat GI dysregulation associated with diarrhea or constipation^2,10^. EC cells transduce signals to afferent sensory nerve fibers within the mucosa that express 5-HT3 receptors, which are excitatory ion channels activated by relatively high (micromolar) concentrations of serotonin^1,2,11^. This serotonergic EC cell-sensory neuron circuit modulates a range of processes, including GI motility, secretion, nausea, and pain^2,12,13^. In contrast, G protein-coupled 5-HT4 receptors on intestinal epithelial cells are activated by relatively low (nanomolar) levels of serotonin, leading to enhanced ion secretion^14,15^. This latter process plays a crucial role in maintaining fluid balance in the gut, aiding digestion, and protecting the intestinal lining by supporting formation of a mucus barrier^16^. However, we do not fully understand the mechanisms under which EC cells achieve the thousand-fold difference in serotonin concentration required to differentially activate 5H-T3 versus 5-HT4 receptor, and whether or how this relates to the excitability status of EC cells under distinct physiological circumstances.
Such potential differential actions raise interesting questions about the temporal and spatial nature of serotonergic signaling in the intestine and how this relates to the arrangement of crypts and villi that define the complex architecture of the gut. Crypts are small invaginations within the epithelium that house stem cells, which are vulnerable to microbes and irritants and thus protected by antimicrobial peptides and a thick layer of mucus^17^. Villi are long finger-like projections that extend from crypts towards the lumen and are more directly exposed to the luminal contents. Interestingly, EC cells change their molecular identity as they migrate from crypts to villi; for example, two distinct members of transient receptor potential (TRP) ion channel family that sense environmental stress are segregated within the crypt-villus axis. TRPA1 (a.k.a. the wasabi receptor), which detects reactive electrophilic irritants, is located predominantly in crypts, whereas TRPM2, which senses intracellular ADP-ribose as a measure of oxidative stress, is found mostly in villi^18–20^. An important question is whether EC cells in these different tissue compartments respond to distinct stimuli to elicit differential physiological actions, and how this relates to distribution of 5-HT receptor subtypes and the dynamics of transmitter release within the crypt-villus architecture. Addressing these important questions requires the development of new quantitative approaches for analyzing signaling in a complex organ structure with spatial and temporal resolution, as has been advocated for the gut^21^.
In this study, we describe a mouse model that allows for direct observation of serotonin release and propagation within the intact crypt-villus architecture. Using this system, we determine how differential release of serotonin from EC cells promotes distinct physiological responses, and how crypt and villus EC cells employ different TRP channels and transmitters to detect and transduce exogenous or endogenous stress signals within and beyond the gut.
Results
Spatiotemporal dynamics of gut serotonin
To monitor real-time release and propagation of serotonin within the crypt-villus architecture, we set out to develop a mouse model in which genetically encoded GPCR-activation-based serotonin sensors are expressed in the intestinal epithelium^22,23^. gGRAB5HT3.0 is an improved sensor whose fluorescence intensity increases upon serotonin binding (Extended Data Fig. 1a)^23^. We developed a transgenic mouse line that expresses gGRAB5HT3.0 and a red-shifted Ca^2+^ indicator, jRGECO1a, following exposure to Cre recombinase (Extended Data Fig. 1b). Crossing this reporter line to Vil1-Cre mice resulted in expression of gGRAB5HT3.0 and jRGECO1a in intestinal epithelial cells (Extended Data Fig. 1c), enabling us to visualize the release and propagation of serotonin within the gut using fluorescence microscopy. It should be noted that expression levels of gGRAB5HT3.0 and jRGECO1a are slightly higher in villi compared to crypts (Extended Data Fig. 1d) and decrease progressively in the proximal and distal colon, consistent with the expression pattern of the Vil1 gene^24^ (Extended Data Fig. 1c). To visualize reporter activation, we removed a section of jejunum, flushed the luminal contents, and then filleted the tissue to create a flat sheet. We imaged the tissue from the smooth muscle side to visualize crypts or from the luminal side to observe villi. When exposed to a high K^+^ solution, EC cells released serotonin, which subsequently activated the gGRAB5HT3.0 sensor, which is expressed in both EC cells and adjacent epithelial cells within crypts and villi (Fig. 1a–b, and Supplementary videos 1 and 2). Serendipitously, we observed the highest fluorescence intensity of gGRAB5HT3.0 in EC cells in either fixed or live tissues, which facilitated their identification during imaging (Fig. 1c–e).
Two modes of serotonin release in crypts
Having established a tool to visualize serotonin release in the intact intestine, we observed an interesting differential profile whereby tonic serotonin release was seen in crypts but not villi (Fig. 2a and Supplementary video 3). This tonic release was similarly observed when crypts were isolated from intact tissue (Extended Data Fig. 2a). We noticed that this differential activity aligned with preferential expression of TRPA1 in crypt EC cells^18–20^. Indeed, in situ hybridization histochemistry revealed robust Trpa1 expression in crypts that diminished progressively from lower to upper villi (Extended Data Fig. 2b). This was further supported by reanalysis of a published single-cell RNA sequencing dataset^20^, which revealed a gradient of Trpa1 expression, with highest levels in crypt EC cells and progressively lower expression towards the villus (Extended Data Fig. 2c). To functionally validate these expression data, we assessed serotonin release from dissociated crypts and upper villi upon stimulation with the TRPA1 agonist, allyl isothiocyanate (AITC). Consistent with the observed Trpa1 expression gradient, AITC-evoked serotonin release was detected in dissociated crypts but not in similarly dissociated upper villi (Extended Data Fig. 2d). To determine whether TRPA1 contributes to tonic serotonin release in crypt EC cells, we examined intestinal organoids derived from Tac1-IRES-Cre;GCaMP5g-IRES-tdTomato mice, which express the Ca^2+^ indicator GCaMP5g and tdTomato specifically in EC cells (Extended Data Fig. 2e). In these organoids, which replicate crypt features^18^, a TRPA1 antagonist, A967079 (A96), diminished the spontaneous activity of EC cells (Fig. 2b). We consistently observed A96-sensitive spontaneous TRPA1 channel openings in single-cell voltage-clamp recordings from EC cells (Extended Data Fig. 2f) and found that A96-sensitive spontaneous membrane depolarizations drove repeated action potentials (Extended Data Fig. 2g). Additionally, spontaneous calcium flux in EC cells was inhibited by tetrodotoxin (TTX), a voltage-gated sodium channel (NaV) blocker (Extended Data Fig. 2h). Taken together, these results demonstrate that low levels of TRPA1 channel opening are sufficient to drive NaV-dependent action potentials, leading to tonic serotonin release from crypt EC cells.
The normalized gGRAB5HT3.0 sensor response suggested that tonic serotonin release is only a fraction of stimulated release (Fig. 2c). To verify this observation, we quantitatively compared tonic and AITC-evoked serotonin release from crypt EC cells in intestinal organoids. For this experiment, we used the low-affinity serotonin sensor, gGRAB5HT2m, whose affinity is more suitable for measuring serotonin within the micromolar range^23^. We positioned a human embryonic kidney (HEK293) cell expressing gGRAB5HT2m adjacent to (5 μm away from) an EC cell in Tac1-IRES-Cre;GCaMP5g-IRES-tdTomato organoids. This allowed us to monitor spontaneous and AITC-evoked calcium signals in EC cells while simultaneously monitoring serotonin release with the juxtaposed biosensor cell (Fig. 2d). At the end of each recording, we applied a maximally effective concentration of serotonin to fully activate the gGRAB5HT2m sensor, a normalization step that enabled us to estimate the local concentration of released serotonin based on the dose-response curve of the sensor (Extended Data Fig. 2i). Indeed, we observed a substantially lower GCaMP signal during the tonic phase compared to the AITC-stimulated phase (Fig. 2e) with a corresponding difference in the level of released serotonin (Fig. 2f). Specifically, tonic serotonin release remained at a minimal level such that the normalized gGRAB5HT2m amplitude surpassed 0.63 (corresponding to 1 μM serotonin) in just 1 of 8 cells examined. Conversely, for AITC-stimulated release, the normalized gGRAB5HT2m amplitude exceeded 0.63 in 6 of 9 cells evaluated.
If tonic serotonin release is in the nanomolar range, then basal EC cell activity should predominantly activate high affinity metabotropic 5-HT receptors, whereas stimulated EC cells should activate both metabotropic and lower affinity ionotropic receptors. To test this hypothesis, we first monitored GCaMP signals in EC cells within organoids to observe their activity while simultaneously measuring whole-cell currents in neighboring HEK293 biosensor cells expressing ionotropic 5-HT3 receptors(Fig. 2g)^11^. As expected, we found that peak 5-HT3 currents were substantially smaller during tonic serotonin release compared to AITC-stimulated release (Fig. 2g–i), consistent with the idea that EC cells activate ionotropic 5-HT3 receptors most robustly when stimulated by agonists.
On the other hand, tonic serotonin release should activate 5-HT4 or 5-HT2 metabotropic receptor subtypes, which exhibit nanomolar sensitivity to serotonin^14,25^. To test this prediction, we developed a biosensor in which HEK293 cells co-express Gq-coupled 5-HT2A receptors and GCaMP8m. When activated, 5-HT2A receptors promote endoplasmic reticulum (ER)-stored Ca^2+^ release, which results in increased GCaMP8m fluorescence. Notably, low-nanomolar concentrations of serotonin repeatedly activated this biosensor without apparent desensitization (Extended Data Fig. 2j). Despite observing a more robust calcium response in EC cells during the AITC-stimulated phase, peak 5-HT2A biosensor responses during tonic and AITC-stimulated release were equivalent (Fig. 2j–l). Taken together, the gGRAB5HT2m, 5-HT3, and 5-HT2A biosensor experiments are in complete agreement and suggest that tonic nanomolar serotonin activates metabotropic, rather than ionotropic 5-HT receptors. Conversely, when crypt EC cells are stimulated by electrophiles (or other agonists), they release micromolar concentrations of serotonin, recruiting both metabotropic and ionotropic 5-HT receptors.
5-HT receptors in the crypt-villus axis
What are the physiological consequences of tonic versus stimulated serotonin release in crypts? We focused on 5-HT3 and 5-HT4 subtypes as the most relevant targets in the intestine^2,10^. We first investigated distribution of the 5-HT4 receptor, which is known to stimulate epithelial chloride secretion, thereby influencing the rate of fluid transfer into the intestinal lumen^15^. Analysis of published single-cell RNA sequencing datasets revealed that 5-HT4 receptors are expressed exclusively within crypts, primarily in progenitor cells with slightly lower expression in Paneth cells (Extended Data Fig. 3a). Consistent with this, in situ hybridization histochemistry showed that 5-HT4 transcripts are expressed in crypts (but not villi) (Fig. 3a–b), specifically by Olfm4-positive progenitor and Lyz-positive Paneth cells (Extended Data Fig. 3b). We therefore would expect that tonically released serotonin from crypt EC cells would activate these receptors. To demonstrate the communication between EC cells and 5-HT4 receptors in crypts, we used an Ussing chamber to measure ion secretion in ex vivo intestinal preparations. Consistent with previous findings, bath applied serotonin stimulated ion secretion in a 5-HT4-dependent manner (Fig. 3c and Extended Data Fig. 3c)^15^. To specifically activate EC cells, we used Tac1-IRES-Cre;ePet-Flp;hM3Dq mice, in which deschloroclozapine (DCZ) triggers serotonin release from EC cells selectively expressing excitatory DREADD receptors^13^. Application of DCZ stimulated ion secretion, demonstrating a functional communication axis between EC cells and 5-HT4 receptor-expressing crypt cells (Fig. 3d and Extended Data Fig. 3c).
In light of these results, we examined the contribution of tonic serotonin release to basal ion secretion using intestinal organoids. Activation of the stimulatory G protein (Gs) pathway is known to induce fluid secretion, leading to organoid swellings^26^. Indeed, acute exposure to serotonin produced organoid swelling in a 5-HT4 (but not 5-HT3) dependent manner (Fig. 3e). Furthermore, a reduction in basal swelling was observed when organoids were incubated with a TRPA1 antagonist (A96) or 5-HT4 antagonist (RS) (Fig. 3f), suggesting that TRPA1-induced tonic serotonin release contributes to organoid swelling via activation of 5-HT4 receptors. Notably, long-term exposure to A96 and RS did not affect overall organoid growth, excluding the potential impact of these drugs on organoid proliferation (Extended Data Fig. 3d). Based on these findings, we conclude that TRPA1-dependent tonic serotonin release from crypt-residing EC cells activates 5-HT4 receptors, predominantly localized in crypts, thereby stimulating ion secretion.
We next asked whether crypt and villus EC cells can target ionotropic 5-HT3 receptors, which require micromolar serotonin for activation. It is known that both intrinsic and extrinsic sensory afferents express ionotropic 5-HT3 receptors^13,27,28^. To further characterize the expression of 5-HT receptors in these sensory neurons, we first reanalyzed a published single-cell RNA sequencing dataset^29^ and found that intrinsic primary afferent neurons predominantly express 5-HT3 receptors, with minimal expression of other 5-HT receptor subtypes (Extended Data Fig. 4a). Next, we performed retrograde tracing from the small intestine mucosa and observed that the majority of labeled extrinsic mucosal afferents originate from nodose ganglia, with only sparse mucosal afferent innervation from dorsal root ganglia (Extended Data Fig. 4b). Single-cell RT-PCR analysis of these traced vagal mucosal neurons confirmed the expression of 5-HT3 receptors and absence of 5-HT4 receptors (Extended Data Fig. 4c). To functionally validate these findings, we performed jejunal afferent recordings in the presence of a selective 5-HT4 receptor antagonist. Consistent with the gene expression data, blocking 5-HT4 receptors had no effect on basal nerve fiber activity (Extended Data Fig. 4d), further supporting the predominant role of 5-HT3 receptors in mediating serotonergic signaling in mucosa-innervating sensory neurons. Collectively, these findings suggest that mucosa-innervating extrinsic sensory neurons require exposure to micromolar concentrations of serotonin for activation via 5-HT3 receptors.
We then set out to determine the precise localization of 5-HT3 receptors along nerve fibers; specifically, whether they are preferentially expressed near EC cells, concentrated at nerve terminals, or broadly distributed along the nerve fibers. For this purpose, we deployed a modified 5-HT3A-specific nanobody (VHH15)^30^ fused to mCherry, which specifically labeled heterologously expressed 5-HT3 receptors or native receptors in nodose and dorsal root ganglia (Fig. 3g and Extended Data Fig. 5a–b). Labeled nerve fibers innervating the mucosa were seen in both the small and large intestine, revealing uniform channel expression along the length of these afferents (Fig. 3h and Extended Data Fig. 5b).
Thus, EC cells in crypts or villi should be capable of activating mucosal afferents so long as the released transmitter reaches micromolar concentrations. To directly assess the propagation of EC-derived serotonin, we first established the distance between the basolateral side of EC cells and the closest nerve fibers (Fig. 3i and Extended Data Fig. 5c), which on average (5.5 ± 4.5 μm) is comparable to that between EC and biosensor cells used in quantification experiments described above, indicating that most EC cells can present micromolar levels of serotonin to mucosal afferents when stimulated. By using Insm1-Cre;gGRAB5HT3.0 sensor mice, which express gGRAB5HT3.0 in both EC cells and nerve fibers (Extended Data Fig. 5d), we could directly visualize the propagation of serotonin between EC cells and nerves. Indeed, we observed that serotonin released from either crypt or villus EC cells readily reached and covered nearby nerve fibers (Extended Data Fig. 5e–f). In summary, our immunohistochemical and serotonin sensor analyses show that both crypt and villus EC cells are sufficiently close to 5-HT3 receptors on mucosal sensory nerve fibers to transmit excitatory serotonergic signals to these afferents.
Mucus degradation unmasks TRPA1
What physiologic circumstances might promote bolus serotonin release from crypt EC cells? As noted above, TRPA1 is an irritant-activated ion channel that responds to a wide range of electrophilic toxicants and inflammatory agents^31^. These include pungent agents from wasabi, garlic, onion, and other members of the Brassica and Allium plant family, which are associated with both potential health benefits and risks^32,33^. Also, environmental toxicants or metabolic byproducts of certain chemotherapeutic drugs are strong electrophiles that elicit severe inflammation in internal organs^34^. We were therefore curious to know whether such dietary electrophiles (AITC, allicin, and cinnamaldehyde) or metabolites (acrolein and 2-pentenal) could activate crypt EC cells. In organoids from Tac1-IRES-Cre;GCaMP5g-IRES-tdTomato mice, each of these agents elicited robust responses in EC cells that were blocked by the TRPA1 antagonist, A96 (Fig. 4a and Extended Data Fig. 6a).
We next examined the effects of electrophiles in freshly prepared gut tissue, where the crypt structure and its protective mucus layer are preserved. Using gGRAB5HT3.0 mice, we were surprised to find that only acrolein, a highly reactive electrophile, robustly stimulated crypt EC cells, whereas the other tested electrophiles were ineffective (Fig. 4b–c, and Extended Data Fig. 6b). We reasoned that if the mucus layer acts as a barrier to irritant access, then its degradation should increase susceptibility of crypt EC cells to weaker electrophiles. To test this hypothesis, we incubated gut tissue with StcE, a mucinase from pathogenic E.coli O157 that digests Muc2, a primary component of intestinal mucus^35,36^. To determine whether StcE digestion enhances crypt access, we exposed gut tissue to a fluorescently tagged electrophile, BODIPY-iodoacetamide (BODIPY-IA). In the presence of an intact mucus layer, BODIPY-IA infiltrated only the villus tips, even after 15 minutes of incubation (Fig. 4d). Remarkably, StcE treatment allowed BODIPY-IA to reach the crypts in as little as 3 minutes, supporting the idea that the mucus layer restricts electrophile access to the crypt region (Fig. 4d). Subsequently, we exposed StcE-treated intestinal tissue from gGRAB5HT3.0 mice to previously ineffective electrophiles and found that they readily activated crypt EC cells after mucus digestion (Fig. 4b–c, and Extended Data Fig. 6c). Thus, we conclude that under normal conditions a protective mucus layer permits only very potent and permeable electrophiles, such as acrolein, to access EC cells from the luminal side and activate TRPA1. However, when this protective mucus barrier is compromised, then other electrophilic irritants, such as those from dietary sources, can gain access to crypt EC cells and stimulate serotonin release.
TRPM2 senses oxidative stress in villi
Given that TRPA1 is preferentially expressed by crypt EC cells (Extended Data Fig. 2b–c), we wondered if villus EC cells can also detect chemical irritants or other sentinels of tissue damage. Reanalysis of a published single-cell RNA sequencing dataset revealed a gradient of Trpm2 expression^20^, with the highest transcript levels in villus EC cells and progressively lower expression toward the crypt (Extended Data Fig. 2c). This finding was corroborated by our in situ hybridization analysis, which showed preferential expression of TRPM2 channels in villus EC cells (Extended Data Fig. 7a). TRPM2 is an excitatory ion channel activated by intracellular ADP-ribose, which is generated during oxidative stress^37^. Indeed, whole-cell patch-clamp recordings revealed ADP-ribose-activated inward currents in EC cells (Fig. 5a). These currents were attenuated by 2-aminoethoxydiphenyl borate (2-APB), which inhibits TRPM2, and reduced upon replacement of extracellular Na^+^ with NMDG^+^. Consistent with the expression pattern of TRPM2, we observed significantly larger ADP-ribose-activated currents in villus versus crypt EC cells (Fig. 5a).
We next asked whether oxidative stress activates villus EC cells. Using dissociated EC cells expressing GCaMP5g, we found that 200 μM H2O2 robustly activated villus EC cells. This activation was blocked by 2-APB, but not by A96, demonstrating TRPM2-dependent activation (Fig. 5b and Extended Data Fig. 7b). Furthermore, in a small intestinal injury model induced by administration of anti-CD3 antibody^38^, which increased epithelial cell apoptosis as indicated by elevated cleaved caspase-3 positive cells (Extended Data Fig. 7c), we observed an increase in serotonin release in villi but not in crypts (Extended Data Fig. 7d–e). Importantly, this increased serotonin release was blocked by 2-APB, confirming TRPM2-dependence. These findings suggest that TRPM2 channels sense inflammatory stress of epithelial cells within villi.
Villus ECs co-release serotonin and ATP
Villus EC cells do not exhibit tonic serotonin release (Fig. 2a), and expression of 5-HT4 receptors in villi is minimal (Fig. 3a–b). Given these findings, we hypothesized that villus EC cells might have distinct neurotransmitter release patterns and targets. While it has been suggested that EC cells release ATP alongside serotonin^39^, we tested this directly using a genetically encoded ATP sensor, gGRABATP1.0^40^. EC cells within organoids or primary isolated EC cells were depolarized with high K^+^ (assessed by GCaMP fluorescence) while simultaneously monitoring the gGRABATP1.0 signal in neighboring biosensor cells (Fig. 5c). We treated organoids with BMP4, a key regulator of crypt-villus axis differentiation^18^, which resulted in loss of spontaneous EC cell activity (Extended Data Fig. 8a), indicative of successful crypt-to-villus transformation. Interestingly, we found that ATP was only released from primary dissociated single villus cells or villus-differentiated organoids, but not from isolated primary crypts or organoids where crypt EC cells predominate (Fig. 5c–d and Extended Data Fig. 8b). Moreover, release was attenuated in the absence of extracellular Ca^2+^, suggesting that ATP is co-released with serotonin via secretory vesicles (Fig. 5d).
To ask whether released nucleotide contributes to sensory nerve fiber activation, we first investigated the sensitivity of mucosal afferents to ATP and serotonin. We traced and dissociated mucosa-innervating vagal neurons from the small intestine and performed single-cell Ca^2+^ imaging (Extended Data Fig. 8c)^41^. Remarkably, all traced mucosal vagal neurons responded to ATP, and a majority also responded to the 5-HT3 receptor-selective agonist, mCPBG (Extended Data Fig. 8d–e). Consistent with these results, single-cell real-time PCR from these neurons revealed expression of both P2X and 5-HT3 receptors (Extended Data Fig. 4c), suggesting that villus EC-derived ATP and serotonin collaborate to activate mucosal vagal afferents. Moreover, most of these neurons did not respond to AITC or H2O2, or express TRPA1 or TRPM2 channels, highlighting the role of EC cells as primary specialized sensors for electrophiles and oxidative stress in the small intestine that couple to mucosal afferents (Extended Data Fig. 4c and 8d–e).
To directly assess the contribution of EC-derived serotonin and ATP to mucosal afferent activation, we then examined nerve fiber activity in Tph1-CreER;lsl-ChR2 mice that express light-activated ion channels specifically in crypt and villus EC cells (Extended Data Fig. 8f–g). Using “flat sheet” ex vivo afferent recording preparations, compound action potentials were measured from the jejunal mesenteric nerve bundle while stimulating EC cells with 470 nm light from the luminal side (Fig. 5e). We observed that 6.3% of nerve fibers became active and 16.6% increased their firing rate after light stimulation. In contrast, preparations from control lsl-ChR2 mice showed no changes in the firing rate after light stimulation, demonstrating that ChR2-expressing EC cells signal to mucosal nerve fibers upon light activation (Fig 5f). We then repeated the recordings in the presence of either the 5-HT3 receptor antagonist alosetron, the P2X receptor antagonist PPADS, or a combination of both (Fig. 5f and Extended Data Fig. 8h). Alosetron alone completely diminished light-activated responses and reduced the percentage of fibers with increased firing to 7.4%. PPADS alone significantly diminished light-evoked spiking, with only 1.5% of fibers showing activation, but had minor effects on fibers with increased firing. Furthermore, the combination of both antagonists blocked almost all responses (Fig. 5f and Extended Data Fig. 8h). In summary, EC cells use both serotonin and ATP to signal to the mucosa-innervating vagal afferents, with the ATP input likely coming from villus EC cells.
Finally, we asked whether specific activation of crypt or villus EC cells leads to nerve fiber activity in a 5-HT3 and/or P2X-dependent manner. We used Nav1.8-Cre;lsl-ChR2 mice, which express ChR2 in Nav1.8-positive sensory neurons, allowing us to optogenetically activate the sensory nerve endings in the jejunum. This enabled us to test if EC cell activation sensitizes the nerve endings to light stimulation^13^, presumably via serotonin or ATP. Using ex vivo jejunal preparations from these mice, we applied either AITC (a weak electrophile) or acrolein (a strong electrophile) and recorded nerve fiber activity (Extended Data Fig. 9a). AITC application did not alter baseline activity or light sensitivity of mucosal afferents (Extended Data Fig. 9b–d) whereas acrolein increased both (Extended Data Fig. 9e–g). This differential response aligns with our ex vivo serotonin sensor imaging results demonstrating that crypt EC cells are responsive to strong but not weak electrophiles (Fig. 4b–c). Notably, the acrolein-induced effects were blocked by alosetron, but not PPADS, indicating that crypt EC cells primarily employ serotonin to activate mucosal afferents in response to strong electrophiles (Extended Data Fig. 9e–g).
To activate villus EC cells, we applied H2O2. This treatment increased both basal nerve activity and light sensitivity (Extended Data Fig. 10a–c). The effect was primarily PPADS-dependent, with alosetron showing only modest impact (Extended Data Fig. 10a–c), suggesting that villus EC cells predominantly use ATP, with a minor contribution from serotonin, to signal mucosal afferents. In summary, our biosensor experiments and nerve fiber recordings demonstrate that crypt and villus EC cells respond to distinct stressors (electrophiles vs. oxidative stress) and favor different transmitters (serotonin versus ATP) to communicate with mucosal afferents.
Discussion
Two modes of serotonin release
In this study, we develop and exploit biosensors to characterize spatial and temporal dynamics of neurotransmitter signaling within the gut architecture with the goal of understanding the relevance of these parameters to homeostatic and protective functions. Previously, high-performance liquid chromatography (HPLC) and enzyme-linked immunosorbent assay (ELISA) were employed to measure gut serotonin^42,43^. However, these methods primarily measure serotonin extracted from entire tissue samples and typically provide single timepoint measurements without spatial information. Voltage amperometry has also been used to detect gut serotonin^44,45^, but these measurements were primarily limited to luminal serotonin or concentrations on the villus surface. A more recent innovation, a tissue-like electrochemical biosensor, was designed to record serotonin in the intact intestine, but this method also predominantly measures luminal serotonin^46^. By comparison, our approach provides enhanced spatial, temporal, and quantitative analyses that greatly enhance our understanding of how specific stimuli trigger transmitter release from sensory cells within a complex anatomical structure – in this case, the crypt-villus architecture with an intact mucus layer. By directing expression of serotonin sensors to all intestinal epithelial cells, we can visualize the extent of diffusion of EC-derived serotonin within the epithelial layer following activation. Another noteworthy feature of gGRAB5HT3.0 is its independence from arrestin-mediated desensitization^23^, which allows for extended (> 30 minutes) imaging, as well as signal normalization upon addition of a saturating agonist concentration, thereby increasing the method’s quantitative robustness.
Using these tools, we found that EC cells within the crypt release serotonin in two ways, including low-level tonic and high-level evoked modes. One important question is what accounts for TRPA1-dependent tonic release? Our patch-clamp experiments show that basal TRPA1 activity can be observed in cultured EC cells and that single-channel events depolarize the membrane sufficiently to activate NaV channels and elicit action potentials. Thus, tonic transmitter release is likely a cell autonomous process driven by low-level TRPA1 activity that is either spontaneous or elicited by tonic low-level production of cellular electrophiles. TRPA1 is also a ‘receptor operated’ channel that can be activated downstream of signaling pathways that increase cytosolic calcium, representing another potential regulatory mechanism^47^. The identification of factors or conditions that support tonic channel activation may provide insights into endogenous or metabolic processes that regulate basal EC cell excitability. In any case, our findings suggest that crypt EC cells constitutively modulate ion secretion through 5-HT4 receptor-expressing cells located within the crypt. Consistent with this, it has been shown that a 5-HT4 antagonist decreases basal ion secretion in the small intestine^15^. Therefore, crypt EC cells may fine-tune ion secretion in response to changes in luminal or endogenous electrophiles that access crypts. Importantly, this low level of tonically released serotonin does not activate 5-HT3 receptors on sensory neurons, suggesting that crypt EC cells control gut secretion without conveying signals to intrinsic or extrinsic sensory networks, except perhaps in extreme pathological circumstances that degrade the protective mucus layer (see below).
It has been proposed that enteroendocrine L cells communicate with sensory neurons through synapse-like contacts^48^, an idea that we subsequently suggested might also apply to interactions between EC cells and primary afferents^4^. While a subset of EC cells and nerve fibers are closely juxtaposed, our current analysis indicates that the majority are too distant to establish bona fide synapses and likely communicate in a paracrine manner. Consistent with our measurements, recent studies demonstrate that most colonic and small intestinal EC cells are similarly too distant to form synapses with spinal and vagal afferents, respectively^49,50^. Their analysis of small intestinal EC cells and vagal afferents measured an average distance of 25.7 μm, which is greater than what we report here. This discrepancy likely reflects the fact that our analysis encompasses all nerve fibers, including intrinsic primary afferents and extrinsic vagal afferent neurons, while their tracing study labels only a subset of vagal afferents. In any case, our findings indicate that despite these distances, communication still occurs as released serotonin diffuses towards the closest nerve fiber, which uniformly express 5-HT3 receptors along their length. Our observations also suggest that when multiple EC cells within the same villus or adjacent crypts are activated simultaneously, they collaborate to stimulate the same nerve fibers. Thus, we propose that sparsely distributed EC cells integrate the information within the local environment in a parallel processing manner, converging their signals to produce a singular output to the nervous system to transmit signals within or outside the gut.
Mucus is a barrier for irritants
The mucus layer is thickest over the crypts, protecting stem cells that regenerate the intestinal epithelium. It therefore makes sense that TRPA1 channels are located preferentially in this protected zone, where they can serve as sentinels for highly reactive irritants such as acrolein, ingestion of which elicits frequent vomiting in dogs^51^. Acrolein, an environmental toxicant, is found in fried foods and alcoholic beverages and produced by catabolism of cyclophosphamide and related chemotherapeutic agents^52^. Acrolein may also be produced by microbial metabolism of glycerol in the gut, representing another pathological scenario in which EC cell activation initiates protective nocifensive signals^52^.
Weaker electrophiles found in foods such as mustard, garlic, and onions do not usually evoke an extreme nocifensive reflex, consistent with our finding that they do not penetrate the mucus layer to stimulate TRPA1 channels on crypt EC cells. In pathological states like colitis or bacterial infection, where the mucus layer is compromised^53^, these dietary electrophiles could potentially breach this protective barrier to activate crypt EC cells and promote nausea. This is consistent with observations that patients with inflammatory bowel disease (IBD) often avoid spicy foods, including mustard and garlic, suggesting heightened exposure of their crypt EC cells to luminal contents^54^. Also, differential sensitivity to endogenous or exogenous electrophiles will likely reflect individual variations in the status of protective mucosal barrier, which is influenced by factors such as microbiota and dietary habits^55^. Importantly, crypt EC cells tonically stimulate ion secretion, which presumably regulates the hydration status of intestinal contents near crypts, regardless of the mucus layer’s condition. This tonic regulation may also be tuned according to the permeability of electrophiles towards the crypts, but as noted above, nerve fibers should only be recruited when electrophiles penetrate the barrier in sufficient quantities to activate TRPA1 channels.
Detecting stress signals in villi
TRPA1 is activated by reactive oxygen species (ROS) such as H2O2 and 4-hydroxynonenal, making it a key physiologic sensor of oxidative stress and cellular redox state^56^. TRPM2 is also activated by H2O2 and our results suggest that these two TRP channel subtypes function as irritant/ROS sensors in crypts versus villi, respectively (Fig. 6). Unlike TRPA1 receptors in crypts, TRPM2 channels in villi are not as well shielded by a thick mucus layer and may therefore serve as ‘first responders’ to oxidative stress. Moreover, the ability of activated villus EC cells to simultaneously release serotonin and ATP augments their capacity to robustly stimulate mucosa-innervating vagal neurons, most or all of which are excited by both transmitters.
While EC cells have long been recognized primarily as serotonin-releasing cells, our data reveal a role for ATP as a transmitter, with an interesting segregation to a topologically specific population of EC cells in villi. How this specification is established remains an interesting question for future studies. Regardless, these findings necessitate a reevaluation of how purinergic receptors contributes to EC-mediated vagal and intrinsic primary afferent activation, and its subsequent impact on physiological responses, including nausea sensations and changes in gut motility. While ATP is widely appreciated as an activator of primary afferents^57^, purinergic receptors are also expressed by other cell types in the gut, such as enteric glia (P2X7), enterocytes (P2X7), and secretomotor neurons (P2Y1)^58^, and it is therefore possible that villus EC cells target these receptors to trigger additional stress responses.
ROS are produced in pathological situations such as IBD and chronic granulomatous disease^59^. Furthermore, chemotherapeutic drugs can rapidly generate ROS during the initial treatment stages, causing damage to the intestinal mucosa^59^. Our findings suggest that ROS produced in these individuals may activate TRPM2 in villus EC cells, triggering GI pain and nausea. With the loss of the protective mucus layer, TRPA1 channels in the crypt may then be recruited, further contributing to nociceptive and neurogenic inflammatory responses. Several TRPA1 antagonists have undergone clinical trials for managing inflammatory pain or airway hypersensitivity^60^; our work now highlights their potential use for treating gastrointestinal symptoms associated with overproduction of reactive irritants (of microbial or inflammatory origin) or reduction of the protective mucosal barrier. The same may pertain to potent and selective TRPM2 inhibitors, which are currently lacking.
Materials and Methods
Experimental model and subject details
Mice
All experimental procedures were conducted in accordance with guidelines approved by the Institutional Animal Care Committees at UCSF, SAHMRI Animal Ethics Committee, and Peking University, and aligned with the NIH Guide for the Care and Use of Laboratory Animals. Subjects were 8–16-week-old mice of both sexes, given ad libitum access to standard lab chow and sterile water. They were housed in a controlled environment under a 12-hour light/dark cycle. For serotonin sensor imaging, Villin-Cre (Jackson Laboratory. Strain no. 021504) and Insm1-GFP-Cre (gift from Corey Harwell. MMRRC ID: 36986) were crossed to the gGRAB5HT3.0-P2A-jRGECO1a reporter line. For nerve fiber recordings, Tph1-CreER^61^ (gifted from Dr. Juanita L. Merchant, University of Arizona College of Medicine) or Nav1.8-Cre (gifted from Dr. Wendy Imlach, Monash University, Australia. Jackson Laboratory, Strain no. 036564) were crossed to the channelrhodopsin 2 (ChR2, Ai32D) reporter line (Jackson Laboratory, Strain no. 012569). GCaMP imaging in organoids used Tac1-IRES-Cre (Jackson Laboratory, Strain no. 021877) crossed with GCaMP5g-IRES-tdTomato mice (gift from Dr. Lily Jan, UCSF. Jackson Laboratory, Strain no. 024477). Excitatory DREADD hM3Dq receptors were expressed in EC cells using Tac1-IRES-Cre;ePet1-Flp;FL-hM3Dq mice. Htr3a-GFP mice (MMRRC ID: 000273) visualized 5-HT3 expressing nerve fibers. Pirt1-Cre mice (gift from Dr. Xinzhong Dong, Johns Hopkins Medicine) were crossed with Ai14 tdTomato reporter mice (Jackson Laboratory, Strain no. 007914) for mucosal nerve fiber visualization.
Epithelial cell isolation and organoid culture
Adult male Tac1-IRES-Cre;GCaMP5g-IRES-tdTomato mice were used to generate intestinal organoids as previously reported^62^, specifically utilizing the upper jejunum to avoid ectopic Tac1-IRES-Cre expression in the lower intestine. Organoids were maintained and passaged every 6 days in organoid growth media (advanced Dulbecco’s modified Eagle’s medium/F12 supplemented with penicillin/streptomycin, 10 mM HEPES, Glutamax, B27 [Thermo Fisher Scientific], 1 mM N-acetylcysteine [Sigma], 50 ng/ml of mouse recombinant epidermal growth factor [Thermo Fisher Scientific], R-spondin1 [10% final volume] and 100 ng/ml of murine Noggin [Peprotech]). For villus organoid differentiation, day 4 organoids were treated with 5 μM IWP2 (Stemgent), 10 μM DAPT (Sigma), 1 μM PD0325901 (Sigma), and 20 ng/mL BMP4 (Peprotech) for 4 days.
Cell lines
HEK293FT cells (Thermo Fisher Scientific) cells were maintained in DMEM, 10% fetal calf serum, and 1% penicillin/streptomycin. The R-spondin 1 expressing HEK293T (Sigma) cells were maintained in DMEM, 20% fetal calf serum, 1% penicillin/streptomycin, and 125 μg/mL Zeocin (Thermo Fisher Scientific) at 37°C, 5% CO2. Zeocin was removed upon production of R-spondin 1 conditioned media. HEK293T cells (ATCC) were grown in DMEM, 10% fetal calf serum, and 1% penicillin/streptomycin at 37°C, 5% CO2 and transfected using Lipofectamine 3000 (Thermo Fisher Scientific) according to manufacturer’s protocol. For biosensor experiments, 200 ng pDisplay- gGRAB5HT2m-IRES-mCherryCAAX (Addgene, #208710), 200 ng pcDNA3–5-HT2A-P2A-GCaMP8m, or 200 ng pDisplay-gGRABATP1.0-IRES-mCherryCAAX (Addgene, #167582) was transfected to HEK293T cells in 24-well plates. For 5-HT3 biosensor experiment, 200 ng pcDNA3–5-HT3A and 20 ng pcDNA3-mApple were co-transfected to HEK293T cells in 24-well plates.
Method details
Generation of gGRAB5HT3.0-P2A-jRGECO1a reporter mouse
The gGRAB5HT3.0-P2A-jRGECO1a reporter mouse was generated with help of Biocytogen Pharmaceuticals Co., Ltd. (Beijing, China). In detail, the CAG-loxP-STOP-loxP-gGRAB5HT3.0-P2A-jRGECO1a-WPRE-bGH sequence was inserted to the Rosa26 locus of mouse embryonic stem (ES) cells using CRISPR/Cas9-mediated homology-directed repair (HDR). Successful targeting was confirmed with PCR. The genetically modified ES cells were injected into eight-cell stage embryos to generate chimeric mice. The chimeric mice were then mated with wild-type mice to obtain germline transmission of the targeted allele. The resulting transgenic mouse line stably expresses both gGRAB5HT3.0 and jRGECO1a under the CAG promoter at the Rosa26 locus upon excision of the floxed stop cassette by Cre recombinase.
Anti-CD3 antibody-induced inflammation model
Mice received a single intraperitoneal injection of 30 μg anti-CD3 antibody (Thermo Fisher Scientific) diluted to 200 μL with physiological saline. Mice were sacrificed for tissue collection after 12 hours.
Ex vivo serotonin sensor imaging
A 1 cm piece of the jejunum was isolated from an 8–16-week old Villin-Cre;gGRAB5HT3.0~-P2A-jRGECO1a or Insm1-GFP-Cre;gGRAB5HT3.0-P2A-jRGECO1a mouse. The isolated tissue was then immediately filleted-open along the mesentery, pinned down to a Sylgard-coated recording chamber, and imaged from the smooth muscle side to observe crypts and from the luminal side to observe villi. Imaging was performed with a Leica SP8 confocal microscope with an HC APO L 20x/1,00 W objective and LAS X software (Leica Microsystems, v3.5.5.19976). The tissue was bath perfused with bubbled room-temperature Krebs buffer (118 mM NaCl, 4.7 mM KCl, 1 mM MgCl2, 2 mM CaCl2, 1.2 mM KH2PO4, 25 mM NaHCO3, and 10 mM D-glucose) at a rate of 1 mL/min. All pharmacological reagents were diluted in Krebs buffer and bath perfused with simultaneous manual application. For recordings of StcE digested tissues, 50 μM StcE was supplied to Krebs buffer. At the end of each recording, gGRAB5HT3.0~ was fully activated by bath applied 20 μM serotonin. Acquired images were analyzed with Fiji software v2.14 (NIH). The regions of interest (ROIs) were drawn around individual crypts or villi and ΔF/F0 was calculated and normalized to serotonin-activated maximum signals. The area under the curve (A.U.C.) was calculated as (Normalized gGRAB5HT3.0 ΔF/F0) for the duration of 5 minutes during baseline or drug application. When measuring the baseline serotonin levels in villi, 20 μM RS 23597–190 was added at the end of recordings to fully quench the sensor.
Serotonin sensor imaging with isolated villi and crypts
Pieces of the jejunum were isolated from an 8–16-week-old Villin-Cre;gGRAB5HT3.0-P2A-jRGECO1a mice. The isolated tissue was then filleted-open along the mesentery. For recordings from isolated villi, villi were scraped off using glass coverslips and resuspended in a 50% Matrigel-Krebs buffer mixture. Matrigel domes (5 μL) were then formed on glass coverslips for imaging. The villi exposed to the surface of the Matrigel domes were identified under a microscope and used for imaging. For recordings from isolated crypts, after villi removal, the tissue was incubated in 10 mL of cold Dulbecco’s Phosphate-Buffered Saline (DPBS) with 30 mM EDTA for 20 minutes, followed by vigorous shaking for 30–60 seconds. Isolated crypts were filtered through 70 μm strainers and plated onto CellTak (Corning)-coated coverslips. Serotonin sensor imaging was performed with an upright microscope equipped with a Grasshopper 3 (FLIR) camera run by the Micro-Manager software (v2.0) and a Lambda LS light source (Sutter). Villi and crypts were maintained under a constant laminar flow of Ringer’s solution applied by a pressure-driven microperfusion system (SmartSquirt, Automate Scientific). All pharmacological reagents were delivered by local perfusion. Acquired images were analyzed with Fiji software (NIH). ROIs were drawn around individual EC cell and ΔF/F0~ was calculated.
Expression and purification of StcE
The pET28b-StcE-Δ35-NHis plasmid (gift from Dr. Carolyn Bertozzi, Stanford University) was transformed into E.coli BL21(DE3). Transformed E.coli cells were cultured in LB medium containing 50 μg/L kanamycin at 37°C for 4 hours. Isopropyl-thio-b-D-galactopyranoside (IPTG) was added to a final concentration of 0.3 mM to induce protein expression. Following an additional incubation at 20°C for 12 hours, cells were harvested by centrifugation and resuspended in purification buffer (500 mM NaCl and 20 mM HEPES-Na [pH 7.5]). Cell extracts were obtained by sonication followed by centrifugation at 36,000 g for 30 minutes. The supernatant was incubated with 2 mL Ni-NTA (Qiagen) for 1 hour at 4°C with gentle mixing. The resin was washed in batch with 5 column volumes of purification buffer, then loaded onto a column and further washed with 5 column volumes of purification buffer + 20 mM imidazole and 10 column volumes of purification buffer + 30 mM imidazole. The column was then eluted with purification buffer + 250 mM imidazole. To remove imidazole, the eluted protein was concentrated to 30 mM and then dialyzed against purification buffer overnight. Purified proteins are then stored at 4°C.
Mucus digestion with StcE
Purified StcE (30 mM) was diluted to 10 mM with double distilled H2O and 1 M HEPES-Na (pH 7.5) solution was added to a final concentration of 20 mM. This dilution was performed immediately before the experiment to avoid precipitation of StcE. The isolated jejunum (1 cm) was incubated in 10 mL of 10 mM StcE for 60 minutes at room temperature with gentle shaking. The StcE solution was exchanged after 30 minutes. Digested tissues were immediately mounted on a recording chamber for gGRAB5HT3.0~ imaging.
BODIPY-IA staining
A 10 mM StcE solution was prepared as described above. Sections of the jejunum (~1 cm) were moved to 10 mL of a solution of 166 mM NaCl + 20 mM HEPES-Na (pH 7.5) with or without 10 mM StcE, and incubated for 30 minutes at room temperature with gentle shaking. Digested tissues were immediately transferred to 10 μM BODIPY-FL-iodoacetamide (Thermo Fisher Scientific) in Dulbecco’s Phosphate-Buffered Saline (DPBS) and incubated for 3–15 minutes at room temperature with gentle shaking. Stained tissues were briefly rinsed with DPBS and fixed with 4% paraformaldehyde (PFA) for 3 hours at 4°C. Fixed tissues were dehydrated in 30% sucrose overnight at 4°C. The tissues were embedded in Tissue-Tek O.C.T. Compound (Sakura Finetek USA) and subsequently sectioned at a thickness of 10 μM on a Leica CM3050 S cryostat. The nuclei were stained with 4,6-diamidino-2-phenylindole (DAPI, 0.5 μg/ml, Thermo Fisher Scientific) and sections were mounted with ProLong Diamond Antifade Mountant (Thermo Fisher Scientific). Confocal images were captured on an inverted Nikon Ti microscope run using Micro Manager 2.0 Gamma^63^, equipped with a Zyla 4.2 CMOS camera (Andor), piezo XYZ stage (ASI), CSU-W1 Spinning Disk with Borealis upgrade (Yokogowa/Andor), Spectra-X (Lumencor), ILE 4 line Laser Launch (405/488/561/640 nm; Andor). Images were taken using a Plan Apo λ 20x / 0.75 using lasers 405, 488, and 561 nm and emission filters 447/60, 525/50, 607/36, for DAPI, GFP, and RFP, respectively. Maximum intensity projections were generated in Fiji v2.14.
Expression and purification of the 5-HT3 nanobody
The 5-HT3 nanobody (VHH15, gift from Dr. Hugues Nury, Institut de Biologie structurale) was engineered to include a C-terminal fusion with mCherry-His6 through an SSGSS linker and a gp64 signal peptide sequence was appended to the N-terminus, facilitating secretion into the insect cell culture media. The resultant plasmid, pFastBac-gp64-VHH15-mCherry-His6, was used to transfect Sf9 cells, generating P1 virus. Sf9 cells were then infected with amplified P2 virus for nanobody expression and harvested at 60 hours post-infection. The culture was centrifuged at 3,500 g for 15 minutes and the supernatant was filtered through a 0.22 μm filter. The filtered supernatant was adjusted to pH 7.5 using 1 M HEPES (pH 8.0), and divalent ions were replenished by adding 5 mM CaCl2 and 2 mM NiSO4. The supernatant was then incubated with 2 mL of Ni-NTA resin for 2 hours at 4°C. The resin was washed in batch with 5 column volumes of VHH15 buffer (500 mM NaCl and 50 mM Tris [pH 8.0]), then loaded onto a column and further washed with 5 column volumes of VHH15 buffer with 20 mM imidazole. The column was then eluted with 3 column volumes of elution buffer (125 mM NaCl, 250 mM imidazole, and 50 mM Tris [pH 7.4]). The elute was concentrated and loaded onto a Superdex 200 10/30 (GE Healthcare) gel filtration column in the buffer 10 mM HEPES (pH 7.5) and 100 mM NaCl. Fractions containing the peak were pooled and concentrated to 10 μM (0.39 mg/mL).
Histology and immunostaining
Immunofluorescence in the small and large intestine was performed using 10 μm cryosections. Blocking was performed with 5% w/v BSA (Sigma), 5% normal serum corresponding to secondary antibody species, and 0.3% Triton-X in PBS at room temperature for 30 minutes. Primary antibodies were incubated overnight at 4°C at the indicated dilutions. Antibodies used were against serotonin (1:5,000, Immunostar), Tuj1 (1:500, Abcam), Collagen IV (1:500 Abcam), cleaved caspase 3 (Asp175) (1:500, Cell Signaling Technology), Lysozyme (1:1000, Agilent), GFP (1:500, Abcam), and mCherry (1:500, Takara). Secondary antibodies from Invitrogen (Alexa Fluor 647 goat anti-rabbit, Alexa Fluor 568 goat anti-rabbit, and Alexa Fluor 488 goat anti-chicken) were incubated at 500 dilution for 2 hours at room temperature at 500 dilution. Z-stack images were taken with a Nikon CSU-W1 spinning disk confocal microscope as described above (UCSF Center for Advanced Light Microscopy). Maximum intensity projections were generated in Fiji v2.14.
Whole-mount staining of EC cells and mucosal nerve fibers
A piece of the jejunum was isolated from 8–16 week-old Pirt1-Cre;Ai14 mice. Pirt1-Cre;Ai14 mice were used because of the high expression level of tdTomato in mucosal nerve fibers. Isolated tissues were washed, filleted-open, and fixed with 4% PFA for 4 hours at 4°C. Blocking was performed with 5% w/v BSA (Sigma), 5% donkey serum, and 0.3% Triton-X in PBS at room temperature for 3 hours. Tissues were then incubated in a primary antibody solution (1:500 rabbit anti-serotonin, Immunostar, and 500 goat anti-collagen IV, Abcam) for two days at 4°C. Following primary incubation, tissues were washed three times in PBS with 0.2% Triton-X and incubated overnight in a secondary antibody solution (1:500 Alexa Fluor 488 donkey anti-rabbit and Alexa Fluor 647 donkey anti-goat). After overnight, tissues were washed three times in PBS with 0.2% Triton-X and mounted with ProLong Diamond Antifade Mountant. Z-stack images were taken with a Nikon CSU-W1 spinning disk confocal microscope as described above, using a Plan Apo VC 100x / 1.4 Oil objective (UCSF Center for Advanced Light Microscopy). Image deconvolution and 3D image reconstruction were conducted using Huygens (Scientific Volume Imaging) and Imaris (Oxford Instruments), respectively. The distances between the basolateral side of EC cells and the nearest dendrites were manually measured in Imaris.
Immunostaining of HEK293T cells with the 5-HT3 nanobody
HEK293T cells were plated on 4-well chamber slides (ibidi) and transfected with either pcDNA3–5-HT3A or the empty pcDNA3 plasmid. After overnight, cells were fixed with 4% PFA for 20 minutes at room temperature. Fixed cells were incubated with VHH15-mCherry in PBS + 0.1% Triton-X (1:20 dilution) for 1 hour at room temperature. After staining, cells were washed with PBS + 0.1% Triton-X three times and mounted with ProLong Diamond Antifade Mountant. Z-stack images were taken with a Nikon CSU-W1 spinning disk confocal microscope as described above (UCSF Center for Advanced Light Microscopy). Maximum intensity projections were generated in Fiji v2.14.
Nodose and dorsal root ganglia (DRG) isolation and immunostaining with the 5-HT3 nanobody
Nodose ganglia and DRGs were harvested from both male and female Htr3a-EGFP mice between 8–16 weeks of age. The dissected ganglia were fixed in 4% PFA for 3 hours at 4°C. Blocking was performed with 5% w/v BSA (Sigma), 5% goat serum, and 0.3% Triton-X in PBS at room temperature for 3 hours. The DRGs were then incubated with VHH15-mCherry (1:20 dilution) in PBS with 0.1% Triton-X overnight at 4°C. Post incubation, the ganglia were briefly washed in PBS and fixed again in 4% PFA for 30 minutes at 4°C. Following a PBS wash, ganglia were transferred to a primary antibody solution (1:500 chicken anti-GFP, Abcam, and 500 rabbit anti-mCherry, Takara) overnight. Ganglia were then washed three times in PBS with 0.2% Triton-X, and moved to a secondary antibody solution overnight (1:500 Alexa Fluor 488 goat anti-chicken and Alexa Fluor 568 goat anti-rabbit). The stained ganglia were then washed three times in PBS with 0.2% Triton-X and mounted with ProLong Diamond Antifade Mountant. Z-stack images were taken with a Nikon CSU-W1 spinning disk confocal microscope as described above (UCSF Center for Advanced Light Microscopy). Maximum intensity projections were generated in Fiji v2.14.
Immunostaining of the intestine with the 5-HT3 nanobody
The jejunum and proximal colon were harvested from male and female Htr3a-EGFP mice between 8–16 weeks of age. Tissues were filleted-open, washed, and incubated in the staining buffer containing VHH15-mCherry (1:30 dilution) and protease inhibitor cocktail (Roche) in Krebs buffer for 2 hours at room temperature. After staining, the tissues were washed three times with DPBS and fixed in 4% PFA for 1 hour at room temperature. Blocking was performed with 5% w/v BSA (Sigma), 5% goat serum, and 0.3% Triton-X in PBS at room temperature for 3 hours. Tissues were then incubated in a primary antibody solution (1:500 chicken anti-GFP, Abcam, and 500 rabbit anti-mCherry, Takara) for two days at 4°C. Following incubation, the tissues were washed three times in PBS with 0.2% Triton-X, and incubated in a secondary antibody solution (1:500 Alexa Fluor 488 goat anti-chicken and Alexa Fluor 568 goat anti-rabbit). After an overnight incubation, the tissues were washed three times in PBS with 0.2% Triton-X and mounted with ProLong Diamond Antifade Mountant. Z-stack images were taken with a Nikon CSU-W1 spinning disk confocal microscope as described above (UCSF Center for Advanced Light Microscopy). Maximum intensity projections were generated in Fiji v2.14.
In situ hybridization
5 or 10 μm cryosections were prepared as described above. Single-molecule RNA-FISH was performed using the RNAscope Multiplex Fluorescent Detection Kit v2 (Advanced Cell Diagnostics) according to manufacturer’s protocol. The following probes were used in this Mm-Tph1-C2 (318701-C2), Mm-Trpa1-C1 (512891), Mm-Trpm2-C1 (316831), Mm-Htr4-cust-C3 (408241-C3), and Mm-Olfm4-C1 (311831). Z-stack images were taken with a Nikon CSU-W1 spinning disk confocal microscope as described above (UCSF Center for Advanced Light Microscopy). Maximum intensity projections were generated in Fiji v2.14.
GCaMP imaging using intestinal organoids
Five days after passage, Tac1-IRES-Cre;GCaMP5g-IRES-tdTomato organoids were removed from Matrigel (Corning) and mechanically broken up with a 1000 μL pipette. The organoid fragments were seeded onto Cell-Tak (Corning)-coated coverslips and placed in a recording chamber containing Ringer’s solution (140 mM NaCl, 5 mM KCl, 2 mM CaCl2, 2 mM MgCl2, 10 mM D-glucose, and 10 mM HEPES-Na [pH 7.4]). EC cells were identified by tdTomato expression. GCaMP imaging was performed with an upright microscope equipped with a Grasshopper 3 (FLIR) camera and a Lambda LS light source (Sutter). Organoids were maintained under a constant laminar flow of Ringer’s solution applied by a pressure-driven microperfusion system (SmartSquirt, Automate Scientific). All pharmacological reagents were delivered by local perfusion. Acquired images were analyzed with Fiji software (NIH). ROIs were drawn around individual EC cell and ΔF/F0 was calculated.
Organoid swelling assay
Organoids were first passaged into organoid culture media lacking N-acetylcysteine to prevent the inhibition of TRPA1 channels. 24 hours post-passage, organoids were treated with either 1 μM serotonin, 10 μM RS 23597–190, or 5 μM A967079 in N-acetylcysteine-free organoid culture media. Images of the organoids were captured at 60 minutes post-serotonin treatment and 12 hours post-A967079 treatment. Cross-sectional areas of the organoids were subsequently measured using Fiji v2.14.
Biosensor experiments
HEK293T cells transiently transfected with biosensor plasmids (gGRAB5HT2m, 5-HT2A receptor and GCaMP8m, 5-HT3 channel, or gGRABATP1.0 sensor) were dissociated with trypsin and washed once with Ringer’s. The dissociated cells were plated on top of intestinal organoids. Individual HEK293T cells were carefully lifted from coverslips and positioned 5 μm from an EC cell using a glass pipette. For the 5-HT3 biosensor experiments, whole-cell configuration was achieved before lifting the cell. The membrane potential was held at −80 mV to measure inward 5-HT3 currents. For 5-HT2A and gGRABATP1.0 biosensor experiments, imaging was performed using an upright microscope equipped with a Grasshopper 3 camera (FLIR) run by the Micro-Manager software (v2.0) and a Lambda LS light source (Sutter Instrument). The entire area of each biosensor cell was used for the calculation of ΔF/F0 values. For gGRAB5HT2m biosensor experiments, imaging was performed on a Leica SP8 confocal microscope with LAS X software (Leica Microsystems, v3.5.5.19976). At the end of each recording, gGRAB5HT2m was fully activated with 500 μM serotonin. In these experiments, only the portion of each biosensor cell membrane within 5 μm of an EC cell was used for the calculation of ΔF/F0, which was then normalized to the maximum signals activated by serotonin. In all the biosensor experiments, the bath solution was static to prevent the washout of endogenously released serotonin, and pharmacological agents were applied manually with a 1000 μL pipette. All images were analyzed using Fiji v2.14.
Electrophysiology
Electrophysiological recordings were performed with an Axopatch 200B amplifier (Molecular Devices) connected to Digidata 1550B (Molecular Devices) connected to the pClamp software (v10.7), sampling at 10 kHz and filtering at 1 kHz. Membrane potentials were corrected for liquid junction potentials. Patch electrodes (3–6 MW) were pulled from borosilicate capillaries (BF-150–110-10, Sutter Instrument). The external solution for both EC cell and 5-HT3 channel recordings was Ringer’s solution. For TRPM2 recordings, the external solution contained 150 mM NaCl, 2 mM CaCl2, 1 mM MgCl2, 5 mM HEPES-Na (pH 7.4), and 10 mM D-glucose. The NMDG external solution was composed of 150 mM NMDG-Cl, 2 mM CaCl2, 1 mM MgCl2, 5 mM HEPES-Na (pH 7.4), and 10 mM D-glucose. The intracellular solution for EC cell recordings consisted of 140 mM K-aspartate, 13.5 mM NaCl, 1.6 mM MgCl2, 0.09 mM EGTA, 9 mM HEPES-K (pH 7.35), 14 mM phosphocreatine-tris, 4 mM MgATP, 0.3 mM Na2GTP. Intracellular solution for 5-HT3 recordings consisted of 140 mM K-gluconate, 5 mM NaCl, 1 mM MgCl2, 10 mM EGTA-K, and 10 mM HEPES-K (pH 7.2). For TRPM2 recordings, the intracellular solution included 150 mM NaCl, 5 mM HEPES-Na (pH 7.4), 5 mM EGTA-Na, 1 mM MgCl2, 5.1 mM CaCl2 (yielding a final free Ca^2+^ concentration of 100 μM), and 500 μM ADP-ribose (Sigma).
EC cell dissociation
EC cells were isolated from the upper half of the small intestine of 8–16-week-old Tac1-IRES-Cre;GCaMP5g-IRES-tdTomato mice. The tissue was cut into approximately 3 cm segments, incubated in 10 mL of cold DPBS with 30 mM EDTA and 1.5 mM DTT on ice for 20 minutes, then transferred to 6 mL of warm DPBS with 30 mM EDTA and incubated at 37°C for 8 minutes. To dissociate the epithelial layer, vigorous shaking was applied for 30–60 seconds. The dissociated epithelium was centrifuged and washed with DPBS containing 10% fetal bovine serum (FBS). The washed epithelium was digested in 10 mL of digestion buffer (HBSS with 0.3 mg/mL dispase II [Sigma] and 0.2 mg/mL DNaseI [Sigma]) at 37°C for 8 minutes, with vigorous shaking at 2-minute intervals. The cells were then washed once with HBSS containing 10% FBS and 0.2 mg/mL DNaseI, filtered through 70 μm and 40 μm strainers, and resuspended in DMEM supplemented with 10% FBS, B27, and 5 μM Y-27632 (Sigma). Cells were plated onto glass coverslips precoated with 5% Matrigel solution. Two days following dissociation, EC cells exhibiting the characteristic polygonal or cone-shaped morphology were predominantly surviving cells from the crypt regions. These crypt-originating EC cells were subsequently utilized for electrophysiological recordings and GCaMP imaging conducted 2–3 days post-dissociation.
Intestinal Isc measurement in mice
The ileum was excised under anesthesia and soaked in isoosmolar solution containing 300 mM mannitol and 10 μM indomethacin. Mucosa was stripped from serosa/muscle layers under a dissection microscope and mounted on Ussing chambers containing Ringer’s solution (120 mM NaCl, 5 mM KCl, 1 mM CaCl2, 1 mM MgCl2, 10 mM D-glucose, 5 mM HEPES-Na [pH 7.4], and 25 mM NaHCO3) on the basolateral side. For apical side, a similar solution was used except 120 mM NaCl was replaced with 60 mM NaCl and 60 mM sodium gluconate, and glucose was replaced with 10 mM mannitol. Compounds were added to both apical and basolateral bathing solutions unless specified otherwise. The solutions were aerated with 95% O2 /5% CO2 and maintained at 37°C during experiments. Short-circuit current (Isc) was measured using an EVC4000 multichannel voltage clamp (World Precision Instruments, Sarasota, FL, USA) connected to the pClamp software (v10.7) via Ag/AgCl electrodes and 3 M KCl agar bridges as previously described^64^.
Retrograde tracing of mucosal afferents from the proximal small intestine
Adult male mice of C57BL/6 background (Jackson Laboratory) aged 16–20 weeks were used. Retrograde tracing using cholera toxin subunit B (CTB, 0.5%) directly conjugated to Alexa Fluor 488 (Invitrogen, Thermo Fisher Scientific, #C2284, Australia) was performed from the lumen of the proximal small intestine (jejunum) using a technique modified from Harrington et al^41^. A small aseptic abdominal incision was made in mice anesthetized with isoflurane (2–4% in oxygen). The proximal small intestine was located, and injections of 5 μL were made through the intestinal wall into the lumen at three sites covering a length of 5 cm. The tracer was expelled completely prior to the needle withdrawal back through the intestine wall, which was gently rubbed together using cotton tip applicators to distribute the tracer throughout the lumen. Injections were made with a 30-gauge needle (HAMC7803–07, point 4; Hamilton Company, Bio-Strategy, Campbellfield, Vic, Australia) attached to Hamilton 5 μL syringe (HAMC7634–01, 5 μL 700 series RN syringe; Hamilton Company, Bio-Strategy). The abdominal incision was then sutured closed, analgesic (buprenorphine, 0.1 mg/kg) and antibiotic (ampicillin, 50 mg/kg) administration were given subcutaneously as mice regained consciousness. Mice were then housed individually and closely monitored for 4 days prior to i) perfuse fixation and nodose/jugular ganglia (vagal ganglia, VG) and spinal dorsal root ganglia (DRG, T8–13) collection or vagal ganglia removal for downstream dissociation and ii) cell picking and single cell RT-PCR or iii) Ca^2+^ imaging studies.
CLARITY processing and imaging of sensory ganglia
CLARITY method, which removes lipids, rendering tissue transparent whilst preserving ultrastructure, to visualize CTB-labelled neurons in intact whole nodose and DRG^65–67^. Four days after retrograde tracing surgery mice were euthanised via Lethabarb overdose, (intraperitoneal administration) and underwent transcardial perfuse fixation as previously described^67^. Following removal of individual VG and DRG and post-fixation (24 hours in 4% PFA at 4°C), ganglia were placed in 4% PFA-hydrogel solution (4% acrylamide, 0.25% VA-044 in PBS, Sigma-Aldrich) at 4°C for 48 hours. Residual oxygen was then removed from samples, as oxygen inhibits hydrogel polymerization, using a standard vacuum pump and desiccation chamber attached to a nitrogen gas supply^65^. Samples were degassed for 20 minutes and exposed to nitrogen gas for 5 minutes before transferred to a 37°C oven until the hydrogel solution had uniformly polymerized (90 minutes). Samples were removed from the hydrogel and underwent passive clearing in 8% SDS/200mM boric acid solution (Sigma-Aldrich), at 37°C. Buffer was changed after 24 hours and again after 72 hours at which point samples were transparent. Ganglia were then placed individually (left and right) into wells filled with RapiClear 1.47 refractive index solution (# RC147001, SunJin Lab, Taiwan) of a 18-well glass slide for at least 4 hours prior to confocal microscopy. Fluorescence was visualized with a confocal laser scanning microscope (Leica TCS SP8X, Wetzlar, Germany). Images (1024 × 1024 pixels) were obtained with 20X oil immersion lenses and 495 nm-excitation and 503/538 nm emission detection settings. Ganglia were optically sectioned (10–15 μm) and projected images reconstructed for each ganglion (230–390 μm). Images were processed using Leica LAS Lite and Image J software. Labelled neurons were manually counted from digital photomicrographs of CLARITY-cleared ganglia using Image J Cell Counter tool. Data was collected from N = 4 male mice, N from 2 ganglia per spinal level (n = 6 NG and T8-T9 with 1 ganglion lost during tissue processing from N = 2 mice and n = 7 T10-T13 DRG with 1 ganglion lost during tissue processing from N = 1 mouse.
Calcium imaging of dissociated nodose ganglia neurons
Retrogradely traced nodose ganglion neurons were isolated from adult mice. Briefly, 4 days after retrograde tracing, mice were euthanized by CO2 inhalation and nodose ganglia were surgically removed and were digested with 4 mg/mL collagenase II (GIBCO, Life Technologies) plus 4 mg/mL dispase (GIBCO) for 30 min at 37°C, followed by 4 mg/mL collagenase II for 10 min at 37°C, similar to that described previously for DRG ^4^. Neurons were then mechanically dissociated into a single-cell suspension via trituration through fire-polished Pasteur pipettes. Neurons were resuspended in DMEM (GIBCO) containing 10% FCS (Invitrogen), 2 mM L-glutamine (GIBCO), 100 mM MEM non-essential amino acids (GIBCO), 100 mg/ml penicillin/streptomycin (Invitrogen) and 100 ng/ml NGF (Sigma). Neurons were spot-plated on coverslips coated with poly-D-lysine (800 mg/ml) and laminin (20 mg/ml) and maintained at 37°C in 5% CO2. After 24 h in culture, neurons were loaded with 2.5 μM Fura-2-AM (Thermo Fisher Scientific) and 0.02% (v/v) pluronic acid for 30 min at room temperature in Ringer’s solution ((NaCl 140 mM, KCl 5 mM, CaCl2 1.25 mM, MgCl2 1 mM, glucose 10 mM, HEPES 10 mM, pH 7.4). After a brief wash, coverslips were transferred to a recording chamber filled with Ringer’s solution at room temperature (about 22°C). Retrogradely traced nodose ganglion neurons were identified by the presence of the CTB-488 tracer and viability was verified by responses to 40 mM KCl. Fura-2-AM fluorescence was measured at 340 nm and 380 nm excitation, and 530 nm emission was measured using an Olympus IX71 microscope in conjunction with a Sutter Lambda 10–3 wavelength switcher and the Chroma filter set no. 49011 (ET480/40x (Ex), T510lpxrxt (BS), ET535/50m (Em)). Fluorescence images were obtained every 5 seconds, using a 4X objective with a monochrome CCD camera (Retiga ELECTRO). Images were taken at baseline and following administration of Adenosine 5’-triphosphate disodium salt (ATP, 10 μM, Sigma Merck), m-Chlorophenylbiguanide hydrochloride (mCPBG, 10 μM, Tocris), Tegaserod maleate (1 μM, Sigma Merck), Capsaicin (50 nM, Sigma Merck), Allyl isothiocyanate (AITC, 1 μM, Sigma Merck), H2O2 (200 μM, Sigma Merck) and KCl (40 mM). Fluorescence traces of cell bodies were extracted using Metafluor software (Molecular Devices). Regions of Interest (ROIs) were manually drawn around the cell bodies of neurons and their fluorescence traces were extracted as 340/380 ratio.
Single-cell RT-PCR on mucosal traced nodose ganglion neurons
Four days after the mucosal retrograde tracing procedure described above, nodose ganglia were removed and enzymatically dissociated. Neurons were allowed to settle for two hours before adding 2 mL of media and preparing the coverslips for single-cell picking. Traced cells were manually picked using a micromanipulator under a microscope equipped with an appropriate fluorescent filter. Cells were under a continuous slow flow of RNA/DNase free PBS to reduce potential contamination. After picking a traced cell, the glass capillary was broken into a tube containing 9 μL lysis buffer with 1 μL DNaseI (TaqMan^®^ Gene Expression Cells-to-CT^™^ Kit; Thermo Fisher Scientific). A bath control was taken and analyzed from every coverslip with other samples. The whole cell lysate was used for cDNA synthesis using SuperScript^™^ VILO^™^ Master Mix (Thermo Fisher Scientific) and diluted 5 for further PCR analysis. PCR was performed according to the manufacturer’s instructions using TaqMan^™^ Gene Expression Master Mix (Thermo Fisher Scientific) for 55 cycles. A target was defined to be present when a typical amplification curve was produced. Predesigned Taqman probes were purchased from Thermo Fisher Scientific (P2rx2, Mm00462952_m1; P2rx3, Mm00523699_m1; Htr3a, Mm00442874_m1; Htr3b, Mm00517424_m1; Htr4, Mm00434129_m1; Trpv1, Mm01246300_m1; Trpa1 Mm01227437_m1 and Trpm2, Mm00663098_m1).
Ex vivo mucosal recordings from sensory nerves innervating the small intestine Jejunum afferent
12–18-week male and female mice were euthanized by CO2 inhalation. A 2 cm piece of the jejunum with intact mesentery segments was removed, opened longitudinally, and pinned down (mucosal side up) in a specialized organ bath. The jejunum was perfused with a modified Krebs buffer (117.9 mM NaCl, 4.7 mM KCl, 25 mM NaHCO3~, 1.3 mM NaH2PO4, 1.2 mM MgSO4, 2.5 mM CaCl2, 11.1 mM D-glucose), bubbled with carbogen (95% O2, 5% CO2) at 34°C. Krebs buffer also contained 1 μM nifedipine (to suppress smooth muscle activity) and 3μM indomethacin (to suppress potential inhibitory actions of endogenous prostaglandins). The free end of the mesentery’s segment was extended into an adjacent recordings compartment, which was subsequently filled with paraffin oil. A whole nerve bundle supplying the jejunal segment was then located within the mesentery, carefully cleaned away, and placed on a platinum recording electrode. Action potentials (AP) generated within the jejunum traveled across the nerve fibers, through the recording electrode, into a differential amplifier, filtered, sampled (20 kHz) using a 1401 interface (CED, Cambridge, UK) run by Spike 2 software (V.5.18), and stored on a PC for off-line analysis.
Optogenetic stimulation of mucosal
Nav1.8-Cre;lsl-ChR2 mice which express channelrhodopsin (ChR2) in sensory afferents but not EC cells were used^4,68^. This transgenic line allowed us to optogenetically activate mucosal afferents in an EC-independent and mechanically-independent manner, as previously described^13^. Using the jejunum ex vivo afferent preparation described above, we recorded the action potentials generated by stimulating a small section of the jejunum (3 mm^2^) with continuous exposures of increasing light (470 nm) intensities, ranging from 0.082–7 mW. This range of light intensities allowed us to determine the light activation threshold of afferents, and their stimulus response profiles to graded light stimuli. We selectively identified mucosal afferents based on their lower optogenetic (and mechanical) stimulation thresholds^13,69^. Each light exposure was applied for 2 sec with a 10 sec interval between exposures. Light was delivered using a High Power Fiber-Coupled LED Light Source (model BLS-FCS-0470–10) and Multimode Fiber Patchcords (Numerical 0.39 NA, Core 400 μm. Catalog # FPC-0400–39-025MA-BP, Mightex, Pleasanton, CA 94566, US). In selected experiments, we perfused the section of the jejunum receiving the light stimuli with the TRPM2 agonist H2O2~ (200 μM), or the TRPA1 agonists AITC (100 μM) or acrolein (100 μM). These agonists were perfused alone or in the presence of the 5-HT3 antagonist alosetron (10 μM), or the P2X antagonist PPADS (10 μM). In selected experiments, we perfused the area receiving the light stimuli with 1 μM of the 5-HT4 antagonist RS 23597–190. A gravity driven perfusion system slowly delivered the perfusion compounds. 5 minutes after perfusion of these compounds, we repeated the illumination protocol (still under continuous perfusion of the compounds) at the increasing light intensities described above. Action potentials generated throughout the recordings were analyzed off-line using the Spike 2 wavemark function and discriminated as single units based on a distinguishable waveform, amplitude, and duration (CED, Cambridge, UK). Data is expressed as i) afferent’s activity induced by compound tested, or by individual light stimulus (AP/sec), or ii) light intensity threshold for action potential activation (mW/mm^2^). The letter ‘n’ represents the number of afferents, whilst ‘N’ represents the number of animals for each study. Data was analyzed to determine if they were distributed normally using D’Agostino & Pearson or Anderson-Darling tests, with subsequent analysis using 2-way ANOVA with Šídák’s multiple comparisons test (for more than 2 groups). For comparison between 2 groups, we used i) Wilcoxon matched-pairs signed rank test, or ii) two-sided paired t test, or iii) two-sided unpaired t test. A P value of ≤ 0.05 was considered statistically significant.
Action potential recording induced by optogenetic stimulation of EC
For this we used a transgenic mouse line that expresses the light sensitive protein ChR2 in EC cells (Tph1-CreER;lsl-ChR2) and their control counterparts line lsl-ChR2. This allowed us to optogenetically activate EC cells in an agonist-independent manner. Using the jejunum ex vivo afferent prep described above, we recorded the spontaneous nerve activity (baseline firing rate) of SI afferents innervating the jejunum. Following 5 min baseline recordings, a section of the jejunum of ~3 mm^2^ was illuminated with a light wavelength of 470 nm, at 100% intensity (equivalent to 7 mW/mm^2^), using a BioLED control module (model BLS-PL04-US) coupled with a High Power Fiber-Coupled LED Light Source (model BLS-FCS-0470–10) and Multimode Fiber Patchcords (Numerical 0.39 NA, Core 400 μm. Catalog # FPC-0400–39-025MA-BP, Mightex, Pleasanton, CA 94566, US). We first applied pulses of light (3 × 20 ms pulses at 5 Hz) and then followed by continuous illumination (3 × 20 sec continuous light). On selected experiments, we perfused the section of the jejunum receiving the light stimuli with either alosetron (10 μM), or PPADS (10 μM), or the combination of both alosetron plus PPADS (10 μM each). Perfusion of these inhibitors were applied by a gravity fed perfusion system, 5 minutes prior to light stimulation. Afferents sensitive to light are defined afferents in which firing rate (AP/sec) after light is more than double that of the firing rate (AP/sec) at baseline. To quantitate the effect of the above-mentioned inhibitors on light-induced activation, we normalized the firing rate (AP/sec) induced by light as a percentage of the firing rate (AP/sec) at baseline. The letter ‘n’ represents the number of afferents, whilst ‘N’ represents the number of animals for each study. Significant differences in the data were determined using Chi-Squared test, and t-test. A P value of < 0.05 was considered statistically significant.
Analysis of single-cell RNA sequencing datasets
For mouse gut epithelial cells, we downloaded the normalized scRNA-seq data from the GEO database GSE224223^20^, where both epithelial and EEC subtypes are well annotated. For mouse ENS, we downloaded the P21 mouse scRNA-seq data from the GEO database GSM4504450^29^, and further performed normalization and annotation based on the original publication. All expression profiling were performed using Seurat 5.0.1. (Satjia Lab, https://satijalab.org/seurat/) on R 4.4.1.
Statistical analysis
Data were analyzed with Prism (Graphpad) and n represents the number of cells, crypts, villi, or independent experiments. Data were considered significant if P ≤ 0.05 using paired or unpaired two-sided Welch’s t-test, Mann-Whitney U test, one-way ANOVA, two-way ANOVA, Kruskal-Wallis test, Wilcoxon matched-pairs signed rank test, and Chi-square test. Statistical parameters are described in figure legends. All significance tests were justified considering the experimental design and we assumed normal distribution and variance, as is common for similar experiments. Sample sizes were chosen based on the number of independent experiments required for statistical significance and technical feasibility.
Extended Data
