Specialized Mechanosensory Epithelial Cells in Mouse Gut Intrinsic Tactile Sensitivity

Abstract

BACKGROUND AND AIMS:

The gastrointestinal (GI) tract extracts nutrients from ingested meals while protecting the organism from infectious agents frequently present in meals. Consequently, most animals conduct the entire digestive process within the GI tract while keeping the luminal contents entirely outside the body, separated by the tightly sealed GI epithelium. Therefore, like the skin and oral cavity, the GI tract must sense the chemical and physical properties of the luminal contents to optimize digestion. Specialized sensory enteroendocrine cells (EECs) in GI epithelium interact intimately with luminal contents. A subpopulation of EECs express the mechanically gated ion channel Piezo2 and are developmentally and functionally like the skin’s touch sensor, that is, the Merkel cell. We hypothesized that Piezo2+ EECs endow the gut with intrinsic tactile sensitivity.

METHODS:

We generated transgenic mouse models with optogenetic sensors in EECs and Piezo2 conditional knockouts. We used a range of reference standard and novel techniques from single cells to living animals, including single-cell RNA sequencing and opto-electrophysiology, opto-organ baths with luminal shear forces, and in vivo studies that assayed GI transit while manipulating the physical properties of luminal contents.

RESULTS:

Piezo2+ EECs have transcriptomic features of synaptically connected mechanosensory epithelial cells. EEC activation by optogenetics and forces led to Piezo2-dependent alterations in colonic propagating contractions driven by intrinsic circuitry, with Piezo2+ EECs detecting the small luminal forces and physical properties of the luminal contents to regulate transit times in the small and large bowel.

CONCLUSIONS:

The GI tract has intrinsic tactile sensitivity that depends on Piezo2+ EECs and allows it to detect the luminal forces and physical properties of luminal contents to modulate physiology.

The somatosensory senses of touch and taste have evolved to critically assess the physical properties of meals, frequently guides ingestion.¹ The skin, tongue, and oral cavity use specialized epithelial mechanosensory cells and sensory structures to determine a meal’s physical composition, including hardness, texture, and viscosity.² Although ingestion takes minutes, digestion takes hours, so most interactions with food products occur post-prandially. The gut keeps the digestive process outside the body and has developed a remarkable ability to determine the physical properties of the luminal contents,³ such as size and shape of luminal particles, by modulating motility⁴ and sensation.³ Consequently, diets that manipulate physical properties, such as with insoluble fiber, form a bedrock of treatments in diseases ranging from gastrointestinal (GI) sensory and motility disorders,^3,5 to metabolic disorders including obesity⁶ and diabetes,⁷ to developmental disorders like autism.⁸

To physically interact with luminal contents effectively, the gut developed a broad range of mechanosensory circuits composed of diverse mechanosensory cells.^9,10 The gut epithelium, which comes in direct contact with luminal contents, harbors specialized sensory enteroendocrine cells (EECs) that are developmentally and functionally similar to the specialized epithelial cells in the oral cavity and skin that contribute to the sensations of taste¹¹ and touch.¹² The EECs sense an impressive range of luminal stimuli—chemicals, such as nutrients and bacterial metabolites¹³; spices and odorants¹⁴; and forces.^15,16 The striking similarities between the Merkel cells in the skin and a subset of EECs led to a hypothesis of their being of common ancestry.¹⁷ We recently found a population of mechanically sensitive EECs in the small bowel that, like Merkel cells, contain the mechanically gated ion channel Piezo2.^15,16 Merkel cells and mechanosensing EECs have common developmental pathways^18,19; both detect mechanical forces through the mechanosensitive ion channel Piezo2,^16,20 are electrically excitable,²¹ and release the neurotransmitter serotonin (5-HT)^22,23 to initiate local and systemic responses.²³

However, although the role of Merkel cells is well understood, mechanosensitive EEC roles in GI functions remain unresolved.^24,25 Removing mucosal serotonin leaves mechanically induced contractions mostly intact,^26,27 but alters the pellets’ physical properties,²⁸ suggesting their role in luminal mechanosensing. We hypothesized that Piezo2 EECs are the gut’s tactile sensors. Using a range of techniques and models, we demonstrate that the Piezo2 EECs contribute to the gut’s ability to sense forces and physical properties of luminal contents to regulate critical digestive functions.

Methods

Full methods are available in the Supplementary Material.

Results

Piezo2⁺ Epithelial Cells Are Enteroendocrine Cells Involved in Neuroepithelial Communication

We were motivated by the recent discoveries that mechanosensitive epithelial cells in the skin and the gut expressed the mechanosensitive ion channel Piezo2.^16,20,29 We wanted to determine the populations of Piezo2+ gut mucosal cells. We generated Piezo2^tomato mice (Piezo2^cre/+:: Rosa26^{tm9(CAG-tdTomato)Hze/+}), in which Piezo2-expressing cells were lineage-traced (Figure 1A). We verified the model by taking a colon cross-section where we saw tomato⁺ cells present in the tunica muscularis, intracrypt, and sporadic epithelial cells (Figure 1B). From colon mucosa, we sorted 1,522,111 ± 142,606 single cells and found that 0.5% ± 0.2% were tomato+ (n = 3, Figure 1C and Supplementary Figure 1). We performed single-cell RNA sequencing on tomato+ cells and found several populations (Figure 1D), but only 3 were actively expressing Piezo2 (Figure 1E), suggesting that other populations expressed Piezo2 transiently, potentially during development.^30,31 Based on the clusters’ enriched genes, we classified the Piezo2+ populations as innate immune, lymph endothelium, and EECs (Figure 1E and F).

Figure 1.

Transcriptional profile of the epithelial Piezo2 enteroendocrine cell subpopulation. (A) Strategy to enrich Piezo2+ cells for single-cell RNA sequencing (scRNAseq). Lineage-traced cells from Piezo2^tomato colon mucosa: tomato+ if ever expressed Piezo2 (red) and some continued to express Piezo2 (red/blue). Tomato+ cells were sorted by fluorescence-activated cell sorting (FACS) and examined by scRNAseq. (B) Representative epifluorescence micrograph of Piezo2^tomato colon showing tomato+ structures and cells: ^&myenteric plexus, ^#intracrypt, and *sporadic epithelial cells. Red line shows mucosal peeling plane. (C) FACS sorting of dissociated cells from Piezo2^tomato colonic mucosa showing tomato+ cell population (from total mean ± SEM, 1,522,111 ± 142,606 cells, tomato+ were 0.2%−0.9%, mean ± SEM, 0.5% ± 0.2%, n = 3). (D) t-distributed stochastic neighbor embedding (tSNE) plot of 10 tomato+ populations. (E) tSNE plot of tomato+ subpopulations showing 3 clusters that actively express Piezo2. (F) Heatmap showing top cluster defining genes and Piezo2 expression across clusters (bottom strip). (G) Epithelial Piezo2+ population expresses EEC-specific developmental factors, a range of chemo- and olfactory receptors signal transduction ion channels, signaling and synaptic molecules, active zone, neuroepithelial communication, and post-synaptic receptors. (H) Pseudotime plot of the Piezo2+ enterochromaffin cell population showing that from root state (cyan), Piezo2 cells segregate into the following 3 states: Piezo2^loTph1^hi, Piezo2^hiTph1^hi, and Piezo2^hiTph1^lo. (I) Comparing transcriptomes of the Piezo2^hi to Piezo2^lo populations suggests that Piezo2^hi is an electrically excitable synaptically connected EEC subpopulation. (J) Colonic organoids from Piezo2^tomato mice show tomato+ cells (magenta). Scale bars: 50 μm (top), 10 μm (bottom). (K) FACS of dissociated Piezo2^tomato colonic organoids. (I) Quantitative polymerase chain reaction of FAC sorted organoid-derived Piezo2^tomato epithelial cells showing they are EECs enriched in Chga, Tph1, and Piezo2. (M) Piezo2 EECs are a unique population of synaptically connected electrically active mechanosensory EECs. Piezo2 epithelial cells are defined by EEC transcription factors (NeuroD1, Insm1, and Nkx2-2) and enriched in receptors for mechanical forces (Piezo2), and metabolites (Olfr78, Ffar2 [short-chain fatty acids]), and signal transduction ion channels (Scn3a, Cacna1h, Cacna1a). Synaptic vesicles (Chga/b, Snap25, Syt7) contain neurotransmitters, especially serotonin (Tph1), and less so others (secretin [Sct], substance P [Tac1], Glp-1 [Gcg and Pcsk1], Pyy, and Somatostatin [Sst]). They are enriched in molecules that form a presynaptic active zone (Pclo, Rimbp2, and Rim2), but also post-synaptic receptors (glutamate [Grin3a]).

The only Piezo2+ epithelial population (Vil1⁺) was EECs (Figure 1D and Supplementary Figure 2), and it expressed Piezo2 most densely compared with all sequenced cells (Figure 1E). Piezo2+ EECs expressed well-described EEC transcription factors (Insm1, Neurod1, Nkx2.2)^32-34 and signaling molecules known to be synthesized and released by EECs (most 5-HT via Tph1, secretin Sct, substance P via Tac1, and some expressed somatostatin Sst, peptide YY Pyy, and GLP-1 via Pcsk1).^35,36 In addition to high expression of Piezo2, Piezo2+ EECs expressed other EEC receptors, described previously, for bacterial metabolites (Ffar2, Olfr78)^37,38 and hormones (Igf2r), as well as ion channels critical for signal transduction (Scn3a, Cacna1a, and Cacna1d)^13,21 (Figure 1G). Our findings not only support the emerging theme that EECs co-express a range of signaling molecules,^36,39 but also suggest that the variety of released signaling molecules in response to luminal forces may allow multifaceted control of a range of complex physiologic processes within the gut and systemically.²³

Piezo2+ EECs expressed markers of neuroepithelial communication, including molecules relevant for synapses and synapse formation, neuroepithelial communication, and even postsynaptic receptors (Figure 1G).^13,40 We proceeded to examine the heterogeneity of this population using pseudotime analysis⁴¹ and found that the Piezo2+ EEC population was divided into the following 2 subpopulations: Piezo2^hi and Piezo2^lo (Figure 1H). The Piezo2^hi population was highly enriched in Gene Ontology cellular components related to electrical excitability and synaptic connectivity (Figure 1I) and Piezo2^lo was not. When we compared the presynaptic machinery between Piezo2^hi EECs and Merkel cells,⁴² we found a significant overlap of >30% (DynaVenn, P < .05) (Supplementary Figure 3). We confirmed our findings of epithelial Piezo2+ EECs in organoids from the Piezo2^tomato mice (Figure 1J), where we found a tomato + epithelial population (Figure 1K and J). When enriched by fluorescence-activated cell sorting, Piezo2 population accounted for 1.1% of total cells (Figure 1K). This population was consistent with Piezo2 EECs through enrichment in both Piezo2 and classical EEC markers (Figure 1L). In all, the Piezo2+^hi EEC population have a genetic landscape to be involved in communication between specialized mechanosensory epithelial cells and sensory neurons, similar to Merkel cells and sensory neurons in the skin (Figure 1M).

Mechanosensitive Enteroendocrine Cells^ReaChR Require Piezo2 for 5-HT Release

EECs are proposed to communicate with intrinsic primary afferent neurons, which regulate motility and secretion, which are critical aspects of digestion²³ (Figure 2A). To selectively stimulate EECs, we integrated the red-shifted channelrhodopsin (ReaChR).⁴³ Then we found that Piezo2+ EECs densely and selectively express NeuroD1 (Supplementary Figure 2), which is a “late” transcription factor in EEC development.⁴⁴ Therefore, to express ReaChR specifically in EECs, we generated EEC^ReaChR (NeuroD1^Cre/+::Rosa26^{LSL-ReaChR-mCitrine/+}) mice. We found that EEC^ReaChR mice had a scattered population of ReaChR+ (Figure2A and B) cells in the GI epithelium (Figure 2A, B, E, and H and Supplementary Figure 4). Through colocalization with chromogranin A (Cga) immunolabeling (Figure 2A and C), a maker of EECs⁴⁵ (Figure 2A and D), we determined that ReaChR+ cells were EECs and via Piezo2 immunolabeling we found that 40% ± 3% were Piezo2+ (Figure 2H-J, n = 4). As we expected in the colon, where 5-HT+ enterochromaffin cells comprise the largest EEC population,⁴⁶ the majority of the EEC^ReaChR cells were 5-HT+ (Figure 2A, E-G). NeuroD1+ EECs are mechanosensitive Piezo2+ EECs (Figure 2A and J),^16,29 and indeed, EEC^ReaChR (Figure 2A and H) colocalized with Piezo2 immunofluorescence (Figure 2A and I).

Figure 2.

EEC^ReaChR cells in the GI tract are EECs that are predominately enterochromaffin cells that express a red light–gated ion channel ReaChR. (A) Model where Piezo2 and ReaChR are expressed by EECs, and activation of EECs by force or red-light releases 5-HT to enteric neurons that control GI physiologic functions. (B–J) Projection of confocal stack images of EEC^ReaChR mouse colon with transgenic ReaChR expression (magenta in panels B, E, and H) colocalizing with immunofluorescence of EEC marker CgA (green in panels C and D), 5-HT (green in panels F and G), and Piezo2 (green in panels I and J). Nuclei counterstained with 4′,6-diamidino-2-phenylindole (blue). Lower panels expanded from within white rectangles. Scale bars = 20 μm. (B) ReaChR expression (magenta), (C) CgA immunofluorescence (green) and (D) colocalization. (E) ReaChR expression (magenta), (F) 5-HT immunofluorescence (green) and (G) colocalization. (H) ReaChR expression (magenta), (I) Piezo2 immunofluorescence (green) and (J) colocalization.

Optogenetics allows selective stimulation of excitable cell populations.⁴⁷ EECs are electrically excitable with hyperpolarized resting potentials compared with other epithelial cells.^13,21 Therefore, we asked whether EEC^ReaChR cells can be activated by both optical⁴³ and mechanical stimuli to drive 5-HT release.¹⁶ We generated EEC^ReaChR colonoids, which contained a population of fluorescent cells that were morphologically EECs (Figure 3A) and made primary cultures for whole-cell voltage-clamp electrophysiology to examine optically induced currents (Figure 3B). We voltage-clamped the EEC^ReaChR cells at a holding potential of −70 mV and used 625 nm optical stimuli to find robust duration-dependent inward opto-currents only in the fluorescent EEC^ReaChR cells (Figure 3C and D) that had linear current–voltage relationships consistent with nonselective cation conductance (Figure 3E). The EEC opto-currents had kinetics consistent with previous reports,⁴³ and when heterologously expressed ReaChR in HEK-293 cells (Supplementary Figure 5).

Figure 3.

Optical and mechanical stimulation activate single EEC^ReaChR cells. (A) Projection of a confocal stack overlayed with DIC of live 3-dimensional colonic organoid derived from EEC^ReaChR mouse with ReaChR (red) channel and transmitted light (gray). (B) Experimental setup showing an organoid-derived primary EEC under whole-cell voltage clamp electrophysiology and stimulated with red light. (C) Representative optically induced whole-cell currents with voltage-clamped electrophysiology. Scale bars = ordinate, 50 pA, abscissa, 1 second. (D) Normalized current dose-duration curve of EEC^ReaChR cell. (E) Current–voltage relationship (I/V) of peak current in EEC^ReaChR cell. Scale bars = ordinate, 200 pA, abscissa, 600 milliseconds. (F) Experimental setup showing mechanical or optogenetic stimulation of EEC^ReaChR cell (red) leading to 5-HT release that is detected and reported by the 5-HT biosensor. (G) Confocal stack image showing EEC^ReaChR cell (red) co-cultured with 5-HT biosensor (5HT3R/R-GECO, green). Scale bar = 20 μm. (H) Representative 5-HT biosensor fluorescence (ΔF/F₀) traces after EEC^ReaChR activation by light (black), block by 5HT₃R inhibitor ondansetron (OND, 0.1 μM, red), and mechanical stimulation (blue). Stimulation period in the line below trace. Scale bars = ordinate, 0.5ΔF/F₀, abscissa, 30 seconds. (I) Average 5-HT biosensor responses to EEC^ReaChR light stimulation (black), with OND (red), and with mechanical stimulation (blue) (individual data points paired, and bars showing mean ± SEM, n = 9, paired t test). (J) Representative 5-HT biosensor fluorescence (ΔF/F₀) traces after EEC activation with light in the presence of Piezo2 blocker Gd³⁺ (blue) mechanical stimulation with Gd³⁺ (red) and mechanical stimulation in washed control bath solution (blue). Stimulation period in the line below trace. Scale bars = ordinate, 0.5ΔF/F₀, abscissa, 30 seconds. (K) Average 5-HT biosensor response to EEC^ReaChR red-light stimulation with Gd³⁺ (purple) mechanical stimulation with Gd³⁺ (green) and washout control mechanical stimulation (blue) (mean ± SEM, n = 15, paired t test). **P < .01, ****P < .0001.

We wondered whether the optical and mechanical stimulation of EEC^ReaChR cells induce 5-HT release, which is critical for neuroepithelial transmission.¹³ We constructed 5-HT biosensors from HEK-293 cells transfected with a genetically modified 5-HT–gated ion channel (5-HT₃R) and a red-fluorescent calcium indicator (R-GECO) (Figure 3F and G).¹⁶ The 5-HT biosensors alone did not respond to optical stimuli, but responded to 1 μM 5-HT (Supplementary Figure 5). In co-cultured organoid-derived EEC^ReaChR cells with 5-HT biosensors (Figure 3F and G), we found that in 67% of EEC^ReaChR cells, optical stimulation led to the activation of 5-HT biosensors (n = 9, Figure 3H and I), and ondansetron (0.1 μM), a 5HT₃R blocker, inhibited biosensors’ optical responses (Figure 3H and I). We wanted to test whether the optically sensitive EEC^ReaChR cells are mechanosensitive EECs. We stimulated EEC^ReaChR mechanically by direct membrane displacement and found that 67% of mechanically stimulated EEC^ReaChR cells released 5-HT (n = 9, Figure 3H and I). Next, we used Gd³⁺, a Piezo2 channel blocker,⁴⁸ which inhibited the mechanosensitive, but not optical, response (Figure 3J and K). These data showed that EEC^ReaChR are mechanosensitive EECs that rely on Piezo2 for 5-HT release, and optogenetic stimulation can bypass pharmacological Piezo2 block.

Piezo2 Enteroendocrine Cells Adjust Motility in Response to Small Luminal Forces

EEC activation is proposed to control propulsive motility.²³ However, EEC roles in GI motility have been disputed^24,25 because previous studies have not directly activated EECs.^26,28 We investigated EEC roles in regulating contractions using the EEC^ReaChR model with optical stimulation in the organ bath (Figure 4A) using optogenetic stimulation protocols optimized in single EECs (Figure 3). To ask whether EECs influence propagating contractions, we used a low-volume intraluminal stimulus in colonic organ bath to induce repetitive contractions (Figure 4A). Optogenetic EEC^ReaChR colon stimulation increased the frequency of propagating contractions (Figure 4B and C), but did not significantly alter contraction force, duration, or propagation (Supplementary Figure 6). The increases in frequency in response to light stimulation were inhibited by 5HT3R and 5HT4R blockade (Supplementary Figure 7).

Figure 4.

Optical stimulation of EEC^ReaChR increases colonic propagating contraction frequency. (A) Experimental setup with organ bath used to measure optically stimulated (λ) propagating contractions through oral (ΔT₁) and aboral (ΔT₂) tension transducers and aboral pressure (ΔP) sensor. (B) Representative traces (ΔT₁, ΔT₂, ΔP) showing baseline, optical stimulation (2Hz, 40% duty cycle, 10 seconds, 90 seconds start to start), and recovery contraction frequency in EEC^ReaChR colon. Gray dashed line at peak of each tension trace (from ΔT1 to ΔT2) follows contraction propagating along the tissue. Optical stimulation shown between T₁ and T₂ traces (red). (C) Average propagating contraction frequency during baseline (gray), optical stimulation (red), and recovery (blue) in EEC^WT and EEC^ReaChR experimental colons (mean ± SEM, EEC^WT n = 6, EEC^ReaChR n = 14, paired t test) (ΔT1 scale bars = ordinate, 0.5 mN, abscissa, 1 minute; ΔT2 scale bars = ordinate, 1 mN, abscissa, 1 minute; and ΔP scale bars = ordinate, 5 cmH₂O, abscissa, 1 minute). **P < .01, ***P < .001.

We wondered whether mechanosensitive EECs altered the contraction frequency in response to luminal mechanical stimuli (Figure 5A). To abolish EEC mechanosensitivity, we used a conditional Piezo2 knockout from EECs (Piezo2^CKO) using the gut epithelium-specific Vil1-Cre driver (Villin1^cre/+ ::Piezo2^f/f).¹⁶ We found no difference in resting colonic contraction frequency between Piezo2^CKO and Piezo2^WT mice (Figure 5B and D). We then asked whether EEC mechanical stimulation could change colonic contraction frequency. We used luminal shear stress, a physiologically relevant stimulus for mucosal contribution to colonic motility. We found that stimulation with small shear forces (10 mL/h) increased the contraction frequency in Piezo2^WT mice, but failed to influence contraction frequency in Piezo2^CKO mice (Figure 5B-E). In contrast, higher-shear forces that we predicted to affect both mucosa and tunica muscularis (25 mL/h) increased contraction frequency in both Piezo2^CKO and Piezo2^WT mice (Figure 5F). Although similar, the responses in Piezo2^CKO tissue, unlike in Piezo2^WT, were highly variable (Figure 5F and Supplementary Figure 8). These data suggest that although Piezo2-dependent EEC mechanosensitivity is not required for colonic motility, it is sufficient to regulate the contraction frequency, especially to small intraluminal mechanical stimuli.

Figure 5.

Mechanosensitive EEC activation by small luminal forces increases colonic contraction frequency. (A) Experimental setup to measure luminal shear (black arrow in the lumen) induced changes in colonic motility reported by a tension transducer (ΔT₁). (B) Contractions intervals, displayed as black ticks, in Piezo2^WT colon basal and intraluminal shear (small force: 10 mL/h and large force: 25 mL/h, for 10 seconds, 90 seconds start to start). (C) Contraction frequency during baseline, intraluminal, and recovery in Piezo2^WT colon (mean ± SEM, 10 mL/h, n = 12; 25 mL/h, n = 6, paired t test). (D) Contractions intervals, displayed as red ticks, in Piezo2^CKO colon at rest and with intraluminal shear flow (small force: 10 mL/h and large force: 25 mL/h, 10 seconds, 90 seconds start to start). (E) Contraction frequency during baseline, intraluminal flow, and recovery in Piezo2^CKO colon (mean ± SEM, 10 mL/h, n = 10; 25 mL/h n = 7, paired t test). (F) Percent increase contraction frequency of Piezo2^WT and Piezo2^CKO colon from baseline to shear (mean ± SEM, Piezo2^WT n = 6, Piezo2^CKO n = 7, 1-tailed t test). (G) Experimental organ bath set up to measure luminal flow (black arrow in lumen) and 625 nm red-light optogenetic (2 Hz, 40% duty cycle) induced changes in colonic motility through a tension transducer (ΔT1). (H) Contraction intervals of tension traces showing basal, luminal flow (10 mL/h) baseline, flow (10 mL/h), and optogenetic stimulation with flow (2 Hz, 40% duty cycle and 10 mL/h) frequency in the presence of luminal Gd³⁺ in EEC^WT (top) and EEC^ReaChR (bottom) colon. (I) Contraction frequency of EEC^WT and EEC^ReaChR colon during baseline, intraluminal flow, baseline with Gd³⁺, Gd³⁺ intraluminal flow, and Gd³⁺ flow with optogenetic stimulation (mean ± SEM, EEC^WT n = 6, EEC^ReaChR n = 7, paired t test). *P < .05, **P < .01, ***P < .001, ****P < .0001.

We then asked whether optical and mechanical stimulation in EEC^ReaChR colons produced similar responses. The organ bath setup was modified to introduce a fiber optic red light for optogenetic stimulation (Figure 5G). We found that 10 mL/h intraluminal flow increased colonic contraction frequency in both EEC^ReaChR and EEC^WT mice (Figure 5H). We then added luminal Gd³⁺ to inhibit Piezo2 channels and found that Piezo2 blockage did not affect propagating contractions without mechanical stimulation and shear stress (10 mL/h) responses were diminished. Finally, we combined optical and mechanical stimulation while blocking the epithelial Piezo2 channels and found that in EEC^ReaChR tissue the diminished response to shear stress can be rescued by optical stimulation (Figure 5I). These data showed that colonic mechanosensitive Piezo2+ EECs adjust contraction frequency in response to small luminal forces generated by the movement of contents.

Mechanosensitive Enteroendocrine Cells Detect Physical Properties of Luminal Contents to Regulate Gut Motility

Next, we asked whether Piezo2-dependent EEC mechanosensitivity influences GI motility in vivo. Considering our in vitro data showing that mechanosensitive EECs directly regulate colonic motility, we tested the hypothesis that Piezo2-dependent EEC mechanosensitivity regulates gut transit (Figure 6A). To determine whole-gut transit, we used a carmine red gavage protocol with remote monitoring of pellet production (Figure 6B).⁴⁹ Piezo2^CKO mice had slowed whole-gut transit by 29% compared with Piezo2^WT mice (time_total Piezo2^CKO 295 ± 15 minutes vs Piezo2^WT 211 ± 26 minutes; *P < .05) (Figure 6B).

Figure 6.

Mechanosensitive Piezo2 EECs are regulators of tactile sensing by the luminal gut. (A) In vivo assays to determine the role of Piezo2 EECs in the gut: whole-gut transit (red), gastric emptying (blue), small intestine (yellow), and colon (purple). (B) Whole-gut transit time for Piezo2^WT and Piezo2^CKO (mean ± SEM, n = 12, n = 28, unpaired t test). (C) Gastric emptying T_1/2 was not different between Piezo2^CKO and Piezo2^WT mice (mean ± SEM, n = 5, n = 7, unpaired t test). (D) Top row, small bowel transit in Piezo2^WT vs Piezo2^CKO for liquids. Bottom row, small bowel transit for 390 μm fluorescent beads (n = 8, n = 6, unpaired t test). Spectral analysis of fluorescence deconstructed by frequency, showing the dominant frequencies of the signa (scale bar = 0.1). (E) Piezo2^WT (left) and Piezo2^CKO (right) pellet length and width, with image of pellets comparing size (mean ± SEM, n = 40, n = 71, unpaired t test) (scale bar = 2.0 mm). (F) Distal colonic transit of Piezo2^WT and Piezo2^CKO mice in 1 mm, 2 mm, and 3 mm beads (mean ± SEM, unpaired t test). *P < .05, **P < .01.

We compared gastric emptying between Piezo2^CKO and Piezo2^WT mice using the validated ¹³C-octanoic acid breath test^50,51 and found no differences (Figure 6C and Supplementary Figure 9). We were not surprised because the rodent stomach expresses Vil1 sparsely.⁵² Then we tested small bowel transit using liquid rhodamine isothiocyanate dextran gavage and profiling the small bowel fluorescence distribution between Piezo2^CKO and Piezo2^WT.⁵³ We found no differences in the fluorescence distribution between the groups when using liquids (geometric center [GC], GC^CKO_liquid = 4.9 ± 0.43, GC^WT_liquid = 4.4 ± 0.38; P > .05) (Figure 6D). We asked whether small particulates would have a different distribution in the small bowel. We gavaged small (390 μm) fluorescent beads and found drastic changes compared with liquid only in Piezo2^WT, with contents being spread out through the entire length of the small bowel (Figure 6D, bottom black). Intriguingly, although GCs between Piezo2^WT and Piezo2^CKO were not different (GC^CKO_solid = 5.4 ± 0.33, GC^WT_solid = 5.6 ± 0.43; P > .05) (Figure 6D), the 2 models’ distribution was remarkably different (Figure 6D, right panels). Although Piezo2^WT distributed the beads across small bowel length, widening the spectral density peak, the Piezo2^CKO small bowel continued to treat the luminal contents like liquids, showing up as a single peak (Figure 6D, right panels).

We proceeded to examine the colon. We looked at the fecal pellets’ dimensions because of previous findings that Tph1 knockout altered fecal pellet dimensions, suggesting a mechanosensory dysfunction.²⁸ Interestingly, we found that pellets had a significantly larger diameter in Piezo2^CKO mice compared with Piezo2^WT (2.3 ± 0.005 mm vs 2.1 ± 0.005 mm; P < .05), but were of similar length (Piezo2^CKO 5.74 ± 0.017 mm vs Piezo2^WT 5.53 ± 0.027 mm; P > .05) (Figure 6E). These results suggested that Piezo2 EECs are critical regulators of pellet size. To test the hypothesis that pellet size altered colonic transit and contributed to slower whole-gut transit slowing in Piezo2^CKO compared with Piezo2^WT, we used colonic bead expulsion assay.⁵⁴ We found that when comparing colonic transit times between different-sized beads, there was no difference in the smallest bead size (1 mm: Piezo2^CKO 39.3 ± 1.9 seconds vs Piezo2^WT 42.3 ± 5.7 seconds) (Figure 6F). As the pellet size grew, Piezo2^CKO transit time was significantly delayed compared with Piezo2^WT (2 mm: Piezo2^CKO 47.4 ± 1.2 seconds vs Piezo2^WT 33.6 ± 0.86 seconds; P < .05; 3 mm: Piezo2^CKO 41.3 ± 1.4 seconds vs Piezo2^WT 26.9 ± 1.0 seconds; P < .05). These results showed that the Piezo2 EECs are a critical element of the gut’s intrinsic mechanosensory system that detects luminal forces and discriminates particle sizes to regulate gut motility.

Discussion

The Gut’s Epithelial Tactile Receptors

The central finding in this study is that a population of epithelial Piezo2+ EECs serve as gut touch receptors and endow the gut with an intrinsic tactile sensitivity that regulates GI tract function and transit. The gut, like skin, is in constant contact with the external world and, therefore, it must determine the physical makeup of luminal contents. We found that a subpopulation of Piezo2 EECs was enriched in mechanisms involved in forming and maintaining synaptic contacts. Based on the shared developmental transcription factors, neurotransmitters, and synaptic connectivity, Piezo2+ EECs join the ranks of other Piezo2+ specialized epithelial mechanosensing cells—Merkel cells in the skin,^20,42 hair cells in the ear,⁵⁵ and neuroepithelial bodies in the lung.⁵⁶ However, skin touch and hearing are well-defined senses with a clear function, but how might gut touch receptors work, and what do they do?

EECs are highly diverse in their expression of receptors, neurotransmitters, and hormones across the length of the gut and along the crypt–villus axis.^35,39 Like other EEC subpopulations, Piezo2+ EECs express G-protein–coupled receptors potentially to complement and tune the mechanosensory responses based on the state of luminal contents, like microbiome.^13,29 For example, EECs are implicated in diseases such as irritable bowel syndrome,⁵⁷ in which there are alterations in motility and visceral mechanosensory dysfunction⁵⁸ and our results raise a possibility that diet-related shifts in microbial metabolites may alter GI motility through alteration of mechanosensitive EECs.

The Piezo2+ EECs also expressed a range of genes that produce neurotransmitters critical for sensory transduction. Most were 5-HT–producing enterochromaffin cells through the expression of Tph1, but subpopulations also express neuropeptides, including substance P, secretin, PYY, and somatostatin. Some of these signaling molecules stimulate motility (5-HT)^23,59,60; others inhibit it (Pyy and Sst),⁶¹ while the roles of others in GI motility are still unclear and likely segment dependent (substance P, secretin). Consequently, the regional repertoire of Piezo2+ EECs may define their impact on physiology. While Piezo2 EECs are present along the entire gut length,^15,16,29 there are fewer Piezo2 EECs in the proximal small bowel, where the bulk of nutrient absorption occurs, than in the distal small bowel, where ileal brake relies on Pyy and Glp1 release to slow the escape of nutrients and solids from the absorptive area.¹⁵ Similarly, Piezo2 EECs were significantly denser in the distal colon and rectum than in the more proximal areas,²⁹ and may contribute to the regulation of transit of contents through the colon. Indeed, the combination of surface receptors and chemical coding by Piezo2 EECs may underlie some regional complexities in GI physiology.

Gut Touch Receptors Regulate Gut Motility

The roles of EECs in gut motility remain unclear.^24,25 Although EECs are epithelial, unlike other epithelial cell types, and like other neuroepithelial sensory cells,⁶² they have hyperpolarized resting potentials and are electrically excitable.²¹ Therefore, to get at the role of these cells in physiology, we made a novel mouse model where EECs were optogenetic (EEC^ReaChR), allowing us, for the first time, to definitively assay EEC roles in GI physiology. We found that EEC^ReaChR had robust dose–stimulus photocurrents, and optogenetic stimulation drove 5-HT release, as did mechanical stimulation in the same EECs. The ability to overcome Piezo2 block by optical stimulation not only confirmed the necessity of Piezo2 for mechanically induced 5-HT, but established a platform to study EEC contributions to physiology.

Therefore, we used optogenetic EEC^ReaChR and conditional gut epithelial Piezo2 knockout (Piezo2^CKO) models to determine the physiologic roles of Piezo2 EECs. Because luminal forces and physical makeup of contents tune EEC hormone release⁶³ and regulate motility,⁶⁴ we focused on this function. We found that the optical stimulation of EEC^ReaChR in the colon most consistently increased the frequency of propagating contractions. Our findings are consistent with a model where mechanosensitive EECs regulate propagating contractions via serotonin receptors,⁶⁰ but they are not necessary to initiate them.²⁶ In turn, colonic contraction frequency regulates gut transit and may contribute to bowel disturbances and abdominal pain in patients with irritable bowel syndrome⁶⁵ and slow transit constipation.

Gut Touch Senses Small Luminal Forces and Physical Properties of Contents

Small luminal forces increased colonic contraction frequency in a Piezo2 EEC-dependent fashion because direct optical stimulation could rescue epithelial Piezo2 blockade. However, larger forces increased contractile frequency independently of epithelial Piezo2, likely by engaging the mechanosensors that reside deeper in the gut wall—pacemaker cells⁶⁶ or enteric neurons.⁶⁷ Therefore, our study suggests that, similar to the layered extrinsic mechanosensory innervation in the skin⁶⁸ and gut,⁶⁹ there is an intrinsic multilayer mechanosensory circuitry built directly into the gut wall.

The skin and oral cavity sense food texture using tactile sensing. We wondered whether Piezo2 EECs adjusted gut motility in response to the physical makeup of luminal contents. Piezo2^CKO had a delay in whole-gut transit, suggesting that GI motor function depends on Piezo2 EECs. To explore regional changes in motility, we independently tested motor function in the stomach, small bowel, and colon. The small bowel is the longest segment of the GI tract and has a critical life-sustaining function of digesting meals and extracting nutrients. The dictum is that mechanical digestion takes place in the stomach,⁷⁰ and the small bowel is responsible for chemical digestion. Yet, the small bowel frequently contains particulates, and studies in humans showed that although transit times of solids and liquids through the small bowel are similar, the small bowel’s length ensures the appropriate distribution of particulates, presumably for mechanical digestion.^71,72 Bolus distribution was similar between Piezo2^WT and Piezo2^CKO. However, the addition of insoluble particulates led to dramatic redistribution across the entire small bowel, possibly by segmenting activity required to physically digest the remaining solids. The small bowel transit time between Piezo2^CKO and Piezo2^WT for particulates was not altered, supporting previous findings that transit times between solids and liquids in the human small bowel are similar.⁷¹ Remarkably, however, Piezo2^CKO led to a complete loss of the ability of the small bowel to redistribute the particulate contents. These results suggest that the gut tactile system regulates the distribution of small bowel contents and possibly segmenting activity in the small bowel.

We found that the colon relied on the gut touch system to respond to the physical sizes of contents, and loss of tactile sensing eliminated the size-dependent acceleration of colon transit. Interestingly, the epithelial 5-HT knockout led to a change in the physical size of fecal pellets,²⁸ which, based on the current findings, is due to Piezo2+ EECs. These findings imply that the diseases may disrupt the gut touch system, such as slow transit constipation, where 5-HT deficiency was described previously,^60,73 and slowed gastric emptying,⁶⁰ and therapy involves targeting 5-HT pathways.⁷⁴ Although constitutive knockout of epithelial 5-HT alone did not delay whole-gut transit in some studies,⁷⁵ we found delays in whole-gut transit in Piezo2^CKO mice compared with Piezo2^WT. We may reconcile this apparent discrepancy in our finding that Piezo2+ EECs release other neurotransmitters in addition to 5-HT, which may provide physiologic redundancy. Therefore, the most effective approach to targeting this system may be via upstream receptor blockade rather than regulation of signaling molecule release or downstream receptors.

Our findings suggest that the gut touch sensors likely collaborate with other mechanosensory mechanisms deeper in the gut wall, likely the enteric⁷⁶ and extrinsic nervous system.^9,10,69 Small particulates mainly engage gut touch, while larger particles may engage both mechanosensors and speed up transit. When we eliminated gut touch mechanosensors, small bowel and colon lost their ability to determine the physical makeup of luminal contents, and discriminate particulate size, respectively. Neuro-epithelial circuits form between EECs and extrinsic sensory neurons,⁷⁷ and possibly between EECs and intrinsic primary afferent neurons.⁹ Studies suggest that both intrinsic primary afferent neurons⁶⁷ and extrinsic sensory neurons⁶⁹ are mechanosensitive. An attractive hypothesis is that the gut touch system may be akin to somatosensory touch, where the epithelial Merkel cell forms a synapse with an Aβ neuron, and cooperative mechanosensitivity is required for touch sensing.^20,62 The wiring of the gut touch circuitry and how it communicates with other intrinsic and extrinsic mechanosensors requires further investigation.

The mechanisms of how the gut feels luminal forces and the physical nature of its contents remained unclear. We described a novel intrinsic gut tactile sensing mechanism that requires specialized epithelial mechanosensors akin to Merkel cells in the skin and oral cavity.

WHAT YOU NEED TO KNOW.

BACKGROUND AND CONTEXT

Enteroendocrine cells (EECs) are epithelial sensors of luminal nutrients, microbial metabolites, and mechanical forces. The roles of mechanically sensitive EECs that express mechanogated ion channel Piezo2 remain poorly understood.

NEW FINDINGS

Piezo2+ EECs may be synaptically connected epithelial sensory cells, and their activation by light (optogenetics) of forces leads to modulation of colonic contraction propagation frequency and an inability to detect small luminal forces and physical sizes of particulates, suggesting that the gastrointestinal (GI) tract has an intrinsic tactile sensitivity that we termed gut touch.

LIMITATIONS

The mechanisms of gut touch require further investigation, and these findings require translation to humans in health and disease.

IMPACT

GI epithelial mechanosensory EECs mediate gut touch, which contributes to the regulation of GI motility and therefore may play essential roles in functional and motility disorders.

Abbreviations used in this paper:

EEC

enteroendocrine cell

GC

geometric center

GI

gastrointestinal

ReaChR

red-shifted channelrhodopsin