RET Signaling Persists in the Adult Intestine and Stimulates Motility by Limiting PYY Release from Enteroendocrine Cells

Abstract

Background & Aims:

RET tyrosine kinase is necessary for enteric nervous system (ENS) development. Loss-of-function RET mutations cause Hirschsprung disease (HSCR), in which infants are born with aganglionic bowel. Despite surgical correction, HSCR patients often experience chronic defecatory dysfunction and enterocolitis, suggesting that RET is important after development. To test this hypothesis, we determined the location of postnatal RET and its significance in gastrointestinal (GI) motility.

Methods:

Ret^CFP/+ mice and human transcriptional profiling data were studied to identify the enteric neuronal and epithelial cells that express RET. To determine whether RET regulates gut motility in vivo, genetic, and pharmacologic approaches were used to disrupt RET in all RET-expressing cells, a subset of enteric neurons, or intestinal epithelial cells.

Results:

Distinct subsets of enteric neurons and enteroendocrine cells expressed RET in the adult intestine. RET disruption in the epithelium, rather than in enteric neurons, slowed GI motility selectively in male mice. RET kinase inhibition phenocopied this effect. Most RET⁺ epithelial cells were either enterochromaffin cells that release serotonin (5-HT) or L-cells that release PYY and GLP-1, both of which can alter motility. RET kinase inhibition exaggerated PYY and GLP-1 release in a nutrient-dependent manner without altering 5-HT secretion in mice and human organoids. PYY receptor blockade rescued dysmotility in mice lacking epithelial RET.

Conclusions:

RET signaling normally limits nutrient-dependent peptide release from L-cells and this activity is necessary for normal intestinal motility in male mice. These effects could contribute to dysmotility in HSCR, which predominantly affects males, and uncovers a mechanism that could be targeted to treat post-prandial GI dysfunction.

Graphical Abstract

Lay summary

Signaling by the RET growth factor receptor in intestinal epithelial cells is important for regulating gut hormone release and propulsive activity, offering a new therapeutic target for motility disorders.

Introduction

RET tyrosine kinase is a transmembrane protein that functions as the signaling co-receptor for several secreted ligands. RET is a proto-oncogene and activating mutations are linked to neoplasia. In contrast, inactivating RET mutations cause developmental defects, most notably Hirschsprung’s disease (HSCR). HSCR is a male-predominant disorder in which infants are born lacking enteric ganglia, typically in the distal colon. RET dysfunction is a major cause of both familial and sporadic forms of HSCR with more than 60% of patients found to have pathogenic alleles leading to diminished RET signaling¹. The neural crest derived progenitors that colonize the bowel require RET for normal migration, proliferation and differentiation into enteric neurons². When RET signaling is deficient, these progenitors fail to fully colonize the bowel, causing congenital aganglionosis and functional bowel obstruction, the cardinal features of this disorder³. The standard of care for HSCR is surgical resection of the aganglionic bowel segment. While surgery resolves the bowel obstruction, chronic defecatory dysfunction and enterocolitis often persist⁴. Deficient RET signaling in the non-resected, ganglionated bowel could play a role in these chronic complications; however, this possibility has not been well explored.

While RET has been studied extensively in fetal development, its functions in the postnatal intestine have received less attention. Identifying these functions is crucial, not just for understanding the chronic complications of HSCR, but also for addressing the prominent GI symptoms in the many settings in which RET signaling is disrupted. These include multiple endocrine neoplasia type 2 (MEN2), irritable bowel syndrome (IBS), and cancer chemotherapy with RET kinase inhibitors^5-7. In the postnatal mouse intestine, there are at least 3 cell types with important roles in homeostasis that are potentially vulnerable to disruptions in RET signaling: immune cells, epithelial cells, and neurons. In the immune system, RET signaling in group 3 innate lymphoid cells stimulates interleukin-22 secretion, which limits epithelial reactivity and vulnerability to inflammation⁸. In the epithelium, RET is expressed by isolated cells scattered throughout the intestine that colocalize with the enteroendocrine cell (EEC) marker chromogranin A (CHGA)^9,10. EECs are diverse cells that release a variety of signals in response to chemical and mechanical cues from the gut lumen. Many EEC-derived signals alter intestinal motility, including 5-HT. In vitro, RET ligands alter 5-HT secretion from colon organoids in a manner that is sensitive to RET kinase inhibition⁹. The significance of this signaling in vivo, and whether it plays a role in causing dysmotility in RET-deficient states like HSCR, however, are unclear.

Among the strongest candidates to mediate the effects of aberrant RET signaling on GI motility are enteric neurons, which are essential for normal gut motor function. While RET is necessary for the initial generation of neurons in the fetal ENS, it is rapidly downregulated in many of these neurons in both mice and humans^2,11. In humans, RET can be detected in subsets of neurons in the myenteric and submucosal plexus layers of the adult duodenum and colon^12,13. At least some of the RET⁺ colonic neurons contain transcripts for choline acetyltransferase (CHAT), indicating that they are cholinergic¹⁴. RET is also expressed in a subset of neurons in the adult mouse ENS¹⁵, but the identity of these neurons and whether RET serves any essential functions in the neurophysiology of the adult gut are undefined. Analogous to observations in EECs, the application of RET ligands to mouse enteric neurons in vitro alters neuropeptide secretion^16,17. Similarly, RET ligand administration to gut explants changes calcium transients in some neurons and alters tissue contractility¹⁸; however, whether these effects are direct and how they relate to motor functions in vivo remain unknown. To define better the role of RET signaling in the postnatal intestine, here we identify the major populations of RET-expressing neurons and EECs. Then, using conditional genetic and gut-restricted pharmacological approaches to disrupt RET signaling in one or more of these cell populations in vivo, we find that RET kinase activity in a subset of EECs limits peptide secretion to regulate small bowel motility.

Methods

Animal studies

Littermate controls and 12–16-week-old male and female mice were used for all experiments, except where noted. Studies were approved by the Institutional Animal Care and Use Committees at GSK, Columbia University Medical Center, or Boston Children’s Hospital and were conducted in accordance with the GSK Policy on the Care, Welfare and Treatment of Laboratory Animals. Measurements of gastrointestinal transit time (GITT), gastric emptying/small intestinal transit (GE/SITT), colonic migrating motor contractions (CMMCs), fecal pellet output (FPO), and fecal pellet composition were done as previously described^19,20. For fasting GITT, food was removed 6 hours prior to dye gavage. For 3-week-old mice, fecal pellets were checked starting 30 minutes after gavage of 100μL of 6% carmine red. To test effects of RET kinase inhibition, mice were administered 10mg/kg GSK408B or vehicle (0.5% hydroxypropylmethylcellulose) twice daily by orogastric gavage for 3.5-5.5 days. The final dose was administered 30 minutes prior to GI motility testing. To assess NPY2R signaling, 2μg/kg BIIE0246 (Tocris 1700) or vehicle (5% Tween-80 in 0.9% saline) was administered once via IP injection 30 minutes before GITT measurement. For fasting/re-feeding experiments, the final dose of GSK408B or vehicle was administered 30 minutes prior to collection of “fasting” plasma.

Quantitation of gene and protein expression

For neuronal quantitation, intestinal segments were immunostained and imaged as whole-mounts as described¹⁹. Ileal segments represent the final 1.5cm of the small intestine. Colon segments represent 2cm proximal to the anus. For epithelial quantitation, 12μm cryosections were immunostained. Antibodies are in Supplementary Table 1. For quantification of PYY/GLP-1 in mouse plasma or human organoid supernatants, custom Mesoscale U-plex plates were used per manufacturer’s instructions. For mouse transcript quantitation, the intestinal epithelium was mechanically isolated and collected into Trizol (Invitrogen). RT-qPCR was performed to measure Ret and Pyy relative to the housekeeping genes Rpl19 or Gapdh. Primer sequences are in Supplementary Table 2. Ret expression in sorted EECs, L-cells and intestinal epithelial cells from mice and humans was assessed from published data²¹ (GSE114853, GSE114913). For scRNA-SEQ analysis of Ret, Chat, and Nos1 in enteric neurons, published datasets of myenteric neurons from 6-8 week old mice²² (GSE153202) and humans²³ (SCP1038) were analyzed.

Statistical analyses

Analyses were performed using SPSS and visualized using GraphPad Prism version 9. Graphs show individual replicates, mean and standard error of the mean (SEM), except where noted. Two-tailed, unpaired t-tests were used to compare pairs of means. For 3 or more groups, ANOVA was used as detailed in Supplementary Methods.

Please see Supplementary Methods for detailed descriptions of mouse strains, antibodies, and experimental protocols.

Results

Specific subsets of neurons in the adult ENS express RET

RET expression is maintained in at least some enteric neurons after fetal development¹⁵; however, the identity of the neurons and the extent to which RET expression varies along the radial and longitudinal axes of the ENS is unknown. To address these knowledge gaps, we characterized Ret expression in the postnatal ENS using Ret^CFP/+ knockin mice, which enabled us to circumvent the limited efficacy of RET antibodies. In adult mice, Ret-expressing cells immunoreactive for the pan-neuronal marker ANNA-1 were abundant within the myenteric and submucosal plexuses (Figure 1A-C). In contrast, Ret expression was undetectable in intraganglionic and intramuscular glia marked by S100B immunoreactivity (Supplementary Figure 1). We quantified the proportions of Ret-expressing enteric neurons across 3 stages of the lifespan: pre-pubertal (4 weeks), mature adult (12-16 weeks) and aged adult (52 weeks). In the mature adult colon, 54.7% of myenteric neurons and 69.2% of submucosal neurons expressed Ret (n=8-9 mice/group). Age did not affect these proportions in either plexus (submucosal p=0.1274; myenteric p=0.1517); moreover, there were no differences in the proportions of Ret-expressing myenteric neurons between the colon and ileum (p=0.3317). These data indicate that Ret is expressed in a large subset of neurons in both plexuses of the postnatal mouse ENS and that the proportion remains relatively stable with age, with little variation along the length of the gut.

Figure 1. Ret is expressed in subsets of mature enteric neurons, including all NOS1⁺ nitrergic neurons and some CALB1⁺ cholinergic neurons.

(A-B) Ret expression in the myenteric (A) and submucosal plexus (B) visualized by immunostaining adult Ret^CFP/+ mouse colon for CFP and the pan-neuronal marker ANNA-1.

(C) The proportions of ANNA-1⁺ neurons that expressed Ret (CFP⁺) in ileums and colons from Ret^CFP/+ mice that were 4, 12-16, or 52 weeks of age in the myenteric (MP) and submucosal plexus (SMP) layers (n=5-9 mice).

(D) Virtually all NOS1⁺ myenteric neurons in adult Ret^CFP/+ mouse ileum expressed Ret.

(E) Proportions of NOS1⁺ myenteric neurons expressing Ret in adult ileum (Ile) and colon (Col).

(F) Proportions of Ret⁺ adult myenteric neurons that were NOS1⁺. *** p<0.001 by unpaired t-test

(G) A subset of CALB1⁺ myenteric neurons in an adult Ret^CFP/+ mouse colon expressed Ret.

(H) Proportions of CALB1⁺ myenteric neurons expressing Ret in the adult intestine.

(I) Proportions of Ret⁺ adult myenteric neurons that were CALB1⁺.

(J) UMAPs of scRNA-SEQ of myenteric neurons from adult mouse colons²² show Ret expression within a large subset of Chat⁺ neurons and the majority of Nos1⁺ neurons.

(K) UMAPs of scRNA-SEQ of myenteric neurons from adult human colons²³ show similar distributions of RET, CHAT and NOS1 expression across clusters.

Scale bars=50μm

To determine whether Ret expression in enteric neurons was stochastic or a fixed feature of specific subtypes, we characterized the Ret⁺ myenteric neurons in mature adult Ret^CFP/+ mice. Nitrergic and cholinergic neurons are considered mutually exclusive populations with developmentally distinct trajectories^11,24. Almost 100% of NOS1-immunoreactive neurons expressed Ret in the ileum and colon, suggesting that Ret expression is a fixed feature of nitrergic neurons (Figure 1D-E). NOS1⁺ neurons constituted 31.6% of all Ret⁺ neurons in the ileum and 51.4% in the colon (Figure 1F). In contrast, only a subset of myenteric neurons immunoreactive for Calbindin (CALB1), a calcium binding protein that marks many cholinergic neurons, expressed Ret (Figure 1G). In the ileum and colon, 74.7% and 53.9% of CALB1+ neurons expressed Ret, respectively (Figure 1H). CALB1⁺ neurons constituted a minority of Ret⁺ neurons, 29.6% in the ileum and 19.0% in the colon (Figure 1I). Analysis of Ret expression in published single cell transcriptional profiling data of myenteric neurons in 6-8-week-old mice was consistent with our observations that Ret expression is enriched in nitrergic neurons and subsets of cholinergic and other neurons (Figure 1J; Supplementary Figure 2). This expression pattern was similar in human adult colonic myenteric neurons (Figure 1K; Supplementary Figure 2), suggesting that RET signaling is likely to have conserved functions in the mammalian ENS.

Genetic depletion of RET in adult mice transiently slows GI motility

RET is required for ENS development and Ret-null mice do not survive beyond birth because of this and other developmental defects²⁵. Even mice that lack RET only in neural crest derivatives do not survive to adulthood due to intestinal aganglionosis²⁶. Thus, the functions of RET in the mature ENS have remained unclear. To determine these functions, we generated Ret^CreER/flox mice (hereafter RET^cKO) in which RET could be conditionally depleted in adult mice after ENS development. At baseline, RET^cKO mice are Ret haploinsufficient. Upon tamoxifen administration to induce Cre recombinase activity, cells that express Ret at the time of induction become Ret-null (Supplementary Figure 3A). To determine the effects of postnatal RET depletion, adult RET^cKO mice and controls were induced with tamoxifen and studied for 4 weeks. RET^cKO mice appeared well and had no weight loss (Supplementary Figure 3B). The overall neuronal density and the proportions of CALB1⁺ and NOS1⁺ neurons in the intestines of RET^cKO mice did not differ from those of littermates with intact Ret (Ret^flox/+ or Ret^flox/flox mice; Supplementary Figure 3C-D), indicating that RET signaling is not essential for the survival of mature enteric neurons. Previous work showed that conditional deletion of RET shortly after colonization of the bowel by ENS precursors caused colonic hypoganglionosis associated with cell death²⁷, suggesting that enteric neurons require RET for maintenance. To test further whether RET is necessary for enteric neuronal survival, we generated RET^cKO-GFP mice in which every Ret-expressing cell that loses functional RET upon induction of Cre activity starts expressing GFP (Figure 2A-B; Supplementary Figure 3E-F). Four weeks after tamoxifen induction in adult RET^cKO-GFP mice, 60.5% of neurons in the ileum and 42.0% in colon were GFP⁺, similar to the proportions in adult Ret^CFP/+ mice, indicating robust and Cre-dependent RET deletion. RET depletion in adult RET^cKO-GFP mice did not alter neuronal densities in the distal ileum or colon (Figure 2C). Although virtually all NOS1⁺ neurons became GFP⁺ in RET^cKO-GFP mice (Figure 2D), their densities were no different in the intestines of RET^cKO-GFP mice compared to Cre-negative littermate controls (Figure 2E). These observations further substantiate that RET signaling is not necessary for the maintenance of enteric neurons in the adult intestine.

Figure 2. Conditional deletion of RET in adult mice causes transient GI dysmotility but does not alter neuronal survival.

(A) The distal colon of a Ret^{CreER/flox-GFP} mouse isolated 4 weeks post-induction (RET^cKO-GFP) shows that a large proportion of myenteric neurons immunoreactive for the pan-neuronal marker HuC/D (grey) and virtually all NOS1⁺ neurons (red) express GFP (cyan), indicating that RET depletion does not alter neuronal survival. No GFP expression is evident in a Ret^+/flox-GFP littermate control (WT). Scale bar=50μm.

(B) Schematic of the Ret^flox-GFP allele in which cDNA for human RET (hRET) is knocked into Exon 1 of the mouse Ret locus, flanked by LoxP sites. Cre activity leads to simultaneous RET loss and induction of GFP expression.

(C) Postnatal RET depletion does not alter myenteric neuronal density in the distal ileum or colon. Open circles, males; closed circles, females.

(D) Virtually all NOS1⁺ myenteric neurons are GFP⁺ 4 weeks after RET depletion in RET^cKO-GFP mice.

(E) The proportions of NOS1⁺ myenteric neurons are unaffected by RET loss.

(F-G) RET depletion slows GI transit time (GITT) in RET^cKO mice, but not in RET^+/− haploinsufficient controls, at 2 weeks post-induction (F). GITT begins to normalize by 4 weeks (G). *p<0.05 by two-way repeated measures ANOVA.

To determine whether RET depletion affected the function rather than the survival of enteric neurons, we measured gastrointestinal transit times (GITT) in RET^cKO mice at 2- and 4-weeks post induction. To account for potential effects of Ret haploinsufficiency, we compared them to Ret^CreER/+ littermates. Although GITT in Ret^CreER/+ mice was not different from baseline at 2 or 4-weeks post-induction, GITT in RET^cKO mice was prolonged by 35% at 2-weeks (Figure 2F). This defect resolved by 4-weeks (Figure 2G), suggesting that RET depletion in adult mice causes transient dysmotility.

Some enteric neurons might be continually replaced in the adult ENS. To determine whether turnover of RET-expressing neurons explained the transient dysmotility in RET^cKO mice, we generated Ret^CreER/+Rosa26^TdTomato/+ mice in which Ret-expressing cells could be permanently labeled with the TdTomato reporter upon tamoxifen administration. One week later, robust TdTomato was evident in both plexuses of the ENS, immune cells in Peyer’s patches, and isolated epithelial cells (Supplementary Figure 4). At 3-4 weeks post-induction, TdTomato expression remained unchanged in the neuronal and immune compartments but was undetectable in the epithelium (Supplementary Figure 4). These observations suggest that RET-expressing epithelial cells turned over within 4 weeks, while most RET-expressing neurons did not. RET^cKO mice, by proxy, are likely to have evolving levels of RET signaling in the gut as RET-expressing cell types turn over at different rates. Normalization of dysmotility in RET^cKO mice by 4-weeks post-induction could thus reflect repopulation of the intestinal epithelium with cells containing intact Ret alleles or mobilization of compensatory mechanisms within RET-deficient neurons.

To establish a more stable model of RET deficiency in the mature ENS, we examined the effects of constitutively depleting RET in a population of differentiated enteric neurons. Given the uniform expression of RET among inhibitory motor neurons and their well-established role in GI motility, we generated mice lacking RET in this population. Inhibitory motor neurons are characterized by expression of VIP and NOS1, and most NOS1-expressing cells in the murine myenteric plexus express VIP^18,24,28. Because Nos1^Cre/+ mice, but not Vip^Cre/+ mice, exhibited Cre activity in the intestinal epithelium in addition to the ENS, we generated Vip^Cre/+Ret^flox/flox mice to determine the effects of depleting RET within inhibitory motor neurons (hereafter RET^VipKO). RET^VipKO mice were born at expected Mendelian ratios, appeared healthy, and grew similarly to Ret^flox/flox littermates (Supplementary Figure 5A-B). As we observed in RET^cKO mice, constitutive depletion of RET in inhibitory motor neurons did not grossly alter the appearance or proportions of these neurons (Supplementary Figure 5C-D), again indicating that RET signaling is not essential for their survival. RET^VipKO mice showed no deficits in GITT, 1-hour FPO, fecal water content or pellet mass, implying that RET signaling in this large population of motor neurons is not essential for gut motility or secretomotor function (Supplementary Figure 5E-H). The functional significance of RET expression in VIP neurons remains to be determined. Taken together, our observations indicate that RET depletion in adult mice causes transient GI dysmotility. This dysmotility was not replicated by constitutively deleting RET in a large population of enteric neurons that are important for motility, suggesting that RET expression in epithelial cells may be more consequential for propulsive activity.

RET signaling in the intestinal epithelium regulates motility selectively in male mice

To determine whether epithelial RET signaling is necessary for normal GI motility, we generated Vil1^Cre/1000 Ret^flox/flox mice (hereafter RET^EpiKO) that would lack RET in the intestinal epithelium but not in the ENS²⁹. RET^EpiKO mice were born at expected Mendelian ratios, had no overt deficits, and grew similarly to littermate controls (Ret^flox/flox mice; Supplementary Figure 6A-B). Ret transcripts were undetectable in the epithelium of RET^EpiKO mice although expression in the non-epithelial compartment remained robust (Supplementary Figure 6C). In contrast to RET^VipKO mice, GI transit was 35% slower in adult RET^EpiKO mice than controls; however, this deficit was only present in males (Figure 3A). Both male and female RET^EpiKO mice exhibited no alterations in 1-hour FPO (Figure 3B), fecal water content or pellet mass (Supplementary Fig 6D-E), all of which primarily measure colonic function. These observations establish that epithelial RET signaling is essential for normal GI motility in males and suggest that its effects are most prominent in the upper GI tract.

Figure 3. Genetic or pharmacologic disruption of intestinal epithelial RET signaling slows GI transit in a sex-dependent manner.

(A) GI transit time in adult mice lacking RET in the intestinal epithelium (vil1^Cre/1000 Ret^flox/flox; RET^EpiKO) is 35% slower in males, but not females, compared to controls (Ret^flox/flox; WT).

(B) 1-hour FPO was no different in RET^EPiKO mice compared to controls.

(C) GI transit time in mice administered a gut-restricted RET kinase inhibitor (Drug) was 33% slower in males, but not females, compared to vehicle-treated controls (Veh).

(D) RET inhibition did not alter 1-hour FPO in either sex.

(E-F) Colonic contraction frequency and gastric emptying were unaffected by RET inhibition.

(G) Small intestinal transit was slowed by RET inhibition. The x-axis represents intestinal segment number (proximal→distal; n=8-9 male mice).

*p<0.05 by two-way ANOVA (A-D) or one-way repeated measures ANOVA (G).

To ascertain whether acute disruption of epithelial RET signaling would have the same effects as chronic genetic depletion, which could have developmental effects, we determined the effects of RET kinase inhibition on GI motility in adult mice. GSK408B belongs to a family of RET kinase inhibitors that are highly specific and have poor systemic absorption, such that enteral administration enables gut-restricted drug activity³⁰. We administered GSK408B or vehicle to wildtype mice twice daily for 3.5 days by orogastric gavage and then assessed GI motility. Remarkably, RET kinase inhibition phenocopied RET^EpiKO mice. Male, but not female, mice displayed 33% longer GI transit times again with no difference in 1-hour FPO (Figure 3C-D). To delineate which segments of the GI tract relied on RET signaling for normal motility, we examined gastric emptying, small intestinal transit, and colonic motility in GSK408B-treated males. Consistent with normal colonic function seen by 1-hour FPO in vivo, detailed analysis of colonic migrating motor contractions (CMMCs) ex vivo did not reveal a difference in contraction frequency or any other feature of these motor behaviors (Figure 3E; Supplementary Figure 6F-J). In the upper GI tract, RET kinase inhibition did not alter gastric emptying but slowed small intestinal transit (Figure 3F-G). In conclusion, both pharmacological and genetic disruption of RET kinase in the epithelium slowed GI motility selectively in males with effects primarily manifested on the small intestine.

The sex-dependent effects of disrupting epithelial RET were striking and resembled the sex-dependent penetrance of RET mutations, a major factor in the disproportionate incidence of HSCR in males. The role of epithelial RET in motility may similarly be established during fetal development or acquired later, such as upon puberty, when gonadal steroids surge and influence gut motility²⁰. Adult male mice had higher levels of Ret expression in the small intestinal epithelium than pre-pubertal males and GI transit times were abnormal in 12-week-old, but not 3-week-old, RET^EpiKO mice (Supplementary Figure 7), suggesting that motility becomes dependent on epithelial RET signaling upon puberty.

Subsets of enteroendocrine cells in the adult intestinal epithelium express Ret

To determine how epithelial RET signaling promotes motility, we examined RET expression in the adult small intestinal epithelium. EECs are sensory epithelial cells that detect chemical and mechanical stimuli from the gut lumen; at least some express RET^9,10. Consistent with these reports, in 12–16-week-old Ret^CFP/+ mice, the majority of Ret-expressing cells in the duodenum and ileum were immunoreactive for the EEC marker CHGA in both sexes (Figure 4A-C). Similar to enteric neurons, Ret expression was not universal among EECs with ~50% of CHGA⁺ cells expressing Ret, suggesting that Ret expression might be characteristic of specific types of EECs (Figure 4B, Supplementary Figure 8A).

Figure 4. The majority of Ret-expressing small intestinal epithelial cells are enteroendocrine cells, including subsets of enterochromaffin and L-cells.

(A) Ret-expressing cells are immunoreactive for CHGA in the Ret^CEP/+ mouse small intestine.

(B) Large subsets of CHGA⁺ epithelial cells in adult male (M) and female (F) Ret^CFP/+ mice express Ret, in the duodenum (Duo) and ileum (Ile).

(C) The majority of Ret-expressing cells in the epithelium express CHGA.

(D) Many small intestinal epithelial cells immunoreactive for serotonin (5-HT) express Ret (arrows) in a Ret^CFP/+ mouse; but some 5-HT⁺ cells are Ret-negative (*).

(E) Subsets of enterochromaffin cells throughout the small intestine express Ret.

(F) Most Ret-expressing small intestinal epithelial cells contain 5-HT. The proportion was higher in the duodenum than the ileum in both sexes.

(G) Many small intestinal epithelial cells immunoreactive for PYY express Ret (arrows) in a Ret^CFP/+ mouse; but some PYY⁺ L-cells are Ret-negative (*).

(H) A subset of PYY⁺ cells expresses Ret throughout the small intestine in both sexes.

(I) Many Ret-expressing small intestinal epithelial cells contain PYY. The proportion of Ret⁺ cells that are L-cells is higher in the ileum and is no different between the sexes.

(J) Analysis of Ret expression in transcriptional profiling data²¹ shows that Ret is highly expressed in enteroendocrine cells (EEC) overall, as well as L-cells in particular, compared to other intestinal epithelial cells (IEC), in both mice and humans.

* p<0.05, ** p<0.01, and **** p<0.0001 by two-way ANOVA (A-I) or one-way ANOVA (J).

EECs are diverse cells that are often classified by the biogenic amines and peptides they release in response to stimuli. Two major types of EECs that regulate GI motility are enterochromaffin cells and L-cells, which release 5-HT and PYY, respectively. In the male duodenum, 39% of enterochromaffin cells marked by 5-HT immunoreactivity expressed Ret, and this proportion was similar in the ileum and in females (Figure 4D-E). Conversely, while 80-90% of Ret⁺ cells expressed 5-HT in the duodenum, this proportion was lower in the ileum (females 72%, males 52%; Figure 4F), highlighting the potential for regional differences in epithelial RET biology. Among L-cells, 33% of PYY⁺ epithelial cells expressed Ret, and this proportion was consistent across sexes and along the length of the small intestine (Figure 4G-H). Conversely, 49% and 61% of Ret⁺ cells expressed PYY in the male and female duodenum, respectively, and these proportions were even higher in the ileum (67% and 84%; Figure 4I). The proportions of 5-HT⁺ or PYY⁺ cells expressing Ret were lower in the colon than in the small intestine (Supplementary Figure 8B-C). Secondary analysis of transcriptional profiling data²¹ from small intestinal EECs, L-cells and intestinal epithelial cells (IEC) also showed that Ret expression is highly enriched in EECs and L-cells in both mice and humans (Figure 4J). In sum, in the mouse and human intestine, Ret is expressed in large subsets of enterochromaffin cells and L-cells, two major types of EECs that influence GI motility.

RET signaling alters GI motility by dampening peptide hormone release

Enterochromaffin cells respond to mucosal deformation by releasing 5-HT onto local nerve endings to stimulate peristalsis³¹. Genetic elimination of these cells or disruption of their mechanosensory apparatus slows GI transit^32,33. Enterochromaffin cells are also a major source of circulating 5-HT and loss of 60-70% of these cells is sufficient to reduce both serum and intestinal levels³³. To determine whether disrupting intestinal RET signaling slows motility in adult males by compromising enterochromaffin cell-derived 5-HT, we assessed fasting and post-prandial levels of 5-HT and its breakdown product 5-HIAA in GSK408B-treated wildtype mice. RET kinase inhibition had no effect on 5-HT or 5-HIAA levels in the blood or intestinal tissues (Supplementary Figure 9), suggesting that disrupting RET signaling in the adult gut does not slow motility by compromising 5-HT secretion.

Unlike enterochromaffin cells which respond to both mechanical and chemical stimuli, L-cells are generally considered chemosensory cells that release peptides in response to luminal nutrients³⁴. These peptides, such as PYY and glucagon-like peptide 1 (GLP-1), signal through endocrine, paracrine and neurocrine mechanisms to influence a variety of gut functions including motility. Although RET kinase inhibition had no effect on circulating 5-HT, it was associated with much higher levels of GLP-1 and PYY in the post-prandial state (Figure 5A-B). These elevated levels were evident in males, but not females, similar to the finding of slowed GI transit, suggesting that the phenotypes might be linked.

Figure 5. Disruption of RET signaling leads to exaggerated PYY and GLP-1 release, slowing GI motility in a sex- and nutrient-dependent manner.

(A-B) Plasma levels of GLP1 and PYY after overnight fast (Fast) or 2 hours post re-feeding (Fed) in mice administered a RET kinase inhibitor (Drug) or vehicle (Veh). RET inhibition does not alter GLP-1 or PYY levels in females. In males, fasting levels of GLP1 and PYY levels are low in both groups, but feeding leads to much higher levels of both hormones in RET inhibitor-treated mice. *p<0.05 by two-way ANOVA.

(C) Male RET^EpiKO mice have similar GI transit times as controls (WT) following a 6-hour fast.

(D) A single dose of the NPY2R antagonist BIIE0246 restores GITT in male RET^EpiKO mice to that of controls (see Figure 3A for comparison).

(E) BIIE0246 does not alter GITT in WT mice.

(F) Male human rectal organoids treated with selpercatinib show a dose-dependent increase in GLP-1 release. *p<0.05 by one-way ANOVA.

(G) RET expression and proposed functions in the mature intestine.

L-cells normally secrete GLP-1 and PYY in a nutrient-dependent manner. To determine whether the dysmotility associated with epithelial RET deficiency is nutrient-dependent, we measured GI transit in male RET^EpiKO and littermate controls after a 6-hour fast. Fasting slowed GITT in all mice, as expected, but also eliminated the effect of RET deficiency; indicating that dysmotility in RET^EpiKO mice is nutrient-dependent (Figure 5C). GLP-1 and PYY mediate the ileal brake phenomenon, in which nutrient-stimulated release of these peptides from L-cells slows upper GI motility^35,36. Blockade of GLP-1 signaling by the selective GLP1R antagonist exendin 9-39 had no effect on GI transit in control or RET^EpiKO mice (Supplementary Figure 10A). PYY signals through two different receptors, NPY1R and NPY2R. Upon secretion from L-cells, PYY is cleaved into an active form (PYY_3-36) that has highest affinity for NPY2R; NPY2R signaling modulates GI motility³⁷. To determine whether exaggerated PYY signaling was responsible for slowed GI transit in male RET^EpiKO mice, we administered a single dose of the NPY2R antagonist BIIE0246 to RET^EpiKO males and littermate controls prior to GITT measurement. NPY2R inhibition was sufficient to completely rescue the motility deficit in RET^EpiKO mice (Figure 5D) at a dose that had no effect on motility in wildtype males (Figure 5E), suggesting that excessive PYY signaling slows GI motility in RET-deficient males. This excessive signaling could occur if supernumerary L-cells are present in the absence of RET or if RET signaling normally limits the production or release of secreted peptides from L-cells. L-cell numbers and Pyy transcript levels were similar in RET^EpiKO mice and controls (Supplementary Figure 10B-C), suggesting that RET most likely affects peptide release. To determine whether RET signaling similarly limits human L-cell secretion, we tested the effects of the RET kinase inhibitor selpercatinib on male intestinal organoids cultured in conditions that support EEC differentiation (Supplementary Figure 10D)³⁸. Selpercatinib caused a dose-dependent increase in GLP-1 levels in the culture medium at 24-hours (Figure 5F), with no effect on 5-HT or PYY (Supplementary Figure 10E-F). Subsequent exposure to the secretagogue forskolin drove further release of GLP-1, but not PYY, indicating that PYY release was already at maximal levels in this system and RET kinase inhibition exerted no additional effect (Supplementary Figure 10G-H). Together, these observations indicate that RET kinase signaling normally restrains peptide secretion from mouse and human L-cells and disruption of this signaling causes exaggerated release. In vivo, this activity is essential for normal regulation of intestinal motility in male mice via PYY signaling to NPY2R (Figure 5G).

Discussion

RET is a potent receptor tyrosine kinase that plays myriad roles in development but its homeostatic functions in the mature intestine have remained underexplored. Here, we report that important subsets of enteric neurons and EECs in the adult mouse and human intestines express RET. Using genetic and pharmacologic approaches to disrupt RET signaling in the adult intestine, we discovered that RET activity in EECs, but not VIP+ neurons, stimulates intestinal motility by regulating post-prandial PYY release in a sex-specific manner.

Ret expression in the fetal ENS becomes restricted to one of two major developmental trajectories for enteric neurons^11,24. We extend these observations to the adult ENS, demonstrating that Ret expression is stably maintained by 55-70% of neurons in the myenteric and submucosal plexuses throughout life. Although mature cholinergic neurons vary in their Ret expression, NOS1⁺ myenteric neurons, most of which are VIP⁺ inhibitory motor neurons, virtually all express Ret. We constitutively depleted RET in VIP⁺ neurons, however, and found no deficits in GI motility or cell survival. The colonic hypoganglionosis previously reported with Ret ablation after ENS colonization²⁷ may indicate a requirement for Ret in precursors that enter the bowel late in development. Alternatively, RET signaling in adult enteric neurons may be important in the context of injury than homeostasis. Given the long-term risk of enterocolitis in patients with HSCR, this possibility is important to explore.

In contrast to our findings in the ENS, three independent approaches to RET depletion in the intestinal epithelium revealed an essential role for RET in EEC regulation of GI motility. Conditional deletion of all RET signaling in adult Ret^CreER/flox mice led to transient dysmotility that resolved as the gut epithelium turned over. The time course of this defect was similar to the transient dysmotility that occurs upon enterochromaffin cell depletion, which resolves as the cells are restored³³. Second, constitutive RET depletion selectively in the intestinal epithelium slowed GI transit by 35% in male mice. Because developmental effects could limit this constitutive model, we used an orthogonal pharmacological approach as a third model. Pharmacologic inhibition of RET kinase in adult mice phenocopied the genetic models. These observations converge to support the conclusion that RET signaling in EECs is necessary for regulating intestinal motility in males.

Although subsets of both enterochromaffin cells and L-cells expressed RET in the adult intestine, RET kinase inhibition altered levels of PYY and GLP-1, but not 5-HT. The reason for this selectivity is not clear. The release of monoamines and peptides might be regulated differently in EECs such that RET affects one but not the other. EECs in both sexes expressed RET but disruption of RET signaling perturbed GI motility and hormone release only in male mice. Differential stress effects or kinetics of hormone release could have contributed to this result, especially because re-feeding effects were blunted in males in our experimental paradigm (Figure 5A-B). Nevertheless, our observations resemble the striking sexual dimorphism of human HSCR, which occurs 4 times more often in males than females³. GI motility was intact in juvenile RET^EpiKO males, suggesting that puberty alters male EECs and/or their cell circuits. Although evidence for androgen effects on EECs is lacking, estrogens have been shown to stimulate endocrine cell secretion³⁹. Moreover, human female EECs exhibit lower thresholds for nutrient-dependent 5-HT release than male EECs in vitro⁴⁰, and mouse female EECs exhibit more tonic activation of viscerosensory circuits in vivo⁴¹. These biological features may converge to make male EEC secretion more RET-dependent, leading to the observed sex-dependence. Broadly, our findings add to the evidence that the cellular-molecular mechanisms governing gut motility are sex-dependent.

Overall, our findings have implications for the many clinical contexts in which RET signaling is disrupted, including HSCR, MEN2, and cancer chemotherapy. Perturbations of RET signaling in EECs could explain chronic GI dysmotility in HSCR as well as the high rates of diarrhea and constipation in clinical trials with RET kinase inhibitors⁵. Conversely, this nutrient-dependent mechanism could be exploited to treat post-prandial symptoms in patients with functional GI disorders. Gut-restricted RET kinase inhibitors are already in clinical trials for IBS and are highly effective in pre-clinical models of visceral hypersensitivity⁷. Anti-nociceptive effects of these compounds may be attributable to their effects on EECs, which directly communicate with visceral afferent neurons⁴¹. In conclusion, we have found that Ret expression is prominent in the adult ENS and intestinal epithelium of mice and humans. Disruption of epithelial RET kinase signaling in mice revealed an essential, sex-dependent function of RET in EECs that impacted intestinal motility. These observations establish a role for RET signaling in GI homeostasis that could be targeted for the treatment of digestive disorders.

What You Need To Know.

BACKGROUND AND CONTEXT

RET is a membrane receptor tyrosine kinase essential for enteric nervous system development but its significance in the mature intestine in neurons and epithelial cells is largely unknown.

NEW FINDINGS

Some adult enteric neurons and enteroendocrine cells express RET but it is dispensable for their survival. Instead, epithelial RET is required for regulating motility in males by limiting PYY release.

LIMITATIONS

The functional significance of RET expression in mature enteric neurons and the mechanism underlying sex-dependence of RET activity in enteroendocrine cells remain unclear.

CLINICAL RESEARCH RELEVANCE

Loss and gain of function mutations in RET can cause the congenital intestinal disorder Hirschsprung disease and various cancers, respectively. These findings offer new insight into why these disorders and their treatments may cause gastrointestinal symptoms, and uncover a pathway that could be targeted to alleviate chronic dysmotility in Hirschsprung disease and minimize the adverse effects of cancer chemotherapy.

BASIC RESEARCH RELEVANCE

RET has been considered an essential survival factor for enteric neurons. These findings show that RET is not necessary for neuronal or enteroendocrine cell survival, but it is important for regulating hormone release from enteroendocrine cells. GLP-1 and PYY are major therapeutic targets for metabolic disorders – this study identifies RET signaling as a cellular mechanism that normally limits their release.

Abbreviations:

CFP

cyan fluorescent protein

CMMC

colonic migrating motor complex

EEC

enteroendocrine cell

ENS

enteric nervous system

FPO

fecal pellet output

GE/SITT

gastric emptying/small intestinal transit time

GI

gastrointestinal

GITT

gastrointestinal transit time

GLP-1

glucagon-like peptide 1

HSCR

Hirschsprung’s Disease

IBS

irritable bowel syndrome

LC/MS

liquid chromatography/mass spectrometry

MEN2

multiple endocrine neoplasia type 2

MP

myenteric plexus

PYY

peptide YY

scRNA-SEQ

single cell RNA sequencing

SMP

submucosal plexus

UMAP

uniform manifold approximation and projection

WT

wildtype

5-HT

serotonin

Data Availability:

The bulk and single-cell RNA sequencing data sets analyzed are from previously published studies and accession numbers are listed in the Methods. All other data are available in the manuscript and the Supplementary Materials. Please contact the corresponding author for any additional information.