Abstract
Familial dysautonomia (FD) is a rare sensory and autonomic neuropathy that results from a mutation in the ELP1 gene. Virtually all patients report gastrointestinal (GI) dysfunction and we have recently shown that FD patients have a dysbiotic gut microbiome and altered metabolome. These findings were recapitulated in an FD mouse model and moreover, the FD mice had reduced intestinal motility, as did patients. To understand the cellular basis for impaired GI function in FD, the enteric nervous system (ENS; both female and male mice) from FD mouse models was analyzed during embryonic development and adulthood. We show here that not only is Elp1 required for the normal formation of the ENS, but it is also required in adulthood for the regulation of both neuronal and non-neuronal cells and for target innervation in both the mucosa and in intestinal smooth muscle. In particular, CGRP innervation was significantly reduced as was the number of dopaminergic neurons. Examination of an FD patient's gastric biopsy also revealed reduced and disoriented axons in the mucosa. Finally, using an FD mouse model in which Elp1 was deleted exclusively from neurons, we found significant changes to the colon epithelium including reduced E-cadherin expression, perturbed mucus layer organization, and infiltration of bacteria into the mucosa. The fact that deletion of Elp1 exclusively in neurons is sufficient to alter the intestinal epithelium and perturb the intestinal epithelial barrier highlights a critical role for neurons in regulating GI epithelium homeostasis.
Significance Statement
This study demonstrates the key role that ELP1 plays in ENS development and maintenance. Previously, we showed that patients with the peripheral neuropathy, familial dysautonomia (FD), and a mouse model of FD have a dysbiotic microbiome and metabolome and impaired gut motility. FD is caused by a mutation in the gene, ELP1, which plays a critical role in modifying tRNAs and hence proteome homeostasis. Using both FD mouse models and an FD patient biopsy, we identify cellular and molecular disruptions in the ENS and intestinal epithelium that may cause the FD intestinal clinical manifestations. In addition to dysregulated neurogenesis, peripheral axons are reduced, which may underlie gut dysmotility and indirectly, lead to breaches in the epithelium barrier integrity.
Introduction
Familial dysautonomia is a recessive developmental and progressive neurodegenerative disorder resulting from mutations in the gene ELP1, previously referred to as IKBKAP (Anderson et al., 2001; Slaugenhaupt et al., 2001). Classic hallmarks of the disease are cardiovascular instability, decreased pain and temperature sensation, progressive gait ataxia, and optic neuropathy, with a key debilitating comorbidity being gastrointestinal (GI) dysfunction. A recent study revealed that 97% of FD patients reported GI symptoms, with nearly half of them considering it their most burdensome FD symptom (Ramprasad et al., 2021). GI problems include decreased gastric motility, diarrhea, constipation, ulcers, and bleeds and can manifest as severe necrotizing enterocolitis in young patients. The molecular, cellular, and pathophysiological mechanisms mediating the FD GI dysfunction have not been investigated.
With the advent of mouse models that recapitulate key clinical phenotypes of human FD, studies have uncovered pathological mechanisms underlying deficits in dorsal root, cranial, and sympathetic ganglia and revealed how Elp1 reduction alters neuronal function (Cheishvili et al., 2011; Dietrich et al., 2011, 2012; George et al., 2013; Jackson et al., 2014; Lefler et al., 2015; Morini et al., 2016; Naftelberg et al., 2016; Ohlen et al., 2017; Goffena et al., 2018; Ueki et al., 2018; L. Li et al., 2020; Cameron et al., 2021; Even et al., 2021; H. F. Wu et al., 2022; Leonard et al., 2022; Shilian et al., 2022; Tolman et al., 2022). ELP1 is a scaffolding subunit of the six-member Elongator complex, which modifies particular tRNAs (Huang et al., 2005; C. Chen et al., 2011; Lin et al., 2013; Karlsborn et al., 2014a,b; M. J. O. Johansson et al., 2018; Hermand, 2020; Jaciuk et al., 2023). Modification of the wobble uridines in the anti-codon sequence with thiol (s2) and a methoxycarbonyl-methyl (mcm5) methyl regulates the translation of both AA- and AG-ending codons. In the context of diminished U34 modifications in the absence of Elp1, the proteome is altered, with genes exhibiting a biased usage of these codons directly impacted (Bauer and Hermand, 2012; Goffena et al., 2018; Tavares et al., 2021). Interestingly, genetic mutations in the Elongator complex subunits, ELP2–4 and ELP6, also cause neurodevelopmental disorders (Gaik et al., 2023), emphasizing the importance of this complex in the nervous system. No studies have been conducted on the requirement for Elp1 or of any Elongator subunit in the developing or adult enteric nervous system.
The GI system is regulated by both an intrinsic, enteric nervous system (ENS) and by extrinsic inputs from the vagus nerve and sympathetic and dorsal root ganglia (reviewed in Gershon, 1998; Furness, 2006; Sharkey and Mawe, 2023). Together, the ENS and extrinsic inputs function as an organized network to ensure homeostatic responses to gut luminal contents, including any potentially injuries or pathogenic insults, in addition to mediating peristalsis and such key GI functions as absorption of fluids and nutrients. Should any arm of this orchestrated network fail to develop normally and/or misfunction, neurogenic pathologies can result such as Hirschsprung's disease and other motility disorders (Barak et al., 2005; Westfal and Goldstein, 2017; Tilghman et al., 2019).
Intestinal homeostasis also involves balanced communication between the ENS, the intestinal epithelium, the gut microbiome and metabolome, and the gut mucosal immune system. Recently, we showed that the gut microbiome and metabolome of FD patients are perturbed relative to healthy relatives (Cheney et al., 2023; Costello et al., 2023). We also demonstrated that the gut microbiome and metabolome are perturbed in an FD mouse model in which Elp1 is selectively deleted in neurons, suggesting that dyscoordination in FD patients results from a faulty ENS and/or altered extrinsic innervation of the GI tract. Given the commonly reported GI symptoms among FD patients (Ramprasad et al., 2021), their markedly different microbiome diversity and the fact that mouse models for FD recapitulate disease hallmarks including reduced weight, body mass index, and gut transit time (Cheney et al., 2023), the goal of this study was to test whether deletion of Elp1 in mice alters the development and/or maintenance of the ENS and gut mucosa.
Materials and Methods
Experimental design and statistical analyses
Mice
This study was conducted in the AALAC-accredited Animal Resource Center at Montana State University under local IACUC-approved protocols (MSU protocol no. 2021-35-81). Mouse lines have all been previously studied and characterized by our lab (George et al., 2013; Chaverra et al., 2017; Tolman et al., 2022) and by others (Danielian et al., 1998; Coppola et al., 2004; Coksaygan et al., 2006; Rossi et al., 2011; Vianna et al., 2012; Lopez et al., 2018; Deal et al., 2021; Leonard et al., 2022) and were generated by crossing Elp1loxp/loxp mice to either Tuba1a-Cre;Elp1+/loxp, or Phox2b-Cre;Elp1+/loxp, or Wnt1-Cre;Elp1+/loxp lines. To identify cholinergic neurons in Tuba1a-Cre;Elp1 conditional knock-out mice and controls, a Chat-EGFP gene was introduced using B6.Cg-Tg(RP23-268L19-EGFP)2Mik/J mice (Jackson Laboratory, Strain #007902). Phox2b-Cre:B6(Cg)-Tg(Phox2b-cre)3Jke/J mice (Strain #016223; Gautron et al., 2012) were purchased from Jackson Laboratory (donation from Dr. Joel Elmquist). The Tuba1-Cre line was a kind gift from Dr. Lino Tessarollo, NCI.
Neuronal quantification and statistical analysis in GI whole mounts
All neurons (Hu+) were counted in 8–10 fields of view (equal number of imaging fields for mutants and controls within each experiment) with each field of view ranging in area from 0.03 mm2/field at 63× to 0.4 mm2/field at 20×, zoom of 1. The entire thickness (z-axis) of the plexus (as defined by Hu+ expression) was imaged. All imaging and image analysis was conducted blind as to the tissue genotype. If two conditions were being compared, unpaired t tests were conducted using Prism software. When comparing more than two conditions, one-way ANOVA followed by Tukey’s post hoc multiple-comparisons analysis was used. ImageJ (ImageJ FIJI software; http://imagej.nih.gov/ij; NIH) was used to quantify area expressing molecule of interest, e.g., CGRP, synapsin, and E-cadherin. Both male and female mice were included in each experiment, and adult mice were 6–10 months of age, except for the mice used for quantifying dopaminergic and cleaved caspase-3 neurons which were 10–13 months old; age-matched controls and mutants were always used in each experiment.
Carbonic anhydrase 1 (CAR1) quantification
Images were taken using a 20× magnification oil immersion objective lens with a Leica TCS SP8 confocal microscope, and a total of six sections per mouse were imaged for analysis (n = 3 FD mice, 3 controls, all mice were 6 months old). Images were randomized and deidentified to avoid bias during analysis. Images acquired were processed and merged Z-stacks (4.01 µm thickness, 0.5 µm step size) using the Leica Application Suite Advanced Fluorescence software. Images were analyzed using ImageJ FIJI software. The percentage of the CAR1+ epithelium area was quantified by tracing, avoiding gaps due to open crypts, images were converted to 8 bit grayscale, and the image threshold was adjusted using the preset “Moments” threshold settings to detect the stain and reduce background signal. The area stained (detected by threshold settings) within the traced region and the total area of the traced region were quantified and used to calculate the percent area stained within the luminal epithelial layer. The same method was repeated for the entire mucosa layer area. Significance was calculated using a one-way ANOVA and Student's t test.
Quantification of alcian blue/PAS stained mucin in crypts
The entire area of the crypt was traced, avoiding the epithelial cells lining the crypts as above. The area stained (detected by threshold) and the total area of the crypt were quantified to find the percent area of crypt stained. A total of 50 crypts per mouse across 4–5 sections were analyzed, and only completely transverse crypts (the crypts go from top to bottom of the mucosa) were used for analysis. Significance was calculated using a Student's t test.
Quantification of alkaline phosphatase brush border staining
The luminal epithelium layer was traced, excluding any stained blood vessels, images were converted to 8 bit grayscale, and the image threshold was adjusted as described above. The area stained (detected by threshold settings) within the traced region and the total area of the traced region were quantified and used to calculate the percent area stained. Significance was calculated using a Student's t test.
Quantification of FISH/MUC2 staining
Images of sections were taken with a 63× magnification oil immersion objection lens with a Leica TCS SP8 confocal microscope, a total of three random spatially different fields per section (in five sections per mouse) were taken (n = 4 FD mice, 4 controls, all 5–7 months of age). Slides were randomized and deidentified prior to imaging and subsequent analysis to prevent bias. Images acquired were processed and merged Z-stacks (5 µm thickness, 1 µm step size) using the Leica Application Suite Advanced Fluorescence software. The ultraviolet (UV) and 488 nm (green) laser channels were optimized for the DAPI and MUC2 stains, respectively. The number of bacteria in the mucosa layer in each image was quantified via the 561 nm (red) laser channel optimized for the FISH stain. Significance was calculated using a one-way ANOVA and Student's t test.
Whole mount immunochemistry
Mice were anesthetized with isoflurane or CO2 prior to cervical dislocation. The small intestine was removed and ±0.5 or 1 cm from the middle point of the ileum was harvested. A superficial longitudinal incision was made along the serosal surface and the longitudinal muscle, and the adherent myenteric plexus (LMMP) was peeled off from the underlying tissue containing the submucosal plexus (SP). The LMMP and SP were placed on a Sylgard dish and pinned down with mucosa facing down. Tissues were fixed with 4% paraformaldehyde for 1.5 h at room temperature. After fixation, tissues were rinsed three times with phosphate buffer (PBS). LMMP tissue was further peeled by gently peeling away strips of longitudinal muscle with forceps. SP was further processed by gently scraping villi with forceps. Tissues were immersed in blocking solution (BS; Animal-Free Blocker 1×, Vector, catalog #SP-5030, 0.5% Triton X-100 and 20% DMSO) overnight, at room temperature with gentle agitation. The tissues were then removed from BS and incubated with the appropriate primary antibody (in BS) at the listed concentration (Table 1) for 3 d at 4°C in the dark, with gentle shaking. Following incubation with primary antibody, tissues were washed three times (15 min wash each) in PBS plus 0.5% Triton X-100 (PBST) at room temperature, with gentle shaking. The tissues were then incubated (BS) in the appropriate secondary antibody (Table 2) with DAPI at 4°C for 24 h, with gentle shaking in the dark. Tissues were again washed three times in PBST at room temperature in the dark. Finally, tissues were overlaid with Vectashield mounting medium (Vector, catalog #H-1000), coverslipped, and imaged using a Leica SP8 confocal microscope. For immunolabeling of embryonic whole mounts, the steps were the same but shortened, e.g., tissue was fixed for 30 min.
Primary antibodies
Secondary antibodies
Immunolabeling of paraffin-embedded colon tissue for luminal mucus preservation
Methods for mucus preservation were adapted from a previously published protocol (M. E. Johansson and Hansson, 2012). A segment of middle colon containing a stool pellet was removed and submerged in freshly prepared methanol-Carnoy's fixative (60% v/v anhydrous methanol, 30% v/v chloroform, 10% v/v glacial acetic acid) and processed for paraffin embedding (Chaverra et al., 2017).
Human antrum tissue and staining
Tissue biopsies were fixed in Zamboni's fixative followed by cryoprotection in 20% sucrose and then sectioned at 60 µm. Samples were obtained from enrolled participants under Institutional Review Board-approved protocols at New York University School of Medicine (NYU School of Medicine; Dr. Horacio Kaufmann, NYU, Department of Neurology) and University of Minnesota Hospitals (Dr. William Kennedy, University of Minnesota, Department of Neurology).
Hematoxylin and eosin staining for gut structure
Frozen, 4% paraformaldehyde-fixed tissue was cryosectioned at 12 µm and collected onto Superfrost Plus slides. H&E staining was performed as described (Chaverra et al., 2017).
Alcian blue periodic acid-schiff (PAS) mucin stain
Paraffin-embedded colons fixed for mucus preservation were sectioned at 5 µm. Deparaffinized slides were stained using the Thermo Scientific Richard-Allan Scientific Alcian Blue/PAS kit (# 87023). Images were taken at 10× magnification using a Leica DMIL inverted fluorescence microscope with a DFC 3000G camera, images were processed and captured using LAS V4.4 computer software, and two spatially different fields per section (4–5 sections total) were imaged per mouse (n = 3 cKO mice, 3 control, all 6 months old). Prior to analysis, images were randomized and deidentified and analyzed with ImageJ FIJI software.
Epithelial brush border staining
Freshly frozen tissue and newly cut tissue sections, within 2 weeks, were required for optimum staining. Paraformaldehyde-fixed and O.C.T.-embedded tissue was cryosectioned at 12 µm and placed onto Superfrost Plus slides. Slides were left to air-dry for 30 min at 37°C in the dark. Slides were then washed with 1× PBS and immersed in freshly prepared color development solution [1% nitro blue tetrazolium (NBT) and 1% 5-bromo-4-chloro-3-indolyl-phosphate (BCIP)] in Tris buffer and then counter stained with FastGreen. Images of sections were taken at 5× magnification using a Zeiss Stemi SV-11 APO stereo microscope with ProgRes Mac Capture Pro software, and 3–6 sections per mouse were imaged (n = 3 cKO mice, 3 controls, all 6–7 months of age). Images were analyzed using ImageJ FIJI software.
Fluorescence in situ hybridization and mucin2 (FISH MUC2) double labeling
The FISH MUC2 stain was adapted from M. E. Johansson and Hansson (2012). Paraffin-embedded tissue sections that were preserved for mucus were deparaffinized via incubation at 60°C for 10 min and then incubated in prewarmed (60°C) xylene substitute solution (Clear-Rite 3) twice for 10 min each. Sections were then incubated in 99.5% ethanol for 5 min and then let to air-dry. A total of 50 µl of hybridization solution (20 mM Tris-HCl, pH 7.4, 0.9 M NaCl, 0.1% w/v SDS, 5% v/v formamide) heated to 50°C was mixed with 0.5 µl (0.5 µg) of the universal EUB338 probe (5′-GCTGCCTCCCGTAGGAGT-3′) and to the sections. Sections were then overlayed with a coverslip and placed in a humidified chamber and incubated at 50°C in the dark overnight. Slides were then incubated in FISH washing buffer (20 mM Tris-HCl, pH 7.4, 0.9 M NaCl) at 50°C in the dark and then washed in PBS and processed as above for immunofluorescent labeling.
FITC-dextran assay for gut barrier integrity
The fluorescein isothiocyanate (FITC)-dextran assay was adapted from Kominsky et al. (2014); 4 kDa FITC-dextran (Sigma Aldrich, 100 mg/ml in PBS) was administered orally in 150 µl volumes to each mouse. After 4 h, the mice were anaesthetized using isoflurane, and blood was collected via cardiac puncture. Serum was obtained after centrifugation at 5,000 × g for 5 min. Serum was then diluted 1:1 v/v in PBS and 100 µl was placed in 96-well microplates for quantification. Fluorescence was quantified by spectrophoto-fluorometry with an excitation of 485 nm (20 nm bandwidth) and an emission wavelength of 528 nm (20 nm bandwidth), and concentrations were calculated using standard concentrations of FITC-dextran in PBS ranging from 0 to 8,000 ng/ml. Serum from mice not administered FITC-dextran was used to determine the background. Significance was calculated using a Student's t test.
Results
Elp1 is required for normal ENS development
The enteric nervous system derives from neural crest cells that migrate into the rostral and caudal ends of the developing gut and from Schwann cell precursors (Kapur et al., 1992; Young et al., 1998; Uesaka et al., 2016; Espinosa-Medina et al., 2017; Nagy and Goldstein, 2017). To determine the requirement for Elp1 in ENS formation, we could not use a null mouse model because total deletion of Elp1 in mice causes a failure in neural tube development and differentiation and death of embryos by embryonic day (E) 10.5 (Y. T. Chen et al., 2009; Dietrich et al., 2011) indicating that Elp1 is required from the earliest stages of nervous system formation. To obviate this early lethality, we generated conditional knock-out (cKO) mice by crossing Wnt1-cre mice to mice in which exon 4 of Elp1 is floxed (Wnt1-Cre;Elp1+/loxp; George et al., 2013). This line was selected because numerous studies have shown that Wnt1-cre is expressed in the vast majority of enteric neural crest cells and in 96% of enteric neurons (Danielian et al., 1998; Lopez et al., 2018; Deal et al., 2021). Earlier studies have established that Elp1 is not required for normal neural crest migration into the developing gut (George et al., 2013; Jackson et al., 2014). However, whether Elp1 plays a role in neurogenesis, neural differentiation, and/or neuronal survival in the ENS has not been investigated.
To determine if Elp1 is required for normal ENS development, we quantified neuronal number in the stomach (foregut) and colon (hindgut) at E17.5, toward the end of development. In the Wnt-1-cre;Elp1loxp/loxp cKO mouse, neuronal number (Hu+ and Tuj1+) in the distal hindgut/colon was reduced by ∼30% compared with littermate controls (Elp1+/loxp; Fig. 1A,B). Although neuron number (Hu+ and Tuji1+) was reduced along the entire cKO stomach, quantification of neuronal number was restricted to the stomach lesser curvature between the esophagus and duodenum due to their pronounced landmarks. This region was also chosen given its relevance for the swallowing impairment that FD patients experience since it is at this junction where branches of the vagus nerve enter the stomach. Moreover, sympathetic axons enter at this site in association with the small arteries that grow within the gut wall. ENS neurons were significantly reduced in the Elp1 cKO stomach by 60% (Fig. 1C,D) relative to that in control littermate stomachs (Fig. 1C–E). There was also a striking reduction in extrinsic sympathetic innervation as defined by TH+ axons (Fig. 1E); fewer TH+ contacts on ENS neurons are visible in the cKO compared with the control (Fig. 1E). Collectively, these data indicate that Elp1 is required for normal neurogenesis in the ENS in both the foregut and hindgut.
Elp1 is required for normal enteric nervous system (ENS) development. A, B, Hu+ enteric neurons (green) in the hindgut of Wnt1-cre;Elp1loxp/loxp cKO are significantly reduced in number compared with control littermate E17.5 mice; p = 0.0138; unpaired t test, N = 4: 4 cKO hindgut and 4 littermate control hindgut whole mounts were imaged across 13 fields each, at 63×. C, D, All Hu+ enteric neurons were also significantly reduced in the esophagus–duodenum junction region of the stomach in the cKO compared with controls at E17.5; p = 0.0006, unpaired t test, N = 3. All neurons were counted in the stomach between the junction with the esophagus and the duodenum. Sympathetic innervation (TH, red) is also reduced in the cKO stomach. E, While not quantified, enteric neurons often appear less evenly distributed in the E17.5 cKO stomach compared with the control and instead, are often more clustered. See Extended Data Figure 1.1 for more details.
Figure 1-1
(A) Hu + neurons (magenta) express Elp1 protein (green) but Elp1 protein levels are reduced in the Wnt1-cre;Elp1loxp/loxp cKO E17.5 small intestine compared to that in the control. (B) Neurons (Hu+) were counted in the entire submucosal plexus in the distal colon of the Phox2b-cre;Elp1 cKO and control littermates at P9. No significant difference in neuronal number was detected between the two genotypes. Download Figure 1-1, TIF file.
To confirm that Elp1 is expressed in the developing ENS, we examined ELP1 protein levels at E17.5 (Extended Data Fig. 1.1). ELP1 protein was expressed in myenteric neurons, and its level was reduced in ENS neurons in the cKO (Extended Data Fig. 1.1).
Elp1 is required for extrinsic innervation of the GI tract
We then asked whether extrinsic innervation of the GI tract requires Elp1. Extrinsic innervation to the GI tract comprises inputs from the vagus nerve (sensory/afferent and parasympathetic/efferent), dorsal root ganglia (DRGs), and sympathetic ganglia. Because of the importance of the vagal innervation and the fact that it is not targeted using the Wnt1-cre, we analyzed the GI tract of a different mouse line, Phox2b-cre;Elploxp/loxp mice. This line was chosen because previous studies by our lab and others indicates its robust expression in the vagal motor and sensory complex but with only sparse expression in the trunk neural crest with the exception of a few sympathetic neurons and reported absence of expression in enteric neurons (Rossi et al., 2011; Vianna et al., 2012; Varin et al., 2019; Tolman et al., 2022). In support of those reports, in our hands, we did not find any reduction in enteric neurons in the colon of Phox2b-cre;Elp1 cKO mice (Extended Data Fig. 1.1B). With this mouse model, we recently showed that neurons in the nodose ganglia, which comprise the sensory component of the vagus nerve, are reduced in number by ∼40% with a similar reduction in the cranial nerve X brainstem nuclei (parasympathetic efferents). Thus both the afferent and efferent arms of the vagus nerve are reduced in the Phox2b-cre;Elp1 cKO mouse line (Tolman et al., 2022). DRG afferents are not affected in this line since Phox2b-cre is not expressed in the DRG (Rossi et al., 2011).
In the mouse, vagal innervation of the ENS is completed during the first week of postnatal development (Murphy and Fox, 2007; Ratcliffe et al., 2011). Therefore we analyzed innervation of the small intestine at postnatal day 9 (P9), the latest time point at which this mouse line is viable. We were particularly interested in the innervation of the villi mucosal lamina propria given their role in mediating communication with luminal and epithelial signals. To visualize vagal fibers, we immunolabeled with an antibody to TrkB since we and others have shown TrkB expression on vagal axons (Ernfors et al., 1992; Wetmore and Olson, 1995; J. T. Erickson et al., 1996; Murphy and Fox, 2010; Tolman et al., 2022) and work by others has demonstrated vagal innervation of villi (Berthoud et al., 1995; Berthoud and Patterson, 1996; Fox et al., 2000; Powley, 2000; Murphy and Fox, 2007; Boesmans et al., 2008; Williams et al., 2016; Hao et al., 2020). We found that TrkB+ axonal innervation of villi was significantly reduced in Phox2b-Cre;Elp1 cKO mice by ∼75% (Fig. 2). These data are consistent with our previous reporting that the vagus nerve is reduced by approximately half in Phox2b-cre;Elp1 cKO mice at P9 (Tolman et al., 2022). Reduction in vagal innervation of the villi mucosal lamina propria would impair detection of epithelial, subepithelial, immune and luminal cues and the reporting of that information to the CNS.
TrkB+ axons are reduced in frequency in the perinatal Phox2b-cre;Elp1loxp/loxp cKO villi mucosa. A–C, TrkB+ (green) positive axons (arrow) are significantly less frequent in villi mucosa in the small intestine at P9 in the Wnt1-cre;Elp1loxp/loxp cKO compared with control littermates; unpaired t test p = 0.0001, n = 5 mice from each condition.
Elp1 is required in the adult GI tract for normal homeostasis
We then assessed whether villi innervation was reduced in the adult small intestine. Since neither the Wnt1-Cre nor Phox2b-Cre Elp1 cKO lines live beyond the perinatal period for reasons that are not clear but most likely due to failures in respiration (George et al., 2013; Tolman et al., 2022), we investigated the ENS of a neuron-selective Cre mouse line we had previously characterized that lives well into adulthood, Tuba1a-Cre;Elp1loxp/loxp (Ueki et al., 2016; Chaverra et al., 2017; Tolman et al., 2022). We and others (Coppola et al., 2004; Coksaygan et al., 2006; Chaverra et al., 2017; Tolman et al., 2022) have shown previously that this cre is active in a subset of PNS neurons, and in the ENS we found cre+ expression in 23–59% of neurons depending on location in the GI tract (Extended Data Fig. 3.1) and no expression in the GI tract outside of the nervous system (Extended Data Fig. 3.1). To verify that the Tuba1a-Cre was still active in the adult vagus nerve, we crossed the Tuba1a-Cre line to a Rosa26mTmG reporter (George et al., 2013) and analyzed the vagus nerve (Extended Data Fig. 3.1). We found widespread GFP expression within the nerve, the majority of which was composed of TrkB+ axons, indicating maintained Cre recombinase activity in the adult vagus nerve of the Tuba1a-Cre;Elp1 line. TrkB immunolabeling of the adult duodenum in the Tuba1a-Cre;Elp1loxp/loxp cKO mouse reveals a 27% reduction in villi containing TrkB+ axons compared with the controls (Elp1+/loxp; Fig. 3A,B; p = 0.053).
Axons are reduced in frequency in the adult Tuba1a-Cre;Elp1loxp/loxp small intestine. A, B, TrkB+ axons in the villi mucosa of adult proximal duodenum are reduced in frequency in the Elp1 cKO compared with control littermates; unpaired t test p = 0.0534, n = 7. C, D, Synapsin+ axons are more fragmented in the adult ileum villi mucosa in the Elp1 cKO compared with control littermates suggesting degeneration; unpaired t test p = 0.0012, n = 4. While both axon bundles (asterisk) and single axons (red arrow) are present in the control, in the mutant individual axons are prevalent (red arrows). E, Whole mount immunolabeling and clearing of proximal duodenum from adult control and Elp1 cKO reveal reduced villi mucosal innervation (Synapsin, magenta) in the cKO compared with the control. F, G, The density of Synapsin+ axons is reduced in the walls of the adult ileum smooth muscle in the cKO compared with littermate controls; unpaired t test p = 0.0447. See Extended Data Figure 3.1 for more details.
Figure 3-1
(A, B) The expression of the Tuba1a-cre (eGFP; green) is restricted to neurons (red, Tuj-1+) in the embryonic (A; E17.5) and adult GI tract (B). (C) Quantification of the percentage of Hu + neurons that express the Tuba1a-cre in the submucosal plexus in the GI tract. (D) Tuba1a-cre is also active in the adult vagus nerve (B), where most of the Tuba1a-cre-eGFP + axons express TrkB (red). Download Figure 3-1, TIF file.
Interestingly, an analysis of synapsin immunolabeling in Tuba1a-Cre;Elp1 tissue indicate that axons within ileum villi tend to be more isolated (i.e., do not form bundles/fascicles) and more fragmented in the Tuba1a-Cre;Elp1loxp/loxp cKO compared with the controls (Elp1+/loxp; Fig. 3C,D); additionally, axon fragment length in this cKO is significantly reduced compared with control. Axon fragmentation is an indication of neuronal degeneration. Furthermore, synapsin immunolabeling in cleared whole mounts of the small intestine supports these findings and reveals a dramatic reduction in axon numbers and lateral branching within villi (Fig. 3E) in this cKO compared with the control. Finally, synapsin staining also showed a 23% reduction in the intramuscular fiber arrays in the intestinal longitudinal smooth muscle in this cKO compared with controls at 6 months (Fig. 3F,G). Collectively these data demonstrate an overall reduction of axons in the small intestine when Elp1 is deleted from extrinsic and intrinsic neurons, manifesting in reduced villi mucosal innervation in perinatal and in the adult, and in reduced smooth muscle innervation in the adult.
Requirement for Elp1 in the adult intrinsic ENS: neurons and non-neuronal cells
We then examined the intrinsic ENS in Tuba1a-Cre;Elp1loxp/loxp Elp1 cKO mice. First, we asked whether non-neuronal cell types were impacted in these cKO adults. Given the critical role of both glial cells and the interstitial cells of Cajal (ICCs) for GI function, both of these cell types were quantified in the muscularis of the proximal duodenum of adult Tuba1a-Cre;Elp1 cKO and control mice, including both of the neuronal plexuses and longitudinal and circular muscles. The ICCs are a fascinating cell population that act as intermediates in neurotransmission between smooth muscle cells and motor axons (Huizinga et al., 2009; Zhu et al., 2014). Double labeling to visualize both the ICCs (using CD117 antibodies) and glial cells (S100b) shows the close juxtaposition of both cell types in the myenteric plexus (Fig. 4A). Quantification of CD117 and S100b staining intensities shows a significant reduction in both ICCs and glial cells throughout the muscularis in the Tuba1a-Cre;Elp1loxp/loxp cKO compared with that in (Elp1+/loxp) controls (Fig. 4B).
Glial cells and ICCs are reduced in number in the adult Tuba1a-Cre;Elp1loxp/loxp proximal duodenum. A, B, Sections were immunolabeled with antibodies to S100b to visualize glia (green) and to CD117 to visualize ICC (red), and the signal intensity of the area of expression quantified (B); multiple unpaired t tests; *p = 0.025, **p = 0.0047, n = 6.
To determine the effects of Elp1 loss on intrinsic neurons in the adult enteric ganglia, ENS neurons were quantified using the neuronal marker Hu (ANNA-1) in the myenteric and submucosal plexuses in adult Tuba1a-Cre;Elp1loxp/loxp cKO mice. In these mice, Cre is active in approximately half of PNS neurons and in the majority of the CNS (Extended Data Fig. 3.1C; Chaverra et al., 2017; Tolman et al., 2022). In contrast to the reduction of enteric neurons found in embryonic Elp1 cKO mice, in the adult, we found a 20% increase in the number of enteric neurons in the myenteric plexus (Fig. 5); while this trend was also present in the submucosal plexus, it was not significant (data not shown).
Loss of Elp1 in the adult Tuba1a-Cre;Elp1loxp/loxp cKO ENS causes an increase in myenteric neuron cell bodies. A, B, The number of Hu+ neurons (green) was determined in whole mounts of the ileum myenteric plexus of the adult Tuba1a-Cre;Elp1loxp/loxp cKO and control mice; unpaired t test p = 0.0280, control n = 4, cKO n = 5 mice; 10 images/mouse quantified with a 20× objective.
Enteric ganglia contain multiple neuronal subtypes that differ in morphology (e.g., Dogiel type I–VII), neurotransmitter and neuropeptide expression, gene expression and electrophysiological properties and exert distinct functions (Furness, 2006; Drokhlyansky et al., 2020; May-Zhang et al., 2021; Morarach et al., 2021; Obata et al., 2022; Dharshika and Gulbransen, 2023). To determine if the loss of Elp1 differentially impacts ENS subpopulations, whole mount analyses were conducted in the ileum on four different neuronal subtypes. Cholinergic neurons are the major excitatory motor subtype in the ENS and comprise ∼60% of ENS neurons (Bian et al., 2003; C. S. Erickson et al., 2014). We found that their absolute number was increased by ∼35% in the myenteric plexus of adult Tuba1a-Cre;Elp1 cKO mice compared with (Elp1+/loxp) controls (Fig. 6A,B). In contrast, the number or percent of VIP+ neurons in the submucosal plexus, which are typically secretomotor (Furness, 2006) did not change (data not shown). We found a significant reduction in the percentage of tyrosine hydroxylase-positive (TH) neurons in the submucosal plexus of Tuba1a-Cre;Elp1 cKO mice compared with (Elp1+/loxp) controls (Fig. 6C,D). Chalazonitis et al. (2020) showed complete colocalization of dopamine and TH in the same neurons in the submucosal plexus of the murine ileum at this age. This means that the signals driving the increase in neurons in the ENS as cKO mice age, including cholinergic neurons, is not affecting the genesis or differentiation pathway for dopaminergic neurons. Interestingly, in an adult model of colitis, an increase in excitatory (calretinin+) neurons was also detected (Belkind-Gerson et al., 2017).
Changes in subtypes of ENS neurons in the adult Tuba1a-Cre;Elp1loxp/loxp cKO mice. A, B, Whole mounts of ileum were immunolabeled with antibodies to (A) ChAT (green) and to neurons (Hu, magenta); cholinergic neurons in the myenteric plexus are increased in the adult Chat-GFP;Tuba1a-Cre;Elp1LoxP/LoxP cKO myenteric plexus (multiple unpaired t tests; *p = 0.0402; ns = 0.075; n = 3) while (C, D) the percentage of dopaminergic neurons is reduced in the submucosal plexus in the Tuba1a-Cre;Elp1loxp/loxp cKO whole mounts. Tyrosine hydroxylase, red; neurons (Hu, green); multiple unpaired t tests; *p = 0.0154; n = 4.
Given the vital communication between intestinal luminal signals and the ENS, including the critical role that CGRP+ axons play in mediating this communication, we quantified CGRP expression in whole mounts of the adult ileum submucosal plexus. Total CGRP expression was quantified which included both intrinsic (neuronal somata and axons) and extrinsic (axonal) CGRP expression and was found to be significantly reduced in the Tuba1a-Cre;Elp1loxp/loxp cKO compared with (Elp1+/loxp) controls (Fig. 7A,B). Since secretion of CGRP from nociceptive-endings and binding to goblet cells induces the release of mucus in response to noxious stimuli (Yang et al., 2022), we then analyzed CGRP+ axons in villi in sections of adult duodenum (Fig. 7C–E). We found that in the adult Tuba1a-Cre;Elp1loxp/loxp cKO mouse, CGRP+ axons in villi were consistently reduced in frequency compared with those in (Elp1+/loxp) controls (N = 3 cKOs, controls; villi from Tuba1a-Cre;Elp1loxp/loxp had 65, 33 and 40% fewer axons than those in controls). We then asked whether CGRP innervation is already reduced in embryonic development and found a significant reduction in innervation of the rectum at E18.5 (Fig. 7F,G). These data demonstrate that Elp1 is required for normal CGRP expression in both the developing and adult GI tract.
CGRP expression is reduced in both the embryonic and adult Elp1 cKO ENS. A, Immunohistochemical staining for CGRP expression (red) is reduced in the adult Tuba1a-Cre;Elp1loxp/loxp cKO ileum. B, Quantification of CGRP protein expression shows significant reductions in the adult cKO (B; unpaired t test p = 0.0227; n = 4). C, CGRP+ axons are also visually reduced in the villi mucosa in the adult cKO proximal duodenum. D, E, CGRP+ axons in cKO and control villi were quantified in (D) and while a trend is detected of reduction in the cKO, due to considerable variability in absolute numbers of immunolabeled axons between experiments in both controls and cKOs, a significant difference between genotypes was only revealed when the changes between genotypes was quantified in a single experiment; the experiment was repeated three times (E). The line connects the average axon number in the control and cKO in the same experiment. F, G, Quantification of CGRP expression in the embryonic Wnt1-cre;Elp1loxp/loxp CKO rectum also shows significant reduction in CGRP innervation (unpaired t test p = 0.0497; n = 3).
We then sought to determine what could cause the increase in neuronal number in the adult Tuba1a-Cre;Elp1loxp/loxp cKO ENS. Neurogenesis in pathological conditions in the adult ENS has been reported in mouse models of colitis (Belkind-Gerson et al., 2015, 2017). These extra neurons have been shown to be generated by increased progenitor cell proliferation and/or transdifferentiation of glial cell types (Heanue and Pachnis, 2011; Laranjeira et al., 2011; Belkind-Gerson et al., 2013; Kulkarni et al., 2017; Middelhoff et al., 2022). Moreover, increased production of neurons has been reported to occur in the context of ongoing neuronal cell death in some studies (Kulkarni et al., 2017; although see also Virtanen et al., 2022). We first asked whether there were changes in ongoing cell death in the Tuba1a-Cre;Elp1loxp/loxp cKO compared with the (Elp1+/loxp) control. We found that rather than being reduced, programmed cell death as defined by cleaved Caspase-3 immunostaining (Hu+/cleaved Caspase-3+) was in fact significantly increased in the adult Tuba1a-Cre;Elp1loxp/loxp Elp1 cKO small intestine, being twice as high in the cKO compared with controls (Fig. 8A,B).
Apoptosis is increased in the adult Tuba1a-Cre;Elp1loxp/loxp Elp1 cKO ileum myenteric plexus. A, B, Ileum whole mounts were immunolabeled with antibodies to neurons (Hu, green) and to cleaved Caspase-3 (red) and the number of cleaved caspase-3-positive neurons determined in 10 fields per mouse in three mice each. B, Quantification shows a significant increase in cleaved caspase-3-positive neurons in the cKO (unpaired t test p = 0.0239). See Extended Data Figure 8.1 for more details.
This finding suggests that the increase in neurons in the Tuba1a-Cre;Elp1loxp/loxp cKO is due to either increased proliferation and/or survival of progenitors and/or transdifferentiation of non-neuronal cells to neurons, rather than to a reduction in cell death. However, quantification of Ki67+ (Fig. 9A–C) cells did not show consistent differences between Tuba1a-Cre;Elp1loxp/loxp cKO and control small intestine (n = 4 cKO; 4 controls; mean =2.0 vs 2.9 Ki67 cells/section; SD = 0.43 vs 2.4; p = 0.49; two-tailed unpaired t test). Sox2+ glial cells have been shown to transdifferentiate into neurons under inflammatory conditions such as colitis (Belkind-Gerson et al., 2017). Moreover, recent work has clarified that the Sox2+ subpopulation of enteric glia that retain the neurogenic capacity during inflammation are intraganglionic within the myenteric plexus (rather than extraganglionic; Guyer et al., 2023). Before downregulating Sox2, Belkind-Gerson et al. (2017) showed that Sox2+ glial cells underwent a transient stage of Sox2 and Hu coexpression. To this end, we quantified the total number of intraganglionic Sox2+; Hu+ cells in the adult myenteric plexus in the Tuba1a-Cre;Elp1loxp/loxp cKO. Not only did we not find an increase in Sox2+;Hu+ neurons in the cKO [35% of Hu+ neurons were Sox2+ in the control (SD = 31.5) and 29.3% of Hu+ cells were Sox2+ in the cKO (SD = 22; p = 0.9; two tailed t test)] but instead found a consistent decrease in the number of Sox2+ cells in the adult Elp1 cKO ileum compared with their paired experimental control (Fig. 9D–F). Sox2+ cells may be reduced in number due to their death and/or to a rapid downregulation of Sox2 as they transdifferentiate into neurons (i.e., Hu+;Sox2−).
Analysis of Ki67+ and Sox2+ cells in the adult Tuba1a-Cre;Elp1loxp/loxp cKO ileum. A–C, Whole mounts were labeled with antibodies to Ki67 (green, A) and the neuronal marker Hu (magenta) and the number of positive cells quantified (B, C); B, unpaired t test p value indicated in graph, n = 4 (C) unpaired t test p value indicated in graph, n = 4. D–F, Whole mounts were immunolabeled with antibodies to Hu (green) and Sox2 (red) the number of Sox2+ cells determined (E) unpaired t test p value indicated in graph. Asterisks (*) indicate Sox2+ intraganglionic cells. N = 4 mice in each genotype group; in each experiment, one Tuba1a-Cre;Elp1loxp/loxp cKO was compared with one control which were immunolabeled and imaged together. However, due to interexperiment variation (Fig. 9E), when pooled together, the differences between cKO and control are not significant. However, given this variation between experiments, if one compares the Tuba1a-Cre;Elp1loxp/loxp cKO to the control within each experiment, there was a consistent and significant reduction in Sox2+ cells in the cKO compared with the control (Fig. 9F). F, Lines connect the number of Sox2+ cells between the control and cKO mouse in the same experiment to indicate the consistent relative reduction in Sox2+ cells in the cKO compared with the control across all 4 mouse pairs (paired t test p = 0.026).
Neuronal Elp1 required for normal GI epithelium
We have shown previously that the adult Tuba1a-Cre;Elp1loxp/loxp cKO mice are thinner than their control littermates (Chaverra et al., 2017) and have prolonged intestinal transit time and an altered gut microbiome and metabolome (Cheney et al., 2023; Costello et al., 2023) and report here that the innervation of villi mucosal lamina propria and smooth muscle in the small intestine is reduced, all indicative of GI dysfunction. While we did not quantify this observation, the mutant GI tract was frequently fragile with reduced muscle tone and vascularization compared with that in control littermates. In response, we investigated the gut epithelium to determine any (indirect) effects on its composition and integrity in the Tuba1a-Cre;Elp1loxp/loxp cKO mice. Histological staining (hematoxylin/eosin) of paraffin sections of the colon show epithelial cell shedding into the lumen of the cKO (Fig. 10A,B), indicative of an infection and/or high turnover of cells. Thickening of the muscularis propria can coincide with cell shedding during an inflammatory response, but interestingly the muscularis propria is thinner in the Tuba1a-Cre;Elp1loxp/loxp than in the control (Elp1+/loxp).
Neuronal-targeted deletion of Elp1 alters colon epithelium. A, H&E staining of the middle colon in littermate controls (top) and Tuba1a-Cre;Elp1loxp/loxp cKOs (bottom) shows reduction of structural integrity in cKO mice epithelium marked by a compromised epithelial border at the lumen interface (a), engorged goblet cells (arrows) in the crypts of the mucosa (b), and a thinner, more fragile muscularis (c). B, IAP stain of the epithelial bush border at the gut-lumen interface (dark purple; arrows; C) shows a significant reduction in the brush border in cKO colons (*p = 0.012). Blood vessels (asterisks) are also stained within the mucosa and muscularis layers. n = 3 cKO mice and 3 controls. D, Alcian blue PAS stain for mucin protein (blue) is significantly increased in FD colon crypts (p = 0.04) n = 3 cKO mice and 3 controls; however, the mucus barrier layer is much less structured in the cKO than in the control (arrows). See Extended Data Figure 10.1 for more details.
Figure 10-1
(A) Goblet cell diameters were measured in the upper, middle, and lower thirds of crypts stained with Alcian Blue PAS. Mean diameters of goblet cells in FD mice (grey, squares) trended to be larger than controls (white, circles) in all three regions (upper, p value = 0.1; middle, p value = 0.08; lower, p value = 0.1). n = 150 cells per region per mouse, 3 Tuba1a-Cre; Elp1loxp/loxp cKO mice and 3 controls. (B-C) No difference in mean crypt length or width in the colon was observed between groups. n = 50 crypts per mouse, 3 cKO mice and 3 controls. All mice were 6 months old. Error bars = ± SEM. Each data point represents the mean measurement for an individual mouse. Download Figure 10-1, TIF file.
Enterocytes secrete intestinal alkaline phosphatase (IAP) into the gut lumen, detoxifying bacterially produced lipopolysaccharides (LPS). Analysis of the colon epithelium using a stain for IAP (Fig. 10C) reveals a significantly diminished brush border in the cKO which is critical for the digestion and absorption of nutrients. Quantification of IAP intensity supports the visual reduction seen in the Tuba1a-Cre;Elp1loxp/loxp cKO compared with controls (p = 0.013). The cKO colon also contained a thinner and more fragile muscularis layer, engorged goblet cells (mucus producing secretory cells) in the crypts of the mucosa layer, and a compromised epithelial layer at the gut-lumen interface.
Given the reduction in enterocyte brush border in the cKO colon epithelium, we wondered if goblet cells were also altered in the Tuba1a-Cre;Elp1loxp/loxp cKO. An alcian blue periodic acid-Schiff stain revealed a dramatic increase in mucin staining within the cKO colon crypts, relative to that in the controls (Fig. 10D). Crypt width and length revealed no significant difference in crypt size between cKO and controls (Extended Data Fig. 10.1). Since these findings suggest retention of mucus in the goblet cells, we measured goblet cell diameters in the upper, middle, and lower thirds of the crypts because goblet cells become larger as they mature and migrate toward the lumen interface from the bottom of the crypts. Though no significant difference in goblet cell diameter was observed, the mean diameters of goblet cells in all three regions of the crypts trended toward being larger in the cKO mice (Extended Data Fig. 10.1). The firm mucus layer that should be present in the gut lumen adjacent to the epithelial border was present in the control colon but was highly irregular and less firm in the cKO, consistent with impaired goblet cell secretion (Fig. 10D).
While mucus secretion is a key defense mechanism against bacterial pathogens, the epithelial layer that resides at the lumen interface just beneath the firm mucus layer also plays an important role in barrier function. As another approach for analyzing mucin, tissue sections were immunolabeled with antibodies to Mucin 2. Mucin 2 staining reveals an increase in mucus in the Tuba1a-Cre;Elp1loxp/loxp cKO crypts and a frequent reduction in the thickness and firmness of mucosal layers lining the lumen (Figs. 10D, 11A). Since the release of mucus is controlled by the contact between CGRP+ nociceptive afferents and goblet cells (Yang et al., 2022), the fact that we see a significant reduction in CGRP expression in the Tuba1a-Cre;Elp1loxp/loxp cKO could explain the alteration in luminal mucus layer organization and structure.
Since the epithelial layer at the luminal interface is a critical frontline defense against enteric pathogens (Goldberg et al., 2008; Ramanan and Cadwell, 2016), we wanted to examine epithelial barrier integrity in the Tuba1a-Cre;Elp1loxp/loxp cKO mouse colon, i.e., does the reduction in the firm mucus layer result in a physiological breach such that bacteria can penetrate across the epithelium? Using a fluorescent in situ hybridization (FISH) bacterial probe (EUB338), we observed an increased number of bacteria present in the mucosa layer of Tuba1a-Cre;Elp1loxp/loxp cKO mouse colons (Fig. 11A). Furthermore, in instances where bacteria were present, the cKO mice consistently had a larger number of bacteria compared with the controls (Fig. 11B).
Epithelial barrier disorganization in the adult Tuba1a-Cre;Elp1loxp/loxp cKO mice. A, Representative images of the firm layer of mucin (MUC2) in the gut lumen. MUC2 (green), FISH (red), and DAPI (blue) stains demarcate the mucus layer and bacterial localizations at the border between the lumen and epithelium; n = 3 different control regions of interest (ROI) and Elp1 cKO mice. Asterisks (*) indicate the localization of the firm mucus layer in the gut lumen of the control which is much less obvious in the cKO. Scale bar, 100 µm. B, Bacteria were frequently observed in cKO colon epithelium, and in larger numbers, compared with controls [no bacteria observed, p = 0.002; 1–5 bacteria observed in region of interest (ROI), p = 0.0006; >10 bacteria observed in FOV, p = 0.09]. n = 4 FD mice and 4 controls. White arrows indicate FISH positive bacteria in the mucosa. Scale bars, 100 µm. Error bars, ±SEM. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001. Each data point represents an individual mouse.
Given these findings, we probed barrier permeability using an oral gavage of FITC-dextran. While there was a trend of increased leakiness in the Tuba1a-Cre;Elp1loxp/loxp cKO gut, the difference was not significant (Fig. 12A; p = 0.08). We then assessed the expression of the colonocyte marker Car1 to determine whether epithelial cell number was altered in Tuba1a-Cre;Elp1loxp/loxp cKO mice; the data revealed no significant difference in Car1 immunostaining in the cKO colon compared with controls (Fig. 13). However, immunolabeling for E-cadherin, a key adherens junction protein mediating epithelial cell adhesion, revealed dramatically reduced expression in the epithelial layer of cKO mouse colons (Fig.12B,C). Collectively, these data provide evidence that reduction of Elp1 in neurons alone is sufficient to alter the integrity and morphology of the gut epithelium. The colons in adult Tuba1a-Cre;Elp1loxp/loxp cKO mice showed impaired mucus secretion and formation of a firm layer, a chemically altered brush border, and reduced adherens junction protein expression near the epithelial-lumen interface and penetration of bacteria across the luminal surface into the mucosal layer. These data have potentially profound implications for FD enteric clinical pathology and function.
E-cadherin expression is significantly reduced in the Tuba1a-Cre;Elp1loxp/loxp cKO colon epithelium. A, Quantification of FITC-dextran assay for gut leakiness shows an increased trend in FITC leakage across the gut wall in the cKOs compared with controls (ns = 0.121) n = 16 control, n = 14 cKO. B, Representative immunofluorescence (IF) images of E-cadherin (green) and DAPI (blue). C, Quantification shows a significant reduction in epithelial area stained for E-cadherin in FD mice. Unpaired t test, **p = 0.007, n = 3. Scale bar, 100 µm. See Extended Data Figure 11.1 for more details.
Epithelial cell number is not altered in Tuba1a-Cre;Elp1loxp/loxp cKO colons. A, The colonocyte marker CAR1 was used to determine whether the number of epithelial cells was reduced in Tuba1a-Cre;Elp1loxp/loxp cKO mice as a result of impaired neuronal signaling. This experiment revealed no significant difference in CAR1 immunostaining in the colon (B).
Reduced innervation in an FD patient's stomach corresponds to alterations seen in the ENS of murine models
The morphology of mucosal innervation from an FD patient's stomach (antrum) biopsy sample was assessed with immunofluorescence analysis (Fig. 14). The mucosal innervation in the control biopsy was well organized with axons oriented both vertically and horizontally extending to the luminal surface of the mucosa. In contrast, the FD patient antrum reveals a much lower density of axons, areas with gaps in innervation, shorter axons that do not extend to the luminal surface of the mucosa, and a more disoriented, erratic pattern of mucosal innervation (Fig. 14).
Reduction of axons in the human FD patient stomach (antrum). Immunolabeling with antibodies to neurons (Pgp9.5, green), basement membrane (Collagen IV, red), and mucus (Ulex, blue) reveal fewer and shortened more erratic axonal branching in the FD patient antrum than in the Healthy/Control subject. A, low magnification view of section through entire biopsy. Rectangles (a’, a’) in A are enlarged and shown in B, rectangles (c’, c’) in A are enlarged and shown in C.
Discussion
Virtually all FD patients suffer from GI dysfunction yet its underlying cellular basis is not understood. The major finding of this study was identifying a critical role for Elp1, the gene that is mutated in FD, in both the development and maintenance of the ENS. In the absence of Elp1, neuronal target innervation of the mucosa and intestinal walls is reduced, which results in alterations in ENS subpopulation numbers. Moreover, we show that loss of Elp1 solely in neurons is sufficient to perturb the integrity of the GI epithelium and alter GI epithelial cell biology. These data also help explain our recent finding that both FD patients and FD mice have a dysbiotic gut microbiome and altered metabolome (Cheney et al., 2023; Costello et al., 2023) and illustrate the diverse roles neurons exert on gut homeostasis.
While the ENS derives from the neural crest and its derivatives, we and others have shown that the neural crest does not require Elp1 for migration to their targets (Hunnicutt et al., 2012; George et al., 2013; Abashidze et al., 2014; Jackson et al., 2014; Naftelberg et al., 2016). Here we show that Elp1 is not required for the formation of grossly normal myenteric and submucosal plexuses. However our data do show that normal numbers of enteric neurons are not achieved by the end of development in Elp1 cKOs (Fig. 1). This same requirement for Elp1 has also been demonstrated in the DRG, trigeminal ganglion, sympathetic ganglia, and placode-derived nodose ganglia (Dietrich et al., 2012; George et al., 2013; Morini et al., 2016; Naftelberg et al., 2016; Leonard et al., 2022; Tolman et al., 2022). In the DRG we showed that neurogenesis was abrogated due to the reduced proliferation of Pax3+ DRG progenitor cells and to the exacerbation of the endogenous programmed cell death (George et al., 2013). Future studies will be required to discern the mechanisms underlying impaired ENS development in FD.
A major finding from this study is the reduction in CGRP+ afferents in both the developing and adult GI tract (Fig. 7). This finding is consistent with reduced pain sensation in FD patients and reductions in TrkA+ pain and temperature afferents in the FD mouse peripheral sensory ganglia (George et al., 2013; Jackson et al., 2014; Leonard et al., 2022). CGRP afferents serve many critical roles in the GI tract in addition to their typical nociceptive function in detecting potential damaging stimuli. Blocking CGRP receptors compromise gastric protection (Lambrecht et al., 1993; Merchant et al., 1995). Their protective function on the mucosa includes inducing vasodilation (Holzer, 2007), and CGRP has also been implicated in the peristaltic reflex (Grider, 1994, 2003; Pan and Gershon, 2000); hence its reduction may contribute to prolonged GI transit time (Cheney et al., 2023) and FD patient GI motility impairment (Ramprasad et al., 2021). The density of CGRP+ villi afferents observed in this study is commensurate with the reported density of vagal extrinsic CGRP+ villi afferents; however, because we quantified total CGRP fluorescence in the submucosal plexus, we cannot exclude that some of the reduced CGRP in the cKO ileum is derived from intrinsic primary afferent neurons (IPANs) in addition to extrinsic inputs.
Recently a key role for CGRP in gut barrier protection has been identified: its release triggers colonic epithelial goblet cells to secrete mucus which forms a major constituent of the mucosal barrier (Yang et al., 2022). Given we find a reduction in CGRP+ axons and an increase in mucin2+ goblet cells, we posit that goblet cells are likely not receiving sufficient signaling to secrete their mucus, thus retaining crypt mucus. Furthermore, we observed that the firm mucus layer is qualitatively thinner and patchy and to some degree absent in the colon lumen of cKO mice. Interestingly, recent work has shown that release of CGRP also influences the composition of the gut microbiome (Lai et al., 2020; Pujo et al., 2023), which is altered in both FD patients and mice (Cheney et al., 2023).
Besides CGRP neurons, we found that dopaminergic (DA) neurons were significantly reduced in the adult Elp1 cKO submucosal plexus (Fig. 6). DA neurons are a key subpopulation in the ENS that affect gut motility (Kusunoki et al., 1985; Walker et al., 2000; Z. S. Li et al., 2004, 2006; Z. Li et al., 2011; Chalazonitis et al., 2020). As for non-neuronal cells, we report that ICCs were reduced in the adult cKO. Loss of ICCs has been associated with reduced vagal afferent nerve endings, suggesting a survival dependency (Huizinga et al., 2009). Smooth muscle ICCs are innervated by motor neurons and connected via gap junctions to muscle (Wang et al., 2013). The reduction in ICCs was most striking in the cKO myenteric plexus which would impact communication between enteric neurons and ICCs and thereby alter intestinal pacemaker activity (Zhu et al., 2014). ICC depletion has been shown to disturb GI motility, resulting in increased total GI transit time (Yamamoto et al., 2008; Sukhotnik et al., 2021) and damage to ICC and/or their reduction is described in almost every GI disorder (Huizinga et al., 2009). Moreover, the combined reduction in the cKO small intestine of both longitudinal muscle intramuscular array fibers (IMA) and of ICCs would certainly impair small intestine motility if this finding was upheld in the human FD GI tract. In addition to ICCs, we report a significant reduction in enteric glia in the Elp1 cKO mouse. Enteric glia are a molecularly and functionally diverse integral cell population essential for GI homeostasis (Neunlist et al., 2013; Zeisel et al., 2018; Boesmans et al., 2021; Progatzky et al., 2021; Rosenberg and Rao, 2021; Seguella and Gulbransen, 2021; Progatzky and Pachnis, 2022; Sharkey and Mawe, 2023). Reductions in glial cells would also have profound effects on the epithelium; they play a key role in mucosal repair after injury (Villanacci et al., 2008; Neunlist et al., 2013; Progatzky et al., 2021; Wallrapp et al., 2022) while their reduction is associated with necrotizing enterocolitis, a disorder that is not uncommon in FD patients (Barak et al., 2005; Kovler et al., 2021).
A surprising finding in the adult cKO was hyperganglionosis, or a significant increase in total neurons in the myenteric plexus of the adult cKO small intestine. This increase could impact gut contractility (Grundy and Schemann, 2006). This finding is interesting as hyperganglionosis is observed in several enteric clinical pathologies including constipation, intestinal neuronal dysplasia (IND), and pseudo-obstruction (Howard et al., 1984; Athow et al., 1991; Puri and Gosemann, 2012; Terra et al., 2017). Pseudo-obstruction is marked by slow intestinal motility and can occur in inflammatory bowel disease (Qualman and Murray, 1994; Villanacci et al., 2008; O’Donnell and Puri, 2011; Puri and Gosemann, 2012; Belkind-Gerson et al., 2015). IND, which is a variant of Hirschsprung Disease (HD), is also marked by hyperganglionosis. Interestingly, polymorphisms in human ELP1 have been associated with HD (Tang et al., 2010; Wang et al., 2022) and depletion of Elp1 in zebrafish during development causes an HD-like phenotype (Cheng et al., 2015). An important goal for future studies is to identify the enteric progenitor cells stimulated to undergo neurogenesis in the adult cKO. ENS glial cells have been shown to undergo neurogenesis in several injurious conditions (Laranjeira et al., 2011; Uesaka et al., 2016; Belkind-Gerson et al., 2017; Kulkarni et al., 2017; D’Errico et al., 2018; Boesmans et al., 2021; Vicentini et al., 2021). We were unable to ascertain the cause of the hyperganglionosis in our Elp1 cKO mutants: it was not due to a failure in apoptosis, nor to a consistent proliferation of progenitor cells. The most parsimonious mechanism is transdifferentiation of glial cells to neurons since we did find a reduction in Sox2+ and S100+ glial cells in the adult mutant small intestine in the same tissue and time as increased neuronal numbers were present. Future studies are required to determine to what extent the increased neurogenesis could result from intestinal inflammation which is often associated with hyperganglionosis in human disorders (Margolis et al., 2011).
Despite increased enteric neuronal cell bodies, we consistently found reduced axon numbers throughout the mouse small intestine and in the FD patient stomach biopsy. In support of a key role for Elp1 in distal target innervation is a recent study showing that Elp1 gene dosage regulates genes involved in axonal projection (Morini et al., 2022). Mucosal nerve deficiencies in the vicinity of enteric ganglia (i.e., in tissue that is not aganglionic) have been found in cases of chronic childhood constipation, suggesting that what may cause the perturbations that lead to constipation, i.e., impaired peristalsis, water secretion, and absorption, is actually not the reduction in ENS neuronal somata but rather to a reduction in axonal outgrowth, branching, and target innervation (Wendelschafer-Crabb et al., 2009). Alternatively, the reduction in axons detected in the adult cKO mice could be due to loss of extrinsic axonal innervation rather than to an impairment of axonogenesis by the supernumerary intrinsic neurons in the murine FD small intestine. Additional studies will be needed to investigate this interesting question. Regardless of the mechanism, reduced axonal innervation in the GI tract could underlie the altered gut microbiome and metabolome of FD patients and mice and the prolonged GI transit time exhibited by both FD mice and patients (Krausz et al., 1994; Martinelli and Staiano, 2011; Ramprasad et al., 2021; Cheney et al., 2023). Gastric mucosa and the submucosal vessels are innervated by IPANs that detect luminal contents, in particular, acid (Laine et al., 2008; Yandrapu and Sarosiek, 2015). Since their activation stimulates release of CGRP, which leads to mucosal protection, reductions of mucosal axons in the stomach could contribute to the GI problems in FD. Although we did not have tissue with which we could quantify ENS neurons in the FD patient biopsy, an analysis of NOS+ inhibitory neurons in the appendix of FD patients showed a significant reduction in patients compared with controls (Bar-Shai et al., 2004), while another study of FD myenteric plexus in the esophagus and stomach showed a reduction in axons (Ariel and Wells, 1985). It is important to note that FD patients frequently take antibiotics for infections that ensue from lung aspiration, and antibiotics in mice can cause reductions in enteric neurons (Muller et al., 2020; Vicentini et al., 2021). In contrast, the mice used in our study were not treated with antibiotics.
Given that the Elp1 deletion in the adult line (Tuba1a-cre;Elp1loxp/loxp) was neuronal specific, one of the most surprising findings reported here were the severe impacts on the colon epithelium (Figs. 10–13) in the adult cKO mutant mice. Our data reveal a reduction in IAP expression in brush border enterocytes, a disruption in goblet cell mucus secretion, disorganized mucus lamination, bacterial breach of the mucosal barrier, and a striking reduction in E-cadherin expression in epithelial cells. These findings implicate a critical role for ENS neurons in maintaining a normally structured intestinal epithelium and epithelial barrier integrity. If these findings are also present in the GI tract of FD patients, they could underlie many of the GI disturbances they experience, including ulcers, GI bleeding, and frequent bacterial infections.
In this study, the changes to epithelium must be non-cell autonomous since Elp1 was exclusively deleted from neurons. Several studies have revealed the close proximity and extensive communication between enteric neurons with their overlying epithelium (Lundgren et al., 2011; Neunlist et al., 2013; Bohórquez et al., 2015; Bellono et al., 2017) with its disruption present in several neurological disorders. Moreover, stimulation of neuronal input provides a protective function on the intestinal epithelial barrier (Lundgren et al., 2011; Neunlist et al., 2013; Cavin et al., 2020). Reduction of IAP in the gut is associated with a weakened gut mucosal defense and increased bacterial translocation (Goldberg et al., 2008). Another epithelial line of defense at the lumen interface are adherens junction proteins, such as E-cadherin, which help prevent infiltration of bacteria. Gut microbiome dysbiosis reported in other neurodegenerative diseases, such as Alzheimer's, Parkinson's, and ALS, is also associated with increased gut permeability (Nandwana et al., 2022). Decreased E-cadherin expression has been reported in an ALS mouse model (S. Wu et al., 2015) and observed in human intestinal samples from patients with necrotizing enterocolitis (Buonpane et al., 2020).
Collectively these data may explain the underlying pathology that mediates several debilitating FD GI problems. High intestinal permeability facilitates excessive passage of toxins which can cause the development of necrotizing enterocolitis (Israel, 1994), a condition that is not uncommon in young FD patients (Ramprasad et al., 2021). Peptic ulcers are also common in FD patients and may result from diminished sensory innervation of gastric blood vessels that might otherwise have protected against acid exposure. GI bleeding is also increased in FD patients and so too might result from Ramprasad et al. (2021) dysregulated mucosal defenses and mucosal blood flow that can result from reduced sensory innervation.
In summary, deletion of Elp1 in the developing ENS impairs ENS embryonic development and GI homeostasis in the adult. In particular we show here that neuronal Elp1 is required for normal mucosal and smooth muscle innervation and regulation of neurogenesis. Coupled with previous data demonstrating gut microbiome dysbiosis in both FD patients and mice, our data reveal many of the pathogenic cellular processes that could impair the GI tract in FD patients and mice. Further studies into the mechanisms linking neuronal dysfunction, gut microbiome dysbiosis, and gut barrier impairment may help guide potential therapeutic strategies to alleviate the debilitating GI symptoms experienced by FD patients.
Footnotes
This study was supported by the National Institutes of Health R01 DK117473 awarded to F.L., S.W., and V.C. and by R15NS090384 awarded to L.G. We also thank Gwen Wendelschafer-Crabb, Department of Neurology, University of Minnesota, for technical support and Dr. Michael D. Gershon and Dr. Alcmene Chalazonitis for helpful discussions when formulating this project.
↵*M.C. and A.M.C. contributed equally to this work.
The authors declare no competing financial interests.
- Correspondence should be addressed to Frances Lefcort at flefcort{at}gmail.com.
