Stress-induced glucocorticoid signaling impairs enteric neurotrophin BDNF–TrkB pathway and drives gastrointestinal dysmotility

Abstract

Stress is a key contributor to gastrointestinal (GI) dysmotility, particularly in patients with disorders of gut-brain interactions. Since GI motility is governed by the enteric nervous system (ENS), stress may act by altering ENS function. While stress activates glucocorticoid signaling via the hypothalamic-pituitary-adrenal axis, the impact of stress-mediated glucocorticoid signaling on ENS biology remains poorly understood. In the central nervous system, glucocorticoids reduce specific isoforms of brain-derived neurotrophic factor (BDNF), impairing signaling through its receptor, tropomyosin related kinase B (TrkB), and contributing to behavioral dysfunction. However, the identity of ENS-specific Bdnf isoforms, their glucocorticoid sensitivity, and the effect of enhanced TrkB signaling on GI motility in stressed animals has not been characterized. Here, using male and female mice, we show that >85% of post-natal ENS Bdnf transcripts are glucocorticoid-responsive isoforms. We also demonstrate that both BDNF and its receptor TrkB (Ntrk2) are expressed by enteric neurons. Stress, in male mice, and administration of dexamethasone (DEXA), a synthetic glucocorticoid receptor agonist, in both male and female mice, cause GI dysmotility, which we demonstrate is associated with significantly reduced Bdnf transcripts in the longitudinal muscle – myenteric plexus tissue in vivo. Dexamethasone exposure also represses Bdnf transcript and mature protein levels in longitudinal muscle – myenteric plexus tissue in vitro. Notably, treatment with HIOC, a selective TrkB agonist, rescues GI transit defects in dexamethasone-treated animals. These findings identify BDNF-TrkB signaling as a key modulator of stress-induced ENS dysfunction and highlight TrkB as a promising therapeutic target for GI dysmotility in disorders of gut-brain interactions.

Stress-associated gastrointestinal (GI) dysmotility, a prevalent issue in irritable bowel syndrome, afflicts large numbers of patients (1, 2, 3). GI motility is regulated by the ENS (4), suggesting that stress-associated GI dysmotility can occur due to ENS dysfunction. Healthy ENS function depends on optimal neurotrophin signaling, where neurotrophins secreted by various gut cell types interact with their receptors on ENS neurons (5, 6, 7). Brain-derived neurotrophic factor (BDNF) and its receptor tropomyosin kinase B (TrkB, coded by Ntrk2) comprise of one such system. In the central nervous system (CNS), stress reduces Bdnf expression, impairing BDNF – TrkB signaling to alter CNS neuronal function and organisms’ behavior (8, 9, 10). However, it remains unclear whether stress-associated GI dysmotility involves reduced Bdnf expression, and if so, by what mechanisms stress alters enteric Bdnf expression.

Rodent studies have demonstrated that BDNF regulates GI motility, as its presence hastens GI transit of luminal contents (11, 12, 13, 14). Mirroring this, decreased BDNF was detected in colonic tissues from patients with slow-transit constipation (11), and exogenous supplementation of BDNF improved GI motility in constipated patients (15). These data suggest that mechanisms regulating Bdnf expression are relevant for pathobiology of GI dysmotility. Further, altered Bdnf transcript levels in distressed irritable bowel syndrome patients (16, 17) suggest that Bdnf dysregulation may extend to stress-induced GI dysmotility. However, how stress affects Bdnf expression in GI motility-regulating cells, and whether normalizing BDNF-TrkB signaling in stressed animals rescues GI motility is not well understood.

The Bdnf gene consists of 8 non-coding (e1-e8) and 1 coding (e9) exons (18). Each non-coding exon possesses a unique promoter, allowing for differential regulation of various Bdnf isoforms (18). Variation in promoter usage underlies context-dependent regulation of Bdnf expression (19, 20). One of these contexts, stress, represses the expression of two specific isoforms – Bdnf-IV and Bdnf-VI – through hypothalamic-pituitary-adrenal axis-dependent modulation of glucocorticoid (GCs) levels (21, 22, 23, 24). GCs activate glucocorticoid receptor (GR), which then bind to glucocorticoid responsive elements in the promoters of exons (e4 and e6) of Bdnf-IV and Bdnf-VI isoforms to downregulate their expression (25). While stress and GC supplementation reduce gut motility in adult mice (26), it remains unknown if repression of specific Bdnf isoforms in the ENS underlies stress-related GI dysmotility. Interrogating this biology has been limited by the unclear identity of BDNF- and TrkB-expressing cells in the gut wall, alongside a lack of profiling of Bdnf isoform expression patterns and their susceptibility to stress and GC stimulation.

To address these gaps, here, we used immunofluorescence, bulk transcriptomics, and quantitative PCR to study the expression of BDNF and TrkB in the longitudinal muscle containing myenteric plexus (LM-MP) tissue, which contains the bulk of the motility-regulating ENS neurons. Further, we used restraint stress and dexamethasone (DEXA), a synthetic GR agonist, to assess how stress and GR stimulation influence Bdnf expression to drive GI dysmotility and whether stimulating TrkB signaling rescues GI motility in a mouse model of GC-induced GI dysmotility. Our results show BDNF and TrkB are expressed by adult enteric neurons, and that stress-responsive isoforms represent the majority of Bdnf isoforms in the post-natal myenteric plexus tissue. We find that stress and DEXA exposure reduce expression of stress-responsive Bdnf isoforms and cause GI dysmotility - which is reversed by using HIOC, a selective agonist of BDNF-receptor TrkB.

Results

Restraint stress in adult mice slows down whole gut motility

We applied restraint stress daily for 1 h for 5 consecutive days on a cohort of adult male C57BL/6 mice and found that restraint stress caused a significant slowing of whole gut motility, as assayed by the carmine-red dye method (n = 5/cohort; mean ± S.E. of whole gut transit time (WGTT) (in min): Control: 93.0 ± 5.612; Stress: 123.0 ± 8.746, p = 0.02, two-tailed unpaired Students’ t test; Fig. 1A). Further, restraint stress caused a significant increase in serum corticosterone levels (n = 4/cohort; mean ± S.E. of corticosterone concentration (in nM): Control: 331.3 ± 72.05; Stress: 633.6 ± 83.96, p = 0.03, two-tailed unpaired Students’ t test; Fig 1B). This replicates a prior study which reports a similar effect of physiological stress on serum corticosterone and whole gut motility (26).

Figure 1.

Restraint stress significantly increases whole gut transit time in adult male mice.A, graphical representation of the mean ± standard error of the whole gut transit times (assayed using the carmine-red dye method) of two cohorts of mice, where the mice in the ‘Stress’ cohort were restrained for an hour every day for 5 days, and the ‘Control’ cohort mice were handled similarly but not restrained. Mice in the Stress cohort showed significantly longer whole gut transit times on average than mice in the control cohort (n = 5/cohort; mean ± S.E. of WGTT (in min): Control: 93.0 ± 5.612; Stress: 123.0 ± 8.746, p = 0.02, ∗p < 0.05. two-tailed unpaired Student’s t test). B, graphical representation of the mean ± standard error of serum corticosterone concentration (assayed using ELISA) of cohorts of ‘Stress’ and ‘Control’ mice. Mice in the Stress cohort showed significant increase in their serum corticosterone levels than mice in the Control cohort (n = 4/cohort; mean ± S.E. of corticosterone concentration (in nM): Control: 331.3 ± 72.05; Stress: 633.6 ± 83.96, ∗p < 0.05, two-tailed unpaired Students’ t test). WGTT, whole gut transit time.

Stress-responsive Bdnf isoforms are expressed in the murine longitudinal muscle - myenteric plexus tissue

Having confirmed that restraint stress delays whole gut transit, we next tested whether stress-responsive Bdnf isoforms are detected in the gut wall. For this, we characterized the expression profile of Bdnf in the motility-regulating LM-MP tissue of post-natal gut across three ages. We isolated LM-MP tissue from the ileum of juvenile (1 month), adult (6 months), and aging (17 months) male and female mice (n = 3/sex for 1-month and 6-month-old mice, n = 2 males and 2 females for 17-month-old mice) and performed bulk RNA sequencing. First, we confirmed that Bdnf transcripts are detected in the ileal tissues of male and female mice at all ages. We next analyzed whether the stress-responsive Bdnf IV (from e4) and Bdnf VI (from e6) are expressed in the ileal LM-MP tissue across ages and in the two sexes. We found that at all profiled ages, Bdnf VI is the dominant isoform, making up 85.3% (95% CI [81.5%, 89.0%]) of all detected Bdnf transcripts (Fig. 2). Similarly, Bdnf V (from e5) expression is detected at all ages and in both male and female tissues (percent representation of Bdnf V at 1-month: 9.7%, 6-months: 9.4%, and 17-months: 6.2%). By contrast, the other stress-responsive isoform Bdnf IV is detected at 1-month and 6-months of age in both male and female ileal LM-MP (percent representation of Bdnf IV at 1-month: 5.3%, and 6-months: 2.3%), but it is not detected in the aging ileal LM-MP – suggesting that the expression of this isoform is reduced or lost with aging. On the other hand, the expression of Bdnf II (e2) is not detected at 1-month of age but is detected at 6-months and 17-months of age (percent representation of Bdnf II at 6-months: 3.3%, and 17-months: 5.1%). 6-months age is the only age we queried in the bulk RNA-seq where we detected the expression of Bdnf-III in the murine ileal LM-MP (representing 2.9% of all Bdnf transcripts at this age) (Extended Figures. 2–1). These data show that stress-responsive Bdnf isoforms (IV and VI) represent 85% of all Bdnf transcripts across age in the murine ileal LM-MP tissue.

Figure 2.

Bdnf-IV and VI together account for >85% of all Bdnf isoforms in the ileal LM-MP at all post-natal ages. Graphical representation of the proportions of various Bdnf isoforms whose transcripts were detected in the murine ileal LM-MP layer in two different sexes across three different ages of 1-month (juvenile), 6-months (mature adult), and 17-months (aging). This representation shows that stress-responsive Bdnf-VI is the dominant isoform found in the murine ileal LM-MP at all ages, with its proportions accounting for >80% of all isoforms. The isoform Bdnf-V is the second most represented isoform at all ages in the murine LM-MP layer, but little is known about the mechanisms that regulate its expression. In addition to Bdnf-VI, Bdnf-IV is another stress-responsive isoform detected in the juvenile and adult mice but is not detected in aging murine ileal LM-MP. In contrast to Bdnf-IV, Bdnf-II isoform is detected in the adult and aging ileal tissues, but not in the juvenile tissues. Finally, Bdnf-III isoform is detected only in the adult ileal LM-MP and not at the other ages. Detailed age-specific isoform specific abundance of various Bdnf isoforms presented in Extended Figures 2–1. BDNF, brain-derived neurotrophic factor; LM-MP, longitudinal muscle – myenteric plexus.

BDNF and TrkB are expressed by myenteric neurons

Prior studies have used immunohistochemical and conditional transgenic mice to establish the expression of BDNF in myenteric ganglia in the post-natal gut (27, 28). However, the exact nature and proportions of adult small intestinal myenteric cells that express BDNF have not been determined. By contrast, using a reporter transgenic mouse, it has been demonstrated that TrkB is expressed by myenteric neurons in the adult gut (29). To support this result, we performed immunohistochemistry with BDNF and Hu antibodies using TrkB-GFP mice to confirm that BDNF and TrkB are expressed by enteric neurons. We observed that BDNF and TrkB are expressed exclusively by enteric neurons (Fig. 3A). On quantification of 448 Hu-expressing cells from 3 mice, we found that ∼24% of all Hu-expressing myenteric cells were immunoreactive for BDNF (average ± standard error of percent of all Hu⁺ cells immunoreactive for BDNF: 23.65 ± 0.51) while a non-overlapping ∼16% of all Hu-expressing myenteric cells expressed TrkB (average ± standard error of percent of all Hu⁺ cells expressing TrkB-GFP: 16.09 ± 1.87) (Fig. 3B). This data shows that BDNF and TrkB are neuronally expressed genes in the myenteric plexus. We further performed immunohistochemistry using anti-BDNF antibodies on formalin fixed paraffin embedded tissue sections from the adult small intestine of human patients without any known motility disorder and again found that while BDNF protein was present in the myenteric ganglia, its abundance was significantly higher in neuron-like cells (Fig. 3C).

Figure 3.

Bdnf is a neuronally expressed gene in the adult murine small intestinal enteric nervous system. Transcriptomic and proteomic evidence of neuronal expression of BDNF. A, 3D projection image obtained by confocal microscopy of a representative murine adult small intestinal myenteric ganglion from a TrkB-GFP mouse, where GFP expression (green) is enhanced by immunostaining with anti-GFP antibodies, and antibodies against BDNF (red) and ANNA-1 antisera containing anti-Hu antibodies (gray) label BDNF-expressing cells and Hu-expressing myenteric neurons, respectively. Immunolabeling shows that TrkB-GFP-expressing cells (green arrow) and BDNF-expressing cells (red arrow) are both myenteric neurons. Nuclei are labeled with DAPI (blue) and scale bar represents 10 μm. B, quantification of BDNF-immunoreactive and TrkB-GFP⁺ neurons in the small intestinal myenteric plexus tissue of 3 adult mice. C, immunostaining a human small intestinal tissue section with antibodies against BDNF also shows that while the secreted form of BDNF is present inside and outside of the myenteric ganglia, the highly enriched presence of BDNF in the cytoplasm of a myenteric neuron (black arrow) suggests its likely source of expression. The scale bar represents 100 μm. BDNF, brain-derived neurotrophic factor.

Restraint stress causes a significant reduction in Bdnf IV expression

Since stress has been observed to reduce Bdnf expression in the CNS (30), we next tested whether restraint stress-induced slowdown of GI motility is associated with alterations in expression of different Bdnf isoforms. Using LM-MP tissues derived from restrained (i.e., stressed) and restraint-naïve control mice (n = 5/cohort) and by using quantitative real time PCR (qRT-PCR) with isoform-specific primers, we found that 5 days of daily restraint stress caused a significant reduction in the expression of stress-responsive Bdnf IV isoform compared to control mice (n = 5/cohort; mean ± S.E. fold change of Bdnf IV: Control: 1.002 ± 0.034; Stress: 0.800 ± 0.044, p = 0.006; two-tailed unpaired Students’ t test) while none of the other tested Bdnf isoforms (I, II, III, V or VI) showed any significant change (Fig. 4).

Figure 4.

Restraint stress causes a significant reduction in Bdnf-IV expression. Graphical representations of the mean ± standard error of fold change in transcript abundance of selected Bdnf isoforms (calculated using the 2^−ΔΔCt method) in small intestinal LM-MP tissues derived from two cohorts of mice – relative to the expression of housekeeping gene Hprt. Mice in the ‘Stress’ cohort were restrained for an hour every day for 5 days, and the ‘Control’ cohort mice were handled similarly but not restrained. Mice in the Stress cohort showed significantly reduced expression of Bdnf-IV isoform while no other isoform showed significant reduction in expression compared to control cohort (n = 5/cohort; mean ± S.E. fold change of Bdnf-IV: Control: 1.002 ± 0.034; Stress: 0.800 ± 0.044, p = 0.006; two-tailed unpaired Students’ t test). BDNF, brain-derived neurotrophic factor; LM-MP, longitudinal muscle – myenteric plexus.

Glucocorticoid receptor is expressed by all cells of the myenteric ganglion

The Thaiss lab showed that stress significantly increases serum corticosterone levels (26). To study whether the increased endogenous glucocorticoid has a direct effect on myenteric neurons, we first tested whether GR is expressed by myenteric neurons. By performing immunostaining on small intestinal myenteric plexus tissue with anti-GR antibodies and ANNA-1 antisera, we observed that GR is expressed ubiquitously by all LM-MP cells, including Hu-expressing cells of the myenteric ganglia (Fig. 5A). This suggests that stress-mediated increase in corticosterone can directly influence all neurons, including Bdnf-expressing myenteric neurons, through their expression of GR.

Figure 5.

DEXA exposure significantly delays whole gut transit while downregulating myenteric BDNF expression.A, representative of a myenteric ganglion and surrounding cells from an adult murine small intestinal LM-MP tissue when immunostained with antibodies against glucocorticoid receptor (red and with ANNA-1 antisera containing anti-Hu antibodies (green) shows that GR is expressed by all cells of the LM-MP tissues. GR shows strong nuclear immunoreactivity in a subset of Hu-expressing myenteric cells (white arrow) while showing relatively weaker nuclear immunoreactivity in other cells (yellow arrow). Nuclei are stained with DAPI (blue). The scale bar represents 10 μm. B, graphical representation of the mean ± standard error of the whole gut transit times (assayed using the carmine-red dye method) of two cohorts of mice in male and female sexes, where the mice in the ‘DEXA’ cohort were dosed once with DEXA (5 mg/kg body weight), and the ‘Control’ cohort mice were dosed with Vehicle (10% DMSO in Saline). Mice in the DEXA cohort showed significantly longer whole gut transit times on average than mice in the control cohort (n = 5/cohort; mean ± S.E. of WGTT (in min): Males: Control: 145.0 ± 20.12; DEXA: 214.0 ± 17.49, ∗p < 0.05; Females: Control: 136.0 ± 17.49; DEXA: 226.0 ± 18.12, ∗∗p < 0.01; two-tailed unpaired Students’ t test). C, cohort of adult male mice dosed with DEXA and sacrificed 4 h later showed a detectable presence in DEXA in their serum and small intestinal tissues, when compared to cohort of adult male mice dosed with vehicle (n = 5/cohort; mean ± S.E. of DEXA conc in tissue (in nmoles/Kg tissue weight): Control: 3.12 ± 4.16; DEXA: 159.4 ± 52.39; mean ± S.E. of DEXA conc in serum (in nM): Control: 6.71 ± 3.03; DEXA: 268.2 ± 59.77; ∗p < 0.05, two-tailed unpaired Students’ t test). D, adult male mice dosed with DEXA and sacrificed 4 h later showed a significant down regulation in the expression of stress and glucocorticoid-responsive Bdnf IV and Bdnf VI transcripts in their small intestinal LM-MP tissues, when compared to age- and sex-matched mice that were dosed with vehicle (n = 3/cohort; mean ± S.E. fold change of Bdnf IV: Control: 1.011 ± 0.104; DEXA: 0.536 ± 0.087, ∗p < 0.05; mean ± S.E. fold change of Bdnf VI: Control: 1.012 ± 0.110; DEXA: 0.422 ± 0.042, ∗∗p < 0.01; two-tailed unpaired Students’ t test). E, small intestinal LM-MP tissues from adult male mice that were cultured with DEXA for 4 h showed a significant reduction in the abundance of BDNF protein as assessed by ELISA, when compared to tissues from age- and sex-matched mice that were cultured with vehicle. (n = 3/cohort; mean ± S.E. BDNF levels (in pg/g of tissue): Control: 473.0 ± 45.32; DEXA: 85.5 ± 32.63, ∗∗p < 0.01; two-tailed unpaired Students’ t test). F, small intestinal LM-MP tissues from adult mice that were cultured with DEXA for 4 h showed a significant reduction in the abundance of both Bdnf IV and Bdnf VI transcripts (Males: n = 3/cohort; mean ± S.E. fold change of Bdnf IV: Control: 1.002 ± 0.047; DEXA: 0.510 ± 0.099; mean ± S.E. fold change of Bdnf VI: Control: 1.018 ± 0.140; DEXA: 0.520 ± 0.079; Females: n = 4/cohort; mean ± S.E. fold change of Bdnf IV: Control: 1.040 ± 0.168; DEXA: 0.525 ± 0.083; mean ± S.E. fold change of Bdnf VI: Control: 1.014 ± 0.097; DEXA: 0.583 ± 0.045; ∗p < 0.05, ∗∗p < 0.01; two-tailed unpaired Students’ t test). G, in separate experiments, small intestinal LM-MP tissues from adult male mice that were cultured with DEXA either for 1 h or for 2 h showed that while 1 h of exposure to DEXA caused no significant change in the expression of either Bdnf IV or VI transcripts (n = 3/treatment; mean ± S.E. fold change of Bdnf-IV: Control: 1.061 ± 0.238; DEXA: 0.795 ± 0.257; mean ± S.E. fold change of Bdnf-VI: Control: 1.086 ± 0.327; DEXA: 0.740 ± 0.083), 2 h exposure of LM-MP to DEXA caused a significant reduction in the abundance of both Bdnf IV and Bdnf VI transcripts (n = 3/treatment at 2 h; mean ± S.E. fold change of Bdnf-IV: Control: 1.020 ± 0.137; DEXA: 0.497 ± 0.049; mean ± S.E. fold change of Bdnf-VI: Control: 1.014 ± 0.114; DEXA: 0.630 ± 0.063), ∗p < 0.05; two-tailed unpaired Students’ t test). BDNF, brain-derived neurotrophic factor; DEXA, dexamethasone; LM-MP, longitudinal muscle – myenteric plexus; WGTT, whole gut transit time.

Dexamethasone exposure significantly delays whole gut transit time while downregulating myenteric BDNF expression

To test whether GR stimulation alone, without restraint stress, similarly alters GI motility, we tested the effect of DEXA, a synthetic fluorinated homolog of hydrocortisone and a mimetic of endogenous glucocorticoid (31), on WGTT. Mice treated with DEXA showed a significant delay in their whole gut motility, when compared to vehicle-treated control mice (n = 5/cohort; mean ± S.E. of WGTT (in min): Males: Control: 145.0 ± 20.12; DEXA: 214.0 ± 17.49, p = 0.03; Females: Control: 136.0 ± 17.49; DEXA: 226.0 ± 18.12, p = 0.007; two-tailed unpaired Students’ t test; Fig 5B). This replicates an independent finding from the Thaiss group that observed significant delay in whole gut transit in DEXA-treated mice, albeit after multiple doses of DEXA (26).

We next asked to what extent DEXA dosing is detected in serum and small intestinal tissue at 4 h post-dose, which is the maximum amount of time required to complete the whole gut motility assay. Using ELISA, we found that both serum and small intestinal tissue showed detectable and significant concentrations of the drug (n = 5/cohort; mean ± S.E. of DEXA concentration in tissue (in nmoles/Kg tissue weight): Control: 3.12 ± 4.16; DEXA: 159.4 ± 52.39, p = 0.04; mean ± S.E. of DEXA concentration in serum (in nM): Control: 6.71 ± 3.03; DEXA: 268.2 ± 59.77, p = 0.011; two-tailed unpaired Students’ t test; Fig 5C). We next asked if Bdnf expression is altered by DEXA exposure, similarly to the effect of restraint-stress. For this, we tested whether exposure to DEXA causes a significant reduction in the expression of stress-responsive Bdnf IV and/or VI isoforms. In small intestinal LM-MP of adult male mice treated with DEXA cohort or vehicle (control cohort) in vivo, we assessed the abundance of these Bdnf isoforms at 4 h post-treatment, a timepoint aligning with the duration of our WGTT experiments. By performing qRT-PCR, we found that expression of both Bdnf IV and VI were significantly reduced in small intestinal LM-MP of mice treated with DEXA when compared to vehicle treated mice in the control cohort (n = 3/cohort; mean ± S.E. fold change of Bdnf IV: Control: 1.011 ± 0.104; DEXA: 0.536 ± 0.087, p = 0.025; mean ± S.E. fold change of Bdnf VI: Control: 1.012 ± 0.110; DEXA: 0.422 ± 0.042, p = 0.007; two-tailed unpaired Students’ t test; Fig. 5D).

Bdnf is an immediate early gene whose expression may be triggered by tissue damage (32, 33). To ensure that our results are not affected by the dissection of tissues immediately before RNA isolation, we cultured adult small intestinal LM-MP tissues from male mice in stem cell media with vehicle (DMSO) or DEXA (5 μM) for 4 h and assessed the effect of DEXA on both BDNF protein expression and Bdnf isoform transcript abundance in LM-MP tissue. By performing BDNF-specific ELISA, we found that LM-MP cultured with DEXA for 4 h showed significant reduction in the levels of BDNF protein when compared to LM-MP cultured with vehicle (n = 3/cohort; mean ± S.E. BDNF levels (in pg/g of tissue): Control: 473.0 ± 45.32; DEXA: 85.5 ± 32.63, p = 0.002; two-tailed unpaired Students’ t test; Fig 5E). We next repeated this assay and performed qRT-PCR with Bdnf isoform-specific probes against LM-MP tissues from male and female mice. Again, we found that exposure to DEXA significantly reduced the expression of both stress-responsive isoforms Bdnf IV and Bdnf VI when compared to control tissues that were similarly cultured but were instead exposed to vehicle (Males: n = 3/cohort; mean ± S.E. fold change of Bdnf IV: Control: 1.002 ± 0.047; DEXA: 0.510 ± 0.099, p = 0.030; mean ± S.E. fold change of Bdnf VI: Control: 1.018 ± 0.140; DEXA: 0.520 ± 0.079, p = 0.028; Females: n = 4/cohort; mean ± S.E. fold change of Bdnf IV: Control: 1.040 ± 0.168; DEXA: 0.525 ± 0.083, p = 0.047; mean ± S.E. fold change of Bdnf VI: Control: 1.014 ± 0.097; DEXA: 0.583 ± 0.045, p = 0.014; two-tailed unpaired Students’ t test; Fig. 5F).

Having observed that DEXA exposure suppresses Bdnf expression in LM-MP tissue at 4 h—the duration of the whole gut transit time assay—we next sought to identify the earliest timepoint at which continued DEXA exposure significantly alters Bdnf transcript levels. In separate experiments, we cultured LM-MP tissues in media containing vehicle (control cohort) or DEXA (DEXA cohort) for 1 h or for 2 h and assessed the transcript abundance of Bdnf-IV and VI at these timepoints independently, when compared to control tissues that were similarly cultured but were instead exposed to vehicle. We found that while there was no significant change in the expression of either Bdnf isoforms at the 1 h timepoint (n = 3/treatment at 2 h; mean ± S.E. fold change of Bdnf-IV: Control: 1.061 ± 0.238; DEXA: 0.795 ± 0.257, p = 0.49; mean ± S.E. fold change of Bdnf-VI: Control: 1.086 ± 0.327; DEXA: 0.740 ± 0.083, p = 0.36; two-tailed unpaired Students’ t test), expression of both Bdnf-IV and VI isoforms were significantly reduced at the 2 h timepoint (n = 3/treatment at 2 h; mean ± S.E. fold change of Bdnf-IV: Control: 1.020 ± 0.137; DEXA: 0.497 ± 0.049, p = 0.023; mean ± S.E. fold change of Bdnf-VI: Control: 1.014 ± 0.114; DEXA: 0.630 ± 0.063, p = 0.043; two-tailed unpaired Students’ t test; Fig. 5G). Given that whole gut transit time experiments were started 30 min after DEXA injection, this would suggest that DEXA -mediated alterations to Bdnf-IV and VI expression occur between the first 30 to 90 min of the 4-h experiment.

Deletion of TrkB in enteric neurons delays motility

We observed that DEXA and stress-induced loss of BDNF is associated with GI dysmotility, and prior data has shown using the hemizygous Bdnf ^+/− mutant mice that reduced BDNF also drives GI dysmotility (11). We next aimed to investigate how loss of TrkB in ENS affects GI motility. Alternate splicing in the Ntrk2 gene produces two TrkB isoforms: A full-length isoform (TrkB.FL) that contains extracellular and intracellular domains and hence carries out intracellular cell signaling in response to extracellular BDNF, or a truncated (TrkB.T1) isoform that lacks the intracellular kinase domain and thus acts as a sink for BDNF (34, 35). To understand the native proportion of these isoformsn in the adult murine ENS, we first queried their relative expression using isoform specific primers (Table 1). Here, we observed that at steady-state, expression of both FL and T1 isoforms is detected in the adult murine small intestinal LM-MP tissue, with TrkB.T1 having significantly greater abundance than the TrkB.FL isoform (n = 3; mean ± S.E. Delta Ct of TrkB.FL: 8.653 ± 0.1771; Delta Ct of TrkB.T1: 5.882 ± 0.5381; p = 0.008; two-tailed unpaired Students’ t test; Fig. 6A).

Table 1.

List of all antibodies and primers used

Resource	Vendor	Cat. No.
BDNF antibody	Abcam	Ab108319
GR antibody	Cell Signaling	CST3660T
GFP antibody	Aves Labs	GFP1020
Primers & Probes
Target	Sequence/Cat. No.	Company
Bdnf IV Fwd	CGTTTACTTCTTTCATGGGCG	IDT
Bdnf IV Rev	AGCTGCCTTGATGTTTACTTTG	IDT
Bdnf IV probe	56-FAM/AGGATGGTC/ZEN/ATCACTCTTCTCACCTGG/3IABkFQ	IDT
Bdnf VI Fwd	ATGCAACCGAAGTATGAAATAACC	IDT
Bdnf VI Rev	GGACCAGAAGCGTGACAAC	IDT
Bdnf VI probe	56-FAM/ACCAGGTGA/ZEN/GAAGAGTGATGACCATCC/3IBkFQ	IDT
Hprt TaqMan assay	Mm00446966_m1	Invitrogen
Uchl1 TaqMan assay	Mm00495900_m1	Invitrogen
Bdnf-I	CAAACAAGACACATTACCTTCCAGC	IDT
Bdnf-II	AGCCAGCGGATTTGTCCGA	IDT
Bdnf-III	CTCCCCGAGAGTTCCG	IDT
Bdnf-V	CTCTGTGTAGTTTCATTGTGTGTTCG	IDT
Bdnf-IX common Rev	ACGTTTACTTCTTTCATGGGCG	IDT
TrkB Fwd	AGCAATCGGGAGCATCTCT	IDT
TrkB.FL Rev	CTGGCAGAGTCATCGTCGT	IDT
TrkB.T1 Rev	TACCCATCCAGTGGGATCTT	IDT
Nos1 Fwd	ATCTGTCTCGCCAGCCATCAGCCA	IDT
Nos1 Rev	GGAGCTTTGTGCAGTTTGCCGTCG	IDT
Chat Fwd	GATGAACGCCTGCCTCCAAT	IDT
Chat Rev	GGCATACCAGGCAGATGCAG	IDT
Hprt Fwd (SyBr)	CAGTCCCAGCGTCGTGATT	IDT
Hprt Rev (SyBr)	GCAAGTCTTTCAGTCCTGTCCAT	IDT

Figure 6.

Conditional mutation of Ntrk2 in enteric nervous system neurons drives depletion of TrkB.Fl isoform resulting in significant GI dysmotility.A, graphical representation of the mean ± standard error of the Delta Ct values of Ntrk2 transcripts encoding full length TrkB.FL and the truncated TrkB.T1 isoforms in the healthy adult murine small intestinal LM-MP tissues shows that the TrkB.T1-encoding transcripts have significantly lower Delta Ct values (suggesting higher abundance) when compared to the transcripts encoding TrkB.FL isoform (n = 3; mean ± S.E. Delta Ct of TrkB.FL: 8.653 ± 0.1771; Delta Ct of TrkB.T1: 5.882 ± 0.5381; ∗∗p < 0.01; two-tailed unpaired Students’ t test). B, graphical representation of the mean ± standard error of the fold change of Ntrk2 transcripts encoding full length TrkB.FL and the truncated TrkB.T1 isoforms in the adult murine small intestinal LM-MP tissues of Mutant Wnt1-cre:TrkB^fl/fl mice compared to Wnt1-cre:TrkB^WT/WTWT control mice, shows a significant decrease in expression of TrkB.FL-encoding transcripts in mutant mice, while level of TrkB.T1-encoding transcripts remained unchanged (n = 4/cohort; mean ± S.E. fold change of TrkB.FL: WT: 1.042 ± 0.1678; mutant: 0.4628 ± 0.1194, ∗p < 0.05; mean ± S.E. fold change of TrkB.T1: WT: 1.069 ± 0.2226; mutant: 1.370 ± 0.3450, ns = p > 0.05; two-tailed unpaired Students’ t test). C, graphical representation of the mean ± standard error of the whole gut transit time (in minutes) of Mutant Wnt1-cre:TrkB^fl/fl mice and Wnt1-cre:TrkB^WT/WTWT control mice shows that the mutant mice have significantly increased transit time of the carmine red dye – suggesting slower GI motility, when compared to the WT control mice (Males: n = 4/cohort; mean ± S.E. of WGTT (in min): WT: 160.0 ± 6.12; mutant: 250.0 ± 15.0; Females: n = 6/cohort; mean ± S.E. of WGTT (in min): WT: 195.0 ± 18.03; mutant: 265.0 ± 0.0; ∗∗p ≤ 0.01; two-tailed unpaired Students’ t test). D, graphical representation of the mean ± standard error of the density of Hu-immunolabeled cells in the small intestinal LM-MP tissues of Mutant Wnt1-cre:TrkB^fl/fl mice and Wnt1-cre:TrkB^WT/WTWT control mice shows that the density of Hu⁺ cells in 20X fields remains unchanged in mutant mice when compared to control WT mice; n = 4/cohort; mean ± S.E. of Hu ⁺ cells per 20× field: WT: 75.99 ± 6.759; mutant: 71.89 ± 10.37; ns = p > 0.05, two-tailed unpaired Students’ t test). E, graphical representations of the mean ± standard error of fold change in transcript abundance of selected neuronal genes Uchl1, Nos1, and Chat (calculated using the 2^−ΔΔCt method) in small intestinal LM-MP tissues derived from two cohorts of adult male mice (Mutant Wnt1-cre:TrkB^fl/fl mice compared to Wnt1-cre:TrkB^WT/WTWT control mice) – relative to the expression of housekeeping gene Hprt. Statistical analyses shows no significant difference in the relative abundance of transcripts of these genes between the two cohorts (n = 4/cohort; mean ± S.E. of fold change in gene expression: Uchl1: WT: 1.005 ± 0.058; mutant: 1.026 ± 0.118; Nos1: WT: 1.055 ± 0.213; mutant: 0.918 ± 0.149; Chat: WT: 1.045 ± 0.179; mutant: 1.525 ± 0.529; ns = p > 0.05; two-tailed unpaired Students’ t test;). LM-MP, longitudinal muscle – myenteric plexus; WGTT, whole gut transit time.

To study the effect of ENS-specific TrkB knockout, we generated Wnt1-cre:TrkB^fl/fl mice, in which TrkB.FL is homozygously knocked out in Wnt1-expressing neural crest and their derivative cells, including in the neural crest-derived cells of the enteric nervous system, through excision of the kinase domain-containing exon 15 (36, 37). Using this model, we first tested the effect of this conditional TrkB mutation in ENS on the expression of the two TrkB isoforms in the gut wall. Profiling the small intestinal LM-MP of adult Wnt1-cre:TrkB^fl/fl mutant mice compared to Wnt1-cre:TrkB^WT/WT WT control mice, we found a significant decrease in expression of transcripts encoding TrkB.FL isoform, while level of transcripts encoding for TrkB.T1 remained unchanged (n = 4/cohort; mean ± S.E. fold change of TrkB.FL: WT: 1.042 ± 0.1678; mutant: 0.4628 ± 0.1194, p = 0.03; mean ± S.E. fold change of TrkB.T1: WT: 1.069 ± 0.2226; mutant: 1.370 ± 0.3450, p = 0.49; two-tailed unpaired Students’ t test; Fig. 6B).

Given our observation that DEXA exposure both reduces Bdnf levels and delays whole gut transit, we hypothesized that loss of TrkB.FL in Wnt1-expressing neural crest-derived cells would likewise reduce motility through decreased BDNF-TrkB signaling in enteric neurons. Consistent with this, we observed a significant increase in WGTT in Wnt1-cre:TrkB^fl/fl mutant mice compared to Wnt1-cre:TrkB^WT/WT WT control mice (Males: n = 4/cohort; mean ± S.E. of WGTT (in min): WT: 160.0 ± 6.12; mutant: 250.0 ± 15.0, p = 0.001; Females: n = 6/cohort; mean ± S.E. of WGTT (in min): WT: 195.0 ± 18.03; mutant: 265.0 ± 0.0, p = 0.01; two-tailed unpaired Students’ t test; Fig 6C).

Finally, we investigated if the conditional Ntrk2 mutation and resulting significant reduction in TrkB.FL isoform expression and delayed transit time is associated with alterations in the overall neuronal structure of the ENS. For this, we performed immunostaining using antibodies against Hu, which is expressed by myenteric neurons and has been previously used to assess changes to ENS structure (38), on small intestinal LM-MP from adult Wnt1-cre:TrkB^fl/fl mutant and Wnt1-cre:TrkB^WT/WT WT control mice. After imaging 7 to 8 20X fields per mouse and on enumerating of numbers of Hu-immunolabeled cells/20X field, we found no significant difference in cell density of Hu⁺ cells in a 20X field (n = 4/cohort; mean ± S.E. of Hu ⁺ cells per 20× field: WT: 75.99 ± 6.759; mutant: 71.89 ± 10.37; p = 0.75, two-tailed unpaired Students’ t test; Fig. 6D).

We further used qPCR against Uchl1 (that encodes for PGP9.5 protein, a known pan-neuronal marker for enteric neurons), Nos1 (that encodes for NOS1 protein, a known marker for inhibitory nitrergic enteric neurons), and Chat (that encodes for ChAT protein, a known marker for excitatory cholinergic enteric neurons) to test whether the loss of TrkB in the Wnt1-cre neural crest-derived neurons effected any changes globally to ENS innervation or to selectively nitrergic or cholinergic enteric neurons. Using adult small intestinal LM-MP tissues from male Wnt1-cre:TrkB^WT/WT and Wnt1-cre:TrkB^fl/fl mice, we found that loss of TrkB causes no significant change in the expression of major neuronal genes (n = 4/cohort; mean ± S.E. of fold change in gene expression: Uchl1: WT: 1.005 ± 0.058; mutant: 1.026 ± 0.118; p = 0.87; Nos1: WT: 1.055 ± 0.213; mutant: 0.918 ± 0.149; p = 0.62; Chat: WT: 1.045 ± 0.179; mutant: 1.525 ± 0.529; p = 0.44; two-tailed unpaired Students’ t test; Fig. 6E).

Targeted activation of TrkB restores motility in DEXA -induced delayed transit models

Our observations that reduced Bdnf expression is associated with slowed whole gut motility in DEXA-treated mice, and that ENS-specific loss of the TrkB.FL isoform, which carries out BDNF signaling, likewise slows motility, suggest that DEXA’s effect on whole gut motility may be mediated by a reduction in BDNF–TrkB signaling. This suggests that increasing TrkB signaling would improve the gut motility of DEXA-treated mice. To test this hypothesis, we studied whether a synthetic and specific agonist of TrkB can restore whole gut motility of DEXA-treated mice. In rodent models, the N-acetyl serotonin-derived compound HIOC has been demonstrated to be a specific agonist of TrkB (39). We found that mice pre-treated with HIOC 30 min before DEXA administration (DEXA + HIOC cohort) had significantly lower whole gut transit times on average than mice treated with DEXA alone (DEXA cohort), while HIOC treatment did not alter whole gut transit in DEXA-untreated mice (n = 9 mice for DEXA cohort; mean ± S.E. of WGTT (in min): 228.3 ± 7.12; n = 9 mice for DEXA + HIOC cohort; mean ± S.E. of WGTT (in min): 150.0 ± 21.94; n = 5 mice for HIOC alone cohort; mean ± S.E. of WGTT (in min): 154.0 ± 13.91; n = 5 mice for No DEXA No HIOC cohort; mean ± S.E. of WGTT (in min): 151.0 ± 15.44. A two-way ANOVA was performed to evaluate the effects of DEXA and HIOC on WGTT. The results indicated a significant main effect for DEXA, F(1, 24) = 4.419, p = 0.0462; a significant main effect for HIOC, F(1, 24) = 4.663, p = 0.0410; and a significant interaction between DEXA and HIOC, F(1, 24) = 5.436, p = 0.0285. Sidak’s multiple comparisons test reveals statistically significant differences between DEXA+/HIOC- and DEXA+/HIOC+ mice (p_adj = 0.0058), between DEXA+/HIOC- and DEXA-/HIOC+ mice (p_adj = 0.0267), and between DEXA+/HIOC- and DEXA-/HIOC+ mice (p_adj = 0.0355). Differences between all other comparisons were found to be not statistically significant (Fig. 7).

Figure 7.

HIOC ameliorates delayed whole gut motility in mice with DEXA -induced dysmotility. DEXA-treated adult male mice that were pre-treated with HIOC - the synthetic and specific agonist of TrkB - showed significant reduction in their WGTT of orally gavaged carmine-red dye, and hence improved whole gut motility, when compared to age- and sex-matched mice that were dosed only with DEXA (n = 9/cohort; mean ± S.E. of WGTT (in min): DEXA only: 228.3 ± 7.12; DEXA + HIOC: 150.0 ± 21.94). By contrast, HIOC dosing of mice in the absence of DEXA treatment does not cause any change in the WGTT of mice (n = 5/cohort; mean ± S.E. of WGTT (in min): Vehicle control (No DEXA + No HIOC): 151.0 ± 15.44; HIOC only: 154.0 ± 13.91. Statistically significant main effects found for DEXA treatment, F(1, 24) = 4.419, p = 0.0462; for HIOC treatment, F(1, 24) = 4.663, p = 0.0410; and for interaction between DEXA and HIOC treatments, F(1, 24) = 5.436, p = 0.0285 (Two-Way ANOVA). Sidak’s multiple comparisons test reveals statistically significant differences between DEXA+/HIOC- and DEXA+/HIOC+ mice; between DEXA+/HIOC- and DEXA-/HIOC+ mice; and between DEXA+/HIOC- and DEXA-/HIOC+ mice. Differences between all other comparisons were found to be not statistically significant (∗p_adj < 0.05, ∗∗p_adj < 0.01, ns = p_adj > 0.05). WGTT, whole gut transit time; DEXA, dexamethasone.

Discussion

Our study establishes that altered BDNF–TrkB signaling in the adult ENS is an important mechanism through which the stress-induced glucocorticoid pathway manifests GI dysmotility. Our study is the first to interrogate the nature of the Bdnf isoforms present in the post-natal gut of juvenile, adult, and aging mice, and establishes that at all ages, the stress-responsive (hence glucocorticoid-responsive) isoforms Bdnf IV and VI together account for the majority of Bdnf isoforms expressed in the gut wall. Further, by demonstrating the localization of BDNF and TrkB-GFP expression in Hu-expressing myenteric cells in the adult murine gut wall, we show that the BDNF–TrkB crosstalk is neuronal in nature. Since stress increases glucocorticoid levels, and glucocorticoids downregulate the expression of Bdnf-IV and Bdnf-VI in other systems (40), we tested and found that both stress and GR agonism using a synthetic GR agonist DEXA significantly downregulated expression of specific Bdnf isoforms in the adult LM-MP. Additionally, we observed that this was associated with a significant slowdown of intestinal motility. By demonstrating that all myenteric cells ubiquitously express GR, we show that these cells are responsive to increases in glucocorticoid levels. Downstream of GR-signaling, we modulated the activity of BDNF-TrkB signaling through ENS-specific reduction of TrkB, demonstrating that this results in slowed whole gut motility, while not effecting overall enteric neuron density in the myenteric plexus. Finally, we established the involvement of BDNF-TrkB signaling in glucocorticoid-mediated GI dysmotility by showing that TrkB agonism significantly improves motility in a mouse model of DEXA -induced GI dysmotility. These results together show that elevated GR activation, an important downstream effect of physiological stress, reduces the expression of stress-responsive Bdnf isoforms to cause GI dysmotility, an effect that can be ameliorated through stimulating TrkB signaling.

Prior studies have shown that stress and the stress hormone corticotrophin-releasing factor in the central nervous system induce a slowdown of gastric, small intestinal, and whole gut transit (41, 42). Through stimulation of the hypothalamic-pituitary-adrenal axis, stress causes increased release of the glucocorticoid cortisol, the effect of which can be mimicked by using the GR agonist DEXA (40). Similar to our findings, Schneider et al. demonstrated that both stress and DEXA increase whole gut transit time of orally-gavaged carmine red dye – suggesting a slowdown of GI motility (26). This may appear at odds with observations of stress-induced increased defecation in thanatosis (43) and studies on stress-induced increases in colonic motility from the Tache and Vanner groups (44, 45, 46). However, this can be reconciled by the fact that the latter studies specifically focused on stress’s effects on colonic motility and did not consider the effects on upper or overall GI motility. Indeed, another study from the Tache group (47) showed that stress slowed down gastric motility while increasing colonic motility. Since the whole gut transit assay reflects dye progression from stomach to expelled feces, we reason that the prolongation in dye transit time in our study and in the Schneider et al. study supports the conclusion that stress impairs gastric and small intestinal motility, and that this delay likely outweighs any acceleration of colonic motility which results in overall slower GI transit.

We found that both a single dose of DEXA and 5 days of restraint stress impaired gastrointestinal motility. Our stress paradigm caused a comparable increase in serum corticosterone levels as reported by the Thaiss group in Schneider et al., but the single dose of DEXA effected significant changes to whole gut motility at a much lower serum and tissue DEXA levels than in the protocol by Thaiss group, which required repeated daily exposures to DEXA. At the end of the treatment, both stress and DEXA significantly reduced the expression of both stress-responsive Bdnf IV isoforms, while dexamethasone additionally reduced Bdnf VI expression. This indicates that DEXA may exert a stronger suppressive effect on Bdnf expression than stress-induced elevations in endogenous corticosterone. Consistent with this, DEXA has a substantially higher affinity for the GR than cortisol, making it a more potent GR agonist (48, 49). It is also plausible that repeated stress led to habituation of mice, leading to a reduced effect on Bdnf-VI expression. It is also plausible that given that expression of Bdnf-IV is activity-dependent and that of Bdnf-VI is not (50), stress-induced GR stimulation mediates the repression of BDNF in enteric neurons by suppressing the activity of BDNF-expressing enteric neurons.

While our bulk RNA-sequencing analysis identified the predominant Bdnf isoforms, the gene’s inherently low expression at steady state likely limits the sensitivity of this approach in capturing all isoforms. This limitation is illustrated by the absence of the Bdnf-I isoform (whose expression is unaffected by stress) in our transcriptomic dataset, despite its detection by isoform-specific qRT-PCR assays. Notably, the absence of Bdnf-IV—the sole glucocorticoid-sensitive isoform downregulated by restraint stress—in aging murine LM-MP tissues may hold important implications for age-related susceptibility to stress-induced dysmotility. Investigating how these isoforms respond to restraint stress in aged mice will be the focus of future studies.

We employed both in vivo DEXA exposure and our established ex vivo culture system (51) to investigate the impact of glucocorticoid signaling on Bdnf isoform expression. Our findings demonstrate that the ex vivo model faithfully recapitulates the in vivo effects of DEXA, establishing it as a robust platform for both assessing how BDNF protein abundance is modified in response to DEXA exposure as well as for future studies examining how diverse exposomes influence the ENS and its surrounding cellular environment.

It is important to note that while stress-mediated changes in glucocorticoid signaling is an important mechanism through which stress manifests dysfunction, stress also drives changes in other hormones in the viscera, which include changes in adrenaline and in the peripheral release of corticotrophin releasing factor from mast cells (52, 53). It is currently unclear how these alternate stress-mediated pathways affect enteric BDNF–TrkB signaling, and future studies will focus on better understanding this interplay. In addition, future studies will test whether the glucocorticoid-mediated changes in the expression of Bdnf-VI and other isoforms occur differentially in an age-associated manner.

Ntrk2 encodes two TrkB isoforms: a full-length TrkB.FL that faithfully carries out canonical BDNF signaling, and a truncated TrkB.T1 isoform that does not do so due to the lack of a functional intracellular kinase domain required for downstream signal transduction (35). As the TrkB.T1 receptor maintains the capacity to bind BDNF, competitive binding of BDNF to Truncated TrkB.T1 isoform serves to limit the availability of BDNF for the full length TrkB.FL isoform, and this quenching of available BDNF through TrkB.T1 modulates TrkB.FL-dependent intracellular signaling (34). In the ENS-specific conditional TrkB mutant model mediated by excision of exon 15 (36, 37), we observed a significant reduction in expression of transcripts encoding the TrkB.FL isoform, while abundance of transcripts encoding TrkB.T1 remained unchanged. As a result of the isoform proportion shifting even more strongly in favor of the truncated isoform, the BDNF’s downstream signaling through TrkB.FL is further reduced, leading to the observed slowing of intestinal motility.

The Wnt1-cre transgenic model deployed by us to drive the conditional mutation of Ntrk2 gene is effective for the neural crest-derived neurons and glial cells of the adult enteric nervous system. We have recently shown that only half of all neurons in the adult murine small intestinal ENS are derived from the Wnt1-expressing neural crest (54). This suggests that the conditional mutation of Ntrk2 is active only in the neural crest-derived subset of ENS neurons. Thus, it is plausible that if TrkB is expressed by mesoderm-derived enteric neurons, their expression will remain unchanged. Yet, the fact that we observed a significant reduction in abundance of TrkB.FL encoding transcript and significant GI dysmotility in the Wnt1-cre:TrkB^fl/fl mice shows the relevance of TrkB expression in the neural crest-derived neurons to GI motility. Notably, exon 15 is the last shared exon of TrkB.FL and TrkB.T1 transcripts (55). While the specific effect of this knockout on TrkB.T1 isoform is unclear, the lack of transcriptional response and its functional role as a dominant negative inhibitor of BDNF suggest that the slowing of intestinal motility is due to loss of full-length TrkB.FL isoform in neural crest-derived neurons of the ENS. Furthermore, the loss of TrkB.FL in neural crest-derived enteric neurons did not change the overall neuronal numbers, or the expression of pan-neuronal marker Uchl1 (coding for PGP9.5). It also did not change the expression of important neuronal genes Nos1 and Chat, suggesting that the overall structure or the substructure of ENS is not altered in this model.

The choice of the Wnt1-cre line for testing the effect of conditional Ntrk2 mutation on GI function was also necessitated by the fact that there is currently no available transgenic cre line specific to a large proportion of ENS cells. Other transgenic lines that have been used by other investigators include the Hand1-cre mouse line (26) that can label non-ENS cells in the gut wall, or the Baf53b-cre line (56) that will also cause loss of TrkB isoforms in central nervous system (57), thus making it difficult to segregate the central nervous system’s effects from the local ENS effects. Future studies would require us to identify the nature of the ENS neurons expressing Bdnf and Ntrk2 and utilize inducible cre systems to interrogate how loss of these genes by specific neuronal subpopulations in the ENS affects GI function.

Alterations and mutations in Bdnf and Ntrk2 have been associated with GI dysmotility (58, 59, 60), and exogenous supplementation of BDNF has been observed to have a prokinetic effect in patients (11, 15). However, the cellular and molecular mechanisms that regulate BDNF levels in these diseases of gut-brain interactions (DGBI) that are often associated with aberrant stress responses were unknown. Our study, which identifies the neuronal nature of BDNF and TrkB, characterizes the major Bdnf isoforms in the LM-MP tissue at three different ages, and studies the effect of stress and GR signaling on Bdnf isoform abundance, is the first to establish the mechanism through which this clinically relevant biomolecule is altered in the gut. Previously, it has been shown that BDNF amplifies the responsiveness of enteric neurons to stimuli (14), suggesting that this neurotrophin regulates the ‘gain’ of TrkB-expressing neurons. Here, using transgenic TrkB-GFP mice, we have estimated this population to be ∼16% of all adult small intestinal Hu⁺ myenteric cells. With our data showing that increasing TrkB signaling in DEXA -treated mice improves GI motility, this suggests that stress-mediated downregulation of BDNF reduces the sensitivity or responsiveness of TrkB-expressing myenteric neurons, leading to dysmotility. Together, this identifies that TrkB is an important therapeutic target for the amelioration of GI dysmotility in DGBI patients.

Thus, by studying an important mechanism through which stress-associated changes in glucocorticoid signaling alters the BDNF–TrkB crosstalk in the adult ENS, our study identifies an important pathway through which stress causes GI dysmotility and provides a putative therapeutic target which can be used for improving GI motility in many DGBI patients.

Experimental procedures

Human tissue procurement and immunohistochemistry

Paraffin sections of the adult small intestinal full-thickness gut from deidentified human tissues were obtained from the Department of Pathology at Beth Israel Deaconess Medical Center, and the Institutional Review Board approved using the deidentified human tissues. The tissues were obtained from adult patients without any known motility disorder, and the deidentified nature of specimens preclude us from providing further identifying information. Antigen retrieval was performed on 5 μm thick sections at 97 °C for 40 min using citrate buffer at pH 6.0, followed by immunostaining with anti-BDNF antibody (abcam, Cat. No. ab108319, dilution 1:500). The tissue sections were then counterstained with DAB chromogen staining, mounted with Cytoseal, and imaged under 40X brightfield. Microscopy was performed using the Nikon Eclipse Ni-U Upright microscope with 40× Nikon Plan Fluor objective (NA: 0.75). Images were acquired using the Nikon DS-Fi3 camera and NIS-Elements BR software (Version 5.20.01; https://www.nisoftware.net/NikonSaleApplication/downloads/CurrentReleaseInstallFiles/NIS-Elements_HC_AR_BR/6_20_02_b2065_Elements_Installation_Utility.zip). During image acquisition, exposure time was adjusted using the Color DS-Fi3 settings to obtain a clear definition of tissue structure. Images were acquired in the TIFF format.

Animals

We used male and female WT C57BL/6J mice from the National Institute on Aging colony (at the ages of 1, 6, and 17 months) for the bulk RNA sequencing experiment. For all other experiments, except for the immunostaining to quantify proportions of BDNF⁺ or TrkB⁺ neurons, we used adult male WT C57BL/6J mice (between the ages of 12 and 20 weeks) from the National Institute on Aging colony. For experiments to qualitatively and quantitatively assess nature of BDNF-expressing and TrkB-expressing neurons, we used 2 to 4 months old male TrkB-GFP mice (kind gift from Dr David Ginty, Harvard), which have been previously used to observe TrkB-expressing neurons in the myenteric plexus (29). For testing the effect of loss of TrkB (Ntrk2) in the ENS on GI function, Wnt1-cre mice, which have been previously used by us to label all neural crest-derived neurons and glial cells in the adult ENS, were crossed with TrkB^fl/WT mice (Jax Strain #022363, kind gift from Dr David Ginty, Harvard University) to generate the Wnt1-cre:TrkB^fl/fl (cause a conditional loss of TrkB from neural crest-derived neurons) and Wnt1-cre:TrkB^WT/WT (control) mice. Mice were housed in a controlled environment under a 12 h light/dark cycle at 22 ± 2 °C, 55 ± 5% humidity with access to food and water ad libitum. All animal procedures were carried out strictly under protocols approved by the Animal Care and Use Committee of Johns Hopkins University and of Beth Israel Deaconess Medical Center in accordance with the guidelines provided by the National Institutes of Health.

Bulk RNA-seq experiments

Tissue isolation for bulk RNA-seq

Mice of three ages (1-month, 6-months, and 17-months) and both sexes (n = 3 per unique condition) were anesthetized with isoflurane and sacrificed by cervical dislocation. For small intestinal tissue isolation, a laparotomy was performed, and the entire small intestine was removed and flushed with ice-cold nuclease-free 1X PBS. The ileum, defined as the distal third of the small intestine, was cut into 3 cm sections for peeling. Next, tissues were placed over a sterile plastic rod and a superficial longitudinal incision was made along the serosal surface and the longitudinal muscle containing myenteric plexus (LM-MP) was peeled off from the underlying tissue using a wet sterile cotton swab. Each gut muscle strip was immediately flash frozen in liquid nitrogen before storage at −80°C. The average time from sacrifice to flash freeze was 10 min. Animals were processed in three batches, balanced across age/sex composition and order of processing.

Library prep & sequencing

RNA isolation, library preparation, and sequencing were completed at the JHU Single Cell and Transcriptomic Core. Samples with RIN ≥ 7.0 were used for library preparation, with all samples being high-quality except one 17-months female (resulting n = 2, average RIN of 17 samples that passed this threshold > 9.0). Frozen tissue samples each in 1.5 ml tubes were submerged in 375 ul of RLT buffer with β-mercaptoethanol. Tissue and buffer were transferred to an ice-cold 2 ml Qiagen sample tube with a 5 mm stainless steel bead and lysed in a TissueLyzer at full speed for 3 min. RNA isolation was completed with the Qiagen RNeasy Mini, Animal Tissue and Cells with DNA Digest protocol. cDNA libraries were prepared with the TruSeq Stranded mRNA Library Prep with unique dual indexes. Paired-end (2x-50 bp) sequencing was completed on a NovaSeq 6000 system.

Preprocessing

Isoform-level transcript abundances were estimated by pseudoalignment via Kallisto ’quant’ (version 0.48.0) to the GENCODE vM27 reference genome assembly. Default parameters were used, except the number of bootstraps was set to 100 to estimate uncertainty (“-b 100”). The median fragment length, as estimated by Kallisto, was 168.7 base pairs. Across all sequenced samples, the median number of reads per sample was 46,756,241 (range: 34.6 M to 58.2 M) and 94.2% of reads were successfully pseudo-aligned to the reference.

Quality control & analysis

For quality control analyses, isoforms were collapsed to gene-level and abundances were normalized by the “scaledTPM” option within tximport, and further analyzed with DESeq2 (v1.30.1, R v4.0.2). Genes were removed from the analysis if they had fewer than 10 counts across all samples or were not detected in at least three samples. At the sample level, PCA and Cook’s distance were used to identify potential outliers. Both approaches identified one 17-months male sample as a source of outlier expression, and this sample also had the lowest RIN (7.0). With this outlier sample removed, isoform-level abundances of the remaining 16 samples were analyzed with IsoformSwitchAnalyzeR (v2.1.2, R v4.2). Isoform fractions were calculated independently for each age group. Counts assigned to each Bdnf isoform were divided by the sum of all Bdnf counts across all samples of that age.

Restraint stress model

Adult male mice were divided into two cohorts, control and stressed. Mice in stressed cohorts were restrained for 1 h every day for 5 consecutive days and were then released back into their cages. The timing of the restraint during the day was varied across the 5 days to reduce habituation. By contrast, the mice in the control group were handled similarly, but were not restrained and were returned to the cage immediately after handling. The WGTT of the mice in response to treatment was assayed on the day after the last day of restraint. In a separate cohort of mice, 2 h after the last restraint stress, the animals were sacrificed, and their small intestinal LM-MP tissues were isolated and snap frozen to analyze changes in expression of various Bdnf isoform transcripts.

DEXA induced model of GI dysmotility

Adult mice were divided into two cohorts, control and DEXA. Mice in the control group were treated with vehicle (10% DMSO in saline, intraperitoneal route) and those in the DEXA group were treated with DEXA (MedChemExpress, Cat. No. HY-14648; 5 mg/kg body weight in 10% DMSO, intraperitoneal route). DEXA is light sensitive so was shielded from bright light, both in powder form and in solution. Mice were first housed in individual cages and after a habituation phase lasting 30 min, they were treated with vehicle or DEXA, which were administered intraperitoneally using a sterile 1 ml insulin syringe. The mice were subjected to WGTT assay 30 min after the treatment.

Another 2 cohorts of mice were similarly used, wherein the control cohort was dosed similarly with vehicle and the DEXA group was dosed similarly with DEXA. 4 h after the treatment, the mice were euthanized, and their small intestinal LM-MP tissues were isolated and immediately snap-frozen in liquid nitrogen.

Effect of TrkB agonist HIOC on DEXA -induced model of GI dysmotility

HIOC (N-[2-(5-hydroxy-1H-indol-3-yl)ethyl]-2-oxopiperidine-3-carboxamide) is a known specific agonist of TrkB. Adult male mice were divided into four cohorts, Vehicle control (No DEXA + No HIOC), HIOC alone, DEXA alone and DEXA + HIOC. Mice were first housed in individual cages and then those in the HIOC alone and DEXA + HIOC group were treated with HIOC (MedChemExpress, Cat. No. HY-101446; 50 mg/kg body weight in 10% DMSO, intraperitoneal route), while those in DEXA group were treated with equal volume of 10% DMSO in saline as a control for HIOC (intraperitoneal route). The mice were then released back to their individual cages for 30 min after which both groups were treated with DEXA (5 mg/kg body weight in 10% DMSO, intraperitoneal route) and again released back to their individual cages for an additional 30 min. Mice in the Vehicle control and HIOC alone groups were not treated with DEXA or its control, but were either treated with HIOC (in the Vehicle control group; 50 mg/kg body weight in 10% DMSO, intraperitoneal route) or with equal volume of vehicle 10% DMSO (in the Vehicle control group). Treatment was administered intraperitoneally using a sterile 1 ml insulin syringe.

The mice were subjected to whole gut transit time assays 30 min after the DEXA treatment.

WGTT

These experiments were started between the times of 8 and 9 AM. Mice were individually caged in plastic cages without bedding and were left undisturbed 30 min. Water was provided to the mice before and during the experiment ad libitum. All mice received oral gavage of 300 μl of 6% (w/v) carmine red (Sigma C1022) in 0.5% (w/v) methylcellulose (Sigma M0512) in sterile saline. The oral gavage was performed such that the entirety of the dye suspension was delivered into the animal’s stomach. The mice were then left undisturbed in their individual cages for 70 min after which their cages were checked every 15 min for the presence of red colored fecal pellets. The time post-gavage for every mouse to produce a red colored fecal pellet was measured, and the difference between the gavage time and red fecal pellet production time was established as the whole gut transit time for that mouse. The experiment was terminated at 250 min post-gavage and the WGTT of any mice that did not expel the red dye at the termination was marked at the value of 265 (i.e. 250 + 15) min. The mean difference in WGTT (in minutes) between cohorts was analyzed statistically.

Mouse small intestine LM-MP tissue isolation for non-sequencing experiments

Mice were anesthetized with isoflurane and sacrificed by cervical dislocation. Mice were placed in dorsal recumbency on the surgical surface and the skin was disinfected with 70% EtOH before opening the abdominal cavity and collecting the small intestine into a Petri dish containing PBS with 1× Pen-Strep. Sterile PBS solution was flushed through the intestine using a 20 ml syringe to remove fecal matter. Entire small intestine was cut into 2 cm pieces into a sterile Petri dish containing Opti-MEM medium with 1X Pen-Strep. small intestine segments were placed over a sterile 1 ml pipette and LM-MP tissue was peeled off from the underlying tissue using a wet sterile cotton swab. Tissues isolated for qPCR-based assessment of Bdnf levels were snap frozen immediately after isolation. Tissues isolated for culture were placed in a fresh sterile ice-cold Opti-MEM medium for further use. Tissues isolated for immunostaining experiments were flattened, fixed in freshly prepared ice-cold 4% paraformaldehyde solution for 5 min (using our established protocol (54, 61)) and then stored in 1X sterile PBS for subsequent processing.

In vitro LM-MP tissue culture and treatments

Isolated LM-MP tissues were incubated in standard tissue culture conditions of 37 °C and 5% CO₂, in a sterile 24 well dish containing Stem cell media (made up of Neurobasal medium containing L-glutamine, B27, and Pen-Strep (all Invitrogen) and Bovine Serum Albumin (Sigma) (38)) and treated with Vehicle (10% DMSO in saline) or DEXA (5 μM in DMSO) for 4 h. After the incubation, tissues were snap frozen in liquid nitrogen & stored at −80 °C for further experimental analysis. To assess how soon after DEXA exposure does the expression of Bdnf IV and VI isoforms significantly vary, we cultured freshly isolated LM-MP in stem cell media that either contained DEXA (5 μM in DMSO) or vehicle ( DMSO) for 1-h or 2-h. Immediately after the culture, the tissues were snap frozen and used for subsequent RNA-based analyses.

RNA isolation and quantitative real-time PCR

Total RNA was extracted from frozen LM-MP tissues with Direct-zol RNA miniprep kit (Zymo research) and 1 μg RNA was converted to cDNA using SuperScript III (Thermo Fisher Scientific, Cat.No.11752050) in a 20 μl reaction. Quantitative real-time PCR was performed on a QuantStudio3 real time PCR system (Life Technologies) with TaqMan Fast advanced qPCR Master Mix (Cat. No. 4444557, Thermo Fisher Scientific) using gene specific TaqMan assay probes. Mouse Hprt served as a housekeeping gene for normalization. Primer and probe information is provided in Table 1. 20 μl qRT-PCR reactions were incubated at 95 °C for 10 min, followed by 40 cycles at 95 °C for 10 s, 60 °C for 10 s, and 72 °C for 40 s. Relative fold changes of genes between treatment groups were calculated using the 2^−ΔΔCt method. Primer information is provided in Table 1, and primers specific to TrkB.FL and T1 were obtained from Tessarollo Lab (55).

Protein isolation and BDNF ELISA

Proteins from frozen LM-MP were extracted using a cell lysis buffer containing RIPA buffer (150 mM NaCl, 50 mM Tris-HCl pH 8.0, 1% NP-40, 0.1% sodium deoxycholate and 0.1% SDS, adjusted with distilled water). Succinctly, for every LM-MP peel, we add 20 μl of 1X RIPA+++ buffer which contains RIPA with Halt Protease Inhibitor (Invitrogen, Cat. No. 78438) at a conc. of 5-10X and Phosphatase Inhibitors 2 and 3 (Sigma Aldrich, Cat. Nos. P5726 and P0044) at a concentration of 2 to 3% to locking sterile Eppendorf tubes and then add 6 to 8 sterile autoclaved silicon beads. Tissue was homogenized using a bead beater in cold room at 7 to 8 speed setting for 5 min, after which the tubes were shaken at 200 to 250 rpm for 30 to 40 min in cold room, and centrifuged at over 12,000 rpm for 20 min at 4 °C. Supernatant thus generated was collected to be used for ELISA. The levels of BDNF in the LM-MP tissues cultured with and without DEXA were measured using the BDNF DuoSet kit (R&D Systems; DY248). In brief, sterile and clean ELISA-grade 96-well plates (Corning) were coated with 100 μl of diluted captures antibody/well and incubated overnight at room temperature. On the second day, the plates were washed and blocked with reagent diluent (300 μl for 1 h at room temperature and 100 μl of protein extract, standards in reagent diluent, or appropriate diluent were added to the prepared wells and incubated for 2 h at room temperature. Following this, 100 μl each of detection antibody, working dilution of streptavidin-HRP, substrate solution, and 50 μl of stop solution were sequentially added to each well. The optical density of each well was measured using a microplate reader set to 540 nm. The BDNF concentration was expressed in picograms of BDNF per microgram of protein.

Serum collection for ELISA assay

Whole blood was collected by cardiac puncture after animals were euthanized. For serum separation, blood samples were allowed to clot at room temperature for 1 h and then centrifuged at 1000 g for 10 min at 4 °C. The serum was carefully transferred into fresh 1.5 ml collection tubes and stored at −80 °C until further use.

Mouse corticosterone ELISA

Mouse serum corticosterone concentrations were measured using mouse corticosterone ELISA kit (Elabscience, E-EL-0161) according to the manufacturer’s instructions. Briefly, 50 μl of serial diluted standards or serum samples (1/10 diluted) and 50 μl of biotinylated detection antibody were incubated in precoated corticosterone ELISA plate for 45 min at 37 °C. The ELISA plate was washed three times with wash buffer, and then incubated with HRP conjugate for 30 min at 37 °C. Subsequently, it was washed 5 times with wash buffer, and then incubated with 90 μl of substrate reagent for 15 min at 37 °C. The reaction was terminated by adding 50 μl of stop solution, and absorbance was measured at 450 nm using a microplate spectrophotometer (Molecular Devices, SpectraMax iD3).

Mouse DEXA ELISA

Mouse serum and small intestine tissue DEXA levels were quantified using a mouse DEXA ELISA kit (Elabscience, E-FS-E009) according to the manufacturer's instructions. In addition to the serum isolation, 2 cm of mouse full thickness ileal tissue were dissected and homogenized in lysis buffer (∼100 mg/ml) using glass beads in a beadruptor for 15 min. The homogenate was then centrifuged at 12,000 g for 15 min at 4 °C. The supernatant was transferred to fresh 1.5 ml collection tubes. A total of 100 μl of serially diluted standards or samples (tissue extract or 1:10 diluted serum) were added to precoated DEXA ELISA plates, along with 50 μl of antibody working solution, and incubated for 30 min at 25 °C in the dark. The plate was washed five times with wash buffer, followed by a 30 min incubation with 100 μl of HRP conjugate at 25 °C. After washing, the color reaction was developed by adding 50 μl each of substrate reagents A and B, and absorbance was measured at 450 nm using a microplate spectrophotometer (Molecular Devices, SpectraMax iD3).

Immunostaining and imaging murine small intestinal LM-MP tissues

Fixed LM-MP tissues were blocked and permeabilized with 5% normal goat serum and 0.5% Triton-X in 1X sterile PBS for 30 min at room temperature. For assessing the co-expression of GR with all enteric neurons, we performed immunostaining of adult C57BL/6 NIA mice with anti-GR (Nr3c1) antibodies (Cell Signaling: CST3660T, 1:1000) and ANNA-1 patient serum containing anti-Hu antibodies (1:1000). For assessing proportions of BDNF-expressing and TrkB-expressing neurons, we immunostained LM-MP tissues from TrkB-GFP mice with antibodies against BDNF (abcam, 1:500), against GFP (Aves Labs, GFP1020, 1:750), and against Hu using ANNA-1 antiserum containing anti-Hu antibodies. After counterstaining with appropriate secondary antibodies and nuclear dye DAPI, the tissues were mounted with Prolong Gold Anti-fade mountant (Invitrogen) and imaged under Leica Stellaris Confocal microscope using 63X PL APO oil immersion objective (NA: 1.40) with z-step set at 0.85 μm. The images were analyzed with Fiji (https://fiji.sc).

Statistics

Statistical analysis to assess differences between group means was conducted using Student’s t test in Prism 10 software (https://www.graphpad.com/).

Data availability

All the transcriptomic data has been deposited in GEO GSE284108 and the code for analysis is available at https://github.com/jaredslosberg/timecourse_lmmp_bulkrnaseq/.

Supporting information

This article contains supporting information.

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.

Reviewed by members of the JBC Editorial Board. Edited by Roger Colbran

Supporting information

Figure S1.