Diminished androgen levels are linked to irritable bowel syndrome and cause bowel dysfunction in mice

Functional gastrointestinal disorders (FGIDs) have prominent sex differences in incidence, symptoms, and treatment response that are not well understood. Androgens are steroid hormones present at much higher levels in males than females and could be involved in these differences. In adults with irritable bowel syndrome (IBS), a FGID that affects 5% to 10% of the population worldwide, we found that free testosterone levels were lower than those in healthy controls and inversely correlated with symptom severity. To determine how this diminished androgen signaling could contribute to bowel dysfunction, we depleted gonadal androgens in adult mice and found that this caused a profound deficit in gastrointestinal transit. Restoring a single androgen hormone was sufficient to rescue this deficit, suggesting that circulating androgens are essential for normal bowel motility in vivo. To determine the site of action, we probed androgen receptor expression in the intestine and discovered, unexpectedly, that a large subset of enteric neurons became androgen-responsive upon puberty. Androgen signaling to these neurons was required for normal colonic motility in adult mice. Taken together, these observations establish a role for gonadal androgens in the neural regulation of bowel function and link altered androgen levels with a common digestive disorder.

3 (Invitrogen). Nuclei were counterstained with DAPI in Vectashield mounting medium (Vector Labs H-1200, Burlingame, CA). To quantify AR + neurons, 1.5cm segments of the colon were immunostained as whole mounts for ANNA-1 and AR. Single planar images were captured at the level of the myenteric plexus (6 fields per segment, identified based on ANNA-1 signal alone). Images were coded, randomized, and quantified using ImageJ by investigators blinded to the experimental condition. To quantify smooth muscle thickness in SHAM and ORCH mice, colons were acutely isolated, gently flushed to remove fecal pellets, and then fixed with 4% paraformaldehyde in phosphate buffered saline while cannulated on bamboo skewers, to ensure equivalent luminal distention across all samples.
Cryosections of colonic tissue (14µm) were immunostained for SMA, mounted in Vectashield with DAPI, and imaged on an Olympus BX41 epifluorescent microscope. Twelve images were captured per animal and muscularis externa thickness was measured using SMA to mark boundaries. For human tissue analysis, IHC was performed on de-identified colon tissue sections obtained from the Columbia University Cancer Center Molecular Pathology Tissue Bank. Tissue sections derived from healthy margin tissue of resection specimens from patients 18-50 years of age. All images of immunohistochemical staining in the manuscript are representative of observations made in a minimum of 3 independent subjects per condition.

Gastrointestinal motility and stool composition analyses
Total gastrointestinal transit time, gastric emptying and small intestinal transit were measured as previously described (20). Fecal pellet output represents the number of pellets spontaneously excreted by each animal in 60-min. This and other in vivo motility tests were always initiated at 9am. For ex vivo imaging of colonic motility, colonic contractile activity was recorded and analyzed using Scribble and Matlab scripts, as previously described (17,20). Briefly, this analysis generates spatiotemporal heat maps in which color represents gut width, the X-axis represents the proximal to distal length of the colon, and the Y-axis represents time. CMMCs were defined as contractions that originated in the proximal colon and successfully propagated at least 50% of the length of the tissue. All parameters were measured from three 15-minute video recordings obtained from each of 6 mice per group, and all individual data points are shown in graphs. The organ bath preparation accommodates 2 colons at a time, and SHAM and ORCH colons were imaged in every session in parallel. To determine the rate and time required for relaxation of individual CMMCs, vertical slice analysis was performed at two positions along the x-axis of the heat map: one at 1/3 rd of the length of the colon (proximal slice) and one at 2/3rds of the length (distal slice). This generates line graphs at these two points showing gut width over time. The green lines shown in Figure 3F and 3H represent the vertical slices from the proximal positions and the pink lines represent the distal positions. From these graphs, the rate of relaxation was calculated from determining the change in gut width over the change in time as averaged from two contractions per video. For stool composition measurements, mice were transferred to clean cages without bedding at 9am and stool output was continuously collected for 1 hour. Animals had free access to both food and water during this time. Total mass of stool collected was "wet" weight per mouse. Fecal pellets were then dried at 45°C for a minimum of 48 hours until a stable mass was reached, and then a "dry" weight was recorded. Percent stool water content = [("wet"-"dry")/"wet"]*100).

Smooth muscle contractility
Mice were euthanized with 40mg/kg sodium pentobarbital. The colons were removed and flushed with modified Krebs buffer (mM: NaCl 137, KCl 2.9, CaCl2 1.8, MgCl2 2.1, NaH2PO4 0.4, NaHCO3 11.9, D-Glucose 5.6, pH 7.4). Connective tissue was removed, and 3-4 mm lengths of transverse annular segments of the distal colon were excised ≥1 cm from the anus and massed. Each colonic ring segment was mounted in a chamber of a DMT 620M Multi-Wire Myograph (DMT, Ann Arbor, MI). The Krebs buffer was changed every 15 min (37°C, bubbled at 95% O2/ 5% CO2), and each tissue was equilibrated at 1 gram (g) resting tension for 1 hour. To demonstrate tissue viability, colonic smooth muscle rings were isometrically contracted with three cycles of increasing log concentrations of acetylcholine (ACh) (10nM-1mM) (Sigma Aldrich, St. Louis, MO). Krebs buffer was exchanged, and tension was readjusted to 1g between each ACh cycle to re-establish the baseline tension. After the final ACh cycle, and following three buffer exchanges, all tissues were pretreated with 1uM tetrodotoxin (EMD Millipore, Jaffrey, NH) to mitigate the pro-relaxant effects of the myenteric nerve plexuses and maximize myogenic contractility. After 20 min pretreatment, the colonic rings were contracted with carbachol 1nM-4uM (Sigma Aldrich, St. Louis, MO) to define the EC50 carbachol of each colonic ring.
The buffer was exchanged, and resting tension returned to 1g. Tetrodotoxin pretreatment was readministered before maximal contraction was induced with 200mM KCl.

Gene expression analysis
Gut tissues from post-pubertal male and female FVB/NJ mice (n=5 mice per sex) were isolated, immediately transferred to TRIzol (Invitrogen), and stored at -80°C for processing. RNA was extracted using the standard phenol/chloroform method per manufacturer's instructions and purified using the RNeasy Micro Kit (Qiagen). Reverse transcription was performed with iScript cDNA Synthesis Kit ACCTGGATGAGAAGGATGAG; Rpl19 Reverse (5'-3'): ACCTTCAGGTACAGGCTGTG. Ar expression was normalized to Rpl19 expression and 2 -Delta CT was used to calculate gene expression changes across tissues. Data was analyzed using two-way ANOVA with multiple comparisons.
AR expression in human colonic myenteric ganglia was analyzed using a publicly available bulk RNAsequencing dataset generated from samples obtained by laser capture microdissection (LCM; n = 6-8 samples per sex and 3 technical replicates per sample) (29). Raw data was downloaded from National Center for Biotechnology Information Gene Expression Omnibus under accession number: GSE153202.
Reads per kilobase per million-mapped Reads (RPKM) value of AR (ENSG00000169083) was extracted from this dataset and graphed by sex.

Statistical analyses of non-human data
For comparisons between pairs of means, unpaired Student's t-tests were used. All graphs display mean ± standard error of the mean (SEM) for each condition with each individual data point shown. "n" refers to number of mice per condition, except where specifically noted. A P value < 0.05 was considered significant and denoted with *. P values < 0.005 are noted with ** and < 0.0005 with ***.
For comparisons of means from 3 of more groups, ANOVA was used. For repeated measures over time compared between two groups, 2-way ANOVA and post-hoc Tukey tests were performed. For muscle contractility measurements, muscle force values were normalized to each colonic ring's mass. Sigmoidal dose-response curves were fitted with nonlinear regression, and best-fit parameters (logEC50, top bound, expressed as value ± SEM) between groups were compared using extra sum-of-squares F test in Prism 4.0 (Graphpad, San Diego, CA).

Human IBS study and statistical analysis
Androgen and sex hormone binding globulin (SHBG) levels were measured in serum collected at the baseline visit of a randomized controlled trial designed to evaluate open label placebo as a therapeutic modality in patients with IBS (Clinical Trials.gov NCT02802241). All subjects in this IRBapproved trial were recruited from a single academic medical center in the United States of America.
Written informed consent was obtained from all subjects. Hormone measurements were obtained using liquid chromatography tandem mass spectrometry (LC-MS/MS) assays and SHBG was measured using a chemiluminescence immunoassay. Assays were performed by the Brigham Research Assay Core (BRAC; Brigham and Women's Hospital, Boston, MA), which is certified by the Center for Disease Control's Hormone Assay Standardization Program (HoST). Measurements were made for all 209 individuals with IBS in the study for whom serum samples were available from the baseline visit and 28 healthy controls (HC). Of these, one individual with IBS was excluded due to an indeterminate measurement. Final analyses were based on the remaining 236 subjects (208 IBS, 28 HC) using SPSS (IBM SPSS Statistics version 27, IBM Corp., Armonk, NY, U.S.A.). Supplemental Table 1 reports descriptive data of study subjects, including age and distribution of IBS-subtypes, as means and standard deviations (SD). This study was not powered to detect subtype-specific differences in hormone levels.
Median values and interquartile ranges (IQR) of all hormone levels are reported in Supplemental Table   2, and distributions are illustrated as violin plots in Figure 1, A and B and Supplemental Figure 7.
Because the distributions of the hormone data were skewed, statistical analysis was conducted after transforming this data using a logarithmic scale (with natural log values). This transformation was applied to all measurements except for percent free testosterone because this measure was already normally distributed. T-tests were used to evaluate the differences in mean log-transformed levels between IBS and HC subjects, separately for males and females.
The Irritable Bowel Syndrome-Symptom Severity Scale (IBS-SSS) is a validated 5-question survey used to generate a composite score based on: severity of abdominal pain, number of days with abdominal pain over the preceding 10 days, presence and severity of abdominal distension, satisfaction with bowel habits, and IBS-related quality of life (14). Each of these 5 components is scored on a scale of 1-100 with a maximum composite score of 500. Partial correlation analysis was used to assess the association between IBS-SSS and percent free testosterone. Student's t-test revealed no difference in mean percent free testosterone between male and female IBS patients, so the groups were combined for correlation analysis to increase power, controlling for age and sex (results shown in Figure 1C).

Supplemental Data
Table S1. Characteristics of human study participants. Table S2. Diminished testosterone levels are associated with the diagnosis of IBS. Figure S1. Gonadal androgen deficiency disrupts colonic motility.      Movie S1. Colonic contractile activity is disorganized and less effective at oral-to-anal propulsion of luminal contents in male mice lacking gonadal function. Features of colonic migrating motor contractions (CMMCs) in colons from SHAM and ORCH mice that were acutely isolated and imaged ex vivo, four or more weeks after surgery. CMMCs were defined as contractions that originated in the proximal colon and successfully propagated at least 50% of the length of the colon. All parameters were measured from three 15-minute video recordings obtained from each of 6 mice per group. Each individual data point is shown. The organ bath preparation accommodates 2 colons at a time, and SHAM and ORCH colons were imaged in every session in parallel. P-values reflect unpaired t-tests and error bars reflect standard error of the mean. * Denotes P < 0.05. ns denotes P > 0.05. A. Time in seconds (sec) required for colons to relax from maximal contraction to baseline gut width following a CMMC was greater in colons from ORCH mice (P = 0.034). B. CMMC speed was no different between colons from SHAM and ORCH mice (P = 0.3755). C. Length of propagation of individual CMMCs was greater in colons from ORCH mice (P = 0.0458), compared to SHAM controls, proportional to the longer colonic length in ORCH mice (see Supplemental Figure 5C). D. Duration of CMMCs was no different between colons from SHAM and ORCH mice (P = 0.1740). E. Rate of colonic relaxation following a CMMC was no different in colons from SHAM and ORCH mice (P = 0.4152).

Supplemental Figure 3. Gonadal androgen deficiency does not grossly alter populations of enteric glia, immune cells or interstitial cells of Cajal in the colon.
A and B. Whole mounts of colons from SHAM and ORCH mice immunostained for MHCII (to label muscularis macrophages) and GFAP (to label enteric glia) and imaged at the level of the myenteric plexus revealed no differences. C and D. Whole mounts of colons from SHAM and ORCH mice immunostained for ANNA-1 (to label enteric neurons) and S100B (to label enteric glia) and imaged at the level of the myenteric plexus revealed no differences. E and F. Cross-sections of colons from SHAM and ORCH mice immunostained for MHCII show no major differences in macrophage numbers or localization. G and H. Cross-sections of colons from SHAM and ORCH mice immunostained for CD3 to label T lymphocyte cells show no major differences in T-cell infiltrates. E -H Cell nuclei marked with DAPI. I -L. Whole mounts of colons from SHAM and ORCH mice immunostained for stem cell factor receptor (SCFR), a marker for interstitial cells of Cajal (ICC), and imaged at the level of the myenteric plexus (I, J) and circular muscle (K, L) revealed no major difference in ICC networks. Panels J and L represent the same field of view of the same tissue specimen imaged at two different z-planes (optical sections) to visualize the two types of ICC networks in the same colon. Scale bars in A -L = 50µm. All images are representative of observations made in a minimum of 3 mice per condition. A. Body mass measured in 10 week old mice, four weeks after bilateral orchiectomy (ORCH) or sham surgery (SHAM), was no different between the groups (N = 6 mice/group). B. Mass of seminal vesicles, a tissue well-known to atrophy in the absence of androgens, is markedly lower in ORCH than SHAM mice 4 weeks after surgery (N = 5 mice/group). C. Colon lengths in ORCH mice are 10% greater compared to SHAM mice, 4-6 weeks after surgery (N = 5 mice/group). D. Neuronal density in the myenteric plexus, measured by whole mount immunostaining of colonic segments from SHAM and ORCH mice for ANNA-1, shows that loss of gonadal function does not alter enteric neuronal number five weeks after surgery (N = 7 mice/group). Unpaired t-tests were used to compare pairs of group means. Error bars reflect standard error of the mean. * Represents P < 0.05, ** P < 0.01, *** P< 0.005, and **** P<0.001. Supplemental Figure 6. Conditional genetic depletion of androgen signaling in the peripheral nervous system. A. Body mass measured in 10 week old AR Wnt1KO mice (cKO) and littermate controls (CTRL; Ar flx/Y ) was no different between the groups (N = 8-10 mice/group). "ns" represents P = 0.9545 resulting from unpaired t-test comparing group means. Error bars reflect standard error of the mean. B, C. Androgen receptor (AR) and pan-neuronal marker (ANNA-1) immunoreactivities show that AR expression is evident in both myenteric neurons and surrounding smooth muscle in colons from 10 week-old AR flx/Y control mice (CTRL), but undetectable in enteric neurons of AR Wnt1KO mice (cKO). Arrow marks an AR + neuron in the CTRL colon. Scale bars = 25 µm. Images are representative of observations made in a minimum of 3 mice per condition. Violin-plots illustrating levels of percent free testosterone and sex hormone binding globulin (SHBG) in healthy controls (HC) and patients with IBS. Median value is indicated as bar in center of each plot. P values are for unpaired t-tests comparing means (% free testosterone) or means of log-transformed values (SHBG) between IBS and HC groups. A and B. Males C and D. Females