Abx-induced depletion of the intestinal microbiota alters the structure and function of the gut in a sex-independent manner
We first examined the effects of depleting the intestinal microbiota on GI morphology and physiology. We treated mice with an Abx regimen (Figure 1a) that effectively reduces the enteric microbiota by over 99%, as we have shown previously [38]. No alterations in body weight were observed over the course of Abx treatment (Figure 1b). Interestingly, anatomical alterations were observed in both sexes in the cecum (Figure 1c) and small intestine (Figure 1d), but not in the colon (Figure 1e), where the absolute number of bacteria are most abundant [39]. Abx-treated mice exhibited larger ceca (Figure 1c) and an increase in small intestinal length (Figure 1d) compared to non-Abx control mice. To determine whether the increased small intestinal length was due to tissue elongation, and not an effect of reduced contractile tone, tissues from both groups were treated with nifedipine, an L-type voltage-gated calcium channel blocker, to induce maximal muscle relaxation. Despite inducing relaxation with nifedipine, small intestinal tissues from Abx-treated mice were still significantly longer than control tissues (Figure S1), suggesting that the intestine lengthens to compensate for the loss of the microbiota.
To evaluate the effects of Abx treatment on intestinal function, we first assessed the fecal pellet output during a novel environment stress test. We also measured the water content in the expelled feces to detect any net alterations in secretion and/or absorption. In both male and female mice, Abx treatment induced higher fecal pellet wet weight compared to the control group (Figure 1f), suggesting an enhanced response to stress in Abx-treated mice. Additionally, an increase in the fecal water content was observed in Abx-treated mice (Figure 1g), suggesting changes in water absorption and/or secretion. Next, we evaluated GI motility using three different approaches. We found that GI motility was slowed in Abx-treated mice, indicated by an increase in whole gut transit time (Figure 1h). To determine the origin(s) of the altered whole gut transit, we performed tests to evaluate small intestinal transit and distal colonic propulsion. Small intestinal motility, assessed by measuring dye migration along the intestine, was significantly reduced in Abx-treated mice (Figure 1i). However, in the distal colon we observed no differences in bead expulsion time between Abx-treated and control groups (Figure 1j). No sex-dependent changes in motility were observed in Abx-treated mice (Figure 1h-j).
Next, we assessed intestinal barrier function by measuring fluorescein-5-6-sulfonic acid (FSA) leakage into serum [27]. Four hours following oral gavage, FSA was measured in the serum as an index of intestinal permeability. Abx-treated mice of both sexes exhibited increased serum FSA compared to control mice, suggesting that depletion of the intestinal microbiota increased intestinal permeability (Figure 1k).
In summary, intestinal microbial depletion altered gut morphology, motility and permeability in both male and female animals.
Indiscriminate loss of enteric neurons is observed after depletion of gut bacteria, triggering enteric neurogenesis in the myenteric plexus
After observing alterations in GI morphology and function following Abx treatment (Figure 1), we sought to determine if these events were associated with neuroanatomical changes in the ENS. Since we observed similar responses in male and female mice regarding Abx-induced alterations in GI function, we performed most of the remaining experiments in male mice. We first evaluated the neuroanatomical changes in two distinct intestinal regions, the distal ileum and proximal colon, analyzing both submucosal and myenteric plexuses (Figure 2 and Figure S2). Strikingly, a loss of HuC/D+ neurons in Abx-treated mice was observed in both the ileum (Figure 2b, 2e) and the colon (Figure 2c, 2f), in the submucosal and myenteric plexuses. This loss of enteric neurons was accompanied by a reduction in the Tuj1+ neuronal fiber density (Figure S3a-b). By evaluating the two major subpopulations of enteric neurons, the nNOS+ nitrergic and the ChAT+ cholinergic neurons, we observed a loss of both the nNOS+ neurons in the myenteric plexus (Figure 2b, 2c; nNOS+ neurons are virtually absent in the submucosal plexus) and ChAT+ neurons in the submucosal and myenteric plexuses (Figure 2e, 2f). To further explore the effect of Abx treatment in subpopulations of enteric neurons, we evaluated the impact of bacterial depletion on calretinin (CALR+) neurons (Figure S3c). After Abx treatment, CALR+ neurons were reduced in the ileal submucosal and myenteric plexuses, and in the colonic submucosal plexus, but interestingly, not in the colonic myenteric plexus (Figure S3d).
Additionally, we analyzed the effects of Abx treatment on enteric glia. We used two approaches to identify the EGC in our experiments (Figure 3a, 3e), a PLP1-eGFP expressing mouse [40], and the immunohistochemical expression of S100B; both well-established EGC markers [41, 42]. We observed a complete overlap in the distribution of PLP1 and S100B expression (Figure S4), supporting their validity as markers of EGC. In PLP1-eGFP mice, we observed no alterations in the number of EGC cell bodies in the mucosa after the Abx treatment in either the ileum or colon (Figure 3b). However, we detected a regional effect of intestinal microbial depletion on EGC in whole-mount preparations. A reduction in the PLP1-eGFP+ area (Figure 3d), and a reduction in EGC number (Figure 3f) were observed in the myenteric plexus in the ileum, but not the colon. The total number of EGC was not altered in the submucosal plexus of either regions (Figure 3f). Together, our data suggest a general loss of neurons in all regions analyzed after Abx-treatment, affecting neuronal subpopulations indiscriminately, and a reduction in the EGC population in the ileal myenteric plexus.
We next tested the hypothesis that the broad loss of neurons in the ENS after Abx treatment would trigger a compensatory response, potentially leading to increased enteric neurogenesis [43, 44]. Sox2 is a transcription factor known as a marker of ENS progenitor cells and is largely expressed by EGC in the adult animal [45, 46]. Thus, we used HuC/D+/Sox2+ double-labeling to evaluate the presence of Sox2-expressing neurons (Figure 4), a putative marker of enteric neurogenesis [45, 46]. We observed that the reduction of HuC/D+ neuronal numbers in the Abx-treated animals was accompanied by reduction in Sox2+ cells in the colonic submucosal plexus (Figure 4b). Surprisingly, we did not detect any HuC/D+/Sox2+ neurons in the submucosal plexus (Figure 4b). In contrast, the decreased number of HuC/D+ neurons in the colon of Abx-treated animals was followed by an increase in Sox2+ cells and HuC/D+/Sox2+ neurons (Figure 4d), indicative of increased neurogenesis in the colon of microbiota-depleted mouse. Further, we tested whether cell proliferation was evident in the ENS by evaluating ileal and colonic sections stained for Ki67, a marker of cell proliferation. As expected, we observed Ki67+ cells at the base of the intestinal epithelial crypts, but we observed no positive staining in the ENS (Figure S5a). Moreover, we tested whether Abx treatment would lead to higher incorporation of 5-ethynyl-2’-deoxyuridine (EdU) into enteric cells, demonstrating de novo DNA synthesis. We did not detect positive EdU incorporation/.staining in the ENS (Figure S5b). Thus, Abx treatment triggered region-specific increased expression of Sox2 in neurons, which is indicative of enteric neurogenesis; however, no markers of proliferation were detected in enteric neurons.
Spontaneous recolonization of bacteria restores altered GI physiology and ENS neuronal loss
We next assessed if the alterations induced by intestinal microbial depletion could be reversed following cessation of Abx treatment. We allowed spontaneous microbial recolonization for 21 days (Figure 5a). To confirm bacterial recolonization, we assessed the bacterial load in the feces of the different groups. Bacterial load in the Abx-withdrawal (microbiota recovery) group was comparable to the control group (Figure S6). The Abx-withdrawal led to a slightly lower body weight as compared to the other two groups by the end of the experiment (Figure 5b). Alterations in gut morphology induced by Abx treatment were normalized following bacteria recolonization (Figure 5c, 5d); Abx-withdrawal mice had reduced cecal weight (Figure 5c) and shorter small intestinal length compared to the Abx-treated group (Figure 5d). No changes in colon length were observed (Figure 5e). The fecal pellet output in response to stress also returned to baseline levels after recolonization (Figure 5f), with a tendency (p=0.10) to recovery in the wet:dry ratio of the feces (Figure 5g).
To further explore the effects of bacterial depletion and recovery on secretion, segments of distal ileum were collected and mounted in Ussing chambers. No differences in ion transport, as measured by changes in short-circuit current (ΔIsc), were observed between Abx-treated and Abx-withdrawal groups when tissues were treated with veratridine, a stimulator of nerve-mediated secretion (Figure 5h, left panel). However, when tissues were treated with carbachol, to directly activate the intestinal epithelium, we observed increased ΔIsc in tissues isolated from Abx-treated mice, which was not observed in the Abx-withdrawal group (Figure 5h, right panel). These results suggest a role for a bacteria-dependent regulation of intestinal ion transport by epithelial cells.
The motility patterns evaluated by three different tests also demonstrated a recovery to initial baseline levels after bacterial recolonization (Figures 5i-k). Lastly, evaluating intestinal permeability, we observed an elevated concentration of FSA in the serum of Abx-treated mice, and this returned to baseline levels in the Abx-withdrawal group (Figure 5l). Here, we also investigated whether the distal ileum may be contributing to the increased gut permeability of the Abx-treated mice. We measured the transepithelial electrical resistance (TER) of the distal ileum in the Ussing chambers. Interestingly, no differences were observed among the three groups (control: 27.17 ± 1.46 Ω/cm2; Abx: 32.18 ± 7.19 Ω/cm2; Abx-withdrawal: 31.39 ± 3.01 Ω/cm2; One-way ANOVA). Overall, spontaneous reconstitution of the intestinal microbiota restored most of the altered physiological parameters induced by Abx treatment.
Next, we sought to determine the effect of the spontaneous bacterial recolonization on neurons and EGC of the ENS (Figure 6). Given the normalization of many of the alterations in GI physiology driven by Abx treatment (Figure 5), and the evidence for increased enteric neurogenesis after Abx treatment (Figure 4), we hypothesized that intestinal microbial recolonization would restore the neuronal population in the ENS. Indeed, we observed a recovery of HuC/D+ neurons in the submucosal plexus and a recovery of HuC/D+ and nNOS+ neurons in the myenteric plexus in the ileum of the Abx-withdrawal group (Figure 6b). A similar tendency was observed in the colon regarding a restoration of HuC/D+ and nNOS+ neurons in the Abx-withdrawal group (Figure 6c and Figure S7). The neuronal recovery was also accompanied by a complete normalization of the S100B+ EGC in the ileal myenteric plexus (Figure 6d, 6e), illustrating the plasticity of the ENS. Moreover, we tested the modulation of neurogenesis after the microbiota recovery (Figure 6f, 6g) and observed the restoration of HuC/D+ neurons as expected, but also a sustained increase in Sox2+ cells and HuC/D+/Sox2+ neurons in the Abx withdrawal group (Figure 6g).
Overall, our data show that the presence of microbial communities in the gut is a major modulator of GI function and ENS integrity in the adult mouse, and that Abx-induced alterations in GI physiology and neuroanatomy of the ENS are largely reversible following spontaneous reconstitution of the microbiota.
LPS supplementation prevents Abx-induced neuronal loss, but does not normalize GI function
Next, we sought to investigate potential mechanisms by which the intestinal microbiota influences GI function and integrity of the ENS. LPS is a component of the cell wall of Gram-negative bacteria and is a known ligand of TLR4, a receptor previously described as important in the regulation of gut motility and ENS neuronal survival [18]. LPS levels are reduced after Abx treatment [18]. Hence, we sought to test whether supplementation with LPS (50 µg/mL) in the drinking water, at a concentration reported to modulate GI motility in mice [47], could attenuate or reverse the changes in GI function and ENS structure (Figure 7a). Briefly, one group received LPS supplementation concomitant with the Abx treatment (LPS group), and a second received LPS supplementation after 14 days of Abx treatment (LPS at d14); a timepoint in which alterations caused by Abx treatment were fully established (Figures 1-3). No differences in body weight were detected among the different groups at the end of the experiment (Figure 7b). Changes in gut morphology caused by Abx treatment, such as an enlarged cecum and longer small intestine, were not reversed by LPS supplementation (Figure 7c-e). Functionally, LPS supplementation had no significant effects on fecal pellet wet weight after novel environment stress or the wet:dry ratio of the feces (Figure 7f-g), nor did it normalize ΔIsc induced by veratridine or carbachol (Figure 7h). As observed in previous experiments, Abx treatment slowed motility (Figure 7i-k), which was not affected by LPS supplementation. Lastly, LPS supplementation had no effect on the alterations in intestinal permeability or TER induced by depletion of the intestinal microbiota (Figure 6i; TER: control: 27.14 ± 2.17 Ω/cm2; Abx: 29.36 ± 2.77 Ω/cm2; Abx + LPS: 32.65 ± 1.14 Ω/cm2; Abx + LPS at d14: 35.85 ± 1.31 Ω/cm2; One-way ANOVA). Taken together, our LPS supplementation regimens (concomitant with Abx administration or after 14 days of Abx treatment), did not normalize the Abx-induced changes in GI structure or physiology.
Although no significant changes in GI function were observed in Abx-treated mice supplemented with LPS, striking neuronal changes were observed (Figure 8). In the ileum, Abx treatment reduced the number of HuC/D+ neurons in both plexuses (Figure 8a, left and middle panels), as observed in our previous experiments. Interestingly, concomitant administration of LPS with Abx prevented the loss of HuC/D+ neurons (Figure 8a, left and middle panels). However, beginning LPS supplementation after 14 days of Abx treatment was not able to reverse the Abx-induced neuronal loss (Figure 8a, left and middle panel). Moreover, a parallel effect was seen on the nitrergic neuronal subpopulation; nNOS+ neurons were reduced in the Abx group compared to control, but this effect was attenuated with concomitant supplementation with LPS (Figure 8a, right panel). Similar trends were observed in the colon (Figure 8b). Both submucosal and myenteric plexuses had a reduction in the number of HuC/D+ neurons in the Abx-treated mice. This effect was attenuated in mice supplemented with LPS during Abx treatment, but not when LPS was given after 14 days of Abx treatment (Figure 8b, left and middle panels). However, the number of nNOS+ neurons were not maintained by concomitant LPS supplementation in the colon (Figure 8b, right panel). Together, these results suggest a role for LPS in neuronal survival, but not in ENS recovery (i.e., neurogenesis).
SCFA supplementation rescues neuronal loss induced by Abx treatment, but has a minor effect on GI function
Finally, we investigated the effects of SCFA supplementation in Abx-treated mice. The SCFA mix (67.5 mM of acetate, 25.9 mM of propionate, 40 mM of butyrate) [48] was administered in the drinking water concomitant with the Abx treatment (Abx + SCFA) or after 14 days of Abx treatment (Abx + SCFA at d14) (Figure 9a).
A slightly reduced body weight was observed in the Abx group by the end of the experiment (Figure 9b). Alterations in gut morphology caused by Abx treatment, such as enlarged cecum (Figure 9c) and longer small intestine (Figure 9d) were not affected by SCFA supplementation. No changes were observed in the colon length (Figure 9e). SCFA supplementation, both concomitant and following 14 days of Abx treatment, reduced the elevated weight of fecal pellet output measured after the novel environment stress (Figure 9f). The wet:dry ratio increase in the feces was not affected by SCFA treatment (Figure 9g). No effects on ion transport in the ileum, nor motility or intestinal permeability were observed in Abx-treated mice supplemented with SCFA (Figure 9h-l; ileum TER: control: 26.77 ± 1.38 Ω/cm2; Abx: 35.84 ± 3.91 Ω/cm2; Abx + SCFA: 33.25 ± 2.98 Ω/cm2; Abx + SCFA at d14: 37.75 ± 2.49 Ω/cm2; One-way ANOVA).
Given the selective effects of SCFA supplementation on the functional parameters altered by Abx-induced intestinal microbiota depletion, we sought to determine if there was an effect on the neuroanatomical alterations in the ENS. Interestingly, a clear effect of the SCFA supplementation was observed in the submucosal plexus of both ileum and colon (Figure 10a, 10b, left panels). A reduced number of HuC/D+ neurons was observed after Abx treatment, which was attenuated in both groups that received SCFA supplementation. In the ileal myenteric plexus, we observed neuronal recovery in mice supplemented with SCFA after 14 days of Abx treatment, but surprisingly, we did not observe a preservation of neurons when SCFA were delivered concomitant with Abx (Figure 10a, middle panel). Additionally, no recovery effect was seen in the nitrergic nNOS+ neuronal subpopulation (Figure 10a, right panel). In the colonic myenteric plexus, mice treated with SCFA at both timepoints had higher number of HuC/D+ and nNOS+ neurons when compared to the Abx group (Figure 10b, right panel). Due to this effect on neuronal recovery observed in the Abx + SCFA at d14 group, we sought to evaluate whether SCFA would also regulate the number of S100B+ EGC in the ileal myenteric plexus, which was the only region where EGC were sensitive to Abx treatment (Figure 3f). Here we observed a similar reduction in the S100B+ EGC numbers in the ileal myenteric plexus, however, no differences were observed after SCFA treatment (S100B+ EGC normalized to control: Control: 100 ± 1.9; Abx: 78.2 ± 4.1 (p < 0.05 compared to Control); Abx + SCFA: 87.2 ± 6.3 (p > 0.05 compared to Abx); Abx + SCFA at d14: 82.6 ± 5.0 (p > 0.05 compared to Abx); One-way ANOVA, followed by Tukey’s multiple comparison test). These results suggest a specific effect of SCFA on enteric neurons.
Considering that SCFA restored the neuronal deficit induced by Abx treatment, but had only a minor effect on GI function, we asked whether a longer treatment regimen (90 days) with SCFA could restore the physiological alterations induced by Abx treatment (Figure 11a). Morphological alterations induced by Abx treatment, such as increased cecum weight and longer small intestine length, were not restored by SCFA treatment (Figure 11b-c). Moreover, long-term SCFA treatment did not restore the impaired motility induced by Abx treatment (Figure 11d-e). We performed the whole gut transit test at two different timepoints, day 42 and day 77, and observed no difference between Abx and Abx+SCFA at d14 group (Figure 11e). Similarly, no changes occurred after SCFA treatment when evaluating small intestine transit (Figure 11d). Interestingly, SCFA restored the Abx-induced increase in intestinal permeability (Figure 11e). These data suggest that SCFA may play a role in regulating barrier function, but their supplementation are insufficient to normalize most physiological functions perturbed by Abx treatment.
Together, we observed a role for SCFA in modulating enteric neurons in a region-specific manner, having the greatest effect in the submucosal plexuses and on colonic myenteric neurons.