Gut microbiota of post-cholecystectomy diarrhea (PCD) patients contributed to diarrhea in humanized microbiome mice. Initially, 20 fecal samples from patients were selected and subjected to 16S rRNA sequence. Decreased α-diversity index (Observed species and ACE index) indicated reduced bacterial richness and evenness in PCD patients, compared to HC and NonPCD patients (Fig. 1A and Supplementary Fig. 1A). Then, obvious discriminations in microbial composition among three groups were revealed by the nonmetric multidimensional scaling (NMDS) model and Principal coordinate analysis (PCoA) (Fig. 1B, C), and this was also confirmed by unweighted pair-group method with arithmetic means (UPGMA) tree (Supplementary Fig. 1B). The dramatically differential fecal microbiota in PCD patients was consistent with our previous results 8.
To investigate whether bacterial changes in PCD patients contribute to diarrhea, humanized gut microbiome mice model (n = 6/donor) was constructed by fecal microbiota transplantation from 20 donors, including 5 HC, 5 NonPCD patients and 10 PCD patients. The experimental process was displayed in Fig. 1D. Dramatic difference of fecal pellets appearance was observed in PCD mice, compared with these from HC and NonPCD (Fig. 1E). The longer and softer feces intuitively implied changed gastrointestinal functions in PCD mice 16. Thereafter, carmine red in methylcellulose solution and fecal water content were applied to evaluate gastrointestinal motility and secretion. The shorter gastrointestinal transit time in PCD mice (Fig. 1F, left panel) indicated augmented intestinal motility, without the length change of whole gastrointestinal tracts (Supplementary Fig. 1C), in contrasted to HC and NonPCD mice. These phenotypes were more obvious in PCD 1, 2, 3, 4, 5 mice (Fig. 1F, right panel). Then, higher absolute fecal water content was found in PCD mice (Fig. 1G). These in vivo results demonstrated that altered gut microbiota from PCD patients contributed to diarrhea.
Then, we sought to investigate which segment of gut contribute to enhanced gastrointestinal motility. Firstly, shortened gastrointestinal transit time and elevated fecal water content (Supplementary Fig. 2A, B) were induced again by mixture PCD fecal bacteria. Then, mice were sacrificed at 4 timepoints (30, 40, 60, 80 min respectively) after carmine solution gavage. Carmine was still transited in small intestine 30 min after gavage and normalized gastrointestinal transit distance was compatible among three groups (Supplementary Fig. 2C, D). At the timepoint of 40 min, the marker was squeezed through ileocecal valve and transited to ascending colon in PCD mice, but it just reached ileocecal valve in HC and NonPCD mice (Fig. 1H). Similarly, red pellets were propelled faster in colon of PCD mice than HC and NonPCD mice at timepoint of 60 min and 80 min after gavage, respectively. The quantified results showed that the colonic propulsion of PCD mice was dramatically greater than HC and NonPCD mice (Fig. 1I). These findings suggested that PCD fecal bacteria primarily facilitated colonic motility, rather than influenced the motility of small intestine.
Tryptophan metabolism was significantly enriched in bacterial predicted function of PCD mice. In light of the irritated colonic propulsion of PCD mice, bacteria in colonic luminal contents were analyzed. The overlapping operational taxonomic units (OTU) Venn diagram displayed that PCD mice possessed the fewest OTU (Fig. 2A), compared to NonPCD and HC mice. Declined microbial diversity and higher beta distance by weighted Unifrac (Fig. 2B and Supplementary Fig. 3A, B) were present in PCD mice. The NMDS and PCoA plot revealed clear segregations of bacterial structure among groups (Fig. 2C, D), which also consisted with UPGMA result (Fig. 2E). Bacterial top 10 phylum abundance in individuals and groups showed over-presented Firmicutes, Verrucomicrobiota and Proteobacteria, but under-presented Bacteroidota in PCD mice (Fig. 2E and Supplementary Fig. 3C, D). These results indicated huge differences of bacterial composition in PCD mice.
Additionally, PICRUSt and Tax4Fun 17 were employed to predict bacterial metagenome functions in metabolism and ecology. Principal coordinate analysis (PCA) analysis indicated divergent cluster of bacterial predicted functions in KEGG pathway level 3 among three groups (Fig. 2F), and tryptophan metabolism et al were significantly enriched in PCD mice (Fig. 2G). After that, these enriched microbial functions were ranked by unsupervised RandomForest analysis, and lysine degradation, tryptophan metabolism and quorum sensing were highlighted (Fig. 2H). Whereas, significance in tryptophan metabolism (Supplementary Fig. 3E) among groups, but not lysine degradation (Supplementary Fig. 3F) was achieved. Since that quorum sensing functions as vital reflector for the inter-relationships among bacteria in microbial system 18 and its potential role was also emphasized in PCD (Fig. 2G, Supplementary Fig. 3G). The remarkably decreased co-occurrence network (Pfdr<0.05, |r| > 0.7) among genera in PCD mice (Supplementary Fig. 4) as well as our previous results in patients 8 indicated that weakened bacterial ecology might be the consequence of decreased microbial diversity. Therefore, we hypothesized that some molecules from tryptophan may underlie accelerated colonic peristalsis in PCD mice.
Elevated serotonin biosynthesis and overexpression of 5-HT receptor in colon of PCD mice. Since 5-hydroxytryptamine (5-HT), also named serotonin, is a fundamental metabolite of tryptophan, which conventionally stimulates intestinal propulsion and segmentation motility 19. The critical role of indigenous fecal microbiota on controlling 5-HT biosynthesis is well documented 15. Thus, we firstly quantified its serum and colon level among three grouped mice and PCD mice displayed approximately doubled circulating and in situ 5-HT (Fig. 3A, B), compared to HC and NonPCD, besides, they were also negatively correlated with gastrointestinal transit time (Fig. 3C, D). As regards this finding, we next compared intestinal enterochromaffin cells (ECs) among three groups, where peripheric serotonin is almost generated. Whereas, compatible colonic mRNA expression of chromogranin A and immunohistochemical (IHC) score in protein level (Supplementary Fig. 5A, B) among groups implied that elevated 5-HT wasn’t derived from over-proliferation of ECs.
Peripheric serotonin metabolism comprising biosynthesis by tryptophan hydroxylase 1 (Tph1) in ECs, reuptake to enterocytes by serotonin selective reuptake transporter (SERT), and eventually catabolism by monoamine oxidase-a (Maoa) 20. Then, these critical intermediates were further evaluated, and upregulated Tph1 with downregulated SERT (Fig. 3E, F) in colon of PCD mice were found, compared to HC and NonPCD, but no significance of colonic Maoa was achieved (Supplementary Fig. 5C). Accordingly, 5-HT metabolism was also examined in small intestine, neither 5-HT level, nor its vital metabolic genes were incompatible among groups (Supplementary Fig. 6A-C). Circulating 5-HT might originate from its overproduction in colon of PCD mice (Supplementary Fig. 6D). Therefore, above results supported that overproduction and down-reuptake of 5-HT accounted for elevated colonic 5-HT in PCD mice, which were regulated by PCD fecal bacteria.
The regulatory role of serotonin on intestinal motility is mediated by 5-HT receptors (5-HTR), of whose family, 5-HT3 and 5-HT4 subtypes in intestinal epithelium and submucosal neurons were primarily responsible for gastrointestinal propulsion and secretion 19. Initially, both colonic mRNA expressions were quantified and our results demonstrated that 5-HTRs were simultaneously overexpressed in colon of PCD mice, in contrast to that of HC and NonPCD mice (Fig. 3G, H), which were also confirmed by increased IHC score (Fig. 3I, J). Intriguingly, those elevated 5-HTRs were mainly located to superficial mucosal epithelium of colons. The 5-HT4R is a G-protein-coupled-receptor and increased downstream signaling cAMP reflected its activation 21, doubled cAMP level in colon and negative association between cAMP level with gastrointestinal transit time (Fig. 3K, L) indicated its activated state. However, neither expressions of 5-HTRs, nor cAMP level in small intestine were significantly changed among three groups (Supplementary Fig. 6E-G), which showed invariant quantity and nonactivated states of 5-HTRs in small intestine. Thus, elevated colonic 5-HT might be a key actuator in accelerating colon motility in PCD mice, which might be mediated by overexpressed epithelial 5-HT3R and 5-HT4R.
The pro-motility effects of PCD fecal microbiota were serotonin-mediated and 5-HT receptors dependent. To boost the confidence of serotonin-stimulated and 5-HTRs-mediated hypermotility in PCD mice, we employed classical Tph1 inhibitor Telotristat ethyl (LX1606) 22 and 5-HTR antagonists alosetron 23, GR113808 24 for 5-HT3R and 5-HT4R blockade respectively. After fecal microbiota transplantation, LX1606, alosetron and GR113808 were administrated to PCD mice (Fig. 4A). The feces of treated mice resembled to those of HC and NonPCD mice, but significantly differ from PCD mice in appearance (Fig. 4B). Subsequently, gut motility was assessed and we found LX1606, alosetron and GR113808 all could extend this index significantly (Fig. 4C), compared to untreated PCD mice. Besides, the elevated fecal water content in PCD mice was partially reserved upon drugs’ administration (Fig. 4D).
To assess the colonic motility among six groups, we retested the propulsive ability of colon 60 min and 80 min after carmine solution gavage. PCD mice exhibited longest transit distance in colon at both timepoints, compared to HC and NonPCD; but it reduced to normal level upon drugs’ administration (Fig. 4E, F), while there was no significance in length of colons among groups (Fig. 4G). These results indicated significant alleviation of accelerated colonic motility in treated mice. Finally, declined cAMP in treated mice indicated sufficient blockade of 5-HTR (Fig. 4H). Taken together, our results confirmed the responsibility of cumulated 5-HT for colonic hyperperistalsis, and demonstrated the therapeutic effects of specifical 5-HTRs blockade on PCD mice.
Microbial metabolites from PCD mice and patients induced 5-HT over-production. To identify the specific contribution of gut microbiota for serotonergic effects, RIN14B cells 25 were incubated with washed fecal bacteria and filtrated solution (Fig. 5A) separately. The compatible 5-HT concentrations in RIN14B cell supernatants co-cultured with washed bacteria and intracellular Tph1 expressions (Supplementary Fig. 7A, B) among three groups indicated the impotency of bacteria in stimulating 5-HT release directly. Nevertheless, when exposed to filtered fecal solutions from PCD mice, excessive 5-HT level with significance was synthesized (Fig. 5B), accompanied with significantly elevated Tph1 expression (Fig. 5C), in contrast to HC and NonPCD. Similarly, these experiments were replicated with washed bacteria and fecal filtrated solutions from 15 donors. Neither 5-HT overproduction in RIN14B, nor the intracellular Tph1 expression could be induced by gut microbiota from 5 PCD individuals (Supplementary Fig. 7C-F). However, when exposed to subjects’ fecal filtrated solutions, the 5-HT concentrations in RIN14B cell supernatant (Fig. 5D), as well as the expression of Tph1 in RIN14B cells (Fig. 5E) increased significantly in grouped PCD patients and individuals. In conclusion, these results proved the responsibility of microbial metabolites for serotonergic effects.
Gut microbial secondary bile acid metabolites stimulated serotonin generation. In accordance, we have elucidated altered fecal BAs metabolism in PCD patients previously 8, but whether changed BAs metabolites contribute to PCD through 5-HT remains unknown. Therefore, fecal BAs metabolism among recipient mice was determined by UPLC/MS (Fig. 5A). The Orthogonal Projections to Latent Structures Discriminant Analysis (OPLS-DA) model indicated dramatically distinctive fecal BAs profiling in PCD mice (Fig. 6A). When analyzing individual BAs metabolites, 11 candidates such as deoxycholic acid (DCA), hyodeoxycholic acid (HDCA) and lithocholic acid (LCA) enriched, but 6 such as cholic acid (CA) reduced in PCD mice (Fig. 6B). Then, the serotonergic effects of these specific metabolites enriched in PCD mice were tested in RIN14B cells. Of these 8 BAs, DCA, HDCA and LCA could double 5-HT level significantly, but taurocholic acid (TCA), chenodeoxycholic acid (CDCA), isoLCA and α-muricholic acid (αMCA) only increase it slightly (Fig. 6C). Furthermore, cumulated 5-HT correspond to overexpressed Tph1 (Fig. 6D) implied the obvious serotonergic effects of DCA, LCA and HDCA. Then, the diarrhea-inducible role of these secondary BAs (DCA, LCA and HDCA) were tested in vivo and we found that they could shorten gastrointestinal transit time for 0.6-fold and elevated fecal water content for 1.3-fold (Fig. 6E, F). Besides, the pro-motility effects on colon of these BAs were found by intracolonic injection of BAs (Fig. 6G, H). Then, in vivo serotonergic effects of them were found by increased colonic 5-HT level and overexpressed TPH1 in BAs-treating mice (Fig. 6I, J), intriguingly, over-expressions of 5-HT receptors with significance were found in colon of BAs-treated mice (Fig. 6K). Therefore, microbial secondary BAs could stimulate 5-HT production in vitro and in vivo.
To further confirm BAs-induced overproduction of serotonin in PCD patients, we profiled fecal BAs metabolism and tryptophan metabolism, and found overabundant BAs metabolites in feces of PCD patients, compared to HC and NonPCD (Supplementary Fig. 8D). The OPLS-DA with good ability displayed an obvious segregation in tryptophan metabolism between NonPCD and PCD patients (Fig. 6L). Then, KEGG enrichment map on tryptophan metabolism pathway to depict the integrative interactions among these metabolites was applied, and serotonin metabolic pathway including tryptophan, serotonin and tryptamine was evidently enriched in feces of PCD patients (Fig. 6M). Besides, the total fecal 5-HT amount was accumulated with marginal significance in PCD patients (Supplementary Fig. 8E), which suggested active serotonin biosynthesis in colon.
Interestingly, these BAs metabolites tested with serotonergic effects were secondary BAs, whose metabolism depended heavily on fecal microbes 13. Thus, we analyzed microbial differences among humanized gut microbiome mice, clostridium spp. and unidentified-Lachnospiraceae encoding 7-dehydroxylases, vital microbial enzymes for BAs synthesis 26, were found overabundant in PCD mice (Supplementary Fig. 9A-D), as well as in feces of PCD patients (Supplementary Fig. 9E, F). Thus, cumulative fecal secondary BAs possibly result from overabundant BAs-transforming bacteria in PCD gut microbiota.
Blocking BAs conjugated TGR5/TRPA1 receptors significantly alleviated PCD gut microbiota-induced diarrhea. DCA, LCA and HDCA are discovered as specific agonists for G protein-coupled bile acid receptor 1 (GPBAR1), or TGR5 27,28, and transient receptor potential ankyrin 1 (TRPA1) enables ECs secrete 5-HT to regulate gastrointestinal motility after TGR5 activation 29,30. Therefore, to verify the diarrhea-inducible roles of bacterial BAs metabolites in PCD, we selectively blocked these two receptors by SBI-115 and HC-030031 31,32. Firstly, we quantified the relative expressions of mRNA for them and found both of them doubled in the colon of PCD mice, compared to that of HC and NonPCD mice (Fig. 7A, B). Then, in situ immunofluorescence (IF) on TGR5 (green) and TRPA1 (red) was applied and elevated fluorescence intensity of them was observed in colon of PCD mice (Fig. 7C, D). These results indicated that over-expressed TGR5 and TRPA1 possibly mediated BAs-induced diarrhea by PCD gut microbiota.
After that, we employed TGR5 antagonist SBI-115 and TRPA1 inhibitor HC-030031 in PCD mice (Fig. 7E). The feces of treated mice resembled to those of HC and NonPCD mice, but significantly differ from PCD mice (Fig. 7F). Then, decreased gastrointestinal transit time was reversed upon these drugs administration in PCD mice, along with lower fecal water content in treated-mice, in contrast to untreated PCD mice (Fig. 7G, H). Besides, the augmented colonic motility of PCD mice was significantly alleviated by SBI-115 and HC-030031 (Fig. 7I, J). Finally, when determining the blocking effects on colonic 5-HT level, we found elevated 5-HT concentration in colon of PCD mice declined in half upon drugs administration (Fig. 7K). Therefore, these evidences proved the pro-motility roles of bacterial secondary BAs metabolites in PCD and implied therapeutical efficacy of TGR5 antagonist and TRPA1 inhibitor on PCD.