FXR Signaling-Mediated Bile Acid Metabolism Is Critical for Alleviation of Cholesterol Gallstones by Lactobacillus Strains

We used a lithogenic-diet (LD)-induced CGS model in mice to investigate whether L. reuteri and L. plantarum could prevent the formation of CGS. Mice were fed LD for 8 weeks, and L. reuteri and L. plantarum were supplemented once a day from the beginning of the experiment. The gallbladders of mice fed a normal diet (ND) were filled with clear bile and were without any cholesterol crystals. Round gallstones or stratified crystals and a small amount of sticky bile occupied the gallbladders of mice fed the LD. The incidence of gallstones reached 100%. Only a few leaflet crystals or some cholesterol particles that were invisible to the naked eye and viscous bile were observed in the gallbladders of mice with L. reuteri and L. plantarum treatments, and the incidence of gallstones was markedly reduced (Fig. 1A and B). We used the grading standards developed by Akiyoshi and colleagues to evaluate the gallstones (24). We found that the grade of the experimental gallstones in mice fed the LD was significantly higher than that of mice fed the ND and was decreased noticeably with L. reuteri and L. plantarum treatments (Fig. 1C). Next, we evaluated the lipid profiles in gallbladder bile and calculated the cholesterol saturation index (CSI) to understand the mechanism involved in the reduction of CG formation by L. reuteri and L. plantarum. Compared with that in the ND group, the bile total cholesterol (TC) level was increased significantly in mice fed the LD, whereas the L. reuteri and L. plantarum treatments decreased the TC level significantly. Compared with that in the ND group, the level of phospholipids (PL) was significantly decreased in the bile of LD mice, whereas the L. reuteri and L. plantarum treatments increased the PL level significantly. However, the level of total bile acids (BAs) in bile was increased significantly and did not alter significantly with L. reuteri or L. plantarum treatment (Fig. 1D). The CSI in the LD group was higher than that in the ND group, and L. reuteri and L. plantarum treatments both reduced it (Fig. 1E), suggesting that L. reuteri and L. plantarum treatments could improve the LD-induced cholesterol hypersaturation in bile.

Several lines of evidence suggest that metabolic syndrome (MS) is associated with the morbidity of gallstones (25, 26). We monitored the weights of mice once a week and found that the weights of LD-fed mice were significantly increased compared with those of ND-fed mice. The weight gains were inhibited in mice receiving L. reuteri and L. plantarum treatments (Fig. 1F). We next calculated the ratio of liver weight to body weight in mice. Compared with ND-fed mice, the ratio of liver weight to body weight was significantly higher in LD-fed mice, and L. reuteri treatment could reverse the change significantly. The ratio in L. plantarum-treated mice did not decrease compared with the ratio in LD-fed mice and was significantly different from that of the ND group (Fig. 1G). The significant increase of the ratio in the L. plantarum-treated group was likely because of the body weight loss after L. plantarum treatment (Fig. 1G). It was reported that insulin resistance was closely related to the onset of gallstones (26, 27). Therefore, we further examined the serum glucose and insulin functions of mice by the oral glucose tolerance test (OGTT). We found that the hyperglycemia of LD-fed mice was significantly improved by L. reuteri and L. plantarum treatments. Furthermore, the glucose tolerance of LD-fed mice was severely impaired compared to that in ND-fed mice. It was effectively improved by L. reuteri and L. plantarum treatments (Fig. S1A and B in the supplemental material). These results suggested that the improvements of lipid metabolism and insulin resistance were accompanied by the reduction of CGS formation by L. reuteri and L. plantarum treatments.

LD administration resulted in significant histopathological and biochemical changes in mice. We found that the connective tissue of the gallbladder’s tunica adventitia was thicker and looser in LD-fed mice than in the ND group, and these alterations were reduced by L. reuteri and L. plantarum treatments (Fig. 1H). Compared with that in the ND group, the gallbladder volume increased significantly in the LD-fed mice, and the change was markedly reversed by L. reuteri and L. plantarum treatments (Fig. 1I). These alterations of the gallbladder suggested that L. reuteri and L. plantarum treatments would be involved in CGS prevention by improving the hypertrophy of the gallbladder wall and the physiological function of the gallbladder. Next, we evaluated the improvement of hepatic steatosis and metabolite abnormalities with L. reuteri and L. plantarum supplementations. Compared with the liver parenchyma in the ND group, there were many vacuoles in the liver parenchyma of LD-fed mice, led by the accumulation of lipid droplets. L. reuteri and L. plantarum treatments led to a marked reduction in liver histological damage (Fig. 1J). Oil red O staining was used to further confirm the change of lipids in the liver parenchyma. Compared with that in the ND group, the liver parenchyma of LD-fed mice showed many lipid droplets stained with oil red. The quantity of lipid droplets was significantly decreased by L. reuteri and L. plantarum treatments (Fig. 1J and K). Furthermore, compared with the levels in the ND group, the levels of serum alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), total cholesterol (TC), and triglycerides (TG) were increased in the LD group, whereas L. reuteri and L. plantarum treatments decreased them significantly (Fig. 1L and M). The levels of high-density lipoprotein (HDL) and low-density lipoprotein (LDL) were increased in the LD group compared with the levels in the ND group. The level of LDL was decreased with L. reuteri and L. plantarum treatments. The level of HDL was further increased in L. plantarum-treated mice (Fig. 1M). These data suggested that the oral administration of L. reuteri and L. plantarum could ameliorate the histological damage in the gallbladder and liver and improve the hepatic steatosis and liver lipid metabolism, which is associated with reduction of CGS formation (28, 29).

Bile acids (BAs), as important signaling molecules, participate in the regulation of BA and cholesterol metabolism (30). Most BAs are reabsorbed from the intestinal epithelium after being processed by gut microbes and then entering the enterohepatic circulation (31, 32). In the current study, we measured the total bile acids and the composition of BAs in mouse liver and serum to understand the mechanism involved in the reduction of CGS formation by L. reuteri and L. plantarum treatments. We found that the level of total liver BAs was significantly increased in LD-fed mice compared with that in ND-fed mice and did not change with L. reuteri and L. plantarum treatments (Fig. 2A). The level of total serum BAs was markedly increased in LD-fed mice compared with the level in ND-fed mice and was significantly decreased by oral L. reuteri treatment (Fig. 2B). BA-targeted metabolomics analysis was employed to determine the alterations in individual BAs in the liver and serum of mice. We found that the compositions of BAs in the liver and serum of LD-fed mice were significantly different from those in ND-fed mice (Fig. 2C and D). We analyzed the top 12 BAs in the serum and liver, respectively, and found that the levels of cholic acid (CA) and taurocholic acid (TCA) in the liver were significantly increased in the LD group compared with their levels in the ND group and that there were more marked increases in CA and TCA with L. reuteri and L. plantarum treatments (Fig. 2C). However, the levels of CA and TCA did not increase in the serum of LD-fed mice or L. reuteri- or L. plantarum-treated mice (Fig. 2D). In addition, we found that taurochenodeoxycholic acid (TCDCA), the most potent endogenous FXR agonist, with a 50% effective concentration (EC₅₀) of 17 μM (33), was increased in the liver and serum of LD-fed mice and that L. reuteri and L. plantarum treatments further upregulated the increased TCDCA level induced by LD feeding. There was a study that demonstrated that β-muricholic acid (β-MCA), a major bile acid in mice, functions as an antagonist of the FXR (34). We found that the levels of β-MCA in the liver and serum were significantly decreased with LD treatment, and there were more marked decreases in β-MCA with L. reuteri and L. plantarum treatments (Fig. 2C and D). Furthermore, tauro-α-muricholic acid (T-α-MCA) (50% inhibitory concentration [IC₅₀] = 28 μM) and T-β-MCA (IC₅₀ = 40 μM) are both known as efficient natural antagonists of FXR (34). We found that the levels of T-α-MCA and T-β-MCA were significantly increased in the LD group compared with the levels in the ND group and that their levels were markedly decreased by L. reuteri and L. plantarum treatments (Fig. 2C and D). It has been reported that downregulation of T-α-MCA and T-β-MCA can reverse the inhibition of the FXR and activate the FXR (33, 35). To further demonstrate that the FXR was activated by L. reuteri and L. plantarum treatments, the ratio of total cholic acids (CAs), deoxycholic acids (DCAs), and chenodeoxycholic acids (CDCAs) (containing taurine-conjugated BAs), which were taken as FXR-agonistic BAs, to total MCAs, which were taken as FXR-antagonistic BAs, was calculated. Compared with that in ND-fed mice, the ratio mentioned above was significantly higher in LD-fed mice, and L. reuteri and L. plantarum treatments could further increase the ratio of hepatic BAs (Fig. 2E). In addition, the ratio of serum BAs did not increase in the LD group compared with that in the ND group; however, the ratio was markedly increased with L. reuteri and L. plantarum treatments (Fig. 2G). These results suggest that FXR might have been activated in LD-fed mice and that the activity of FXR was further enhanced by L. reuteri and L. plantarum treatments. It has been reported that taurine-conjugated BAs have higher hydrophilicity and a higher affinity for FXR, which is beneficial to prevent the formation of CGS (33). The total taurine-conjugated BAs, including TCA, TCDCA, tauroursodeoxycholic acid (TUDCA), taurohyodesoxycholic acid (THDCA), T-α-MCA, and T-β-MCA, were significantly upregulated in the liver and serum of LD-fed mice compared with those of ND-fed mice, and there were more marked increases of the hepatic taurine-conjugated BAs with L. reuteri and L. plantarum treatments (Fig. 2F and H). These results suggest that the L. reuteri and L. plantarum treatments might reduce the formation of CGS by increasing the FXR-agonistic BAs, decreasing the FXR-antagonistic BAs, and upregulating the hydrophilicity of BAs.

FXR and the associated signaling molecules were detected to further determine whether FXR signaling was activated in the ileum and liver. We examined the mRNA and protein levels of FGF15 in the ileum and found that the FGF15 expression in the ileum was significantly increased in LD-fed mice compared with the level in ND-fed mice. Furthermore, there were more marked increases of the mRNA and protein levels of FGF15 with L. reuteri and L. plantarum treatments, and the expression of FGF15 in the ileum detected by immunofluorescence was consistent with the mRNA and protein levels (Fig. 3A to C). There were no significant differences in the mRNA and protein levels of FXR in the ileum and liver between the ND and LD groups, and there were still no significant changes after L. reuteri and L. plantarum treatments (Fig. S2A to D).

Furthermore, the mRNA level of liver Shp was significantly increased in LD-fed mice and further increased by L. reuteri and L. plantarum treatments (Fig. 3D). It is worth noting that the mRNA and protein levels of liver fibroblast growth factor receptor 4 (FGFR4), an important receptor in the FXR signaling pathway, were found to be markedly reduced in LD-fed mice compared with the levels in ND-fed mice and that the levels were reversed by L. reuteri and L. plantarum treatments (Fig. 3E and F). These results indicated that the FXR-FGF15-FGFR4 signaling pathway could be impaired with LD treatment and that L. reuteri and L. plantarum administration could repair it. To study the downstream signaling pathway of FXR, we assessed the expression of cholesterol 7α-hydroxylase (CYP7A1), sterol 12α-hydroxylase (CYP8B1), sterol 27-hydroxylase (CYP27A1), and oxysterol 7α-hydroxylase (CYP7B1), which are important catalytic enzymes for the synthesis of BAs from cholesterol (36, 37). Then, we found that CYP7A1, a rate-limiting enzyme of the classical pathway of BA synthesis (38), was insignificantly downregulated in LD-fed mice compared with its level in ND-fed mice and that the level of CYP7A1 was markedly reduced with the L. reuteri and L. plantarum treatments, suggesting that L. reuteri and L. plantarum treatments can inhibit BA synthesis via the classical pathway (Fig. 3G and H). The mRNA and protein levels of CYP8B1, another important enzyme in the classical pathway, showed no significant change with the LD, L. reuteri, and L. plantarum interventions (Fig. S2E and F). Furthermore, the mRNA and protein levels of CYP27A1, a rate-limiting enzyme in the alternative pathway, were not decreased in LD-fed mice compared with the levels in ND-fed mice, and the protein level of CYP27A1 was markedly increased with L. reuteri and L. plantarum treatments, indicating that the alternative pathway was activated in mice with L. reuteri and L. plantarum treatments (Fig. 3G and H). Finally, the expression of CYP7B1 protein increased in the LD group and reversed with L. reuteri treatment (Fig. 3G and H).

Obstruction of the transport of BAs to the gallbladder is an additional risk for the formation of gallstones (39). The multidrug-resistance-associated protein homologs (MRPs) in the basolateral membrane of hepatocytes are responsible for excreting BAs (40). We found that the mRNA expression of MRP4, a transporter that mediates the coefflux of conjugated BAs (41), was significantly decreased by the LD and that the L. reuteri and L. plantarum treatments markedly upregulated it (Fig. 3I). The mRNA level of hepatic MRP3 was significantly increased in LD-fed mice, and there was a more marked increase with L. reuteri or L. plantarum treatment (Fig. 3I). Also, the mRNA level of hepatic multidrug resistance protein 2 (MDR2) was not statistically significantly upregulated in LD-fed mice compared with the level in the ND group and there was a significant increase with L. reuteri or L. plantarum treatment (Fig. 3I). It was suggested that L. reuteri and L. plantarum intervention improved biliary phospholipid secretion (39). However, the expression of MRP2 was not significantly changed in LD-fed mice regardless of L. reuteri or L. plantarum treatment (Fig. S2G). Furthermore, the mRNA and protein expression levels of the bile salt export pump (BSEP) were detected and were found to be downregulated in LD-fed mice compared with the levels in ND-fed mice, and this decrease could be reversed by L. reuteri and L. plantarum treatments (Fig. 3I and J).

To further study the secretion of cholesterol in the liver, we assessed the expression of hepatic cholesterol excretion genes encoding ATP-binding cassette subfamily G members 5 and 8 (ABCG5/8) (42). We first found that the mRNA levels of the ABCG5/8 genes were increased in LD-fed mice compared with the levels in ND-fed mice and that they could be reversed by L. reuteri treatment but not L. plantarum treatment (Fig. S2H), suggesting that the L. reuteri and L. plantarum treatments exerted different effects on cholesterol transport. Next, we found that the mRNA and protein levels of sterol-regulatory element binding protein 2 (SREBP2) and SREBP cleavage-activating protein (SCAP) were markedly upregulated in LD-fed mice compared with their levels in ND-fed mice and that the L. reuteri and L. plantarum treatments significantly reversed these effects (Fig. 2K and L).

Collectively, these data demonstrate that BA synthesis was inhibited through the FXR-FGF15-FGFR4 and FXR-SHP signaling pathways in LD-fed mice and further markedly inhibited by L. reuteri and L. plantarum treatments. Moreover, the alternative pathway was more advantageous than the classical pathway with L. reuteri and L. plantarum supplements. The excretion of bile acid into the gallbladder was promoted by L. reuteri and L. plantarum. Consistently, cholesterol synthesis was inhibited by downregulating the expression of SCAP and SREBP2 (43), and cholesterol efflux was suppressed in the L. reuteri-treated group.

To further validate our hypothesis that L. reuteri and L. plantarum alleviated CGS by rebuilding the gut microbiota, we collected feces from mice for analysis of the V3-V4 region of the 16S rRNA gene with an Illumina platform. We masked L. reuteri and L. plantarum from the data set for the distance calculation to avoid the two probiotic strains contributing to the separation and used analysis of similarities (ANOSIM) to compare the similarities of the bacterial communities between samples at the operational taxonomic unit (OTU) level based on Bray-Curtis analysis. Principal coordinate analysis (PCoA), a visual method used to study the differences between bacterial communities, indicated that the OTUs of the LD group were significantly different from those of the ND group and that oral administration of L. reuteri and L. plantarum shifted the microbial composition of the LD-fed mice toward that of the ND group at the OTU level (Bray-Curtis ANOSIM, R = 0.6668, P = 0.001) (Fig. 4A). The distance between groups was larger than all the intragroup distances, which indicated that the differences between groups were significantly greater than those within the groups (Fig. 4B). At the phylum level, after 8 weeks of feeding the LD, the average read counts of Firmicutes, the most common phylum in most people (44), increased from 5,198 to 14,172. The percentage of Firmicutes also increased significantly in the LD group compared with that in the ND group, and the read counts of Firmicutes in LD-fed mice decreased from 14,172 to 10,706 and 10,307 with oral L. reuteri and L. plantarum administration, respectively; however, the percentages in L. reuteri- and L. plantarum-treated mice did not decrease compared with that in the LD group (Fig. 4C and D). The average read counts and the percentage of Bacteroidetes decreased significantly in the LD group compared with those in ND group, and L. plantarum treatment decreased them significantly (Fig. 4C and D). Additionally, we found that the read count of Verrucomicrobia in the LD group decreased from 2,128 to 249 compared with the count in the ND group, and it increased markedly to 6,720 in L. plantarum-treated mice (Fig. 4C). We took a closer look at the Firmicutes/Bacteroidetes ratio and found that it was significantly increased in LD-fed mice compared with the ratio in ND-fed mice and that the increase was reversed by L. reuteri treatment. Contrary to L. reuteri treatment, L. plantarum treatment did not induce a decrease in the Firmicutes/Bacteroidetes ratio (Fig. 4D).

At the genus level, the Venn diagrams indicated that the proportion at approximately 48.61% of the total gut bacteria was identical among the four groups (Fig. S3A), revealing that the core microbiota remained stable. We also found that two species, from the Roseomonas and Catabacter genera, were specific to LD-fed mice (Fig. S3A). At the family level, the percentages of community abundance of Muribaculaceae and Akkermansiaceae in the community bar plot analysis showed that both were present in lower relative abundances in LD-fed mice than in ND-fed mice (Fig. 4E and F). The abundance of Muribaculaceae recovered slightly with L. reuteri treatment, and that of Akkermansiaceae was greatly increased with L. plantarum treatment (Fig. 4E and F). Muribaculaceae occupies a major proportion of the community in feces of healthy mice and plays a critical role in improving metabolic disorders (45, 46). In recent years, a number of studies have revealed that Akkermansiaceae has an important function in improving metabolic syndrome and reducing intestinal inflammation (47, 48), and our results showed that the reductions in Akkermansiaceae in LD-fed mice were reversed by L. plantarum treatment. Furthermore, we observed a significantly increased abundance of Erysipelotrichaceae in the LD group compared to its abundance in the ND group, which was reversed by L. reuteri and L. plantarum treatments (Fig. 4E and F). The reduction in Erysipelotrichaceae is beneficial to lipid metabolism (49). As the Kruskal-Wallis H test bar plot in Fig. 3F shows, Lactobacillaceae, which was significantly decreased in LD-fed mice, was increased by L. reuteri and L. plantarum interventions. We further identified the differences in Lactobacillus abundances at the species level. The Kruskal-Wallis H test bar plot of Lactobacillus showed that L. reuteri was increased significantly with L. reuteri treatment and that L. plantarum subsp. plantarum was increased significantly with L. plantarum treatment (Fig. S3B). We further examined the species (top 50 abundances) of gut bacteria in mouse feces and found that the abundances of several species were similarly changed in the L. reuteri and L. plantarum treatments. However, L. reuteri and L. plantarum modulated some species differently. For example, Pediococcus acidilactici, Akkermansia, Erysipelotrichaceae, and Erysipelatoclostridium increased in the L. plantarum-treated group but not in the L. reuteri-treated group, while Parasutterella, Desulfovibrio sp. ABHU2SB, and Allobaculum decreased in the L. plantarum-treated group but not in the L. reuteri-treated group (Fig. 4G).

These results indicate that L. reuteri and L. plantarum treatments manipulated the gut microbiota composition at the phylum, genus, family, and species levels in LD-fed mice, leading to increased numbers of metabolism-promoting flora and inducing a shift in the microbiota abundance patterns from LD-induced metabolic disorders to healthy conditions in different ways.

To further investigate the correlation between the abundances of different gut microbiota and the factors related to the occurrence of CGS, Spearman’s correlation analysis was performed between the top 50 most abundant species and families that were changed by the two strains and several CGS-related parameters, including the incidence and grade of CGS (Fig. 5A, Fig. S4A), blood biochemical indexes (Fig. 5B, Fig. S4B), expression levels of FXR pathway-related genes (Fig. 5C, Fig. S4C), and BAs in serum (Fig. 5D, Fig. S4D) and liver (Fig. 5E, Fig. S4E) in all the groups of mice. At the species level, four species and six species were significantly positively correlated with the incidence and grade of CGS, respectively, and four and two species were negatively correlated with them (Fig. 5A). Some opportunistic pathogenic bacteria that were found to be associated with metabolic disorders in previous studies were also found to be positively associated with the incidence or grade of CGS, e.g., Desulfovibrionaceae and Ruminococcaceae (Fig. 5A) (9, 50). We found that Akkermansia, Muribaculaceae, and Muribaculum were negatively correlated with both the incidence and the grade of CGS (Fig. 5A). At the family level, Akkermansiaceae was negatively correlated with both the incidence and the grade of CGS and Lactobacillaceae and Muribaculaceae were negatively correlated with the incidence of CGS, while Desulfovibrionaceae were positively correlated with the incidence of CGS (Fig. S4A). Next, we used Spearman’s correlation analysis to investigate the relationships between several serum biochemical indexes and gut microbiota. The results showed that at the species level, 10 of the 50 species were significantly negatively correlated with serum ALT, AST, or ALP, including bacteria belonging to Akkermansia, Muribaculaceae, Lactobacillaceae, and Muribaculum (Fig. 5B), and at the family level, 10 of the 50 species were negatively correlated with serum ALT, AST, or ALP, namely, Muribaculaceae, Akkermansiaceae, Lactobacillaceae, etc. (Fig. S4B), suggesting that these bacteria might be involved in repairing liver injury during the process of CGS formation. We found that some gut microbes were significantly positively correlated with serum TG, TC, HDL, LDL, and Glu at the species level, including Desulfovibrio, Dubosiella, Roseburia, and Parasutterella (Fig. 5B), and at the family level, Erysipelotrichaceae, Defluviitaleaceae, Atopobiaceae, and others were significantly positively correlated with serum TG, TC, HDL, LDL, and Glu also (Fig. S4B), suggesting that these bacteria might have an impact on metabolic disorders. Furthermore, when we analyzed the correlations between gut microbiota and the mRNA levels of liver Cyp7a1, Cyp7b1, Cyp8b1, Cyp27a1, Fxr, Fgfr4, Shp, Mrp3, Mrp4, Mdr2, and Bsep and ileum Fxr and Fgf15, we found that Akkermansia and Lactobacillus were significantly positively correlated with liver Fgfr4 at the species level (Fig. 5C). The abundance of Enterobacteriaceae was significantly positively correlated with the ileum FXR signaling-related gene ileum Fgf15 and genes related to BA export into bile, including Bsep, Mdr2, and Mrp3, at the species and family levels (Fig. 5C, Fig. S4C). However, Muribaculaceae showed a negative correlation with liver Shp (Fig. 5C, Fig. S4C). These results suggested that the gut microbiota have an impact on the expression profile of genes involved in FXR signaling pathways and BA synthesis and exportation. Finally, we found that several gut bacteria were significantly correlated with the composition of BAs in serum and liver, suggesting that gut microbiota played a vital role in BA metabolism (Fig. 5D and E, Fig. S4D and E). These data demonstrated that the CGS phenotype, serum biomarkers, FXR signaling pathway-related genes, and the composition of BAs were significantly correlated with gut microbiota and that the protective effects on CGS formation exerted by L. reuteri and L. plantarum treatments might be due to their regulating the gut microbiota.

The data presented above indicate a critical involvement of FXR in L. reuteri- and L. plantarum-mediated prevention of gallstones in LD-fed mice. We hypothesized that the beneficial effects of L. reuteri and L. plantarum treatments could be abolished by the inhibition of FXR activation. To test this hypothesis, mice fed the ND or LD and administered the L. reuteri or L. plantarum treatments for 8 weeks were orally administered Z/E-guggulsterone (Z-Gu), a global FXR inhibitor, or glycine-β-muricholic acid (Gly-MCA), an intestine-specific FXR inhibitor (17). Compared to the round gallstones and sticky bile observed in the gallbladders of LD-fed mice, we found only a few leaflet crystals or some cholesterol particles in the gallbladders of L. reuteri- and L. plantarum-treated mice, and the incidence and the grade of gallstones could be reduced by L. reuteri and L. plantarum treatments. However, Z-Gu almost completely reversed these protective effects, and granular stones filled the gallbladder. Gly-MCA also weakened the protective effects in reducing gallstone formation, and several leaflet crystals were suspended in bile. Under the Z-Gu intervention, the quantity and quality of gallstones were higher, and the incidence of gallstones was higher than in the Gly-MCA intervention (Fig. 6A and B), suggesting that it was the global FXR that reduced CGS formation, not the intestinal FXR alone. Then, we calculated the ratio of liver weight to body weight in mice and found that inhibiting FXR by Z-Gu and Gly-MCA treatment could significantly reverse the L. reuteri-induced reduction of the liver weight-to-body weight ratio, but there were no significant changes in L. plantarum-treated mice (Fig. 6C). Hematoxylin and eosin (H&E) staining and oil red O (ORO) staining were performed to examine hepatic steatosis. Consistently, we found that Z-Gu abolished the protective effects of the L. reuteri and L. plantarum treatments and that Gly-MCA slightly blunted these effects (Fig. S5A). The changes in relative lipid content shown by liver oil red O staining were consistent with the tissue slice staining results (Fig. S5B), suggesting that global FXR was necessary for L. reuteri or L. plantarum to improve liver fat metabolism and that intestine-specific FXR played only a synergistic role. In addition, the L. reuteri- and L. plantarum-induced reductions in the volume of the gallbladder and the thickness of the gallbladder’s tunica adventitia were markedly reversed by the Z-Gu and Gly-MCA interventions, and the results showed that inhibiting global FXR, but not intestine-specific FXR, could make the gallbladder wall thicker than LD feeding alone (Fig. 6D and E). A likely reason is that inhibiting global FXR instead of intestine-specific FXR may lead to gallbladder wall thickening.

Serum ALT, AST, and ALP were analyzed, and we found that both the Z-Gu and Gly-MCA interventions reversed the decreases in ALT, AST, and ALP caused by L. reuteri or L. plantarum treatment and that Z-Gu was more effective (Fig. 6F), indicating that inhibiting global FXR may almost completely abolish the protective effects against liver injury from L. reuteri and L. plantarum treatments in LD-fed mice. Moreover, we found that the L. reuteri and L. plantarum treatments effectively reduced serum TC and that the reduction was significantly reversed by the intestine-specific FXR inhibitor Gly-MCA (Fig. 6G). However, the results showed that the reductions of serum TC and TG by L. reuteri and L. plantarum were not significantly affected by the global FXR inhibitor (Fig. 6G, Fig. S5C). Consistent with previous research results (51), the Z-Gu and Gly-MCA interventions abolished the protective effect of L. reuteri and L. plantarum treatments on serum glucose (Fig. S5D). Furthermore, the protein expression levels of FGF15 were detected and were markedly decreased with Z-Gu and Gly-MCA treatments compared with the levels in LD-fed mice with or without L. reuteri and L. plantarum treatments.

Taken together, these data demonstrate that the L. reuteri and L. plantarum treatments ameliorate experimental gallstones at least partly in an FXR-dependent manner.

Male C57BL/6J mice (about 8 weeks old and not siblings) weighing 20 to 22 g were purchased from Shanghai SLAC Laboratory Animal Co. Ltd. (Shanghai, China). All mice were maintained under 12-h light/dark cycles at 22°C with unlimited access to a standard rodent diet and water and were housed under specific-pathogen-free (SPF) conditions. The mice were allowed to acclimate for at least 1 week before experiments. All animal experiments were approved (SYXK 2013-0050) by the Animal Ethics Committee of Shanghai Jiao Tong University School of Medicine, Shanghai, China, and were conducted under the “3R” principle (reduction, replacement, and refinement). A lithogenic-diet-induced mouse model of CGS was used in this study. Mice were fed a standard chow diet (normal diet [ND], containing 0.02% cholesterol) or a lithogenic diet (LD, containing 15% butter fat, 28% casin, 1.5% cholesterol, and 0.5% cholic acid) for 8 weeks (61). To determine the effects of the administration of Limosilactobacillus reuteri strain CGMCC 17942 (L. reuteri herein) and Lactiplantibacillus plantarum strain CGMCC 14407 (L. plantarum herein), 48 mice were randomly allocated into four groups and 12 mice in each group were randomly divided into four plastic cages (n = 12 per group and n = 3 per cage). The mice in the ND group, fed the normal diet, received 200 μL of normal saline, which served as the control, and the mice in the LD group, fed the lithogenic diet, received 200 μL of normal saline for 8 weeks. The mice in the L. reuteri group and the L. plantarum group, which were fed the LD, received 200 μL of a bacterial suspension of either L. plantarum or L. reuteri at a dosage of 10⁹ CFU/day for 8 weeks. Five mice were randomly selected from each group to have serum and fecal samples collected for bile acid analysis and gut microbiota sequencing. Next, to further verify the role of FXR, 48 mice were randomly allocated into 8 groups. Six mice in each group were randomly divided into two plastic cages (n = 6 per group and n = 3 per cage), and the mice in four of the groups were treated as described above. The mice in another two groups were fed the LD and orally administered 200 μL of L. reuteri or L. plantarum suspension along with the intestine-specific FXR inhibitor Gly-β-MCA (10 mg/kg of body weight), and the mice in the final two groups were fed the LD and orally administered 200 μL of L. reuteri or L. plantarum suspension along with the global FXR antagonist Z/E-guggulsterone (10 mg/kg) for 8 weeks (17).

L. plantarum and L. reuteri were incubated with de Man-Rogosa-Sharpe liquid medium (MRS broth; BD, USA) under anaerobic conditions at 37°C for 24 h to reach the early stationary phase. The cultures were centrifuged at 5,000 × g for 5 min at 4°C, and bacterial cells were resuspended in sterilized saline solution (10⁹ cells/200 μL) and stored at −80°C as a stock solution. Each aliquot was thawed for 1 h to room temperature before it was administered to a mouse by oral gavage. To determine the number of bacterial cells, serial dilutions of the bacterial suspensions were inoculated on fresh MRS agar plates under anaerobic conditions.

The mice were treated for 8 weeks after a week of adaptation. The body weight and food intake of the mice were checked once a week (57). At the beginning of the 8th week, fecal samples were collected from the mice and stored in liquid nitrogen for RNA extraction and subsequent quantification of the gut microbiota as previously described (62). At the end of the 8th week, the mice were fasted for 5 h but allowed free access to water. Blood samples of mice were collected by heart puncture and centrifuged at 400 × g for 20 min at 4°C for subsequent biochemical analysis. The animals were sacrificed by exsanguination after being anaesthetized with pentobarbital. The liver, gallbladder, and ileum were harvested and divided into two parts; one part was immediately snap-frozen in liquid nitrogen and the other part was fixed in 4% neutral paraformaldehyde at room temperature for 24 h.

The liver and gallbladder specimens fixed in paraformaldehyde were used for paraffin embedding. Paraffin-embedded sections (4 μm) were stained with hematoxylin and eosin (H&E), and a 20× objective was used over 5 separate fields by two expert liver pathologists who were blinded to the treatment groups to identify the percentages of hepatocytes affected by steatosis. Hepatic lipid accumulation was measured by oil red O (ORO) staining as described previously (63). Frozen liver sections (6 μm) were stained with ORO lipid stain (Abcam, USA), and the percentages of hepatocytes involved in steatosis were assessed by relative lipid content, which was quantified by ImageJ.

Bile acid (BA) concentrations were determined using negative electrospray liquid chromatography-tandem mass spectrometry (LC-MS/MS). Chromatographic separation was achieved on an Acquity ultraperformance liquid chromatography (UPLC) BEH C₁₈ column (100-mm inner diameter, 1.7 μm; Waters Corp.). Serum and liver tissues were accurately measured, ethanol (−20°C) was added for precipitation, and the mixture vortex mixed for 60 s and then centrifuged. For liver tissues, 100-mg glass beads were added and oscillated before centrifuging. The supernatant was filtered through a 0.22-μm membrane before LC-MS/MS. The mobile phase consisted of a mixture of 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B). Gradient elution was applied for the following phases: 0 to 4 min, 25% B; 4 to 9 min, 25 to 30% B; 9 to 14 min, 30 to 36% B; 14 to 18 min, 36 to 38% B; 18 to 24 min, 38 to 50% B; 24 to 32 min, 50 to 75% B; 32 to 35 min, 75 to 100% B; and 35 to 38 min, 100 to 25% B. The flow rate was 0.25 mL/min.

The 40 kinds of BA standards (Yuanye Bio-Technology, Shanghai, China) were prepared in methanol and equally mixed to generate working standard solutions (clear without any precipitation). The working standard solutions were determined by LC-MS/MS, and the total ion chromatography (TIC) analysis of the working standard solutions is shown in Fig. S6. We used a signal-to-noise ratio of 10 (S/N = 10) to determine the limit of quantitation (LOQ) of BA standards and selected the appropriate standard curve concentration to make the concentration curve, as shown in Table S2. The supplier numbers, CAS numbers, and International Chemical Identifier (InChI) keys for the standard analytes, their retention times, and the mass transition ion pairs (m/z) (precursor ion and product ion) are presented in Table S3. Working standard solutions were freshly prepared before using LC-MS/MS. The concentration of the working standard solution was taken as the abscissa and the peak area as the ordinate to examine the linear range and draw the standard curve. The linear regression equation was obtained for each BA (correlation coefficient r of >0.99) and used for performing quantitative analysis on all samples.

The levels of metabolites in bile were measured by enzymatic assays using a microplate reader (BioTek, Winooski, VT, USA). The levels of total cholesterol (TC) and bile acids (BAs) in bile were measured following the instructions in the kit (Jiancheng Bioengineering Institute, Nanjing, China). The phospholipid levels in bile were measured following the instructions in the kit (Wako Pure Chemicals, Osaka, Japan). The cholesterol saturation index calculation (CSI) for gallbladder bile was performed as described previously (64).

Immunofluorescence was used to detect ileal FGF15 in the paraffin-embedded ileum tissue sections. The process of dewaxing, hydration, antigen retrieval, permeabilization, blocking of nonspecific binding, and incubation with antibodies was performed as previously described (62). The sections were imaged with a Leica TCS SP8 confocal microscope (×400).

Total proteins from liver and ileum tissues were extracted as previously described (62). Total proteins (40 μg per lane) were separated by 10% SDS-PAGE and then electrotransferred to nitrocellulose membranes. The membranes were incubated with primary antibodies against FXR (NR1H4, 1:1,000; Abcam, USA), SREBP2 (1:1,000; Abcam, USA), CYP7A1 (1:1,000; Abcam, USA), CYP8B1 (1:1,000; Abcam, USA), CYP7B1 (1:1,000; Proteintech Group, USA), CYP27A1 (1:1,000; Proteintech Group, USA), BSEP (1:200; Santa Cruz Biotechnology, USA), and β-actin (1:1,000; Cell Signaling Technology, USA) overnight at 4°C. After washing with 0.1% Tween 20–phosphate buffer solution (PBST) three times, the membranes were incubated with goat anti-mouse or goat anti-rabbit IRDye 700 or 800 calcofluor white (CW)-labeled secondary antibodies for 1 h at 37°C and imaged with an Odyssey infrared scanner (LI-COR, Lincoln, NE, USA). Quantification was performed using the LI-COR software Image Studio.

Total RNA from liver and ileum tissues was extracted using TRIzol (Invitrogen, CA, USA) homogenization followed by isopropanol incubation as previously described, and the purity of RNA products was determined to be between 1.8 and 2.0 according to the 260/280 ratio. One thousand nanograms of RNA was subjected to reverse transcription (RT) to generate cDNA using a commercial PrimeScript RT reagent kit (TaKaRa, Japan). The synthesized cDNA was used for quantitative PCR (qPCR) to determine the relative expression of targeted genes using gene-specific, intron-spanning primers (Table 1). Hieff qPCR SYBR green master mix (Yeasen Biotech, Shanghai, China) was used to perform qPCR to analyze mRNA transcripts, and all reactions were performed using the ABI Prism 7900HT sequence detection system (Applied Biosystems, CA, USA). The fold changes in the expression of each target gene were compared to the amount of the housekeeping gene β-actin using the cycle threshold (2^−ΔΔCT) method and are represented as the fold change relative to the value for the control group. Each target gene analyzed in the tissues was analyzed in triplicate in experiments that were repeated 3 times.

At the end of the experiment, we randomly selected 5 mice from the 12 mice in each group to obtain stool samples for gut microbiota detection; the stool samples from the other mice were frozen in liquid nitrogen. Mouse stool samples were collected within 15 min of defecation and stored in liquid nitrogen within 1 h. Bacteria DNA was extracted using QIAamp DNA stool kits (Qiagen, CA, USA). The 16S rRNA gene was PCR amplified using primers binding to the V3-V4 regions. The resulting amplicons were extracted from 2% agarose gels, purified using gel extraction (AxyPrep DNA gel extraction kit; Axygen Biosciences, Union City, CA, USA), and then quantified and sequenced on the Illumina MiSeq platform (Illumina, San Diego, USA) with paired-end 300-nucleotide reads. The 16S rRNA sequencing data were analyzed by using the Quantitative Insights Into Microbial Ecology (QIME) platform (version 1.9.1).

The raw 16S rRNA gene sequencing reads were demultiplexed, quality filtered by using fastp version 0.20.0, and merged by using FLASH version 1.2.7 (65, 66) with the following criteria: (i) no contaminant sequences were allowed; (ii) 300-bp reads were truncated at any site receiving an average quality score of <20 over a 50-bp sliding window, with parts of truncated reads shorter than 50 bp and parts of reads containing ambiguous characters both being discarded; (iii) only overlapping sequences longer than 10 bp were assembled, according to their overlapped sequence, with a maximum mismatch ratio of the overlap region of 0.2 and reads that could not be assembled being discarded; and (iv) samples were distinguished according to the barcode and primers, the sequence direction adjusted, exact barcode matching performed, and 2-nucleotide mismatch in primer matching allowed. Operational taxonomic units (OTUs) with a 97% similarity cutoff (67) were clustered using UPARSE version 7.1, and chimeric sequences were identified and removed.

The nonmicrobiome data obtained were analyzed statistically to determine the level of significance using one-way analysis of variance (ANOVA) along with the post hoc Tukey test. Tukey’s post hoc test was performed for data with F at a P value of <0.05 and no significant variance inhomogeneity. Statistical analysis was performed using SPSS 24 (SPSS, Inc., Chicago, IL, USA). GraphPad Prism 7.0 (GraphPad Software, Inc., La Jolla, CA, USA) was used to shape the experimental data, and the data are presented as the mean values ± standard errors of the means (SEM) (n = 5 to 12). A P value of <0.05 was considered statistically significant. For the gut microbiota data analysis, the abundances of various taxons were evenly subsampled before analysis and the taxonomy of each representative OTU sequence was analyzed by RDP Classifier version 2.2 against the 16S rRNA database using a confidence threshold of 0.7 (68). Principal coordinate analysis (PCoA) based on Bray-Curtis dissimilarity was performed to provide an overview of gut microbial dynamics in response to LD and L. reuteri or L. plantarum treatments. Similarities of gut microbiota between samples were compared by ANOSIM and Adonis based on Bray-Curtis at the phylum level. Statistical comparisons of the relative abundances in different groups were analyzed by using the Kruskal-Wallis H test and Tukey-Kramer test (with 95% confidence intervals and P values of <0.05; multiple-testing P values were adjusted with the false discovery rate [FDR]). The relationships between the top 50 most abundant organisms at the species level, the incidence and grades of CGS, and the serum BAs, liver BAs, serum biochemical indexes, and levels of CGS-related mRNA expression were analyzed by Spearman’s correlation analysis, and the read counts of the abundances of various taxons were transformed to the centered log ratio for Spearman correlation analysis (69). The Spearman’s correlation coefficient was |R|≥0.5. P values are as follows: 0.01 < P ≤ 0.05, 0.001 < P ≤ 0.01, and P ≤ 0.001.

Z/E-guggulsterone and glycine-β-muricholic acid were purchased from MedChemExpress (MCE, USA). The primary antibodies against CYP7A1, CYP8B1, and FXR were purchased from Abcam, USA. β-Actin was purchased from Cell Signaling Technology (CST, USA). The primary antibodies against CYP7B1, CYP27A1, and FGFR4 were from Proteintech Group, USA. The primary antibodies against FGF15 and BSEP were from Santa Cruz Biotechnology, USA, and all the other reagents were from Sigma-Aldrich Chemical (St. Louis, MO, USA).

The sequence files and metadata for all samples used in this study have been deposited into the NCBI Sequence Read Archive (SRA) database (accession number PRJNA680858↗).

The sequence files and metadata for all samples used in this study have been deposited into the NCBI Sequence Read Archive (SRA) database (accession number PRJNA680858↗).