The animal experiments were conducted in two different centers (West China Hospital and Chaoyang Hospital) under the same conditions to confirm the reliability of the results. Six-week-old male C57BL/6 mice were obtained from Beijing Vital River Laboratory Animal Technology Corporation Ltd. and the Experimental Animal Center of Sichuan University. All animals were housed in specific pathogen-free (SPF) conditions under constant temperature and humidity with a 12 h light-dark cycle. Standard diet and sterile water were supplied ad libitum during the initial 7-day adaptation period. All animal procedures were approved by the Animal Ethics Committees of Beijing Chao-Yang Hospital and the Animal Ethics Committee of West China Hospital of Sichuan University (Ref. No. 20211277A).

Mice were randomly divided into different groups of 6-10 animals each before the start of the experiment. Colitis was induced by adding 2.5% (w/v) dextran sulfate sodium (DSS, M.W: 36000-50000, MP Biomedicals, Canada) to the drinking water for 7 days, then replaced with sterile water. Tacrolimus (Tacro, Astellas Ireland Corporation Ltd, Shenyang, China) at different dosages (0.1 mg/kg, 1 mg/kg or 10 mg/kg) was administered via oral gavage in sterile water once a day and maintained until the end of experiments. Lactobacillus plantarum 550 (Lacto) in powder form was isolated from pickle and provided as a gift from Technology Research Institute of Shuxi Condiments of Sichuan Cuisine Corporation Ltd (Chengdu, China), 1×10⁸ CFU in 200 μL sterile water/mouse/day was given daily via oral gavage 3 days before DSS administration and maintained until the end of experiments. Mice in the control group were administered an equal volume of sterile water by gavage. Body weight and water consumption were recorded daily. The blood in the stool was detected by Pyramidon semiquantitative assay using a Fecal Occult Blood test kit (BA2020B, BaSO, Zhuhai, China) and recorded as “-”, “+”, “++”, “+++”, or “++++”. The disease activity index (DAI) was evaluated by scoring the body weight loss (%), stool consistency and blood in feces as described in the literature(Jang et al., 2019), with minor modifications ( Supplementary Table 1 ). Mice were sacrificed at the end of the treatment, and the colon lengths were measured. Cecal contents were collected, snap frozen and kept at -80 °C for 16S rRNA sequencing and targeted bile acid metabolomics assays. Colonic mucosa was isolated and frozen until transcriptomics analysis. Sections of the proximal and distal colon were fixed in 4% paraformaldehyde and stained with hematoxylin and eosin (H&E). Histological scores were calculated according to the criteria described in the literature (a score from 0 to 4 was given for each colon segment [proximal or distal] based on the severity of inflammation, and a combined score [proximal and distal] provided a total colonic histological score per mouse) (Liu et al., 2017).

The total genomic DNA was extracted from cecal contents of mice using the TGuide S96 Magnetic Soil and Stool DNA Kit (TIANGEN Biotech Corporation Ltd, Beijing, China) according to the manufacturer’s instructions. The full-length 16S rRNA gene was amplified with universal primers (27F: 5’-AGRGTTTGATYNTGGCTCAG-3’; 1492R:5’-TASGGHTACCTTGTTASGACTT-3’) under the following conditions: 95 °C for 2 min; 25 cycles of 98 °C for 10 s, 55 °C for 30 s, and 72 °C for 90 s; and a final step at 72°C for 2 min. The purified PCR products were sequenced on the PacBio platform by Biomarker Technologies Corporation (Beijing, China). The circular consensus sequencing (CCS) reads were generated from the corrected original subreads by SMRT Link (version 8.0) following the setting parameters: minPasses ≥ 5 and minPredictedAccuracy ≥ 0.9. After barcode and primer identification followed by chimera removal, high-quality reads were obtained and clustered as operational taxonomic units (OTUs) with similarity over 97% using USEARCH (version 10.0). Species annotation and taxonomic and diversity analyses were conducted with QIIME2 (version 2020.06). Chao1 and Shannon indexes were used to reflect alpha diversity, while nonmetric multidimensional scaling (NMDS) analysis was carried out to show the dissimilarity of beta diversity. The vegan package in R language was applied for ANOSIM analysis to measure the significant differences in beta diversity between groups. Linear discriminant analysis effect size (LEfSe) analysis with LDA score > 3 and statistical analysis of metagenomic profiles (STAMP) were performed to identify the significantly different bacteria between groups at the genus level.

The high-throughput targeted quantification of bile acids in mice cecal contents was performed at Biomarker Technologies Corporation (Beijing, China) by UHPLC-MS/MS. Aliquots (25 mg) from 35 samples were weighed and added 1000 μL of precooled extract solution containing 0.1% formic acid and isotopically labeled internal standard mixture (acetonitrile-methanol-water, 2:2:1) was added per tube. After vortexing for 30 s, the mixture was homogenized at 35 Hz for 4 min, followed by sonication for 5 min in an ice-water bath. The latter 9-min circle was repeated three times, and then the tubes were incubated at -40°C for 1 h and centrifuged at 12000 rpm and 4°C for 15 min. The resulting supernatants were collected for UHPLC-MS/MS analysis.

An UHPLC System (Vanquish, ThermoFisher Scientific) equipped with a Waters ACQUITY UPLC BEH C18 column (150 *2.1 mm, 1.7 μm, Waters) was used for chromatographic separation. Mobile phase A included 1 mmol/L ammonium acetate and 0.1% acetic acid in water, and the mobile phase B was acetonitrile. The column temperature was set at 50°C, while the autosampler temperature was maintained at 4°C. The injection volume was 1 μL. Mass spectrometry in parallel reaction monitoring (PRM) mode was carried out by a Q Exactive HFX mass spectrometer (Thermo Fisher Scientific). The ion source parameters were as follows: spray voltage = +3500/-3100 V, sheath gas (N2) flow rate = 40, aux gas (N2) flow rate = 15, sweep gas (N2) flow rate = 0, aux gas (N2) temperature = 350°C, capillary temperature = 320°C.

A total of 39 bile acids were identified in this study (detailed in Supplementary Table 2 ). The calibration standard solutions were diluted stepwise with a dilution factor of 2 before being subjected to UHPLC-PRM-MS analysis. The final concentration (cF, nmol/L) of the samples measured was obtained by multiplying the directly calculated concentration (cC, nmol/L) output from the system by the dilution factor. The targeted metabolite concentration (cM, nmol/kg) in the tissue samples was calculated according to the following formula (VF: final volume of sample; m: sample mass): C M [ n m o l · k g − 1 ] = C F [ n m o l · L − 1 ] · V F [ μ L ] m [ m g ]

Total RNA from mice colonic mucosa was extracted using the mirVana miRNA Isolation Kit (Ambion) following the manufacturer’s protocol. RNA integrity was assessed by an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). Samples with an RNA integrity number (RIN) ≥ 7 were subjected to the subsequent analysis. The TruSeq Stranded mRNA LT Sample Prep Kit (Illumina, San Diego, CA, USA) was used to construct the libraries. Transcriptome sequencing and analysis were conducted by OE Biotech Corporation Ltd. (Shanghai, China).

The libraries were sequenced on the Illumina HiSeq X Ten platform. Approximately 50 megabytes of raw reads for each sample were generated. Raw data were preprocessed using Trimmomatic. After removing the reads containing poly N and the low-quality reads, clean reads were obtained and mapped to the mouse genome (GRCm38.p6) using HISAT2. The fragments per kilobase of exon model per million mapped fragments (FPKM) value of each gene was calculated using Cufflinks, and the read counts of each gene were obtained by HTSeq-count. Differentially expressed genes (DEGs) were identified using the DESeq (2012) R package, followed by gene set enrichment analysis (GSEA). P value < 0.05 and fold change > 2 or fold change < 0.5 were set as the thresholds for significantly differential expression. To demonstrate the expression pattern of genes in different groups and samples, hierarchical cluster analysis of DEGs was performed. R software based on the hypergeometric distribution was employed for Gene Ontology (GO) enrichment and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of DEGs. Immune infiltration analysis was performed using Immune Cell Abundance Identifier (ImmuCellAI).

Data are presented as the mean ± standard error of the mean (SEM). The GraphPad Prism (version 8.0; GraphPad Software, San Diego, CA, USA) was utilized for statistical analyses. Normality of distribution was confirmed using the Shapiro-Wilk test, and differences among groups were assessed using the Student’s t-test or two-way ANOVA when appropriate, with the Tukey’s post-hoc test. The log-rank test was used to compare survival rates. P < 0.05 was considered statistically significant.

Compared to the control mice, body weight loss, increased DAI scores and decreased survival rates were significantly observed in mice orally treated with 2.5% DSS for 7 days ( Figures 1A, B, E ). The dosage of tacrolimus (10 mg/kg) was determined by our previous study, which is the highest tolerated dose that most closely resembled the serum concentration in humans (Zhang et al., 2018). No significant difference in body weight, DAI scores, survival rate or colon length was observed in the tacrolimus and control groups, supporting the safety of tacrolimus at this dosage. Body weight, DAI scores and survival rates in the DSS colitis model were significantly (P<0.0001, P<0.0001, P<0.01, respectively) ameliorated by tacrolimus treatment ( Figures 1A, B, E ). Furthermore, colitis manifestations, including the shortened length, disruption of crypt-villus structure, goblet cell loss and immune cell infiltration, as determined by morphological changes and pathological scores were significantly (P<0.01) ameliorated in the DSS + Tacro group compared to the DSS group ( Figures 1C, D, F, G ). All these results indicated that tacrolimus ameliorated DSS-induced colitis in both clinical manifestation and pathological changes.

Lower dosages of tacrolimus (0.1 mg/kg and 1 mg/kg) were also used in the acute DSS colitis model in our study. Tacrolimus at these lower dosages attenuated the increased DAI scores and improved survival in colitis ( Supplementary Figures 1B, E ), whereas no significant differences in weight change and colon length were observed in the low-dose DSS + Tacro group compared to the DSS group ( Supplementary Figures 1A, C, D ). These data also supported the effectiveness of tacrolimus in the DSS mouse model. For the consistency and effectiveness of the study, tacrolimus was used at a dosage of 10 mg/kg in the following study.

To investigate the effect of tacrolimus on the diversity and composition of the gut microbiome in DSS-treated mice, 16S rRNA amplicon sequencing of DNA extracted from the cecal contents of mice was performed. The α-diversity of gut microbiota was not significantly different among the Control, Tacro, DSS and DSS + Tacro groups, manifested as the Chao1 index or Shannon index in Figure 2A .

According to the β-diversity of the gut microbiota, NMDS analysis was used, and there was significant (P<0.001) separation of microbial communities among the control, DSS and DSS + Tacro groups. The DSS group was located distant from the control group, while the position of the DSS + Tacro group was closer to the control group than to the DSS group ( Figure 2B ). Consistent with a previous study (Shin et al., 2015), DSS administration resulted in an obvious increase in the relative abundance of pylum_Proteobacteria and Deferribacteres and a decrease in the Firmicutes/Bacteroidetes ratio. All these changes were reversed by tacrolimus treatment ( Figure 2C ). LEfSe is a commonly used method to identify potential biomarkers with significant differences among groups. In our study, the abundances of the genera_Oscillospira, Bacteroides, Escherichia, Enterococcus, Turicibacter and Mucispirillum were significantly enriched, whereas the abundances of the genera_Lactobacillus, Allobaculum and Clostridium were significantly decreased in the DSS group compared to the normal control group ( Figure 2D ). Among the bacteria that differed significantly between the control and DSS groups at the genus level, Lactobacillus occupied the largest mean proportion, approximately 30% ( Figure 2E ). Correlation analysis showed that the DAI scores of colitis were negatively correlated with the abundance of the genus_Lactobacillus, as shown in Figure 2F (R=-0.81; P < 0.001) and Supplementary Figure 2A (R=-0.76; P < 0.001).

In the DSS + Tacro group, LEfSe analysis found that the increased abundance of the genera_Oscillospira, Escherichia, Turicibacter, Mucispirillum, SMB53, Bilophila as well as the decreased abundance of the genera_Lactobacillus and Allobaculum in the colitis model were significantly reversed by tacrolimus treatment ( Figure 2G ). Tacrolimus remarkably (P < 0.05) restored the abundance of the genera_ Lactobacillus, Oscillospira, Turicibacter, etc. ( Figures 2H, I ). Among these significant changes in the microbiota, the abundance of the genus_Lactobacillus showed the most significant changes, with the highest proportion (approximately 15%) in the DSS + Tacro group ( Figure 2H ). To validate our findings, we performed the same research in another center and confirmed similar trends, both showing a decrease in the genus_Lactobacillus in the DSS group and a restoration of its abundance when treated with tacrolimus ( Supplementary Figures 2A, B ). This strengthened our confidence in the findings, although the composition of gut microbiota is not similar in two centers.

Lacto, a strain of the genus Lactobacillus plantarum, was recently isolated from pickle of Sichuan cuisine by our group and determined by 16S ribosomal DNA identification. Body weight changes in the DSS + 1Tacro (1 mg/kg) group and the DSS + Lacto group were not significantly different from those in the DSS group, whereas the combination of Lacto in the DSS + 1Tacro + Lacto group maintained the body weight at a higher level, even closer to that in the DSS + 10Tacro (10 mg/kg) group ( Figure 3A ). Moreover, Lacto further prevented body weight loss in DSS + 10Tacro mice ( Figure 3A ) compared to monotherapy with tacrolimus in the DSS model (P < 0.05). As determined by DAI, the combination of Lacto with either tacrolimus at doses of 1 mg/kg or 10 mg/kg significantly (P < 0.01 and P < 0.0001, respectively) decreased DAI scores compared to the DSS group ( Figure 3B ). Colon length, which is negatively related to the inflammation of colitis (Liu et al., 2017), was measured, and a significantly increased length was reported in the DSS + 10Tacro (P < 0.01) as well as the DSS + Tacro + Lacto groups (P < 0.05, P < 0.01, respectively), with the most increased length in the DSS + 10 Tacro + Lacto group ( Figures 3C, D ). Survival rates were significantly (P < 0.01) increased in the DSS + 10Tacro and DSS + 10Tacro + Lacto groups, with the latter group showing better survival ( Figure 3E ). Compared to the DSS group, cotreatment with tacrolimus and Lacto obviously (P < 0.05 and P < 0.01, respectively) relieved the inflammation of the colonic mucosa, as determined by the histological scores ( Figures 3F, G ). Overall, Lacto, combined with tacrolimus, further ameliorated colitis in a mouse model in clinical and histological manifestations.

RNA sequencing data revealed that at the transcriptional level, the immune and inflammation-related signaling pathways, including the allograft rejection, IFN-γ response, IFN-α response, IL6 JAK STAT3 signaling, inflammatory response, TNFα signaling via NF-κB, complement and IL2 STAT5 signaling were significantly upregulated after DSS administration ( Figure 4A ). Most of these pathways, including the allograft rejection, TNFα signaling via NF-κB, IL6 JAK STAT3 signaling, inflammatory response and complement were significantly downregulated by tacrolimus ( Figure 4B ). When tacrolimus was combined with Lacto, the signaling pathways, including IFN-γ and IFN-α response, allograft rejection, IL2 STAT5 signaling and the inflammatory response, were further downregulated ( Figure 4C ), implying that the combined therapy of tacrolimus and Lacto further inhibited the inflammatory response in the colonic mucosa of colitis mice compared to the tacrolimus monotherapy group. The results of the gene set variation analysis are visualized in Figure 4D . The range of colors indicates the relative abundance of the signaling pathways mentioned above for each sample.

To demonstrate the profile of immune cells in the colonic mucosa, we analyzed the immune cell infiltration levels based on transcriptome sequencing data. As shown in Figure 4E , the proportion of monocytes and macrophages, especially the M1 subtypes, was elevated in the DSS group, while the M2 subtypes were slightly reduced with DSS intervention, in accordance with published reports overviewing the opposing roles of M1 and M2 macrophages in DSS-induced colitis (Lin et al., 2014). To a certain extent, tacrolimus suppressed the increase in the proportion of monocytes and macrophages, whereas coadministration of tacrolimus and Lacto significantly (P < 0.05) inhibited the expansion of these innate immune cells ( Figure 4E ) but not other immune cells, including dendritic cells (DCs), type 1 conventional dendritic cells (cDC1), type 2 conventional dendritic cells (cDC2), plasmacytoid dendritic cells (pDCs), B cells, memory B cells, B1 cells, follicular B cells, germinal center B cells, CD8⁺ cytotoxic T cells (Tc), naïve CD8⁺ T cells, Tγδ cells (Tgd), mast cells, M2 macrophages, NK cells, basophils, granulocytes and eosinophils ( Supplementary Figure 3 ). Altogether, our results suggested that the downregulation of inflammatory signaling pathways may be one of the possible mechanisms accounting for the alleviation of colonic mucosal inflammation after Lacto supplementation.

Next, to explore whether the effect of Lacto supplementation on colitis remission had some relationship with alterations in the gut microbiome, similar to that previously observed in the tacrolimus-treated group (Zhang et al., 2018; Jiao et al., 2020), we compared the results of 16S rRNA sequencing in the presence and absence of Lacto. Increased α-diversity of the gut microbiome was observed in the combination treatment group compared to the tacrolimus-treated group, as reflected by the Chao1 and Shannon indexes ( Figure 5A ). The results from NMDS and ANOSIM analysis clearly indicated that the bacterial communities of these two groups differed from each other ( Figure 5B ). In contrast with tacrolimus treatment, the relative abundance of the pathogenic bacteria Proteobacteria was further reduced, while the abundance of Firmicutes and Bacteroidetes, the two major phyla composing the gut microbiota community in healthy humans, was better maintained by Lacto supplementation ( Figure 5C ). However, Lacto alone did not significantly affect the bacterial diversity or cause beneficial changes in the composition of gut microbiota in colitis ( Figures 5D–F ).

Bile acid metabolism is a biological process in which specific intestinal floras are actively involved (Wahlström et al., 2016). Recently, gut microbiome-derived bile acids have been reported to play a critical role in IBD patients and experimental colitis(Lavelle and Sokol, 2020; Sinha et al., 2020; Thomas et al., 2022). Considering the significant changes in diversity and composition after the combination of tacrolimus and Lacto, bile acids were also detected in our study through high-throughput sequencing. Among all the bile acids analyzed in the Con, DSS and DSS + Tacro groups, the cecal taurochenodeoxycholic acid (TCDCA) level was most strongly correlated with the relative abundance of the genus_Lactobacillus ( Figure 5G ). Furthermore, a significant positive correlation (R=0.82; FDR<0.001) between the TCDCA concentration and the abundance of genus_Lactobacillus was observed in our experiments ( Figure 5H ).

The TCDCA levels that remarkably (P < 0.01) declined after DSS intervention were not restored by tacrolimus treatment. However, the combined treatment of tacrolimus plus Lacto drastically (P < 0.05) increased the TCDCA concentrations in comparison with tacrolimus treatment ( Figure 5I ). In addition, the cecal TCDCA concentrations of mice were negatively (R=-0.83; P<0.001) correlated with the DAI score ( Figure 5J ), suggesting an association between this bile acid and colitis severity.

Bile acids can function as signaling molecules to activate their receptors, thereby affecting intestinal inflammation (Perino et al., 2021; Yang et al., 2021). As presented by our transcriptome analysis, DSS administration perturbed the expression of the bile acid-sensing receptors, including G-protein-coupled bile acid receptor 1 (Gpbar1, TGR5), farnesoid X receptor (FXR), vitamin D (1,25-dihydroxyvitamin D3) receptor (VDR), and retinoid X receptor (RXR), to various degrees, all of which were mitigated somewhat by tacrolimus treatment except TGR5, and this effect was amplified further by coadministration of Lacto ( Figure 5K ). Collectively, our findings demonstrated that the combination of tacrolimus and Lacto improved the diversity and composition of the gut microbiome as well as bile acid metabolism in colitis mice, which possibly contributed to the alleviation of intestinal inflammation. The detailed mechanisms remain to be further explored.