Novel Gut Microbiota Patterns Involved in the Attenuation of Dextran Sodium Sulfate-Induced Mouse Colitis Mediated by Glycerol Monolaurate via Inducing Anti-inflammatory Responses

ABSTRACT Inflammatory bowel disease (IBD) is a type of immune-mediated chronic and relapsing inflammatory gastrointestinal symptoms. IBD cannot be completely cured because of the complex pathogenesis. Glycerol monolaurate (GML), naturally found in breast milk and coconut oil, has excellent antimicrobial, anti-inflammatory, and immunoregulatory functions. Here, the protective effect of GML on dextran sodium sulfate (DSS)-induced mouse colitis and the underlying gut microbiota-dependent mechanism were assessed in C57BL/6 mice pretreated or cotreated with GML and in antibiotic-treated mice transplanted with GML-modulated microbiota. Results showed that GML pretreatment has an advantage over GML cotreatment in alleviating weight loss and reducing disease activity index (DAI), colonic histological scores, and proinflammatory responses. Moreover, the amounts of Lactobacillus and Bifidobacterium and fecal propionic acid and butyric acid were elevated only in mice pretreated with GML upon DSS induction. Of note, fecal microbiota transplantation (FMT) from GML-pretreated mice achieved faster and more significant remission of DSS-induced colitis, manifested as reduced DAI, longer colon, decreased histological scores, and enhanced colonic Foxp3+ regulatory T cells (Tregs) and ratio of serum anti-inflammatory/proinflammatory cytokines, as well as the reconstruction of microbial communities, including elevated Helicobacter ganmani and decreased pathogenic microbes. In conclusion, GML-mediated enhancement of Bifidobacterium and fecal short-chain fatty acids (SCFAs) could be responsible for the anticolitis effect. FMT assay confirmed that gut microbiota modulated by GML was more resistant to DSS-induced colitis via elevating beneficial H. ganmani and establishing Treg tolerant phenotype. Importantly, colitis remission induced by GML is associated with novel gut microbiota patterns, even though different microbial contexts were involved.

beneficial anticolitis effect of GML in dextran sulfate sodium (DSS)-induced colitis in C57BL/6 mice and elucidate the important aspect of mechanisms involved.

RESULTS
GML pretreatment prevents DSS-induced acute colitis. In the pilot experiments, dose-response curves of DSS and GML were performed. DSS (2.5%) and GML (6 mg/ mouse/day) were considered to be the optimal dosages for formal experiments (see Fig. S1 in the supplemental material). In the present study, DSS resulted in severe diarrhea and obvious weight loss (Fig. 1B), reduction in the colon length ( Fig. 1D and E), and histological signs of experimental colitis, including colonic epithelial damage and chronic inflammatory infiltrate ( Fig. 1F and G). Of note, GML pretreatment improved the body weight significantly at day 7 post-DSS application in comparison to the DSS group, whereas GML cotreatment led to less weight loss at the earlier 4-day point but not at days 5 to 7 compared to the DSS group (Fig. 1B). There was 80% survival in mice from DSS and co-GML1DSS groups, while GML pretreatment improved survival to 100% upon DSS induction (Fig. S2). The disease activity index (DAI) remained null for the vehicle (Veh) group and was remarkably lower in GML-pretreated mice than in mice in the DSS group (P , 0.001 at day 6 and P , 0.01 at day 7) (Fig. 1C). The shortening of average colon length in mice of the DSS group (53.7 6 5.64 mm) was significantly reversed in GMLpretreated mice (63.7 6 4.94 mm; P , 0.01) and GML-cotreated mice (62.5 6 6.37 mm; P , 0.05) ( Fig. 1D and E). Moreover, blind histological assessment of colonic micrographs revealed that pretreatment with GML significantly reversed DSS-induced epithelial damage, chronic inflammatory infiltrate, and the extent of inflammation, resulting in a large decrease of overall histological score (P , 0.01). However, cotreatment with GML did not improve the histological parameters of colitis ( Fig. 1F and G).
GML pretreatment prevents DSS-induced microbial dysbiosis. Previous studies revealed that microbial dysbiosis was observed in IBD. To investigate whether GML (I and J) Volcano plot of differentially expressed genes (DEGs) (I) and heatmap representation of DEG levels from colon tissue between pre-GML1DSS and DSS groups as measured by RNA-Seq technology (J). Twenty-three genes were upregulated and 8 genes were downregulated (log 2 jfold changej $ 1.45, adjusted P , 0.05) between pre-GML1DSS and DSS groups. Expression level was represented as each gene's fragments per kilobase million (FPKM) value. *, P , 0.05; **, P , 0.01; ***, P , 0.001, versus Veh group and #, P , 0.05, ##, P , 0.01, versus DSS group for panels A to H, determined by a one-way ANOVA followed by Bonferroni's post hoc test.
Novel Gut Microbiota Patterns and Colitis Remission ® could reverse the dysbiosis, we further evaluated the diversity and composition of fecal microbiota in different groups. Compared with the Veh group, microbial community in the DSS group exhibited reduced a-diversity as assessed by Observed_species (P = 0.069), Chao1 (P , 0.05), Shannon (P = 0.07), and Simpson (P , 0.05) indices. However, the Shannon and Simpson indices in the pre-GML1DSS and co-GML1DSS groups were similar to those of the Veh group (Fig. 3A). Using the operational taxonomic unit (OTU)-level Bray-Curtis dissimilarity matrix, the permutational multivariate analysis of variance (PERMANOVA) indicated a statistical separation among the four groups (P = 0.003; Fig. S4A). The overall difference in the microbial structure (b-diversity) in the pre-GML1DSS (P , 0.01) and co-GML1DSS (P , 0.05) groups was distinct from that of the DSS group, based on weighted but not unweighted UniFrac distances (Fig. 3B). Referring to the taxonomic shifts, phylum-level decreases in the relative abundance of Bacteroidetes and increases in Firmicutes, and higher ratios of Firmicutes/ Bacteroidetes, were observed in mice pretreated with GML relative to animals in the Veh group (P , 0.05 for all). Proteobacteria were more abundant in the co-GML1DSS group than in other groups (Fig. 3C). At the genus level, DSS increased the level of Turicibacter and decreased Chryseobacterium ( Fig. 3D and E), whereas GML pretreatment inhibited the increase of Turicibacter but possessed higher percentages of Romboutsia, Lactobacillus, and Bifidobacterium (specifically Bifidobacterium animalis), while GML cotreatment was associated with elevated Romboutsia and Chryseobacterium ( Fig. 3D and E and Fig. S4C and D). Similarly, as shown in Fig. S4B, the linear discriminant analysis (LDA) effect size (LEfSe) algorithm ranked Erysipelotrichia (from class to family) and genus Turicibacter as the main bacterial taxa of the DSS group, while the Veh group was enriched with family Muribaculaceae within Bacteroidetes, in contrast with those of the pre-GML1DSS group and co-GML1DSS group, which were dominated by Firmicutes and Proteobacteria, respectively. Furthermore, GML pretreatment rescued the concentrations of microbederived SCFAs. Results showed that the concentrations of butyric acid tended to decrease in the DSS group relative to the Veh group (P = 0.065), but mice pretreated with GML exhibited increased butyric acid (P = 0.058) and propionic acid (P , 0.05) compared to mice in the DSS group. Decreased acetic acid and isovaleric acid and unchanged isobutyric acid, valeric acid, and hexanoic acid were observed in the DSS group compared with the Veh group. Consequently, the total SCFAs were lower in the DSS group than the Veh group (P , 0.001) but higher in the pre-GML1DSS than the DSS group (P , 0.05). However, a significant reduction in propionic acid, butyric acid, and total SCFAs was observed in the co-GML1DSS group compared with Veh and pre-GML1DSS groups (Fig. 3F). These data indicated that GML-moderated DSS-induced experimental colitis was closely related to the reconstruction of microbial patterns. Since mice in the pre-GML1DSS group were more resistant to DSS-induced colitis than those in the co-GML1DSS group, which in part excluded the direct effect of GML against the colitis, we concluded that the beneficial effects of GML pretreatment could be mediated by the differential microbial taxa, including Lactobacillus and Bifidobacterium, and the recovery of butyric acid and propionic acid.
FMT from GML-treated donor mice prevented DSS-induced colitis in antibiotictreated mice. Fecal microbiota transplant (FMT) experiments were thus performed in antibiotic-treated mice, which were modeled to colitis by DSS application during the FMT course (Fig. 4A), therefore allowing for a quantifiable effect of GML-modulated microbiota on protection against the colitis-related syndromes. Results showed that FMT from GML-treated donor mice significantly attenuated the DAI (Fig. 4C), colon length shortening ( Fig. 4D and E), and histological symptoms and scores ( Fig. 4F and G) compared with FMT from vehicle-treated mice. It should be mentioned that FMT from vehicle-treated mice contained a healthy microbial community as well, which may illustrate the undifferentiated weight between the G-FMT1DSS and V-FMT1DSS groups during the colitis period. In contrast, a tendency higher in the body weight percentage was maintained in the G-FMT1DSS group relative to the V-FMT1DSS group at days 11 to 14 (Fig. 4B).
FMT from GML-treated mice reshaped the gut microbiota in antibiotic-treated mice upon colitis induction. As shown in Fig. S6A to C, there were no differences in the microbial load and diversity between donor mice. Of importance, in line with the higher Bifidobacterium level in mice in the pre-GML1DSS group ( Fig. 3D and E), a higher abundance of Bifidobacterium (P , 0.05) was detected in GML-treated donor mice relative to vehicle-treated mice, which further indicated that Bifidobacterium could be responsible for the anticolitis effect of GML pretreatment, although it was not maintained in the homologous recipient mice after FMT ( Fig. 6C and D). In terms of recipient mice, we first demonstrated that antibiotic (ABX) treatment resulted in at least 10 4 -fold depletion of the gut bacterial load (Fig. S6D), significant decrease in the number of observed OTUs (Fig. S6E), and change in the microbial compositions at the phylum and genus levels ( Fig. S6F and G). After FMT, the a-diversity showed no difference between the G-FMT1DSS and V-FMT1DSS groups, as reflected by Observed_species, Chao1, Shannon, and Simpson indices (Fig. S6B). Principal-coordinate analysis (PCoA) differentiated the microbiota in the G-FMT1DSS group from those in the V-FMT1DSS group, based on OTU-level Bray-Curtis dissimilarity matrix (PERMANOVA, P = 0.011 [ Fig. 6A]), although b-diversity based on weighted and unweighted UniFrac distance showed no difference (Fig. S6C). Specifically, phylogenetic analysis indicated distinguishable microbial profiles between G-FMT1DSS and V-FMT1DSS groups. Phylumbased distinct microbial communities were explained by richer unidentified_Bacteria (P , 0.001), higher Firmicutes/Bacteroidetes ratio (P , 0.05), and reduced levels of Proteobacteria (P , 0.05) and Deferribacteres (P , 0.001) in the G-FMT1DSS group than in the V-FMT1DSS group (Fig. 6B). When looking at the genus level, Helicobacter (specifically species Helicobacter ganmani) mainly contributed to the increase of unidentified_Bacteria and was equally abundant in the G-FMT1DSS group (P , 0.001, Fig. 6C to E and Fig. S7). In contrast to Helicobacter, a tendency of lower Alloprevotella (P = 0.054, within Bacteroidetes) and significantly reduced Mucispirillum (P , 0.001, within Deferribacteres) and Klebsiella and Cupriavidus (P , 0.05 and P , 0.001, respectively, both within Proteobacteria) were observed in the G-FMT1DSS group compared with the V-FMT1DSS group ( Fig. 6C to E). Of note, the fecal microbial communities of the recipient mice were generally distinct from those of their corresponding donor mice. These differential taxa, including the higher abundance of Helicobacter and lower abundance of Alloprevotella, Mucispirillum, Klebsiella, and Cupriavidus seen in the G-FMT1DSS group, were not observed in GML-treated donor mice. Taken together, colitis remission induced by FMT from GML-treated mice could be partly ascribed to microbial FIG 3 Legend (Continued) abundances of gut microbiota at the phylum level. (D) Heatmap depicting the log 10 jrelative abundancej of top 30 microbial genera, when the taxa with the relative abundance as zero were expressed as 24. (E) Relative abundance of five differential microbial genera. (F) The production of microbial metabolite SCFAs from feces among the four different groups. Each dot represents an individual mouse (n = 9 to 10 mice/group). *, P , 0.05; **, P , 0.01; ***, P , 0.001, versus Veh group; #, P , 0.05; ##, P , 0.01, versus DSS group; †, P , 0.05; † †, P , 0.01, versus pre-GML1DSS group, determined by a nonparametric factorial Kruskal-Wallis test followed by the Mann-Whitney U test when P was ,0.05. Novel Gut Microbiota Patterns and Colitis Remission ® reconstruction in the recipient mice without closely resembling the microbial pattern in the donor mice. Previous studies have highlighted that the inflammation could also influence the reshaping of the gut microbiota (10); however, in the current study, GML-modulated microbiota was more resistant to DSS-induced colitis.
FMT from GML-treated mice differentiates the microbial phylotypes, butyrate metabolism, and promoted Treg phenotype in antibiotic-treated mice upon colitis induction. The LEfSe analysis indicated that mice receiving microbiota from GMLtreated donor mice had higher levels of phylum Firmicutes, class Clostridia, order Clostridiales, and Helicobacter (from phylum to genus), whereas mice receiving microbiota from vehicle-treated mice had richer phylum Proteobacteria, class Erysipelotrichia (Turicibacter), and order Enterobacteriales (Escherichia coli) (Fig. S8). The dominance of specific members of Clostridia involved in the butyrate metabolism has been acknowledged (25). As expected, there was a trend toward higher butyric acid production in the G-FMT1DSS group than in the V-FMT1DSS group (P = 0.07), while other SCFAs showed no difference (Fig. 6F). Furthermore, epidemiological data have linked the protective benefit of Helicobacter pylori, a species of Helicobacter, against the relative risk of IBD (26). And Helicobacter hepaticus, another species of Helicobacter, specifically promoted the differentiation of colonic RORgt-expressing induced Treg (iTreg) cells, which restore eubiosis by facilitating microbiota colonization and suppressing colitis-related inflammation (27). Thus, the colonic lamina propria lymphocytes (LPLs) were analyzed with flow cytometry. Results revealed that FMT from GML-treated mice upregulated the levels of CD4 1 CD25 1 Foxp3 1 Treg cells ( Fig. 5A and B, P , 0.05), as well as dramatically increasing the transcriptional levels of Foxp3 and Rorgt (Fig. 5C). Our findings  (D) Relative abundance of two differential microbial genera between V-donor group and G-donor group. (E) Relative abundance of six differential microbial genera between V-FMT1DSS group and G-FMT1DSS group. (F) Fecal SCFA quantification at day 15 between donor mice, as well as between V-FMT1DSS and G-FMT1DSS groups. Each dot represents an individual mouse (n = 6 mice/group for donor mice, n = 11 mice/group for recipient mice). $, P , 0.05, versus V-donor group. §, P , 0.05, and § § §, P , 0.001, versus V-FMT1DSS group, determined by a nonparametric factorial Kruskal-Wallis test followed by the Mann-Whitney U test when P was ,0.05.

Novel Gut Microbiota Patterns and Colitis Remission
® were parallel to the previous data showing that the specific microbial taxa (such as Clostridia and Helicobacter), butyric acid, Tregs, and the cross talk between them may have a predominant role in preventing DSS colitis (28).
Correlations among GML-induced gut microbiota, significantly altered colitisrelated parameters, and inflammation-related factors. We subsequently correlated significantly altered colitis-related parameters (including body weight, colon length, DAI, and histological scores) and pro-/anti-inflammatory cytokines with specific differential microbial taxa (Fig. 7). Spearman rank correlation analysis showed that the positive correlation between Turicibacter abundances and histological scores and the negative correlation between Turicibacter abundances and body weight and colon length were coincident in two independent experiments. Notably, the abundance of Turicibacter was remarkably increased by DSS induction but was reversed by GML pretreatment or FMT from GML-treated donor mice, suggesting that Turicibacter could be identified as one of the symbolic microbial taxa of DSS-induced colitis. Moreover, a significant increase of two gut commensals, Lactobacillus and Bifidobacterium, in the pre-GML1DSS group was correlated with lower levels of colonic il6, il23p19, and il17 mRNA FIG 7 The Spearman correlation between the specific or differential microbial taxa and the colitis-related indicators among Veh, DSS, pre-GML1DSS, and co-GML1DSS groups (A) and between V-FMT1DSS and G-FMT1DSS groups (B). Red squares indicate positive correlations, and green squares indicate negative correlations.~, 0.05 , P , 0.1, means that correlations remained trending significant, and *, P , 0.05, and **, P , 0.01, mean statistically significant after correction for false-discovery rate. and lower circulating IL-6 and IL-17 levels, indicating negative correlations between Lactobacillus and Bifidobacterium and colonic and systemic inflammation (Fig. 7A). A good correlation between Chryseobacterium and colitis-related parameters and inflammatory status is presented in Fig. 7A. Referring to FMT assay (Fig. 7B), in accordance with the predicted capacity of Helicobacter, which was more abundant in the G-FMT1DSS group than the V-FMT1DSS group, to promote the differentiation of Tregs, we observed a positive correlation between Helicobacter (H. ganmani) and proportion of Foxp3 1 Tregs (R 2 = 0.709, P , 0.01) and Foxp3 (R 2 = 0.502, P = 0.0506) and rorgt (R 2 = 0.730, P , 0.01). Similar correlations between Clostridia and Treg and rorgt were observed. In contrast, Cupriavidus (Cupriavidus metallidurans), Klebsiella (Klebsiella pneumoniae), Mucispirillum (Mucispirillum schaedleri), and E. coli, which were lower in the G-FMT1DSS group than the V-FMT1DSS group, were positively correlated with histological scores, DAI, and proinflammatory cytokines but negatively correlated with body weight, colon length, and anti-inflammatory factors significantly (Fig. 7B). These observations indicated that the distinct gut microbiota patterns reshaped by GML pretreatment and/or FMT trial were both linked to the moderation of colitis-related parameters and the regulated production of colonic inflammation-related factors.

DISCUSSION
The present study demonstrated for the first time that GML can ameliorate DSSinduced colitis in C57BL/6 mice. DSS resulted in colonic inflammatory response and epithelial ulceration, while GML dramatically reversed DSS-mediated inflammation and injury. Furthermore, both DSS and GML pretreatment increased expression of colonic il10, indicating that the favorable effect of GML was not mediated directly via IL-10, which is crucial for preventing spontaneous colitis and maintaining homeostasis (29). Previous research also found increased expression of colonic il10 in DSS-treated mice (30). Of interest, GML pretreatment seemed to be more effective in colitis management than GML cotreatment, although GML cotreatment exhibited a favorable effect on weight change in the early days of DSS application. One possible explanation could be that GML might have a protective effect in the early stage of inflammation. When the inflammatory response was aggravated and the mucosal layer was impaired, GML would be inadequate to rescue the disturbance and, instead, might be detrimental because of its emulsifying property. However, the intestinal permeability analysis showed no obvious difference among different groups, indicating that GML exerts its anticolitis effect in some other way. Specifically, the distinct microbial profiles and microbe-derived SCFAs between pre-GML1DSS and co-GML1DSS groups might be responsible for the case. Mice in the pre-GML1DSS group but not the co-GML1DSS group contained a lower abundance of inflammation-related Turicibacter and higher proportions of Bifidobacterium and Lactobacillus, which ae well-known beneficial SCFAproducing microbes and have anti-inflammatory and anti-colorectal cancer effects in humans and mice (31,32). Similarly, a previous study demonstrated that prebiotic inulin supplementation promoted indigenous Bifidobacterium and Lactobacillus, which were associated with the decreased risk of IBD development (33). Similarly to IBD patients who have typically decreased microbial diversity and attenuated ability to produce SCFAs (2), DSS resulted in significant decline in the microbiota diversity and total SCFAs, especially acetic acid and butyric acid. In contrast, GML, regardless of pretreatment or cotreatment, reversed the DSS-induced dysbiosis in microbial diversity. Accordingly, butyric acid and propionic acid, which were marginally restored in the pre-GML1DSS group relative to the DSS group, could contribute to ameliorating experimental colitis (28). The aforementioned results were consistent with the previous reports that Bifidobacterium has been regarded as an acetate-to-butyrate converting organism, and the synthetic butyrate, in turn, promoted the growth of Bifidobacterium and Lactobacillus in the gut (34,35). It has been verified that Bifidobacterium and butyrate have beneficial effects on ameliorating mucosal inflammation and related symptomology (8, 31, 32). Notably, the higher Bifidobacterium in the pre-GML1DSS group was in line with the higher Bifidobacterium in GML-treated donor mice, indicating that GML-modulated specific microbial taxa were responsible for its ant-inflammatory potential. Indeed, FMT from GML-treated mice has an advantage over that from vehicle-treated mice in achieving colitis remission, and in shifting the microbial communities and immune response toward better homeostasis. In contrast, different gut microbiota patterns were involved in the GML prevention experiment and FMT trial.
Similarly to the previous FMT trials (36,37), the variations of gut microbiota in the recipient mice explain the distinct effects on colitis, which further provided a perspective to identify novel microbial members that are responsible for the anticolitis effect. Mice receiving microbiota from vehicle-treated mice were associated with dominant Erysipelotrichaceae (Turicibacter) and Enterobacteriaceae (E. coli) and significantly increased Alloprevotella, Mucispirillum, Klebsiella, and Cupriavidus, which have been shown to facilitate intestinal inflammation or colorectal cancer (38,39). LEfSe analysis of DSS and V-FMT1DSS groups highlighted that Erysipelotrichaceae (Turicibacter) and Enterobacteriaceae (E. coli) could be deemed the indicator phylotypes of active colitis. Consistently, similar change in Erysipelotrichaceae has been described in IBD patients and inflammation-related mouse models (40), and Enterobacteriaceae (E. coli) could be selectively enhanced by inflammatory response (41). In contrast, FMT from GMLtreated mice during colitis resulted in a trend of increase of Firmicutes (represented by Clostridia) and a marked decrease of Proteobacteria (represented by Klebsiella, Cupriavidus, and Enterobacteriaceae), in parallel with clinical remission by FMT or other intervention trials on mouse colitis model (42). The previous results implied that members of Clostridia were protective against DSS-induced colitis by restoring butyric acid and colonic Foxp3 1 Tregs (25) and could be used as a potential biomarker to guide therapy (39,43). As expected, transplant of GML-mediated microbiota promoted butyric acid concentration, shaped Foxp3 1 Treg differentiation in the LPLs, and shifted cytokine profiles at the gene and protein levels, including IL-10 and IL-6, which may in part be responsible for the significant amelioration of colitis-related status and symptoms caused by DSS. Moreover, IL-6 plays a vital role as a signal marker in restraining Treg maturation and triggering Th17-mediated inflammation (44). Thus, the marked reduction of IL-6 further supported the conclusion that GML-altered microbiota maintains Treg/Th17 balance upon DSS induction. Of importance, genus Helicobacter (specifically H. ganmani), dominant in the G-FMT1DSS group compared with the V-FMT1DSS group, was nonpathogenic in mice and humans and uncorrelated with colitis (45). Previous observations reported that treatment with Tregs resulted in colitis remission in TRUC mice by preferentially facilitating the proliferation of H. ganmani (46). Here, the augment of Helicobacter and H. ganmani may contribute to the induction of RORgt 1 Foxp3 1 Treg cells, both of which were responsible for the mechanism whereby FMT from GML-treated mice achieved a faster remission of DSS-induced colitis. Moreover, the positive correlation between H. ganmani and RORgt 1 Foxp3 1 iTregs fit in exactly with the prospect raised previously that the homeostasis in IBD patients could be reestablished by exploring or managing nonpathogenic microorganisms to induce iTreg responses (27,47). The current experimental data motivated several trials of H. ganmani as IBD therapy.
In conclusion, our findings revealed that GML-modulated microbiota, whether retained within the conventional mice or moved by FMT into the recipient mice, were similarly capable of preventing DSS-induced colitis, despite the fact that distinct gut microbiota patterns were involved in these two conditions. As has been demonstrated recently, certain amounts of clades from different taxa may share similar functions, such as SCFA metabolism and inflammation suppression, because of the functional redundancy of the microbial community (40,48,49). For example, Bifidobacterium enriched in the pre-GML1DSS group and GML-treated donor mice and Helicobacter enriched in the G-FMT1DSS group can potentially promote colonic indoleamine 2,3dioxygenase production and Treg differentiation, both of which play vital roles in attenuating inflammation (50,51).

MATERIALS AND METHODS
Animals, colitis induction, and GML treatment. Male C57BL/6 mice (4 to 5 weeks old) were purchased from the SLAC Laboratory Animal Co. Ltd. (Shanghai, China) and housed under specific-pathogenfree conditions with food and water ad libitum. Food-grade glycerol monolaurate (GML) was obtained from Kangyuan Food Science and Technology Co., Ltd. (Hangzhou, China). DSS (dextran sodium sulfate; molecular weight [MW] 36,000 to 50,000; MP Biomedicals, USA) was dissolved in the drinking water to a concentration of 2.5%. Mice were randomized to four groups (n = 15/group) as follows: Veh group, mice received vehicle (days 0 to 28); DSS model group, vehicle (days 0 to 28) and DSS application on days 21 to 28; GML pretreatment group (pre-GML1DSS), GML (days 0 to 21), vehicle, and DSS (days 21 to 28); GML cotreatment group (co-GML1DSS), GML (days 0 to 28) and DSS added (days 21 to 28). Vehicle (50% polyethylene glycol 400 and 50% distilled water) and GML (6 mg/mouse/day, dissolved in vehicle) were administered daily by gavage (300 ml/mouse). Weight of mice was recorded daily during DSS induction and expressed as a percentage of the weight at day 21 (at which time DSS was applied) defined as 100%. Fresh feces were harvested at day 4 after DSS application. At day 28, mice were euthanized (Fig. 1A). Blood was collected, and serum was generated by centrifugation (3,000 Â g for 20 min at 4°C). Mice were euthanized, and colon was quickly removed for length measurement and collected for further assessment. All animal experiments were in accordance with federal guidelines and approved by the Institutional Animal Care and Use Committee of Zhejiang Chinese Medical University (China, approval no. IACUC-20180319-04).
Histological assessment. Colonic samples were fixed overnight in 10% formalin, dehydrated, and embedded into paraffin according to a standard procedure. The paraffin sections of colonic samples were subjected to hematoxylin and eosin (H&E) staining at core facilities, Zhejiang University School of Medicine (Hangzhou, China). The extent of colitis was assessed according to the previously described protocols (30). The following parameters were examined on a scale of 0 to 4, respectively: (i) epithelial damage-0, normal appearance; 1, mild abnormality; 2, probable erosion or multifocal abnormalities; 3, unequivocal erosion or multifocal abnormalities; 4, ulcer or severe diffusion; (ii) chronic inflammatory infiltrate-0, no infiltration; 1, mild but unequivocal infiltrate; 2, infiltrate, moderate increase; 3, infiltrate, marked increase; 4, accumulation in lumen; (iii) extent of inflammation-0, none; 1, mild diffusion around the crypt area; 2, moderate diffusion in the mucosa; 3, diffusion in submucosa; 4, transmural diffusion, pronounced edema. Histopathology scoring was calculated in a blind manner for a combined score of 0 to 20.
Antibiotic-treated mice and fecal microbiota transplantation (FMT). To deplete the gut microbiota, mice were treated with antibiotic (ABX)-containing water, including ampicillin (1 g liter 21 ), neomycin (1 g liter 21 ), metronidazole (1 g liter 21 ), and vancomycin (0.5 g liter 21 ) for 4 weeks (53). Microbiota depletion was confirmed by anaerobic culture of 10-fold-diluted fecal suspension on Columbia blood agar base containing 5% sterile defibrinated sheep blood. The littermates administered vehicle or GML intragastrically for 4 weeks were used as donor mice and divided into the V-donor and G-donor groups, respectively. Fresh feces collected from each group were pooled and homogenized in the sterile phosphate-buffered saline (PBS), respectively. The fecal supernatant obtained by centrifugation was orally inoculated into ABX-treated recipient mice, respectively (250 ml/mouse, daily), for 3 days, whereupon ABX-treated recipient mice consumed 2.5% DSS-containing water to induce colitis (Fig. 4A). Correspondingly, the recipient mice were divided into V-FMT1DSS and G-FMT1DSS groups, respectively. Feces and serum were collected. Mice were sacrificed at day 15, and colon was quickly collected for length measurement and further assessment.
Quantification of mRNAs by RT-qPCR. Total RNA from colonic samples was extracted using the FastPure Cell/Tissue total RNA isolation kit (Vazyme Biotech Co., Ltd.). The cDNA generated by HiScript II Q RT SuperMix for qPCR (1gDNA wiper) (Vazyme Biotech Co., Ltd.) was analyzed using primers for the indicated genes. Real-time quantitative PCR (RT-qPCR) was performed with 2ÂChamQ SYBR Color qPCR Master Mix (Vazyme Biotech Co., Ltd.) and run in a LightCycler 480 system (Roche) according to the manufacturer's instructions. Genes Foxp3, Il6, Tnfa, Il23p19, Il10, Il17a, Il22, Il1b, Tgfb1, Rorgt, Zo1, Occludin, and Jam-1 were evaluated, and b-actin was used as the housekeeping gene for colonic tissue. Relative expression of mRNA levels was quantified using the threshold cycle (2 2DDCT ) method as described previously (54). The primers are shown in Table 1.
Gut microbiota analysis. Fecal bacterial DNA was extracted with the QIAamp DNA stool minikit (Qiagen, Venlo, Netherlands) based on the manufacturer's instructions. The concentration and quality of the resulting DNA were determined with a UV-visible (UV-Vis) spectrophotometer (NanoDrop 2000c; Thermo Scientific). 16S rRNA amplicon sequencing was performed at the Novogene Bioinformatics Institute (Tianjin, China) with an Illumina IonS5 XL platform. The V3-V4 hypervariable regions in the bacterial DNA were PCR amplified from each sample with a primer set (F341, 59-ACTCCTACGGGRSGCAGCAG-39, and R806, 59-GGACTACVVGGGTATCTAATC-39) containing a unique barcode, which was used to tag PCR products from respective samples (55). The resulting sequences were analyzed according to previous work (53,56).
In detail, the resulting sequencing reads were demultiplexed and quality filtered by the following procedures: (i) reads that had more than three consecutive low-quality base calls were discarded using Cutadapt v1.9.1 and were assigned to respective sample according to the unique barcodes, (ii) adaptors and barcodes were trimmed and reads with average quality below Q20 were removed, and (iii) chimeras were dumped using the Userch algorithm (57,58). The processed sequences with a 97% pairwise identity threshold were clustered into the same operational taxonomic units (OTUs) using Uparse v7.0.1001 (59) and classified taxonomically using the SILVA 132 database (60). a-Diversity was estimated using the Chao1, Shannon, Simpson, and Observed_species indices and was analyzed using Quantitative Insights into Microbial Ecology software (QIIME v1.9.1). b-Diversity was evaluated using principal-coordinate analysis (PCoA) plots based on weighted UniFrac and the Bray-Curtis distance (61). To determine phylum, class, order, family, genus, and species OTUs, data were evaluated using a permutational multivariate analysis of variance (PERMANOVA) for analysis of Bray-Curtis indices. Linear discriminant analysis (LDA) effect size (LEfSe) algorithm was applied to identify specific taxa that responded to different treatments (53). Statistical analyses were performed with R (version 2.15.3).
Measurements of fecal short-chain fatty acid (SCFA) concentrations. Fecal samples were used to measure SCFA concentrations via gas chromatographic analysis as previously described (62). In detail, fecal pellets from single mice were weighed and homogenized in ultrapure water. The pH of fecal suspension was adjusted to 2 to 3 by adding 5 M HCl with intermittent vortexing within 15 min. The clear supernatant was collected after centrifugation for 20 min at 3,000 Â g. 2-Ethylbutyric acid (TEBA), used as the internal standard, was added into the resulting supernatant at a final concentration of 1 mmol liter 21 . A capillary column (30 m, 0.53 mm, 0.50 mm) with a free fatty acid phase (DB-FFAP 125-3237; J&W Scientific, Agilent Technologies Inc.) and the Shimadzu GC-2014 system were applied for gas chromatographic analysis (22).
LPL isolation. For evaluation of regulatory T cells (Tregs), lamina propria lymphocytes (LPLs) in the colonic samples were isolated as described previously (30). In brief, Peyer's patches were discarded and the whole colon was opened longitudinally, minced into 0.5-cm pieces, and digested in EDTA-containing Hanks' balanced salt solution (HBSS; 0.5 mol liter 21 EDTA) for 30 min at 37°C with rigorous shaking to remove intraepithelial lymphocytes and epithelial cells. Remaining tissue pieces were digested twice for 15 min in complete medium (Dulbecco's Modified Eagle Medium [DMEM], 10% fetal calf serum [FCS]) supplemented with 1 mol liter 21 HEPES buffer, 145 mg ml 21 collagenase II (Worthington Biochemical), and 50 mg ml 21 DNase I (Sigma-Aldrich) with pipetting using a 1-ml syringe. The resulting cell suspensions were filtered through a 70-mm cell strainer and centrifuged for 5 min at 1,000 Â g. Cell pellets were layered onto a 40:80% Percoll gradient and centrifuged at 2,000 Â g at room temperature for 20 min in a Thermo Sorvall ST16R centrifuge. The LPLs collected from the 40:80% interface were resuspended in PBS at a density of 1 Â 10 6 prior to staining for flow cytometry.

TABLE 1 Primer sequences used in the qRT-PCR assays
Flow cytometry. The surface staining followed routine procedures (8,30), and the intracellular staining was performed by use of the Foxp3 staining kit (eBioscience), according to the manufacturer's instructions. LPL suspensions were stained for surface molecules by incubating with anti-mouse CD4-FITC (eBioscience), anti-mouse CD25-allophycocyanin (APC) (eBioscience), and rat IgG2a isotype control-phycoerythrin (PE) (eBioscience) on ice for 20 min and then washed with PBS buffer. Fixable viability dye (FVD)-eFluor 450 (eBioscience) was used to exclude dead cells. Cells were then fixed in fixation and permeabilization working solution for 30 min, followed by washing and resuspension in permeabilization buffer prior to staining with anti-mouse Foxp3-PE antibody for at least 30 min on ice. Flow cytometric analysis was performed on an LSR Fortessa analyzer (BD Biosciences), and data were analyzed using FlowJo software (Tree Star).
Transcriptional profiles in the whole-colon tissue. Total RNA extracted from the colonic tissue and with an integrity number larger than 7 was subjected to RNA sequencing, which was performed by the Novogene Bioinformatics Institute (Tianjin, China) with the Illumina HiSeq platform (55). Briefly, raw reads were processed through Trimmomatic v0.3.0 to remove reads containing adaptors or poly-N and reads with low quality. The resulting clean reads were aligned to mouse genome (mm9), and the reads mapped to genomic features were counted by HISAT2 v2.0.1. The differential expression of genes and the KEGG pathway were analyzed by using HTSeq v0.6.0 and clusterProfiler R package, respectively.
Statistical analysis. Data were expressed as means with standard deviation (SD). The lowercase letter 'n' means the number of mice. Data sets among Veh, DSS, pre-GML1DSS, and co-GML1DSS groups were evaluated for whether they followed the normal distribution by the Kolmogorov-Smirnov test. The result showed that the data fitted the normal distribution. The Levene test of homogeneity of variance was further performed. When the data fitted the homogeneity of variance, one-way ANOVA followed by Bonferroni's post hoc test was performed. Nonparametric data among the four groups were analyzed with the Kruskal-Wallis test followed by the Mann-Whitney U test when P was ,0.05. A two-tailed Student t test was applied to compare V-donor group with G-donor group or V-FMT1DSS group with G-FMT1DSS group. IBM SPSS 21.0 (Chicago, IL, USA) and GraphPad Prism 6.0 (San Diego, CA, USA) were used to perform statistical analysis. The outliers were detected and eliminated with the ROUT method (Q = 1%) using GraphPad Prism 6.0. Differences were considered to be statistically significant when P was ,0.05.
Data availability. All the relevant data which support our findings are available from the authors on reasonable request; some have already been included in the paper and the supplemental material. Raw sequence data of microbiota which support the findings in our study have been deposited into NCBI's Sequence Read Archive under accession number PRJNA766304. Raw sequence data of transcriptome that support our findings have been deposited in NCBI's Sequence Read Archive under accession number PRJNA766430.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.