Total Flavonoids of Glycyrrhiza uralensis Alleviates Irinotecan-Induced Colitis via Modification of Gut Microbiota and Fecal Metabolism

Irinotecan (CPT-11)-induced gastrointestinal toxicity strongly limits its anticancer efficacy. Glycyrrhiza uralensis Fisch., especially flavonoids, has strong anti-inflammatory and immunomodulatory activities. Herein, we investigate the protective effect of the total flavonoids of G. uralensis (TFGU) on CPT-11–induced colitis mice from the perspective of gut microbiota and fecal metabolism. The body weight and colon length of mice were measured. Our results showed that oral administration of TFGU significantly attenuated the loss of body weight and the shortening of colon length induced by CPT-11. The elevated disease activity index and histological score of colon as well as the up-regulated mRNA and protein levels of TNF-α, IL-1β, and IL-6 in the colonic tissue of CPT-11–treated mice were significantly decreased by TFGU. Meanwhile, TFGU restored the perturbed gut microbial structure and function in CPT-11–treated mice to near normal level. TFGU also effectively reversed the CPT-11–induced fecal metabolic disorders in mice, mainly call backing the hypoxanthine and uric acid in purine metabolism. Spearman’s correlation analysis further revealed that Lactobacillus abundance negatively correlated with fecal uric acid concentration, suggesting the pivotal role of gut microbiota in CPT-11–induced colitis. Since uric acid is a ligand of the NOD-like receptor family pyrin domain containing 3 (NLRP3) inflammasome, TFGU was further validated to inhibit the activation of NLRP3 inflammasome by CPT-11. Our findings suggest TFGU can correct the overall gut microbial dysbiosis and fecal metabolic disorders in the CPT-11–induced colitis mice, underscoring the potential of using dietary G. uralensis as a chemotherapeutic adjuvant.


INTRODUCTION
Irinotecan (CPT-11) is an effective chemotherapeutic agent used for the treatment of colon, colorectal, lung as well as other solid tumors (1). However, its gastrointestinal toxicity can lead to a variety of symptoms such as diarrhea and intestinal mucositis, even threatening the lives of patients, which may also compromise its therapeutic effects (2,3). CPT-11-induced colitis involves multiple mechanisms, among which enterohepatic circulation involving bacterial b-glucuronidase is recognized as the most important part (4)(5)(6)(7). Fluoroquinolone (i.e., ciprofloxacin, CIF) prophylactic regimens have been shown to be highly effective against chemotherapy-induced bacteremia from gut bacteria (8). Since concerns over antibiotic resistance, there is no satisfactory therapeutic intervention that can prevent or treat CPT-11-induced gastrointestinal toxicity.
Glycyrrhiza uralensis Fisch. has been used as a natural sweetener and herbal medicine for inflammatory diseases (9). The total flavonoids of G. uralensis (TFGU) has been demonstrated to treat ulcerative colitis in mouse model due to its antioxidant activity through nuclear factor-erythroid 2related factor 2 (Nrf2) pathway and anti-inflammatory activity through NF-kB pathway (10). TFGU comprises a variety of flavonoids including liquiritigenin, isoliquiritigenin, liquiritin, isoliquiritin, and glycyrrihizin, which are considered as the major active constituents of G. uralensis. For example, isoliquiritigenin and glycyrrhizin could modulate the Toll-like receptor 4/myeloid differentiation protein 2 complex at the receptor level, leading to suppress lipopolysaccharide-induced activation of signaling cascades and cytokine production (11). Besides, liquiritigenin, isoliquiritigenin and isoliquiritin could mediate the anti-inflammatory responses of lipopolysaccharideinduced macrophage activation via Nrf2 and heme oxygenase-1 (HO-1), as well as prevent IkBa phosphorylation and degradation (12). Importantly, isoliquiritigenin mainly distributes in gastrointestinal tract and has potential to ameliorate the dextran sulfate sodium-induced colitis through inhibiting MAPK pathway (13) and inhibit colitis-associated tumorigenesis through hampering M2 macrophage polarization mediated by the interplay between prostaglandin E 2 (PGE 2 ) and interleukin-6 (IL-6) (14). However, whether TFGU can treat CPT-11-induced colitis and the underlying mechanism remain to be elucidated.
Growing body of research has highlighted that gut microbial homeostasis is helpful for host to maintain the integrity of gut epithelial barrier and modulate the metabolism and immune system, whereas gut microbial dysbiosis involved in inflammation-related diseases (15,16). CPT-11-induced colitis was causally associated with the gut microbial b-glucuronidase, suggestive of the participation of gut microbiota in the gastrointestinal toxicity of CPT-11 (6). Host and gut microbiota co-metabolism maintain the health of body under normal circumstances (17). Previous studies have demonstrated that CPT-11 treatment leads to a series of serum metabolic dysfunctions, mainly affecting the biosynthesis of phenylalanine, tyrosine and tryptophan as well as primary bile acids and short-chain fatty acids (SCFAs) (18,19). Therefore, it is meaningful to investigate the protective mechanisms of TFGU against CPT-11induced colitis from the modulation of the gut microbiota and fecal metabolism.
In this study, we first investigated the effects of TFGU on CPT-11-treated mice. Then, the gut microbiota and fecal metabolites were analyzed with the 16S rRNA sequencing and gas chromatography-mass spectrometry (GC-MS)-based untargeted metabolomics, respectively. Finally, Western blotting method was applied to preliminary investigate the effect of TFGU on the colonic expression of NOD-like receptor family pyrin domain containing 3 (NLRP3) inflammasome.

Preparation of CPT-11 and TFGU
The CPT-11 solution (6 mg/ml) was prepared according to the previous research (20): 34 ml L-(+)-lactic acid dissolved in 40 ml injection water (85°C), and subsequently the 0.277 mg CPT-11 and 0.225 g D-sorbitol were added, heated to 90°C and stirred until it turned to clear. After the pH was adjusted to 3.2 to 3.6, the solution was sterile filtered through 0.22 mm microfiltration membrane filtration and stored in dark until administration. TFGU was prepared according to our previous extraction process (Supplementary Figure 1) with the yield of 3% (w/w). The purity of TFGU expressed as liquiritin equivalents was 80.12% and its chemical compositions were characterized and provided in Supplementary Table 1 and Supplementary Figure 2. Moreover, HPLC analysis indicated that the major peaks identified by comparing with standard compounds were liquiritin apioside (27.06 mg/g), liquiritin (19.43 mg/g), naringin (1.65 mg/g), liquiritigenin (2.51 mg/g) quercetin (0.21 mg/g), licochalcone A (13.85 mg/g) and glabridin (23.27 mg/g) in TFGU (20). TFGU was suspended in 0.5% (w/v) sodium carboxymethyl cellulose to prepare the desired concentration for animal use.

Animal Experiments
Six-week-old male C57BL/6 mice (weighing 20 ± 2g) were purchased from Charles River Laboratories (Nanjing, China) and housed under controlled environment conditions with 25 ± 2°C and a 12-h dark-light cycle throughout the experimental period. The experiment was conducted in accordance with the Laboratory Animal Management Regulations, and the protocol was approved by the Animal Ethics Committee of China Pharmaceutical University (Nanjing, China). After one-week adaptation, mice were randomly divided into 6 groups (n = 6 mice/group) including control (Con, fed with vehicle), model (Mod, fed with CPT-11), CIF-and TFGU-treated groups. Besides Con group, 40 mg·kg −1 CPT-11 was intraperitoneally injected in mice to incur colitis. CIF (40 mg·kg −1 ) and TFGU (135 mg·kg −1 ) were given orally at day 1 before the colitis model established, while these treatments were 30 min prior to the CPT-11 administration from day 2 to 10. Mice in Con and Mod groups were given vehicle (i.e., 0.5% sodium carboxymethyl cellulose) in the same manner. Body weight was measured every day. At day 10, the fresh feces were collected, and the colon tissue was quickly removed by cutting at the pubis symphysis and at the cecum and the entire colon length was measured.

Disease Activity Index (DAI) Calculating and Histological Analysis
DAI was calculated as the accumulated value of the following three parameters: a) body weight loss (0 point = no loss, 1 point = 1-5% loss, 2 point = 5%-10% loss, 3 point = 10%-15% loss, 4 point = over 15% loss); b) diarrhea (0 point = normal, 2 point = loose stools, 4 point = watery diarrhea); and c) hematochezia (0 point = no bleeding, 2 point = slight bleeding, 4 point = gross bleeding). The formalin-fixed colonic tissues were sliced at 5 mm thickness and then subjected to hematoxylin and eosin (H&E) staining to assess: 1) leukocyte infiltration, 2) vascular congestion and erosion, and 3) anabrosis of epidermal cells. Each of the three parts was scored from 0 to 4 according to the severity of intestinal inflammation. The histological scores were accumulated from the scores of the three parts.

Quantitative Polymerase Chain Reaction (qPCR) Analysis
Total RNA was isolated with the Trizol agent. cDNA was synthesized following a program for cDNA elongation. The amplification program of real-time qPCR was performed as follow: 1 cycle of 95°C for 90 s followed by 39 cycles of 95°C for 10 s, 60°C for 30 s, and 72°C for 30 s. 65°C to 95°C was heated at a gradient of 0.5°C for 5 s as a dissolution curve. The primer sequences were shown in Table 1.

Enzyme-Linked ImmunoSorbent Assay (ELISA)
The homogenates of colonic tissues were performed on ice: 500 ml RIPA lysate (containing 1 mM PMSF) was added to 10 mg minced colonic tissue, cell lysis on a cell disrupter for 30 s, centrifuged at 4°C and 13000 rpm for 10 min, and the upper liquid was taken. The proteins of colonic tissue homogenates were detected by the BCA ™ protein assay kit. The protein levels of pro-inflammatory cytokines such as TNF-a, IL-1b, and IL-6 in colonic tissues were quantified by ELISA kits according to the manufacturer's instructions.

Western Blotting Analysis
The loading quantity of protein was diluted to 20 to 50 mg according to the results of BCA ™ protein assay kit. And then, mixed with protein loading buffer at a ratio of 1:4. After vortexed, the proteins were heated in a 100°C for 3 to 5 min to fully denaturation. All samples were stored at −80°C before use. The procedure of Western blotting was referred to the instructions for details. Gray value analysis of gel strips was performed by ImageJ using GAPDH as the internal reference.

Illumina PE250 High Throughput Sequencing
The total DNA of the feces were extracted by the DNA extraction kit. The DNA extraction was subjected to PCR amplification of the 16S rRNA gene V3-V4 region and high-throughput sequencing of Illumina PE250 by Shanghai Lingen Biomedical Technology Co., Ltd. (Shanghai, China). Specific primers with barcode were synthesized for the 16S rRNA gene. The PCR amplification was carried out using TransStart Fastpfu DNA polymerase kit. The program as follow: one cycle of 95°C for 300 s followed by 27 cycles of 95°C for 30 s, 55°C for 30 s, and 72°C for 45 s, together with three replicates per sample. The PE reads were first spliced according to the overlap relationship. Meantime, the quality of the sequence was quality-controlled and filtered, and the OTU cluster analysis and species taxonomic analysis were performed after distinguishing the samples. Based on the above analysis, alpha diversity and beta diversity indices were calculated with QIIME (Version 1.7.0). To mine the microbial diversity difference among samples, significance test was performed by linear discriminate analysis effect size (LEfSe). To predict the functional profiles of the microbial communities, phylogenetic investigation of communities by reconstruction of unobserved states (PICRUSt) analysis was performed, estimating the functions based on 16S rRNA sequencing data, Kyoto Encyclopedia of Genes and Genomes (KEGG) databases, and KEGG Orthology (KO) databases.

Fecal Metabolomics Analysis
The effects of TFGU on the fecal metabolism in CPT-11-induced colitis mice were assayed by a GC-MS-based untargeted

Gene
Primer sequence metabolomics. Fecal samples were thawed at room temperature and 50 mg samples were homogenized for 10 min in tenfold volumes of ultrapure water and methanol, respectively. After 10 min of centrifugation (13,000 rpm, 4°C), 200 ml of each supernatant was blended and spiked with 10 µl methanol, which contains 1,2-13 C myristic acid [internal standard (IS), 300 µg/ ml], vortexed for 3 min. After concentration at 50°C for 2 h, 60 µl methoxy pyridine (10 mg/ml) was added and the mixture was vortexed for 30 s, reacted in shaker at 30°C for 1.5 h (450 rpm). Then the 60 µl derivatization reagent BSTFA was added and reacted at 37°C for 1 h (450 rpm), vortexed for 10 s. Finally, after centrifugation at 13,000 rpm for 10 min (4°C), 1 ml supernatant was injected into GC-MS. The separation was carried out on a Thermo Scientific TraceGOLD TG-5MS capillary column (0.25 mm × 30 m, 0.25 mm). The temperature programmed condition was 60°C at 0 to 1 min, then went up to 320°C in 5 min with the gradient of 20°C/min. The split was 24 ml/min, the split ratio was 20, the temperature was 300°C, and the carrier flow was 1.20 ml/min. The MS analysis was performed in a TSQ 8000 mass spectrometer via electrospray ionization (ESI) interface (Thermo Scientific Inc, Waltham, USA) in positive ion mode. The ion source temperature and the MS transfer line temperature were 300°C. The initial time of scan at 3.72 min with a mass scan range from 50 to 500 Da.

Statistical Analysis
All results were presented as mean ± standard deviation (SD). The metabolomics data processing was performed with Metaboanalysis 3.0. And the orthogonal partial least squares discriminant analysis (OPLS-DA) was conducted by SIMCA-P (version 13.0, Umetrics, Sweden). Statistical evaluations were performed using independent sample T test or one-way ANOVA by SPSS 19.0 (SPSS Inc., Chicago, USA). Statistically significant differences were considered the results with P-values < 0.05. Spearman's correlation was used to show the relations between parameters. The correlation coefficient is always between −1 and +1. The linear relationship will be better if the absolute value of correlation coefficient is closer to 1.

TFGU Attenuated CPT-11-Induced Colitis in Mice
Oral administration of TFGU could significantly improve the weight loss of colitis mice induced by CPT-11 (P < 0.01), similar with the positive drug CIF ( Figure 1A). Considering other features of diarrhea and visible fecal blooding, DAI was calculated and indicated that CPT-11 successfully caused colitis and TFGU and CIF decreased the colitis-related symptoms ( Figure 1B). CPT-11 typically caused colonic shortening while such change was significantly improved by TFGU as well as CIF (P < 0.01 or P < 0.05, Figures 1C, D). Histopathological analysis showed CPT-11 lead to complete disruption of tissue architecture as evidenced by erosion of the mucosa and submucosa, loss of intestinal crypt structure, and massive infiltration of inflammatory cells (Figure 2A). Administration of TFGU and CIF partly rescued the destructed tissue architecture, as evidenced by restoration of crypt structure and reduction of inflammatory cell infiltration. Quantitatively, CPT-11 decreased the length of colon villus and increased histopathological score, whereas TFGU and CIF turned these numbers around (Figures 2B, C).
To examine the effect of TFGU on the pro-inflammatory cytokines in CPT-11-treated mice, the mRNA and protein levels of TNF-a, IL-1b, and IL-6 were measured in colonic tissues  ( Figures 3A, B). Compared with the Con group, CPT-11 significantly elevated the mRNA expressions of TNF-a and IL-1b in the colonic tissues (P < 0.01), which were reversed by TFGU significantly (P < 0.01 or P < 0.05). The IL-6 mRNA level has no significant difference between groups ( Figure 3A). As shown in Figure 3B, TNF-a and IL-6 were remarkably upregulated after CPT-11 challenge (P < 0.01 or P < 0.05). After TFGU treatment, these pro-inflammatory cytokines were significantly downregulated (P < 0.01 or P < 0.05), indicating that TFGU could attenuate the colonic inflammation in colitis mice induced by CPT-11.

TFGU Regulated the Gut Microbial Structure and Function in CPT-11-Treated Mice
In order to examine whether the protective effect of TFGU is associated with gut microbiota, we sequenced the bacterial 16S rRNA V3-V4 region in feces. Venn diagram showed that 181 OTUs were present in all groups, while 18, 7, 17, 7 OTUs uniquely present in Con group, Mod group, CIF group, TFGU group, respectively (Supplementary Figure 3). The value of observed species in the Mod group was significantly lower than that in the Con group (P < 0.01), whereas TFGU group possessed similar observed species with the Con group ( Figure 4A). As expected, CIF further decreased observed species compared with the Mod group, which was consistent with its properties to deplete gut microbiota as a broad-spectrum antibiotic. Likewise, alpha diversity indices for Shannon and Simpson returned to normal levels in TFGU group while CIF exhibited lower alpha diversity indices than the Mod group ( Figures 4B, C). The beta diversity was assessed by PCoA on weighted UniFrac distance matrices, which revealed that TFGU obviously reversed the overall structure of gut microbiota to normal, whereas CIF contributed substantially to a phylogenetically unique microbiota distinct from other groups ( Figure 4D). The composition of the gut microbiota of each group at the level of phylum was shown in Figure 4E. CIF exhibited a significantly negative impact on the Bacteroidetes/Firmicutes ratio compared with other groups (P < 0.01, Supplementary Figure 4)   obvious influence on the Bacteroidetes/Firmicutes ratio compared with Con group. There was a significantly lower abundance of bacteria belonging to the order Lactobacillales in Mod group (P < 0.01), which comprises the lactic acid bacteria with well-known probiotic properties (21), whereas it showed an increasing trend in TFGU group with no significant difference in comparison with Con group (Supplementary Figure 5A). At the genus level, CPT-11 treatment decreased the abundances of several SCFAsproducing bacteria (including Odoribacter, Enterorhabdus and Ro se buria ) and probiotics (mainly La c t o b a c i l l u s) (Supplementary Figures 5B, C), but allowed the proliferation of Muribaculum in mice (Supplementary Figure 6). The abundances of Muribaculum, Roseburia, Anaerotruncus and Lactobacillus were call-backed to some extent after TFGU treatment, suggesting a more robust activation of the adaptive immune system (22). We then used the LDA ( Figure 4F) and LEfSe (Supplementary Figure 7) analyses for microbial biomarker discovery in all groups. The b-glucuronidase secreting bacteria (including Bacteroides) as well as inflammation-promoting bacteria (mainly Peptostreptococcaceae) (23) acted as the representative characters in CPT-11-treated mice; SCFAs-producing genera including Blautia and Alistipes, as well as Papillibacter were characteristic in TFGU group; and CIF exclusively presented Lachnoclostridium, Akkermansia, Anaeroporobacter, and Klebsiella. The PICRUSt analysis used to predict whether TFGU treatment could modulate the functional profile of gut microbiota. Compared with other groups, TFGU group has come to resemble more closely that of the Con group at KEGG levels II and III ( Figures 5A, C), including a decrease in bacterial pathogenesis (i.e., cell motility, bacterial secretion system, bacterial motility proteins) (Figures 5B, D). However, an  increased capacity for basic metabolism (carbohydrate metabolism) was observed in Mod and CIF groups. In these cases, TFGU could modulate the gut microbiota of CPT-11-induced colitis mice, resulting in gut microbiota structure and function back to that of normal mice.

TFGU Has Significant Callback Effect on Fecal Metabolic Disorders Induced by CPT-11
Since TFGU could change the gut microbial structure and function, we examined its effect on fecal metabolism in CPT-11-treated mice. As shown in Figure 6A, a clear grouping trend among the Con, Mod, CIF, and TFGU groups could be observed in positive ion mode (R2X: 0.859, Q2: 0.436) in OPLS-DA plot. Particularly, TFGU could reverse the abnormal metabolism of endogenous metabolites induced by CPT-11. CIF significantly disturbed the fecal metabolism in colitis mice. The OPLS-DA score scatter plots displayed a clear separation between the Con and Mod groups in positive ion mode (R2Y: 0.969, Q2: 0.855) ( Figure 6B). The results of 300 permutations exhibited no overfitting in OPLS-DA models ( Figure 6C). Twenty endogenous metabolites were found in the fecal sample met the conditions of P < 0.05 and variable importance in project (VIP) > 1 between Con and Mod groups ( Figure 6D). The effective of the TFGU on the regulation of some differential metabolites (including Nmethylalanine, uracil, hypoxanthine, lysine, N-acetyl-Dmannosamine and uric acid) in the feces were listed in Figure 7.

Potential Relations Between Fecal Metabolites and Gut Microbiota
To comprehensively analyze the relations among fecal metabolites and gut microbiota, a correlation matrix was generated by calculating the Spearman's correlation coefficient. As shown in Figure 8, aromatic amino acids (including tyrosine and phenylalanine) and branched-chain amino acids (i.e., valine, leucine and isoleucine) were negatively correlated with Roseburia, Lactobacillus, family Lactobacillaceae, class Bacilli, and order Lactobacillales. Seven bacterial strains, including phylum Verrucomicrobia, class Verrucomicrobiae, order Verrucomicrobiales, family Peptostreptococcaceae and  7 | Effects of TFGU on peak areas of the potential metabolites of between the Mod and Con groups for fecal samples. Date are shown as mean ± SD (n = 6 mice/group). # P < 0.05, ## P < 0.01 vs. Con group, while *P < 0.05, **P < 0.01 vs. Mod group.
Akkermansiaceae, genera Akkermansia and Romboutsia, had positive correlations with nineteen metabolites except for thymine. Additionally, uric acid was negatively correlated with Lactobacillus. These relations suggested that gut microbiota could affect fecal metabolites in CPT-11-induced colitis mice.

TFGU Could Significantly Inhibit the Colonic Expression of the NLRP3 Inflammasome Complex Activated by CPT-11
Massive evidence has indicated that uric acid could activate inflammasome NLRP3, while the activation of NLRP3 inflammasome was one of the pathological mechanisms of CPT-11-induced delayed diarrhea (1). So, it is necessary to investigate the protective effect of TFGU against CPT-11induced colitis through the NLRP3 inflammasome complex. As depicted in Figure 9A, CPT-11 could activate inflammasome NLRP3, promote caspase-1 to cleaved the precursor of IL-1b and release a large amount of mature IL-1b into colonic tissue, resulting in colitis. The activation of NLRP3, Cleaved caspase-1/Procaspase-1, and Cleaved IL-1b/ProIL-1b induced by CPT-11 was significantly inhibited after TFGU treatment ( Figure 9B, P < 0.01 or P < 0.05). The above results indicated that the TFGU could significantly inhibited the expression of related proteins in the inflammasome NLRP3 pathway.

DISCUSSION
Over the past decade, the growing body of experimental evidence has pointed to the beneficial actions of plant flavonoids at the gastrointestinal tract (25). In the present study, TFGU attenuated CPT-11-induced body weight loss and colon length shortening, but also down-regulated the pro-inflammatory cytokines (including TNF-a, IL-1b, and IL-6) in colonic tissues so as to partially inhibit intestinal mucositis. The involvement of gut microbiota in the gastrointestinal toxicity of CPT-11 has been reported (26). In our study, CPT-11 significantly decreased microbial diversity in mice, thereby inciting gut microbial dysbiosis and intestinal inflammation. Since b-glucuronidase plays a very important role in CPT-11-induced colitis, the bacterial species shown to exhibit b-glucuronidase activity mainly belong to the colitogenic Bacteroides (Bacteroidetes phylum) and Clostridium (Firmicutes phylum) (27,28) were expanded in CPT-11-induced colitis mice. The maladjustment of these bacteria was partially corrected by TFGU treatment. In addition, TFGU could increase Lactobacillus and butyrateproducing Roseburia, both of which were beneficial for intestinal homeostasis in colitis (29,30). Notably, SCFAs including butyrate can maintain the integrity of gut epithelial barrier through modulating immune response and oxidative stress (31), the content of which were significantly reduced in ulcerative colitis patients (32). CIF as antibiotics has shown efficacy in the treatment of CPT-11-induced colitis, whether this effect is due to depletion of gut microbiota or some other mechanism is still up to debate. We found that the negative impact of CIF on gut microbiota was partially associated with an increased ratio of Firmicutes to Bacteroidetes, which is a hallmark feature of the gut microbial dysbiosis (33).
Besides gut microbial dysbiosis, CPT-11 also caused significant fluctuations in the levels of hypoxanthine and uric acid in the purine metabolism. Uric acid is a ligand of the NLRP3 inflammasome that induces inflammatory mediators such as IL-1b and IL-18 and is thus associated with inflammatory disorders (34). The elevated uric acid in the intestine during an inflammatory response could exacerbate intestinal disease. Treatment with uric acid alone worsened disease and increased gut permeability in dextran sulfate sodium-induced colitis mice (35). In our study, TFGU treatment significantly reduced the fecal uric acid in colitis mice, consistent with the downregulation of purine catabolism. In a previous study, isoliquiritigenin among TFGU potently inhibited the activation of NLRP3 inflammasome, and its inhibitory effect was stronger than that of parthenolide, a known inhibitor of the NLRP3 inflammasome (36). We found that TFGU obviously inhibited the activation of NLRP3 inflammasome in CPT-11-induced colitis mice. These findings support the interpretation that the therapeutic effect of TFGU on CPT-11-induced gastrointestinal toxicity is partially mediated by a decreased fecal uric acid as well as NLRP3 inflammasome inhibition. Noteworthy, since hypoxanthine and uric acid are host-microbial co-metabolites, their pool can be affected by both microbial and host metabolism (Supplementary Figure 8B). The uric acid lowering effect of TFGU seems independent with gut microbial purine metabolism based on the PICRUSt results. Recently, Lactobacillus gasseri was shown to decrease purine levels in the intestine, and Lactobacillus supernatant had urate-lowering effects (37). Our correlation results also revealed that Lactobacillus abundance negatively correlated with fecal uric acid concentration. To rule out the possibility that difference in uric acid concentration was mainly caused by bacteria, fecal microbiota transplantation is expected to conduct.
It has been reported that five metabolic pathways (phenylalanine, tyrosine and tryptophan biosynthesis, glycine, serine and threonine metabolism, alanine, aspartate and glutamate metabolism, phenylalanine metabolism, and primary bile acid biosynthesis) in serum were identified associated with CPT-11 exposure (19), comparable with our fecal metabolism findings. The previous literature has showed that phenylalanine, tyrosine, and tryptophan biosynthesis is related to inflammation reaction (18), which may partially explain why this metabolic pathway has the greatest impact on CPT-11-induced colitis metabolism. With regard to specific metabolites, aromatic amino acids (including phenylalanine and tyrosine) and branched-chain amino acids (i.e., valine, leucine and isoleucine) were significantly up-regulated in CPT-11-induced colitis mice, while TFGU has callback effect to alleviate the fluctuation of CPT-11 on these amino acids although with no significant difference. In a case-control study, significantly increased levels of six amino acids (histidine, tryptophan, tyrosine, phenylalanine, leucine, and valine) were found in the feces of patients with inflammatory bowel disease compared to controls (38). Similarly, greater abundance of branched-chain amino acids and a range of other amino acids (i.e. lysine, alanine, tyrosine, phenylalanine, and glycine) in the fecal samples from patients with active Crohn's disease and ulcerative colitis were found when compared to controls (39). Whether these differences result from malabsorption/increased loss due to inflammation or reflect an increase of the producing bacteria needs to be elucidated.

DATA AVAILABILITY STATEMENT
The data sets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ Supplementary Material.

ETHICS STATEMENT
The animal study was reviewed and approved by the animal ethics committee of China Pharmaceutical University (Nanjing, China).