The mouse model of radiation-induced intestinal injury was constructed by previously described methods (22). The C57BL/6 mice and Stat3^△IEC mice were gavaged with B. fragilis (1 × 10⁹ CFU per mouse) or B. fragilis-derived capsular polysaccharide (PSA) (50 μg/mouse) for 14 days during the pretreatment stage, then administered to mice with radiation. Radiation was given using the irradiator (MultiRad, Faxitron, USA) at 26 Gy/min with 225 kVp X-rays using a 0.3 mm copper filter. (1) The control group is the following: Mice were treated with phosphate buffer (PBS) orally. (2) TAI group: Mice were treated with PBS, and exposed to 10 Gy total of abdominal irradiation. (3) TBI group: Mice were treated with PBS, and exposed to 4.5 Gy total body irradiation. (4) TAI/TBI + ZY-312 group: Mice were treated with B. fragilis (1 × 10⁹ CFU per mouse) through an oral route, dissolved in sterile water in 0.2 ml volume per mice for 15 consecutive days after 4.5 Gy TBI or 10 Gy TAI. All experiments were performed using 6- to 8-week-old sex-matched mice. C57BL/6 mice were purchased from SPF (Beijing) Biotechnology Co., Ltd. (Beijing, China), and STAT3 conditional gene knockout mice (Stat3^△IEC mice, the primers for gene identification were showed in Supplementary Table 2) were purchased from GemPharmatech Co., Ltd. (Jiangsu, China).

Bacteroides fragilis strain ZY-312 was cultured in 5% fetal bovine serum, 9.5 ml of tryptic soy broth, and 100 μl of passaging solution at 37°C for 24 h under anaerobic conditions.

The small intestinal tissue of mice was fixed in 4% paraformaldehyde for 24 h, embedded in paraffin, and cut into sections at a thickness of 5 μm (The histopathology associated index of intestine was evaluated though the criterion in Supplementary Table 1). For immunohistochemical staining, the intestinal sections were subjected to deparaffinization, hydration, antigen retrieval, quenching of endogenous peroxidase, and blocking procedures. All slices were then incubated with the primary antibodies against pSTAT3 (CST #9145) and Ki-67 (Abcam, England, Ab16667) at 4°C overnight followed by incubation with biotinylated secondary antibodies for 30 min and visualization using a 3,3′-Diaminobenzidine Kit (ZSGB-BIO, Beijing, China). For immunofluorescence staining for the detection of MUC2 (Abcam, England, Ab272692), Lgr5 (NBP1-28904SS), ZO-1 (Abcam, England, Ab221547), and Claudin-1 (Abcam, England, Ab242370), the colonic sections were processed by the methods described above. Periodic Acid-Schiff stain (PAS) staining (Biossci, China) was performed following the manufacturer’s protocols.

The small intestinal was isolated from mice. Then SI was washed with PBS and cut into 5 mm pieces, followed by digestion with buffer (5 mM EDTA and 2 mM dithiothreitol in Hanks balanced salt solution; Sigma, St. Louis, MO, USA) at 37°C for 30 min on a rotating platform. Through the digestion stopping stage with Hanks balanced salt solution, the solution was filtered through a 70 μm cell strainer. Lastly, the filtered solution was centrifuged for 10 min at 400 g (4°C) to obtain SI epithelial cells (23).

Western blotting was performed followed by previous protocols (24). The proteins were separated by SDS–PAGE and analyzed by immunoblotting with rabbit polyclonal antiserum to pSTAT3 (CST #9145, Danvers, MA, USA), STAT3 (CST #4904), PCNA (CST #13110), P38 (CST #9212), pmTOR (CST #4511), mTOR (CST #4511), pJAK2 (CST #4695), pNF-KB (CST#4370), Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (CST #5174), ZO-1 (Proteintech 21773-1-AP), Claudin-1 (Proteintech 13050-1-AP), MUC2 (Abcam, England, 272692-10), Lgr5 (Abcam, England, 75850) antibodies were used to measure colonic epithelial protein expression.

The fecal microbiota composition of mice was determined by 16S rRNA gene amplification. Briefly, TGuide S96 Magnetic Soil/Stool DNA Kit (Tiangen Biotech Co., Ltd., Beijing, China) was used to extract genomic DNA from feces. DNA concentration and integrity were measured by a microplate reader (GeneCompang Limited, synergy HTX, United States) and agarose gel electrophoresis, respectively. The 16S rRNA gene V3–V4 region was amplified from the genomic DNA in a 25 μl reaction using the universal bacterial primers: (338F, 50–ACTCCTACGGGAGGCAGCA–30, and 806R, 50–GGACTACHVGGGTWTCTAAT–30). The PCR products were quantified by using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, United States) and purified with Monarch^® DNA gel Extraction Kit (New England Biolabs, USA). Sequencing was performed on an Illumina NovaSeq 6000 with two paired-end read cycles of 250 bases each (Illumina Inc., San Diego, CA, United States; Biomarker Co., Ltd., Beijing, China). The Trimmomatic software v0.33 was used to analyze the raw data. After trimming, amplicon sequence variants (ASVs) were generated by using USEARCH 10.0 with a 97% similarity cutoff. The representative read of each ASV was selected by using the Quantitative Insights into Microbial Ecology (QIIME) package. All representative reads were annotated and blasted against the Silva database (Version 132) using the Ribosomal Database Project (RDP) classifier (confidence threshold was 70%). The microbial richness and diversity in fecal content samples were estimated using the alpha diversity that includes the Shannon index. The UniFrac distance matrix performed by QIIME software was used for the unweighted UniFrac Principal coordinates analysis (PCoA), and phylogenetic tree construction. Linear discriminant analysis effect size (LEfSe) based linear discriminant analysis (LDA) and cladogram were generated to assess differentially abundant microbial taxa. The 16S rRNA gene amplicon sequencing and analysis were conducted by Biomarker Technologies Co., Ltd. (Beijing, China).

Statistical analyses were performed using GraphPad Prism 8.0 (GraphPad Software). All variables are expressed as Means ± SEM, as noted in the figure legends. Two-way ANOVA was used to compare body weights. The one-way ANOVA with Tukey’s tests was used for comparisons among three or more groups. While non-parametric data were analyzed with a Mann–Whitney U-test. P < 0.05 was considered as significance.

To evaluate the effect of B. fragilis strain ZY-312 on radiation-induced intestinal injury, we constructed the corresponding mouse model for total body radiation (TBI) and administered the mice with B. fragilis orally (Figures 1A,B). Compared with the TBI group, mice receiving B. fragilis (TBI + ZY-312 group) had significantly lower weight loss (Figure 1C) and slower SI and colon length shortening (Figure 1D). To explore the effect of B. fragilis on the IECs, the histopathological results revealed that B. fragilis supplementation attenuated SI epithelial cell damage (Figure 1E), reduced overall HAI, and resulted in longer lengths of SI crypts (Figure 1F). Furthermore, we found that B. fragilis promoted SI mucus secretion in radiation-induced intestinal injury (Figure 1G). Concurrently, we constructed a radiation-induced intestinal injury mouse model for total abdominal radiation (TAI) and administered the mice with B. fragilis and B. fragilis-derived capsular polysaccharide (PSA) (Figures 1A,B). Compared with the TAI group, mice receiving B. fragilis (TAI + ZY-312 group) had significantly lower weight loss (Figure 2A) and slower intestinal (SI and colon) length shortening (Figures 2B–C). To explore the effect of B. fragilis and PSA on the IECs, the histopathological results revealed that B. fragilis supplementation and PSA obviously attenuated SI epithelial cell damage (Figure 2D), reduced overall histopathology-associated index (HAI) (Figure 2E), and resulted in longer lengths of SI crypts (Figure 2F). Thus, B. fragilis may alleviate abdominal radiation-induced intestinal injury in mice through B. fragilis-derived PSA. And it suggested that B. fragilis promoted the integrity of SI epithelium to alleviate radiation-induced intestinal injury in mice.

Leucine-rich repeat-containing G-protein coupled receptor 5 (LGR5+) ISCs cells are considered active multipotent intestinal stem cells because they have the capacity to divide and differentiate into different types of intestinal epithelial cells, such as goblet cells and enterocytes (11). Thus the degree of stem cell injury is an important indicator for damage repair after radiation. In addition, intestinal epithelial tight junctions, including zonula occludens-1 (ZO-1), occludins, and claudins, are also highly sensitive to ionizing radiation with the disruption of IECs (25). Next, we attempted to determine whether B. fragilis has the capacity to alleviate the destruction of SI IECs after radiation. B. fragilis was found to promote SI IECs proliferation (Figure 3A). Regarding the types of IECs affected, the small intestinal histological results showed that B. fragilis promoted stem cell proliferation (Figure 3B), goblet cell secretion (Figure 3C), and tight junction protein expression (Figure 3D). This indicated that B. fragilis promoted stem cell and goblet cell regeneration, and accelerated tight junction repair in radiation-induced intestinal injury in mice.

Certain types of intestinal microbiota reside in the mucus layer (26), and mucus has the capacity of resisting pathogenic bacteria in colitis (27). B. fragilis has a positive influence on mucus secretion, which suggests that it may alter the intestinal microbiota, thus relieving radiation-induced intestinal injury in mice. Hence we analyzed intestinal microbiota by 16S rRNA gene sequencing. The main intestinal microbiota phyla were Firmicutes, Bacteroidetes, and Proteobacteria. And the microbial community bar plot showed the relative abundance of phylum Firmicutes increased and that of phylum Bacteroidetes decreased gradually in the TAI + ZY-312 group compared to the TAI group, although there was no statistically significant difference (Figures 4A,B). At the genus level, the genera of the Lachnospiraceae_NK4A136_group producing butyrate increased in the TAI + ZY-312 group compared to the TAI group, despite it did not show a statistically significant difference (Figures 4C,D). Furthermore, we analyzed the microbiota diversity (Shannon), and conducted principal component analysis (PCA), which reflects the difference in gut microbiota between and within groups. We found that the microbiota diversity between the TAI and TAI + ZY-312 groups did not change significantly (Figures 4E,F). And the BugBase results showed that the abundance of anaerobic microbiota was increased in the TAI and TAI + ZY-312 groups compared to the control group (Figure 4G). LEfSe and phylogenetic tree are analytical methods that could screen the statistical difference biomarker among different groups. The LEfSe results showed that the abundance of the Clostridiales_vadinBB60_group was increased in the TAI + ZY-312 group (Figures 4H,I). It indicated that B. fragilis participated in modulating intestinal microbiota, although it did not significantly alter the intestinal microbiota in microbiota abundance and diversity.

It has been reported that the short-chain fatty acids (SCFAs), including butyrate, can activate downstream JAK/STAT (28), p38 (29), NF-κB (30), PI3K/AKT (31), Wnt/β-catenin (32), and other signaling pathways to promote intestinal mucosa proliferation and repair. B. fragilis increased the abundance of the genera of the butyrate-producing Lachnospiraceae_NK4A136_group, which are beneficial for intestinal mucosa regeneration. Next, we attempted to explore the underlying mechanism of B. fragilis in the intestinal mucosa repair after radiation injury. Thus, we detected some classic proliferation pathways including signal transducer and activator of transcription (STAT3) (33), pmTOR, pNF-κB, and p38 signaling. The western blot results showed that B. fragilis activated STAT3 phosphorylation in the intestinal tissue, while B. fragilis did not affect the other classic proliferation pathways, such as pmTOR, pNF-κB, and p38 (Figure 5A). Meanwhile, we found that B. fragilis increased the expression of the proliferation index PCNA, goblet cell marker MUC2, stem cell indicator Lgr5, and tight junction protein claudin-1 (Figure 5A).

The regulation of STAT3 is complex, as it is involved in signal transduction pathways in numerous cell types under various conditions. Pickert et al. (34) reported that interleukin (IL)-22 motivated STAT3 signaling pathway in IECs to repair mucosal wound in mice with experimental colitis. Notably, STAT3 phosphorylation in immune cells may exert different inflammation effects in colitis (35–37). We attempted to identify the specific layer in the small intestines where STAT3 phosphorylation is triggered after B. fragilis administration in radiation-induced intestinal injury. We found that B. fragilis mainly promoted STAT3 phosphorylation in the SI IECs but not in the non-epithelial layer (Figures 5B–D), suggesting that B. fragilis upregulated the STAT3 signaling pathway in SI IECs. It suggested that B. fragilis activated STAT3 phosphorylation with the upregulation with PCNA, MUC2, Lgr5, and claudin-1 in SI IECs.

To verify whether B. fragilis promoted SI IECs regeneration through the STAT3 signaling pathway, we constructed a radiation-induced intestinal injury model in Stat3^△IEC (STAT3 defects in SI epithelium) mice and administered the mice B. fragilis orally (Supplementary Figure 1). Compared with the Stat3^△IEC mice receiving B. fragilis (TAI + ZY-312/Stat3^△IEC group), wild-type mice receiving B. fragilis (TAI + ZY-312/WT group) had significantly lower weight loss (Figure 6A), slower intestinal (SI and colon) length shortening (Figure 6B), attenuated SI epithelial cells damage (Figure 6C), reduced overall HAI (Figure 6D), and had longer lengths of SI crypts (Figure 6E). This suggested that B. fragilis relieved radiation-induced intestinal injury through the STAT3 signaling pathway in SI IECs.

We found that SI IECs in Stat3^△IEC mice after abdominal radiation injury had a very low expression of pSTAT3, while the SI IECs from wild-type mice receiving B. fragilis after abdominal radiation injury had a higher expression of pSTAT3 (Figure 7A). In this context, we discovered that mice receiving B. fragilis in WT mice (TAI + ZY-312/WT group) upregulated the expression of PCNA, MUC2, Lgr5, ZO-1, and claudin-1 (Figures 7A,B), while mice receiving B. fragilis in Stat3^△IEC mice (TAI + ZY-312/Stat3^△IEC group) showed little expression of the markers above (Figures 7A,B). This suggested that B. fragilis requires the STAT3 signaling pathway to promote SI IECs proliferation, stem cell regeneration, mucus secretion, and tight junction proteins expression.

Overall, our results showed that B. fragilis promoted intestinal barrier integrity through the STAT3 signaling pathway in a mouse model of radiation-induced intestinal injury (Figure 8).