L. plantarum Q7 (GenBank: CP019712-16) was isolated from traditional fermented yak yogurt made by herdsmen in Qinghai province, China (21). The strain was cultured in MRS medium at 37°C overnight. Q7-EVs were isolated from the culture supernatants using ultrafiltration combined with ultracentrifugation (22). In brief, the culture supernatants were centrifuged at 8000×g for 30 min to remove cell debris and impurities. After filtered by 0.22 μm filter, the supernatant was concentrated using centricon-plμs-70 100-kDa ultrafiltration tube (Millipore) and ultracentrifuged at 100000×g for 2 h (Hitachi, Ltd., Tokyo, Japan). The precipitate was aspirated gently by adding PBS and centrifuged at 100000×g for 1 h, then resuspended the pellet with PBS. Q7-EVs were collected and stored at -80°. The morphology and the size distribution of Q7-EVs were observed by Transmission electron microscope (H-7650, Hitachi, Ltd., Tokyo, Japan) and Dynamic light scattering (Nanotrac Wave II, Microtrac, Inc., Montgomeryville, USA).

4-5 weeks old SPF male C57BL/6J mice were purchased from the Laboratory Animal Breeding Center of Pengyue (Jinan, China). Mice were housed in an animal care facility and provided with sufficient food (Beijing Keao Xieli Feed Co., Ltd., Beijing, China) and water. Colitis was induced by 3.5% dextran sulfate sodium (DSS, Yeasen Biotech, Shanghai, CN) administration in the drinking water (23) and the mice were treated Q7-EVs by gavage (23, 24). The mice were acclimated for a week and randomly divided into four groups: control group (PBS, daily), DSS+PBS group (PBS, daily), DSS+10 μg Q7-EVs group (0.5 mg/kg body weight Q7-EVs, daily), DSS+20 μg Q7-EVs group (1 mg/kg body weight Q7-EVs, daily). The mice in control group were administered untreated drinking water for 18 days. The DSS group and treatment group were pre-garaged with PBS or Q7-EVs for 10 day and received 3.5% DSS in drinking water from 11^th day to 18^th day. All experimental processes were approved by Animal Ethics Committee of Ocean University of China (permission number: spxy20200720215).

Body weight, stool characteristic and rectal bleeding were recorded every two days. DAI scores were calculated according to previously described methods (25–27). The details of each score were listed in Supplementary Materials ( Table S1 ). The weight of spleen was determined, and the spleen index was defined as the ratio of spleen weight to body weight as previous report (28).

Colon tissue was fixed in 4% paraformaldehyde, dehydrated with gradient alcohol solution and embedded in paraffin. Paraffin-embed samples were cut into continuous longitudinal sections, and then stained by H&E. The histological scoring was measured according to the criteria listed in Table S2 (29).

According to the manuscript’s protocol, the expression level of IL-6, IL-1β, and TNF-α in the serum was determined by ELISA kit (Nanjing Jiancheng Biotechnology, Nanjing, China). Multiskan FC (Thermo Scientific, Waltham, MA, USA) was used to detect the absorbance of each sample at 450 nm.

The colon tissue was put into Trizol reagent and homogenized using a tissue grinder (Wuhan Sevelco Co., Ltd., Wuhan, China). 1 μg RNA was reverse transcribed via a high-capacity cDNA reverse transcription kit (Applied Biosystems, Foster City, USA). Genes were quantitated using the StepOnePlus real-time PCR instrument (Applied Biosystems, Foster City, USA). The specific primers used were shown in Table S3 . The 2^-ΔΔCt method was utilized to normalize the expression results. The glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as the housekeeper gene to normalize each mRNA.

Fecal DNA extraction and high-throughput sequencing were performed by Personal Biotechnology Co., Ltd (Shanghai, China). Briefly, the Fast DNA SPIN extraction kit (MP Biomedicals, Santa Ana, CA, USA) was used to extract genomic DNA from mice feces. The highly variable V3-V4 region of the bacterial 16S rDNA gene was selected for amplification. The library was qualified based on the TruSeq Nano DNA Library Prep Kit from Illumina, and Illumina NovaSeq platform was used for paired-end sequencing. The filtered sequences were clustered as operational taxa (OTU) using the UCLUST algorithm with a similarity threshold of 97%. The OTU abundance and the six-level taxonomic classification from phylum to species were obtained for analysis.

The experimental data were analyzed by GraphPad Prism version 7.0 and SPSS 25.0 statistical software. Comparisons of various anatomical measurements by two-way analysis of variance (ANOVA), followed by Fisher’s LSD post hoc tests. PCA and PCoA analyzed were used R language pack. The linear discriminant analysis (LDA) was applied to identify the differences of microbiota compositions between different groups by using an LDA score threshold of >3.5.

EVs were isolated from the supernatant of L. plantarum Q7 by ultrafiltration combined with ultracentrifugation ( Figure 1A ). The Q7-EVs were characterized by Transmission electron microscope and Dynamic light scattering. The results showed that Q7-EVs were double-layer membrane-enclosed nanoparticles with spherical morphology ( Figure 1B ). It was observed that Q7-EVs ranged from 70 nm to 500 nm in size ( Figure 1C ), the mean size was 185.5 ± 65.4 nm, which was consistent with the size of previous reports (30).

The effect of Q7-EVs on DSS-induced colitis was evaluated in C57BL/6J mice ( Figure 2A ) (31). The body weight, DAI, length of the colon, and colonic histomorphology were analyzed. Compared with the control group, the weight loss could be slowed by Q7-EVs intervention since the 6^th day. The weight loss of 22.6% was observed in the DSS+PBS group, while the weight loss of 12.8% and 12.6% were found in the 10 μg and 20 μg Q7-EVs group on the 8^th day, respectively. The weight loss could be slowed by Q7-EVs intervention with no dose correlation ( Figure 2B ). DAI scores of Q7-EVs (10 μg and 20 μg) treated mice were less than that of DSS+PBS group mice on the 8^th day (p < 0.05) ( Figure 2C ). Colon length was regarded as a key indicator to assess the severity of acute colitis induced by DSS (31, 32). It was confirmed that Q7-EVs treatment improved the shortening of the colon significantly (p < 0.05) ( Figures 2D, E ). The colon length in the DSS+PBS group was 4.16 ± 0.24 cm. After treatment with Q7-EVs, the average length of the colon was observed to be 5.14 ± 0.07 and 5.19 ± 0.34 cm, respectively. The spleen is one of important immune organ of the body, which exerted a regulatory effect on the immunity. It was found that the Q7-EVs reduced the spleen index of DSS-induced colitis mouse model ( Figure S1 ).

H&E staining showed that DSS treatment caused damage to the colon structure, while the pathological features were improved by oral Q7-EVs ( Figure 3A ). It suggested that Q7-EVs could reduce the colonic pathology score of mice and alleviate colitis in mice significantly ( Figure 3B ).

It was observed that there was no significant difference in the effects of two doses of Q7-EVs (10 μg and 20 μg) on the body weight, colon length and colonic pathology score. Therefore, 10 μg Q7-EVs was selected in the following experiments. It was reported that the increase of pro-inflammatory cytokines was a hallmark of DSS-induced colitis (33). The level of inflammatory cytokines in colon tissues was detected (34). It was found that the expression levels of IL-6, IL-1β, TNF-α, and IL-2 were increased in the colon tissue of the DSS+PBS group mice. In contrast, these cytokines were significantly reduced in the colon under the intervention of Q7-EVs ( Figure 4 ). A similar trend was found for the determination of IL-6, IL-1β, and TNF-α in serum ( Figure S2 ).

It was found that microbial richness, evenness and diversity were significantly decreased in the DSS+PBS group in comparison with that of the control group, whereas Q7-EVs could restore the species diversity, richness and community evenness ( Figures 5A–F ).

Venn diagram was used to display the composition information of species in each group. It was observed that 332 OTUs overlapped in all three groups, 197 OTUs were present in both the control and DSS+PBS groups, 279 OTUs in the DSS+PBS and Q7-EVs groups, and 1163 OTUs in the control and DSS+Q7-EVs groups ( Figure 5G ). Compared with the DSS+PBS group, intestinal microbial composition in the DSS+Q7-EVs group was more similar to the control group. The results showed that the gut microbiota of the DSS+PBS group was significantly different from that of the control group, and Q7-EVs alleviated the shift of gut microbiota induced by DSS ( Figures 5H, I ). LEfSe analysis was demonstrated that 37 key phylotypes were determined as distinguished biomarkers on the genus ( Figure 6A ). The control group had 10 dominant microorganisms, the DSS+PBS group and the Q7-EVs group had 14 dominant microorganisms, respectively ( Figure 6B ).

The composition of the gut microbiota at the phylum level was shown in Figure 7A . Compared with the control group, Bacteroidetes, and Verrucomicrobia in the DSS+PBS group were significantly decreased, while Proteobacteria, Epsilonbacteraeota were increased significantly. These changes of gut microbiota were recovered in the Q7-EVs group. The abundance of Proteobacteria was decreased in the DSS+Q7-EVs group. The Firmicutes/Bacteroidetes (F/B) ratio in the DSS+PBS group was higher than that of the control group, and was restored after the intervention of Q7-EVs ( Figure 7B ), which could be used as a biomarker for intestinal inflammation (31). The difference of gut microbiota in each group at the genus level was shown in Figure 7C . Compared with the DSS+PBS group, Bifidobacterium, Rikenellaceae_RC9_gut_group, Akkermansia, Muribaculaceae, Lactobacillus and Alitipes in the DSS+Q7-EVs group were recovered to the level of the control group ( Figure S3 ).

Correlation analysis was conducted to determine the relationship between the gut microbiota and inflammatory cytokines. Proteobacteria (Helicobacter, Escherichia-shigella, Klebsiella, Oscillibacter) and Firmicutes (Clostridium_sensu_stricto_1) were positively correlated with inflammatory cytokines. Muribaculaceae, Bacteroides, Akkermansia, Rikenellaceae_RC9_gut_group, Faecalibaculum, Bifidobacterium, Alistipes, Eubacteriu_fissicatena_ group, Dubosiella, Ruminococcaceae_UCG-014 and Anaeroplasma were negatively correlated with inflammatory cytokines ( Figure 7D ). Combined with the LEfSe analysis, the dominant microorganisms in the DSS+Q7-EVs group contained more anti-inflammatory bacteria.