The L. plantarum G83 (CCTCC M2016245) was isolated from giant panda feces. It is available at the China Center for Type Culture Collection (CCTCC, Wuhan, CHN).

The feces samples were collected from three healthy sub-adult pandas (Chengdu Research Base of Giant Panda Breeding, CHN) and mixed. The fecal inoculum was prepared as described by Zeng (32). In brief, 140 ml pre-reduced PBS was added into 60 g mixed fresh fecal sample. The inoculum was homogenized by vortex mixer. Thereafter, the inoculum was sequentially passed through 2.0-, 1.0- and 0.5-mm diameter stainless steel strainers to remove larger particles and to concentrate filtrate to 10 ml. The giant panda microbiota inoculum (PFM) was prepared before gavage daily.

The experiment was completed under the supervision of the Institutional Animal Care and Use Committee of the Sichuan Agricultural University. Seven-week-old female SPF-C57BL/6J mice, purchased from the Beijing Vital River Laboratory Animal Technology Co. Ltd, were separated into five groups equally. Three mice were housed per cage and provided with free access to sterile distilled water and food. One group of mice was set up as a control group (Blank). The remaining mice were provided sterile water with ampicillin 600 mg/L, vancomycin 600 mg/L, and neomycin 600 mg/L from days 1 to 7. After the antibiotics cocktail treatment, two different methods (A and B) were used to generate the giant panda microbiota-associated (PMA) mouse ( Supplementary Figures 1A, B ). In method A, the mice were gavaged with 0.2 ml/day panda feces microbiota inoculum three times (on days 8, 9, and 10) after sterile water was provided (PMA mouse, BlankT group). We continued monitoring the body weight from days 0 to 13. Fecal pellets were collected on days 0, 7, 10 and 13. In method B, the mice were gavaged with the microbiota inoculum as mentioned in method A, followed by the administration of 5.0% (w/v) DSS (molecular mass 3.6–5.0 kDa; MP Biomedical, Irvine, CA, USA) in drinking water for 3 days from days 11 to 13 (DPMA mouse, BlankD group). We monitored the body weight of the subjects from days 0 to 16. Fecal pellets were collected on days 1, 7, 13, and 16.

DPMA mice were prepared following the method-B mentioned above, then the mice were separated into five different groups ( Supplementary Figure 1C ): 1) the control group (no treatment received, BlankD), 2) mesalazine treatment group (94 mg/d, switched into the equivalent dose to the mouse, TreatMz group), and 3) three 0.2 ml/d L. plantarum G83 treatment groups [namely, 1.0 × 10⁶ CFU/ml (TreatLG group), 1.0 × 10⁷ CFU/ml (TreatMG group), 1.0 × 10⁸ CFU/ml (TreatHG group)] from days 14 to 16. On day 16, the mice were euthanized and the blood, colon, spleen, payer’s patches, mesenteric lymph nodes, and colon content were collected for further analysis. Fecal pellets were collected on days 1, 7, 13, and 16. We monitored the body weight and the disease activity score from days 1 to 16.

The fecal pellets were serially diluted from 0 to 10⁻⁶ in 10-fold and the diluted samples were plated on BHI (for total bacteria), TPY (for Bifidobacteria), EC (for Enterococcus), EMB (for Escherichia coli), and MRS (for Lactobacillus) agar. EC and EMB agar plates were maintained in the aerobic incubator, while the TPY agar was incubated in an anaerobic chamber for anaerobic microorganisms. One BHI plate was maintained aerobically and the another was maintained anaerobically. Following incubation, colony-forming units (CFU) counts were performed the next day.

The protein lysate was obtained from tissue by culture cell total protein extraction reagent (BOSTER, Wuhan, CHN) and mammalian tissue protein extraction reagent (BOSTER, Wuhan, CHN), with protease inhibitor and phosphatase inhibitor cocktail, respectively. Protein concentration was determined by a BCA protein analysis kit (BIOMED, Beijing, CHN). The primary antibodies included the following: rabbit anti-β-actin (#20536-1-AP; Proteintech, Wuhan, CHN), rabbit anti-ZO-1 (#21773-1-AP; Proteintech, Wuhan, CHN), rabbit anti-occludin (#27260-1-AP; Proteintech, Wuhan, CHN), rabbit anti-Claudin 1 (#bs-1428R; Bioss, Beijing, CHN), rabbit anti-NF-κB P65 (#bs-0465R; Bioss, Beijing, CHN), rabbit anti-phospho-NF-κB p65 (#3033T; CST, Danvers, MA USA), rabbit anti-p38 (#ab27986; Abcam, Cambridge, MA, USA), and rabbit anti-phospho-p38(#ab47363; Abcam, Cambridge, MA, USA). The secondary antibody used was goat anti-rabbit (#bs-0296G; Bioss, Beijing, CHN).

The colon tissue samples were fixed in 4% paraformaldehyde fix solution and embedded in paraffin. Sections (5-μm thick) were stained with hematoxylin and eosin (HE). Colonic epithelial slices were scored following Kyoko’s method (33). The score included epithelial damage and inflammatory cell infiltration, and the total score ranged from 0 to 12. For immuno-histochemical staining, the tissues were processed following the conventional method. Subsequently, the primary antibody was added to the slide and incubated overnight at 4°C in a histochemistry staining tray. The HRP-conjugated secondary antibody was then added for 50 min, and the DAB was added for 3 min. Finally, the slides were counterstained with hematoxylin, dehydrated, and sealed for visualization. The images were captured using an optical microscope (Olympus, Tokyo, Japan), and the positively stained cells were scored by ImageJ (NIH, Stapleton, NY, USA).

Flow cytometry was performed to measure the following: CD3⁺ T cells (PerCP/Cy5.5 anti-mouse CD3, clone 17A2, Biolegend, San Diego, CA, USA), CD4⁺ T cells (CD3⁺ CD4⁺, FITC anti-mouse CD4, clone RM4-5, Biolegend, San Diego, CA, USA), CD8⁺ T cells (CD3⁺CD8⁺, PE anti-mouse CD8a, clone 53-6.7, Biolegend, San Diego, CA, USA). Th1 (PerCP-Cy5.5 anti-mouse CD4, clone RM4-5, FITC anti-mouse IFN-GMA, clone XMG1.2, BD Biosciences, San Jose, CA, USA), Th2 (APC anti-mouse IL-4, clone 11B11, BD Biosciences, San Jose, CA, USA), and Th17 (PE anti-mouse IL-17A, clone TC11-18H10.1, BD Biosciences, San Jose, CA, USA) phenotypes were also characterized. Lymphocytes were collected from the peripheral blood (PB), Peyer’s patches (PPs), mesenteric lymph nodes (MLNs), and spleen on day 16. In the surface phenotype assay, 1.0 × 10⁶ cells were blocked with 10 μl rat serum for 30 min at 4°C and stained with the indicated antibody for 30 min at 4°C in the dark. Cells for the intracellular cytokine assay were stimulated with phorbol 12-myristate 13-acetate (MilliporeSigma, Burlington, MA, USA), monensin (MilliporeSigma, Burlington, MA, USA), and ionomycin (MilliporeSigma, Burlington, MA, USA) for 4 h. The cells were labeled with surface markers, before being fixed, permeabilized, and labeled with the indicated intracellular antibody for 30 min at 4°C in the dark. Cells were acquired on the BD FACS-Verse™ flow cytometer (BD Biosciences, San Jose, CA, USA) and the data was analyzed using the FlowJo 10.0.7 software (BD Biosciences, San Jose, CA, USA).

About 0.7 g of the colonic content (mass per sample recorded) was added in a 2 ml centrifuge tube with 1.5 ml ultrapure water. It was thoroughly mixed, set aside for 30 min, and centrifuged at 4°C, 12,000×g, for 15 min. Approximately 1 ml supernatant was mixed with 0.2 ml metaphosphoric acid solution [25% (w/v)] and 23.3 μl crotonic acid solution (210 mmol/L) before being centrifuged at 4°C, 12,000×g for 10 min after incubation at 4°C for 30 min. Following centrifugation, 0.3 ml of supernatant was added to 0.9 ml of chromatographic methanol (1:3 dilution) and centrifuged at 4°C, 12,000×g, for 5 min. The supernatant was filtered through 0.22 μm syringe filter. Shimadzu 2010 Plus AF gas chromatograph was used to determine SCFAs.

The total RNA was isolated from colon tissue using TRIzol Reagent (Invitrogen, Carlsbad, CA, USA). The first-strand cDNA was synthesized from the total RNA using RevertAid First Strand cDNA Synthesis Kit (ThermoFisher, Waltham, MA, USA). The q-PCR was performed using iTaq™ Universal SYBR Green on the CFX Connect™ Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA). Primers and annealing temperature are shown in Supplementary Table 1 . The result of relative gene expression was normalized to that of β-actin and evaluated through the 2^−△△Ct method.

The total microbial DNA was obtained from the colon content or fecal pellets by E.Z.N.A.^® Stool DNA Kit (OMEGA, Norcross, GA, USA) and quantified by NanoDrop™ 2000 Spectrophotometer (ThermoFisher, Waltham, MA, USA). PCR amplification of V3–V4 regions of bacterial 16S rRNA genes was performed using universal primer sequences 341F (5’-barcode-CCTAYGGGRBGCASCAG) and 806R (5’-barcode-GGACTACNNGGGTATCTAAT). The PCR products were purified by QIAquick Gel Extraction Kit (Qiagen, Germantown, MD, USA). Thereafter, the amplicon libraries were prepared using Ion Plus Fragment Library Kit (ThermoFisher, Waltham, MA, USA) and the sequencing was performed on Ion S5™ XL system (ThermoFisher, Waltham, MA, USA).

Data were analyzed using Novogene cloud analysis platform (https://magic.novogene.com). Raw FASTQ files were edited to exclude adaptor contamination (Cutadapt, V1.9.1, http://cutadapt.readthedocs.io/en/stable/). Chimeras were detected using UCHIME Algorithm (http://www.drive5.com/usearch/manual/uchime_algo.html). The clean reads were clustered into operational taxonomic units (OTUs) by Uparse software (v7.0.1001, http://www.drive5.com/uparse/) at 97% identity. Alpha diversity and beta diversity were evaluated using QIIME (Version 1.9.1). Differentially abundant taxa were identified using linear discriminant analysis (LDA) effect size (LEfSe) (https://huttenhower.sph.harvard.edu/lefse/). Functional annotation of prokaryotic taxa (FAPROTAX) was performed to identify ecologically relevant functions based on the current literature on cultured strains (https://pages.uoregon.edu/slouca/LoucaLab/archive/FAPROTAX/lib/php/index.php). DESeq2 was used to detect differentially abundant microbes at the genus level (https://bioconductor.org/packages/release/bioc/html/DESeq2.html).

Panda-microbiota associated (PMA) mice were generated by pre-treatment with an antibiotic cocktail for 7 days and then gavage administration of panda microbiota for 3 days. We continuously monitored total bacteria, Bifidobacteria, Enterococcus, Escherichia coli, and Lactobacillus by viable bacterial count from days 0 to 13. Fecal microbiota were mostly eliminated by day 7 of the antibiotic cocktail ( Supplementary Table 2 ). By day 13, there were no differences observed in the absolute counts of detected bacteria aside from the higher total anaerobe count in the Blank and BlankT group ( Supplementary Table 2 ). To assess for similarity of microbiome diversity and composition of the PMA mice with the donor panda microbiota, 16S rRNA sequencing was performed of the panda gut microbiome (Panda group), non-transplanted mice (Blank group), and PMA-mice (BlankT group). Chao1, Shannon, and Simpson indices were measured to assess the alpha diversity of the gut microbiome ( Figure 1 ). On day 13 ( Figures 1A – C ), the Shannon index was the only alpha diversity metric with a significant difference. As shown in the figure, the Shannon index in the Blank and BlankT groups was higher than that of the Panda group, while no difference was observed between the Blank and BlankT groups. More differences in alpha diversity were observed on day 16 ( Figures 1D – F ): the Chao1, Shannon, and Simpson indices were higher in the BlankT group compared to the Panda group, but only the Shannon index showed a significant difference. There continued to be no differences observed between the Blank and BlankT groups. The BlankT group had lower abundance of Enterococcus, Lactobacillus and Parasutterella than the Blank group and higher abundance of Enterobacteriacaeae (including Klebsiella, Morganella and unclassified members of this family) at day 13 by DESeq2 analysis ( Supplementary Figures 2A, B and Supplementary Tables 3A, B ) on day 13. Data from day 16 showed a similar result as that on day 13, including increased Enterobacteriacaeae in the BlankT group (Klebsiella, Providencia, Proteus, and unclassified members) ( Supplementary Figures 2C, D and Supplementary Tables 3C, D ). The PCoA and NMDS results ( Figures 1G – N ) indicated that the gut microbiome of the BlankT group separated from the Panda group and the Blank group, which was statistically significant based on Anosim and Adonis analysis ( Supplementary Table 4 ) on both days 13 and 16. These results suggest that the gut microbiome of the PMA-mouse is different from the non-transplanted mice and that this difference may persist.

The DSS-induced colitis mouse model has similar clinical features to pandas with a high abundance of Enterobacteriaceae and inflammation caused by an epithelial barrier disruption with exposure of mucosal immune cells to microbial antigens. DSS (5%) was supplied in the drinking water for 3 days following panda microbiota administration to generate a DPMA mouse to model intestinal inflammation in pandas. After DSS oral administration (on day 13), the BlankT and BlankD groups had higher alpha diversity than the Panda group ( Figures 1A – C ). Beta diversity results ( Figures 1G–N ) showed that the different groups separated from each other, which was significant by Anosim and Adonis analyses ( Supplementary Table 5 ). Furthermore, taxonomic summary plots at the family level showed that the proportion of Enterobacteriaceae and Enterococcaceae was higher in the BlankD and Panda groups than the BlankT group ( Supplementary Figures 3A, B ). DESeq2 analysis at the genus level similarly demonstrated enrichment of Enterococcus and unclassified Enterobacteriaceae ( Supplementary Table 6 ). On day 16, the panda group had lower alpha diversity by the Shannon index compared to the BlankT group but there were otherwise no significant differences across the three groups ( Figures 1D – F ). The PCoA and NMDS ( Figures 1K – N ) results indicated that there were significant differences among groups based on Anosim and Adonis analyses ( Supplementary Table 6 ) on days 13 and 16. At the family level, Enterobacteriaceae and Enterococcaceae were found in much greater abundance in the BlankD and Panda groups compared to the BlankT group ( Supplementary Figures 3C, D ). Viable bacterial counts of E. coli (a member of the Enterobacteriacaeae family) were significantly greater in BlankD compared to BlankT; both groups had higher E. coli and Enterococcus as well as lower Bifidobacterium than panda microbiome ( Supplementary Table 7 ).

To identify genera affected by intestinal inflammation in the DPMA-mouse, we further compared the intestinal microbiota between PMA-mouse (BlankT group) and DPMA-mouse (BlankD group). On day 13, LefSe analysis ( Supplementary Figure 4A ) indicated that, at the genus level, the abundance of Enterorhabdus, Bifidobacterium, Roseburia, Dubosiella, Faecalibaculum, and Akkermansia were decreased, and the abundance of Bacteroides, Enterococcus, Clostridioides, Romboutsia, Clostridium_methylpentosum, Turicibacter, Veillonella, Klebsiella, Proteus, Providencia, Pseudomonas, and Parasutterella were increased after DSS administration. On day 16, LefSe analysis ( Supplementary Figure 4B ) revealed that the abundance of Enterorhabdus, Alistipes, Roseburia, Butyricicoccus, Caproiciproducens, Dubosiella, Faecalibaculum, Anaeroplasma, and Akkermansia were decreased, and the abundance of Enterococcus, Clostridioides, Paeniclostridium, Romboutsia, Flavonifractor, Veillonella, Klebsiella, Lelliottia, Morganella, Proteus, and Providencia were increased after DSS administration. As differential genera varied between days 13 and 16, we decided to utilize genera with the same change trend on both days 13 and 16 as marker genera for intestinal inflammation. This included Enterorhabdus spp., Roseburia spp., Dubosiella spp., Faecalibaculum spp., Akkermansia spp., Enterococcus spp., Clostridioides spp., Romboutsia spp., Veillonella spp., Klebsiella, spp. Proteus spp. and Providencia spp.

Based on prior reports that Lactobacillus strains can enhance intestinal barrier function, we investigated whether the panda-derived L. plantarum G83 strain would have this effect on the DPMA-mouse model. Three different doses of L. plantarum G83 were provided to DPMA mice after administration of 5% DSS drinking water for 3 days (TreatLG, TreatMG, and TreatHG as described in the Material and Methods). As a treatment positive control, we also included a group receiving mesalazine, an anti-inflammatory medication that ameliorates DSS-induced colitis (34). First, we performed a histopathologic analysis of the colon tissue ( Figure 2 ). Results showed that the epithelial cell integrity of the colon tissue was destroyed by DSS administration, and a large amount of inflammatory cell infiltration was observed in the BlankD group. There was, however, significantly reduced ulcer damage, intestinal epithelial disruption, and inflammatory cell infiltration by histopathology scoring in the mesalamine (TreatMz) and medium dose L. plantarum G83 treatment group (TreatMG) compared with the BlankD group. We next investigated the effect of L. plantarum G83 on tight junction proteins in the DPMA mouse. The expression of ZO-1 and Occludin protein but not Claudin-1 were significantly increased in the TreatMG and TreatHG groups by immunohistochemistry ( Figure 3 ). To further assess intestinal permeability, the concentration of D-lactic acid in the serum was measured ( Figure 4 ). Compared to the BlankD group, only the TreatMG group had a significantly lower concentration of D-lactic acid. Altogether, data from these experiments support that L. plantarum G83 can modulate epithelial barrier integrity in the DPMA mouse.

DSS induced colitis symptoms in the DPMA mice, namely, weight loss, diarrhea, and rectal bleeding were monitored by a disease activity index (DAI). Reduced DAI scores were observed at day 16 in the DPMA mice receiving L. plantarum G83, which was significant for the TreatMG and TreatHG groups ( Figures 5A, B ). L. plantarum G83 treated mice also exhibited less weight loss than the mice in the BlankD group which again was significant only for the TreatMG and TreatHG groups ( Figures 5C, D ). Colon myeloperoxidase (MPO) activity—a marker of colonic inflammation—was significantly reduced in the TreatMz and TreatMG groups compared to BlankD ( Figure 6A ). Based on these results, we aimed to determine whether L. plantarum G83 could reduce serum and colonic levels of inflammatory cytokines in the DPMA mice. Proinflammatory cytokines in the serum, including TNF-α and IFN-γ, were significantly inhibited by L. plantarum G83 administration but less than the mesalazine treated group ( Figures 7A, B ). IL-10, which has anti-inflammatory effects, was promoted by L. plantarum G83 treatment in the medium high-dose group ( Figure 7C ). We did not observe any difference in IL-10 between the other three groups and BlankD. Colonic TNF-α was inhibited by L. plantarum G83 administration at all three doses while unexpectedly, mesalazine failed to reduce TNF-α secretion in the colon tissue ( Figure 7D ). IFN-γ levels in colon tissue did not significantly differ among the groups ( Figure 7E ). All treatment groups failed to promote colonic IL-10 except for the medium dose of the L. plantarum G83, which showed a higher concentration of IL-10 ( Figure 7G ). Conversely, the TreatMG group exhibited a lower colonic IL-4 concentration compared to the BlankD group, and all other groups showed no difference ( Figure 7F ). Secretory immunoglobulin A (sIgA) in luminal content increased only in the TreatLG group, and there were no significant differences among the other groups ( Figure 6B ). Overall, these results demonstrated that L. plantarum G83 exhibits an anti-inflammatory effect in the DPMA mice.

We hypothesized that the beneficial effects of L. plantarum G83 on intestinal inflammation and barrier integrity in DPMA mice were due to reduced dysbiosis. To assess this, we compared the colonic luminal microbiota of the DPMA mouse after receiving the different treatments. As shown in Figure 8 , L. plantarum G83 treated groups had a lower abundance of Peptostreptococcaceae, Enterobacteriaceae, and Enterococcaceae, and a higher abundance of Muribaculaceae and Fusobacteriaceae compared to the BlankD group. Higher relative abundance of Lactobacillaceae was observed in all L. plantarum G83 treatment groups compared to the TreatMz group ( Figure 8A ). At the genus level, the relative abundance of Fusobacterium increased in all three G83 treatment groups, and the relative abundance of Lactobacillus increased in all L. plantarum G83 treatment groups compared to the TreatMz group ( Figure 8B ). DESeq2 analysis showed that Fusobacterium enrichment and Enterococcus depletion were statistically significant in all three G83 dose groups ( Supplementary Table 8 ). When we analyzed the gut microbiota of the TreatMz group, we found that the microbiota of the TreatMz group showed a drastic change, different from any of all other groups. The relative abundance of Lactobacillus was greatly reduced compared to other groups. The TreatMz, TreatLG, and TreatHG groups had increased microbial richness compared with the BlankD group by the Chao1 index ( Figure 8C ). The treatment groups did not show statistically significant differences in Shannon index or Simpson index compared to BlankD ( Figures 8D, E ). Beta diversity analysis ( Figures 8F, G ) showed that there was a significant difference between each L. plantarum G83 treated group and the BlankD or TreatMz group; no significant difference was observed between the BlankD and TreatMz groups based on Anosim and Adonis analyses ( Supplementary Table 9 ). To understand the effect of L. plantarum G83 on DSS-induced microbial shifts, we performed LefSe analysis to analyze changes in the DSS biomarker genera in all five groups. Enterococcus spp., Klebsiella spp., Proteus spp. and Providencia spp. were more abundant in both the BlankD and TreatMz groups ( Supplementary Figure 5 ). These genera belong to the family Enterobacteriaceae, typically considered as a marker of high risk for enteritis in pandas. Veillonella spp. only increased in the BlankD group. Besides these changes in the previously defined DSS biomarker genera, LefSe also demonstrated enrichment of Bifidobacterium spp. in the Treat MG and TreatHG groups ( Supplementary Figure 5F ). Q-PCR quantitation and viable bacterial count were performed to determine changes in absolute abundance of specific microbial taxa. All three L. plantarum G83 treatment groups had reduced Enterobacteraceae by q-PCR compared to the BlankD and TreatMz groups, and TreatLG and TreatMG also had reduced Proteobacteria ( Supplementary Figure 6 ). Lower abundance of E. coli was detected in all three L. plantarum G83 treatment groups compared to the BlankD group ( Supplementary Figure 7 ). The treatMG and TreatHG groups also had greater Bifidobacterium spp., consistent with the 16S sequencing analysis result. FAPROTAX was used to predict the function of the bacteria in the intestinal microbiota. The heatmap result showed that the BlankD group had a high abundance of genes associated with human_pathogen_diarrhea, human_pathogen_all, human_pathogen_pneumonia, human_pathogen_septicemia, ureolysis, and nitrate metabolism, while the mesalazine and L. plantarum G83 treatment groups had a very low abundance of genes in these pathways ( Supplementary Figure 8 ). Overall, L. plantarum G83 improved microbial dysbiosis, with decreased levels of Enterobacteriaceae and disease-related genes.

To further investigate the functional consequences of the altered microbiome induced by L. plantarum G83 treatment, we measured the concentration of SCFAs, including acetic acid, propionic acid, butyric acid, and valeric acid (valeric acid and isovaleric acid) in the colon lumen using high-performance gas chromatography (HPGC). The medium dose of L. plantarum G83 increased luminal concentration of all detected SCFAs including acetic acid, propionic acid, butyric acid, and valeric acid compared to the BlankD, TreatMz, and TreatLG groups ( Figure 9 ). The high dose of G83 had similar concentrations of acetic acid and butyric acid as the medium dose, but lower concentrations of propionic acid and valeric acid. Furthermore, the TreatMz and the TreatLG groups had numerically increased SCFA concentrations compared to the BlankD group, but only propionic acid in the TreatLG group showed significance. Together, the medium and high dose of L. plantarum G83 administration showed a strong capability to increase colonic luminal SCFAs.

We hypothesized that alterations in the microbiome and microbial metabolites including SCFAs would affect intestinal immune cell signaling pathways including the NF-κB/p65 pathway that regulate the inflammatory response. To investigate this, we compared colonic tissue expression of Toll-like receptors (TLRs) and the NF-κB/p65 pathway related genes among the groups by q-PCR. L. plantarum G83 groups had increased TLR2 gene expression, whereas mesalazine did not alter TLR2 gene expression ( Figures 10A – E ). The expression level of the TLR4 and TLR5 genes was significantly decreased by mesalazine treatment. We did not observe any effects of L. plantarum G83 on TLR4 and TLR5 gene expression, except that the medium dose of L. plantarum G83 showed increased TLR4 gene expression. Interestingly, TLR9, the receptor for bacterial CpG-DNA, was inhibited by both mesalazine and L. plantarum G83. There was no difference among groups in TLR1 expression.

To confirm the effect of L. plantarum G83 on the inflammatory response, we measured expression levels of genes related to the NF-κB pathway, which is critically important in the immune response. Mesalazine significantly reduced the gene expression level of MyD88, NF-κB, iKKβ, TNF-α, IL-1β, and IL-6 ( Figures 10F–K ). The L. plantarum G83 treated group also showed reduced expression of NF-κB, iKKβ, TNF-α, IL-1β, and IL-6. Aside from the medium dose, L. plantarum G83 treatment did not affect MyD88 gene expression.

We noted that there was an increased TLR expression despite a reduction in the MyD88 expression. To investigate this, we further tested several TLR negative regulators, including interleukin 1 receptor-like 1 receptor (IL1RL1) and IL-1R-associated kinase M (IRAK-M). The IL1RL1 gene was promoted by L. plantarum G83 with the highest gene expression level observed in the medium dose treatment group. All treatment groups downregulated IRAK-M gene expression ( Figure 11 ).

We further investigated NF-κB activity by assessing phosphorylation status by Western blot for p-NF-κB-p65/NF-κB-p65 and p-MAPK-p38/MAPK-p38. In the present study, L. plantarum G83 treatment led to reduced NF-κB-p65 phosphorylation which was significant for the TreatMG and TreatHG groups. However, p-MAPK-p38 did not significantly differ in any of the groups compared to BlankD ( Figure 12 ). In summary, these data indicated that the L. plantarum G83 inhibited the NF-κB/p65 pathway in the colon of DPMA mice.

Flow cytometry was then performed to determine the polarization of different T cell subsets in the peripheral blood, Peyer’s Patches, mesenteric lymph nodes (MLN), and spleen. In the peripheral blood, the ratio of CD3⁺CD4⁺/CD3⁺CD8⁺ T cells was reduced by high dose L. plantarum G83 compared to the other groups ( Figure 13A ). In contrast, the CD3⁺CD4⁺/CD3⁺CD8⁺ ratio in spleen was increased in the mesalazine, low and medium dose L. plantarum G83 groups compared to untreated DPMA mice ( Figure 13D ). The CD3⁺CD4⁺/CD3⁺CD8⁺ ratio was also elevated in Peyer’s patches in the medium and high dose L. plantarum G83 groups compared to the other groups ( Figure 13B ). The medium dose L. plantarum G83 group also had increased CD3⁺CD4⁺ T/CD3⁺CD8⁺ ratio in the mesenteric lymph nodes compared to BlankD ( Figure 13C ). We then assessed T helper cell (CD4⁺) subsets in the spleen including Th1, Th2, and Th17 cells. We found that the ratio of Th1/Th2 cells was increased in the low dose L. plantarum G83 group compared to BlankD ( Figure 13E ). The low dose L. plantarum G83 group and the mesalazine group also had higher Th1/Th17 ratio compared to BlankD ( Figure 13F ). Taken together, these results suggest that L. plantarum G83 treatment could affect T cell polarization in DPMA mice.