Bifidobacterium longum (Item No.: JBLC-141) was purchased from Shandong Zhongke Jiayi Bioengineering Co. Ltd. (Shandong, China). Pentobarbital sodium solution, sodium chloride (NaCl) solution (saline), sterilized water, suture needles, syringes, and gauze were purchased from the First Hospital of Lanzhou University, China. Tissue scissors, hemostatic forceps, forceps, arterial clips, and other surgical instruments were provided by the Laboratory of the First Hospital of Lanzhou University, China.

All animal protocols were approved by the Ethics Committee of the First Hospital of Lanzhou University (approval number: LDYYLL2024-537).

Thirty-two 6-week-old male Sprague–Dawley rats were purchased from Lanzhou Veterinary Research Institute. The rats were housed in a clean animal rearing room at 20–25°C, 20–40% humidity, under a 12-h light/dark cycle with free access to food and water. After a one-week acclimatization period, the rats were randomly divided into four groups (n = 8/group): normoxia plus saline (SS), normoxia plus probiotic (SB), hypoxia plus saline (LS), and hypoxia plus probiotic (LB). Each group was prophylactically administered 2 mL of 0.9% saline or an equal volume of a solution containing 1 × 10¹¹ CFU/mL B. longum via intragastric administration for five days. Twenty-four hours after the last administration, the LS and LB groups were placed in the plateau low-pressure oxygen simulation chamber. This chamber increased to an altitude of 6,000 m (oxygen [O₂] content: 13%, barometric pressure: 354 mmHg) at an average speed of ~5 m/s to simulate continuous highland hypoxia for six days. During continuous exposure to hypoxia, the cabin was opened once daily for 30 min to provide food and water, operate gavage, measure the body weight, and collect feces on the day of end of the experiment. The SS and SB groups were maintained in a normoxia environment outside the chamber of the Basic Medical Laboratory of the 940th Hospital of the United Logistics Force (Lanzhou, 1,500 m above sea level, air pressure: 635 mmHg).

On Day 6 of exposure in the plateau hypoxia simulation chamber, the rats were anesthetized with sodium pentobarbital solution (1%, 50 mg/kg); blood was collected from the abdominal aorta, from which serum was prepared and stored at −80°C. Fresh colon tissues and the intestinal contents were collected from each group of rats after they were euthanized, and half of the colon tissue samples were clipped, frozen rapidly in liquid nitrogen, and stored at −80°C. The other half of the samples were fixed by immersion in a 4% paraformaldehyde solution and stored for further examination. To minimize pain and distress as much as possible in the rats, all samples were collected under anesthesia induced using a sodium pentobarbital solution, which ended in their death.

Serum levels of interleukin (IL)-6, tumor necrosis factor α (TNF-α), and IL-10 were detected using enzyme-linked immunosorbent assay (ELISA) kits (Boster, Wuhan, China). Serum levels of diamine oxidase (DAO) and d-lactic acid (DLA) were detected using biochemical analysis kits (Elabscience, Wuhan, China), following the manufacturer’s instructions.

The malondialdehyde (MDA) and superoxide dismutase (SOD) contents of colonic tissues were determined using appropriate commercial kits according to the manufacturer’s instructions (Elabscience, Wuhan, China). OCLN, claudin-1 (CLDN1), and zonula occludens-1 (ZO-1) levels in the colon tissues were determined using analysis kits according to the manufacturer’s instructions (Meilian, Shanghai, China).

Rat colonic samples were fixed in a 4% paraformaldehyde solution, embedded in paraffin, cut into 4-μm slices, and stained with hematoxylin and eosin (H&E). Images were acquired for analysis under a positive light microscope (Nikon Eclipse E100; Nikon, Tokyo, Japan) at 100× and 200× magnification.

The terminal deoxynucleotidyl transferase deoxyuridine dUTP nick-end labeling (TUNEL) method was used to detect apoptosis of intestinal tissue. Fixed intestinal tissues were routinely paraffin-embedded, sectioned, and dewaxed; a proteinase K working solution (proteinase K storage solution diluted with phosphate-buffered saline [PBS] 1:9) was applied drop-wise and incubated for 30 min at 37°C. Subsequently, the tissues were washed with PBS (pH 7.4) thrice for 5 min each, shaken dry, and membrane-breaking working solution was added drop-wise. They were incubated for 20 min and washed with PBS thrice for 5 min each. Proteinase K working solution was added for 20 min at room temperature. The sections were subsequently incubated for 2 h at 37°C, washed with PBS thrice for 5 min each, and incubated in 3% hydrogen peroxide solution prepared with methanol (hydrogen peroxide to methanol = 1:9) for 15 min in the dark. The sections were washed with PBS thrice, dried, and an appropriate amount of reagent 3 (converter-PP) was added (converter-POD), followed by incubation at 37°C for 30 min and washing thrice with PBS for 5 min each. The sections were shaken dry, freshly prepared 3,3′-diaminobenzidine (DAB) color development solution was applied under the microscope to control the color development time, and the sections were rinsed with tap water to terminate the color development. This was followed by re-staining with Harris hematoxylin for approximately 3 min, washing with tap water, and applying 1% hydrochloric acid alcohol for a few seconds. The sections were rinsed with tap water, treated with ammonia, rinsed in tap water, and washed in ammonia water again to restore the blue color. Finally, the sections were dehydrated, dried, sealed, and scanned using a digital section scanner to collect images and data.

Caspase 3 Activity (Elabscience, Wuhan, China) and B-cell lymphoma 2 (BCL-2) (Meilian, Shanghai, China) levels in the colonic tissues were determined using analysis kits according to the manufacturer’s instructions.

Total protein was extracted from rat colonic tissue samples using abrasive beads and radioimmunoprecipitation assay (RIPA) lysis buffer. Protein concentration was measured using a bicinchoninic acid (BCA) assay kit (both from Servicebio, Wuhan, China). Proteins were electrophoresed using 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred onto a polyvinylidene fluoride membrane. After incubation with skim milk to block non-specific binding, the membrane was incubated with primary antibodies at 4°C overnight. The membrane was washed with Tris-buffered saline (TBST), incubated with a secondary antibody for 30 min at room temperature, and washed once with TBST. Protein bands were visualized using the enhanced chemiluminescence (ECL) luminescent liquid combo kit (Servicebio), and data were analyzed using the AIWBwell analysis software (Servicebio). The antibody sources, catalog numbers, and dilutions are listed in Supplementary Table S1.

Microbial DNA was extracted from feces using the TGuide S96 magnetic soil/stool DNA kit (Tiangen Biotech Co., Ltd., Beijing) according to the manufacturer’s instructions. The 338F: 5’ACTCCTACGGGAGGCAGCA-3′ and 806R: 5’GGACTACHVGGGTWTCTAAT-3′ universal primer set was used to amplify the V3–V4 region of the 16S rRNA gene from the genomic DNA extracted from each sample, with the Illumina Novaseq 6,000 sequencing system (Illumina, Santiago CA, USA). Bioinformatics analysis was performed using the BMK Cloud (Biomarker Technologies Co., Ltd., Beijing, China). Based on the quality of a single nucleotide, raw data were filtered using Trimmomatic (version 0.33).

Primer sequences were identified and removed using the Cutadapt software (version 1.9.1). Paired-end reads obtained from the previous steps were assembled using USEARCH (version 10), followed by chimera removal using UCHIME (version 8.1). Sequences with similarity ≥97% were clustered into the same operational taxonomic unit (OTU); OTUs with re-abundance <0.005% were filtered. Taxonomic annotation of the OTUs was performed based on the naïve Bayes classifier in QIIME2 using the SILVA database (release 132) with a confidence threshold of 70%. Alpha diversity was calculated and displayed using QIIME2 and R software programs, respectively.

Beta diversity was determined using QIIME to evaluate the degree of similarity between the microbial communities from different samples. Principal coordinate analysis (PCoA), heat maps, unweighted pair group method with arithmetic mean (UPGMA), and non-metric multidimensional scaling were used to analyze the beta diversity. Furthermore, linear discriminant analysis (LDA) effect size (LEfSe) was used to identify significant taxonomic differences among groups. A logarithmic LDA score of 3.5 was set as the threshold for discriminative features. To explore the dissimilarities in the microbiome among different factors, redundancy analysis was performed with the R software using the “vegan” package.

The experimental data were statistically analyzed and graphed using GraphPad Prism version 10 (GraphPad Software, Inc., San Diego, CA, USA), and data were expressed as mean ± standard error of the mean. Comparisons between multiple groups of quantitative data were performed using a one-way analysis of variance (ANOVA). Results with a p < 0.05 were considered statistically significant.

H&E staining of the rat colonic tissues (Figure 1A) showed that the SS group had normal intestinal tissue structure. The epithelial cells of the mucosal layer were tightly arranged, with no substantial erosion or detachment; numerous crypts and goblet cells were observed. Furthermore, the tissue was not substantially infiltrated by inflammatory cells. In contrast, the intestinal tissue of the LS group exhibited an abnormal structure with eroded and some detached epithelial cells in the mucosal layer. In addition, the crypts were disordered, with reduced goblet cells and increased inflammatory cells in the mucosal layer. The LB group exhibited a normal intestinal tissue structure with tightly arranged epithelial cells in the mucosal layer and structurally normal crypts. Moreover, the goblet cells were abundant, and no obvious inflammatory cell infiltration. The SB group also showed a normal intestinal tissue structure, with intact epithelial cells in the mucosal layer and numerous goblet cells, comparable to that of the SS group.

Serum IL-6 levels were significantly higher (p < 0.0001; Figure 1K) in the LS group than in the SS group, whereas that of IL-10 was significantly lower (p < 0.001; Figure 1L). Pretreatment of LB group with JBLC-141 significantly decreased IL-6 levels (p < 0.0001) and slightly increased IL-10 (p > 0.05) compared with the LS group.

ELISA results revealed that colonic tissue levels of OCLN, CLDN1, and ZO-1 were significantly lower in the LS group than in the SS group (all p < 0.0001; Figures 1B–D). JBLC141 pretreatment significantly increased the expression of these proteins in the LB group (OCLN: p < 0.01, CLDN1: p < 0.01, and ZO-1: p < 0.05, respectively) compared with the LS group. These results were corroborated by western blotting (Figures 1E–H). Furthermore, the protein expression level analysis indicated good intestinal barrier repair.

The detection of rat serum DAO and DLA levels (Figures 1I,J) showed significantly higher levels in the LB group than in the SS group (both p < 0.0001). Their levels were significantly lower in the JBLC-141-pretreated LB group than in the LS group (p < 0.05 and p < 0.001, respectively).

TUNEL analysis of intestinal tissue apoptosis (Figures 2A,B) showed no obvious apoptotic cells in the SS group. In contrast, numerous brownish-yellow apoptotic signals were observed in the mucosal layer of the LS group. Meanwhile, significantly fewer apoptotic signals were detected in the LB group than in the LS group. The colonic structure of the SB group was comparable to that of the SS group, with no apoptotic signals observed.

Detection of Caspase 3 Activity and B-cell lymphoma 2 (BCL-2) levels in the colonic tissues (Figures 2F,G) showed that the Caspase 3 Activity level in the LS group was significantly higher than in the SS group (p < 0.0001). Their abundance was significantly lower in the LB group than in the LS group (p < 0.001). The BCL-2 level was significantly lower in the LS and LB groups (both p < 0.0001) than in the SS group and significantly higher in the LB group than in the LS group (p < 0.0001). These results were corroborated by western blot analysis (Figures 2C–E).

To determine the antioxidant effects of JBLC-141, MDA and SOD levels were detected in hypoxic rat colonic tissues. A significantly higher MDA level (p < 0.0001; Figure 3A) and lower SOD activity (p < 0.0001; Figure 3B) was observed in the LS group than in the SS group. JBLC-141 pretreatment reversed in the LB group (MDA, p < 0.0001; SOD, p < 0.001).

Western blotting revealed significantly higher KEAP-1 levels in the colonic tissues of the LS group compared with the SS group (p < 0.01) and significantly lower levels of NRF2 (p < 0.0001; Figures 3C–F) and heme oxygenase 1 (HO-1; p < 0.0001). These changes in KEAP-1 (p < 0.05), NRF2 (p < 0.05), and HO-1 (p < 0.001) were significantly more suppressed in the LB group than in the LS group after JBLC-141 pretreatment. These results were corroborated by biochemical assays (Figures 3G–I).

We further investigated the correlation between the intestinal flora composition and barrier damage in each experimental group. The Venn diagrams showed that 455 genera were identified in all samples, with 318, 260, 265, and 322 specific genera identified in the SS, LS, LB, and SB groups, respectively. Among the four groups, 185 genera were identified, with 53, 19, 28, and 59 unique to the SS, LS, LB, and SB groups, respectively (Figure 4A). Hence, B. longum JBLC-141 had a regulatory effect on the intestinal flora of rats in the low-oxygen environment of the plateau.

Microbial diversity analysis showed that the Shannon and Chao indices of the LS group were lower than those of the SS group, indicating that plateau hypoxia decreased the diversity and abundance of the intestinal flora. However, no significant difference was observed between the LB and LS groups (Figure 4B). PCoA and UPGMA analyses of bacterial communities showed a significant separation between the LS group and the SS and SB groups. Therefore, plateau hypoxia decreased changes in the structure of intestinal bacterial communities. Meanwhile, JBLC-141 pretreatment relatively restored the structure to that of the SS and SB groups (Figures 4C,D).

Analysis of the microbial phyla composition identified Firmicutes, Bacteroidetes, Desulfobacterota, Proteobacteria, and Actinobacteria as the major phyla in each group, accounting for >99% of the total microbial population. The relative Actinobacteria population in the LS group was reduced but recovered after JBLC-141 pretreatment (Figure 4E). Analysis of the top 20 phylogenetic compositions showed that the LS group contained unclassified_Muribaculaceae, unclassified_Clostridia_UCG-014, unclassified_Oscillospiraceae, unclassified_Eubacterium_coprostanoligenes_group, Dubosiella, Allobaculum, Ligilactobacillus, and NK4A214_group. Furthermore, this group exhibited a reduced and increased abundance of Lactobacillus, Prevotella, Bacteroides, Prevotellaceae_NK3B31_group, Phascolarctobacterium, Christensenellaceae_R_7 group, and Blautia. Notably, JBLC-141 pretreatment of the LB group reversed these trends (Figure 4F). Biomarkers that differed significantly between the groups were identified. Further analysis of the rat intestinal flora in each group using LEfSe showed that g_Lactobacillus, f_Tannerellaceae, g_Parabacteroides, p_Proteobacteria, c_Gammaproteobacteria, s_unclassified_Parabacteroides, s_unclassified_Prevotellaceae, g_unclassified_Prevotellaceae, s_unclassified_Blautia, f_Sutterellaceae, o_Burkholderiales, and s_Parabacteroides_spp were significantly enriched in the LB group. In contrast, f_Prevotellaceae, g_Prevotella, s_Lactobacillus_intestinalis, s_unclassified_Prevotella, s_unclassified_Roseburia, and g_Roseburia were significantly enriched in the LS group (Figures 4G,H).

Correlation analysis was conducted using hierarchical clustering based on Spearman’s correlation of differential flora using oxidative stress-related and intestinal barrier function indicators. At the genus level, the serum pro-inflammatory factor, IL-6, and the indicator of intestinal barrier permeability, DLA, were significantly negatively correlated with Allobaculum, Dubosiella, unclassified_RF39, and NK4A214_group. In addition, IL-6 was significantly and positively correlated with Prevotella_9, whereas DLA was significantly and positively correlated with unclassified_Clostridia_UCG-014. DLA was also negatively correlated with unclassified_Clostridia_UCG_014, unclassified_[Eubacterium]_coprostanoligenes_group, and significantly positively correlated with Parabacteroides, Roseburia, and Bacteroides.

SOD, which removes reactive oxygen radicals from colon tissue, and colon tissue tight junction proteins ZO-1, OCLN, and CLDN1 were significantly and positively correlated with Allobaculum, Dubosiella, and unclassified_RF39. Intestinal tissue ZO-1, OCLN, and CLDN1 were negatively correlated with Bacillus, Lactobacillus, Parabacteroides, and Roseburia. The colonic tissue oxidative stress product, MDA, content was negatively correlated with unclassified_Oscillospiraceae, unclassified_Muribaculaceae, and unclassified_Desulfovibrionaceae and positively correlated with Phascolarctobacterium.