The potential probiotic strain used in this study was isolated and confirmed previously in our lab. Briefly, strains of lactic acid bacteria (LAB) were isolated from fermented yoghurt, confirmed by whole genome sequencing (WGS) and compared with the GenBank. A total of 292 LAB strains were isolated, the L. rhamnosus P118 strain (PRJNA848987) for this study was selected based on preliminary results obtained in vitro and in silico. To obtain the inoculum of the LAB, L. rhamnosus strain P118 was propagated twice in the DeMan, Rogosa, and Sharpe (MRS, Oxoid Ltd, Hants, UK) broth at 37°C for 16h under shaking. The concentration of bacterial suspension was measured by spectrophotometer and presented as equivalent to CFU/mL. Moreover, a spontaneous novobiocin-resistant S. Typhimurium SL1344 kept in our lab was grown overnight in the Luria-Bertani (LB) broth at 37 °C in an orbital shaking incubator at 180 rpm/min, sub-cultured twice, and then the CFU/mL was determined by spectrophotometer.

A total of 160 chickens, 80 were used for clinical trials and 80 were used to observe survival and growth. At the same time, the same grouping method and feeding conditions were used. One hundred and sixty 1-day-old healthy female Babcock chicks (License No.: 2021-0005) were purchased from a standardized hatchery (ChunPai hatchery, CiXi, Zhejiang, China). Four groups of chicks were randomly divided (n=40 in each group), and each chick was given its own unique number. In each group, 20 chicks are used to count the weight, growth status and survival rate, and the other 20 chicks are used to monitor clinical symptoms, and obtain tissue samples and other biomaterials used in subsequent experiments. The treatment groups were as follows: (I) the negative control (no L. rhamnosus P118 pretreatment and no Salmonella infection, designed as C); (II) the L. rhamnosus P118-pretreated group (10⁸ CFU L. rhamnosus P118, designed as P); (III) the Salmonella-infected group (10⁸ CFU S. Typhimurium SL1344, designed as S); and (IV) the L. rhamnosus P118-pretreated and Salmonella-infected group (10⁸ CFU L. rhamnosus P118 and 10⁸ CFU S. Typhimurium SL1344, designed as P+S) ( Figure 1A ). For the duration of the experiment, all chicks received free access to water and a starter feed without antibiotics. Throughout this study, all experimental protocols were approved by the Ethics Review Committee of Experimental Animal Welfare of Zhejiang University (Ethical Approval ZJU20190093) and the safety procedures were followed.

At the five- and seven-day post-infection periods, six chicks from each group were selected randomly. Under respiratory anesthesia (Matrx VMR, Midmark, OH, USA), chicks were subjected to cardiac blood collection using vacuum blood collection vessels (BD Vacutainer ®, NJ, USA). Blood samples were incubated for one hour at room temperature before centrifugation at 2,000 g for 10  min to separate serum. Chickens were euthanized by cervical dislocation under anesthesia. One arm of cecum contents, duodenum, liver, spleen and fecal were collected for Salmonella enumeration. To prepare for the follow-up observation of intestinal ultrastructure, a segment (~1 cm) of cecum and duodenum were fixed in 2.5% glutaraldehyde solution (electron microscope fixing solution). Duodenum and jejunal mucosa were scraped using a glass slide (10 cm length), and segments were stored at −80°C for total RNA isolation. Meanwhile, a segment from another arm of the cecum was collected and separated into two parts. One part was fixed in 4% paraformaldehyde, while the other part was kept in liquid nitrogen for further research.

The tissue samples (duodenum, cecum, liver, spleen, and feces) were aseptically collected from the euthanized animals at 5- and 7-days post-infection (dpi). Samples were serially diluted in sterile saline (0.9% NaCl) and aliquots were transferred to xylose lysine deoxycholate (XLD) agar (Land Bridge Co., Ltd., Beijing, China) plates supplemented with 50μg/mL nalidixic acid (BBI life science, Shanghai, China), which were incubated under aerobic conditions at 37°C for 24 h. Logarithmically transformed (log10) Salmonella counts were expressed in colony-forming units per gram (CFU/g). For the samples that showed no growth in the agar plate, the backup samples from the same source were partially diluted and transferred to buffered peptone water (BPW) (Land Bridge Co., Ltd., Beijing, China) incubated at 37°C for 24 hours under shaking. Then, they were plated onto brilliant green agar (Land Bridge Co., Ltd., Beijing, China) plates to detect Salmonella.

Tissue blocks from the duodenum, cecum and liver were fixed in the 4% paraformaldehyde buffer for 24 hours, rinsed with PBS repeatedly for 5 times, and then embedded in paraffin. Tissues were cut into serial sections (~5 μm thickness) and stained with hematoxylin and eosin (HE staining), the best overall quality section was used for histopathological analysis. According to Bandara et al. (39) and Memon et al. (40), two blinded investigators were scored histological inflammation and clinical appearance, including the degree of inflammation of liver and intestinal tissues, crypt damage, and the percentage of involvement of inflammatory area in each slide ( Table 1 , Figures 2A–C ).

We performed samples according to the Zhejiang University Bio-ultrastructure analysis Lab’s rules for animal sample preparation. In brief, the samples underwent double fixation, infiltration, embedding, ultrathin sectioning and staining, and were observed in H-7650 TEM (Hitachi, Ltd., Tokyo, Japan). Reagents used in the preparation process as glutaraldehyde, PBS, ethanol, acetone, lead citrate and methylene blue are all produced by Sinopharm Chemical Reagent Co., Ltd., Shanghai, China. Osmic acid, Spurr resin embedding agent and uranyl acetate are produced by Structure Probe, Inc. West Chester, PA, USA.

Histomorphological status was examined by HE-stained sections as described previously. Then the paraffin sections are dewaxed to water, and the tissue sections are placed in EDTA buffer (pH=9.0) at 95°C for antigen repair and maintained in sub boiling state for 15 minutes. It is important to avoid excessive evaporation of the buffer and to dry the slices as little as possible during this process. After natural cooling, the slices were placed in PBS (pH=7.4) and shaken for 15 min for decolorization. Then they were put into the 3% hydrogen peroxide solution and incubated at room temperature in the dark for 25 min to block endogenous peroxidase. Afterwards, they were placed in PBS (pH=7.4), shaken and washed for 15min. Further, 3% BSA was added dropwise to cover the slices evenly, and the serum was blocked at room temperature for 30 min. Then the blocking solution was removed, the diluted primary antibody (1:50) drops were added (Servicebio® Co., Ltd., Wuhan, China) to the slices, and the slice flat was placed in the wet box for incubation at 4°C overnight. The next day, the slice was placed in PBS (pH=7.4), shaken and washed for 15 min. After the sections were dried, the tissues were covered with secondary antibodies (1:500, HRP labelled) corresponding to the primary antibody and incubated at room temperature for 50 min. Next, after washing and drying the slices, the freshly prepared diaminobenzidine (DAB) color developing solution was added, the control of color development time was done under the microscope (the positive is brown-yellow), and slices were washed with distilled water to stop the color development.

Then the hematoxylin was counterstained for 3 min, washed with distilled water, and treated with hematoxylin differentiation solution for 10 s. Washing with distilled water, the hematoxylin returned to blue. Finally, to dehydrate the slice and make it transparent, the slices were put into 75% alcohol for 5 min, 85% alcohol for 5 min, 100% ethanol I for 5 min, 100% ethanol II for 5 min, n-butanol for 5 min, xylene I for 5 min. The slices were taken out from xylene, dried slightly, and sealed. After completing the above steps, the microscopic examination, image acquisition and analysis were operated. In the visual field, hematoxylin-stained nuclei are blue, and the positive expression of DAB is brownish-yellow.

Finally, the apoptotic cells and some tight junction protein aggregates were quantitatively analyzed with the aid of a light microscope (Olympus, Tokyo, Japan), which was equipped with a digital camera and an image analysis program (Image-Pro ® Plus version 6 software, Media Cybernetics, MD, USA). The data were showed as the number of stained cells per 1,000 μm2 in the villus and crypt regions.

The endotoxin activity (ET) was measured according to the instruction of the kit (Xiamen Tachypleus amebocyte lysate Reagent Co., Ltd., Fujian, China), and the activity was expressed in EU/mL. The cytokine D-lactic acid in serum was determined according to the procedure of the commercial ELISA kit (Jiancheng Bioengineering Institute, Nanjing, China). The competition method was used to detect the content of D-lactic acid in the sample. The serum was added into the enzyme-labelled hole pre-coated with antibody, and the recognition antigen labelled by horseradish peroxidase (HRP) was added. When incubated at 37°C for 1 hour, they compete with solid-phase antibodies to form immune complexes. After washing with PBS, the bound HRP catalyzes TMB (tetramethylbenzidine) to turn blue, then terminates the reaction under the action of sulfuric acid to turn yellow. There is an absorption peak at the wavelength of 450 nm. The absorbance value is inversely related to the antigen concentration in the sample.

The contents of immune-related factors IL-1β and IL-18 in serum were determined according to the operation manual of the commercial ELISA Kit (Solarbio® Sci & Tech Co., Ltd., Beijing, China). In short, they use the enzyme-linked immunosorbent assay based on the double antibody sandwich method to coat the monoclonal antibodies against chicken IL-1β and IL-18 on the enzyme label plate and add the standard and sample, respectively. The IL-1βand IL-18 in the sample will fully bind to the coated antibody. After washing with PBS, add a biotinylated secondary antibody, and the secondary antibody will specifically bind to IL-1β and IL-18. After washing with PBS, TMB is added, and the combined HRP catalyzes TMB to turn blue. After adding sulfuric acid termination solution, it turns yellow. The absorbance of the reaction pore sample was measured at 450 nm. The contents of IL-1β and IL-18 in the sample were positively correlated with the OD value. The contents of IL-1β and IL-18 in serum can be obtained by drawing a standard curve and four-parameter fitting calculation.

Total RNA of cecum samples was extracted using RNA Easy Fast animal tissue total RNA Extraction Kit (TIANGEN® Biotech, Beijing, China) according to the manufacturer’s protocol. The concentration and purity of total RNA were quantified with a NANODrop® ND-1000 UV-VIS spectrophotometer (Thermo Scientific, Wilmington, DE, USA) and agarose gel electrophoresis. The quantitative real-time PCR assay was performed with the Stratagene Mx3005P (with Mxpro 4.10 software) detection system (Agilent Technologies, Santa Clara, CA, USA) according to optimized PCR protocols using the SYBR qPCR Master Mix kit (Vazyme Biotechnology Co., Ltd., Nanjing, China). The primer pairs for amplifying genes encoding tight junction protein-related genes Claudin-1 and ZO-1, apoptosis-related gene Bax, and inflammasome-related gene IL-1β, and IL-18 are presented in Table 2 . GAPDH was used as an internal reference. The qPCR conditions were an initial denaturation step at 95°C for 30 s, 40 cycles at 95°C for 5 s, and annealing and extension temperature at 55-60°C ( Table 2 ) for 35 s. Each biological sample was run in triplicate. The method of 2−ΔΔCt (46) was used to calculate relative gene expression levels between different samples.

The data from the experiment were subjected to ANOVA after determining variance homogeneity by using the SPSS 16.0 software (IBM, New York, USA). The figures for data visualization were performed using GraphPad Prism 9.3 (GraphPad Software Inc., San Diego, CA). The analysis of other data was performed using Student’s t-test. The difference was considered to be statistically significant as P < 0.05, and the data were expressed as the means ± SEM.

To explore the effect of L. rhamnosus P118 on reducing Salmonella infection in chicks, according to our experimental plan, we used 160 female chicks, of which 80 were used to observe survival and growth, with 20 in each group. Another 80 were used for clinical trials ( Figure 1A ). They were equally divided into four groups and received the same environment and care as noted. It can be seen that on the 15th day, the mortality of chicks in group S was close to 50.00%, higher than that observed in group P+S pretreated with probiotic P118 (15.00%) (P <0.05), while no death was observed in groups C and P ( Figure 1B ). As shown in Figure 1C , there was no statistical difference in daily gain between groups C and P (P >0.05). On the other hand, chicks infected with Salmonella grow slowly, especially after six days. There was no significant difference in body weight between the treatment groups at or before the age of 6 days. However, from the age of 7 days to the end of the experiment, the difference between the groups was significant (P <0.05) and gradually increased. Due to the early intervention of L. rhamnosus P118, even after the challenge, the adverse situation of Salmonella infection in the P+S group was reversed.

To determine the effect of the oral Salmonella challenge, the loads of S. Typhimurium in the liver, spleen, duodenum, cecum and feces were determined. S. Typhimurium was not detected in the tissue samples from the C and P groups. In comparison with the S group, the number of S. Typhimurium colonized in the tissues and feces was significantly lower in the P+S group (P < 0.05) ( Figures 1D, E ). Moreover, we found that when compared with five dpi, the clinical symptoms and organ bacterial loads of chicks on day seven dpi were significantly (P <0.05) different between the groups ( Figures 1D, E ). Therefore, the pathological sections, serum biochemical indexes, determination of immune factors, immunohistochemical analysis, full-length 16S rRNA sequencing and other tests mentioned later are all based on the chicks’ samples on day seven dpi.

In order to determine if liver and intestinal integrity were affected by S. Typhimurium infection, the morphology of samples was monitored. Only the S group showed large number of heterophils, indicating a mild inflammatory response at seven dpi. Hematoxylin-eosin-stained pathological sections showed that chicks in group S presented a disordered structure of hepatic cord, unclear structure of hepatic lobules, narrowing or even disappearance of hepatic sinuses, loose spacing of hepatocytes and a large number of pink eosinophilic particles ( Figure 2A ). S. Typhimurium caused severe damage to the villi morphological structure of the duodenum and cecum ( Figure 2B, C ). The integrity of the intestinal inner wall of the duodenum in group S was damaged, eosinophils were found in the inner wall cells, duodenal mucous gland cells disappeared, the morphology of intestinal villi was seriously degraded, some crypt structures were unclear, and the relative depth of the surviving crypts increased ( Figure 2B ). In group S, the cecal villi were also significantly shorter, some epithelium on the mucosal surface was missing, inflammatory cells in lamina propria were increased, and neutrophil infiltration was prominent. Still, the overall structure was visible ( Figure 2C ). It significantly reduced the height of the duodenal villi and increased the depth of recess ( Figures 2D, E , Table S2 ). Moreover, S. Typhimurium strongly increased (P < 0.05) the appearance score, inflammation score, diarrhea score and crypt damage score. L. rhamnosus P118 pretreatment (P + S) significantly (P < 0.05) reversed the trend and decreased all the group’s scores compared with the S group. ( Figure 2F , Table S1 ).

Based on this result, we attempted to visualize microvilli ultrastructure in high-resolution transmission electron microscopy (TEM) to examine the effects of L. rhamnosus P118 stimulation on intestinal microvillus development ( Figure 3A ). The results showed that the microvilli of the chicks treated with L. rhamnosus P118 had a significantly 1.3 times longer length (1208.4 ± 28.02 nm) ( Table S2 ) and more perpendicular to the top surface of intestinal cells compared with the C group ( Figures 3A, B ). The density of microvilli was higher than P+S (P< 0.05) ( Figure 3C , Table S2 ).

To determine that the morphological changes in the chicken intestinal tract are indeed affected by barrier function-related proteins, claudin-1 and ZO-1 were used for immunohistochemical staining of the duodenum. Figures 4A, B shown that these two proteins will be dyed brown yellow. The arrow in the figure refers to a part of the stained tight junction protein.

To investigate why S. Typhimurium and L. rhamnosus P118 changed the intestinal permeability, and the expression of TJ genes claudin-1 and ZO-1 were measured by qPCR. As shown in Figure 4C , there was a significant difference in claudin-1 and ZO-1 at the mRNA level among all four groups (P < 0.05). In comparison with the C group, Salmonella infection (group S) significantly (P <0.05) decreased the mRNA levels of ZO-1 and claudin-1( Figures 4C, D ). The down-regulation of ZO-1 and claudin-1 due to Salmonella infection was eliminated in the P+S group. Moreover, in the group P, the ZO-1 expression was significantly (P <0.05) higher than that in group C ( Figure 4D ).

To assess the intestinal permeability among different treated groups, levels of mediators D-lactic acid and endotoxin (ET) in serum samples were determined. As shown in Figures 5A, B S. Typhimurium infection significantly (P <0.05) increased intestinal permeability compared with the C group. However, the D-lactic acid and ET in the serum of the chicks in the P+S group were significantly lower than those in the S group (P < 0.05). There was no significant difference between the P group and the P+S group in endotoxin (P > 0.05) ( Table S3 ).

To elucidate changes in the inflammatory responses induced by Salmonella infection, ELISA was used to measure the levels of cytokines IL-1β and IL-18 in serum samples ( Figures 5C, D ). Compared with the C group, Salmonella infection significantly increased the levels of cytokines in serum (P < 0.05). However, after probiotic pretreatment, the levels of IL-1β and IL-18 were significantly decreased in the P+S group compared with the S group (P < 0.05) ( Figures 5C, D , Table S4 ).

Apoptosis also affects intestinal barrier function. Apoptotic cells were found in the S group, and P+S group, as shown in Figure 6A , the cytoplasm of cells dyed brown-yellow is apoptotic cells containing Bax gene expression products. The more cells stained per unit area, the higher the frequency of apoptotic events. After Salmonella infection, the number of apoptotic cells in group S was significantly (P < 0.05) higher than that in the control group ( Figure 6B ).

To investigate whether Salmonella infection caused apoptosis in chicks, immunohistochemical staining and qPCR were also used to detect the levels of apoptosis-related proteins and mRNA. Compared with group S, the process of apoptosis was significantly (P <0.05) reduced after probiotic pretreatment. The change in the trend of the mRNA expression level of Bax is mutually confirmed with the above results, it was highly expressed in group S and significantly (P < 0.05) higher than in other treatment groups. Similarly, L. rhamnosus P118 pretreatment improved this trend ( Figure 6C ).

In terms of the genus of full-length 16s rRNA sequencing ( Figure 7A ), we highlight the marked increase in Lactobacillus spp. in the feces of chicks pretreated with L. rhamnosus P118 (32.15%) in group P compared with S (1.10%), P+S (10.10%), and C (6.79%) groups. Variations in the abundance of Salmonella across the groups were also observed: group P+S showed significantly decreased abundance (9.88%) compared with group S (19.85%), while the Salmonella spp. was not detected in group C and group P. Interestingly, in the Salmonella infection group, the proportion of Salmonella is only the second, and the dominant genus is still Enterococcus, and the sum of the two genera reaches 70.27% in top ten genera, which dramatically reduces the living space of other genera. Relatively, the proportion of various genera in the P+S group is balanced.

The corresponding alpha diversity analysis is shown in Figures 7B, C . Using the Chao 1 and ace indexes analysis, the abundance of microbes in the guts of chicks was further validated. The species diversity in chicks after Salmonella challenge (S group) was significantly (P < 0.05) decreased, while in the P+S group, it was increased obviously (P < 0.05) compared with the group C (ace indexes). Investigation of species and ace indexes showed that pretreatment with L. rhamnosus P118 significantly increased species richness and stabilized the proportion of balanced gut microbiota compared with controls ( Figure 7A, C ).

LEfSe analysis is shown in Figure 7D , which presents OTU at different taxonomic levels that are significantly different between group S and the other groups ( Figure S1 ) (P < 0.05). On day seven dpi, when compared with the P+S, the S group showed an increase in the relative abundance of Erysipelatoclostridium and the cecum contents ( Figure 7E , P<0.05) and a decrease in Lactobacillus spp. abundance in the cecum. At the same time, compared with the C group, at the genus level, it is the same as S vs P+S, but the relative abundance of Erysipelatoclostridium is still significant different ( Figure 7F ).