Genetically engineered probiotic E. coli Nissle 1917 enhances protection against Salmonella via increased adhesion and systemic T-cell responses Authors: Ewa Carolak, Joanna Czajkowska, Wiktoria Waszczuk, Agata Dutkiewicz, Anna Ewa Kedzierska, Krzysztof Grzymajlo Published: 2026 ABSTRACT The rise in antimicrobial resistance underscores the need for innovative strategies to combat gastrointestinal infections. Probiotics such as Escherichia coli Nissle 1917 (EcN) offer promising options, but the molecular mechanisms underlying their protective effects remain unclear. We introduce a G66R point mutation in FimH, creating a high-binding EcN variant that more effectively prevents Salmonella Typhimurium attachment and induces a distinct host transcriptional profile, shifting toward adaptive rather than innate inflammatory signaling. In vivo, EcNG66R pretreatment significantly reduced intestinal colonization, fecal shedding, and systemic spread, and prevented splenic enlargement compared with EcNWT. Protection was associated with a marked expansion of CD4+ and CD8+ T cells, essential for clearing intracellular pathogens. EcNG66R further enhanced “readiness” in the spleen under non-infected conditions, without adverse effects on host physiology. EcNG66R thus functions as a dual-action probiotic—improving competitive exclusion while priming cytotoxic T-cell-mediated protection—and provides a promising platform for developing next-generation microbe-based therapies. Introduction Gastrointestinal infections remain a major global health concern, with Salmonella spp. among the leading causes of foodborne illness worldwide. These pathogens cause a wide range of symptoms, from mild gastroenteritis to severe systemic disease, especially in vulnerable groups such as children, the elderly, and immunocompromised individuals. The infection starts with invasion of the intestines, then Salmonella spreads to distal tissues, including mesenteric lymph nodes, liver, and spleen1. The spleen, a vital lymphoid organ where immune responses to blood-borne pathogens are initiated, becomes a key site for bacterial replication during systemic salmonellosis1. The host defense against Salmonella infection involves an interplay between immediate innate immunity and adaptive immune memory. Although initial bacterial control relies on rapid innate responses, long-term protection depends on establishing antigen-specific B and T cell populations capable of mounting recall responses upon re-exposure. The cellular immune response, mediated by helper CD4+ and cytotoxic CD8+ T cells, is particularly important for eliminating this facultative intracellular pathogen from infected macrophages and epithelial cells2. With the increasing emergence of multidrug-resistant pathogens, conventional antibiotic therapy faces growing limitations, necessitating the development of innovative preventive and therapeutic strategies. Probiotic bacteria have gained significant attention as an effective approach to prevent or reduce gastrointestinal infections. Their beneficial effects occur through various mechanisms, including competitive exclusion for adhesion sites and nutrients, production of antimicrobial substances, and modulation of the host immune response3. Escherichia coli Nissle 1917 (EcN) is one of the best-characterized probiotics, originally isolated from a German soldier who remained healthy during a severe shigellosis outbreak in 19174. EcN (serotype O6:K5:H1) has been extensively studied and clinically applied for over a century5–7. EcN possesses unique genetic and phenotypic characteristics that distinguish it from pathogenic E. coli strains. It lacks typical virulence factors while expressing beneficial features such as microcin production, which inhibits enteropathogenic bacteria in the inflamed gut8. Additionally, EcN does not produce enterotoxins or cytotoxins6 and exhibits acid and bile tolerance, crucial for survival in the gastrointestinal tract5. Yet despite extensive clinical application and documented therapeutic benefits, the molecular mechanisms driving its protective effects against enteric pathogens remain only partially characterized9. Bacterial adhesion to intestinal epithelial cells represents a critical determinant in the probiotic–pathogen competitive relationship. In E. coli, including probiotic strains, adhesion is primarily mediated by fimbrial structures6. Type 1 fimbriae (T1F) are among the most studied adhesive organelles, with the FimH adhesin located at their tip serving as the primary binding molecule. FimH is a mannose-specific lectin that recognizes and binds to mannosylated glycoproteins on the host cell surface. The binding properties of FimH can vary significantly between strains, with even single amino acid substitutions capable of dramatically altering binding affinity and specificity10. Therefore, enhancing FimH-mediated adhesion may strengthen a probiotic’s ability to compete with pathogens for epithelial binding sites, thus reinforcing colonization resistance. However, improved adhesion may also influence directly or indirectly other probiotic functions, including host signaling, immune interactions, and microbial fitness in the gut. FimH functions as an immunomodulatory molecule beyond its adhesive role. By acting as a direct TLR4 ligand, it activates MyD88-dependent NF-κB signaling and drives pro-inflammatory cytokine production, including IFN-β and IL-1β, in epithelial cells and macrophages11–13. FimH stimulates NK cell cytotoxicity, promotes dendritic cell maturation, and enhances antigen-specific T cell responses14,15. In the Salmonella infection model, FimH contributes directly to the induction of intestinal inflammation during bacterial invasion16. The present study aimed to assess the effect of genetically enhanced EcN with improved adhesion properties. Among several point mutations that increase EcN binding to epithelial cell lines in vitro, we chose the G66R mutant in the FimH adhesin. We examined EcN’s high-binding phenotype in protecting against Salmonella Typhimurium infection and found that the mutant strain protects the host more effectively. We thoroughly characterized the phenotypic changes caused by the G66R mutation in the probiotic context and evaluated its influence on infection outcome, host immune responses, and protection against pathogen challenge. Results Point mutations in the fimH gene of E. coli Nissle 1917 improve its ability to attach to intestinal epithelial cell lines while causing minor alterations in host gene expression One potential mechanism of probiotic action is competing with pathogens for environmental niches and binding sites on host intestinal cells. Here, we selected high-binding phenotypes of pathogenic E. coli17–19, and introduced point mutations responsible for those phenotypes into the fimH gene (G66R, T158P, and Y186R) of E. coli Nissle 1917 (EcN) (Table 1, Fig. 1A). Mutant strains showed no detectable differences in growth dynamics compared to the wild-type strain (data not shown). Subsequently, we assessed its adhesion properties using two intestinal epithelial cell lines (Fig. 1B, C). The number of adhered bacteria increased significantly, approximately 4.1 times for EcNG66R (p < 0.0001), 1.6 times for T158P (p < 0.05), and 3.1 times for Y186R (p < 0.001) compared to the EcNWT strain during interaction with the MIEC cell line (Fig. 1B). Notably, this binding pattern stayed consistent during interaction with the IPEC-J2 cell line (Fig. 1C). To determine whether the improved binding could be replicated by simply increasing the bacterial dose, we conducted adhesion assays using different MOI levels for the mutant with the most significant difference in adhesion. Importantly, even at elevated MOI, EcNWT did not achieve binding levels comparable to the EcNG66R (Fig. 1D). The probiotic mechanisms of action involve the activation or silencing of innate responses in intestinal cells17,18,20. Therefore, we assessed whether the altered binding pattern influences the expression levels of selected cytokines associated with gastrointestinal infections. Interestingly, among the tested cytokines, the only statistically significant difference was observed for the IL-6 gene, with decreased expression in the MIEC cell line in the presence of the selected mutant compared to EcNWT (Fig.1E) (p < 0.05). Expression levels of IL-1β, IL-10, IL-12, IL-19, and IFN-γ showed no significant differences between mutant and WT treatments (Supplementary Fig. 3A). These minimal differences led us to investigate how the EcNG66R affects the overall MIEC transcriptional response using RNA sequencing (RNA-seq). After DEG analysis, we observed 143 and 43 genes significantly up- and downregulated, respectively, in EcNWT compared to the untreated control (Fig. 1F). The EcNG66R mutant increased the expression of 280 genes and decreased the expression of 119 genes compared to untreated cells (Fig. 1G). Venn diagram analysis showed overlap in responses between the two strains: 129 genes were upregulated by both, with 151 genes uniquely upregulated by EcNG66R and only 14 by EcNWT. Only 16 genes were commonly suppressed, while 103 genes were uniquely downregulated by EcNG66R and 27 genes by EcNWT (Supplementary Fig. 3B, Supplementary Table 4).Fig. 1A point mutation in the fimH gene affects the binding properties of E. coli Nissle 1917 and modifies host cell response.A Structural visualization of introduced mutations in the fimH gene using ColabFold (implementation of AlphaFold2. Amino acid substitutions are highlighted on the protein structure (the pilin domain (gray), the lectin domain (red)). Original residues (blue), mutated residues (green), from top to bottom: G66R, T158P, Y186R. Binding of EcN and its mutants to mice (B) or porcine (C) intestinal epithelial cell lines at MOI 100. D Binding of EcNWT and EcNG66R to MIEC at MOI 10, 100, and 1000. E Relative expression of IL-6 following incubation of the MIEC with E. coli Nissle and its fimH point mutants. Volcano plots showing (DEGs) between: F MIEC untreated versus MIEC after exposure to EcNWT; G MIEC untreated versus MIEC after exposure to EcNG66R; H MIEC after exposure to EcNG66R and the MIEC after exposure to EcNWT. Genes significantly up-regulated—red (right side), genes significantly down-regulated—blue (left side). Not significant genes—gray. Labeled genes—top 20 most significant by p-value. Vertical dashed lines mark the ±1log₂FC threshold, and the horizontal dashed line indicates the −log₁₀(padj) threshold. Statistical differences between strains and various conditions were analyzed by one-way ANOVA with Tukey’s post-hoc test, preceded by a normality test, using GraphPad Prism version 11.0.0 for Windows, GraphPad Software, Boston, MA, USA. www.graphpad.com.Table 1Bacterial strains used in this studyNameStrainMutation in fimHSourceEcNWTEscherichia coli Nissle 1917Wild-typeSchierack lab strain collectionEcNG66REscherichia coli Nissle 1917 with a point mutation (G66R) in fimH gene66 (Gly-Arg)GGC to CGC;originally reported in ref. 17This studyEcNTI58PEscherichia coli Nissle 1917 with a point mutation (T158P) in fimH gene158 (Thr-Pro)ACT to CCT;originally reported in ref. 19This studyEcNY186REscherichia coli Nissle 1917 with a point mutation (Y186R) in fimH gene186 (Tyr-Arg)TAT to CGT,originally reported in ref. 18This studyS. TmWTS. Typhimurium SL1344Wild-typeMonack lab strain collectionS. TmWT_mCherryS. Typhimurium SL1344 with mCherry plasmidn/aThis studyE. coli DH5α λpirEscherichia coli DH5α with integrated conjugal transfer functionsn/aMonack lab strain collectionE. coli S17-1 λpirE. coli S17-1 with integrated conjugal transfer functionsn/aMonack lab strain collectionEcNWT-EryREscherichia coli Nissle 1917; EryRWild-typeThis studyEcNG66R-EryREscherichia coli Nissle 1917 with a point mutation (G66R) in fimH gene; EryR66 (Gly-Arg)This studyEcNWT-NalREscherichia coli Nissle 1917; NalRWild-typeThis studyEcNG66R-NalREscherichia coli Nissle 1917 with a point mutation (G66R) in fimH gene; NalR66 (Gly-Arg)This study A point mutation in the fimH gene affects the binding properties of E. coli Nissle 1917 and modifies host cell response. A Structural visualization of introduced mutations in the fimH gene using ColabFold (implementation of AlphaFold2. Amino acid substitutions are highlighted on the protein structure (the pilin domain (gray), the lectin domain (red)). Original residues (blue), mutated residues (green), from top to bottom: G66R, T158P, Y186R. Binding of EcN and its mutants to mice (B) or porcine (C) intestinal epithelial cell lines at MOI 100. D Binding of EcNWT and EcNG66R to MIEC at MOI 10, 100, and 1000. E Relative expression of IL-6 following incubation of the MIEC with E. coli Nissle and its fimH point mutants. Volcano plots showing (DEGs) between: F MIEC untreated versus MIEC after exposure to EcNWT; G MIEC untreated versus MIEC after exposure to EcNG66R; H MIEC after exposure to EcNG66R and the MIEC after exposure to EcNWT. Genes significantly up-regulated—red (right side), genes significantly down-regulated—blue (left side). Not significant genes—gray. Labeled genes—top 20 most significant by p-value. Vertical dashed lines mark the ±1log₂FC threshold, and the horizontal dashed line indicates the −log₁₀(padj) threshold. Statistical differences between strains and various conditions were analyzed by one-way ANOVA with Tukey’s post-hoc test, preceded by a normality test, using GraphPad Prism version 11.0.0 for Windows, GraphPad Software, Boston, MA, USA. www.graphpad.com. Table 1Bacterial strains used in this studyNameStrainMutation in fimHSourceEcNWTEscherichia coli Nissle 1917Wild-typeSchierack lab strain collectionEcNG66REscherichia coli Nissle 1917 with a point mutation (G66R) in fimH gene66 (Gly-Arg)GGC to CGC;originally reported in ref. 17This studyEcNTI58PEscherichia coli Nissle 1917 with a point mutation (T158P) in fimH gene158 (Thr-Pro)ACT to CCT;originally reported in ref. 19This studyEcNY186REscherichia coli Nissle 1917 with a point mutation (Y186R) in fimH gene186 (Tyr-Arg)TAT to CGT,originally reported in ref. 18This studyS. TmWTS. Typhimurium SL1344Wild-typeMonack lab strain collectionS. TmWT_mCherryS. Typhimurium SL1344 with mCherry plasmidn/aThis studyE. coli DH5α λpirEscherichia coli DH5α with integrated conjugal transfer functionsn/aMonack lab strain collectionE. coli S17-1 λpirE. coli S17-1 with integrated conjugal transfer functionsn/aMonack lab strain collectionEcNWT-EryREscherichia coli Nissle 1917; EryRWild-typeThis studyEcNG66R-EryREscherichia coli Nissle 1917 with a point mutation (G66R) in fimH gene; EryR66 (Gly-Arg)This studyEcNWT-NalREscherichia coli Nissle 1917; NalRWild-typeThis studyEcNG66R-NalREscherichia coli Nissle 1917 with a point mutation (G66R) in fimH gene; NalR66 (Gly-Arg)This study Bacterial strains used in this study 66 (Gly-Arg) GGC to CGC; originally reported in ref. 17 158 (Thr-Pro) ACT to CCT; originally reported in ref. 19 186 (Tyr-Arg) TAT to CGT, originally reported in ref. 18 To better illustrate that the differences between the probiotic strains are modest, significant DEGs were ranked by the standard deviation of log2 (fold change). The top 50 most variable genes for each condition were plotted (Supplementary Fig. 3C), revealing that, even among these most strongly affected genes, expression changes rarely exceeded ~2-fold differences between strains. To compare the transcriptional programs triggered by EcNWT and EcNG66R, we conducted direct differential expression analysis. We identified 14 genes with significantly higher expression in EcNG66R-treated cells, primarily pseudogenes and ncRNAs, and 17 genes with higher expression in EcNWT-treated cells (Fig. 1H/Supplementary Table 6). EcNG66R mainly upregulated GAPDH and non-coding RNAs. EcNWT induced a coordinated cellular defense program characterized by antioxidant protection, regulated inflammatory signaling, and improved protein trafficking. Importantly, when we specifically examined immune-related genes, we found no significant differences between the two strains: among 765 immune-associated genes analyzed, indicating that the FimH G66R mutation does not fundamentally alter the core immunological transcriptional response in MIEC cells. EcNG66R and EcNT158P mutants inhibit Salmonella Typhimurium cell attachment more effectively than EcNWT Given the critical role of competitive exclusion in probiotic function, we re-evaluated all three mutants to test whether their increased adhesion improves protection against pathogens (Fig. 2A). Pretreatment with EcNG66R and EcNT158P significantly reduced STm binding compared with EcNWT (Fig. 2B), with EcNG66R showing the most potent protective effect. This consistent superior performance of the EcNG66R mutant across both adhesion and pathogen-exclusion assays reinforced our decision to select this variant for further investigation. To assess potential immune modulation, the expression of cytokines previously examined in probiotic-treated cells was re-evaluated by qPCR in MIEC cells infected with STm, with or without probiotic pretreatment (Supplementary Fig. 4A). Notably, pre-treatment with either EcNWT or EcNG66R mutants significantly reduced STm-induced IL-6 expression compared to STm infection alone (Fig. 2C) (p < 0.01). However, no significant differences in IL-6 expression were observed between EcNWT and mutant EcNG66R pre-treatments in the presence of STm (Fig. 2C). Expression levels of IL-1β, IL-10, IL-12, IL-19, and IFN-γ remained unchanged (Supplementary Fig. 4A). To identify genes uniquely affected by each treatment condition, we compared all three setups (STm alone, EcNWT + STm, EcNG66R + STm) against uninfected control cells (Fig. 2D, E). Venn diagram analysis revealed sets of uniquely regulated genes: 234 genes were downregulated in EcNWT + STm, 114 in EcNG66R + STm, and 18 in STm alone. Similarly, 170 genes were upregulated in EcNWT + STm, 66 in EcNG66R + STm, and 32 in STm alone (Supplementary Fig. 4B). This suggests that pre-treatment with either probiotic strain significantly alters the cellular response to Salmonella infection. Further analysis of gene expression variability across the three infection conditions highlighted the transcripts that responded most dynamically to treatment. The top 50 differentially expressed genes were selected based on the magnitude of fold-change variation (Supplementary Fig. 4C). Notably, several inflammatory and stress-response genes, including Ccl2, Cxcl1, Egln3, Cxcl2, and Cebpd, showed the strongest modulation. When directly compared, infection with STm in the presence of EcNWT resulted in 24 upregulated genes, whereas STm infection in the presence of EcNG66R resulted in 84 upregulated genes (Fig. 2F). Analysis of immune-related genes specifically revealed that among 791 immunological genes examined, 11 showed significant differential expression between the two treatment groups, with nine genes upregulated and two genes downregulated in EcNG66R + STm compared to EcNWT + STm (Fig. 2G).Fig. 2Probiotics with enhanced adhesion properties affect Salmonella–host cell interactions.A Schematic illustrating the experimental workflow for the competitive exclusion assay and subsequent analysis. Created in BioRender. Grzymajlo, K. (2026) https://BioRender.com/zhk97cx. B Inhibition of STm adhesion to the MIEC preincubated with different EcN variants. The bars show the reduction in STm adhesion level, and the background shading indicates the adhesion level of STm alone. C Relative expression of IL-6 after incubation of the MIEC with EcNWT and EcNG66R followed by infection with STm. Volcano plot showing DEGs between: MIEC after exposure to STm versus: D MIEC after exposure to EcNWT and then STm; E MIEC after exposure to EcNG66R and then STm; F MIEC after exposure to EcNWT and STm versus the MIEC after exposure to EcNG66R and STm. Genes significantly up-regulated—red (right side), genes significantly down-regulated—blue (left side). Not significant genes—gray. Labeled genes- top 20 most significant by p-value. Vertical dashed lines indicate the ±1log₂FC threshold, and the horizontal dashed line indicates the −log₁₀(padj) threshold. G Top differentially expressed immunological genes identified by RNA-seq analysis in the MIEC after exposure to EcNWT and STm versus the MIEC after exposure to EcNG66R and STm. The bar plot displays immune-related genes - upregulated (red) and downregulated (blue), with the highest absolute log₂ fold change. Statistical significance was assessed using a one-way ANOVA with Tukey’s post hoc test for data that met the assumptions of normality, whereas the Kruskal–Wallis test with Dunn’s post hoc test was used when the data did not meet these assumptions. Statistical analyses were performed using GraphPad Prism version 11.0.0 for Windows, GraphPad Software, Boston, MA, USA. www.graphpad.com. Probiotics with enhanced adhesion properties affect Salmonella–host cell interactions. A Schematic illustrating the experimental workflow for the competitive exclusion assay and subsequent analysis. Created in BioRender. Grzymajlo, K. (2026) https://BioRender.com/zhk97cx. B Inhibition of STm adhesion to the MIEC preincubated with different EcN variants. The bars show the reduction in STm adhesion level, and the background shading indicates the adhesion level of STm alone. C Relative expression of IL-6 after incubation of the MIEC with EcNWT and EcNG66R followed by infection with STm. Volcano plot showing DEGs between: MIEC after exposure to STm versus: D MIEC after exposure to EcNWT and then STm; E MIEC after exposure to EcNG66R and then STm; F MIEC after exposure to EcNWT and STm versus the MIEC after exposure to EcNG66R and STm. Genes significantly up-regulated—red (right side), genes significantly down-regulated—blue (left side). Not significant genes—gray. Labeled genes- top 20 most significant by p-value. Vertical dashed lines indicate the ±1log₂FC threshold, and the horizontal dashed line indicates the −log₁₀(padj) threshold. G Top differentially expressed immunological genes identified by RNA-seq analysis in the MIEC after exposure to EcNWT and STm versus the MIEC after exposure to EcNG66R and STm. The bar plot displays immune-related genes - upregulated (red) and downregulated (blue), with the highest absolute log₂ fold change. Statistical significance was assessed using a one-way ANOVA with Tukey’s post hoc test for data that met the assumptions of normality, whereas the Kruskal–Wallis test with Dunn’s post hoc test was used when the data did not meet these assumptions. Statistical analyses were performed using GraphPad Prism version 11.0.0 for Windows, GraphPad Software, Boston, MA, USA. www.graphpad.com. EcNG66R offers improved protection against systemic salmonellosis by reducing bacterial colonization and preventing infection-associated pathology Based on the in vitro effects shown by EcNG66R, we tested whether the mutant provides better protection for the host than EcNWT in a Salmonella-susceptible BALB/c mouse model. We administered a high dose (108) of the pathogen by oral gavage to induce systemic infection via the gastrointestinal route (Fig. 3A). Six days after infection, mice in both groups showed symptoms of salmonellosis (e.g., reduced appetite, weight loss, ruffled fur, hunched posture, lethargy), but mice pre-treated with EcNG66R showed milder clinical signs throughout the infection: they had better coat condition, more natural posture, and higher spontaneous activity compared to EcNWT-pretreated animals. Blood analysis revealed significantly higher glucose levels in the mice pre-treated with EcNG66R and then infected with STm (average 6.5 mmol/l) compared to the EcNWT + STm group (average 3.1 mmol/l; p < 0.05) (Supplementary Fig. 5A). The EcNG66R + STm group also had significantly higher magnesium levels (average 1.4 versus 1.0 mmol/l in EcNWT + STm; p < 0.05), while potassium, sodium, and iron levels showed trends toward differences between groups, but these did not reach statistical significance (Supplementary Fig. 5A).Fig. 3EcNG66R protects BALB/c mice more efficiently against Salmonella Typhimurium infection.A Schematic representation of the animal infection experiment workflow. Created in BioRender. Grzymajlo, K. (2026) https://BioRender.com/zhk97cx. B Percentage change in body weight following STm infection in BALB/c mice pretreated with either EcNWT (blue) or EcNG66R (green). (Top left) Comparison of organ weights (in grams) from mice infected with STm mice pretreated with EcNWT (blue)(n = 8) or EcNG66R (green)(n = 5). C Spleen, D liver, E mesenteric lymph nodes (MLN), gallbladder, gut, and intestine. Each data point represents an individual mouse, and horizontal lines indicate the mean ± SEM. F STm bacterial counts in feces (CFU/g) over 6 days post-infection (D1–D6) comparing EcNWT(blue dots) and EcNG66R (green dots) treatments. Bacterial loads in organs: G spleen, H liver, I MLM, gallbladder, gut, and intestine. Data are expressed as CFU/g, with each symbol representing an individual animal—EcNWT (blue dots)(n = 8) and EcNG66R (green dots)(n = 5), and bars indicating mean ± SEM. For logarithmic presentation, uncolonized samples (no detectable bacteria) were assigned a threshold value of 1 CFU/g. J Relative expression of IL-10 in the spleen of mice post-infection with STm with pretreatment with EcNWT (blue) and EcNG66R (green). K Volcano plot showing DEGs between spleens from mice pretreated with EcNWT + STm versus EcNG66R + STm. Genes with TPM > 1 in both groups, |log₂(fold change)| > 1, and adjusted p-value < 0.05 are plotted. Genes significantly up-regulated (red) (right side) and genes significantly down-regulated (blue) (left side). Not significant genes—gray. Labeled genes—top 20 most significant by p-value. Vertical dashed lines indicate the ±1 log₂FC threshold, and the horizontal dashed line indicates the -log₁₀(padj) threshold. L Top differentially expressed immunological genes identified by RNA-seq analysis in spleens from mice pretreated with EcNWT versus EcNG66R, both followed by STm infection. The bar plot shows the top 20 immune-related genes with the highest absolute log₂ fold change between the experimental conditions analyzed. Only genes with adjusted p ≤ 0.05, |log₂FC| ≥ 1, and TPM ≥ 1 were included. Red bars indicate genes upregulated under the tested condition, while blue bars represent genes downregulated in comparison. Statistical comparisons were performed using: 2-way ANOVA with multiple comparison test (F), unpaired Student’s t-test for data that met the assumptions of normality, and Mann–Whitney for the data that did not meet the assumptions of normality (C–E, G–I), after the removal of outliers (ROUT, Q = 1%), using GraphPad Prism version 11.0.0 for Windows, GraphPad Software, Boston, MA, USA, www.graphpad.com. The number of mice per group varies from 6 to 9. EcNG66R protects BALB/c mice more efficiently against Salmonella Typhimurium infection. A Schematic representation of the animal infection experiment workflow. Created in BioRender. Grzymajlo, K. (2026) https://BioRender.com/zhk97cx. B Percentage change in body weight following STm infection in BALB/c mice pretreated with either EcNWT (blue) or EcNG66R (green). (Top left) Comparison of organ weights (in grams) from mice infected with STm mice pretreated with EcNWT (blue)(n = 8) or EcNG66R (green)(n = 5). C Spleen, D liver, E mesenteric lymph nodes (MLN), gallbladder, gut, and intestine. Each data point represents an individual mouse, and horizontal lines indicate the mean ± SEM. F STm bacterial counts in feces (CFU/g) over 6 days post-infection (D1–D6) comparing EcNWT(blue dots) and EcNG66R (green dots) treatments. Bacterial loads in organs: G spleen, H liver, I MLM, gallbladder, gut, and intestine. Data are expressed as CFU/g, with each symbol representing an individual animal—EcNWT (blue dots)(n = 8) and EcNG66R (green dots)(n = 5), and bars indicating mean ± SEM. For logarithmic presentation, uncolonized samples (no detectable bacteria) were assigned a threshold value of 1 CFU/g. J Relative expression of IL-10 in the spleen of mice post-infection with STm with pretreatment with EcNWT (blue) and EcNG66R (green). K Volcano plot showing DEGs between spleens from mice pretreated with EcNWT + STm versus EcNG66R + STm. Genes with TPM > 1 in both groups, |log₂(fold change)| > 1, and adjusted p-value < 0.05 are plotted. Genes significantly up-regulated (red) (right side) and genes significantly down-regulated (blue) (left side). Not significant genes—gray. Labeled genes—top 20 most significant by p-value. Vertical dashed lines indicate the ±1 log₂FC threshold, and the horizontal dashed line indicates the -log₁₀(padj) threshold. L Top differentially expressed immunological genes identified by RNA-seq analysis in spleens from mice pretreated with EcNWT versus EcNG66R, both followed by STm infection. The bar plot shows the top 20 immune-related genes with the highest absolute log₂ fold change between the experimental conditions analyzed. Only genes with adjusted p ≤ 0.05, |log₂FC| ≥ 1, and TPM ≥ 1 were included. Red bars indicate genes upregulated under the tested condition, while blue bars represent genes downregulated in comparison. Statistical comparisons were performed using: 2-way ANOVA with multiple comparison test (F), unpaired Student’s t-test for data that met the assumptions of normality, and Mann–Whitney for the data that did not meet the assumptions of normality (C–E, G–I), after the removal of outliers (ROUT, Q = 1%), using GraphPad Prism version 11.0.0 for Windows, GraphPad Software, Boston, MA, USA, www.graphpad.com. The number of mice per group varies from 6 to 9. In the EcNG66R-treated group, the average mouse weight six days after STm infection dropped to ~95% of the initial weight, whereas in the EcNWT-treated group, it was about 84% (Fig. 3B). Importantly, mice pre-treated with the EcNG66R mutant did not develop splenomegaly, a typical syndrome of systemic salmonellosis21, in contrast to the EcNWT pre-treated group (Fig. 3C). In this group, we also observed an enlarged liver and intestine mass compared to the EcNG66R group (Fig. 3D, E). No differences were detected in the mass of the gallbladder, lymph nodes, or gut (Fig. 3E). During the first 3 days after infection, most mice in the EcNG66R group did not exhibit fecal shedding, whereas all mice in the EcNWT-treated group shed feces from the first day. Over the following days (D5, D6 pi), shedding levels were significantly lower (p < 0.05) in the EcNG66R group compared to EcNWT (Fig. 3F). Further validation of the protective effect of EcNG66R was performed by 16S rRNA gene sequencing of fecal samples and intestines from infected mice (Supplementary Table 5A, B). At day 1 post-infection, STm was detected only in EcNWT-treated mice (6 reads, 0.01039%), while EcNG66R-treated mice showed none. By day 4 post-infection, the difference widened: Salmonella abundance in EcNWT + STm feces reached 5.823% (335 reads), whereas in EcNG66R + STm mice, Salmonella constituted only 0.1503% (88 reads)—an ~39-fold. Total fecal Enterobacteriaceae followed the same pattern, reaching 0.913% (536 reads) in EcNWT + STm (driven primarily by Salmonella expansion) versus 0.2107% (125 reads) in EcNG66R + STm (~4.3-fold reduction) at day 4. Analysis of intestinal tissue colonization at day 6 post-infection confirmed the profound protective effect of EcNG66R (Supplementary Table 5B). EcNWT + STm mice showed substantial Salmonella colonization (3997 reads; 8.391%), indicating robust mucosal colonization, whereas EcNG66R + STm mice had only three reads (0.00554%)—a ~1500-fold reduction. Total Enterobacteriaceae abundance mirrored this pattern: 17.55% in EcNWT + STm versus only 0.02618% in EcNG66R + STm, representing an ~670-fold decrease. Differences in overall microbial composition between groups were more noticeable in intestinal tissue than in fecal samples, where family-level profiles remained broadly similar between EcNWT + STm and EcNG66R + STm animals across all time points (Supplementary Fig. 7A, B). In intestinal tissue, both groups were dominated by Muribaculaceae, Lachnospiraceae, and Lactobacillaceae, and the main difference between groups was the relative abundance of Enterobacteriaceae, which was significantly higher in the EcNWT + STm group, consistent with the pathogen colonization data above. In our experimental model, pretreatment with the EcNG66R mutant resulted in a significant decrease in Salmonella burden across most tested tissues compared with EcNWT pretreatment (Fig. 3G–I). In the spleen, bacterial levels were reduced by ~7300-fold (Fig. 3G)(p < 0.01), while in the liver fell 1700-fold (Fig. 3H)(p < 0.05). Reductions occurred in the MLN (350-fold; ns), gallbladder (60-fold; p < 0.05), and intestine (5800-fold; p < 0.05) (Fig. 3I). In the gut, bacterial loads decreased 13,000-fold (p < 0.05). Importantly, several mice in the EcNG66R-pretreated group showed a complete absence of Salmonella in multiple organs. In contrast, all mice in the EcNWT-pretreated group had detectable bacterial colonization in all examined tissues. qPCR analysis of infected spleens showed no significant differences in IL-6, IL-15, and IFN-γ expression between EcNWT + STm and EcNG66R + STm groups, while IL-10 was significantly higher in EcNWT + STm mice (p < 0.05) (Fig. 3J, Supplementary Fig. 5B). To understand the molecular basis for superior protection conferred by EcNG66R, including reduced organ colonization and prevention of splenomegaly, we performed RNA-seq analysis of spleens from BALB/c mice pretreated with either EcNWT or EcNG66R mutant, followed by STm infection. The transcriptomic analysis revealed differences between the two groups, with 695 differentially expressed genes (369 upregulated and 326 downregulated in EcNWT + STm relative to EcNG66R + STm) out of 27,466 total genes analyzed (Fig. 3K). Analysis of immune-related genes identified 1504 candidates, of which 223 were differentially expressed between treatment groups (Fig. 3L). The EcNG66R + STm group upregulated 125 immune genes, while EcNWT + STm upregulated 98, among which we identified a core neutrophil effector program—Ctsg (cathepsin G), Elane (neutrophil elastase), Mpo (myeloperoxidase), and Prtn3 (proteinase 3)—key components of neutrophil extracellular traps (NETs) and bacterial killing. In addition, EcNWT + STm mice displayed strong induction of the neutrophil-recruiting chemokines Cxcl2 and Cxcl3, potent chemoattractants and central mediators of acute inflammation. Additional inflammatory chemokines and cytokines, including Cxcl1, Cxcl11, Cxcl16, Ifng, IL-1b, and IL-10, showed increased expression trends (p < 0.05 but padj > 0.05). Upregulation was further observed for Selp (P-selectin), Itga2b (integrin α2b), Arg1 (arginase 1—typically associated with M2 macrophage polarization and immunoregulation), and Ms4a3. In contrast, the 125 immune genes upregulated in EcNG66R + STm mice revealed a different immunological strategy. EcNG66R + STm mice showed significant upregulation of CD8+ T-cell markers (Cd8a, Cd8b1) and the T-cell/NK-cell recruiting chemokine Xcl1 (lymphotactin) along with its receptor Xcr1. The response in EcNG66R + STm mice was dominated by extensive T-cell activation, characterized by increased expression of multiple T-cell receptor chains (Trbv1, Trbv13-1, Trbv13-2, Trbv14, Trbv19, Trbv20, Trgc2, Trgv1, Trbc1, Trbc2, Trac, Trav14d-3-dv8), co-receptors (Cd3d, Cd3e, Cd3g, Cd5, Cd7, Cd28), and transcriptional regulators (Bcl11b, Eomes, Lef1, Bach2, Themis). The upregulation trend of IL-15 (padj = 0.084, p < 0.05), a critical cytokine for CD8+ T-cell and NK cell survival and proliferation. B-cell responses were likewise enhanced, with elevated expression of Cd19, Cd22, Cd79a, Cd79b, Ms4a1 (CD20), Fcer2a (CD23), Cr2 (CD21), and several immunoglobulin genes (Ighv1-34, Ighv1-47, Ighv1-63, Ighv1-73, Ighv5-15, Ighv9-3), with Igkv2-112 maintaining its upregulation from the condition lacking STm. Increased expression of Spib, Ikzf3, and Bank1 further supports B-cell activation. Additionally, EcNG66R + STm mice showed a strong induction of interferon-stimulated genes (Mx1, Mx2, Ifi206, Ifi208, Ifit3, Ifit3b, Ifit1bl1, Oas1b), as well as chemokines Ccl21a (secondary lymphoid chemokine) and Ccl24 (eosinophil chemotactic protein). EcNWT and EcNG66R pretreatment elicits immune modulation and differential protection against Salmonella Typhimurium infection Flow cytometric analysis of spleen samples revealed distinct immunological profiles depending on the probiotic pretreatment strain administered prior to STm infection. The frequencies of Th cells were similar in both groups (Fig. 4A and B). However, frequencies of Th CD103+ cells within CD3+ and Th cells (p < 0.01) were lower in the EcNG66R + STm group compared to the EcNWT + STm group (Fig. 4D and E). Frequencies of Tc cells only within CD3+ T cells (p < 0.05) were higher in the EcNG66R + STm group (Fig. 4F and G), indicating a stronger cytotoxic T cell response following infection in mice pretreated with the EcNG66R. Similarly, the frequencies of Tc CD103+ cells only within CD3+ cells were higher in the EcNG66R + STm group compared to the EcNWT + STm group (Fig. 4I). There were no changes in the frequencies of the total pool of dendritic cells or resident dendritic cells in both groups (Fig. 4K–M). In contrast, mice pretreated with the EcNWT displayed significantly elevated frequencies of migratory dendritic cells within viable and CD3− cells (p < 0.01) (Fig. 4N–P).Fig. 4EcNG66R modifies the immune response in mice infected with Salmonella.The data shown the percentages of : T helper (Th, CD3+CD4+) cells A among viable cells, B among CD3+ cells; CD103+ Th cells (CD3+CD4+CD103+), C among viable cells, D among CD3+ cells and E among Th cells; T cytotoxic (Tc, CD3+CD8a+) cells F among viable cells and G among CD3+ cells; CD103+ Tc cells (CD3+CD8a+CD103+), H among viable cells, I among CD3+ cells, J among Tc cells, K dendritic cells (DCs, CD3−CD11c+) among viable cells; resident dendritic cells (rDCs, CD3−CD11c+CD8a+), L among viable cells and M among DCs; migratory dendritic cells (mDCs, CD3−CD11c+CD103+), N among viable cells, O among CD3− and P among DCs; Q F4/80+ macrophages (CD3−F4/80+) among viable cells; M1 macrophages (CD3−F4/80+CD86+), R among viable and S F4/80+ macrophages; M2 macrophages (CD3−F4/80+CD206+), T among viable cells and among, U F4/80+ macrophages, and V the specific fluorescence intensity of CD86 on the surface on F4/80+ macrophages in the spleen cells after Salmonella infection in mice pretreated with EcNWT (n = 7) and EcNG66R (n = 7). Values are presented as mean ± SEM. Statistical significance: for normally distributed data analyzed with an unpaired Student’s t-test, for non-normally distributed data analyzed by the Mann–Whitney U test, using GraphPad Prism version 11.0.0 for Windows, GraphPad Software, Boston, MA, USA. www.graphpad.com. EcNG66R modifies the immune response in mice infected with Salmonella. The data shown the percentages of : T helper (Th, CD3+CD4+) cells A among viable cells, B among CD3+ cells; CD103+ Th cells (CD3+CD4+CD103+), C among viable cells, D among CD3+ cells and E among Th cells; T cytotoxic (Tc, CD3+CD8a+) cells F among viable cells and G among CD3+ cells; CD103+ Tc cells (CD3+CD8a+CD103+), H among viable cells, I among CD3+ cells, J among Tc cells, K dendritic cells (DCs, CD3−CD11c+) among viable cells; resident dendritic cells (rDCs, CD3−CD11c+CD8a+), L among viable cells and M among DCs; migratory dendritic cells (mDCs, CD3−CD11c+CD103+), N among viable cells, O among CD3− and P among DCs; Q F4/80+ macrophages (CD3−F4/80+) among viable cells; M1 macrophages (CD3−F4/80+CD86+), R among viable and S F4/80+ macrophages; M2 macrophages (CD3−F4/80+CD206+), T among viable cells and among, U F4/80+ macrophages, and V the specific fluorescence intensity of CD86 on the surface on F4/80+ macrophages in the spleen cells after Salmonella infection in mice pretreated with EcNWT (n = 7) and EcNG66R (n = 7). Values are presented as mean ± SEM. Statistical significance: for normally distributed data analyzed with an unpaired Student’s t-test, for non-normally distributed data analyzed by the Mann–Whitney U test, using GraphPad Prism version 11.0.0 for Windows, GraphPad Software, Boston, MA, USA. www.graphpad.com. After the STm challenge, macrophage responses varied depending on probiotic exposure. In mice pretreated with EcNWT and then challenged (EcNWT+STm), the frequency of splenic F4/80+ macrophages among viable cells was higher compared to EcNG66R + STm group (p < 0.01)(Fig. 4Q). The M1 macrophages within viable (p < 0.05)(Fig. 4R) but not within F4/80+ cells were increased in the EcNWT + STm group (Fig. 4S), however the expression of CD86 activation marker (calculated as MFI) was higher in EcNG66R + STm group (p < 0.01)(Fig. 4V). The M2 macrophages among viable and F4/80+ cells were similar in both examined groups (Fig. 4T, U). Differential splenic immune cell profiles between EcNWT and EcNG66R-treated mice following probiotic treatment Given the significant differences in protection against Salmonella challenge observed between EcNWT and EcNG66R-pretreated mice, including prevention of splenomegaly, reduced pathogen loads, and improved clinical outcomes with the EcNG66R mutant, we sought to understand the baseline immunological differences established by these probiotic strains prior to infection. To do so, we evaluated the effects of sequential oral administration of EcN (WT and G66R) in Salmonella-susceptible BALB/c mice without pathogen challenge by comparing overall health, physiological features, and systemic immune profiles (Supplementary Fig. 6A). Oral administration of probiotic strains does not affect mice’s behavior, weight, or weight of internal organs, and does not cause any visible organ changes (Supplementary Fig. 6B). Bacterial enumeration from isolated systemic organs (spleen, liver, mesenteric lymph nodes) revealed no detectable CFU. Moreover, whole-body bioluminescence imaging revealed luminescent signals across animals, with no significant differences in intensity observed between EcNWT- and EcNG66R-treated animals, and signals remained confined to the gastrointestinal region with no evidence of systemic dissemination (data not shown). To directly evaluate the relative intestinal colonization fitness of the two strains, we performed a streptomycin pretreatment competitive index experiment (Supplementary Fig. 6C). EcNG66R was consistently recovered at higher levels than EcNWT from fecal samples at all examined time points (geometric mean CI across all time points = 4.37). The competitive advantage of EcNG66R was most prominent during the first half of the experiment (GM CI at D1 + 3h = 10.83, D2 = 8.81, D2 + 3h = 7.56, D3 = 8.06). Tissue-associated CI collected on day 5 was consistently above 1 across all examined intestinal compartments (gut CI = 2.03, intestine CI = 1.08, cecum CI = 1.85), confirming that EcNG66R colonization advantage extended to the mucosal level. (Supplementary Fig 6C). 16S rRNA analysis revealed comparable Enterobacteriaceae levels between the EcNWT and EcNG66R treatment groups across multiple time points (Supplementary Table 5C), indicating that the EcNG66R mutant does not alter baseline Enterobacteriaceae populations prior to pathogen challenge. Overall community composition was similar across treatment groups, with no major shifts in dominant families observed after administration of either probiotic strain (Supplementary Fig. 7A). Blood ferrum concentration was significantly decreased after EcNWT, but not after EcNG66R treatment, whereas blood glucose level was significantly decreased after treatment with both probiotic variants compared to untreated mice (Supplementary Fig. 6D). qPCR analysis of uninfected spleens showed no significant differences in IL-6, IL-10, IL-15, and IFN-γ expression between EcNWT and EcNG66R treatment groups (Supplementary Fig. 6E). Flow cytometric analysis of splenic leukocytes revealed distinct immune cell phenotype frequencies between EcNWT and EcNG66R probiotic-treated mice. The overall distribution of T cell populations differed significantly between the two groups (Fig. 5A–J). CD3+ lymphocytes were notably higher in EcNG66R-treated mice (p < 0.001, not shown), along with increased frequencies of Th cells within viable cells (p < 0.001)(Fig. 5A), indicating an increased potential for tolerogenic T-cell responses. Despite no significant differences in Th CD103+ cells within viable cells (Fig. 5C), there were elevated frequencies of Th CD103+ cells within CD3+ (p < 0.001) and Th cells (p < 0.01) populations in EcNWT in comparison to the EcNG66R group (Fig. 5D and E). Tc cells frequencies within viable and CD3+ cells were higher (p < 0.0001) in EcNG66R group (Fig. 5F and G). Furthermore, the percentages of CD103+ Tc cells among viable (p < 0.0001), CD3+ (p < 0.0001), and Tc cells (p < 0.01) were elevated in EcNG66R-treated animals (p < 0.0001), suggesting enhanced expansion or retention of cytotoxic effector subsets (Fig. 5H–J). In contrast, EcNWT-treated mice exhibited a significantly higher frequency of dendritic cells within viable cells (p < 0.0001), along with higher frequencies of resident dendritic cells (CD11c+CD8+) within viable cells but not within CD11c+ cells (Fig. 5K-M). Similarly, EcNWT-treated mice had significantly elevated migratory dendritic cells (CD11c+CD103+) within viable (p < 0.01), CD3− and CD11c+ cells (p < 0.01) (Fig. 5N–P). Together, these findings demonstrate that probiotic treatment caused different immune modulation in the spleen. EcNWT-treated mice mainly showed enhanced dendritic and regulatory T-cell activity, whereas EcNG66R-treated mice displayed an enrichment of effector T-cell subsets, especially cytotoxic T cells. Analysis of myeloid cells showed that EcNWT-treated mice also had higher levels of F4/80+ macrophages within viable cells (p < 0.05) (Fig. 5Q). Both M1 (pro-inflammatory) and M2 (anti-inflammatory) macrophage frequencies among viable cells (p < 0.0001 and p < 0.001, respectively) and F4/80+ cells (p < 0.0001 and p < 0.001, respectively) were present in higher frequencies in EcNWT-treated animals (p < 0.01) (Fig. 5R–U), suggesting increased macrophage recruitment. Interestingly, the CD86 MFI was again higher in the G66R group (p < 0.01) (Fig. 5V).Fig. 5EcNG66R boosts the immune response in mice.The data shown the percentages of: T helper (Th, CD3+CD4+) cells A among viable cells, B among CD3+ cells; CD103+ Th cells (CD3+CD4+CD103+), C among viable cells, D among CD3+ cells, E among Th cells; T cytotoxic (Tc, CD3+CD8a+) cells, F among viable cells, G among CD3+ cells; CD103+ Tc cells (CD3+CD8a+CD103+), H among viable cells, I among CD3+ cells, J among Tc cells; dendritic cells (DCs, CD3−CD11c+), K among viable cells; resident dendritic cells (rDCs, CD3−CD11c+CD8a+), L among viable cells, M among DCs; migratory dendritic cells (mDCs, CD3−CD11c+CD103+), N among viable cells, O among CD3−, P among DCs, Q F4/80+ macrophages (CD3−F4/80+) among viable cells; M1 macrophages (CD3−F4/80+CD86+), R among viable, S F4/80+ macrophages; M2 macrophages (CD3−F4/80+CD206+), T among viable cells, among U F4/80+ macrophages, V the specific fluorescence intensity of CD86 on the surface on F4/80+ macrophages in the mice spleens cells after treatment with EcNWT (n = 7) and EcNG66R (n = 7). Values are presented as mean ± SEM. Statistical significance: for normally distributed data analyzed with an unpaired Student’s t-test, for non-normally distributed data analyzed by the Mann–Whitney U test. Statistical analyses were performed using GraphPad Prism version 11.0.0 for Windows, GraphPad Software, Boston, MA, USA. www.graphpad.com. EcNG66R boosts the immune response in mice. The data shown the percentages of: T helper (Th, CD3+CD4+) cells A among viable cells, B among CD3+ cells; CD103+ Th cells (CD3+CD4+CD103+), C among viable cells, D among CD3+ cells, E among Th cells; T cytotoxic (Tc, CD3+CD8a+) cells, F among viable cells, G among CD3+ cells; CD103+ Tc cells (CD3+CD8a+CD103+), H among viable cells, I among CD3+ cells, J among Tc cells; dendritic cells (DCs, CD3−CD11c+), K among viable cells; resident dendritic cells (rDCs, CD3−CD11c+CD8a+), L among viable cells, M among DCs; migratory dendritic cells (mDCs, CD3−CD11c+CD103+), N among viable cells, O among CD3−, P among DCs, Q F4/80+ macrophages (CD3−F4/80+) among viable cells; M1 macrophages (CD3−F4/80+CD86+), R among viable, S F4/80+ macrophages; M2 macrophages (CD3−F4/80+CD206+), T among viable cells, among U F4/80+ macrophages, V the specific fluorescence intensity of CD86 on the surface on F4/80+ macrophages in the mice spleens cells after treatment with EcNWT (n = 7) and EcNG66R (n = 7). Values are presented as mean ± SEM. Statistical significance: for normally distributed data analyzed with an unpaired Student’s t-test, for non-normally distributed data analyzed by the Mann–Whitney U test. Statistical analyses were performed using GraphPad Prism version 11.0.0 for Windows, GraphPad Software, Boston, MA, USA. www.graphpad.com. To explain the systemic immunomodulatory differences between EcNWT and EcNG66R, we conducted a comparative RNA-seq analysis of spleens from BALB/c mice. DEG analysis showed minimal transcriptional divergence between the two strains, with only 31 genes meeting our significance criteria out of 27,604 total genes analyzed (Supplementary Table 7). Notably, this limited transcriptional divergence between EcNWT and EcNG66R aligns with similar minimal differences observed in our earlier in vitro MIEC cell line analysis. The EcNG66R mutant-induced upregulation of 21 genes and downregulation of 10 genes compared to wild-type EcN (Supplementary Table 7 and Supplementary Fig. 6F). The EcNG66R mutant increased the expression of several inflammatory mediators that were suppressed by EcNWT. These included Acod1 (aconitate decarboxylase 1), a metabolic enzyme essential for itaconate production and inflammatory macrophage polarization, as well as the acute-phase proteins Saa3 (serum amyloid A3) and Lcn2 (lipocalin 2). Additionally, EcNG66R treatment led to higher expression of Mmp8 (matrix metalloproteinase 8), a neutrophil-derived protease involved in tissue remodeling and inflammatory cell recruitment, and IL-1r2 (interleukin 1 receptor type 2), a decoy receptor that regulates IL-1 signaling. The increased expression of multiple immunoglobulin variable chain genes (Ighv1-11, Ighv14-4, Igkv2-112) and T-cell receptor genes (Trbv30, Trav3-1) in the EcNG66R group suggests enhanced adaptive immune activation compared to wild-type EcN. Analysis of immune-related genes revealed that only 11 genes showed significant differential expression between the two treatments, with nine upregulated and two downregulated in EcNG66R compared to EcNWT (Igkv4-62, Gm8893) (Supplementary Fig. 6G). Overall, our results show that EcNG66R serves as an immune primer and stimulates T-cell responses more effectively than EcNWT, resulting in better protection against Salmonella Typhimurium infection, as demonstrated by lower fecal shedding and decreased internal organ loads. Discussion The various mechanisms of probiotic action include adhesion, competition, or immune response modulation, and modern molecular biology techniques enable genetic modifications and the enhancement of specific features to offer even more comprehensive protection3. Escherichia coli Nissle 1917 (EcN) is one of the first human commensals identified for its role in protecting against enteric pathogen infections4, but the mechanisms and conditions behind its protective function remain unclear. In this study, we introduced three point mutations (G66R, T158P, and Y186R) into the fimH gene of the EcN to modify adhesive properties and examine their impact on probiotic function and protection against Salmonella Typhimurium (STm). All three mutations targeted specific regions of the FimH protein: G66R is located in the lectin domain near the mannose-binding pocket17,22, T158P is in the interdomain linker connecting the lectin and pilin domains19, and Y186R is found within the pilin domain18 (see Fig. 1A), and therefore may differentially affect the adhesin’s conformational flexibility, ligand-binding strength, and/or specificity. It has previously been shown that point mutations in the fimH can affect either the strength or the binding specificity to host cell structures17–19,23,24, not only in E. coli but also in various Salmonella serovars25–27, with diverse ligands identified, including β1 and α3 integrins, CD48, collagens, laminins, and fibronectin28–30. For example, the S62A substitution completely transforms the receptor repertoire, enabling bacterial penetration of tissue barriers and systemic dissemination24, while the E89K mutation redirects adhesion from yeast cells to uroepithelial cells23. Despite significant differences in binding levels, the mutant strain triggers only modest transcriptional changes in vitro, primarily associated with stress response, inflammatory regulation, and cellular trafficking. Most notably, EcNG66R reduced IL-6 expression compared with wild-type, suggesting that the mutant may engage alternative signaling cascades that promote tolerance rather than acute inflammation. Across multiple models, EcN consistently lowers IL-6 levels. Oral EcN reduces IL-6 and IFN-γ in acute intestinal inflammation-engineered31 EcN variants further suppress IL-6, TNF-α, and CXCL-1 in DSS colitis32, and curli-TFF3-producing EcN strains induce pronounced reductions of IL-6, IL-17A, and TNF-α in colonic tissue33. EcNG66R reduces STm adhesion more effectively than EcNWT, which aligns with the broadly described mechanism of probiotic-mediated competitive exclusion of epithelial binding sites34–37, decreasing pathogen load, preventing infection, and maintaining gut barrier integrity38–40. Beyond binding competition, Salmonella infections trigger pro-inflammatory responses (IL-6, IL-1β, IL-12) that drive IFN-γ production and macrophage activation41. In our data, both strains reduced IL-6 expression during infection compared with STm alone. This reflects strain-specific immunomodulatory strategies in which Salmonella typically triggers robust IL-6 and IL-8 production42. Differences induced by two strains align with findings that different probiotic strains employ mechanisms to suppress inflammation by reduction of IL-6, IL-8, and IL-1β, and secretion of regulatory cytokines IL-10 and IL-1943–45. These co-interaction experiments provided the first indication that EcNG66R not only competes more effectively for physical binding sites but also programs epithelial cells to respond differently to pathogenic challenge, acting as a dual-action probiotic. Previous studies have shown that receptor-binding specificity may affect the type and strength of immune responses triggered by FimH variants. For example, the T158P mutation in adherent-invasive E. coli enhances binding to CEACAM6 receptors on intestinal cells. This results in increased IL-1β levels and significant intestinal inflammation in Crohn’s disease models19. The G66R mutation in uropathogenic E. coli strains activates bladder epithelial cells through CD14 and Toll-like receptors, triggers apoptosis, and attracts additional immune cells17. However, at this point, we cannot tell whether the enhanced epithelial cell attachment and immune modulation are directly related or act in parallel. Salmonella Typhimurium is a major foodborne pathogen causing gastroenteritis and can progress to life-threatening systemic infections, particularly in vulnerable populations. BALB/c mice represent a naturally susceptible strain that develops systemic STm infection following oral administration46. In this model, bacteria breach the intestinal barrier and disseminate via Peyer’s patches through MLN to the spleen and liver47. We employed a high infectious dose to establish intestinal colonization and subsequent systemic progression, creating conditions to evaluate probiotic efficacy. In our model, in vitro adhesion of EcNG66R translates to enhanced intestinal colonization in vivo, as confirmed by the streptomycin pretreatment competitive index experiment—a model that depletes resident microbiota to specifically enrich for epithelial colonizers48. The progressive decline in recovery of both strains, coupled with increasing inter-individual variability, may reflect the progressive reassertion of colonization resistance as the microbiota recovers from streptomycin treatment, a phenomenon well-documented for E. coli strains competing for locally available nutrients in the intestinal mucus layer49,50. It was shown before that increased adherence to the intestinal cell surface does not necessarily translate into better protection against pathogens. For example, an EcN with the envZ^P41L mutation showed up to 10-fold higher colonization of the streptomycin-treated mouse intestine, but this did not improve its effectiveness at preventing colonization by the enterohemorrhagic E. coli strain EDL93350. Our results indicate that EcNG66R pre-treatment more effectively limits infection at the critical early stage of intestinal colonization (reducing shedding), thereby limiting systemic spread (lower enumeration of STm in systemic organs). The significant difference in mucosal colonization likely explains the subsequent decrease in systemic dissemination, as Salmonella must first establish intestinal colonization before reaching systemic sites by breaching the epithelial barrier or through M cell transcytosis51. Salmonella transmission depends critically on fecal shedding, as superspreaders drive pathogen spread. Shedding intensity is determined by bacterial virulence factors, host genetics, and microbiota composition, and antibiotic disruption can convert low-shedders into supershedders52. Host genetics significantly influence these dynamics—BALB/c mice (lacking protective SLC11A1) develop acute infections with high shedding, while 129 Sv mice exhibit chronic disease46,53. Probiotic interventions that effectively suppress fecal shedding—as achieved with the EcNG66R mutant—hold significant epidemiological value by interrupting transmission cycles in vulnerable populations. 16S rRNA sequencing data provide important ecological context for interpreting differences in protection between groups. Both probiotic strains maintained similar gut microbial communities with no major restructuring following administration, consistent with the established principle that a single introduced strain typically produces quantitative shifts in specific taxa rather than wholesale ecological reorganization54. Community differences were more apparent in intestinal tissue than in fecal samples, reflecting the partial correspondence between fecal and mucosa-associated microbiota55. The main distinction between infected groups was a marked relative expansion of Enterobacteriaceae in intestinal tissue of EcNWT + STm mice driven by Salmonella mucosal colonization, which was virtually absent in EcNG66R + STm animals—confirming that EcNG66R limits pathogen expansion within an otherwise ecologically comparable gut environment. Body weight loss constitutes one of the most sensitive and biologically meaningful proxies of systemic infection severity in murine salmonellosis, correlating with the extent of bacterial dissemination, cytokine-driven metabolic deterioration, and overall host physiological compromise56. In our model, EcNG66R-pretreated mice showed significantly less weight loss compared to the EcNWT-pretreated group, indicating a noticeably milder disease course despite identical infectious challenge. This distinction was further supported by the absence of splenomegaly in EcNG66R-pretreated animals—a characteristic of severe Salmonella infection in susceptible mouse strains1—alongside preserved liver and intestinal mass and lower bacterial burden across tissues. Together, these clinical and pathological parameters provide a coherent picture of reduced disease severity conferred by EcNG66R pretreatment. Blood metabolite analysis further highlights the physiological consequences of differential protection. Salmonella infection perturbs host glucose metabolism through increased glycolysis and pathogen-driven disruption of enteroendocrine signaling, driving systemic hypoglycemia57,58. Probiotics can help counteract this imbalance. In uninfected mice, both probiotic strains lowered blood glucose, whereas during Salmonella infection, EcNG66R-pretreated mice maintained significantly higher blood glucose compared to the EcNWT group. This indicates that EcNG66R effectively counteracts the infection-driven metabolic crash, preventing the dangerous hypoglycemia57,58. Moreover, EcNWT treatment significantly depleted blood iron in uninfected mice, potentially reflecting siderophore-mediated iron scavenging 9. EcNG66R did not deplete circulating iron, and this difference was no longer detectable during infection, indicating that systemic iron preservation is not the primary driver of EcNG66R superior protection. The significantly higher serum magnesium levels in EcNG66R + STm mice might also be relevant. At the immune level, adequate magnesium supports T helper, B cell, and macrophage responses to lymphokines59, whereas hypomagnesemia promotes a pro-inflammatory state via IL-6 upregulation and exacerbates systemic inflammation60—suggesting that its preservation may reinforce the expanded T cell populations observed in EcNG66R-pretreated mice while limiting excessive inflammatory damage. EcNG66R + STm mice mounted a lymphocyte-centered immune strategy characterized by CD8+ T cell expansion and dendritic cell recruitment, along with reduced splenic macrophage numbers, suggesting a shift toward cytotoxic immune responses that may enhance pathogen clearance. This outcome validates the concept that pathogen-specific adaptive immunity can be more effective than broad innate inflammatory responses in controlling intracellular bacteria such as Salmonella61. It was demonstrated that noncirculating tissue-resident memory T cells (both CD4+ and CD8+) in nonlymphoid tissues are vital components of immunity to Salmonella infection, with effective protection correlating specifically with expanded Salmonella-specific memory T cells that display markers of tissue residence, such as P2X7, ARTC2, LFA-1, and CD101 2. The expansion of CD103+ tissue-resident Tc cells by EcNG66R may serve as a mechanism for establishing long-lasting, tissue-specific immunity that cannot be achieved solely through circulating memory cells. CD11c+CD103+ dendritic cells are key in intestinal Salmonella infection: they are recruited from the lamina propria into the epithelium in a TLR- and chemokine-dependent manner and extend dendrites to capture luminal bacteria62. In Salmonella infection, CD103+ DCs help direct CD4+ T-cell differentiation toward immune regulation, including FoxP3+ Treg expansion63,64. Although early responses are dominated by flagellin-specific CD4+ T cells, CD103+ DCs tend to promote mucosal homeostasis62,64. During Salmonella infection, EcNG66R-treated mice maintained significantly lower total macrophage frequencies compared to EcNWT-treated animals, yet showed higher activation states, reflected in higher expression of CD86 (calculated as MFI). This qualitative shift—fewer total macrophages but enhanced activation—may suggest more efficient immune control via adaptive mechanisms, reducing the need for massive macrophage recruitment while ensuring that the present macrophages are functionally optimized. Moreover, observed expansion of F4/80+ macrophages in the EcNWT pretreated group explains the pathological splenomegaly characterized by macrophage accumulation in white pulp and hemophagocytic activity described in severe Salmonella infections1. EcNWT-triggered response may be focused on neutrophil pathway activation, as RNA-seq results show this myeloid-dominated response, with upregulation of the core neutrophil effector program: Ctsg (cathepsin G), Elane (neutrophil elastase), Mpo (myeloperoxidase), and Prtn3 (proteinase 3)—all essential components of neutrophil extracellular traps (NETs) and bacterial killing machinery65. The upregulation of Cxcl2, Cxcl3, Selp (P-selectin), Itga2b (integrin α2b), and multiple other factors involved in neutrophil recruitment and tissue infiltration also supports this hypothesis. In contrast, EcNG66R drives extensive upregulation of diverse T-cell receptor chains (Trav, Trbv genes), co-receptors (Cd3d, Cd3e, Cd3g, Cd5, Cd7, Cd28), and coordinated expression of T-cell transcription factors (Bcl11b, Eomes, Lef1, Bach2, Themis), indicating not merely quantitative T-cell expansion but qualitative programming of effector functions. The selective upregulation of the XCL1–XCR1 chemokine axis marks a significant shift in how the immune response is directed. XCR1 is exclusively expressed on CD8α+ dendritic cells in mice, a specialized subset that excels at cross-presenting exogenous antigens via MHC class I molecules to prime CD8+ T cell responses66,67. This cross-presentation pathway is essential for initiating cytotoxic T lymphocyte responses against intracellular bacteria like Salmonella that do not directly infect antigen-presenting cells, but rather remain sequestered within phagosomes where antigens are typically restricted to MHC class II presentation. This enhanced cross-presentation may overcome a key virulence strategy of Salmonella: during infection with virulent S. Typhimurium, CD8+ T cell priming is considerably delayed (peaking around day 21 rather than day 7), with the phagosomal lifestyle allowing escape from host CD8+ T cell recognition and conferring a survival advantage to the pathogen68. The early immunological programming established by EcNG66R—with expanded XCR1+ dendritic cells and pre-positioned tissue-resident CD8+ T cells—likely accelerates this otherwise delayed priming process, eliminating the temporal window that Salmonella typically exploits for establishing chronic infection. By contrast, the neutrophil-recruiting CXC chemokines (Cxcl2, Cxcl3) upregulated in EcNWT + STm mice lack this precision, recruiting cells with potent but non-specific antimicrobial mechanisms that inflict collateral damage while failing to establish durable adaptive immunity. The critical role of adaptive immunity in controlling even beneficial commensal bacteria is underscored by studies showing that E. coli Nissle can translocate and disseminate systemically in mice lacking both microbiota and adaptive immunity (germfree Rag1−/− mice, 100% mortality), while reconstitution with CD4+ T cells prior to challenge prevents both translocation and mortality69. These findings reveal two potentially distinct immunological strategies. EcNWT pretreatment prompts a quick, primarily innate immune response marked by neutrophil recruitment and myeloid effector activation—a frontline defense that is strong but temporary and nonspecific. In contrast, EcNG66R redirects the immune response toward adaptive mechanisms, in which early inflammatory signals activate and expand cytotoxic T cells, preparing the host for a more targeted and lasting response upon encountering a pathogen. The enhanced protection against Salmonella systemic dissemination observed in EcNG66R-treated mice prompted us to investigate the differences among probiotic strains in the mouse model. In both groups treated with probiotics, we did not observe any changes in mouse behavior, activity patterns, or organ pathology throughout the experimental period, which is a critical safety feature for probiotic strains3. In the splenic compartment under non-infected conditions, EcNG66R induced a slightly more inflammatory environment than EcNWT, characterized by upregulation of Acod170 and associated metabolic reprogramming genes. This mild inflammatory state, alongside antimicrobial effectors like Lcn29 (lipocalin 2, which sequesters bacterial siderophores) and immunomodulatory molecules (Saa3, IL-1r2), suggests that EcNG66R establishes a state of metabolic “readiness” or immune priming that enables rapid, effective antimicrobial responses upon pathogenic challenge71,72. Infections with Salmonella often lead to a decline in both CD4+ and CD8+ T lymphocyte populations in spleen and systemic lymphoid tissues, impairing adaptive immune responses and enabling bacterial dissemination. By contrast, exposure or immunization with attenuated Salmonella frequently elicits robust expansion of antigen-specific CD4+ and CD8+ T cells and boosts host immune control1. Remarkably, mice treated with EcNG66R showed significantly higher frequencies of CD3+ T cells, with increases observed in both CD4+ helper T cell and CD8+ cytotoxic T cell subsets. Notably, wild-type EcN has been previously shown to increase T helper and T cytotoxic cell frequencies in both systemic and intestinal tissues73. It seems that the G66R mutation may enhance adaptive immunity, which in turn limits Salmonella colonization and systemic spread—adding another protective layer to the reduced adhesion and competitive exclusion. This suggests that EcNG66R may function as an immunological primer—similar to a live attenuated vaccine—establishing CD8+ T cell surveillance that enables rapid pathogen recognition and elimination upon subsequent Salmonella challenge, explaining the enhanced protection22,73. The expansion of CD8+ T cell populations by EcNG66R is particularly significant, given that CD8+ T cells are critical for long-term immunity to Salmonella, providing both immediate cytotoxic effector functions and durable immunological memory that neutrophil responses cannot achieve. As demonstrated by vaccine studies, Salmonella vectors elicit robust antigen-specific CD8+ T cell responses with elevated IFN-γ production22,74. Moreover, CD8+ T cells have been shown in specific MHC backgrounds to accelerate Salmonella clearance, sometimes even more potently than CD4+ cells, highlighting that both T-cell arms contribute significantly to pathogen control51. The ability of EcNG66R to expand both CD4+ and CD8+ T-cell compartments prior to pathogen exposure may equip the host with adaptive effector functions, including early IFN-γ-dependent cytotoxicity and CD4+-mediated sterilizing immunity, which are known to be required to limit subsequent infection51,74,75. Beyond the T cell compartment, differences in the population of antigen-presenting cells were observed between EcNG66R- and EcNWT-treated mice. EcNG66R treatment resulted in significantly lower total frequencies of F4/80+ macrophages in the spleen. Salmonella exploits macrophages as an intracellular replication niche76, where the pathogen survives within modified phagosomes called Salmonella-containing vacuoles (SCVs), using type III secretion systems to prevent lysosomal fusion and reprogram host cell metabolism77. Salmonella preferentially replicates within hemophagocytic macrophages, which exhibit an anti-inflammatory M2 phenotype that creates a permissive environment for bacterial survival78. During severe Salmonella infection, massive macrophage expansion and hemophagocytosis characterize pathological splenomegaly, with F4/80+ macrophage infiltration into white pulp contributing to tissue damage and providing bacterial sanctuaries for persistent infection1. The absence of this macrophage expansion in EcNG66R-treated mice—and the corresponding lack of splenomegaly observed in this group—may indicate the protective effect of the modified probiotic. This phenotype synergizes with the enhanced CD4+ and CD8+ T cell responses induced by EcNG66R, suggesting a shift toward adaptive, rather than macrophage-dominated immunity. Collectively, we demonstrated that the G66R mutation in FimH of E. coli Nissle 1917 enhances the probiotic strain’s ability to protect the host against Salmonella infection, likely through a dual mechanism of competitive exclusion and immune priming. This leads to limited Salmonella intestinal colonization, lower fecal shedding, decreased systemic dissemination, and reduced infection pathology. However, to fully translate these observations into broader contexts, several limitations and avenues for future research must be addressed. While the protective phenotype of EcNG66R is robust, certain mechanistic links remain correlative rather than causal: although we observed a significant expansion of T-cells, this study does not provide direct functional evidence of their absolute requirement for protection. We cannot yet formally decouple the specific contribution of these subsets from potential CD4+ help or innate immunity. The specific receptor recognized by the G66R variant has not yet been confirmed or identified. While our in vitro data support the hypothesis of expanded receptor recognition, the biochemical identity of these targets remains elusive. Currently, part of our mechanistic insights relies on RNA-Seq transcriptomics, with no additional proteomic or functional verification across all physiological compartments. Methods Bacteria, plasmids, and growth conditions Bacterial strains and plasmids used in this study are listed in Tables 1 and 2;Table 2Plasmids used in this studyPlasmidCharacteristicReferencepEMGSuicide plasmid; KmR80pEMG-G66RSuicide plasmid; KmR; with insert carrying G66R point mutation in fimHThis studypEMG-T158PSuicide plasmid; KmR; with insert carrying T158P point mutation in fimHThis studypEMG-Y186RSuicide plasmid; KmR; with insert carrying Y186R point mutation in fimHThis studypSW-2Plasmid for m-toluate-inducible expression of the I-SceI enzyme; GmR80pFPV-mCherryPlasmid contains the gene for the mCherry protein AmpR85p16SluxTemperature-sensitive p16Slux plasmid (containing the luxABCDE operon of Photorhabdus luminescens), EryR82 Table 2Plasmids used in this studyPlasmidCharacteristicReferencepEMGSuicide plasmid; KmR80pEMG-G66RSuicide plasmid; KmR; with insert carrying G66R point mutation in fimHThis studypEMG-T158PSuicide plasmid; KmR; with insert carrying T158P point mutation in fimHThis studypEMG-Y186RSuicide plasmid; KmR; with insert carrying Y186R point mutation in fimHThis studypSW-2Plasmid for m-toluate-inducible expression of the I-SceI enzyme; GmR80pFPV-mCherryPlasmid contains the gene for the mCherry protein AmpR85p16SluxTemperature-sensitive p16Slux plasmid (containing the luxABCDE operon of Photorhabdus luminescens), EryR82 Plasmids used in this study Escherichia coli Nissle 1917 (EcN), its mutants, and Salmonella Typhimurium SL1344 (STm) strains were grown in Luria-Bertani broth (LB). Unless otherwise specified, bacterial cultures were typically grown at 37 °C for 16 h in static conditions. Salmonella strains with the mCherry plasmid were grown in LB medium supplemented with 100 µg/mL of ampicillin. For animal experiments, STm was grown in LB medium supplemented with 50 µg/mL of streptomycin. Before all the experiments, STm, EcN, and its mutants were passaged three times to induce T1F production79. Mutants generation A two-step recombineering method was used for precise genomic edits in EcN, as described earlier, with minor modifications80,81 (Supplementary Fig. 1). PCR was performed with Q5 High-Fidelity DNA Polymerase (New England Biolabs M0491S) for all cloning procedures according to the manufacturer’s instructions. To construct the pEMG-G66R, pEMG-T158P, pEMG-Y186R suicide plasmids (Table 2), the flanking DNA fragments surrounding the target chromosomal regions were amplified via PCR and inserted into the pEMG suicide vector between I-SceI restriction sites. The pEMG derivative was mobilized from E. coli S17-1λpir into EcN through conjugation. Transconjugants were selected on M9-glucose minimal medium supplemented with kanamycin (50 μg/ml). Positive EcN::pEMG transconjugant colonies were transformed with the pSW-2 plasmid and stimulated on LB agar containing gentamicin (20 μg/ml) and m-toluate (1 mM), which induced I-SceI expression. Transformants that had lost the integrated pEMG plasmid were identified based on their kanamycin-sensitive phenotype. Candidate clones were screened by replica-plating on LB agar with and without kanamycin (50 μg/mL), and PCR-verified sensitive colonies with appropriate primers. Sanger sequencing confirmed the presence of the targeted point mutations in the fimH gene. Generation of bioluminescent EcN strains Bioluminescent EcN strains were generated using the p16Slux system as described by Riedel et al. 82. Plasmid p16Slux, containing the luxABCDE operon, was isolated from E. coli SURE cells and electroporated into erythromycin-susceptible recipient strains. Transformants were selected on LB agar with erythromycin (500 μg/ml) at 30 °C and confirmed by plasmid isolation and gel electrophoresis. Chromosomal integration of the lux operon was induced by cultivating transformants at 42 °C for 24 h under erythromycin selection. Integration was verified by colony PCR using primers 16S_rev_XhoI (CTGATCTCGAGGGCGGTGTGTACAAGG)82 and 16S_fwd_int (ATTAGCTAGTAGGTGGGGTAACGGCTCACCTAGG)82, which target the expected 1150 bp fragment, resulting in a 1150 bp amplicon. The bioluminescence of positive integrants was measured using a SPARK Plate Reader (TECAN, Switzerland) with serial dilutions of overnight cultures, and strains with comparable luminescence levels were selected. Generation of nalidixic acid-resistant strains Nalidixic acid-resistant mutants of EcNWT and EcNG66R were created through direct selection on solid medium. Overnight cultures of each strain were plated (50–100 µl) onto LB agar supplemented with nalidixic acid at increasing concentrations (10, 25, 50, 75, 100, and 200 µg/ml) and incubated for 20 h at 37 °C. Colonies that appeared on plates containing 25–75 µg/ml nalidixic acid were re-streaked onto LB agar with 50 µg/ml nalidixic acid to verify the stability of the resistance phenotype. Cell cultures and preparation of the co-cultures Porcine intestinal epithelial cell line IPEC-J2 was cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Biowest, France) supplemented with 10% fetal bovine serum (FBS; Biowest), 2 mM glutamine, and antibiotics (penicillin, streptomycin; Biowest). The mouse intestinal epithelial cell line (MIEC) was grown in muINTEPI Medium (InSCREENeX, Germany). Both at 37 °C with 5% CO2. All cell lines were either passaged twice a week or had the medium replaced every other day. For all experiments, epithelial cells were seeded in 24-well or 6-well plates at a density of 2 × 105 cells or 1 × 106 per well the day before the assay and used when monolayers were established. Probiotic adhesion and competitive exclusion assay Adhesion assays were prepared as previously described83 with minor modifications. Cell monolayers were prepared as described above, washed with PBS, and 400 or 800 μl of DMEM without FBS or antibiotics was added to each well. For probiotic adhesion and MOI-dependent adhesion assays: EcNWT and EcN mutants were adjusted to MOI 10, 100 and 1000 and added to the appropriate wells. DMEM alone served as a control. The plates were incubated for 2 h at 37 °C and processed as described below. For competitive exclusion assays, after the initial incubation, cells were washed with PBS to remove non-attached probiotic bacteria. Subsequently, 400 μL of DMEM and 100 μL of STm suspension (MOI 100) were added, followed by a further 2-h incubation at 37 °C. Afterwards, cells were washed three times with PBS and prepared for further analysis. For adhesion quantification, PBS was removed from each well, and 200 μL of sterile 1% PBST was added. The plates were incubated for 10 min at room temperature on a rocking shaker. Then, 800 μL of PBS was added to each well, and the cells were evenly suspended throughout the entire volume. Serial dilutions were prepared in 96-well plates and plated on LB agar plates with or without 100 μg/mL ampicillin, followed by overnight incubation at 37 °C. Ampicillin selection was used to distinguish STm from EcN in competitive exclusion assays. Colony counts were performed, and CFU/mL were calculated for each sample and strain. Total RNA isolation and qPCR Total RNA was extracted from the MIEC cell line after co-culture with EcNWT and its mutants using TRIzol™ Reagent (Invitrogen™, Thermo Fisher Scientific, USA) following the manufacturer’s protocol. The concentration and purity (A260/A280 ratio) were assessed using a NanoDrop 2000c (Thermo Fisher Scientific). Total RNA (2 µg per reaction) was reverse transcribed into complementary DNA (cDNA) using the High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific), following the manufacturer’s instructions. The resulting cDNA served as a template for quantitative PCR (qPCR), performed on a QuantStudio™ 5 Real-Time PCR System (Applied Biosystems, USA) using EvaGreen dye (Biotium, USA), amplification conditions as follows: initial denaturation at 95 °C for 3 min, followed by 35 cycles of denaturation at 95 °C for 15 s, annealing at 56 °C for 25 s, and extension at 72 °C for 30 s. A melt curve analysis was performed over a temperature range of 60–95 °C, with appropriate step settings, to verify product specificity. All reactions were conducted in technical triplicate, with at least three biological replicates per condition. Relative gene expression levels were calculated using the comparative ΔΔCt method and expressed as relative quantification (RQ) values. Gene expression was normalized to the Actb housekeeping gene and calibrated to untreated MIEC cells or spleen tissue from uninfected mice. Primer sequences are provided in Supplementary Table 2. Total RNA-sequencing Total RNA was extracted as described above. RNA integrity and potential contamination were evaluated with the Multina system (microchip electrophoresis system for DNA/RNA analysis MCE™-202 MultiNA)(SHIMADZU, Japan). RNA samples from three independent biological replicates per condition (both cell culture and tissue samples) were pooled and subsequently submitted to Novogene for transcriptome sequencing. Upon receipt, all RNA samples underwent additional quality control assessments by Novogene to confirm RNA integrity and purity prior to library preparation and sequencing. RNA sequencing data analysis RNA-seq data were analyzed in R (2025.05.1) in RStudio using the packages readxl, openxlsx, dplyr, ggplot2, stringr, tidyr, pheatmap, RColorBrewer, tibble and ggrepel—custom functions for filtering, categorization, and identification of differential genes (DEGs). Comprehensive summaries were generated to quantify the number of expressed genes, significantly regulated genes, and genes meeting combined TPM, fold-change, and padj criteria. Raw reads were converted to TPM, and genes with TPM > 1 in both samples were retained for further analysis. Unique genes were defined as those with high expression in one sample (>5 TPM) and low expression in the other (<2 TPM). DEGs were identified at |log₂FC| thresholds > 1 and padj < 0.05, classifying them as up- or down-regulated. Genes that did not meet the thresholds were considered nonessential. Immune-related genes were identified by matching DEGs with GO annotations and KEGG pathways (29 files), using keywords and GO terms related to the regulation and activation of the immune response. For better characterization of expression changes, a difference score (range of changes × max |log₂FC|) was calculated for each gene, including only genes with TPM > 1 and padj < 0.05. The 50 genes with the highest scores were selected and plotted in a scatter plot with log₂FC = 0 and ±1 threshold lines, ordered by mean absolute expression change. In vivo experiments The animal experimental procedures were approved by the Local Ethics Committee for Animal Experimentation in Wrocław (Resolution No. 060/2022/P1 dated January 18, 2023). Female BALB/c mice (5–6 weeks old) were obtained from Anlab (Prague, Czech Republic). Throughout the experiment, animals were maintained under controlled environmental conditions. The mice underwent a 5-day acclimation period followed by gentle handling sessions prior to experimental procedures. Mice were randomly assigned to four different groups: EcNWT + PBS, EcNWT + STm, EcNG66R + PBS, and EcNG66R + STm. All four groups received the appropriate probiotic bacterial suspension (EcNWT or EcNG66R) at 10⁸ CFU in 0.1 ml PBS via intragastric gavage using a stainless steel ball-tipped gavage needle. This treatment was repeated daily for 5 consecutive days at 24-h intervals. To monitor the colonization and persistence of the probiotic, fecal samples were collected on days 1 and 5 of probiotic administration, homogenized in PBS, serially diluted, and plated on LB, LB with streptomycin, and MacConkey Agar to count viable bacteria. On the 6th day, feed and water were withheld for 2 h for all animals before oral administration of a single infectious dose of STm of 10⁸ CFU in 0.1 mL PBS via intragastric gavage to the EcNWT + STm, EcNG66R + STm groups and 0.1 mL of PBS to EcNWT + PBS, EcNG66R + PBS groups. Animals were monitored daily for clinical signs, including coat condition, locomotor activity, posture, and changes in feeding and drinking behavior. Body weights were recorded daily throughout the experiment, and fecal samples were collected for further analysis. After infection, mice were monitored using the Lago X bioluminescence imaging system (Spectral Instruments Imaging, USA) with Aura Imaging Software v.4.0 to track luminescent bacterial signals. For whole-body imaging, mice were anesthetized with 5% isoflurane and maintained under anesthesia during image acquisition. Mice were euthanized 6 days after STm infection, and their spleens, livers, lymph nodes (MLN), gallbladders, small and large intestines, were collected and weighed. Blood was collected at necropsy for further analysis. Tissue samples were homogenized in 1 ml of PBS, diluted, and plated on LB supplemented with streptomycin to determine bacterial loads. Additionally, spleen tissues were immediately placed in RNAlater solution (Thermo Fisher Scientific) and stored at −80 °C until further processing. Collected feces were homogenized and plated daily after infection as described above. To directly compare the intestinal colonization of EcNWT and EcNG66R, antibiotic-marked strains were used: EcNWT and EcNG66R carrying an erythromycin resistance marker (EcNWT-EryR, EcNG66R-EryR) and EcNWT and EcNG66R carrying nalidixic acid resistance (EcNWT-NalR, EcNG66R-NalR). Mice received a single oral dose of streptomycin (20 mg per mouse in 0.1 ml PBS) via intragastric gavage 24 h before starting probiotic treatment. Mice were randomly divided into two groups (4 mice each) and given a mixed probiotic suspension daily for 5 days, as described above: Group 1 received EcNWT-EryR and EcNG66R-NalR; Group 2 received EcNG66R-EryR and EcNWT-NalR (5 × 10⁸ CFU of each strain; total of 10⁹ CFU per mouse in 0.1 ml PBS). Fecal samples were collected twice daily—at 3 and 24 h post-administration. On day 5, at the study endpoint, mice were euthanized, and the small intestine and cecum were excised and processed as described above. Fecal and tissue homogenates were cultured on LB agar supplemented with erythromycin (100 µg/ml) or nalidixic acid (100 µg/ml) for differential enumeration. The competitive index (CI) was calculated for each mouse at each time point using the formula: CI = (EcNG66R_out/EcNWT_out)/(EcNG66R_in/EcNWT_in), where CFU recovered from fecal samples is divided by CFU in the administered inoculum. A CI > 1 indicates a colonization advantage for EcNG66R; a CI < 1 indicates the opposite. Samples with zero CFU for either strain were excluded, as well as one extreme outlier (CI > 500). CI values from both groups were combined for analysis. The geometric mean CI was calculated at each time point and across all time points. Splenocyte isolation Spleens harvested from BALB/c mice and mechanically disrupted using sterile syringe plungers through 40 μm cell strainers (BD Falcon, USA) in 2 ml of red blood cell lysis solution (0.83% NH₄Cl), centrifuged (300×g, 10 min, 4 °C), and washed twice with a sorting buffer (PBS, 2% FBS, and 2 mM EDTA). Cells were cryopreserved in RPMI-1640 medium (Biowest) supplemented with L-glutamine, penicillin/streptomycin, HEPES, and 10% FBS, with 50% FBS and 10% DMSO at −80 °C. Flow cytometry Cells were rapidly thawed at 37 °C with gradual addition of warm RPMI-1640 medium, centrifuged (300×g, 5 min, RT), washed with fresh medium (200×g, 10 min, RT), and PBS (300×g, 5 min, RT). Cell concentration was adjusted to 1 × 10⁶ cells per 100 μl PBS and stained for 15 min, RT with Zombie Green (BioLegend, USA) to assess viability and blocked for 10 min, RT for non-specific binding with Mouse FcX (BioLegend). Then the cells were incubated for 20 min, 4 °C, with antibodies listed in Supplementary Table 3 and appropriate isotype controls. After two washes with FACS buffer (PBS with 0.5 mM EDTA, 0.002% sodium azide, 1% FBS), cells were fixed with 1% fixation buffer (Biolegend) (20 min, 4 °C). Following two additional washes, samples were analyzed by flow cytometry on the BD FACSLyric™ Clinical Flow Cytometry System (Becton Dickinson), acquiring at least 200,000 events per sample at a flow rate of 600-80 events. Cytometer Setup and Tracking beads (CS&T Research Beads, Becton Dickinson) and compensation beads (BD™ FC Beads 7-Color Kit) were used for automated quality assurance, compensation adjustments, and control of machine performance. Data were analyzed using FlowJo™ software version 10.8.1 (Becton Dickinson), acquiring at least 200,000 events per sample. The percentage of T helper (CD3+CD4+), T cytotoxic (CD3+CD8+), M1 macrophages (CD3−F4/80+CD86+), M2 macrophages (CD3−F480+CD206+), resident (CD3−CD11c+CD8+), and migratory (CD3−CD11c+CD103+) dendritic cells (CD3−CD11c+) was calculated as the total number of viable cells, CD3+ cells, or the parent population during gating. The relative levels of CD86 in F4/80+ macrophages were calculated as follows: median fluorescence of cells stained with anti-CD86 antibody minus median fluorescence of cells stained with isotype-matched control (Supplementary Table 3). The gating strategy is presented in Supplementary Fig. 2A and B. 16s rRNA gene sequencing For 16S rRNA gene sequencing, samples were collected across various experimental groups and time points to evaluate bacterial community composition during probiotic colonization and after Salmonella infection. For the probiotic colonization study (uninfected mice, fecal samples were collected at three time points: day 5 of probiotic administration (D5), day 6 from the first administration, and one day after stopping probiotic treatment (D6), and day 10 from the first administration (D10) from both EcNWT and EcNG66R treatment groups. For the Salmonella challenge experiment, fecal samples were collected at day 5 of probiotic administration before pathogen challenge (D5), day 1 pi (D1), and day 4 pi from both EcNWT + STm and EcNG66R + STm groups (D4). Additionally, intestinal tissue samples were collected during necropsy at day 6 post-infection from infected mice (EcNWT + STm and EcNG66R + STm) and at the equivalent time point from uninfected probiotic-treated control mice (EcNWT and EcNG66R). For each experimental group and time point, individual samples from mice within the same treatment group were pooled and sent to GenXone S.A. for DNA extraction, 16S rRNA gene library preparation, and sequencing. Mutation visualization Predicted 3D structures of the Escherichia coli Nissle 1917 FimH protein (wild-type and single-point mutants G66R, T158P, and Y186R) were generated using ColabFold (implementation of AlphaFold2). The obtained PDB models were visualized, annotated, and rendered in UCSF ChimeraX84. Mutated residues were highlighted and labeled, and domain coloring was applied to distinguish the receptor-binding and pilin domains. High-resolution images were rendered using the ray-tracing option in ChimeraX for figure preparation. Statistical analysis Statistical analyses were performed using GraphPad Prism version 11.0.0 (GraphPad Software, San Diego, CA, USA). Specific statistical tests used for each experiment are indicated in the figure legends. P-values < 0.05 were considered statistically significant and are indicated as follows: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Supplementary information Supplementary Information Supplementary Information