C57BL/6 J mice were maintained in the facilities of the “Forschunginstitute fur Experimentelle Medizin” (FEM, Charite – Universitätsmedizin, Berlin, Germany), under SPF conditions. Age and sex matched mice aged between 10 and 12 weeks were used.

To eradicate the commensal gut microbiota, mice were transferred to sterile cages and were subjected to an ampicillin plus sulbactam antibiotic regimen (2 g/L plus 1 g/L, respectively; purchased from Dr. Friedrich Ebert Arzneimittel, Ursensollen, Germany) as a drinking solution for a period of 8 weeks. These mice were kept and handled under strict aseptic conditions. Microbiota depletion was confirmed using cultural and molecular analyses of fecal samples, as described earlier (Heimesaat et al., 2022).

Fresh fecal samples were collected from five individual healthy human volunteers and murine animals, respectively, after screening to ensure their exclusion of enteropathogenic bacteria, parasites, and viruses. Fecal samples were pooled separately for humans and mice and dissolved in an equal volume of sterile phosphate-buffered saline (PBS, Thermo Fischer Scientific, Waltham, MA, United States). Pooled samples were aliquoted and stored at −80°C until further use. Prior to gavage experiments, aliquots of the pooled fecal samples were thawed, and bacterial communities were analyzed using both culture-dependent and culture-independent methods with 16S rRNA gene sequencing. Quantification of total bacterial load and identification of individual bacterial groups were also done using cultural and culture-independent analyses. Starting a week prior to infection (i.e., on days −7, −6, and − 5), mice were gavaged with 300 μL of the respective fecal suspension as described in details earlier (Bereswill et al., 2011; Shayya et al., 2023), to study the effects of bacterial communities on host physiology and housed under SPF conditions with ad libitum access to food and water.

Cultural analysis, biochemical identification, and molecular detection of luminal bacterial communities from stomach, duodenum, ileum, and colon were performed as previously described (Weschka et al., 2021; Shayya et al., 2023). In brief, DNA extracts and plasmids were quantified using Quat-iT PicoGreen reagent (Invitrogen, Paisley, UK) and adjusted to a concentration of 1 ng per μL. Abundance of the major bacterial groups within the gut microbiota was assessed using quantitative real time polymerase chain reaction (qRT-PCR) with group-specific primers targeting the 16S rRNA gene (Tib MolBiol, Berlin, Germany). The number of 16S rRNA gene copies per μg of DNA in each sample was determined, and the frequencies of the respective bacterial groups were computed relative to the eubacterial (V3) amplicon.

For infection experiments, we utilized the C. jejuni strain 81-176, which was derived from frozen stocks and inoculated on karmali agar plates (Oxoid, Wesel, Germany). The bacteria were grown under microaerophilic conditions at 37°C in a closed container containing gas packs (CampyGen, Oxoid, Wesel, Germany) for 48 h to obtain optimal growth conditions. To establish an infection model, we infected SPF, SAB, mma, and hma mice with 10⁹ viable C. jejuni bacterial cells via oral gavage in a total volume of 300 μL of PBS on two consecutive days (days 0 and 1).

Mice were sacrificed by CO₂ asphyxiation on day 21 post infection. Luminal samples from the gastrointestinal tract (stomach, duodenum, ileum, colon) of each mouse were harvested for a set of post-experimental analyses, including cultural and molecular analyses. Respective samples were homogenized in sterile PBS (Thermo Fisher Scientific, Waltham, MA, United States) and serial dilutions were plated onto karmali agar (Oxoid, Wesel, Germany) and incubated under microaerophilic conditions for 48 h at 37°C as described earlier in details (Heimesaat et al., 2022).

Targeted metabolomic analysis of fecal samples from d0 was performed by Biocrates Lifesciences AG (Innsbruck, Austria) using their McP Quant 500 platform. Fecal samples were collected from each subject and immediately frozen at −80°C until further analysis. Samples were shipped on dry ice to Biocrates for metabolomic analysis. Biocrates’ MxP^® Quant 500 product uses mass spectrometry (MS) to quantify over 500 metabolites in a single run, including amino acids, biogenic amines, organic acids, and lipids. The quality of the samples was assessed using internal standards, and data were preprocessed to remove contaminants, drift, and other systematic errors. Lipids and hexoses were analyzed using flow injection analysis-tandem mass spectrometry (FIA-MS/MS) on a SCIEX 5500 QTRAP^® instrument (AB SCIEX, Darmstadt, Germany) with an electrospray ionization (ESI) source. Small molecules were analyzed using liquid chromatography–tandem mass spectrometry (LC–MS/MS) on a Xevo^® TQ-XS instrument (Waters, Milford, MA, United States). To prepare the samples for analysis, a 96-well based sample preparation device was used, which contained inserts impregnated with internal standards. A predefined amount of the sample was added to the inserts, followed by the addition of a phenyl isothiocyanate (PITC) solution to derivatize some of the analytes (e.g., amino acids). After the derivatization was completed, the target analytes were extracted using an organic solvent, followed by a dilution step. The extracted metabolites were then analyzed by FIA-MS/MS and LC–MS/MS using multiple reaction monitoring (MRM) to detect the analytes. Concentrations of the metabolites were calculated using an appropriate mass spectrometry software (Sciex Analyst^® and Waters MassLynx^™) and imported into Biocrates’ MetIDQ^™ software for further analysis. The metabolite concentrations were then determined in each sample by normalizing to the weight and dilution factor from the original metabolome extraction protocol, then provided in “μM,” which were subsequently exported as an excel file. Individual metabolites were grouped by classes and visualized using GraphPad Prism (V9; San Diego, CA, United States).

The minimum inhibitory concentration (MIC) values for deoxycholic acid (Sigma-Aldrich, St. Louis, Missouri, United States; 30,960) were determined using the broth microdilution method in BactoTM Brain Heart Infusion (BHI) (BD Biosciences, Heidelberg, Germany) broth supplemented with 5% fetal bovine serum (FBS) (Biochrom, Berlin, Germany). The bacterial inoculum of C. jejuni 81-176 was prepared from overnight growth on karmali agar plates (Oxoid, Wesel, Germany) and suspended in 5 mL of BHI broth supplemented with 5% FBS to an optical density at 600 nm (OD600) of 0.1. Two-fold serial dilutions of the test compounds were prepared in 96-well plates in the range of 1–2,048 mg/L to a final volume of 200 μL per well, and 10 μL of the bacterial suspension was added to the respective wells. The plates were incubated for 48 h under microaerophilic conditions at 37°C, and then the OD was measured at 600 nm using a microplate reader (Infinite M Flex, Tecan, Switzerland). The MIC was defined as the lowest concentration of the compound that completely inhibited bacterial growth.

We conducted statistical analyses to evaluate the significance of observed differences between groups. Prior to analysis, we first assessed the normality of the data using the Anderson-Darling test. Medians and significance levels were then calculated using GraphPad Prism (V9; San Diego, CA, United States). For pairwise comparisons of the groups, we employed the Student’s t-test for normally distributed data and the Mann–Whitney test for non-normally distributed data. Moreover, to account for the possibility of errors resulting from multiple comparisons, we utilized the one-sided ANOVA with Tukey’s post correction for normally distributed data, and the Kruskal-Wallis test with Dunn’s post correction for non-normally distributed data. A two-sided probability (p) value of ≤0.05 was considered statistically significant.

The animal studies were conducted in adherence to the European animal welfare guidelines (2010/64/EU) after receiving authorization from the local commission for animal experiments (“Landesamt für Gesundheit und Soziales.” LaGeSo, Berlin; under registration number G0172/16). Clinical conditions of mice after infection were surveyed daily.

To establish CR against C. jejuni in SAB mice, and to confirm the role of the specific murine or human gut microbiota therein, we introduced human or murine gut microbiota to SAB mice by FMT via gavage. One week after reconstitution, these mice were perorally infected with 10⁹ viable C. jejuni on day (d) 0 and d1 along with conventional SPF mice and SAB mice as controls with and without CR, respectively. This setting enabled us to analyze CR by assessing the kinetics of pathogen shedding in fecal samples and in distinct gastrointestinal compartments.

Cultural analysis of fecal samples from SPF mice exhibited effective clearance of the pathogen within a few days of infection, indicative for a strong CR against C. jejuni given a complete absence of colonization (Figure 1A). In contrast, CR was abrogated in SAB mice which revealed a robust and persistent colonization of the colon by C. jejuni, with pathogen densities ranging from 10⁸ to 10⁹ colony-forming units per gram (CFU/g) during the experimental period (Figure 1B). Conversely, SAB mice reconstituted with murine gut microbiota mma showed a protective CR with sporadic colonization occurring in only a few individuals (Figure 1C). On the other hand, hma mice displayed absence of CR as indicated by susceptibility to C. jejuni intestinal colonization, maintaining high median pathogen loads between 10⁷ and 10⁹ CFU/g of feces throughout the experiment (Figure 1D).

Furthermore, we investigated pathogen numbers along the gastrointestinal tract in the distinct mouse cohorts. As anticipated, CR against C. jejuni observed in SPF and mma mice extended throughout the gut, as demonstrated by significantly higher C. jejuni loads in the stomach, duodenum, ileum, and colon of SAB mice compared to SPF and mma mice, and in hma mice compared to mma mice (p < 0.01–0.001) (Figures 2A–D).

Hence, these results demonstrate that the two mice groups - with and without CR each – were excellently suited to investigate the metabolite patterns in the intestinal milieu associated with presence and absence of murine CR against C. jejuni.

To further ensure the quality of our mice models with and without CR, we quantified the major microbial groups in the gut microbiota by qRT-PCR based on 16S rRNA analysis before infection. Our objective was to elucidate the differences in microbial communities between mice harboring human or murine gut microbiota known from earlier studies (Bereswill et al., 2011; Mrazek et al., 2019) to ensure that the different microbiota transplants had successfully engrafted in the mouse intestines. Our analysis revealed that hma mice harbored higher total eubacterial loads compared to SPF mice (p < 0.05), while no significant differences in the gene numbers were observed between SPF and mma, and between hma and mma mice (Figure 3A). Moreover, significantly higher enterobacterial loads were observed in hma mice (~10³ gene numbers/ng DNA) prior to infection compared to SPF and mma mice (~10 gene numbers/ng DNA) (p < 0.001) (Figure 3B). Differences in enterococci fecal loads were evident, with hma mice exhibiting significantly higher loads (~10² gene numbers/ng DNA) than both murine models (~10⁻² gene numbers/ng DNA) (p < 0.001) (Figure 3C). Similarly, in line with the distinctive species composition of the gut microbiota, hma mice harbored significantly lower fecal loads of lactobacilli (~10⁻² gene numbers/ng DNA) compared to the resistant models, SPF and mma mice (~10⁵ gene numbers/ng DNA) (p < 0.01–0.001) (Figure 3D). Interestingly, mma mice displayed higher bifidobacterial loads (~10⁴ gene numbers/ng DNA) (p < 0.01–0.001) compared to both SPF and hma mice (~10³ and 10⁴ gene numbers/ng DNA, respectively) (Figure 3E). Furthermore, the levels of Bacteroides/Prevotella species were significantly elevated in hma mice (~10⁶ gene numbers/ng DNA) compared to SPF and mma mice (~10⁴ gene numbers/ng DNA) (p < 0.001) (Figure 3F). Similarly, hma mice also exhibited significantly higher loads (~10⁵ gene numbers/ng DNA) of the Clostridium coccoides group than SPF and mma mice (~10⁴ gene numbers/ng DNA) (p < 0.001) (Figure 3G). On the other hand, both hma and mma mice displayed higher fecal loads (~10⁵ gene numbers/ng DNA) of the Clostridium leptum group compared to SPF mice (~10⁴ gene numbers/ng DNA) (p < 0.01–0.001) (Figure 3H). Lastly, SPF and mma mice exhibited significantly higher loads (~10⁶ gene numbers/ng DNA) (p < 0.001) of Mouse Intestinal Bacteroides compared to hma mice, where these bacteria were nearly absent (Figure 3I). Hence, murine CR against C. jejuni is associated with higher loads of lactobacilli and Mouse Intestinal Bacteroides, but with lower loads of enterobacteria, enterococci, Bacteroides/Prevotella species, and Clostridium coccoides group. The fact that these results were comparable to observations in earlier studies confirm the successful establishment of the murine and human microbiota in the intestines of mice following oral FMT, and that CR was associated with a specific murine microbiota composition. Finally, species-specific differences in the microbiota compositions associated with presence and absence of CR are further used as a basis for the interpretation of results obtained from metabolomic analyses.

In consideration of C. jejuni relying on amino acids as a primary nutrient and energy source within the gut (Wright et al., 2009), we conducted a comprehensive analysis of the amino acid concentrations in fecal samples of mice with and without CR prior to infection. The resulting metabolomics assessment revealed notable variations in the composition of amino acids among the different cohorts (Figure 4). As could be expected due to the absence of intestinal bacteria, the SAB mice exhibited significantly higher total concentrations of free amino acids compared to the other mice groups (p < 0.05–0.001) (Supplementary Figure S1). Concerning individual amino acids, we focused first on the amino acids essential for C. jejuni growth. Most importantly, levels of cysteine which is required for C. jejuni growth in the absence of serine, were significantly reduced in SPF mice and in mma as compared to hma and to SAB mice (p < 0.05–0.001) (Figure 4H). Notably, this was the only amino acid that was significantly reduced in mma mice in comparison to hma mice. Furthermore, the abundance of other amino acids essential for fueling C. jejuni metabolism including serine, glutamine, asparagine, threonine, and proline, was markedly lower in SPF, mma, and hma mice compared to SAB mice (p < 0.01–0.001) (Figures 4A–E). On the other hand, aspartic acid concentrations were lower in SPF and mma mice compared to SAB mice (p < 0.05), and glutamic acid levels were lower only in mma mice compared to SAB (p < 0.05), nevertheless a trend toward lower concentrations of both amino acids in mma mice compared to hma mice was observed (Figures 4F,G). Similarly, this pattern extended to arginine with significantly lower concentrations in SPF mice compared to SAB mice and in mma mice compared to hma and to SAB mice (p < 0.05–0.001) (Supplementary Figure S2A).

Moreover, SPF mice exhibited substantially lower concentrations of other amino acids, including leucine, isoleucine, lysine, histidine, phenylalanine, tryptophan, tyrosine, and valine, as compared to SAB mice, with a trend observed toward lower concentrations in mma mice compared to hma mice (p < 0.05–0.001) (Supplementary Figures S2B–I). Noteworthy, the levels of these amino acids were also significantly lower in both hma and mma mice compared to SAB mice (p < 0.05–0.001) (Supplementary Figures S2B–I). Additionally, glycine concentrations were significantly lower in SPF and mma mice compared to SAB mice, with a trend observed toward lower levels in mma mice compared to hma mice (p < 0.05–0.01) (Supplementary Figure S2J). Similarly, methionine levels were significantly lower in SPF mice compared to SAB mice, and a trend toward lower concentrations in mma compared to hma mice was observed (p < 0.05) (Supplementary Figure S2K). On the other hand, no significant differences were detected for alanine levels between the different mouse cohorts, yet a trend toward lower concentrations in SPF mice compared to SAB mice and in mma mice compared to hma mice was observed (Supplementary Figure S2L). Furthermore, the amino acid-related metabolites sarcosine and phenylalanine betaine were also evaluated and revealed a distinct pattern. Sarcosine levels were significantly higher in SPF and mma mice compared to SAB and hma mice, respectively (p < 0.001 and p < 0.05, respectively) (Supplementary Figure S3A). Similarly, phenylalanine betaine levels were significantly elevated in SPF mice compared to SAB mice (p < 0.001), while mma mice exhibited higher levels compared to hma mice (p < 0.001) (Supplementary Figure S3B). Of note, mma mice also exhibited significantly lower levels of these metabolites compared to SAB mice (p < 0.01–0.001) (Supplementary Figures S3A,B). Hence, comparisons of metabolomic signatures among our mice cohorts revealed that murine CR against C. jejuni is significantly associated with a shortage in amino acids creating an unfavorable environment for C. jejuni.

A comprehensive targeted metabolomic analysis of the bile acid composition in fecal samples of mice on d0 prior to C. jejuni 81-176 infection revealed significant differences between the distinct mouse cohorts. The mice with CR, namely SPF and mma mice, exhibited a bile acid profile characterized by a predominant secondary bile acid pool compared to the mice without CR, the SAB and hma mice, respectively. This was evident by a significantly lower ratio of primary to total bile acids in SPF mice compared to SAB mice, and in mma mice compared to hma and SAB mice (p < 0.001) (Figure 5A), as well as the significantly higher ratio of secondary to total bile acids in SPF mice compared to SAB mice, and in mma mice compared to hma and SAB mice (p < 0.01–0.001) (Figure 5B). Additionally, the ratio of secondary bile acids to primary bile acids, an indicator of secondary bile acid synthesis, was found to be significantly upregulated in the SPF mice compared to SAB mice (p < 0.001), and in mma mice compared to hma and to SAB mice (p < 0.01 and p < 0.001, respectively) (Figure 5C).

It is of note that the secondary bile acid deoxycholic acid, which is highly toxic to C. jejuni was significantly associated with CR in the mice groups. It was elevated in SPF mice compared to SAB mice, as well as in mma mice compared to hma and SAB mice (p < 0.05–0.001) (Figure 6E).

Moreover, it was reported that tauroconjugation of cholic acid is associated with a higher bacterial 7-α-dehydroxylation into deoxycholic acid (Eldere et al., 1996). Taurine-conjugated bile acids were predominant in the bile acid pool of SAB mice and hma mice, leaving SPF mice with significantly lower levels compared to SAB mice, and mma mice with lower levels compared to hma and SAB mice (p < 0.05–0.001) (Figure 6A). In particular, significantly lower concentrations of taurocholic acid and tauromuricholic acids were observed in SPF and mma mice compared to SAB mice, and in mma mice compared to hma mice for taurocholic acid (p < 0.01–0.001) (Figures 6B,C), with a trend toward lower concentrations for tauromuricholic acids (Figure 6C). Remarkably, the primary bile acid cholic acid exhibited a distinct pattern, with SPF mice harboring significantly higher concentrations compared to SAB mice, and mma mice harboring significantly lower concentrations than hma mice (p < 0.05–0.01) (Figure 6D). Moreover, the cholic acid concentrations were significantly lower in SAB mice compared to mma and hma mice (p < 0.05 and p < 0.001, respectively) (Figure 6D).

Hence, the results from bile acid profiling highlight distinct metabolomic signatures associated with CR against C. jejuni which is characterized by a predominance of secondary bile acids.

In light of the pivotal antimicrobial roles of fatty acids and their potential impact on the composition of the gut microbiota, we conducted an in-depth investigation into the fatty acid profiles present in the colonic milieu of our distinct mouse models on day 0 before infection. Through comprehensive metabolomic analysis, we unraveled distinct patterns of fatty acid concentrations that distinguished between mouse groups exhibiting CR or susceptibility to C. jejuni colonization. Remarkably, our findings revealed that SPF mice exhibited significantly higher concentrations of free fatty acids in the colon compared to SAB mice (p < 0.001, Figure 7A). A similar trend was observed for mma mice, demonstrating significantly elevated levels of free fatty acids compared to hma and to SAB mice (p < 0.001, Figure 7A). Moreover, when examining the sum of monounsaturated and polyunsaturated fatty acids, SPF mice displayed significantly higher concentrations of both types of fatty acids compared to SAB mice (p < 0.001), while mma mice had significantly higher concentrations compared to hma and SAB mice (p < 0.001) (Figures 7B,C). Notably, monounsaturated fatty acid concentrations were found to be significantly higher (p < 0.001) in hma mice compared to SAB mice (Figure 7B).

Further delving into the differentially expressed fatty acids, particularly within the monounsaturated fatty acid class, we observed significant differences in octadecenoic acid and eicosenoic acid. Octadecenoic acid levels were significantly elevated in SPF and in mma mice compared to SAB and hma mice, respectively (p < 0.001) (Figure 8A). Additionally, hma mice displayed higher concentrations of octadecenoic acid compared to SAB mice (p < 0.001) (Figure 8A). A similar pattern emerged for eicosenoic acid, with SPF and mma mice exhibiting substantially elevated levels of this fatty acid compared to SAB and hma mice, respectively (p < 0.01–0.001) (Figure 8B), whereas no statistically significant difference was observed between the two susceptible groups (Figure 8B).

Interestingly, the distinctive patterns observed between resistant and susceptible groups extended to specific polyunsaturated fatty acids, although not all fatty acids followed these patterns. Octadecadienoate, eicosadienoic acid, and eicosatrienoic acid exhibited different patterns, with SPF mice harboring higher fecal concentrations of these fatty acids compared to susceptible models SAB mice, and mma mice harboring higher concentrations compared to hma and SAB mice (p < 0.05–0.001) (Figures 8C–E). On the other hand, although SPF and mma mice exhibited significantly higher fecal concentrations of arachidonic acid compared to SAB mice (p < 0.001 and p < 0.05, respectively), a trend toward higher concentrations in mma mice compared to hma mice was still observed (Figure 8F). Furthermore, SPF, mma, and hma mice displayed higher colonic concentrations of docosahexaenoic acid and eicosapentaenoic acid compared to SAB mice (p < 0.05–0.001) (Figures 8G,H). Nevertheless, no significant differences were observed in colonic levels of the latter fatty acids between mma and hma mice (Figures 8G,H). Hence, these results indicate that elevated free fatty acids represent another metabolomic signature significantly associated with murine CR against C. jejuni.

In addition to investigating the role of bile acids, amino acids, and fatty acids, we explored the metabolomic landscape of other classes of metabolites present in the colon of our mouse models before infection. Specifically, we delved into the analysis of phosphatidylcholines and diglycerides, which hold potential relevance to host-microbe interactions, in addition to the crucial components of nucleic acids, namely purines. Our investigation into purine metabolites revealed distinct profiles that distinguished between mice cohorts with and without CR (Figure 9A). Mma mice exhibited significantly higher levels of purine metabolites compared to hma mice (p < 0.001) (Figure 9A). Interestingly, although no significant differences were observed between SAB and the resistant cohorts, a trend was still observed, and SAB mice displayed higher concentrations of purines compared to hma mice (p < 0.05) (Figure 9A). In particular, xanthine concentrations in SPF, mma, and SAB mice were significantly higher compared to hma mice (p < 0.01–0.001) (Supplementary Figure S4A). On the other hand, hypoxanthine displayed a different pattern, as SPF mice harbored significantly higher levels compared to SAB mice, and this was also the case for mma mice compared to hma and SAB mice (p < 0.01–0.001) (Supplementary Figure S4B). Notably, SAB mice exhibited a higher xanthine synthesis computed by the ratio of xanthine to hypoxanthine compared to both SPF and mma mice (p < 0.001) (Supplementary Figure S4C).

Moreover, these distinct patterns extended to phosphatidylcholines, diglycerides, and triglycerides. SPF and mma mice exhibited significantly higher concentrations of all three classes of the metabolites compared to SAB mice (p < 0.05–0.001) (Figures 9B–D). Nevertheless, compared to hma mice, only diglycerides were measured at significantly higher concentrations in mma mice (p < 0.05), albeit a trend toward higher levels of phosphatidylcholines and triglycerides in mma mice compared to hma mice could still be observed (Figures 9B–D).