Suppression of gut colonization by multidrug-resistant Escherichia coli clinical isolates through cooperative niche exclusion

Abstract

Human gut colonization by multi-drug resistant Enterobacterales (MDR-E) poses a risk for subsequent infections. Because of the collateral damage antibiotics cause to the microbiota, microbiome-based interventions aimed at promoting decolonization have garnered interest. In this study, we evaluate the strain-specific potential of 430 commensal Escherichia coli isolates to inhibit the growth of an MDR E. coli strain. Comparative analyses using in vitro, ex vivo, and mouse models reveal that only a subset of commensal strains can facilitate gut decolonization. Bioinformatic and experimental analyses of the antagonism among representative strains demonstrate that both direct and indirect carbohydrate competition contribute to niche exclusion between E. coli strains. Finally, the combination of a protective E. coli strain with a Klebsiella oxytoca strain enhances the inhibitory potential against metabolically diverse MDR E. coli strains and additional MDR-E species, highlighting that rationally designed metabolically complementary approaches can contribute to developing next-generation probiotics with broad-spectrum activity.

Fig. 1. Commensal E. coli strains show different competitive effects.

A Schematic overview of experimental workflow. Commensal and MDR E. coli strains are spiked into isolated cecal contents of GF animals in a 10:1 ratio. After 24 h of anaerobic co-cultivation, CFUs of MDR E. coli were quantified by plating on selective agar plates. Created with Biorender. B Fold change of co-cultures to control of all strains (n = 430). Strains marked in red were selected for further in vivo experiments. C Fraction of competitive and non-competitive isolates in each phylogroup. D The phylogenetic relationship of commensal E. coli isolates is visualized in a taxonomic tree. Different rings (from outer to inner ring) show the presence of colibactin, virulence genes, AMR genes, phylogroup, competitive phenotype, and log-fold reduction.

Fig. 2. Specific commensal E. coli strains enable decolonization of MDR E. coli in a preventive and therapeutic manner.

A Workflow of in vivo experiments. For the prophylactic model (B, C, G, H), SPF mice were treated with ampicillin in their drinking water three days before oral gavage with commensal E. coli/ PBS. After four days, mice were orally gavaged with 10⁸ CFUs E. coli MDR1 (MHH). For the therapeutic model (D–F), colonization order was reversed. Fecal samples were taken to monitor fecal colonization levels and 16S rRNA gene sequencing at days 1, 3, 6, 9, 14, 21, 28, and 42. Created with Biorender. B Resulting fecal colonization levels of E. coli MDR1 after different time points of colonization in the prophylactic model. Data represents the geometric mean and SEM of two to three independent experiments with n = 9–17 mice per group. C Clearance kinetics of E. coli MDR1 (clearance = CFU/g below the detection limit in feces). P-values represent the Log-rank (Mantel-Cox) test with *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. D Resulting fecal colonization levels of E. coli MDR1 in the therapeutic model. Geometric mean and SEM of two independent experiments with n = 8 mice per group. P-values represent the Log-rank (Mantel-Cox) test with *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. E Clearance kinetics of E. coli MDR1 after different time points of colonization (clearance = CFU/g below the detection limit in feces). P-values represent the Log-rank (Mantel-Cox) test with *p < 0.05. F Resulting fecal burden of E. coli MR102 after different time points of colonization. CFU/g could be identified by selective plating on MacConkey base supplemented with 10 g/L D-Maltose. Mean and SEM of two independent experiments with n = 8 (PBS) and n = 9 (MR102) mice per group. G Resulting fecal burden of E. coli MDR2 (NRZ 21236, ST131) after different time points of colonization in the prophylactic model. Two independent experiments with n = 7–9 mice per group. P-values indicate a nonparametric Kruskal-Wallis test *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. H Clearance kinetics of E. coli MDR2 after different time points of colonization (clearance = CFU/g below the detection limit in feces). P-values represent the Log-rank (Mantel-Cox) test with *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Fig. 3. The protective effect is dependent on the microbial context.

A β-diversity of fecal samples of mice treated with PBS, MR102, MR158, MR103, MK192 or LK192 at day-7 or 6 was analyzed using the Bray-Curtis dissimilarity matrix and NMDS. B β-diversity of fecal samples of mice treated with PBS or MR102, was analyzed from day −7 to day nine using the Bray-Curtis dissimilarity matrix and NMDS. C, D α-diversity represented by observed OTUs or Shannon diversity. P-values indicate a nonparametric Kruskal-Wallis test *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. E β-diversity using pairwise wUnifrac distance after antibiotic treatment. P values indicate a nonparametric Kruskal-Wallis test *p < 0.05. Mean and SD with n = 8 (PBS) and n = 10 (MR102) mice per group (F) β-diversity of fecal samples of slow and rapid clearance mice at day nine was analyzed using the Bray-Curtis dissimilarity matrix and NMDS. G α-diversity on day nine of slow and rapid clearance phenotype mice represented by observed OTUs and Shannon index n = 16 rapid-clearer and n = 10 slow-clearer in box with whiskers (min to max). H Analysis of differentially abundant bacterial species in mice that showed rapid clearance vs. slow clearance phenotype by LefSe.

Fig. 4. Distinct carbohydrate utilization is involved in intra-species competition.

A The heatmap shows an AUC of 72 h of growth curves of E. coli strains in MM9 supplemented with 5 g/L of single carbon sources (n = 39). Results of three independent experiments performed in triplicates. B Commensal E. coli MR102 colonization levels at different time points. Mean and SEM of two independent experiments with n = 4 (PBS) and n = 8 (MR102 and MR102+Cellobiose) mice per group. P-values indicate a nonparametric Kruskal-Wallis test p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. C Clearance kinetics of E. coli MDR1 after different time points of colonization (clearance = CFU/g below the detection limit in feces). P-values represent the Log-rank (Mantel-Cox) test with 0.05. D CFU/g of E. coli MDR1 of single mice at different time points of colonization. Geometric mean and SEM of two independent experiments with n = 4 (PBS) and n = 7 (MR102) and n = 8 (MR102+Cellobiose) mice per group. mice per group. P-values indicate a nonparametric Kruskal-Wallis test p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Fig. 5. Protective E. coli strain is superior to MDR E. coli in direct competition for specific carbohydrates.

A Results of competition assay in Biolog® Phenotypic microarray competitive index (CFU/ml E. coli MDR1/ CFU/ml E. coli MR102) and OD₆₀₀ of the respective strains of all carbon sources that both strains could utilize. One dot represents data from three independent experiments. B Results of competition assay in Biolog® Phenotypic microarray competitive index (E. coli MDR1/ E. coli MR102) of all carbon sources that could be utilized by at least one of the strains. One dot represents data from three independent experiments. C Competition assay in MM9 supplemented with 5 g/L of the respective carbon source under aerobic and anaerobic conditions. P values indicate a nonparametric Kruskal-Wallis test *p < 0.05. Mean and SD with one dot representing data from one experiment performed in duplicates or triplicates. D Resulting fecal burden of E. coli MDR1 after different time points of colonization. Geometric mean and SD of two independent experiments with n = 8 (PBS), n = 9 (WT) and n = 10 (ΔmanA and manA::manA) mice per group. E Commensal E. coli MR102 colonization levels at different time points. Geometric mean and SD of two independent experiments withof two independent experiments with n = 8 (PBS), n = 9 (WT) and n = 10 (ΔmanA and manA::manA). F Clearance kinetics of E. coli MDR1 after different time points of colonization (clearance = CFU/g below the detection limit in feces). P-values represent the Log-rank (Mantel-Cox) test with 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Fig. 6. A Combination of metabolically diverse Enterobacteriaceae can enlarge the target spectrum of MDR-E.

A Heatmap summarizing the results of an ex vivo competition assay of five commensal E. coli strains, K. oxytoca, and combinations of E. coli and K. oxytoca against a panel of 18 MDR E. coli strains and one EPEC strain. Fold reduction of CFU/ ml in co-cultures to MDR E. coli control is classified as non-protective (100-fold reduction). B Venn diagrams displaying carbon source overlap of E. coli MDR3 (NRZ 55652, ST167), K. oxytoca MK01, and E. coli MR102 or EcN. C Resulting fecal burden of E. coli MR102 and K. oxytoca MK01 in co-colonization after different time points of colonization. Geometric mean and SD of one experiment with n = 5 (co-colonization MR102 and MK01, n = 4 (single-colonization MR102 and n = 3 (single-colonization MK01) mice per group. D–G Resulting fecal burden of MDR-E after different time points of colonization. Geometric mean and SD of two independent experiments with n = 9 (D), n = 8 (PBS, MK01, MR102), n = 10 (mix) (E), n = 8 (PBS, MK01), n = 9 (MR102, mix) (F), n = 9 (PBS, MK01), n = 8 (MR102) and n = 10 (mix) (G) mice per group.