Gut microbiota-derived acetate mitigates Escherichia fergusonii-induced gut dysbiosis exacerbating Klebsiella pneumoniae pulmonary infection in mice

Abstract

Escherichia fergusonii is an emerging food-borne opportunistic pathogen that disrupts gut homeostasis and exacerbates the severity of other organ diseases. However, the effects of gut colonization by E. fergusonii on the gut-lung axis remain unclear. Here we show that the abundance of Escherichia increases in microbiota-depleted mice and is enriches in patients with K. pneumoniae-associated pneumonia. Both alterations show a statistically significant correlation with the severity of lung infection. In the murine model, gut colonization by E. fergusonii induces specific gut dysbiosis, decreases intestinal and pulmonary acetate levels, and exacerbates carbapenem-resistant K. pneumoniae-induced pneumonia. Notably, fecal microbiota transplantation and acetate administration restore immune responses, ameliorate gut dysbiosis, increase acetate concentrations in intestinal and pulmonary tissues, and improve disease outcomes in mice with pneumonia. This study demonstrates that E. fergusonii-induced gut dysbiosis modifies the gut-lung immune axis in resistant K. pneumoniae-induced pneumonia and proposes a potential intervention strategy involving microbiota-derived acetate.

Murine gut colonization modeling of E. fergusonii sheds light on this emerging opportunistic pathogen in disrupting the gut microbiota, reducing acetate levels, and exacerbating subsequent CRKPassociated pneumonia through the gut–lung immune axis.

Introduction

Klebsiella pneumoniae is a common opportunistic pathogen causing both nosocomial and community-acquired pneumonia worldwide¹. The emergence of carbapenem-resistant K. pneumoniae (CRKP) presents substantial challenges to public health because of limited therapeutic options and high mortality rates². While antibiotics remain essential for managing lung infections, their overuse accelerates the development of drug resistance and disrupts the balance and diversity of the gut microbiota in patients, thereby complicating infection management efforts^3,4.

Accumulating evidence has established that intestinal microbes confer protective effects against pulmonary infectious diseases. Specifically, these symbiotic microorganisms alleviate lung injury induced by bacterial pathogens, as well as that caused by influenza virus infection^5–7. By modulating the gut microbiota composition and adjusting the levels of intestinal short-chain fatty acids (SCFAs), the activity of immune cells can be altered, leading to remodeling of the lung immune environment. This mechanism reduces pulmonary inflammatory responses and injury⁸. In contrast, gut dysbiosis can alter the lung metabolome and influence immune responses to pulmonary infections^9,10. The immunological connection between the gut and lung, termed the “gut-lung axis”, enables bacterial metabolite translocation across the gastrointestinal tract barrier into the bloodstream, thereby modulating pulmonary immune responses¹¹. The gut microbiota is a crucial component of the host’s defense mechanism against pathogens. Dysbiosis within the gut ecosystem facilitates the invasion and colonization by pathogenic organisms, thereby increasing the susceptibility to opportunistic infections. Several murine models of bacterial lung infection indicate that gut colonization by carbapenemase-producing Enterobacterales exacerbates lung immune responses by modulating the gut-lung axis^12,13.

Tang et al.¹⁴ compared the gut microbiota composition between control mice and those with K. pneumoniae-induced pulmonary infection mice. Escherichia was the predominant genus in the pulmonary infection group, suggesting the potential involvement of Escherichia in the progression of K. pneumoniae-induced pulmonary infection. Previous meta-analyzes and animal experiment studies have shown that the significant presence of intestinal Escherichia may serve as a potential diagnostic microbial biomarker of lung infections and correlates positively with the inflammatory factor IL-1 level in lung tissues¹⁵.

Escherichia fergusonii is the closest phylogenetic relative of Escherichia coli within the genus Escherichia and is recognized as an emerging opportunistic pathogen in both humans and animals^16,17. E. fergusonii is frequently found in food products and food animals¹⁸, indicating potential risks to food safety and public health. Notably, Jung et al.¹⁹ reported that gut colonization with E. fergusonii enhanced the infectivity of GII.12 human norovirus in gnotobiotic pigs. These characteristics suggest that E. fergusonii may significantly impact the health of humans and animals by disrupting gut homeostasis and enhancing the virulence of other pathogens. However, the mechanisms by which E. fergusonii modulates the microbiome and metabolome along the gut-lung axis require further investigation.

In the present study, we observed a significant increase in the proportion of Escherichia in gut microbiota-depleted mice compared to that in the control group. We also observed elevated E. fergusonii levels in fecal samples from patients with K. pneumoniae-associated pneumonia compared to those in healthy controls, and this increase was correlated with disease severity. Therefore, based on the principle of the gut-lung axis and combined with the findings of the current study, we designed this investigation to examine how gut colonization by E. fergusonii exacerbates pulmonary CRKP infection.

Results

Impact of alterations in gut microbiota composition on the host defence against CRKP-induced lung infection

To investigate the role of the intestinal microbiota during Klebsiella-induced pneumonia, we administered a combination of broad-spectrum antibiotics (ampicillin, neomycin, metronidazole, and vancomycin) to wild-type mice to deplete the gut microbiota, as shown in Fig. 1A. H&E staining and immunostainings for Ki67 (proliferation restricted to the crypt), MUC2 (goblet cell differentiation), and Chromogranin A (neuroendocrine cell differentiation) revealed neither infection symptoms nor histological abnormalities in intestinal sections from untreated and microbiota-depleted mice. These results indicate that, while broad-spectrum antibiotic treatment did not affect intestinal structure and epithelial integrity (Fig. 1B, C), it caused extensive elimination of commensal bacteria from the gut environment (Fig. 1D–H). As shown in Fig. 1D, E, antibiotic-treated mice exhibited reduced α-diversity indices (Shannon and Simpson) and altered bacterial community structures compared to untreated controls (p = 0.001). Based on the indicator analysis at the phylum and genus levels, we identified specific taxonomic shifts induced by antibiotic treatment. At the phylum level, Proteobacteria abundance significantly increased, while Bacteroidetes and Firmicutes abundance decreased in the microbiota-depleted group (Fig. 1F). At the genus level, Escherichia, a member of the Proteobacteria phylum, showed increased abundance, while Lactobacillus abundance decreased in antibiotic-treated mice compared to untreated mice (Fig. 1G). LEfSe analysis also confirmed that Escherichia was the dominant genus in the gut microbiota of the antibiotic-treated group (Fig. 1H). Notably, FMT partially reversed the depletion of the gut microbiota in microbiota-depleted mice, although it remained associated with the elevated abundance of Proteobacteria, Firmicutes, and Escherichia and reduced abundance of Lactobacillus compared to that in the Con group.

Fig. 1. Antibiotic-induced alterations in the gut microbiome profile of mice.

A Experimental design: A cohort of 54 male wild-type mice was assigned to three groups. One group received a combination of broad-spectrum antibiotics (metronidazole, neomycin, ampicillin, and vancomycin) in their drinking water for 3 weeks. Another group underwent fecal microbiota transplantation (FMT) after antibiotic treatment for 4 consecutive days, while the third group served as untreated controls. B Representative histological sections of the small intestine from untreated and microbiota-depleted mice show preserved epithelial integrity with hematoxylin and eosin (H&E) staining. Ki67 staining indicates restricted proliferation in the crypts, MUC2 staining shows goblet cell differentiation, and Chromogranin A staining depicts neuroendocrine cell differentiation in both groups (representative images out of 6 mice/group), scale bars: 100 μm. C Histopathological staining scores for intestinal sections. D α-Diversity assessment with Shannon and Simpson indices shows reduced diversity in antibiotic-treated mice compared to that in control mice. E Principal coordinate analysis (PCoA) biplot illustrating β-diversity based on the Bray-Curtis dissimilarity index highlights distinct clustering among the three groups. F Relative abundance at the phylum level for Proteobacteria, Bacteroidetes, and Firmicutes, with significant changes observed in antibiotic-treated mice. G Relative abundance at the genus level for Escherichia and Lactobacillus shows a notable increase in Escherichia abundance and a decrease in Lactobacillus abundance in the microbiota-depleted group. H LEfSe analysis results indicate Escherichia as the dominant genus in the antibiotic-treated group. All data are expressed as mean ± SEM of biologically independent samples from different mice (n = 6 mice/group in one independent experiment); ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, and ^****p < 0.0001 (Kruskal–Wallis ANOVA test followed by two-sided Mann–Whitney U test).

To evaluate the impact of the intestinal microbiota composition on host defense against lung infection, mice received intranasal instillation of CRKP on day 25 (Fig. 2A) and were assessed by survival analysis, lung histological scoring, bacterial load quantification, and immune response profile analysis after 24 h. Survival analysis revealed numerically lower mortality rates in the untreated group (16.0%) and the FMT group (33.3%) compared to those in the microbiota-depleted mice group (50.0%). However, the absence of statistical significance (Log-rank χ² = 1.394, p = 0.238) may be attributed to the relatively limited sample size (n = 6/group) or to individual variability in the host response to FMT (Fig. 2B). Subsequently, the severity of inflammation and tissue injury induced by CRKP infection was quantified by histological scoring of H&E-stained lung sections. The infected microbiota-depleted mice developed severe pulmonary inflammatory lesions, characterized by inflammatory cell infiltration in the alveolar parenchyma and perivascular area, alveolar septal thickening, and interstitial edema; however, FMT partially restored the inflammation levels in these mice (Fig. 2C). Consistent with this observation, microbiota-depleted mice showed higher bacterial loads in their lungs (p = 0.024) and blood (p = 0.0049) after CRKP challenge compared to untreated mice; however, FMT improved bacterial clearance from both lungs (p = 0.049) and blood (p = 0.0084) following CRKP infection (Fig. 2D). Intranasal infection with K. pneumoniae caused increased production of the cytokines TNF-α and IL-1β in lung homogenates from the microbiota-depleted mice compared to that in untreated mice. However, FMT significantly reduced the levels of TNF-α (p < 0.0001) and IL-1β (p = 0.0028), compared to those in microbiota-depleted mice (Fig. 2E). The levels of other cytokines are shown in Supplementary Fig. 1. These findings demonstrate that gut microbiota alterations, particularly increased abundance of Escherichia, may influence host defense against CRKP-induced pneumonia.

Fig. 2. FMT improves outcomes and reduces inflammation in CRKP-induced pneumonia.

A Experimental design: Four days post-treatment, mice received an intranasal challenge with 25 μL CRKP suspension at 10⁵ CFU concentration, while the non-infected control group (n = 12 male wild-type mice) received intranasal instillation of sterile PBS. B Survival curves following CRKP infection showed a numerically higher survival rates in the FMT group compared to those in microbiota-depleted mice (Log-rank [Mantel–Cox] test, n = 6 mice/group). C Representative lung slides depicting pulmonary inflammation and total lung histopathological scores, with FMT reducing inflammation levels in microbiota-depleted mice, scale bars: 50 μm. D Quantification of the CRKP bacterial load in the left lung and blood samples reveals decreased bacterial burden in the FMT group. E Lung cytokine levels measured by ELISA show that FMT significantly reduced TNF-α and IL-1β levels compared to those in microbiota-depleted mice. All data are expressed as mean ± SEM of biologically independent samples from different mice (n = 6 male mice/group from one independent experiment); ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, and ^****p < 0.0001 (Kruskal–Wallis ANOVA test followed by two-sided Mann–Whitney U test).

Association between E. fergusonii abundance and the severity of K. pneumoniae-associated pneumonia

To examine the impact of E. fergusonii on the microbiome along the gut-lung axis, qPCR was initially performed to evaluate the absolute copy numbers of E. fergusonii in fecal samples from healthy controls (n = 25) and patients with K. pneumoniae-associated pneumonia (critically ill patients: n = 30; non-critically ill patients: n = 24). The univariate analysis of demographic and clinical characteristics and multivariate analysis of risk factors associated with the severity of K. pneumoniae-associated pneumonia are summarized in Supplementary Tables 1 and 2, respectively. To demonstrate the detection range of the qRT-PCR analysis, tenfold serial dilutions containing 1 × 10⁴ to 1 × 10⁹ genomic DNA/mL were assayed in triplicate for the Ef gene of the E. fergusonii strain (Fig. 3A). Representative results were reported as threshold cycle (C_T) number versus log starting quantities of DNA. The standard curve analysis presented in Fig. 3B demonstrates a correlation coefficient of 0.9936, indicating a highly reliable linear relationship. Total microbial cell counts in fecal samples, as determined by 16S rRNA gene quantification, were significantly reduced in both non-critically ill (p = 0.006) and critically ill patient groups (p < 0.001) compared with healthy controls (Supplementary Fig. 2). This finding is consistent with a progressive disruption of the gut microbiome with increasing disease severity. To minimize variations in total bacterial DNA content across individual patients, the E. fergusonii copy numbers were normalized to total 16S and spike-in plasmid copy numbers. Significantly higher E. fergusonii copy numbers were detected in critically ill pneumonia patients compared to those in both healthy controls (p < 0.0001) and non-critically ill patients with pneumonia (p < 0.0001) (Fig. 3C). Spearman’s correlation analysis revealed a positive association between the copy numbers of E. fergusonii and the severity of K. pneumoniae-associated pneumonia (adjusted Spearman’s r = 0.808; p < 0.001; Fig. 3D).

Fig. 3. Absolute quantification of E. fergusonii in fecal samples from healthy controls (n = 25) and patients with K. pneumoniae-associated pneumonia (critically ill patients, n = 30; non-critically ill patients, n = 24).

A Representative qPCR amplification curves for E. fergusonii, with relative fluorescence units (RFU) plotted against the number of PCR cycles. B Standard curves for absolute quantification of E. fergusonii. C Absolute abundance of E. fergusonii in fecal samples, normalized to total 16S rRNA gene copies and spike-in recovery. Data are presented as Log₁₀ copies/g of stool. D Spearman’s correlation analysis between E. fergusonii copy numbers (normalized to total 16S rRNA gene copies and spike-in recovery) and the severity of K. pneumoniae-associated pneumonia. Shaded area indicates the 95% confidence interval. Adj., adjusted; ^****p < 0.0001 (Kruskal–Wallis ANOVA test followed by two-sided Mann–Whitney U test).

FMT alleviates gut dysbiosis due to gut colonization by E. fergusonii

To investigate whether colonization by E. fergusonii disrupts the gut microbiota in a murine model and whether such dysbiosis can be ameliorated by FMT, we designed an experiment to establish gut dysbiosis through intraperitoneal administration of clindamycin (to which both E. fergusonii and K. pneumoniae are naturally resistant), which facilitated the stable colonization of E. fergusonii in mice. The experimental design and main study groups are shown in Fig. 4A. We characterized the gut microbiota alterations associated with antibiotic (clindamycin) or E. fergusonii exposure. The monitoring of E. fergusonii load in stool, measured by CFU and qPCR, indicated that antibiotic administration maintained stable gut colonization by E. fergusonii for 21 days (Fig. 4B, and Supplementary Fig. 3), consistent with previous reports²⁰. Mice exposed to both clindamycin and E. fergusonii showed no signs of infection or histological evidence of colitis, indicating colonization rather than infection (Fig. 4C). The 16S rRNA amplicon analysis revealed several key findings regarding the impact of E. fergusonii and the subsequent effects of FMT. Mice colonized by E. fergusonii demonstrated the lowest α-diversity index among all groups, indicating significant disruption in the gut microbiome. However, this α-diversity index increased significantly following FMT (p < 0.0001, Fig. 4D). β-diversity analysis, assessed through PCoA with Bray-Curtis dissimilarity distance, revealed that the E. fergusonii colonization (C-Fer) and Con groups clustered distinctly; the results highlighted substantial differences in microbial composition. Notably, FMT partially restored this clustering, bringing the microbiome closer to that of the Con group (p = 0.001, Fig. 4E). Furthermore, the ratio of Firmicutes to Bacteroidetes was significantly higher in E. fergusonii-colonized mice than in mice exposed to either E. fergusonii or clindamycin alone. This imbalance decreased significantly following FMT (p < 0.0001, Fig. 4F). At the family level, Muribaculaceae emerged as the most abundant bacterial group across all five groups. Notably, its abundance was significantly reduced in the E. fergusonii colonization group compared to that in mice exposed to either E. fergusonii or clindamycin alone. However, FMT increased its abundance (p = 0.0002, Fig. 4G). Gut colonization by E. fergusonii was associated with a specific dysbiosis profile, characterized by a consistently marked decrease in the relative abundances of Muribaculaceae and Lachnospiraceae_group when compared with gut dysbiosis induced by clindamycin alone. The relative abundance of Enterobacteriaceae was increased in mice treated with clindamycin or colonized by E. fergusonii, compared with controls (Supplementary Data 1). At the genus level, the relative abundance of Escherichia/Shigella was numerically increased in E. fergusonii-colonized mice and clindamycin-treated mice, relative to the control group (Supplementary Data 2). Taken together, these results suggest that E. fergusonii colonization may be closely associated with gut dysbiosis. Thus, FMT appears to demonstrate the potential to restore the gut microbiome, highlighting its beneficial effect against dysbiosis induced by E. fergusonii colonization.

Fig. 4. FMT improves gut dysbiosis induced by the asymptomatic gut colonization of E. fergusonii.

A Experimental plan and group overview for establishing gut dysbiosis through intraperitoneal clindamycin administration and subsequent E. fergusonii colonization. B Quantification of E. fergusonii load in stools for 21 days (n = 9 mice/group in one independent experiment) shows sustained colonization in treated mice. C Representative images of colon sections stained with H&E on day 21 before lung infection (representative images out of 3 mice/group) indicate no signs of infection or colitis in colonized mice. Scale bars: 50 μm. D–G Gut microbiota 16S rRNA analysis on day 21 (n = 5 male mice/group from the same experiment), including α-diversity (Shannon index), PCoA biplot of β-diversity (Bray–Curtis dissimilarity index), relative abundance at the phylum level, and relative abundance at the family level, highlights significant improvements in microbiome composition following FMT. All data are expressed as mean ± SEM from independent samples from different mice. ^*p < 0.05, ^***p < 0.001, and ^****p < 0.0001 (Kruskal–Wallis ANOVA test followed by two-sided Mann–Whitney U test).

Pre-existing intestinal colonization by E. fergusonii is associated with a worsened outcome of subsequent CRKP infection, which is improved by FMT

To assess the effect of E. fergusonii colonization or antibiotic treatment on the host’s defense against lung infections, mice were intranasally infected with CRKP on day 21 and evaluated 24 h later (Fig. 5A). The CRKP lung infection severity was significantly worse in mice colonized by E. fergusonii, as evidenced by increased lung histological scores (p < 0.0001, Fig. 5B), clinical disease severity scores (p < 0.0001, Fig. 5C), bacterial load of CRKP in lung homogenates (p < 0.0001, Fig. 5D), higher mortality rates (50% versus 16.7%, p = 0.0014, Fig. 5E), and elevated serum cytokine levels (Supplementary Fig. 4). However, these adverse effects were significantly or partially alleviated following FMT treatment, which improved histological and clinical scores, reduced bacterial load, decreased mortality rates, and normalized serum TNF-α levels (Fig. 5B–E, and Supplementary Fig. 4). Interestingly, the Clin group exhibited a modest increase in bacterial load, serum TNF-α levels, and mortality compared to the Con group. This finding suggests that, while clindamycin-induced microbiome disruption did not significantly worsen infection outcomes in the absence of E. fergusonii, it still exhibited some negative effects.

Fig. 5. Pre-existing intestinal colonization by E. fergusonii is correlated with exacerbated outcomes of subsequent CRKP infection, which was ameliorated by FMT.

A Experimental protocol: On day 21, mice received an intranasal challenge with 25 μL CRKP suspension at 10⁵ CFU concentration, while the non-infected control group (n = 12 male wild-type mice) received intranasal instillation of sterile PBS. B Representative images of lung tissue sections stained by H&E (representative images out of 6 mice/group) and total lung histopathological scores show reduced inflammation in the FMT group. Scale bars: 50 μm. C Clinical disease severity score (n = 6 male mice/group in one independent experiment) indicates improved outcomes in the FMT group. D Bacterial load of CRKP in the left lung (n = 6 male mice/group in one independent experiment) reveals reduced bacterial load in the FMT group. E Survival curves showed a numerically higher survival rates in the FMT group (Log-rank [Mantel–Cox] test, n = 6 mice/group in one independent experiment). F Representative flow cytometry plots; percentages of CD3, CD4, Treg, and Th17 cells; and the Treg/Th17 ratio in the lungs (n = 3 male mice/group in one independent experiment) indicate restored immune balance following FMT. All data are expressed as mean ± SEM of biologically independent samples from different mice; ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, and ^****p < 0.0001 (one-way ANOVA followed by Tukey’s post-hoc test).

Based on these data, we examined the lung immune response profile of mice infected with CRKP by flow cytometry. The gating strategy is reported in Supplementary Fig. 5. The immune response profile analysis indicated that E. fergusonii colonization resulted in decreased proportions of CD3 (p < 0.0001), CD4 (p < 0.0001), and regulatory T cells (Treg, p < 0.0001) as well as a reduced Treg/Th17 ratio (p < 0.0001), while the proportion of Th17 cells (p < 0.0001) increased. FMT restored this balance by increasing Treg (p < 0.0001) and decreasing Th17 (p = 0.0042) cell populations (Fig. 5F). Similar immune response patterns were observed in spleen tissues, with comparable results to those observed in lung tissues (Supplementary Fig. 5). These findings suggest that FMT may improve the outcomes of CRKP infection in mice previously colonized by E. fergusonii, potentially by modulating the Treg and Th17 cell balance.

Decreased acetate levels in the gut and lungs are associated with E. fergusonii gut colonization, which is restored by FMT

Given that the relationship between colonization-induced gut dysbiosis and lung infection severity may involve microbial metabolites such as SCFAs, which protect against lung infections, we determined SCFA concentrations in the gut and lungs by GC-MS²¹. The results showed that E. fergusonii colonization (the C-Fer group) was associated with significantly lower levels of acetate (p = 0.0137), propionate (p = 0.0227), butyrate (p = 0.03), isovaleric acid (p = 0.0082), and total SCFAs (p = 0.0127) in the caecum compared to those in infected control mice. In contrast, the FMT group exhibited significantly increased levels of acetate (p = 0.0106), butyrate (p = 0.0003), and total SCFAs (p = 0.0116) in mice (Fig. 6A). Longitudinal monitoring showed significant differences in acetate concentrations between the C-Fer group and the Clin group (day 7: t = 8.748, p = 0.001; day 14: t = 3.291, p = 0.030; day 21: t = 2.450, p = 0.070; two-tailed t test). A subsequent analysis of SCFA concentrations in the lungs revealed distinct clustering in the principal component analysis model. As shown in Fig. 6B, the C-Fer group and the FMT group were significantly separated, indicating substantial changes in lung SCFA composition following FMT intervention. FMT significantly increased the levels of acetate (p = 0.0231) and total SCFAs (p = 0.0213) (Fig. 6C). We observed significant longitudinal changes in lung acetate levels between the C-Fer and the Clin groups (day 7: t = 3.688, p = 0.021; day 14: t = 3.254, p = 0.031; day 21: t = 1.730, p = 0.159; two-tailed t test). Based on the variable importance in projection (VIP) scores, acetate emerged as the key lung SCFA, showing the highest degree of discrimination between the Con and C-Fer groups as well as between the C-Fer and FMT groups (Fig. 6D). This finding is consistent with the results of Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis, which identified fatty acid biosynthesis as the most significantly affected pathway (Fig. 6E). Overall, these results suggest that E. fergusonii gut colonization is associated with specific alterations in the gut and lung SCFA profiles. The decreased acetate levels in the lungs of C-Fer group mice may be related to CRKP lung infection severity.

Fig. 6. Decreased concentrations of short-chain fatty acids (SCFAs) in the gut and lungs are associated with E. fergusonii gut colonization.

A Quantification of SCFAs in caecum lumen samples on day 22 shows a significant increase in acetate, butyrate, and total SCFA concentrations following FMT, as well as the temporal dynamics of acetate over the 22-day period. B Principal component analysis (PCA) score plot shows distinct clustering of lung SCFA profiles across the five experimental groups. C Quantification of SCFAs in lung samples (n = 3 male mice/group) on day 22 shows a significant increase in acetate and total SCFA concentrations following FMT, as well as the temporal dynamics of acetate over the 22-day period. D Variable importance in projection (VIP) score plot highlights the top 3 discriminating SCFAs in the lungs (VIP > 0.5), with acetate emerging as the key discriminator. E KEGG pathway-based quantitative enrichment analysis reveals fatty acid biosynthesis as the most significantly affected pathway following FMT treatment. All data are expressed as mean ± SEM of biologically independent samples from different mice (n = 3 mice/group in one independent experiment); ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, and ^****p < 0.0001 (one-way ANOVA followed by Tukey’s post-hoc test).

Oral supplementation of acetate enhances outcomes in lung infection

To evaluate the potential therapeutic effects of acetate on E. fergusonii-colonized mice, acetate was added to their drinking water for 7 days before CRKP infection (Fig. 7A). This intervention decreased mortality (Fig. 7B) and significantly reduced the histological score (p < 0.0001) in E. fergusonii-colonized mice infected with CRKP (Fig. 7C). Although no significant difference was observed in the clinical disease score following acetate supplementation, a downward trend was noted (Fig. 7D). Interestingly, our preliminary data indicated that the Bifidobacterium animalis subspecies lactis BB-12 (BB-12) effectively increased intestinal acetate levels and exerted a protective effect against lung infection, which is consistent with the results of direct acetate supplementation (Supplementary Fig. 6). Additionally, we assessed whether acetate supplementation enhanced the lung immune response profile. As expected, acetate supplementation increased the proportion of CD3 (p = 0.0039), CD4 (p = 0.0280), and Treg cells (p = 0.0134) as well as the Treg to Th17 ratio (p = 0.0296) following E. fergusonii colonization (Fig. 7E). In contrast, acetate supplementation reduced the proportion of Th17 cells (p = 0.0002, Fig. 7E). Similar immune response patterns were observed in spleen tissues, reflecting the findings noted in lung tissues (Supplementary Fig. 7). These results indicate that acetate supplementation may improve the outcomes of CRKP infection in E. fergusonii-colonized mice, potentially through modulating the balance of Treg and Th17 cells.

Fig. 7. Oral supplementation of acetate improves lung infection outcomes.

A Experimental protocol for supplementing acetate (100 mM) in drinking water for 7 days in E. fergusonii-colonized mice. B Survival curves showed a numerically higher survival rates in the acetate group (Log-rank [Mantel–Cox] test, n = 5 mice/group in one independent experiment). C Representative images of lung tissue sections stained by H&E (representative images of 5 mice/group in one independent experiment) and total lung histopathological scores show reduced inflammation in acetate-supplemented mice. Scale bars: 50 μm. D Clinical disease severity score (n = 5 mice/group in one independent experiment) shows a trend toward lower scores in acetate-supplemented mice. E Representative flow cytometry plots; percentages of CD3, CD4, Treg, and Th17 cells; and the Treg/Th17 ratio in the lungs (n = 3 mice/group in one independent experiment) reveal enhanced immune profiles following acetate supplementation. All data are expressed as mean ± SEM of biologically independent samples from different mice; ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, and ^****p < 0.0001 (one-way ANOVA followed by Tukey’s post-hoc test).

Discussion

K. pneumoniae is a Gram-negative pulmonary pathogen responsible for a variety of community- and hospital-acquired infections. Infections caused by CRKP pose a major threat to public health and are associated with high mortality rates, particularly among immunocompromised and critically ill patients²². However, broad-spectrum antibiotic treatments and intensive care procedures that induce gut dysbiosis facilitate the colonization and proliferation of multidrug-resistant enteric bacteria in the host intestine, thereby increasing mortality risk in patients²³. The intestinal microbiota and its associated metabolites play a crucial role in preventing pneumonia by inhibiting the colonization of potentially pathogenic bacteria and regulating immune responses. In the present study, we demonstrated that asymptomatic gut colonization by E. fergusonii (ATCC 35469) caused gut dysbiosis, altered the levels of SCFAs in both gut and lungs, modified the immune response profile, and increased the severity of CRKP (KPC carbapenemase, ST15)-induced lung infection. Our study found that FMT and acetate treatment restored the immune response profile and improved outcomes of lung infection in mice, suggesting a potential strategy to enhance immune defense against pulmonary infections through microbiota-derived metabolites such as acetate. Nevertheless, the translational potential and therapeutic value of acetate in humans remain to be further validated.

E. fergusonii belongs to the genus Escherichia and the family Enterobacteriaceae²⁴. This atypical bacterial species can cause infections in both animals and humans, including bacteremia, urinary tract infections, and diarrhea in humans and pneumonia in beef cattle^16,25. Additionally, E. fergusonii has also been detected frequently in foods and animals for consumption, suggesting its widespread presence in the food chain^26,27. E. fergusonii can alter the gut microbiome profile and biological functions, potentially contributing to the progression of other organ diseases in rats²⁸. To examine the potential association between K. pneumoniae-associated pneumonia and increased E. fergusonii abundance, we compared the levels of E. fergusonii between patients with K. pneumoniae-associated pneumonia and healthy controls. The results indicated that significantly higher copy numbers of E. fergusonii were detected in critically ill pneumonia patients than in both healthy controls and non-critically ill pneumonia patients.

Subsequently, we determined whether gut colonization by E. fergusonii exacerbates gut dysbiosis in C57BL/6J mice. Given the elevated risk of bacterial pathogen colonization following microbiota disruption, we used clindamycin to promote successful E. fergusonii colonization. Indeed, alteration of the gut microbiota induced by clindamycin alone was associated with decreased α-diversity and reduced abundance in Rikenellaceae and Marinifilaceae in our study. The unexpected increase in Lactobacillaceae abundance may reflect strain-specific resistance to clindamycin. A notable distinction between E. fergusonii colonization and clindamycin treatment alone was the increase in the Firmicutes/Bacteroidetes ratio, accompanied by decreased abundances of Muribaculaceae and Lachnospiraceae (but an increase in Lachnospiraceae NK4A136). Bhattacharya et al.²⁹ suggested that an elevated Firmicutes/Bacteroidetes ratio is typically associated with lung inflammation, while previous studies indicated a negative correlation between Muribaculaceae and lung inflammation^30,31. Lachnospiraceae are recognized for their beneficial roles in other animal models by nourishing enterocytes and providing host nutrients. Moreover, the abundance of the Lachnospiraceae NK4A136 group has been demonstrated to be associated with reduced amino acid metabolism³².

As predicted, we observed that CRKP-induced lung infections were notably more severe in mice with E. fergusonii gut colonization. This was evident through several key observations: elevated lung histopathological and clinical scores, increased bacterial loads, higher mortality rates, and disrupted Treg/Th17 cell balance. Specifically, the bacterial load of CRKP in lung homogenates and blood was markedly higher in mice colonized with E. fergusonii. This finding implies that gut colonization by E. fergusonii may compromise the host’s ability to effectively eliminate CRKP. Consequently, the pathogen shows enhanced proliferation and dissemination, leading to more severe infection. The concurrent effects of antibiotic-induced dysbiosis and E. fergusonii colonization appear to exacerbate inflammation beyond the impact of either condition independently. This synergistic effect is biologically plausible, as antibiotics can induce gut dysbiosis, resulting in systemic inflammation. The presence of E. fergusonii further aggravates this inflammatory state, potentially by modulating the gut-lung axis and enhancing the inflammatory response in the lungs. The primary manifestations of this exacerbation include intensified inflammatory responses in the lungs and a disrupted Treg/Th17 balance^6,33,34. Moreover, impaired immune responses may increase the susceptibility of the patient to various pathogenic infections, potentially disrupting intestinal microbiota balance. FMT in mice colonized by E. fergusonii significantly improved the outcomes of CRKP lung infection. This observation aligns with recent studies indicating that FMT could be a promising therapeutic intervention for conditions such as coronavirus disease 2019 (COVID-19) by restoring the damaged gut microbiome and modulating the immune response^35–37. Additionally, FMT is an effective preventive measure against bacterial infections in patients or mice colonized by multidrug-resistant bacteria^12,38. However, while FMT can improve the gut microbial environment and reduce inflammation, it may not completely counteract the synergistic inflammatory effects of E. fergusonii colonization and antibiotic treatment.

SCFAs are a potent metabolic component that influences the gut-lung axis³⁹, which is crucial for immune system development and lung mucosal function, providing protection against infections and pulmonary damage⁴⁰. SCFA analysis is critical for understanding the mechanisms underlying gut microbiota-mediated immune responses in infections. Our results indicate that E. fergusonii gut colonization was associated with reduced SCFA concentrations in the gut, which is consistent with the observed gut dysbiosis and subsequent alterations in the immune response. Although E. fergusonii colonization may not directly deplete SCFAs, antibiotic-induced gut dysbiosis facilitating E. fergusonii colonization likely resulted in the reduction of SCFA-producing bacteria. These beneficial bacterial populations, typically diminished during dysbiosis, are primarily responsible for SCFA production. Consequently, the disruption of the gut microbiota composition leads to reduced SCFA levels. Notably, the Muribaculaceae family, known to correlate positively with gut SCFA concentrations, exhibited a significant reduction in its abundance in our study model. Thus, the observed effects primarily originate from specific dysbiosis associated with decreased SCFA production. Notably, FMT partially restored gut microbiome composition and subsequently increased SCFA concentrations in mice previously colonized by E. fergusonii. Furthermore, the analysis of lung SCFA concentrations revealed that E. fergusonii gut colonization was associated with reduced levels of acetate (derived from the gut microbiota), a key mediator of the gut-lung axis in lung injury cases⁴¹. Notably, acetate supplementation has emerged as a potential therapeutic approach, as it reduces lung infection severity and enhances the immune response profile in mice. BB-12, a well-characterized probiotic strain, has been shown to alleviate the disruptive impacts of antibiotics on fecal SCFA levels and gut microbiota composition⁴². Our preliminary data (Supplementary Fig. 6) showed that BB-12 effectively elevated intestinal acetate levels and exerted a protective effect against lung infection, which is consistent with the results of direct acetate supplementation. As reported previously, acetate enhances host antiviral response⁴³, maintains the defensive functions of intestinal epithelial cells⁴⁴, modulates lung injury⁴⁵, and promotes the production of Treg cells^46,47. Under in vitro conditions, acetate directly inhibited the growth of antibiotic-resistant Enterobacterales, including carbapenemase-producing K. pneumoniae⁴⁸. Our findings indicate that microbiota-derived acetate reduces lung infection severity in E. fergusonii-colonized mice by promoting the generation of Treg cells. This observation suggests that acetate plays a crucial role in mediating the functioning of the gut-lung axis and could serve as a potential therapeutic intervention for lung infections exacerbated by gut colonization of specific pathogens.

The current study has several limitations that require careful consideration. First, the investigation focused exclusively on the colonization effects of E. fergusonii, potentially overlooking the influence of other members of the genus Escherichia. Although E. fergusonii is recognized for its ability to disrupt gut homeostasis and enhance the pathogenicity of other pathogens, other species within this genus may also contribute to gut dysbiosis and subsequent lung infection outcomes. Second, we performed a cross-sectional analysis of the patient-based data with a relatively small sample size, which may have introduced potential confounding variables and affected the generalizability of our findings. A larger, longitudinal study is necessary to thoroughly examine the dynamic changes in the gut microbiota and their correlation with the severity of K. pneumoniae-associated pneumonia over time. Third, while acetate was identified as a key mediator in the gut-lung axis when investigating mechanisms underlying exacerbated lung infections, other metabolic pathways and metabolites that might also play significant roles were not comprehensively evaluated. Future research should implement more extensive metabolomic approaches to provide a broader understanding of the microbiota-derived factors influencing lung infection outcomes.

The results of the present study demonstrate a direct link between asymptomatic E. fergusonii colonization and increased severity and mortality of subsequent CRKP lung infection due to specific gut dysbiosis. An increase in Escherichia abundance was observed in microbiota-depleted mice, along with an increase in E. fergusonii population in patients with K. pneumoniae-associated pneumonia, confirming the adverse impact of asymptomatic gut E. fergusonii colonization on subsequent lung infection outcomes. However, FMT and acetate treatment restored the immune response profile and improved lung infection outcomes. In summary, this study emphasizes the importance of the gut-lung axis in mitigating CRKP-induced pneumonia and reveals a potential metabolic immunomodulator involving microbiota-derived acetate.

Methods

Mice

Male C57BL/6J mice, aged 6–8 weeks and weighing 16–18 g, were obtained from the Laboratory Animal Center of Dalian Medical University (Dalian, China). These mice were housed in the specific pathogen-free (SPF) animal center of Dalian Medical University. Each cage contained five mice, and all animals were maintained under SPF conditions with free access to sterilized pure water and food. We have complied with all relevant ethical regulations for animal use. The experimental procedures and animal protocols were approved by the Institutional Animal Ethics Committee.

Participants and faeces sample collection

We conducted an observational study at the Second Affiliated Hospital of Dalian Medical University (Dalian, China). From April 20, 2024, to June 29, 2024, fecal samples were collected from patients with hospital-acquired K. pneumoniae-associated pneumonia following admission, and the samples were stored at −80 °C for further analysis. The diagnosis of hospital-acquired K. pneumoniae-associated pneumonia was confirmed by X-ray or computed tomography. Additionally, the diagnosis was validated by isolating K. pneumoniae from lower respiratory tract secretions, including sputum, bronchoalveolar lavage fluid, bronchoscopic needle aspiration, or blood. The exclusion criteria were as follows: (a) patients initially diagnosed to have concurrent bloodstream, abdominal, intestinal, or urinary tract infections; (b) K. pneumoniae colonization that did not meet the infection criteria; (c) outpatients or patients seeking emergency care; and (d) pregnant or lactating women. Patients with K. pneumoniae-associated pneumonia who received clinical indication of selective intestinal decolonization with oral non-adsorbable antibiotics during the study period were excluded from the analysis. Patient data were extracted from an electronic medical record system. The data were anonymized to protect privacy and were used solely for research purposes. All ethical regulations relevant to human research participants were followed. The study protocol was approved by the ethics committee of the hospital, and informed consent was obtained from all participants or their close relatives.

A total of 54 patients with hospital-acquired K. pneumoniae-associated pneumonia, comprising 30 critically ill patients and 24 non-critically ill patients, and 25 healthy controls were enrolled in this study. The diagnostic criteria for critically ill patients required the presence of at least one major criterion or at least three minor criteria⁴⁹, and patients with community-acquired pneumonia were excluded from the study.

Quantification of E. fergusonii in fecal samples by real-time PCR

Genomic DNA was extracted from fecal samples collected from human subjects and E. fergusonii-colonized mice using the E.Z.N.A.® Stool DNA Kit (Omega Bio-Tek, Norcross, GA, USA) according to the manufacturer’s protocol. Quantitative real-time PCR (qPCR) was conducted using specific primers Ef-F (AGATTCACGTAAGCTGTTACCTT) and Ef-R (CGTCTGATGAAAGATTTGGGAAG) and universal 16S primers 16S-F (TTGACGTTACCCGCAGAAGAA) and 16S-R (GCTTGCACCCTCCGTATTACC) in 384-well plates on the QuantStudio 6 Flex real-time PCR system (Life Technologies, Carlsbad, CA, USA) with ChamQ Universal SYBR qPCR Master Mix (Vazyme, Nanjing, China). The specificity of the primers was verified by the BLAST search program (www.ncbi.nlm.nih.gov/BLAST) and qualitative PCR. The resulting amplicons were separated on 1% agarose gels and visualized after ethidium bromide staining⁵⁰. The recombinant plasmid pTZREP-529, harboring the Toxoplasma gondii REP-529 repetitive element, was used as an exogenous spike-in control to monitor and normalize DNA extraction and amplification recovery⁵¹. Total bacterial load in fecal samples was assessed using a qPCR assay targeting the 16S rRNA gene⁵². Each reaction contained 4 μL template DNA, 25 μL 2×ChamQ Universal SYBR qPCR Master Mix, forward and reverse primers each at a final concentration of 1 μM, and nuclease‑free water to adjust the total reaction volume to 50 μL. Thermal cycling was performed with an initial denaturation at 95 °C for 30 s, followed by 40 cycles consisting of 95 °C for 10 s and 60 °C for 30 s. Melting‑curve analysis was performed following amplification to confirm specific target amplification. Each run included negative controls (without DNA) and extraction controls (without pellets). All qPCR assays were performed in triplicate. The absolute copy numbers of E. fergusonii gene in human fecal samples were normalized to those of the 16S rRNA gene, determined using the corresponding standard curves, and further corrected based on the recovery rate of the pTZREP-529 spike-in control.

Bacterial strains

The standard strain E. fergusonii ATCC 35469 was obtained from Shanghai Preservation Biotechnology Center (Shanghai, China). The CRKP strain (producing KPC carbapenemase, sequence type [ST] 15 clonal lineage), sourced from the clinical laboratory in a university-affiliated hospital in China, was utilized for mouse lung infection studies. The strains were cultured in Luria-Bertani (LB) broth (Solarbio, Beijing, China) at 37 °C with overnight agitation. The bacterial concentration was determined by measuring the optical density at 600 nm and enumeration of colony-forming units (CFUs). Freeze-dried Bifidobacterium animalis subspecies lactis BB-12 (BB-12) was obtained from the Culture Collection of Dalian Medical University (Dalian, China) and stored at −20 °C until experimental use. Viable BB‑12 cell counts from freeze‑dried powder were quantified via the plate count method. Following temperature equilibration at 23 °C, 500 mg powder was resuspended in 0.9% (w/v) saline and homogenized with a Smasher homogenizer (BioMérieux, Craponne, France). Serial dilutions were plated onto cysteine‑supplemented De Man, Rogosa and Sharpe (MRS) agar (Solarbio, Beijing, China). Plates were incubated anaerobically at 37 °C for 48 h in GasPak jars (BBL, Cockeysville, USA), and viable colonies were subsequently counted⁵³.

Animal experimental study design

For the CRKP-induced pneumonia experiments after gut microbiota disruption, mice were randomly assigned to three infected groups (n = 18 per group) and the non-infected group (n = 12): (i) untreated group; (ii) microbiota-depleted group; (iii) fecal transplant group; (iv) non-infected group, intranasally instilled with sterile phosphate-buffered saline (PBS). Following antibiotic treatment, fecal microbiota transplantation (FMT) was performed in the fecal transplant group for 4 consecutive days. Six mice from each group were euthanized. The experimental design is shown in Fig. 1A. All remaining mice were inoculated with CRKP (Fig. 2A). Subsequently, six mice from each group were euthanized 2 days after CRKP infection, while the remaining six mice were monitored for 6 days.

For the CRKP-induced pneumonia experiments following gut colonization of E. fergusonii, mice were randomly assigned to five infected groups (n = 18 per group) and the non-infected group (n = 12): (i) control group (Con); (ii) E. fergusonii exposure group (Fer); (iii) clindamycin treatment group (Clin); (iv) E. fergusonii colonization group (C-fer); and (v) fecal transplant group (FMT); (vi) non-infected group, intranasally instilled with sterile PBS. Six mice from each group were euthanized before CRKP infection. The experimental design and main study groups are shown in Fig. 4A. All remaining mice were inoculated with CRKP (Fig. 5A). Subsequently, six mice from each group were euthanized 2 days after CRKP infection, while the remaining six mice were monitored for 6 days. Saline (1 mL) containing 3 × 10⁹ CFU of E. fergusonii was administered daily through gavage for 14 days, starting from day 1 in the Fer, C-Fer, and FMT groups. The Con and Clin groups received 1 mL of sterile saline. Gut dysbiosis was induced through an intraperitoneal injection of clindamycin at 10 mg/kg on days 0, 7, and 14¹². Intraperitoneal injection ensured rapid absorption of the antibiotic into the bloodstream and its systemic distribution, enabling more effective depletion of the intestinal microbiome. FMT was performed on days 16 and 20 in the FMT group. On day 21, acute lung infection was induced through intranasal instillation of CRKP at 4 × 10⁶ CFU/mL concentration in 25 μL of bacterial suspension⁵⁴.

For the acetate supplementation experiments, mice were assigned to receive either 100 mM sodium acetate in drinking water (n = 13 per group, Fig. 7A) or approximately 1 × 10⁷ CFU of B. lactis BB-12 (BB-12 group, Supplementary Fig. 6). Sodium acetate was administered in drinking water from days 15 to 21, with solutions changed twice weekly, and no decrease in water intake. For the BB-12 group, mice received daily oral gavage of 100 µL BB-12 suspension (resuspended in sterile saline) over the identical period (days 15 to 21), with voluntary ingestion from the tip of a gavage bulb⁵³. For all survival analyzes, mice were humanely euthanized via intraperitoneal injection of pentobarbital sodium (100 mg/kg) when body weight loss exceeded 20%. In vivo studies, no blinding of group allocation was performed; mice were randomly assigned to experimental groups using simple random sampling. Sample size was calculated based on previous data on K. pneumoniae infection experiments.

Microbiota depletion

To deplete the gut microbiota, mice were administered a combination of broad-spectrum antibiotics (metronidazole 1 mg/mL, neomycin sulfate 1 mg/mL, ampicillin 1 mg/mL, and vancomycin 0.5 mg/mL, Solarbio) in their drinking water for 3 weeks, with solutions changed twice weekly⁵⁵. The antibiotic therapy was discontinued 4 days before the infection.

FMT

SPF C57BL/6J mice were housed in aseptically autoclaved empty cages (without bedding) to ensure normal bowel movement. Fresh fecal pellets (200 mg) were collected from each donor cage using sterile filter paper, immediately resuspended in 3 mL of sterile PBS, and homogenized for 2 min with a sterile glass pestle under aseptic conditions. Subsequently, the homogenate was centrifuged at 800 × g for 3 min at 4 °C, and the resulting supernatant was carefully collected as the fecal microbiota suspension for oral supernatant transplantation. Mice in the FMT group were administered 200 μL of fresh fecal supernatant per mouse via oral gavage daily for 4 consecutive days, whereas for the microbiota-depleted group, an equal volume of sterile physiological saline was administered via oral gavage using the identical regimen to replace the fecal supernatant.

Model of infectious pneumonia

Mice were infected intranasally with approximately 10⁵ CFU of CRKP in 25 μL PBS after being anesthetized with 3–5% isoflurane inhalation through a vaporizer. Non-infected mice received intranasal administration of an equal volume of PBS⁵⁶. Clinical symptoms were monitored and scored based on the established parameters¹².

Pathological examination and immunohistochemical assay

Lung and colon tissues were fixed in 4% paraformaldehyde solution and embedded in paraffin. Hematoxylin (0.5% w/v, Solarbio, Beijing, China) and 1% eosin solution (H&E, Solarbio)-stained paraffin sections (4 μm) were quantitatively scored for histopathology^57,58. Lung sections were assessed for bronchitis, edema, interstitial inflammation, intra-alveolar inflammation, pleuritis, and endothelialitis. A blinded experienced pathologist scored each parameter on a scale of 0 to 4, with 0 indicating absent and 4 indicating severe. The total histological score was calculated by summing individual parameter scores, with a maximum possible score of 24. Colon sections were evaluated for mononuclear cell infiltration, polymorphonuclear leukocyte infiltration, epithelial hyperplasia, and epithelial injury. These parameters were graded on a scale of 0 to 3, with 0 indicating absent, 1 indicating mild, 2 indicating moderate, and 3 indicating severe, with a maximum possible score of 12.

Paraffin-embedded colon sections obtained following broad-spectrum antibiotic treatment were prepared for immunohistochemical analyzes. The sections were incubated overnight at 4 °C with the following primary antibodies diluted in PBS containing 1% bovine serum albumin: anti-Ki67 (dilution 1:1000; ab15580, Abcam, Cambridge, UK), mucin2 H300 (dilution 1:500; sc-15334, Santa Cruz Biotechnology, Santa Cruz, USA), and anti-Chromogranin A (dilution 1:200; 10529-1-AP, Proteintech, Rosemont, USA). After three PBS washes, the sections were incubated with appropriate HRP-conjugated secondary antibodies (dilution 1:100; ab205718, Cambridge, Abcam, UK) for 1 h at room temperature, and staining was developed using diaminobenzidine.

Clinical disease score

At 48 h post infection with CRKP, mice were assessed for pneumonia severity using a clinical disease score system based on the following parameters: fur appearance (0–2 scale), activity level (0–2 scale), and body temperature (0 for 30–35 °C; 2 for <30 °C). The total clinical disease score was calculated by summing these parameters and incorporating the percentage of weight loss divided by 10⁵⁹.

Microbiota analysis

The gut bacterial analysis was performed by amplifying the V3-V4 region of the 16S rRNA gene and sequencing using the Illumina MiSeq platform. Microbial genomic DNA was extracted from murine faeces using the QIAamp DNA Stool Mini Kit (QIAGEN, Courtabeuf, France) in accordance with the manufacturer’s protocol. The V3-V4 regions of the 16S rRNA gene were amplified using the following specific primers: 338F, 5′-ACTCCTACGGGAGGCAGCAG-3′ and 806R, 5′-GGACTACHVGGGTWTCTAAT-3′. Sequencing was conducted on an Illumina MiSeq platform. Library controls included a no-template control (water) and a DNA extraction blank control. Operational taxonomic units (OTUs) present in more than half of the fecal samples were identified as core OTUs. Community diversity (α-diversity) was evaluated using the Shannon and Simpson indices. β-diversity was analyzed by principal coordinate analysis (PCoA) with Bray-Curtis distance using the “microbial” R package⁶⁰. To assess the percentage of variation due to the treatment and its statistical significance, the Permutational Multivariate Analysis of Variance (PERMANOVA) test was employed using the Vegan package (v2.5-3). Taxonomic differences between groups at multiple levels were evaluated using Linear Discriminant Analysis Effect Size (LEfSe) with an LDA score threshold of 4 to identify the specific microbial taxa responsible for these variations⁶⁰. The Illumina sequencing, data processing, and data analysis were described in detail in the Supplementary Information.

Bacterial load quantification

To determine the bacterial load of CRKP in blood and lung homogenates, tenfold serial dilutions were prepared using sterile isotonic saline by plating 100 μL onto a selective medium (MacConkey agar [Solarbio, Beijing, China] with 1 μg/mL cefotaxime and 0.125 μg/mL meropenem)⁶¹. These dilutions were cultured for bacterial quantification at 37 °C for 12 h. The bacterial counts in the tissues were calculated based on the CFU value. For the isolation and quantification of E. fergusonii isolates, fecal samples were collected and cultured on MacConkey agar (Oxoid, Basingstoke, UK) and Simmons citrate agar (Becton-Dickinson, Oxford, UK) supplemented with 2% adonitol. All agar plates underwent aerobic incubation at 37 °C, and we examined the plates for the presence of yellow colonies at 24 h and 48 h. Suspected colonies were selected for biochemical testing and further confirmed using the API 20E system according to the manufacturer’s instructions (BioMerieux, Basingstoke, UK). Bacterial load in fecal samples was quantified and expressed as CFU/g of feces⁶².

ELISA

The levels of TNF-α, IL-1β, IL-6, IL-10, IL-17, IL-22, and CXCL-1 in lung homogenates and serum were quantified using ELISA kits (BD Biosciences, San Jose, CA, USA) in accordance with the manufacturer’s protocols. Cytokine concentrations in lung homogenates were normalized to total protein content, and absorbance was measured at 450 nm to calculate analyte concentrations against fitted standard curves.

Flow cytometry analysis

Lung and spleen tissues were collected aseptically and homogenized in PBS. A total of 2 × 10⁶ cells were used for flow cytometry analysis. Surface staining was performed in the presence of Fc-blocking antibodies (anti-CD16/CD32, clone: 2.4G2, 553141, BD Biosciences, San Diego, CA, USA; 1:50) by using fluorescein isothiocyanate-labeled anti-mouse cluster of differentiation 3 (CD3; clone: 145–2C11, 553061, BD Biosciences, San Diego, CA, USA; 1:200) and PE-Cy5-labeled anti-mouse cluster of differentiation 4 (CD4; clone: GK1.5, BD Biosciences, San Diego, CA, USA; 1:100) antibodies for 30 min at 4 °C. The cells were then fixed, permeabilized, and stained with allophycocyanin-labeled anti-mouse Foxp3 (clone: FJK-16s, BD Biosciences, San Diego, CA, USA; 1:20) or PE-labeled anti-mouse IL-17A (clone: SCPL1362, BD Biosciences, San Diego, CA, USA; 1:20) antibodies. Data were acquired using an Accuri^TM C6 Flow Cytometer (BD Biosciences, San Diego, CA, USA) and analyzed with Flow Plus software (v1.0.264.15).

SCFA analysis

Gas chromatography-mass spectrometry (GC–MS) was used to measure SCFA concentrations in feces and lung tissue samples. For each 20 mg sample, 800 µL of phosphoric acid (0.5% v/v) solution was added. The mixture was frozen and ground at 50 Hz for 3 min in two cycles, followed by 10 min of ultrasonication. After centrifugation, 0.2 mL of supernatant was combined with N-butanol solvent containing the internal standard 2-ethylbutyric acid (10 μg/mL) for extraction. The sample was vortexed for 10 s, ultrasonicated for 10 min, and centrifuged at 4 °C. The supernatant was collected and subjected to GC–MS analysis of SCFAs by using an Agilent 8890B gas chromatograph coupled to a 7000D mass spectrometer with an HP-FFAP capillary column (30 m × 0.25 mm × 0.25 μm). Helium (99.999%) was used as the carrier gas at a 1 mL/min flow rate. The GC column temperature was initially maintained at 80 °C, increased to 120 °C at 20 °C/min, further increased to 160 °C at 5 °C/min, and held at 220 °C for 3 min. Sample injection volume was 1 µL in the splitting mode (10:1) with inlet temperature at 180 °C. Data were acquired using MassHunter quantitative analysis software (v10.0.707.0; Agilent Technologies, Santa Clara, CA, USA).

Statistics and reproducibility

All data are presented as mean ± standard error of the mean (SEM). Statistical analyzes were conducted with one-way ANOVA followed by Tukey’s post-hoc test or t-test using GraphPad Prism (v7.04, GraphPad Software, Inc., La Jolla, CA, USA). The Kaplan–Meier log-rank test was used for comparing survival between the groups. Spearman’s rank correlation coefficient (Spearman’s r) was calculated to assess the relationship between E. fergusonii copies in patients’ fecal samples and the severity of K. pneumoniae-associated pneumonia. A p-value < 0.05 was considered statistically significant.

Ethical statement

All animal experiments conducted in this study were approved by the Dalian Medical University Animal Care and Use Committee (Approval No. AEE18001). The human observational study protocol was approved by the ethics committee of the hospital (Approval No. KY2024-051-01), and informed consent was obtained from all participants or their close relatives.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Y.L. Conceptualization, Investigation, Formal analysis, Writing-original draft. J.Y. Investigation, Validation, Data curation, Project administration. Q.L. Validation, Data curation, Project administration. S.G. Validation, Data curation, Project administration. J.H. Validation, Data curation, Project administration. F.L. Investigation, Software. Y.W. Investigation, Software. C.W. Validation, Data curation, Project administration. D.W. Conceptualization, Investigation, Writing-review & editing. X.H. Validation, Data curation, Project administration. M.Z. Validation, Data curation, Project administration. E.L. Conceptualization, Software, Investigation, Supervision, Writing-review & editing. Y.C. Conceptualization, Funding acquisition, Investigation, Supervision, Writing-review & editing. All authors read and approved the final manuscript.

Peer review

Peer review information

Communications Biology thanks Dennis Nurjadi and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editors: Sabina Leanti La Rosa and Rupali Sathe. A peer review file is available.

Funding

This work was supported by the Scientific Research Project of the Education Department of Liaoning Province (grant number LJ212410161048) and the General Project of Liaoning Provincial Department of Education (JYTMS20230566). This work was also supported by the Liaoning Provincial Program for Top Discipline of Basic Medical Sciences and Youth Talent Cultivation Fund Project of Dalian Medical University.

Data availability

The sequencing datasets generated during and/or analyzed in the current study are deposited in the GenBank Sequence Read Archive (SRA) under the accession number PRJNA1125855 (submission ID: SUB14542092). The underlying source data for all graphs presented in the manuscript have been provided as Supplementary Data 3. Other data will be available from the corresponding author upon request.

Code availability

No custom software or code was developed for this study. Standard publicly available tools and packages were used for data processing and statistical analysis, as detailed in the Methods section.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s42003-026-10793-w.