Whole genome comparisons reveal gut-to-lung translocation of Escherichia coli and Burkholderia cenocepacia in two cases of ventilator-associated pneumonia in ICU patients

Abstract

Background

Identifying the sources of pathogenic bacteria causing ventilator-associated pneumonia (VAP) in intensive care unit (ICU) patients is crucial for developing effective prevention and treatment strategies. However, the scarcity of reported cases with confirmed sources limits the ability to evaluate and manage VAP, which remains a major challenge for healthcare systems globally.

Methods

Pathogens were isolated from endotracheal aspirate (ETA) samples of VAP patients using conventional culture techniques. Whole-genome comparisons, based on average nucleotide identity (ANI), were performed to identify genetically identical strains by comparing pulmonary isolate genomes with gut metagenome-derived bacterial genomes. Mouse models of pneumonia and colitis were used to validate the translocation of pathogenic bacteria from the gut to the lungs. Metagenomic analysis was performed to characterize the gut microbiome and resistome.

Results

Pathogenic isolates were obtained from the ETA samples of seven VAP patients, with one isolate per sample. Among these, Escherichia coli (Ec1) and Burkholderia cenocepacia (Bc1) from two patients were genetically identical to strains in their respective gut microbiota, with ANI values above 99%, indicating gut-to-lung translocation. The Ec1 strain demonstrated increased resistance to cefazolin while remaining susceptible to gentamicin, amikacin, and kanamycin, compared to previously reported pneumonia-associated E. coli strains. The Bc1 strain showed elevated resistance to macrolides, chloramphenicols, and tetracyclines relative to pneumonia-associated B. cenocepacia strains. Metagenomic analysis revealed a highly individualized gut microbiota composition among VAP patients. Notably, the translocated bacteria were not dominant within their gut microbiota. Additionally, these patients showed a marked increase in the total abundance of antibiotic resistance genes (ARGs) in their gut microbiota. The translocation ability of the Ec1 strain was validated in a mouse pneumonia model, where it caused more severe lung damage. Furthermore, elevated levels of Escherichia-Shigella were detected in the lung tissues of colitis mice, suggesting that gut-to-lung bacterial translocation may occur in a severely inflamed host, potentially leading to pneumonia.

Conclusions

This study demonstrates the gut-to-lung translocation of E. coli and B. cenocepacia, highlighting their role in the development and progression of VAP in ICU patients. These findings provide valuable insights for implementing targeted prevention and treatment strategies for VAP in ICU settings.

Graphical abstract

Introduction

Ventilator-associated pneumonia (VAP) is a common nosocomial infection affecting lung tissues, typically occurring 48 h or more after the intubation of mechanical ventilation [1]. It is one of the most prevalent infections acquired in intensive care unit (ICU) and remains a significant contributor to morbidity and mortality [2]. According to the National Healthcare Safety Network, VAP accounts for 7–32% of healthcare-associated infections and 10% of pediatric device–related infections [3]. Pathogenic bacteria frequently implicated in VAP include both Gram-positive and Gram-negative species, such as Staphylococcus aureus, Pseudomonas aeruginosa, Klebsiella, and Enterobacter species [3, 4]. The precise sources of these pathogens are often unclear, posing challenges for the development of effective preventive strategies against VAP.

The etiology, epidemiology, pathophysiology, and treatment of VAP have been extensively studied for preventive and therapeutic purposes [5]. Previous studies often suggest that bacteria may reach the lungs through fecal contamination or by ascending migration from the intestines into the stomach, followed by aspiration into the lungs [6]. Despite significant efforts to enhance hygiene and reduce ICU-acquired infections, the incidence of VAP remains persistently high [7]. Beyond the reported causes, other overlooked factors may contribute to the development of ICU-acquired lung infections, warranting further investigation.

Recent studies have proposed the possibility of internal bacterial translocation from the gut across the intestinal barrier into the systemic or lymphatic circulation, ultimately reaching the lungs [8,9,10]. This hypothesis is supported by microbiome-level analyses comparing bacterial compositions in the lungs and gut of patients with acute respiratory distress syndrome, as well as in post-stroke mouse models [9, 10]. A notable study further substantiated this concept by identifying P. aeruginosa translocating from the gut to the lungs in an ICU patient through whole-genome comparisons of isolates obtained from traditional culture methods [11]. Despite these advances, the detection and characterization of translocated bacteria remain challenging due to the lack of effective screening labels for target organisms, making their isolation from complex gut microbial communities labor-intensive. Moreover, microbiome-level comparisons often fail to provide strain-level details, which are essential for understanding antibiotic resistance, physiological traits, and guiding therapeutic interventions. Consequently, gut-to-lung bacterial translocation via the gut-lung axis may be a contributing factor to VAP, the limited reports of this phenomenon in VAP patients could be attributed to methodological limitations.

Escherichia coli, predominantly from the B2 phylogenetic group, has been identified as a causative agent of ICU-associated pneumonia, characterized by a high virulence factor gene content and low antimicrobial resistance [12]. Recent data indicate that Enterobacteriaceae, particularly E. coli, are frequently implicated in ICU-associated pneumonia, with an incidence of approximately 6.64–10.52 isolates/1000 patient-days [13]. Burkholderia cenocepacia, a prominent member of the Burkholderia cepacia complex, is the second most common pathogen associated with lung infections in cystic fibrosis patients [14, 15]. Its virulence is underpinned by several key features, including the ability to acquire foreign DNA, regulate gene expression via quorum sensing, compete for iron during infection, and mediate antimicrobial resistance and inflammation through its membrane and surface polysaccharides [14]. Despite the clinical relevance of E. coli and B. cenocepacia in VAP, the specific sources of these pathogens within ICU environments remain unclear.

In this study, we employed whole-genome comparisons between pulmonary isolate genomes and gut metagenome-derived genomes to illustrate the gut-to-lung translocation. To further validate this translocation, we utilized a mouse pneumonia model colonized with human-originated E. coli and a DSS-induced colitis mouse model. These findings provide novel insights into possible pathways contributing to the development of VAP.

Materials and methods

Study population

This study included 36 eligible participants, consisting of seven VAP patients and 29 healthy controls. The VAP patients were recruited from the ICU at Tianjin Third Central Hospital, Tianjin, China, between December 2020 and June 2021. Inclusion criteria for VAP patients were individuals of any genders meeting the diagnostic criteria for VAP [16], while those with gastrointestinal disorders were excluded. All VAP patients underwent gut microbiota analysis, and their gut metagenome-derived draft genomes were compared with respective pulmonary isolate genomes using FastANI (https://github.com/ParBLiSS/FastANI). Only pathogens with ANI values ≥ 99% were selected for phenotypic experiments, those with ANI values < 99% were excluded. Gut microbiota datasets for healthy Chinese controls were obtained from the European Bioinformatics Institute (ERP005860) [17]. Control participants, age-matched to the VAP patients, were excluded if they had hypertension, diabetes, obesity, metabolic syndrome, inflammatory bowel disease, nonalcoholic fatty liver disease, celiac disease, cancer, or had received antibiotics/probiotics within 8 weeks of enrollment. Detailed participant information is provided in Table S1.

Isolation of pulmonary bacteria, nanopore sequencing and analysis

Endotracheal aspirate (ETA) samples were collected and stored at 4℃. MacConkey Agar Medium (HopeBio, HB8458, China) was used for pathogen isolation. Genomic DNA was extracted using the Bacterial Genome DNA Kit (CWbiotech, CW0552, China) following the manufacturer’s instructions. DNA was then sent to Biomarker Technologies (Beijing, China) for sequencing on the Nanopore PromethION48 platform. Approximately 1 µg of DNA was employed for library construction with the Rapid Barcoding Sequencing Kit (Oxford Nanopore Technologies, SQK-RBK004, UK). Indexed libraries were loaded onto the flow cell and sequenced. Raw reads were base-called using guppy [18] and demultiplexed with qcat v. 1.1.0 (https://github.com/nanoporetech/qcat). Filtered subreads were assembled with Canu v1.5 [19], followed by correction using Racon v3.4.3 [20]. Circularization and adjustment of starting points were performed with Circlator v1.5.5 [21], and final error correction was done using Pilon v1.22 [22]. The resulting high-accuracy genome was used for subsequent analysis.

Stool genomic DNA extraction

Fecal samples were collected by ICU physicians and immediately stored at -80 °C by assistants for further experiments. DNA was extracted from approximately 100 mg of fecal material using the QIAamp Fast DNA Stool Mini Kit (QIAGEN, 51604, Germany) following the manufacturer’s instructions. DNA quality and integrity were assessed by 1.0% agarose gel electrophoresis, and concentrations were measured using a NanoPhotometer (IMPLEN, NanoPhotometer N60/N50, Germany).

Metagenomic sequencing and taxonomic profiling

Approximately 1 µg of DNA per fecal sample was used for library construction, followed by paired-end sequencing on an Illumina NovaSeq 6000 platform (PE150) at Novogene Bioinformatics Technology Co., Ltd. (Beijing, China). Raw sequencing data were trimmed, and human reads (hg38 human reference genome: http://hgdownload.cse.ucsc.edu/goldenPath/hg38/bigZips/hg38.fa.gz) were removed using Bowtie2 v2.4.4 (http://bowtie-bio.sourceforge.net/bowtie2/index.shtml). High-quality sequences were then assembled using Megahit (https://github.com/voutcn/megahit) [23]. Taxonomic profiling was performed using MetaPhlAn 3.0 [24] to assess the relative abundance of microbial species. A total of 258 bacterial species were identified in VAP patients, and 188 in healthy individuals (Table S2).

ARG annotation

For pathogens, ARG annotation was performed using CARD [25], a curated resource for AMR gene identification. Genomic sequences were analyzed using the Resistance Gene Identifier (RGI) tool provided by CARD, which utilizes the curated Antibiotic Resistance Ontology to predict ARGs based on sequence homology and protein family classification. Default parameters were applied to ensure accurate annotation, and only hits with strict or perfect matches to reference genes were retained. For gut metagenomes, the ARGs were determined using the ARGs-OAP v2.0 pipeline [26, 27]. As stated in the reference, the formula for calculating ARG abundance (copies of ARGs/16S rRNA gene) is as follows:

$$\begin{aligned}&ARG\,abundance\cr&\quad=\sum\nolimits_1^n\frac{\frac{N_{ARG-like\,sequence}\times{L}_{reads}}{L_{ARG\,reference\,sequence}}}{\frac{N_{ARG-like\,sequence}\times{L}_{reads}}{L_{16\text{S}\,sequence}}}\end{aligned}$$

Bacterial sampling and detection in the ward environment

Sterile cotton swabs, pre-moistened with pre-cooled sterile 1× PBS, were used to sample surgical instruments, countertops, floors, and doorknobs within the ward. Each swab was then placed into 200 µl of pre-cooled 1 × PBS in a 1.5 mL tube, ensuring full immersion. The resulting solution was spread onto LB agar plates and incubated at 37 °C for 24 h. After incubation, colony growth was assessed to determine the presence of pathogenic bacteria.

Assembling draft genomes based on reference genomes, and average nucleotide identity (ANI) estimate

The pulmonary pathogen genomes were established as reference genomes to extract corresponding draft genomes from gut microbiota. Indexing of reference genomes and alignment of fecal metagenomic data reads to these reference genomes were conducted with Bowtie2 v2.4.4. The aligned reads were assembled using SPAdes v3.13.1 (https://github.com/ablab/spades). Pairwise comparisons of complete pathogen genomes and gut metagenome-derived genomes were conducted using FastANI (https://github.com/ParBLiSS/FastANI). Genome mappings were visualized using an R script with the “genoPlotR” package, available in the repository (https://github.com/ParBLiSS/FastANI/tree/master/scripts).

De Novo assembly, Binning, and ANI estimate

Processed sequencing reads were de novo assembled using Megahit (https://github.com/voutcn/megahit), generating contigs with a minimum length of 200 bp. For binning, Bowtie2 (http://bowtie-bio.sourceforge.net/bowtie2/index.shtml), Samtools [28], and MetaBAT2 [29] were employed sequentially. Bowtie2 was utilized to align the reads to the assembled contigs, while Samtools converted and sorted the resulting alignment files. MetaBAT2 then partitioned the contigs into distinct bins based on coverage depth. To assess the quality and completeness of the assembled bins, checkM was applied [30]. The obtained bins were subsequently compared against reference pathogen genomes using BLASTn [31]. Finally, pairwise comparisons between complete pathogen genomes and draft genomes were performed using FastANI (https://github.com/ParBLiSS/FastANI).

Minimum inhibitory concentration (MIC) measurements

MIC measurements were conducted as previously described to assess the antibiotic resistance of translocated isolates [32]. E. coli Ec1 and B. cenocepacia Bc1 isolates were cultured overnight in 2 mL of LB broth (HopeBio, HB0128, China). The saturated cultures (20 µL) were then inoculated into 2 mL of fresh LB broth and sub-cultured to an OD of ~ 0.5 at 600 nm. The cultures were then transferred to 96-well plates containing antibiotics (kanamycin (KAN), amikacin (AMK), gentamicin (GEN), ampicillin (AMP), imipenem (IPM), cefazolin (CZO), ceftazidime (CAZ), polymyxin B (poly B), polymyxin E (poly E), chloromycetin (CM), levofloxacin (LEVO), fosfomycin (FOS), erythrocin (ERY), azithromycin (AZM), moxifloxacin (MXF), tetracycline (TET)) in log2 serial dilutions, with an initial density of 5 × 10⁵ CFUs/mL. After 24 h of incubation at 37 °C with shaking at 234 rpm, the OD600 was recorded.

Infection of galleria Mellonella larvae

G. mellonella killing assays were performed using K12 (standard E. coli K12 MG 1655), Ec1, Bc (standard B. cenocepacia ATCC BAA-245), and Bc1 strains. The K12 and Bc strains were purchased from Beijing Biosea Biotechnology co., LTD. Overnight cultures of each isolate were washed with 1 × PBS and adjusted to a concentration of 1 × 10⁷ CFUs/mL. Ten larvae, each approximately 3 cm in length, were individually injected with 10 µL of the bacterial suspension for each strain. The larvae were then incubated in a 90 cm culture dish in the dark at 37 °C for 96 h. The quantity of dead larvae was recorded every 12 h. The assay was conducted with three biological replicates.

Growth curves

Bacterial growth curves were monitored using a microplate reader (BMG Labtech, POLARstar Omega, Germany). Each well of the honeycomb microplates was loaded with 198 µL of fresh LB broth and inoculated with 2 µL of the tested strains (K12, Ec1, Bc, and Bc1 strains, 5 × 10⁸ CFUs/mL) [33]. The cultures were incubated at 37 °C with shaking at 200 rpm, and the bacterial concentration (OD600) was recorded every 0.5 h for a total of 24 h. Bacterial growth curve analysis and key metrics, including intrinsic growth rate, lag time, and maximum population density, were obtained using the Grofit package in R [34].

Animal experiments

All animal experiments followed ethical guidelines and were approved by the Institute of Radiation Medicine, Chinese Academy of Medical Sciences (Approval No. IRM/2-IACUC-2205-002). Eight-week-old male C57BL/6J mice (21 g), from Beijing Vital River Laboratory Animal Technology Co., Ltd, were acclimatized for one week and housed under standard specific-pathogen-free conditions at the Institute of Radiation Medicine. The housing conditions at 24–25 °C, 50–55% humidity, and a 12-hour light/dark cycle. For the pneumonia model, mice (n = 5 per group) were randomly assigned to four groups: Cont (Standard diet, ad libitum water for 30 days, followed by intratracheal instillation of 1 × PBS on day 31), Pa (Standard diet and water ad libitum for 30 days, followed by intratracheal instillation of P. aeruginosa (P8 W, 5 × 10⁷ CFUs) on day 31), Ec (Standard diet and water ad libitum for 30 days, with Ec1 gavage every other day, followed by intratracheal instillation of 1 × PBS on day 31), and EcPa (Standard diet and water ad libitum for 30 days, with Ec1 gavage every other day, followed by P. aeruginosa instillation on day 31). To minimize contamination, oral cavity of each mouse was carefully cleaned with a fine 75% alcohol-soaked cotton swab prior to the P. aeruginosa challenge. Mice were sacrificed 12 h after the instillation, and tissue samples were collected and stored at -80 °C for further analysis. For the colitis model, mice (n = 5 per group) were assigned to Control (Standard diet and water ad libitum for 7 days) and 5% DSS (Standard diet and water ad libitum added with 5% DSS for 7 days) groups. After euthanasia, the Mesenteric Lymph Nodes (MLNs) and lung tissues were immediately homogenized for bacterial plating and cultivation. The remaining tissue samples were collected and stored at -80 °C for subsequent analysis.

Isolation of E. coli Ec1 from mouse tissues and whole-genome sequencing

After euthanasia, samples from the oral cavity, lungs and feces of mice were collected and stored in 1 × PBS solution. Ec1 (ampicillin-resistant E. coli strain) was selectively isolated using E. coli chromogenic medium (Hopebio, HB7001, China) with 100 mg/L ampicillin. Representative isolates underwent DNA extraction and purification using a Bacteria genomic DNA kit (CWbiotech, CW0552, China) following the manufacturer’s instructions. A total of 1 µg of DNA was used for whole-genome sequencing at Biomarker Technologies. For whole-genome comparisons of isolates from mouse tissues, FastANI (https://github.com/ParBLiSS/FastANI) was applied.

Histological analysis

Colon and lung tissues were fixed in 4% paraformaldehyde, embedded in paraffin, and sectioned to 5 μm. The sections were stained with Hematoxylin and Eosin (H&E) (Solarbio, G1120, China). The colon histological damage score was based on evaluations of crypt architecture, inflammatory cell infiltration, muscle thickening, goblet cell depletion, and crypt abscesses [35]. Total lung damage score was the sum of scores for each criterion (lung injury areas, collapsed alveoli, hyperplasia and metaplasia of bronchial epithelial cells, mucus plugs defined as mucus material in bronchi and total inflammatory cells) [36].

Flow cytometry

For Th17 cell staining, lymphocytes isolated from the spleen were incubated with a cell stimulation cocktail (plus protein transport inhibitors) (Thermo Fisher Scientific, 00-4975-93, USA) at 37 °C and 5% CO2 for 6 h. Surface staining was performed using FITC-anti-mouse CD3 (Thermo Fisher Scientific, 11-0032-80, USA), Percp-anti-mouse CD4 (BioLegend, 100537, USA), and Percp-anti-mouse CD45 (BioLegend, 103129, USA). For intracellular staining, cells were fixed and permeabilized with fixation/permeabilization solution (BD Biosciences, 554722, USA), followed by staining with APC-anti-mouse IL-17 A (BioLegend, 506915, USA). For Treg cell staining, lymphocytes were isolated from the spleen, MLNs and colon lamina propria. Surface markers FITC-anti-mouse CD3 (Thermo Fisher Scientific, 11-0032-80, USA), Percp-anti-mouse CD4 (BioLegend, 100537, USA), and APC-anti-mouse CD45 (BioLegend, 147707, USA) were used, along with the intranuclear staining marker PE-anti-mouse Foxp3 (Miltenyi Biotec, 130-111-678, Germany). Stained cells were then analyzed for proportions.

Statistical analyses

Statistical analyses were performed using R v4.1.0 (R Foundation for Statistical Computing, Austria, https://www.R-project.org/. ) and GraphPad Prism v8.2.1.441 (GraphPad Software, USA). The Shannon and Simpson indices were calculated using the vegan R package [37]. Principal Coordinates Analysis (PCoA) based on the weighted UniFrac distance or Bray‒Curtis dissimilarity was performed and visualized using the vegan [37], ade4 [38], and ggplot2 [39] R packages. The ANOSIM function in the vegan package [37], with 999 permutations, was employed to assess statistical significance. Heatmaps were generated using the ggplot2 R package [39]. Procrustes analysis was performed using the vegan [37] and ggplot2 [39] packages. Distinguishing biomarkers between groups were identified using the LEfSe analysis web tool (http://huttenhower.sph.harvard.edu/galaxy).

Results

Characteristics of pathogenic strains isolated from ETA samples

Two strains, E. coli (Ec1) and B. cenocepacia (Bc1), were isolated from endotracheal aspirate (ETA) samples from two out of seven VAP patients in ICU. Subsequent findings identified these strains as translocation bacteria. Antibiotic resistance testing revealed that Ec1 exhibited the highest resistance to ampicillin, imipenem, cefazolin, and azithromycin (Fig. 1A). Whole-genome sequencing of Ec1 identified 13 antibiotic resistance genes (ARGs), including genes for efflux pumps, β-lactamases, and membrane-repair mechanisms, contributing to resistance to multiple antibiotic classes (Fig. 1B). Additionally, Ec1 contained 62 potential virulence genes related to adhesion, invasion, and host survival, such as genes for flagella, pilus, capsule, and siderophores (Fig. 1C). Infection of G. mellonella larvae with Ec1 resulted in a survival rate of 41.7%, significantly lower than the 63.6% survival rate of E. coli K12 (Fig. 1D). Ec1 also showed faster growth and higher maximum population density than K12, with no significant difference in lag time and biofilm formation (Fig. 1E-I). Ec1 harbored four plasmids (Fig. S1).

For Bc1, the highest resistance was observed to ampicillin, imipenem, cefazolin, and ceftazidime (Fig. 1A). Whole-genome analysis identified efflux pump genes and β-lactamase genes (Fig. 1B), as well as virulence factors such as genes for flagella, capsule, and quorum-sensing systems (Fig. 1C). G. mellonella larvae infected with Bc1 exhibited a survival rate of 18.2%, which was lower than the 25.4% survival rate for standard B. cenocepacia (Bc) (Fig. 1D). Growth analysis showed Bc1 had a shorter lag phase and higher maximum biomass compared to Bc, although the growth rate was not significantly different (Fig. 1E-H). Biofilm formation between Bc1 and Bc was also not significantly different (Fig. 1I). Bc1 contained two plasmids (Fig. S2).

Pathogens isolated from the other five patients included P. aeruginosa (Pa1), Enterococcus faecium (Ef1), Klebsiella pneumoniae (Kp1), and B. cenocepacia (Bc2, Bc3), with their characteristics shown in Fig. S3. Comparisons of their genomes with draft genomes from their respective gut metagenomes revealed ANI values below 99%, suggesting that these isolates were likely not to be translocated bacteria.

Fig. 1

Characteristics of Ec1 and Bc1 strains isolated from ETA samples of two VAP patients. (A) MIC values for the Ec1 and Bc1 strains against various antibiotics are represented by solid circles. Aminoglycosides: kanamycin (KAN), amikacin (AMK), gentamicin (GEN); β_lactams: ampicillin (AMP), imipenem (IPM), cefazolin (CZO), ceftazidime (CAZ); Polypeptides: polymyxin B (poly B), polymyxin E (poly E); Chloramphenicols: chloromycetin (CM); Fluoroquinolones: levofloxacin (LEVO); Phosphonate: fosfomycin (FOS); Macrolides: erythrocin (ERY), azithromycin (AZM); quinolone: moxifloxacin (MXF); Tetracyclines: tetracycline (TET). (B) ARG profiles of Ec1 and Bc1 strains and their resistance to the antibiotic classes (orange present, azure absent). (C) Binary heatmap showing the presence (orange) or absence (azure) of virulence genes in Ec1 and Bc1 strains. (D) Survival curves of G. mellonella larvae infected with K12, Ec1, Bc and Bc1 strains for 96 h. The blue, green, orange, and purple curves represent the survival rates of larvae infected with K12, Ec1, Bc and Bc1, respectively. (E) Growth curves of K12 (blue), Ec1 (green), Bc (orange), and Bc1 (purple) strains, with eight replicates per strain (n = 8). (F) Growth rate differences among K12, Ec1, Bc and Bc1 strains, calculated using the R package (Grofit) from the growth curve data. (G) Box plot illustrating the lag time (h) during growth for K12, Ec1, Bc, and Bc1 strains. (H) Box plot illustrating the maximum optical density (OD600) reached during growth for K12, Ec1, Bc, and Bc1 strains. (I) Box plot displaying biofilm formation (OD570) of K12, Ec1, Bc and Bc1 strains, with eight replicates per strain (n = 8). Statistics by the unpaired two-tailed Student’s t test. Values shown are the mean ± SD. *** p < 0.001. ns: no significance. K12: standard E. coli K12 MG1655, Ec1: translocated E. coli, Bc: standard B. cenocepacia, Bc1: translocated B. cenocepacia

Characteristics of gut microbiota in VAP patients

We analyzed the gut microbiota composition in seven VAP patients at multiple taxonomic levels. Healthy controls displayed a higher proportion of Bacteroidetes (48.32-83.01%) and a lower proportion of Firmicutes (15.00-48.38%). In contrast, VAP patients exhibited distinct microbiota patterns. Patients 1, 2, and 3 had microbiota compositions dominated by > 98% Firmicutes, while patients 4 and 7 exhibited higher proportions of Bacteroidetes (Fig. 2A). At the family level, Bacteroideaceae predominated in healthy controls (40.66-79.52%), whereas VAP patients showed greater variability. Enterococcaceae were most abundant in patients 1, 2, and 6, while patient 3’s microbiota was almost entirely composed of Leuconostocaceae (Fig. 2B). At the genus level, Enterococcus was abundant in several patients, with Weissella dominating in patient 3 (Fig. 2C). Species-level analysis revealed greater inter-patient diversity, but E. faecium was predominant in patients 1, 2, and 6 (Fig. S4).

Fig. 2

High interindividual diversity in fecal microbiota composition at the phylum (A), family (B) and genus (C) level in VAP patients. Each bar represents the microbiota composition of one individual. The data are presented as the percentage of bacterial relative abundance in each sample

We also performed statistical comparisons between the gut microbiota of VAP patients and healthy controls. α-diversity analysis revealed significant differences, with VAP patients showing a reduced Shannon index (Fig. 3A). PCoA and ANOSIM confirmed distinct β-diversity between VAP patients and controls (Fig. 3B and C, R = 0.778, p < 0.001). At the phylum level, VAP patients exhibited an increased in Firmicutes and a decreased in Bacteroidetes, leading to a 4.43-fold increase in the Firmicutes/Bacteroidetes ratio (Fig. 3D). At the family level, Enterococcaceae were enriched, while Bacteroidaceae declined (Fig. 3E). Enterococcus abundance was higher at the genus level, with reductions in Bacteroides and short-chain fatty acid-producing genera such as Roseburia and Faecalibacterium (Fig. 3F). LEfSe analysis further highlighted these microbiota alterations (Fig. 3G). Additionally, pathogen species and abundance were elevated in VAP patients, with notable increases in E. faecium and other four pathogens (Fig. 3H-J). Overall, VAP patients showed gut microbiota alterations, characterized by a depletion of beneficial bacteria and an overgrowth of pathogenic species.

Fig. 3

Gut microbiota composition in VAP patients compared to healthy controls. (A) Box plot showing differences in gut microbial α-diversity, measured by the Shannon index, between VAP patients and healthy controls. (B) PCoA plot illustrating the distribution of gut microbiota composition based on β-diversity using weighted UniFrac distances. (C) ANOSIM analysis indicating distinct separation in gut microbiota composition between VAP patients and healthy controls (R = 0.778, p < 0.001). (D) Stacked bar plot of the average relative abundance of bacterial phyla in fecal samples from VAP patients and healthy controls. (E) Box plot comparing the relative abundance (%) of Enterococcaceae and Bacteroidaceae between the two groups. (F) Stacked bar plot depicting the relative abundance (%) of the top 20 bacterial genera in VAP patients and healthy controls. (G) Cladogram from LEfSe analysis highlighting microbial taxa enriched in VAP patients (green nodes) and healthy controls (red nodes). Yellow nodes represent taxa with no significant differences between groups. (H) Venn diagram of pathogenic species in fecal samples, showing 47 species in VAP patients, 39 species in healthy controls, and 25 species shared between the groups. (I) Box plot comparing the relative abundance (%) of pathogenic bacteria in the gut between VAP patients and healthy controls. (J) Heatmap of significantly different pathogenic bacteria between VAP patients and healthy controls across individual samples. Statistical analysis was performed using an unpaired two-tailed Student’s t test. Data are presented as mean ± SD. ** p < 0.01, *** p < 0.001

Gut resistome in VAP patients

Therapeutic antibiotics use affects the antibiotic resistance of intestinal flora. Analysis of fecal metagenome datasets identified 294 ARG subtypes in VAP patients (Table S3). While the number of ARG subtypes didn’t differ significantly between VAP patients (82–195 subtypes) and healthy controls (75–162 subtypes) (Fig. 4A), the relative abundance of ARGs, measured as copies per 16 S rRNA gene, was markedly higher in VAP patients (0.06–4.91) compared to healthy controls (0.10–0.88) (Fig. 4B). Shannon diversity indices indicated comparable α-diversity between the two groups (VAP: 2.17–2.98, controls: 1.77–3.80, Fig. 4C). However, PCoA based on Bray-Curtis dissimilarity revealed significant differences in resistome between the two groups (Fig. 4D).

In VAP patients, ARGs conferring resistance to bleomycin, rifamycin, fosfomycin, vancomycin, bacitracin, aminoglycosides, multidrug, β-lactam, macrolide-lincosamide-streptogramin (MLS), tetracycline, and chloramphenicol classes were significantly enriched (Fig. 4E, Fig. S5, and Table S4). Notably, Tet(Q/W/O/X2) and Erm(B) genes were present in higher abundance compared to controls (Fig. 4F).

Fig. 4

Variations in the gut resistome of VAP patients compared to healthy controls. (A) Comparison of ARG subtype species between VAP patients and healthy controls. No significant difference was observed between the two groups (p = 0.6308). (B) The total relative abundance of ARGs was significantly higher in the gut of VAP patients compared to healthy controls, as indicated by the copies of ARGs/16S rRNA gene. (C) Box plot illustrating the Shannon index for ARGs in the gut microbiota of VAP patients and healthy controls. Although no significant difference was detected (p = 0.1956), a trend toward lower diversity was observed in patients. (D) PCoA plot based on Bray-Curtis dissimilarity, displaying the distribution of ARG profiles in the gut microbiota of VAP patients and healthy controls. (E) Heatmap presenting the profiles of different ARG types, assessed by log¹⁰ ARGs/16S rRNA, highlighting distinct antibiotic resistance patterns between VAP patients and healthy controls. Statistical differences are marked in red next to each ARG type. (F) Histogram showing linear discriminant analysis (LDA) scores from LEfSe analysis of ARGs in the gut microbiome. The LDA score is greater than 4.0. Statistics by the unpaired two-tailed Student’s t test. Values shown are the mean ± SD. ** p < 0.01. *** p < 0.001. ns: no significance

Taxonomic annotation of ARG-carrying contigs revealed Firmicutes, Proteobacteria, Actinobacteria, and Bacteroidetes as the primary reservoirs among 345 ARG-carrying contigs. Firmicutes accounted for 57% of these bacteria, predominantly carrying resistance genes for multidrug, MLS, aminoglycoside, tetracycline, and vancomycin classes (Fig. S6). Procrustes analysis revealed a significant correlation between ARGs and gut pathogens (Table S5) in VAP patients (M² = 0.774, p < 0.001, 9999 permutations) (Fig. 5A). Spearman correlation analysis further revealed positive associations between ARGs and 25 pathogens, including members of the Enterococcus genus. E. faecium exhibited strong correlations with nine ARG types, including aminoglycoside, bacitracin, β-lactam, chloramphenicol, fosfomycin, MLS, multidrug, tetracycline, and vancomycin (Fig. 5B). These findings underscore the role of pathogenic bacteria, particularly Enterococcus species, as major reservoirs of ARGs, which may compromise the effectiveness of antibiotics in VAP patients.

Fig. 5

Correlation between gut pathogen abundance and ARG profiles. (A) Multivariate Procrustes analysis comparing the gut pathogen composition with ARG type profiles of the participants. Shorter lines between an ARG eigenvalue (triangle) and its corresponding pathogen eigenvalue (round) indicate minor discordance between datasets for that sample. Significant correlations (p < 0.001, 9999 permutations) were detected. (B) Heatmap illustrating Spearman correlation analysis, representing statistically significant correlations between gut pathogens and ARG types. Blue squares indicate significant positive correlations (r > 0.5, p < 0.05), while orange squares indicate significant negative correlations (r < − 0.5, p < 0.05). * p < 0.05. ** p < 0.01. *** p < 0.001

Gut-to-lung translocation of pathogenic bacteria in VAP patients

The ICU ward operates under positive pressure and undergoes twice-daily disinfection, with no pathogenic bacteria detected in the environment (e.g., surgical instruments, countertops, floors, and doorknobs). Stringent precautions were implemented to minimize procedural contamination. Based on these controls, we hypothesized that bacterial translocation via the gut-lung axis contributed to VAP. To test this hypothesis, we attempted to isolated identical strains from patient feces using antibiotic resistance markers but were unsuccessful. Instead, we utilized gut metagenome sequencing datasets to assemble bacterial genomes and performed whole-genome ANI analysis to compare bacterial genomes between the lungs and gut. ANI values greater than 99% between two genomes indicate a common ancestor, as previously reported [40]. Our comparisons revealed genetically identical strains between the lungs and the gut. Using reference-based methods to obtain draft genomes, the ANI values were 99.96% for E. coli and 99.41% for B. cenocepacia (Fig. 6A and B). When draft genomes were obtained through de novo assembly and binning, the ANI values were 99.92% for E. coli and 99.30% for B. cenocepacia (Fig. 6C and D). These findings suggest that gut bacteria migrate to the lungs and contribute to the onset of VAP in these patients.

Fig. 6

ANI analysis of pulmonary isolate genomes and metagenome-derived genomes. The upper layer represents the assembled draft genomes, while the lower layer displays the reference genomes of the isolated strains. (A) The ANI value between the Ec1 genome and its corresponding assembled draft genome obtained utilizing reference genome-based methods is 99.96%. (B) The ANI value between the Bc1 genome and its corresponding assembled draft genome obtained utilizing reference genome-based methods is 99.41%. (C) The ANI value between the Ec1 genome and its corresponding assembled draft genome based on de novo assembly and binning is 99.92%. (D) The ANI value between the Bc1 genome and its corresponding assembled draft genome based on de novo assembly and binning is 99.30%. Ec1: translocated E. coli, Bc1: translocated B. cenocepacia

Gut-to-lung translocation of pathogenic bacteria in pneumonia and colitis mouse models

To confirm gut-to-lung translocation and explore the underlying mechanisms in mice, we colonized the gut with Ec1 and then induced lung infection through intratracheal infusion of P. aeruginosa (Fig. 7A). Ec1, identifiable using E. coli chromogenic medium with 100 mg/L ampicillin, was isolated from the lungs and gut of EcPa mice but was absent in oral samples (Fig. 7B and C). These findings indicate the translocation of Ec1 from the gut to the lungs. This conclusion is further supported by comparative genomic analysis. In the EcPa group, genomes of fecal isolates (n = 3, one isolate per mouse) showed over 99.9% ANI values with the original Ec1 strain (Fig. 7D and S7). Similarly, lung isolates (n = 6, two isolates per mouse) also exhibited over 99.9% ANI values with the original Ec1 (Fig. 7E and S8). Moreover, ANI values between lung and fecal isolates exceeded 99.9% (Fig. 7F-K), providing compelling evidence for Ec1 translocation in mice with P. aeruginosa-induced pneumonia.

Fig. 7

Gut-to-lung translocation of antibiotic-resistant E. coli Ec1 in P. aeruginosa-induced pneumonia mice. (A) Experimental design and group assignments for mouse experiments. (B) Quantification of resistant E. coli Ec1 load in stool samples on days 0 and 31, cultured on E. coli chromogenic medium containing 100 mg/L ampicillin in the Cont, Ec, Pa, and EcPa groups (n = 5). (C) Quantification of resistant E. coli Ec1 load in the oral cavity and lung tissues after sacrifice, detected using E. coli chromogenic medium with 100 mg/L ampicillin across experimental groups (n = 5). (D) Representative image of ANI analysis comparing the genomes of resistant E. coli isolated from fecal samples of EcPa mice with the original Ec1 strain (n = 3). (E) Representative image of ANI analysis comparing the genomes of resistant E. coli isolated from lung tissues of EcPa mice with the original Ec1 strain (n = 6). (F-K) ANI analysis comparing resistant E. coli genomes isolated from lungs and fecal samples of EcPa mice, fecal isolates: n = 3, one isolate/mouse, lung isolates: n = 6, two isolates/mouse. (L) Representative H&E-stained lung sections from Cont, Ec, Pa, and EcPa groups. Scale bars = 50 μm (n = 5). (M) Disease activity index scores of lung injury across experimental groups (n = 5). (N) Histological score of lung injury based on H&E staining (n = 5). (O) Quantification of resistant E. coli Ec1 load in the MLNs after sacrifice, detected using E. coli chromogenic medium with 100 mg/L ampicillin across groups (n = 5). Statistics by one-way ANOVA with Tukey’s post-hoc test. Values shown are the mean ± SD. * p < 0.05. ** p < 0.01. Ec1: translocated E. coli

To assess the impact of Ec1 colonization and translocation on lung injury, we performed H&E staining of lung tissues, which revealed more pronounced inflammatory infiltrates in the alveolar areas of EcPa mice compared to controls (Fig. 7L). Disease severity, evaluated using the disease activity index (Fig. 7M) and histological score (Fig. 7N), was higher in the EcPa group, suggesting that Ec1 colonization and translocation exacerbates lung inflammation. In the Ec group, slight lung damage was observed, possibly due to the translocation of pathogen-associated molecular patterns or immune cells, resulting in minor inflammation. Notably, antibiotic-resistant E. coli was cultured from the mesenteric lymph nodes (MLNs), while blood cultures remained negative, suggesting that gut bacteria may translocate via the lymphatic system (Fig. 7O). In the Pa group, we observed a reduction in Treg cell populations in the spleen, MLNs, and colon lamina propria, alongside an increase in Th17 cell populations in the spleen compared to control (Fig. 8A-D), reflecting an imbalanced immune response induced by pneumonia. These immune disturbances were even more pronounced in the EcPa group (Fig. 8A-D), suggesting that Ec1 colonization and translocation further exacerbated immune dysregulation. Histological analysis of the colon revealed shorter crypts and more severe damage in the EcPa group (Fig. 8E-H), indicating compromised gut barrier integrity and potential bacterial translocation. Mild colon injury was also observed in the Pa group, likely due to systemic dissemination of P. aeruginosa.

Fig. 8

Immunological changes and gut integrity in P. aeruginosa-induced lung infection mouse. (A-C) Representative flow cytometry plots and proportions of Tregs identified as FoxP3⁺ within CD3⁺CD4⁺CD45⁺ T cells in the spleen (A), MLNs (B), and colonic lamina propria (C) across experimental groups (n = 3). (D) Representative flow cytometry plots and proportions of Th17 cells identified as IL-17 A⁺ within CD3⁺CD4⁺CD45⁺ cells in the spleen across groups (n = 3). (E) Representative H&E-stained colon sections from Cont, Ec, Pa, and EcPa groups. Scale bars = 50 μm (n = 5). (F) Disease activity index scores indicating colon injury across groups (n = 5). (G) Histological score of colon injury from H&E-stained sections (n = 5). (H) Submucosa thickness (µm) of the colon in each group (n = 5). Statistics by one-way ANOVA with Tukey’s post-hoc test. Values shown are the mean ± SD. * p < 0.05. ** p < 0.01

To further investigate primary lung infection resulting from gut-to-lung translocation, we utilized a 5% DSS-induced colitis model [41]. DSS treatment led to crypt disruption and thickening of the muscularis mucosa and submucosa, confirming the successful establishment of colitis model (Fig. 9A). In DSS-treated mice, bacterial cultures from the MLNs were positive, whereas no cultures were detected in control mice (Fig. 9B). Blood cultures from both groups remained negative, indicating that the lymphatic system may also act as a pathway for gut-to-lung translocation in colitis mice. 16 S rRNA sequencing of lung tissues revealed no significant changes in α-diversity, but the Shannon index was decreased, and the Simpson index was increased in DSS-treated mice, indicating lung microbial dysbiosis (Fig. 9C and D). β-diversity analysis based on Bray-Curtis distances showed significant differences in lung microbiota composition between DSS-treated and control groups (Fig. 9E). Notably, gut-associated bacteria, Enterobacteriaceae and Escherichia-Shigella, were significantly enriched in the lung tissues of DSS-treated mice (Fig. 9F and G). Positive bacterial cultures also detected in lung samples (Fig. 9H). Histological analysis of lung tissues demonstrated abscess formation, increased histological scores, and thickened alveolar septa (Fig. 9I). Additionally, total protein levels in bronchoalveolar lavage fluid (BALF) were significantly elevated in DSS-treated mice compared to controls (Fig. 9J). These findings suggest that Enterobacteriaceae and Escherichia-Shigella translocate to the lungs via lymphatic circulation in DSS-induced colitis mice, contributing to the initiation of lung inflammation.

Fig. 9

Gut-to-lung translocation via the lymphatic system in DSS-induced colitis mice. (A) H&E staining of the distal colon to assess inflammation severity, with histological score, muscularis mucosae thickness, and submucosal thickness measurements. Scale bars = 50 μm (n = 5). (B) CFUs cultured from MLNs of Control and 5% DSS-treated mice (n = 5). (C, D) Box plots showing the α-diversity of lung microbiota in Control and 5% DSS groups, assessed using the Shannon (C) and Simpson indices (D) (n = 3). (E) PCoA plot displaying β-diversity of lung microbiota based on Bray-Curtis distances (n = 3). (F, G) Stacked bar plots presenting average relative abundance of bacteria taxa at the family (F) and genus (G) levels in lung microbiota from Control and 5% DSS-treated mice (n = 3). (H) CFUs cultured from whole lung samples of Control and 5% DSS-treated mice (n = 5). (I) H&E staining of lung sections to assess inflammation severity, with histological score and alveolar septum thickness measurements. Scale bars = 50 μm (n = 5). (J) Airway inflammation assessed by total protein concentration in bronchoalveolar lavage fluid (n = 5). Statistics by one-way ANOVA with Tukey’s post-hoc test. Values shown are the mean ± SEM. * p < 0.05. ** p < 0.01. *** p < 0.001

Discussion

VAP is one of the most prevalent in ICU-acquired infections and remains a major contributor to hospital-acquired infection-related mortality, drawing considerable attention [2]. However, methodological limitations have hindered a comprehensive understanding of the origins of VAP-associated pathogens. Traditional bacterial isolation from complex gut microbial communities is labor-intensive and often impractical due to the lack of effective screening markers for translocated bacteria. Additionally, conventional microbiome analysis methods frequently fail to resolve strain-level critical phenotypic traits of pulmonary pathogens. To address these challenges, our study integrates culture-based isolation with whole-genome comparison approaches, linking pulmonary isolate genomes with gut metagenome-derived bacterial genomes. This approach enables the identification of gut-derived bacteria in the lungs without extensive culturing and allows for strain-level characterization of translocated pathogens. By providing detailed insights into their antibiotic resistance and virulence potential, our findings advance the understanding of bacterial translocation along the gut-lung axis and its role in VAP pathogenesis.

The ICU environment where patients were treated adhered to stringent infection control protocols. A positive-pressure ventilation system was designed to prevent the ingress of external airborne contaminants [42]. Regular and thorough disinfection procedures were implemented twice a day, encompassing patient areas, medical equipment, and surrounding environments. Environmental samples collected from surgical instruments, countertops, floors, and doorknobs tested negative for pathogens, effectively ruling out external contamination as a source of infection. Additionally, rigorous measures were taken to prevent procedural contamination. ICU medical staff strictly followed hand hygiene protocols and regularly change diapers and bed linens to prevent contamination by fecal microorganisms. Comprehensive oral care protocols, including the use of antiseptic mouthwash and plaque removal, effectively reduced the bacterial load in the oral cavity [43, 44]. Intubation was performed via a small neck incision, thereby bypassing the oral and pharyngeal regions where bacterial colonization is common [45]. This approach, combined with patient positioning practices such as head elevation [46], further mitigated the risk of aspiration or migration of oral or gastrointestinal bacteria into the lungs. Although gastroesophageal reflux was considered, the endotracheal tube cuff in the intubation device seals the trachea, preventing the aspiration of gastric contents, food particles, saliva, and other substances into the airway, thereby reducing the risk of aspiration pneumonia [47]. Consequently, pulmonary infections observed in this study could not be attributed to environment contamination or aspiration of oral or gastrointestinal flora. Instead, our results suggest the direct migration of bacteria from the gut into the lungs.

From the ETA samples of seven patients, we isolated the dominant strain in each case. Among these, only E. coli Ec1 (isolated from patient 7) and B. cenocepacia Bc1 (isolated from patient 2) were identified as translocated bacteria through comparative genomic analysis, establishing them as key contributors to the development of VAP. Comparative genomics analysis of the remaining five isolate genomes with draft genomes yielded ANI values below 99%, likely due to insufficient coverage of the gut metagenomes [48]. Assembling fecal metagenomic-derived genomes based on reference genomes introduces potential confirmation bias, as only reads aligning exclusively to the ETA isolates are selected, omitting non-matching gene regions. And unlike de novo-assembled MAGs, using pulmonary genomes as reference genomes results in the alignment of a substantial number of identical reads from fecal metagenomic to reference genome, whether it’s the ancestor of this strain or not. Results from the mouse pneumonia model indicated that gut-to-lung bacterial translocation persisted following pneumonia onset, underscoring the need for physicians to consider ongoing translocation when devising therapeutic strategies for VAP. This study specifically verified the translocation of Ec1 from the gut to the lungs. Future investigations will focus on the translocation of Bc1 in mice, aiming to elucidate deeper mechanisms, including lung signaling interactions and shifts in microbial composition. In colitis models, we observed that gut-associated bacteria, Enterobacteriaceae and Escherichia-Shigella, appeared to translocate to the lungs via lymphatic circulation, potentially contributing to lung infections. However, the positive bacterial cultures in the lungs cannot be conclusively attributed to gut translocation. An alternative hypothesis is that these bacteria may naturally exist in the lungs at low, undetectable levels under normal conditions and proliferate in response to disease-induced changes. These findings highlight the intricate microbial interactions along the gut-lung axis and emphasize the importance of further research to clarify the mechanisms underlying these dynamics.

Although our study demonstrated that gut bacteria translocate to the lungs via lymphatic circulation in pneumonia and colitis mouse models, the precise mechanisms and routes of bacterial translocation in VAP patients remain unclear. Future investigations should focus on elucidating whether blood circulation, lymphatic drainage, or a combination of both serves as the primary route in the translocation process.

E. coli and B. cenocepacia have been recognized as pathogens associated with pneumonia [12, 49]. Previous studies have reported E. coli strains linked to pneumonia exhibiting resistance to multiple antibiotics, including ampicillin, levofloxacin, amoxicillin, ceftriaxone, imipenem, ciprofloxacin, cefotaxime, ceftazidime, gentamicin, amikacin, ofloxacin, piperacillin, and colistin [50,51,52,53]. In comparison, the Ec1 strain demonstrated increased resistance to cefazolin but remained susceptible to gentamicin, amikacin, and kanamycin. The Burkholderia genus is known for its intrinsic resistance to polymyxin, attributed to modified lipopolysaccharides [54]. Additionally, certain Burkholderia species exhibit resistance to fluoroquinolone and aminoglycoside due to alterations in DNA gyrase and the activity of efflux pumps [55]. Genomic analysis of B. cenocepacia has previously identified β-lactamase resistance genes (PenB, AmpR, AmpC, and AmpD) [56], and multidrug efflux pump genes (AmrAB-OprA, BpeAB-OprB, and BpeEF-OprC) [55]. Notably, the Bc1 strain in our study displayed a distinct ARG profile, encompassing macrolide (AmrA, AmrB), β-lactam (BlaOXA-192), tetracycline (AdeF), chloramphenicol (CeoA, CeoB), fluoroquinolone (AdeF, AmrA, AmrB, CeoA, CeoB), and aminoglycoside (AmrA, AmrB, ceoA, ceoB) resistance genes. These results highlight the substantial antibiotic resistance observed in B. cenocepacia and underscore the importance of considering these resistance profiles during the initial stages of empirical therapy.

Metagenomic analysis revealed that the gut microbiota composition of VAP patients was highly individualized, consistent with previous study on ICU patients [57]. However, an increased abundance of pathogenic bacteria was observed in these patients. This pathogen is of particular concern due to its association with high burdens of ARGs and its ability to produce toxins, which can complicate clinical management and worsen outcomes in VAP patients [58]. Interestingly, the translocated bacteria identified in our study were not the predominant members of the gut microbiota. This observation contrasts with previous research, which identified dominant gut species, K. pneumoniae and E. faecium, as translocated pathogens responsible for bloodstream infections [59]. This divergence underscores the complexity of gut-to-lung translocation and suggests that the mechanisms underlying this process in VAP may differ from those observed in bloodstream infections. Despite these complexities, the findings of this study provide valuable insights into the gut microbiome and resistome of VAP patients, offering potential guidance for antibiotic selection and therapeutic interventions.

Despite these results, further research involving larger longitudinal sampling and mechanistic studies is needed to confirm our findings, clarify the accuracy and mechanisms of gut-to-lung translocation, and further elucidate the role of the gut microbiota in the pathogenesis of VAP.

Our study highlights the critical role of gut-to-lung bacterial translocation in the development of lung infections among VAP patients. For the first time, we have identified the translocation of two specific species, E. coli and B. cenocepacia, from the gut to the lungs. These findings provide valuable insights that can inform the optimization of antibiotic therapy and the development of alternative therapeutic strategies targeting the gut microbiota for the treatment of VAP in ICU patients.

Conclusion

This study underscores the pivotal role of gut-to-lung bacterial translocation in the pathogenesis of VAP. Utilizing whole-genome comparison techniques, we successfully traced the migration of E. coli and B. cenocepacia from the gut to the lungs. These findings not only deepen our understanding of VAP etiology but also facilitate the development of more targeted and effective therapeutic interventions.