Introduction

Acinetobacter baumannii (A. baumannii) is one of the common opportunistic pathogens that can cause widespread and severe nosocomial infections, which are especially fatal in immunocompromised patients who have underlying diseases or have undergone major surgery1. A. baumannii can exert virulence through membrane porin2, capsular polysaccharide3, phospholipase4, outer membrane vesicles5, iron acquisition systems6 and protein secretion systems7.

The protein secretion system is a crucial virulence factor in A. baumannii. Currently, four types of secretion systems have been found in A. baumannii: type I secretion system (T1SS), type II secretion system (T2SS), type V secretion system (T5SS), and type VI secretion system (T6SS)8. These secretion systems not only transport virulence factors to the surrounding environment or directly transfer them to adjacent cells, 9but also hold significant importance in antibacterial resistance10. Among these, the T6SS is capable of directly injecting toxic effector molecules into the cells of both prokaryotic and eukaryotic hosts11, thereby playing a crucial role in interactions between bacteria and their hosts12. The structure of the T6SS resembles a contractile bacteriophage tail device. Hemolysin co-regulated protein (Hcp), valine-glycine repeat protein G (VgrG), proline-alanine-alanine-arginine (PAAR), TssB and TssC form the phage-like tail tube, tail spike and sheath components, respectively, which together form the tube-sheath complex of the T6SS. Hcp is a hexameric tubular structure characterised by helical symmetry, capable of carrying effector molecules and interacting with substrates13. VgrG, a trimeric structure located at the tip of the tail tube, is sharpened by the conical PAAR protein, which facilitates the secretion of effector proteins14. TssB and TssC form a contractile sheath that provides the energy required for the transport of effector proteins15. As Hcp and VgrG are released into the extracellular environment after activation of the system, they can serve as molecular markers of T6SS activity16,17.

Metabolomics enables the comprehensive and systematic analysis of all small molecule metabolites in an organism, thereby identifying the role of specific metabolites in disease and searching for biomarkers of disease onset, progression and severity18. Metabolomics can be divided into targeted and untargeted analytical approaches. Untargeted metabolomics has been widely applied in bacterial research and has proven to be a successful research method. In terms of bacterial virulence, studies have shown that key secondary metabolites linked to virulence, including rhamnolipids, alkyl quinolones, and phenazines, are notably elevated in highly virulent strains of Pseudomonas aeruginosa. Furthermore, a predictive model for distinguishing between Pseudomonas aeruginosa strains has been developed by integrating metabolite features with machine learning19. In terms of bacterial-host interactions, Nicole Aiosa and colleagues studied the changes in the metabolome of Burkholderia thailandensis co-cultured with mammalian host cells and identified the roles of infection-related metabolites in bacterial virulence and host defence, providing new insights for in-depth analysis of bacterial-host interactions20. In addition, untargeted metabolomics can also be used to study biofilms, bacterial stress responses and environmental adaptation21,22.

However, there remains a scarcity in untargeted metabolomics research on A. baumannii. Therefore, further exploration of A. baumannii’s metabolite profile, metabolic pathways, and their associations with virulence, antibiotic resistance, and other factors holds significant academic and clinical value.

In this study, we constructed ATCC17978Δhcp, ATCC17978ΔvgrG, and ATCC17978ΔhcpΔvgrG, with the wild-type A. baumannii strain (ATCC17978) used as a control. We performed untargeted identification and differential comparison of metabolites in different A. baumannii strains using liquid chromatography-tandem mass spectrometry (LC-MS/MS) to investigate the effects of key T6SS genes hcp and vgrG on the metabolic activity of A. baumannii.

Methods

Bacterial strains, cells and growth conditions

A. baumannii strain ATCC17978 was kindly provided by Sir Run Run Shaw Hospital, affiliated with the Zhejiang University School of Medicine. The mutant strains ATCC17978Δhcp, ATCC17978ΔvgrG, and ATCC17978ΔhcpΔvgrG were constructed in our laboratory using homologous recombination technology. Human pulmonary alveolar epithelial cells (HPAEpiC) were purchased from Wuhan Saios Biotechnology Co., Ltd.

A. baumannii was cultured in Luria–Bertani (LB) broth or on blood agar plates at 37 °C. HPAEpiC were cultured in specialized epithelial cell medium supplemented with 2% fetal bovine serum (FBS) and 1% penicillin-streptomycin (100 µg/mL each) at 37 °C under 5% CO2.

Non-targeted metabolomics analysis

HPAEpiC were seeded in culture flasks and cultured in antibiotic-containing medium until reaching 80–90% confluence, followed by replacement with serum- and antibiotic-free medium. Concurrently, four strains of A. baumannii were cultured overnight in LB broth, harvested by centrifugation, and resuspended in serum- and antibiotic-free epithelial cell medium. Based on the preliminary real-time cell analysis (RTCA) experimental results from our research group (a sharp decrease in growth index after 12 h of co-culture), a bacterial suspension with a multiplicity of infection (MOI) of 100:1 was used for co-culturing with HPAEpiC for 12 h. After co-culture, the culture medium was collected and centrifuged at 2,000 rpm for 10 min to harvest a portion of the cell pellet. Subsequently, the sample was centrifuged at 13,000 rpm for 10 min to collect the bacterial pellet. Meanwhile, the co-culture supernatant was filtered through a 0.22 μm membrane to obtain it. The adherent cells were treated with trypsin and then centrifuged at 2,000 rpm for 10 min to collect the remaining cell pellet. All samples were divided into nine comparison groups (Table 1), and subjected to non-targeted metabolomics analysis by Mingke Biotechnology Co., Ltd. (Hangzhou, China).

LC-MS/MS was performed using a Waters Acquity I-Class PLUS UPLC system coupled with an Xevo G2-XS QToF high-resolution mass spectrometer, equipped with a UPLC HSS T3 column (1.8 μm, 2.1 × 100 mm). Mass spectral data were acquired in MSe mode under the control of MassLynx V4.2 software, enabling simultaneous dual-channel data acquisition at low and high collision energies. Raw data were processed using Progenesis QI software for peak extraction, alignment, and preprocessing, with metabolite annotation conducted against the METLIN online database, public databases, and a custom database.

Bioinformatic analyses were performed in R. Principal component analysis (PCA) utilized the prcomp package, while orthogonal partial least squares-discriminant analysis (OPLS-DA) was implemented via the ropls package23. Volcano plots were generated using ggplot224. Metabolites were annotated and classified based on the Human Metabolome Database (HMDB)25and pathway enrichment analysis was conducted using the Kyoto Encyclopedia of Genes and Genomes (KEGG)26,27.

For statistical analysis, univariate analysis employed Student’s t-test and fold change (FC) to evaluate the significance of differential metabolites. Multivariate analysis included unsupervised PCA and supervised OPLS-DA. Metabolites with variable importance in projection (VIP) scores > 1 and p < 0.05 in the OPLS-DA model were considered statistically significant.

Table 1 Metabolite comparison of different co-culture groups.
Full size table

S/AB/C represent the supernatant, bacterial component, and cell component from the co-culture system, respectively. WT/H/V/HV represent ATCC17978, ATCC17978Δhcp, ATCC17978ΔvgrG, and ATCC17978ΔhcpΔvgrG, respectively.

Biofilm assay

Biofilm quantification was performed using the crystal violet microplate assay28. The collected A. baumannii strains that had been co-cultured with HPAEpiC were resuscitated on blood agar plates. Fresh single colonies were then suspended in sterile saline to achieve 0.5 McFarland turbidity, followed by 1:100 dilution in LB broth. Aliquots of 200 µL were dispensed into 96-well plates with three replicates per strain and incubated statically at 37 °C for 24 h. After aspirating the bacterial cultures, wells were gently washed with PBS to remove planktonic bacteria, air-dried, and fixed with methanol for 10 min. Biofilms were stained with 200 µL of 0.1% crystal violet for 10 min, washed with PBS to remove excess stain, and air-dried again. Finally, the bound dye was solubilized with 200 µL of 95% ethanol, and absorbance was measured at 570 nm using a microplate reader to quantify biofilm formation.

Scanning electron microscopy (SEM)

The collected A. baumannii strains that had been co-cultured with HPAEpiC were individually inoculated into LB broth and cultured overnight at 37 °C with shaking at 180 rpm. Bacteria were harvested by centrifugation and washed twice with PBS. Subsequently, the samples were fixed overnight at 4 °C in a 2.5% glutaraldehyde solution and washed three times (15 min each) with 0.1 M phosphate buffer (pH 7.0). The samples were then subjected to gradient dehydration by sequential immersion in 50%, 75%, and 90% ethanol solutions for 10 min each, followed by three immersions (10 min each) in 100% anhydrous ethanol. The ethanol was then replaced by sequential immersion in 50%, 75%, and 90% t-butanol solutions for 10 min each, followed by three immersions (10 min each) in 100% t-butanol to remove residual ethanol. The processed samples were placed in small metal cups and stored at − 20 °C for later use. After freeze-drying, the samples were mounted on stubs with conductive carbon adhesive and sputter-coated with gold for 30 s using a Hitachi E1010 ion sputter. Finally, the samples were observed using a Hitachi SU8010 field-emission SEM.

Bacterial adherence assay

HPAEpiC were seeded in 24-well plates and cultured in epithelial cell medium containing dual antibiotics until reaching approximately 90% confluence, followed by replacement with serum- and antibiotic-free medium for subsequent experiments. Meanwhile, four strains of A. baumannii were individually inoculated into LB broth and cultured at 37 °C with shaking at 180 rpm for 18 h. The bacterial suspensions were adjusted to 0.5 McFarland standard, centrifuged, and resuspended in serum- and antibiotic-free medium. Based on previous RTCA (Real-Time Cell Analysis) results from our research group (showing the most significant growth index at 5 h of co-culture), HPAEpiC were co-cultured with each A. baumannii strain at an MOI of 100:1 for 5 h. After co-culture, the supernatant was discarded and cells were washed twice with PBS. Sterile distilled water was added to lyse the cells for 30 min, and the lysates were serially diluted and plated onto LB agar plates for overnight culture at 37 °C, followed by colony counting. Each experimental group was set up with three independent replicate wells. The bacterial adhesion rate was calculated using the formula: adhesion rate = (adherent bacterial count/inoculated bacterial count) ×100%.

Expression quantification by reverse transcription quantitative polymerase chain reaction (RT-qPCR)

The four strains of A. baumannii that had been co-cultured with HPAEpiC were resuscitated on blood agar plates. Subsequently, single colonies from each strain were individually inoculated into LB broth and cultured overnight at 37 °C with shaking at 180 rpm. After incubation, bacterial pellets were collected by centrifugation at 12,000 rpm for 5 min. Total RNA was then extracted using the Modified Bacterial RNA Extraction Kit (SparkJade) following the manufacturer’s instructions. The concentration of the extracted RNA was measured using the Nanodrop 2000 Nucleic Acid Analyzer (Thermo). Subsequently, cDNA was synthesized through reverse transcription using the HiScript III All-in-one RT SuperMix Perfect for qPCR kit (Vazyme) on a Model 9902 PCR machine (Applied Biosystems).

Specific primers targeting biofilm-associated genes of A. baumannii (including bap, csuA, csuB, bfmS, bfmR, pgaA, and pgaD) were designed based on sequence information obtained from the NCBI database, with 16 S rRNA serving as the reference gene (Table 2). qPCR was performed using the ChamQ SYBR Color qPCR Master Mix kit (Vazyme) on an ABI 7500 Real-Time PCR System (Applied Biosystems). Relative changes in gene expression levels were calculated using the 2−ΔΔCt method. To ensure experimental reproducibility and reliability, all procedures were performed with three biological replicates.

Table 2 The primers used in this study for detection of biofilm related genes.
Full size table

Results

Quality control data analysis

The base peak intensity (BPI) chromatograms of the seven quality control (QC) samples in this experiment show excellent overlap in peak retention times and peak areas, indicating good instrument stability (Supplementary Figure S1).

Multivariate statistical analysis of samples

Using unsupervised PCA, there was no overlap between the samples of different comparison groups, and they showed significant differences (Supplementary Figure S2). Based on these results, supervised OPLS-DA analyses were further performed. The R²Y and Q²Y values of the OPLS-DA models for each comparison group were both greater than 0.5 and close to 1, indicating good stability, fit, and predictive ability of the models. This made them suitable for screening differential metabolites and further reinforced the separation trends between the comparison groups, with more tightly concentrated distributions within each group (Supplementary Figure S3). The results of permutation tests indicated that the slope for the Q²Y fitting regression line associated with each comparison group was positive, thus validating the statistical significance of the model. Additionally, all the blue points were located above the red points, indicative of a strong independence between the training and testing datasets utilized for the modeling process (Supplementary Figure S4). In conclusion, these results collectively indicate significant metabolic differences between the two sample groups in each comparison group.

Differential metabolite screening and identification

Metabolites were annotated and classified using HMDB, and those with VIP > 1 and p < 0.05 were selected as statistically significant differential metabolites29. In the comparative classification of differential metabolites, we will exclude the unclassified part.

Supernatant comparison groups

SH vs. SWT: In positive ion mode, we identified 229 differential metabolites, including 157 upregulated and 72 downregulated (Supplementary Figure S5A). These differential metabolites were classified into 10 groups, with lipids and lipid-like molecules constituting the largest proportion (14.41%) (Supplementary Figure S6A). In negative ion mode, we identified 305 differential metabolites, including 204 upregulated and 101 downregulated (Supplementary Figure S5B). These differential metabolites were classified into 9 groups, with organic acids and derivatives constituting the largest proportion (11.80%) (Supplementary Figure S6B).

SV vs. SWT: In positive ion mode, we identified 103 differential metabolites, including 67 upregulated and 36 downregulated (Supplementary Figure S5C). These differential metabolites were classified into 9 groups, with lipids and lipid-like molecules constituting the largest proportion (15.53%) (Supplementary Figure S6C). In negative ion mode, we identified 144 differential metabolites, including 84 upregulated and 60 downregulated (Supplementary Figure S5D). These differential metabolites were classified into 9 groups, with organic acids and derivatives constituting the largest proportion (12.50%) (Supplementary Figure S6D).

SHV vs. SWT: In positive ion mode, we identified 194 differential metabolites, including 151 upregulated and 43 downregulated (Supplementary Figure S5E). These differential metabolites were classified into 10 groups, with lipids and lipid-like molecules constituting the largest proportion (11.86%) (Supplementary Figure S6E). In negative ion mode, we identified 270 differential metabolites, including 202 upregulated and 68 downregulated (Supplementary Figure S5F). These differential metabolites were classified into 9 groups, with organic acids and derivatives constituting the largest proportion (12.96%) (Supplementary Figure S6F).

Bacterial comparison groups

ABH vs. ABWT: In positive ion mode, we identified 74 differential metabolites, including 58 upregulated and 16 downregulated (Fig. 1A). These differential metabolites were classified into 8 groups, with organic acids and derivatives and organic oxygen compounds both constituting the largest proportion (13.51%) (Fig. 2A). In negative ion mode, we identified 193 differential metabolites, including 153 upregulated and 40 downregulated (Fig. 1B). These differential metabolites were classified into 9 groups, with organic acids and derivatives constituting the largest proportion (11.92%) (Fig. 2B).

Fig. 1
figure 1

Volcano plots of each comparison group in bacteria. In these plots, every point corresponds to a metabolite. The x-axis indicates the fold change in abundance across the groups being compared, while the y-axis illustrates the significance of the p-value. Larger points indicate higher VIP values. Downregulated differential metabolites are represented by blue points, upregulated differential metabolites are shown in red, and gray points signify metabolites that were identified but lacked significant differences. Additionally, five common differential metabolites were selected and labeled.

Full size image
Fig. 2
figure 2

Differential metabolite classification plots of each comparison group in bacteria.

Full size image

ABV vs. ABWT: In positive ion mode, we identified 52 differential metabolites, including 14 upregulated and 38 downregulated (Fig. 1C). These differential metabolites were classified into 7 groups, with organic oxygen compounds constituting the largest proportion (21.15%) (Fig. 2C). In negative ion mode, we identified 122 differential metabolites, including 50 upregulated and 72 downregulated (Fig. 1D). These differential metabolites were classified into 9 groups, with organic oxygen compounds constituting the largest proportion (13.11%) (Fig. 2D).

ABHV vs. ABWT: In positive ion mode, we identified 37 differential metabolites, including 21 upregulated and 16 downregulated (Fig. 1E). These differential metabolites were classified into 7 groups, with organic oxygen compounds constituting the largest proportion (21.62%) (Fig. 2E). In negative ion mode, we identified 111 differential metabolites, including 55 upregulated and 56 downregulated (Fig. 1F). These differential metabolites were classified into 9 groups, with organic acids and derivatives and organic oxygen compounds both constituting the largest proportion (11.71%) (Fig. 2F).

Cellular comparison groups

CH vs. CWT: In positive ion mode, we identified 107 differential metabolites, with 34 upregulated and 73 downregulated (Supplementary Figure S7A). These differential metabolites were classified into 9 groups, with lipids and lipid-like molecules constituting the largest proportion (14.02%) (Supplementary Figure S8A). In negative ion mode, we identified 209 differential metabolites, with 51 upregulated and 158 downregulated (Supplementary Figure S7B). These differential metabolites were classified into 9 groups, with organic acids and derivatives constituting the largest proportion (12.44%) (Supplementary Figure S8B).

CV vs. CWT: In positive ion mode, we identified 88 differential metabolites, with 54 upregulated and 34 downregulated (Supplementary Figure S7C). These differential metabolites were classified into 10 groups, with organic acids and derivatives and organoheterocyclic compounds both constituting the largest proportion (13.64%) (Supplementary Figure S8C). In negative ion mode, we identified 148 differential metabolites, with 90 upregulated and 58 downregulated (Supplementary Figure S7D). These differential metabolites were classified into 9 groups, with organic acids and derivatives constituting the largest proportion (14.86%) (Supplementary Figure S8D).

CHV vs. CWT: In positive ion mode, we identified 115 differential metabolites, with 30 upregulated and 85 downregulated (Supplementary Figure S7E). These differential metabolites were classified into 9 groups, with lipids and lipid-like molecules constituting the largest proportion (18.26%) (Supplementary Figure S8E). In negative ion mode, we identified 229 differential metabolites, with 52 upregulated and 177 downregulated (Supplementary Figure S7F). These differential metabolites were classified into 10 groups, with organic acids and derivatives constituting the largest proportion (13.54%) (Supplementary Figure S8F).

Differential metabolite KEGG enrichment analysis

Classification of the differential metabolites revealed a high degree of consistency between the major differential components of the supernatant and the cells. The subsequent pathway analysis will focus solely on the analysis of bacterial components.

In the comparisons of the bacterial components among the three groups, five common differential metabolites with statistical significance (p < 0.05) were identified from the most abundant organic acids and derivatives, as well as organic oxygen compounds. These metabolites are N-acetyl-d-glucosamine 6-phosphate, 6-hydroxypseudooxynicotine, 3-deoxy-D-manno-octulosonate, N-Acetylneuraminic acid, and N-acetylmuramoyl-Ala. Pathway annotation of these five metabolites using the KEGG database revealed five common differential metabolic pathways with statistical significance (p < 0.05): amino sugar and nucleotide sugar metabolism, O-antigen nucleotide sugar biosynthesis, phosphotransferase system (PTS), nicotinate and nicotinamide metabolism, and lipopolysaccharide biosynthesis. Among these, PTS showed significant statistical differences (p = 0.01, p = 0.04, p = 0.03) in ATCC17978Δhcp, ATCC17978ΔvgrG, and ATCC17978ΔhcpΔvgrG (Fig. 3). In the KEGG annotation pathway map and differential abundance score (DAS) plot of the PTS pathway, when comparing ATCC17978 with ATCC17978Δhcp (ABH vs. ABWT), the expression of NAG-6P, β-Glucoside, and Fructose was upregulated in the wild-type strain, and the overall expression of PTS was upregulated (DAS = 0.43), indicating that the deletion of hcp leads to a downregulation of PTS expression (Figs. 4A and 5A). When comparing ATCC17978 with ATCC17978ΔvgrG (ABV vs. ABWT), NAG-6P expression was downregulated and β-Glucoside expression was upregulated in the wild-type strain, with no significant overall change in the PTS expression trend (DAS = 0), suggesting that the deletion of vgrG does not significantly affect PTS function (Figs. 4B and 5B). When comparing ATCC17978 with ATCC17978ΔhcpΔvgrG (ABHV vs. ABWT), the expression of NAG-6P and Fructose was upregulated in the wild-type strain, and the overall expression of PTS was upregulated (DAS = 0.29), indicating that the combined deletion of hcp and vgrG leads to a downregulation of PTS expression (Figs. 4C and 5C).

Fig. 3
figure 3

Bar plots of pathway enrichment analysis for differential metabolites in each bacterial comparison group. The x-axis indicates the quantity of differential metabolites linked to each pathway, while the y-axis displays the names of these pathways. The intensity of the colors in the bar chart reflects the P-value, and the bars’ lengths signify the number of metabolites associated with each pathway. PTS was marked with a red box.

Full size image
Fig. 4
figure 4

Adapted from KEGG pathway database (www.kegg.jp/kegg/kegg1.html).

KEGG annotation pathway maps of PTS in each bacterial comparison group. Red points indicate upregulation of metabolite content, while green points indicate downregulation of metabolite content.

Full size image
Fig. 5
figure 5

DAS plots of each bacterial comparison group. The x-axis shows the DAS, while the y-axis depicts the names of pathways exhibiting differential expression. The lengths of the line segments correspond to the absolute values of the DAS, and the sizes of the circular markers indicate the quantity of differential metabolites within each pathway. The circular markers located on the left side of the axis indicate an overall downregulation tendency in the expression of this pathway. Conversely, the circular markers positioned on the right side of the axis signify an overall upregulation tendency in the pathway’s expression. The color of the circles indicates the magnitude of the p-values.

Full size image

Biofilm assay

Quantitative crystal violet staining revealed that compared to the wild-type strain, ATCC17978Δhcp exhibited significantly reduced biofilm formation (p < 0.01), while ATCC17978ΔvgrG showed a non-significant decrease (p > 0.05). Notably, ATCC17978ΔhcpΔvgrG demonstrated significantly enhanced biofilm-forming capacity (p < 0.0001) (Fig. 6).

Fig. 6
figure 6

(A) Crystal violet staining of biofilms (n = 3); (B) The absorbance of biofilm formation was measured at OD570 nm, followed by a one-way ANOVA for statistical analysis. **p < 0.01; ****p < 0.0001; ns, not significant.

Full size image

SEM

The biofilm formation capacity of four A. baumannii strains was investigated using SEM at ×10,000 and ×30,000 magnifications. The results showed that all four strains maintained their typical rod-shaped morphology and aggregated to form clustered structures. The surface architecture of the wild-type strain was dense and uniform, with intact and plump bacterial morphology, resulting in structurally stable, continuous, and dense biofilms. The surface of ATCC17978Δhcp exhibited varying degrees of damage, leading to loosely structured biofilms. ATCC17978ΔvgrG maintained relatively intact bacterial architecture, while ATCC17978ΔhcpΔvgrG not only preserved complete morphological integrity but also secreted more abundant extracellular polymeric substances (EPS), forming structurally robust biofilms with significantly enhanced density (Fig. 7).

Fig. 7
figure 7

SEM images of A. baumannii at ×10,000 and × 30,000 magnifications. Arrows highlight damaged bacterial surfaces.

Full size image

Bacterial adherence assay

The adhesion assay results demonstrated that compared to the wild-type strain, the adhesion ability of the three mutant strains decreased. However, the adhesion capacity of ATCC17978ΔhcpΔvgrG was still stronger than that of ATCC17978Δhcp and ATCC17978ΔvgrG (Fig. 8).

Fig. 8
figure 8

The rate of bacterial adhesion. Statistical analyses were conducted by one-way ANOVA. ***p < 0.001; ****p < 0.0001.

Full size image

Expression quantification by RT-qPCR

Analysis of biofilm-related gene expression revealed that compared to the wild-type strain, ATCC17978Δhcp exhibited a slight increase in bfmS expression but showed downregulation trends in other biofilm-related genes (p > 0.05). ATCC17978ΔvgrG demonstrated downregulation of bap, csuB, and bfmR genes alongside upregulation of csuA, bfmS, pgaA, and pgaD genes, with only the csuA expression change being statistically significant (p < 0.01). Consistent with the crystal violet biofilm assay results, all tested genes in ATCC17978ΔhcpΔvgrG showed significantly elevated expression levels compared to the wild-type strain (p < 0.05) (Fig. 9).

Fig. 9
figure 9

Relative expression levels of biofilm-associated genes. Statistical analyses were conducted by one-way ANOVA. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; ns, not significant.

Full size image

Discussion

In this study, we found notable variations in metabolite expression among different A. baumannii gene knockout strains (ATCC17978Δhcp, ATCC17978ΔvgrG, ATCC17978ΔhcpΔvgrG) compared to the wild-type strain of A. baumannii (ATCC17978) during co-culture with HPAEpiC through non-targeted metabolomics analysis. In terms of the number of differential metabolites, our results suggest that supernatant > cells > bacteria, which is consistent with the characteristic that prokaryotes (A. baumannii) have weaker secretion capabilities compared to eukaryotes (HPAEpiC). Regarding the types of differential metabolites, the metabolism of lipids and lipid-like molecules, as well as organic acids and their derivatives were predominantly affected in the supernatant and cellular triple-pair comparison groups. In the bacterial three-pair comparison groups, the metabolism of organic acids and their derivatives, as well as organic oxygen compounds, were predominantly affected. Although metabolites in the supernatant are co-secreted by both the bacteria and the cells, the amount of metabolites secreted by the cells was significantly higher than that secreted by the bacteria. Moreover, cell lysis leads to the release of intracellular components into the supernatant during co-culture. This resulted in the main components of the supernatant and cells being highly similar, making it difficult to distinguish them clearly. Therefore, the pathway analysis focused exclusively on the bacterial components.

In the three bacterial component comparison groups, five statistically significant common differential metabolic pathways were identified. Among these, PTS exhibited significant statistical differences in ATCC17978Δhcp, ATCC17978ΔvgrG, and ATCC17978ΔhcpΔvgrG (p = 0.01, p = 0.04, p = 0.03). However, in ATCC17978ΔvgrG, no significant change in overall PTS expression was observed, which may be attributed to the upregulated NAG-6P expression coupled with downregulated β-Glucoside expression.

The PTS is a unique and complex system in bacteria, which generally includes the histidine phosphocarrier protein (HPr), enzyme I (EI), and a variable number of sugar-specific enzyme II complexes (EIIA, EIIB, EIIC, EIID) that depend on the species30. It primarily facilitates the phosphorylation of different sugars and their derivatives via a phosphorelay cascade, subsequently transporting them into the interior of the cell. In addition, PTS is involved in several physiological functions, including central bacterial carbon and nitrogen metabolism31, virulence regulation32, stress response mediation33, biofilm formation34 and antibiotic resistance35thus playing a key role in the overall regulation of bacterial metabolism. Among these functions, PTS is critical for bacterial biofilm formation. For example, in Streptococcus mutans, deletion of the manL gene, which encodes EIIABMan in the mannose PTS, leads to impaired PTS function and severely hinders biofilm formation. Similarly, in Bacillus cereus, deletion of the ptsH gene, which encodes HPr, also impairs PTS function and disrupts biofilm formation36,37. Studies on Vibrio cholerae have demonstrated that PTS participates in biofilm formation through multiple independent pathways38.

We conducted systematic phenotypic analyses of four A. baumannii strains after co-culture, including biofilm assays, SEM observations, adhesion assays, as well as biofilm-related gene expression detection. Comparative analysis with the wild-type strain revealed distinct phenotypic and genotypic alterations in the mutant strains. The results showed that compared to the wild-type strain, ATCC17978Δhcp exhibited significantly reduced biofilm formation capacity and adhesion ability, while ATCC17978ΔvgrG only showed decreased adhesion with no significant difference in biofilm formation. Notably, ATCC17978ΔhcpΔvgrG demonstrated enhanced biofilm formation but reduced adhesion ability relative to the wild-type. SEM observations further revealed surface damage in ATCC17978Δhcp, whereas both ATCC17978ΔvgrG and ATCC17978ΔhcpΔvgrG maintained bacterial surface structure integrity comparable to the wild-type. At the gene relative expression level, RT-qPCR analysis detected upregulation of all examined biofilm-related genes in the double mutant, while only the csuA gene showed increased expression in ATCC17978ΔvgrG among the single mutants.

These results demonstrate that while there is indeed a correlation between bacterial biofilm formation capacity and adhesion ability, the adhesion ability is not entirely determined by biofilm-forming capability. The integrity of bacterial surfaces and the production of EPS play a crucial role in biofilm formation. Moreover, the expression levels of genes associated with biofilm formation are consistent with the biofilm-forming capacity.

Our research team previously conducted a study on ATCC17978Δhcp and found that its ability to form biofilms was diminished compared to the wild-type strain. However, the biofilm formation ability was restored in the hcp complemented strain39. Existing literature also supports that vgrG deletion does not significantly impact A. baumannii biofilm formation, which aligns with our current observations40. Surprisingly, ATCC17978ΔhcpΔvgrG double mutant displayed a unique phenotypic reversal: significantly enhanced biofilm formation coupled with coordinated upregulation of biofilm-associated genes.

The formation mechanism of bacterial biofilms is highly complex. These results demonstrate that while the PTS influences biofilm formation, it does not serve as the sole regulatory mechanism. The significant changes in the PTS may also influence other phenotypic alterations in bacteria. We hypothesize that dual deletion triggers specific stress-response pathways as a compensatory survival strategy, ultimately leading to reinforced biofilm production. However, this hypothesis still requires further experimental validation.

Conclusion

The hcp gene plays a key role in biofilm formation, while the vgrG gene has a comparatively minor impact. The simultaneous absence of both the hcp and vgrG genes unexpectedly significantly promotes the formation of biofilms.