Synergetic attenuated inflammatory response and increased apoptosis confer high pathogenicity of Acinetobacter baumannii

ABSTRACT

Acinetobacter baumannii (A. baumannii) has emerged as a major global healthcare threat, ranking among the leading causes of both hospital-acquired and community-acquired infections. Its extensive antibiotic resistance complicates treatment, underscoring the need for therapeutic targets that elicit robust immune responses. This study identified two multidrug-resistant clinical strains of A. baumannii A.ba-R (rough) and A.ba-S (smooth). A.ba-R has thick and dense capsular polysaccharide, and exhibits high pathogenicity by suppressing host inflammatory responses and inducing apoptosis to evade immune surveillance. Conversely, A.ba-S possesses a thin capsular polysaccharide and triggers pyroptosis via inflammasome activation, causing pronounced inflammation. RNA-sequencing revealed that capsule polysaccharide, transmembrane secretion systems, and nutrient acquisition systems of A.ba-R might contribute to the induction of apoptosis. Our study reveals different roles of apoptosis and pyroptosis in the pathogenicity of A. baumannii and provides potential therapeutic targets against A. baumannii infections.

Introduction

Acinetobacter baumannii (A. baumannii), a Gram-negative, obligately aerobic coccobacillus, is a leading cause of hospital-acquired infections worldwide. Recognized as a critical global health threat, it has been listed by the Infectious Diseases Society of America (IDSA) as one of the six priority “ESKAPE” pathogens requiring urgent development of new antibiotics [1,2]. Moreover, the U.S. Centers for Disease Control and Prevention (CDC) further classifies multidrug-resistant (MDR) A. baumannii as a “Serious” public health threat due to its rapid acquisition of resistance mechanisms [3,4]. The global spread of multidrug resistance in A. baumannii is critically driven by its inherent ability to undergo conjugation, which facilitates horizontal gene transfer. Approximately 45% of isolates worldwide are classified as MDR, exhibiting resistance to multiple antibiotic classes, including β-lactams, macrolides, fluoroquinolones, and aminoglycosides [5–7]. Given its escalating resistance, identifying novel therapeutic and immunomodulatory targets is imperative to combat this resilient pathogen [8].

Host cell death represents a fundamental defense mechanism against bacterial infections [9]. Apoptosis, a genetically regulated non-inflammatory cell death, plays a critical role in eliminating senescent, damaged, or infected cells [10–12]. As one of the predominant cell death modes during infection, apoptosis helps restrict microbial replication and dissemination [13]. Pathogens have evolved diverse strategies to modulate apoptotic pathways for survival. Intracellular bacteria typically suppress host apoptosis to maintain their replicative niche [14]. For instance, Mycobacterium tuberculosis secretes effector proteins that inhibit Bax/Bak-mediated mitochondrial apoptosis, prolonging macrophage survival to establish persistent infection [15]. Salmonella exploits the PI3K/Akt pathway via the SopB effector to block caspase-8 activation and evade immune clearance [16]. Extracellular bacteria predominantly secrete toxins or surface effector molecules and exploit apoptosis to disrupt host barrier or immune surveillance, such as Group A streptococci (GAS) uses pore-forming cytolysin streptolysin O (SLO) to trigger macrophage apoptosis, promoting immune evasion and virulence [17,18]. Staphylococcus aureus induces apoptotic death in infected macrophages to facilitate bacterial dissemination [19].

Inflammasomes are pivotal components of innate immunity that detect microbial patterns and danger signals, which mediate the secretion of pro-inflammatory cytokines IL-1β and IL-18, and induction of pyroptosis, an inflammatory form of programmed cell death [20–22]. By eliminating intracellular pathogen replication niches and facilitating phagocytic clearance, pyroptosis serves as a crucial antimicrobial defense mechanism [23,24]. Among characterized inflammasomes, the NLRP3 inflammasome is particularly significant in pulmonary bacterial infections, including those caused by Klebsiella pneumoniae, Streptococcus pneumoniae, Staphylococcus aureus, Mycobacterium tuberculosis, Pseudomonas aeruginosa, and Legionella pneumophila [25–27]. Notably, the activation magnitude and consequences of NLRP3 signaling vary depending on bacterial virulence, leading to distinct pathological outcomes [28–31]. Emerging evidence highlights the critical role of NLRP3 inflammasome activation in host defense against clinically significant A. baumannii pulmonary infections [32,33].

Our previous work demonstrated that A. baumannii infection triggers concurrent apoptosis, pyroptosis and necroptosis, where type I interferon (IFN-I) plays a regulatory role in host defense against A. baumannii infection by modulating pyroptotic and necroptotic pathways [34]. However, the potential contribution of these distinct cell death modalities to the pathogenicity of clinically isolated MDR A. baumannii strains remains to be elucidated.

In this study, we characterized two clinically relevant A. baumannii strains (A.ba-R and A.ba-S) with distinct surface architectures. Compared with A.ba-S, A.ba-R possessed a thicker capsular polysaccharide (CPS) and exhibited enhanced resistance to both complement-mediated serum killing and phagocytosis by bone marrow-derived macrophages (BMDMs). In murine infection models (bloodstream and 5-Fluorouracil (5-FU)-induced immunosuppression), A.ba-R demonstrated significantly greater pathogenicity, which correlated with its ability to induce extensive host cell apoptosis while suppressing inflammatory responses. In contrast, A.ba-S triggered robust NLRP3 inflammasome activation and pyroptosis, resulting in effective host defense and reduced virulence. Mechanistically, A.ba-R upregulates the expression of genes involved in histamine metabolism, CPS synthesis and bacterial secretion system, facilitating immune evasion. Alcian blue staining further confirmed increased CPS production in A.ba-R, and the isolated CPS was shown to induce apoptosis in BMDMs. Furthermore, A.ba-R exhibited a significantly lower tissue burden in Caspase-3^-/- mice than in wild-type (WT) controls, underscoring the critical role of caspase-3-mediated apoptosis in its pathogenicity. These findings delineate a novel virulence strategy whereby A.ba-R enhances its pathogenicity through apoptosis induction and immune suppression, while A.ba-S elicits protective inflammasome activation. This work identifies promising molecular determinants associated with A. baumannii virulence and reveals potential therapeutic targets against MDR strains.

Materials and methods

Ethics approval and consent to participate

Prior to participation, all human subjects provided written informed consent. All procedures were conducted in accordance with the Shandong University Research Ethics Review Committee (Approval No.: ECSBMSSDU2021-1–086). Studies involving human participants adhere to the Declaration of Helsinki (https://www.wma.net/policiespost/wma-declaration-of-helsinki-ethical-principles-formedical-research-involving-human-subjects/).

WT and knockout mice were SPF-clean and kept under specific pathogen-free conditions (12 h/12 h light and dark cycle, 22 °C ± 2 °C) in the Model Animal Research Center at Shandong University, Jinan, Shandong Province, China. All animal experiments were conducted in strict accordance with the guidelines for the care and use of laboratory animals and were approved by the Ethics Committee of Scientific Research of Shandong University (Approval No.: ECSBMSSDU2021-2–171). All studies were reported in adherence with the ARRIVE guidelines (https://doi.org/10.6084/m9.figshare.30397396).

Bacterial culture and infection of mice

Clinical A. baumannii strains (A.ba-R and A.ba-S) were isolated from sputum samples obtained from Qilu Hospital, Shandong University. Strain A.ba-R was obtained in August 2021, and strain A.ba-S was isolated in November 2014. Both samples were acquired during routine hospital diagnostic procedures. Bacterial cultures were prepared by overnight incubation in brain heart infusion (BHI) medium with shaking, followed by 3–4 h of sub-culture to mid-log phase. Cells were then washed and resuspended in sterile phosphate-buffered saline (PBS) for inoculation. Eight- to-ten-week-old, same-gender WT and Caspase-3^−/− C57BL/6J mice were infected via tail vein injection or intranasal infection (4.0 ×10⁸ colony-forming units (CFU) per mouse). For survival analysis, WT mice received 4.0 × 10⁹ CFU via the tail vein injection, and body weight and survival were monitored daily. In accordance with the IACUC protocol, survival monitoring was continued until mice were moribund. For bacterial load determination, mice were euthanized by carbon dioxide inhalation at predetermined time points post-infection (20 h for the tail vein injection group and 48 h for the intranasal infection group), and brain, spleen, and lung tissues were collected [35,36]. Tissue homogenates were serially diluted and plated onto BHI agar plates, incubated overnight at 37 °C, and subjected to colony counting [34]. Experiments involving the intravenous infection of mice with A. baumannii were conducted in three-independent experiments with similar results. Statistical analysis of tissue bacterial load was performed using a two-tailed Student’s t-test. Survival curves were analyzed using the log-rank (Mantel – Cox) test.

5-FU-induced myelosuppression model

To establish the myelosuppression model, male WT C57BL/6J mice received a single intraperitoneal injection of 5-FU (2 mg per mouse), with a control group receiving an equal volume of dimethyl sulfoxide (DMSO). Daily monitoring of body weight and clinical status was performed throughout the study. Bone marrow cells were collected from femurs and tibias at designated timepoints (days 3, 5, 7, and 9 post-injection) for nucleated cell counting after erythrocyte lysis, enabling evaluation of 5-FU-mediated myeloid suppression. This experiment was conducted in triplicate.

On day 5 post-5-FU administration, WT male C57BL/6J mice were intranasally challenged with A. baumannii (4.0 ×10⁸ CFU per mouse). Post-infection monitoring included daily assessment of clinical parameters [37]. At 48 h post-infection, mice were euthanized with carbon dioxide for lung collection and bacterial load quantification [38]. Tissue homogenates were serially diluted and plated onto BHI agar plates, incubated overnight at 37 °C, and subjected to colony counting [34]. This infection model was conducted in triplicate with similar results. Statistical analysis was performed using a two‑tailed Student’s t‑test for tissue bacterial load and two‑way ANOVA for body weight curves.

Antimicrobial susceptibility test (AST)

Antimicrobial susceptibility testing, including minimum inhibitory concentration (MIC) determination and interpretation, was performed in accordance with Clinical and Laboratory Standards Institute (CLSI) guidelines [39]. For most antimicrobial agents, susceptibility was assessed automatically using the Vitek 2 Compact 60 system with AST-GN335 cards. The inoculum was prepared by first adjusting a bacterial suspension to a 0.5 McFarland standard in 0.45% saline. Subsequently, 145 µL of this suspension was added to 3 mL of 0.45% saline for final standardization before card inoculation and system processing [40]. Exceptions were colistin and cefiderocol, which were tested separately via broth microdilution. Specifically, the MIC for cefiderocol was determined using iron-depleted cation-adjusted Mueller-Hinton broth (ID-CAMHB), prepared in accordance with established protocols and CLSI recommendations [39,41]. All experiments were performed as three independent biological replicates, each with three technical replicates.

Growth assay

A. baumannii strains were pre-cultured overnight in BHI medium. The cultures were then diluted in fresh BHI medium to an initial optical density at 600 nm (OD₆₀₀) of 0.02 and incubated at 37 °C with shaking at 220 rpm for 14 h. Bacterial growth was monitored every hour by measuring the OD₆₀₀. Growth rates were calculated from the resulting curves using GraphPad Prism software. The experiment was performed in triplicate, each comprising three technical replicates. Data were analyzed by two-way ANOVA.

Assessment of bacterial resistance to serum killing

Fasting serum samples were collected from healthy donors at Qilu Hospital, Shandong University during November and December 2023 and March 2026. Donors included individuals of both sexes, aged 20 to 30 years, with normal hepatic and renal function and biochemical parameters. All samples were obtained as part of routine hospital diagnostic procedures. The bacterial resistance to serum killing assay was completed concurrently with sample collection. Heat-inactivated serum (56 °C, 30 min) served as the negative control. Bacterial suspension (1 ×10⁸ CFU/mL) was mixed with test or control sera (200 μL) and incubated at 37 °C for 2 h. To determine CFU counts, the mixture was serially diluted and plated onto BHI agar and incubated for colony counting. Total CFU was calculated as (colony count × dilution factor × 200 µL) / (plated volume). Bacterial survival rate was then assessed by calculating the ratio of final total CFU to the initial inoculum CFU. For flow cytometry and western blot analysis of complement C3b deposition, the mixture was centrifuged, washed, and resuspended in PBS. A portion of the sample was used for western blot analysis, while the remainder was incubated with a rabbit anti-C3b polyclonal antibody (Proteintech, 21337–1-AP) for 30 min. After washing with PBS, the samples were incubated with a donkey anti-rabbit fluorescent secondary antibody (Biolegend, 406416) for 30 min, followed by another wash with PBS prior to analysis on a BD LSR Fortessa flow cytometer (BD Biosciences). Three independent experimental replicates were performed, each with three technical replicates. Statistical analysis was performed using a two-tailed Student’s t-test.

Preparation of BMDMs, treatment, and bacterial infection

Bone marrow cells isolated from WT C57BL/6J mice were differentiated into BMDMs by 5-day culture in L929-conditioned DMEM/F-12 medium containing 10% FBS, 1% non-essential amino acids, and 1% penicillin-streptomycin. Before infection, BMDMs were pretreated for 3 h with specific inhibitors: z-VAD (Calbiochem, 627610), N-acetylcysteine (NAC, Sigma, A9165), or Z-DEVD-FMK (MCE, HY-12466). Both WT and inhibitor-pretreated BMDMs were then infected with A.ba-R or A.ba-S for indicated durations prior to lysis for RNA (3 h and 6 h) or protein (8 h and 12 h) analysis. Separately, following treatment with CPS (50 µg/mL) extracted from A.ba-R for the indicated times (8 h and 12 h), WT BMDMs were lysed for protein analysis. Three independent experimental replicates were performed.

Transmission electron microscopy (TEM)

For TEM analysis, A.ba-R and A.ba-S were fixed with 2.5% glutaraldehyde in PBS. Infected BMDMs were fixed with 2% paraformaldehyde and 2.5% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.4) at 37 °C for 1 h. All samples were processed, embedded, and sectioned for TEM analysis. Imaging was performed using a Hitachi TEM system operated at 80.0 kV. Infected BMDMs were imaged at 4.0k magnification (Zoom-1, HC-1) at Jinan Central Hospital (Jinan, China), while bacterial samples were imaged at 2.0k magnification (Zoom-1, HC-1) at Servicebio (Wuhan, China). Capsule thickness was quantified by taking five independent measurements at random locations per isolate. The experiment was conducted in three independent replicates and statistical significance was assessed using two-tailed Student’s t-tests.

Bacterial killing assay

Following a 2-hour infection with A.ba-R or A.ba-S at the specified multiplicity of infection (MOI), BMDMs were treated with apramycin (100 µg/mL, 45 min) followed by PBS washing to remove extracellular bacteria. To quantify intracellular bacteria, cells were resuspended in 200 µL of PBS, and the resulting suspension was serially diluted and plated on BHI agar for CFU enumeration. Total CFU was calculated as: (colony count × dilution factor) / (plated volume) × 200 µL. The experiment included three independent replicates with four technical replicates each. Statistical significance was assessed using two-sided Student’s t-tests.

Immunoblot analysis and antibodies

Protein samples from in vivo (20 h post tail vein injection and 48 h post intranasal instillation) and in vitro (8 h and 12 h post-infection) experiments were resolved by 12% SDS-PAGE and transferred onto polyvinylidene fluoride (PVDF) membranes. After blocking, membranes were incubated with the following primary antibodies: anti-caspase-1 (AdipoGen, AG-20B-0042), anti-caspase-3 (CST, 9662), anti-cleaved-caspase-3 (CST, 9661), and anti-GAPDH (CST, 2118). HRP-conjugated secondary antibodies (anti-rabbit, CST, 7074; anti-mouse, CST, 7076) were then applied. Protein molecular weights were estimated using a 180 kDa pre-stained protein ladder (Vazyme, MP102-01). For both tissue and cellular samples, infected conditions were compared with their respective uninfected controls. The experiment was performed in three independent replicates.

Quantitative real-time PCR (qRT-PCR)

Total RNA was isolated from bacteria cultures, uninfected BMDMs and infected BMDMs at 3 h and 6 h post-infection using TRIzol Reagent (Invitrogen, Thermo Fisher Scientific, 15596018). cDNA was reverse transcribed using M-MLV Reverse Transcriptase (Promega, M1701). qRT-PCR was performed on the Roche LightCycler 96 Real-Time Detection System. Bacterial 16S rRNA and host Hprt (hypoxanthine guanine phosphoribosyltransferase) genes were used as internal references for normalizing bacterial and host gene expression, respectively. Normalization was performed using the ΔCt method, where the ΔCt value was calculated as the Ct value of the target gene minus the Ct value of the internal control gene. The primer sequences used are listed in Supplemental Table S5. Infected samples were compared with uninfected BMDM (media) controls. The experiment was conducted with three independent biological replicates, each containing four technical replicates. Statistical analysis was performed using two-way ANOVA for host gene expression and two‑tailed Student’s t-tests for bacterial gene expression.

Library preparation, genome sequencing, and assembly

High-molecular-weight genomic DNA was extracted from A. baumannii isolates following standard protocols and subjected to quality and quantity assessment using Qubit fluorometry and agarose gel electrophoresis. Complete genome sequencing was performed using a hybrid strategy combining long-read Oxford Nanopore Technologies (ONT) sequencing and short-read second-generation sequencing.

For long-read sequencing, libraries were prepared using the ONT Native Barcoding Kit (SQK-NBD114.96) according to the manufacturer’s instructions and sequenced on the PromethION platform (Oxford Nanopore Technologies) with R10 chemistry (SUP-V5). For short-read sequencing, Illumina-compatible libraries were constructed and sequenced on the DNBSEQ-T7 platform (MGI Tech), generating paired-end reads.

Raw sequencing reads from both platforms were subjected to quality control to remove low-quality reads and adaptor sequences. Long and short reads were jointly used for de novo genome assembly with Flye, followed by iterative polishing using the short-read data. Circularization of the chromosome was confirmed based on read overlap at contig ends [42]. Plasmid sequences were identified from the assembled contigs using Plasmer, and putative plasmid contigs were further annotated by BLAST searches against the PLSDB database. The final assemblies were evaluated by mapping both long- and short-read data back to the assembled genomes to assess read alignment rates and average coverage depth. Multilocus sequence typing (MLST) was conducted on both strains according to the Pasteur scheme using the MLST software. Genome annotation was performed using Prokka to predict protein-coding genes and RNA features. Genome completeness and contamination were assessed using CheckM2, and taxonomic assignment was confirmed with GTDB-Tk. The resulting assemblies represent complete circular genomes. The genome sequence datasets have been deposited in GenBank under accession numbers PRJNA1401946 and PRJNA1401942. The scalable circular genome maps were generated using the Java package CGView.

To assess the genomic similarity between the bacterial strains, the Average Nucleotide Identity (ANI) was calculated using fastANI. To identify genomic variations between the two strains, the A.ba-R genome was aligned to the A.ba-S reference genome using MUMmer. Structural and sequence variants, including single-nucleotide polymorphisms (SNPs), were subsequently identified using SyRI. Genome-wide SNP density was calculated using a sliding window approach with a 10-kb window size, and a density map was generated based on the chromosomal coordinates of identified SNP loci. To assess gene-specific variation, SNP loci were mapped to the A.ba-S gene annotation, and the number of SNPs per gene was quantified. The top 10 genes with the highest SNP burden were selected for further functional classification and visualization.

RNA-sequencing (RNA-seq) and analysis

Total RNA was extracted from uninfected BMDMs and BMDMs infected with either A.ba-R or A.ba-S at 3 h post-infection. RNA-seq libraries were prepared and sequenced by Novogene Corporation (Beijing, China) on an Illumina platform, generating paired-end reads. Raw reads were quality-filtered to remove adaptors, poly-N reads, and low-quality reads, resulting in high-quality clean reads. For host – pathogen dual RNA-seq analysis, these clean reads were aligned to the Mus musculus reference genome (GRCm39; annotation: Mus_musculus.GRCm39.105.gtf) and to the A. baumannii reference genome NZ_CP058289 using Bowtie2 (v2.2.3) [43]. Gene expression was quantified with HTSeq (v0.6.1) and normalized as Fragments Per Kilobase of transcript per Million mapped reads (FPKM) [44,45]. Differentially expressed genes were identified using the DEGSeq R package (with edgeR normalization), with significance defined as an adjusted P < 0.05 (Benjamini – Hochberg method) and |log₂ fold change| ≥1 [46]. Significant genes were subjected to Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and gene set enrichment analysis (GSEA) using clusterProfiler (v4.10.1), and visualized via heatmaps using pheatmap. Gene set variation analysis (GSVA) was performed on RNA-seq expression data to estimate pathway activity scores using the GSVA R package. All data are publicly available at Figshare (https://doi.org/10.6084/m9.figshare.30397396).

K-locus identification and genomic localization

The CPS synthesis locus (K-locus) of A. baumannii was identified using Kaptive [47]. The complete circular chromosome sequence of strain NZ_CP058289 was used as input and compared with the A. baumannii K-locus reference database to determine the K-locus type [47,48].

To define the precise genomic location of the identified K-locus, the corresponding reference locus sequence was retrieved from the Kaptive database and aligned to the NZ_CP058289 genome using minimap2. The start and end coordinates of the K-locus were determined from this alignment.

Genomic coordinates of the glycosyltransferase genes detected in the RNA-seq analysis were extracted from the GenBank annotation file of NZ_CP058289. These coordinates were then compared with the K-locus boundaries to confirm whether each glycosyltransferase gene resides within the K-locus region.

Enzyme-linked immunosorbent assay (ELISA)

Cytokine concentrations in both in vivo (at 20 h post‑tail vein injection and 48 h post‑intranasal instillation) and in vitro samples (12 h post-infection) were quantified using commercial ELISA kits (ELISA MAX™ Standard Sets, BioLegend) following the manufacturer’s instructions. Specifically, mouse interleukin IL-1β (Cat# 432601) and IL-6 (Cat# 431301) levels were measured to assess inflammatory responses. For both tissue and cellular samples, cytokine levels under infected conditions were compared with those of their respective uninfected controls. The experiment included three independent biological replicates. For in vitro assays, each biological replicate comprised four technical replicates. Statistical analysis was performed using two-way ANOVA (for cellular samples with multiple factors) or two-tailed Student’s t-tests (for tissue sample comparisons), as appropriate.

Lactate dehydrogenase (LDH) release assay

Cell culture supernatants were collected at 8 h post-infection. Using supernatants from uninfected cells as a control, LDH activity was assessed with the CytoTox 96® Non-Radioactive Cytotoxicity Assay (Promega, G1780) according to the manufacturer’s instructions. Three independent experimental replicates were performed, each with three technical replicates. Statistical significance was determined using two-tailed Student’s t-test.

India ink staining

A bacterial colony from an overnight BHI agar culture was suspended in a drop of sterile saline on a microscope slide and air-dried at room temperature. The smear was then saturated with 1% crystal violet for 5 minutes, rinsed with deionized water, and air-dried. Subsequently, a drop of India ink was applied for negative staining, followed by a final air-drying step [49]. Slides were visualized using an optical microscope (ZEISS Axio Imager A2). Three independent experimental replicates were performed.

Purification of bacterial capsules and SDS-PAGE detection

Bacterial cells were cultivated overnight in 10 mL of BHI medium at 37 °C with shaking. The cells were harvested by centrifugation (12,000 × g for 3 min), and CPS was extracted from the pellet using a Bacterium Capsular Polysaccharide Extraction Kit (Beijing BioRab Tech, KL230801) according to the manufacturer’s protocol. The absorbance of the extracted CPS was measured at 315 nm using a 96-well microplate reader, and the concentration of the extracted CPS was quantified against a standard curve of known CPS concentrations [50]. For visualization, CPS-containing samples were normalized based on their concentration, mixed with an equal volume of SDS sample buffer, and boiled at 100 °C for 5 min. Equal volumes of the normalized samples were then separated by SDS-PAGE, and the gel was stained with 0.1 % (w/v) alcian blue to visualize the CPS. Three independent experimental replicates were performed.

Statistical analysis

All data were analyzed with GraphPad Prism 8.0.2 and are presented as the mean ± standard error of the mean (SEM). Statistical analyses were performed using two-way ANOVA, two-sided Student’s t-test and log-rank (Mantel – Cox) test. P ≤ 0.05 were considered statistically significant.

Results

Comparative genomic analysis and plasmid profiling of two clinical A. baumannii strains with divergent pathogenicity

To determine the genomic difference between two clinical A. baumannii strains, A.ba-R and A.ba-S, complete genome sequencing was performed. The assembled genomes comprised a single circular chromosome of approximately 3.94 Mb for A.ba-R and 4.07 Mb for A.ba-S, with similar genomic architecture, coding density, and GC content profiles (Supplemental Figure S1a). ANI analysis revealed that both A.ba-R and A.ba-S shared 97.8% identity with the A. baumannii reference genome NZ_CP058289 from GenBank. A.ba-R and A.ba-S shared an ANI of 99.85%, confirming they are highly related A. baumannii strains (Supplemental Figure S1b). MLST analysis further demonstrated that both strains belonged to sequence type ST2 (Supplementary Figure S1c). Plasmid identification revealed that A.ba-R harbored two plasmids (pSSA12-1 and p2AB5075), whereas A.ba-S contained only one (pKCRI-28–1) (Supplementary Figure S1d). Functional annotation of the plasmids revealed distinct characteristics (Supplemental Table S1). In A.ba-R, both pSSA12-1 and p2AB5075 carry genes involved in plasmid replication and maintenance, conjugation, toxin-antitoxin systems, and defense/stress adaptation. In A.ba-S, pKCRI-28–1 harbors genes associated with plasmid replication and maintenance, mobile genetic elements, defense systems, and transcriptional regulation. However, these plasmid-borne genetic features did not correlate with the differences in CPS production and virulence between the two strains.

To further explore genetic variations underlying the phenotypic differences between the two strains, we performed SNP analysis using A.ba-S as the reference genome. Genome-wide comparison identified 950 SNP sites between the two strains (Supplementary Table S2). These substitutions were predominantly transition mutations, with A–G and C–T being the most frequent (Supplementary Figure S1e). SNP density analysis revealed that the top 10 most variable genes were primarily associated with three functional categories: capsule biosynthesis/exopolysaccharide precursor synthesis, virulence/cell envelope and stress response (Supplementary Figure S1f). These genetic variations, particularly in capsule biosynthesis and virulence-associated genes, may account for the differences in capsule composition and virulence between the two strains.

The clinically isolated A. baumannii strain A.ba-R is less sensitive to serum killing and more resistant to macrophage phagocytosis than A.ba-S

On Columbia blood agar, A.ba-S formed flat, gray colonies, while A.ba-R colonies were raised and translucent (Supplemental Figure S2a). No morphological differences were observed on LB or BHI agar (Supplemental Figure S2a). The growth rates were comparable between A.ba-R and A.ba-S in BHI media (Supplemental Figure S2b). To further characterize these clinical isolates, we assessed their antibiotic resistance profiles and serum survival capabilities. AST performed according to CLSI guidelines showed that both A.ba-R and A.ba-S exhibited MDR profiles, with MICs exceeding clinical breakpoints for most agents tested [39]. Both A.ba-R and A.ba-S remained susceptible to colistin and cefiderocol (Supplemental Figure S2c). Notably, despite their similar resistance profiles, A.ba-R demonstrated significantly greater resistance to serum killing than A.ba-S (Figure 1(a)). Quantitative analysis by western blotting (Supplementary Figure S2d) and flow cytometry (Figure 1(b) and Supplementary Figures S2e, f) further confirmed that A.ba-R exhibited lower C3b deposition on its surface, a key virulence trait that enables evasion of complement-mediated lysis and enhances bloodstream survival [51]. To investigate the phagocytosis resistance, we compared the intracellular survival of A.ba-R and A.ba-S in BMDMs across multiple MOIs. The bacterial killing assay revealed fewer A.ba-R than A.ba-S within BMDMs (Figure 1(c)). Consistent with this, at 4 h post-infection, TEM analysis of BMDMs revealed that phagocytosis of A.ba-S was significantly higher than that of A.ba-R, with an approximately 10-fold difference in bacterial uptake (30 ± 8.26 vs 3 ± 1.08 bacterial cells per macrophage) (Figures 1(d,e)). TEM also revealed distinct surface ultrastructure: A.ba-R was surrounded by a thick, fibrous, capsule-like layer, whereas A.ba-S exhibited a sparse, electron-dense structure (Figures 1(f,g)). These results indicate that A.ba-R exhibits dual resistance to both serum killing and phagocytosis, likely attributable to the presence of a fibrous capsular-like layer on its surface. This structural feature may enhance its clinical virulence by facilitating improved immune evasion.

Figure 2.

A.ba-R exhibits enhanced resistance to host clearance in vivo. a. Bacterial burden in brains and spleens of WT mice 20 h post-intravenous infection with 4.0 × 10⁸ CFU A.ba-R or A.ba-S (n = 5 mice per group). b. IL-1β and IL-6 levels in spleens and sera at 20 h post-infection, measured by ELISA (n = 5 mice per group). c. Immunoblot analysis of caspase-3 (cleaved) and caspase-1 (p20) activation in brains and spleens of uninfected, A.ba-R- and A.ba-S-infected mice. d, e. Immunocompromised WT mice (5-FU-treated) were infected intranasally with 4.0 × 10⁸ CFU A.ba-R or A.ba-S. Body weight change (d), and lung bacterial burden (e) were assessed at 48 h post-infection (n = 5 mice per group). f. IL-1β and IL-6 levels in lungs and sera at 48 h post-infection (n = 5 mice per group). g. Immunoblot analysis of caspase-3 (cleaved) and caspase-1 (p20) in lungs (e). Survival curves of WT mice intravenously infected with 4 × 10⁹ CFU of A.ba-R or A.ba-S (n = 8 mice per group). Data are representative of 3 independent experiments with similar results (a-g) or 2 independent experiments with similar results (h). Data represent mean ± SEM for (a, b, d-f), statistical analysis: 2-sided Student’s t-test without correction for multiple comparisons for (a, b, e, f), two-way ANOVA for (d), log-rank (Mantel–Cox) test for (h), * P < 0.05, ** P < 0.01, *** P < 0.01, **** P < 0.0001.

Figure 1.

Enhanced serum resistance and reduced phagocytosis in clinical MDR A. baumannii strain A.ba-R. a. Survival rates of A.ba-R and A.ba-S after 2 h treatment with human sera or heat-inactivated sera. Each dot represents an independent experiment (n = 3 biologically independent samples). b. C3b deposition on A.ba-R and A.ba-S was measured by flow cytometry after 2 h treatment with human sera or heat-inactivated sera. Each dot represents an independent experiment (n = 3 biologically independent samples). c. Intracellular bacterial counts in WT BMDMs infected with A.ba-R or A.ba-S at the indicated MOI for 2 h (n = 4 biologically independent samples). d, e. TEM analysis of WT BMDMs infected with A.ba-R or A.ba-S (MOI = 40). Magnification: 4.0k (Zoom-1, HC-1). Scale bars: 1 µm (d). Intracellular bacterial counts are quantified in (e). f, g. TEM images of A.ba-R and A.ba-S. Magnification: 2.0k (Zoom-1, HC-1). Scale bars: 1 µm (f). Quantification of the thickness of the CPS in A.ba-R and A.ba-S in f (g). Data are from 3 independent experiments (a, b, c) or representative of 3 independent experiments with similar results (d-g). Data represent mean ± SEM for (a, b, c, e, g), * P < 0.05, *** P < 0.001, **** P < 0.0001(2-sided Student’s t-test uncorrected for multiple-comparisons).

A.ba-R has higher pathogenicity than A.ba-S

To assess the pathogenicity of A.ba-R and A.ba-S in mice, we intravenously infected WT mice with either A.ba-R or A.ba-S (4.0 × 10⁸ CFU per mouse, n = 5 per group). Notably, mice infected with A.ba-R exhibited markedly higher bacterial burdens in both the brain and spleen compared with those challenged with A.ba-S (Figure 2(a)), whereas the production of IL-1β and IL-6 was significantly lower in the spleens and sera of A.ba-R-infected mice than in A.ba-S-infected mice (Figure 2(b)). Western blot analysis further revealed caspase-3 activation in the brain and spleen of A.ba-R-infected mice, but enhanced caspase-1 activation in the spleen of A.ba-S-infected mice (Figure 2(c)). To further investigate the pathogenicity in a severe infection model, we used 5-FU-treated immunocompromised mice (Supplemental Figure S3) [52]. Following 5-FU treatment, mouse body weight decreased initially and began recovering on days 4–5, while bone marrow cell counts declined continuously, reaching their nadir on days 3–7 before recovering on days 8–9 (Supplemental Figures S3a, b). This model exhibited increased susceptibility to A. baumannii infection (Supplemental Figures S3c-e). Mice were intranasally infected with A.ba-R or A.ba-S on day 5 post-5-FU injection (n = 5 per group), a time point of initial weight recovery but maximal bone marrow depletion. In this model, A.ba-R-infected mice exhibited significantly greater weight loss than A.ba-S-infected mice at day 7 (Figure 2(d)). At 48 h post-infection, the lung bacterial loads demonstrated a statistically significant elevation in A.ba-R-infected mice compared with their A.ba-S-exposed counterparts (Figure 2(e)). Consistent with the pathophysiological patterns observed in the intravenous challenge model, the production of IL-1β and IL-6 was also markedly lower in the lungs and sera of A.ba-R-infected mice compared with those of A.ba-S-infected mice (Figure 2(f)). Similarly, lung tissues from A.ba-R-infected mice exhibited remarkable caspase-3 activation, whereas those from A.ba-S-infected mice exhibited more caspase-1 activation (Figure 2(g)). To assess differences in mortality, WT mice were intravenously infected with an extremely high inoculum dose of A.ba-R or A.ba-S (4.0 × 10⁹ CFU per mouse; n = 8 per group). All mice infected with A.ba-R succumbed by 30 h post-infection, whereas mice infected with A.ba-S all survived beyond day 3 (Figure 2(h)). Collectively, these findings demonstrate that the A.ba-R exhibits enhanced evasion of host clearance mechanisms compared with A.ba-S. This amplified resistance may be attributed to the ability of A.ba-R to suppress inflammatory responses and simultaneously induce apoptosis, thereby facilitating its persistence and exacerbating its virulence in the host.

A.ba-R induces less inflammatory responses than A.ba-S

To further elucidate the mechanisms underlying the pathogenicity differences between A.ba-R and A.ba-S, we conducted RNA-seq on BMDMs infected with these two strains for 3 h (Supplemental Table S3). Compared with uninfected controls (Media), 391 genes were upregulated in A.ba-R-infected BMDMs and 276 genes were upregulated in A.ba-S-infected BMDMs (Supplemental Figure S4a). Additionally, compared with A.ba-S-infected BMDMs, A.ba-R-infected BMDMs specifically upregulated 63 genes and downregulated 10 genes (Figure 3(a)). GO enrichment analysis of the 63 upregulated genes in A.ba-R-infected BMDMs revealed significant enrichment in pathways related to negative regulation of chemotaxis, cytokine production, and canonical NF-κB signaling (Figure 3(b)). This pattern suggests a suppression of inflammatory responses. The enrichment of pathways associated with Ras/MAPK signaling and apoptosis indicates their involvement in the induction of apoptosis (Figure 3(b)). We next focused on the 6,198 genes that were commonly downregulated in both A.ba-R- and A.ba-S-infected BMDMs compared with uninfected controls. Functional enrichment analyses revealed that these genes are predominantly involved in core cellular processes, including cell cycle progression, chromosome segregation, DNA replication, ribosome biogenesis, mitochondrial translation, and oxidative phosphorylation (Supplementary Figure S4b).

Figure 4.

Apoptosis promotes A.ba-R pathogenicity in vivo. a. LDH analysis of WT BMDMs infected with A.ba-R and A.ba-S (MOI = 50) for 8 h (n = 3 biologically independent samples). b. Immunoblot analysis of caspase-1 (p20), caspase-3 (cleaved) in WT BMDMs infected with A.ba-R or A.ba-S (MOI = 50) at indicated time points. c. Caspase-1 (p20) expression in WT BMDMs infected with A.ba-S (MOI=50, 12 h) treated with z-VAD (25 µM), NAC (2 mg/mL) or Z-DEVD (50 µM). d. IL-1β gene expression in A.ba-R- and A.ba-S-infected BMDM from (c) and (e) (n = 4 technical replicates; 3 independent experiments). e. Caspase-3 (cleaved) expression in WT BMDMs infected with A.ba-R (MOI = 50, 12 h) with indicated inhibitors. f. Bacterial burden in brains and spleens of WT and Caspase-3-/- mice 20 h post intravenous infection with 4.0 x 108 CFU A.ba-R (n = 6 mice for WT, n = 4 mice for Caspase-3-/-). g. Immunoblot analysis of caspase-3 (cleaved) and caspase-1 (p20) in brains and spleens of uninfected and A.ba-R-infected WT and Caspase-3-/- mice. h. IL-1β and IL-6 levels in spleens and sera at 20 h post-infection (n = 6 mice for WT, n = 4 mice for Caspase-3-/-). Data are representative of 3 independent experiments with similar results (a-h), data represent mean ± SEM for (a, d, f, h), 2-sided Student’s t test without multiple-comparisons correction for (a, f, h), two-way ANOVA for (d), * P < 0.05, *** P < 0.01, **** P < 0.0001.

Figure 3.

A.ba-R suppresses inflammatory responses in BMDMs. a. Venn diagram of overlapping DEGs across treatment conditions. b. GO enrichment analysis of upregulated genes in A.ba-R- vs. A.ba-S-infected BMDMs. Gene ratio represents pathway-associated gene proportion; color intensity indicates significance (P-value adjusted (P adjust)) c. Heatmap of genes upregulated in A.ba-S-versus A.ba-R-infected BMDMs. d. qRT-PCR validation of chemokine Ccl2/7/8/12 expression in infected BMDMs with A.ba-R or A.ba-S at 50 MOI for indicated time (n = 4 technical replicates; 3 independent experiments). e. Analysis of GSEA enrichment of apoptosis-related genes in BMDMs infected with A.ba-R relative to WT BMDMs infected with A.ba-S. The enrichment score (es) plot (top): the peak furthest from the vertical distance ×=0 axis represents the es value. This peak appears at the front end of the set of categorized apoptotic genes (es > 0), indicating that the apoptotic pathway is upregulated in A.ba-R-infected BMDMs. The barcode plot (bottom) marks the positions of all genes involved in the apoptotic process within the gene sorting list using vertical lines. The concentration of these vertical lines at the front end of the gene sorting list further supports the upregulation of the apoptotic gene sets pathway. The gray area in the plot illustrates the distribution of rank values, which are calculated from the log2 Fold change values of all genes after sorting. f. GSVA heatmap of pyroptosis and apoptosis pathway activity in infected BMDMs with A.ba-R and A.ba-S. g. qRT-PCR validation of apoptosis-related gene expression in infected BMDMs with A.ba-R or A.ba-S at 50 MOI for indicated time (n = 4 technical replicates; 3 independent experiments). Data are representative of 3 independent experiments with similar results (d, g), data represent mean ± SEM for (d, g), two-way ANOVA for (d, g), ** P< 0.01, *** P < 0.001, **** P < 0.0001.

In contrast, KEGG analysis of the 232 genes commonly upregulated by both strains showed enrichment in general immune activation pathways, including TNF signaling, NF-κB, and cytokine-cytokine receptor interaction (Supplementary Figure S4c). GO analysis further confirmed that both strains are recognized by multiple pattern recognition receptors, triggering a cellular response to bacteria (Supplementary Figure S4d). However, despite this common recognition, the induction of common inflammatory cytokines was significantly more pronounced in A.ba-S-infected BMDMs than in A.ba-R-infected BMDMs (Supplementary Figure S4e). Strikingly, among the ten genes significantly upregulated upon A.ba-S infection, four encoded chemokines (Ccl2, Ccl7, Ccl8, and Ccl12) implicated in immune cell recruitment. The remaining six genes encompassed diverse functions: immune modulation (Il10), immediate-early transcriptional response (Egr1), cellular stress and apoptosis (Htra4), as well as metabolism and cell cycle regulation (Gatm, Rgcc, Nyap2) (Figure 3(c)). This expression profile suggests that A.ba-S infection potently stimulates immune cell recruitment and inflammatory responses. qRT-PCR independently confirmed the upregulation of these chemokines in the infected BMDMs (Figure 3(d)). Collectively, these results indicate that A.ba-S infection promotes a robust pro-inflammatory response, whereas A.ba-R infection predominantly activates apoptotic signaling and suppresses inflammation, highlighting a fundamental divergence in their interactions with host.

To determine the role of cell death in the pathogenicity of A.ba-R and A.ba-S, we employed GSEA. This revealed a significantly higher activation of the apoptosis pathway in A.ba-R-infected BMDMs compared with A.ba-S-infected cells (Figure 3(e)). We further quantified pathway enrichment scores using GSVA, which confirmed significant activation of apoptotic pathways in A.ba-R-infected BMDMs (Figure 3(f)), recapitulating the previous observations (Figures 2(c,g)), In contrast, pyroptosis-related markers, particularly those associated with inflammatory cascades, were significantly attenuated in A.ba-R-infected BMDMs relative to their A.ba-S-infected counterparts (Figure 3(f)). qRT-PCR analysis confirmed that A.ba-R-infected BMDMs exhibited a pro-apoptotic transcriptional signature: Bcl2 expression was significantly reduced, whereas pro-apoptotic-associated genes (Bax, DFFEB) were upregulated relative to A.ba-S-infected cells (Figure 3(g)).

A. ba-R triggers apoptosis whereas A.ba-S triggers pyroptosis

To further resolve cell death mechanisms underlying strain-specific pathogenicity, BMDMs were infected with A.ba-R and A.ba-S and harvested at 8 h and 12 h post-infection for comparative analysis. A.ba-S-infected BMDMs exhibited significantly more cell death than A.ba-R-infected BMDMs as measured by LDH release (Figure 4(a)). Proteolytic processing of pyroptosis executor caspase-1 and apoptosis marker caspase-3 was systematically profiled. Immunoblot analyses revealed opposed activation patterns: A.ba-S-infected BMDMs exhibited robust caspase-1 activation evidenced by pronounced cleaved caspase-1 p20 fragment generation, indicating canonical pyroptotic pathway engagement, whereas caspase-3 processing remained undetectable (Figure 4(b)). Conversely, A.ba-R-infected BMDMs elicited predominant caspase-3 activation, concurrent with attenuated caspase-1 maturation (Figure 4(b)), thereby validating the strain-specific cell death modality dichotomy observed in vivo and transcriptional profiling data. Furthermore, BMDMs were subjected to pharmacological inhibition of key apoptotic/pyroptotic pathways, followed by quantitative profiling of caspase-1 and caspase-3 activation. Complementary to caspase activation profiling, IL-1β secretion – a cardinal biomarker of pyroptotic inflammation, was quantified by ELISA in inhibitor-treated BMDMs following bacterial challenge. In BMDMs infected with A.ba-S, both the ROS scavenger NAC and pan-caspase inhibitor z-VAD effectively suppressed pyroptotic lysis (Figure 4(c)) and subsequent IL-1β secretion (Figure 4(d)). Conversely, A.ba-R-infected BMDMs demonstrated caspase-3-dependent apoptotic dominance, where z-VAD and the specific caspase-3 inhibitor Z-DEVD-FMK abrogated apoptosis (Figure 4(e)). Notably, while A.ba-R infection attenuated overall IL-1β production, residual cytokine release remained sensitive to z-VAD and Z-DEVD inhibition (Figure 4(d)).

To mechanistically dissect apoptosis-dependent pathogenesis, WT and Caspase-3^-/- mice were intravenously challenged with A.ba-R (4.0 ×10⁸ CFU per mouse). WT mice exhibited significantly higher bacterial burdens in the brain and spleen compared with those of Caspase-3^-/- counterparts (Figure 4(f)). Corroborating these observations, immunoblot analysis of infected tissues revealed robust apoptotic progression in WT mice but a complete ablation of apoptotic signaling in Caspase-3^-/- mice (Figure 4(g)). Pyroptotic signaling was undetectable in either WT or Caspase-3^-/- mice infected with A.ba-R (Figure 4(g)). Furthermore, the production levels of IL-1β and IL-6 in the spleens and sera were comparable between infected WT and Caspase-3^-/- mice (Figure 4(h)). Collectively, our findings demonstrate that caspase-3-dependent apoptosis mediates immune evasion and contributes to the in vivo pathogenesis of A.ba-R.

Bacterial secretion system and CPS may play important roles in A.ba‑R‑induced apoptosis

To identify bacterial virulence determinants, we performed comparative transcriptomic analysis of A.ba-R and A.ba-S (Supplemental Table S4). Relative to A.ba-R, A.ba-S showed differential expression of 200 upregulated and 87 downregulated genes (Figure 5(a)). KEGG pathway analysis revealed distinct metabolic profiles: A.ba-R-infected samples showed enhanced activities in histamine metabolism, bacterial secretion systems, and amino/nucleotide sugar metabolism (Figure 5(b)). Conversely, A.ba-S-infected samples exhibited significant upregulation of valine, leucine and isoleucine degradation pathways, along with phenylalanine and propionate metabolism (Figure 5(c)). Further enrichment analysis demonstrated significantly higher expression of bacterial secretion system-related genes in A.ba-R compared with A.ba-S. Furthermore, genetic analysis identified significant upregulation of glycosyltransferase genes involved in CPS synthesis in A.ba-R (Figure 5(d)). To address whether the observed transcriptional differences could be attributed to gene presence or absence, we examined the genomic localization of glycosyltransferase genes that were highly expressed in A.ba-R but not in A.ba-S, and found that these genes were absent in the genome of A.ba-S. (Supplementary Figure S5a). These findings are consistent with the genomic SNP analysis (Supplementary Figure S1f) and the distinct surface ultrastructural differences observed by TEM (Figure 1(f)). Notably, these upregulated genes mapped to the KL3 capsule biosynthesis locus according to the A. baumannii reference genome NZ_CP058289 (Supplemental Figure S5b). qRT-PCR validation of differentially expressed genes confirmed the RNA-seq results, showing elevated expression of secretion system components (tssH [HTZ92_RS10275], Hcp [HTZ92_RS10320], DotU [HTZ92_RS10260], tssM [HTZ92_RS10295]) (Figure 5(e)) and CPS synthesis genes (Glyco-1[HTZ92_RS16950], Glyco-2 [HTZ92_RS16955], Glyco-3 [HTZ92_RS16935], and Glycof-4 [HTZ92_RS16940]) (Figure 5(f)) in A.ba-R. These findings are consistent with the established role of CPS in A. baumannii pathophysiology [53]. India ink staining visually confirmed the capsule presence in A.ba-R but not A.ba-S (Supplemental Figure S5c). To functionally validate the CPS expression differences, CPS extracted from A.ba-R was separated by SDS-PAGE and stained intensely with alcian blue, whereas the material from A.ba-S yielded no detectable signal, indicating a marked difference in CPS production between the strains (Figure 5(g)). Furthermore, exposure of BMDMs to purified A.ba-R CPS triggered caspase-3 cleavage, indicating activation of the apoptotic pathway (Figure 5(h)). These findings suggest that A.ba-R’s enhanced virulence stems from coordinated upregulation of bacterial secretion systems and CPS, which may act synergistically to promote apoptosis induction, immune evasion, and infection establishment.

Figure 5.

Bacterial secretion systems and CPS contribute to A.ba-R-induced apoptosis. a. Volcano plot of differentially expressed genes between A.ba-R and A.ba-S. Numbers indicate upregulated/downregulated genes (fold change > 2, P< 0.05). b, c. Top 10 enriched pathways for upregulated genes in A.ba-R (b) and A.ba-S (c). bubble size represents gene count; color intensity indicates significance(P-value); x-axis shows gene ratio. d. Heatmap of bacterial secretion system and glycosyltransferase related differentially expressed in A.ba-R versus A.ba-S. e, f. qRT-PCR validation of secretion system (tssH, Hcp, DotU, tssM) (e) and glycosyltransferase (Glyco-1/2/3, glycof-4) genes (f) expression in A.ba-R versus A.ba-S (n = 4 technical replicates; 3 independent experiments). g. Analysis of polysaccharides in early postexponential phase cell lysates of A.ba-R and A.ba-S separated by SDS-PAGE and stained with alcian blue. h. Caspase-3 (cleaved) expression in WT BMDMs treated with A.ba-R (MOI = 50) or CPS (50 µg/mL) for indicated time. Data represent 3 independent experiments with similar results (e-h), data represent mean ± SEM for (e, f), **** P < 0.0001 (2-sided Student’s t-test without multiple-comparisons correction).

Discussion

As a Gram-negative bacterium with extensive multidrug resistance, A. baumannii poses a serious global public health challenge. Clinical strains of this pathogen exhibit marked heterogeneity and have been observed to modulate cell death programs through diverse virulence factors [54,55]. CPS, a high-molecular-weight carbohydrate matrix that encases the bacterial surface, forms a critical physical barrier [56]. This barrier effectively shields the pathogen from host innate immune effectors and limits antibiotic penetration, thereby contributing significantly to its antimicrobial resistance [57].

We identified two clinical A. baumannii strains A.ba-R and A.ba-S with distinct virulence profiles. A.ba-R exhibited robust CPS production, which may contribute to its resistance to serum killing and phagocytosis, suggesting a potential role for the capsule in immune evasion. In contrast, A.ba-S showed minimal CPS expression and was highly susceptible to these host defenses. In both in vitro and in vivo models, A.ba-R predominantly induced apoptosis while suppressing inflammatory responses, which correlated with its high pathogenicity. Conversely, A.ba-S primarily triggered pyroptosis and robust inflammatory responses, demonstrating significantly attenuated virulence. In vivo infection of Caspase-3^-/- mice confirmed the critical role of caspase-3-mediated apoptosis in A.ba-R’s enhanced pathogenicity. Furthermore, treatment of BMDMs with purified A.ba-R CPS was sufficient to induce apoptosis. Integrated with RNA-seq data, we propose that A.ba-R’s virulence is mediated through the coordinated actions of its CPS, multiple transmembrane secretion systems, and nutrient acquisition systems, which collectively promote non-inflammatory apoptosis. This unique combination of virulence factors enhances pathogenicity through three synergistic mechanisms: suppression of host inflammatory responses, induction of apoptosis, and facilitation of host niche adaptation.

During A.ba-R infection, we observed significant upregulation of genes involved in amino sugar and nucleotide sugar metabolism. These pathways are essential for the biosynthesis of bacterial surface structures [58,59]. Specifically, amino sugars (e.g. N-acetylglucosamine) are critical precursors for cell wall biosynthesis, while nucleotide sugars are substrates for polysaccharide production, including CPS and biofilm matrix components [60]. We propose that this metabolic reprogramming enhances bacterial virulence through multiple mechanisms: CPS-mediated immune evasion, biofilm-associated antibiotic resistance, and potentially improved bacterial adhesion via modified cell surface structures [53].

CPS, a key virulence factor ubiquitous among pathogens, contributes to bacterial pathogenesis through multiple mechanisms: (i) modulating biofilm dynamics to promote persistent infections [53]; (ii) conferring resistance to antimicrobial peptides and complement-mediated killing [61]; (iii) inhibiting phagocytosis by innate immune cells while enhancing intracellular survival [62]; and (iv) providing broad antimicrobial resistance [63]. In this study, the enhanced virulence isolate A.ba-R showed a consistent set of phenotypes: a prominent CPS layer, increased survival in host environments, and resistance to both serum killing and phagocytosis. These features are likely to facilitate the establishment of persistent infection. These findings are consistent with established literature demonstrating that CPS shields A. baumannii from host immunity by reducing complement C3 deposition and opsonin binding, thereby diminishing phagocytic uptake by neutrophils and macrophages [64,65]. Clinically, CPS-expressing isolates exhibit greater serum resistance and virulence in vivo compared with CPS-deficient mutants [65]. Similarly, Philip et al. reported that clinical isolates with thick capsules possess superior resistance to desiccation and disinfectants, alongside enhanced pathogenicity in mice, whereas capsular mutants are rapidly cleared [66]. Notably, exposure to sub-inhibitory concentrations of carbapenems can trigger CPS overproduction, which enhances complement evasion and increases mortality in systemic infection models [67].

Furthermore, CPS can directly induce apoptosis via the mitochondrial-dependent pathway [68], playing a crucial role in dictating the host cell death modality. Strains with a thick capsule are more likely to induce non-inflammatory apoptosis, whereas those with sparse CPS preferentially trigger inflammatory pyroptosis [69,70] (Figures 2(c, g), 4(b)). Apoptosis mitigates inflammation and may promote long-term bacterial survival by limiting immune activation. In contrast, pyroptosis results in the release of proinflammatory mediators and facilitates robust immune clearance [32]. In this study, we demonstrate that the heavily encapsulated A.ba-R strain triggers robust caspase-3 cleavage in primary macrophages, and its virulence is markedly attenuated in Caspase-3^-/- mice, establishing apoptosis as a critical virulence accelerator. Conversely, the thin-CPS A.ba-S strain elicits caspase-1 activation, IL-1β release, and rapid bacterial clearance. Thus, by modulating CPS thickness, A. baumannii can toggle between silent apoptotic persistence and immunogenic pyroptotic elimination, representing a sophisticated tunable strategy for niche adaptation.

During bacterial infection, apoptosis operates as a double-edged sword: it can restrict microbial dissemination or be subverted to favor pathogen survival [9]. Group A Streptococcus secretes streptolysin S to trigger macrophage apoptosis, thereby silencing cytokine production, enhancing bloodstream survival, and increasing virulence during systemic infections [17]. Similarly, Yersinia species employ YopJ to induce macrophage apoptosis, which facilitates the establishment of infection and evasion of host immune responses [71]. Likewise, Shigella flexneri promotes invasion and dissemination by inducing apoptosis in intestinal epithelial cells, thereby disrupting intestinal mucosal integrity [72]. Our study demonstrates that A.ba-R infection triggers apoptosis in both BMDMs and infected mouse tissues. Key observations supporting this conclusion include markedly reduced bacterial loads in the brains and spleens of Caspase-3^-/- mice compared with WT controls and the absence of apoptotic signals in Caspase-3^-/- tissues, as confirmed by western blot analysis. These findings collectively suggest that A.ba-R exploits host apoptosis as a dissemination strategy. While the exact mechanism of apoptotic induction requires further investigation, potential contributors may include CPS-mediated pro-apoptotic signaling, effector molecules from bacterial secretion systems, and the synergistic action of multiple virulence factors. Future studies should aim to identify the specific bacterial apoptogenic factors, characterize the targeted host cell pathways, and develop mechanism-based therapeutic interventions against A.ba-R infections.

Funding Statement

This work was supported by the National Natural Science Foundation of China [2022YFC2305000, 82125021, 82472293].

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

All data and supplementary tables that support the findings of this research are publicly available in the Figshare repository via the persistent DOI: https://doi.org/10.6084/m9.figshare.30397396 [73].

Supplemental data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/21505594.2026.2658909