The bacterial strains and plasmids utilized in this study and their characteristics are outlined in Table 1 . The bacterial A. baumannii ATCC 17978 strain utilized in this study was UN variant harboring AbaAL44 fragment (Wijers et al., 2021). The bacterial strains were consistently maintained in Luria-Bertani (LB) broth or agar. Carbenicillin at a concentration of 100 μg/mL, Kanamycin at 50 μg/mL, and Hygromycin at 300 µg/mL were added when necessary (Chen et al., 2017; Hua et al., 2021). Human Bronchial Epithelial Cell line (HBEpiC) and alveolar epithelial cell line A549 were cultured in Dulbecco’s modified Eagle’s medium (DMEM). The human macrophage cell line THP-1 was maintained in RPMI 1640 and supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin. These strains and cells were supplied by the State Key Laboratory of Pathogen and Biosecurity in China. Plasmid pYMAb2-hyg was offered by the Key Laboratory of Microbial Technology and Bioinformatics of Zhejiang Province.

Genomes of A. baumannii ATCC 17978 were downloaded from the National Center for Biotechnology Information Public Database (https://ftp.ncbi.nlm.nih.gov) (GenBank: GCA_001077675.1) and verified by Sanger sequencing for the targeted genes. Amino Acid Sequence of PAAR-M protein ACX60_RS11625 and PAAR-C protein ACX60_RS09070 were aligned using multiple sequence alignment tools in https://toolkit.tuebingen.mpg.de/tools/ClustalΩservice.

The tagP gene ACX60_RS09070 of A. baumannii ATCC 17978 strain was deleted from the genome as previously described with minor modifications (Tucker et al., 2014). Briefly, a 2167bp DNA fragment was constructed using fusion PCR, which included 368 bases upstream and 600 bases downstream of the ACX60_RS09070 coding sequence (CDS). Primers 09070upf/r and 09070dnf/r were used to amplify the flanking regions, while primers Kanf/r were used to amplify the kanamycin resistance cassette from the pKD4 plasmid. The purified PCR product was then electroporated into A. baumannii carrying RecAb on pMMB67EH (pAT02 plasmid), which was maintained with carbenicillin. The kanamycin cassette in successful recombinants was deleted using pMMB67EH expressing the FLP recombinase (pAT03 plasmid). PCR was used to confirm the loss of the kanamycin cassette and the plasmids pAT02 and pAT03, resulting in the generation of the tagP null A. baumannii ATCC 17978ΔtagP::FRT (ΔtagP). A pair of primers 09070Rvf/r amplifying the ACX60_RS09070 and its 197bp upstream sequence with promoter were used to construct complement strains, and the pYMAb2-hyg plasmid was used as previously described to generate complement strain ATCC 17978ΔtagP::com (ΔtagP::com) and ATCC 17978ΔtagP-pYMAb2-hyg (ΔtagP-p) (Hua et al., 2021). PCR and sequencing were used to confirm the accuracy sequence of the tap gene. The primers adopted in this experiment were listed in the Supplementary Table 1 .

The overnight cultures were 1:100 diluted with 60 mL of fresh LB in a 200 mL Erlenmeyer flask and maintained at 37°C with a shaking speed of 200 r/min. The absorbance at OD₆₀₀ was automatically measured every 2 min for 36 h, and the results of the area under the curve from three trials were presented as the mean and standard deviation.

The biofilm formation ability of the strains was assessed using a previously established method (Tucker et al., 2014), with slight modifications that the 96-well polystyrene cell culture plate was incubated for 72 hours at 28°C without shaking, the bacterial fluid in twelve parallel repeating wells were determine the optical density (OD)₆₂₀ to estimate the total cell biomass before being removed, and the remaining biofilms were rinsed and stained with 0.1% crystal violet according to the previously established method (Tucker et al., 2014), biofilm quantities were subjected to OD₅₉₀/OD₆₂₀ ratio.

The overnight cultures were 1:100 diluted with LB broth, and 1 μL of the bacterial suspension was placed at the center of the motility assay plate according to the reference (Chen et al., 2017). For swarming motility test, plates with 10 g/L tryptone, 5 g/L NaCl, and 0.3% Noble agar (Becton Dickinson, Sparks, MD, USA) were used; for twitching motility test, Mueller-Hinton (MH) medium with agar at a concentration of 0.4%.

LB medium containing a concentration gradient of 1.0, 2.0, and 3.0 M glucose (C₆H₁₄O₆) was utilized to create a hyperosmotic environment; likewise, a concentration gradient of 10.0%, 15.0%, and 20.0% sodium chloride to form a high salinity environment; LB medium with pH values of 7.0, 6.0, 5.0, 4.0, and 3.5 to create different pH environments. 100 μL of the bacteria with an OD of 1.0 were added to 1.9 mL of the above medium for simulating at the setting time points. Pellets from 1 mL of the bacterial mixture were collected and washed twice with PBS before resuspended in the same volume of 0.01M pH7.4 PBS for plate counting. For the hydrogen peroxide (H₂O₂) resistance experiment, 200 mM and 20 mM H₂O₂ were added into PBS instead of LB medium. For the heat and cold temperature experiments, shake the bacteria in LB medium at 50°C for 30 min and at 4°C for 24 h, respectively. Further experiments with bacteria kept in pH 3.5 LB medium for 5, 15, and 30 minutes or PBS containing 20 mM H₂O₂ for 15, 30, 45, and 60 minutes were carried out to investigate the survival difference between the wild-type and the knockout strains.

The minimal inhibitor concentrations (MICs) of common antibiotics ( Supplementary Table 2 ) against A. baumannii were measured by the broth microdilution test card (GN13, bioMérieux, Lyons, France) according to the instructions of the manufacturer and the Clinical and Laboratory Standards Institute (CLSI, 2021; Humphries et al., 2021). The MICs of Gentamicin, Tigecycline and Colistin were further confirmed by the E-test (bioMérieux). E. coli ATCC 25922 was used as the quality control.

Adhesion and invasion ability to HBEpiC and A549 cells were assayed. As recommended by Tang J et al (Tang et al., 2020), firstly, about 5 × 10⁵ cells per well were seeded in six-well plates in the culture medium of DMEM (Dulbecco’s modified Eagle’s medium) supplemented with 10% fetal bovine serum (Gibco) in a 37°C incubator with 5% CO₂. For the infection experiments, cells adhesion and tightly connected growth until covered the bottom of the plate up to 80%, followed by bacterial infection with a multiplicity of infection (MOI) of 20. The infected cells were incubated for 2 h, then the infected monolayers were washed three times with PBS and lysed in 500 μL of 0.5% Triton X-100 before bacterial adhesion determination. To measure their invasion capability, cells in each well were supplemented with 200 μg/mL of gentamicin and incubated for 15 min; after that, dilutions of the lysates were plated on the LB agar for measurement. At each time point, two six-well plates were used for each strain, with three wells setting as controls without bacterial infection to exclude contamination by other bacteria during the experimental process. Three wells were used for adhesion experiments, three wells were used for invasion experiments, and the remaining one parallel well was used for Wright-Giemsa staining. Each strain was tested in triplicate.

Cell cultures were performed in accordance with the protocol in the references (Subashchandrabose et al., 2016; Sycz et al., 2021). Initially, 2 × 10⁵ cells were inoculated in each well of six-well plates and grew for 24 h before the test, where the culture medium was RPMI 1640 with 10% fetal bovine serum and 100 ng/mL PMA. After washing, the cells were co-cultured with the bacteria at a MOI of 50. Three technical replicates were exercised to assess the phagocytic ratio at 30 min and the survival ratio at 4, 8, and 12 h, and four identical plates were prepared for each assay at different time points (T₃₀, T₂₄₀, T₄₈₀, and T₇₂₀); dilutions of the bacteria were incubated with THP-1 cells for 30 min to allow for phagocytosis; 200 μg/mL Gentamicin was used to eliminate the extracellular bacteria. Phagocytes engulfing bacteria were collected at 15 min or 4, 8, and 12 h after the addition of gentamicin. At each time point, cells were lysed with 0.5% Triton X-100 and plated on the LB agar.

The serum resistance assay was determined following the procedure in previous research (Subashchandrabose et al., 2016). In brief, the overnight cultures were adjusted to a final concentration of 0.5 McFarl with PBS, and then 1:10 diluted to a final volume of 100 μL in either 90% serum or 90% heat-inactivated serum. Dilutions were plated at 0, 60, 120, and 180 min. The complement strains were cultured overnight in LB without antibiotic selection.

Expression of Hcp and its polyclonal antibody were prepared as previously described (Liu et al., 2016). Briefly, the hcp gene fragment from ATCC 17978 genomic DNA was amplified using HcpF/R primers ( Supplementary Table 1 ), which introduced the Nde1 and BamH1 restriction sites. The amplified fragments were then inserted into the pET-28a(+) expression vector with the addition of 0.5 mM IPTG to induce protein expression at 22°C. Following sequential chromatographic separation, Hcp was achieved with a high purity of > 95%, assessed by Coomassie Blue-stained SDS-PAGE. After that, the recombinant protein was diluted in sterile PBS and then mixed with Freund’s complete adjuvant for injection into the auricular vein of female New Zealand rabbits. The anti-Hcp polyclonal antibodies were used to detect native Hcp in the whole-cell lysates or supernatants by Western blotting with reference to the literatures (Shneider et al., 2013; Liu et al., 2016), and the protein RNAP was set as the control. Bacterial LB cultures were incubated at 28°C for 20 hours, followed by centrifugation to separate the bacterial cells and culture supernatant. 200 µL bacterial pellet was washed with 1 mL PBS for three times, then resuspended in 1 mL lysis buffer (6 mM NaCl, 1 mM NaH₂PO4, 10 mM imidazole) containing a protease inhibitor at a ratio of 100:1 and sonicated to prepare the whole-cell lysates. Supernatant was filtered through a filter with 0.22 µm membrane. Then 5 µL of bacterial protein or 20 µL of supernatant was mixed with loading buffer to prepare WB samples.

The competition of bacteria of WT, ΔtagP, with E. coli MG1655 and MJ109 harboring pKD4 was assayed according to steps in the publication (Basler et al., 2013). Briefly, the killer and prey cells were mixed in equal proportions (1:1 ratio) and spotted onto the LB-agar plates. After incubating at 37°C for 4 h for the mixture with MJ109 and 6 h for the mixture with MG1655, the mixed cells were recovered in fresh LB and vigorously vortexed to release cells from the agar. The mixtures were then diluted and plated onto LB-agar plates with 50 μg/mL Kanamycin or 15 μg/mL Gentamicin for prey survival selection.

Following the protocol presented in the reference (Tang et al., 2020), A. baumannii bacteria were adjusted using PBS to match the turbidity of 0.5 McFarland. Twenty G. melonella of the same size were randomly selected to test the virulence of each bacterial strain. 20 μL of the suspension of each bacterium strain or PBS was injected into each G. melonella. After that, the G. melonella was preserved in a dark environment at 37°C for 4 days. The G. melonella was considered dead if it did not respond to gentle probing.

RNA sequencing analysis was conducted by Macro & Micro-test Bio-Tech Co., Ltd. (Beijing, China) (Hua et al., 2017). In brief, A. baumannii ATCC 17978 wild-type and the tagP null mutant strains were cultured overnight in LB broth at a temperature of 28°C. The bacterial cultures were then diluted 1:100 in fresh LB broth and incubated at 28°C for 4~5 hours and collected by centrifugation at 2,500 × g at 4°C for 15 minutes. RNA extraction was performed with the Pure Link™ RNA Mini Kit (Invitrogen, Carlsbad, CA, USA). The resulting sequencing reads were aligned to the ATCC 17978 genome using Bowtie2 v2.3.4.3 (Langmead and Salzberg, 2012). HTSeq v0.6.1 was employed to quantify the number of the reads mapped to each gene. Differential gene expression analysis was conducted using the DESeq R package (version 1.18.0) (Robinson et al., 2010). P-values were adjusted with the Benjamini and Hochberg method. Genes with an adjusted P-value < 0.05 and a log2 fold change > 1 were considered to be differentially expressed.

LC-MS proteomic analysis was executed by PTM BioLab Co., Inc. (Hangzhou, China) (Song et al., 2020). Cultures were prepared as the same for RNA-Seq analysis. Three replicate cultures were used for each group. The resulting MS/MS data were processed with the MaxQuant engine (v.1.5.2.8, https://www.maxquant.org) searching against the proteomic database of ATCC 17978 in NCBI with GenBank accession number GCA_001077675.1. The parameters and the false discovery rate were set referred to the literature (Fei et al., 2022).

The biofilm formation and surface motility ability of A. baumannii was closely related to Csu pili (Chen et al., 2017; Moon et al., 2017). Since the biofilm formation and surface motility ability of the tagP null mutant strain was significantly reduced, we compared the expression of Csu pili genes between wild-type and the tagP null mutant strains cultured in LB broth at a temperature of 37°C by qRT-PCR assay. The RNAs were reverse transcribed into cDNA using SynScript III RT SuperMix (Tsingke Biotechnology Co., Ltd., Beijing, China). The genes of Csu pili gene cluster for qRT-PCR analysis were csuAB, csuA, csuB, csuC, csuD and csuE. The level of RNA polymerase β-subunit (rpoB) mRNA was used as a control. The primers for qRT-PCR were listed in the Supplementary Table 4 .

Student’s t-test was employed to compare the mean value variances between the two groups. Chi-square and Fisher’s exact test were utilized for qualitative data analysis. Mean values and SEM from the three independent experiments were analyzed by one-way ANOVA, followed by Dunnett’s test for multiple comparisons. Survival curves were constructed using the Kaplan-Meier method and assessed using a log-rank (Mantel-Cox) test. All statistical analysis and plotting were based on R 4.3.0, and the Tidyverse and ggsurvey packages were used for data visualization (Xu et al., 2021). For all tests, a P value of < 0.05 was considered statistically significant.

Animals in this study were treated following China’s guidelines for laboratory animal welfare and were conducted in accordance with the regulations outlined in laboratory animal permit No. SCXK (Jing) 2021-0006, obtained from Beijing Vital River Laboratory Animal Technology Co. Ltd. Human serum collection was approved by the ethical committee of Xiangya Hospital of Central South University (No. 202308645).

Among the genes encoding PAAR proteins in A. baumannii, ACX60_RS11625 (encoding PAAR-M) is situated in the main T6SS cluster ( Figure 1A ), while ACX60_RS09070 or tagP gene (encoding PAAR-C) resides in the auxiliary cluster ( Figure 1B ). Sequence alignment revealed that conserved sequence existed in their N-terminal ( Figure 1C ) while PAAR-M lacks the complex C-terminal structure possessed by PAAR-C, which suggests that the two PAAR proteins may have different biological functions.

The impact of the tagP gene on the growth of A. baumannii was investigated by comparing the in vitro growth rates of A. baumannii ATCC 17978 wild-type and the tagP null mutant strains. The growth kinetics of each strain in LB broth at 37°C was shown in Figure 2A . There was no significant difference in growth between the presence (wild-type) and absence (ΔtagP) of the tagP gene in A. baumannii.

After incubating at 28°C for 72 hours, biofilm formation between A. baumannii strains with and without tagP gene was observed and compared, including the film on the surface of LB liquid, the adhesion to polystyrene microtiter plate well, and the biofilm quantification using crystal violet staining. The results manifested that the wild-type strain formed clear biofilms and adhered to the microtiter plate wells, while the tagP null mutant strain did not form visible biofilms on the surface of the culture or the wells. Quantitative analysis using crystal violet confirmed significantly lower biofilm formation in the tagP null mutant strain compared to that in the wild-type strain ( Figure 2B ). Notably, biofilm formation was restored in the tagP complementation strain, in contrast, ΔtagP-p strain did not restore this phenotype. Therefore, we could conclude that tagP is crucial for A. baumannii to develop biofilm.

Although A. baumannii is commonly considered “non-motile”, its surface motility can be triggered under specific conditions and involved in its virulence. In order to investigate the impact of tagP on the surface motility of A. baumannii, the bacteria in the exponential phase were inoculated onto motility agar plates. Subsequent observation revealed that both twitching and swarming surface associated motility were significantly reduced in the tagP deletion mutant, whereas the complemented strains but not the ΔtagP-p strains exhibited similar surface motility to the wild-type ones ( Figure 2C ). These findings suggest that tagP participates in the surface motility of A. baumannii.

In various simulated extreme environments, the deletion of tagP was not found to significantly affect the adaptation of A. baumannii to changes in osmotic pressure ( Figure 3A ), salt concentration ( Figure 3B ), or temperature ( Figures 3C, D ); however, it showed lower tolerance than the wild-type strain to H₂O₂ stimulation ( Figure 3E ) or to acidic conditions with pH 3.5 ( Figures 3F, G ), while the complemented strain was able to restore tolerance in such conditions.

The MICs of ceftazidime (CAZ), cefepime (FEP), meropenem (MEM), imipenem (IMP), piperacillin/tazobactam (TZP), ticarcillin/clavulanic acid (TIM), piperacillin (PIP), levofloxacin (LVX), ciprofloxacin (CIP), gentamicin (GEN), tobramycin (TOB), amikacin (AMK), sulphamethoxazole (SXT), doxycycline (DOX), minocycline (MH), tigecycline (TGC) and colistin (COL) were measured. Of all results, there was no significant difference observed between wild-type and tagP null mutant strains ( Supplementary Table 2 ).

HBEpiC bronchial epithelial cells and A549 alveolar epithelial cells were used to assess the tagP gene on the adhesion and invasion capacity of A. baumannii. It was observed that all of the tagP null mutant strains, the complemented strains, and the wild-type strains exhibited adhesion to HBEpiC and A549 cells; moreover, the tagP null mutant strains showed significantly lower adherence than the wild-type strain ( Figures 4A, B ), and the adherence was restored in the complemented strains. On the other side, all strains showed weak invasiveness toward either HBEpiC or A549 cells, and no significant differences were found among them ( Figures 4C, D ).

The gentamicin protection assay was used to test whether A. baumannii mutants with impaired in vivo fitness resulted in compromised intramacrophage survival. The results revealed that within 30 min of phagocytosis, the tagP null mutant strains exhibited a significantly low phagocytosis rate compared to the wild-type strain ( Figure 5A ), while the complemented strains were restored from the phagocytosis defect. As some groups have shown that some strains of Acinetobacter can replicate inside macrophages (Sato et al., 2019; Sycz et al., 2021), both the wild-type and the null strains survived after 4, 8, and 12 h, and without significant difference in growth rate between them ( Figure 5B ). Thus, finally, a significantly higher load of the wild-type strains was isolated from macrophages ( Figure 5C ).

Heat inactivation is a common method to deactivate complement, the main factor that kills bacteria in serum. Investigation of the survival difference between the wild-type and the tagP null mutant strains in the natural human serum or the heat-inactivated serum proved that the wild-type strain had a significant reduction in viability in natural serum but insignificant difference in the heat-inactivated serum after 60, 120, and 180 min upon exposure ( Figure 6 ), indicating that complements may involve killing the bacteria in serum. Obviously, compared to the parental strain, the tagP null mutant strains appeared to be more susceptible to natural serum with the extension of exposure time ( Figure 6A ). However, such survival reduction was not discovered in the heat-inactivated serum ( Figure 6B ), suggesting that the tagP null mutant strain is highly susceptible to complement mediated killing and the tagP gene may specifically contribute to the complement resistance.

We assessed whether TagP affects the expression and secretion of Hcp in A. baumannii. As shown in Figure 7 , Hcp protein was identified in the whole-cell lysates of the wild type strain and the complemented strain, but not in the tagP null mutant and ΔtagP-p strain. Western blotting failed to detect Hcp protein in the culture supernatants of all strains including the wild type strain. Weber et al. discovered that two regulators encoded by the plasmid pAB3 can suppress the expression and secretion of Hcp in ATCC 17978 (Weber et al., 2015). Our results are consistent with their findings, and further imply that the presence of tagP gene is critical for the expression of Hcp in A. baumannii.

To test whether tagP influences the killing activities against the E. coli prey, we used MJ109 and MG 1655 as prey cells to conduct multiple repeated experiments, and found out that the deletion of tagP did not lead to significant changes in killing efficiency ( Figure 8 ). Weber et al. proved that the presence of pAB3 plasmid suppressed the bacteria-killing ability of ATCC17978, the pAB3^- ATCC17978 only showed residual kill ability to E. coli (Weber et al., 2015). In our assays, the wild type strain and the tagP null mutant failed to kill the high titers of E. coli. Our results demonstrated that strains absence of tagP still possess the bacterial antagonistic ability, implying that multiple effectors might be involved in the prokaryotic competitive killing in A. baumannii other than Hcp or T6SS.

G. mellonella is often used to study host-pathogen interactions. The G. mellonella model is considered reliable for studying the virulence and pathogenesis of A. baumannii (Peleg et al., 2009). As shown in Figure 9 , significant differences (P < 0.05) in virulence targeting G. mellonella were uncovered between the wild-type and the tagP null mutant strains, where the G. mellonella infected with wild-type strain exhibited a mean death rate of 55%, while less than 20% when infected with the tagP null mutant strains.

A total of 501 upregulated and 213 downregulated TagP-dependent genes were identified by RNA-seq analysis satisfying the criteria of log2 Fold change > 1 and padj < 0.05 ( Supplementary Table 3 , tab A and B, respectively, and Supplementary Table 5 ). Based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) classifications, genes were broadly distributed among the KEGG categories ( Supplementary Figure 1 ). Nevertheless, the majority of them correlative to tagP fell into the “microbial metabolism in diverse environment category”, where the upregulated genes were mainly gathered in the amino acid metabolism pathway, while the downregulated clustered in the glycometabolism and ribosome. Simultaneously, quantitative proteomic analysis determined a total of 108 upregulated and 104 downregulated TagP-dependent proteins as presented in Supplementary Table 3 , tab C and D, respectively, including 61 upregulated and 59 downregulated proteins fulfilling the criteria of log2 Ratio > 1 and P value < 0.05. In addition, these differentially expressed genes went through enrichment analysis to further determine their biological functions ( Supplementary Figure 2 ).

Genes within the T6SS main cluster that exhibited significant changes in transcription levels were shown in Table 2 ; significant expression changes ranged from tssH (ACX60_RS11620) to asaA (ACX60_RS11680), indicating a closely functional relationship between tagP and the T6SS genes; TssB, TssC, Hcp, and AsaA exhibited consistent changes at the protein level.

Twenty-six genes outside the T6SS main cluster were found to be consistently upregulated at both the transcription and the protein level, and 31 genes were downregulated, as presented in Tables 3 , 4 . Notably, H-NS was significantly upregulated in the tagP null mutant. Several membrane proteins, including the OmpA family protein, Trimeric autotransporter adhesin Ata, type 1 fimbrial protein, surface antigen protein SurA1, and a gene cluster (ACX60_17480-17515) responsible for lipopeptide encoding, were significantly downregulated.

To investigate whether Tap affects the expression of Csu pili genes, which were closely related to the biofilm formation and motility ability of A. baumannii, we utilized qRT-PCR to detect the expression of Csu genes in A. baumannii with or without TagP under dynamic culture conditions. The results ( Supplementary Figure 3 ) showed that the Csu gene cluster, including CsuAB to CsuE, was significant downregulation in the tagP null mutant strains.