Vaccination with Acinetobacter baumannii adhesin Abp2D provides protection against catheter-associated urinary tract infection

Abstract

Catheter-associated urinary tract infections (CAUTIs) contribute greatly to the burden of healthcare-associated infections. Acinetobacter baumannii is a Gram-negative bacterium with high levels of antibiotic resistance that is of increasing concern as a CAUTI pathogen. A. baumannii expresses fibrinogen-binding adhesins (Abp1D and Abp2D) that mediate biofilm formation on catheters, which become coated with fibrinogen upon insertion. Here we develop a protein subunit vaccine against the Abp1D and Abp2D receptor binding domains (RBD) and show that vaccination significantly reduces bacterial titers in a female mouse model of CAUTI. We further demonstrate that immunity to Abp2D_RBD alone is sufficient for protection. Mechanistically, we define the B cell response to Abp2D_RBD vaccination, demonstrate that passive immunization with Abp2D_RBD-immune serum transfers immunity to naïve mice, and show that Abp2D_RBD-immune serum inhibits bacterial binding to fibrinogen-coated catheters. This work represents an antibiotic-sparing strategy for the prevention of A. baumannii CAUTI which has an important role in the global fight against antimicrobial resistance.

Acinetobacter baumannii is a multi-drug-resistant pathogen of urgent international concern. Here, the authors develop a protein subunit vaccine which prevents A. baumannii catheter-associated urinary tract infections in mice by inhibiting Abp2D, a key adhesive virulence factor.

Introduction

The U.S. Centers for Disease Control and Prevention (CDC) estimates that 1 out of every 31 hospitalized patients in the U.S. will acquire at least one healthcare-associated infection (HAI) over the course of their care¹. HAIs are problematic because they lead to an increased burden of morbidity and mortality for patients, cost the US healthcare system an estimated $35.7 billion per year, and contribute to antibiotic overuse that drives the development of antimicrobial resistance². One of the most common types of HAIs are catheter-associated urinary tract infections (CAUTIs). The risk of CAUTI increases by 3–7% with each subsequent day of catheterization and approaches 100% in patients catheterized for >30 days^3,4. More than 30 million Foley (urinary) catheters are used annually in the United States^5,6. Because catheter use is so ubiquitous, CAUTIs rank in the top 3 causes of HAIs in most epidemiologic studies and make up 20−40% of all HAIs^7,8. Unfortunately, despite concerted infection mitigation efforts by public health agencies and healthcare facilities, the rate of CAUTIs continues to rise and increased by 5% from 2019-2021 in the United States¹.

Acinetobacter baumannii is a Gram-negative bacterium implicated in multiple types of HAIs. While it is best known for causing ventilator-associated pneumonia, it is increasingly recognized as an important CAUTI pathogen⁹. Several single-center studies have identified A. baumannii as a leading cause of CAUTI in their facilities^9,10. Indeed, recent studies have demonstrated that 17% of published A. baumannii isolates originated in the urinary tract and up to 2% of the healthy population may exhibit A. baumannii asymptomatic bacteriuria^11,12. At the same time, the level of antimicrobial resistance identified in A. baumannii isolates is on the rise, with most isolates resistant to at least one antibiotic class and many isolates displaying multi-drug resistance¹³. As a result, the CDC and World Health Organization have classified carbapenem-resistant A. baumannii as an “urgent threat,” which is the highest threat level¹⁴. Together, the prevalence of A. baumannii coupled with its multi-drug resistance profile have emphasized the critical need for antibiotic-sparing therapeutics for A. baumannii CAUTI.

Many Gram-negative bacteria produce hair-like proteinaceous fibers called pili, tipped by specialized adhesins that recognize receptors with stereochemical specificity to determine host and tissue tropisms. The class of pili most implicated in host-pathogen interactions are the chaperone-usher pathway (CUP) pili¹⁵. CUP adhesins are two-domain proteins with an amino terminal receptor binding domain (RBD) and a pilin domain that links the adhesin to the pilus rod. Importantly, new therapies that neutralize the function of the receptor binding domain of CUP pilus adhesins in a variety of pathogens have been successful in preclinical models and early human clinical trials^16–25. Our recent studies of A. baumannii CAUTI pathogenesis have revealed that A. baumannii CUP adhesins are critical in catheter colonization and thus may represent promising drug targets^9,26. When a urinary catheter is inserted into the bladder, it induces inflammation and leads to deposition of host proteins, such as fibrinogen, onto the surface of the implant^{24,25,27–29}. A. baumannii have evolved two CUP pili, Abp1 and Abp2, tipped with fibrinogen binding adhesins Abp1D and Abp2D respectively. The majority of published A. baumannii genomes encode one or both of these pilus operons²⁶. Both of these adhesins have been shown to bind fibrinogen and to be critical in a mouse model of CAUTI²⁶. Therefore, we hypothesized that targeting the Abp1D/Abp2D adhesins with an adhesin-based vaccine, comprised of their receptor binding domains Abp1D_RBD and Abp2D_RBD, would prevent A. baumannii CAUTI pathogenesis. We chose a vaccine platform that elicits systemic, rather than mucosal, immunity because the unique pathophysiology of CAUTI allows serum IgG to flow into the bladder lumen at high levels²⁵ and because systemic vaccines have a track record of efficacy against other mucosal pathogens^20,21,30,31.

Here, we present evidence that vaccination with recombinant Abp2D_RBD provides protection from A. baumannii CAUTI in a mouse model. Immunity conferred by a previous A. baumannii infection was used to establish a baseline for vaccine performance. We demonstrate that a vaccine formulation containing both adhesins provided protection from A. baumannii CAUTI that was equal to natural infection. Further, we show that immunity to Abp2D_RBD alone was sufficient for protection. We demonstrate that our Abp2D_RBD vaccine generates robust memory B cell and bone marrow plasma cell responses, that antibody-mediated protection is transferable to naïve mice via passive immunization, and that Abp2D_RBD-immune serum directly inhibits binding of A. baumannii to fibrinogen-coated catheters. Finally, we show that abp2d sequences across the population of published A. baumannii genomes are dominated by a handful of strains with the ACICU allele used in our vaccine being the most common. This work establishes that an adhesin-based vaccine may be a promising strategy for multidrug-resistant A. baumannii CAUTI and could directly contribute to the arsenal of antibiotic-sparing therapeutics needed to meet the urgent threat of antibiotic-resistant A. baumannii.

Results

A history of infection provides protection against subsequent A. baumannii CAUTI despite minimal adhesin-specific IgG responses

We have previously shown that adhesins Abp1D and Abp2D are critical virulence factors for A. baumannii CAUTI pathogenesis²⁶. However, prior studies did not examine the immune response to Abp1D/Abp2D during infection, nor did they consider whether a history of infection would provide any degree of protection from a subsequent challenge infection. To investigate this question, we catheterized C57BL/6 mice, infected with A. baumannii strain ACICU or mock-infected with PBS, and then treated with apramycin at week 5 to clear the infection (Fig. 1a). We then analyzed the serum and bladders of treated mice, termed convalescent mice, for IgG specific to Abp1D_RBD and Abp2D_RBD to determine if an immune response was elicited against these adhesins. While two individuals developed a low level of Abp1D or Abp2D-specific IgG (AUC ~ 0.005), most animals did not generate any appreciable antigen-specific IgG response (AUC < 0.002) (Fig. 1b). The low levels of antigen-specific IgG to the Abp1 and 2 adhesins was not surprising, as this has been observed in other pilus systems²⁰. Each pilus is comprised of thousands of rod subunits tipped by a single adhesin. When mice are infected with whole bacterial cells or immunized with whole pili, the antibody response is skewed towards the much more abundant rod subunit and the response to the adhesin is minimal^20,24. Despite the low adhesin-specific IgG response, CAUTI mice treated with antibiotics and subsequently re-infected displayed a reduction in bladder and catheter bacterial titers of 1.1 log (bladder) and 1.3 log (catheter) compared to naïve mice (Fig. 1c). There was no significant reduction in kidney infection, however we do not consider the murine CAUTI model a reliable model for ascending kidney infection as the transurethral catheterization and infection process results in a supraphysiologic bolus of refluxing bacteria into the kidney that does not recapitulate what happens in humans (Supplementary Fig. 1a). The lack of IgG titers suggests that factors other than adhesin-specific IgG, such as epithelial trained immunity or T cell-mediated immunity, are likely responsible for the modest protective effect^32–37.

Fig. 1. Immunity from prior A. baumannii infection.

a Seven to nine week old C57BL/6 mice were catheterized and infected with A. baumannii strain ACICU or mock-infected with PBS. Urine was collected to monitor infection status. Mice received 10 days of oral apramycin (1 g/L) at week 5 to clear the infection. At week 8, mice were catheterized and challenged with A. baumannii strain ACICU and sacrificed 24 h post infection. b Bladder homogenates and serum from weeks 3, 5, and 8 were assayed for Abp1D_RBD- (purple) and Abp2D_RBD-specific IgG (teal) by ELISA. Heatmaps were generated by calculating the area under the curve for each serum/tissue sample. c Bladder and catheter titers were enumerated for mock (dark circles) and ACICU-infected mice (open circles). Dashed lines indicate limit of detection. Solid lines indicate median +/− 95% CI. Data shown are pooled from two independent experiments, n = 16 (mock) and n = 17 (ACICU). Two-tailed Mann-Whitney U-test. Created in BioRender. Timm, M. (2025) https://BioRender.com/8ycspp0. Source data are provided as a Source Data file.

Immunization with A. baumannii Abp1/Abp2 adhesins provides protection from CAUTI

Mutations in abp1/abp2 attenuate virulence^9,26. Thus, based on work with other adhesin-based vaccines^20,24,25, we hypothesized that immunization with the A. baumannii adhesins, Abp1D and Abp2D, might confer increased protection relative to natural immunity. We purified the Abp1D and Abp2D receptor binding domains (RBDs) as previously described²⁶ and used the proteins to immunize C57BL/6 mice. Four weeks after the final immunization, mice were catheterized and infected with strain ACICU (Fig. 2a). Although IgG levels in the urinary tract are low at baseline, the inflammation elicited by catheterization leads to an influx of serum IgG, which has been shown to provide protection in other CAUTI vaccine models^24,25. Therefore, we collected serum at: (i) week 4 (prior to 2nd immunization); (ii) week 8 (prior to 3rd immunization); and (iii) week 12 (at time of sacrifice), to test for Abp1D_RBD- and Abp2D_RBD-specific IgG. We also tested bladder and kidney homogenates collected at sacrifice for Abp1D_RBD- and Abp2D_RBD-specific IgG. We note that antibody titers in these organs likely largely reflect the effect of increased blood flow and vascular permeability caused by catheterization and should not be interpreted to indicate local production of antigen-specific IgG. All animals produced a strong IgG response against both Abp1D_RBD and Abp2D_RBD. This response was enhanced with each subsequent immunization to an AUC > 0.08 at week 12. (Fig. 2b and Supplementary Fig. 1b). Since IgA is secreted in the urine and may play a role in immunity in the urinary tract³⁸, we also assayed for antigen-specific IgA and found that it was undetectable in urine, serum and tissues (Supplementary Fig. 2).

Fig. 2. Immunization with Abp1D and Abp2D provides protection from CAUTI.

a Seven to nine week old C57BL/6 mice received 3 adjuvanted doses of 50 ug Abp1D_RBD and 50 ug Abp2D_RBD or buffer alone (mock). Serum was collected at weeks 4 and 8 prior to immunizations, and at week 12 following sacrifice. Four weeks after the third dose, mice were catheterized and challenged with A. baumannii strain ACICU. Mice were sacrificed 24 h after infection. b Bladder homogenates and serum from week 4, week 8, and the day of sacrifice were assayed for Abp1D_RBD (purple) and Abp2D_RBD-specific IgG (teal) by ELISA. Heatmaps were generated by calculating area under the curve for each serum/tissue sample. c Bacterial titers were enumerated from bladders and catheters for mock (circles) and immunized mice (triangles). Dashed lines indicate limit of detection. Solid lines indicate median +/− 95% CI. Data shown are pooled from two independent experiments, n = 19 per group. Two-tailed Mann-Whitney U-test. d Normalized reduction in bacterial titers in the bladders of convalescent (n = 17, open circles) and Abp1D_RBD/Abp2D_RBD immunized mice (n = 19, triangles). Bars indicate median +/- 95% CI. Two-tailed Mann-Whitney U-test. e ELISA AUCs of Abp1D_RBD (purple) and Abp2D_RBD-specific IgG (teal) in serum from convalescent (n = 17, open circles) vs. Abp1D_RBD/Abp2D_RBD immunized mice (n = 19, triangles). Bars indicate mean +/- SD. Created in BioRender. Timm, M. (2025) https://BioRender.com/s2cuhoy. Source data are provided as a Source Data file.

Mice that received Abp1D_RBD/Abp2D_RBD immunizations had significantly lower bacterial titers in bladder tissue (2.5 log reduction) and on the catheter surface (1.4 log reduction) than mock immunized animals (Fig. 2c). The magnitude of the phenotype was comparable to that observed in convalescent mice (Fig. 2d). The degree of antigen-specific IgG in serum and bladder tissue was also much greater in immunized mice compared to convalescent mice (AUC > 0.08 vs. <0.005) (Fig. 2e). These data demonstrate that Abp1D_RBD/Abp2D_RBD vaccination produces greater immunity against two key CAUTI virulence factors and provides a similar level of protection as a prior infection.

Immunization with Abp2D_RBD, but not Abp1D_RBD, is required for protection from A. baumannii CAUTI

As mentioned above, A. baumannii deficient in either Abp1D, Abp2D, or both are attenuated in a CAUTI model²⁶. To test whether immunity to both adhesins is required for protection from CAUTI, we immunized mice with each adhesin individually (Fig. 3a). We tested the cross-reactivity of the IgG response in immunized animals (Fig. 3b) because Abp1D_RBD and Abp2D_RBD share both structural homology and 70% sequence identity²⁶. Each mouse in the Abp1D_RBD-immunized group produced a strong Abp1D_RBD-specific IgG response, but the degree of cross-reactivity with Abp2D_RBD varied between animals, with only ~30% displaying strong cross-reactivity. Similarly, each mouse in the Abp2D_RBD-immunized group produced a strong Abp2D_RBD-specific IgG response, with strong Abp1D_RBD cross-reactivity occurring in ~50% of individuals. Upon catheterization and infection, mice that received Abp2D_RBD immunizations were protected from infection, with a statistically significant decrease in bladder (3.5 log) and catheter (2.1 log) bacterial titers compared to mock-immunized animals (Fig. 3c) and no change in kidney titers (Supplementary Fig. 1c). However, despite high serum levels of Abp1D_RBD-specific IgG (AUC > 0.08), mice that received Abp1D_RBD immunizations were not protected from infection, with bladder and catheter titers equivalent to mock-immunized animals (Fig. 3c). This suggests that the lack of protection from CAUTI in Abp1D_RBD-immunized animals is not due to a lack of immunogenicity. We further explored whether cross-reactivity to Abp1D_RBD may play a role in protection in Abp2D_RBD-immunized animals and found that there was no significant correlation between either serum or bladder Abp1D_RBD IgG and bacterial titers (Supplementary Fig. 3). Our studies indicate that immunization with Abp2D_RBD alone is both necessary and sufficient for protection from A. baumannii CAUTI in this model.

Fig. 3. Immunization with Abp2D, not Abp1D, drives protection from CAUTI.

a Seven to nine week old C57BL/6 mice received 3 adjuvanted doses of 50 ug Abp1D_RBD, Abp2D_RBD, or buffer alone (mock). Serum was collected at weeks 4 and 8 prior to immunizations, and at week 12 following sacrifice. Four weeks after the third dose, mice were catheterized and challenged with A. baumannii strain ACICU and sacrificed 24 h post-infection. b Bladder homogenates and serum from week 4, week 8, and the day of sacrifice were assayed for Abp1D_RBD (purple) and Abp2D_RBD-specific IgG (teal) by ELISA. Heatmaps were generated by calculating area under the curve for each serum/tissue sample. c Bacterial titers were enumerated from bladders and catheters from mock (black), Abp1D_RBD- (purple) and Abp2D_RBD-immunized animals (teal). Dashed lines indicate limit of detection. Solid lines indicate the median +/− 95% CI. Data shown are pooled from two independent experiments, n = 20 (Mock), n = 17 (Abp1D_RBD), n = 18 (Abp2D_RBD). Kruskal-Wallis test with Dunn’s multiple comparisons correction. Created in BioRender. Timm, M. (2025) https://BioRender.com/i4jnxjq. Source data are provided as a Source Data file.

Abp2D_RBD vaccine generates antigen-specific bone marrow plasma cells and splenic memory B cells

A successful vaccine elicits an antibody response that is both high-affinity and long-lasting. We evaluated the immunogenicity of our Abp2D_RBD protein subunit vaccine by examining memory B cells and bone marrow plasma cells in immunized animals (Fig. 4a). Abp2D_RBD-specific memory B cells were detectable in the spleens of all immunized animals (Fig. 4b–d) by flow cytometry. Antigen-specific memory B cells were defined as lymphocytes/single cells/live/CD4^- CD19⁺/IgD^lo/GL7^- CD38⁺/IgM^- IgG1⁺/Abp2D_RBD-bio-SA-APC-Fire750⁺ (Supplementary Fig. 4). In addition, all immunized animals had detectable Abp2D_RBD-specific antibody-secreting cells in their bone marrow as assayed by ELISpot (Fig. 4e, f). We also tested serum and tissue IgG levels in these mice, which were Abp2D_RBD-vaccinated but did not undergo CAUTI (Fig. 4g and Supplementary Fig. 1d). Serum IgG titers were similar to those seen in earlier cohorts. However, bladder IgG levels were reduced (AUC ≤ 0.04), likely due to a lack of catheterization, which is known to induce significant inflammation and IgG influx²⁵. The presence of antigen-specific bone marrow plasma cells, memory B cells, and high levels of serum IgG indicate that Abp2D_RBD vaccination generates all of the hallmarks of immunity of an effective vaccine.

Fig. 4. Vaccination with Abp2D generates antigen-specific memory B cells and bone marrow plasma cells.

a Seven to nine week old C57BL/6 mice received 3 adjuvanted doses of 50 ug Abp2D_RBD or buffer (mock) and were sacrificed 4 weeks after the third dose. b Splenic memory B cells (Live/CD4^- CD19⁺/IgD^lo/GL7^- CD38⁺/IgM^- IgG1⁺) were stained with Abp2D_RBD-biotin and detected with SA-APC-Fire750 via flow cytometry. c, d Quantification of Abp2D_RBD+ splenic memory B cells as (c) % of IgM-/IgG1+ memory B cells and (d) total Abp2D_RBD+ memory B cells per spleen. Mock n = 10, Abp2D n = 15. Bars indicate mean +/- SD. Two-tailed unpaired t-test with Welch’s correction. e Bone marrow was assayed for antigen-specific bone marrow plasma cells via ELISpot. Representative wells from mock-immunized animals and Abp2D_RBD-immunized animals are shown. f Quantification of Abp2D_RBD-specific bone marrow plasma cells in mock (n = 7) and Abp2D_RBD-immunized animals (n = 15). Dotted line indicates the limit of detection (3 cells per 1e6 bone marrow cells). Bars indicate mean +/- SD. Two-tailed unpaired t-test with Welch’s correction. g Bladder homogenates and serum from week 4, week 8, and the day of sacrifice were assayed for Abp1D_RBD- (purple) and Abp2D_RBD-specific IgG (teal) by ELISA. Heatmaps were generated by calculating area under the curve for each serum/tissue sample. Data are pooled from two independent experiments. Created in BioRender. Timm, M. (2025) https://BioRender.com/b7xpqbu. Source data are provided as a Source Data file.

Passive immunization with serum from Abp2D_RBD-immunized mice protects naïve mice from CAUTI

If the immunity conferred by the Abp2D_RBD vaccine is due to serum IgG, then transferring IgG from immunized animals to naïve animals should also provide protection from a CAUTI challenge. To test the degree to which immunity conferred by the Abp2D_RBD vaccine is antibody-mediated, we administered serum pooled from mock immunized or Abp2D_RBD-immunized animals (Fig. 5a). Low doses of Abp2D-immune serum showed a trend towards reduced bacterial titers which did not achieve statistical significance (Supplementary Fig. 5). Mice that received a high dose of serum from Abp2D_RBD immunized animals had a statistically significant reduction in bladder (1.8 log) and catheter (1.6 log) bacterial titers compared to mock immunized animals (Fig. 5b). The magnitude and incidence of kidney infection was also reduced but this difference was not statistically significant (Supplementary Fig. 1e). Serum levels of adhesin-specific IgG in infected mice reached an AUC of >0.08, equivalent to the levels detected in the serum pools used for passive immunizations (Fig. 5c). These data demonstrate that humoral rather than cellular immunity is the likely driver of protection in our vaccine model.

Fig. 5. Passive immunization with serum from immunized mice protects naïve mice from CAUTI.

a Seven to nine week old C56/Bl6 mice received two 400 ul doses of pooled serum at 24 h and 0 h prior to catheterization and infection with A. baumannii strain ACICU. Serum was pooled from mock or Abp2D_RBD-immunized animals. b Mice were sacrificed 24 h post-infection and bacterial titers enumerated from bladders and catheters. Dashed lines indicate limit of detection. Solid lines indicate the median +/− 95% CI. Data shown are pooled from two independent experiments, n = 14 (Mock), n = 13 (Abp2D_RBD). Two-tailed Mann-Whitney U-test. c Serum and bladder homogenates from infected mice were assayed for Abp1D_RBD- (purple) and Abp2D_RBD-specific IgG (teal) by ELISA. The serum pools used for immunizations were also tested and are shown on the left of each heatmap. Heatmaps were generated by calculating the area under the curve. Created in BioRender. Timm, M. (2025) https://BioRender.com/4dsvxgc. Source data are provided as a Source Data file.

Serum from immunized mice blocks binding of A. baumannii to fibrinogen-coated catheters in vitro

Serum IgG contributes to immunity through a variety of mechanisms. To determine whether direct inhibition of the Abp2D-fibrinogen interaction was responsible for the phenotype we observe in immunized mice, we coated catheters with fibrinogen in vitro and incubated them with A. baumannii in the presence and absence of protective serum (Fig. 6a and Supplementary Fig. 6). Catheters incubated with Abp2D_RBD-immune serum had a significant reduction in bacterial binding to the fibrinogen-coated catheter surface, with an A700/A800 ratio (representing the proportion of fibrinogen-coated surface on which A. baumannii was detected) of 0.45 compared to 0.79 for mock serum (Fig. 6b). These data suggest that the efficacy of the Abp2D_RBD vaccine may be partially mediated by the direct blocking of the Abp2D-fibrinogen interaction.

Fig. 6. Abp2D-immune serum reduces in vitro binding of A. baumannii to fibrinogen-coated catheters.

a Silicone catheter material was coated with fibrinogen and incubated with A. baumannii + PBS (left), serum from mock-immunized mice (middle), or serum from Abp2D-immunized mice (right). Fibrinogen is stained in green and A. baumannii is stained in red. Bottom panels show a merge of the green and red channels. Catheter material was fixed and incubated with primary and secondary antibodies as a negative control for catheter autofluorescence, and used to set the threshold for each channel which was then applied equally to all experimental samples. Representative images from a single biological replicate are shown; a total of 3 biological replicates each with 4–6 technical replicates per condition were performed (additional replicates are shown in Supplementary Fig. 6). b Ratio of the red (A700) to green (A800) mean fluorescence intensity, n = 15 (PBS and Abp2D), n = 14 (Mock). One catheter from each group with poor fibrinogen saturation was excluded from statistical analysis following testing with Grubbs’ method. Bars represent mean +/- SD. One-way ANOVA with Tukey’s multiple comparisons test. Created in BioRender. Timm, M. (2025) https://BioRender.com/7csyj69. Source data are provided as a Source Data file.

The ACICU abp2d allele is prevalent among sequences isolates but does not elicit cross-reactive immunity

Multiple abp2d alleles exist within the population of published Acinetobacter genomes. To assess the potential coverage of our vaccine, we interrogated 5201 Acinetobacter genomes from the European Nucleotide Archive, of which 4791 contained an abp2d gene. The ACICU allele, which we used to generate our Abp2D_RBD vaccine, is the most common variant found in 2573 genomes (54%) (Fig. 7a). Since source data are not universally available for ENA sequence data, we also queried a database of A. baumannii clinical isolates from urine, blood and other body sites (acinetobase.vib.be) which demonstrated a similar proportion of ACICU allele dominance (27 out of 43 abp2d-containing strains) (Supplementary Fig. 8a).

Fig. 7. The ACICU abp2d sequence dominates in published A. baumannii genomes.

a Percent of each abd2d allele found in 4791 A. baumannii genomes from the ENA database. b Percent identity relative to the ACICU sequence at each nucleotide (orange) and amino acid (blue) position for the full-length abp2d gene. c Serum pooled from mice immunized with Abp2D_ACICU was assayed for binding to whole bacterial cells from clinical and reference strains by ELISA. Strains HUC85-130 were isolated from human urine or catheters, while strains ACICU and UPAB1 are reference strains. Binding to E. coli strain UTI89 was considered background and is shown by the dotted line. Bars represent mean +/- SD from 3 biological replicates. Strains HSS22-01-KS1 and KS2 were isolated from urologic stents and contain genes for Abp1D but not Abp2D. d Mice immunized with Abp2D_ACICU were challenged with strain UPAB1 or strain HUC130-01. Data are pooled from 2 independent experiments, n = 15 per group. Dashed lines represent the limit of detection, and solid lines represent the median +/− 95% CI. Kruskal-Wallis test with Dunn’s multiple comparisons correction. Created in BioRender. Timm, M. (2025) https://BioRender.com/p5ut530. Source data are provided as a Source Data file.

To test whether vaccination with Abp2D_ACICU might provide protection against diverse strains, we chose representatives from our collection of clinical A. baumannii CAUTI isolates from urine and catheters³⁹, which had 68–87% amino acid sequence identity to Abp2D_ACICU. We first used pooled serum from Abp2D_RBD-immunized mice to interrogate binding to whole bacterial cells via ELISA (Fig. 7c). The highest degree of cross-reactivity (AUC > 0.04) was observed for strain HUC86-02 (87% identity to ACICU); however, this strain was unable to be tested in vivo due to poor infectivity in our mouse model. Thus, we selected strains UPAB1 and HUC130-01 (70.7% and 72.6% identity to ACICU, respectively, AUC > 0.02 for both strains) for in vivo testing. Mice immunized with Abp2D_ACICU were challenged with each strain and 24 h bladder, catheter, and kidney titers were enumerated (Fig. 7d). There was no significant difference in bacterial titers between mock immunized and Abp2D_RBD immunized mice. These data suggest that a multivalent vaccine will likely be required to provide cross-strain protection from A. baumannii CAUTI, as is true for vaccines against many other bacterial and viral pathogens.

Discussion

Catheter-associated urinary tract infections are the second most common cause of healthcare-associated infections. Although A. baumannii causes a small percentage of all CAUTIs, these infections are often multi-drug resistant and frequently life-threatening for affected patients, leading the CDC to label A. baumannii as a “pathogen of urgent concern.”¹⁴ Thus, there is a critical need to develop novel antibiotic-sparing strategies to combat this infection. Here we demonstrate that a vaccine targeting the adhesin that mediates the interaction between A. baumannii and its host ligand provides protection from CAUTI. Our Abp2D_RBD vaccine elicits many features of a successful immune response, including production of memory B cells, bone marrow plasma cells, and high levels of serum IgG. While several vaccine strategies have been attempted for A. baumannii with mixed results⁴⁰, to our knowledge this is the first report of an A. baumannii vaccine that is effective in preventing CAUTI in a mouse model.

A. baumannii Abp1 and Abp2 pili are both capable of binding to fibrinogen, and both play a role in CAUTI²⁶. Thus, we expected that immunity to both Abp1D and Abp2D would be required in order to fully “neutralize” the bacteria and prevent adhesion to the catheter. It was therefore surprising that immunity to Abp1D_RBD proved to be unnecessary for protection from CAUTI. Although mice immunized with Abp1D_RBD generated high serum levels of Abp1D_RBD-specific IgG, including instances of IgG capable of cross-reacting with Abp2D_RBD, there was no protective effect (Fig. 3). One possible explanation for this finding is that Abp1D plays a smaller role in CAUTI pathogenesis in A. baumannii strain ACICU than was previously appreciated and that fibrinogen binding in vivo is dominated by Abp2D. This would be consistent with existing data on the dominance of Abp2 over Abp1 pili in a different urinary A. baumannii isolate, UPAB1²⁶. It is also possible that differences in protein conformation may be playing a role. While the two adhesins share a great deal of structural similarly, the anterior loop of the binding pocket is held in an “open” conformation by a K34-E87 salt bridge in Abp2D, while this salt bridge is absent in Abp1D, allowing the protein to more readily adopt both “open” and “closed” conformations²⁶. Protein conformation can have significant consequences for vaccine efficacy by directing an antibody response towards more or less biologically relevant epitopes^41,42. Therefore, we hypothesize that our Abp1D vaccine may have been sub-optimal and that further exploration and refinement of the protein’s conformation may be able to elicit a different result. Future studies will also investigate the nature of the Abp1D-specific antibodies found in some Abp2D-immunized animals, which may help to elucidate the most immunologically relevant epitopes.

The ACICU allele which we used to generate our Abp2D_RBD vaccine is found in >50% of the published genomes we analyzed, including in clinically relevant strains. Although immunization with this allele did not provide cross-reactive immunity against strains with different abp2d sequences, our sequence analysis demonstrates that >85% coverage could be achieved by vaccinating against the top 6 alleles. This is consistent with many other diseases which require multi-valent immunization to achieve broad protection. Licensed vaccines for pathogens ranging from Streptococcus pneumoniae to human papillomavirus include several of the most common antigenic variants in one vaccine. Future studies will investigate whether a combination vaccine of multiple Abp2D_RBD alleles will prevent CAUTI caused by diverse A. baumannii strains. In addition, since polymicrobial CAUTI is common^39,43 and fibrinogen-binding adhesins have been identified in many of the top CAUTI pathogens, a combination vaccine incorporating adhesins such as E. coli FimH^21,44, E. faecalis EbpA^24,25, S. aureus ClfA/B²⁹ and A. baumannii Abp2D could provide powerful protection against some of the most persistent catheter colonizers.

Although we demonstrated that our Abp2D_RBD vaccine produces a robust antigen-specific IgG response and that this immunity is transferable through serum, one limitation of this study is that we have not definitively identified which specific properties of the antibody response are providing the protection from challenge. We have shown that Abp2D-immune serum reduces binding of A. baumannii to fibrinogen-coated catheters, so it is likely that this direct “neutralization” of bacterial binding is playing a role. However, antibodies can also promote infection clearance through other mechanisms such as opsonization and complement activation. Future studies will attempt to establish which properties of the Abp2D_RBD antibody response are most essential for protection and optimize Abp2D_RBD immunizations to maximize efficacy.

Vaccines have an important role to play in reducing the incidence of disease and decreasing opportunities for natural selection of antibiotic-resistance. However, it is difficult to predict which patients may develop an A. baumannii infection and therefore challenging to identify who would most benefit from vaccination. One potential patient cohort is chronically catheterized patients. CAUTI risk increases by 3-7% for each day of catheterization, leading to an almost 100% probability of CAUTI in patients who remain catheterized over the long term³. Once established, CAUTI can be highly recurrent in spite of repeated antibiotic administration³⁹. Thus, chronically catheterized patients may be good candidates for prophylactic vaccination against CAUTI pathogens such as A. baumannii. However, perhaps the greatest potential benefit of an A. baumannii vaccine lies in the developing world. The highest relative burden of deaths associated with antibiotic-resistant A. baumannii occurs in low and middle income countries¹³. Health centers in Somalia and Kuwait report that A. baumannii accounts for up to 25% of CAUTIs in their facilities^11,12. In this setting, the storage conditions required for a protein subunit vaccine (e.g., simple refrigeration) present an advantage over more modern vaccine modalities²⁷.

Our findings highlight how basic research into microbial pathogenesis, such as the identification of pili implicated in CAUTI, can be translated into effective, antibiotic-sparing therapeutics. An Abp2D_RBD vaccine has the potential to reduce A. baumannii CAUTI incidence in vulnerable patient populations and represents a promising strategy in the fight against these antimicrobial resistant infections.

Methods

Ethics statement

All animal experimentation was conducted according to the National Institutes of Health guidelines for the housing and care of laboratory animals. All experiments were performed in accordance with institutional regulations after review and approval by the Animal Studies Committee at Washington University School of Medicine in St Louis, Missouri.

General bacteriology

Bacterial stocks were maintained as glycerol stocks at -80 °C. Strains were streaked on LB-agar plates and incubated at 37 °C for 14-18 h, at which time colonies were selected and used to inoculate liquid low-salt LB media (10 g tryptone, 5 g NaCl, and 5 g yeast extract per L). All bacterial cultures used in this study were grown statically at 37 °C for 24 h followed by 1:1000 dilution and subculture for an additional 18-20 h. Bacteria were spun down at 3000 x g, washed 1x in PBS, resuspended at the specified OD₆₀₀, and kept on ice until use. A. baumannii strain ACICU, representative of global clone 2⁴⁵, was used for all experiments presented in Figs. 1–6, Supplementary Figs. 1–3 and 6. Additional A. baumannii isolates used in Fig. 7 are shown in Supplementary Table 1.

General murine model specifications and handling

Exclusively female mice were used in this study due to the anatomic difficulty of catheterizing male mice. Mice were housed in standard Washington University barrier animal housing with a 12/12 light dark cycle at 22 degrees Celcius and 30% humidity. All mice used in this study were of strain C57BL/6, wild type, purchased from Charles River Lab Grantee. Mice were between 7 and 9 weeks old at the initiation of experiments.

Protein purification and labeling

Protein was expressed and labeled as previously described²⁴. Briefly, cells were harvested in a large-scale fermenter format from C600 containing expression plasmids, grown to mid-logarithmic phase, and induced with 0.1 mM IPTG for 1 h. The culture was subsequently harvested, and the periplasm isolated generally as described previously³⁵. Receptor binding domain protein constructs and mutants were purified by cobalt affinity chromatography, eluted at ~150 mM imidazole with a gradient of 1xPBS to 1xPBS/300 mM imidazole. Protein-containing fractions were pooled and run on a Source 15S (Tm GE) cation-exchange column and eluted at 30 mM NaCl with a gradient of 20 mM MES pH 5.7 to 20 mM MES pH 5.7/200 nM NaCl. Purified protein was subsequently dialyzed or buffer exchanged into 20 mM MES pH 5.8 + 50 mM NaCl. Where required, protein was biotinylated using the EZ-Link NHS-PEG4 biotinylation reagent (Thermo Scientific) and diluted in H₂O to 100 mM. Protein was either dialyzed or buffer exchanged into 1× PBS. Biotinylation reagent was added at a 20 fold molar excess for 2 h at 4 °C under rocking. Biotinylated protein was subsequently dialyzed into PBS, removing the excess biotin reagent.

Murine immunizations

All immunizations were prepared by mixing 50 μg/mouse of Abp1D_RBD or Abp2D_RBD 1:1 by volume with Addavax, a squalene oil-in-water adjuvant (Invivogen) to a total volume of 50 μL/mouse. Mock immunizations were prepared by mixing buffer 1:1 with Addavax. C57BL/6 mice were obtained from Charles River Laboratories and were 7–9 weeks old at the first immunization. Mice were immunized intramuscularly in the hind flank at weeks 0, 4, and 8, for a total of 3 immunizations of 50 μg/protein each. For dual immunization experiments, each mouse received 50 μg of Abp1D_RBD in the left hind flank and 50 µg of Abp2D_RBD in the right hind flank at each time point. Blood was collected at weeks 4 and 8 by submandibular or submental collection prior to the administration of the immunization.

Murine CAUTI model

Exclusively female mice were used for all experiments due to the anatomic difficulty of catheterizing male mice. Mice were catheterized and infected as described previously⁹ with one modification: 2 doses of A. baumannii were administered because this model was found to more reliably infect 19–21 week old C57BL/6 mice (the age following 3 immunizations) than a single dose. Briefly, mice were anesthetized with 4% isoflurane/0.8% oxygen by inhalation. A short piece of silicone tubing (4–5 mm) (Braintree Scientific #ID0.012xOD0.025) was transurethrally inserted into the bladder and immediately followed by 2 doses of 2 × 10⁸ CFUs of A. baumannii strain ACICU in 50 μL of PBS (OD600 ~ 13) 3 h apart. 24 h after infection, mice were anaesthetized and humanely sacrificed by cervical dislocation. Blood was collected from the inferior vena cava, allowed to clot for 30 min at room temperature, and spun down to remove red blood cells and clotting factors. Serum was removed to a new tube and frozen at -20 °C until analysis. Catheters were removed from bladders, placed into 1 mL of sterile PBS, and processed by vortexing for 30 s, sonicating for 10 min, and vortexing for an additional 30 s to remove biofilm and bacteria from the catheter surface. Bladders and kidney pairs were both placed into tubes containing sterile stainless steel beads and sterile PBS (1 mL for bladders, 800 μL for kidney pairs) and homogenized at 4 °C using the MP Biomedical Fastprep-24 homogenizer. The homogenization settings used were 1 min shaking at 4 m/s, 5 min of rest, followed by an additional 1 min of shaking. Bladder, kidney, and catheter samples were serially diluted and plated on selective media (LB + 100 μg/L Ampicillin). Plates were incubated at 37 °C for 12–16 h and bacterial cfus enumerated. Remaining bladder and kidney homogenates were frozen at -20 °C for additional analyses. Samples from mice which spontaneously expelled their catheters prior to sacrifice were excluded from analyses.

Convalescent infection model

Seven to nine week old C57BL/6 mice were catheterized and infected as described above. One group of mice received 2 doses of 2 × 10⁸ cfus of ACICU, and the other group received sterile PBS. Urines were collected at days 3, 7, 10, 14, and weekly thereafter. Blood was collected by submandibular or submental collection at weeks 3, 5, and 8, and at time of sacrifice. At week 5, mice were treated with 1 g/L Apramycin for 10 days to clear bacteriuria. At week 8, mice were again catheterized and infected, then sacrificed at 24 h post infection as described above.

ELISAs

All ELISAs were performed using Grenier Microlon high-binding plates (Grenier Bio-One #655085). Plates were coated with 100 µLof 1 µg/mL Abp1D_RBD, Abp2D_RBD, or E. faecalis EbpA^NTD (used as a negative control for anti-HIS antibodies) in PBS and incubated overnight at 4 °C. The following morning plates were washed 1x with 200 µLPBS containing 0.05% Tween-20 (PBS-T). Plates were blocked with 300 µLof PBS-T containing 10% fetal bovine serum (P10) for 1.5 h at room temperature. Serum, bladder, and kidney homogenates were diluted 1:30 into 75 µL P10 and then serially diluted 1:3 and incubated for 1 h at room temperature. Plates were washed 3x in PBS-T. Goat anti-mouse-IgG-HRP secondary antibody (Southern Biotech Cat# 1030-05) was diluted 1:1000 in P10 and 100 µL was added to each well and incubated for 1 h at room temperature in the dark. Plates were washed 3x with PBS-T followed by 3x with PBS, developed with 100 µL developing reagent and quenched with 100 µL of 1 M HCl. Developing reagent consists of 10 mL phosphate-citrate buffer (Sigma Cat# P4809), 4 mg o-Phenylenediamine dihydrochloride (Sigma Cat#P8787), and 33 µL 3% H₂O₂ per plate. Plates were read using the BioTek ELx800 plate reader on the OD490 setting. Graphpad Prism 10 was used to calculate area under the curve for each sample. AUCs were baseline corrected by subtracting the AUC binding to the negative control protein, EbpA^NTD24, which contains the same 6x His tag used to purify Abp1D_RBD and Abp2D_RBD but is otherwise structurally unique. IgA ELISAs were performed exactly as described above except that the secondary antibody was Goat anti-mouse IgA HRP (Southern Biotech Cat#1040-05).

ELISpot

PVDF-membrane plates (Millipore Sigma #MSIPN4W50) were prepared by activating with 50 µL of 35% ethanol for 30 s followed by washing 3x with PBS. Plates were coated with 100 µL of 5 µg/mL Abp2D_RBD or anti-mouse IgG (positive control) in PBS and incubated overnight at 4 °C. The next morning, plates were washed 3x with PBD + 0.05% Tween-20 (PBS-T) and blocked with 200 µL of RPMI media containing 10% fetal bovine serum (R10) for 2 h at 37 °C and 5% CO₂. Mice that were immunized as described above were sacrificed 4 weeks after the third immunization. Bone marrow was collected from both femurs into R10, washed, and resuspended to a concentration of 1 × 10⁷ cells/mL. 5 × 10⁵ cells were added to the first well and serially diluted. Plates were incubated for 4 h at 37 °C and 5% CO₂. Plates were washed 1x with PBS and 3x with PBS-T. 100 µL of biotinylated anti-mouse IgG (Southern Biotech Cat#1030-08) diluted 1:1000 in PBS containing 2% fetal bovine serum and 2 mM EDTA was added to the plate and incubated overnight at 4 °C. The next day, plates were washed 3x with PBS-T. HRP-conjugated streptavidin (Jackson Immunoresearch Cat# 016-030-084) was diluted 1:5000 in PBS + 2%FBS/2 mM EDTA, 100 µL added to each well, and the plates incubated for 1.5 h at room temperature in the dark. Plates were washed 3x with PBS-T followed by 1x with PBS. Developing solution was prepared by diluting 3 mg of 3-through a 0.45 PVDF membrane. Just prior to use, 100 µL of 3% H₂O₂ was added to the mixture. 100 µL of developing solution was added to each well and allowed to incubate until spots were visible, ~5 min. Developing solution was removed and plates washed under DI water to halt the reaction. Plates were dried overnight at room temperature and imaged using the CTL ImmunoSpot imager (Cellular Technology Limited). Spots were counted using the CTL ImmunoSpot automatic counting program version 7.0.38.2 with default parameters.

Flow cytometry

Mice that were immunized as described above were sacrificed 4 weeks after the third immunization. Spleens were collected into RPMI containing 2% fetal bovine serum and manually homogenized using the back of a syringe plunger. Cells were filtered through 75 um mesh, washed 1x, and counted. 2 × 10⁷ splenocytes were stained for flow cytometry. All washes for the staining process were performed in PBS containing 2% fetal bovine serum and 2 mM EDTA. Cells were incubated with CD16/32 (Biolegend Cat# 101302) and 5.875 µg/mL of biotinylated Abp2D_RBD for 10 min, then washed 3x. A cocktail containing the following antibodies in the specified dilutions was prepared in BD Brilliant Staining Buffer (BD Cat. # 563794), all sourced from BioLegend unless otherwise indicated: Zombie NIR 1:400 (Cat#423105), CD19-BV750 1:200 (Cat#115561), CD4-BV570 1:100 (Cat#100542), IgD-BV711 1:200 (Cat#405731), IgM-BV605 1:50 (Cat#406523), IgG1-BV510 1:100 (Cat#406621), Fas-PE 1:200 (BD Cat# 554258), GL7-PcpCy5.5 1:50 (Cat# 144610), CD38-PE-Cy7 1:200 (Cat#102718), CD138-BV421 1:200 (Cat#142508), and streptavidin-APC-Fire-750 1:200 (Cat#405250). Invitrogen UltraComp eBeads were used for single colors. Flow cytometry data was collected using the Cytek Aurora with 4 laser 16V-14B-10YG-8R configuration and processed on FlowJo10 for Mac.

Passive immunization model

Serum was collected from mice that were either mock immunized or immunized with Abp2D_RBD as described above and frozen until pooling. An equal volume of serum from each individual within a group was combined to form the serum pools. Serum pools were sterile filtered and frozen in aliquots at -20 °C until use. Two pools were prepared: (i) Mock immunized, and (ii) Abp2D_RBD immunized. Naïve, 7–9 week old C56Bl/6 mice received 1 dose of 400 µL pooled serum 24 h prior to catheterization and infection as described above. Mice received a second dose of 400 µL pooled serum immediately prior to infection. Mice were sacrificed at 24 hpi and tissue titers enumerated as described above.

in vitro catheter imaging

Catheter staining procedures were modified from those previously described²⁸. Silicone catheter material (Braintree Scientific #ID0.012xOD0.025) was cut to length (4-5 mm) and incubated in 1 mL of 10 mg/mL human fibrinogen that was free from plasminogen and von Willebrand factor (Enzyme Research Laboratories #FIB2) in sterile PBS. Catheters were incubated in fibrinogen solution for 24 h at 37 °C, rotating. Catheters were fixed in 1 mL 10% neutral buffered formalin (Epredia #5701) for 30 min at room temperature, then washed 3x in sterile PBS and stored in sterile PBS until use. Bacteria were grown as described above and resuspended to an OD of 0.5 in sterile PBS. Serum from Abp2D- or Mock-immunized mice was added to the bacterial suspension at a 1:20 ratio. Fibrinogen coated catheters were added and incubated for 1 hr at 37 °C, rotating. Catheters were fixed in 1 mL 10% neutral buffered formalin (Epredia #5701) for 30 min at room temperature, then washed 3x in sterile PBS. Catheters were blocked overnight at 4 °C in blocking buffer (1.5% bovine serum albumin, 0.1% sodium azide), shaking. The next day, goat anti-Fibrinogen antibody (Sigma-Aldrich Cat# F8512) and rabbit anti-Acinetobacter antibody⁹ were diluted in blocking buffer (1:500 and 1:1000, respectively), added to the catheters and incubated for 2 hrs at RT, rotating. Catheters were washed 3x in PBS containing 0.05% Tween (PBS-T). Secondary antibodies (LI-COR Donkey Anti-Goat IRDye 800CW Ref#926–32214 and LI-COR Donkey Anti-Rabbit IRDye 680LT Ref#926–68023) were diluted 1:10,000 in blocking buffer, wrapped in foil to protect from light and incubated 1 hr at RT, rotating. Catheters were washed 3x in PBS-T and allowed to dry overnight at 4 °C. Catheters were imaged using the Odyssey Imaging System (LI-COR Biosciences) using the following settings: preset - protein gel; focus offset – 0.5 mm; resolution – 42 μm; quality – high; intensity – 5.0. Images were analyzed using Odyssey Infrared Imaging software (version 3.0.16) to measure relative infrared fluorescence at 800 nm and 700 nm, corresponding to fibrinogen and Acinetobacter, respectively. Background fluorescence was assessed using catheters incubated in PBS and fixed/washed identically to experimental samples. The threshold for each channel was adjusted such that background fluorescence of the PBS-control catheter was minimized, and this threshold was applied equally to all experimental samples within a replicate.

Fluorescence quantification

Fiji Version 2.16.0/1.54 g was used to manually define a region of interest around each imaged catheter and quantify the mean fluorescence intensity at 800 nm and 700 nm as has been previously described⁴⁶. Since the degree of fibrinogen coating varied between catheters, fluorescence was normalized as follows: A700/A800 ratio = (mean fluorescence at 700 nm)/(mean fluorescence at 800 nm). Grubbs method with alpha = 0.05 was used to remove outliers prior to statistical testing, which removed one catheter from each condition (corresponding to the catheters with very minimal fibrinogen deposition). Excluded catheters are indicated with an asterisk in Supplementary Fig. 6. A one-way ANOVA with Tukey’s multiple comparisons test was used to test for statistical significance.

Clinical A. baumannii isolates

5 of the 7 clinical A. baumannii strains used in this study have been previously published³⁹. Strains were isolated from the urine and/or catheters of patients undergoing standard-of-care catheter removal or from patients undergoing urologic surgery for the removal of kidney stones or stents. Informed consent was obtained from all patients. This study was approved by the Washington University School of Medicine (WUSM) Internal Review Board (approval #201410058) and performed in accordance with WUSM’s ethical standards and the 1964 Helsinki declaration and its later amendments.

Whole bacterial cell ELISAs

Bacterial strains (Supplementary Table 1) were grown as described above, washed 1x in sterile PBS, and resuspended to an OD₆₀₀ of 1.0. 100 µL of the bacterial suspension was added to an ELISA plate (Grenier Bio-One #655085) and incubated 4 hrs at 4 °C (temp). Plates were centrifuged at 2,800 rpm for 5 min, the supernatant decanted, and fixed with 100 µL of 10% neutral buffered formalin (Epredia #5701) for 20 min at room temperature. Plates were washed 3x with PBS + 0.05% Tween-20 and used immediately for ELISAs per the protocol described above. Graphpad Prism 10 was used to calculate area under the curve for each sample. AUCs were baseline corrected by subtracting the AUC binding by the mock serum pool from the Abp2D_RBD serum pool.

abp2d sequencing

Genomic DNA was isolated from the HUC strains listed in Supplementary Table 1 using the Wizard Genomic DNA purification kit (Promega #A1125) and sequenced using the Plasmidsaurus hybrid ONT+ Illumina bacterial genome sequencing service. abp1d and abp2d sequences were identified using Proksee⁴⁷ and aligned in Geneious version 2024.0.3 (Biomatters). Sequences were deposited in GenBank under the accession numbers listed in Supplementary Table 1.

abp2d sequence analysis

To interrogate for the presence of ACICU abp2D sequence amongst Acinetobacter baumannii strains, a publicly available and searchable k-mer database constructed from 661 K bacterial genomes in the European Nucleotide Archive (Umbrella project PRJEB46036) was queried using the COBS search index with a threshold value of 0.80^48,49. Genomes were previously assembled and refined with Kraken2⁵⁰ and Bracken⁵¹, to determine taxonomic species. A total of 5201 genomes were identified as A. baumannii. The ACICU abp2D sequence was used as a BLAST query for the custom database from the selected A. baumannii assemblies⁵². Homologous gene sequences were then filtered to remove nucleotide sequences which did not cover >80% of the ACICU abp2D sequence. This BLAST-based search strategy differed from the search strategy used in Tamadonfar et al., which was more stringent and identified only sequences which were nearly identical to the ACICU abp2d allele²⁶. A total of 4791 out of 5201 A. baumannii genomes were found to contain an abp2d allele. Since source data are not universally available for ENA strains, we also interrogated a database of Belgian clinical isolates available at acinetobase.vib.be and processed these genomes similar to above. Nucleotide and amino acid sequence alignments were aligned using the MAFFT algorithm⁵³, BLOSUM62 scoring matrix, and PAL2NAL⁵⁴ as previously described²⁵. Final alignments for data visualization were performed in Geneious v2024.0.3 using the built-in Clustal Omega 1.2.3 alignment tool and Geneious neighbor-joining tree builder. Percent identity for each base pair was calculated in Geneious. For nucleotide alignments, the curve was smoothed in GraphPad Prism v10.4.1 with a 5-neighbor 2^nd-order smoothing function. No smoothing was performed for amino acid alignments.

Statistical analysis

All statistical tests were performed using built-in statistical functions of GraphPad Prism 10. Analysis of bacterial titer data, which are non-parametric, was performed using the Mann-Whitney U-test for comparisons of 2 groups or the Kruskal-Wallis test with Dunn’s multiple comparisons test for comparisons of 3 or more groups. Analyses of immunologic data in Fig. 4 were performed using an unpaired t-test with Welch’s correction for unequal standard deviation. Analysis of mean fluorescence intensity in Fig. 6 and Supplementary Fig. 6 was performed using a one-way ANOVA with Tukey’s multiple comparisons test. Categorical kidney infection incidence data presented in Supplementary Fig. 1 were analyzed using Fisher’s exact test. Correlations in Supplementary Fig. 3 were performed using Spearman’s rank correlation.

Reporting summary

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

Author contributions

M.R.T., K.O.T., T.M.N., K.W.D., A.H.E., and S.J.H. designed experiments; K.O.T., K.W.D., and J.S.P. directed all protein cloning and purification efforts; M.R.T. carried out experiments; M.R.T., K.O.T., T.M.N., J.B.V., K.W.D., A.H.E., and S.J.H. analyzed the data; and M.R.T., T.M.N., and S.J.H. wrote the paper with input from all authors.

Peer review

Peer review information

Nature Communications thanks Timothy Keys, Shawna McCallin, and John Boyce for their contribution to the peer review of this work. A peer review file is available.

Data availability

All data supporting the findings of this study are available within the paper, the Supplementary Information, and the provided Source Data. Novel bacterial strains described in this study have not been publicly deposited due to IRB limitations but are available upon request. abp2d sequences have been deposited in GenBank under accession numbers PV425425, PV425426, PV425427, PV425428, and PV425429 as described in Supplementary Table 1. Source data are provided with this paper.

Competing interests

The authors have additional information to disclose. M.R.T., K.O.T., K.W.D., J.S.P., A.H.E., and S.J.H. are listed on provisional patent applications regarding therapeutics targeting Abp1D and Abp2D. S.J.H. consults for Fimbrion Therapeutics, QureTech Bio, and Sequoia Sciences. AHE has received consulting and speaking fees from InBios International, Fimbrion Therapeutics, RGAX, Mubadala Investment Company, Moderna, Pfizer, GSK, Danaher, Third Rock Ventures, Goldman Sachs and Morgan Stanley and is the founder of ImmuneBio Consulting. The remaining authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-62402-9.