Biofilm formation in Acinetobacter baumannii isolates from companion animals and horses: associations with host, clinical origin, antimicrobial resistance, and clonal lineage

1 Introduction

Acinetobacter baumannii is a Gram-negative bacterium of the family Moraxellaceae and is recognized as an important cause of nosocomial and community-associated infections, including urinary and respiratory tract infections, bacteremia, sepsis, and wound infections (Antunes et al., 2014; Batt and Tortorello, 2014; Almasaudi, 2018). Beyond its role as an opportunistic pathogen, A. baumannii is particularly notable for its pronounced ability to form biofilms, a trait that substantially contributes to its persistence in clinical environments and to its success as a healthcare-associated pathogen (Tomaras et al., 2003; Pour et al., 2011; Longo et al., 2014; Eze et al., 2018). Biofilm formation (BF) enhances bacterial survival under adverse environmental conditions, promotes long-term colonization, and facilitates persistence on both biotic and abiotic surfaces, including medical devices such as urinary catheters and endotracheal tubes (Tomaras et al., 2003; Longo et al., 2014; Eze et al., 2018). Through BF, A. baumannii can establish protected microbial communities that exhibit increased tolerance to host immune responses and antimicrobial treatment, thereby exacerbating urinary and respiratory tract infections in both humans and animals (Pour et al., 2011).

The clinical relevance of BF is further amplified by the remarkable antimicrobial resistance (AMR) profile of A. baumannii. The species possesses multiple intrinsic and acquired resistance mechanisms and frequently exhibits a multidrug-resistant (MDR) phenotype. These mechanisms include the production of drug-modifying enzymes, overexpression of efflux pumps, alterations in membrane permeability, and modifications of antibiotic target sites (Wareth et al., 2019). Although studies directly addressing the relationship between BF and AMR remain limited, several virulence-associated genes—in this context commonly referred to as biofilm-associated genes (BAGs)—have been linked to BF and may contribute to both bacterial persistence and resistance development (Choi et al., 2009; Eze et al., 2018; Pakharukova et al., 2018; Sarshar et al., 2021; Kim et al., 2022; Roy et al., 2022). The combination of strong biofilm-forming capacity and multidrug resistance has placed A. baumannii among the ESKAPE pathogens, a group of clinically important bacteria frequently associated with treatment failure due to antibiotic resistance (Rice, 2008).

Treatment of A. baumannii infections is particularly challenging in the context of biofilm-associated infections, where antimicrobial penetration and efficacy may be reduced. Antimicrobial therapy is generally guided by antimicrobial susceptibility testing (AST) results, both in human and veterinary medicine (van der Kolk et al., 2019). Therapeutic management is complicated by widespread resistance to penicillins, cephalosporins, and tetracyclines (Bundesamt für Verbraucherschutz und Lebensmittelsicherheit, 2023; Schwabe and Paffrath, 2023), and by the increasing prevalence of MDR isolates, which significantly limits available treatment options (Lee et al., 2017; Nocera et al., 2021). In human medicine, mild to moderate infections may be treated with minocycline monotherapy (Lee et al., 2017), whereas severe infections typically require combination therapy, often including carbapenems (e.g., imipenem or meropenem) or ampicillin-sulbactam combined with aminoglycosides (Isler et al., 2019). In cases of MDR infections where alternative antibiotics are ineffective, colistin or tigecycline are recommended (Giguère et al., 2013; Wong et al., 2017; European Medicines Agency, 2019). In response to the global rise of MDR A. baumannii, the World Health Organization (WHO) designated this pathogen as a critical-priority organism for research and development of new therapeutic strategies in its 2017 Priority Pathogens List (World Health Organization, 2017).

In veterinary medicine, A. baumannii is increasingly recognized as an opportunistic pathogen associated with a wide range of clinical infections in animals, including wound, urinary tract, respiratory, postoperative, and device-associated infections. In companion animals and horses, infections are reported most frequently from wounds and the urinary tract (Vaneechoutte et al., 2000; Endimiani et al., 2011; Nocera et al., 2021). Comparative genomic studies have further demonstrated a close relationship between isolates from humans and animals (Wareth et al., 2019; Yehya et al., 2025).

To date, twelve international clones (ICs), designated IC1 to IC12, have been described (Karah et al., 2023; Shelenkov et al., 2023; an Xu et al., 2024; Karah et al., 2025). Among these, IC1, IC2, and IC7 are of particular epidemiological importance, as they represent predominant lineages associated with hospital- and community-acquired infections worldwide (Higgins et al., 2010; Lupo et al., 2017; Jacobmeyer et al., 2022). Several studies have demonstrated that A. baumannii isolates from cats, dogs, and horses frequently belong to the same worldwide clones - particularly IC1, IC2, and IC7 - as human clinical isolates, suggesting possible transmission routes or shared reservoirs between human and animal populations (Higgins et al., 2010; Endimiani et al., 2011; Zordan et al., 2011; Nocera et al., 2021; Castillo-Ramírez et al., 2025).

Given the critical role of BF in bacterial persistence, virulence, and potential therapeutic failure, this study aimed to investigate the association between the biofilm-forming capacity of A. baumannii isolates with their host species, clinical origin, AMR, and assignment to international clones. A better understanding of these associations may contribute to the development of more targeted therapeutic strategies and improved infection management in both human and veterinary medicine.

2 Materials and methods

2.1 Sampling and identification of bacterial strains

A total of 507 A. baumannii isolates from cats (n = 134), dogs (n = 289), and horses (n = 84) were randomly collected from clinical samples submitted to our veterinary microbiological diagnostic laboratory between 2008 and 2024. Most samples originated from Germany (n = 465), while the remaining samples were obtained from France (n = 22), Italy (n = 18), the Netherlands (n = 1), and Belgium (n = 1).

The isolates were obtained from different clinical sources: wounds and abscesses (n = 173), urinary tract (n = 102), medical devices, including catheters, implants, and tubes (n = 40), nasopharynx (n = 39), respiratory tract (n = 38), genital tract (n = 24), skin, hair, and paw/hoof (n = 24), and various other clinical sites, including ears, eyes, feces, gastrointestinal tract, and internal organs (n = 67).

Species identification was performed using matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) with the Microflex LT/SH mass spectrometer and the MALDI Biotyper database version 3.3.1.0 (both Bruker Daltonics, Bremen, Germany). Species identification was further confirmed by ribosomal multilocus sequence typing (rMLST) using the online tool “Identify species” provided by PubMLST (https://pubmlst.org/species-id; accessed 29. October 2025). All isolates were stored at -70 °C in Brain Heart Infusion broth (Oxoid, Wesel, Germany) supplemented with 30% glycerol.

2.2 Antimicrobial susceptibility testing

Antimicrobial susceptibility testing (AST) was performed using the VITEK^® 2 system (AST card GN-97; bioMérieux, Nürtingen, Germany). All isolates were tested using the same instrumentation and laboratory workflow throughout the study period (2008–2024), ensuring methodological consistency. Susceptibility results were interpreted retrospectively using the current clinical breakpoints as specified below to allow direct comparison across the entire dataset. The following antimicrobial agents were tested: ampicillin (AMP), amoxicillin–clavulanic acid (AMC), cephalexin (CEX), cefpodoxime (CPD), cefovecin (VEC), ceftiofur (CTU), imipenem (IPM), amikacin (AMK), gentamicin (GEN), enrofloxacin (ENR), marbofloxacin (MAR), tetracycline (TET), nitrofurantoin (NIT), chloramphenicol (CHL), and trimethoprim/sulfamethoxazole (SXT).

As veterinary-specific clinical breakpoints for Acinetobacter spp. are not available, human clinical breakpoints from the CLSI supplement M100 (Clinical and Laboratory Standards Institute, 2025) were applied, except for AMC, CPD, NIT, and CHL (Enterobacterales breakpoints., CLSI M100), ENR and MAR (ciprofloxacin breakpoints for Acinetobacter spp., CLSI M100), VEC and CTU (Enterobacterales breakpoints., CLSI VET01S (Clinical and Laboratory Standards Institute, 2024)), AMP and CEX (Enterobacterales breakpoints, EUCAST (European Committee on Antimicrobial Susceptibility Testing, 2025)). Intrinsic resistance was interpreted according to CLSI guideline definitions for Acinetobacter spp (Clinical and Laboratory Standards Institute, 2025). Escherichia coli ATCC 25922 and Pseudomonas aeruginosa ATCC 27853 were used as quality control reference strains for AST performed on the VITEK^® 2 compact automated system, as recommended by the manufacturer.

2.3 Biofilm assay

Biofilm production was quantified using a crystal violet assay with minor modifications based on O’Toole et al. (2011) (O’Toole, 2011). Overnight cultures of A. baumannii isolates were harvested from blood agar plates (blood agar base by Merck Chemicals, GmbH, Darmstadt, Germany, supplemented with 5% sheep blood) and incubated in 3 ml Mueller-Hinton Broth (MHB) at 37 °C with shaking at 130 rpm for 20 h. The bacterial suspension was adjusted to an OD₆₀₀ of 0.05 with MHB, and 200 µl of each dilution was transferred into 96-well microtiter plates (U-Profile, Rotilabo^® by Carl Roth, Karlsruhe, Germany). Plates were sealed with a Breathe-Easy^® sealing membrane (Diversified Biotech) and incubated at 28 °C for 24 h and 48 h. The incubation temperature was selected based on preliminary experiments (data not shown), which demonstrated more robust and reproducible BF at 28 °C than at 37 °C. Similar incubation conditions have been used in previous studies investigating BF in A. baumannii (Malaka de Silva et al., 2018; Eze et al., 2018). After incubation, bacterial growth was measured photometrically using the Multiskan FC plate reader (Thermo Fisher Scientific) at OD₅₇₀ to account for planktonic biomass. Supernatants were removed and each well was washed three times using 300 µl distilled water. Adherent bacterial cells were fixed by adding 250 µl of 99% methanol for 15 min. After removal of methanol, biofilms were stained with 0.1% crystal violet solution (Merck, Darmstadt, Germany) for 30 min. Following three washing steps with 300 µl of distilled water, the stain was dissolved in 300 µl of an ethanol-acetone solution (80:20) on a horizontal shaker at 300 rpm for 45 min. The amount of eluted crystal violet was used as a surrogate for the total biofilm mass and was measured photometrically using the Multiskan FC plate reader (Thermo Fisher Scientific) at OD₅₉₅. To account for differences in bacterial growth between isolates, the OD₅₉₅ values were normalized to the corresponding OD₅₇₀ measurements, yielding specific biofilm formation (SBF = OD₅₉₅/OD₅₇₀). Each isolate was analyzed in three technical replicates, and the entire assay was repeated independently three times. SBF values represent the mean of the three biological replicates. Positive (A. baumannii ATCC 19606) and negative (A. baumannii ATCC 17978) control strains were included in each independent experiment to monitor assay performance and inter-assay reproducibility (Qi et al., 2016; Malaka de Silva et al., 2018). The SBF threshold (SBF_T) was defined as three standard deviations above the mean SBF of the negative control (ATCC 17978). The results were classified as follows (Stepanović et al., 2000): SBF ≤ SBF_T, non-biofilm producers; SBF_T < SBF ≤ (2 × SBF_T), weak biofilm producers; (2 × SBF_T) < SBF ≤ (4 × SBF_T), moderate biofilm producers; (4 × SBF_T) < SBF, strong biofilm producers.

2.4 Whole-genome sequencing

Genomic DNA for whole-genome sequencing (WGS) was extracted using the Master Pure™ DNA Purification Kit (Biozym Scientific GmbH, Hessisch Oldendorf, Germany). Sequencing libraries were prepared using the Nextera XT Library Preparation Kit (Illumina GmbH, Munich, Germany) for a 250-bp paired-end sequencing on an Illumina MiSeq platform (Illumina Inc., San Diego, CA, USA) with a minimum coverage of 100-fold. FASTQ files were quality-trimmed prior to de novo assembly and annotation using SPAdes v.3.15.1 (Bankevich et al., 2012) and RAST v.2.0 (http://rast.nmpdr.org/; accessed 29 October 2025) (Klotz et al., 2018).

2.5 Multilocus sequence typing and assignment of international clones

Whole-genome sequences were used to identify multilocus sequence types (MLST) according to the Pasteur scheme using Ridom SeqSphere+ v.10.0.5 (Ridom GmbH, Münster, Germany). International clones (ICs) IC1, IC2, and IC3 were initially identified by multiplex PCR as described by Turton et al (Turton et al., 2007). MLST assignments were used as additional evidence for IC classification where sequence types with established IC associations were available. To enable the classification of isolates beyond IC1 to IC3 and to confirm IC assignments, a whole-genome-based comparison with representative published genomes of IC1 to IC12 was performed using a previously described approach categorize (Unger et al., 2022). Briefly, reference genomes representing the established ICs were included in the analysis, and isolates were assigned to IC lineages based on their clustering with the corresponding reference genomes in the phylogenetic tree.

2.6 Identification of biofilm-associated genes

Biofilm-associated genes (BAGs) were identified using a combined in silico workflow involving Ridom SeqSphere+ v.10.0.5 (Ridom GmbH, Münster, Germany), the”Multiple Genome Analysis” function of BacWGSTdb 2.0, and Geneious R8 v.8.1.9 (Biomatters Ltd., Auckland, New Zealand) (Ruan and Feng, 2016; Ruan et al., 2020; Feng et al., 2021). A combined approach was applied to account for the considerable sequence variability observed in some BAGs and to increase the robustness of gene detection across genetically diverse isolates.

The analyzed BAGs included csuA/B/C/D/E (encoding the chaperon-usher pilus system Csu), ompA (outer membrane protein A), bap (biofilm-associated protein), blp1 and blp2 (Bap-like proteins), AbaIR (quorum-sensing system consisting of abaI [acyl-homoserine lactone synthase AbaI] and abaR [DNA-binding HTH domain-containing protein AbaR]), the pgaABCD gene cluster encoding PNAG (polysaccharide poly-β-(1,6)-N-acetylglucosamine), and bfmR/bfmS (two-component regulatory system) (Choi et al., 2009; Longo et al., 2014; Pakharukova et al., 2018; Sarshar et al., 2021; Kim et al., 2022; Roy et al., 2022). A threshold of 90% for both nucleotide identity and length coverage was used for the prediction of AMR genes and BAGs.

Initial BAG screening was performed using tools implemented in SeqSphere+ and BacWGSTdb 2.0. SeqSphere+ was used for automated gene detection using “A. baumannii VFDB” whereas BacWGSTdb 2.0 provided homology-based detection using its curated virulence-associated database. For manual validation, a predefined set of biofilm-associated genes (including bap, blp1, blp2, csuABCDE, pgaABCD, ompA, abaI/abaR, and bfmRS) and corresponding reference genomes were compiled from publicly available A. baumannii genomes retrieved from the NCBI database. Geneious was used to confirm gene presence or absence by sequence mapping and comparison with the respective reference sequences. Results obtained with SeqSphere+ and BacWGSTdb were cross-validated, and discordant findings were resolved by BLASTn analysis against the corresponding reference sequences. A gene was considered present when both nucleotide identity and sequence coverage were ≥90%.

2.7 Statistical analysis

Statistical analyses were performed using IBM SPSS Statistics version 31 (IBM Corporation, Armonk, NY, USA) and R version 4.4.1 (R Foundation for Statistical Computing, Vienna, Austria). All statistical tests were two-sided, and p < 0.05 was considered statistically significant. Group comparisons were conducted using Chi-Square-tests and Fisher`s exact tests for categorical variables, and t-tests and analysis of variance (ANOVA) for continuous variables. Bonferroni correction was applied for multiple pairwise comparisons where appropriate. Bowker’s test of symmetry was applied for paired analyses of temporal changes between 24 h and 48 h measurements. Effect sizes were interpreted according to Cohen’s conventions. To account for potential confounding between host species, clinical origin, clonal lineage, antimicrobial resistance, and biofilm-associated genes, a multivariable Firth logistic regression model was applied with strong BF as the binary outcome. Results are presented as adjusted odds ratios with 95% confidence intervals.

3 Results

3.1 Characteristics of the bacterial isolates and phylogenetic analysis

Most A. baumannii isolates were recovered from wounds and abscesses (n = 173), with a significantly higher prevalence in dogs (n = 119) compared to cats (n = 33) and horses (n = 21) (p < 0.0001). A similar pattern was observed for isolates obtained from medical devices (n = 40), of which most originated from dogs (n = 35), while only four isolates were obtained from cats and one from a horse (p < 0.0001). The second most common clinical origin was the urinary tract (n = 102), with similar numbers in dogs (n = 57) and cats (n = 44), but only one isolate from a horse. Eight isolates were found in each of the following clinical sites in horses: the nasopharynx (n = 39), the respiratory tract (n = 38), and the genital tract (n = 24). In dogs, 21 isolates originated from the nasopharynx, 19 from the respiratory tract, and 12 from the genital tract. The situation was similar in cats: Here, 10 isolates were obtained from the nasopharynx, 11 from the respiratory tract, and 4 from the genital tract. In contrast, isolates from hair, paw/hoof (n = 24) were found significantly more in horses (n = 20) than in cats (n = 0) or dogs (n = 4) (p < 0.0001). However, this observation should be interpreted with caution, as these patterns may, at least in part, reflect sampling bias due to unequal sample sizes and underlying sampling distribution among cats (n = 134), dogs (n = 289), and horses (n = 84).

All 507 isolates were screened for assignment to international clones IC1-IC12. IC1 (n = 181) and IC2 (n = 118) were the most prevalent lineages, followed by IC3 (n = 39) and IC7 (n = 21), whereas IC5 (n = 2) and IC8 (n = 8) were detected only rarely. No isolates were assigned to IC4, IC9, IC10, IC11, or IC12. A total of 138 isolates could not be assigned to any of the investigated ICs and are hereafter referred to as non-IC1-IC12 (Table 1). Detailed information on isolate origin, host species, and isolate distribution is provided in Supplementary Table 1.

Table 1. International clone (IC) assignment of 507 A. baumannii isolates tested in this study according to host species.

No isolates were assigned to IC4, IC6, or IC9–IC12; therefore, these ICs are not shown as separate columns.

The distribution of ICs differed significantly between host species (p < 0.0001). Horses showed a markedly higher proportion of non-IC1-IC12 isolates, whereas isolates from dogs were dominated by IC1 and IC2. Cats displayed an intermediate pattern, with a comparatively higher proportion of IC7. Pairwise comparisons revealed no significant difference between cats and dogs (p = 0.15), whereas both species differed significantly from horses (p < 0.0001).

A significant association between clinical source and IC distribution was observed (p < 0.0001). Isolates from wounds and abscesses were predominantly assigned to IC1, whereas urinary tract isolates showed an increased proportion of IC7 and included the only IC5 isolates detected in this study (Table 2). However, because only two IC5 isolates were identified, this observation should be interpreted with caution. Medical device-associated isolates were enriched for IC2 and IC3, whereas isolates from the nasopharynx and genital tract were mainly assigned to non-defined IC lineages. Pairwise comparisons confirmed significant differences particularly between wound- or medical device-associated isolates and isolates originating from the genital tract or nasopharynx (p < 0.0001).

Table 2. Distribution of international clones (ICs) among clinical sources of 507 A. baumannii isolates.

“Others” includes the clinical origins ear, eye, feces, gastrointestinal tract, and internal organs. Dashes (–) indicate that no isolates were assigned. No isolates were assigned to IC4, IC6, or IC9–IC12; therefore, these ICs are not shown as separate columns.

3.2 Antimicrobial susceptibility

Antimicrobial susceptibility profiles were determined for 507 A. baumannii isolates against 16 antimicrobial agents representing seven antibiotic classes (Figure 1). Most isolates exhibited high levels of resistance. All isolates were either intrinsically resistant or showed acquired resistance to AMP, AMC, CEX, VEC, CTU, NIT, and CHL; only one isolate showed intermediate susceptibility to CTU. Resistance to CPD was detected in 500 isolates (98.6%). AMK was the most effective antimicrobial agent, with 97.8% (n = 496) of isolates classified as susceptible. Resistance to IPM was observed in 31.0% (n = 157) of isolates. In addition, resistance was detected in 66.7% (n = 338) of isolates for GEN, 68.6% (n = 348) for TET, 79.3% (n = 402) for ENR, and 79.5% (n = 403) for MAR. Minimum inhibitory concentration (MIC) data of the individual isolates are provided in Supplementary Table 1.

Within host species, differences in resistance were observed for 6 of the 17 antimicrobial agents. While 71.9% of canine isolates and 70.2% of feline isolates were resistant to GEN, only 42.9% of horse isolates showed resistance to this agent. Similar patterns were observed for ENR, MAR, TET, and SXT (Table 3). A significant difference was observed for IPM: resistance was detected in 31.1% of canine and 49.3% of feline isolates, but in only 1.2% of equine isolates.

Table 3. Antimicrobial resistance phenotypes of 507 A. baumannii isolates stratified by host species, shown as absolute numbers and percentages within each species.

*Intrinsic resistance according CLSI guideline definitions for Acinetobacter spp (Clinical and Laboratory Standards Institute, 2025). Dashes (–) indicate that no isolates were assigned.

According to the definition proposed by Magiorakos et al. (2012), excluding intrinsic resistance, isolates were classified as either non-MDR or MDR, with MDR defined as resistance to agents from three or more antimicrobial classes (Magiorakos et al., 2012). Based on this definition, 20.5% (n = 106) of isolates were classified as non-MDR and 79.5% (n = 401) as MDR (p < 0.0001).

Overall, MDR prevalence differed significantly among host species (p < 0.0001), with 42.9% in equine isolates 89.6% in feline isolates, and 85.5% in canine isolates. No significant difference was observed between cats and dogs (p = 0.34), whereas MDR rates in horses were significantly lower than those in cats and dogs (p < 0.0001 each).

3.3 Occurrence of biofilm-associated genes

All 507 isolates were screened for the presence of biofilm-associated genes (BAGs), with detailed distributions provided in Supplementary Table 1. Several BAGs were highly prevalent across all host species, including bfmR, bfmS, and multiple components of the pgaABCD and csu loci, suggesting the presence of a conserved core gene set associated with BF. However, not all BAGs were universally detected, and host-associated differences were observed for selected genes (Table 4).

Table 4. Occurrence (%) of biofilm-associated genes (BAGs) in 507 A. baumannii isolates, stratified by host species.

The regulatory genes bfmR and bfmS were identified in 100% of isolates. Likewise, abaI, abaR, csuA/B, csuC, csuD, csuE, ompA, and pgaABCD were detected in more than 90% of isolates across all host species. In contrast, csuA and csuB were present in only 68.4% and 67.1% of isolates, respectively; however, these differences were not statistically significant between host species (p = 0.18).

The bap gene showed variable presence, being detected in 26.9% of feline, 39.1% of canine, and 57.1% of equine isolates. The blp1 gene was present in 81.3% of feline and 85.8% of canine isolates, but significantly less frequently in equine isolates (48.8%; p = 0.033 and p = 0.006, respectively). In contrast, the blp2 gene was detected in all isolates. Overall, feline and canine isolates showed largely comparable BAG profiles, whereas equine isolates were distinguished primarily by a lower prevalence of blp1.

BAG distribution was further assessed in relation to ICs. Most BAGs showed a conserved distribution across ICs; however, selected genes showed marked lineage-specific variation (Table 5). The bap gene was predominantly associated with IC2 (114/118, 96.6%), IC5 (2/2, 100%), and non-IC1-IC12 isolates (81/138, 58.7%). The blp1 gene was detected in all IC1, IC2, IC3, and IC5 isolates, but was absent in IC7 and IC8 isolates and occurred at lower frequency in non-IC1-IC12 isolates (42.0%, p < 0.0001).

Table 5. Distribution (%) of biofilm-associated genes (BAGs) among international clones (ICs) in 507 A. baumannii isolates.

Overall, while most BAGs were widely distributed among the analyzed isolates, specific genes – particularly bap and blp1 – demonstrated pronounced lineage-associated distribution patterns.

3.4 Specific biofilm formation

Using the crystal violet assay, specific biofilm formation (SBF) was determined for all 507 A. baumannii isolates. The SBF threshold (SBF_T) was calculated separately for each incubation time, resulting in values of 0.7141 at 24 h and 0.6491 at 48 h. No significant difference in SBF_T was observed between the two incubation times (p = 0.24). Across all isolates, SBF showed a slight but significant increase from 24 h to 48 h (p < 0.0001). After stratification by host species, significant differences between the two incubation times were also observed for feline, canine, and equine isolates (all p < 0.0001). Canine isolates showed the most pronounced increase in SBF, whereas equine isolates exhibited the smallest increase (Figure 2).

Among feline isolates, the overall BF classification remained relatively stable over time. At 24 h, the most common category was moderate BF (n = 50, 37.3%), followed by strong BF (n = 30, 22.4%). Between 24 h and 48 h, the number of strong biofilm producers increased slightly (from n = 30 to n = 39), whereas the number of moderate producers decreased (from n = 50 to n = 36). Paired analysis revealed no statistically significant change in BF category distribution over time (p = 0.12) (Figure 3). Canine isolates were predominantly classified as moderate biofilm producers at 24 h (n = 128, 44.3%), with a substantial increase in strong biofilm producers at 48 h (from n = 46 to n = 85). Paired analysis showed a significant shift toward stronger BF, driven mainly by transitions from moderate to strong (p < 0.0001). Equine isolates showed the most pronounced temporal shift in BF classification. At 24 h, only one of 84 isolates was classified as a strong biofilm producer, whereas moderate and weak producers dominated. By 48 h, the number of strong biofilm producers increased markedly (from n = 1 to n =20), and paired analysis confirmed a highly significant shift in BF classification (p < 0.0001). The SBF values for individual isolates, together with their corresponding classifications, are provided in Supplementary Table 1.

3.5 Biofilm formation in relation to clinical origin

Biofilm formation was also analyzed in relation to clinical origin. After 24 h of incubation, a significant association between clinical origin and BF classification was observed (p < 0.001), and this association remained significant after 48 h (p < 0.001) (Figure 4). Isolates derived from wounds and abscesses exhibited higher proportions of moderate and strong biofilm producers than isolates from most other clinical origins. Urinary tract and medical device-associated isolates likewise showed high proportions of moderate and strong biofilm producers. In contrast, isolates from the genital tract were predominantly classified as weak or non-biofilm producers. From 24 h to 48 h, a shift toward stronger biofilm production was observed across several clinical sources, particularly among isolates from wounds/abscesses and the urinary tract. Medical device-associated isolates also exhibited high proportions of moderate and strong biofilm producers at both time points, whereas genital tract isolates remained predominantly weak or non-biofilm producers.

Paired analysis of temporal changes between 24 h and 48 h revealed a highly significant overall shift in BF categories (p < 0.001), indicating a non-symmetric transition pattern. The predominant transition was from moderate to strong BF, whereas reverse transitions were comparatively rare. This effect was particularly pronounced among isolates from wounds/abscesses, the urinary tract, and medical device-associated samples (Table 6).

Table 6. Distribution (%) of 507 A. baumannii isolates across four categories (non, weak, moderate, and strong biofilm producers) after 24 h and 48 h of incubation, stratified by clinical origin.

Wound, abscess: n = 173; urinary tract: n = 102; medical devices: n = 40; nasopharynx: n = 39; respiratory tract: n = 38; genital tract: n = 24; skin, hair, paw/hoof: n = 24; others: n = 67 (including ear, eye, feces, gastrointestinal tract, and internal organs). Dashes (–) indicate that no isolates were assigned.

Stratified analyses by clinical origin revealed significant temporal shifts for isolates from wounds/abscesses and the urinary tract (both p < 0.001). After Bonferroni correction for multiple testing (adjusted α = 0.00625), both groups remained statistically significant. Isolates from medical device-associated samples showed a borderline effect (p = 0.006), which did not remain significant after Bonferroni correction. No significant temporal changes were observed for isolates from the genital tract, nasopharynx, or skin/hair/paw/hoof (all p ≥ 0.61).

3.6 Association between biofilm formation and antimicrobial resistance

To investigate the association between BF and phenotypic AMR, acquired resistance to CPD, VEC, CTU, IPM, AMK, GEN, ENR, MAR, TET, and SXT was analyzed. Intrinsic resistance was excluded, and isolates with intermediate susceptibility were classified as resistant.

All isolates showed resistance to at least three of the antimicrobial agents included in this analysis (Table 7). BF was significantly associated with the number of AMR at both 24 hours and 48 hours (p < 0.0001). Overall, BF increased with the number of AMR, despite minor deviations from a consistent gradient. Low-resistance isolates (3–5 agents) were mainly non- or weak biofilm producers, whereas highly resistant isolates (8–10 agents) were strongly enriched among moderate and particularly strong biofilm producers.

Table 7. Distribution (%) of acquired phenotypic resistance among 507 A. baumannii isolates according to biofilm formation category (non, weak, moderate, and strong) at 24 h and 48 h.

CPD, cefpodoxime; VEC, cefovecin; CTU, ceftiofur; IPM, imipenem; AMK, amikacin; GEN, gentamicin; ENR, enrofloxacin; MAR, marbofloxacin; TET, tetracycline; SXT, trimethoprim-sulfamethoxazole. Resistance to 0 to 2 antibiotics is not shown, as all isolates were resistant to at least 3 antibiotics.

A more detailed analysis of BF in relation to MDR status confirmed this pattern, with isolates stratified into non-MDR and MDR groups across BF categories and incubation times (Figure 5). At 24 h, MDR isolates were predominantly classified as moderate or strong biofilm producers, whereas non-MDR isolates were mainly weak or moderate producers and only rarely classified as strong (p < 0.001). After 48 h, the proportion of strong biofilm producers increased in both groups but was markedly higher among MDR isolates, with 33.4% (n = 134) classified as strong producers, compared with 7.5% (n = 8) of non-MDR isolates (p < 0.001). Paired analysis of temporal changes from 24 h to 48 h revealed only minor shifts among non-MDR isolates, primarily from moderate to strong BF, which did not remain significant after Bonferroni correction (adjusted α = 0.00625, p = 0.112). In contrast, MDR isolates showed a pronounced and highly significant shift from moderate to strong BF (p < 0.001).

Given the long sampling period, the distribution of biofilm phenotypes and MDR isolates was evaluated by year of isolation. Although year-to-year fluctuations were observed, no clear temporal trend in BF or MDR occurrence was evident across the study period (2008–2024), indicating that the main findings were not attributable to a particular time interval.

3.7 Biofilm formation across international clones

A direct comparison of ICs based on SBF values showed that IC1 isolates exhibited significantly higher BF than all other ICs (p < 0.001). In addition, IC2 isolates demonstrated a significant increase in BF from 24 h to 48 h (p = 0.015), which was also observed for non-IC1-IC12 isolates (p < 0.001). Greater variability in SBF values was observed among IC1, IC2, and non-IC1-IC12 isolates (Figure 6).

Classification into BF categories supported these findings. IC1 isolates showed the highest biofilm-forming capacity, with a substantial proportion of strong biofilm producers already present after 24 h (n = 77, 42.5%), increasing further after 48 h (n = 108, 59.7%) (Table 8). The number of isolates analyzed remained constant; the increase in the proportion of strong biofilm producers from 24 h to 48 h reflects temporal biofilm maturation rather than differences in sample size. In contrast, IC2 isolates showed no strong BF at 24 h but developed a considerable proportion of strong producers after 48 h (n = 31, 26.3%). IC3 isolates predominantly exhibited moderate BF at both time points (61.5% at 24 h and 74.4% at 48 h) and did not include any strong biofilm producers. Similarly, IC7 and IC8 isolates were restricted to non-, weak, or moderate BF categories. Non-IC1-IC12 isolates showed a heterogeneous distribution, with strong BF observed only after 48 h and at low frequency (n = 5, 3.6%).

Table 8. Distribution of international clones (ICs) among 507 A. baumannii isolates by biofilm formation category (non, weak, moderate, strong) after 24 h and 48 h. Dashes (–) indicate that no isolates were assigned.

3.8 Relationship between biofilm-associated genes and biofilm formation

As 14 of the 18 investigated BAGs were present in at least 90% of all isolates, subsequent univariable analyses focused on genes with lower prevalence, namely bap (38.9%), blp1 (78.5%), csuA (68.4%), and csuB (67.1%). The presence of these genes was analyzed in relation to BF categories across all 507 A. baumannii isolates (Figure 7).

The presence of blp1 was strongly associated with BF. It was detected in 100% of strong biofilm producers at 24 h, and in 98.6% at 48 h. In contrast, blp1 was present in 78.8% and 59.3% of non-biofilm producers at 24 h and 48 h, respectively, whereas among weak biofilm producers blp1 was identified in 59.3% at 24 h and 67.4% at 48 h. Conversely, bap exhibited an inverse association with BF. It was frequently detected in non-biofilm producers (63.6% at 24 h and 60.2% at 48 h), whereas its prevalence was markedly lower in strong biofilm producers (0% at 24 h and 25.0% at 48 h).

Fisher’s exact test (strong vs. non-biofilm producers) identified significant associations for bap at both time points (both p < 0.0001) and for blp1 at 48 h (p = 0.04), but not at 24 h (p = 0.32).

For csuA and csuB, no consistent associations with BF were observed. Although csuA showed a weak association at 24 h (p = 0.041), this was not confirmed after 48 h (p = 0.74). Likewise, csuB did not differ significantly across BF categories at either time point (24 h: p = 0.48; 48 h: p = 0.46).

Because univariable analyses do not account for potential confounding, the investigated variables were subsequently evaluated in a multivariable logistic regression model.

3.9 Multivariable analysis of biofilm formation

To investigate independent predictors of BF, a Firth-penalized logistic regression model was performed including host species, clinical origin, IC, MDR phenotype, and BAGs. The overall model was statistically significant (likelihood ratio test: p < 0.001), indicating that the included variables collectively contributed to the prediction of strong BF.

Host species remained significantly associated with BF. Compared with feline isolates (reference category), equine isolates showed significantly higher odds of strong BF (OR = 7.80, 95% CI: 3.00–∞, p < 0.001), whereas no significant difference was observed between canine and feline isolates (p = 0.65).

Among the ICs, IC1 exhibited a strong relationship with BF (p < 0.001), consistent with its high biofilm-forming capacity observed in the univariable analyses. IC2 showed a weaker, borderline association (p = 0.067).

The inverse association between the bap gene and BF remained significant in the multivariable model (p = 0.003), with bap-positive isolates showing lower odds of strong BF than bap-negative isolates. In contrast, blp1 did not retain significance in the multivariable model.

Clinical origin and MDR phenotype were not significantly related to BF after adjustment for the other covariates, suggesting that the associations observed in the univariable analyses were at least partly attributable to host species and clonal background.

4 Discussion

Acinetobacter baumannii is widely recognized as an opportunistic pathogen of major clinical relevance in both human and veterinary medicine, largely due to its ability to acquire multidrug resistance (MDR) and to persist under adverse conditions, particularly through biofilm formation (BF). In this study, we systematically assessed BF in A. baumannii isolates from companion animals (cats, dogs, and horses) and examined its relationship with host species, clinical origin, international clones (ICs), antimicrobial resistance (AMR), and biofilm-associated genes (BAGs).

A first key finding is that BF is highly prevalent but heterogeneous, with variation across host species and infection sites. While canine isolates showed a temporal increase in BF and equine isolates a delayed but pronounced shift toward strong BF, feline isolates remained comparatively stable. Likewise, isolates derived from wounds or abscesses, urinary tract infections, and device-associated samples were more frequently moderate to strong biofilm producers than those from the genital tract. Together, these observations indicate that BF is not uniformly expressed across A. baumannii isolates and varies according to ecological context, particularly host environment and infection niche. However, because host species, clinical origin, and clonal lineage are closely intertwined in this dataset, the observed associations cannot be interpreted as evidence of causal relationships and may largely reflect underlying population structure and confounding factors. This interpretation is consistent with previous reports demonstrating strong BF in canine and feline A. baumannii isolates and linking it to persistence in chronic infections and on abiotic surfaces, where prolonged survival provides a selective advantage (van der Kolk et al., 2019; Lysitsas et al., 2024; Ababneh et al., 2026). In both veterinary and human medicine, BF enhances bacterial adherence, persistence, and tolerance to environmental and antimicrobial stressors, especially in infections of the urinary and respiratory tract (Tomaras et al., 2003; Pour et al., 2011; Rajangam and Narasimhan, 2024). However, our findings should be interpreted cautiously, as only A. baumannii-positive samples were included and underlying differences in sampling populations across host species were not controlled for.

A second important finding is the strong IC-dependent pattern of BF. IC1 and IC2 were the predominant lineages and exhibited the highest biofilm-forming capacity, with IC1 showing consistently strong BF and IC2 displaying pronounced maturation over time. In contrast, other ICs and non-IC1-IC12 isolates showed weaker or delayed BF. This supports the concept that specific clonal lineages are predisposed to enhanced persistence traits. Previous studies have similarly demonstrated that globally dominant ICs - particularly IC1 and IC2 – combine MDR with virulence-associated traits, including BF, which likely contributes to their epidemiological success (Diancourt et al., 2010; Nocera et al., 2021; Shelenkov et al., 2023; Ayswarya et al., 2026). Our findings are further supported by reports such as Kishii et al. (2020), who observed increased biofilm maturation in IC2-, IC6-, and IC7-associated strains (Kishii et al., 2020). Nevertheless, the limited number of isolates in less common ICs (e.g., IC5 and IC8) in our study restricts the robustness of lineage-specific conclusions.

Closely linked to clonal background, our data also support an association between BF and MDR, with MDR isolates tending to exhibit higher biofilm-forming capacity. This aligns with numerous studies reporting a positive relationship between antimicrobial resistance and BF (Badave and Kulkarni, 2015; Bardbari et al., 2017; Ranjbar and Farahani, 2019; Yang et al., 2019; Asaad et al., 2021; Sherif et al., 2021; Jacobmeyer et al., 2022). However, in the multivariable model, MDR status was not independently associated with strong BF after adjustment for host species and clonal lineage. Therefore, although functional relationships between antimicrobial resistance and BF may exist, the patterns observed in the present dataset are likely shaped to a considerable extent by lineage-associated effects. Bardbari et al. (2017) investigated BF in clinical and environmental A. baumannii isolates and identified a significant association for both groups between MDR and BF (Bardbari et al., 2017). Also Asaad et al. (2021) demonstrated that MDR and XDR phenotypes were significantly associated with biofilm-formation capacity (Asaad et al., 2021). However, few studies have also shown an inverse relationship between BF and MDR phenotype of clinical A. baumannii isolates (Qi et al., 2016; Wang et al., 2018; Shenkutie et al., 2020). Qi et al. (2016) identified 197 of 272 clinical A. baumannii isolates as MDR and reported that 8.7% of non-MDR and 91.3% MDR isolates were negatively correlated with BF (Qi et al., 2016). Such discrepancies likely reflect differences in study design, strain composition, and experimental conditions. Mechanistically, the co-occurrence of MDR and BF may be explained by shared regulatory pathways, stress responses, or co-selection processes that favor both persistence and resistance traits (Longo et al., 2014; Harding et al., 2018). Nevertheless, the observational nature of the present study precludes conclusions regarding causality, and the observed patterns may be influenced by residual confounding between MDR phenotype, clonal lineage, host species, and clinical origin. From a clinical perspective, this combination is particularly concerning, as it may further reduce treatment efficacy and promote chronic or recurrent infections.

At the genetic level, BF in A. baumannii is governed by complex regulatory networks involving quorum sensing (AbaI/AbaR), two-component systems (BfmRS), and structural components such as pili (Csu system) and outer membrane or matrix-associated proteins (e.g., OmpA, Bap) (Eze et al., 2018; Harding et al., 2018; Kim et al., 2022; Rajangam and Narasimhan, 2024). These components contribute to distinct stages of biofilm development, ranging from initial attachment to biofilm maturation and stability. Despite the well-established functional roles of these gene products, most BAGs – including abaI, abaR, bfmR, bfmS, csuA/B, csuC, csuD, csuE, ompA, and pgaABCD - were highly conserved across isolates in the present study. This indicates that gene presence alone is insufficient to explain the observed phenotypic variability in BF. This observation is consistent with previous reports showing widespread distribution of key BAGs, such as ompA, bap, and csuE, among both susceptible and MDR isolates, without a consistent correlation to BF strength (Thummeepak et al., 2016; Yang et al., 2019; Sherif et al., 2021). Together with evidence that BF is highly strain-dependent and influenced by environmental conditions (Eze et al., 2018), these findings underscore the importance of regulatory mechanisms, gene expression dynamics, and lineage-specific effects in shaping biofilm phenotypes.

In this context, blp1 emerged as a notable exception. Although blp1 showed a strong association with BF in univariable analyses, this relationship was not retained in the multivariable model. The apparent association therefore seems to be driven largely by the lineage-associated distribution of the gene, which was predominantly confined to IC1, IC2, and IC3. Consequently, the present data do not support the interpretation of blp1 as an independent driver of the enhanced biofilm-forming capacity or apparent success of these lineages. Nevertheless, this should not be interpreted as evidence against a potential role of blp1 in BF. Indeed, experimental studies have demonstrated reduced biofilm production following disruption of blp1 or blp2 in A. baumannii strain AYE (Gregorio et al., 2015). Taken together, these findings suggest that the association observed in the present dataset primarily reflects clonal background, whereas the experimentally demonstrated contribution of blp1 to BF may be masked by lineage-associated effects.

Skerniškytė et al. (2019) also showed that blp1 was present in IC1 and IC2 isolates; however, the blp1 variant identified in IC2 differed from the originally described gene (Skerniškytė et al., 2019). This suggests that allelic variation within BAGs – rather than mere gene presence or absence – may play an important role in modulating BF and could occur in a lineage-specific manner.

Notably, while the presence and variation of blp1 have been described for IC1 and IC2, data on IC3 remain scarce. The detection of blp1 in IC3 isolates provides novel insight into its distribution in this less well-characterized lineage. Nevertheless, further studies are required to determine whether other ICs besides IC3 harbor blp1 variants or whether variants of other BAGs exist.

In contrast, bap showed an inverse association with BF in our dataset, which contrasts with its commonly described role as a key biofilm-promoting factor. In previous studies, bap expression was positively correlated with strong BF and has been implicated in both BF and antimicrobial resistance phenotypes (Loehfelm et al., 2008; Yang et al., 2019; Kakavan et al., 2025). This discrepancy likely reflects differences in gene regulation, allelic variation, or environmental influences rather than gene presence alone. Similar inconsistencies have been reported previously, including the presence of bap in non-biofilm-forming isolates and its absence in biofilm-positive strains (Ghasemi et al., 2018; Gautam et al., 2023; Smitran et al., 2023). These findings underscore the complexity of BF regulation and reinforce the notion that genotype-phenotype relationships in A. baumannii are highly context-dependent.

Other BAGs, including csuA and csuB, did not show consistent associations with BF, which may be explained by the primary role of the Csu pili system in early-stage attachment rather than mature BF (Tomaras et al., 2003; Eze et al., 2018). Overall, these results emphasize that BF should be considered a dynamic and multifactorial phenotype driven by regulatory networks and environmental conditions rather than the mere presence of individual genes.

From a clinical perspective, the predominance of strong biofilm producers among isolates from wounds and abscesses, urinary tract infections, and medical devices highlights the relevance of BF in veterinary infections. Biofilm-associated growth likely promotes persistence, reduces antimicrobial susceptibility, and increases the likelihood of chronic and recurrent infections (Costerton et al., 1999; Hall-Stoodley et al., 2004). This is particularly relevant for A. baumannii, a pathogen characterized by high MDR rates and the ability to survive under antimicrobial pressure, partly through biofilm-associated tolerance mechanisms such as persister cell formation (Tacconelli et al., 2018; Shenkutie et al., 2020; Mohamed et al., 2023). The co-occurrence of MDR and strong BF, especially in canine and feline isolates, further underscores the therapeutic challenges in veterinary medicine, where treatment options are often limited and animals may serve as sources of resistant strains. These findings support the potential value of anti-biofilm strategies as adjuncts to antimicrobial therapy.

Finally, several limitations should be considered. Although the overall number of isolates was high, subgroup sizes for certain host species and ICs were limited, reducing statistical power. In addition, all isolates originated from a single diagnostic laboratory, which may limit generalizability. Future multicenter studies integrating genomic and phenotypic data are needed to validate these findings and to further elucidate the interplay between clonal lineage, resistance, and BF.

5 Conclusions

Biofilm formation in A. baumannii from companion animals represents a common and clinically relevant phenotype associated with clonal lineage, infection site, and antimicrobial resistance. The strong biofilm-forming capacity of dominant lineages, particularly IC1 and IC2, together with their frequent multidrug resistance, is consistent with the hypothesis that lineage-associated traits contribute to persistence and epidemiological success.

The lack of consistent associations between most biofilm-associated genes and biofilm formation indicates that gene presence alone is insufficient to explain phenotypic variability, emphasizing the role of regulatory mechanisms and genetic background. While bap remained independently associated with biofilm formation, the direction of this association differed from previous reports, underscoring the complexity of genotype-phenotype relationships underlying biofilm formation. In contrast, the association observed for blp1 appeared to be largely explained by its lineage-associated distribution, highlighting the importance of clonal background when interpreting relationships between biofilm-associated genes and biofilm formation.

Clinically, the convergence of biofilm formation and multidrug resistance underscores the challenges associated with the treatment of A. baumannii infections in companion animals and supports the consideration of anti-biofilm strategies as adjuncts to antimicrobial therapy. Future studies should focus on regulatory pathways and gene expression dynamics, ideally within multicenter frameworks integrating genomic and phenotypic data. Such approaches will be essential to disentangle the relative contributions of clonal background, regulatory mechanisms, and environmental influences to biofilm formation and persistence in A. baumannii.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fbrio.2026.1874938/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: George Cosmin Nadăş, University of Agricultural Sciences and Veterinary Medicine of Cluj-Napoca, Romania

Reviewed by: Jayaseelan Murugaiyan, SRM University, India

Adriana Belas, Lusofona University, Portugal

Footnotes

AMR, antimicrobial resistance; AST, antimicrobial susceptibility testing; BAG, Biofilm-associated gene; BF, Biofilm formation; CGE, Center for Genomic Epidemiology; IC, International Clone; MALDI-TOF MS, matrix-assisted laser desorption/ionization- time of flight mass spectrometry; MDR, Multidrug-resistance; MLST, multilocus sequence typing; NCBI, National Center for Biotechnology Information; SBF, Specific Biofilm Formation; ST, sequence type; VAG, virulence-associated gene; WGS, whole genome sequencing.