This study included nine in-patients enrolled by the Department of Orthopedics at Qingdao Huangdao District Traditional Chinese Medicine Hospital in a rural area of Qingdao, China, between January 2021 and August 2024, meeting the case criteria as previously described (Wang et al., 2024a, 2024b). All patients were local residents with no recent travel history outside the region. Exudate specimens were collected upon clinical examination.

This study was conducted in accordance with the ethical principles of the 1964 Declaration of Helsinki and later amendments. Ethical approval was granted by the Ethics Committees of both Qingdao University and Qingdao Huangdao District Traditional Chinese Medicine Hospital on January 1, 2021. According to administrative policy at the time, no registration number was assigned.

Bacterial isolates were retrieved by streaking exudate swabs on blood agar plates (Babio, Jinan, China) and incubated at 37°C. Bacterial species identification and antimicrobial susceptibility tests were performed using the VITEK 2 COMPACT system (bioMérieux, France) according to the manufacturer’s constructions. The results were interpreted following the Clinical and Laboratory Standards Institute (CLSI) guidelines (M100, Ed31-35, Malvern, Pennsylvania, USA). The antibiotics tested included imipenem, meropenem, ampicillin/sulbactam, piperacillin/tazobactam, ceftazidime, cefotaxime, ceftriaxone, cefoperazone/sulbactam, cefepime, gentamicin, tobramycin, amikacin, ciprofloxacin, levofloxacin, trimethoprim/sulfamethoxazole, minocycline, polymyxin B, and tigecycline. A. baumannii ATCC 19606 was used as the quality control.

Whole-genome sequencing (WGS) was performed on one ABC isolate from each patient using the Illumina NovaSeq 6000 platform (Sangon Biotech, Shanghai, China), generating paired-end reads of 150 bp. Raw reads were quality-filtered and trimmed using Trimmomatic v0.36 to remove adapters and low-quality bases. Filtered reads were de novo assembled using SPAdes v3.15. Gene predictions and annotations were conducted using Prokka v1.14 with default bacterial settings. Phylogenetic analysis was performed for the nine clinical isolates along with publicly available genomes of A. baumannii, A. pittii, and A. soli downloaded from the National Center for Biotechnology Information (NCBI). These genomes were filtered using pyANI v0.2.10 to retain those with ≥ 95% genome-wide average nucleotide identity (ANI) to the clinical isolates under study. All protein-coding genes were clustered into orthologous groups (OGs) based on sequence similarity using MMseqs2 v13 with default parameters. Multiple sequence alignments were first generated at the protein level using MAFFT v7.475, and codon-based alignments were then obtained by back-translation using PAL2NAL v14. Poorly aligned or ambiguously aligned regions were removed with Gblocks v0.91. The resulting alignments were concatenated into a single dataset using in-house Python scripts. A maximum-likelihood phylogenetic tree was constructed using FastTree v.2.1 and visualized with the Interactive Tree of Life (iTOL).

Multilocus sequence typing (MLST) was performed via the pubMLST.org online platform using the A. baumannii scheme. Antibiotic resistance genes (ARGs) were identified using the Comprehensive Antibiotic Resistance Database (CARD) with the criteria “perfect and strict hits only”. Virulence factors of Pathogenic Bacteria database (VFDB) was used to screen for the presence of virulence genes (sequence identity ≥ 70%, query coverage ≥ 70%, E-value < 1e-5). Mobile genetic elements (MGEs) were identified with the following tools under default parameters: COPLA for plasmid classification, CRISPRCasFinder for CRISPR-Cas system detection, ISfinder for insertion sequences, MobileElementFinder for transposable elements (identity ≥ 90%, coverage ≥ 95%,E-value < 1e-5), and PHASTER for prophage prediction.

The nine cases included in this study had no overlapping hospital stays or evident epidemiological connection. Only Patient 1 shared a brief overlap with patients 2 and 7 during hospitalization, but their Acinetobacter isolates were phylogenetically distant or belonged to different species ( Table 1 , Figure 1 ).

Strains isolated from patients 1 to 6 were all initially identified as ABC by culture, and further clarified as A. baumannii by WGS ( Figure 1A ). Patient 1 was admitted to the ICU in critical condition following a severe traffic accident and underwent invasive skeletal traction upon arrival. She remained in the ICU for the first eight days and underwent five additional surgeries during a 90-day hospitalization. Yellowish, fish-smelling discharge was first observed at the fixation pinhole on day 9. Eight subsequent bacterial cultures from day 19 to 44 were all positive for ABC. The isolate included in this study, HD12, was obtained on day 26 and identified as carbapenem-resistant A. baumannii (CRAB) with a high level of multidrug resistance ( Figure 2 ). Notably, resistance to cefoperazone/sulbactam emerged by day 42, with the minimum inhibitory concentration increasing from 16 to ≥64 µg/mL after 20 days of intravenous treatment. Phylogenetic analysis showed that HD12 clustered with 17 other sequences from human sources in China, derived from various sample types including abdominal fluid (1), blood (5), bronchoalveolar lavage fluid (4), cerebrospinal fluid (1), sputum (3), kidney preservation fluid (2), and one unspecified, while HD12 was the only one obtained from a wound ( Figure 1B ).

Patient 2 was admitted with osteomyelitis with purulent yellow discharge lasting for about one month. Initial cultures were negative, but sinus tract sampling on day 4 yielded Staphylococcus aureus, and an incision exudate on day 10 was positive for ABC by culture (isolate HD8). Phylogenetic analysis placed HD8 within a subclade of 10 genomes from six countries across four continents, originating from humans (3), ICU furniture (2), and birds (5). Additionally, none of the twenty sequences in the adjacent subclades were from China, while 55% (11/20) were human-associated ( Supplementary Figure S1A ).

A. baumannii isolates HD136 and HD167, obtained from patients 3 and 4 about one year apart, were phylogenetically related ( Supplementary Figure S1B ). Patient 3 was admitted with a fractured left femur after falling from a height. Skeletal traction was performed on the day of admission in the patient ward, and internal fixation was carried out on day 5. Isolate HD136 was cultured from the incision secretion on day 11. Together with other four strains from China, it belonged to a subclade of 17 genomes, all recovered from human across 11 countries of five continents with 71% (12/17) from Asia. Patient 4, admitted for implant removal, had drainage cultivated daily starting on postoperative day 6. HD167, isolated on day 9, belonged to the neighbor subclade of HD136, which predominantly composed of human isolates (10/12), with most originating from China (8/12).

Patient 5 underwent surgery upon admission due to the bleeding left middle finger with open fracture, presenting with a 3-cm irregular ring-shaped laceration and severe contamination. Purulent exudate appeared on day 8, and ABC was identified by culture. A. baumannii HD181 was recovered on day 13. In the phylogenetic tree, it clustered within a subclade comprising human-/avian-retracted (8 or 4 out of 14) sequences from diverse geographical regions, including China. The three most closely related isolates were avian-associated, one from Germany and two from Poland ( Supplementary Figure S1C ).

Patient 6 was hospitalized for hip replacement 40 days post injury. ABC was first identified from serosanguinous drainage from the surgical incision 6 days after surgery. Isolate HD239, as the only one from wound exudate, was fitted into a subclade of 13 isolates all collected from China, including eight from humans, one from Andrias davidianus (Chinese giant salamander), and two from Apis mellifera (honeybee) ( Supplementary Figure S1D ).

Isolates HD11 and HD13 were recovered from patients 7 and 8, respectively, both upon admission. Identified as ABC by culture and A. pittii by WGS, these two isolates showed distinct genomic profiles ( Figure 1C ). Patient 7 was hospitalized to remove orthopedic implants in the right ankle one year post surgery, presenting with a 1-cm unhealed wound and viscous yellow drainage. Culture revealed a polymicrobial infection including ABC, Proteus mirabilis (extended-spectrum beta-lactamase, ESBL-positive), and Klebsiella pneumoniae. HD11 was phylogenetically placed at the edge of A. pittii tree with an ANI of 94.6%, but shared a common ancestor with 96% of the 349 A. pittii genomes analyzed. It was most closely related to one soil isolate from the USA (GCA_003053325.1) and one human wound isolate from China (GCA_022449185.1) ( Figure 1D ). Patient 8, admitted four days after trauma, was diagnosed with a skin and soft tissue infection. Isolate HD13 clustered within a phylogenetic subclade where approximately 84% (26/31) of strains were human-associated, primarily from the USA (10) and China (9) ( Figure 1E ).

Patient 9 was admitted four hours after trauma with injured left extremity. Upon admission, isolate HD108 was recovered from a heavily contaminated deep wound, along with Escherichia coli, and was initially reported as ABC by culture, but later identified as A. soli by WGS ( Figure 1F ). A total of 35 published A. soli genomes fulfilled the phylogenetic analysis criteria of this study. Among the 36 genomes in the A. soli phylogenetic analysis, including HD108, 83% (30/36) were from human sources, primarily blood (15) and sputum (11), mostly collected in Japan (12) and China (11). Of note, HD108 was closely related to four human isolates from Japan, two environmental samples from India, and one avian isolate from China.

Among the six A. baumannii clinical isolates (HD8, HD12, HD136, HD167, HD181, and HD239), two A. pittii isolates (HD11 and HD13), and one A. soli isolate (HD108), genomic analysis identified no plasmid in any isolate. Intact prophages were detected in all isolates except A. baumannii HD8 and A. soli HD108. Notably, A. baumannii HD12, HD136, and A. pittii HD11 carried ARGs within predicted prophages. In HD11, the β-lactamase encoding gene bla _OXA-421 and efflux pump (EP) genes adeABCR were located within prophage regions. While other EP genes, adeL-adeFGH and adeIJK, were present in all six A. baumannii strains, adeABC genes were exclusively detected in HD12 ( Figure 3 ).

A. baumannii HD136 and HD181 harbored a type I-F CRISPR-Cas system. Furthermore, HD136 contained one compositive transposon carrying quinolone-resistant genes abaQ and adeL. A clear correlation between genotypic and phenotypic profiles of antimicrobial resistance (AMR) was observed in both A. baumannii HD12 and A. soli HD108, where HD12 displayed MDR and HD108 appeared susceptible to all antibiotics tested ( Figures 2 , 3 ). In contrast, the remaining five A. baumannii isolates and the two A. pittii strains harbored ARGs conferring resistance to various antimicrobial classes. However, apart from HD239, which was resistant to cefotaxime, they all exhibited intermediate resistance to ceftriaxone only, while A. pittii HD11 was phenotypically susceptible to all examined drugs ( Figures 2 , 3 ).

β-lactamase encoding genes, bla _OXA and bla _ADC, were detected in all isolates except A. soli HD108. All A. baumannii strains carried intrinsic bla _OXA-51-like genes, while both A. pittii carried bla _OXA-213-like alleles ( Figure 3 ). Of note, A. baumannii HD12 bore both bla _OXA-23, the most common OXA-type carbapenem-hydrolyzing β-lactamase, and the intrinsic OXA-51-like bla _OXA-66 (Bush and Bradford, 2020). Insertion sequence ISAba1 bracketed bla _OXA-23 and bla _OXA-66 in either opposite or same orientation, and was also present downstream of bla _ADC-30 in the reverse orientation (data not shown). HD181 harbored bla _OXA-374, bla _ADC-6, and mcr-4.3 with ISAba13 located downstream of bla _OXA-374 (data not shown). Virulence factors (VFs) involved in adherence, immune evasion, and stress survival were detected across all species, with minor differences in non-canonical VFs, such as katAB and ureB ( Supplementary Figure S2 ).

Sequence types (STs) of the isolates were shown in Figure 1 . Worth noting is that ST3401 and ST3402 were newly assigned to A. baumannii HD167 and A. pittii HD11, respectively. Corroborated by its marginalized phylogenetic status ( Figure 1A ), HD181 cannot be assigned a valid ST type because its gdhB gene lacked a suitable match in the multi-locus sequence typing scheme of A. baumannii on pubMLST.org. Additionally, no ST assignment was available for A. soli ( Table 1 ).