The strain C. portucalensis JH112 was isolated in September 2023 from a blood specimen of a 82-year-old male patient with acute myocardial infarction at Jinhua Central Hospital, Zhejiang Province, China. The sample was cultured on Columbia blood agar plates at 37°C for 24 h, followed by purification. Initial identification was performed using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF/MS, bioMérieux, France), which suggested Citrobacter braakii. To achieve accurate species-level identification, the 16S rRNA gene was sequenced and compared using the EzBioCloud database¹ (Chalita et al., 2024). Average Nucleotide Identity (ANI) analysis further confirmed the strain as C. portucalensis. In addition, ribosomal multilocus sequence typing (rMLST) was performed using the PubMLST species identification tool² (Jolley et al., 2012). This method indexes variation across 53 ribosomal protein subunit genes (rps genes) and confirmed the species identity as C. portucalensis, with 100% match to the reference profile.

Minimum inhibitory concentrations (MICs) for multiple antibiotics were determined using the broth microdilution method, including agents from the carbapenem (meropenem and ertapenem), cephalosporin (ceftazidime and cefepime), aminoglycoside, fluoroquinolone, sulfonamide, tigecycline, and colistin classes. MIC interpretations were based on the CLSI 2023 guidelines, except for tigecycline and colistin, which were interpreted according to EUCAST standards. E. coli ATCC 25922 was used as the quality control strain.

The classification of MDR and extensively drug-resistant (XDR) phenotypes was based on the criteria which consider only acquired resistance to antimicrobial agents. Intrinsic resistance mechanisms were excluded from the categorization (Magiorakos et al., 2012).

Carbapenemase production was initially screened using the NG-Test Carba 5 (NG Biotech), which detects KPC, NDM, IMP, VIM, and OXA-type carbapenemases. Positive results were confirmed by PCR amplification and Sanger sequencing for genotype identification.

Conjugation assays were performed using a liquid mating protocol. The donor strain JH112 was mixed with sodium azide-resistant E. coli J53 recipient cells at a 1:3 ratio and incubated statically in LB broth at 37°C for 18 h. Transconjugants were selected on LB agar plates containing sodium azide (100 μg/ml) and meropenem (2 μg/ml) to suppress the donor and select for transconjugants. Positive colonies were confirmed by antimicrobial susceptibility testing and PCR amplification of resistance genes. Conjugation frequency was calculated as the number of transconjugant colony-forming units (CFU) per donor CFU.

Plasmid DNA was extracted from C. portucalensis JH112, using the Qiagen Plasmid Midi Kit. The mixed plasmid preparation was electroporated into E. coli DH5α using standard conditions (2.5 kV, 200 Ω, 25 μF). Transformants were selected on LB agar plates supplemented with meropenem (2 μg/ml). Colonies growing on selective media were screened by PCR for bla_NDM–1 and bla_KPC–2 to determine which plasmid(s) had been acquired.

Genomic DNA was extracted using the TIANamp Bacterial DNA Kit (Tiangen Biotech, Beijing, China). Whole-genome sequencing of strain JH112 was conducted using a hybrid approach. Long-read sequencing was performed on the Oxford Nanopore MinION platform, and Illumina short-read sequencing was additionally performed for this strain to improve base accuracy. Hybrid genome assembly was carried out using Unicycler v0.4.9 (Wick et al., 2017), which integrates both long and short reads, and SPAdes v3.13.0 (Bankevich et al., 2012). Genome annotation was carried out with Prokka v1.14.6 (Seemann, 2014), and completeness was evaluated using BUSCO v4.1.2 (Simão et al., 2015). Resistance genes were annotated using the ResFinder (Florensa et al., 2022) and CARD databases (Alcock et al., 2020). Plasmid replicon types were identified with PlasmidFinder (Carattoli et al., 2014). Insertion sequences (ISs) were annotated using the ISfinder database (Siguier et al., 2006). Plasmid mobility elements, including oriT, relaxase, and type IV secretion system (T4SS) components, were detected using the oriTfinder tool (Li et al., 2018).

To determine the phylogenetic position of JH112 within the C. portucalensis population, all available C. portucalensis genome sequences were downloaded from the PubMLST database (as of December 2024) (Jolley et al., 2018). MLST was performed using the BIGSdb platform (Jolley and Maiden, 2010). Minimum spanning tree (MST) analysis based on MLST profiles was conducted with GrapeTree to evaluate the genetic relationships among different sequence types (STs). In addition, core genome alignment-based phylogenetic trees were visualized using the Interactive Tree of Life (iTOL)³ to determine the evolutionary clustering of JH112 among global isolates.

To characterize the genetic environment of bla_KPC–2 and bla_NDM–1, nucleotide sequences were aligned using BLASTn against representative reference plasmids in the NCBI database. Local synteny and structural variations were visualized using Easyfig v2.2.3. Circular maps of complete plasmid structures were generated using the Proksee platform⁴ to illustrate homologous regions, structural rearrangements, insertion sequences, and MGEs.

An 82-year-old male patient with end-stage renal disease, chronic heart failure, hypertension, and type 2 diabetic nephropathy was admitted to Jinhua Central Hospital in August 2023 following acute cardiovascular decompensation. He was diagnosed with ST-segment elevation myocardial infarction and transferred to the ICU after emergency coronary intervention.

Despite empirical antimicrobial treatment with ceftriaxone, linezolid, and piperacillin/tazobactam, the patient’s condition remained critical. On hospital day 3, blood cultures yielded two Gram-negative bacilli, initially identified by MALDI-TOF MS as C. braakii and Aeromonas caviae. The Citrobacter isolate was later confirmed by 16S rRNA sequencing and whole-genome sequencing as C. portucalensis, while the Aeromonas isolate could not be revived for further characterization.

Antimicrobial susceptibility testing revealed that the C. portucalensis strain exhibited an MDR profile, including resistance to carbapenems, cephalosporins, aminoglycosides, and fluoroquinolones. The A. caviae isolate also showed MDR in preliminary testing. The patient’s condition deteriorated despite intensive care, and following family consent, life-sustaining therapy was withdrawn. The patient died on 5 September 2023.

Antimicrobial susceptibility testing indicated that C. portucalensis JH112 exhibited an MDR phenotype. The strain showed high-level resistance to various β-lactams, aminoglycosides, quinolones, and combination agents, with MICs as follows: ceftazidime (>128 μg/ml), cefepime (>128 μg/ml), meropenem (32 μg/ml), ertapenem (128 μg/ml), gentamicin (4 μg/ml), rifampicin (>128 μg/ml), tobramycin (>128 μg/ml), and ceftazidime-avibactam (>128 μg/ml). The isolate remained susceptible only to tigecycline (0.25 μg/ml) and polymyxin B (<0.125 μg/ml) (Table 1).

To investigate the population structure of C. portucalensis and determine the evolutionary placement of the clinical isolate JH112, we performed two complementary analyses. First, a MST was constructed using all STs available in the PubMLST database as of December 2024. JH112 was classified as ST252, which has only one other recorded isolate in the database. Interestingly, ST252 occupied a central node within the MST, connected to multiple other STs through single-locus variants, suggesting that it may represent an ancestral or intermediary genotype within the current population structure (Figure 1). However, due to the limited number of isolates per ST and the incomplete epidemiological metadata in the database, further inferences about its prevalence or clinical importance remain tentative.

To further explore the genomic relationship of JH112 to other C. portucalensis strains, we constructed a core genome-based maximum likelihood phylogenetic tree using all publicly available genomes from PubMLST. The result showed that JH112 clustered closely with environmental isolates 2SOAct20merA (Czechia) and LACPHL-BACT202400417 (United States), both of which share allelic similarities but belong to distinct STs (ST1248 and ST1323, respectively). Despite their phylogenetic proximity, these STs differ by at least one of the seven canonical MLST loci, underscoring the limitations of classical MLST in capturing fine-scale evolutionary relationships. The absence of defined clonal complexes for these STs in the PubMLST database further reflects the early diversification phase of this species. Overall, the tree supports a model of high genomic heterogeneity within C. portucalensis, with no evidence of dominant epidemic clones or geographic clustering, consistent with sporadic emergence from diverse ecological niches (Figure 2).

Whole-genome sequencing of JH112 revealed a 4,985,931 bp chromosome (GC content: 51.6%) harboring two resistance genes: bla_CMY–49 and qnrB1. CMY-49 encodes an AmpC-type β-lactamase conferring resistance to various cephalosporins and is not inhibited by commonly used β-lactamase inhibitors. qnrB1 encodes a pentapeptide repeat protein that protects DNA gyrase, contributing to low-level quinolone resistance.

Two plasmids were identified: pJH112-2NDM (IncHI2/IncHI2A, 340,718 bp) and pJH112-3KPC (IncFII(Yp), 110,729 bp). pJH112-2NDM carries multiple resistance genes, including bla_NDM–1, two copies of bla_OXA–1, aac(3)-IIa, aac(3)-IId, aac(6’)-Ib3, msr(E), mph(E), catB3, ARR-3, and sul1, forming a complex MDR island. Additionally, it harbors tellurite resistance genes (terA, terC, terD, terE, terZ, and terW), facilitating survival under heavy metal stress. The presence of MGEs such as ISAba125, IS26, ISVsa5, and ISYps3 suggests a history of recombination and potential for HGT (Table 2 and Figure 3).

To further investigate the evolutionary relationships of the identified plasmids, phylogenetic analyses were performed using representative plasmid sequences retrieved from NCBI. The phylogenetic tree based on the complete plasmid sequences revealed that pJH112-3KPC formed a distinct lineage, separate from other known KPC-2–bearing plasmids, supporting its unique structural configuration and evolutionary divergence. In contrast, pJH112-2NDM clustered closely with pNDM-CRNMS3 and pF3221-NDM, indicating shared ancestry and potential derivation from a common IncHI2 backbone. These findings are consistent with the >90% sequence identity observed in pairwise alignment and highlight differing evolutionary pressures and mobility potential between the two plasmids (Figure 4).

The genetic context of bla_NDM–1 includes an upstream ISAba125–IS26–YokD structure, forming a typical Tn125-like transposon. This structure is incomplete in reference plasmids pNDM-CRNMS3 and pF3221-NDM, indicating that pJH112-2NDM may represent an earlier acquisition state or an independently evolved module. Downstream, a gene cluster composed of FrmB, FrmA, FrmR, ISVsa5, bla_OXA–1, IS26, and ISYps3 was identified, with potential mobility mediated by IS26 and ISYps3. Comparative analysis showed structural variations compared to reference plasmids, suggesting divergent evolutionary paths.

Together, these findings indicate that bla_NDM–1 is embedded within a complex and modular resistance island resembling a Tn125-like transposon, with extensive flanking insertion sequences (ISAba125, IS26, ISVsa5, and ISYps3) that may enhance environmental recombination and HGT, especially under antibiotic or heavy metal selective pressure. This structure differs substantially from the bla_KPC–2 genetic context, reflecting divergent mobility mechanisms.

pJH112-3KPC harbors bla_KPC–2 and bla_TEM–1, and was the only plasmid successfully transferred in conjugation experiments. It contains a complete conjugation module, including oriT, relaxase, and a fully functional T4SS. Conjugation assays confirmed successful transfer to E. coli J53, and the resulting transconjugants remained susceptible to ceftazidime-avibactam, indicating that the bla_NDM–1-carrying plasmid was not co-transferred. PCR confirmed the transfer of bla_KPC–2 only. The conjugation frequency of pJH112-3KPC was estimated at approximately 3.6 × 10^–3 transconjugants per donor cell, suggesting an efficient horizontal transfer potential under laboratory conditions.

BLAST analysis revealed that no known plasmid shares complete structural identity with pJH112-3KPC, though partial homology (up to 66% coverage and 99.9% nucleotide identity) was observed with plasmids from various Enterobacteriaceae species, indicating a potentially novel plasmid backbone. Upstream of bla_KPC–2, the ISKpn27–tnpR–ISE63 module was present; however, typical accessory elements such as hin2, ISKpn19, and ISEc15 were absent and replaced by an IS5 insertion, suggesting IS5-mediated recombination. Downstream, six conserved ORFs, hin1, and TnAs1 were fully retained, indicating a structurally complete transposition region (Figure 5).

Notably, pJH112-3KPC also carries a complete O-antigen biosynthesis gene cluster, including wzm, wzt, wbbM, wbbN, wbbO, and glf, which is uncommon in Enterobacterales plasmids and may have originated from Klebsiella species. While not classified as traditional virulence factors, these genes are involved in lipopolysaccharide biosynthesis and may enhance immune evasion and environmental fitness, suggesting that the plasmid confers both AMR and potential virulence traits.

In contrast to pJH112-2NDM, the bla_KPC–2 locus is integrated within a compact and structurally complete transposition region with functional conjugation machinery, supporting high transferability. The absence of typical accessory elements and the presence of IS5 insertion suggest independent evolution and recombination events. This plasmid thus represents a distinct mobile platform combining resistance and potential virulence features.

Although oriT and relaxase elements were present on pJH112-2NDM, no transconjugants were obtained under standard laboratory conditions, suggesting impaired or inactive conjugation functionality under the tested parameters. To further evaluate its mobility potential, three independent electroporation assays were conducted using purified plasmid DNA from JH112 and E. coli DH5α as the recipient. Despite selection on meropenem-containing media, no transformants harboring bla_NDM–1 were recovered, while a small number of colonies carrying bla_KPC–2 were detected, indicating that only pJH112-3KPC was successfully introduced. These results further support that pJH112-2NDM was non-transmissible under the tested conditions, likely reflecting impaired conjugation function.

In recent years, the prevalence of carbapenem-resistant Citrobacter species has shown a notable upward trend globally, with increasing reports from Europe, the United States, and several Asian countries including China, India, and South Africa (Nobrega et al., 2023; Sommer et al., 2024). The dissemination of resistance genes such as bla_NDM and bla_VIM has been frequently associated with these isolates. Citrobacter spp. are commonly found in humid hospital environments, sewage systems, and ICU drainage networks. Their exceptional environmental persistence enables them to serve as difficult-to-eradicate reservoirs of AMR genes and act as critical mediators of HGT (Sellera et al., 2022).

This study reports a clinical bloodstream isolate, C. portucalensis JH112, recovered from a patient with multiple comorbidities, including end-stage renal disease, hypertension, diabetes mellitus, and coronary artery disease. Notably, a MDR A. caviae was co-isolated from the same patient, although subsequent recovery failed. The coexistence of multiple MDR pathogens under such critical infection settings may facilitate interspecies gene transfer and the acceleration of resistance evolution through community-level gene exchange. This observation is consistent with previous studies linking Citrobacter bacteremia with polymicrobial infections and chronic disease comorbidities (Hashimoto et al., 2021; Goncalves et al., 2024). Members of the genus Aeromonas possess strong integrative capacities and may act as collectors or transport platforms for resistance genes, promoting the assembly and dissemination of mobile resistance islands in collaboration with species such as Citrobacter (Carusi et al., 2024).

Multilocus sequence typing analysis revealed that JH112 belongs to ST252, a rarely reported ST. Although C. portucalensis exhibits high ST diversity, most isolates display sporadic distribution, with no dominant epidemic clones identified to date, unlike high-risk lineages such as ST11 or ST258 in K. pneumoniae (Li et al., 2021; Zhang et al., 2024). The presence of multiple MDR plasmids in such strains suggests that this species may currently be in a “pre-epidemic” evolutionary phase, with the potential to emerge as a dominant MDR clone under selective pressure in high-risk settings like the ICU.

A major concern highlighted in this study is the coexistence of two distinct carbapenemase-encoding plasmids within a single C. portucalensis isolate. From a clinical and epidemiological perspective, this configuration represents a high-risk resistance profile. The plasmid harboring bla_KPC–2 demonstrated successful conjugative transfer under laboratory conditions, underscoring its potential for rapid dissemination within nosocomial environments, particularly in settings with intense antimicrobial selective pressure.

Such dual-carbapenemase producers pose a major therapeutic challenge, as they often exhibit resistance to nearly all β-lactams, including advanced β-lactam/β-lactamase inhibitor combinations like ceftazidime-avibactam and meropenem-vaborbactam, which are typically reserved for either KPC or NDM producers alone. Dual producers such as this strain are associated with poor clinical outcomes and are increasingly recognized as a significant global threat (Lee et al., 2022). This narrows available treatment options to toxic or less effective agents such as tigecycline or colistin, particularly concerning for bloodstream infections or immunocompromised patients. Consequently, infections caused by such strains are associated with increased treatment failure, prolonged hospitalization, and higher mortality.

Remarkably, one of the plasmids (pJH112-3KPC) also carried a complete O-antigen biosynthesis gene cluster, which may have been acquired from Klebsiella spp. Though not conventional virulence determinants, these genes contribute to lipopolysaccharide synthesis and may enhance immune evasion, colonization potential, and host adaptation (Hsieh et al., 2012). The fusion of resistance determinants with putative pathogenicity elements suggests that this plasmid may represent a hybrid resistance-adaptation platform with multifactorial advantages in clinical settings.

Structurally, pJH112-2NDM retained the ISAba125–IS26–YokD module upstream of bla_NDM–1, forming a typical Tn125-like transposon not observed in comparator plasmids pNDM-CRNMS3 and pF3221-NDM. This structural configuration may reflect an earlier evolutionary acquisition event or an independent integration pathway, consistent with previous reports highlighting the role of Tn125 and ISAba125 in facilitating the global dissemination of bla_NDM (Khan et al., 2017). The downstream presence of FrmA/B/R, bla_OXA–1, IS26, and ISYps3 forms a complex resistance locus, with Frm genes potentially enhancing bacterial stress tolerance. The bla_OXA–1 gene and associated IS elements may constitute a functional composite transposon, further increasing the plasticity of the resistance region.

From a clinical and epidemiological standpoint, the carriage of multiple resistance determinants—including aac(3)-IIa, aac(3)-IIb, aac(6’)-Ib3, msr(E), mph(E), catB3, ARR-3, sul1—and tellurite resistance genes (terA–terW) greatly expands the bacterium’s potential to survive in hostile environments, such as those exposed to antibiotics or heavy metals in hospital settings. The dense array of MGEs (ISAba125, IS26, ISVsa5, and ISYps3) not only reflects a recombinogenic genome structure but also poses a risk for further HGT events across different species and ecological niches.

Although pJH112-2NDM harbors recognizable oriT and relaxase elements, no conjugative transfer was observed under standard laboratory conditions. This failure is likely multifactorial. The plasmid’s T4SS is incomplete, spanning only 9 kb and lacking auxiliary genes essential for conjugation. Its large size (>340 kb) may also impose a significant metabolic burden on the host, reducing mobilization efficiency. Furthermore, IncHI2 plasmids are known to require low-temperature induction for expression of their conjugation machinery, while our experiments were conducted at 37°C—conditions that may suppress transfer activity (Forns et al., 2005). In addition, recipient incompatibility or entry exclusion systems may contribute to the observed transfer failure. These findings align with prior research by Li et al. (2022), which reported that only 3.1% of dual-carbapenemase-producing isolates were capable of simultaneous plasmid co-transfer, emphasizing the potential role of interplasmid incompatibility and regulatory constraints. While pJH112-2NDM appears non-conjugative under current conditions, its structurally intact resistance region—rich in MGEs—suggests latent mobility potential under selective pressure or future recombination, warranting ongoing genomic surveillance and infection control vigilance.

Additionally, the chromosomal presence of bla_CMY–49 and qnrB1 suggests intrinsic resistance to β-lactams and quinolones. While CMY-type β-lactamases are typically plasmid-borne, chromosomally encoded variants, such as bla_CMY–37 in C. freundii (Ahmed and Shimamoto, 2008) and bla_CMY–190 in Citrobacter youngae (Liu et al., 2025), have been documented. Moreover, both C. freundii and Enterobacter cloacae are known to carry inducible chromosomal AmpC β-lactamase genes (Sader et al., 2021). Chromosomal carriage of AmpC may facilitate inducible expression under β-lactam exposure and contribute to treatment failure (Tebano et al., 2024).

Taken together, C. portucalensis JH112 represents a clinically relevant MDR isolate with dual-carbapenemase production, structural novelty, and potential for dissemination. This study provides a comprehensive “strain–plasmid–gene” analysis of its resistance architecture and highlights the need for enhanced surveillance of cryptic resistance reservoirs in the context of One Health.