Emergence of Polymyxin Resistance Driven by a PhoQ Mutation in KPC-2-Producing Klebsiella pneumoniae

TL;DR

This study identifies a specific mutation in the PhoQ protein that causes polymyxin resistance in a dangerous type of Klebsiella pneumoniae.

Contribution

The study reveals that the L96P mutation in PhoQ is a novel driver of polymyxin resistance in CRKP.

Findings

The L96P mutation in PhoQ significantly upregulates genes like phoP/Q, pmrD, and arnBCADTEF.

The PhoQ L96P mutation increases colistin and polymyxin B resistance in Klebsiella pneumoniae.

The mutation's role was confirmed through site-directed mutagenesis and antimicrobial testing.

Abstract

Background: The emergence of polymyxin-resistant, carbapenem-resistant Klebsiella pneumoniae (CRKP) presents a critical challenge to clinical management. This study aimed to delineate the molecular mechanisms driving the acquisition of polymyxin resistance in CRKP. Methods: We analyzed polymyxin-susceptible and polymyxin-resistant CRKP isolates obtained from a single patient. Antimicrobial susceptibility testing was performed to determine the minimum inhibitory concentrations. Whole genome sequencing was employed to identify variations in two-component systems and to screen for mcr genes, which were involved in polymyxin resistance. Differential gene expression was assessed using RNA sequencing and validated by quantitative real-time PCR. Furthermore, site-directed mutagenesis was utilized to confirm the causal role of specific mutations in conferring the resistant phenotype. Results: An L96P mutation in the PhoQ protein was found in the polymyxin-resistant CRKP isolate. Compared with the PhoQ wild-type, this mutation significantly upregulated expression of phoP/Q, pmrD, and arnBCADTEF operon-related genes. A corresponding L96P mutant was subsequently constructed in the polymyxin-susceptible ATCC 13883 strain via site-directed mutagenesis. Antimicrobial susceptibility testing confirmed that the PhoQ L96P mutation elevates the minimal inhibitory concentrations of colistin and polymyxin B to 64 mg/L and >32 mg/L, respectively, from a baseline of 0.5 mg/L. Conclusions: The PhoQ L96P mutation is a pivotal driver of polymyxin resistance in CRKP, primarily mediated through the upregulation of the arnBCADTEF operon.

Figures

Full text

1. Introduction

Carbapenem-resistant Klebsiella pneumoniae (CRKP) is a leading cause of hospital-acquired infections and has become a global health concern [1,2,3,4]. CRKP exhibits high resistance to most antimicrobial agents, with susceptibility restricted to a few options, including tigecycline, eravacycline, polymyxins, novel β-lactam/β-lactamase inhibitor combinations (BLBLIs), and cefiderocol [5,6,7,8,9,10]. The combination of ceftazidime-avibactam, polymyxins, and tigecycline, referred to as the “triple therapy,” has become the last-line treatment for infections caused by carbapenem-resistant Enterobacterales (CRE) [11]. However, these drugs may undergo susceptibility changes during treatment, leading to therapeutic failure [12,13,14,15,16].

Polymyxins, including polymyxin E (colistin) and polymyxin B, are cyclic polypeptide antibiotics that act by disrupting the bacterial cell membrane, leading to rapid bactericidal effects. These antibiotics are particularly effective against multidrug-resistant pathogens such as Acinetobacter baumannii, Pseudomonas aeruginosa, and CRE. Previous studies have suggested that clinicians in China should consider polymyxin as a last-resort antibiotic [17]. The increasing use of polymyxins has led to a significant rise in resistance. Research indicates that between 2014 and 2019, the resistance rate of CRKP to polymyxins increased from 0.9% to 4.5% [18]. Furthermore, the 2022 China Antimicrobial Resistance Surveillance Report revealed that the polymyxin resistance rate in CRKP had reached 7.7%. Furthermore, the emergence of polymyxin resistance during treatment can significantly reduce the drug’s efficacy and increase the risk of adverse effects [15,19,20]. Polymyxin-resistant CRKP has been identified as an independent risk factor associated with higher mortality rates and poor clinical outcomes in hospitalized patients [21]. Consequently, enhanced surveillance of polymyxin-resistant strains has become an urgent priority, and investigating the underlying mechanisms of polymyxin resistance remains a critical area of research.

Polymyxin resistance in Klebsiella pneumoniae (K. pneumoniae) is primarily mediated by modifications in lipopolysaccharide (LPS) by substituting phosphate groups with positively charged molecules such as 4-amino-4-deoxy-L-arabinose (L-Ara4N) or phosphoethanolamine (pEtN), which are often regulated by two-component systems (TCSs), including pmrA/B, phoP/Q, and crrA/B, as well as the mgrB gene, which negatively regulates the phoP/Q system. In addition, plasmid-mediated mobile polymyxin resistance genes, such as mcr and its derivatives [22], play a significant role in the horizontal dissemination of resistance.

This study aims to elucidate the in vivo evolutionary mechanism of polymyxin resistance by integrating multiple technologies, such as whole genome sequencing (WGS), gene expression analysis, and site-directed mutagenesis on polymyxin susceptible and resistant isolates from a single patient, thereby delivering valuable insights for resistance control.

2. Results

2.1. Isolation of Strains and Case Information

A 31-year-old patient was diagnosed with Burkitt’s lymphoma in an external hospital and admitted to Tongji Hospital for CAR-T treatment. Upon admission, laboratory results showed the following: procalcitonin 2.56 ng/mL, high-sensitivity C-reactive protein 160.9 mg/L, erythrocyte sedimentation rate 119 mm/h, ferritin 8143.5 µg/L, and interleukin-6 (IL-6) 101 pg/mL. Fungal elements were detected in the fecal sample. The patient’s absolute neutrophil count progressively declined, suggesting a high risk of severe infection. Empirical antimicrobial therapy was initiated, including imipenem/cilastatin, linezolid, caspofungin, and cefoperazone/sulbactam.

The 4600 copies/mL of cytomegalovirus were detected in the blood after 14 days of hospitalization. Concurrent antiviral treatment with cytomegalovirus immunoglobulin and ganciclovir was also initiated. On day 18, blood culture identified CRKP (designated as 5B1). In response, the antimicrobial regimen was adjusted with tigecycline (100 mg bid) and polymyxin B (50wu q12h) added, and imipenem/cilastatin was discontinued.

By day 35, a rectal swab identified another CRKP isolate (designated as 5RS1). On day 38, Stenotrophomonas maltophilia was detected in urine culture, and levofloxacin and minocycline were added based on AST results. Given the patient’s multiple high-risk factors—including severe neutropenia and extensive disruption of the skin-mucosal barrier (evidenced by ecchymosis, surgical wounds, and invasive indwelling catheters)—anti-Gram-positive coverage was continued empirically despite the absence of confirmatory culture results. Based on a balanced assessment of efficacy and potential adverse effects, linezolid was switched to teicoplanin. On day 39, Candida species were isolated from the drainage fluid of the nephrostomy tube, prompting further optimization of the antimicrobial regimen. The patient’s treatment regimen at that point included ganciclovir, polymyxin B, teicoplanin, tigecycline, minocycline, cefoperazone/sulbactam, and levofloxacin. Temporary bladder irrigation with amphotericin B was performed on days 46 and 47.

Despite the extensive antimicrobial therapy, the patient’s condition deteriorated on day 47. Stenotrophomonas maltophilia was still present in the urine, and a new CRKP isolate (designated as 5B2) was recovered from blood. The patient developed signs of septic shock and rapidly progressive Burkitt lymphoma, ultimately leading to multiple organ failure. The patient succumbed to his condition on day 49. A timeline summarizing the patient’s antibiotic treatment and pathogen isolation is provided in Figure 1.

Notably, the initial CRKP isolates (5RS1 and 5B1) were sensitive to polymyxins, with minimal inhibitory concentrations (MICs) of 1 mg/L for colistin (COL) and 0.5 mg/L for polymyxin B (PMB) (Table 1). However, the latter isolate (5B2) exhibited significant resistance to polymyxins, with MICs exceeding 128 mg/L for COL and 32 mg/L for PMB (Table 1), indicating the emergence of polymyxin resistance in CRKP during treatment.

2.2. Relationship Between Polymyxin Resistance, Efflux Pump Inhibitors, and Plasmids

Unfortunately, the 5B1 isolate was not preserved in our laboratory. Given the similar antibiotic susceptibility profiles observed in the 5RS1 and 5B1 isolates, we selected 5RS1 and 5B2 isolates for further investigation. Co-cultivation of high-level polymyxin-resistant isolate (5B2) with efflux pump inhibitors, including carbonyl cyanide m-chlorophenyl hydrazone (CCCP, a final concentration of 20 mg/L) and Phe-Arg-β-naphthylamide (PAβN, a final concentration of 50 mg/L), showed no reduction in the MICs of either COL or PMB, as detailed in Table 1.

Subsequently, the 5B2 isolate was co-incubated with the recipient strain Escherichia coli (E. coli) EC600, and a transconjugant (designated 5B2-trans) was selected on Mueller–Hinton agar containing meropenem and rifampicin. PCR amplification and sequencing confirmed the presence of the blaKPC-2 gene in the transconjugant. The MIC values for carbapenems (imipenem and meropenem) in the transconjugant were consistent with those of the donor strain. However, the MIC values for COL and PMB in the transconjugant were aligned with those of the recipient strain, at 1 and 0.5 mg/L, respectively (Table 1).

Interestingly, no transconjugants were obtained on plates containing both polymyxin and rifampicin. These findings suggest that carbapenem resistance in the 5B2 isolate is transferable via plasmid conjugation, whereas polymyxin resistance appears to be non-transferable through this mechanism.

2.3. Molecular Mechanism of Polymyxin Resistance

The isolates 5RS1 and 5B2 were identified as ST11 and harbor the IncFIB_(K), IncR/IncFII_pHN7A8, and ColRNAI plasmids. The structural details of their chromosomes and the three plasmids are provided in Table 2 and Figure S1. They showed no differences in terms of the acquired resistance genes (Table 2, Figure S1). Both isolates carried chromosomal resistance genes, including the oqxA and oqxB efflux pump genes, the β-lactam resistance gene blaSHV-182, the sulfonamide resistance gene sul1, the streptomycin resistance gene aadA2, and the fosfomycin resistance gene fosA6. In addition, the IncFIB_(K)_ plasmid harbored several resistance genes, including sul1, aadA16, the azithromycin resistance gene mph(A), the dihydrofolate reductase resistance gene dfrA27, the rifampin resistance gene arr-3, the quinolone resistance gene aac(6′)-Ib-cr, the tetracycline resistance gene tet(D), the chloramphenicol resistance gene floR, and the β-lactam resistance gene blaCTX-M-27. The carbapenem resistance gene blaKPC-2 was located on the IncR/IncFII_pHN7A8_ plasmid, embedded in the structure of IS26-ISKpn6-blaKPC-2-ISKpn27-Tn3-IS26. This plasmid also carried the β-lactam resistance gene blaTEM-1B and the aminoglycoside resistance gene rmtB.

Through a comparative analysis of the unique genes and variant types between the two isolates, we identified a nonsynonymous mutation in the phoQ gene that may be associated with polymyxin resistance (Table S1). The 5B2 isolate harbored a leucine-to-proline missense mutation at position 96 in phoQ (L96P; ctg → ccg). This residue is evolutionarily highly conserved, and the mutation is predicted to severely impair PhoQ protein function, with a prediction confidence score of 1.0. Functional validation confirmed that the ATCC 13883 phoQ L96P mutant exhibited an increase in the MICs of COL and PMB, from 0.5 mg/L to 64 mg/L and >32 mg/L, respectively, compared to the wild-type. In addition to the PhoQ c.287T > C (p.Leu96Pro) mutation, we detected several other genetic alterations in the 5B2 isolate, including a disruptive inframe deletion in sn-glycerol-3-phosphate acyltransferase (PlsB) (c.2279C > T, p.Ala760Val), 1-acyl-sn-glycerol-3-phosphate acyltransferase (PlsC) (c.461_466delGCGGCC, p.Arg154_Gly155del), redox-sensitive transcriptional activator (SoxR) (c.362A > G, p.Asp121Gly), and its associated regulatory protein (SoxS) (frameshift variant c.264delA, p.Leu88fs).

To further investigate the resistance mechanism, we performed transcriptome sequencing of the polymyxin-sensitive strain 5RS1 (wild-type PhoQ) and the resistant strain 5B2 (PhoQ L96P), with three biological replicates per group. This yielded a total of 14.47 Gb of high-quality clean data. The clean data per sample ranged from 2.26 to 2.46 Gb, with Q30 scores between 94.62% and 95.03% and an average GC content of 54.41% (Table S2). The mapping rates to the reference genome K. pneumoniae HS11286 were 89.0% to 92.0% (Table S2). Principal component analysis demonstrated excellent reproducibility within sensitive and resistant strains (Figure S2), indicating a stable and reliable experimental process free from significant technical bias. Compared with the 5RS1 isolate, the 5B2 isolate exhibited significantly elevated expression of phoP/Q, pmrD, and the downstream arnBCADTEF operon (Figure 2A). Additionally, the expression of genes involved in lipid A modification, including pagP (lipid A palmitoyl transferase) and lpxO (lipid A hydroxylase), was highly upregulated. In contrast, PlsB, PlsC, the pmrCAB operon, and the AcrAB-TolC efflux pump showed only slight increases in expression, with no statistically significant differences. Conversely, SoxR and SoxS were markedly downregulated. Subsequently, qPCR results confirmed the significant upregulation of genes such as phoP/Q and pmrD (p < 0.05), supporting the RNA-Seq findings (Figure 2B).

GO and KEGG analyses jointly indicate that differentially expressed genes are significantly enriched in specific pathways, such as antibiotic response and membrane modification, which enhances the reliability of sequencing results. GO enrichment analysis of the significantly upregulated genes highlighted several biological processes related to polymyxin resistance (Figure 3A,B). Among these, genes associated with “lipid A biosynthesis” (GO: 0009245) include pagP, arnE, arnT, arnD, arnA, and arnC. Additionally, those involved in the “lipopolysaccharide biosynthesis” pathway (GO: 0009103) included arnE, arnD, arnA, arnC, and arnB. Notably, genes linked to the “antimicrobial response” process (GO: 0046677) included macB, fabI, pmrD, tolC, arnD, arnA, arnC, and arnB. Furthermore, KEGG pathway analysis further supported these findings, revealing that genes involved in “cationic antimicrobial peptide resistance” (ko01503) included dsbA, cpxA, cpxR, degP, nlpE, pagP, phoQ, phoP, pmrD, tolC, arnE, arnT, arnD, arnA, arnC, and arnB. Genes associated with “lipopolysaccharide biosynthesis” (ko00540) included pagP, lpxO, and arnT (Figure 3C,D).

3. Discussion

The spread of CRKP represents a critical public health threat. This challenge is compounded when CRKP develops resistance to polymyxins, a last-line therapy, pushing patients toward untreatable infections. In this context, our study, utilizing a series of homologous CRKP clinical isolates from an individual patient, uncovers a novel mutation responsible for the development of polymyxin resistance.

The emergence of polymyxin resistance during polymyxin therapy has become increasingly common, often arising before the completion of the treatment course [15]. Previous studies have shown that polymyxin resistance typically develops in patients with carbapenem-resistant but polymyxin-susceptible Acinetobacter baumannii infections during polymyxin treatment [23]. Furthermore, CRKP infections treated with polymyxins are particularly prone to developing polymyxin resistance, especially when treatment involves prolonged courses or low-dose regimens. This resistance is typically driven by the accumulation of genetic mutations [15,24]. However, the genetic and phenotypic changes that occur in bacteria during polymyxin therapy remain incompletely characterized. Some studies have reported that CRKP can acquire resistance to both tigecycline and polymyxins via chromosomal mutations following treatment [12,13]. In our case, CRKP transitioned from being polymyxin-susceptible to polymyxin-resistant during therapy. This shift not only diminished the efficacy of polymyxin but also heightened the risk of adverse reactions.

The currently recommended polymyxin dosing regimen includes an initial loading dose of 2.5 mg/kg, followed by 1.5 mg/kg every 12 h [25]. In our study, the patient received a daily dose of polymyxin below the recommended dose, potentially resulting in suboptimal drug concentrations that failed to suppress the emergence of resistant bacterial subpopulations during treatment. Optimizing the polymyxin therapy regimen is essential for minimizing the risk of resistance development. To limit the enrichment of resistant strains, it is critical to adhere to established dosing guidelines, ensuring antibiotic concentrations remain above the mutation prevention threshold. Proper dosing is fundamental in mitigating the development of resistance caused by underdosing or inadequate drug exposure.

Polymyxin resistance in Gram-negative bacteria primarily involves several key mechanisms [8,26], including: (1) modification of lipopolysaccharide (LPS), (2) formation of capsule polysaccharides, (3) overexpression of efflux pumps such as AcrAB-TolC and MdtEF in E. coli and KpnEF, KexD in K. pneumoniae, (4) overexpression of outer membrane proteins like OprH and OmpW, and (5) overexpression of polymyxin resistance gene mcr, which is plasmid-encoded [22]. Among them, LPS modification is mainly regulated by TCSs, including crrA/B, phoP/Q, and pmrA/B [27,28]. Notably, TCSs can be activated by external cationic signals (e.g., low Mg^2+^, polymyxin exposure) or internal genetic mutations [29]. These mutations include insertion sequences (IS) that inactivate the mgrB gene [12,30], as well as specific amino acid substitutions in PhoQ [13], such as A21S [31] and L247P [32]. The mutation of phoQ leads to its constitutive activation, further activating phoP, which directly upregulates the arnBCADTEF operon or sequentially activates pmrA and pmrCAB operons through pmrD. The pmrCAB operon encodes the pEtN transferase pmrC (also known as eptA) [28]. The arnBCADTEF operon encodes a series of enzymes responsible for the synthesis of L-Ara4N. This phosphorylation cascade ultimately modifies lipid A by adding L-Ara4N or pEtN, which reduces its net negative charge. This decrease in negative charge thereby diminishes the binding affinity of cationic polymyxins to the outer membrane, reduces drug penetration, and leads to resistance [33,34].

In this study, we identified a phoQ L96P mutation in the 5B2 isolate. This mutation is relatively rare and was first described in a polymyxin-resistant CRKP strain (NJST258 1) in 2014 [35]. It was later identified in three polymyxin-resistant K. pneumoniae strains isolated from a single wastewater treatment plant in 2015 [35,36,37]. Notably, the above studies identified the phoQ L96P mutation in polymyxin-resistant strains solely through genomic sequencing, without establishing a functional genetic link. In contrast, our study demonstrates, via gene expression analysis and site-directed mutagenesis, that the phoQ L96P mutation is directly responsible for polymyxin resistance, as evidenced by a 128-fold increase in polymyxin MICs.

RNA sequencing analysis of the phoQ L96P mutant revealed a dramatic increase in the expression of phoP/Q and the arnBCADTEF operon compared to the phoQ wild-type strain, consistent with previous research [27,32,38]. However, no significant increase in the expression of the pmrCAB operon was observed. These results suggest that the phoQ L96P mutation may mediate the synthesis and transfer of L-Ara4N to LPS by phoP/Q-arnBCADTEF pathway, thereby contributing to polymyxin resistance. Previous studies have found that inhibiting fatty acid biosynthesis restores susceptibility to polymyxin in clinically relevant resistant bacteria in vivo by inducing stress responses and altering membrane composition [39,40]. PlsB and PlsC utilize two acyl-ACP molecules to generate phosphatidic acid, which serves as the precursor for all cytoplasmic membrane phospholipids and glycolipids [41]. Therefore, we speculate that mutations/upregulation of plsB/plsC may alter the membrane phospholipid composition, indirectly reducing polymyxin binding. Supporting this view, both GO and KEGG analyses revealed significant enrichment of differentially expressed genes in pathways related to “lipid A biosynthesis”, “lipopolysaccharide biosynthesis”, “antimicrobial response”, and “cationic antimicrobial peptide resistance”, processes that collectively remodel the physicochemical properties of the outer membrane [42]. The soxRS system functions as a bacterial defense system that senses external stress and stimulates soxRS global regulation [43]. Telke et al. previously reported that elevated soxRS levels can induce polymyxin heteroresistance via the AcrAB-TolC efflux pump [44]—a finding that may seem inconsistent with our observations—earlier work indicates that bacteria can optimize metabolic resource allocation to prioritize essential processes, thereby enhancing fitness in complex host environments [45,46,47]. Hence, we propose that the loss-of-function mutation and transcriptional downregulation of soxRS in the present clinical isolate likely reflect an adaptive metabolic optimization rather than a random event. Once a potent, constitutively active resistance mechanism such as the PhoQ L96P mutation is established, the soxRS-mediated AcrAB-TolC efflux pathway may become energetically costly or functionally redundant. Taken together, our findings suggest that the clinical isolate has achieved an exceptionally reinforced membrane shield through a combination of superimposed genetic mechanisms—including PhoQ L96P-driven activation of the phoP/Q-arnBCADTEF pathway and multiple modifications of lipid A and membrane phospholipids, coupled with metabolic fine-tuning via burden reduction (i.e., downregulation of SoxRS). This synergistic integration likely underlies the extreme MIC phenotype observed in the clinical isolate compared to the engineered PhoQ L96P mutant. Further investigations will be required to fully dissect these complex interactions.

Although this study has established the critical role of the phoQ L96P mutation in conferring high-level polymyxin resistance, its scope is limited by the analysis of isolates from a single patient. Additionally, the complete molecular mechanism requires further elucidation. Key limitations include the need for direct confirmation of the mutation-mediated lipid A modifications (e.g., L-Ara4N or pEtN) using mass spectrometry, as well as a deeper investigation into the conformational changes it induces and the specific regulatory network involved.

4. Materials and Methods

4.1. Species Identification and Antimicrobial Susceptibility Testing

This study included three CRKP isolates, one obtained from a rectal swab (designated 5RS1) and two from blood samples (designated as 5B1 and 5B2, respectively) of the same patient. Isolates were identified by Autof ms1000 (Autobio, Zhengzhou, China), and antimicrobial susceptibility testing (AST) was performed by the broth microdilution method [48]. Eravacycline and tigecycline were interpreted using Food and Drug Administration breakpoints (https://www.fda.gov/). The remaining antimicrobial agents were evaluated according to the Clinical and Laboratory Standards Institute (CLSI) M100 Ed35 guidelines [48]. Notably, the breakpoint for aztreonam-avibactam was referenced to aztreonam.

4.2. Plasmid Conjugation Assay

A broth mating experiment was conducted using the rifampin-resistant E. coli C600 as a recipient to assess the transferability of polymyxin resistance, following the methodology as described in previous studies [49]. Transconjugants were selected on Mueller–Hinton (MH) agar plates supplemented with 500 μg/mL rifampin and 2 μg/mL colistin to confirm the successful transfer of resistance.

4.3. WGS and Bioinformatics Analysis

The genomic DNA of all CRKP isolates was extracted from overnight cultures grown in Luria-Bertani (LB) broth using the SDS method [50]. For Illumina sequencing, genomic DNA was fragmented to 350 bp using a Covaris ultrasonicator (Covaris, Woburn, MA, USA), followed by library preparation with the NEBNext Ultra DNA Library Prep Kit (New England Biolabs, Ipswich, MA, USA). Libraries were purified using AMPure XP beads (Beckman Coulter Life Sciences, Indianapolis, IN, USA), analyzed for size distribution with Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA), quantified via quantitative real-time PCR (qPCR), and sequenced on NovaSeq 6000 platform (Illumina, Inc., San Diego, CA, USA) with 2 × 150 bp paired-end reads. In parallel, approximately 10 kb DNA fragments were recovered using the BluePippin system (Sage Science, Beverly, MA, USA), followed by end repair and adapter ligation with SQK-LSK109 kit (Oxford Nanopore Technologies plc, Oxford, UK) for sequencing on PromethION platform (Oxford Nanopore Technologies, Oxford, UK). Raw sequencing data were quality-controlled and filtered using Fastp (https://github.com/OpenGene/fastp, accessed on 15 December 2025, v0.23.2), and Nanopore data with quality scores above 7 were assessed and summarized via NanoPlot (https://github.com/wdecoster/NanoPlot, accessed on 15 December 2025, v1.29.1), ensuring a minimum sequencing depth of 300× for all isolates. Subsequent genome polishing and de novo assembly were performed with Unicycler (https://github.com/rrwick/Unicycler, accessed on 15 December 2025, v0.4.5), with annotations generated using Prokka (https://github.com/tseemann/prokka, accessed on 15 December 2025, v1.14.6) and NCBI Prokaryotic Genome Annotation Pipeline (PGAP). Antimicrobial resistance genes and plasmid incompatibility (Inc.) types were identified and analyzed using ABRicate (https://github.com/tseemann/abricate, accessed on 15 December 2025, v1.2.0).

Comparative genomic analysis and visualization were conducted using Easyfig (http://mjsull.github.io/Easyfig/, accessed on 15 December 2025, v2.2.5) and BRIG (https://sourceforge.net/projects/brig/, accessed on 15 December 2025, v0.95). Single-nucleotide polymorphisms (SNPs) and insertion–deletion mutations (InDels) were detected by Snippy (https://github.com/tseemann/snippy, accessed on 15 December 2025, v4.6.0) and validated through polymerase chain reaction (PCR) and Sanger sequencing. Based on a multiple sequence alignment of proteins across different strains, PolyPhen-2 (http://genetics.bwh.harvard.edu/pph2/ accessed on 15 December 2025, v2) was used to predict the potential functional impact of missense mutations on protein activity.

4.4. RNA Sequencing (RNA-Seq)

Total RNA was extracted from overnight cultures grown in LB broth using RNAprep pure Bacteria Kit (Tiangen Biotech, Beijing, China). A transcriptome library was prepared using the VAHTS Universal V6 RNA-seq Library Prep Kit (Vazyme, Nanjing, China) and sequenced on the Illumina platform. Raw sequencing reads were subjected to quality control using Trimomatic (https://github.com/usadellab/Trimmomatic, accessed on 15 December 2025, v0.39). Differential gene expression was assessed using DESeq2 (https://bioconductor.org/packages/release/bioc/html/DESeq2.html, accessed on 15 December 2025, v1.22.2), with genes considered significantly differentially expressed if they had an adjusted p-value < 0.05 and an absolute fold change (FC) > 2. Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed to identify the biological functions and pathways associated with the differentially expressed genes.

4.5. Quantitative Real-Time PCR (qPCR)

The gene expression differences in TCSs were validated using qPCR with three biological replicates for each sample, following the procedures and primer sequences outlined in a previous study (Table S3) [32]. Statistical analysis was performed using an independent sample t-test, with a p value < 0.05 considered statistically significant.

4.6. Site-Directed Mutagenesis

PhoQ mutant was generated in K. pneumoniae ATCC 13883 (GenBank accession: NZ_JOOW01000000) via allelic exchange using a suicide vector system. Genomic DNA from the wild-type strain (apramycin-susceptible) served as the template to amplify the upstream and downstream homologous arms of the phoQ gene using primer pairs phoQ-mut-F/phoQ-19-R and phoQ-19-F/phoQ-mut-R, respectively. These fragments were ligated into the EcoRI-HindIII-digested pUC19 vector via seamless cloning, yielding the recombinant plasmid ATCC13883-mut-phoQ-UD-pUC19. After sequence verification, the mutation-containing fragment was amplified from this plasmid using primers mut-phoQ-F/R and cloned seamlessly into the SmaI-linearized pApr73sacB vector (conferring sucrose sensitivity). Colony PCR screening with primers Apr-F/Apr-R confirmed the successful construction of the mutant plasmid phoQ-mut-pApr73sacB (conferring apramycin resistance). The resulting plasmid was electroporated into ATCC 13883 competent cells. Positive single-crossover recombinants (apramycin-resistant, sucrose-susceptible) were selected on LB agar containing 100 mg/L apramycin. These recombinants were then cultured in LB medium supplemented with 10% sucrose to select for double-crossover mutants (apramycin-susceptible, sucrose-resistant). The constructed mutant strain was verified by PCR with primers phoQ-CX-F/phoQ-JD-R and subsequent Sanger sequencing. The confirmed mutant, designated ATCC13883-mut-phoQ, was subjected to AST. All primers used in this study are listed in Table S3.

5. Conclusions

Our study offers valuable insights into the mechanisms influencing the transition of polymyxin sensitivity in the host. It underscores the importance of enhancing the monitoring of antibiotic susceptibility in clinical practice to quickly identify and address potential shifts in polymyxin sensitivity, thereby aiding in the control of drug resistance emergence and spread.