Membrane perturbation by the last-resort antibiotic polymyxin B drives biphasic regulation of horizontal gene transfer

Abstract

Although it is increasingly recognized that anthropogenic chemicals modulate horizontal gene transfer (HGT), the nature of these interactions is often more complex than a simple promotion or inhibition. The potential for a single chemical to exert opposing, concentration-dependent effects represents a critical and less-explored frontier in microbial ecology. Here, we investigate the last-resort antibiotic polymyxin B, a membrane-targeting peptide, and reveal a concentration-dependent, biphasic regulation of plasmid conjugation. Subinhibitory concentrations (0.125–0.5 mg/l) consistently inhibited the transfer of antibiotic resistance genes (ARGs) by up to 65.4%, whereas bactericidal concentrations (≥ 1 mg/l) strongly promoted it by up to 15.9-fold. This regulatory switch is driven by distinct physiological states: low-level exposure triggers defensive responses including reduced membrane permeability, whereas high-level exposure causes catastrophic membrane damage, inducing a synergistic stress response involving oxidative damage (> two-fold ROS increase) and a surge in cellular energy (up to 83.0% ATP increase) that facilitates HGT. High-concentration polymyxin B also promotes plasmid transfer in complex microbial communities derived from activated-sludge biofilms. Our findings reveal a new paradigm for the interaction between chemical stressors and microbial evolution, demonstrating that the ecological impact of contaminants on HGT cannot be predicted by monotonic models and highlighting the role of environmental hotspots in shaping the dissemination of antibiotic resistome.

Introduction

The global proliferation of antimicrobial resistance (AMR) represents one of the most pressing public health crises of the 21st century [1]. The emergence of multidrug-resistant and extensively drug-resistant Gram-negative bacteria has severely compromised the efficacy of many front-line antibiotics, making once-manageable infections increasingly difficult to treat [2]. Although the selective pressure from the overuse and misuse of antibiotics in clinical and agricultural settings is a primary driver of this crisis [3], the environmental dissemination of resistance is sustained by robust evolutionary mechanisms. Horizontal gene transfer (HGT) stands out as a fundamental engine of microbial evolution, allowing for the rapid exchange of genetic material, including antibiotic resistance genes (ARGs), across diverse bacterial populations and even taxonomic boundaries [4]. Among the pathways of HGT, bacterial conjugation is a particularly efficient route for spreading complex arrays of ARGs. This process involves the transfer of mobile genetic elements, such as plasmids [5], through direct cell-to-cell contact. The frequency and trajectory of conjugation are not static; they are dynamically modulated by a myriad of selective pressures, including exposure to a wide range of anthropogenic chemicals released into the environment [6].

The last-resort antibiotic polymyxin B, a cationic antimicrobial peptide, has re-emerged as a critical salvage therapy in an era of escalating resistance to front-line agents. Originally discovered in the 1940s but later deprioritized due to toxicity concerns, its use is now increasing substantially for treating life-threatening infections caused by multidrug-resistant and extensively drug-resistant pathogens, such as carbapenem-resistant Enterobacteriaceae [7, 8]. Its efficacy stems from a unique mode of action; unlike antibiotics with specific intracellular targets, polymyxin B exerts its bactericidal effect through a direct, physical perturbation of the bacterial outer membrane. This process is initiated by an electrostatic interaction with the lipid A moiety of lipopolysaccharides (LPS), displacing the divalent cations (Ca²⁺ and Mg²⁺) that stabilize the membrane structure and leading to a loss of integrity [9]. However, the long-term sustainability of this crucial antibiotic is challenged by two interconnected factors: the global spread of plasmid-mediated resistance (mcr) genes in clinical settings [10] and its significant release into the environment. Following intravenous administration, a substantial portion (60%–70%) of a polymyxin B dose is excreted unchanged in urine, a pathway potentially augmented by improper disposal of unused medication [11]. This constitutes a primary source for its entry into hospital and municipal wastewater systems. Consequently, aquatic environments, particularly wastewater treatment plants (WWTPs), are exposed to a wide spectrum of polymyxin B concentrations, from low, subinhibitory levels (μg/l) in diluted effluents to high, bactericidal levels (mg/l) closer to discharge points [12].

The presence of this wide concentration gradient poses a fundamental challenge to our understanding of its environmental impact. The direct interaction of polymyxin B with the cell membrane, the very interface where the intricate machinery of conjugation operates, suggests a more intimate and complex influence on HGT than that of other antibiotics. This raises a critical question about the nature of its dose–response relationship. Although many environmental stressors are reported to have monotonic effects on HGT (i.e. a consistent promotion or inhibition) [13–19], it is plausible that polymyxin B could exert a nonlinear, biphasic effect. At low, sublethal concentrations, membrane stress might trigger adaptive physiological responses, such as reinforcing the cell envelope or altering biofilm formation, which could hinder the conjugation process. Conversely, at high, bactericidal concentrations, catastrophic membrane failure could paradoxically facilitate plasmid influx and activate broad stress-response pathways (like the SOS response) known to promote HGT. Despite the profound implications of such a dual role, research has remained narrowly focused on clinical resistance mechanisms [20, 21]. The potential for a single, critical antibiotic to act as both an inhibitor and a promoter of ARGs dissemination depending on its environmental concentration represents a major blind spot in current ecological risk models.

This study is designed to illuminate this blind spot through a systems-level investigation that integrates molecular mechanisms with ecological consequences. We systematically dissect the biphasic effects of polymyxin B on ARGs conjugation across multiple scales of biological complexity. We begin by establishing the phenomenon in controlled in vitro systems and then proceed to a deep mechanistic exploration using a suite of advanced tools. Transcriptomic analysis (RNA-seq), coupled with functional assays for membrane permeability, membrane potential, oxidative stress (ROS), and cellular energetics (ATP), is employed to unravel the underlying molecular pathways. To establish environmental relevance, these mechanistic insights are then rigorously tested and contextualized in situ within complex microbial communities using biofilms derived from activated sludge and authentic water samples. By bridging the gap from molecular perturbation to community-level ARG dynamics and ecosystem processes, our objective is to construct a holistic model of polymyxin B’s environmental role, providing a new framework for evaluating the ecological risks of membrane-active antimicrobials.

Materials and methods

Bacterial strains, plasmids, and culture conditions

Two distinct bacterial conjugation systems were established to investigate the effects of polymyxin B. For planktonic assays, a well-characterized system was employed using Escherichia coli DH5α as the donor strain. This strain harbored the broad-host-range IncP-1 plasmid RP4-8, which carries genes conferring resistance to ampicillin and gentamicin. The recipient strain was E. coli K802, which is resistant to rifampicin, allowing for effective counter-selection of the donor. Both strains were routinely cultured at 37°C with shaking in Luria–Bertani (LB) broth, a standard medium for their optimal growth (Supplementary Text S1). To establish a baseline for exposure experiments, the minimum inhibitory concentration (MIC) of polymyxin B was determined for both strains using a standard microbroth dilution method and was found to be 1 mg/l for each (Supplementary Text S2; Supplementary Figs S1 and S2). This value was used to define the subinhibitory and supra-inhibitory concentration ranges for all subsequent experiments.

For biofilm experiments requiring clear visual tracking, a second system was utilized. The donor strain was Pseudomonas putida KT2440, a well-known environmental bacterium. This strain was specifically engineered for dual-fluorescence tracking to unambiguously identify conjugation events. It chromosomally encodes the lacI^q repressor and a red fluorescent protein (RFP), and it harbors a modified RP4 plasmid tagged with a gene for green fluorescent protein (GFP) and conferring resistance to ampicillin, kanamycin, and tetracycline [18, 22, 23]. A critical component of this system is the chromosomal LacI^q protein that efficiently represses GFP expression in the donor strain itself, ensuring that donor cells are exclusively red and preventing signal overlap.

Biofilm model system

To simulate a complex, environmentally relevant microbial habitat, biofilms were cultivated using activated sludge as the inoculum [24]. The sludge was sourced from a municipal WWTP in Hefei, China, a known hotspot for microbial diversity and HGT events. Biofilms were grown in six-well plates for 48 h to allow for the development of a mature community structure. For controlled mechanistic studies, a defined synthetic wastewater was used as the growth medium (Supplementary Table S1). For real-world validation assays, this was replaced with 0.20 μm-filtered water collected from a local river, a lake, and the influent of the same municipal WWTP. Filtration removed indigenous microorganisms and preserved the unique water chemistry of each environment.

Plasmid conjugation transfer experiments

To assess the effect of polymyxin B on HGT in a planktonic system, conjugation assays were performed using E. coli donor and recipient strains. Briefly, cells from overnight cultures were harvested by centrifugation, washed twice with phosphate-buffered saline (PBS, pH 7.2) to remove residual nutrients, and resuspended in fresh PBS. All conjugation experiments were conducted in this non-nutritive buffer to ensure that observed changes in cell populations were attributable to conjugation events and selective pressure, rather than to confounding effects from bacterial growth. Equal volumes of donor and recipient suspensions were combined and exposed to a range of polymyxin B concentrations (0, 0.125, 0.25, 0.5, 1, and 1.5 mg/l). After a standard 8-h mating period at 37°C without shaking, the mixtures were serially diluted and plated on selective agar plates to enumerate transconjugant and recipient colonies, from which conjugation frequencies were calculated (Supplementary Text S3). The successful transfer of the intact RP4-8 plasmid to transconjugants was verified using polymerase chain reaction (PCR) amplification of plasmid-specific genes (Supplementary primers in Table S2) and agarose gel electrophoresis (Supplementary Text S4).

To assess HGT in a more structured community, the engineered P. putida donor strain and polymyxin B were introduced into the pre-established activated sludge biofilms. After a 24-h co-incubation period to allow for interaction and conjugation, the outcomes were analyzed. For controlled experiments in synthetic wastewater, conjugation events were both visualized with confocal laser scanning microscopy (CLSM, Leica SP8) and precisely quantified using analytical flow cytometry (CytoFLEX, Beckman Coulter), which can differentiate between red, green, and nonfluorescent cells (Supplementary Text S5). For the validation experiments in real-world water matrices, the extent of conjugation was evaluated by CLSM, followed by quantitative image analysis using ImageJ to obtain integrated densities for green (transconjugants) and red (donors). We analyzed absolute green values and the green to red ratio, the latter controlling for variations in the donor population size caused by the antibiotic’s toxicity. During CLSM imaging, the excitation wavelength and emission wavelength for red fluorescence (donor bacteria) and green fluorescence (transconjugants) were 561 ± 25–650 ± 60 nm and 488 ± 20–525 ± 40 nm, respectively.

Analytical and mechanistic assays

To elucidate the mechanisms behind polymyxin B’s effects, a suite of phenotypic and molecular analyses was performed. To test the hypothesis that polymyxin B alters intercellular electrostatic interactions, the zeta potential of bacterial cells was analyzed using a Zetasizer Nano ZS (Malvern). The role of oxidative stress as a potential trigger for HGT was investigated by quantifying intracellular reactive oxygen species (ROS) production with a 2′,7′-dichlorofluorescein diacetate (DCFH-DA) probe, and the effect on cellular energy required for conjugation was determined by measuring intracellular ATP content using a bioluminescence-based ATP Assay Kit. To assess direct physical impacts on the cell envelope, membrane integrity was measured using propidium iodide (PI) staining and flow cytometry, and any resulting morphological changes were visualized via high-resolution scanning and transmission electron microscopy (SEM; Zeiss GeminiSEM 500 and TEM; Hitachi H-7650, respectively). Detailed methods for each assay are provided in the Supplementary Materials (Supplementary Text S6–S11).

Transcriptomic analysis (RNA-Seq)

To obtain a global, unbiased view of the cellular response to polymyxin B, we performed transcriptomic analysis on the E. coli conjugation system. Briefly, RNA-seq was performed on the mixed donor–recipient conjugation system used in the assays. Donor (E. coli DH5α carrying RP4-8) and recipient (E. coli K802) were combined in PBS lacking carbon and nitrogen sources to establish growth-independent conditions. Conjugation mixtures were incubated with final polymyxin B concentrations of 0 mg/l (control), 0.125 mg/l (subinhibitory), and 1 mg/l (MIC) at 37°C for 6 h. Following incubation, cells were harvested by centrifugation (6000 rpm, 5 min), washed with sterile PBS, and pellets were flash-frozen in liquid nitrogen and stored at −80°C. Total RNA extraction, library preparation, and sequencing were conducted by Novogene Co., Ltd. (Beijing, China). To identify genes and metabolic processes associated with the modulation of conjugation, we defined differentially expressed genes as those exhibiting a Benjamini–Hochberg adjusted P-value (P_adj) < .05 and a |log₂FoldChange| ≥ 0.585 (see Supplementary Text S12 for details).

Statistical analysis

All experiments were performed with at least three independent biological replicates to ensure statistical power, and data are presented as mean ± standard deviation (SD). Statistical significance between each treatment group and the no-drug (0 mg/l) control group was determined using one-way analysis of variance (ANOVA) followed by Dunnett’s multiple comparisons test. Prior to analysis, data were log₁₀-transformed where necessary to meet the assumptions of normality and homogeneity of variance [25–27]. If these assumptions were still violated after transformation, a robust Welch’s ANOVA with a Games–Howell post hoc test was employed instead. A P-value <.05 was considered the threshold for significance, with specific levels denoted throughout the manuscript as ^* (P < .05), ^** (P < .01), and ^*** (P < .001). All statistical analyses were performed using IBM SPSS Statistics 20.

Results

Polymyxin B exerts a concentration-dependent, biphasic effect on plasmid conjugation

To establish the fundamental dose–response relationship of polymyxin B’s effect on HGT, we conducted quantitative assays in a defined E. coli conjugation system. The results revealed a significant concentration-dependent, biphasic regulatory effect of polymyxin B on the conjugation transfer frequency of plasmids (Fig. 1A and B). At subinhibitory concentrations (0.125–0.5 mg/l), polymyxin B significantly suppressed conjugation, reducing transfer frequencies by 43.7%–65.4% compared to the untreated control. In contrast, at and above the MIC (1 mg/l), polymyxin B significantly promoted plasmid transfer, with the frequency increasing from 5.90 × 10⁻³ to 9.39 × 10⁻², an increase of ~15.9-fold over the control. At ≥MIC (therapeutic) levels, killing in PBS over 8 h is substantial but incomplete; a quantifiable survivor subpopulation persists (Supplementary Fig. S3). Absolute CFU counts show that these survivors undergo increased conjugation, indicating that high-intensity stress can enrich plasmid transfer precisely within the cells most likely to endure. Confirmation via PCR and gel electrophoresis (Supplementary Fig. S4) verified that these outcomes were due to authentic changes in conjugation efficiency and not spontaneous mutation. Furthermore, the killing kinetics curves of various concentrations of polymyxin B on the donors, recipients, and transconjugants during 8 h (Supplementary Text S13) demonstrated that there was no significant difference in the survival status of the bacterial cultures in each group (Supplementary Figs S5 and S6). This rules out the possibility that the transfer of the plasmid confers a survival advantage to the recipient bacteria, confirming that the observed increase in frequency is due to enhanced conjugative transfer. These results provide the core evidence for a paradoxical, biphasic regulatory role of polymyxin B on ARGs conjugation and set the stage for subsequent mechanistic and environmental validation.

Figure 1.

Concentration-dependent effects of polymyxin B on the conjugative transfer of plasmid-mediated ARGs. (A) Experimental workflow for assessing plasmid-mediated ARGs transfer. (B) Conjugation frequency in E. coli pure cultures exposed to varying polymyxin B concentrations. (C) Representative confocal microscopy images showing GFP-marked plasmid transfer in P. putida KT2440-activated sludge biofilms at different polymyxin B concentrations. Conjugation frequencies were log₁₀-transformed for statistical analysis, whereas the figures display nontransformed data. Error bars represent the standard deviation; ^*P < .05, ^**P < .01, ^***P < .001.

Validation of the biphasic effect in complex aquatic systems

To ascertain whether the biphasic effect observed in pure culture possessed environmental relevance, we next validated the phenomenon in more complex, ecologically pertinent systems. Within a biofilm model using an activated sludge community, both CLSM imaging (Fig. 1C) and flow cytometry analysis (Supplementary Fig. S7) showed that the biofilm results partially mirrored the planktonic system. 0.125 mg/l polymyxin B did not change transconjugant formation, whereas 1 mg/l increased conjugation frequency by 2.3-fold relative to the control. Consistent results were also observed when the experiments were conducted in authentic municipal wastewater, river, and lake water, where high concentrations of polymyxin B increased the ratio of green (transconjugants) to red (donors) integrated fluorescence density by 2.2–3.0-fold (Fig. 2; Supplementary Fig. S8). Together, these experiments confirm that the impact of high-stress conditions caused by high concentrations of polymyxin B on gene transfer is more robust in complex microbial communities across various aquatic matrices. This finding underscores the potential risk of enhanced dissemination of ARGs under such conditions and highlights the necessity of incorporating complex microbial communities and realistic environmental settings into ecotoxicological assessments.

Figure 2.

Effects of polymyxin B on the conjugative transfer of plasmid-mediated ARGs in actual aquatic environments and microbial communities. Fluorescence signals of biofilm conjugation systems exposed to different concentrations of polymyxin B in water samples collected from (A) influent of a municipal WWTP, (B) lake water, and (C) river water.

Polymyxin B differentially modulates bacterial surface charge and aggregation

To begin elucidating the mechanisms behind the dual effect, we first investigated factors influencing intercellular contact, a prerequisite for conjugation. Subinhibitory polymyxin B concentrations significantly decreased the Zeta potential of E. coli, making the surface more electronegative (Fig. 3A and B), which would increase electrostatic repulsion, hindering the close contact required for conjugation. Conversely, MIC and higher concentrations of polymyxin B significantly increased the Zeta potential (making it less negative). This is attributable to the cationic nature of polymyxin B neutralizing negatively charged LPS moieties [8, 28], thereby reducing electrostatic repulsion and facilitating cell aggregation crucial for conjugation. This observation aligns with studies showing surface charge alterations affect E. coli adhesion and ARGs transfer [29, 30]. Pearson correlation analysis also revealed a strong and highly significant positive correlation between Zeta potential and conjugation frequency for both donor (r = 0.88, P < .001) and recipient (r = 0.78, P < .001) cells (Supplementary Tables S3 and S4). Furthermore, transcriptomic data revealed that subinhibitory concentrations of polymyxin B downregulated the c-di-GMP synthase gene (dgcM), a modulation which typically inhibits biofilm formation [31, 32], whereas the MIC concentration activated the EnvZ/OmpR two-component system [14, 33], upregulated the autoaggregation factor flu_1 [34], and downregulated flagellar synthesis genes (flhC, flhD) (Fig. 3C, D, and E). Additionally, the downregulation of pdeD (c-di-GMP degradation) [35] and csgD (curli synthesis) [36] further points toward the maintenance of cell clusters. These opposing effects on cell surface properties and aggregation-related gene expression align with the observed biphasic impact on conjugation frequency.

Figure 3.

Influence of polymyxin B on bacterial intercellular contact and biofilm formation. (A, B) Zeta potential changes in donor (E. coli DH5α) and recipient (E. coli K802) bacteria at varying polymyxin B concentrations. (C) Schematic diagram of the biofilm formation pathway in E. coli showing differentially expressed genes (adapted from KEGG) at 0.125 mg/l polymyxin B. (D, E) Fold changes in E. coli biofilm-associated gene expression at 0.125 mg/l and 1 mg/l polymyxin B, respectively. Error bars represent the standard deviation; ^*P < .05, ^**P < .01, ^***P < .001.

Cell membrane permeability is modulated in a concentration-dependent manner

Given that polymyxin B is a known membrane-active peptide, we next assessed whether changes in cell envelope permeability could be a key driver [37]. Membrane permeability analysis (PI staining) showed that subinhibitory concentrations decreased the proportion of membrane-compromised cells by 17.1%–26.7% (Fig. 4A and B). This suggests a potential adaptive response reducing membrane permeability under mild stress, possibly hindering plasmid uptake [38], consistent with conjugation inhibition. However, MIC and higher concentrations of polymyxin B increased the proportion of compromised cells by up to three-fold. Electron microscopy showed that sub-MIC polymyxin B left cell envelopes intact (Supplementary Fig. S9), whereas ≥MIC exposure produced pronounced damage, including surface blebbing and wall thinning (Fig. 4C and D). Consistent with this significant membrane stress, exposure to polymyxin B at MIC induced widespread upregulation of membrane transporter systems (e.g. ABC transporters dppB, dppC, dppD, oppF), outer membrane protein machinery (lspA, lolA_1, lolB) [39], and efflux pumps (tolB) [40]) indicative of increased transport activity (Fig. 4E and F). Concurrently, upregulation of genes controlling LPS (arnC_3, pagP, eptA_1) [41, 42] and peptidoglycan (dacC, ddlB, murC, murG, ybjG) biosynthesis [43] at the MIC (Supplementary Table S5) likely reflects a compensatory response to maintain envelope integrity amidst polymyxin B-induced damage. Collectively, these data demonstrate that polymyxin B exerts opposing, concentration-dependent effects on membrane permeability, providing a direct mechanistic link to its biphasic regulation of conjugation.

Figure 4.

Impact of polymyxin B on bacterial membrane permeability and morphology. (A, B) Membrane permeability changes in donor (E. coli DH5α) and recipient (E. coli K802) bacteria at varying polymyxin B concentrations (PI staining, flow cytometry). Representative (C) SEM and (D) TEM images of E. coli cells exposed to varying polymyxin B concentrations. (E, F) Fold changes in E. coli gene expression at 1 mg/l polymyxin B: (E) membrane proteins and (F) ABC transporters. For Fig. 4A, Welch’s ANOVA was used to assess the significance, followed by the Games–Howell test for post hoc comparisons. Error bars represent the standard deviation; ^*P < .05, ^**P < .01, ^***P < .001.

Concentration-dependent oxidative stress responses are elicited

As oxidative stress is a known promoter of HGT [44, 45], we then quantified intracellular ROS levels to determine its role. No significant change in ROS levels was observed at subinhibitory concentrations. However, at and above the MIC, ROS levels increased significantly by approximately two-fold (Fig. 5A and B). Further transcriptomic analysis revealed the upregulation of oxidative stress-related pathways and genes at the MIC (Fig. 5C and D). Specifically, genes associated with antioxidant enzymes (e.g. the superoxide dismutases sodC, sodB; the catalase katE) [46] and the oxidative stress response regulator (iscR) were significantly upregulated (Fig. 5E) at the MIC of polymyxin B. This was further supported by KEGG pathway analysis showing upregulation of genes involved in antioxidant metabolism [glutathione (pepA, pepD, pepN) and ascorbate (ulaD, ulaE, ulaF, ulaG)] (Fig. 5F and G). This establishes oxidative stress as a key mechanism for the promotion of conjugation at the MIC of polymyxin B, whereas ruling it out as a cause for the inhibition observed at subinhibitory levels.

Figure 5.

Polymyxin B induces oxidative stress and modulates antioxidant defense in E. coli. (A, B) Intracellular ROS production in donor (E. coli DH5α) and recipient (E. coli K802) bacteria at varying polymyxin B concentrations. (C, D) Schematic diagrams based on KEGG pathways showing upregulation of glutathione (C) and ascorbate (D) metabolism genes in E. coli at 1 mg/l polymyxin B. (E–G) Fold changes in E. coli gene expression at 1 mg/l polymyxin B: (E) ROS generation, (F) glutathione metabolism, and (G) ascorbate metabolism. For Fig. 5A and B, Welch’s ANOVA was used to assess the significance, followed by the Games–Howell test for post hoc comparisons. Error bars represent the standard deviation; ^*P < .05, ^**P < .01, ^***P < .001.

Cellular energy metabolism is differentially impacted

Considering that conjugation is an energy-intensive process requiring ATP [19], we investigated whether polymyxin B differentially impacts cellular energy metabolism. Subinhibitory concentrations of polymyxin B did not significantly affect intracellular ATP levels. However, at and above the MIC, ATP levels increased substantially by 37.6%–82.8% (Fig. 6A). Transcriptomic analysis elucidated the basis for this energy surge, revealing that at the MIC, core metabolic pathways for energy generation were significantly upregulated (Fig. 6B and C). Specifically, genes encoding components of the primary ATP-generating pathway, oxidative phosphorylation via the electron transport chain (ETC), were significantly upregulated (Fig. 6D). These included genes for ETC Complex I (ndhC, nuoB, nuoC, nuoE, nuoF, nuoG, nuoH, nuoI, nuoJ, nuoK, nuoL, nuoM, nuoN) and Complex IV (cyoA, cyoB, cyoC, cyoD, cyoE) [47]. Concurrently, genes for ATP synthase (Complex V), which utilizes the proton gradient to synthesize ATP (atpA, atpD, atpG, atpH), were also upregulated. Furthermore, pathways feeding into central metabolism were stimulated (Fig. 6C). Genes driving fatty acid degradation via beta-oxidation (fadA, fadB, fadE, fadI, fadJ) were upregulated (Fig. 6E), providing acetyl-CoA. Additionally, key enzyme genes within the TCA cycle (acnA, sucA, sucB, sucC, sucD), which processes acetyl-CoA to fuel the ETC [48], were also significantly upregulated (Fig. 6E). This coordinated metabolic reprogramming to boost cellular energy production provides a clear energetic basis for the enhanced conjugation observed under bactericidal polymyxin B stress.

Figure 6.

Polymyxin B modulates energy metabolism in E. coli. (A) Intracellular ATP in the E. coli donor-recipient coculture across varying polymyxin B concentrations, reported to reflect the overall energy state of the conjugation system. (B, C) Schematic diagrams of (B) oxidative phosphorylation pathway, and (C) fatty acid degradation (outer) and TCA cycle (inner) pathways in E. coli at 1 mg/l polymyxin B. (D, E) Gene expression fold changes in E. coli at 1 mg/l polymyxin B related to: (D) oxidative phosphorylation and (E) fatty acid degradation and TCA cycle. Error bars represent the standard deviation; ^*P < .05, ^**P < .01, ^***P < .001.

Discussion

The spread of ARGs via plasmid conjugative transfer is a critical public health concern. Environmental stressors, including antibiotics, are known to modulate this process. Our study unveils a previously overlooked biphasic effect of the last-resort antimicrobial peptide polymyxin B on plasmid conjugation transfer. We demonstrate that sub-MIC concentrations of polymyxin B significantly inhibit plasmid transfer, whereas supra-MIC concentrations promote it. This divergent effect contrasts with the single promoting or inhibiting effect of most stressors [13–19] and even differs from the low-promotion/high-inhibition pattern reported for a few other stressors [44, 49], highlighting a unique and agent-specific response.

The biphasic effect of polymyxin B on conjugation is likely to stem from the inherent physical and chemical properties of polymyxin B as a cationic amphiphilic peptide. At lower concentrations, the polymyxin B molecules adhere to the cell membrane surface through electrostatic interactions but are insufficient to cause membrane integrity disruption. Cells adjust their surface properties by reducing membrane permeability and increasing membrane electronegativity to cope with environmental stress, which indirectly affects cell-to-cell contact and conjugation efficiency (Fig. 7). However, at higher concentrations, a large number of polymyxin B molecules insert into the membrane structure, forming stable transmembrane channels through detergent-like effects, resulting in leakage of cell contents and loss of membrane integrity [50, 51]. This strong stress triggers multilevel responses in bacteria. Severe changes such as enhanced bacterial membrane permeability, decreased membrane electronegativity, increased ROS production, and accelerated ATP generation may promote the exchange of genetic material between cells (Fig. 7). Alternatively, high antibiotic concentration pressure selects for drug-resistant populations with higher conjugation activity. Moreover, under severe envelope damage, cells can display a short-lived metabolic burst [52, 53]. ATP rises transiently via substrate-level phosphorylation from intracellular reserves [54, 55] as ATP-consuming processes are suppressed, reflecting last-ditch repair attempts.

Figure 7.

Proposed mechanisms of polymyxin B’s concentration-dependent dual effects on plasmid-mediated ARGs conjugation.

We extended our investigation beyond the E. coli model system by conducting conjugation assays in activated-sludge biofilms under authentic water matrices. Promotion at 1 mg/l polymyxin B still existed, whereas inhibition at 0.125 mg/l was weaker and not statistically significant, suggesting attenuation by matrix effects and limited power. Although this experiment in complex environmental samples is still a short-term experiment and cannot fully simulate the long-term exposure in the environment, the effect under supra-MIC is directly valuable for understanding the ecological risks of sudden high-concentration pollution events such as pharmaceutical wastewater leakage. Moreover, due to the drug disposal and drug emissions from patients’ urine [11], hospital drainage may experience a brief high drug concentration (low drainage volume and low dilution period [12]), which makes our findings directly relevant to the risk assessment in these hotspot areas. By contrast, sub-MIC exposures are highly relevant for forecasting long-term ecological effects, as these levels are ubiquitous and impose sustained selection pressure [56]. Under such conditions, polymyxin B is unlikely to cause large, immediate increases in ARG transfer.

The number of transconjugants in the group exposed to subinhibitory concentrations of polymyxin B was reduced compared to the control group, whereas the number of recipient bacteria remained almost unchanged (Supplementary Fig. S3). In the MIC and above concentration groups, the number of transconjugants initially increased to be comparable to that of the control group, but then decreased (Supplementary Fig. S3). At this concentration, polymyxin B could kill 95% or more of the bacteria, reducing the population of the recipient bacteria and increasing the conjugation frequency. However, the increase in the probability of conjugation events in this surviving small bacterial population raises many concerns. Supra-MIC levels, which correspond to clinical doses [57], still permit bacterial survival. These surviving cells are likely to acquire higher-level resistance and may act as seeds for resistance dissemination [58]. Therefore, studying supra-MIC effects is critical not only for understanding the upper limits of antibacterial efficacy but also for assessing the potential spread of drug resistance following treatment failure.

The findings of this study should be interpreted with several limitations in mind. First, our experiments address acute, short-term exposure. Over longer periods, additional ecological processes such as community-level tolerance and microbial degradation of polymyxin B may influence outcomes. Second, although our nongrowing model effectively isolated the stress-driven response, active bacterial growth introduces additional complexity regarding cell susceptibility and survivor dynamics. Future studies are needed to elucidate how these factors interact to modulate conjugation.

In summary, we demonstrate a biphasic effect of polymyxin B on plasmid conjugation, with a robust promotion at higher concentrations reproduced across model systems and complex environmental matrices. These results indicate that antibiotic impacts on plasmid transfer in environmental communities are more nuanced than commonly assumed and that conjugation among survivor subpopulations under high antibiotic stress warrants particular attention. Environmental risk assessments should therefore incorporate concentration-dependent responses, with explicit consideration of surviving populations and their potential to propagate ARGs.

Author contributions

Meng-Qi Liang (Conceptualization, Methodology, Investigation, Formal analysis, Writing—original draft), Li Yuan (Conceptualization, Supervision, Funding acquisition, Writing—review & editing), Qian-He Liu (Methodology), Jing Wu (Resources), Dong-Feng Liu (Resources), and Guo-Ping Sheng (Conceptualization, Supervision, Funding acquisition, Writing—review & editing). All authors reviewed and approved the final version of the manuscript.

Conflicts of interest

None declared.

Funding

This work was supported by the National Natural Science Foundation of China (52370208, U24A20513, and 52530001), the Key Research and Development Program of Anhui Province (2023t07020017), and University of Science and Technology of China-Xinjiang Normal University Counterpart Cooperation and Development Joint Fund (KY2400002504). We also acknowledge the Instruments Center for Physical Science, University of Science and Technology of China, for technical support.

Data availability

The raw RNA sequencing data generated in this study have been deposited in the NCBI Sequence Read Archive database under accession number PRJNA1283993. All other data supporting the findings of this study are contained within the paper and its Supplementary Information files.