Introduction

The widespread distribution of antibiotic resistant bacteria (ARB) and antibiotic resistance genes (ARGs) in aquatic environments has become a global public health issue1,2,3. Totally 127 ARGs were detected in Maozhou River, China and the absolute abundance of ARGs could reach 1.23 × 105 ~ 8.89 × 106 copies·mL−14. Even a total of 297 ARGs could be detected in drinking water with the relative abundance of ~103 copies ARGs per cell5. Drug-resistant infections have caused an annual death toll of 700,000 worldwide, which was projected to reach 10 million by 20506. The advanced oxidation processes (AOPs) based on chlorine, ozone, UV irradiation, hydrogen peroxide and persulfate have been considered a vital step to eliminate waterborne ARB and ARGs, but their effectiveness remains controversial7,8,9,10. Over 67.5 mg·L−1 ozone was needed to achieve the removal of 1.2 ~ 2.7 logs ARGs, while the dosage of 177.6 mg·L−1 was required for the similar removal efficiency of ARGs from municipal wastewater11,12. One of the important factors hindering efficient elimination of waterborne antibiotic resistance is the delay between ARB inactivation and ARGs reduction13. During traditional AOPs, ARB are vulnerable to the oxidants and can be readily inactivated, subsequently releasing intracellular ARGs (iARGs) into the environment14. The released ARGs may horizontally transfer into new hosts, resulting in the emergence of new ARB15,16. Conventional disinfection with chlorine or chloramine has been reported to promote the horizontal gene transfer (HGT)17. High reagent dosage is therefore commonly required to achieve efficient elimination of waterborne ARGs, but it is impractical due to high cost and the generation of disinfection byproducts18. Therefore, alternative approaches to traditional AOPs are urgently needed for economic and efficient mitigation of antibiotic resistance dissemination (Fig. 1a).

Fig. 1: Research flowchart for developing a dual-zone strategy to eliminate waterborne antibiotic resistance.
figure 1

A Main problems and bottlenecks of currently employed advanced oxidation processes (AOPs) in the elimination of waterborne antibiotic resistance. B Research framework to achieve a dual-zone elimination strategy based on trans-membrane peroxymonosulfate (PMS) activation by piezoelectric membrane. C Zone-specific elucidation of the processes, outcomes, and mechanisms during the elimination of waterborne antibiotic resistance.

Ultrafiltration (UF) membrane separation has been widely employed in water purification and wastewater reclamation due to its high efficiency and environmental friendliness. However, traditional UF membranes could block ARB to some extent, but are mostly ineffective at removing free extracellular ARGs (eARGs) from water due to relatively large pore sizes, potentially producing ARB-rich retentate and ARB/eARGs mixed permeate with a high risk of antibiotic resistance dissemination19,20. Photocatalytic reactive or electroactive membranes have recently been developed to improve ARGs removal through in-situ radical generation21,22. However, the removal mainly occurred near the surface of catalytic membranes, but the trans-membrane process has not yet been fully utilized as a second reaction zone to facilitate ARGs elimination from the permeate.

The piezoelectric membrane can simultaneously produce ROS on the membrane surface and in the membrane interior upon application of external forces (e.g., ultrasonic irradiation or periodic hydraulic pressure), endowing the membrane with dual-zone catalytic capability near the membrane surface and during the trans-membrane transport process23,24. Additionally, the separated electron-hole pairs on a piezoelectric membrane can be further employed to activate peroxymonosulfate (PMS) to generate more free radicals25. These characteristics may enable the piezoelectric membrane to be an excellent alternative for traditional UF membranes for efficient elimination of ARB and ARGs in existing membrane-based water treatments. Although the piezoelectric membranes have been applied for the removal of organic pollutants such as pharmaceutical and personal care products26, dyes27, and antibiotics28 in recent years, the application potential of the piezoelectric membrane on the ARGs elimination is rarely reported.

Herein, we fabricated polytetrafluoroethylene/zinc oxide nanorods (PTFE-ZnO) piezoelectric electrospun nanofiber membrane with a three-dimensional network structure to develop a dual-zone strategy for the elimination of waterborne antibiotic resistance (Fig. 1b). The abundant inter-connected pores within the membrane interior can work as piezocatalytic channels to achieve excellent trans-membrane catalytic capability. Based on the trans-membrane piezoelectric activation of PMS (Supplementary Fig. 1), a dual-zone strategy was developed to enhance the efficiency of in-situ ROS generation during the elimination of waterborne antibiotic resistance (Fig. 1c). In the first stage of elimination near the membrane surface (zone 1), the ROS generated from the piezoelectric membrane surface can completely inactivate ARB in the retentate. Subsequent trans-membrane processing (zone 2) further enhanced ARGs elimination in the permeate, primarily due to facilitated interactions between ROS and ARGs as well as minimized ROS competition from inactivated ARB in piezocatalytic channels. This dual-zone strategy demonstrates that the ROS generated within the two consecutive reduction zones have diverse functions in the elimination of ARB and ARGs, which can be highly in accordance with the delay between ARB inactivation and ARGs reduction and meanwhile guide the development of more efficient techniques to mitigate antibiotic resistance. The dominant ROS for ARB inactivation and ARGs elimination through the dual-zone strategy were also identified to explore potential mechanisms. Collectively, the piezoelectric membrane is highly effective at eliminating waterborne ARB and ARGs and has great potential as a substitute for traditional UF membranes to achieve efficient mitigation of antibiotic resistance dissemination.

Results and discussion

Piezocatalytic channels in PTFE-ZnO nanofiber membrane

To enhance the mitigation of antibiotic resistance dissemination in the permeate, the trans-membrane piezocatalytic capability of the membrane was intensified by forming numerous piezocatalytic channels within the membrane interior using an electrospinning technique (Fig. 2a). The nanofibers of the pristine electrospun nanofiber membrane were composed of nano-sized PTFE particles stuck together by a polyethylene oxide (PEO) polymer. After sintering, the PEO was removed and the PTFE particles melted together, forming open three-dimensional network structures in the membrane matrix (Fig. 2b). During membrane filtration, the tortuous and inter-connected pores can function as piezocatalytic channels to continuously generate ROS for effective elimination of waterborne ARB and ARGs. Although the membranes before and after sintering had a similar average pore size of ~1.70 μm and total porosity of ~45%, a narrowed pore size distribution was observed after sintering (Fig. 2c). The results indicate that the sintering procedure can effectively homogenize pore size of the membrane without sacrificing porosity. The uniform size of piezocatalytic channels within the membrane interior could boost the piezocatalytic degradation efficiency during the trans-membrane process due to enhanced interactions between in-situ generated ROS and target pollutants29.

Fig. 2: Structural characteristics of PTFE-ZnO nanofiber membrane.
figure 2

A Schematic illustration of the synthesis of PTFE-ZnO nanofiber membrane with uniformly sized piezoelectric channels. B Morphology of nanofiber membranes and ZnO nanorods before and after sintering. C Pore size distribution of nanofiber membranes before and after sintering. D Elemental mapping images of C, F, O and Zn elements of PTFE-ZnO nanofiber membrane.

Due to excellent thermal stability, ZnO nanorods, which are typical piezoelectric inorganic materials, were chosen as additives to enhance the piezoelectric function of the PTFE nanofiber membrane (Supplementary Fig. 2). The ZnO nanorods can maintain their original structure and morphology after sintering (Fig. 2b). The energy dispersive X-ray spectroscopy (EDS) (Supplementary Fig. 2) and elemental mapping (Fig. 2d) confirmed the uniform distribution of ZnO nanorods in membrane skeletons. In addition, the ZnO nanorods appeared to be exposed at the surface of membrane skeletons, leading to the formation of wrinkles on the skeleton surface. The increased surface roughness after sintering could provide more reactive sites for the generation of ROS during the trans-membrane piezocatalytic process30. The hydrophilicity and water flux of the membrane was also improved by the addition of ZnO nanorods (Supplementary Table 2).

Piezoelectric PMS activation by PTFE-ZnO nanofiber membrane

The formation of abundant uniform piezocatalytic channels within the membrane interior can enhance the interactions between in-situ generated ROS and waterborne ARB and ARGs. Efficient generation of ROS is critical to effective elimination of waterborne ARB and ARGs as they can simultaneously oxidize various cellular substances, such as proteins, lipids and nucleic acids13. For piezoelectric materials, the piezo-charges formed on the material surface during the application of external forces are responsible for ROS generation31. According to the piezoelectric amplitude determined by the PFM analysis (Fig. 3a–c), the prepared PTFE-ZnO nanofiber membrane exhibited satisfactory piezoelectric responses to both ultrasonic irradiation (40 kHz, 180 W) and hydraulic pressure generated by pumping DI water through the membrane.

Fig. 3: Piezoelectric properties of PTFE-ZnO nanofiber membrane.
figure 3

A PFM amplitude mapping of the PTFE-ZnO nanofiber membrane in the pristine state. B PFM amplitude mapping of the PTFE-ZnO nanofiber membrane after hydraulic-driven filtration. C PFM amplitude mapping of the PTFE-ZnO nanofiber membrane treated by ultrasonic irradiation. D Current outputs of the PTFE-ZnO nanofiber membrane in peroxymonosulfate (PMS) and deionized (DI) water under ultrasonic irradiation. E Current outputs of the PTFE-ZnO nanofiber membrane in PMS under hydraulic-driven filtration with different water fluxes. FH Electron spin resonance (ESR) signals for TEMP-1O2, DMPO/DMSO-•O2, DMPO-•OH and DMPO-SO4•−, respectively, over the PTFE-ZnO nanofiber membrane in PMS or DI water under ultrasonic irradiation.

In comparison to hydraulic pressure, ultrasonic irradiation was more effective in forming piezoelectric electrets in PTFE and then producing enhanced piezoelectric effect24. This was further confirmed by monitoring the membrane’s transient piezoelectric current response on an electrochemical workstation equipped with a three-electrode system. When periodically switching on and off the ultrasonic irradiation, a reproductive current signal with a rectangular pulse type was detected between the membrane immersed in DI water and the counter electrode (Fig. 3d). The piezoelectric current density represents the intensity of electron mobility, an important index for the piezocatalytic performance32. The current density was significantly increased with the presence of PMS, indicating that the PTFE-ZnO nanofiber membrane can work as a potential PMS activator to generate more ROS (Fig. 3d)33,34,35. The PMS can also be activated by the PTFE-ZnO nanofiber membrane under periodic hydraulic pressure23,35. Although the generated piezoelectric current density was notably lower than under ultrasonic irradiation (Fig. 3e), it was found to be proportionately enhanced with increasing water flux during membrane filtration, suggesting that the frequency of applied external forces could affect the performance of PTFE-ZnO nanofiber membrane in piezoelectric PMS activation.

The electron spin resonance (ESR) analysis verified the generation of more ROS during the piezoelectric activation of PMS by the PTFE-ZnO nanofiber membrane under ultrasonic irradiation in comparison to the ultrasonic-driven piezoelectric PTFE-ZnO membrane system without PMS activation. The triplet peaks with equal intensity for TEMP-1O2, sextuplet peaks for DMPO/DMSO-·O2, quadruplet peaks with a relative intensity of 1:2:2:1 for DMPO-•OH and weak sextuplet peaks for DMPO-SO4·− were detected during the piezoelectric PMS activation (Fig. 3f–h). In contrast, only weak ESR signals corresponding to TEMP-1O2 were observed in the absence of PMS (Supplementary Fig. 3). These results clearly demonstrated that the PTFE-ZnO nanofiber membrane can effectively activate PMS to generate abundant ROS for ARB and ARGs elimination.

The ROS (e.g., 1O2, ·O2, SO4·- and ·OH) can function as strong oxidative species to eliminate waterborne microbial contaminants9,36,37. The piezoelectric activation of PMS by the PTFE-ZnO nanofiber membrane under ultrasonic irradiation was observed to exhibit the optimal elimination performance, primarily attributed to enhanced ROS generation. As the strategy was found to be equally effective at eliminating two different ARG genes (blaTEM-1 and aac(3)-II), blaTEM-1 was used as a representative gene for subsequent demonstration. Nearly complete ARB inactivation and ~2.0 log removal of ARGs were achieved within 30 min when the PTFE-ZnO nanofiber membrane was attached in the beaker containing a mixture of PMS and ARB solutions under ultrasonic irradiation (named as the US + PMS + M system) (Supplementary Figs. 4 and 5). In comparison, only ~1.0 log ARGs was eliminated by the employment of ultrasonic irradiation for PMS activation without the presence of the PTFE-ZnO nanofiber membrane (named as the US + PMS system), demonstrating the enhancement of PMS activation by PFTE-ZnO nanofiber membrane. Although ultrasonic irradiation could generate •OH radicals from water dissociation in cavitation bubbles, the amount of free radicals is insufficient for effective ARB inactivation or ARGs elimination, and even a significant increase of eARGs was observed after ultrasonic irradiation, due to facilitated release of iARGs from ruptured ARB cells38,39 (Supplementary Figs. 4b and 5b). Nevertheless, ultrasonic irradiation could effectively trigger the piezoelectric response of PTFE-ZnO nanofiber membrane for PMS activation.

The above results confirmed the potential application of piezoelectric PMS activation by the PTFE-ZnO nanofiber membrane in the mitigation of waterborne antibiotic resistance. However, in this batch study, the PTFE-ZnO nanofiber membrane only functioned as a heterogenous PMS activator to induce the ROS-dominant reduction of ARB and ARGs near the membrane surface. Due to the absence of trans-membrane process, its internal piezocatalytic channels were not utilized as an additional zone for enhanced ARGs elimination. Therefore, our subsequent work employed the PFTE-ZnO nanofiber membrane in filtration mode to construct a dual-zone elimination strategy to inactivate ARB in the retentate near the membrane surface (zone 1) and enhance ARGs elimination in the permeate during the trans-membrane process (zone 2) (Fig. 4a, b).

Fig. 4: Mechanisms and performances of the dual-zone elimination strategy based on piezoelectric membrane filtration.
figure 4

A Schematic illustration of the configuration of a cross-flow piezoelectric membrane filtration device. B Schematic presentation of the mechanism underlying the developed dual-zone strategy for the elimination of antibiotic resistance. C Removal rates of ARB and ARGs in the retentate and permeate during trans-membrane piezoelectric PMS activation by PTFE-ZnO nanofiber membrane. D Estimated rate constants of antibiotic resistance elimination during the two reaction zones. E Radical quenching tests conducted to identify the roles of multiple ROS in the removal of ARB from the retentate and permeate.

Complete inactivation of ARB and effective reduction of ARGs in retentate

As the first reduction zone of our dual-zone elimination strategy, the piezoelectric activation of PMS (1 mM) by ultrasonic-driven PTFE-ZnO nanofiber membrane filtration achieved complete inactivation of ARB (approximately 7.0 log) and partial elimination of ARGs (0.72 log for iARGs and 0.48 log for eARGs) from the retentate (Fig. 4c and Supplementary Fig. 6). The horizontal gene transfer (HGT) frequency was reduced by ~7.0 log in the retentate in comparison to the feed solution, and further became not detectable in the permeate, demonstrating the effective control of antibiotic resistance dissemination after the treatment (Supplementary Fig. 7). The blaTEM-1 was used as a representative gene. Since the membrane sieving effect resulted in concentrated ARB in the retentate (Supplementary Table 3), the ARB inactivation is critical to subsequent harmless disposal of the retentate. The higher removal rate of iARGs versus eARGs might be attributed to the release of iARGs from ruptured ARB cells, which is evidenced by the increased abundance of eARGs in the retentate from the ultrasonic-driven piezoelectric PTFE-ZnO membrane filtration system without PMS activation (Supplementary Fig. 8). Therefore, the piezoelectric PMS activation by the PTFE-ZnO membrane filtration can achieve complete inactivation of ARB and effective reduction of ARGs in the retentate and allows for subsequent harmless disposal after treatment.

To explore the mechanism of ARB inactivation and ARGs elimination in the retentate, radical quenching tests were conducted to understand the roles of ROS by employing a variety of radical scavengers, including methanol (MeOH, scavenger of SO4·−/·OH), isopropanol (IPA, scavenger of •OH), superoxide dismutase (SOD, scavenger of ·O2), L-histidine (L-his, scavenger of 1O2) and potassium iodide (KI, scavenger of piezo-induced h+)40,41. The addition of MeOH, IPA or KI caused negligible inhibition of ARB inactivation, while the SOD caused slight inhibitory activity (Fig. 4e). By contrast, the ARB inactivation was almost completely inhibited by L-his, demonstrating the crucial role of 1O2 in the ARB inactivation in the retentate. Since the iARGs are easily released from inactivated ARB cells and converted to eARGs, the ROS, including 1O2, •O2, SO4•−, •OH and h+, were all involved in iARGs elimination by causing the leakage of iARGs from ARB cells and subsequent oxidative degradation42 (Supplementary Fig. 9). However, it was observed that L-his and KI produced stronger inhibitory effects on the total ARGs (tARGs) than iARGs in the retentate. This indicated that the quenching of 1O2 and h+ could cause a relatively increased abundance of eARGs, suggesting their dominant roles in eARGs reduction. The important contribution of h+ to the ARGs elimination provided clear evidence of the function of piezoelectric nanofiber membrane in the ROS generation. Considering the strong effect of 1O2 on ARB inactivation and ARGs reduction, the roles of ·O2, SO4·−, ·OH and h+ were further evaluated by adding their corresponding radical quenchers together with L-his into the solution (Supplementary Fig. 10). When1O2 was pre-quenched, the addition of MeOH, IPA, SOD and KI caused more significant inhibition of ARGs elimination than ARB inactivation, confirming that ·O2, SO4·−, ·OH and h+ mainly participated in the ARGs elimination rather than ARB inactivation in the retentate.

To further explore the mechanism for the ARB/ARGs reduction in the retentate, the pathways for ROS generation in the retentate through the piezoelectric PMS activation by PTFE-ZnO nanofiber membrane filtration were identified through the ESR and radical quenching test results (details given in Supplementary Notes). The piezoelectric property of PTFE-ZnO nanofiber membrane is triggered by ultrasonic irradiation or periodic hydraulic pressure to form separated electron-hole pairs (e and h+) on the membrane surface (Eq. (1)), which can effectively activate PMS to generate multiple ROS through following reactions.

$${\rm{Membrane}}+{\rm{US}}\to {{\rm{e}}}^{-}+{{\rm{h}}}^{+}$$
(1)
$${{{\rm{HSO}}}_{5}}^{-}+{{\rm{e}}}^{-}\to \cdot {\rm{OH}}+{{{\rm{SO}}}_{4}}^{2-}$$
(2)
$${{\rm{O}}}_{2}+{{\rm{e}}}^{-}\to \cdot {{{\rm{O}}}_{2}}^{-}$$
(3)
$${{{\rm{HSO}}}_{5}}^{-}+{{\rm{H}}}_{2}{\rm{O}}\to {{\rm{H}}}^{+}+\cdot {{{\rm{O}}}_{2}}^{-}+{{{\rm{SO}}}_{4}}^{2-}$$
(4)
$$\cdot {{{\rm{O}}}_{2}}^{-}+{{\rm{H}}}_{2}{\rm{O}}\to {}^{1}{\rm{O}}_{2}+{{\rm{OH}}}^{-}$$
(5)
$$\cdot {{{\rm{O}}}_{2}}^{-}+\cdot {\rm{OH}}\to {{\rm{OH}}}^{-}+{}^{1}{\rm{O}}_{2}$$
(6)
$${{{\rm{HSO}}}_{5}}^{-}+{{\rm{h}}}^{+}\to {{\rm{H}}}^{+}+{{{\rm{SO}}}_{5}}^{\cdot -}$$
(7)
$$\cdot {{{\rm{O}}}_{2}}^{-}+{{\rm{H}}}_{2}{\rm{O}}\to {}^{1}{\rm{O}}_{2}+{{\rm{OH}}}^{-}$$
(8)

When the stream flows through the membrane surface, the 1O2 radicals that are mainly generated from PMS activation (Eqs. (4‒8)) are responsible for complete inactivation of ARBs and play a crucial role in effective reduction of both iARGs and eARGs. The ·O2- radicals can be formed primarily via Eq. (5) and secondarily by Eq. (3). In addition, Eqs. (2) and (8) drive the generation of ·OH and SO4·- radicals, respectively. The ROS including ·O2, ·OH, SO4·− and h+ could all participate in ARGs elimination, but mostly via the generation of 1O2 (Eqs. (4 and 68)). Collectively, these ROS facilitate complete inactivation of ARB and effective reduction of ARGs in the retentate.

Enhanced ARGs reduction during trans-membrane piezocatalytic process

Although complete inactivation of ARB and effective elimination of ARGs can be achieved in the retentate, the residual ARGs (especially eARGs) can still cause antibiotic resistance dissemination once they enter into the permeate. The trans-membrane process constitutes a second zone of the dual-zone strategy for enhanced ARGs elimination (Fig. 4b). This is mainly attributed to the presence of piezocatalytic channels within the membrane interior, which could facilitate the interaction between in-situ generated ROS and target ARGs during the trans-membrane piezocatalytic process29. Higher PMS activation efficiency was also observed during the trans-membrane process (Supplementary Fig. 11). In particular, the removal of iARGs could reach 3.69 log (>99.98%) in the permeate (Fig. 4c and Supplementary Fig. 12a). When the inactivated ARB passes through piezocatalytic channels, the generated ROS could further destroy cell structures by degrading the biomolecules (e.g., lipids and proteins), forming more cell debris in the permeate (Supplementary Fig. 13) and resulting in the release of iARGs from ruptured ARB cells22,43. Different from the processes in the retentate, only 1O2 and h+ were observed to influence iARGs reduction in the permeate, and the quenching of ·O2, SO4·− and ·OH radicals exhibited negligible effects (Supplementary Fig. 9). When 1O2 was pre-quenched, the majority of ARB would pass through the membrane in a live state, and then the addition of MeOH, IPA, SOD and KI could further suppress ARB inactivation and iARGs reduction in the permeate (Supplementary Fig. 10). These results suggested that 1O2 and h+ played primary roles in the elimination of both iARGs and eARGs during the trans-membrane piezocatalytic process, while ·O2, SO4·− and ·OH could participate in ARGs reduction and ARB inactivation when the intact ARB passes through the membrane.

After trans-membrane piezocatalysis, the abundance of eARGs was decreased from 1.4 × 104 copies·mL−1 in the retentate to 5.0 × 102 copies·mL−1 in the permeate, suggesting that the in-situ generated ROS (especially 1O2) in piezocatalytic channels can effectively degrade both the released iARGs during the trans-membrane process and residual eARGs originating from the retentate (Supplementary Fig. 12b). The efficient ARGs elimination in the permeate can be stably maintained during a continuous 3-cycle operation process (Supplementary Fig. 14a). Although Zn leaching clearly occurred at the beginning stage, it was decreased with the operation time, far lower than the permissible concentration value of 1.0 mg·L−1 in drinking water. (Supplementary Fig. 14b). Only a slight reduction in ARB inactivation and tARGs removal was observed when the tap water was used as feed solution, and there is no detectable change in the membrane morphology after the treatment, demonstrating the great application potential of the proposed system (Supplementary Figs. 15 and 16).

Due to enhanced interaction between ROS and ARGs in piezocatalytic channels and minimized ROS competition from ARB, enhanced removal of both iARGs and eARGs was achieved during the trans-membrane piezocatalytic process than in the retentate (Fig. 4d and Supplementary Table 4). Moreover, the significantly improved reduction of eARGs is likely attributed to the involvement of multiple ROS as demonstrated in Supplementary Fig. 9b. To further confirm the enhanced reduction of eARGs during the trans-membrane process, the plasmid pUC57 encoded with blaTEM-1 was chosen as representative eARGs and treated with piezoelectric PMS activation by PTFE-ZnO nanofiber membrane filtration. It was clearly observed that a majority of plasmid was removed during the trans-membrane process, demonstrating the key role of piezocatalytic channels in the elimination of antibiotic resistance from the permeate (Supplementary Fig. 17).

Herein, we demonstrate the satisfactory performance of trans-membrane PMS activation by PTFE-ZnO nanofiber membrane filtration for the control of antibiotic resistance dissemination. Different from previously reported catalytic membranes, the piezoelectric PTFE-ZnO nanofiber membrane can construct one more reduction zone for efficient elimination of ARB and ARGs due to the formation of abundant piezocatalytic channels with uniform pore size within the membrane interior. The dual-zone strategy achieved not only complete ARB inactivation and effective ARGs reduction in the retentate near the membrane surface (zone 1), but also enhanced elimination of ARGs (especially eARGs) in the permeate during the trans-membrane process (zone 2). Moreover, in comparison to traditional AOPs, the in-situ generated ROS in zone 2 can achieve higher removal of ARGs due to enhanced interactions between ROS and target ARGs as well as minimized ROS competition from ARB. In addition, our findings demonstrated that the ROS generated within the two consecutive reduction zones have diverse functions in ARB and ARGs elimination, providing good guidance for developing more efficient techniques to mitigate antibiotic resistance. Note that the hydraulic-driven pieozoelectric membrane filtration could also achieve an attractive ARGs elimination with a lower unit energy consumption (Supplementary Fig. 22). Collectively, the piezoelectric membrane-based dual-zone elimination strategy has great potential in membrane-based water treatment processes for effective mitigation of antibiotic resistance dissemination.

Methods

Fabrication of PTFE-ZnO nanofiber membrane

ZnO nanorods that were synthesized according to the method in Supplementary Notes and polyethylene oxide (PEO) were added into PTFE aqueous dispersion under sonication for at least 1 h. After mechanical stirring at room temperature for 3 h, the mixture with the composition as illustrated in Supplementary Table 1 was electrospun on the aluminum foil at an applied voltage of 21 kV, collection distance of 15 cm, feed rate of 0.6 mL·h−1 and electrospinning time of 4 h. The obtained nanofiber membrane was sintered at 380 °C for 1 h in air atmosphere to remove PEO and melt PTFE particles together.

Characterization of ZnO nanorods and PTFE-ZnO nanofiber membrane

The size and fine structure of ZnO nanorods were determined by transmission electron microscope (TEM, JEOL, JEM-2010). The field emission scanning electron microscope (FESEM) images with the energy dispersive X-ray spectroscopy (EDX) were investigated by Hitachi S4800. Pore size distribution of the membrane was measured by mercury intrusion porosimetry (Micro-CT, YXLON). X-ray diffraction (XRD) patterns were measured by Bruker D2 phaser using Cu Kα radiation at 2θ scanning range of 10–80°. The thermogravimetric analysis (TGA) was conducted with a Metter-Toledo TGA/SDTA851E. The piezoelectric properties of PTFE-ZnO nanofiber membranes were characterized by scanning probe microscope (SPM, Dimension FastScan, Bruker) with the function of piezo-response force microscopy at an applied voltage of 1.0 V. The piezoelectric current was monitored by electrochemical workstation (CHI760E, CH Instruments) with a three-electrode system according to the Supplementary Notes. The membrane was attached on a ring-shaped electrode and inserted into a filtration unit with one side of membrane exposed to background solution. The porosity and hydrophilicity of the membranes were measured by the gravimetric method and a contact angle meter (Attension Theta, Biolin Scientific, Switzerland), respectively. The water flux of the membranes was determined by using a dead-end filtration module with DI water as the feed solution. The permeate volume was recorded in the predetermined time intervals under the operation pressure of 5 bar. The above-mentioned membrane properties were illustrated in Supplementary Table 2.

Identification of reactive oxygen species

Reactive oxygen species (ROS) were detected by the electron spin resonance (ESR) with 5,5′-dimethyl-1-pyrroline-n-oxide (DMPO) and 2,2,6,6-tetramethyl-4-piperidino (TEMP) as the spin trapping agents. A piece of PTFE-ZnO nanofiber membrane with an area of 1 cm2 was added into 1-mL deionized (DI) water or PMS solution (1 mM). TEMP (50 mM), DMPO (50 mM) or DMPO/DMSO (50 mM/100 mM) was added to detect the signals of singlet oxygen (1O2), hydroxyl radical (·OH)/sulfate radical (SO4·−) or superoxide radical (·O2) by ESR, respectively. Different radical quenchers, including methanol (MeOH), Isopropanol (IPA), superoxide dismutase (SOD), L-histidine (L-his) and potassium iodide (KI), were added into PMS solution to identify the role of ·OH/ SO4·−, ·OH, ·O2, 1O2 and h+ in the reduction of ARB and ARGs, respectively. The relative molar concentration ratio of each quencher to PMS was 100:1.

Reduction of ARB and ARGs in batch study

The antibiotic resistant bacterium (E. coli bio-56954) containing amoxicillin resistance gene (blaTEM-1) and gentamicin resistance gene (acc(3)-II) was selected as the model ARB (Details available in Supplementary Notes). The ARB stock solution (~109 CFU·mL−1) was prepared according to the Supplementary Notes. For the reduction of ARB and ARGs under ultrasonic irradiation, fifty mL ARB suspension at a concentration of ~107 CFU·mL−1 in a 100-mL glass beaker was treated by ultrasonic irradiation for 30 min. After the treatment, the samples were taken for quantitative determination of ARB and ARGs abundance. The ARGs elimination efficiency was observed to be elevated by increasing the PMS concentration ranging 0.5–2 mM as shown in Supplementary Fig. S18, and 1 mM PMS was chosen in this study to exploit the mechanism. Similar to the antibiotic resistance elimination under ultrasonic irradiation in the batch study, we also investigated the reduction of ARB and ARGs by other different systems, including the addition of 1 mM PMS into ARB suspension under stirring, the addition of 1 mM PMS into ARB suspension under ultrasonic irradiation and the addition of 1 mM PMS combined with PTFE-ZnO nanofiber membrane (a size of 4 cm2) into ARB suspension under ultrasonic irradiation.

Reduction of ARB and ARGs in filtration mode

The PTFE-ZnO nanofiber membrane was placed into a lab-scale cross-flow membrane filtration apparatus. The ARB suspension with a concentration of ~107 CFU·mL−1 or PMS solution was pumped through the filter with a peristaltic pump. The power and frequency of ultrasonic cleaner were 180 W and 40 kHz, respectively. The samples were taken from the outlet of retentate and permeate, and then sodium thiosulfate was immediately added to quench the reaction.

ARB enumeration and ARGs determination in ARB solution

An aliquot of 100 μL sample was serially diluted with PBS and spread on sterile Luria-Bertani (LB) agar plates. After incubation at 37 °C for overnight, the colonies were manually counted with a detection limit of 1 CFU·mL−1. An aliquot of 1 mL sample was filtered through 0.22 μm mixed cellulose ester membrane. The DNA on the membrane was extracted for the determination of iARGs by using the Ezup Column Bacteria Genomic DNA Purification Kit (Sangon Biotech, China). The DNA in the filtrate was extracted for the determination of eARGs by using the SanPrep Column DNA Gel Extraction Kit (Sangon Biotech, China). Quantitative determination of ARGs was conducted with the real-time quantitative polymerase chain reaction (qPCR) according to the standard method as illustrated in Supplementary Notes. The logarithmic inactivation of ARB or logarithmic reduction of ARGs before and after the membrane was determined as log C0/Ce, and the removal percentage of ARGs before and after the membrane = 100 × (C0Ce)/C0, where \({C}_{0}\) and \({C}_{e}\) represent the culturable ARB concentration or ARGs abundance in feed solution and samples taken from the outlets of retentate and permeate, respectively. The removal efficiency of ARGs during the trans-membrane transport process was calculated according to the abundance difference of ARGs in the retentate and permeate.

Membrane integrity of ARB

The membrane integrity of ARB was analyzed by Flow Cytometry (FCM, Beckman, Gallios). Samples were stained with SYBR Green I and propidium iodide (PI) prior to the FCM analysis. The SYBR Green I was used to determine the total ARB cells with both intact and damaged membranes, while PI was able to stain cells with damaged membranes only. During the FCM analysis, SYBR Green I and PI stains would produce green fluorescent with intact cell membranes and red fluorescent with damaged membranes. The morphological structure of ARB cells was also observed by SEM. The samples were filtered through 0.2 μm nuclepore track-etched membrane (Whatman) and the membrane was immersed into 2.5 wt% glutaraldehyde to fix the cells at 4 °C for overnight. After rinse with PBS, the ARB cells were dehydrated by using a series of ethanol/PBS solutions (30%, 50%, 70%, 90% and 100% ethanol) and finally treated with a mixture of tertiary butyl alcohol (TBA) and ethanol (1:1, v/v). The membrane was finally freeze-dried for at least 24 h prior to the SEM analysis.

Degradation of plasmid

Plasmid pUC57 encoded with blaTEM-1 was chosen as the model eARGs (Details available in Supplementary Notes). Sample DNA was collected by using SanPrep Column DNA Gel Extraction Kit (Sangon Biotech) and the abundance of blaTEM-1 gene was quantified by qPCR and semi-quantified by gel electrophoresis (Details available in Supplementary Notes) at 100 V for 30 min using the gel electrophoresis system (Benchmark Scientific).