Phage-guided rapid enrichment of copper-protocatechuic acid nanoparticles for targeted antibacterial therapy

Abstract

The effective treatment of bacterial infections requires rapid and selective enrichment of antibacterial agents on bacterial surfaces within infectious microenvironments. Transition metal ion-based antibacterial agents show broad-spectrum bactericidal activity, but they usually rely on passive diffusion and stochastic bacterial contact, limiting their ability to establish sufficient local doses within short treatment windows. Herein, an M13 phage-mediated targeting strategy was developed for the rapid enrichment and delivery of copper ion–protocatechuic acid nanoparticles (CP NPs). Through phage display, an M13 phage targeting methicillin-resistant Staphylococcus aureus (MRSA-targeting phage, MTP) and an M13 phage targeting Pseudomonas aeruginosa (P. aeruginosa-targeting phage, PTP) were identified. MTP and PTP retained the typical filamentous morphology of M13 phage and selectively adhered to corresponding bacteria within 1 min. Protocatechuic acid (PCA) was then used as a polyphenolic bridge for phage-surface adhesion and Cu²⁺ coordination, yielding CP@MTP and CP@PTP, which integrate bacterial recognition, metal–polyphenol assembly, and copper-based antibacterial activity into filamentous delivery scaffolds. After only 10 min of bacterial exposure, CP@MTP selectively reduced MRSA by 3 log, while CP@PTP reduced P. aeruginosa by 1.7 log, likely owing to rapid phage-mediated CP NP enrichment on bacterial surfaces. Transcriptomic analysis further revealed bacterial stress responses induced by targeted copper delivery. In zebrafish larval tail-fin infection models, CP@MTP and CP@PTP effectively eradicated corresponding bacteria and reduced neutrophil recruitment. In a murine P. aeruginosa pneumonia model, CP@PTP prolonged lung retention, reduced pulmonary bacterial burden, and alleviated inflammatory lung injury. This study establishes targeted M13 phages as spatial regulators for rapid and precision antibacterial therapy.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12951-026-04803-6.

Introduction

The effective treatment of bacterial infections requires antibacterial strategies that can rapidly act at pathogen surfaces in complex infectious microenvironments [1]. In infected tissues, bacteria are often embedded in mucus, biofilms, necrotic debris, or inflammatory matrices [2, 3], which restrict drug transport and reduce effective contact between antibacterial agents and bacterial cells [4]. This challenge is particularly critical in acute wound and pulmonary infections, where bacterial proliferation and inflammatory injury can progress rapidly [5, 6]. Therefore, antibacterial materials should not only exhibit potent bactericidal activity, but also establish an effective local dose at bacterial surfaces within a short therapeutic window. Conventional antibiotic therapy is increasingly challenged by bacterial adaptation and resistance development [7]. Antibacterial agents that act through a single or limited number of biochemical targets can impose strong selective pressure, allowing bacteria to survive through target modification, drug efflux, enzymatic degradation, or metabolic compensation [8, 9]. By contrast, transition-metal ions have long been used for antimicrobial disinfection because they can attack bacteria through multiple parallel mechanisms [10, 11]. Among them, copper ions (Cu²⁺) can damage bacterial membranes, disturb metal-ion homeostasis, impair protein function, disrupt energy metabolism, and induce oxidative stress [12–14]. This multi-target antibacterial mode makes copper-based agents less dependent on a single vulnerable pathway and may reduce the likelihood of resistance development compared with conventional single-target antibiotics [15, 16].

Recent studies on copper-induced cell death have further highlighted the biological consequences of copper overload [17]. In bacterial systems, excessive copper can disrupt metal homeostasis, induce redox pressure, damage proteins, and interfere with central metabolism, producing a broad stress state that is difficult to compensate through a single resistance mechanism. However, dissociative Cu²⁺ and conventional copper-containing nanoparticles usually lack bacterial recognition [18, 19]. They mainly rely on passive diffusion and stochastic contact with bacterial cells, making it difficult to generate a sufficient local copper concentration on bacterial surfaces within short treatment windows. Increasing the total copper dose may improve bacterial killing, but it also increases nonspecific exposure and potential cytotoxicity [20]. Thus, copper-based antibacterial therapy requires a strategy that can rapidly enrich copper at target bacterial surfaces without simply increasing the administered dose.

Bacteriophages provide a useful biological scaffold for bacterial recognition [21]. Phage display is a well-established technique for screening target-binding peptides, proteins, or phages, and has been widely applied in ligand discovery and biomedical engineering [22, 23]. In this technique, foreign gene fragments are inserted into genes encoding phage coat proteins, allowing the corresponding peptides or proteins to be displayed on the phage surface [24, 25]. A phage display library is incubated with a target, after which weakly bound or unbound phages are removed by washing [26]. The retained phages are then eluted, amplified in host bacteria, and subjected to further selection rounds, eventually yielding phages with specific affinity for the target [22, 27].

In this study, methicillin-resistant Staphylococcus aureus (MRSA) and Pseudomonas aeruginosa (P. aeruginosa) were selected as representative Gram-positive and Gram-negative bacterial strains, respectively. Instead of using predefined molecular receptors, whole bacterial cells were used as targets for M13 phage display screening. This strategy enabled the enrichment of bacteria-binding phages without prior knowledge of specific surface receptors. M13 is a non-lytic filamentous phage whose natural host is F-pilus-bearing Escherichia coli ER2738 [28], therefore, the selected M13 phages were not expected to directly infect or lyse MRSA or P. aeruginosa. Instead, they were used as targeting carriers. After screening, the obtained phages could be further modified because the M13 capsid surface contains amino, carboxyl, N-terminal, and other hydrophilic or hydrophobic groups [29]. Metal-polyphenol assembly provides a mild and versatile route to introduce copper-based antibacterial components onto this phage surface.

Here, a targeted M13 phage-mediated platform was constructed for rapid enrichment of Cu²⁺-protocatechuic acid nanoparticles (CP NPs) on bacterial surfaces (Scheme 1). Phage display was used to identify an MRSA-targeting phage (MTP) and a P. aeruginosa-targeting phage (PTP). Protocatechuic acid (PCA) was attached to MTP or PTP as a polyphenolic bridge and then coordinated with Cu²⁺, forming CP NPs on the phage surface. The resulting CP NPs-modified MTP and PTP were denoted CP@MTP and CP@PTP, respectively. A Cu²⁺/PCA mixture at the same Cu²⁺ equivalent concentration was denoted CP. The central design goal was to reshape the spatial distribution of copper-based antibacterial agents rather than simply increase the total copper dose. Unlike CP, which remains largely dispersed during short treatment windows, CP@MTP and CP@PTP rapidly enriched CP NPs on the corresponding bacterial surfaces, producing enhanced bactericidal effects after short bacterial exposure. The antibacterial efficacy of CP@MTP and CP@PTP was further evaluated in zebrafish tail-fin infection models and a murine P. aeruginosa pneumonia model. Together, this study establishes a phage-guided metal-polyphenol platform for rapid, targeted, and multi-mechanistic antibacterial therapy with reduced reliance on conventional antibiotic-like single-target killing.

Scheme. 1.

Schematic illustration of the preparation of Cu²⁺-PCA nanoparticle (CP NPs)-modified targeted phages (CP@MTP and CP@PTP), their targeted antibacterial process, and their therapeutic application in zebrafish larval infection and mouse pneumonia models. Created in BioRender. Ji, J. (2026) https://BioRender.com/2go6kka

Results

Screening and characterization of targeted phages

A Ph.D.™ -7 phage display peptide library was used for screening. This M13-based library displays random heptapeptides on the pIII capsid protein [26]. MRSA and P. aeruginosa were selected as representative Gram-positive and Gram-negative bacterial models, respectively, to screen MRSA-targeting phage (MTP) and P. aeruginosa-targeting phage (PTP). Briefly, the Ph.D.™ -7 library was incubated with MRSA or P. aeruginosa to allow target-binding phages to adhere to bacterial surfaces. Weakly bound and unbound phages were removed by washing, whereas bacteria-retained phages were eluted and amplified by infecting E. coli ER2738. The amplified phages were then subjected to the next selection round. After 6 binding-washing-elution-amplification cycles, MTP and PTP were obtained. Transmission electron microscopy (TEM) showed that both MTP and PTP retained the typical filamentous morphology of M13 phage, with a length of approximately 1 μm and a width of approximately 5 nm (Fig. 1a, b). MTP showed preferential adhesion to Gram-positive bacteria, including MRSA, Enterococcus faecalis (E. faecalis), and Staphylococcus epidermidis (S. epidermidis), whereas PTP showed stronger adhesion to Gram-negative bacteria, including P. aeruginosa and Escherichia coli (E. coli) (Fig. S1). These results indicated that the targeting behavior of MTP and PTP depends on the type of bacterial strains. MTP showed strong targeting ability to Gram-positive bacteria. PTP showed strong targeting ability to Gram-negative bacteria. Sanger sequencing further confirmed that both phages carried defined foreign inserts, corresponding to the heptapeptide sequences GSLVRTA for MTP and ANARLPV for PTP (Fig. S2, S3). Their targeting ability was then evaluated by TEM and fluorescence colocalization. TEM images showed abundant MTP adhesion on the MRSA surface, with some phages adopting an approximately vertical orientation, suggesting that the displayed peptide region might interact with MRSA surface structures (Fig. 1b). In contrast, PTP showed no obvious adhesion to MRSA. Conversely, PTP was markedly enriched on the surface of P. aeruginosa, whereas MTP showed limited binding. These results established MTP and PTP as matched targeting phages and non-targeting controls in the two bacterial models: MTP served as the targeting phage for MRSA and the non-targeting control for P. aeruginosa, whereas PTP served as the targeting phage for P. aeruginosa and the non-targeting control for MRSA.

Fig. 1.

Preparation and characterization of targeted phages and Cu²⁺-PCA nanoparticle-modified targeted phages (CP@MTP and CP@PTP). (a) Schematic illustration of the preparation of CP@MTP and CP@PTP. Created in BioRender. Ji, J. (2026) https://BioRender.com/2go6kka. (b) TEM images of MTP and PTP after incubation with MRSA or P. aeruginosa. Bacterial pellets were collected by centrifugation after 10 min of incubation for TEM observation. (c) CLSM colocalization images of DAPI-labeled MRSA or P. aeruginosa after incubation with FMTP or FPTP for 1 min. (d) Phage titers of MTP and PTP determined by plaque assays, showing titers on the order of 10¹¹ PFU/mL, and TEM images of CP@MTP and CP@PTP. (e) UV-vis absorption spectra of CP@MTP and CP@PTP. (f) FTIR spectra of CP@MTP and CP@PTP. (g) High-resolution Cu 2p XPS spectrum of CP@MTP

Fluorescence colocalization further confirmed the target ability of MTP and PTP. Fluorescein isothiocyanate-labeled MTP (FITC-labeled MTP, FMTP) and FITC-labeled PTP (FPTP) were incubated with MRSA and P. aeruginosa, respectively. Confocal laser scanning microscopy (CLSM) showed that the green fluorescence of FMTP rapidly overlapped with DAPI-labeled MRSA, whereas FPTP produced no obvious signal on MRSA (Fig. 1c). Conversely, FPTP labeled P. aeruginosa within a short time, whereas FMTP showed only weak binding. Clear targeted adhesion was observed within 1 min, indicating that the selected phages possessed minute-scale bacterial-recognition ability. Because carrier amplifiability is essential for subsequent material preparation, phage plaque assays were used to measure the amplification titers of MTP and PTP. Both phages retained strong infectivity toward E. coli ER2738, and their titers reached the order of 10¹¹ PFU/mL after a single round of amplification (Fig. 1d and Fig. S4), which provided sufficient carrier material for constructing CP NP-modified targeted phages.

Construction and characterization of CP@MTP and CP@PTP

Although MTP and PTP provided selective bacterial-recognition scaffolds, rapid bacterial killing required the introduction of a copper-based bactericidal module. Direct Cu²⁺ loading onto the phage scaffold (named Cu@MTP and Cu@PTP) was first evaluated. After MTP or PTP was incubated with Cu²⁺ and dialyzed to remove unbound ions, the resulting Cu@MTP and Cu@PTP retained only 8.6 and 6.5 µg/mL Cu²⁺, respectively, even when the Cu²⁺ feeding concentration was increased to 64 µg/mL (Fig. S5). This limited retention indicated that direct interactions between Cu²⁺ and phage surface proteins were insufficient for efficient copper loading. To overcome this limitation, protocatechuic acid (PCA) was introduced as a polyphenolic bridge for phage-surface adhesion and Cu²⁺ coordination. PCA was first incubated with MTP or PTP at titers of approximately 10¹¹ PFU/mL, allowing it to attach to the phage surface through interactions between polyphenolic hydroxyl groups and capsid-protein amino groups (Fig. 1a). Cu²⁺ was then added to coordinate with surface-bound PCA, forming Cu²⁺-PCA nanoparticles (CP NPs) on the phage scaffold. After dialysis removed unbound PCA, Cu²⁺, and free small-molecule components, CP NP-modified targeted phages were obtained and denoted CP@MTP and CP@PTP. Compared with direct Cu²⁺ loading, PCA-mediated metal-polyphenol assembly markedly improved copper retention. At a Cu²⁺ feeding concentration of 32 µg/mL, the Cu²⁺ loading efficiency reached approximately 92% for CP@MTP and 81% for CP@PTP. This loading efficiency was markedly higher than Cu²⁺ loading alone. Further increasing the feeding concentration to 64 µg/mL reduced the loading efficiency, suggesting that effective loading approached saturation near 32 µg/mL (Fig. S6).

Transmission electron microscopy (TEM) showed numerous particulate structures on CP@MTP and CP@PTP. These particles were approximately 20 nm in diameter and were distributed along the M13 filamentous scaffold (Fig. 1d). This morphology supported the formation of CP NPs after Cu²⁺ coordination with PCA on the phage surface. A small number of free particles were also observed in the TEM background, suggesting that part of PCA and Cu²⁺ assembled into free CP NPs in solution. Nevertheless, the dominant phage-associated structures were CP NPs distributed on MTP or PTP. Spectroscopic and elemental analyses further verified the material composition. Ultraviolet-visible (UV-vis) spectra showed PCA-related absorption in CP@MTP and CP@PTP. The absorption at 256 nm was assigned to PCA, whereas the band near 286 nm was attributed to the overlap between phage protein absorption and polyphenol absorption (Fig. 1e). Fourier-transform infrared (FTIR) spectroscopy showed absorption changes at 1650 and 1605 cm^-1 in CP@MTP compared with MTP, supporting the involvement of PCA in phage-surface assembly (Fig. 1f). X-ray photoelectron spectroscopy (XPS) detected Cu 2p signals in CP@MTP and CP@PTP, together with characteristic Cu²⁺ satellite peaks, indicating the presence of Cu²⁺ in the loaded nanoparticles (Figs. 1g and S7-S9). Zeta-potential measurements showed that both phages remained negatively charged after CP NP loading, suggesting that nanoparticle modification did not reverse the overall surface charge (Fig. S10). Together, these results indicated that PCA acted as the key bridge for Cu²⁺ retention and CP NP formation on targeted phage surfaces. This design integrated bacterial recognition by MTP/PTP, surface adhesion and coordination by PCA, and copper-based antibacterial activity into a single M13 filamentous scaffold, thereby forming a targeted copper-delivery system.

In vitro antibacterial activity of CP@MTP and CP@PTP

The above results demonstrated that CP NPs were successfully constructed on MTP and PTP surfaces. To determine whether the modified phages could still recognize target bacteria and deliver antibacterial components, rapid in vitro antibacterial assays were performed. MRSA and P. aeruginosa were separately incubated with H₂O, MTP, PTP, CP, CP@MTP, and CP@PTP for 10 min. Unbound materials were then removed, and the bacteria were further incubated before colony counting. As shown in Fig. 2a and c, and S11, MTP and PTP alone didn’t show obvious antibacterial activity against MRSA, confirming that non-lytic M13 phage could not directly kill non-host bacteria. CP and non-targeted CP@PTP showed weak antibacterial effects against MRSA. In contrast, CP@MTP reduced MRSA viability by 3 orders of magnitude, indicating that MTP-mediated delivery markedly improved the rapid antibacterial efficiency of CP NPs against MRSA. Similarly, for P. aeruginosa, only CP@PTP showed a strong antibacterial activity and reduced bacterial counts by 1.7 orders of magnitude, whereas CP and non-targeted CP@MTP showed limited antibacterial effects (Fig. 2a and d, and S11). This targeting-dependent antibacterial activity showed that CP@MTP and CP@PTP exhibited stronger antibacterial activity within a short time only when matched with their corresponding target bacteria. Live/dead staining was consistent with colony counting. MRSA treated with CP@MTP showed strong red fluorescence from dead bacteria, and P. aeruginosa treated with CP@PTP also displayed intense red fluorescence. In contrast, red fluorescence of dead-bacteria was weak in the CP and non-targeted phage groups (Fig. 2b). Crystal violet staining showed that CP@MTP and CP@PTP efficiently removed MRSA and P. aeruginosa biofilms, respectively, with stronger effects than the non-targeted groups (Fig. S12). These results directly showed that targeted delivery greatly improved the bactericidal efficiency of Cu²⁺ over a short time.

Fig. 2.

In vitro antibacterial activity of CP@MTP and CP@PTP. (a) Representative agar plate images of MRSA and P. aeruginosa after incubation with CP@MTP or CP@PTP for 10 min. (b) LIVE/DEAD fluorescence staining images of MRSA and P. aeruginosa after incubation with CP, CP@MTP, or CP@PTP for 10 min. Green: live bacteria; Red: dead bacteria. (c) Bacterial counts of MRSA after incubation with MTP, CP, CP@MTP, or CP@PTP for 10 min. (d) Bacterial counts of P. aeruginosa after incubation with PTP, CP, CP@MTP, or CP@PTP for 10 min. (e) Bacterial counts of MRSA after incubation with CP@MTP or CP@PTP at different Cu²⁺ equivalent concentrations (0, 5, 10, 15, and 20 µg/mL) for 10 min. (f) Bacterial counts of P. aeruginosa after incubation with CP@MTP or CP@PTP at different Cu²⁺ equivalent concentrations (0, 5, 10, 15, and 20 µg/mL) for 10 min. (g) Bacterial counts of MRSA after incubation with CP@MTP or CP@PTP for different times (0, 10, 30, 60, and 120 min). (h) Bacterial counts of P. aeruginosa after incubation with CP@MTP or CP@PTP for different times (0, 10, 30, 60, and 120 min). The control group received an equal volume of sterile water. The Cu²⁺ equivalent concentration in panels (a)-(d) and (g), (h) was 16 µg/mL. A Cu²⁺/PCA direct mixture at the same Cu²⁺ equivalent concentration was denoted CP

The effect of equivalent Cu²⁺concentration on antibacterial performance was further examined. As the Cu²⁺concentration in CP@MTP or CP@PTP increased, the bactericidal effect of the targeted delivery groups against their corresponding bacteria gradually strengthened (Fig. 2e, f). At equivalent Cu²⁺ concentration of 20 ug/mL, non-targeted phage-mediated metal delivery reduced bacterial counts by only approximately 1 order of magnitude, whereas targeted delivery showed much stronger antibacterial activity. This result indicated that antibacterial efficiency depended not only on copper dose but also strongly on phage targeting. Incubation-time experiments further confirmed the rapid action of this system (Fig. 2g, h). In the non-targeted phage groups, antibacterial activity gradually increased as incubation time was extended. When the incubation time reached 120 min, bacterial counts decreased by approximately 2 orders of magnitude. In contrast, the targeted delivery groups reduced bacterial counts by approximately 2.5 orders of magnitude within 10 min. These findings showed that targeted phages markedly shortened the onset time of copper-based antibacterial materials, converting Cu²⁺ from long-duration passive-contact antibacterial agents into minute-scale actively delivered antibacterial agents.

Mechanistic analysis of rapid bactericidal activity

The rapid bactericidal activity of CP@MTP and CP@PTP was primarily attributed to their ability to reshape the spatial distribution of free Cu²⁺. CP mainly rely on diffusion and random contact during short treatment windows, making it difficult to establish a sufficiently high local concentration on bacterial surfaces. In contrast, MTP and PTP recognized their corresponding bacteria within 10 min and delivered the loaded CP NPs directly to bacterial surfaces (Fig. 3a). Thus, targeted phages did not simply increase the total antibacterial dose; rather, they redistributed dispersed Cu²⁺ in solution into locally enriched nanostructures at the bacteria-material interface very quickly. To examine this process, electron microscopy was first used to visualize the interactions between CP NPs-modified phages and bacteria. Transmission electron microscopy (TEM) showed that abundant CP@MTP adhered to the MRSA surface after 10 min of coincubation. The bacterial boundary became blurred, accompanied by apparent structural disruption and intracellular-content leakage. Similarly, CP@PTP strongly adhered to the surface of P. aeruginosa and was associated with visible outer-membrane damage (Fig. 3b, c). Scanning electron microscopy (SEM) further showed extensive coverage of MRSA by CP@MTP and marked enrichment of CP@PTP on P. aeruginosa, together with bacterial shrinkage and rupture. These observations indicated that targeted M13 phages rapidly concentrated CP NPs onto the corresponding bacterial surfaces, whereas non-targeted groups showed much weaker local coverage. Cryogenic TEM (cryo-TEM) was then used to assess bacterial ultrastructure under a near-hydrated state. After targeted delivery, both MRSA and P. aeruginosa showed severe disruption of the cell wall/membrane envelope, with shrinkage, collapse, and internal structural disorder. By contrast, non-targeted groups exhibited less material attachment and milder morphological damage. Together, these imaging results supported that local Cu²⁺ enrichment was closely associated with rapid bacterial-envelope disruption.

Fig. 3.

In vitro antibacterial mechanism of CP@MTP and CP@PTP. (a) Schematic illustration of the interaction processes of CP and CP@MTP with MRSA. Created in BioRender. Ji, J. (2026) https://BioRender.com/2go6kka. (b) TEM, SEM, and cryo-TEM images of MRSA after incubation with CP, CP@MTP, or CP@PTP for 10 min. Bacterial pellets were collected by centrifugation for imaging. (c) TEM, SEM, and cryo-TEM images of P. aeruginosa after incubation with CP, CP@MTP, or CP@PTP for 10 min. Bacterial pellets were collected by centrifugation for imaging. (d) Copper content in MRSA pellets after incubation with CP, CP@MTP, or CP@PTP for 10 min. (e) Copper content in P. aeruginosa pellets after incubation with CP, CP@MTP, or CP@PTP for 10 min. (f) K⁺ concentration in the supernatant after MRSA was incubated with CP, CP@MTP, or CP@PTP for 10 min. (g) K⁺ concentration in the supernatant after P. aeruginosa was incubated with CP, CP@MTP, or CP@PTP for 10 min. The control group received an equal volume of sterile water. The Cu²⁺ equivalent concentration in panels (b)-(g) was 16 µg/mL

To quantify this targeted enrichment, copper content in bacterial pellets was measured after treatment. Under the same incubation time and Cu²⁺ dose, MRSA treated with CP@MTP contained approximately 4-fold more copper than that treated with CP@PTP. Conversely, P. aeruginosa treated with CP@PTP contained approximately 4-fold more copper than that treated with CP@MTP (Fig. 3d, e). These results confirmed that targeted M13 phages redirected CP NPs from nonspecific dispersion toward the corresponding bacterial microenvironment, thereby increasing bacteria-associated copper accumulation. Because copper accumulation was accompanied by marked envelope damage, K⁺ release was further measured to evaluate membrane permeability (Fig. 3f, g). K⁺ is mainly distributed inside bacterial cells, and its release into the supernatant indicates increased membrane permeability or impaired membrane integrity. Compared with CP, targeted phage-mediated metal delivery increased K⁺ release by approximately 4-fold, indicating stronger disruption of the bacterial membrane barrier. This result was consistent with the membrane rupture, bacterial shrinkage, and intracellular-content leakage observed by electron microscopy. Based on these results, the rapid antibacterial mechanism of CP@MTP and CP@PTP can be summarized as a targeting-induced local Cu²⁺ enrichment process. Targeted phages first adhered to corresponding bacterial surfaces within 10 min and redistributed CP NPs from a dispersed solution state to a bacteria-surface-enriched state. This local enrichment increased bacteria-associated copper accumulation and induced Cu²⁺ overload, which further damaged bacterial envelope structures, enhanced membrane permeability, promoted K⁺ leakage, and caused intracellular-content leakage, bacterial shrinkage, and lytic death. This mechanism explained why targeted delivery achieved rapid antibacterial activity within 10 min, whereas non-targeted groups required prolonged incubation to approach a similar effect. More importantly, it showed that the key value of this system lies in rapidly changing the spatial concentration distribution of antibacterial agents, rather than simply increasing the total antibacterial dose.

Transcriptomic analysis

To elucidate the antibacterial mechanisms of CP@MTP and CP@PTP, transcriptome sequencing was performed on MRSA and P. aeruginosa after material treatment. Volcano plots showed extensive transcriptional reprogramming in both bacterial species (Fig. 4a, b). In MRSA, 1296 differentially expressed genes (DEGs) were identified, accounting for 55% of all detected genes. In P. aeruginosa, 3297 DEGs were identified, accounting for 61%. These results indicated that CP@MTP and CP@PTP induced strong bacterial stress responses. Gene Ontology (GO) enrichment analysis showed that the DEGs in both bacteria were mainly enriched in translation, ribosome-related processes, and protein synthesis. This suggested that material treatment caused global metabolic and protein-synthesis remodeling in bacteria (Fig. 4c, d). This change more likely reflected an adaptive response to metal stress and cellular damage rather than direct ribosomal targeting by the materials. Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis further supported this interpretation (Fig. 4e, f). In MRSA, pathways such as ABC transporters and Staphylococcus aureus infection were significantly enriched, suggesting that transmembrane transport, metal homeostasis, and virulence-associated processes were affected. In P. aeruginosa, the pathways more closely associated with material treatment included oxidative phosphorylation, carbon metabolism, and lipopolysaccharide biosynthesis, indicating that CP@PTP markedly disturbed the respiratory chain, energy metabolism, and outer-membrane lipopolysaccharide (LPS) homeostasis.

Fig. 4.

Transcriptomic sequencing analysis. (a) Volcano plot of MRSA in the control and CP@MTP groups. (b) Volcano plot of P. aeruginosa in the control and CP@PTP groups. (c) GO enrichment analysis of MRSA in the control and CP@MTP groups. (d) GO enrichment analysis of P. aeruginosa in the control and CP@PTP groups. (e) KEGG enrichment analysis of MRSA in the control and CP@MTP groups. (f) KEGG enrichment analysis of P. aeruginosa in the control and CP@PTP groups. (g) Heatmap of selected genes in MRSA in the control and CP@MTP groups. (h) Heatmap of selected genes in P. aeruginosa in the control and CP@PTP groups. The control group received an equal volume of sterile water. The Cu²⁺ equivalent concentration in panels was 16 µg/mL

Heatmaps of mechanism-related genes revealed both shared stress responses and species-specific features. In MRSA, upregulation of metal homeostasis genes, including copA, copZ, and znuB [30], indicated that CP@MTP disrupted not only copper handling but also broader metal homeostasis involving zinc and other heavy metals (Fig. 4g). This imbalance was accompanied by increased expression of oxidative stress and repair genes, such as azoR, ahpC, and radA [31], suggesting elevated redox pressure and damage repair demand. In parallel, altered expression of isaA and lrgA indicated disturbed cell wall remodeling and autolysis related responses, consistent with impaired cell wall homeostasis [32, 33]. Changes in biofilm and virulence regulators further suggested that CP@MTP interfered with MRSA adhesion, colonization, stress adaptation, and pathogenic regulation. Meanwhile, P. aeruginosa showed a distinct response pattern (Fig. 4h). Classical copper resistance and multi metal efflux genes, such as pcoB and czcA, were downregulated, whereas compensatory metal transport genes, including oprC and znuB, were upregulated [34, 35]. These results suggested that CP@PTP caused copper detoxification failure and broader metal homeostasis disruption, rather than simply activating metal efflux. Oxidative stress genes, such as katA and ahpC [36], were upregulated, together with changes in respiratory chain genes, indicating redox stress associated with disturbed electron transport and energy metabolism. As a Gram-negative bacterium, P. aeruginosa mainly responded through outer membrane and LPS remodeling. Upregulation of lpxC and waaC suggested compensatory envelope remodeling rather than a simple increase in membrane permeability [37]. Meanwhile, downregulation of lasI, pelA, and pcrV indicated suppression of quorum sensing [38], biofilm formation, and virulence secretion systems. And, qPCR analysis showed that CP@MTP induced stronger upregulation of ahpC and copA in MRSA, while CP@PTP induced stronger upregulation of oprC and katA in P. aeruginosa, compared with the control and non-targeted groups (Fig. S13).

Both bacteria showed disrupted metal homeostasis, redox stress, and altered virulence and biofilm regulation. MRSA displayed more prominent cell wall remodeling and autolysis responses, whereas P. aeruginosa showed stronger disruption of copper detoxification, respiratory chain activity, and outer membrane/LPS remodeling. Overall, the transcriptomic results suggested that CP@MTP and CP@PTP exerted antibacterial effects through multiple mechanisms and were not likely to cause resistance. The resistance-development tendency of CP@MTP and CP@PTP was compared with clinically used antibiotics over three treatment cycles. The minimal inhibitory concentration (MIC) of vancomycin and meropenem increased by more than 4-fold and 8-fold, respectively, whereas CP@MTP and CP@PTP showed smaller MIC increases of about 3-fold and 2-fold (Fig. S14), suggesting a lower tendency to induce drug resistance. Together, these findings support a multitarget antibacterial mechanism for phage copper metal materials, involving copper overload, metal homeostasis disruption, oxidative stress, envelope damage, and inhibition of biofilm formation and virulence.

Targeted bacterial clearance on a zebrafish larval tail-fin infection model

Before in vivo anti-infection evaluation, the biocompatibility of CP@MTP and CP@PTP was assessed. Dissociative Cu²⁺ showed obvious cytotoxicity at 20 µg/mL, whereas CP@MTP and CP@PTP almost didn’t show detectable cytotoxicity at the same equivalent Cu²⁺concentration (Figs. S15, S16). This result indicated that immobilizing Cu²⁺ as phage-associated CP NPs reduced the cytotoxicity associated with dissociative Cu²⁺ exposure. Based on this improved biocompatibility, a zebrafish larval tail-fin infection model was used to evaluate the targeted delivery platform in a living local-infection microenvironment (Fig. 5a) [39, 40]. Fluorescent imaging confirmed that both FITC-labelled MRSA and FITC-labelled P. aeruginosa colonized the injured tail-fin region, indicating successful establishment of the infection models (Fig. S17). In uninfected larvae with tail-fin injury, neither FMTP nor FPTP produced obvious green fluorescence at the wound site, suggesting limited nonspecific accumulation of the targeted phages to injured tissue alone (Fig. S18). By contrast, FMTP was markedly enriched at MRSA-infected tail fins, whereas FPTP showed only weak signals (Fig. 5b, c). Conversely, FPTP accumulated strongly at P. aeruginosa-infected regions, whereas FMTP showed limited accumulation. These results showed that MTP and PTP retained selective recognition of their corresponding target bacteria in a complex in vivo environment.

Fig. 5.

Targeted bacterial clearance in zebrafish larval tail fins. (a) Schematic illustration of the establishment and treatment of the zebrafish larval tail-fin bacterial infection model. Created in BioRender. Ji, J. (2026) https://BioRender.com/2go6kka. (b) Fluorescence images of MRSA-infected zebrafish larvae after incubation with FMTP or FPTP for 15 min. (c) Fluorescence images of P. aeruginosa-infected zebrafish larvae after incubation with FMTP or FPTP for 15 min. (d) Bacterial counts in MRSA-infected zebrafish larvae after incubation with CP, CP@MTP, or CP@PTP for 15 min. (e) Bacterial counts in P. aeruginosa-infected zebrafish larvae after incubation with CP, CP@MTP, or CP@PTP for 15 min. (f) SEM images of the infected tail-fin regions in MRSA-infected zebrafish larvae after incubation with CP, CP@MTP, or CP@PTP for 15 min. (g) Quantification of neutrophil recruitment at the tail-fin region in MRSA-infected zebrafish larvae at 4 and 16 h after different treatments. (h) Quantification of neutrophil recruitment at the tail-fin region in P. aeruginosa-infected zebrafish larvae at 4 and 16 h after different treatments. (i) Fluorescence images showing neutrophil recruitment at the tail-fin region in MRSA-infected zebrafish larvae at 4 and 16 h after different treatments. (j) Fluorescence images showing neutrophil recruitment at the tail-fin region in P. aeruginosa-infected zebrafish larvae at 4 and 16 h after different treatments. The control group received an equal volume of sterile water. The Cu²⁺ equivalent concentration used for treatment was 16 µg/mL

The therapeutic efficacy of CP@MTP and CP@PTP was then evaluated. In the MRSA infection model, CP@MTP reduced bacterial counts in the infected tail-fin region by approximately 3.5 orders of magnitude, whereas free CP NPs showed weak antibacterial effect and non-targeted CP@PTP reduced bacterial counts by only approximately 0.4 orders of magnitude (Fig. 5d). In the P. aeruginosa infection model, CP@PTP reduced bacterial counts by approximately 2.8 orders of magnitude, whereas free Cu²⁺ and non-targeted CP@MTP reduced bacterial counts by only approximately 0.6 and 0.8 orders of magnitude, respectively (Fig. 5e). These results demonstrated that phage-mediated targeted nanoparticles remained substantially more effective delivery of Cu²⁺ than free Cu²⁺ or non-targeted delivery in vivo. Consistent with the bacterial-count results, scanning electron microscopy (SEM) showed abundant filamentous phages and particulate structures retained at MRSA-infected regions after CP@MTP treatment, accompanied by obvious MRSA shrinkage and rupture (Fig. 5f). Similarly, CP@PTP accumulated extensively in P. aeruginosa-infected regions and induced clear bacterial morphological damage (Fig. S19). These morphological observations supported the targeted retention of CP NPs-modified phages and their local antibacterial effect at infected tissues.

The effect of targeted bacterial clearance on infection-induced inflammation was further examined using Tg(lyz: EGFP) zebrafish larvae. After MRSA infection, the control, CP, and non-targeted CP@PTP groups showed extensive neutrophil recruitment at 4 h, with cell numbers exceeding 60 at the tail-fin wound. In contrast, the CP@MTP group showed fewer than 40 neutrophils at the wound site (Fig. 5g, h). At 16 h, neutrophil numbers did not further increase in the CP@MTP group, whereas the control and non-targeted groups maintained stronger inflammatory responses, with cell numbers remaining above 50. Similarly, in the P. aeruginosa infection model, CP@PTP significantly reduced neutrophil recruitment at the infected tail-fin region (Fig. 5i, j). These results indicated that targeted treatment not only reduced bacterial burden, but also alleviated infection-induced inflammation by rapidly removing the local infectious source.

In vivo therapeutic efficacy in a murine -induced pneumonia model

To further evaluate the therapeutic efficacy of this targeted delivery strategy in a complex mammalian infection environment, an acute P. aeruginosa pneumonia model was established in mice. P. aeruginosa is a clinically important pulmonary pathogen that can rapidly proliferate in lung tissue and induce strong inflammatory injury [41]. Compared with the zebrafish tail-fin model, this pneumonia model provided a more stringent setting for assessing antibacterial efficacy in deep infected tissues. The experimental procedure is shown in Fig. 6a. P. aeruginosa was first inoculated into mouse lungs by bronchial administration to establish acute pneumonia. After infection was established, mice were randomly divided into 4 groups and treated by bronchial administration for two consecutive days. CP were used to evaluate the pulmonary antibacterial effect of free Cu²⁺, whereas CP@MTP served as a non-targeted phage-delivery control. CP@PTP was used to determine whether PTP-mediated targeting could improve the treatment of P. aeruginosa lung infection.

Fig. 6.

Targeted treatment of P. aeruginosa pneumonia in mice. (a) Schematic illustration of the establishment and treatment of the mouse P. aeruginosa pneumonia model. Created in BioRender. Ji, J. (2026) https://BioRender.com/2go6kka. (b) Fluorescence images of lung tissues at different time points (0.1, 4, 12, 24, and 36 h) after bronchial administration of FITC-labeled CP@MTP (F-CP@MTP) and FITC-labeled CP@PTP (F-CP@PTP). (c) Images of lung tissues after treatment with CP, CP@MTP, or CP@PTP in the mouse P. aeruginosa pneumonia model. (d) Kaplan-Meier mortality curves of mice in different treatment groups (log-rank test, p = 0.0659). (e) Body-weight changes of mice in different treatment groups. (f) P. aeruginosa counts in lung homogenates collected from different treatment groups on day 3. (g) WBC counts in peripheral blood from different treatment groups on day 4. (h) Neu# counts in peripheral blood from different treatment groups on day 4. (i) Quantification of alveolar area in lung tissues from different treatment groups. (j) H&E, Masson, and EVG staining images of lung tissues from different treatment groups. The control group received an equal volume of sterile water. The Cu²⁺ equivalent concentration used for treatment was 20 µg/mL

Prior to therapeutic evaluation, CP@MTP and CP@PTP were administered to healthy mice for biosafety assessment. Hematoxylin and Eosin (H&E) staining of major organs showed no obvious pathological abnormalities, indicating good in vivo biosafety (Fig. S20). The retention of CP@MTP and CP@PTP in infected lung tissue was first compared. Lung fluorescence imaging showed that CP@PTP remained detectable in infected lungs for up to 36 h, whereas the CP@MTP fluorescence signal largely disappeared 12 h after administration (Fig. 6b). This difference indicated that PTP-mediated recognition of P. aeruginosa enhanced local retention of the material in pulmonary infectious foci, while non-targeted CP@MTP was more readily cleared from the lungs. This improved retention translated into better therapeutic outcomes. Gross lung images showed marked congestion, edema, and dark-red lesions in the control group, indicating severe infection and inflammatory injury. These abnormalities were partially alleviated in the CP and CP@MTP groups but remained evident. In contrast, lung color and morphology in the CP@PTP group were closer to those of healthy tissue (Fig. 6c). Survival analysis further showed that mortality was 50% in the control group at the end of treatment, but decreased to 10% after CP@PTP treatment, which was lower than that in the CP and CP@MTP groups (Fig. 6d). Consistently, the body-weight loss induced by P. aeruginosa infection was markedly alleviated after CP@PTP treatment (Fig. 6e).

The antibacterial effect was then quantified by colony counting of lung homogenates. Large numbers of P. aeruginosa remained in the control group, indicating persistent infection. CP partially reduced the pulmonary bacterial burden, while CP@MTP showed weaker antibacterial activity than CP@PTP. By contrast, CP@PTP significantly decreased the bacterial burden in lung tissue (Fig. 6f), demonstrating that PTP-mediated targeting improved the antibacterial efficiency of CP NPs in the pulmonary infection environment. In parallel, routine blood analysis showed that white blood cell (WBC) and neutrophil (Neu) counts were markedly increased after infection, consistent with an acute systemic inflammatory response. After CP@PTP treatment, WBC and Neu levels nearly reduced to those of healthy mice (Fig. 6g, h), indicating effective attenuation of infection-induced systemic inflammation.

Histological analysis on day 4 further supported the lung-protective effect of CP@PTP. H&E staining showed large and evenly distributed alveolar spaces in healthy lung tissue (Fig. 6i, j). After infection, the alveolar area decreased from 79% to 40%, accompanied by alveolar septal thickening, inflammatory-cell infiltration, epithelial-cell detachment, and tissue disorganization. CP and CP@MTP partially alleviated these pathological changes, whereas CP@PTP better preserved alveolar architecture and markedly reduced inflammatory infiltration. Masson staining further showed that abnormal collagen deposition was lower in the CP@PTP group than in the control and non-targeted treatment groups. Elastic van Gieson (EVG) staining showed broken, disordered, and unevenly distributed elastic fibers in the control group, whereas elastic-fiber organization was clearly improved after CP@PTP treatment. Collectively, these results demonstrated that CP@PTP reduced pulmonary bacterial burden and alleviated P. aeruginosa pneumonia-induced inflammatory injury, alveolar destruction, collagen deposition, and elastic-fiber damage. In the end, the potential of CP@MTP/CP@PTP in combination with clinically used antibiotics was investigated. Checkerboard assays showed that the FIC index was below 0.5 for CP@MTP combined with vancomycin and CP@PTP combined with meropenem, indicating synergistic antibacterial activity (Fig. S21). Based on this result, the therapeutic effect of CP@PTP plus meropenem was further evaluated in the mouse P. aeruginosa pneumonia model. Compared with single treatments, the combination therapy achieved stronger bacterial clearance, reduced inflammation (Fig. S22), suggesting that CP@PTP could serve as an effective adjuvant to antibiotic therapy.

Conclusion

This study developed a M13 phage-mediated targeted platform for the rapid enriched delivery of Cu²⁺. MTP and PTP were identified to target MRSA and P. aeruginosa through M13 phage display, respectively. Both phages retained the typical filamentous morphology of M13, showed excellent amplifiability and selective adhesion to their corresponding bacterial targets within a short time. To introduce Cu²⁺ onto these recognition scaffolds, PCA was used as a polyphenolic bridge for phage-surface adhesion and Cu²⁺ coordination. PCA was first attached to MTP or PTP through interactions between polyphenolic hydroxyl groups and capsid-protein amino groups. Subsequent Cu²⁺ coordination produced CP NPs on the phage surface, yielding CP@MTP and CP@PTP. This mild metal-polyphenol assembly integrated bacterial recognition, copper-based antibacterial activity, and the one-dimensional M13 scaffold into a single targeted delivery system. CP@MTP and CP@PTP enriched CP NPs on their corresponding bacterial surfaces within 10 min in vitro, thereby reshaping the local concentration distribution of antibacterial agents. This targeted enrichment increased bacteria-associated copper accumulation and induced membrane permeabilization, K⁺ release, intracellular-content leakage, and bacterial lysis. Transcriptomic analysis confirmed that the antibacterial effect arose from coordinated disruption of metal handling, redox balance, envelope integrity, energy metabolism, and pathogenic regulation. CP@MTP and CP@PTP effectively cleared MRSA and P. aeruginosa infections in vivo, respectively, in zebrafish larval tail fins and reduced neutrophil recruitment. In a mouse P. aeruginosa pneumonia model, CP@PTP prolonged retention in infected lung tissue, reduced pulmonary bacterial burden and animal mortality, and alleviated inflammatory lung injury. Overall, this work showed that targeted M13 phages can convert metal-based antibacterial agents from passive diffusion to active targeting, providing a material design principle for rapid and precision therapy of drug-resistant bacterial infections.

Author contributions

F. L.: conceptualization, data curation, writing – original draft; Y.H.:methodology; H. C.: validation, methodology, data curation; Y. C.: validation; S. W.: methodology; J. J.: funding acquisition, writing – review & editing; Q. J.: project administration, funding acquisition, writing – review & editing.

Funding

The Pioneer and Leading Goose R&D Program of Zhejiang (2024C03080), the National Natural Science Foundation of China (52293381).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.