Metallothionein‐Inspired Dual‐Stage Ion‐Regulatory Coatings With Infection‐Triggered Bactericidal Activity and Long‐Term Antifouling Protection

ABSTRACT

The long‐term failure of implantable stents originates from static protection that cannot adapt to the temporal shift of primary risks from early‐stage infection and late‐stage biofilm–mineral occlusion. Inspired by the dynamic metal homeostasis of metallothioneins, we propose a β‐hydroxy thioether–based ion‐regulatory network that programs competing binding sites with distinct kinetic and thermodynamic preferences, thereby embedding an intrinsic functional clock into an otherwise static material. As a proof of concept, silver is employed as a representative ion to demonstrate how conventional metal bactericides can be transformed into stage‐adaptive defense factors. Initially, the system is dominated by labile hydroxyl complexation, triggering silver ion release under infection‐associated acidification and enabling rapid bactericidal clearance. Over time, the system shifts toward thermodynamically favored thioether coordination, which becomes dominant and reprograms the coating, resulting in bacterial repulsion and mineral exclusion at sub‐bactericidal ion levels. The stage‐adaptive protection is validated in methicillin‐resistant Staphylococcus aureus‐infected wound and long‐term (3‐month) bladder indwelling models. This work establishes time‐programmable ion regulation as an extensible design concept for adaptive biomaterials capable of coping with evolving biological environments.

A metallothionein‐inspired β‐hydroxy thioether network embeds a functional clock into a static interface through kinetic–thermodynamic coordination competition. Infection triggers transient ion release for rapid bacterial clearance, followed by thermodynamic anchoring that suppresses adhesion and mineralization. The interface thus evolves autonomously with time, illustrating an extensible design concept for adaptive biointerfaces.

1. Introduction

Ensuring the long‐term safety of implantable medical devices remains a critical challenge in modern medicine. Millions of patients worldwide rely on vascular stents, biliary drains, and urinary stents, which often remain in place for weeks, months, or even a lifetime [1, 2]. However, their durability is fundamentally limited by static protection strategies that cannot adapt to the time‐dependent shift in dominant biological risks, leading to repeated interventions, secondary surgeries, and even life‐threatening sepsis. Urinary stents exemplify this dilemma: acute bacterial infection dominates immediately after implantation, as surgical injury facilitates microbial invasion and local calcium dysregulation [3, 4, 5, 6, 7]. Over time, mineral‐driven biofilm maturation progressively becomes the primary failure mode during long‐term indwelling, driven by sustained urinary calcium exposure, crystal nucleation, and biofilm consolidation [4, 8, 9, 10, 11, 12]. Therefore, an ideal coating should not merely combine multiple functions, but rather autonomously switch its dominant protective role in synchrony with the evolving biological timeline, ensuring protection that is both potent in the early stage and persistent over long‐term implantation.

Current coating strategies remain fundamentally constrained by static designs that lack intrinsic temporal programmability, regardless of whether they rely on multifunctional combinations or single‐mechanism formulations [9, 13]. As a result, such systems cannot adapt to the stage‐dependent challenges encountered during implantation. In practice, antibacterial coatings provide early protection but suffer from cytotoxicity and resistance upon prolonged activity, without transitioning to a benign, anti‐adhesive state [14]. Conversely, surface‐modification—such as wettability tuning or anti‐encrustation designs—offer long‐term fouling resistance but fail to address acute infection risks at early stages [9, 15, 16, 17]. At a deeper level, this functional mismatch originates from the chemical logic of static surfaces, in which bactericidal motifs often accelerate calcification [18, 19], whereas mineralization‐resistant chemistries disrupt the hydration layer required for durable anti‐adhesion [20, 21]. Together, these limitations reflect a fundamental issue: static surface chemistries encode fixed functions rather than time‐evolving rules, thereby precluding autonomous functional transitions across the implantation timeline.

Metallothionein (MT) provides a compelling biomimetic model for time‐programmed functionality, as its behavior is governed by competition between kinetically and thermodynamically distinct metal–thiol interactions [22, 23, 24]. Its cysteine‐rich domains enable hierarchical metal regulation: essential ions such as Zn²⁺ and Cu²⁺ undergo reversible, kinetically controlled binding and stimulus‐responsive release, whereas toxic ions like Cd²⁺ and Hg²⁺ drive a thermodynamically favored, irreversible sequestration (Scheme 1A). Notably, MT shows negligible affinity for hard ions like Ca²⁺, effectively excluding them from this competitive landscape [22, 23, 25]. In essence, MT does not merely bind metals, but encodes a time‐dependent regulatory logic in which coordination kinetics dictate fast responses, while thermodynamic stability governs long‐term fate. This intrinsic strategy—sorting metals by binding kinetics and final stability—provides a transferable design principle for constructing synthetic systems capable of autonomous functional evolution over time.

SCHEME 1.

(A) Schematic illustration of a metallothionein (MT)‐inspired strategy for constructing stage‐adaptive ion‐regulated β‐hydroxy thioether (HT)–based networks. (B) Time‐programmed protective modes of PHT_S−PU stents after urinary implantation. The coating autonomously transitions from an early infection‐responsive bactericidal mode driven by Ag⁺ release (Mode I) to a long‐term passive resistance mode characterized by chemotaxis‐suppressed bacterial adhesion and ion‐regulated Ca²⁺ exclusion (Mode II), matching the shift in primary risk from acute infection to biofilm–mineral occlusion.

Building on this biomimetic principle, we designed a time‐programmed, stage‐adaptive ion‐regulation network using a β‐hydroxy thioether (HT) multi‐site binding module, in which a built‐in functional clock governs autonomous functional evolution. Through competition between kinetically labile hydroxy–silver and thermodynamically stable thioether–silver coordination, a static metal bactericide is converted into a self‐regulating defense system. During early implantation, infection‐induced acidification disrupts hydroxyl complexation, triggering on‐demand Ag⁺ release for rapid bactericidal action while suppressing infection‐driven calcification. Over time, the system shifts toward thermodynamically preferred thioether–silver coordination, dampening silver release and stabilizing sub‐bactericidal retention. This state maintains long‐term suppression of bacterial surface engagement and inhibits calcium nucleation, effecting a transition from an acute “kill‐and‐block” mode to a chronic “anti‐adhesion and antifouling” mode (Scheme 1B). Applied to polyurethane stents, this network provides programmable dual‐stage protection, resolving the longstanding conflict between early antibacterial efficacy and long‐term biocompatibility. More broadly, programming functional evolution through competitive binding kinetics offers a versatile route to adaptive biointerfaces.

2. Results and Discussion

2.1. Construction of Metallothionein‐Inspired Nanogels as a Multi‐Site Ion‐Regulatory Network

To construct a multi‐site platform for stabilizing adaptive‐binding metal ions, polymeric nanogels (PHT NGs) were synthesized via a thiol–epoxy “click” reaction between pentaerythritol tetra(3‐mercaptopropionate) (PETMP) and poly(ethylene glycol) diglycidyl ether (PEG‐DGE) under ambient conditions (Figure 1A). Catalyzed by DBU at a fixed thiol‐to‐epoxy molar ratio of 1:1 [26], this reaction intrinsically generates HT motifs, embedding hydroxyl and thioether groups as chemically distinct yet spatially coupled metal‐binding sites [27, 28]. Network dimensionality was controlled by the total reactant concentration. Above 30%, macroscopic gelation occurred, whereas dilution below this threshold yielded discrete NGs (Figure 1B), reflecting a transition from percolated networks to isolated coordination domains. Consistent with this structural confinement, diluted formulations (5%) produced NGs with narrow size distributions (a polydispersity index of ≈0.13) and uniform spherical morphology of ≈50 nm (Figure 1C,D and Figure S1), suitable for dynamic ion access and exchange. Attenuated total reflectance Fourier‐transform infrared (ATR‐FTIR) spectroscopy confirmed the thiol–epoxy reaction by loss of the thiol (2553 cm⁻¹) and epoxy (757, 843, 914, and 1256 cm⁻¹) [29, 30, 31]. Concurrently, a broad O–H stretching band appeared in the 3306–3580 cm⁻¹ region (Figure S2). X‐ray photoelectron spectroscopy (XPS) further verified the C–S bond formation with S 2p signals at 163.0 and 164.1 eV (Figure 1E and Figure S3) [26]. Together, these results establish a chemically well‐defined NG network enriched in HT motifs, providing a robust and programmable metal‐binding framework for subsequent ion‐regulation studies.

FIGURE 1.

Characterization and biological evaluation of PHT_S NGs and PHT_S‐PU stents: (A) Schematic illustration of the preparation of PHT_S NGs and the co‐depositing with PDA onto PU stents to form the PHT_S‐PU coating. (B) Hydrodynamic diameter and polydispersity index of PHT NGs at various total monomer concentrations, measured by DLS. (C) Zeta potential and size distribution of PHT and PHT_S NGs; (D) SEM and TEM images showing the morphology of PHT and PHT_S NGs. (E,F) High‐resolution XPS spectra of the S 2p (E) and Ag 3d (F) regions; (G) Hydrodynamic diameter and polydispersity index of PHT_S NGs solutions stored at 4°C for different time periods. (H) Photographs and SEM cross‐sectional images of pristine PU, PDA‐PU, and PHT_S‐PU stents. (I) Elemental mapping images and (J) XPS deconvolution spectra of the S 2p region for PHT_S‐PU. (K) Live/dead staining of 3T3 and HK‐2 cells after 24 h incubation with stent extract of PHT‐PU and PHT_S‐PU. (L) Immunofluorescence staining of γ‐H2AX (red) and DAPI (blue) in HK‐2 cells after exposure to PHT_S‐PU stents to evaluate DNA damage. Etoposide was used as the positive control. (M) Heatmap of serum biochemical markers (ALT, AST, BUN, CREA, LDH) in mice 24 h after intravenous injection of PHT and PHT_S NGs. PBS was used as the negative control. Data are presented as mean ± SD (n ≥ 3).

PHT NGs prepared at 5% concentration were used to fabricate silver‐loaded NGs (PHT_S NGs) by introducing AgNO₃ followed by NaBH₄ reduction (Figure 1A). The reduction produced a characteristic brown coloration, consistent with in situ formation of nanosilver. Compared with the parent PHT NGs, PHT_S NGs exhibited an increased hydrodynamic diameter (84 to 143 nm) while maintaining a narrow size distribution (Figure 1C). Transmission electron microscopy (TEM) further revealed ≈113 ± 5 nm dried NGs containing uniformly dispersed 5–10 nm nanoparticles embedded within the matrix, and a plasmon band at 410 nm confirmed the presence of Ag nanoparticles (Figure S4) [32]. Beyond physical entrapment, surface‐sensitive spectroscopy supports chemical stabilization of silver by the network. XPS showed the expected Ag 3d doublet (367.7/373.7 eV) and a shifted S 2p feature at 161.6 eV, consistent with thioether–silver coordination (Figure 1E,F) [33]. Together with a silver loading of ≈10.3 wt% (Figure S5), these results suggest that HT motifs provide multi‐site binding that anchors silver species while limiting particle coalescence. Accordingly, PHT_S NGs remained colloidally stable in PBS (pH 7.4) for at least 7 d, with negligible changes in size and dispersity (Figure 1G and Figure S6). This stabilization is achieved without sacrificing biosafety: murine fibroblasts (3T3) and human uroepithelial cells (HK‐2) maintained >90% viability even at 3000 µg mL⁻¹, and hemolysis remained <10% at the same dose (Figures S7 and S8). Collectively, these data indicate that the HT–enriched network functions as a chemically “buffered” coordination host, stabilizing metal species under neutral conditions, suppressing aggregation, and providing a robust structural basis for subsequent ion‐regulated, stage‐adaptive functionality.

2.2. Dopamine‐Assisted Nanogel‐to‐Coating Deposition for Programmable Stent Interfaces

We next evaluated PHT_S NGs as precursors for constructing functional coatings on polyurethane (PU) stents. Co‐deposition was achieved by immersing PU substrates in Tris buffer (pH 8.5) containing PHT_S NGs and dopamine (DA) at room temperature (Figure 1A). Compared with the dark brown appearance of conventional polydopamine (PDA) coatings, PHT_S‐modified surfaces exhibited a lighter yellow‐brown color, suggesting altered polymerization and assembly behavior (Figure 1H). Morphological analysis revealed that PDA alone formed a thin (≈61 nm) but rough and aggregated layer, whereas PHT_S deposition produced a compact and uniform coating with a thickness (≈124 nm) comparable to the NG diameter, consistent with near‐monolayer NG immobilization on the PU surface (Figure S9). Elemental mapping confirmed the colocalization of sulfur, nitrogen, and silver across the coating, indicating the successful transfer of the NG metal‐binding network from solution to the solid interface (Figure 1I). XPS further identified thioether and thioether–metal coordination signatures, together with minor SO_x species likely arising from oxidative interactions during DA polymerization (Figure 1J and Figure S10). The resulting PHT_S‐PU surfaces were hydrophilic, with a reduced water contact angle (≈51°) and an increased surface potential (≈+18 mV), reflecting the combined contributions of hydroxyl‐rich PEG segments and coordinated metal species (Figures S11 and S12). We attribute the improved coating uniformity to multivalent hydrogen‐bonding and coordination interactions between PHT_S NGs and catechol/amine groups of PDA, which moderate dopamine polymerization and suppress uncontrolled aggregation [27]. These results demonstrate that PHT_S NGs can be directly converted into robust and homogeneous surface layers, establishing DA‐assisted NG‐to‐coating deposition as an effective strategy for translating ion‐regulatory network architectures from bulk solution to implant‐relevant interfaces.

To assess the biocompatibility of PHT_S‐PU, we evaluated cytotoxicity, genotoxicity, hemocompatibility, and in vivo systemic safety. The coating supported >90% viability of murine fibroblasts (3T3) and human uroepithelial cells (HK‐2), with hemolysis remaining below 10%, indicating good cytocompatibility and blood compatibility (Figure 1K and Figure S13). Notably, the coating retained its morphology after 30 d in PBS (pH 7.4), confirming structural robustness under physiologically relevant conditions (Figure S14). Genotoxicity analysis using DAPI/γ‐H2AX staining revealed pronounced DNA damage only in the etoposide‐treated positive control, whereas PHT_S‐PU showed no detectable γ‐H2AX foci or nuclear abnormalities, comparable to PBS‐treated controls (Figure 1L). Consistently, short‐term implantation of PHT_S‐PU stents in healthy mice (3 d) did not perturb hepatic or renal function markers, blood cell counts, or leukocyte distributions, all of which remained within normal physiological ranges (Figure 1M and Figure S15). These results demonstrate that PHT_S‐PU exhibits favorable biocompatibility and hematological safety both in vitro and in vivo. Mechanistically, this safety profile arises from the chemically buffered metal‐binding environment of the HT‐network, which stabilizes metal species under neutral conditions, together with the DA‐assisted NG‐to‐coating architecture that ensures coating integrity during implantation—thereby minimizing cytotoxic and genotoxic risks.

2.3. Mode I: Infection‐Triggered Ion Release Enables Early‐Stage Active Antibacterial Clearance

Post‐surgical injury creates a transient window of high infection risk during early implantation [34, 35]. To evaluate the first 24 h antibacterial performance of PHT_S‐PU under clinically relevant complexity, we challenged the coatings with mono‐, dual‐, and polymicrobial cultures in TSB using extended‐spectrum β‐lactamase‐producing Escherichia coli (ESBL EC), methicillin‐resistant Staphylococcus aureus (MRSA), Pseudomonas aeruginosa (PA), and Proteus mirabilis (PM) (Figure S16). These mixed communities are intrinsically harder to eradicate due to biofilm‐assisted protection and interspecies cooperation [36]. As shown in Figure 2B and Figure S17, after 24 h co‐incubation, PHT_S‐PU eliminated >99% of planktonic bacteria in all mono‐species models, and still achieved >97% reduction in dual‐species (Gram‐negative + MRSA) and multi‐species (ESBL EC + PM + PA + MRSA) settings, indicating robust activity that extends beyond single‐pathogen conditions. Mechanistically, PHT_S‐PU elevated intracellular oxidative stress even at a sublethal dose (400 µg mL⁻¹), increasing reactive oxygen species (ROS) by 311–619% relative to the silver‐free control (Figure 2C). Consistent with ROS‐driven killing, scanning electron microscope (SEM) and live/dead staining revealed pronounced membrane deformation and loss of integrity in treated bacteria, whereas controls retained intact envelopes (Figure 2D and Figures S18 and S19). These data establish a rapid, broad‐spectrum bactericidal mode within the first 24 h, effectively eradicating both individual pathogens and polymicrobial communities in metabolically and structurally diverse infection settings.

FIGURE 2.

Mode I: Infection‐triggered ion release enables early‐stage active antibacterial clearance: (A) Schematically illustrates the Mode I antibacterial mechanism of the metallothionein‐inspired HT‐–network during early implantation. (B) Viable bacterial counts of a mixed culture of ESBL EC, MRSA, PA, and PM after treatment with PDA‐PU and PHTs‐PU at 37°C for 24 h. (C) Intracellular ROS levels in ESBL EC, MRSA, PA, and PM after 12 h of incubation with PHTs‐PU. (D) SEM images showing bacterial morphology after treatment with PHTs‐PU for 24 h. (E) Long‐term antibacterial activity of PU, PDA‐PU, and PHTs‐PU against MRSA after 30 d of continuous soaking in PBS. (F) Antibacterial performance against MRSA over five repeated treatment cycles using PU, PDA‐PU, and PHTs‐PU. (G) Photographs of inhibition zones against ESBL EC, MRSA, PA, and PM for PU, PDA‐PU, and PHTs‐PU discs in agar diffusion assays. (H) Cumulative Ag⁺ release profiles from PHTs‐PU under neutral (pH 7.4) and acidic (pH 5.5) conditions, with or without H₂O₂‐induced oxidative stress. (I) Cyclic Ag⁺ release behavior of PHTs‐PU under alternating pH environments (pH 5.5 and 7.4). (J) Time‐resolved XPS analysis of thioether–silver coordination evolution in PHT_S‐PU coatings, based on quantitative deconvolution of the S 2p spectra. (K) SEM images of PDA‐PU and PHTs‐PU surfaces after incubation with mixed bacterial strains ESBL‐EC, MRSA, PA, and PM in AU for 1 and 7 d at 37°C. Green indicates bacterial colonization; yellow represents mineral deposition. (L) ICP‐MS analysis of Ca²⁺ content on PDA‐PU and PHTs‐PU surfaces after incubation in bacteria‐laden AU for various durations. (M) Viable bacterial counts of ESBL‐EC, MRSA, PA, PM, or their mixture in AU after incubation with PHTs‐PU at 37°C for different time points. Arrows indicate survival rates below 0.1%. (N) pH variations of AU containing mixed bacterial strains in the presence of PDA‐PU and PHTs‐PU coatings, with bare PU as a control. Statistical significance is shown for comparisons between PDA‐PU and PHTs‐PU. Data are presented as mean ± SD (n ≥ 3). Statistical significance was evaluated using unpaired two‐tailed Student's t‐test. *p < 0.05, **p < 0.01, ***p < 0.001; n.s., not significant.

To determine whether this rapid bactericidal mode persists throughout the clinically relevant early implantation window (≈2 weeks) under continuous fluid exposure and recurrent contamination, we evaluated residual activity after prolonged soaking and under repeated challenge (Figure S20). After pre‐immersion in PBS (pH 7.4), PHT_S‐PU retained strong antibacterial efficacy without detectable coating deterioration, sustaining a 4–8‐log reduction against MRSA for 15 d and remaining active against ESBL EC for ≈20 d (Figure 2E and Figures S21–S23). In contrast, PDA‐PU and PHT‐PU controls showed negligible effects, underscoring the necessity of the silver‐bearing NG module. Beyond static pre‐soaking, we further tested robustness under repetitive high‐load exposure by repeatedly challenging the same samples with ESBL EC or MRSA (10⁶ CFU mL⁻¹, 24 h per cycle under alternating infection‐relevant and neutral conditions) (Figure S24). PHT_S‐PU maintained ≈7‐log killing over five consecutive cycles, indicating that antibacterial function is not exhausted by a single burst event but can be repeatedly mobilized within the acute implantation period (Figure 2F and Figures S25–S27). Collectively, these results position PHT_S‐PU as a durable early‐stage antibacterial defense that withstands both prolonged physiological conditioning and repeated infection insults.

To mechanistically reconcile strong early killing with biosafety, we examined whether PHT_S‐PU acts through continuous diffusive silver leakage or a regulated release process. Zone‐of‐inhibition assays produced only minimal halos (<0.5 mm) against ESBL EC, MRSA, PA, and PM, in sharp contrast to conventional silver coatings and inconsistent with sustained diffusive release (Figure 2G and Figure S28) [37]. Instead, inductively coupled plasma mass spectrometry (ICP‐MS) revealed a pronounced pH‐gated behavior (Figure 2H): under neutral conditions (pH 7.4), cumulative release over 24 h was only 0.18 µg cm⁻², whereas mild acidification (pH 5.5) increased release to ≈2.0 µg cm⁻² (>11‐fold). Under a matched cycling scheme with comparable duration and cycle number to the antibacterial challenge (Figure 2F), repeated pH switching reproducibly reactivated release, yet the released amount progressively decayed and became minimal after the fifth cycle, mirroring the antibacterial reuse profile (Figure 2I). This correspondence indicates that release behavior constitutes the primary response to environmental triggering, while the system remains intrinsically self‐limiting through multi‐site complexation encoded by the HT‐network. At neutral pH, cooperative hydroxyl complexation and thioether coordination restrain silver mobility and suppress uncontrolled ion leakage. Under acidic microenvironments associated with infection, protonation weakens hydroxyl‐mediated complexation, facilitating oxidative dissolution of nanosilver and triggering on‐demand Ag⁺ release (Figure 2A). With repeated triggering, residual silver becomes increasingly dominated by thermodynamically stable thioether anchoring, while passivation further suppresses oxidation, naturally tapering release over time. This progressive stabilization is corroborated by time‐resolved XPS analysis, which reveals a time‐dependent increase in the thioether–silver coordination fraction of PHT_S‐PU coatings during surface residence, increasing from ≈10% at day 1 to ≈20% by day 7, followed by only a minor change up to day 15, indicative of a transition toward a stabilized coordination state (Figure 2J and Figure S29). This structural evolution provides a direct basis for the observed attenuation in ion release and antibacterial activity over repeated cycles. This preference is consistent with the higher polarizability and lower electronegativity of thioether sulfur, which facilitate more effective orbital overlap and impart partial covalent character to thioether–silver interactions, rendering them thermodynamically favored over hydroxyl–silver complexation. Together, PHT_S‐PU implements a defense mode that is responsive yet self‐limiting, where infection‐triggered ion release dominates the early‐stage response, while subsequent stabilization progressively attenuates this activity.

2.4. Mode I: Suppression of Infection‐Driven Mineral Nucleation during Acute Defense

Although macroscale encrustation typically develops over prolonged implantation, bacterial infection can trigger nucleation initiation within hours to days by concentrating Ca²⁺ at negatively charged membranes/extracellular polymeric substances (EPS) and creating localized supersaturation [38, 39]. To model this acute, infection‐driven risk, PDA‐PU and PHT_S‐PU were incubated in artificial urine (AU, 37°C) for 7 d with ESBL EC, MRSA, PA, PM, or a polymicrobial consortium (Figure S30). Under mixed infection, PDA‐PU rapidly accumulated dense crystalline deposits by SEM—hallmarks of infection‐accelerated scaling—whereas PHT_S‐PU remained largely free of crystalline or membrane‐like structures (Figure 2K). This morphological difference was supported quantitatively: calcium deposition on PDA‐PU increased over time to ≈30 µg cm⁻² by day 7, while PHT_S‐PU consistently stayed below 1.5 µg cm⁻², indicating that the coating suppresses early nucleation rather than merely delaying late‐stage buildup (Figure 2L). Importantly, the observed suppression of mineral nucleation originates from infection control rather than an independent interfacial process. PHT_S‐PU nearly eradicated bacteria in single‐strain cultures and still reduced survival to ≈21% under polymicrobial conditions, yet maintained low calcium deposition even in the most challenging mixed community (Figure 2M and Figures S30 and S31). Mechanistically, AU containing PDA‐PU became strongly alkaline (≈pH 9.2 within 15 h), consistent with urease‐positive P. mirabilis–driven ammonia production that promotes calcium phosphate precipitation (Figure 2N). In contrast, PHT_S‐PU preserved a near‐neutral milieu, suggesting that rapid antibacterial clearance suppresses urease‐mediated alkalinization and thereby removes the chemical driving force for nucleation. These results indicate that during acute defense, PHT_S‐PU intercepts infection‐driven encrustation at its earliest step by controlling bacterial activity, thereby blocking the microbial pH shift and subsequent Ca²⁺ nucleation cascade. This process preserves stent cleanliness and patency in the critical early post‐implantation period.

2.5. Mode II: Thermodynamically Anchored Silver Drives Persistent Suppression of Bacterial Adhesion

Although stent coatings may initially kill bacteria, even sparse early adhesion can rapidly trigger EPS secretion and biofilm maturation, stabilizing polymicrobial communities and diminishing bactericidal efficacy [11, 40, 41]. Long‐term protection therefore depends less on sustained killing than on preventing adhesion once the acute infection challenge has passed. To interrogate this late‐stage requirement under conditions where infection‐triggered release is largely exhausted, we tested PHT_S‐PU after five consecutive antibacterial challenges, when thermodynamic anchoring dominates. The coatings were then exposed to an aggressive 30 d colonization regimen in TSB with exponentially growing ESBL EC or MRSA (10⁹ CFU mL⁻¹, refreshed every 48 h; Figure 3A). SEM and confocal laser scanning microscopy (CLSM) showed rapid biofilm establishment on PU and PDA‐PU, culminating in near‐complete surface coverage by day 30, while PHT‐PU displayed only transient delay and was ultimately colonized (Figure 3B–D). In sharp contrast, PHT_S‐PU remained essentially free of detectable biomass throughout the 30 d (“zero colonization”), despite cumulative Ag⁺ release of only ≈200 ng cm⁻²—far below bactericidal levels. These results indicate a distinct Mode II behavior: thermodynamically stabilized silver establishes a low‐leaching interfacial state that persistently suppresses bacterial adhesion, thereby preventing adhesion‐driven biofilm initiation under repeated high‐burden exposure.

FIGURE 3.

Mode II: Thermodynamically anchored silver enables persistent suppression of bacterial adhesion at the long‐term stage: (A) Schematic illustration of the long‐term bacterial adhesion assay, simulating post‐acute implantation conditions in which infection pressure subsides while biofilm risk progressively dominates under repeated bacterial exposure. (B, C) Fluorescent live/dead staining and SEM images of PU, PDA‐PU, PHT‐PU, and PHT_S‐PU stents after incubation with 1 × 10⁹ CFU mL⁻¹ (B) MRSA and (C) ESBL EC for 1, 15, and 30 d. (D) Quantification of the fluorescent staining area on PU‐based stent surfaces after incubation with MRSA and ESBL EC for different durations. Arrows indicate coverage below 0.1%. (E) BSA adsorption on PU‐based surfaces evaluated by BCA assay. (F) Schematic illustration of the Mode II anti‐biofilm mechanism. Data are presented as mean ± SD (n ≥ 3). One‐way ANOVA with Tukey's post hoc test was used to evaluate statistical significance. For panel D, significance is shown for comparisons at 30 d. *p < 0.05, **p < 0.01, ***p < 0.001; n.s., not significant.

To probe the basis of late‐stage suppression of bacterial adhesion, we first examined resistance to protein adsorption, a prerequisite for initial surface conditioning (Figure 3E). Using bovine serum albumin (BSA) as a representative protein, conventional PU and PDA‐PU accumulated substantial protein layers (5 to 15 µg cm⁻²), whereas PHT‐PU and PHT_S‐PU suppressed adsorption to near‐background levels (<1 µg cm⁻²). This behavior is consistent with a highly hydrated interface arising from dense hydroxyl and PEG segments, which energetically disfavors nonspecific protein attachment. This coordination‐stabilized, low‐release state maintains a weakly perturbed ionic environment and a stable hydration layer, together limiting protein adsorption and reducing the probability of bacterial approach and surface engagement. By minimizing formation of a protein conditioning layer, the PHT_S‐PU surface reduces the physicochemical cues required for bacterial anchoring, thereby delaying adhesion and biofilm initiation under late‐stage conditions where silver species are thermodynamically immobilized rather than bactericidally active (Figure 3F).

To further resolve the origin of late‐stage suppression of bacterial adhesion, we analyzed bacterial behavior at both single‐cell and collective levels using PHT_S‐PU samples preconditioned through five antibacterial release cycles. Single‐cell tracking by CLSM revealed that on PU and PDA‐PU surfaces, E. coli displayed persistent, directional swimming with stable velocities (3–5 µm s⁻¹), accompanied by frequent turning events (>500 deg s⁻¹, 10–13 events per cell)—features consistent with active chemotactic behavior and surface exploration (Figure 4A–C and Table 1). On PHT‐PU, trajectories became more extended (mean squared displacement reaching ≈2500 µm² after 30 s) without significant changes in speed or turning dynamics, suggesting prolonged surface residence driven by hydration rather than active avoidance (Figure 4D). In contrast, PHT_S‐PU fundamentally altered bacterial motility. Cells exhibited confined, non‐directional displacements with markedly reduced speed and strongly dampened turning activity, indicative of disrupted chemotactic signaling and ineffective surface sensing. This jitter‐like, frequently reversing motion indicates disrupted motility and loss of directional persistence, preventing sustained surface exploration and stable attachment. Importantly, this single‐cell impairment translated to the population level: while PU, PDA‐PU, and PHT‐PU supported robust swarming and translocation of P. mirabilis, PHT_S‐PU effectively blocked collective migration across agar bridges (Figure 4E,F). Together, these observations indicate that the coordination‐stabilized interfacial state does not act through bactericidal stress, but instead limits bacterial motility and navigation required for stable surface engagement. This effect manifests as impaired directional persistence at the single‐cell level and reduced collective migration, thereby sustaining late‐stage suppression of bacterial adhesion even at ion concentrations far below bactericidal thresholds.

FIGURE 4.

Disrupted bacterial motility and chemotactic signaling under Mode II adhesion suppression: (A) Representative motility trajectories of ESBL EC on PU, PDA‐PU, PHT‐PU, and PHT_S‐PU surfaces. (B) Time‐dependent changes in linear speed. (C) Rate of change in direction index (RCDI). (D) Mean squared displacement (MSD) over time. Horizontal dashed lines indicate thresholds for motility acceleration (>5 µm s⁻¹) and directional turning events (>500 deg s⁻¹). (E) Representative images from bridge‐based swarming assays showing swarming inhibition of PM on PU‐based surfaces. (F) Quantification of swarming zone areas after 24 h incubation at 37°C. (G) Gene Ontology (GO) enrichment analysis of ESBL EC after 24 h exposure to PHT_S‐PU. (H) Heatmaps of differentially expressed genes associated with three KEGG functional modules: flagellar assembly, bacterial chemotaxis, and biofilm formation. Untreated ESBL EC served as the control. (I) Schematic of Mode II adhesion suppression, in which thermodynamically anchored silver disrupts chemotactic signaling and motility rather than inducing bactericidal stress. Data are presented as mean ± SD (n ≥ 3).

TABLE 1.

Quantitative analysis of ESBL EC mobility on different stent surfaces.

Samples	Acceleration events a	Turning events b
PU	3 ± 1.7	10 ± 1.9
PDA‐PU	6 ± 2.1	13 ± 3.6
PHT‐PU	7 ± 3.0	11 ± 3.2
PHTS‐PU	0	2 ± 1.4

^a Number of motility runs exhibiting a peak velocity greater than 5 µm s⁻¹.

^b Number of directional turning events with a rate of change exceeding 500 deg s⁻¹.

To elucidate the molecular basis of late‐stage anti‐adhesion, we performed transcriptomic analysis (RNA‐seq) on ESBL EC after 24 h exposure to PHT_S‐PU preconditioned through five release cycles, mimicking long‐term use. Gene Ontology (GO) enrichment revealed a focused suppression of processes central to cellular growth and information flow—most prominently translation and core metabolic functions—pointing to a global downshift in biosynthetic capacity rather than acute lethality (Figure 4G). The enrichment of ribosome‐related terms further implies disruption of intracellular assemblies that support rapid protein turnover, a prerequisite for motility and surface adaptation. Consistent with this systems‐level attenuation, KEGG analysis showed coordinated downregulation across three adhesion‐critical modules: flagellar assembly, chemotaxis, and biofilm formation (Figure 4H). Genes associated with flagellar assembly (e.g., flgE, fliC, and motA) were consistently downregulated, indicating impaired formation of the filament, hook, and motor apparatus, in line with the reduced swimming speed observed in single‐cell tracking. Suppression of key chemotaxis regulators (e.g., cheA, cheY, tsr, tar, and motB) further disrupted signal transduction, corresponding to decreased turning frequency and constrained directional motility. In parallel, downregulation of biofilm‐related genes (e.g., csgA/csgC, wcaJ, rcsC/rcsD, gspE–gspK, luxS, flu, and bcsA) impaired processes from initial adhesion to matrix production, thereby limiting surface colonization and collective migration. Together with the observed defects in single‐cell navigation and collective migration, these transcriptomic signatures point to coordinated pathway‐level alterations rather than stress‐induced killing (Figure 4I). Thermodynamically anchored silver thus enforces a low‐ion, non‐lethal state that selectively disables the molecular machinery required for adhesion, chemotaxis, and biofilm development, providing a mechanistic basis for sustained suppression of bacterial adhesion under sub‐bactericidal conditions.

2.6. Mode II: Preferential Ion Coordination Enables Long‐Term Resistance to Biofilm–Mineral Occlusion

During long‐term indwelling, urinary stents experience continuous exposure to calcium‐rich flow, where sustained ion supply and shear collectively undermine antifouling surfaces by destabilizing hydration layers and promoting crystal nucleation [40, 42]. To interrogate anti‐encrustation performance under these clinically relevant conditions, PHT_S‐PU coatings were first preconditioned through five antibacterial challenge cycles and then evaluated in sterile, dynamic crystallization assays that simulate prolonged urinary flow. Three urine models were used: AU, oxalate‐enriched urine (AU + 0.52 mm Na₂C₂O₄), and calcium‐enriched urine (AU + 2.20 mm CaCl₂). PU stents with different coatings were continuously perfused with AU at 37°C for 30 d using a peristaltic pump, simulating long‐term flow to evaluate resistance to calcium deposition (Figure 5A). Under continuous perfusion, conventional PU, PDA‐PU, and PHT‐PU rapidly accumulated crystalline deposits, with visible scaling emerging within 7 d and progressing to near‐complete surface coverage and lumen obstruction by day 30 (Figure 5B and Figure S32). In contrast, PHT_S‐PU exhibited markedly delayed and suppressed crystallization, remaining largely deposit‐free for 7–15 d even under oxalate‐ or calcium‐enriched conditions, and maintaining lumen patency at time points when other coatings failed (Figure 5C,D). Quantitative analysis corroborated this behavior: surface‐bound calcium on control coatings exceeded 30 µg cm⁻², whereas PHT_S‐PU consistently limited calcium accumulation to below 6 µg cm⁻² after 30 d (Figure 5E). Elemental mapping further confirmed that deposits on control surfaces were calcium phosphate/oxalate in nature, while PHT_S‐PU showed only background calcium signals (Figure 5F and Figure S33). These results indicate that under long‐term dynamic flow, where hydration‐based antifouling alone becomes insufficient, PHT_S‐PU sustains resistance to mineral encrustation through suppression of calcium nucleation and growth governed by the underlying coordination state. This behavior preserves stent patency during prolonged indwelling.

FIGURE 5.

Mode II: Preferential ion coordination enables long‐term resistance to biofilm–mineral occlusion: (A) Schematic of a dynamic encrustation model simulating urine flow through PU‐based stents, driven by a peristaltic pump and using AU. (B) SEM images of PU‐based stent surfaces after exposure to flowing standard AU, oxalate‐enriched AU (AU + 0.52 mm Na₂C₂O₄), or calcium‐enriched AU (AU + 2.20 mm CaCl₂) for 7 and 30 d. (C) SEM images of intraluminal encrustation on PU‐based stents after 30 d of continuous AU flow. (D) Time to complete lumen occlusion for PU‐based stents under dynamic AU flow. (E) ICP‐MS analysis of Ca²⁺ deposition on PU‐based stent surfaces after treatment with standard, oxalate‐enriched, and calcium‐enriched AU for different durations. (F) SEM and EDS elemental mapping of the encrusted layer on PDA‐PU stents, showing spatial distribution of C, N, O, P, Ca, and Mg. (G) ¹H NMR spectra of 2‐hydroxyethyl methyl sulfide in the absence or presence of equimolar AgNO₃ or Cu(NO₃)₂. (H) ITC binding isotherms and (I) corresponding enthalpy changes for the interactions between PHT or PHT_S NGs and various ligands in aqueous solution. (J) Real‐time adsorption profiles of Ag⁺ and Cu²⁺ on PHT‐PU monitored by QCM via frequency shifts. (K) Proposed Mode II mechanism, in which thermodynamically anchored silver establishes a coordination‐selective network that excludes calcium from nucleation sites, thereby sustaining long‐term resistance to mineral occlusion under continuous flow. Data are presented as mean ± SD (n ≥ 3). Statistical significance was evaluated using one‐way ANOVA with Tukey's post hoc test. *p < 0.05, **p < 0.01, ***p < 0.001; n.s., not significant.

To elucidate the governing molecular origin of long‐term anti‐encrustation, we examined the ion‐selective coordination behavior of HT motifs toward calcium. Using 2‐hydroxyethyl methyl sulfide as a structural analog, proton nuclear magnetic resonance spectroscopy (¹H NMR) showed pronounced chemical shifts upon addition of Ag⁺, whereas Ca²⁺ induced negligible perturbation (Figure 5G), indicating intrinsically weak interaction between calcium and the thioether framework. This selectivity was quantified by isothermal titration calorimetry (ITC): Ag⁺ binding to PHT NGs was strongly exothermic (ΔH ≈ −75 kJ mol⁻¹), while Ca²⁺ produced little measurable heat (Figure 5H,I), establishing a clear affinity hierarchy (Ag⁺ ≫ Ca²⁺). Importantly, this hierarchy translated into competitive exclusion. When silver was pre‐anchored within the network (PHT_S NGs), subsequent Ca²⁺ titration generated no additional binding signal, indicating that silver occupies the dominant coordination sites and blocks calcium association. An analogous effect was observed at the interface level using quartz crystal microbalance (QCM): although the PHT coating could initially adsorb both ions, silver saturation eliminated subsequent calcium uptake (Figure 5J). These results demonstrate that HT motifs encode a preferential coordination rule that governs ion selection under mode II conditions, whereby thermodynamically anchored silver precludes calcium binding at potential nucleation sites. This silver‐directed selectivity provides a molecular basis for sustained resistance to mineral encrustation under long‐term, calcium‐rich flow conditions.

Taken together, these findings identify preferential coordination as the governing rule of Mode II interfacial state. Once infection‐responsive release subsides, thermodynamically anchored silver fixes the interfacial coordination landscape, transforming the coating from a reactive antibacterial surface into a stable, ion‐selective barrier. In this regime, calcium exclusion is not achieved through dynamic release or surface hydration, but through a locked coordination state that resists ion exchange under continuous calcium exposure. Importantly, this thermodynamic endpoint complements the earlier kinetically driven Mode I response, enabling a seamless transition from active defense to passive persistence across the implantation timeline (Figure 5K). More broadly, the HT framework illustrates a transferable design principle: by encoding coordination hierarchy rather than specific functionality, adaptive biointerfaces can be programmed to evolve from responsive to stable states. Such a rule‐based strategy is not limited to silver, but could be extended to other biologically relevant soft metal ions to address diverse long‐term challenges in implantable systems.

2.7. In Vivo Validation of Dual‐Stage Functional Conversion across Infection and Long‐Term Implantation Models

To validate the proposed dual‐stage functional conversion in vivo, we deliberately avoided a single integrated preclinical model. Simultaneously imposing an acute, high‐burden infection and a 3‐month indwelling implant in small animals introduces excessive physiological stress, increases mortality, and obscures stage‐specific outcomes [43, 44]. Instead, we adopted two complementary, time‐resolved models aligned with the distinct biological demands of Mode I and Mode II. An MRSA‐infected full‐thickness murine skin wound—prone to rapid biofilm formation—was used to interrogate early‐stage antibacterial efficacy and its impact on inflammation and tissue repair (Figure 6). In parallel, a 3‐month rat bladder indwelling stent model was employed to assess late‐stage suppression of bacterial adhesion, resistance to mineral encrustation, and systemic/local biocompatibility under continuous urinary flow (Figure 7). This paired‐model strategy enables a clean, mechanistically faithful evaluation of stage‐adaptive performance across the implantation timeline.

FIGURE 6.

In vivo validation of early‐stage (Mode I) infection control using PHT_S‐PU dressing: (A) Schematic illustration of the preparation of PHT_S‐PU wound dressings. (B) Timeline of the MRSA‐infected wound model establishment and the corresponding treatment schedule. (C) Representative images of wound healing progression and (D) quantification of relative wound area over time following treatment with PDA‐PU and PHT_S‐PU dressings; untreated PU dressing served as the control. (E) Body weight changes of mice during the treatment period. (F) Photographs and (G) quantification of bacterial CFUs in wound tissue after different treatments at indicated time points. Representative histological and immunological assessments of infected wounds on days 3, 7, and 11, including: (H) Hematoxylin and eosin (H&E) staining for tissue morphology; (I) Immunohistochemical staining for pro‐inflammatory cytokines IL‐6 and TNF‐α; and (J) Quantification of IL‐6 and TNF‐α positive areas. Data are presented as mean ± SD (n ≥ 3). Unpaired two‐tailed Student's t‐test was used to evaluate statistical significance, which is shown for comparisons between control and PHTs‐PU. *p < 0.05, **p < 0.01, ***p < 0.001; n.s., not significant.

FIGURE 7.

In vivo validation of long‐term (Mode II) suppression of bacterial adhesion and anti‐encrustation in a rat bladder indwelling model: (A) Timeline of the long‐term bladder implantation procedure. (B) Surgical implantation of PU and PHT_S‐PU stents into the rat bladder. (C) Photographs and (D) quantification of bacterial CFUs in urine and on explanted stents at days 15 and 90. (E) SEM images and (F) quantification of Ca²⁺ and Mg²⁺ deposition on PU and PHT_S‐PU stent surfaces after 15, 30, and 90 d of implantation. (G) Ag⁺ concentrations in rat blood and urine over time post‐implantation. (H,I) Serum biochemical markers of liver and kidney function in rats implanted with PU and PHT_S‐PU stents at the indicated time points. (J) Representative H&E staining images of liver and kidney tissues after 15 and 90 d of implantation. (K) Representative H&E, CD3, and CD68 staining images of bladder tissues, and (L) Quantification of CD3‐positive and CD68‐positive areas at days 15 and 90. Data are presented as mean ± SD (n ≥ 3). Unpaired two‐tailed Student's t‐test was used to evaluate statistical significance, which is shown for comparisons between control and PHTs‐PU. *p < 0.05, **p < 0.01, ***p < 0.001; n.s., not significant.

We assessed early‐stage anti‐infective efficacy using a murine full‐thickness skin wound model infected with MRSA, a setting characterized by rapid biofilm establishment. PHT_S NGs were co‐deposited with DA onto porous PU to fabricate wound dressings, which were applied after infection establishment (Figure 6A,B). Relative to uncoated PU and PDA‐PU controls, PHT_S NGs dressings markedly accelerated wound closure, achieving ≈91% healing by day 7 without affecting body weight (Figure 6C,D). Body weights remained consistent across all groups (Figure 6E). This functional improvement correlated with a >7‐log reduction in MRSA burden, whereas controls retained high bacterial loads consistent with biofilm persistence (Figure 6F,G). Tissue analyses further linked bacterial control to a favorable healing milieu: PHT_S‐PU–treated wounds showed reduced inflammatory infiltration and enhanced collagen deposition, alongside pronounced suppression of pro‐inflammatory cytokines IL‐6 and TNF‐α compared with controls (Figure 6H–J). Together, these outcomes indicate that during the acute phase, PHT_S‐PU not only suppresses infection but also reshapes the local inflammatory environment to support tissue regeneration—validating Mode I performance in vivo.

To evaluate late‐stage performance in a clinically relevant indwelling setting, PHT_S‐coated PU stents were implanted into rat bladders for up to 3 months, with uncoated commercial PU stents as controls (Figure 7A,B and Figure S34). By day 15, controls developed clear implant‐associated infection, with urine bacterial burdens exceeding 5 log CFU mL⁻¹ and pronounced surface colonization accompanied by early mineral deposits (Figure 7C–E). In contrast, PHT_S‐PU stents showed no overt urinary infection and markedly reduced bacterial adhesion and crystallization. This divergence widened over time: after 3 months, control stents were heavily encrusted, whereas PHT_S‐PU exhibited only sparse microcrystals and remained largely deposit‐free. Quantitatively, Ca and Mg accumulation was reduced by an order of magnitude or more (Ca: 62 to 2 µg cm⁻², Mg: 18 to 0.3 µg cm⁻², Figure 7F), supporting durable suppression of encrustation under continuous urinary flow. Importantly, this long‐term protection was achieved without systemic metal exposure or organ toxicity. Silver remained below the detection limit in both blood and urine (<0.01 ppm), serum liver/kidney markers stayed within normal ranges, and histology confirmed intact liver and kidney architecture (Figure 7G–J). Locally, bladder tissues showed only mild inflammatory changes, with minimal CD3⁺ T‐cell and CD68⁺ macrophage signals (<0.1% area; Figure 7K,L). These in vivo data establish that, following early‐stage antibacterial action, the coating transitions into a low‐leaching, persistent state that suppresses bacterial adhesion and resists mineral encrustation during long‐term implantation while preserving systemic and local biocompatibility.

3. Conclusion

Inspired by the selective metal‐ion homeostasis of metallothioneins, we developed a β‐hydroxy thioether–based multi‐site binding network that converts static metal bactericides into time‐programmed, stage‐adaptive defense factors. Using silver as a representative ion, this system embeds an intrinsic functional clock into an otherwise static coating through competition between kinetically labile hydroxyl complexation and thermodynamically stable thioether anchoring. During early implantation, infection‐associated acidification weakens hydroxyl‐mediated complexation, triggering on‐demand silver release that enables rapid bactericidal clearance, thereby suppressing infection‐driven mineral nucleation. As infection pressure subsides, strong thioether–silver coordination and interfacial passivation establish a thermodynamically locked, low‐leaching state, which resists calcium accumulation and prevents biofilm initiation. At sub‐bactericidal levels, immobilized silver further sustains long‐term suppression of bacterial adhesion by disrupting chemotactic signaling and collective migration rather than relying on continued killing. Applied to PU stents, this dual‐stage functional conversion was validated across complementary in vivo models, including an MRSA‐infected murine skin wound and a 3‐mon rat bladder indwelling model, demonstrating effective early infection control, durable resistance to encrustation, and favorable systemic and local biocompatibility. Beyond silver, the underlying ion‐regulation principle offers an extensible design concept for programming adaptive biointerfaces using other biologically relevant soft metal ions, potentially providing a versatile blueprint for long‐acting, biocompatible coatings capable of coping with evolving biological environments.

4. Experimental Section

The detailed materials and methods were described in the Supporting Information.

Funding

This work was supported by the National Natural Science Foundation of China (U25A20259, 52473132), the National Key R&D Program of China (2023YFB3812402).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: adma73872‐sup‐0001‐SuppMat.pdf.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.