Visible-light-responsive supramolecular enzyme mimics for combating antimicrobial resistance

Abstract

The global rise of antimicrobial resistance has created an urgent need for alternative strategies that can overcome the limitations of conventional drug-resistance mechanisms. Light-assisted antimicrobial therapy, particularly photodynamic inactivation, offers spatial and temporal control while minimizing the likelihood of resistance. Since biomedical applications demand visible-light-responsive catalysts that enable deeper tissue penetration and minimize photodamage, designing systems that avoid reliance on UV excitation is essential. This work reports a supramolecular oxidase mimic “suprazyme” based on benzohydrazide aggregation-induced emission luminogens (AIEgens) with systematic π-extension that act as efficient metal-free photo-responsive antimicrobial agents. By extending the aromatic core from benzene (G₀Ben) to naphthalene (G₀Nap) and anthracene (G₀Ant), we progressively narrowed the band gap (3.1 eV to 2.0 eV), red-shifting the absorption into the visible region and enhancing the reactive oxygen species generation under white light. Among the series, G₀Ant exhibited the most robust oxidase-like activity, efficiently producing superoxide radicals without requiring exogenous H₂O₂. The assemblies displayed excellent stability against variations in ionic strength, pH, and temperature, outperforming natural oxidase enzymes such as laccase. Critically, G₀Ant demonstrated potent light-activated antibacterial efficacy against both Gram-positive methicillin resistant Staphylococcus aureus (MRSA) and Gram-negative (Escherichia coli) strains, causing severe membrane disruption while showing minimal dark toxicity and good biocompatibility toward mammalian cells. These findings establish the π-extension of benzohydrazide-based AIEgens as a rational design principle to engineer visible-light-responsive suprazymes for safe, sustainable, and effective antimicrobial therapy.

---

Introduction

Rising antimicrobial-resistant bacteria have become one of the most urgent healthcare challenges of this century, raising concerns regarding infection management and global public health.^1,2 Conventional antibiotics are losing their efficacy at an increasing rate, demanding the need for innovative antimicrobial strategies that can bypass traditional resistance mechanisms.^3,4 Light-assisted antimicrobial therapy, particularly photodynamic inactivation (PDI), has emerged as a promising alternative owing to its ability to generate reactive oxygen species (ROS) upon light activation.^5–7 These ROS, including superoxide and singlet oxygen, can disrupt cellular membranes, nucleic acids, and proteins in a multi-targeted manner, greatly reducing the chances of resistance development. Compared to antibiotic-based approaches, PDI offers spatial and temporal control, minimal invasiveness, and broad-spectrum activity, making it a powerful tool for next-generation antimicrobial therapeutics.⁶

Artificial enzyme mimics using supramolecular materials (suprazymes) have recently gained significant attention as functional materials in catalysis, therapeutics, diagnostics, and environmental remediation due to their ease of synthesis, robustness, and cost-effectiveness compared to native enzymes.^8–16 Among these, oxidase-mimicking systems are particularly attractive since they utilize molecular oxygen as a minimal substrate to generate ROS without requiring added peroxides.¹⁷ Unlike peroxidase mimics that rely on toxic H₂O₂ species, oxidase mimics present compatible catalytic routes.¹⁸ Recent research advances have shown the intersection of photo-responsive supramolecular materials with enzyme mimicking abilities, resulting in light-activated suprazymes capable of generating ROS.^19–21 These photo-responsive suprazymes present a particularly promising approach for combating drug-resistant bacteria through PDI. Inorganic semiconductors such as TiO₂ and ZnO have been investigated as photo-activated agents owing to their stability and photocatalytic properties.^22,23 However, their reliance on UV light, limited tunability and potential cytotoxicity restricts their biomedical translation. A critical factor for the real-world applicability of materials for PDI is the choice of excitation wavelength, while many reported systems require UV irradiation, visible-light-responsive catalysts are far more suitable for biomedical applications such as antimicrobial therapy and wound healing, where deeper tissue penetration and minimal photodamage are essential.²⁴ Moreover, visible light provides an abundant source for therapeutic activation.^25,26 Tremendous efforts have been made to narrow the large band gap of TiO₂ through modifications, including surface deposition of metals like Ag, Au, and Pd, as well as doping with other metal ions such as Ce, Fe, Cu, Ti, Mo and Rh, thus increasing its visible light photocatalytic activity.^27–30

Organic chromophores generally possess higher molar absorption coefficients than inorganic semiconductors and can be chemically tailored for bandgap tuning, supramolecular assembly, and enhanced biocompatibility.³¹ In our previous study, we reported a metal-free benzohydrazide-based suprazyme exhibiting efficient photo-responsive oxidase-like activity.¹³ We demonstrated that the molecular self-assembly into higher-order assembled structures can create robust catalytic scaffolds with improved aggregation-induced emission (AIE) properties, higher photostability and enhanced ROS generation. Thus, a key challenge lies in understanding and manipulating the structure–function relationship of organic assemblies to further tailor their optical properties for visible-light activation. This understanding will enable the development of efficient therapeutics capable of eradicating drug-resistant bacterial strains under white light, offering high efficacy with minimal side effects. In this work, we demonstrate that the systematic π-extension of benzohydrazide-based AIEgens allows for the precise tuning of their electronic structures, effectively narrowing their band gaps, thereby red-shifting their photo-responsiveness into the visible region. This strategy of structural engineering not only enhances the ROS generation efficiency under visible light, but also broadens the practical scope of metal-free suprazymes as safe, sustainable, and effective platforms for light-assisted antimicrobial therapy against drug-resistant bacteria. We demonstrate that modification of the benzohydrazide core from benzene (G₀Ben) to naphthalene (G₀Nap) to anthracene (G₀Ant) progressively extends absorption into the visible region, thereby markedly enhancing the catalytic efficiency under visible light photoirradiation. The molecules spontaneously form stable colloidal assemblies through π–π stacking and hydrogen-bonding interactions in the presence of water, which serve as the catalytically active suprazyme state. The catalytic activity is mediated by in situ generated O₂˙⁻ from assemblies, resembling the behavior of natural enzymes such as laccase, which produces ROS such as O₂˙⁻ and OH˙ during substrate oxidation.³² Notably, G₀Ant exhibits strong antimicrobial activity under white light, effectively eradicating both drug-susceptible E. coli and methicillin-resistant Staphylococcus aureus (MRSA) strains. In the future, the G₀Ant assembly can be further adapted for topical antimicrobial therapy, biofilm eradication, surface disinfection, and incorporation into biocompatible nanocarrier platforms for localized photodynamic treatment.

Results and discussion

Molecular design strategy for tuning the electronic properties

To extend the photophysical utility of benzohydrazide-based AIEgens, we designed two new derivatives of the compound ‘G₀’ from our previous work (herein G₀Ben, Fig. 1A)¹³ by incorporating extended aromatic moieties to enhance the π–π interactions and shift the absorption range of the system. The molecules N′,N″-((1E,1′E)-1,4-phenylenebis(methaneylylidene))di(2-naphthohydrazide) (G₀Nap, Fig. 1A) and N′,N″-((1E,1′E)-1,4-phenylenebis(methaneylylidene))di(9-anthracenylhydrazide) (G₀Ant, Fig. 1A) were synthesized in good yields, following the procedures described in the supporting information (Schemes S1–S3).³³ The purity of the products was confirmed by ¹H NMR, ¹³C NMR spectroscopy and mass spectrometry analysis (Fig. S1–S6). The incorporation of π-conjugated aromatic units was expected to extend the conjugation length, as well as intensify the intermolecular π–π stacking, thereby red-shifting the optical response into the visible range.³⁴ The UV-Vis absorption spectra of G₀Nap and G₀Ant (20 µM) in DMF indeed showed significant bathochromic shifts compared to G₀Ben. UV-Vis absorption spectra were recorded in pure DMF to ensure complete molecular dissolution and to avoid scattering effects arising from aggregation. While G₀Ben exhibited an absorption maximum at 344 nm, G₀Nap displayed an absorption band in the range of 320–420 nm with a distinct maximum at ∼385 nm, whereas G₀Ant showed an even stronger red shift with absorption in a wide region of up to ∼500 nm (Fig. 1B). The systematic modulation in the absorption of the molecules confirmed that extending the aromatic backbone effectively shifts the n–π and π–π transitions into the visible-light region. Increasing the water fraction (up to 90% H₂O in DMF–water mixture) resulted in changes in the absorption profiles of the molecules. Beyond a threshold water content, significant broadening of the spectrum could be observed, indicating the enhanced interactions among molecules leading to aggregation and self-assembly formation (Fig. S7). However, the colloidal assemblies driven by mainly π–π interactions remain well-dispersed without precipitation under experimental conditions (Fig. S8).

Fig. 1 (A) Chemical structures of G₀Ben, G₀Nap and G₀Ant. (B) UV-Vis absorption spectra of the three molecules in 100% DMF. (C) FL spectra of G₀Ben, G₀Nap and G₀Ant in a DMF–water mixture (90% H₂O v/v). Excitation: 350 nm, 370 nm and 430 nm, respectively. (D) Time-domain fluorescence intensity decay of the assemblies of G₀Ben, G₀Nap and G₀Ant in 90% H₂O (v/v), exhibiting the highest lifetime of the G₀Ant assembly when excited using a 415 nm (visible) laser. (E) FESEM images of the G₀Ben, G₀Nap and G₀Ant self-assemblies. Conditions: G₀X (X = Ben, Nap, and Ant) = 20 µM and 90% H₂O (v/v).

When the water fraction was increased to the solutions of G₀Ben, G₀Nap and G₀Ant, the fluorescence intensity increased with increasing water fractions (Fig. S9), which is a typical characteristic of AIEgens. Notably, G₀Ant shows an initial enhancement in the emission intensity with increasing water content due to the onset of aggregation, followed by a red-shift and decrease at higher water fractions, which is consistent with tighter packing and aggregation-caused quenching (Fig. S9C). A similar aggregation (absorption/emission) behaviour could be observed with increasing water fraction in the DMSO–water mixture of G₀Ant (Fig. S10). Furthermore, the critical aggregation concentration (CAC) of G₀Ant was determined by increasing its concentration in a solution of 90% H₂O (v/v) DMF–water mixture. The solutions were excited at 430 nm, and the corresponding emission at 560 nm was recorded. The fluorescence started increasing with increasing concentrations of the G₀Ant assembly. CAC was determined via the intersection point of the linear parts of the titration curves, which gave the value of ∼4.5 µM (Fig. S11). Together, photoluminescence (PL) studies revealed that G₀Ben exhibited blue emission (λ_max ∼ 424 nm) and G₀Nap exhibited cyan-green emission (λ_max ∼ 495 nm), while G₀Ant showed orangish emission (λ_max ∼ 560 nm) in 90% H₂O (DMF:H₂O mixture v/v) (Fig. 1C). The aggregates were also visibly emissive under light irradiation (Fig. 1C). The observed bathochromic shift in the emission spectra can be attributed to the increased π-conjugation that effectively lowers the HOMO–LUMO energy gap and thereby red-shifts the emission band. Moreover, when excited with a near-visible laser at 415 nm, G₀Ant exhibited the longest excited-state lifetime (1.9 ns), followed by G₀Nap (2.1 ns) and the shortest for G₀Ben (0.2 ns) (Fig. 1D), demonstrating the effective stabilization of the excited state of G₀Ant. Field emission scanning electron microscopy (FESEM) imaging revealed distinct supramolecular morphologies, ranging from small fibrous assemblies in G₀Ben and G₀Ant to flake-like architectures in G₀Nap (Fig. 1E). ¹H NMR titration was attempted to probe the aggregation-induced π–π stacking, where the addition of 10% D₂O to G₀Ben or G₀Nap dissolved in DMSO-d₆ resulted in upfield shifts of aromatic protons, indicative of enhanced π–π interactions (Fig. S12).^13,32 In contrast, G₀Ant exhibited strong aggregation and poor solubility even in neat DMSO-d₆, leading to precipitation at the concentrations required for NMR measurements (Fig. S13). The wide-angle XRD profiles of the three assemblies exhibit characteristic reflections in the wide-angle region that are commonly attributed to aromatic π–π stacking interactions (Fig. S14).³³ All three systems exhibit such features, indicating the presence of aggregation-driven aromatic stacking within the self-assembled structures. Differences in the position and broadening of these reflections suggest variations in the π–π stacking organization among the assemblies, which can be attributed to differences in the nature and extent of the aromatic cores. Collectively, these findings establish the π-extension of benzohydrazide AIEgens as a robust strategy to engineer visible-light-responsive suprazymes with enhanced catalytic efficiency, as discussed later.

We calculated the ground state (S₀) HOMO and the first excited state (S₁) LUMO energy levels for the three molecules (G₀Ben, G₀Nap, and G₀Ant) using density functional theory (DFT) (Fig. 2A). As expected, all three molecules exhibited band gaps within the typical semiconductor range, which is highly promising for photocatalytic applications such as the generation of ROS. The band gap systematically decreased from 3.70 eV in G₀Ben to 3.28 eV in G₀Nap and further to 2.88 eV in G₀Ant (Fig. 2A). This reduction in the energy gap arises from the progressive extension of the π-conjugated system with increasing aromatic ring size, which enhances electronic delocalization. As a result, the energy difference between S₀ and S₁ decreases, consistent with the experimentally observed bathochromic shift in absorption (Fig. 1B) and the enhanced excitation lifetime under visible light irradiation.

Fig. 2 (A) DFT studies demonstrating the HOMO–LUMO characteristics and the corresponding decreasing band gap of G₀Ben, G₀Nap and G₀Ant in the gaseous state. (B) Band gaps of G₀Ben, G₀Nap and G₀Ant estimated from the UV-vis spectra. (C) Photocurrent response of the G₀Ant suprazyme in a DMF–water mixture (90% H₂O v/v) dried on a silicon wafer upon alternate light illumination.

To further validate the theoretical predictions, the experimental band gaps were estimated from Tauc-plot analysis of the UV-Vis absorption spectra (Fig. 2B). The plots clearly revealed a decrease in band gap values from G₀Ben (3.12 eV) to G₀Nap (2.75 eV) and further to G₀Ant (2.08 eV), which is consistent with the DFT results (Fig. 2B). The observed trend further confirms that extension of the π-conjugated framework effectively narrows the energy gap, thereby facilitating visible-light excitation. Such reduced band gaps not only support efficient light harvesting under white-light irradiation, but also underline the potential of these benzohydrazide derivatives in photocatalytic and ROS-generating applications. The photoelectrochemical properties of the G₀Ant assemblies were further investigated to probe their ability to generate excited-state charge carriers under visible-light irradiation. A thin film of G₀Ant coated on a silicon wafer substrate exhibited a pronounced photocurrent response when illuminated with a broadband white-light source at an applied bias of 4 V (Fig. 2C). The on–off switching cycles demonstrated stable and reproducible photocurrent generation, confirming efficient electron excitation and transfer within the supramolecular framework. This observation directly supports the proposed mechanism, wherein photoexcited electrons harvested from visible light are utilized to activate molecular oxygen, leading to the formation of ROS.

Light-induced oxidase mimetic activity

The oxidase-like activity of the self-assembled G₀Ant system, driven by ROS production, was investigated. The oxidase-like activity of the system was evaluated at concentrations above its CAC (Fig. S11) using 3,3′,5,5′-tetramethylbenzidine (TMB) as the chromogenic substrate under white light irradiation (Fig. 3A). In acetate buffer (100 mM, pH 4.0), the reaction mixture containing G₀Ben/ G₀Nap/ G₀Ant with TMB underwent a colour change from colourless to blue upon photo-irradiation with a maximum absorbance at 652 nm, indicating the oxidation of TMB catalyzed by the assembled structures. Importantly, this oxidation occurred in the absence of hydrogen peroxide, confirming that the systems exhibit intrinsic oxidase-like activity rather than peroxidase-like activity. As shown in Fig. 3B, all three derivatives exhibit photo-oxidase activity, but with markedly different efficiencies. Among them, G₀Ant consistently produced the highest absorbance of oxidized TMB (TMB_ox) across the visible region, peaking around 456 nm and maintaining a high level under white-light irradiation. In contrast, G₀Nap showed moderate activity, while G₀Ben exhibited the lowest activity. The high response of G₀Ant under white light is particularly significant, as it enables the use of a broad and readily available light source for activating the system. Moreover, G₀Ant shows negligible oxidase-like activity in the monomeric form (0% H₂O) and appears only in the assembled state (90% H₂O), demonstrating a structure–function relationship typical of supramolecular enzyme mimic “suprazyme” (Fig. S15). Consequently, white light was chosen for all further studies to exploit the superior activity of the G₀Ant assembly under practical irradiation conditions. The control experiments in Fig. 3C demonstrate that, in the absence of light, no appreciable oxidation of TMB is observed for any of the G₀X systems (X = Ben, Nap, Ant). Under white-light irradiation, G₀Ant in combination with TMB produced a clear increase in TMB_ox, confirming the light-induced oxidase activity. Neither the assemblies alone nor TMB alone generated any appreciable signal under identical conditions (Fig. S16). These data confirm that the observed activity arises from the self-assembled G₀Ant and is light-dependent. Furthermore, the catalytic activity is largely retained even at water contents as high as 98% in the DMF–water mixture, indicating that only a minimal amount of co-solvent is required for application (Fig. S17). Moreover, comparable catalytic activity was observed when DMF was replaced with DMSO as the co-solvent (Fig. S18). Direct comparison of the three derivatives under identical white-light conditions further established G₀Ant as the most efficient system (Fig. 3C). The rate of TMB oxidation with G₀Ant was significantly higher than with G₀Nap or G₀Ben, corroborating the wavelength-screening results. This justified selecting G₀Ant for all subsequent mechanistic and kinetic studies.

Fig. 3 (A) Schematic of the light-induced oxidase activity of the self-assembled G₀Ant structures. (B) Oxidation of TMB (0.5 mM) in acetate buffer (100 mM and pH 4.0) in the presence of 20 µM of G₀Ben, G₀Nap and G₀Ant irradiated with different light sources, including 370 nm, 390 nm, 427 nm, 456 nm, and white light for 10 min. (C) Rate of TMB (0.5 mM) oxidation in the presence of G₀Ben, G₀Nap and G₀Ant after white light irradiation in acetate buffer (100 mM and pH 4.0) for 10 min. (D) The initial rate (V₀) of the catalytic reaction with G₀Ant at different substrate concentrations. The solid red line is derived from fitting the Michaelis–Menten equation. (E) Amount of oxidized TMB by G₀Ant upon white light irradiation monitored at different time intervals. (F) The cumulative increase in the amount of oxidised TMB upon alternatively exposing the assembly of G₀Ant to light. The rate of product formed is almost the same with each on–off event, as compared to continuous light irradiation. (G) Rate of TMB oxidation with different concentrations of G₀Ant in acetate buffer (100 mM and pH 4.0) for 10 min. (H) G₀Ant activity across varying serum concentrations measured using TMB oxidation. All the data present the mean ± SD for n technical replicates (n = 3).

Kinetics of the oxidase activity of G₀Ant

Kinetic analysis of TMB oxidation by the G₀Ant assembly under white light followed Michaelis–Menten behaviour (Fig. 3D). The fitted parameters (V_max = 2.63 µM min⁻¹, K_M = 0.0219 mM, R² = 0.9601) indicate a high catalytic efficiency and strong substrate affinity compared to small-molecule oxidase mimics. These values highlight the potent enzyme-like activity of the self-assembled system. Time-course measurements under continuous white-light irradiation (Fig. 3E) revealed a steady increase in the TMB_ox concentration for G₀Ant, whereas the control remained negligible. The rapid and sustained formation of TMB_ox underscores the robust catalytic activity of the assembly. The photo-responsiveness of the G₀Ant system was probed by alternating exposure to white-light irradiation and darkness (Fig. 3F). TMB_ox formation followed a staircase-like profile, with the reaction rate increasing during light ‘on’ phases and staying constant during ‘off’ phases. This switchable behaviour confirms that the oxidase-like activity can be externally modulated by light, a key advantage for controlled applications. An increase in the G₀Ant concentration produced a corresponding rise in activity (Fig. 3G and Fig. S19), supporting the conclusion that the G₀Ant assembly is responsible for the catalytic behaviour. The oxidase activity assay for G₀Ant was conducted in differently diluted human serum to examine its enzymatic performance in real-world conditions (Fig. 3H). The unchanged catalytic activity of the G₀Ant suprazyme in serum demonstrates its promise as a viable enzyme mimic for biomedical applications.

Mechanism and robustness of the suprazyme

To identify the type of ROS involved, different scavengers were added during TMB oxidation under white light (Fig. 4A). The activity was not suppressed by hydroxyl radical or singlet oxygen scavengers (IPA, L-His, NaN₃), but decreased markedly with hole (h⁺) quenchers (KI, EDTA) and under N₂ atmosphere, indicating that superoxide anion (O₂˙⁻) could be the primary ROS generated by the G₀Ant assembly under white light.^35,36 UV-Vis spectra of N,N,N′,N′-tetramethyl phenylenediamine (TMPD) oxidation (Fig. 4B) further supported the generation of O₂˙⁻. Only the combination of G₀Ant, TMPD, and white light produced the characteristic absorbance peaks of oxidized TMPD, confirming photo-induced electron transfer from the assembly to O₂ and subsequent O₂˙⁻ production.³⁷ O₂˙⁻ production was also verified using dihydrorhodamine (DHR) as a fluorescent probe (Fig. 4C). Under white-light irradiation, G₀Ant rapidly converted non-fluorescent DHR into its bright fluorescent product, whereas controls lacking G₀Ant or light showed negligible signal. This assay confirms efficient photo-induced O₂˙⁻ generation by the G₀Ant assembly under white light.

Fig. 4 (A) Effect of different scavengers on the TMB oxidation via assembly of G₀Ant under white light irradiation. Experimental conditions: 0.5 mM TMB, 20 µM G₀Ant, 5 mM KI, 0.5 mM L-His, 20 mM IPA and 1.0 mM NaN₃, 10 mM EDTA, N₂ purging: 10 min, and white light irradiation time: 10 min. (B) Absorbance spectra of TMPD in the presence of G₀Ant as a catalyst with irradiation of light resulting in the formation of TMPD⁺˙. (C) Fluorescence spectra for O₂˙⁻ using DHR as an indicator under white light irradiation (20 mW cm⁻²) with DHR as the control. (D) Effect of varying NaCl concentrations (0–500 mM) on the photo-oxidase activity of 20 µM G₀Ant relative to the natural enzyme laccase. (E) Comparison of the percentage oxidase activity of 20 µM G₀Ant at varying pH levels (3–10) with laccase. (F) Comparison of the percentage photo-responsive oxidase activity of G₀Ant (20 µM) and laccase across a temperature range of 4–80 °C. The error bars represent the mean ± SD for n = 3 independent measurements.

Natural enzymes often suffer from instability and loss of activity under harsh conditions. To evaluate whether the G₀Ant suprazyme offers superior stability compared to the natural oxidase enzymes, such as laccase that also uses O₂˙⁻-mediated oxidation, we assessed its activity under extreme conditions, including high ionic strength, varying pH, and elevated temperatures. Under conditions of high ionic strength, a clear contrast in activity was observed between the G₀Ant suprazyme and the natural enzyme. Laccase exhibited a ∼50% loss of activity at 100 mM NaCl, whereas G₀Ant preserved more than 80% of its activity even at 500 mM NaCl (Fig. 4D). The activity loss of the natural enzyme is attributed to a salting-out effect, which alters the active-site structure and charge distribution of the enzyme.³⁸ Testing the oxidase activity in buffers of varying pH revealed maximal activity from pH 3–6, which is better than the natural enzyme (Fig. 4E). The acyl hydrazone linkage in G₀Ant makes it more stable and less prone to hydrolysis under a broad pH range.^39–41 The catalytic activity of the suprazyme was further evaluated across a temperature range of 4–80 °C. Whereas laccase activity dropped sharply above 60 °C, the G₀Ant suprazyme maintained approximately 95% of its activity even at 80 °C (Fig. 4F). The G₀Ant suprazyme retains over 98% of its catalytic efficiency even after using for five consecutive cycles, demonstrating excellent stability (Fig. S20). Collectively, the high ionic strength, pH, recyclability and thermal stability of G₀Ant underscore its suitability as a resilient and efficient catalyst for practical applications.

Antibacterial activity of G₀Ant

The antibacterial efficacy of the G₀Ant suprazyme was evaluated against Gram-positive (MRSA) and Gram-negative (E. coli) bacteria with and without white-light irradiation (Fig. 5A and B) using the plate count method. In the absence of the material, both MRSA and E. coli exhibited dense bacterial growth. Addition of G₀Ant alone caused only a negligible and slight reduction in colony density, indicating limited dark toxicity (Fig. S21). However, when bacteria were treated with G₀Ant (80 µM) under white-light irradiation for 10 minutes, a dramatic reduction of bacterial colonies was observed, with the plates appearing almost clear. This result highlights the synergistic effect of the self-assembled G₀Ant and light, demonstrating its potent photo-induced antibacterial action through ROS. FESEM images of MRSA and E. coli after treatment also revealed severe membrane disruption, cell shrinkage, and surface roughness compared to those of the untreated bacteria, which showed a smooth, intact morphology, providing direct visual evidence of bacterial killing by the photo-activated G₀Ant assembly (Fig. 5A and B). The strong activity under broad-spectrum white light rather than narrow-band lasers underscores the practical relevance of the system for real-world antimicrobial applications. The antibacterial effect of the G₀Ant assembly under white-light irradiation was quantified by colony counting at varying concentrations (Fig. 5C and D). At a concentration of 80 µM G₀Ant, no colony of MRSA survived under the light irradiation. Likewise, a concentration-dependent decline was observed for E. coli, with complete eradication at 80 µM. The data clearly demonstrate the potent photo-antibacterial activity of the assembly. Together, the colony-forming unit (CFU) counts and SEM images corroborate the strong bactericidal action of G₀Ant under white-light irradiation.

Fig. 5 (A) Photographs of bacterial colonies obtained by incubating Gram-positive bacteria MRSA with G₀Ant (80 µM) in the dark and when exposed to white light; the bacterial samples not treated with these compounds served as controls. The corresponding FESEM images of MRSA are presented below the plates. (B) Photographs of bacterial colonies obtained by incubating Gram-negative bacteria E. coli with G₀Ant (80 µM) in the dark and when exposed to white light; the bacterial samples not treated with these compounds served as controls. The corresponding FESEM images of E. coli are presented below the plates. (The blue arrows in the SEM images represent the bacteria with collapsed and destroyed membranes). (C) The bactericidal activity of G₀Ant at different concentrations (0, 20, 40, 60, 80 and 100 µM) against MRSA using the colony counting method, after irradiation with white light. (D) The bactericidal activity of G₀Ant at different concentrations (0, 20, 40, 60, 80 and 100 µM) against E. coli using the colony counting method, after irradiation with white light. The error bars represent the mean ± SD for n = 3 independent measurements.

To evaluate the biocompatibility of G₀Ant, the cytotoxicity was assessed using the MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl tetrazolium bromide) assay on a non-cancerous human embryonic kidney (HEK293) cell line. The detailed methodology is provided in the supporting information. HEK293 cells were exposed to the material for 24 hours before MTT treatment. A concentration range of 0–150 µM was used for the treatments. G₀Ant was found to be cytocompatible under light irradiation in the range of 0–80 µM, which is approximately equal to the minimal bactericidal concentration (MBC) value of 80 µM (Fig. S22). G₀Ant also shows a biocompatibility of >90% in the absence of light. The observed low cytotoxicity indicates that G₀Ant can provide an effective enzyme mimicking scaffold with potential biomedical applications.

Conclusion

In this study, we demonstrate a molecular design strategy for developing metal-free, visible-light-responsive suprazymes by π-extension of benzohydrazide AIEgens. The stepwise modification from benzene to anthracene cores enabled precise control over band energies, resulting in enhanced ROS generation under clinically relevant white-light irradiation. The optimized derivative, G₀Ant, exhibited effective oxidase-like activity, high catalytic stability under harsh conditions, and potent bactericidal activity against drug-resistant pathogens, while maintaining low cytotoxicity toward mammalian cells. Importantly, its ability to function without added peroxides and to operate efficiently under broad-spectrum visible light underscores its translational potential for real-world biomedical applications such as antimicrobial therapy and wound healing. The small amount (2–5%) of organic co-solvent present in the colloidal dispersion formulation does not hinder its practical biological applications. This work highlights supramolecular structural engineering as a powerful tool for creating robust artificial enzyme systems, paving the way for next-generation light-assisted therapeutics to treat antibiotic-resistant bacteria.

Conflicts of interest

The authors declare no competing financial interest.

Data availability

The data supporting this study are available from the corresponding authors upon reasonable request.

The supplementary information (SI) contains experimental procedures, molecular characterization, photophysical and structural analysis, enzyme-mimicking and ROS studies, antimicrobial assays, recyclability tests, and cytotoxicity evaluations. See DOI: https://doi.org/10.1039/d5tb02325d.

Acknowledgements

P. R. thanks the Indian Council of Medical Research (ICMR) (Grant no. IRPSG-2024-01-02893) and MoE-STARS/STARS-2/2023-0651 for financial support. S. R. acknowledges major financial support from the Science and Engineering Research Board (SERB) (CRG/2022/009021) and the Department of Biotechnology (BT/PR49984/MED/32/911/2023). The authors are grateful to the Department of Science and Technology (DST-FIST: SR/FST/PSII009/2010) for the equipment facility at MRC. R. M. and R. P. are thankful to MoE for the doctoral research fellowship. A. K. and A. M. are grateful to the Prime Minister Research Fellowship for their doctoral research.

References

1. C. S. Ho, C. T. H. Wong, T. T. Aung, R. Lakshminarayanan, J. S. Mehta, S. Rauz, A. McNally, B. Kintses, S. J. Peacock, C. de la Fuente-Nunez, R. E. W. Hancock and D. S. J. Ting, Lancet Microbe, 2025, 6, 100947.
2. M. A. Salam, M. Y. Al-Amin, M. T. Salam, J. S. Pawar, N. Akhter, A. A. Rabaan and M. A. A. Alqumber, Healthcare, 2023, 11.
3. A. Arshadi Edlo, K. Akhbari and D. J. Henry, Mater. Adv., 2025, 6, 7662.
4. M. Capuozzo, A. Zovi, R. Langella, A. Ottaiano, M. Cascella, M. Scognamiglio and F. Ferrara, Antibiotics, 2024, 13, 1112.
5. B. Ran, L. Ran, Z. Wang, J. Liao, D. Li, K. Chen, W. Cai, J. Hou and X. Peng, Chem. Rev., 2023, 123, 12371.
6. Y. Liu, R. Qin, S. A. J. Zaat, E. Breukink and M. Heger, J. Clin. Transl. Res., 2015, 1, 140.
7. J. M. Soares, V. V. Yakovlev, K. C. Blanco and V. S. Bagnato, Sci. Rep., 2024, 14, 28268.
8. P. Hyziuk, M. Flaibani, P. Posocco and V. Sashuk, Chem. Sci., 2024, 15, 16480.
9. S. Abedanzadeh, Z. Moosavi-Movahedi, N. Sheibani and A. A. Moosavi-Movahedi, Biochem. Eng. J., 2022, 183, 108463.
10. M. Solra, K. Jaiswal, A. Srivastava, S. Mishra, A. Kamra, M. De and S. Rana, Adv. Healthcare Mater., 2025, 2500835.
11. O. S. Tiwari, V. Rawat, H. Zhang, S. Chibh, S. Rencus-Lazar, C. E. Diesendruck and E. Gazit, J. Pept. Sci., 2024, 30, e3626.
12. B. Gruber, E. Kataev, J. Aschenbrenner, S. Stadlbauer and B. König, J. Am. Chem. Soc., 2011, 133, 20704.
13. A. Kamra, R. Mahashaya, P. Rajamalli and S. Rana, Small, 2025, 21, 2411416.
14. S. Bhattacharya, K. Snehalatha and V. P. Kumar, J. Org. Chem., 2003, 68, 2741.
15. P. Dowari, M. Kumar Baroi, T. Das, B. Kanti Das, S. Das, S. Chowdhuri, A. Garg, A. Debnath and D. Das, J. Colloid Interface Sci., 2022, 618, 98.
16. S. Xu, H. Wu, Y. Liu, B. Zhang and Z.-G. Wang, Small, 2025, 21, 2501778.
17. L. Zhou, J. Yan, C. Cui, Y. Xu, K. Zhang, M. Du, Z. Zhang, X. Wu and B. Li, Small Struct., 2025, 2500195.
18. R. Qu, L. Shen, Z. Chai, C. Jing, Y. Zhang, Y. An and L. Shi, ACS Appl. Mater. Interfaces, 2014, 6, 19207.
19. S. Bhattacharyya, S. R. Ali, M. Venkateswarulu, P. Howlader, E. Zangrando, M. De and P. S. Mukherjee, J. Am. Chem. Soc., 2020, 142, 18981.
20. G. Li, W. Tian, C. Zhong, Y. Yang and Z. Lin, ACS Appl. Mater. Interfaces, 2022, 14, 21750.
21. M. Talebi, K. Dashtian, R. Zare-Dorabei, H. Ghafuri, M. Mahdavi and F. Amourizi, Anal. Chim. Acta, 2023, 1247, 340924.
22. W. Q. Wei, D. Liu, Z. Wei and Y. F. Zhu, ACS Catal., 2017, 7, 652.
23. J. Zhang and J. Liu, Nanoscale, 2020, 12, 2914.
24. Y. Lin, X. Liu, Z. Liu and Y. Xu, Small, 2022, 18, 2201184.
25. X. Wang, L. He, Q. Cai, Z. Su, Z. Huang, H. Sun, Y. Lai, Z. Chen, J. Ye, Y. Yu, Z. Zou, H. Huang and H. Zhu, Chem. Sci., 2025, 16, 15638.
26. S. Zhen, F. Ma, H. Chen, L. Li, M. Li, Z. Zhao and B. Z. Tang, Aggregate, 2025, 6, e70044.
27. R. Hossain, J. Hollow, V. Chesterfield, D. Langley and A. Apblett, Mater. Adv., 2025, 6, 7821.
28. A. Zielińska-Jurek, E. Kowalska, J. W. Sobczak, W. Lisowski, B. Ohtani and A. Zaleska, Appl. Catal., B, 2011, 101, 504.
29. J. Song, M. Zheng, X. Yuan, Q. Li, F. Wang, L. Ma, Y. You, S. Liu, P. Liu, D. Jiang, L. Ma and W. Shen, J. Mater. Sci. Technol., 2017, 52, 6976.
30. Y. Yalçın, M. Kılıç and Z. Çınar, Appl. Catal., B, 2010, 99, 469.
31. T. Schembri, J. Albert, H. Hebling, V. Stepanenko, O. Anhalt, K. Shoyama, M. Stolte and F. Würthner, ACS Cent. Sci., 2025, 11, 452.
32. P. Rajamalli, S. Atta, S. Maity and E. Prasad, Chem. Commun., 2013, 49, 1744.
33. P. Rajamalli and E. Prasad, Soft Matter, 2012, 8, 8896.
34. L. Brandner and T. J. J. Müller, Front. Chem., 2023, 11, 1124209.
35. Y. Wang, X. Feng, J. Cao, X. Zheng, X. Gong, W. Yu, M. Wang and S. Shi, Angew. Chem., Int. Ed., 2024, 63, e202319139.
36. Y. Xia, W. He, J. Li, L. Zeng, T. Chen, Y. Liao, W. Sun, J. Lan, S. Zhuo, J. Zhang, H. Yang and J. Chen, Anal. Chem., 2019, 91, 8406.
37. S. Yu, R. X. Zhu, K. K. Niu, N. Han, H. Liu and L. B. Xing, Chem. Sci., 2024, 15, 1870.
38. A. M. Hyde, S. L. Zultanski, J. H. Waldman, Y. L. Zhong, M. Shevlin and F. Peng, Org. Process Res. Dev., 2017, 21, 1355.
39. P. A. Sreelakshmi, R. Mahashaya, S. Leitherer, U. Rashid, J. M. Hamill, M. Nair, P. Rajamalli and V. Kaliginedi, J. Am. Chem. Soc., 2024, 146, 35242.
40. A. E. Dascalu, L. Halgreen, A. Torres-Huerta and H. Valkenier, Chem. Commun., 2022, 58, 11103.
41. J. Xu, Y. Liu and S.h. Hsu, Molecules, 2019, 24, 3005.

---

Footnote

† These authors have contributed equally.

---