Light-triggered molecular mechanotherapy of tumor using membrane-mimicking conjugated oligoelectrolytes

Abstract

A class of light-mediated mechanotherapeutic agents was developed on the basis of conjugated oligoelectrolytes (COEs), which mimic the topology of lipid membranes and intrinsically exhibit excellent biocompatibility. Low-dose white light irradiation (20 milliwatts per square centimeter for 10 minutes) substantially decreased the half-maximal inhibitory concentration of the optimized COE against A549 cancer cells from more than 256 to 0.6 micromolar. Typical photodynamic and photothermal effects were not responsible for the potent anticancer efficacy. Biophysical and photophysical experiments using vesicle models revealed that COEs can induce mechanical force likely by molecular conformation change within lipid membranes under light exposure, supporting the mechanotherapeutic mechanism by which COEs after excitation can physically disrupt cell membrane. Investigation of two other COEs with similar spectral properties but different backbone architectures revealed that their mechanotherapeutic efficacy is dependent on molecular topology. These results highlight the potential to develop light-responsive mechanotherapeutic agents based on membrane-mimicking COE platform for cancer treatment.

A type of mechanotherapeutic agents disrupts cancer cell membranes via molecular motion upon photoactivation.

INTRODUCTION

Cancer remains a leading global health challenge, with the frequent development of drug resistance further complicating treatment and imposing notable economic burdens (1, 2). One major mechanism of resistance is the up-regulation of DNA repair pathways in cancer cells, enabling them to survive DNA damage induced by conventional therapies such as chemotherapy and radiation (3, 4). Additional mechanisms, including genetic mutations that prevent drug binding, activation of alternative survival pathways, and expression of efflux pumps that expel therapeutic agents, further limit the effectiveness of targeted therapies and immunotherapies, hindering long-term therapeutic success (5, 6). Emerging techniques like photodynamic therapy (PDT) and photothermal therapy (PTT) offer promising alternatives by providing highly selective, minimally invasive cancer treatments with reduced systemic side effects, leveraging precise spatiotemporal control through light exposure (7–9). However, resistance to these therapies can still develop; for instance, cancer cells may activate antioxidant defenses to neutralize reactive oxygen species (ROS) generated by PDT or up-regulate heat shock proteins (HSPs) to withstand thermal stress induced by PTT (10, 11). Overcoming these challenges necessitates therapeutic strategies that bypass traditional resistance mechanisms, such as physically disrupting fundamental structures of cancer cells (12, 13).

Molecular mechanotherapy, which uses mechanical forces to directly and irreversibly disrupt cancer cell membranes, represents a promising strategy to minimize the potential for resistance development (14). Unlike PDT and PTT, which rely on ROS or heat generation, molecular mechanotherapy directly disrupts cell membranes through physical forces, making it less dependent on the tumor microenvironment and inherently resistant to common adaptive mechanisms (15, 16). In this context, Tour and colleagues designed molecular machines on the basis of Feringa’s unidirectional light-triggered motors that drill pores into cell membranes, causing irreversible damage and subsequent cell death (17–19). The recent development of cyanine-based molecular jackhammers further demonstrates that light-activated vibronic-driven membrane disruption can effectively eradicate cancer cells, underscoring the translational potential of this approach (20, 21). Building on these pioneering studies, there is ongoing interest in expanding the range of mechanotherapeutic agents by exploring alternative molecular platforms that combine potent anticancer efficacy with excellent biocompatibility.

Membrane-mimicking conjugated oligoelectrolytes (COEs) are molecules characterized by linear hydrophobic conjugated backbones adorned with side chains that terminate in hydrophilic groups, closely resembling the structure of lipid bilayers (22, 23). COEs have been used as fluorogenic and lipophilic probes due to their excellent aqueous solubility and high biocompatibility with cellular membranes. These properties enable precise detection of extracellular vesicles via flow cytometry and stable tracking of liposomes in vivo, effectively addressing issues such as false positives and dye leakage associated conventional cyanine-based dyes (like PKH26) (24–27). Beyond fluorescence, the conjugated backbones of COEs can convert absorbed photoenergy into mechanical outputs, such as molecular torsions or other conformation variations, making them promising candidates for mechanotherapeutic agents.

In this study, we explored the application opportunities of the previously synthesized membrane-mimicking COE named BT in the mechanotherapeutic model (24). Upon irradiation with the built-in laser of a confocal microscope (~3 mW), BT molecules intercalated into cancer cell membranes, rapidly inducing cell expansion and rupture. The addition of ROS scavengers did not compromise phototherapeutic efficacy, and no substantial temperature elevation was recorded. These observations suggest that membrane disruption was attributed to the mechanical action of BT rather than photothermal or photodynamic effects under low-dose light irradiation. The emerging of distinct excited-state species and morphological changes in large multilamellar vesicles (LMVs) upon irradiation further corroborate the proposed mechanotherapeutic mechanism. Two additional COEs with similar spectral properties but different backbone structures were investigated, suggesting that the bent backbone with small rotational barrier enhances mechanotherapeutic efficacy. Moreover, BT under irradiation induced cancer cell pyroptosis and elicited a potent immune response, demonstrating high therapeutic efficacy in subcutaneous tumor-bearing mice. This work potentially reveals unexplored avenues for developing membrane-targeting mechanotherapeutic agents to broaden cancer therapeutic options.

RESULTS

Distinguishing the therapeutic mechanism

The irradiation of light-absorbing molecules can give rise to multiple effects, including photothermal and photodynamic responses, which open opportunities to influence cancer treatment outcomes. To provide supporting evidence of a mechanotherapeutic action from these confounding effects, we used two previously synthesized membrane-mimicking COEs, Ben and S6 (Fig. 1A) (24). Ben contains a heterogeneous backbone including benzene, thiophene, and vinyl units, while S6 has a regio-regular oligo-phenylenevinylene backbone. Density functional theory (DFT) calculations were used to optimize the molecular conformations, revealing that Ben has a highly twisted backbone with a large dihedral angle of 24° between the dithiophenebenzene “core” and the stilbene “wing.” In contrast, S6 exhibits a planar backbone, with the dihedral angle between the benzene and vinyl groups reduced to nearly 0°. These two COEs serve as molecular models to preliminarily investigate the relationship between chemical structure and mechanotherapeutic efficacy.

Fig. 1. Characterizations of photoelectronic properties of COEs.

(A) Chemical structure of COEs and their density functional theory (DFT) optimized backbones. (B) FRET analyses between donor fluorophore Ben and acceptor Nile Red in lipid bilayers. Signal intensity was measured in arbitrary units (a.u.). (C and D) Representative time-evolution snapshots of Ben (C) and S6 (D) in lipid bilayers at 0 (left) and 200 (right) ns, as predicted by MD. (E) Normalized absorption and emission spectra of 10 μM COEs in phosphate-buffered saline (PBS). (F) Cyclic voltammograms of Ben and S6 measured in n-Bu₄NPF₆ acetonitrile solution (0.1 M); Fc, ferrocene reference; SCE, saturated calomel electrode as the reference electrode. (G) Relative emission intensity of fluorescent probe HPF treated with Ben or S6. Excitation wavelength is 490 nm. FI_t/FI_t0, fluorescence intensity ratio (at 515 nm) after/before light exposure. (H) Relative absorbance degradation curves of ABDA treated with Ben or S6. The irradiation was performed using white light at 20 mW cm⁻². A/A₀, absorption value ratio (at 410 nm) after/before light exposure.

The dynamic, amorphous, and nanoscale nature of lipid bilayer membranes poses challenges for directly observing the interactions of COEs. Building on previous studies that used polarized fluorescence microscopy and colocalization with lipophilic probes, we used a fluorescence resonance energy transfer (FRET) assay to validate the membrane-intercalating properties of COEs (28, 29). Large unilamellar vesicles (LUVs) composed of 1-palmitoyl-2-oleoyl-glycero-3-phosphocholine (POPC) and 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-(1′-rac-glycerol) sodium (POPG) at an 85:15 molar ratio served as model membranes (25). These vesicles were loaded with Nile Red and extruded through a 100-nm membrane to create the Nile Red–containing LUVs (NR-LUVs; see preparation details in the Supplementary Materials). Upon adding Ben to the NR-LUV solution in phosphate-buffered saline (PBS), the fluorescence intensity of Nile Red increased, while Ben’s emission decreased (Fig. 1B). Because FRET occurs only at close distances, this energy transfer from donor (Ben) to acceptor (Nile Red) indicates that Ben spontaneously intercalated into the lipid bilayer, positioning its conjugated backbone within the hydrophobic membrane core. A similar FRET behavior was observed for S6 (fig. S1). Molecular dynamic simulations (MD) further confirmed that both COEs can span over the lipid membrane, with their polar termini interacting with the phosphate head groups on both leaflets of the bilayer (Fig. 1, C and D, and fig. S2) (30). Ben adopts a more bent backbone conformation within lipid bilayers, in contrast to the more linear structure of S6. Nevertheless, the unique membrane-mimicking topology of both COEs confers good biocompatibility, as evidenced by half-maximal inhibitory concentration (IC₅₀) values exceeding 256 μM against A549 cells (fig. S3). Additionally, the six charged side chains of these molecules provide excellent aqueous solubility, and no notable nano-aggregation was detected in PBS, as quantified by an ultrafiltration method (fig. S4). This favorable biocompatibility and aqueous solubility highlight the advantages of COEs over conventional cyanine-based lipophilic molecules (25).

Despite differences in backbone configuration, Ben and S6 exhibited similar spectral properties in PBS, with absorption and emission peaks at ~400 and 550 nm, respectively (Fig. 1E). Cyclic voltammetry (CV) measurements determined the highest occupied molecular orbital and lowest unoccupied molecular orbital energy levels to be −2.6/−4.8 eV for Ben and −2.7/−5.0 eV for S6 (Fig. 1F and fig. S5). DFT calculations also supported both COEs have similar photoelectronic properties (fig. S6). The generation of ROS by Ben and S6 in PBS was evaluated using three different probes under white light at 20 mW cm⁻² (see spectrum in fig. S7) irradiation for 2 min: hydroxyphenyl fluorescein (HPF) for hydroxyl radicals (Fig. 1G and fig. S8), 9,10-anthracenediylbis(methylene)-dimalonic acid (ABDA) for singlet oxygen (Fig. 1H and fig. S9), and dihydrorhodamine 123 (DHR123) for superoxide anions (fig. S10) (31, 32). The results show that S6 generates ROS more efficiently than Ben at the same concentration, with ~2.0-, 1.7-, and 1.7-fold higher generation rates for hydroxyl radicals, singlet oxygen, and superoxide anions, respectively. This enhanced ROS production could be attributed to the higher absorption coefficient of S6, which is about 1.3 times greater than that of Ben (fig. S11).

We investigated their cellular uptake before evaluating the intracellular ROS generation of COEs. After adding 10 μM S6 to A549 cells in Dulbecco’s modified Eagle’s medium (DMEM), S6 stained the cell membranes within the first few minutes, followed by internalization into cytoplasm after 1 hour of incubation (fig. S12). In contrast, under the same conditions, no distinct “fluorescent hollow” pattern was observed before the internalization of Ben, likely because its bent backbone hinders effective membrane intercalation. Colocalization experiments were conducted using the mitochondrion-targeting stain MitoTracker Red and lysosome-targeting stain LysoTracker Red for cells preincubated with S6 or Ben for 24 hours (figs. S13 and S14). In agreement with previous findings, both S6 and Ben preferentially accumulated in lysosomes relative to mitochondria (33). Intracellular ROS production was assessed using the nonfluorescent probe 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA), which converts to the fluorescent compound 2′,7′-dichlorofluorescein (DCF) upon reacting with ROS (Fig. 2A). Compared to that in the control before white light treatment, bright green fluorescence from DCF was observed in both Ben- and S6-incubated cells after irradiation (34). Analysis by flow cell cytometry further indicated that S6 exhibited a greater capacity to generate ROS in cells compared to Ben under the same conditions (fig. S15).

Fig. 2. In vitro ROS generation and phototoxicity of COEs.

(A) Intracellular ROS evaluation using DCFH-DA probe for A549 cells treated by 5 μM COEs and white light irradiation (5 min, 20 mW cm⁻²) with PBS as control. (B) Micrographs of Ben- or S6-incubated A549 cells with or without light treatments, followed by double staining using calcein-AM (for viable cells, presented in green) and propidium iodide (PI; for dead cells, presented in red). Scale bars, 50 μm. (C) Live and dead cell subpopulation analyses using flow cytometry. (D and E) Viability of A549 cells that preincubated with different concentrations of Ben or S6 and treated by white light irradiation at 20 mW cm⁻² for 10 min. (F) Photothermal effects of 20 μM Ben or S6 in PBS under 10 min of white light exposure at 20 mW cm⁻². Neat PBS was used as control. (G) DFT-evaluated relative potential energy of Ben and S6 backbones along the dihedral between central core and stilbene wing segments, from trans-planar to cis-planar.

An inverse trend was observed when assessing cytotoxic effects using a Live/Dead Cell Double Staining assay, which uses calcein-AM and propidium iodide (PI) to stain viable and dead cells, respectively (Fig. 2B). Without light irradiation, cells incubated with either Ben or S6 predominantly exhibited fluorescence from calcein-AM. After white light treatment, fluorescence from PI became predominant in the Ben-incubated cells. In contrast, under the same conditions, the S6-incubated cells remained viable despite S6’s stronger ROS generation capability. Quantitative analysis using flow cytometry revealed that ~96% of cells, whether preincubated with Ben or S6, exhibited good viability before light exposure (Fig. 2C). Ten minutes of white light irradiation compromised around 97% of cells in the Ben-treated group, while the viability percentage in the S6-treated group remained nearly unaffected. Cell viability was further evaluated using the Cell Counting Kit-8 assay (Fig. 2, D and E). Under dark conditions, neither Ben nor S6 exhibited substantial toxicity. Upon white light irradiation (12 J cm⁻²), the IC₅₀ for A549 cells treated with Ben decreased to 1.8 μM, whereas no substantial phototoxicity was observed in the S6-treated cells in the test window.

The weaker ROS generation efficiency but greater anticancer efficacy of Ben suggests that additional light-responsive anticancer mechanisms, rather than photodynamic effects, are involved (35). Nonsubstantial temperature increase was found for both Ben and S6 in PBS compared to that in the control, arguing against substantially different photothermal effects (Fig. 2F). We hypothesize that the prominent phototoxicity of Ben is attributed to its mechanical effects on cell membrane. As shown in Fig. 2G, the relative potential energy scan calculated using DFT indicates that the rotational barrier between the dithiophenebenzene core and stilbene wing in the Ben backbone is 3.3 kcal mol⁻¹, substantially lower than that of S6 (5.7 kcal mol⁻¹). These results imply that the heterogeneous and bent backbone of Ben may enhance mechanical actions and membrane disruption outputs, leading to more potent anticancer efficacy compared to the linear S6. Notably, ROS can also contribute to cancer cell ablation. When irradiation intensities were increased to 40 or 60 mW cm⁻², substantial phototoxicity was observed for S6, with IC₅₀ values reduced to 7.7 and 0.9 μM, respectively (fig. S16). The poor anticancer performance of S6 in the above studies is probably due to the ROS generated being below the cell-killing threshold under the low-dose light irradiation.

Membrane-targeting molecular mechanotherapy

To further enhance mechanotherapeutic efficacy, a COE molecule named BT was synthesized by replacing the central benzene unit of Ben with the electron-deficient benzothiadiazole (Fig. 3A) (24). DFT simulations indicate a twisted backbone with a similar dihedral angle of 24° (fig. S17), and MD simulations confirmed that BT can also span lipid membranes (fig. S18), while CV revealed a reduced bandgap of 1.9 eV for BT. This narrow bandgap allows for longer wavelength absorption, extending beyond 600 nm in the visible region (Fig. 3B), enabling BT to absorb more photoenergy and convert it into mechanical outputs, particularly when using routine white light as the exposure source. Additionally, its intrinsic fluorescence response, with an emission peak at 700 nm, facilitates biodistribution analysis in biological contexts (36). Notably, substantial intracellular ROS generation was detected in BT-incubated A549 cells after white light irradiation at 20 mW cm⁻² using the DCFH-DA probe (fig. S19). The production rates of hydroxyl radicals, singlet oxygen, and superoxide anions for BT in PBS, measured with HPF, ABDA, and DHR123 probes, were 21, 12, and 60% lower than those of Ben, respectively (fig. S20). By contrast, BT exhibited more potent cell-killing capability under irradiation, despite its lower ROS production efficiency. For example, cell damage evaluation using a lactate dehydrogenase (LDH) release assay indicated that BT increased LDH release by 62%, compared to 24% for Ben under the same conditions (Fig. 3C) (37).

Fig. 3. Light-triggered membrane intercalation and disruption of BT.

(A) Chemical structure of BT and its cyclic voltammogram in 0.1 M n-Bu₄NPF₆ acetonitrile solution. (B) Normalized absorption and emission spectra of 10 μM BT in PBS. (C) Quantitative analysis of LDH release from A549 cells subjected to COE and light treatments (n = 3). (D) Time-lapse micrographs of A549 cells treated with 10 μM BT in PBS. During the observation intervals, cells in the gray box were additionally irradiated by the built-in 514-nm laser using microscope’s bleaching model. Scale bars, 50 μm. (E) Zoom-out micrographs after the bleaching experiments in (D). Scale bars, 100 μm. (F) Relative emission intensity at 650 nm for LUVs (POPC:POPG, 85:15) treated by 10 μM BT in PBS with or without additional 488-nm excitation during detection intervals. I/I₀, fluorescence intensity ratio (650 nm) after/before light exposure.

Confocal microscopy was used to probe any mechanical disruption of the cell membranes (see movies S1 and S2). After adding 20 μM BT, micrographs of A549 cells in PBS were captured every 10 s. During the observation intervals, cells in the observation window (marked with a gray box in Fig. 3D) were continuously exposed using the microscope’s bleaching model and a built-in 514-nm laser (~3 mW, measured using a laser power meter), resulting in higher irradiation doses for these cells compared to the surrounding area. At 100 s, fluorescence patterns indicating membrane intercalation of BT molecules began to appear in the photobleaching region. As exposure continued, the cells underwent a series of morphological changes, including cellular swelling and membrane blebbing, characteristic of pyroptosis (38, 39). This accumulated cellular stress compromised plasma membrane integrity, leading to cell rupture and release of intracellular contents by 300 s. Subsequently, cells in the remaining observation area, which experienced relatively reduced irradiation dose, initiated similar membrane disruptions, resulting in cell lysis. In contrast, cells without BT treatment remained intact under the same laser exposure for 30 min.

After performing bleaching experiments, we switched the objective lens from 40× to 10× to investigate the cells outside the observation area in the BT-treated group (Fig. 3E). Bright fluorescence was recorded for cells within the previous observation region, highlighting the excellent photostability of membrane-intercalated BT. In contrast, cells outside this region remained intact and poorly labeled, indicating that the membrane partitioning of BT was driven by light exposure. Further biophysical experiments using LUVs were conducted to measure emission intensity kinetics after BT treatment with a fluorometer (Fig. 3F). Under continuous 488-nm light exposure, the fluorescence intensity increased by more than 100% over 30 min due to the progressive membrane intercalation of BT molecules. When the exposure was turned off during detection intervals, the fluorescence increase was reduced to ~50%. These results demonstrate that the membrane-intercalation rate of BT is highly light responsive, facilitating controlled activation for selective cancer treatment.

The effects of BT on membrane permeability were assessed in A549 cells using 4′,6-diamidino-2-phenylindole (DAPI), a blue fluorescent DNA stain that is impermeable to intact cell membranes but penetrates compromised ones (Fig. 4) (20). After a 6-hour incubation, the membrane-mimicking BT appeared to integrate into lipid trafficking pathways, distributing across organelles and cellular membranes. Under 20 min of white light irradiation, a progressively increasing DAPI fluorescence was observed, reaching levels 20 times higher than that in BT-treated cells without exposure. Flow cytometry analysis indicated that the DAPI-positive cell subpopulation increased from 1 to 56% after both BT treatment and light exposure (fig. S21) that light-activated BT under light-activation compromised cell membrane permeability. Similar membrane permeabilization assays were performed for Ben and S6 using PI, a red fluorescent, cell-impermeant dye, to avoid spectral overlap with COEs. Under the same white light irradiation, notable PI fluorescence was detected in the nuclei of cells preincubated with Ben (Fig. 4 and fig. S22). In contrast, no noticeable membrane permeabilization was observed in the S6 group, despite S6’s higher ROS production capability. To further investigate the role of ROS, we supplemented the BT-treated cells with ROS scavengers before light exposure, including 10 mM N-acetylcysteine and 5 mM vitamin C (see their scavenging effects in fig. S23) (40, 41). Compared to that in controls, no substantial difference was observed in the DAPI-positive cell subpopulation as analyzed by flow cytometry (Fig. 5A). These results indicate that the compromised membrane permeability is attributed to the mechanical disruption caused by COEs rather than ROS effects.

Fig. 4. Membrane permeabilization analysis of COE-treated A549 cells.

Confocal images of COE-treated A549 cells stained by cell-impermeant DNA dye. (A and B) Cells treated with 10 μM BT and (A) exposed to white light irradiation or (B) maintained in the dark, then stained with DAPI. (C) Ben-treated or (D) S6-treated cells following white light irradiation, then stained with PI. Relative to the Dark group, cells in the Light group were exposed to white light at 20 mW cm⁻² during observation intervals. Scale bars, 50 μm. Time-dependent fluorescence intensities of DNA dye were presented in the right panel with three cells statistically analyzed.

Fig. 5. Mechanistic investigation of light-triggered mechanical membrane disruption.

(A) Dead cell subpopulation evaluated using flow cytometry for BT-treated A549 cells supplemented by ROS scavengers, including N-acetylcysteine (NAC) or vitamin C (ViC). (B) Calcein leakage experiments for COE-treated LUVs under white light exposure. FI_t/FI_t0, fluorescence intensity ratio (515 nm) after/before light exposure. (C) Membrane fluidity assessed by measuring the generalized polarization (GP) value using Laurdan indicator for BT-treated LUVs with or without white light irradiation at 20 mW cm⁻² (n = 3). LUVs without COE treatment were used as control. (D) Schematic representation of COE-containing multilamellar lipid sample fabricated by drop casting on silicon wafer for XRD characterizations. (E and F) XRD curves of untreated (E) or BT-containing (F) multilamellar lipid sample subjected to white light irradiation for 20 min. (G and H) Evolution of femtosecond transient absorption spectra for BT-containing LUVs following 465-nm excitation. (I) Selected kinetic traces and corresponding fits at 860 and 890 nm. ΔA (mOD), difference in optical density before and after pumping.

Mechanistic studies using lipid models

Biophysical experiments using artificial membranes were conducted to investigate the mechanical effects of COEs on lipid bilayers. The integrity of lipid membranes in the LUV model was assessed using calcein leakage assays (42), in which calcein encapsulated within liposomes remains in a fluorescence-quenched state but fluoresces upon release into diluted solution after leakage (Fig. 5B). Under the same white light irradiation, Ben-containing LUVs exhibited substantial calcein leakage, whereas LUVs treated with S6 showed calcein fluorescence levels similar to the control (i.e., without COE treatment). Consistent with the mammalian cell phototoxicity results, BT induced the most pronounced calcein leakage due to its potent mechanical disruption compromising membrane integrity.

We further assessed changes in lipid bilayer fluidity using Laurdan, a fluorescent dye sensitive to membrane phase transitions (43), which exhibits emission maxima at 440 nm in the gel phase and 490 nm in the liquid phase (Fig. 5C). Due to spectral overlap between the blue-fluorescent COEs and Laurdan, only BT-treated LUVs were evaluated in this assay. Before white light irradiation, no substantial change was observed in the generalized polarization (GP) values (see calculation formula in the Supplementary Materials) compared to that in the blank LUVs. Upon light exposure, the GP values of BT-treated LUVs gradually decreased over time, indicating that the lipids became more disordered and membrane fluidity increased.

To gain insight into the light-responsive membrane disruption process, in situ x-ray diffraction (XRD) measurements were conducted (Fig. 5D). Multilamellar phospholipid membranes on silicon wafer substrates were fabricated by drop-casting lipid solutions in organic solvents (44). The diffraction peak at q = 0.18 Å⁻¹ corresponds to a bilayer thickness of 3.5 nm, while the peak at q = 0.52 Å⁻¹ represents an interlamellar spacing of 1.2 nm (Fig. 5E) (44). After white light irradiation, the diffraction peak intensities increased, probably due to heat generated by the silicon substrate during exposure, which caused the as-casted lipids to transition into a more ordered phase. The BT-containing lipid film was prepared by mixing the COE and lipid solutions before drop casting. Compared to the control (without COE treatment), the BT-treated lipid film exhibited more pronounced diffraction peaks, suggesting that membrane intercalation of BT enhanced lamellar alignment (fig. S24). Additionally, differential scanning calorimetry (DSC) experiments revealed higher phase transition temperatures during both heating and cooling processes for BT-supplemented liposomes compared to controls (fig. S25) (45). These results demonstrate that BT, without light irradiation, can rigidify and stabilize bilayer structures, consistent with other elongated membrane-mimicking COEs (46). However, upon light irradiation (20 mW cm⁻²), the diffraction intensities gradually decreased with increased exposure time, and the film became fully amorphous at 20 min, with all diffraction peaks disappearing (Fig. 5F). During the membrane collapse, the lipid membrane thickness decreased from the initial 3.5 nm (q = 0.18 Å⁻¹) to 2.9 nm (q = 0.22 Å⁻¹) at 10 min, implying that light-activated BT may undergo mechanical actions during the membrane disruption process. Femtosecond transient absorption spectroscopy was used to investigate the excited-state dynamics of BT in LUV membranes. Upon 465-nm excitation, a ground-state bleach and stimulated emission band spanning 550 to 750 nm was observed, along with a broad excited-state absorption (ESA) extending beyond 750 nm (Fig. 5, G and H). Notably, the transient spectra evolved on an ultrafast timescale, with a new ESA peak emerging with a ~5-ps time constant (Fig. 5I), indicating the formation of a distinct species likely resulting from a torsional or other conformational relaxation of BT backbone in the excited state.

To visualize the impacts of molecular mechanical forces on lipid membranes, we prepared the COE-loaded LMVs (see the Supplementary Materials for details). Time-lapse fluorescence microscopy was performed by capturing micrographs every 10 s to monitor morphological changes in the LMVs (47). Fluorescence collected from the COEs outlined the membrane structures of the LMVs, indicating that the COEs were uniformly distributed within the lipid bilayers (Fig. 6). During the observation intervals, the BT-treated LMVs within the bleaching region (marked with a yellow box) were additionally irradiated using the microscope’s built-in 514-nm laser (~3 mW). This led to notable morphological alterations: the LMV rapidly divided into two independent vesicles after exposure. In contrast, the LMVs outside the bleaching region showed no noticeable shape changes. Similar bleaching experiments were conducted for LMVs treated with Ben and S6, using an excitation wavelength of 405 nm at a power of ~3 mW during the regional bleaching process. The Ben-treated LMVs in the irradiation region exhibited apparent envelope modifications within 40 s. Conversely, the S6-treated LMV showed no substantial morphological changes during the entire observation period, although S6 is more capable of generating ROS. These results clearly demonstrate the mechanical forces exerted by COEs can drive physical movements and alter the morphologies of lipid membranes.

Fig. 6. Morphology changes of COE-containing LMVs under laser irradiation.

Time-lapse micrographs of COE-containing LMVs (composed of 20 mM lipids and 20 μM COEs) in PBS. During the observation intervals, LMVs in the yellow box were additionally irradiated by the built-in laser using microscope’s bleaching model. (A) BT-containing LMVs exposure to 514-nm laser (~3 mW), (B) Ben-containing LMVs, or (C) S6-containing LMVs exposure to 405-nm laser (~3 mW). LMVs in the gray box without additional irradiation were used as control. Scale bars, 20 μm.

In vivo tumor therapy

Before antitumor animal experiments, in vitro phototoxicity of BT was assessed using A549 cancer cells (Fig. 7A). After 10 min of white light irradiation (20 mW cm⁻²), IC₅₀ substantially decreased from more than 256 to 0.6 μM for BT (fig. S26), outperforming Ben’s 1.8 μM under identical conditions (Fig. 2D). Previous time-series microscopy revealed pyroptosis characteristics, including rapid cell swelling, including rapid cell swelling, blebbing, and rupture in cells treated with BT and irradiation (Fig. 3D). To elucidate the programmed cell death pathway, a Western blot assay was performed to detect pyroptosis-specific caspase-1 activation and gasdermin D (GSDMD) cleavage, with glyceraldehyde 3-phosphate dehydrogenase (GAPDH) as the loading control (Fig. 7B). The results showed substantial up-regulation of cleaved caspase-1 and pyroptosis-inducing N-terminal fragment of GSDMD (GSDMD-N) (48). Additionally, cells subjected to both BT and irradiation treatments (the “BT + light” group in Fig. 7C) exhibited significantly increased secretion of proinflammatory cytokines such as interleukin-1β (IL-1β) and IL-18 compared to untreated controls. These findings suggest that the mechanical force generated by BT upon white light irradiation induces oxidative stress in A549 cells, activating caspase-1 (49). This activation promotes the secretion of proinflammatory cytokines, thereby triggering the pyroptosis pathway. Simultaneously, GSDMD is cleaved into its N-terminal fragments, which translocate to the cell membrane, forming pores that compromise cellular integrity and facilitate the release of intracellular contents, including IL-1β and IL-18 (50, 51). Concomitantly, annexin V–allophycocyanin (APC)/DAPI staining confirmed that BT induced apoptosis or necrosis in cells with light irradiation, suggesting the activation of multiple cell death pathways (fig. S27).

Fig. 7. Programmed cell death and in vivo therapeutic effect of light-activated BT.

(A) Viability of BT-incubated A549 cancer cells after white light irradiation at 20 mW cm⁻² for 10 min. (B) Expression levels of cleaved caspase-1 and GSDMD-N in A549 cells as assessed by Western blot analysis. (C) IL-18 and IL-1β content detected using ELISA for supernatant of BT-treated A549 cell culture after light treatment (n = 3). (D) Representative images of HMGB1 and CRT immunofluorescence staining for A549 cells after BT and irradiation treatments. Scale bar, 50 μm. (E) Tumor volume growing trend within 2 weeks after intratumoral injection of BT, followed by light treatment (n = 4). (F) Photographs of tumors collected from mice after 2 weeks of treatment. (G) Hematoxylin and eosin (H&E)–staining analysis of tumor sections after BT and irradiation treatments. (H) Representative confocal images of HMGB1, CRT, and CD8 levels in 4T1 tumor sections. Scale bars, 50 μm. (I and J) TFN-α and INF-α levels in serum after BT treatment as detected by ELISA (n = 3). Data are presented as means ± SD. *P < 0.05, **P < 0.01, and ***P < 0.001 were determined using one-sample t tests. TNF-α, tumor necrosis factor–α; IFN-γ, interferon-γ.

To explore the effects of light-responsive BT on immune activation, we further assessed the pyroptosis-mediated release of damage-associated molecular patterns (DAMPs) (52), specifically the chaperone protein calreticulin (CRT) and the inflammatory mediator high mobility group box 1 (HMGB1). As illustrated in Fig. 7D, CRT accumulated to cell surface in the BT + light group, enhancing the recognition and clearance of dying cells by the immune system. Additionally, HMGB1, initially localized in the nucleus, was released into the extracellular space following BT and irradiation treatments. Similar expressions of CRT and HMGB1 were still observed after supplementation with the ROS scavenger vitamin C (fig. S28). In addition, the adenosine triphosphate level in A549 cells after BT treatment and light exposure was significantly lower than that in the PBS group (fig. S29) (53). These findings suggest that the immunogenic cell death (ICD) process was likely induced by the mechanical force of BT on the membrane under irradiation, rather than typical ROS, thereby initiating an adaptive immune response driven by pyroptosis.

To investigate the in vivo therapeutic efficacy of BT, we established a subcutaneous tumor model in BALB/c mice using the murine-derived 4T1 cell line. In vitro cell viability assays indicated that BT under light irradiation exhibits similar and potent anticancer performance against 4T1 cells (fig. S30). Following intratumoral injection of BT solution in PBS (100 μl, 100 μg ml⁻¹, or 48 μM), in vivo fluorescence imaging over 24 hours suggested that the COE molecules stably retain within the tumor tissue (fig. S31). In therapeutic experiments, the tumor sites were irradiated with white light at 80 mW cm⁻² for 10 min at 2 hours post–BT injection (Fig. 7E). Compared to the controls, 4T1 tumor growth was effectively inhibited, in the BT + light group with declined tumor volumes (Fig. 7F). Hematoxylin and eosin (H&E) staining revealed notable changes in cell morphology with nuclear condensation in tumor sections (Fig. 7G). We further assessed the photothermal effect in vivo, in addition to previous in vitro ROS scavenging experiments (Fig. 5A). No substantial temperature elevation was observed in the tumor, either with or without BT injection, supporting the conclusion that the efficacious antitumor outcome is primarily attributed to the mechanotherapy of BT (fig. S32).

Furthermore, immunofluorescence staining of tumor sections revealed substantial CRT exposure on cell surface and extracellular release of HMGB1 in the BT + light group (Fig. 7H), suggesting that light-activated BT induced pyroptosis and promoted ICD process (54, 55). Moreover, substantially increased tumor infiltration of CD8⁺ T cells was observed in the BT + light group compared to that in the PBS control. Serum levels of immune-related inflammatory cytokines, including tumor necrosis factor–α and interferon-γ, were also substantially elevated (Fig. 7, I and J), confirming the activation of a systemic antitumor immune response induced by pyroptosis (56, 57). These findings demonstrate that BT induces an ICD-mediated antitumor immune response in vivo, highlighting the potential of molecular mechanotherapy using COEs for cancer therapy.

The in vivo toxicity of BT was evaluated by monitoring the body weights of mice throughout the experimental period, with negligible changes observed (fig. S33). Additionally, no obvious cellular damage was detected in major organs, including the heart, liver, spleen, lungs, and kidneys, using H&E staining (fig. S34). Routine blood tests showed no obvious alterations in blood cells (fig. S35). Meanwhile, blood biochemical analyses demonstrated no substantial hepatic or renal toxicity compared to untreated controls, as evidenced by representative markers of liver and kidney function (fig. S36). These results suggest that BT has good biocompatibility.

DISCUSSION

In this study, we show that membrane-targeting COEs can behave as mechanotherapeutic agents for cancer therapy. The best performing COE molecule, BT, demonstrated good biocompatibility with cell membranes, effectively stabilizing bilayer structure as evidenced by XRD and DSC characterizations. Upon low-dose light exposure (20 mW cm⁻² for 10 min), the IC₅₀ for BT-treated A549 cancer cells substantially decreased from over 256 to 0.6 μM. This was visually confirmed through confocal microscopy, which revealed that BT molecules intercalate into cancer cell membranes and rapidly induce cell expansion and rupture when activated by a built-in 514-nm laser (~3 mW). Biophysical experiments show that certain COEs can induce physical movement of LMV membranes under light irradiation, generating mechanical actions within lipid bilayers. Additional studies indicated that the mechanical stress generated by BT on the cell membrane after photoactivation could activate cell pyroptosis, thereby inducing the immune response and enhancing therapeutic efficacy in antitumor animal models. In situ XRD characterizations revealed a process of membrane thinning followed by collapse upon light exposure. Femtosecond transient spectrum evolution of BT further indicated that new excited-state species were formed, probably due to the molecular conformation changes caused by light activation. Given its reliance on physical membrane disruption, this mechanism is expected to be applicable to bacterial membranes, offering a potential strategy for antimicrobial resistance.

The studies of two additional COEs (Ben and S6) with similar spectral properties but different backbone architectures are in line with a mechnotherapeutic effect, rather than conventional photodynamic or photothermal effects contributed to the anticancer efficacy. The bent backbone structure of Ben was more favorable for mechanotherapeutic outcomes compared to the linear structure of S6. We hypothesized that absorption of light induces more pronounced torsional or conformational changes in bent COE molecules that generate greater mechanical stress in the membranes than their linear counterparts and thereby enhance structural disruption. COEs with bent backbones exhibited difficulty in spontaneously partitioning into cell membranes, requiring assistance of light irradiation, providing a promising strategy for controlled and selective cancer treatment. These newly discovered light-responsive molecular “machines,” with their distinctive membrane-mimicking structures, offer potential for diverse biological applications, such as controlling transmembrane transport for on-demand drug release and modulating ion channels. Future advances in rational molecular design enabling activation by near-infrared (NIR) light could further enhance tissue penetration and broaden the therapeutic scope of COE-based mechanotherapeutics to deeper pathological sites. Furthermore, integration of molecular mechanotherapy with complementary modalities such as photodynamic or PTT holds promise for synergistically improving antitumor efficacy and addressing therapeutic resistance.

MATERIALS AND METHODS

Cellular bleaching imaging experiment

A549 cells were seeded in 35-mm confocal dishes (BioSharp) at a density of 6000 cells per dish and cultured for 24 hours to ensure adequate cell adhesion. Following this, the cells were subjected to different treatments: In the BT group, the medium was replaced with PBS containing 20 μM BT; and in the control group, the medium was replaced with neat PBS. The cells were then observed using a confocal laser scanning microscope (CLSM, LSM 780) with “time series” and “bleaching” models selected. The excitation wavelength for BT was set at 514 nm, with emission collected in the range of 600 to 700 nm. In the bleaching region, the bleaching interactions were set as 20, and the time series were 200 based on the built-in 514-nm laser. As a result, the total experimental duration was 31 min and 41 s. The laser power intensity was measured using a Laser Power Meter (CNI Laser, TP100, China) to be ~3 mW for microscope built-in 514-nm laser.

DAMPs of A549 tumor cells

HMGB1 release into the nucleus and CRT exposure on the membrane were observed using CLSM. Specifically, A549 cells were first cultured in a confocal dish, and, after 12 hours of incubation, BT water solution (2.5 μg ml⁻¹) diluted with DMEM complete medium was added to the confocal dish. After coincubation for 6 hours, the culture medium was removed, and the cells were washed with PBS buffer, followed by the addition of fresh complete medium. The cells were then irradiated with white light (20 mW cm⁻²) for 10 min. One hour later, the cells were washed with PBS buffer and fixed with 4% paraformaldehyde for 20 min. Subsequently, the cells were permeabilized with 0.1% Triton X-100 (Beyotime Biotechnology Co. Ltd.) for 30 min and then blocked with 4% Paraformaldehyde Fix Solution (Beyotime Biotechnology Co. Ltd.) for 20 min. The primary antibody HMGB1 or CRT was then added, and the cells were incubated overnight at 4°C. The cells were then incubated with Alexa Fluor 488–labeled anti-rabbit secondary antibody at room temperature for 2 hours. After washing with PBS buffer, Hoechst 33342 was added for nuclear staining. Last, all samples were visualized using CLSM (for Alexa Fluor 488–labeled HMGB1 and CRT: λ_exc = 488 nm, emission collected from 500 to 530 nm; and for Hoechst 33342: λ_exc = 405 nm, emission collected from 410 to 510 nm). The following antibodies were used: HMGB1 (Abcam, ab18256) and CRT (Abcam, ab32063).

In vivo antitumor assay

Five-week-old female BALB/c mice were obtained from the Animal Center of South China Agricultural University and housed in a specific pathogen–free laboratory. All animal experiments were conducted in accordance with the approved protocol from the local ethics committee and the regulations of the Institutional Animal Care and Use Committee of South China Agricultural University (approval number SCAU-2023D062). A mouse tumor model was established by subcutaneously injecting a suspension of 4T1 breast cancer cells (2 × 10⁶ cells in 100 μl) into the right flank of each mouse. When the tumor volume reached ~60 to 70 mm³, the mice were randomly divided into four groups (four mice per group): PBS, PBS + light, BT, and BT + light. The concentration of injected BT was 0.1 mg/ml with an administered volume of 100 μl, and the white light was administered at an intensity of 80 mW cm⁻² for 10 min. The weight of the mice and the tumor volume were recorded every 2 days following treatment, with tumor volume calculated using the formula: Volume = (length × width²)/2. After 14 days, the mice were euthanized, and the tumors were excised, photographed, weighed, and fixed using paraformaldehyde for subsequent H&E-staining experiments.

Supplementary Materials

The PDF file includes:

Supplementary Text

Figs. S1 to S36

Legends for movies S1 and S2

Other Supplementary Material for this manuscript includes the following:

Movies S1 and S2