Oxidative photocatalysis on membranes triggers non-canonical pyroptosis

The intracellular membrane in eukaryotes is a vital platform for organelle interactions, signalling, metabolic reactions, and biosynthesis, with functions mediated by membrane proteins and lipids that are susceptible to reactive oxygen species. Oxidative damage to intracellular membranes disrupts organellar function and can trigger programmed cell death implicated in diverse diseases. While lipid peroxidation via hydroxyl radicals and ferroptosis has been studied, cellular responses to intracellular membrane-localised ROS generation and oxidative stress directed toward membrane proteins remain unclear. Photocatalysis offers spatiotemporal control over oxidative stress, and organelle-targeted ROS generators have been used; however, no photocatalyst had been developed to simultaneously localise across intracellular membranes (ER, Golgi, mitochondria, vesicles, nucleus), limiting the study of membrane protein oxidation. The authors aimed to develop an intracellular membrane-localised photocatalyst to induce oxidative damage on membrane proteins and elucidate the resultant cell death signalling, hypothesizing that intracellular membrane-focused oxidation could trigger non-canonical pyroptosis.

Prior work has established that intracellular membranes and their proteins/lipids are vulnerable to ROS, with lipid peroxidation linked to ferroptosis via hydroxyl radicals. Photocatalytic tools have been used to generate ROS within specific organelles to probe cell death pathways. Nonetheless, a system capable of broadly targeting intracellular membranes had not been realized, presenting a gap in understanding how membrane protein oxidation shapes cellular responses and death signalling. This study builds on advances in photoredox catalysis and organelle-targeted photosensitisers to investigate endotoxin-independent triggers for pyroptosis.

• Photocatalyst design and photophysics: Synthesized an amphiphilic organic photocatalyst (BTP) composed of benzothiadiazole (acceptor) and triphenylamine (donor), promoting charge separation. Measured ground and excited-state redox potentials via cyclic voltammetry and optical band gap. Estimated E* redox potentials indicated that under 450 nm irradiation, BTP undergoes reductive quenching by water to generate H2O2 and •OH and oxidative quenching by O2 to generate O2•− and oxidize amino acids.
• ROS generation assays: Fluorescence quenching of BTP by water in acetonitrile mixtures assessed PET from H2O to BTP*. H2O2 production quantified with DPD/HRP in PBS under blue LED (λmax 450 nm, 66.7 mW·cm−2) up to 150 min under normoxia vs hypoxia (Ar-bubbled); control in DMSO. Hydroxyl radical detection by hydroxyphenyl fluorescein (HPF) fluorescence under various conditions (including Ar-bubbled, normoxia, DMF, and with added H2O2) after blue light dose (2 J·cm−2). EPR spectroscopy with BMPO (10 mM) spin trapping confirmed BMPO–OH adduct upon BTP + H2O2 irradiation, with Fenton reaction as positive control.
• Cellular localisation and membrane proximity: Assessed uptake of BTP in HeLa cells at 37 °C vs 4 °C and with NaN3 to infer energy-dependent internalisation. Colocalisation with organelle markers (ER, Golgi, mitochondria) by confocal and structured illumination microscopy (SIM) before and after photocatalysis. Determined membrane proximity via emission peak shifts in aqueous vs bicelle (lipid bilayer) systems and in cells by lambda scans.
• Cellular ROS and lipid oxidation: Monitored intracellular ROS with H2DCF-DA and O2•− with dihydroethidium following BTP photocatalysis. Lipid oxidation profiled by extracting cellular lipids and analysing by UPLC-MS; monitored changes in chromatograms and oxidation of 15:0-18:1 phosphatidylcholine.
• Membrane protein stability assays: In vitro assays on E. coli rhomboid protease GlpG (6 TM helices). SDS-PAGE to quantify loss of monomer band and aggregation after BTP + blue light (λpeak 450 nm, 30 J·cm−2; BTP 100–150 μM). Thermal denaturation profiling (10 min at 30–89.9 °C) post-photocatalysis. Single-molecule magnetic tweezers with GlpG reconstituted in bicelles; repeated force ramps (1–50 pN) under blue light (450 nm, 9.16 mW·cm−2) with/without BTP (10–20 μM) to measure unfolding forces and irreversibility at critical cycle C0; infrared light control (850 nm).
• Proteomics of oxidised proteins: Label-free quantitative LC–MS/MS on HeLa cells across four conditions (BTP±, hv±; 450 nm, 10 J·cm−2; BTP 4 μM) after on-filter digestion and high-pH fractionation. Two-stage database searches: initial Sequest/Scaffold for protein IDs with O-Met as variable mod; secondary MODplus search of previously unidentified spectra for extensive oxidative modifications across 17 amino acids (FPOP-type). Quantified precursor intensities of oxidative modifications, computed fold changes vs combined controls, and applied criteria (O-Met screen: p<0.05, FC>2; multi-AA screen: p<0.01, FC>4). Classified proteins by GO cellular location (membrane-specific, membrane-cytosolic, cytosolic) and biological process; built functional networks for PQC-related categories.
• Cation mobilisation and organelle stress: Monitored mitochondrial Ca2+ with Rhod-2 and MitoTracker during time-lapse confocal imaging under light (445 nm, 0.3 mW), and by flow cytometry pre/post irradiation (450 nm, 10 J·cm−2). Assessed mitochondrial morphology by live-SIM, and membrane potential via TMRE fluorescence after BTP (10 μM) and light (450 nm, 10 J·cm−2). Measured intracellular K+ efflux using ION K Green-2 by flow cytometry.
• Cell death assays and pyroptosis markers: Viability by Calcein AM/PI staining and MTT in HeLa; MTT in Panc-1 and MiaPaca-2 under normoxia/hypoxia. LDH release assays comparing BTP photocatalysis to LPS+nigericin and Ce6+hv controls; inhibitor studies with z-VAD-fmk (pan-caspase) and Liproxstatin-1 (ferroptosis inhibitor). Western blots for GSDMD cleavage (NT fragment) in cells and media, caspase-1/3/4/5 processing, and IL-1β/IL-18 secretion; ATP release quantified. Validated dependency on GSDMD using WT vs GSDMD−/− iBMDMs for morphology and LDH release.

• BTP photoredox cycles: Excited BTP mediates water oxidation (E*− ~1.47 V) to H2O2 and •OH and oxygen reduction (E*+ ~−1.36 V) to O2•− and amino acid oxidation. BTP fluorescence quenched with increasing water content; excited-state lifetime decreased, supporting reductive quenching by water.
• ROS generation: H2O2 produced from water upon blue light irradiation; generation accelerated under hypoxia (oxygen competes as a quencher). HPF assay confirmed •OH production, enhanced under hypoxia and with H2O2; diminished in non-aqueous DMF. EPR detected BMPO–OH adduct after BTP photocatalysis (significant signal, P=0.00049). BTP did not generate singlet oxygen.
• Membrane localisation and stress: BTP uptake is energy-dependent; localises to ER and Golgi initially and relocalises post-photocatalysis to mitochondria and plasma membrane. Emission spectra shifts indicate proximity to lipid bilayers in cells. Cellular assays (DCF, DHE) show increased ROS and O2•−. Lipid oxidation was minimal: no new UPLC-MS peaks and no change in 15:0-18:1 PC oxidation, suggesting limited diffusion of polar •OH into lipid interior.
• Membrane protein destabilisation: In vitro GlpG assays showed decreased monomer band to 31 ± 9% at 150 μM BTP with light and increased aggregates; thermal denaturation midpoint dropped from 75 °C (no BTP) to 35 °C (150 μM BTP). Single-molecule tweezers revealed irreversible loss of unfolding forces to <15 pN at a critical cycle (C0) after BTP + blue light, not recoverable by fresh bicelles, indicating irreversible oxidative damage to the native fold.
• Proteomics: O-Met screen identified predominantly membrane-localised oxidation (339 membrane-specific vs 40 cytosolic proteins; 24.5% vs 3.6%). Organellar distribution of oxidised membrane proteins: mitochondria 31.1%, ER 19.1%, nucleus 21.1%, Golgi 7.7%. Multi-AA oxidation search (p<0.01, FC>4) confirmed highest oxidation in membrane-specific proteins across residues, notably Trp and His. A conservative set of 250 oxidised membrane proteins was enriched in ER, mitochondria, nucleus, and Golgi; 58.1% mapped to PQC-related organelles (ER, GA, mitochondria). Ninety-seven proteins clustered into PQC networks: UPR/ERAD, ER–Golgi transport, mitochondrial trafficking/transport, and lipid metabolism; GO enrichment indicated impacts on ER and mitochondrial UPR.
• Cation mobilisation and organelle dysfunction: Mitochondrial Ca2+ increased markedly during irradiation (Rhod-2), with matrix swelling and dynamic fission/fusion; TMRE signal decreased, indicating loss of membrane potential. Intracellular K+ efflux observed post-photocatalysis.
• Cytotoxicity and pyroptosis: Nearly complete HeLa cell death within 24 h after BTP photocatalysis; enhanced cytotoxicity in hypoxic Panc-1 and MiaPaca-2 consistent with increased •OH under hypoxia. Lytic morphology with plasma membrane swelling/blebbing; robust LDH release exceeding LPS+nigericin and Ce6+hv controls. z-VAD-fmk reduced LDH release, whereas Liproxstatin-1 had limited effect, distinguishing from ferroptosis. GSDMD-NT increased in cells and media; GSDMD knockout abrogated morphology changes and LDH release. Caspase-4/5 cleavage observed, with minimal caspase-1 activation; ATP release increased; little cleaved IL-1β/IL-18 detected, consistent with lack of active caspase-1. Together, data support non-canonical inflammasome activation (caspase-4/5) leading to GSDMD-mediated pyroptosis.

The study demonstrates that directing oxidative photocatalysis to intracellular membranes preferentially oxidises membrane proteins, destabilising their folds and perturbing protein quality control within ER, Golgi, and mitochondria. The resulting maladaptive UPR, mitochondrial dysfunction, and Ca2+ influx/K+ efflux create a cellular milieu that activates non-canonical inflammasome caspases-4/5, leading to GSDMD cleavage, pore formation, ionic dysregulation, and lytic cell death consistent with pyroptosis. This establishes an endotoxin-independent route to pyroptosis driven by membrane protein oxidation rather than lipid peroxidation or canonical inflammasome activation. The findings clarify how spatially confined ROS at membranes can reprogram death signalling toward pyroptosis and highlight hypoxia-enhanced photocatalytic efficacy, relevant to tumour microenvironments.

An amphiphilic, membrane-localised organic photocatalyst (BTP) was developed that, upon visible-light activation, oxidises water to generate H2O2 and •OH even under hypoxia and selectively oxidises membrane proteins. Single-molecule and bulk assays show irreversible destabilisation of membrane protein folds. Proteomics reveals widespread oxidation of PQC-related membrane proteins across ER, Golgi, and mitochondria, culminating in maladaptive stress responses, cation flux, activation of caspase-4/5, GSDMD cleavage, and non-canonical pyroptosis. These results introduce intracellular membrane-focused oxidation as a controllable, endotoxin-independent trigger of pyroptosis, offering a potential strategy for spatiotemporal immune activation in hypoxic tumours. Future work should delineate precise causal links between specific oxidised targets and caspase-4/5 activation, evaluate in vivo efficacy and safety, and optimise photocatalyst delivery and light dosing for therapeutic applications.

The study primarily uses in vitro and cell culture systems without in vivo validation. While proteomics and biophysical data indicate widespread membrane protein oxidation and folding destabilisation, the direct molecular pathway linking specific oxidised proteins to caspase-4/5 activation remains correlative. Canonical inflammasome activation was minimal, and cleaved IL-1β/IL-18 secretion was not efficiently detected, limiting assessment of inflammatory cytokine outputs. Lipid peroxidation appeared limited under the experimental conditions, but this may vary across cellular contexts. Broader generalisability across cell types and detailed causal relationships among ER stress, ion fluxes, and caspase-4/5 activation require further investigation.