IntroductionThe study of cell death has evolved conceptually over time. Initially, it began with morphological descriptions of necrosis and apoptosis.1,2 Later, molecular analyses of regulated pathways such as necroptosis and pyroptosis were added.3,4 Recently, the focus has shifted towards mechanisms driven by metabolic reprogramming and metal homeostasis regulatory networks, including ferroptosis and cuproptosis, as well as other forms of cell death.5 Each conceptual breakthrough, frequently proven by a pivotal chemical probe (e.g., Nec-1 or Erastin) or transformative technological innovation, has progressively refined the ontological framework of regulated cell death.5,6 This expanding landscape continues to provide novel therapeutic targets and fundamentally reshape our understanding of cellular fate. From the perspectives of pharmacology and medicinal chemistry, the expanding molecular map of RCD is a therapeutic gold mine. These pathways often rely on druggable nodal proteins such as kinases, proteases, or metabolic enzymes. The clinical success of venetoclax, a B-cell lymphoma 2 (BCL-2) inhibitor, alongside the intensive development of RIPK1 or NLRP3 inhibitors, underscores the transformative potential of modulating cell death in disease treatment.3,7,8 However, extensive work is needed to transform this potential into safe and effective therapies.The apoptotic pathway is currently the most mature field for clinical translation, with drugs already approved for market and several new-generation drugs entering the clinical stage (see “Targeting apoptosis” section). These include IAP antagonists, BCL-2 inhibitors, and BH3 mimetics. RIPK1 and RIPK3 are key regulatory kinases in the necroptosis pathway, and their inhibitors have therapeutic potential in inflammatory diseases. To date, several RIPK1 kinase inhibitors have entered clinical trials (see “Clinical Trial Progress Targeting Necroptosis” section), with indications such as autoimmune diseases (psoriasis, ulcerative colitis, rheumatoid arthritis and inflammatory bowel disease (IBD)), infection-related inflammatory responses (severe COVID-19), haematological diseases (haemophagocytic lymphohistiocytosis (HLH), acute graft-versus-host disease (aGVHD)), and neurodegenerative diseases (Alzheimer’s disease (AD), amyotrophic lateral sclerosis (ALS)). However, no RIPK1 inhibitor has successfully entered the market, let alone any of the less studied RIPK3 and MLKL inhibitors. Ferroptosis inhibitors are emerging compounds that have developed rapidly in recent years, especially in the context of AI-assisted drug screening. Ferroptosis inhibitors have shown clinical potential in treating neurodegenerative diseases and oral mucositis related to oxidative stress (see “Clinical trial progress targeting metal-dependent cell death” section). However, there is as yet no drug with a “ferroptosis inhibitor” as its officially approved treatment indication. Cuproptosis, which is a recently discovered form of cell death, likewise lacks nonspecific “cuproptosis inhibitors” entering the clinical stage. Clinical translation has focused more on how to overcome the resistance of solid tumours to copper death. Therefore, at the clinical research level, progress has involved mainly iron death inducers and copper death inducers (for killing tumours) rather than inhibitors (see “Clinical trial progress targeting metal-dependent cell death” section). The targets of these cell death pathways hold considerable therapeutic promise. However, clinical research on drugs that target a single pathway at a specific molecular node has progressed slowly. This slow progress highlights a critical barrier to clinical translation. To convert this potential into safe and effective therapies, a deep and integrated understanding of the complex crosstalk among different cell death pathways, as well as their context-dependent functional roles in various diseases, is urgently needed.The current focus has shifted towards the systematic dissection of dynamic interaction networks among cell death pathways and the development of targeted combination therapies. This trend is largely driven by integrated advances in chemical biology tools, CRISPR-based functional genomic screening, and multiomics analysis technologies. The evolution of the field reflects not only a fundamental transformation in research paradigms but also its central role in advancing therapeutic innovation under the framework of precision medicine. For instance, in oncology, integrative metabolomic and epigenomic analyses have revealed that ferroptosis inducers and immune checkpoint inhibitors can act synergistically by modulating lipid peroxidation levels and T-cell function within the tumour microenvironment.9 In neurodegenerative disease, the combination of single-cell transcriptomics and live-cell imaging has revealed dynamic crosstalk between excitotoxicity pathways and mitochondrial quality control networks.10 The application of cryo-electron microscopy (EM), spatial transcriptomics, and artificial intelligence-based predictive modelling is expected to shift the focus further from static pathway characterisation to real-time analysis of dynamic regulatory networks.In this review, we aim to systematically bridge this gap between fundamental biology and clinical translation. We first delineate the core molecular mechanisms of the major RCD pathways. We will subsequently analyse their established and emerging roles in disease pathogenesis. This discussion focuses on the translational pipeline, critically evaluating current pharmacological strategies, from approved drugs to promising clinical candidates that target these pathways. Finally, we explore future frontiers, including leveraging pathway crosstalk for combination therapies, overcoming drug resistance, and employing advanced delivery systems for precision targeting. Overall, this review provides a comprehensive roadmap for harnessing the power of cell death regulation in modern medicine.Timeline of the discovery of multiple forms of RCDThe understanding of cell death has evolved through distinct historical phases, each marked by conceptual and technological breakthroughs (Fig. 1). The field originated in the 19th century with morphological descriptions of passive necrosis.11 A pivotal milestone was reached in 1972 when Kerr, Wyllie, and Currie formally defined apoptosis, thereby establishing the fundamental paradigm of RCD.12 The 1990s ushered in the molecular era, facilitating the identification of key regulatory molecules, including caspases-3/8/9, antiapoptotic members of the BCL-2 family, and proapoptotic proteins such as Bax/Bak, as promising therapeutic targets.13,14 The concept of necroptosis was established as a regulated form of necrosis following the 2005 identification of necrostatin-1 as a specific inhibitor.3 Pyroptosis was first named in 2001 and later underwent substantial mechanistic clarification,15 culminating in the 2015 identification of gasdermin D as its executioner protein.4 These advances were complemented by the recognition of the NLRP3 inflammasome as a master regulator, which opened new therapeutic avenues for targeting inflammation-driven pathologies.16 A conceptual transition towards metabolism- and ion homeostasis–regulated cell death pathways was initiated in 2012 with the establishment of ferroptosis17 as a distinct entity, followed by the mechanistic elucidation of cuproptosis in 2022.18 This phase is marked by drug repurposing, including the use of sulfasalazine for ferroptosis induction19 and elesclomol for cuproptosis modulation,20 alongside the development of foundational chemical probes such as erastin and RSL3.21 These findings have inspired efforts to expand investigations to encompass additional metal- and ion-associated cell death modalities. The concept of zinc-induced cytotoxicity gained significant attention in 2020, with subsequent studies identifying lysosomal membrane permeabilisation (lysozincrosis) as its defining characteristic and recognising disulfiram as a functional zinc ionophore for experimental interrogation.22 Recently, sodium overload-driven cell death has been reconceptualised beyond mere ionic disruption and identified as a mechanistically defined process involving specific ion channels (e.g., ENaC and ASIC1) and mitochondrial impairment.23 In 2025, Fu et al. coined the term NECSO for this mode of cell death and established it as an RCD pathway triggered by necrocide 1 (NC1) via TRPM4‑dependent sodium influx.24Fig. 1Full size imageTimeline of the discovery of multiple forms of RCDs. The key findings, landmark studies, and core molecular mechanisms according to each modality are indicatedElucidation of the early mechanisms governing regulated cell death laid a critical foundation for subsequent discoveries. Autophagy and lysosomal cell death emerged as well-defined regulated cell death modalities in the early 21st century,25 and their core molecular machinery was extensively characterised throughout the 2010s. These advances thus provide an essential framework for understanding organelle-mediated cell death. Moreover, entosis, a unique “cell-in-cell” (CIC) invasive death process, is regulated through Rho and Rho-associated kinase (ROCK)-mediated mechanisms.26,27 The concept of “Parthanatos”, which involves PARP-1 hyperactivation and AIF-driven chromatin degradation, was proposed in 2009.28,29 In parallel, the discovery landscape of novel regulated cell death modalities has been actively expanding. In 2017, oxeiptosis emerged as a caspase-independent, reactive oxygen species (ROS)-sensitive pathway mediated by the KEAP1, PGAM5, and AIFMI axis.30 In 2018, alkaliptosis was identified as a pH-dependent death mechanism mediated through CA9 inhibition.31 Most recently, disulfidptosis was established as a novel form of cell death driven by disulfide stress, which induces cytoskeletal collapse under glucose starvation.32 In 2025, Wang et al. described a distinct lytic cell death modality termed “mitoxyperiosis”, elicited by innate immune signalling and metabolic disruption. In this form, mitochondria undergoing BAX/BAK1/BID-dependent oxidative stress maintain prolonged plasma membrane contact, leading to local oxidative damage.33 Collectively, these discoveries highlight the ongoing diversification of regulated cell death mechanisms closely linked to metabolic, ionic, and redox alterations.Core pathways of regulated cell deathsApoptosisApoptosis involves not only the passive termination of cellular function but also a highly coordinated, energy-dependent process.12 The defining characteristics of apoptosis involve the “silent” disassembly of cells, including cell membrane blebbing, chromatin condensation, nuclear fragmentation, and the formation of apoptotic bodies.34 Early research described apoptosis as a linear signalling cascade, whereas modern systems biology has revealed it to be a complex decision-making process.35 The time required from the initiation of apoptosis to the complete disruption of cellular structure is relatively long, and the process is regulated by both the concentration of stimuli and differences in cell types. Apoptosis can be triggered by a variety of physiological and pathogen-induced stimuli, and it is initiated primarily by two converging pathways: the intrinsic (mitochondrial) pathway and the extrinsic (death receptor) pathway35 (Fig. 2).Fig. 2Full size imageCanonical intrinsic and extrinsic apoptotic signalling cascades. Apoptosis is governed by two distinct yet convergent pathways. The intrinsic (mitochondrial) pathway (left panel) is elicited by intracellular stressors, such as DNA damage, growth factor deprivation, or cellular stress, which induce the activation of proapoptotic BH3-only proteins (e.g., PUMA, BIM, BID, and BAD). These sensors antagonise antiapoptotic BCL-2 family members (BCL-2, BCL-XL, BCL-W, and MCL1), thereby liberating the effectors BAX and BAK to undergo oligomerisation. This process induces MOMP, resulting in the release of cytochrome c. Subsequently, cytochrome c associates with APAF1 and procaspase-9 to assemble the apoptosome, facilitating the activation of the initiator caspase-9. In contrast, the extrinsic pathway (right panel) commences with the binding of transmembrane death receptors (FASR, DR4/5, TNFR1) to their respective extracellular ligands (FASL, TRAIL, TNF). This interaction leads to the recruitment of adaptor proteins (FADD and TRADD) and RIPK1 to form the DISC, which drives the proteolytic activation of caspase-8. Crosstalk between the two pathways occurs via the caspase-8-mediated cleavage of BID into tBID, which is subsequently translocated to the mitochondria to amplify the death signal. Both cascades ultimately converge on the activation of the executioner caspases-3 and -7, culminating in systematic cellular dismantling. By FigdrawMechanism of intrinsic apoptosisThe pathway initiating intrinsic apoptosis acts as an intracellular sensor that responds to stimuli, including DNA damage, growth factor deprivation, oxidative stress, and endoplasmic reticulum (ER) unfolded protein stress. Centring on mitochondrial outer membrane permeabilisation (MOMP), this pathway drives the process of cellular self-destruction.36 Whether MOMP occurs is tightly regulated by the BCL-2 family of proteins, which exert mutual inhibitory effects on one another on the mitochondrial surface.37 BCL-2 is a mammalian protein that is homologous to the CED-9 gene of Caenorhabditis elegans,38 and members of this family serve as key upstream regulators of intrinsic apoptosis.The BCL-2 family consists mainly of the following proteins: (1) the proapoptotic proteins Bcl-2-associated X protein (BAX) and Bcl-2 homologous antagonist/killer (BAK); (2) the prosurvival proteins BCL-2, Bcl-2-like protein 1 (BCL-XL), myeloid cell leukaemia 1 (MCL1), Bcl-2-like protein 2 (BCL-W), and Bcl-2-related protein A1 (BFL-1)39,40,41; and (3) BH3-only proapoptotic proteins p53-upregulated modulator of apoptosis (PUMA), BH3-interacting domain death agonist (BID), phorbol-12-myristate-13-acetate-induced protein 1 (NOXA), BCL-2-associated agonist of cell death (BAD), Bcl-2-modifying factor (BMF), and Bcl-2-interacting mediator of cell death (BIM).32,42,43,44 All members of the BCL-2 family have one or more BCL-2 homology (BH) domains, which regulate the interactions between proapoptotic and prosurvival family members and thus control the process of apoptosis.38BH3-only proapoptotic proteins function as both stress sensors and initiation switches in apoptosis; however, they cannot directly induce apoptosis alone but instead trigger the intrinsic apoptotic pathway by regulating other members of the BCL-2 family. When cells are subjected to stress stimuli, BH3-only proapoptotic proteins are released via transcriptional or posttranslational regulation of the BCL-2 family. These proteins bind to prosurvival proteins and inhibit their activity, which disturbs the balance between proapoptotic and prosurvival proteins and impairs the prosurvival effect.45 This inhibition releases the key proapoptotic effectors BAX and BAK. In normal cells, BAX shuttles between the cytoplasm and the mitochondrial outer membrane (MOM), while BAK is constitutively inserted into the mitochondrial outer membrane.46 Upon activation, however, these proteins undergo conformational changes, notably the exposure of their N-terminal epitopes, and then oligomerise to form annular pores.46 Notably, some BH3-only proteins, including BIM, can also directly activate BAX and BAK to initiate the above process.45Recent cryo-EM studies have elucidated the structure of BAX oligomeric pores on the mitochondrial outer membrane (MOM), revealing that the size of their apertures is adjustable from 10 nm to more than 100 nm.47 These large outer membrane pores do not directly rupture the inner mitochondrial membrane (IMM). Instead, they enable regulated IMM herniation: the intact inner membrane bulges outwards through the MOM pores, releasing not only cytochrome c (approximately 12 kDa) at the early apoptotic stage but also mitochondrial DNA (mtDNA) at the late stage. Notably, this process represents a programmed step of apoptosis, rather than unregulated mitochondrial rupture leading to necrosis.48 These large pores increase mitochondrial membrane permeability and drive the release of proapoptotic factors. Cytochrome c and direct IAP-binding protein with low pI (DIABLO; also known as SMAC), both normally located in the mitochondrial intermembrane space, are released into the cytoplasm, where they promote the activation of downstream caspases.49,50 In the presence of deoxyadenosine triphosphate (dATP)/adenosine triphosphate (ATP), apoptotic peptidase-activating factor 1 (APAF1), a protein that interacts with cytochrome c, undergoes nucleotide-dependent conformational changes to form a heptameric wheel-like structure.51,52 APAF1 contains a caspase recruitment domain (CARD), which interacts with the CARD of procaspase-9 to facilitate the assembly of the apoptosome complex.53 This assembly subsequently recruits procaspase-9. Unlike effector caspases, procaspase-9 is activated not via proteolytic cleavage but through conformational changes induced by dimerisation within the apoptosome complex.54 Notably, X-linked inhibitor of apoptosis protein (XIAP), a key member of the inhibitor of apoptosis protein (IAP) family, is competitively bound by its natural antagonist DIABLO/SMAC. This competitive binding displaces XIAP from its interaction with caspase-9 and caspase-3/7, thereby releasing activated caspases from XIAP-mediated inhibition.55 Specifically, XIAP acts as a “brake” in apoptosis, whereas DIABLO/SMAC serves as the “key” to release this brake. A proteolytic cascade then ensues, in which caspase-9 mediates the cleavage of other downstream caspases (caspase-3 and caspase-7). These activated caspases cleave hundreds of protein substrates, ultimately disrupting cellular architecture.56Mechanism of extrinsic apoptosisExtrinsic apoptosis, also referred to as death receptor-mediated apoptosis, is a process whereby cells sense specific extracellular death signals and initiate an intracellular protease cascade through cell-surface death receptors, ultimately leading to cell death.34 This initiating signal is triggered by the binding of several death ligands to their corresponding cell-surface death receptors, such as Fas ligand (FASL) to the Fas receptor (FasR)57; tumour necrosis factor (TNF) to TNF receptors 1 and 2 (TNFR1/2)58; and TNF-related apoptosis-inducing ligand (TRAIL) to its death receptors (DR4 and DR5).59 Upon ligand–receptor binding, the adaptor molecules FAS-associated death domain (FADD) and TNFRSF1A-associated via death domain (TRADD) are recruited. FADD further recruits procaspase-8 through its death effector domain, leading to the formation of the death-inducing signalling complex (DISC), in which procaspase-8 is activated.60 Activated caspase-8 directly cleaves the precursors of caspase-3 and caspase-7 to execute the apoptotic programme.61Notably, the cellular dependence on mitochondria differs widely across cell types during extrinsic apoptosis. In thymocytes, the DISC is stably assembled and generates high levels of active caspase-8, which directly cleaves and activates the effector caspase-3.62 However, the extrinsic pathway alone is often insufficient to induce cell death in hepatocytes and most cancer cells, as DISC formation is inefficient and results in the production of only trace amounts of caspase-8.36 To achieve effective cell death, the apoptotic signal must be amplified. Caspase-8 cleaves the BH3-only protein BID into its truncated form (tBID). tBID then translocates to the mitochondria to activate BAX/BAK, thereby promoting mitochondrial outer MOMP and engaging the intrinsic apoptotic pathway.63Tumour necrosis factor receptor 1 (TNFR1), activated by TNF, also plays a critical role in the execution of extrinsic apoptosis. After the binding of TNF to TNFR1, Complex I is first formed at the cell membrane and consists of TNFR1–TRADD–TNF receptor-associated factor 2 (TRAF2)–receptor-interacting protein kinase 1 (RIPK1)–cellular inhibitor of apoptosis proteins (cIAPs). This complex primarily activates NF-κB, which upregulates antiapoptotic and prosurvival proteins, including c-FLIP and Bcl-2, thereby suppressing apoptosis.64 When prosurvival signalling is inhibited through cIAP inactivation or NF-κB pathway blockade, membrane-bound Complex I disassembles. TRADD and RIPK1 dissociate from the receptor, translocate into the cytoplasm, and assemble to form Complex IIa.65 Within Complex IIa, procaspase-8 undergoes oligomerisation and autocleavage to generate mature, active caspase-8. Active caspase-8 then directly cleaves and activates caspase-3/7, initiating the apoptotic cascade.66NecroptosisNecroptosis is a form of lytic RCD characterised by organelle swelling, plasma membrane rupture, and the release of intracellular components that can trigger an innate immune response.51,67 The activation of cell-surface receptors, including TNFR1, Fas (a death receptor, also known as CD95), DR4, DR5, and Toll-like receptors (including TRL3 and TRL4), induces necroptosis.52 RIPK1, receptor-interacting protein kinase 3 (RIPK3), and mixed lineage kinase-like protein (MLKL) serve as central mediators in the execution of necroptosis.68 In addition, necroptosis is involved in the pathogenesis and progression of various diseases, and identifying novel drugs that target the regulation of necroptosis represents a vital strategy for treating related diseases.Initiation of necroptosisDuring necroptosis, cells swell, rupture, and release substantial contents, namely, damage-associated molecular patterns (DAMPs). These endogenous biomolecules are normally present in cells but are sequestered in specific subcellular structures, such as the nucleus, mitochondria, and lysosomes, under cellular homeostasis. When necroptosis occurs, cellular integrity is disrupted. This is a key distinction from apoptosis: Apoptotic cells form apoptotic bodies that are phagocytosed, whereas necroptotic cells rupture directly.69 These endogenous molecules are then released extracellularly, triggering “danger recognition”. The released substances include (1) those derived from the nucleus, which are involved mainly in DNA/chromatin regulation, such as high-mobility group box 1 (HMGB1); (2) those derived from the cytoplasm, which mainly perform certain physiological functions in the cytoplasm, such as energy metabolism and protein folding, represented by ATP; (3) those derived from mitochondria, such as mtDNA; and (4) several other molecules, such as interleukin-1α (IL-1α) and urea.70,71DAMPs can stimulate pattern recognition receptors (PRRs). The PRR family includes Toll-like receptors (TLRs), RIG-I-like receptors (RLRs), NOD-like receptors (NLRs), and C-type lectin receptors (CLRs), which act as initial sensors of infection and mediate pathogen clearance by coordinating inflammatory responses in diseased cells.72 Frequently, identical PRRs are involved in the corecognition of DAMPs and pathogen-associated molecular patterns (PAMPs).73 This observation implies that the response triggered by pathogens resembles inflammatory reactions linked to cellular stress, injury, or death. Multiple stimuli acting on the cell surface, including TNF, FasL, TRAIL, double-stranded RNA (dsRNA), interferon-γ (IFN-γ), ATP, pathogens and Z-nucleic acid binding protein 1 (ZBP1), can induce necroptosis.74 Each inducer has distinct effects on downstream signalling pathways.TNF is a central cytokine in the inflammatory response that not only induces cellular inflammation but also regulates cell death.75 TNF predominantly exerts its biological effects via two specific receptor subtypes, TNFR1 and TNFR2. Specifically, TNFR1 contains a death domain (DD) and mediates cellular inflammation, cell survival, apoptosis, and necroptosis.76 It also functions as a molecular switch, transitioning to an alternative signalling mode upon activation.In the absence of TNF, TNFR1 located on the cell surface spontaneously forms trimers on the plasma membrane. Upon specific binding of TNF to TNFR1, TNFR1 trimers recruit various proteins, including TRADD, RIPK1, cIAP1, cIAP2, TRAF2, TRAF5 and linear ubiquitin chain assembly complex (LUBAC).64,77 Together with these recruited proteins, TNFR1 assembles into Complex I. In this process, TRADD acts primarily as a “molecular bridge”: It first binds with TNFR1 and subsequently mediates the recruitment of FADD through its DD, thereby bridging the upstream receptor (TNFR1) with downstream signalling pathways.78 Further, cIAP1/2 (core members of the cIAP family) contain three baculoviral IAP repeat (BIR) domains (the core apoptotic inhibitory motif) and a RING domain (which indicates E3 ubiquitin ligase activity).79 Upon binding to TRAF2, cIAP1/2 maintains stability by suppressing its ubiquitination. In Complex I, LUBAC also plays a vital role in the entire signalling cascade. LUBAC is composed primarily of three subunits (HOIP, HOIL-1L, and SHARPIN) that act synergistically to catalyse the formation of linear ubiquitin chains.80 In addition, previous studies have demonstrated that cIAP1/2 can directly ubiquitinate RIPK1, whereas RIPK1 is in a deubiquitinated state within the necrosome.81 However, subsequent studies have indicated that LUBAC catalyses the conjugation of Lys63-linked ubiquitin moieties to Lys377 of RIPK1. This modification directly reubiquitinates RIPK1 within the necrosome and activates the nuclear factor-κB (NF-κB) pathway.77 These results demonstrate that Complex I can modulate prosurvival and pro-death pathways by regulating its assembly components.Upon polyubiquitination, RIPK1 modulates the transition from Complex I to Complex II.68 This process is counterregulated by the deubiquitinase CYLD, which specifically removes K63-linked ubiquitin chains from RIPK1 to facilitate the formation of Complex II.82 Upon deubiquitination, RIPK1 is released from Complex I into the cytoplasm and is recruited to Complex II, also known as the DISC. This complex assembles with TRADD (liberated upon TNFR1 internalisation), FADD, and caspase-8, triggering caspase-8–mediated apoptosis.64However, the need for TRADD in Complex II assembly remains controversial. In the presence of the pan-caspase inhibitor Z-VAD.fmk, TNF-induced necroptosis is abrogated by TRADD inhibition, indicating that TRADD is essential.83 Conversely, other studies demonstrate that Complex II can still form after TRADD knockdown.84 These conflicting observations suggest the existence of compensatory mechanisms that remain to be elucidated.Downstream of Complex II, caspase-8 acts as a key molecular switch to determine cell fate.85 When caspase-8 is inhibited, whether by deletion or pharmacological inhibition, apoptosis is blocked, and cells commit to necroptosis.77 Under these conditions, RIPK1 and RIPK3 are activated, phosphorylated, and oligomerise with the downstream substrate MLKL to form the necrosome.86,87 Activated RIPK3 then phosphorylates MLKL, which translocates to the plasma membrane and executes necroptosis.88The reliance on pharmacological caspase inhibitors (such as Z-VAD.fmk) in early in vitro models to switch the mode of cell death from apoptosis to necroptosis sparked a conceptual controversy: whether necroptosis represents a bona fide physiological process in vivo or merely an experimental artefact.3 However, compelling genetic and evolutionary evidence has since established necroptosis as a genuine biological response. From an evolutionary perspective, necroptosis is currently considered a highly conserved “failsafe” host defence mechanism, which is particularly evident in the host–pathogen evolutionary arms race. For instance, to evade host defence, many viruses (murine cytomegalovirus, MCMV) encode specific inhibitors of caspase-8 to subvert apoptosis. In response, the host organism deploys the RIPK3–MLKL-dependent necroptotic pathway to rapidly eliminate infected cells and release DAMPs to mobilise innate immunity. The physiological relevance of this pathway is further substantiated by the fact that certain viruses have evolved specialised countermechanisms, such as the MCMV-encoded M45 protein (vIRA), which explicitly targets the RHIM domains of RIPK1 and RIPK3 to block necroptosis.89 These results indicate that necroptosis is not merely an artefact induced by drugs in vitro; otherwise, it does not exert selective evolutionary pressure in vivo.Furthermore, in terms of context dependence, in vivo murine models demonstrate that physiological necroptosis is tightly regulated. The embryonic lethality of caspase-8- or FADD-deficient mice is completely rescued by the codeletion of Ripk3 or Mlkl.90 These observations indicate that the fundamental role of the FADD/caspase-8 complex during normal embryonic development is to continuously suppress necroptosis. When this default suppression is naturally compromised in specific pathological contexts, such as severe ATP depletion during ischaemia‒reperfusion injury, necroptosis rapidly transitions from a dormant backup pathway to a primary driver of tissue damage. In this process, necrosis occurs and plays a functional role in the pathophysiological process of ischaemic renal injury.91 Therefore, necroptosis, exclusively as a pharmacological artefact, should be recognised as an interconnected, compensatory network designed to ensure organismal survival under fluctuating physiological stresses.In addition to TNF-induced necroptosis, necroptotic signalling can be triggered by members of the PRR family, including plasma membrane-associated TLRs and cytoplasmic NLRs.92 For example, viral or bacterial nucleotides, lipoproteins, lipopolysaccharides (LPS), or peptidoglycans, and the PRR family respond by inducing inflammation or cell death.93Execution of necroptosisRIPK1, a member of the serine/threonine kinase family, acts as a central regulator of cell survival, apoptosis, and necroptosis.94 Its core structure comprises a kinase domain and a death domain. Upon binding upstream signals transmitted by TNFR1, RIPK1 is activated and interacts with RIPK3 via its receptor-interacting protein homotypic interaction motif (RHIM) domain to form an amyloid-like structure (Fig. 3).95,96 During this process, RIPK1 also binds to other proteins containing the RHIM domain, including TIR-domain-containing adaptor-inducing interferon-β (TRIF) and ZBP1, which function as alternative initiators of necroptotic signalling.Fig. 3Full size imageSignalling cascades and execution mechanisms of necroptosis. Necroptosis is initiated by diverse stimuli, including cytokines (TNF, FASL, and TRAIL), pathogens (LPS and viruses), and their corresponding receptors (TNFR1, FASR, DR4/5, and TLR4) or intracellular sensors (ZBP1). Upon TNFR1 binding, membrane-bound Complex I assembles to drive NF-κB activation and promote cell survival. Deubiquitylation of RIPK1 facilitates its transition into cytosolic Complex II, which comprises RIPK1, FADD, TRADD, and caspase-8. When caspase-8 activity is inhibited, RIPK1 recruits RIPK3 to form the necrosome, triggering the phosphorylation and oligomerisation of MLKL. The translocation of phosphorylated MLKL to the plasma membrane initiates membrane destabilisation. Critically, the terminal execution of necroptosis is orchestrated by distinct effectors: SIGLEC12 undergoes dephosphorylation and mediates plasma membrane rupture, whereas NINJ1 oligomerises to form pores that facilitate DAMP release. Conversely, the ESCRT-III machinery serves as a negative regulator by repairing membrane damage to preserve cellular integrity. By FigdrawThe assembly of the necrosome relies on the interaction between RIPK1 and RIPK3; however, recent structural analyses have revealed that this complex forms a stable, heteroamyloid fibril architecture rather than a simple disordered aggregate.97 Notably, the specific recruitment mechanisms within this supramolecular structure vary significantly across species, a difference with crucial implications for translational medicine. In murine models, ZBP1 can directly engage RIPK3 to trigger necroptosis. In contrast, groundbreaking studies in human cells have demonstrated that ZBP1 cannot bind RIPK3 directly. Instead, RIPK1 acts as an obligatory adaptor protein to bridge this interaction, facilitating the assembly of a functional necrosome.98 This species-specific dependency underscores the complexity of the RHIM-dependent signalling network and suggests that the regulation of necroptosis in humans involves additional checkpoints to prevent aberrant activation.After necrosome assembly, the signal is propagated to the terminal effector, MLKL.95 In its dormant state, MLKL exists in the cytosol as a monomer, folded in a conformation where its C-terminal pseudokinase domain suppresses the lethal potential of its N-terminal four-helix bundle (4HB) domain.88 The activation process is initiated when RIPK3 phosphorylates MLKL at specific serine/threonine residues (Ser358 and Thr357 in humans; Ser345 in mice) located within the activation loop of the pseudokinase domain.88 This phosphorylation acts as a molecular “release mechanism”, inducing a substantial conformational change that unlatches the 4HB domain from the pseudokinase regulatory region.99 Liberated from autoinhibition, MLKL rapidly forms higher-order oligomers, which act as the active units for necroptosis execution.100 Finally, these phosphorylated oligomers are actively transported via a Golgi–microtubule network to accumulate at specific hotspots on the plasma membrane.99However, the phosphorylation of RIPK3 and MLKL alone does not guarantee the bona fide completion of necroptosis. For example, metabolites such as acrolein can robustly induce RIPK3 and MLKL phosphorylation, but they effectively abort terminal necroptosis by targeting chaperone proteins (HSP90) to inhibit MLKL oligomerisation.101 Thus, confirming MLKL oligomerisation is strictly necessary to define true necroptotic activity. Once oligomerised, MLKL actively translocates to the plasma membrane via a highly regulated Golgi–microtubule-dependent network.99 At the membrane, the positively charged MLKL 4HB domain electrostatically binds to negatively charged phosphatidylinositol phosphates (PIPs), allowing its amphipathic helices to insert into the lipid bilayer and compromise membrane integrity.102For decades, MLKL oligomers were thought to directly form the lytic pores responsible for cell death.103 However, recent paradigm-shifting discoveries have redefined the final moments of necroptotic cell death. While MLKL insertion indeed permeabilises the membrane to small ions, causing an influx of calcium (Ca2+) and sodium (Na+), it does not solely cause the catastrophic plasma membrane rupture (PMR) that characterises necroptosis.104 Instead, the ion flux and membrane stress mediated by MLKL activate a separate cell-surface protein, Ninjurin-1 (NINJ1).104 In response to MLKL-induced cellular swelling and membrane damage, NINJ1 polymerises into extensive filament rings that essentially ‘slice’ the plasma membrane, releasing membrane discs and leading to the explosive release of large intracellular DAMPs.105 Thus, MLKL acts as the “key” to initiating cell death, whereas NINJ1 functions as the “battering ram” that physically disrupts the cellular boundary. Subsequent studies revealed that during necroptosis, cells still undergo rupture even in the absence of NINJ1, suggesting the existence of another executor specifically responsible for plasma membrane rupture (PMR) in necroptosis.104 Noh et al. identified SIGLEC12 as a key mediator of PMR during necroptosis.106 Upon MLKL activation, SIGLEC12 interacts with MLKL and is specifically cleaved by TMPRSS4 at Arg410, releasing an active fragment containing the NISI domain. This fragment subsequently assembles into fibrillar structures at the plasma membrane, directly mediating membrane rupture and thereby completing the terminal steps of necroptosis. Notably, this mechanism is species-specific: SIGLEC12-mediated PMR is essential in human cells but dispensable in mouse cells (knockout of mouse Siglec12 does not affect PMR during necroptosis).Furthermore, this lethal process is subject to a final layer of negative regulation. Cells possess an intrinsic repair mechanism mediated by the endosomal sorting complexes required for transport (ESCRT-III) machinery.107 In the early stages of MLKL activation, the ESCRT-III complex can recognise MLKL-damaged membrane segments and excise them in the form of extracellular vesicles (or “bubbles”), thereby delaying cell death and sustaining membrane integrity.99 Consequently, the execution of necroptosis is determined by a critical threshold: Cell death occurs only when the rate of MLKL activation and subsequent NINJ1-mediated rupture overwhelms the capacity of the ESCRT-III repair machinery.PyroptosisPyroptosis is a distinct, regulated, and lytic form of RCD characterised by its robust proinflammatory outcome.108 The term “pyroptosis”, originating from the Greek pyro (fire) and ptosis (falling), was initially coined to describe caspase-1-dependent death in macrophages infected with intracellular pathogens.15 Currently, pyroptosis is critically dependent on the pore-forming activity of cleaved gasdermin proteins.85 Upon activation, these proteins oligomerise to form pores in the plasma membrane, leading to the release of inflammatory cytokines (IL-1β and IL-18) and to osmotic imbalance, cell swelling, and eventual lytic rupture (Fig. 4).85,109 This update broadens the scope beyond the original inflammasome-caspase-1 paradigm to include various upstream activating proteases and pathways.Fig. 4Full size imageMajor activation pathways of pyroptosis and their regulatory networks. Pyroptosis is activated through three major pathways, namely, the canonical, noncanonical, and alternative pathways, with extensive cross-regulation among them. Canonical inflammasome pathway: PRRs such as NLRP1/3, NLRC4, and AIM2 recognise PAMPs or DAMPs and then recruit the adaptor protein ASC, which in turn recruits and activates procaspase-1. Activated caspase-1 cleaves pro-IL-1β and pro-IL-18 to generate mature proinflammatory cytokines and cleaves GSDMD at Asp275 (human)/Asp276 (mouse) to release the N-terminal pore-forming domain, thereby triggering pyroptosis. Noncanonical inflammasome pathway: LPS from Gram-negative bacteria enters the host cytoplasm and directly binds to the CARD of caspase-4/5 (human)/caspase-11 (mouse), inducing their oligomerisation and autocatalytic activation. These activated caspases then cleave GSDMD to initiate pyroptosis. Caspases in this pathway cannot directly process pro-IL-1β and pro-IL-18; instead, potassium efflux triggered by GSDMD pores activates the NLRP3 inflammasome, which recruits and activates caspase-1 to mediate the maturation and release of proinflammatory cytokines. Alternative pyroptosis pathway: GZMA cleaves GSDMB to activate it in a species-specific pathway (GSDMB is absent in mice). GZMB, which is delivered into target cells by cytotoxic lymphocytes, cleaves GSDME either directly or indirectly (by activating caspase-3) to induce pyroptosis, with the expression level of GSDME serving as a molecular switch for this cell death transition. Caspase-8 can cleave GSDMD (during Yersinia infection) or GSDMC (in the hypoxic tumour microenvironment) to trigger pyroptosis under specific conditions. These pathways act as critical links between the adaptive immune system and pyroptosis and are involved in processes such as tumour therapy and immune surveillance. By FigdrawGasdermin family: the final executor of inflammatory cell deathThe entire execution phase is governed primarily by the gasdermin protein family. In humans, this family includes six members: GSDMA, GSDMB, GSDMC, GSDMD, GSDME, and PJVK. Among these, GSDMA, GSDMB, GSDMC, GSDMD and GSDME are directly involved in the execution of pyroptosis through their pore-forming activities, whereas PJVK has distinct functions related to peroxisome homeostasis.110,111 Mice lack a GSDMB orthologue but possess multiple Gsdma and Gsdmc genes alongside GSDMD and GSDME.112 All pyroptosis-related gasdermins share a conserved two-domain architecture: an N-terminal pore-forming domain and a C-terminal repressor domain, connected by a flexible linker region.113 Under normal physiological conditions, the C-terminal domain folds over and interacts with the N-terminal domain, maintaining the protein in an autoinhibited state. This intramolecular interaction masks critical hydrophobic and basic residues on the N-terminal domain required for membrane binding and oligomerisation.113The activation of gasdermins requires proteolytic cleavage within the linker region, which results in autoinhibition. Various proteases can perform this cleavage depending on the specific signalling pathway involved. After cleavage, the liberated N-terminal fragment undergoes conformational rearrangement and exposes membrane‑binding motifs with high selectivity for specific acidic phospholipids.85,113 These lipids include phosphatidylinositol phosphates, such as PtdIns(4)P and PtdIns(4,5)P2, which are enriched in the inner leaflet of the plasma membrane, and cardiolipin, which is present in mitochondrial and bacterial membranes.85 Upon membrane anchoring, the N-terminal fragments oligomerise to form large transmembrane pores.114 Cryo-EM revealed that the gasdermin A3 membrane pore assembles into a ring‑shaped β‑barrel comprising approximately 27 subunits, with an inner diameter of 18–22 nm.115 The core function of gasdermin pores is to mediate the unconventional secretion of proinflammatory cytokines (e.g., IL-1β and IL-18).109 These cytokines lack a classic signal peptide for ER-Golgi-dependent secretion. The gasdermin pores in the plasma membrane provide a direct conduit for their release into the extracellular space, amplifying local and systemic inflammatory responses.116 The leakage of other intracellular contents, including alarmins and DAMPs, further propagates inflammatory responses.117 Therefore, gasdermin-mediated pore formation is the defining and indispensable event that connects upstream danger sensing to the lytic and highly inflammatory phenotype of pyroptosis.118Canonical inflammasome pathway: caspase-1-mediated GSDMD activationThe canonical inflammasome pathway is the best-established mechanism for driving pyroptosis. This pathway centres on the assembly of a multiprotein complex, called the inflammasome, which acts as a molecular platform for the activation of caspase‑1.119 Inflammasome assembly is triggered by the detection of specific danger signals in the cytosol. A typical inflammasome complex comprises three core components: a sensor protein, an adaptor protein, and an effector caspase.112 The sensor proteins are pattern recognition receptors that detect PAMPs or host-derived DAMPs.119 Major inflammasome sensors include several NLRs and other cytosolic sensors. The NLR family pyrin domain-containing 1 (NLRP1) inflammasome is activated by bacterial factors such as anthrax lethal toxin and muramyl dipeptide.120 The NLR family CARD domain-containing 4 (NLRC4) inflammasome responds to bacterial flagellin and components of type III secretion systems.121 The absent in melanoma 2 (AIM2) inflammasome is a non-NLR sensor that directly binds cytosolic double-stranded DNA from viruses or bacteria.122 The pyrin inflammasome is activated by modifications to host Rho GTPases mediated by bacterial toxins, such as Clostridium difficile toxin B.123Among the members of the NLR family, the NLRP3 inflammasome is the most extensively studied because of its activation by a remarkably wide array of diverse stimuli. These activators include extracellular ATP, uric acid crystals, cholesterol crystals, amyloid-β aggregates, silica, asbestos, ROS, and various bacterial, viral, and fungal pathogens.124 Activation of the NLRP3 inflammasome is a tightly regulated two-step process.125 The first step is priming, which is often initiated by TLR agonists such as LPS. This signalling, which is typically mediated through the NF-κB pathway, transcriptionally upregulates NLRP3, pro-IL-1β, and pro-IL-18. The second step is activation, triggered by a specific stimulus such as potassium efflux, mitochondrial reactive oxygen species (mtROS) production, or lysosomal damage. These events promote the oligomerisation of NLRP3. Upon activation, most inflammasome sensors (except NLRC4, which can directly recruit caspase-1) recruit the adaptor protein apoptosis-associated speck-like protein containing a CARD (ASC).126,127 ASC contains two death domains: a pyrin domain (PYD) and a caspase activation and recruitment domain (CARD).128 The sensor interacts with ASC via PYD–PYD homotypic interactions. ASC then polymerises into large filamentous structures visible as specks under microscopy. These ASC specks recruit procaspase-1 via CARD–CARD interactions, bringing multiple procaspase-1 molecules into proximity.126 This proximity-induced autocatalytic cleavage activates caspase-1.Activated caspase-1 has two primary substrates that define the canonical pyroptosis pathway. First, it cleaves the inactive cytokine precursors pro-IL-1β and pro-IL-18 at specific aspartate residues to generate mature, highly proinflammatory cytokines.129 Second, caspase-1 cleaves GSDMD. Cleavage of human GSDMD occurs after aspartate 275 (Asp275), whereas cleavage of mouse GSDMD occurs after Asp276.85 This proteolytic event separates the N-terminal pore-forming domain from the C-terminal inhibitory domain. The released GSDMD-N-terminal fragments rapidly translocate to the inner leaflet of the plasma membrane, where they bind to specific lipids, oligomerise, and insert to form characteristic pores. This pathway is a fundamental host defence mechanism, primarily occurring in immune cells such as macrophages to eliminate intracellular pathogen niches and alert neighbouring cells.130Noncanonical inflammasome pathway: caspase-4/5/11-mediated GSDMD activationThe noncanonical pyroptosis pathway provides a direct cellular response to cytosolic Gram-negative bacterial infections and operates independently of canonical inflammasome sensors and caspase-1. The central trigger is bacterial LPS, which enters the host cell cytosol, while the key effectors are inflammatory caspases that directly sense LPS, such as caspase-11 in mice and its human orthologues caspase-4 and caspase-5.131This pathway is initiated by the cytosolic delivery of LPS, which is mediated by bacterial secretion systems, outer membrane vesicles, or interferon-inducible guanylate-binding proteins (GBPs).132 The cytosolic lipid A moiety of LPS is directly recognised and bound by the CARD of caspase-11/4/5.131 This binding drives caspase oligomerisation and autoactivation, triggering pyroptosis without the involvement of conventional inflammasome sensors. Activated caspase-4, -5, or -11 then cleaves GSDMD at the same site targeted by caspase-1.4 A key distinction from the canonical pathway is that caspase-11 is inefficient at processing pro-IL-1β and pro-IL-18 into their mature forms.133 Therefore, while the noncanonical pathway directly induces cell lysis via GSDMD, it does not directly mediate the release of these two cytokines. Instead, the noncanonical pathway engages the canonical machinery to achieve full inflammatory output. The pores formed by GSDMD-N induce rapid efflux of potassium ions (K+) from the cytosol, a well-established potent activator of the NLRP3 inflammasome.134 This activates NLRP3 inflammasome assembly and caspase-1 activation. Activated caspase-1 then processes pro-IL-1β and pro-IL-18, and these mature cytokines are released through the preexisting GSDMD pores.134An additional regulatory layer involves pannexin-1 channels. Caspase-11 can cleave and activate pannexin-1 channels in the plasma membrane, leading to the release of cellular ATP. This extracellular ATP acts in an autocrine or paracrine manner on purinergic P2X7 receptors on the same or neighbouring cells.135 P2X7 receptor activation is another potent inducer of K+ efflux that further amplifies NLRP3 inflammasome activation, creating a positive feedback loop to increase inflammation.136 In the disease setting of gram-negative bacterial infections, this GSDMD-dependent pathway is crucial for defence; however, its robust inflammatory output and dysregulation critically contribute to severe systemic pathologies such as septic shock.136Alternative pyroptosis pathways mediated by apoptotic caspases and granzymesIn addition to the inflammatory caspases associated with inflammasomes, other proteolytic systems can cleave and activate specific gasdermins. These alternative pathways illustrate how distinct gasdermins shape inflammatory outputs in diverse biological contexts, particularly during cancer therapy and immune surveillance, thereby expanding the physiological and pathological relevance of pyroptosis.GSDME in cancer settingsThe apoptotic effector caspase-3 can induce pyroptosis by cleaving GSDME (also known as DFNA5).137 This switch frequently occurs during cancer chemotherapy. Chemotherapeutic agents such as cisplatin, paclitaxel, and doxorubicin activate caspase‑3 through canonical apoptotic signalling. In cells expressing high levels of GSDME, activated caspase-3 cleaves GSDME after it reaches Asp270.138 This cleavage acts as a critical molecular switch in tumour tissues, altering the morphological outcome from a noninflammatory, apoptotic phenotype (characterised by cell shrinkage and apoptotic bodies) to a highly inflammatory, pyroptotic phenotype (characterised by cell swelling and lysis).139 Thus, GSDME expression acts as a molecular switch that determines whether caspase‑3 activation leads to apoptosis or pyroptosis.GSDME and GSDMB in immune surveillanceSimilarly, granzyme B (GZMB), a serine protease released by cytotoxic lymphocytes (natural killer cells and cytotoxic T cells), directly cleaves GSDME at the same site as caspase-3.140 GZMB is delivered into target cells via perforin, enabling immune cells to induce pyroptosis independently of the target cell’s intrinsic caspase activation.141 GZMB can also promote pyroptosis by cleaving and activating procaspase-3 within target cells, after which caspase-3 cleaves GSDME.142 Furthermore, other granzymes target different gasdermins. Granzyme A (GZMA), another protease from cytotoxic lymphocytes, cleaves human GSDMB at Lys229 and Lys244.143 Notably, GSDMB is absent in mice, highlighting species-specific differences in pyroptosis execution. Together, these granzyme-mediated pathways form a critical link between the adaptive immune system and pyroptosis, allowing cytotoxic lymphocytes to eliminate target tissues while simultaneously releasing potent inflammatory signals that amplify antitumour or antiviral immunity.144GSDMC and GSDMD via caspase-8 in specific disease modelsThe apoptotic initiator caspase-8 also participates in pyroptosis induction under specific conditions.145 During Yersinia infection, the bacterial effector protein YopJ inhibits the host kinase TAK1, blocking prosurvival signalling and promoting the assembly of the RIPK1-FADD-caspase-8 complex and the activation of caspase-8.146 Activated caspase-8 then cleaves GSDMD to induce macrophage pyroptosis,147 representing an alternative route for GSDMD activation that is independent of the canonical inflammasome pathway. Furthermore, in certain cancer models, distinct gasdermins shape the inflammatory output of hypoxic tumour tissues. The execution of pyroptosis by caspase-8 depends on the prior transcriptional upregulation of GSDMC. For example, under hypoxic conditions in tumour cells, nuclear PD-L1 and phosphorylated STAT3 drive the transcription of GSDMC. Subsequently, TNF signalling activates caspase-8, which cleaves this transcriptionally upregulated GSDMC protein to induce pyroptosis.148Apart from caspase-8-induced pyroptosis, human monocytes possess a distinct TLR4-TRIF-RIPK1-FADD-CASP8-dependent alternative NLRP3 inflammasome pathway. When activated by LPS alone, this cascade drives the release of IL-1β but does not trigger pyroptosis. The absence of potassium efflux and ASC pyroptosome formation leads to limited caspase-1 activity, which cannot effectively cleave gasdermin to cause cell lysis. Moreover, caspase-8 restrains necroptosis signalling, ultimately uncoupling the inflammatory response from cell death in human innate immune cells.149Metal-dependent cell deathMetals, encompassing bulk elements (e.g., Na, K, Mg and Ca) and trace elements (e.g., Fe, Mn, Co, Cu, Zn and Mo), are indispensable for organismal development and cellular physiology.150 The precise homeostasis of these ions is a critical determinant of cell fate. Deviations from equilibrium, whether through deficiency or excess, can initiate specific forms of RCD by dysregulating core signalling networks. For instance, redox-active metals such as Fe²⁺/³⁺, Cu⁺/²⁺ and Mn²⁺/³⁺/⁴⁺ are integral to electron transfer and energy metabolism.151 However, their accumulation induces excessive ROS generation, disrupts the intracellular redox balance, and thereby elicits ferroptosis or other RCD pathways. Beyond their redox roles, metals such as Fe, Cu and Zn are crucial for maintaining the structural integrity of proteins and macromolecules.152 Specifically, Cu⁺ overload compromises the stability of iron‒sulfur cluster proteins, a hallmark of cuproptosis.18 As enzymatic cofactors, Ca²⁺, Zn²⁺, and Mg²⁺ are fundamental to catalytic activity and metabolic processes.153 Notably, Zn²⁺ and Mg²⁺ can inhibit key glycolytic enzymes, creating a state of metabolic starvation that promotes apoptosis.154 In contrast, dysregulated Ca²⁺ signalling triggers caspase activation, driving both apoptosis and pyroptosis.155 Collectively, these mechanisms support the therapeutic potential of strategically manipulating intracellular metal ion homeostasis, which represents a promising approach for regulating cell fate through the precise induction or suppression of specific RCD pathways.FerroptosisFerroptosis is a form of RCD driven by iron-dependent lipid peroxidation within mitochondria, which causes inactivation of glutathione peroxidase 4 (GPX4), membrane destabilisation, and oxidative cellular collapse.17 GPX4 activity relies critically on glutathione (GSH), whose biosynthesis requires cysteine and glutamate, and its uptake is regulated by the cystine/glutamate antiporter system xc−. Ferroptosis is characterised by three pivotal hallmarks: (1) oxidative damage to membrane phospholipids containing polyunsaturated fatty acids (PUFAs), (2) the availability of redox-active iron, and (3) a compromised capacity to repair lipid hydroperoxides (LOOHs).156 Ferroptosis is morphologically and biochemically distinct from apoptosis and necroptosis, with characteristic abnormal mitochondrial changes.157 Its discovery connects redox biology, iron metabolism, and disease pathophysiology, and it is triggered by the dysregulation of three interrelated metabolic pathways: iron metabolism, glutathione biosynthesis, and lipid metabolism (Fig. 5).158Fig. 5Full size imageThree classical ferroptosis metabolic pathways. Ferroptosis is caused by excessive lipid peroxidation and occurs mainly on PUFA-PL. The synthesis of polyunsaturated fatty acid phospholipids is regulated by ACSL4, LPCAT3 and ACC. Ferroptosis is triggered by the iron-mediated Fenton reaction. Therefore, the sensitivity to ferroptosis is closely related to iron metabolism, involving processes such as iron uptake (e.g., through transferrin), iron storage (e.g., through ferritin), and iron release (e.g., through NCOA4). System Xc−-GSH–GPX4 acts as a defence mechanism against ferroptosis. The Xc−-GSH–GPX4 axis contributes to the reduction of phospholipid hydroperoxides to corresponding alcohols (PL-OHs), preventing and degrading the occurrence of peroxidation. The inhibition of system Xc− (e.g., by erastin) depletes GSH, leading to the unchecked accumulation of phospholipid hydroperoxides and triggering ferroptosis. By FigdrawMechanism of ferroptosis execution—Iron metabolism and ROS productionIron accumulation is a fundamental prerequisite for initiating ferroptosis. Consequently, genes and proteins involved in iron metabolism can modulate cellular sensitivity to this form of cell death by controlling intracellular iron levels.158 There are two types of iron ions within the cell: Fe2+ and Fe3+. Among them, Fe2+ is the main ion involved in the Fenton reaction.159 When the activity of GPX4 is diminished or lost, toxic LOOH cannot be reduced to nontoxic lipid alcohols. These accumulated peroxides, in the presence of redox-active Fe²⁺, initiate and propagate lipid peroxidation, leading to a surge in the generation of hydroxyl radicals and ROS, ultimately resulting in ferroptosis.160 Cellular Fe²⁺ synthesis primarily occurs through two pathways (Fig. 5). The first and predominant route begins with transferrin receptor 1 (TFR1)–mediated endocytosis of extracellular Fe³⁺-bound transferrin. Within endosomes, Fe³⁺ is reduced to Fe²⁺ by six-transmembrane epithelial antigen of prostate 3 (STEAP3) and subsequently transported into the cytosolic labile iron pool (LIP) by divalent metal transporter 1 (DMT1/NRAMP2). This mobilised Fe²⁺ fuels ROS generation and acts as a cofactor for activating lipoxygenases (LOXs), thereby initiating lipid peroxidation.161 The second pathway involves the ferritinophagy-mediated release of Fe²⁺. Excess Fe³⁺ is stored in ferritin heavy chain 1 (FTH1), which can be mobilised and reduced to Fe²⁺ when needed. This Fe²⁺ release is further amplified through ferritinophagy, a selective autophagy process mediated by the binding of FTH1 to nuclear receptor coactivator 4 (NCOA4),162 ultimately driving mitochondrial lipid peroxidation.163 In addition, mitochondrial metabolism contributes to ROS generation.5Multiple intracellular molecules play pivotal roles in ferroptosis by modulating Fe²⁺ levels. For instance, transferrin and its receptor facilitate iron import, thereby triggering ferroptosis.164 Iron-responsive element-binding protein 2 (IREB2) increases cellular iron uptake and ferroptosis susceptibility by stabilising the transcription of DMT1 and TFR1 while concurrently regulating the function of ferroportin (FPN).165 Conversely, heat shock protein B1 (HSPB1) reduces cellular iron load by suppressing TFR1 expression.166 Other known regulators of TFR1 include ubiquitin-specific protease 7, microRNA-214, and the atypical cyclin-like protein SPY1.164,167 Ferroportin 1 (FPN1), the principal mammalian iron exporter, is a potential therapeutic target in ferroptosis prevention via degradation regulation.168 The expression of FPN1 is negatively regulated by hepcidin, which induces FPN1 degradation and promotes ferroptosis. In non-small-cell lung cancer, the deubiquitinase USP35 inhibits ferroptosis by interacting with and stabilising FPN.169 In summary, cellular iron homeostasis is governed by the systems responsible for iron uptake (TFR1), iron storage (ferritin), and iron export (FPN1), which thus regulate sensitivity to ferroptosis.The convergence of iron overload and dysregulated lipid metabolism drives excessive ROS generation, leading to oxidative stress and activation of nuclear factor erythroid 2-related 2 (NRF2).170 The activation of NRF2 orchestrates a comprehensive cytoprotective response: It transcriptionally upregulates a series of genes, such as haem oxygenase-1 (HO-1), GPX4, and FTH1.170 Concurrently, NRF2 modulates iron homeostasis through three key effectors, namely, the HECT/RLD domain-containing E3 ligase HERC2, vesicle-associated membrane protein 8 (VAMP8), and nuclear receptor coactivator 4 (NCOA4), to balance ferritin synthesis and turnover. Together, these actions establish a dual defence mechanism against ferroptosis in malignant cells: inhibition of phospholipid peroxidation and sequestration of redox-active iron. By precisely regulating LIP, this integrated NRF2 network functions as a master regulator of cellular susceptibility to ferroptosis.Mechanism of ferroptosis execution–the system Xc–GSH–GPX4 axis and cystine deprivationFerroptosis is governed primarily by the system Xc− and GPX4 pathways. System Xc−, a cystine/glutamate antiporter, is composed of dimers of solute carrier family 3 member 2 (SLC3A2) and solute carrier family 7 member 11 (SLC7A11) embedded in the plasma membrane.171 SLC7A11 imports cystine, which is essential for the synthesis of GSH. As a potent reducing agent, GSH plays dual, interconnected roles in this pathway: As an essential cofactor, it enables GPX4 to reduce toxic lipid hydroperoxides (e.g., R-OOH) to inert alcohols (e.g., R-OH), which are concomitantly oxidised to glutathione disulfide (GSSG), thereby preventing ferroptosis caused by the accumulation of lipid ROS.17 GSSG is subsequently recycled back to GSH by glutathione reductase (GSR) in an NADPH-dependent manner (Fig. 5). GPX4, a selenoprotein, functions as a key phospholipid hydroperoxidase that detoxifies lipid peroxides in cellular membranes, thereby preventing a lethal peroxidation cascade.172 Inhibition of system Xc− (e.g., by erastin) depletes GSH, whereas direct targeting of GPX4 (e.g., by RSL3) disables the final enzymatic defence, leading to both unchecked accumulation of phospholipid hydroperoxides, particularly those derived from arachidonoyl and adrenoyl phosphatidylethanolamines, and subsequent ferroptotic demise.172,173 The system Xc−–GSH–GPX4 pathway is also significantly modulated by the transcription factor NRF2. Notably, NRF2 directly upregulates the expression of SLC7A11 as well as both the catalytic and regulatory subunits of glutamate–cysteine ligase.174 These targets work together to sustain cellular GSH levels, thereby supporting the activity of GPX4. Conversely, p53 suppresses ferroptosis through a dual mechanism involving direct binding to the SLC7A11 promoter and interaction with ubiquitin-specific peptidase 7 (USP7). These combined actions reduce histone H2B monoubiquitination at the SLC7A11 promoter, resulting in its transcriptional repression.175 Consequently, this pathway disrupts GSH biosynthesis and promotes ferroptosis.176Beyond this GPX4-related core axis, parallel regulatory nodes fine-tune ferroptotic susceptibility. Ferroptosis suppressor protein 1 (FSP1) is an NADPH-dependent oxidoreductase that effectively inhibits lipid peroxidation by reducing coenzyme Q10 (CoQ10) to its hydroquinone form (CoQH₂).177 The FSP1–CoQ10–NAD(P)H axis constitutes a GPX4-independent, mitochondria-agnostic ferroptosis suppression system in which FSP1 catalyses the regeneration of the antioxidant ubiquinol.178 Additionally, two key pathways, namely, the GTP cyclohydrolase 1-tetrahydrobiopterin (GCH1–BH4) pathway and the dihydroorotate dehydrogenase-coenzyme Q (DHODH-CoQ) pathway, contribute to the lipid antioxidant network and exert compensatory resistance mechanisms against ferroptosis.164,179,180 Specifically, the GCH1–BH4 pathway inhibits ferroptosis through lipid remodelling.179 BH4, a product of GCH1-catalysed reactions, acts as a lipophilic radical-trapping antioxidant. It can directly neutralise lipid radicals, suppress phospholipid peroxidation, and promote the reduction of CoQ10 to increase the lipid antioxidant capacity. Thus, the GCH1–BH4 pathway is a GPX4-independent compensatory resistance mechanism that selectively protects polyunsaturated phospholipids from peroxidation. On the other hand, DHODH, located on the inner mitochondrial membrane, coordinates with GPX4 to inhibit ferroptosis by reducing CoQ to ubiquinol, which constitutes a critical compensatory ferroptosis resistance pathway under GPX4-inhibited conditions.180 Furthermore, there are novel GPX4- and FSP1-independent mechanisms for suppressing ferroptosis: the membrane-bound O-acyltransferase domain-containing protein 1 and 2-monounsaturated fatty acid (MBOAT1/2-MUFA) system and the sterol-C5-desaturase-7-dehydrocholesterol (SC5D-7-DHC) system.181,182 MBOAT2 catalyses the transfer of MUFAs to lysophospholipids (e.g., lysophosphatidylethanolamine, Lyso-PE), generating MUFA-incorporated phosphatidylethanolamine (PE-MUFA) while reducing the cellular pool of peroxidation-susceptible polyunsaturated fatty acid-containing phosphatidylethanolamine (PE-PUFA). The presence of endogenous or exogenous MUFAs is essential for this process.181 In the SC5D-7-DHC system, 7-DHC is an intermediate in cholesterol biosynthesis and has been confirmed to be a potent antiferroptotic metabolite. Cancer cells accumulate 7-DHC, which serves as a potent lipophilic antioxidant. By shielding phospholipids in both the plasma and mitochondrial membranes from peroxidation, 7-DHC suppresses ferroptosis and facilitates tumour survival.183 Consequently, targeting 7-DHC biosynthesis has emerged as a promising therapeutic strategy for cancer treatment. Collectively, these pathways form an important lipid antioxidant network independent of GPX4, providing effective compensatory resistance to prevent ferroptosis.Mechanism of ferroptosis execution—Lipid peroxideLipid peroxidation (LPO) refers to the biochemical process in which the lipids of cell membranes or membrane-containing organelles are oxidised by free radicals (including ROS and ·OH). Lipid peroxides and their derivatives, including malondialdehyde (MDA) and 4-hydroxynonenal (4-HNE), are indispensable biomarkers of ferroptosis.184 Polyunsaturated fatty acid phospholipids (PUFA-PLs) play a crucial role in the process of phospholipid peroxidation, whereas monounsaturated fatty acids (MUFAs) have antiferroptotic activity.185 Therefore, the synthesis pathway of PUFA-PLs plays a pivotal role in governing ferroptosis occurrence. The synthesis of MUFA-PLs and PUFA-PLs originates from the acetyl-CoA carboxylase (ACC)-catalysed conversion of acetyl-CoA to malonyl-CoA, which is essential for PUFA generation. Long-chain acyl-CoA synthetase 4 (ACSL4) and lysophosphatidylcholine acyltransferase 3 (LPCAT3) subsequently cooperate to catalyse the incorporation of PUFAs into phospholipids to form PUFA-PLs.5 These PL-PUFAs are then susceptible to peroxidation by iron-dependent lipoxygenases and ROS, producing phospholipid hydroperoxides (PUFA-PL-OOH) that ultimately trigger ferroptosis186 (Fig. 5). The activities of ACSL3 and ACSL4 significantly influence cellular sensitivity to ferroptosis. Wang et al. demonstrated that ACSL4 knockout inhibits ferroptosis and alleviates renal injury in mice.187Mechanism of ferroptosis execution—Metabolic states and cell densityBeyond the core execution pathways, cellular susceptibility to ferroptosis is profoundly modulated by broader physiological contexts, notably metabolic states and cell density. Under conditions of energy stress (glucose starvation), AMP-activated protein kinase (AMPK) is activated to maintain energy homeostasis. AMPK confers resistance to ferroptosis by phosphorylating and inhibiting ACC, thereby halting the biosynthesis of PUFAs and depleting the substrates required for lipid peroxidation.145Furthermore, spatial confluence and intercellular interactions strictly dictate ferroptosis sensitivity. High cell density and established cell–cell contacts suppress ferroptosis through the E-cadherin–NF2–Hippo signalling axis. In confluent cells, the activation of this pathway restricts the nuclear translocation of the transcriptional coactivator Yes-associated protein (YAP). This inhibition subsequently downregulates the expression of key ferroptosis promoters, including ACSL4 and TFR1.188 Thus, the metabolic microenvironment and physical tissue architecture act as crucial upstream gatekeepers that fine-tune the cellular threshold for ferroptosis.CuproptosisCopper (Cu), an essential trace element, acts as a key cofactor for various metabolic enzymes and is involved in vital biological processes such as mitochondrial respiration, antioxidant defence, iron metabolism, and the biosynthesis of diverse biomolecules.189 Disulfiram, initially approved for alcohol dependence treatment,190 was later found to exert anticancer effects potentiated by copper supplementation.191 These complexes induce cell death by disrupting protein homeostasis, with cellular alterations similar to those caused by proteasome inhibitors such as MG132 and bortezomib.150 In 2022, Tsvetkov et al. defined this unique form of cell death as cuproptosis.18 This process is characterised by intracellular copper accumulation, mitochondrial lipoylated protein aggregation, Fe‒S cluster protein destabilisation, and proteotoxic stress.192Like the mitochondrion-dependent intrinsic apoptotic pathway, cuproptosis is characterised primarily by mitochondrial damage. Notably, cuproptosis inducers, such as elesclomol and DSF, can elicit apoptosis-like morphological alterations.193 However, cuproptosis operates independently of BAX/BAK, whereas the intrinsic apoptotic pathway requires the involvement of these proteins. In contrast, ferroptosis exhibits distinct mitochondrial morphological alterations (e.g., shrinkage, increased membrane density and cristae loss).194 Tian et al. utilised transmission electron microscopy (TEM) to elucidate the unique morphological characteristics of cuproptosis: mitochondrial swelling with inner–outer membrane dissociation, complete cristae disintegration, abnormal inner membrane folding and rupture, reduced outer membrane density, protein aggregation, copper particle deposition in the endoplasmic reticulum, and localised plasma membrane damage.195Copper metabolismMaintaining cellular copper homeostasis depends on a finely regulated network of transport proteins, molecular chaperones, and regulatory factors196 (Fig. 6a).Fig. 6Full size imageSchematic diagram of copper homeostasis. a Cu²⁺ is first reduced to Cu⁺ by the STEAP family before it is absorbed into intestinal epithelial or somatic cells via SLC31A1. In the cytoplasm, CCS delivers Cu⁺ to SOD1 to form homodimers for ROS neutralisation and transports Cu⁺ into the nucleus to activate transcription factors such as HIF-1α. Mitochondrial copper transport relies on the ability of COX17 to deliver Cu⁺ to MT-CO2, a process that is additionally assisted by SCO1, SCO2, and COA6; moreover, COX17 also supplies MT-CO1 via COX11. ATOX1 transports copper to the trans-Golgi network. ATP7A and ATP7B are crucial for copper efflux: At elevated intracellular copper levels, they undergo conformational changes and translocate from the trans-Golgi network to the plasma membrane via exocytosis to pump copper out of the cell. b Schematic diagram of cuproptosis and lipoylation. Cu2⁺ is transported into cells via the copper ionophore elesclomol and subsequently imported into mitochondria. Within the mitochondria, FDX1 reduces Cu2⁺ to Cu⁺ and releases it from elesclomol. Key metabolic enzymes (e.g., DLAT) then undergo lipoylation, which is catalysed by FDX1, LIAS, and other enzymes. Cu⁺ directly binds to these lipoylated proteins, inducing their oligomerisation. In parallel, Cu⁺ causes the loss of Fe‒S clusters. These aberrant events collectively induce proteotoxic stress, ultimately leading to cell death. By FigdrawCopper uptake is mediated primarily by the high-affinity transporter CTR1 (Slc31a1), which forms Cu⁺ complexes that are modulated by extracellular conditions.197 This step depends on the reduction of extracellular Cu²⁺ to Cu⁺ by six-transmembrane epithelial antigens of the prostate (STEAP) reductases, although other efficient in vivo reductases might also assist.198 Following reduction, CTR1-specific translocation is highly selective for Cu⁺.199 Additionally, zinc transporter 1 (ZnT1) serves as a functional alternative for cellular copper uptake, enabling Zn²⁺ to competitively inhibit Cu²⁺ to alleviate copper overload, particularly under copper-stress conditions. Moreover, although divalent metal transporter 1 (DMT1) was initially proposed to mediate divalent copper uptake with minimal activity,200 conflicting evidence exists for its direct physiological role in copper transport, and further substantiation is needed.201Upon entering cells, copper is rapidly sequestered by dedicated chaperones, including copper chaperone for superoxide dismutase (CCS), cytochrome c oxidase copper chaperone (COX17), and antioxidant protein 1 (ATOX1).202 CCS mediates copper incorporation into superoxide dismutase 1 (SOD1) to scavenge ROS and maintain cellular redox homeostasis in the cytoplasm and mitochondria, and it delivers copper to MEK1/2 to regulate kinase activity.189 COX17 is a key chaperone involved in mitochondrial copper transport to maintain mitochondrial function. Localised to the intermembrane space203 and potentially assisted by nonprotein ligands, COX17 delivers Cu⁺ to COX11 via COX19 and COA4.204,205 It also cooperates with SCO1/2 and COA6 to facilitate the formation of Complex IV, with COX16 and COA6 assisting in copper transfer, impacting tumour metabolism.206 The cytoplasmic copper chaperone (ATOX1) delivers Cu⁺ to ATP7A/B, facilitating entry into the secretory pathway and supporting cuproenzyme synthesis or promoting exocytosis.207 Furthermore, metallothioneins (MTs) safely sequester excess copper to prevent oxidative damage, utilising their high affinity for metals to increase cellular copper tolerance.208Cellular copper efflux is mediated primarily by ATP7A and ATP7B, which regulate cuproptosis. ATP7A facilitates systemic distribution, while ATP7B excretes excess copper into bile to regulate whole-body homeostasis. Their mutations cause Menkes disease and Wilson disease, respectively, as both transporters utilise ATP hydrolysis to maintain homeostasis.209,210Mechanism of cuproptosisThe core mechanism of cuproptosis involves the binding of excess copper ions to lipoylated proteins associated with the mitochondrial tricarboxylic acid (TCA) cycle, which triggers protein aggregation, the loss of iron‑sulfur (Fe‑S) cluster proteins, and mitochondrial dysfunction, ultimately leading to cell death via proteotoxic stress.18 This process relies on mitochondrial metabolism and is regulated at multiple levels by copper homeostasis, key enzymes, and signalling pathways, as detailed below (Fig. 6b).(1) Lipoylation regulation mediated by FDX1-LIAS and copper ion activationCopper exists primarily in two oxidation states, Cu⁺ and Cu²⁺, with distinct cellular activities. Extracellular Cu²⁺ modulates membrane receptor–growth factor interactions, and after membrane transfer, it is reduced to Cu⁺ by metalloreductases, influencing growth factor receptor activation via protein structure or phosphorylation.211 Under physiological conditions, the intracellular copper concentration is tightly controlled at low levels by conserved homeostatic mechanisms (e.g., CTR1-mediated uptake and ATP7A/ATP7B-mediated efflux) to prevent copper‑induced toxicity.212 However, disrupted copper homeostasis (e.g., copper ionophore-mediated influx, CTR1 overexpression, glutathione depletion, and Wilson disease) leads to excessive intracellular Cu⁺ accumulation and cuproptosis.The transport of mitochondrial copper, including that of SLC25A3 and COX17/19, is mediated mainly by COX17, which transports Cu+ from the cytoplasm to the inner mitochondrial membrane.213,214,215 During cuproptosis, elesclomol acts as a copper ionophore that transports excess Cu²⁺ into mitochondria,216 where mitochondrial ferredoxin 1 (FDX1) functions as a key upstream regulator, initiating cell death through two distinct mechanisms (Fig. 6b). First, FDX1 serves as a reductase within the mitochondrial matrix, reducing the elesclomol‑Cu²⁺ complex to cytotoxic Cu⁺. FDX1 also directly interacts with elesclomol to facilitate Cu⁺ release.217 The liberated Cu⁺ induces the aggregation of lipoylated TCA cycle enzymes, such as dihydrolipoamide S‑acetyltransferase (DLAT), resulting in proteotoxic stress and energy metabolism disruption. Second, FDX1 directly binds to lipoic acid synthase (LIAS), promoting its interaction with lipoyl carrier proteins (e.g., GCSH) to drive protein lipoylation and destabilise mitochondrial Fe–S cluster proteins.18 Notably, FDX1, LIAS (both Fe–S cluster proteins), and total cellular protein lipoylation are markedly downregulated during cuproptosis.22 This downregulation is a consequence, not a trigger, of cell death, and their loss protects cells from copper stress. In elesclomol‑Cu²⁺-induced cellular models, reductions in FDX1, LIAS, or lipoylated proteins are frequently employed as surrogate markers to monitor cuproptosis induction. However, given that FDX1 has also been identified as a direct target of elesclomol,175 future studies are needed to determine whether it serves as a universal regulator of cuproptosis, particularly under conditions induced by other copper ionophores. Emerging research indicates that the elesclomol–Cu2+ complex can still dissociate and that Cu²⁺ can be reduced to Cu⁺ in the cytosol even in FDX1-knockout cells.217 These findings suggest that FDX1 is not the only intracellular copper regulator and that elesclomol cytotoxicity may involve FDX1-independent pathways. Therefore, the reliability of FDX1/LIAS reductions as cuproptosis biomarkers in vivo or in clinical specimens requires further validation. FDX1 drives sulfinylation and disrupts the stability of Fe–S cluster proteins through LIAS binding, independent of its indirect effects on Fe–S cluster biosynthesis. FDX2, the sole electron donor in mitochondrial Fe–S cluster assembly, is a critical component of the Fe–S cluster biogenesis machinery176Lipoylation is a conserved lysine posttranslational modification limited to a few TCA cycle-associated proteins (e.g., DLAT).218 In human cells, only five mitochondrial proteins have been found to undergo this modification: DLAT, pyruvate dehydrogenase complex component X (PDHX), dihydrolipoyllysine residue succinyltransferase (DLST), dihydrolipoamide branched-chain transacylase (DBT), and glycine cleavage system H protein (GCSH).218 These proteins constitute essential components of key energy metabolism complexes, namely, the pyruvate dehydrogenase complex (PDHC, comprising DLAT and PDHX), α-ketoglutarate dehydrogenase complex (α-KGDHC, which includes DLST), branched-chain α-keto acid dehydrogenase complex (BCKDHC, containing DBT), and glycine cleavage system (GCS, featuring GCSH). Increased lipoylation of these subunits correlates with a marked increase in the enzymatic activity of their respective complexes.219 Under physiological conditions, copper binding to lipoylated mitochondrial proteins may have regulatory functions (e.g., transient allosteric modulation via mild aggregation). However, copper overload (from ionophores or pathology) disrupts this balance, leading to copper-induced protein aggregation and cell death.220(2) Other regulatory mechanisms and pathways associated with cuproptosisCuproptosis is characterised by FDX1-mediated protein lipoylation, Fe–S cluster protein depletion, and the oligomerisation of mitochondrial lipoylated proteins (e.g., DLAT). It is regulated by an intricate regulatory network composed of multidimensional signalling axes (e.g., ATF3–PER2–HSP70–AKT, TRIM21–ID1 (inhibitor of DNA binding 1)-TCF12-SLC31A, SERPINB3-MAPK), epigenetic modifications (e.g., HIF-1α, isocitrate dehydrogenase 1 (IDH1)/c-MYC/DLAT) and metabolic reprogramming (e.g., haem synthesis, Nudix hydrolase 21 (NUDT21) lactylation) in diverse diseases. Additionally, the synergistic effect between cuproptosis and ferroptosis has been identified (metal-regulatory transcription factor 1 (MTF1)), along with disease-specific resistance and sensitivity mechanisms. In the tumour microenvironment and different cancer types, under hypoxic conditions, hypoxia inducible factor1 alpha (HIF-1α) mediates cuproptosis resistance by activating pyruvate dehydrogenase kinase 1 and 3 (PDK1/3) to repress DLAT expression and binding to the hypoxia response element (HRE) of the metallothionein 2A (MT2A) promoter to promote mitochondrial copper ion chelation; it also forms a positive feedback loop with copper ions to increase its own stability.221 In group 3 medulloblastoma disease, the IDH1/c-MYC/DLAT axis epigenetically sustains high DLAT expression, increasing cellular sensitivity to copper ionophores.222 In oral squamous cell carcinoma, the ATF3–PER2–HSP70–AKT axis drives cuproptosis by inhibiting AKT to upregulate DLAT, SLC31A1 and other key molecules.223 In colorectal cancer, lncRNA PVT1 epigenetically activates FDX1 transcription to drive cuproptosis.224 In acute myeloid leukaemia, inhibition of haem biosynthesis induces abnormal copper accumulation and initiates cuproptosis, with a synthetic lethal interaction observed between haem deficiency and glycolysis.225 In gastric cancer, cuproptosis synergises with ferroptosis, and MTF1 inhibits this synergistic effect as well as cuproptosis by regulating Fe‒S cluster assembly and iron metabolism (e.g., upregulating FTH1 and activating TRIM31 to ubiquitinate NCOA4).226 In nontumour settings, excessive copper sulfate triggers myocardial cuproptosis via oxidative stress and inflammatory responses, which can be effectively alleviated by CoQ10.227 In radiotherapy-resistant tumours, the downregulation of BTB and CNC homology 1 (BACH1) expression relieves transcriptional repression of the expression of the metallothioneins MT1E/X, which in turn inhibits cuproptosis through chelation of copper ions. Radiotherapy induces tumour cell cuproptosis by activating SLC31A1 and depleting mitochondrial GSH, and copper ionophores can reverse radioresistance.228 Combining cuproptosis inducers with axis-targeted drugs, immune checkpoint inhibitors, or nanodelivery systems and using regulatory axes as biomarkers to predict therapeutic efficacy provides novel targets and strategies for precision disease treatment.Other metal homeostasis–regulated cell deathNecrosis by sodium overload (NECSO)NECSO is a novel form of necrotic cell death driven by excessive intracellular sodium accumulation. While preliminary studies have outlined its pathological cascade, this newly proposed pathway still requires broader validation by independent research groups before being universally recognised alongside fully established RCD modalities.23,24 Its core mechanism relies on the aberrant activation of transient receptor potential cation channel subfamily M member 4 (TRPM4), which disrupts mitochondrial ion homeostasis and energy metabolism, ultimately leading to cellular osmotic imbalance and rupture. NECSO is significantly distinct from apoptosis, autophagy, and other forms of RCD. It is characterised by sustained TRPM4 activation, uncontrolled sodium influx, cell swelling, membrane blebbing, and depolarisation, culminating in necrotic cell death.23,24The molecular cascade underlying NECSO can be divided into three sequential stages, as illustrated in Fig. 7. NECSO is initiated by the specific and irreversible activation of human TRPM4 (hTRPM4) via its small-molecule agonist necrocide 1 (NC1). NC1 binds with high affinity (Kd≈0.88 μM) to a vanilloid-binding pocket (VBP)-like cavity within the transmembrane (TM) region of hTRPM4, involving key residues such as W864, H908, and F936. This binding triggers sustained channel opening and robust Na⁺ influx. Notably, because of structural differences in the TM region between species, NC1 fails to activate mouse TRPM4 (mTRPM4). The resulting TRPM4-dependent Na⁺ overload drives a central mitochondrial damage cascade. Increased cytosolic Na⁺ activates the mitochondrial Na⁺/Ca²⁺ exchanger (NCLX), leading to mitochondrial Na⁺ accumulation and Ca²⁺ efflux. This Ca²⁺ efflux directly inhibits respiratory chain complex II and III activity and disrupts the TCA cycle, partly through reducing the activation of Ca²⁺-dependent enzymes (e.g., pyruvate dehydrogenase and isocitrate dehydrogenase). These events lead to severe mitochondrial dysfunction, suppression of oxidative phosphorylation (OXPHOS), and cellular energy depletion. In the downstream execution phase, energy exhaustion impairs Na⁺/K⁺-ATPase function, abolishing physiological ion gradients and increasing intracellular osmotic pressure. This osmotic imbalance promotes water influx, a process facilitated by aquaporin 4 (AQP4), leading to progressive cell swelling, plasma membrane rupture, and necrotic death. This pathological cascade can be attenuated by several interventions, including TRPM4 inhibitors (e.g., clotrimazole), NCLX antagonists (e.g., CGP37157), dihydropyridines (DHPs), or high-molecular-weight osmoprotectants (e.g., PEG8K).23Fig. 7Full size imageSchematic overview of NECSO. Under physiological conditions, the TRPM4 channel remains closed, allowing healthy mitochondria to maintain the ion gradients necessary for Na⁺/K⁺-ATPase function. Following NC1 treatment, TRPM4 channels open and mediate excessive Na⁺ influx. Excessive amounts of accumulated Na⁺ enter mitochondria via the NCLX exchanger, leading to suppressed oxidative phosphorylation (OXPHOS), decreased mitochondrial Ca²⁺, and subsequent disruption of the TCA cycle. This mitochondrial dysfunction impairs ATP production, which in turn suppresses Na⁺/K⁺-ATPase activity and collapses ionic gradients. Consequently, elevated intracellular osmotic pressure drives water influx, potentially facilitated by AQP4, resulting in cell swelling, progressive plasma membrane damage, and eventual cell lysis. By FigdrawLysozincrosisLysozincrosis is a zinc ion‑dependent form of regulated cell death.229 Its core mechanism involves aberrant lysosomal zinc release leading to cytosolic zinc overload, followed by mitochondrial swelling and dysfunction, energy metabolism collapse, and oxidative stress. Notably, this process cannot be blocked by classical inhibitors of apoptosis, ferroptosis, or other established cell death pathways.Lysozincrosis is initiated by excessive zinc release from lysosomes, which subsequently triggers severe mitochondrial damage and energy collapse. Under physiological conditions, lysosomes serve as primary intracellular zinc storage sites. Nonselective mucolipin TRP channel 1 (TRPML1) is localised to the lysosomal membrane and is permeable to both Ca²⁺ and Zn²⁺, thereby participating in intracellular zinc homeostasis. In certain tumour cells, such as metastatic melanoma cells, TRPML1 is highly expressed. Specific agonists (e.g., ML-SAs) can hyperactivate TRPML1, inducing massive zinc release from lysosomes into the cytosol and resulting in zinc overload.229 The accumulated cytosolic zinc overload targets mitochondria and inhibits electron transport chain complexes, TCA cycle enzymes, and key glycolytic enzymes (e.g., GAPDH), ultimately causing mitochondrial membrane potential (MMP) loss and rapid ATP depletion.229,230 Simultaneously, excess zinc can induce opening of the mitochondrial permeability transition pore (mPTP), releasing pro-death factors such as cytochrome C and thereby amplifying the cell death signal.230Other types of cell deathDisulfidptosisDisulfidptosis is a novel form of regulated cell death induced by disulfide stress. Its core mechanism is the abnormal intracellular accumulation of disulfide bonds, leading to cytoskeletal collapse. Notably, the induction of disulfidptosis is strongly dependent on specific metabolic states and coordinated signalling regulatory networks231 (Fig. 8). SLC7A11-mediated cystine uptake is known to suppress ferroptosis17 but promotes cell death under glucose starvation conditions. In 2023, Liu et al. named this form of cell death modality “disulfidptosis”.231 The core trigger of disulfidptosis is high expression of SLC7A11, which drives the excessive import of extracellular cystine. Under glucose starvation or inhibition of glucose transporters (GLUTs), the pentose phosphate pathway (PPP) is compromised, resulting in reduced NADPH production. This severely compromises the intracellular reductive capacity, resulting in the accumulation of unreduced cystine. Consequently, the thiol balance is disrupted, inducing a state of disulfide stress. Mechanistically, this stress preferentially targets actin cytoskeleton proteins, including actin (ACTB), filamin-A/-B (FLNA/B), myosin-9 (MYH9), and talin-1 (TLN1). The formation of aberrant disulfide bonds within these proteins induces F-actin contraction, detachment from the plasma membrane, and eventual cytoskeletal collapse. Disulfidptosis is coordinately regulated by multiple pathways: (1) The Rac–WAVE regulatory complex (WRC)–actin-related protein 2/3 (Arp2/3) axis facilitates actin polymerisation, providing structural substrates for disulfide crosslinking and thus facilitating disulfidptosis.231 (2) Metabolic enzymes play opposing roles: Glucose-6-phosphate dehydrogenase (G6PD) generates NADPH via the PPP to antagonise disulfide accumulation, whereas lactate dehydrogenase B (LDHB) binds and inhibits G6PD activity, thereby promoting disulfidptosis. (3) The STAT3–LDHB pathway increases disulfide stress by transcriptionally activating LDHB expression.231,232 (4) ER stress is concurrently activated during glucose deprivation, and the inhibition of ER stress exacerbates disulfide accumulation and disulfidptosis, whereas direct induction of ER stress exerts a protective effect.233Fig. 8Full size imageSchematic diagram of SLC7A11-mediated disulfidptosis under glucose deprivation. Disulfidptosis is driven by high rates of SLC7A11-mediated cystine uptake in SLC7A11-high cancer cells under glucose-deprived conditions. Under glucose deprivation, the lack of glucose-derived NADPH, a critical reducing agent in the pentose phosphate pathway (PPP), creates a vulnerable metabolic state. In SLC7A11-high cancer cells, elevated cystine uptake through SLC7A11 leads to intracellular NADPH depletion, excessive accumulation of cystine and other disulfide molecules, and aberrant disulfide bond formation among actin cytoskeletal proteins. This culminates in the collapse of the actin network and ultimately triggers disulfidptosis. Rac1-WRC-mediated branched actin polymerisation and pseudopod formation may provide a structural scaffold that supports disulfide crosslinking between actin cytoskeletal proteins, thereby facilitating the progression of disulfidptosis. In exhausted CD8+ T cells, STAT3 facilitates LDHB expression and the LDHB–G6PD interaction, thus triggering G6PD inactivation and mediating disulfidptosis. By FigdrawAlkaliptosisAlkaliptosis is a type of RCD driven by intracellular alkalinisation and is characterised by pH-dependent regulated necrosis. Its defining feature is a sustained elevation in cytosolic pH, which sets it apart from other forms of RCD that typically involve intracellular acidification and are regulated in collaboration by ion channels and transport proteins both inside and outside the cells.234 Its core mechanism involves the disruption of cellular pH homeostasis and lysosomal function by inducer molecules (e.g., JTC801) that activate both the NF-κB–CA9 pathway and the ATP6V0D1–STAT3 pathway, ultimately leading to the selective induction of cancer cell death.31 Alkaliptosis is selectively cytotoxic towards cancer cells, including pancreatic ductal adenocarcinoma and melanoma cells, whereas normal cells (e.g., hepatocytes and fibroblasts) exhibit strong resistance to this cell death process.31 Mechanistically, alkaliptosis is distinct from apoptosis (no caspase activation or PARP cleavage), necroptosis (no RIPK3‑MLKL interaction), ferroptosis (no GPX4 degradation), and autophagy (no ATG gene dependence). Consistent with these differences, inhibitors targeting these classical death pathways fail to block alkaliptosis.31 To date, JTC801 is the only known specific inducer of alkaliptosis, and its activity depends on key structural moieties, including the amine groups at the 4- and 6-positions of its quinoline ring and the phenoxymethyl group. JTC801 exerts its pro-alkaliptotic effects through two distinct molecular axes, independent of its known target OPRL1. First, JTC801 promotes IKKβ phosphorylation, leading to NF‑κB activation. Activated NF-κB induces the transcriptional suppression and degradation of carbonic anhydrase 9 (CA9), a key pH‑regulatory enzyme, thereby directly disrupting intracellular acid‑base homeostasis and driving alkalinisation.31 Second, JTC801 directly binds to the V‑ATPase V0 subunit D1 (ATP6V0D1) and stabilises this protein by inhibiting proteasomal degradation. Stabilised ATP6V0D1 facilitates the formation of an ATP6V0D1‑STAT3 complex, which sequesters STAT3 and prevents its phosphorylation and nuclear translocation. This sequestration impairs lysosomal acidification, increases lysosomal membrane permeability, and thereby potentiates cell death. Depletion of ATP6V0D1 attenuates JTC801‑induced alkaliptosis, whereas STAT3 inhibition restores sensitivity.235 The cancer-selective lethality of alkaliptosis stems from the differential expression of pH-regulatory molecules (e.g., CA9 and ATP6V0D1) between malignant and normal tissues. This difference in expression enables JTC801 to preferentially induce profound alkalinisation in cancer cells, culminating in the induction of lethal alkaliptosis.OxeiptosisOxeiptosis is a recently identified ROS-sensitive, caspase-independent, and noninflammatory regulated cell death pathway. Its activation is predominantly triggered by the accumulation of high concentrations of ROS, with hydrogen peroxide (H₂O₂) as the most prominent inducer.30 Oxeiptosis is characterised by distinct, caspase-independent morphological changes that resemble those of apoptosis, and these changes are closely associated with ROS accumulation. Specifically, affected cells exhibit plasma membrane blebbing, a typical apoptotic feature, along with nuclear enlargement, rounding, swelling, and partial detachment from the culture substrate. TEM further confirmed that membrane blebbing and nuclear enlargement are key ultrastructural hallmarks of oxeiptosis. These features distinguish oxeiptosis from classical cell death pathways while retaining apoptosis-like morphological traits.30,236The core molecular mechanism of oxeiptosis is mediated by the KEAP1–PGAM5–AIFM1 signalling axis, which is indispensable for the execution of this cell death process.30 The knockdown of any of these three molecules can significantly abrogate oxeiptosis induction. Specifically, under conditions of high oxidative stress, KEAP1 undergoes a functional switch from mediating cytoprotective NRF2 activation to dissociating from its binding partner PGAM5, a mitochondrial serine–threonine phosphatase.237 The released PGAM5 then translocates into the mitochondria, where it specifically dephosphorylates AIFM1 at the Ser116 residue (a hallmark molecular event of oxeiptosis). Dephosphorylated AIFM1 retains its localisation in the mitochondria to trigger characteristic cellular changes, such as membrane blebbing and nuclear enlargement, ultimately leading to cell death.238 Notably, oxeiptosis is distinct from classical cell death pathways, including apoptosis, necroptosis, ferroptosis, and autophagy. Oxeiptosis is not affected by specific inhibitors of these other pathways; instead, it can be completely blocked by ROS scavengers (e.g., N-acetyl-l-cysteine and catalase) that specifically neutralise intracellular ROS or H₂O₂.236ParthanatosParthanatos, a distinct form of poly (ADP-ribose) polymerase-1 (PARP-1)-dependent cell death, is an irreversible, programmed necrotic pathway activated by PARP‑1 hyperactivation following DNA damage.239 Major triggers, such as oxidative/nitrosative stress (ROS/ONOO⁻), DNA‑damaging agents, excitotoxicity, ischaemia–reperfusion, or metabolic disturbances such as hyperglycaemia‑hyperlipidaemia, all induce DNA lesions either directly or indirectly.34 These stimuli can initiate Parthanatos through various mechanisms; for instance, impaired Fe–S cluster synthesis disrupts mitochondrial function, elevates ROS levels, exacerbates DNA damage, and thereby promotes the progression of this death pathway.240 Morphologically and biochemically, Parthanatos is defined by caspase‑independent execution and the absence of characteristic cellular swelling. Key signatures include the nuclear translocation of apoptosis‑inducing factor (AIF) and extensive DNA fragmentation mediated by PAAN/MIF.239 During this process, the nuclei undergo chromatin condensation and large‑scale DNA cleavage (∼50 kb) without karyorrhexis or apoptotic DNA laddering. The integrity of the plasma membrane is lost, which is marked by the externalisation of phosphatidylserine but no formation of apoptotic bodies or cellular swelling, features that distinguish Parthanatos from apoptosis and necroptosis. Mitochondria undergo membrane potential depolarisation, cristae disorganisation, and fission while maintaining outer membrane integrity, permitting the release of full‑length 62‑kDa AIF. Additionally, typical caspase‑associated morphological changes and prominent cytoplasmic vacuolisation are absent, although structural interactions linked to PAR‑polymer‑dependent mitochondrial‑nuclear signalling are evident.239After DNA damage, PARP-1 is hyperactivated and consumes NAD⁺ to synthesise abundant poly(ADP-ribose) (PAR) polymers.241 Under physiological conditions, PARP-1 participates in DNA repair. However, excessive activation leads to the accumulation of PAR, which functions as a “death signal” that directly drives cell death rather than merely through energy depletion.242 PAR subsequently translocates from the nucleus to mitochondria, where it binds to the mitochondrion-localised AIF to trigger the release of full-length AIF and its nuclear translocation.239 This pathway is tightly regulated by multiple factors, including PAR glycohydrolase (PARG)-mediated PAR degradation and hydrogen sulfide (H₂S)-induced cysteine desulfurase (NFS1) S-sulfhydration at C383, which restores Fe–S cluster synthesis.240 Notably, AIF itself lacks nuclease activity and relies entirely on PAAN/MIF to perform DNA cleavage, which constitutes the hallmark terminal event of Parthanatos.243 Consistent with its morphological and biochemical features, Parthanatos is caspase-independent and lacks cellular swelling, distinguishing it from apoptosis and necroptosis.239 It can be blocked by PARP inhibitors (e.g., DPQ and ABT-888) and the PAAN/MIF nuclease-specific inhibitor PAANIB-1. Moreover, Parthanatos plays a critical role in various diseases, including stroke, Parkinson’s disease, and diabetic cardiomyopathy, making PARP-1, AIF, and PAAN/MIF important potential therapeutic targets.243MitoxyperiosisMitoxyperiosis is a newly proposed distinct lytic cell death mechanism induced by innate immune signalling (e.g., TLR ligand/TNF stimulation) and metabolic disruption (e.g., carbon starvation, innate immune activation and metabolic disruption (IIAMD)).33 Originally presented as being independent of caspase activity and distinct from pyroptosis, PANoptosis, necroptosis, ferroptosis, and oxeiptosis, this framework is still evolving and needs broader independent validation. Crucially, a closer mechanistic evaluation indicated that mitoxyperiosis shares several overlapping features with established pathways, including apoptosis, secondary pyroptosis, and ferroptosis. The involvement of caspase activation, lipid peroxidation, and partial sensitivity to ferrostatin-1 treatment implies that it may intertwine with these classical modalities, and further investigation is needed to demarcate its exact boundaries.Despite these mechanistic overlaps, the primary morphological signature proposed to define mitoxyperiosis involves a characteristic “oxidative stress–mitochondria–plasma membrane” ternary spatial pattern. During mitoxyperiosis, mitochondria undergoing BAX/BAK1/BID-dependent oxidative stress maintain prolonged plasma membrane contact, leading to local oxidative damage. The morphological hallmarks of mitoxyperiosis include sustained mitochondrial–plasma membrane contact (lasting >20 min), leading to local oxidative damage. Other reported hallmarks include abnormal cytoskeletal and pseudopod dynamics, localised plasma membrane injury and rupture, and the colocalisation of lipid peroxidation foci with mitochondria at the cell periphery.33,244The process of mitoxyperiosis is propelled by upstream signalling cascades: IIAMD engages the TLR-MyD88 or TNFR pathway, which relies on BAX/BAK1/BID to elicit mitochondrial dysfunction (manifested as diminished membrane potential) and GSH exhaustion, consequently precipitating intracellular oxidative stress characterised by aberrant ROS accumulation. Notably, mTORC2, rather than mTORC1, acts as a central regulatory hub: its activation suppresses RhoA-GTP activity and reduces actin polymerisation, thereby impairing lamellipodium retraction and attenuating cytoskeletal dynamics. Subsequently, oxidatively stressed mitochondria are retained in sustained proximity to the plasma membrane, a hallmark event that permits the sustained delivery of ROS to localised membrane domains and induces site-specific oxidative damage (e.g., lipid peroxidation denoted by 4-HNE adduct accumulation), which constitutes the defining characteristic of mitoxyperiosis. Ultimately, this site-restricted membrane damage culminates in plasma membrane permeabilisation and cell lysis. Notably, mTOR inhibition restores cytoskeletal function, which promotes lamellipodium formation. This mobilisation facilitates the displacement of mitochondria away from the plasma membrane, preserving membrane integrity and attenuating mitoxyperiosis.33,244NETotic cell deathNeutrophil extracellular trap (NET)-associated cell death, termed NETosis, was first formally defined by Fuchs et al. in 2007245 after the initial discovery of bactericidal neutrophil extracellular traps by Brinkmann et al. in 2004.246 NETosis is initiated upon neutrophil activation and is characterised by the release of NETs, which are composed of decondensed chromatin, histones, and intracellular functional proteins (e.g., neutrophil serine proteases (NSPs) and myeloperoxidase (MPO)), ultimately leading to cellular dysfunction or death.247 On the basis of activation mechanisms and cellular consequences, NETosis is primarily classified into two subtypes: Suicidal NETosis, which accounts for 20–25% of NET release events; depends on NADPH oxidase (NOX) activation and nuclear DNA release; and is typically induced by phorbol-12-myristate-13-acetate (PMA), LPS, and bacteria.248,249 In contrast, vital NETosis involves vesicular transport of mitochondrial DNA and is triggered by TLR4-activated platelets and bacterial components, allowing cells to retain phagocytic and migratory functions.248,250Neutrophil activation signals are categorised into two core pathways on the basis of NOX activity. In the NOX-dependent pathway, ligand binding (e.g., LPS) to TLR/IgG-Fc receptors triggers endoplasmic reticulum calcium release, activating protein kinase C (PKC) to phosphorylate the NOX2 subunit, assemble NADPH oxidase, and generate ROS that initiate downstream cascades.245,246 The NOX-independent pathway is directly induced by calcium ionophores (e.g., ionomycin), which increase intracellular Ca²⁺ levels to activate peptidylarginine deiminase 4 (PAD4) and initiate chromatin relaxation without the involvement of ROS.251 Additionally, cytokines (TNF, IL-8), immune complexes, and chemokines (e.g., G-CSF) secreted by tumour cells can increase NETosis.252 NET formation depends on glycolysis for energy supply, with exogenous glucose and glutamine significantly promoting NET release.253 Chromatin decondensation relies on two synergistic mechanisms: First, elevated intracellular Ca²⁺ activates PAD4, which converts arginine residues on histones to citrulline, reducing histone–DNA binding affinity.254 Second, NSP-mediated histone hydrolysis and cytoskeletal remodelling occur. Upon ROS activation, neutrophil elastase degrades F-actin to relieve nuclear membrane constraints and then is translocated to the nucleus to hydrolyse histones (e.g., H3 and H4) alongside proteinase 3 (PR3), accelerating chromatin decondensation.255 At the level of release mechanisms, suicidal NETosis is driven by ROS-mediated disruption of the lysosomal and nuclear membranes; then, decondensed chromatin fuses with cytoplasmic components and is released via plasma membrane rupture.245 In vital NETosis, mitochondrial DNA is packaged into vesicles budding from the nuclear envelope, which fuse with the plasma membrane for release while maintaining membrane integrity.248 The role of GSDMD in NET extrusion remains controversial. Some studies suggest that GSDMD cleavage facilitates nuclear pore formation to promote DNA release, whereas recent evidence indicates that NETosis can occur entirely independently of GSDMD cleavage and pyroptotic pathways.256,257Entotic cell deathEntotic cell death, named entosis, is a type of RCD characterised by the active invasion of viable homotypic cells into adjacent cells via adhesion molecules, leading to the formation of cell-in-cell (CIC) structures.258 The typical morphological features of entosis include a crescent-shaped nucleus in host cells and the internalised cell enclosed by a large entotic vacuole, which distinguishes it essentially from cell cannibalism and emperipolysis.259The Rho–ROCK–actomyosin contraction axis is the core molecular regulatory pathway of entosis, with the Rho family of small GTPases (RhoA/B/C, Rac1-3, and Cdc42) as key regulators. Their activity cycle (GTP-bound active state/GDP-bound inactive state) is modulated by guanine nucleotide exchange factors (GEFs) and GTPase-activating proteins (GAPs). Rho-GTP activates downstream ROCK, which phosphorylates myosin light chain 2 (MLC2) and inhibits its phosphatase, thereby promoting the interaction between actin and myosin II to increase contractility.260 RhoGAP p190A locally inhibits Rho activity to form a Rho activity gradient, which generates mechanical tension, driving cellular invasion together with the assembly of a uropod-like actin-rich structure at the posterior pole of the invading cell.26,261,262 Intercellular adherens junctions are essential for entosis: E-cadherin, P-cadherin and α-catenin are core molecules for CIC structure formation, and their deletion completely abrogates this process. The ring-like junction structure formed by E-cadherin and vinculin persists throughout the entire invasion process, stabilising cell‒cell contact and coordinating actomyosin contraction.262,263 Invasive efficiency is also affected by the difference in membrane stiffness between internalised and host cells, and cells with higher RhoA activity and a difference in stiffness sufficient to overcome the energy barrier are more prone to complete invasion.263 Collectively, the initiation, progression and fate regulation of entosis are jointly modulated by microenvironmental signals, molecular pathways and cellular mechanical properties and are closely associated with tumours and other diseases.Lysosome-dependent cell deathLysosome-dependent cell death (LCD), also referred to as autolysis, is characterised by primary lysosomal membrane permeabilisation (LMP) as the initiating event. Following LMP, lysosomal cathepsins are released into the cytosol to trigger cell death, with the optional involvement of mitochondrial outer MOMP and caspases.25,34 LMP can be induced by diverse endogenous stimuli, including oxidative stress, calcium overload, p53/STAT3-mediated signalling and exogenous factors (e.g., lysosomotropic agents, TRAIL, and pathogen infection).264 This process is regulated by BCL-2 family proteins,265 ESCRT-mediated membrane repair and sphingolipid metabolism.266The execution of LCD relies on cytosolic cathepsins that induce cell death through direct degradation of cellular structural/functional proteins, cleavage of BID to activate MOMP, degradation of antiapoptotic BCL-2 family members, and cross-activation of other RCD pathways (e.g., necroptosis, pyroptosis, and ferroptosis).267 This process is tightly controlled by endogenous inhibitors (e.g., cystatins and serpins), lysosomal membrane stabilisers (e.g., HSP70), and pharmacological agents targeting cathepsins. Importantly, it plays critical roles in physiological tissue remodelling and various pathological processes, including inflammation, neurodegeneration, and cancer.25,268,269,270Autophagic cell deathAutophagic cell death (ACD) is a subtype of RCD that relies on the autophagic machinery composed of core autophagy-related gene (Atg) products as lethal effectors.271 Its key characteristic is that abnormal activation of the autophagic pathway directly mediates cell death.272 The initiation of ACD depends on the activation of the autophagic initiation complex (ULK1/2–ATG13–FIP200) and the abnormal enhancement of autophagic flux triggered by stress signals such as nutrient deprivation and growth factor signal inhibition. Additionally, long noncoding RNAs and glucocerebrosidase can further promote the activation of lethal autophagy by regulating Atg gene expression or lipid metabolism.273,274 Changes in the cellular morphology of ACD are characterised by the accumulation of autophagy-related structures and the progressive degradation of cellular components.272 Among these features, extensive cytoplasmic vacuolisation, formed by the fusion of autophagosomes and autolysosomes, is the most typical morphological marker of ACD.272The execution of ACD involves multiple coordinated molecular mechanisms, with the selective degradation of antiapoptotic proteins (e.g., BCL-2 and c-IAP1/2) and key metabolic proteins through the autophagic pathway.275 In addition, excessive mitophagy-induced energy metabolism disorders, ferritinophagy-mediated ferroptosis activation,276 and the cross-regulation of autophagosomes as assembly platforms for necrosomes collectively amplify the lethal effect.277 This process is regulated by transcription factors (e.g., TFEB), the MAPK signalling pathway, and subtype-specific mechanisms (e.g., plasma membrane Na⁺/K⁺-ATPase activity dependent on autosis).34 ACD plays crucial roles in various pathophysiological processes, including developmental remodelling, tumour therapeutic response, and ischaemic injury.Crosstalk between cell death pathwaysMetal-related death networkFerroptosis and cuproptosisWhile ferroptosis and cuproptosis differ significantly in terms of their initiating ions and core effector proteins, emerging research has revealed a substantial and sophisticated “dialogue” between these two modalities at critical cellular metabolic hubs and ion homeostasis (Fig. 9). This interaction constitutes not merely a mechanistic overlap but also a multilayered network encompassing shared subcellular targets, interlinked antioxidant defences, interdependent metabolic pathways, and the consequent potential for synergistic therapeutic intervention.Fig. 9Full size imageInteractive crosstalk between ferroptosis and cuproptosis. These two metal-dependent cell death processes form a complex interactive regulatory network in cellular metabolism and ion homeostasis. Mutual influence of metabolic flows: Copper overload disrupts the TCA cycle and Fe–S cluster biogenesis, weakens antioxidant defences, creates conditions for lipid peroxidation, and promotes ferroptosis. Both forms of cell death can induce mitochondrial ROS production and synergistically amplify cytotoxicity. Shared regulatory nodes: Copper ions drive ferroptosis by degrading GPX4 via TAX1BP1-mediated autophagy; ferroptosis inducers stabilise FDX1 and inhibit GSH synthesis, increasing the aggregation of acylated DLAT. Glutathione depletion may be a common mechanism between these two types of cell death. Interplay of ion homeostasis: MTF1 regulates iron-copper metabolism and Fe–S cluster stability by upregulating FTH1, TRIM31 and ISCA2, thereby simultaneously inhibiting both ferroptosis and cuproptosis. By Figdraw(1) Mutual influence of metabolic flow: Cuproptosis can prime cells for ferroptosis by disrupting the TCA cycle and impairing Fe–S cluster protein biogenesis; these processes directly compromise mitochondrial bioenergetics and metabolic integrity. This mitochondrial dysfunction undermines the cell’s antioxidant defences, exemplified by the compromised generation of NADPH via the pentose phosphate pathway. Redox imbalance indirectly amplifies cellular vulnerability to lipid peroxidation, thereby fostering a metabolic milieu conducive to ferroptosis initiation.278 Copper overload directly induces ROS production within the mitochondria, whereas ferroptosis elicits a robust lipid peroxidation cascade that itself generates substantial amounts of ROS. These dual ROS sources converge in the mitochondrial compartment, exerting an additive effect that synergistically drives cells beyond the death threshold.279(2) Shared regulatory nodes: Copper ions bind to the cysteine residues (C107 and C148) of GPX4, promoting GPX4 degradation via the Tax1-binding protein 1 (TAX1BP1)-mediated autophagic pathway and consequently driving ferroptosis.280 Copper chelators (tetrathiomolybdate, TTM; triethylenetetramine, TETA) specifically suppress erastin-induced ferroptosis without affecting other modes of cell death.280 Conversely, the induction of ferroptosis can potentiate cuproptosis. The combination of ferroptosis inducers (sorafenib or erastin) with elesclomol significantly increased lipoylated DLAT aggregation in hepatoma cells, outperforming elesclomol alone.281 Mechanistically, both sorafenib and erastin stabilise FDX1 by inhibiting its mitochondrial protease-mediated degradation and simultaneously inhibiting the synthesis of GSH, suggesting that the depletion of GSH might be a common mechanism underlying these two types of cell death. BSO, an inhibitor of GSH synthesis, is also known to induce both ferroptosis282 and cuproptosis.18(3) Interplay of ion homeostasis: MTF1, a central regulator of metal ion homeostasis, functions as a critical nexus linking ferroptosis and cuproptosis by bidirectionally regulating iron and copper metabolism. (1) It regulates key iron metabolism pathways. FTH1 binds intracellular free Fe²⁺ to reduce Fe²⁺-mediated lipid peroxidation.162 MTF1 directly binds to the FTH1 promoter to transcriptionally upregulate FTH1 expression, thereby lowering free Fe²⁺ levels and inhibiting ferroptosis induced by agents such as FINO2/ES-Cu.226 NCOA4, a key ferritinophagy cargo receptor, promotes ferritin degradation and Fe²⁺ release.162 MTF1 transcriptionally activates the E3 ubiquitin ligase TRIM31, which catalyses NCOA4 ubiquitination and degradation, reducing ferritinophagy, stabilising FTH1, and indirectly inhibiting ferroptosis.226 (2) It regulates Fe‒S cluster assembly and copper ion binding. Fe–S clusters are key complexes for the mitochondrial respiratory chain and iron metabolism. Excess copper can compete for binding to the metal-binding sites of Fe–S cluster assembly proteins, inhibiting Fe–S cluster biogenesis and exacerbating cuproptosis.18 MTF1 directly binds to the iron‒sulfur cluster assembly protein 2 (ISCA2) promoter and transcriptionally activates ISCA2, a core component of the [4Fe-4S] cluster assembly machinery.226 ISCA2 stabilises Fe–S clusters, antagonises copper ion-induced damage to Fe–S clusters, maintains iron homeostasis by regulating the iron-starvation response, and indirectly inhibits cuproptosis.283 Additionally, ISCA2 upregulates FTH1 to reduce Fe²⁺ accumulation and inhibit ferroptosis.284 Conversely, ISCA2 knockdown downregulates DLAT and FTH1 expression, increases DLAT protein aggregation, elevates intracellular Fe²⁺ levels and cooperatively activates both cuproptosis and ferroptosis.226DMT1, which is primarily responsible for the transport of Fe²⁺, may also facilitate intracellular divalent copper uptake. Iron deficiency upregulates DMT1 expression without altering the function of the copper transporter CTR1. In an iron-deficient rat model, cerebrospinal fluid (CSF) and brain parenchyma copper levels increased by 55%, and ventriculocisternal perfusion experiments demonstrated increased copper clearance efficiency.285 These findings reveal the complex regulatory mechanisms of the metal ion metabolic network in neurodegenerative diseases.Ferroptosis and necroptosisFerroptosis and necroptosis share critical intersections despite their divergent molecular machinery. Necroptosis is a caspase-independent pathway driven by the RIPK1–RIPK3–MLKL axis, leading to plasma membrane rupture.95 In contrast, ferroptosis is an iron-dependent form of cell death characterised by the lethal accumulation of lipid peroxides and the failure of GPX4 to detoxify lipid ROS.286 Accumulating evidence has revealed that these two pathways are functionally coupled, with intricate regulatory crosstalk and mutual compensation. Functional interchangeability between ferroptosis and necroptosis has been demonstrated in the pathogenesis of acute kidney injury.287 Genetic ablation of Acsl4, a key gene that governs ferroptosis, renders cells resistant to ferroptosis while increasing their susceptibility to necroptosis. Conversely, MLKL knockout in mice increases the sensitivity of renal tubular epithelial cells to ferroptosis.287Specific knockout of the Gpx4 gene in murine haematopoietic cells induces severe anaemia, a phenotype that can be rescued by concurrent ablation of the Ripk3 gene.288 These results indicate that GPX4 deficiency not only triggers lipid peroxidation but also induces RIPK3-dependent necroptosis. Moreover, heat shock protein 90 (HSP90) acts as a dual modulator of these two death pathways: It binds and stabilises RIPK1 to promote necroptosis while simultaneously mediating GPX4 degradation via chaperone-mediated autophagy to facilitate ferroptosis.289 Thus, the inhibition of HSP90 exerts a synergistic effect by blocking both forms of cell death. Iron overload, an initiator of ferroptosis, also promotes necroptosis during ischaemia‒reperfusion injury.290 Iron overload induces mPTP opening, which exacerbates RIPK1 phosphorylation and thereby primes cells for necroptosis. Collectively, these findings emphasise the need to consider the coordinated effects of ferroptosis and necroptosis in related diseases; blockade of one pathway may drive cells to “escape” via the alternative death route, highlighting the clinical significance of targeting their crosstalk.Furthermore, the metabolic context intricately dictates the signalling and crosstalk between these cell death modalities. Mechanistic distinctions exist between chemical-induced and metabolite-induced ferroptosis. Classical chemical inducers, such as RSL3, trigger ferroptosis and lipid peroxidation without inducing the significant phosphorylation of RIPK3 and MLKL.101,172 In contrast, metabolites derived from the catabolism of polyamines or cyclophosphamide, such as acrolein, initiate ferroptosis through multipronged mechanisms, including the direct modification of ferroptosis-related proteins (FTH1) and the decreased expression of antioxidant defences such as GPX4 and TXN1.101 Importantly, during metabolite-induced ferroptosis, these metabolic byproducts concurrently activate necroptotic signalling markers, driving the robust phosphorylation of RIPK3 and MLKL. However, despite the presence of these activated markers, the execution of necroptosis is halted because the metabolites suppress the essential oligomerisation of MLKL and its subsequent translocation to the plasma membrane, potentially by targeting HSP90.291 This highlights a complex layer of metabolic regulation in which specific byproducts act as a switch, driving cells towards ferroptosis while simultaneously blocking necroptotic signalling.Ferroptosis and pyroptosisPyroptosis and ferroptosis, while morphologically and biochemically distinct forms of RCD, involve intricate crosstalk that is tightly orchestrated by shared regulatory factors and overlapping signalling cascades, among which ROS constitute the most important core hub.292 During pyroptosis, either inflammasome activation or gasdermin family protein-mediated membrane pore formation directly impairs mitochondrial integrity and function, triggering the robust generation of mitochondrial ROS (mtROS).293 These mtROS not only directly promote ferritinophagy to release iron ions into the cytoplasm, expanding the labile iron pool (LIP),276 but also inhibit the activity of antioxidant factors in the ferroptosis pathway, creating conditions for the initiation of ferroptosis.294 Conversely, lipid peroxidation products (such as 4-HNE) generated during ferroptosis can modify key sites of the NLRP3 inflammasome, promoting its assembly and activation, thereby triggering caspase-1-dependent pyroptosis and forming a ROS-mediated bidirectional amplification loop.295 Autophagy acts as a core regulatory hub with bidirectional and selective regulatory effects on both. On the one hand, autophagy can inhibit the excessive activation of pyroptosis by degrading critical pyroptotic mediators (e.g., NLRP3 inflammasome components and GSDMD) or eliminating ROS-overproducing damaged mitochondria. On the other hand, autophagy can directly promote the occurrence of ferroptosis through pathways such as ferritinophagy (releasing iron ions), mitophagy (regulating mitochondrial ROS production), and chaperone-mediated autophagy (CMA, degrading GPX4).296 This selective regulation enables autophagy to serve as a key switch for the transition between pyroptosis and ferroptosis. The cGAS–STING pathway regulates pyroptosis and ferroptosis through multiple mechanisms. Upon activation, this pathway can recruit NLRP3 to the endoplasmic reticulum and mediate its deubiquitination, a critical step for inflammasome activation that initiates pyroptosis.297Moreover, pyroptosis-induced membrane rupture and ferroptosis-induced mitochondrial damage can release nuclear DNA and mtDNA, respectively, which in turn activate the cGAS–STING pathway.298,299 In addition, the NF-κB pathway can promote pyroptosis by transcriptionally regulating NLRP3 and pro-IL-1β while upregulating the expression of SLC7A11 to increase GSH synthesis and inhibit ferroptosis.300,301This interplay has significant pathological implications in diseases such as cancer, cardiovascular diseases, and chronic inflammation. In cancer, the transition between the two modalities is closely associated with therapeutic resistance, and targeted regulation can increase therapeutic efficacy.302 In cardiovascular diseases such as heart failure and atherosclerosis, pyroptosis and ferroptosis synergistically exacerbate tissue damage.303 Therefore, targeting the key pathways involved in their interplay, including mtROS, autophagy, and the cGAS–STING pathway, has emerged as a potential therapeutic strategy for related diseases.Ferroptosis, cuproptosis and disulfidptosisFerroptosis, cuproptosis, and disulfidptosis are intricately interconnected through shared metabolic regulators, intersecting pathways, and mutual regulatory mechanisms. Their interplay hinges on core metabolic networks involving cystine–GSH metabolism, ROS homeostasis, and glucose–NADPH metabolism, providing crucial theoretical support for combination cancer therapy. The interplay of these three modalities is mediated by a core regulatory network consisting of three key factors: SLC7A11, GSH, and p53. SLC7A11, a cystine/glutamate antiporter that plays a multifaceted role, imports cystine for GSH synthesis to suppress ferroptosis.17 In contrast, under glucose deprivation or NADPH depletion, its activity leads to aberrant cystine accumulation, triggering disulfidptosis.231 The synthesised GSH chelates intracellular copper ions, thereby inhibiting cuproptosis.18 GSH serves as a central hub in this network. It acts as an essential cofactor for GPX4 to degrade lipid hydroperoxides and block ferroptosis and functions as an intracellular copper chelator to restrain cuproptosis.18 Its synthesis relies on SLC7A11-imported cystine, and NADPH depletion during disulfidptosis indirectly reduces GSH production,231 increasing sensitivity to ferroptosis. p53 regulates these pathways bidirectionally, primarily through transcriptional control of genes such as SLC7A11; p53 deficiency elevates intracellular ROS levels, increasing susceptibility to both ferroptosis and cuproptosis, and indirectly influences disulfidptosis by modulating glucose metabolism.304,305 These core factors further integrate three intersecting pathways—cystine metabolism, ROS homeostasis, and glucose–NADPH metabolism—to form a coordinated and counterbalanced network. Cystine metabolism serves as the common entry point. When NADPH is sufficient, cystine is preferentially directed towards GSH synthesis to suppress ferroptosis; conversely, NADPH deficiency redirects cystine towards the induction of disulfidptosis.231,306 Excess copper ions disrupt this metabolic equilibrium, impairing GSH production and establishing a self‑amplifying cascade of “copper accumulation–GSH depletion–enhanced ferroptosis and disulfidptosis”.18 ROS homeostasis acts as the central driver. Ferroptosis relies on iron-dependent lipid peroxidation and ROS accumulation5; cuproptosis involves Cu⁺-induced ROS generation and direct mitochondrial damage18,307; and disulfidptosis indirectly promotes ROS buildup via disulfide stress.231 The ROS-generating pathways of these three types of death are complementary and mutually amplifying. Glucose–NADPH metabolism provides critical metabolic support. Cells with high SLC7A11 expression require glucose to fuel the PPP for NADPH production to maintain cystine reduction, and glucose deprivation directly triggers disulfidptosis.231 Moreover, NADPH also suppresses ferroptosis by reducing GSSG to GSH and by supporting the FSP1-mediated regeneration of reduced coenzyme Q (CoQH₂) from CoQ, and its metabolic dysregulation also influences cellular sensitivity to cuproptosis.18,308,309Integration of signalling pathwaysPyroptosis, apoptosis and necroptosis (PANoptosis)In recent years, studies have identified a distinct, coordinated form of cell death termed PANoptosis310 (Fig. 10). Characterised by the concurrent activation of apoptosis, pyroptosis and necroptosis, this unique cell death modality cannot be abrogated by inhibiting only one of these individual pathways.310 Unlike the fortuitous co-occurrence of different cell death programmes, PANoptosis is a genetically programmed, coordinated process orchestrated by the PANoptosome, a multicomponent protein complex. The scaffold-like architecture of this complex plays a pivotal role: It integrates both sensor and effector molecules, ensuring that if a single signalling node is impaired by pathogenic insult, the cell can still undergo lysis and execute its suicide programme to eliminate the threat.311Fig. 10Full size imageMolecular mechanisms of ZBP1-mediated PANoptosis. Upstream activation: Viral infections and sterile stimuli (such as alcohol and IFN) serve as the primary triggers. Sterile stimuli promote the accumulation of Z-nucleic acids (Z-NAs) via the JNK signalling pathway, which subsequently activates ZBP1. Simultaneously, these stimuli inhibit ADAR1 (a host negative regulator that normally prevents Z-NAs from activating ZBP1). The relief of ADAR1 inhibition enables Z-NAs to act as direct ligands, leading to the robust activation of ZBP1. Downstream execution (left—canonical pathway): Activated ZBP1 mediates the assembly of a canonical PANoptosome complex comprising ZBP1, caspase-8, RIPK3, caspase-1, and NLRP3. This multimeric complex drives diverse death cascades: Caspase-1 cleaves GSDMD into its pore-forming N-terminal fragment (N-GSDMD), apoptotic caspase-3/7 cleaves GSDME into N-GSDME, and RIPK3 mediates the phosphorylation of MLKL (p-MLKL). Downstream execution (right—alternative pathway): An alternative, caspase-1/NLRP3-independent PANoptosome can also assemble, consisting solely of ZBP1, caspase-8, and RIPK3. This complex induces membrane rupture primarily through caspase-3/7-mediated GSDME cleavage and MLKL phosphorylation. By FigdrawThe assembly of the PANoptosome is triggered by specific innate immune sensors, most notably ZBP1. During viral infections (e.g., influenza A virus or SARS-CoV-2), ZBP1 detects viral nucleic acids and recruits the adaptor RIPK3, caspase-8, and NLRP3 inflammasome components to form the ZBP1-PANoptosome.312 Upon activation, the NLRP3 inflammasome assembles into a functional inflammasome complex, which in turn activates caspase‑1 and subsequently promotes the generation of the proinflammatory cytokines IL‑1β and IL‑18.313 This complex drives a signalling pathway that simultaneously activates gasdermin D-N (mediating pyroptosis), caspase-3/7 (mediating apoptosis), and MLKL (mediating necroptosis).314 However, neither NLRP3 nor caspase‑1 is essential for influenza A virus (IAV)-induced PANoptosis because ZBP1-dependent activation of caspase-8/-3/-7, RIPK3 and MLKL is also sufficient to trigger PANoptosis.315Importantly, to prevent aberrant activation of this lethal pathway and maintain tissue homeostasis, ZBP1-mediated PANoptosis is tightly regulated by the RNA-editing enzyme ADAR1. A recent study highlighted the crucial regulatory role of ADAR1–ZBP1 interactions in suppressing spontaneous PANoptosis.316 Mechanistically, ADAR1 utilises its adenosine-to-inosine (A-to-I) base modification activity to prevent the accumulation of endogenous immunostimulatory double-stranded RNAs and Z-form RNAs (Z-RNAs). Loss of ADAR1 function leads to the accrual of endogenous Z-RNAs, particularly those enriched in the 3′ untranslated regions of interferon-stimulated mRNAs, which bind to the Zα domain of ZBP1 and trigger robust PANoptosis.317 Furthermore, ADAR1 can physically restrict ZBP1-mediated immune responses via direct Zα–Zα domain interactions that limit ZBP1 binding to RIPK3. This interaction serves as a critical checkpoint that, when pharmacologically targeted using interferons and nuclear export inhibitors, can induce PANoptosis to promote tumour regression.318Conversely, under sterile inflammatory conditions, disruption of this ADAR1–ZBP1 regulatory axis drives severe pathologies. A prominent example is dietary alcohol-induced liver injury. Recent evidence has demonstrated that ethanol and interferons synergistically activate JNK signalling, which consequently suppresses ADAR1 expression. This alcohol-induced repression of ADAR1 releases the brake on endogenous Z-RNA formation, triggering ZBP1-dependent PANoptosis and causing intense inflammatory cytokine release and liver damage. These studies critically expand the disease relevance of PANoptosis from infectious diseases to metabolism-driven, sterile inflammatory disorders.319Additionally, studies using TAK1-deficient models have demonstrated that either inactivation (or depletion) or hyperactivation (or overexpression) of TAK1 can induce cell death and inflammatory responses.147 Even when upstream survival signals are blocked, the PANoptosome is still capable of triggering robust inflammation and cell death, highlighting its function as a “backup” defence mechanism.310 Increasing evidence suggests that the pathological relevance of PANoptosis extends significantly to the pathogenesis of cytokine storm syndromes, highlighting its broad translational potential.320,321 Cytokine storm syndromes drive pathology in life-threatening conditions such as severe viral infections (e.g., SARS-CoV-2), CAR T-cell toxicity, HLH, and systemic inflammatory disorders.320 Central to this pathology is a self-amplifying positive feedback loop in which synergistic cytokine signalling drives disease progression. Specifically, the synergism of TNF and IFN-γ triggers inflammatory cell death by activating the JAK/STAT1/IRF1 axis, which induces nitric oxide (NO) production and subsequently drives caspase-8/FADD-mediated PANoptosis.321 This PANoptosis-mediated lysis releases abundant proinflammatory cytokines and damage-associated molecular patterns (DAMPs) into the microenvironment, inducing paracrine PANoptosis in neighbouring cells and perpetuating a lethal cytokine storm. This discovery has positioned the PANoptosome and synergistic cytokine signalling pathways as high-priority therapeutic targets: Blocking specific nodes within this complex (such as ZBP1 or Caspase-8) or neutralising TNF and IFN-γ can ameliorate tissue damage and prevent mortality in conditions ranging from COVID-19 and HLH to CAR T-cell toxicity, without completely compromising host defence.312Autophagy and apoptosisApoptosis, characterised by caspase activation and chromatin condensation, is a canonical mode of RCD.34 In contrast, autophagy primarily acts as a cytoprotective process that degrades damaged organelles to recycle nutrients and alleviate cellular stress.322 Despite their distinct features, mounting evidence supports intricate crosstalk between these two modalities323 (Fig. 11). The BCL-2-Beclin-1 complex, formed via the BH3 domain of BCL-2 and the BH3-like domain of Beclin-1, serves as a core regulatory hub in the crosstalk.275,324 Disrupting their interaction directly reshapes cell death progression: Novel small molecules derived from the SW076956 scaffold bind the BH3-binding groove of BCL-2 with nanomolar affinity, selectively disrupting the BCL-2-Beclin-1 interaction to specifically trigger autophagy without off-target apoptosis.325Fig. 11Full size imageMolecular crosstalk between autophagy and apoptosis signalling pathways. Molecular crosstalk between autophagy and apoptosis. The BCL-2/Beclin-1 complex acts as a central rheostat for cell fate. Autophagy is initiated upon the dissociation of this complex, driven by specific posttranslational modifications (e.g., JNK/DAPK phosphorylation and TRAF6 ubiquitination) and the competitive binding of BH3-only proteins. In parallel, intrinsic apoptosis occurs via BCL-2 inhibition and subsequent BAX/BAK activation, whereas extrinsic apoptosis is triggered by death receptor-mediated Caspase-8 activation. The binding of FADD to the ATG5/ATG12/ATG16 complex, together with the interaction between p62 and LC3, results in the recruitment of caspase‑8 to autophagosomes, thereby initiating and amplifying apoptotic signals. FAP‑1 is targeted for autophagic degradation in autophagosomes, resulting in the release of caspase‑8 and the initiation of apoptotic signalling. By FigdrawUnder nutrient deprivation or stress conditions, c-Jun N-terminal kinase 1 (JNK1) is activated and phosphorylates BCL-2 at Ser69, Thr70, and Ser87. This phosphorylation event weakens the interaction between BCL-2 and the BH3 domain of Beclin-1.326 In contrast, elevated glucose levels diminish AMPK-mediated JNK1 activation, which in turn strengthens the BCL-2–Beclin-1 interaction and inhibits autophagy.327 Beclin-1 phosphorylation modulates its association with BCL-2, thereby exerting either positive or negative regulatory control over autophagy. Death-associated protein kinase (DAPK) is a Ca²⁺/calmodulin (CaM)-regulated serine/threonine kinase that phosphorylates Beclin-1 at Thr119, consequently promoting autophagy.328 In addition, Beclin-1 is regulated by tumour necrosis factor receptor-associated factor 6 (TRAF6). Signalling via toll-like receptor 4 and IFN-γ induces TRAF6-mediated ubiquitination of Beclin-1 at Lys117 within the BH3 domain, further enhancing autophagy.329The adaptor protein FADD binds to death receptors through its death domain and recruits caspase-8. The recruitment of caspase-8 to autophagosomes is accomplished through multiple mechanisms. In the first mechanism, ATG5 interacts with the death domain of FADD via its central region, and the binding affinity between ATG5 and FADD is further influenced by ATG12 and ATG16. The association of FADD with the ATG5/ATG12/ATG16 complex allows caspase-8 to be recruited to the autophagosomal membrane, initiating and amplifying apoptotic signals.330 An alternative mechanism involves interaction between p62 and LC3. LC3 contributes to the fusion of autophagosomes with lysosomes and mediates the recruitment of caspase-8 to autophagosomes, with p62 acting as a core mediator to maintain the interaction among these three components.331 Another regulatory mode that governs the crosstalk between autophagy and apoptosis is known as autophagic degradation, in which FAP-1 serves as a regulatory molecule.332Ginsenoside Rg1 has been shown to disrupt the BCL-2–Beclin-1 complex, promoting autophagy (elevating the LC3-II/LC3-I ratio and reducing p62 levels) while suppressing apoptosis and exerting cytoprotective effects on starved H92C cardiomyocytes.333 In diabetic cardiomyopathy, a high-glucose environment inhibits AMPK activity and the JNK1–BCL-2 signalling pathway, strengthening the BCL-2–Beclin-1 interaction to suppress cardiac autophagy and promote cardiomyocyte apoptosis. Notably, metformin can reverse these pathological changes: It activates AMPK to induce BCL-2 phosphorylation and dissociation from Beclin-1, restoring cardiac autophagic flux, reducing apoptosis, and improving cardiac structure and function in diabetic mice.327Silica nanoparticles (SNPs) induce testicular toxicity by reducing the Leydig cell count, impairing cell viability, and increasing apoptosis in a dose-dependent manner while activating Beclin-1-mediated autophagy. Importantly, Beclin-1 mitigates SNP toxicity by suppressing the caspase-8 apoptotic pathway. Consistently, Beclin-1 knockdown or pharmacological inhibition of autophagy (e.g., 3-methyladenine and chloroquine) activates caspase-8-driven apoptosis, significantly increasing the apoptosis rate.334 In herpes simplex virus type 1 (HSV-1) infection, caspase-8 is indispensable for apoptosis in mouse embryonic fibroblasts (MEFs) and U937 cells, and it facilitates viral release by inhibiting autophagy, as reflected by downregulated Beclin-1 expression, reduced conversion of LC3-I to LC3-II, and impaired p62 degradation. Conversely, caspase-8 knockdown enhances autophagic activity and attenuates viral release.335 Crosstalk between autophagy and apoptosis acts as a programmable molecular switch that is regulated by proteins (Bcl-2–Beclin-1 and caspase-8–Beclin-1). These findings indicate that targeting a single pathway alone is insufficient, and future combination strategies tailored to specific disease contexts may provide novel therapeutic insights.Organelle-level dialogueMitochondria as integration hubsMitochondria, traditionally recognised as the “powerhouse of the cell,” have evolved into central integration hubs that converge and coordinate multiple RCD pathways. Through their dual roles in energy metabolism and signal transduction, mitochondria integrate upstream stress signals, mediate the crosstalk between distinct cell death modalities, and shape cellular fate in physiological and pathological contexts.Apoptosis is centrally orchestrated by mitochondria through the intrinsic pathway. The BCL-2 family (prosurvival: BCL-2, BCL-XL; proapoptotic multidomain: BAX, BAK, BOK; proapoptotic BH3-only: BID, BIM, PUMA) regulates mitochondrial outer MOMP, representing the irreversible commitment step in intrinsic apoptosis.34,37 Upon MOMP, mitochondrial intermembrane space (IMS) proteins, such as cytochrome c, are released into the cytosol, where they form apoptosomes with APAF1 and procaspase-9 to activate downstream effector caspases (caspase-3/-7) and trigger orderly cell dismantling.53 Notably, mitochondria blur the line between apoptotic and nonapoptotic death by linking MOMP to inflammatory signalling.336 Under caspase inhibition, the MOMP-induced release of mtDNA and double-stranded RNA (dsRNA) activates the cGAS–STING and MAVS-dependent type I interferon pathways,337 respectively, bridging apoptosis to inflammatory cell death. Additionally, nonlethal MOMP (minority or incomplete MOMP) represents a convergent signalling state that modulates cell survival, genome stability, and innate immunity, further expanding the integrative role of mitochondria beyond cell death.336Mitochondria serve as both signalling platforms and effectors for inflammatory RCD modalities (pyroptosis, necroptosis, and PANoptosis) by propagating proinflammatory signals. In pyroptosis, mitochondrial dysfunction (e.g., mtROS overproduction and mtDNA leakage) activates the canonical NLRP3 inflammasome, triggering caspase-1-mediated cleavage of GSDMD.85 The N-terminal domain of GSDMD forms pores in the plasma membrane and mitochondrial membranes, exacerbating mtROS release and mtDNA efflux to amplify inflammation.338 Conversely, pyroptotic signalling (e.g., caspase-3-mediated GSDME cleavage) can directly disrupt mitochondrial integrity,138 creating a positive feedback loop. Mitochondrial quality control (MQC) mechanisms such as mitophagy restrict pyroptosis by clearing damaged mitochondria,339 whereas excessive mitophagy may promote inflammasome activation, highlighting context-dependent crosstalk. In necroptosis, mtROS promotes RIPK1 autophosphorylation and necrosome assembly,340 whereas RIPK3 phosphorylates MLKL to induce mitochondrial membrane permeabilisation, further increasing ROS production.95 Mitochondrial dynamics also regulate necroptosis: Excessive dynamin-related protein 1 (DRP1)-mediated fission promotes necroptosis by fragmenting mitochondria, whereas long optic atrophy 1 (L-OPA1)-dependent fusion suppresses it by preserving mitochondrial integrity.341,342Additionally, mitochondrial metabolites (e.g., citrate) can recruit DRP1 to mitochondria, linking fission to necroptotic signalling in acute lung injury (ALI).343 In PANoptosis, mtROS accumulation and mtDNA oxidation trigger the assembly of PANoptosomes (e.g., ZBP1-, AIM2-, and RIPK1-PANoptosomes), which activate caspases (caspase-1/-8/-3) and MLKL.311,344 Mitophagy inhibits PANoptosis by reducing mtROS and mtDNA leakage, whereas impaired mitophagy exacerbates the vicious cycle of mitochondrial damage and inflammatory cell death.344The role of mitochondria in iron, copper, and redox metabolism positions them as key regulators of metabolism-linked RCD (ferroptosis, cuproptosis, disulfidptosis). Mitochondria contribute to ferroptosis via impaired mitophagy (e.g., FUNDC1 deficiency)-mediated iron overload, reduced GPX4 activity, and oxidative stress amplified by imbalanced mitochondrial dynamics (e.g., excessive fission).303,345 Mitochondrial biogenesis regulators, such as PPARα coactivator-1 (PGC-1α), can either promote ferroptosis (by enhancing lipid peroxidation)346 or suppress it (via antioxidant responses),347 depending on the cellular context. In cuproptosis, excessive intracellular copper disrupts mitochondrial OXPHOS by destabilising Fe‒S clusters, activating the FDX1-mediated lipoylation of DLAT, and inducing mitochondrial dysfunction.18 Mitophagy attempts to clear copper-laden mitochondria, but impaired mitophagy leads to cumulative damage and cuproptosis.348 In cancer, targeting mitochondrial copper metabolism (e.g., via SLC31A1 upregulation) induces cuproptosis, offering a therapeutic avenue for apoptosis-resistant tumours.349 In disulfidptosis, mitochondrial ROS production exacerbates NADPH depletion by disrupting the reduction of cystine to cysteine.231 Although direct crosstalk is less characterised, functional enrichment analysis links disulfidptosis to mitophagy and mitochondrial fusion pathways.350Bridging role of ER stressThe ER is a key organelle that regulates protein folding and cellular homeostasis and activates the unfolded protein response (UPR) to restore balance or induce cell death during ER stress.351 Dysfunction serves as a critical bridging pathway, playing a significant connective role within the cellular death regulatory network. It can not only trigger apoptotic cascades352 but also directly functionally interact with the process of ferroptosis by modulating lipid peroxidation and iron metabolism.353Ferroptosis inducers act as initiating factors of ER stress. Classic inducers (e.g., artesunate (ART), erastin, sorafenib and (1S,3R)-RSL3) inhibit System xc⁻, abrogating extracellular cystine uptake and causing severe intracellular cystine deficiency. This impairs GSH synthesis, disabling the primary antioxidant defence system against lipid peroxidation and leading to massive lipid peroxide accumulation.354 Concurrently, insufficient cystine uptake disrupts ER lumenal redox and calcium homeostasis, impairs secretory protein posttranslational modification, and triggers the UPR and ER stress.355 Among the three canonical upstream sensor pathways of ER stress, namely, the inositol-requiring protein 1 (IRE1), activating transcription factor 6 (ATF6) and protein kinase RNA-like ER kinase (PERK) pathways, the PERK pathway functions as the core regulatory axis in transducing ferroptotic signals to apoptotic signals, acting as a key hub for the crosstalk between ferroptosis and apoptosis.356Upon ER stress induction, PERK is activated and phosphorylates eukaryotic initiation factor 2α (eIF2α), which globally represses general protein translation while selectively upregulating the expression of the transcription factor activating transcription factor 4 (ATF4). As a master transcription factor in the ER stress response, ATF4 activates the expression of its downstream target genes, including C/EBP homologous protein (CHOP), tribbles pseudokinase 3 (TRIB3) and asparagine synthetase (ASNS). Among these targets, CHOP is the key molecule that mediates proapoptotic signal transduction: it binds to the promoter region of the proapoptotic molecule p53-upregulated modulator of apoptosis (PUMA) in a p53-independent manner, thereby inducing the specific expression of PUMA.357 This step represents the critical molecular link connecting ER stress to the crosstalk between ferroptosis and apoptosis. Notably, ferroptosis inducers alone induce PUMA expression via ER stress but fail to trigger apoptosis, as PUMA remains biochemically inactive because of the dynamic balance between PUMA and antiapoptotic BCL-2 family proteins (e.g., BCL-2 and BCL-XL). Thus, ferroptosis inducers sensitise tumour cells to apoptosis without inducing an overt apoptotic phenotype.358,359Combined treatment with the ferroptosis inducer and extrinsic apoptosis inducer TRAIL disrupts the pro- and antiapoptotic balance, triggering robust synergistic lethal effects, with ER stress-mediated PUMA activation as the central event.360 This complex activates caspase-8, which cleaves BID into tBID to initiate the upstream signals of the mitochondrial apoptotic pathway.361 This process involves c-Jun N-terminal kinase (JNK)/Akt-mediated modulation of antiapoptotic proteins and Beclin-1, which drives PUMA activation. Subsequently, activated PUMA synergises with tBID to promote Bax/Bak oligomerisation, leading to MOMP.362 MOMP releases cytochrome c into the cytosol, activating the caspase-9/caspase-3 cascade and inducing apoptosis.53 Furthermore, ferroptosis-induced lipid peroxidation exacerbates mitochondrial membrane damage, resulting in the formation of a positive feedback loop that amplifies the cell death response. Notably, Bax has been identified as a key molecular mediator of mitochondrial pathway activation in this type of crosstalk.363In summary, the ER acts as a core bridge in ferroptosis‒apoptosis crosstalk, with the ER stress-mediated PERK‒eIF2α‒ATF4‒CHOP‒PUMA signalling axis as the key link. In addition, the regulation of PUMA activity by other molecules (e.g., JNK and Beclin-1) reflects the molecular regulatory mechanism of this crosstalk.363 These findings not only improve the theoretical understanding of RCD but also provide a novel paradigm for tumour therapy. Specifically, the combined application of ferroptosis and apoptosis inducers can target and activate the ER stress-mediated crosstalk pathway, significantly increasing the efficacy of cancer cell elimination.Pharmacological intervention and translational progressPro-death therapies: eliminating pathological cellsCombination strategies in cancer therapyApoptosis in cancersUnder physiological conditions, the initiation and execution of apoptosis are tightly regulated, with the dynamic balance between proapoptotic and antiapoptotic signals dictating the fate of cell survival and death. Disruption of this balance and subsequent inhibition of apoptosis prevent the effective elimination of abnormally proliferating cells, ultimately leading to tumourigenesis.364 Notably, suppressing apoptosis serves as a critical strategy for cancer cells to escape oncogene-induced apoptosis.The evasion of apoptosis during tumourigenesis is closely associated with an imbalance in the ratio of proapoptotic to antiapoptotic proteins. The t(14;18) chromosomal translocation of the BCL-2 gene in lymphoma leads to transcriptional activation and subsequent overexpression of BCL-2.365 Ectopic overexpression of BCL-2 in B lymphocytes drives lymphomagenesis in murine models, and sustained inhibition of apoptosis further promotes the progression and malignant transformation of established lymphomas.365 Strikingly, BCL-2 overexpression abrogates MYC-induced apoptosis, further facilitating tumourigenesis and malignant progression.366 Additionally, BIM, a proapoptotic protein, acts as a critical tumour suppressor in MYC-induced murine B-cell leukaemia, with its deletion markedly accelerating tumour development in Eμ-Myc transgenic mice.367 Collectively, these findings demonstrate that the BCL-2 protein family is crucial in the initiation and progression of multiple cancer types.In addition to the BCL-2 family, IAPs are strongly involved in cancer pathogenesis. IAPs modulate the apoptotic process primarily by inhibiting caspase activity and regulating multiple cellular signalling pathways, and they additionally participate in cell proliferation, migration, inflammation, and immune regulation.368 Preclinical studies have shown that the IAP inhibitor LCL161 has robust antitumour efficacy in murine models of BRCA1-mutant ovarian cancer.369 In acute myeloid leukaemia (AML), overexpression of cIAP2 is correlated with poor overall survival in patients583. XIAP overexpression is also prevalent in advanced breast cancer, at expression levels significantly higher than those in normal breast tissue,370 and increased XIAP expression is associated with shortened overall survival in hepatocellular carcinoma patients.371 These clinical observations highlight the potential of IAP-targeted therapeutics for various cancer types.p53, a nuclear phosphoprotein encoded by the TP53 gene, is a master regulator of the cellular stress response. Its core biological functions include maintenance of genomic stability, modulation of cell fate decisions, and suppression of cancer cell proliferation. Dysregulation of p53 functions, mostly caused by TP53 gene mutations, is linked to the development of many human cancers.372 As the first identified tumour suppressor gene associated with apoptosis, aberrant p53 expression and function are frequently correlated with adverse clinical outcomes in patients with cancer.373 Several oncogenes, including MYC and E2F transcription factor 1 (E2F1), can induce the expression of alternative reading frames (ARFs). ARF then binds to murine double minute 2 (MDM2), abrogating the MDM2–p53 interaction and leading to p53 accumulation and subsequent activation of proapoptotic proteins.374 Upon activation by various extrinsic cellular stressors, p53 transactivates the expression of a panel of proapoptotic proteins, including PUMA, NOXA, BAK, APAF1 and TRAIL. However, p53 is functionally inactivated in most cancer types, precluding its normal proapoptotic function.375TP53 gene mutation and the negative regulation of p53 by MDM2 are the major causes of p53 inactivation. A landmark study revealed the first allele-specific small molecule that selectively binds to cysteine in the p53 Y220C mutant, restoring the thermal stability and transactivation activity of mutant p53.376Necroptosis in cancersNecroptosis plays a complex, context-dependent role in tumour initiation and progression, exerting either tumour-suppressive or oncogenic effects in different settings. The induction of necroptosis represents a promising strategy for overcoming apoptosis resistance and triggering immunogenic cell death (ICD). Necroptotic signalling releases danger signals and secondary messengers that modulate immune cells within the tumour microenvironment (TME), enhance antigen presentation, and couple cell death with antitumour immune activation, thereby increasing therapeutic efficacy.377RNA-dependent adenosine deaminase 1 (ADAR1) masks the immunogenicity of ZBP1-driven necroptosis. ADAR1 depletion leads to Z-RNA accumulation, ZBP1 activation, and subsequent RIPK3-mediated necroptosis, which significantly increases the efficacy of immune checkpoint blockade (ICB) in melanoma models.317 These observations indicate that the activation of necroptosis can revitalise antitumour immunity against immunologically “cold” tumours with low immune infiltration. In hepatocellular carcinoma (HCC), elevated expression of core necroptotic components (RIPK1, RIPK3, and phosphorylated MLKL) is positively correlated with increased CD8⁺ cytotoxic T-cell infiltration and favourable patient survival, supporting the immune-stimulatory role of necroptosis in tumours.378Furthermore, underappreciated crosstalk occurs between ferroptosis and necroptosis during cancer therapy: Ferroptosis-inducing metabolites generated from chemotherapy can severely suppress necroptosis. For instance, acrolein, a toxic byproduct of chemotherapeutics such as cyclophosphamide, activates ferroptosis but antagonises necroptosis by inhibiting MLKL oligomerisation, despite proper upstream phosphorylation. This metabolic interference promotes chemoresistance.101 Consequently, coadministering acrolein-trapping agents such as hydralazine removes this barrier, restoring MLKL oligomerisation and significantly enhancing the in vivo efficacy of chemotherapeutics.Conversely, necroptosis can facilitate tumour progression and metastasis under certain pathological conditions. Plasma membrane rupture during necroptosis results in the release of proinflammatory factors that may establish a tumour‑permissive TME. For instance, in breast cancer, MLKL-dependent necroptosis triggers ADAM-mediated ectodomain shedding of membrane proteins (e.g., E-cadherin), which directly promotes cell migration and metastasis.379 The soluble fragments generated from this shedding further suppress T‑cell‑mediated antitumour immunity. In gastric cancer (GC), single-cell and spatial transcriptomic analyses revealed that necroptosis acts as a dynamic driver of lymph node metastasis rather than merely a terminal cell death event. In metastatic lymph nodes, upregulated expression of necroptosis-related genes such as CHMP3 reshapes the TME into an immunosuppressive state through the MDK-NCL signalling pathway.380 A prognostic risk model based on necroptosis-related long noncoding RNAs also effectively stratifies GC patients, with high risk scores linked to aberrant immune infiltration and poor clinical outcomes.381 In pancreatic ductal adenocarcinoma, the necroptotic machinery contributes to an immunosuppressive TME, suggesting that targeting RIPK3 and other key regulators may yield therapeutic benefits.382Therefore, necroptosis-targeted cancer therapies require precise stratification: activating necroptosis may reinvigorate antitumour immunity in immunologically “cold” tumours, whereas inhibiting necroptosis may limit inflammation-driven metastasis and progression in certain malignancies.Pyroptosis in cancersPyroptosis plays a crucial antitumour role in cancer primarily through directly eliminating tumour cells and reshaping the antitumour immune microenvironment.108 On the one hand, the activation of inflammasomes (e.g., NLRP3 and AIM2) can directly trigger pyroptosis in tumour cells via gasdermin-mediated membrane pore formation, leading to rapid lysis of malignant cells. For instance, the AIM2 inflammasome increases the sensitivity of AML cells to differentiation therapy, whereas inhibition of inositol-requiring enzyme 1α (IRE1α) increases taxane-induced pyroptosis in triple-negative breast cancer (TNBC).383 On the other hand, tumour cells release proinflammatory cytokines (IL-1β and IL-18) and DAMPs (e.g., HMGB1 and ATP) during pyroptosis, which act as danger signals to recruit and activate dendritic cells (DCs).384 Activated DCs then present tumour-associated antigens to CD8⁺ T cells, initiating systemic cell-mediated antitumour immunity.385 This process establishes an “in situ vaccine” effect, converting immunologically “cold” tumours into inflamed “hot” tumours. It reshapes the TME by reducing the infiltration of immunosuppressive cells such as myeloid-derived suppressor cells (MDSCs) and regulatory T cells (Tregs) while polarising tumour-associated macrophages towards the proinflammatory M1 phenotype.386 Additionally, the N-terminal fragment of GSDMD (GSDMD-NT) induces mitochondrial damage in tumour cells.293 Pyroptosis significantly enhances antitumour immune responses, reverses the immunosuppressive state, increases the efficacy of immune checkpoint inhibitors, and hinders tumour cell migration and the formation of premetastatic niches.Conversely, under specific pathological conditions, pyroptosis promotes tumour progression by fostering a protumour inflammatory microenvironment and enabling tumour immune escape. Persistently activated inflammasomes can drive chronic inflammation, and the long-term release of cytokines (e.g., IL-1β and IL-6) promotes an immunosuppressive TME by recruiting MDSCs and Tregs while simultaneously promoting tumour cell proliferation, angiogenesis (e.g., via upregulation of VEGF expression), and epithelial‒mesenchymal transition (EMT).387,388 Moreover, pyroptosis can induce the polarisation of M2-type tumour-associated macrophages (TAMs) and upregulate the expression of immune checkpoint molecules (e.g., PD-L1), thereby inhibiting effector T-cell function. Some tumour cells precisely regulate the pyroptosis pathway through epigenetic modifications (e.g., METTL3-mediated m⁶A modification) or metabolic regulation (e.g., SIRT7-mediated regulation of NLRP3), thereby achieving survival, increasing stemness, and developing resistance to chemotherapy and radiotherapy.389,390 Furthermore, the pyroptosis of immune cells directly impairs antitumour defence. Chemotherapy-induced pyroptosis of MDSCs can further suppress cytotoxic T lymphocyte (CTL) function by stimulating IL-17 production.391 In some tumours, the abnormal expression of pyroptosis-related molecules (e.g., GSDMD and NLRP3) is closely associated with increased tumour malignancy and poor patient prognosis, highlighting the context-dependent nature of its role.392Metal-dependent cell death in cancerMetal-dependent RCD, orchestrated by cascade signalling pathways, has emerged as a pivotal regulatory mechanism in cancer progression and therapy.393 Essential and trace metal ions (e.g., Fe, Cu, Zn, Ca, and Mn) serve as core mediators, with their dyshomeostasis (excess or deficiency) disrupting redox balance, lipid metabolism, and mitochondrial function to trigger cancer cell death. Notably, metal-dependent regulated cell death connects metal biology with tumorigenesis, offering unique therapeutic benefits through the targeting of cancer-specific metal metabolism reprogramming. As a tumour-suppressive mechanism, it eliminates malignant cells, yet cancers frequently evolve adaptive strategies to evade such death. For instance, ferroptosis, driven by Fe²⁺-catalysed lipid peroxidation via the Fenton reaction, is suppressed in aggressive tumours through upregulation of the SLC7A11/GPX4 antioxidant axis172 or NRF2-mediated iron homeostasis regulation through HECT and RLD domain-containing E3 ubiquitin protein ligase 2 (HERC2, E3 ubiquitin ligase for NCOA4 and FBXL5) and VAMP8 (mediating autophagosome–lysosome fusion).170 In the hypoxic tumour microenvironment, HIF-1α-dependent lactate production confers ferroptosis resistance through a pH-dependent mechanism that acts in parallel with the classical SLC7A11 and FSP1 systems. HIF-1α also transcriptionally upregulates SLC1A1, a key glutamate transporter, to increase cystine uptake, thereby further promoting ferroptosis resistance.394 Conversely, Cu⁺/Cu²⁺ overload induces cuproptosis by disrupting lipoylated TCA cycle proteins and Fe–S cluster biosynthesis, yet elevated Cu levels in tumours also promote proliferation and angiogenesis via AKT/ERK signalling.395 Metal-dependent cell death further modulates tumour metastasis; melanoma cells exploit enriched amounts of lymphatic GSH and oleic acid to resist ferroptosis, facilitating lymphatic dissemination,396 whereas Zn²⁺-mediated lysozincrosis inhibits metastatic melanoma growth by triggering lysosomal Zn²⁺ release and mitochondrial dysfunction.229The TME is dynamically regulated by metal metabolism, resulting in reciprocal crosstalk between cancer cells and stromal components. Iron dyshomeostasis shapes antitumour immunity: Fe²⁺ overload in TAMs promotes M2 polarisation and immunosuppression, while CD8⁺ T-cell-derived IFN-γ synergises with arachidonic acid to increase tumour ferroptosis and cytotoxicity.397 Tumours might evade ferroptosis in the TME, and SLC13A3 serves as an itaconate transporter in tumour cells. Through SLC13A3-mediated uptake of itaconate derived from TAMs, tumour cells activate the NRF2–SLC7A11 pathway and consequently evade immune-mediated ferroptosis.398 Copper ions modulate the TME by inducing M1 macrophage polarisation, promoting vascular endothelial growth factor (VEGF) expression, and regulating T-cell receptor (TCR) signalling to promote antitumour immunity. Zinc and manganese ions also play important roles in immunomodulation.399 Zn²⁺ activates the NF-κB–MAPK pathway to enhance T-cell and NK cell function,400 whereas Mn²⁺ triggers cGAS‒STING pathway activation in dendritic cells, increasing type I interferon secretion and adaptive immunity.401 These interactions highlight metalloptosis as a key node linking metal ion metabolism to TME remodelling, providing opportunities for metalloimmunotherapy.Targeting metal-dependent cell death has emerged as a promising strategy for cancer therapy, with approaches including small-molecule inducers, metal nanoparticles (MNPs), and combination regimens. Ferroptosis inducers (e.g., erastin and RSL3) disrupt system Xc⁻ or GPX4 to deplete GSH and accumulate lipid peroxides,172 whereas copper ionophores (e.g., disulfiram and elesclomol) promote Cu⁺ overload and cuproptosis. MNPs, including iron oxide, copper-based, and calcium-based nanoparticles, offer unique advantages for targeted delivery; they increase metal ion accumulation in tumours, synergise with photothermal/photodynamic therapy, and reverse drug resistance by bypassing cancer cell antioxidant defences.393 For example, Cu-based MNPs induce cuproptosis via DLAT oligomerisation and Fe–S cluster depletion, whereas iron sulfide nanoparticles amplify ferroptosis by activating the Fenton reaction.402 Moreover, combining metal ion–regulated cell death inducers with chemotherapy, radiotherapy, or immune checkpoint inhibitors (e.g., anti-PD-L1) synergistically enhances antitumour efficacy by overcoming apoptosis resistance and remodelling the immunosuppressive TME.403 Notably, targeting metalloptosis regulators (e.g., SLC7A11, FSP1, and ATP7B) effectively reversed resistance to sorafenib, cisplatin, and other conventional drugs. Despite substantial progress, clinical translation faces challenges, including the precise control of intracellular metal levels to avoid off-target toxicity, tumour-specific delivery, and the elucidation of crosstalk between metalloptosis and other RCD pathways.Applications in infectious diseasesNecroptosis in infectious diseasesNecroptosis plays a dual role in the pathogenesis of infectious diseases. It not only serves as a significant host defence strategy to eradicate the intracellular habitats of pathogens but also contributes to the development of maladaptive immunopathology. For instance, infection with Salmonella Typhimurium triggers type I interferon signalling, which in turn promotes necroptosis in macrophages. The death of wild-type macrophages can be prevented by the specific inhibition of RIPK1 or the knockdown of RIPK3.404 During influenza A virus infection, ZBP1 acts as a sensor for viral RNA, recruiting RIPK3 to initiate necroptosis.405 Notably, recent studies on SARS-CoV-2 have demonstrated that hyperactive necroptosis contributes significantly to the “cytokine storm” and severe lung damage observed in COVID-19 patients.406Pyroptosis in infectious diseasesPyroptosis plays a dual regulatory role in the pathogenesis of infectious diseases caused by diverse microorganisms, such as bacteria, viruses, and fungi. During bacterial infections, Streptococcus pneumoniae utilises surface protein A (PspA) and pneumolysin (PLY) to activate the NLRP3 or AIM2 inflammasome, and this induced pyroptosis can either limit bacterial colonisation or aggravate inflammatory damage in the lungs.407 Gram-negative pathogens, such as Salmonella and Shigella, modulate pyroptosis via type III secretion system (T3SS) components. Salmonella flagellin and T3SS needle protein activate the NAIP/NLRC4 inflammasome,408 whereas Shigella MxiI induces macrophage pyroptosis through the NAIP2/NLRC4 axis. In viral infections, the envelope protein domain III and nonstructural protein 1 (NS1)409 of dengue virus, the nucleic acids of influenza A virus, and the spike protein and Nsp5 of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) all engage the inflammasome-pyroptosis axis.410,411 This axis is involved in either viral replication or host immune clearance. Fungal components, including the hyphal form, ergosterol and candidalysin of Candida albicans, also induce pyroptosis via inflammasomes, contributing to fungal survival or increased host defence.412 Pathogens have evolved sophisticated strategies to evade pyroptosis-mediated immune surveillance, enabling immune escape and persistent infection. At the bacterial level, the Salmonella effector SopF blocks pyroptotic signalling by inhibiting the activation of the NLRP3 and NLRC4 inflammasomes, whereas Shigella OspC3 directly targets GSDMD to suppress its cleavage, facilitating bacterial dissemination.413,414Therefore, targeting key molecules in the pyroptosis signalling pathway has emerged as a promising therapeutic strategy for infectious diseases. Inhibitors of NLRP3 (e.g., tranilast, OLT1177), caspase-1 (e.g., VX765), and GSDMD (e.g., disulfiram, dimethyl fumarate) have demonstrated significant efficacy in different models of bacterial, viral, and fungal infections, reducing inflammatory tissue damage, lowering pathogen loads, and improving survival rates.415 The further in-depth exploration of pathogen-specific pyroptosis regulatory mechanisms, clarification of functional differences in pyroptosis across various cell types, and development of highly specific targeted modulators will provide more precise therapeutic avenues for combating refractory microbial infections.Anti-death therapies: protecting healthy cellsDevelopment of neuroprotective agentsApoptosis in neurodegenerative diseasesProgressive loss of specific neuronal groups represents a defining feature of neurodegenerative pathologies, with oxidative stress, neuroinflammation and RCDs closely linked to their pathogenesis. Apoptosis plays a critical role in the initiation and progression of neurodegenerative diseases, despite being indispensable for the normal development and maturation of the nervous system.416Alzheimer’s disease (AD), the most common neurodegenerative disorder in elderly people, causes progressive dementia and poses a severe global health burden. Its typical pathological features include neuritic plaques composed of amyloid-β (Aβ) protein, neurofibrillary tangles, and hyperphosphorylated Tau protein.417 The pathological progression of AD is further accompanied by mutations in the amyloid precursor protein (APP), presenilin-1 (PSEN1) and presenilin-2 (PSEN2) genes.417,418Excessive neuronal apoptosis exacerbates neurological dysfunction, directly contributing to the onset and progression of the disease. Modulating the balance between proapoptotic and antiapoptotic proteins is a key strategy to ameliorate aberrant apoptosis in AD. Aβ deposition disrupts the balance of BCL-2 family proteins, increasing proapoptotic protein levels and increasing the risk of neuronal death.419 For instance, in Aβ-induced SH-SY5Y cells, miR-16-5p directly targets the mRNA encoding BCL-2, and its overexpression downregulates BCL-2 expression and induces excessive cellular apoptosis.420 Similarly, an in vivo mouse study demonstrated that oligomerisation of Aβ results in increased Bim expression, decreased BCL-2 levels, triggered BAX activation, and induced neuronal cell death, effects that are effectively abrogated by BAX inhibition or knockout.421 Consistent with these findings, treatment of primary neurons with Aβ1-42 for 72 h rapidly and sustainably downregulated BCL-2 expression and upregulated BAX expression.422 Collectively, these findings confirm a strong association between BCL-2 family imbalance and AD pathogenesis.Parkinson’s disease (PD) is characterised by the progressive, selective loss of midbrain substantia nigra pars compacta (SNpc) dopaminergic (DA) neurons and the formation of Lewy bodies composed of abnormally aggregated α-synuclein.423 Overexpression of α-synuclein leads to its abnormal accumulation on the neuronal mitochondrial membrane, eliciting oxidative stress and the release of cytochrome c into the cytoplasm, thereby initiating mitochondria-mediated apoptosis.424Huntington’s disease (HD) is an autosomal dominant neurodegenerative disorder caused by abnormal expansion of CAG trinucleotide repeats in the huntingtin (HTT) gene on chromosome 4. This expansion produces mutant HTT (mHTT) protein, accompanied by progressive impairments in motor function and cognitive impairments.425 Further investigations have indicated that mHTT disrupts the balance between pro- and antiapoptotic proteins and markedly upregulates the expression of the proapoptotic proteins BIM and BAX, thereby inducing neuronal apoptosis. However, deletion of BIM leads to a decrease in both mHTT protein levels and neuronal cell death in experimental models.426 In addition, mitochondrial fragmentation and cristae structural alterations have been identified in in vitro HD cell models, which increase the apoptotic susceptibility of neurons in these models.427Necroptosis in neurodegenerative diseasesWhile apoptosis was historically viewed as the primary mechanism of neuronal cell death, accumulating evidence suggests that necroptosis plays a pivotal detrimental role in the pathophysiology of major neurodegenerative conditions, including AD, ALS, and PD.428In postmortem brain tissues from AD patients, the expression levels of RIPK1 and MLKL are significantly elevated, which correlates with reduced neuronal survival and decreased brain weight.429 Necroptosis is closely linked to microglial dysfunction. RIPK1 expression is upregulated in AD brains, and its activation mediates a disease-associated microglial (DAM) phenotype.430 This study revealed that inhibiting RIPK1 kinase activity reduced the amyloid burden and restored memory function in mouse models, suggesting that necroptosis signalling drives the inflammatory response to Aβ. In addition, Aβ oligomers activate microglia to release TNF, which induces neuronal necroptosis via the RIPK1-MLKL pathway. MLKL knockout or the inhibition of RIPK3 (via GSK’872) reverses Aβ oligomer-induced cognitive impairment.431The coexpression of p-Tau and p-MLKL in AD patient brains suggests that Tau aggregation may trigger necroptosis.432 In p-Tau-overexpressing cells and TauP301S mice, p-Tau induces neuronal death and microglial activation via the RIPK1–RIPK3–MLKL and NF-κB pathways, whereas Nec-1 s (a RIPK1 inhibitor) suppresses this process and improves cognitive function.433ALS is characterised by the deterioration of both upper and lower motor neurons, leading to muscle weakness and eventual paralysis.434 Necroptosis is equally critical in patients with ALS, particularly genetically linked cases. Optineurin (OPTN), a gene frequently mutated in ALS, normally suppresses necroptosis by promoting the proteasomal degradation of RIPK1. Loss-of-function mutations in OPTN lead to the accumulation of RIPK1 and subsequent necroptosis-dependent demyelination and axonal degeneration, identifying a direct genetic link between the necroptotic machinery and ALS pathogenesis.435 TANK-binding kinase 1 (TBK1), a serine/threonine kinase, is another core regulator of necroptosis.436 Whole-exome sequencing of ALS patients revealed a significant enrichment of heterozygous loss-of-function mutations in the gene encoding TBK1, and subsequent studies confirmed the pathogenic effects of TBK1 mutations in ALS.437 Notably, necroptosis is activated in TBK1-mutant zebrafish, and treatment with necroptosis inhibitors increases their survival rate.438In PD models, necroptotic machinery actively mediates axonal degeneration in dopaminergic neurons,439 suggesting that targeting necroptosis could preserve neuronal connectivity early in the disease course, offering a broader therapeutic window than previously anticipated. Integrated bioinformatics and experimental validation revealed differentially expressed necroptosis-related genes in PD.440 However, the pharmacological inhibition of this pathway requires careful optimisation: While necroptosis is activated in rotenone-induced PD models, the use of inhibitors such as Nec-1 must balance efficacy against potential off-target toxicity, highlighting the complexity of translating mechanistic insights into clinical treatments.441Pyroptosis in neurodegenerative diseasesPyroptosis has emerged as a critical mediator in the pathogenesis of neurodegenerative diseases.442 Central to this process is the activation of inflammasome complexes (e.g., NLRP1, NLRP3, AIM2) in central nervous system (CNS) resident cells, including microglia, astrocytes, and neurons. This activation leads to the subsequent cleavage of GSDMD and GSDME, resulting in membrane pore formation, cell lysis, and the release of proinflammatory cytokines (IL-1β and IL-18) and DAMPs. In AD, Aβ plaques and hyperphosphorylated tau trigger NLRP3 inflammasome activation in microglia and astrocytes, promoting GSDMD-mediated pyroptosis that exacerbates neuroinflammation and neuronal loss.443 In PD, α-synuclein aggregates activate microglial NLRP3 inflammasomes. This activation drives GSDMD-dependent pyroptosis and dopaminergic neuron degeneration in the substantia nigra, whereas the genetic ablation of Gsdmd in mouse models mitigates motor deficits and neuroinflammation.444 In ALS and multiple sclerosis (MS), pyroptosis contributes to disease progression through distinct cell-type-specific mechanisms. In ALS, aggregated TDP-43 and mutant SOD1 induce neuronal GSDME cleavage, leading to mitochondrial damage and axon retraction, whereas microglial GSDMD activation amplifies neuroinflammation.445 In MS, GSDMD and GSDME are upregulated in demyelinating lesions, with GSDMD mediating oligodendrocyte death and GSDME impairing microglial myelin debris clearance.446 Notably, pyroptosis activation is not limited to these classic neurodegenerative disorders; recent evidence links it to HIV-associated neurocognitive disorders (HAND), where HIV-1 proteins such as gp120 induce NLRP3/GSDMD-dependent pyroptosis in microglia, contributing to neuronal injury and cognitive impairment even under combination antiretroviral therapy.447The interplay between pyroptosis and various CNS cell types, as well as the blood–brain barrier (BBB), further amplifies its pathogenic effect on neurodegeneration. Astrocytic pyroptosis, mediated by NLRP3/caspase-1/GSDMD signalling, leads to cell swelling, membrane pore formation, and the release of neurotoxins. This not only disrupts BBB integrity and exacerbates neuroinflammation in conditions such as depression and ischaemic stroke448 but also worsens BBB dysfunction when combined with pyroptosis in brain microvascular endothelial cells. Proinflammatory cytokines and ROS released by glial cells further promote endothelial cell lysis and increased BBB permeability, facilitating peripheral immune cell infiltration and neurotoxin leakage into the CNS.449 Neurons are also susceptible to pyroptosis. For example, mitochondrial toxins and aggregated proteins (e.g., TDP-43 in ALS) induce caspase-3-dependent GSDME cleavage, triggering mitochondrial damage, neurite retraction, and cell death.442,445 Furthermore, clinical evidence of elevated inflammasome components and cleaved GSDMs in the peripheral blood and CSF of patients highlights the potential of pyroptosis mediators as diagnostic biomarkers.443 These findings position pyroptosis as a converging therapeutic hub for developing disease-modifying treatments across neurodegenerative disorders.Metal homeostasis in neurodegenerative diseasesIron, manganese, copper, zinc, and other metal ions are essential micronutrients for physiological processes, including energy metabolism and protein synthesis, with homeostasis tightly regulated by multiple proteins and molecular mechanisms.450 However, imbalances of metal ions (either excess or insufficiency) can induce neuronal death through oxidative stress, ferroptosis, cuproptosis, cell senescence, or neuroinflammation, which are closely involved in the pathological mechanisms of neurodegenerative diseases such as PD, AD, ALS, and HD.450Different neurodegenerative diseases exhibit characteristic metal ion distribution abnormalities. In PD, significant iron deposition occurs in regions such as the substantia nigra (SN) and red nucleus (RN), and the iron/ferritin ratio in the CSF can serve as a potential marker for disease progression.451 AD patients have elevated levels of iron, copper, and zinc in the cerebral cortex, hippocampus, and other regions, which colocalise with Aβ plaques and tau tangles.452,453 ALS patients present iron accumulation in the left precentral gyrus and thalamus, with abnormal levels of multiple metal ions in the CSF.454,455 In HD, iron deposition in the basal ganglia positively correlates with disease severity, accompanied by dysregulated manganese, copper and zinc distribution.456Targeting metal homeostasis is a promising therapeutic direction for treating neurodegenerative diseases. Metal ion chelators (e.g., desferrioxamine, clioquinol, quercetin, and deferiprone) exert neuroprotective effects by chelating excessive metal ions, reducing oxidative stress, and inhibiting pathological protein aggregation. Some chelators (e.g., deferiprone) have shown potential in clinical trials because they improve patients’ motor function and reduce brain iron levels.457 However, challenges such as heterogeneous efficacy and the need to balance systemic iron reserves remain. Future research should focus on optimising the targeting ability and BBB penetration of these agents to increase their clinical applicability.Applications in cardiovascular and autoimmune diseasesNecroptosis in cardiovascular diseasesNecroptosis is a critical driver of pathogenesis across a spectrum of cardiovascular diseases.458 In ischaemia‒reperfusion injury, elevated RIPK3 expression during oxidative stress results in the phosphorylation of calmodulin-dependent protein kinase II (CaMKII), which subsequently triggers the opening of the mPTP, leading to myocardial rupture and cell death.459 A recent study revealed that RIPK3 is actively secreted from cardiomyocytes during myocardial infarction and functions as a DAMP.460 This extracellular RIPK3 binds to the receptor for advanced glycation end products (RAGE) to activate CaMKII, thereby exaggerating inflammatory signalling and cell death, while its antibody-mediated neutralisation significantly ameliorates cardiac injury.460In addition to myocardial pathologies, necroptosis plays a pivotal role in vascular pathologies. In advanced atherosclerosis, macrophage necroptosis within lesions directly contributes to the formation of unstable necrotic cores.461 Specifically, atherogenic ligands such as oxidised low-density lipoprotein (oxLDL) trigger the RIPK3–MLKL pathway, and inhibition of this pathway significantly stabilises plaques by reducing systemic inflammation.461 Similarly, in the pathogenesis of aortic diseases, RIPK3-mediated necroptosis in vascular smooth muscle cells is a prerequisite for the development of aortic dissections and aneurysms. Its deletion effectively prevents smooth muscle cell necrosis and subsequent aortic dilation.462Pyroptosis in cardiovascular diseasesPyroptosis is a key driver of tissue damage, inflammatory amplification, and pathological remodelling across the spectrum of cardiovascular diseases (CVDs). Defined by membrane pore formation, rapid cell lysis, and the release of proinflammatory cytokines (IL-1β and IL-18) and DAMPs, pyroptosis is executed primarily through the cleavage of GSDMD and GSDME by activated caspases, creating a self-sustaining inflammatory cascade within the cardiovascular microenvironment.463 In the heart and vasculature, pyroptosis occurs in a cell-type-specific manner, affecting cardiomyocytes, vascular endothelial cells, smooth muscle cells, and infiltrating macrophages.463,464 The upstream activation of this pathway is triggered by a diverse array of stressors prevalent in CVDs, including oxidative stress, metabolic dysfunction, cholesterol accumulation, ischaemic injury, and neurohormonal activation. These stressors converge to upregulate inflammasome components and prime cells for pyroptosis.465 This priming step, which is regulated by NF-κB signalling, upregulates the expression of pro-caspases and proinflammatory cytokine precursors.Pyroptosis contributes to the pathogenesis of diverse CVDs, including atherosclerosis, acute myocardial infarction (AMI), heart failure (HF), pulmonary hypertension (PH), diabetic cardiomyopathy (DCM), myocarditis, pericarditis, and hypertension.108 In atherosclerosis, cholesterol crystals and oxLDL induce NLRP3 inflammasome activation in macrophages and vascular cells, promoting pyroptosis, foam cell formation, and inflammatory cell infiltration. Genetic deletion of GSDMD or NLRP3 reduces plaque size and necrotic core expansion in animal models. In HF, persistent inflammatory stress activates the NLRP3-caspase-1-GSDMD axis, leading to continuous cardiomyocyte pyroptosis and myocardial fibrosis, resulting in impaired contractile function.466 In DCM, hyperglycaemia-induced NLRP3 activation and GSDME-dependent pyroptosis contribute to myocardial fibrosis and dysfunction. Additionally, pyroptosis mediates cardiotoxicity from chemotherapeutic agents (e.g., doxorubicin), participates in myocarditis (via viral-induced canonical and noncanonical inflammasome activation), and perpetuates pericarditis through the release of IL-1α/β from dying cells.465 In hypertension, angiotensin II (Ang II) upregulates NLRP3 activity, driving vascular smooth muscle cell pyroptosis and vascular remodelling.467Targeting the pyroptotic pathway has emerged as a transformative therapeutic strategy for treating CVDs, with robust preclinical evidence and promising clinical translation. Small-molecule inhibitors, including MCC950 (a selective NLRP3 inhibitor)468 and disulfiram (a GSDMD pore blocker),469 have demonstrated efficacy in preclinical models of CVDs by reducing pyroptosis and inflammatory cytokine release. Despite these advances, critical challenges remain, including the need for tissue-specific targeting to prevent systemic immunosuppression and the translation of preclinical findings to diverse patient populations.Necroptosis in autoimmune diseasesSystemic lupus erythematosus (SLE) is a chronic inflammatory disorder involving multiple organ systems, with autoantigen accumulation as one of its typical hallmark features.470 Dysregulated necroptosis in lymphocytes and macrophages leads to an overload of necrotic debris, which serves as a source of autoantigens and fuels the production of antinuclear antibodies.470 The inhibition of RIPK1 has been shown to alleviate nephritis and reduce autoantibody levels in murine lupus models, underscoring the pathogenic role of this pathway in SLE progression.471 Necroptosis is also involved in inflammatory bowel disease (IBD), particularly ulcerative colitis. Excessive necroptosis of intestinal epithelial cells (IECs) compromises the integrity of the gut barrier. The P2Y14 receptor regulates this process; its deficiency or inhibition protects against IEC necroptosis via the cAMP/PKA/CREB/RIPK1 pathway, thereby ameliorating colitis.472Pyroptosis in autoimmune diseasesPyroptosis acts as a critical driver of autoimmune pathogenesis by disrupting immune homeostasis and amplifying tissue-damaging inflammation. Central to this process is the cleavage of GSDM family proteins (e.g., GSDMD and GSDME) into N-terminal fragments that form membrane pores. This triggers cell swelling, rupture, and the release of proinflammatory mediators, including IL-1β, IL-18, and HMGB1, that perpetuate immune cell recruitment and sustained inflammation, which are hallmarks of autoimmune disorders110,112Accumulating clinical and preclinical evidence underscores the context-dependent and organ-specific roles of pyroptosis across diverse autoimmune conditions. In the CNS, activation of the caspase-11/GSDMD pathway in astrocytes and oligodendrocytes contributes to neuroinflammation and demyelination in MS.473 In skin and hair disorders, pyroptosis plays distinct pathogenic roles. Caspase-11-driven GSDMD cleavage exacerbates imiquimod-induced psoriasis-like inflammation, whereas defective caspase-11-dependent lysosome acidification is implicated in SLE.474 In patients with SLE, elevated levels of cleaved GSDMD and IL-18 in peripheral blood mononuclear cells (PBMCs) correlate with disease activity. Additionally, pyroptosis-fuelled neutrophil extracellular trap (NET) formation amplifies autoantigen presentation.475 In IBD, dual activation of canonical NLRP3 and noncanonical caspase-11 inflammasomes in mesenchymal stromal cells and epithelial cells pyroptosis disrupts mucosal barrier integrity.476 This disruption enables commensal bacteria to exacerbate intestinal inflammation.The pathogenic relevance of pyroptosis extends beyond these key disorders, with consistent links to immune dysregulation across organ systems. In joint-specific autoimmune diseases such as RA, caspase-5 polymorphisms and neutrophil accumulation driven by pyroptosis-related mediators further highlight organ-specific contributions.477 These findings demonstrate that pyroptosis operates through conserved molecular mechanisms, including canonical (NLRP3-caspase-1-GSDMD) and noncanonical (caspase-4/5/11-GSDMD) inflammasome activation. These pathways adapt to tissue-specific contexts to drive disease progression. The release of proinflammatory cytokines and DAMPs during pyroptosis creates a feed-forward loop that sustains immune activation, breaks immune tolerance, and amplifies tissue injury, solidifying pyroptosis as a unifying pathogenic mechanism across the autoimmune disease spectrum.Applications in metabolic diseasesNecroptosis in metabolic diseasesNecroptosis has been identified as a driver of sterile inflammation in metabolic disorders.478 By releasing DAMPs, it triggers robust immune responses, thereby exacerbating tissue injury in obesity, fatty liver disease and atherosclerosis.478 In obesity, inflammation in adipose tissue is a precursor to systemic insulin resistance. RIPK1 gene variants are associated with obesity in humans; critically, therapeutic silencing of RIPK1 in a diet-induced obesity mouse model reduces fat mass and increases insulin sensitivity by modulating the immune microenvironment within adipose tissue.479 These findings highlight RIPK1 as a genetic driver and potential therapeutic target for metabolic dysfunction.Hepatic necroptosis is also a key feature of metabolic dysfunction-associated steatotic liver disease (MASLD), also known as nonalcoholic fatty liver disease (NAFLD).480 MASLD often progresses to nonalcoholic steatohepatitis (NASH), and studies have shown that RIPK3 expression is significantly upregulated in liver tissues of patients with NASH and in corresponding murine models.481 Specific inhibition of RIPK1 kinase activity attenuates liver injury, inflammation, and fibrosis in murine models of steatohepatitis.482In type I diabetes, T-cell–mediated β-cell death is a causal driver, with necroptosis playing an important role in this pathological process.483 Notably, a study demonstrated that Nec-1, a specific inhibitor of necroptosis, can prevent pancreatic islet cell loss under inflammatory stress, a protective effect that may enhance short-term β-cell survival.484 Furthermore, RIPK1-dependent regulation of necroptosis governs the development of islet inflammation and glucose intolerance in murine pancreatic islet cells.346 Conversely, conflicting findings suggest that necroptosis regulated by RIPK1 kinase activity is not associated with glucose tolerance in pancreatic islet cells.485 These contradictory observations indicate more intricate underlying mechanisms that warrant in-depth investigation, including the correlation between cellular model selection and disease model severity. In type II diabetes, chronic inflammation in adipose tissue acts as a key contributor to systemic insulin resistance. It has been reported that RIPK3 in adipose tissue plays critical roles in maintaining tissue homeostasis and suppressing inflammatory responses.486 Collectively, these findings illustrate that the roles and biological functions of RIPK1 and RIPK3 in diabetes are tightly modulated by pathological context dependency and tissue specificity.Pyroptosis in metabolic diseasesPyroptosis has emerged as a central pathogenic mechanism in diverse metabolic diseases, where its dysregulation amplifies chronic inflammation and disrupts metabolic homeostasis. This process is driven mainly by the activation of inflammasomes, particularly NLRP3 and AIM2,487 in response to multiple metabolic stressors, such as hyperglycaemia, dyslipidaemia (e.g., saturated fatty acids, oxLDL), mitochondrial dysfunction, and DAMPs such as HMGB1.488 In type II diabetes, high glucose and free fatty acids drive NLRP3 inflammasome activation in pancreatic β-cells, podocytes, and vascular endothelial cells, exacerbating insulin resistance and complications such as diabetic nephropathy, retinopathy, and cardiomyopathy.489 In obesity, adipocyte pyroptosis and the subsequent release of DAMPs promote adipose tissue inflammation and insulin resistance.490 In osteoporosis, osteoblast pyroptosis mediated by the NLRP3–CASP1–GSDMD axis disrupts bone homeostasis by inhibiting osteogenesis and promoting osteoclast activation.491Lipid species have distinct effects on cell death pathways. Saturated fatty acids (e.g., palmitic acid) induce pyroptosis in hepatocytes, cardiomyocytes, and pancreatic acinar cells via endoplasmic reticulum stress and mitochondrial dysfunction. In contrast, ω-3 polyunsaturated fatty acids (e.g., docosahexaenoic acid) inhibit NLRP3 inflammasome assembly through G protein-coupled receptor (GPR) 120 (GPR120)/β-arrestin2 signalling.492 Collectively, these findings underscore pyroptosis as a convergence point of metabolic stress, lipid dysregulation, and inflammation, with pathway-specific and cell-type-specific mechanisms all offering novel therapeutic avenues for mitigating metabolic disease progression and complications.Clinical and preclinical drug pipelineTargeting apoptosisBCL-2 inhibitorsVenetoclax (ABT-199; Table 1) is the first selective small-molecule BCL-2 inhibitor approved for clinical use by the FDA. Its mechanism of action, elucidated in early studies, involves blocking the antiapoptotic protein BCL-2 to release proapoptotic proteins (e.g., BIM), thereby inducing MOMP and subsequent apoptosis for cancer cell clearance.493Table 1 Representative clinical-stage drugs targeting the cellular apoptotic pathwayFull size tableIn early clinical trials of venetoclax,494 56 patients with relapsed/refractory chronic lymphocytic leukaemia (R/R-CLL) or small lymphocytic lymphoma (SLL) were treated with 8 doses ranging from 150 to 1200 mg daily. In an expanded cohort, 60 patients received weekly stepwise dose escalation, with a maximum daily dose of 400 mg. Treatment outcomes demonstrated therapeutic activity of venetoclax across all dose levels: 92 out of 116 treated patients (79%) achieved clinical responses, with response rates of 71–79% observed particularly in patients with 17p-deleted CLL and unmutated IGHV. In terms of safety, tumour lysis syndrome (TLS) initially occurred in 3 patients, with one fatal case. However, no TLS was reported after dose adjustment. Other adverse events included mild diarrhoea (52%), upper respiratory tract infection (48%), and nausea (47%) (NCT01328626). In another clinical trial,7 elderly patients with AML exhibited poor prognosis and limited responses to hypomethylating agents. In this study, 431 AML patients aged ≥75 years were randomised into two groups: azacitidine plus venetoclax (286 cases) and azacitidine plus placebo (145 cases). All patients received standard-dose azacitidine, with the combination group additionally administered venetoclax at 400 mg daily in 28-day cycles. The results indicated that azacitidine combined with venetoclax significantly prolonged overall survival and improved response rates in elderly AML patients, albeit with a higher risk of febrile neutropenia (NCT02993523).Despite promising clinical efficacy, drug resistance inevitably develops in most patients, particularly those with AML and R/R-CLL, leading to disease relapse. In CLL patients with disease progression during venetoclax treatment, a BCL-2 G101V mutation was detected. This mutation reduces the binding affinity of venetoclax by approximately 180-fold, thereby preventing the drug from displacing proapoptotic mediators from intracellular BCL-2 and conferring acquired resistance in primary patient cells.495 In 100 AML patients treated with venetoclax plus azacitidine, in vitro and in vivo experiments revealed that monoblastic AML (FAB-M5 subtype) was a resistance driver. The underlying mechanism involves the loss of BCL-2 expression in these AML cells, which switches to MCL1-dependent oxidative phosphorylation (OXPHOS) for survival, as venetoclax cannot inhibit MCL1.496 These findings suggest that combining venetoclax with a selective MCL1 inhibitor may improve AML treatment efficacy, as combination therapy may address the limitations of single-agent venetoclax resistance.Lisaftoclax, an oral, second-generation BCL-2 inhibitor approved by the NMPA in 2025, is primarily indicated for the treatment of haematologic malignancies, including CLL/SLL. For CLL/SLL treatment, lisaftoclax is administered via a 5-day dose-escalation regimen, reaching a maintenance dose of 600 mg on day 6. Multiple clinical trials of lisaftoclax are still ongoing. A phase II clinical study demonstrated improved efficacy of lisaftoclax (600 mg daily, 28-day cycles) in 72 patients with R/R-CLL/SLL refractory to Bruton’s tyrosine kinase (BTK) inhibitors (NCT05147467). In addition, phase III clinical trials for the use of lisaftoclax to treat chronic lymphocytic leukaemia and AML are currently ongoing (NCT06104566; NCT06389292).Sonrotoclax (BGB-11417) is a potent, highly selective second-generation BCL-2 inhibitor that inhibits tumour growth more effectively than venetoclax. Notably, sonrotoclax evades the resistance caused by the BCL-2 G101V mutation via a novel binding mode that targets the P2 pocket of BCL-2.497 Further mechanistic investigations have revealed that sonrotoclax in combination with radiotherapy induces immunogenic ferroptosis in cancer cells, and the most significant antitumour effect is achieved when sonrotoclax is combined with both radiotherapy and immunotherapy (anti-PD-L1).498 Ongoing clinical trials include a phase III head-to-head study (NCT06073821) comparing sonrotoclax (BGB-11417) plus zanubrutinib (BGB-3111) versus venetoclax plus obinutuzumab in 652 CLL patients.The first-generation BH3 mimetic ABT-737 represented a breakthrough in targeting BCL-2 family protein–protein interactions but lacked oral bioavailability, limiting its clinical application.499 On the basis of the structural scaffold of ABT-737, rational design yielded the second-generation BH3 mimetic navitoclax (ABT-263), which has EC50 values of 8 nM and 30 nM in BCL-2-dependent and BCL-XL-dependent cells, respectively, and significant oral bioavailability.499 Early phase I clinical studies have shown significant clinical improvements in CLL patients, but despite its antitumour activity, its therapeutic window is severely limited by dose-limiting thrombocytopenia. Although navitoclax has encountered obstacles in haematologic malignancy treatment, it holds broad prospects for myelofibrosis (MF) therapy. A pivotal phase II clinical study (NCT03222609) evaluated navitoclax in combination with ruxolitinib in MF patients who were refractory to or had an inadequate response to ruxolitinib monotherapy and demonstrated significantly prolonged overall survival and improved total symptom scores.BCL-XL inhibitorsBeyond BCL-2 inhibitors, BCL-XL has emerged as a critical therapeutic target in apoptosis. However, the clinical development of BCL-XL inhibitors has long been hindered by dose-limiting thrombocytopenia, as exemplified by navitoclax (ABT-263), a dual BCL-XL/BCL-2 inhibitor that induces severe on-target platelet toxicity. To overcome this limitation, proteolysis-targeting chimaera (PROTAC) technology has been employed to selectively degrade BCL-XL via E3 ligases that are minimally expressed in platelets. The first-in-class BCL-XL PROTAC, DT2216, conjugates a BCL-XL ligand with a von Hippel‒Lindau (VHL) E3 ligase ligand. By leveraging the low expression of VHL in platelets, DT2216 achieves tissue-specific degradation of BCL-XL, effectively preventing thrombocytopenia while maintaining potent antitumour activity.500 Notably, DT2216 has entered phase I clinical trials for relapsed/refractory solid malignancies (NCT04886622), with preliminary results indicating a favourable safety profile and manageable, transient thrombocytopenia. This PROTAC-based strategy represents a breakthrough in mitigating the on-target toxicity of BCL-XL inhibition and provides a promising therapeutic avenue for BCL-XL–dependent cancers.ABBV-155 is an ADC uniquely characterised by the conjugation of a BCL-XL inhibitory payload to a B7H3-directed antibody. Clinical evaluations have demonstrated that this approach dissociates on-target antitumour activity from dose-limiting haematological toxicity, achieving therapeutic benefits without inducing significant thrombocytopenia.501 Nanoparticle drug delivery technology, in which dendrimer-based nanocarriers are used to encapsulate the dual BCL-2/BCL-XL inhibitor AZD4320, significantly enhances the therapeutic window via precise modulation of pharmacokinetic properties. This approach improves tumour-specific accumulation and optimises drug release kinetics, thereby effectively reducing platelet toxicity. Notably, the resulting candidate, AZD0466, has demonstrated significant therapeutic efficacy in preclinical models of both acute myeloid leukaemia and T-cell acute lymphoblastic leukaemia.502MDM2 inhibitorsAlrizomadlin (APG-115), a novel second-generation MDM2 inhibitor developed by Ascentage Pharma, is designed to overcome the limitations of first-generation agents.503 APG-115 binds highly specifically to the p53-binding pocket of the MDM2 protein. Structural biology studies have shown that APG-115 has extremely high binding affinity for MDM2 (Ki