image: A selection of major milestones in the study of extracellular ATP-mediated cell death and AICD is presented. The timeline summarizes the independent discovery and early biochemical characterization of ATP, the subsequent proposal of ATP-mediated energy transfer and the purinergic neurotransmission hypothesis, the evolving understanding of the roles of extracellular ATP in membrane permeabilization and cell death, the identification of P2X7 receptor- and inflammasome-associated mechanisms, the expansion of AICD mechanisms and their cancer-related translational relevance, and the conceptualization of AICD, its extension across disease contexts, and the proposal of the AICD–AICS continuum framework. The figure is designed to capture the historical trajectory and conceptual development of AICD-related research. The cited studies represent selected milestones rather than an exhaustive review of research in this area.
Credit: Doblin Sandai, Wangzheqi Zhang, Xiaokun Li and Hao Pan / Medicine Discovery
A review published in Medicine Discovery applies a four-level evidence-grading framework to the ATP-induced cell death framework and finds that 17 of the 38 AICD-Core genes rest on direct experimental evidence, while 11 are supported indirectly, 6 only in a context-dependent or mechanistically supportive way, and 4 by evidence that is currently insufficient
ATP is best known as the energy currency of the cell. Released into the space around cells after injury, inflammation or cellular stress, however, it also acts as a damage-associated signal that shapes intercellular communication and immune regulation and, in some circumstances, determines whether a cell survives. When extracellular ATP accumulates abnormally, purinergic receptors are over-stimulated and intracellular ATP homeostasis is disturbed; the result can be irreversible damage and death rather than adaptation. Research over the past decade has gathered these processes under a single label — ATP-induced cell death, or AICD — and, increasingly, under a single curated list of genes.
That list is where the problem begins. Existing databases that annotate regulated cell death-related genes, pathways and disease associations, including XDeathDB, RCDdb and FerrDb V3, are organised largely around defined cell death modalities rather than around the upstream ATP abnormalities that may precede them. As ATP-linked gene sets grow, so does the risk that inclusion in a database is read as evidence that a gene actually drives ATP-induced cell death.
A review published in Medicine Discovery addresses this gap directly. Rather than adding another list of candidate genes, the review summarises what is known about how AICD works, scores the strength and directness of the evidence behind each core gene, and describes ATPexGen, the public database that accompanies the framework.
AICD is a framework, not a single death programme
The authors describe AICD as an operational conceptual framework rather than a mode of cell death in its own right. "Instead of being an independent mode of cell death running parallel to apoptosis, pyroptotic cell death, necrotic cell death or ferroptosis, AICD encompasses upstream ATP abnormalities as well as cellular stress associated with multiple cell death outcomes," they write. What unites the framework is the starting point — abnormal ATP — rather than the ending.
On current evidence, the review identifies two general routes into AICD. The first is excessive extracellular ATP acting through purinergic receptors: P2 receptors, and the P2X7 receptor (P2X7R) in particular, respond to high extracellular ATP by triggering transmembrane ion flux, changes in membrane permeability and downstream inflammatory or death-associated signalling. The second is disruption of intracellular ATP homeostasis, which is associated with perturbed energy balance, mitochondrial dysfunction, endoplasmic reticulum stress and redox imbalance. The two are not mutually exclusive and can act through overlapping stress and death-response pathways.
Downstream of these triggers, the review maps a network rather than a line. Sustained P2X7R stimulation produces calcium influx and potassium efflux and increases membrane permeability. The membrane proteins PANX1 and ANO6/TMEM16F influence ATP release and phospholipid scrambling, and PANX1-dependent ATP release can amplify local purinergic signalling. Store-operated calcium entry mediated by STIM1 and ORAI1 sustains intracellular calcium elevation, while ROCK1-dependent cytoskeletal remodelling contributes to the membrane blebbing of dying cells. Inflammasome activation — with NLRP3 as the best-characterised ATP-responsive sensor and NLRP1 contributing in specific cellular contexts — and downstream caspase signalling widen the range of outcomes further: caspase-1 is associated largely with inflammasome-associated inflammatory cell death, whereas caspase-3, caspase-7 and caspase-8 are often involved in apoptotic or overlapping death responses. Mitochondrial dysfunction, BAX-dependent signalling, cytochrome c release, reactive oxygen species and impaired glutathione- and thioredoxin-dependent antioxidant buffering provide a second point of convergence.
The decisive variable, the authors argue, is context. Which mechanism dominates depends on cell type, receptor expression, ATP concentration and duration of exposure, and on the state of downstream stress- and death-response pathways. "Instead of resulting in a single deterministic cell-death phenotype, these context-dependent mechanisms may lead to apoptotic, pyroptotic, necrotic, ferroptotic, or other forms of cell death," the review states.
Separating database classification from evidence strength
The central methodological contribution of the review is a four-level ordinal evidence score, applied here to the 38 genes in the AICD-Core set. A score of 3 indicates direct experimental evidence linking an ATP-related perturbation to a cell-death phenotype; 2 indicates indirect experimental support for an ATP-related cell-death mechanism; 1 indicates context-dependent or mechanistically supportive evidence; and 0 indicates that current evidence is insufficient to establish a defined gene-specific association with ATP-related cell death.
Across the 38 core genes, the results are uneven. Seventeen genes scored 3, drawn from ATP sensing and membrane remodelling, inflammasome activation, caspase signalling, oxidative stress and stress-responsive kinase pathways — among them P2RX7, PANX1, ANO6, NLRP3, NLRP1, CASP1, CASP3 and CASP9, the mitogen-activated protein kinases MAPK1, MAPK3, MAPK8, MAPK14 and MAP3K5, CYBB, P2RX4 and the P2Y receptors P2RY2 and P2RY11. Eleven scored 2, including BAX, CYCS, CASP7, CASP8, ROCK1, TNF and several P2X and P2Y receptors. Six scored 1 — ORAI1, STIM1, P2RX3, P2RY4, P2RY6 and P2RY14 — and four scored 0: MYL9, P2RY5, CCNE2 and CCND2.
The authors are explicit that these categories are not a ranking of biological importance. "Core membership should not be seen as implying that each gene has been shown to contribute directly to ATP-induced cell death," they write. The Core/Extended distinction reflects the database's manual curation framework, based on mechanistic relevance and the available gene-level information, and was not set by applying a prespecified quantitative evidence threshold. Making weaker associations visible, the review argues, is more informative than assuming equivalent confidence across all core genes: the four score-0 records identify entries that require further verification and that may be retained, reclassified or removed as new experimental evidence appears.
The review also draws a boundary between ATP itself and related nucleotides. Several P2Y receptors are activated preferentially by ADP, UTP, UDP or UDP-sugars rather than by ATP, making their involvement nucleotide-associated rather than ATP-specific. "These associations remain informative for identifying potentially relevant mechanistic interfaces, but they should not be construed as equivalent to direct AICD evidence," the authors write.
A searchable resource, with limits stated up front
The gene sets are collated in ATPexGen, part of the ATPexSenDB system (http://www.atpexsendb.com/), which also hosts a companion database for ATP-induced cellular senescence. ATPexGen v1.0, released in May 2026, contained 38 AICD-Core and 216 AICD-Extended genes as of 4 August 2026. Users can query the database by gene symbol or keyword, filter genes across six mechanistic categories — purinergic receptors, ionic imbalance, mitochondrial dysfunction, endoplasmic reticulum stress, inflammasome activation and membrane rupture — and retrieve entry-level annotations for gene symbol, AICD module, cell death function and ATP-related mechanism. The Core and Extended datasets can be downloaded in CSV, TXT and JSON formats, and currently filtered results exported as an Excel file, for use in functional enrichment analysis, network analysis and candidate gene screening.
The review sets out clear limits on how far the resource should be taken. The literature was synthesised narratively rather than systematically, and because the gene collection was built iteratively, no single reproducible search date or complete Boolean search strategy can be reconstructed. Molecular relationships are recorded as conservative narrative mechanism annotations rather than as a dedicated structured interaction field with assigned regulatory polarity, and should not be read as a complete PubMed-wide interaction map. The current release does not present all evidence-related information as uniformly structured entry-level fields, and gene-specific update dates were not captured in a consistently retrievable manner.
Most importantly, the authors warn against treating inclusion as proof of causality. "Neither database inclusion nor mechanistic annotation should be taken as conclusive evidence that any individual gene has a causal role in AICD across diseases, cell types or experimental models," they write.
What comes next
Future curation, the authors state, will be driven by the accumulation of sufficiently substantive evidence rather than by a fixed calendar interval. New candidate genes will be evaluated on their ATP-related mechanistic relevance and on the strength and directness of the available gene-level evidence. Candidates with stronger support for central AICD-related mechanisms will generally be considered for AICD-Core, while broader, indirect or context-dependent associations will typically remain in AICD-Extended; unresolved candidates will stay pending review rather than receiving a definitive classification. The evidence score is intended to serve as one structured evidence layer for re-evaluating Core/Extended classification in future releases, rather than as a basis for retrospectively redefining the v1.0 gene sets, and each release will document gene additions, removals, reclassifications and major annotation changes so that gene-level changes remain traceable.
The review also identifies what would most improve the field: experiments that directly connect a defined ATP perturbation with directed manipulation at the gene level and with standardised cell-death endpoints, and more consistent reporting of ATP concentration, exposure time, cell type, metabolic state and the methods used to assess cell death. Such evidence, the authors suggest, would help distinguish direct mediators of AICD from downstream modifiers and broader nucleotide-associated mechanisms.
"Separating database classification from evidence strength may therefore enhance interpretation of gene–AICD associations and suggest priorities for future validation," the authors conclude. The review reports no new experimental data and establishes no new causal relationships: the evidence scores represent a conservative assessment of the published literature, and are not a substitute for experimental validation.
Article information
ATP-Induced Cell Death: Mechanistic Framework, Evidence-Graded Gene Landscape, and the ATPexGen Resource.
Medicine Discovery, 2026.
Original article link: https://www.sciencedirect.com/science/article/pii/S3117928226000032
DOI: 10.1016/j.medd.2026.100003
About the journal
Medicine Discovery (ISSN: 3117-9282) is a peer-reviewed, international, open-access journal published by Zhujiang Hospital of Southern Medical University in collaboration with KeAi Publishing and distributed on the ScienceDirect platform. The journal is dedicated to publishing high-quality translational medical research that bridges mechanistic discovery and clinical implementation, and covers five sections spanning mechanisms and systems medicine, translational and clinical research, digital medicine and data-driven healthcare, therapeutic innovation and biomedical engineering, and methodology, standards, and implementation science.
Website: https://www.sciencedirect.com/journal/medicine-discovery
Submit: https://www.editorialmanager.com/medd/default2.aspx
Method of Research
Literature review
Subject of Research
Cells
Article Title
ATP-induced cell death: Mechanistic framework, evidence-graded gene landscape, and the ATPexGen resource