The review applies an author-developed framework with four axes: attribution of the DNA source, causality at the signaling node, specificity of the cellular lineage and relevance of the disease endpoint. Each axis is scored from zero to two. A complete preclinical causal chain requires the highest score in all four categories within the same model and specified lineage. The authors emphasize that this is a stringent operational benchmark, not a validated risk-of-bias instrument, quantitative ranking of targets or formal systematic evidence-grading system.

At the molecular level, productive cGAS–STING signaling requires more than mitochondrial stress, oxidative DNA damage or an abundance of cellular DNA. The relevant DNA must become accessible in the cytosol, engage cGAS, generate 2′3′-cyclic GMP-AMP and activate the endoplasmic-reticulum-resident adaptor STING. Downstream signaling includes TBK1, IRF3, type I interferon and NF-κB. By contrast, AIM2 recognizes cytosolic double-stranded DNA and recruits ASC to activate caspase-1, mature IL-1β and IL-18 and potentially trigger gasdermin-D-dependent pyroptosis. The pathways may respond to the same DNA pool, but neither is an obligatory intermediate of the other.

The strongest endothelial evidence comes from oscillatory-shear-stress models, where endothelial cells exposed to disturbed flow showed cytosolic double-stranded DNA, predominantly mitochondrial DNA, together with cGAS-associated DNA and cGAMP. Endothelial STING deficiency or knockdown and pharmacological STING inhibition reduced senescence, permeability and lesion-associated outcomes. These results support STING-node dependence during early endothelial dysfunction, but they do not complete the causal chain: endothelial-specific Cgas loss with same-cell rescue was not tested, and the same mitochondrial-DNA event has not been established in atherosclerotic plaque endothelium.

Evidence for vascular smooth muscle cells points to a DNA-damage-associated STING stress program rather than a proven driver of plaque failure. DNA double-strand breaks, telomere damage, senescence and extracellular-trap release have each been associated with smooth-muscle-cell stress, and lineage evidence from mice with telomere dysfunction links cytosolic DNA and STING–TBK1–IRF3 signaling to dedifferentiation and greater plaque burden. In chronic kidney disease-associated atherosclerosis, mitochondrial damage, DNA–cGAS colocalization and activation of TBK1 and IRF3 were reported in fibrous-cap smooth muscle cells. Nevertheless, smooth-muscle-specific Cgas perturbation, lineage-matched rescue and direct testing of cap thickness, collagen organization and necrotic-core endpoints remain missing.

In macrophages and hematopoietic cells, whole-body perturbations, bone-marrow transfer and systemic pharmacology support STING-node-dependent inflammatory amplification in mouse atherosclerosis. These experiments nominate a hematopoietic contribution but do not identify a macrophage-autonomous cGAS route. Lipid stress, defective efferocytosis, mitochondrial remodeling and inflammatory cell death may increase self-DNA availability, yet the responsible DNA species, same-cell binding to cGAS and production of cGAMP have not been consistently demonstrated in plaque macrophages. A model linking gasdermin-D-mediated mitochondrial perforation, mitochondrial DNA leakage and STING signaling also complicates pathway ordering, because mitochondrial damage may precede rather than follow STING activation.

AIM2 has a separate evidentiary profile. In JAK2V617F clonal hematopoiesis, AIM2 is associated with caspase-1 and IL-1β inflammatory outputs, although the DNA ligand and its relationship to cGAS–STING remain unresolved. In advanced-plaque models, AIM2 perturbation was associated with thicker fibrous caps, greater collagen and fewer cell-death or inflammatory features. Diabetes models indicate contributions from both hematopoietic AIM2 and NLRP3 to lesion growth, but do not establish AIM2–gasdermin-D as the unique mechanism underlying necrotic-core expansion. These findings do not identify AIM2 as a general downstream pathway or establish it as a universal target for chronic plaque stabilization.

Acute post-stroke evidence creates a separate setting. Neutrophil extracellular trap formation is the best-documented source of the circulating cell-free DNA signal that amplifies AIM2 signaling in plaques, with links to inflammation, thrombosis and recurrent vascular events. The route by which this DNA enters the cytosol of plaque sensor cells is unresolved. In this model, neutrophils appear to supply the ligand rather than serve as the principal cGAS or AIM2 effector within the lesion. The findings therefore support event amplification after stroke, not an explanation of chronic baseline plaque vulnerability. They also illustrate why pathway inhibitors must be judged against the specific event and endpoint they are intended to affect.

Human evidence remains mainly spatial and tissue-level. Carotid endarterectomy studies have detected STING in plaques, enriched staining in endothelial and immune-cell-rich regions and measured 2′3′-cGAMP in five plaques but not in four macroscopically normal arteries. These observations establish coexistence in human lesions, not the cell producing cGAMP or a complete intracellular DNA-to-sensor mechanism. The review argues that a decisive human study would need to distinguish mitochondrial from nuclear DNA, measure DNA–cGAS proximity or binding, quantify cGAMP in microdissected plaque regions and assess STING, TBK1 and IRF3 activation in the same annotated cells.

Pharmacological tractability is treated as a separate gate from biological evidence. STING inhibitors have reduced atherosclerotic burden in murine models, and human-cGAS inhibitors have suppressed DNA-triggered cytokine output in cultured human macrophages, but neither establishes efficacy in human plaques. Species selectivity is critical because some compounds activate murine but not human STING. Broader innate-immune functions also make infection, tissue repair, macrophage efferocytosis and endothelial reendothelialization essential safety endpoints rather than optional checks. The review’s practical conclusion is deliberately conservative: test STING blockade where only STING-node dependence is established, consider cGAS inhibition after same-cell DNA-to-cGAS-to-cGAMP evidence, and evaluate AIM2 or inflammasome strategies in settings where the relevant DNA source and acute or advanced-plaque phenotype are clearly matched.

Subject of Research: Evidence standards for cytosolic DNA-sensing pathways in atherosclerosis

Article Title: An evidence-to-claim framework for cGAS–STING and AIM2 in atherosclerosis

Article References: Lu, L., Lin, H., Zhang, P., Li, Y., & Zhang, J. (2026). An evidence-to-claim framework for cGAS–STING and AIM2 in atherosclerosis. Frontiers in Cell and Developmental Biology, 14, Article 1999070. https://doi.org/10.3389/fcell.2026.1999070

Image Credits: AI Generated

DOI: 10.3389/fcell.2026.1999070

Keywords: atherosclerosis, cGAS–STING, AIM2, cytosolic DNA sensing, mitochondrial DNA, endothelial dysfunction, vascular smooth muscle cells, macrophage inflammation, inflammasomes, plaque biology, evidence-to-claim, framework