Mitochondrial DNA release activates cGAS–STING signalling and sustains type I interferon responses in aged microglia
H. Okada¹, M. Reyes¹ ², S. Lindqvist³, J. Adeyemi¹, R. Castellanos-Ortiz² *
Introduction. Innate immune sensing of cytosolic DNA is a conserved defence against viral and bacterial infection. The enzyme cGAS binds double-stranded DNA in a sequence-independent manner and catalyses synthesis of 2′3′-cGAMP from ATP and GTP. cGAMP engages STING, a four-pass transmembrane protein resident in the ER, which in its apo state forms a domain-swapped dimer with an open ligand-binding pocket. Ligand binding closes a β-sheet lid over the pocket and rotates the ligand-binding domain by approximately 180° relative to the transmembrane domain, permitting side-by-side packing of adjacent dimers. Whether this structural transition is limiting under sterile, age-related stress has not been established.
Microglia from aged animals exhibit elevated mitochondrial ROS and reduced TFAM occupancy, both of which have been associated with nucleoid instability. We therefore asked whether mtDNA escape is sufficient to engage cGAS in these cells, and at which step of the pathway signalling is amplified. Using quantitative imaging of STING puncta, non-reducing gel electrophoresis and phospho-specific immunoblotting for TBK1 (S172) and IRF3 (S396), we resolved sensing, trafficking and transcriptional stages in primary cells isolated from 3- and 22-month-old C57BL/6J mice.
Results. Cytosolic mtDNA, quantified by qPCR of MT-ND1 in digitonin-extracted fractions, was elevated 3.8-fold in aged microglia (Fig. 1b). cGAS co-localised with extranuclear DNA foci in 61 ± 7% of aged cells. Strikingly, total STING abundance did not differ between groups (Fig. 2, lower panel), whereas high-molecular-weight STING species increased 4.6-fold and co-localised with GM130-positive Golgi membranes. TBK1 phosphorylation scaled with oligomer abundance (r = 0.82) rather than with STING expression (r = 0.11). Treatment with H-151, a covalent STING antagonist that blocks palmitoylation at Cys91, reduced oligomeric STING by 83% and TBK1 phosphorylation by 76%, while total STING and cGAMP concentrations were unchanged (Fig. 3a–c). IFNB1 transcripts fell to levels observed in young cells within 48 h.
Discussion. Our findings argue that the amount of STING in aged microglia is not the relevant variable. Instead, signalling output is set by how efficiently ligand-bound STING assembles into higher-order clusters after its conformational switch. This distinction matters for therapeutic design: strategies that lower STING expression may be less effective than those that prevent its oligomerisation or trafficking to the Golgi. Limitations include the use of primary cultures, which may not capture the full complexity of microglial states in situ, and the modest cohort size. Future work should test whether mtDNA release is itself reversible with interventions that restore mitochondrial membrane integrity, and whether cluster size predicts cognitive decline in longitudinal cohorts.
Methods. Primary microglia were isolated by CD11b magnetic separation from pooled cortices (n = 6 per group). Cytosolic fractions were prepared by digitonin permeabilisation (25 µg/mL, 10 min, 4 °C) and verified by absence of COX IV. STING oligomers were resolved on 4–12% gradient gels under non-reducing conditions…
