🧬 Aging Wiki

mitochondrial-rna-leakage

Properties1
aliasesmtRNA-MAVS axis, cytosolic mt-dsRNA, mitochondrial dsRNA leakage, mtRNA-driven SASP

9 min read

Mitochondrial RNA leakage

The mitochondrial RNA leakage axis is a damage-associated molecular pattern (DAMP) pathway in which mitochondrial-derived double-stranded RNA (mt-dsRNA) escapes from the mitochondrial matrix into the cytoplasm, is recognized by the cytosolic RNA sensors RIG-I and MDA5, activates the OMM-tethered adaptor mavs, and drives type-I interferon production and SASP amplification. Three independent labs have converged on this mechanism as a major driver of cellular senescence (cell-level) and inflammaging (organism-level), establishing the mtRNA arm as a mechanistic bridge from the mitochondrial-dysfunction hallmark to the cellular-senescence hallmark.

This axis is distinct from but complementary to the parallel mtDNA–cgas-sting arm. The two arms together establish mitochondria as a dual-modality DAMP source in aged cells — releasing both DNA (sensed via cgas→sting) and RNA (sensed via RIG-I/MDA5→mavs) in response to mitochondrial damage.

Mechanism

1. Origin of cytosolic mt-dsRNA

The mitochondrial genome is transcribed bidirectionally from both heavy and light strands, generating long complementary antisense transcripts that anneal into long mt-dsRNA species in the mitochondrial matrix. Under physiological conditions, the matrix-localized 3′-to-5′ exoribonuclease PNPase (encoded by PNPT1) degrades these dsRNA species before they can escape; bi-allelic loss-of-function in PNPT1 unleashes mt-dsRNA cytoplasmic accumulation and drives a type-I interferonopathy with ~90-fold IFN-β mRNA induction 1.

2. Cytosolic leakage mechanisms

Two pathways permit mt-dsRNA exit from mitochondria:

  • miMOMP (minority mitochondrial outer-membrane permeabilization) — sub-apoptotic bax/bak1 pore formation during senescence; the OMM becomes selectively permeable to mtRNA without full-scale cell death. In senescent fibroblasts and FFC-diet MASH mouse hepatocytes, BAX/BAK double-knockout abolishes cytosolic mt-dsRNA accumulation, RIG-I/MDA5 expression upregulation, MAVS aggregate formation (~5 aggregates/cell senescent vs ~0 in proliferating; p=0.0001), and SASP cytokine secretion 2.
  • SEC61A1-dependent leakage — recently described in aging brain; the SEC61 translocon contributes to mtRNA escape independently of BAX/BAK. Sec61a1 knockdown alleviates mt-dsRNA cytoplasmic accumulation and rescues cognitive aging in wild-type mice 3.

3. Cytosolic RNA sensing

Cytosolic mt-dsRNA is recognized by:

  • RIG-I (DDX58) — binds short dsRNA with 5′-triphosphate ends
  • MDA5 (IFIH1) — binds longer dsRNA (>1 kb), particularly relevant for the long bidirectional mtRNA transcripts

In senescent cells, both sensors are transcriptionally upregulated, and RNA-immunoprecipitation experiments confirm direct binding of RIG-I and MDA5 to mtRNA species 2.

4. MAVS signalosome assembly and downstream signaling

RIG-I/MDA5 CARD-CARD interaction with mavs on the outer mitochondrial membrane triggers MAVS prion-like polymerization into amyloid-like fibrils (~1 ng activated MAVS converts ~16 ng endogenous MAVS into aggregates in 30 min) 4. The activated signalosome recruits TBK1/IKKε, which phosphorylate irf3/irf7 → IFN-α/β transcription. A parallel arm drives NF-κB → pro-inflammatory cytokine transcription (IL-6, IL-8, CXCL chemokines) that constitutes the SASP secretome 5. Full signaling architecture on the rig-i-mavs-pathway page.

Independent confirmations — 3-lab convergence (2024–2026)

PaperLabModel systemKey result
Victorelli et al. 2025 Nat CommunPassos (Newcastle/Mayo)IR- and replicatively-senescent MRC5 + IMR90 fibroblasts; FFC-diet MASH mouse with hepatocyte-specific Bax/Bak dKOBAX/BAK dKO abolishes cytosolic mtRNA, RIG-I/MDA5 upregulation, MAVS aggregates, and SASP. STING inhibition > MAVS knockdown for SASP reduction in this model
López-Polo et al. 2024 Nat CommunSerrano (CNIO/IRB Barcelona)Senescent human fibroblastsmt-dsRNA accumulation drives senescent-cell SASP via cytosolic RNA sensing; independent replication of the core mechanism 6
Zhang et al. 2026 Cell ResearchBeijing (multi-institutional)Naturally aging C57BL/6 mouse brain; 5xFAD AD mouse model; human AD brain tissueSEC61A1-mediated mtRNA leakage → MAVS → neuroinflammation → cognitive decline (without motor deficit); Sec61a1 or Mavs knockdown rescues cognitive aging in WT mice 3

The convergence across 3 labs, 2 disease contexts (MASH + AD/cognitive aging), and 3 organ systems (liver, brain, cellular fibroblasts) establishes the mt-dsRNA-MAVS-SASP axis as a robust, reproducible aging mechanism.

Relationship to the parallel cGAS-STING (mtDNA) arm

The mtRNA-MAVS axis is mechanistically parallel to the better-characterized mtDNA–cGAS–STING axis:

FeaturemtRNA arm (this page)mtDNA arm (cgas-sting)
DAMP speciesmt-dsRNAcytosolic mtDNA fragments
SensorsRIG-I (DDX58), MDA5 (IFIH1)cgas
Adaptormavs (OMM-tethered)sting (ER-tethered)
Downstream kinaseTBK1, IKKεTBK1 (predominantly)
Effector TFsIRF3, IRF7, NF-κBIRF3, NF-κB
OutputIFN-α/β, SASP cytokinesIFN-β, SASP cytokines
Leakage mechanismBAX/BAK miMOMP; SEC61A1BAX/BAK miMOMP; mitochondrial permeability transition

Victorelli 2025 directly compared the two arms in their MASH model and found STING inhibition reduces SASP more than MAVS knockdown — the cGAS-STING (DNA) arm appears dominant in their system. However, the two arms are not redundant: each responds to a different DAMP species, and the relative dominance likely varies by tissue and damage modality. In Zhang 2026’s aging-brain context, the MAVS arm appears load-bearing for the cognitive phenotype, suggesting tissue-specific arm dominance.

Aging-context significance

This axis provides a mechanistic bridge for the long-observed correlation between mitochondrial dysfunction and cellular senescence (López-Otín hallmark co-occurrence). The causal chain:

  1. Mitochondrial damage (oxidative stress, mtDNA mutations, electron transport chain decline) → bax/bak1 miMOMP or SEC61A1-mediated permeability
  2. mtRNA + mtDNA leakage into cytoplasm
  3. RIG-I/MDA5 (RNA arm) + cgas (DNA arm) sensor engagement
  4. mavs + sting signalosomes drive IFN + NF-κB
  5. SASP amplification → paracrine senescence + tissue inflammation

The axis connects multiple hallmarks:

Therapeutic angles

Intervention classMechanismAging-context tractability
BAX/BAK inhibitorsBlock miMOMP-mediated mtRNA leakageBroadly toxic — BAX/BAK required for canonical apoptosis; selective subapoptotic blockade not yet achieved clinically. translation-blocked-safety
PNPT1/PNPase augmentationEnhance matrix mt-dsRNA degradationUntested; Dhir 2018 establishes PNPase loss as causal in monogenic disease (proof-of-principle) but no augmentation strategy exists. preclinical-only
RIG-I/MDA5 antagonistsBlock cytosolic RNA sensingTool compounds exist; no clinically advanced antagonist (most RIG-I pharmacology has pursued agonism for antiviral/oncology indications). preclinical-only
MAVS inhibitorsBlock adaptor signalosome assemblyNo clinical inhibitor; CARD-domain oligomerization is challenging to drug. Tier 4 directly; modulated indirectly via TBK1/IKKε.
TBK1/IKKε inhibitorsBlock downstream IFN productionamlexanox (FDA-approved 1997 + Phase 2 T2D primary HbA1c endpoint hit, Oral 2017) — the only clinically advanced node in this axis. Shared with the cGAS-STING arm.
JAK inhibitorsBlock downstream IFN-α/β signalingjak1 inhibitors (ruxolitinib, baricitinib) suppress SASP in aged mice (Xu 2015) — tier-1 aging-context pharmacology.
SenolyticsEliminate senescent cells generating the damage signalbcl-xl-targeting (navitoclax) addresses downstream rather than the mtRNA axis directly.

The mtRNA arm is currently less clinically tractable than the cGAS-STING arm (where STING antagonists are advancing for autoimmune indications), but the TBK1/IKKε bottleneck offers a shared pharmacological handle pointing back to amlexanox.

Knowledge gaps

  • Human in-vivo evidence is limited — most data are from cellular models or mouse tissue. Human cohort evidence for the mtRNA arm specifically (vs the cGAS-STING arm, which has multiple human cohort signals) is sparse. needs-human-replication
  • Tissue-context dominance — Victorelli 2025 (MASH model) shows cGAS-STING > MAVS for SASP; Zhang 2026 (aging brain) shows MAVS is load-bearing for cognitive phenotype. Which tissues default to which arm, and what determines the balance, is unresolved. dose-response-unclear
  • Quantification of mtRNA species — the exact mtRNA species (anti-light-strand transcripts, mature processed mt-mRNAs, ncRNAs) that escape and activate RIG-I/MDA5 are not fully characterized. Dhir 2018 implicates long bidirectional transcripts; whether shorter mtRNA species also contribute is unclear. no-mechanism
  • Intervention tractability — BAX/BAK inhibition is broadly toxic; PNPase augmentation is untested; direct RIG-I/MDA5/MAVS antagonists are not clinically advanced. The shared TBK1/IKKε bottleneck (amlexanox) is the most tractable handle. long-term-unknown
  • Integration with mitophagy — mitophagy clearance of damaged mitochondria should prevent the leakage substrate, but the quantitative coupling between mitophagy efficiency and mtRNA arm activity is not fully mapped. Jiménez-Loygorri 2024 establishes the link in aged retina via type-I IFN suppression. no-mechanism

Cross-references

Footnotes

Footnotes

  1. doi:10.1038/s41586-018-0363-0 · Dhir A et al. · Nature 2018 · in-vivo (human + mouse) · model: primary fibroblasts from 4 PNPT1-mutation patients + hepatocyte-specific PNPase KO mice · ~90-fold IFN-β mRNA induction upon PNPase depletion; PNPase-depleted cells show cytoplasmic dsRNA distribution; MDA5 (not RIG-I) is primary sensor; MAVS knockdown abrogates IFN-β induction; classified as type I interferonopathy · PDF locally available · verified 2026-05-13 (MAVS protein page verification pass) ↩

  2. doi:10.1038/s41467-025-66159-z · Victorelli S et al. · Nat Commun 2025 · PMID 41398033 · in-vitro + in-vivo · model: IR- and replicatively-senescent MRC5/IMR90 fibroblasts; FFC-diet MASH mouse with hepatocyte-specific Bax/Bak dKO · cytosolic mtRNA accumulation, RIG-I/MDA5 upregulation, MAVS aggregates (~5 vs ~0 per cell; p=0.0001), and SASP all abolished by BAX/BAK dKO; STING inhibition > MAVS knockdown for SASP reduction · gold OA; PDF locally available · verified 2026-05-13 (RIG-I/MAVS pathway verification pass) ↩ ↩2

  3. doi:10.1038/s41422-026-01224-w · Zhang [et al.] · Cell Research 2026 · PMID 41692872; PMC13092635 (OA release 2027-05-01) · in-vivo · model: naturally aging C57BL/6 mouse brain + 5xFAD AD mice + human AD brain tissue · SEC61A1-mediated mtRNA leakage → MAVS → neuroinflammation → cognitive decline (no motor deficit); Sec61a1 or Mavs knockdown rescues cognitive aging in WT mice · abstract verified 2026-05-13 via PubMed efetch (full PDF not yet OA) ↩ ↩2

  4. doi:10.1016/j.cell.2011.06.041 · Hou F et al. · Cell 2011 · in-vitro · MAVS prion-like polymerization on OMM; K63-Ub4 requirement; ~1 ng PK-MAVS converts ~16 ng endogenous MAVS in 30 min · PDF locally available · verified 2026-05-13 ↩

  5. doi:10.1016/j.cell.2005.08.012 · Seth RB, Sun L, Ea CK, Chen ZJ · Cell 2005 · in-vitro + in-vivo · MAVS discovery paper; CARD domain (10–77); TM domain (514–535) OMM-essential for signaling; NF-κB + IRF3 + IRF7 activation · PDF locally available · verified 2026-05-13 ↩

  6. doi:10.1038/s41467-024-51363-0 · López-Polo V et al. · Nat Commun 2024 · in-vitro · model: senescent human fibroblasts (Serrano lab) · independent replication: mt-dsRNA accumulation drives senescent-cell SASP via cytosolic RNA sensing; consistent with Victorelli 2025 mechanism ↩

  7. doi:10.1038/s41467-024-45044-1 · Jiménez-Loygorri JI et al. · Nat Commun 2024 · in-vivo · urolithin A 2.3 mg/kg/day i.p. × 8 weeks in aged mice; mitophagy preservation suppresses type-I IFN signaling in aged retina · PDF locally available · verified 2026-05-13 (type-I-IFN pathway verification pass) ↩

Graph View

Backlinks