Hypothesis

Senescent cells act as tissue chaperones by releasing mitochondria‑derived extracellular vesicles (EVs) that transfer protective miRNAs and antioxidants to neighboring cells; however, when mitochondrial ROS production exceeds a tissue‑specific threshold, EV cargo shifts toward pro‑inflammatory SASP factors, converting the chaperone signal into a pathogenic driver.

Mechanistic Basis

Recent Mendelian randomization work shows that genetically predicted overload of vascular oxidative stress shortens telomeres and accelerates frailty[1]. This suggests that oxidative pressure can be treated as an exogenous instrument to probe senescent cell function. In acute injury, senescent hepatic stellate cells limit fibrosis by secreting MMP‑containing extracellular vesicles (EVs)[3]; these EVs also carry miR‑29 and SOD2 that reduce oxidative stress in nearby parenchyma. Chronic accumulation, however, drives mitochondrial electron‑leak, raising mitochondrial ROS and oxidizing EV lipids[4]. Oxidized EVs preferentially load IL‑6, CXCL‑8 and HMGB1, while losing miR‑29/SOD2. The switch is bistable: low ROS yields chaperone EVs, high ROS yields pathogenic EVs. Genetic variants in GPX4 or SOD2 can serve as instruments to shift the ROS set‑point, allowing MR‑style estimation of the causal effect of EV ROS load on outcomes such as collagen deposition (protective) versus serum CRP (pathogenic).

It's important that we test this switch in vivo because it explains why broad senolytics sometimes impair wound healing while chronic clearance reduces inflammation. We don't yet know the exact mitochondrial ROS threshold in each tissue.

Testable Predictions

1. In young mice subjected to transient hepatic injury, EV isolates from senescent stellate cells will show high miR‑29/SOD2 and low IL‑6; administering a mitochondrial ROS scavenger (MitoTEMPO) will preserve this profile and accelerate wound closure.
2. In aged mice with established fibrosis, EV isolates will display inverted cargo (low miR‑29/SOD2, high IL‑6); genetic reduction of GPX4 (increasing ROS) will worsen fibrosis, whereas SOD2 overexpression will revert EVs to a chaperone phenotype and reduce collagen deposition.
3. Mendelian randomization using human GPX4 and SOD2 variants as instruments will predict opposite associations: alleles linked to higher GPX4 activity (lower ROS) will correlate with lower fibrosis scores but higher frailty indices, reflecting loss of pathogenic SASP; alleles linked to lower SOD2 will correlate with higher frailty and higher serum IL‑6, indicating a shift toward pathogenic EVs.
4. Single‑cell EV RNA‑seq from human skin biopsies will reveal a bimodal distribution of EV‑associated miR‑29 versus IL‑6 transcripts, with the breakpoint aligning with plasma 8‑iso‑PGF2α levels, a marker of oxidative stress.

Experimental Design

• In vitro: Induce senescence in primary human fibroblasts with sub‑lethal doxorubicin; treat parallel cultures with MitoTEMPO or antimycin A to modulate ROS. Harvest EVs, quantify miR‑29, SOD2, IL‑6, CXCL‑8 by qPCR and ELISA; assess recipient cell oxidative stress and proliferation.
• In vivo: Use p16‑3MR mice to label senescent cells; after CCl₄‑induced liver injury, administer MitoTEMPO or overexpress SOD2 via AAV8. Isolate EVs from liver, perform proteomics and small‑RNA seq, measure hydroxyproline content and serum ALT/AST.
• Human MR: Summary‑level data from GWAS of GPX4 (rs12345) and SOD2 (rs67890) linked to fibrosis (MetaAnalysis of NAFLD) and frailty (Frailty Index GWAS). Conduct inverse‑variance weighted MR, test for pleiotropy with MR‑Egger.
• Clinical correlation: Collect skin punch biopsies from young and elderly donors; isolate EVs, run Nanopore RNA‑seq for miR‑29 and IL‑6, correlate with plasma 8‑iso‑PGF2α.

If the ROS‑dependent EV switch holds, senolytics timed to chaperone‑phase depletion will impair repair, whereas ROS‑lowering adjuvants will preserve beneficial EV signaling while still clearing pathogenic senescent cells. This reframes the senolytic debate from "remove all" to "modulate the EV signaling state".