Hypothesis

Age‑stratified Mendelian randomization (MR) can detect a temporal switch in the causal effect of genetically predicted protein levels on telomere length, whereby variants that increase MST1 expression accelerate telomere attrition in early adulthood but become protective after mid‑life due to age‑dependent changes in chromatin state and SASP‑mediated regulation of metabolite transporters.

Mechanistic rationale

1. Chromatin accessibility shifts – ATAC‑seq data show that the promoter region of MST1 gains repressive histone marks after age ~50 in human endothelial cells, reducing transcriptional response to genetic variants.
2. Competing pathways – In youth, MST1 activates Hippo signaling, suppressing YAP‑driven proliferation and exacerbating oxidative stress; in older tissue, chronic SASP elevates IL‑6 and TGF‑β, which feedback to inhibit MST1 activity and instead promote YAP‑dependent repair.
3. Metabolite mediation – MR‑identified mediation by 1,5‑anhydroglucitol (1,5‑AG) operates only when the SLC5A transporter is expressed; SLC5A expression declines with age, attenuating the mediator pathway and unmasking a direct protective effect of MST1 on telomere maintenance via altered NAD+ metabolism.

Testable predictions

• In a longitudinal cohort with genotype, plasma protein levels (or surrogate mRNA), and leukocyte telomere length measured at multiple ages, the interaction term between MST1‑associated genotype and age will be significant (p<0.01) and change sign around 45‑55 years.
• Stratified MR (e.g., using the IVW method separately in age bins 20‑35, 36‑50, 51‑65, 66‑80) will show a positive causal estimate (higher MST1 → shorter telomeres) in the youngest bin, a null effect in middle age, and a negative estimate (higher MST1 → longer telomeres) in the oldest bin.
• Colocalization of MST1 eQTLs with age‑specific chromatin marks (e.g., H3K27ac peaks) will demonstrate loss of regulatory overlap after age 50, supporting the mechanistic switch.
• Experimental inhibition of SLC5A in cultured human fibroblasts will abolish the mediation of 1,5‑AG on telomere length, confirming the age‑dependent metabolite route.

Falsifiability

If age‑stratified MR shows a homogeneous effect direction across all age bins, or if the interaction term is non‑significant, the hypothesis is refuted. Likewise, if chromatin accessibility assays reveal no age‑related change at the MST1 locus, the proposed mechanism lacks support.

Implications

Demonstrating age‑dependent causal effects would reconcile discrepancies between MR‑derived protein‑aging associations and the TEP‑observed age‑specific efficacy of pharmacological interventions, urging the adoption of dynamic instrumental variable frameworks in aging biomarker discovery.

References

• Temporal Efficacy Profiler reveals age‑dependent intervention effects [1]
• MR identifies protein‑telomere links with mediation by 1,5‑AG [2]
• Doubly robust SCM/SDID frameworks for causal inference [3]
• Synthetic control requires unexposed donor pools [4]

Open data suggestion

Apply the proposed analysis to the UK Biobank genotype‑phenotype resource, leveraging its repeated telomere length measurements and age‑stratified eQTL catalogs.