Autophagy‑driven splicing feedback creates a bistable switch that locks cells into senescence or resilience

Hypothesis

Autophagy does not merely degrade p16^INK4a; it also modulates the activity of splicing regulators hnRNP A1/A2 and SF2/ASF through nutrient‑sensing pathways, thereby biasing CDKN2A alternative splicing toward the anti‑senescence p14^ARF isoform. When autophagic flux is high, increased amino‑acid recycling activates mTORC1, which phosphorylates hnRNP A1/A2, reducing its nuclear import and shifting the hnRNP A1/A2 : SF2/ASF ratio in favor of SF2/ASF. This shift promotes p14^ARF production, which in turn sustains autophagy by inhibiting p53‑mediated transcriptional repression of autophagy genes. Conversely, when autophagy falters, depleted amino‑acid pools lower mTORC1 activity, hnRNP A1/A2 accumulates in the nucleus, the ratio shifts toward hnRNP A1/A2, and p16^INK4a splicing predominates. The newly synthesized p16 further impairs autophagy by sequestering p62, establishing a double‑negative feedback loop that locks the cell into a senescent state. Thus autophagy acts as a rheostat that simultaneously controls p16 protein stability and the transcriptional‑splicing output of the CDKN2A locus, creating a bistable switch between resilience and senescence.

Mechanistic Basis

• Autophagy‑derived amino acids activate mTORC1 → phosphorylation of hnRNP A1/A2 → cytoplasmic retention.
• Cytoplasmic hnRNP A1/A2 cannot compete with SF2/ASF for pre‑mRNA binding → increased p14^ARF isoform.
• p14^ARF stabilizes MDM2, limiting p53 activity and preserving transcription of autophagy genes (e.g., LC3, ATG5).
• Loss of autophagy reduces amino‑acid signaling, mTORC1 activity drops, hnRNP A1/A2 relocates to nucleus.
• Nuclear hnRNP A1/A2 promotes exon skipping that yields p16^INK4a.
• Accumulated p16 binds p62, inhibiting autophagosome formation and further reducing flux.

Testable Predictions

1. Flux manipulation – Treating cells with rapamycin (to inhibit mTORC1) under nutrient‑rich conditions will increase nuclear hnRNP A1/A2, raise the hnRNP A1/A2 : SF2/ASF ratio, and elevate p16^INK4a mRNA isoform ratio without changing total CDKN2A transcription. Conversely, Torin1 washout or leucine supplementation should shift splicing toward p14^ARF and reduce p16 protein levels only when autophagy is functional.
2. p62 sequestration assay – Overexpressing p16^INK4a in autophagy‑competent cells will decrease LC3‑II conversion and increase p62‑positive puncta; this effect will be abolished in p62‑knockout cells, confirming that p16 impairs autophagy via p62 sequestration.
3. Isoform‑specific rescue – Expressing a p14^ARF construct resistant to splicing regulation in autophagy‑deficient (ATG5 KO) cells will restore mTORC1 activity, reduce nuclear hnRNP A1/A2, and lower p16 levels, rescuing the senescence phenotype despite blocked autophagic flux.
4. In vivo read‑out – Mice with a knock‑in reporter that fluoresces green for p16^INK4a and red for p14^ARF will show a shift from red‑dominant to green‑dominant signal in tissues where autophagy is genetically inhibited (e.g., Atg7‑LKO) and this shift will be preventable by systemic leucine supplementation or mTORC1 activation.

If any of these predictions fail—for instance, if altering autophagy flux does not change the hnRNP A1/A2 : SF2/ASF ratio or isoform balance, or if p16 overexpression fails to suppress autophagic flux in a p62‑dependent manner—then the hypothesis that autophagy governs a splicing feedback loop controlling CDKN2A isoform choice would be falsified.