The age-related drop in autophagy isn't just a byproduct of overactive mTORC1. I suspect it’s actually a self-reinforcing, damaging feedback loop where CDK5-mediated phosphorylation of HSP90AA1 acts as a molecular "lock" on the promoters of ATG5, ATG7, and ATG12. In aging tissues, CDK5 accumulates at the nuclear periphery and acts as a "phospho-sink." This sequesters HSP90AA1 in a shape that can’t facilitate TFEB nuclear import. The result is a form of epigenetic entrapment: TFEB isn't just missing from the nucleus; it’s functionally evicted from its high-affinity CLEAR-motif binding sites. This leaves the chromatin locked in a repressive state, preventing rapid reactivation even if you acutely inhibit mTORC1.

While mTORC1 suppresses TFEB through S211 phosphorylation [PMID: 39158307], the CDK5-HSP90AA1 axis [PMID: 36854744] functions as a distal gatekeeper. My hunch is that when HSP90AA1 fails to interact with TFEB, TFEB defaults to a low-affinity state at ATG gene promoters. Over time, this failure to bind recruits Histone Deacetylases (HDACs), which remodel the local chromatin architecture. This permanently lowers the "autophagy floor," which explains why traditional drugs like rapamycin often fall short in late-stage aging—the system is structurally silenced.

To test this, we need to move past simple fractionation. My experimental plan involves three steps:

1. ChIP-Re-ChIP Analysis: I’ll use sequential ChIP on aged versus young mouse tissue (liver or neural) to look for TFEB and H3K9me3 at ATG5/7/12 loci. If TFEB eviction is linked to heterochromatin formation, we’ll see it here.
2. Phospho-mimetic HSP90AA1 Knock-in: We should generate an HSP90AA1(S595D) constitutive mimic model. If I’m right, these mice will show premature aging and stalled autophagic flux regardless of their mTORC1 or AMPK status, proving this CDK5-HSP90AA1 interaction is a primary bottleneck.
3. Proximity Ligation Assays (PLA): I plan to quantify the physical interaction between TFEB and HSP90AA1 in senescent cells after treating them with CDK5 inhibitors.

If this holds up, it shifts the focus away from global TFEB activation as the primary goal. Instead, we might need a two-step approach: first, inhibit the CDK5-HSP90AA1 switch to "unlock" the promoters, and only then inhibit mTORC1 to drive nuclear translocation. Simply forcing TFEB into the nucleus without addressing the HSP90AA1-mediated block at the promoter level likely explains why autophagy-inducing drugs have shown such inconsistent results in neurodegeneration trials.