Hypothesis\n\nProtein aggregates are not inert garbage but form phase‑separated hubs that retain active UPR sensors (IRE1, PERK, ATF6) while excluding the transcriptional machinery needed for downstream response. This spatial uncoupling creates a local ‘signaling‑on, transcription‑off’ state that buffers proteotoxic stress by sequestering misfolded clients without triggering energetically costly chaperone synthesis. Over time, hub maturation reduces sensor availability, contributing to the decoupled UPR observed in senescent cells.\n\n### Mechanistic model\n\n1. Nucleation – Misfolded proteins expose hydrophobic patches that drive liquid‑liquid phase separation with chaperones and UPR sensors.\n2. Maturation – As the hub ages, hydrophobic cores become more ordered, converting the liquid droplet into an amyloid‑like solid that still presents sensor luminal domains on its surface.\n3. Selective exclusion – The dense interior sterically hinders large transcriptional complexes (RNA Pol II, splicing factors) while permitting small signaling molecules and sensor cytosolic domains to remain active.\n4. Feedback attenuation – Persistent sensor activity without transcriptional output leads to chronic IRE1‑XBP1 splicing (as seen in senescent cells) without downstream ATF6/XBP1‑s target induction, reproducing the decoupled stress response.\n\n### Testable predictions\n- Aggregates will colocalize with phosphorylated IRE1 and PERK but show depletion of RNA Pol II and spliced XBP1‑s transcripts at sub‑micron resolution.\n- Artificial induction of reversible aggregates (e.g., optogenetic clustering) will increase IRE1 phosphorylation without raising BiP or CHOP expression.\n- Dissolving aggregates with chemical chaperones (4‑PBA) or disaggregases (Hsp110) will restore transcriptional output and reduce senescence markers.\n- In vivo, regions of high aggregate burden in aged mouse brain will display a spatial mismatch between UPR sensor activation (phospho‑IRE1) and downstream chaperone mRNA levels.\n\n### Experimental approach\n- Apply expansion microscopy combined with multiplexed error‑robust FISH (MERFISH) to map UPR sensor proteins, aggregate markers (p62, ubiquitin), and nascent transcripts at ~0.05 µm resolution in primary fibroblasts and brain slices from young, middle‑aged, and old mice.\n- Use live‑cell FRET‑based IRE1 activity reporters alongside MS2‑MCP labeling of XBP1‑s mRNA to monitor real‑time signaling vs transcription within individual aggregates.\n- Perturb aggregate dynamics with optogenetic Cry2‑clustering of a model misfolded protein and quantify changes in sensor phosphorylation, transcriptional bursting, and senescence‑associated β‑galactosidase.\n- Validate findings in the 4.2 million‑cell mouse brain atlas dataset by correlating spatial aging clock scores with aggregate‑sensor transcriptomics mismatches.\n\nIf aggregates function as ordered signaling hubs, disrupting their phase‑separated state should exacerbate proteotoxic stress rather than alleviate it, directly challenging the view that they are merely the end‑point of proteostatic collapse.
