Hypothesis\n\nIn aged neurons, stochastic epigenetic erosion increases transcriptional noise at specific loci, leading to probabilistic activation of the ATF4‑DDIT3‑CEBPG degeneration regulon. We're proposing that this noise can flip a switch that upregulates “eat‑me” signals such as complement component C1q and exposed phosphatidylserine. Simultaneously, age‑associated H3K4me3 and H3K27ac enrichment at microglial inflammatory genes primes phagocytic receptors (e.g., CR3) for heightened uptake of these marked neurons. Consequently, neurons that are both epigenetically noisy and metabolically expensive—identified by low synaptic activity and high ATP demand—are selectively eliminated, coupling cellular inefficiency to an active quality‑control mechanism.\n\n## Mechanistic Rationale\n\n- Epigenetic drift in aging produces cell‑to‑cell variability in chromatin accessibility (Aging brains exhibit widespread epigenetic erosion...).\n- In injury models, ATF4, Ddit3 and Cebpg constitute a core degeneration transcriptional network that is antagonized by survival factors Pax6, Meis2 and Pou4f1/2 (Injury models in retinal ganglion cells...).\n- Aged microglia display priming via H3K4me3 and H3K27ac at inflammatory promoters, positioning them for rapid response (Aged microglia show epigenetic priming...).\n- We propose that the noise‑driven ATF4 module directly induces expression of complement C1q and ligands for microglial phagoptosis, creating a selective “eat‑me” tag that is recognized by the primed microglial state. It's this link that turns stochastic epigenetic variation into a targeted clearance signal.\n\n## Testable Predictions\n\n1. Single‑cell multi‑omics from aged mouse cortex will reveal a subpopulation of neurons with high ATF4‑DDIT3‑CEBPG signature, elevated C1q mRNA, and increased chromatin variance at ATF4 enhancers.\n2. Pharmacological inhibition of ATF4 (e.g., with ISRIB) or genetic knockdown of Cebpg will reduce neuronal loss in aged brains without globally suppressing microglial inflammation. It's expected that overall cytokine levels remain unchanged.\n3. Disrupting microglial H3K4me3 deposition (using a CRISPR‑dCas9‑KRAB targeting the C3ar1 promoter) will blunt phagocytosis of ATF4‑high neurons despite their presence.\n4. Neurons identified as ATF4‑high will show lower expression of synaptic activity genes (e.g., Synapsin1) and higher mitochondrial respiration markers, linking metabolic cost to susceptibility. They can't be rescued by generic anti‑oxidants if the ATF4 pathway is intact.\n\n## Experimental Approach\n\n- Perform snRNA‑seq + snATAC‑seq on cortical samples from young (3 mo) and aged (24 mo) mice.\n- Identify neuronal clusters co‑expressing ATF4 regulon and complement genes; quantify chromatin accessibility variance at ATF4 binding sites.\n- Validate C1q surface deposition by immunofluorescence and phagoptosis markers (e.g., annexin V) in situ.\n- Apply ISRIB via osmotic pump to aged mice for 4 weeks; assess neuron numbers, microglial phagocytic cups (Iba1+/CD68+ contacts with NeuN+ cells), and behavioral performance. It's a short‑term intervention that should reveal causal links.\n- Use microglia‑specific H3K4me3 demethylase (KDM5B) overexpression to test priming requirement. This approach doesn't require genetic ablation of microglia, preserving overall immune surveillance.\n\n## Potential Outcomes\n\nIf predictions hold, neuronal loss in aging reflects an active, noise‑triggered quality‑control pathway rather than passive damage. Failure to observe ATF4‑dependent C1q upregulation or lack of rescue by ATF4 inhibition would falsify the hypothesis, supporting alternative models of stochastic degeneration.