Hypothesis

Age-related epigenetic decay in neurons increases stochastic transcriptional noise, which the brain counters by strengthening reliance on established predictive models; the observed behavioral rigidity is therefore a compensatory response to noisy internal representations, not a direct result of chromatin over-consolidation.

Mechanistic Rationale

Signal et al. (2024) show that aged cortical neurons lose H3K27me3 at embryonic developmental superenhancers and lose H3K27ac at synaptic signaling enhancers, while H3K4me3 remains stable Signal et al., 2024. This pattern reflects erosion of both repressive and activating marks, leading to two concurrent molecular consequences:

1. Loss of H3K27me3 permits low‑level transcription from normally silenced developmental loci, generating aberrant transcripts.
2. Loss of H3K27ac reduces expression of synaptic proteins, weakening signal‑dependent plasticity.

The ectopic transcripts produce transcriptional noise that interferes with precise neuronal signaling. Networks exposed to elevated noise increase their prior variance in Bayesian inference models, shifting the cost‑benefit balance toward preserving existing weights rather than updating them. Behaviorally, this appears as reduced flexibility and pattern entrenchment, mimicking over‑consolidation but actually arising from a homeostatic attempt to maintain signal‑to‑noise ratio.

In contrast, glioblastoma exhibits targeted H3K27me3 loss coupled with H3K4me3 gain at proliferation drivers, a coordinated epigenetic reprogramming that supports adaptive plasticity Kozłowska et al., 2019. Aged neurons lack this coordinated activation, supporting the drift‑rather‑than‑program model.

Testable Predictions

• Restoring H3K27me3 levels at developmental superenhancers in aged neurons will reduce ectopic transcription without altering global H3K4me3.
• Enhancing H3K27ac at synaptic enhancers will rescue synaptic gene expression and improve behavioral flexibility.
• Combined epigenetic stabilization will decrease transcriptional noise (measured by nascent RNA seq variability) and correlate with improved performance on reversal learning tasks.
• If rigidity stems from compensatory over‑reliance on priors, then artificially increasing environmental uncertainty (e.g., variable reward schedules) will exacerbate behavioral deficits in aged animals unless epigenetic noise is first reduced.

Experimental Approach

1. Pharmacological/Epigenetic Intervention: Treat aged mice (24 mo) with an EZH2 agonist (to boost H3K27me3) and/or a p300/CBP activator (to restore H3K27ac). Include vehicle and young controls.
2. Molecular Readouts: Perform ChIP‑seq for H3K27me3, H3K27ac, H3K4me3; quantify ectopic transcripts from developmental loci via RNA‑seq; assess synaptic gene expression.
3. Noise Assessment: Use single‑cell nascent RNA sequencing (EU‑labeling) to measure transcriptional variance across neurons.
4. Behavioral Tests: Measure reversal learning, set‑shifting, and novelty‑seeking before and after treatment.
5. Uncertainty Challenge: Introduce a probabilistic reversal learning paradigm to test whether increased environmental volatility worsens performance in untreated aged mice but not in epigenetically rescued animals.

Falsification would occur if EZH2/p300 modulation fails to reduce ectopic transcripts or transcriptional noise, or if behavioral flexibility does not improve despite molecular rescue, indicating that the observed rigidity is not a compensatory response to epigenetic drift‑induced noise.