Hypothesis: Age-related increase in epigenetic noise preferentially silences transcription factors that reside in the peripheral positions of gene regulatory networks, and this peripheral silencing drives the observed modular loosening of co‑expression modules; consequently, restoring youthful expression of these peripheral TFs, rather than central hub TFs, yields the greatest improvement in network information transmission and phenotypic rejuvenation.

Rationale

• Epigenetic noise manifests as increased variance in DNA methylation at specific loci, a process captured by the EVORA approach that detects outlier changes rather than average shifts EVORA approach detects cancer risk markers through outlier DNAm changes.
• Peripheral network positions are enriched for age‑associated methylated genes that show low connectivity and centrality, suggesting these loci are intrinsically vulnerable to rewiring Peripheral localization appears intrinsic....
• Master regulators such as EZH2 can restore youthful expression patterns when overexpressed, indicating that certain TFs act as bottlenecks for information flow EZH2 emerged as particularly potent.

Novel Mechanistic Insight

We propose that epigenetic noise does not act randomly; instead, it preferentially targets promoter and enhancer regions of peripheral TFs because these regions reside in chromatin compartments with higher baseline accessibility and lower nucleosome density. Increased stochastic methylation/demethylation at these sites alters the binding affinity of the TFs themselves and of co‑activator complexes, leading to a functional silencing that is not accompanied by reduced TF protein levels but by loss of DNA‑binding activity. Because peripheral TFs sit at the edges of modules, their weakened activity reduces cross‑talk between neighboring genes, causing the observed decline in module co‑expression (modular loosening). Central hub TFs, by contrast, are buffered by redundant interactions and higher-degree connectivity, making their activity more resistant to noise‑induced fluctuations.

Testable Predictions

1. In aged tissues, peripheral TFs will show a significant increase in cell‑to‑cell variance of promoter methylation compared with central TFs, measurable by single‑cell bisulfite sequencing.
2. Restoring the expression of a peripheral TF (e.g., a liver‑specific TF identified as low‑centrality) in aged mice will improve mutual information between target genes by >15 % and reduce fibrosis more effectively than overexpression of a central hub TF of similar expression level.
3. Artificially increasing epigenetic noise at peripheral TF promoters in young cells (using dCas9‑TET1 or dCas9‑DNMT3A to induce stochastic methylation changes) will recapitulate the modular loosening signature seen in aged networks, without altering overall TF protein abundance.
4. Knock‑down of a peripheral TF in young cells will increase expression variance of its downstream module and decrease network information transmission, mimicking the aging phenotype.

Experimental Design

• Single‑cell multi‑omics: Perform scRNA‑seq coupled with scBS‑seq on young (3 mo) and aged (24 mo) mouse livers. Compute methylation variance per TF promoter and correlate with network centrality (derived from co‑expression maps). Expect higher variance for low‑centrality TFs in aged cells.
• AAV‑mediated TF overexpression: Select one peripheral TF (low centrality, high age‑associated methylation variance) and one central TF (high centrality, low variance). Deliver via AAV8 to aged mice; assess liver fibrosis, hepatic fat, glucose tolerance, and compute gene‑regulatory information transfer using the mutual information metric from the single‑gene knock‑in study single gene knock‑ins can restore information transmission....
• Noise induction: Use CRISPR‑epigenetic editors to introduce randomized methylation changes at promoters of a peripheral TF in primary hepatocytes from young mice. Measure resulting changes in module co‑expression and information flow.
• Loss‑of‑function: siRNA knock‑down of the same peripheral TF in young hepatocytes; evaluate whether module loosening and reduced information transfer occur.

Falsifiability

If peripheral TFs do not show higher methylation variance than central TFs, or if restoring peripheral TFs fails to outperform central TFs in rescuing network information and phenotypes, the hypothesis would be refuted. Likewise, if inducing noise at peripheral TF promoters does not reproduce modular loosening, the proposed causal chain would be invalid.

Implications

Validating this hypothesis would shift the focus from global epigenetic drift to locus‑specific noise that targets network periphery, explaining why certain master regulators (e.g., EZH2) have disproportionate rejuvenative power. It would also provide a mechanistic basis for the EVORA outlier‑detecting strategy: the same stochastic methylation jumps that serve as cancer risk markers may, in normal aging, preferentially hit peripheral TFs and initiate functional decline.