Background

Aging reduces mutual information (MI) between transcription factors (TFs) and target genes due to input distribution mismatch, reflecting uncoordinated TF activity rather than DNA sequence damage [https://pubmed.ncbi.nlm.nih.gov/41278781/]. This decay is amplified by network centralization and loss of stabilizing feedback motifs [https://www.fightaging.org/archives/2025/03/gene-regulatory-networks-in-the-design-of-approaches-to-slow-aging/]. Two epigenetic noise layers have been identified: TET2‑mediated 5hmC loss that diminishes TF binding site accessibility (e.g., SRF/myocardin) [https://pdfs.semanticscholar.org/60f1/469eedf107e7f9854c9625b9fa4ba78e22f8.pdf] and dysregulation of transcription elongation factors (NELF, SPT6, ELOA) that alters isoform‑specific RNA production and can rejuvenate fibroblasts when ELOA is depleted [https://news.feinberg.northwestern.edu/2025/12/18/exploring-the-connection-between-gene-expression-and-aging/]. However, how these layers converge to produce the observed TF input mismatch remains unclear.

Hypothesis

We propose that the chromatin remodeler CHD4 (a core subunit of the NuRD complex) couples TET2‑dependent 5hmC loss and elongation factor dysregulation by modulating nucleosome occupancy at enhancer‑promoter hubs. In aged cells, reduced 5hmC decreases CHD4 recruitment to certain enhancers, leading to aberrant nucleosome positioning that sterically hinders TF binding and simultaneously impedes the recruitment of pausing factors (NELF) and processivity factors (SPT6/ELOA). This dual defect creates a coordinated shift in TF input distribution—increasing network centralization around a few hub TFs while eroding stabilizing feedback loops—thereby driving the measurable decline in MI. Restoring CHD4 activity at specific loci should realign TF input distributions and recover MI without globally altering TF expression.

Testable Predictions

1. Correlation: In multiple aged tissues (muscle, liver, fibroblasts), CHD4 occupancy at TF‑target enhancers will inversely correlate with local 5hmC levels and directly correlate with nucleosome density (measured by MNase‑seq) and RNA polymerase II pausing indices (NELF/SPT6 ChIP‑seq).
2. Causality: Acute depletion of CHD4 in young cells will recapitulate the aged MI loss profile (↓MI, ↑network centrality, ↓feedback motif strength) and increase senescence markers; conversely, targeted overexpression of CHD4 (via dCas9‑CHD4 fusion) in aged cells will rescue MI by ≥8% and reduce senescence-associated secretory phenotype (SASP).
3. Specificity: Rescue will be most pronounced for mitochondrial‑linked hub TFs (Ppara, Esrra, Ppargc1b) identified in silico knock‑in studies [https://arxiv.org/html/2601.04016v1][https://pubmed.ncbi.nlm.nih.gov/41542164/], while global TF overexpression will produce smaller MI gains and risk off‑target network effects.
4. In vivo validation: Liver‑specific CHD4 overexpression in 20‑month‑old mice will improve hepatic histology and circulating ALT/AST levels, paralleling the rejuvenation observed with EZH2 upregulation [https://www.nih.gov/news-events/nih-research-matters/manipulating-gene-activity-reverse-aging], and will increase hepatic MI measured by single‑cell TF‑target correlation.

Experimental Design

• Omics profiling: Perform scATAC‑seq, scRNA‑seq, CUT&RUN for CHD4, 5hmC (hMeDIP), NELF, SPT6, and ELOA in young (3 mo) and aged (24 mo) mouse muscle, liver, and fibroblasts. Compute MI between TF activity (inferred from motif accessibility) and target gene expression; assess network centrality (betweenness) and feedback motif density (e.g., feed‑forward loops).
• Perturbations: Use CRISPRi to knock down CHD4 in young primary fibroblasts; use AAV‑dCas9‑CHD4 to target enhancers of Ppara, Esrra, Ppargc1b in aged fibroblasts and in vivo mouse liver. Include controls (non‑targeting dCas9, scrambled sgRNA).
• Readouts: MI recalculation after each perturbation, SA‑β‑gal staining, SASP cytokine ELISA, mitochondrial respiration (Seahorse), and histological scoring.
• Statistical tests: Linear mixed‑effects models linking CHD4 occupancy, 5hmC, nucleosome density, and MI; rescue significance set at p < 0.01 with Bonferroni correction for multiple tissues.

Potential Implications

If validated, this hypothesis positions CHD4 as a mechanistic nexus that translates epigenetic noise (5hmC loss) and transcriptional elongation dysregulation into a unified loss of transcriptional information flow. It suggests that locus‑specific chromatin remodeling, rather than broad TF modulation, could efficiently restore network coherence and delay senescence, offering a precise synthetic‑biology route to prolongevity interventions.