Hypothesis

Protein aggregation initially sequesters dysfunctional mismatch repair (MMR) proteins in aging colonic crypts, preserving genomic stability; however, as aggregates mature they impair crypt stem cell function and promote the expansion of dMMR clones that seed field cancerization.

Mechanistic Model

1. Early protective phase – Age‑related decline in chaperone activity and proteasome capacity increases the pool of misfolded MMR subunits (MLH1, MSH2, MSH6, PMS2). Transient oligomeric aggregates act as sinks that lock away these defective proteins, preventing dominant‑negative interactions with the remaining wild‑type complexes and preserving mismatch repair activity. This aligns with the idea that aggregates represent the proteome’s last attempt at order [[Ending Aging.md]].
2. Transition to pathology – Persistent oxidative stress and declining autophagic flux cause aggregates to grow into insoluble, amyloid‑like deposits. These deposits sequester not only misfolded MMR proteins but also essential co‑factors (e.g., EXO1, PCNA) and chaperones (HSP70, HSP90), thereby reducing the functional MMR pool below a critical threshold. Moreover, aggregate surfaces aberrantly activate integrin‑FAK signaling, driving Wnt‑independent crypt hyperplasia and creating a permissive niche for clonal expansion.
3. Propagation of field cancerization – Crypts harboring aggregate‑laden, MMR‑deficient stem cells acquire a growth advantage under inflammatory cytokines (IL‑6, TNF‑α) that are elevated in aged mucosa. The resulting patches of dMMR epithelium spread laterally via fission, producing the epigenetically altered field observed in >50% of normal‑appearing mucosa adjacent to tumors [[3]].

Testable Predictions

• In colonic crypts from young donors (<30 y), MMR proteins will be diffusely soluble; in crypts from aged donors (>65 y) a fraction will co‑localize with detergent‑insoluble aggregates, and this fraction will increase with epigenetic age acceleration (measured by GrimAge) [[1],[4]].
• Pharmacological enhancement of autophagy (e.g., spermidine) or inhibition of amyloid formation (e.g., EGCG) will reduce insoluble MMR burden and delay the onset of dMMR clones in aged mouse colon organoids, whereas proteasome inhibition will accelerate aggregate formation and MMR loss.
• Inducible expression of an aggregation‑prone MLH1 mutant will recapitulate the protective‑to‑pathopathic transition: early aggregates will correlate with preserved MMR activity, while later aggregates will correlate with loss of repair and increased microsatellite instability.
• Spatial transcriptomics of aggregate‑positive crypts will show up‑regulation of senescence‑associated secretory phenotype (SASP) genes and Wnt‑independent growth pathways, preceding field‑wide methylation changes.

Experimental Approach

1. Human tissue – Obtain colonic biopsies from colonoscopy‑screened individuals stratified by age and epigenetic age acceleration. Perform sequential extraction (soluble, Triton‑X‑100 soluble, SDS‑soluble) followed by immunoblot for MLH1/MSH2 and immunofluorescence for aggregate markers (p‑Ser129‑like ubiquitin, Thioflavin‑S). Correlate insoluble MMR load with dMMR status (MSH2/MSH6 loss) and with GrimAge acceleration.
2. Organoid models – Derive colon organoids from young and aged donors; treat with autophagy modulators or aggregation inhibitors; assess MMR function using a fluorescent reporter of microsatellite stability and measure clone formation over 30 days.
3. Mouse models – Cross a knock‑in of aggregation‑prone MLH1 (e.g., MLH1‑ΔNLS) with a Villin‑CreERT2 line; induce expression in adult mice; longitudinally monitor MMR immunofluorescence, tumor burden, and methylation arrays.
4. Readouts – Quantify aggregate burden (filter trap assay), MMR activity (in vitro mismatch repair assay), senescence (p16, SASP cytokines), and field methylation (EPIC array) to test the predicted trajectory.

If data show that early aggregation preserves MMR function while later aggregation drives its loss, the hypothesis gains support; conversely, if aggregation merely correlates with loss without a protective phase, the model is refuted.