EPIGENETIC STATES ACROSS IMMUNE MEMORY, TRAINING AND EXHAUSTION: A CLAIM-CALIBRATED SYNTHESIS FROM MECHANISM TO THERAPEUTIC REPROGRAMMING

Immune memory cannot be explained solely by antigen-receptor clonotypes. Immune cells can also retain regulatory configurations whose effects depend on lineage, tissue and the conditions under which recall occurs. In this mechanistic review, we evaluate the literature available through July 2026 on DNA methylation, histone modification, chromatin accessibility, three-dimensional genome organization, non-coding RNA, and immunometabolism in adaptive memory, trained and tolerized innate responses, exhaustion, and disease-associated inflammatory memory. The evidence is interpreted within a five level framework. The foundation is composed of descriptive maps. Prospective prediction and functional recall are distinct but complementary pathways to causal perturbation and, ultimately, clinical usefulness. The level provided is the strongest inference supported by one study. It is not a formal bias assessment, nor a numeric quality score. Single-cell, longitudinal and perturbational studies are increasingly suggesting that the fate of the immune system is determined not by a single epigenetic mark, but by the order of incoming signals, the tissue context and cell identity. Interventions under investigation include chromatin-modifying drugs, manipulation of trained immunity, combinations with checkpoint blockade, and locus-specific epigenome editing. But persistent accessibility is not enough to suggest memory, predictive association is not enough to infer causation, and a causal regulatory element is not necessarily a safe therapeutic target. Hence, clinical translation will rely on longitudinal validation, cell-resolved biomarkers, confirmed target engagement, controllable persistence, selective delivery, and careful monitoring of unintended immune effects.