Divergent exhaustion networks define memory CD8+ T cell subset fate under chronic stimulation and reveal targets for epigenetic reprogramming

Abstract T cell exhaustion (TEX) impairs the efficacy of cancer immunotherapies. Following activation from an acutely resolved infection or vaccination, naïve CD8+ T cells (TN) differentiate into heterogeneous effector (TEFF) and memory (TMEM) subsets, each with distinct functional capacities that shape their persistence and antitumor potential. In contrast, during a chronic infection or cancer, TN differentiate into heterogenous TEX subsets. Whether and how non-TN subsets undergo exhaustion upon chronic antigen encounter remains unknown. This is a key knowledge gap, since most T cells used in immunotherapies such as CART cells and indeed most T cells in adult humans are not TN. While epigenetic reprogramming is known to underpin transitions in TMEM and TEX states from TN, the molecular drivers of exhaustion (particularly when arising from distinct differentiation stages) remain poorly defined. We hypothesized that chronic stimulation, a hallmark of tumor microenvironments, accelerates epigenetic remodeling in the more terminally differentiated TMEM subsets, predisposing them to rapid dysfunction. In contrast, less differentiated subsets retain greater chromatin plasticity and resilience, allowing them to resist exhaustion and preserve functional capacity over time. To investigate this, we sort purified human peripheral CD8+ T cell subsets from healthy donors and subjected them to repetitive chronic stimulation for over 3 weeks. In addition to naïve (TN) CD8+ cells, we studied stem cell memory (TSCM), central memory (TCM), effector memory (TEM1 and TEM2), and terminally differentiated effector memory (TEMRA) cells. TEMRA and TEM2 cells exhibited accelerated functional decline, marked by impaired proliferative capacity and diminished tumor clearance compared to all other subsets. In contrast, less differentiated subsets (TSCM, TCM) demonstrated greater durability and sustained antitumor activity compared to TN, suggesting a differentiation state-dependent hierarchy in exhaustion susceptibility. Longitudinal multiomic profiling of human CD8+ T cells across differentiation states and stimulation time points were used to map the transcriptional and epigenetic trajectories that underlie T cell exhaustion. By integrating dynamic gene regulatory networks with changes in chromatin accessibility, we identified how subset identity and differentiation state shape susceptibility or resilience to exhaustion and the capacity to maintain functional plasticity. These analyses revealed subset-specific regulatory networks, including exhaustion priming enhancers and resilience-associated chromatin modules, that dictate long-term cell fate. By decoding these networks, this work provides a blueprint for engineering interventions that reprogram T cell differentiation trajectories or preserve epigenetic flexibility in therapeutic contexts. Targeting these regulatory circuits may enable more durable T cell function, delay exhaustion onset, and amplify the efficacy of immunotherapies of hematologic malignancies, ultimately bridging the gap between transient responses and lasting remission in patients.