Reprogramming Exhausted Cytotoxic T Cells Through Transcription Factor Overexpression

Abstract

T cells are part of the immune system’s adaptive defence; they specifically target and kill both virally infected and cancerous cells through antigen recognition. During an acute infection, effector T cells (TEFF) mount a robust response to ultimately clear all virally infected cells. Since T cells can respond rapidly to infection with high proliferative and cytotoxic potential, the immune system has evolved to contain inhibitory signals with multiple T cell intrinsic and extrinsic cues to quickly halt T cell response after viral clearance. While important for defending the body from widespread autoimmune disease, in the context of chronic viral infection or cancer these inhibitory signals curtail the T cell response. One of the intrinsic inhibitory signals arises from repeated stimulation of the T cell receptor (TCR), pushing T cells down a differentiation path broadly termed T cell exhaustion. Eventually T cells are driven to a terminal exhausted state (TEX) resulting in diminished T cell proliferation and incomplete tumor/viral clearance. The concept of using T cells directly as a therapy for cancer, adoptive T cell therapy (ACT), has proven incredibly successful in recent years but has been limited to effectiveness in haematological malignancies. In solid tumors, T cells are driven to the TEX state by chronic antigen stimulation and immune checkpoint signalling. Approaches that have attempted to prevent T cell exhaustion have had tremendous success for cancer immunotherapy as exemplified by the best-selling drug Keytruda, an immune checkpoint inhibitor. Understanding and developing approaches to improve the T cell mediated response during chronic antigen stimulation has the potential to not only improve cancer and viral therapy but also elicit higher cure rates for patients.Modern genome engineering technologies have dramatically advanced ACT by programming TEX cells into desirable phenotypic states. Transcription factors (TFs) are part of the regulatory logic that wires cells to distinct lineages by either directly binding to genomic DNA or interacting with other DNA binders. By targeting TFs, researchers have been able to identify key drivers of cell differentiation trajectories. Given our lab’s expertise in high-throughput genetic perturbation screening to drive cells to specific states, we applied a library of all full-length human TFs (multiplexed Overexpression of Regulatory Factors – mORF) to diverse T cell stimulation contexts. We pursued two main objectives: 1. To screen for TFs that either drive or prevent the TEX state from forming. 2. To screen for TFs that could reverse the TEX state. To discover and characterize novel TF regulators that drive or prevent T cell differentiation to TEX, we screened with a lentiviral mORF library in chronically (inducing TEX formation) or acutely (inducing TEFF formation) TCR-stimulated cytotoxic CD8+ T cells. Given the central role of the TF TOX in driving and maintaining the TEX state in CD8+ T cells, we used TOX protein expression as a readout by isolating high versus low TOX expressing T cells through fluorescence activated cell sorting (FACS). Comparing screen results in the context of acute versus chronic stimulation facilitated identification of factors specific to the TEX state. Over 100 TF ORFs were enriched in TOX-high and TOX-low bins, including both known and novel regulators of TOX. To comprehensively look at the transcriptomic and epigenetic effects top TFs had on acutely or chronically stimulated T cells, we sub-pooled on TFs that emerged as hits uniquely regulating TOX during chronic and screened in chronic versus acute stimulation setting with a final Phorbol myristate acetate and ionomycin (PMA/IO) stimulation using Simultaneous High-throughput ATAC and RNA expression with sequencing (SHARE-seq) as a readout. We generated a high-resolution ~770,000 single-cell dataset of chronically/acutely/PMA/IO stimmed T cells with matched TF perturbation, RNA-seq, and ATAC-seq. TFs were programming T cells in distinct ways, either driving T cells to endogenous or completely synthetic states. Computational analysis revealed KLF2 to be driving repression of the TEX state through direct repression of the TOX locus and enriching in the TEFF chromatin program. Individual validation of KLF2 in the chronic stimulation setting with flow cytometry confirmed a dramatic improvement in effector function via significant upregulation of IFN-γ and Granzyme B, highlighting KLF2 overexpression as a potential strategy to improve T cell function during chronic stimulation. We were able to identify many other TFs that altered TEX formation, but we also sought to identify TFs that could reprogram TEX cells into a more functional state for improved ACT. Tumor-infiltrating lymphocytes (TILs) are a promising type of ACT to treat solid tumors. TILs are manufactured by expanding and reinfusing tumor-reactive T cells from tumor biopsies. Efficacy of TIL therapies has been limited by the heterogeneity of expanded TIL products and the high prevalence of dysfunctional TEX CD8+ T cells. While a subset of CD8+ TILs co-expressing CD103 and CD39 are enriched for tumor-reactive TILs across multiple cancer types, these cells are often in the TEX state with low proliferative potential. To identify regulators of TEX proliferation within TILs, we applied the mORF library to isolated CD8+ and CD8+ CD103+ CD39+ TILs. RELB emerged as the dominant driver of human TIL expansion with a skew towards CD8+ cells. TCR diversity was maintained after multiple days of in vitro expansion driven by RELB. Transcriptome profiling of multiple RELB-expressing TIL subtypes revealed a shift towards a memory/costimulatory-like phenotype. Using a HER2-targeting CAR and tumor co-culture model, RELB conferred improved persistence after multiple tumor challenges in vitro and improved solid tumor control in mouse xenografts in vivo. Finally, co-culture of RELB-overexpressing TILs with patient-matched tumor organoids showed an increase in TIL product polyfunctionality, tumor reactivity, and tumor killing. By applying overexpression of all human TFs to diverse T cell states throughout TEX formation, we have generated multiple datasets that implicate many reprogramming factors that form, prevent, and even reverse TEX state. Our approaches led to the development of the first trimodal (RNA+ATAC+TF perturbation) single-cell dataset in different T cell stimulation settings that could be used as both a resource and as a direction for human T cell reprogramming. We also conducted the first high-throughput genetic perturbation screen in exhausted human TILs and identified a novel role for a known TF regulator of T cell function. Novel strategies to modulate intrinsic drivers of T cell exhaustion need to be developed. Through our work, we have contributed to bolstering the T cell engineering toolset that exists to improve ACT and immunotherapy. With cell state reprogramming, we hope to achieve the level of success ACT has seen in hematological malignancies for solid tumor indications and pave the way for curative therapies for metastatic disease.