Loss of TET2 Uncouples Proliferative and Effector Functions in CAR T Cells

Chimeric antigen receptor (CAR) T cells have opened a new paradigm for the treatment of leukemia and lymphoma. Their production, however, is laborious, requiring tens of millions of CAR T cells per infusion. This constraint could be significantly alleviated if safe and more efficacious T cells could be generated. In a patient with chronic lymphocytic leukemia, treated with anti-CD19 CAR T cells, a recent report described the emergence of a single T cell clone that at its expansion peak accounted for 94% of circulating CAR T cells, coinciding with the development of cytokine release syndrome and tumor regression (Fraietta et. al. Nature 2018). Insertional mutagenesis in this T cell had disrupted an allele of TET2, an epigenetic regulator mediating the oxidation of 5-methylcytosine. The other allele appeared to bear an inherited hypomorphic variant, resulting in the near complete loss of TET2 function in this clone. To understand the mechanisms accounting for this chance clinical finding, we investigated the effect of TET2 loss in human T cells engineered to express different chimeric receptors.

Using CRISPR/Cas9, we edited TET2 in T cells engineered to express a CD19-specific second-generation CAR encompassing the costimulatory domain of either CD28 or 4-1BB (Rv-1928z and Rv-19BBz). TET2 disruption enhanced the in vivo anti-tumor activity of Rv-19BBz but not Rv-1928z CAR T cells tested under stress test conditions using limiting CAR T cell doses (as previously described in a human B cell acute lymphoblastic leukemia (B-ALL) NALM6 model, Zhao et. al. Cancer Cell 2015). Since Rv-1928z induces potent effector differentiation but limited persistence compared to Rv-19BBz, we hypothesized that loss of TET2 could amplify the expansion and persistence of 4-1BB-costimulated T cells but not override the differentiation program imparted by Rv-1928z.

To test this hypothesis, we utilized two orthogonal approaches known to limit exhaustion and increase persistence of CD28-costimulated CAR T cells, Rv-1928z co-expressed with 4-1BB ligand (Rv-1928z-41BBL) and 1928z driven by the TRAC promoter (TRAC-1928z). Disruption of TET2 enhanced the anti-tumor efficacy of both these CAR T cells and promoted acquisition of a central memory phenotype. However, over time (50-200 days), TET2-edited TRAC-1928z and Rv-1928z-41BBL attained a hyper-proliferative phenotype ultimately requiring euthanasia due to splenomegaly and extensive CAR T cell accumulation in various organs. Post-mortem analysis found no evidence of NALM6 in these mice. This was in contrast to stress test studies with Rv-1928z and Rv-19BBz where most mice succumbed to NALM6 progression. These observations established an essential role for CAR signaling in determining the phenotypic outcome of TET2 loss in T cells.

To examine the long-term effects of TET2 disruption in the context of all 4 receptors, we treated human B-ALL bearing mice with curative doses of all 4 CAR T cells and followed them for up to 200 days. We found that all 4 CAR expressing TET2-edited T cells could eventually attain a hyper-proliferative phenotype, but with varying frequency depending on the CAR design (Rv-1928z-41BBL and TRAC-1928z > Rv-19BBz > Rv-1928z). To assess their effector function, NALM6-bearing mice were infused with adoptively transferred hyper-proliferative TET2-edited CAR T cells. Strikingly, these T cells were unable to elicit any tumor control, despite their maintaining a central memory phenotype as assessed by flow cytometry. This loss of effector function was observed for all 4 CAR T cell types, suggesting a discrepancy between function and flow cytometric phenotype.

Transcriptional, methylation and genome accessibility studies revealed a unique T cell state wherein the proliferative program is uncoupled from effector response. We identified a unique transcriptional and epigenetic signature that is manifested in a loss of effector function while maintaining robust proliferation. This state stands in contrast to the classically described T cell exhaustion state where loss of effector function is preceded by loss of proliferative ability. TET2 disruption thus promotes a CAR T cell proliferative program that depends on the CAR design but does not in itself enhance anti-tumor activity.