TET2, tumor control, and CAR T cell hyperproliferation

Abstract

A recent study by Jain et al. published in Nature followed up on evidence suggesting that depletion of 5-methylcytosine dioxygenase TET2 in chimeric antigen receptor CAR T cells could enhance their expansion, persistence, and antitumor efficacy. Their findings are cautionary, but offer hope of a path forward.

CAR T cell therapy has revolutionized the treatment of specific cancers such as leukemias and lymphomas. However, the antitumor efficacy of CAR T cells is often hampered by their limited expansion and poor persistence in patients. Understanding the mechanisms that regulate T cell biology in tumors will help in designing more effective CAR T cells for therapeutic use.

A previous case report [1], in which a patient with chronic lymphocytic leukemia (CLL) was successfully treated with CD19-targeted CAR T cells, found that tumor clearance correlated with delayed expansion of a single CAR T cell clone (Figure 1). Tumor clearance was followed by contraction of the dominant clone and of total CAR T cells, resembling the resolution of a classical CD8⁺ T cell effector response. Genetic analysis of the dominant, expanded clone established that the CD19-CAR viral vector had integrated into one allele of the TET methylcytosine dioxygenase 2 (TET2) gene in a way that produced truncated transcripts coding for nonfunctional proteins. TET2 encodes one of three TET methylcytosine dioxygenases that regulate chromatin structure enzymatically by oxidizing 5-methylcytosine to 5-hydroxymethylcytosine, 5-formylcytosine, and 5-carboxylcytosine, thereby promoting DNA demethylation, and also nonenzymatically by recruiting other chromatin-modifying proteins [2]. The second TET2 allele of the expanded clone carried a preexisting point mutation, E1879Q at the protein level, that diminished, but did not abolish, TET2 catalytic activity. The analysis did not rule out the possibility that the observed clonal expansion, tumor clearance, and clonal contraction might depend on an undetected alteration in another cellular pathway that cooperated with decreased TET2 function. Nevertheless, these observations raised the tantalizing possibility that TET2 deletion might promote vigorous CAR T cell expansion and robust antitumor immunity without triggering uncontrolled CAR T cell growth. Indeed, loss of Tet2 function has been shown to increase the effector function of CD8⁺ tumor-infiltrating lymphocytes at early times in the tumor and to enhance antitumor immunity [3].

Figure 1. The discussed instances of human CAR T cells with impaired or no TET2 function.

(Upper panel) Timeline of treatment and response in CLL Patient-10 [1]. A first infusion of CD19-specific CAR T cells elicited partial regression of the leukemia, followed by renewed tumor growth. A second infusion initially stabilized tumor size without causing regression. Several weeks after the second infusion, a single CAR T cell clone with diminished TET2 function expanded rapidly, cleared the tumor, and then contracted in numbers leaving behind memory CAR T cells. The precise level of residual TET2 enzymatic activity in the expanded clone was not determined, and whether TET2 nonenzymatic functions were impaired is unknown. (Lower panel) Timeline of treatment, response, and late CAR T cell hyperproliferation in a mouse model that tested TET2-deficient human CAR T cells [4]. Human NALM6 leukemia cells were implanted into an immunodeficient NSG mouse, followed by adoptive transfer of CD19-specific human CAR T cells. In the condition depicted, the T cell TET2 locus had been subjected to CRISPR–Cas9 editing that rendered a substantial fraction of the T cells TET2-deficient. CAR T cells at a sufficient dose rapidly cleared the tumor. However, many weeks later, some CAR T cell clones deficient in both TET2 alleles proliferated extensively, effectively replacing the original leukemia with a CAR T cell-based disease. Abbreviations: CAR T cells, chimeric antigen receptor T cells; CLL, chronic lymphocytic leukemia; NALM6, human B cell acute lymphocytic leukemia cell line; NSG mouse, NOD/SCID/IL2Rγ^null mouse.

In a recent study published in Nature, Jain et al. [4] tested this possibility directly. The authors implanted human NALM6 acute lymphocytic leukemia cells into immunodeficient NOD/SCID/IL2Rγ^null (NSG) mice, followed by transfer of TET2-edited (sgTET2) or TET2 wild-type CD19-targeted human CAR T cells (Figure 1). TET2 deletion in T cells expressing both a CD19 CAR (termed 1928z) and 4-1BBL led to improved tumor rejection compared to control CAR T cells. The earliest deviation in tumor growth for sgTET2 cells compared to controls was at around day 16 after T cell adoptive transfer. Several weeks after complete tumor clearance, the mice suffered splenomegaly and infiltration of the liver and kidneys by oligoclonal CAR T cells. These hyperproliferative TET2-edited CAR T cells were less effective than the fresh CAR T cell product against NALM6 cells in vitro and were ineffective in controlling NALM6 tumors in vivo when administered at a dose that would be curative using fresh CAR T cells. Late hyperproliferation of TET2-deleted T cells expressing the CD19 1928z CAR alone, without 4-1BBL, was observed at lower frequency, although TET2 deletion had no early benefit in antitumor activity in this case.

The observations seemed to validate a concern, extrapolated from hematopoietic stem/progenitor cells (HSPCs), that rare TET2-deleted CAR T cell clones might exhibit pathological proliferation. It is well established that TET2 loss-of-function mutations occurring in HSPCs predispose to clonal hematopoiesis and the subsequent development of myeloid and lymphoid malignancies, even though the mutations are not oncogenic by themselves [5]. However, it remained unclear whether engineering TET2 loss-of-function mutations in fully developed peripheral blood T cells would carry a liability of T cell lymphoma.

Why did the CAR T cells in the study by Jain et al. continue to proliferate in the absence of antigen? It does not appear that the cells had acquired a cell-intrinsic commitment to uncontrolled proliferation. The hyperproliferative cells had no obvious cancer driver mutations and failed to survive when transferred at 2 × 10⁶ cells/mouse into secondary NSG recipient mice, unless the mice additionally received human IL7 and human IL15. Given the much smaller numbers of CAR T cells present in the secondarily grafted mice than in the donor mice, and the different context, a likely scenario is that hyperproliferative CAR T cells in the donor mice were sustained by a cell-extrinsic signal either from a subset of the CAR T cells themselves or from a conducive environment in the ‘cured’ donor mouse.

The hyperproliferative phenotype was linked to elevated levels of BATF3 and MYC in the CAR T cells, echoing findings for some human T cell and B cell lymphomas where lymphocyte proliferation depended on upstream factors that drove BATF3 expression and on cooperating cellular signaling pathways [6–8]. Similar adjunct contributors are probably also necessary in the hyperproliferative TET2-deficient T cells, because BATF3 overexpression by itself does not turn peripheral T cells into T cell lymphomas [9]. The failure of hyperproliferative CAR T cells to engraft upon secondary transfer in the study by Jain et al. also weighs against a cell-intrinsic cancerous state, because either BATF3 expression was not maintained in the new signaling environment of the recipient mouse – continued BATF3 expression was not tested – or BATF3 expression was maintained and did not support proliferation of the transferred cells.

Two aspects of the new study may be worth exploring further. The authors [4] infused a CAR T cell product harboring a vast number of cells deficient in both TET2 alleles, whereas eradication of CLL in the previous case report [1] started with a single cell that still retained some TET2 catalytic function. There is thus still a case for testing CAR T cells with reduced, but not abolished, TET2 function. A key question is whether extremely low numbers of such cells in a CAR T cell infusion would replicate the success of the dominant expanded clone in the CLL patient. If only a single cell – as in the CLL patient – or a few cells are required, then an extremely low statistical risk posed by rare transformation to hyperproliferation might be tolerable when weighed against the other risks inherent to cancer and CAR T cell therapy. Another aspect specific to the Jain et al. CAR T cell product is that it contained CD4⁺ CAR T cells that would proliferate in the tumor. In principle, oligoclonally expanded CD4⁺ CAR T cells might have provided an extrinsic signal for the proliferation of some CD8⁺ CAR T cell clones, a precedent being that TET2-deficient clonally expanded B cells aid the survival and proliferation of TET2-deficient angioimmunoblastic T cell lymphomas [10,11]. It is unlikely that a TET2-impaired CD4⁺ clone was accidentally produced in the CLL patient. In this scenario, a CD8⁺ T cell-only CAR T cell product might avoid the liability of uncontrolled proliferation.

Importantly, improved tumor clearance by TET2-deficient cells reflected different mechanisms in the two studies (Figure 1) – enhanced early effector activity at a time of only modest CAR T cell expansion in the humanized mouse study [4], contrasted with the delayed massive expansion of a single clone possessing adequate antitumor function in the CLL patient [1]. It could be beneficial to harness either mechanism in CAR T cells, but the challenge will be to accomplish that without provoking excessive CAR T cell proliferation.