CD28 CAR T Cells for the Treatment of T Cell Malignancies

Introduction: While synthetic immunotherapy has been introduced into the standard of care treatment of patients with B-lineage malignancies, patients with T-lineage malignancies have not gained benefit from this novel approach, yet. The main reason is the challenging choice of target antigen in T-lineage malignancies such as pediatric T cell acute lymphoblastic leukemia (T-ALL). As activating mutations of CD28 have already been described in T cell malignancies, we set out to quantify surface expression of CD28 and other costimulatory molecules on primary pediatric T-ALL samples and explore the utility of this functionally relevant molecule as a novel target antigen.

Methods: Bone marrow (BM) samples of pediatric T-ALL patients (n=54) at time of diagnosis and healthy control individuals (n=14) were collected. We quantified surface expression of CD28 and 25 additional co-stimulatory or co-inhibitory molecules by flow cytometry after gating on T cell precursors (living singlets, CD45^dim/SSC-A^low and CD7⁺). Subsequently, we generated a set of 20 CD28 directed second-generation chimeric antigen receptors (CARs) based on five different monoclonal antibodies with variation in hinge domain and light chain / heavy chain chronology. We analyzed specific CAR activity against CD28 expressing T-ALL cell lines after retroviral transduction into primary human T cells. Additional CD28 knockout (KO) by CRISPR/Cas9 could prevent T cell fratricide. CAR constructs were subjected to in vivo testing in a T-ALL NSG mouse model with transplantation of 7.5e4 CCRF-CEM cells transduced with firefly luciferase on day 0 and transfer of 2.5e6 CD28 KO T cells on day 3 harboring either of both CD28 CAR molecules. We used CD7 CAR T cells as positive control, CD19 CAR T cells and CD28 KO T cells without CAR as negative controls.

Results: We observed significant upregulation of CD28 on T-ALL leukemia when compared to healthy BM donors (Figure 1A, mean 68.8% vs. 3.7%, p=0.0002). We confirmed upregulation of CD28 in published T-ALL RNA-expression data sets. Interestingly, other co-stimulatory molecules such as CD127 were upregulated, too (mean 50.0% vs. 14.6%, p=0.0001), while co-inhibitory molecules such as CD160, TIGIT and TIM-3 were strongly downregulated (>10fold and p<0.0001 each). We hypothesized that this could imply a functional relevance of CD28 in T-ALL. Therefore, we performed co-culture assays of monocyte derived dendritic cells (CDs) expressing CD28 ligands CD80 and CD86 with different T-lineage leukemia cell lines. The presence of DCs significantly increased proliferation of CD28⁺ T-ALL cell lines under stress conditions. In order to test the feasibility of CD28 CAR T cells, we generated CD28 KO primary T cells without CAR. When tested in co-culture assays with leukemia cells and T cell engagers, CD28 KO T cells showed unchanged cytotoxic capacity and target-dependent activation illustrating that CD28 KO T cells retain short-term effector functions. Based on cytotoxic capacity and CAR T cell expansion, we identified two lead CD28 CAR candidates out of a pool of 20 CD28 CARs. After CD28 KO, CD28 CAR-T cells showed in vitro expansion comparable to CD19 CAR T cells. Next, we asked how CD28 CAR T cells compare to CD7 CAR T cells, which are currently evaluated clinical trials for T-ALL therapy. Therefore, we generated CD7 KO T cells that were subsequently transduced with a functional CD7 CAR. In vitro, both CD28 CAR T cells and CD7 CAR T cells showed killing of CD28⁺CD7⁺ CCRF-CEM cells of >90% at an effector:target (E:T) ratio of 0.2:1, with CD7 being outperformed by one of the CD28 CARs while outperforming the other CD28 CAR at an E:T ratio of 0.04:1 (Figure 1B). Finally, we went on to validate the functionality of CD28 CAR T cells in vivo. We observed significantly prolonged survival of mice treated with CD28 CAR T cells and CD7 CAR T cells when compared to CD19 CAR T cells or control T cells. No difference in survival between CD7 and CD28 CAR T cell treated mice could be observed in two independent experiments.

CONCLUSION: We identify CD28 as novel target antigen for pediatric T-ALL and provide evidence that CD28 is an immunotarget with functional relevance for T-ALL. We demonstrate the feasibility of CD28 CAR T cell generation and that these novel CAR T cells perform equally well as CD7 CAR T cells both in vitro and in vivo. Novel and functionally relevant CAR targets will facilitate clinical development of immunotherapy for T-lineage malignancies.