Combinatorial Antigen Targeting Strategy for Acute Myeloid Leukemia

Introduction: Efforts to safely and effectively treat acute myeloid leukemia (AML) by targeting a single leukemia associated antigen with chimeric antigen receptor T (CAR T) cells have had limited success. We determined whether combinatorial expression of chimeric antigen receptors directed to two different AML associated antigens would augment tumor eradication and prevent relapse in targets with heterogeneous expression of myeloid antigens.

Methods: We generated CD123 and CD33 targeting CARs; each containing a 4-1BBz or CD28z endodomain. We analyzed the anti-tumor activity of T cells expressing each CAR alone or in co-transduction with a CLL-1 CAR with CD28z endodomain and CD8 hinge previously optimized for use in our open CAR-T cell trial for AML (NCT04219163). We analyzed CAR-T cell phenotype, expansion and transduction efficacy by flow cytometry and assessed function by in vitro and in vivo activity against AML cell lines expressing high, intermediate or low levels of the target antigens (Molm 13= CD123 high, CD33 high, CLL-1 intermediate, KG1a= CD123 low, CD33 low, CLL-1 low and HL60= CD123 low, CD33 intermediate, CLL-1 intermediate/high) For in vivo studies we used NOD.SCID IL-2Rg^-/-3/GM/SF (NSGS) mice with established leukemia, determining antitumor activity by bioluminescence imaging.

Results: We obtained high levels of gene transfer and expression with both single (CD33.4-1BBʓ, CD123.4-1BBʓ, CD33.CD28ʓ, CD123.CD28ʓ, CLL-1 CAR) and double transduction CD33/CD123.4-1BBʓ or CD33/CD123.CD28ʓ) although single-transductants had marginally higher total CAR expression of 70%-80% versus 60-70% after co-transduction. Constructs containing CD28 co-stimulatory domain exhibited rapid expansion with elevated peak levels compared to 41BB co-stim domain irrespective of the CAR specificity. (p<0.001) (Fig 1a). In 72h co-culture assays, we found consistently improved anti-tumor activity by CAR Ts expressing CLL-1 in combination either with CD33 or with CD123 compared to T cells expressing CLL-1 CAR alone. The benefit of dual expression was most evident when the target cell line expressed low levels of one or both target antigens (e.g. KG1a) (Fig 1b) (P<0.001). No antigen escape was detected in residual tumor. Mechanistically, dual expression was associated with higher pCD3ʓ levels compared to single CAR T cells on exposure to any given tumor (Fig 1c). Increased pCD3ʓ levels were in turn associated with augmented CAR-T degranulation (assessed by CD107a expression) in both CD4 and CD8 T cell populations and with increased TNFα and IFNɣ production (p<0.001 Fig 1d). In vivo, combinatorial targeting with CD123/CD33.CD28ʓ and CLL-1 CAR T cells improved tumor control and animal survival in lines (KG1a, MOLM13 and HL60) expressing diverse levels of the target antigens (Fig 2).

Conclusion: Combinatorial targeting of T cells with CD33 or CD123.CD28z CARs and CLL-1-CAR improves CAR T cell activation associated with superior recruitment/phosphorylation of CD3ʓ, producing enhanced effector function and tumor control. The events that lead to increased pCD3ʓ after antigen engagement in the dual transduced cells may in part be due to an overall increase in CAR expression but may also reflect superior CAR recruitment after antigen engagement. We are now comparing the formation, structure, and stability of immune synapses in single and dual targeting CARs for AML.