A Novel Dual-Antigen Targeting Approach Enables Off-the-Shelf CAR NK Cells to Effectively Recognize and Eliminate the Heterogenous Population Associated with AML

Abstract

While acute myeloid leukemia (AML) treatment has improved, relapse is still common, even after reduced intensity conditioning allogeneic transplantation. Natural killer (NK) cells are a promising allogeneic cell immunotherapy without the risk of graft-versus-host disease or cytokine release syndrome. We and others have reported that haploidentical NK cells can safely result in 30-50% complete remissions in advanced AML patients when given with high-dose lymphodepleting chemotherapy and exogenous cytokine support. However, while NK cell natural cytotoxicity has a clinical signal in AML, increased specificity and multi-dosing are likely to enhance the efficacy and durability of this treatment strategy. A number of stress ligands, including the MHC-I polypeptide-related sequence A and B (MICA/B), bind to the NK cell activating receptor NKG2D and initiate NK cell killing of transformed and infected cells, and several groups have shown promising results in clinical trials by developing NK cells that carry a NKG2D chimeric antigen receptor (CAR). However, the distal α1 and α2 domains of MICA/B that are recognized by NKG2D can be proteolytically cleaved from the surface of the tumor by metalloproteases, allowing for escape from NK cell-mediated detection and elimination. Wucherpfennig and colleagues have described that the membrane proximal α3 domain remains on the cell surface and may be a potential target for immunotherapy across many cancers, including AML. However, since AML is heterogenous, we hypothesized that clinical success would require dual-targeting and selected CD33 as the ideal complementary targeting approach since CD33 is a validated marker expressed on greater than 80% of AML blasts. To this end, our lab has developed a Tri-specific Killer Engager (TriKE) capable of agonistically ligating the CD16 Fc receptor to CD33 found on AML along with IL-15 co-stimulation to further activate the NK cell response. Here, we demonstrate the efficacy of a multiplexed-gene edited induced pluripotent stem-cell (iPSC) derived NK cell (iNK) product that expresses a high-affinity, non-cleavable version of CD16 (hnCD16), a membrane-bound IL-15 fusion receptor, CD38 knockout to enhance metabolic fitness, and a CAR against the α3 domain of MICA/B (α3 MICA/B) to drive a potent response against AML alone and when combined with anti-CD33 TriKE. In the initial study, the iNK cell backbone (iNK cells without the α3 MICA/B CAR) induced potent activity against the AML cell line HL60, and displayed further enhancement of activity with the addition of anti-CD33 TriKE (GTB-3650), representing combined effects of natural cytotoxicity and antibody-dependent cellular cytotoxicity. To show specificity for α3 MICA/B targeting, the AML cell line THP-1 was stained for the presence of MICA/B by flow cytometry using the 6D4 clone that recognizes the α1/2 domains of MICA/B or the 7C6 clone that uniquely recognizes the α3 domain +/- proteolytic cleavage with trypsin (Fig. 1A-B). Significant expression of the α3 domain was observed on THP-1 cells using the 7C6 antibody, which was highly expressed even after protease treatment. On the other hand, the α1/2 domains of MICA/B were detectable at lower levels using the 6D4 antibody and were undetectable after protease treatment (Fig. 1A-B). To assess a dual targeting approach, THP-1 cells were used as targets in live imaging functional assays under standard conditions (Effector:Target [E:T] 2:1). The iNK cell backbone had modest natural cytotoxicity at 3 hours that was significantly enhanced by the addition of α3 MICA/B CAR. The effect was further improved with more rapid killing kinetics when combined with the anti-CD33 TriKE (Fig. 1C). To evaluate the efficacy of dual-targeting in a more stringent physiologic manner, we mimicked "stress” conditions by minimizing the E:T ratio to 0.25:1. While all effector conditions induced immediate killing in 4 hours at the low E:T ratio, sustained tumor control was only observed with dual-antigen targeting (Fig. 1D). Studies with primary AML targets, +/- preincubation of decitabine and all-trans retinoic acid, known to upregulate NKG2D ligands in AML, are in progress and will be discussed. In summary, dual-targeting strategies using off-the-shelf CAR NK cells targeting α3 MICA/B in combination with antigen-specific TriKE targeting CD33 represent an ideal clinical strategy to enhance efficacy and durability of treatment in advanced AML. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal