Evaluation of a Bifunctional Sirpα-CD123 Fusion Antibody for the Elimination of Acute Myeloid Leukemia Stem Cells

Background: Despite considerable advances in the development of novel strategies for the treatment of acute myeloid leukemia (AML) the relapse rate is still high with only limited treatment options. Relapse occurs due to the persistence of chemotherapy-resistant leukemic stem cells (LSCs), which re-initiate outgrowth of the disease, highlighting the need of targeting LSCs to improve overall survival. Immunotherapies represent a promising strategy to target chemotherapy-resistant LSCs in AML. LSCs are characterized by the expression of the interleukin-3 receptor α, also known as CD123. CD123 is expressed on AML blasts and LSCs, and shows only a moderate expression on normal hematopoietic stem cells, claiming CD123 as a suitable target antigen (Haubner et al, Leukemia 2019). CD47, known as a marker of self, is also highly expressed on LSCs as immune escape mechanism. CD47 transmits a "don't eat me" signal upon its interaction with the myeloid-specific signal regulatory protein alpha (SIRPα) receptor on macrophages, thus inhibiting phagocytosis. In order to efficiently eliminate LSCs and provide AML patients a possibility for prolonged relapse-free survival, we have designed a bifunctional antibody that specifically targets CD123 and simultaneously blocks CD47. Importantly, our strategy restricts the benefits of the CD47 blockade to CD123 positive AML cells. Thus, we hypothesize a lower risk for on-target off-leukemia toxicity.

Methods: The bifunctional SIRPα-CD123 antibody was generated by fusing the endogenous extracellular domain of SIRPα, which functions as the CD47 blocking domain, to an CD123 antibody CD123. We assessed the selective binding of the bifunctional antibody to CD123⁺CD47⁺ AML-derived cells and the ability to block CD47 on CD123⁺ cells in vitro. Furthermore, the biological activity of the SIRPα-CD123 antibody was examined using the AML-derived cell line MOLM-13, patient-derived xenografted (PDX) AML cells as well as primary cells from patients with newly diagnosed or relapsed AML.

Results: We engrafted the endogenous SIRPα V-like domain to an antibody targeting CD123, which improved the binding of the bifunctional SIRPα-CD123 antibody to AML cells compared to a conventional CD123 antibody (MFI ratio^CD123 = 2.46 0.25 vs MFI ratio^SIRPα-CD123 = 4.44 0.60). The SIRPα-CD123 antibody enhanced the elimination of the AML-derived MOLM-13 cells by antibody-dependent cellular cytotoxicity (EC₅₀^CD123 = 38.5 pM vs EC₅₀^SIRPα-CD123 = 10.1 pM, n = 9). Additionally, the cytotoxicity was confirmed using primary patient-derived AML cells ex vivo. Further, an improved ex vivo cytotoxicity towards AML PDX cells was observed with the SIRPα-CD123 antibody (% lysis at 100 nM: 14.27 5.40 vs 42.94 10.21 for CD123 and SIRPα-CD123 antibodies respectively, n = 3). With regards to the inhibition of CD47 signaling, we were able to show a blockade of CD47 on CD123⁺CD47⁺ positive cells by the SIRPα-CD123 antibody. Correspondingly, a significant increase in phagocytosis of primary patient-derived AML cells mediated by monocyte-derived macrophages was observed in allogenic as well as autologous settings (% phagocytosis, normalized to isotype control and maximum phagocytosis in an autologous setting: 20.11 4.59 vs 90.37 6.22, n = 5 for CD123 and SIRPα-CD123 antibodies, respectively). We were further able to show a preferential binding to MOLM-13 in the presence of a 20-fold excess of red blood cells indicating a potential low on-target off-leukemia toxicity. Taken together, our in vitro data supports the elimination of the CD123⁺CD47⁺ positive AML LSC compartment by a synergistic effect of avidity-dependent binding to CD123 and CD47 and the simultaneous inhibition of the innate immune CD47-SIRPα signaling pathway.

Conclusions: The SIRPα-CD123 is a bifunctional antibody with the potential to deplete CD123⁺CD47⁺ AML LSCs by a dual mode of action mechanism resulting in NK cell dependent cytotoxicity and macrophage-mediated phagocytosis. By combining a high affinity binding to CD123⁺ cells and a low affinity CD47 blockade that is restricted to CD123⁺ cancer cells we effectively minimize the risk for CD47-related on-target off-leukemia toxicity. The results of our in vitro assays using AML cell lines are consistent with the data from PDX and primary AML samples and support further preclinical testing of the SIRPα-CD123 antibody in vivo.