Potent Activity of Duocarmyin- and Mmaf-Conjugated FLT3-Directed Antibody-Drug Conjugates Towards Acute Myeloid Leukemia Stem Cells <i>in Vitro</i> and <i>In Vivo</i>

Acute myeloid leukemia (AML) therapy failure due to relapse or refractory disease is likely attributed to the persistence of resting leukemic stem cells (LSCs) in the patient's bone marrow. As those cells do not divide, they have a decreased vulnerability towards standard chemotherapy. There is thus a high need to develop efficient and specific anti-LSC compounds that advance to the clinic. These agents must have the potential to specifically target LSCs and to eradicate them despite their resting and drug-resistant state. These prerequisites can be elegantly met using antibody-drug conjugates (ADCs) with a carefully selected target and payload. Herein, we describe the development of two FLT3-targeting ADCs either utilizing a DNA-damaging or a microtubule-targeting drug as payload and an in-depth investigation of their capability to eradicate LSCs. In payload studies with arrested and proliferating cell lines, we show sustained activity of the DNA-damaging agent duocarmycin (DUBA) in the resting cells, while the efficacy of microtubule-targeting agents was diminished. These results suggested that DUBA could be a promising payload for the construction of an anti-LSC ADC. When conjugated to the humanized FLT3-mAb 20D9h3, both DUBA and monomethyl auristatin F (MMAF) efficiently eradicated FLT3-positive AML cell lines in vitro by proliferation-inhibition and apoptosis induction. While 20D9h3-DUBA arrests cells in G1/S and activates ATR-CHK1 DNA damage repair pathway, 20D9h3-MMAF halts the cell cycle in G2/M, in line with the payload's mechanism of action. In contrast to gemtuzumab ozogamicin, both ADCs do not lose efficacy in cells overexpressing the drug-resistance transporter ATP-dependent translocase (ABCB1). We went on to analyze the potency of both ADCs towards early and very early leukemic progenitors using patient-derived xenograft (PDX) and primary AML cells from three different donors. To this end, we used colony-forming unit (CFU) assays, long-term culture initiating cell (LTC-IC) assays and leukemia-initiating cell (LIC) assays. Overall, 20D9h3-DUBA was highly potent in eradicating colonies and preventing engraftment with effective doses as low as 0.025 µg/ml. Despite not being as effective as 20D9h3-DUBA in colony-assays, 20D9h3-MMAF pre-treatment of AML-393 (0.3 µg/ml) and AML-388 (1 µg/ml) PDX cells was also able to prevent leukemic outgrowth in 4/5 and 5/5 NSG mice, respectively. This is remarkable, as microtubule-targeting agents are generally not regarded as ideal payload choice to eliminate quiescent stem cells. Importantly, healthy CD34-positive bone marrow cells analyzed in the same assays were not harmed by either ADC at the effective doses highlighting the excellent therapeutic window provided by the FLT3 target. The maleimide-technology used in 20D9h3-DUBA is associated with short-comings such as tendency to aggregate, loss of linker-payload due to thiol-exchange reaction and carboxylesterase 1c (CES1c) cleavage specifically in rodents complicating in vivo analyses. In contrast, 20D9h3-MMAF is conjugated by the P5 technology, leading to stable and hydrophilic ADCs. Despite those difficulties, both 20D9h3-MMAF and 20D9h3-DUBA led to strong tumor reductions in AML-388 xenograft mouse models after intravenous treatment with 2 x 3 mg/kg. In summary, we provide evidence that FLT3-targeting ADCs with both DUBA and MMAF as payload sucessfully eradicate AML LSCs. Further advancements will encompass linker improvements of 20D9h3-DUBA to eventually unite the ADC's particularly high potency towards AML LSCs with an excellent linker stability.