Document resource
Background Natural Killer (NK) cell immunotherapies are being implemented into the clinic for treatment of hematologic malignancies and solid tumors; however, our ability to discern which of these immunotherapies will work best is hindered by the rudimentary in vivo systems we currently employ. To address this issue we generated Acute Myeloid Leukemia (AML) and Ovarian Cancer (OC) models using three different mouse strains, with different immune complexity through use of humanization and transgene expression, and evaluated efficacy of our Tri-specific Killer Engager (TriKE) platform in said strains. 1 Methods Mouse models employed in our studies included NSG mice engrafted with enriched human NK cells (xeno-NSG; least complex), CD34 humanized standard NSG mice (huNSG; intermediate complexity), and CD34 humanized NSG-FLT3-IL15 (huNSG-F15; highest complexity). For our AML tumor model, HL60-luciferase cells were injected intravenously while OC model OVCAR8-luciferase cells were injected intraperitoneally. In xeno-NSG models, enriched NK cells were injected 3 (AML) or 4 (OC) days after tumor engraftment. Mice were treated 3x/week with CD33-targeting (AML) or B7H3-targeting (OC) TriKE, and tumor bioluminescence (BLI) was evaluated weekly. Fluids and tissues from different compartments were harvested for a 42 marker CyTOF analysis (Standard Biotools Helios system) and multiplex-immunofluorescence imaging on the phenocycler-fusion system (Akoya Biosciences) at different time points.Results BLI confirmed greatest tumor control in the huNSG-F15 groups by the TriKE ( figure 1A). Evaluation of immune reconstitution (figure 1B) in a TriKE dose escalation study with AML model indicated a robust induction of NK cell expansion in a dose dependent manner, but not CD8 expansion: indicating targeted activity of the IL-15 moiety within the TriKE. Multiplex immunofluorescence imaging of xeno-NSG and huNSG-F15 spleens visually confirmed greater immune complexity and infiltration in the huNSG-F15 group (figure 2). Blood harvested for flow cytometry and CyTOF are being used to examine immune composition across early and late timepoints. Future experiments will utilize harvested tissue to perform further imaging.Conclusions The responses seen in the huNSG-F15 strain, compared to the other strains, indicate that immune complexity is a key factor in evaluating NK cell therapeutics. We postulate that this more complex approach is necessary to better elucidate how patients will respond to NK cell immunotherapies, and gives us a tool to understand why certain immunotherapies succeed or fail. Future work will involve comparing correlates within these strains to ongoing clinical trials in order to fully evaluate the value of these strains.Acknowledgements Mice used in this study were kindly provided by Jackson labs, Bar Harbor, ME. Services provided by the University Flow Cytometry Resource and the Comparative Pathology Shared Resource at the University of Minnesota.Reference Phung, Shee Kwan, et al. ‘Bi-specific and tri-specific NK cell engagers: the new avenue of targeted NK cell immunotherapy.’ Molecular Diagnosis & Therapy. 2021;25:5:577–592. doi:10.1007/s40291-021-00550-6Ethics Approval IACUC # 2207-40255AAbstract 970 Figure 1Evaluation of tumor control and immune composition in different mouse models of AML and OC. A) NSG, huNSG, and huNSG-F15 mice were injected intravenously with 750,000 HL-60lucs (top) or 200,000 OVCAR8lucs (bottom). For NSG mice, 750,000 magnetically enriched (CD3/CD19 depleted) NK cells were added 3 (AML) or 4 (OC) days after tumor engraftment. At that time treatment with 50 μg cam16-15-33 (AML) or 50 μg cam16-15-camB7H3 (OC) Trikes were dosed IP 3x/week through the duration of the study. Day 1 was considered the start of dosing. BLIs were obtained using an IVIS imager every 7 days. Negative controls did not receive tumor. B) 100 μl of blood was obtained at day 14 post start of TriKE treatment in HL60luc engrafted mice, and staned and evaluated for presence of human NK cell (huCD45+ CD56+ CD3-) or T cells (huCD45+ CD56- CD3+) by flow cytometryAbstract 970 Figure 2Evaluation of immune infiltration indicates greater complexity within the huF15 model when compared to the NSG model. Spleens of NSG and huNSG-15 at day 14 were fixed and processed for histology. Tissues were mounted onto Superfrost (Fisher scientific) slides and stained with a 30 -plex antigen panel according to the Akoya formalin-fixed paraffin embedded sample protocol. Images were generated off the Phenocycler-fusion multiplexing microscope system and processed for analysis in the HALO image analysis platform (Indica Labs)