Abstract Successful transplantation of autologous gene-modified hematopoietic stem and progenitor cells (HSPCs) requires efficient ablation of resident hematopoietic stem cells. Considering that conventional myeloablative conditioning regimens are associated with nonhematologic toxicities, we evaluated CD45-directed radioimmunotherapy (RIT) using the α-emitter astatine 211 (211At) before the transplantation of ex vivo gene-edited autologous HSPCs in nonhuman primates as an alternative. We humanized the CD45 antibody, BC8, and labeled it with 211At. As a model, mobilized CD34+ HSPCs were multiplex gene-edited using an adenine base editor, modifying the HBG promoter to reactivate fetal hemoglobin and deleting CD33. Two animals each received 300 or 400 μCi/kg of 211At. In contrast to historic controls conditioned with total body irradiation, CD45-RIT animals did not show any noticeable nonhematopoietic toxicities and were almost entirely transfusion independent with rapid recovery of neutrophils and platelets. Dose-dependent engraftment of gene-edited cells was enabled by 211At. A new single-cell sequencing assay revealed up to 70% combined monoallelic and biallelic gene-editing efficiency in the blood, consistent with complete replacement of the bone marrow stem cell compartment. Assessment by bulk analysis underestimated the frequency of gene-edited cells, highlighting the importance of a single-cell readout. Single-cell sequencing further confirmed stable and unbiased contribution of multiplex-edited HSPCs to all mature lineages in the blood, providing high-resolution data assuring successful replacement upon autologous HSPC gene therapy. The levels of edited cells remained stable for the entire follow-up of >18 months. Collectively, these studies identify 211At-CD45 RIT as a targeted alternative for myeloablative conditioning for autologous gene therapy.

The selection of genetically engineered immune or hematopoietic cells in vivo after gene editing remains a clinical problem and requires a method to spare on-target toxicity to normal cells. Here, we develop a base editing approach exploiting a naturally occurring CD33 single nucleotide polymorphism leading to removal of full-length CD33 surface expression on edited cells. CD33 editing in human and nonhuman primate hematopoietic stem and progenitor cells protects myeloid progeny from CD33-targeted therapeutics without affecting normal hematopoiesis in vivo, thus demonstrating potential for improved immunotherapies with reduced off-leukemia toxicity. For broader application to gene therapies, we demonstrate highly efficient (>70%) multiplexed adenine base editing of the CD33 and gamma globin genes, resulting in long-term persistence of dual gene-edited cells with HbF reactivation in nonhuman primates. Using the CD33 antibody-drug conjugate Gemtuzumab Ozogamicin, we show resistance of engrafted, multiplex edited human cells in vivo, and a 2-fold enrichment for edited cells in vitro. Together, our results highlight the potential of adenine base editors for improved immune and gene therapies.

Introduction: Pivekimab sunirine (PVEK; IMGN632), a CD123 antibody-drug conjugate carrying an alkylating monoamine indolinobenzodiazepine pseudodimer (IGN), is currently under clinical investigation for acute myeloid leukemia (AML) and other CD123-expressing neoplasms. Here, we studied multiple potential modulators of PVEK's anti-leukemia activity. Methods: We tested potential variables that might modulate the in vitro cytotoxic effects of PVEK against human acute myeloid and lymphoblastic cells, using genetically and functionally well-defined human acute leukemia cell line models and primary leukemia cells from adults with AML. Results: We first sought to determine the quantitative relationship between CD123 expression and PVEK-induced cytotoxicity using 2 human AML cell lines (ML-1, MOLM-13) endogenously expressing CD123 and generated sublines in which CD123 was deleted via CRISPR/Cas9. We then used these CD123KO sublines to lentivirally express human CD123 at different level. Despite very low levels of CD123 expression on parental cells, PVEK induced cell death in a dose dependent manner in both cell lines, as determined flow cytometrically after 3 days of drug exposure. This effect was entirely dependent on the presence of CD123 as indicated by the resistance of the CD123KO subline to PVEK. However, in both engineered ML-1 and MOLM-13 cells, there was a quantitative dependence of PVEK-induced cytotoxicity on the amount of CD123 displayed on both cell lines, demonstrating the importance not only of CD123 expression per se (with minimal amount of CD123 being required and sufficient for the drug's anti-AML activity), but also of the level of CD123 expression for PVEK-induced cytotoxicity. Across a panel of 13 human acute myeloid and lymphoblastic leukemia cell lines expressing varying levels of CD123, sFGN849 and PVEK were broadly active against TP53 wild-type cell lines and exerted dose-dependent cytotoxicity. sFGN849 and PVEK also dose-dependently induced cell death in cell lines with TP53 alteration. However, they were significantly less active against TP53 mutated/deleted than TP53 wild-type acute leukemia cell lines. When accounting for the varying sFGN849 sensitivity across cell lines, there was a statistically significant relationship between PVEK-induced cytotoxicity and CD123 expression (R2=0.4469, P=0.0125). This association remained in both the TP53 wild-type cell lines (R2=0.7654, P=0.0099) and the TP53mutated/deleted cell lines (R2=0.6709, P=0.0461). A similar relationship was found across 9 primary AML cell samples (R2=0.6724, P=0.0068). Since the above data suggested that TP53 alterations impaired the ability of PVEK to induce cytotoxicity in acute myeloid and lymphoblastic leukemia cells, we used TP53wild-type EOL-1, MOLM-13, MV4;11, REH, and RS4;11 cells and deleted TP53 from bulk cells via CRISPR/Cas9 and then exposed these engineered cells to idasanutlin to enrich for the population of TP53KO cells. sFGN849-induced cytotoxicity was significantly lower in TP53KOcells than parental TP53wild-type cells in all 5 cell line pairs, demonstrating the importance of functional TP53 for PVEK-induced cytotoxicity. In contrast, overexpression of P-glycoprotein or breast cancer resistance protein (BCRP) only minimally affected the cytotoxic activity of sFGN849 and PVEK in sublines engineered to overexpress these ABC transporter proteins relative to parental cell counterparts. Finally, to study their impact of anti-apoptotic BCL-2 family proteins on PVEK-induced cytotoxicity, we generated sublines EOL-1, MOLM-13, and MV4;11 cells in which we singly over-expressed either BCL-2, BCL-xL, and found sFGN849 activity was reduced in sublines overexpressing one of these proteins relative to parental cells. Conclusions: PVEK has broad anti-leukemia activity against human acute myeloid and lymphoblastic leukemia cells. Important modulators for the drug's cytotoxic effects we identified include CD123 expression levels, TP53 alterations, and overexpression of anti-apoptotic BCL-2 family proteins but surprisingly not activity of ATP transporter proteins. The clinical relevance of these findings should be explored.

This phase 1 study investigated the addition of gemtuzumab ozogamicin (GO) to intensive chemotherapy with cytarabine, daunorubicin, and midostaurin in 21 patients with newly diagnosed (ND) FMS-like tyrosine kinase 3 (FLT3)-mutated acute myeloid leukemia (AML). Four dose levels of GO were evaluated. The use of GO was tolerable, with all dose-limiting toxicities similar to those seen in standard-of-care treatment. After induction, the median time to platelet recovery was 26 days, and the median time to absolute neutrophil count (ANC) recovery was 27 days. The maximum tolerated dose was cytarabine 100 mg/m2 on days 1 to 7, midostaurin 50 mg twice daily on days 8 to 21, daunorubicin 60 mg/m2 on days 1 to 3, and GO 3 mg/m2 on days 1 and 4. For the 18 patients who were evaluable for response after induction therapy, 16 patients (76%) achieved a composite complete response (complete remission [CR] + CR with incomplete hematologic recovery), and 2 (10%) had stable disease. Of the 14 patients who proceeded to consolidation, 5 discontinued the study for transplant, 1 for disease progression, and 1 for physician discretion. Seven patients completed consolidation therapy, all of whom achieved a CR. In total, 13 of the 21 patients (62%) received a hematopoietic stem cell transplant. Our results show that GO can safely be combined with intensive chemotherapy with midostaurin in ND, FLT3-mutated AML. This trial was registered at www.clinicaltrials.gov as #NCT03900949.

Background: Improved survival of some patients with AML with the antibody-drug conjugate gemtuzumab ozogamicin (GO) validates CD33 as a therapeutic target, but GO is often ineffective. As one limitation, hP67.6 (used in GO) and almost all other CD33 antibodies recognize the membrane-distal V-set domain. This may be problematic given existence of a CD33 variant lacking this domain. Moreover, we have shown that CD33/CD3 bispecific antibodies (BiAbs) and chimeric antigen receptor (CAR)-modified T cells binding CD33 closer to the cell membrane exert greater T-cell mediated cytotoxicity than those binding distally. Here, we investigated whether this principle applies to CD33-targeted therapies harnessing natural killer (NK) cells. Methods: We compared NK cell-mediated cytotoxicity against human leukemia cells expressing full-length CD33 (CD33FL) vs. isogenic cells expressing similar levels of a CD33 molecule lacking the membrane proximal C2-set domain (CD33ΔE3-4), bringing the V-set domain into immediate cell membrane proximity. We studied 3 types of therapeutics: 1) unconjugated V-set-directed CD33 IgG1 antibodies; 2) CD33Vset/CD16a BiAbs; and 3) CD33V-set-directed CAR-NK cells. NK cell-mediated cytotoxicity was determined flow cytometrically in co-culture assays by quantifying cell numbers and proportion of non-viable target leukemia cells. Results: First, we investigated antibody-dependent cell-mediated cytotoxicity (ADCC) of CD33V-set antibodies with NK-92 cells transduced with high-affinity CD16a (NK-92CD16a) and primary human NK cells as effectors. Three different CD33 antibodies (including lintuzumab) elicited dose-dependent cytotoxicity against human leukemia cells overexpressing CD33FL or CD33∆E3-4 but not against cells lacking CD33. Importantly, the CD33V-set antibodies exhibited greater ADCC against cells expressing CD33∆E3-4 than cells expressing CD33FL. We then tested the effect of membrane proximity on the efficacy of CD33Vset/CD16a BiAbs in IgG4-scFv format. Like CD33V-set antibodies, the CD33Vset/CD16a BiAb (using sequences from lintuzumab) induced greater cytotoxicity against cells expressing CD33∆E3-4 than cells expressing CD33FL. In a third series of experiments, we investigated the effect of membrane proximity on the efficacy of human NK cells (KHYG-1 cells) transduced with CD33V-set-directed CARs. Both the lintuzumab- and hP67.6-based CAR-NK cells demonstrated dose-dependent, CD33-specific cytotoxicity. Both CAR-NK cell products exhibited greater cytotoxicity against cells expressing CD33∆E3-4 than cells expressing CD33FL. Consistent with this notion of greater activation upon membrane proximal engagement, CD33V-set CAR-NK cells showed higher intracellular TNFα and IFNγ levels after co-culture with CD33∆E3-4-expressing cells compared to CD33FL-expressing cells. Our observations provided the rationale to explore CD33C2-set-directed therapeutics. We previously generated a panel of murine and human antibodies that bind the C2-set domain regardless of the presence or absence of the V-set domain (“CD33PAN” antibodies). In the presence of NK-92CD16a cells, all 4 CD33PAN antibodies we tested induced CD33-specific ADCC against human AML cell lines. Finally, we generated CD33C2-set-directed CAR-NK cells using the sequences from 5 CD33PAN antibodies. Compared to non-targeting CAR-NK cells, CD33PAN CAR-NK cells induced substantially greater cytotoxicity against AML cell lines as well as primary human AML cells with a range of CD33 molecules expressed on leukemic cells. Conclusions: Our data indicate decreasing the distance between CD33 binding epitope and leukemia cell membrane enhances the efficacy of CD33-directed NK cell therapies. These findings are the first to show that membrane proximity may matter for ADCC efficacy even for a relatively small target antigen like CD33, with a change in epitope position by ~4 nm sufficing to yield differences. Our studies are also the first to show an advantage of targeting CD33 membrane-proximally to enhance CAR-NK efficacy. Together, our findings support the further development of CD33C2-set-directed NK cell-based therapies for AML and other CD33-expressing neoplasms.

Abstract: The treatment of monogenetic disorders, such as hemoglobinopathies and lysosomal storage diseases, has markedly improved with the advent of cell and gene therapies, particularly allogeneic or gene-modified autologous stem cell transplantations. However, therapeutic efficacy is reliant on maintaining engraftment above a critical threshold. To maintain such engraftment levels, we and others have pursued approaches to shield edited cells from antibody or chimeric antigen receptor (CAR) T-cell–mediated selection. Here, we focused on CD33, which is expressed early on hematopoietic stem and progenitor cells (HSPCs) as well as on myeloid progenitors. Rhesus macaques were engrafted with HSPCs edited to ablate CD33 using either CRISPR/CRISPR-associated protein 9 or adenine base editor. Both editing strategies showed similar post-transplant recovery kinetics and yielded equivalent levels of engraftment. We then created a V-set domain–specific CAR construct (CAR33), validated its functionality in vitro, and treated both animals with autologous CAR33 T cells. CAR33 T cells expanded after infusion and caused specific depletion of CD33WT but not CD33null progeny, leading to a transient enrichment for gene-edited cells in the blood. No depletion was seen in the bone marrow stem cell compartment with CD34+CD90+ HSCs expressing lower levels of CD33 in comparison to monocytes. Thus, we show proof of concept and safety of an epitope editing–based enrichment/protection strategy in macaques.

BACKGROUND/OBJECTIVES:Approved for treatment of acute leukemia, gemtuzumab ozogamicin (GO) and inotuzumab ozogamicin (InO) are antibody-drug conjugates (ADCs) that deliver a toxic calicheamicin (CLM) derivative. The resistance mechanisms to GO/InO remain incompletely understood. METHODS:We performed a genome-wide clustered regularly interspaced short palindromic repeat (CRISPR)/Cas9 screen for CLM sensitivity genes, and then performed confirmatory cytotoxicity assays. RESULTS:Several DNA damage pathway regulation genes were identified, most notably TP53. Across 13 acute leukemia cell lines, the six TP53-mutant cell lines (TP53MUT) were indeed 10- to 1000-fold less sensitive to CLM than the seven TP53WT cell lines. In five TP53WT/KO syngeneic cell line pairs we generated, TP53KO cells were significantly less sensitive to CLM than their TP53WT counterparts. In TP53WT but not TP53MUT cells, the MDM2 inhibitor and p53 activator, idasanutlin, enhanced CLM cytotoxicity, demonstrating that decoupling of cells from MDM2-p53 regulation sensitizes leukemia cells to CLM. The ATM inhibitors AZD1390 and lartesertib also significantly enhanced CLM efficacy but did so independent of the TP53 status. In contrast, neither an ATR inhibitor, Chk1/Chk2 inhibitor, Chk2 inhibitor, or a PARP inhibitor significantly impacted CLM-induced cytotoxicity across the thirteen cell lines. Together, our studies identify ATM, MDM2, and TP53-which are in the same cellular response to DNA damage pathway-as key modulators of CLM-induced cytotoxicity in acute leukemia cells. CONCLUSIONS:These results support further evaluation of combination therapies with corresponding small-molecule inhibitors (currently pursued for therapy of other cancers) toward clinical testing as novel strategies to increase the efficacy of CLM-based ADCs such as GO and InO.

Background/Objective: Current treatments for eosinophilic and mast cell disorders are often ineffective. One promising target to improve outcomes is sialic acid-binding immunoglobulin-like lectin-8 (Siglec-8). As limitations, there are few Siglec-8 monoclonal antibodies (mAbs) available to date, and Siglec-8-directed treatments have so far primarily focused on unconjugated mAbs, which may be inadequate, especially against mast cells. Methods: Here, we used transgenic mice to raise a diverse panel of fully human mAbs that either recognize the V-set domain, membrane-distal C2-set domain, or membrane-proximal C2-set domain of full-length Siglec-8 as a basis for novel therapeutics. Results: All mAbs were efficiently internalized into Siglec-8-expressing cells, suggesting their potential to deliver cytotoxic payloads. Tool T cell-engaging bispecific antibodies (BiAbs) and chimeric antigen receptor (CAR)-modified natural killer (NK) cells using single-chain variable fragments from Siglec-8 mAbs showed highly potent cytolytic activity against Siglec-8-positive cells even in cases of very low target antigen abundance, whereas they elicited no cytolytic activity against Siglec-8-negative target cells. Siglec-8V-set-directed T cell-engaging BiAbs and Siglec-8V-set-directed CAR-modified NK cells induced substantially greater cytotoxicity against cells expressing an artificial smaller Siglec-8 variant containing only the V-set domain than cells expressing full-length Siglec-8, consistent with the notion that targeting membrane-proximal epitopes enhances effector functions of Siglec-8 antibody-based therapeutics. Indeed, unconjugated Siglec-8C2-set mAbs, Siglec-8C2-set-directed T cell-engaging BiAbs, and Siglec-8C2-set-directed CAR-modified NK cells showed high antigen-specific cytolytic activity against Siglec-8-positive human cell lines and primary patient eosinophils. Conclusions: Together, these data demonstrate Siglec-8-directed immunotherapies can be highly potent, supporting their further development for eosinophilic and mast cell disorders.

Increasing efforts are focusing on natural killer (NK) cell immunotherapies for AML. Here, we characterized CC-96191, a novel CD33/CD16a/NKG2D immune-modulating TriNKET®. CC-96191 simultaneously binds CD33, NKG2D, and CD16a, with NKG2D and CD16a co-engagement increasing the avidity for, and activation of, NK cells. CC-96191 was broadly active against human leukemia cells in a strictly CD33-dependent manner, with maximal efficacy requiring the co-engagement of CD16a and NKG2D. A frequent CD33 single nucleotide polymorphism, R69G, reduced CC-96191 potency but not maximal activity, likely because of reduced CD33 binding. Similarly, the potency, but not the maximal activity, of CC-96191 was reduced by high concentrations of soluble CD33; in contrast, the soluble form of the NKG2D ligand MICA did not impact activity. In the presence of CD33+ AML cells, CC-96191 activated NK cells but not T cells; while maximum anti-AML efficacy was similar, soluble cytokine levels were 10- to >100-fold lower than with a CD33/CD3 bispecific antibody. While CC-96191-mediated cytolysis was not affected by ABC transporter proteins, it was reduced by anti-apoptotic BCL-2 family proteins. Finally, in patient marrow specimens, CC-96191 eliminated AML cells but not normal monocytes, suggesting selectivity of TriNKET-induced cytotoxicity toward neoplastic cells. Together, these findings support the clinical exploration of CC-96191 as in NCT04789655.

Current CD33-targeted immunotherapies typically recognize the membrane-distal V-set domain of CD33. Here, we show that decreasing the distance between T cell and leukemia cell membrane increases the efficacy of CD33 chimeric antigen receptor (CAR) T cells. We therefore generated and optimized second-generation CAR constructs containing single-chain variable fragments from antibodies raised against the membrane-proximal C2-set domain, which bind CD33 regardless of whether the V-set domain is present (CD33PAN antibodies). CD33PAN CAR T cells resulted in efficient tumor clearance and improved survival of immunodeficient mice bearing human AML cell xenografts and, in an AML model with limited CD33 expression, forced escape of CD33negleukemia. Compared to CD33V-set CAR T cells, CD33PAN CAR T cells showed greater in vitro and in vivo efficacy against several human AML cell lines with differing levels of CD33 without increased expression of exhaustion markers. CD33PAN moieties were detected at a higher frequency on human leukemic stem cells, and CD33PAN CAR T cells had greater in vitro efficacy against primary human AML cells. Together, our studies demonstrate improved efficacy with CAR T cells binding CD33 close to the cell membrane, providing the rationale to investigate CD33PAN CAR T cells further toward possible clinical application.

Background: Ex vivo gene-modified autologous cell products are increasingly explored to cure monogenic disorders (e.g. sickle cell disease, thalassemia). Efficient removal of bone marrow (BM)-resident hematopoietic stem/progenitor cells (HSPCs) is a requirement for successful engraftment of such cell products. To accomplish this, various conditioning regimens have been utilized, each with their own unique profile of target specificity and on- and off-target toxicities. Currently most widely used is busulfan but unwanted non-hematologic toxicities have provided the impetus to develop more specific approaches that spare normal tissues better. Monoclonal antibodies (mAbs) conjugated with either small molecule toxins or radionuclides to selectively target phenotypically distinct cells have gained attention for this purpose. Of particular interest as target is CD45, a glycoprotein expressed at very high copy number exclusively on almost all blood cells. CD45-targeted radioimmunotherapy (RIT) with the beta-emitter, iodine-131, has proven effective to augment conditioning before allogeneic hematopoietic cell transplantation (HCT). Here, we evaluated CD45-directed RIT using a the highly potent alpha emitter, astatine-211 (211At) as payload as sole conditioning agent before autologous HCT of ex vivo gene-edited HSPCs in a nonhuman primate (NHP) HCT and gene therapy model. Methods: A humanized version of the human/NHP cross-reactive CD45 mAb, BC8 (HuBC8), was conjugated with isothiocyanatophenethyl-ureido-closo-decaborate(2-) (HuBC8-B10), a boron cage molecule to enable subsequent labeling with 211At as done in our early phase clinical trials. Simultaneously, NHP CD34+ cells were mobilized with G-CSF/AMD3100, collected via leukapheresis, and cryopreserved after being gene modified ex vivo using adenine base editors to reactivate fetal hemoglobin (HBG) production as well as to delete CD33. 72 hours after administration of a single dose of 211At-labeled HuBC8-B10, gene-edited CD34+ cells were thawed and infused. Animals received 300 µCi/kg (n=2) or 400 µCi/kg (n=1) of 211At with 0.5 mg/kg of HuBC8-B10. All animals were monitored for toxicities, and blood count recovery as well as engraftment of gene-modified HSPCs and blood lineages which were assessed serially using flow cytometry and next generation sequencing. Results: The editing efficiency of CD33 and HBG in the infusion product ranged from 50-77% and 15-38%, respectively, with no measurable impact on cell viability or erythro-myeloid differentiation potential of edited cells in colony-forming cell assays. A total of 2-4x106 CD34+ cells/kg were infused into the animals and rapid recovery of neutrophils and platelets seen in between 6-8 and 10-13 days, respectively. Transient weight loss over the first 20-30 days was seen (n=3) and gastric ulcer treatment needed (n=1). An 211At dose of 300 µCi/kg led to incomplete myeloablation with neutrophils remaining above 400/µl, whereas full myeloablation was seen at a dose of 400 µCi/kg. Ablation of monocytes, lymphocytes, and platelets was seen in all animals, whereas CD45-negative erythrocytes were spared and the hemoglobin remaining stable throughout the study. Two animals were entirely transfusion-independent and the third animal receive a single platelet transfusion. Dose-dependent engraftment of gene-editing in the peripheral blood was seen with 20-40% CD33-negative cells and 5-10% HBF reactivation in the two animals receiving 211At at 300 µCi/kg as compared to 80% of CD33-negative cells and 20% HBF reactivation in the single animal receiving 211At at 400 µCi/kg. Full recovery of the BM stem cell compartment was confirmed at 3-month post-transplant by flow cytometry. Conclusion: CD45-targeted alpha emitter-based RIT with 211At enables stable engraftment of ex vivo gene-edited autologous stem cell products. Our studies identify 211At-CD45 RIT as a targeted alternative for myeloablative conditioning followed by autologous transplantation of gene-modified HSPCs. 211At-HuBC8-B10 is well tolerated with only minimal adverse reactions observed in NHPs. While 211At-CD45 RIT demonstrates selective depletion of CD45-positive white blood cells as well as HSPCs in the BM, CD45-negative erythrocytes are spared, likely contributing to the minimal supportive care needs needed following autografting.

Background: CD45-targeted radioimmunotherapy (RIT) has long been explored as augmentation of conditioning before hematopoietic cell transplantation (HCT). The CD45 monoclonal antibody (mAb) most exploited for this purpose is BC8. Validating this approach, BC8 labeled with iodine-131 (131I-apamistamab [Iomab-B]) followed by allogeneic HCT was recently shown to improve outcomes of older adults with relapsed/refractory AML relative to conventional care. However, as a murine mAb, BC8 has important limitations, including substantial infusion toxicities and development of human anti-mouse mAbs, which preclude BC8 redosing and future use of any other murine mab. To overcome these limitations, we humanized BC8 and tested its anti-tumor properties in vivo using the potent alpha-emitter, astatine-211 (211At), as payload. Methods and Results: BC8 was humanized by grafting its complementarity-determining regions (CDRs) into the human variable domain germline heavy and light chain sequences with highest homology to BC8's sequences. Two variants were generated, one containing murine CDRs only and the other additionally including 2 murine residues within the non-CDR human light chain variable region and 4 murine residues within the non-CDR human heavy chain variable region. In competitive ELISA assays, CDR-grafted BC8 with murine back mutations (“HuBC8”) showed only minimally reduced binding to human CD45 compared to chimeric BC8 (ChiBC8) or murine BC8, and flow cytometrically determined binding of HuBC8 to CD45+ leukemia cell lines was similar to that of ChiBC8. For comparative in vivo testing, BC8, ChiBC8, and HuBC8 were conjugated with isothiocyanatophenethyl-ureido-closo-decaborate(2-) (B10-NCS), a boron cage molecule used for subsequent labeling with 211At as done in our early phase clinical trials with 211At-based RIT. For assessment of CD45+ cell targeting, NOD-Rag1null IL2rɣnull/J (NRG) mice were injected with CD45+ MOLM-13 cells in the flank to generate human AML cell flank tumors, followed by a single infusion of BC8 or HuBC8 labeled with 10 µCi of 211At. Tissues were harvested at 24 hours for analysis on a gamma counter and demonstrated similar tumor cell accumulation of radiolabeled BC8 and HuBC8. In vivo efficacy studies with ChiBC8, HuBC8, and non-binding isotype control mAbs (13R4; all human IgG4 frameworks) were performed by injection of 0.2x106 luciferase-transduced MOLM-13 or ML-1 cells into tail veins of NRG mice to generate disseminated human AML. 2 days later, groups of 8 mice were treated with radiolabeled mAbs (40 µCi 211At/animal); one group did not receive any mAb. In both leukemia models, ChiBC8 and HuBC8 extended the survival of treated mice relative to 13R4 (P<0.0001), without significant difference between ChiBC8 and HuBC8 (for MOLM-13: median survival 39.5 days [13R4]) vs. 60 days [ChiBC8] vs. 59 days [HuBC8]; for ML-1: 73 days vs. 138 days vs. 114 days), demonstrating potent in vivo anti-tumor efficacy of 211At-NCS-HuBC8 RIT. Since Fc engineering to minimize Fc receptor interactions has been shown to improve RIT and its therapeutic ratio (i.e. tumor-to-normal cell targeting), we then compared ChiBC8/IgG1 with ChiBC8 using frameworks with reduced Fc binding properties (IgG4, IgG4PAA, IgG2m4, and IgG2σ). There were no significant differences in CD45+ cell targeting in NRG mice bearing MOLM-13 flank tumors. However, there were substantial differences in anti-leukemia efficacies between different antibody frameworks, with ChiBC8/IgG1 labeled with 20 µCi 211At leading to longest survival of NRG mice whereas 211At-NCS-ChiBC8 with IgG4 ProAlaAla, IgG2m4, or IgG2σ did not extend survival of mice beyond what was accomplished with 211At-NCS-13R4 (P<0.0001; median survival 28.5 days [13R4/IgG1]) vs. not reached [ChiBC8/IgG1] vs. 95.5 days [ChiBC8/IgG4] vs. 36 days [ChiBC8/IgG2m4] vs. 27.5 days [ChiBC8/IgG4PAA] vs. 25 days [ChiBC8/IgG2σ]). Increasing the specific activity further increased the anti-leukemia efficacy of 211At-NCS-HuBC8. Conclusions: 211At-NCS-HuBC8 shows similar in vivo CD45+ cell targeting properties and anti-leukemia efficacy as murine 211At-NCS-BC8. Our studies identify the antibody framework (with greatest efficacy so far observed with IgG1) and specific activity as critical factors for the efficacy of 211At-NCS-HuBC8. Together, these study support further development of HuBC8 as antibody for possible clinical RIT applications.

Current immunotherapeutic targets are often shared between neoplastic and normal hematopoietic stem and progenitor cells (HSPCs), leading to unwanted on-target, off-tumor toxicities. Deletion or modification of such targets to protect normal HSPCs is, therefore, of great interest. Although HSPC modifications commonly aim to mimic naturally occurring phenotypes, the long-term persistence and safety of gene-edited cells need to be evaluated. Here, we deleted the V-set domain of CD33, the immune-dominant domain targeted by most anti-CD33 antibodies used to treat CD33-positive malignancies, including acute myeloid leukemia, in the HSPCs of two rhesus macaques, performed autologous transplantation after myeloablative conditioning, and followed the animals for up to 3 years. CD33-edited HSPCs engrafted without any delay in recovery of neutrophils, the primary cell type expressing CD33. No impact on the blood composition, reconstitution of the bone marrow stem cell compartment, or myeloid differentiation potential was observed. Up to 20% long-term gene editing in HSPCs and blood cell lineages was seen with robust loss of CD33 detection on myeloid lineages. In conclusion, deletion of the V-set domain of CD33 on HSPCs, progenitors, and myeloid lineages did not show any adverse effects on their homing and engraftment potential or the differentiation and functionality of myeloid progenitors and lineages.

On-target toxicity to normal cells is a major safety concern with targeted immune and gene therapies. Here, we developed a base editing (BE) approach exploiting a naturally occurring CD33 single nucleotide polymorphism leading to removal of full-length CD33 surface expression on edited cells. CD33 editing in human and nonhuman primate (NHP) hematopoietic stem and progenitor cells (HSPCs) protects from CD33-targeted therapeutics without affecting normal hematopoiesis in vivo , thus demonstrating potential for novel immunotherapies with reduced off-leukemia toxicity. For broader applications to gene therapies, we demonstrated highly efficient (>70%) multiplexed adenine base editing of the CD33 and gamma globin genes, resulting in long-term persistence of dual gene-edited cells with HbF reactivation in NHPs. In vitro , dual gene-edited cells could be enriched via treatment with the CD33 antibody-drug conjugate, gemtuzumab ozogamicin (GO). Together, our results highlight the potential of adenine base editors for improved immune and gene therapies. Graphical abstract:

Click to increase image sizeClick to decrease image size Disclosure statementAll authors declare no competing conflict of interest.Data availability statementFor original data and reagents, please contact the corresponding author (rwalter@fredhutch.org).Additional informationFundingResearch reported in this publication was supported by the National Institutes of Health/National Cancer Institute (NIH/NCI; R01-CA172582).

Supplemental Figures 1-3, Tables 1-3. Supplemental Figure 1. Highly variable expression of MMRN1 in pediatric AML patient specimens. Supplemental Figure 2. Clinical outcome in patients with high and low MMRN1 expression. Supplemental Figure 3. Relationship between bone marrow blast percentage and MMRN1 expression in the AAML0531 patient cohort. SUPPLEMENTAL TABLE 1. Comparison of Baseline Characteristics of Patients with Low (<0.5) vs. High (≥0.5) MMRN1 Expression in AAML03P1. SUPPLEMENTAL TABLE 2. Univariate and Multivariate Regression Models of OS, EFS, and RR for the AAML0531 Study, with Outcome Censored at Time of HCT. SUPPLEMENTAL TABLE 3. Relative MMRN1 expression in paired samples of CD34+/CD33- and CD34+/CD33+ cells.

Radioimmunotherapy (RIT) has long been pursued to improve outcomes in acute leukemia and higher-risk myelodysplastic syndrome (MDS). Of increasing interest are alpha-particle-emitting radionuclides such as astatine-211 ( 211 At) as they deliver large amounts of radiation over just a few cell diameters, enabling efficient and selective target cell kill. Here, we developed 211 At-based RIT targeting CD123, an antigen widely displayed on acute leukemia and MDS cells including underlying neoplastic stem cells. We generated and characterized new murine monoclonal antibodies (mAbs) specific for human CD123 and selected four, all of which were internalized by CD123+ target cells, for further characterization. All mAbs could be conjugated to a boron cage, isothiocyanatophenethyl-ureido- closo -decaborate(2-) (B10), and labeled with 211 At. CD123+ cell targeting studies in immunodeficient mice demonstrated specific uptake of 211 At-labeled anti-CD123 mAbs in human CD123+ MOLM-13 cell tumors in the flank. In mice injected intravenously with MOLM-13 cells or a CD123 NULL MOLM-13 subline, a single dose of up to 40 µCi of 211 At delivered via anti-CD123 mAb decreased tumor burdens and substantially prolonged survival dose dependently in mice bearing CD123+ but not CD123– leukemia xenografts, demonstrating potent and target-specific in vivo anti-leukemia efficacy. These data support the further development of 211 At-CD123 RIT toward clinical application.