Acute Myeloid Leukemia Blasts Research Articles

Lactate Utilization Provides a Metabolic Escape to Resist the Antileukemic Activity of BET Inhibition in Acute Myeloid Leukemia

Acute myeloid leukemia (AML) is a molecularly heterogeneous malignant disease of the bone marrow resulting from arrest of hematopoietic precursors and uncontrolled proliferation of myeloblasts. Inhibition of the BET-family co-activator BRD4 (BETi) shows encouraging pre-clinical activity but limited clinical activity as a single agent. We previously demonstrated that MYC suppression by BETi induces metabolic and mitochondrial changes in AML blasts leading to a collapse in mitochondrial respiration. AML blasts that are resistant to BETi undergo comparable mitochondrial respiration despite impaired glycolysis and glutaminolysis. The metabolic pathway allowing BETi-resistant AML blasts to maintain metabolic integrity is unclear. Therefore, to identify how AML blast cells metabolically bypass BET inhibition, we used two different cell lines: BETi-resistant MOLM-13 and BETi-susceptible MV-4-11 cells. AML cell lines were treated with BETi for 48 hr and global untargeted metabolic mass spectrometry revealed that glycolysis, glutaminolysis, and fatty acid metabolism were not leveraged to maintain tricarboxylic acid (TCA) cycle activity (Fig. 1A). Rather, an increased NADH/NAD+ ratio and intracellular lactate accumulation suggested that lactate was being utilized as an alternative carbon source to sustain mitochondrial respiration. In support, MOLM-13 cells treated with BETi accumulated higher levels of mitochondrial lactate dehydrogenase, which preferentially converts lactate into pyruvate to fuel the TCA cycle. The heightened capacity of MOLM-13 cells to utilize lactate better maintained cellular viability but failed to support cellular replication, suggesting that lactate utilization promotes cellular senescence during BET inhibition. Therefore, we employed three chemical inhibitors of lactate utilization to couple with BETi: AZD3965 to prevent extracellular lactate uptake, UK5099 to block lactate import into the mitochondria, and oxamate to broadly impair lactate dehydrogenase activity. All three chemical inhibitors of lactate utilization were sufficient to render MOLM-13 cells susceptible to BET inhibition, as mitochondrial respiration, ATP production, and cellular viability were decreased to levels comparable to MV-4-11 cells. A similar correlation was observed in patient samples where AML blasts with increased resistance to BETi were predisposed to utilize lactate as an alternative carbon source and rendered susceptible to BET inhibition with lactate utilization inhibitors. Furthermore, these results were translated in vivo as low-dose combination of BETi and oxamate significantly reduced burdens of BETi-resistant cells in murine models (Fig. 1B). These results demonstrate that some AML blasts metabolically adapt to overcome BET inhibition by consuming lactate but that combinations of BETi and lactate utilization inhibitors can stymie this resistance. Recognizing the metabolic switching to utilize lactate in AML therapy resistance, and studying lactate utilization inhibitors in clinical settings, especially in BETi-based therapies should be considered. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

A Phase 1, Open-Label, Safety, Tolerability, and Efficacy Study of NC525 in Subjects with Advanced Myeloid Neoplasms

Background With the exception of certain select subtypes, Acute Myeloid Leukemia (AML) remains incurable for a majority of patients with current therapies. Despite recent drug approvals targeting actionable molecular mutations and BCL-2, relapse remains inevitable for most patients with AML due to leukemia-initiating cells and progenitors {i.e., leukemic stem cells (LSCs)} which are resistant to conventional cytotoxic drugs. Novel agents that preferentially target LSCs to lessen or eliminate the risk for relapse while avoiding significant myelosuppression by sparing normal hematopoietic stem cells (HSCs), are highly sought-after anti-leukemia therapeutics. Leukocyte Associated Immunoglobulin-like Receptor 1 (LAIR-1) is an immune inhibitory receptor that is present on most immune cell subsets, including T cells, NK cells, monocytes, macrophages, dendritic cells, and granulocytes. LAIR-1, a surface leukocyte receptor containing two immune receptor tyrosine-based inhibitory motifs (ITIM), binds to endogenous collagen ligands and plays an important role in inducing an immunosuppressive microenvironment. In AML, the engagement of LAIR-1 inhibits intracellular downstream survival signals, leading to decreased GM-CSF-mediated survival and proliferation of AML blasts. LAIR-1 is expressed on AML blasts and LSCs regardless of FAB subtypes, cytogenetics or molecular mutations. Recent work demonstrates that myeloid leukemia cell lines, primary AML blasts, monoblasts and monocytes, and MDS-EB-2 blasts all express LAIR-1. Critically, while high expression of LAIR-1 is characteristic of LSCs, this receptor is minimally detectable on normal HSCs, making it an ideal anti-leukemic target. NC525 is a humanized immunoglobulin gamma 1, kappa (IgG1κ) monoclonal antibody that targets LAIR-1-expressing AML blasts and LSCs. Extensive IND-enabling preclinical studies have shown the following features of NC525: (i) Inhibition of colony formation of AML CD34+ bone marrow cells; (ii) sparing of colony formation and differentiation ability of healthy CD34+ HSCs; (iii) direct ex vivo killing of AML blasts without killing healthy leukocytes; (iv) suppression of AML in both MV4-11 and THP-1 CDX models; (v) restriction of MV4-11 in a humanized murine model, without depletion or suppression of human T cells; (vi) restriction of MV4-11 tumor growth in murine blood, spleen, and bone marrow; and (vii) reduction of AML disease burden in primary patient-derived AML xenograft (PDX) models. Finally, NC525 completely inhibits leukemia propagation in secondary transplant models, suggesting that NC525 is capable of eradicating LSCs. Furthermore, NC525 destroys leukemic cells by ligation-mediated apoptosis of leukemic blasts and stem cells without evidence of an immunomodulatory effect on other immune subsets. Methods This is an open-label, non-randomized, Phase 1 dose escalation study to determine the safety and tolerability of NC525 in adult patients with advanced myeloid neoplasms. This study will also assess the clinical benefit of NC525. Eligible patients must have relapsed or refractory AML, myelodysplastic syndrome (MDS), or chronic myelomonocytic leukemia (CMML) that has progressed after standard therapy. Participants will receive NC525 by intravenous infusion on Days 1 and 15 of each 28-day cycle. During Phase 1, the safety and tolerability of NC525 will be evaluated by exploring escalating dose levels of NC525. In addition, the pharmacologically active dose (PAD), defined as the dose that provides a maximal biologic effect, such as achieving a complete remission (CR), CR with partial hematological recovery (CRh) or CR with incomplete hematological recovery (CRi), and the maximum tolerated dose (MTD) of NC525 will be determined. Safety and tolerability will be further assessed through the expansion of a few dose levels to determine an optimal recommended Phase 2 Dose (RP2D) and administration schedule of NC525, based on the PK/PD profiles. Taken together, preclinical data suggest that targeting LAIR-1 in AML and other myeloid malignancies may be an effective therapy for these diseases with the advantage of specifically eradicating leukemic stem cells and sparing normal HSCs.

Functional Drug Screen Identifies Thiostrepton, NAMPT Inhibitors and Metformin As Potential Candidates to Target the Macrophage Tumor Microenvironment in Acute Myeloid Leukemia

While it is increasingly becoming clear that cancer cells reside in symbiosis with diverse stromal and immune cells, the tumor supportive microenvironment (TSM) in acute myeloid leukemias (AML) remains poorly understood. We and others previously identified that the presence of tumor-supportive macrophages (often with an M2-like phenotype) was associated with poor prognosis and a more aggressive course of disease in vivo using various AML models. Therefore, tools to target the tumor supportive microenvironment, or to convert an M2-polarized TSM into a tumor suppressive one, are of great interest. Here, we investigated the effects of therapeutic drugs on AML blasts and their TSM specifically focusing on the macrophage compartment. Drug screens using 20 candidate compounds including metabolic inhibitors (targeting glycolysis and mitochondrial respiration) were performed on human AML cell lines (n=16), primary AML samples (n=52), and CD34+ cells from healthy cord blood donors (n=5). Experiments were performed in co-culture using MS5 stromal cell as a supportive feeder. Effects on cell viability and mitochondrial membrane potential (MMP) were evaluated by flow cytometry on blasts (CD45+CD34+) as well as on M1 or M2 macrophages (CD80+ or CD163+CD206+ within the CD45+CD34- fraction). Effects on glycolysis and oxygen consumption were evaluated by Seahorse. Human primary monocytes isolated from healthy donors were also used for co-culture experiments and M1 and M2 macrophages were generated as previously described (Weinhäuser et al, 2022). We identified that conventional drugs such as cytarabine and venetoclax had significant cytotoxic effects on AML blasts, albeit with heterogeneous efficacies dependent on AML subtype, while M2 macrophages remained largely unaffected. Functionally, M2 populations displayed higher MMP levels compared to AML blasts, presumably contributing to their increased drug resistance. Among all the evaluated compounds, only thiostrepton (an antibiotic that also inhibits FOXM1), KPT-9274, and metformin displayed cytotoxic effects on both the AML blasts and M2 macrophages, associated with reductions of MMP levels in both compartments. Various studies have aimed to identify compounds that would repolarize tumor supportive M2 macrophages into tumor suppressive M1 macrophages, including thiostrepton. Using healthy M2 macrophages, we could not detect consistent signs of re-polarization (evaluated by flow cytometry and gene expression profile) upon thiostrepton treatment. In fact, none of the 20 tested compounds was able to efficiently repolarize M2 into M1 macrophages. Furthermore, when we evaluated the effects of thiostrepton pretreated M2 macrophages on short-term and long-term AML cultures, we could not detect a significant reduction in AML cell viability or proliferation as observed for M1 co-cultures. However, we did notice a pronounced cytotoxic effect on the M2 macrophages themselves when a higher dose of thiostrepton (>5 µM) was administered. Metabolically, thiostrepton-treated macrophages also displayed decreased mitochondrial respiration. Next, we determined the ED50 of a large panel of AML cell lines (n=16), which ranged from 1.91-38.67 µM suggesting that tumor cells themselves display considerable sensitivity to thiostrepton. Cell line specific ED50 values were correlated to their transcriptomes and gene set enrichment studies revealed that thiostrepton resistance was associated with the term "ACTIVATION OF GENE EXPRESSION BY SREBF and SREBP", suggesting that fatty acid metabolism can be targeted to enhance thiostrepton efficiency. Surprisingly, the expression of its target FOXM1 did not correlate with the cytotoxic effects induced by thiostrepton in AML cells but was down-regulated in M2-macrophages upon treatment. Thiostrepton-treated healthy CD34+ cells remained largely unaffected in terms of cell viability and differentiation capacity, suggesting that thiostrepton may have a favorable therapeutic window. In conclusion, our screening setup allows for the identification of compounds that can target both leukemic blasts as well as their tumor supportive microenvironment, and thiostrepton, KPT-9274 and metformin emerged as potentially promising candidates. We hope that (combinatorial) targeted drug treatments that also take the tumor microenvironment into account can help to improve AML treatment and prevent relapse. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Sting Agonists Activate Bone Marrow Macrophages and Synergise with Anti-CD47 Blockade Improving Survival in Animal Models of Acute Myeloid Leukemia

The bone marrow (BM) microenvironment regulates acute myeloid leukemia (AML) initiation, proliferation, and chemotherapy resistance. Previous research by our group has identified activation of the stimulator of interferon genes (STING) pathway within BM macrophages supresses leukemic growth. This contrasts with solid tumours where infiltrating macrophages release immunosuppressive signals, allowing cancer progression. AML blasts also regulate the action of macrophages by activating the 'don't eat me’ signal via interactions between CD47 and SIRPα on macrophages. Recent studies have led to clinical trials investigating the potential of blocking CD47 in patients with AML. Here we hypothesise and investigate whether activation of the STING pathway in macrophages of the AML tumor microenvironment will enhance the anti-leukemic effects of anti-CD47 antibody in disease models of AML. Initially we found that activation of STING pathway, using DMXAA or cGAMP, in bone marrow derived macrophages (BMDM) induced clearance (via phagocytosis) of MN1 and HOXA9/Meis1 cells (two murine AML models) from the BM of animals with AML. To determine if the STING agonist activated the STING pathway we performed in vitro and in vivo experiments. In vitro BMDM treated with DMXAA or cGAMP induced STING gene expression signature (upregulation of GBP, IFIT3 and IRF3). For in vivo assessment of STING activation in macrophages, C57Bl/6 animals were treated with DMXAA for 24h, followed by isolation by FACS of BM macrophages (GR1-, F4/80+/ CD115int). RNA extracted and qPCR showed upregulation of STING gene expression signature. Next using two established AML in vivo models (HOXA9/Meis1 and MN1) we investigated the effect of STING agonist on AML progression. Results show that treatment with STING agonist significantly improved survival of mice with AML. Moreover, the numbers and activity of cytotoxic T cells was not changed between treatment groups. Next, we explored the role of anti-CD47 blockade in combination with STING agonist. In vitro CD47 blockade synergised with DMXAA to significantly increase phagocytosis of AML by macrophages. In vivo we found that mice with AML receiving the combined treatments (DMXAA and anti-CD47) demonstrated significantly prolonged survival, over animals given either treatment alone. In summary, we report that BM macrophages activated via the STING pathway supresses AML expansion through phagocytosis and not T cell mediated cytotoxicity. A process which is enhanced by anti-CD47 blocked on AML blasts. The results reported here provide a molecular mechanistic rationale for the clinical evaluation of targeting STING and CD47 in patients with AML.

Retrospective Lineage Tracing of Pediatric Acute Myeloid Leukemia Using Single-Cell Whole Genome Sequencing

Background Acute myeloid leukemia (AML) accounts for 15-20% of all pediatric acute leukemias and is a very heterogeneous disease, both genetically and phenotypically. Although survival has increased gradually during the last decades, relapse-rates remain around 20-30%, and children who relapse have a relatively poor outcome. Still, the exact cause of pediatric AML (pAML) remains unclear. Previously, we found an increased clonal mutation burden in leukemic blasts of a subset of pAML cases, including the ones characterized by RUNX1-RUNX1T1 translocations. Here, we study the mutation accumulation and clonal evolution of pAML to elucidate the timing of driver accumulation and mutational processes underlying leukemogenesis in children. Methods All pAML bone marrow samples were acquired via the biobank of the Princess Máxima Center for pediatric oncology after ethical approval was obtained. We used primary template-directed amplification (PTA) to accurately analyze the complete genomes of single cells at nucleotide resolution. For PTA-mediated whole genome sequencing (WGS), we sorted single leukemic blasts, hematopoietic stem cells (HSCs), multipotent progenitor cells (MPPs) and several differentiated cell types. Subsequently, WGS data was used to study mutation accumulation and to perform phylogenetic lineage tracing of different cell types of the same patient. Results Thus far, we have analyzed the single-cell WGS data of 18 cells from two AML samples characterized by NUP98-NSD1 and RUNX1-RUNX1T1 fusions. We identified multiple cells that immunophenotypically resembled HSCs, MPPs and more mature cell types from different lineages, that shared the fusion gene with the AML blasts (Figure 1). Moreover, a subset of these cells shared one or more additional driver mutations with the immunophenotypically defined bulk leukemic blasts. Subsequent mutation analysis showed that cells with the oncogenic fusion had an increased somatic mutation burden compared to cells without the fusion. Accordingly, we studied mutational signatures to identify the processes underlying this mutation burden. Interestingly, besides the expected mutational signatures, we found that the increased mutation burden was mainly characterized by SBS18, a signature that is associated with damage induced by oxidative stress. This process was active throughout the leukemic lineage, whereas unrelated HSCs and MPPs of the same patients did not display this signature. This finding indicates that the process was specific for the leukemic lineage and therefore likely cell intrinsic. Next, we used driver and passenger mutations to construct phylogenetic lineage trees and found that the accumulation of drivers occurred stepwise. We show that the oncogenic fusion was the first driving event in different patients, followed by accumulation of additional driver mutations. For example, in one patient, based on the clock-like mutation burdens in the separate branches, we could estimate that the first drivers were already present when the patient was between 7.5 to 9 years old (Figure 1, branch A), while this patient was diagnosed at 15.2 years. We have previously identified similar patterns in multiple MLL-rearranged therapy-related AML (t-AML) patients. Both in the t-AML and pAML cases, some leukemic blasts did not have any additional genetic drivers compared to the phenotypical stem and progenitor cells. Conclusions We used single-cell WGS to identify pre-leukemic cells in bone marrow samples of pAML patients, which shared both the oncogenic rearrangement and additional driver mutations with the AML blasts. These pre-leukemic cells showed an increased mutation load, caused by distinct mutational processes. In addition, these cells were immunophenotypically and functionally stem-cell like, contributing to hematopoiesis. Our analyses show that, despite their aberrations, these cells still have differentiation potential. Interestingly, the first driver aberrations of pAML originated years before diagnosis. These data suggest that genetic driving aberrations may not be the only underlying mechanism of AML development in children. Noncoding drivers, epigenetic changes or environmental influences may play an important role in the final steps towards leukemogenesis. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Longitudinal Multimodal Atlas of AML Surface Antigens Enables Adaptive and Combinatorial Targeting

Acute myeloid leukemia (AML) is a clinically heterogeneous disease characterized by chromosomal abnormalities, gene mutations, and inter- and intra-patient genetic and phenotypic variability. Therapies specifically targeting cell surface antigens are promising treatments for AML, but surface antigen expression variability and potential antigen escape mechanisms may impact efficacy of single-targeting therapies. Delineating surface antigen expression uniformity and density between patients and during disease progression from diagnosis to relapse will enable rational prioritization of AML antigens for adaptive and combinatorial targeting. Here, we elucidate surface antigen expression patterns for candidate therapeutic targets by performing multimodal integrative analysis of AML surface proteins and transcriptomes of matched diagnosis and relapse clinically annotated samples. To identify potential therapeutic AML targets, we compiled a comprehensive list of protein surface targets reported in clinicaltrials.gov, literature, and large proteomics mass spectrometry datasets of AML patient bone marrow samples (deBoer et. al. Cancer Cell 2018; Jayavelu et al, Cancer Cell 2022). To focus on surface antigens and reduce the risk of targeting antigens that may have off-tumor solid tissue toxicity, we then excluded antigens that lack plasma cell membrane localization or are expressed on healthy non-hematologic tissues. We adapted Cellular Indexing of Transcriptomes and Epitopes by Sequencing (CITE-seq) to generate longitudinal single-cell multimodal readouts of both the transcriptome and targeted surface proteins of matched diagnostic and relapsed samples from 42 adults with AML (Figure 1). Antibody derived tags (ADTs) against selected antigens were used at saturating conditions. Single cell libraries were prepared with 10X Genomics 5' Feature Barcoding Library Kit v2 and sequenced on Illumina NextSeq 2000. Cellranger v6.0.0 and Seurat v3 were used for mapping, read quantification, and analysis, respectively. We leveraged AML cell lines with known expression of benchmark antigens to assess the accuracy of the ADT readout by correlating it with antibody bound per cell (ABC) values derived from the traditional flow cytometry-based QuantiBRITE assay. Patient bone marrow data were projected onto low dimensional space using UMAP to visualize the population structure of the data. We observed separation of AML blasts from healthy cell lineages and delineation of blast gating based on CD45 expression. k-nearest neighbors clustering identified phenotypically diverse clusters within AML blasts within each patient sample, enabling quantification of specific antigen-positive clusters and antigen co-expression patterns. We found strong correlation between ADT readout and corresponding ABC values for cell surface protein expression on AML blasts of our benchmark antigens (CD33, CD123, CLL1, and ADGRE2). We then proceeded to systematically rank 87 AML surface antigens in blast populations based on ADT readout. Moreover, transcriptome profiling allowed investigation of antigens with limited antibody availabilities, exploration into novel cell states, and identification of signaling pathways that may associate with the presence of particular surface antigens. Identification and analysis of leukemic stem cells are ongoing. Healthy bone marrow and peripheral blood will be analyzed alongside AML samples to elucidate antigen target expression during normal hematopoiesis. In summary, our multimodal analysis of AML surface antigens informs cell surface clonal heterogeneity of patient AML and builds a reference atlas of expression patterns and intensities of potential surface AML antigen targets. In ongoing work, we are analyzing correlations between clonal surface antigen expression in longitudinal analysis of diagnosis and relapse samples as well as correlation with clinical parameters. This comprehensive surface antigen expression profiling will inform selection of therapeutically targetable and consistently expressed antigens to address AML heterogeneity in future immunotherapy targeting studies. Note: J.E. and M.U. share co-first authorship. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

M2 Macrophages Drive Resistance to Phagocytosis and Improve Mitochondrial Metabolism in Acute Myeloid Leukemia Facilitating Leukemic Transformation and In Vivo Engraftment

Acute myeloid leukemia (AML) is a heterogeneous disease, which also extends to its tumor microenvironment (TME). Using single cell RNA-seq data (GSE116256), we performed differential gene expression analysis on macrophages from AML patients (AML-derived macrophages, AdMs) compared to healthy bone marrow macrophages. Thus, 306 upregulated genes in AdMs were identified as involved in different biological pathways such as proliferation, inflammation and cellular metabolism. Down-regulated genes (232) were mainly associated with the regulation of apoptosis. Single cell gene set enrichment analysis associated AdMs with an M2-like phenotype and an immunosuppression transcriptional program. We then investigated the macrophage landscape in a large cohort of AML patients by flow cytometry (n = 61) and immunohistochemistry, which revealed a large heterogeneity whereby AML patients with a predominant M2-like TME showed poor prognosis. Next, we evaluated the immune function of AdMs compared to healthy donor-derived macrophages. Thus, we showed that AdMs displayed decreased phagocytic capacity towards AML cells compared to healthy macrophages. Since AdMs might directly descend from the mutated leukemic clones we evaluated whether these mutations could affect the immune function of macrophages. Cord blood-derived CD34+ cells were lentivirally transduced with FLT3-ITD, differentiated towards macrophages, and these FLT3-ITD+ macrophages displayed decreased phagocytic activity. We questioned whether endogenous murine macrophages that are still present in immunodeficient mouse strains that are often used for human studies might form a barrier that needs to be taken to achieve efficient engraftment of human cells. Macrophages were isolated from NSG mice and were used in phagocytosis assays. The percentage of phagocytosed primary AML blasts by murine macrophages ranged between 3.2-51.4%, similar to phagocytosis levels observed when using human macrophages for the same AML samples. While performing these phagocytosis assays, we noted that the cells that remained and were not phagocytosed had an altered appearance and were very viable, as if these cells had gained fitness. We tested this further by exposing leukemic blasts to M2 macrophages for 48 h, after which the remaining cells were analyzed functionally. We observed that the 48 h "training" of primary leukemic blasts significantly enhanced their mitochondrial metabolism (in part mediated via mitochondrial transfer from macrophages to leukemic blasts), enhanced their in vitro and in vivo homing capacity and that they became resistant to phagocytosis. As a most striking example we used primary APL cells that are notoriously difficult to engraft in NSG mice, which we hypothesized could be in line with their high intrinsic susceptibility to macrophage-mediated phagocytosis. APL cells were exposed to M2 macrophages for 48h resulting in two cellular fates: 91.7% of cells were phagocytosed, but the remaining 8.3% APL blasts were able to induce fatal leukemia when transplanted into NSGS mice in all evaluated cases (n = 6), while prior to training these cells did not engraft long-term and were not able to confer disease. We also tested the tumor supportive capacity on M2 macrophages by performing intrafemoral injection in NSGS mice after which non-trained APL blasts were injected intravenously. Again, efficient engraftment and fatal leukemia development were observed, but only in mice co-injected with M2 macrophages. Strategies to target the supportive AdM subpopulation would be of interest. In our scRNA-seq analysis, we identified the metabolic enzyme Nicotinamide phosphoribosyltransferase (NAMPT) to be upregulated in AdMs. Ex vivo studies revealed that while conventional chemotherapeutics, AraC/venetoclax, efficiently target the AML blasts (CD34+/CD117+) with minimum effects on the AdMs, NAMPT inhibition with KPT-9274 was able to efficiently target both the AML blasts and the AdM fractions. Functionally, treatment of M2 macrophages with KPT-9274 shifted their polarization towards an M1 phenotype, with reduced support to AML cells in co-culture assays. In conclusion, we uncover the heterogeneity in the macrophage landscape in AML patients, show the functional relevance of M2-polarized macrophages for leukemic transformation, and provide alternative approaches for targeting aimed at the tumor-supportive microenvironment.

The Role of Quiescence in Acute Myeloid Leukemia Growth

Acute myeloid leukemia (AML) is the most common leukemia in adults and its prognosis is usually poor. Despite the introduction of targeted treatments in recent years, AML remains a high-mortality disease with a 5-year survival of less than 30%. Emerging evidence suggest that therapy failure and leukemia relapse are due to the intra-tumoral genomic and biological heterogeneity. At biological level, AML is hierarchically organized with leukemia stem cells (LSCs) at the apex. LSCs are a rare cell sub-population with self-renewal capacity, responsible for leukemia initiation and maintenance. Standard AML therapies have limited effects on LSCs, mainly due to their property of exiting the cell cycle and enter quiescence. Preliminary data obtained in our group have shown that different leukemia-initiating oncogenes (NPMc+, PML-RARα and MLL-AF9) share the property of enforcing quiescence in pre-leukemic hematopoietic stem cells (HSCs), and that this is critical for the progression and maintenance of the leukemia clone. Underlying molecular mechanisms, however, are unknown. To this end, we have performed an in vivo reverse genetic screening to identify quiescence-related genes indispensable for leukemia growth. Among the identified hits, Socs2, Stat1 and Sytl4 silencing prevented AML outgrowth in vivo. Notably, Socs2 and Stat1 interference increased proliferation, prevented accumulation of quiescent blasts and induced apoptosis. Interestingly, Socs2 and Stat1 silencing exerted no effects on in vitro proliferation, cell cycle distribution or survival of AML blasts, suggesting that their effects are largely dependent on the in vivo leukemia context. Single-cell RNAseq (scRNAseq) analysis of Socs2-interfered blasts showed marked down-regulation of genes characterizing the dormant status of quiescent HSCs, suggesting that loss of quiescence in Socs2-interfered blasts is linked to the loss of their regenerative potential. Since dormancy and self-renewal potential in HSCs are antagonized by prolonged stress signals, we analyzed scRNAseq data for activation of the integrated stress response (ISR). Strikingly, Socs2-interfered cells showed aberrant activation of ATF4, UPR and autophagic transcriptional programs, which was evident in both proliferating and cell cycle restricted blasts, at variance with control blasts that showed increased ISR expression specifically in proliferating cells. Since elevated levels of UPR may act as danger signals and favor immune-mediated clearance in vivo, we analyzed immune-related signaling pathways. Consistently, Socs2-silenced blasts showed a dramatic down-regulation of specific immune check-point molecules, including CD24a, galectin 9 and VISTA, which are involved in the regulation of B cells, T cells, NK cells and macrophages. Based on these findings, we hypothesized that Socs2 regulates the resolution of stressful signals that eventually accumulate in hyper-proliferating AML blasts by allowing cells to enter quiescence and activate pathways of ISR resolution, including upregulation of immune check-point molecules. If blasts fail to enter quiescence, as in in the absence of Socs2 expression, cells maintain sustained ISR activation and downregulate immune check-point molecules, thus triggering immune-mediated cell death. Consistently, the anti-leukemic effect of Socs2-silencing was partially rescued upon propagation into immunocompromised mice. As well, macrophage depletion prolonged disease latency of Socs2-interfered blasts in immunocompetent mice. These findings provide preliminary evidence of the existence, in AML blasts, of an adaptive response to hyperproliferation that involves ISR activation, induction of quiescence and immune evasion. Targeting this adaptive response, as by Socs2 interference, may activate potent mechanisms of AML immune clearance.

CD25 Targeting Eliminates Regulatory T Cells and CD25+ Blasts in Acute Myeloid Leukemia

Background Acute Myeloid leukemia (AML) is a heterogeneous disease that requires novel targeted treatment options tailored to the patients' specific microenvironment and blast phenotype. CD25 Mab (also referred to as RG6292 and RO7296682) is an afucosylated, IL-2 non-blocking human IgG1 antibody, shown to efficiently deplete immunosuppressive regulatory T cells (Tregs) in humans and solid tumour models (Kolben 2021) whilst allowing binding of IL-2 to effector T cells and the induction of anti-tumour adaptive immune responses (Solomon, Amann et al. 2020). CD25 Mab binds to CD25+ target cells and its crystallisable fragment (Fc) to Fc receptors (FcR) expressed on the surface of effector cells, such as FcgRIIIa on Natural Killer (NK) cells, monocytes and macrophages. It mediates killing of target cells through antibody-dependent cellular cytotoxicity (ADCC) and antibody-dependent cellular phagocytosis (ADCP). CD25 Mab is currently being investigated in a phase I dose escalation monotherapy study (NCT04158583) or in combination with atezolizumab (NCT04642365). In AML patients, a higher frequency of immunosuppressive bone marrow Tregs was observed as compared to that of healthy donors and their abundance correlated with poorer outcome (Szczepanski, Szajnik et al. 2009, Han, Dong et al. 2018, Dong, Han et al. 2020). In addition, CD25 is also expressed on a subset of AML cells and may be restricted to leukemic stem cells (LSC) or cells with a progenitor phenotype (Angelini, Ottone et al. 2015, Bertolini, Kageyama et al. 2018, Aref, Azmy et al. 2020). Considering the critical role of LSC in the propagation of the disease and the high relapse rate in AML, it would be important to understand if CD25 targeting could deepen responses and help prevent relapse in patients. In this study, we hypothesized that CD25 Mab may have a dual mode of action whereby it depletes suppressive Tregs and has a direct cytotoxic effect on CD25+ AML cells. Methods We conducted an in-depth characterization of bone marrow and/or blood samples of 37 AML patients and healthy donors by high dimensional flow cytometry (up to 39 cellular markers) and RNA sequencing using computational analysis. Moreover, we performed ex vivo ADCC assays using allogeneic NK cells isolated from healthy donors and AML patient material to test the cytotoxic potential of CD25 Mab on Tregs and CD25+ AML cells, as compared to isotype control antibody. Results We found that the abundance of Tregs and CD25+ AML cells correlated strongly with that of the blood in 11 patients with time-matched samples, indicating that blood samples could be used for the identification of predictive biomarkers. Moreover, we observed a strong enrichment in the prevalence of CD25 expressing AML cells in patients bearing internal tandem duplication mutations in the FLT3 gene (FLT3-ITD). Furthermore, all four patients treated with a hypomethylating agent in combination with venetoclax displayed detectable levels of CD25+ AML cells. We adopted a patient-centric approach (n=14 patients) to study AML clusters with CD25 expression, due to the high inter-patient heterogeneity. Interestingly, CD25+ clusters most commonly expressed CD34, an LSC marker indicating a preferential expression in immature AML cells. Finally, ex vivo treatment of primary AML patient samples with CD25 Mab led to the specific killing of CD25+ AML cells and Tregs by allogeneic NK cells, as compared to isotype control treatment. Taken together, these results provide a proof of concept of CD25 Mab's dual mode of action in AML patient samples. Conclusions Using high dimensional flow cytometry and computational analysis, we provide a deep characterization of AML patient samples and highlight the heterogeneity in the pattern of expression of common AML targets currently under clinical investigation. We dissected the expression of CD25 on healthy T cells and malignant cells and demonstrated that the blood offers a window into the bone marrow composition and could be used to ascertain Treg and CD25+ AML cells prevalence. Mode of action studies demonstrated that CD25 Mab depletes suppressive Tregs and has a direct cytotoxic effect on the CD25+ AML cells. CD25 targeting represents an attractive target for the treatment of AML, especially in patients where CD25 is expressed on LSC or immature AML cells. Our study warrants further exploration of CD25 Mab as combinatorial treatment with, for instance, FLT3 inhibitors or venetoclax.

Establishing Immunocompetent Leukemia Models to Investigate the Impact of CAR T Cells on the Immune Microenvironment and Bone Marrow Niche

Background Chimeric antigen receptor (CAR) T cells need to overcome limited expansion and persistence within the highly immunosuppressive tumor microenvironment (TME) to successfully eliminate acute myeloid leukemia (AML) in the clinic. CD123 has been identified as a promising target for AML because of its high expression on leukemia stem cells and AML blasts, with a clear increased differential expression when compared to healthy hematopoietic progenitors (HPCs). To date, preclinical evaluation of CD123-CAR T cells have used human xenograft models, which lack a functional immune system. This cross-species evaluation of CAR T cells inadequately addresses interactions between CAR T cells and the immune microenvironment, a limitation of many CAR T cell therapy studies. We have generated immunocompetent murine leukemia models to evaluate murine CD123 (mCD123)-specific CAR T cells and thus investigate the impact of the AML TME on CAR T cell dysfunction. Methods and Results We used biopanning of a murine phage display library on recombinant mCD123 to select a panel of mCD123-specific single chain variable fragments (scFvs). 4 scFvs, encoding 'top binders', were subcloned into a MSGV1 retroviral vector encoding a CD28.z-CAR to generate 4 mCD123-CARs. Second-generation mCD123-CAR T cells were generated by retroviral transduction of activated splenocytes. Only one mCD123-CAR construct was expressed on the cell surface as judged by flow cytometric analysis using F(ab')2 (55.8±11.5% N=7) or recombinant mCD123 (59.0±12.1% N=6) for detection, and T cells expressing this CAR were selected for further analysis. To test the antigen specificity of mCD123-CAR T cells, we expressed full-length mCD123 on 2 murine leukemia cell lines: C1498, a spontaneous myelomonocytic AML cell line, and a Arf-null, BCR-ABL1 preB-ALL. In addition, we analyzed a panel of virally induced NUP98::KDM5A AML generated by retroviral transduction of murine HPCs and determined all samples naturally expressed mCD123 antigen at different densities (range 38-77% N=4). We co-cultured mCD123-CAR in the presence of mCD123+ leukemia cell lines, with mB7-H3-CAR, SP6-CAR, or non-transduced T cells serving as controls. mCD123-CAR T cells recognized CD123+ targets, as evidenced by significantly increased cytokine secretion and cytolytic activity in comparison to controls (N=6, p<0.01). Antitumor activity was maintained in repeat stimulation assays for up to 5 stimulations. To determine in vivo toxicity of mCD123-CAR T cells, non-tumor bearing C57BL/6 mice were given 200 mg/kg cyclophosphamide (Cy) and one effector T cell infusion generated from transgenic luciferase mice (N=5) and monitored by bioluminescence imaging for 10 days. Bone marrow and spleen were harvested 10 days after injection for flow cytometry evaluation and colony formation unit (CFU) readouts. We detected no significant differences in toxicities between mCD123-CAR and control T cells. We next evaluated the anti-leukemia activity of mCD123-CAR T cells. C57BL/6 mice were injected with preB-ALLArf-/CD123+ cells on day 0 and received 100 mg/kg of Cy on day 2. On day 4, they were injected with 10x106 mCD123-CAR T cells. Mice receiving tumor only, tumor + Cy, or SP6-CAR served as controls. mCD123-CAR T cells had significant anti-leukemia activity resulting in a prolonged survival advantage in comparison to controls (N=5 per group, p<0.001). We focused on delineating the TME in our established models and preliminary analysis of comprehensive flow cytometry and single cell transcriptomic data on bone marrow. Our findings revealed that the TME of our established immune competent leukemia CAR T cell therapy models recapitulate human disease, having an increase in dying tumor cells within the bone marrow of mice treated with mCD123-CAR, with a number of other populations (e.g. dendritic cells, regulatory T cells) approximating the tumor-only microenvironment profile. Conclusions We have developed functional mCD123-CAR T cells targeting AML and ALL and have established immunocompetent models to test them. mCD123-CAR T cells demonstrated potent anti-leukemic activity in vitro and have anti-leukemia activity with minimal to no toxicity in vivo. These findings will provide insight into the impact of the AML TME on CAR T cell functionality, and inform future strategies to improve CAR T cells in the immunosuppressive TME.

Senti-202, a Selective, Off-the-Shelf, Preclinical CAR-NK Cell Therapy with CD33 and/or FLT3 Activating CAR, Healthy Cell Protection from Endomucin (EMCN) Inhibitory CAR and Calibrated Release IL-15 for Hematologic Malignancies Including AML

Background: Patients with CD33 and/or FLT3 malignancies, which encompass most myeloid malignancies such as acute myeloid leukemia (AML), have a grim prognosis and a high unmet need. There are no approved cell therapies for patients with AML due to paucity of targets and an immunosuppressive bone marrow milieu. SENTI-202, a preclinical CAR-NK cell is engineered with a logic-gated gene circuit to overcome these challenges. Methods: SENTI-202 is engineered with a FLT3 OR CD33 NOT EMCN logic-gated gene circuit and a proprietary calibrated release IL-15 (crIL-15). These 3 chimeric proteins are engineered onto a natural killer (NK) cell (healthy adult peripheral blood-derived) to create an off-the-shelf chimeric antigen receptor (CAR) NK cell therapy. The bivalent CD33 OR FLT33 (OR GATE) activating CAR (aCAR) enables concurrent targeting of CD33+ and/or FLT3+ AML cells which kills both AML blasts and leukemic stem cells (LSCs) in vitro and in vivo. This concurrent targeting of two tumor antigens could potentially provide longer remission and less chance of relapse. The NOT EMCN (NOT GATE) inhibitory CAR (iCAR) protects healthy EMCN+ hematopoietic stem cells (HSCs) and early hematopoietic progenitor cells (HPCs) from off-tumor toxicity to potentially aid in post-treatment reconstitution of a healthy hematopoietic system. The crIL-15 provides both autocrine and paracrine IL-15 stimulation to the CAR-NK cells (and surrounding immune cells) to promote cell expansion, persistence, and tumor killing. While previous presentations (Garrison et al., 2021) focused on the performance of individual SENTI-202 components in a cell, this presentation focuses on functional data from the complete logic-gated gene circuit and the crIL-15 together in a cell (i.e., the final CAR-NK with all 3 genetic elements). Results: We demonstrated delivery of all circuit components using a single retroviral vector. We then investigated activity of the resulting CAR-NK cells using a traditional in vitro cytotoxicity assay (E:T=1:2; 20hrs) in which SENTI-202 showed >90% killing of leukemia cells (e.g. SEM; P≤0.001), maintained significant serial killing potential (3 rounds; P≤0.0001; P≤0.001; P≤0.001), as well as robust killing against primary AML blasts and LSC-enriched target cell populations. Importantly, SENTI-202 also exhibited significant tumor cell killing and improved survival (P≤0.001) in an MV4-11 xenograft AML mouse model. We investigated the ability of SENTI-202 to reduce off-tumor toxicity toward EMCN+ healthy cells, including primary HSCs. Within in vitro assays, SENTI-202 provided >50% protection to FLT3+ CD33+ model healthy cells that also expressed the EMCN safety antigen, and similar significant protection was maintained through serial protection assays (3 rounds; P≤0.0001; P≤0.001; P≤0.0001). As further evidence of the selectivity imparted using the OR/NOT logic gating approach, in vitro assays with SENTI-202 resulted in killing of leukemia cells but not primary healthy human HSCs and HPCs (~42% protection) (P≤0.01). Finally, in the presence of SENTI-202 in vivo, the NOT GATE protected and enabled significant (P≤0.0001) expansion of EMCN+ FLT3+ CD33+ model healthy cells. Conclusions: In summary, preclinical evaluation of SENTI-202 demonstrated selective killing of CD33 and/or FLT3 expressing AML cell lines while protecting healthy HSCs and HPCs. This selectivity is imparted by the novel logic-gated gene circuit engineered into healthy adult NK cells. Clinical evaluation of SENTI-202 is planned to evaluate the safety and efficacy in patients with hematologic malignancies with high unmet need, including AML.

Modeling Flotetuzumab-Associated CRS in AML Using inVitro and inVivo Preclinical Models

Introduction: Flotetuzumab (FLZ) is a bispecific CD123 x CD3 DART® molecule that is being investigated for the treatment of relapsed and refractory acute myeloid leukemia (AML). FLZ induces artificial immunological synapses between CD3 on cytotoxic T-cells and CD123, which is expressed on AML blasts along with off-target hematopoietic and non-hematopoietic tissues. T-cell activation and expansion requires cytokine release from both innate and adaptive immune cells, which when excessive, can result in cytokine release syndrome (CRS), a potentially life-threatening toxicity that manifests with fever, hypotension, organ dysfunction, respiratory failure, and/or coagulopathy. FLZ-mediated anti-leukemia efficacy is often limited by CRS. Therefore, developing pre-clinical models by which to better understand the mechanisms of CRS induction, including collaborating cell types that produce IL-6 and other inflammatory cytokines, is essential to understanding the biology of CRS and developing mechanistically plausible preventative and treatment strategies. MacroGenics has developed a murine genetic knock-in model that expresses a chimeric murine CD3 epsilon (CD3e) chain bearing the human CD3e epitope recognized by FLZ (hCD3e). These mice are immune competent, maintain normal signaling through the T-cell receptor and expand and secrete cytokines upon binding to FLZ and human CD123. To validate this model, we explore whether T-cells harvested from the spleens of immune competent C57Bl/6 hCD3e mice will bind FLZ and kill human CD123-expressing target cells in vitro. Methods: A murine AML cell line (AML1-GFP FLT3-ITD, DNMT3AR878H) was transduced with human CD123 (hCD123) using a lentiviral vector construct and single cell clones expressing high levels of hCD123 were expanded (AMLGFP-hCD123). Spleens from hCD3eHOM C57Bl/6 mice were harvested for isolation of T-lymphocytes via negative immunomagnetic selection on an AutoMACS machine. In vitro killing assays were set up using purified C57Bl/6 hCD3eHOM T-cells and two clones of AMLGFP-hCD123 cells. The effector to target ratio was 1:1, with 2.5x105 cells per well in a 24-well plate. Flow cytometry/FACS was then performed at 24 hours to assess killing of AMLGFP-hCD123 clones 4 and 14 and T-cell subset number and phenotype. Antibodies to the following antigens were used for FACS: mCD4, hCD3 (epitope SP38), mCD25, mCD8, mCD44, mCD62L, hCD123, mCD3, mCD127, mCD69, and mCD45, in addition to endogenous GFP. Results: At 24 hours, target cell killing of clones 4 and 14 was noted (Figure 1). Compared to the controls treated with FLZ or T-cells alone, targets treated with T-cells + FLZ showed a 2.3- and 2.5-fold decrease in the number of AMLGFP-hCD123 targets in clones 4 and 14, respectively, at 24 hours (Figure 1). In terms of T-cell phenotypic changes, the percentage and MFI expression of activation markers, namely CD25 and CD69, increased for CD4+ and CD8+ subsets in the treatment group compared to controls (data not shown). T-cell subsets were characterized as naïve (CD62L+/CD44-), central memory (CD62L+/CD44+), and effector memory (CD62L-/CD44+), as shown in Figure 2. For CD4+ and CD8+ subsets co-cultured with targets, the addition of FLZ induced phenotypic changes away from naïve and towards central and effector memory subsets (Figure 2). In the two control groups, naïve T-cells predominated at the 24-hour time point (Figure 2). Conclusions: The in vitro killing assay shows that T-cells from the spleens of immune competent C57Bl/6 hCD123eHOM mice recognize FLZ and hCD123 on transduced murine leukemia cells. Co-culture with T-cells + target cells + FLZ showed killing at 24 hours compared to controls. T-cell activation and proliferation were noted based on cell surface markers. The shift in phenotype from naïve to central and effector memory subsets suggests T-cell expansion upon leukemia antigen recognition and portends killing of target cells. Pending in vivo studies will evaluate if FLZ can activate endogenous T cells, induce CRS, and kill AMLGFP-hCD123 targets in the huCD3eHOM mice. If successful, we will use the huCD3eHOM model to test the efficacy of various therapies to mitigate CRS and preserve FLZ-mediated T-cell cytotoxicity against AMLGFP-hCD123 targets in huCD3eHOM mice. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

The Novel Bispecific Innate Cell Engager (ICE®) AFM28 Efficiently Directs Allogeneic NK Cells to CD123+ Leukemic Stem- and Progenitor Cells in AML

Introduction: Acute myeloid leukemia (AML) is a common hematologic malignancy characterized by expansion of undifferentiated myeloid cells in the bone marrow (BM) and peripheral blood. Although AML can be treated with curative intent, residual leukemic stem cells (LSCs) are frequently reservoirs for relapse, which is associated with a poor prognosis. Moreover, especially elderly patients are often not eligible for intensive treatment regimen. Therefore, there is a high need for novel treatment strategies, which are capable of eliminating LSCs in order to induce long-term remissions, suppress measurable residual disease (MRD) and prevent relapse. A promising emerging approach is the adoptive transfer of allogeneic natural killer (NK) cells, which was shown to be clinically active and could induce complete remissions in relapsed/refractory (R/R) AML. Innate Cell Engagers (ICE®) may further augment the activity of NK cell therapies, in particular against LSCs and hold the potential to increase frequency and durability of responses. The tetravalent bispecific CD123/CD16A ICE®, AFM28, directs the cytotoxic activity of CD16A-expressing NK cells towards CD123-expressing (CD123+) AML blasts and LSCs. In this study, we evaluated the efficacy of AFM28 in pre-clinical models of AML. Methods: Antibody-dependent cell-mediated cytotoxicity (ADCC) data were obtained by a 4h calcein release assay using AML cell lines and primary human NK cells (E:T ratio 2.5:1) in the presence of titrated antibodies. CD64 and CD123 expression levels on AML cell lines were evaluated by flow cytometry. Intracellular FACS was used to detect STAT5/pSTAT5. Proliferation of TF-1 cells in the presence of IL-3 or GM-CSF was measured via CellTiter Glo Assay (Promega). Using primary BM samples of n=6 AML patients and n=4 healthy donors, ADCC assays and colony-forming unit (CFU) assays were performed. BM-MNC (ADCC assay) or enriched CD34+ cells (CFU assay) and allogeneic NK cells (E:T ratio 1:1) were treated with AFM28 or an Fc-enhanced anti-CD123 antibody comparator at different concentrations of AFM28 (0-500 or 0/10/100/1000 pM, respectively) for 24h. Subsequently, ADCC assays were analyzed by flow cytometry and CFU assays were evaluated after 7-14 days. Systemic anti-tumor activity of AFM28 in vivo was assessed using in-life imaging (BLI) in a T cell-depleted huFcgR-C57BL6 transgenic mouse model against intravenously administered luciferase-transfected murine AML cells expressing human CD123 as tumor antigen. Results: Pre-clinical in vitro models using a panel of AML cell lines showed efficacious ADCC against CD123+ tumor cells by allogeneic NK cells in the presence of AFM28. ADCC induction by AFM28 was independent of leukemic cell mutational profiles and even targeted cells with low levels of CD123 surface expression. Of note, expression of the high-affinity IgG receptor CD64 on AML cell lines completely abolished ADCC induction by an Fc-enhanced anti-CD123 antibody whereas AFM28 also showed high ADCC efficacy against all tested CD64+ cell lines. In addition, AFM28 exerted an NK cell-independent inhibitory effect on the IL-3-induced phosphorylation of STAT5 and abolished proliferation of CD123+ TF-1 cells, indicating an antagonistic effect on IL-3R signaling. In ex vivo models using primary AML patient cells, AFM28 demonstrated strong ADCC mediated by allogeneic NK cells. CFU assays of patient-derived AML BM samples performed subsequent to incubation with allogeneic NK cells with and without AFM28 showed significantly reduced numbers of outgrowing colonies in AFM28 treated samples, indicating that AML LSCs and progenitor cells were eliminated. In the in vivo efficacy study with huFcgR-C57BL6 mice, treatment with AFM28 at 15 mg/kg b.w. led to complete inhibition of tumor growth in 5/5 animals throughout a treatment period of 42 days. In contrast, all untreated control mice (6/6) developed systemic disease. Conclusion: AFM28 induced highly potent and selective lysis of CD123+ leukemic cells, including LSCs and progenitor cells, by allogeneic NK cells in vitro and in vivo. The capacity to eradicate LSCs holds the promise for durable responses and the potential for long-term remissions in patients with R/R AML. AFM28 is currently being prepared for a first-in-human clinical investigation as monotherapy and in combination with allogeneic NK cells.

LigandCD70.CAR As a Platform for Dual-Targeting CAR T Cells for Acute Myeloid Leukemia

Chimeric antigen receptor (CAR) T cells have demonstrated remarkable potency in acute lymphoblastic leukemia (ALL), but similar success has yet to be seen in acute myeloid leukemia (AML). Due to the variable expression of targetable antigens on AML blasts and leukemic stem cells, treatment with single-antigen targeting immunotherapy may select for antigen-negative clones that ultimately cause disease progression or relapse. An appropriate combination of two different CAR-binders and co-stimulatory domains could both increase tumor control and reduce antigen escape. We utilized a binder derived from human CD27 protein, the only known ligand of CD70, to generate the LigandCD70.CAR, which demonstrated superior in vivo activity against AML compared to conventional antibody-derived CARs (Sauer et al 2021). As the LigandCD70.CAR consists of a full length CD27 fused with CD3-zeta, this design takes advantage of the physiologic receptor-ligand interaction of CD70/CD27 to bind with high affinity to CD70+ cells and provides CD27-mediated costimulation to the T cells. These characteristics combined with the high expression of CD70 on more than 80% of AML make the LigandCD70.CAR a promising candidate for combination with other CARs to target AML. We evaluated the LigandCD70.CAR in combination with a CLL-1.CAR and a CD123.CAR, as both have shown preclinical efficacy and are currently being tested in clinical trials. We co-transduced activated T cells with LigandD70.CAR and CLL-1 CAR to generate LigandCD70/CLL-1 bispecific CAR (CLL-1.BiCAR) T cells. CLL-1.BiCAR T cells had comparable in vitro expansion to single-CAR expressing T cells but with enhanced cytotoxicity against CLL-1+ AML cell lines compared to single CLL-1.CARs, even if CD70 expression on tumor cells was low. In xenograft AML models of CD70-low CLL-1-high tumors (HL60), CLL-1.BiCAR T cells enhanced tumor control, survival and T cell persistence compared to single CLL-1.CAR T Cells (Figure 1). In order to determine whether the LigandC70.CAR can enhance other CARs, we co-transduced ligandCD70.CAR with CD123.CAR (Riberdy et al 2020) to generate a Ligand.CD70/CD123 bispecific CAR (CD123.BiCAR). The in vitro expansion and cytotoxicity was comparable to single-CAR expressing T cells. Treatment with CD123.BiCAR T cells significantly improved in vivo survival and tumor control in a xenograft model of AML with CD70-low/CD123-high target cells. In summary, LigandCD70.CAR co-expressed with either CLL-1.CAR or CD123.CAR demonstrate enhanced anti-tumor efficacy compared to single-targeting CLL-1.CAR and CD123.CAR, even in CD70-low tumors. The simultaneous targeting of two AML antigens may thus mitigate the risk of antigen-negative relapse. These data indicate that the Ligand70.CAR combined with other AML targeting CARs could be an effective therapeutic option for AML. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Single-Cell Map of Childhood Acute Myeloid Leukaemia Using Variational Auto-Encoders

Acute myeloid leukaemia (AML) is a haematological cancer in the bone marrow, with accumulation and expansion of immature cells of the myeloid lineage. Standard treatment of childhood AML is chemotherapy, which does not achieve durable remission in many patients. Personalised medicine including immunotherapies against surface antigens can potentially target chemotherapy resistant cells and achieve long-term remission. Identifying AML blasts and suitable targets for AML therapy are hampered by the heterogeneity and complex clonal and phenotypic composition of the cancer, as well as its complex evolution as the disease progresses into relapse. Aberrant expression of surface antigens on AML blasts measured by flow cytometry is an established strategy to monitor measurable residual disease. Although the likelihood to find blasts is increased as more markers are included in a staining, manual interrogation of all possible combinations exceeding 10 markers becomes extremely difficult. We aim to identify malignant cells based on expression of 23 surface markers normally or aberrantly expressed during hematopoiesis using spectral flow cytometry and machine learning. The goal of this high-dimensional approach is to distinguish leukaemia cells from normal myeloid cells with high confidence, and to place the blasts along the normal myeloid developmental trajectory. We built a single-cell AML map from 20 paediatric AML patients enrolled in the Children's Oncology Group Phase III trial, AAML1031. All patients were treated with standard chemotherapy on Arm A and consented to provide tissue for research. Cryopreserved samples were obtained for three time-points: diagnosis, remission, and relapse. We assumed full remission samples to be healthy and used them to train variational auto-encoders with different hyper-parameters in a cross validated manner. These models subsequently learned to encode the expression of healthy cells in a reduced dimension and to reconstruct their expression from this latent space. We selected the top-10 models that best reconstruct the healthy cells and used it to encode and reconstruct cells from non-remission samples. Since the model works well for encoding and reconstructing healthy cells, we expect a good reconstruction of healthy cells in non-remission samples. For malignant cells we expect a poor reconstruction, as their expression pattern differs from healthy cells. Thus, we use the reconstruction error of cells from non-remission samples to classify them as healthy or malignant. We annotated some cells with clear aberrant expression as malignant and shuffled them with remission cells not used for model development. This resulted in multiple synthetic mixtures with known percentages of malignant cells (ranging from 10 to 90% blasts). Using these synthetic mixtures, we evaluated how well we can classify malignant cells based on their mean-squared reconstruction error. We can achieve an average area under the receiving operator curve of 0.9 over all synthetic mixtures (Fig. 1). The latent space of the variational auto-encoder captures the developmental trajectory of myeloid lineage bone marrow cells (Fig. 2), and we can use this to identify where among the developmental trajectory the leukaemia occurs. We aim to use the trajectory assignment to study the time evolution of the disease in terms of the distribution of malignant cells across the myeloid lineage as well as investigate differences between diagnosis and relapse. We expect that this work will reveal patterns of evolution from diagnosis to relapse that will inform the development of novel strategies to predict and prevent relapse. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Different Subsets of Exhausted and Senescent-like T Cells Correlate with Response and Survival in Patients with Acute Myeloid Leukemia (AML) Undergoing Chemotherapy

Introduction Novel immunotherapies have changed the treatment paradigm for several oncohematologic diseases. In AML, these strategies have found limited application so far, in part due to incomplete understanding of bone marrow (BM) immune microenvironment. Several studies have identified T-cell exhaustion and senescence as two dominant states of dysfunction. We hypothesized that T cell dysfunctional states will predict response to induction chemotherapy in patients with AML. Methods We analyzed serial BM samples from healthy donors (HD); and AML patients responding (Res) and not responding (NonRes) to chemotherapy. We performed flow-cytometry in 10 Res, 6 NonRes (at diagnosis and at time of response assessment) and 12 HD. We subsequently performed single-cell RNA sequencing (scRNAseq) to analyze BM of 2 HD, 3 Res, and 2 NonRes at diagnosis and response assessment. This was followed by a scRNAseq-guided 26-color spectral flow-cytometry panel (12 Res and 10 NonRes). In-vitro assays were conducted to study the impact of blasts and metabolites on CD8+ T cells differentiation. Results Flow-cytometry identified a PD1+CD28+ CD8+ T-cell subset significantly (p<0.01) increased in Res versus NonRes at baseline and post-chemotherapy. Unsupervised clustering of scRNAseq data identified 6 main CD8+ subsets: naive, GZMK+IL7R+, intermediate GZMK+ (int-GZMK), intermediate(Int), senescent-like (SL), and terminally exhausted (Ttex). The GZMK+IL7R+, int-GZMK and Ttex expressed ITGAE and ENTPD1, supporting the idea that these subsets are tumor reactive and exhaustion is antigen-driven. Additionally, GZMK+IL7R+ cells had a progenitor-exhausted signature expressing stemness (TCF7, IL7R), anti-apoptotic (BCL2) and exhaustion (PDCD1, TIGIT) markers. This stem-like signature may allow long-term in-vivo persistence and anti-tumor immune surveillance consistent with the significant increase of GZMK+IL7R+ and Int-GZMK in Res vs NonRes. Conversely, Ttex and SL were significantly increased in NonRes. Trajectory inference algorithms identified two pathways with different end stages (SL; Ttex) demonstrating that senescence and exhaustion coexist but are different states in AML patients (Figure 1A). At trajectory's divergence, GNLY, NKG7, PRF1 and GZMB were the most upregulated genes in the "senescent” lineage while GZMK was upregulated in the "exhaustion” lineage. At the clonal level, the most expanded subsets were GZMK+IL7R+ and int-GZMK in Res and SL in NonRes, revealing a strong relationship between CD8+ T-cell subsets abundance and clonal expansion. The same subsets were confirmed by spectral flow-cytometry. GZMK+IL7R+cells were significantly increased in Res vs NonRes and SL where significantly increased in NonRes vs Res. In this larger cohort we corroborated at the protein level the differentiation path found in scRNAseq, confirming senescent and terminally exhausted CD8+ T cells as end differentiation T-cell states in AML (Figure 1A). Patients with a higher-than-median CD8+ GZMK+IL7R+/SL had a significantly (p<0.05) prolonged overall survival, affirming the clinical relevance of the differential abundance of the identified subsets in AML (Figure 1B). Gene-set enrichment analysis on scRNAseq showed increased hypoxia in NonRes VS Res, suggesting that anaerobic (non-oxidative) metabolism could promote senescence or terminal exhaustion. In vitro experiments with flow cytometry-sorted BM T cells and AML blasts demonstrated that adding lactic acid promotes CD8+ T cells senescence (p<0.05) whereas sodium bicarbonate prevents the development of senescence (p<0.05). These effects appear to be AML blast-specific since co-culture of HD T cells with AML blasts resulted in similar phenotypic changes. Conclusion Our study provides novel insights in understanding T-cell differentiation in the BM microenvironment of AML patients. We found that terminal exhaustion and senescence are derived from two different differentiation paths and that the ratio between stem-like and senescent CD8+ T cells is predictive of survival. Based on our scRNAseq and in-vitro results, we hypothesize that blasts and diverse metabolic microenvironments in Res and NonRes may have a different impact on T-cell differentiation and be responsible for different patients outcomes. Further studies could explore strategies that promote stem-like functional T cells to gain long-term AML control. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Developing a Humanized Murine Model for Co-Transplantation of Acute Myeloid Leukemia

Humanized mouse models and primary disease xenografts have produced significant translational insight into the biology of the bone marrow (BM) niche and aided drug discovery in acute myeloid leukemia (AML). To enable the study of cell-cell interactions between hematopoietic stem and progenitor cells (HSPCs) and AML as an emerging source of drug resistance, we describe two complementary approaches to humanized AML xenotransplantation models. One strategy utilizes the sequential transplantation of humanized hCD34+ cells followed by human AML cells. Recipient NBSGW mice (NOD.Cg-Kit W-41J Tyr + Prkdcscid Il2rgtm1Wjl/ThomJ), known for high rates of medullary engraftment, were conditioned with 15mg/kg busulfan followed by injection with either granulocyte colony-stimulating (G-CSF)-mobilized or BM-derived human CD34+ HSPCs. Flow cytometric analysis of the BM following injection of hCD34+ cells revealed efficient BM engraftment by both G-CSF- and BM- derived hCD34+ cells, demonstrated by BM human chimerism (hCD45%) of 63.7±10.0% by 8-weeks post-injection. Further BM hCD45 subtype analyses revealed B-cell (hCD45+ hCD19+) dominant repopulation (78.3±4.48%), followed by myeloid cell (hCD45+ hCD33+) repopulation (18.12±4.19%). T-cell repopulation was predictably minimal (0.008±0.01%). At 9- and 15- weeks post hCD34+ cells injection, GFP-expressing MOLM-14 AML cells (MOLM-14-GFP; 1E5, 1E6, or 2E6 cells) were injected. Flow cytometric analyses of MOLM-14-GFP BM chimerism up to 21 days post-injection showed complete absence of MOLM-14-GFP cells in the BM with no significant changes in overall hCD45 chimerism and hCD45 subtype composition compared to a PBS-injected control. These observations are consistent with a potential role for residual healthy hematopoietic cells in suppressing AML engraftment in the BM, previously shown in congenic studies. To test this hypothesis, we developed a second humanized AML model designed to investigate immunotherapeutic approaches for AML, where AML blasts are engrafted first, then followed by a human immune cell challenge. FA-AML1 or Kasumi-1 AML cells were injected into NSG mice (NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ) at 8 weeks of age. Injected mice were monitored for AML engraftment by HLA-DR expression in peripheral blood (PB). Engraftment occurred 6-12 weeks post-injection. After AML engraftment, we reconstituted a human immune system in the mice with a one-time injection of peripheral blood mononuclear cells (PBMCs), confirmed by flow cytometry for CD3+ T-cells (14.7±9.2%), CD33+ myeloid cells, and CD19+ B-cells. AML burden in the PB decreased 5-10% upon PBMC engraftment. In this model, a PB burden of <10% AML engraftment confers durable leukemic remission, whereas at >60% AML engraftment, leukemic cells persist in the periphery for over 100 days. Flow cytometric analysis demonstrated immune cell numbers peak in week 1 post-injection (30%), then decrease by week 4 to less than 1% in both the BM and PB, but persist at this low level for over 100 days. Altogether, these models confirm the ability of human hematopoietic cells to actively suppress leukemic engraftment in the BM microenvironment. More broadly, we demonstrate the development of two complementary humanized AML co-transplantation models that can attain dynamic ranges of human immune cell and AML cell engraftment to facilitate preclinical therapeutic development studies.

Downregulation of the Mutated WT1 Alleles in Malignant Hematopoietic Cells Via Nonsense-Mediated mRNA Decay Contradicts Their Prognostic Impact

Background The Wilms Tumor 1 (WT1) gene is highly expressed in almost all types of malignancies including acute myeloid leukemia (AML). It is also mutated in up to 20% of AML blasts in pediatric and adult patients. Both, the increased expression and also the presence of mutations are correlated negatively with chance of survival in different AML patient cohorts. We and others have previously showed devastating survival outcomes in children with AML blasts harboring WT1 mutations in combination FLT3-ITD and NUP98-NSD1 (Niktoreh et al., 2019). As WT1 can act as both, oncogene and tumor suppressor gene, this study was designed to analyze the WT1 genomic locus mutations in acute leukemia blasts. Methods and Results Screening for WT1 sequence alterations. Using the TruSight Myeloid Sequencing PanelTM (Illumina), we initially screened 35 leukemia and lymphoma cell lines for genomic alterations in 54 genes. This panel allowed us to detect even low variant-allele-frequency (VAF) alterations in the mutational hotspots of WT1, exons 7 and 9. In addition to a known WT1 mutation in CTS cells (AML M1), we discovered novel alterations in three other malignant cell lines. KASUMI-6 cells (AML M2) harbored the identical mutation as CTS cells, a 2-bp deletion in WT1 exon 7 leading to frameshift with a premature termination codon (c.1202_1203delGA, p.Arg401IlefsTer3). The U937 histiocytic lymphoma cells carried the known TP53 mutation and also an unknown heterozygous nonsense mutation in WT1 exon 7, changing the arginine 369 into a premature stop codon (c.1105C>T, p.Arg369Ter). Interestingly, MOLM-1 cells (CML in blast crisis) harbored a p.Arg462Gln alteration. Next, we performed Sanger sequencing for the complete coding region of WT1, including the upstream promoter as well as the 5'UTR and 3'UTR regions, all 10 exons and the adjacent 150 bp intronic sequences at the exon-intron boundaries. Sanger sequencing confirmed the mutations discovered via NGS and revealed single bp alterations in the WT1 promoter region of CTS cells and in the 5'UTR of K-562 cells. We also detected a recurring variation in intron 3 of WT1 (c.872+16G>A) in the cell lines K-562, NB-4, M07e and UOC-M1; the functional implications of the alterations are yet to be determined. Functional impact of WT1 mutations. To understand the potential impact of the WT1 sequence alterations, we studied the WT1 mRNA expression in these cells in the presence of cycloheximide (CHX), an inhibitor of the nonsense-mediated mRNA decay (NMD) machinery. As shown in figure 1A, U937 cells were used as negative control cells, as they are known to have no expression of the WT1 mRNA. Despite the presence of mutations in CTS, KASUMI-6 and MOLM-1 cells, all three cell lines expressed elevated levels of WT1, well within the range of 30 other malignant cell lines without any alterations/mutations in the WT1 locus (data not shown). To differentiate the WT1 expression between the mutated and the wildtype allele, we performed a fragment length analysis on a capillary electrophoresis system using the complimentary DNA of CHX-treated cell lines. For these analyses, we employed a multiplex PCR strategy using ABL1 as housekeeping gene and a primer pair binding in exons 4 and 10 of WT1. As delineated in figure 1B, the mutated alleles in CTS and KASUMI-6 cells, which are 2-bps shorter, could be easily separated from the wildtypes (data for KASUMI-6 not shown). Additionally, since our primer pair also amplified the alternatively spliced exon 5 and the alternative splice donor at the end of exon 9, leading to inclusion or exclusion of three amino acids (KTS), we were able to separately analyze the four major WT1 isoforms from both alleles. As shown in figure 1B, the expression levels of the WT1 mutated alleles of CHX-treated cells increased 2- to 3-fold compared to DMSO-treated cells in CTS and 3- to 4-fold in KASUMI-6 (data not shown). Conclusion Careful analyses of the WT1 genomic locus in 35 malignant hematological cell lines revealed previously unknown heterozygous loss-of-function WT1 mutations in KASUMI-6, MOLM-1 and U-937 cells. mRNA expression analysis demonstrated that these mutations are strongly regulated by NMD machinery, leading to (almost) complete downregulation of the mutated WT1 allele. These results strongly question any direct effect of WT1 mutations in the leukemic blasts and therefore cannot explain the role of mutations in WT1 as a prognostic factor in AML. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Programmed Death-1 Homolog (PD-1H/VISTA) Blockade Confers a Synergistic Anti-Leukemia Effect with PD-1 Blockade

Monoclonal antibodies (mAbs) to block the interaction of PD-1/PD-L1 (herein anti-PD therapy) have been at the forefront of cancer therapy. Despite its success in many cancers, anti-PD therapy as a single agent has shown modest response rates in acute myeloid leukemia (AML), indicating different mechanisms of immune evasion in human AML. Programmed Death-1 Homolog (PD-1H, also known as VISTA) is a co-inhibitory immunoglobulin broadly found in hematopoietic cells. We previously found that PD-1H surface protein was expressed on AML blasts; AML blast PD-1H inhibited T cell immunity, leading to immune evasion in AML (Kim et al, ASH 2017). Now, we demonstrate that PD-1H is highly expressed on normal myeloid cells and on T cells in AML bone marrow. To determine the role of host immune cell PD-1H in immune evasion of AML, AML cells (C1498) were transplanted into B6 mice with wild-type (WT) or PD-1H knockout (KO) immune cells to assess in vivo proliferation using bioluminescence. In vivo AML growth was delayed in PD-1H KO mice compared with WT mice (mean radiance in PD-1H WT vs. PD-1H KO mice on day 24: 4.4x108 vs. 5.1x105, N=5 per group, p=0.04). To elucidate which immune cell subsets expressing PD-1H contribute to immune evasion in AML, we generated lineage-specific conditional KO mice that lacked expression of PD-1H in either T cells (Cre-Lck) or myeloid cells (Cre-LysM) (macrophages, granulocytes). Following transplantation with AML cells, we assessed in vivo proliferation in both conditional KO mice and littermate controls using bioluminescence. We found that in vivo proliferation of AML was significantly inhibited by myeloid cell-specific KO of PD-1H, compared with littermate controls (mean radiance in Cre-LysM-PD-1Hfl/fl vs. Cre-LysM+PD-1Hfl/fl on day 23: 1.7x109 vs. 1.7x106, N=9 (combined from two independent experiments), p=0.03). T cell-specific KO of PD-1H showed a trend toward anti-leukemia effects, but the effect was not statistically significant (mean radiance in Cre-Lck-PD-1Hfl/fl vs. Cre-Lck+PD-1Hfl/fl on day 34: 2.5x108 vs. 2.2x107, N=6, p=0.1). Taken together, these results show that PD-1H on host myeloid cells rather than on T cells contributes to immune evasion in AML. Consistent with reduced in vivo AML growth in PD-1H KO mice, anti-mouse PD-1H (mPD-1H) mAb treatment significantly inhibited in vivo AML growth compared with isotype control in a syngeneic AML model, while anti-mouse PD-1 (mPD-1) mAb modestly reduced in vivo AML growth. However, treatment with a combination of anti-mPD-1H mAb and anti-mPD-1 mAb synergistically suppressed in vivo growth of AML in WT B6 mice, compared with mice treated with either mAb alone (mean radiance with anti-mPD-1H mAb vs. the combination of anti-mPD-1 mAb and anti-mPD-1H mAb on day 21: 2x106 vs 2.4x105, N=5 per group, p=0.017; anti-mPD-1 mAb vs. the combination: p=0.00053). Finally, we determined the effect of human PD-1H (hPD-1H) in a humanized AML model. NSG-SGM3 mice were first reconstituted with purified human T cells and then subcutaneously injected with human leukemia cells (THP1) expressing or not expressing hPD-1H. Two weeks later, PD-1H+ AML tumors were significantly bigger than PD-1H- AML tumors (PD-1H+ AML tumors: 78.8 mm3 vs. PD-1H- AML tumors: 8.5 mm3, N=5, p<0.0001). Interestingly, treatment with anti-hPD-1H mAb significantly reduced the size of PD-1H+ AML tumors, accompanied by increased T cell infiltration, while anti-hPD-1 mAb did not significantly inhibit tumor growth (anti-hPD-1H mAb: 7.2 mm3 vs. anti-hPD-1 mAb: 51.4 mm3, N=5, p=0.0032). More importantly, the combination treatment of anti-hPD-1H mAb with anti-hPD-1 mAb synergistically reduced the size of PD-1H+ AML tumors (0 mm3, N=5, p=0.0035 vs. anti-hPD-1 alone). These findings extend and validate the results in syngeneic mouse leukemia models by showing that hPD-1H mAb can reverse immune evasion induced by leukemia-associated PD-1H in a humanized AML model. Our studies reveal that host immune cell PD-1H in addition to AML blast PD-1H induces immune evasion in syngeneic and humanized AML mouse models. More importantly, PD-1H blockade by a specific mAb reverses this immune evasion. Finally, PD-1H blockade is potentially synergistic with anti-PD therapy. Our studies implicate PD-1H as a potential target for AML immunotherapy.

Safety and Anti-Leukemic Activity of CD123-CAR T Cells in Pediatric Patients with AML: Preliminary Results from a Phase 1 Trial

Background The prognosis of pediatric patients with relapsed or refractory (r/r) Acute Myeloid Leukemia (AML) remains dismal. Evaluation of novel therapies such as chimeric antigen receptor (CAR) T cells is therefore urgently needed. CD123 is overexpressed on AML blasts and leukemic stem cells (LSC), making it an attractive target for CAR T-cell directed therapy. Methods and Results We designed a 'bridge to allogeneic hematopoietic cell transplant (HCT)’ Phase 1 study to evaluate the safety and feasibility of treating pediatric patients with r/r AML with CD123-CAR T-cell on four dose levels (DL): DL1: 3x105/kg, DL2: 1x106/kg, DL3: 3x106/kg, DL4: 1x107/kg (NCT04318678). CD123-CAR T-cells were generated from CD4/CD8-selected autologous leukapheresis products using a lentiviral vector, which encoded a CD123-CAR with CD28.z signaling domain and a CD20 safety switch. Patients received lymphodepleting chemotherapy with Fludarabine and Cyclophosphamide followed by a single CAR T cell infusion. To date we have enrolled 12 patients. The median age of treated patients was 17 years (range 12-21years). Except for the first patient who had primary refractory disease, all other patients had relapsed following HCT (n=1-4 previous HCTs). Despite patients being heavily pre-treated, we successfully manufactured products on all patients. The infused products had a predominantly CD4+ effector memory immunophenotype, with a median CD123-CAR+/CD20+ expression of 60.2% (range 46.1-79.4%, N=6). All infused CAR T cell products exhibited potent antitumor activity in vitro when cultured against CD123+ targets (including autologous blasts for 2 patients). We infused 2 patients on DL1 and 3 patients on DL2. Two patients were infused on single patient protocols. All infusions were well tolerated, without any adverse infusion events. Post infusion, we observed isolated fevers that resolved within 24 hours (possible Grade 1 Cytokine Release Syndrome (CRS)). No Grade 2 ≥ CRS or neurotoxicity was seen. Multiplex cytokine analysis showed statistically significant increase of IL15 after lymphodepletion without significant increase in Th1 cytokines or IL6 after T cell infusion. While transient cytopenias were observed, patients did not develop persistent marrow aplasia. No dose limiting toxicities were noted. Disease evaluation was performed at 4-6 weeks following CAR T cell infusion. The 2 patients on DL1 showed no response. In the 3 patients on DL2, we observed: reduction in blast percentage without complete remission (CR) in 1 patient, no response in 1 patient and CR in 1 patient. The patient with CR had isolated extramedullary disease, showing complete resolution of all lesions by week 4 by PET imaging (Figure 1). She recurred 2 months following infusion, coinciding with loss of CAR T-cell detection by qPCR. She received a second infusion off study and again achieved a short-lived CR. One additional patient was infused on dose level 2 off protocol. She achieved morphological CR at day 28 with low level minimal residual disease (0.19%). Correlative studies revealed CD123-CAR T cell expansion in patients on DL2 (Figure 2), but not on DL1. However, peak expansion on DL2 was low in comparison to published reports on CAR T cells for Acute Lymphoblastic Leukemia. Detailed phenotypic analysis revealed that CD123-CAR T cell products were predominantly effector memory (CD45RO+CCR7-) with <2% naïve-like (CD45RO-CCR7+ or CD45RO-CD62L+) T cells. In addition, PD1, TIM3 and CD39 were expressed at a median of 43, 81 and 16% in CD4+ and 3, 72 and 45% in CD8+ T cells. This phenotype could be completely reversed by inhibiting CAR signaling with dasatinib during the manufacturing process, resulting in a CAR T cell product dominated by naïve-like T cells that did not express TIM3, PD1 or CD39, leading to enhanced T-cell effector function. Mechanistic analysis revealed that the observed phenotype could be explained by the transient, increased cell surface expression of CD123 on T cells upon activation during manufacturing, and not by tonic CAR signaling. Conclusions Our initial clinical experience with CD123-CAR T cells for pediatric patients with r/r AML demonstrates feasibility, safety, and evidence of anti-leukemic activity. Upcoming CD123-CAR T cell products will be manufactured in the presence of dasatinib to limit T-cell differentiation and exhaustion as we continue to explore our CD123-CAR T cell therapy approach for pediatric r/r AML. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal