Zelenirstat Inhibits N-Myristoyltransferases to Disrupt Src Family Kinase Signaling and Oxidative Phosphorylation, Killing Acute Myeloid Leukemia Cells

Abstract

Acute myeloid leukemia (AML) is a hematologic malignancy with limited treatment options and a high likelihood of recurrence after chemotherapy. We studied N-myristoylation, the myristate modification of proteins linked to survival signaling and metabolism, as a potential therapeutic target for AML. N-myristoylation is catalyzed by two N-myristoyltransferases (NMT), NMT1 and NMT2, with varying expressions in AML cell lines and patient samples. We identified NMT2 expression as a marker for survival of patients with AML, and low NMT2 expression was associated with poor outcomes. We used the first-in-class pan-NMT inhibitor, zelenirstat, to investigate the role of N-myristoylation in AML. Zelenirstat effectively inhibits myristoylation in AML cell lines and patient samples, leading to degradation of Src family kinases, induction of endoplasmic reticulum stress, apoptosis, and cell death. Zelenirstat was well tolerated in vivo and reduced the leukemic burden in an ectopic AML cell line and in multiple orthotopic AML patient-derived xenograft models. The leukemia stem cell-enriched fractions of the hierarchical OCI-AML22 model were highly sensitive to myristoylation inhibition. Zelenirstat also impairs mitochondrial complex I and oxidative phosphorylation, which are critical for leukemia stem cell survival. These findings suggest that targeting N-myristoylation with zelenirstat represents a novel therapeutic approach for AML, with promise in patients with currently poor outcomes.

Figure 1.

NMTs are variable and sensitive to zelenirstat in AML. Levels of NMT1 and NMT2 were assayed in a variety of AML cell lies (A and B) and patient samples (C and D) by Western blot (A and C) or flow cytometry (B and D), presented as mean fluorescence intensity ± SEM. Click chemistry using alkyne-myristate–labeled cell lysates was used to determine myristoylation levels in MV-4-11 AML cell line (E), cells from patients with AML (F), and PBMCs extracted from healthy individuals (G) following treatment with zelenirstat for 1 hour.

Figure 2.

NMT2 and MISS-54 are prognostic markers in AML. Kaplan–Meier survival of patients in the TCGA-LAML project. Patients were separated into quartiles based on RNA sequencing expression of NMT1 (A) or NMT2 (B) and survival of the top and bottom expressing groups compared. Curve comparison performed using the log-rank test. C and D, Boxplots of NMT2 expression categorized by NPM1 or IDH1 mutant status compared via t test. E, MISS-54 for patients in TCGA-LAML compared with all other TCGA project tumors via t test. F, Kaplan–Meier survival of patients in TCGA-LAML in top and bottom quartiles of MISS-54 score. Curve comparison performed using the log-rank test. G, MISS-54 scores of patients in TCGA-LAML classified by cellular hierarchy composition compared via t test.

Figure 3.

Zelenirstat kills AML cells in vitro and in vivo. A, AML cell lines, (B) patient cells, or healthy cells were incubated with zelenirstat in vitro for 96 hours and viability measured by CellTiter-Blue assay. C, MV-4-11 cells were injected into the flank of immunodeficient mice. After 10 days, mice received indicated dose of zelenirstat daily via s.c. injection. Each group contains 10 mice. D, Patient-derived AML cell model CTG-3439 was injected into NOG-EXL mice via i.v. tail vein injection. Mice received indicated concentrations of zelenirstat via oral gavage following a 4-day on, 3-day off regimen. Tail vein bleeds were collected at indicated time points, and maximal blood volume and bone marrow flushes collected following euthanasia at day 42. Samples were analyzed by flow cytometry for the presence of human CD45. Terminal samples are pooled and represent all animals in group. Treatments were compared by mixed-effects analysis. E, Primary patient AML samples were injected into NOD/SCID gamma mice via intrafemoral xenograft and treated with 50 mg/kg zelenirstat via oral gavage 3× weekly. After 4 weeks of treatment, mice were sacrificed and bone marrow flushes collected, and samples analyzed via flow cytometry for the presence of human CD45. Each dot represents one mouse.

Figure 4.

LSC-enriched populations seem particularly sensitive to zelenirstat. OCI-AML22 cells were cultured with zelenirstat for 48 (A) or 72 (B) hours and then analyzed for viability and immunophenotype via flow cytometry. Cell counts were expressed as a percentage of vehicle control and normalized. IC₅₀ defined by least squares fit model.

Figure 5.

Zelenirstat disrupts signaling in AML and induces apoptosis. MV-4-11 was incubated with zelenirstat for 48 hours, stimulated with 100 ng/mL FL and SCF, and then lysed. Lysates were probed for HCK, phosphorylated SFKs, and phosphorylated Stat5 by Western blotting (A). MV-4-11 was incubated with zelenirstat for up to 72 hours, then stimulated with 100 ng/mL FL and SCF, and lysates probed for BiP and cleaved caspase 3 (B). Band intensity was determined using Image Studio Lite, and each sample normalized to actin. Differences between treatment groups determined using one-way ANOVA (A and C). Samples from patients with AML and healthy PBMCs were incubated with indicated concentrations of zelenirstat for 96 hours and then stained with annexin V and propidium iodide. Cells were analyzed via flow cytometry and live/dead/apoptotic percentages determined by FlowJo (D). FL, FLT3 ligand.

Figure 6.

Zelenirstat targets AML cell metabolism. MV-4-11 cells were treated with zelenirstat at indicated concentrations for 48 hours and then metabolic activity analyzed via Seahorse glycolytic rate assay. Data shown as time course (A) of average normalized OCR of three biological replicates, and value comparison at each stage of the assay (B). C, Pretreated AML cell lines were analyzed for cellular ATP content. D, OCR of MV-4-11 cells treated with indicated zelenirstat concentrations measured using Resipher real-time oxygen sensors. Data represent average peak OCR achieved by three biological replicates. E–H, Mitochondria were isolated from MV-4-11 and U937 cells treated with zelenirstat for 48 hours, then mitochondrial membranes were solubilized, and proteins separated by blue-native PAGE. Complex I activity was visualized in-gel using nitro blue tetrazolium activity buffer (E and F), or proteins transferred to nitrocellulose membranes and probed for NDUFB11 (E and G). Lysates of isolated mitochondria separated by SDS-PAGE were also probed for NDUFAF4 (E and H) and VDAC1/2. Band intensities were normalized to VDAC1/2 levels and quantified (F–H). Complex I activity of MV-4-11 cells treated with zelenirstat was analyzed via Abcam Complex I Activity Assay (I). Data expressed as change in A450 over assay period of 1 hour. Error bars represent ± SEM.