Arginine dependency is a therapeutically exploitable vulnerability in chronic myeloid leukaemic stem cells

Abstract To fuel accelerated proliferation, leukaemic cells undergo metabolic deregulation, which can result in specific nutrient dependencies. Here, we perform an amino acid drop‐out screen and apply pre‐clinical models of chronic phase chronic myeloid leukaemia (CML) to identify arginine as a nutrient essential for primary human CML cells. Analysis of the Microarray Innovations in Leukaemia (MILE) dataset uncovers reduced ASS1 levels in CML compared to most other leukaemia types. Stable isotope tracing reveals repressed activity of all urea cycle enzymes in patient‐derived CML CD34+ cells, rendering them arginine auxotrophic. Thus, arginine deprivation completely blocks proliferation of CML CD34+ cells and induces significantly higher levels of apoptosis when compared to arginine‐deprived cell lines. Similarly, primary CML cells, but not normal CD34+ samples, are particularly sensitive to treatment with the arginine‐depleting enzyme, BCT‐100, which induces apoptosis and reduces clonogenicity. Moreover, BCT‐100 is highly efficacious in a patient‐derived xenograft model, causing > 90% reduction in the number of human leukaemic stem cells (LSCs). These findings indicate arginine depletion to be a promising and novel strategy to eradicate therapy resistant LSCs.

A Indicated cell lines grown for 72 h in complete medium or medium deficient in arginine and cell number recorded (per well).Cells were seeded at 10,000 cells (dashed line) in 200 ll/well in three replicate plates.Mean, and SEM are plotted.B Indicated cell lines grown for 72 h in complete medium or medium deficient in arginine and viability measured.Three independent experiments are shown with mean and SEM.Live cells (Annexin VÀ, 7-AADÀ) were analysed.C Representative plot showing propidium iodide (RNASE-PI) staining after culturing K562 cells with or without arginine for 72 h.Three independent experiments are shown with mean, and SD.D Indicated cell lines grown for 16 h in complete medium or medium deficient in arginine and ASS1 protein levels were quantified.One representative plot is shown in top panel.Below, 3-4 independent experiments are shown with mean and SEM.E ASS1 expression from the MILE study.The dotted line revers for average of Healthy Bone Marrow; ** refers to P < 0.0021, *** refers to P < 0.0002 and ****refers to P < 0.0001.F ASS1 expression in indicated stem-cell enriched datasets.CP: chronic phase, AP: accelerated phase, BP: blast phase.HPC and LPC refer to normal and leukaemic progenitor cells respectively.CMP, GMP, MEP and MPP refer to common myeloid, granulocyte-macrophage, megakaryocyte-erythrocyte, and multipotent progenitors respectively.Markers used in these studies were E-MTAB-2581: CD34, CD38, GSE47927E-GEOD-47927: CD34  from three independent experiments.Mean and SEM is plotted.E Volcano plot from LC-MS analysis of CML CD34 + cells (n = 4 patients) treated with BCT-100 (100 ng/ml) for 24 h.Blue denotes q-value < 0.1, red denotes q-value < 0.05.F Pathway analysis of (E).Blue denotes q-value < 0.1, red denotes q-value < 0.05 (Benjamini & Hochberg).G Analysis of 13 C 5 glutamine labelling in de novo synthesis of pyrimidines one CML CD34 + samples that was treated with BCT-100 (100 ng/ml) for 24 h.H Volcano plot from LC-MS analysis of normal CD34 + cells (n = 3 biological samples) treated with BCT-100 (100 ng/ml) for 24 h.I L-arginine from (E) (4 patient samples) and (H) (n = 3 biological samples).Mean and SEM is plotted.Data information: An ordinary one-way ANOVA with Tukey's correction for multiple comparisons were performed on data from (A-C) (data from (C) was log-transformed to ensure normality).Kruskal-Wallis test with the Benjamini and Hochberg false discover correction was used to analyse data for (D).Metabolanalyst was used to calculate q-values after mean-centering and R-Log transforming data for (E) and (H).For (F), pathway analysis was conducted using Metabolanalyst after meancentering and R-Log transforming data.The Globaltest and relative betweenness centrality were used on Homo sapiens KEGG database.A Kruskal-Wallis test was used to analyse data for (I).A Indicated cell lines, normal or CML CD34 + cells were treated for 24 h with BCT-100 (1,000 ng/ml for cell lines and 100 ng/ml for primary samples).ASS1 counts from indicated cell lines (three independent experiments), normal samples (n = 3 biological replicates) and CML patient samples (n = 4 biological replicates) are shown.Mean and SD is plotted.B PCA for samples in (A).C PCA for primary samples only.D Volcano plot showing differentially expressed genes between normal and CML CD34 + cells, red denotes q-value < 0.05, blue denotes q-value < 0.1.E Volcano plot showing differentially expressed genes between normal CD34 + vehicle and BCT-100 treated cells, red denotes q-value < 0.05.F Venn diagram showing differentially expressed genes from indicated comparisons.G Volcano plot showing differentially expressed genes between Control and ASS1 KD K562 cells, red denotes q-value < 0.05, blue denotes q-value < 0.1.H Volcano plot showing differentially expressed genes between Control K562 cells treated with BCT-100 or vehicle, red denotes q-value < 0.05, blue denotes q-value < 0.1.I Volcano plot showing differentially expressed genes between ASS1 KD K562 cells treated with BCT-100 or vehicle, red denotes q-value < 0.05, blue denotes q-value < 0.1.J Downregulated KEGG pathways from GSEA analysis on genes from the instersect of KD-BCT and Control-BCT are shown.Here pi values (computed by multiplying log 2 fold change by Àlog 10 (q-value)) from Control-BCT were used.In red are sets with corrected P-values less than false discovery rate threshold (P < 0.25: dotted line).K Downregulated KEGG pathways from GSEA analysis on genes from the instersect of KD-BCT and Control-BCT are shown.Here pi values from KD-BCT were used.In red are sets with corrected P-value less than false discovery rate threshold (P < 0.25: dotted line).L The upregulated pathways corresponding to (J) are shown.M The upregulated pathways corresponding to (K) are shown.DESEQ2 and GSEA were used as described in methods.

Figure EV1 .
Figure EV1.Absence of ASS1 is a feature of CML patient samples.

▸Figure EV4 .
Figure EV4.Pharmacological arginine depletion uniqely effects CML patient samples with drastic effects on both control and ASS1 KO cells.

Figure EV5 .
Figure EV5.Pharmacological arginine depletion is human effective in CML LSC xenograft model.A Dosing strategy for escalation study.Serum samples were taken prior to first dose, each first escalation dose and 24 h after last dose.B Dosing strategy for PDX experiment.C Gating strategy is shown for flow cytometry analysis.D The percentage of CD45 + cells from single cells is shown.Biological replicate data from all mice, average and SD are plotted.Vehicle: n = 8 mice, Imatinib: n = 8 mice, BCT-100: n = 6 mice, Combo: n = 7 mice.E The percentage of CD34 + cells (from CD45) is shown.Average and SD are plotted.Biological replicate data from all mice, average and SD are plotted.Vehicle: n = 8 mice, Imatinib: n = 8 mice, BCT-100: n = 6 mice, Combo: n = 7 mice.F The absolute number of CD34 + cells is shown.Average and SD are plotted.Biological replicate data from all mice, average and SD are plotted.Vehicle: n = 8 mice, Imatinib: n = 8 mice, BCT-100: n = 6 mice, Combo: n = 7 mice.Data information: A Kruskal-Wallis test was used to analyse data for (D-F) with significant changes (adjusted P-value, all P-values in (D) are > 0.5) or relevant comparisons are shown.
ASL: argininosuccinic acid lyase.B Results from 48 h tracing of indicated labelled amino acid in CML CD34 + patient samples ( 13 C 6 arginine,13