Repurposing Nelarabine to Induce Differentiation of Acute Myeloid Leukemia

The success of all-trans retinoid acid (ATRA) in acute promyelocytic leukemia (APL) pioneered the concept of differentiation therapy. However, comparable approaches to overcome differentiation blockage for non-APL acute myeloid leukemia (AML) are hampered by lack of an effective drug discovery platform. Recently, we analyzed gene signatures of compounds (ATRA, arsenic trioxide, zalcitabine, and sodium butyrate) that trigger myeloid differentiation in the NCI-60 collection datasets and identified CD38 as the top gene upregulated by differentiation induction. We next initiated an in silico screen in the DTP database of >20,000 compounds to identify compounds that increase CD38 levels. Among those retrieved from "CellMiner" with CD38 as input, we assessed the top 193 available from NCI (r>0.6, p=0) for effects on differentiation utilizing a conditional murine myeloid differentiation-arrest model overexpressing estrogen receptor-HoxA9 (ER-HoxA9) fusion proteins (Cell, 2016). We identified NSC755985 (Nelarabine, NEL) in that screen.

NEL is an orphan drug approved to treat relapsed or refractory T-cell acute lymphoblastic leukemia (ALL). NEL at clinically achievable doses (C_max: 6.73 μM~ 26.91 μM, at a proposed adult dosing schedule of 1,500 mg/m²/day) markedly induced primary AML cell differentiation and death while sparing normal hematopoietic cells (AML vs normal, IC₅₀: 14.7 ± 4.3 μM, n=7, vs 45.3 ± 1.3 μM, n=3), suggesting a therapeutic window in AML. Ex vivo NEL treatment compromised BM engraftment of CD34⁺ cells from one primary AML specimen in immunodeficient NSG mice at 12 weeks post-transplant (human CD45⁺ cells in BM: NEL 0.73% vs vehicle 33.42%, n=6/group, p<0.01). NEL administration in vivo (130 mg/kg/day, i.v. for 5 consecutive days) reduced leukemic burden of NSG mice xenografted with luciferase-expressing U937 cells (Radiance: NEL 2.83×10⁷ vs vehicle 1.65×10⁸ photons/s/cm², n=9/group,p<0.01) and extended mouse survival.

Transcriptome analyses in U397 cells and primary AML specimens revealed that NEL treatment upregulated RAS-related pathways. NEL-elicited RAS activation was confirmed by pull-down assay using a GST-Raf1-RBD affinity probe, followed by blotting with a pan-RAS antibody. We performed functional analysis by infecting ER-HoxA9 cells with lentiviral vector expressing oncogenic RAS and observed enhanced myeloid differentiation, as evidenced by increased CD11b/GFP levels relative to MOCK-infected controls. Given that NEL's active metabolite Ara-GTP perturbs guanine nucleotide metabolism, we asked if NEL-evoked RAS activation was associated with accrual of intracellular GTP. HPLC/MS analyses of U937 cells showed that NEL treatment resulted in a marked increase in GTP (approximately 5-fold higher than baseline at sub-millimolar levels) which was secondary to Ara-GTP. Importantly, either electroporation of GTP into U937 cells or indirect introduction of GTP by addition of guanine utilizing purine salvage pathways activated RAS and recapitulated differentiation induction phenotypes.

To determine whether AML cells with higher RAS activity exhibited greater NEL sensitivity, we pretreated U937 cells with a RAS agonist KRA-553 or ectopically expressed RAS mutants and observed enhanced NEL inhibitory effects in both cases. We also observed enhanced vulnerability to NEL treatment in MLL-AF9 transformed murine hematopoietic cells from Kras^{Lox-Stop-Lox (LSL) G12D/+}/Vav-Cre mice (Blood, 2009) versus Cre⁺ counterparts. Relevant to AML line THP-1 which is poorly responsive to NEL (IC₅₀>100 µM), we observed extremely low levels of Ara-GTP, no GTP increase or RAS hyperactivation after NEL treatment; Ara-GTP is inactivated by SAM domain and HD domain-containing protein 1 (SAMHD1), a dNTP hydrolase, whose high expression reportedly underlies NEL resistance in T-ALL. Indeed, SAMHD1 deletion remarkably increased RAS activity in THP-1 cells treated with NEL, thereby fully reversing NEL resistance.

Our study provides a preclinical basis for testing NEL efficacy in a large cohort of AML patients, given that RAS activity is generally high in AML, or even against other malignancies harboring RAS mutations, which are considered "undruggable". Additionally, further study to test whether SAMHD1 inhibition enhances NEL efficacy against RAS active cancers is warranted.