Abstract
Asparaginase, a bacterial enzyme that depletes the nonessential amino acid asparagine, is an integral component of acute leukemia therapy. However, asparaginase resistance is a common clinical problem whose biologic basis is poorly understood. We hypothesized, based on the concept of synthetic lethality, that gain-of-fitness alterations in the drug-resistant cells had conferred a survival advantage that could be exploited therapeutically.
To identify molecular pathways that promote fitness of leukemic cells upon treatment with asparaginase, we performed a genome-wide CRISPR/Cas9 loss-of-function screen in the asparaginase-resistant T-ALL cell line CCRF-CEM. Cas9-expressing CCRF-CEM cells were transduced with a genome-wide guide RNA library (Shalem et al. Science343, 84-87, 2014), treated with either vehicle or asparaginase (10 U/L), and guide RNA representation was assessed. Our internal positive control, asparagine synthetase, was the gene most significantly depleted in asparaginase-treated cells (RRA significance score = 1.56 x 10-7), followed closely by two regulators of Wnt signaling, NKD2 and LGR6 (RRA score = 6 x 10-6and 2.19 x 10-5, respectively). To test how these genes regulate Wnt signaling in T-ALL, we transduced CCRF-CEM cells with shRNAs targeting NKD2 or LGR6, or with an shLuciferase control. Knockdown of NKD2 or LGR6 increased levels of active β-catenin, as well as the activity of a TopFLASH reporter of canonical Wnt/β-catenin transcriptional activity (P < 0.0001), indicating that NKD2 and LGR6 are negative regulators of Wnt signaling in these cells. We then validated the screen results using shRNA knockdown of NKD2 or LGR6, which profoundly sensitized these cells to asparaginase (P< 0.0001) and potentiated asparaginase-induced apoptosis (P < 0.0001).
Inhibition of glycogen synthase kinase 3 (GSK3) is a key event in Wnt-induced signal transduction. Thus, we tested whether CHIR99021, an ATP-competitive inhibitor of both GSK3 isoforms (GSK3α and GSK3β), could phenocopy the effect of Wnt pathway activation. Pharmacologic inhibition of GSK3 induced significant sensitization to asparaginase across a panel of cell lines representing distinct subtypes of treatment-resistant acute leukemia, including T-ALL, AML and hypodiploid B-ALL (Fig. 1a, b). Importantly, GSK3 inhibition did not sensitize normal hematopoietic progenitors to asparaginase, suggesting a leukemia-specific effect.
Wnt-induced sensitization to asparaginase was independent of β-catenin and mTOR activation, because genetic and pharmacologic manipulation of these Wnt targets had no effect on asparaginase response. Instead, it was mediated by Wnt-dependent stabilization of proteins (Wnt/STOP), which inhibits GSK3-dependent protein ubiquitination and proteasomal degradation (Acebron et al. Mol Cell54, 663-674, 2014, Taelman et al. Cell143, 1136-1148, 2010). Indeed,Wnt-induced sensitization to asparaginase was completely blocked by the transduction of leukemia cells with FBXW7 (P < 0.0001), whose overexpression can reverse Wnt/STOP (Acebron et al. Mol Cell54, 663-674; 2014), or by expression of a hyperactive proteasomal subunit ΔN-PSMA4 (P < 0.0001), which globally increases protein degradation (Choi et al. Nat Commun7, 10963, 2016).
Although GSK3α and GSK3βare redundant for many of their biologic functions, genetic or pharmacologic inhibition of GSK3α fully phenocopied Wnt-induced sensitization to asparaginase (P < 0.0001), whereas selective inhibition of GSK3β had no effect. We then leveraged the recently developed GSK3α-selective small molecule inhibitor BRD0705 (Wagner et al. Sci Transl Med10, 2018) to test the in vivo therapeutic potential of our findings. Immunodeficient NRG mice were injected with leukemic cells from a primary asparaginase-resistant T-ALL patient derived xenograft, and treatment was begun after confirmation of leukemic engraftment (n=16 mice per group). In vivo, this PDX proved completely resistant to asparaginase or BRD0705 monotherapy, whereas the combination was highly efficacious (median survival of 17 days in vehicle, vs. median not reached at 60 days in combo-treated mice; P < 0.0001; Fig. 2a, b). The combination was also well-tolerated, with no appreciable weight changes or increases in serum bilirubin levels.
Our findings provide a molecular rationale for activating Wnt/STOP signaling to improve the therapeutic index of asparaginase.
Disclosures
No relevant conflicts of interest to declare.
Prediction and analysis of multiple protein lysine modified sites based on conditional wasserstein generative adversarial networks
