Background Relapse remains as a formidable challenge for acute lymphoblastic leukemia (ALL), particularly because it is associated with dramatic drug resistance and thus dismal prognosis. It is crucial to understand the genetic basis of relapsed ALL to develop new therapeutic strategies. Recently, somatic mutations in NT5C2 were identified as one of the most common genomic lesions specific to relapsed ALL and were directly linked to thiopurine resistance (Nat Genet 2013, Nat Med 2013, Nat Commun 2015). A nucleotidase, NT5C2 dephosphorylates purine monophosphate nucleotides and is important for cellular nucleic acid homeostasis. However, the exact mechanisms by which NT5C2 mutations influence thiopurine resistance in ALL are not completely understood.Methods Wildtype and mutant human NT5C2 (p.R39Q, p.R238W, and p.R367Q) were expressed in E. Coli and purified by affinity chromatography followed by gel filtration. Nucleotidase activity was measured by quantifying the release of free phosphate with the Malachite Green assay. NT5C2 enzyme kinetics (Vmax and Km) were determined for endogenous purine nucleotide substrates (inosine monophosphate [IMP] and guanosine monophosphate [GMP]) and thiopurine drug metabolite substrates (thioinosine monophosphate [TIMP] and thioguanosine monophosphate [TGMP]). Human ALL cell lines (Nalm6 and REH) stably overexpressing wildtype or mutant NT5C2 were established by lentiviral transduction. Thiopurine transport was characterized in vitro by quantifying the influx and efflux of 14 C labeled mercaptopurine (MP) and its metabolites.Results To characterize NT5C2 substrate specificity, we first determined the nucleotidase activity of wildtype and mutant proteins with endogenous purine nucleotides (IMP and GMP) and thiopurine metabolites (TIMP and TGMP). Wildtype NT5C2 exhibited nucleotidase activity against all 4 nucleotide metabolites, but IMP and TIMP were clearly the preferred substrates with up to 4-fold higher reaction velocities compared to GMP and TGMP, respectively. All leukemia-derived NT5C2 mutations resulted in substantial increase in nucleotidase activity across substrates, with the p.R367Q mutant consistently exhibiting the greatest gain of activity. By comparison, mutant NT5C2 proteins showed similar substrate preference of IMP over GMP (e.g., reaction rate of 26.6 pmol/min and 10.4 pmol/min for IMP and GMP, respectively, with the p.R367Q NT5C2). However, this inosine over guanosine bias in substrate specificity was reversed with thiopurine metabolites: leukemia-derived mutant NT5C2s were substantially more efficient to dephosphorylate TGMP than TIMP (e.g., reaction rate of 27.8 pmol/min and 41.5 pmol/min for TIMP and TGMP, respectively, with the p.R367Q NT5C2). The "mutant to wildtype" ratio of NT5C2 nucleotidase activity (e.g., the relative increase of Vmax) was significantly greater for thiopurine metabolites than what was observed when endogenous purine nucleotides were used as substrates (e.g., 15.2- and 1.6-fold increase in Vmax for TIMP and IMP with the p.R367Q NT5C2). These results suggest that ALL relapse-associated NT5C2 mutations likely confer not only increase in endogenous nucleotidase activity but also novel enzymatic functions that specially affect thiopurine metabolism.In addition to the increase in thiopurine resistance, human ALL cell lines expressing mutant NT5C2 also showed significant reduction in intracellular levels of thiopurine metabolites after drug treatment, pointing to influence of NT5C2 mutations on thiopurine drug transport. Using 14 C labeled MP, we traced both the influx and efflux of drug metabolites in isogenic cell lines expressing wildtype or mutant NT5C2 s. In both Nalm6 and REH cells, intracellular uptake of MP was reduced by 51% in mutant cells compared to cells with wildtype protein. In contrast, the level of MP metabolites in the extracellular medium was 3 times higher with mutant cells, reflecting increased drug efflux. Therefore, expression of mutant NT5C2 resulted in a net reduction of intracellular exposure to thiopurine, suggesting disturbance of the drug transport pathway as a distinct mechanism for NT5C2-mediated thiopurine resistance.ConclusionsNT5C2 mutations specifically influence thiopurine metabolism and transport, and therefore are critical determinants for drug resistance in relapsed ALL. DisclosuresNo relevant conflicts of interest to declare.