Resistance to the isocitrate dehydrogenase 1 mutant inhibitor ivosidenib can be overcome by alternative dimer-interface binding inhibitors

Abstract

Ivosidenib, an inhibitor of isocitrate dehydrogenase 1 (IDH1) R132C and R132H variants, is approved for the treatment of acute myeloid leukaemia (AML). Resistance to ivosidenib due to a second site mutation of IDH1 R132C, leading to IDH1 R132C/S280F, has emerged. We describe biochemical, crystallographic, and cellular studies on the IDH1 R132C/S280F and R132H/S280F variants that inform on the mechanism of second-site resistance, which involves both modulation of inhibitor binding at the IDH1 dimer-interface and alteration of kinetic properties, which enable more efficient 2-HG production relative to IDH1 R132C and IDH1 R132H. Importantly, the biochemical and cellular results demonstrate that it should be possible to overcome S280F mediated resistance in AML patients by using alternative inhibitors, including some presently in phase 2 clinical trials.

Natural and synthetic 2-oxoglutarate derivatives are substrates for oncogenic variants of human isocitrate dehydrogenase 1 and 2

“…By contrast, the R140Q IDH2 -value for 2OG is ∼6- and ∼14-fold higher than that for R132H IDH1 and R172K IDH2, respectively, indicating a lower affinity of the R140Q IDH2 variant for 2OG. The R132H IDH1 -value for 2OG is in the range of values previously reported using absorbance assays ( 10 , 53 , 54 ). Comparison of / values implies 2OG is ∼21-fold and ∼9-fold less efficient substrate for R140Q IDH2 than R172K IDH2 and R132H IDH1, respectively ( Table 2 , entry A).…”

“…Crystallization trials were initiated to investigate the binding modes of the 2OG derivatives. Structures with R132H IDH1, R172K IDH2, or R140Q IDH2 were not obtained, however, 3-butyl-2OG ( 4 ) was crystallized in complex with R132C/S280F IDH1 in the presence of catalytically inactive Ca(II), substituting for Mg(II), and NADPH under reported conditions ( 53 ). The structure of the R132C/S280F IDH1:Ca(II): 4 complex was solved by molecular replacement using a R132C/S280F IDH1 structure (PDB ID: 7PJM ( 53 )) as a search model (2.35 Å resolution; C 2 2 2 1 ; Fig.…”

NADP(H)-dependent biocatalysis without adding NADP(H)

“…6 B , corresponding to the maximum turnover frequencies of IDH1 and FNR, respectively, under nanoconfinement. For comparison, solution assays performed under comparable conditions gave k cat values of 86 s −1 for IDH1 ( 41 ) and 51 s −1 for FNR ( SI Appendix , Fig. S10 ).…”

“…Ferredoxin NADP + -reductase (FNR) from Chlamydomonas reinhardtii and human IDH1 (wild-type and R132H) were expressed and purified as previously described ( 8 , 41 ). Nanoporous ITO electrodes were prepared as described previously ( 8 ), and the enzymes were loaded by applying concentrated mixtures (at the specified ratios) to the electrode surface for approximately 45 min before rinsing thoroughly with buffer solution.…”

“…Such a high binding affinity for nicotinamide may appear detrimental for enzymes that use it as an exchangeable cofactor, but it should be borne in mind that a copurified complex represents just one state among many, within or outside the catalytic cycle. Indeed, IDH1 enzymes do not have particularly low K m values for NADP(H) (they lie in the range 10 to 100 µM) ( 20 , 41 ) compared to other NAD(P)(H)-dependent dehydrogenases, suggesting that different conformations may vary in their affinities for NADP(H): accordingly, transient kinetic studies have shown that ~50% of enzyme-bound NADPH is released from wild-type IDH1 after it undergoes a conformational change induced by Mg 2+ and isocitrate binding ( 14 ). The copurified NADP(H) remains bound even when large volume buffer exchanges are used (~12 million-fold dilution)—its release occurring only after Mg 2+ and substrate bind ( 14 ).…”