Radioligand Binding to IDH1 Enzymes Revealed as Method-Dependent in New Study

The radiofluorinated cancer drug ivosidenib, labeled with fluorine-18 and known as [18F]AG-120, binds specifically to both the wild-type and the R132H mutant form of the soluble enzyme isocitrate dehydrogenase 1 at nanomolar radioligand concentrations, according to a new research article published in EJNMMI Radiopharmacy and Chemistry on 22 September 2026. The study, led by Winnie Deuther-Conrad, Thu Hang Lai, Andreas Maurer and Barbara Wenzel, set out to evaluate a suite of established biochemical and biophysical approaches for characterizing radioligand binding to a soluble intracellular protein, a task that poses particular methodological challenges compared with the well-established assay formats available for membrane-associated targets. The central and perhaps most consequential finding is that the apparent affinity of the radioligand for its target depends strongly on the assay format used to measure it, with estimated dissociation constants varying by roughly two- to thirty-fold across the methods investigated.

Isocitrate dehydrogenase 1 is a cytosolic, NADP+-dependent enzyme that catalyzes the oxidative decarboxylation of isocitrate to 2-oxoglutarate, coupling central metabolism to cellular redox homeostasis. Mammalian cells express three IDH isoenzymes: cytosolic IDH1, mitochondrial NADP+-dependent IDH2, and mitochondrial NAD+-dependent IDH3. In glioma, mutations in IDH1 or IDH2 occur in approximately 70 to 80 percent of adult WHO grade 2 and grade 3 diffuse gliomas, with the heterozygous R132H substitution in IDH1 representing the predominant variant. This mutation confers a neomorphic enzymatic activity: instead of its normal role, the mutant enzyme reduces 2-oxoglutarate to the oncometabolite D-2-hydroxyglutarate in an NADPH-dependent reaction. The resulting accumulation of 2-hydroxyglutarate to high intracellular concentrations competitively inhibits multiple 2-oxoglutarate-dependent dioxygenases involved in epigenetic regulation and cell differentiation, contributing to the metabolic and epigenetic alterations characteristic of IDH-mutant tumors.

Paradoxically, IDH-mutant gliomas generally carry a more favorable prognosis than IDH-wildtype gliomas, reflecting distinct tumor biology and greater sensitivity to chemo- and radiotherapy. The high intracellular 2-hydroxyglutarate concentrations also provide a basis for non-invasive assessment of IDH mutation status by proton MR spectroscopy, although definitive determination of IDH status still relies on molecular characterization of tumor tissue, which is essential for accurate diagnosis and treatment stratification. IDH status constitutes a defining molecular feature of distinct adult-type diffuse glioma entities in the current WHO classification and has become therapeutically actionable. This has motivated the development of small-molecule inhibitors targeting mutant IDH1, which predominantly bind allosterically at the interface between the two enzyme subunits of the dimer and suppress the neomorphic production of 2-hydroxyglutarate. Several such inhibitors have entered clinical development, including ivosidenib, the first mutant-IDH1-selective small molecule approved for cancer therapy.

The clinical relevance of IDH mutation status has in turn motivated efforts to image the mutated enzyme directly in the body. Positron emission tomography radioligands targeting mutant IDH1 could, in principle, enable direct visualization of the target protein and provide spatial information on IDH1-mutant tumor tissue. Several radioligand development efforts using structurally different low-molecular-weight mutant IDH1 inhibitors have been reported, and a radiofluorinated analogue of AG-120 was synthesized in the authors’ laboratories and previously evaluated in vitro and in vivo. In that earlier work, [18F]AG-120 showed approximately thirty-fold higher uptake in U251 glioblastoma cells overexpressing IDH1R132H than in cells transfected with wild-type IDH1. Yet when the same cell lines were used to generate glioblastomas in rats, PET imaging revealed only moderately higher accumulation in the IDH1R132H tumors, raising the question of whether differences in the interaction of the radioligand with the two enzyme variants in the cellular system might contribute to this discrepancy.

Adding to the puzzle, prior label-free biophysical and biochemical studies had shown that the functional selectivity of allosteric mutant IDH1 inhibitors does not necessarily reflect preferential binding to the mutant enzyme. Instead, selectivity may arise from differential effects on metal ion and substrate binding in the two variants, and investigations of AG-120 itself reported comparable binding affinities for both isoforms in the high-nanomolar range. Whether this binding profile also held for [18F]AG-120 at tracer concentrations in the cellular system used in the earlier uptake studies had not been established. The authors therefore emphasize that the aim of the present study was not to establish a universally applicable protocol for determining ligand affinity or specificity for IDH1R132H, but to evaluate a set of established approaches using the resources available in their laboratory, employing standard radiopharmaceutical equipment.

The experiments drew on two complementary sources of the enzyme: cytosolic fractions from stably transfected U251 glioblastoma cells overexpressing either wild-type IDH1 or IDH1R132H, and purified recombinant His-tagged versions of both variants immobilized on nickel-coated plates via metal-chelate binding. Radioligand binding was assessed by size-exclusion chromatography, rapid glass fiber filtration, and autoradiography, while thermal shift assays and Western blotting provided additional evidence for target engagement and target identity. The high sensitivity of radioactivity detection allowed binding to be assessed at low, tracer-like radioligand concentrations. All binding measurements were performed in the presence of 10 mM magnesium ions but without added NADP+, NADPH, isocitrate or 2-oxoglutarate, a simplification chosen to avoid additional equilibria associated with cofactor and substrate occupancy, although the inhibitory potency of allosteric inhibitors may depend on the conformational and cofactor state of the enzyme.

The first line of evidence came from thermal shift-Western blot assays on the cytosolic fractions. Incubation with 18 micromolar AG-120 increased the apparent aggregation temperature of both enzyme variants compared with vehicle controls, by 3.6 degrees Celsius for wild-type IDH1 and 9.8 degrees Celsius for IDH1R132H, corresponding to mean values from two independent experiments. The baseline aggregation temperatures measured in the cytosolic fractions, 50.5 plus or minus 2.0 degrees Celsius for the wild-type enzyme and 45.6 plus or minus 1.5 degrees Celsius for the mutant, were comparable to values previously reported for the corresponding purified homodimeric enzymes using fluorescence-based thermal shift analysis, both in absolute terms and in the difference between the two variants. This confirmed that AG-120 engages both IDH1 variants within the cytosol-enriched preparations from the transfected cells.

To link the radioactive signal specifically to the enzyme, the researchers incubated the cytosolic fractions with nanomolar [18F]AG-120, separated the proteins by native polyacrylamide gel electrophoresis, and analyzed the gels by autoradiography followed by Western blotting. In representative experiments the radioactive signal co-localized with the corresponding IDH1 or IDH1R132H band detected by antibody staining, supporting an association of the radioligand with the respective enzyme variant. However, the authors caution that the contribution of co-migrating proteins cannot be excluded, and that co-migration after native PAGE does not rule out binding to an off-target protein or complex with similar electrophoretic mobility. Additional radioactive signals in the cytosolic fractions, absent from the pure radioligand sample, appeared in regions of more slowly migrating proteins and may reflect nonspecific binding to cytosolic proteins other than the target enzyme. The authors therefore describe these results as supportive, but not definitive, evidence for IDH1 as the binding target.

Quantitative estimates of affinity came from homologous competition experiments in which a fixed, low concentration of [18F]AG-120 was competed with increasing concentrations of unlabeled AG-120, with nonspecific binding defined by co-incubation with the pan-IDH1 inhibitor BAY1436032. Size-exclusion chromatography on Sephadex G-25 showed the radioligand co-eluting with protein, and competition reduced the radioactive peak, but reliable inhibition curves could not be generated because substantial displacement occurred only at competitor concentrations roughly one thousand-fold higher than the radioligand. Moreover, ligand depletion was severe: in total binding samples, more than 95 percent of the applied radioligand was protein-bound for IDH1wt fractions and about 70 percent for IDH1R132H fractions, so the free radioligand concentration could not be assumed to equal the total. Apparent KD values derived from this analysis, 541 nanomolar for IDH1wt and 40 nanomolar for IDH1R132H, are therefore unreliable, although the data suggested a greater proportion of competitor-displaceable binding in the mutant preparations, roughly 80 percent versus about 30 percent.

Rapid glass fiber filtration, using polyethyleneimine-preincubated filters to retain soluble proteins, reduced ligand depletion so that the bound fraction remained below 10 percent of the applied radioligand, making this the most robust quantitative approach in the study. Homologous inhibition curves from these experiments yielded apparent KD values of 19.9 nanomolar for IDH1wt-U251 and 9.37 nanomolar for IDH1R132H-U251 cytosolic fractions, from two independent experiments per enzyme. A control assay using cytosol from parental U251 cells expressing endogenous IDH1 at physiological levels showed no measurable reduction in radioligand binding with BAY1436032 or AG-120, although low-level specific binding to endogenous IDH1 could not be excluded. With immobilized recombinant His-tagged enzymes on nickel-coated plates, apparent KD values for IDH1R132H ranged from approximately 50 to 200 nanomolar across four experiments and varied between protein batches, while a single evaluable experiment for IDH1 gave 182 nanomolar. Across all formats, apparent affinities were consistently higher, and the specific binding fraction larger, for the mutant enzyme, but the limited number of independent experiments precludes a robust comparison, and the authors conclude that the resulting values should be regarded as method-dependent apparent affinities rather than definitive measures of intrinsic ligand affinity. They also note that these binding differences alone appear insufficient to explain the earlier discrepancy between cellular uptake and tumor accumulation, pointing to factors such as tumor perfusion or the tumor microenvironment as possible contributors that were not investigated here.

Subject of Research: Binding of the radioligand [18F]AG-120 to wild-type and mutant isocitrate dehydrogenase 1 assessed by multiple biochemical and biophysical assay formats

Article Title: Methods for assessing binding of a small-molecule radioligand to soluble enzyme isocitrate dehydrogenase 1 variants

Article References: Deuther-Conrad, W., Lai, T. H., Maurer, A., & Wenzel, B. (2026). Methods for assessing binding of a small-molecule radioligand to soluble enzyme isocitrate dehydrogenase 1 variants. EJNMMI Radiopharmacy and Chemistry, 11(1), Article 70. https://doi.org/10.1186/s41181-026-00502-7

Image Credits: AI Generated

DOI: 10.1186/s41181-026-00502-7

Keywords: IDH1, IDH1R132H, ivosidenib, AG-120, [18F]AG-120, radioligand binding, PET tracer, glioma, oncometabolite, 2-hydroxyglutarate, thermal shift assay, size-exclusion chromatography