Correcting glucose-6-phosphate dehydrogenase deficiency with a small-molecule activator

Abstract

Glucose-6-phosphate dehydrogenase (G6PD) deficiency, one of the most common human genetic enzymopathies, is caused by over 160 different point mutations and contributes to the severity of many acute and chronic diseases associated with oxidative stress, including hemolytic anemia and bilirubin-induced neurological damage particularly in newborns. As no medications are available to treat G6PD deficiency, here we seek to identify a small molecule that corrects it. Crystallographic study and mutagenesis analysis identify the structural and functional defect of one common mutant (Canton, R459L). Using high-throughput screening, we subsequently identify AG1, a small molecule that increases the activity of the wild-type, the Canton mutant and several other common G6PD mutants. AG1 reduces oxidative stress in cells and zebrafish. Furthermore, AG1 decreases chloroquine- or diamide-induced oxidative stress in human erythrocytes. Our study suggests that a pharmacological agent, of which AG1 may be a lead, will likely alleviate the challenges associated with G6PD deficiency.

Fig. 1

Canton G6PD (R459L) variant is biochemically different from WT G6PD. a Enzymatic scheme of G6PD activity. b A linear map of G6PD domain structure with most common variants indicated. c Catalytic activity of recombinant WT G6PD and Canton G6PD enzymes with kinetic parameters (n = 5, ^****p < 0.0001). d Thermostability of WT G6PD and Canton G6PD enzyme (n = 3, ^**p = 0.003). e G6PD protein levels and residual G6PD activity (normalized to NT (no treatment) of each enzyme) after incubation with chymotrypsin for 1 h (n = 3 for protein level assay, ^**p = 0.0046; n = 2 for enzyme assay, ^*p = 0.024). f Protein stability assessment with cycloheximide treatment (50 μg mL⁻¹), blocking de novo protein biosynthesis, in lymphocytes derived from corresponding subjects (n = 3, ^*p = 0.013). Protein levels were normalized to the level of each enzyme at 0 h (no treatment). g G6PD activity was lower in cell lysates with Canton variant (n = 4, ^****p < 0.0001). h, i, j Lymphocytes with Canton variant generated less GSH and more reactive oxygen species (ROS) and were less viable (n = 4, (n = 3 for Fig. 1i), ^*p < 0.05, ^**p < 0.01). Error bars represent mean ± SEM. Statistical differences were calculated by two-tailed unpaired Student’s t-test. NT: no treatment, WT: wild-type, Chy: chymotrypsin

Fig. 2

Canton mutation (R459L) loses essential inter–helical interactions. a Structural overlay of WT G6PD (green) and Canton variant (orange). Structural NADP⁺ is shown as spheres, and arrows and circles indicate G6P and catalytic NADP⁺-binding sites (G6P and catalytic NADP⁺ were not observed in our structures). b (Left) Inter-helical interactions through R459 on the αn helix in WT G6PD and side chains of D181 and N185 on the adjacent helix (αe). (Right) Canton mutation loses such interactions, leading to displacements of the helix (αe) and a loop containing K171, P172, F173, G174 and R175 that precedes the helix. c, d Mutations of R459-interacting residues on the αe helix showed Canton mutation-like activity and thermostability (n = 3, ^****p < 0.0001, one-way ANOVA) and were also susceptible to chymotrypsin treatment (n = 4 for D181A, ^****p < 0.0001; n = 3 for N185A, ^*p = 0.026, two-tailed unpaired Student’s t-test). Error bars represent mean ± SEM. NT no treatment; WT: wild-type; Chy: chymotrypsin

Fig. 3

AG1 (activator of G6PD) induces biochemical changes in the Canton variant. a Increased activity of Canton G6PD enzyme by AG1 (n = 5, ^***p = 0.0002, two-tailed unpaired Student’s t-test) and b a dose response curve of AG1. c AG1 changed kinetic parameters of Canton G6PD. d AG1 promoted dimerization of Canton G6PD (n = 3). e Size-exclusion FPLC (calibrated Superdex 75 10/300 GL column) profile of purified Canton G6PD in the presence of G6P or AG1. f AG1 reduced proteolytic susceptibility of Canton G6PD (n = 3, ^***p < 0.001, ^*p < 0.05, one-way ANOVA). The protein levels were normalized to the non-treated (NT) enzyme level. g Cycloheximide-chase assay using lymphocytes carrying the Canton variant (n = 3). Protein levels were normalized to the level of each enzyme at time 0 h. h, i, j AG1 increased G6PD activity in cell lysates with the Canton variant (n = 4, ^**p = 0.0032, two-tailed unpaired Student’s t-test), mildly enhanced a GSH level (n = 7, ^*p = 0.0282, two-tailed unpaired Student’s t-test) and reduced a ROS level in culture (n = 6, ^*p = 0.0452, two-tailed unpaired Student’s t-test). k AG1 increased viability of lymphocytes carrying the Canton variant (n = 6, ^**p = 0.003, two-tailed unpaired Student’s t-test). l, m AG1 activated other major G6PD variants, including A⁻ (V68M, N126D; blue spheres), Mediterranean (S188F, orange spheres), and Kaiping (R463H, yellow spheres) variants, respectively (n = 4, ^****p < 0.0001, ^**p < 0.01, ^*p = 0.011, one-way ANOVA), and promoted their dimerization (n = 3). Purple spheres in the structure represent the side chain of R459. n Cycloheximide-chase assay using fibroblasts carrying the Mediterranean variant (n = 4, ^*p = 0.0437, two-tailed unpaired Student’s t-test). o, p AG1 significantly decreased a ROS level (n = 6, ^***p = 0.0001, ^****p < 0.0001, one-way ANOVA) and increased a GSH level in those cultures (n = 3, ^*p = 0.0214, ^****p < 0.0001, one-way ANOVA). 100 μM and 1 μM of AG1 were used for in vitro assays and cell-based assays, respectively. 5% DMSO (stock) was used as vehicle. For FPLC assay, 500 μM AG1, 200 μg of Canton G6PD recombinant enzyme and 10 mM G6P were used. Cells were subjected to serum starvation for 48 h. Error bars represent mean ± SEM. MW: molecular weight, FPLC: fast protein liquid chromatography, NT: no treatment, Veh: vehicle, WT: wild-type, Med: Mediterranean fibroblast

Fig. 4

AG1 attenuates ROS-induced pericardial edema in a G6PD-dependent manner. a Embryos were treated at 24 hpf with 1 μM AG1 with and without chloroquine (CQ; 100 μg mL⁻¹) and scored at 32 hpf. Representative phenotypic images of pericardial edema and pooling (magenta arrows) are provided on the left (scale bar: 300 μm). Embryo orientation is lateral view, anterior left. Raw counts used for chi-square analysis and calculated p value are included in table below. b ROS levels in individual WT embryos from three independent clutches. Embryos were treated at 24 hpf for 5 h before ROS measurement. Error bars represent mean ± SD (^***p < 0.001, ns = not statistically significant, p > 0.99, Kruskal–Wallis multiple comparison test, adjusted p value using Dunn’s test). c G6PD activity and NADPH levels were measured using the lysates of pooled embryos (from two independent clutches). Error bars represent mean ± SEM (^***p = 0.0003, two-tailed unpaired Student’s t-test). d Embryos were injected with either sgRNA targeting exon 10 of g6pd (Guide alone) or sgRNA + Cas9 protein (Guide + Cas9, G6PD KO (knockout)) to generate G6PD F0 crispants. Representative phenotypic images of pericardial edema and pooling (magenta arrows) are provided on the left (scale bar: 300 μm). Treatment conditions are the same as in a. Raw counts used for chi-square analysis and calculated p value are included in table below. e ROS levels in individual embryos with sgRNA or sgRNA + Cas9 (G6PD KO) protein injection. Treatment conditions and the statistics are the same as in b (^*p = 0.0267, ^**p < 0.01, ^****p < 0.0001, ns = not statistically significant). f G6PD activity and NADPH levels were measured with lysates of pooled embryos from three independent experiments. Error bars represent mean ± SEM of the replicate measurements (^*p < 0.05, ^**p < 0.01, one-way ANOVA for G6PD activity measurement and two-tailed unpaired Student’s t-test for NADPH measurement). Veh: vehicle, KO: knockout, CQ: chloroquine

Fig. 5

AG1 reduces hemolysis upon exposure to oxidative stressors. a AG1 reduced the extent of hemolysis of 5% erythrocyte suspension treated with either 1 mM chloroquine (CQ) for 4 h under light or 1 mM diamide for 4 h at 37 °C (n = 7 independent blood samples, ^*p = 0.0372, ^**p = 0.0019, ^****p < 0.0001, one-way ANOVA). b–d AG1 significantly increased GSH levels and reduced ROS levels, when 5% erythrocyte suspension was exposed to either 1 mM chloroquine or diamide for 3 h at 37 °C, which was consistent with increased G6PD activity (n = 9–11, ^*p < 0.05, ^**p < 0.01, ^****p < 0.0001, one-way ANOVA). e Band 3 protein was clustered (cBand3) when 5% erythrocyte suspension was treated with chloroquine, which was alleviated by AG1 treatment. Each lane represents one individual sample, and quantification is provided in Supplementary Fig. 7a. f–h AG1 (1 μM) improved storage of erythrocytes (5% suspension) at refrigerated temperature by reducing hemolysis and concomitant protein leakage for 28 days (n = 13 independent blood samples, ^*p < 0.05, two-tailed unpaired Student’s t-test), which corresponded with increased G6PD activity (n = 4, ^*p = 0.0323, ^***p = 0.0003, one-way ANOVA). Each sample was re-treated with AG1 every week. Representative hemolysis phenotypic images are provided below f. Error bars represent mean ± SD. NT: no treatment, CQ: chloroquine, cBand3: clustered band 3 protein