Origin of Elevated S-Glutathionylated GAPDH in Chronic Neurodegenerative Diseases

Abstract

H₂O₂-oxidized glyceraldehyde-3-phosphate dehydrogenase (GAPDH) catalytic cysteine residues (C_c(SH) undergo rapid S-glutathionylation. Restoration of the enzyme activity is accomplished by thiol/disulfide S_N2 displacement (directly or enzymatically) forming glutathione disulfide (G(SS)G) and active enzyme, a process that should be facile as C_c(SH) reside on the subunit surface. As S-glutathionylated GAPDH accumulates following ischemic and/or oxidative stress, in vitro/silico approaches have been employed to address this paradox. C_c(SH) residues were selectively oxidized and S-glutathionylated. Kinetics of GAPDH dehydrogenase recovery demonstrated that glutathione is an ineffective reactivator of S-glutathionylated GAPDH compared to dithiothreitol. Molecular dynamic simulations (MDS) demonstrated strong binding interactions between local residues and S-glutathione. A second glutathione was accommodated for thiol/disulfide exchange forming a tightly bound glutathione disulfide G(SS)G. The proximal sulfur centers of G(SS)G and C_c(SH) remained within covalent bonding distance for thiol/disulfide exchange resonance. Both these factors predict inhibition of dissociation of G(SS)G, which was verified by biochemical analysis. MDS also revealed that both S-glutathionylation and bound G(SS)G significantly perturbed subunit secondary structure particularly within the S-loop, region which interacts with other cellular proteins and mediates NAD(P)⁺ binding specificity. Our data provides a molecular rationale for how oxidative stress elevates S-glutathionylated GAPDH in neurodegenerative diseases and implicates novel targets for therapeutic intervention.

Keywords: glutathione; glyceraldehyde-3-phosphate dehydrogenase; hydrogen peroxide; molecular dynamic simulation; neurodegenerative disease; oxidative stress; redox signaling.

Figure 1

(A) The gluconeogenic enzyme GAPDH assay standard curve data generated by following the decrease in NADH absorption over a 60 min time window in the assay coupled to 3-PGA kinase phosphorylation of 3-PGA. The absorption decrease in the absence of GAPDH (blank) was first fitted using linear regression. The interpolated values at each timepoint were added to all concentrations of GAPDH data points from the standard curve and the samples to be measured. (B) First, all data sets were corrected for the time-dependent decrease in absorption of the blanks. The data were fitted to a monoexponential decay measured on three separate days. The rate constants calculated from the data in panel (A) are directly proportional to the amount of enzyme in the assay in the three experiments, and the slopes are not significantly different. (C) As a test of the validity of the data fit model, the 60 min data were truncated in steps of 10 min and the results each successive truncation plotted. (D) The SEM’s for the rate constants for each truncation obtained from the exponential fit successively declined from a maximum at 10 min to a minimum at 60 min, demonstrating that the data modeling is robust. The value of the exponential rate constant is independent of the assay start time, convenient for manual multiwell microtiter plate kinetic assays. (E) The four thiols used in this assay (DTT, BME, G(SH), and Cysteine) are examined for interference in the GAPDH assay. BME interference is observed and noted at concentrations used in the study. (F) IOB oxidized GAPDH is reactivated by DTT at various concentrations. Reactivation by 500 μM DTT does not significantly differ from full activation at 1 mM used to assess the % recovery of oxidized GAPDH under various experimental conditions.

Figure 2

(A) Dose response for reactivation of S-glutathionylated GAPDH by G(SH) in the absence (red bars) and presence of 1 mM DTT (blue bars). (B) Time course of reactivation of S-glutathionylated GAPDH by 1 mM G(SH) in the absence (red bars) and presence of 1 mM DTT (blue bars). (C) Dose response of reactivation of S-cysteinylated GAPDH by cysteine in the absence (red bars) and presence of 1 mM DTT (blue bars). (D) Dose response of reactivation of S-mercaptoethanolylated GAPDH by either BME in the absence (red bars) and DTT (blue bars). The data (Mean ± SD, n = 3) are from representative experiments. The green bars in Panels A, B and C represent incubation without G(SH) in the presence of 1 mM DTT.

Figure 3

MDS ligand interactions within and around the active site of an isolated subunit of solvated h-GAPDH (1u8F) represented in two dimensions. (A) A catalytic cysteine residue (C152(SH) GAPDH subunit is converted to C_c(SOH), the initial step of the oxidation of GAPDH by H₂O₂ and a molecule of G(SH) is added to the model. After MDS, the G(SH)-C152(SH) sulfur-sulfur are separated by 4.4 Å, within range of the disulfide bond formation. The H-bond donor and acceptor network are shown in green and blue dotted lines, and the bond lengths are obviously distorted by a 2D rendering. The atomic bond distances and interaction energies are tabulated in Supplementary Tables S1 and S2. The solid black line delineates the van der Walls molecular cross section of G(SH). (B) S_N2 nucleophilic attack by G(SH) on sulfenic acid is simulated by the formation of the disulfide bond in MOE, forming C152S-glutathione. (C) A second molecule of G(SH) is placed within the active site pocket, and after MDS it is docked in an antiparallel alignment with C152 S-glutathione. S_N2 nucleophilic attack by G(SH) on the S-glutathionylated structure forms a docked G(SS)G ligand and reducesC152(SH) with their nearest sulfur-sulfur distance of 3.56 Å, such that the three sulfurs atoms can undergo thiol-disulfide exchange resonance.

Figure 4

Distribution of S-glutathionylated and bound G(SH) and G(SS)G as a function of time of incubation with 1 mM G(SH). G(SH) and G(SS)G were measured using the Promega oxidized/reduced protocol and reagent kit. (a ) Time course of the decline in total mol of S-glutathionylated subunits GAPDH/mol tetramer measured following the removal of all unbound G(SH) and G(SS)G after subunit denaturation and washing in the retentate following Microcon® spin separation (see text for details). (b ) Time course of unbound reduced G(SH) recovered and measured in the eluate following Microcon® spin separation. (c ) Time course of unbound G(SS)G recovered and measured in the eluate following Microcon® spin separation. Note: the data represents 2 mol equivalents of G(SH) derived from 1 mol equivalent of G(SS)G in the Promega protocol). (d ) The total number of G(SH) equivalents bound to GAPDH over the time course of the incubation of S-glutathionylated GAPDH with 1 mM G(SH) is shown as the sum of measurements. (a), (b), (c) The associated cumulative SD of the triplicate samples. The black dashed line represents the theoretical maximal G(SH) binding capacity of the four active sites within the GAPDH tetramer. The data show the combined mean values ± SD from two separate experiments with technical triplicates.

Figure 5

The protein sequence alignment and structure superposition tool in MOE is used to align the native h-GAPDH subunit structure with the output from MDS, and the energy minimizes subunit structures for C152S-glutathionylated subunit (Panel [A]) and the native h-GAPDH subunit structure with the output from MDS and the energy minimizes subunit structure for G(SS)G subunit docked within the active site (Panel [B]). The figure shows a truncated version of the full 335-amino acid sequence data shown in Supplementary Figures S2 and S3, to emphasize the region of greatest perturbation and includes all residues in the S-loop region of the subunit (residues 180–203). The root mean square distance (RMSD) for each alignment column (i.e., residue pair) used during the superposition. The RMSD value is represented by a black vertical bar above the pairwise aligned sequences. Closely matching RMSD values are highlighted by a horizontal line cutoff below 2.0 Å, while a larger RMSD excursion (5–13 Å) indicates poor atomic coordinate superposition. The linearized secondary structural features the following: Red α-helix; yellow β-sheet; and blue 2-5 residue turns; while no color is random coil for the three structure superpositions shown below Panel [B].

Figure 6

Ribbon diagram of the superimposed subunit structures of GAPDH (A) native, (B) S-glutathionylated, and (C) a subunit with tightly bound G(SS)G in the active site. The figure shows the clear transposition of the S-loop region (residues A180–L203) using of the two modified subunits compared to the native enzyme. The indole side chain of tryptophane W196 is shown as a visual marker for the residue displacement (native subunit is red, S-glutathionylated subunit is green, subunit with G(SS)G bound is black). The short α-helix (W196–R200) in the native subunit is converted to coil and turns in the modified subunits. The S-loop region is of interest because it is a site of protein binding and also mediates NAD(P)⁺ cofactor binding selectivity.