Macromolecular crowding and bicarbonate enhance the hydrogen peroxide-induced inactivation of glyceraldehyde-3-phosphate dehydrogenase

Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) is an asymmetric homotetrameric NAD⁺-dependent enzyme that catalyzes the oxidative phosphorylation of glyceraldehyde-3-phosphate (G3P) to 1,3-biphosphoglycerate, with concurrent formation of NADH. Together with the glycolytic role of GAPDH, this abundant cytosolic enzyme (c.f. estimated intracellular levels are low micromolar but can reach up to 75 µM in skeletal muscle cells [1]) has other non-glycolytic (moonlighting) activities in different cell organelles, affecting, amongst others, mRNA stability, cell trafficking and gene regulation [2]. Recent attention has focused on the role of GAPDH as a reversible metabolic switch in cells exposed to elevated (supraphysiological) steady-state levels of oxidants (e.g. cancer cells, or cells exposed to oxidative insults) [3,4]. GAPDH is particularly sensitive to oxidative post-translational modification by a number of oxidants including the two-electron oxidant, and signaling molecule hydrogen peroxide (H₂O₂) [5]. This inactivation arises from modification of the active site cysteine (Cys, position 152 in the human sequence, position 150 in the rabbit isoform), though alteration of other residues, such as a nearby Cys (Cys₁₅₄ in the human isoform), can also modulate its enzymatic activity [6,7]. Inactivation of GAPDH modulates its function (e.g. switching from a dehydrogenase to an acyl phosphatase activity [8]) and triggers the re-routing of the carbon flux from glycolysis to the oxidative phase of the pentose phosphate pathway (oxPPP). The oxPPP pathway is essential for the reduction of NADP⁺ to NADPH which is an essential reducing cofactor for the glutathione reductase, glutaredoxin and thioredoxin enzyme systems that play a key role in cellular defenses against oxidative stress [3]. Therefore, rapid and efficient H₂O₂-induced modification of GAPDH is essential for cellular signaling of cellular stress, and metabolic adaptation to these conditions.

In the light of these data, we hypothesized that a combination of both macromolecular crowding and the presence of bicarbonate, might have additive or synergistic effects on H₂O₂-induced inactivation of GAPDH. This hypothesis was tested in studies using rabbit muscle GAPDH (which has 95% sequence homology with the human isoform and a conserved catalytic site [18]), with this exposed to physiologically-relevant H₂O₂ doses in the absence or presence of different crowding agents (both small molecule and high molecular mass polymers with well-known physical and chemical properties [19]), and also in the absence or presence of 25 mM bicarbonate (and therefore physiological levels of CO₂).

Initial studies examined the effects of macromolecular crowding alone on H₂O₂-induced inactivation of GAPDH, with these data compared with dilute solutions (i.e. 100 mM phosphate buffer solution containing 0.1 mM DTPA). The effect of H₂O₂ on GAPDH activity (0.8 µM monomer units) was determined after the pre-incubation of the enzyme in the absence or presence of 0.35, 0.7, 1.4 and 2.8 µM H₂O₂ for 3 h at 37°C (oxidant-to-protein molar excesses of ∼0.4 ∼0.9, ∼1.8 and ∼3.5 respectively; concentrations and molar ratios likely to reflect the physiological to pathological range). A steady increase in absorbance at 340 nm, due to reduction of NAD⁺ to NADH), was observed over time on incubation with 0.2 mM G3P and 0.5 mM NAD⁺ (Figure 1). A similar kinetic profile was observed for GAPDH pre-incubated with 0.7 µM H₂O₂ under the same conditions. In contrast, pre-incubation with 2.8 µM H₂O₂ resulted in a significant decrease in GAPDH activity. This effect was confirmed by determining the initial slopes of the kinetic profiles, from which the GAPDH activity was determined relative to the controls (vide infra).

A previous study has reported that macromolecular crowding does not alter the specific activity of GAPDH, and only the monomer:dimer:tetramer equilibrium [20]. To confirm these activity data, we determined the effect of the crowding agent dextran 35 on enzyme activity. The presence of 60 mg ml⁻¹ dextran 35 (at the time point at which activity was determined) did not alter the time course of NAD⁺ reduction, when compared with data obtained under dilute conditions (Figure 2A). This was confirmed for GAPDH samples pre-incubated with 0.7 and 2.8 µM H₂O₂ under dilute conditions, with the activity measurements recorded in the absence or the presence of 60 mg ml⁻¹ dextran (Figure 2B). However, when the oxidation of GAPDH was carried out under crowded conditions (i.e. in the presence of 120 mg ml⁻¹ dextran 35), an enhanced inactivation of the enzyme was determined (Figure 2B). Incubation of GAPDH with 0.7 µM H₂O₂ under dilute conditions resulted in a loss of ∼7% of the activity compared with the activity of the native, non-oxidized, enzyme, whilst incubation with 0.7 µM H₂O₂ in the presence of 120 mg ml⁻¹ dextran 35 resulted in a ∼16% decrease in activity (Figure 2B,C). The effect of dextran 35 on H₂O₂-mediated inactivation of GAPDH was more pronounced with a 3.5-fold molar excess of H₂O₂ over protein concentration. A residual activity of ∼82 ± 1.4% (i.e. ∼18% loss of the activity) was determined under dilute conditions, whereas a residual activity of 50 ± 1.4% was determined for GAPDH samples incubated with 2.8 µM H₂O₂ under crowded conditions (Figure 2C).

To investigate whether this effect was specific to this linear polymer, or was a general effect of macromolecular crowding, similar experiments were carried out with the branched polymer Ficoll 70. The presence of Ficoll during the measurement of GAPDH activity did not alter the kinetic profiles (data not shown). However, oxidation of GAPDH with 2.8 µM H₂O₂ in the presence of 120 mg ml⁻¹ Ficoll 70 resulted in a significantly greater loss of activity than under dilute conditions (Figure 3). A residual activity of 50 ± 9.1% was determined under these conditions, with this data showing no statistical differences when compared with the data obtained with dextran 35 under identical conditions, suggesting that this is a general crowding effect of these macromolecules, rather than a structure-dependent activity.

The effect of small molecules, which increase solution viscosity, but which do not have the space filling (exclusion phenomenon) effect of macromolecules such as dextran or Ficoll was also examined. Initial experiments with glycerol (120 mg ml⁻¹) induced a complete loss of GAPDH activity when added to the measurement buffer (data not shown). This is ascribed to a protein unfolding activity of this compound. In contrast, the presence of 120 mg ml⁻¹ sucrose during activity measurements did not induce any significant effects on enzyme activity relative to dilute conditions (data not shown). In contrast with the data obtained with the macromolecule crowding agents, incubation of GAPDH with increasing concentrations of H₂O₂ in the presence of 120 mg ml⁻¹ sucrose did not enhance oxidant-induced inactivation of GAPDH (Figure 4).

Bicarbonate has been reported previously to modulate H₂O₂-induced GAPDH inactivation [10]. This was confirmed under our experimental conditions. Due to the slow rate of reaction of H₂O₂ with CO₂ (second-order rate constant, k, ∼0.02 M⁻¹ s [21]) to give H₂CO₄/HCO₄⁻ (pK_a ∼3.4 [22]), H₂O₂ was pre-incubated with 25 mM sodium bicarbonate in 100 mM phosphate buffer (pH 7.4) for >7 min before addition of the enzyme. An enhancement in H₂O₂-induced inactivation was observed in the presence of 25 mM bicarbonate (Figure 5A,). Thus, incubation of 0.8 µM GAPDH with 1.4 µM H₂O₂ in the absence or presence of 25 mM bicarbonate resulted in an activity decrease of ∼4% and ∼15%, respectively. This difference was enhanced at higher concentrations of H₂O₂ (2.8 µM) with a decrease in GAPDH activity of ∼18% and 28.5% detected in the absence and presence of 25 mM bicarbonate, respectively (Figure 5A).

As both macromolecular crowding and bicarbonate appear individually to contribute to the inactivation of GAPDH induced by H₂O₂ (see above), further studies were carried out to determine whether these effects were alternative, additive or synergistic. Figure 5B shows the GAPDH activity (as a %) on incubation with increasing doses of H₂O₂ in the presence of 120 mg ml⁻¹ dextran 35 in the absence, or presence, of 25 mM bicarbonate. A greater inactivation of the enzyme was detected with the presence of both bicarbonate and dextran in the reaction system. Thus, an activity of ∼50 ± 1.4% was determined for GAPDH incubated with 2.8 µM H₂O₂ in the presence of 120 mg ml⁻¹ dextran 35, and an activity of ∼31 ± 0.8% was determined for system containing both 120 mg ml⁻¹ and 25 mM bicarbonate during GAPDH exposure to H₂O₂ (Figure 5C). Incubation of GAPDH with 2.8 µM H₂O₂ in the presence of Ficoll 70 had a similar effect, with an enzyme activity of ∼50 ± 9.1% with only the crowding agent present (Figure 3) and a residual activity of ∼23 ± 4.2% with both Ficoll 70 and bicarbonate present (Figure 5C).

GAPDH plays an important role in the metabolic adaptation of cells undergoing oxidative stress [3], as GAPDH inactivation re-routes the carbon flux from glycolysis to the oxidative branch of the pentose phosphate pathway, thereby enhancing the production of reducing equivalents (in the form of NAPDH) and synthetic intermediates that provide tolerance of damage by way of an enhanced capacity of antioxidant defense and repair systems [23]. However, an open question is to how GAPDH is efficiently inactivated in the crowded cytosol of cells where the total amount of proteins ranges between 2 and 4 million proteins per µm³ [24], and the average concentration of intracellular proteins thiols is 10–40 mM [25]. In particular, this enzyme would not be expected, based on prior kinetic data (see Introduction), to be readily inactivated due to the slow rate of reaction of H₂O₂ with GAPDH. Different hypotheses have been proposed to rationalize this conundrum, and explain how reaction with H₂O₂ (or other oxidants) might enable its activity as a redox switch. Redox relays have been proposed as a possible explanation (i.e. inactivation of GAPDH via the intervention of another protein with which H₂O₂ reacts much more rapidly). Such processes have been identified for a number of other proteins that act as ‘redox switches’ (e.g. oxidation of STAT3 via oxidized peroxiredoxin-2, thioredoxin/ASK1 by oxidized peroxiredoxin-1 and others; [26,27]), but GAPDH has not been shown (so far) to be selectively oxidized via this type of process. However, a recent study has reported that inactivation of GAPDH can be mediated by superoxide dismutase (SOD1) [28]. This is believed to occur via a proximity-model (i.e. due to interactions between the enzymes) in which the generation of localized H₂O₂ by SOD1 may induce efficient oxidation of GAPDH. In addition, it has been proposed that the composition of the reaction media may be of relevance as bicarbonate, an important buffering agent in the cell cytosol, can modulate the reactivity of H₂O₂, via the formation of the more reactive adduct species peroxymonocarbonate, $HCO4−$ [21]. Consistent with this proposal, inactivation of GAPDH by H₂O₂ has been shown to be enhanced in the presence of bicarbonate, via a mechanism where GAPDH facilitates in situ formation of $HCO4−$ [10].

We have recently reported that macromolecular crowding, an intrinsic property of biological systems, can modulate the mechanism(s) of reaction of H₂O₂ with GAPDH, with the data obtained consistent with a greater extent of hyperoxidation of the active site Cys residue under crowded, compared with dilute conditions [17]. In the light of these data, we investigated the effects of macromolecular crowding, bicarbonate, and a mixture of these (which is likely to most accurately reflect the in vivo situation) on the activity of GAPDH.

Low concentrations of H₂O₂ (0.8 µM GAPDH with 0.7 µM H₂O₂) under dilute conditions had no significant effects on GAPDH activity, in agreement with previous reports [10,29] (Figure 1). Higher concentrations of H₂O₂ (2.8 µM H₂O₂ corresponding to a 3.5-fold molar excess over the protein concentration), gave modest, but significant, extents of inactivation, consistent with the established kinetic data. This range of H₂O₂ concentrations was then used to determine whether macromolecular crowding and/or bicarbonate enhance inactivation. Dextran 35 (final concentration 60 mg ml⁻¹ when the activity of the enzyme was examined) did not alter the kinetics of reduction of NAD⁺ by GAPDH (as measured by the absorbance increase at 340 nm; Figure 2A) with this data being in agreement with previous data suggesting that macromolecular crowding does not alter the specific activity of GAPDH [20]. This lack of effect of dextran 35 at the moment of determining the enzyme activity was also observed for samples of GAPDH previously oxidized under dilute conditions (Figure 2B). Thus, no differences in residual activity (% values compared with controls) were observed for samples oxidized under dilute conditions, when the activity measurements were determined in the absence or presence of the crowding agent. However, when the exposure to H₂O₂ was carried out under crowded conditions (i.e. in the presence of 120 mg ml⁻¹ dextran 35) an enhanced loss of the activity was detected (Figure 2B,C). Similar behavior was observed across all the H₂O₂ concentrations tested (0.35–2.8 µM). These novel data indicate that crowding modulates both the oxidation mechanism (as reported in [17]), and also enzyme activity. This effect of macromolecular crowding can be rationalized by increased interaction between small molecules (in this case H₂O₂) and the target protein [30]. It has been reported that besides an increased susceptibility of the active site Cys (Cys150 in the rabbit isoform) to oxidation, the catalytic site of GAPDH possesses a binding site for H₂O₂ in the shallow cavity formed by this Cys and neighboring proton-donating residues [7]. Thus, macromolecular crowding may contribute to interactions that decrease the activation energy by stabilizing the transition state formed during the (two-electron nucleophilic) attack of the thiol of the catalytic Cys on H₂O₂, thereby increasing the rate of this reaction.

Similar experiments with the branched polymer Ficoll 70 (Mw ∼70 000 g mol⁻¹), which has similar chemical properties to (linear) dextran 35, with both being polyols, demonstrated that this enhanced inactivation is not particular to dextran (Figure 3). This general phenomenon appears to arise from the excluded volume phenomenon as high concentrations of sucrose (Mw 342 g mol⁻¹), which enhances solution viscosity but would not otherwise be expected to impede interactions in the same manner as a large polymer, had minimal effects on H₂O₂-induced GAPDH inactivation (Figure 4). However, it should also be considered that the large steric bulk of dextran 35 or Ficoll 70 may alter GAPDH structure dynamics, and may enhance weak attractive forces. Similar findings have been reported for complexes formed between bovine serum albumin and poly(diallyldimethylammonium chloride) where the crowding agent polyethylene glycol (Mw ∼8000 g mol⁻¹) enhanced complex formation, whilst sucrose did not [31].

The current experiments have confirmed that the presence of 25 mM bicarbonate, under the conditions employed, enhances H₂O₂-induced GAPDH inactivation under dilute conditions (absence of crowding agent; Figure 5A). The additional presence of dextran 35, to induce macromolecular crowding, resulted in a greater loss of enzyme activity when compared with the absence of bicarbonate (Figure 5B,C). Thus, after incubation of GAPDH (0.8 µM) with 2.8 µM H₂O₂ residual activity levels of ∼50% and ∼30% were determined under macromolecular crowding conditions in the absence and the presence of 25 mM bicarbonate, respectively (Figure 5C). Similar data was obtained when Ficoll was used as the crowding agent.

Despite these novel observations, a number of important, but as yet unresolved, questions remain to be explored including: (1) whether (and how) macromolecular crowding modulates the in situ formation of $HCO4−$ by GAPDH, as suggested by Winterbourn et al. [10]; (2) whether macromolecular crowding modulates the effects of both the GAPDH substrate (G3P) and cofactor (NAD⁺) on H₂O₂-induced inactivation, and particularly in the presence of bicarbonate; and (3) whether macromolecular crowding and bicarbonate alter the oxidation pathways of GAPDH [17].

The data presented here demonstrate that both macromolecular crowding and buffer composition enhance H₂O₂-induced GAPDH inactivation, with the combination of these factors resulting in additive effects. These data may help explain how GAPDH is efficiently modified and inactivated in the cytosol of cells under conditions of oxidative stress, with these changes being consistent with its role as a metabolic switch under (patho)physiological conditions [3]. These data suggest that the second-order rate constant reported for reaction of H₂O₂ with GAPDH, determined under dilute conditions, may underestimate the rate constant for enzyme oxidation under biological conditions, by a considerable margin. Further experiments appear to be warranted to determine the magnitude of this difference in rate constants, though such measurements are likely to be complicated by the non-homogeneous effects that occur under crowded conditions, which may hinder traditional kinetic analyses.

GAPDH (rabbit muscle), catalase (bovine liver), G3P, NAD⁺, H₂O₂, dextran 35 (average molecular mass 35 000–45 000 g mol⁻¹), sucrose, glycerol, diethylenetriaminepentaacetic acid (DTPA) and other biochemicals were from Sigma–Aldrich (Søborg, Denmark). Ficoll® PM70 (average molecular mass ∼70 000 g mol⁻¹) was purchased from VWR (Søborg, Denmark). H₂O₂ was standardized by measuring its absorbance at 240 nm using a molar absorptivity of 39.4 M⁻¹ cm⁻¹. To generate $HCO4−$⁠, H₂O₂ was diluted into bicarbonate-containing buffer (25 mM sodium bicarbonate in 100 mM sodium phosphate buffer, pH 7.4, containing 0.1 mM DTPA), and incubated for at least 7 min before use. All solutions were prepared using ultrapure water.

GAPDH (0.8 µM in monomer units) was incubated with 0.35, 0.7, 1.4 or 2.8 µM H₂O₂ (i.e. 0.4–3.5-fold molar excess over the GAPDH monomer concentration) for 3 h at 37°C in 100 mM sodium phosphate buffer (pH 7.4) containing 0.1 mM DTPA. For the experiments carried out in the presence of sodium bicarbonate (25 mM), buffers were prepared on the day of the experiment with the pH adjusted to 7.4 before use. The solutions used in the experiments under crowded conditions were prepared by mixing stock solutions prepared in 100 mM sodium phosphate buffer (pH 7.4) containing 0.1 mM DTPA and 120 mg ml⁻¹ dextran 35, Ficoll® 70, sucrose or glycerol. The reaction volume was typically between 50 and 100 µl, with the oxidation reactions stopped at the required time points by adding catalase (20 µg ml⁻¹) to remove excess H₂O₂.

GAPDH activity was measured by diluting control and treated samples to an enzyme concentration of 0.4 µM (i.e. 1:1 dilution) in 100 mM Tris buffer (pH 8.0) containing 0.4 mM G3P, 1 mM NAD⁺, 15 mM sodium arsenate and 0.2 mM EDTA in Greiner UV-Star® 96 well plates (VWR) [10]. Thus, for samples prepared under crowded conditions (i.e. in the presence of 120 mg ml⁻¹ dextran 35, Ficoll® 70, sucrose, or glycerol), a final concentration of 60 mg ml⁻¹ of the crowding agents was present when the activity was measured. Initial slopes were determined from the increase in absorbance at 340 nm due to NADH formation using a SpectraMax® i3 plate reader (Molecular Devices, U.S.A.). Activity data are presented as a percentage relative to the respective control (i.e. incubation in the absence of H₂O_2, and/or bicarbonate, under both dilute and crowded conditions). Plate readings were performed separately for each experimental condition to avoid any lag time between the addition of substrate to the enzyme and the start of the plate reading.

Activity data are presented as mean ± SD from at least six individual experiments, with the figures created using OriginPro. Statistical significance was determined by one-way ANOVA using Dunnett's post hoc test. Significance was assumed at the P < 0.05 level.

Experimental data will be made available on reasonable request to the authors.

The authors declare that there are no competing interests associated with the manuscript.

Eduardo Fuentes-Lemus: Conceptualization, Software, Formal analysis, Supervision, Validation, Investigation, Visualization, Methodology, Writing — original draft, Writing — review and editing. Rebecca H. J. Bloemen: Software, Formal analysis, Validation, Investigation, Visualization, Writing — review and editing. Rafael Radi: Conceptualization, Visualization, Writing — review and editing. Michael J. Davies: Conceptualization, Supervision, Funding acquisition, Visualization, Methodology, Project administration, Writing — review and editing.

Authors acknowledge funding from the Novo Nordisk Foundation (Laureate grant: NNF20SA0064214) to M.J.D.

GAPDH

glyceraldehyde-3-phosphate dehydrogenase

SOD1

superoxide dismutase