Sickle hemoglobin disturbs normal coupling among erythrocyte O2 content, glycolysis, and antioxidant capacity

Abstract

Energy metabolism in RBCs is characterized by O2-responsive variations in flux through the Embden Meyerhof pathway (EMP) or the hexose monophosphate pathway (HMP). Therefore, the generation of ATP, NADH, and 2,3-DPG (EMP) or NADPH (HMP) shift with RBC O2 content because of competition between deoxyhemoglobin and key EMP enzymes for binding to the cytoplasmic domain of the Band 3 membrane protein (cdB3). Enzyme inactivation by cdB3 sequestration in oxygenated RBCs favors HMP flux and NADPH generation (maximizing glutathione-based antioxidant systems). We tested the hypothesis that sickle hemoglobin disrupts cdB3-based regulatory protein complex assembly, creating vulnerability to oxidative stress. In RBCs from patients with sickle cell anemia, we demonstrate in the present study constrained HMP flux, NADPH, and glutathione recycling and reduced resilience to oxidative stress manifested by membrane protein oxidation and membrane fragility. Using a novel, inverted membrane-on-bead model, we illustrate abnormal (O2-dependent) association of sickle hemoglobin to RBC membrane that interferes with sequestration/inactivation of the EMP enzyme GAPDH. This finding was confirmed by immunofluorescent imaging during RBC O2 loading/unloading. Moreover, selective inhibition of inappropriately dispersed GAPDH rescues antioxidant capacity. Such disturbance of cdB3-based linkage between O2 gradients and RBC metabolism suggests a novel mechanism by which hypoxia may influence the sickle cell anemia phenotype.

Figure 1

Comparison of osmotic fragility, fraction of reduced RBC membrane thiol, and reducing equivalent recycling in normal and SSRBCs after oxidant exposure. (A) NaCl-induced osmotic lysis at baseline and after oxidant exposure (HX/XO) for normal and SSRBCs; half-maximal effective concentration (EC₅₀) increased for SSRBCs alone (n = 3-10). *P < .05, mean ± SEM, is plotted. (B) Free (reduced) membrane thiol for normal and SSRBCs at baseline and after oxidant exposure (HX/XO; n = 6-8). *P < .05 for % loss, normal versus sickle cell. (C) GSH and NADPH E_hc for normal and SSRBCs at baseline and during oxidant exposure (n = 4-6). *P < .05 for normal versus sickle cell (baseline); +P < .05 for normal versus sickle cell (HX/XO exposed comparison); #P < .05 for normal versus sickle cell (HX/XO exposed comparison). At baseline, note that relative to normal RBCs, SS membrane thiol redox poise favors oxidation and GSH and NADPH reducing power is diminished (ie, less negative E_hc). Antioxidant system failure in SSBRCs is evidenced by the rate and degree to which GSH and NADPH reducing power is lost under O₂⁻ loading. Moreover, after oxidant exposure, SSRBCs suffer membrane thiol depletion and lose resilience to osmotic fragility (ie, rightward shift).

Figure 2

Glucose metabolism in SSRBCs is dysregulated with EMP bias and resilience to oxidative loading is restored by EMP blockade. HMP flux as a function of O₂ content was determined using ¹H-NMR spectroscopy to monitor positional ¹³C-enrichment in lactate isotopomers generated by RBCs incubated with [2-¹³C]-glucose (± MB to stimulate the HMP). (A) Representative spectrum from normal, unstimulated RBCs is shown (supplemental Figure 5 illustrates spectra for all conditions studied). Per sample, 256 acquisitions were recorded, resulting in a lactate S/N exceeding 1000/1. During acquisition, continuous-wave ¹³C-decoupling of lactate C1 was performed using Bayesian algorithms.⁵⁰ Lactate methyl isotopomer signals were deconvoluted and fit to: a doublet D (3-bond ¹H-¹H coupling only) arising from [¹²C]-lactate, a doublet of doublets Q3 (3-bond ¹H-¹H coupling and one bond ¹H-¹³C coupling) arising from HMP-generated [3-¹³C]-lactate, and a doublet of doublets Q2 (3-bond ¹H-¹H coupling and 2 bond ¹H-¹³C coupling) arising from EMP-generated [2-¹³C]-lactate. This approach yielded signal amplitude estimates with a 12% SD for the lowest magnitude (Q3) to ≤ 2% for the highest magnitude (D) peaks. Proportional HMP flux was calculated from peak areas attributable to the pathway-specific lactate isotopomers (eg, Q2→EMP, Q3→HMP, see insets and Table 1). (B) Glucose uptake rate plotted against lactate generation for all conditions studied; note grouping among normal and SSRBC populations, which demonstrate differing lactate/glucose relationships. Both glucose uptake and lactate generation accelerated with MB stimulation. (C) In normal RBCs, HMP flux increased significantly during MB stimulation and deoxygenation blunted this increase. SSRBCs demonstrated similar O₂ content–dependent variation in HMP flux, but the HMP increase with MB stimulation was significantly less robust than in normal RBCs. HMP/EMP bias was rebalanced in SSRBCs by incubation with the selective GAPDH inhibitor KA (please refer to Figure 7 for pathway schema); RBCs were then presented with an oxidative challenge. (D) This manipulation stabilized GSH-reducing power in SSRBCs during oxidant loading (mean ± SEM; n = 3-4). *P < .05, reported as E_hc, see supplemental Figure 3 for E_hc components. (E) KA treatment also restored membrane thiol protection in SSRBCs under oxidant loading (n = 3-8; mean ± SEM). *P < .05.

Figure 3

Binding of GAPDH and Hb to RBC membrane. Membranes from fully oxygenated normal RBCs were prepared by hypotonic lysis and alkaline washing. Membranes referred to as “control” were taken from the final (7th) wash. Equal volumes of these membranes were further processed to produce stripped (ie, removal of associated extrinsic membrane proteins) or shaved membranes (ie, digestion of exofacial and cytoplasmic protein domains). (A) To study GAPDH inhibition, the 3 preparations were incubated in enzyme activity buffer, pelleted, and the activity remaining in supernatant was determined (n = 3; mean ± SEM). +P < .05 for control versus shaved; *P < .05 for shaved versus stripped. (B) Samples from each wash (containing progressively less heme) were run in this GAPDH activity assay. The percentage GAPDH inhibition varied inversely with membrane-bound heme (n = 3; R² = 0.94). (C) Membrane-bound heme was monitored during repeated alkaline washing of membranes purified from fully oxygenated normal or SSRBCs. SS membranes retained heme more than did normal membranes (n = 3-5; mean ± SEM). Note similar globin findings in Figure 4A. *P < .05.

Figure 4

Model of inverted RBC membrane attached to silica beads. RBC membranes from normal and SSRBCs were prepared by hypotonic lysis and alkaline washing; GAPDH was depleted by further salt washing. (A) Differences were observed in membrane protein banding from normal and SSRBCs. Note that despite salt washing, Hb globin (Band 9) remained associated with SS membranes, which is consistent with membrane-associated heme in Figure 3C. Inverted GAPDH-depleted membranes were subsequently attached to silica beads. (B) Immunofluorescence imaging of glycophorin A (an exofacial label) and cdB3 (a cytoplasmic label) was used to determine membrane orientation (Zeiss Axioskop microscope, 40× objective; AxioCam with AxioVision acquisition software; imaging medium Prolong Gold). (C) Confocal immunofluorescence images of (non–GAPDH-depleted) inverted membrane coated beads demonstrate free access to cdB3, GAPDH, and Hb (FluoViewFV1000 Olympus confocal microscope, PlanApo 60×/1.40 oil objective, 2.4× zoom; Flouview acquisition software; imaging medium Prolong Gold). (D) The model was further characterized by comparing membrane protein, membrane volume, and B3 number (as a function of bead count) between preparations from normal and SSRBCs. Although the membrane protein load per bead was greater for SS preparations, the membrane volume and B3 count was invariant. (E) To evaluate GAPDH binding to the bead model, we determined the ratio between membrane B3 and membrane-bound GAPDH as a function of [GAPDH] in solution; as expected, this relationship is linear. (F) GAPDH binding specificity for B3 was qualitatively evaluated by confocal immunofluorescence, which illustrates variation in GAPDH binding ± preincubation with antibody to the cdB3 NH₂ terminus (GAPDH-binding site; FluoViewFV1000 Olympus confocal microscope, PlanApo 60×/1.40 oil objective; Flouview acquisition software; imaging medium Prolong Gold).

Figure 5

Avid O₂-responsive HbS binding to RBC membrane defeats potential control of GAPDH activity. RBC membranes depleted of endogenous GAPDH were attached to silica beads. (A) To study membrane-based inhibition of free GAPDH, progressive numbers of membrane coated beads were incubated with a GAPDH activity buffer (containing a fixed amount of GAPDH). Beads were pelleted and residual enzyme activity in the supernatant was assayed (n = 3-9; mean ± SEM). *P < .05 for normal versus SS. (B) To quantitate GAPDH binding to membrane, progressive amounts of GAPDH (in activity buffer) were incubated with membrane-coated beads (bead number fixed at 3 × 10⁸), beads were pelleted, and membrane-bound GAPDH was quantitated by immunoblot and densitometry. SSRBC membranes quenched and bound significantly less GAPDH than control RBC membranes (n = 4; mean ± SEM) *P < .05 for normal versus SS. In separate experiments, beads coated with membrane from oxygenated or deoxygenated normal RBCs were incubated with oxygenated (HbO₂% > 85%) or deoxygenated (HbO₂% < 15%) HbAo or HbS and then incubated with GAPDH. Enzyme activity was determined in the supernatant. (C) No difference was observed in GAPDH quenching after incubation of bead-attached membrane from oxygenated RBCs with either oxy-HbAo or HbS (n = 4; mean ± SEM). (D) However, beads coated with washed membrane from deoxygenated RBCs and incubated with deoxy-HbS quenched significantly less GAPDH than those incubated with deoxy-HbAo (n = 6; mean ± SEM). *P < .05 for deoxy-HbAo versus deoxy-HbS.

Figure 6

O₂-dependent cytosol/membrane migration of GAPDH is absent in SSRBCs. Confocal immunofluorescent imaging was used to observe GAPDH migration in intact RBCs during O₂ loading/unloading. Washed RBCs were either fully oxygenated (HbO₂% > 95%) or deoxygenated (HbO₂% < 20%), fixed under appropriate gas tensions, probed for GAPDH (green) or cdB3 (yellow; see “Methods”), and confocal images were obtained (FluoViewFV1000 Olympus confocal microscope, PlanApo 60×/1.40 oil objective; Flouview acquisition software; imaging medium Prolong Gold). Membrane-based location of cdB3 is easily visualized in all images from both SS and normal RBCs. SSRBCs fail to demonstrate normal O₂ content–dependent migration of GAPDH from cytosol to membrane, as demonstrated in normal RBCs on oxygenation.

Figure 7

Simplified scheme of cdB3-based control of glucose metabolism in RBCs. (A) Energy metabolism in RBCs proceeds through either the EMP pathway (orange arrows) or the HMP pathway (blue arrows; also known as the pentose shunt). Both share G6P as initial the substrate. The HMP is the sole source of NADPH in RBCs and generates fructose-6-phosphate (F6P) or glyceraldehyde-3-phosphate (G3P), which rejoin the EMP before glyceraldehyde-3-phosphate dehydrogenase (G3PD/GAPDH), a key regulatory point. The EMP generates NADH (used by met-Hb reductase) and ATP (to drive ion pumps) and 2,3-DPG (to modulate Hb p50). Hydrogen peroxide (H₂O₂) and O₂⁻ are the principal endogenous reactive oxygen species (ROS) encountered by RBCs. Both ROS may be generated internally (not shown); however, only H₂O₂ can cross the membrane directly. O₂⁻ enters RBCs through Band 3 (anion exchange protein 1 or AE-1). H₂O₂ and O₂⁻ are ultimately reduced to water by catalase or GSH peroxidase (GPx). (B) O₂ content modulates EMP and HMP balance via reciprocal binding for cdB3 between deoxy-Hb and key EMP enzymes (PFK, Aldo, GAPDH, PK, and LDH).^, In oxygenated RBCs (right half of stylized oxygen dissociation plot), sequestration to cdB3 inactivates these EMP enzymes, resulting in HMP dominance and maximal NADPH (and thus GSH) recycling capacity. In deoxygenated RBCs (left half of oxygen dissociation plot), deoxy-Hb binding to cdB3 disperses these EMP enzymes, creating G6P substrate competition and thereby constraining HMP flux, limiting NADPH and GSH recycling capacity, and weakening resilience to ROS such as O₂⁻. (C) In SSRBCs, SSHb binds abnormally avidly to the RBC membrane at cdB3.^, We hypothesized that this biases normal EMP/HMP cycling (disfavoring HMP), rendering SSRBCs vulnerable to oxidant attack. In support of this, we found that by blocking the EMP with KA (at the point normally inhibited by cdB3-GAPDH binding) restored resilience to oxidative loads, presumably by lifting the G6PD substrate constraint.