Irreversible AE1 tyrosine phosphorylation leads to membrane vesiculation in G6PD deficient red cells

Abstract

Background: While G6PD deficiency is one of the major causes of acute hemolytic anemia, the membrane changes leading to red cell lysis have not been extensively studied. New findings concerning the mechanisms of G6PD deficient red cell destruction may facilitate our understanding of the large individual variations in susceptibility to pro-oxidant compounds and aid the prediction of the hemolytic activity of new drugs.

Methodology/principal findings: Our results show that treatment of G6PD deficient red cells with diamide (0.25 mM) or divicine (0.5 mM) causes: (1) an increase in the oxidation and tyrosine phosphorylation of AE1; (2) progressive recruitment of phosphorylated AE1 in large membrane complexes which also contain hemichromes; (3) parallel red cell lysis and a massive release of vesicles containing hemichromes. We have observed that inhibition of AE1 phosphorylation by Syk kinase inhibitors prevented its clustering and the membrane vesiculation while increases in AE1 phosphorylation by tyrosine phosphatase inhibitors increased both red cell lysis and vesiculation rates. In control RBCs we observed only transient AE1 phosphorylation.

Conclusions/significance: Collectively, our findings indicate that persistent tyrosine phosphorylation produces extensive membrane destabilization leading to the loss of vesicles which contain hemichromes. The proposed mechanism of hemolysis may be applied to other hemolytic diseases characterized by the accumulation of hemoglobin denaturation products.

Figure 1. Time course of AE1 tyrosine phosphorylation, oxidation and clustering.

GSH levels measured in control (CTRL) and G6PD deficient (G⁻) RBCs upon diamide treatment, expressed as µM (Fig. 1A). Amounts of hemichromes (HMC) measured in isolated membranes of control and G6PD deficient RBCs (Fig. 1B). HMCs were quantified by Vis spectrometry and expressed as nmoles/ml. Quantitative densitometry of AE1 phosphorylation levels (Fig. 1C) and of the oxidatively cross-linked AE1 (Fig. 1D) in control and G⁻ RBCs. Quantification of AE1 phosphorylation levels and of oxidized AE1 was performed with an IR fluorescence detection scanner (Odyssey, Licor, USA) of anti-phosphotyrosine (apTyr) and anti-AE1 (aBd3) western blots with Odyssey V3.0 software and expressed as fluorescence arbitrary units. The western blots in figure 1C and 1D show the areas used for AE1 tyrosine phosphorylation and oxidation quantifications. Membrane proteins were solubilized in presence (Fig. 1C) or absence (Fig. 1D) of the reducing agent (DTT). Percentages of clustered AE1 after gel filtration separation of the high molecular weight protein complexes (Fig. 1E). Membrane proteins were extracted from control and G6PD deficient RBCs by 1% triton-X100, the supernatant was applied to a 40×1 cm column filled with Sepharose CL-6B to isolate the high molecular weight membrane protein complexes. AE1 was quantified by eosine maleimide fluorescence detection. Clustered AE1 was quantified as percentage of total AE1 eluted from the column. The chromatogram in figure 1E shows the two peaks corresponding to clustered and non-clustered AE1. The western blots in figure 1E show the presence of aggregated AE1 (aBd3), its tyrosine phosphorylation (apTyr) and Syk (aSyk) in the high molecular weight fraction. Membrane proteins were solubilized in absence (lanes 1 and 2) or in presence (lane 3) of the reducing agent (DTT). Control and G6PD deficient RBCs were treated with 0.25 mM diamide (Dia) in presence or absence of Syk inhibitors 10 µM (Syk I.) at different incubation times (0–600 minutes). Values are means of 3 experiments. All differences observed between control and G6PD deficient RBCs were significant (p<0.01). In Figure 1C and 1E (only in G6PD deficient and after 60 minutes) the changes caused by Syk inhibitors were significant (p<0.01).

Figure 2. Time course of G6PD deficient RBC lysis.

A. Quantification of haemoglobin released from G6PD deficient (G⁻) RBCs that were treated with 0.25 mM diamide (Dia) in presence or absence of Syk inhibitors 10 µM (Syk I.) and o-vanadate 1 mM (OV) at different incubation times (0–600 minutes). Values are means of 3 experiments and are expressed as nmoles/ml. The changes caused by Syk inhibitors (after 120 minutes) and OV (after 60 minutes) were significant (p<0.01). B. Anti band 3 (aBd3) and anti-phosphotyrosine (apTyr) western blots of G⁻ RBCs treated with diamide 0.25 mM (lanes 1 and 2) in absence or presence of Syk inhibitors (lane3) or o-vanadate (lane 4). Membrane proteins were solubilized in presence of the reducing agent (DTT).

Figure 3. Confocal microscopy images of control and G6PD deficient RBCs.

Panels A D: hemichrome (HCM) autofluorescence; Panels B, E staining with anti-band 3 antibody (aBd3); Panels C, F staining with anti-phosphotyrosine antibody (apTyr). Panels A, B, C control RBCs (CTRL). Panels D, E, F G6PD deficient red cells (G⁻). Yellow arrows indicate clusters containing hemicromes and co-stained with anti AE1 and anti-phosphotyrosine antibodies. Bar: 20 µm.

Figure 4. Vesicle characterization.

A. Confocal image of isolated vesicles using hemichrome auto-fluorescence (excitation 488 nm, emission 630–750 nm). Bar: 10 µm. B. SDS PAGE of isolated vesicles and mass spectrometry identification of haemoglobin in the two most intense gel bands (indicated by the arrows). C. Corresponding western blot with anti AE1 (aBd3) (lanes 1 and 2) and anti-phosphotyrosine (apTyr) (lanes 3 and 4) antibodies under reducing (DTT) and non reducing conditions. D. Quantification of AE1 contained in vesicles released from G6PD deficient RBCs (G⁻). Quantification of AE1 was performed by IR fluorescence detection of anti-phosphotyrosine and anti-band 3 western blots (Odyssey, Licor, USA) with Odyssey V3.0 software and expressed as fluorescence arbitrary units. RBCs were treated with 0.25 mM diamide (dia) in presence or absence of Syk inhibitors 10 µM (Syk I.) at different incubation times (0–600 minutes). Values are means of 3 experiments. The changes caused by Syk inhibitors (after 120 minutes) were significant (p<0.01).

Figure 5. Schematic representation of the proposed mechanism.

AE1 is dynamically associated with the cytoskeleton through ankyrin binding in untreated red cells (A). In G6PD deficient RBCs diamide causes irreversible AE1 disulfide cross-linking and its phosphorylation by Syk kinase, diamide also causes progressive hemichrome formation. In control RBCs AE1 oxidation and phosphorylation are transient and no hemichrome formation is observed (B). Hemichromes bind to AE1 and promote the clustering of phosphorylated AE1 (C). Large aggregates of AE1 and hemichromes are released in vesicles (D).