Na' and K' Transport Damage Induced by Oxygen Free Radicals in Human Red Cell Membranes*

The treatment of human erythrocytes with phenazine methosulfate (PMS) results in a sustained intracellular production of oxygen free radicals as shown by the reduction of nitroblue tetrazolium and methemoglobin production. Inhibition of superoxide dismutase by di- ethyldithiocarbamate further enhances nitroblue tetrazolium reduction indicating an increase in the PMS- dependent oxygen free radicals. NADH is one of the hydrogen donors able to react with PMS and O2 in order to generate oxygen free radicals in cyclic reac-tions which last far more than 4 h. The attack of red cell membranes by oxygen free radicals greatly alters their chemical structure. We observed lipid peroxidation, as shown by the increase in malondialdehyde above endogenous levels and met- hemoglobin binding to the cell membrane. PMS treatment markedly modifies the ionic equilibrium of the erythrocytes. At low PMS concentrations they lose intracellular K'. This is a prehemolytic effect, since hemolysis occur at high PMS concentrations. Both toxic effects are enhanced by superoxide dismutase inhibition with diethyldithiocarbamate, strongly suggesting that they result from the attack of cell membranes by oxygen free radicals. A study of the effect of PMS on the different transport pathways for K' across the human red cell membrane showed that at low PMS concentrations there is a specific increase in passive K+ permeability with no major effect on the specific K' carriers such as the (Na+,K+)-pump or the (Na+,K+)- studied by adding different concentrations of PMS, DDC, or DIDS to the (Na',K')-Ringer medium. For all the with DDC or for 10 min with DIDS. PMS was then added to this DDC or DIDS experiments, the cells were preincubated for 30 min medium. Samples were taken at different time intervals. In two control experiments, we studied the effect of 1 mM DDC on purified human erythrocyte superoxide dismutase (Sigma) and on the superoxide dismutase of intact The addition a medium units/ml of purified superoxide dismutase PMS-NADH-NBT sufficient the superoxide dismutase activity. of intact inhibits 78 activity The effect of human blood superoxide dismutase (Sigma) and bovine liver catalase (Sigma) on PMS-dependent K+ fluxes was studied using the above protocol. Inhibition of endogenous catalase was carried out following a previous established protocol (29).

The treatment of human erythrocytes with phenazine methosulfate (PMS) results in a sustained intracellular production of oxygen free radicals as shown by the reduction of nitroblue tetrazolium and methemoglobin production. Inhibition of superoxide dismutase by diethyldithiocarbamate further enhances nitroblue tetrazolium reduction indicating an increase in the PMSdependent oxygen free radicals. NADH is one of the hydrogen donors able to react with PMS and O2 in order to generate oxygen free radicals in cyclic reactions which last far more than 4 h.
The attack of red cell membranes by oxygen free radicals greatly alters their chemical structure. We observed lipid peroxidation, as shown by the increase in malondialdehyde above endogenous levels and methemoglobin binding to the cell membrane.
PMS treatment markedly modifies the ionic equilibrium of the erythrocytes. At low PMS concentrations they lose intracellular K' . This is a prehemolytic effect, since hemolysis occur at high PMS concentrations. Both toxic effects are enhanced by superoxide dismutase inhibition with diethyldithiocarbamate, strongly suggesting that they result from the attack of cell membranes by oxygen free radicals. A study of the effect of PMS on the different transport pathways for K' across the human red cell membrane showed that at low PMS concentrations there is a specific increase in passive K+ permeability with no major effect on the specific K' carriers such as the (Na+,K+)-pump or the (Na+,K+)cotransport system. The increase in passive K+ permeability has the following properties: (i) it is not affected by ethylene glycol bis(P-aminoethyl ether)-N,Nflfltetraacetic acid, quinine, or the replacement of C1by NO3-; (ii) it is enhanced by inhibition of the anion carrier with 4,4'-diisothiocyano-2,!2'-disulfonic acid stilbene; (iii) it is partially inhibited by external superoxide dismutase; and (iv) it is not affected by external catalase or by inhibition of endogenous catalase with 3-amino-1,2,4-triazole.
Our results strongly suggest that the PMS-dependent erythrocyte K' loss results from membrane lipid peroxydation by oxygen free radicals, particularly 0 $ The generation of free radicals following irradiation (1, 2), disturbance of tissue oxygen supply (3)(4)(5)(6)(7), inflammation (8-lo), and other physiopathological conditions (11,12) may be extremely noxious to the structure and function of cell mem-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby solely to indicate this fact. marked "advertisement" in accordance with 18 U.S.C. Section 1734 branes. This membrane damage appears to result from the peroxidation of unsaturated lipids (13) and the inactivation of "SH groups of functional proteins (14). On the other hand, one of the main effects of cardiac and cerebral hypoxia is a loss of intracellular K' (5,15). It is possible that some mechanism involving the interaction of oxygen-free radicals with some membrane K' carrier or pathway may mediate this K' loss (16,17). Unfortunately, the molecular mechanism of this phenomenon is not well known given the difficulty in studying the different transport systems in cells of the brain and myocardium.
The human red cell contains several transport systems for K' which are well characterized at the molecular level and which appear to be representative of those existing in other cells (18). Furthermore, it has been recently described that phenazine methosulfate may react in vitro with hydrogen donors such as NADH leading to the formation of superoxide anions (0 5) and other oxygen free radicals (19). We have thus applied this method for the intracellular generation of oxygen free radicals in human erythrocytes, and we have studied the different transmembrane K' pathways under these conditions.

MATERIALS AND METHODS
Preparation of Red Cells-Venous blood, collected in heparinized tubes, was centrifuged at 1750 X g for 10 min, and the plasma and huffy coat were removed. Red cells were then washed twice with cold isosmotic NaCl or 4 times with isosmotic MgCle, depending upon the type of experiment to be performed.
Qualitative Indicator of Free Radicals-A 50% suspension of washed red cells in cold (Na',K')-Ringer medium was resuspended in the same medium containing 50 PM NBT' to give a final hematocrit of 4% (hematocrit was measured in an aliquot of the original suspension of washed red cells). The (Na',K')-Ringer medium contained: 137.5 mM NaCI, 4.2 m~ KCl, 1 mM MgCL, 2.5 mM Na' phosphate (pH 7.4, at 37 "C), 10 mM Tris-4-morpholinepropanesulfonic acid (pH 7.4, at 37 "C), and 10 mM glucose. All steps were carried out at 4 "C.
Triplicate plastic tubes containing 2 ml of red cell suspension were incubated for 0, %, 1, 135, 2, 3, and 5 h at 37 "C. To stop the reaction, tubes were transferred to 4 O C for 1 min and then centrifuged for 4 min at 4 "C. The supernatant was carefully removed (avoiding pellet contamination), and the absorbance of reduced NBT was measured at 560 nm (19).
The effect of PMS, DDC, or DIDS on oxygen free radical production was studied using the above protocol. Reduction of NBT under such conditions was studied by adding different concentrations of PMS, DDC, or DIDS to the (Na',K')-Ringer medium. For all the with DDC or for 10 min with DIDS. PMS was then added to this DDC or DIDS experiments, the cells were preincubated for 30 min medium. Samples were taken at different time intervals.
In two control experiments, we studied the effect of 1 mM DDC on purified human erythrocyte superoxide dismutase (Sigma) and on the 3 108

Oxygen Free Radicals and
Erythrocyte K+ Loss endogenous superoxide dismutase of intact red cells. The addition of 1 mM DDC to a medium containing 5 units/ml of purified superoxide dismutase and PMS-NADH-NBT is sufficient to inhibit 100% of the superoxide dismutase activity. Moreover, the incubation of intact human red cells with 1 mM DDC for 30 min at 37 "C, inhibits 78 -C 10% of initial superoxide dismutase activity (mean f range of three healthy donors). Superoxide dismutase activity was measured using the procedure of Winterbourn et al. (41).
Measurement of Intracellular NADH in Human Erythrocytes-Washed red cells were suspended in cold (Na',K')-Ringer medium (in which glucose was replaced by 10 mM deoxyglucose) with different concentrations of PMS. The final hematocrit was about 10%. Triplicate plastic tubes were incubated for 4 h at 37 "C. To stop the reaction, tubes were transferred to 4 "C for 1 min. The red cell pellet was washed twice with isosmotic NaCl and then hemolyzed with distilled water to a final volume of 2.5 ml. 0.25 ml of 10 N NaOH was added to the hemolysates, which were immediately vortexed. 0.275 ml of 30% phosphoric acid was then added and the tubes vortexed for a second time. Tubes were centrifuged for 10 min at 2000 X g. 2 ml of supernatant were carefully removed. 0.2 ml of 10 N NaOH was added, and the tubes were vortexed. 0.21 ml of 30% phosphoric acid was then added, and the tubes were vortexed for a second time. Each procedure of protein precipitation with NaOH and HnPOd was carried out in less than 1 min. This extremely exothermic reaction is probably responsible for the protein precipitation. We developed this procedure of protein precipitation because the classical methods (trichloracetate, heat, etc.) were unable to preserve NADH content. NADH was measured in the supernatant by fluorimetry (excitation, 366 nm; emission, 455 nm) ( 3 4 ) . The calibration curve has been obtained by measurement of different NADH concentrations exposed under the same conditions to NaOH and H:lP04. This calibration curve is comparable to one obtained in H20.
Malondialdehyde Measurement-Washed red cells were diluted in a cold Mg", sucrose medium to give a final hematocrit of about 25%.
Triplicate plastic tubes containing 2 ml of red cell suspension were incubated for 2 h at 37 "C. To stop, the reaction tubes were transferred to 4 "C for 1 min and then centrifuged for 4 min at 1750 X g at 4 "C. The red cell pellet was washed twice with isosmotic NaCl and then hemolyzed with 30 ml of cold Na phosphate 5 mM, pH 7.4. Membranes were centrifuged for 20 min at 30,000 X g. The supernatant was removed, and the membranes were resuspended in 30 ml of 50 mM Na' phosphate, pH 7.4, at 4 "C. After recentrifugation, membranes were resuspended in 5 m~ Na' phosphate, pH 7.4, recentrifuged, and the supernatant removed. 4 ml of %2 N H2S04, 1 ml of TBA solution (30) and 20 pl of 10% Acationox (a detergent from Scientific Product, IL) were added to the membrane pellet the TBA solution contained 0.67 g of thiobarbituric acid, 100 ml of 0.1 N NaOH, and 100 ml of glacial acetic acid. Membrane suspensions were transferred to 95 "C for 1 h. A control without TBA was run at the same time. After the incubation, the tubes were rapidly cooled, and 5 ml of 1-butanol were added. A chromophoric product of MDA was extracted and centrifuged for 15 rnin at 2000 X g. Its absorption was measured in the butanol phase at 532 nm. MDA content was calculated from the difference between tubes with and without TBA.
The effect of 2 mM PMS and 2 mM PMS + 1 mM DDC on membrane MDA was studied using the above protocol. Methemoglobin Measurement-The MetHb content of PMStreated erythrocytes was measured using a similar protocol to that of NBT. After PMS incubation, the red cells were washed twice with isosmotic NaCl and then hemolyzed with distilled water. Membranes were removed by centrifugation for 10 min at 2000 X g, and MetHb was measured in the supernatant by absorbance at 630 nm (20-29). The MetHb bound to the red cell membrane was measured in ghosts resuspended in a medium containing 0.05% Acationox.
Measurement of Cation Movements-The techniques for measuring the different Na'-and K+-transport pathways in human erythrocytes have been developed in great detail elsewhere (21-23). Briefly, (Na',K+)-pump fluxes are represented by the fraction of total Na' efflux inhibited by 0.1 mM ouabain, (Na+,K+)-cotransport fluxes are represented by the fraction of Na' and K' efflux in a Mg", sucrose, ouabain medium inhibited by 1 mM furosemid or 0.05 mM bumetanide, and the passive Na' and K+ permeabilities are represented by the ouabain-and bumetanide-resistant fraction of Na+ and K+ efflux in a Mg"', sucrose medium.
The effect of PMS was studied by adding different concentrations of PMS to the efflux medium. The effect of DIDS or DDC was llME (hours) FIG. 1. Generation of oxygen free radicals in PMS-treated erythrocytes. The appearance of oxygen free radicals was followed spectroscopically by reduction of nitroblue tetrazolium (19). Inhibition of superoxide dismutase by DDC further increases the concentration of oxygen free radicals. A similar experiment has been repeated five times. The experiment represented in this figure was performed in Mg", sucrose medium instead of (Na',K')-Ringer medium. studied by preincubating the cells with these compounds for 10 or 30 min, respectively. PMS was then added to this medium. Samples were taken at different time intervals including zero time for flux determination.
The effect of human blood superoxide dismutase (Sigma) and bovine liver catalase (Sigma) on PMS-dependent K+ fluxes was studied using the above protocol. Inhibition of endogenous catalase was carried out following a previous established protocol (29).

PMS-
treated Human Red Cells-It has been previously reported that PMS may react with hydrogen donors such as NADH to produce superoxide anion (0 5) and other oxygen free radicals (19). Such a generation of free radicals can be followed spectroscopically by the reduction of NBT. In recent experiments we observed that PMS-like compounds may easily enter erythrocytes.' Thus, we decided to see whether oxygen free radicals may be generated inside human red cells by incubation with PMS. Fig. 1 shows that in PMS-treated erythrocytes, there is a net increase in the reduced form of NBT above control levels. In 6 different experiments, we observed that reduced NBT appears immediately after PMS incubation and increases linearly in time for at least 3-5 h, suggesting that PMS reacts with intracellular hydrogen donors to generate oxygen free radicals (Fig. 11). Further support of this mechanism came from the fact that when the erythrocyte superoxide dismutase is inhibited by DDC, there is an increase in reduced NBT (Fig. 1).
In control experiments, we observed that PMS is only able to transform hemoglobin into MetHb if NADH is present in the medium (see also Ref. 22). This indicates that this reaction is mediated by the oxygen free radicals generated by the reaction between PMS, NADH, and 0 2 (19). In PMS-treated erythrocytes, there is a marked and rapid MetHb production (Fig. 2, A and B) further suggesting the generation of intracellular oxygen free radicals.
The Effect of PMS on the Intracellular NADH Content of Human Red Cells-It is thought that one of the intracellular substrates capable of reacting with PMS to generate oxygen free radicals i s NADH. One experimental problem in demonstrating a reaction between PMS and NADH is that this latter is regenerated by glycolysis.
' R. P. Garay Thus, we measured the intracellular content of NADH after PMS treatment in erythrocytes in which glycolysis was inhibited by the addition of deoxyglucose (Table I)   now well established that oxygen free radicals are able to peroxidate unsaturated phospholipids of different membrane preparations (24, 25). This phenomenon is currently studied by measuring one of the products of this reaction: MDA. Table I1 shows that in ghosts prepared from PMS-treated erythrocytes, there is an increase in MDA content above endogenous levels which is further increased by the inhibition of superoxide dismutase by DDC. MetHb-binding to the Red Cell Membrane-A small fraction of the formed MetHb in PMS-treated erythrocytes remains fiimly bound to the red cell membrane (Table 111). The amount of bound MetHb is further increased by inhibition of superoxide dismutase with DDC, suggesting a reaction between MetHb and some product of the reaction between the membrane and the oxygen free radicals.
The Effect of PMS on the Intracellular Na' and K+ Content of Human Erythrocytes-The addition of PMS to human erythrocytes incubated in physiological (Na+,K+)-Ringer medium induces a loss of cell K+ which is accompanied by a less important gain in intracellular Na'. This phenomenon is enhanced by the inhibition of superoxide dismutase by DDC (Fig. 3, A and B ) .
The Effect of PMS on the Erythrocyte (Na+,K+)-pump-Ouabain-sensitive Na' efflux was measured in fresh human erythrocytes treated with different concentrations of PMS. Fig. 4 shows that the (Na+,K+)-pump is resistant to PMS treatment because high PMS concentrations are required to inhibit the (Na',K+)-pump. In 8 healthy donors, the PMS concentration for half-maximal pump inhibition varied from 0.8 to 3 mM (mean:l.l mM) (Fig. IO). Fig. 5 shows that the molecular mechanism of action of PMS involves inhibition of the maximal rate of the (Na+,K+)pump without any observable effect on the apparent affinity for intracellular Na'.
The Effect of PMS on the Erythrocyte (Na+,K+)-Cotransport System-Bumetanide-sensitive Na+-efflux was measured in fresh human erythrocytes treated with different concentrations of PMS (Fig. 4). We observed that the (Na+,K+)-cotransport system could be inhibited by PMS concentrations lower than those used for inhibiting the (Na+,K+)-pump. In 6 healthy donors, the PMS concentration for half-maximal inhibition varied from 0.1 to 0.5 mM (mean: 0.33 mM) (Fig. IO). Fig. 6 shows that the molecular mechanism of PMS action on this cotransport system is similar to that on the (Na+,K+)pump, i.e. inhibition of the maximal rate without affecting the apparent affinity for intracellular Na'.
The Effect of PMS on the Passive IC+ and Na' Permeabilities-Passive K+ permeability, but not passive Na+ permeability, increases with very low doses of PMS (Fig. 4). Indeed, Fig. 10 shows that in 24 healthy donors, the passive K+ permeability is stimulated by PMS concentrations as low as 0.01 to 0.03 mM. This is the transport parameter most affected by PMS. Fig. 7 shows that the PMS-dependent increase in passive K+ permeability is linear with the increase in intracellular K'   concentration and thus corresponds to a real increase in the basal K' permeability. In addition, Fig. 8 shows that the increase in passive K' Permeability is not a "Gardos effect" because it is not affected by extracellular Ca", ethylene glycol bis (P-aminoethyl ether)-N,N,N',N'-tetraacetic acid, quinine, or carbocyanin (23). Fig. 9 shows that the inhibition of superoxide dismutase by DDC strongly enhances the effect of PMS on passive K' permeability. This corresponds to a toxic effect of oxygen free radicals on the red cell membrane as further shown by the appearance of hemolysis at high PMS and DDC concentrations ( Fig. 9 inset). Inhibition of intracellular catalase by preincubation with 3amino 1,2,4-triazole (29) had no effect on the PMS-dependent passive K' permeability.
The PMS effect is slightly inhibited by addition of superoxide dismutase to the incubation medium (Table IV) and not modified by external catalase.
The Effect of DIDS on Passive K" Permeability a n d NBT Reduction in PMS-treated Erythrocytes-It has been shown that the very active anion carrier of human red cells is able to

I N T R A C E L L U L A R Na' C O N C E N T R A T I O N (mmoles/l. cells)
FIG. 5. inhibition of the maximal rate of (Na+,K+)-pump by oxygen free radicals. Theoretical curves were constructed by the method described in Ref. 21    sucrose medium the percentage of hemolysis is reduced suggesting a "free radical scavenger" effect of the sucrose.    (Table V).
DIDS showed a slight inhibitory effect on PMS-dependen1 NBT reduction (Table V). This result suggested to us thal NBT is able to react with both intra-and extracellular oxyger free radicals. Indeed, this hypothesis was confirmed by exper. iment showing that NBT may easily enter the erythrocytes approaching equilibrium in less than 5 min.

DISCUSSION
The results presented in this paper show that PMS treat. ment of human red cells results in a sustained generation 0 1 intraceklar oxygen free radicals for more than 5 h. This phenomenon is partially masked by the very active erythro.

Radicals and Erythrocyte K+ Loss
cyte superoxide dismutases which destroy such radicals (27). Indeed, superoxide dismutase inhibition by DDC reveals a significant production of oxygen free radicals in PMS-treated erythrocytes. The production of oxygen free radicals is further demonstrated by the appearance of MetHb. In fact, PMS treatment of human hemoglobin does not result in MetHb formation unless NADH is present in the incubation medium, suggesting a secondary reaction with oxygen free radicals.
The chemical nature of the erythrocyte substrates reacting with PMS is not yet completely understood. We observed here a decrease in the NADH content of PMS-treated erythrocytes under conditions in which glycolysis and thus NADH regeneration are inhibited by deoxyglucose. A likely mechanism of oxygen free radical production is the cyclic reaction shown previously in a cell-free system by Nishikimi et al. (19): However, PMS may be reduced by other erythrocyte metabolites such as glutathion which exists at higher concentrations than that of NADH and is in redox equilibrium. This is at present under investigation.
The presence of oxygen free radicals is extremely noxious to the cell membrane. Several authors have reported peroxidation of unsaturated lipids by oxygen free radicals in different membrane preparations (13,24-26). Indeed, we observed here that the PMS treatment of human red cells results in lipid peroxidation as shown by the increase in MDA above endogenous levels. The PMS-dependent membrane MDA is further increased by superoxide dismutase inhibition as expected for a lipid peroxidation dependent on the intracellular oxygen free radicals. On the other hand, lipid peroxidation is certainly not the only noxious effect of oxygen free radicals on the cell membrane. In fact, a small but significant fraction of the formed MetHb is fiimly bound to the red cell membrane. A similar phenomenon has been previously described under conditions in which MetHb is produced in the presence of MDA, suggesting a cross-linkage mechanism between MDA, amino groups of MetHb, and amino groups of the membrane (26). The relative participation of lipid peroxidation, MetHb binding, and other mechanisms on the erythrocyte membrane damage induced by oxygen free radicals deserves further investigation.
The main purpose of this paper is the kinetic study of the effect of oxygen free radicals on the different K' transport systems of human red cell membranes. Indeed, PMS treatment modifies the ionic equilibrium of human erythrocytes in a manner which resembles that previously observed under physiopathological conditions involving generation of oxygen free radicals (3-10). We observed that low PMS concentrations result in a quasi-specific increase in passive K' permeability without any major effect on any specific K+ carrier such as the (Na+,K+)-pump, or the (Na',K')-cotransport system. This increase in passive K' permeability explains the cell K' loss. It is certainly not a Gardos effect (28) because it is not inhibited by quinine, carbocyanin, or ethylene glycol bis(Paminoethyl ether)-N,N,N',N'-tetraacetic acid, nor is it carriermediated because it is independent of C1-, which is thought to be a cosubstrate or cofactor of most carrier-mediated K' transport systems. It thus appears likely to be the consequence of an increase in the K' leak through the lipid bilayer. Indeed, the PMS-dependent increase in passive K' permeability and cell K' loss are both markedly increased by superoxide dismutase inhibition, further suggesting a mechanism involving the formation of transient K' channels through membrane regions containing peroxidated phospholipids. The relative specificity for K' has been previously observed under other circumstances (23) and may be explained by the fact that steric hindrance through the membrane channel is higher for the Na' ion than for the K' ion in the hydrated state.
It is interesting to note that sickled red cell membranes show a higher susceptibility to lipid peroxidation (31), increased binding of hemoglobin (32), and altered K' permeability (33). Thus, these cells show some properties similar to our PMS-treated erythrocytes.
It appears from our results that PMS induces the intracellular generation of oxygen free radicals such as 0 i!. The PMS-dependent 0 ; can be released from the cells into the incubation medium by the anion carrier. This is shown by (i) a slight protective effect of external superoxide dismutase and In conclusion, the PMS treatment of human red cells mimics the membrane peroxidation and K' loss of certain physiopathological conditions in which oxygen free radicals are generated (Fig. 11). This system could thus represent a useful model for studying these mechanisms at the molecular level. of H202.