Interaction of Ascorbate and (cid:97) -Tocopherol in Resealed Human Erythrocyte Ghosts TRANSMEMBRANE ELECTRON TRANSFER AND PROTECTION FROM LIPID PEROXIDATION*

A role for ascorbate-derived electrons in protection against oxidative damage to membrane lipids was investigated in resealed human erythrocyte ghosts. Incuba- tion of resealed ghosts with the membrane-impermeant oxidant ferricyanide doubled the ghost membrane con- centration of F 2 -isoprostanes, a sensitive marker of lipid peroxidation. Incorporation of ascorbate into ghosts during resealing largely prevented F 2 -isoprostane formation due to extravesicular ferricyanide. This protec- tion was associated with a rapid transmembrane oxidation of intravesicular ascorbate by extravesicular ferricyanide. Transmembrane electron transfer, which was measured indirectly as ascorbate-dependent ferri- cyanide reduction, correlated with the content of (cid:97) -to-copherol in the ghost membrane in several respects. First, ascorbate resealed within ghosts protected against ferricyanide-induced oxidation of endogenous (cid:97) -tocopherol in the ghost membrane. Second, when exogenous (cid:97) -tocopherol was incorporated into the ghost membrane during the resealing step, subsequent ferri- cyanide reduction was enhanced. Last, incubation of intact erythrocytes with soybean phospholipid lipo- somes, followed by resealed ghost preparation, caused a proportional decrease in both the membrane content of (cid:97) -tocopherol and in ferricyanide reduction. Incorporation of exogenous (cid:97) -tocopherol during resealing of ghosts prepared from liposome-treated cells completely restored the ferricyanide-reducing capacity of the ghosts. These results suggest that the transmembrane transfer of ascorbate-derived electrons in erythrocyte ghosts is dependent in part on (cid:97) -tocopherol and that such transfer may help to protect the erythrocyte mem- brane against oxidant stress originating outside the cell.

A role for ascorbate-derived electrons in protection against oxidative damage to membrane lipids was investigated in resealed human erythrocyte ghosts. Incubation of resealed ghosts with the membrane-impermeant oxidant ferricyanide doubled the ghost membrane concentration of F 2 -isoprostanes, a sensitive marker of lipid peroxidation. Incorporation of ascorbate into ghosts during resealing largely prevented F 2 -isoprostane formation due to extravesicular ferricyanide. This protection was associated with a rapid transmembrane oxidation of intravesicular ascorbate by extravesicular ferricyanide. Transmembrane electron transfer, which was measured indirectly as ascorbate-dependent ferricyanide reduction, correlated with the content of ␣-tocopherol in the ghost membrane in several respects. First, ascorbate resealed within ghosts protected against ferricyanide-induced oxidation of endogenous ␣-tocopherol in the ghost membrane. Second, when exogenous ␣-tocopherol was incorporated into the ghost membrane during the resealing step, subsequent ferricyanide reduction was enhanced. Last, incubation of intact erythrocytes with soybean phospholipid liposomes, followed by resealed ghost preparation, caused a proportional decrease in both the membrane content of ␣-tocopherol and in ferricyanide reduction. Incorporation of exogenous ␣-tocopherol during resealing of ghosts prepared from liposome-treated cells completely restored the ferricyanide-reducing capacity of the ghosts. These results suggest that the transmembrane transfer of ascorbate-derived electrons in erythrocyte ghosts is dependent in part on ␣-tocopherol and that such transfer may help to protect the erythrocyte membrane against oxidant stress originating outside the cell.
Erythrocytes possess redundant and overlapping mechanisms for protection of their cytoplasmic contents against oxidative damage, including catalase (1), superoxide dismutase (2), and low molecular weight antioxidants such as GSH and ascorbate (3). Defenses against oxidant damage in the plasma membrane are poorly understood but relate primarily to prevention and reversal of peroxidation of unsaturated fatty acids in the lipid bilayer (4). Both lipophilic and cytoplasmic factors are thought to protect the erythrocyte membrane against lipid peroxidation. Foremost is ␣-tocopherol (4,5), but ubiquinol-10 (6, 7), membrane protein sulfhydryls (8), and a GSH-dependent phospholipid hydroperoxidase (9,10) are also likely to contribute. Although ascorbic acid does not directly retard peroxidation of membrane lipids (11), it may do so indirectly by reducing the tocopheroxyl free radical in the lipid bilayer (12,13).
Beyond their ability to use intracellular reducing potential to protect the cell membrane from oxidant stress, erythrocytes have long been known to reduce extracellular oxidants such as ferricyanide (14 -16), methemoglobin (17), and ferricytochrome c (18). This export of electrons across the cell membrane has been attributed to activity of a transmembrane NADH:(acceptor) oxidoreductase (E.C. 1.6.99.3) (19,20). There are probably several such activities in erythrocyte membranes (20). For example, a transmembrane enzyme has been purified (21) that can be differentiated from the prominent inward facing cytochrome b 5 reductase (22) by the presence of carbohydrate, lower substrate affinities, and lack of a flavin as a cofactor. The transmembrane enzyme has also been implicated in the export of electrons derived from ascorbate (15,16,23). A physiologic function of this enzyme has not been established. One possibility is that in reticulocytes it provides electrons for reduction of Fe 3ϩ on transferrin (24); the resulting Fe 2ϩ is then released and can be taken up by the cell (25). It has also been suggested that such an enzyme may use electrons from intracellular NADH to reduce lipid hydroperoxides in the membrane (26). Ascorbate might help to protect the lipid bilayer, either by donating electrons to such an enzyme or by recycling the lipophilic antioxidant ␣-tocopherol.
The goals of the present study were to evaluate the mechanism of ascorbate-dependent trans-or intramembrane electron transfer in human erythrocytes and to determine whether such transfer protects against membrane lipid peroxidation. To minimize the effects of endogenous cytoplasmic antioxidants and to ensure study of transbilayer transfer, resealed erythrocyte ghosts containing ascorbate were used. The results show that ascorbate resealed within ghosts protects against ferricyanideinduced loss of endogenous ␣-tocopherol and against peroxidation of membrane lipids. The protective effect of intravesicular ascorbate involves a transmembrane transfer of ascorbate-derived electrons, the extent of which depends on the membrane content of ␣-tocopherol. pared from heparinized blood drawn from normal volunteers. The cells were washed three times in 10 volumes of phosphate-buffered saline (PBS), which consisted of deionized water containing 140 mM NaCl and 12.5 mM Na 2 HPO 4 at pH 7.4. The buffy coat of white cells was carefully removed with each wash. Resealed impermeable erythrocyte ghosts were prepared from intact cells as described by Steck and Kant (27).

Materials-Solid
Measurement of Extra-and Intravesicular Ascorbate Concentrations-After incubations as noted, resealed ghosts were pelleted by centrifugation, and ascorbate concentrations in the supernatant and pellet were measured. Ghosts in the pellet (0.2 ml) were lysed by diluting with 4 volumes of PBS containing 0.3% Triton X-100 (v/v). Protein removal without hemoglobin denaturation was carried out by the ultrafiltration method of Iheanacho et al. (28). The original supernatant or hemolyzed ghost solution was transferred to a Centricon-10 filter apparatus (Amicon, Inc., Beverly, MA) and centrifuged at 4°C for 30 min at 5000 ϫ g. This resulted in 0.3-0.4 ml of a clear ultrafiltrate, which was either assayed directly or diluted with 4 volumes of ice-cold methanol and microcentrifuged in the cold for 1 min. An aliquot of the resulting supernatant was assayed for ascorbate by HPLC using the ion-pairing method of Pachla and Kissinger (29). Separation was carried out on a Waters DeltaPak C 18 column (300 m, 5 ), with a 4-mm guard column of the same packing material. The mobile phase consisted of 80 mM ammonium acetate, 1 mM tridecylamine, and 15% methanol, pH 5.2. At a flow rate of 1 ml/min, ascorbate was eluted at 5-6 min, with UV detection at 254 nm. The assay sensitivity was 50 pmol/sample.
To estimate the relative amounts of ascorbate and its oxidation products, ghosts were resealed in the presence of 400 M of unlabeled ascorbate, 0.05 Ci of [ 14 C]ascorbate, and 0.2 Ci of L-[ 3 H]glucose. Sample preparation and HPLC analysis were carried out as described in the previous paragraph, except that the postdetector eluate was collected in 1-ml volumes with a fraction collector. Aliquots of these fractions (0.5 ml) were added to 5 ml of liquid scintillation fluid and counted under dual-label conditions with efficiency correction. Both To assess the effects of the resealing process on ascorbate added during resealing, ghosts were resealed in the presence of 400 M unlabeled ascorbate and [ 14 C]ascorbate, and ultrafiltrates of the ghost contents and hemolysate were analyzed by HPLC with scintillation counting of the column eluates as outlined above. At the end of the resealing period, the ghosts typically retained 200 -300 nmol of ascorbate/ml of intravesicular ghost volume. After the hour of resealing at 37°C the ascorbate concentration in the hemolysate outside the ghosts had decreased to 130 -150 M. The decrease in ascorbate both inside the ghosts and in the hemolysate was due largely to its conversion to [ 14 C]DHA, which was present in equal concentrations inside and outside the ghosts at the end of resealing (data not shown). Variable amounts of 2,3-[ 14 C]diketo-1-gulonic acid were also observed.
Measurement of Radiolabeled Ascorbate Efflux-Erythrocyte ghosts were resealed for 1 h at 37°C in the presence of 400 M freshly prepared unlabeled ascorbate, 0.05 Ci of [ 14 C]ascorbate, and 0.2 Ci of tracer L-[ 3 H]glucose. After the usual centrifugation washes at 15,000 ϫ g in 40 volumes of PBS to remove residual cytoplasmic contents, 0.2 ml of packed ghosts were rapidly diluted with mixing into 1 ml of PBS containing 5 mM D-glucose at 37°C. At the indicated times, 0.2-ml aliquots of the ghost suspension were removed and microfuged for 60 s to separate ghosts and buffer. The assay was stopped by removing 0.15 ml of the supernatant above the ghost pellet and adding it to scintillation fluid for radioactive counting by dual-label methods with efficiency correction. Efflux of radiolabel for both radionuclides was corrected for extravesicular radioactivity present at time zero, and calculated as a fraction of the total incorporated into the ghosts. This value was subtracted from 1 to obtain the fraction of radioactivity remaining in the cells at a given time point. Where noted, the distribution of [ 14 C]ascorbate and its oxidation products in ghosts and efflux medium was analyzed by HPLC with postcolumn fraction collection and scintillation counting as described in the previous section.
Measurement of ␣-Tocopherol in Erythrocyte Lysates and Ghost Membranes-An aliquot (0.1 ml) of erythrocyte lysate or a 0.2-ml pellet of packed ghosts was mixed with 10 l of methanol containing 0.5 mg of butylated hydroxytoluene, followed by 5-10 volumes of methanol containing 20 mM ascorbate. Butylated hydroxytoluene and ascorbate were added to prevent oxidation of ␣-tocopherol upon lysis of the ghosts with methanol (30). The lysate was mixed and allowed to sit on ice for 10 min before centrifugation and assay of the supernatant for ␣-tocopherol. Recovery of ␣-tocopherol was not improved by hexane extractions (30), so these were not routinely performed. Extraction of erythrocyte ␣-tocopherol in the presence of pyrogallol (31) also did not improve ␣-tocopherol recovery over the method described above. The lysate and ghost membrane extracts were chromatographed by isocratic HPLC in 95% methanol on a Waters DeltaPak C 18 column (300 m, 5 ) with a 4-mm guard column of the same packing material. Tocopherols were quantified using a Kratos model 980 fluorometric detector, with excitation set at 210 nm and emission limited to wavelengths greater than 335 nm by a cut-off filter (32). Under these conditions, ␥-tocopherol eluted at 6.2 min, and ␣-tocopherol eluted at 7.3 min. The assay sensitivity for ␣-tocopherol was 2-5 pmol. Peak areas were integrated using a Shimadzu C-R6A Chromatopac recorder and compared with freshly prepared standards of ␣-tocopherol.
Assay of Ferricyanide Reduction-Reduction of ferricyanide was measured by adding 0.2 ml of resealed ghosts to 0.8 ml of PBS containing the indicated concentrations of ferricyanide and other additives. Incubation was carried out in a shaking water bath at 37°C. At the times noted, 0.2-ml aliquots of ghosts were removed and microfuged for 1 min at 13,000 rpm. The ferrocyanide content of an aliquot of the supernatant was measured spectrophotometrically, using an extinction coefficient of 10,500 M Ϫ1 ⅐ cm Ϫ1 for the 1,10-phenanthroline complex with ferrocyanide (33). Correction was made for a small amount of background absorbance of the ghost-conditioned buffer. The amount of ferrocyanide generated was expressed relative to the intracellular water space of the ghosts (15), which was confirmed in separate studies to be 70% of the packed ghost volume.
Incubation of Erythrocytes with Phospholipid Liposomes-Liposomes were prepared by dissolving 25 mg of soybean L-␣-phosphatidylcholine (Sigma Type II-S) in 0.25 ml of chloroform, drying this solution under a stream of nitrogen, and removing residual solvent under house vacuum for 2 h. The lipid was resuspended in 50 mM Tris-HCl, pH 7.4, and liposomes were prepared by sonication for 3-5 min on ice in an Utrasonics model W-220F sonicator set at maximal microtip power. The liposomes (20 M phospholipid) were incubated for 2 h at 37°C with washed erythrocytes at a 20% hematocrit in PBS containing 5 mM D-glucose. Erythrocytes were pelleted by centrifuging at 1000 ϫ g for 5 min and washed twice more in 25 volumes of PBS before preparation of resealed ghosts. Aliquots of the liposome-containing supernatant were sampled for measurement of ␣-tocopherol and lipid hydroperoxides (34). The content of ␣-tocopherol was measured in ghosts, as was the ability of ghosts to reduce 1 mM ferricyanide in the presence of 5 mM D-glucose for 15 min at 37°C. Incubation of erythrocytes with liposomes caused a small increase in hemolysis measured at 540 nm after removal of cells. Compared with cells incubated in the absence of liposomes, the increase in absorbance was about 3-fold. Cells incubated in the presence of liposomes to which exogenous ␣-tocopherol had been added during preparation (1.8 nmol/mol of phospholipid) did not hemolyze. This agrees with previous results reported by Niki et al. (35) for the same type of liposome. In cells incubated with liposomes not containing added ␣-tocopherol, less than 1% of the ␣-tocopherol lost from the ghost membranes was recovered in the liposomes removed from the cells, suggesting loss through oxidation. However, no increase in the formation of lipid hydroperoxides was detected in either the liposomes or in erythrocyte membranes (results not shown).
Measurement of Lipid Peroxidation-For measurement of lipid peroxidation, resealed ghosts were lysed in 10 volumes of deionized water, and ghost membranes were prepared by three centrifugation washes at 15,000 ϫ g in 10 ml of PBS. F 2 -isoprostanes were extracted from ghost membranes, purified by TLC, derivatized, separated by selected ion monitoring gas chromatography, and quantified by a gas chromatography/mass spectrometric assay employing stable isotope dilution techniques, as described previously (36). Lipid hydroperoxides in liposomes and erythrocyte ghosts were measured using the FOX2 assay (34). Thiobarbituric acid-reactive substances were measured as described previously (37).
Data and Statistical Analyses-Curve fitting was carried out by nonlinear least-squares regression in the graphics software package FigP (Biosoft Inc., Cambridge, United Kingdom). Statistical significance was assessed by one-way analysis of variance using the statistical software package Sigmastat (Jandel Scientific, St. Louis, MO).

RESULTS AND DISCUSSION
To induce an oxidant stress on the erythrocyte membrane, resealed ghosts that had been depleted of cytosolic contents were incubated for 15 min with 1 mM ferricyanide at 37°C.
Ferricyanide has long been known not to penetrate the intact erythrocyte membrane (38). Similarly, the agent also failed to oxidize the residual hemoglobin present in resealed ghosts (results not shown), indicating that any effects observed are restricted to the outer surface of the membrane. However, ferricyanide is a relatively mild oxidant, and it did not increase the ghost membrane content of either thiobarbituric-reactive substances or lipid hydroperoxides (results not shown). Therefore, F 2 -isoprostanes were measured to provide a more sensitive measure of membrane lipid peroxidation in the ghosts. F 2 -isoprostanes are derived from free radical-mediated oxidation of arachidonic acid and have been shown to provide specific and extremely sensitive estimates of lipid peroxidation, both in vivo (39) and in vitro (40). In contrast to lipid hydroperoxides, F 2 -isoprostanes are chemically stable end products that can remain esterified in tissues or be released by the actions of phospholipases (36). Ghosts resealed in the presence or absence of ascorbate had low levels of F 2 -isoprostanes, and ferricyanide treatment more than doubled F 2 -isoprostane formation (Table  I). In ghosts resealed to contain ascorbate, the membrane content of F 2 -isoprostanes was halved. These results show that ferricyanide does oxidize a small amount of arachidonic acid that is presumably located in the outer membrane bilayer and that intravesicular ascorbate protects against this oxidation. The findings also indicate the enhanced sensitivity of the F 2isoprostane assay compared with other measures of lipid peroxidation. Although the improved sensitivity of this assay is due in part to the ease of oxidation of arachidonic acid, previous results suggest that this level of oxidation would not be detected by measuring changes in the membrane content of arachidonic acid (41).
Possible mechanisms for ascorbate-dependent protection of the ghost membrane against ferricyanide-induced lipid peroxidation were investigated. It is unlikely that the observed protection was due simply to consumption of most of the oxidant by intravesicular ascorbate, since in these incubations ferricyanide was present in at least a 17-fold molar excess over ascorbate. Whereas ferricyanide apparently does not cross the ghost membrane, ascorbate may efflux from the ghosts and consume ferricyanide directly in the medium. To assess the latter, timedependent efflux of radiolabel from ghosts that had been loaded with [ 14 C]ascorbate was measured, with the results shown in Fig. 1 (42)(43)(44). However, in intact cells the net efflux of [ 14 C]ascorbate is slower than observed in these ghosts (42,44). This difference is most likely due to the limited ability of the ghosts to recycle DHA to ascorbate compared with intact cells, so that any DHA generated rapidly exits on the glucose transporter (42,43).
When ghosts resealed with [ 14 C]ascorbate/L-[ 3 H]glucose were incubated with ferricyanide, there was rapid release of about 70% of the carbon-14, with no effect on the efflux of L-[ 3 H]glucose (Fig. 1). Under similar conditions, 1 mM ferricyanide also decreased the measured intravesicular ascorbate content to less than 5% of control concentrations within 15 min. HPLC analysis showed that following ferricyanide treatment, 90% of the radioactivity inside the ghosts was 2,3-[ 14 C]diketo-1-gulonic acid, and the remainder was [ 14 C]dehydroascorbate and [ 14 C]ascorbate. No [ 14 C]ascorbate was detected outside the ghosts. These results suggest that extravesicular ferricyanide oxidizes intravesicular ascorbate to DHA, which can then either escape the ghost on the glucose transporter (42)(43)(44) or undergo irreversible degradation to 2,3-diketo-1-gulonic acid (45). The latter, having undergone ring opening and containing a negative charge, is trapped within the ghosts and accounts for the residual carbon-14 radioactivity observed in the efflux experiment of Fig. 1.
The separation of ferricyanide and ascorbate across the ghost membrane implies a transmembrane transfer of ascorbatederived electrons, which could be estimated as the extent of ferricyanide reduction outside the ghosts. Ferricyanide reduction occurred rapidly over the same time course as loss of radiolabel from ferricyanide-treated ghosts and was largely complete following 15 min of incubation (results not shown). Several features of this ascorbate-to-ferricyanide electron transfer are shown in Table II. Ghosts resealed to contain ascorbate were from 3-5-fold more effective in reducing extravesicular ferricyanide than ghosts not containing ascorbate.  When ghosts were incubated with 100 M DHA, a small but significant increase in ferricyanide reduction was also observed. No increase in intravesicular ascorbate content was detected by direct assay, but after 5 min of incubation with 400 M [ 14 C]DHA under the conditions of Table II, 90% of the carbon-14 present in ghosts was [ 14 C]ascorbate. This suggests that the mechanism for the DHA-induced increase in ferricyanide reduction involved uptake of DHA, intravesicular reduction of DHA to ascorbate, and transmembrane reduction of extravesicular ferricyanide. Although glucose was required for the small DHA effect, omission of glucose did not affect the much larger extent of ferricyanide reduction in ghosts resealed to contain ascorbate.
The addition of ascorbate oxidase to the ghosts several minutes before ferricyanide was added did not affect ferricyanide reduction (Table II). In control studies not shown, the addition of 400 M ascorbate with ascorbate oxidase under the conditions of Table II decreased the expected amount of ferricyanide reduction by about two-thirds, indicating the ability of the enzyme to compete with ferricyanide for extravesicular ascorbate. The failure to see an ascorbate oxidase-induced decrease in ferricyanide reduction in ghosts containing ascorbate confirms the observations shown in Fig. 1, which indicated little direct efflux of ascorbate across the ghost membranes during the experiment. This further suggests a transmembrane transfer of ascorbate-derived electrons in the reduction of ferricyanide, rather than ferricyanide-induced efflux of ascorbate, followed by direct oxidation by ferricyanide outside the ghosts.
Ghosts resealed in the presence of 4 mM GSH showed a doubling in ferricyanide reduction (Table II). GSH and ascorbate together had an additive effect on the extent of ferricyanide reduction. A role for GSH in recycling ascorbate was investigated by measuring intravesicular ascorbate concentrations just before the addition of ferricyanide. No ascorbate was detected in ghosts resealed with GSH alone, but in ghosts resealed with both 400 M ascorbate and 4 mM GSH, ascorbate depletion was almost completely prevented (data not shown). This suggests that in ghosts resealed with GSH and ascorbate, much of the enhanced ability to reduce ferricyanide was due to increased intravesicular ascorbate concentrations. Direct participation of GSH in the transmembrane electron transfer process cannot be ruled out, however.
A common mechanism was sought to account for both transfer of ascorbate-derived electrons and ascorbate-dependent pro-tection of the outer bilayer leaflet from peroxidation in response to ferricyanide treatment. It has previously been observed that ascorbate alone does not directly transfer reducing equivalents to membrane lipids or hydroperoxides (11). A more likely possibility is that ␣-tocopherol might mediate the effect. It is well established that ␣-tocopherol can intercept free radicals in lipid bilayers (4) and that ascorbate can recycle ␣-tocopherol from its free radical in cell membranes (46,47). As shown in Fig. 2, increasing amounts of intravesicular ascorbate were found to preserve endogenous ␣-tocopherol in resealed ghosts over a range of extravesicular ferricyanide concentrations. This protection was partial at a 400 M loading concentration of ascorbate and more complete at a loading concentration of 2 mM, with an associated lag phase. The ascorbatedependent protection of endogenous ␣-tocopherol against oxidation by ferricyanide fits with the hypothesis that ␣-tocopherol is the proximal electron donor to ferricyanide and that ascorbate in turn recycles ␣-tocopherol.
It is also possible that ␣-tocopherol contributes to the transmembrane movement of electrons from ascorbate to ferricyanide. It has been shown that ␣-tocopherol can directly transfer ascorbate-derived electrons to ferricyanide across artificial liposomal bilayers (48), although transbilayer movement of ␣-tocopherol is slow relative to quinones with short hydrophobic tails (49). To determine whether ascorbate-dependent ferricyanide reduction requires ␣-tocopherol, two types of experiment were carried out. In the first, ␣-tocopherol was added to the ghost membranes during the resealing step. Only about 10% of added ␣-tocopherol was recovered in the cell lysate before resealing of the ghosts (see Table III legend), possibly because of oxidation or failure to stay in solution. However, 7-8% of the total available ␣-tocopherol was found in the ghosts, resulting in a 10-fold increase in the ␣-tocopherol content (Table III). The presence of exogenous ␣-tocopherol in the ghost membranes tripled ferricyanide reduction in ghosts not containing ascorbate and doubled it in ascorbate-containing ghosts (Table III). It should be noted that the incubation time of ghosts with ferricyanide was 5 min in these experiments rather than 15 min as in the experiments shown in Table II and Fig. 2. A shorter incubation time was used to ensure that ascorbate depletion was not limiting for ferricyanide reduction. Although ␣-tocopherol intercalates into the membrane bilayer and can  2. Protection against loss of endogenous ␣-tocopherol by intravesicular ascorbate in resealed erythrocyte ghosts. Ghosts were resealed in the absence (closed circles) or presence of 400 M ascorbate (closed squares) or 2 mM ascorbate (closed triangles) and incubated for 15 min at 37°C with 5 mM D-glucose and the indicated concentration of ferricyanide. The membrane content of ␣-tocopherol was determined and is expressed as a percentage of an untreated control. Data are from three to six experiments Ϯ S.E. The difference between control and both sets of ascorbate-supplemented ghosts was significant at the p Ͻ 0.01 level across the three ferricyanide doses.
bind specifically to the erythrocyte surface (50), based on the measured content of ␣-tocopherol in membranes not treated with ferricyanide (Table III), simple direct oxidation of ␣-tocopherol could have accounted for only a small portion of the observed increase in ferricyanide reduction.
In the second type of experiment designed to test for a role of ␣-tocopherol in facilitating the transfer of ascorbate-derived electrons across the ghost membrane, ferricyanide reduction was measured in ghosts depleted of endogenous ␣-tocopherol. Depletion of endogenous ␣-tocopherol was accomplished by incubating intact erythrocytes with soybean lecithin liposomes for 2 h followed by centrifugation washes to remove the liposomes. As shown by the first pair of bars in Fig. 3, this treatment decreased both the endogenous ␣-tocopherol content and the ferricyanide-reducing capacity of subsequently prepared ascorbate-containing resealed ghosts by 40 -50%. A similar decrease compared with control was observed when ascorbate was omitted during the resealing step (results not shown). Whether ␣-tocopherol was transferred from erythrocytes to the liposomes (51,52) or oxidized by the liposomes (53) is controversial. Although lipid peroxidation was not detected in either liposomes or ghosts following incubation of intact cells with liposomes in the present studies, the failure to detect quantitative transfer supports an oxidative loss. If depletion of ␣-tocopherol by liposome treatment in intact cells is the cause of the decrease in ferricyanide reduction in resealed ghosts, then adding back ␣-tocopherol during the resealing phase should reverse the effect. As shown by the second pair of bars in Fig.  3, ghosts exposed to ␣-tocopherol and ascorbate during resealing had the expected increases in measured ␣-tocopherol content and ferricyanide-reducing capacity (compare to results shown in Table III). The addition of both ␣-tocopherol and ascorbate during resealing of ghosts derived from liposometreated cells increased the ␣-tocopherol content and the capacity to reduce ferricyanide (third pair of bars in Fig. 3). Although the ␣-tocopherol content of such ghosts did not attain the level observed in ␣-tocopherol-treated ghosts prepared from cells not exposed to liposomes, the ferricyanide-reducing capacity was fully restored (compare the second and third pairs of bars in Fig. 3). Thus, both ␣-tocopherol depletion and repletion correlate with ferricyanide reduction in these ghosts, which supports the possibility that endogenous ␣-tocopherol might play a role in mediating the transmembrane transfer of ascorbatederived electrons.
Ferricyanide reduction by erythrocytes has usually been attributed to a transmembrane oxidoreductase that uses intra-cellular NADH as the electron donor (19,20). Ascorbate has also been implicated as an electron donor for such an enzyme (15,16), and recent results from this laboratory indicate that ascorbate may be the preferred substrate (23). The present results further suggest that ␣-tocopherol, either directly or in concert with an enzyme, may also be involved in the electron transfer as well as in protection of membrane lipids against oxidation by external ferricyanide.

TABLE III
Effects of ␣-tocopherol addition on ferricyanide reduction in resealed ghosts Ghosts were resealed in the presence or absence of the indicated antioxidant. Ascorbate was added at an initial concentration of 400 M. Although ␣-tocopherol was added initially to a calculated concentration of 73 M (0.6% isopropyl alcohol), its content in the lysate before ghost preparation ranged between 6 and 16 M. Following resealing and centrifugation washes, ghosts were incubated for 5 min at 37°C with 5 mM D-glucose with or without 1 mM ferricyanide. The membrane ␣-tocopherol content in ghosts not treated with ferricyanide and the amount ferricyanide reduced are expressed as a function of the intracellular ghost volume, Ϯ S.E. from the indicated number (n) of experiments. One asterisk indicates p Ͻ 0.05 versus ghosts not containing ascorbate, and two asterisks indicates p Ͻ 0.05 versus ghosts treated with the ␣-tocopherol but not ascorbate. Ghost 3. Parallel changes in ␣-tocopherol content and ferricyanide reduction in ghosts from liposome-treated erythrocytes with and without ␣-tocopherol treatment. Liposome treatment, ␣-tocopherol addition, and incubation conditions are described under "Experimental Procedures." Shown are the relative increases in ␣-tocopherol content (open bars) and ferricyanide reduction (hatched bars) observed in ghosts prepared from cells that had been treated with liposomes but not resealed in the presence of exogenous ␣-tocopherol (first pair of bars), in ghosts from control cells that were treated with ␣-tocopherol during resealing (second pair of bars), and in ghosts that were both prepared from liposome-treated cells and treated with ␣-tocopherol during resealing (third pair of bars). The data are from four experiments Ϯ S.E. and are expressed relative to responses observed in ghosts not exposed to either liposomes or ␣-tocopherol. All ghosts were resealed in the presence of 400 M ascorbate. An asterisk indicates p Ͻ 0.05 compared with both other treatments.