Asymmetric Metabolism of Phosphatidylethanolamine in the Human Red Cell Membrane*

The incorporation of labeled fatty acids into phos- phatidylethanolamine (PE) on the two sides of the human red cell membrane was studied by use of the vectorial probe trinitrobenzene sulfonate (TNBS). A small population of PE molecules on the outer surface of the membrane has a 4-fold higher turnover rate than the remaining PE molecules. This effect is greatest with palmitic acid, less with linoleic and linolenic acids, and not seen with stearic acid. By use of the hydrophobic penetrating probe fluorodinitrobenzene (FDNB), we find a second larger population of PE and phosphati-dylserine (PS) molecules which reacts with FDNB and has a higher specific activity than the PE and PS molecules which do not react. With human polymorpho- nuclear cells, the labeled PE molecules inside the cell have a higher specific activity than the PE molecules located on the outer cell surface. These results suggest that there are heterogeneous populations of PE and PS on both halves of the red cell membrane which show different metabolic turnover rates of their fatty acids.

The incorporation of labeled fatty acids into phosphatidylethanolamine (PE) on the two sides of the human red cell membrane was studied by use of the vectorial probe trinitrobenzene sulfonate (TNBS).
A small population of PE molecules on the outer surface of the membrane has a 4-fold higher turnover rate than the remaining PE molecules. This effect is greatest with palmitic acid, less with linoleic and linolenic acids, and not seen with stearic acid. By use of the hydrophobic penetrating probe fluorodinitrobenzene (FDNB), we find a second larger population of PE and phosphatidylserine (PS) molecules which reacts with FDNB and has a higher specific activity than the PE and PS molecules which do not react. With human polymorphonuclear cells, the labeled PE molecules inside the cell have a higher specific activity than the PE molecules located on the outer cell surface. These results suggest that there are heterogeneous populations of PE and PS on both halves of the red cell membrane which show different metabolic turnover rates of their fatty acids.
The red cell membrane has an asymmetric distribution of its phospholipids (1-5). The question arises how this asymmetry is made and maintained and whether these phospholipids are differently metabolized and renewed. Renooij and van Golde (6) found that the specific activities of the more unsaturated classes of PC' of rat red cells labeled from P,:" in vivo were higher on the outer layer of the red cell membrane as compared to the inner layer, but that the disaturated PC had the highest activity on the inner layer. They proposed that the acylation activity of the erythrocyte was directed toward the formation of disaturated PC on the inner layer of the membrane. Renooij et al. (7,8) reported that the incorporation of different fatty acids into PC of isolated rat red cells occurred predominantly on the inner layer of the membrane. They state that two mechanisms may account for the renewal of phospholipids in the red cell membrane, these being the acylation of lysophospholipids which occurs in the membrane and the exchange of phospholipids between serum lipoproteins and the red cell membrane (9). Shohet (10, 11) also * This research was supported in part by funds from an Institutional Biomedical Research Support Grant RR05403 and National Institutes of Health Grant HBL02063. The costs of publication of this article were defrayed in part by the payment of page charges. This ance with 18 U.S.C. Section 1734 solely to indicate this fact. article must therefore be hereby marked "aduertisement" in accord- reported that these two processes occur in two different compartments in the red cell membrane.
In order to further explore the origin of the asymmetry of the red cell membrane, we have studied the incorporation of different fatty acids into the various phospholipids of the isolated human red cell. We have found an asymmetric labeling of P E on the two halves of the membrane by use of vectorial chemical probes. Our results show that a small population of P E molecules on the outer surface of the red cell membrane has a more rapid turnover than the other PE molecules either on the outer or inner surface of the membrane. We also provide evidence for heterogeneous populations of PE and PS in the red cell membrane. With human PMN cells, the labeled P E molecules inside the cell have a higher specific activity than the PE molecules on the outer cell surface.

MATERIALS AND METHODS
Human blood was obtained from the Red Cross Blood Bank and used within 2 weeks. The cells were washed three times with Krebs-Ringer phosphate (KRP) buffer, pH 7.4, containing 5 mM glucose and 1% BSA prior to use. The white cells and platelets were removed from the red cells by three washings in KRP buffer and centrifugation. Human PMN cells were prepared by the method of Boyum (12). The cells after appropriate incubation with labeled fatty acids or with chemical probes were lysed in 10 mM Tris buffer containing 1 mM EDTA by the method of Dodge et al. (13). The ghosts were extracted with chloroform/methanol by the Folch et al. method (14) and the lipids separated by two-dimensional or one-dimensional TLC (15,16).
The lipid spots were eluted with hot methanol/HCl and lipid P was determined by the method of Harris and Popat (17). The reaction of cells with TNBS or FDNB was carried out in 120 mM NaCl and 40 mM NaHCOn buffer, pH 8.5.

RESULTS
The nonpermeable probe TNBS was used to measure the specific activity of PE and PS on the outer surface of the red cell membrane. The PE and PS which do not react with TNBS at 0 "C are localized on the inner surface of the membrane and/or are tightly bound to membrane protein (18). The time course profiie in Fig. 1 shows that PE on the outer membrane surface has a much higher specific activity than the remaining PE in the membrane when intact cells are incubated with [3H]palmitate. These PE molecules represent only 1-2% of the total PE molecules. These results demonstrate an asymmetric rapid turnover of a small population of PE molecules. The amount of PS which reacts with TNBS at 0 "C is too small to measure. The labeling of red cells with TNBS was, therefore, done at 21 "C in order to react sufficient PS molecules for analysis. At 21 "C, there is slow penetration of TNBS into the cell so that saturation of PE labeling is not attained as it is at 0 "C ( Fig. 2). However, we did carry out some experiments with TNBS labeling at 21 "C in order to compare the results with those at 0 "C.
In order to see whether the PE and PS molecules localized on the inner membrane surface also have heterogz-~O U S pop-    Table I show that the PE and PS molecules which react with FDNB or TNBS have a higher specific activity than the remaining unreactive PE and PS molecules. Since FDNB reacts on both sides of the membrane, the increase in specific activity of the DNP-PE and possibly DNP-PS molecules is due in part to a small population of molecules localized on the outer membrane surface as seen in both Fig. 1 and Table I. However, since the extent of reaction of PE and PS with FDNB is much greater than that of TNBS, the results show that another population of PE and PS molecules on the inner surface of the membrane or more deeply buried in the membrane is accessible to FDNB and TNBS and has a higher turnover of their fatty acids than the remaining PE and PS molecules which do not react with these probes.
In order to see whether the labeling of PE and PS was due to acylation of lyso-PE and lyso-PS in the membrane, the red cells were pretreated with purified phospholipase A2 and then incubated with [3H]palmitate. The results in Table I1 show that phospholipase action led to a small decrease in the amount of PC, PE, and PS and that phospholipase enhanced the labeling of PC, PE, and PS. The results in Table I11 show that phospholipase treatment also enhanced the specific activity of the PE which reacted with TNBS from 309 to 371 pmol of [3H]palmitate incorporated/pmol of PE. These results suggest that the acylation of lyso-PE may account in part for the labeling of PE. Our previous studies (18) as well as the results in Table I1 show that phospholipase A2 treatment has no significant effect on the extent of reaction of PE with TNBS.
Since C3H]palmitate was incorporated into PE in an asymmetric manner, we tested I4C-labeled stearate, linoleate, and linolenate to see whether these fatty acids behave like palmitate. The data in Table IV show that palmitate, linoleate, and linolenate had a more rapid turnover in TNP-PE as compared to unreacted PE, whereas stearate did not. The  Effect ofphospholipase AI treatment on the content and labeling of red cell phospholipids Twenty aliquots of 0.5 ml of packed red cells were incubated in 19.5 ml of KRP buffer, pH 7.4, containing 5 mM glucose and 1% BSA. Ten samples served as controls without phospholipase AI treatment and 10 samples were treated with phospholipase AI for 1 h at 37 "C. The cells were washed and resuspended in 20 ml of KRP buffer containing 50 pCi of [3H]palmitate and incubated for 1 h at 37 "C. The cells were washed twice with KRP buffer. The samples were suspended in 20 ml of 120 mM NaCl and 40 mM NaHC03, pH 8.5, containing 2 mM TNBS and reacted for 1 h at 21 "C. The cells were washed once with KRP buffer, ghosts prepared, and lipids extracted and analyzed by TLC. Incorporation values are the mean f S. D. for n = 5. The nanomole values are averages of duplicate lipid P analyses.    ) were added to two tubes each. All tubes were incubated at 37 "C for 1 h. The cells were centrifuged, washed twice with KRP buffer, and resuspended in 20 ml of 120 mM NaCl and 40 mM NaHC03, pH 8.5, containing 2 mM TNBS and incubated at 0 "C for 1 h. The cells were washed once with KRP buffer, ghosts were prepared, and lipid extracted and analyzed by TLC. Values are the mean * S. D. for n = 4. greatest effect was seen with palmitate. The red cell, therefore, can discriminate between these fatty acids. As seen in Table  IV, the per cent of total PE which reacted with TNBS was the same in 4 samples. However, the total amount of 14Cfatty acid which was incorporated was highest with ["C] palmitate and lowest with [14C]stearate even though only 17 as assessed by the TNBS reaction at 0 "C 1.3 x lo7 PMN cells were incubated with 10 pCi of [3Hlpalmitic acid in 9 ml of KRP buffer, pH 7.4, containing 5 mM glucose and 1% BSA for 1 h at 37 "C. The cells were washed twice with KRP buffer and suspended in ice-cold 2 ml of 120 mM NaCl and 40 mM NaHC03 buffer, pH 8.5, containing 2 mM TNBS. The cells were reacted for 1 h at 0 "C, washed with KRP buffer, and the lipids were extracted and analyzed by TLC. Values are the mean f S. D. for n = 2. To see whether hormones influenced the asymmetric labeling of PE from ['Hlpalmitate, experiments were carried out in which the effect of isoproterenol and insulin was tested. These hormones had no effect on the enhanced labeling of TNP-PE from [3H]palmitate as compared to controls (data not provided).
It is possible that the incorporation of the fatty acids into PE and PS is due in part to small amounts of contaminating white cells on platelets. The red cells were washed three times to remove white cells and platelets. As a control study, we prepared and purified PMN cells and incubated them with ['Hlpalmitate for 1 h at 37 "C followed by TNBS labeling at 0 "C for 1 h. The results in Table V show that unlike the red cell, the TNP-PE had a lower specific activity than the unreacted PE. TNBS reaction of PE at 0 "C with PMN cells was much higher than that of red cells (25% as compared to 1-2%). Since PMN cells contain several internal membranes, the unreacted 'H-labeled PE molecules can arise from any or all of these membranes, whereas the TNP-PE we believe represents PE molecules on the outer surface of the plasma membrane. These results with PMN cells make it unlikely that the different asymmetric labeling pattern in red cells is due to white cell contamination.

DISCUSSION
The incorporation of fatty acids into red cell membrane phospholipids has been reported to occur in part by acylation of lysophospholipids (7,8). However, this is not the only pathway for incorporation of fatty acids into red cell phospholipids since exchange reactions can also occur (9). The asymmetric incorporation of fatty acids into PC on the inside of the membrane has been reported by Renooij et al. (7,8). There may be other processes which have to be considered in regard to the turnover and metabolism of red cell membrane phospholipids. Do certain phospholipids have a different asymmetric metabolism and are there discrete populations of phospholipids on either surface of the membrane which have a different turnover than other phospholipids of the same type? We studied this problem by labeling PE and PS in intact red cells with radioactive fatty acids and using vectorial probes to sense certain populations of PE and PS and analyzing their specifk activities. We show that a small population of PE molecules on the outer surface of the red cell membrane has a larger turnover rate than the other PE molecules in the membrane. This was demonstrated by use of TNBS to label PE on the outer surface of the membrane (2, 18). The asymmetric labeling pattern of PE was most prominent when the TNBS reaction was done at 0 "C under conditions where TNBS does not penetrate into the cell. We also show by use of the penetrating hydrophobic probe FDNB that other pop-

Phosphatidylethanolamine Turnover in Red Cells
ulations of PE and PS molecules which are reactive with FDNB have a higher turnover rate than the PE and PS molecules which do not react with FDNB. Therefore, there are heterogeneous domains of PE and PS on either side of the membrane which are metabolized differently. Evidence for heterogeneous domains of PE has been obtained earlier in our studies on the effect of TNBS and FDNB on K' and P, transport in red cells (19). Our studies provide further evidence to an earlier finding of Rowe (20) who found two ethanolamine-phospholipid fractions in human red cells which differed in their rates of incorporation of ' ' ' Pi and [I4C]acetate.
The significance of our findings to the origin and maintenance of the asymmetry of phospholipids in the red cell membrane can be only speculative at this time. PE has been shown to be asymmetrically distributed in the red cell membrane, being localized primarily on the inner half of the membrane (19). We find that a very small population of P E molecules on the outer half of the membrane has a 4-fold more rapid turnover of their fatty acids than the remaining PE molecules. On the other hand, Renooij et al. (7,8) found that PC which is localized primarily on the outer half of the membrane has a population of molecules on the inner half of the membrane which has a more rapid turnover of their fatty acids. The enzymes responsible for the incorporation of fatty acids into these phospholipids may be asymmetrically arranged in the membrane. These enzymes which incorporate fatty acids into PE may show a preference for palmitic acid over stearic, linoleic, and linolenic acid. Whether PE molecules rich in palmitic acid have a different disposition in the membrane remains to be determined. The effect of fatty acid composition on the physical and biochemical properties of the various phospholipid classes in cell membranes is not well understood. Do those phospholipids with certain fatty acids interact differently with certain membrane proteins which somehow regulate their metabolism and transmembrane mobility? It is also possible that the fatty acids are incorporated into PE molecules on the inner surface of the membrane and that these newly synthesized molecules are rapidly translo-cated to the outer surface. If this is so, then it remains to be explained what mechanisms allow for the preferential translocation.