150
Phospholipid Composition of Cultured Human Endothelial Cells Eric J. Murphy a,*, Laurie Joseph b,1, Ralph Stephens b and Lloyd A. Horrocks a aDepartment of Medical Biochemistry and Neuroscience Program and bDepartment of Pathology, The Ohio State University, Columbus, Ohio 43210
Detailed analyses of the phospholipid compositions of cultured human endothelial cells are reported here. No significant differences were found between the phospholipid compositions of cells from human artery, saphenous and umbilical vein. However, due to the small sample sizes, relatively large standard deviations for some of the phospholipid classes were observed. A representative composition of endothelial cells is: phosphatidylcholine 36.6%, choline plasmalogen 3.7%, phosphatidylethanolamine 10.2%, ethanolamine plasmalogen 7.6%, sphingomyelin 10.8%, phosphatidylserine 7.1%, lysophosphatidyleholine 7.5%, phosphatidylinositol 3.1%, lysophosphatidylethanolamine 3.6%, phosphatidylinositol 4,5-bisphosphate 1.8%, phosphatidic acid 1.9%, phosphatidylinositol 4phosphate 1.5%, and cardiolipin 1.9%. The cells possess high choline plasmalogen and lysophosphatidylethanolamine contents. The other phospholipids are within the normal biological ranges expected. Phospholipids were separated by high-performance liquid chromatography and quantified by lipid phosphorus assay. Lipids 27, 150-153 (1992).
(9). These decreases in PtdCho may be due to the hydrolysis of choline plasmalogen (7,8) or the receptor mediated breakdown of PtdCho (10,11). Regardless, these receptor mechanisms all involve the degradation of membrane phospholipids resulting in the release of arachidonic acid for eicosanoid formation (12-17). Partial phospholipid compositions of endothelial cells have been reported for different species (5,18-25). However, in order to better understand the role of phospholipids in receptor function, the complete and detailed phospholipid composition of endothelial cells needs to be known. In the present study, the phospholipid compositions of human endothelial cells from three vascular sources were compared in order to determine whether phospholipid composition is dependent upon the vascular source This report is the first to describe the simultaneous extraction and high-performance liquid chromatography (HPLC) separation of plasmalogens and polyphosphoinositides from cultured cells.
MATERIALS AND METHODS Phospholipids, and particularly the polyphosphoinositides, play an important role in endothelial cell receptor function. Stimulation of the bradykinin receptor has been shown to affect polyphosphoinositide turnover in endothelial cells (1,2). This turnover is mediated by the combined action of phospholipase C and diacylglycerol lipase on the polyphosphoinositides, thereby releasing arachidonic acid (3,4). Several alternative pathways have been proposed for arachidonic acid release in endothelial cells. These include the action of phospholipase A2 (5,6) and a combined calcium independent phospholipase A1 and lysophospholipase pathway (4). In other cell systems, decreases in choline plasmalogen have been linked to bradykinin receptor activation (7). Plasmalogen hydrolysis has been implicated in several receptor mechanisms which result in release of arachidonic acid (8). Phosphatidylcholine (PtdCho) is a major contributor of arachidonic acid following thrombin stimulation (6) or stimulation by the calcium ionophore A23187 *To whomcorrespondenceshouldbe addressed at the Department of Medical Biochemistry, The Ohio State University, 1645 Nell Avenue, Room 471, Columbus, OH 43210. 1Present address: Department of Orthopedic Surgery,ThomasJefferson University, Philadelphia, PA 19107. Abbreviations: CerPCho,sphingomyelin;Gpl, glycerophospholipid; HAE, human arterial endothelial;HPLC, high-performanceliquid chromatography;HSVE, humansaphenousveinendothelial;HUVE, human umbilical vein endothelial; lysoPtdCho,lysophosphatidylcholine;lysoPtdEtn,lysophosphatidylethanolamine;PlsCho,choline plasmalogen; PtdCho, phosphatidylcholine;PlsEtn, ethanolamine plasmalogen; PtdEtn, phosphatidylethanolamine;Ptd2Gro, cardiolipin; PtdIns, phosphatidylinositol;PtdIns4P phosphatidylinositol 4-phosphate; PtdIns(4,5)P2, phosphatidylinositol4,5-bisphosphate; PtdOH, phosphatidic acid; PtdSer, phosphatidylserine;TLC,thinlayer chromatography. LIPIDS, Vol. 27, no. 2 (1992)
Three vascular sources were used to establish separate human endothelial cell cultures. Cells were isolated from cadaver aorta artery, from bypass patient's saphenous vein, and from newborn umbilical veins by treating the luminal surface with 0.1% type II collagenase (26). The cells were plated on human fibronectin coated glass tissue culture flask (23 cm 2) containing M-199 media Gibco (Gaithersburg, MD) supplemented with 10% fetal bovine serum, 150 ~g/mL endothelial cell growth factor (27), 90 ~g/mL Na-heparin (28), and antibiotics. The ceils in a flask upon reaching con fluency were split and plated on 100-ram glass cell culture plates. Ceils were used from passage 5 through 12. Their endothelial origin was confirmed by the presence of factor VIII antigen (29). Confluent cells were extracted using n-hexane/2-propanol (3:2, v/v) (30). The use of this extraction method avoids the use of toxic solvents, decreases the amount of extractable protein, and avoids the loss of water-soluble phospholipids. Prior to extraction the medium was removed and the cells were washed with two 3-mL portions of cold PBS buffer to remove traces of medium. The plates were immediately placed upon dry ice prior to extraction and frozen to minimize membrane damage and activation of acylhydrolases during extraction (31). Two 3-mL aliquots of n-hexane/2-propanol (3:2, v/v) were used to extract the lipids from the cells. The initial aliquot was added to the frozen cells which were subsequently removed from the plate by using a Teflon cell scraper. The second aliquot was used to rinse the plate. Most cellular debris was removed by filtration through glass wool. The extracts were kept under N2 to minimize auto-oxidation of lipids. Extracts were stored at - 2 0 ~ Prior to phospholipid separation, the cells were subjected to ultrafiltration using a Rainin 0.2 gm Nylon filter (Woburn, MA). The sample was dried under nitrogen and redissolved in a known
151
COMMUNICATION volume of n-hexane/2-propanol/water (3:2, + 5.5% by volume) prior to high-performance liquid chromatography (HPLC). The cell extracts were separated into major phospholipid classes by HPLC. Solvents used were H P L C grade nhexane and 2-propanol from E.M. Science (Cherry Hill, NJ). Solvents were filtered through a 0.5-{nn Millipore FHt y p e Nylon filter (Bedford, MA) and degassed. Solvent A was n-hexane/2-propanol (3:2, v/v), solvent B was nhexane/2-propanol/water (3:2 + 5.5% by volume). Water was purified using a Millipore water purification system. The H P L C i n s t r u m e n t consisted of two Altex (Berkeley, CA) 100A pumps, an Altex 420/421 controller and an Altex model 210 injection port. The D u p o n t Zorbax Silica column (4.6 m m • 250 m m 5-6 tin1; Wilmington, DE) was maintained at a constant t e m p e r a t u r e of 34~ using a Jones chromatography heating block (Columbus, OH). An ISCO (Lincoln, NE) V4 ultraviolet (UV) variable wave length detector was used to detect p e a k s at 205 nm. The chromatographic procedure (32) permitted the separation of the m a j o r phospholipids including the separation of the acidic phospholipids phosphatidylserine (PtdSer) and phosphatidylinositol (PtdIns) as well as the resolution of lysophosphatidylethanolamine (lysoPtdEtn). The polyphosphoinositides were also separated (33). Retention times for each phospholipid class were according to polarity with the elution order being from least polar to m o s t polar. The lower limit of sensitivity was 100 nmol of injected lipid phosphorus. Plasmalogens were separated by H P L C after acidic hydrolysis (34). The ethanolamine glycerophospholipid (EtnGpl) and choline glycerophospholipid (ChoGpl) peaks were collected, dried under N2, and then inverted over 5 drops of concentrated hydrochloric acid in a test tube cap for two minutes. This caused hydrolysis of the alkenyl ether bond of the plasmalogens while the alkylacyl and diacyl bonds remained intact. The sample was reextracted and rechromatographed utilizing the same separation procedure as used in the initial separation. The alkylacyl and diacyl fractions were then collected and subjected to alkaline hydrolysis in 2 m L of 0.1 M N a O H in methanol. After 15 min the reaction was terminated by addition of 1 m L of ethyl formate (35). Cleavage of the alkali-labile ester linkage caused formation of glycerophosphocholine or glycerophosphoethanolamine and of alkylglycerophosphocholine or alkylglycerophosphoethanolamine. The products were separated following extraction of the methanol p h a s e with chloroform/isobutanol/water (4:2:3, v/v/v) followed by centrifugation to facilitate phase separation. The top layer contained glycerophosphocholine or glycerophosphoethanolamine, the b o t t o m layer contained the ether lysophospholipids. Phospholipid classes were q u a n t i t a t e d by m e a s u r i n g lipid phosphorus (36). RESULTS
The compositions of h u m a n saphenous vein (HSVE), h u m a n umbilical vein (HUVE) and h u m a n arterial endothelial cells (HAE) are reported in Table 1. L y s o P t d E t n represented an abnormally high mole percentage of the total lipid phosphorus in the cells from all sources. The polyphosphoinositide values are also reported. No statistically significant differences in the proportions of
TABLE 1 Phospholipid Composition of Cultured Human Endothelial Cellsa
Ptd2Gro PtdEtn+ PtdOH PtdIns lysoPtdEtn PtdSer PtdCho+ CerPCho Ptdlns4P lysoPtdCho Ptdlns4,5P 2
HUVE n = 12 • S.D.
HSVE n = 10 X S.D.
2.60 _ 2.88 17.63 • 1.98 2.37 +- 3.09 1.90 • 3.09 3.12 • 3.36 5.36 • 2.70 42.76 - 6.26 8.91 _ 2.69 1.56 _+ 1.75 7.53 _ 5.04 1.24 • 0.94
2.06 • 1.05 16.86 -+ 3.11 1.81 +.1.64 2.99 • 2.21 5.26 - 2.22 8.86 - 1.76 39.03 • 6.56 10.52 _+ 2 . 3 9 1.19 -- 0.94 8.07 • 4.68 1.99 +- 1.05
HAE n ---- 12 • S.D. 1.08 • 19.06 • 1.56 • 4.26 • 2.54 • 7.15 • 39.12 • 12.96 • 1.90 • 7.02 • 2.09 •
0.85 3.78 0.76 2.31 2.07 2.87 3.92 5.04 1.83 4.84 1.26
aValues are mole % of total phospholipid. Tukey's multiple comparison test was used for statistical evaluation. +PtdEtn and PtdCho represent the sum of alkenylacyl, alkylacyl and diacyl fractions. Abbreviations are: CerPCho, sphingomyelin; HAE, human arterial endothelial; HSVE, human saphenous vein endothelial; HUVE, human umbilical vein endothelial; lysoPtdCho, lysophosphatidylcholine; lysoPtdEtn, lysophosphatidylethanolamine; PlsCho, choline plasmalogen; PtdCho, phosphatidylcholine; PlsEtn, ethanolamine plasmalogen; PtdEtn, phosphatidylethanolamine; Ptd2Gro, cardiolipin; PtdIns, phosphatidylinositol; PtdIns4P, phosphatidylinositol 4-phosphate; PtdIns(4,5)P2, phosphatidylinositol 4,5-bisphosphate; PtdOH phosphatidic acid; PtdSer phosphatidylserine.
TABLE 2 Composition of Ethanolamine and Choline Glycerophospholipid SubclasSes a
Diacyl Alkenylacyl Alkylacyl Diacyl Alkenylacyl Alkylacyl
EtnGpl class (%)
Total Gpl (%)
45.2 _+ 1.4 40.5 +_ 1.9 14.3 _ 0.5
8.0 7.1 2.5
ChoGpl 78.4 11.6 10.0
class (%) +_ 1.1 • 0.9 _ 1.6
Total Gpl (%) 33.5 5.0 4.3
aThe values are means _+ standard deviations. For EtnGpl and ChoGpl, n = 4. Abbreviations are: EtnGpl, ethanolamine glycerophospholipid; ChoGpl, choline glycerophospholipid; and Gpl, glycerophospholipid.
phospholipid classes were found between the various sources of endothelial cells. The m o s t i m p o r t a n t finding is the high l y s o P t d E t n level seen in all three sources. Because no differences were found between the m a j o r phospholipids from various sources, the plasmalogen content of only the H U V E cells was determined. The P l s E t n content was 40.5% of the E t n G p l and the PlsCho content was 11.6% of ChoGpl. These values represent 7.1% and 5.0% respectively of the total lipid phosphorus. The compositions of the choline and ethanolamine glycerophosph~ lipid subclasses are listed in Table 2. The a m o u n t of lipid phosphorus injected onto the colu m n was at or below the desirable lower limit of sensitivity (100 nmol) due to the small sample sizes available from LIPIDS, Vol. 27, no. 2 (1992)
152
COMMUNICATION the cultures. Thus, the variability between samples increases especially for phospholipid classes that represent a small percentage of total lipid phosphorus, DISCUSSION The purpose of this study was to determine the phospholipid composition of human endothelial cells in culture and the dependence of this composition on the vascular source The separation of phospholipids by H P L C allows for good recovery of all phospholipids, improved phosphorus determination and detection of smaller changes in phospholipid composition with statistical significance due to a lower variance than based on thin-layer chromatography (TLC) (37}. The extraction procedure used permits a good recovery of phospholipids while minimizing the amount of protein in the extract. The procedure increases polyphosphoinositide recovery and avoids the need for a twophase system which can cause losses of lysoglycerophospholipids (30}. Freezing the cells prior to extraction decreases the mechanical trauma incurred by the cells during their removal from the plates (31). Hence~ our values should represent more closely the actual composition of the cells in culture Several papers have reported partial phospholipid compositions of endothelial cells from various vascular sources and species (5,19-25). A composition of 25.3% P t d E t n , 36.3% PtdCho and 12.8% CerPCho in H U V E cells has been reported by Rastogi and Nord~y (19), which is similar to our data. A plasmalogen content in H U V E cells of 42.5% P l s E t n in E t n G p l and 6.8% PlsCho in ChoGpl has been reported by Blank e t al. (22). However, their values for the choline glycerophospholipid diacyl (88.6%) and alkylacyl (4.6%} subclasses differ from our values of 78.4% and 10%, respectively. The reported values for the two ethanolamine glycerophospholipid subclasses were 52.6% for the diacyl and 4.9% for the alkylacyl fractions (22). Our values were 45.3% and 14.3% for the diacyl and alkylacyl subclasses, respectively. These differences may reflect the different methodologies used to quantitate the E t n G p l and ChoGpl subclasses. Values for the proportions of choline and ethanolamine glycerophospholipid subclasses reported by Takamura e t al. (25) also differ from our values. The differences could be due to the different methodology used to separate the subclasses. The reported phospholipid compositions for b o t h umbilical vein and arterial endothelial cells (25) are also considerably different from our values. Phospholipid separations by TLC usually have lower recoveries t h a n H P L C separations (our recoveries are 98.3% + 3), and TLC may not completely resolve all classes of phospholipids because of the lower number of theoretical plates. Values for P t d E t n and PtdCho composition from rat and human brain endothelial cells (24) compare closely to our values, although a difference exists in CerPCho contents with 20.4% and 17.0% for rat and human brain, respectively, compared with our value of 10.8%. The difference may be due the fact t h a t the cells were freshly isolated, and not cultured. Isolated rat brain endothelial cells have 46.3% P l s E t n in E t n G p l (18), which is close to our value The polyphosphoinositide content of endothelial cells has not been reported previously for any species. Guinea pig brain contains 0.58% P t d I n s 4 P and 2.58% P t d I n s (4,5)P 2 (38). Our values are within this r a n g e Our LIPIDS, Vol. 27, no. 2 (1992)
data compare favorably to several literature values for the partial phospholipid composition of endothelial cells from various species and vascular sources and are the first to include polyphosphoinositides. The cells exhibited a high l y s o P t d E t n and a relatively high lysoPtdCho content. The position of the f a t t y acid was not determined. Regardless of f a t t y acid position, these lysophospholipids are known to have detergent properties on cellular membranes (39): High concentrations of endogenous lysophospholipids are generally assumed to increase membrane fluidity and to give rise to subtle c h a n g e s in cellular surface. The high levels in human endothelial cells may be related to the apparent ease of thrombosis (40). Increases in membrane fluidity may also cause the cell membrane to be more susceptible to mechanical damage accounting for the observed morphm logical changes such as crater formation in endothelial cells (41}. High levels of lysolipid may modulate the effect of injury and interaction of the endothelium with various hormones (42) because lysophospholipids can alter the functioning of membrane-bound enzymes (39,43). Endothelial cells play a major role in vascular homeostasis (40). In order to understand the lipid metabolism of endothelial cells, the basic phospholipid composition must be known. It has been hypothesized t h a t lipid composition varies with vascular source based on observations of ultrastructural differences depending on vessel size and location (19). In this study no dependence of phospholipid composition on vascular source was found. However, it should be noted t h a t in some phospholipid classes the standard deviations were high. This is especially true for phospholipids which make up a small percentage of the total phospholipids. Differences in phospholipid classes molecular species dependent upon vascular source were not ruled out. ACKNOWLEDGMENTS We thank Diana Carter for her help in the preparation of the manuscript, Patrick Vaccarofor providing the tissues, Mark Armeni for assistance in cell culture and extraction procedures, Lynda Tit~ terington for growing the cells, Robert Hancock for assistance in the statistical calculations, and Laura Dugan for her advice with the HPLC separations. This work was supported in part by National Institutes of Health research grant NS-10165. REFERENCES 1. Derian, C.K., and Moskowitz, M.A. (1986) J. BioL Chem. 261, 3831-3837. 2. Derian, C.K., and Moskowitz, M.A. (1985) Fed. Proa 44, 4862. 3. Moscat, J., Moren~ F., Herrero, C., Iglesias, S., and GorciaBaz:reno, P. (1986} Biochem. Biophys. Res. Commun. 139, 1098-1103. 4. Martin, T.W., and Wysolmerski, R.B. (1987)J. Biol. Chem. 262, 13086-13092. 5. Hong, S.L., and Deykin, D. (1982)J. BioL Chem. 257, 7151-7154. 6. Thomas, J.M.F., Hullin, F., Chap, H., and Douste-Blazy,L. (1984) Thromb. Res. 34, 117-123. 7. Horrocks, L.A., Yec~Y.K., Harder, H.W., Mozzi, R., and Goracci, G. (1986) in Advances in Cyclic Nucleotide and Protein Phosphorylation Research (Greengard, E, and Robinson, G.A., eds.) pp. 263-292, Raven Press, New York. 8. Horrocks, L.A., Harder, H.W., Mozzi, R., Goracci, G., Francescangeli, E., Porcellati, S., and Nenci, G.G. (1986) in Enzymes of Lipid Metabolism (Freysz, L., Dreyfus, H., Massarelli, R., and Gatt, S., eds.) Vol. 2, pp. 707-711, Plenum Press, New York.
153
COMMUNICATION 9. Thomas, J.M.F., Chap, H., and DousteBlazy, L. (1981) Biochem. Biophys. Res. Commun. 103, 819-824. 10. L6ffelholz, K. (1989) Blochem. Pharmacol. 38, 1543-1549. 11. Pelech, S.L., and Vance, D.E. (1989) Trends Biochem. Sci. 14, 28-30. 12. Marcus, A.J., Weksler, B.B., and Jaffe, E.A. (1978)J. BioL Chem. 253, 7138-7141. 13. Nawroth, P.O., Stern, D.M., and Dietrich, M. (1985)Blur50, 7-11. 14. Ingerman-Wojenski, C., Silver, M.J., Smith, J.B., and Macarck, E. (1981)J. Clir~ Invest. 67, 1292-1296. 15. Yamaja Setty, B.N., Stuart, M.J., and Walenga, R.W. (1985) Biochim. Biophys. Acta 833, 484-494. 16. Gormam R.R., Oglesby, T.D., Bundy, G.L., and Hopkins, N.K. (1985) Platelets and Vascular Occlusion 72, 708-712. i7. Buchanan, M.R., Butt, R.W., Magas, Z., Van Ryn, J., Hirsh, J., and Nazir, D.J. (1985) Thrombosis andHaemostasis 53, 306-311. 18. Selivonchick, D.P., and Roots, B.I. (1976) Lipids 12, 165-169. 19. Rastogi, B.K., and Nord4y, A. (1980) Thromb. Res. 18, 629-641. 20. Matheson, D.F., Oei, R., and Roots, B.I. (1980) Neurochem. Res. 5, 43-59. 21. Wey, H.E., Jakubowski, J.A., and Deykin, D. (1986) Biochim. Biophys. Acta 878, 380-386. 22. Blank, M.L., Spector, A.A., Kaduce, T.L., and Snyder, F. (1986) Blochim. Biophys. Acta 877, 211-215. 23. Matheson, D.F., Oei, R., and Roots, B.I. (1980) Neurochem. Res. 5, 683-695. 24. Siakotos, A.N., Rouser, G., and Fleischer, S. (1969) Lipids 4, 234-239. 25. Takamura, H., Kasai, H., Arita, H., and Kito, M. (1990) J. Lipid Res. 31, 709-717. 26. Gimbrone, M.A., Cotran, R.S~, and Folkman, J. (1974)J. CeUBioL 60, 673-684. 27. Maciag, T., Cerundolo, J., Ilsley, S., Kelly, P.R., and Forand, R. (1979) Proc. Natl. Acad. ScL USA 76, 5674-5678.
28. Thornton, S.C., Mueller, S.N., and Levine, E.M. (1983) Science 222, 623-624. 29. Jaffe" E.A., Hoyer, L.W., and Wachmar, R,L. (1973) J. Clin. Invest. 52, 2757-2764. 30. Hara, A., and Radin, N.S. (1978) Anal Biochem. 90, 420-426. 31. Demediuk, P., Anderson, D.K., Horrocks, L.A., and Means, E.D. (1985) In Vitro Cell. Develop. BioL 21, 569-574. 32. Dugan, L.L., Demediuk, P., Pendley, II, C.E., and Horrocks, L.A. (1986) J. Chromatogr. 378, 317-327. 33. Dugan, L.L., Bazan, N.G., and Horrocks, L.A. (1986) Trans. Am. Soc. Neurochem. 17, 138. 34. Murphy, E.J., Stephens, R., Jurkowitz, M.S., and Horrocks, L.A. (1987) J. Neurochem. 48, 560-B. 35. Horrocks, L.A., and Sun, G.Y. (1972) in Research Methods in Neurochemistry (Rodnight, R., and Marks, N., eds.) Vol. I, pp. 223-231, Plenum Press, New York. 36. Rouser, G., Siakotos, A., and Fleischer, S. (1969)Lipids 1, 85-86. 37. Horrocks, L.A., Dugan, L.L., Flynn, C.J., Goracci, G., PorceUati, S., and Yec~Y.K. (1987) in Lecithin." Technologica~ Biological and Therapeutic Aspects, (Ansell, G.B., and Hanin, I., eds.) pp. 3-16, Plenum Press, New York. 38. Hawthorne, J.N., and Kai, M. (1970) in Handbook of Neurochemistry (Lajtha, A., eeL)pl~ 491-508, Plenum Press, New York. 39. Weltzien, H.U., Richter, G., and Ferber, E. (1979) J. BioL Cherr~ 254, 3652-3657. 40. Moncada, S., and Vane, J.R. (1979) N. Engl. J. Med. 300, 1142-1147. 41. Svendson, W., Dalen, H., Moland, J., and Engodal, H. (1985) Arteriosclerosis 58, 27-37. 42. Ryan, U.S., and Ryan, J.W. (1985) Chest 88, 2035-2075. 43. Weltzien" H.U. (1979) Biochim. Biophys. Acta 559, 259-287. [Received July 9, 1990, and in revised form December 27, 1991; Revision accepted December 28, 1991]
LIPIDS, Vol. 27, no. 2 (1992)
Phospholipid Composition of Cultured Human Endothelial Cells Eric J. Murphy a,*, Laurie Joseph b,1, Ralph Stephens b and Lloyd A. Horrocks a aDepartment of Medical Biochemistry and Neuroscience Program and bDepartment of Pathology, The Ohio State University, Columbus, Ohio 43210
Detailed analyses of the phospholipid compositions of cultured human endothelial cells are reported here. No significant differences were found between the phospholipid compositions of cells from human artery, saphenous and umbilical vein. However, due to the small sample sizes, relatively large standard deviations for some of the phospholipid classes were observed. A representative composition of endothelial cells is: phosphatidylcholine 36.6%, choline plasmalogen 3.7%, phosphatidylethanolamine 10.2%, ethanolamine plasmalogen 7.6%, sphingomyelin 10.8%, phosphatidylserine 7.1%, lysophosphatidyleholine 7.5%, phosphatidylinositol 3.1%, lysophosphatidylethanolamine 3.6%, phosphatidylinositol 4,5-bisphosphate 1.8%, phosphatidic acid 1.9%, phosphatidylinositol 4phosphate 1.5%, and cardiolipin 1.9%. The cells possess high choline plasmalogen and lysophosphatidylethanolamine contents. The other phospholipids are within the normal biological ranges expected. Phospholipids were separated by high-performance liquid chromatography and quantified by lipid phosphorus assay. Lipids 27, 150-153 (1992).
(9). These decreases in PtdCho may be due to the hydrolysis of choline plasmalogen (7,8) or the receptor mediated breakdown of PtdCho (10,11). Regardless, these receptor mechanisms all involve the degradation of membrane phospholipids resulting in the release of arachidonic acid for eicosanoid formation (12-17). Partial phospholipid compositions of endothelial cells have been reported for different species (5,18-25). However, in order to better understand the role of phospholipids in receptor function, the complete and detailed phospholipid composition of endothelial cells needs to be known. In the present study, the phospholipid compositions of human endothelial cells from three vascular sources were compared in order to determine whether phospholipid composition is dependent upon the vascular source This report is the first to describe the simultaneous extraction and high-performance liquid chromatography (HPLC) separation of plasmalogens and polyphosphoinositides from cultured cells.
MATERIALS AND METHODS Phospholipids, and particularly the polyphosphoinositides, play an important role in endothelial cell receptor function. Stimulation of the bradykinin receptor has been shown to affect polyphosphoinositide turnover in endothelial cells (1,2). This turnover is mediated by the combined action of phospholipase C and diacylglycerol lipase on the polyphosphoinositides, thereby releasing arachidonic acid (3,4). Several alternative pathways have been proposed for arachidonic acid release in endothelial cells. These include the action of phospholipase A2 (5,6) and a combined calcium independent phospholipase A1 and lysophospholipase pathway (4). In other cell systems, decreases in choline plasmalogen have been linked to bradykinin receptor activation (7). Plasmalogen hydrolysis has been implicated in several receptor mechanisms which result in release of arachidonic acid (8). Phosphatidylcholine (PtdCho) is a major contributor of arachidonic acid following thrombin stimulation (6) or stimulation by the calcium ionophore A23187 *To whomcorrespondenceshouldbe addressed at the Department of Medical Biochemistry, The Ohio State University, 1645 Nell Avenue, Room 471, Columbus, OH 43210. 1Present address: Department of Orthopedic Surgery,ThomasJefferson University, Philadelphia, PA 19107. Abbreviations: CerPCho,sphingomyelin;Gpl, glycerophospholipid; HAE, human arterial endothelial;HPLC, high-performanceliquid chromatography;HSVE, humansaphenousveinendothelial;HUVE, human umbilical vein endothelial; lysoPtdCho,lysophosphatidylcholine;lysoPtdEtn,lysophosphatidylethanolamine;PlsCho,choline plasmalogen; PtdCho, phosphatidylcholine;PlsEtn, ethanolamine plasmalogen; PtdEtn, phosphatidylethanolamine;Ptd2Gro, cardiolipin; PtdIns, phosphatidylinositol;PtdIns4P phosphatidylinositol 4-phosphate; PtdIns(4,5)P2, phosphatidylinositol4,5-bisphosphate; PtdOH, phosphatidic acid; PtdSer, phosphatidylserine;TLC,thinlayer chromatography. LIPIDS, Vol. 27, no. 2 (1992)
Three vascular sources were used to establish separate human endothelial cell cultures. Cells were isolated from cadaver aorta artery, from bypass patient's saphenous vein, and from newborn umbilical veins by treating the luminal surface with 0.1% type II collagenase (26). The cells were plated on human fibronectin coated glass tissue culture flask (23 cm 2) containing M-199 media Gibco (Gaithersburg, MD) supplemented with 10% fetal bovine serum, 150 ~g/mL endothelial cell growth factor (27), 90 ~g/mL Na-heparin (28), and antibiotics. The ceils in a flask upon reaching con fluency were split and plated on 100-ram glass cell culture plates. Ceils were used from passage 5 through 12. Their endothelial origin was confirmed by the presence of factor VIII antigen (29). Confluent cells were extracted using n-hexane/2-propanol (3:2, v/v) (30). The use of this extraction method avoids the use of toxic solvents, decreases the amount of extractable protein, and avoids the loss of water-soluble phospholipids. Prior to extraction the medium was removed and the cells were washed with two 3-mL portions of cold PBS buffer to remove traces of medium. The plates were immediately placed upon dry ice prior to extraction and frozen to minimize membrane damage and activation of acylhydrolases during extraction (31). Two 3-mL aliquots of n-hexane/2-propanol (3:2, v/v) were used to extract the lipids from the cells. The initial aliquot was added to the frozen cells which were subsequently removed from the plate by using a Teflon cell scraper. The second aliquot was used to rinse the plate. Most cellular debris was removed by filtration through glass wool. The extracts were kept under N2 to minimize auto-oxidation of lipids. Extracts were stored at - 2 0 ~ Prior to phospholipid separation, the cells were subjected to ultrafiltration using a Rainin 0.2 gm Nylon filter (Woburn, MA). The sample was dried under nitrogen and redissolved in a known
151
COMMUNICATION volume of n-hexane/2-propanol/water (3:2, + 5.5% by volume) prior to high-performance liquid chromatography (HPLC). The cell extracts were separated into major phospholipid classes by HPLC. Solvents used were H P L C grade nhexane and 2-propanol from E.M. Science (Cherry Hill, NJ). Solvents were filtered through a 0.5-{nn Millipore FHt y p e Nylon filter (Bedford, MA) and degassed. Solvent A was n-hexane/2-propanol (3:2, v/v), solvent B was nhexane/2-propanol/water (3:2 + 5.5% by volume). Water was purified using a Millipore water purification system. The H P L C i n s t r u m e n t consisted of two Altex (Berkeley, CA) 100A pumps, an Altex 420/421 controller and an Altex model 210 injection port. The D u p o n t Zorbax Silica column (4.6 m m • 250 m m 5-6 tin1; Wilmington, DE) was maintained at a constant t e m p e r a t u r e of 34~ using a Jones chromatography heating block (Columbus, OH). An ISCO (Lincoln, NE) V4 ultraviolet (UV) variable wave length detector was used to detect p e a k s at 205 nm. The chromatographic procedure (32) permitted the separation of the m a j o r phospholipids including the separation of the acidic phospholipids phosphatidylserine (PtdSer) and phosphatidylinositol (PtdIns) as well as the resolution of lysophosphatidylethanolamine (lysoPtdEtn). The polyphosphoinositides were also separated (33). Retention times for each phospholipid class were according to polarity with the elution order being from least polar to m o s t polar. The lower limit of sensitivity was 100 nmol of injected lipid phosphorus. Plasmalogens were separated by H P L C after acidic hydrolysis (34). The ethanolamine glycerophospholipid (EtnGpl) and choline glycerophospholipid (ChoGpl) peaks were collected, dried under N2, and then inverted over 5 drops of concentrated hydrochloric acid in a test tube cap for two minutes. This caused hydrolysis of the alkenyl ether bond of the plasmalogens while the alkylacyl and diacyl bonds remained intact. The sample was reextracted and rechromatographed utilizing the same separation procedure as used in the initial separation. The alkylacyl and diacyl fractions were then collected and subjected to alkaline hydrolysis in 2 m L of 0.1 M N a O H in methanol. After 15 min the reaction was terminated by addition of 1 m L of ethyl formate (35). Cleavage of the alkali-labile ester linkage caused formation of glycerophosphocholine or glycerophosphoethanolamine and of alkylglycerophosphocholine or alkylglycerophosphoethanolamine. The products were separated following extraction of the methanol p h a s e with chloroform/isobutanol/water (4:2:3, v/v/v) followed by centrifugation to facilitate phase separation. The top layer contained glycerophosphocholine or glycerophosphoethanolamine, the b o t t o m layer contained the ether lysophospholipids. Phospholipid classes were q u a n t i t a t e d by m e a s u r i n g lipid phosphorus (36). RESULTS
The compositions of h u m a n saphenous vein (HSVE), h u m a n umbilical vein (HUVE) and h u m a n arterial endothelial cells (HAE) are reported in Table 1. L y s o P t d E t n represented an abnormally high mole percentage of the total lipid phosphorus in the cells from all sources. The polyphosphoinositide values are also reported. No statistically significant differences in the proportions of
TABLE 1 Phospholipid Composition of Cultured Human Endothelial Cellsa
Ptd2Gro PtdEtn+ PtdOH PtdIns lysoPtdEtn PtdSer PtdCho+ CerPCho Ptdlns4P lysoPtdCho Ptdlns4,5P 2
HUVE n = 12 • S.D.
HSVE n = 10 X S.D.
2.60 _ 2.88 17.63 • 1.98 2.37 +- 3.09 1.90 • 3.09 3.12 • 3.36 5.36 • 2.70 42.76 - 6.26 8.91 _ 2.69 1.56 _+ 1.75 7.53 _ 5.04 1.24 • 0.94
2.06 • 1.05 16.86 -+ 3.11 1.81 +.1.64 2.99 • 2.21 5.26 - 2.22 8.86 - 1.76 39.03 • 6.56 10.52 _+ 2 . 3 9 1.19 -- 0.94 8.07 • 4.68 1.99 +- 1.05
HAE n ---- 12 • S.D. 1.08 • 19.06 • 1.56 • 4.26 • 2.54 • 7.15 • 39.12 • 12.96 • 1.90 • 7.02 • 2.09 •
0.85 3.78 0.76 2.31 2.07 2.87 3.92 5.04 1.83 4.84 1.26
aValues are mole % of total phospholipid. Tukey's multiple comparison test was used for statistical evaluation. +PtdEtn and PtdCho represent the sum of alkenylacyl, alkylacyl and diacyl fractions. Abbreviations are: CerPCho, sphingomyelin; HAE, human arterial endothelial; HSVE, human saphenous vein endothelial; HUVE, human umbilical vein endothelial; lysoPtdCho, lysophosphatidylcholine; lysoPtdEtn, lysophosphatidylethanolamine; PlsCho, choline plasmalogen; PtdCho, phosphatidylcholine; PlsEtn, ethanolamine plasmalogen; PtdEtn, phosphatidylethanolamine; Ptd2Gro, cardiolipin; PtdIns, phosphatidylinositol; PtdIns4P, phosphatidylinositol 4-phosphate; PtdIns(4,5)P2, phosphatidylinositol 4,5-bisphosphate; PtdOH phosphatidic acid; PtdSer phosphatidylserine.
TABLE 2 Composition of Ethanolamine and Choline Glycerophospholipid SubclasSes a
Diacyl Alkenylacyl Alkylacyl Diacyl Alkenylacyl Alkylacyl
EtnGpl class (%)
Total Gpl (%)
45.2 _+ 1.4 40.5 +_ 1.9 14.3 _ 0.5
8.0 7.1 2.5
ChoGpl 78.4 11.6 10.0
class (%) +_ 1.1 • 0.9 _ 1.6
Total Gpl (%) 33.5 5.0 4.3
aThe values are means _+ standard deviations. For EtnGpl and ChoGpl, n = 4. Abbreviations are: EtnGpl, ethanolamine glycerophospholipid; ChoGpl, choline glycerophospholipid; and Gpl, glycerophospholipid.
phospholipid classes were found between the various sources of endothelial cells. The m o s t i m p o r t a n t finding is the high l y s o P t d E t n level seen in all three sources. Because no differences were found between the m a j o r phospholipids from various sources, the plasmalogen content of only the H U V E cells was determined. The P l s E t n content was 40.5% of the E t n G p l and the PlsCho content was 11.6% of ChoGpl. These values represent 7.1% and 5.0% respectively of the total lipid phosphorus. The compositions of the choline and ethanolamine glycerophosph~ lipid subclasses are listed in Table 2. The a m o u n t of lipid phosphorus injected onto the colu m n was at or below the desirable lower limit of sensitivity (100 nmol) due to the small sample sizes available from LIPIDS, Vol. 27, no. 2 (1992)
152
COMMUNICATION the cultures. Thus, the variability between samples increases especially for phospholipid classes that represent a small percentage of total lipid phosphorus, DISCUSSION The purpose of this study was to determine the phospholipid composition of human endothelial cells in culture and the dependence of this composition on the vascular source The separation of phospholipids by H P L C allows for good recovery of all phospholipids, improved phosphorus determination and detection of smaller changes in phospholipid composition with statistical significance due to a lower variance than based on thin-layer chromatography (TLC) (37}. The extraction procedure used permits a good recovery of phospholipids while minimizing the amount of protein in the extract. The procedure increases polyphosphoinositide recovery and avoids the need for a twophase system which can cause losses of lysoglycerophospholipids (30}. Freezing the cells prior to extraction decreases the mechanical trauma incurred by the cells during their removal from the plates (31). Hence~ our values should represent more closely the actual composition of the cells in culture Several papers have reported partial phospholipid compositions of endothelial cells from various vascular sources and species (5,19-25). A composition of 25.3% P t d E t n , 36.3% PtdCho and 12.8% CerPCho in H U V E cells has been reported by Rastogi and Nord~y (19), which is similar to our data. A plasmalogen content in H U V E cells of 42.5% P l s E t n in E t n G p l and 6.8% PlsCho in ChoGpl has been reported by Blank e t al. (22). However, their values for the choline glycerophospholipid diacyl (88.6%) and alkylacyl (4.6%} subclasses differ from our values of 78.4% and 10%, respectively. The reported values for the two ethanolamine glycerophospholipid subclasses were 52.6% for the diacyl and 4.9% for the alkylacyl fractions (22). Our values were 45.3% and 14.3% for the diacyl and alkylacyl subclasses, respectively. These differences may reflect the different methodologies used to quantitate the E t n G p l and ChoGpl subclasses. Values for the proportions of choline and ethanolamine glycerophospholipid subclasses reported by Takamura e t al. (25) also differ from our values. The differences could be due to the different methodology used to separate the subclasses. The reported phospholipid compositions for b o t h umbilical vein and arterial endothelial cells (25) are also considerably different from our values. Phospholipid separations by TLC usually have lower recoveries t h a n H P L C separations (our recoveries are 98.3% + 3), and TLC may not completely resolve all classes of phospholipids because of the lower number of theoretical plates. Values for P t d E t n and PtdCho composition from rat and human brain endothelial cells (24) compare closely to our values, although a difference exists in CerPCho contents with 20.4% and 17.0% for rat and human brain, respectively, compared with our value of 10.8%. The difference may be due the fact t h a t the cells were freshly isolated, and not cultured. Isolated rat brain endothelial cells have 46.3% P l s E t n in E t n G p l (18), which is close to our value The polyphosphoinositide content of endothelial cells has not been reported previously for any species. Guinea pig brain contains 0.58% P t d I n s 4 P and 2.58% P t d I n s (4,5)P 2 (38). Our values are within this r a n g e Our LIPIDS, Vol. 27, no. 2 (1992)
data compare favorably to several literature values for the partial phospholipid composition of endothelial cells from various species and vascular sources and are the first to include polyphosphoinositides. The cells exhibited a high l y s o P t d E t n and a relatively high lysoPtdCho content. The position of the f a t t y acid was not determined. Regardless of f a t t y acid position, these lysophospholipids are known to have detergent properties on cellular membranes (39): High concentrations of endogenous lysophospholipids are generally assumed to increase membrane fluidity and to give rise to subtle c h a n g e s in cellular surface. The high levels in human endothelial cells may be related to the apparent ease of thrombosis (40). Increases in membrane fluidity may also cause the cell membrane to be more susceptible to mechanical damage accounting for the observed morphm logical changes such as crater formation in endothelial cells (41}. High levels of lysolipid may modulate the effect of injury and interaction of the endothelium with various hormones (42) because lysophospholipids can alter the functioning of membrane-bound enzymes (39,43). Endothelial cells play a major role in vascular homeostasis (40). In order to understand the lipid metabolism of endothelial cells, the basic phospholipid composition must be known. It has been hypothesized t h a t lipid composition varies with vascular source based on observations of ultrastructural differences depending on vessel size and location (19). In this study no dependence of phospholipid composition on vascular source was found. However, it should be noted t h a t in some phospholipid classes the standard deviations were high. This is especially true for phospholipids which make up a small percentage of the total phospholipids. Differences in phospholipid classes molecular species dependent upon vascular source were not ruled out. ACKNOWLEDGMENTS We thank Diana Carter for her help in the preparation of the manuscript, Patrick Vaccarofor providing the tissues, Mark Armeni for assistance in cell culture and extraction procedures, Lynda Tit~ terington for growing the cells, Robert Hancock for assistance in the statistical calculations, and Laura Dugan for her advice with the HPLC separations. This work was supported in part by National Institutes of Health research grant NS-10165. REFERENCES 1. Derian, C.K., and Moskowitz, M.A. (1986) J. BioL Chem. 261, 3831-3837. 2. Derian, C.K., and Moskowitz, M.A. (1985) Fed. Proa 44, 4862. 3. Moscat, J., Moren~ F., Herrero, C., Iglesias, S., and GorciaBaz:reno, P. (1986} Biochem. Biophys. Res. Commun. 139, 1098-1103. 4. Martin, T.W., and Wysolmerski, R.B. (1987)J. Biol. Chem. 262, 13086-13092. 5. Hong, S.L., and Deykin, D. (1982)J. BioL Chem. 257, 7151-7154. 6. Thomas, J.M.F., Hullin, F., Chap, H., and Douste-Blazy,L. (1984) Thromb. Res. 34, 117-123. 7. Horrocks, L.A., Yec~Y.K., Harder, H.W., Mozzi, R., and Goracci, G. (1986) in Advances in Cyclic Nucleotide and Protein Phosphorylation Research (Greengard, E, and Robinson, G.A., eds.) pp. 263-292, Raven Press, New York. 8. Horrocks, L.A., Harder, H.W., Mozzi, R., Goracci, G., Francescangeli, E., Porcellati, S., and Nenci, G.G. (1986) in Enzymes of Lipid Metabolism (Freysz, L., Dreyfus, H., Massarelli, R., and Gatt, S., eds.) Vol. 2, pp. 707-711, Plenum Press, New York.
153
COMMUNICATION 9. Thomas, J.M.F., Chap, H., and DousteBlazy, L. (1981) Biochem. Biophys. Res. Commun. 103, 819-824. 10. L6ffelholz, K. (1989) Blochem. Pharmacol. 38, 1543-1549. 11. Pelech, S.L., and Vance, D.E. (1989) Trends Biochem. Sci. 14, 28-30. 12. Marcus, A.J., Weksler, B.B., and Jaffe, E.A. (1978)J. BioL Chem. 253, 7138-7141. 13. Nawroth, P.O., Stern, D.M., and Dietrich, M. (1985)Blur50, 7-11. 14. Ingerman-Wojenski, C., Silver, M.J., Smith, J.B., and Macarck, E. (1981)J. Clir~ Invest. 67, 1292-1296. 15. Yamaja Setty, B.N., Stuart, M.J., and Walenga, R.W. (1985) Biochim. Biophys. Acta 833, 484-494. 16. Gormam R.R., Oglesby, T.D., Bundy, G.L., and Hopkins, N.K. (1985) Platelets and Vascular Occlusion 72, 708-712. i7. Buchanan, M.R., Butt, R.W., Magas, Z., Van Ryn, J., Hirsh, J., and Nazir, D.J. (1985) Thrombosis andHaemostasis 53, 306-311. 18. Selivonchick, D.P., and Roots, B.I. (1976) Lipids 12, 165-169. 19. Rastogi, B.K., and Nord4y, A. (1980) Thromb. Res. 18, 629-641. 20. Matheson, D.F., Oei, R., and Roots, B.I. (1980) Neurochem. Res. 5, 43-59. 21. Wey, H.E., Jakubowski, J.A., and Deykin, D. (1986) Biochim. Biophys. Acta 878, 380-386. 22. Blank, M.L., Spector, A.A., Kaduce, T.L., and Snyder, F. (1986) Blochim. Biophys. Acta 877, 211-215. 23. Matheson, D.F., Oei, R., and Roots, B.I. (1980) Neurochem. Res. 5, 683-695. 24. Siakotos, A.N., Rouser, G., and Fleischer, S. (1969) Lipids 4, 234-239. 25. Takamura, H., Kasai, H., Arita, H., and Kito, M. (1990) J. Lipid Res. 31, 709-717. 26. Gimbrone, M.A., Cotran, R.S~, and Folkman, J. (1974)J. CeUBioL 60, 673-684. 27. Maciag, T., Cerundolo, J., Ilsley, S., Kelly, P.R., and Forand, R. (1979) Proc. Natl. Acad. ScL USA 76, 5674-5678.
28. Thornton, S.C., Mueller, S.N., and Levine, E.M. (1983) Science 222, 623-624. 29. Jaffe" E.A., Hoyer, L.W., and Wachmar, R,L. (1973) J. Clin. Invest. 52, 2757-2764. 30. Hara, A., and Radin, N.S. (1978) Anal Biochem. 90, 420-426. 31. Demediuk, P., Anderson, D.K., Horrocks, L.A., and Means, E.D. (1985) In Vitro Cell. Develop. BioL 21, 569-574. 32. Dugan, L.L., Demediuk, P., Pendley, II, C.E., and Horrocks, L.A. (1986) J. Chromatogr. 378, 317-327. 33. Dugan, L.L., Bazan, N.G., and Horrocks, L.A. (1986) Trans. Am. Soc. Neurochem. 17, 138. 34. Murphy, E.J., Stephens, R., Jurkowitz, M.S., and Horrocks, L.A. (1987) J. Neurochem. 48, 560-B. 35. Horrocks, L.A., and Sun, G.Y. (1972) in Research Methods in Neurochemistry (Rodnight, R., and Marks, N., eds.) Vol. I, pp. 223-231, Plenum Press, New York. 36. Rouser, G., Siakotos, A., and Fleischer, S. (1969)Lipids 1, 85-86. 37. Horrocks, L.A., Dugan, L.L., Flynn, C.J., Goracci, G., PorceUati, S., and Yec~Y.K. (1987) in Lecithin." Technologica~ Biological and Therapeutic Aspects, (Ansell, G.B., and Hanin, I., eds.) pp. 3-16, Plenum Press, New York. 38. Hawthorne, J.N., and Kai, M. (1970) in Handbook of Neurochemistry (Lajtha, A., eeL)pl~ 491-508, Plenum Press, New York. 39. Weltzien, H.U., Richter, G., and Ferber, E. (1979) J. BioL Cherr~ 254, 3652-3657. 40. Moncada, S., and Vane, J.R. (1979) N. Engl. J. Med. 300, 1142-1147. 41. Svendson, W., Dalen, H., Moland, J., and Engodal, H. (1985) Arteriosclerosis 58, 27-37. 42. Ryan, U.S., and Ryan, J.W. (1985) Chest 88, 2035-2075. 43. Weltzien" H.U. (1979) Biochim. Biophys. Acta 559, 259-287. [Received July 9, 1990, and in revised form December 27, 1991; Revision accepted December 28, 1991]
LIPIDS, Vol. 27, no. 2 (1992)
