Abstract In endothelial cells (EC), caveolae or plasma- lemmal vesicles (PVs) represent a structurally and bio- chemically specialized membrane microdomain.
Springer-Verlag 1998
Cell Tissue Res (1998) 293:101±110
REGULAR ARTICLE
Anca Gafencu ´ Mihaela Stanescu Aurel Mircia Toderici ´ Constantina Heltianu Maya Simionescu
Protein and fatty acid composition of caveolae from apical plasmalemma of aortic endothelial cells Received: 17 September 1997 / Accepted: 12 February 1998
Abstract In endothelial cells (EC), caveolae or plasmalemmal vesicles (PVs) represent a structurally and biochemically specialized membrane microdomain. Since few data are available on the biochemical composition of PVs of large vessel endothelium, we have designed experiments to isolate this domain and to analyze its chemical components. A highly purified apical membrane fraction was obtained from cultured bovine aortic EC by using cationic colloidal silica (silica-ap), or the EC were surface-radioiodinated and a cell homogenate was prepared. Detergent treatment (Triton X-100; TX) and mechanical disruption of both the silica-ap fraction and cell homogenate followed by ultracentrifugation on a sucrose gradient gave detergent-soluble and detergent-insoluble membranous fractions. The lowest density TX-insoluble fraction appeared morphologically as distinct vesicles (caveolae; 60 nm average diameter; PVs fraction). Biochemical characterization of the PVs fraction (by comparison with the soluble fraction) revealed the presence, at high concentration, of specific caveolar markers, viz., caveolin (both isoforms, the 24-kDa form being conspicuously more abundant) and Ca2+-ATPase. By contrast, angiotensin-converting enzyme and alkaline phosphodiesterase were present almost exclusively in the TX-soluble fraction. The glycoproteins in the PVs fraction were of apparent molecular weights 52, 68, 95, and 114 kDa. Analysis of the fatty acid composition revealed more palmitoleic and stearic acid in the PVs fraction then in the TX-soluble fraction. Thus, in comparison with the plasmalemma proper, the PVs fraction (1) is detergent-insoluble; (2) contains caveolin in two isoforms; (3) contains Ca2+-ATPase at high concentration; (4) contains a set of specific glycoproteins; and (5) is enriched in palmitoleic and stearic acids. This work was supported by a grant from the Romanian Academy. A. Gafencu ´ M. Stanescu ´ A.M. Toderici ´ C. Heltianu M. Simionescu ( ) Institute of Cellular Biology and Pathology, ªNicolae Simionescuº, 8 P.B. Hasdeu Street, RO 79691 Bucharest, Romania Tel.: +40-1-4111144, Fax: +40-1-4111143
)
Key words Aortic endothelium ´ Plasmalemmal vesicles ´ Detergent insoluble domains ´ Fatty acids ´ Cell culture ´ Bovine
Introduction A general characteristic of endothelial cells (EC) of various origins is the presence of an unusually large number of plasmalemmal vesicles (PVs; also named caveolae). Cumulated data have shown that PV have numerous functions, including the transport of soluble and/or membranebound proteins into the cell by endocytosis, or across the cell by transcytosis (for a review, see Simionescu and Simionescu 1991). Previous experiments have demonstrated that PVs represent a structurally and biochemically differentiated microdomain of the EC surface (Simionescu et al. 1981a). Attempts to isolate caveolae from EC have been based on (1) extraction with Triton X-100 (TX) of total lung homogenate (Lisanti et al. 1994), (2) detergent extraction accompanied by mechanical disruption (shearing) of the lung capillary EC surface attached to cationic silica (Schnitzer et al. 1995a, b, c) and (3) non-detergent treatment of a purified aortic EC plasma membrane fraction disrupted by sonication (Shaul et al. 1996). In all these cases, the probes obtained have been further subjected to sucrose density gradients. One reportedly reliable method for isolating the PVs fraction from the membranous (noncaveolar) domain involves the attachment of cationic silica to the apical membrane surface followed by detergent (TX) treatment and mechanical disruption (Schnitzer et al. 1995a). Recently, a method based on specific immuno-isolation on anti-caveolin-coated beads of lung microvascular EC plasmalemma has been described (Stan et al. 1997). The direct effect of TX on aortic EC structures has been demonstrated by electron microscopy; short exposure to the detergent solubilizes the plasmalemma proper almost completely, whereas the PVs membrane remains intact (Moldovan et al. 1995). Little is known about the
102
protein composition of PVs from the macrovascular endothelium, or whether there are similarities to those of microvascular EC. Moreover, no data exist concerning the lipid composition of caveolar domain. We have used the TX-based method to isolate detergent-resistant domains (caveolae) from either apical plasmalemmal attached to cationic silica or from cell-surface radioiodinated aortic EC. As previously reported (Schnitzer et al. 1995c), this fraction contains caveolar material. We report here the protein and fatty acid (FA) composition of this PVs fraction.
Materials and methods Chemicals Chemicals were obtained from the following sources: Dulbecco's modified Eagle's medium (DMEM) and fetal calf serum (FCS) from Gibco BRL (Gaithersburg, Md., USA), Na125-I from Dupont, NEN (Dreieich, Germany), Triton X-100 and Epon 812 from Serva (Heidelberg, Germany), sucrose from Merck (Darmstadt, Germany), sodium polyacrylate (Mw 240 000) from Aldrich Chemical Co. (Steinheim, Germany), anti-caveolin (mouse monoclonal antibody) from Transduction Laboratories (Lexington, N.Y., USA), blotting detection kit for mouse antibodies and enhanced chemiluminescence (ECL) for glycoprotein detection from Amersham Interntional (Buckinghamshire, England), tetramethyl-ammonium-hydroxide (METH-PREP I) from SUPELCO (Bellefonte, USA), trimethyl silyl hydroxide (TMSH) from Macherey Nagel (Düren, Germany), and 4,6 diamidino-2-phenylindole (DAPI) from Molecular Probes (Leiden, The Netherlands). All other reagents were from Sigma (St. Louis, Mo.). Cationic colloidal silica was kindly provided by Dr. Bruce Jacobson (Department of Biochemistry and Molecular Biology, University of Massachusetts). Goat antibody against rabbit pulmonary angiotensin-converting enzyme (ACE) was a gift from Dr. R.L. Soffer (Dept. Molec. Biol., Albert Einstein Coll. Med., New York, USA). Cell culture EC were isolated from bovine aorta by collagenase digestion and cultured on glass Petri dishes in DMEM supplemented with 15% FCS, and 100 U penicillin, 100 mg streptomycin, 50 mg neomycin/ ml (Jinga et al. 1986). The cells were used between the 3rd and 5th passage. Isolation of EC apical plasma membrane Generally, the protocol described for the isolation of the apical plasmalemma from cell monolayers grown in culture was used (Stolz and Jacobson 1992), except that Nycodenz or Metrizamide was replaced by 70% sucrose. Briefly, cultured EC were washed with phosphate-buffered saline (PBS) containing Ca2+ and Mg2+, followed by coating buffer (CB) consisting of 20 mM morpholino-ethane-sulfonic acid (MES), 135 mM NaCl, 0.5 mM CaCl2, 1 mM MgCl2, pH 5.5, for 5 min. Subsequently, the cells were sequentially incubated with 1% colloidal cationic silica and polyacrylic acid (PAA; 10 min each), with CB washes in between. The cells were then exposed to 2.5 mM imidazole, pH 7.0 (lysis buffer), and after 30 min, the lysate (in the dish) was forced through a 23 G needle attached to a syringe, the resulting homogenate containing solely the apical plasmalemma (ap) was collected and centrifuged at 900 g for 10 min. The pellet was resuspended in lysis buffer and, after centrifugation on a 70% sucrose cushion at 28000 g for 60 min, the silicacoated apical plasma membranes (silica-ap) were deposited at the bottom of the tube. Aliquots from the interface layer and pellet were
processed for electron microscopy (see below). Except for the first wash in PBS, the procedure was performed at 4 C. The presence of DNA in all probes were tested with DAPI as described by Brunk et al. (1979). The silica-ap preparation was used to isolate further the detergent-resistant and detergent-soluble domains of the plasma membrane. Preparation of TX-insoluble membrane domains of EC Three preparations were used to obtain detergent-insoluble membrane domains: (1) the silica-ap preparation, obtained as described above, (2) the whole cell homogenate (ch), and (3) the ch obtained from EC whose apical membrane proteins had previously been radioiodinated. To prepare cell homogenates, confluent EC were rinsed with 25 mM MES buffer containing 150 mM NaCl, pH 6.5 (MBS), scraped from the culture dishes, and centrifuged at 1000 g for 10 min. The pellet was resuspended in TX (see below). To obtain ch from EC whose surface proteins were radioiodinated, the confluent cells (on a plastic culture dish) were rinsed with MBS and incubated with a mixture containing 200 mCi Na125I, 60 mg lactoperoxidase (LPO), and 20 ml 0.05% H2O2 in ml MBS, for 15 min, at 20 C, with mild stirring (Hubbard and Cohn 1972). The monolayers were then extensively washed with cold MBS, scraped from the dishes, and processed as above. From probes (1±3), the pellets were used to isolate detergent-insoluble membrane domains according to the method described by Lisanti (1994). Briefly, in each case, the probes were mechanically disrupted in the presence of cold 1% TX (final concentration) in MBS by passing (10±20 times) the mixtures through a 24 G needle, at 4 C. Sucrose in MBS (40% final concentration) was subsequently added. The probes were placed in centrifuge tubes, and a 5%± 30% sucrose gradient (in the same buffer) was layered above. After centrifugation at 270000 g, in an SW 50.1 rotor (Beckman Instruments), for 20 h at 4 C, aliquots (300 ml) were collected from the top to the bottom of each tube and assayed for turbidity (determined by spectrophotometry at 595 nm absorbance), sucrose concentration (by refractometry), and radioactivity (by spectrometry). Characterization of membrane fractions Electron microscopy Based on the turbidity assay, the TX-insoluble fractions were fixed in 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.4, at 20 C. After 10 min, the probes were centrifuged at 12000 g, for 15 min, and the resulting pellets were rinsed, postfixed in 1% OsO4 (in the same buffer), for 1 h, at 4 C, stained en bloc in 1% tannic acid (Simionescu and Simionescu 1976), dehydrated, and embedded in Epon 812. Thin sections were cut on a Reichert ultramicrotome (Reichert Sci. Instruments, Buffalo, N.Y.) and examined with a Philips 400-HM electron microscope. The radii of spherical vesicles were determined or the perimeters of vesicles of various shapes were measured with a podometer (Keufer and Fisser, Basel), and the diameter was calculated. Enzyme assays The starting preparations and the fractions obtained from the sucrose gradient were tested for the following enzyme markers:angiotensinconverting enzyme (ACE; Conroy et al. 1976), 5©-nucleotidase (5©NT), Ca2+-ATPase, alkaline phosphatase (AP; Casale et al. 1984), alkaline phosphodiesterase (APD; Heltianu et al. 1994). Analysis of membrane proteins SDS polyacrylamide gel electrophoresis and autoradiography Aliquots of the initial probes and of the fractions obtained from sucrose gradients were exposed to 15% trichloroacetic acid (final con-
103 centration) and centrifuged at 12000 g, for 5 min. The sediment of proteins was solubilized in Laemmli sample buffer and processed for SDS polyacrylamide gel electrophoresis (SDS-PAGE; Laemmli 1970). The gels were stained with either silver nitrate or Coomassie brilliant blue. In the experiments in which the EC surface was radiolabeled before preparation of ch (probe 3), dried gels were exposed to X-ray film (Kodak X-Omat), at 70 C, for 3±30 days. Immunoprecipitation From surface radioiodinated EC (probe 3), the membrane fractions obtained on the sucrose gradient were solubilized in 50 mM TRISHCl buffer, pH 7.5, containing 1% Nonidet P40, 2 mM EDTA, 150 mM NaCl, 0.2 mM phenylmethane sulfonyl fluoride, 1 mM pepstatin, and 5 mM benzamidine, and then incubated with antiACE or anti-caveolin antibodies for 5 h, at 4 C, on a shaking platform (Antohe et al. 1991). The immune complexes formed were exposed to protein G ± Streptococcus cells (protein G ± group C Streptococcus sp., Sigma), for 18 h at 4 C, with mild stirring. The immunoprecipitates were centrifuged at 12000 g, for 5 min, extensively Fig. 1 Electron micrographs showing isolated endothelial cell apical plasmalemma adsorbed to cationic silica (a, b) and the vesiculated membrane resistant to extraction within Triton X-100 (c). In a and b, the various size particles of cationic silica decorate evently the isolated plasmalemma proper and the diaphragms of plasmalemmal vesicles (v). The vesicle (caveolae) membranes are not decorated by the silica beads. On TX treatment and ultracentrifugation (to remove the silica beads and soluble proteins), only highly purified insoluble membranous vesicles of different dimensions and sizes are present (c). Bars 100 nm
washed with the same buffer as above, and then processed for SDS-PAGE. Coomassie-stained gels were dried and subjected to autoradiography. Immunoblotting For all probes, proteins separated by SDS-PAGE were electrotransferred to nitrocellulose membranes (NC) by using a semi-dry system with a graphite electrode apparatus (Andersen 1984). The NC blots were stained with Ponceau S, and the position of tansferred proteins was marked. The discolored blots were exposed to 1% bovine serum albumin (BSA) in TRIS buffer saline containing 0.05% Tween 40 (24 h at 4 C), then incubated with anti-caveolin antibody for 1 h at 22 C, and visualized with a blotting detection kit for mouse antibodies, as described by the manufacturer. Protein concentration This was determined by using a bicinchoninic acid assay kit and BSA as the standard.
104
Lipids were isolated from membrane fractions by methanol/chloroform extraction as previously reported (Bligh and Dyer 1959). Briefly, 0.8 ml of each fraction obtained from the sucrose gradient (in MBS buffer) was mixed with 2 ml methanol for 24 h, and then 2 ml chloroform was added over 2.5 h (in two steps). The probes were mixed with 1 ml tridistilled H2O and incubated overnight. To facilitate extraction, the mixture was acidified with 0.15 ml HCl. To separate the organic and aqueous phase, the mixture was centrifuged at 1000 g for 10 min. The aqueous (upper) layer and the protein band were removed, and the remaining organic phase was evaporated to dryness under nitrogen. The lipid samples were then prepared for gas chromatographic analysis (GC) by transmethylation (Fourie and Basson 1990). The dried probes were solubilized in 150 ml transmethylation reagent (chloroform:METH-PREP I ± 2:1, v/v), sonicated for 5 min at 80 C, and then aliquots of 1 ml were injected into a gas chromatographic system (Carlo Erba Instruments, Type HRGC 5300, Milan, Italy), equipped with a SP 2340 column (30 m length, 0.32 mm film thickness; Supelco, Switzerland). The starting temperature of the column was 70 C for 2 min; this was then increased at a rate of 5 C/min up to 180 C, at which time the chromatographic run was performed. FAs were detected by a flame ionization detector (FID; Carlo Erba Instruments, Type EL 580, Milan, Italy) and quantitated by using a CromJet Integrator (Spectra Physics, San Jose, Calif., USA). The following FA ratios were measured: 12:0, 14:0, 16:0, 16:1, 18:0, 18:1, and 18:2. The unsaturated index was calculated as in Ricci et al. (1987).
Results Purification of apical plasmalemma of EC The apical plasma membrane was obtained from cultured EC sequentially exposed to colloidal cationic silica and PAA (for fixation of the former to the cell surface), followed by cell lysis. On centrifugation of the ensuing homogenate, most internal membranes and soluble proteins were removed with the supernatant. The pellet was resuspended in lysis buffer and, after centrifugation on a sucrose cushion, a band at the interface and a pellet were obtained. The DNA assay revealed that, following centrifugation of the crude material, most nuclei (92%) remained within the sucrose and interface band, whereas a small percentage (about 8%) were spun down with the pellet. These results were confirmed by morphological studies. Examination by electron microscopy of the pellet and the interface band showed that most of the silica-ap was found in the pellet where it appeared as long membrane sheets with various size colloidal silica particles attached. Cationic silica was bound preferentially to the exoplasmic side of the plasmalemma proper (Fig. 1a), whereas the
0.12
30 II
0.10
I
20
0.08 IV
10
0.06 0.04
Sucrose gradient (%)
Determination of membrane FA composition
×103 1500
40 III
Absorbance (OD595)
NC blots containing proteins of TX-insoluble and TX-soluble membrane fractions obtained from ch and silica-ap were used. The protocol described by the manufacturer for total carbohydrate labeling was followed. The method involves oxidation of carbohydrate moieties that react with biotin hydrazide, with subsequent binding of streptavidin-horseradish-peroxidase conjugate to biotinylated glycoproteins. The complexes were detected by enzymatic reaction with ECL detection reagents.
0.14
0 0
5
10
500 375
cpm
Glycoprotein composition of membrane fractions
250 125 0
15
Fraction number
Fig. 2 Turbidity and radioactivity peaks of the fractions isolated from TX-exposed endothelial cell (EC) homogenates. The cultured EC were surface-radioiodinated, homogenized in TX, and then ultracentrifuged on a continuous sucrose gradient (filled circles). Absorbance measurements at 595 nm (open squares) indicate three lightscattering bands (I±III), at 7%±11%, 11%±17% and 17%±24% sucrose density (TX-resistant fractions). Fraction IV collected at 30%±40% sucrose density has the lowest turbidity and represents the detergent-soluble fraction. Radioactivity (filled squares) is found in fractions I, II, and especially in IV
PVs were not decorated (Fig. 1a, b). The PVs varied in size and shape compared with those observed in EC in culture. Similar results were recently reported for lung microvascular EC (Schnitzer et al. 1995a, c; Stan et al. 1997). Some silica-ap was also present at the interface band together with a large number of cell nuclei. The results obtained with our experimental procedure (in which the apical membranes were separated on a sucrose cushion) were similar to those of the original method in which Nycodenz or Metrizamide was used (Chaney and Jacobson 1983; Jacobson et al. 1992; Stolz et al. 1997). Isolation of TX-insoluble membrane domains The detergent-insoluble membrane domains (PV-rich fraction) were obtained from the isolated EC apical plasmalemma and from the ch treated with TX and further centrifuged on a sucrose density gradient. The results obtained when ch (300 ml aliquots collected after centrifugation) was analyzed for turbidity, sucrose concentration, and radioactivity (probe 3) are shown in Fig. 2. The turbidity assay (absorbance at 595 nm) indicated that, starting from the top of the tube, there were three light-scattering bands (ch1±ch3); the fourth fraction (ch4) showed no absorbance at 595 nm. The sucrose density of these fractions corresponded to 7%±11%, 11%±17% and 17%±24% for ch1±ch3, respectively, and 30%±40% for ch4 (Fig. 2). Examination by electron microscopy of the TX-insoluble fractions (ch1±ch3) revealed the presence of a heterogeneous collection of membrane-bound vesicles of various diameters (40±400 nm) and shapes (data not shown). In the case of ch obtained from surface radioiodinated EC, a small amount of radioactivity was found in
105
the ch1 and ch2 fractions, most of the radioactivity being found in ch4 fraction (Fig. 2). These data indicated the presence of TX-insoluble apical membrane proteins in the ch1 and ch2 fractions and of detergent-soluble surface membrane proteins in the ch4 fraction. In the case of silica-ap treated under the same conditions as above, only one light-scattering band (ap1), at 7%±11% sucrose and corresponding to ch1, was obtained. By electron microscopy, it was found that ap1 was composed of uniform vesiculated membranes devoid of other contaminants (Fig. 1c). From two separate experiments, calculated vesicle diameters in the ap1 (and ch1) fraction indicated that, out of 100 profiles ranging in size from 40 and 150 nm, approximately 78% presented an average outer diameter of about 60 nm, being comparable in size and shape to isolated PVs reported by others (Schnitzer et al. 1995a, c; Stan et al. 1997). The TX-soluble fraction (ap4) was found at the same sucrose concentration as ch4. The pellet obtained after ultracentrifugation contained cationic silica free of plasma membranes (data not shown). Of the total cellular protein, the three fractions of TXinsoluble membrane domains contained approximately 1%±2% protein. Similar results were obtained for caveolin-rich membrane domains isolated from lung (Lisanti et al. 1994). Clear probes obtained at 30%±40% sucrose density, representing TX-soluble proteins (the ch4 and ap4 fractions), contained approximately 70% protein. The pellet consisted of about 20% of cellular proteins; the 24%±30% sucrose fraction contained approximately 8% protein. Biochemical analysis of membrane fractions Protein composition The proteins from all fractions and from the corresponding starting materials were separated by SDS-PAGE (Fig. 3). From all proteins of the silica-ap preparation (Fig. 3A, lane 1), the 55-kDa and 68-kDa polypeptides were also found in the TX-insoluble (ap1) fraction (Fig. 3A, lane 2); numerous polypeptide bands (12± 95 kDa) were present in the ap4 fraction (Fig. 3A, lane 3). We assumed that the small number of bands revealed on SDS-PAGE of ap1 was attributable to low protein concentration. To circumvent this problem, apical membrane proteins of EC whose surface had previously been radioiodinated were subjected, under the same conditions, to SDS-PAGE followed by autoradiography. In these experiments from all radiolabeled membrane proteins (Fig. 3B, lane 1), polypeptide bands of approximately 63, 68, 95, 114, 131, and 166 kDa were detected in the detergent-insoluble fraction (Fig. 3B lane 2); both TX-resistant fractions (ch1 and ch2) had identical autoradiographic protein patterns. The main bands found in the soluble fraction (ch4) were at 16, 18, 45, 80, and 116 kDa (Fig. 3B, lane 3). When the dried gels were exposed to the X-ray film for a long time, some of the faint bands found in ch also
kDa
kDa 205
kDa
116 116
116
66
66
66
45
45 45
24 24 24
18
18
a
1
2
3
b
18
1
2
3
c
1
2
3
4
5
Fig. 3a±c Electrophoretic pattern of endothelial cell (EC) apical membrane, cell homogenate, and the derived membrane fractions. a SDS-PAGE of silica-isolated EC apical plasmalemma (lane 1) and of the TX-insoluble and TX-soluble fractions (lanes 2, 3). b Autoradiography of the EC homogenate obtained after cell-surface radioiodination (lane 1) and of the TX-insoluble and TX-soluble fractions (lanes 2, 3). c Protein electrophoretic pattern of the cell homogenate (lane 1), the TX-insoluble fractions (lanes 2±4), and the soluble fraction (lane 5). Molecular weight markers were: myosin (205 kDa), galactosidase (116 kDa), albumin (66 kDa), ovalbumin (45 kDa), trypsinogen (24 kDa), and beta-lactoglobulin (18.4 kDa)
became evident on the autoradiograms of the TX-insoluble domains. The major 66-kDa polypeptide found in ch was particularly abundant in ch1 and was not observed in the soluble ch4 fraction (Fig. 3B, lanes 2, 3, respectively); this radioiodinated polypeptide, probably representing surface-adsorbed albumin, was also previously reported to be present in the vesicle fraction (Lisanti et al. 1994). SDS-PAGE of the total cell homogenate and the derived fractions (ch1±ch4) showed that, of the total ch (Fig. 3C, lane 1), the major proteins of apparent Mr 12, 18, 27, 44, and 68 kDa were highly enriched in the TX-insoluble fractions (Fig. 3C, lanes 2±4); in the detergentsoluble fraction, the main polypeptide bands were found between 40±60 kDa (Fig. 3C, lane 5). Unstained gels similar to those presented in Fig. 3C were transferred to NC membranes and tested for the presence of caveolin. The immunoblots revealed that all TX-insoluble fractions (ch1, ch2, and ch3) were highly enriched in caveolin that appeared as two polypeptides at the expected positions of 21 and 24 kDa; interestingly, the 24-kDa band appeared significantly more abundantly than the 21-kDa band (Fig. 4, lanes a±c). These peptides correspond to the two isoforms of caveolin (alpha-caveolin, 24 kDa, and beta-caveolin, 21 kDa) recently reported (Scherer et al. 1995). Similar results have also been obtained for 3T3-L1 fibroblasts stably expressing myc-
106 kDa
Ca²+ -ATPase
100
ACE
80
Total activity (%)
29
18
a
a
b
c
d
60
40
20
0
b
1
2
ap1
3
Fig. 4a, b Immunodetection of caveolin in the endothelial plasmamembrane-derived fractions. a The proteins of the ch1±ch4 fractions were separated by SDS-PAGE (gels were loaded with 40 mg protein/ lane), transferred to nitrocellulose membrane, and further incubated with anti-caveolin antibody followed by blot-developing kit. Caveolin is present in all three TX-resistant fractions (lanes a±c) and absent in the TX-soluble fraction (lane d). Note the two isoforms of caveolin (arrows) of which the higher Mr polypeptide is more abundant (lanes a±c). b Caveolin was determined in apical plasma membrane attached to silica (1), TX-insoluble PVs (2), and the detergentsoluble membrane fraction (3) by immunodot analysis. Note the enhancement of caveolin in ap1 (2) and its absence in ap4 (3)
tagged caveolin-2 immunoreacted with antibody against the C-terminal caveolin-1, which recognizes both the alpha and beta isoforms (Scherer et al. 1995, 1996). The presence of caveolin in ch3 (as shown in Fig. 2, this does not contain a peak of radioactivity after cell surface iodination) suggests that this fraction may represent either internalized plasmalemmal vesicles and/or caveolin-rich membrane domains present in the trans Golgi-network (Kurschalia et al. 1992). The TX-soluble fraction (ch4) did not contain the 21-kDa or 24-kDa polypeptide in detectable amounts (Fig. 4, lane d). The pellet obtained after sucrose gradient centrifugation of the cell homogenate showed a slight reaction with caveolin antibody (data not shown). This suggests that some of the TX-insoluble membrane domains are associated with other cell components that are pelleted down (Lisanti et al. 1994). The presence of caveolin in ap1 fraction was also revealed by dot blot experiments that documented the existence of the protein in silica-ap, its enrichment in the ap1 fraction, and its absence in the ap4 (TXsoluble) fraction (Fig. 4b). From the experiments in which EC surface proteins were radioiodinated, aliquots of TX-resistant complexes (see Fig. 3B, lane 2, for the protein pattern) were used for the immunoprecipitation assay with the anti-caveolin antibody. Unexpectedly, the autoradiograms of gels containing the immune complexes did not reveal any protein bands (data not shown). This result can be explained by the known hydrophobic hairpin structure of caveolin; this is inserted into the cytoplasmic aspect of the plasmalemma, with little or no exposure to the exoplasmic side of the membrane, a location that presumably obstructs the
ap4
Fig. 5 The distribution of the caveolar marker, Ca2+-ATPase, and of the membrane marker, angiotensin-converting enzyme (ACE) on endothelial apical membrane-derived fractions obtained after Triton X100 (TX) treatment. ap1 TX-insoluble fraction, ap4 TX-soluble fraction. Note the high level of Ca2+-ATPase in ap1 and of ACE in ap4 Table 1 The specific activity of enzyme markers in endothelial cell homogenate (ch) and in the TX-insoluble (ch1±ch3) and TX-soluble (ch4) fractions. All enzyme activities are expressed in nmol/min per mg, except AP, which is expressed in mmol/min per mg. The percentage of each enzyme present in the various fractions is given in parentheses. Note the low amount of 5©-NT and AP in the caveolar (ch1) fraction (5©-NT 5©-nucleotidase, AP alkaline phosphatase, ACE angiotensin-converting enzyme, APD alkaline phosphodiesterase) Enyzmes
ch
Ca2+-ATPase
2.22
5©NT
0.05
AP
26.00
ACE
1.46
APD
5.80
ch1
ch2
ch3
ch4
10.75 (32.8) 25.50 (15.4) 4.42 (13.8) 0.11 (6.5) 3.30 (9.6)
13.75 (42.0) 76.30 (46.2) 27.30 (85.5) 0.08 (4.7) 5.60 (16.3)
7.32 (22.4) 62.99 (37.5) 0.10 (0.3) 0.05 (2.9) 7.30 (21.2)
0.91 (2.8) 0.60 (0.4) 0.10 (0.3) 1.46 (85.9) 18.20 (53.2)
access of radioiodination reagents to this protein (Dupree et al. 1993; Parton and Simons 1995). Under the experimental conditions used, apical membranes from 1108 cells (31 mg protein) represented 3.3% (1 mg) of the total amount of protein. The yield of the PV-rich fraction derived from silica-bound apical membranes was approximately 10% (100 mg protein). The distribution of caveolar and noncaveolar membrane markers in the starting material and in the fractions obtained from the equilibrium sucrose density from gradient centrifugation are presented in Fig. 5 and Table 1. Upon TX exposure of silica-ap, the ap1 fraction (detergentinsoluble) was particulary enriched in Ca2+-ATPase compared with the ap4 fraction (TX-soluble); conversely, the latter was especially rich in ACE whose activity was almost lacking in ap1 fraction (Fig. 5). The results obtained are in good agreement with immunodetection studies on microvascular EC (Schintzer et al. 1995a; Stan et al. 1997).
107 Fig. 6a±c Identification of membrane glycoproteins from TX-insoluble and TX-soluble membrane fractions obtained from apical plasmalemma (a, b) or homogenates (c, d) of endothelial cells. The detergent insoluble ap1 and ch1 (lanes a, c, respectively) and soluble ap4 and ch4 (lanes b, d, respectively) fractions were subjected to SDS-PAGE and transferred to nitrocellulose membranes; the blots were incubated with the ECL glycoprotein detection kit
ap1
kDa
12 : 0
ap4
14 : 0 16 : 0 116
16 : 1
97
18 : 0 18 : 0
66
18: 2
45
0
10
a
20
30
40
% of total FA ch1
24
12 : 0
18
14 : 0
ch2 ch3 ch4
16 : 0
a
b
c
d
Because of the low quantity of proteins that could be obtained from isolated silica-apical membrane, other membrane markers were assessed in the TX-insoluble and TX-soluble fractions obtained from the cell homogenate (Table 1). In all fractions (ch1±ch4), the percentage of Ca2+-ATPase, 5©NT, and AP was much higher in the detergent-insoluble (ch1±ch3) than in TX-soluble membanes (ch4), a result similar to that reported in other studies (Lisanti et al. 1994; Schnitzer et al. 1995c). However, in the ch1 fraction, which corresponds to the bouyant density of ap1 and represents the pure caveolar fraction, 5©NT and AP (the GPI-linked proteins) were in much smaller amounts (three and six time less, respectively) than in the ch2 fraction (Table 1). The noncaveolar markers ACE and APD were abundant (86% and 53%, respectively) in ch4. A small amount of APD was noted in ch2 and ch3 (16% and 21%, respectively). The data obtained for the distribution of ACE were confirmed by immunoprecipitation experiments in which the ch1, ch2, and ch4 fractions isolated from surface radioiodinated EC homogenate treated with TX (see Fig. 2B for the protein profile) were incubated with anti-ACE antibody. The results showed that, out of all the fractions tested, the antibody immunoprecipitated a polypeptide of approximately 220±240 kDa (ACE) only from the ch4 fraction (TX-soluble; not shown). Analysis of the glycoprotein composition of all the fractions obtained from ap and ch was performed by using an ECL detection system. As shown in Fig. 6, the ap1 fraction exhibited three major glycoproteins of 52, 95, and 114 kDa (lane a), and the ch1 fraction displayed two glycoproteins of 52 and 68 kDa (lane c). Except for
16 : 1 18 : 0 18 : 0 18: 2 0
10
b
20
30
40
% of total FA
Fig. 7a, b Fatty acid (FA) composition of apical endothelial cell (EC) membrane fractions (a) and of fractions isolated from EC homogenate (b). Lipids extracted from TX-insoluble (ap1) and soluble (ap4) fractions obtained from apical plasmalemma, and from detergent-insoluble (ch1±ch3) and detergent-soluble (ch4) fractions were analyzed for FA content by gas chromatography. FA were detected by a flame ionization detector. The FA content was calculated by integration of the chromatographic peak areas and expressed as a percent of the total. Left Carbon chain length and number of double bonds
the glycoprotein of 52 kDa, the other glycoproteins were also found in the microvascular EC (Ghitescu et al. 1997; Stan et al. 1997). More glycoproteins were detected in the TX-soluble fractions: ap4 and ch4 (Fig. 6, lanes b and d; respectively). Lipid composition The FA content of all fractions was determined by GCFID (Fig. 7). The percent composition of each FA was calculated by integration of the chromatographic peak areas. A comparison of the TX-soluble and TX-insoluble fractions revealed that palmitoleic (16:1) and stearic
108
(18:0) acids were significantly more prevalent in ap1 than in ap4 (4.2 and 1.5 times, respectively), whereas palmitic (16:0) and oleic (18:1) acids were more prevalent in ap4 than in ap1 (1.2 and 1.5 times, respectively; Fig. 7a). Other FA (12:0, 14:0, 18:2) were similarly distributed in both fractions. The FA composition in all detergent-insoluble fractions (ch1±ch3) obtained from ch was almost identical (Fig. 7b). Comparing the TX-insoluble and TX-soluble fractions, the stearic acid content in the ch1±ch3 fractions was higher than in ch4, and the oleic acid content in the ch1±ch3 fractions was lower than in ch4, a result similar to that obtained in ap-derived fractions. By contrast, palmitoleic acid was detected in minor amounts in all ch1±ch4 fractions, and the palmitic acid content in ch1± ch3 was higher than that in ch4 (Fig. 7b). The unsaturated index (UI) for the ap fractions was almost similar in detergent-insoluble and detergent-soluble membrane domains. The UI for the ch fractions was significantly lower for the insoluble ch1±ch3 fractions (0.63, 0.58, and 0.78, respectively) than for the soluble ch4 fraction (1.2). Similar results were obtained when the lipids extracted were prepared for GC by transmethylation with TMSH, and the FA methyl esters were detected by mass spectrometry.
Discussion Previous reports have demonstrated the existence of structurally and biochemically differentiated microdomains on the luminal plasmalemma of microvascular EC; these microdomains differ in their glycoprotein composition and anionic site distribution (Simionescu et al. 1981b). Moreover, recently accumulated data show variability in the resistance of the EC plasma membrane to short-term exposure to TX. Thus, after exposure of total lung homogenate to TX, detergent-insoluble and TX-soluble membrane fractions have been isolated and partially characterized (Lisanti et al. 1994). TX-soluble and TX-resistant microdomains have also been isolated from silica-coated luminal plasmalemma of microvessels. The latter also contain a low density caveolar fraction rich in caveolin and Ca2+-ATPase (Schnitzer et al. 1995a, b, c). In cultured aortic endothelial cells, the effect of TX on the cell membrane has been demonstrated by electron microscopy, showing that, after a short detergent exposure of 10 min at 4 C, the plasmalemma proper is almost completely solubilized but that intact bona fide vesicles and uncoated pits are also present (Moldovan et al. 1995). Despite the structural evidence for the existence of TX-soluble and TX-insoluble microdomains of large vessel EC, the biochemical characterization of these domains is scanty; the only data reported indicate that the PVs fraction contains caveolin, endothelial nitric oxide synthase, and calmodulin (Garcia-Cardena et al. 1996; Shaul et al. 1996). By applying the same technique as for microvessel PVs (Schnitzer et al. 1995a) to cultured aortic EC, we
have characterized the protein, glycoprotein, and FA composition of detergent-soluble and detergent-insoluble membrane domains. Based on previous reports (Schnitzer et al. 1995a, b) and on the data obtained in this study, the TX-insoluble (ap1 and ch1) fractions can be safely considered as PVs (or caveolae) fractions. The procedure used consists of (1) the isolation of ap by cationic silica, which has the advantage of avoiding potential contamination with cytoplasmic vesicles or with Golgi- or trans-Golgi-derived vesicles, which also contain caveolin (Dupree et al. 1993; Kurzchalia et al. 1992); (2) the purification of caveolin-rich domains based on their resistance to short-term exposure to TX and the low buoyant density of caveolae because of their high content of glycosphingolipids (Dupree et al. 1993). Morphological analysis of the PVs fraction shows good preservation of the lipid bilayer and of the overall membrane structure. Because of the small quantity of silica-isolated ap, as an alternative method the apical membrane proteins have been radioiodinated and the cell homogenate preparation subjected to a similar TX treatment as that above. In this case, two radioactive TX-insoluble fractions (ch1 and ch2) are obtained, whereas only one (ap1) detergent-insoluble fraction is found in the case of silica-ap. We assume that ch2 represents internalized vesicles based on recent data showing that, when aortic EC in culture are incubated with LPO (and processed by diaminobenzidine cytochemistry), some cytoplasmic vesicles in close vicinity to the plasmalemma are labeled in addition to the apical plasmalemmal vesicles and plasmalemma proper (Stanescu et al. 1996). The third fraction obtained (ch3) is not radiolabeled but has the same structural characteristics as ch1 and ch2, being rich in caveolin; it can be assumed that the ch3 fraction contains cytoplasmic vesicles and caveolin-rich vesicles derived from the trans-Golgi network. The vesicular structures obtained by the above methods vary in diameter. We suppose that those vesicles in the range of 60±90 nm represent single vesicles, whereas those with a larger diameter are derived from defolded clustered vesicles that are open to the apical front. Caveolin, the main marker for caveolae, is present in significant amounts in both the ap1 and ch1 fractions. Electrophoretic analysis of ch1 shows that the band corresponded to alpha caveolin is much more abundant than the beta isotype. Similar results, i.e., differences between the ratio of the two caveolin isoforms, have been found in fat cells and their precursors (Scherer et al. 1994, 1996). The caveolin distribution in macrovascular EC (this study) is different from that reported for microvascular EC, where the proportion between the high and the low Mr caveolin forms is similar in isolated TX-resistant vesicles (Lisanti et al. 1994; Schnitzer et al. 1995c). One can speculate that the different expression of the two caveolin isoforms in EC of various vascular beds is an indication of the existence of fine biochemical differences between these two vesicle populations and possibly between PVs within the same EC, as previously suggested (Scherer et al. 1996). The specific PVs chemical composition
109
may be related to their various functions in EC (Simionescu and Simionescu 1991; Heltianu et al. 1997). In addition to caveolin, the PVs fraction contains a high concentration of Ca2+-ATPase, whereas GPI-linked proteins (5©NT, AP) are found in low quantities. These data are similar to those reported for microvessel EC-derived caveolae (Schnitzer et al. 1995c). The glycoprotein pattern of detergent-insoluble fractions shows the presence of glycoproteins of 52, 68, 95, and 114 kDa. With the exception of the 52-kDa glycoproteins, the others were also found in PVs isolated from microvascular EC (Stan et al. 1997; Ghitescu et al. 1997). To our knowledge, data on the FA composition of the plasma membrane and PVs fractions in EC are limited. Some results obtained from epithelial cells demonstrate that the apical plasma membrane contains a high quantity of glycolipids of which 21% are concentrated in surface caveolae. For EC, the only data reported refer to the FA composition of glycosphospholipids (Pacifici et al. 1994). In our experiments, we have found that palmitoleic (16:1) and stearic (18:0) acids are particularly concentrated in caveolae (ap1 fraction). In the TX-soluble membrane domain (ap4 fraction), the main FAs are palmitic (16:0) and oleic (18:1) acids. This is an additional indication that PVs have a distinct biochemical composition from the rest of the EC apical plasmalemma. Taken together, these results show that the PVs fraction (1) is insoluble in TX; (2) expresses caveolin isoforms that appear in a different protein ration; (3) is enriched in Ca2+-ATPase; (4) contains glycoproteins of an apparent Mr of 52, 68, 95, and 114 kDa, which differ from the plasmalemma proper; and (5) is enriched in palmitoleic and stearic acids in comparison with the plasma membrane where palmitic and oleic acids are more abundant. These data confirm and extend the concept that the plasmalemmal vesicles (caveolae) represent a distinct membrane microdomain of endothelial cells, and that fine biochemical differences may exist between vesicles in the micro- and macro-vasculature. Acknowledgements We are grateful to Dr. Bruce Jacobson (Department of Biochemistry and Molecular Biology, University of Massachusetts) for kindly supplying us with the cationic colloidal silica. The excellent technical assistance of F. Chiritescu, F. Georgescu (radiobiochemistry) I. Manolescu (cell culture), N. Dobre (photography), M. Scheian (graphics), and M. Daju (word processing) is gratefully acknowledged.
References Andersen JK (1984) Electroblotting of multiple gels: a simple apparatus without buffer tank for rapid transfer of proteins from polyacrylamide gels to nitrocellulose. J Biochem Biophys Methods 10:203±209 Antohe F, Heltianu C, Simionescu M (1991) Albumin-binding proteins of endothelial cells: immunocytochemistry detection of the 18 kDa peptide. Eur J Cell Biol 56:34±42 Bligh EG, Dyer WJ (1959) A rapid method of total lipid extraction and purification. Can J Biochem Physiol 37:915±919 Brunk C, Jones K, James TW (1979) Assay for nanogram quantities of DNA in cellular homogenates. Anal Biochem 92:497±500
Casale TB, Friedman M, Parada N, Plekes J, Kalliner M(1984) Preparation of human lung purified plasma membrane fraction. Confirmation of enzyme markers, electron microscopy and histamine H1 receptor binding. J Membr Biol 79:33±39 Chaney LK, Jacobson BS (1983) Coating cells with colloidal silica for high yield isolation of plasma membrane sheets and identification of transmembrane proteins. J Biol Chem 258:10062± 10072 Conroy JM, Hoffman H, Kirk ES, Hirzel HO, Sonnenblick EH, Soffer RL (1976) Pulmonary angiotensin-converting enzyme. J Biol Chem 251:4828±4832 Dupree P, Parton RG, Raposo G, Kurzchalia TV, Simons K (1993) Caveolae and sorting in the trans-Golgi network of epithelial cells. EMBO J 12:1597±1605 Fourie PC, Basson DS (1990) Application of a rapid transesterification method for identification of individual fatty acids by gas chromatography on three different nut oils. J Am Oil Chem Soc 67:18±20 Garcia-Cardena G, Oh P, Liu J, Schnitzer JE, Sessa WC (1996) Targeting of nitric oxide synthase to endothelial cell caveolae via palmitoylation: implications for nitric oxide signaling. Proc Natl Acad Sci USA 93:6448±6453 Ghitescu L, Crine P, Jacobson BS (1997) Antibodies specific to plasma membrane of rat lung microvascular endothelium. Exp Cell Res 232:47±55 Heltianu C, Dobrila L, Stanescu M, Antohe F, Simionescu N, Silverstein SC, Simionescu M (1994) A method for selective radiolabeling of lung endothelium plasmalemmal vesicles, in situ. Eur J Cell Biol 4:61±70 Heltianu C, Serban G, Alexandru V, Gafencu A, Simionescu N, Simionescu M (1997) Expression of transferrin receptors in endothelial cells transfected by electroporation. Eur J Cell Biol 72:79±89 Hubbard AL, Cohn ZA (1972) The enzymatic iodination of the red cell membrane. J Cell Bio 55:390±405 Jacobson BS, Schnitzer JE, McCaffery M, Palade GE (1992) Isolation and partial characterization of the luminal plasmalemma of microvascular endothelium from rat lungs. Eur J Cell Biol 58:296±306 Jinga VV, Bogdan I, Fruchter J (1986) Experimental model for the quantitative estimation of transendothelial transport in vitro; a two-compartment system. Rev Roum Morphol Physiol 23: 227±230 Kurzchalia TV, Dupree P, Parton RG, Kellner R, Virta H, Lehnert M, Simons K (1992) VIP 21, a 21-kD membrane protein is an integral component of trans-Golgi-network-derived transport vesicles. J Cell Biol 118:1003±1014 Laemmli UK (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227:680±685 Lisanti MP, Scherer PE, Vidugiriene J, Tang ZL, Hermanowski-Vosatka A, Tu Y-H, Cook RF, Sargiacomo M (1994) Characterization of caveolin-rich membrane domains isolated from an endothelial-rich source: implications for human disease. J Cell Biol 126:111±126 Moldovan N, Heltianu C, Simionescu N, Simionescu M (1995) Ultrastructural evidence of differential solubility in Triton X-100 of endothelial vesicles and plasma membrane. Exp Cell Res 219:309±313 Pacifici EHK, McLeod L, Peterson H, Sevanian A (1994) Linoleic acid hydroperoxide-induced peroxidation of endothelial cell phospholipids and cytotoxicity. Free Radic Biol Med 17:285± 295 Parton RG, Simons K (1995) Digging into caveolae. Science 269: 1398±1399 Ricci S, Patoia L, Berrettini M, Binaglia L, Scarcella MG, Bucaneve G, Vecchini A, Carloni I, Agostini L, Parise P, Favero AD (1987) Fatty acid pattern of red blood cell membranes and risk of ischemic brain infarction: a case-control study. Stroke 18:575±578 Scherer PE, Lisanti MP, Baldini G, Sargiacomo M, Corley-Mastick C, Lodish HF (1994) Introduction of caveolin-rich vesicles. J Cell Biol 127:1233±1243
110 Scherer PE, Tang Z-L, Chun MC, Sargiacomo M, Lodish HF, Lisanti MP (1995) Caveolin isoforms differ in their N-terminal protein sequence and subcellular distribution. J Biol Chem 270:16395± 16401 Scherer PE, Okamoto T, Chun MC, Nishimoto I, Lodish HF, Lisanti MP (1996) Identification sequence, and expression of caveolin-2 defines a caveolin gene family. Proc Natl Acad Sci USA 93:131±135 Schnitzer JE, Oh P, Jacobson BS, Dvorak AM (1995a) Caveolae from luminal plasmalemma of rat lung endothelium: microdomains enriched in caveolin, Ca2+-ATPase, and inositol triphosphate receptor. Proc Natl Acad Sci USA 92:1759±1763 Schnitzer JE, McIntosh DP, Dvorak AN, Liu J, Oh P (1995b) Separation of caveolae from associated microdomains of GPI ± anchored proteins. Science 269:1435±1439 Schnitzer JE, Liu J, Oh P (1995c) Endothelial caveolae have the molecular transport machinery for vesicle budding, docking, and fusion including VAMP, NSF, SNAP, annexins and GTPases. J Biol Chem 270:14399±14404 Shaul PW, Smart EJ, Robinson LJ, German Z, Yuhanna IS, Ying Y, Anderson RG, Michel T (1996) Acylation targets endothelial nitric-oxide synthase to plasmalemmal caveolae. J Biol Chem 271:6518±6522 Simionescu M, Simionescu N, Palade GE (1981b) Differentiated microdomains on the luminal surface of the capillary endothelium. II. Partial characterization of their anionic sites. J Cell Biol 90:614±621
Simionescu M, Simionescu N (1991) Endothelial transport of macromolecules: transcytosis and endocytosis. A look from cell biology. Cell Biol 25:1±79 Simionescu N, Simionescu M (1976) Galloylglucoses of low molecular weight as mordant in electron microscopy. I. Procedure and evidence for mordanting effect. J Cell Biol 70:608±621 Simionescu N, Simionescu M, Palade GE (1981a) Differentiated microdomains on the luminal surface of the capillary endothelium. I. Preferential distribution of anionic sites. J Cell Biol 90:605± 613 Stan RV, Roberts WG, Predescu D, Ihida K, Saucan L, Ghitescu L, Palade GE (1997) Immunoisolation and partial characterization of endothelial plasmalemmal vesicles (caveolae). Mol Biol Cell 8:595±605 Stanescu M, Heltianu C, Simionescu M (1996) Different protein composition of vesicle membrane and plasmalemma proper of lung capillary endothelia. Rev Roum Biochim 33:125±134 Stolz DB, Jacobson BS (1992) Examination of transcellular membrane protein polarity of bovine aortic endothelial cells in vitro using the cationic colloidal silica microbeads membrane isolation procedure. J Cell Sci 103:39±51 Stolz DB, Branton D, Jacobson BS (1997) Plasma membrane isolation using the cationic colloidal silica isolation technique. Cell Biol 72:79±89