Endothelial Albumin Binding Proteins Are Membrane-associated Components Exposed on the Cell Surface*

The heterobifunctional, photoactivatable, thiol- cleavable cross-linker sulfosuccinimidyl 2-(p-azido-salicy1amido)ethyl- 1,3’-dithiopropionate (SASD) was radioiodinated and used to determine whether endothelial albumin binding proteins (ABP) recently iden-tified (Ghinea, N., Fixman, A., Alexandru, D., Popov, D., Hasu, M., Ghitescu, L., Eskenasy, M., Simionescu, M., and Simionescu, N. (1988) J. Cell Biol. 107, 231- 239) are plasma membrane-associated components exposed on the cell surface. Microvascular endothelial cells (MEC) freshly isolated from rat epididymal fat were incubated with ‘261-2-~-azidosali~y1amido)ethyl- 1,3‘-dithiopropionate (ASD)- albumin conjugate which upon photolysis by U V light was cross-linked to the receptor proteins. By cleaving the disulfide linkages of the cross-linker with 5% B-mercaptoethanol and the ligand-receptor interactions with 0.1% sodium dodecyl sulfate, the radioiodinated ASD moiety remained attached to the receptor peptides which were further detected by 5% sodium dodecyl sulfate-polyacrylamide gel electrophoresis followed by autoradiography. In parallel, samples were examined by

The heterobifunctional, photoactivatable, thiolcleavable cross-linker sulfosuccinimidyl 2-(p-azido-salicy1amido)ethyl-1,3'-dithiopropionate (SASD) was radioiodinated and used to determine whether endothelial albumin binding proteins (ABP) recently identified ( upon photolysis by U V light was cross-linked to the receptor proteins. By cleaving the disulfide linkages of the cross-linker with 5% B-mercaptoethanol and the ligand-receptor interactions with 0.1% sodium dodecyl sulfate, the radioiodinated ASD moiety remained attached to the receptor peptides which were further detected by 5% sodium dodecyl sulfate-polyacrylamide gel electrophoresis followed by autoradiography. In parallel, samples were examined by ligand blotting with albumin-gold complex. The results showed that in these experimental conditions ABP are represented by two major peptides of 31 and 18 kDa and two minor bands of 73 and 56 kDa. Densitometric scanning showed that the two major bands constitute more than 70% of the total ABP. The four peptides were not apparent if the samples were not UV-irradiated. The binding of the radioiodinated ligand to ABPs was reduced by -82% in the presence of excess competitive unlabeled albumin. When MEC were incubated with unlabeled SASD and exposed to UV light, the autoradiographic banding pattern obtained was similar to that of either radioiodinated receptor proteins or MEC not treated with SASD. This indicated that the four albumin binding peptides are distinct proteins of the endothelial cell plasmalemma. revealed that the capillary endothelium of murine lung, heart, skeletal muscle, and adipose tissue contains specific albumin binding sites restricted mainly to plasmalemmal vesicles (1-4). Monomeric albumin detected by the immunogold technique was also localized to plasmalemmal vesicles as well as to the plasmalemma proper (5). Radioiodinated albumin was found to bind to cultured microvascular endothelial cells in a specific, saturable, and reversible manner (6).
Using the ligand blotting technique, we have recently reported the identification of albumin binding proteins (ABP)' in microvascular endothelial cells (MEC) freshly isolated from rat epididymal fat (7,8). Under the conditions used for cell extraction and SDS gel electrophoresis, the ABP were represented by two pairs of polypeptides with major components of 31 and 18 kDa. ABP electrophoretically transferred to nitrocellulose membranes bound specifically and with high affinity (& = 60 X lo-' M) to monomeric and polymeric albumin. Cell surface iodination of MEC followed by electroblotting and autoradiography showed that bands with the same molecular weight as those stained by albumin-gold were also labeled with radioiodine. Since theoretically it was possible that the same band may contain two different polypeptides, one reactive with albumin-gold and the other radioiodinatable, these results were not taken as definitive evidence that the detected ABP were membrane-associated. Therefore, we conducted new experiments in which albumin was crosslinked to its plasmalemma1 receptor proteins by an iodinatable, photoactivatable, and thiol-cleavable reagent, sulfosuccinimidyl 2-(p-azidosalicylamido)ethyl-1,3'-dithiopropionate (SASD). This heterobifunctional cross-linker in organic solvents was used to characterize receptor-ligand interactions and derivatize noniodinatable ligands. Iodination of SASD in aqueous solution with preservation of its reactive moieties makes this reagent a valuable tool for the identification of cell surface receptors, such as the receptor for phytohemagglutinin (9). In the present study we have used this crosslinker to determine whether endothelial ABP (7,8) are components of the plasma membrane.

Endothelial Albumin Binding Proteins
R-2 x-ray films used for gel autoradiography from Azomures, Tg. Mures, Romania. Monomeric albumin was prepared from commerical BSA by gel filtration on Sephadex G-200.
Bovine immunoglobulin G, albumin-free, was kindly prepared by Dr. Szegli (Institute Dr. I. Cantacuzino, Bucharest, Romania). All other chemicals employed were of analytical grade.
Albumin-Gold Complex-A. Au was prepared as described previously (2).
Microvascular Endothelial Cells-MEC were freshly isolated from rat epididymal fat pads according to Refs. 11 and 12; these cells express albumin binding sites (8). The cell suspension obtained was devoid of adipocytes and may contain up to a maximum of 15% pericytes as contaminant; in most preparations, the pericytes represented less than 10% of the cell population. We assumed that even if pericytes contain ABP, their potential contribution to the detected ABP should be minor. The endothelial origin of ABP is supported by the previous evidence of albumin-specific binding to MEC both in situ (1)(2)(3)(4)7) and in culture (6). Cells were maintained for 10-12 h in culture medium "199, supplemented with 15% fetal calf serum, 2 mg/ml histamine, 100 pg/ml heparin, 20 pg/ml endothelial cell growth supplement, 100 units/ml penicillin, and 100 pg/ml streptomycin. At the start of the experiments, the culture medium was discarded and the cells extensively washed in PBS.

Methods
Since SASD contains a photolabile azido group, before photolysis all experiments were carried out in a dark room under red light using vessels covered with aluminum foil.
Preparations of Radiolabeled Putative ABP Radioiodination of SASD-the protocol followed was that described in Ref. 9 with slight modifications. Borosilicate glass tubes were first coated with 400 pg of Iodogen (for each tube). SASD was dissolved in dimethyl sulfoxide (1 mg/50 pl) to which 950 pl of 0.2 M borate buffer, pH 8.0, was added to reach a solvent to buffer ratio of 1:20 (v/ v). One hundred microliters (185 nmol) of dissolved SASD was diluted to 1 ml with 200 mM borate buffer, pH 8.0, followed by addition of 2 mCi of NaIZ5I (carrier-free). The reaction was terminated after 1 min by emptying the reaction tube. The solution was used directly for coupling to albumin.
Binding and Cross-linking of lZ5I-ASD-A to Endothelial ABP-To prevent nonspecific binding, MEC were preincubated with 2% polyvinylpyrrolidone in PBS for 1  Cleavage of '251-ASD-A .ABP Complex-Cell-containing samples were washed with PBS (3 min X 2). To cleave both the disulfide linkages of the cross-linker and ligand-receptor binding, the samples were treated for 1 h at 37 "C with 5% p-mercaptoethanol and 0.1% SDS in PBS. (In separate experiments we found out that, at a concentration of 0.176, SDS dissociates the albumin-receptor complex but fails to extract ABP from the membrane (data not shown).) As reported in Ref. 13 on exposure of MEC to UV light the labeled moiety of ASD is expected to be intra-or intermolecularly linked with albumin. To eliminate these radiolabeled secondary products, the cells were washed with PBS (3 min X 2) followed by centrifugation at 13,500 x g.
After cleavage of the cross-linked, radiolabeled ligand-receptor complex, the sample and the control cells (see below) were collected in PBS containing 5% SDS, 5% 0-mercaptoethanol, 5 mM benzamidine hydrochloride, and 0.25 mM phenylmethylsulfonyl fluoride. To check the effect of detergent concentration on ABP extraction, we also conducted experiments using 10% SDS. The cells were homogenized for 1 min with an ultrasonic cell disruptor (Heat Systems Ultrasonics Inc., Plainview, NY), employing the microtip with the amplitude at setting 6. The homogenate was centrifuged at 13,500 x g for 5 min in a Beckman Microfuge B and the supernatant used for protein determination (13) and electrophoresis.
Controk-To check for albumin cross-linking to its receptor proteins, two sets of experiments were carried o u t (i) incubation of MEC with "'1-ASD-A in the presence of a 300-fold excess of unlabeled albumin as competitor; and (ii) incubation of MEC with L251-ASD-A without photoactivation (the ligand-receptor complex was not crosslinked).
Cross-linking of MEC Surface Components-To check whether within the endothelial cell membrane the ABP peptides detected are or are not associated with one another, we incubated MEC with unlabeled SASD. Based on published reports indicating that SASD can introduce covalent links between two proteins (e.g. ligand/receptor) (9, 13), we assumed that in the absence of a specific ligand, the photoactivated SASD can covalently link juxtaposed membrane proteins including ABP. If these peptides are parts of a multisubunit receptor they should be cross-linked and appear in subsequent SDS-PAGE as high molecular weight complexes. If the ABP peptides are separate entities within the plasmalemma, their electrophoretic pattern should be similar to that obtained with the radiolabeled SASDalbumin conjugate (previous experiment in this paper).
Briefly, after washing with PBS (3 min X 3), MEC (3 mg of protein) were suspended in 1 ml of 20 mM borate buffer, pH 8.0, and incubated with 2 mM SASD in the same buffer as above for 30 min at 22 "C. The unbound SASD was removed by washing cells in PBS (3 min X 2). The cells were then irradiated with UV light from a 100-W xenon lamp (as above) for 70 s on ice, washed twice with PBS, and homogenized in PBS containing 5% SDS, 5 mM benzamidine, and 0.25 mM phenylmethylsulfonyl fluoride. After centrifugation at 13,500 X g for 5 min, the supernatant was used for protein assay and SDS-PAGE (without 0-mercaptoethanol). The resolvedproteins were then electrotransferred to nitrocellulose membranes and incubated with albumin-gold complex. Controls consisted in using protein extracts from freshly isolated MEC not treated with SASD.
Identification of ABP SDS-PAGE and Autoradiography-Homogenate samples were concentrated by precipitation, on ice, with trichloracetic acid at 10% final concentration, in the presence of 0.015% sodium deoxycholate. After 30 min, samples were centrifuged at 13,500 X g for 5 min, the supernatant decanted and the pellet dissolved in solubilization buffer (198 mM Tris, 107 mM H3P01, pH 6.8, containing 5% SDS, 5% .j 3mercaptoethanol and 420 mM sucrose. 8-Mercaptoethanol was omitted in experiments requiring nonreducing conditions. Aliquots of 5 pl were counted in a Beckman 4000 y counter. Electrophoresis was conducted according to Ref. 15, the acrylamide concentration being 5-15% in the resolving gel and 3.75% in the stacking gel. Gels were calibrated using a Sigma high molecular weight kit supplemented with myoglobin (horse heart), and the molecular weights were calculated by the calibration curve method (16). The gels (13 X 17 cm and 1.4 mm thick) were electrophoresed overnight at 5 mA and then stained with 0.2% Coomassie Brilliant Blue R-250 in 45% methanol, 10% acetic acid. Autoradiograms of the dried gels were made on Azoix R-2 x-ray film.
Electroblotting-Incubation with Albumin-Gold-The transfer of electrophoretically separated proteins to nitrocellulose membrane was performed in 41 mM Tris, 40 mM borate buffer, pH 8.3, for 3 h at 200 mA in an electroblotting chamber (17). The blots were washed with PBS (10 min x 3) containing 1 mM CaC12 and, to prevent nonspecific binding, were preincubated with 2 mg/ml bovine IgG in PBS for 12 h at 22 "C. The strips were then incubated with A.Au in PBS (concentration corresponding to Aki6% = 1.0). Visualization Of A. Au complex binding to the ABP bands was enhanced by the silver method recommended by Janssen Life Science Products (Beerse, Belgium).

RESULTS AND DISCUSSION
Cross-linking of monomeric albumin to its putative endothelial receptor proteins (ABP) was achieved by using SASD, a heterobifunctional, iodinatable, photoactivatable, and cleavable reagent. SASD has one chemically and one photochemically reactive group connected by a thiol-cleavable linkage. Fig. 1 illustrates the sequence of steps involved in the present study. SASD was first radioiodinated in the 2-(pazidosalicylamido) moiety using Na1251 and Iodogen (Step I ) .
The radioiodinated SASD was attached to monomeric albumin (BSA) by acylation for 60 min at room temperature, which resulted in the loss of the sulfosuccinimidyl moiety. The radioiodinated 2-(p-azidosalicylamido)-1,3'- After extensive washing with PBS, the '9-ASD-A was allowed to bind to the albumin receptor proteins (ABP) to which it was cross-linked by photolysis with UV light for 70 s at 0 "C (Step 4). After washing with PBS, the disulfide linkages of the cross-linker were cleaved with 5% 8-mercaptoethanol, and albumin-receptor interactions were dissociated with 0.1% SDS. As a result, the radioiodinated ASD moiety remained attached to the receptor peptides (ABP) (Steps 5 and 6). After additional washings with PBS, the labeled ABP were extracted, subjected to electrophoresis and visualized by autoradiography. Fig. 2A shows the banding pattern obtained with MEC subjected to the protocol depicted in Fig. 1. Four bands appeared (lam b): two major bands at 31 and 18 kDa, and two minor bands at 73 and 56 kDa. Densitometry indicated the following approximate relative contributions of each peptide to the total radioactivity associated with the ABP: 7, 21, 39, and 33% for the 73,56,31, and 18-kDa peptides, respectively. The two major bands (31 and 18 kDa) therefore constitute more than 70% of the total ABP. The four bands did not appear if MEC were not exposed to UV light ( l a n e c). As already mentioned, the binding of the radiolabeled ligand to ABP was reduced in the presence of 300-fold excess of unlabeled BSA (by -82%), suggesting that the binding is susceptible to competitive inhibition. with the A. Au complex followed by silver enhancement. When MEC were incubated with unlabeled SASD and exposed to UV light ( l u n e d), the banding pattern obtained was the same as that observed with untreated MEC ( l a n e e ) . The polypeptides revealed were the same as in the experiments with radioiodinated receptor (lane b). Assuming that SASD can covalently link adjacent membrane proteins, these findings suggest that the four albumin binding peptides may be distinct entities rather than parts of an oligomer. However, stronger evidence is needed to elucidate convincingly the in vivo interaction between the ABP peptides.
By using a detergent concentration higher (5%) than in our previously reported experiments (1% SDS) (a), in addition to the two major peptides (31 and 18 kDa) revealed in those conditions, two other minor peptides (73 and 56 kDa) were also extracted from MEC plasma membranes. However, when the SDS concentration was raised to IO%, no additional bands appeared. Densitometry of lanes d and e in Fig. 2B showed that the two minor bands extracted at high detergent concentration contribute less than 30% of the total mass of detected ABP. Cultured MEC are reported to express a single 60-kDa glycoprotein that interacts weakly but specifically with albumin (18). These observations suggest that under the experimental conditions used in this study, ABP (represented by two major and two minor polypeptides) appear to be distinct membrane components exposed on the endothelial cell surface. Their exact interactions remain to be elucidated. The findings also suggest that two minor ABP may occur either in small concentration or have a lower affinity for albumin than the 31and 18-kDa peptides, or both. Kinetic studies to define the binding properties of each ABP and the detection of their location to different microdomains of the endothelial plasma membrane are currently being conducted.