The voltage-dependent anion channel is a receptor for plasminogen kringle 5 on human endothelial cells.

Human plasminogen contains structural domains that are termed kringles. Proteolytic cleavage of plasminogen yields kringles 1-3 or 4 and kringle 5 (K5), which regulate endothelial cell proliferation. The receptor for kringles 1-3 or 4 has been identified as cell surface-associated ATP synthase; however, the receptor for K5 is not known. Sequence homology exists between the plasminogen activator streptokinase and the human voltage-dependent anion channel (VDAC); however, a functional relationship between these proteins has not been reported. A streptokinase binding site for K5 is located between residues Tyr252-Lys283, which is homologous to the primary sequence of VDAC residues Tyr224-Lys255. Antibodies against these sequences react with VDAC and detect this protein on the plasma membrane of human endothelial cells. K5 binds with high affinity (Kd of 28 nm) to endothelial cells, and binding is inhibited by these antibodies. Purified VDAC binds to K5 but only when reconstituted into liposomes. K5 also interferes with mechanisms controlling the regulation of intracellular Ca2+ via its interaction with VDAC. K5 binding to endothelial cells also induces a decrease in intracellular pH and hyperpolarization of the mitochondrial membrane. These studies suggest that VDAC is a receptor for K5.

Angiogenesis is essential for tumor growth (1)(2)(3)(4). Vascular endothelial growth factor (VEGF) 1 is a potent mitogen promoting endothelial cell proliferation (5,6), whereas angiostatin inhibits this process in vitro and suppresses tumor growth in vivo (7,8). Angiostatin is a fragment of plasminogen (Pg) consisting of either the first three or four of its kringles (7). Pg kringle 5 (K5) also suppresses growth factor-stimulated angiogenesis via cell cycle G 1 arrest and induction of apoptosis (9 -12); however, the cellular receptor(s) mediating these effects are unknown. K5 confers on Pg the capacity to bind to human umbilical vein endothelial cells (HUVEC) with high affinity (13). K5 also mediates binding of Pg to the Pg activator streptokinase (SK) (14,15). Sequence similarities between SK and the mitochondrial human voltage-dependent anion channel (VDAC1) exist; specifically, the region comprising SK residues Tyr 252 -Lys 283 are homologous to VDAC1 residues Tyr 224 -Lys 255 (16). We raised antibodies against peptides contained within these regions and used them to identify VDAC1 on the HUVEC surface by flow cytometry (FACS). Receptor binding assays demonstrated that K5 binds with high affinity to sites on these cells. K5 inhibits VEGF-stimulated HUVEC proliferation and induces a decrease in cytosolic pH and an increase in the potential of isolated mitochondria. Highly purified VDAC1 binds to K5 after reconstitution of the receptor into liposomes. Our data suggest that VDAC1 is a receptor for K5 on the cell surface.
Proteins-Human Pg was resolved into its isoforms, Pg 1 and 2 (17,18). Pg 2 was digested with elastase and fractionated by gel and affinity chromatography to obtain mini-Pg, followed by digestion of mini-Pg with pepsin to obtain K5 (10,19,20). Gel electrophoresis (10 -20% gradient gel, nonreducing conditions) identified a doublet of ϳ12 kDa, which was identified by mass spectrometry as K5. Amino-terminal sequence analysis yielded the sequence LPTVETPSEE, corresponding to Pg residues 450 -459, confirming the identification of K5 (21). Reduction/alkylation of K5 was performed by incubating 20 g of K5 with 1 mM dithiothreitol for 30 min followed by incubation with 5 mM iodoacetamide for 30 min, both at room temperature, and removal of these reagents by dialysis versus 10 mM Hepes, pH 7.5. Iodination of K5 was performed with 125 I-labeled Bolton-Hunter reagent (specific activity, 500 -700 cpm/ng).
Antibodies-Antibodies to SK were raised in rabbits, and the IgG fraction specific against the SK sequence Glu 263 -Lys 283 was purified by immunoaffinity on a resin containing this peptide conjugated to activated carboxyhexyl-Sepharose (Amersham Biosciences). The antibodies against the 21-amino acid sequence Lys 235 -Lys 255 of VDAC1 conjugated to keyhole limpet hemocyanin (22) were prepared in rabbits by CO-VANCE (Denver, PA). The IgG fraction specific to VDAC1 was purified by immunoaffinity on a resin containing the VDAC1 peptide conjugated to carboxyhexyl-Sepharose. The monoclonal antibody 20B12 against human mitochondrial VDAC1 was from Molecular Probes, Inc.
Endothelial Cell Proliferation Assay-HUVEC from Clonetics (San Diego, CA) were grown in Dulbecco's modified Eagle's medium containing 20% bovine serum, 100 units/ml penicillin/streptomycin, 2.5 g/ml amphotericin B, 2 mM glutamine, 5 units/ml sodium heparin, and 200 g/ml endothelial cell growth supplement (23). The cells were washed with phosphate-buffered saline and dispersed in a 0.05% trypsin solu-tion. The cells were resuspended in medium (25 ϫ 10 3 cells/ml) and plated in 96-well culture plates (0.2 ml/well). After 24 h at 37°C, the medium was replaced with 0.2 ml of Dulbecco's modified Eagle's medium, 5% bovine serum, 1% antibiotics, and the test samples were applied. Cell proliferation was determined at 24 h using bromodeoxyuridine labeling and a colorimetric immunoassay (Roche Applied Science). The results were expressed as percentages of control proliferation determined in the presence of VEGF (10 ng/ml) and the absence of K5.
Flow Cytometry-HUVEC were detached from the culture flasks (75 cm 2 ) by incubation for 5 min at 37°C with Ca 2ϩ and Mg 2ϩ -free phosphate-buffered saline containing 4 mM EDTA and pelleted. The cells (1 ϫ 10 7 /ml) were washed with phosphate-buffered saline before resuspension in ice-cold Phenol Red-free Hanks' balanced salt solution (HBSS), 1% BSA, 0.3 mg/ml goat IgG, and 0.01% NaN 3 (staining buffer). The cell suspensions (100 l) were incubated 30 min with dilutions of rabbit polyclonal anti-human SK peptide IgG, anti-human VDAC1 peptide IgG, or the murine anti-human mitochondrial VDAC1 monoclonal antibody. The cells were washed with ice-cold staining buffer, pelleted, and resuspended in 100 l of ice-cold staining buffer. The cell suspensions were incubated in the dark with an AF488-conjugated for 30 min to goat anti-rabbit or mouse IgG from Molecular Probes, Inc. The cells were washed twice with ice-cold staining buffer, resuspended in ice-cold 1% paraformaldehyde, and stored in the dark at 4°C until analysis by FACS. The mean relative fluorescence after excitation at ϭ 495 nm was determined for each sample on a FACSVantage SE flow cytometer (BD Biosciences) and analyzed with CELLQUEST® software (BD Biosciences).
Ligand Binding Analysis-The cells were grown in tissue culture plates until the monolayers were confluent. The cells were washed in HBSS. The binding assays were performed at 4°C in RPMI 1640 containing 2% BSA. Increasing concentrations of 125 I-K5 were incubated with cells for 60 min in 96-well strip plates. Free and bound ligand were separated by aspirating the incubation mixture and washing the cell monolayers rapidly thrice with RPMI 1640 containing 2% BSA. The wells were stripped from the plates and radioactivity determined. The bound ligand was calculated after subtraction of nonspecific binding measured in the presence of 50 mM p-aminobenzamidine. The K d and B max of K5 were determined by fitting the data directly to the Langmuir isotherm using the statistical program SYStat® for Windows.
Antibody Binding Studies-The binding assays were performed in HUVEC grown in 96-well strip plates. The cells were washed in HBSS and incubated with increasing concentrations of 125 I-labeled anti-human VDAC1 peptide IgG for 90 min at 25°C in RPMI 1640 containing 2% BSA. The cells were rinsed with RPMI 1640, and the wells stripped from the plates were inserted in plastic tubes to determine radioactivity. IgG bound was calculated after subtraction of nonspecific binding measured in the presence of 50 M nonlabeled IgG. The B max of the anti-VDAC1 IgG was then calculated.
Measurements of Intracellular Free Ca 2ϩ Concentration and Cytosolic pH i -HUVEC [Ca 2ϩ ] i was measured by digital imaging microscopy using the fluorescent indicator Fura-2/AM (24). For measurements of pH i , HUVEC were incubated overnight in Dulbecco's modified Eagle's medium on glass coverslips and then washed with HBSS with 0.1 M sodium bicarbonate, pH 7.1. The cells were incubated for 20 min with 2 M 2Ј,7Ј-bis-(2-carboxyethyl)-5-(and -6)-carboxyfluorescein (BCECF) in HBSS, rinsed with buffer thrice, and placed on the fluorescent microscope stage. Intracellular pH (pH i ) was measured by a digital video imaging technique in cells stimulated by the ligands, which were added after obtaining a stable base line (25).
Gel Electrophoresis-Electrophoresis was performed in 0.1% SDS employing a discontinuous Laemmli buffer system (26). The gels were stained with 0.25% Coomassie Brilliant Blue R-250. Transfer to nitrocellulose membranes was carried out by the Western blot method (27). The dye-conjugated M r markers (Bio-Rad) used were of M r 38,100, 28,400,18,200,9,200, and 4,300.
Purification of VDAC1 From 1-LN Cells-It is difficult to obtain large numbers of cultured HUVEC; however, we found that 1-LN cells are a good source of this protein. 1-LN cells were grown in RPMI 1640 supplemented with 10% (v/v) fetal bovine serum, 100 units/ml penicillin G, and 100 ng/ml streptomycin in 20 culture flasks (150 cm 2 ). After detaching with 10 mM EDTA in HBSS and pelleting, the cells were suspended in 10 ml of 20 mM Hepes, pH 7.2, 0.25 M sucrose containing the proteinase inhibitors (each at 0.5 mg/ml) antipapain, bestatin, chymostatin, trans-epoxysuccinyl-L-leucylamido-(4-guanidino)butane (E-64), leupeptin, pepstatin, o-phenanthroline, and aprotinin. The cells were lysed by sonication on ice (five 10-s bursts with 30-s intervals). The homogenate was centrifuged at 800 ϫ g for 15 min, followed by centrifugation at 50,000 ϫ g for 1 h. The pellet containing cell membranes was resuspended in 20 mM Tris-HCl, pH 8.0, containing 1% (v/v) Triton X-100 to solubilize membranes and centrifuged at 50,000 ϫ g for 30 min to remove insoluble materials. VDAC was sequentially purified to homogeneity using gel filtration on Sephadex G-150 and immunoaffinity chromatography with an anti-VDAC peptide IgG conjugated to Sepharose 4-B (see Fig. 1C).
Incorporation of VDAC1 into Liposomes and Binding of K5 to the Reconstituted Receptor-Purified VDAC1 was reconstituted into liposomes (28,29) as follows. 50 l of a suspension of liposomes (8 M L-␣-phosphatidylcholine, 8 M phosphatidylethanolamine, 8 M ␤-oleoyl-␥-palmitate and 6.9 M cholesterol) in 5% Me 2 SO were mixed with VDAC1 (5 g) and incubated with agitation for 30 min at room temperature. The concentration of Me 2 SO was reduced to 0.5% with 50 mM Tris-HCl, pH 7.4. After the addition of 125 I-K5 (10 nM) and incubation for another 30 min at room temperature, the mixture was filtered through a Sephadex G-75 column (55 ϫ 2 cm). To study inhibition of K5 binding to VDAC1 reconstituted into liposomes, the mixture was incubated with the specific anti-VDAC1 IgG for 30 min at room temperature before the addition of 125 I-K5.
The kinetic parameters of K5 binding to VDAC1 on reconstituted liposomes were performed on large unilamellar liposomes (0.4 m in diameter) prepared by extrusion of multilamellar vesicles through 0.4-m defined polycarbonate filters (Nucleopore, Pleasanton, CA) (30). For these experiments, proteoliposomes containing VDAC1 or BSA were prepared by mixing the proteins (50 g) in 2.5 mM Hepes, pH 7.4, 145 mM NaCl, and 0.3 mM N-dodecyl-␤-D-maltopyranoside with N-dodecyl-␤-D-maltopyranoside saturated (0.6 mM) liposomes at a 1:3 volume ratio of protein preparations to liposomes (31). The detergent was removed after three 2-h incubations at 4°C of the proteoliposomes with 10 mg of Biobeads SM2 (Bio-Rad) followed by three 30-min centrifugations at 100,0000 ϫ g. Phospholipid phosphate was then determined phospholipid) in 2.5 mM Hepes, pH 7.4, containing 145 mM NaCl. Filtration and determination of kinetic parameters were carried out as described above.
Preparation of Mitochondria-Mitochondria from 1-LN cells were isolated (33), and the protein levels were estimated using the bicinchonic acid method (34).
Mitochondrial Membrane Potential-Membrane potential (⌬) was determined at room temperature using DSMPϩ, a fluorescent indicator of membrane potential (35). The assay consisted of a final volume of 2 ml containing sucrose (250 mM), Hepes (10 mM), EGTA (2.5 mM) pH 7.4, mitochondria (0.2 mg) cellular protein, DSPM ϩ 2 (nmol), rotenone (1 g), sodium succinate (10 M), pH 7.4, and increasing concentrations of K5. The mixture was incubated for 20 min with K5 prior to addition of DSPMϩ. The fluorescence intensity was measured (excitation ϭ 489 nm, emission ϭ 566 nm) in a Shimadzu RF-5301PC spectrofluorometer (Shimadzu Corporation, Kyoto, Japan). Maximal fluorescence in the absence of K5 was obtained with mitochondria incubated with DSPMϩ alone. The results are the means of fluorescence determinations from three experiments.

Analyses of Sequence Similarities between SK and Human
VDAC1-The regions of sequence similarity between streptokinase and human VDAC1 were identified by the BLAST program provided by the Swiss Institute of Bioinformatics (Fig.  1A). The topology prediction for helical transmembrane proteins was solved with use of the hidden Markov model, also provided by the Swiss Institute of Bioinformatics, and shows a loop through the outer mitochondrial membrane spanning VDAC1 residues 247 QTLKPGIKL 255 (Fig. 1B). We raised rabbit antibodies to the SK peptide 263 EINNTDLISLEYKYV-LKKGEK 283 and the VDAC1 peptide 236 KVNNSSLIGLGY-TQTLKPGIK 255 . A Coomassie Brilliant Blue stain of the purified VDAC (Fig. 1C, lane 1) shows a major band of M r ϳ32,000. A blot binding assay with a rabbit anti-VDAC1 (peptide Lys 235 -Lys 255 ) IgG shows reactivity with this protein (Fig. 1C,  lane 2). Similarly, the purified VDAC1 showed reactivity with the anti-SK (peptide Glu 263 -Lys 283 ) IgG (Fig. 1C, lane 3), confirming the structural relatedness between VDAC1 and SK. However, the purified VDAC1 did not show any reactivity with 125 I-K5 when electroblotted to a nitrocellulose membrane (Fig.  1C, lane 4).
Binding of K5 to VDAC1 Incorporated into Liposomes-The experiments described above demonstrate a significant impact on the ability of purified receptor to bind to K5; therefore, the purified VDAC1 was incorporated into liposomes and gel filtration on Sephadex G-75 employed to identify and separate the reactants (Fig. 2). 125 I-K5 eluted at a column volume of 100 -120 ml (Fig. 2A). When 125 I-K5 was incubated with solubilized VDAC1 (2 g), the radiolabeled material eluted in the same fractions as above, suggesting no reactivity between K5 and solubilized receptor (Fig. 2B). When VDAC1 was incorporated into liposomes and then reacted with K5, the radiolabeled material eluted from the column as two peaks, one of them corresponding to the void volume where VDAC1 elutes and the other corresponding to the elution volume of unreacted K5 (Fig.  2C). These data indicate that K5 binds to VDAC1 when this receptor is incorporated into a lipid membrane. Binding of K5 to membrane-incorporated VDAC1 was inhibited by anti-VDAC1 (peptide Lys 235 -Lys 255 ) IgG, demonstrating that this is the region responsible for binding to K5 (Fig. 2D).
K5 binds to VDAC1 proteoliposomes (Fig. 3A) in a dose-dependent manner with high affinity (K d of 22 Ϯ 3.1 nM). The binding is specific for VDAC1 because control proteoliposomes prepared with BSA or empty liposomes show little specific binding (Fig. 3A). Binding of 125 I-K5 to VDAC1 proteoliposomes is inhibited by unlabeled K5 or anti-VDAC1 IgG (Fig.  3B), suggesting that VDAC1 is a receptor for K5.
Analyses of VDAC1 on the Cell Surface of HUVEC by Flow Cytometry-As determined by FACS, HUVEC reacted with an antibody against the SK peptide (Fig. 4A) as well as an antibody against the VDAC1 peptide (Fig. 4B) or a murine antibody against human mitochondrial VDAC1 (Fig. 4C), as expected because mitochondrial and plasma membrane VDAC1 share the same primary structure (35). FACS analysis of HUVEC reacted with K5 (0.1 M) prior to reaction with antibody against the VDAC1 peptide (Fig. 4D) shows inhibition of binding of this antibody, suggesting that both K5 or the IgG compete for the same binding site. Taken together, these experiments show that VDAC1 is not only expressed on the surface of HUVEC but also establishes the structural relationship between SK and VDAC1 hypothesized by McCabe et al. (16).
Inhibition of Endothelial Cell Proliferation by K5-K5 inhibited VEGF-dependent HUVEC proliferation in a dose-depend-ent manner (Fig. 5A). As observed previously (10), the antiendothelial cell proliferation of K5 was abolished after reduction/alkylation of the protein, suggesting that the formation of appropriate disulfide bridges is essential to maintain its activity.
Binding of K5 to HUVEC-K5 binds to these cells in a dose-dependent manner with high affinity (K d of 28 Ϯ 1.37 nM) and to a large number of sites (12.6 Ϯ 0.56 ϫ 10 5 binding sites/cell) (Fig. 5B). The value of the K d is comparable with that determined for binding of K5 to VDAC1 reconstituted in proteoliposomes. Electrophoretic separation of proteins in a HU-VEC lysate followed by a blot binding assay with a rabbit anti-VDAC1 IgG (Fig. 5B, inset) shows only one band of M r ϳ32,000. Binding of K5 to HUVEC is inhibited by Pg, Pg peptides containing K5, or by an IgG fraction against VDAC1 peptide showing structural relatedness to SK (Fig. 5C).
Binding of Anti-VDAC1 Peptide IgG to HUVEC-125 I-Labeled anti-VDAC1 IgG bind to HUVEC in a dose-dependent manner to a large number of sites (B max of 11.6 ϫ 10 5 binding sites/cells) (Fig. 5D). This value is comparable with that determined for the binding of K5 to HUVEC, suggesting VDAC1 as a unique receptor for K5 on the cell surface.
Effect of K5 Binding on HUVEC [Ca 2ϩ ] i and pH i -We also investigated whether K5 binding to HUVEC produced changes in [Ca 2ϩ ] i or pH i and compared these changes with those produced by Pg 2. Pg 2 (100 nM) added to HUVEC induces a transient rise in [Ca 2ϩ ] i lasting about for 90 s before returning to base line (Fig. 6A). Pg 2 also induced a rise in pH i , which was continuous for 400 s (Fig. 6B). A similar concentration (100 nM) of K5 induced a small a rise in [Ca 2ϩ ] i (Fig. 6C) and produced a continuous decrease in pH i during the same time period (Fig.  6D). Incubation of HUVEC with K5 followed by Pg 2 shows a decreased stimulation in [Ca 2ϩ ] i (Fig. 6E); however, the decrease in pH i induced by K5 is abolished after the addition of Pg (Fig. 6F). Incubation of HUVEC with anti-VDAC1 peptide IgG prior to the addition of K5 causes no change in [Ca 2ϩ ] i (Fig. 6G) or pH i (Fig. 6H and Table I).
Effect of K5 on Mitochondrial Membrane Potential-The flu-  orescent dye DSPMϩ was used as an indicator of a coupled membrane potential (⌬) (35). We observed that K5 induces a concentration-dependent increase in ⌬ with mitochondria isolated from 1-LN cells (Fig. 7). DISCUSSION In search of an endothelial cell receptor for K5, a potent suppressor of growth factor-stimulated angiogenesis (9 -12), we investigated the functional relationship between SK and VDAC1, two proteins displaying sequence homologies (16) in a region of SK responsible for binding of K5 (14,15). Our experiments, including direct binding of K5, specific antibody competition, and reconstitution of the membrane protein into liposomes, suggest that VDAC1 is a receptor for K5 on the cell surface. VDACs are a group of proteins that form channels through the outer mitochondrial membrane (35,36) and are also observed in human skeletal muscle, B lymphocyte, and hematopoietic cell plasma membranes (37,38). This is the first study reporting expression of VDAC1 in HUVEC plasma membranes. A receptor binding assay on endothelial cells using 125 I-K5 was previously reported by another group (40); however, no specific binding of K5 to endothelial cells was detected (40). We obtained similar negative results when we used iodination techniques producing tyrosine modification of K5. The integrity of Tyr 512 is required for reactivity of K5 with its ligands (41). The same residue is also essential for the interaction of K5 with Pg residue Lys 50 , which stabilizes Pg in a closed conformation when the protein is in the circulation (42).
Endothelial cell proliferation is preceded by an increase in cytosolic pH i , leading to angiogenesis and the repair of injured endothelial cells (43). Cell proliferation is also dependent on modulation of [Ca 2ϩ ] i (43). Our data suggest that K5 interferes with both cytosolic pH i and [Ca 2ϩ ] i in HUVEC. Because the most abundant protein of the mitochondrial outer membrane is VDAC1 (44), we investigated the interaction of K5 with isolated mitochondria. Regulation of apoptosis involves selective mitochondrial membrane permeabilization (45). Mitochondria are organelles with two well defined compartments: the matrix, surrounded by the inner membrane, and the intermembrane space, surrounded by the outer membrane. The inner membrane is folded into numerous christae to increase its surface area and contains the protein complexes from the electron transport chain, ATP synthase, and the adenine nucleotide translocator. To function properly, the inner membrane is almost impermeable under physiological conditions, thereby allowing the respiratory chain to create an electrochemical gradient (⌬⌿). The ⌬⌿ results from the respiration-driven, electron transport chain-mediated pumping of protons out of the inner membrane and is indispensable for driving ATP synthase, which phosphorylates ADP to ATP. In the aptoptotic program, mitochondrial ⌬⌿ may be either decreased or increased, depending on the mechanism affecting VDAC1 (45). An increase in ⌬⌿ transfers charges to the outer membrane, resulting in the closure of VDAC1 (46,47). An early feature of apoptosis is intracellular acidification, which induces hyperpolarization of mitochondrial outer membrane (47,48). The intracellular acidification induced in HUVEC by K5 via interaction with VDAC1 suggests a mechanism by which K5 may cause apoptosis of endothelial cells (12).
It is thought that VDAC1 is concentrated in caveolae and caveolae-related domains of the plasma membrane (39). Caveolin-1 and -2 are abundantly expressed in normal human endothelial cells (49). VEGF induces endothelial cell proliferation via a mechanism that produces a significant reduction in the expression of caveolin-1. In the absence of VEGF, angiostatin (kringles 1-3) does not affect endothelial cell proliferation or alter the levels of expression of caveolin-1 (50). However, in the presence of VEGF, angiostatin blocks VEGF-induced downregulation of caveolin-1 (50). Caveolae are structures that cluster groups of fibrinolytic proteinases, thus providing a favorable environment for proteinase cooperation. The urokinasetype plasminogen activator receptor is localized in caveolae (51). Receptors for Pg such as annexin II and the ganglioside GM1 are also localized in caveolae (52)(53)(54) along with metalloproteinase 2 (55). Annexin II and the ganglioside GM1 bind Pg via the L-lysine binding site in kringle 1, whereas VDAC binds Pg via a site in K5. In situ degradation of receptor-bound Pg by metalloproteinases may generate anti-angiogenic peptides (kringles 1-3 and K5) (56), which once generated may control angiogenesis via alternative pathways.