A Truncated Plasminogen Activator Inhibitor-1 Protein Blocks the Availability of Heparin-binding Vascular Endothelial Growth Factor A Isoforms*

We have made deletions of the porcine plasminogen activator inhibitor-1 (PAI-1) gene to obtain recombinant truncated PAI-1 proteins to examine functions of the PAI-1 isoforms. We previously reported that one recombinant truncated protein, rPAI-123, induces the formation of angiostatin by cleaving plasmin. The rPAI-123 protein is also able to bind urokinase plasminogen activator and plasminogen and then reduce the amount of plasmin that is formed. We have now prepared three different truncated rPAI-1 proteins and demonstrate that PAI-1 conformations control the release of heparin-binding vascular endothelial growth factor (VEGF) isoforms. The rPAI-123 isoform can regulate the functional activity of heparan sulfate-binding VEGF-A isoforms by blocking the activation of VEGF from heparan sulfate. The rPAI-123 conformation induced extensive apoptosis in cultured endothelial cells and thus reduced the number of proliferating cells. The rPAI-123 isoform inhibited migration of VEGF-stimulated sprouting from chick aortic rings by 65%, thus displaying a role in anti-angiogenic mechanisms. This insight into anti-angiogenic functions related to PAI-1 conformational changes could provide potential intervention points in angiogenesis associated with atherosclerotic plaques and cancer.

Angiogenesis is the formation of new capillary blood vessels as outgrowths of pre-existing vessels. This tightly regulated process normally occurs during development (1,2), tissue repair and remodeling (1,2), and abnormally in pathologic diseases (3). At the onset of angiogenesis, the quiescent endothelium is destabilized into migratory, proliferative endothelial cells. The angiogenic (activated) endothelium is maintained primarily by positive regulatory molecules. In the absence of such molecules, the endothelium remains in a differentiated, quiescent state that is maintained by negative regulatory molecules, which act as angiogenesis inhibitors (4,5). Normally, the negative and positive activities are balanced to maintain the vascular endothelium in quiescence (5,6). A shift in the balance of the positive and negative regulatory molecules can alter the differentiated state of the endothelium from the nonangiogenic, quiescent to the angiogenic state (5). In the switch to pro-angiogenesis, the quiescent endothelial cells are stimulated to migrate toward a chemotactic stimulus, lining up in a tube (sprout) formation (6). They also secrete proteolytic enzymes that degrade the endothelial basement membrane, thus allowing the migrating endothelial cells to extend into the perivascular stroma to begin a new capillary sprout. The angiogenic process is characterized by increased proliferation of endothelial cells to form the extending capillary (6 -9).
Vascular endothelial growth factor (VEGF) 1 is a mitogenic factor that stimulates pro-angiogenic properties, including endothelial cell migration and proliferation. VEGF induces the expression of plasminogen activator proteolytic pathway proteins that participate in cellular invasive and remodeling processes (10 -12). VEGF-A RNA can undergo alternative splicing to produce four isoforms (13)(14)(15). Three of those isoforms, VEGF-A 165 , VEGF-A 189 , and VEGF-A 206 , bind to heparin. VEGF-A 189 and VEGF-A 205 isoforms have a greater affinity for heparin (16,17) than VEGF-A 165 , which can also be active in the soluble form (16). Pro-VEGF affinity for heparin/heparan sulfate appears to be important in the regulation of the availability of VEGF at the cell surface (18), where it can interact with its tyrosine kinase receptors to exert its activity (19). VEGF-A 165 can be released from heparan sulfate in an inactive or active form (20,21). Plasmin and uPA cleave pro-VEGF into an active form of varied sizes depending upon the isoform and the activator molecule (21). Soluble pro-VEGF-A 165 can be cleaved by plasmin to produce a smaller active molecule that can directly bind VEGFR-1 and VEGFR-2 independent of heparan sulfate binding (17).
There are naturally occurring molecules that serve as negative regulators of angiogenesis. Angiostatin, one of those negative regulators, is a 38-to 45-kDa cleavage product of plasminogen, containing kringle domains 1-4 (K1-4) (22,23). Plasminogen, the precursor of plasmin, is activated when it is cleaved at the carboxyl terminus by plasminogen activators. The amino terminus contains five consecutive kringle domains, each of ϳ9 kDa. The greatest inhibitory activity of angiostatin is contained within kringles 1-3 (24) and kringles 1-5 (25). The mechanism for angiostatin inhibition of endothelial cell growth in vitro and angiogenesis in vivo is unclear.
It has been shown that, when PAI-1 is cleaved between residues P and PЈ in the RCL, PAI-1 is converted to a substrate (31,32). In the cleaved conformation, the RCL is partially inserted into the ␤-sheet of strand A, thus making the structure of cleaved and inactive PAI-1 more similar to each other than to active PAI-1. However, it has been demonstrated that there are distinct conformational differences between latent and cleaved PAI-1 (40).
We have recently reported that, when a truncated porcine PAI-1 protein, rPAI-1 23 , is incubated with plasminogen and urokinase plasminogen activator (uPA), it induces formation of an angiostatin-like protein that has proteolytic activity (55). In this reaction, angiostatin is formed from cleaved plasmin. uPA enhances the formation of the angiostatin-like protein by increasing the amount of available plasmin. The proteolytic activity of the 36-kDa angiostatin is ultimately inhibited by increasing amounts of rPAI-1 23 that are available for binding uPA and/or plasminogen. In this second mechanism, rPAI-1 23 reduces the numbers of uPA/plasminogen interactions, thus reducing the amount of plasmin produced. Cultured endothelial cells exposed to rPAI-1 23 exhibit a decrease in proliferation, increased apoptosis, and decreased migration in the presence of VEGF. This truncated PAI-1 appears to expose sites that participate in a functional role for PAI-1 in generating angiostatin fragments from plasmin.
We have previously shown by zymography the importance of rPAI-1 23 interactions with uPA, plasminogen, and plasmin that result in angiostatin formation. We also demonstrated that rPAI-1 23 blocks migration of VEGF-stimulated endothelial cells (55). Because pro-VEGF is able to form complexes with heparan sulfate at the endothelial cell surface where it can be cleaved by uPA or plasmin to produce active VEGF, we considered the potential importance of PAI-1 binding domains for heparin, heparan sulfate, uPA at amino acids 128 -145, and plasmin that may affect VEGF activation. We further engineered the porcine PAI-1 gene to obtain three additional trun-cated PAI-1 proteins, rPAI-1, for functional studies. There are no known functional differences between human and porcine PAI-1 (61,62). Two of the rPAI-1 proteins, rPAI-1 Hep23 and rPAI-1 ⌬23 , are encoded by DNA sequences identical to the rPAI-1 23 (55) except that rPAI-1 Hep23 contains the codons for the entire heparin binding domain, which have been completely deleted in rPAI-1 ⌬23 . The rPAI-1 24 protein is the only rPAI-1 isoform that contains the reactive center loop (RCL) on the carboxyl terminus (amino acids 320 -351). Amino acid residues 262-379 in mature poPAI-1 were deleted from the carboxyl termini of rPAI-1 23 , rPAI-1 Hep23 , and rPAI-1 ⌬23 . The RCL was within the deleted region and, therefore, enabled examination of the importance of the uPA site at residues 128 -145, in the absence of the primary uPA site at Arg-346. We demonstrate here that the rPAI-1 23 conformation is able to block the availability of heparan sulfate-binding VEGF-A, thus inhibiting the angiogenic activity of endothelial cells in vitro and sprouting in an organ culture.
The PCR conditions for all three genes were as described previously (55). The PCR-amplified rPAI-1 DNA fragments were double-digested with EcoRI and XbaI (Roche Molecular Biochemicals, Indianapolis, IN) to activate the incorporated restriction enzyme sites. The restricted DNA was ligated into a Pichia pastoris yeast shuttle vector, pGAPZ␣A (Invitrogen, Carlsbad, CA). The TOP 10 strain of Escherichia coli was transformed by electroporation as described (55). Following an overnight incubation at 37°C, colonies were selected and grown in low salt LB broth for 5-7 h at 37°C. The DNA from each colony growth was isolated (Qiagen, Inc., Valencia, CA) to identify a clone containing each gene insert. Positive isolates from restriction enzyme digests were verified by sequencing. Each recombinant protein was expressed in P. pastoris as described (55) and purified by affinity chromatography. The deletions of poPAI-1 are shown diagramatically in Fig. 1. The sequences of the rPAI-1 23 , rPAI-1 ⌬23 , rPAI-1 Hep23 , and rPAI-1 24 DNA matched the known sequence of the corresponding segment of porcine PAI-1 DNA. Each rPAI-1 protein corresponded to its expected molecular weight (data not shown).

Anti-angiogenic Effects of rPAI-1 Proteins
Characterization of the rPAI Protein Interactions with Heparin, uPA, and Plasminogen-The functionality of the rPAI-1 proteins was first tested by incubating each truncated protein with uPA and plasminogen to assess the proteolytic activity of the products of the reaction, as described (55). The functionality of each rPAI-1 isoform was then tested in a reaction with heparin bound to Sepharose beads. First, the rPAI-1 23 , rPAI-1 ⌬23 , rPAI-1 Hep23 , and rPAI-1 24 proteins (20 g) were each incubated with 50 l of heparin-bound Sepharose beads (Amersham Biosciences, Piscataway, NJ) for 2 h at 37°C. The unbound protein was separated from the heparin-Sepharose-bound protein complex by microcentrifugation at 4°C for 15 min. The proteins bound to the Sepharose beads were washed in TBS/Tween 20 followed by TBS washes.
Identification of Plasminogen Kringle Domains-Protein products from the reaction of rPAI-1, uPA, and plasminogen were electrophoresed on a 4 -20%, SDS, non-reducing polyacrylamide gel, and transferred to nitrocellulose. The blots were probed with an antibody (1 g/ml) specific for plasminogen kringles 1-3 (R&D Systems, Inc., Minneapolis, MN). Following a 1-h, room temperature incubation with the primary antibody, a secondary rabbit anti-goat IgG HC ϩ LC (Pierce, Rockford, IL) polyclonal antibody at a concentration of 1 g/ml was incubated with the anti-kringle probed membrane for 1 h at room temperature. A horseradish peroxidase-conjugated antibody (donkey anti-rabbit IgG, Amersham Biosciences, Arlington Heights, IL) diluted 1:5000 amplified the binding reaction, which was ultimately detected by addition of a chemiluminescent substrate (Amersham Biosciences).
Proliferation-Bovine endothelial cells, treated with exogenous rPAI-1 protein, were plated into six-well culture plates at a density of 1.0 ϫ 10 4 /ml to assess their proliferative properties in the presence of rPAI-1 proteins. The cells were trypsinized and counted on a hemacytometer plate at 48 and 96 h after adding exogenous rPAI-1. To further ascertain the proliferative properties of the rPAI-1-treated cells, a Brd-Urd labeling assay was performed using a FITC-labeled BrdUrd-specific antibody. The addition of propidium iodide (PI) enabled a microscopic count of the proliferating cells relative to total number of cells. BrdUrd was added to the culture medium to obtain a final concentration of 10 M. The cells were incubated for 30 min at 37°C in a CO 2 incubator. The cells were washed twice in phosphate-buffered saline (PBS) containing 1% bovine serum albumin. FITC-conjugated anti-BrdUrd was diluted 2.5-fold in 0.5% Tween 20/PBS and added directly to the cell culture medium for 30 min at room temperature. The cells were washed in PBS and incubated with PI for 1 min. Incorporation of BrdUrd was calculated by counting the number of cells containing FITC stain (green) or PI stain (red) in five fields per sample in triplicate experiments.
rPAI-1 Effect on Tubule Formation in a Chick Aortic Arch Ring Assay-Aortic arches were removed (R. Auerbach, University of Wisconsin, personal communication) from fertilized chicken eggs (Oliver Merrill & Sons, Londonderry, NH) at day 14 of embryonic development. The eggs were cracked into a sterile 100-mm culture dish. The embryo was removed from its surroundings by cutting away the associated membranes and yolk sac. The chick embryo was placed ventral side up to surgically expose the heart and aortic arches. The heart and aortic arch were removed and placed into a sterile culture dish containing PBS to which 1% penicillin-streptomycin (Invitrogen, Gaithersburg, MD) was added. Arches, from which the surrounding adventitia had FIG. 1. Truncated rPAI-1 DNA relative to full-length human and porcine PAI-1 genes. This figure defines the regions of poPAI-1 sequences that correspond to the huPAI-1. The sequences of huPAI-1 that code for domains that interact with specific molecules in the proteolytic, fibrinolytic, and adhesion processes are indicated with arrows and bars. The bars are of two separate designs to indicate overlapping domains. The poPAI-1 and huPAI-1 were aligned to locate the same coding regions in poPAI-1. Deletions of the poPAI-1were based on the aligned sequences of the two genes. The deletions that were made in poPAI-1 to obtain rPAI-1 23 , rPAI-1 ⌬23 , rPAI-1 24 , and rPAI-1 Hep23 are shown. been removed, were cut into 0.8-mm sections. Each arch was placed onto 1-5 l of Matrigel (66) that was deposited on the bottom of a six-well culture plate just prior to adding the ring. An additional 300 l of ice-cold Matrigel was spread in a circle surrounding each aortic arch. The Matrigel was allowed to solidify before adding 2 ml of human endothelial-SFM basal growth medium (Invitrogen). An rPAI-1 protein and bovine brain extract (BBE) (Clonetics, San Diego, CA), at 30 nM and 10%, respectively, were added to each well and incubated at 37°C, 5% CO 2 . To assess the characteristics of the new sprouts that could be ascribed to VEGF and the characteristics that were inhibited by rPAI-1 23 , VEGF-A (100 ng/ml), and rPAI-1 23 (30 nM) were added to the culture medium containing the aortic rings. At 48 h, additional medium containing BBE or VEGF and rPAI-1 protein was added to the aortic rings. Growth at 37°C was continued for an additional 48 h. Quantitative evaluation of tubule formation was performed by a blinded observer on a scale of 1-5 (least to maximum sprouting).
Biochemical Interactions of a VEGF-heparin Complex with rPAI-1 Proteins, uPA, and Plasmin-VEGF was isolated from bovine aortic endothelial cells by first incubating the cells overnight in serum-free DMEM. The medium was changed before adding 100 g/ml heparin (Sigma, St. Louis, MO) for 4 h at 37°C. The serum-free DMEM containing the heparin-VEGF complex was precipitated in 80% ethanol. The serum-free medium from which the VEGF-heparin complex was isolated, was probed for VEGF-A 121 (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) to ensure that this non-heparin binding VEGF isoform did not precipitate with the VEGF-heparin complex. VEGF-A 121 was not detected. The heparin-VEGF complex was incubated with rPAI-1 23 , rPAI-1 ⌬23 , rPAI-1 24 , or rPAI-1 Hep23 protein (3, 15, and 30 nM) for 2 h at 37°C. Either uPA (0.25 IU) or plasmin (made from 1.0 IU plasminogen and 0.25 IU uPA) was added to the VEGF-heparin-rPAI-1 reaction. After an additional 2-h incubation at 37°C with uPA or plasmin, DTT at a final concentration of 0.1 M was added to one half of the reaction mixture for 3.5 h at 37°C. An equal volume of each reaction mixture was denatured at 95°C, then electrophoresed on a 4 -20%, SDS-polyacrylamide gel. The VEGF in each sample was visualized on a nitrocellulose membrane probed with a monoclonal antibody to human VEGF-A 165, 189, 206 (BD Pharmingen, San Diego, CA). A secondary rabbit anti-mouse IgG HCϩLC (Pierce, Rockford, IL) polyclonal antibody was diluted to 1 g/ml in TBS, pH 8.0, containing 5% skim milk, and incubated with the anti-VEGF probed membranes for 1 h at room temperature. A horseradish peroxidase-conjugated antibody (donkey anti-rabbit IgG, Amersham Biosciences, Arlington Heights, IL) diluted 1:500 amplified the binding reaction, which was ultimately detected by addition of a chemiluminescent substrate (Amersham Biosciences).
VEGF-A in the Culture Medium of rPAI-1-treated Endothelial Cells-Bovine aortic endothelial cells were seeded into six-well culture plates and grown to confluence in DMEM, as described. The confluent (quiescent) cells were treated with a single dose of rPAI-1 protein at a final concentration of 1.2 nM. At 6, 12, 15, 24, 30, 48, and 72 h after the onset of treatment, the culture medium was removed from cells treated with each rPAI-1 protein. Equivalent amounts of protein were incubated with 0.1 M DTT for 2.5 h at 37°C. The protein samples were electrophoresed on a 4 -20% gradient, SDS-polyacrylamide gel. The proteins were transferred to a nitrocellulose membrane and probed for VEGF-A 165, 189, 206 , as described.
VEGF-A Bound to Heparan Sulfate in rPAI-1-treated Endothelial Cells-Bovine aortic endothelial cells were seeded into six-well culture plates and treated with rPAI-1 proteins, as described. Following a 6and 12-h incubation with the rPAI-1 proteins, the culture medium was removed, the cell layer was washed twice in HBSS. The cells were incubated for 1 h at 37°C in 1 ml of HBSS-containing 0.05 IU of heparinase III (Sigma, St. Louis, MO). Following the incubation, the HBSS-containing proteins released during the enzymatic digest were collected. The proteins were concentrated in 80% ethanol. An equivalent amount of protein from each sample was incubated with 0.1 M DTT at 37°C for 2.5 h prior to electrophoresis in a 4 -20% gradient SDSpolyacrylamide gel. The separated proteins were transferred to an immunoblot and probed for VEGF-A using an antibody specific for epitopes common to VEGF-A 165, 189, 209 (BD Pharmingen, San Diego, CA). In another series of experiments the immunoblots were probed simultaneously with two competing antibodies to VEGF-A; one antibody was specific for the active site of VEGF-A 165 (R&D Systems, Inc., Minneapolis, MN) and the second antibody was raised against an epitope common to VEGF-A 165, 189, 206 . The binding reactions occurred at 4°C for 15 h. The binding reaction was further amplified, as described. The chemiluminescent detection was performed, as described.
Apoptosis in rPAI-1 23 -treated BAEC versus PAEC-Bovine aortic endothelial cells (BAEC) and porcine aortic endothelial cells (PAEC) were selected for comparison of apoptosis based on the expression of VEGFR-1 and VEGFR-2. BAECs express both receptors, and PAECs do not express either receptor (67). BAEC and PAEC were seeded into T-25 flasks containing DMEM supplemented with 10% fetal bovine serum, penicillin/streptomycin (100 IU/ml), and L-glutamine (0.292 mg/ml). The cells were incubated at 37°C, 5% CO 2 until the cells reached confluence, at which time fresh culture medium containing 1.2 nM rPAI-1 23 was added to the cells. The treated cells continued to grow for an additional 36 h before harvesting for analysis of apoptosis in an Annexin V assay, as described.

rPAI-1 Protein Interactions and Anti-angiogenic Activity in Vitro
rPAI Protein Interactions with Heparin, uPA, and Plasminogen-We first analyzed the effects of the recombinant truncated PAI-1 molecules on protease activity in the presence of 1) uPA and plasminogen or 2) heparin, uPA, and plasminogen. The zymogram, Fig. 2, lanes 4 -7, shows that the interaction of three of the four truncated PAI-1 proteins with uPA and plasminogen resulted in production of one or more proteolytic fragments that migrate at 34 -38 kDa. The rPAI-1 23 protein (lane 6) induces formation of two 34-to 38-kDa proteolytic angiostatin fragments from plasmin. However, the rPAI-1 ⌬23 (lane 5) and rPAI-1 24 (lane 7) proteins each have a single band that corresponds to one of the 34 -38-kDa fragments visualized in the rPAI-1 23 products. In the case of rPAI-1 ⌬23 (lane 5), a proteolytic band appears at or near the size of the lower proteolytic fragment induced by rPAI-1 23 cleavage of plasmin. The rPAI-1 24 (lane 7) protein induces a proteolytic fragment at or near the molecular mass corresponding to the larger of the two plasmin cleavage products induced by rPAI-1 23 . The rPAI-1 Hep23 does not produce a proteolytic fragment at 34 -38 kDa (lane 4). The rPAI-1 23 protein (partial heparin binding domain) maintains its activity when bound to heparin (lane 3). Similarly, the proteolytic activity associated with a reaction mix containing uPA, plasminogen, and rPAI-1 Hep23 (complete heparin domain) is not altered in the presence of heparin (lane 2). The rPAI-1 ⌬23 protein (lacks a heparin binding domain) and the rPAI-1 24 protein (contains the RCL and a partial heparin domain) did not demonstrate proteolytic plasmin cleavage products when incubated with heparin. The proteolytic proteins near 80-kDa correspond to plasmin. In lanes 3 and 6 there are proteolytic proteins near 50 kDa that may represent a different plasmin cleavage product containing a greater number of plasminogen kringle domains. The function of the rPAI-FIG. 2. rPAI-1 affinities for heparin and resulting proteolytic activity. Each of the rPAI-1 proteins (in excess) were: 1) incubated with uPA (0.5 IU) for 1 h, at 37°C before adding plasminogen (Plg) (1 IU) for an additional 1 h incubation at 37°C or 2) incubated with heparinbound Sepharose beads for 1 h, at 37°C before the incubation with uPA and plasminogen to assess the function of each rPAI-1 protein and the importance of the heparin-binding domain in the rPAI-1 function. The reacted samples were electrophoresed on a 1.3% casein polyacrylamide gel to visualize the proteolysis associated with each rPAI-1 protein. 1 Hep23 isoform (complete heparin domain) in a reaction with uPA and plasminogen as compared with the rPAI-1 23 isoform (partial heparin binding domain) shows that binding to heparin does not alter the inability of rPAI-1 Hep23 to mediate the formation of proteolytic fragments at a molecular mass near 34 kDa. These experiments show that a full heparin-binding domain can block the ability of PAI-1 to induce proteolytic proteins corresponding to or near the molecular mass of angiostatin containing K1-3.
Proliferation-Bovine aortic endothelial cells treated with rPAI-1 protein were assessed for their continued ability to proliferate after exposure to rPAI-1 proteins. The control cells proliferated at a rate of approximately one doubling in 48 h (Fig. 5A). The rPAI-1 24 -treated cells did not double in number during the first 48 h, but doubled in the subsequent 48 h. The rPAI-1 23 -and rPAI-1 ⌬23 -treated cells did not increase in number in the 96 h test period. In fact, the rPAI-1 23 -treated cells decreased their number by 80% between 48 and 96 h. This result was not surprising due to the high rate of apoptosis that occurred during the first 3 days of exposure to rPAI-1 23 . There were 46% fewer apoptotic rPAI-1 ⌬23 -treated cells. Nearly 100% of the rPAI-1 23 -and rPAI-1 ⌬23 -treated endothelial cells examined in a BrdUrd labeling assay incorporated BrdUrd into DNA (Fig. 5B); thus, supporting low density in rPAI-1 23 -and rPAI-1 ⌬23 -treated cells as a result of reduced number and not loss of proliferation capability. The heparan sulfate binding rPAI-1 protein supports full proliferation and cell survival, whereas the poor or non-heparan sulfate binding proteins impair cell number.
Evaluation of rPAI-1 Effect on New Tubule Formation in a Chick Aortic Arch Ring Assay-The aortic rings from 14-day chick embryos formed tubules in Matrigel. The tubules proliferated and migrated extensively when stimulated with BBE, as shown in Fig. 6B. Here we tested the ability of the recombinant PAI-1 proteins on in vitro angiogenesis. The rings that were exposed to rPAI-1 Hep23 -BBE, shown in Fig. 6D, had a proliferation and migration rate at least equivalent to the control (Fig.  6B). In contrast, after 3 days of exposure to rPAI-1 23 -BBE, new tubules extended from the aortic rings to ϳ50% of the length measured in the control or rPAI-1 Hep23 -treated rings (data not shown). By day 4, that difference was 65% as shown in Fig. 6A. The newly formed tubules from the aortic rings treated with rPAI-1 ⌬23 migrate about 50% less than the control on day 3 (data not shown). That difference remains nearly the same on day 4 (Fig. 6C). There is a greater amount of proliferation near the periphery of the ring of rPAI-1 ⌬23 -treated samples when compared with the rPAI-1 23 -treated rings in Fig. 6A. The branches of the rPAI-1 23 -and rPAI-1 ⌬23 -treated rings are more flattened and tightly connected. Their branches appear to fuse as they extend in parallel. The differences observed in the newly formed tubules in the rPAI-1-treated samples are consistent with the data presented in Figs. 2-5, where it is shown   FIG. 3. Verification of plasminogen kringle 1-3 domains in  rPAI-1, uPA and plasminogen reactions. Each rPAI-1 protein (3, 15, and 30 nM) was incubated with uPA for 1h at 37°C before adding plasminogen for a 2 nd 1h incubation at 37°C. The final products of the reaction mixture were transferred to nitrocellulose and probed for plasminogen kringles 1-3. In the legend, the rPAI-1 proteins are identified as their subscript: rPAI-1 23 ϭ 23; rPAI-1 ⌬23 ϭ ⌬23; rPAI-1 24 ϭ 24; rPAI-1 Hep23 ϭ Hep23.  4. Apoptosis of rPAI-1-treated endothelial cells. Bovine aortic endothelial cells were grown to confluence in DMEM containing 10% FBS before adding equivalent amounts of rPAI-1 proteins to fresh culture medium. The endothelial cells were exposed to the rPAI-1 protein for 24 h. The cells were then analyzed for apoptosis in an Annexin V assay.
Migration appeared to be VEGF-dependent. By day 4, The  Fig. 3A were verified by performing a BrdUrd labeling assay to determine the numbers of cells that were undergoing DNA synthesis. Both rPAI-1 23 -and rPAI-1 ⌬23 -treated cells declined in number between 48 and 96 h, even though they were proliferating as determined by the BrdUrd assay.
FIG. 6. Ex vivo evaluation of rPAI-1 effect on endothelial cell tubule formation in a chick aortic arch ring assay. The aortic arch, surgically removed from 14-day-old chick embryos, was sectioned into rings of ϳ0.8 mm. The rings were placed on Matrigel to promote tubule formation in endothelial-SFM basal growth medium containing 10% bovine brain extract (BBE). Either rPAI-1 23 , rPAI-1 ⌬23 , or rPAI-1 Hep23 was added (30 nM) to the medium containing BBE. Growth of tubules took place at 37°C, 5% CO 2 for 4 days. At day 2, additional medium, rPAI-1, and BBE was added. In B and D, the migration and proliferation of the tubules extend across the entire field at a ϫ20 magnification. In A and C, the tubule length begins at the edge of the ring seen as black/dark gray and defined by arrows. A, rPAI-1 23 /BBE-treated; B, BBE-treated; C, rPAI-1 ⌬23 /BBE-treated; and D, rPAI-1 Hep23 /BBE-treated. rPAI-1 23 -VEGF-treated rings (Fig. 7B) display a reduction in the rate of migration similar to that measured in the rPAI-1 23 / BBE-treated samples (Fig. 6A). The significant differences in the structure of the VEGF-treated control tubules (Fig. 7B) and the rPAI-1 23 -VEGF-treated samples (Fig. 7A) should be noted. These data show that rPAI-1 23 and rPAI-1 ⌬23 are able to inhibit the migratory function of new sprouts from chick aortic rings stimulated with BBE. The rPAI-1 23 protein inhibits the migration of new sprouts stimulated with VEGF. In the rPAI-1 23and rPAI-1 ⌬23 -treated aortic rings we observed apoptosis of sprouting endothelial cells. The apoptosis appears to result in breakage of the tubule.

Analyses of Biochemical Interactions of a VEGF-Heparin Complex with rPAI-1 Proteins, uPA, and Plasmin
We next determined if the rPAI-1 proteins could regulate the release or activation of VEGF-A from a complex with heparin. Such information could provide insight into the mechanisms by which rPAI-1 23 inhibits the migration of VEGF-stimulated sprouts from chick aortic rings. Therefore, variable concentrations of each rPAI-1 protein were incubated with a VEGFheparin complex to examine the release and/or activation of VEGF-A. VEGF activation and release from the complexes were then tested in the presence of activator molecules uPA or plasmin. Western blots of the complexes were probed with an antibody to heparin-binding VEGF-A isoforms (Fig. 8, A-D,  and Fig. 9, A-C).
Release or Activation of VEGF-A 165, 189, 206 -In this set of experiments, blot membranes were probed for released and/or activated VEGF-A 165, 189, 206 from an rPAI-1-VEGF-heparin complex containing activator molecules uPA or plasmin. When rPAI-1 23 is part of the complex, there are only traces of the 46-kDa activated VEGF-A 165, 189, 206 (Fig. 8A, lanes 5-13). Fig.   8B shows the effect of rPAI-1 ⌬23 in a reaction mixture containing VEGF-heparin and uPA or plasmin. VEGF is released from heparin in the absence (lanes 6 -8) or presence of either uPA (lanes 9 -11) or plasmin (lanes 12-14). Similarly, the presence of rPAI-1 24 in the mixture with VEGF-heparin results in the release of active VEGF at all concentrations of rPAI-1 24 as shown in Fig. 8C, lanes 6 -14. In the rPAI-1 24 -containing reactions, the VEGF release occurs in the presence of uPA (lanes 9 -11) or plasmin (lanes [12][13][14]. Reaction mixtures containing rPAI-1 Hep23 (Fig. 8D, lanes 6 -14) show activated VEGF in reactions containing low concentration of rPAI-1 Hep23 in the absence (lane 6) or presence of either uPA (lane 9) or plasmin (lane 12). At higher concentrations of rPAI-1 Hep23 , there is blockage of VEGF release and an increase in VEGF-containing fragments between 60 and 80 kDa (Fig. 8D, lanes 7, 8, 10, 11,  13, and 14). The high molecular mass fragments containing VEGF have cross-reactivity with VEGF-B 186 (data not shown) and are indicative of the 60-to 62-kDa active VEGF-B 186 homodimer (23). The greater than 80-kDa molecular mass containing VEGF also exists in samples containing rPAI-1 23 , rPAI-1 ⌬23 , and rPAI-1 24 . However, the samples containing rPAI-1 Hep23 clearly have a greater amount of VEGF-A complexed at a high molecular mass. These experiments show that rPAI-1 ⌬23 and rPAI-1 24 do not block the activation or release of VEGF from a complex with heparin. Although rPAI-1 23 and rPAI-1 24 both have a partial heparin-binding domain, only rPAI-1 24 has a RCL. The RCL may alter the conformation such that the partial heparin domain in rPAI-1 24 is obscured and unable to block the release of activated VEGF. On the other hand, both rPAI-1 23 and rPAI-1 Hep23 are able to block the release of VEGF from a complex with heparin. These data suggest that the heparin-binding domain in each of these two isoforms participates in blocking the release of active VEGF-A and that partial FIG. 7. Evaluation of rPAI-1 23 effect on VEGF-stimulated tubule formation in a chick aortic arch. The VEGF contribution to the migration and/or proliferation of tubules formed from the aortic rings was assessed by adding VEGF (100 ng/ml) as the sole growth factor in place of BBE, which contains various growth factors. The outgrowth/migratory capability of the tubules was inhibited by ϳ65% when rPAI-1 23 was an additive to the medium containing VEGF. The pattern of tubule formation, extension, and branching is very different in the rPAI-1 23 -treated samples (B). A, VEGF-treated; B, rPAI-1 23 /VEGF-treated. heparin binding is more effective in blocking VEGF-A activation and/or release. These findings suggest that the ability of rPAI-1 23 to block growth and vessel sprouting may be due to its ability to block VEGF activation.
In VEGF-heparin complexes containing lower concentrations of rPAI-1 23 protein, there are activated VEGF fragments between 30 and 42 kDa, as shown in lanes 3 and 4, which correspond to the reported sizes for processed VEGF-A 165, 189, 206 . The fragments in Fig. 9A (lanes 3 and 4) that migrate at 36 -38 kDa are also in reactions containing uPA (Fig. 8B, lanes 4 and 5) or plasmin (Fig. 8C, lanes 2 and 5). The most predominant VEGF fragment released from a VEGF-heparin-rPAI-1 23 complex is seen near 50 kDa (mature VEGF) in the absence of uPA and plasmin. When the rPAI-1 23 protein concentration is increased to 30 nM, the active VEGF at 50 kDa is absent and the products are ϳ30, 38 -42 kDa, which corresponds in size to uPA-matured VEGF-A 189 and VEGF-A 165 , or plasmin-activated VEGF-A 189 (Fig. 9, A, lane 5, and B and C, lanes 6). The data obtained from the reducing reactions demonstrate that the rPAI-1 23 conformation blocks the release of multiple forms of matured, activated, and processed (uPA and plasmin cleaved) VEGF-A fragments reported for heparin-binding VEGF-A 165, 189 isoforms. In mixtures containing a VEGF-heparin-rPAI-1 Hep23 complex, the associated VEGF remains at a higher molecular mass (greater than 80 kDa) (data not shown), except when rPAI-1 Hep23 is at a low concentration. The rPAI-1 Hep23 protein conformation maintains VEGF-A 165, 189, 206 in a complex with heparin.

VEGF-A in the Culture Medium of rPAI-1-treated Endothelial Cells
The results of the analysis of VEGF-A contained within the culture medium of rPAI-1-treated endothelial cells is shown in Fig. 10 (A-C). The rPAI-1 23 -treated cells (Fig. 10A, lanes 3-8) primarily contain VEGF-A fragments at a molecular mass greater than 50 kDa. A small fraction of VEGF fragments at a molecular mass less than 50 kDa (lanes 4 -8) are also observed in the rPAI-1 23 -treated samples. The fragments greater than 50 kDa are representative of mature or pro-VEGF and those less than 50 kDa correspond to active VEGF. In the culture medium samples collected from rPAI-1 ⌬23 -treated cells (Fig.  10A, lanes 9 -14) there is an abundance of VEGF-A fragments at a molecular mass of ϳ36 -45 kDa at all time points. This molecular mass corresponds to active or uPA/plasmin processed, active VEGF-A. All rPAI-1 ⌬23 -treated samples also contain mature or pro-VEGF at a molecular mass greater than 50 kDa. One of the inactive fragments in the 15-h time (lane 10) is absent in lanes 9, 11-14. The culture media from all untreated (Fig. 10B, lanes 4 -8) and rPAI-1 24 -treated cells (Fig. 10B, lanes  9 and 10), contain active VEGF and a small amount of processed, activate VEGF. The culture medium samples from the rPAI-1 Hep23 -treated cells (Fig. 10C) primarily contain inactive VEGF and lesser amounts of processed, active VEGF. The results of these experiments clearly show that VEGF in rPAI-1 ⌬23 -treated culture medium contains a much greater amount of active VEGF, mostly representative of uPA or plasmin processed VEGF. Similar VEGF fragments are absent or present to a lesser degree in the untreated or rPAI-1 23 -, rPAI-1 Hep23 -, and rPAI-1 24 -treated cells. The culture media from rPAI-1 24treated and untreated cells contain a greater amount of active, unprocessed VEGF than the media from rPAI-1 23 -, rPAI-1 ⌬23 -, or rPAI-1 Hep23 -treated cells. These data correspond with the biochemical analysis of VEGF release from a complex with heparin and rPAI-1 proteins (Fig. 8, A-D). The molecular mass of all of the VEGF fragments correspond to dimeric VEGF-A despite the rigorous reducing conditions applied to all samples. Such results have been reported by others (20).

VEGF-A Bound to Heparan Sulfate in rPAI-1-treated Endothelial Cells
In the series of experiments shown in Fig. 11 (A and B), two different antibodies were used to delineate differences in heparan sulfate-bound VEGF-A isoforms in rPAI-1-treated cells. In Fig. 11A, immunoblots were probed for VEGF-A with an antibody specific for epitopes common to VEGF-A 165, 189, 205 . The results of this set of experiments showed that, at the 12-h time point, the rPAI-1 Hep23 - (Fig. 11A, lane 6) and rPAI-1 24 -treated (lane 11) cells contained two fragments of VEGF-A at a molecular mass near 58 kDa. In the rPAI-1 23 -treated (lanes 3 and 4), rPAI-1 ⌬23 -treated (lanes 8 and 9), and untreated (lane 7) samples, VEGF-A is absent or barely visible. However, when two competing antibodies specific for VEGF-A were simultaneously incubated with immunoblots containing proteins released from a heparinase digest (Fig. 11B), the rPAI-1 23 -treated (lane 1), rPAI-1 ⌬23 -treated (lane 10), and untreated (lane 6) cells show VEGF-A released from heparan sulfate. Among those samples, the rPAI-1 23 -treated cells at the 6-h time point had the greatest amount of detectable VEGF-A released in the enzymatic digest. The VEGF is seen as two distinct fragments with a small difference in molecular mass. Each fragment corresponds to the molecular mass of dimeric VEGF, despite the rigorous reducing conditions. By 12 h, VEGF is not detected in rPAI-1 23 -treated cells (lane 2). However, at the 12-h time point, the rPAI-1 ⌬23 -treated (lane 10) and the untreated cells (lane 6) show a variable molecular mass in VEGF released as a result of the digest. The VEGF fragments in the rPAI-1 23 -treated samples are very close to 50 kDa, whereas, the VEGF fragments in the untreated and the rPAI-1 ⌬23 -treated cells correspond to a molecular mass that is more representative of VEGF-A 189 or VEGF-A 206 . The rPAI-1 Hep23 -and rPAI-1 24 -treated cells do not show the release of heparan sulfate-bound VEGF when the blot is probed with competing antibodies (lanes 5, 6, 13, and 14). The results of these sets of experiments show that different VEGF-A isoforms were released with the heparinase digest, depending upon the rPAI-1 treatment. The use of competing antibodies exposed binding sites in rPAI-1 23 -treated, rPAI-1 ⌬23 -treated, and untreated cells that were not detected with the single antibody probe for VEGF-A 165, 189, 205 . The most pronounced difference is the intensity and molecular mass of VEGF-A at the 6-h time point in rPAI-1 23 -treated cells. It corresponds to a molecular mass near that reported for dimeric mature/pro-VEGF-A 165 and/or mature VEGF-A 189 . Because the VEGF fragments became intensely visible upon competition with an antibody specific for the active site of VEGF-A 165 , the data suggest that the fragments slightly above 50 kDa are VEGF-A 165 . These data also show that the conformation of VEGF is altered as a result of rPAI-1 23 treatment. It should also be noted that the presence of ECM-associated VEGF-A at the 6-h time point is coordinate with the absence of activated VEGF in the culture medium of rPAI-1 23 -treated cells at 6 h (Fig. 10A, lane 3). The VEGF-A fragments that become visible at the 12-h time points in untreated or rPAI-1 ⌬23 -treated cells are at a greater molecular mass that would correspond to pro-VEGF-A 189 or pro-VEGF-A 206 , the isoforms that have the greatest affinity for heparan sulfate. Similarly, the VEGF detected in all samples is representative of dimeric VEGF, despite the reducing conditions. FIG. 9. Western blot analysis of VEGF-A 165, 189, 206 activation in a VEGF-heparin complex with rPAI-1 proteins, uPA, or plasmin. Each rPAI-1 protein (3, 15, and 30 nM) was incubated with a VEGF-heparin complex for 2 h at 37°C before adding either uPA (0.25 IU) or plasmin. The incubation continued at 37°C for 2 additional h. To determine the heparin-binding VEGF-A fragments that were part of a complex with heparin and rPAI-1 proteins, the samples were reduced in 0.1 M DTT at 37°C for 3.5 h. The samples were electrophoresed on a 4 -20% polyacrylamide gel. Membranes containing the transferred protein were probed for human VEGF 165,189,206 . This figure shows the results of the biochemical reactions that include rPAI-1 23 Fig. 12 shows the results of the Annexin V measurement of apoptosis in bovine and porcine aortic endothelial cells treated with rPAI-1 23 protein. These data show that rPAI-1 23 treatment resulted in 300% more apoptotic bovine aortic endothelial cells as compared with porcine aortic endothelial cells. Because PAEC do not express VEGFR-1 or VEGFR-2, these experiments clearly suggest that most (but not all) of the apoptosis induced by rPAI-1 23 is through a mechanism that blocks/inhibits VEGF receptors 1 and/or 2.

DISCUSSION
Truncated PAI-1 proteins were used as biological tools to study anti-angiogenic functions of PAI-1. We demonstrate that, in the absence of the reactive center loop that was contained within deleted residues 262-379, the rPAI-1 Hep23 isoform, which contains the full heparin (heparan sulfate)-binding domain, does not inhibit angiogenic functions. Cultured endothelial cells treated with rPAI-1 Hep23 did not undergo apoptosis, and they proliferated at a rate comparable to controls. Angiogenic tubules from embryonic chick aortic rings migrated and proliferated at levels equivalent to the control samples.
When varied concentrations of rPAI-1 Hep23 were reacted with VEGF-A and heparin, higher concentrations prevented the release of VEGF-A from that complex (Fig. 8D). VEGF-A 206 appears to be the isoform bound to heparan sulfate in the extracellular matrix of bovine aortic endothelial cells after the initial 12 h of rPAI-1 Hep23 treatment. The culture medium primarily contains inactive populations of VEGF-A isoforms, as well as small amounts of activated VEGF-A 165 and/or VEGF-A 189 . Immunoblots probed for VEGF-B show that the rPAI-1 Hep23 -VEGF-heparin complex also contains VEGF-B (data not shown). This interesting aspect of rPAI-1 Hep23 is being investigated.
The rPAI-1 24 isoform, which only has deletions at the amino terminus and, therefore, has the RCL, did not: 1) block the release of activated VEGF-A isoforms in vitro or in biochemical reactions, 2) induce apoptosis in BAEC, and 3) inhibit the migration and proliferation of angiogenic tubules in an organ culture. This suggests that the residues at the carboxyl terminus introduce a conformation in rPAI-1 24 that alters the accessibility of important binding domains that enable rPAI-1 23 to function in anti-angiogenic mechanisms.
The rPAI-1 ⌬23 isoform, in which the entire heparin-binding domain and residues 262-379 have been deleted, is able to: 1) mediate the production of proteolytic fragments that contain angiostatin kringles 1-3, 2) induce apoptosis in 22% of the treated BAEC, and 3) reduce tubule migration by 50% in an organ culture. However, rPAI-1 ⌬23 does not block the release of activated heparin-binding VEGF-A in biochemical reactions. That evidence is supported by the in vitro data (Fig. 10A), which shows an abundance of active VEGF-A of variable sizes in the culture medium of endothelial cells treated with rPAI-1 ⌬23 . The VEGF-A fragments correspond in size to those reported for active VEGF-A 165 and uPA or plasmin-activated VEGF-A 165 and/or VEGF-A 189 . The interesting aspect of uPA and plasmin activated VEGF-A 165, 189 is their ability to bind directly to VEGFR-1 and VEGFR-2 independently of heparan sulfate. A comparison of the levels of activated VEGF-A in the culture medium of rPAI-1 ⌬23 -and rPAI-1 Hep23 -treated cells would suggest that plasmin-and uPA-activated VEGF-A are either not important in the pro-angiogenic mechanisms or the active VEGF is bound to VEGF receptors in rPAI-1 Hep23treated cells. In contrast, the culture media from rPAI-1 ⌬23treated cells contain substantial amounts of activated VEGF-A that are either: 1) representative of a percentage of the total pool of active VEGF where a portion is bound and unbound or 2) the activated VEGF-A is available to bind the receptors, but apoptosis occurs through a mechanism other than VEGF inhi-bition, thereby, reducing the number of VEGF receptors. The data would suggest that cleavage of plasmin into angiostatin is the apoptosis mechanism that may limit VEGF receptor numbers. The active VEGF-A may bind the surviving endothelial cells, enabling them to proliferate. Such a mechanism would account for the rPAI-1 ⌬23 and rPAI-1 23 differences in cell number (Fig. 5), percentage of apoptotic cells (Fig. 4), and migration of sprouting tubules in an organ culture (Fig. 6).
The rPAI-1 23 isoform, which also has residues 262-379 deleted, but contains a partial heparin domain, is able to: 1) produce, in a reaction with uPA and plasminogen, proteolytic fragments corresponding to angiostatin kringles 1-3; 2) induce apoptosis in 39% of the adherent-treated BAEC; 3) inhibit tubule migration by 65% in an organ culture; 4) block the release of heparin-binding VEGF-A isoforms in biochemical reactions (Figs. 8 and 9); 5) block the release of VEGF-A 165 and/or VEGF-A 189 from heparan sulfate following 6 h of rPAI-1 23 treatment; and 6) reduce the activation of VEGF-A 165 and/or VEGF-A 189 .
The culture medium of rPAI-1 23 -treated cells contains small amounts of activated VEGF when compared with that found in rPAI-1 ⌬23 -treated cells. Because the rPAI-1 23 -treated cells are undergoing a higher level of apoptosis after 24 h of treatment, a larger pool of activated, receptor-bound VEGF is not likely. The collective data suggest that rPAI-1 23 partial binding to heparan sulfate introduces a conformational change in rPAI-1 23 that enables it to control uPA and/or plasmin activation of pro-VEGF bound to heparan sulfate. We demonstrated in Fig.  2 that rPAI-1 23 bound to heparin is able to maintain its ability to cleave plasmin. However, plasmin-activated VEGF has reduced mitogenic activity and by itself would not account for the high level of anti-angiogenic activity associated with rPAI-1 23 (17). The heparan sulfate-bound VEGF-A in the rPAI-1 23treated cells shows that the active site of VEGF-A 165 is exposed, which demonstrates that rPAI-1 23 bound to heparan sulfate alters the conformation of VEGF-A. The conformational change could result in one of the following mechanisms: 1) rPAI-1 23 binds VEGF-A or 2) rPAI-1 23 binds uPA at PAI-1 residues 110 -145 to prevent uPA from activating heparan sulfate-bound or soluble VEGF-A. Both mechanisms are currently under investigation.
The rPAI-1 23 and rPAI-1 ⌬23 isoforms demonstrate anti-angiogenic properties. However, there are observed and measured differences between the two isoforms. Based on the differences presented, we propose two mechanisms that contribute to the additional apoptosis in rPAI-1 23 -treated BAEC. One mechanism is early blockage of VEGF activation, which blocks VEGFregulated survival signaling pathways to result in apoptosis of endothelial cells. PAEC, which do not express VEGFR-1 or VEGFR-2, were used as a tool to make a distinction between the apoptosis resulting from rPAI-1 23 blockage of VEGF functional activity in BAEC and a second apoptosis mechanism. The results of those experiments (Fig. 12) support a mechanism whereby rPAI-1 23 blocks activation of VEGF-A to prevent (or limit) VEGF binding to its receptors. The 65% reduction in migration of VEGF-stimulated tubules from rPAI-1 23 -treated embryonic chick aortic rings supports a mechanism that inhibits VEGF-A function.
The combined data suggest a second apoptosis mechanism induced by rPAI-1 23 that is less significant than the apoptosis related to blockage of VEGF functional activity. Angiostatin and other angiogenesis inhibitors are associated with the induction of apoptosis (68 -72). The role of rPAI-1 23 and rPAI-1 ⌬23 in the production of angiostatin kringles 1-3 in reactions with uPA and plasminogen suggests that production of angiostatin may be one mechanism that induces apoptosis in FIG. 12. Apoptosis of rPAI-1 23 -treated BAEC and PAEC. Bovine aortic endothelial cells (BAEC) and porcine aortic endothelial cells (PAEC) were grown in DMEM containing 10% FBS at 37°C, 5% CO 2 until the cells reached confluence, at which time fresh culture medium containing 1.2 nM rPAI-1 23 was added to the cells. The treated cells were incubated at 37°C for 36 h, harvested, and analyzed for apoptosis in an Annexin V assay. rPAI-1 23 -and rPAI-1 ⌬23 -treated endothelial cells. Thus, rPAI-1 23 has two potential anti-angiogenic mechanisms that result in enhanced apoptosis: blockage of VEGF function and induction of angiostatin production through cleavage of plasmin. We have not reported the specific binding interactions that occur in a reaction containing rPAI-1 23 , uPA, and plasminogen, 2 but we do show that rPAI-1 23 is required in the reaction in order for plasmin cleavage to occur. The cleavage of plasmin in this reaction results in production of angiostatin kringles 1-3.
Others have shown that PAI-1 can be cleaved at the P1 and PЈ1 residues of the RCL, turning the inhibitor into a substrate with a molecular mass of 39kDa (42,73,74). The amount of substrate PAI-1 (RCL cleaved, inactive) is increased during the interaction of PAI-1 with thrombin in the presence of heparin and vitronectin (42). A smaller cleaved PAI-1 fragment (less than 31 kDa) has been shown to be produced as a result of adding heparin and thrombin to a pre-existing PAI-1⅐thrombin complex (42) or in reaction of PAI-1 with plasmin (43). The deletions at the carboxyl terminus exclude the RCL in three of the rPAI-1 isoforms that we have utilized in these studies. Each of these proteins functions differently with respect to proand anti-angiogenic mechanisms. Therefore, we can conclude that the functional activity of these proteins is not solely dependent upon the absence of the reactive center loop. The structural difference in rPAI-1 23 , rPAI-1 ⌬23 , and rPAI-1 Hep23 is the heparin (heparan sulfate)-binding domain, which seemingly accounts for the differences in functional activity of the three proteins. The large amounts of activated VEGF in the culture medium of rPAI-1 ⌬23 -treated cells strongly support the importance of partial PAI-1 binding to heparan sulfate with respect to controlling activation of VEGF.
Data from biochemical and in vitro studies demonstrate that PAI-1 is cleaved by thrombin, matrix metalloproteinase-3, and plasmin. However, there have not been studies to examine the presence of cleaved PAI-1 products in vivo. Identification of cleaved PAI-1 in correlation with tumor growth and atherosclerotic plaque development would provide valuable information with respect to the controversial pro-or anti-angiogenic role of PAI-1 in these two diseases.
Our studies provide evidence that PAI-1 conformational changes expose binding domains that participate in anti-angiogenic mechanisms. Such evidence supports the importance in evaluating the anti-angiogenic role of PAI-1 in tumor growth (76,77) and atherosclerotic plaques (50,75) and as a potential therapeutic molecule.