Annexin A2-S100A10 Heterotetramer, a Novel Substrate of Thioredoxin*

The binding of plasminogen activators and plasminogen to the cell surface results in the rapid generation of the serine protease plasmin. Plasmin is further degraded by an autoproteolytic reaction, resulting in the release of an angiostatin, A61 (Lys78–Lys468). Previously, we demonstrated that the annexin A2-S100A10 heterotetramer (AIIt) stimulates the release of A61 from plasmin by promoting the autoproteolytic cleavage of the Lys468–Gly469 bond and reduction of the plasmin Cys462–Cys541 disulfide (Kwon, M., Caplan, J. F., Filipenko, N. R., Choi, K. S., Fitzpatrick, S. L., Zhang, L., and Waisman, D. M. (2002) J. Biol. Chem. 277, 10903–10911). Mechanistically, it was unclear if AIIt promoted a conformational change in plasmin, resulting in contortion of the plasmin disulfide, or directly reduced the plasmin disulfide. In the present study, we show that AIIt thiols are oxidized during the reduction of plasmin disulfides, establishing that AIIt directly participates in the reduction reaction. Incubation of HT1080 cells with plasminogen resulted in the rapid loss of thiol-specific labeling of AIIt by 3-(N-maleimidopropionyl)biocytin. The plasminogen-dependent oxidation of AIIt could be attenuated by thioredoxin. Thioredoxin reductase catalyzed the transfer of electrons from NADPH to the oxidized thioredoxin, thus completing the flow of electrons from NADPH to AIIt. Therefore, we identify AIIt as a substrate of the thioredoxin system and propose a new model for the role of AIIt in the redox-dependent processing of plasminogen and generation of an angiostatin at the cell surface.

The plasmin(ogen) system consists of five major components (reviewed in Refs. [1][2][3][4]. First, plasminogen, a single chain inactive zymogen, consists of two functional regions, the kringle-containing domains and the catalytic site (5). The kringle domains are protein-protein interaction domains that interact with the C-terminal lysines of cell-surface plasminogen receptors. Second, the plasminogen activators, urokinase-type plasminogen activator (uPA) 1 and tissue plasminogen activator (tPA), are highly specific serine proteases that cleave plasminogen into the two-chain active enzyme plasmin. tPA mediates mainly intravascular thrombolysis, whereas uPA is involved in pericellular proteolysis during cell migration, wound healing, and tissue remodeling. Third, the inhibitors of plasminogen activators consist of four proteins, PAI-1, PAI-2, PAI-3, and nexin. These inhibitors form stoichiometric complexes with the plasminogen activators, resulting in the loss of plasminogen activator activity. Fourth, the uPA receptor is a 55-60-kDa glycoprotein that is anchored in the cell membrane by a glycosylphosphatidylinositol moiety (6 -9). Primarily, the uPA receptor is thought to enhance and direct uPA proteolytic activity (10 -13). Increased uPA activity has been shown to be correlated with tumor invasiveness and to be a prognostic indicator of disease recurrence and metastasis in multiple types of cancer (14,15). Fifth, the plasminogen receptors consist of a group of cell-surface protein and nonprotein molecules that bind plasminogen with low affinity (K d ϭ 0.3-2 M) and high capacity (10 4 to 10 7 binding sites/cell). Interestingly, although many plasminogen receptors have been identified, only a small subset of cellular plasminogen receptors, those that possess a C-terminal lysine residue and are co-localized with the plasminogen activator-plasminogen activator receptor complex, participate in cell-surface plasminogen activation (reviewed in Refs. 16 -18). Candidate plasminogen receptors possessing Cterminal lysines include S100A10 (19 -21), cytokeratin-8 (22)(23)(24)(25), TIP49a (26), integrin ␣ M ␤ 2 (27), and ␣-enolase (28 -31).
Once formed, cell-surface plasmin must be tightly regulated to avoid inappropriate extracellular proteolysis and cellular damage. For example, plasmin produced at the cell surface and released into the extracellular milieu is rapidly inactivated by the plasmin inhibitor ␣ 2 -antiplasmin (32)(33)(34). In addition, plasmin undergoes an autoproteolytic reaction, resulting in the cleavage and inactivation of the molecule (35). Interestingly, several fragments produced by the plasmin autoproteolytic reaction are biologically active and possess anti-angiogenic activity (36,37). Recent studies have reported the amino acid sequence of the primary plasmin fragment produced at the cell surface as a consequence of plasmin autoproteolysis (38). This protein, a recent addition to the angiostatin family of proteins, is referred to as A 61 and has a molecular mass of 61 kDa and the amino acid sequence Lys 78 -Lys 468 of plasminogen (reviewed in Ref. 39).
The annexin A2-S100A10 heterotetramer (AIIt) is a Ca 2ϩbinding protein that is composed of two annexin A2 and two S100A10 subunits (reviewed in Refs. 40 -45). AIIt is anchored to the plasma membrane as a result of the interaction of the annexin A2 subunit with the plasma membrane lipids. The S100A10 subunit protrudes away from the membrane into the extracellular medium (46). The S100A10 subunit functions as the plasminogen regulatory subunit of AIIt (19,(47)(48)(49)(50). The C-terminal lysine of this subunit participates in the stimulation of tPA-dependent plasminogen activation by AIIt (19,47,49). The S100A10 subunit binds tPA (K d ϭ 0.45 M), plasminogen (K d ϭ 1.81 M), and plasmin (K d ϭ 0.36 M) (48). Removal of the C-terminal lysines from the S100A10 subunit blocks both tPA and plasminogen binding to S100A10 (19,48,49). Although the C-terminal lysines of S100A10 form tPA-and plasminogen-binding sites, the location of the plasmin-binding site is unknown. In contrast to the S100A10 subunit, the annexin A2 subunit does not bind tPA or plasminogen, but binds plasmin (K d ϭ 0.78 M) (48).
The mechanism by which plasminogen is converted to angiostatin(s) is controversial (reviewed in Ref. 39). We proposed a novel mechanism in which AIIt regulates the conversion of plasminogen to an angiostatin, A 61 (51). Initially, we showed that AIIt stimulates the plasminogen activator-dependent conversion of plasminogen to plasmin (20,50). Subsequently, we reported that AIIt stimulates plasmin autoproteolysis in vitro (21,42,52). Most recently, we demonstrated that AIIt stimulates both the autoproteolytic cleavage of plasmin at the Lys 468 -Gly 469 bond and the reduction of the plasmin Cys 462 -Cys 541 disulfide (38,51). The reduction of the Cys 462 -Cys 541 disulfide, which allowed classification of AIIt as a plasmin reductase, is particularly important in the regulation of A 61 formation because this reaction is necessary to release A 61 from the proteolyzed plasmin molecule. However, it was unclear if the thiols of AIIt directly participated in the plasmin reduction reaction or if the conformational change induced in plasminogen by AIIt caused an increase in the dihedral strain energy of the plasmin disulfides and opportunistic attack of these disulfide(s) by hydroxyl ions. The latter mechanism has been suggested for the thiol-independent reductase activity of phosphoglycerate kinase (53).
In this study, we have further characterized the plasmin reductase activity of AIIt and show that AIIt thiols directly participate in the reduction of plasmin disulfides. We show that incubation of HT1080 fibrosarcoma cells with plasminogen results in the rapid oxidation of AIIt. The oxidation of AIIt is dependent on the generation of plasmin, as the formation of catalytically deficient plasmin at the cell surface fails to oxidize AIIt. We also show that oxidized AIIt is reduced by the thioredoxin system. This work is the first direct demonstration of the participation of AIIt in the redox-dependent processing of plasminogen at the cell surface and also identifies AIIt as a new substrate for the thioredoxin system.

EXPERIMENTAL PROCEDURES
Materials-HT1080, DU145, and PC-3 cells were obtained from American Type Culture Collection. Transfected HT1080 cells were prepared as described (54). Glu-plasminogen and plasmin were purchased from American Diagnostica Inc. The catalytically deficient plasminogen (Pg cd ) mutant S741C was purified as described (55). Two-chain uPA was a generous gift from Dr. H. Stack (Abbott). Human recombinant thioredoxin-1 and the catalytically deficient thioredoxin-1 (Trx cd ) mutant C32S were purified from Escherichia coli as described previously (56). Rat thioredoxin reductase-1 was purified from rat liver as described previously (57). AIIt was purified from bovine lung as described (58). AIIt was subjected to limited plasmin digestion as described previously (19). Phosphoglycerate kinase was a generous gift from Dr. P. J. Hogg (Center for Thrombosis and Vascular Research, University of New South Wales, Sydney, Australia). NADPH, NADP ؉ , horseradish perox-idase-conjugated streptavidin, and protein-disulfide isomerase were purchased from Calbiochem. 3-(N-Maleimidopropionyl)biocytin (MPB) was purchased from Molecular Probes, Inc. Reduced glutathione, oxidized glutathione, L-cysteine, and streptavidin-agarose were purchased from Sigma. L-Lysine-Sepharose was purchased from Amersham Biosciences. Anti-annexin A2, anti-annexin A2 light chain, and anti-thioredoxin monoclonal antibodies were purchased from BD Biosciences. Anti-thioredoxin reductase polyclonal antibody was purchased from Upstate Biotechnology, Inc. Horseradish peroxidase-conjugated antimouse secondary antibody was purchased from Santa Cruz Biotechnology, Inc.
Dialysis of Proteins Used in the Reaction-After purification or reconstitution, AIIt, thioredoxin, protein-disulfide isomerase, and phosphoglycerate kinase were dialyzed against 20 mM Tris-HCl (pH 7.5) and 140 mM NaCl under argon gas to prevent possible oxidation.
Detection of Free Thiols in Extracellular Annexin A2 and S100A10 -Cultured HT1080, DU145, and PC-3 cell monolayers were washed five times with Dulbecco's modified phosphate-buffered saline (DPBS) containing 0.1 g/liter CaCl 2 , 0.2 g/liter KCl, 0.2 g/liter KH 2 PO 4 , 0.1 g/liter MgCl 2 , 8 g/liter NaCl, 2.16 g/liter Na 2 HPO 4 , 1 g/liter D-glucose, and 0.33 mM g/liter sodium pyruvate and scraped. All DPBS solutions used for this procedure were dialyzed under argon gas, and cells were incubated with rotating. The scraped cells were incubated with DPBS, 1 M plasminogen in DPBS, or 1 M Pg cd mutant in DPBS for the indicated times. Then, 100 M MPB was added to each incubation mixture, incubated at room temperature for 30 min, and then incubated with 200 M reduced glutathione at room temperature for 30 min. Cells from each incubation were washed three times with DPBS; lysed in phosphate-buffered saline (PBS) containing 1% Nonidet P-40, 1 mM Pefabloc, 1 g/ml aprotinin, 1 g/ml leupeptin, and 1 mM benzamidine on ice for 10 min; and centrifuged. The protein concentration of the supernatant was measured, and 200 g of each supernatant was incubated with 50 l of streptavidin-agarose beads at 4°C for 1.5 h. The agarose beads were washed three times with PBS containing 0.1% Tween 20 (TPBS). The bound proteins were eluted by boiling in reducing SDS-PAGE sample buffer and subjected to Western blotting using anti-annexin A2 and anti-annexin A2 light chain antibodies as described below.
In some experiments, the scraped cells were incubated with DPBS or 1 M plasminogen at 37°C for 30 min. Each incubation mixture was then divided into five portions, and each portion was incubated with DPBS, 1 M thioredoxin, 5 M thioredoxin, 1 M Trx cd mutant, or 5 M Trx cd mutant at 37°C for 1 h. Then, 100 M MPB was added to each incubation mixture, and the same procedures as described above were followed. Detection of Extracellular Annexin A2 and S100A10 -Cultured cell monolayers were surface-biotinylated by incubation with PBS containing 0.5 mg/ml sulfosuccinimidobiotin 2-(biotinamido)ethyl-1,3-dithiopropionate (sulfo-NHS-biotin) at room temperature for the indicated times. The monolayers were washed five times with PBS, scraped, and centrifuged at 300 ϫ g for 5 min. The cell pellet was incubated with PBS containing 1% Nonidet P-40, 1 mM Pefabloc, 1 g/ml aprotinin, 1 g/ml leupeptin, and 1 mM benzamidine on ice for 10 min. The mixture was then centrifuged at 1800 ϫ g for 10 min, and the protein concentration of the supernatant was measured. 100 g of each supernatant was incubated with 50 l of streptavidin-agarose beads at 4°C for 2 h. The agarose beads were washed three times with TPBS. The bound proteins were eluted by boiling in reducing SDS-PAGE sample buffer and subjected to Western blotting using anti-annexin A2 and anti-annexin A2 light chain antibodies as described below.
Plasmin Reductase Assay-Plasminogen (4 M) was incubated with uPA (0.075 M) and various combinations of AIIt, thioredoxin, thioredoxin reductase, NADPH, HT1080-conditioned medium, glutathione, or cysteine in buffer containing 20 mM Tris (pH 7.5) and 140 mM NaCl at 37°C for the indicated times. To label any free thiol groups of produced protein(s), the reaction mixture was incubated with 100 M MPB at room temperature for 30 min. Then, the reaction mixture was diluted with SDS-PAGE sample buffer and subjected to reducing SDS-PAGE, followed by Western blotting with horseradish peroxidase-conjugated streptavidin as indicated below.
In some experiments, the MPB-reacted mixture was incubated with L-lysine-Sepharose at room temperature for 30 min to purify A 61 . The matrix was extensively washed with PBS, and the bound proteins were eluted by boiling the resin with SDS-PAGE sample buffer. Each sample was subjected to nonreducing SDS-PAGE, followed by Western blotting with horseradish peroxidase-conjugated streptavidin as described below.
Preparation of HT1080-conditioned Medium and Detection of Thioredoxin and Thioredoxin Reductase in the Conditioned Medium-HT1080 cells were cultured to 70 -80% confluency. The monolayer was washed three times with PBS, replaced with cystine-free DMEM, and incubated overnight in an CO 2 incubator. The conditioned medium was collected, filtered, and frozen until used. In certain occasions, the conditioned medium was boiled at 65°C for 20 min and centrifuged, and the supernatant was used for the plasmin reductase assay. To detect thioredoxin and thioredoxin reductase in the conditioned medium, the indicated volume of boiled conditioned medium was subjected to SDS-PAGE, followed by Western blotting with antithioredoxin monoclonal antibody or anti-thioredoxin reductase polyclonal antibody as described below.
Electrophoresis and Western Blotting-Samples were diluted with SDS-PAGE sample buffer, subjected to SDS-PAGE, and electrophoretically transferred to nitrocellulose membrane (0.45-m pore size) at 4°C for 1 h. The membrane was blocked in TPBS with 5% skim milk at room temperature for 1 h and incubated overnight at 4°C with 0.13 g/ml anti-annexin A2 light chain monoclonal antibody, 0.13 g/ml antiannexin A2 monoclonal antibody, 0.5 g/ml anti-human thioredoxin monoclonal antibody, or 1.5 g/ml anti-rat thioredoxin reductase-1 polyclonal antibody in TPBS with 5% skim milk. The blot was extensively washed with TPBS and then incubated at room temperature for 1 h with 0.16 g/ml horseradish peroxidase-conjugated goat anti-mouse or anti-rabbit secondary antibody in TPBS with 5% skim milk. In the case of MPB-reacted protein samples, the membrane was blocked and incubated at room temperature for 1 h with 0.1 g/ml horseradish peroxidase-conjugated streptavidin in TPBS with 5% skim milk. The membrane was extensively washed with TPBS and visualized by enhanced chemiluminescence.
Thioredoxin Reductase Assay-NADPH oxidation coupled to thioredoxin reduction was monitored at 30°C as a decrease in absorbance at 340 nm using a Hewlett-Packard Model 8453 UV-visible spectrophotometer equipped with a thermostable cell holder and a multicell transport. The assay was carried out in a 200-l reaction mixture containing 50 mM Tris-HCl (pH 7.5), 100 mM NaCl, 200 M NADPH, 50 nM thioredoxin reductase, 3 M wild-type thioredoxin or Trx cd mutant, 3.5 M AIIt, 0.4 M uPA, and 20 M plasminogen.

AIIt Is Oxidized at the Cell Surface-Previous studies have
shown that incubation of cells with plasminogen results in the formation of the autoproteolytic plasminogen fragment A 61 by a three-step reaction (38,51). These reactions involve the uPAdependent conversion of plasminogen to plasmin, followed by the autoproteolytic cleavage of the Lys 468 -Gly 469 bond of plasmin (38). The Cys 462 -Cys 541 disulfide of the autoproteolyzed plasmin is then cleaved, resulting in the release of A 61 (Lys 78 -Lys 468 ). AIIt stimulates all three reactions, and the intrinsic plasmin reductase activity of AIIt is responsible for the reduction of the plasmin Cys 462 -Cys 541 disulfide (51).
To investigate the redox status of AIIt during this reaction, we incubated HT1080 cells with plasminogen and then labeled the cell-surface proteins with the substantially membrane-impermeable, free sulfhydryl-reactive reagent MPB (59). The reaction of free sulfhydryl-containing proteins with MPB results in the biotinylation of extracellular proteins that contain free thiols. The biotinylated protein fraction was collected by streptavidin-agarose pull-down and subjected to Western blot analysis with antibodies against the annexin A2 and S100A10 subunits of AIIt. Therefore, only AIIt that is labeled with MPB will be detected on the Western blot, and the loss of MPB labeling of AIIt will correspond to the loss of AIIt on the Western blot. As shown in Fig. 1A, in the absence of plasminogen, both subunits of cell-surface AIIt were labeled by MPB. In contrast, incubation of HT1080 cells with plasminogen resulted in a rapid loss of MPB labeling of the subunits of AIIt. Because MPB labels only thiols, but not disulfides or oxidized FIG. 1. Plasminogen-dependent oxidation of AIIt at the surface of HT1080 cells. A, HT1080 cells were incubated with 1 M plasminogen for the indicated times, labeled with MPB, washed, and lysed. The cell lysate including MPB-labeled cell-surface proteins was then immunoprecipitated with streptavidin-agarose, and the bound proteins were subjected to SDS-PAGE, followed by Western blotting using anti-annexin A2 (AnxA2) antibody and anti-annexin A2 light chain antibody. B, shown is the detection of extracellular annexin A2 and S100A10. HT1080 cell monolayers were surface-biotinylated with sulfo-NHS-biotin at room temperature for the indicated times. The monolayers were washed, scraped, and centrifuged. The cell pellet was then lysed, and the supernatant was immunoprecipitated with streptavidin-agarose. The other experimental procedures were the same as described for A. AIIt (first lane) and plasmin-treated AIIt (second lane) standards (Std.) are also shown. C, HT1080 cells were incubated with DPBS, 1 M plasminogen (Pg), or 1 M Pg cd mutant for the indicated times and then labeled with MPB. The other experimental procedures were the same as described for A. The AIIt standard is also shown (first lane). D, HT1080 cells were incubated with DPBS or 1 M plasminogen at 37°C for 30 min. Each incubation mixture was then divided into five portions, and each portion was incubated with DPBS, 1 or 5 M thioredoxin (Trx), or 1 or 5 M Trx cd mutant at 37°C for 1 h. Then, each reaction was incubated with MPB, and the other experimental procedures were as described for A. thiols, this observation suggests that AIIt is oxidized during plasminogen activation. However, whether or not the oxidation of AIIt results in the formation of disulfides or sulfenic or sulfinic acid derivatives of the thiols cannot be discriminated by this assay.
Our previous studies (38,51,54) have shown that plasminogen is rapidly converted to plasmin by HT1080 cells. To rule out the possibility that the loss of MPB labeling is due to the proteolysis of AIIt by plasmin generated from plasminogen, the cells were incubated with plasminogen for various times, followed by cell-surface labeling with sulfo-NHS-biotin. This reagent labels the primary amine groups of cell-surface proteins with biotin and is therefore used as a general protein-labeling reagent. The biotinylated cell-surface proteins were collected with streptavidin-agarose and analyzed by Western blotting with antibodies against both annexin A2 and S100A10. As shown in Fig. 1B, significant plasminogen-dependent proteolysis of cell-surface AIIt was not observed even after 24 h of incubation of cells with plasminogen. As a control, we included a lane containing AIIt that had been digested with plasmin. It was shown previously that annexin A2 can be cleaved by plasmin in vitro between Lys 27 and Ala 28 (18). The loss of these 27 amino acids from annexin A2 was easily visualized by SDS-PAGE (Fig. 1B, second lane). Thus, plasmin-dependent proteolysis of AIIt at the cell surface did not occur under these experimental conditions.
The plasminogen-dependent oxidation of AIIt at the cell surface could have been caused by an enhanced susceptibility of AIIt to oxidation due to a conformational change in the molecule as a result of its interaction with plasminogen or plasmin. We tested this possibility by incubation of HT1080 cells with recombinant mutant plasminogen in which the active-site serine (Ser 741 ) was replaced with cysteine (Pg cd ). The activation of this mutant plasminogen results in the generation of plasmin, which is catalytically deficient and possesses only 0.0125% of the activity of the wild-type molecule (55). We observed that incubation of HT1080 cells with Pg cd did not cause the oxidation of AIIt (Fig. 1C). Thus, our results suggest that the conversion of plasminogen to catalytically active plasmin occurs prior to oxidization of AIIt.
The two principal systems that maintain the cellular thiol/ disulfide redox state are the GSH and thioredoxin systems (60). In the extracellular environment, GSH is thought to serve primarily not as an antioxidant, but as a substrate for glutamyltransferase, the first enzyme in the pathway to provide cysteine from GSH degradation. In contrast, thioredoxin contains a dithiol motif at its active site, which is ideally suited for reduction of protein disulfides, sulfoxides, and sulfenic acids. We therefore examined the possibility that thioredoxin might be involved in maintaining the redox status of AIIt. As shown in Fig. 1D, the addition of thioredoxin to HT1080 cells did not affect the MPB labeling of AIIt. However, the addition of thioredoxinafterplasminogentreatmentreversedtheplasminogendependent oxidation of AIIt in a dose-dependent manner. Notably, the Trx cd mutant (with the active-site cysteine (Cys 32 ) replaced with serine) could not reverse the plasminogen-dependent oxidation of AIIt, suggesting that the redox activity of thioredoxin, but not the binding followed by the subsequent conformational changes, plays a role in mediating this activity. These results confirm that incubation of HT1080 cells with plasminogen results in the oxidation of AIIt thiols and also identify thioredoxin as a potential regulator of the redox status of AIIt.
Role of the S100A10 Subunit of AIIt in the Plasminogen-dependent Oxidation of AIIt-Previous work from our laboratory established that S100A10 is a key cellular plasminogen regulatory protein and that the loss of S100A10 from the cell surface causes a dramatic decrease in cellular plasmin generation (54,61). To examine the consequence of over-or underexpression of the S100A10 subunit in the plasminogen-dependent oxidation of AIIt, we used our well characterized HT1080 cell lines (54). The cells were stably transfected with a vector encoding a sense or antisense S100A10 construct. The transfected HT1080 cells were incubated in the presence or absence of plasminogen and labeled with MPB, and the MPB-labeled extracellular proteins were collected by pull-down with streptavidin-agarose. Western blot analysis of MPB-labeled AIIt showed that the overexpression of extracellular S100A10 did not affect the levels of MPB-labeled annexin A2 or S100A10 (Fig. 2). In addition, the antisense-mediated loss of S100A10 from the cell surface did not affect the labeling of annexin A2 by MPB in the absence of plasminogen (Fig. 2). However, the antisense-mediated underexpression of cell-surface S100A10 resulted in increased labeling of the annexin A2 subunit in the presence of plasminogen. Thus, the presence of S100A10 appears to be required for the plasminogen-dependent oxidation of AIIt. Because the underexpression of S100A10 from the surface of HT1080 cells also results in an 80 -90% loss of the plasmin-generating capacity of HT1080 cells (54), it is reasonable to propose that S100A10 is required to produce the plasmin that serves as the substrate for the AIIt oxidation reaction.
We also examined the universality of the plasminogen-dependent oxidation of AIIt. As shown in Fig. 3, both DU145 and PC-3 prostate carcinoma cells also demonstrated rapid oxidation of AIIt upon the addition of plasminogen. In contrast, the addition of the Pg cd mutant to these cells failed to result in the oxidation of AIIt. Therefore, similar to the results observed with HT1080 cells, the plasminogen-dependent oxidation of AIIt at the surface of prostate cancer cells requires catalytically active plasmin.

FIG. 2.
Dependence of the plasminogen-dependent oxidation of AIIt on the S100A10 subunit. HT1080 cells were stably transfected with a vector encoding a sense or antisense S100A10 construct as described previously (54). Normal HT1080 cells and each clone were incubated with DPBS, 1 M plasminogen (Pg), or 1 M Pg cd mutant for the indicated times and then labeled with MPB. The other experimental procedures were the same as described in the legend to Fig.  1A. The AIIt standard (Std.) is also shown. AnxA2, annexin A2.

In Vitro Studies of the Plasminogen-dependent Oxidation of
AIIt and Its Subunits-We utilized a cell-free assay to more fully characterize the redox status of AIIt during plasmin generation. For these experiments, we incubated various combinations of uPA, plasminogen, and AIIt; and after incubation for various times, the reactants were labeled with MPB, subjected to SDS-PAGE, and analyzed by Western blotting using horseradish peroxidase-conjugated streptavidin. As shown in Fig.  4A, both subunits of AIIt were labeled by MPB in vitro, and the MPB labeling of AIIt was unaffected when AIIt was incubated with either uPA or plasminogen alone. However, the incubation of AIIt with uPA and plasminogen resulted in the rapid loss of MPB labeling of AIIt, suggesting that AIIt is oxidized during plasmin generation. In contrast, the incubation of AIIt with uPA and the Pg cd mutant did not affect the MPB labeling of AIIt. Thus, as observed in the experiments performed with the cells, the in vitro data also suggest that the generation of catalytically active plasmin is required for the oxidation of AIIt.
The loss of MPB labeling could have been due to the proteolysis of AIIt by the plasmin generated by the reaction or to the oxidation of AIIt. To discriminate between these possibilities, AIIt was incubated with uPA and plasminogen, and the reactants were subjected to SDS-PAGE, followed by Coomassie Blue staining. We observed that AIIt staining was lost when AIIt was incubated with uPA and plasminogen (data not shown). However, it was unclear if the proteolysis of AIIt was a result of the oxidation of AIIt. To discriminate between these possibilities, AIIt was incubated with uPA and plasminogen in the presence or absence of the thioredoxin system. The reactants were then labeled with MPB, and the MPB-labeled proteins were collected with streptavidin-agarose and subjected to SDS-PAGE, followed by Western blotting with antibodies against annexin A2 and S100A10. When AIIt was incubated in the presence of uPA, plasminogen, and the thioredoxin system (thioredoxin, thioredoxin reductase, and NADPH), only slight proteolysis of AIIt was observed (Fig. 4B). Interestingly, the proteolysis of AIIt was not inhibited by the addition of either thioredoxin or thioredoxin reductase alone. This suggests that the presence of the complete thioredoxin system is required to prevent the proteolysis of AIIt. Because the thioredoxin system functions as a thiol redox control system to maintain proteins in their reduced state, the ability of this redox system to block the proteolysis of AIIt suggests that the oxidation of AIIt occurs prior to the proteolysis of AIIt by plasmin and that the oxidation of AIIt triggers its proteolysis in vitro.
The Conditioned Medium Contains AIIt Reductase Activity-The major extracellular redox systems are the glutaredoxin and thioredoxin systems (60). To examine the possibility that extracellular redox systems might be involved in the redox regulation of AIIt, we incubated the conditioned medium from HT1080 cells with uPA, plasminogen, and AIIt. As shown in Fig. 5A, the conditioned medium partially blocked the plasminogen-dependent oxidation of AIIt. Interestingly, the AIIt reductase activity of the conditioned medium was stable to boiling. The thioredoxin system was initially described as an endogenous, heat-stable, glucocorticoid receptor-activating factor (62,63) and later shown to be secreted by cultured cells (64). We therefore examined the conditioned medium for the presence of this redox system. As shown in Fig. 5 (B and C), Western blot analysis confirmed the presence of both thioredoxin and thioredoxin reductase in the boiled conditioned medium.
In addition to the thioredoxin system, HT1080 cells have been shown to secrete other reductases and reducing equivalents into the extracellular environment. These include glutathione, cysteine, protein-disulfide isomerase, and phosphoglycerate kinase (59, 60, 65). We therefore tested these reductases as potential AIIt reductases. As shown in Fig. 6, only the mM Tris (pH 7.5) and 140 mM NaCl at 37°C for 30 min. The reaction mixture was then incubated with MPB, and the MPB-labeled proteins were collected with streptavidin-agarose and subjected to reducing SDS-PAGE, followed by Western blotting with antibodies against annexin A2 (AnxA2) and S100A10. The AIIt standard (Std.) is also shown.
thioredoxin system was capable of preventing the plasminogendependent oxidation of AIIt. It was interesting to note that thioredoxin alone was capable of partially preventing the oxidation of AIIt, although partial proteolysis of AIIt did occur.

Regulation of the Plasmin Reductase Activity of AIIt by the
Thioredoxin System-We showed originally that the binding of AIIt to plasmin results in the reduction of the plasmin Cys 462 -Cys 541 disulfide and the release of A 61 from plasmin (51). Our current observations that AIIt was oxidized during this reaction (Fig. 1) and that the oxidation of AIIt could be reversed by the thioredoxin system (Fig. 4B) presented the possibility that AIIt might participate in cycles of oxidation/reduction. Because AIIt stimulates the release of an angiostatin (A 61 ), we tested the possibility that the thioredoxin system might stimulate the AIIt-dependent formation of A 61 by reversibly reducing AIIt that is oxidized during plasmin reduction. As shown in Fig. 7, the AIIt-dependent release of A 61 was stimulated by the thioredoxin system. In contrast, under our assay conditions, the thioredoxin system did not activate the release of A 61 in the absence of AIIt. Interestingly, although the amount of A 61 produced in the presence of AIIt and 1 M thioredoxin was only slightly enhanced compared with the amount released in the absence of thioredoxin, the addition of thioredoxin reductase and NADPH potentiated the effect of thioredoxin to an amount similar to that observed upon the inclusion of 5 M thioredoxin (Fig. 7). This suggests that thioredoxin reductase and NADPH reduce the oxidized thioredoxin and by this mechanism potentiate thioredoxin activity.
Flow of Reducing Equivalents from the Thioredoxin System to Plasmin through AIIt-Our data support the concept that AIIt can be maintained in a reduced state by the thioredoxin system. Thus, we envisioned the redox cascade in which thioredoxin reductase utilizes NADPH to reduce thioredoxin that has become oxidized subsequent to its reduction of AIIt. Therefore, thioredoxin might serve as an AIIt reductase, whereas AIIt serves as a plasmin reductase. We directly tested this possibility by measuring the oxidation of NADPH during the plasmin reductase reaction. Because the oxidation of NADPH to NADP ϩ results in loss of absorbance at 340 nm, it can be spectrophotometrically monitored. As shown in Fig. 8, we observed that the addition of AIIt and the thioredoxin system to uPA and plasminogen resulted in an initial rapid decrease in absorbance at 340 nm, suggesting that NADPH is oxidized by the reaction. In contrast, NADPH was not oxidized when the Trx cd mutant was substituted for thioredoxin. Similarly, NADPH was not oxidized when plasminogen or thioredoxin was omitted from the reaction (Fig. 8) or when the Pg cd mutant was substituted for plasminogen (data not shown). The rate of NADPH oxidation was considerably decreased and 140 mM NaCl at 37°C for 2 h. The reaction mixture was then incubated with MPB, and the generated A 61 was collected with L-lysine-Sepharose at room temperature for 30 min. The matrix was extensively washed, and the bound proteins were subjected to nonreducing SDS-PAGE, followed by Western blotting with horseradish peroxidase-conjugated streptavidin.
when AIIt was omitted from the reaction. Therefore, these results suggest that AIIt participates in a redox cascade involving the flow of electrons from the thioredoxin system to plasmin. Fig. 9 illustrates pictorially the flow of reducing equivalents from the thioredoxin system to plasmin through AIIt for the generation of A 61 . DISCUSSION Folkman (66 -68) originally postulated that tumor growth and metastasis are critically dependent on the development of an adequate blood supply. The growth of new blood vessels, a process called angiogenesis, is mediated by signaling molecules released from both tumor and host cells. Folkman demonstrated that a plasminogen fragment accumulates in the circulation during the growth of a Lewis lung carcinoma and disappears when the tumor is resected. The molecule, named angiostatin, was shown to be produced by the primary tumor and to inhibit the neovascularization and growth of its remote metastases by blocking the tumor-stimulated growth of microvascular endothelial cells (69). Although Ͼ10 years have passed since this seminal discovery, the exact structure of angiostatin(s), its mechanism of formation, and its mode of action remain controversial (reviewed in Ref. 39).
Angiostatin formation has been postulated either to involve the direct cleavage of plasminogen by several distinct proteases such as metalloelastase, gelatinase B (matrix metalloproteinase-9), stromelysin-1 (matrix metalloproteinase-3), matrilysin (matrix metalloproteinase-7), cathepsin D, and prostate-specific antigen or, alternatively, to involve a three-step mechanism entailing the conversion of plasminogen to plasmin by uPA, the autoproteolytic cleavage of plasmin, and the release of the resultant plasmin fragment by cleavage of disulfide bonds (reviewed in Ref. 39). The cleavage of the plasmin disulfide bonds is accomplished by free sulfhydryl group donors such as N-acetylcysteine and glutathione or by hydroxyl ions at alkaline pH. Alternatively, the plasmin disulfide bond cleavage by hydroxyl ions can be facilitated at neutral pH by a plasmin reductase such as phosphoglycerate kinase (53) and by cellsurface actin (70). Our laboratory has proposed that AIIt is a key plasminogen receptor that regulates the formation of the angiostatin A 61 at the surface of many tumor cells (reviewed in Refs. 18 and 39).
This investigation was aimed at identifying the exact mechanism by which AIIt reduces the plasmin disulfides and stimulates the release of A 61 . Our demonstration that the cellular processing of plasminogen results in the oxidation of AIIt at the cell surface is important because it establishes that AIIt participates in three plasminogen-dependent reactions, viz. the conversion of plasminogen to plasmin, plasmin autoproteolysis, and the cleavage of plasmin disulfide bonds. Thus, these experiments have established the exact mechanism by which AIIt regulates cell-surface plasminogen and stimulates A 61 production. Because the thiols of AIIt are oxidized during A 61 production, we can rule out the possibility that AIIt induces a conformational change in plasminogen, resulting in an increase in the plasminogen disulfide dihedral strain energy, which then increases its susceptibility to be attacked by hydroxyl ions. A similar mechanism has been suggested for the plasmin reductase activity of phosphoglycerate kinase (53). We can now postulate that the plasmin reductase activity of AIIt is a consequence of the reduction of plasmin disulfides by the thiols of AIIt. Because we observed that catalytically deficient plasmin is not reduced by AIIt, we can also conclude that autoproteolysis of plasmin occurs before reduction.
Hogg and co-workers (36,53,71) have proposed a novel mechanism for the cellular production of angiostatin in which phosphoglycerate kinase, secreted into the extracellular milieu, binds plasminogen and induces a conformational change in plasmin that facilitates the attack on the Cys 512 -Cys 536 disulfide bond by hydroxyl ions, resulting in the formation of a sulfenic acid at Cys 512 and a free thiol at Cys 536 . The Cys 536 thiol was then postulated to exchange with the Cys 462 -Cys 541 disulfide bond, resulting in the formation of a new disulfide bond between Cys 536 and Cys 541 and a free thiol at Cys 462 . The final step in the reaction involves cleavage of the Arg 530 -Lys 531 bond by an unidentified serine protease(s). Soff and co-workers (70) proposed a distinct mechanism for the cellular production  9. Model of the mechanism of AIIt-stimulated A 61 production from plasminogen. The C-terminal lysine residues of the S100A10 subunit of AIIt bind plasminogen (Pm), resulting in the co-localization of the plasminogen⅐AIIt complex with the uPA⅐uPA receptor complex. uPA catalyzes the conversion of plasminogen to plasmin, resulting in the formation of the plasmin⅐AIIt complex. AIIt then stimulates the autoproteolysis of the plasmin Lys 468 -Gly 469 bond, followed by the reduction of the plasmin Cys 462 -Cys 541 disulfide by AIIt thiols. AIIt oxidized by this reaction is reduced by reducing equivalents that flow through the thioredoxin (Trx) system. TrxR, thioredoxin reductase-1.
of an angiostatin involving cell-surface ␤-actin. They proposed that ␤-actin mediates the binding of plasmin to the cell surface and its autoproteolysis to an angiostatin. They also reported that antibodies to actin reduce membrane-dependent generation of their angiostatin by 70% and that the addition of actin to in vitro generated plasmin results in stoichiometric conversion to their angiostatin in a cell-free system. In contrast, our model of angiostatin formation involves thiol-dependent reduction of autoproteolyzed plasmin by AIIt.
The loss of MPB labeling of cell-surface AIIt, observed upon addition of plasminogen to cells, could be due to a decrease in the accessibility of the AIIt thiols to MPB or to oxidation of these thiols, resulting in either the formation of disulfides or the conversion of these thiols to sulfenic acid. However, the key observation that inactive plasmin does not promote the oxidation of AIIt suggests that a conformational change due to the binding of plasminogen or plasmin to AIIt is an unlikely cause of AIIt oxidation. Furthermore, the demonstration that thioredoxin reverses the plasminogen-mediated loss of MPB labeling of AIIt is further evidence that the plasminogen-dependent oxidation of AIIt is due to the interaction of AIIt thiols with the disulfides of plasmin because plasminogen and plasmin possess only disulfides. The observation that thioredoxin can prevent plasminogen-dependent oxidation of AIIt is important because it shows that AIIt is not irreversibly oxidized during plasmin reduction. However, it is still unclear if the plasminogen-dependent oxidation of AIIt results in the formation of disulfides or the conversion of AIIt thiols to sulfenic acid.
It was also interesting that, among the reductants examined, only thioredoxin could prevent AIIt oxidation, whereas other reductants, including protein-disulfide isomerase, glutathione, cysteine, and phosphoglycerate kinase, could not. Thioredoxin is a small redox protein of 12 kDa and has a single active-site sequence, Trp-Cys-Gly-Pro-Cys (reviewed in Ref. 72). Proteindisulfide isomerase has two thioredoxin-like domains containing a distinct active-site sequence, Trp-Cys-Gly-His-Cys, and catalyzes the formation, rearrangement, or reduction of protein disulfide bonds (reviewed in Ref. 73). Both proteins have been shown to be secreted by cells (64,74,75). In view of their redox potentials (thioredoxin, Ϫ270 mV; and protein-disulfide isomerase, Ϫ150 mV), thioredoxin and protein-disulfide isomerase have been considered to serve primarily as a reductant and an oxidant, respectively. These might be the reason why only thioredoxin could reduce oxidized AIIt.
Thioredoxin reductase is a flavoenzyme that catalyzes the reduction of thioredoxin using NADPH as a cofactor (reviewed in Refs. 76 and 77). Thioredoxin reductase is a heat-stable enzyme that can be secreted from cells (78). Thus, the thioredoxin/thioredoxin reductase system is present in the extracellular milieu. Our finding that thioredoxin reductase potentiates the reversal of the plasminogen-dependent oxidation of AIIt is not surprising. Thioredoxin regeneration depends on the proper reduction by thioredoxin reductase and NADPH; thus, the two proteins act in synergy. It is therefore reasonable to suspect that thioredoxin reductase promotes the AIIt reductase activity of thioredoxin because thioredoxin reductase is available in extracellular fluids for the maintenance of reduced and active thioredoxin.
In summary, our laboratory has demonstrated that AIIt stimulates the tPA-or uPA-dependent conversion of plasminogen to plasmin and that the C-terminal lysine residues of the S100A10 subunit mediate this function in vitro (19,20,50). We also showed that AIIt co-localizes with the uPA⅐uPA receptor complex and that the loss of S100A10 from the cell surface of fibrosarcoma and colorectal cells results in a dramatic loss of plasmin generation, thus establishing AIIt as a major regulator of cellular plasmin production (54,61). Plasmin activity was also shown to be regulated by AIIt. Specifically, we showed that AIIt stimulates plasmin autoproteolysis by an intramolecular mechanism (21). We utilized N-and C-terminal sequencing and mass spectrometry to characterize the major plasmin fragment produced by plasmin autoproteolysis and identified the Lys 78 -Lys 468 fragment, which we subsequently named A 61 (38). A 61 was shown to be as potent as other angiostatins (38). We also showed that AIIt not only stimulates plasmin autoproteolysis by facilitating cleavage of the plasmin Lys 468 -Gly 469 bond, but also facilitates the reduction of the plasmin Cys 462 -Cys 541 disulfide in vitro (38,51). We have shown here that the reduction of the plasmin Cys 462 -Cys 541 disulfide results in the oxidation of AIIt, thus supporting the concept that the thiols of AIIt directly participate in the reduction of plasmin disulfides and the generation of A 61 . Our observation that the addition of plasminogen to cells results in the oxidation of AIIt at the cell surface confirms that AIIt is an important cellular regulator not only of plasmin production, but also of plasmin-derived A 61 generation at the cell surface. Finally, we have shown that the plasminogen-dependent oxidation of AIIt is not irreversible and that oxidized AIIt can be reduced by the thioredoxin/ thioredoxin reductase system. Future studies will be required to establish which thiols of AIIt participate in this oxidation and reduction reaction.