PLASMIN REDUCTION BY PHOSPHOGLYCERATE KINASE IS A THIOL-INDEPENDENT PROCESS

disulfide bonds by PGK has been explored in this study. We show that none of the PGK Cys residues are directly involved in plasmin reduction but that alkylation/oxidation of the fast-reacting Cys or conformational changes in the same region of the protein inhibit reductase activity.


Introduction
Disulfide bonds of certain cell surface proteins can interchange between the oxidized and reduced state (1)(2)(3). These observations suggest that the function of some secreted proteins may be controlled by interchange of one or more disulfide bonds (1). The reduction of disulfide bonds in plasmin by a tumor cell-derived protein was the first example of disulfide exchange in a secreted soluble protein (4)(5)(6). A second example is disulfide exchange in von Willebrand Factor, which is facilitated by thrombospondin-1 (7). Plasmin reduction is the first step in formation of the tumor angiogenesis inhibitor, angiostatin.
Tumor expansion and metastasis is dependent on tumor neovascularization, or angiogenesis (8). Angiogenesis is balanced by several protein activators and inhibitors (9).
One such inhibitor is angiostatin (10), which is an internal fragment of the plasma zymogen, plasminogen. Plasminogen contains five consecutive kringle domains followed by a serine proteinase module. Urokinase-or tissue-plasminogen activator convert plasminogen to plasmin by hydrolysis of a single peptide bond in the serine proteinase module. Plasmin is processed in the conditioned medium of tumor cells producing angiostatin fragments consisting of kringle domains 1-4½, 1-4 and 1-3 (6 and references therein). 14). All three kringle-containing fragments have been shown to inhibit endothelial cell proliferation in vitro and angiogenesis in vivo (15 and references therein).
The plasmin reductase was recently purified from human fibrosarcoma cell conditioned medium and shown to be the glycolytic enzyme, phosphoglycerate kinase (PGK; ATP:3phospho-D-glycerate 1-phosphotransferase, EC 2.7.2.3) (6). Plasma of mice bearing fibrosarcoma tumors contained several-fold more PGK than mice without tumors and administration of PGK to tumor-bearing mice caused an increase in plasma levels of angiostatin and decrease in tumor vascularity and rate of tumor growth. Solid tumors employ PGK and other glycolytic enzymes to facilitate anaerobic production of ATP. These findings indicate that PGK plays an additional role in tumorigenesis by initiating extracellular formation of angiostatin from plasmin.
PGK is the sixth enzyme of the glycolytic pathway where it catalyses the high-energy phosphoryl transfer reaction from the acid anhydride bond of 1,3-bisphosphoglycerate to the β-phosphate of MgADP. PGK also influences DNA replication and repair in mammalian cell nuclei (16,17), stimulates viral mRNA synthesis in the cytosol (18) and extends through the cell wall of C. albicans (19). The human enzyme has a molecular mass of ~45 kDa and consists of a single polypeptide chain of 418 residues. Crystallographic studies of the yeast 6 disulfide bonds by PGK has been explored in this study. We show that none of the PGK Cys residues are directly involved in plasmin reduction but that alkylation/oxidation of the fastreacting Cys or conformational changes in the same region of the protein inhibit reductase activity.

Consequence of mutation of all seven PGK Cys residues to Ala for plasmin reductase activity
The seven Cys in wt PGK were mutated to Ala in a cumulative fashion. The two fastreacting Cys, Cys379,380, were mutated first and the resulting C379,380A PGK cDNA was then used to mutate Cys367. The corresponding C379,380,367A PGK cDNA was then used to mutate Cys316 and so forth. The Cys were mutated in the order, Cys379, 380, 367, 316,   Fig. 2A). This result suggested that Cys50, which was the last Cys to be mutated, was required for plasmin reductase activity. To test this hypothesis, the C50A PGK mutant was made and assayed for plasmin reductase activity.
This C50A mutant had similar specific activity as wt PGK ( Fig. 2A). The same qualitative results were observed for all the PGK mutants when plasmin reduction was measured by resolving the MPB-labeled proteins on SDS-PAGE and blotting with streptavidin-peroxidase to detect the labeled angiostatin fragments (Fig. 2B).

Figure 2
These results indicated that neither Cys50 nor any of the other PGK Cys were required for plasmin reduction. We hypothesised that the Cys-less PGK had lost plasmin reductase activity due to secondary effects of the mutations on the integrity of the PGK tertiary structure. This theory was tested by examining the susceptibility of the mutant PGK's to proteolysis by plasmin.
Consequence of mutation of PGK Cys residues on the susceptibility of PGK to proteolysis wt or mutant PGK's were incubated with plasmin and proteolysis of the PGK's examined by SDS-PAGE. wt PGK and the C379,380A, C379,380,367,316A, and C50A PGK mutants were resistant to plasmin proteolysis (Fig. 3). In contrast, the C379,380,367A, C379,380,367,316,108A, C379,380,367,316,108,99A and C379,380,367,316,108,99,50A PGK mutants were proteolysed by plasmin to different extents. In particular, the 6 and 7 Cys to Ala mutants were completely degraded by plasmin during the incubation. The proteolysis was plasmin-dependent as all the PGK's remained intact after incubation with plasmin that had been inactivated with Val-Phe-Lys-chloromethyl ketone (not shown).   (Fig. 5B), or at different temperatures in pH 7.4 buffer for 30 min (Fig. 5C). The thiols in the reduced plasmin/angiostatin were labeled with MPB and detected using streptavidin-peroxidase.

Figure 5
The plasmin reductase activity of PGK was optimal at pH 7 at early time points of incubation. There was no obvious effect, however, of pH on plasmin reduction after 60 min incubation. Increasing NaCl concentrations reduced plasmin reduction, although the effects were relatively modest. Reductase activity was reduced by 73% when the NaCl concentration was increased from 0 to 2 M. There was no substantial effect of temperature on plasmin reductase activity of PGK between 20 and 50 o C.

Effect of 3-PG and ATP-induced conformational change in PGK on the plasmin reductase activity
The reactivity of the fast-reacting Cys in pig muscle PGK are reduced upon binding of 3-PG and/or MgATP to PGK (31). This is a result of conformational changes in the enzyme induced by substrate binding (21). The consequences of substrate-induced conformational changes in PGK for plasmin reductase activity was tested. PGK was incubated with plasmin in the absence or presence of 3-PG and/or ATP and 1 mM MgCl 2 in HEPES/Tween-buffered saline for 30 min at 37 o C. The thiols in the reduced plasmin/angiostatin were labeled with MPB and detected using streptavidin-peroxidase.
Incubation of PGK with 1 mM 3-PG or MgATP reduced plasmin reductase activity bỹ 60%, while incubation with 1 mM 3-PG and MgATP reduced activity by ~90% (Fig. 6A).  Gallic and ellagic acid competitively inhibit PGK kinase activity (32,33). Gallic acid appears to bind to the same site as MgATP (32). Both gallic and ellagic acid at 100 µM concentration inhibited plasmin reduction by PGK by >80% (not shown). NAD(H), another adenine nucleotide, also inhibited plasmin reduction by >60% at 10 mM concentration, presumably through binding to the same site on PGK as MgATP (not shown).

Discussion
Protein reductant active sites typically contain a redox active dithiol/disulfide with the sequence, CysGlyXCys (34). The Cys thiols cycle between the reduced dithiol and oxidized disulfide bond in coordination with a dithiol or disulfide of a protein substrate. This can result in reduction, formation or interchange of disulfide bonds in the protein substrate.
There are seven Cys in PGK, none of which are involved in disulfide bonds, and only two of the seven are nearby in the primary or tertiary structure (22)(23)(24)(25). This suggests that the mechanism by which PGK reduces disulfide bonds in plasmin is unconventional. We have reported that the plasmin reductase activity of PGK is inhibited by NEM and iodoacetamide (6), which implies a role for one or more of the PGK Cys residues in plasmin reduction. In this study we have explored the role of all 7 PGK Cys, and in particular the two fast-reacting Cys, in reduction of plasmin disulfide bonds.
The seven Cys in PGK were mutated to Ala in a cumulative fashion in the order, Cys379, 380, 367, 316, 108, 99 and 50. The two fast-reacting Cys, Cys379,380, were mutated first and the resulting cDNA was then used to mutate Cys367, and so forth. The specific plasmin reductase activity of the mutant PGK's, with the exception of the Cys-less PGK, was similar to that of the wild-type protein. Some mutations were shown to change the tertiary structure of PGK as measured by susceptibility to proteolysis by plasmin. For instance, the Cys-less PGK was rapidly degraded by plasmin which probably accounted for the loss of plasmin reductase activity. These results implied that PGK Cys were not directly involved in plasmin reduction. The question remained, therefore, why alkylation of Cys379,380 inhibited plasmin reduction. This question was explored by reacting the Cys residues with different alkylating/oxidising agents and examining the consequences for plasmin reductase activity.
Carboxymethylation, but not methylation, of the fast-reacting Cys of pig PGK inactivates the kinase activity (35). Reaction of the pig enzyme with DTNB also inactivates the kinase activity (27), while reaction with HgCl 2 or bBBr reduces kinase activity by up to 80% (30). HgCl 2 reacts with free thiols and can facilitate oxidation of a dithiol to a disulfide bond.
bBBr is a homobifunctional alkylating agent that can cross-link thiols in close proximity.
The fast-reacting thiols of PGK were also reacted with TT and GSSG. Accessible thiols react with these compounds to form mixed disulfides with thiosulfate and glutathione, respectively.
Reaction of PGK with all four alkylating/oxidising agents reduced the number of reactive thiols per mol of PGK to ~0.2-0.5. The plasmin reductase activity of the modified proteins was reduced to 7-35% of control. These results indicate that alkylation of the fast-reacting thiols perturb plasmin reductase activity, which is consistent with our earlier report of inhibition of plasmin reductase activity by NEM and iodoacetamide (6). Neither changes in pH, ionic strength nor temperature markedly affected plasmin reductase activity, which implies that the reaction of PGK and plasmin involved predominantly hydrophobic  Cys462-Cys541 and Cys512-Cys536 disulfide bonds in K5 must have been cleaved to release microplasmin from K1-4 and they proposed that the increased -OH ion concentration at alkaline pH was responsible for cleaving the Cys462-Cys541 disulfide bond. We have suggested that the mechanism of plasmin proteolysis at alkaline pH is the same as the mechanism of proteolysis facilitated by PGK at neutral pH (5).
We propose, therefore, that PGK facilitates cleavage of the Cys512-Cys536 disulfide bond by -OH, which results in formation of a sulfenic acid at position 512 and a free thiol at Cys536. The Cys536 thiol is then available to exchange with the Cys462-Cys541 disulfide bond resulting in formation of a new disulfide at Cys536-Cys541 and a free thiol at Cys462.
Kringle 5 is then susceptible to proteolysis at Arg530-Lys531, Arg474-Val475 and/or Lys467-Gly468. This is the simplest sequence of events that can explain all the available data. For instance, cleavage of only the Cys512-Cys536 would not enable release of the kringle 1-4 angiostatin fragments from plasmin. The free thiol that is labeled by MPB in the three angiostatin fragments would be Cys462. We do not exclude cleavage of the Cys483-Cys524 disulfide bond, although this is not required to explain the experimental observations. PGK presumably binds to plasmin and induces a conformational change in kringle 5 that facilitates -OH attack on the Cys512-Cys536 disulfide bond. This suggests that other molecules that interact with plasmin might also facilitate cleavage of the kringle 5 disulfides.
There is recent evidence to support this hypothesis. Interaction of a truncated porcine plasminogen activator inhibitor-1 (residues 80-265), but not full length protein, with plasmin has been shown to result in generation of kringle-containing angiostatin fragments (45). It is possible that the truncated protein is facilitating the same sequence of events in plasmin that are achieved by PGK.       Numbering is based on the sequence of human plasminogen (791 residues) beginning at Glu1, excluding the 19 amino acid signal peptide that ends at Met.