Autoactivation of avian urokinase-type plasminogen activator (uPA). A novel mode of initiation of the uPA/plasmin cascade.

In contrast to mammalian urokinase-type plasminogen activator (uPA), which is produced and maintained in zymogen form, avian uPA is found in the active two-chain form in cultures of normal and transformed chicken cells in the absence of plasmin, the putative natural activator of pro-uPA. Recombinant chicken uPA (ch-uPAwt) synthesized in two distinct expression systems also presents in the active two-chain form. In addition, conversion to the active uPA in both natural and recombinant expression systems could be prevented by uPA-specific inhibitors including a monoclonal antibody that uniquely inhibits the catalytic activity of ch-uPA. Most significantly, an active site mutant of avian uPA (ch-uPAS353A) that lacks catalytic activity is produced and maintained in single-chain form. Furthermore, the single-chain ch-uPAS353A mutant can be converted to the two-chain form by purified active ch-uPAwt. These results strongly indicate an autocatalytic mechanism of activation of ch-uPA. Autoactivation appears to be an intrinsic property of ch-uPA and may be the initiating molecular event in uPA-mediated proteolytic cascades.

With few exceptions, serine proteases are secreted as singlechain zymogens and require limited proteolysis for conversion to a two-chain form that possesses full enzymatic activity. Accordingly, mammalian urokinase-type plasminogen activator (uPA) 1 is secreted and maintained as a zymogen pro-uPA that possesses very low intrinsic activity (1)(2)(3)(4)(5)(6)(7). The precipitating event in the initiation of the uPA/plasmin proteolytic cascade remains elusive, as the natural mechanism of activation of uPA is not well understood. Whereas the activation of pro-uPA is presumed to be mediated primarily by plasmin in a feedback mechanism (2,8,9), plasmin itself must be generated from plasminogen by active uPA and thus cannot be envisioned as the single initiating factor in the proteolytic process. In addition, in plasminogen-deficient mice, uPA is found in its active two-chain form (10), suggesting that mechanisms other than activation by plasmin are responsible for uPA activation. There is accumulating evidence that the single-chain human uPA through interactions with substrates, inhibitors, cell surface receptors, or fibrin clots (7,8,(11)(12)(13) is capable of acquiring transient conformations that are similar to its active two-chain configuration, and the resulting enhanced catalytic activity of the zymogen may serve as the initiating event of this proteolytic cascade. Definitive proof for this mechanism of pro-uPA activation, however, is lacking.
In striking contrast to mammalian cells, which generally produce, secrete, and maintain uPA in the single-chain, zymogen form, transformed chicken embryo fibroblasts (RSVCEF) consistently produce active two-chain uPA in the complete absence of plasminogen and plasmin (14). It was suggested that malignant transformed cells possess a unique structural or enzymatic process that brings about uPA activation (14). We investigated this plasmin-independent activation of avian uPA in a number of cell culture systems, including recombinant expression systems, and present evidence that the activation of chicken uPA is not dependent on malignant transformation but rather is an autocatalytic event that appears to reflect an intrinsic property of avian uPA.

EXPERIMENTAL PROCEDURES
Cell Cultures-Primary cultures of chick embryo fibroblasts (CEF) were prepared from 11-day old embryos as described (14). Secondary cultures of CEF were infected with the BH strain of Rous sarcoma virus (RSV) generating highly transformed cultures of RSVCEF that were propagated, maintained, and characterized as described previously (14). Cultures of the chicken macrophage cell line, HD11, obtained from M. Hayman (SUNY, Stony Brook, NY) and the duck embryo cell line, CCL 141, obtained from ATCC (Rockville, MD) were maintained under identical culture conditions as CEF and RSVCEF. The human epidermoid carcinoma cell line HEp3 and the mouse myeloma cell line NS0 were cultured and maintained as described previously (15).
Expression of Recombinant Chicken uPA (ch-uPA)-The cDNA encoding ch-uPA wt was cloned into Bluescript KS ϩ (Stratagene, La Jolla, CA) as described previously (16). The ch-uPA wt insert was ligated into the EcoRI and BclI restriction sites of the pEE 12 expression vector, a cytomegalovirus promoter-driven expression vector obtained from C. Bebbington (Celltech, Slough, UK). The ch-uPA wt -pEE 12 vector was transfected by electroporation into mouse myeloma NS0 cells. The transfected cultures were grown in the absence of glutamine under limiting dilution to select for cells expressing glutamine synthetase encoded by the pEE 12 vector (17). Selected colonies were screened for ch-uPA wt production by casein-plasminogen zymography and immunoblotting. High level ch-uPA wt -expressing colonies were picked, subcloned, expanded in culture, and harvested in serum-free medium. A subclone, 2F7NS0/ch-uPA wt produced 0.1 g/ml of ch-uPA wt in serumfree conditioned medium.
The ch-uPA wt cDNA also was transferred from Bluescript into pVL1392 a baculovirus-based transfer vector that contains the complete polyhedron gene locus of the Autographa californica nuclear polyhedrosis virus (Pharmingen, San Diego, CA). The ch-uPA wt -containing vector was co-transfected with BaculoGold viral DNA into Sf9 insect cells. Recombinant baculovirus containing the ch-uPA wt insert was amplified through sequential infections of Sf9 cells and then used to infect Hi5 insect cells, a high level protein producer insect cell line * This work was supported in part by Grant CA76039 from the National Cancer Institute. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
(Invitrogen, San Diego, CA). Infected cells expressing and secreting ch-uPA wt protein were harvested and analyzed for the production and zymogen status of ch-uPA.
Diisopropyl Fluorophosphate (DFP) Treatment and Casein-Plasminogen Zymography-Treatment with the irreversible serine protease inhibitor, DFP, can differentiate between single-chain zymogen uPA, which is resistant to DFP and two-chain active uPA, which is completely inhibited by DFP. Samples of conditioned media are incubated at 20°C with 5 mM DFP for 1 h followed by a second 1 h treatment with 5 mM DFP. The samples are then electrophoresed on SDS-polyacrylamide gels (12%) under nonreducing conditions alongside untreated samples or solvent (isopropyl alcohol) only treated samples. The gels are then washed two times with 2.5% Triton-X100 followed by a 1 h washing in H 2 O. The washed gel is placed over an underlay consisting of 1% agarose, 5% nonfat dry milk (casein) and 50 g/ml chicken plasminogen and incubated at 37°C until pronounced zones of lysis appear at the electrophoretic position of uPA.
Reducing SDS-PAGE, Western Blotting, and Immunodetection-Samples of conditioned media are electrophoresed under nonreducing (no ␤ME) or reducing (5% v/v ␤ME) conditions on SDS-polyacrylamide (12%) gels. The gels are transferred at 30 volts to nitrocellulose membranes. The membranes are blocked for 2 h in 5% casein-0.1% Tween 20 then incubated overnight at 4°C with 1 g/ml rabbit polyclonal antibody raised against ch-uPA. After washing, the membranes are incubated with horseradish peroxidase-conjugated goat anti-rabbit secondary antibody and following extensive washing are developed by ECL chemiluminescence (Amersham Pharmacia Biotech). 35 S-Labeling of ch-uPA and Immunoprecipitation-Cell cultures that are expressing recombinant wild type or mutant ch-uPA are washed twice with serum-free Dulbecco's modified Eagle's medium and placed in serum-free Dulbecco's modified Eagle's medium containing one-tenth the normal amount of methionine and cysteine and supplemented with [ 35 S]methionine (1100 Ci/mmol) and [ 35 S]cysteine (1100 Ci/mmol) to a total of 75 Ci/ml. The cultures are incubated for 18 -20 h to radiolabel the expressed proteins, and the conditioned medium is harvested, centrifuged (3000 rpm for 10 min) and immunoprecipitated to capture the radiolabeled ch-uPA. Immunoprecipitation is carried out by adding 50 g/ml purified rabbit anti-ch-uPA polyclonal antibody to the radiolabeled conditioned medium and incubating overnight at 4°C. Following incubation, protein A-Sepharose beads are added (1:20 v/v) and incubation continued for an additional 4 -6 h. The mixture is centrifuged to capture the immune complexes, washed extensively, resuspended in 7.5 M urea-1ϫ SDS-PAGE sample buffer with or without reducing agent (␤ME), electrophoresed on SDS-polyacrylamide (12%) gels and processed for autoradiography as described previously (18).
Protein Purification and Analysis-Chicken plasminogen was purified from chicken plasma by lysine-Sepharose affinity chromatography as described (19). The ch-uPA wt was purified by benzamidine-Sepharose affinity chromatography and gel filtration chromatography as described (19). Polyclonal antibodies to ch-uPA and a neutralizing monoclonal antibody to ch-uPA were purified by affinity chromatography as described previously (14,20). Human plasminogen, plasmin, uPA, and antibodies to human uPA were from American Diagnostica (Greenwich, CT).

Construction and Expression of the Active Site Mutant of ch-uPA-
The active site serine residue (Ser 353 ) of ch-uPA was mutated to alanine utilizing the method of splicing by overlap extension and polymerase chain reaction (21). The internal 5Ј primer, 5Ј-GGGAGATGCTGGCG-GCCC-3Ј, was used to incorporate the nucleotide change (TCT 3 GCT) that mutates Ser 353 to alanine. The resulting cloned DNA was sequenced to verify the mutation and subcloned into both the pEE 12 and baculovirus expression vector for expression in NS0 and Hi5 cells, respectively, as described above. The expressed mutant, ch-uPA S353A , was completely inactive catalytically in both the plasminogen-coupled chromogenic substrate assay (16) and the zymographic underlay assay.

RESULTS AND DISCUSSION
To investigate the mechanism of activation of chicken uPA, the cDNA was subcloned into a baculoviral expression vector and in a mammalian expression vector for expression in two distinct, heterologous culture systems: Hi5 insect cells and NS0 mouse myeloma cells, respectively. Cells that synthesized and secreted ch-uPA were selected, expanded in culture, and incubated to release ch-uPA into the conditioned medium. All conditioned media were obtained under serum-free conditions in the absence of plasminogen and in the presence of aprotinin, a potent plasmin inhibitor. The recombinant ch-uPA from the two expression systems was tested for its zymogen status by two independent methods. In the first method, which tests the conformational status of the active site, conditioned media from cultures of RSVCEF, Hi5-ch-uPA wt and NS0-ch-uPA wt are left untreated or treated with DFP, an irreversible active site inhibitor of serine proteases that targets only the active form of the enzymes. Subsequently, the samples are subjected to SDS-PAGE and underlay zymography. Active two-chain uPA would be sensitive to DFP, resulting in the diminution of zones of lysis in the zymograph. In contrast, single-chain zymogen uPA would be resistant to inhibition by DFP, displaying a zone of lysis comparable to that created by the untreated sample. As shown in Fig. 1A, recombinant ch-uPA produced in both expression systems (lanes 3-6) is as sensitive to DFP as the native RSVCEF-produced ch-uPA (lanes 1-2), indicating that ch-uPA presents entirely in its active form.
That the expressed recombinant ch-uPA is in the two-chain active protease form was determined by an alternative method based on protein structure, namely electrophoresis on SDSpolyacrylamide gels under nonreducing or reducing conditions, followed by Western blotting and probing with a polyclonal anti-ch-uPA antibody (Fig. 1B, lanes 1-6). Upon reduction, zymogen ch-uPA would be maintained as a 48-kDa single-chain form, whereas two-chain uPA would dissociate into a 35-kDa B-chain and a 12-14-kDa A-chain. The recombinant ch-uPA secreted in both the insect and the mammalian expression systems (lanes 3 and 4, and 5 and 6, respectively) exhibits only the 35-kDa B-chain upon reduction (A-chain is not immunoreactive with this antibody), with little or no evidence of the 48-kDa single-chain form, indicating that recombinant ch-uPA is found in the two-chain form identically to the native ch-uPA expressed by RSVCEF (lanes 1 and 2).
In contrast to avian uPA, human uPA obtained from conditioned medium of human tumor cells is maintained in the single-chain zymogen form, as seen by resistance to DFP treatment (Fig. 1A, lanes 7 and 8). Furthermore, recombinant human uPA expressed in the same insect cell expression system was found to be secreted and maintained in single-chain zymogen form (Fig. 1B, lanes 7 and 8), under identical conditions to those that allowed complete conversion of ch-uPA to the active two-chain form, indicating that there is no ubiquitous singlechain uPA-converting enzyme in the recombinant expression system that had activated the chicken pro-uPA.
Thus, the recombinant ch-uPA produced by two heterologous expression systems is found to be in the active two-chain form, indicating that ch-uPA is an intrinsically unstable zymogen that converts rapidly and efficiently to the active two-chain uPA form independent of the culture system in which it is expressed. Interestingly, this intrinsic conversion process generates the same two-chain uPA molecule as plasmin-mediated activation of uPA. Recombinant ch-uPA harvested from NS0ch-uPA wt cultures was purified and subjected to SDS-PAGE under reducing conditions. The B-chain was blotted onto polyvinylidene difluoride membrane and subjected to amino-terminal sequencing. The first 8 residues obtained were Ile, Val, Gly, Gly, Ser, Gln, Ala, Glu, corresponding to ch-uPA sequence 153 IVGGSQAE 160 , indicating that the cleavage site is at Lys 152 -Ile 153 , the homologous position in all mammalian uPA's that is cleaved during plasmin-mediated activation.
Inhibition of the conversion process using highly specific reagents indicates that the conversion of ch-uPA requires its own catalytic activity. A monoclonal antibody to ch-uPA added to cultures expressing ch-uPA wt (Fig. 2) prevents the conversion reaction, yielding a DFP-insensitive chicken uPA zymogen (lanes 3 and 4). This monoclonal antibody has been shown to be highly specific in blocking the catalytic activity of ch-uPA (20). Low molecular weight inhibitors of uPA activity such as benzamidine and nitrophenyl guanidinobenzoate also prevent chicken pro-uPA conversion, whereas serine protease inhibitors specific for plasmin-or trypsin-like enzymes such as ⑀ACA, SBTI, and trasylol had no inhibitory effect on conversion in any of the ch-uPA-expression systems (data not shown). Thus, only specific inhibitors of uPA effectively impede conversion of ch-uPA, suggesting that the catalytic activity of the enzyme is necessary for its own conversion. To test this hypothesis directly, we designed, constructed, and expressed an active site mutant chicken uPA devoid of catalytic activity. This mutant, in which the active site serine (Ser 353 ) was replaced by alanine (ch-uPA S353A ), was expressed both in the Hi5/baculoviral expression system and in the NS0 mammalian expression system. The mutant ch-uPA S353A was indistinguishable from wildtype ch-uPA on SDS-PAGE under nonreducing conditions and was shown to lack any measurable activity on underlay zymography or in a chromogenic substrate assay (data not shown). However, when the expressed ch-uPA S353A mutant is assessed for single-chain or two-chain status by Western blotting, it presents in the 48-kDa single-chain form under reducing conditions (Fig. 3, lane 2), whereas the wild-type ch-uPA wt expressed in parallel cultures exhibits only the two-chain form (Fig. 3, lane 1), dissociating under reducing conditions into a 35-kDa B-chain and a 14-kDa A-chain. Therefore, abrogating the catalytic activity of chicken uPA by mutagenesis of the active site serine residue completely ablates conversion to the two-chain uPA under conditions identical to those permissive for the generation of the active two-chain form of the wild-type enzyme. This result is consistent with and confirmatory to the preventive effects of the exogenously added uPA-specific inhibitors (Fig. 2) and strongly supports an autocatalytic mechanism of conversion.
To ensure that the serine to alanine mutation of chicken uPA did not alter its conformation so as to make the activation site (Lys 152 -Ile 153 ) inaccessible, or the entire molecule incapable of conversion, the mutant ch-uPA S353A was treated with plasmin, the putative natural activator of uPA. Indeed, the mutant chicken uPA is converted by plasmin in a dose-dependent manner (Fig. 3, lanes 3 and 4) to yield a B-chain indistinguishable from that of ch-uPA wt . These data attest that the mutant is conformationally capable of being converted but remains in the single-chain form under the identical conditions that allowed for complete conversion of the wild-type chicken uPA to the two-chain form. These results also further support the absence of any distinct converting enzyme in these expression systems. If such a convertase existed, it should be, like plasmin, equally capable of converting ch-uPA S353A as it would ch-uPA wt , yet no two-chain ch-uPA S353A has ever been observed.
To demonstrate that ch-uPA autoactivation could be recapitulated in vitro, the inactive ch-uPA S353A mutant was radiolabeled and utilized as a single-chain substrate for conversion by purified active ch-uPA wt . Fig. 4 illustrates that ch-uPA wtcontaining conditioned medium (lane 2) as well as 50 -150 ng of purified ch-uPA wt (lanes 3 and 4) can process 1 g of single- FIG. 1. Native and recombinant chicken uPA is found in the active two-chain form, whereas human uPA is maintained in the single-chain zymogen form. A, detection by DFP sensitivity and zymography. uPA-containing samples of serum-free conditioned medium (CM) harvested from RSVCEF cultures (lanes 1 and 2), Hi5 insect cells infected with recombinant baculovirus containing the chicken uPA cDNA (Hi5/ch-uPA, lanes 3 and 4), and NS0 mouse myeloma cells expressing recombinant chicken uPA (NS0/ch-uPA, lanes 5 and 6) were treated with DFP, an irreversible serine protease inhibitor (lanes 2, 4, and 6) or with solvent control (lanes 1, 3, and 5). The samples were then subjected to SDS-PAGE followed by casein-plasminogen underlay zymography. All CM were obtained in the absence of plasminogen and in the presence of the plasmin inhibitor aprotinin. Identically treated uPA-containing samples of serum-free CM from cultures of a human epidermoid carcinoma cell line, HEp3 (lanes 7 and 8) were used as the single-chain uPA control, unreactive to DFP. Mobilities of chicken and human uPA are indicated. B, detection by Western blotting and immunostaining of nonreduced or reduced uPA. uPA-containing samples of serum-free CM from RSVCEF cultures (lanes 1 and 2), Hi5/ch-uPA cultures (lanes 3 and 4), and NS0/ch-uPA cultures (lanes 5 and 6) were subjected to SDS-PAGE under nonreducing (lanes 1, 3, and 5) or reducing (lanes 2, 4, and 6) conditions, transferred to nitrocellulose membrane and probed with a polyclonal anti-chicken uPA antibody (lanes 1-6). Human uPA expressed in the baculovirus system (Hi5/huPA) (lanes 7 and 8) was used as a single-chain control and probed with a rabbit anti-human uPA polyclonal antibody (lanes 7 and 8). Molecular weight markers are indicated on the left. High molecular weight (HMW) and low molecular weight (LMW) uPA species observed under nonreducing conditions are marked on the left. The 35-kDa B-chain and the 50-kDa single-chain uPA observed under reducing conditions are marked on the right as B or scuPA, respectively. The ϳ14-kDa A-chain position is marked on the right as well, although the A-chain is not detectable by the anti-chicken uPA antibody on Western blots under reducing conditions. The presence of the A-chain is demonstrated when radiolabeled ch-uPA is immunoprecipitated by this antibody and subjected to SDS-PAGE under reducing conditions (see Fig. 4).

FIG. 2.
Anticatalytic monoclonal anti-ch-uPA IgG inhibits conversion of native ch-uPA in normal CEF cultures. Normal CEF were treated with PMA to induce uPA and cultured in serum-free medium in the absence (lanes 1 and 2) or presence (lanes 3 and 4) of the anti-ch-uPA IgG, DMC8, an anticatalytic monoclonal antibody specific for chicken uPA (20). The conditioned media were harvested and treated with DFP (lanes 2 and 4) or with solvent control (lanes 1 and 3) then subjected to SDS-PAGE and casein-plasminogen underlay zymography until clear zones of lysis appear. Molecular weight markers are indicated in kDa on the left, and the electrophoretic mobility of chicken uPA is shown on the right. chain ch-uPA (lane 1) to the two-chain form (lanes 2 and 4). This result demonstrates that active ch-uPA wt has the catalytic capability and specificity to autolytically convert inactive single-chain ch-uPA to two-chain ch-uPA. In concert with the evidence for ch-uPA activation in multiple independent expression systems (Fig. 1), the inhibitory effect on conversion of an anticatalytic monoclonal antibody (Fig. 2) and that the active site mutation abrogates conversion (Fig. 3), the ability of ch-uPA wt specifically to convert ch-uPA S353A argues strongly for an autocatalytic mechanism of conversion of ch-uPA.
Finally, it appears that this plasmin-independent generation of active two-chain avian uPA is not unique to the transformed chicken fibroblast culture system. Normal avian macrophages stimulated with phorbol ester also produce active two-chain uPA in the absence of plasmin and in the presence of aprotinin (Fig. 5A, lane 2), similar to RSVCEF (lane 1). Furthermore, a normal duck cell line, cultured as well in the absence of plasmin and in the presence of the plasmin inhibitor, also produces the active (DFP-sensitive) form of avian uPA (Fig. 5B, lanes 3  and 4). Thus, the unusual occurrence of active two-chain uPA appears to be a distinct feature of the avian fibrinolytic system. This is the first description of autoactivation of a protease of the fibrinolytic cascade. There are, however, other studies that document a mechanism of autoactivation for factor VII (22,23) and factor XII (24 -26), the two serine proteases that initiate the intrinsic and extrinsic coagulation cascade, respectively. Furthermore, the initiation of the classical pathway of the complement cascade, namely activation of C1, also is an autocatalytic event (27,28). It is interesting to note that activation of PA is believed to be the initiating molecular event in the fibrinolytic cascade, possibly indicating a pattern that the primary initiation of biologically significant proteolytic cascades may be an autocatalytic event.
Mammalian uPA is controlled post-translationally at multiple levels, including synthesis and secretion in a stable zymogen form and tight regulation of its catalytic activity and turnover by PA-specific serpins PAI-1 and PAI-2 (29 -31). This study demonstrates that avian uPA is capable of autoactivation, thus precluding the regulatory requirement of another protease to initiate pro-uPA activation. In addition, previous studies in our laboratory have shown that chicken uPA is resistant to inhibition by the PA-inhibiting serpins PAI-1 and PAI-2 (16). Thus, it appears that chicken uPA lacks the two major post-translational regulatory mechanisms that control mammalian uPA. The physiological significance of this apparent lack of regulation in the avian fibrinolytic/proteolytic system is not known, but it appears from studies on both avian embryo development and the promoter region of the uPA gene  (50 and 150 ng, in lanes 3 and 4, respectively). After overnight incubation, the samples were immunoprecipitated utilizing a polyclonal rabbit anti-ch-uPA antibody, electrophoresed on 12% SDS gels under reducing conditions, and subjected to autoradiography. Molecular weight markers in kDa are indicated on the left, and uPA forms under reducing conditions are marked on the right. that chicken uPA is regulated spatially and temporally at the level of expression rather than post-translationally (32)(33)(34)(35). Furthermore, since the existence and location of specific avian PAIs have never been reported, it is difficult to assess how the intrinsic autoactivation of avian uPA is controlled catalytically and spatially at the tissue level.
In conclusion, chicken uPA constitutes an interesting natural variant of uPA that can evade most regulatory mechanisms that constrain the behavior of mammalian uPA. Identification of the structural motif(s) that dictate the autoactivating propensity of chicken uPA will not only yield mechanistic insight into the process but also will substantiate the corresponding structural motifs in human uPA that apparently stabilize the zymogen and limit pro-uPA autoactivation. This unusual uPA variant has already shed light on the activation modality of proteolytic cascades in the avian system and may continue to confer insights into the structural/functional features of serine proteases as well as on the exquisite control systems that have developed in the course of mammalian evolution.