Production and characterization of recombinant human plasminogen(S741C-fluorescein). A novel approach to study zymogen activation without generation of active protease.

A variant of recombinant plasminogen with the plasmin active site serine (S741) replaced by cysteine was produced and labeled with fluorescein at this residue to provide the derivative Plg(S741C-fluorescein). Studies of cleavage, conformation, and fibrin-binding properties of the derivative showed it to be a good model substrate to study plasminogen activation. Both in solution and in a fully polymerized fibrin clot, cleavage of the single chain zymogen to the two-chain "plasmin" molecule was accompanied by a 50% quench of fluorescence intensity. This change allows facile, continuous monitoring of the kinetics of cleavage. Measurements of cleavage by single chain t-PA within intact, fully polymerized 3 microM fibrin yielded apparent kcat and Km values of (0.08 s-1, 0.52 microM) and (0.092 s-1, 0.098 microM) for [Glu1]- and [Lys78]Plg(S741C-fluorescein), respectively. These values are similar to those obtained by others with plasma plasminogen. The approach used here might generally be useful in simplifying the analysis of zymogen activation kinetics in cases where the product (protease) has a great influence on its own formation via positive or negative feedback loops.

Construction of Plasminogen(S741C)-The full-length plasminogen cDNA (18) was inserted as a BalI-SphI fragment into the multiple cloning site (mp-18) of pATA-18 as described (19). Site-directed mutagenesis was achieved by the polymerase chain reaction overlap-extension technique (20). We employed the two partly complementary oligonucleotides 2 and 3 (Table I) to change the codon for Ser 741 (AGT) to Cys (TGT), whereas amplification of the 760 base pairs 3Ј-end of the plasminogen cDNA was achieved by oligonucleotides 1 and 4. The mutated fragment was used to substitute the "wild-type" EcoRV-SphI fragment of plasminogen in pATA-18. The absence of undesired mutations was verified by DNA sequencing of the entire fragment, using oligonucleotides 1, 3, and 4.
Construction of Stable Cell Lines-Stable cell lines expressing [Glu 1 ]Plg(S741C) were constructed essentially as described for native Plg (19). In brief, cDNA for human [Glu 1 ]Plg(S741C) was removed from the pATA-18 plasmid by digestion with HindIII and blunt ends were generated with T4-DNA polymerase. The expression vector pNUT (21) was digested with SmaI, treated with alkaline phosphatase, and ligated to the [Glu 1 ]Plg(S741C) cDNA. The correct insertion and identity of [Glu 1 ]Plg(S741C) cDNA was evidenced by sequencing, using oligonucleotides 5 and 6. BHK cells were cultured in Dulbecco's modified Eagle's medium/F-12, supplemented with 5% newborn bovine serum from which (bovine) plasminogen was removed by passage over lysine-Sepharose. Baby hamster kidney (BHK-21) cells were transfected with the pNUT -Plg(S741C) plasmid according to the calcium phosphate precipitate method (22). Sixteen hours after transfection, the growth medium was supplemented with 0.44 mM methotrexate to select for pNUT (dihydrofolate reductase). Two weeks after transfection, individual clones were screened for [Glu 1 ]Plg(S741C) production by a sandwich enzyme-linked immunosorbent assay for human plasminogen (Affinity Biologicals, Ancaster, Ontario, Canada). Typical production levels were 10 g/10 6 cells/day. Production, Purification, and Fluorescein Labeling of [Glu 1 ]Plg-(S741C)-Cell lines were grown in 500-cm 2 triple flasks (Nunc) for large scale production. At confluence, the cells were washed and the selection medium was replaced by serum-free Opti-MEM I, supplemented with 50 M ZnCl 2 . Conditioned media were collected every other day; supplemented with 1 M dansyl-Glu-Gly-Arg-chloromethyl ketone, 10 mM Tris-HCl, pH 8.0, and 1 mM EDTA; and loaded onto lysine-Sepharose. After washing the column with PBS (20 mM NaP i , pH 7.4, 150 mM NaCl) until the A 280 of the effluent was Ͻ0.002, [Glu 1 ]Plg(S741C) was eluted with PBS, 10 mM 6-aminohexanoic acid. Fractions containing Ͼ0.25 mg/ml [Glu 1 ]Plg(S741C) based on A 280 were pooled. Labeling was performed directly on this pooled fraction (typically 10 -15 ml, 6 -10 M [Glu 1 ]Plg(S741C)) by adding 150 l of 20 mM 5-IAF in N,NЈ-dimethylformamide. The reaction was continued for 30 min at room temperature in the dark. Excess label was removed from this incubation mixture on a 1-ml DEAE-fast flow column (Pharmacia, Uppsala, Sweden), equilibrated and run in PBS, 0.05% Tween 80, resulting in elution of Plg(S741C-fluorescein) in the flow-through and binding of 5-IAF to the resin. Subsequently, the protein was concentrated and freed from remaining traces of 5-IAF and 6-aminohexanoic acid (6-AHA) by applying to a 5-ml DEAE-fast flow column after a 1:5 dilution with 20 mM Tris-HCl, pH 8.0. The column was washed with 20 mM HEPES-NaOH, pH 7.4, and the labeled protein eluted with HBST (20 mM HEPES-NaOH, pH 7.4, 150 mM NaCl, 0.02% Tween 80) and stored in aliquots at Ϫ70°C. All chromatographic steps were performed in the dark, and control runs were performed with "wild-type" plasminogen to determine the absence of nonspecific labeling. The amount of fluorescein incorporation was determined spectrophotometrically using an extinction coefficient at 495 nm for fluorescein of 84,000 M Ϫ1 cm Ϫ1 (Molecular Probes). The concentration of the labeled protein was determined from absorbance at 280 nm, after correction for the contribution of fluorescein (A 280 ϭ 0.19⅐A 495 ). Typical incorporation levels were at 0.9 mol/mol. [Glu 1 ]Plg(S741C-fluorescein) was converted to [Lys 78 ]Plg-(S741C-fluorescein) by adding 5 l of 188 M plasmin to 2.5 ml of 15 M plasminogen in HBST supplemented with 5 mM 6-aminohexanoic acid. After 90 min, plasmin was removed by binding to 1 ml of aprotininagarose. The flow-through, containing the [Lys 78 ]Plg(S741C-fluorescein), was treated with 10 M valylphenylalaninyllysyl chloromethyl ketone for 1 h. The absence of traces of plasmin was verified by incubating aliquots of the protein at 37°C with 1 mM S2251 for 2 h, in which period no increase of the absorbance at 405 nm was observed. Samples of Plg(S741C-fluorescein) (Glu 1 and Lys 78 form) were subjected to urea/ acetic acid 7.5% PAGE (23) at 120 V in a minigel system and mobilities were compared to native plasminogen. Gels were stained with Coomassie Brilliant Blue and destained.
Fibrin-binding Assay-Binding of [Glu 1 ]-or [Lys 78 ]Plg(S741C-fluorescein) forms to a fibrin clot was measured as follows. To a series of microcentrifuge tubes containing 2 l of 30 nM ␣-thrombin in HBST, 0.5 M CaCl 2 equilibrated at 37°C, were added 98 l of HBST with a fixed concentration of Plg(S741C-fluorescein) and various concentrations of fibrinogen. Clotting of the fibrinogen was complete within 1 min, after which the incubation was continued for 10 min at 37°C. Subsequently, the fibrin clot was pelleted by centrifugation for 1 min at 10,000 ϫ g and the supernatant was removed immediately. The amount of non-fibrinbound Plg(S741C-fluorescein) was determined by quantitation of fluorescence intensity of the supernatant, as follows. Aliquots of 25 and 75 l of each supernatant were added to 75 and 25 l of HBST in 96-well fluorescence plates, and fluorescence intensity was measured at excitation and emission wavelengths of 495 and 535, respectively, employing a 530-nm emission cut-off filter. Fluorescence intensities were converted to concentration of plasminogen in the supernatant by comparison to the supernatant of an otherwise identically treated sample that did not contain fibrin. The latter was identical to the intensity of untreated plasminogen, directly diluted from the stock solution. Fluorescence intensities obtained in this way were linear with respect to plasminogen concentration as established by measuring, under identical conditions, the fluorescence intensities in wells that contained serial dilutions of fluorescent plasminogen. In an alternative experiment, we determined the binding of various concentrations of Fluorescence intensities were measured at excitation and emission wavelengths of 495 and 535, respectively, employing a 530-nm emission cut-off filter. [Glu 1 ]-or [Lys 78 ]Plg(S741C-fluorescein) (90 l) was added to wells with HBST and equilibrated until a stable fluorescence signal was obtained. Then, 6-AHA was added at various concentrations and again a stable signal was obtained. Next, 15 units of u-PA were added to give a final volume of 100 l and fluorescence intensity was followed at 1-min intervals. The time courses of cleavage of fluorescent plasminogen were subjected to nonlinear regression analysis according to a single exponential decay process, from which the rate constants and, therefore, rates of the reaction were determined. This analysis was justified since rates of cleavage activation were linear with respect to both the u-PA and [Glu 1 ]-or [Lys 78 ]Plg(S741C-fluorescein) concentration. Second order rate constants were obtained from the pseudo first order rate constants by assuming 60,000 units/mg u-PA and a molecular weight of 55,000.
Fluorescent Plasminogen Cleavage by t-PA-Experiments to meas-    1 ]Plg(S741C) was expressed in stably transformed BHK cells at production levels of 10 g of plasminogen/ml of serum-free medium (10 6 cells)/day. [Glu 1 ]Plg(S741C) was purified from the BHK conditioned medium to apparent homogeneity by conventional affinity chromatography on lysine-Sepharose. Analysis of the protein on acid urea gels ( Fig.  1) showed the absence of detectable amounts of degraded products. Laser densitometry of Coomassie-stained acid/urea gels showed that the recombinant plasminogen displays a similar ratio of the two glycoforms: 35% plasminogen-I (glycosylated at Asn 288 and Thr 345 ) and 65% plasminogen-II (glycosylated at Thr 345 only) as observed for plasma plasminogen (24) and recombinant wild-type plasminogen (19).

Production and Characterization of [Glu 1 ]Plg(S741C)-Recombinant [Glu
The production levels in BHK cells and yields after purification on lysine-Sepharose of [Glu 1 ]Plg(S741C) were identical to those for wild-type recombinant plasminogen (19). This suggests that no gross alterations in the structure result from the introduction of a new, free cysteine replacing serine 741. To further substantiate the validity of this plasminogen variant as a model, we analyzed by intrinsic fluorescence one of the most striking properties of [Glu 1 ]plasminogen, namely its tight, activation-resistant conformation (25,26). The increase of this intrinsic fluorescence (Fig. 2) has a sigmoidal relationship with increasing concentrations of the lysine analog 6-AHA as shown previously for plasma derived plasminogen (27) and "wild-type" recombinant plasminogen (19), indicating the tight (activationresistant) conformation of the Glu 1 form of this variant.

Plg(S741C) Lacks Detectable Proteolytic Activity-
The absence of amidolytic and proteolytic activity in the variant plasminogen molecule (S741C) was substantiated by "activation" with the three plasminogen activators u-PA, t-PA, and streptokinase as follows. [Glu 1 ]Plg(S741C) at 5 M was incubated with 150 units/ml u-PA in the presence of the chromogenic substrate S2251 (1.0 mM), and the absorbance at 405 nm was followed in time for 20 h. The progress curve did not vary from similar experiments without plasminogen, whereas identical experiments with native plasminogen indicated a lower detection limit of 0.5 nM plasmin. Incubations of 5 M [Glu 1 ]Plg(S741C) with 5 or 50 units/ml streptokinase, which generates an active site in plasminogen by 1:1 complex formation rather than cleavage at Arg 561 -Val 562 , were performed in a similar fashion as described (19), and gave an amidolytic activity (S2251) corresponding to 0.0125% of the input [Glu 1 ]Plg(S741C), when compared to the native plasminogenstreptokinase complex in an identical experiment. Finally, 5 M [Glu 1 ]Plg(S741C) was included in a clot lysis experiment with 3 M fibrin and 5 nM t-PA, as described previously (19,28). The turbidities of these clots were stable for 16 h, whereas in control experiments run simultaneously at little as 0.1 nM native plasminogen can be detected by a decrease in turbidity due to lysis of the fibrin. Based on these assays we conclude that [Glu 1 ]Plg(S741C) when "activated" does not possess sufficient intrinsic amidolytic activity to perturb the experiments that are described in this paper. Changes in intrinsic fluorescence of [Glu 1 ]Plg(S741C) upon binding of 6-AHA were quantified as described for recombinant and plasma plasminogen (19). Excitation and emission wavelengths were at 290 and 340 nm, respectively. The increase in fluorescence shows a sigmoidal relation to 6-AHA concentration. The corresponding Hill plot is shown in the inset as reported (19). The Hill coefficent was 2.0, and the transition midpoint was 0.52 mM.
The interaction of [Glu 1 ]Plg(S741C-fluorescein) with 6-AHA was analyzed by quantifying the change in fluorescence intensity of the reporter group (Fig. 3). Contrary to the results on the intrinsic fluorescence change (Fig. 2), the intensity decreases (21.4% at saturation) and can be described best by a model that embodies binding to a single binding site with K d of 2.52 mM. The Hill coefficient for the inferred binding suggested minimal cooperativity (h ϭ 1.1). The decrease in fluorescence intensity of [Lys 78 ]Plg(S741C-fluorescein) was Ͻ 1.5% at 10 mM 6-AHA. Intrinsic (Trp) fluorescence cannot be studied for these molecules due to interference by the fluorescein label, precluding a direct comparison with plasma plasminogen or [Glu 1 ]Plg-(S741C). Comparison of the results of Figs. 2 and 3, however, suggests that the interaction between 6-AHA and plasminogen measured by intrinsic fluorescence, with a Hill coefficient of 2.0 and a transition midpoint at 0.52 mM 6-AHA, is different from the interaction measured by extrinsic fluorescence.
The existence of an activation-resistant conformation has been used to rationalize the weak fibrin binding of [Glu 1 ]-as compared to [Lys 78 ]plasminogen. To determine whether the Glu 1 and Lys 78 forms of the fluorescent plasminogen exhibit these differences in affinity for fibrin, we measured their binding to fully polymerized clots (Fig. 4). Fibrin binding of [Glu 1 ]Plg(S741C-fluorescein) was weak with an estimated K d of 30 M, whereas that of [Lys 78 ]Plg(S741C-fluorescein) was stronger (K d ϭ 1.2 M, n ϭ 1.8 sites/fibrin). These trends are identical to those of the "wild-type" recombinant plasminogen species (19), and the values of the binding parameters are similar to those reported for the plasma plasminogen forms (K d ϭ 38 M and 0.32, respectively) (29).
The Activation Cleavage of Plg(S741C-fluorescein) by u-PA in Solution-[Glu 1 ]Plg(S741C-fluorescein) was treated with 6-AHA, and the initial decrease in fluorescence was measured. Then urokinase was added and the progressive decrease over time was monitored continually. An example is shown in Fig.  5A. Functional homogeneity of the fluorescent plasminogen is suggested by the coincidence of data and the line obtained by linear regression to the equation for first order decay. In this experiment, samples were withdrawn at regular intervals after the addition of urokinase and subjected to SDS-PAGE. The fluorescent bands were photographed (Fig. 5A, inset), and the gel was stained with Coomassie Blue, destained, and scanned with a densitometer. As the inset of Fig. 5A indicates, the progressive decline in intensity correlates with cleavage and, as expected, the fluorescence is associated exclusively with either the zymogen or the light chain, with no visible fluorescence in the heavy chain. Although the data are not shown, densitometry indicated that the decline in fluorescence after the addition of urokinase was linear in the extent of cleavage of [Glu 1 ]Plg(S741C-fluorescein), and linear regression of intensity values to the extent of the reaction predicted an overall drop in intensity (including that due to 6-AHA) of 54% at 100% completion of the reaction.
Notably, the maximal decrease in fluorescence intensity of [Glu 1 ]Plg(S741C-fluorescein) upon activation at different 6-AHA concentrations was inversely proportional to the initial change upon adding 6-AHA (data not shown), making the combined effects of 6-AHA and u-PA constant at about a 50% decrease. From this observation we suggest that the [Glu 1 ] plasmin molecule has a relaxed conformation similar to [Lys 78 ] plasminogen, and therefore no effect of 6-AHA on the overall intensity change is observed. This is in agreement with results on the properties described for active site blocked Glu-plasmin, which are similar to Lys-plasminogen, with respect to, for example, affinity for fibrin (30). The influence of 6-AHA on the rate of plasminogen activation by urokinase in solution has been well documented and resulted in the proposal of a tight, activation-resistant conformation of [Glu 1 ]plasminogen in the absence of such lysine analogs (26,31). A similar effect of 6-AHA on activation rates is shown for [Glu 1 ]Plg(S741C-fluorescein) in Fig. 5B. The decrease at higher concentrations of 6-AHA has been shown to result from the direct inhibitory effect of lysine analogs on the activity of u-PA (31). The maxi-  Fig. 7A. Upon initialization of the reactions, an initial decrease in intensity (ϳ10%) occurred because of dilution. Although a small additional change (4.8%) followed the polymerization of fibrin at the high input level of fibrinogen, as evidenced by the difference in control (minus t-PA) signals of Fig. 7A, the magnitude of the subsequent decreases when the reactions approached completion were virtually identical at both the high and low fibrin concentrations. The relative lack of influence of fibrin polymerization on the signal, potentially due to, for example, light scattering can most likely be attributed to the plate reader format whereby both the excitation and emission optics are above the sample and the sample well is quite reflective. The lines of Fig. 7A are the regression lines obtained by fitting the data to the equation for first order decay, a procedure that is justified because of the relatively low input concentration of [Glu 1 ]Plg(S741C-fluorescein). The good fit of the data to the equation implies functional homogeneity. The regression analysis indicated a 50% decrease in intensity upon completion of the reaction. The data from similar experiments with [Lys 78 ]Plg(S741C-fluorescein) did not fit as well to the first order decay equation (due to low K m ), but monitoring for extended periods indicated a 40% decline in intensity at completion of the reaction (data not shown), at both low and high input concentrations of fibrinogen. In the absence of the fluorescein-labeled protein, the signal was negligible (7.0% or less than that with the fluorescent protein over the range of protein concentrations studied). This blank value, however, was subtracted from all relevant data.
In order to measure initial rates, activator concentrations were decreased so that the approximately linear portion of the reaction could be measured. An example is shown in Fig. 7B. In this case the magnitude of signal change encompassed by the exhibited data is about 10% of the total signal, which corresponds to about 20% consumption of the substrate. Over this range the rate was essentially constant. In typical measurements of initial rates, data such as those of Fig. 7B were subjected to linear regression to determine the slope. This approach was employed to obtain the apparent k cat and K m  (Fig. 8). DISCUSSION We describe the production of a variant of plasminogen in which the serine of the plasmin catalytic triad has been replaced by cysteine. This enabled the introduction of a fluorescein label at the position of this residue in plasminogen to produce [Glu 1 ]Plg(S741C-fluorescein). Characterization of the activating cleavage, conformation, and fibrin-binding properties of this variant showed it to be a good model substrate to study plasminogen activation. Both in solution and in a fully polymerized fibrin clot, the activation of the single chain zymogen to the two-chain "plasmin" molecule was accompanied by an approximate 50% quench of the fluorescence intensity of the fluorescein reporter group (40% for the Lys 78 form), indicating a substantial change in the micro-environment of this probe upon occurrence of the activation cleavage. Results for active proteases, which had been labeled via protein-chemical approaches, had already indicated the sensitivity of this position within the protease domain (9 -11). We show that this approach yields a variant of the zymogen, at high levels of expression, which differs from the native zymogen only by a serine to cysteine substitution. In the case of plasminogen, we did not find indications of malfolding as a result of this change, since production levels in BHK cells as well as other properties of this variant were identical to those obtained for "wild-type" plasminogen (19).
Recently Bock and co-workers (9) described the production of a fluorescent plasminogen derivative similar to that reported here. They treated plasma plasminogen with streptokinase and covalently modified the active site with a thioester chloromethyl ketone. The thioester was subsequently hydrolyzed with hydroxylamine, and the thiol group was covalently modified with an anilinonaphthylsulfonate moiety. The fluorescent derivative was then separated from streptokinase. The active site-modified fluorescent derivative yielded readily measured spectral changes upon interaction with lysine analogues, with streptokinase, and upon cleavage to the plasmin derivative. , including one with HBST only, all yielded an intensity value of 7.5, and this has been subtracted from the data indicated. The initial decrease upon addition of the 10-l aliquot to start the reactions is consistent with dilution. As the subsequent data indicate, fibrin polymerization only marginally affected the initial (control) signal, which was 94.9 at 50 nM fibrin (å) and 90.4 at 3.0 M fibrin (Ç). The lines through the data obtained with t-PA (q, E) are regression lines that resulted upon fitting the experimental points by nonlinear regression to the equation I ϭ I o Ϫ ⌬I max (1 Ϫ exp(Ϫk⅐t)), where I is the intensity, ⌬I max is the maximal decrease in intensity, k is the first order rate constant, and t is the time. In both instances the ⌬I max was 50% of the initial (control) intensity. B, an example of the use of linear regression to determine the initial rate of cleavage of 1. that the derivative is a good surrogate for unmodified plasminogen whereby the properties of plasminogen can be monitored. The cleaved form of the derivative yielded plasmin activity of about 1.0% of native plasmin. The presently described derivative Plg(S741C-fluorescein) has many of the properties of the derivative described by Bock et al. (9). Its fluorescence properties are sensitive to lysine analogues and cleavage by plasminogen activators. In addition, it appears to be a good derivative for analyzing the interactions and reactions of plasminogen. Unlike the derivative prepared with streptokinase, however, the presently described derivative exhibits no detectable plasmin activity when cleavaged by plasminogen activators, whether or not cysteine 741 is modified with a fluorescent probe. This lack of activity is undoubtedly the result of the cysteine for serine substitution at position 741. Thus, Plg(S741C) appears ideally suited as a tool to produce plasminogen derivatives that can be labeled with thiol specific probes and that generate no plasmin activity upon cleavage by plasminogen activators.
Plg(S741C-fluorescein) enabled us for the first time to do steady state determinations of activation rates in a fully polymerized fibrin clot without solubilization of the clot. Since no active protease was formed, none of the positive feedback loops of fibrinolysis have taken place. The stimulation factors (rate fibrin absent/rate fibrin present) were several hundredfold for both forms of plasminogen. In a very detailed study in which plasminogen activation rates were deduced from rates of release of isotope from 125 I-labeled fibrin films due to generated plasmin, k cat  It has been suggested that the conversion of single-chain t-PA to two-chain t-PA, [Glu 1 ]plasminogen to [Lys 78 ]plasminogen, and intact fibrin to proteolytically modified fibrin, which are all plasmin-catalyzed, might result in substantial acceleration of the rate of plasminogen activation. The kinetic parameters presented in the present study, however, are very similar to those obtained in assays in which all of the mentioned feedback loops do occur. Hence, our results indicate that the effects of the proposed feedback loops in a polymerized fibrin clot might be limited under non-pathological conditions. Differences are restricted to decreased K m values for plasminogen activation, being 0.52 M ([Glu 1 ]plasminogen, intact fibrin, single chain t-PA; this study) to minimally 0.02 M ([Lys 78 ]plasminogen, two chain t-PA, plasmin-modified fibrin; Ref. 2). At physiological levels of fibrin (9 M) and plasminogen (1.5 M), this will result in a 25% increase in rate of plasminogen activation.
Adverse conditions like clot retraction, however, might lead to plasminogen depletion, and therefore the decrease in K m could have a more dramatic effect.
In conclusion, the approach to label a zymogen at serine of the corresponding active site of the enzyme via mutation to cysteine and incorporation of a fluorescent, cysteine-specific probe might have wide applicability, since it enables steady state activation studies through a readily measured signal. In addition, it simplifies the interpretation of kinetics in cases where the product (protease) has a great influence on its own formation via positive or negative feedback loops.