Staphylokinase as a Plasminogen Activator Component in ...

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Feb. 1999, p. 506–513 0099-2240/99/$04.0010 Copyright © 1999, American Society for Microbiology. All Rights Reserved.

Vol. 65, No. 2

Staphylokinase as a Plasminogen Activator Component in Recombinant Fusion Proteins S. J. SZARKA,1 E. G. SIHOTA,1 H. R. HABIBI,2

AND

S.-L. WONG1*

Department of Biological Sciences, Division of Cellular, Molecular, and Microbial Biology,1 and Division of Zoology,2 University of Calgary, Calgary, Alberta, Canada T2N 1N4 Received 4 May 1998/Accepted 28 October 1998

The plasminogen activator staphylokinase (SAK) is a promising thrombolytic agent for treatment of myocardial infarction. It can specifically stimulate the thrombolysis of both erythrocyte-rich and platelet-rich clots. However, SAK lacks fibrin-binding and thrombin inhibitor activities, two functions which would supplement and potentially improve its thrombolytic potency. Creating a recombinant fusion protein is one approach for combining protein domains with complementary functions. To evaluate SAK for use in a translational fusion protein, both N- and C-terminal fusions to SAK were constructed by using hirudin as a fusion partner. Recombinant fusion proteins were secreted from Bacillus subtilis and purified from culture supernatants. The rate of plasminogen activation by SAK was not altered by the presence of an additional N- or C-terminal protein sequence. However, cleavage at N-terminal lysines within SAK rendered the N-terminal fusion unstable in the presence of plasmin. The results of site-directed mutagenesis of lysine 10 and lysine 11 in SAK suggested that a plasmin-resistant variant cannot be created without interfering with the plasmin processing necessary for activation of SAK. Although putative plasmin cleavage sites are located at the C-terminal end of SAK at lysine 135 and lysine 136, these sites were resistant to plasmin cleavage in vitro. Therefore, C-terminal fusions represent stable configurations for developing improved thrombolytic agents based on SAK as the plasminogen activator component. derivative (scFv) of fibrin-specific monoclonal antibody MA15C5 resulted in a bifunctional product that exhibited a 13-fold increase in thrombolytic potency in vitro (19). Using the latter approach overcomes the heterogeneity problem and the need for postpurification processing inherent to chemical cross-linking. A second factor that limits effective thrombolytic treatment is the elevated clot-forming activity of thrombin observed in patients (24). Active clotting antagonizes the thrombolytic effort and can result in reocclusion even after successful degradation of the initial clot (16, 25). The strategies used to limit clot reformation include blocking the key protease thrombin by using specific inhibitors, such as heparin or hirudin. SAK is a recently rediscovered plasminogen activator (9) that is isolated from bacteriophage (1, 29) or lysogenic strains (11) of Staphylococcus aureus. SAK is a plasminogen activator that exhibits many desirable thrombolytic properties in vivo. Limited clinical trials have shown that SAK is as effective as t-PA at achieving early perfusion in myocardial infarction patients (62 and 57%, respectively [34]). In addition, SAK exhibits better fibrin specificity than t-PA (without having direct affinity for fibrin) and is capable of dissolving platelet-rich clots (8, 21). These natural properties support the conclusion that SAK should be used as a thrombolytic agent for clinical treatment. Since early reperfusion is not obtained in 38% of treated patients, the thrombolytic effectiveness of SAK can still be improved. SAK lacks any affinity for fibrin and the ability to inhibit thrombin, two functions which would supplement and potentially improve its thrombolytic potency. Both of these functions can be combined with SAK by constructing translational fusions with fibrin-binding or anti-thrombin protein domains. There have been no previous reports describing SAK in the context of a fusion protein. Since protein extensions to u-PA affected its activity (38) and lysine 10 within SAK has been shown to be sensitive to plasmin cleavage (10, 33), it was necessary to first evaluate the utility of SAK as a fusion component. To do this, we constructed model fusion proteins by

As a clinical treatment for myocardial infarction, activating endogenous plasminogen effectively dissolves pathologic clots and saves patient lives. Plasminogen activators, identified in and cloned from human sources (tissue plasminogen activator [t-PA] and urokinase-like plasminogen activator [u-PA]) and bacterial sources (streptokinase and staphylokinase [SAK]), can catalyze the conversion of circulating plasminogen to the active protease plasmin (14, 35). The plasminogen activators currently used for clinical thrombolytic treatment include bacterial streptokinase and human u-PA and t-PA (7). One detrimental property of these agents, at pharmacological doses, is the fact that they lack fibrin specificity or exhibit only partial fibrin specificity. Consequently, the side effects of treatment involve systemic activation of plasminogen, which results in proteolytic degradation of several plasma proteins. Reductions in blood fibrinogen and other clotting components expose patients to a serious risk of secondary hemorrhaging complications. The clot specificity of thrombolytic proteins can be enhanced by chemical cross-linkage to an anti-fibrin-specific monoclonal antibody, 59D8. Acquired fibrin affinity has resulted in 10- to 29-fold increases in clot-localized activity for both thrombolytic proteins (t-PA [28] and u-PA [3]) and an anti-thrombin protein (hirudin [2]). These results demonstrate that engineering fibrin affinity can improve the effectiveness of thrombolytic agents. Since chemical cross-linking creates a heterogeneous mixture of products, this approach is unsuitable for pharmacological agents used in human patients. In addition, chemically cross-linking two proteins may result in a reduction in the specific activity of one or both of the protein subunits (26). A translational fusion between u-PA and a single-chain * Corresponding author. Mailing address: Department of Biological Sciences, Division of Cellular, Molecular and Microbial Biology, University of Calgary, 2500 University Drive, N.W., Calgary, Alberta, Canada T2N 1N4. Phone: (403) 220-5721. Fax: (403) 289-9311. E-mail: slwong@ucalgary.ca. 506

VOL. 65, 1999

SAK AS A FUSION PROTEIN

507

FIG. 1. (A) Amino acid substitutions in the SAK variants constructed. Only the amino acids that are different from the amino acids in the SAK 42D (13) sequence in panel B are indicated; the remaining amino acids are the same as the amino acids in SAK 42D. The region deleted from DN10SAK is indicated by a line. (B) Components of expression-secretion plasmid pSAKP. The amino acids boardering the SacB signal sequence and SAK junction are indicated, as are the primer sites used for PCR amplification of the different SAK site-directed variants. Lysine residues described in the text are numbered. The signal peptidase cleavage site is indicated by the open arrow. The primary plasmin cleavage site in SAK is indicated by the solid arrow.

joining SAK to the small functional domain hirudin variant 1 (HV1) in both N- and C-terminal configurations. These proteins were produced by using an engineered Bacillus subtilis expression system (41) via secretion. Hirudin was selected as a fusion partner because of its compatibility with the secretion system. To develop successful SAK fusions, the SAK domain in the proteins should retain full biological activity and the fusion proteins should be resistant to plasmin digestion. Therefore, these two criteria were used to evaluate the fusion proteins obtained. In addition, SAK variants with substitutions of the N-terminal lysines were constructed and characterized in order to create a plasmin-resistant form of SAK suitable for Nterminal fusion. MATERIALS AND METHODS

FIG. 2. Flowcharts for construction of the HV1-SAK (A) and SAK-HV1 (B) expression vectors. The restriction enzyme sites used in construction (see text) are indicated. P43, P43 promoter; SacB SP, levasucrase signal peptide, including a ribosomal binding-translation initiation site; linker, synthetic linker peptide.

Construction of SAK expression vectors. A seven-protease-deficient strain, B. subtilis WB700 (trpC2 nprE aprE epr bpf Dmpr::ble DnprB::bsr Dvpr::ery), was used as an expression host for production and secretion of recombinant proteins (43). SAK 42D (1) was produced by using plasmid pSAKP (Fig. 1B) (42). Additional SAK sequence variants were generated by incorporating different site-directed nucleotide changes into primers used for PCR amplification of SAK (Table 1 and Fig. 1). To replace the lysine residue occupying the 10th or 11th position (Lys-10 or Lys-11) with one of four selected amino acids, forward primer sets A/E10SAKF, H/R10SAKF, A/E11SAKF, and H/R11SAKF were designed with twofold redundancy. Mixed codons within each redundant primer set (G[A, C]A or C[A, G]T) were used to encode either glutamic acid, alanine, histidine, or arginine instead of lysine. The same strategy was used to create a triple-alanine substitution at Lys-6, Lys-8, and Lys-10 (A6A8A10SAK) and a deletion variant lacking the first 10 amino acids of mature SAK (DN10SAK). The primers were also designed to facilitate cloning of the SAK PCR products as in-frame translational fusions with the modified levansucrase secretion signal peptide (sacB signal peptide) (39). Processing at the signal peptidase cleavage site allowed production of secreted SAK proteins with no additional N-terminal amino acids. The reverse primer SAKR engineered a PstI site 140 bp downstream of the SAK translation termination codon. PCR amplification with Taq polymerase (Pharmacia, Baie d’Urfe´, Quebec, Canada) or Vent polymerase (New England Biolabs, Mississauga, Ontario, Canada) was performed by using the manufacturers’ instructions. The amplification conditions were as follows: 30 cycles consisting of 94°C for 1 min, 52°C for 1 min, and 72°C for 1 min. The amplified SAK sequences were purified, digested with HindIII and PstI, and cloned into pWB705HM. pWB705HM is a variant of pWB705 (41) in which the second downstream HindIII restriction enzyme site has been removed by an end fill and religation treatment. pWB705HM contains the P43 promoter (36) and a modified sacB signal sequence (39) cloned into pUB18 (40), a derivative of pUB110 (23). The final clones were confirmed by DNA sequence determination.

To create the HV1-SAK expression plasmid pBsHV1SAK, the cassette from pSAKP (P43 promoter, sacB signal peptide, SAK) was subcloned as a EcoRISphI fragment into a derivative of the Escherichia coli vector pUC18 (in which the HindIII site had been destroyed by an end fill reaction) to generate pEcSAKP (Fig. 2A). A DNA fragment encoding the HV1 linker peptide was created by PCR amplification of a synthetic hirudin sequence (catalog no. BBG 45; R & D Systems, Minneapolis, Minn.) with primers HV1SAKF and HV1SAKR. Primer HV1SAKF modified the DNA sequence upstream of the first codon in hirudin in order to facilitate proper fusion to the sacB secretion signal sequence. Primer HV1SAKR extended HV1 with a synthetic peptide linker-encoding sequence, creating an in-frame fusion between the two domains. The HV1 linker PCR fragment was digested with HindIII and XhoI and inserted between the sacB signal peptide and SAK sequences within pEcSAKP to create pEcHV1SAK. The entire HV1-SAK expression cassette was subcloned by using EcoRI and SphI from pEcHV1SAK into pWB705HM to generate pBsHV1SAK for expression in B. subtilis. The SAK-HV1 expression plasmid pBsSAKHV1 was constructed by first incorporating a linker peptide-encoding sequence upstream of the translation termination codon of SAK (Fig. 2B). To do this, primers SAKHV1R and SAKHV1F were used to amplify the entire pSAKP plasmid by inverse PCR (37). Then, the synthetic hirudin sequence (R&D Systems) was used as a template for PCR amplification with primers HV1F and M13-20. The amplified HV1 sequence was digested with BamHI and ligated to the inverse PCR product cut with BamHI and BclI. The final constructs were confirmed by DNA sequencing. Purification of SAK proteins. For protein purification, B. subtilis strains were grown in 0.5-liter batch cultures in superrich medium (18) supplemented with 10 mg of kanamycin per liter at 37°C for 5.5 h; the cultures were shaken at 300 rpm. Bacterial cells were removed from each culture by centrifugation in a Sorvall type GSA rotor at 9,200 3 g for 20 min at 4°C. The culture supernatant was collected,

508

SZARKA ET AL.

APPL. ENVIRON. MICROBIOL.

TABLE 1. Primer sequences used for fusion protein construction and site-directed mutagenesisa

a All primer sequences are 59 to 39. M 5 A or C; R 5 A or G. Nucleotides that are different from the nucleotides in the template coding sequence are indicated by double underlining.

and EDTA and phenylmethylsulfonyl fluoride were added to concentrations of 10 and 2 mM, respectively. The supernatant was salt fractionated at 4°C with ammonium sulfate at 45% saturation prior to precipitation of SAK protein at 65% saturation. Salted samples were centrifuged as described above at 19,600 3 g. The final pellets were resuspended in 15 ml of S-200 buffer (10 mM Tris-HCl, 1 mM EDTA, 250 mM NaCl; pH 8.5). Each dissolved sample was on a Sephacryl HR S-200 column (2.5 by 100 cm) at 4°C separated with a flow rate of 24.5 cm h21. Fractions (9 ml) were collected, and aliquots were analyzed to determine yield and purity by protein gel electrophoresis and Coomassie blue staining. The fractions selected were pooled and dialyzed against cation-exchange buffer (1 mM EDTA, 4.8 mM citric acid, 10.4 mM disodium phosphate; pH 5.0) at 4°C prior to separation on a carboxymethyl cellulose (Whatman CM-52; Fisher Scientific, Nepean, Ontario, Canada) column (2.5 by 4.0 cm) (29). Separate CM-52 matrix was used for each SAK variant to prevent cross-contamination of samples. Each bound sample was washed with 3 column volumes of cationexchange buffer and then eluted with 5 volumes of a linear 0 to 0.5 M NaCl gradient in buffer. Fractions containing SAK protein were dialyzed against 0.1 mM EDTA–10 mM Tris-HCl (pH 7.5) at 4°C. The final samples were lyophilized and resuspended in sufficient 10% glycerol to achieve 10-fold concentration prior to storage at 280°C. To purify HV1-SAK and SAK-HV1 fusion proteins, cultures were grown and isolated culture supernatants were salt fractionated as described above. The pellets obtained from ammonium sulfate precipitation were resuspended in 4 ml of S-100 buffer (50 mM acetic acid, 250 mM NaCl; pH 5.0), and the soluble fraction was separated on a Sephacryl S-100 size exclusion column (2.5 by 38 cm) at a linear flow rate of 12.2 cm h21. Fractions (3 ml) were collected and were analyzed by using gel electrophoresis and Coomassie blue staining. Selected fractions were pooled and dialyzed against cation-exchange buffer (50 mM acetic acid, pH 4.5) at 4°C before separation on a Pharmacia Mono S HR 5/5 column with a 0 to 1 M NaCl gradient. Selected Mono S fractions were dialyzed against an anion-exchange buffer (20 mM bis-Tris, pH 5.8) before separation on a Mono Q HR 5/5 column with a 0 to 1 M NaCl gradient. Each purified sample was lyophilized to a volume of 100 ml, and final buffer exchange and size exclusion separation were performed with a Superdex 75 HR 10/30 column equilibrated with 20 mM bis-Tris–20 mM NaCl (pH 6.5). Purified protein fractions were pooled and stored at 280°C. Plasminogen purification. Human plasminogen was isolated from sodium citrate-treated plasma obtained from the Red Cross of Calgary, Canada. As described by Deutsch and Mertz (12), frozen plasma was mixed with an equal volume of phosphate-buffered saline (PBS) (136 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 1.8 mM KH2PO4; pH 7.4), and then diisopropyl fluorophosphate (Sigma, Oakville, Ontario, Canada) was added to a final concentration of 5.5 mM. The plasma solution was stirred at room temperature for 5 h and then dialyzed for 16 h against 4 liters of PBS at 4°C. Plasminogen obtained from the diisopropyl fluorophosphate-treated plasma was affinity purified on a lysine Sepharose (Pharmacia) column (1.5 by 11 cm) and was eluted with 200 mM ε-aminocaproic acid in PBS. The lysine analog was removed from purified plas-

minogen samples by dialysis for 16 h at 4°C against PBS before aliquots were stored at 280°C. N-terminal processing of SAK with plasmin. The method used to process SAK with plasmin was adapted from the method of Ueshima et al. (33). Equimolar amounts of SAK and plasminogen (1.5 mM each) were combined in 100 mM Tris-HCl (pH 7.5) and incubated at 15°C for 20 min. At each time point, a 7.5-ml aliquot was removed, mixed with an equal volume of 23 sample loading buffer, and held at 95°C. Time point and control samples were separated by electrophoresis on a 0.5-mm-thick Tris-Tricine polyacrylamide gel for 5 h at a constant voltage of 10 V cm21. Tricine gel electrophoresis was performed by using the method of Scha¨gger and von Jagow (30), as modified by Chan et al. (4). Processed proteins were transferred to a polyvinylidene difluoride membrane by electroblotting for N-terminal sequencing of individual protein bands as described by Matsudaira (22). The HV1-SAK and SAK-HV1 processing reactions were performed at 37°C. The processed intermediates were separated on a sodium dodecyl sulfate–12% polyacrylamide gel and transferred to a polyvinylidene difluoride membrane for sequencing (22). N-terminal sequencing was performed at the protein microsequencing laboratory at the University of Victoria. Processed products were also detected by Western blotting with polyclonal antibodies against SAK (42), anti-hirudin polyclonal antibodies (Cedarlane Laboratories, Hornby, Ontario, Canada), and monoclonal antibody MATG 108 against the C terminus of hirudin (20). Plasminogen activation reaction. To monitor the rate of activation of plasminogen by SAK (32), reaction mixtures containing 1 mM plasminogen, 5 nM SAK (including variant and fusion proteins), and 100 mM Tris-HCl (pH 7.5) were incubated at 37°C. At various times 5-ml aliquots were removed and mixed with 35 ml of stop buffer (700 mM NaCl, 100 mM Tris-HCl; pH 7.5). When all of the aliquots had been collected, 10 ml of the chromogenic substrate N-p-tosylGly-Pro-Lys nitroanilide (Sigma) was added to a final concentration of 1 mM. Plasmin cleavage of the substrate was monitored with a Ceres model UV900HDi plate reader (Bio-Tek Instruments Inc., Winooski, Vt.) by determining the increase in absorbance at 405 nm over a 10-min period at 37°C. Plasmin activity (the change in absorbance at 405 nm per minute) was plotted versus the activation time (in minutes), and linear rates of plasmin generation relative to the SAK control were calculated. The means and standard deviations were determined from triplicate assays.

RESULTS Activity and processing of HV1-SAK fusion protein. An Nterminal fusion to SAK (HV1-SAK) containing 79 additional amino acids was constructed by using the HV1 protein (13) and a joining linker peptide sequence (Fig. 3A). N-terminal sequencing of the first five amino acids (VVYTD) of the pu-

VOL. 65, 1999


509

FIG. 3. (A) Amino acids at boundaries and junctions for each protein domain in HV1-SAK. The HV1 (13) and SAK 42D (1) protein sequences have been published previously. The arrow indicates the plasmin processing site between Lys-10 and Lys-11 in SAK. (B) Coomassie blue-stained protein gel containing HV1-SAK samples which were processed by plasmin at various reaction times (0 to 32 min) and comparative protein samples, including plasminogen (Plgn), HV1-SAK, DN10SAK, and SAK. Samples were prepared in a reducing buffer. (C) Anti-SAK Western blot detection of processed HV1-SAK, DN10SAK, and SAK.

rified protein confirmed that the signal sequence was removed correctly from this fusion protein by the B. subtilis signal peptidase. The rate of plasminogen activation by HV1-SAK was 106.75% 6 8.9% of the rate of plasminogen activation by SAK alone. To determine the stability of the fusion protein in the presence of plasmin, the integrity of purified HV1-SAK was monitored in a reaction mixture containing equimolar plasminogen at 37°C for 32 min (Fig. 3B). Rapid (,2-min) and complete processing of the 30-kDa HV1-SAK fusion protein occurred concomitantly with the activation of plasminogen to plasmin. Following cleavage, a 16-kDa protein comigrated with the DN10SAK protein (a recombinant SAK which lacked the first 10 N-terminal amino acids and was equivalent to the plasmin-processed form of SAK), and N-terminal sequencing of the first five amino acids (KGDDA) of this protein identified the protein as processed SAK. Anti-SAK antibodies confirmed that the 16-kDa SAK domain was separated from the 30-kDa fusion protein (Fig. 3C). The Western blot results also verified that the 29-kDa protein (reaction time, $2 min) was not related to the original 30-kDa HV1-SAK fusion (reaction time, 0 min). The 29-kDa protein likely represents the B chain of plasmin as its appearance correlated with the conversion of plasminogen to plasmin when preparations were analyzed under reducing conditions. Activity and processing of engineered SAK variants. To engineer plasmin resistance into the HV1-SAK fusion, poten-

tial plasmin cleavage sites within SAK were altered by sitedirected mutagenesis. Since plasmin processing has been reported only for the N-terminal region of SAK, we produced sequence variants which had amino acid substitutions in one or more of the N-terminal lysine residues. These variants were analyzed in a nonfusion context in order to establish whether cleavage of SAK by plasmin could be prevented while the specific activity of SAK as a plasminogen activator was maintained. The initial SAK variant kept Lys-11 (previously shown to be important by Gase et al. [15]), while Lys-6, Lys-8, and Lys-10 were changed to alanines (A6A8A10SAK) (Fig. 1A). The alanine substitutions in A6A8A10SAK delayed processing, and they also reduced the activity of this variant compared to both SAK and DN10SAK (data not shown). Since Lys-11 was present in both DN10SAK and A6A8A10SAK, the differences in structure and activity implied that cleavage after Lys-10 by plasmin was necessary for activity. A systematic analysis in which we examined plasmin processing at position 10 and the specificity for lysine at position 11 was carried out by using a series of amino acid substitutions representing neutral (Ala), positively charged (His or Arg), and negatively charged (Glu) side groups at either Lys-10 or Lys-11 (Fig. 1A). Processing of SAK variants with substitutions at Lys-10 or Lys-11. Each of the eight SAK variants was purified to homogeneity for analysis and comparison to SAK and DN10SAK. Due to the rapid processing of SAK by plasmin at 37°C, the

510

SZARKA ET AL.

FIG. 4. Coomassie blue-stained protein gels showing plasmin processing of SAK and SAK variants after reaction for 0 to 20 min. Lanes contained the following protein samples: plasminogen (lane Plgn), the SAK variant (lane SAKp), and DN10SAK (lane DN10SAK). Processing of the wild-type SAK is shown in the top gel. The SAK variants used in this study are indicated beside the corresponding gels.

reactions were performed at 15°C in order to isolate and identify processed intermediates. In reaction mixtures containing an equimolar amount of plasminogen, the SAK variants converted plasminogen to plasmin at rates comparable to the SAK rates. With the SAK control, plasmin was first observed between 4 and 6 min (Fig. 4). In comparison, plasmin was detected by 2 min for variant R11SAK and by 4 min for variants DN10SAK, A11SAK, E11SAK, H11SAK, or A10SAK. With sequence variants E10SAK, H10SAK, and R10SAK plasmin was observed by 6 min. Except for the DN10SAK reaction, the appearance of plasmin in each of the reactions directly correlated with the appearance of lower-molecular-weight forms of SAK. Tricine-polyacrylamide gel electrophoresis allowed separation of full-length SAK from intermediate and final processed forms. Figure 4 shows that like SAK, the position 11 mutants (A11SAK, E11SAK, H11SAK, and R11SAK) were converted directly to the final processed form, indicating that processing occurred at the most sensitive (i.e., primary) cleav-


age site. Amino-terminal protein sequencing of processed A11SAK confirmed that cleavage occurred following Lys-10, the site previously identified for SAK cleavage (33). The processing patterns obtained for position 10 variants were more complex. Like the position 11 variants, R10SAK also exhibited direct conversion from the full-length form to the final processed form. N-terminal amino acid determination confirmed that cleavage occurred after Arg-10 and yielded DN10SAK with Lys-11 at the new N terminus. Therefore, Arg-10 can functionally replace Lys-10 in SAK processing by plasmin, which is consistent with the trypsinlike protease activity of plasmin (17). H10SAK and E10SAK had transient intermediate-molecular-weight forms, which were fully processed by the final time point. For E10SAK, N-terminal sequencing performed with the gel-purified intermediate and final processed forms confirmed that substitution of glutamic acid at Lys-10 resulted in plasmin cleavage initially after Lys-6 (DN6SAK; intermediate band at 6 min) and then after Lys-11 (DN11SAK; final band at 25 min). A10SAK was processed more slowly, and both intermediate and final forms were still present after 20 min. N-terminal sequencing of the intermediate form obtained at 8 min revealed a mixture of amino acids representing both DN6SAK and DN8SAK forms. The final processed form obtained after an extended 60-min reaction was confirmed to have the DN11SAK sequence. Activities of SAK variants with substitutions at Lys-10 or Lys-11. The rate of plasminogen activation for each of the sequence variants was determined under excess-plasminogen conditions. The SAK, DN10SAK, R11SAK, and R10SAK reactions began with a linear increase in plasmin activation for the initial 8 min. The relative rates of plasminogen conversion compared to the SAK control value (100% 6 9.6%) were 127% 6 6.4%, 172% 6 12.4%, and 234% 6 61.4% for DN10SAK, R11SAK, and R10SAK, respectively. A10SAK had slower kinetics, with a lag period of 16 min before the maximum activation rate (71.6% 6 7.4%) was reached. The remaining SAK variants (E10SAK, H10SAK, A11SAK, E11SAK, and H11SAK) exhibited slow, extensive lag periods and did not achieve a linear activation rate after incubation for 60 min. Although the activation rates were too low to be calculated, the qualitative results indicated that there was significant decline in SAK activity with the position 11 changes. Specific activity and plasmin processing of SAK-HV1 fusion protein. To evaluate the alternative orientation, a SAK-HV1 fusion (Fig. 5A) was constructed and purified. N-terminal sequencing of the first five amino acids (SSSFD) of this protein confirmed that the signal sequence was properly removed. The purified fusion was assayed for SAK specific activity. Like HV1-SAK, SAK-HV1 had a relative plasminogen activation rate (115.4% 6 6.2%) equivalent to the rate of SAK alone. In addition to the processing site between Lys-10 and Lys-11, SAK also contains a lysine doublet at the C terminus, Lys-135 and Lys-136. Although cleavage after Lys-135 or Lys-136 by plasmin has not been reported, the close proximity of these sites to the C-terminal end may have precluded detection. To monitor the plasmin sensitivity of C-terminal lysine residues in the SAK-HV1 fusion, purified protein was incubated with an equimolar amount of plasminogen, and the preparation was monitored for 32 min (Fig. 5B). To determine the extent of plasmin processing under physiologically relevant conditions, both fusion protein reactions were conducted at 37°C. Due to comigration of the plasmin B chain and the SAK-HV1 fusion protein on a reducing gel (Fig. 5C), separation was performed on a nonreducing gel to maintain the covalent disulfide linkage between plasmin A and B chains and thereby allow the two peptides to be resolved together as a higher-molecular-weight

VOL. 65, 1999


511

FIG. 5. (A) Amino acids at the boundaries and junctions for the protein domains in SAK-HV1. The solid arrow indicates the primary plasmin processing site between Lys-10 and Lys-11 in SAK. The open arrows indicate the Lys-135 and Lys-136 residues that were tested for plasmin sensitivity. (B and C) Coomassie blue-stained protein gels containing HV1-SAK processed by plasmin for 0 to 32 min. The proteins used for comparison were plasminogen (Plgn), HV1-SAK, DN10SAK, and SAK. Samples were prepared under nonreducing (B) and reducing (C) buffer conditions. (D) Samples representing the 32-min time point from panel B were detected with anti-SAK polyclonal antibodies (a-SAK), anti-hirudin polyclonal antibodies (a-Hir), and anti-hirudin C-terminal specific monoclonal antibody MAT 108 (a-C-Hir).

complex separate from the SAK-HV1 fusion protein (Fig. 5B). Like HV1-SAK plasminogen activation (Fig. 3B), plasminogen activation by SAK-HV1 was evident after 2 min (Fig. 5C). Since processing after Lys-10 was expected to occur in SAKHV1, additional cleavage after Lys-135 or Lys-136 should have liberated a SAK domain with a molecular weight equivalent to that of DN10SAK. No processed fusion product comigrated near DN10SAK (Fig. 5B), and therefore no cleavage occurred at Lys-135 or Lys-136. However, partial cleavage of SAK-HV1 yielded a protein product whose apparent molecular mass was 4 to 5 kDa less than the molecular mass of the original fusion protein. Western blot detection was performed with the 32-min reaction products by using polyclonal anti-SAK antibodies, polyclonal anti-hirudin antibodies, or monoclonal anti-hirudin antibodies specific for the C terminus. The antibody patterns confirmed that the degraded fusion product lacked the C-terminal end of hirudin due to cleavage within HV1 (Fig. 5D). To strengthen this conclusion, N-terminal sequences of the two processed SAK-HV1 intermediates prepared after a 32-min processing reaction were determined. Both intermediates had a sequence

(KGDDA) identical to that of DN10SAK. Therefore, the SAKHV1 processing profile which we obtained supported the conclusion that residues Lys-135 and Lys-136 in the SAK-HV1 fusion protein are resistant to cleavage by plasmin for more than 32 min at 37°C. DISCUSSION We used specific activity and resistance to plasmin-mediated cleavage in vitro as criteria to evaluate the utility of SAK as the plasminogen activator component in a fusion protein. Two model fusion proteins, HV1-SAK and SAK-HV1, both converted plasminogen to plasmin at rates comparable to the rate obtained with SAK alone. The normal level of SAK activity observed with HV1-SAK can be explained by rapid processing and conversion to DN10SAK. In contrast, the C-terminal fusion SAK-HV1 was stable in the presence of plasmin. This indicates that the C-terminal lysine residues in SAK are resistant to plasmin cleavage. The recently reported structure of SAK shows that both the N terminus and the C terminus are accessible and directed away from the protein core (27). This ar-

512

SZARKA ET AL.

rangement may explain why extending the protein sequence is tolerated and results in no or minimal disruption of the structure and function of SAK. Although plasmin cleavage of SAK at Lys-10 (33) and the requirement for Lys-11 (15) have been observed previously, it was not clear whether proteolysis was necessary or occurred only because of the susceptibility of Lys-10 (29) within the unstructured and exposed N-terminal domain (27). Our analysis indicated that full activation of SAK occurred through essential plasmin processing at Lys-10 which made Lys-11 the new N terminus. Compared to SAK, activity is reduced in SAK variants which are defective in correct processing or lack a new N-terminal lysine. In addition, the positively charged side chain in arginine can functionally substitute for lysine at the position 10 cleavage site or can fulfill the N-terminal requirement. Schlott et al. (31) used single end point processing analysis to identify the final processed forms for a set of SAK variants, including R10SAK, H10SAK, R11SAK, H11SAK, and E11SAK. The results of our SAK variant analysis and the results of Schlott et al. (31) are complementary and lead to the conclusion that active SAK requires an N-terminal proximate lysine. In addition, unique subsets of SAK variants in each report contributed to the total number of different amino acid substitutions tested at both position 10 and position 11 in SAK. In contrast, we were able to identify individual processing intermediates by combining a slower reaction rate (15°C) and a time course study with high-resolution gel electrophoresis. These intermediates were used to identify and define different plasmin sensitivities for the N-terminal lysines in SAK. The DN10SAK variant which we produced in B. subtilis was correctly processed by the signal peptidase during secretion and exhibited activity comparable to that of full-length SAK. The DN10SAK produced as a cytoplasmic protein in E. coli by Schlott et al. (31) retained the initiation methionine and thus was a DN9M10SAK variant. Schlott et al. (31) reported that the DN9M10SAK protein had activity comparable to that of SAK, suggesting that the N-terminal lysine can accommodate an extra amino acid upstream without a reduction in activity. Although the requirement for an N-terminal lysine has been independently confirmed, the functional role and precise proximal requirements still need to be defined. This can be achieved by measuring the activity of SAK variants designed to test the lysine position relative to the N terminus. Our time course study of plasmin processing by position 10 variants verified that Lys-10 is the preferred plasmin cleavage site and that once the molecule is cleaved, the remaining Lys-11 is stable. In the absence of a plasmin substrate at position 10, plasmin cleaves after the secondary sites at Lys-6, Lys-8, and Lys-11, and there is eventual processing to the smallest stable form, DN11SAK. Therefore, removal of the Lys-10 processing site is not sufficient to make SAK resistant to plasmin, and removal of all four N-terminal lysines should inactivate SAK. Consequently, creation of a plasmin-resistant SAK variant for construction of N-terminal fusions does not appear to be possible without directly compromising SAK activity. Our results and the results of other workers (31) show that the requirement for an N-terminal lysine in SAK depends on the relative amounts of plasminogen and SAK activator. Under physiologically relevant conditions when the activator is limiting, the activation rates of SAK variants with nonconservative substitutions of Lys-11 (A11SAK, E11SAK, and H11SAK) are significantly reduced. Models that describe SAK activation of plasminogen indicate that SAK-plasmin complexes are capable of cleaving both free and SAK-bound plasminogen (9). Gase et al. (15) have suggested that DN10SAK-plasmin, but not


SAK-plasmin, can activate plasminogen directly. Therefore, lysine at the new N terminus of DN10SAK may be necessary for plasmin to utilize plasminogen as a substrate. In contrast, conversion of plasminogen to plasmin in reaction mixtures containing equimolar amounts of plasminogen and activator shows little dependence on the N-terminal amino acid of SAK. Under equimolar conditions, SAK-bound plasminogen should predominate over free plasminogen. Since SAK-bound plasminogen can be used as a substrate by both processed SAK-plasmin and unprocessed SAK-plasmin complexes, SAK position 10 or 11 variants would be expected to activate plasminogen at rates similar to the SAK rate under equimolar conditions. For the N-terminal fusion protein HV1-SAK, the rapid separation of the HV1 and SAK domains by plasmin in vitro suggests that the function of the fusion protein (i.e., colocalization of the two activities) is not realized in vivo during exposure to plasmin. For this reason, N-terminal fusions to SAK are not useful for designing improved thrombolytic agents for use in vivo. Testing SAK-HV1 for plasmin processing confirmed the resistance of Lys-135 and Lys-136 to plasmin cleavage in vitro. Sako (29) observed a loss of activity after substitution in the C-terminal region of SAK. Also, Gase et al. (15) have shown that deletion of one or both of the C-terminal lysines (Lys-135 and Lys-136) eliminates SAK activity. Removal of Lys-135 and/or Lys-136 transformed normally soluble SAK protein into an insoluble protein during E. coli expression, suggesting that the terminal amino acids are involved in protein folding or stabilization of the tertiary structure. Both C-terminal lysine residues were present in the SAK-HV1 fusion. In addition, an 18-amino-acid linker peptide connecting SAK to HV1 was included to minimize interdomain interference during protein folding. We encountered sensitivity of hirudin to plasmin digestion in a portion of the SAK-HV1 protein population. Chatrenet and Chang (6) observed that trypsin sensitivity at Lys-36 in refolded hirudin was proportional to the amount of nonactive hirudin, and Chang observed that native hirudin is resistant to many proteases (5). Based on hirudin disulfide shuffling in vitro (6), the absence of a disulfide isomerase results in 42% of the hirudin being recovered as nonnative isomers. Together, these findings suggest that the observed cleavage of HV1 in the SAK-HV1 population may be due to the accumulation of nonnative conformations as a result of the absence of disulfide isomerase activity in the extracellular environment of B. subtilis. In the process of determining the utility of SAK in recombinant fusion proteins, we characterized the plasmin cleavage site in SAK and its role in SAK activity. Cleavage at Lys-10 is important for creating a new N-terminal residue (Lys-11) with a lysine functional group. The required processing of the SAK N-terminal domain precludes developing a plasmin-resistant form for the construction of N-terminal SAK fusions. Plasmin resistance and wild-type activity indicate that SAK can accommodate additional protein sequences fused to its C-terminal end. These results should allow rational development of improved thrombolytic agents based on SAK. ACKNOWLEDGMENTS We thank Sandy Kielland of the University of Victoria for determining N-terminal sequences of various SAK derivatives and the Red Cross of Canada (Calgary) for supplying plasma. Teaching and research assistantships for E.G.S. from the Department of Biological Sciences of the University of Calgary are greatly appreciated. This work was supported by a research grant from Heart and Stroke Foundation of Canada (Alberta and Northwest Territories) to S.-L.W. S.-L.W. is a senior medical scholar of the Alberta Heritage Foundation for Medical Research.

VOL. 65, 1999

SAK AS A FUSION PROTEIN REFERENCES

1. Behnke, D., and D. Gerlach. 1987. Cloning and expression in Escherichia coli, Bacillus subtilis, and Streptococcus sanguis of a gene for staphylokinase—a bacterial plasminogen activator. Mol. Gen. Genet. 210:528–534. 2. Bode, C., M. Hudelmayer, P. Mehwald, S. Bauer, M. Freitag, E. von Hodenberg, J. B. Newell, W. Kubler, E. Haber, and M. S. Runge. 1994. Fibrintargeted recombinant hirudin inhibits fibrin deposition on experimental clots more efficiently than recombinant hirudin. Circulation 90:1956–1963. 3. Bode, C., M. S. Runge, S. Schonermark, T. Eberle, J. B. Newell, W. Kubler, and E. Haber. 1990. Conjugation to antifibrin Fab9 enhances fibrinolytic potency of single-chain urokinase plasminogen activator. Circulation 81: 1974–1980. 4. Chan, D. W., C. J. Veillette, and S. P. Lees-Miller. Autophosphorylation sites in the human Ku antoantigen. Submitted for publication. 5. Chang, J. Y. 1990. Production, properties, and thrombin inhibitory mechanism of hirudin amino-terminal core fragments. J. Biol. Chem. 265:22159– 22166. 6. Chatrenet, B., and J. Y. Chang. 1992. The folding of hirudin adopts a mechanism of trial and error. J. Biol. Chem. 267:3038–3043. 7. Collen, D. 1997. Thrombolytic therapy. Thromb. Haemostasis 78:742–746. 8. Collen, D., F. De Cock, and J. M. Stassen. 1993. Comparative immunogenicity and thrombolytic properties toward arterial and venous thrombi of streptokinase and recombinant staphylokinase in baboons. Circulation 87: 996–1006. 9. Collen, D., and H. R. Lijnen. 1994. Staphylokinase, a fibrin-specific plasminogen activator with therapeutic potential? Blood 84:680–686. 10. Collen, D., B. Schlott, Y. Engelborghs, B. Van Hoef, M. Hartmann, H. R. Lijnen, and D. Behnke. 1993. On the mechanism of the activation of human plasminogen by recombinant staphylokinase. J. Biol. Chem. 268:8284–8289. 11. Collen, D., Z. A. Zhao, P. Holvoet, and P. Marynen. 1992. Primary structure and gene structure of staphylokinase. Fibrinolysis 6:226–231. 12. Deutsch, D. G., and E. T. Mertz. 1970. Plasminogen: purification from human plasma by affinity chromatography. Science 170:1095–1096. 13. Dodt, J., H.-P. Mu ¨ller, U. Seemu ¨ller, and J.-Y. Chang. 1984. The complete amino acid sequence of hirudin, a thrombin specific inhibitor. FEBS Lett. 165:180–184. 14. Emeis, J. J., J. H. Verheijen, H. K. Ronday, M. P. M. de Maat, and P. Brakman. 1997. Progress in clinical fibrinolysis. Fibrinolysis Proteolysis 11: 67–84. 15. Gase, A., M. Hartmann, K. H. Guhrs, A. Rocker, D. Collen, D. Behnke, and B. Schlott. 1996. Functional significance of NH2- and COOH-terminal regions of staphylokinase in plasminogen activation. Thromb. Haemostasis 76:755–760. 16. Granger, C. B., R. M. Califf, and E. J. Topol. 1992. Thrombolytic therapy for acute myocardial infarction. Drugs 44:293–325. 17. Groskopf, W. R., L. Summaria, and K. C. Robbins. 1969. Studies on the active center of human plasmin. Partial amino acid sequence of a peptide containing the active center serine residue. J. Biol. Chem. 244:3590–3597. 18. Halling, S. M., F. J. Sanchez-Anzaldo, R. Fukuda, R. H. Doi, and C. F. Meares. 1977. Zinc is associated with the b subunit of DNA-dependent RNA polymerase of Bacillus subtilis. Biochemistry 16:2880–2884. 19. Holvoet, P., Y. Laroche, H. R. Lijnen, R. Van Cauwenberge, E. Demarsin, E. Brouwers, G. Matthyssens, and D. Collen. 1991. Characterization of a chimeric plasminogen activator consisting of a single-chain Fv fragment derived from a fibrin fragment D-dimer-specific antibody and a truncated singlechain urokinase. J. Biol. Chem. 266:19717–19724. 20. Koch, C., O. Whitechurch, P. Cordier, and C. Roitsch. 1993. Anti-hirudin monoclonal antibodies directed toward discontinuous epitopes of the hirudin amino-terminal and epitopes involving the carboxy-terminal hirudin amino acids. Anal. Biochem. 214:301–312. 21. Lijnen, H. R., B. Van Hoef, L. Vanderbossche, and D. Collen. 1992. Biochemical properties of natural and recombinant staphylokinase. Fibrinolysis 6:214–225. 22. Matsudaira, P. 1987. Sequence from picomole quantities of proteins electroblotted onto polyvinylidene difluoride membranes. J. Biol. Chem. 262: 10035–10038. 23. McKenzie, T., T. Hoshino, T. Tanaka, and N. Sueoka. 1986. The nucleotide

24.

25.

26.

27. 28. 29. 30. 31. 32.

33. 34.

35. 36. 37. 38. 39. 40. 41.

42. 43.

513

sequence of pUB110: some salient features in relation to replication and its regulation. Plasmid 15:93–103. Merlini, P. A., D. Ardissino, K. A. Bauer, L. Oltrona, A. Spinola, P. Diotallevi, R. D. Rosenberg, and P. M. Mannucci. 1995. Activation of the hemostatic mechanism during thrombolysis in patients with unstable angina pectoris. Blood 86:3327–3332. Ohman, E. M., R. M. Califf, E. J. Topol, R. Candela, C. Abbottsmith, S. Ellis, K. N. Sigmon, D. Kereiakes, B. George, and R. Stack. 1990. Consequences of reocclusion after successful reperfusion therapy in acute myocardial infarction. TAMI Study Group. Circulation 82:781–791. Phaneuf, M. D., C. K. Ozaki, M. T. Johnstone, J. P. Loza, W. C. Quist, and F. W. LoGerfo. 1994. Covalent linkage of streptokinase to recombinant hirudin: a novel thrombolytic agent with antithrombotic properties. Thromb. Haemostasis 71:481–487. Rabijns, A., H. L. De Bondt, and C. De Ranter. 1997. Three-dimensional structure of staphylokinase, a plasminogen activator with therapeutic potential. Nat. Struct. Biol. 4:357–360. Runge, M. S., C. Bode, G. R. Matsueda, and E. Haber. 1987. Antibodyenhanced thrombolysis: targeting of tissue plasminogen activator in vivo. Proc. Natl. Acad. Sci. USA 84:7659–7662. Sako, T. 1985. Overproduction of staphylokinase in Escherichia coli and its characterization. Eur. J. Biochem. 149:557–563. Scha ¨gger, H., and G. von Jagow. 1987. Tricine-sodium dodecyl sulfatepolyacrylamide gel electrophoresis for the separation of proteins in the range from 1 to 100 kDa. Anal. Biochem. 166:368–379. Schlott, B., K.-H. Gu ¨hrs, M. Hartmann, A. Ro ¨cker, and D. Collen. 1997. Staphylokinase requires NH2-terminal proteolysis for plasminogen activation. J. Biol. Chem. 272:6067–6072. Schlott, B., M. Hartmann, K. H. Guhrs, E. Birch-Hirschfeid, H. D. Pohl, S. Vanderschueren, F. Van de Werf, A. Michoel, D. Collen, and D. Behnke. 1994. High yield production and purification of recombinant staphylokinase for thrombolytic therapy. Bio/Technology 12:185–189. Ueshima, S., K. Silence, D. Collen, and H. R. Lijnen. 1993. Molecular conversions of recombinant staphylokinase during plasminogen activation in purified systems and in human plasma. Thromb. Haemostasis 70:495–499. Vanderschueren, S., L. Barrios, P. Kerdsinchai, P. Van den Heuvel, L. Hermans, M. Vrolix, F. De Man, E. Benit, L. Muyldermans, D. Collen, and F. Van de Werf. 1995. A randomized trial of recombinant staphylokinase versus alteplase for coronary artery patency in acute myocardial infarction. Circulation 92:2044–2049. Verstraete, M. 1995. The fibrinolytic system: from petri dishes to genetic engineering. Thromb. Haemostasis 74:25–35. Wang, P.-Z., and R. H. Doi. 1984. Overlapping promoters transcribed by Bacillus subtilis s55 and s37 RNA polymerase holoenzyme during growth and stationary phase. J. Biol. Chem. 259:8619–8625. Williams, J. 1989. Optimized strategies for the polymerase chain reaction. BioTechniques 7:762–769. Wnendt, S., E. Janocha, J. Schneider, and G. J. Steffens. 1996. Construction and structure-activity relationships of chimeric prourokinase derivatives with intrinsic thrombin-inhibitory potential. Protein Eng. 9:213–223. Wong, S.-L. 1989. Development of an inducible and enhancible expression and secretion system in Bacillus subtilis. Gene 83:215–223. Wong, S.-L., L.-F. Wang, and R. H. Doi. 1988. Cloning and nucleotide sequence of senN, a novel ‘Bacillus natto’ gene that regulates expression of extracellular protein genes. J. Gen. Microbiol. 134:3269–3276. Wong, S.-L., X.-C. Wu, L.-P. Yang, S.-C. Ng, and N. Hudson. 1995. Production and purification of antibody using Bacillus subtilis as an expression host, p. 100–107. In H. Y. Wang and T. Imanaka (ed.), Antibody expression and engineering—1995. American Chemical Society, Washington, D.C. Ye, R., J.-H. Kim, B.-G. Kim, S. Szarka, E. Sihota, and S.-L. Wong. 1999. High level production of intact, biologically active staphylokinase from Bacillus subtilis. Biotechnol. Bioeng. 62:87–96. Ye, R., L. P. Yang, and S.-L. Wong. 1996. Construction of protease deficient Bacillus subtilis strains for expression studies: inactivation of seven extracellular proteases and the intracellular LonA protease, p. 160–169. In Proceedings of the International Symposium on Recent Advances in Bioindustry 1996. The Korean Society for Applied Microbiology, Seoul, Korea.