Haemostasis 1991 ;21(suppl 1): 41-48
© 1991 S. Kargcr AG, Basel 0301-0147/91 /0217-0041 $2.75/0
Structure-Function and Refolding Studies of the Thrombin-Specific Inhibitor Hirudin Paul H. Johnson, PingSze, Richard C. Winant, Debra Hudson, Peter Underhill, Jerome B. Lazar, Cris Olsen, Ron Almquist Molecular Biology Department and Bio-Organic Chemistry Laboratory, SRI International, Menlo Park, Calif., USA
Key Words. Recombinant hirudin • Protein folding • Reverse-phase HPLC • Synthetic peptides • Thrombin exosite
Introduction Hirudin is well established as the most potent and specific known inhibitor of throm bin [see review article, ref. 1]. It is an effective anticoagulant and antithrombotic agent in animals and humans [2, 3] and may be par ticularly useful in the clinical treatment of arterial thrombosis and disseminated intra vascular coagulation and in extra- corporeal
blood circulation systems. Hirudin should be more effective than heparin as an adjunct to fibrinolytic therapy in preventing reocclu sion, a condition that may be aggravated by the ability of fibrin and its degradation prod ucts to form a ternary complex with heparin and thrombin, resulting in a large decrease in the sensitivity of thrombin to inhibition by antithrombin III [4]. In addition, hirudin ap pears to have an additional and novel antico
Downloaded by: Thomas Jefferson University Scott Library 147.140.27.100 - 5/30/2018 6:35:33 AM
Abstract. We have developed a novel expression and purification system that yields recombinant desulfo-hirudin (HV-1) with high specific activity (10,000 antithrombin units/mg) and an inhibition constant (Kj) for human a-thrombin of 0.2 pM. Reduced and denatured hirudin rapidly refolds to the native, fully active conformation at high concentra tion (> 5 0 mg/ml) by incubation at pH 10. Analytical gel filtration studies at neutral pH suggest that hirudin is a multimer. Initial binding of hirudin to thrombin appears to be followed by dissociation of the hirudin multimer to give a tight-binding 1:1 hirudimthrombin complex. Thrombin inhibition studies showed that hirudin synthetic peptide fragments 42-65 and 51-65 [but not (Ala22)-6-28, containing two of the three disulfide bonds formed in native hirudin] were similarly effective in inhibiting thrombin cleavage of fibrinogen (IC50 = 4.9 and 6.0 |iM, respectively, at a thrombin concentration of 1 \iM). We conclude that hiru din has unusual structural and refolding properties and that its mechanism of inhibition involves noncovalent interaction with multiple sites on thrombin. The interaction of hirudin (specifically the region of Lys-47) with the basic specificity pocket of thrombin may contrib ute to the binding but is not essential for its inhibitory activity.
42
Johnson/Sze/Winant/Hudson/Underhill/Lazar/Olsen/Almquist
Materials and Methods Protein Preparations Recombinant hirudin (r-hirudin) variant 1 (HV-1) was produced from a synthetic gene using an Escheri chia coli expression system based on the genetic regu latory elements o f the colicin El operon [1, 9]. The protein was purified to homogeneity by anion ex change chromatography and reverse-phase high-pres sure liquid chromatography (RP-HPLC) [10]. Human a-thrombin was generously provided by Dr. John Fenton, New York State Department o f Health. Peptide Synthesis Synthetic peptide fragments from both the aminoand carboxy-terminal regions o f hirudin were pre
pared by solid-phase techniques [11] using a Beckman Model 990 C peptide synthesizer. After synthesis, the peptides were cleaved from the resin using anhydrous hydrogen fluoride in the presence o f 10% p-cresol at - 5 °C for 1 h. The peptides were separated from the various organic side products by washing the resin with ether and then extracting the peptide with 10% acetic acid in water. Hirudin 51-65 (His-Asp-Gly-AspPhe-Glu-Glu-Ile-Pro-Glu-Glu-Tyr-Leu-Gln) and hiru din 42-65 (Gly-Glu-Gly-Thr-Pro-Lys-Pro-Gln-SerHis-Asn-Asp-Gly-Asp-Phe-Glu-Glu-Ile-Pro-Glu-GluTyr-Leu-Gln) were prepared on Boc-Gln-O-resin. Fol lowing synthesis, removal o f the protecting groups, and cleavage o f peptide from the resin, the peptide extracts were lyophilized and purified by preparative RP-HPLC (13-26% acetonitrile/0.1 % trifluoroacetic acid, linear gradient, 4 liters total solvent, 30 ml/min) using Vydac 218TPB1015 C i8 silica. [Ala22]-hirudin 6-28 (Cys-Thr-Glu-Ser-Gly-GlnAsn-Leu-Cys-Leu-Cys-Glu-Gly-Ser-Asn-Val-Ala-GlyGln-Gly-Asn-Lys-Cys) was prepared on Boc-Cys-S-(4methyl benzyl)-0-resin. Cys16 was protected as the 4methyl benzyl derivative, and Cys6 and Cys14 were protected as the acetoamidomethyl derivatives. After synthesis and cleavage with hydrogen fluoride the Cys6, Cys14 acetoamido methy-protected peptide ex tract was diluted with water, adjusted to pH 8.1 with ammonium hydroxide, and shaken slowly for 6 days to air-oxidize Cys16 and Cys28 to form this disulfide bond. The peptide was lyophilized and purified by RP-HPLC as described above except that a 5-17% gradient was used. Product fractions were combined, lyophilized, and dissolved in methanol-water (4:1) and added dropwise over 15 min to a rapidly stirring solution of 0.1 M I2 in methanol. The resulting solu tion was stirred an additional 10 min, and then 0.5 M ascorbic acid in 1.0 M citrate buffer, pH 5.0, was added dropwise until the solution remained colorless. The methanol was evaporated under reduced pres sure and the mixture applied to XAD2 resin (Alltech Associates). The resin was washed with water, and the peptide was eluted with methanol. The methanol was evaporated, and the peptide (containing two disulfide bonds at 6-14 and 16-28) was purified by RP-HPLC as described above. Assay Methods The ability o f r-hirudin and hirudin synthetic fragments to inhibit thrombin activity was deter mined using a clotting assay with purified human
Downloaded by: Thomas Jefferson University Scott Library 147.140.27.100 - 5/30/2018 6:35:33 AM
agulant effect: it can displace factor Xa from the vascular endothelium [5, 6] and thereby affect the activation of prothrombin. Although originally isolated from the sali vary gland of the medicinal leech Hirudo medicinalis, hirudin has now been produced in bacteria and yeast by recombinant DNA methods [1]. Nuclear magnetic resonance spectroscopy has provided important struc tural information about hirudin at atomic resolution [7, 8], These developments have enabled the application of various biochemi cal, genetic engineering, and molecular mod eling techniques to understand the structural and functional properties of hirudin and its mechanism of interaction with thrombin. The results reported here provide new infor mation on the structural and refolding prop erties of hirudin. We demonstrate the appli cation of (1) reverse phase-HPLC as an effi cient, high-resolution method for analyzing hirudin conformational differences; (2) im proved techniques using a computer-controlled kinetic microtiter plate reader for analyzing thrombin-hirudin inhibition ki netics, and (3) synthetic peptide fragments for mapping functional domains of hirudin.
Recombinant Hirudin
Protein Folding Analysis Hirudin was reduced and denatured by incuba tion with 0.1 M dithiothreitol in 6 M guanidine hydrochloride (GndHCl) at a pH between 8 and 10
for 1 h at 37 °C or 2 h at room temperature. The solu tion was adjusted to pH 3 and purified by RP-HPLC on a Vydac C4 column using gradient elution (1530%) in 0.065% trifluoroacetic acid and water. Hiru din fractions were lyophilized and stored at - 2 0 °C. Protein folding was initiated by dissolving reduced hirudin in reaction buffer at pH 10 and incubating for various times. The folding reaction was stopped by adjusting the pH to 3, and the reaction mixture was analyzed by RP-HPLC and thrombin inhibition anal ysis as described above. Analytical Gel Filtration Molecular weight (MW) analysis o f hirudin was performed on a Superóse 12 column controlled by an FPLC system (Pharmacia). All experiments per formed under different pH or solvent conditions were independently calibrated with MW markers includ ing: ovotransferrin (77,000), bovine serum albumin (66.000) , ovalbumin (42,000), (3-lactoglobulin A (36.000) , chymotrypsinogen A (25,000), trypsinogen (24.000) , soybean trypsin inhibitor (20,000), a-lactalbumin (14,000), ribonuclease A (13,500), cytochrome C (11,500), aprotinin (6,500), intact insulin (5,650), and insulin B chain (3,400). Hirudin MW was calcu lated from standard curves (log MW as a function of elution position) o f marker proteins determined by linear least squares fit o f the data.
Results and Discussion Thrombin Inhibition by r-Hirudin We have used a novel expression system based on modified genetic regulatory ele ments derived from the colicin El operon [13] to produce high yields of active hirudin HV-1 in E. coli [9]. Thrombin inhibition was measured by two different assay systems em ploying either fibrinogen or a chromogenic peptide as the thrombin substrate. Figure la shows the results of the fibrino gen clotting assay for a typical preparation of r-hirudin in which the calculated IC50 value is 0.5 nM. The IC50 was determined from the linear least squares fit of the data and calcu lated as the hirudin concentration at which
Downloaded by: Thomas Jefferson University Scott Library 147.140.27.100 - 5/30/2018 6:35:33 AM
fibrinogen (Sigma) or using a chromogenic substrate. For both assays we developed the application o f a computer-controlled kinetic microtiter plate reader to process up to 96 reactions simultaneously. For the clotting assay, hirudin was diluted to 50 pi in assay buffer (50 mM Tris-HCl, 120 mAf NaCl, 0.5% PEG 6000, pH 7.4), added to 50 pi of human a-thrombin (0.2 pmol) in assay buffer containing 40 mAf calcium chloride (final thrombin concentration equal to 1 aA/), and incubated for 5 min in a 96-well microtiter plate at room temperature. The reaction was initiated by addition o f 100 pi o f fibrinogen (10 mg/ml) in assay buffer (without PEG) and the solu tion mixed for 10 s. The turbidity o f the reaction mix ture was monitored at 405 nm using the Vma, Kinetic Microplate Reader (Molecular Devices Corporation). Data acquisition and processing were accomplished by a microcomputer interfaced to the microtiter plate reader using software provided by Molecular Devices Corporation. One antithrombin unit (ATU) is the amount o f hirudin that neutralizes one NIH unit o f thrombin at 37 °C, using fibrinogen as substrate. Ac tivity measurements reported here were performed at 37 °C with the chromogenic substrate S-2238 (KabiVitrum) and converted to thrombin activity using 1.25 A4os/min/thrombin unit. Inhibition constants (K¡) were determined in steady-state velocity experiments using S-2238. Reac tions were carried out at room temperature in plastic microtiter plates at a concentration o f 0.4 nAf throm bin in 50 mA/bis-Tris, 0.1 % PEG 6000, 0.1 AfNaCl, 0.5 mg/ml Brij 35 at pH 7.8; reactions were initiated by addition o f S-2238 to a final concentration o f 296 pAf in 0.3 ml. Hirudin was varied over a range that included several concentrations above and below the thrombin concentration. Data points represent the mean values from three separate determinations. Data acquisition was accomplished using the kinetic microtiter plate reader as described above. The soft ware used for tight-binding inhibition analysis was provided by Dr. Stuart Stone (Friedrich Miescher Institute, Basel, Switzerland). In a separate study (data not shown), we demonstrated that initiation o f the reaction by either substrate or thrombin yielded comparable estimates o f K¡.
43
44
Johnson/Sze/Winant/Hudson/Underhill/Lazar/Olsen/Almquist
the initial velocity of the thrombin reaction was inhibited by 50%. The concentration of hirudin was determined by comparison with standards whose concentrations and activi ties had been determined by amino acid composition analysis and chromogenic sub strate assay, respectively. Inhibition con stants were calculated from measurements of steady-state velocities as a function of hiru din concentration fitted by nonlinear regres sion to the initial rate equation of Morrison [12] for tight-binding inhibition kinetics: 2-T -V ,/V 0 = [(Ki + H - T ) 2 + 4-K '-T ]05- K' - H + T,
Protein Folding Although improper folding is a common problem for intracellular production of disulfide-bond-containing proteins in E. coli, we discovered that r-hirudin, which contains three disulfide bonds [1], was recovered in its native conformation and fully active form in extracts following cell disruption. To fur ther characterize the folding properties of hirudin, we have studied the kinetics of fold ing in vitro as a function of protein concen tration and altered solvent conditions. Pro tein folding kinetics can be conveniently measured using RP-HPLC. Figure 2 illus trates the difference in elution properties be tween oxidized (refolded) and reduced rhirudin HV-1. The two forms are reproducibly separated by an elution time difference of 8 min.
Fig. 1. Fibrinogen clotting assay (a) and tightbinding inhibition analysis using a chromogenic-substrate assay (b). Thrombin concentration was 1 \iM in a and 0.4 \iM in b. The curve represents the best fit by nonlinear regression analysis to the initial-rate equa tion.
Figure 3a shows the kinetics of hirudin refolding at a protein concentration of 50 mg/ml (7.2 mM) and molar ratios of hirudin and oxidized:reduced glutathione (2:1) equal to 0.42 and 4.2. In the presence of excess redox agent, greater than 95% of the hirudin folds to its native, fully active conformation in less than 1 h. Reducing glutathione con centration by a factor of 10 reduces the fold ing rate similarly such that the reaction pro ceeds to completion within approximately lOh. Figure 3b demonstrates that in the ab sence of any added oxidation-reduction buff er, hirudin refolds to completion within 24 h at pH 10. The reaction rate is reduced by the
Downloaded by: Thomas Jefferson University Scott Library 147.140.27.100 - 5/30/2018 6:35:33 AM
where H = hirudin and T = thrombin concen trations, Vj and V0 are initial rates of the inhibited and uninhibited reactions, and Ki = Kj7(l+[substrate]/Km). Figure lb shows the results for a standard r-hirudin prepara tion with a calculated Kj = 0.2 pA/and a spe cific activity of 10,000 ATU/mg.
addition of greater than 1 M GndHCl, but even in the presence of 6 M GndHCl more than 25% of the hirudin refolds to the native conformation within 24 h. The ability of hi rudin to refold at extremely high concentra tion even in the absence of an added oxida
tion-reduction agent and the presence of high GndHCl concentrations is unusual for a protein and suggests that hirudin may pos sess some uncommon structural and/or chemical properties. The apparent absence of formation of hirudin folding interme-
Downloaded by: Thomas Jefferson University Scott Library 147.140.27.100 - 5/30/2018 6:35:33 AM
absorbance at 215 nm using two different scales differing 5-fold in sensitivity, a Reduced r-hirudin; elution time was 27 min; dithiothreitol elutes at the solvent front; the arrow indicates the position o f oxidized r-hirudin. b Oxidized r-hirudin; elution time was 19 min; the arrow indicates the position o f reduced r-hirudin.
Johnson/Sze/Winant/Hudson/Underhill/Lazar/Olsen/Almquist
Fig. 3. Refolding kinetics o f r-hirudin. Hirudin was reduced in 0.1 M dithiothreitol and 6 M GndHCl at pH 9.5, dialyzed against 0.1 N acetic acid, lyophilized, and dissolved in 50 mM sodium carbonate/ bicarbonate buffer, 1 mM EDTA, pH 10.0. Protein (50 mg/ml) was refolded by incubation at 25 °C. The extent o f refolding was quantitated by RP-HPLC and thrombin inhibition assays, a Effect o f 3 m M and 30 mM glutathione (2:1 molar ratio o f oxidized and
reduced forms) on refolding kinetics o f 50 mg/ml hirudin, b Influence o f GndHCl concentration on hirudin refolding kinetics in the absence o f gluta thione. Fig. 4. Apparent molecular weight o f r-hirudin determined by analytical gel filtration using a supe róse 12 column. Molecular weight as a function o f pH in 0.5 M NaCl (a) and GndHCl concentration (A/) at pH 7 (b).
diates with nonnative disulfide bonds indi cates a strong tendency of hirudin to assume a native-like conformation even in the pres ence of denaturing solvents.
values at neutral pH up to 16 kD [1], Fig ure 4a shows the results of analytical sizeexclusion chromatography performed as a function of pH in 0.5 MNaCl. The apparent MW varies, from approximately 24 kD at pH 7-10, down to approximately 10 kD at pH 3.0. Figure 4b shows a similar MW tran sition as a function of GndHCl concentra tion; in 6 M GndHCl hirudin has a calcu lated MW corresponding exactly to mono mer size (7 kD). The MW value of r-hirudin
Quaternary Structure Early studies on the hydrodynamic prop erties of natural hirudin, using ultracentrifu gation and gel filtration methods, suggested a variation in apparent MW values from 7 kD (monomer size) at low pH to higher MW
Downloaded by: Thomas Jefferson University Scott Library 147.140.27.100 - 5/30/2018 6:35:33 AM
46
47
Recombinant Hirudin
Mapping Functional Domains with Synthetic Peptides Synthetic peptides provide an important approach to mapping functional domains of hirudin and identifying determinants for thrombin interaction. We chemically synthe sized three fragments corresponding to resi dues 6-28 (with Ala22 replacing Cys22), 4265, and 51-65. For the fragment 6-28 ana logue, Cys-6:Cys-14 and Cys-16:Cys-28 were oxidized to form two of the three disulfide bonds in the corresponding amino-terminal core region of hirudin. Table 1 compares the thrombin inhibi tory properties of the synthetic fragments with those of intact r-hirudin (1-65) using both the fibrinogen-clotting assay and the chromogenic-substrate assay. Fragments 4265 and 51-65 were similarly (although distinguishably) active in inhibiting thrombin cleavage of fibrinogen but were inactive (at 333 \\.M) in inhibiting thrombin hydrolysis
Table 1. Structure-activity correlations o f hirudin and synthetic peptides Peptide
IC50, gAf Fibrinogen
K„ pA/ Substrate S-22383
r-Hirudin (1-65) s-Hirudin (42-65) s-Hirudin (51-65) s-HirudinAla22-(6-28)
0.0005 4.9 6.0
0.2 NA NA
NA
NA
a
NA = No measurable activity at 333 pAf. D-Phe-L-Pipecolyl-Arg-p-nitroanilide.
of a peptidyl p-nitroanilide substrate. The fragment 6-28 analogue was not inhibitory using either substrate at concentrations up to approximately 0.4 mM. There were no syn ergistic effects of the fragment 6-28 ana logue on either of the carboxy-terminal frag ments in inhibiting thrombin (data not shown). Unlike recombinant hirudin modi fied at Lys-47, which shows a 3- to 10-fold increase in the Kj depending on the specific amino acid substitution [1], the presence of Lys-47 did not significantly enhance the ac tivity of the carboxy-terminal synthetic pep tide. Although these results may possibly in dicate that the amino terminal regional of hirudin is necessary for conformational sta bilization of the region in the vicinity of Lys47, the results support the conclusion that the possible interaction of Lys-47 with thrombin does not make a major contribu tion to the binding free energy. The ability of carboxy-terminal hirudin peptides to inhibit thrombin activity using fibrinogen as substrate, but not using a small chromogenic substrate, suggests that the car boxy-terminal domain of hirudin interacts
Downloaded by: Thomas Jefferson University Scott Library 147.140.27.100 - 5/30/2018 6:35:33 AM
at neutral pH is similar (26.5 kD) to that we previously determined using a different chromatographic medium [1], In addition, we have shown that in these two different chromatographic systems, the apparent MW of the thrombin-hirudin complex corre sponds to the sum of the MW of thrombin and hirudin monomer; the complex will form at neutral pH in 1 M, but not in 1.5 M, GndHCl [1, data not shown]. These results further support the conclusion that hirudin may be multimeric in solution but binds to thrombin as a monomer. It must be noted, however, that these MW measurements un der neutral, nondenaturing conditions do not account for possible effects due to the asymmetry of the hirudin molecule, or its possible interaction with the chromato graphic media.
48
Johnson/SzeAVinant/Hudson/Underhill/Lazar/Olsen/Almquist
Acknowledgments We thank Jon Miller and Margaret Fahnestock for their critical review o f this manuscript. This work was supported by SRI Internal Research and Develop ment projects 870D32XJC and 391D32YSB and by funds provided by the Cigarette and Tobacco Surtax Fund o f the State of California through the TobaccoRelated Disease Research Program o f the University o f California.
References 1 Johnson PH, Sze P, Winant R, Payne PW, Lazar JB: Biochemistry and genetic engineering o f hiru din. Semin Thromb Hemost 1989;15:302-315. 2 Markwardt F, Nowak G, Sturzebecher J, Griessbach U, Walsman P, Vogel G: Pharmacokinetics and anticoagulant effect o f hirudin in man. Thromb Haemost (Leipz) 1986;52:150-163. 3 Markwardt F, Fink E, Kaiser B, Klocking HP, Nowak G, Richter M, Sturzebecher J: Pharmaco logical survey of recombinant hirudin. Pharmazie 1988;43:202-207. 4 Hogg PJ, Jackson CM: Fibrin monomer protects thrombin from inactivation by heparin-anti thrombin III: Implications for heparin efficacy. Proc Natl Acad Sei USA 1989;86:3619-3623. 5 Friedberg RC, Hagen P-O, Pizzo SV: The role o f endothelium in factor Xa regulation: The effect o f
6
7
8
9
10
11
12
13
14
15
16
plasma proteinase inhibitors and hirudin. Blood 1988;71:1321-1328. Pizzo SV, Friedberg RC, Sze P, Winant R, Hud son D, Lazar JB, Johnson PH: Recombinant hiru din displaces human factor Xa from its endothe lial binding sites. Thromb Res 1990;57:803-806. Folkers PJM, Clore GM, Driscoll PC, Dodt J, Kohler S, Gronenbom AM: Solution structure o f recombinant hirudin and the Lys-47 -> Glu mu tant: A nuclear magnetic resonance and hybrid distance geometry-dynamical simulated annealing study. Biochemistry 1989;28:2601-2617. Haruyama H, Wuthrich K: Conformation o f re combinant desulfatohirudin in aqueous solution determined by nuclear magnetic resonance spec troscopy. Biochemistry 1989;28:4301-4312. Sze P, Lazar JB, Winant RC, Hudson D, Sohel I, Johnson PH: Cloning and expression o f a syn thetic gene for the thrombin specific inhibitor, hirudin, submitted. Winant RC, Lazar JB, Johnson PH: Chemical modifications and amino acid substitutions in re combinant hirudin that increase hirudin-throm bin affinity. Biochemistry, in press. Erickson BW, Merrifield, RB: Solid phase peptide synthesis; in Neurath H, Hill RL (eds): The Pro teins. New York, Academic Press, 1976, vol 2, pp 255-527. Morrison JF: Kinetics o f the reversible inhibition o f enzyme-catalyzed reactions by tight binding inhibitors. Biochim Biophys Acta 1969; 185:269— 286. Waleh NS, Johnson PH: Structural and functional organization o f the colicin E 1 operon. Proc Natl AcadSci USA 1985;82:8389-8393. Fenton II JW, Olson TA, Zabinski MP, Wilner GD: Anion binding exosite o f human alphathrombin and fibrin(ogen) recognition. Biochem istry 1988;27:7106-7112. Noe G, Hofsteenge J, Rovelli G, Stone SR: The use of sequence specific antibodies to identify a secondary binding site in thrombin. J Biol Chem 1988;263:11729-11735. Lazar JB, Winant RC, Johnson PH: Hirudin: Amino-terminal residues play a major role in the interaction with thrombin. J Biol Chem, in press. Dr. Paul Johnson Molecular Biology Department SRI-20501, 333 Ravenswood Avenue Menlo Park, CA 94025 (USA)
Downloaded by: Thomas Jefferson University Scott Library 147.140.27.100 - 5/30/2018 6:35:33 AM
with a region of thrombin distinct from the catalytic site. This region is likely to be the anion binding exosite, a region rich in basic amino acids that is implicated in fibrinogen binding [14] and contains the p-cleavage site. An antibody to a synthetic peptide repre senting this region is able to block hirudin binding to a-thrombin [15]. On the basis of these results, together with implications from our molecular-modeling studies [ 1, 16], hirudin appears to interact with thrombin at multiple sites including the catalytic site, the apolar binding site, and the anion-binding exosite.
© 1991 S. Kargcr AG, Basel 0301-0147/91 /0217-0041 $2.75/0
Structure-Function and Refolding Studies of the Thrombin-Specific Inhibitor Hirudin Paul H. Johnson, PingSze, Richard C. Winant, Debra Hudson, Peter Underhill, Jerome B. Lazar, Cris Olsen, Ron Almquist Molecular Biology Department and Bio-Organic Chemistry Laboratory, SRI International, Menlo Park, Calif., USA
Key Words. Recombinant hirudin • Protein folding • Reverse-phase HPLC • Synthetic peptides • Thrombin exosite
Introduction Hirudin is well established as the most potent and specific known inhibitor of throm bin [see review article, ref. 1]. It is an effective anticoagulant and antithrombotic agent in animals and humans [2, 3] and may be par ticularly useful in the clinical treatment of arterial thrombosis and disseminated intra vascular coagulation and in extra- corporeal
blood circulation systems. Hirudin should be more effective than heparin as an adjunct to fibrinolytic therapy in preventing reocclu sion, a condition that may be aggravated by the ability of fibrin and its degradation prod ucts to form a ternary complex with heparin and thrombin, resulting in a large decrease in the sensitivity of thrombin to inhibition by antithrombin III [4]. In addition, hirudin ap pears to have an additional and novel antico
Downloaded by: Thomas Jefferson University Scott Library 147.140.27.100 - 5/30/2018 6:35:33 AM
Abstract. We have developed a novel expression and purification system that yields recombinant desulfo-hirudin (HV-1) with high specific activity (10,000 antithrombin units/mg) and an inhibition constant (Kj) for human a-thrombin of 0.2 pM. Reduced and denatured hirudin rapidly refolds to the native, fully active conformation at high concentra tion (> 5 0 mg/ml) by incubation at pH 10. Analytical gel filtration studies at neutral pH suggest that hirudin is a multimer. Initial binding of hirudin to thrombin appears to be followed by dissociation of the hirudin multimer to give a tight-binding 1:1 hirudimthrombin complex. Thrombin inhibition studies showed that hirudin synthetic peptide fragments 42-65 and 51-65 [but not (Ala22)-6-28, containing two of the three disulfide bonds formed in native hirudin] were similarly effective in inhibiting thrombin cleavage of fibrinogen (IC50 = 4.9 and 6.0 |iM, respectively, at a thrombin concentration of 1 \iM). We conclude that hiru din has unusual structural and refolding properties and that its mechanism of inhibition involves noncovalent interaction with multiple sites on thrombin. The interaction of hirudin (specifically the region of Lys-47) with the basic specificity pocket of thrombin may contrib ute to the binding but is not essential for its inhibitory activity.
42
Johnson/Sze/Winant/Hudson/Underhill/Lazar/Olsen/Almquist
Materials and Methods Protein Preparations Recombinant hirudin (r-hirudin) variant 1 (HV-1) was produced from a synthetic gene using an Escheri chia coli expression system based on the genetic regu latory elements o f the colicin El operon [1, 9]. The protein was purified to homogeneity by anion ex change chromatography and reverse-phase high-pres sure liquid chromatography (RP-HPLC) [10]. Human a-thrombin was generously provided by Dr. John Fenton, New York State Department o f Health. Peptide Synthesis Synthetic peptide fragments from both the aminoand carboxy-terminal regions o f hirudin were pre
pared by solid-phase techniques [11] using a Beckman Model 990 C peptide synthesizer. After synthesis, the peptides were cleaved from the resin using anhydrous hydrogen fluoride in the presence o f 10% p-cresol at - 5 °C for 1 h. The peptides were separated from the various organic side products by washing the resin with ether and then extracting the peptide with 10% acetic acid in water. Hirudin 51-65 (His-Asp-Gly-AspPhe-Glu-Glu-Ile-Pro-Glu-Glu-Tyr-Leu-Gln) and hiru din 42-65 (Gly-Glu-Gly-Thr-Pro-Lys-Pro-Gln-SerHis-Asn-Asp-Gly-Asp-Phe-Glu-Glu-Ile-Pro-Glu-GluTyr-Leu-Gln) were prepared on Boc-Gln-O-resin. Fol lowing synthesis, removal o f the protecting groups, and cleavage o f peptide from the resin, the peptide extracts were lyophilized and purified by preparative RP-HPLC (13-26% acetonitrile/0.1 % trifluoroacetic acid, linear gradient, 4 liters total solvent, 30 ml/min) using Vydac 218TPB1015 C i8 silica. [Ala22]-hirudin 6-28 (Cys-Thr-Glu-Ser-Gly-GlnAsn-Leu-Cys-Leu-Cys-Glu-Gly-Ser-Asn-Val-Ala-GlyGln-Gly-Asn-Lys-Cys) was prepared on Boc-Cys-S-(4methyl benzyl)-0-resin. Cys16 was protected as the 4methyl benzyl derivative, and Cys6 and Cys14 were protected as the acetoamidomethyl derivatives. After synthesis and cleavage with hydrogen fluoride the Cys6, Cys14 acetoamido methy-protected peptide ex tract was diluted with water, adjusted to pH 8.1 with ammonium hydroxide, and shaken slowly for 6 days to air-oxidize Cys16 and Cys28 to form this disulfide bond. The peptide was lyophilized and purified by RP-HPLC as described above except that a 5-17% gradient was used. Product fractions were combined, lyophilized, and dissolved in methanol-water (4:1) and added dropwise over 15 min to a rapidly stirring solution of 0.1 M I2 in methanol. The resulting solu tion was stirred an additional 10 min, and then 0.5 M ascorbic acid in 1.0 M citrate buffer, pH 5.0, was added dropwise until the solution remained colorless. The methanol was evaporated under reduced pres sure and the mixture applied to XAD2 resin (Alltech Associates). The resin was washed with water, and the peptide was eluted with methanol. The methanol was evaporated, and the peptide (containing two disulfide bonds at 6-14 and 16-28) was purified by RP-HPLC as described above. Assay Methods The ability o f r-hirudin and hirudin synthetic fragments to inhibit thrombin activity was deter mined using a clotting assay with purified human
Downloaded by: Thomas Jefferson University Scott Library 147.140.27.100 - 5/30/2018 6:35:33 AM
agulant effect: it can displace factor Xa from the vascular endothelium [5, 6] and thereby affect the activation of prothrombin. Although originally isolated from the sali vary gland of the medicinal leech Hirudo medicinalis, hirudin has now been produced in bacteria and yeast by recombinant DNA methods [1]. Nuclear magnetic resonance spectroscopy has provided important struc tural information about hirudin at atomic resolution [7, 8], These developments have enabled the application of various biochemi cal, genetic engineering, and molecular mod eling techniques to understand the structural and functional properties of hirudin and its mechanism of interaction with thrombin. The results reported here provide new infor mation on the structural and refolding prop erties of hirudin. We demonstrate the appli cation of (1) reverse phase-HPLC as an effi cient, high-resolution method for analyzing hirudin conformational differences; (2) im proved techniques using a computer-controlled kinetic microtiter plate reader for analyzing thrombin-hirudin inhibition ki netics, and (3) synthetic peptide fragments for mapping functional domains of hirudin.
Recombinant Hirudin
Protein Folding Analysis Hirudin was reduced and denatured by incuba tion with 0.1 M dithiothreitol in 6 M guanidine hydrochloride (GndHCl) at a pH between 8 and 10
for 1 h at 37 °C or 2 h at room temperature. The solu tion was adjusted to pH 3 and purified by RP-HPLC on a Vydac C4 column using gradient elution (1530%) in 0.065% trifluoroacetic acid and water. Hiru din fractions were lyophilized and stored at - 2 0 °C. Protein folding was initiated by dissolving reduced hirudin in reaction buffer at pH 10 and incubating for various times. The folding reaction was stopped by adjusting the pH to 3, and the reaction mixture was analyzed by RP-HPLC and thrombin inhibition anal ysis as described above. Analytical Gel Filtration Molecular weight (MW) analysis o f hirudin was performed on a Superóse 12 column controlled by an FPLC system (Pharmacia). All experiments per formed under different pH or solvent conditions were independently calibrated with MW markers includ ing: ovotransferrin (77,000), bovine serum albumin (66.000) , ovalbumin (42,000), (3-lactoglobulin A (36.000) , chymotrypsinogen A (25,000), trypsinogen (24.000) , soybean trypsin inhibitor (20,000), a-lactalbumin (14,000), ribonuclease A (13,500), cytochrome C (11,500), aprotinin (6,500), intact insulin (5,650), and insulin B chain (3,400). Hirudin MW was calcu lated from standard curves (log MW as a function of elution position) o f marker proteins determined by linear least squares fit o f the data.
Results and Discussion Thrombin Inhibition by r-Hirudin We have used a novel expression system based on modified genetic regulatory ele ments derived from the colicin El operon [13] to produce high yields of active hirudin HV-1 in E. coli [9]. Thrombin inhibition was measured by two different assay systems em ploying either fibrinogen or a chromogenic peptide as the thrombin substrate. Figure la shows the results of the fibrino gen clotting assay for a typical preparation of r-hirudin in which the calculated IC50 value is 0.5 nM. The IC50 was determined from the linear least squares fit of the data and calcu lated as the hirudin concentration at which
Downloaded by: Thomas Jefferson University Scott Library 147.140.27.100 - 5/30/2018 6:35:33 AM
fibrinogen (Sigma) or using a chromogenic substrate. For both assays we developed the application o f a computer-controlled kinetic microtiter plate reader to process up to 96 reactions simultaneously. For the clotting assay, hirudin was diluted to 50 pi in assay buffer (50 mM Tris-HCl, 120 mAf NaCl, 0.5% PEG 6000, pH 7.4), added to 50 pi of human a-thrombin (0.2 pmol) in assay buffer containing 40 mAf calcium chloride (final thrombin concentration equal to 1 aA/), and incubated for 5 min in a 96-well microtiter plate at room temperature. The reaction was initiated by addition o f 100 pi o f fibrinogen (10 mg/ml) in assay buffer (without PEG) and the solu tion mixed for 10 s. The turbidity o f the reaction mix ture was monitored at 405 nm using the Vma, Kinetic Microplate Reader (Molecular Devices Corporation). Data acquisition and processing were accomplished by a microcomputer interfaced to the microtiter plate reader using software provided by Molecular Devices Corporation. One antithrombin unit (ATU) is the amount o f hirudin that neutralizes one NIH unit o f thrombin at 37 °C, using fibrinogen as substrate. Ac tivity measurements reported here were performed at 37 °C with the chromogenic substrate S-2238 (KabiVitrum) and converted to thrombin activity using 1.25 A4os/min/thrombin unit. Inhibition constants (K¡) were determined in steady-state velocity experiments using S-2238. Reac tions were carried out at room temperature in plastic microtiter plates at a concentration o f 0.4 nAf throm bin in 50 mA/bis-Tris, 0.1 % PEG 6000, 0.1 AfNaCl, 0.5 mg/ml Brij 35 at pH 7.8; reactions were initiated by addition o f S-2238 to a final concentration o f 296 pAf in 0.3 ml. Hirudin was varied over a range that included several concentrations above and below the thrombin concentration. Data points represent the mean values from three separate determinations. Data acquisition was accomplished using the kinetic microtiter plate reader as described above. The soft ware used for tight-binding inhibition analysis was provided by Dr. Stuart Stone (Friedrich Miescher Institute, Basel, Switzerland). In a separate study (data not shown), we demonstrated that initiation o f the reaction by either substrate or thrombin yielded comparable estimates o f K¡.
43
44
Johnson/Sze/Winant/Hudson/Underhill/Lazar/Olsen/Almquist
the initial velocity of the thrombin reaction was inhibited by 50%. The concentration of hirudin was determined by comparison with standards whose concentrations and activi ties had been determined by amino acid composition analysis and chromogenic sub strate assay, respectively. Inhibition con stants were calculated from measurements of steady-state velocities as a function of hiru din concentration fitted by nonlinear regres sion to the initial rate equation of Morrison [12] for tight-binding inhibition kinetics: 2-T -V ,/V 0 = [(Ki + H - T ) 2 + 4-K '-T ]05- K' - H + T,
Protein Folding Although improper folding is a common problem for intracellular production of disulfide-bond-containing proteins in E. coli, we discovered that r-hirudin, which contains three disulfide bonds [1], was recovered in its native conformation and fully active form in extracts following cell disruption. To fur ther characterize the folding properties of hirudin, we have studied the kinetics of fold ing in vitro as a function of protein concen tration and altered solvent conditions. Pro tein folding kinetics can be conveniently measured using RP-HPLC. Figure 2 illus trates the difference in elution properties be tween oxidized (refolded) and reduced rhirudin HV-1. The two forms are reproducibly separated by an elution time difference of 8 min.
Fig. 1. Fibrinogen clotting assay (a) and tightbinding inhibition analysis using a chromogenic-substrate assay (b). Thrombin concentration was 1 \iM in a and 0.4 \iM in b. The curve represents the best fit by nonlinear regression analysis to the initial-rate equa tion.
Figure 3a shows the kinetics of hirudin refolding at a protein concentration of 50 mg/ml (7.2 mM) and molar ratios of hirudin and oxidized:reduced glutathione (2:1) equal to 0.42 and 4.2. In the presence of excess redox agent, greater than 95% of the hirudin folds to its native, fully active conformation in less than 1 h. Reducing glutathione con centration by a factor of 10 reduces the fold ing rate similarly such that the reaction pro ceeds to completion within approximately lOh. Figure 3b demonstrates that in the ab sence of any added oxidation-reduction buff er, hirudin refolds to completion within 24 h at pH 10. The reaction rate is reduced by the
Downloaded by: Thomas Jefferson University Scott Library 147.140.27.100 - 5/30/2018 6:35:33 AM
where H = hirudin and T = thrombin concen trations, Vj and V0 are initial rates of the inhibited and uninhibited reactions, and Ki = Kj7(l+[substrate]/Km). Figure lb shows the results for a standard r-hirudin prepara tion with a calculated Kj = 0.2 pA/and a spe cific activity of 10,000 ATU/mg.
addition of greater than 1 M GndHCl, but even in the presence of 6 M GndHCl more than 25% of the hirudin refolds to the native conformation within 24 h. The ability of hi rudin to refold at extremely high concentra tion even in the absence of an added oxida
tion-reduction agent and the presence of high GndHCl concentrations is unusual for a protein and suggests that hirudin may pos sess some uncommon structural and/or chemical properties. The apparent absence of formation of hirudin folding interme-
Downloaded by: Thomas Jefferson University Scott Library 147.140.27.100 - 5/30/2018 6:35:33 AM
absorbance at 215 nm using two different scales differing 5-fold in sensitivity, a Reduced r-hirudin; elution time was 27 min; dithiothreitol elutes at the solvent front; the arrow indicates the position o f oxidized r-hirudin. b Oxidized r-hirudin; elution time was 19 min; the arrow indicates the position o f reduced r-hirudin.
Johnson/Sze/Winant/Hudson/Underhill/Lazar/Olsen/Almquist
Fig. 3. Refolding kinetics o f r-hirudin. Hirudin was reduced in 0.1 M dithiothreitol and 6 M GndHCl at pH 9.5, dialyzed against 0.1 N acetic acid, lyophilized, and dissolved in 50 mM sodium carbonate/ bicarbonate buffer, 1 mM EDTA, pH 10.0. Protein (50 mg/ml) was refolded by incubation at 25 °C. The extent o f refolding was quantitated by RP-HPLC and thrombin inhibition assays, a Effect o f 3 m M and 30 mM glutathione (2:1 molar ratio o f oxidized and
reduced forms) on refolding kinetics o f 50 mg/ml hirudin, b Influence o f GndHCl concentration on hirudin refolding kinetics in the absence o f gluta thione. Fig. 4. Apparent molecular weight o f r-hirudin determined by analytical gel filtration using a supe róse 12 column. Molecular weight as a function o f pH in 0.5 M NaCl (a) and GndHCl concentration (A/) at pH 7 (b).
diates with nonnative disulfide bonds indi cates a strong tendency of hirudin to assume a native-like conformation even in the pres ence of denaturing solvents.
values at neutral pH up to 16 kD [1], Fig ure 4a shows the results of analytical sizeexclusion chromatography performed as a function of pH in 0.5 MNaCl. The apparent MW varies, from approximately 24 kD at pH 7-10, down to approximately 10 kD at pH 3.0. Figure 4b shows a similar MW tran sition as a function of GndHCl concentra tion; in 6 M GndHCl hirudin has a calcu lated MW corresponding exactly to mono mer size (7 kD). The MW value of r-hirudin
Quaternary Structure Early studies on the hydrodynamic prop erties of natural hirudin, using ultracentrifu gation and gel filtration methods, suggested a variation in apparent MW values from 7 kD (monomer size) at low pH to higher MW
Downloaded by: Thomas Jefferson University Scott Library 147.140.27.100 - 5/30/2018 6:35:33 AM
46
47
Recombinant Hirudin
Mapping Functional Domains with Synthetic Peptides Synthetic peptides provide an important approach to mapping functional domains of hirudin and identifying determinants for thrombin interaction. We chemically synthe sized three fragments corresponding to resi dues 6-28 (with Ala22 replacing Cys22), 4265, and 51-65. For the fragment 6-28 ana logue, Cys-6:Cys-14 and Cys-16:Cys-28 were oxidized to form two of the three disulfide bonds in the corresponding amino-terminal core region of hirudin. Table 1 compares the thrombin inhibi tory properties of the synthetic fragments with those of intact r-hirudin (1-65) using both the fibrinogen-clotting assay and the chromogenic-substrate assay. Fragments 4265 and 51-65 were similarly (although distinguishably) active in inhibiting thrombin cleavage of fibrinogen but were inactive (at 333 \\.M) in inhibiting thrombin hydrolysis
Table 1. Structure-activity correlations o f hirudin and synthetic peptides Peptide
IC50, gAf Fibrinogen
K„ pA/ Substrate S-22383
r-Hirudin (1-65) s-Hirudin (42-65) s-Hirudin (51-65) s-HirudinAla22-(6-28)
0.0005 4.9 6.0
0.2 NA NA
NA
NA
a
NA = No measurable activity at 333 pAf. D-Phe-L-Pipecolyl-Arg-p-nitroanilide.
of a peptidyl p-nitroanilide substrate. The fragment 6-28 analogue was not inhibitory using either substrate at concentrations up to approximately 0.4 mM. There were no syn ergistic effects of the fragment 6-28 ana logue on either of the carboxy-terminal frag ments in inhibiting thrombin (data not shown). Unlike recombinant hirudin modi fied at Lys-47, which shows a 3- to 10-fold increase in the Kj depending on the specific amino acid substitution [1], the presence of Lys-47 did not significantly enhance the ac tivity of the carboxy-terminal synthetic pep tide. Although these results may possibly in dicate that the amino terminal regional of hirudin is necessary for conformational sta bilization of the region in the vicinity of Lys47, the results support the conclusion that the possible interaction of Lys-47 with thrombin does not make a major contribu tion to the binding free energy. The ability of carboxy-terminal hirudin peptides to inhibit thrombin activity using fibrinogen as substrate, but not using a small chromogenic substrate, suggests that the car boxy-terminal domain of hirudin interacts
Downloaded by: Thomas Jefferson University Scott Library 147.140.27.100 - 5/30/2018 6:35:33 AM
at neutral pH is similar (26.5 kD) to that we previously determined using a different chromatographic medium [1], In addition, we have shown that in these two different chromatographic systems, the apparent MW of the thrombin-hirudin complex corre sponds to the sum of the MW of thrombin and hirudin monomer; the complex will form at neutral pH in 1 M, but not in 1.5 M, GndHCl [1, data not shown]. These results further support the conclusion that hirudin may be multimeric in solution but binds to thrombin as a monomer. It must be noted, however, that these MW measurements un der neutral, nondenaturing conditions do not account for possible effects due to the asymmetry of the hirudin molecule, or its possible interaction with the chromato graphic media.
48
Johnson/SzeAVinant/Hudson/Underhill/Lazar/Olsen/Almquist
Acknowledgments We thank Jon Miller and Margaret Fahnestock for their critical review o f this manuscript. This work was supported by SRI Internal Research and Develop ment projects 870D32XJC and 391D32YSB and by funds provided by the Cigarette and Tobacco Surtax Fund o f the State of California through the TobaccoRelated Disease Research Program o f the University o f California.
References 1 Johnson PH, Sze P, Winant R, Payne PW, Lazar JB: Biochemistry and genetic engineering o f hiru din. Semin Thromb Hemost 1989;15:302-315. 2 Markwardt F, Nowak G, Sturzebecher J, Griessbach U, Walsman P, Vogel G: Pharmacokinetics and anticoagulant effect o f hirudin in man. Thromb Haemost (Leipz) 1986;52:150-163. 3 Markwardt F, Fink E, Kaiser B, Klocking HP, Nowak G, Richter M, Sturzebecher J: Pharmaco logical survey of recombinant hirudin. Pharmazie 1988;43:202-207. 4 Hogg PJ, Jackson CM: Fibrin monomer protects thrombin from inactivation by heparin-anti thrombin III: Implications for heparin efficacy. Proc Natl Acad Sei USA 1989;86:3619-3623. 5 Friedberg RC, Hagen P-O, Pizzo SV: The role o f endothelium in factor Xa regulation: The effect o f
6
7
8
9
10
11
12
13
14
15
16
plasma proteinase inhibitors and hirudin. Blood 1988;71:1321-1328. Pizzo SV, Friedberg RC, Sze P, Winant R, Hud son D, Lazar JB, Johnson PH: Recombinant hiru din displaces human factor Xa from its endothe lial binding sites. Thromb Res 1990;57:803-806. Folkers PJM, Clore GM, Driscoll PC, Dodt J, Kohler S, Gronenbom AM: Solution structure o f recombinant hirudin and the Lys-47 -> Glu mu tant: A nuclear magnetic resonance and hybrid distance geometry-dynamical simulated annealing study. Biochemistry 1989;28:2601-2617. Haruyama H, Wuthrich K: Conformation o f re combinant desulfatohirudin in aqueous solution determined by nuclear magnetic resonance spec troscopy. Biochemistry 1989;28:4301-4312. Sze P, Lazar JB, Winant RC, Hudson D, Sohel I, Johnson PH: Cloning and expression o f a syn thetic gene for the thrombin specific inhibitor, hirudin, submitted. Winant RC, Lazar JB, Johnson PH: Chemical modifications and amino acid substitutions in re combinant hirudin that increase hirudin-throm bin affinity. Biochemistry, in press. Erickson BW, Merrifield, RB: Solid phase peptide synthesis; in Neurath H, Hill RL (eds): The Pro teins. New York, Academic Press, 1976, vol 2, pp 255-527. Morrison JF: Kinetics o f the reversible inhibition o f enzyme-catalyzed reactions by tight binding inhibitors. Biochim Biophys Acta 1969; 185:269— 286. Waleh NS, Johnson PH: Structural and functional organization o f the colicin E 1 operon. Proc Natl AcadSci USA 1985;82:8389-8393. Fenton II JW, Olson TA, Zabinski MP, Wilner GD: Anion binding exosite o f human alphathrombin and fibrin(ogen) recognition. Biochem istry 1988;27:7106-7112. Noe G, Hofsteenge J, Rovelli G, Stone SR: The use of sequence specific antibodies to identify a secondary binding site in thrombin. J Biol Chem 1988;263:11729-11735. Lazar JB, Winant RC, Johnson PH: Hirudin: Amino-terminal residues play a major role in the interaction with thrombin. J Biol Chem, in press. Dr. Paul Johnson Molecular Biology Department SRI-20501, 333 Ravenswood Avenue Menlo Park, CA 94025 (USA)
Downloaded by: Thomas Jefferson University Scott Library 147.140.27.100 - 5/30/2018 6:35:33 AM
with a region of thrombin distinct from the catalytic site. This region is likely to be the anion binding exosite, a region rich in basic amino acids that is implicated in fibrinogen binding [14] and contains the p-cleavage site. An antibody to a synthetic peptide repre senting this region is able to block hirudin binding to a-thrombin [15]. On the basis of these results, together with implications from our molecular-modeling studies [ 1, 16], hirudin appears to interact with thrombin at multiple sites including the catalytic site, the apolar binding site, and the anion-binding exosite.
