Biochemistry 1998, 37, 6645-6657
6645
Novel Natural Product 5,5-trans-Lactone Inhibitors of Human R-Thrombin: Mechanism of Action and Structural Studies† Malcolm P. Weir,*,‡ Susanne S. Bethell,‡ Anne Cleasby,‡ Callum J. Campbell,‡ Richard J. Dennis,‡ Clive J. Dix,‡ Harry Finch,‡ Harren Jhoti,‡ Christopher J. Mooney,‡ Shila Patel,‡ Chi-Man Tang,‡ Malcolm Ward,‡ Alan J. Wonacott,‡ and Christopher W. Wharton§ Glaxo Wellcome Medicines Research Centre, Gunnels Wood Road, SteVenage, Hertfordshire SG1 2NY, U.K., and School of Biochemistry, UniVersity of Birmingham, Birmingham B15 2TT, U.K. ReceiVed October 9, 1997; ReVised Manuscript ReceiVed February 18, 1998
ABSTRACT: High-throughput screening of methanolic extracts from the leaves of the plant Lantana camara identified potent inhibitors of human R-thrombin, which were shown to be 5,5-trans-fused cyclic lactone euphane triterpenes [O’Neill et al. (1998) J. Nat. Prod. (submitted for publication)]. Proflavin displacement studies showed the inhibitors to bind at the active site of R-thrombin and R-chymotrypsin. Kinetic analysis of R-thrombin showed tight-binding reversible competitive inhibition by both compounds, named GR133487 and GR133686, with respective kon values at pH 8.4 of 1.7 × 106 s-1 M-1 and 4.6 × 106 s-1 M-1. Electrospray ionization mass spectrometry of thrombin/inhibitor complexes showed the tight-bound species to be covalently attached, suggesting acyl-enzyme formation by reaction of the active-site Ser195 with the trans-lactone carbonyl. X-ray crystal structures of R-thrombin/GR133686 (3.0 Å resolution) and R-thrombin/GR133487 (2.2 Å resolution) complexes showed continuous electron density between Ser195 and the ring-opened lactone carbonyl, demonstrating acyl-enzyme formation. Turnover of inhibitor by R-thrombin was negligible and mass spectrometry of isolated complexes showed that reversal of inhibition occurs by reformation of the trans-lactone from the acyl-enzyme.The catalytic triad appears undisrupted and the inhibitor carbonyl occupies the oxyanion hole, suggesting the observed lack of turnover is due to exclusion of water for deacylation. The acyl-enzyme inhibitor hydroxyl is properly positioned for nucleophilic attack on the ester carbonyl and therefore relactonization; furthermore, the higher resolution structure of R-thrombin/GR133487 shows this hydroxyl to be effectively superimposable with the recently proposed deacylating water for peptide substrate hydrolysis [Wilmouth, R. C., et al. (1997) Nat. Struct. Biol. 4, 456-462], suggesting the R-thrombin/GR133487 complex may be a good model for this reaction.
R-Thrombin is a serine protease that belongs to the trypsin family and is generated by autocatalytic and factor Xa cleavage of the circulating plasma protein prothrombin at the final step of the coagulation cascade; it has a central role in the hemostatic process, where it displays both coagulant (3) and anticoagulant properties (4). The major coagulant effects are the cleavage of fibrinogen to release fibrin, the activation of platelets through proteolysis of the platelet thrombin receptor leading to platelet aggregation (5), and conversion of factors V and VIII to factors Va and VIIIa, respectively, in a positive feedback loop, causing amplification of the cascade (3). X-ray crystallographic studies have shown R-thrombin to possess the classical trypsin-like fold and the Ser195/His57/Asp1021 catalytic triad at the active site (6). In addition, the structure contains insertion loops that help impart specificity by participating in substrate and inhibitor binding as well as a unique secondary binding site, † Coordinates have been deposited with the Brookhaven Protein Data Bank under codes 1AWF (GR133487) and 1AWH (GR133686). ‡ Glaxo Wellcome Medicines Research Centre. § University of Birmingham. 1 The amino acid residue numbering follows the chymotrypsinogen system of Bode et al. (9).
the exosite, which is distant from the active site. R-Thrombin utilizes this secondary site when it binds fibrinogen and the leech anticoagulant protein hirudin in a bidentate manner (7). One region of the R-thrombin structure that appears to determine substrate specificity is a loop close to the active site, referred to as the sixties loop, residues from which help define the S1′ and S2 pockets2 in particular (9). The S1 pocket prefers basic residues such as Arg and Lys due to Asp189, which forms the base of the pocket. Serine proteases have attracted a great deal of interest as drug targets due to their widespread involvement in biological processes, and many serine protease inhibitors have been designed or discovered that exploit the catalytic mechanism to form reversible or irreversible covalent adducts [reviewed by Powers and Harper (10) and Wharton (11)]. Classes of irreversible inhibitor include β-lactam elastase inhibitors (12-14), isocoumarins (15), isatoic anhydrides (16), enol lactones (17), ynenol lactones (18), 6-chloro-2-pyrones (19), and the chloromethyl ketone R-thrombin inhibitor PPACK,3 which forms an adduct to His57 (20). Boronate, trifluo2 Naming of enzyme pockets (S1, S2, ...) corresponding to substrate/ inhibitor residues (P1, P2, ...) follows the notation of Schechter and Berger (8).
S0006-2960(97)02499-9 CCC: $15.00 © 1998 American Chemical Society Published on Web 04/23/1998
6646 Biochemistry, Vol. 37, No. 19, 1998
Weir et al. EXPERIMENTAL PROCEDURES
FIGURE 1: Triterpenes GR133686 and GR133487 were isolated from the leaves of the plant Lantana camara; extracts of the leaves showed activity as inhibitors of R-thrombin in an in vitro fibrinformation assay and the active components were purified and characterized from this source as described in O’Neill et al. (1). The activity of these compounds against a range of serine proteases is shown in Table 2. The structure of the peptide choloromethyl ketone inhibitor PPACK (20) used here for mechanistic studies is shown for comparison.
romethyl ketone, and aldehyde inhibitors have been reported to form reversible covalent adducts to Ser195 that mimic the catalytic tetrahedral intermediate (21-24). Thrombin’s central role in clot formation combined with this high level of knowledge of serine protease inhibition makes it a prime target for antithrombotic inhibitor design, and many inhibitors have been produced both by mechanism-based approaches and by design of competitive inhibitors that bind exclusively by noncovalent interactions (24-27). In a search for novel R-thrombin inhibitors by highthroughput screening of natural products, methanolic extracts of leaves from Lantana camara (commonly known as wild sage) were found potently to inhibit human R-thrombin and to a lesser extent R-chymotrypsin and other serine proteases; fractionation and purification showed the active constituents to be 5,5-trans-fused cyclic lactone euphane triterpenes (Figure 1, 1). We describe here the structural and mechanistic basis for R-thrombin inhibition by these molecules, which is shown to be via reversible acylation of Ser195 by the lactone moiety and which provides a basis for generalization of this novel template to the design of serine protease inhibitors. In addition, these observations show the precise location of the hydroxyl group for inhibitor deacylation (relactonization), which may coincide with the position of the deacylating water molecule for peptide substrate hydrolysis (2). 3 Abbreviations: EDTA, ethylenediaminetetraacetic acid; ESI-MS, electrospray ionization mass spectrometry; FT-IR, Fourier transform infrared spectroscopy; MS/MS, tandem mass spectrometry; PEG, poly(ethylene glycol); pNA, p-nitroanilide; pNp, p-nitrophenyl; PPACK, D-Phe-L-Pro-L-Arg chloromethyl ketone; Eoc-, N-ethoxycarbonyl; Boc, tert-butyloxycarbonyl; boroarginine, boronic acid analogue of arginine; RP-HPLC, reversed-phase high-pressure liquid chromatography; tPA, tissue plasminogen activator.
Materials. Freeze-dried human prothrombin complex intermediate was purchased from Bio Products Laboratory, Dagger Lane, Elstree, U.K. Oxyuranus scutellatus snake venom, N-trans-cinnamoylimidazole, and sodium phosphate were purchased from Sigma. Bovine pancreatic chymotrypsin, bovine trypsin, and human plasmin were from Sigma; factor Xa was from Boehringer; factors XIa and XIIa were from Enzyme Research Laboratories Inc.; and human tPA, cathepsin G, and neutrophil elastase were from Calbiochem, Nottingham, U.K. Specific peptide p-nitroanilide substrates for elastase and cathepsin G were from Sigma; tPA substrate was from American Diagnostica Inc.; and substrates for chymotrypsin, trypsin, factors Xa, XIa, and XIIa, plasmin, and R-thrombin were from Kabivitrum. Ultrapure proflavin hemisulfate was purchased from ICN Biomedicals. The heparin-Sepharose CL-6B resin was from Pharmacia, PEG 6000 was from Fluka, Tris-HCl was from BDH, and the sodium chloride was from Fisons. The euphane triterpene compounds GR133686 and GR133487 (Figure 1) were isolated from the leaves of the Lantana camara plant as previously described (1). All buffers were made up with Milli-Q-grade water. PPACK was from Calbiochem. Molecular weight cutoff filters were from Millipore, Watford, U.K. R-Thrombin ActiVation and Purification. Human R-thrombin was activated and purified following the procedure of Fenton et al. (28) and later by the modified procedure of Ngai and Chang (29). One vial of the prothrombin complex intermediate was resuspended at room temperature in 20 mM Tris-HCl and 0.1 M NaCl buffer, pH 7.5, and diluted until the concentration of prothrombin was 0.2-0.4 mg/mL, estimated using an extinction coefficient at 280 nm of ) 5.2 × 104 M-1 cm-1. The concentration of CaCl2 was made up to 10 mM by adding 1 mL of a 1 M CaCl2 solution per 100 mL of prothrombin solution. Freeze-dried snake venom (10 mg) was dissolved in 1 mL of water and added to the prothrombin under constant stirring. The activation was monitored by the increase in absorbance at 405 nm following the release of nitroaniline from the cleavage of the chromogenic substrate S2238 (30). When the rate of absorption change became constant, the activated solution was placed on ice. The activation was usually complete in 30-45 min. The activated R-thrombin was loaded onto a heparinSepharose column equilibrated with 20 mM Tris-HCl and 0.1 M NaCl buffer, pH 7.5, at a flow rate of 5 mL/min. After loading, the column was washed until all the unbound inactive material was eluted. The activated R-thrombin was eluted from the column with a linear salt gradient from 0 to 100% of 20 mM Tris-HCl and 1 M NaCl, pH 6.0, and pooled fractions were stored at -70 °C. ProflaVin Binding. Proflavin displacement studies were performed to assess inhibitor active-site binding (31). The buffer used was 50 mM sodium phosphate and the pH was either 6.0 or 8.4 and was adjusted by varying the ratio of the mono- and dibasic phosphate salts. Fresh stocks of proflavin were prepared daily and the concentration was determined by measuring the absorption at 444 nm and using an extinction coefficient of ) 3.79 × 104 M-1 cm-1. Binding and displacement studies of proflavin to R-thrombin were conducted on a double-beam Cecil 6600 instrument
R-Thrombin Inhibition by 5,5-trans-Lactones with a thermostated 10 mm path length cuvette holder to which a Julabo U3 circulating water bath was attached and set at 25 °C. The binding studies were conducted by adding proflavin in 0.5 µM increments to the sample and reference cuvettes. The sample cuvette contained 8.8 µM R-thrombin. The displacement experiments were performed by titrating the compound GR133686 into a 10 µM R-thrombin/13 µM proflavin solution. Fourier Transform Infrared Spectroscopy. The instrument used was a Nicolet 60SX Fourier transform instrument attached to an IBM personal computer running the Nicolet PC/IR software. The cell used was a 75 µm path length, CaF2 demountable liquid cell with Teflon spacers. The experiments were carried out using bovine pancreatic R-chymotrypsin as a model for thrombin. Protein was dissolved in unbuffered D2O to give approximately 3 mL of a 50 mg/ mL stock solution. After 24-48 h, the pD was adjusted to 7.4 by adding DCl or NaOD. Ten microliters of a 45 mM inhibitor solution in acetonitrile was added over a period of 2 min to give a 1:1 enzyme-to-inhibitor ratio. The reference samples were prepared in similar fashion, leaving out the compound. To ensure that the protein concentration in each sample and reference set was identical, each set was prepared from the same stock, which was centrifuged for 2 min in an Eppendorf centrifuge prior to division into two 0.4 mL aliquots. The sample was loaded into the cell and 3200 scans at 2 cm-1 resolution were collected over a period of about 5 min. The reference spectrum was subtracted from that of the sample to give the difference spectrum of the effect of the inhibitor complex. The extent of acylation was monitored independently by titrating a 40 µL aliquot of the sample with N-trans-cinnamoylimidazole halfway through the data collection (32). Enzyme Kinetics. Inhibition versus a range of serine proteases was initially assessed by 15-min IC50 measurements. Enzyme and inhibitor were preincubated in 96-well microplates for 15 min, and reactions were initiated by addition of peptidyl p-nitroanilide substrates to a final volume of 200 µL. Activity was measured at 405 nm as the initial rate of substrate hydrolysis, and the concentration of inhibitor reducing activity to 50% of uninhibited control levels (IC50) was determined. Conditions for the following proteases were essentially as described in the associated references: R-thrombin (33), tissue plasminogen activator (34), factor Xa (35), cathepsin G and human neutrophil elastase (36), and plasmin (37). Conditions and substrates for chymotrypsin, trypsin, factor XIa, and factor XIIa were (a) chymotrypsin, 50 mM Tris-HCl and 50 mM CaCl2, pH 8.4, assayed against 200 µM MeO-Suc-Arg-Pro-Tyr-pNA; (b) trypsin, 50 mM TrisHCl and 15 mM CaCl2, pH 8.4, assayed against 160 µM N-benzoyl-Ile-Glu-Gly-Arg-pNA; (c) factor XIa and factor XIIa, 28 mM barbital, 125 mM NaCl, and 1 mM EDTA, pH 7.4, assayed against 400 µM pyroGlu-Pro-Arg-pNA and 200 µM D-Pro-Phe-Arg-pNA, respectively. Detailed kinetic analysis of R-thrombin was carried out as follows. Specific chromogenic substrate S2238 was dissolved in water at 2 mM, stored at 4° C, and diluted in the appropriate buffers for use. For experiments at pH 8.4 the buffer was 50 mM Tris-HCl, 0.5 M NaCl, and 0.1% PEG 6000 (Km ) 1.91 µM), and at pH 6.0 the buffer was 50 mM Na2HPO4/NaH2PO4, 0.75 M NaCl, and 0.1% PEG 6000 (Km ) 10.6 µM). Progress curves were measured at 25 °C in a
Biochemistry, Vol. 37, No. 19, 1998 6647 final volume of 1 mL in 1 cm path length cuvettes at a substrate concentration of 0.2 mM except where stated, and reactions were monitored at 405 nm in a Beckman DU70 spectrophotometer. The molar extinction coefficient was determined at 405 nm using a 0.1 mM p-nitroaniline standard (Kabivitrum) as 1.02 × 104 M-1 cm-1 at pH 6.0 and as 1.0 × 104 M-1 cm-1 at pH 8.4. These values were used to calculate product release in progress curve experiments. Enzyme Kinetics: Inhibition Progress CurVes. Preliminary experiments indicated that GR133686 and GR133487 displayed time-dependent inhibition of R-thrombin and R-chymotrypsin. Depending on the inhibition mechanism, the pseudo-first-order rate constant kobs, estimated from inhibition progress curves, varies with inhibitor concentration [I] in a characteristic fashion (38). A linear relationship
Scheme 1 kon
E + I {\ } EI k off
implies simple time-dependent inhibition (Scheme 1), where kobs has the relationship
kobs ) koff + kon[I]/(1 + [S]/Km)
(1)
A hyperbolic plot of kobs versus [I] indicates a mechanism displaying saturation kinetics typical of alternate substrate or slow-binding inhibition, in which enzyme and inhibitor rapidly equilibrate to form a noncovalent complex, E‚I, which then isomerizes relatively slowly to the tight or stable complex, EI* (Scheme 2), where kobs has the relationship
kobs ) k4 + k3[I]/Ki(1 + [S]/Km + [I]/Ki)
(2)
koff and k4 were measured in separate experiments as described below.
Scheme 2 Ki
k3
} EI* E + I {\} E‚I {\ k 4
Experiments generating progress curve data of the timedependent inhibition of R-thrombin by GR133487 and GR133686 were initiated by adding 990 µL of a preformed mixture of substrate and inhibitor at varying concentrations to 10 µL of human R-thrombin to give a final concentration of 1 nM R-thrombin. Progress curve data were collected by recording A405 for 60 min or for shorter periods at higher inhibitor concentrations and transferred using Beckman data capture software. Experiments were carried out under pseudo-first-order conditions ([I] > 25[E]). Control reactions always gave a linear increase in absorbance well in excess of those used for analysis, but typically less than 10% of substrate was consumed. The range of inhibitor concentrations used was defined empirically to give measurable values of kobs. The integrated rate equation describing product concentration as a function of time in the presence of a slow-binding inhibitor (38), eq 3, was fitted to the data of each curve by nonlinear regression using the RS/1 (BBN Software Products Corp.) procedure FITFUNCTION, giving values of kobs at each concentration of inhibitor:
6648 Biochemistry, Vol. 37, No. 19, 1998
A405 ) Vst + (V0 - Vs)(1 - e-kobst)/kobs + c
Weir et al.
(3)
where V0 is the initial rate of change of absorbance, Vs is the steady-state rate, kobs is the pseudo-first-order rate constant for transition from V0 to Vs, and c is the initial absorbance at 405 nm. Enzyme Kinetics: Concentrated Incubations and Dissociation of Preformed ComplexessMeasurement of k4. Thrombin (2 µM) was mixed with an equal volume of GR133487 (2.5 µM at pH 8.4, 10 µM at pH 6.0, for 60 min), or 5 µM thrombin was mixed with an equal volume of GR133686 (20 µM at pH 8.4, 40 µM at pH 6.0, for 60 min). Assay of residual activity of these mixtures confirmed that thrombin activity was minimal. Progress curves of the recovery of enzymatic activity were carried out by extensive dilution into a solution of S2238, 250 µM, and observing the recovery of activity by monitoring absorbance at 405 nm. Final thrombin concentrations were 0.2 nM (GR133487 at pH 8.4), 0.5 nM (GR133487 at pH 6.0), 0.1 nM (GR133686 at pH 8.4), and 0.05 nM (GR133686 at pH 6.0). The low final concentration of R-thrombin for GR133686 at pH 6.0 was necessary because of the much longer time course of recovery that was observed for these conditions. Equation 3 was fitted to the data of individual progress curves to determine kobs, which for convenience is referred to as kdis when obtained from dissociation progress curves. Equations 1 and 2 define kdis. In these experiments the extensive dilutions (4000-100 000-fold) made to allow recovery of total enzyme activity give a low final value of [I]. This combines with a large value of 1 + [S]/Km such that kdis tends to koff (eq 1) or k4 (eq 2) and thus becomes a direct measure of the limiting dissociation rate constant. That the final concentration of inhibitor was negligible was confirmed for GR133487 at pH 8.4 by performing dissociation experiments over a range of inhibitor concentrations, with no detectable variation in kdis. Enzyme Kinetics: Residual ActiVity under Single-TurnoVer Conditions. Experiments were performed at pH 8.4 by mixing a final concentration of 50 nM GR133487 with increasing amounts of R-thrombin, 10-100 nM, and assaying residual activity at each combination of enzyme:inhibitor at 30, 90, and 180 min after mixing. Over part of the range of this experiment R-thrombin was in significant excess over GR133487, providing several samples under single-turnover conditions. Formation and Isolation of R-Thrombin/Inhibitor Complexes. R-Thrombin/inhibitor complexes were separated from free inhibitor by ultrafiltration as follows. R-Thrombin, in 300 µL aliquots at a concentration of 16 µM, was buffered with 50 mM Na2HPO4/NaH2PO4, pH 6, and 0.75 M NaCl. Inhibitors were dissolved in a small amount of CH3CN followed by excess buffer to a final concentration of 1.6 mM. The incubation solution was prepared by addition to the R-thrombin of 15 µL of inhibitor solution. The vessel was then incubated at 37 °C for 5 min, and 25 µL was removed and injected onto RP-HPLC. The remaining 275 µL was transferred to a centrifugal filtration cartridge with a nominal molecular weight cutoff of 10 000 kDa, which had been prewashed with buffer leaving the membrane just wet, and spun through with further loadings of buffer (2 × 300 µL), leaving around 50 µL each time; after the second wash the
volume in the vessel was made back to 300 µL and the total was transferred to a fresh tube. A 25 µL aliquot was subjected to RP-HPLC. CompetitiVe Incubation with PPACK. One mole equivalent of PPACK dissolved in buffer was added to the remainder of the R-thrombin/inhibitor complex isolated as above. The mixture was incubated at 37 °C and aliquots were chromatographed at hourly intervals from 1 to 4 h after PPACK addition. The eluted components were collected for mass spectrometric analysis. A further aliquot containing a molar excess of fresh thrombin was added to a 50 µL portion of R-thrombin/inhibitor/PPACK incubation mixture at the 4 h time point. RP-HPLC. Chromatography was carried out using a Pharmacia FPLC system at a flow rate of 1 mL/min on a 4.6 × 100 mm Pharmacia PRO-RPC column using a linear gradient with a 5 min hold time at 5% eluent B followed by 5-60% eluent B over 50 min. Eluent A was 0.1% aqueous trifluoroacetic acid, and eluent B contained 0.1% TFA in 60% CH3CN/H2O. Detection was performed at 252 nm; the absorbance ranges are shown in the chromatograms. Samples were dried in a Savant vacuum centrifuge prior to mass spectrometry. Mass Spectrometry. Mass spectrometry was performed by infusion of the RP-HPLC fractions dissolved in CH3CN/ H2O/AcOH (50:50:1) at 1 µL/min into an Analytica electospray ionization source fitted to a Finnigan TSQ-700. The mass spectrometer was scanned over the m/z range 8001800 for R-thrombin and m/z 400-800 for the inhibitors. A scan cycle time of 30 s was used. MS/MS experiments utilized unit resolution for parent selection of m/z 753, and -20 V for collision potential. RP-HPLC/MS data on autolytically cleaved R-thrombin samples was recorded on a Sciex API III mass spectrometer scanned over the range m/z 900-2400 in 0.3 Da steps at 0.5 ms/step. X-ray Crystallography: R-Thrombin/GR133686. Human R-thrombin was inhibited using a 1:5 molar ratio of protein to GR133686 dissolved in acetonitrile. Crystallization experiments were performed using vapor diffusion techniques together with the McPherson screen to scan potential conditions (39). Crystals suitable for X-ray analysis were grown from 8-12% PEG 4000 and 100 mM sodium acetate, pH 5.0-5.5, at 4 °C using an initial protein concentration of 36 mg/mL. Crystals grew as orthorhombic rods in 7 days to a typical size of 350 × 100 µm and belong to space group P212121 (a ) 65.6 Å, b ) 102.8 Å, c ) 119.7 Å) with two molecules per asymmetric unit. A data set was collected from one crystal using a FAST area detector system mounted on an Enraf-Nonius FR581 rotating anode and processed using MADNES and CCP4 programs (Table 1, 40). The structure was solved using molecular replacement techniques with a well-refined 1.9 Å structure of a R-thrombin/PPACK complex (coordinates courtesy of Professor W. Bode, Max Planck Institute, Munich); the search model consisted only of protein atoms. The rotation function using X-PLOR (41) resulted in two peaks 7.5σ above the mean using 3-15 Å data. These two orientations were then optimized independently using the X-PLOR Patterson-correlation function. The translation function was then calculated with a data range 3-10 Å and using the two orientations independently; a top peak of ∼13σ resulted in each run. Refinement was performed using X-PLOR and the final R-factor was 20.2%
R-Thrombin Inhibition by 5,5-trans-Lactones
Biochemistry, Vol. 37, No. 19, 1998 6649
Table 1: Statistics for X-ray Data Collection and Refinement of the R-Thrombin/Inhibitor Complexes GR133686
GR133487
resolution (Å) R-merge (%)
Data Collection P212121 a ) 65.6, b ) 102.8, c ) 119.7 3.0 12.4
C2 a ) 71.9, b ) 71.8, c ) 73.0a 2.2 6.0
model atoms R-factor (%)
Refinement 4840 (2 solvent) 20.2
2757 (230 solvent) 19.9
bonds (Å) angles (deg) dihedrals (deg) impropers (deg)
Geometryb 0.017 (0.018) 3.6 (3.6) 29.9 (30.1) 2.1 (1.9)
0.016 3.1 29.3 1.9
space group unit cell (Å)
a β ) 101.4°. b The values in parentheses are for the second molecule in the asymmetric unit for the thrombin/GR133686 complex.
at 3 Å with good stereochemistry for the model (Table 1). X-ray Crystallography: R-Thrombin/GR133487. Hirugen-thrombin was prepared according to a modification of the method of Skrzypczak-Jankun et al. (7). Crystallization was carried out by the vapor diffusion method using 10 mg/ mL of hirugen-thrombin inhibited by GR133487 (in excess of 100Ki), 12.5% PEG 4000, 50 mM Hepes, pH 7, and 100 mM NaCl in the drop and 30% PEG 4000, 100 mM Hepes, pH 7, and 1.6 M NaCl in the well. After 24 h equilibration time the drops were seeded with microcrystals of hirugenthrombin following which the NaCl concentration in the well was slowly increased over several days. Crystals grew at 20 °C to typical sizes of 300 × 300 × 225 µm and were characterized as belonging to space group C2 (Table 1). A data set was collected using the in-house FAST area detector (as above) to 2.2 Å resolution using one crystal. The R-thrombin/GR133487 structure did not require a separate structure solution as the crystals were isomorphous to the native hirugen-thrombin structure that had previously been solved using molecular replacement in our laboratory using the R-thrombin/PPACK search model. The refinement was performed using X-PLOR and the final R-factor was 19.9% at 2.2 Å with good geometry for the model (Table 1). RESULTS The strained nature of the 5,5-trans-lactone ring system was suggestive of protease inhibition by ring opening and acylation of the active-site serine. Binding at the active site of R-thrombin and R-chymotrypsin was first shown by displacement of the dye proflavin monitored by difference absorption spectroscopy; proflavin is known to bind in the active site of serine proteases with a low micromolar Kd and with a concomitant shift in absorption maximum from 444 to 465 nm (31, 42). It has been shown to bind to an apolar site close to the catalytic site of R-thrombin (43). Titration of GR133686 into 10 µM R-thrombin/proflavin complex at pH 8.3 showed displacement of proflavin proportional to inhibitor concentration (Figure 2), indicating tight binding within the mixing time of the experiment (20-30 s). Similar results were seen with thrombin at pH 6 and with R-chymotrypsin. No change in absorbance was seen during 2 h at 20 °C with a 2-fold excess of GR133686 over R-thrombin,
FIGURE 2: Proflavin displacement from R-thrombin by GR133686. Titration of proflavin into a 10 µM solution of R-thrombin causes a stoichiometric change in the (465 - 440 nm) difference absorption of the dye as shown in the inset figure. Proflavin is quantitatively displaced by the tighter-binding GR133686, indicating that the inhibitor binds in the active site of R-thrombin. Table 2: IC50 Data for GR133487 and GR133686 versus a Range of Serine Proteinases IC50a (µM) enzyme
GR133487a
GR133686
R-thrombin R-chymotrypsin trypsin tPA factor Xa cathepsin G plasmin factor XIa neutrophil elastase factor XIIa
0.004 0.01 0.12 >10 >10 >10 5.6 1.1 >10 >10
0.004 0.07 0.07 1.0 >10 2.2 3.5 0.7 >10 >10
a IC 50 values (means of three determinations) are the inhibitor concentrations reducing enzyme activity to 50% control level after 15 min.
indicating that the tightly bound complex was relatively stable. Kinetic Analysis of R-Thrombin Inhibition. The timedependent onset of inhibition of R-thrombin by GR133487 and GR133686 was initially observed in IC50 determinations, which at pH 8.4 also showed that inhibition was tight-binding and selective for R-thrombin over a number of other serine proteases (Table 2), although it is noteworthy that these compounds are also potent R-chymotrypsin inhibitors despite the difference in P1 specificity of these enzymes (44). The kinetics of R-thrombin inhibition were examined in detail. A typical family of inhibition progress curves for GR133487 at pH 8.4 is shown in Figure 3a; similar families of curves were obtained at pH 8.4 for GR133686 and at pH 6.0 for both compounds. GR133686 at pH 6.0 and GR133487 at both pH 6.0 and 8.4 showed evidence of saturation of kobs with [I], as shown in Figure 3b, typical of the Scheme 2 mechanism. GR133686 at pH 8.4, however, showed a linear increase of kobs with [I] and was therefore analyzed according to Scheme 1. Because of the small magnitude of koff and k4 compared to kobs, the former parameters were measured directly from dissociation progress curves . In these experiments enzyme and inhibitor were mixed under conditions favoring production of the tight complex, EI*, such that dissociation could be initiated by extensive dilution into a solution of chromogenic substrate with the enzyme almost completely
6650 Biochemistry, Vol. 37, No. 19, 1998
Weir et al.
FIGURE 3: (a) Progress curves of slow-binding R-thrombin inhibition by GR133487 at pH 8.4. Reactions were initiated by addition of inhibitor and substrate, 990 µL, to enzyme solution, 10 µL, in 50 mM Tris-HC1 buffer containing 0.5 M NaCl and 0.1% PEG 6000. Final inhibitor concentrations for each curve were as labeled, and final enzyme concentration was 1 nM. Except at higher inhibitor concentrations (micromolar), initial rates of substrate hydrolysis were similar to rates with uninhibited enzyme (control). Solid lines are the integrated rate equation (eq 3) fitted to the data of each curve from which the pseudo-first-order rate constant for inhibition, kobs, is derived. (b) Replot of kobs versus [I] for inhibition of R-thrombin by GR133487 at pH 8.4. kobs approaches a limit at higher concentrations of GR133487, indicating saturation of the enzyme. The solid line is eq 2 fitted to the data, taking into account the predetermined value of k4. Similar plots were obtained for GR133487 and GR133686 at pH 6.0. For GR133686 at pH 8.4, the relationship between kobs and [I] was linear and was analyzed by fitting eq 1 to the data. (c) Progress curves of dissociation of R-thrombin activity from preformed inhibited complexes with GR133487 (b) and GR133686 (0) at pH 8.4. Buffer conditions were as in panel a. For conditions see main text. Solid lines are eq 3 fitted to the data of each curve. (d) Residual R-thrombin activity in enzyme-inhibitor complexes with GR133487 at pH 8.4. R-Thrombin at the final concentrations indicated was mixed with GR133487 at a final concentration of 50 nM. Plots of residual activity determined at 30 (O), 90 (b) and 180 min (0) after mixing are superimposable, indicating minimal breakdown of the inhibitor. Table 3: Summary of R-Thrombin Inhibition Kinetic Data for GR133487 and GR133686 and Some Comparative Inhibition Data (49-55) GR133487,b
pH 8.4 GR133487,b pH 6.0 GR133686,c pH 8.4 GR133686,b pH 6.0 antithrombin III heparin-antithrombin III protease nexin 1 PPACK Eoc-Phe-Pro-azaLys-p-Np Boc-(D)Phe-Pro-boroArg-OH fluorescein mono-p-guanidinobenzoate hydrochloride 4-chloro-7-guanidino-3-methoxyisocoumarin
kon (s-1 M-1)
k4 or koff (s-1)
Ki
k3 (s-1)
Ki*
t1/2a (min)
1.71 × 0.037 × 106 4.6 × 106 0.296 × 106 7.4 × 103 1.4 × 106 1.8 × 106 1.2 × 107 >1.6 × 107 9.3 × 106 2.77 × 103 2.9 × 105
8.4 × 1.3 × 10-4 1.3 × 10-4 7.1 × 10-6 2.6 × 10-6 2.3 × 10-6 1.7 × 10-5 na
24 nM 0.97 µM 28 pM 0.83 µM 1.4 mM 4.0 µM 3.4 µM 0.13 µM
