- Email: [email protected]
Novel interactions of large P3 moiety and small P4 moiety in the binding of the peptide mimetic factor VIIa inhibitor
BBRC Biochemical and Biophysical Research Communications 326 (2005) 859–865 www.elsevier.com/locate/ybbrc
Novel interactions of large P3 moiety and small P4 moiety in the binding of the peptide mimetic factor VIIa inhibitorq Shojiro Kadono*, Akihisa Sakamoto, Yasufumi Kikuchi, Masayoshi Oh-eda, Naohiro Yabuta, Kazutaka Yoshihashi, Takehisa Kitazawa, Tsukasa Suzuki, Takaki Koga, Kunihiro Hattori, Takuya Shiraishi, Masayuki Haramura, Hirofumi Kodama, Yoshiyuki Ono, Toru Esaki, Haruhiko Sato, Yoshiaki Watanabe, Susumu Itoh, Masateru Ohta, Toshiro Kozono Fuji Gotemba Research Labs, Chugai Pharmaceutical Co., Ltd., 1-135 Komakado, Gotemba, Shizuoka 412-8513, Japan Received 12 November 2004 Available online 7 December 2004
Abstract Selective factor VIIa-tissue factor complex (FVIIa/TF) inhibition is seen as a promising target for developing new anticoagulant drugs. A novel peptide mimetic factor VIIa inhibitor, ethylsulfonamide-D -biphenylalanine-Gln-p-aminobenzamidine, shows 100fold selectivity against thrombin in spite of its large P3 moiety, unlike previously reported FVIIa/TF selective inhibitors. X-ray crystal structure analysis reveals that the large P3 moiety, D -biphenylalanine, and the small P4 moiety, ethylsulfonamide, make novel interactions with the 170-loop and Lys192 of FVIIa/TF, respectively, accompanying ligand-induced conformational changes of the 170-loop, Gln217, and Lys192. Structural comparisons of FVIIa with thrombin and amino acid sequence comparisons among coagulation serine proteases suggest that these interactions play an important role in achieving selective inhibition for FVIIa/TF. 2004 Elsevier Inc. All rights reserved. Keywords: Factor VIIa; Blood coagulation; Serine protease; X-ray crystallography; Drug design
The blood coagulation cascade is divided into extrinsic and intrinsic coagulation pathways. Factor VIIa (FVIIa) in complex with tissue factor (TF) initiates the extrinsic coagulation pathway. This complex activates factors IX to IXa (FIXa) and X to Xa (FXa), which in turn activate factor X to Xa and prothrombin to thrombin, respectively [1]. Thrombin cleaves fibrinogen to fibrin, which forms blood clots with activated platelets. Inappropriate thrombus formation in blood vessels causes cardiovascular diseases (myocardial infarction, stroke, pulmonary embolism, and so on), which are the most common q Abbreviations: FVIIa, factor VIIa; TF, tissue factor; sTF, soluble domain of tissue factor; FIXa, factor IXa; FXa, factor Xa; FXIa, factor XIa; FXIIa, factor XIIa; APC, activated protein C. * Corresponding author. Fax: +81 550 87 5326. E-mail address: [email protected] (S. Kadono).
0006-291X/$ - see front matter 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.bbrc.2004.11.108
causes of mortality in the industrialized world [2]. Recent studies on blood coagulation have suggested that selective inhibition of extrinsic coagulation provides effective anticoagulation and low risk of bleeding compared with other antithrombotic mechanisms [3–5]. Thus, specific FVIIa/TF complex inhibition, which blocks only extrinsic coagulation, is seen as a promising target for developing new anticoagulant drugs [6,7]. Recently, we reported on propylsulfonamide-D -ThrMet-p-aminobenzamidine (compound 1, Fig. 1), which showed submicromolar inhibition for FVIIa/TF (IC50 = 130 nM) and 26-fold selectivity against thrombin [8]. In addition, another peptide mimetic inhibitor, benzylsulfonamide-D -Ile-Gln-p-aminobenzamidine (compound 2, Fig. 1), showed potent inhibition for FVIIa/TF (IC50 = 25 nM) and 6-fold selectivity against
860
S. Kadono et al. / Biochemical and Biophysical Research Communications 326 (2005) 859–865
Fig. 1. Chemical structures of peptide mimetic factor VIIa inhibitors.
thrombin. Crystal structures of compounds 1 [8] and 2 (S. Kadono et al., unpublished data) bound to FVIIa/ sTF revealed that interactions with the S2 and S3 sites of FVIIa/sTF would play an important role in improving selectivity against thrombin. Compounds 1 and 2 have relatively small P3 moieties, D -threonine and D -isoleucine, respectively, which fit the small S3 site of FVIIa/ sTF consisting of Thr99, Pro170I, Trp215, and Gln217 [8]. On the other hand, thrombin has a large and hydrophobic S3 site consisting of Leu99, Ile174, and Trp215 which fits a large P3 moiety, e.g., D -phenylalanine. Therefore, the introduction of the large P3 moiety results in good hydrophobic interactions with thrombin [7–11] and a reduction in selectivity against thrombin [7,8]. Further optimization was continued because compounds 1 and 2 have insufficient selectivity against thrombin. Consequently, as shown in Table 1, compound 3 which has small D -alanine moiety in P3 showed relatively high selectivity against thrombin (>59-fold), whereas the introduction of D -alanine led to low binding affinity for FVIIa/TF (IC50 = 1700 nM). Compound 4,
which has the D -phenylalanine moiety in P3, showed submicromolar inhibition for FVIIa/TF (IC50 = 180 nM), whereas the introduction of D -phenylalanine led to low selectivity against thrombin (13-fold). These data are consistent with the structure–activity relationships of FVIIa/TF selective inhibitors; FVIIa/TF has a small S3 site which fits a small P3 moiety [7,8], whereas thrombin has a large S3 site which fits a large P3 moiety [9–11]. Interestingly, the introduction of the larger P3 moiety, D -biphenylalanine (compound 5, Fig. 1), led to relatively high selectivity against thrombin. This high selectivity is not consistent with the structure–activity relationships of FVIIa/TF selective inhibitors. Therefore, the 3D structural data of compound 5 bound to FVIIa/TF help us to understand new interactions that improve specificity for FVIIa/TF. In this study, we determined the X-ray crystal structure of compound 5 bound to FVIIa/sTF in order to reveal interactions with FVIIa/sTF and the inhibitor. Based on this structure, key interactions for improving selectivity against thrombin, likely a crucial factor in reducing risk of bleeding, and other coagulation serine proteases will be discussed. These results will provide valuable information for the structure-based drug design of selective inhibitors for FVIIa/TF.
Materials and methods Compound synthesis. The synthesis and structure–activity relationship of peptide mimetic FVIIa/TF inhibitors will be reported elsewhere. Inhibition assays. The inhibition activities for human FVIIa/TF and thrombin were measured using chromogenic substrates as described previously [8]. Human factor Xa (FXa) inhibition was measured as follows: A 10% (v/v) DMSO solution of test compound (20 ll) was mixed with 20 ll buffer (500 mM Tris–HCl, pH 8.4, 1500 mM NaCl), 20 ll S-2222 (5 mM, Daiichi Pure Chemical), and 120 ll distilled water in a 96-well assay plate. The reaction was initiated by the addition of 20 ll human factor Xa solution (50 mU/ml, Enzyme Research Laboratories) and the absorbance at 405 nm was then monitored to measure the initial velocity of the reaction. Percent inhibition at each concentration was determined from the experimental and control samples. IC50 was calculated from the concentration–reaction activity curve of each test compound. Human factor XIa (FXIa) inhibition was measured as follows: A 10% (v/v) DMSO solution of test compound (20 ll) was mixed with 100 ll buffer (200 mM Tris–HCl, pH 7.2, 300 mM NaCl), 20 ll S-2366 (2 mM, Daiichi Pure Chemical), and 40 ll distilled water in a 96-well
Table 1 IC50 values of peptide mimetic inhibitors (P4-P3-Gln-p-aminomethylbenzamidine)
2 3 4 5
P3
P4
FVIIa/TF IC50 (nM)
Thrombin IC50 (nM)
FXa IC50 (nM)
FXIa IC50 (nM)
APC IC50 (nM)
D -Ile
SO2Benzyl SO2Et SO2Et SO2Et
25 1700 180 93
150 (6) >100,000 (>59) 2400 (13) 9400 (101)
330 (13) — — 74,300 (799)
35 (1.4) — — 460 (4.9)
83 (3.3) — — 2750 (30)
D -Ala D -Phe D -BiphenylAla
Values in parentheses refer to ratio against FVIIa/TF IC50.
S. Kadono et al. / Biochemical and Biophysical Research Communications 326 (2005) 859–865 assay plate. The reaction was initiated by the addition of 20 ll human factor XIa solution (5 nM, Enzyme Research Laboratories) and the absorbance at 405 nm was then monitored to measure the initial velocity of the reaction. Percent inhibition at each concentration was determined from the experimental and control samples. IC50 was calculated from the concentration–reaction activity curve of each test compound. Human activated protein C (APC) inhibition was measured as follows: A 10% (v/v) DMSO solution of test compound (20 ll) was mixed with 40 ll buffer (200 mM Tris–HCl, pH 8.0), 40 ll S-2366 (2 mM, Daiichi Pure Chemical), 20 ll NaCl (1 M), 20 ll CaCl2 (20 mM), and 40 ll distilled water in a 96-well assay plate. The reaction was initiated by the addition of 20 ll human activated protein C solution (1 lg/ml, Sigma) and the absorbance at 405 nm was then monitored to measure the initial velocity of the reaction. Percent inhibition at each concentration was determined from the experimental and control samples. IC50 was calculated from the concentration–reaction activity curve of each test compound. Coagulation assays. Prothrombin time (PT) was measured as follows: 100 ll human plasma (Dade Behring) containing the test compound was incubated at 37 C for 3 min and was then mixed with 100 ll Thromborel S (Dade Behring). The plasma clotting time was then measured with KC-10A coagulometer (Amelung). The concentration for 2-fold prolongation of PT (2· PT) was calculated from the concentration–reaction activity curve of each test compound. Activated partial thromboplastin time (APTT) was measured as follows: 100 ll human plasma (Dade Behring) containing the test compound was incubated at 37 C for 1 min and was then mixed with 100 ll APTT reagent (SIGMA). After 3 min, the human plasma was mixed with 100 ll CaCl2 (20 mM). The plasma clotting time was then measured with KC-10A coagulometer (Amelung). The concentration for 2-fold prolongation of APTT (2· APTT) was calculated from the concentration–reaction activity curve of each test compound. Computational modeling. The computational model of compound 5 bound to thrombin was built based on the X-ray crystal structure of D Phe-Pro-Arg chloromethyl ketone bound to thrombin (PDB code 1PPB) [9]. After the modification of the P1, P2, P3, and P4 moieties, and the manual docking of compound 5 to the thrombin active site using QUANTA (Accelrys), energy minimization was performed using CHARMm (Accelrys). Cloning, expression, purification, and crystallization. Purified human FVIIa/sTF and crystals of human FVIIa/sTF in complex with compound 5 were prepared as described previously [8,12]. Data collection. After soaking in a cryoprotectant solution of 10% PEG 5000, 100 mM cacodylate buffer, pH 5.0, 100 mM NaCl, 5 mM CaCl2, and 30% (v/v) glycerol, the crystal was frozen at 100 K in a nitrogen gas stream. X-ray diffraction data on the FVIIa/sTF crystal in complex with compound 5 were collected on an R-axis IV (RIGAKU) mounted on a copper rotating-anode X-ray generator ultraX 18 (RIGAKU) equipped with a Confocal Mirror (Osmic). The data were processed using DENZO and SCALEPACK [13]. Structure determination and refinement. The crystal of FVIIa/sTF in complex with compound 5 is isomorphous with the crystal of FVIIa/ sTF in complex with D -Phe-Phe-Arg chloromethyl ketone [12]. The model phases of FVIIa/sTF bound to compound 5 were improved by rigid body refinement with CNX (Accelrys), using the previously published structure of FVIIa/sTF in complex with D -Phe-Phe-Arg chloromethyl ketone (PDB code 1DAN) [14] as a starting model. The inhibitor molecule was identified by the difference Fourier method. Model building of the protein and inhibitor was performed with QUANTA (Accelrys) and the structure was refined with CNX (Accelrys). X-ray diffraction data collection and refinement statistics are shown in Table 2. Atomic coordinates of FVIIa/sTF bound to compound 5 have been deposited with the RCSB Protein Data Bank [15], with Accession code 1WTG.
861
Table 2 X-ray data collection and refinement statistics FVIIa/sTF/compound 2 Data collection Space group ˚) Cell parameters (A a b c ˚) Resolution (A Total reflections Unique reflections Completeness (final shell) (%) Rmergea (final shell) (%)
P212121 71.28 82.32 123.38 50.0–2.2 115,301 36,773 97.7 (94.9) 7.9 (22.8)
Refinement ˚) Resolution (A 20.0–2.2 Reflections 36,205 R-factorb (Rfree)c (%) 20.4 (25.1) Rms deviation from ideal ˚) Bond lengths (A 0.005 Bond angles () 1.32 P P P a Rmerge = hkl j| (hkl) ÆI (hkl)æ|/ hkl|Ij (hkl)|, where Ij (hkl) and |I (hkl)| are the intensity of measurement j and the mean intensity for the reflection with indices hkl, respectively. P P b R-factor = hkl|Fcalc (hkl)| |Fobs (hkl)|/ hkl|Fobs (hkl)|, where Fobs and Fcalc are the observed and calculated structure factor amplitudes, respectively. c Rfree is calculated with 7% of the reflection set aside randomly.
Results and discussion Inhibition activity and selectivity of compound 5 As shown in Table 1, compound 5, in spite of its large P3 moiety, showed relatively high selectivity against thrombin (101-fold) and maintained submicromolar inhibition activity for FVIIa/TF (IC50 = 93 nM). The introduction of the D -biphenylalanine moiety in P3 and the ethylsulfonamide moiety in P4 led to 17-fold improvement of selectivity against thrombin compared with that of compound 2 which has the relatively small P3 moiety D -isoleucine and the large P4 moiety benzylsulfonamide. This improvement of selectivity is not consistent with previously reported structure–activity relationship studies of factor VIIa selective inhibitors [7,8]. The blood coagulation cascade is a complex of many serine proteases including factor VIIa, thrombin, factors IXa, Xa, XIa (FXIa), XIIa (FXIIa), activated protein C (APC), and so on [1]. A lack of selectivity against coagulation serine proteases other than thrombin likely leads to a reduction of selectivity for extrinsic coagulation. Therefore, selectivity against FIXa, FXa, FXIa, FXIIa, and APC was evaluated using chromogenic assays. Of these serine proteases, FIXa and FXIIa were slightly inhibited by compounds 2 and 5 (IC50 > 100 lM). Selectivity against FXa, FXIa, and APC is improved by the introduction of the D -biphenylalanine in P3 and the ethylsulfonamide moiety in P4 (Table 1). The improvement of selectivity of compound 5 against FXa, FXIa, and
862
S. Kadono et al. / Biochemical and Biophysical Research Communications 326 (2005) 859–865
APC over compound 2 is 61-, 2.5-, and 8.9-fold, respectively. Especially, the improvement of selectivity against FXIa is low. In addition, the selectivity of compound 5 for extrinsic coagulation inhibition was evaluated with in vitro assays using human plasma, the prothrombin time (PT) assay, and the activated partial thromboplastin time (APTT) assay in order to confirm effects on blood coagulation from the improvement of specificity for FVIIa/TF inhibition. Theoretically, completely specific inhibition on FVIIa/TF, which blocks only extrinsic coagulation, should result in prolongation of only PT without affecting APTT, thus making the concentration ratio for 2fold prolongation of APTT and PT (2· APTT/2· PT) infinitely large. The measured 2· APTT/2· PT ratio of compound 5 was 6.2, which is higher than that of a
thrombin selective inhibitor, Argatroban [16] (0.4, inhouse data), a FXa selective inhibitor, DX-9065a [17] (2.2, in-house data), and compound 1 (2.7). This means that the introduction of D -biphenylalanine in P3 and ethylsulfonamide in P4 leads to the improvement of selectivity for extrinsic coagulation inhibition by improving specificity for FVIIa/TF inhibition. Structure of FVIIa/sTF in complex with compound 5 As shown in Fig. 2A, compound 5 binds to FVIIa/sTF at the S1, S2, and S3 sites. The amidino group of the benzamidine moiety in P1 makes a salt bridge with the side chain of Asp189, and is stabilized further by hydrogen bonds with the main chain of Gly219 and the side chain of Ser190 in the S1 site of FVIIa/sTF. The peptidyl chain
Fig. 2. (A) Crystal structure of compound 5 bound to factor VIIa/sTF with the Fo Fc omit map contoured at 2.5r. Carbons of compound 5 are shown in green. The coordinates of the 170-loop, Lys192, and Gln217 in the crystal structure of FVIIa/sTF bound to compound 1 are superposed (red). (B) Computational model of compound 5 bound to thrombin. Carbons of the computational model of compound 5 are shown in orange. The coordinates of compound 5, the 170-loop, Lys192, and Gln217 in the crystal structure of FVIIa/sTF bound to compound 2, are superposed (green). This figure was created using PyMol (Delano Scientific, http://www.pymol.org).
S. Kadono et al. / Biochemical and Biophysical Research Communications 326 (2005) 859–865
atoms of compound 5 make hydrogen bonds with the main chain of Ser214 and Gly216, forming a short antiparallel b-sheet. In the S2 site of FVIIa/sTF, the side chain of Asp60 makes a hydrophilic pocket with the side chain oxygen of Tyr94 and the main chain oxygen of Thr98, which is a unique feature in the active site of FVIIa/ sTF. The amide group of the glutamine side chain in P2 of compound 5 makes strong hydrogen bonds with the side chain carboxylate of Asp60, the side chain oxygen of Tyr94, and the main chain oxygen of Thr98. The large P3 moiety, D -biphenylalanine, not only occupies the S3 site of FVIIa/sTF but also makes additional interactions with the 170-loop of FVIIa/sTF, which is composed of residues Gln170A-Ser170H, accompanying ligand-induced conformational changes of the 170-loop and the Gln217 side chain from the structure of FVIIa/ sTF bound to compound 1 which has the small P3 moiety, D -threonine [8]. One of the benzene rings of D -biphenylalanine near the Ca atom mainly makes hydrophobic interactions with Pro170I and Gln217; the other distal benzene ring of D -biphenylalanine from the Ca atom makes hydrophobic interactions with Val170E, Gly170F, Asp170G, and Ser170H of the 170-loop. This is the first case in which interactions between inhibitors and Val170E, Gly170F, Asp170G, and Ser170H of the 170-loop have been observed. In addition, a clear electron density of the residues 170D–170H in the 170-loop is observed, whereas a weak electron density of the residues 170D–170H in the 170-loop was observed in the crystal structure of FVIIa/sTF in complex with compound 1 [8]. This loop is known to be flexible [18] and to show a conformational change when FVIIa is activated from FVII [19] and, thus, the ligand-induced conformational changes of the 170-loop and Gln217 appear to arise easily by the occupation of the large P3 moiety. The ethyl group in P4 of compound 5 does not occupy the S1-sub site of FVIIa/sTF and loses hydrophobic interactions with the S1-sub site consisting of a Cys191-Cys220 disulfide bridge, Gln143, Asp146, Gly219, and Lys192, whereas the propyl group in P4 of compound 1 occupies the S1-sub site of FVIIa/sTF and makes hydrophobic interactions [8]. Instead, the ethyl group in P4 makes intramolecular CH–p interactions with the biphenyl group in P3. In addition, one of the oxygen atoms of the sulfonamide moiety in P4 not only makes a hydrogen bond with the main chain nitrogen of Gly219, which is conserved in the binding of compound 1 to FVIIa/sTF
863
[8], but also makes a new charge–dipole interaction with Lys192 of FVIIa/sTF accompanying a ligand-induced conformational change of Lys192 from the structure of FVIIa/sTF bound to compound 1. Structural basis for selectivity against thrombin The crystal structure of compound 5 bound to FVIIa/sTF revealed that the D -biphenylalanine moiety in P3 makes novel interactions with the 170-loop and that the ethylsulfonamide moiety in P4 adopts a novel conformation and makes an additional charge–dipole interaction with Lys192 of FVIIa/sTF. The contacts on the large area of the S3 site and the 170-loop would lead to the relatively potent binding affinity of compound 5 for FVIIa/TF. The charge–dipole interaction with Lys192 is expected to be energetically more preferable than hydrophobic interactions between the less hydrophobic ethyl group and the S1-sub site of FVIIa/sTF. The modeling study of compound 5 bound to thrombin suggests that the D -biphenylalanine moiety in P3 of compound 5 is too large to sit in the S3 site of thrombin. Thus, the conformation of D -biphenylalanine bound to thrombin is probably the same as that bound to FVIIa/sTF (Fig. 2B). The benzene ring near the Ca atom, which forms hydrophobic interactions with Pro170I and Gln217 in the S3 site of FVIIa/sTF, is expected to make similar hydrophobic interactions with Ile174 and Glu217 in the S3 site of thrombin. However, the distal benzene ring, which interacts with the 170loop of FVIIa/sTF, cannot make similar interactions with thrombin because thrombin does not have the long amino acid insertion corresponding to the 170-loop of FVIIa (Figs. 2B and 3). Thus, this difference in the length of the 170-loop would result in a difference in the contact area from the D -biphenylalanine moiety in the S3 sites of FVIIa/sTF and thrombin, and this difference in contact area would lead to a relative reduction of the binding affinity for thrombin and an improvement in selectivity against thrombin. In addition, thrombin has Glu192 as the residue corresponding to Lys192 of FVIIa/sTF and, thus, thrombin is not expected to be able to make such a charge–dipole interaction with the sulfonamide group (Figs. 2B and 3). Thrombin is expected to make hydrophobic interactions with the less hydrophobic ethyl group in the S1-sub site,
Fig. 3. Sequence alignment of human blood coagulation serine proteases around the 170-loop and residue 192.
864
S. Kadono et al. / Biochemical and Biophysical Research Communications 326 (2005) 859–865
which may be weaker than the charge–dipole interaction between Lys192 of FVIIa/TF and the sulfonamide group. Thus, the difference in the residue at position 192 would result in a difference in the binding mode of the P4 moiety in FVIIa/TF and thrombin, and this binding mode difference may lead to a relative reduction of the binding affinity for thrombin and an improvement in selectivity against thrombin.
proteases suggested that these interactions would be characteristic of binding with FVIIa/TF and be responsible for the improvement of selectivity against a wide variety of coagulation serine proteases, which would result in desirable improvement of selectivity for extrinsic coagulation inhibition, which may lead to low risk of breeding. This structural information would be important for the rational and rapid development of specific inhibitors for FVIIa/TF.
Structural basis for selectivity against coagulation serine proteases other than thrombin Acknowledgments Amino acid sequence comparisons among coagulation serine proteases show that FVIIa has the longest amino acid insertion at the 170-loop (Fig. 3). Coagulation serine proteases other than FVIIa do not have residues corresponding to Val170E, Gly170F, or Asp170G in FVIIa, which make hydrophobic interactions with the distal benzene ring of the D -biphenylalanine moiety in P3. These coagulation serine proteases will have difficulty making hydrophobic interactions with the distal phenyl ring of the D -biphenylalanine moiety similar to thrombin. Consequently, the introduction of the D biphenylalanine moiety in P3 would lead to the improvement of selectivity not only against thrombin but also against other coagulation serine proteases. In addition, Lys192 of FVIIa, which makes a charge– dipole interaction with the sulfonamide group in P4, is not conserved among coagulation serine proteases (Fig. 3). Lys192 of FVIIa is changed to Glu or Gln in serine proteases other than FVIIa and FXIa. For these coagulation serine proteases, other than FVIIa and FXIa, it is difficult to make the charge–dipole interaction with the sulfonamide group similar to thrombin. Consequently, the introduction of the small P4 moiety, ethylsulfonamide, likely leads to the improvement of selectivity not only against thrombin but also against coagulation serine proteases other than FXIa. A slight improvement of the selectivity of compound 5 against FXIa over compound 2 is likely responsible for the existence of Lys192 in FXIa which can make the charge-dipole interaction with the sulfonamide group in P4.
Conclusion X-ray crystal structure analysis of ethylsulfonamidebound to FVIIa/sTF revealed that the large P3 moiety D -biphenylalanine and the small P4 moiety ethylsulfonamide make novel interactions with the 170-loop and Lys192, respectively, accompanying ligand-induced conformational changes of the 170-loop, Lys192, and Gln217 and a novel binding conformation of the ethylsulfonamide moiety. Structural comparisons with thrombin and amino acid comparisons among coagulation serine
D -biphenylalanine-Gln-p-aminobenzamidine
We thank K. Esaki for measuring inhibition activities and F. Ford for proofreading the manuscript. References [1] E.W. Davie, K. Fujikawa, W. Kisiel, The coagulation cascade: initiation, maintenance, and regulation, Biochemistry 30 (1991) 10363–10370. [2] E. Braunwald, R.M. Califf, C.P. Cannon, K.A. Fox, V. Fuster, W.B. Gibler, R.A. Harrington, S.B. King III, N.S. Kleiman, P. Theroux, E.J. Topol, F. Van de Werf, H.D. White, J.T. Willerson, Redefining medical treatment in the management of unstable angina, Am. J. Med. 108 (2000) 41–53. [3] L.A. Harker, S.R. Hanson, J.N. Wilcox, A.B. Kelly, Antithrombotic and antilesion benefits without hemorrhagic risks by inhibiting tissue factor pathway, Haemostasis 26 (Suppl. 1) (1996) 76–82. [4] J. Himber, D. Kirchhofer, M. Riederer, T.B. Tschopp, B. Steiner, S.P. Roux, Dissociation of antithrombotic effect and bleeding time prolongation in rabbits by inhibiting tissue factor function, Thromb. Haemost. 78 (1997) 1142–1149. [5] J.A. Szalony, B.B. Taite, T.J. Girard, N.S. Nicholson, R.M. LaChance, Pharmacological intervention at disparate sites in the coagulation cascade: comparison of anti-thrombotic efficacy vs. bleeding propensity in a rat model of acute arterial thrombosis, J. Thromb. Thrombolysis 14 (2002) 113–121. [6] O. Klingler, H. Matter, M. Schudok, M. Donghi, J. Czech, M. Lorenz, H.P. Nestler, H. Szillat, H. Schreuder, Structure-based design of amidinophenylurea-derivatives for factor VIIa inhibition, Bioorg. Med. Chem. Lett. 14 (2004) 3715–3720. [7] J.J. Parlow, B.L. Case, T.A. Dice, R.L. Fenton, M.J. Hayes, D.E. Jones, W.L. Neumann, R.S. Wood, R.M. Lachance, T.J. Girard, N.S. Nicholson, M. Clare, R.A. Stegeman, A.M. Stevens, W.C. Stallings, R.G. Kurumbail, M.S. South, Design, parallel synthesis, and crystal structures of pyrazinone antithrombotics as selective inhibitors of the tissue factor VIIa complex, J. Med. Chem. 46 (2003) 4050–4062. [8] S. Kadono, A. Sakamoto, Y. Kikuchi, M. Oh-eda, N. Yabuta, T. Koga, K. Hattori, T. Shiraishi, M. Haramura, H. Kodama, T. Esaki, H. Sato, Y. Watanabe, S. Itoh, M. Ohta, T. Kozono, Crystal structure of human factor VIIa/tissue factor in complex with peptide mimetic inhibitor, Biochem. Biophys. Res. Commun. 324 (2004) 1227–1233. [9] W. Bode, I. Mayr, U. Baumann, R. Huber, S.R. Stone, J. ˚ crystal structure of human alphaHofsteenge, The refined 1.9 A thrombin: interaction with D -Phe-Pro-Arg chloromethylketone and significance of the Tyr-Pro-Pro-Trp insertion segment, EMBO J. 8 (1989) 3467–3475. [10] D.W. Banner, P. Hadvary, Crystallographic analysis at 3.0-A resolution of the binding to human thrombin of four active sitedirected inhibitors, J. Biol. Chem. 266 (1991) 20085–20093.
S. Kadono et al. / Biochemical and Biophysical Research Communications 326 (2005) 859–865 [11] M.R. Wiley, N.Y. Chirgadze, D.K. Clawson, T.J. Craft, D.S. Gifford-Moore, N.D. Jones, J.L. Olkowski, L.C. Weir, G.F. Smith, D -Phe-Pro-p-amidinobenzylamine: a potent and highly selective thrombin inhibitor, Bioorg. Med. Chem. Lett. 6 (1996) 2387–2392. [12] D. Kirchhofer, A. Guha, Y. Nemerson, W.H. Konigsberg, F. Vilbois, C. Chene, D.W. Banner, A. DÕArcy, Activation of blood coagulation factor VIIa with cleaved tissue factor extracellular domain and crystallization of the active complex, Proteins 22 (1995) 419–425. [13] Z. Otwinowski, W. Minor, Processing of X-ray diffraction data collected in oscillation mode, Methods Enzymol. 276 (1997) 307– 326. [14] D.W. Banner, A. DÕArcy, C. Chene, F.K. Winkler, A. Guha, W.H. Konigsberg, Y. Nemerson, D. Kirchhofer, The crystal structure of the complex of blood coagulation factor VIIa with soluble tissue factor, Nature 380 (1996) 41–46. [15] H.M. Berman, J. Westbrook, Z. Feng, G. Gilliland, T.N. Bhat, H. Weissig, I.N. Shindyalov, P.E. Bourne, The Protein Data Bank, Nucleic Acids Res. 28 (2000) 235–242.
865
[16] R. Kikumoto, Y. Tamao, T. Tezuka, S. Tonomura, H. Hara, K. Ninomiya, A. Hijikata, S. Okamoto, Selective inhibition of thrombin by (2R,4R)-4-methyl-1-[N2-[(3-methyl-1,2,3,4-tetrahydro-8-quinolinyl+++)sulfonyl]-L -arginyl]-2-piperidinecarboxylic acid, Biochemistry 23 (1984) 85–90. [17] S. Katakura, T. Nagahara, T. Hara, M. Iwamoto, A novel factor Xa inhibitor: structure-activity relationships and selectivity between factor Xa and thrombin, Biochem. Biophys. Res. Commun. 197 (1993) 965–972. [18] K. Sichler, D.W. Banner, A. DÕArcy, K.P. Hopfner, R. Huber, W. Bode, G.B. Kresse, E. Kopetzki, H. Brandstetter, Crystal structures of uninhibited factor VIIa link its cofactor and substrate-assisted activation to specific interactions, J. Mol. Biol. 322 (2002) 591–603. [19] C. Eigenbrot, D. Kirchhofer, M.S. Dennis, L. Santell, R.A. Lazarus, J. Stamos, M.H. Ultsch, The factor VII zymogen structure reveals reregistration of beta strands during activation, Structure (Camb.) 9 (2001) 627–636.
Novel interactions of large P3 moiety and small P4 moiety in the binding of the peptide mimetic factor VIIa inhibitorq Shojiro Kadono*, Akihisa Sakamoto, Yasufumi Kikuchi, Masayoshi Oh-eda, Naohiro Yabuta, Kazutaka Yoshihashi, Takehisa Kitazawa, Tsukasa Suzuki, Takaki Koga, Kunihiro Hattori, Takuya Shiraishi, Masayuki Haramura, Hirofumi Kodama, Yoshiyuki Ono, Toru Esaki, Haruhiko Sato, Yoshiaki Watanabe, Susumu Itoh, Masateru Ohta, Toshiro Kozono Fuji Gotemba Research Labs, Chugai Pharmaceutical Co., Ltd., 1-135 Komakado, Gotemba, Shizuoka 412-8513, Japan Received 12 November 2004 Available online 7 December 2004
Abstract Selective factor VIIa-tissue factor complex (FVIIa/TF) inhibition is seen as a promising target for developing new anticoagulant drugs. A novel peptide mimetic factor VIIa inhibitor, ethylsulfonamide-D -biphenylalanine-Gln-p-aminobenzamidine, shows 100fold selectivity against thrombin in spite of its large P3 moiety, unlike previously reported FVIIa/TF selective inhibitors. X-ray crystal structure analysis reveals that the large P3 moiety, D -biphenylalanine, and the small P4 moiety, ethylsulfonamide, make novel interactions with the 170-loop and Lys192 of FVIIa/TF, respectively, accompanying ligand-induced conformational changes of the 170-loop, Gln217, and Lys192. Structural comparisons of FVIIa with thrombin and amino acid sequence comparisons among coagulation serine proteases suggest that these interactions play an important role in achieving selective inhibition for FVIIa/TF. 2004 Elsevier Inc. All rights reserved. Keywords: Factor VIIa; Blood coagulation; Serine protease; X-ray crystallography; Drug design
The blood coagulation cascade is divided into extrinsic and intrinsic coagulation pathways. Factor VIIa (FVIIa) in complex with tissue factor (TF) initiates the extrinsic coagulation pathway. This complex activates factors IX to IXa (FIXa) and X to Xa (FXa), which in turn activate factor X to Xa and prothrombin to thrombin, respectively [1]. Thrombin cleaves fibrinogen to fibrin, which forms blood clots with activated platelets. Inappropriate thrombus formation in blood vessels causes cardiovascular diseases (myocardial infarction, stroke, pulmonary embolism, and so on), which are the most common q Abbreviations: FVIIa, factor VIIa; TF, tissue factor; sTF, soluble domain of tissue factor; FIXa, factor IXa; FXa, factor Xa; FXIa, factor XIa; FXIIa, factor XIIa; APC, activated protein C. * Corresponding author. Fax: +81 550 87 5326. E-mail address: [email protected] (S. Kadono).
0006-291X/$ - see front matter 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.bbrc.2004.11.108
causes of mortality in the industrialized world [2]. Recent studies on blood coagulation have suggested that selective inhibition of extrinsic coagulation provides effective anticoagulation and low risk of bleeding compared with other antithrombotic mechanisms [3–5]. Thus, specific FVIIa/TF complex inhibition, which blocks only extrinsic coagulation, is seen as a promising target for developing new anticoagulant drugs [6,7]. Recently, we reported on propylsulfonamide-D -ThrMet-p-aminobenzamidine (compound 1, Fig. 1), which showed submicromolar inhibition for FVIIa/TF (IC50 = 130 nM) and 26-fold selectivity against thrombin [8]. In addition, another peptide mimetic inhibitor, benzylsulfonamide-D -Ile-Gln-p-aminobenzamidine (compound 2, Fig. 1), showed potent inhibition for FVIIa/TF (IC50 = 25 nM) and 6-fold selectivity against
860
S. Kadono et al. / Biochemical and Biophysical Research Communications 326 (2005) 859–865
Fig. 1. Chemical structures of peptide mimetic factor VIIa inhibitors.
thrombin. Crystal structures of compounds 1 [8] and 2 (S. Kadono et al., unpublished data) bound to FVIIa/ sTF revealed that interactions with the S2 and S3 sites of FVIIa/sTF would play an important role in improving selectivity against thrombin. Compounds 1 and 2 have relatively small P3 moieties, D -threonine and D -isoleucine, respectively, which fit the small S3 site of FVIIa/ sTF consisting of Thr99, Pro170I, Trp215, and Gln217 [8]. On the other hand, thrombin has a large and hydrophobic S3 site consisting of Leu99, Ile174, and Trp215 which fits a large P3 moiety, e.g., D -phenylalanine. Therefore, the introduction of the large P3 moiety results in good hydrophobic interactions with thrombin [7–11] and a reduction in selectivity against thrombin [7,8]. Further optimization was continued because compounds 1 and 2 have insufficient selectivity against thrombin. Consequently, as shown in Table 1, compound 3 which has small D -alanine moiety in P3 showed relatively high selectivity against thrombin (>59-fold), whereas the introduction of D -alanine led to low binding affinity for FVIIa/TF (IC50 = 1700 nM). Compound 4,
which has the D -phenylalanine moiety in P3, showed submicromolar inhibition for FVIIa/TF (IC50 = 180 nM), whereas the introduction of D -phenylalanine led to low selectivity against thrombin (13-fold). These data are consistent with the structure–activity relationships of FVIIa/TF selective inhibitors; FVIIa/TF has a small S3 site which fits a small P3 moiety [7,8], whereas thrombin has a large S3 site which fits a large P3 moiety [9–11]. Interestingly, the introduction of the larger P3 moiety, D -biphenylalanine (compound 5, Fig. 1), led to relatively high selectivity against thrombin. This high selectivity is not consistent with the structure–activity relationships of FVIIa/TF selective inhibitors. Therefore, the 3D structural data of compound 5 bound to FVIIa/TF help us to understand new interactions that improve specificity for FVIIa/TF. In this study, we determined the X-ray crystal structure of compound 5 bound to FVIIa/sTF in order to reveal interactions with FVIIa/sTF and the inhibitor. Based on this structure, key interactions for improving selectivity against thrombin, likely a crucial factor in reducing risk of bleeding, and other coagulation serine proteases will be discussed. These results will provide valuable information for the structure-based drug design of selective inhibitors for FVIIa/TF.
Materials and methods Compound synthesis. The synthesis and structure–activity relationship of peptide mimetic FVIIa/TF inhibitors will be reported elsewhere. Inhibition assays. The inhibition activities for human FVIIa/TF and thrombin were measured using chromogenic substrates as described previously [8]. Human factor Xa (FXa) inhibition was measured as follows: A 10% (v/v) DMSO solution of test compound (20 ll) was mixed with 20 ll buffer (500 mM Tris–HCl, pH 8.4, 1500 mM NaCl), 20 ll S-2222 (5 mM, Daiichi Pure Chemical), and 120 ll distilled water in a 96-well assay plate. The reaction was initiated by the addition of 20 ll human factor Xa solution (50 mU/ml, Enzyme Research Laboratories) and the absorbance at 405 nm was then monitored to measure the initial velocity of the reaction. Percent inhibition at each concentration was determined from the experimental and control samples. IC50 was calculated from the concentration–reaction activity curve of each test compound. Human factor XIa (FXIa) inhibition was measured as follows: A 10% (v/v) DMSO solution of test compound (20 ll) was mixed with 100 ll buffer (200 mM Tris–HCl, pH 7.2, 300 mM NaCl), 20 ll S-2366 (2 mM, Daiichi Pure Chemical), and 40 ll distilled water in a 96-well
Table 1 IC50 values of peptide mimetic inhibitors (P4-P3-Gln-p-aminomethylbenzamidine)
2 3 4 5
P3
P4
FVIIa/TF IC50 (nM)
Thrombin IC50 (nM)
FXa IC50 (nM)
FXIa IC50 (nM)
APC IC50 (nM)
D -Ile
SO2Benzyl SO2Et SO2Et SO2Et
25 1700 180 93
150 (6) >100,000 (>59) 2400 (13) 9400 (101)
330 (13) — — 74,300 (799)
35 (1.4) — — 460 (4.9)
83 (3.3) — — 2750 (30)
D -Ala D -Phe D -BiphenylAla
Values in parentheses refer to ratio against FVIIa/TF IC50.
S. Kadono et al. / Biochemical and Biophysical Research Communications 326 (2005) 859–865 assay plate. The reaction was initiated by the addition of 20 ll human factor XIa solution (5 nM, Enzyme Research Laboratories) and the absorbance at 405 nm was then monitored to measure the initial velocity of the reaction. Percent inhibition at each concentration was determined from the experimental and control samples. IC50 was calculated from the concentration–reaction activity curve of each test compound. Human activated protein C (APC) inhibition was measured as follows: A 10% (v/v) DMSO solution of test compound (20 ll) was mixed with 40 ll buffer (200 mM Tris–HCl, pH 8.0), 40 ll S-2366 (2 mM, Daiichi Pure Chemical), 20 ll NaCl (1 M), 20 ll CaCl2 (20 mM), and 40 ll distilled water in a 96-well assay plate. The reaction was initiated by the addition of 20 ll human activated protein C solution (1 lg/ml, Sigma) and the absorbance at 405 nm was then monitored to measure the initial velocity of the reaction. Percent inhibition at each concentration was determined from the experimental and control samples. IC50 was calculated from the concentration–reaction activity curve of each test compound. Coagulation assays. Prothrombin time (PT) was measured as follows: 100 ll human plasma (Dade Behring) containing the test compound was incubated at 37 C for 3 min and was then mixed with 100 ll Thromborel S (Dade Behring). The plasma clotting time was then measured with KC-10A coagulometer (Amelung). The concentration for 2-fold prolongation of PT (2· PT) was calculated from the concentration–reaction activity curve of each test compound. Activated partial thromboplastin time (APTT) was measured as follows: 100 ll human plasma (Dade Behring) containing the test compound was incubated at 37 C for 1 min and was then mixed with 100 ll APTT reagent (SIGMA). After 3 min, the human plasma was mixed with 100 ll CaCl2 (20 mM). The plasma clotting time was then measured with KC-10A coagulometer (Amelung). The concentration for 2-fold prolongation of APTT (2· APTT) was calculated from the concentration–reaction activity curve of each test compound. Computational modeling. The computational model of compound 5 bound to thrombin was built based on the X-ray crystal structure of D Phe-Pro-Arg chloromethyl ketone bound to thrombin (PDB code 1PPB) [9]. After the modification of the P1, P2, P3, and P4 moieties, and the manual docking of compound 5 to the thrombin active site using QUANTA (Accelrys), energy minimization was performed using CHARMm (Accelrys). Cloning, expression, purification, and crystallization. Purified human FVIIa/sTF and crystals of human FVIIa/sTF in complex with compound 5 were prepared as described previously [8,12]. Data collection. After soaking in a cryoprotectant solution of 10% PEG 5000, 100 mM cacodylate buffer, pH 5.0, 100 mM NaCl, 5 mM CaCl2, and 30% (v/v) glycerol, the crystal was frozen at 100 K in a nitrogen gas stream. X-ray diffraction data on the FVIIa/sTF crystal in complex with compound 5 were collected on an R-axis IV (RIGAKU) mounted on a copper rotating-anode X-ray generator ultraX 18 (RIGAKU) equipped with a Confocal Mirror (Osmic). The data were processed using DENZO and SCALEPACK [13]. Structure determination and refinement. The crystal of FVIIa/sTF in complex with compound 5 is isomorphous with the crystal of FVIIa/ sTF in complex with D -Phe-Phe-Arg chloromethyl ketone [12]. The model phases of FVIIa/sTF bound to compound 5 were improved by rigid body refinement with CNX (Accelrys), using the previously published structure of FVIIa/sTF in complex with D -Phe-Phe-Arg chloromethyl ketone (PDB code 1DAN) [14] as a starting model. The inhibitor molecule was identified by the difference Fourier method. Model building of the protein and inhibitor was performed with QUANTA (Accelrys) and the structure was refined with CNX (Accelrys). X-ray diffraction data collection and refinement statistics are shown in Table 2. Atomic coordinates of FVIIa/sTF bound to compound 5 have been deposited with the RCSB Protein Data Bank [15], with Accession code 1WTG.
861
Table 2 X-ray data collection and refinement statistics FVIIa/sTF/compound 2 Data collection Space group ˚) Cell parameters (A a b c ˚) Resolution (A Total reflections Unique reflections Completeness (final shell) (%) Rmergea (final shell) (%)
P212121 71.28 82.32 123.38 50.0–2.2 115,301 36,773 97.7 (94.9) 7.9 (22.8)
Refinement ˚) Resolution (A 20.0–2.2 Reflections 36,205 R-factorb (Rfree)c (%) 20.4 (25.1) Rms deviation from ideal ˚) Bond lengths (A 0.005 Bond angles () 1.32 P P P a Rmerge = hkl j| (hkl) ÆI (hkl)æ|/ hkl|Ij (hkl)|, where Ij (hkl) and |I (hkl)| are the intensity of measurement j and the mean intensity for the reflection with indices hkl, respectively. P P b R-factor = hkl|Fcalc (hkl)| |Fobs (hkl)|/ hkl|Fobs (hkl)|, where Fobs and Fcalc are the observed and calculated structure factor amplitudes, respectively. c Rfree is calculated with 7% of the reflection set aside randomly.
Results and discussion Inhibition activity and selectivity of compound 5 As shown in Table 1, compound 5, in spite of its large P3 moiety, showed relatively high selectivity against thrombin (101-fold) and maintained submicromolar inhibition activity for FVIIa/TF (IC50 = 93 nM). The introduction of the D -biphenylalanine moiety in P3 and the ethylsulfonamide moiety in P4 led to 17-fold improvement of selectivity against thrombin compared with that of compound 2 which has the relatively small P3 moiety D -isoleucine and the large P4 moiety benzylsulfonamide. This improvement of selectivity is not consistent with previously reported structure–activity relationship studies of factor VIIa selective inhibitors [7,8]. The blood coagulation cascade is a complex of many serine proteases including factor VIIa, thrombin, factors IXa, Xa, XIa (FXIa), XIIa (FXIIa), activated protein C (APC), and so on [1]. A lack of selectivity against coagulation serine proteases other than thrombin likely leads to a reduction of selectivity for extrinsic coagulation. Therefore, selectivity against FIXa, FXa, FXIa, FXIIa, and APC was evaluated using chromogenic assays. Of these serine proteases, FIXa and FXIIa were slightly inhibited by compounds 2 and 5 (IC50 > 100 lM). Selectivity against FXa, FXIa, and APC is improved by the introduction of the D -biphenylalanine in P3 and the ethylsulfonamide moiety in P4 (Table 1). The improvement of selectivity of compound 5 against FXa, FXIa, and
862
S. Kadono et al. / Biochemical and Biophysical Research Communications 326 (2005) 859–865
APC over compound 2 is 61-, 2.5-, and 8.9-fold, respectively. Especially, the improvement of selectivity against FXIa is low. In addition, the selectivity of compound 5 for extrinsic coagulation inhibition was evaluated with in vitro assays using human plasma, the prothrombin time (PT) assay, and the activated partial thromboplastin time (APTT) assay in order to confirm effects on blood coagulation from the improvement of specificity for FVIIa/TF inhibition. Theoretically, completely specific inhibition on FVIIa/TF, which blocks only extrinsic coagulation, should result in prolongation of only PT without affecting APTT, thus making the concentration ratio for 2fold prolongation of APTT and PT (2· APTT/2· PT) infinitely large. The measured 2· APTT/2· PT ratio of compound 5 was 6.2, which is higher than that of a
thrombin selective inhibitor, Argatroban [16] (0.4, inhouse data), a FXa selective inhibitor, DX-9065a [17] (2.2, in-house data), and compound 1 (2.7). This means that the introduction of D -biphenylalanine in P3 and ethylsulfonamide in P4 leads to the improvement of selectivity for extrinsic coagulation inhibition by improving specificity for FVIIa/TF inhibition. Structure of FVIIa/sTF in complex with compound 5 As shown in Fig. 2A, compound 5 binds to FVIIa/sTF at the S1, S2, and S3 sites. The amidino group of the benzamidine moiety in P1 makes a salt bridge with the side chain of Asp189, and is stabilized further by hydrogen bonds with the main chain of Gly219 and the side chain of Ser190 in the S1 site of FVIIa/sTF. The peptidyl chain
Fig. 2. (A) Crystal structure of compound 5 bound to factor VIIa/sTF with the Fo Fc omit map contoured at 2.5r. Carbons of compound 5 are shown in green. The coordinates of the 170-loop, Lys192, and Gln217 in the crystal structure of FVIIa/sTF bound to compound 1 are superposed (red). (B) Computational model of compound 5 bound to thrombin. Carbons of the computational model of compound 5 are shown in orange. The coordinates of compound 5, the 170-loop, Lys192, and Gln217 in the crystal structure of FVIIa/sTF bound to compound 2, are superposed (green). This figure was created using PyMol (Delano Scientific, http://www.pymol.org).
S. Kadono et al. / Biochemical and Biophysical Research Communications 326 (2005) 859–865
atoms of compound 5 make hydrogen bonds with the main chain of Ser214 and Gly216, forming a short antiparallel b-sheet. In the S2 site of FVIIa/sTF, the side chain of Asp60 makes a hydrophilic pocket with the side chain oxygen of Tyr94 and the main chain oxygen of Thr98, which is a unique feature in the active site of FVIIa/ sTF. The amide group of the glutamine side chain in P2 of compound 5 makes strong hydrogen bonds with the side chain carboxylate of Asp60, the side chain oxygen of Tyr94, and the main chain oxygen of Thr98. The large P3 moiety, D -biphenylalanine, not only occupies the S3 site of FVIIa/sTF but also makes additional interactions with the 170-loop of FVIIa/sTF, which is composed of residues Gln170A-Ser170H, accompanying ligand-induced conformational changes of the 170-loop and the Gln217 side chain from the structure of FVIIa/ sTF bound to compound 1 which has the small P3 moiety, D -threonine [8]. One of the benzene rings of D -biphenylalanine near the Ca atom mainly makes hydrophobic interactions with Pro170I and Gln217; the other distal benzene ring of D -biphenylalanine from the Ca atom makes hydrophobic interactions with Val170E, Gly170F, Asp170G, and Ser170H of the 170-loop. This is the first case in which interactions between inhibitors and Val170E, Gly170F, Asp170G, and Ser170H of the 170-loop have been observed. In addition, a clear electron density of the residues 170D–170H in the 170-loop is observed, whereas a weak electron density of the residues 170D–170H in the 170-loop was observed in the crystal structure of FVIIa/sTF in complex with compound 1 [8]. This loop is known to be flexible [18] and to show a conformational change when FVIIa is activated from FVII [19] and, thus, the ligand-induced conformational changes of the 170-loop and Gln217 appear to arise easily by the occupation of the large P3 moiety. The ethyl group in P4 of compound 5 does not occupy the S1-sub site of FVIIa/sTF and loses hydrophobic interactions with the S1-sub site consisting of a Cys191-Cys220 disulfide bridge, Gln143, Asp146, Gly219, and Lys192, whereas the propyl group in P4 of compound 1 occupies the S1-sub site of FVIIa/sTF and makes hydrophobic interactions [8]. Instead, the ethyl group in P4 makes intramolecular CH–p interactions with the biphenyl group in P3. In addition, one of the oxygen atoms of the sulfonamide moiety in P4 not only makes a hydrogen bond with the main chain nitrogen of Gly219, which is conserved in the binding of compound 1 to FVIIa/sTF
863
[8], but also makes a new charge–dipole interaction with Lys192 of FVIIa/sTF accompanying a ligand-induced conformational change of Lys192 from the structure of FVIIa/sTF bound to compound 1. Structural basis for selectivity against thrombin The crystal structure of compound 5 bound to FVIIa/sTF revealed that the D -biphenylalanine moiety in P3 makes novel interactions with the 170-loop and that the ethylsulfonamide moiety in P4 adopts a novel conformation and makes an additional charge–dipole interaction with Lys192 of FVIIa/sTF. The contacts on the large area of the S3 site and the 170-loop would lead to the relatively potent binding affinity of compound 5 for FVIIa/TF. The charge–dipole interaction with Lys192 is expected to be energetically more preferable than hydrophobic interactions between the less hydrophobic ethyl group and the S1-sub site of FVIIa/sTF. The modeling study of compound 5 bound to thrombin suggests that the D -biphenylalanine moiety in P3 of compound 5 is too large to sit in the S3 site of thrombin. Thus, the conformation of D -biphenylalanine bound to thrombin is probably the same as that bound to FVIIa/sTF (Fig. 2B). The benzene ring near the Ca atom, which forms hydrophobic interactions with Pro170I and Gln217 in the S3 site of FVIIa/sTF, is expected to make similar hydrophobic interactions with Ile174 and Glu217 in the S3 site of thrombin. However, the distal benzene ring, which interacts with the 170loop of FVIIa/sTF, cannot make similar interactions with thrombin because thrombin does not have the long amino acid insertion corresponding to the 170-loop of FVIIa (Figs. 2B and 3). Thus, this difference in the length of the 170-loop would result in a difference in the contact area from the D -biphenylalanine moiety in the S3 sites of FVIIa/sTF and thrombin, and this difference in contact area would lead to a relative reduction of the binding affinity for thrombin and an improvement in selectivity against thrombin. In addition, thrombin has Glu192 as the residue corresponding to Lys192 of FVIIa/sTF and, thus, thrombin is not expected to be able to make such a charge–dipole interaction with the sulfonamide group (Figs. 2B and 3). Thrombin is expected to make hydrophobic interactions with the less hydrophobic ethyl group in the S1-sub site,
Fig. 3. Sequence alignment of human blood coagulation serine proteases around the 170-loop and residue 192.
864
S. Kadono et al. / Biochemical and Biophysical Research Communications 326 (2005) 859–865
which may be weaker than the charge–dipole interaction between Lys192 of FVIIa/TF and the sulfonamide group. Thus, the difference in the residue at position 192 would result in a difference in the binding mode of the P4 moiety in FVIIa/TF and thrombin, and this binding mode difference may lead to a relative reduction of the binding affinity for thrombin and an improvement in selectivity against thrombin.
proteases suggested that these interactions would be characteristic of binding with FVIIa/TF and be responsible for the improvement of selectivity against a wide variety of coagulation serine proteases, which would result in desirable improvement of selectivity for extrinsic coagulation inhibition, which may lead to low risk of breeding. This structural information would be important for the rational and rapid development of specific inhibitors for FVIIa/TF.
Structural basis for selectivity against coagulation serine proteases other than thrombin Acknowledgments Amino acid sequence comparisons among coagulation serine proteases show that FVIIa has the longest amino acid insertion at the 170-loop (Fig. 3). Coagulation serine proteases other than FVIIa do not have residues corresponding to Val170E, Gly170F, or Asp170G in FVIIa, which make hydrophobic interactions with the distal benzene ring of the D -biphenylalanine moiety in P3. These coagulation serine proteases will have difficulty making hydrophobic interactions with the distal phenyl ring of the D -biphenylalanine moiety similar to thrombin. Consequently, the introduction of the D biphenylalanine moiety in P3 would lead to the improvement of selectivity not only against thrombin but also against other coagulation serine proteases. In addition, Lys192 of FVIIa, which makes a charge– dipole interaction with the sulfonamide group in P4, is not conserved among coagulation serine proteases (Fig. 3). Lys192 of FVIIa is changed to Glu or Gln in serine proteases other than FVIIa and FXIa. For these coagulation serine proteases, other than FVIIa and FXIa, it is difficult to make the charge–dipole interaction with the sulfonamide group similar to thrombin. Consequently, the introduction of the small P4 moiety, ethylsulfonamide, likely leads to the improvement of selectivity not only against thrombin but also against coagulation serine proteases other than FXIa. A slight improvement of the selectivity of compound 5 against FXIa over compound 2 is likely responsible for the existence of Lys192 in FXIa which can make the charge-dipole interaction with the sulfonamide group in P4.
Conclusion X-ray crystal structure analysis of ethylsulfonamidebound to FVIIa/sTF revealed that the large P3 moiety D -biphenylalanine and the small P4 moiety ethylsulfonamide make novel interactions with the 170-loop and Lys192, respectively, accompanying ligand-induced conformational changes of the 170-loop, Lys192, and Gln217 and a novel binding conformation of the ethylsulfonamide moiety. Structural comparisons with thrombin and amino acid comparisons among coagulation serine
D -biphenylalanine-Gln-p-aminobenzamidine
We thank K. Esaki for measuring inhibition activities and F. Ford for proofreading the manuscript. References [1] E.W. Davie, K. Fujikawa, W. Kisiel, The coagulation cascade: initiation, maintenance, and regulation, Biochemistry 30 (1991) 10363–10370. [2] E. Braunwald, R.M. Califf, C.P. Cannon, K.A. Fox, V. Fuster, W.B. Gibler, R.A. Harrington, S.B. King III, N.S. Kleiman, P. Theroux, E.J. Topol, F. Van de Werf, H.D. White, J.T. Willerson, Redefining medical treatment in the management of unstable angina, Am. J. Med. 108 (2000) 41–53. [3] L.A. Harker, S.R. Hanson, J.N. Wilcox, A.B. Kelly, Antithrombotic and antilesion benefits without hemorrhagic risks by inhibiting tissue factor pathway, Haemostasis 26 (Suppl. 1) (1996) 76–82. [4] J. Himber, D. Kirchhofer, M. Riederer, T.B. Tschopp, B. Steiner, S.P. Roux, Dissociation of antithrombotic effect and bleeding time prolongation in rabbits by inhibiting tissue factor function, Thromb. Haemost. 78 (1997) 1142–1149. [5] J.A. Szalony, B.B. Taite, T.J. Girard, N.S. Nicholson, R.M. LaChance, Pharmacological intervention at disparate sites in the coagulation cascade: comparison of anti-thrombotic efficacy vs. bleeding propensity in a rat model of acute arterial thrombosis, J. Thromb. Thrombolysis 14 (2002) 113–121. [6] O. Klingler, H. Matter, M. Schudok, M. Donghi, J. Czech, M. Lorenz, H.P. Nestler, H. Szillat, H. Schreuder, Structure-based design of amidinophenylurea-derivatives for factor VIIa inhibition, Bioorg. Med. Chem. Lett. 14 (2004) 3715–3720. [7] J.J. Parlow, B.L. Case, T.A. Dice, R.L. Fenton, M.J. Hayes, D.E. Jones, W.L. Neumann, R.S. Wood, R.M. Lachance, T.J. Girard, N.S. Nicholson, M. Clare, R.A. Stegeman, A.M. Stevens, W.C. Stallings, R.G. Kurumbail, M.S. South, Design, parallel synthesis, and crystal structures of pyrazinone antithrombotics as selective inhibitors of the tissue factor VIIa complex, J. Med. Chem. 46 (2003) 4050–4062. [8] S. Kadono, A. Sakamoto, Y. Kikuchi, M. Oh-eda, N. Yabuta, T. Koga, K. Hattori, T. Shiraishi, M. Haramura, H. Kodama, T. Esaki, H. Sato, Y. Watanabe, S. Itoh, M. Ohta, T. Kozono, Crystal structure of human factor VIIa/tissue factor in complex with peptide mimetic inhibitor, Biochem. Biophys. Res. Commun. 324 (2004) 1227–1233. [9] W. Bode, I. Mayr, U. Baumann, R. Huber, S.R. Stone, J. ˚ crystal structure of human alphaHofsteenge, The refined 1.9 A thrombin: interaction with D -Phe-Pro-Arg chloromethylketone and significance of the Tyr-Pro-Pro-Trp insertion segment, EMBO J. 8 (1989) 3467–3475. [10] D.W. Banner, P. Hadvary, Crystallographic analysis at 3.0-A resolution of the binding to human thrombin of four active sitedirected inhibitors, J. Biol. Chem. 266 (1991) 20085–20093.
S. Kadono et al. / Biochemical and Biophysical Research Communications 326 (2005) 859–865 [11] M.R. Wiley, N.Y. Chirgadze, D.K. Clawson, T.J. Craft, D.S. Gifford-Moore, N.D. Jones, J.L. Olkowski, L.C. Weir, G.F. Smith, D -Phe-Pro-p-amidinobenzylamine: a potent and highly selective thrombin inhibitor, Bioorg. Med. Chem. Lett. 6 (1996) 2387–2392. [12] D. Kirchhofer, A. Guha, Y. Nemerson, W.H. Konigsberg, F. Vilbois, C. Chene, D.W. Banner, A. DÕArcy, Activation of blood coagulation factor VIIa with cleaved tissue factor extracellular domain and crystallization of the active complex, Proteins 22 (1995) 419–425. [13] Z. Otwinowski, W. Minor, Processing of X-ray diffraction data collected in oscillation mode, Methods Enzymol. 276 (1997) 307– 326. [14] D.W. Banner, A. DÕArcy, C. Chene, F.K. Winkler, A. Guha, W.H. Konigsberg, Y. Nemerson, D. Kirchhofer, The crystal structure of the complex of blood coagulation factor VIIa with soluble tissue factor, Nature 380 (1996) 41–46. [15] H.M. Berman, J. Westbrook, Z. Feng, G. Gilliland, T.N. Bhat, H. Weissig, I.N. Shindyalov, P.E. Bourne, The Protein Data Bank, Nucleic Acids Res. 28 (2000) 235–242.
865
[16] R. Kikumoto, Y. Tamao, T. Tezuka, S. Tonomura, H. Hara, K. Ninomiya, A. Hijikata, S. Okamoto, Selective inhibition of thrombin by (2R,4R)-4-methyl-1-[N2-[(3-methyl-1,2,3,4-tetrahydro-8-quinolinyl+++)sulfonyl]-L -arginyl]-2-piperidinecarboxylic acid, Biochemistry 23 (1984) 85–90. [17] S. Katakura, T. Nagahara, T. Hara, M. Iwamoto, A novel factor Xa inhibitor: structure-activity relationships and selectivity between factor Xa and thrombin, Biochem. Biophys. Res. Commun. 197 (1993) 965–972. [18] K. Sichler, D.W. Banner, A. DÕArcy, K.P. Hopfner, R. Huber, W. Bode, G.B. Kresse, E. Kopetzki, H. Brandstetter, Crystal structures of uninhibited factor VIIa link its cofactor and substrate-assisted activation to specific interactions, J. Mol. Biol. 322 (2002) 591–603. [19] C. Eigenbrot, D. Kirchhofer, M.S. Dennis, L. Santell, R.A. Lazarus, J. Stamos, M.H. Ultsch, The factor VII zymogen structure reveals reregistration of beta strands during activation, Structure (Camb.) 9 (2001) 627–636.