Molecular structure, lipophilicity, solubility, absorption, and polar surface area of novel anticoagulant agents

Journal of Molecular Structure: THEOCHEM 916 (2009) 76–85

Contents lists available at ScienceDirect

Journal of Molecular Structure: THEOCHEM journal homepage: www.elsevier.com/locate/theochem

Molecular structure, lipophilicity, solubility, absorption, and polar surface area of novel anticoagulant agents Milan Remko * Comenius University Bratislava, Faculty of Pharmacy, Department of Pharmaceutical Chemistry, Odbojarov 10, SK-832 32 Bratislava, Slovakia

a r t i c l e

i n f o

Article history: Received 28 July 2009 Received in revised form 4 September 2009 Accepted 6 September 2009 Available online 12 September 2009 Keywords: Factor Xa inhibitors Direct inhibitors of thrombin Molecular structure Solvent effect Lipophilicity Solubility

a b s t r a c t The methods of theoretical chemistry have been used to elucidate molecular properties of factor Xa inhibitors (rivaroxaban, apixaban, otamixaban, betrixaban, razaxaban, and DX-9065a) and direct inhibitor of thrombin (dabigatran). The geometries and energies of these drugs have been computed using HF/ 6-31G(d), Becke3LYP/6-31G(d) and Becke3LYP/6-31++G(d,p) model chemistries. In the case of the Xa inhibitors (rivaroxaban, apixaban, otamixaban, betrixaban, razaxaban, and DX-9065a) the fully optimized most stable conformers possess characteristic L-shape structure. Water has a remarkable effect on the geometry of the anticoagulants studied. The anticoagulant drugs exhibit the largest stability in solvent as expected. Computed partition coefficients (ALOGPS method) for drugs studied varied between 1.7 and 3.9. Neutral compounds are described as lipophilic drugs. Rivaroxaban is drug with lowest lipophilicity. The anticoagulants studied are only slightly soluble in water, their computed solubilities from interval between 5 and 70 mg/L are sufficient for fast absorption. Experimentally determined solubility of rivaroxaban (8 mg/L) is very well interpreted by calculation. Rivaroxaban with PSA value 88 belongs to the anticoagulants with increased absorption. Direct thrombin inhibitor dabigatran is molecule with high total number of proton donor and proton acceptor groups (15), high PSA (150) and lowest absorption of the compounds studied. Ó 2009 Elsevier B.V. All rights reserved.

1. Introduction Anticoagulants are key drugs for the prophylaxis and treatment of thromboembolic disorders [1–5]. Commonly used anticoagulants include parenterally administered unfractionated heparin and low molecular weight heparins, and the orally administered vitamin K antagonists (warfarin) [1,6]. These drugs are not targeted, i.e. they inhibit more than one enzyme in the coagulation cascade [1–6]. Heparin-based anticoagulants are indirect inhibitors that enhance the proteinase inhibitory activity of a natural anticoagulant, antithrombin [6]. Although effective, their use has been hampered by numerous limitations [4]. There is a growing interest in new, orally active anticoagulants with significant advantages to current agents such as heparin and warfarin for the treatment and prevention of thrombotic diseases. The new anticoagulants under investigation for venous thromboembolism treatment target factor Xa (fXa) or thrombin [1–5]. Inhibitors of factor Xa block thrombin generation, whereas thrombin inhibitors block the activity of thrombin, the enzyme that catalyses the conversion of fibrinogen to fibrin [1,3]. Besides synthetic indirect factor Xa inhibitors from the family of glycosaminoglycans (fondaparinux, idraparinux) * Tel.: +421 2 50117225; fax: +421 2 50117100. E-mail address: [email protected] 0166-1280/$ - see front matter Ó 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.theochem.2009.09.011

numerous direct, selective factor Xa inhibitors are currently at various stages of development in different therapeutical indications [1–4]. Small-molecule synthetic compounds such as rivaroxaban, razaxaban, apixaban, betrixaban are members of a new class of orally available active-site-directed factor Xa inhibitors. Small-molecule direct thrombin inhibitors (dabigatran etexilate) inhibit thrombin directly by directly binding to the active catalytic site [1–3]. Despite a great deal of experimental evidence for the relationship betweenthechemicalandpharmaceuticalpropertiesofnewanticoagulants targeting factor Xa or thrombin and their biological activity, there is no single experimental study concerned with the systematic comparative experimental investigation of the physico-chemical and pharmacokinetic parameters of these medicinally useful new anticoagulants. Quantitative structure activity relationships of factor Xa inhibitorswerediscussedquiterecently[7],andtheX-raycrystalstructuresofrivaroxaban[8],DX-9065a[9],apixaban[10],otamixaban[11] andrazaxaban[12]incomplexeswithfactorXawereusedtoclarifythe binding mode of these ligands. The molecular structure of six monomeric structural units (1-OMe DIdoA-2SNa2 (unit A), 1-OMe GlcNS6SNa2 (unitD),1,4-DiOMeGlcNa(unitE),1,4-DiOMeGlcN-S3S6SNa3 (unitF), 1,4-DiOMeIdoA-2SNa2 (unitG), and 1,4-DiOMeGlcN-S6SNa2 (unitH)),fourdimericstructuralunits(D–E,E–F,F–G,andG–H),twotrimericstructuralunits(D–E–F,andF–G–H)andpentamerD–E–F–G–H

M. Remko / Journal of Molecular Structure: THEOCHEM 916 (2009) 76–85

(fondaparinux) of heparin has been previously investigated [13–17] using the density functional theory. The absence of experimental data of novel synthetic anticoagulants targeting factor Xa or thrombin presents a challenge to the application of computational modeling techniques in order to enhance our understanding of the subtle biological effects of these anticoagulants. In this paper we have used the results of largescale theoretical calculations for the study of the molecular structure, pKa, lipophilicity, solubility, absorption, and polar surface area of factor Xa inhibitors (rivaroxaban, apixaban, otamixaban, betrixaban, razaxaban, and DX-9065a) and direct inhibitors of thrombin (dabigatran). The results of theoretical studies of these drugs were compared with the available experimental data and discussed in relation to the present theories of action of these agents. 2. Computational details Theoretical calculations of the rivaroxaban, apixaban, otamixaban, betrixaban, razaxaban, DX-9065a, dabigatran, and dabigatran etexilate (Fig. 1) were carried out with the Gaussian 03 computer code [18] at the ab initio SCF (HF [19]) and density functional theory (DFT, Becke3LYP [20–24]) levels of theory using the 6–31G(d) and 6–31++G(d,p) basis sets. In order to evaluate the conformational behavior of these systems in solvent, we carried out optimization calculations in the presence of water. The methodology used in this work is centered on two solvation methods, PCM [25,26] and Onsager [27] models. The structures of all gas-phase species were fully optimized at the HF/6–31G(d) and Becke3LYP/6– 31G(d) levels of theory without any geometrical constraint. In order to check the correctness of the B3LYP calculated relative energies using the double-f basis set, we also performed calculations of the anticoagulant species, using the larger basis set 6–31++G(d,p) implemented in the Gaussian 03 package of computer codes [14,15]. The structures of all condensed-phase (SCRF) species were fully optimized without any geometrical constraint at the DFT level of theory applied. Lipophilicity and water solubility calculations were carried out using web-based VCCLAB [28–30]. For calculations of molecular polar surface areas the fragmentbased method of Ertl and coworkers [31,32] incorporated in the Molinspiration Cheminformatics software [33] was used. 3. Results and discussion 3.1. Molecular structures Conformational search using theoretical methods for such large systems was in the past limited to use some of the available forcefield methods [34]. Rapid advances in computer hardware and software and in quantum medicinal chemistry have brought high-performance computing and graphic tools within the reach of most academic and industrial laboratories, thus facilitating the development of useful approaches to rational drug design. Quantum chemical calculations are now applied successfully in medicinal chemistry and drug design to determine accurately molecular structures and properties for use in a wide variety of CADD studies [35]. It is common in the computational study of drugs to use structural data obtained from X-ray crystallography or NMR spectroscopy as guides to the quality of theoretical computations. In the absence of the experimental published data about molecular conformations of drugs, as an alternative to analyzing small molecule crystal structures the conformations of drugs bound to their protein targets can be examined [36]. Rivaroxaban, otamixaban and DX-9065a are chiral molecules and may be present as racemates. However, under the development process of these drugs a clear preference for particular enantiomer was observed,

77

indicating a specific interaction with fXa. The calculations for these drugs were carried out with the biologically most active enantiomers only. In the absence of the small molecule crystal structures, we examined the conformations of ligands bound to their fXa and trombin targets by studying the macromolecular crystal structures deposited in the Protein Data Bank. In the context of drug design, the conformation a small molecule adopts when bound to a pharmaceutical target is of fundamental importance. The relative orientation of anticoagulant moiety defined by individual dihedral angles (a, b, c, d, e, f, g, and h, Fig. 1) was taken from the experimental data for available X-ray data of the crystal structures deposited in the Protein Data Bank (PDB) [37] (complexes of the Xa factor with rivaroxaban (PDB: 2W26), apixaban (PDB: 2P16), otamixaban (PDB: 1KSN), razaxaban (PDB: 1Z6E), DX-9065a (PDB: 1FAX), and complex of the ethylester of dabigatran with trombin (PDB: 1HTS). A particular ligand conformations observed in these complexes may be due to an intermolecular interaction that is not present in solution or vapor phases. All compounds, but apixaban possess amide functionality, which imparts certain conformational rigidity to the overall structure of molecules studied. The possible relaxation of the geometry of a ligand upon dissociation from the receptor may bring new information about the conformation of drug in isolated state. An analysis of the harmonic vibrational frequencies at the HF level of theory of the optimized species revealed that all the structures obtained were minima (no imaginary frequencies). The Cartesian coordinates (Å) of all gas-phase drugs investigated, optimized at the B3LYP/6–31++G(d,p) level of theory, are given in Table A of the electronic Supporting Information. The geometries optimized at the B3LYP/6–31++G(d,p) level of theory are shown in Fig. 2. In the case of the Xa inhibitors (rivaroxaban, apixaban, otamixaban, betrixaban, razaxaban, and DX-9065a) the fully optimized most stable conformers possess characteristic L-shape structure. Examination of the space models of the B3LYP computed structures using two basis sets of the drugs investigated shows that the increase of the basis set gives essentially the same results. The effect of bulk solvent is treated with two solvation methods (the Onsager [27] and PCM [25,26]) for comparison. Initial calculations were carried out, for computational reasons, using the SCRF formalism of Wong et al. [38–41]. The radii of the cavities used in this approach were chosen after a volume calculation of each molecule, and the dielectric constant of water (e = 78.5) was used. The placing of the isolated molecules into a spherical cavity within a dielectric medium of the Onsager model of solvation does not represent the realistic situation in the biological medium; it seems helpful in revealing the main role of the solvent in intermolecular electrostatic interactions. The second, PCM (conductor-like polarizable model), defines the cavity by the envelope of spheres centered on the atoms or the atomic groups [25,26]. The whole concept of using such macroscopic properties as dielectric constants in microscopic computations has been criticized [42,43]. Despite all these valid criticisms, continuum-based methods of solvation are used extensively and successfully in a variety of problems [44,45]. It has been shown previously [46] that the conductor-like polarizable method reproduces hydration energies with accuracy in the order of a few kcal/mol but mostly (70% of the cases) even better than one kcal/mol. The factor Xa inhibitors (rivaroxaban, apixaban, otamixaban, betrixaban, razaxaban, and DX-9065a) studied do not possess common pharmacophore functionality. However, the intense research in this field accumulated new results, which have been summarized in a number of publications [1–12]. The binding of inhibitor to fXa is characterized by two general interaction sites S1 and S4 (Fig. 3). Based on the analysis of the binding in the S1 pocket fXa inhibitors are categorized into two classes. One class of early inhibitors mimics the natural interaction between arginine of prothrombin in the fXa active site with the Asp189 of the S1 pocket of fXa. Such compounds

78

M. Remko / Journal of Molecular Structure: THEOCHEM 916 (2009) 76–85

O

13

O

10 4

1

5

8

2

O

N

6

11

7

N

3

15

N H

9

Cl

S 12 14

MeO

O

H 3C O

CH3 N

12

4

11

O

O 4

1

10

5 6

3

7

5

1

N N

N

N

9

8 2

N

2

O

4

N

5

O8

3

6

9

11 13

8

CH3 H3C

4

N

16

15

16

1

18

NH2

HN

14

N14

2

17

16

13

NH

3

15

NH 15

5

12

11

9

6

O N

O10

14

Otamixaban 17

N O

7

OMe

12

NH2

12

Razaxaban

10

N H

N

F

CH3 O

7

11 N

N H 13

O

2

7

6

N

Apixaban 1

10

8

3

NH2

+

CF3

9O

Rivaroxaban

N

13

CH3

NH

12 11 9

Cl

4

2 3

N

HN 8

O5

1

LY-517717 O 10

7 6

Betrixaban

N H

OMe

HO

O CH3 N N

N

9

4

2

NH 11 12

17

N H

8

7

1

10

N

6

3

15 16

O5

13

Dabigatran

NH2

14

H 3C

H 3C O

O CH3

O

O

N N

N 2 1

9

4

N

6

3

7

O

8

10

N

17

11 12

N H

5

15 13

14

16

NH2

Dabigatran Etexilate Fig. 1. Structure and atom labeling in the anticoagulant drugs studied.

often contain phenylamidine group mimicking this interaction (otamixaban, betrixaban, DX-9065a). The second category of compounds utilized neutral aryl groups bind into the S1 pocket (e.g.

aminobenzisoxazole for razaxaban, chlorothiophene moiety of rivaroxaban). In addition to the S1 pocket, a second major binding site of fXa is a narrow hydrophobic channel (the S4 pocket) defined by the

79

M. Remko / Journal of Molecular Structure: THEOCHEM 916 (2009) 76–85

Fig. 2. B3LYP/6–31++G(d,p) optimized structures of the anticoagulants investigated.

Tyr 228 Important interaction

O

S

Cl

Coplanar rings

O N H O

N

S1

N

Pocket

O 3.3 Å

S4

Perpendicular arrangement

O

Hydrogen bond

Hydrophobic pocket 2.0 Å

Gly 219 Fig. 3. Binding model of rivaroxaban. S1 and S4 pocket represents binding sites typical for synthetic direct fXa inhibitors.

aromatic rings of Tyr99, Phe174, and Trp215 [8]. Dabigatran is a novel reversible direct thrombin inhibitor. It interacts with the active site of thrombin composed of the specificity (S1), proximal (S2)

and distal (D) pockets. The amidine group of dabigatran interacts with the aspartic acid (Asp189) of the S1 pocket. The central methylbenzimidazole part is bound to thrombin by a hydrophobic

80

M. Remko / Journal of Molecular Structure: THEOCHEM 916 (2009) 76–85

interaction with the S2 pocket, and the 2-pyridyl group is positioned between Leu99 and Ile174 in the D-pocket [47]. The relative molecular orientation of individual drugs studied is described by different number of rotatable bonds. These dihedral angles of the fully optimized anticoagulants are given in Table 1 together with the available X-ray structures of these drugs in the bound state (at the receptor). Values for these dihedral angles for individual drugs are different (Table 1), and no general conclusions about pharmacophore functionality can be deduced. Thus, the structure of drugs studied will be discussed individually. According to our calculations in the gas state, the equilibrium geometries computed at the HF level of theory are in general agreement with the DFT results obtained with the ‘‘standard” 6-31G(d) basis set. However, the ab initio SCF and DFT optimized dihedral angles of some drugs studied exhibit large differences (within the 10°– 20°). In order to study the basis set effect on the geometry of the anticoagulants investigated within the DFT theory we also carried out calculations using the larger basis set 6–31++G(d,p). The extension of the basis set in the DFT calculations resulted in only small changes in the equilibrium geometry of the drugs studied. The optimized dihedral angles using two basis sets within the DFT theory fit one another to within about 2°–5° (Table 1). Water has a remarkable effect on the geometry of the anticoagulants studied (Table 1). Table 2 shows the results obtained for calculations performed in both, vacuum and that based on the solvation method used. The anticoagulant drugs exhibit the largest stability in solvent as expected, since they hold considerable dipole moment (Table 2). The energy difference between gas phase and solvated phase was significant for the both solvation models employed in this work. The solvated phase energies within the Onsager model were obtained after full geometry optimization in water. However, some anticoagulants failed to optimize geometries in water within the PCM formalism. Table 2 contains water stabilization energies using single-point PCM calculations and in vacuo fully optimized geometries. The comparison of the single-point PCM and available water stabilization energies obtained after full geometry optimization in water indicate that in the case of neutral drugs the optimization of geometry in aqueous medium does not significantly change the solvation energy. The difference in water stabilization energy is very small (around 5 kJ/mol). The PCM provided substantially more stable structures than Osanger’s model. Experimentally, small molecule drug conformations are commonly studied using X-ray crystallography. The absence of the published experimental X-ray structural data of anticoagulants studied presents a challenge to the application quantum chemical methods in order to obtain information about the stable conformations of these drugs in the gas phase and in solution. A comparison of the ab initio SCF calculated conformational energies of drug molecules with the conformer distribution in the solid state routinely show a good correlation [48]. Moreover, previous investigations of the protein–ligand complexes revealed similar torsion angles distributions for fragments when the bound and unbound distributions were compared [49]. For the reason of comparison and analysis of theoretically determined conformations and proteinbound conformations of anticoagulants studied we also present the available structural data for bound anticoagulants on the factor Xa receptor. 3.1.1. Rivaroxaban The 1,3-oxazolidin-ring system of rivaroxaban (5-chloro-N[[(5S)-2-oxo-3-[4-(3-oxomorpholin-4-yl)phenyl]-1,3-oxazolidin-5yl]methyl]thiophene-2-carboxamide) has in position 5 a chiral carbon atom and possesses two enantiomers termed R and S with a clear preference for (S)-configuration [8]. Optimized molecular conformation of the (S)-rivaroxaban (Fig. 4) provides the L-shape, which is needed for factor Xa binding. Experimentally determined

3D structure of the (S)-rivaroxaban studied corresponds to the bound molecule at the protein, therefore the general structural motifs of drug can be compared with results for isolated molecule from theoretical methods only. The experimental values for the dihedral angles in the rivaroxaban–fXa complex are well interpreted by the corresponding angles computed for the solvated (S)-rivaroxaban (Fig. 4). The main difference in the molecular structure of bound and unbound rivaroxaban arises from the position of the morpholinone end moiety, dihedral angle a[C(1)–N(2)– C(3)–C(4)]. The carbonyl group of this moiety effects mainly a planarization of the morpholinone ring and brings it into a rather perpendicular arrangement to the aryl ring [8]. The DFT calculations suggest more planar arrangement of the morpholine and aryl rings. The dihedral angle a[C(1)–N(2)–C(3)–C(4)] is about 51°–55° (DFT method), and for (S)-rivaroxaban in water solution, and/or at receptor increases to about 67°–77° (Table 1). Oxazolidone and aryl rings are almost coplanar (the dihedral angle b[C(5)–C(6)– N(7)–C(8)]) (Table 1). 3.1.2. Apixaban Apixaban (1-(4-methoxyphenyl)-7-oxo-6-[4-(2-oxopiperidin1-yl)phenyl]-4,5-dihydropyrazolo[5,4-c]pyridine-3-carboxamide) is one of the most rigid molecules of the anticoagulants studied. Its 3D structure (Fig. 2) is governed by three dihedral angles (a, b, and c), Table 1. The 3D structure of the bound apixaban at the factor Xa receptor, the gas-phase structure and solvated apixaban are substantially different. In general, the coordination of the apixaban to its fXa receptor (PDB file 2P16) leads to the perpendicular arrangement of the phenyllactam moiety (dihedral angle a[C(1)–N(2)–C(3)–C(4)]). The same perpendicular arrangement forms also bicyclic pyrazole scaffold and p-methoxyphenyl moiety of bound apixaban (dihedral angle c[C(9)–N(10)–C(11)– C(12)]. The phenyllactam and the bicyclic pyrazole scaffold of bound apixaban are also in almost perpendicular arrangement (dihedral angle b[C(5)-C(6)-N(7)-C(8)]), Table 1. Values of these dihedral angles in the gas-phase structure and solvated apixaban are quite different. The unbound apixaban is in gas phase and in water solution substantially more planar, the planarization effect is especially considerable for the moiety containing phenyl ring, bicyclic pyrazole scaffold and p-methoxyphenyl group (dihedral angles b, and c). The central part in the planarization of this apixaban moiety plays the carbonyl group of the central lactam group of the bicyclic pyrazole scaffold. This group effects a planarization of whole moiety via electrostatic intramolecular interaction with acidic hydrogen atoms of the neighboring rings (Fig. 2). The C-3 carboxamido substituent and the pyrazole scaffold are in mutual planar arrangement. 3.1.3. Otamixaban (R,R)-Otamixaban (methyl (2R,3R)-2-(3-carbamimidoylbenzyl)3-{[4-(1-oxidopyridin-4-yl)benzoyl]amino}butanoate) is structurally very flexible molecule, and its space arrangement is defined by 8 dihedral angles (Fig. 1). The P4 phenylpyridyl-N-oxide and benzamidine groups, responsible for the interaction of otamixaban with the fXa receptor, are connected by the system of single bonds enable large structural flexibility of this drug on receptor. A comparison of the fully optimized isolated molecule and the bound otaximaban (PDB file 1K3N) indicates that the largest structural rearrangement resulting in the biologically active conformation is related with the dihedral angles e, f, and g (Table 1). The gas-phase conformations of the phenylpyridyl-N-oxide and benzamidine groups of otamixaban are also preserved in bound drug. The amidine group of the benzamidine moiety is twisted out of the aromatic ring with the dihedral angle l[C(15)–C(16)–C(17)–N(18)] from a relatively narrow interval of 155°–160° (Table 1).

81

M. Remko / Journal of Molecular Structure: THEOCHEM 916 (2009) 76–85 Table 1 Optimized dihedral anglesa of the drugs studied. Dihedral angle

a[C(1)–N(2)–C(3)–C(4)] b[C(5)–C(6)–N(7)–C(8)] c[C(8)–C(9)–C(10)–N(11)] d[C(9)–C(10)–N(11)–C(12)] e[C(9)–C(10)–N(11)–O(13)] f[C(10)–N(11)–C(12)–C(14)] g[N(11)–C(12)–C(14)–C(15)]

a[C(1)–N(2)–C(3)–C(4)] b[C(5)–C(6)–N(7)–C(8)] c[C(9)–N(10))–C(11)–C(12)]

a[C(1)–C(2)–C(3)–C(4)] b[C(5)–C(6)–C(7)–O(8)] c[C(5)–C(6)–C(7)–N(9)] d[C(6)–C(7)–N(9)–C(10)] e[C(7)–N(9)–C(10)–C(11)] f[N(9)–C(10)–C(11)–C(12)] g[N(9)–C(10)–C(11)–C(13)] h[C(10)–C(11)–C(13)–C(14)] l[C(15)–C(16)–C(17)–N(18)]

X-ray

a[C(1)–C(2)–C(3)–C(4)] b[C(2)–C(3)–C(4)–C(5)] c[C(3)–C(4)–C(5)–C(6)] d[C(7)–C(8)–O(9)–C(10)] e[C(8)–O(9)–C(10)–C(11)] f[C(11)–N(12)–C(13)–N(14)] g[C(a)–C(b)–C(c)–N(d)]

a[C(1)–C(2)–N(3)–C(4)] b[C(2)–N(3)–C(4)–O(5)] c[C(2)–N(3)–C(4)–C(6)] d[N(3)–C(4)–C(6)–C(7)] e[N(8)–C(9)–C(10)–N(11)] f[C(9)–C(10)–N(11)–C(12)] g[C(10)–N(11)–C(12)–C(13)] h[C(14)–C(15)–C(16)]–N(17)]

DFT – CPCM

(S)-Rivaroxaban 57.7 4.9 59.7 106.3 2.1 178.1 0.6

54.7 2.4 60.9 100.9 5.5 174.8 9.4

66.9 7.8 59.2 106.7 5.0 175.3 2.6

2p16 84.2 116.6 94.1

104.5 70.5 62.3

125.4 50.0 52.8

Apixaban 116.8 55.0 57.8

129.6 54.3 53.2

112.4 58.7 62.3

1k3n 36.7 20.7 158.2 177.9 127.9 179.5 54.6 159.9 156.5

42.7 25.2 154.7 176.3 79.1 170.4 64.8 162.6 153.2

32.9 22.9 157.2 176.9 77.3 171.2 64.4 162.8 158.0

(R,R)-Otamixaban 35.3 21.9 157.9 177.7 76.8 170.3 65.1 162.6 154.9

33.6 24.9 156.3 177.4 59.4 177.5 55.8 144.7 160.5

5.7 2.1 178.2 145.6 155.0 1.1 179.7 22.6 35.9 147.8

3.8 2.8 178.1 155.7 172.1 3.7 176.9 17.9 35.2 149.3

Betrixaban 2.4 3.9 176.9 154.8 170.1 3.2 177.3 15.5 38.8 145.5

2.2 3.0 177.8 156.4 173.7 1.9 179.0 17.9 36.4 148.8

7.8 2.4 178.0 152.3 160.8 1.8 178.7 24.7 43.3 141.4

1z6e 83.5 115.3 6.7 172.7 163.8 81.4

66.3 115.8 2.4 176.3 145.2 69.3

56.9 127.5 0.2 178.4 147.7 58.1

Razaxaban 58.2 122.8 2.3 176.3 145.6 62.7

58.2 127.8 0.3 178.3 148.9 59.0

57.9 128.2 0.6 178.0 154.2 74.2

1fax 44.0 55.6 57.2 127.4 41,4 169.2 179.8

77.6 68.6 62.2 177.5 81.9 175.2 154.2

80.7 67.6 65.9 176.0 82.1 177.1 159.0

(S,S)-DX9065a 79.4 67.9 62.0 177.9 81.5 178.1 156.3

78.6 66.3 64.7 178.2 81.0 178.2 157.3

75.2 66.2 62.4 178.0 81.1 178.2 156.9

1KTSb 83.0 170.8 4.5 132.6 31.1 79.3 33.3 176.7

142.3 145.6 39.2 33.5 125.7 82.7 18.2 156.0

146.9 152.0 33.5 31.6 123.4 80.8 18.4 160.3

Dabigatran 144.7 152.7 32.6 34.2 125.7 78.1 16.3 158.1

146.4 150.6 34.7 35.6 122.4 77.4 15.7 161.1

142.8 146.1 38.8 33.8 124.6 82.4 18.2 155.7

147.5 152.4 33.0 32.8 122.2 80.9 18.5 160.8

Dabigatran etexilate 145.5 153.5 31.6 35.0 122.9 79.6 16.9 160.7

146.2 151.3 33.7 37.4 118.7 78.2 15.6 162.7

b[C(2)–N(3)–C(4)–O(5)] c[C(2)–N(3)–C(4)–C(6)] d[N(3)–C(4)–C(6)–C(7)] e[N(8)–C(9)–C(10)–N(11)] f[C(9)–C(10)–N(11)–C(12)] g[C(10)–N(11)–C(12)–C(13)] h[C(14)–C(15)–C(16)]–N(17)] a

DFT – Onsager

51.1 4.3 62.1 103.9 1.7 178.2 5.6

a[C(1)–C(2)–N(3)–C(4)]

b

B3LYP/6–31++g(p,d)

74.7 15.7 61.9 105.8 3.8 176.7 8.7

b[C(2)–N(3)–C(4)–O(5)] c[C(2)–N(3)–C(4)–C(6)] d[N(3)–C(4)–C(6)–C(7)] e[C(6)–C(7)–N(8)–C(9)] f[C(7)–N(8)–C(9)–O(10)] g[C(7)–N(8)–C(9)–C(11)] h[N(8)–C(9)–C(11)]–C(12)] l[C(13)–C(14)–C(15)–N(16)] m[C(13)–C(14)–C(15)–N(17)]

b[C(5)–C(6)–N(7)–C(8)] c[C(6)–N(7)–C(8)–O(9)] d[C(6)–N(7)–C(8)–C(10)] e[N(7)–C(8)–C(10)–C(11)] f [C(10)–N(11)–C(12)–C(13)]

B3LYP/6–31g(d)

2w26 77.9 19.7 63.2 96.3 0.5 178.4 1.3

a[C(1)–C(2)–N(3)–C(4)]

a[C(1)–N(2)–C(3)–C(4)]

HF/6–31g(d)

For definition of dihedral angles see Fig. 1. Dabigatran ethylester.

82

M. Remko / Journal of Molecular Structure: THEOCHEM 916 (2009) 76–85

Table 2 The Becke3LYP/6–31G(d) solvent stability of the anticoagulant agents investigateda. No. 1 2 3 4 5 6 7 8 a b c d e f

Drug (S)-Rivaroxaban Apixaban (R,R)-Otamixaban Betrixaban Razaxaban (S,S)-DX-9065a Dabigatran Dabigatran etexilate

Cavity valueb (a0) 5.59 5.92 5.98 5.44 5.63 5.58 6.06 6.64

DEc Onsager 2.9 5.3 7.6 1.2 1.9 3.5 4.1 7.8

DEc,d PCM

Gas-phase dipole momentf e

93.8 (95.5) 93.3 (97.8)e 122.2 72.6 (76.1)e 113.1 (119.2)e 123.7 148.8 130.1

3.29 7.49 3.96 2.19 2.84 3.70 4.39 6.87

Water as solvent. Ångström (Å). kJ/mol. Single-point PCM calculation. Full geometry optimization in the solvent. Debye (D).

er stabilization effect of intramolecular hydrogen bond of the NAH. . .O@C type (Fig. 2). The N,N-dimethylamidine group is twisted out of the aromatic ring by about 35°. Water solvent does not have pronounced effect on the overall shape of betrixaban (Table 1).

Fig. 4. (A) Molecular superimposition of the Becke3LYP optimized molecular structure of (S)-rivaroxaban (green) and (S)-rivaroxaban from the cocrystal with coagulation factor Xa, PDB 2W26 (blue). (B) Molecular superimposition of the Becke3LYP optimized molecular structure of (S)-rivaroxaban (green) and in solution optimized (S)-rivaroxaban (violet). (C) Molecular superimposition of the (S)rivaroxaban from the cocrystal with coagulation factor Xa, PDB 2W26 (blue) and in solution optimized (S)-rivaroxaban (violet). For simplicity the hydrogen atoms are omitted. (For interpretation of color mentioned in this figure legend the reader is referred to the web version of the article.)

3.1.4. Betrixaban Betrixaban (N-(5-chloropyridin-2-yl)-2-[[4-(N,N-dimethylcarbamimidoyl)benzoyl]amino]-5-methoxybenzamide) [50] is another fXa inhibitor containing benzamidine moiety. Its conformational structure is, like in otamixaban, determined by eight dihedral angles a, b, d, e, f, h, l, and m (Fig. 1). Both amide groups are practically planar (dihedral angles b, c, f, and g). HF and DFT methods describe differently the conformation of the amide groups on the central methoxyphenyl ring. The DFT method prefers more planar arrangement, which is apparently due by high-

3.1.5. Razaxaban Razaxaban (2-(3-amino-1,2-benzoxazol-5-yl)-N-[4-[2-(dimethylaminomethyl)imidazol-1-yl]-2-fluorophenyl]-5-(trifluoromethyl) pyrazole-3-carboxamide) is a predecessor of apixaban that was discontinued based on less that optimal pharmacological properties [51]. The discovery of the pyrazole scaffold was an important milestone in search for new molecules targeting fXa and proved to be crucial in the evolution of orally bioavailable fXa inhibitors such as razaxaban and apixaban [10,12]. The dimethylaminomethylimidazole moiety and proximal phenyl ring in vapor-phase razaxaban has a dihedral angle a of about 57° (DFT calculation). The amide group is planar and rotated out of the phenyl ring by about 122° (DFT, angle b). The DFT calculated angle between the plane of the pyrazole moiety and the aminobenzisoxazole plane of razaxaban is about 60° (Table 1). Water has only slight effect on the geometry of the razaxaban. The biologically active conformation of the fXa bound razaxaban (PDB file 1Z6E) differs from the gas-phase structure and/or solvated molecule in two points. The imidazole ring of the phenylimidazole moiety is at the receptor in a perpendicular arrangement to the proximal phenyl ring (dihedral angle a[C(1)–N(2)–C(3)–C(4)] = 83.5°). Almost perpendicular arrangement is also observed for the moiety containing the pyrazole scaffold and the aminobenzisoxazole group (dihedral angle f [C(10)–N(11)–C(12)–C(13)] = 81.4°). 3.1.6. DX-9065a (S,S)-DX-9065a ((+)-2S-2-[4-[[(3S)-1-acetimidoyl-3-pyrrolidinyl]oxy]phenyl]-3-[7-a midino-2-naphthyl]propanoic acid) belongs to the class of early inhibitors mimicked the natural interaction of the arginine of prothrombin and Asp189 of the S1 specificity pocket in the fXa active site. The pyrrolidine ring of this drug binds to the aryl binding site S4 of fXa. In the gas-phase structure the amidine group of the naphthamidine moiety is rotated out of the naphthalene ring system by about 160°, dihedral angle g[C(a)–C(b)–C(c)–N(d)], Fig. 1. The connecting carbon chain of the propanoic acid moiety orients the second pharmacophoric group containing phenyl and pyrrolidine almost perpendicularly to the naphthalene ring plane allowing L-shape conformation needed for good fXa inhibitory activity. The pyrrolidine ring is in perpendicular position with respect to the plane of the phenyl group (dihedral angle e[C(8)–O(9)–C(10)–C(11)] is about 82°, Table 1). The biologically active conformation of the bound DX-9065a at the fXa receptor (PDB file 1FAX) is quite different.

83

M. Remko / Journal of Molecular Structure: THEOCHEM 916 (2009) 76–85

Largest changes upon interaction with the active site are observed in the flexible region of substituted propionic acid (dihedral angles a, and b) and around ether bond (dihedral angles d, and e). Amidine group in the bound conformation of DX-9065a is situated in the plane of the naphthalene moiety (Table 1). 3.1.7. Dabigartan and dabigatran etexilate Dabigatran (3-[[2-[[(4-carbamimidoylphenyl)amino]methyl]-1methylbenzimidazole-5-carbonyl]-pyridin-2-ylamino]propanoic acid) is direct inhibitor of thrombin that binds to thrombin with high affinity and specificity. Since dabigatran is not orally active in clinical praxis it is used in the form of a double prodrug dabigatran etexilate (ethyl-3-[[2-[[[4-(N0 -hexoxycarbonylcarbamimidoyl) phenyl]amino]methyl]-1-methylbenzimidazole-5-carbonyl]-pyridin-2-ylamino]propanoate) [52,53]. In the isolated molecule the 2-pyridyl group and carboxamide moiety are in mutual nonplanar configuration with the dihedral angle a[C(1)–C(2)–N(3)-C(4)] of about 145°. Pyridyl and benzimidazole rings are in mutual perpendicular orientation. The substituted amino group of the linker between benzimidazole and benzanidine is almost coplanar with the aromatic ring (dihedral angle g[C(10)–N(11)–C(12)–C(13)] = 16.3°, DFT calculation). The amidine group of the benzamidine substituent is twisted out of the benzene ring by 158° (dihedral angle g, Fig. 1). Overal conformation of dabigatran exhibits typical Z-shape that was also observed in the crystal structure of the bound prodrug of dabigatran (ethylester of dabigatran) on the thrombin active site [47]. The dabigatran core structure of the double prodrug dabigatran etexilate remains upon structural modification in the carboxyl and amidine parts unaffected (Table 1). The optimal geometry of dabigatran and dabigatran exetilate computed with the Onsager method do not considerably differ in water solution from those obtained for isolated molecules (Table 1). 3.2. Lipophilicity Poor solubility and poor permeability are among the main causes for failure during drug development [54–56]. It is therefore important to determine these physico-chemical properties associated with a drug, before synthetic work is undertaken. The computed log P values (P is the partition coefficient of the molecule in the water – octanol system), together with the experimental data, are shown in Table 3. The ALOGPs method is part of the AlogPS 2.1 program [57] used to predict lipophilicity [58] and aqueous solubility [59,60] of compounds. The lipophilicity calculations within this program are based on the associative neural network approach and the efficient partition algorithm. The LogKow (KowWin) program [61] estimates the log octanol/water partition coefficient (log P) of organic chemicals and drugs using an atom/ fragment contribution method developed at Syracuse Research Corporation [62]. The XLOGP2 is atom-additive method applying corrections [63,64]. Available experimental log P value of rivaroxaban was extracted from the literature [65]. It is best reproduced by the associative neural network ALOGPs method (Table 3). Com-

Table 3 Calculated partition coefficients of the anticoagulants.

puted partition coefficients (ALOGPS method) for drugs studied varied between 1.7 and 3.9. Neutral compounds are described as lipophilic drugs. Rivaroxaban is drug with lowest lipophilicity. The variation in the lipophilicity of the series of fXa inhibitors studied is low and corresponds well to the lack of importance of hydrophobic effect derived from QSAR studies [7]. Dabigatran, biologically active direct inhibitor of thrombin, exhibits moderate lipophilicity (Table 3) and it’s binding to the thrombin active site results, apart from the salt bridge with Asp 189, solely from hydrophobic interaction [47,52]. Much greater lipophilicity possesses double prodrug dabigatran exetilate that is rapidly converted into the active form upon absorption from the gastrointestinal tract. 3.3. Solubility Log S – an intrinsic solubility in neutral state is indicative of a compound’s solubility (S). As the experimental solubilities of most compounds under study are not known, the log S values were calculated using ALOGPS a predictor. This method uses E-state indices as descriptors and a neural network [59] as the modeling ‘‘engine”. The computed solubilities are presented in Table 4. Previous calculations of chemically different biologically active compounds have been shown, that these methods well reproduce known experimental solubilities [66,67]. Drug solubility is one of the important factors, which affect the movement of a drug from a site of administration into the blood. Knowing of drug solubility is important. It is well known that insufficient solubility of drugs can lead to poor absorption [68]. Investigation of the rate-limited steps of human oral absorption of 238 drugs (including warfarin) has been shown [68] that the absorption of a drug is usually very low if the calculated solubility is <0.0001 mg/L. The neutral anticoagulants studied are only slightly soluble in water, their computed solubilities from interval between 5 and 70 mg/L are sufficient for fast absorption. Experimentally determined solubility of rivaroxaban (8 mg/L) [8] is very well interpreted by calculation (Table 4). DX-9065a and dabigatran are zwitterions of high water solubility and poor oral bioavailability [47,69]. 3.4. Absorption, polar surface area, and ‘‘rule of five” properties High oral bioavailability is an important factor for the development of bioactive molecules as therapeutic agents. Passive intestinal absorption, reduced molecular flexibility (measured by the number of rotatable bonds), low polar surface area or total hydrogen bond count (sum of donors and acceptors), are important predictors of good oral bioavailability [70,71]. Properties of molecules such as bioavailability or membrane permeability have often been connected to simple molecular descriptors such as log P (partition coefficient), molecular weight (MW), or counts of hydrogen bond acceptors and donors in molecule [72]. Lipinski [73] used these molecular properties in formulating his ‘‘Rule of Five”. The rule states that most molecules with good membrane permeability

Table 4 Calculated solubilities of the anticoagulants studied.

No.

Drug

ALOGPS

KoWWIN

XLOGP2

No.

Drug

ALOGPS

1 2 3 4 5 6 7 8

(S)-Rivaroxaban Apixaban (R,R)-Otamixaban Betrixaban Razaxaban (S,S)-DX-9065a Dabigatran Dabigatran Etexilate

1.74 2.23 2.12 2.86 3.90 2.61 2.37 5.17

2.18 2.71 0.97 3.86 1.81 1.69 1.95 5.58

0.57 1.51 5.06 5.04 3.31 4.81 2.73 5.36

1 2 3 4 5 6 7 8

(S)-Rivaroxaban Apixaban (R,R)-Otamixaban Betrixaban Razaxaban (S,S)-DX-9065a Dabigatran Dabigatran etexilate

4.64 3.83 5.35 4.44 4.07 4.49 3.68 5.13

(10.04 mg/L) (67.8 mg/L) (2.02 mg/L) (16.29 mg/L) (45.11 mg/L) (14.26 mg/L) (97.47 mg/L) (4.66 mg/L)

84

M. Remko / Journal of Molecular Structure: THEOCHEM 916 (2009) 76–85

Table 5 Calculated absorption (%ABS), polar surface area (PSA) and Lipinski parameters of the anticoagulants studied.

a

No.

Drug

%ABS

Volume

PSA

NROTB

n ON acceptors

n OHNH donors

Log P, calcda

Formula weight

1 2 3 4 5 6 7 8

(S)-Rivaroxaban Apixaban (R,R)-Otamixaban Betrixaban Razaxaban (S,S)-DX-9065a Dabigatran Dabigatran etexilate

77.8 69.8 62.7 70.1 66.5 65.3 55.8 54.5

351.74 406.55 407.73 392.76 423.70 410.68 419.63 583.46

88.18 110.77 130.73 107.41 120.04 123.50 150.22 154.05

5 5 9 7 7 8 9 17

8 9 8 8 10 7 10 12 (viol.)

1 2 4 3 3 5 5 3

1.49 2.15 2.71 3.57 3.01 3.01 2.35 5.37

435.89 459.51 446.51 451.91 528.47 (viol.) 444.53 471.52 627.75 (viol.)

Range of log P values obtained by three theoretical methods (Table 3) (ALOGPS, KoWWIN, XLogP2).

have log P 6 5, molecular weight 6500, number of hydrogen bond acceptors 610, and number of hydrogen bond donors 65. This rule is widely used as a filter for drug-like properties. Table 5 contains calculated percentage of absorption (%ABS), molecular polar surface area (PSA) and Lipinski parameters of the anticoagulants investigated. Magnitude of absorption is expressed by the percentage of absorption. Absorption percent was calculated [68] using the expression: %ABS = 109  0.345 PSA. Polar surface area (PSA) was determined by the fragment-based method of Ertl and coworkers [31–33]. One ligand (razaxaban) violated ‘‘rule of five” (too high molecular weight). It was discontinued for further development in 2005 because another oral fXa inhibitor (apixaban) with more safety profile was under development by the same company [74]. The structural redesign of razaxaban issued in reduction of molecular weight, lowering of lipophilicity, reducing the total number of proton donor and proton acceptor groups and increase in absorption of apixaban compared to the parent razaxaban (Table 5). Low number of rotatable bonds in rivaroxaban and apixaban indicates that these ligands upon binding to fXa receptor should change their conformation only slightly. However, apixaban under binding undergoes appreciable conformational change, which is needed to accommodate phenyllactam moiety and bicyclic pyrazole scaffold in the proper space of the fXa receptor cavity (Table 1). Its relatively high value of polar surface area (110.7) results in worsening of the absorption in comparison with rivaroxaban (Table 5). Indeed, rivaroxaban with PSA value 88 belong to the anticoagulants with increased absorption. Direct thrombin inhibitor dabigatran is very flexible molecule with high total number of proton donor and proton acceptor groups (15), high PSA (150) and lowest absorption of the compounds studied (Table 5). Dabigatran (like otamixaban and DX-9065a) contain ionizable acidic and/or basic functional groups, and at physiological conditions are present in the form of charged species. Ionization of acidic or basic groups and high PSA of these drugs are not compatible with their oral application. Otamixaban and DX-9065a are administered intravenously. Oral administration of dabigatran is enabled through its double prodrug eliminating charges and thus increasing the lipophilicity of the molecule (Table 5).

(ii)

(iii)

(iv)

(v)

(vi)

B3LYP computed structures using two basis sets of the drugs investigated shows that the increase of the basis set gives essentially the same results. Water has a remarkable effect on the geometry of the anticoagulants studied. The anticoagulant drugs exhibit the largest stability in solvent as expected. The energy difference between gas phase and solvated phase was significant for the both solvation models employed in this work. Computed partition coefficients (ALOGPS method) for drugs studied varied between 1.7 and 3.9. Neutral compounds are described as lipophilic drugs. Rivaroxaban is drug with lowest lipophilicity. The neutral anticoagulants studied are only slightly soluble in water, their computed solubilities from interval between 5 and 70 mg/L are sufficient for fast absorption. Experimentally determined solubility of rivaroxaban (8 mg/L) is very well interpreted by calculation. Rivaroxaban with PSA value 88 belongs to the anticoagulants with increased absorption. Direct thrombin inhibitor dabigatran is very flexible molecule with high total number of proton donor and proton acceptor groups (15), high PSA (150) and lowest absorption of the compounds studied. This work yields quantities that may be inaccessible or complementary to experiments and represents the first theoretical approach in which both the gas-phase and solvated phase properties of clinically useful second-generation anticoagulants were evaluated. Such investigations may be, due to the present recognition of the important potential commercial value of accurate prediction of geometry, lipophilicity, solubility, and absorption as very important factors for the designing of highly active ligands selectively acting on fXa and thrombin receptors, useful in design of new drugs in the prevention and treatment of a broad variety of conditions, including the prevention of venous thromboembolism, manifesting as deep-vein thrombosis or pulmonary embolism, in patients undergoing major orthopaedic or general surgery, acutely ill nonsurgical patients and cancer patients with improved properties and intellectual property value.

4. Conclusions This theoretical study set out to determine stable conformations, solvent effect, lipophilicity, solubility, absorption and polar surface area of eight ligands acting on fXa and thrombin receptors for which a relatively small amount of experimental physicochemical data exist, considering its pharmacological importance. Using the theoretical methods the following conclusions can be drawn. (i) In the case of the Xa inhibitors (rivaroxaban, apixaban, otamixaban, betrixaban, razaxaban, and DX-9065a) the fully optimized most stable conformers possess characteristic L-shape structure. Examination of the space models of the

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.theochem.2009.09.011. References [1] [2] [3] [4] [5]

D. Kubitza, S. Haas, Expert Opin. Investig. Drugs 15 (2006) 843. A.G.G. Turpie, Arterioscler. Thromb. Vasc. Biol. 27 (2007) 1238. P.L. Gross, J.I. Weitz, Arterioscler. Thromb. Vasc. Biol. 28 (2008) 380. S. Haas, J. Thromb. Thromb. 25 (2008) 52. F.F. van Doormaal, H.R. Büller, S. Middeldorp, Crit. Rev. Oncol. Hematol. 66 (2008) 145. [6] U.R. Desai, Med. Res. Rev. 24 (2004) 151.

M. Remko / Journal of Molecular Structure: THEOCHEM 916 (2009) 76–85 [7] Ch.A. Kontogiorgis, D. Hadjipavlou-Litina, Med. Res. Rev. 24 (2004) 687. [8] S. Roehrig, A. Straub, J. Pohlmann, Th. Lampe, J. Pernerstorfer, K.-H. Schlemmer, P. Reinemer, E. Perzborn, J. Med. Chem. 48 (2005) 5900. [9] H. Brandstetter, A. Kühne, W. Bode, R. Hubert, W. Von der Saal, K. Wirthensohn, R.A. Engh, J. Biol. Chem. 271 (1996) 29988. [10] D.J.P. Pinto, M.J. Orwat, S. Koch, K.A. Rossi, R.S. Alexander, A. Smallwood, P.C. Wong, A.R. Rendina, J.M. Luettgen, R.M. Knabb, K. He, B. Xin, R.R. Wexler, .P.Y.S. Lam, J. Med. Chem. 50 (2007) 5339. [11] K.R. Guertin, Ch.J. Gardner, S.I. Klein, A.L. Zulli, M. Czekaj, Y. Gong, A.P. Spada, D.L. Cheney, S. Maignan, J.-P. Guilloteau, K.D. Brown, D.J. Colussi, V. Chu, Ch.L. Heran, S.R. Morgan, R.G. Bentley, Ch.T. Dunwiddie, R.J. Leadley, H.W. Pauls, Bioorg. Med. Chem. Lett. 12 (2002) 1671. [12] M.L. Quan, P.Y.S. Lam, Q. Han, D.J.P. Pinto, M.Y. He, R. Li, Ch.D. Ellis, Ch.G. Clark, Ch.A. Teleha, J.-H. Sun, R.S. Alexander, S. Bai, J.M. Luettgen, R.M. Knabb, P.C. Wong, R.R. Wexler, J. Med. Chem. 48 (2005) 1729. [13] M. Remko, W.C. von der Lieth, J. Chem. Inf. Mod. 46 (2006) 1194. [14] M. Remko, W.C. von der Lieth, J. Chem. Inf. Mod. 46 (2006) 1687. [15] M. Remko, M. Swart, F.M. Bickelhaupt, J. Phys. Chem. B 111 (2007) 2313. [16] M. Remko, P.T. van Duijnen, W.C. von der Lieth, J. Mol. Struct. Theochem. 814 (2007) 119. [17] M. Remko, W.C. von der Lieth, J. Phys. Chem. A 111 (2007) 13484. [18] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman, Jr. J.A. Montgomery, T. Vreven, T.K.N. Kudin, J.C. Burant, J.M. Millam, S.S. Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G.A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J.E. Knox, H.P. Hratchian, J.B. Cross, C. Adamo, C. Jaramillo, R. Gomperts, R.E. Stratmann, O. Yazyev, A.J. Austin, R. Cammi, C. Pomelli, J.W. Ochterski, P.Y. Ayala, K. Morokuma, G.A. Voth, P. Salvador, J.J. Dannenberg, V.G. Zakrzewski, S. Dapprich, A.D. Daniels, M.C. Strain, O. Farkas, D.K. Malick, A.D. Rabuck, K. Raghavachari, J.B. Foresman, J.V. Ortiz, Q. Cui, A.G. Baboul, S. Clifford, J. Cioslowski, B.B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R.L. Martin, D.J. Fox, T. Keith, M.A. Al-Laham, C.Y. Peng, A. Nanayakkara, M. Challacombe, P.M.W. Gill, B. Johnson, W. Chen, M.W. Wong, C. Gonzalez, J.A. Pople, Gaussian 03, Revision D.01, Gaussian, Inc., Wallingford CT, 2004. [19] W.J. Hehre, L. Radom, P.v.R. Schleyer, J.A. Pople, Ab Initio Molecular Orbital Theory, Wiley, New York, 1986. [20] W. Kohn, L.J. Sham, Phys. Rev. A 140 (1965) 1133. [21] For a recent perspective, see: E.J. Baerends, Theor. Chem. Acc. 103 (2000) 265. [22] A.D. Becke, Phys. Rev. A38 (1988) 3098. [23] A.D. Becke, J. Chem. Phys. 98 (1993) 5648. [24] C. Lee, W. Yang, R.G. Parr, Phys. Rev. B37 (1988) 785. }u } man, J. Chem. Soc. Perkin Trans. 2 (1993) 799. [25] A. Klamt, G. Schu [26] M. Cossi, N. Rega, G. Scalmani, V. Barone, J. Comp. Chem. 24 (2003) 669. [27] L. Onsager, J. Am. Chem. Soc. 58 (1936) 1486. [28] http://www.vcclab.org. [29] I.V. Tetko, J. Gasteiger, R. Todeschini, A. Mauri, D. Livingstone, P. Ertl, V.A. Palyulin, E.V. Radchenko, N.S. Zefirov, A.S. Makarenko, V.Y. Tanchuk, V.V. Prokopenko, J. Comput. Aided Mol. Des. 19 (2005) 453. [30] I.V. Tetko, P. Bruneau, H.-W. Mewes, D.C. Rohrer, G.I. Poda, Drug Discov. Today 11 (2006) 700. [31] P. Ertl, P. Selzer, in: J. Gasteiger (Ed.), Handbook of Chemoinformatics: From Data to Knowledge, Wiley-VCH, Weinheim, 2003, pp. 1336–1348. [32] P. Ertl, B. Rohde, P. Selzer, J. Med. Chem. 43 (2000) 3714. [33] http://www.molinspiration.com. [34] C.S. Winters, L. Shields, P. Timmins, P. York, J. Pharm. Sci. 83 (1994) 300.

85

[35] P. Carloni, F. Alber (Eds.), Quantum Medicinal Chemistry, Wiley-VCH GmbH & Co. KGaA, Weinheim, 2003. [36] K.A. Brameld, B. Kuhn, D.C. Reuter, M. Stahl, J. Chem. Inf. Mod. 48 (2008) 1. [37] Protein Data Bank. Available from: http://www.rscb.org/pdb/ (accessed May 2009). [38] M.W. Wong, M.J. Frisch, K.B. Wiberg, J. Am. Chem. Soc. 113 (1991) 4776. [39] M.W. Wong, M.J. Frisch, K.B. Wiberg, J. Am. Chem. Soc. 114 (1992) 523. [40] M.W. Wong, M.J. Frisch, K.B. Wiberg, J. Chem. Phys. 95 (1991) 8991. [41] M.W. Wong, M.J. Frisch, K.B. Wiberg, J. Am. Chem. Soc. 114 (1992) 1645. [42] P.Th. van Duijnen, A.H. De Vries, Int. J. Quant. Chem. Quant. Biol. Symp. 29 (1995) 523. [43] C.N. Schultz, A. Warshel, Proteins 44 (2001) 400. [44] J. Tomasi, B. Mennucci, R. Cammi, Chem. Rev. 105 (2005) 2999. [45] J. Kongsted, P. Söderhjelm, U. Ryde, J. Comput. Aided Mol. Des. 23 (2009) 395. [46] C.C. Pye, T. Ziegler, Theor. Chem. Acc. 101 (1999) 396. [47] N.H. Hauel, H. Nar, H. Priepke, U. Ries, J.-M. Stassen, W. Wienen, J. Med. Chem. 45 (2002) 1757. [48] F.H. Allen, S.E. Harris, R. Taylor, J. Comput. Aided Mol. Des. 10 (1996) 247. [49] H.J. Boehm, G. Klebe, Angew. Chem. Int. Ed. 35 (1996) 2588. [50] P. Zhang, W. Huang, L. Wang, L. Bao, Z.J. Jia, S.M. Bauer, E.A. Goldman, G.D. Probst, Y. Song, T. Su, J. Fan, Y. Wu, W. Li, J. Woolfrey, U. Sinha, P.W. Wong, S.T. Edwards, A.E. Arfsten, L.A. Clizbe, J. Kanter, A. Pandey, G. Park, A. Hutchaleelaha, J.L. Lambing, S.J. Hollenbach, R.M. Scarborough, B.-Y. Zhu, Bioorg. Med. Chem. Lett. 19 (2009) 2179. [51] P.C. Wong, E.J. Crain, C.A. Watson, R.R. Wexler, P.Y.S. Lam, M.L. Quan, R.M. Knabb, J. Thromb. Thrombolysis 24 (2007) 43. [52] D. Kikelj, Patophysiol. Haemost. Thromb. 33 (2003/2004) 487. [53] A.C. Spyropoulos, Thromb. Res. 123 (2008) S29. [54] A. Avdeef, Curr. Top. Med. Chem. 1 (2001) 277. [55] T.T. Oprea, J. Comput. Aid. Mol. Des. 16 (2002) 325. [56] U. Norinder, A.S. Bergström, ChemMedChem 1 (2006) 920. [57] I.V. Tetko, V.Y. Tanchuk, J. Chem. Inf. Comput. Sci. 42 (2002) 1136. [58] I.V. Tetko, V.Y. Tanchuk, A.E.P. Villa, J. Chem. Inf. Comput. Sci. 41 (2001) 1407. [59] I.V. Tetko, V.Y. Tanchuk, T.N. Kasheva, A.E.P. Villa, J. Chem. Inf. Comput. Sci. 41 (2001) 1488. [60] K.V. Balakin, N.P. Savchuk, I.V. Tetko, Curr. Med. Chem. 13 (2006) 223. [61] W.M. Meylan, P.H. Howard, J. Pharm. Sci. 84 (1995) 83. [62] http://www.syrres.com/. [63] R.X. Wang, Y. Fu, L.H. Lai, J. Chem. Inf. Comput. Sci. 37 (1997) 615. [64] R.X. Wang, Y. Gao, L.H. Lai, Drug Discov. Des. 19 (2000) 47. [65] EMEA File Xarelto. Available from: www.emea.europa.eu. [66] M. Remko, M. Swart, F.M. Bickelhaupt, Bioorg. Med. Chem. 14 (2006) 1715. [67] T. Taskinen, J. Yliruusi, Adv. Drug Deliv. Rev. 55 (2003) 1163. [68] Y.H. Zhao, M.H. Abraham, J. Lee, A. Hersey, Ch.N. Luscombe, G. Beck, B. Sherborne, I. Cooper, Pharm. Res. 19 (2002) 1446. [69] Y. Fujii, M. Takahashi, H. Morita, H. Kikuchi, Y. Aramaki, G.L. Amidon, Drug Metab. Pharmacokinet. 22 (2007) 26. [70] D.F. Veber, S.R. Johnson, H.(Y. Cheng, B.R. Smith, K.W. Ward, K.D. Kapple, J. Med. Chem. 45 (2002) 2615. [71] H.H.F. Refsgaard, B.F. Jensen, P.B. Brockhoff, S.B. Padkj?r, M. Guldbrandt, M.S. Christensen, J. Med. Chem. 48 (2005) 805. [72] I. Muegge, Med. Res. Rev. 23 (2003) 302. [73] C.A. Lipinski, F. Lombardo, B.W. Dominy, P.J. Feeney, Adv. Drug Delivery Rev. 23 (1997) 3. [74] A.C. Spyropoulos, Expert Opin. Investig. Drugs 16 (2007) 431.