Crystal structures, absolute configurations and molecular docking studies of naftopidil enantiomers as α1D-adrenoceptor antagonists.

Chiral drug naftopidil (NAF), a specific α1D-adrenoceptor (AR) antagonist for the treatment of benign prostatic hyperplasia, was used in racemic form for several decades. Our recent work declared that NAF enantiomers showed the same antagonistic effects on the α1D-AR, but the binding mechanism of these two stereochemical NAF isomers to the α1D receptor remained unclear. Herein, we reported the crystallographic structures of optically pure NAF stereoisomers for the first time and unambiguously determined their absolute configurations. The crystal data of R and S enantiomers matched satisfactorily the pharmacophore model for α1D-selective antagonists. Based on the constructed α1D homology model, molecular docking studies shed light on the molecular mechanism of NAF enantiomers binding to α1D-AR. The results indicated that NAF enantiomers exhibited the very similar binding poses and occupied the same binding pocket.

Introduction

Benign prostatic hyperplasia (BPH) is a progressive condition characterized by a nodular enlargement of the prostate resulting in obstruction of the urethra. Emerging contenders to current therapies is focusing on drug targets which are able to relax prostatic smooth muscle in a similar way to the α1-adrenoceptor (AR) antagonists (α1A-, α1B- and α1D-AR), as this appears to be the most effective mechanism of action2, 3.

Naftopidil (NAF, Fig. 1) is a chiral drug with high selectivity for α1A- and α1D-AR over than for α1B subtype, and exhibits significant clinical efficacy for alleviating lower urinary tract symptoms (LUTS) associated with BPH. However, NAF still remains to be used under racemic form. We also know that the physiochemical and biochemical properties of racemic mixtures and individual stereoisomers can differ significantly. Additionally, stereoselective metabolism of chiral compounds can influence pharmacokinetics, pharmacodynamics, and toxicity. Appropriate chiral antidotes must be selected for therapeutic benefit and to minimize adverse events.

Figure 1

Chemical structure of naftopidil. The asterisk (*) indicates the chiral center.

Individual NAF enantiomers could be obtained by enantioselective synthesis7, 8 and hydrolytic kinetic resolution, but their crystal structures had not been reported so far. We herein described the crystallographic structures of (+)/(–)-NAF and determined their absolute configurations based on single-crystal X-ray diffraction analysis. Moreover, molecular docking studies explored the molecular mechanisms of NAF enantiomers binding to the homology-modeled α1D-AR, which helps to rationally explain their antagonistic activities. This work would provide valuable information for the relationships between stereostructures of chiral molecules and bioactivities.

Materials and methods

All reagents and solvents were of analytical grade and commercially available. The 1H NMR spectra were recorded on a Bruker Avance instrument using CDCl3 as a solvent and TMS as an internal standard, and coupling constants (J) were quoted in Hz. Optical rotation measurements were obtained using a Rudolf AUTOPOL IV polarimeter. Single-crystal X-ray diffraction data were collected on a Rigaku RAPID II diffractometer with Cu Kα radiation (λ=1.54178 Å).

Chemistry

(+)/(–)-NAF isomer (ee purity >99.5%) was purchased from Boehringer Mannhei (Ingelheim, Germany). The structure of (+)-NAF was characterized by 1H NMR and high-resolution mass spectrometry (HR-MS). 1H NMR (300 MHz, CDCl3) δ 8.19 (d, J=9.3 Hz, 1H), 7.82 (d, J=9.2 Hz, 1H), 7.57–7.43 (m, 3H), 7.37 (t, J=7.9 Hz, 1H), 7.09 (dd, J=10.2, 6.6 Hz, 1H), 6.91 (t, J=7.5 Hz, 3H), 6.82 (d, J=7.6 Hz, 1H), 5.66 (s, 1H), 4.92 (s, 1H), 4.49–4.30 (m, 1H), 4.24–4.04 (m, 1H), 3.88 (s, 5H), 3.54 (s, 4H), 3.46 (d, J=8.1 Hz, 2H), 3.28 (d, J=15.8 Hz, 2H). HR-MS (ESI) m/z Calcd. for C24H28N2O3 [M + H]+, 393.2100; Found, 393.2104.

X-ray crystallography

Suitable crystals of NAF enantiomers were obtained by slowly evaporating a mixture of dichloromethane and n-hexane solution at ambient temperature. High-quality colorless crystals were mounted on a glass fiber in a random orientation. The data were collected by an R-AXIS RAPID II diffractometer equipped with graphite-monochromatic Cu Kα radiation (λ=1.54178 Å) by using the ω scan mode. The structures were solved by direct methods using Olex2 software, and the non-hydrogen atoms were located from the trial structure and then refined anisotropically with SHELXL using a full-matrix least squares procedure based on F. The weighted R factor, wR and goodness-of-fit S values were obtained based on F2. The hydrogen atom positions were fixed geometrically at the calculated distances and allowed to ride on the parent atoms. Crystallographic data excluding structure factors have been deposited at the Cambridge Crystallographic Data Center (CCDC). CCDC 1023461 contains the supplementary crystallographic data for this paper.

Homology modeling and molecular docking

The amino acidic sequences of the human α1D receptor were retrieved from SwissProt database (entry code: P25100, ADA1D_HUMAN).

The homology model of α1D subtype was successfully produced by our previous work, and then submitted to be energy optimization by using CHARMMing program. Structural evaluation and stereochemical analyses were performed using PROCHECK, PROVE, CRYST and Ramchandran plot. PyMOL software was employed for checking and validating protein structures after model refinement.

The crystallographic structures of NAF enantiomers were saved in mol2 format. The preparation of the pdbqt files was done by standard procedure using AutoDock Tools 1.4.6. The docking procedures were performed in AutoDock Vina using the default scorning function. The binding site was identified according to previous studies. Exhaustiveness was set to 100, and number of output conformations was set to 20. The searching seed was random. The calculated geometries were ranked in terms of free energy of binding and the best poses were selected for further analysis.

Results and discussion

Crystal structures of NAF enantiomers

(+)/(–)-NAF were converted to their hydrochloride salts, i.e., NAF·2(HCl), with [α]25D +23.7° (c 0.439, CH3OH) and –24.0° (c 0.481, CH3OH), respectively. Both enantiomers (+)-NAF·2(HCl) and (–)-NAF·2(HCl) crystallized in the monoclinic space group P21 with one crystallographically independent molecule in the asymmetric unit. Their representative crystal structures are presented in Fig. 2 and exhibit good mirror symmetry. Crystal data and structural refinement are shown in Table 1.

Figure 2

Crystallographic structures of (R)-(+)-NAF·2(HCl) (upper) and (S)-(–)-NAF·2(HCl) (lower). Displacement ellipsoids are drawn at the 30% probability level.

Table 1

Crystal data and structural refinement of compounds (R)-(+)-NAF·2(HCl) and (S)-(–)-NAF·2(HCl).

Compd.	(R)-(+)-NAF·2(HCl)	(S)-(–)-NAF·2(HCl)
Chemical formula	C24H28N2O3·2(HCl)	C24H28N2O3·2(HCl)
MW	465.40	465.40
Crystal system, space group	Monoclinic, P21	Monoclinic, P21
a, b, c (Å)	11.777(2), 5.7595(12), 17.464(4)	11.776(2), 5.7561(12), 17.462(4)
α, β, γ (°)	90, 95.03(3), 90	90, 95.02(3), 90
V (Å3)	1180.0(4)	1179.1(4)
Z	2	2
ρcalc (g/cm3)	1.310	1.311
μ (mm-1)	2.697	2.700
F (000)	492.0	492.0
Crystal size (mm3)	0.3×0.2×0.2	0.3×0.2×0.2
Radiation	Cu Kα (λ=1.54178)	Cu Kα (λ=1.54178)
θ range (°)	3.767 to 68.212	3.768 to 68.220
Tmin/Tmax	0.550/0.583	0.549/0.583
Reflections collected/unique/observed	19516/3984/2759	21209/4089/1825
Goodness-of-fit on F2	1.119	1.181
R1/wR2 [I≥2σ (I)]	0.0552/0.1203	0.1085/0.2593
Δρmax/Δρmin (e Å–3)	0.49/—0.28	0.54/—0.40
Flack/Hooft parameters	0.012(14)/0.022(14)	0.04(2)/0.086(17)

The dihedral angle between benzene ring and naphthalene plane was 18.0(3)° for (+)-NAF and 18.1(5)° for (–)-NAF, respectively. The piperazine ring indicated a chair-type geometry. Interestingly, intermolecular H-bonds (O–H···Cl, N–H···Cl and C–H···Cl, Fig. 3 and Table 2) played critical roles in stabilizing the packing structures. In addition, NAF in the crystal was assembled in a way to yield a high density (1.310 g/cm3), in which one and/or two overlapping molecules were regularly arranged and were thought to be kept in balance by intermolecular van der Waals forces.

Figure 3

Crystal packing of (R)-NAF·2(HCl) (A) and (S)-NAF·2(HCl) (B) along the b axis. Black dashed lines show the intermolecular H-bonds.

Table 2

Intermolecular hydrogen bonds for compounds (+)/(–)-NAF·2(HCl) (Å, °).

D---H···A	D---H		H···A	D···A	D---H···A (°)
(+)-NAF·2(HCl)
O(1)---H(1)···Cl(1)a		0.82	2.44	3.1767(7)	151
N(2)---H(2)···Cl(2)b		0.98	2.04	3.0016(6)	168
C(10)---H(10B)···Cl(1)c		0.97	2.72	3.4258(7)	130
(–)-NAF·2(HCl)
N(1)---H(1 A)···Cl(2)d		0.98	2.00	2.9686(7)	172
O(1)---H(1B)···Cl(2)e		0.82	2.72	3.1788(7)	117
C(10)---H(10B)···Cl(2)f		0.97	2.71	3.4208(8)	130

Symmetry code: ax,—1 + y, z; b1 —x,—1/2 + y, 1 —z; c1 —x, 1/2 + y, 1 —z; d1 + x, y, z; e1 —x, —1/2 + y, —z; f1 + x, —1 + y, z.

We could determine the absolute structure of (+)-NAF·2(HCl) based on the calculated Flack parameter 0.012(14). The Hooft parameter of 0.022(14) was also sufficient to confirm the absolute structure. The absolute configuration of the chiral center of (+)-NAF·2(HCl) was thus determined to be R. Similarly, (–)-NAF·2(HCl) was also unambiguously assigned to be S since the small Flack and Hooft parameters 0.04(2) and 0.086(17), respectively.

Interestingly, crystal data of NAF enantiomers were satisfied with the pharmacophoric model for selective α1D-AR antagonists (Table 3). It can be seen that the measured distances of PI-HY1 (5.6 Å) and PI-HBA (4.2 Å) were nearly equal to that of the α1D model.

Table 3

Visualization of pharmacophoric features of NAF based on Barbaro׳s model and comparison of important distances between pharmacophoric features in reported subtype-selective α1-AR antagonists and crystallographic structures of NAF enantiomers. Colour legend: green, hydrophobic features (HY); blue, positive ionizable (PI); rose, hydrogen bond donor (HBD); red, hydrogen bond acceptor (HBA).

X-ray structure	Distance (Å)
	PI-HY1	PI-HBA	PI-HY3	PI-HBD
α1A-AR antagonists	5.5	7.1	–	–
α1B-AR antagonists	6.2	–	7.8	4.9
α1D-AR antagonists	5.4	4.5	–	–
(R)-NAF	5.6	4.2	6.8	3.1
(S)-NAF	5.6	4.2	6.9	3.1

– Not applicable.

Molecular docking studies

(R)-NAF, (S)-NAF and racemic NAF showed similar α1D-AR antagonistic effects with the pA2 values of 7.85, 8.03 and 7.93, respectively. Although (R)- and (S)-NAF exhibited the same high affinity towards α1D-AR, the binding mechanisms of NAF enantiomers to the α1D receptor were still unclear. Then molecular docking analysis was performed, which might shed light on the antagonistic properties of NAF enantiomers over α1D-AR.

α1D-AR is a member of the G protein—coupled receptors (GPCRs) family that are constructed by seven transmembrane (TM) helices, N- and C-terminal fragments, and intra- and extracellular loop (ECL) regions21, 22. Molecular docking was performed on α1D receptor constructed by homology model building using the AutoDock-vina program since the accurate 3D structures of α1D-AR with high resolutions has not been obtained yet. To achieve the reliable docking results, the lowest energy conformations of NAF enantiomers were extracted from their crystal structures and the α1D-AR model was submitted to be energy optimization by using CHARMMing program. The top ranked poses of (R)-NAF and (S)-NAF (Fig. 4A and B) both positioned in the hydrophobic pocket involving TM 2, 3, 6 and 7 with the same calculated binding energies (—9.0 kcal/mol). The OH group of (R)-NAF formed a hydrogen bond (2.6 Å) with Glu190 in the ECL2 region that has been reported to be essential for GPCR activation. The methoxyl at the arylpiperazine moiety formed an H-bond with Thr189 (3.0 Å between the oxygen atom of methoxyl group and the hydroxyl oxygen atom of Thr189). The protonated piperazine moiety formed an electrostatic interaction (3.1 Å) with Thr189 of ECL2, and the benzene ring was mainly engaged in hydrophobic interactions with Phe185 and Trp175 residues. Additionally, the naphthalene moiety was placed in the hydrophobic region among TM5, TM6 and TM7, and contacted via hydrophobic interactions with residues Phe304, Phe305 and Phe324. As compared to the binding mode for (R)-NAF-α1D complex, (S)-NAF showed very similar binding behavior (Fig. 3B). On the basis of the similar binding poses and binding energies of (R)/(S)-NAF with α1D receptor, we can rationally explain the similar antagonistic activities towards α1D. On the other hand, it indicated that the α1D homology model was feasible and useful for virtual screening of the α1D-selective blockers. Furthermore, we inferred that residues Glu190 and Thr189 played an important role in recognizing the α1D subtype, especially for arylpiperazine-based antagonists.

Figure 4

(A) The top-ranked docking poses of (R)-NAF (yellow carbons) and (S)-NAF (magenta carbons) into the putative binding sites of α1D-AR. (B) α1D (surface) –ligand (stick) complex. The two antagonists are shown in stick representation. The receptors are shown in cartoon representation with red alpha helices and green loops. The seven TM helices are labeled by 1, 2, 3, 4, 5, 6 and 7, respectively. Dashed lines represent the hydrogen bonds or electrostatic interactions.

Conclusions

In this work, we reported the crystallographic structures of NFA enantiomers for the first time, and unambiguously determined their absolute configurations based on the Flack and Hooft parameters. In crystal packing, specific intermolecular hydrogen bonds [O–H···Cl, N–H···Cl and C–H···Cl] are found to stabilize the three-dimensional structure. Furthermore, NAF enantiomers fitted well with the ligand-based pharmacophore model for α1D-selective antagonists.

(R)- and (S)-NAF exhibited the similar antagonistic activities towards α1D-AR, but the underlying mechanisms still remain unclear. Molecular docking studies revealed the binding modes of NAF enantiomers to the α1D receptor associated with their antagonistic effects. Docking results indicated that the OH group via H-bond contacted with Glu190 in the ECL2, which might play an important role in the recognition of α1D-AR. The arylpiperazine part was placed on the entrance of hydrophobic pocket, and the naphthalene moiety entered into a deep hydrophobic region surrounded by TM 5, 6 and 7. Binding mode of (R)-NAF was very similar to that of (S)-NAF, which was consistent with our previous report that the enantiomers had the same antagonistic potency for α1D-AR. Details of NAF enantiomers binding mode provide valuable clues for the design of selective α1D-AR antagonists in the future.