Skip to content
Publicly Available Published by De Gruyter December 8, 2016

Invariom-based comparative electron density studies of iso-sildenafil and sildenafil

  • Birger Dittrich EMAIL logo and Peter Luger EMAIL logo

Abstract

The title compounds have raised considerable medical and broad public interest in that sildenafil is used as an agent against male erectile dysfunction; iso-sildenafil is not in clinical use. A comparison of their structural and electronic properties therefore seems of interest. The electron densities of iso-sildenafil and the cationic and neutral forms of sildenafil were examined by the application of the invariom formalism relying on diffraction data reported in the literature. The electron-density distributions obtained were subjected to topological analysis using the quantum theory of atoms in molecules (QTAIM) formalism to yield bond topological and atomic properties. Moreover, molecular Hirshfeld surfaces and electrostatic potentials (ESPs) were calculated. A number of structural and electronic differences were thus identified between sildenafil and the iso-analog. In both sildenafil structures, the phenyl ring and the pyrazolopyrimidine fragment are practically coplanar (planar conformation), whereas in the iso-analog they exhibit an angle of 44° (inclined form). Related to differences in molecular structure are completely different hydrogen bonding patterns and differences in the ESPs, the latter ones being influenced by different methylation at the pyrazolopyrimidine fragment. Iso-sildenafil is present as a hydrogen-bond dimer in the crystal, and the ESP of this dimer is dominated by a surrounding positive potential.

1 Introduction

It raised exceptional attention in the broad public, when sildenafil (Viagra), an agent against male erectile dysfunction, came into use in 1998. Sildenafil is an inhibitor of phosphodiesterase 5 (PDE5). This enzyme degrades cyclic guanidine monophosphate which in turn supports the blood flow in the penis and hence stabilizes an erection.

Since then the average turnover of this sought-after drug was 1.7 billion US$ per year, until the patent expired in June 2013. Subsequently, generics appeared on the market. Moreover, related compounds with the same indication were introduced: tadalafil (Cialis) in 2002, vardenafil (Levitra) in 2003 and avanafil (Spedra) in 2013.

The first single-crystal X-ray structure analysis of a sildenafil compound was published in 1999 on iso-sildenafil [1], which differs from sildenafil only by methylation at N2 (instead of N1) of the pyrazolopyrimidine fragment (Fig. 1) and which is not in clinical use.

Fig. 1: Formulae of iso-sildenafil (I), top, sildenafil free base (II), middle, and protonated sildenafil in the saccharinate salt (III), bottom.
Fig. 1:

Formulae of iso-sildenafil (I), top, sildenafil free base (II), middle, and protonated sildenafil in the saccharinate salt (III), bottom.

Despite worldwide attention it surprisingly took until 2005 for a first structure determination by X-ray diffraction of sildenafil itself (as a citrate salt) to appear [2], followed 1 year later by a detailed study on 12 solvates of sildenafil saccharinate [3]. It then took further 6 years until in 2012 a structure analysis of the free base of sildenafil was reported [4]. In 2013, a group of further solvate structures were published for sildenafil, co-crystallized with seven different dicarboxylic acids [5]. Recently, sildenafil salicylate [6] and citrate [7] structure analyses were reported (CSD version 5.37, Feb. 2016 [8]).

Macromolecular structure determinations of the catalytic domain of human PDE5 complexed with sildenafil, tadalafil and vardenafil were published in 2003 [9].

Electron-density distributions (EDs) of the neutral and cationic forms of sildenafil and of the iso-analog were generated by an application of the invariom formalism [10], [11]. This provides, in addition to molecular connectivity and conformation, further information on the two compounds. We made use of the known low-resolution X-ray diffraction data sets of iso-sildenafil (I) from [1] (213 K, (sinθ/λ)max=0.63 Å−1, CCD refcode CAXZEG [8]), sildenafil free base (II) from [4] (190 K, (sinθ/λ)max=0.65 Å−1, CCD refcode QEGTUT [8]) and sildenafil saccharinate acetonitrile (III) from [3] (100 K, (sinθ/λ)max=0.59 Å−1, CCD refcode QEMLIE [8]), for this comparative ED study (see Fig. 1).

2 Application of the invariom formalism

The known atomic parameters were taken from the independent atom models (IAM) reported in [1], [4] and [3] to establish starting parameters for subsequent aspherical-atom refinements. For aspherical modeling of iso-sildenafil (I), sildenafil (II) and the sildenafil saccharinate acetonitrile complex (III), the software InvariomTool [12] was applied, which automatically selects and assigns the corresponding invarioms after analyzing the identity and chemical neighborhood of each atom. Scattering factors up to the hexadecapolar level, including κ parameters [13], were taken from the invariom library and were assigned to all atoms. Bond distances to hydrogen atoms were elongated to values from energy-minimized structures of the respective model compounds also used in a scattering-factor assignment. After invariom transfer, refinement of positional and displacement parameters (anisotropic for non-hydrogen atoms, isotropic for hydrogen atoms) and fixed multipolar parameters was carried out making use of the xd2006 program suite [14] until convergence was achieved.

In all refinements the quantity Σhw(h)(|Fo(h)|2–|Fc(h)|2)2 was minimized by using the statistical weight w(h)=1/σ2(Fo(h)2) and only those structure factors which met the criterion   Fo22σ(Fo2) were included.

Neutrality constraints were applied to (I) and (II). For (III) no charge transfer was allowed between the sildenafil cation and the saccharinate anion, and the acetonitrile solvent molecule was kept neutral. Selected crystallographic data and figures of merit are summarized in Table 1.

Table 1:

Selected crystallographic and refinement data for (I), (II) and the cation of (III).a

Compound(II)(III)
FormulaC22H30N6O4SC22H30N6O4S(C22H31N6O4S)(C7H4NO3S)(C2H3N)
Crystal systemMonoclinicMonoclinicTriclinic
Space group (No.)P21/n (14)P21/c (14)P1̅ (2)
Z442
V, Å32408.5(6)2420.0(3)1637.1(3)
(sinθ/λ)max, Å−10.630.650.59
T, K213190100
Unique reflections396353865672
Observed reflections [Fo2≥2σ(Fo2)]285221545304
Invariom refinement:
R(F)0.03970.06290.0284
Rall(F)0.07430.26590.0303
Rw(F)0.02220.04510.0343
R(F2)0.03980.06090.0622
Rall(F2)0.05050.11180.0623
Rw(F2)0.04230.08200.0687
Min/max ∆ρ, e Å−3−0.239/0.241−0.281/0.412−0.378/0.308
Gof1.251.382.58
Nref/Nv6.86.611.3

aFor further data, see [1], [3] and [4].

3 Results and discussion

3.1 Structural properties

The structures of the molecules (I) and (II) and of the cation of (III) as present in the crystal structures are shown in Fig. 2 in Ortep representations [15], [16]. The atomic numbering scheme used in the literature was kept for all three models.

Fig. 2: Ortep representations [15], [16] of (I), (II) and the cation of (III).
Fig. 2:

Ortep representations [15], [16] of (I), (II) and the cation of (III).

In addition to the chemical difference between sildenafil and the iso-analog (as mentioned above), the molecular conformations in the solid state are different, which can best be described by the seven torsion angles along the free rotatable bonds, which are τ1 and τ2 at the propyl group, τ3 and τ4 at the ethoxy group, τ5 and τ6 at the bonds from the phenyl ring via the sulfonyl S atom to the piperazine ring, and τ7 linking the phenyl ring and the pyrazolopyrimidine fragment (see Table 2). The most important difference is seen for τ7, which is close to zero in (II) and in (III) (planar form), but is 44.2(3)° in (I) (inclined form). This has consequences on the possible formation of an intramolecular hydrogen bond (HB) as will be discussed below. Further conformational differences, as expressed by τ2, are found in the rotation of the propyl group; otherwise, the molecular conformations of (II) and (III) are alike and comparable.

Table 2:

Selected torsion angles for (I), (II) and the cation of (III).a

Torsion angle(1) sequence(II) sequence(III) sequence
τ1C3–C31–C32–C33−59.8(4)C4–C18–C22–C21−174.8(5)C25–C21–C22–C23−178.5(1)
τ2C9–C3–C31–C3287.3(3)C3–C4–C18–C2245.0(7)C19–C25–C21–C22−170.2(1)
τ3C12–O12–C121–C122177.3(2)C1–O1–C13–C17−176.8(3)C16–O5–C17–C18178.7(1)
τ4C13–C12–O12–C12113.2(3)C8–C1–O1–C13−6.3(5)C15–C16–O5–C1718.1(2)
τ5C15–S1–N21–C2269.4(2)C9–S1–N3–C1464.5(3)C13–S1–N2–C2−66.7(1)
τ6N21–S1–C15–C16−98.8(2)N3–S1–C9–C11−84.9(3)N2–S1–C13–C1281.8(1)
τ7C12–C11–C5–N644.2(3)C1–C6–C2–N20.9(6)C16–C11–C27–N41.0(2)

aAtom numbering as in the original literature and as shown in Fig. 2.

3.2 Electron-density-derived properties

Invariom-based ED examination goes beyond “spherical” structure analysis with the IAM. The invariom approach provides non-spherical ED for a given structure and thus permits us to compute ED-derived properties, for both qualitative and quantitative methods of evaluation. Figure 3 gives a qualitative impression of the static deformation density in the main molecular plane of (III), consisting of the phenyl and the pyrazolopyrimidine rings. In contrast to IAM refinements, electron-density accumulations on all covalent bonds and in lone pair regions become visible.

Fig. 3: Static deformation electron density in the common plane of the phenyl and the pyrazolopyrimidine rings of the cation of (III). Blue/black/red contour lines for positive/zero/negative densities. Contour intervals at 0.1 e Å−3.
Fig. 3:

Static deformation electron density in the common plane of the phenyl and the pyrazolopyrimidine rings of the cation of (III). Blue/black/red contour lines for positive/zero/negative densities. Contour intervals at 0.1 e Å−3.

Quantitative ED properties of the title compounds (I), (II) and the cation of (III) were obtained from an analysis according to Bader’s quantum theory of atoms in molecules (QTAIM) formalism [17]. This analysis provides both bond critical points (BCPs, defined by the property that the gradient ∇ρ(r) vanishes at this point) and atomic properties by integration over the atomic basins bound by the zero-flux surfaces of the gradient vector field, which subdivide a structure into transferable substructures.

BCPs were located on all covalent bonds and on the H(donor)···X(acceptor) (X=N, O) linkages of the HBs. Values of the ED, Laplacians and ellipticities [ρ(rBCP), ∇2ρ(rBCP), ε] were calculated, providing information about the strength and nature of a bond. Selected averages over these properties for C–C bonds and selected properties of bonds involving heteroatoms are given in Table 3. Atomic properties were calculated using the TOPOND algorithm as implemented [19] in the XD subprogram XDprop [14]. A selection of atomic volumes and charges for (I)–(III) is given in Table 4. Only those atoms where |q |>0.30 e for one of the contributing structures are listed.

Table 3:

Averaged bond topological properties for (I), (II) and the cation of (III), for C–C bonds and some selected bonds involving heteroatoms.

BondLength, Åρ(rBCP), e Å−32ρ(rBCP), e Å−5εaNbC–C bond order nbc
C–C (aromatic)1.404(19)2.09(6)−17.0(9)0.22(2)271.58
C–C (single)1.511(21)1.74(5)−12.2(13)0.03(2)151.10
S–C1.756(6)1.47(1)−9.0(3)0.09(1)3
S–N1.693(13)1.67(2)−14.2(10)0.16(1)3
S–O1.429(4)2.21(1)+11.4(9)0.05(1)6
N–N1.350(3)2.42(1)−6.9(3)0.16(2)3
C(sp3)–N1.463(10)1.78(3)−9.7(16)0.09(4)12
C(sp3)–Nd1.490(4)1.86(2)−11.9(2)0.01(1)3

aThe ellipticity ε is defined by (λ1/λ2)−1, with λ1 and λ2 being the two principal negative curvatures of ρ(r) at a BCP. It is a measure for the asphericity and hence the double-bond character of a bond.

bN=number of entries contributing to the average.

cThe bond order nb was calculated as nb=exp[C1(ρ(rBCP)−C2)], with C1=1.0229 and C2=1.6459 [18].

dOnly (III).

Table 4:

Charges q and volumes Vtot for the atoms in (I)/(II)/(III). Only atoms with |q| ≥ 0.30 e in either (I), (II) and the cation of (III) are listed.

Atomaq, eVtot, Å3
S1/S1/S12.95/2.94/2.895.87/5.67/5.79
O1/O3/O7−1.31/−1.31/−1.3021.77/17.46/19.99
O2/O4/O6−1.30/−1.31/−1.3020.15/−21.49/17.78
O7/O2/O4−0.91/−0.90/−0.9317.39/19.03/19.24
O12/O1/O5−0.93/−0.92/−0.9113.18/13.67/13.82
N1/N4/N6−0.44/−0.63/−0.6316.90/11.08/9.73
N2/N6/N5−0.63/−0.44/−0.468.78/16.47/14.60
N4/N1/N3−0.79/−0.77/−0.7616.92/14.90/15.31
N6/N2/N4−0.82/−0.89/−0.9913.15/13.81/14.97
N21/N3/N2−0.79/−1.02/−0.9912.81/11.95/11.20
N24/N5/N1−0.81/−0.85/−0.9511.36/11.27/8.56b
C5/C2/C270.63/0.65/0.648.04/8.01/7.71
C7/C7/C261.07/1.05/1.076.64/7.03/7.39
C9/C3/C190.31/0.28/0.288.77/7.97/8.51
C12/C1/C160.37/0.38/0.389.00/8.61/8.33
C24/C19/C30.31/0.21/0.198.73/10.01/9.24
C121/C13/C170.29/0.30/0.317.74/7.48/7.74
H6/H1/H1AAc0.47/0.51/0.522.66/3.49/1.92

aAtom numbering according to (I), (II) and (III).

bProtonated in the cation of (III).

cAll further hydrogen atoms have charges and volumes in the range 0.03–0.13 e and 6–11 Å3, respectively.

It turned out that, as expected, the obtained covalent bond properties are practically the same for chemically equivalent bonds of (I)–(III), which also holds for the atomic properties of atoms with the same nearest neighbors.

For the 15 C(sp3)–C(sp3) bonds, the average ρ(rBCP) value of 1.511(21) e Å−3 yields a bond order of 1.10 [17], [18]. From 27 averaged C(sp2)–C(sp2) bonds, a bond order of 1.58 is derived. As to N–C(sp3) bonds, the three bonds to the protonated nitrogen atom in (III) should in principle be considered separately. However, except for the slightly longer N–C bond in (III) there is no further difference in the bond-topological descriptors when compared to (I) and (II).

Strong charge concentration is seen for the hetero atoms (Table 4). The strongest positive charge is carried by the sulfur atom with q≈3 e. Such pronounced Bader charges are not uncommon. In a recent study on C/Si analogs of haloperidol and venlafaxine, positive charges close to +3 e were found at the Si atoms [20]. The strongly positive charge at the sulfur atom is compensated by the negative charges of the direct neighbor atoms. Their charge sum is −3.58/−3.82/−3.79 e for (I)/(II)/(III), so that an excess charge of more than −0.5 e is found in the direct environment of the sulfur atom.

The pyrazole and the ethoxy oxygen atoms both have charges close to −0.9 e. Consequently, the adjacent pyrimidine carbon atom has a positive charge of≈+1 e, whereas the charge at the ethoxy oxygen atom is balanced by the charge sum of the two adjacent carbon atoms being 0.66/0.68/0.69 e for (I)/(II)/(III).

All nitrogen atoms carry a significant negative charge, which also holds for the protonated nitrogen atom of the piperazine ring of (III). Its difference to the corresponding nitrogen atoms in (I) and (II) is visible in the atomic volumes, which is by 2.8 Å3 smaller for (III) compared to (I) and (II). A small difference with respect to the charges can be seen if the group charge of all atoms of the piperazine ring is considered. It is negative, −0.53 and −0.40 e, for (I) and (II), but slightly positive at +0.19 e for (III). As observed earlier [21], [22], [23], protonation causes a delocalization of charge over a wide range in the environment of the atom in question.

In the pyrazole ring, where the methylation is different between (I) and (II)/(III), only marginal differences in bond-topological and atomic properties are seen for the endocyclic carbon atoms (Fig. 4). On the one hand, the methylated nitrogen atoms have comparable atomic properties for (I) and (II)/(III), which also holds on the other hand for the non-methylated ones. There is thus a considerable difference in atomic properties between the methylated and the non-methylated nitrogen atoms. We note a volume difference of more than 5 Å3, accompanied by a difference of the negative charges of about 0.2 e.

Fig. 4: ρ(rBCP) values in e Å−3 and atomic properties (charges in e, volumes in Å3, both in red rectangular frames) for the pyrazole ring in (I) (left) and in (II)/(III) (right).
Fig. 4:

ρ(rBCP) values in e Å−3 and atomic properties (charges in e, volumes in Å3, both in red rectangular frames) for the pyrazole ring in (I) (left) and in (II)/(III) (right).

Intra- and intermolecular contacts in terms of potential HBs are summarized in Table 5. It was tried to determine all BCPs on all possibly attractive interactions listed in Table 5. It turned out that a strong intramolecular HB between the pyrazole N–H and the ethoxy oxygen exists with a properly identified BCP for both sildenafil structures (II) and (III). This is not the case for iso-sildenafil (I) in the solid state. Although the geometric situation indicates close N…O and H…O contacts typically found for HBs, no critical point is found at this site, so that from the QTAIM analysis no intramolecular HB exists for (I). However, this situation differs for the structure obtained from a gas-phase geometry optimization with the functional/basis-set combination B3LYP/6-31G(d,p), where coplanar rings and an intramolecular HB are indeed found.

Table 5:

Summary of hydrogen bonding topologies for (I), (II) and (III).

TypeD–H···AD···A, ÅH···A, Åρ(rBCP), e Å−32ρ(rBCP), e Å−5
(I) intraN6–H6···O12a2.768(3)2.41(3)
(I) interN6–H6···O7b2.861(3)1.79(3)0.183.03
(I) interC14–H14···O1c3.323(3)2.37(3)0.060.94
(II) intraN2–H1···O1a2.644(3)1.96(3)0.152.85
(II) interC8–H2···O2d3.225(3)2.19(3)0.091.33
(III) intraN4–H2AA···O5a2.634(2)1.91(2)0.162.99
(III) sil-sacchN1–H1AA···N7a2.739(2)1.79(2)0.243.25
(III) sil-sacchC3–H3B···O1e3.297(2)2.35(2)0.070.96
(III) sil-sacchC15–H15···O3f3.294(2)2.37(2)0.060.93
(III) acet-silC30–H30A···O4g3.249(2)2.32(2)0.081.11
(III) acet-sacchC30–H30B···O3a3.202(2)2.27(2)0.091.28

aSymmetry code: x, y, z.

bSymmetry code: −x, 2−y, −z.

cSymmetry code: 1/2−x, 1/2+y, 1/2−z.

dSymmetry code: −x, 1/2+y, −1/2−z.

eSymmetry code: 1−x, 1−y, 1−z.

fSymmetry code: 1+x, y, z.

gSymmetry code: 1−x, 1−y, 1−z.

Obviously the above-mentioned non-co-planarity along the torsion τ7 is unfavorable for the formation of an intramolecular HB in the solid state. Instead, a strong intermolecular HB was found for (I) generating a dimeric structure via a crystallographic inversion center. For (II) only a weak C–H…O intermolecular HB exists, so that solid-state packing of (I) and (II) are completely different.

Figure 5 illustrates this hydrogen bonding situation for (I) and (II). In addition, Fig. 5 also shows the corresponding Hirshfeld surfaces. Molecular surfaces and the corresponding surface properties play an important role in drug–receptor recognition processes, because the mutual recognition takes place via surface interactions in a first step. Hirshfeld surfaces are defined by the ratio of the molecular ED to the crystal density [25], [26] when equal to 0.5. When the aspherical ED is mapped by a color code onto this surface, ED concentrations are emphasized so that sites and strengths of intermolecular interactions become visible (see Figs. 5 and 6) [27]. For (I), two strong deeply colored signals are seen at the donor and acceptor sites of the intermolecular HB, and no further significant signals exist. For (II), the rather weak HB C(8)–H(2)···O(2) (Table 5) is illustrated by corresponding weak signals on the Hirshfeld surfaces. There are a few further weak ED concentrations visible on this surface of (II). They are all based on weak intermolecular approaches in the crystal structure, so the Hirshfeld surface is a fine instrument to make intermolecular contacts visible.

Fig. 5: Dimeric and non-dimeric structures for (I) and (II), Schakal representations (left) [24] and corresponding Hirshfeld surfaces (right) [25], [26], drawn with Moliso [27].
Fig. 5:

Dimeric and non-dimeric structures for (I) and (II), Schakal representations (left) [24] and corresponding Hirshfeld surfaces (right) [25], [26], drawn with Moliso [27].

Fig. 6: Hydrogen bond contacts in the cation–anion complex (III) (left) [15], [16] and the corresponding illustration on the Hirshfeld surface of the cation (right) [27].
Fig. 6:

Hydrogen bond contacts in the cation–anion complex (III) (left) [15], [16] and the corresponding illustration on the Hirshfeld surface of the cation (right) [27].

In case of (III), no cation···cation HB exists, but a quite strong N–H···N HB links the cation and the saccharinate anion. In addition, some further weak C–H···O contacts are listed in Table 5. They are all displayed as ED concentrations on the corresponding Hirshfeld surface in Fig. 6.

For the understanding of molecular polarization, reactivity behavior and the intermolecular substrate/acceptor recognition, the electrostatic potential (ESP) plays a helpful role. That is why it was calculated by using the method of Volkov et al. [28] with the XDprop subprogram of XD2006 [14] and color coded onto the 0.0067 e Å−3 (=0.001 a.u.) electron-density isosurface (see Figs. 79) [27].

Fig. 7: Electrostatic potentials of (I) and (II) mapped onto the electron-density isosurface ρ=0.0067 e Å−3; the representation was generated with Moliso [27].
Fig. 7:

Electrostatic potentials of (I) and (II) mapped onto the electron-density isosurface ρ=0.0067 e Å−3; the representation was generated with Moliso [27].

Fig. 8: Electrostatic potential of the iso-sildenafil dimer of (I) [27]. For further details see the legend of Fig. 7.
Fig. 8:

Electrostatic potential of the iso-sildenafil dimer of (I) [27]. For further details see the legend of Fig. 7.

Fig. 9: Electrostatic potential of the sildenafil saccharinate acetonitrile complex (III). Left: representation of the cation only. Right: complete complex, consisting of the cation, the anion and the acetonitrile solvent molecule [27].
Fig. 9:

Electrostatic potential of the sildenafil saccharinate acetonitrile complex (III). Left: representation of the cation only. Right: complete complex, consisting of the cation, the anion and the acetonitrile solvent molecule [27].

The ESP surfaces of (I) and (II) show considerable differences in the regions where the methylation is different (Fig. 7). At the N–H unit, which is the donor of the intermolecular HB, a strongly positive potential exists in (I), accompanied by negative ESP on the accepting oxygen. Both pronounced potentials are not seen in (II), where the intermolecular HB does not exist; see the discussion above. Figure 8 shows the ESP surface of the iso-sildenafil dimer generated by the intermolecular HB. We note that the dimer is enveloped by an almost positive ESP shell. It can be speculated that this surface property reduces the interaction ability so that the iso form is pharmaceutically less active.

The ESP of the sildenafil saccharinate acetonitrile complex (III) is displayed in Fig. 9 in representations of the cation only (Fig. 9, left) and the complete complex (Fig. 9, right). The color bar scale (Fig. 9, left) shows that in the cation the ESP surface is completely positive with the strongest positive region above the hydrogen of the protonated nitrogen in the piperazine ring facing the saccharinate anion. In the representation of the complete complex (Fig. 9, right), the charge separation and hence the polarization between cation and anion is nicely visible with positive and negative surfaces, whereas the neutral acetonitrile solvent has an ESP surface close to zero.

4 Conclusions

A number of differences in structure and electron density distribution have been identified in this comparative study between iso-sildenafil and sildenafil.

The molecular structures differ not only in the region where they are chemically diverse. There is also a significant conformational difference in that for both the neutral and the cationic form of sildenafil the phenyl ring and the bicyclic pyrazolopyrimidine system are practically coplanar, which allows an intramolecular N–H···O HB. This HB does not exist in the crystal structure of iso-sildenafil, where the two ring systems in question are inclined by a torsion angle of 44.2(3)°. Accordingly, no intramolecular BCP could be located for the atom sequence N–H···O. Instead, a strong intermolecular HB is formed to establish a dimer via a crystallographic inversion center. All intermolecular interactions are clearly visible on the corresponding Hirshfeld surfaces.

Differences in the ESPs were also found. They are related to the different methylation, and to the above-mentioned differences in hydrogen bonding. The cationic–anionic structure of the sildenafil saccharinate acetonitrile complex is reflected in the ESP. The ESP of the iso-sildenafil dimer is dominated by a shell of positive potential around it.

In all 24 sildenafil small-molecule structures determined by X-ray diffraction known until today, all structures but one exist in the planar form. Obviously this planar form is energetically favored for the small-molecule structures of sildenafil, enabling the intramolecular N–H···O HB between the pyrazolopyrimidine donor NH and the oxygen acceptor of the ethoxy group. This planarity was even mentioned in the literature as a requirement for optimal bioactivity [1]. However, the interaction with the receptor is probably quite different. In the PDE5 complex [9], the sildenafil molecule in the active site pocket of the receptor exists in the inclined form; the τ7 torsion angle is 49.2°. Instead of an intramolecular HB, the amide moiety of the pyrazolo fragment is reported to be involved in a bidentate hydrogen bonding interaction with a Gln residue of the receptor [9]. Surprisingly, this HB pattern is similar to the one of the inclined form of iso-sildenafil in its crystal structure, although the intramolecular HB is again present in the gas-phase structure of iso-sildenafil after geometry optimization.

Although for the small-molecule structures of sildenafil and iso-sildenafil differences concerning structural and electronic properties exist which could explain the different pharmacological properties, findings are not unambiguous (see above). In contrast to examples discussed earlier [23], [29], [30], [31], the properties of the small molecules in the sildenafil/iso-sildenafil structures are obviously not suited to mimic the biologically relevant drug–receptor interactions.

Nevertheless we do believe that the invariom formalism can routinely deliver valuable information for biologically interacting systems with moderate effort, offering new options for drug research in addition to considering steric interactions only. The spatial resolution provided by small-molecule structures is complementary to that achievable in macromolecular structure determination, and the accurate electron density model can lead to further insight, provided that the small-molecule contacts and conformation in the crystal are similar to the situation in the respective drug–receptor interaction.

Acknowledgments

The authors are grateful to the Deutsche Forschungsgemeinschaft for financial support (project DI 921/6-1). They also thank Dr. C. Maichle-Mössmer (Tübingen, Germany) and Dr. R. Banerjee (Hyderabad, India) for having e-mailed .res and .hkl files of (I) and (III) to us. Corresponding data for (II) were obtained from the CSD.

References

[1] M. M. El-Abadelah, S. S. Sabri, M. A. Khanfar, W. Voelter, C. Maichle-Mössmer, Z. Naturforsch. 1999, 54b, 1323.10.1515/znb-1999-1016Search in Google Scholar

[2] H. S. Yathirajan, B. Nagaraj, P. Nagaraja, M. Bolte, Acta Crystallogr. 2005, E61, o489.Search in Google Scholar

[3] R. Banerjee, P. M. Bhatt, G. R. Desiraju, Cryst. Growth Des.2006, 6, 1468.10.1021/cg0601150Search in Google Scholar

[4] D. Stepanovs, A. Mishnev, Z. Naturforsch. 2012, 67b, 491.10.5560/znb.2012-0073Search in Google Scholar

[5] P. Sanphui, S. Tothadi, S. Gangulu, G. R. Desiraju, Mol. Pharmaceutics2013, 10, 4687.10.1021/mp400516bSearch in Google Scholar PubMed

[6] D. Stepanovs, M. Jure, A. Mishnev, Mendeleev Commun.2015, 25, 49.10.1016/j.mencom.2015.01.018Search in Google Scholar

[7] A. Abraham, D. C. Apperley, S. J. Byard, A. J. Ilott, A. J. Robbins, V. Zorin, R. K. Harris, P. Hodgkinson, CrystEngComm.2016, 18, 1054.10.1039/C5CE02234GSearch in Google Scholar

[8] F. H. Allen, Acta Crystallogr. 2002, B58, 380.10.1107/S0108768102003890Search in Google Scholar PubMed

[9] B.-J. Sung, K. Y. Hwang, Y. H. Jeon, J. I. Lee, Y.-S. Heo, J. H. Kim, J. Moon, J. M. Yoon, Y.-L. Hyun, E. Kim, S. J. Eum, S.-Y. Park, J.-O. Lee, T. G. Lee, S. Ro, J. M. Cho, Nature2003, 425, 98.10.1038/nature01914Search in Google Scholar PubMed

[10] B. Dittrich, T. Koritsanszky, P. Luger, Angew. Chem. Int. Ed.2004, 43, 2718.10.1002/anie.200353596Search in Google Scholar PubMed

[11] B. Dittrich, C. B. Hübschle, K. Pröpper, F. Dietrich, T. Stolper, J. J. Holstein, Acta Crystallogr. 2013, B69, 91.10.1107/S2052519213002285Search in Google Scholar PubMed

[12] C. B. Hübschle, P. Luger, B. Dittrich, J. Appl. Crystallogr. 2007, 40, 623.10.1107/S0021889807016524Search in Google Scholar

[13] N. K. Hansen, P. Coppens, Acta Crystallogr. 1978, A34, 909.10.1107/S0567739478001886Search in Google Scholar

[14] A. Volkov, P. Macchi, L. J. Farrugia, C. Gatti, P. R. Mallinson, T. Richter, T. Koritsánszky, XD2006 – A Computer Program for Multipole Refinement, Topological Analysis of Charge Densities and Evaluation of Intermolecular Energies from Experimental and Theoretical Structure Factors, University of Buffalo, NY (USA), University of Milano (Italy), University of Glasgow (UK), CNRISTM, Milano (Italy), Middle Tennessee State University, TN (USA) 2006.Search in Google Scholar

[15] C. K. Johnson, Ortep, Oak Ridge Thermal Ellipsoid Plot Program for Crystal Structure Illustrations, Rep. ORNL-3794, Oak Ridge National Laboratory, Oak Ridge, TN (USA) 1965.Search in Google Scholar

[16] A. L. Spek, Acta Crystallogr. 2009, D65, 148.Search in Google Scholar

[17] R. F. W. Bader, Atoms in Molecules – A Quantum Theory, Clarendon Press, Oxford, 1994.Search in Google Scholar

[18] P. Luger, M. Messerschmidt, S. Scheins, A. Wagner, Acta Crystallogr. 2004, A60, 390.10.1107/S0108767304014825Search in Google Scholar

[19] A. Volkov, C. Gatti, Y. Abramov, P. Coppens, Acta Crystallogr. 2000, A56, 25210.1107/S0108767300001628Search in Google Scholar

[20] P. Luger, B. Dittrich, R. Tacke, Org. Biomol. Chem.2015, 13, 9093.10.1039/C5OB00728CSearch in Google Scholar

[21] S. Mebs, S. Grabowsky, D. Förster, J. Phys. Chem.2010, A114, 10185.10.1021/jp100995nSearch in Google Scholar

[22] S. Scheins, M. Messerschmidt, W. Morgenroth, C. Paulmann, P. Luger, J. Phys. Chem.2007, A111, 5499.10.1021/jp0709252Search in Google Scholar

[23] P. Luger, M. Weber, B. Dittrich, Future Med. Chem. 2012, 4, 1399.10.4155/fmc.12.85Search in Google Scholar

[24] E. Keller, J. S. Pierrard, Schakal99, A Computer Program for the Graphical Representation of Molecular and Crystallographic Models, University of Freiburg, Freiburg (Germany) 1999.Search in Google Scholar

[25] J. J. McKinnon, A. S. Mitchell, M. A. Spackman, Chem. Eur. J.1998, 4, 2136.10.1002/(SICI)1521-3765(19981102)4:11<2136::AID-CHEM2136>3.0.CO;2-GSearch in Google Scholar

[26] M. A. Spackman, J. J. McKinnon, D. Jayatilaka, CrystEngComm.2008, 10, 377.Search in Google Scholar

[27] C. B. Hübschle, P. Luger. J. Appl. Crystallogr.2006, 39, 901.10.1107/S0021889806041859Search in Google Scholar

[28] A. Volkov, H. F. King, P. Coppens, L. J. Farrugia, Acta Crystallogr. 2006, A62, 400.10.1107/S0108767306026298Search in Google Scholar PubMed

[29] M. W. Shi, A. N. Sobolev, T. Schirmeister, B. Engels, T. C. Schmidt, P. Luger, S. Mebs, B. Dittrich, Y.-S. Chen, J. M. Bak, D. Jayatilaka, C. S. Bond, M. J. Turner, S. G. Stewart, M. A. Spackman, S. Grabowsky, New J. Chem. 2015, 39, 1628.10.1039/C4NJ01503GSearch in Google Scholar

[30] G. Klebe, Structure Correlation and Ligand/Receptor Interactions, in Structure Correlation, Vol. 2 (Eds.: H.-B. Bürgi, J. D. Dunitz), Wiley-VCH, Weinheim, 1994, chapter 13, pp. 543–603.10.1002/9783527616091.ch13Search in Google Scholar

[31] M. Mladenovic, M. Arnone, R. F. Fink, B. Engels, J. Phys. Chem.2009, B113, 5072.10.1021/jp809537vSearch in Google Scholar PubMed

Received: 2016-8-17
Accepted: 2016-9-8
Published Online: 2016-12-8
Published in Print: 2017-1-1

©2017 Walter de Gruyter GmbH, Berlin/Boston

Downloaded on 24.4.2024 from https://www.degruyter.com/document/doi/10.1515/znb-2016-0178/html
Scroll to top button