Solid-state tautomeric structure and invariom refinement of a novel and potent HIV integrase inhibitor

Abstract

The conformation and tautomeric structure of (Z)-4-[5-(2,6-difluoro­benzyl)-1-(2-fluoro­benzyl)-2-oxo-1,2-dihydro­pyridin-3-yl]-4-hy­droxy-2-oxo-N-(2-oxopyrrolidin-1-yl)but-3-enamide, C₂₇H₂₂F₃N₃O₅, in the solid state has been resolved by single-crystal X-ray crystallography. The electron distribution in the mol­ecule was evaluated by refinements with invarioms, aspherical scattering factors by the method of Dittrich et al. [Acta Cryst. (2005), A61, 314–320] that are based on the Hansen–Coppens multipole model [Hansen & Coppens (1978 ▶). Acta Cryst. A34, 909–921]. The β-diketo portion of the mol­ecule exists in the enol form. The enol –OH hydrogen forms a strong asymmetric hydrogen bond with the carbonyl O atom on the β-C atom of the chain. Weak intra­molecular hydrogen bonds exist between the weakly acidic α-CH hydrogen of the keto–enol group and the pyridinone carbonyl O atom, and also between the hydrazine N—H group and the carbonyl group in the β-position from the hydrazine N—H group. The electrostatic properties of the mol­ecule were derived from the mol­ecular charge density. The mol­ecule is in a lengthened conformation and the rings of the two benzyl groups are nearly orthogonal. Results from a high-field ¹H and ¹³C NMR correlation spectroscopy study confirm that the same tautomer exists in solution as in the solid state.

Comment  

The retroviral enzyme HIV-1 integrase is essential for HIV replication and is a significant target for the discovery and development of anti-HIV therapeutic agents (Moir et al., 2011 ▶; Frankel & Young, 1998 ▶; Nair & Chi, 2007 ▶; Pommier et al., 2005 ▶). Research efforts on anti-HIV integrase inhibitors for the treatment of acquired immunodeficiency syndrome (AIDS) have resulted in several anti-HIV agents, two of which, raltegravir and elvitegravir, have been approved by the US Food and Drug Administration for the clinical treatment of HIV–AIDS (Nair et al., 2006 ▶; Summa et al., 2008 ▶; Min et al., 2010 ▶; Shimura & Kodama, 2009 ▶; Taktakishvili et al., 2000 ▶). The crystallographic structures of some HIV integrase inhibitors have been reported [see, for example, Rhodes et al. (2011 ▶) and Newton et al. (2005 ▶)]. As resistance, toxicity and drug–drug inter­actions are recurring issues with all classes of anti-HIV drugs, the discovery of new anti-HIV active integrase inhibitors remains a significant scientific challenge. The compound (Z)-4-[5-(2,6-difluoro­benzyl)-1-(2-fluoro­benzyl)-2-oxo-1,2-dihydro­pyridin-3-yl]-4-hy­droxy-2-oxo-N-(2-oxopyrrolidin-1-yl)but-3-enamide, (1), discovered in our laboratory, is an integrase inhibitor which possesses potent (low nM) anti-HIV activity against a diverse set of HIV-1 isolates and also against HIV-2 and SIV. However, this compound can exist in three possible forms (I, II and III) with respect to the β-diketo functionality (see Scheme 1). In order to determine which tautomeric form is dominant in the solid state, the single-crystal X-ray structure of integrase inhibitor (1) was undertaken. An invariom refinement (Dittrich et al., 2005 ▶; Hansen & Coppens, 1978 ▶) was performed to examine the electrostatic properties of the mol­ecule in further detail.

The mol­ecular structure of (1) contains a variety of distinct groups, including a central pyridinone ring, fluoro­benzene rings, a keto–enol group and a 2-oxopyrrolidin-1-yl group (Fig. 1 ▶ and Table 1 ▶). The scattering factors of some fragments were not yet present in the invariom database (Dittrich et al., 2006 ▶) and so were calculated here using quantum mechanics. This method of refinement led to C—H distances somewhat longer [1.013 (15)–1.107 (12) Å] than ordinarily expected from an X-ray determination but closer to distances determined from neutron diffraction (Allen et al., 2006 ▶).

Figure 1.

The mol­ecular structure of the dominant tautomeric form of (1) present in the crystalline state. Displacement ellipsoids are drawn at the 50% probability level. Dashed lines indicate intra­molecular hydrogen bonds.

Table 1. Selected geometric parameters (Å, °).

F36—C24	1.3484 (12)
F37—C31	1.3467 (13)
F38—C35	1.3444 (13)
O8—C7	1.2102 (11)
O9—C5	1.2178 (12)
O11—C10	1.2678 (11)
O14—C13	1.3033 (12)
O21—C16	1.2245 (11)
N1—N6	1.3802 (11)
N1—C2	1.4605 (13)
N1—C5	1.3608 (12)
N6—C7	1.3531 (13)
C7—C10	1.5271 (14)
C10—C12	1.3986 (14)
C12—C13	1.3979 (14)
C12—H12	1.038 (12)
C13—C15	1.4682 (14)
C15—C16	1.4553 (13)
C15—C20	1.3784 (13)
C18—C19	1.3676 (14)
C19—C20	1.4078 (14)

O11—C10—C7	117.89 (9)
O11—C10—C12	125.07 (9)
C7—C10—C12	117.02 (9)
C10—C12—C13	119.72 (9)
C10—C12—H12	120.1 (7)
C13—C12—H12	120.1 (7)
O14—C13—C12	120.11 (9)
O14—C13—C15	116.09 (9)
C12—C13—C15	123.79 (9)

The refined position of the H atom in the strong intra­molecular O—H⋯O hydrogen bond indicated that it is not symmetrically located between the two O-atom centers but rather favors atom O14 over O11, i.e. form I is the dominant form in the solid. This is reflected by the longer C—O bond for O14 [1.3033 (12) Å, compared with 1.2678 (11) Å for O11]. However, this is not a completely localized H atom and there is some residual disorder. This was seen in the residual electron-density map about atom O11, with a maximum peak height of 0.24 e Å⁻³ between atoms O11 and H14 (Fig. 2 ▶). A deformation electron-density map (with contour step values of ±0.05 e Å⁻³) is shown in Fig. 3 ▶. The hydrogen bond is not linear; the O14—H14⋯O11 angle is 155.5 (16)° (Table 2 ▶). The H atom on atom N6 donates an inter­molecular hydrogen bond to atom O9ⁱ (N6—H6⋯O9ⁱ; Table 2 ▶), forming a centrosymmetric dimer. The slightly acidic α-C—H hydrogen of the keto–enol group donates a weak intra­molecular hydrogen bond to carbonyl atom O21 (C12—H12⋯O21; Table 2 ▶).

Figure 2.

A residual electron-density map in the plane of the atoms of the keto–enol group. Contours are drawn at 0.05 e Å⁻³ inter­vals. Solid lines represent positive contours and dashed lines negative contours.

Figure 3.

A deformation electron-density map in the plane of the atoms of the keto–enol group. Contours are drawn at 0.05 e Å⁻³ inter­vals. Solid lines represent positive contours and dashed lines negative contours.

Table 2. Hydrogen-bond geometry (Å, °).

D—H⋯A	D—H	H⋯A	D⋯A	D—H⋯A
N6—H6⋯O9i	1.000 (14)	1.924 (14)	2.8014 (12)	144.8 (10)
O14—H14⋯O11	1.092 (18)	1.492 (18)	2.5270 (11)	155.5 (16)
C12—H12⋯O21	1.038 (13)	2.135 (12)	2.8346 (13)	122.7 (9)

Symmetry code: (i) .

Since a complete static electron-density distribution is available from the invariom model scattering factors, various properties like the dipole moment and electrostatic potential [V(r)] can be derived from the electron density. They were calculated using the program XDPROP in XD2006 (Volkov et al., 2006 ▶). The results could provide information on the capacity of this mol­ecule to inter­act with a protein-binding site. The dipole moment (p) for the mol­ecule calculated from the multipole populations is 12.06 D. The electrostatic field is the force that a hypothetical proton would be subjected to if it were present. The electrostatic potential (a scalar quantity) at a given point can be defined as the amount of work that is needed to bring a unit of charge from infinity to that point. A composite of the positive (0.2 e Å⁻³) and negative (−0.06 e Å⁻³) electrostatic potential isosurfaces plotted with the program Molekel (Molekel, 2009 ▶) is shown in Fig. 4 ▶. Regions of strong positive potential are shown in lighter grey and negative potential in dark grey. Atom H14, involved in a strong intra­molecular hydrogen bond, shows a strong positive potential, while adjacent atom O11 shows a strong negative potential.

Figure 4.

Positive (0.2 e Å⁻³) and negative (−0.06 e Å⁻³) electrostatic potential isosurfaces of (1) shown in light and dark grey, respectively.

The results of a high-field ¹H and ¹³C NMR correlation spectroscopy study (COSY, HSQC, HMQC, HMBC and NOESY) are consistent with the structure observed in the solid state.

Experimental  

The integrase inhibitor was prepared from the coupling of the corresponding diketo acid (Seo et al., 2011 ▶) and 1-amino-2-pyrrolidin­one p-toluene­sulfonate. Compound (1) crystallized from dichloro­methane as yellow prisms (yield 78%; m.p. 448–449 K). UV (CH₃OH, λ, nm): 401 (∊ 9, 139), 318 (∊ 6, 225). HRMS calculated for C₂₇H₂₂F₃N₃O₅: [M + H]⁺ 526.1590; found: 526.1589.

Crystal data  

• C₂₇H₂₂F₃N₃O₅

• M _r = 525.48

• Triclinic,

• a = 8.8182 (10) Å

• b = 11.5986 (13) Å

• c = 12.2741 (14) Å

• α = 98.594 (2)°

• β = 90.904 (2)°

• γ = 106.435 (2)°

• V = 1188.3 (2) ų

• Z = 2

• Mo Kα radiation

• μ = 0.12 mm⁻¹

• T = 173 K

• 0.75 × 0.65 × 0.35 mm

Data collection  

• Bruker APEXII area-detector diffractometer

• Absorption correction: empirical (using intensity measurements) (SADABS; Sheldrick, 2008b ▶) T _min = 0.841, T _max = 1.000

• 22962 measured reflections

• 5447 independent reflections

• 5175 reflections with I > 3σ(I)

• R _int = 0.029

Refinement  

• R[F ² > 2σ(F ²)] = 0.035

• wR(F ²) = 0.067

• S = 2.00

• 5175 reflections

• 431 parameters

• All H-atom parameters refined

• Δρ_max = 0.24 e Å⁻³

• Δρ_min = −0.19 e Å⁻³

The program InvariomTool (Hübschle et al., 2007 ▶) was used to prepare master and input files for an invariom refinement with the XDLSM program of the XD2006 suite (Volkov et al., 2006 ▶). The program assigns invarioms to all atoms in a given crystal structure by examining the connectivity in terms of nearest or next-nearest neighbors. Nonspherical valence scattering contributions for atoms in an environment of single bonds were obtained from theoretical calculations on model compounds that included nearest-neighbour atoms, whereas for H atoms and atoms in a delocalized chemical environment the model compounds also included the next-nearest neighbor atoms. Several fragments were not present in the invariom database. Therefore, new scattering factors for these fragments, including the keto–enol group, were calculated from geometry optimizations at the B3LYP/D95++(3df,3pd) level of theory and included in the invariom database to increase its coverage of chemical environments. These calculations were performed using GAUSSIAN09 (Frisch et al., 2009 ▶). Full-matrix least-squares refinements on F ² using complete multipole expansions were carried out with the program XDLSM using statistical weights. Only reflections with intensities I > 3σ(I) were included in the refinement. Initially, bond lengths involving H atoms were set to the X—H distances obtained from model compounds that included the next-nearest neighbors of the H atom of inter­est. However, in the final cycles, these atom positions were refined freely. Positional and displacement (anisotropic for non-H atoms) parameters, but not multipoles, were refined. However, a hexa­deca­polar level of the multipole expansion was used for all atoms. A mol­ecular electroneutrality constraint was applied. The introduction of invarioms improved R(F) from 0.0511 to 0.0349 while using the same weighting scheme {w = 1/[σ²(F _o ²)]} as the spherical-atom refinement, and improved the goodness-of-fit value from 2.832 to 2.001.

Data collection: APEX2 (Bruker, 2011 ▶); cell refinement: SAINT (Bruker, 2009 ▶); data reduction: SAINT; program(s) used to solve structure: SHELXS97 (Sheldrick, 2008a ▶); program(s) used to refine structure: XD2006 (Volkov et al., 2006 ▶); molecular graphics: XD2006, Molekel (Molekel, 2009 ▶) and OLEX2 (Dolomanov et al., 2009 ▶); software used to prepare material for publication: XD2006.