INTRODUCTION
The 3,4-diaminopyridine (3,4-DAP) molecule, whose crystal structure was recently established (Rubin-Preminger and Englert, Reference Rubin-Preminger and Englert2007), can be used pharmaceutically (amifampridine) for the treatment of various diseases including the Lambert-Eaton myasthenic syndrome (LEMS) (McEvoy et al., Reference McEvoy, Windebank, Daube and Low1989). Tartrate and phosphate salts of 3,4-DAP were shown to have very high stability in storage conditions complying with international recommendations contrarily to 3,4-DAP (Guyon et al., Reference Guyon, Pradeau, Le Hoang and Houri2002). The 3,4-DAP phosphate (3,4-DAPP) formulation has obtained the orphan medicinal product status both in the European Union and in the Unites States (Quartel et al., Reference Quartel, Turbeville and Lounsbury2010).
The aim of the present work is to provide an answer for the 3,4-DAPP to the usual questions for pharmaceutical compounds: Is a single phase structure established by crystallographic investigation or are there polymorphs in the sample produced? Is there formation of a salt or cocrystallization of the API (active pharmaceutical ingredient) with H3PO4? In spite of the absence of a suitable single crystal, solving the structure of 3,4-DAPP is realized from powder diffraction data by using methodologies which continuously have demonstrated their efficiency during repeated blind tests (Le Bail et al., Reference Le Bail, Cranswick, Adil, Altomare, Avdeev, Cerny, Cuocci, Giacovazzo, Halasz, Lapidus, Louwen, Moliterni, Palatinus, Rizzi, Schilder, Stephens, Stone and van Mechelen2009).
STRUCTURE DETERMINATION, RIETVELD REFINEMENTS, AND DFT STRUCTURE OPTIMIZATION
The powder diffraction pattern, recorded from a sample as received from the SERATEC Company (www.serateclab.com), could be indexed in a monoclinic cell by the MCMAILLE software (Le Bail, Reference Le Bail2004). The FoMs (figures of merit) for the pattern selected for this study were found to be M 20 = 24.7 (de Wolf, Reference de Wolf1968), F 20 = 55.4 (0.0049, 73) (Smith and Snyder, Reference Smith and Snyder1979), McM 20 = 362.0 (Le Bail, Reference Le Bail, Clearfield, Reibenspies and Bhuvanesh2008), and Z = 8. Using Le Bail fitting (Le Bail, Reference Le Bail2005) for evaluating the reliability of the indexing and for intensities extraction, the I2/c space group was determined unambiguously. The structure solution was done in the direct space using the ESPOIR software (Le Bail, Reference Le Bail2001) moving in the unit cell, by a Monte Carlo process, a 3,4-diaminopyridine molecule and a PO4 tetrahedron as rigid bodies (for a total of 12 degrees of freedom). This led to R = 14% on the first 100 “|F obs|” extracted at low diffracting angles (<42°2θ). Refinements were undertaken by the Rietveld (Reference Rietveld1969) FULLPROF software (Rodriguez-Carvajal, Reference Rodriguez-Carvajal1993). The hydrogen atom positions were suggested by the SHELXL software (Sheldrick, Reference Sheldrick2008) in the hypothesis of cocrystallization. By using soft restraints on N-C and C-C interatomic distances, the 3,4-diaminopyridine molecule was maintained in the shape similar to that established by Rubin-Preminger and Englert (Reference Rubin-Preminger and Englert2007). The O and P atoms of the H3PO4 group were refined freely, but all H atoms were fixed at their positions suggested by the procedure implemented in the SHELXL program.
Refined atomic coordinates were further optimized by energy minimization in the solid state using the VASP program (Kresse and Hafner, Reference Kresse and Hafner1993; Kresse and Furthmüller, Reference Kresse and Furthmüller1996a, Reference Kresse and Furthmüller1996b). The electron exchange-correlation interaction was described in the generalized gradient approximation (Perdew and Wang, Reference Perdew and Wang1992). Plane waves formed the basis set, and calculations were performed using the projector-augmented wave method (Blöchl, Reference Blöchl1994; Kresse and Joubert, Reference Kresse and Joubert1999) and atomic pseudopotentials
(Kresse and Hafner, Reference Kresse and Hafner1994). The energy cut-off controlling the accuracy of the calculation was set to 500 eV, representing an extended basis set and consequently highly accurate calculations. The positions of all atoms were optimized by means of the conjugated gradient method in the Γ-point of the Brillouin zone with the unit-cell parameters fixed. In the course of the calculation, one of the hydrogen atoms from the H3PO4 group jumped to the molecule ring N atom forming
pyridinium cation. This new conformation was then confirmed by the independent calculations starting with a different initial geometry. The final formula of the title compound hence is [C5H3(NH)(NH2)2]+ (H2PO4)−. The same behavior conducting to a 3,4-pyridinium salt is observed when the phosphoric acid is replaced by succinic (Fun and Balasubramini, Reference Fun and Balasubramani2009), squaric (Koleva et al., Reference Koleva, Tsanev, Kolev, Mayer-Figge and Sheldrick2007), or tartaric (Koleva et al., Reference Koleva, Kolev, Tsanev, Kotov, Mayer-Figge, Seidel and Sheldrick2008) acids.
The basic crystallographic data, experimental, and Rietveld-refinement details are summarized in Table I. The Rietveld fit, atoms numbering, and crystal packing are depicted in Figures 1–3, respectively. The atomic coordinates obtained by both methods are listed in Table II, and the essential bond distances, bond angles are detailed in Table III. Hydrogen bonds geometry is summarized in Table IV, and X-ray powder diffraction data are given in Table V. Both sets of atomic coordinates were deposited (CIF) at the Crystallography Open Database (www.crystallography.net) (Gražulis et al., Reference Gražulis, Chateigner, Downs, Yokochi, Quirós, Lutterotti, Manakova, Butkus, Moeck and Le Bail2009) and in the supplementary date.
DISCUSSION
According to the optimized atomic coordinates, the molecule of 3,4-DAP+ is essentially planar with the maximum
deviation from the least-squares plane fit to all nonhydrogen atoms being only 0.039 Å (due to N3). The optimized bond distances and bond angles are in good agreement with the values derived from the single crystal data (Rubin-Preminger and Englert, Reference Rubin-Preminger and Englert2007), though it is N2 which is significantly displaced by 0.131 Å from the plane in the 3,4-DAP. The phosphate group in 3,4-DAPP is a slightly deformed tetrahedra with a minimum spread of the O-P-O bond angles
(Table III). As expected, the P-O bond distances fall into two groups, and the protonated P-O vertices show the expected lengthening of the P=O bonds (Blessing, Reference Blessing1988; Mahmoudkhani and Langer, Reference Mahmoudkhani and Langer2002; Demir et al., Reference Demir, Yilmaz and Harrison2003; Smrčok et al., Reference Smrčok, Havlíček, Kaman and Rundlőf2009; Balamurugan et al., Reference Balamurugan, Jagan and Sivakumar2010; Kaman et al., Reference Kaman, Smrčok, Císařová and Havlíček2011; Marouani et al., Reference Marouani, Al-Deyab and Rzaigui2011). The structure of the 3,4-DAPP consists of dihydrogen phosphate tetrahedra linked by moderately strong (Jeffrey, Reference Jeffrey1997) O-H…O hydrogen bonds [Table IV and Figure 4(a)] involving all oxygen
a 1/2 − x, 1/2 − y, 1/2 − z.
b 1/2 − x, −1/2 − y, 1/2 − z.
c x, −y, 1/2 + z.
d −x, −y, 1 − z.
a −1/2 + x, 1/2 − y, z.
atoms to the chains running along the b-axis. Antiparallelly, π–π stacked 3,4-DAP+ cations form molecular columns in the spaces between the chains. The molecules are mutually slipped, but the perpendicular distances of the molecules are only ∼3.31 Å. Although the dominant interaction of the molecules with their surroundings is electrostatic, their bonding is further enhanced by the formation of N-H…O and C-H…O hydrogen bonds [Table IV and Figure 4(b)]. As a result, the architecture of hydrogen bonds comprise oxygen atoms, which are either acceptors or both donors and acceptors. Specifically, while the O1 and O2 atoms play only the role of acceptors of relatively strong O-H…O and N-H…O hydrogen bonds, the O3, except for playing a donor role in the phosphate chain, also receives a pair of a strong N-H and a weak C-H bonds. Finally, the donor O4 atom is also acceptor of a weak C-H…O bond. The N3…O1(O2) separations are comparable to the estimated mean values (∼2.85 Å) of the distribution of N…O contact distances extracted from CCSD database (Kumara Swamy et al., Reference Kumara Swamy, Kumaraswamy and Kommana2001). On the contrary, the hydrogen bonds involving the
N2 atom are remarkably longer and hence weaker. The lengths of two C-H…O are comparable, but the bond angle in C5-H3…O4 is close to linear, and the bond is hence probably somewhat stronger.
CONCLUSION
In the absence of a suitable single crystal, modern powder diffraction methodologies have the power sometimes to reveal crystal structures, including those of important pharmaceutical compounds. However, peak overlapping leads to unfavorable data/parameter ratios, and the crystal structures are often quite inaccurate without using restraints and constraints and even when using them. Moreover, not locating clearly H atoms may lead to difficulties to give the correct names to molecular compounds (in the cocrystallization hypothesis, the name would have been 3,4-diaminopyridine phosphoric acid). The presence of a single phase (no polymorph) in the sample studied is thus confirmed. Optimizing the atomic coordinates by quantum chemical calculations looks to be a complementary approach that will probably be soon generalized in such cases.
ACKNOWLEDGEMENT
One of the authors (L.S.) wishes to express his thanks to Slovak Grant Agency VEGA for a partial financial support of this study under the contract No. 2/150/09.