Received 2 July 2013
Crystal structure analysis and sublimation thermodynamics of bicyclo derivatives of a neuroprotector family
The crystal structures of three new structurally related drug-like bicyclo derivatives are correlated with measured thermodynamic quantities for their sublimation and melting processes. The sublimation thermodynamics are determined using the temperature dependencies of the vapour pressure, and the melting processes are examined using differential scanning calorimetry. The three compounds contain a common N-(3-thia-1-azabicyclo[3.3.1]non-2-ylidene)aniline core, with either a CH3, F or CF3 substituent at the 4-position of the aniline ring. Lattice energy calculations are made using both the PIXEL and Coulomb-London-Pauli (CLP) models, and the conformational flexibility of the molecules is examined using gas-phase density functional theory (DFT) calculations. The experimentally measured crystal lattice energies (H0sub) decrease in the order: CH3 > F > CF3. The calculated lattice energies using the PIXEL approach are in good agreement with the experimental values, and the partitioned intermolecular interaction energies suggest that dispersion contributions dominate the crystal structures of all three compounds. The sublimation energies and melting points are inversely correlated for the three molecules, with the melting points increasing in the order CF3 < F < CH3.
There are three main steps in rational drug design and development, namely the discovery of the lead compound, identification of the pharmacophore and optimization (Rekka & Kourounakis, 2008). The latter stage involves structural modification of the pharmacophore, guided by structure-activity relationships in order to find the best compound for a particular purpose. By changing the nature and position of substituent groups, it is possible to fine-tune the electronic and structural features of a potential drug and, as a consequence, its physicochemical properties and pharmacological activity. During the optimization phase, information about the molecular conformation of the drug becomes crucial for understanding its structure-activity relationships. This information is most clearly available from single-crystal X-ray structures, which provide detailed information on the molecular conformation in the solid state. In addition, knowledge of the crystal structure of a compound, as well as analysis of the intermolecular interactions, can provide important insights for improvement of the physicochemical properties, for example aqueous solubility. The difficulty lies in the fact that solubility is not only influenced by the properties of an isolated molecule (such as molecular volume and hydrogen-bond donor/acceptor strengths) but also involves the energy necessary to disrupt the crystal lattice to bring it into solution. Small changes to the substituents on the drug molecule can yield significant changes in its crystal structure, and therefore it is important to analyse crystal structures of closely related potential drug compounds in order to identify structures with the desired physicochemical properties, as well as biological activity.
Bicyclic compounds based on the bicyclo[3.3.1]nonane core can be used as prospective drug candidates because of their established biological activity. The literature describes applications as antibiotic, antimicrobial, antibacterial (Rosen et al., 2009), antiviral (Hrebabecký, Dracínský, Palma et al., 2009; Hrebabecký, Dracínský & Holý, 2009; Sála et al., 2010), anti-influenza, antifungal, cytotoxic and anti-cancer agents (Abdel-Hafez, 2008; Kiss et al., 2009; Koketsu et al., 2002; Nogueira et al., 2004). Such compounds also possess thymoleptic (mood-enhancing) properties, affect the blocking activity of membrane pumps of the central and peripheral monoamine neurones, and are potent in blocking the uptake into the central nervous system (Nakamura & Lipton, 2010). Ohta et al. (2007) and Zefirova et al. (2010) have reported on the radioprotective ability of 1,3-thiazine bicyclo analogues, and they emphasized that the lipophilicity of such compounds can be strongly enhanced by `insertion' of the parent structure into a bicyclo[3.3.1]nonane core. In addition, Shingo et al. (2010) have clearly demonstrated that a substituted tert-amino group in the structure of the bicyclic compound affects physicochemical properties such as aqueous solubility.
As the object of this study, we have synthesized three new drug-like compounds with a bicyclic structure and different substitution groups in the para position: N-(3-thia-1-azabicyclo[3.3.1]non-2-ylidene)-4-(trifluoromethyl)aniline (1); 4-fluoro-N-(3-thia-1-azabicyclo[3.3.1]non-2-ylidene)aniline (2); 4-methyl-N-(3-thia-1-azabicyclo [3.3.1]non-2-ylidene) aniline (3). In this paper, we report on the molecular and crystal structures of these compounds and measurements of the thermodynamic aspects of their sublimation and melting processes. The measured crystal lattice energies are correlated with the crystal structures using gas-phase and crystal structure calculations.
The compounds were synthesized as indicated below by reaction of 3-(bromomethyl)-piperidine hydrobromide with the corresponding arylisothiocyanate. A solution of sodium bicarbonate (1.85 g, 22 mmol) in the minimal amount of water was added dropwise to a stirred solution of the arylisothiocyanate (10 mmol) and (3-bromomethyl)piperidine hydrobromide (2.56 g, 10 mmol) in 30 ml of methanol. When the formation of the precipitate was complete, it was filtered and recrystallized from dioxane. Spectroscopic characterization is provided in the supporting information.1 Single crystals were grown by slow evaporation from n-hexane solution for (1) and (2), or dioxane solution for (3).
Single-crystal X-ray measurements were carried out using a Bruker Kappa APEX II CCD diffractometer with graphite-monochromated Mo K radiation ( = 0.7107 Å). The structures were solved by direct methods and refined by a full-matrix least-squares procedure. The CF3 group in compound (1) exhibits rotational disorder, which was modelled by including two orientations, with refined site occupancy factors 0.54 (5):0.46 (5). The C-F distances were restrained to a common refined value with s.u. 0.01 Å, and the FF distances were restrained to 1.633 times that value, also with s.u. 0.01 Å. Anisotropic displacement parameters were refined for all F atoms. The resulting displacement ellipsoids are highly prolate, but these were preferred over the possible inclusion of further partially occupied F atom sites.
Sublimation experiments were carried out by the transpiration method, as described elsewhere (Zielenkiewicz et al., 1999). In brief, a stream of an inert gas passes above the sample at a pre-determined slow constant flow rate under a constant temperature in order to saturate the carrier gas with the vapour of the tested substance. The vapour condenses at some point downstream, and the sublimate mass and purity are determined. The vapour pressure over the sample at this temperature can be calculated based on the amount of the sublimated sample and the volume of the inert gas used. The equipment was calibrated using benzoic acid. The standard value of sublimation enthalpy obtained was = 90.5 ± 0.3 kJ mol-1, which is in good agreement with the value recommended by IUPAC: = 89.7 ± 0.5 kJ mol-1 (Cox & Pilcher, 1970). The saturated vapour pressures were measured at each temperature five times with the standard deviation being within 3-5%. Since the saturated vapour pressure of the tested compounds is low, it may be assumed that the heat capacity change of the vapour with temperature is negligibly small.
The experimentally determined vapour pressure data may be described in (lnP, 1/T) coordinates in the following way
with , where P0 = 105 Pa.
Sublimation data are obtained at elevated temperatures for experimental reasons. However, in comparison with effusion methods, the temperatures in our case are much lower, which makes the extrapolation to room temperature conditions easier. In order to improve this extrapolation further, heat capacities (Cp,cr298) of the crystals were estimated using the additive scheme proposed by Chickos & Acree (2002). The values of Cp,cr298 for (1), (2) and (3) were 348.8, 304.2 and 316.0 J mol-1 K-1, respectively. Heat capacity was introduced as a correction for recalculation of the sublimation enthalpy at 298 K () according to the equation (Chickos & Acree, 2002)
The values of applied for correction of the sublimation enthalpies were 1.4, 2.2 and 1.9 kJ mol-1 for (1), (2) and (3), respectively.
Differential scanning calorimetry (DSC) was carried out using a DSC 204F1 Phoenix (Netzsch, Germany). DSC runs were performed in an atmosphere of flowing (25 ml min-1) dry Ar gas of high purity (99.996%) using standard Al sample pans and a heating rate of 10 K min-1. The DSC was calibrated using five standards (Hg, biphenyl, In, Sn and Bi). The sample mass was measured with an accuracy of 0.01 mg using a Sartorius M2P balance.
Geometric optimization of isolated molecules was carried out using the GAUSSIAN03 program at the B3LYP/6-311++G(d,p) level of theory (Frisch et al., 2003). Since no imaginary frequency was found, all of the optimized structures were characterized as minima. The crystal structures were analysed using the PIXEL approach developed by Gavezzotti (2005). This method provides quantitative determination of crystal lattice energies and pairwise intermolecular interactions, with a breakdown of these energies into coulombic, polarization, dispersion and repulsion terms. The classical atom-atom potential was also calculated using the CLP (Coulomb-London-Pauli) model (Gavezzotti, 2011). For compound (1), only one disorder component was included for the calculation. The free molecular volume in the crystal lattice was estimated on the basis of the X-ray diffraction data and van der Waals molecular volume (Vvdw) calculated using the spatial descriptors in Materials Studio (Accelrys, 2011). For compound (1), again only one disorder component was included. The molecular free volume (Vfree) is calculated as
where V is the volume of the unit cell and Z is the number of molecules in the unit cell. In general, Vfree is a composite value that comprises: (a) molecular volume determined by the conformational state of the molecule in the crystal; (b) free volume as a result of molecular packing in the crystal. We have previously found this parameter to be useful for analyzing relationships between crystal structures and thermodynamic quantities (Surov et al., 2009). To describe changes in the molecular packing density, the parameter has also been introduced (Surov et al., 2009). It shows the change of Vfree per molecule in the crystal relative to Vvdw.
| || Figure 1 |
Molecular structures of (1), (2) and (3), with displacement ellipsoids shown at 50% probability for non-H atoms. For (1), both disorder components of the CF3 group are shown.
Fig. 2 shows an overlay of the molecules in (1), (2) and (3), based on the non-H atoms of the 3-thia-1-aza-bicyclo fragment. It is evident that all molecules have a similar conformation in the crystal. The conformational differences are caused mainly by rotation of the phenyl fragment about the C4-N1 bond, while the bicycle part of the molecule remains unchanged. The conformational differences are quantified by the torsion angle C8-N1-C4-C3: molecules (1) and (2) exhibit almost identical values (71.6 versus 67.9°), while the angle decreases to 47.1° for (3).
| || Figure 2 |
Overlay of the molecules based on the non-H atoms of the 3-thia-1-aza-bicyclo fragment. Colour code: (1) red, (2) blue, (3) green. H atoms are omitted, and only one component of the disordered CF3 group is shown for (1).
Packing diagrams for (1), (2) and (3) are shown in Fig. 3. Inspection of the structures shows two distinct arrangements for the 3-thia-1-aza-bicyclo and phenyl fragments. The first is observed in compound (1), where a layer-like structure is formed by molecules, with a `head-to-tail' orientation within each layer. In contrast, the structures of (2) and (3) are constructed from alternating layers of the 3-thia-1-aza-bicyclo and phenyl fragments. The angle between the least-squares planes of the neighbouring phenyl rings in (2) and (3) is 66.3 and 79.6°, respectively.
| || Figure 3 |
Packing diagrams for (1), (2) and (3). H atoms are omitted. Both components of the disordered CF3 group are shown for (1).
In order to measure the crystal lattice energies and thermochemical characteristics of the compounds, sublimation and DSC experiments were carried out. The temperature dependencies of the saturated vapour pressure are summarized in Table 2. The thermodynamic functions of sublimation and fusion processes are presented in Table 3. The results suggest that the crystal lattice energy () decreases in the order: (3) > (2) > (1).
+ln(P (Pa)) = (29.0 ± 0.3) - (10868 ± 93)/T; R2 = 0.999; = 2.6 × 10-2; n = 15.
§ln(P (Pa)) = (31.5 ± 0.2) - (11928 ± 65)/T; R2 = 0.999; = 2.1 × 10-2; n = 15.
Although the differences between the measured lattice energies of (1), (2) and (3) are small, they reflect the crystal structures of the compounds, and in particular the parameter (Table 1 and §2.5). It is found that the crystal lattice energy () increases while decreases, which corresponds to a packing density increase. A similar dependence is observed for the Gibbs energy of sublimation (). The melting temperature of the compounds (Table 3), however, shows the opposite trend, decreasing as decreases. We believe that this indicates a different mechanism of crystal destruction under melting conditions, compared with sublimation conditions. Similar results have been obtained for a range of structurally related spiro compounds [phenyl-(1-thia-3-aza-spiro[5.5]undec-2-en-2-yl)-amines] studied by our group recently (Ol'khovich et al., 2012): replacing CH3 with CF3 at the phenyl ring of those compounds also causes the sublimation enthalpy () to decrease but the melting temperature to increase.
In order to study the conformational flexibility of the molecules for the crystal-to-gas transfer process, geometrical optimization was performed at the B3LYP/6-311**(d,p) level of theory. Molecules were extracted from the crystal structure and used as the starting point for the optimization procedure. For compound (1), only one component of the disordered CF3 group was retained. Selected geometric parameters for the non-optimized and optimized molecules are shown in Table 4. The calculations show that all of the optimized molecules adopt almost identical conformations. Thus, the character of the substituent has only a minor influence on the conformational state of the molecules after the optimization procedure, and it can be concluded that all geometrical differences between the molecules studied in the crystals are caused by optimization of the packing energy. The maximum distortion of the C8-N1-C4-C3 torsion angle between the crystal and optimized conformations is observed for compound (1) (13.9°), while the minimum distortion is observed for compound (3) (-8.6°). The other selected geometric parameters show a similar trend.
The calculated lattice energies obtained using the PIXEL and CLP methods were compared with the experimental sublimation enthalpies, corrected by adding a constant contribution of 2RT to approximate the zero-point energy and thermal correction to 298 K (Gavezzotti & Filippini, 1997)
The results are summarized in Table 5. The PIXEL approach shows reasonable agreement with the experimental crystal lattice energies. The calculated energies are overestimated by 10% on average, which is within the confidence interval declared by Gavezzotti as an average percent discrepancy (Gavezzotti, 2005). However, the CLP method (based on atom-atom potentials) fails to predict the crystal lattice energy order for the compounds studied. Since the PIXEL results agree with the experimental data, they should provide a sound indication of the energy contributions to the total packing energy. As expected, the dispersion energy dominates the crystals of all of the compounds, while the electrostatic (coulombic and polarization) terms play a lesser role. Replacement of the CF3 group in (1) by the CH3 group in (3) appears to increase the attractive part of the crystal lattice energy (the sum of Coulombic, polarization and dispersion terms) by ca 63 kJ mol-1. On the other hand, the repulsive energy increases by a factor of two, with the result that the lattice energy of (3) gains only ca 7 kJ mol-1 compared with (1). It should be stressed that these results do not take into account the rotational disorder observed for the CF3 group in the crystal structure of (1).
+ElattCLP is the total CLP crystal lattice energy.
§Elattexp is the experimental crystal lattice energy calculated according to equation (6).
Geometric optimization of the isolated molecules shows that the varying CF3, F or CH3 substituent in (1), (2) or (3) has very little influence on the molecular conformation. The molecular conformation is slightly more variable in the crystal, attributed to optimization of the crystal lattice energy. The crystal structures of (2) and (3) are similar, with alternating layers of 3-thia-1-aza-bicyclo and phenyl fragments, while the crystal structure of (1) contains layers with molecules having a `head-to-tail' orientation within each layer. Thus, introduction of the CF3 substituent in place of F or CH3 has a significant influence on the crystal structure. The measured crystal lattice energies are directly correlated with the defined parameter, which corresponds to a direct correlation between and packing density. However, the melting temperatures show the opposite correlation, decreasing with increasing packing density. Calculation of the crystal lattice energy using the PIXEL approach compares quite well with the experimental values, and suggests that the dispersion energy dominates all three crystal structures, while the electrostatic (coulombic and polarization) terms play a lesser role.
This work was supported by a grant from the President of the RF, No. MK-7097.2012.3.
Abdel-Hafez, Sh. H. (2008). Eur. J. Med. Chem. 43, 1971-1977.
Accelrys (2011). Materials Studio. Accelrys Inc., San Diego, CA, USA.
Bruker (2008). APEX2 and SADABS. Bruker AXS Inc., Madison, Wisconsin, USA.
Chickos, J. S. & Acree, W. E. (2002). J. Phys. Chem. Ref. Data, 31, 537-698.
Cox, J. D. & Pilcher, G. (1970). Thermochemistry of Organic and Organometallic Compounds. London: Academic Press.
Enraf-Nonius (1993). CAD-4. Enraf-Nonius, Delft, The Netherlands.
Frisch, M. J. et al. (2003). GAUSSIAN03, Revision B. 03. Gaussian Inc., Pittsburgh, PA, USA.
Gavezzotti, A. (2005). Z. Kristallogr. 220, 499-510.
Gavezzotti, A. (2011). New J. Chem. 35, 1360-1368.
Gavezzotti, A. & Filippini, G. (1997). Theoretical Aspects and Computer Modeling of the Molecular Solid State, edited by A. Gavezzotti, pp. 61-97. Chichester: Wiley and Sons.
Hrebabecký, H., Dracínský, M., De Palma, A. M., Neyts, J. & Holý, A. (2009). Collect. Czech. Chem. Commun. 74, 487-502.
Hrebabecký, H., Dracínský, M. & Holý, A. (2009). Collect. Czech. Chem. Commun. 74, 1425-1441.
Kiss, L., Nonn, M., Forró, E., Sillanpää, R. & Fülöp, F. (2009). Tetrahedron Lett. 50, 2605-2608.
Koketsu, M., Tanaka, K., Takenaka, Y., Kwong, C. D. & Ishihara, H. (2002). Eur. J. Pharm. Sci. 15, 307-310.
Nakamura, T. & Lipton, S. A. (2010). Cell Calcium, 47, 190-197.
Nogueira, C. W., Zeni, G. & Rocha, J. B. T. (2004). Chem. Rev. 104, 6255-6285.
Ohta, S., Matsuda, S., Gunji, M. & Kamogawa, A. (2007). Biol. Pharm. Bull. 30, 1102-1107.
Ol'khovich, M. V., Sharapova, A. V., Blokhina, S. V., Perlovich, G. L. & Proshin, A. N. (2012). J. Chem. Eng. Data, 57, 3452-3457.
Rekka, E. & Kourounakis, P. (2008). Chemistry and Molecular Aspects of Drug Design and Action, edited by E. Rekka & P. Kourounakis. Boca Raton: CRC Press.
Rosen, J. D., German, N. & Kerns, R. J. (2009). Tetrahedron Lett. 50, 785-789.
Sála, M., De Palma, A. M., Hrebabecký, H., Nencka, R., Dracínský, M., Leyssen, P., Neyts, J. & Holý, A. (2010). Bioorg. Med. Chem. 18, 4374-4384.
Sheldrick, G. M. (2008). Acta Cryst. A64, 112-122.
Shingo, K., Yoshiyuki, H., Yasuo, N., Makio, S., Masahiko, T., Masahide, Y., Kimiye, B. & Shinichi, U. (2010). Bioorg. Med. Chem. 18, 3925-3933.
Surov, A. O., Terekhova, I. V., Bauer-Brandl, A. & Perlovich, G. L. (2009). Cryst. Growth Des. 9, 3265-3272.
Zefirova, O. N., Nurieva, E. V., Chupakhin, V. I., Semenova, I. S., Peregud, D. I., Onufriev, M. V. & Gulyaeva, N. V. (2010). Mendeleev Commun. 20, 323-325.
Zielenkiewicz, W., Perlovich, G. L. & Wszelaka-Rylik, M. (1999). J. Therm. Anal. Calorim. 57, 225-234.