When computation meets experiment : a combined approach towards understanding crystalline materials properties

Abstract

The solid-state behaviour of an active pharmaceutical ingredient has a significant influence on its function and effectiveness in the body. Due to such importance, a crystallographic study was conducted to analyse and predict such behaviour in three prominent antivirals, namely ganciclovir, famciclovir and valacyclovir (hydrochloride form), all of which act as nucleoside analogue polymerase inhibitors against herpesviruses. This investigation incorporates the complementary use of multiple computational approaches, novel or otherwise, which offer a direct comparison of results and provide a more holistic understanding. A new scoring system for the quantification of molecular flexibility was devised, given the proven lack of appropriateness of the number of rotatable bonds parameter to represent such molecular feature exhaustively. This was followed by conformational analysis, which provided the parameters of the accessible conformational space of each molecule. In addition, this investigation entailed a detailed exploration of each crystalline antiviral, particularly the administered form, from intramolecular and intermolecular levels, including full interaction maps and energy frameworks. This analysis provided an insight of how features at these levels can manifest their impacts on supramolecular traits, such as the conformational polymorphism between GCV form I and II. The crystallographic behaviour was also characterised through the use of variable temperature powder X-ray diffraction and simultaneous thermal analysis. Both intermolecular analysis and experimental work supported the thermodynamic stability of GCV and FCV form I, with minimal risk of polymorphism, as opposed to VCV⸱HCl. This study displayed how polymorphism is governed by the present degree of stability, which in return is influenced by the HB donors to acceptors ratio, HB coordination number per group and the strength of other non-bonding contacts, beyond HB. Different polymorph and co-crystal formation prediction techniques were validated against parallel experimental work and examined to understand better the extent of their applicability. Results demonstrated how the HB propensity model approach is not adequate to classify the thermodynamic stability of forms whose non-bonding network is heavily composed of non-conventional HB interactions and other contacts beyond HB, as GCV and FCV form I. The prediction of co-crystallisation was conducted through four distinct approaches, which were accompanied by the testing of multiple settings so as to improve the quality of outcomes. These considerations were related to the inputted molecular conformations and the selection of interactions. The ranking based on the excess enthalpy evaluated through the COSMO approach, resulted to be an adequate indicator of the suitability of coformers, especially for FCV co-crystals. The highest overall success rate was associated with the HB energy approach at 74.47%, followed by molecular complementarity (68.97%) and HB propensity model (65.38%) procedures, all of which performed better for GCV trials.