Zwitterionic or Not? Fast and Reliable Structure Determination by Combining Crystal Structure Prediction and Solid-State NMR

Abstract

When it comes to crystal structure determination, computational approaches such as Crystal Structure Prediction (CSP) have gained more and more attention since they offer some insight on how atoms and molecules are packed in the solid state, starting from only very basic information without diffraction data. Furthermore, it is well known that the coupling of CSP with solid-state NMR (SSNMR) greatly enhances the performance and the accuracy of the predictive method, leading to the so-called CSP-NMR crystallography (CSP-NMRX). In this paper, we present the successful application of CSP-NMRX to determine the crystal structure of three structural isomers of pyridine dicarboxylic acid, namely quinolinic, dipicolinic and dinicotinic acids, which can be in a zwitterionic form, or not, in the solid state. In a first step, mono- and bidimensional SSNMR spectra, i.e., 1H Magic-Angle Spinning (MAS), 13C and 15N Cross Polarisation Magic-Angle Spinning (CPMAS), 1H Double Quantum (DQ) MAS, 1H-13C HETeronuclear CORrelation (HETCOR), were used to determine the correct molecular structure (i.e., zwitterionic or not) and the local molecular arrangement; at the end, the RMSEs between experimental and computed 1H and 13C chemical shifts allowed the selection of the correct predicted structure for each system. Interestingly, while quinolinic and dipicolinic acids are zwitterionic and non-zwitterionic, respectively, in the solid state, dinicotinic acid exhibits in its crystal structure a “zwitterionic-non-zwitterionic continuum state” in which the proton is shared between the carboxylic moiety and the pyridinic nitrogen. Very refined SSNMR experiments were carried out, i.e., 14N-1H Phase-Modulated (PM) pulse and Rotational-Echo Saturation-Pulse Double-Resonance (RESPDOR), to provide an accurate N–H distance value confirming the hybrid nature of the molecule. The CSP-NMRX method showed a remarkable match between the selected structures and the experimental ones. The correct molecular input provided by SSNMR reduced the number of CSP calculations to be performed, leading to different predicted structures, while RMSEs provided an independent parameter with respect to the computed energy for the selection of the best candidate.

Organic NMR crystallography: enabling progress for applications to pharmaceuticals and plant cell walls

“…46 Brouwer and Mikolajewski have recently presented GIPAW calculations along with 1 H DQ and 1 H- 13 C heteronuclear correlation MAS NMR spectra for glucose, to identify trends in the 1 H chemical shift with hydrogen bonding parameters, 47 noting that Shen et al have presented GIPAW calculations to complement 17 O MAS NMR experiments for the same sugar molecule. 48 Chierotti and co-workers have combined experiment such as 1 H DQ and 1 H- 13 C heteronuclear correlation as well as 14 N-1 H MAS NMR spectra, and GIPAW calculation to study co-crystals of the pharmaceutical ethionamide, 49 probe tautomerism in the pharmaceutical mebanazole, 50 identify zwitterions, in combination with CSP, in isomers of pyridine dicarboxylic acid, 51 and to analyse leucopterin, the white pigment in butterfly wings, including a 1 H DQ MAS spectrum at 1 GHz. 52 Working together with scientists at AstraZeneca and Pfizer, Brown and co-workers have presented NMR crystallography studies of a range of pharmaceutical molecules.…”