Date Published: March 01, 2017
Publisher: International Union of Crystallography
Author(s): Colin R. Groom, Jason C. Cole.
Small-molecule crystal structures are of tremendous value in understanding protein–ligand complexes, both individually and as a collection.
Combining macromolecular crystallography with small-molecule crystallography yields valuable complementary information on any ligands that are present, information which can help to address many of the challenges in generating small molecules for use in a biological context. At its simplest, the availability of a small-molecule crystal structure, from either powder diffraction or single-crystal diffraction studies, gives insight into a potential conformation of a molecule. Beyond this, it reveals the interactions that a molecule makes, both with itself and neighbouring molecules. Contrast and comparison of these conformations and interactions with those made with a protein are often revealing.
The conformation of a molecule ‘in the solid form’ (i.e. a small-molecule single crystal or powder) will be one of a repertoire of conformations that the molecule can adopt. This conformation will be a compromise between the energetic minima of the conformation and the interactions that the molecule can make. Also influencing the conformation may be considerations regarding the generation of symmetry, to allow a repeating lattice and the kinetic accessibility of a particular crystalline form. However, the compromise between conformation and interactions to give either a small-molecule lattice or a protein–ligand complex are fundamentally the same. What are sometimes referred to as the effects of mysterious ‘crystal packing forces’ are in fact both rare and explicable (Cruz-Cabeza et al., 2012 ▸). The beautiful balance seen between adjusting conformation to optimize interactions is common in small-molecule and protein systems. It is usually more appropriate to think of a molecule being gently pulled from its exact energetically minimal conformation to create the best fit to a protein binding site or the most optimal lattice than to think of a molecule being pushed into an unfavourable conformation owing to unfavourable interactions. The maxim that ‘proteins don’t strain ligands, protein crystallographers do’ is worth keeping in mind (Liebeschuetz et al., 2012 ▸; Rupp et al., 2016 ▸). Should the conformation a molecule in a small-molecule lattice be similar to that when bound to a protein, this can give vital information to the structural biologist. An illustrative example is that of HCV NS5B inhibitors developed by Bristol-Myers Squibb (Gentles et al., 2014 ▸). Key to the activity of these molecules are the dihedral angles within the cyclopropylindolobenzazepine rings, specifically the angle between the fused methoxy-substituted phenyl moiety and the indole ring (Fig. 1 ▸). Comparison of the small-molecule structure (CSD Refcode MIYWIC) and protein–ligand complex (PDB entry 4nld) shows this similarity.
Thus far, we have focused on the influence of conformation on the affinity and properties of molecules. Although reasonably strong binding to a target receptor is indeed important, many other properties influence the use of a compound as a protein binder. Foremost among these is solubility. In a drug-discovery context, solubility influences absorption, bioavailability and target exposure (Di et al., 2009 ▸; Williams et al., 2013 ▸). Poorly soluble compounds may also make it difficult to reconcile the observed in vitro, cellular and in vivo activities. For the macromolecular crystallographer, the difficulties are no less severe. It can prove difficult to generate protein–ligand complexes, either by co-crystallization or through soaking experiments, for poorly soluble compounds. Even if a compound is more soluble in organic conditions, or with solvents such as DMSO, these can prevent crystallization. It is, therefore, important for the macromolecular crystallographer to understand the drivers behind solubility and, more importantly, to understand how the skills of a structural scientist can be brought to bear on this problem.
It is somewhat surprising that, given the commitment in obtaining a crystal structure of a protein–ligand complex, more resources are not invested in generating small-molecule crystal structures of these ligands. At its very simplest, this is an effective way of confirming the precise chemical structure of the material thought to be under study. Whether commercially sourced or not, the question ‘is this what it says on the bottle?’ should always be considered (Halford, 2012 ▸).