Expanding the horizons of G protein-coupled receptor structure-based ligand discovery and optimization using homology models

Abstract

With >800 members in humans, the G protein-coupled receptor (GPCR) super-family is the target for more than 30% of the marketed drugs. The recent boom in GPCR crystallography has enabled the solution of ∼30 different GPCR structures, which boosted the identification and optimization of novel modulators and new chemical entities through structure-based strategies. However, the number of available structures represents a small part of the human GPCR druggable target space, and its complete coverage in the near future seems unlikely. Homology modelling represents a reliable tool to fill this gap, and hence to vastly expand the horizons of structure-based drug discovery and design. In this Feature Article, we show from a wealth of retrospective and prospective studies that in spite of the pitfalls of and standing challenges in homology modelling, structural models have been critical for the blossoming and success of GPCR structure-based lead discovery and optimization endeavours; in addition, they have also been instrumental in characterizing receptor-ligand interaction, guiding the design of site-directed mutagenesis and SAR studies, and assessing off-target effects. Considering though their current limitations, we also discuss the most pressing issues to develop more accurate homology modelling strategies, with a special focus on the integration of computational tools with biochemical, biophysical and QSAR data, highlighting methodological aspects and recent progress. The teachings of the three GPCR Dock community-wide assessments and the fresh developments in GPCR classes B, C and F are commented. This is a fast growing and highly promising field of research, and in the coming years, the use of high-quality models should enable the discovery of a growing number of potent, selective and efficient GPCR drug leads with high therapeutic potential through receptor structure-based strategies.

Dynamic roles for the N-terminus of the yeast G protein-coupled receptor Ste2p

“…Inappropriate and / or altered GPCR functions cellular are directly or indirectly associated with many diseases including schizophrenia, Alzheimer's, cancer, blindness, obesity, hypertension, and diabetes [2, 6, 7]. Not surprisingly, GPCRs are the therapeutic targets for a large portion of currently prescribed drugs [8-12]. …”

Computational chemistry in drug lead discovery and design

“…Although small domains might be modeled ab initio , this is an unrealistic strategy for standard proteins. In this case, in silico structural characterization through homology modeling appears as a sound technique which could expand approximately by two orders of magnitude the number of protein structures . Needless to say, structural models could also expand the horizons to characterize protein structural properties, function and mechanisms, and to assess target druggability (cf.…”

Open challenges in structure-based virtual screening: Receptor modeling, target flexibility consideration and active site water molecules description

“…It is thus clear that, in the years to come, a complete coverage of the druggable target space will not be available for structure-based drug lead discovery and optimization endeavors. In this scenario, accurate in silico homology modeling appears as a reliable alternative to fill the gap and expand by even two orders of magnitude the number of available structures [30,31].…”

“…In case of poor sequence identity (either global or at the binding site), structural conservation should be examined whenever possible. For example, the average low sequence identity in GPCRs (around 20%, less than the 30-50% threshold generally necessary to develop good homology models [39]) is compensated by the conserved seven-transmembrane fold, which allows the generation of high-quality models usable in drug design campaigns [31,[40][41][42][43][44]. In cases of low sequence similarity, multiple sequence alignment might help in improving the quality of the final model [45][46][47], though it should be stressed that a correct sequence alignment is necessary -though hardly sufficient-to develop reliable homology models.…”