High-throughput structure determination with X-ray crystallography has demonstrated its value to accelerate drug discovery for over a decade. It can facilitate not only the optimisation of lead compounds and target identification, but also lead discovery through increasing screening capabilities. But despite the progress in genome sequencing, robotics and bioinformatics, the structures and functional mechanisms of membrane proteins are still relatively equivocal. Membrane proteins account for approximately 30% of all proteins (Wallin & Heijne, 1998) and are targeted by an estimated 60% of all drugs (Overington et al., 2006). Their crystal structures, however, only comprise ~3% of those in the Protein Data Bank (PDB). So, what challenges remain in their structural determination, and will the recent resurgence in cryo electron microscopy (cryo-EM) be critical to furthering our insight into these proteins?
To begin with we must consider the hurdles which must be overcome for the crystallisation of a protein. These lie firstly in the overexpression of the protein (Seddon et al., 2004); secondly in the extraction of this protein out of the membrane in a non-aggregated, stable state; and finally, in making sufficient crystal contacts with the protein crystal for structural determination, a problem augmented by membrane proteins due to their tendency to exhibit inherent flexibility (Bill et al., 2011). Strategies have been employed to overcome some of these problems, including mutagenic or chimeric approaches (Abdul-Hussein et al., 2013; Caffrey, 2015), with the overall intention to improve membrane stability, increase crystal contacts and fix conformational states. However, these are not universally applicable and many membrane proteins have proven to be resistant to crystallisation irrespective of focused efforts.
So when does it become necessary to look for an alternative? Cryo-EM may not have provided one previously as high resolution protein structures were limited by both EM hardware and smaller membrane proteins’ tendency to aggregate. Yet, this method had always evaded some of the major challenges of crystallography, for example reducing the amount of protein needed in comparison to crystal studies (μg vs. mg scale). And, with existing advancements in the EM hardware, in terms of detectors and microscopes, the use of this technique for the study of membrane protein structure is being readily adopted. Crucially, cryo-EM specimens are made by fast freezing biological samples from the solution directly in liquid nitrogen temperature. This ability to maintain the protein complex in its soluble state permits the structure to be examined in a state much more closely resembling its native state. Cryo-EM can now obtain near-atomic resolution structures of macromolecular complexes up to several MDa in size and, in combination with image classification algorithms, not only are high resolution 3D maps now possible, but also the evaluation of several conformational states of the same sample.
This changing landscape in recent years has permitted the high resolution structural determination of a number of membrane proteins. One key group are the ryanodine receptors (RyR’s), which facilitate the intracellular release of Ca2+, integral to muscle contraction. They are implemented in cardiac arrhythmias and have surfaced as potential therapeutic targets for heart failure (Betzenhauser & Marks, 2010; Anderson & Marks, 2010). More specifically the RyR1 channel, a tetrameric channel with a molecular mass >2.2 MDa, lends itself to EM structural determination over x-ray crystallography. Nevertheless, until a couple of years ago the highest resolution structure obtained was a mere ~1 nm resolution (Ludtke et al., 2005), which could only give an indication of overall channel architecture. This channel has since been determined to a much higher ~4 Å (Yan et al., 2015), an impressive feat when considering that it is the largest of all known ion channels. From this, several new important functional domains and changes in channel conformation have been identified, helping to deduce potential channel gating mechanisms.
It must not be overlooked that cryo-EM does pose its own, new challenges. However, the exciting opportunities offered by cryo-EM has motivated a focused effort to improve application, from sample preparation, to data collection and processing, to the modelling of the 3D maps. Some challenges still remain: one of the major factors preventing the attainment of even higher resolution structures is radiation damage caused by exposure to high energy electrons (Glaeser & Taylor, 1978). Even at very small electron doses (~3 e–/Å2) damage has been shown on some charged side chains (Grant & Grigorieff, 2015a) and damage to the specimen caused by doses above ~10 e–/Å2 result in a loss of diffraction from 2D and 3D crystals (Baker et al., 2010). It is therefore thought that the early frames of a direct detector movie contain the highest resolution signal, however at present are not able to be recovered due to limitations in the hardware.
Still in its early stages, cryo-EM is unlikely to replace X-ray crystallography entirely and it is also unlikely that there will be a method unanimous to all membrane proteins. It may be that a combination of techniques could be used in conjunction with each other. Examples of this include the docking of X-ray crystallographic structures within cryo-EM maps, or solving the phases of X-ray crystallographic diffraction data using cryo-EM maps. Although X-ray crystallography has proved to be a powerful tool, cryo-EM is enabling the structural study of some membrane targets which had previously been deemed unattainable. And with the advent of new technologies, the future for membrane protein structural studies is filled with potential.
Blog written by Victoria Miller
Abdul-Hussein S, Andréll J, Tate CG. 2013. Thermostabilisation of the serotonin transporter in a cocaine-bound conformation. J Mol Biol 425:2198-2207
Anderson DC & Marks AR. 2010. Fixing ryanodine receptor Ca leak – a novel therapeutic strategy for contractile failure in heart and skeletal muscle. Drug Discov Today Dis Mech 7:e151-157
Baker LA, Smith EA, Bueler SA, Runimsteim JL. 2010. The resolution dependence of optimal exposures in liquid nitrogen temperature electron cryomicroscopy of catalase crystals. J Struct Biol 169:431-437
Betzenhauser MJ & Marks AR. 2010. Ryanodine receptor channelopathies. Pflugers Arch 460:467-480
Bill RM, Henderson PJF, Iwata S, Kunji ERS, Michel H, Neutze R, et al. 2011. Overcoming barriers to membrane proteins in vitro. Methods Mol Biol 601:219-245
Caffrey M. 2015. A comprehensive review of the lipid cubic phase or in meso method for crystallising membrane and soluble proteins in vitro. Methods Mol Biol 601:219-245
Glaeser RM & Taylor KA. 1978. Radiation damage relative to transmission electron microscopy of biological specimens at low temperature: a review. J Microscopy 112:127-138
Grant T & Grigorieff N. 2015a. Measuring optimal exposure for single particle cryo-EM using a 2.6 Å reconstruction of rotavirus VP6. eLife 4:e06980
Ludtke SJ, Serysheva II, Hamilton SL, Chiu W. 2005. The pore structure of the closed RyR1 channel. Structure 13:1203-1211
Overington JP, Al-Lazikani B, Hopkins AL. 2006. How many drug targets are there? Nat Rev Drug Discov 5:993-996
Seddon AM, Curnow P, Booth PJ. 2004. Membrane proteins, lipids and detergents: not just a soap opera. Biochem Biophys Acta 1666:105-117
Wallin E & Heijne von, G. 1998. Genome-wide analysis of integral membrane proteins from eubacterial, archaean, and eukaryotic organisms. Protein Sci 7:1029-1038
Yan Z, Bai X-C, Yan C, Wu J, Li Z, Xie T, et al. 2015. Structure of the rabbit ryanodine receptor RyR1 at near-atomic resolution. Nature 517:50-55