38 39 D I A M O N D L I G H T S O U R C E A N N U A L R E V I E W 2 0 2 0 / 2 1 D I A M O N D L I G H T S O U R C E A N N U A L R E V I E W 2 0 2 0 / 2 1 Biological Cryo-Imaging Group eBIC Visualising amembrane receptor complex to enhance drug development Related publication: LeeY.,WarneT., Nehmé R., Pandey S., Dwivedi-Agnihotri H., Chaturvedi M., Edwards P. C., García-Nafría J., Leslie A. G.W., Shukla A. K. &Tate C. G. Molecular basis of β-arrestin coupling to formoterol-bound β1-adrenoceptor. Nature 583 , 862–866 (2020). DOI: 10.1038/ s41586-020-2419-1 Publication keywords: Cryo-EM; GPCR; Membrane protein; Structure; Arrestin A third of all FDA approved drugs target G protein-coupled receptors (GPCRs) but often have side effects. GPCRs have two signalling pathways - the G protein-coupled pathway and the arrestin-mediated pathway - and one may be the therapeutic pathway whilst the other may produce side effects. To understand this phenomenon at the molecular level, researchers needed to determine the structure of a GPCR coupled to arrestin and compare its structure to the G protein-coupled state, with the same agonist bound to both receptors. The GPCR-arrestin complex is highly mobile, which means it would not form crystals suitable for X-ray diffraction experiments. Therefore, the team used cryogenic-electron microscopy (cryo-EM) at the Electron Bio-Imaging Centre (eBIC) at Diamond Light Source to determine its structure at a 3.3Å resolution. Producing a stable GPCR-arrestin complex suitable for structure determinationwas still a challenge. Arrestin coupling to a GPCR requires the presence ofmembrane lipids, so they prepared the receptor in lipid nanodiscs. The receptor also has to be phosphorylated, so they developed a method for ligating phosphorylated peptides to the end of the receptor. Their results allowed them to identify two regions of the GPCR that could be used in the development of ‘biased’ agonists that signal predominantly through either the G protein or arrestin pathways. The structures will be an invaluable input into structure-based drug design, a powerful tool for the development of new therapeutic drugs. The human body relies on an extensive network of signalling molecules (agonists) such as hormones and neurotransmitters to co-ordinate bodily functions. GPCRs are the major family of receptors on the surface of cells that bind hormones. Upon a GPCR binding a hormone, there is a subsequent conformation change on the intracellular face of the receptor resulting in binding and activation of G proteins and arrestins, which then activate downstream signalling cascades to alter the cells’biochemistry. The pivotal role GPCRs play in intercellular communication makes them ideal drug targets. Over 80 years of drug discovery has resulted in an incredible array of molecules that bind to GPCRs and either activate or inhibit them. Currently 34% of FDA approved drugs target GPCRs for the treatment of diverse conditions such as high blood pressure, pain, asthma, heart conditions and migraines 1 . However, many drugs have side effects that limit their use and there is a continual drive to understand the molecular pharmacology of GPCRs and how to develop new drugs. Structural biology of GPCRs offers the opportunity for targeted drug development through structure-based drug design, and there are now examples of such therapeutics in clinical trials 2 . There are many reasons why drugs have side effects, such as they bind to other receptors or they are metabolised to compounds that have a different pharmacology. However, some side effects of drugs at GPCRs arise through the ability of agonists to signal through both the G protein pathway and the arrestin pathway. In many cases, only one of these pathways produces the therapeutic effect, whilst the other pathway produces side effects. Thus, if a therapeutic could be developed that could signal down only the therapeutic pathway, a so-called biased agonist, then side effects could be significantly reduced 3 .To understand the molecular basis of biased signalling, it is necessary to determine the structure of a GPCR bound to an agonist and coupled to either β-arrestin or a G protein. This is what we have done in the current work. In the present study, we determined the structure of the β 1 -adenoceptor (β 1 AR) coupled to β-arrestin 1 (βarr1) by cryo-EM. Preparation of the sample was not straightforward because βarr1 couples to only a phosphorylated receptor and requires lipids around the receptor for efficient coupling. We ligated a phosphorylated peptide containing six phosphoresidues (V 2 R 6P ) to the C-terminus of the receptor by sortase mediated ligation and then inserted the purified receptor into lipid nanodiscs. The complex was then formed by adding the agonist formoterol, βarr1 and a conformation-specific antibody F ab fragment, F ab 30, that binds to βarr1 and locks it in an active receptor-coupled state. The additional mass provided by the F ab was also essential for good alignment of particles during image processing. The sample was then imaged by cryo-EM, although the sample had to be tilted at 30˚ to reduce the effects during processing of insufficient different views of the complex. Computational processing of the images generated a 3D structure with an overall resolution of 3.3 Å which allowed modelling of the whole receptor (Fig. 1). The β 1 AR- βarr1 structure showed that the arrestin made three distinct interactions. Firstly, there was a direct interaction with the lipids within the nanodisc, explaining the requirement of βarr1 for lipids during coupling to a GPCR. Secondly, there were extensive interactions between βarr1 and each of the six phosphate groups in the phosphorylated C-terminus. Thirdly, there were extensive interactions between β 1 AR and βarr1, and in particular the finger loop of βarr1 inserted into a cleft in the cytoplasmic face of the receptor at a site that is known to couple G proteins (Fig. 2). We could compare the interfaces between βarr1 and β 1 AR with the interface between β 2 AR and the G protein G s , as β 2 AR is known to be virtually identical to β 1 AR in multiple different conformations. This showed that there were distinct differences in both the positions of transmembrane helices and residues involved in binding to either a G protein or arrestin. The junction between a receptor and either arrestin or G protein is therefore distinct (Fig. 2) and could be targeted by small molecules that would result in preferential binding of either transducer, resulting in biased signalling. Agonists bind to the orthosteric binding site in the extracellular half of the receptor and there is allosteric coupling to the intracellular surface of the receptor to promote transducer coupling. Coupling of a G protein to a receptor increases its affinity for agonists by decreasing the volume of the binding site by up to 40% and thus increasing the number and/or strength of receptor- ligand interactions 4 . However, arrestin coupling does not result in such a large increase in affinity, suggesting that there were structural differences in the orthosteric binding site depending whether arrestin or a G protein were coupled. To determine this, we also solved a structure by X-ray crystallography ofβ 1 ARboundtothesameagonist(formoterol)andcoupledtoaconformation- specific nanobody Nb80, which is a G protein mimetic. Comparing the two binding sites identified subtle changes in the position of transmembrane helices that resulted in fewer formoterol- β 1 AR hydrogen bonds and van der Waals interactions when β 1 AR was coupled to βarr1 compared to when it was coupled to Nb80. These, and other changes in the orthosteric binding site, such as the different position of extracellular loop 3, would be sufficient to develop novel ligands that had biased signalling towards arrestin rather than a G protein. The new structure of β 1 AR coupled to βarr1 and its comparison with G protein-coupled states allows an understanding of the molecular basis for biased signalling. This will aid the design of new and improved therapeutics targeting GPCRs that will have reduced side effects through activating preferentially only desirable intracellular pathways. References: 1. Hauser A. S. et al. Trends in GPCR drug discovery: New agents, targets and indications. Nat. Rev. Drug Discov. 16 , 829–842 (2017). DOI: 10.1038/ nrd.2017.178 2. Congreve M. et al. Impact of GPCR Structures on Drug Discovery. Cell 181 , 81–91 (2020). DOI: 10.1016/j.cell.2020.03.003 3. Kenakin T. Biased receptor signaling in drug discovery. Pharmacol. Rev. 71 , 267–315 (2019). DOI: 10.1124/pr.118.016790 4. Warne T. et al. Molecular basis for high-affinity agonist binding in GPCRs. Science. 364 , 775–778 (2019). DOI: 10.1126/science.aau5595 Funding acknowledgement: Medical Research Council MC_U105197215, European Research Council (EMPSI 339995) and Sosei Heptares. We acknowledge Diamond Light Source for access and support of the cryo-EM facilities at the UK’s national Electron Bio-imaging Centre (eBIC) [under proposal BI23268], funded by theWellcome Trust, MRC and BBRSC. Corresponding author: Dr Christopher G. Tate, MRC Laboratory of Molecular Biology,
[email protected] Figure 2: (a,b) Differences in engagement of the finger loop of arrestin and the C-terminal α 5 helix of the G protein G s ; (c,d) Surface representations of the β 1 AR-βarr1 complex and β 2 AR-G s complex highlighting the distinct junctions at the receptor-transducer interfaces suitable for the development of biased drugs (pink oval); (e) Superposition of the formoterol-β 1 AR-βarr1 complex with the formoterol-β 1 AR-Nb80 complex around the orthosteric binding site. Side chains (sticks) that interact with formoterol (sticks) are shown with hydrogen bonds (dashed lines, pale blue). Transmembrane helices and extracellular loop 3 (ECL3) are labelled. Figure panels are reproduced from Lee et al. (2020). Figure 1: (a) Selected 2D class averages of the β 1 AR-βarr1 complex with the orientation of the complex indicated for two examples (red and blue boxes); (b) Overall structure of the β 1 AR-βarr1. The F ab fragment has been removed for clarity. Figure panels are reproduced from Lee et al. (2020).