April 15, 2007 (Vol. 27, No. 8)
Susan Aldridge, Ph.D.
New Technologies and Approaches Facilitate the Discovery of Ligands
G protein-coupled receptors (GPCRs) are transmembrane proteins that transduce an extracellular signal—ligand binding—into an intracellular signal—activation of G-protein—which in turn activates various key cellular pathways. GPCRs represent the largest protein family in the human genome and play a vital role in many biochemical and physiological processes, from cell-to-cell communication to the regulation of mood and behavior.
It is hardly surprising that GPCRs have long been an important target in drug discovery, initially in cardiovascular and cancer indications, but increasingly also in metabolic disorders, such as diabetes.
Next month, Informa Life Sciences is sponsoring a conference in Lisbon that sums up the latest developments in GPCR-based drug discovery. GEN talked to a number of scientists who will be making presentations at the Lisbon meeting.
PerkinElmer Life and Analytical Sciences (www.las.perkinelmer.com) has a well-established range of both ligand-binding and cell-based functional GPCR assays. The company is now focused upon developing and expanding these assays in line with advances in the understanding of these targets.
“People are looking at identifying allosteric regulators of GPCRs—this is now an important concept,” says Richard Eglen, Ph.D., vp and general manager, discovery and research reagents. “There is also interest in modulating constitutive GPCR activity using inverse agonists, and due to the fact that GPCRs function as assemblies, such as dimers,” researchers are also beginning to study several new GPCR signaling pathways, including their role in intracellular kinase pathways.
Consequently, PerkinElmer is now applying its bead-based AlphaScreen® technology to a new high-throughput assay of phosphorylation by ERK1/2, kinases that act as signal transduction mediators in many of the pathways regulated by GPCR. This cell-based, homogeneous assay thus expands the range of HTS techniques available to identify GPCR ligands and detect potential differences in GPCR pharmacology.
PerkinElmer’s recent acquisitions of Euroscreen Products and Evotec Technologies further extends the company’s range of GPCR-screening technologies. Evotec brings the Opera ™ HCS platform, which provides high-content analysis with the throughput required for primary and secondary screening. This new tool will allow the study of protein-protein interactions, including GPCR dimerization.
Euroscreen brings in the AequoScreen™ cell-based GPCR-screening platform, which is based upon a luminescence technology and will be applicable to the HTS detection of GPCR ligands. The large assay window notably allows the detection of allosteric regulators of GPCRs.
“PerkinElmer has long had all the classical GPCR-screening tools, such as radiolabeled binding and cell-based cAMP assays,” says Dr. Eglen. “With these new technologies, we can now look at emerging aspects of GPCRs as well, collectively providing the largest instrument and reagent solutions available for research in these important drug discovery targets.”
The Euroscreen acquisition also allows PerkinElmer to offer a large range of GPCRs, in either cells or membranes. This is particularly important to researchers wishing to profile the action of their novel drugs across the GPCR superfamily.
Today Euroscreen pursues a dual program—developing assays and screening technologies, such as AequoScreen™ for GPCRs and also working on an internal pipeline of drugs against GPCR targets.
Orphan GPCRs, for which ligands and function remain unknown, are attractive potential drug targets (there are over a hundred in the human genome waiting to be explored) and the company uses AequoScreen to help identify relevant ligands.
For instance, it is currently working on various leucocyte chemoattractant GPCRs, such as ChemR23, which is found in dendritic cells, macrophages, and NK cells. Euroscreen has succeeded in isolating ChemR23’s natural ligand, chemerin, from human inflammatory fluids.
The role of these GPCRs in human disease is being explored, although more complex animal models will be needed because there are no obviously homologous genes in the mouse, so it is not possible to create knockouts. Meanwhile, chemical libraries are being screened for leads against these new GPCRs for both target validation and drug development.
Meanwhile, DiscoveRx (www.discoverx.com) provides a range of assay solutions for GPCRs and other important drug targets, all based upon its enzyme fragment complementation (EFC) assay technology. EFC is a new way of measuring protein-protein interactions. An enzyme acceptor fragment is fused to one protein, an enzyme donor fragment to another protein. If the two proteins bind, they make up a-galactosidase, which hydrolyzes a substrate that produces a chemiluminescent or fluorescent signal.
The company has now applied EFC to its new b-arrestin assays for GPCR binding, which are lead products in the PathHunter™ cell-based assay portfolio. Arrestins are key players in the desensitization of most GPCRs after ligand binding. Keith Olson, Ph.D., vp of product and market development at DiscoveRx, explains how GPCR stimulation triggers G-protein dissociation and GPCR kinase phosphorylation. The latter, which only phosphorylates active GPCRs, in turn triggers the recruitment of arrestin from the cytoplasm to the membrane. Here it binds to the phosphorylated C-terminus of GPCR, causing its internalization for either recycling or degradation. “This is a generic system applicable to virtually all GPCRs,” he says.
The DiscoveRx cell-based assay involves the fusion of the enzyme acceptor to b-arrestin and fusion of the donor (known as the ProLink) to the GPCR. When the arrestin binds to the GPCR, complementation between the enzyme acceptor and the ProLink occurs, leading to generation of a readout signal. Dr. Olson explains that this new format offers a number of advantages: generic implementation for any GPCR, reduced interference in comparison to second messenger and reporter gene assays, and greatly improved ease of use compared to imaging approaches.
“The assay is applicable to primary screening, secondary screening, selectivity profiling, and deorphanization of novel GPCRs,” he says. Currently the b-arrestin assay is applicable in around 70 GPCR-driven cell lines, and this will be extended to cover more than 200 GPCR targets over the next few months.
Virtual GPCR Testing
Finally, at Bayer Schering Pharma (www.schering.de), the computational chemistry group is validating a virtual approach, called machine learning, to speed the discovery of GPCR ligands and also to learn more about the druggability of GPCR targets. Machine learning is a type of artificial intelligence that can extract rules and patterns from massive data sets, using data mining and statistical techniques.
According to Antonius Ter Laak, Ph.D., this new approach can transfer existing knowledge out of the data set and into the drug discovery process by generating rules of more general meaning than the more conventional ligand discovery approaches. These rules can then be used across various projects. Bayer Schering has become interested in machine learning over the last three years and has, for example, developed several ADME property prediction tools in a collaboration with the Fraunhofer Institute.
When applied to GPCRs, the machine-learning dataset contains GPCR ligand information in the form of highly mathematical molecular descriptors and, currently covers around 30,000 ligands known to bind to over 200 different GPCR targets. Dr. Ter Laak adds that the best way of using the results of the machine-learning process is as a filter for a large ligand library before synthesis—so that molecules unlikely to bind are not synthesized.
Many different types of molecules act as GPCR ligands—it is really not possible just to point to specific or obvious features of a molecule that determine whether it will bind or not. The Bayer Schering Computational Chemistry group has found that there is a combination of subtle properties that make a compound more likely to bind to the binding pocket of GPCRs. “What we get from machine learning is a rough measure of GPCR ligand likeness,” explains Dr. Ter Laak.
Bayer Schering has also been using a chemogenomics approach to learn more about new orphan GPCR targets. “We know that certain ligands bind to subfamilies of GPCRs,” says Dr. Ter Laak. “Now we want to learn which type of molecule will bind to a particular subfamily.” GPCRs are notoriously difficult to crystallize and most knowledge is based upon the crystal structure of rhodopsin and its retinal binding site. “In several cases, based on transmembrane binding site properties, we have been able to find close neighbor GPCRs with known endogenous or synthetic ligands,” says Dr. Ter Laak.
They now have a database of important binding site residues of all the GPCRs in the human genome. With this annotated database of 30,000 GPCR ligands, it has been possible to start asking additional questions about features that link sequence space with ligand space.