technology feature
Alternative routes 930 The plot thickens 931 Without a mark 931 Making a match 932 Tracking a moving target 932 GPCRs: insane in the membrane Michael Eisenstein
They are the quintessential drug targetbut the dynamic structures and highly elaborate mechanisms of G protein–coupled receptors continue to keep experts in both industry and academia on their toes.
It is inevitable that any article about horrible name,” says Terry Kenakin of which also target GPCRs. “Our screen- G protein–coupled receptors (GPCRs) will GlaxoSmithKline’s Department of Assay ing is biased toward functional screens begin with a statistic reflecting the impor- Development and Compound Profiling. just because they’re higher-throughput, tance of these proteinswhich transmit “They’re seven-transmembrane domain although the false positive rate is high,” essential signals from a wide range of hor- receptors that bind a lot of other signaling says Roth. “Radioligand binding assays mones and neurotransmittersas drug proteins, and to just think of G proteins is tend to be a lot more specific, but obvi- targets. In fact, the numbers are striking: really limiting.” ously there’s information that can’t be 40–50% of all marketed drugs are thought gained from them.” to modulate GPCR activity. Equally strik- Intercepting signals The most mature functional assays are ing, however, is the fact that so many of On a positive note, this maturing under- those that quantify second-messenger these drugs were formulated with a lim- standing of GPCRs has enabled the devel- activity induced via G-protein activa- ited understanding of the deep complex- opment of increasingly sophisticated and tion. Activation is often accompanied by ity underlying GPCR biology. “There are high-throughput methods for receptor an influx of calcium ions, and a variety so many challenges in identifying high- characterization and drug discovery. With of sensitive assays are available for high- affinity, selective GPCR drugs with the more than 100 functionally opaque ‘orphan’ throughput screening of changes in Ca2+ desired efficacy that it’s amazing we have GPCRs and many more whose function is at in GPCR-expressing cells upon treatment anything at all,” says Brian Kobilka of best partially understood, researchers have with a compound or compounds of inter- Stanford University. their work cut out for them. est. PerkinElmer offers a pair of photo- The traditional model of GPCR func- For initial assessment of potential proteins that serve as effective indicators: © All rights reserved. 2009 Inc. Nature America, tion is complicated enough. Ligand bind- ligands, binding assays are still a typi- ing at the receptor extracellular domain cal first step, and the use of radiolabeled induces intracellular domain rearrange- ligand molecules remains the standard ments that allow a GTP-bound heterotri- approach for quantifying specificity. “You meric G-protein αβγ complex, which can can get some very precise measurements, comprise hundreds of individual subunit not just of the receptor expression density combinations, to bind. The G-protein but also on-off rates of ligand association complex then hydrolyzes GTP to GDP and and dissociation,” says Richard Eglen, dissociates into α and βγ subunits, each of president of Bio-discovery at PerkinElmer. which then activates distinct downstream As a less hazardous alternative for high- signal cascades. throughput screening, nonradioactive flu- This, however, pales in comparison to orescence-based platforms are also avail- the byzantine complexity that scientists able, such as PerkinElmer’s DELFIA and have subsequently uncovered regarding LANCE assays and the recently launched the remarkable flexibility of GPCR struc- Tag-lite system from Cisbio. ture and signaling activity, which is, in However, even a highly informative turn, closely dependent on cellular context binding assay needs to be followed up and interactions with activity-modulating by more detailed mechanistic studies, binding partners. Even their name is mis- points out Bryan Roth, whose team at the leading, as evidenced by the recent deter- University of North Carolina is dedicated mination that some GPCRs do not even to characterizing the specificity and func- Arthur Christopoulos and colleagues are use G proteins, leading some to renounce tional properties of both pharmaceutical investigating and designing molecules that the family moniker altogether. “It’s a compounds and drugs of abuse, many of modulate GPCR activity allosterically.
nature methods | VOL.6 NO.12 | DECEMBER 2009 | 929 technology feature
Eglen. The FLIPRTETRA instrument from MDS Analytical Technologies is suitable for high-throughput analysis of such O + 2 aequorin and Photina assays, but it also (Slow) enables fluorescence-based analysis via the widely-used FLIPR Calcium Flux Assay. However, this only covers a subset of receptor activities. “Calcium mobiliza- Coelenterzaine Apoaequorin Aequorin tion assays are a well-established meth- (Fast) od for Gq-coupled GPCRs, but Gi- and 3 Ca2+ Gs-coupled GPCRs have been tradition- ally assayed via cyclic AMP (cAMP) activ- ity,” explains David Yamane, senior direc- + Co2 ++ Light (466 nm) tor of drug discovery marketing for MDS Analytical, which offers the CatchPoint immunoassay for quantifying cAMP pro- duction. Promega’s GloSensor assay also Coelenteramide Apoaequorin PerkinElmer represents a sensitive and straightforward tool for detecting GPCR-induced increases Aequorin-based assays rely on calcium to facilitate the enzymatic production of coelentarimide, a in adenylate cyclase activity using an engi- bioluminescent compound. neered luciferase enzyme whose capacity to produce luminescence is directly acti- aequorin and Photina, both of which the presence of calcium. “They’re engi- vated by cAMP binding. “It [has] a fan- are jellyfish-derived enzymes that con- neered into the cell line, and they give a tastic dynamic range, and we can theo- vert the chemical substrate coelentera- very bright luminescent response as the retically monitor the cAMP signaling in zine into luminescent coelenteramine in intracellular calcium is elevated,” explains real-time,” says Arthur Christopoulos of Monash University.
Alternative routes As mentioned above, GPCR activity does not end with G protein–mediated signal- ing, and researchers who stop their inves- tigations at the second-messenger level are likely to miss the full story. For example, most activated GPCRs bind the protein β-arrestin, and although this was once © All rights reserved. 2009 Inc. Nature America, thought to be primarily a mechanism for restricting receptor activation, it is now clear that arrestins trigger signal cascades of their own. “You can inhibit G proteins altogether, and you’ll still get arrestin activation,” says Tom Wehrman, director of cell biology at DiscoveRx. “And there are actually two GPCRs, CXCR7 and C5L2, that don’t actu- ally couple to any G protein but do activate arrestin.” DiscoveRx offers the PathHunter assay, which monitors receptor-arrestin interaction via enzyme fragment comple- mentation1. In this approach, one segment of the β-galactosidase enzyme is linked to the GPCR of interest and the other is attached to β-arrestin, and only when the two interact is full enzymatic activity restored, enabling processing of a biolumi- nescent substrate. “The main advantage is MDS Analytical Instruments Analytical MDS that you’re actually monitoring the bind- The FLIPRTETRA instrument from MDS Analytical Instruments enables high-throughput screening of ing event: it’s a stoichiometric readout calcium flux for GPCR studies. on receptor activation,” says Wehrman.
930 | VOL.6 NO.12 | DECEMBER 2009 | nature methods technology feature
Invitrogen also uses a fusion protein sys- across cell monolayers cultured atop elec- tem in their Tango arrestin assay, with a trodes and changes in the ratio of voltage protease-tagged arrestin that specifically to current across that monolayera prop- cleaves a transcription factor fused to the erty known as impedanceresulting from target receptor upon binding, releasing drug-induced rearrangements in cellular it to enable activation of a downstream organization are tracked. “Changes in cell reporter. adherence, shape, volume, and cell-cell Other assays are also emerging that interactions that occur upon exposure of gauge more pathway-independent cells to a stimulus, such as a ligand, con- responses to receptor activation, such as tribute to the impedance signal,” explains MAP kinase activity or receptor endocy- Yamane. “These factors, individually or tosis, but ultimately, experts agree that the collectively, affect the flow of current key to success lies in effectively combining and thereby influence the magnitude and readouts (for example, cAMP and Ca2+ or characteristics of the impedance signal.” G protein–dependent and –independent Corning’s EPIC system implements an activation) to overcome the biases inher- alternative, optical approach using a reso- ent to any assay-based screening strategy nant waveguide biosensor, which quanti- Corning’s EPIC system tracks changes in the and derive a more complete overall pic- fies alterations in the refraction index at distribution of molecules at the cell surface as ture of receptor activity. “You must use an indicator of GPCR-triggered cellular activity. the membrane-sensor interface of cul- multiple functional endpoints,” asserts Courtesy of Y. Fang, Corning. tured cells. In the context of GPCR screen- Christopoulos. “In my lab, nobody can ing, these changes result from molecular ask any questions without doing at least to yield mature cells that are appropriate rearrangements induced by receptor acti- three functional assays and one binding for the study of a given GPCR under nor- vation. “Cell signaling, particularly medi- assay where possible.” mal or disease-specific conditions. In the ated through receptors, often involves end, however, animal models remain the protein trafficking, microfilament remod- The plot thickens ultimate goal. “If you keep introducing eling, cell adhesion alterations and mor- Even with many approaches available, complexities, you’re going to get a lot of phological changes of cells, all of which uncovering the true nature of receptor interesting observations but you’ll proba- can lead to [substantial] dynamic mass activity is a challenge. The same receptor bly go down mechanistic paths that aren’t redistribution,” explains Ye Fang, senior may not only activate different down- relevant,” says Christopoulos. “You’re bet- research manager at Corning. “Such stream pathways in response to different ter off taking a small, judicious sampling redistribution is not random; instead, it ligandsa phenomenon known as ‘func- of compounds in vivo sooner and then is tightly regulated and is often dynamic tional selectivity’ or ‘biased agonism’but coming back to deconvolute.” both spatially and temporally, and the might even respond differently to the same Roth, whose team has developed cell biosensor simply acts as a noninvasive ligand depending on the cellular context. lines for screening purposes that express monitor to record [this] in real time.” This complexity is exacerbated by the nearly 250 different GPCRs, concurs. On one hand, label-free platforms offer © All rights reserved. 2009 Inc. Nature America, existence of allosteric modulators, which “There are so many ways to be misled by multiple advantages, enabling screening of bind outside the active site but can pro- using an overexpressed receptor in a non- virtually any cell type that can be success- foundly alter receptor ligand specificity or native cell line,” he says. “If we make a dis- fully cultured in real time and in a path- kinetics. “If you put an allosteric modula- covery, we try as best we can to validate it way-independent manner. “[Label-free tor on the receptor, you know you’re going in a native tissue and then preferably with systems] can act as a single platform for to make a new conformation, and so you a wild-type and a knockout mouse.” all classes of GPCRs,” says Fang. However, could engender functional selectivity even some screeners are skeptical about how with agonists that don’t normally have it,” Without a mark effectively these data can be interpreted. says Christopoulos. “We have examples of A newer alternative that has given hope to “They’re really what most of us would pathway-biased allosteric modulation.” some are the so-called ‘label-free’ screen- consider to be ‘black box’ readouts,” says Such subtleties can potentially be lost in ing technologies, which are designed to Roth. “All you can basically say is that you an overly engineered and artificial assay sys- enable screening of receptor activity in added a drug and something happened to tem. “In many cases, you identify a GPCR as virtually any cell line or primary cell cul- the cell.” a target based on physiological data, and the ture, without the need for receptor engi- In some cases, this may save time and receptor might be expressed in the brain in neering or other genetic modification. resources. “When you do a cyclic AMP a particular neuron, but then you perform “That’s going to be the coming thing: assay, you know what’s happening there a high-throughput screen in over-express- we’ll re-emerge into the old way of doing but you don’t know what else is happen- ing, immortalized cell lines that are nothing things, in real time with human tissues ing; I don’t know if that’s any better than like the cell in which the receptor normally and primary cells,” says Kenakin. “Tissues getting a generic response and not under- resides,” says Kobilka. that are not healthy but actually model the standing exactly what’s happening,” says One potential solution might be the illness we’re trying to cure.” Kenakin. “There are tools you can use to directed differentiation of embryonic In the CellKey system from MDS get at that, like [small interfering] RNA stem or induced pluripotent stem cells Analytical a constant voltage is applied or enzyme inhibitors.” For some scientists,
nature methods | VOL.6 NO.12 | DECEMBER 2009 | 931 technology feature
Brian Shoichet’s team at the University of (Rescriptor) to the histamine H4 GPCR. California at San Francisco is at the fore- Shoichet believes this is only the begin- front in this regard, having recently applied ning of what promises to be a fruitful, their molecular docking algorithm, DOCK, parallel avenue for drug research. “I think to the Kobilka group’s β2-adrenergic recep- that going back to the classical pharma- tor structure to identify potential ligands cology view of typing receptors by the from among one million commercially ligands that bind to them as an overarch- available compounds5. “By our stan- ing idea is a game-changer,” he says. dards, it was an unbelievable success,” says Shoichet. “We found that about 25% of the Tracking a moving target molecules we predicted to bind, did bind, With so many technological resources at which is about five- to tenfold better than hand for functional analysis and ligand dis- we do with non-GPCR sites. The best hit covery, most scientists working with GPCRs we got was a nine nanomolar inverse ago- agree that aggressive pursuit of structural nist, … and it turns out that this is the best data is now a top priority, although it would inverse agonist that’s been characterized to be a mistake to assume that even ‘solved’ date.” Members of the Christopoulos lab structures can be considered a closed book. are likewise gaining momentum in their ab “I think the binding pocket is going to dif- initio drug design efforts, using the steady fer for different active states of the recep- increase in available structural data to facili- tor, so you have to have other structures,” Recent successes in GPCR crystallization, tate the design of allosteric modulators that says Kobilka, adding that ongoing investi- including this structure of the β2-adrenergic specifically target peptide-binding GPCRs. gations with nuclear magnetic resonance receptor from Brian Kobilka and colleagues, are “Transmembrane regions are where the (NMR) spectroscopy in his and other labs helping propel the field forward. Courtesy of crystal structures are getting quite good, … may complement crystallography and help B. Kobilka. and we believe those will be novel allosteric capture some of the more dynamic details sites ripe for targeting,” he says. of receptor behavior. “The methodology though, this technology is simply still too Shoichet and Roth have also been work- and instrumentation for high-resolution new and for now, too expensive for any ing together to explore the potential of structure determination of GPCRs by NMR categorical assessment. “I don’t know how applying computational methods to a [spectroscopy] aren’t quite there yet, but far it’s going to go in telling us things, … more ‘classical’ pharmacological analysis, NMR [spectroscopy] can provide a lot of but I don’t want to be too negative,” says finding targets that match a given drug interesting information about the ligand- Christopoulos. “Once things settle down, rather than the other way around. In a induced movement of specific domains we’ll see where this ends up.” recently published study6, they analyzed and the dynamics of different parts of the thousands of pharmaceutical compounds, protein,” he says. Making a match comparing them chemically to known In the meantime, every new structure The floppy, complex structure of GPCRs receptor ligands and using those similari- helps. “There are still so few that each one © All rights reserved. 2009 Inc. Nature America, has made them exceedingly difficult to ties to reveal new information about drug is really going to affect a lot of people,” crystallize; for many years, the structure specificity and potential off-target effects. says Shoichet. “It really brings the field a of rhodopsin was the only high-resolution According to Shoichet, the results revealed big bolt of energy.” structure available, making direct chemi- insights that would have been overlooked 1. Wehrman, T., Kleaveland, B., Her, J.H., Balint, cal screening the only game in town for by conventional genomic or proteomic R.F. & Blau, H.M. Proc. Natl. Acad. Sci. USA drug discovery. “For a virtual library analysis. “Things that look unrelated by 99, 3469–3474 (2002). screen, you have to get the contacts exactly sequence look highly related by the ligands 2. Warne, T. et al. Nature 454, 486–492 (2008). 3. Cherezov, V. et al. Science 318, 1253–1254 right, and so you’re looking at less than they bind,” he says. “For instance, seroton- (2007). two-angstrom resolution in the binding ergic receptors have seven major subtypes; 4. Jaakola, V.P. et al. Science 322, 1211–1217 pocket,” says Christopoulos. “Most virtu- serotonin 3 is an ion channel, and all the (2008). al screens that aren’t based on real crystal others are GPCRs; … their sequences are 5. Kolb, P. et al. Proc. Natl. Acad. Sci. USA 106, 6843–6848 (2009). structures fail.” completely different, but they’re recog- 6. Keiser, M.J. et al. Nature advance online This is starting to change, with recent nizing similar molecules.” Their findings publication, doi:10.1038/nature08506 2 (1 November 2009). structures for the β 1-adrenergic , also enabled them to identify the physi- 3 4 β2-adrenergic and A2A adenosine recep- ological basis for known drug side-effects, tors potentially heralding accelerated prog- such as the capacity of fluoxetine (Prozac) Michael Eisenstein is a freelance science ress in GPCR crystallography and thus new to act as a beta-blocker and unexpected journalist based in Brooklyn, New York, opportunities for in silico drug discovery. affinity of the antiviral drug delavirdine USA ([email protected]).
932 | VOL.6 NO.12 | DECEMBER 2009 | nature methods technology feature
Suppliers guide: companies offering reagents and hardware for GPCR -based research Company Web address Applied Biosystems http://www.appliedbiosystems.com/ Assay Designs http://www.assaydesigns.com/ Axxam http://www.axxam.com/ BD Biosciences http://www.bdbiosciences.com/ BioFocus DPI http://www.biofocusdpi.com/ BioTek http://www.biotek.com/ BMG Labtech http://www.bmglabtech.com/ Caliper Life Sciences http://www.caliperls.com/ Cambridge Bioscience http://www.bioscience.co.uk/ CEREP http://www.cerep.fr/ ChanTest http://www.chantest.com/ Cisbio http://www.htrf.com/ Corning http://www.corning.com/ DiscoveRx http://www.discoverx.com/ GE Healthcare http://www.gehealthcare.com/ Geneservice http://www.geneservice.co.uk/ GenScript http://www.genscript.com/ Hamamatsu http://www.hamamatsu.com/ Heptares Therapeutics http://www.heptares.com/ Invitrogen http://www.invitrogen.com/ Meso Scale Discovery http://www.meso-scale.com/ Millipore http://www.millipore.com/ Molecular Devices (MDS Analytical) http://www.moleculardevices.com/ Multispan http://www.multispaninc.com/ patoBios http://www.patobios.com/ PerkinElmer Life Sciences http://las.perkinelmer.com/ Promega http://www.promega.com/ R&D Systems http://www.rndsystems.com/ Roche Applied Science http://www.roche-applied-science.com/ TGR Biosciences http://www.tgr-biosciences.com.au/ © All rights reserved. 2009 Inc. Nature America, Thermo Scientific http://www.thermo.com/
nature methods | VOL.6 NO.12 | DECEMBER 2009 | 933