Corresponding author S. Herlitze: Allgemeine Zoologie und Neurobiologie, Ruhr-Universität Bochum, Bochum D-44780, Germany. Email: email@example.com
G-Protein-coupled receptors (GPCRs) integrate extracellular cues into intracellular signals to modulate the cellular state. Owing to their diverse modulatory functions, GPCRs represent one of the major drug targets of the pharmaceutical industry. Until now, the characterization and control of GPCRs and their intracellular signalling cascades have mainly relied on chemical compounds, which either activate or inhibit GPCR pathways, albeit with limited receptor and cell-type specificity. Recently, new approaches have been developed to control signalling cascades in cell- and receptor-type-specific ways. The chemical approach focuses on GPCR design and activation by an inert chemical compound, whereas the physical approach uses designer GPCRs and activation by physical stimuli, such as light.
While action potentials determine and encode information within the millisecond to second time range, the cellular state is modulated over long time periods, in particular by G-protein-coupled receptors (GPCRs) that activate or inhibit intracellular signalling pathways.
The GPCRs consist of seven transmembrane domains (TMs), which are connected via intra- and extracellular protein domains. The TMs and the extracellular protein domains are involved in ligand binding and activation of the receptors, whereas the four intracellular domains, which connect TM1–2 (i1 loop), TM3–4 (i2 loop), TM5–6 (i3 loop) and the C-terminal domain (CT), are involved in receptor trafficking and subcellular localization, as well as selective coupling to a G-protein. Three main GPCR-coupled pathways can be distinguished, i.e. Gi/o, Gq and Gs. The Gi/o pathway modulates, for example, ion channel conductances, such as Ca2+ and K+ channels, while the Gq pathway mediates Ca2+ release from intracellular stores and activates ion conductances, such as transient receptor potential channels. The Gs pathway in particular uses cAMP as its second messenger.
Disturbances within these intracellular signalling pathways are often responsible for the development and manifestation of neuropathological disease states. Therefore, the need for new tools to dissect intracellular signals and to understand their involvement in behaviour modulation has given rise to the development of designer GPCRs. Newly developed designer GPCRs can be categorized as those activated by inert ligands and those activated by light. The main focus of this review is the development of light-activated GPCRs and their comparison with chemically gated receptors.
Chemically activated GPCRs: RASSLs, DREADDS and the AlstR/AL system
The GPCRs belong to one of the largest protein superfamilies in the human genome and therefore stimulate a wide variety of signalling pathways. According to their second-messenger system they are categorized into four families: Gi/o, Gs, Gq/11 and G12 (Wettschureck & Offermanns, 2005). The structural similarities and diversity of GPCRs make it difficult to specifically modify one particular GPCR subtype or one individual second-messenger system in vivo. Pharmaceutical compounds, such as chemical agonists and antagonists, allow for the control of GPCRs with more or less specificity (Nichols & Roth, 2009). To selectively control GPCR signalling pathways, novel designer receptors had to be developed.
RASSLs. Engineered receptors, named RASSLs (receptors activated solely by synthetic ligands), lack the ability to respond to their endogenous ligands and are instead activated by synthetic compounds. The first attempt to design a RASSL was made by Strader et al. (1991) based on the Gi-coupled β2-adrenergic receptor (β2AR). Site-directed mutagenesis experiments indicated that Asp113 was the crucial ligand-binding site for this receptor and suggested that substitution of the Asp113β-carboxymethyl side-chain with a β-hydroxymethyl side-chain of a Ser residue would result in a mutant receptor, which would be activated by compounds capable of forming hydrogen bonds. Indeed, this RASSL prototype was insensitive to its endogenous ligand and could be activated by ketone ligands and catechol esters, both of which have no affinity for the wild-type receptor (Strader et al. 1991). Nevertheless, due to a low potency (half-maximal effective concentration, EC50= 118 μm), low receptor affinity and unknown pharmacokinetics, the novel system has limitations for in vivo applications.
The next generation of RASSLs, developed in particular for in vivo applications, were based on the human κ-opioid receptor. The Gi/o-coupled RASSL Ro1 (RASSL based on opioid receptor no. 1) can be activated by nanomolar concentrations of synthetic small-molecule agonists, such as spiradoline, and posseses decreased binding affinity and signalling response to opioid peptides (Coward et al. 1998). Modified RASSLs derived from Ro1 have been expressed in at least six tissues of transgenic mice in order to analyse the effects of Gi signalling pathway activation (Redfern et al. 1999, 2000; Mueller et al. 2005; Sweger et al. 2007; Guettier et al. 2008).
DREADDs. Despite the advantages of RASSLs for GPCR effector pathway research, the system still exhibits some problems. Binding affinity experiments indicate x-fold lower affinity values for RASSLs to their native ligands, but it is not clear to what extent the remaining affinity for the ligand affects the GPCR activity for in vivo applications. Furthermore, it is not clear whether the synthetic ligands might influence other proteins in the same neuronal circuit. Therefore, another generation of RASSLs has been developed. These GPCRs, termed DREADDS (designer receptors exclusively activated by designer drugs), are solely activated by a synthetic, biologically inert drug (Armburster et al. 2007). The first DREADD was a member of the human muscarinic receptor family. Two amino acid exchanges (Y149C3.33/A239G5.46) within the human muscarinic ACh receptor M3 (hM3), a Gq-coupled GPCR, abolished the affinity for the native ligand ACh but allowed activation by the small drug-like molecule clozapine-N-oxide (CNO; Armburster et al. 2007). Clozapine-N-oxide is a pharmacologically inactive molecule with no binding affinity to any class of receptor (Bender et al. 1994). The hM3 receptor indicated that a few mutations are sufficient to create a DREADD that ultimately favours the artificial over the native ligand, thereby facilitating the investigation of GPCR function in vivo and in vitro. With this knowledge, it was also possible to create a DREADD human muscarinic ACh 4 (hM4D), a Gi/o coupled receptor, which was potent enough to activate inwardly rectifying potassium channels (GIRKs) in transfected human embryonic kidney cells (HEK cells) and neuronal hippocampal cells (Armburster et al. 2007). Subsequently, a whole family of designer muscarinic receptors, hM1–5D, were created and found to be exclusively activated by CNO and not by ACh or carbachol, clearly following the initial efforts of designer engineering of GPCRs. Thus, DREADDS cover the major G-protein signalling pathways, Gq (Armburster et al. 2007), Gi/o (Armburster et al. 2007) and Gs (Guettier et al. 2009).
The AlstR/AL system. Another genetic approach for manipulating neuronal activity has been developed by Lechner et al. (2002). Briefly, an allatostatin receptor–allatostatin (AlstR/AL) system was used to quickly and reversibly silence ferret cortical neurons in vitro. The allatostatin receptor is a Drosophila GPCR involved in the regulation of juvenile hormone synthesis in insects. It is activated by the ligand allatostatin, but does not respond to similar ligands of related mammalian receptors (Birgul et al. 1999). The allatostatin receptor has been shown to activate mammalian GIRK channels via the Gi/o pathway in Xenopus oocytes (Birgul et al. 1999) and has been used for silencing neurons in different in vitro and in vivo preparations (Tan et al. 2006). Certain disadvantages are associated with the AlstR/AL system, which include temporal resolution and the removability of AL from specific tissues, owing to limited diffusion characteristics (Tan et al. 2006).
In summary, RASSLs, DREAADs and AlstRs represent powerful tools in probing G-protein receptor signalling. The advantage of designer GPCRs is to selectively activate individual GPCR subtypes without side-effects. However, the overall spatiotemporal resolution is low in vivo because of slow application kinetics, washout and ligand degradation.
Light-activated GPCRs: ChARGe, vRh and OptoXR
To overcome limitations of application and washout of chemical compounds in studies that examine GPCR function in vivo, new tools had to be developed. The idea was to activate GPCRs by stimuli that penetrate cells and tissues, such as sound, magnetic fields or light.
ChARGe. The chARGe system was the first approach to control GPCR pathways by light (Zemelman et al. 2002). Essentially, the activation of invertebrate rhodopsins might allow depolarization of cells outside the Drosophila eye, as invertebrate rhodopsins couple to the Gq/11 class of G-proteins and ultimately open endogenous non-selective cation channels. By expression of 10 proteins belonging to the Drosophila visual phototransduction cascade, three structural components were identified. The name chARGe refers to the required proteins: the Drosophila rhodopsin NinaE, the αq subunit of the corresponding G-protein and arrestin-2. Photostimulation of Xenopus oocytes expressing chARGe evoked currents and revealed that only initial loading with retinal is required to activate Gq pathways. ChARGe was further tested in hippocampal neurons, and illumination with 400–600 nm reliably elicited action potential firing with latencies of the order of seconds. Since the system consists of three separate genes, chARGe has never been tested in vivo.
RO4 and vRh-CT. Based on the idea that ion channels such as presynaptic Ca2+ channels and postsynaptic GIRKs are modulated by the Gi/o pathway in a membrane-delimited manner and that modulation occurs within the millisecond to second time scale, vertebrate rhodopsin (vRh) has successfully been used to control these signalling pathways in neurons (Li et al. 2005; Oh et al. 2010). In the vertebrate eye, vRh signals through the G-protein transducin, which belongs to the Gi subfamily. The system was established using the rat rhodopsin 4 (RO4) (Li et al. 2005) and was found to inhibit presynaptic Ca2+ channels and activate GIRK channels in HEK293 cells when stimulated by 475 nm light (Li et al. 2005; Fig. 1A). After expresssion in hippocampal neurons, the green fluorescent protein-tagged RO4 hyperpolarizes the cell membrane, reduces action potential firing somatodendritically and decreases synaptic transmitter release presynaptically (Fig. 1A). Expression and light activation in the embryonic chicken spinal cord leads to synchronization of neuronal network activity, a process involved in axon path-finding and synapse formation. Thus, RO4 is a useful tool to control neuronal function pre- and postsynaptically on a millisecond to second time scale via unselective activation of the Gi/o pathway within subcellular structures.
The next generation of light-activated receptors developed allow for the precise control of a GPCR signalling pathway within a subcellular structure and in a GPCR type-specific way. The C-termini (CT) of many GPCRs contain protein domains for subcellular targeting, G-protein binding and selectivity. Therefore, tagging vRhs, such as RO4, with the CTs of GPCRs should result in GPCR subtype-specific anchoring. Oh et al. (2010) demonstrated that tagging vRh with the CT of the 5-HT1A receptor (Rh-CT5–HT1A), a GPCR important for the regulation of the serotonergic transmitter system and a drug target for anxiety and depression, resulted in a functional light-activated GPCR that is also targeted in 5-HT1A-specific manner in hippocampal neurons and serotonergic neurons from the dorsal raphe nucleus (Fig. 1B). In the presence of Rh-CT5-HT1A, but not Rh alone, the endogenous 5-HT1A receptor response was reduced, suggesting that Rh-CT5-HT1A competes with the 5-HT1A receptor for subcellular targeting and anchoring slots in the membrane of neurons. The Rh-CT5-HT1A could also rescue 5-HT1A receptor-like responses in 5-HT1A receptor knock-out mice and control firing of neurons within the dorsal raphe nucleus (Fig. 1B).
OptoXRs. The possibility of exchanging the intracellular loops of vRh to produce light-activated chimeric receptors had recently been demonstrated by Dr Khorana's group (Kim et al. 2005). His group replaced the intracellular loops of the vRh with the β2-adrenergic receptor loops and converted vRh into a light-activated Gs-coupled β2-adrenergic-like receptor (Kim et al. 2005). Using the same chimeric approach, an α1a-adrenergic receptor like opsin-GPCR was engineered to control the Gq pathway (Airan et al. 2009). The Gs- and Gq-coupled chimeric vRhs were designated optoXRs. Expression and light activation of the optoXRs in HEK cells demonstrated their functionality and the specific activation of different signalling pathways. The resulting elevated cAMP levels from the optical stimulation of opto-β2AR were comparable with pharmacological stimulation of the wild-type β2AR. Likewise, activation of opto-α1AR resulted in an upregulation of inositol trisphosphate, similar to levels achieved in pharmacological studies. Additionally, the expression of the Opto-β2R in adult neurons of the nucleus accumbens caused a decrease in network firing during optical stimulation in brain slices, whereas activation of opto-α1AR increased network activity. The light-control of optoXRs was then used to modulate the behaviour of freely moving mice. Mice expressing opto-α1AR in accumbens neurons were conditioned by delivery of light pulses in a place preference assay. Indeed, photostimulation was sufficient to drive the place preference as expected for the activation of this signalling pathway.
The future of light-activated GPCRs: general considerations
Vertebrate versus invertebrate rhodopsin. The existing light-activated GPCR systems (ChARGe, RO4, OptoXRs and Rh-CT) have been developed by using either vRh or invertebrate rhodopsin (iRh). Both of these photoreceptor classes are comprised of many different members, which can be activated by different wavelengths of light. The photoreceptors can be categorized into three groups, with peak sensitivities for ultraviolet light [300–400 nm (UV)], blue light [400–500 nm (B)] and long-wavelength light [500–600 nm (LW)]. This provides the possibility of choosing from three pigments to activate differentially and/or simultaneously three different GPCRs by UV/blue, cyan/green and yellow/red light. However, iRh and vRh differ fundamentally in their photocycle. In vRh, light induces the isomerization from 11-cis-retinal to all-trans-retinal, which in turn induces a conformational change in the bound opsin and activates the coupled G-protein. The restoration of functional rhodopsin in vertebrates is mediated by isomerases in the nearby tissue, which are able to recycle all-trans-retinal to 11-cis-retinal. Therefore, a steady supply of retinal has to be guaranteed in heterologous expression systems to gain functional receptors (Brueggemann & Sullivan, 2002). Interestingly, in cultured hippocampal neurons, brain slices, spinal cord preparations or in vivo in rats or mice, an additional supply of retinal is not necessary (Li et al. 2005; Zhang et al. 2006; Airan et al. 2009).
In contrast, iRh possesses an intrinsic photocycle, which enables the receptors to restore 11-cis-retinal by a photochemical reaction (Byk et al. 1993). Therefore, retinal is only needed for the initial loading of rhodopsin with its chromophore (Zemelman et al. 2002), which has advantages for in vivo applications in comparison to vRh. The present disadvantage of iRh is that three proteins of the invertebrate visual system are required to be expressed to reconstitute functional Gq signalling in vertebrates. The potential to solve this problem by engineering new iRh has to be investigated.
Signalling specificity. A critical factor for the proper function of light-activated GPCRs for control of intracellular signalling is that their activation kinetics should resemble kinetics of cell-type-specific signalling cascades. In addition, the receptor should be trafficked and localized to the area, where GPCR signalling can be controlled. Indeed, Rh-CT5-HT1A replaces endogenous 5-HT1A receptors and substitutes their function in knock-out mice (Oh et al. 2010). It is important that endogenous GPCRs can be replaced by the engineered GPCRs, so that interference with endogenous receptors can be minimized. The functional substitution of endogenous receptors by Opto-GPCRs will offer the opportunity to correlate different receptor signalling pathways with modulation of behaviour without creating knock-out animals for each receptor. In combination with viral delivery systems and cell-type-specific promotors to target Opto-GPCRs into the tissue and cell type of choice, these designer GPCRs may guide therapeutic developments to treat disease states.
Several genetic approaches have now been developed to control neuronal activity and neuronal signals. The choice between one or the other depends on the application and any spatiotemporal resolution requirement. Pharmacological control via GPCRs has low temporal resolution, since compound diffusion and removal are slow or impossible. However, DREADDs, for example, have the advantage that only small structural changes within the ligand-binding side are sufficient to convert the receptor, and intracellular domains necessary for signal transduction remain intact. Conversely, light-based methods offer a high spatiotemporal resolution. Since Rhs can be activated by different wavelengths, various Rhs can be combined and expressed in one or more cell types to control the Gi, Gq and Gs pathways simultaneously or separately.
This work was supported by the Deutsche Forschungsgemeinschaft (HE2471/8-1) and by the National Institutes of Health (MH081127). We would like to thank Davina Gutierrez for helpful comments and proofreading.