|Home | About | Journals | Submit | Contact Us | Français|
We are creating families of designer G-protein-coupled receptors (GPCRs) to allow for precise spatiotemporal control of GPCR signaling in vivo. These engineered GPCRs, called receptors activated solely by synthetic ligands (RASSLs), are unresponsive to endogenous ligands but can be activated by nanomolar concentrations of pharmacologically inert, drug-like small molecules. Currently, RASSLs exist for the three major GPCR signaling pathways (Gs, Gi, Gq). These new advances are reviewed here to help facilitate the use of these powerful and diverse tools.
G-protein-coupled receptors (GPCRs) are an ideal vehicle for engineering synthetic signaling systems. These receptors function as signaling switches throughout the body and regulate virtually every physiological response1-3. GPCRs are also the largest gene family targeted for drug discovery4. GPCRs stimulate a variety of G protein pathways; for example, Gs stimulates cAMP production, Gi inhibits cAMP production, and Gq stimulates phospholipase C and releases intracellular calcium stores. Since GPCRs have a relatively simple modular design and are encoded by small genes (usually <1.5 kb), engineered GPCRs can be easily transferred without loss of their functionality into different tissues and species. Finally, designer GPCRs could be useful for regulating physiologic processes and engineering tissues with stem cells and other technologies.
Attempts to engineer GPCRs that are activated solely by pharmacologically inert drug-like molecules have met with varying degrees of success (Table 1). Engineered receptors and engineered receptor/ligand pairs have been created by several approaches and bear different names—receptors activated solely by synthetic ligands (RASSLs)5, therapeutic receptor-effector complexes (TRECs)6, neoceptors7, and designer receptors exclusively activated by designer drugs (DREADDs)8. Here we will refer to them as RASSLs, and we propose a consensus nomenclature (Table 1) for those in widespread use or in development. This nomenclature links the name of the parent receptor to the major G-protein signaling pathway activated by the receptor.
In the first attempt to make a designer GPCR, Strader and colleagues developed a series of compounds to selectively activate a mutant version of the ••-adrenergic receptor (••-AR) that was unresponsive to its natural hormone9. This group focused on D1133.32, which is conserved among all biogenic amine GPCRs and is critical for binding terminal amine groups (Fig. 1). Mutating D1133.32 to Ser greatly reduced activation of the •2-AR by biogenic amines. Remarkably, this mutation enabled the newly synthesized butanone derivative, 1-(3',4'-dihydroxyphenyl)-3-methyl-1-butanone (L-185,870), to activate the mutant receptor but not the wildtype (WT) receptor, albeit with relatively low potency (EC50=118 •M)9. Although this original report was inspiring, the synthetic agonist had low affinities and unknown pharmacokinetics that rendered it impractical for in vivo use.
The first engineered receptor or RASSL activated by an agonist with nanomolar affinity suitable for in vivo use was reported by Coward et al.5, who took advantage of potent synthetic drugs originally developed as potential analgesics, such as •-opioid receptor (KOR) agonists (e.g., spiradoline). The first RASSL was created by introducing mutations in the KOR that abrogated signaling via the natural peptide ligands, yet preserved stimulation by spiradoline5. This engineered human RASSL (hRO-i; h=human, R=RASSL, O=Opioid, i=Gi; referred to as Ro15, a Gi-coupled receptor, has been expressed in at least six tissues (Fig. 2) in transgenic animals. Diverse phenotypes were induced, including ligand-dependent heart-rate modulation10, bitter and sweet taste sensations11,12, and ligand-independent cardiomyopathy13, hydrocephalus14, and osteopenia15. These exciting results fueled efforts to develop RASSLs with improved ligand pharmacology and a greater range of signaling responses.
Once scientists understood that RASSLs could be designed to work with existing drugs, new RASSLs soon emerged from studies of a wide variety of receptors, including the 5-HT4 serotonin16, •2-adrenergic6, H1-histamine17, A3 adenosine7, 5-HT2A-serotonin18, and MC4-melanocortin19 receptors. In most, key residues essential for binding the native ligand were targeted by site-directed mutagenesis. For example, introduction of the D1003.32A mutation (analogous to the D1133.32A mutation in the •2-AR9) into the Gs-coupled human 5-HT4 serotonin receptor abolished the ability of 5-HT to activate the receptor but did not affect the activity of many synthetic agonists, including carbazimidamides, benzamides, benzimidazolones, and aryl ketones; this RASSL, previously known as Rs1, will be referred to as hRS-s16. Many of these synthetic agonists have drug-like properties, nanomolar affinities, and readily penetrate the central nervous system and can therefore be used effectively in vivo16,20. In addition, hRS-s is activated by antagonists of the 5-HT4 receptor, which have fewer in vivo side effects than 5-HT4 receptor agonists. Intriguingly, when expressed in osteoblasts of young mice, hRS-s dramatically alters bone growth in vivo21, presumably due to constitutive activation of the Gs pathway. These experiments provide valuable insights into the specific cellular and temporal factors that allow Gs signaling to induce bone growth.
This same mutagenesis approach of replacing conserved residues of the binding pocket with alanines has been applied to other biogenic amine receptors. Thus, mutation of a conserved serine residue in the fifth transmembrane region (S2045.46A) led to a sizeable loss of affinity and efficacy of (-)-adrenaline at •2A-adrenoceptors (Gi/o-coupled receptors), while the mutated receptor could still be activated by synthetic agonists (UK14304, clonidine) or even by antagonists of the WT receptor (atipamezole, SKF86466)22. Similarly, a Gq/11-coupled RASSL was obtained by introducing the F4356.55A mutation into the histamine H1 receptor17. This RASSL could be activated by high concentrations of endogenous histamine and had improved affinity and potency for 2-phenylhistamines, a class of synthetic H1R agonists (2-[3-chlorophenyl]histamine). Interestingly, alternative substitutions at this position (F4356.55) resulted in RASSLs with different levels of constitutive activity; F4356.55A (hRH-s) had the lowest level of constitutive signaling.
Despite these noteworthy advances, first-generation RASSLs were not ideal for experimentation. The ligands of first-generation RASSLs activated endogenous receptors (e.g., •-opioid, H1-histamine, 5-HT4-serotonin, MC4-melanocortin), and had low affinities for the mutated receptor (A3-adenosine neoceptor; ••-AR TREC; 5-HT2A serotonin RASSL). Moreover, profound phenotypes induced by constitutive activity were observed upon RASSL overexpression in vivo (RO-i and RS-s). Finally, development of new RASSLs by repeated cycles of directed mutagenesis was labor intensive and did not consistently yield receptors with ideal agonist affinities or controlled levels of constitutive signaling. To overcome these inherent difficulties, we developed a generic approach to create a new class of RASSLs that have low constitutive activity and respond specifically to drug-like, pharmacologically inert small molecules8.
We used a well-established yeast mutagenesis system to produce hundreds of thousands of mutant hM3 muscarinic receptors and screened them for signaling characteristics of an “ideal” RASSL8. After multiple rounds of mutagenesis and iterative screening, we isolated mutants that had lost the ability to respond to the natural ligand (acetylcholine) but gained the ability to respond with nanomolar potency to clozapine-N-oxide (CNO), a pharmacologically inert, bioavailable23 synthetic compound (Fig. 3). This new class of RASSLs was designated DREADDs (ref. 6). We will refer to this first DREADD as “hRMD-q” (RASSL M3 DREADD, Gq-coupled; referred to as hM3-D in ref. 6). The RMD-q receptor is insensitive to acetylcholine but activates the Gq pathway to induce calcium mobilization upon binding of CNO. Analogous mutations in the closely related M4 muscarinic receptor, which is Gi-coupled and inhibits cAMP accumulation, led to another RASSL/DREADD that we call hRMD-i (referred to as hM4D8). Intriguingly, when activated by CNO, hRMD-i silenced hippocampal neurons via G••-mediated activation of G-protein inwardly rectifying K+ (GIRK) channels8. It is likely that hRMD-i will be widely used to induce neuronal silencing in vivo via indirect activation of GIRKs. More recently, (Guettier, J.M., et al abstract, International Group on Insulin Secretion, St. Jean Cap-Ferrat, France; 2007), made chimeras of the rat equivalent of the hRMD-q that incorporates the second and third intracellular loops of the Gs-coupled •1 adrenergic receptor to create a Gs-coupled RASSL (rRMD-s, see table 1). Thus, CNO can be used to activate the Gs, Gi, or Gq signaling pathways, depending on which of the new RASSLs is utilized.
With the current DREADD-type RASSLs, only two point mutations were required to create hRMD-I and hRMD-q. By contrast, the rRMD-s required two point mutations and swapping of two intracellular loops. Creating other DREADD-type RASSLs by directed molecular evolution will likely require at least multiple point mutations based on our own unpublished experience (Pei et al, in preparation). Thus far, all of the point mutations have been found in or near predicted binding sites for orthosteric ligands.
The general method we devised evolves GPCR ligand specificity toward pharmacologically `inert” ligands (e.g., drug-like compounds without known molecular targets). This technique is likely to be widely used to create designer GPCRs, owing to the availability of strains of yeast (Saccharomyces cerevisiae) engineered to express and respond to human GPCRs24,25. When these GPCR-expressing yeast are activated by an agonist, the signal induces the expression of a variety of selectable markers under control of a Fus-1 promoter. This system allows for the facile screening and optimization of millions of mutant GPCRs in a relatively short time26. Dozens of human GPCRs have been expressed in yeast27, thereby opening up the potential to create families of designer GPCRs activated by specific small molecules.
RASSLs may be valuable in controlling growth and ensuring appropriate control of function for experimental or therapeutic tissue engineering. GPCR signaling is essential for the growth and differentiation of many tissues2. For example, the 5-HT2B serotonin receptor is required for cardiac development and cell-cycle progression28-30. Ectopic signaling via GPCRs can promote abnormal growth3, leading to human disease. For instance, drug-induced valvular heart disease may be caused by excessive stimulation of cardiac 5-HT2B receptors31. One can envision the use of RASSLs to activate discrete signaling pathways to promote the proper growth and differentiation of engineered tissues.
Another potential use of RASSLs is to gain precise control of signaling in neurons and other tissues. Currently, this control in defined neuronal populations can be facilitated by expressing RASSLs in a neuron-specific manner. We reported that CNO-mediated activation of the Gi-coupled hRMD-i induces neuronal silencing when expressed in hippocampal neurons8. When expressed in hippocampal neurons, hRMD-q induces neuronal excitation and intracellular Ca++ release in CNO-dependent fashion (Rogan and Roth, manuscript in preparation). Using these two engineered muscarinic receptors, one could gain precise bi-directional control of neuronal firing in vitro and in vivo. These modified receptors could also be used in other excitable tissues, such as cardiac pacemaker cells, where Gs stimulation speeds diastolic depolarization and accelerates heart rate, and Gi stimulation slows heart rate. Expression of different Gs (e.g., hRS-s or rRMD-s) and Gi (e.g., RO-i or hRMD-i) RASSLs in pacemaker cells could allow for the precise regulation of heart rate without affecting cardiac muscle function.
In studies of first-generation RASSLs expressed in vivo, constitutive signaling (constitutive activity) has often produced the most profound effects. Overexpression of a Gi-coupled RASSL (hRO-i) in cardiomyocytes led to cardiomyopathy13, while overexpression in osteoblasts led to osteoporosis15. Recently a Gs RASSL (hRS-s) expressed in osteoblasts induced marked bone growth15. Constitutive activity is a common property of native GPCRs32,33 and is essential for the normal function of certain GPCRs34. Therefore, RASSLs with different levels of constitutive activity (high and low) will be needed to recapitulate normal GPCR functions. Because of the potential ligand-independent effects, RASSL expression ideally should be controlled through conditional expression systems (e.g., Tet or Cre). With these systems, a single RASSL transgenic line can be used to drive expression in diverse tissues with tighter temporal control. The second-generation RASSLs (hRMD-q, hRMD-i, rRMD-s) were created to lack constitutive activity8, and thus far, their overexpression in mice has not elicited baseline phenotypes (Rogan, Roth, Guettier and Wess, unpublished observations). These second-generation RASSLs will be most useful for studies in which ligand-dependent effects (rather than baseline phenotypes) are sought.
The RASSL field has undergone dramatic growth in the past decade, but many more challenges lie ahead. For instance, the ideal family of RASSLs would respond to a clinically approved, biologically inert drug (e.g., antibiotic or antiviral) that has no intrinsic effect on human cells, allowing tissue engineering without the drug safety studies needed for relatively new compounds such as CNO. The optimal series of RASSLs would also selectively couple to each of the GPCR pathways, including noncanonical pathways, such as those involving arrestins, GRKs, and intracellular kinases29. Furthermore, each RASSL would have different constitutive responses, desensitization properties, and subcellular targeting that could be fine-tuned with simple mutations. For many of these goals, we will need spatiotemporal control of RASSL expression to allow for direct comparisons of RASSL actions that could be applicable to virtually any tissue. Several groups are now testing a variety of approaches (BAC transgenics, knock-ins, and inducible systems) to meet these new challenges and provide new tools to RASSL researchers. With these highly refined tools, biologists will have a better understanding of how to use RASSLs and GPCR signaling pathways for tissue engineering.
The cross-disciplinary nature of RASSL-related research fosters a highly collaborative community that makes protocols, reagents, and transgenic animals publicly available whenever possible. Even though individual members of the RASSL community initially created tools specifically for their own research, the potential uses of these tools go well beyond any individual project. Indeed, precisely because GPCR signaling is important to such a wide swath of biology, it is impossible for us to accurately predict how and where RASSLs will be ultimately used.
We anticipate that our RASSL delivery systems will be deployed for a wide range of tissue engineering applications in neurological disease, pain perception, immunology, bone metabolism, and diabetes. In each case, our efforts will provide enabling technologies to rapidly advance those fields. For instance, in many neurological diseases (e.g., Parkinson's disease), RASSLs may be useful for correcting the imbalance of neural pathways, in a manner that could complement the surgical/electrical approaches in current clinical practice. Similarly, many groups envision using tissue-engineering approaches to study pain perception pathways. RASSLs, which selectively modulate neuronal firing, should be ideal for this application. Although GPCRs are clearly important in bone metabolism, many key receptors signal via multiple pathways and exhibit constitutive signaling. RASSLs allow researchers to stimulate discrete signaling pathways in bone metabolism. Finally, in diabetes, GPCRs play a role in the growth, development, and function of insulin-secreting pancreatic •-cells35. Dissecting the precise roles of different G-protein signaling pathways in •-cell function should be of considerable therapeutic interest.
Perhaps most importantly, the use of RASSL technology may shed light on relatively unknown aspects of GPCR signaling. For instance, many researchers are investigating nonclassical signaling responses of GPCRs, such as signaling by G12/13, arrestins, receptor kinases, regulators of G protein signaling, Wnt receptor signals, and scaffolding proteins29. It should be of interest to create two RASSLs that only differ in their ability to activate the arrestin pathways. Expression of these two RASSLs in the same spatial and temporal pattern would then allow determination of the true physiological roles of arrestin signaling.
Some of these non-G-protein signaling pathways could prove essential for robust tissue engineering and for uncovering the pathways responsible for stem cell differentiation. One can envision scenarios in which RASSLs are selectively expressed in different stem cell lineages and then activated (with an exogenous ligand or by overexpression) to determine which pathways are responsible for lineage choices and tissue differentiation.
We would like to thank our funding agencies and the members of our labs who have contributed to the RASSL projects. In particular, we appreciate support by the US National Institutes of Health; HL60664-07 (BRC), DK072071 (RAN), U19MH82441 (BLR), a NARSAD Distinguished Investigator Award (BLR), the American Heart Association pre-doctoral fellowship program (0415005Y to WCC), and the Veterans Affairs Merit Review Program (to BPH), the California Institute of Regenerative Medicine, Gladstone Institute CIRM Fellowship Program (Grant T2-00003 to ECH). The Gladstone Institutes received support from National Center for Research Resources Grant RR18928-01. The authors thank J. Ng, T. Nguyen, D. Srivastava, R.W. Mahley, H. Zahed, B. Phillips, M. Spindler, G. Howard, and S. Ordway for providing valuable technical assistance and discussions.
Highlighted Papers (numbers refer to order in citations) 5. Coward et al., PNAS 1998: First use of the term RASSL and first description of a RASSL that responds to a drug with nanomolar potency allowing potential in vivo use.
8. Armbruster et al., PNAS, 2007. First application of directed evolution for engineering RASSLs/DREADDs that respond to an experimentally inert ligand (CNO).
10. Redfern et al., Nature Biotech 1999: First use of a RASSL in a transgenic animal. Expression of hRO-i (previously known as Ro1) in the heart allows control heart rate within seconds after administration of spiradoline.
11. Mueller et al., Nature 2005. First use of a RASSL to decode a sensory pathway in vivo. RASSL expression in the sweet or bitter taste buds of transgenic mice allows the RASSL ligand (spiradoline) to induce a sweet (attractive) or bitter (aversive) responses.