Orthosteric- and allosteric-induced ligand-directed trafficking at GPCRs

Abstract

Many orthosteric agonists differentially activate downstream effectors of GPCRs. Such defined induction of signaling has strongly supported the hypothesis termed ‘ligand-directed trafficking of receptor signaling’ (LDTRS). More recently, subtype-selective GPCR activators, such as allosteric agonists and positive allosteric modulators, have also exhibited the capacity to activate specific signaling pathways. Based on this finding, it may be possible to achieve ligand-specific receptor active states that optimize the biological responses specific to GPCRs. This review discusses recent studies in which both orthosteric and allosteric compounds have been demonstrated to induce LDTRS.

Introduction

GPCRs represent one of the largest classes of drug targets, with up to 40% of currently marketed drugs acting at these proteins [1,2]. GPCRs are active in almost every organ system and present a wide range of opportunities as therapeutic targets for many disorders, including cancer, cardiac dysfunction, CNS disorders and obesity [1]. Consequently, GPCRs have remained primary targets for intense research and drug development in many academic and industrial laboratories. These receptors have a common architecture at their transmembrane core, consisting of seven α-helices separated by intracellular and extracellular loops [3]. It is generally accepted that signaling through a GPCR involves neurotransmitter or ligand binding and subsequent modulation of a large variety of intracellular effectors, such as protein kinases and ion channels [4], via the activation of single or multiple G-protein families [5-8]. In addition, GPCRs can signal effectors independently of G-protein action to induce cellular behaviors, such as cellular proliferation and apoptosis, by induction of β-arrestin-dependent signaling [7, 9]. Given this multiplicity of potential interacting partners, it is clear that the measurement of a single response is insufficient to understand the complexities of signaling coupled to GPCR activation.

Ligand-directed trafficking of GPCR signaling: Inconsistent with a linear model of efficacy

Traditionally, detailed experimentation in tissues was used for the rigorous characterization and quantification of pharmacological responses [10,11]. These data led researchers to suggest that agonist-relative efficacy is a unique property of a drug, and that efficacy is independent of the effector pathway measured [10,11]. Consequently, to explain the complex behavior of many biologically active compounds, a model of linear efficacy was proposed, suggesting that the full complement of cellular responses coupled to a given GPCR was equally elicited. Although there have been accounts of many compounds adhering to this mode of action, it is becoming increasingly evident that some compounds differ in their ability to stimulate separate stimulus-response pathways coupled to GPCRs, suggesting effector pathway-dependent effects (ie, agonist-directed trafficking of receptor signaling [ADTRS]) [3,12-14]. For example, structurally unrelated orthosteric (competitive) agonists of serotonin 5-HT_2A/2C receptors have demonstrated differential phospholipase C (PLC) and phospholipase A (PLA) activation, manifested as changes in the rank orders of potency or efficacy of compounds [15]. Similarly, some agonists of μ-opioid receptors have exhibited the ability to induce effector activation in the absence of receptor internalization, suggesting non-linear agonist-induced cellular behaviors [16-18]. This phenomenon is also not restricted to orthosteric agonists. Many reports have confirmed that allosteric ligands that interact with sites on the receptor that are distinct from the endogenous ligand (orthosteric) binding site can induce functionally specific cellular effects, suggesting allosteric ligand-induced differential signaling efficacies [19-22]. Based on these results, a working definition of efficacy can no longer be explained by a linear model, but will need to include differential efficacies for various signaling pathways downstream of GPCRs [14,23]. To accommodate this need, researchers have attempted to modify the conceptual framework underlying receptor theory. Many of these ideas can be classified using the general descriptive terms that include ADTRS, ‘agonist-directed signaling’, ‘ligand-bias’ and ‘functional selectivity’. For simplicity, this review will use the term ‘ligand-directed trafficking of receptor signaling’ (LDTRS).

Divergent GPCR signaling and multiplicity of effector activation

As noted, ligand-bound receptors can induce the activation of many signaling cascades. The dynamics of such responses are generally governed at multiple levels of the response process and, thus, differential activation of cellular responses downstream of ligand binding can diverge from multiple points. In many cases, signals can diverge at the level of the G-protein. Many datasets have confirmed that, once activated, Gα and Gβγ subunits can simultaneously, and synergistically, activate or inhibit downstream effectors (Figure 1A). The best-characterized example of such dual effector regulation is the simultaneous inhibition of adenylyl cyclase via Gα subunits and activation of PLC via Gβγ subunits. This phenomenon has been confirmed downstream of multiple GPCRs, including the 5-HT_1A receptors in HeLa cells [24]and dopamine D₂ receptors in LtK- fibroblasts [25]. Importantly, pertussis toxin (PTX) blocks these responses, suggesting that both pathways are affected by G-proteins belonging to the Gα_i/o family (reviewed in reference [26]). Receptors have also been demonstrated to couple to multiple G-protein families, a finding that provides an additional level of signaling diversity downstream of GPCRs (Figure 1B). In some cases, up to four unrelated classes of G-proteins can be activated; for example, using human thyroid gland, Laugwitz et al demonstrated that the human thyrotropin receptor induced the incorporation of photoreactive [α-³²P]guanosine triphosphate (GTP)-azidoanilide into immunopreciptated Gα_i/o, Gα_s, Gα_q/11 and Gα₁₂ proteins [27]. Furthermore, pretreatment with PTX caused a 35% increase in the accumulation of cAMP, suggesting functional coupling to both Gα_i and Gα_s G-proteins [27]. Similarly, receptor promiscuity for three G-protein families (Gα_i/o, Gα_s and Gα_q/11) has been demonstrated for corticotropin-releasing hormone receptors [28], gonadotropin-releasing hormone receptor [29] and muscarinic M₁ AChRs [6,30,31]. In each example, second messenger (eg, cAMP or inositol 1,4,5-trisphosphate [IP₃]) production was induced, suggesting the functional interaction of receptors with multiple G-protein families. G-protein coupling promiscuity has also been demonstrated for neurotensin receptors [32]and for histamine H₂ receptors [33]. Together, these results suggest that multiple effectors can be stimulated by agonist-induced receptor activation via an interaction with multiple G-proteins.

Figure 1. Multiple mechanisms of signaling divergence downstream of GPCRs.

Signals can diverge downstream of agonist binding as a result of the dissociation of the Gβγ and Gα subunits. (B) Receptors can promiscuously couple to multiple G-protein families, providing an additional branch point for signaling. (C) Agonist binding can result in a multiplicity of effector coupling by the activation of either G-protein responses or β-arrestin responses.

(Adapted with permission from Elsevier Ltd and Hermans E: Biochemical and pharmacological control of the multiplicity of coupling at G-protein-coupled receptors. Pharmacol Ther (2003) 99(1):25-44. © 2003 Elsevier Ltd)

In addition to coupling to multiple G-protein families, several studies now confirm effector activation via β-arrestin-mediated protein scaffolding (Figure 1C). For example, stimulation of angiotensin II (AngII) type 1 (AT₁) receptors in HEK293 cells increased the binding of both cRaf-1 and extracellular regulated kinase 2 (ERK2) to β-arrestin 2, an association thought to enhance cRaf-1 signaling and MEK-dependent activation of ERK2 [34]. Similarly, agonist-induced β-arrestin recruitment can act as a scaffold for other members of the MAPK family, including p38 protein kinases and c-Jun NH₂ terminal kinases [35]. Interestingly, β-arrestin-dependent signaling to MAPKs can be dissociated from G-protein-dependent signaling. For example, full-length parathyroid hormone (PTH) induces a biphasic ERK response through type 1 PTH-related peptide receptors. The initial phase of this response is sensitive to PKA and PKC inhibitors, suggesting G-protein-mediated effects, while the delayed response is sensitive to siRNA-mediated β-arrestin depletion, suggesting signaling mediated exclusively by β-arrestins [36].

Thus, GPCRs can couple promiscuously to multiple effectors via interaction with G-proteins and/or β-arrestins. A functionally selective agonist could, theoretically, stimulate any combination of these responses, resulting in LDTRS (Figure 2A). To explain this behavior, a model with multiple active receptor states has been proposed in which distinct receptor conformations are coupled with distinct effectors [37-41]. In contrast, models of receptor behavior that include two active states better explain classical pharmacological predictions [42-44]. According to these models, ligands indifferently stabilize an intermediate active conformation, allowing simultaneous activation of all signaling pathways (Figure 2B).

Figure 2. Potential responses downstream of functionally selective GPCR agonists.

(A) Agonist 1 stimulates multiple responses downstream of a GPCR, whereas agonist 2 favors the induction of the G-protein response 1. (B) Agonist 3 acts as a non-biased ligand that has no preference for the response it elicits. (C) In the presence of a positive allosteric modulator (PAM), an agonist induces a receptor conformation favorable for G-protein response 2.

Orthosteric ligand-induced functional selectivity of GPCRs

Some of the first reports of LDTRS demonstrated that orthosteric agonists induce non-equivalent effects across multiple signaling pathways. For example, at the M₁ AChR, the orthosteric agonist carbachol was demonstrated to stimulate both Gα_q-mediated PLC activation and Gα_s-mediated cAMP accumulation, whereas rigid analogs of ACh, including cevimeline (AF-102B), AF-150 and AF-151, did not stimulate cAMP production [45]. These findings have also been extended beyond M₁ AChRs. For example, using a [³⁵S]GTPγS binding/immunoprecipitation strategy in M₂- or M₄-expressing CHO cells, a later study demonstrated that the partial agonist pilocarpine was more effective in activating Gα_i3 than Gα_i1/2 G-proteins [30]. In addition, pilocarpine induced the activation of Gα_i3 subunits, but not Gα_q/11subunits, in CHO cells expressing M₃ AChRs, suggesting differential activation of two families of G-proteins downstream of muscarinic (m)AChRs [30].

As noted, a study using a series of orthosteric agonists of 5-HT_2A/2C subtypes demonstrated the differential induction of [³H] inositol phosphate (IP) accumulation and [¹⁴C]arachidonic acid (AA) release [15]. In this study, the 5-HT_2C receptor agonist 3-trifuoromethylphenyl-piperazine (TFMPP) favored IP generation rather than AA release, whereas lysergic acid diethylaminde (LSD) favored AA release rather than IP generation [15]. Importantly, the study simultaneously measured AA release and IP accumulation, two mainly independent cellular responses, revealing a reversal in potency and efficacy for these two compounds and indicating actual LDTRS. Furthermore, these effects were observed in expression systems with little to no receptor reserves for either response, permitting the intrinsic activity of each agonist to be quantified by comparing the ratio of the maximal response of each drug to that of the full agonist 5-HT [15].

Allosteric ligand-induced functional selectivity of GPCRs

In addition to orthosteric ligand-induced effects, recent investigations have suggested that compounds that allosterically bind and modulate GPCRs can induce LDTRS. These compounds, typically referred to as allosteric activators or modulators, have exhibited selectivity for individual receptor subtypes and are providing exciting new approaches for the development of therapeutic agents. Allosteric activators include agonists, positive allosteric modulators (PAMs) and negative allosteric modulators (NAMs), and have been identified for multiple GPCR families, including metabotropic glutamate receptors (mGluRs), mAChRs, and adenosine and serotonin receptors [46-48]. PAMs and NAMs are two classes of allosteric modulator that have been demonstrated to modify receptor conformations, resulting in altered binding and/or functional properties of the orthosteric ligands. PAMs and NAMs are unlike agonists in that they are inactive in the absence of orthosteric ligands. It is also apparent that these modulators possess a rich repertoire of behaviors that extend beyond simple modification to orthosteric agonist affinity [37]. Specifically, these modulators may induce functionally selective receptor conformations more frequently than orthosteric ligands (Figure 2C). While this phenomenon is still poorly understood, a recent study demonstrated that two structurally diverse potentiators of M₁ AChRs are distinct in their abilities to enhance the activation of downstream effectors [49]. In this study, the compounds VU-0029767 and VU-0090157 (both Figure 3) potentiated ACh-induced calcium mobilization, as is typically expected for an allosteric potentiator. However, VU-0090157 potentiated ACh-induced activation of phospholipase D (PLD) and phosphoinositide (PI) hydrolysis, while VU-0029767 had little or no effect on PLD activation or PI hydrolysis, suggesting allosteric modulator-induced functional selectivity. As M₁-mediated signaling to PLD typically involves the activation of Gα₁₂ or small G-proteins such as R-Ras [50,51], it was concluded that when bound simultaneously by ACh and VU-0029767, M₁ AChRs adopt receptor conformations that are unable to form productive signaling complexes with non-Gα_q signaling partners [49]. In a similar study, the differential activation of receptor responses coupled to mGluR5 were demonstrated upon binding of the allosteric potentiator N-(4-chloro-2-[(1,3-dioxo-1,3-dihydro-2H-isoindol2-yl) methyl]phenyl)-2-hydroxybenzamide (CPPHA; originating from Merck & Co Inc; Figure 3) [52]. Interestingly, CPPHA potentiated glutamate-induced ERK phosphorylation at low glutamateconcentrations, but decreased ERK responses at high glutamate concentrations. At the M₄ AChR, the PAM LY-2033298 (Lilly Research Centre Ltd; Figure 3) was also demonstrated to stimulate pathway-specific allosteric modulation [53]. Using an operational model of allosterism and agonism, Leach et al demonstrated that the functional cooperativity between LY-2033298 and ACh was significantly greater for ERK1/2 and GSK-3β phosphorylation when compared with [³⁵S]GTPγS binding [53]. Similarly, a NAM of the CRTH2R (chemoattractant-receptor-homologous molecule on T-helper 2 cells) has been described that is inactive with respect to coupling to G-protein-linked pathways, but is a potent antagonist of G-protein-independent β-arrestin coupling to the same receptor [54]. Together, these results provide multiple examples of allosteric modulator-induced receptor conformations that are non-equivalent across different signaling pathways.

Figure 3. Selected functionally selective positive allosteric modulators of GPCRs.

More recently, it has become evident that there is an additional class of allosteric ligand, referred to as allosteric agonists, that is capable of inducing the selective activation of receptor responses. These agonists bind GPCRs at topographically distinct sites and can induce cellular responses in the absence of endogenous ligand. Many cases of the differential activation of receptor responses by putative allosteric agonists have been observed at mAChRs. For example, a multiple assay approach that included [³⁵S]GTPγS binding, as well as measurements of IP₃ generation and adenylyl cyclase activity in M₁-expressing CHO cells, was used to demonstrate that the allosteric agonists AC-42 (4-N-butyl-1-[4-(2-methylphenyl)-4-oxo-1-butyl] piperidine; ACADIA Pharmaceuticals Inc/Meiji Seika Kasisha Ltd; Figure 4) and 77-LH-28-1 (1-[3-(4-butyl-1-piperidinyl)propyl]-3,4-dihydro-2(1H)-quinolinone; GlaxoSmithKline plc; Figure 4) did not significantly induce [³⁵S]GTPγS binding to Gα_i1/2 when compared with the orthosteric agonists oxotremerine-M, pilocarpine and arecoline [19]. In contrast, Gα_q/11 and Gα_s-mediated signaling was induced by both types of agonists, suggesting that the allosteric agonists activated only distinct subsets of G-proteins [19]. A subsequent study by the same researchers demonstrated that the orthosteric agonists caused significant receptor internalization and downregulation, whereas prolonged exposure of M₁-expressing CHO cells to AC-42 did not significantly alter either cell-surface or total cellular M₁ AChR expression; 77-LH-28-1 caused some degree of receptor internalization, but did not induce receptor downregulation in these cells [22]. Similarly, the M₁-selective allosteric agonists TBPB ((1-(1′-2-methylbenzyl)-1,4′-bipiperidin-4-yl)-1H -benzo [d] imidazol-2(3H)-one; Merck & Co Inc; Figure 4) and AC-260584 (4-[3-(4-butylpiperidin-1-yl)-propyl]-7-fuoro-4H-benzo[1,4] oxazin-3-one; ACADIA Pharmaceuticals/Meiji Seika Kaisha; Figure 4) were demonstrated to be functionally coupled to the ERK pathway and to induce calcium release; however, these compounds did not induce significant β-arrestin recruitment or internalization of the M₁ AChR [20]. These effects are reminiscent of those of the prototypical analgesic opioid agonist morphine, which has been demonstrated to activate μ-opioid receptor-mediated effectors in the absence of receptor internalization in HEK293 cells [18]. However, in contrast to the reported lack of M₁ AChR internalization following the addition of AC-260584, a recent report indicated that AC-260584 induced prolonged internalization in the absence of recycling of M₁ AChRs in HEK293 cells [21].

Figure 4. Selected functionally selective allosteric agonists of GPCRs.

In addition to agonists of the M₁ AChR, other putative allosteric agonists have been demonstrated to be functionally selective at different GPCRs. For example, in a study by Griffin et al, the partial agonist McNa-343 (4-[[[(3-chlorophenyl) amino]carbonyl] oxy]-N,N,N-trimethyl-2-butyn-1-aminium chloride; Figure 4) differentially activated effectors downstream of M₂ AChRs [55]. In this study, concentration-response curve analysis was used to estimate the intrinsic relative activity of McN-A-343; this compound had 10-fold greater activity for stimulating M₂-Gα₁₅ responses compared with M₂-Gα_i responses [55]. A more recent study suggests that McNa-343 is a bitopic molecule composed of an orthosteric agonist moiety coupled to an allosteric modulator that induces the differential activation of M₂ AChR responses by occupying both the orthosteric and allosteric binding sites [56]. This bitopic behavior can, theoretically, arise from either switching between two different binding modes or simultaneous occupation of both the orthosteric and allosteric sites.

Ligand-directed trafficking of receptor signaling at β-arrestins

It is important to reiterate that agonist-biased effects are not limited to cellular responses that are mediated by G-proteins; many orthosteric agents have demonstrated functional selectivity for β-arrestin-mediated scaffolding. For example, the β-adrenergic receptor antagonist carvedilol exhibited negative efficacy for Gα_s-mediated signaling, but positive efficacy for β-arrestin-mediated ERK activation [57]. Similarly, the AngII analog Sar¹Ile⁴Ile⁸-AngII, which acts as an agonist for the AT₁ receptor, induced ERK activation in the absence of reported G-protein activation at AT₁ receptors [58]. While there have been no reported examples of allosteric ligand-induced effector activation mediated exclusively by β-arrestin scaffolding, this pathway provides an additional arm of signaling that may aid the understanding of the potential functional effects that occur downstream of allosteric modulator-stabilized receptor states.

Conclusion

Given the examples of ligand-induced differential efficacies described in the preceding sections, it is clear that classifying compounds on the basis of their ability to modulate a single effector system does not provide a comprehensive description of the potential of a particular compound. Therefore, a multi-assay approach in a similar cellular background will be required to understand the pluridimensionality of GPCR agonists and modulators. Furthermore, estimates of the relative signaling efficacies will be required to compare functional selectivity between agonists. Such an approach was recently used to identify allosteric site mutations of the M₂ AChR that provide ligand-selective signaling bias [59]. In terms of desensitization, the findings that allosteric and orthosteric agonists induce differential receptor endocytosis suggest that the intrinsic activity of an agonist and receptor endocytosis may not always be fundamentally linked. These findings are particularly interesting, given previous reports of a tight correlation between the intrinsic activity of an agonist and its propensity to induce receptor endocytosis [60].

The most parsimonious explanation for LDTRS is that the receptor conformation stabilized by a given ligand can be active toward one set of downstream effectors, while being inactive for another set of effectors [61]. A multi-state model in which distinct receptor conformations are involved in the coupling with distinct effectors has been proposed [37-41]. Based on this model, creative drug design could help to stabilize the most productive active receptor states for a particular pathway. In terms of drug discovery, the induction of LDTRS by orthosteric or allosteric ligands will change the current paradigm for proper target validation. Accordingly, investigators may find it helpful to generate as much information as possible on the most constructive and therapeutically relevant biological responses for a given disease, prior to beginning a search for a potential therapeutic ligand for a specific receptor subtype. Such findings could help prevent some of the undesired side effects commonly associated with therapeutic compounds by avoiding the activation of ancillary pathways unrelated to the targeted therapeutic process.