Being mindful of seven-transmembrane receptor ‘guests’ when assessing agonist selectivity

Abstract

The time-honored approach of quantifying agonist selectivity through measurement of agonist affinity with binding and efficacy through potency ratios in model assays for prediction of effect in therapeutic systems can fall short of providing useful answers for functionally selective agonists. Agonists are now known to have pluridimensional efficacies that are associated with selected signalling pathways coupled to the receptor. This necessitates specifically tailored assay formats to measure pre-determined efficacies of ligands to characterize agonist selectivity fully. If such assays can access signalling that directly emanates from the interaction of the agonist-bound receptor and a cytosolic signalling protein, then the Black/Leff operational model can be used to specifically quantify ‘transduction ratios’ (τ/K_A) that fully characterize selective activation of signalling pathways by a given agonist. As whole-cell processing of pleiotropic signalling cascades imposes cell-specific phenotypic agonist profiles, ultimately the assessment of agonist selectivity is most reliably done in the therapeutically relevant primary cell system.

The paper in this issue of the BJP by Jillian Baker (Baker, 2010) emphasizes the importance of determining both the affinity and efficacy of selective agonists in assessing overall selectivity of agonism. The value of this approach in determining agonist receptor selectivity is clear, as agonist potency can be controlled by affinity or efficacy, or both. This basic pharmacological principle can be extended in light of our present knowledge of pleiotropic signalling through seven transmembrane receptors (7TMRs). Specifically, it is apparent that agonists activating receptors coupled to multiple signalling proteins can produce biased activation of some signalling pathways over others (see Kenakin, 2006; Mailman, 2007). Under these circumstances, conventional whole-cell measures of affinity (binding) and efficacy (whole-cell potency ratios) can be misleading. It is therefore worth discussing the concept of agonist selectivity in the functionally selective world.

7TMRs are allosteric systems mediating the vectorial transduction of energy from one locus on the receptor to another.It is useful to think of 7TMR selective agonism in terms of classical allosteric terminology, thus, agonists are modulators, the receptor protein the conduit and cytosolic signalling proteins the guests (see Figure 1). It should be noted that considering the separate binding loci of agonists and signalling proteins, all agonists, whether they bind to the natural endogenous agonist binding site or other binding sites, are allosteric modulators with respect to the signalling protein. In addition, the terms modulator and guest are interchangeable in terms of energy flow. Thus, for an agonist as a modulator, the guest may be a G-protein. However, when viewed from the cytosol, the modulator could also be the G-protein and the agonist the guest; both viewpoints are pharmacologically relevant (vide infra). The receptor can be seen as rolling on an energy landscape as it adopts multiple conformations in response to changes in energy of the system; the various energy wells on this landscape represent favoured receptor conformations, some of which may mediate cellular signalling (Fraunfelder et al., 1991; Onaran et al., 2002; Hilser et al., 2006). As receptors take conformational excursions away from canonical native structures, the addition of an agonist modulator will necessarily bias the sampled conformations of the receptor and produce a new energy landscape for it to traverse. Under these circumstances, an allosteric modulator essentially produces a ‘new receptor’ with potentially different capabilities to interact with cytosolic signalling proteins. For example, a molecule may stabilize a conformation that favourably binds to and activates a G-protein, and thus induces visible agonism. A new energy landscape may not contain conspicuous energy wells for binding to standard signalling molecules, but rather may bias the receptor towards conformations mediating other behaviours such as in cases where non-G-protein signalling antagonists actively internalize receptors (Roth and Chuang, 1987) or only produce extracellular signal-regulated kinase (ERK) signalling through β-arrestin (see Azzi et al., 2003). These effects blur distinctions between classical ‘agonist’ and ‘antagonist’ labels for molecules (Kenakin, 2008). Thus, a given molecule can have numerous efficacies (pluridimensional efficacy, Galandrin and Bouvier, 2006), and this leads to what is commonly termed functional selectivity or agonist ‘bias’.

Figure 1.

Classical allosteric system for 7TMR agonism. All agonists become the ‘modulator’, the receptor is the ‘conduit’ and the cytosolic signalling molecules are the ‘guests’. Within this scheme, the Black/Leff operational model can be used to quantify selective activation of pathways through identification of unique log(τ/K_A) ratios.

The nature of the guest molecule dictates the strength and the type of signal produced by an agonist, but it also determines the affinity of the modulator agonist for the receptor. As mentioned previously, allosteric energy has reciprocal vectorial properties. Thus, if a molecule increases the affinity of the receptor for a given guest (i.e. G-protein), the same is true in the reverse direction, namely the guest will increase the affinity of the receptor for the modulator. An example of this effect has been shown experimentally with [³H]-dimethyl-W84 and [³H]-NMS binding on muscarinic M₂ receptors (Trankle et al., 1999). The influence of different coupling molecules (i.e. G-proteins) on the affinity of the muscarinic receptor for the agonist carbachol was elegantly demonstrated by Birdsall et al. (1980) in the form of a 56-point [³H]-propylbenzylcholine displacement by the agonist carbachol curve spanning seven orders of magnitude. The different affinities of carbachol with each G-protein couple are made manifest by the extremely broad range of the total binding curve demonstrating the fact that different guests can confer different affinities on the receptor for the agonist. For this reason, functional selectivity must be described in terms of the affinity (and activating properties) of the agonist-bound receptor for the signalling protein and the affinity of the receptor for the agonist. The identity of the guest can be extremely important to the interaction of the agonist with the receptor. For example, the modulator eburnamonine binds to the muscarinic M₂ receptor; if the guest molecule is acetylcholine, the binding is enhanced by a factor of 15. If the guest molecule is pilocarpine, the affinity is decreased by a factor of 25 (Jakubic et al., 1997). For these reasons, the signalling pathway activated by the agonist-bound receptor also controls the affinity of the receptor for the agonist.

A most useful tool for characterization of agonism is the Black/Leff operational model (Black and Leff, 1983), as it involves both affinity (K_A, the equilibrium dissociation constant of the agonist–receptor complex) and efficacy (K_E, the virtual equilibrium constant for the complex formed between the ‘activated’ receptor and cellular stimulus–response machinery). This latter term embodies both tissue sensitivity and the intrinsic efficacy of the agonist in the form of τ which is the receptor density divided by K_E. Therefore, selective activation of a pathway can be described by both K_A and τ; a convenient single parameter to do this is log(τ/K_A) (Kenakin, 2009). As shown in Figure 1, if the activated (agonist-bound) receptor directly interacts with a range of signalling proteins, then different cellular signals may result each quantified by unique τ/K_A ratios. The difficulty comes in accessing these selective activations to characterize ‘selective’ agonism with tools that measure macro-affinity (i.e. total agonist binding) and cell-based efficacy (which may be a mixture of pathways emphasized in agonist-specific ways). Under these circumstances, conventional methods to measure affinity (i.e. binding) or selective efficacy (agonist potency ratios from whole-cell response) become blunt instruments in dissecting pathway selectivity. Radioligand binding of agonist–receptor interaction in membranes may result in measures of a ‘high-affinity’ state which may be an amalgam of receptor–G-protein complexes, only one of which may be relevant to a given signalling pathway. For instance, the affinity of [¹²⁵I]-human calcitonin for the human calcitonin receptor is 400-fold greater than the EC₅₀ for calcium transient response in HEK cells; presumably, the known pleiotropic interaction of the calcitonin receptor with G_i, G_q and G_s protein, and the absence of GTP in the binding assay contribute to the high-affinity binding that may not be relevant for functional response (Watson et al., 2000). Similarly, total cellular response may be a crude indicator of agonist–receptor signalling. For instance, activation of ERK1/2 can be achieved through both G-protein and β-arrestin (non-G-protein) routes. Thus, for angiotensin receptors, equiactive ERK responses to angiotensin II (G-protein +β-arrestin activation) and the analog SII (only β-arrestin activation) belie the separate signalling pathways these ligands may activate in any given system (Wei et al., 2003).

The observation of functionally selective agonism has considerably muddied the waters with respect to quantifying relative potency in one system and assuming that it will be applicable to all systems. This is because the cell can influence agonist activity by virtue of modifying the relative stoichiometry of signalling components, that is, a β-arrestin-dependent agonist will be more active in cells with high levels of β-arrestin, as opposed to those with lower levels of β-arrestin. Two potential (and concurrent) strategies to address this problem could be envisioned. The first may be to characterize therapeutically relevant selectivity in the therapeutically relevant primary cell, perhaps through label-free technology. A second may be to quantify agonist bias through the operational model with systems that mediate signalling directly interacting with the agonist-bound receptor, and relate this scale to therapeutically observed phenotypic agonist activity. While, ultimately, this will not furnish direct measures of therapeutically relevant selectivity, it may furnish a scale through which medicinal chemists may actively optimize useful biased activity.

Glossary

Abbreviations

dimethyl-W64

N,N-bis[9-(1,3-dihydro-1,3-dioxo-4-methyl-2H-isoindol-2-yl)propyl]-N,N,N-tetramethyl-1,6-hexanediaminium diiodide

GTP

guanosine triphosphate

HEK

human embryonic kidney

NMS

N-methylscopolamine

SII

Sar¹,Ile⁴,Il⁸-Angiotensin II

Conflict of interest

The author declares no conflict of interest in this work.