Emerging structural insights into GPCR–β-arrestin interaction and functional outcomes

Abstract

Agonist-induced recruitment of β-arrestins (βarrs) to G protein-coupled receptors (GPCRs) plays a central role in regulating the spatio-temporal aspects of GPCR signaling. Several recent studies have provided novel structural and functional insights into our understanding of GPCR–βarr interaction, subsequent barr activation and resulting functional outcomes. In this review, we discuss these recent advances with a particular emphasis on recognition of receptor-bound phosphates by βarrs, the emerging concept of spatial positioning of key phosphorylation sites, the conformational transition in βarrs during partial to full-engagement, and structural differences driving functional outcomes of βarr isoforms. We also highlight the key directions that require further investigation going forward to fully understand the structural mechanisms driving βarr activation and functional responses.

Introduction

G protein-coupled receptors (GPCRs) constitute a large family of cell surface proteins with more than 800 members, and they represent one of the most prominent classes of drug targets [1]. They recognize a wide range of signals including hormones, peptides, lipids, and chemicals to initiate an intricate network of downstream signaling that is crucial for cellular physiology and homeostasis [2]. There are three distinct identifying features of GPCRs that include a seven transmembrane (7TM) topology, coupling of heterotrimeric G-proteins, and phosphorylation by GRKs and recruitment of β-arrestins (βarrs) [3]. While there are some examples that exhibit a slight variation on this theme, this paradigm is mostly conserved across the members of this family [3]. Binding of βarrs is typically believed to desensitize the receptors by terminating G-protein coupling, mediate receptor endocytosis, and to initiate downstream signaling pathways [3,4]. Considering the multifunctionality of βarrs in the context of GPCR activation and signaling, there continues to be an ever-expanding interest in understanding the structural and mechanistic aspects of GPCR-βarr interaction leading to βarr activation and resulting functional outcomes [5].

Our current understanding of GPCR–βarr interaction is rationalized in terms of two primary driving forces, that is, agonist-induced receptor phosphorylation predominantly by GPCR kinases (GRKs), and activationdependent opening of the intracellular face of the receptor transmembrane core [5,6] (Figure 1a). The contribution of receptor phosphorylation in driving βarr interaction continues to be an interesting area of investigation as it likely drives the ability of βarrs to interact with and regulate a large repertoire of GPCRs despite poorly conserved primary sequence [5]. It has been demonstrated for multiple GPCRs that the pattern of phosphorylation referred to as bar-code entails specific conformational signatures on βarrs, which in turn determine βarr-mediated functional responses [7–9]. In this review, we discuss recent advances that uncover previously unanticipated mechanistic aspects of GPCR–βarr interaction such as decisive contribution of single phosphorylation sites, structural differences in βarrs between partially-engaged and fully-engaged complexes, and conformational differences between the activated βarr isoforms.

Figure 1. GPCR-βarr interaction and decisive contribution of single phosphorylation site in V₂R-βarr interaction.

(a). Agonist-induced activation of GPCRs results in receptor phosphorylation by GRKs followed by βarr binding and activation in a biphasic manner. Distinct patterns of receptor phosphorylation drive functionally-specific βarr conformations. (b). Superimposition of the crystal structures of βarr1 in complex with V₂Rpp^WT and V₂Rpp^T360A reveals repositioning of the phosphopeptides, especially in the proximal segment. The structural snapshots were designed based on PDB IDs 4JQI and 7DFA, respectively. (c). Mutation of Thr³⁶⁰ results in the disruption of a key interaction with the Lys²⁹⁴ in the lariat loop of βarr1, which in turn leads to a dramatic alteration in agonist-induced βarr1 trafficking pattern.

Phosphorylation pattern and spatial positioning of key sites

It is generally believed that multi-site phosphorylation of GPCRs determines the binding affinity of βarrs and the stability of GPCR–βarr complexes. Interestingly however, emerging data now underscores the decisive contribution of even single phosphorylatable residues, which are specifically positioned in the cluster of phosphorylation sites. This is best exemplified through a comprehensive characterization of the vasopressin receptor (V₂R) using a site-directed mutagenesis approach [10]. While the mutation of some of the phosphorylation sites in V₂R such as Thr³⁴⁷ and Ser³⁵⁰ do not appear to have a major effect on βarr recruitment, Thr³⁶⁰Ala mutation (V₂R^T360A) leads to a substantial decrease in βarr recruitment as measured by co-immunoprecipitationbased direct physical interaction assay and luciferasebased reporter assay [10]. Interestingly, V₂R^T360A mutation dramatically alters the trafficking pattern of βarrs as they do not localize to the endosomal vesicle unlike the wild-type receptor, even after sustained agonist-stimulation, although they efficiently translocate to the plasma membrane [10]. Moreover, this single site mutation also results in a dramatic decrease in agonist-induced ERK1/2 MAP kinase activation although G-protein-coupling remains unaffected [10].

Similar observations have also been reported for several other GPCRs including the apelin receptor [11] and the chemokine receptor CXCR7 [12]. For the apelin receptor, apelin-induced βarr recruitment is substantially reduced by the mutation of a single phosphorylation site (Ser³³⁹) while the mutation of other sites does not have a significant influence [11]. Interestingly, βarr recruitment in response to another agonist called elabela is sensitive to mutation of another phosphorylation site namely Ser³³⁵ [11]. Similar to V₂R^T360A, reduced βarr recruitment to apelin receptor for the single site mutants, also results in significantly attenuated level of ERK1/2 MAP kinase phosphorylation, especially at late time-points of agonist-stimulation [11]. Along the same lines, CXCL12-induced βarr recruitment for CXCR7, also referred to as atypical chemokine receptor subtype 3 (ACKR3), is most sensitive to a single phosphorylation site, that is, Thr³⁵² while the mutations of other sites exert only moderate effect individually [12].

What is the structural basis of this intriguing observation? Molecular dynamics (MD) simulation studies on V₂Rpp-βarr1 crystal structure revealed that Thr³⁶⁰Ala mutation disrupts a pivotal interaction with Lys²⁹⁴ in the lariat loop of βarr, which in turn leads to a smaller inter-domain rotation, a hallmark of βarr activation [10]. A subsequent study combining MD simulation and spectroscopy also converged to the same conclusion [13]. Moreover, an elegant study reporting the crystal structure of βarr1 in complex with multiple phosphopeptides derived from V₂R, including V₂Rpp^T360 which essentially mimics V₂R^T360A mutation provides additional structural insights [14](Figure 1b–c). In the crystal structure of βarr1–V₂Rpp^T360 complex, the proximal segment of the phosphopeptide (i.e. residues 353e360) was observed to have been repositioned compared to the V₂Rpp (Figure 1b). This repositioning of the phosphopeptide expectedly alters the interaction of the phosphate groups present in this segment of the phosphopeptide with Lys/Arg of βarr1 including the interaction with Lys²⁹⁴ [14](Figure 1c). These changes also influence the regions of βarr1 involved in the interaction with signaling partners such as c-Raf-1, MEK1, and c-Src, which may provide a potential structural explanation for compromised downstream signaling upon Thr³⁶⁰Ala mutation in V₂R[14]. Now that a recent study has comprehensively identified the binding sites of Raf1 and MEK1 on βarr1, it would be interesting to experimentally probe this hypothesis further. Nonetheless, these recent studies now establish the critical importance of spatial positioning of individual phosphorylation sites in GPCR–βarr interaction and signaling. This important advance should be integrated with the current conceptual framework of phosphorylation bar-code and how it influences βarr-mediated functional response. It is worth noting that a recent study has reported that Lys²⁹⁴Ala mutation in βarr1 does not attenuate interaction with some GPCRs although its interaction with V₂R was not measured and a conformational response was not explored [15].

Transitioning from tail-engaged to fully-engaged conformation

While phosphorylation and activation of GPCRs were originally proposed as two major determinants for βarr binding based on a number of biochemical and biophysical studies [16], direct evidence for biphasic interaction mechanism first emerged through negative-staining-based electron microscopy studies [17]. A chimeric version of the β2 adrenergic receptor (β2AR) harboring the carboxyl-terminus of V₂R, referred to as β₂V₂R, exhibited two distinct binding modes with βarr1 in a complex stabilized by an antibody fragment [17]. While one of these is mediated solely through the phosphorylated carboxyl-terminus of the receptor, the other involves an additional interaction with the receptor transmembrane core [17]. These two binding modes have been referred to as the “partially-engaged” or “tail-engaged” and “fully-engaged” or “core-engaged” conformations, and they have subsequently served as experimental framework to dissect the functional outcomes associated with the tail-interaction and the core-engagement in GPCR–βarr complexes [18–21].

While it is intuitive that these two binding arrangements are likely to impart distinct structural changes in βarrs, direct evidence to support this notion has emerged recently through an NMR spectroscopy analysis of partially-and fully-engaged β₂V₂R–βarr1 complexes assembled in nanodisc [22](Figure 2a). This study is based on measuring the chemical shift differences in fifteen isoleucine residues distributed across βarr1 as readout of conformational changes upon its interaction with the receptor [22](Figure 2a). A direct comparison of βarr1 in the basal state with that in tail-engaged and fully-engaged complexes suggests that βarr1 is partially-activated in the tail-engaged conformation and core-engagement results in additional conformational changes as reflected by chemical shift differences for Ile158 located in the N-domain of βarr1, and Ile²⁴¹ and Ile³¹⁷ in the C-domain (Figure 2b). While Ile¹⁵⁸ positions in the proximity of phosphorylated carboxyl-terminus of the receptor, Ile²⁴¹ and Ile³¹⁷ are at the core-interaction interface. Interestingly, a previous study has also provided evidence to suggest a significant structural change in βarr2 during transition from tail-only to core-interaction based on synthetic antibody sensors and fluorescence spectroscopy [23]. Taken together, these studies now establish that core-engagement with the receptor imparts additional conformational changes in βarrs than those induced by receptor phosphorylation alone, and they also offer a plausible explanation for functional segregation associated with the two binding modes between GPCRs and βarrs as reported in several studies [18–21].

Figure 2. Structural differences in barr1 between partially-engaged and fully-engaged complexes.

(a) Schematic representation of the partially-and fully-engaged β₂V₂R–βarr1 complexes reconstituted in nanodisc and studies using methyl-TROSY NMR spectroscopy. The lower panel shows the position of fifteen residues in βarr1 used in this study and the three isoleucine residues exhibiting significant differences between partially-and fully-engaged conformations are highlighted in blue. (b) Differences in the chemical shift for selected isoleucine (Ile) residues in basal, partially-engaged and fully-engaged βarr1 conformations as measured by NMR spectroscopy are shown. The position of the corresponding isoleucine residues in βarr1 are indicated in the table. These data are adapted from a previous publication [22].

Structural differences between barr isoforms

An intriguing aspect of βarr-mediated signaling and regulation of GPCRs is the functional divergence of the two isoforms, that is, βarr1 and 2 in the context of receptor desensitization, endocytosis and signaling [24]. While differences between receptor-bound conformation of βarr isoforms has been proposed as the underlying mechanism [23], relatively limited structural coverage of activated βarr2 restricts the mechanistic inference of this phenomenon at high-resolution. The structure of βarr2 in complex with a phosphopeptide derived from the CXCR7 (referred to as CXCR7pp) has shed some light on the differences between the activated conformations of βarr1 and 2 [25](Figure 3a). For example, CXCR7pp-bound βarr2 exhibits a smaller inter-domain rotation compared to the V₂Rpp-bound βarr1 although three-element interaction and polar core disruption are apparent [25] (Figure 3b). In addition, the positioning of the proximal segment of CXCR7pp also appears to be slightly different than V₂Rpp leading to the identification of a different binding pocket in βarr2 that accommodates phosphorylated Ser³³⁵ of CXCR7pp (Figure 3c). These findings raise the question if the difference in the inter-domain rotation and phosphate-binding pocket are indeed a specific structural feature of activated βarr2, or, it is an outcome of a limited number of phosphates in CXCR7pp compared to V₂Rpp. It should be noted however that isolated phosphopeptides may exhibit a slightly different binding mode and/or orientation compared to the full receptor, and therefore, additional structural coverage of βarr2 in complex with full GPCRs and phosphopeptides derived from additional receptors, may help clarify this point. In this context, it is also important to highlight recent studies that have utilized multiple biochemical and biophysical approaches to suggest that βarr1 and 2 may engage ERK2 differently in terms of overall binding interface andconformation[26,27]. While these studies probed βarr-ERK2 interaction in-vitro without a GPCR directly being involved, pre-activated mutants of βarrs generated through carboxyl-terminus truncation were used. Thus, the findings described in these studies are quite interesting as they may provide potential mechanistic explanation for differential contribution of βarr isoforms in ERK1/2 activation as reported for several GPCRs.

Figure 3. Structural differences between phosphopeptides-bound βbarr isoforms.

(a) Superimposition of V₂Rpp-bound βarr1 and CXCR7pp-bound βarr2. The structural snapshots are generated in ChimeraX [44] using previously determined crystal structures (PDB ID 4JQI and 6K3F) [25,45]. (b) Surface representation of V₂Rpp-bound βarr1 and CXCR7-bound βarr2 to depict the difference in the inter-domain rotation between βarr1 and 2. (c)V₂Rpp and CXCR7pp are differently positioned in the corresponding crystal structures and an additional phosphate-binding site is identified in βarr2 that is not apparent in V₂Rpp-bound βarr1 structure.

The knowledge gaps and emerging directions

In the current paradigm, Ser/Thr residues in the carboxyl-terminus of GPCRs are typically considered as primary sites of phosphorylation governing βarr binding although for some GPCRs, phosphorylation sites in the 2nd and 3rd intracellular loop (ICL2 and ICL3) have also been identified and characterized [28]. Interestingly, there are several examples such as the muscarinic, dopaminergic and serotonergic receptors, which contain a relatively short carboxyl-terminus but a significantly longer ICL3 with a series of phosphorylatable Ser/Thr residues (Figure 4a). Therefore, it would be very interesting to probe if such receptors also engage βarrs in a biphasic manner involving distinct steps of phosphorylation-dependent and transmembrane core-dependent binding interfaces. Considering the close proximity of phosphorylation sites (i.e. in ICL3) and the TM5-TM6 interface that forms a major part of the core interaction, it is plausible that partially-engaged and fully-engaged complexes of these receptors may be significantly different than those visualized for chimeric GPCR constructs such as β₂V₂R[17], β₁V₂R[29] and M₂V₂R[30] or NTS1Re–βarr1 complex stabilized using cross-linking [31].

Figure 4. Schematic depiction of key knowledge gaps and emerging directions.

(a) While most GPCRs harbor phosphorylation sites in their carboxylterminus, several others have a very short carboxyl-terminus and phosphorylatable Ser/Thr residues are predominantly located in the 3rd intracellular loop. The commonality and differences in the binding modes of barrs for these two types of GPCRs remains to be explored in detail. (b) Arrestin-coupled receptors (ACRs) robustly recruit βarrs despite lacking an inherent ability to functionally couple to G-proteins. While βarr conformations for ACRs vs. GPCRs are qualitatively distinct from each other, structural details at high-resolution remains to be determined. (c) The structural and mechanistic basis of catalytic activation of βarrs upon their transient interaction with selected GPCRs, despite rapid dissociation of the receptor from their complexes, remains to be explored.

Another interesting paradigm where our current understanding is rather limited is that displayed by the arrestin-coupled receptors (ACRs) [32]. While natural agonists of GPCRs drive the coupling of both G-proteins and βarrs, there are several examples of βarr-biased ligands, which preferentially trigger the coupling of βarrs but not G-proteins [33]. More importantly, there are multiple 7TM receptors that do not couple to G-proteins, even upon stimulation with their native agonists, but exhibit robust βarr-coupling and downstream signaling [32,34–36]. These receptors present an interesting model system to dissect the structural and conformational mechanisms driving distinct βarr-mediated functional outcomes. For example, considering the lack of G-proteins-coupling to ACRs, it is conceivable that βarrs do not have to mediate their desensitization. Therefore, βarrs are likely to engage with ACRs primarily through phosphorylation-dependent mechanism without a significant involvement of the receptor core, which appears to be essential for mediating receptor desensitization (Figure 4b). Still however, as βarrs are involved in their endocytosis and signaling, they must adopt an active-like conformation, a situation that can be viewed as being analogous to βarr-biased ligands activating prototypical GPCRs. Although a recent study has established distinct βarr conformations induced by selected ACRs compared to their prototypical GPCR counterparts [32], the fine structural differences between βarr complexes of ACRs vs. GPCRs remain to be visualized at higher resolution.

Yet another paradigm that has emerged in the past couple of years is the catalytic activation of βarrs observed in cellular context [37,38]. It appears that even a transient interaction with GPCRs is sufficient to drive an active-like conformation in βarrs that may sustain even after the dissociation of the receptor, and more importantly, drive downstream signaling such as ERK1/2 MAP kinase activation [37–39](Figure 4c). This conformational memory of βarrs visualized in live cells using a combination of high-resolution imaging and genetically encoded sensors challenges the notion that a physical complex of GPCRs and βarrs is a prerequisite to drive functional outcomes. Still however, structural features associated with such a conformational signature in βarrs remain to be deciphered using structural and biophysical approaches. The release of the carboxyl-terminus of βarrs from the N-domain is considered the key step in βarr activation [40,41] and accordingly, truncated βarr constructs have been observed to exhibit a measurable level of βasal activation and comparatively smaller dependence on receptor phosphorylation [42]. It is therefore conceivable that a similar mechanism may facilitate catalytic activation, where transient interaction with activated and phosphorylated GPCRs drives the C-terminal release of βarrs, a conformation that is stabilized further and sustained, potentially through additional interaction with the membrane bilayer potentially through c-edge loops [43].

Conclusion and future perspective

The recent advances have provided important insights into the interaction of βarrs with activated and phosphorylated GPCRs, especially with respect to decisive contribution of specifically positioned individual phosphorylation sites, conformational changes associated with the transition of partially-engaged complexes to fully-engaged complexes, and potential structural mechanism of functional divergence between the βarr isoforms. Going forward, structural visualization of additional GPCR–βarr complexes, especially those with native GPCRs, with both βarr isoforms in different binding modes and conformations would be important. Moreover, an improved structural mapping of interaction interfaces between βarrs with their binding partners, direct visualization of catalytic activation of βarrs, especially at higher-resolution than currently available, and specific contribution of lipids in spatio-temporal control of GPCR–βarr engagement are also key areas that require focused efforts. Together, these studies should illuminate the intricacies of how βarrs recognize and regulate a large repertoire of GPCRs, and help us better understand the structural and functional diversity encoded in the GPCR–βarr system