Molecular determinants of orexin receptor-arrestinubiquitin complex formation

Introduction

The orexin system plays a critical role in maintaining and integrating primordial physiological functions including sleep-wake transitions and metabolic signals controlling energy homeostasis, as well as modulation of addictive behaviour processes and dependencies (Sakurai and Mieda, 2011; Kim et al., 2012).

Both orexin receptors (OX₁ and OX₂; receptor nomenclature follows Alexander et al., 2013) typically couple with the G_q subclass of G proteins upon stimulation, resulting in release of inositol phosphates and elevation of intracellular calcium levels (Sakurai et al., 1998). In addition, it is evident that both these receptors robustly recruit members of the multi-adaptor protein family, the arrestins (Milasta et al., 2005; Dalrymple et al., 2011).

The formation of arrestin-bound GPCR complexes can confer a wide range of regulatory and cellular signalling functions to the complex. Activated arrestin-bound complexes can desensitize certain signalling cascades, such as those mediated through Gα, and also propagate distinct signalling processes through scaffolding various precursors to pathways resulting in MAPK, c-Src and Akt activation (DeWire et al., 2007). Association with β-arrestin also influences compartmentalization of complexes in the cellular milieu including internalization, recycling and degradation through their ability to bind other structural and regulatory proteins (Moore et al., 2006). Within these complexes, ubiquitination of both receptors and β-arrestins introduces another level of regulation that can influence compartmentalization, trafficking and signalling properties of the complex (Becuwe et al., 2012).

Recruitment and stability of GPCR-β-arrestin interactions is dependent on the affinity of β-arrestin for the GPCR. This typically occurs through phosphorylation of serine and threonine residue clusters in the C-terminal tail of GPCRs by GPCR kinases (GRKs). Phosphorylation of these residues provides the necessary chemical energy to promote high affinity interaction between receptors and β-arrestins (Gurevich and Gurevich, 2006). The presence or absence of clusters of these residues can influence the temporal stability of GPCR-arrestin interactions. This has been characterized for a number of GPCRs including the β₂-adrenoceptors, vasopressin V_1A and V₂ receptors, μ- and δ-opioid receptors, thyrotropin-releasing hormone TRH₁ receptors, angiotensin II AT_1A receptors and dopamine D₂ receptors and may broadly promote separation of GPCRs towards different downstream trafficking and signalling outcomes based upon their relative degree of arrestin association (Oakley et al., 1999; 2000; 2001,,; Kafi et al., 2011).

Using BRET techniques, we have shown that both OX₁ and OX₂ receptors display relatively high stability in forming complexes with both β-arrestins (Dalrymple et al., 2011). However, in contrast to other receptors that display stable arrestin association, tangible differences between the BRET kinetics of the OX₁ and OX₂ receptor-β-arrestin complexes were only observed upon prolonged measurement. Extended BRET (eBRET) assays displayed kinetic profiles between β-arrestin, ubiquitin and OX₁ receptors that were more transient over a period of 4 h of orexin A stimulation, compared with profiles between β-arrestin, OX₂ receptors and ubiquitin which exhibited a more robust and stable kinetic profile (Dalrymple et al., 2011). In addition, temporal ERK1/2 phosphorylation could be similarly subtype specifically distinguished between the orexin receptors over extended periods of agonist stimulation. These long-term departures of orexin receptor-β-arrestin BRET kinetics suggest a mechanism for differential orexin receptor subtype function.

To gain an insight into such possible mechanisms involved in functional orexin receptor subtype distinctions, the contribution of the C-terminal tail of OX₂ receptors was investigated through bulk substitution and site-specific mutagenesis. Previous studies that investigated molecular determinants involved in GPCR-β-arrestin interactions revealed the nature of serine/threonine cluster sites primarily phosphorylated by GRKs (Oakley et al., 1999; 2000; 2001,,) and specifically for OX₁ (Milasta et al., 2005). Using these principles, the contributions of both the entire C-terminal tail as well as three putative GRK phosphorylation sites within the C-terminal tail of OX₂ receptors were assessed to investigate the formation and stability of OX₂ receptor-arrestin-ubiquitin complexes and to demonstrate key structural features that defined subtype-specific functions of orexin receptors.

Methods

Results

Discussion and conclusions

We have previously shown that the two orexin receptor subtypes display diverging BRET kinetic profiles when forming β-arrestin and ubiquitin complexes, with differences in cellular localization and signalling also being observed (Dalrymple et al., 2011). OX₂ receptors displayed greater stability over time when forming β-arrestin complexes compared with OX₁ receptors, along with a sustained ability to maintain phosphorylated ERK1/2 while being unable to rapidly recycle upon internalization. Prior to that study, OX₁ receptors had been found to colocalize with β-arrestin1 upon orexin A stimulation using confocal microscopy (Evans et al., 2001) and specific sites in these receptors were shown to be involved in β-arrestin interaction (Milasta et al., 2005). Our aim was therefore to establish the molecular determinants responsible for OX₂ receptor interactions with β-arrestin and to understand the molecular basis for the differences in OX₁ and OX₂ receptor function that we had observed, in terms of receptor-arrestin-ubiquitin complex stability.

The OX₁ctOX₂ chimera (Figure 1) was generated to investigate whether the C-terminal tail of OX₂ receptors could bestow greater stability to the OX₁ receptor-β-arrestin complex. However, ligand-induced BRET assays indicated that BRET-tag orientation made a substantial difference to our ability to detect a β-arrestin proximity BRET signal specifically for this chimera, in contrast to the wild-type receptors (Figure 3). The cell surface expression of the Venus-tagged OX₁ctOX₂ chimera was significantly reduced, compared with that of OX₁ and OX₂ receptors (Figure 2A), which may contribute to the reduction in β-arrestin proximity BRET signal observed. However, as the inositol phosphate generation was not affected (Figure 2C), it is unlikely that cell surface expression alone was responsible for the almost complete abolition of this BRET signal. Indeed, this finding implies that the conformation of the OX₂ C-terminal tail is different (and therefore orients the BRET-tag differently) when attached to the rest of the OX₁ receptor, compared with the conformation when attached to the rest of the OX₂ receptor.

In contrast, although the cell surface expression may have been slightly reduced with the Rluc8-tagged OX₁ctOX₂ chimera (Figure 2B), a BRET signal was observed for proximity to the Venus-tagged β-arrestins (Figure 3C, D). Notably, the resultant kinetic profile of Rluc8-tagged OX₁ctOX₂ was similar to that of OX₁ and not OX₂ receptors. These findings indicate that the primary structure of the OX C-terminal tail is not wholly responsible for determining the stability of the interaction with β-arrestins.

OX₁ and OX₂ receptors share significant homology in primary structure except for the N-terminus, distal region of the C-terminal tail and the third intracellular loop (ICL3) (Sakurai et al., 1998; Voisin et al., 2003). There are very few differences in ICL1 and 2, with those present not likely to have a major effect on β-arrestin recruitment. Regarding ICL3, the only putative high affinity GRK phosphorylation site is conserved (T250, T251, S252 in OX₁ and T258, S259, S260 in OX₂) and is in the proximal region of the loop that is itself highly conserved. Beyond this, OX₁ ICL3 has two serines and a threonine spread through the loop, whereas OX₂ ICL3 has three serines and two threonines, again not clustered. Therefore, from the perspective of primary structure, there are no obvious differences between the ICL3 of OX₁ and OX₂ receptors in terms of high affinity GRK phosphorylation sites. However, it is likely that the conformation of ICL3 differs between the receptor subtypes and how configuration of this with the C-terminus influences high affinity β-arrestin interaction is certainly worthy of future investigation.

It is our hypothesis that the lack of gain of function observed with the OX₁ctOX₂ mutant with respect to β-arrestin binding is due to disruption of secondary or tertiary structure, within or between the intracellular domains. This indicates that appropriate GRK phosphorylation sites need to be not only present, but correctly positioned and orientated for both phosphorylation and β-arrestin interaction. Therefore, although there are notable examples of C-terminal tail chimeras adopting the characteristics of the substituted C-terminal tail, such as β₂-adrenoceptor-V₂-tail or V₂-β₂-adrenoceptor tail chimeras (Oakley et al., 1999; Shenoy and Lefkowitz, 2003; Tohgo et al., 2003) as well as chimeras of β₂-adrenoceptor-AT_1A (Anborgh et al., 2000) and NK₁-PAR2 receptors (Pal et al., 2012), our data show that this is not always the case. Indeed, we have previously published work investigating the effect of extracellular loop substitution on ligand binding and signalling properties of the gonadotrophin-releasing hormone receptor (Pfleger et al., 2008), where interactions between loops appeared to play a role. It is likely that similar interactions between the intracellular loops and C-terminal tail are involved in configuring intracellular binding sites for β-arrestin. Therefore, although the C-terminal tail of OX₂ receptors may function well in the spatial context of the rest of the OX₂ receptor, it may not when set amongst the intracellular loops of the OX₁ receptor. It is also possible that homo- or heteromerization could play a role.

To further elucidate the molecular determinants of orexin receptor-arrestin interactions, three serine/threonine clusters in the C-terminal tail of OX₂ receptors were analysed for their ability to affect OX₂ receptor-β-arrestin binding strength and stability (Figure 4). The BRET-tag orientation of the OX receptor constructs appears to have some influence on their relative cell surface expression levels (Figure 6A, B) and general stability of the kinetic profiles (Figure 5), however, taking this into account, the overall effects of the mutations relative to wild-type are largely consistent, regardless of BRET-tag orientation (Figure 5). Mutation of any of the cluster sites in isolation had little detrimental impact on the initial strength of the receptor-arrestin interaction, with the possible exception of the Δ406 mutant with Rluc8-tagged receptors. This contrasts with receptors such as AT_1A and oxytocin receptors that have a similar complement of phosphorylation clusters, but only require mutation of a single cluster to significantly disturb arrestin translocation (Oakley et al., 2001).

As GRK phosphorylation increases receptor affinity for β-arrestins as a consequence of introducing negative charge, it was postulated that the negatively charged glutamate within the 399 cluster could act as a phosphate mimic, as described previously (Gurevich and Gurevich, 2006). However, our data from the Δ399-E402Q mutant compared to Δ399 (and indeed the triple cluster mutant with and without E402Q) provide evidence against any such role for this residue (Figure 8).

A previous study revealed that mutation of the single serine/threonine cluster in the distal end of the C-terminal tail of OX₁ receptors severely impaired arrestin translocation, but surprisingly mutation of the proximal cluster had little effect (Milasta et al., 2005). Upon comparison of the primary structures of OX₁ and OX₂ receptors, the serine/threonine clusters at 399 and 427 in OX₂ receptor are similarly present at corresponding sites in OX₁ receptor (Figure 10). However, an additional cluster present in OX₂ receptors at position 406 is absent in the same corresponding region of OX₁ receptors (Figure 10). Our findings indicate that mutation of the 406 serine/threonine cluster has the greatest effect on destabilizing the OX₂ receptor-β-arrestin interaction, this being particularly clear with Venus-tagged receptor and Rluc8-tagged β-arrestin 2 (Figures 5C and 7). Indeed, this profile exhibits similarities to that observed in our previous study with OX₁ receptors (Dalrymple et al., 2011). This alteration in kinetics is also demonstrated through arrestin-ubiquitin kinetics where the profiles of OX₂ Δ406 and OX₁ receptors overlap (Figure 9), in contrast to the separation in β-arrestin-ubiquitin kinetics observed for the wild-type receptors (Dalrymple et al., 2011). These data indicate that, in addition to the potential role of C-terminus/ICL3 configuration discussed above, the 406 cluster may contribute to the differentiation of the orexin receptor subtypes, by conferring upon OX₂ receptors the ability to form more stable complexes with β-arrestin and ubiquitin.

Two of the three hypothesized GRK phosphorylation cluster sites, 406 and 427, appear to be involved in achieving stable OX₂ receptor-β-arrestin complexes. Mutation of these clusters in combination appears to render substantial loss-of-function with regard to β-arrestin interaction strength, despite no apparent reduction in cell surface expression as determined by assessment of proximity to the Kras cell surface marker. These findings imply a degree of redundancy, and hint at another potential molecular mechanism to explain the previously observed difference in receptor-arrestin-ubiquitin complex stability with OX₁ compared to OX₂ receptors (Dalrymple et al., 2011). As a result of our findings, we propose a simplistic model that may help to explain some of the complex stability differences, particularly as observed in Figures 5C, 7 and 9. This model would suggest that OX₁ receptors are phosphorylated on the distal serine/threonine cluster, as reported by Milasta et al. (2005), but OX₂ receptors are phosphorylated on the two clusters that we have designated 406 and 427 (Figure 10). Perhaps only one of these sites in the OX₂ receptor requires phosphorylation in order for the receptor to adopt a high-affinity state for arrestin binding. As the receptors are internalized, it is hypothesized that dephosphorylation of the distal serine/threonine cluster of OX₁ receptor switches the receptor to a lower-affinity state for arrestin binding, whereas OX₂ receptor requires both sites to be dephosphorylated for this to occur. This may then extend the time during which individual receptors remain in the high-affinity state for arrestin binding, which is consistent with the OX₂ receptor-arrestin-ubiquitin complex being more stable over time and OX₂ receptors recycling more slowly to the plasma membrane as a consequence (Dalrymple et al., 2011). This is of course not the only possible explanation, and our findings with the C-terminal chimera indicate a role for other parts of the intracellular domain as well, as discussed above.

This study therefore provides fundamental insights into the molecular determinants that govern orexin receptor subtype-specific arrestin-ubiquitin complex formation and stability. Our findings, and the conceptual models they have elicited, provide further potential molecular explanations for our previous observations (Dalrymple et al., 2011) and enable us to begin making key correlations between structure and function.

Conflict of interest

None.