In this issue of the Biochemical Journal, Xu et al. describe how they use a spot peptide array to identify a unique sequence within β-arrestin-2 that is required for both multimerization and ERK1/2 (extracellular-signal-related kinase 1/2) scaffolding. They provide evidence that dimers may serve as more than just ‘storage forms’ of β-arrestins, incapable of interacting with receptors but, rather, perhaps, adding to the specificity of G-protein-coupled-receptor signalling. They show that key charged residues within this dimerization interface of β-arrestin-2 block association with ERK1/2 and subsequent activation of ERK1/2 by β2-adrenergic receptors, while internalization is unaffected. They suggest that self-association may serve as a means of sheltering scaffolding sites on β-arrestins from specific binding partners to prevent constitutive activation of key signalling pathways. These studies enhance our understanding of how β-arrestins can juggle their roles as scaffolds of multiple pathways in response to diverse signals.
β-Arrestins have emerged as incredibly diverse molecules, capable of inhibiting GPCR (G-protein-coupled receptor) signalling to promote desensitization, while simultaneously promoting other downstream signals through localized activation and inhibition of various enzymatic activities [1,2]. It is now widely accepted that the outcome of β-arrestin recruitment to a GPCR varies widely, depending on the receptor, the cell type or the agonist. For some receptors, one β-arrestin predominantly runs the desensitization/internalization side of the show, while the other one scaffolds MAPKs (mitogen-activated protein kinases). For other receptors, both β-arrestins seem to do everything . Over the past few years, the number and diversity of β-arrestin binding partners has increased exponentially [2,3], and although information confirming these interactions and probing their physiological relevance has been forthcoming, one aspect of this story has remained elusive. That is: how do these two proteins carry out so many diverse roles and how do they know when and where to do each of them? There is evidence that (i) β-arrestins can adopt different conformations, (ii) that their interaction with different receptors involves distinct binding sites, (iii) that conformational changes can affect accessibility of β-arrestin to different binding partners and (iv) that β-arrestins can form both homo/heterodimers and tetramers [2,4–6].
Although a differential function for monomeric β-arrestins and β-arrestin homo- and hetero-dimers would seem to be a probable means of accounting for some of their diverse cellular roles, there is limited evidence for a physiological role for β-arrestin oligomers. One stumbling block in elucidating a role for β-arrestin self-association is the apparent concentration required for spontaneous oligomerization. Earlier reports on retinal rod arrestin demonstrated that it crystallizes as a tetramer and forms multimers under physiologically relevant concentrations in solution. However, cellular concentrations of β-arrestins-1 and -2 are much lower than those of rod arrestin. Thus, for some time, researchers believed that β-arrestin oligomers do not form at physiological concentrations and are likely to be negligible unless one proposes the existence of microdomains with extremely high localized concentrations or the presence of other binding partners that promote self-association . Studies by Storez et al. in 2005 demonstrated the existence of constitutive β-arrestin homo- and hetero-dimers in live cells using standard BRET (bioluminescence resonance energy transfer), FLIM (fluorescence lifetime imaging microscopy) and co-immunoprecipitation ; however, other studies have suggested that binding to InsP6 might be required for these dimers to form [5,7]. This would suggest that altering cellular levels of InsP6 might change the equilibrium between monomeric and oligomeric β-arrestins.
Although the current evidence points to the existence of β-arrestin oligomers under physiological conditions, we are still left with the question of their functional significance. In fact, there is still considerable controversy over whether or not β-arrestin oligomers are active or serve as a means of suppressing β-arrestin function until its actions are required by the cell. For rod arrestin, β-arrestin oligomers can bind microtubules, but not their cognate GPCR, rhodopsin, whereas monomeric arrestin binds rhodopsin and mediates desensitization . These findings led investigators to propose that arrestin oligomers were ‘storage forms’ of the molecules. In the proposed structure of inactive β-arrestins, intramolecular interactions between the N- and C-termini are stabilized by a polar core of charged amino acids, preventing interaction of β-arrestin with certain binding partners such as clathrin . Activation by association with a GPCR is thought to disrupt this polar core, thus exposing the C-terminal clathrin-binding site. Studies by Nobles et al.  showed that interaction of β-arrestin-1 with a phosphopeptide derived from the vasopressin II receptor exposes positively charged amino acids involved in InsP6 binding and avails the C-terminus to clathrin binding. If multimers represent inactive ‘storage forms’ of β-arrestin, does receptor activation disrupt them and is the release of monomeric β-arrestin necessary for their function? Can β-arrestin-dependent functions then be toned down if other cellular pathways resulting in increased InsP6-induced multimerization are in place? Alternatively, can β-arrestin dimerization be induced by receptor activation and will these multimers preferentially carry out a subset of possible β-arrestin functions upon receptor activation? Extrapolating on this model, since various binding partners appear to interact with distinct sites on β-arrestins, it stands to reason that self-association could alter which domains are available for binding to different proteins, including the activating receptor itself.
In this issue of the Biochemical Journal, Xu et al.  used a spot peptide array to pinpoint a specific region within β-arrestin-2 distinct from those sites involved in InsP6 binding that are important for self-association. With this information, they were then able to create mutants deficient in dimerization, as indeed disruption of charged residues within this region prevents β-arrestin-2 self-association. Although these data do not rule out the possibility the InsP6 is still required for β-arrestin self-association, they suggest that there are sites of direct contact between β-arrestin molecules. Such domains may either serve to stabilize InsP6-bound dimers or they might provide an alternative means of β-arrestin dimerization, resulting in a potentially distinct conformation. They then probed the functional role of β-arrestin dimers in the signalling by the prototypical GPCR β2AR (β2-adrenergic receptor). Quite unexpectedly, mutations that disrupt β-arrestin-2 oligomerization had no discernable effect on receptor internalization, but markedly inhibited ERK1/2 (extracellular-signal-related kinase 1/2) activation by β2AR and association of ERK1/2 with β-arrestin-2. There are several novel implications in these findings. They suggest that β-arrestin self-association may only impact on a subset of β-arrestin-dependent events. At first glance, it might appear that β-arrestin dimers are required for ERK1/2 activation and not receptor internalization. However, the authors also show that ERK1/2 binds to a region overlapping the dimer interface, supporting the converse hypothesis, namely that only β-arrestin monomers are capable of binding to, and thus promoting activation of, ERK1/2. This would lead one to propose a model wherein, in its inactive state, β-arrestin multimers occlude the ERK1/2 binding sites. Presumably, upon activation, the multimers disperse, unveiling the ERK1/2 binding sites and allowing β-arrestin-dependent ERK1/2 activation. This prediction raises a number of interesting questions. First, does receptor activation decrease β-arrestin dimerization? This may be difficult to answer, because many GPCRs themselves dimerize, thus they could in effect force the formation of complexes containing multiple β-arrestins that are distinct from the β-arrestin dimers described here. Secondly, if the sites identified in the study by Xu et al.  co-operate with InsP6 to stabilize β-arrestin dimers, are there physiological conditions leading to decreased InsP6 levels that allow constitutive signalling from β-arrestin-2 to ERK1/2?
To make the story even more interesting, these same mutations do not affect association of β-arrestin-2 with another MAPK family member, namely JNK3 (c-Jun N-terminal kinase 3). A complex of β-arrestin-2, JNK3 and its upstream activators, MKK4 (MAPK kinase 4) and ASK1 (apoptosis signal-regulating kinase 1), was demonstrated to form constitutively in mammalian cells and was recruited to membrane-bound vesicles upon angiotensin-II-type-Ia-receptor activation. Furthermore, both visual and non-visual arrestins appear to interact with JNK3 in a manner that is independent of receptor-mediated conformational changes . Thus JNK3 may represent a subset of signalling proteins that can interact with β-arrestins in monomeric and dimeric forms. The binding site for JNK3 is within a non-overlapping region with the dimer interface and localized to a region that might remain exposed even in β-arrestin multimers. Since the effect of β-arrestin-multimerization on signalling is relatively selective, it will be very interesting to examine how these same mutations affect interaction of β-arrestins with various other binding partners. For example, Src binds to a proline-rich region within the N-termini of β-arrestins that might also be unaffected by multimerization. Other proteins such as LIMK [LIM (Lin11, Isl-1 and Mec-3) domain kinase], PI3K (phosphoinositide 3-kinase), filamen, Ral-GDS (Ral GDP dissociation stimulator), PP2A (protein phosphatase 2A) and NF-κB (nuclear factor κB) [1,2] may also be differentially regulated by β-arrestin multimerization.
Despite the wealth of evidence that we need to integrate oligomerization of β-arrestins into our evaluation of how they regulate different cellular processes, whether monomeric or oligomeric forms have specific roles in signal perpetuation that can be distinguished from their roles in signal down-regulation has not been fully investigated. In previous studies by Storez et al. , activation of angiotensin II type 1a receptor promoted translocation of forced β-arrestin-2 dimers into vesicles, suggesting that they might be capable of receptor trafficking and/or scaffolding of signalling molecules. In other studies, mutations preventing InsP6 binding to β-arrestin-2 reduced interaction with the ubiquitin ligase Mdm2 (murine double-minute 2) and inhibited the p53-dependent anti-proliferative effects of β-arrestin-2 . Together these studies suggest that the β-arrestin dimers might be functionally active and not merely inhibitory or storage forms of the molecules . In fact, it is equally possible that, if multimers can compete with some proteins for specific binding sites on β-arrestin, they may be conformationally favourable for interaction with others. At any rate, these results provide the first clue as to how some of the distinctly independent functions of β-arrestins are sorted by the cell. Finally, if we superimpose these findings by Xu et al.  on the recent paradigm shift that β-arrestins mediate G-protein-independent signals and may therefore be viable therapeutic targets for pathway-specific drugs , we might predict that one could target a specific subset of arrestin-directed events.