FRET (fluorescence resonance energy transfer) and co-immunoprecipitation studies confirmed the capacity of β-arrestin 2 to self-associate. Amino acids potentially involved in direct protein–protein interaction were identified via combinations of spot-immobilized peptide arrays and mapping of surface exposure. Among potential key amino acids, Lys285, Arg286 and Lys295 are part of a continuous surface epitope located in the polar core between the N- and C-terminal domains. Introduction of K285A/R286A mutations into β-arrestin 2–eCFP (where eCFP is enhanced cyan fluorescent protein) and β-arrestin 2–eYFP (where eYFP is enhanced yellow fluorescent protein) constructs substantially reduced FRET, whereas introduction of a K295A mutation had a more limited effect. Neither of these mutants was able to promote β2-adrenoceptor-mediated phosphorylation of the ERK1/2 (extracellular-signal-regulated kinase 1/2) MAPKs (mitogen-activated protein kinases). Both β-arrestin 2 mutants displayed limited capacity to co-immunoprecipitate ERK1/2 and further spot-immobilized peptide arrays indicated each of Lys285, Arg286 and particularly Lys295 to be important for this interaction. Direct interactions between β-arrestin 2 and the β2-adrenoceptor were also compromised by both K285A/R286A and K295A mutations of β-arrestin 2. These were not non-specific effects linked to improper folding of β-arrestin 2 as limited proteolysis was unable to distinguish the K285A/R286A or K295A mutants from wild-type β-arrestin 2, and the interaction of β-arrestin 2 with JNK3 (c-Jun N-terminal kinase 3) was unaffected by the K285A/R286A or L295A mutations. These results suggest that amino acids important for self-association of β-arrestin 2 also play an important role in the interaction with both the β2-adrenoceptor and the ERK1/2 MAPKs. Regulation of β-arrestin 2 self-association may therefore control β-arrestin 2-mediated β2-adrenoceptor-ERK1/2 MAPK signalling.
β-Arrestin 1 and 2 are ubiquitously expressed members of the arrestin protein family. Although long appreciated to play key roles in the desensitization of function of GPCRs (G-protein-coupled receptors), recent studies have begun to demonstrate a much wider ranging set of roles for the β-arrestins, including interactions with receptors outside of the GPCR family and the generation of distinct, G-protein-independent signals [1–3]. Many of the recently described functions of β-arrestins, including their ability to regulate activation of members of the MAPK (mitogen-activated protein kinase) families [4–7], reflect the capacity of β-arrestins to act as scaffolds for a substantial number of cellular polypeptides. Indeed, proteomic analysis of β-arrestin-interacting proteins has indicated upwards of 200 distinct polypeptides residing in distinct cellular compartments and with wide-ranging functions in signal transduction, cellular organization and nucleic acid binding . These include polypeptides involved in the production and destruction of the secondary messengers cAMP  and diacylglycerol . Many of these interactions have been uncovered via combinations of yeast two-hybrid screens and immunoprecipitation/pulldown studies, and sites of interaction have been mapped by mixtures of protein truncation and mutagenesis techniques. Recently, however, the application of peptide array-based techniques has allowed detailed mapping of potential sites of interactions between β-arrestin 2 and protein-interaction partners such as the cAMP-specific phosphodiesterase PDE4D5 [11,12]. Such screens rapidly focus attention on key regions, and even individual amino acids, that may contribute to or define protein–protein interaction surfaces. Protein self-association is an extremely common theme in biology  and recent studies have indicated that β-arrestin 1 and β-arrestin 2 are able to both homo- and hetero-dimerize [14,15]. β-Arrestins have long been known to interact with phosphoinositides  and recent studies have indicated that, at least in part, binding of IP6 (inositol 1,2,3,4,5,6-hexakisphosphate) to positively charged regions in both the N- and C-terminal domains contributes to β-arrestin 1 and 2 dimerization . In that study the 2.9 Å (1 Å=0.1 nm) atomic level structure of a β-arrestin 1–IP6 complex indicated the existence of two IP6-binding sites per protein monomer. Isothermal titration calorimetry and [3H]IP6-binding assays confirmed two-site binding with a low affinity IP6-binding site in the N-domain and a high-affinity site in the C-domain of the arrestin. Therefore IP6 appears to act as a bridge between β-arrestin monomers to generate dimers and, potentially, as has also been observed for members of the GPCR superfamily  that interact with arrestins, to generate higher-order β-arrestin structures and arrays. Structural models of β-arrestin–β-arrestin interactions  also predict that direct protein–protein contacts are likely to contribute to the stability and effectiveness of β-arrestin dimerization/oligomerization.
To identify regions and specific amino acids important in β-arrestin 2 self-association that are distinct from the IP6-binding domains we have employed combinations of spot-immobilized peptide arrays with subsequent supporting mutagenesis and combinations of FRET (fluorescence resonance energy transfer) and co-immunoprecipitation studies. Mutation of positively charged residues in a surface-exposed, polar core region between the N- and C-terminal IP6-binding domains, which was identified via peptide array studies, limited the effectiveness of β-arrestin 2 self-association. Such mutants of β-arrestin 2 also interacted poorly with both the prototypic β-arrestin-interacting GPCR, the β2-adrenoceptor and with the ERK1/2 (extracellular-signal-regulated kinase 1/2) MAPKs. This resulted in an inability of the modified forms of β-arrestin 2 to support β2-adrenoceptor-mediated, β-arrestin 2-dependent ERK1/2 MAPK phosphorylation. These studies indicate that sites other than the IP6-binding domains are important for β-arrestin 2 self-association, that the quaternary structure of β-arrestin 2 plays a key role in β2-adrenoceptor and ERK MAPK interactions and suggest that regulation of β-arrestin 2 self-association may, therefore, control β-arrestin 2-mediated β2-adrenoceptor-ERK1/2 MAPK signalling.
Antibodies and radiochemicals
Monoclonal anti-VSV-G (vesicular-stomatitis virus-G) antibody and a polyclonal anti-FLAG antiserum were from Sigma. The polyclonal anti-β2-adrenoceptor antibody (H-73) and one of the β-arrestin antibodies used were from Santa Cruz Biotechnology. Monoclonal p44/42 MAPK (ERK1/2) and phospho-p44/42 MAPK (phospho-ERK1/2) antibodies were from Cell Signalling Technology. A second anti-β-arrestin antibody was a gift from Dr R. J. Lefkowitz (Howard Hughes Medical Institute, Duke University, Durham, NC, U.S.A.). Lipofectamine™ 2000 transfection reagent was from Invitrogen. Protease inhibitor cocktail tablets were from Roche. [3H]CGP-12177 (44 Ci·mmol−1) was purchased from GE Healthcare.
Cell culture and transfection
HEK (human embryonic kidney)-293 cells were maintained in DMEM (Dulbecco's modified Eagle's medium) supplemented with 0.292 g/litre L-glutamine and 10% (v/v) newborn calf serum at 37 °C in a 5% CO2 humidified atmosphere. A HEK-293 cell line stably expressing the human β2-adrenoceptor has been described previously  and was maintained as above but with the addition of 800 μg/ml G418 (Geneticin). Cells were grown to 50–70% confluence before transient transfection. Transfections were performed using Lipofectamine™ 2000 reagent (Invitrogen) according to the manufacturer's protocol.
Plasmid construction and site-directed mutagenesis
Bovine β-arrestin 2 was used as a template to generate β-arrestin–eCFP (where eCFP is enhanced cyan fluorescent protein) and β-arrestin–eYFP (where eYFP is enhanced yellow fluorescent protein) via overlapping PCR. The first set of primers were 5′-ACTATGACGACCAGTTCTGTGTGAGCAAGGGCGAGGAGCT-3′ (forward) and 5′-AGCTCCTCGCCCTTGCTCACACAGAACTGGTCGTCATAGT-3′ (reverse). The overlapping PCR products were amplified by the second set of primers with KpnI and ApaI sites and were inserted into pCDNA3, the primers were 5′-CGATGGTACCCCATGGGGGAGAAACCCGGGACCAGGGT-3′ (forward; the KpnI site is underlined) and 5′-TGCTGGGCCCTTACTTGTACAGCTCGTCCATGCCGAGAG (reverse; the ApaI site is underlined). eCFP–β-arrestin-2 was a gift from Dr Juan F. Lopez-Gimenez (University of Glasgow). VSV-G–β-arrestin-2 was generated by PCR and inserted into the KpnI–ApaI site of pCDNA3. The primers were 5′-CGATGGTACCCCGCCACCATGTACACCGATATAGAGATGAACCGCCTTGGAAAGGGGGAGAAACCCGGGACCAGGGTC-3′ (forward with VSV-G sequence; the KpnI site is underlined) and 5′-TGATGGGCCCTCAACAGAACTGGTCGTCATAGTCCTCGT (reverse; the ApaI site is underlined). FLAG–β-arrestin-2 was generated by PCR and also inserted into the KpnI–ApaI site of pCDNA3. The primers were 5′-CGATGGTACCCCATGGACTACAAGGACGACGATGATAAGGGGGAGAAACCCGGGACCAGGGT-3′ (forward with FLAG sequence; the KpnI site is underlined) and 5′-TGATGGGCCCTCAACAGAACTGGTCGTCATAGTCCTCGT (reverse; the ApaI site is underlined).
Site-directed mutagenesis was performed using the QuikChange® II site-directed mutagenesis kit (Stratagene) according to the manufacturer's protocol. The K285A/R286A double mutation of β-arrestin-2 was introduced using the primers 5′-TCAGCAACAACCGGGAGGCGGCCGGCCTCGCTCTGGAT-3′ (forward) and 5′-ATCCAGAGCGAGGCCGGCCGCCTCCCGGTTGTTGCTGA-3′ (reverse). The primers for the K295A mutation of β-arrestin-2 were 5′-CTCTGGATGGGAAGCTCGCGCACGAGGACACCAACC-3′ (forward) and 5′GGTTGGTGTCCTCGTGCGCGAGCTTCCCATCCAGAG-3′ (reverse).
Live cell FRET imaging
HEK-293 cells grown on poly-D-lysine-treated glass coverslips were transiently transfected with combinations of eCFP- and eYFP-tagged forms of β-arrestin 2. At 18–24 h after transfection, cells were transferred to a microscope chamber containing Hepes-buffered saline solution [130 mM NaCl, 5 mM KCl, 1 mM CaCl2, 1 mM MgCl2, 20 mM Hepes and 10 mM D-glucose (pH 7.4)]. eCFP- and eYFP-expressing cells were excited using an Optoscan monochromator (Cairn Research) which was set to 500/5 and 430/12 nm for the sequential excitation of eYFP and eCFP respectively. A dual dichroic mirror (86002v2bs; Chroma) was used to reflect the excitation light through a Nikon ×40 (NA=1.3) oil-immersion Fluor lens. The resultant eCFP and eYFP fluorescence emission signals were detected using a high-speed filterwheel device (Prior Instruments) containing the following band pass emitters: HQ470/30 m for eCFP and HQ535/30 m for eYFP.
Bleedthrough coefficients were measured from cells expressing either eCFP or eYFP alone and were quantified by dividing the amount of fluorescence detected in the FRET channel (e.g. FRETeCFP-eYFP) by the fluorescence detected from each fluorescent protein, at its own filter channel (e.g. eCFP or eYFP). Corrected FRET (FRETc) was calculated using a pixel-by-pixel methodology using the equation
where eCFP, eYFP and FRET values correspond to background corrected images obtained through the eCFP, eYFP and FRET channels. B and A correspond to the values obtained for the eCFP (donor) and eYFP (acceptor) bleedthrough coefficients respectively. To correct the FRET levels for the varying amounts of donor (eCFP) and acceptor (eYFP), normalized FRET (FRETN) was calculated using the equation
The crystal structure (PDB number: 1ZSH) of bovine arrestin-2 (β-arrestin 1) with bound IP6 was used to locate basic residues from array peptide-57 implicated in arrestin self-association. Rendered surfaces were generated using PyMol (DeLano Scientific LLC; http://www.pymol.org).
Peptide arrays and alanine scans
A β-arrestin-2 peptide library was produced by automatic SPOT synthesis as described previously [20,21]. It was synthesized on continuous cellulose membrane supports on Whatman 50 cellulose membranes using Fmoc (fluoren-9-ylmethoxycarbonyl) chemistry with the AutoSpot-Robot ASS 222 (Intavis Bioanalytical Instruments). Alanine-scanning peptide libraries were constructed by taking the residues in positive spots and sequentially changing each residue to alanine (or, if an alanine was the natural amino acid at that position, to aspartate). The interaction of spotted peptides with purified recombinant GST (glutathione transferase) and GST–β-arrestin-2 fusion proteins was determined by overlaying the cellulose membranes with 10 μg/ml recombinant protein. Bound recombinant proteins were then detected following wash steps with rabbit anti-GST and anti-β-arrestin antisera, and detection was performed with a secondary anti-rabbit HRP (horseradish peroxidase)-coupled antibody. Similar studies explored interactions of the β-arrestin-2 peptide library with ERK2 (Millipore 14-198). The relative intensity of interaction of the recombinant proteins with the arrayed peptides was estimated by densitometry.
Cells were harvested 24 h after transfection and resuspended in immunoprecipitation buffer (150 mM NaCl, 0.01 mM NaPO4, 2 mM EDTA, 0.5% Triton X-100 and 5% glycerol plus protease inhibitor cocktail tablets). The cell pellet was lysed on a rotating wheel for 30 min at 4 °C. Samples were then centrifuged for 15 min at 20000 g at 4 °C, and the supernatant was transferred to a fresh tube with Protein G/A beads (Sigma) to preclear the samples. After incubation on a rotating wheel for 1 h at 4 °C, the samples were re-centrifuged at 20000 g at 4 °C for 1 min, and the protein concentration of the supernatant was determined. Samples containing equal amounts of protein were incubated for 2 h with 30 μl of anti-VSV-G agarose beads (Sigma) at 4 °C on a rotating wheel. Samples were then washed three times with immunoprecipitation buffer. After addition of 2×SDS loading buffer and heating to 90 °C for 5 min, both immunoprecipitated samples and cell lysate controls were resolved by SDS/PAGE and subsequently immunoblotted to detect proteins of interest.
Limited proteolysis studies
HEK-293 cells were transfected with wild-type β-arrestin-2 or its mutants, 24 h after transfection, cells were lysed in lysis buffer [150 mM NaCl, 0.01 mM NaPO4, 2 mM EDTA, 0.5% Triton X-100 and 5% glycerol plus protease inhibitor cocktail tablets (pH 8.0)] on a rotating wheel for 30 min at 4 °C. Samples were then centrifuged for 15 min at 20000 g at 4 °C. Protein (60 μg) from the supernatant fraction was incubated with 0.6 μg of sequencing grade trypsin (Promega) at 37 °C for various times as indicated. Digestion was terminated by the addition of 2×SDS loading buffer and heating at 95 °C for 5 min. Samples were then resolved by SDS/PAGE and subsequently immunoblotted with a β-arrestin-2 antibody (Santa Cruz Biotechnology).
HEK-293 cells stably expressing the β2-adrenoceptor  were transfected transiently with VSV-G-tagged forms of wild-type and mutated β-arrestin-2. Cells were then treated with or without isoprenaline. Subsequently, the membrane-permeable and reversible cross-linker DTSP [dithiobis(succinimidyl propionate)] was added at a final concentration of 2 mM. The cells were then incubated under gentle agitation at room temperature (18 °C) for 30 min and washed twice with 50 mM Tris/HCl (pH 7.4), in PBS to neutralize unreacted DTSP. The cells were then lysed and immunoprecipitated as above.
β2-Adrenoceptor internalization studies were performed as in . Briefly, cells were co-transfected with PCDNA3-β2-adrenoceptor plus β-arrestin 2 or its mutants in 24-well plates. At 24 h after transfection, cells were treated with 10 μM isoprenaline or vehicle for 30 min. The cells were then washed twice with ice-cold PBS (pH 7.4) followed by addition of 200 μl of binding mix (10 nM [3H]CGP-12177) in Krebs-Ringer-Hepes buffer [130 mM NaCl, 5 mM KCl, 1.2 mM MgSO4, 1.2 mM CaCl2, 20 mM Hepes,1.2 mM Na2PO4, 10% (w/v) glucose and 0.1% BSA (pH 7.4)], and incubated at 4 °C for 90 min. Non-specific binding was assessed by the presence of 10 μM propranolol. All experiments were terminated by the removal of the binding medium and washing of the cells with ice-cold PBS. To detach cells from the plate, 0.5 ml of 0.5 mM EDTA in PBS was added and this volume plus a further 0.5 ml wash of the wells of the microtitre plate were counted.
ERK1/2 MAPK phosphorylation assays
HEK-293 cells stably expressing the β2-adrenoceptor were grown in 6-well plates and transiently transfected with various forms of β-arrestin 2. The cells were rendered quiescent by serum starvation for 12 h prior to stimulation for the indicated times with or without isoprenaline. Cells were then placed on ice and solubilized directly in 500 μl of Laemmli sample buffer. The samples were then sonicated for 30 s, heated for 15 min at 95 °C, and microcentrifuged for 5 min before fractionation of the proteins on SDS/PAGE. ERK1/2 MAPK phosphorylation was detected by protein immunoblotting using a phospho-ERK1/2-specific antibody (Cell Signalling Technology). The nitrocellulose membranes were subsequently stripped of immunoglobulins and re-probed using an anti-ERK1/2 antibody to assess the equivalence of protein loading.
β-Arrestin 2 has been reported to self-associate [14,15]. To examine the basis of such interactions, we set out to gain insight into amino acid residues that are important for this interaction. The aim was to generate self-association-compromised β-arrestin 2 point mutants and determine how interfering with self-association might affect function. We chose two complementary strategies, FRET and classical co-immunoprecipitation experiments to detect self-association. For FRET experiments, bovine β-arrestin 2 was tagged at the C-terminus with eYFP and with eCFP at either the N or C-terminus. Following transfection of either individual forms or pairs of constructs into HEK-293 cells, imaging of eCFP and eYFP autofluorescence demonstrated each of the constructs to be predominantly cytoplasmic and excluded from the nucleus (Figure 1A and results not shown). Measurements of eCFP-to-eYFP FRET demonstrated a significant signal that was consistent with the self-association of β-arrestin 2. The FRET signal obtained was substantially greater when the fluorescent protein tags were each located at the C-terminus of β-arrestin 2 (Figure 1B) than when C-terminally tagged eCFP–β-arrestin 2 and N-terminally tagged β-arrestin 2–eYFP were co-expressed (Figure 1B). Although other interpretations are possible, this observation suggests that β-arrestin 2 self-association might be sterically constrained to favour conformations where the C-termini are juxtaposed.
FRET studies indicate that β-arrestin 2 is able to self-associate
Spot-immobilized peptide array analysis has proved to be a powerful novel technology to gain insight into sites that putatively underpin protein–protein interaction. Indeed, we have recently exploited this technique to identify sites that define the interaction of the PDE4D5 isoform with the signalling scaffold proteins, β-arrestin and RACK1 (receptor for activated C-kinase 1) [11,12] and between PDE4B1 and the scaffold protein, DISC1 (disrupted in schizophrenia 1) whose gene disruption is implicated in schizophrenia . 25-mer peptides covering the entire primary sequence of β-arrestin 2 were spot synthesized on cellulose membranes with a 5-amino-acid moving frameshift. These immobilized peptide arrays were probed with recombinantly expressed GST–β-arrestin 2 or, as a control, with GST alone. Subsequent to washing, the peptide arrays were exposed to either an anti-β-arrestin antiserum or an anti-GST antiserum and interactions between GST–β-arrestin 2 and the target peptides were monitored immunologically. Neither the anti-GST nor β-arrestin antiserum identified interactions on the peptide array probed with GST alone (Figure 2 and results not shown). However, both the anti-GST and anti-β-arrestin antisera identified specific spots, a number of which were non-contiguous, on the peptide arrays that had been overlaid with GST–β-arrestin 2 (Figure 2 and results not shown). A number of the spots identified by the two antisera were the same and these represent regions potentially involved in β-arrestin 2 self-association. This included spot 57 (amino acids 281–305) (Figure 2).
Detection of potential β-arrestin 2 self-association elements using spot-immobilized peptide arrays
The peptide array represents linear 25-amino-acid segments of β-arrestin 2 that are not necessarily surface-exposed in the mature folded protein. However, regions involved in self-association must necessarily be at the surface of the protein. We therefore overlaid the residues of β-arrestin 2 identified via the peptide array on the surface of β-arrestin 1 (PDB number: 1ZSH) (Figure 3) in order to define regions identified by the peptide array studies that are surface-exposed and, hence, most likely to be available for protein interactions. These included residues within the sequence encompassing amino acids 281–305 of β-arrestin 2, but not a number of the other peptides suggested by the array interaction studies (results not shown). We subsequently narrowed the search for key interacting amino acids within amino acids 281–305 of β-arrestin 2 by generating further spot-immobilized peptide arrays that incorporated an alanine scan throughout this sequence (Figure 4). For residues 289 and 302 that are naturally alanine residues, these were replaced by aspartate residues. Three clear and distinct sets of patterns were observed in the alanine scan of this region. Compared with the wild-type sequence, certain alterations, e.g. N282A, L288A, L290A and L294A, had little or no effect on the capture of GST–β-arrestin 2, whereas certain alterations, e.g. D281A, E284A and D291A, increased the interaction. Most importantly, changes of R283A, K285A, R286A, K293A and K295A all decreased or even ablated interaction with the β-arrestin 2 probe (Figure 4). The point mutations that substantially decreased these interactions were then overlaid on to the surface of the β-arrestin 1 structure (Figure 3). Of these, Arg283, Lys285, Arg286 and Lys295 were clearly surface-exposed and Lys285, Arg286 and Lys295 formed a contiguous ribbon (Figure 3) in a region between the N- and C-terminal IP6-binding domains that have recently been suggested to play an important role in β-arrestin self-association . It should be noted that, while a role of IP6 for β-arrestin 2 self-association has been demonstrated , this is likely to provide a priming/facilitating role that is substantiated by direct amino-acid-mediated interactions. No IP6 was employed in the peptide array analyses, suggesting that these experiments identify core β-arrestin 2 self-association surfaces.
A model of β-arrestin 2 highlighting surface residues potentially involved in self-association
An alanine scanning peptide array of residues 281–305 of β-arrestin 2 identifies a number of key amino acids involved in self-association
To explore the roles of Lys285, Arg286 and Lys295 in β-arrestin 2 self-association in more detail, both K285A/R286A β-arrestin 2 and K295A β-arrestin 2 mutants were C-terminally tagged with either eCFP or eYFP. Following transfection into HEK-293 cells, eCFP-to-eYFP FRET was again measured. Compared with the FRET signal obtained when wild-type forms of β-arrestin 2–eCFP and –eYFP were co-expressed, co-expression of K285A/R286A β-arrestin 2–eCFP with wild-type β-arrestin 2–eYFP resulted in a lower normalized FRET signal and this was further reduced following co-expression of K285A/R286A β-arrestin 2–eCFP with K285A/R286A β-arrestin 2–eYFP (Figure 5). The differences were statistically significant. The co-expression of K295A β-arrestin 2–eCFP with either wild-type β-arrestin 2–eYFP or K295A β-arrestin 2–eYFP resulted in a trend towards a reduced FRET signal (Figure 5) but this was not statistically significant. To further explore the importance of Lys285, Arg286 and Lys295 in the self-association of β-arrestin 2, wild-type and mutated forms of β-arrestin 2 were N-terminally epitope-tagged with either the FLAG or VSV-G sequences. Following co-transfection of wild-type FLAG– and VSV-G–β-arrestin 2 into HEK-293 cells, immunoprecipitation with an anti-VSV-G antiserum resulted in effective co-immunoprecipitation of FLAG–β-arrestin 2 (Figure 6). Interestingly, in the SDS/PAGE conditions used, approx. 50% of the co-immunoprecipitated FLAG–β-arrestin 2 migrated with a molecular mass of 45 kDa, consistent with the anticipated size of the monomeric protein. However, most of the rest of the FLAG–β-arrestin 2 migrated with a molecular mass close to 90 kDa, consistent with the maintained presence of an SDS-resistant dimer (Figure 6). There was also some evidence for a yet higher level organization of β-arrestin 2, as FLAG-immunoreactive material of a higher apparent molecular mass was also detected in the co-immunoprecipitated samples (Figure 6). Co-transfection of FLAG– and VSV-G–K285A/R286A β-arrestin 2 into HEK-293 cells demonstrated that the mutations were without effect on construct expression levels. However, co-immunoprecipitation of β-arrestin 2 was substantially decreased for the various mutant forms (Figure 6). Similar reductions in, but not abolition of, co-immunoprecipitation was observed with tagged forms of K295A β-arrestin 2 (Figure 6) and this was not further affected by combination of the K285A/R286A and K295A mutations (Figure 6). It was clearly possible that alterations in the observed FRET and co-immunoprecipitation studies reflected alterations in folding of the β-arrestin 2 mutants. Although it is impossible to exclude potential folding and maturation effects without detailed analysis, we employed limited proteolysis to detect any gross perturbation of β-arrestin 2 structure associated with the mutations studied. We were unable to detect differences in the rate or pattern of limited tryptic fragmentation (results not shown).
FRET studies suggest altered self-association of K285A/R286A β-arrestin 2 whereas K295A β-arrestin 2 is less affected
K285A/R286A β-arrestin 2 and Lys295Ala β-arrestin 2 display markedly reduced capacities to co-immunoprecipitate
β-Arrestin 2 is known to interact with many GPCRs in both a constitutive and receptor-agonist-dependent manner. The β2-adrenoceptor has been perhaps the most extensively studied. To explore functional consequences of the β-arrestin 2 self-association interface mutations, HEK-293 cells stably expressing the β2-adrenoceptor  were transfected transiently with wild-type VSV-G–β-arrestin 2. Following treatment of the cells with the cell-permeant cross-linker DTSP, VSV-G immunoprecipitation allowed co-immunoprecipitation of the β2-adrenoceptor (Figure 7A). As anticipated from the known role of β-arrestin 2 in interactions with agonist-activated GPCRs, the amount of β2-adrenoceptor co-immunoprecipitated in such experiments was increased substantially when cells were first exposed to the β-adrenoceptor agonist isoprenaline (10 μM for 10 min) (Figure 7A). In contrast, when the β2-adrenoceptor-expressing HEK-293 cells were transfected with either VSV-G–K285A/R286A β-arrestin 2 or VSV-G–K295A β-arrestin 2, the amount of β2-adrenoceptor that could be co-immunoprecipitated with the VSV-G-tagged β-arrestin 2 was reduced substantially, whether the studies were performed with or without pre-exposure to isoprenaline (Figure 7A). Despite the reduced interaction of the β-arrestin 2 mutants with the β2-adrenoceptor this was insufficient to limit their ability to enhance isoprenaline-mediated internalization of the β2-adrenoceptor from the surface of HEK-293 cells when the wild-type and mutated forms of β-arrestin 2 were transfected transiently into these cells (Figure 7B).
K285A/R286A β-arrestin 2 and K295A β-arrestin 2 interact poorly with the β2-adrenoceptor but do not limit internalization of the receptor
It has been suggested that interactions with β-arrestin 2 can be an important element in the ability of many GPCRs, including the β2-adrenoceptor, to elicit the phosphorylation and activation of the MAPKs ERK1 and ERK2. Even without introduction of VSV-G–β-arrestin 2, addition of isoprenaline to the β2-adrenoceptor expressing HEK-293 cells resulted in a rapid enhancement of ERK1/2 phosphorylation (Figure 8). Transfection of wild-type VSV-G–β-arrestin 2 resulted in a marked enhancement of ERK1/2 phosphorylation in response to isoprenaline (Figure 8). In contrast, introduction of either VSV-G–K285A/R286A β-arrestin 2 or VSV-G–K295A β-arrestin 2 was unable to increase isoprenaline-mediated ERK1/2 phosphorylation (Figure 8). Because the effects of the K285A/R286A and K295A β-arrestin 2 mutations on the ability of isoprenaline to promote ERK1/2 phosphorylation were virtually complete, whereas interaction of these mutated forms with the β2-adrenoceptor was only partially compromised, we explored the contribution of this region of β-arrestin 2 to direct interactions with ERK2. Spot-immobilized peptide arrays, akin to those employed above to define sites of β-arrestin 2 self-association, were overlaid with recombinant GST–ERK2. Following washing, the arrays were probed with an anti-ERK1/2 antibody. Peptides 54 and particularly 55 (amino acids 271–295) of the β-arrestin 2 sequence were identified (Figure 9). A peptide array incorporating an alanine scan through this segment of β-arrestin 2 identified a number of amino acids, including both Lys285 and Arg286, where alteration to alanine reduced the interaction with GST–ERK2 (Figure 9). Furthermore, replacement of Lys295 with an alanine residue in such arrays virtually eliminated detectable interactions with ERK2 (Figure 9). Following transfection of HEK-293 cells with VSV-G-tagged β-arrestin 2, immunoprecipitation of cell lysates with anti-VSV-G caused the co-immunoprecipitation of ERK1/2 (Figure 10). Transfections with either VSV-G–K285A/R286A β-arrestin 2 or VSV-G–K295A β-arrestin 2 resulted in similar levels of expression of these forms of β-arrestin 2 as wild-type but, following anti-VSV-G immunoprecipitation, substantially lower levels of ERK1/2 were co-immunoprecipitated (Figure 10). This suggests an important role of this surface-exposed region of β-arrestin 2 in interactions with ERK1 and ERK2. Loss of interaction with ERK1 and ERK2 was selective. Neither VSV-G–K285A/R286A β-arrestin 2 nor VSV-G–K295A β-arrestin 2 displayed reduced ability to interact with JNK3 (c-Jun N-terminal kinase 3) in co-immunoprecipitation studies (Figure 11).
K285A/R286A β-arrestin 2 and K295A β-arrestin 2 are unable to enhance β2-adrenoceptor-mediated ERK1/2 MAPK phosphorylation
Detection of potential ERK2 interaction sites in β-arrestin 2 using spot-immobilized peptide arrays
Interactions between β-arrestin 2 and ERK1/2: K285A/R286A β-arrestin 2 and K295A β-arrestin interact poorly
Both K285A/R286A β-arrestin 2 and K295A β-arrestin 2 interact with JNK3 as well as wild-type β-arrestin 2
Dimerization or multimerization is one of the most common themes in biology  and can involve either two or more copies of the same polypeptide (homodimerization/multimerization), or interactions between different polypeptides with varying degrees of relatedness (heterodimerization/multimerization). Recent studies utilizing FRET have demonstrated the capacity of β-arrestin 1 to both self-associate and to interact with the closely related and also ubiquitously expressed arrestin, β-arrestin 2 . The FRET studies we report in the present study confirm the capacity of β-arrestin 2 to self-associate and, although indirect, suggest that β-arrestin 2 self-association might be sterically constrained to favour conformations where the C-termini are juxtaposed. We also demonstrate that mutations of positively charged amino acids at the surface limit β-arrestin 2 self-association and interfere with interactions with the ERK1/2 MAPKs and with the β2-adrenoceptor. As these same elements of β-arrestin 2 are important for each of these interactions this may explain why cytoplasmic β-arrestin 2 is not in a conformation to constitutively activate ERK1/2. Indeed, ERK1/2 would probably be shielded from interactions with β-arrestin 2 until the arrestin was in monomeric state. Although interactions between β-arrestin 2 and ERK1/2 have previously been reported , until now the site of this interaction on β-arrestin 2 has not been mapped. Equally, the mutated forms of β-arrestin 2 in the present study appear to interact relatively poorly with the β2-adrenoceptor. However, it remains to be established whether this is a general feature of β-arrestin 2–GPCR interactions or that the specific mutations we have explored in the present study are relevant only to interactions with the β2-adrenoceptor. Importantly, the mutations introduced did not modulate interaction with all proteins that are well-known to be scaffolded by β-arrestin 2. For example, interactions with JNK3 were unaltered, suggesting, as did the lack of variation in limited proteolysis studies, that the mutations generated in β-arrestin 2 did not simply result in a generalized unfolding or denaturation. Given that the JNK3 interaction has previously been mapped to the RRS motif of β-arrestin 2, between amino acids 195–202 [1,5], the lack of effect of the K285A/R286A and K295A mutants on interactions with JNK3 are not unexpected and are consistent with no overall disruption of this scaffold.
Although many proteins that are central to signal transduction processes are known to have the capacity to self-associate and/or to interact with other polypeptides, in many cases little is known about the key amino acids involved in the interaction interface(s). Such information is, however, of utmost importance to inform the rational design of mutants that may limit protein–protein interaction and hence permit an assessment of the functional importance of these interactions. Recently, spot-immobilized peptide arrays have provided a valuable approach to focus attention on regions of proteins that may contribute to such interactions [11,12,20–23]. Importantly, we have shown via co-immunoprecipitation studies that effective β-arrestin 2 self-association is important for interaction with the β2-adrenoceptor and, therefore, potentially with other GPCRs. Moreover, in recent studies Boularan et al.  have shown that interfering with the oligomerization of β-arrestin 2 by mutation of IP6-binding sites limits effective interactions of the arrestin with the protooncogene Mdm2 (murine double minute 2), a previously characterized β-arrestin 2 binding partner [25–27]. Given the vast array of β-arrestin-interacting partners identified via proteomic approaches , it is evident that deciphering the importance of β-arrestin 2 monomers compared with oligomers in scaffolding protein complexes, and the identity of β-arrestin subpopulations characterized by specific cohorts of partners in a particular complex, will be a substantial undertaking. Furthermore, although the traditional role of β-arrestins in agonist-induced GPCR desensitization and internalization has been supplemented in recent times by an appreciation of their role in the initiation of distinct, G-protein-independent signalling cascades [1–4], the contribution of the β-arrestin monomer/dimer/oligomer equilibrium in defining these distinct processes remains to be established. The results provided in the present study indicate that β-arrestin 2 self-association may be critically important for transmitting signals from the β2-adrenoceptor to the ERK1/2 MAPKs and, in conjunction with the recent results of Boularan et al. , demonstrate that the effects on downstream effectors may be limited to specific cohorts of scaffolded proteins, thereby instilling pathway selectivity.
These studies were supported by grants G040005, G9811527 and G0600765 from the Medical Research Council (U.K.). We thank Dr John Pediani for assistance with the FRET imaging studies.
enhanced cyan fluorescent protein
extracellular-signal-regulated kinase 1/2
enhanced yellow fluorescent protein
fluorescence resonance energy transfer
human embryonic kidney
c-Jun N-terminal kinase 3
mitogen-activated protein kinase
Present address: AstraZeneca R&D Charnwood, Bakewell Road, Loughborough, Leicestershire LE11 5RH, U.K.