Hetero-oligomers of α2A-adrenergic and μ-opioid receptors do not lead to transactivation of G-proteins or altered endocytosis profiles

Y.Q. Zhang, L.E. Limbird


Complexes of α2A-ARs (α2A-adrenergic receptors) and MORs (μ-opioid receptors), probably hetero-oligomers, were detected by co-immunoisolation after extraction from HEK-293 cells (human embryonic kidney 293 cells). Functional communication between these receptors is revealed by α2A-AR activation of a pertussis toxin-insensitive Giα subunit (termed as Gi1) when fused with the MOR and evaluated in membranes from pertussis toxin-treated cells. However, the α2A-AR does not require transactivation through MOR, since quantitatively indistinguishable results were observed in cells co-expressing α2A-AR and a fusion protein of Gi1 with the first transmembrane span of MOR (myc–MOR-TM1). Functional cross-talk among these α2A-AR–MOR complexes does not occur for internalization profiles; incubation with adrenaline (epinephrine) leads to endocytosis of α2A-AR but not MOR, while incubation with DAMGO ([D-Ala,NMe-Phe,Gly-ol]enkephalin) leads to endocytosis of MOR but not α2A-AR in cells co-expressing both the receptors. Hence, α2A-AR and MOR hetero-oligomers, although they occur, do not have an obligatory functional influence on one another in the paradigms studied.

  • α2A-adrenergic receptor
  • G-protein-coupled receptor
  • hetero-oligomer
  • μ-opioid receptor


An emerging literature has demonstrated the homo- and hetero-oligomerization of GPCRs (G-protein-coupled receptors). In some cases, formation of these oligomers has been paralleled by functional changes among the participating GPCRs, such as changes in ligand selectivity [1], altered downstream signalling [2] or changes in the profiles of receptor-mediated endocytosis [3]. These findings have been summarized in a number of excellent reviews [46].

We were particularly interested in the possible hetero-oligomerization of α2A-ARs (α2A-adrenergic receptors) with MORs (μ-opioid receptors) or δ-opioid receptors, since there is considerable synergism of these receptors in vivo. For example, mutation of the α2A-AR in the mouse by homologous recombination to a defective receptor structure, D79N (Asp79→Asn) α2A-AR, results not only in a remarkably diminished sensitivity to the antinociceptive response to α2-agonists but also in a significant shift in the sensitivity to agonists directed at MOR or δ-opioid receptor [7]. One molecular mechanism for this α2-adrenergic–opioid synergism might relate to formation of hetero-oligomers among these α2A-ARs and opioid receptors. Consequently, we undertook studies to determine whether hetero-oligomers existed among α2A-AR and MOR and, if they occurred, their functional relevance. During the course of these studies, Jordan et al. [8] reported the existence of detectable hetero-oligomerization among α2A-AR and MOR after overexpression in fibroblasts or primary cultures of hippocampal or spinal cord neurons.



cDNA encoding the FLAG-tagged mouse MOR cDNA and cDNA encoding the hMOR (human MOR)–Gi1 G-protein (Cys-351 of Gi1 was mutated to Ile) fusion (hMOR–Gi1) were gifts from Dr L. Devi (Mt Sinai School of Medicine, New York, NY, U.S.A.) and Dr G. Milligan (University of Glasgow, Scotland, U.K.) respectively. The first transmembrane domain (TM1) of the MOR was fused with Gi1 in-frame (Myc–TM1μMOR–Gi1) and was generated from hMOR–Gi1 in several steps. First, hMOR–Gi1 was subcloned in-frame into pCMV-Myc vector (BD Biosciences, Palo Alto, CA, U.S.A.) to generate Myc–hMOR–Gi1 and, subsequently, an ApaI site was created just before the start codon of the Gi1 in Myc–hMOR–Gi1 using site-directed mutagenesis. Finally, PCR was used to amplify the first transmembrane domain of hMOR (amino acids 1–105) from Myc–hMOR–Gi1 using the N-terminal primer (5′-TCTGCTCTAAAAGCTGCGGAATTG-3′) and the C-terminal primer (5′-TAGGGCCCGGTGGCAGTCTTCATCTTGGTG-3′) containing an ApaI site (underlined). This PCR product was cloned into Myc–hMOR–Gi1 to substitute the Myc-fused full-length hMOR with only TM1 of the vector. In this fusion (Myc–TM1–Gi1), a two-amino-acid spacer (Gly-Pro) containing an ApaI site was introduced immediately after the codon for the 105th amino acid of the hMOR and before the start codon of the Gi1.

Cell culture and transfection

HEK-293 cells (human embryonic kidney 293 cells) were maintained in Dulbecco's modified Eagle's medium, supplemented with 2 mM glutamine, 10% (v/v) foetal calf serum and PenStrep. Cells were grown in 60 or 100 mm dishes and transfection was performed using either FuGENE 6 (Roche Diagnostics Corporation, Indianapolis, TN, U.S.A.) for immunoprecipitation or a calcium phosphate transfection kit (Invitrogen, Carlsbad, CA, U.S.A.) for GTP[S] binding assay, according to the manufacturer's instructions. Cells were analysed 48 h after FuGENE 6 transfection and 72 h after calcium phosphate transfection. For the GTP[S] binding assay, cells were treated overnight with pertussis toxin at 100 ng/ml.

Co-immunoisolation and Western blotting

HEK-293 cells co-expressing MOR (FLAG–MOR) and α2A-AR (HA–α2A-AR, where HA stands for haemagglutinin) were lysed for 1 h in lysis buffer [1.0% Nonidet P40, 150 mM NaCl, 10 mM Tris/HCl, pH 8.0, 10% (v/v) glycerol, 1 mM EDTA and 100 mM iodoacetamide] containing protease inhibitors (10 μM PMSF, 1 μg/ml leupeptin, 1 μg/ml soya-bean trypsin inhibitor, 1 μg/ml pepstatin A and 10 μg/ml aprotinin) on ice. Receptor complexes were immunoisolated using anti-HA rat monoclonal antibody (Roche Molecular Biochemicals, Indianapolis, IN, U.S.A.), collected with Protein A–agarose beads and analysed by Western blotting using anti-HA mouse monoclonal antibody (HA.11; BabCO, Denver, PA, U.S.A.) or mouse anti-FLAG antibody (M1; Sigma).

[35S]GTPγS binding assay

Cells transiently expressing the HA–α2A-AR, co-expressing HA–α2A-AR and hMOR–Gi1 or co-expressing HA–α2A-AR and Myc–TM1μMOR–Gi1 were harvested and GTP[S] binding was performed as described in [9], but with minor modifications. Membranes containing 100 fmol of [3H]-RX821002 (α2A-AR) or 2 μM [3H]diprenorphine (MOR) binding, assayed as described previously [10,11], were added to an assay buffer {20 mM Hepes (pH 7.4), 3 mM MgCl2, 100 mM NaCl, 1 μM guanosine 5′-diphosphate, 0.2 mM ascorbic acid and 50 nCi of [35S]GTPγS} containing either 100 μM adrenaline (epinephrine) or 1 μM DAMGO ([D-Ala,NMe-Phe,Gly-ol]enkephalin) or both. Non-specific [35S]GTPγS binding was defined as binding detected in the presence of 100 μM GTP[S]. Reactions were incubated at 30°C for 15 min and were terminated by vacuum filtration through Whatman GF/B filters.


FLAG–MORs were transiently transfected into HEK-293 cells stably expressing porcine HA–α2A-AR plated on to glass coverslips. On the day of the study, cells were washed with serum-free medium and incubated, at 4°C, with mouse anti-HA antibody (or mouse M1 anti-FLAG antibody) to identify cell-surface α2A-AR or MOR respectively. Cells were warmed and incubated for the indicated time periods in the absence or presence of the agonist adrenaline or DAMGO for 0–30 min. At the end of the incubation, cells were fixed, permeabilized and incubated with ALEXA-Fluor 568-conjugated secondary antibody. Cell imaging was performed using an LSM510 laser screening confocal microscope. Panels shown (Figure 3) are representative of at least 12 separate images from three separate experiments (except for ‘no drug, 30 min’, which was evaluated in only two experiments).


Studies published during the course of these experiments demonstrated the formation of hetero-oligomers between the α2A-AR and MOR after overexpression in HEK-293 or Madin–Darby canine kidney II cells on the basis of co-immunoprecipitation and proximity-based BRET assays [8]. We have corroborated those findings (Figure 1A) after the expression of epitope-tagged HA–α2A-AR and FLAG–MOR in HEK-293 cells, and we report that the oligomerization was agonist-independent.

Figure 1 Hetero-oligomerization of α2A-AR and MOR

The cDNAs encoding wild-type (A) and mutant (B) HA–α2A-AR and wild-type FLAG–MOR (A, B) were co-expressed in HEK-293 cells, and the proteins were extracted and immunoisolated as described in the Experimental section. IP, immunoprecipitation; αHA, antibody directed against the HA epitope; IB, immunoblot; αFLAG, antibody directed against the FLAG epitope.

Also noticeable in our studies is a substantial increase in the amount of homo-oligomerization of the HA–α2A-AR and, thus, probably also in the amount of α2A-AR–MOR hetero-oligomerization (Figure 1A, cf. the relative concentrations of oligomers and monomers in lanes 1 and 2), when an anti-HA antibody (a bifunctional molecule) was added to the detergent extract, a prerequisite for co-immunoisolation strategies. Consequently, the observations that α2A-AR and MOR co-immunoprecipitate should be interpreted as qualitative in nature and do not permit a quantitative assessment of the fraction of the cellular HA–α2A-AR that is engaged in either homo- or hetero-oligomers, since the addition of antibody for effecting immunoisolation itself enriches the degree of apparent oligomerization detected through SDS/PAGE.

Figure 1(B) reveals that the detection of hetero-oligomerization of α2A-AR and MOR in immunoisolates occurs with wild-type α2A-AR as well as mutated HA–α2A-AR lacking the 3i loop (Δ3i HA–α2A-AR) or C-tail (ΔCtail HA–α2A-AR), indicating that, as for the β2-AR oligomers [12], sequences in or near the bilayer are probably critical for homo- and hetero-oligomerization. Our results also confirm that these interactions can be detected only if HA–α2A-AR and FLAG–MOR are expressed in the same cell. Co-association of FLAG–MOR in HA–α2A-AR immunoisolates does not occur when mixed detergent lysates are processed for immunoisolation and is only detected in immunoisolates derived from detergent extracts of cells co-expressing HA–α2A-AR and FLAG–MOR.

GPCR activation of proximal signal transduction events can be examined by measuring agonist-elicited increases in [35S]GTPγS binding. We exploited a modification of an experimental strategy used by Carrillo et al. [13] and introduced by McLean et al. [9], which measured agonist-stimulated [35S]GTPγS binding as the signal output to assess whether or not transactivation of G-proteins through associated GPCR occurs after hetero-oligomerization. HA–α2A-AR was co-expressed with the myc–hMOR–Gi1 fusion protein, where Gi1 designates a mutant Gi1α subunit that was mutated to eliminate sensitivity to pertussis toxin [9,13]. Under these experimental conditions, agonist activation of [35S]GTPγS binding in pertussis toxin-treated cells would solely reflect activation of the mutant Gi1α subunit. For DAMGO to activate [35S]GTPγS, binding would be straightforward, since the hMOR is introduced to the cells as a fusion protein with Gi1. For α2A-AR to activate [35S]GTPγS binding, however, there is a requirement for ‘cross-talk’ between the α2A-AR and the hMOR–Gi1 fusion protein, as expected for α2A-AR–MOR hetero-oligomers. As shown in Figure 2, agonist activation of α2A-AR resulted in detectable activation of [35S]GTPγS binding only in cells co-expressing the hMOR–Gi1 fusion protein. The finding that adrenaline treatment of cells expressing the HA–α2A-AR alone does not lead to a detectable stimulation of [35S]GTPγS binding proves that pertussis toxin pretreatment was sufficient to eliminate agonist-evoked [35S]GTPγS binding to endogenous G-proteins. As expected, DAMGO activation of [35S]GTPγS binding occurs in cells co-expressing hMOR–Gi1 and HA–α2A-AR. Co-administration of adrenaline and DAMGO caused an approximately additive increase in [35S]GTPγS binding compared with stimulation detected in response to either agonist alone.

Figure 2 Cross-activation of pertussis toxin-insensitive Gi (Gi1) by α2A-AR requires the co-expression of a membrane-targeted G-protein α subunit

HEK-293 cells co-expressing α2A-AR and either a MOR–Gi1 fusion protein or TM1–Gi1, a fusion of Gi1 with the first transmembrane span of MOR, were harvested and evaluated for [35S]GTPγS binding as described in the Experimental section. Epi, adrenaline (agonist at α2A-AR); DAMGO, agonist at MOR.

One interpretation of the finding that agonist activation of HA–α2A-AR causes enhanced [35S]GTPγS binding to a G-protein fused with the co-expressed MOR (hMOR–Gi1) is that transactivation of Gi1 occurs through hMOR, secondary to α2A-AR-induced conformational changes propagated through MOR after agonist binding to the α2A-AR. However, as emphasized by Molinari et al. [14], the fusion of G-proteins to GPCR necessarily enriches their concentration in the proximity of the receptor and any GPCR in association with that receptor. This enrichment in local G-protein concentration, rather than transactivation, may be responsible for the enhanced receptor-mediated G-protein activation in cells co-expressing GPCR and GPCR–G-protein fusions. To test this latter hypothesis, we prepared an epitope-tagged (myc) fusion protein of TM1 of the MOR fused with Gi1, analogous to the approach employed by Molinari et al. [14].

Immunofluorescence studies revealed that the myc–TM1–Gi1 protein achieved cell-surface expression in a manner indistinguishable from the expression of FLAG–MOR alone (results not shown). Our functional studies revealed that co-expression of HA–α2A-AR and myc–TM1–Gi1 resulted in an activation of [35S]GTPγS binding by adrenaline indistinguishable from activation in cells co-expressing HA–α2A-AR and the hMOR–Gi1 fusion protein. Since no full-length MOR was expressed in these cells, DAMGO was without effect, and additivity of [35S]GTPγS binding activation was not observed in response to adrenaline+DAMGO, as expected. These findings argue against transactivation as the explanation for acquisition of adrenaline-stimulated [35S]-GTPγS binding in pertussis toxin-treated cells co-expressing HA–α2A-AR and hMOR–Gi1; instead, they suggest that the ability of α2A-AR to activate Gi1α in pertussis toxin-treated cells expressing the hMOR–Gi1α fusion protein results from enrichment of Gi in the local environment shared by the co-expressed α2A-AR and hMOR–Gi1, probably organizing as hetero-oligomers, and mimicked by myc–TM1–Gi1.

Preliminary studies evaluating receptor-activated MAPK (mitogen-activated protein kinase) as a measure of α2-AR–MOR cross-talk due to hetero-oligomerization did not reveal a change in the time course, extent or concentration–response relationships for the α2-agonist (adrenaline or UK14304) or MOR agonist (DAMGO) activation of p42/44 MAPK phosphorylation (results not shown), in contrast with the findings of Jordan et al. [8]. Subtle methodological differences or differences in the extent of oligomerization in our studies may be responsible for our findings.

One final assessment of the functional consequences of α2A-AR–MOR hetero-oligomerization was their inter-dependent effects on agonist-evoked receptor redistribution. Agonist activation of HA–α2A-AR in HEK-293 cells leads to limited redistribution (endocytosis) over 60 min [15]. In contrast, DAMGO activation of MOR is followed by rapid receptor redistribution and marked internalization [16]. As shown in Figure 3, we observed that the α2 agonist, adrenaline, could cause the redistribution of HA–α2A-AR in cells co-expressing HA–α2A-AR and FLAG–MOR. DAMGO could elicit marked internalization of FLAG–MOR. However, adrenaline did not alter MOR distribution nor did DAMGO alter the distribution of the α2A-AR. These findings are evidence for independent internalization of these receptors despite their co-expression and cross-talk in the activation of [35S]GTPγS binding in HEK-293 cells.

Figure 3 Independent agonist-evoked endocytosis of HA–α2A-AR and FLAG–MOR co-expressed in HEK-293 cells

FLAG–MOR was co-expressed transiently in HEK-293 cells permanently expressing HA–α2A-AR and agonist-evoked redistribution was evaluated over a period of 0–30 min, as described in the Experimental section. Data are representative of at least 12 images under each condition obtained from two (no drug, 30 min) or three (all other conditions) separate experiments. Epi, adrenaline (epinephrine).


The hetero- and homo-oligomerization of GPCR has been studied extensively, principally in heterologous systems [13,8,9,12,13,17], but also in native cells and tissues [1]. A variety of experimental strategies confirm the proximity of these receptors to target membranes, consistent with the interpretation that oligomers are forming.

A number of studies also suggest that hetero-oligomers can manifest properties distinct from either of the protomer GPCRs in the complex [1,18].

In the present study, we were intrigued to explore the hetero-oligomerization of α2A-AR with MOR, since there is a profound synergism of these receptors and their pathways in vivo, especially in anti-nociception [7]. The recent findings of Jordan et al. [8] that juxtaposition of these receptors in surface microcompartments occurs was confirmed by our studies, on the basis of both co-immunoisolation (Figure 1) and cross-talk between α2A-AR and the MOR–Gi1α fusion protein, permitting α2-AR-activated [35S]GTPγS binding to the mutant Gi1α subunit in pertussis toxin-treated cells (Figure 2). However, our results also indicate that α2A-AR activation of [35S]GTPγS binding under these circumstances does not require transactivation through MOR, since quantitatively indistinguishable findings are observed in cells co-expressing α2A-AR and a fusion protein of Gi1 and the first transmembrane span of MOR, termed myc–TM1–Gi1. Similarly, agonist-induced endocytosis of HA–α2A-AR and FLAG–MOR occur independent of one another in cells co-expressing both the receptors. Hence, either α2A-AR and MOR ‘juxtaposition’ in a shared membrane microenvironment occurs and true hetero-oligomerization does not or the hetero-oligomers can behave functionally and morphologically independent of one another.

Taken together, our results are consistent with the conclusion that there is physical interaction between α2A-ARs and MORs, at least after overexpression in heterologous cells. However, our results are not consistent with the assertion that there are functional consequences of these hetero-oligomers distinct from each receptor functioning alone, at least for the regulation of GTP binding to cognate G-proteins or for agonist-accelerated receptor internalization. As such, hetero-oligomerization by itself is unlikely to be the molecular underpinning of adrenergic–opioid synergism in vivo.


We are grateful to Dr Q. Wang and Dr A. Brady for considerable experimental advice during the course of these experiments and to Dr G. Milligan and Dr L. Devi for providing cDNAs encoding FLAG–MOR and hMOR–Gi1 respectively. L.E.L. was supported by NIH grant no. HL43671.


  • Signalling Outwards and Inwards: A Focus Topic at BioScience2004, held at SECC Glasgow, U.K., 18–22 July 2004. Edited by J. Challiss (Leicester, U.K.), A. Harwood (University College London, U.K.), M. Humphries (Manchester, U.K.), C. Isacke (Institute of Cancer Research, London, U.K.), R. Liddington (Burnham Institute, La Jolla, CA, U.S.A.), T. Palmer (Glasgow, U.K.), K. Siddle (Cambridge, U.K.), C. Sutherland (Dundee, U.K.), H. Wallace (Aberdeen, U.K.) and M. Welham (Bath, U.K.).

Abbreviations: α2A-AR, α2A-adrenergic receptor; DAMGO, [D-Ala,NMe-Phe,Gly-ol]enkephalin; GPCR, G-protein-coupled receptor; HA, haemagglutinin; HEK-293 cells, human embryonic kidney 293 cells; MAPK, mitogen-activated protein kinase; MOR, μ-opioid receptor; hMOR, human MOR; TM1, first transmembrane domain


View Abstract