| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J. Biol. Chem., Vol. 281, Issue 9, 5416-5425, March 3, 2006
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
1
2
From the
Centro de Estudios Farmacológicos y Botánicos, Consejo Nacional de Investigaciones Científicas y Técnicas, Buenos Aires 1414, Argentina and the Department of Pharmacology, University of Washington School of Medicine, Seattle, Washington 98195
Received for publication, July 11, 2005 , and in revised form, November 22, 2005.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
Additional pharmacological evidence that mAChR can form dimers was provided by Maggio et al. (8), who used chimeric 
2-adrenergic/M3 muscarinic receptors composed of the first five transmembrane domains (TMD) of one receptor and the last two TMD of the other. No binding or signaling was detected when either fusion protein was expressed alone, but coexpression of both chimeras rescued binding activity and signaling properties of both adrenergic and muscarinic ligands. These results were interpreted as intermolecular interactions between two inactive receptors now engaged in a dimeric complex with restored receptor functional properties. However, such trans-complementation results have been challenged because of the fact that those intermolecular interactions were seen by studying mutant receptors and not wild type receptors (1). Coimmunoprecipitation of solubilized epitope-tagged M3 mAChR and Western blot analysis has also been employed to demonstrate the existence of mAChR oligomers in a direct fashion. This biochemical approach revealed that the M3 mAChR is capable of forming disulfide-linked as well as noncovalent dimers/multimers (9). Although coimmunoprecipitation/immunoblotting assays provide fairly convincing evidence for mAChR dimer formation, such studies have been criticized for the possibility that the observed receptor aggregation could result from an artifact because of the presence of concentrated preparations of highly hydrophobic molecules following the solubilization process (1). Over the last decade, several groups have used biophysical assays based on light resonance energy transfer to assess GPCR oligomerization (1, 2, 10, 11). Fluorescence resonance energy transfer and its variation, bioluminescence resonance energy transfer (BRET), have been used extensively to overcome technological limitations present in pharmacological and biochemical methods because they have the distinct advantage of monitoring real time interactions between proteins synthesized in their correct location in living cells (10). In the classical BRET methodology, the first protein is fused to Renilla luciferase (RLuc), and the second protein partner is fused to a fluorescent protein (yellow fluorescent protein (YFP)). If both partners interact, resonance energy transfer between RLuc (donor) and the YFP (acceptor) can be measured upon addition of the RLuc substrate coelenterazine (12).
In this study, the occurrence of constitutive homo- and hetero-oligomers among M1, M2, and M3 mAChR was investigated in living cells expressing physiological levels of these receptors by BRET. BRET saturation experiments were conducted to estimate the relative ability of each receptor subtype to engage in homo- and heterotropic interactions. The nature of such homo-oligomers, the proportion of receptor molecules forming those oligomeric complexes, and the effect of muscarinic agonist on M1, M2, and M3 mAChR homo- and heterodimerization were also addressed. Finally, we examined the implications of oligomerization on mAChR function by determining the effect of M2/M3 heterodimerization on agonist-induced M3 mAChR regulation.
|
EXPERIMENTAL PROCEDURES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
Receptor Constructs—mAChR-RLuc and mAChR-YFP fusion proteins were constructed by ligating the Renilla luciferase (RLuc) and the YFP moieties to the C-terminal end of the receptors. Mouse M1, porcine M2, and human M3 mAChR coding sequences without their stop codons were amplified by using sense and antisense primers harboring unique restriction sites. The fragments were then subcloned in-frame into pRL-CMV-RLuc (Promega) or pEYFP-N1 (Clontech, Palo Alto, CA). Restriction sites used were as follows: XhoI/BamHI (M1-RLuc, M1-YFP), XhoI/SacII (M2-RLuc, M2-YFP), EcoRV/BamHI (M3-RLuc), and EcoRI/BamHI (M3-YFP). pEYFP was cut with AgeI and BsrGI, and the YFP fragment was inserted into the AgeI/BsrGI site from human Smoothened-GFP (Smo-GFP) to give Smo-YFP. Smo-GFP was provided by Dr. A. Kaykas (University of Washington). The HA-M3 mAChR expression plasmid was obtained from University of Missouri-Rolla cDNA Resource Center (Rolla, MO). All other constructs have been described elsewhere (14, 15).
Cell Culture and Transfection—HEK 293 or JEG-3 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal bovine serum and penicillin (100 units/ml)/streptomycin sulfate (0.1 mg/ml) at 37 °C in a humidified 10% CO2 environment. Transient transfections were performed on cells that were at 70–80% confluence using the calcium phosphate precipitation protocol (16).
Pharmacological Characterization of Constructs—HEK 293 cells grown in 15-cm dishes were transfected with RLuc-tagged, YFP-tagged, or nontagged mAChR (12.5-25 µg of DNA/plate). Following a 36-h period post-transfection, cells were washed with ice-cold PBS, and membranes were obtained as described previously (17). Saturation binding studies were performed by incubating membranes with increasing concentrations of [3H]QNB at 37 °C for 90 min in 50 mM sodium/potassium phosphate, pH 7.4. Atropine (1 µM) was used to define nonspecific binding. The reaction was stopped by rapid filtration over GF/C glass-fiber filters (Whatman). Filters were then washed four times with cold phosphate buffer, and the retained radioactivity was determined by scintillation counting. Curves were analyzed by nonlinear regression analysis using Prism 4 software (GraphPad, San Diego).
BRET Assays—Thirty six hours after transfection, HEK 293 cells were distributed in 96-well microplates (white opaque Isoplate, Wallac) washed with phosphate-buffered saline (PBS) and resuspended in PBS/glucose 0.1% in the presence or absence of 10 µM carbachol at 25 °C. Coelenterazine h was added at a final concentration of 5 µM, and light intensity was sequentially integrated in the 510–590 and 440–500 nm windows using a Victor2 multiplate reader (PerkinElmer Life Sciences). BRET ratio was defined as ((emission at 510–590)/(emission at 440–500)) – Cf, where Cf corresponded to (emission at 510–590)/(emission at 440–500) for the receptor RLuc constructs expressed alone in the same experiments.
Fluorescence and Luminescence Measurements—Transiently transfected HEK 293 cells were harvested, washed, and resuspended as described above and then distributed in 96-well microplates (either clear-bottomed or white opaque Isoplates for fluorescence and luminescence determinations, respectively). Both determinations were also carried out in a Victor2 multiplate reader. The total fluorescence of cells was measured using an excitation filter of 485 nm and an emission filter of 535 nm. For assessing the activity of Renilla luciferase, total luminescence was determined on samples incubated with 5 µM coelenterazine. For both measurements, the means of duplicate wells were calculated. Total fluorescence values were then divided by the background determined in wells containing untransfected cells. The background values for total luminescence were negligible, and they were subtracted from sample values.
Titration of Donor and Acceptor Fusion Proteins—Expression levels of mAChR-RLuc and mAChR-YFP constructs were monitored by measuring luminescence and fluorescence, respectively, as described above. This procedure is based on the observation that luminescence and fluorescence levels of several receptor-RLuc and receptor-YFP (or receptor-GFP) fusion proteins have been found to be linearly correlated with receptor numbers (18). Because this correlation is an intrinsic property of each fusion protein, we tested whether our mAChR fusion proteins followed a linear correlation pattern. Thus, we expressed each mAChR-RLuc or mAChR-YFP at different levels in HEK 293 cells and assessed the relationship between luciferase activity or fluorescence, respectively, and the amount of mAChR-binding sites in the same cells. Thirty six hours after transfection, 
20,000 HEK 293 cells were incubated with 1 nM [3H]QNB in PBS, 0.1% glucose for 90 min at 25 °C. Nonspecific binding was determined in the presence of 1 µM atropine. The reaction was stopped by rapid filtration over GF/C glass-fiber filters, which were then washed four times with cold PBS. Retained radioactivity was determined as described above. Luminescence and fluorescence (arbitrary units) were plotted against total binding sites, and linear regression curves were obtained (supplemental Fig. 2). These standard curves generated for each single experiment were used to transform fluorescence and luminescence values into femtomoles of receptor. Thus, the fluorescence/luminescence ratios were transformed into (receptor-YFP)/(receptor-RLuc) ratios, which allowed us to determine accurate BRETmax and BRET50 values (19). Protein concentration of samples was determined to control for the number of cells and also to express receptor numbers in femtomole/mg of total cell protein, using the standard method described by Lowry et al. (20).
Receptor Down-regulation Assays—Determination of agonist-induced M2 or M3 mAChR down-regulation was accomplished using the binding of the membrane-permeable muscarinic ligand [3H]QNB to intact cells expressing either mAChR subtype, as described previously (21). Briefly, JEG-3 cells were transiently transfected with either M2 or FLAG-M3 mAChR DNA constructs. Forty eight hours later cells were incubated either with 1 mM carbachol for different times (within a 0–12-h-range) or in the presence of various concentrations of carbachol for 12 h at 37 °C. Cells were then washed with ice-cold PBS and labeled with a saturating concentration of [3H]QNB (1 nM) in PBS for 90 min at 37 °C. Nonspecific binding was determined in the presence of 1 µM atropine. Labeled cells were washed with ice-cold PBS and filtered onto GF/C membrane filters (Whatman). Filters were then washed with ice-cold PBS, transferred to scintillation vials, and combined with scintillation fluid before the determination of radioactivity by scintillation counting.
Down-regulation of Coexpressed M2 and M3 mAChR—JEG-3 cells were transfected with M2 mAChR, FLAG-M3 mAChR, or both DNA constructs at different proportions following the procedure indicated above. Forty eight hours post-transfection, cells were treated with 10 µM carbachol for 12 h at 37 °C. Cells were then washed with ice-cold PBS, harvested in 50 mM sodium/potassium phosphate buffer, pH 7.4, supplemented with protease inhibitors (10 µg/ml leupeptin, 70 µg/ml phenylmethanesulfonyl fluoride, and 0.25 µg/ml pepstatin A), and were glass-glass homogenized. Membranes were obtained as described previously (17), and the total number of mAChR in crude membrane homogenates was determined using the binding of [3H]QNB as described by Halvorsen and Nathanson (22). The numbers of M2 and FLAG-M3 mAChR were measured by the immunoprecipitation assay described previously (23) using anti-M2 and anti-FLAG monoclonal antibodies, respectively.
Alternatively, control experiments were performed by cotransfecting cells with FLAG-M3 mAChR and HA-M3 mAChR at different relative expression levels, as determined by immunoprecipitation using anti-FLAG and anti-HA antibodies (23). The extent of FLAG-M3 down-regulation after carbachol treatment was then determined as indicated above.
Data Analysis—BRET saturation curves were analyzed using Prism 4 software. Isotherms were fitted using a nonlinear regression equation assuming a single binding site, which provided BRETmax and BRET50 values. The correlation between fluorescence or luminescence and receptor density was analyzed by linear regression curve fitting with the same software. Statistical significance of differences in M2 versus M3 mAChR regulation was assessed by the two-tailed unpaired Student's t test.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
|
To determine whether M1, M2, and M3 mAChR can form homodimers in intact cells, we assessed the ability of coexpressed RLuc-tagged mAChR and YFP-tagged mAChR to interact with one another by BRET saturation curve analysis. HEK 293 cells were transiently cotransfected with a constant amount of the RLuc construct and increasing amounts of the YFP construct, and the transfer of energy between both fusion proteins was measured upon addition of the Renilla luciferase substrate coelenterazine h. The expression levels of each RLuc-tagged receptor were assessed by detection of total luminescence from the previous luciferase reaction, and the expression levels of the YFP-tagged receptors were monitored by determination of total fluorescence. Because the expression levels of RLuc constructs were found to be modified by the cotransfection of increasing amounts of their YFP partners, BRET signals were plotted as a function of the fluorescence/luminescence ratio. Coexpression of M1, M2, or M3 donor molecules with their homologous acceptor partners resulted in classical saturation isotherms (Fig. 1, A–C), in which BRET signals increased as a hyperbolic function of the concentration of the acceptor, reaching an asymptote that corresponds to the saturation of all BRET donors by acceptor molecules. This observation suggests that M1, M2, and M3 mAChR form constitutive homodimers in live cells. To ensure that these BRET signals did not result from a spurious interaction between RLuc and YFP, we performed control assays in which each RLuc-tagged mAChR was coexpressed with increasing amounts of soluble YFP. These saturation assays resulted in quasi-linear curves showing marginal BRET signals, which indicate a nonspecific interaction between both coexpressed partners in each case. The selectivity of the interaction between donor and acceptor mAChR was further assessed to rule out the possibility of nonspecific BRET signals because of transmembrane protein overexpression. The distantly related seven transmembrane signaling receptor Smoothened fused to YFP (Smo-YFP) was cotransfected with M1, M2, or M3 mAChR-RLuc, which resulted in a dramatically decreased BRET ratio for the three mAChR tested. In contrast, coexpression of Smo-RLuc with increasing amounts of Smo-YFP yielded significantly higher BRET levels (supplemental Fig. 1), indicating that Smo-RLuc is effectively capable of forming constitutive dimers in intact cells when coexpressed with a specific partner. BRET data were not influenced by the relative position of RLuc or YFP on the receptor pairs because identical results were obtained for the reciprocal orientation.4 These data confirm our previous observation, indicating that M1, M2, and M3 donor and acceptor fusion proteins are engaged in specific interactions as constitutive homodimers.
Although the saturation curves shown previously suggest that M1, M2, and M3 donor and acceptor specifically interact with each other, they do not provide enough information about the binding parameters required for proper quantitative analysis of receptor-receptor interactions. Therefore, we conducted a new set of saturation experiments in which the amount of each receptor effectively expressed in transfected cells was monitored, for each individual experiment, by correlating both total luminescence and total fluorescence with the number of [3H]QNB-binding sites. The analysis of such correlation shows a linear relationship between total luminescence and receptor density with no significant differences among the slopes obtained for the three mAChR receptor subtypes (supplemental Fig. 2). Correlation analysis between total fluorescence and total receptor numbers also yielded linear curves. However, the slope obtained for M1-YFP was significantly higher than that from either M2-YFP or M3-YFP. The linear regression equations shown in supplemental Fig. 2 were used to transform fluorescence and luminescence units to receptor numbers, and BRET signals were plotted as a function of (receptor-YFP number)/(receptor-RLuc number) ratio. These new BRET saturation curves also behaved as hyperbolic functions reaching a saturation level (Fig. 2). According to classical analysis of BRET-monitored protein-protein interaction assuming one binding site, the concentration of the acceptor giving 50% of the energy transfer (BRET50) accounts for the relative affinity between both interacting protomers, and the maximal BRET signal depends on the total number of dimers formed as well as on the distance between the donor and the acceptor within the dimer (19). When comparing the curves obtained for M1, M2, and M3 mAChR homodimers (Table 2), no significant differences in BRET50 values were found, indicating that the affinities of protomers for one another are similar among the three mAChR subtypes studied. On the other hand the BRETmax value for the M1 donor-acceptor pair was found to be significantly different from that for M2 and M3 homodimers. Indeed, the BRETmax for the M1-RLuc/M1-YFP pair was over twice as high as the other BRETmax values. This could indicate that either a higher percentage of the M1 subtype is engaged in homodimerization or that energy transfer between RLuc and YFP within the M1 homodimer is more efficient than that for M2 and M3 homodimers. Because the relative affinities between each of the partners were found to be very similar among the three receptor subtypes (Table 2), the second hypothesis is more likely. Moreover, the fact that correlation curves between total fluorescence and receptor density yielded a higher slope value for M1 suggests that YFP functions as a more efficient acceptor when it is fused to the M1 receptor subtype, which may result in higher BRET values for this pair.
|
|
|
|
Detailed quantitative analysis of 
2-adrenergic receptor homodimerization by Mercier et al. (19) suggested that it is possible to estimate the proportion of receptors engaged in dimerization. Assuming a free equilibrium between donor and acceptor fusion proteins, only 50% of the dimers are expected to generate BRET if both partners are expressed at equimolar concentrations, because each partner can also generate 25% homodimers between identical molecules. Thus, if all receptors form dimers, the maximal BRET ratio detected at equimolar expression levels should be BRETmax/2. We determined experimental BRET values corresponding to equimolar expression of donor/acceptor at different expression levels of total receptors (Table 2, legend). Predicted percentage values of M1, M2, and M3 mAChR existing as dimers are shown in Table 2 and suggest that a population of a single subtype of M1, M2, or M3 mAChR is predominantly composed of constitutive homodimers.
|
|
Evidence that agonists can induce changes in BRET signal has been reported for several GPCR. Eventually, such changes in energy transfer either reflect variations in dimer formation or just conformational changes within pre-existing dimers (2). In contrast, other studies have failed to show BRET changes upon agonist treatment (3). Therefore, we investigated the effect of a muscarinic agonist on BRET by performing BRET saturation curves on HEK 293 cells cotransfected with all possible combinations of M1, M2, and M3 donor and acceptor fusion proteins. Carbachol (10 µM) treatment during 5, 15, or 30 min of transfected cells revealed no significant changes in BRET signal for any of the receptor pairs tested (Tables 4 and 5 and see supplemental Figs. 3 and 4). These results suggest that agonist treatment does not modify homo- or heterodimer formation or even promote permanent conformational changes that could modify BRET within pre-existing dimers.
|
|
(h) M2, 2.4 ± 0.1; M3, 2.5 ± 0.2) (Fig. 5A, inset). Carbachol-induced differential down-regulation of M2 and M3 mAChR was confirmed by concentration-response curves, which again showed significantly different maximal effects (Emax (%) M2, 65 ± 4; M3, 24 ± 3, p < 0.02) but similar EC50 values for both receptor subtypes (EC50 (µM) M2, 2.2 ± 0.1; M3, 1.8 ± 0.2) (Fig. 5B). Given that both M2 and M3 mAChR are coexpressed in many tissues, including the gastrointestinal tract, uterus, urinary bladder, ciliary body, and airways (28) and that they were also found to form heterodimers (Fig. 3), we asked whether M2/M3 heterodimerization could modify the differential pattern of M2/M3 receptor regulation in JEG-3 cells coexpressing both receptor subtypes. Thus, we treated cells expressing M2 mAChR alone, FLAG-tagged M3 mAChR alone, or both receptors at different proportions with 10 µM carbachol for 12 h and determined the extent of subtype-specific receptor down-regulation by immunoprecipitation of solubilized [3H]QNB-labeled M2 or FLAG-M3 mAChR using anti-M2 or anti-FLAG monoclonal antibodies, respectively. Agonist stimulation of cells expressing M2 mAChR alone resulted in a 43.4 ± 5.4% decrease in cellular receptor numbers, whereas treatment of cells expressing only FLAG-M3 mAChR led to a 15.0 ± 4.6% loss of [3H]QNB-binding sites, as assessed by immunoprecipitation, which was consistent with the binding assay results described above (Fig. 6). Most strikingly, coexpression of JEG-3 cells with FLAG-M3 mAChR and increasing proportions of M2 mAChR led to a concentration-dependent increase in FLAG-M3 mAChR down-regulation, with a maximal 43.5 ± 3.3% decrease in receptor number, as compared with control untreated cells. In contrast, the extent of M2 mAChR down-regulation remained unchanged within the same range of receptor ratio values.4
To rule out the possibility that the coexpression of the M2 receptor increases the FLAG-M3 receptor down-regulation because of an unanticipated effect of the FLAG tag, we performed control experiments by cotransfecting the FLAG-tagged M3 mAChR in the presence of increasing amounts of a non-FLAG-tagged M3 mAChR (HA-M3). In this case, the extent of FLAG M3 down-regulation was not significantly altered at any HA-M3/FLAG-M3 expression ratio tested (Fig. 6).
Taken together, these data indicate that coexpression of M2 and M3 mAChR facilitates M3 receptor regulation by forming M2/M3 heterodimers that are more susceptible of being regulated by agonist exposure than M3 alone.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
Resonance energy transfer approaches (fluorescence resonance energy transfer and BRET) have become systems of choice to monitor protein-protein interactions in living cells (2). Because BRET between RLuc and YFP can occur only if both molecules are within 100Å of each other, the occurrence of BRET between RLuc- and YFP-tagged fusion proteins indicates a molecular proximity that is consistent with dimerization (12). In this study we use the term dimer as the simplest form of oligomerization, and we eventually discuss the possibility that BRET could also result from the existence of larger complexes.
Our results from BRET saturation analysis show that M1, M2, and M3 can form homo-oligomers in intact HEK 293 cells. Indeed, the interaction between mAChR-RLuc and mAChR-YFP partners within each oligomer studied has proved to be saturable and specific. The specificity of mAChR oligomerization was demonstrated by BRET saturation curves showing the inability of M1, M2, and M3 receptors to heterodimerize with an unrelated seven transmembrane domain protein, the signaling receptor Smoothened (Smo). The potential inability of Smo-YFP to engage in oligomerization was ruled out by the demonstration that this fusion protein interacts with Smo-RLuc to yield a saturable BRET signal consistent with Smo homodimerization (supplemental Fig. 1), as was suggested previously (30). Further evidence supporting the specificity of mAChR oligomerization is derived from a -galactosidase complementation approach by Kaykas et al. (15) who have recently demonstrated that the unrelated GPCR Frizzled 1 and Frizzled 2 can homodimerize and even heterodimerize with each other, but not with M1 or M2 mAChR. More recently, it has been reported that µ, 
, or 
opioid receptors do not form detectable hetero-oligomers with M2 mAChR by using the BRET2 technology (31).
|
|
2-adrenergic, opioid, melatonin, and adenosine A2A receptors exist as dimeric complexes in the plasma membrane (19, 25, 31, 32). In contrast, our data present a different view from previous biochemical reports suggesting the existence of M1, M2, and M3 mAChR higher order homo-oligomers (9, 33). In particular, the recent work by Park and Wells (34) has suggested the existence of trimeric or even larger aggregates in extracts from solubilized M2 mAChR expressed in Sf9 cells. Because of the highly hydrophobic nature of mAChR TMD and the extreme levels of protein expression reached in Sf9 cells, it could be argued that detection of oligomers by this methodology results from artifactual aggregation following cell lysis and solubilization (10). The results shown in this report using the BRET methodology suggest that M1, M2, and M3 mAChR expressed in live cells are populations composed of a predominant fraction of homodimers and a minor fraction of monomers. However, we cannot rule out the existence of higher order complexes that cannot be detected by BRET because of distance constraints. Although most previous reports in favor of the existence of mAChR oligomers have agreed that these GPCR can form homodimers, the existence of constitutive mAChR heterodimers remains controversial. By using trans-complementation experiments, Maggio et al. (35) provided pharmacological evidence for the formation of M2/M3 heterodimers. Upon coexpression of wild type M2 or M3 mAChR with gene fragments originating from M3 or M2 mAChR, respectively, these authors determined the presence of two populations of binding sites as follows: one representing the wild type M2 or M3 receptors, and the other identified as a heterodimeric M2/M3 receptor. Recently, Novi et al. (36) reported that M2 and M3 receptors can interact physically by showing that HA-tagged M3 mAChR could be immunoprecipitated with anti-Myc antibodies, and Myc-M2 mAChR were immunoprecipitated with anti-HA antibodies when both receptors were coexpressed in COS-7 cells. These data have been challenged by other reports, which failed to provide evidence for the existence of mAChR heterodimers. For instance, Griffin et al. (37) compared the pharmacological antagonism of M2 and M3 mAChR expressed in Chinese hamster ovary cells both in isolation or in combination, and their binding and signaling data were consistent with those expected for independent receptors and not heteromeric complexes or the exhibition of a novel pharmacology. Moreover, biochemical evidence against M3/M1 and M3/M2 mAChR heterodimerization was provided by Zeng and Wess (9), who reported no detectable coimmunoprecipitation of an HA epitope-tagged M3 mAChR with M1 or M2 mAChR in lysates from COS-7 cells cotransfected with either receptor pair. It is tempting to speculate that the biochemical approach that failed to show evidence for the existence of M3/M1 and M3/M2 heterodimers (9) was carried out using mutant M3 receptors lacking most of the i3 loop, which might hamper the interaction with wild type M1 or M2 mAChR. In fact, trans-complementation assays by Maggio et al. (35) have shown that removal of a large fragment of the i3 loop from M3 can prevent the interaction between this mutant receptor and a truncated form of M2, and that such interaction can be rescued by using the wild type M3 receptor instead of the short M3 form. These data also appear to be consistent with the hypothesis proposed by Gouldson et al. (38), who suggested that GPCR dimers are generated by swapping TMD and proposes the integrity of the i3 loop as a requirement for receptor dimerization. However, biochemical data showing that the shortening of the i3 from M3 did not prevent M3 homodimerization (9) suggest that the presence of the full-length i3 loop is not required for the formation of M3 receptor homodimers.
In this study we attempted to resolve the discrepancy between pharmacological and biochemical data by using quantitative BRET analysis of receptor-receptor interaction to determine whether mAChR can form heterodimers in living cells. In our BRET assays the wild type sequence of receptors is conserved. In addition, receptor constructs are expressed within a preserved cellular environment and at their proper location at the cell surface. Our results from BRET saturation curves clearly indicate that M1, M2, and M3 can form heterodimers. BRET50 values from M1, M2, and M3 mAChR heterodimerization curves (Table 3), although significantly higher than those determined for mAChR homodimers (Table 2), are also indicative of high affinity heterodimers. In fact, similar high affinity values have been reported for other GPCR homo- or hetero-oligomers (18, 19, 39, 40). Therefore, our data suggest that the coexpression of two mAChR subtypes in a given tissue results in a heterogeneous combination of homo- and heterodimers, whose proportions depend on both the relative affinity between interacting species and their expression levels, and a minor complement of monomers.
Agonist binding to some GPCR induces or stabilizes the dimerized state, which is the form of the receptor that productively interacts with G proteins. Other GPCR exist as constitutive dimers, which are already formed in the endoplasmic reticulum, and some of them even undergo monomerization upon agonist stimulation (1–3). There are, however, divergent results for a number of GPCR depending on the method used to assess dimerization (3). By using SDS-PAGE and Western blot analysis of membrane extracts from COS-7 cells expressing a mutant M3 mAChR, Zeng and Wess (9) showed that agonist-induced receptor stimulation had no significant effect on M3 mAChR dimerization. Our results were consistent with such findings, because carbachol treatment of HEK 293 cells expressing homologous or heterologous combinations of M1, M2, and M3 mAChR fusion proteins revealed no significant changes in the BRET signal, as compared with control untreated cotransfected cells. BRET saturation curve analysis has shown that M1, M2, or M3 mAChR may exist as homogeneous populations predominantly composed of high affinity homodimers. Alternatively, a heterogeneous population of two mAChR subtypes exists as a combination of constitutive homo- and heterodimers, with a minimal proportion of monomers. Therefore, we can conclude that the absence of a significant change in the BRET signal following agonist stimulation indicates that muscarinic agonists do not induce monomerization/disaggregation of oligomers or even conformational changes that may modify the stability of the oligomeric complexes. However, we cannot rule out the possibility that agonist stimulation induces assembly/disassembly events without affecting the steady state proportion of mAChR receptors engaged in homo-/heterodimers.
The functional consequences of mAChR homo- and heterodimerization have not been extensively addressed. Novi et al. (41) have recently demonstrated that the paired activation of the two components of M3 mAChR homodimer is required for induction of extracellular signal-regulated kinase 1/2 phosphorylation, whereas single activation of part of a heterodimer is sufficient for G protein activation. In addition, these authors reported that the recruitment of -arrestin-1 by homodimeric M3/M3 and heterodimeric M2/M3 mAChR requires the paired activation of the single receptor components within the dimer (36, 41).
Although these results contribute to the understanding of the role of oligomerization on -arrestin-mediated signaling, the functional implications of mAChR heterodimerization on receptor regulation have not yet been investigated. The M2/M3 heterodimer pair is particularly interesting to analyze for several reasons. (a) Coexpressed M2 and M3 mAChR subtypes mediate autonomic regulation of smooth muscle contraction in many tissues. (b) Some studies have argued against the existence of mAChR heterodimers. In particular, the presence of M2/M3 constitutive heterodimers has been challenged by recent reports and is still a matter of controversy (9, 35–37). (c) M2 and M3 mAChR regulation has typically been examined using cell systems expressing a single mAChR subtype. This model differs from the physiological situation, in which different mAChR subtypes coexist, even together with less or nonrelated receptors.
Much has been learned about short term desensitization (G protein uncoupling, internalization/sequestration), which regulates GPCR within seconds to minutes after agonist-induced activation. Considerably less is known about biochemical mechanisms that regulate GPCR activity upon sustained agonist exposure. Long term regulation of GPCR, or "down-regulation," is thought to be important to physiological adaptation induced by endogenous ligands as well as to the clinical effects of exogenously administered drugs that are used in a chronic or repeated manner. Moreover, mutational analyses of mAChR and other GPCR have demonstrated that short term and long term receptor regulation pathways can be differentiated. In particular, we have shown previously that the long term down-regulation and the rapid sequestration of M1 and M2 mAChR are independent processes involving distinct molecular mechanisms (42, 43). Tsuga et al. (44) have also investigated differential regulatory mechanisms of M2 mAChR upon agonist stimulation and demonstrated that internalization depends on the integrity of i3 (which becomes phosphorylated by GRK2 (G protein-coupled receptor kinase 2)), although down-regulation may occur through a GRK2 facilitating pathway and i3 independent pathway. The differential "short" versus"long" term regulation may be critical when choosing a cell system to assess mAChR regulation. In fact, M1 and M2 mAChR expressed in HEK 293 cells undergo agonist-induced internalization but not down-regulation (45, 46).
JEG-3 choriocarcinoma cells have proved to be a validated model to study the differential regulation of muscarinic receptor subtypes (14, 21, 43, 47). When M1 or M2 mAChR are expressed in these cells, sustained agonist stimulation leads to consistently higher levels of receptor down-regulation for M2 than for M1. Schlador et al. (21) identified five amino acids ("VTILA") located in TMD6, TMD7, and i3 that are essential for M2 mAChR regulation via a dynamin-independent mechanism. These residues are not present in M1 mAChR, but substitution of this motif into the M1 mAChR is sufficient for conversion to the down-regulation-competent M2 phenotype.
Because the M3 mAChR subtype also lacks the VTILA motif, we suspected that this receptor would show an impaired receptor regulation profile compared with M2. Consistent with our predictions, M2 underwent agonist-induced down-regulation to a much higher extent than M3. Because we demonstrated previously that most of the M2 and M3 mAChR exist as constitutive homodimers (Figs. 1 and 2) and that agonist exposure does not induce changes in M2 and M3 homodimerization (Table 4), these results indicate that a significant fraction of the M2 homodimer undergoes down-regulation, whereas most, if not all, of the M3 homodimer remains at the cell surface.
Our previous results also suggest that a heterogeneous population of both M2 and M3 mAChR is composed of M2 and M3 homodimers together with a significant fraction of M2/M3 heterodimers, whose proportion depends on the expression levels of both receptor subtypes and the relative affinity values for both homodimers and the M2/M3 heterodimer. If such a heterogeneous population of receptors undergoes sustained agonist exposure, we would predict that M2 and M3 homodimers undergo differential down-regulation. To determine the fate of M2/M3 heterodimers, we first transfected JEG-3 cells with various proportions of M2/FLAG-M3, then subjected those cells to long term carbachol exposure, and finally determined the levels of total, M2, and M3 mAChR down-regulation using a subtype-specific mAChR immunoprecipitation protocol described previously (17). We observed that the extent of M3 mAChR down-regulation increased as the M2/M3 expression level ratio was raised. Because the formation of M2/M3 heterodimers is promoted by gradually increasing the expression levels of M2 (relative to M3) (Fig. 4), and carbachol does not modify M2/M3 heterodimerization (Table 5), our results suggest that the increase in M3 mAChR down-regulation is facilitated by the formation of M2/M3 heterodimers.
The mechanism by which M2/M3 heterodimers down-regulate more quickly than M3 homodimers remains to be elucidated. However, we can speculate that the VTILA residues, which are essential for M2 mAChR internalization and down-regulation in JEG-3 cells, may be involved in M2/M3 heterodimer internalization and down-regulation. According to this hypothesis, the endocytic machinery could recognize the VTILA motif present in the M2 component and drive the whole M2-M3 complex into the endocytic pathway.
In summary, we have demonstrated that M1, M2, and M3 mAChR form high affinity homo- and heterodimers in living cells and that the stability of those oligomers is not changed by short term agonist stimulation. In addition, our data suggest that agonist-induced M3 mAChR down-regulation is enhanced as a result of M2/M3 heterodimerization, and our data support the notion that heterodimerization has potential implications on GPCR long term regulation.
|
FOOTNOTES |
|---|
The on-line version of this article (available at http://www.jbc.org) contains supplemental Figs. 1–4.
1 Researcher for Consejo Nacional de Investigaciones Científicas y Técnicas.
2 To whom correspondence should be addressed: Dept. of Pharmacology, University of Washington, Box 357750, Seattle, WA 98195-7750. Tel.: 206-543-9457; Fax: 206-616-4230; E-mail: nathanso{at}u.washington.edu.
3 The abbreviations used are: GPCR, G protein-coupled receptor; mAChR, muscarinic acetylcholine receptor; TMD, transmembrane domain; BRET, bioluminescence resonance energy transfer; RLuc, Renilla luciferase; YFP, yellow fluorescence protein, GFP, green fluorescence protein; Smo, smoothened; QNB, quinuclidinyl benzilate; i3, third intracellular loop; PBS, phosphate-buffered saline; HA, hemagglutinin.
4 J. C. Goin and N. M. Nathanson, unpublished observations.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
