Distinct Agonist Regulation of Muscarinic Acetylcholine M2-M3 Heteromers and Their Corresponding Homomers*

Background: Muscarinic receptors can form both homo- and hetero-oligomers. Results: Co-expression of M2 and M3 receptors resulted in concurrent detection of both homomer and heteromer interactions and regulation of M2-containing forms by agonist. Conclusion: Co-existing receptor oligomers display differential regulation. Significance: Oligomers of closely related receptors display distinct properties that may be targeted therapeutically.

Members of the family of muscarinic acetylcholine receptors constitute models for understanding more broadly the superfamily of rhodopsin-like G protein-coupled receptors (GPCRs) 3 in terms of signaling, structure and pharmacology (1)(2)(3). The existence of complexes between muscarinic receptors, in the form of homomers and heteromers has been reported previously (4 -9) and the basis and importance of dimerization/oligomerization involving members of this group of GPCRs has been discussed extensively (10 -12).
The growing availability of crystal structures of different rhodopsin-like GPCRs has, in many cases, shown potential interaction interfaces between monomeric units (13)(14)(15). However, it remains uncertain if these are of physiological significance or simply reflect the most effective way of producing a crystal lattice. Moreover, it is clear that purified and reconstituted monomeric units of such receptors are able to interact with heterotrimeric G proteins in a manner that is regulated by guanine nucleotides and, therefore, in a functionally relevant manner (16 -17). In addition to this, there are widely conflicting views on the stability of GPCR-GPCR interactions (18 -21), whether this varies substantially within closely related groups of GPCRs, and on the effects or otherwise of receptor ligands on such interactions (see Ref. 11 for review). Furthermore, although it is widely accepted that co-expression of pairs of GPCRs that are able to interact may result in the concurrent presence of each of heteromers containing both GPCRs as well as the corresponding homomers, this has been challenging to demonstrate directly (22). Herein, we use co-expression of forms of the human muscarinic M 2 and M 3 receptors to explore these issues. We demonstrate concurrent detection of M 2 -M 2 , M 3 -M 3 , and M 2 -M 3 interactions at the surface of cells and distinct agonist regulation of these interactions.
VSV-SNAP-hM 2 WT cDNA constructs were produced by introducing the metabotropic glutamate 5 receptor (mGluR5) signal sequence followed by either the VSV and SNAP tags or the hemaglutinin (HA) and CLIP tags into the N terminus of the hM 2 WT or hM 3 RASSL receptor, respectively (4,23).

Generation of Flp-In TM T-REx TM 293 Cells Stably Expressing Muscarinic Receptor
Constructs-Cells were maintained in complete Dulbecco's modification of Eagle's medium (DMEM) without sodium pyruvate, 4500 mg⅐l Ϫ1 glucose, and L-glutamine, supplemented with 10% (v/v) fetal bovine serum, 1% (v/v) penicillin/streptomycin mixture, 200 g⅐ml Ϫ1 hygromycin B, and 10 g⅐ml Ϫ1 blasticidin in a humidified atmosphere. Single stable Flp-In TM T-REx TM 293 cell lines able to inducibly express the different cDNA constructs were generated as described previously (4,(22)(23). To constitutively co-express a second receptor construct in these cells they were transfected with the appropriate cDNA construct, as described above, and antibiotic-resistant clones selected using 1 mg⅐ml Ϫ1 G418. All such cell lines were initially screened by fluorescence microscopy for receptor expression based on covalent binding of SNAP-or CLIP-tagged fluorophores and subsequently by measuring specific binding of [ 3 H]QNB in cell membrane preparations.
Cell Membrane Preparations-Cells treated or not with doxycycline, were harvested after 24 h, in ice-cold phosphatebuffered saline (PBS) and pellets were frozen at Ϫ80°C for a minimum of 1 h. Pellets were thawed and resuspended in icecold 10 mM Tris, 0.1 mM EDTA, pH 7.4 (TE) buffer, supplemented with Complete TM protease inhibitor mixture (Roche Diagnostics). Cells were passed through a 25-gauge needle (5-10 times) and then homogenized on ice, by 50 strokes in a glass-on-teflon homogenizer. Homogenized cells were centrifuged at 200 ϫ g for 5 min at 4°C. The supernatant fraction was removed and transferred to microcentrifuge tubes and subjected to further centrifugation at 90,000 ϫ g for 45 min at 4°C. The pellets were resuspended in TE buffer, and protein concentration was assessed. Membrane preparations were either used directly or kept at Ϫ80°C until required.
Radioligand Binding Studies-Binding using various concentrations of [ 3 H]QNB was carried out using 5 g of membrane protein per reaction in assay buffer (20 mM HEPES, 100 mM NaCl, 10 mM MgCl 2 , pH 7.4). Nonspecific binding was defined in the presence of 10 M atropine. Reactions were incubated for 2 h at 30°C. Bound ligand was separated from free by vacuum filtration through GF/C filters (Brandel Inc.). The filters were washed twice with assay buffer, and bound ligand was estimated by liquid scintillation spectrometry.
Cell Lysate Preparation and Immunoblotting-Cells were harvested, washed twice in ice cold PBS, and pelleted by centrifugation. The pellets were resuspended in radio-immunoprecipitation buffer (50 mM HEPES, 150 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 10 mM NaF, 5 mM EDTA, 10 mM NaH 2 PO 4 , 5% ethylene glycol, pH 7.4), supplemented with Complete TM protease inhibitors mixture. Resuspended cells were then placed on a rotating wheel for 30 min at 4°C, and subsequently centrifuged at 21,000 ϫ g, for 15 min at 4°C. Supernatants were collected, and the protein concentration of the lysates determined. Samples were heated at 60 -65°C in 1ϫ Laemmli buffer (10% w/v SDS, 10 mM dithiothreitol, 20% v/v glycerol, 0.2 M Tris-HCl, 0.05% w/v bromphenol blue, pH 6.8). The required amount of protein lysate was then loaded on 4 -12% NuPAGE TM Novex Bis-Tris gels (Life Technologies). Following electrophoresis, proteins were transferred onto a nitrocellulose membrane, blocked, and subsequently incubated with the primary antibody/antiserum in 5% fat-free milk TBST (2 mM Tris-base, 15 mM NaCl, and 0.1% v/v Tween 20, pH 7.4) at 4°C, overnight. After 5 ϫ 5 min washing steps with TBST, the appropriate horseradish peroxidase-conjugated IgG secondary antibody was incubated with the membrane at room temperature for 1 h. Immunoblots were developed using enhanced chemiluminescence solution (Pierce).
Epifluorescence Imaging of Living Cells-Cells were seeded on poly-D-lysine pre-coated cover slips (0.0 mm thickness) to 500,000 cells per cover slip and incubated overnight in the presence or absence of doxycycline in complete DMEM. Cells that expressed HA-CLIP-hM 3 RASSL receptor were labeled with 5 M CLIP-Surface 488 while those expressing VSV-SNAP-hM 2 WT were labeled using 5 M SNAP-Surface 549 (New England Biolabs) in complete DMEM for 30 min at 37°C in 5% CO 2 . Cells were washed three times with complete DMEM and once with HEPES physiological saline solution (130 mM NaCl, 5 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 20 mM HEPES, pH 7.4, and 10 mM D-glucose). Cover slips were imaged using an inverted Nikon TE2000-E microscope (Nikon Instruments, Melville, NY) equipped with a 40ϫ (numerical aperture-1.3) oil-immersion Pan Fluor lens and a cooled digital Photometrics Cool Snap-HQ charge-coupled device camera (Roper Scientific, Trenton, NJ).
Homogeneous Time-resolved FRET (htrFRET)-Cells were grown to 100,000 per well on poly-D-lysine pre-treated 96-well solid black bottom plates (Greiner Bio-One). Cells were induced with doxycycline at the stated concentration for 24 h to express the receptor(s) of interest. After 24 h induction, cell surface receptor expression was monitored by adding 10 nM SNAP-Lumi4Tb or 20 nM CLIP-Lumi4Tb. After incubation at 37°C/5% CO 2 for 1 h, cells were washed three times with labeling medium (Cisbio Bioassays), and the fluorescence output was read at 620 nm using a PheraStar FS (BMG Lab technologies).
In htrFRET experiments various combinations of energy donor:acceptor were used to detect either homomers or heteromers. Detection of hM 2 WT homomers was carried out by labeling with 5 nM SNAP-Lumi4Tb with varying concentrations of SNAP-Red. hM 3 RASSL homomers were detected by labeling with 10 nM CLIP-Lumi4Tb and varying concentrations of CLIP-Red. Heteromeric interactions between hM 2 WT and hM 3 RASSL were detected using 5 nM SNAP-Lumi4Tb with varying concentrations of CLIP-Red, or the reverse combination, 10 nM CLIP-Lumi4Tb with varying concentrations of SNAP-Red. Labeling reactions were carried out for 1 h at 37°C/5% CO 2 . Cells were then washed three times with 100 l per well labeling medium and plates were either read directly after this or further processed to test the effect of receptor ligands. For the latter experiments, ligands were added to the plates after the washing step and subsequently incubated at the noted temperature and times prior to measurements using a PheraStar FS. Both the emission signal from the SNAP-Lumi4Tb or CLIP-Lumi4Tb (620 nm) and the FRET signal emanating from the acceptor SNAP-Red or CLIP-Red (665 nm) were recorded. Specific 620 nm fluorescence, 665 nm FRET or 665:620 ratio values are shown as the difference between signals obtained from induced and un-induced cells.
Triple Labeling htrFRET-Cells were plated, grown, and treated with doxycycline in the same way as described in the previous section. The cells were then simultaneously labeled with three different, but spectrally compatible htrFRET substrates. One donor was used at a time, either SNAP-Lumi4Tb (5 nM) or CLIP-Lumi4Tb (10 nM), in combination with SNAP-Green (100 nM) and CLIP-Red (100 nM). The substrates were prepared at 3ϫ the final concentrations in labeling buffer and 25 l of each was added per well. Cells were incubated for 1 h at 37°C/5% CO 2 and washed three times with labeling buffer. Ligands were added to the plates and incubated at the set time points after which the plates were read using a PheraStar FS. Two different protocols were used to measure the fluorescence output corresponding to energy transfer to the two acceptors, CLIP-Red at 665 nm and SNAP-Green at 520 nm. The donor emission at 620 nm originating from either SNAP-Lumi4Tb or CLIP-Lumi4Tb was also measured in both protocols.
Inositol Monophosphate Accumulation Assay-A suspension of 10,000 cells per assay point was prepared in stimulation buffer (10 mM HEPES, 1 mM CaCl 2 , 0.5 mM MgCl 2 , 4.2 mM KCl, 146 mM NaCl, 5.5 mM glucose, and 50 mM LiCl, pH 7.4) and incubated with ligands for 1 h at 37°C/5% CO 2 in a white Proxiplate-384 Plus (PerkinElmer). After stimulation, cells were lysed in a mixture of detection reagents prepared in lysis buffer according to the manufacturer's instructions (IP-One Tb kit, Cisbio Bioassays) and incubated for a further hour at room temperature. htrFRET was then measured using a PheraStar FS and changes in inositol monophosphate levels were calculated as ratio of 665/620 nm signals.
cAMP Inhibition Assay-A suspension of 4,000 cells per assay point was prepared in Hank's Balanced Salt Solution (HBSS). Cells were co-incubated with forskolin (5 M) and ligands for 30 min in a white Proxiplate-384 Plus. This step was followed by lysis of cells using a mixture of detection reagents prepared in lysis buffer according to manufacturer's instructions (cAMP dynamic 2 kit, Cisbio Bioassays) and incubation for 1 h at room temperature. htrFRET was measured on a PheraStar FS and the reduction of cAMP levels was calculated as ratio of 665/620 nm.

Results
To explore aspects of the potential oligomerization of the wild type (WT) human (h) muscarinic M 2 acetylcholine receptor, a construct (VSV-SNAP-hM 2 WT) was generated in which the extracellular N-terminal domain was modified to incorporate both the VSV peptide epitope tag and the SNAP protein tag sequences. This was cloned into the doxycycline-inducible locus of Flp-In TM T-REx TM 293 cells and a transfected population selected. Doxycycline-regulated expression of this construct was assessed in three distinct ways. Firstly, immunoblotting with an anti-SNAP/CLIP antiserum of SDS-polyacrylamide gel electrophoresis (SDS-PAGE) resolved lysates of cells that had been maintained for 24 h in the presence of different concentrations of doxycycline identified specific induction of the receptor construct as a polypeptide with apparent molecular mass in the region of 80 kDa (Fig. 1A). No equivalent species was detected either in lysates of these cells grown in the absence of doxycycline or in lysates of parental, non-transfected Flp-In TM T-REx TM 293 cells (Fig. 1A). Secondly, doxycycline-induced expression and effective cell surface delivery of the construct was defined by fluorescence emission at 620 nm following excitation at 337 nm, subsequent to adding the SNAP-tag label SNAP-Lumi4Tb to intact cells. This reflects covalent attachment of the label to the SNAP tag of the construct. This is located in the extracellular milieu because the N-terminal domain of cell surface targeted GPCRs is anticipated to be outside the cell (Fig. 1B). Third, specific binding of concentrations of the muscarinic antagonist [ 3 H]QNB, close to We have previously characterized VSV-and SNAP-tagged forms of both the WT muscarinic hM 3 acetylcholine receptor and a chemically engineered, Receptor Activated Solely by Synthetic Ligand (RASSL) variant (4,(23)(24). This form is not able to bind or respond effectively to acetylcholine or related synthetic analogs. Rather, it is activated by the usually inert chemical ligand clozapine N-oxide (CNO) (23)(24). Now, a modification of this construct to generate HA-CLIP-hM 3 RASSL in which the N-terminal VSV-and SNAP-tags were replaced with the HA peptide epitope tag and the CLIP protein tag sequence was generated. This was also cloned into the doxycycline-inducible locus of Flp-In TM T-REx TM 293 cells. Doxycycline-induced expression and cell surface delivery of this construct was also characterized by immunoblotting to detect each of the CLIP-and HA-tags ( Fig. 2A) and, now, by the binding of CLIP-Lumi4Tb (Fig. 2B). As anticipated from the substantially larger third intracellular loop of the hM 3 receptor compared with hM 2 , the apparent molecular mass of the predominant form of HA-CLIP-hM 3 RASSL identified by the SNAP/CLIP antiserum was in the region of 110 kDa ( Fig. 2A). Such RASSL forms of muscarinic receptors display modestly reduced affinity for many antagonist ligands (24), including [ 3 H]QNB, compared with the equivalent WT receptor. Preliminary studies indicated the K d of [ 3 H]QNB for HA-CLIP-hM 3 RASSL to be in the region of 2.5 nM. Therefore, by measuring the specific binding of a substantially higher concentration of [ 3 H]QNB (15 nM) than used for VSV-SNAP-hM 2 WT it was also possible to quantify expression of HA-CLIP-hM 3 RASSL (Fig. 2C). Noticeably, although the anti-SNAP/CLIP antiserum identified two forms of HA-CLIP-hM 3 RASSL, the HA antiserum identified only the more rapidly migrating and less prominent form ( Fig. 2A). Pretreatment of cells during the period of receptor induction with the de novo N-glycosylation inhibitor tunicamycin demonstrated the form with lower mobility, which was not identified by the anti-HA antiserum, to be the mature N-glycosylated form. Moreover, equivalent studies indicated that VSV-SNAP-hM 2 WT was also N-glycosylated in the absence of tunicamycin treatment (Fig. 3) and that these mature forms of the receptors were the predominant species present.
In cells induced to express VSV-SNAP-hM 2 WT addition of a single concentration of SNAP-Lumi4Tb, as potential energy donor, along with varying concentrations of SNAP-Red, as potential energy acceptor, generated bell-shaped homogeneous time-resolved (htr)FRET signals. These were detected as emission at 665 nm following excitation at 337 nm and are consistent with VSV-SNAP-hM 2 WT existing, at least in part, as cell surface homo-dimers/oligomers ( Fig. 4A) (4). By contrast, no such signals were produced in the absence of doxycycline-induced receptor expression (Fig. 4A). To define that these htrFRET signals reflected relevant homomeric protein-protein interactions, and not simply proximity due to the level of receptor expression causing crowding or bystander effects, we performed equivalent experiments in Flp-In TM T-REx TM 293 cells able to express the monomeric transmembrane protein CD86 (25) (Fig. 4A). This polypeptide was also modified to introduce both the VSV-and SNAP-tag sequences into the extracellular N-terminal domain. Here, addition of a combination of SNAP-Lumi4Tb and varying concentrations of SNAP-Red did not result in significant htrFRET signal in cells induced to express VSV-SNAP-CD86. Indeed, the signal was indistinguishable from cells in which expression of this construct was not induced (Fig. 4A). These experiments were carefully designed to result in cell surface expression of the same amount of VSV-SNAP-CD86 as VSV-SNAP-hM 2 WT. This was measured directly by the level of binding of SNAP-Lumi4Tb to each of the receptors, as in Fig. 1B, as fluorescence at 620 nm following excitation as 337 nm (Fig. 4B). Therefore, VSV-SNAP-hM 2 WT is present within homo-oligomers at expression levels in which such signals are not produced by a well characterized monomeric protein.
Addition of a single concentration of CLIP-Lumi4Tb, as potential energy donor, along with varying concentrations of CLIP-Red to cells induced to HA-CLIP-hM 3 RASSL also generated bell-shaped htrFRET signals (Fig. 4C). This was lacking in cells not induced to express HA-CLIP-hM 3 RASSL (Fig. 4C). These results also are consistent with homo-dimeric/oligomeric HA-CLIP-hM 3 RASSL interactions (Fig. 4C), confirming previous reports of hM 3 -hM 3 interactions (4).
To explore the potential for co-expressed hM 2 and hM 3 to exist within heteromeric complexes, cells able to express VSV-SNAP-hM 2 WT only following addition of doxycycline, were further transfected with HA-CLIP-hM 3 RASSL and clones constitutively expressing this receptor variant isolated. A substantial number of clones were characterized in preliminary studies.
These identified examples in which the levels of constitutively expressed HA-CLIP-hM 3 RASSL remained constant while expression of varying levels of VSV-SNAP-hM 2 WT could be achieved by cell maintenance in the presence of different concentrations of doxycycline. A representative clone is shown in Fig. 5. Cell surface VSV-SNAP-hM 2 WT and HA-CLIP-hM 3 RASSL were imaged individually following addition of the cell impermeant dyes SNAP-surface 549 or CLIP-surface 488. As shown in Fig. 5A the CLIP-tagged receptor was present both with and without doxycycline treatment while the SNAP-tagged receptor was only present following doxycycline treatment. Merging of these images indicated clear co-localization of the two receptors at the resolution of light microscopy (Fig. 5A). Levels of binding of CLIP-Lumi4Tb (reflecting the presence of HA-CLIP-hM 3 RASSL) to these cells were constant over a range of doxycycline concentrations. By contrast, binding of SNAP-Lumi4Tb (reflecting the appearance of VSV-SNAP-hM 2 WT) increased with increasing concentrations of doxycycline (Fig. 5B). To better quantify the relative levels of VSV-SNAP-hM 2 WT and HA-CLIP-hM 3  culated to occupy some 87-90% of HA-CLIP-hM 3 RASSL and more than 98% of VSV-SNAP-hM 2 WT in membranes prepared from cells treated or not with doxycycline. This defined that HA-CLIP-hM 3 RASSL was present at 1632 Ϯ 650 fmol⅐mg protein Ϫ1 . Moreover, because after treatment with 5 ng⅐ml Ϫ1 doxycycline the combined level of expression of muscarinic receptors was 4678 Ϯ 1481 fmol⅐mg protein Ϫ1 (Fig. 5C), these studies indicated the hM 2 WT could be expressed at up to twice the total level of hM 3 RASSL. In parallel sets of immunoblots of SDS-PAGE-resolved samples, anti-VSV antibodies only detected protein of the appropriate molecular mass, corresponding to VSV-SNAP-hM 2 WT, following treatment of the cells with doxycycline (Fig. 5D). Immunoblots using the combined anti-SNAP/CLIP antiserum confirmed that a polypeptide(s) in the region of 80 kDa (VSV-SNAP-hM 2 WT) was expressed in a doxycycline-dependent manner by these cells, while a polypeptide(s) in the region of 110 kDa (HA-CLIP-hM 3 RASSL) was expressed constitutively (Fig. 5D).
Using these cells, without doxycycline treatment, homomeric HA-CLIP-hM 3 RASSL interactions were clearly detected as htrFRET signal at 665 nm following addition of combinations of CLIP-Lumi4Tb and CLIP-Red (Fig. 6A). Interestingly, such interactions were maintained when the VSV-SNAP-hM 2 WT construct was also expressed, i.e. following treatment with doxycycline (Fig. 6A). By contrast, and as anticipated, no htrFRET signal corresponding to VSV-SNAP-hM 2 WT ho- momers was detected in the absence of doxycycline, because this receptor is absent. However, htrFRET signal corresponding to VSV-SNAP-hM 2 WT homomers appeared at the cell surface following doxycycline treatment of the cells (Fig. 6B). Importantly, in the doxycycline-induced cells addition of combinations of SNAP-Lumi4Tb and CLIP-Red also demonstrated the proximity of hM 2 WT and hM 3 RASSL, potentially within heteromeric oligomers (Fig. 6C). Moreover, hM 2 WT-hM 3 RASSL hetero-interactions were also detected when the labeling protocol was reversed to use a combination of CLIP-Lumi4Tb and, therefore, HA-CLIP-hM 3 RASSL as the energy donor, and SNAP-Red and, therefore, VSV-G-SNAP hM 2 WT as energy acceptor (Fig. 6C). Immunoprecipitation of VSV-SNAP-hM 2 WT with anti-VSV antibodies resulted in co-immunoprecipitation of anti-HA immunoreactivity, corresponding to HA-CLIP-hM 3 RASSL, only after doxycycline treatment had resulted in the co-expression of the two receptors (Fig. 6D). hM 2 is linked predominantly to Pertussis toxin-sensitive, G ifamily G proteins while hM 3 is usually largely associated with signaling via G q/11 -family G proteins. Moreover, although the acetylcholine mimetic carbachol is able to activate WT muscarinic receptors, it is reported to display very low potency at RASSL forms of this receptor family (23,24). This was confirmed in cells induced to express VSV-SNAP hM 2 WT in the constitutive presence of HA-CLIP-hM 3 RASSL. Here carbachol was able to effectively inhibit forskolin-stimulated cAMP production with pEC 50 ϭ 6.9 Ϯ 0.1 (mean Ϯ S.E., n ϭ 4). However, in cells not induced to express VSV-SNAP-hM 2 WT and, therefore, with only HA-CLIP-hM 3 RASSL present, little inhibition of forskolin-stimulated cAMP levels was noted at concentrations of carbachol up to 1 M (Fig. 7A). By contrast, both in the absence (pEC 50 ϭ 8.10 Ϯ 0.08) and presence (pEC 50 ϭ 8.00 Ϯ 0.18) (means Ϯ S.E., n ϭ 5 in each case) of VSV-SNAP hM 2 WT, CNO was able to potently stimulate the production of inositol monophosphates (Fig. 7B). This is a downstream indicator of G q /G 11 activation. Interestingly, although not reaching statistical significance, there was a trend toward higher inositol monophosphate production in response to CNO when the two receptors were co-expressed (Fig. 7B). This did not reflect a direct effect of CNO on the hM 2 WT receptor orthosteric binding pocket because neither with nor without doxycycline induction was carbachol able to cause a significant accumulation of inositol monophosphates in these cells (Fig. 7B). Importantly, however, these studies did define the functionality of the expressed constructs and confirmed the previously established selectivity of the agonist ligands in this setting (23,24).
Potential effects of ligands on the organization or stability and regulation of GPCR oligomers is a complex topic in which a range of observations have been reported (11). In cells induced with doxycycline to allow co-expression of VSV-SNAP-hM 2 WT and HA-CLIP-hM 3 RASSL, as noted above, coaddition of a combination of SNAP-Lumi4Tb and CLIP-Red resulted in detection of htrFRET signal, consistent with interactions between the two receptors (Fig. 8A). Over a period of 40 min, exposure to a concentration (10 M) of the muscarinic antagonist atropine that is sufficient to occupy fully both the hM 2 WT and hM 3 RASSL constructs, had no greater effect on the heteromer signal than addition of vehicle (Fig. 8A). By con- trast, addition of either carbachol alone, or a combination of carbachol and CNO, resulted in a substantial and rapid decline in htrFRET signal corresponding to hM 2 WT-hM 3 RASSL interactions (Fig. 8A). Unlike carbachol, CNO was unable to produce such an effect when applied alone (Fig. 8A). When equivalent studies were performed using a combination of SNAP-Lumi4Tb and SNAP-Red to detect hM 2 receptor homomers, both carbachol alone and carbachol plus CNO now resulted instead in an extensive increase in htrFRET signal (Fig.  8B). Once again neither CNO alone, nor atropine had any effect on the hM 2 homomer htrFRET signal compared with vehicletreated cells (Fig. 8B). Unlike the hM 2 WT homomer, htrFRET signal corresponding to the hM 3 RASSL homomer was not affected in these cells in a ligand-dependent manner (Fig. 8C). This was the case whether or not expression of VSV-SNAP-hM 2 WT had been induced (Fig. 8C). Importantly, the effects of carbachol on both the hM 2 WT homomer (pEC 50 ϭ 5.5 Ϯ 0.2) and hM 2 WT-hM 3 RASSL heteromer (pEC 50 ϭ 5.2 Ϯ 0.3) (means Ϯ S.E., n ϭ 4 in each case) interactions were concentration-dependent (Fig. 9).
As an extension to these studies we attempted to identify concurrently in the same cells both homo-and hetero-interactions involving VSV-SNAP-hM 2 WT. SNAP-and CLIP-Lumi4Tb have broad emission spectra. As such, upon excitation at 337 nm they can potentially transfer energy to both Green (with htrFRET output at 520 nm) and Red (with htrFRET output at 665 nm) energy acceptors. This potentially allows concurrent dual color detection of multiple interactions of the energy donor-tagged receptor. We, therefore, initially added a mixture of SNAP-Lumi4Tb and both CLIP-Red and SNAP-Green to cells induced to co-express VSV-SNAP-hM 2 WT and HA-CLIP-hM 3 RASSL. Such studies were indeed able to identify interactions of the energy donor-labeled hM 2 WT with both energy acceptor labeled hM 2 WT and hM 3 RASSL receptors concurrently (Fig. 10, A and B). Moreover, as in the individual htrFRET experiments reported above, concurrent analysis of the two distinct interactions of the energy donor-labeled hM 2 WT receptor showed an equivalent carbachol-mediated decrease in hM 2 WT-hM 3 RASSL heteromeric interactions (Fig.  10A) and increase in hM 2 WT-hM 2 WT homomeric interactions (Fig. 10B). Once again, the muscarinic antagonist atropine was without effect (Fig. 10, A and B). Finally, in cells induced to co-express the hM 2 WT and hM 3 RASSL receptors, labeling of hM 3 RASSL with the energy donor CLIP-Lumi4Tb and proportions of both hM 2 WT and hM 3 RASSL respectively with SNAP-Green and CLIP-Red, the hM 2 WT-hM 3 RASSL heteromeric interactions were again specifically decreased by treatment with carbachol (Fig. 10C). By contrast hM 3 RASSL homo-interactions were once more unperturbed by addition of any of CNO, carbachol, or atropine (Fig. 10D).

Discussion
Although it is well established that monomers of the individual subtypes of muscarinic acetylcholine receptors can exist in proximity to one another (4, 26 -28) and, indeed, have the capacity to generate dimers and/or higher-order oligomers (6 -7, 9, 18, 27-28), a broad range of issues around such interactions remain unresolved. Among these are the stability (19) or otherwise (8,18) of dimeric interactions, the overall dimensions and organization of dimeric/oligomeric complexes (7,9,25) and the implications of this for the details of interaction with heterotrimeric G proteins and downstream signal transduction (11)(12). Moreover, as muscarinic subtypes are expressed at markedly different levels in different cells and tissues this may, as suggested by some (25) but not other (19 -20) reports on both muscarinic and other rhodopsin-like family GPCRs, affect the extent of their dimerization/oligomerization. Furthermore, distinct muscarinic receptor subtypes may be coexpressed in physiologically relevant cells (29 -30). Although the capacity for heteromeric interactions between various muscarinic receptor pairs has been explored to some degree (26 -27), the propensity for this to occur concurrently with homomerization, and its implications for function, have been little explored to date. For example, M 2 and M 3 receptors are co-expressed in smooth muscle but the functional importance of this for the integration of signaling remains uncertain. Within the current studies we have, therefore, addressed a number of these issues by combinations of biochemical, biophysical, and chemical biology approaches.
Central to these studies was the use of cell surface hrtFRET, based on the incorporation of SNAP-and CLIP-tags (31-32), into various muscarinic receptor constructs. Such tagging allowed the covalent incorporation of hrtFRET-competent fluorophores into the extracellular N-terminal region of the receptors via linkage to the engineered SNAP and CLIP protein tags. Importantly, such large scale modification of the N-termi- nal domain of either the hM 2 or hM 3 receptor did not affect their basic ligand pharmacology. Of equal importance was the introduction of RASSL-inducing mutations into the hM 3 receptor constructs (23)(24). Particularly for the muscarinic receptor family, such modified GPCRs are also frequently denoted as DREADDs (Designer Receptors Exclusively Activated by Designer Drugs) (33). The associated alteration in agonist pharmacology so produced allowed for selective agonist occupancy and activation of the hM 2 WT receptor (with the acetylcholine mimetic carbachol) and the hM 3 RASSL receptor (with the muscarinic RASSL agonist CNO) in cells co-expressing the two receptor subtypes. This was confirmed by demonstrating both that carbachol-mediated inhibition of cAMP levels was observed only following induced expression of the hM 2 receptor and not when the hM 3 RASSL receptor was expressed alone, and that CNO, but not carbachol, was able to promote the production of inositol monophosphates via the hM 3 RASSL receptor, both in the absence and presence of the hM 2 WT receptor. By contrast, the antagonist atropine was able to bind to both receptors with similar affinity.
In cells able to express only either the SNAP-tagged hM 2 WT receptor or the CLIP-tagged RASSL form of the hM 3 receptor htrFRET studies provided clear evidence for homomeric interactions of each subtype. Although this was anticipated from previous work, in cells constitutively expressing hM 3 RASSL receptors, induced expression of the hM 2 WT receptor now resulted in detection of hM 2 -hM 2 interactions as well as hM 2 -hM 3 interactions at the surface of these cells without eliminating hM 3 -hM 3 interactions. The most obvious interpretation of these results is that receptor homomers can co-exist with relevant heteromers perhaps, as suggested by Herrick-Davis et al., as stable and distinct dimers (20). However, it is important to note that others have suggested such interactions to be more dynamic (18,34). Moreover, concurrent monitoring of hM 2 -hM 2 and hM 2 -hM 3 interactions in dual color studies, in which a single energy donor and two distinct energy acceptor reagents were added concurrently, also provided evidence for each of these interactions. This is the first time that such an approach has been used to examine multiple interaction partners of a GPCR simultaneously.
A common concern in studies on interactions involving cell surface transmembrane proteins is that high level expression may result in apparent interactions based on proximity that reflect the levels of expression achieved. We addressed this in two distinct ways. Firstly, for all the studies performed we generated and utilized stably transfected cell lines able to express the receptor(s) of interest in a controlled, inducible manner. Generally, studies that rely entirely on transient transfection protocols encounter challenges due to high level expression of the receptors, often incompletely processed, within subsetsof the cell population. Herein, we demonstrated that the bulk of each of the muscarinic receptor subtype constructs was appropriately N-glycosylated, as anticipated for mature, correctly trafficked GPCRs. More importantly we also generated an equivalent cell line able to inducibly express VSV-SNAP-CD86. CD86 is recognized as a monomeric single transmembrane domain protein (25). Expression of this construct to the same level as used to study VSV-SNAP-hM 2 WT generated no specific htrFRET signal upon addition of a combination of SNAPtag energy donor and acceptor species. This provided comfort that the signals produced at these levels of expression of VSV-SNAP-hM 2 WT did indeed reflect true receptor-receptor interactions.
The further key outcome of these studies is that the agonist carbachol was able to change energy transfer signals corresponding to both hM 2 -hM 2 and hM 2 -hM 3 interactions. By contrast this ligand had no effects on hM 3 -hM 3 interactions. This latter feature was hardly surprising as the hM 3 RASSL constructs used in these studies were modified to have minimal affinity for carbachol (23)(24). However, in the case of the hM 2 -hM 2 and hM 2 -hM 3 interactions the directionality of the effect of carbachol was completely different. Both in cells expressing only the hM 2 WT receptor construct, and those expressing both the hM 2 WT and the hM 3 RASSL receptors, carbachol increased the htrFRET signal corresponding to hM 2 -hM 2 homomers and did so in both a time-and concentration-dependent manner. Moreover, the EC 50 for the ligand in producing these changes in htrFRET was very similar to the affinity of carbachol at the hM 2 WT receptor. This is consistent with the effects reflecting receptor occupancy. By contrast carbachol decreased the htrFRET signal corresponding to hM 2 WT-hM 3 RASSL interactions. This was, however, once again both time-and concentration-dependent. It could be argued in the hM 2 -hM 3 co-expressing cells that the effect of carbachol was to diminish hM 2 -hM 3 interactions and that this then resulted in greater hM 2 -hM 2 interactions, i.e. to promote a heteromer to homomer transition. However, although these effects of carbachol could also be detected in triple labeling, 'dual color' studies in which the effects on the receptor complexes were measured concurrently, further studies will be required to support such a conclusion. Perhaps surprisingly, unlike carbachol, CNO was unable to influence htrFRET signals corresponding to hM 3 -   Fig. 9 induced to co-express VSV-SNAP-hM 2 WT and HA-CLIP-hM 3 RASSL were incubated with a combination of each of SNAP-Lumi4Tb, CLIP-Red, and SNAP-Green (A, B) and htrFRET signal measured at both 665 nm (SNAP-Lumi4Tb to CLIP-Red (hM 2 WT-hM 3 RASSL heteromer)) (A) and 520 nm (SNAP-Lumi4Tb to SNAP-Green (hM 2 WT homomer)) (B). Treatment with carbachol but not atropine reduced the hM 2 WT-hM 3 RASSL heteromer signal (A) and concurrently increased the hM 2 WT homomer signal (B). Equivalent studies used a combination of CLIP-Lumi4Tb and both SNAP-Green and CLIP-Red (C, D). Treatment with carbachol only reduced the hM 2 WT-hM 3 RASSL heteromer (520 nm) signal (C) while neither carbachol, atropine nor CNO had any effect on the hM 3 RASSL homomer (665 nm) signal (D). Data represent means Ϯ S.E., n ϭ3. Statistical significance as follows: *, p Ͻ 0.05, **, p Ͻ 0.001, and ***, p Ͻ 0.0001 when compared with vehicle.
RASSL-hM 3 RASSL interactions to any greater extent than addition of vehicle. This may reflect greater stability of hM 3 -hM 3 homomeric interactions compared with either hM 2 -hM 2 homomers or hM 2 -hM 3 heteromers. However, although identified as a highly selective activator of RASSL forms of muscarinic receptor subtypes, CNO is of course not a direct equivalent of carbachol. This is despite CNO acting as an apparently high efficacy agonist that, in a wide range of assays, shows broad similarity in capacity to activate and regulate the hM 3 RASSL as either carbachol or acetylcholine do at the wild type hM 3 receptor (23). Although there may be differences in details of efficacy or bias of CNO at the hM 3 RASSL receptor in end points that have not been assessed previously that may account for this difference, a distinct explanation is that the hM 2 and hM 3 receptors differ in the basis or stability of their homomeric interactions. It is notable in this regard that Calebiro et al. have provided evidence for markedly different stability and propensity of ␤ 1 -and ␤ 2 -adrenoceptors to form dimers and higherorder oligomers (25), even though these receptors are highly homologous and are activated by the same hormones. This is not the first set of studies to suggest a capacity of ligand to alter the organization and/or stability of a muscarinic receptor homomer. Although muscarinic toxin 7, a highly selective allosteric peptide ligand of the M 1 subtype, binds (35)(36) in a very different manner to carbachol or CNO (23), it has been reported to stabilize M 1 receptor homomers (35)(36). It has also been suggested that the selective M 1 receptor antagonist pirenzepine can promote dimerization of this receptor (37).
Beyond possible differences in efficacy, one further observation that is difficult to provide a clear explanation for was the marked difference in the effects of carbachol and CNO on hM 2 WT-hM 3 RASSL interactions and, thus, on hM 2 -hM 3 heteromers. Although difficult to demonstrate without making further alterations in the ligand binding pocket to alter ligand pharmacology, as has been done for the ␤ 2 -adrenoreceptor (38) and the leukotriene B(4) receptor (39), which, to some extent invalidates the basis of the experiment, it is anticipated that a ligand effect across the interface of a receptor homo-dimer/ oligomer should be symmetric. Therefore, an effect of ligand binding to one protomer is anticipated to be reciprocated by (the same) agonist occupancy of the other protomer. Herein, carbachol effects on hM 2 WT-hM 3 RASSL receptor interactions were not recapitulated by CNO. This may simply reflect that the hM 2 and hM 3 receptors are, of course, distinct species or that the makeup of hM 2 WT-hM 3 RASSL receptor heteromers is not simply 1:1 in oligomeric (7, 28) rather than dimeric configurations. No-matter the basis for the lack of symmetry here, this is topic that requires and deserves further consideration in the future.
Notwithstanding this final point, the current studies offer a broad range of novel insights into differences in ligand regulation of hM 2 -hM 2 versus hM 3 -hM 3 interactions and provide a one donor plus two acceptors strategy to concurrently assess interactions of a protein with more than a single partner. The molecular basis for the noted differences in ligand regulation between closely related receptors will provide a drive for future analysis.