Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

G protein-coupled receptors are transmembrane proteins that mediate a variety of signaling processes, such as neurotransmission, hormonal response, olfaction, and light transduction. Muscarinic receptors are members of this family and molecular cloning has revealed the existence of five different subtypes (Kubo et al., 1986; Bonner et al., 1987, 1988; Peralta et al., 1987) that show a high degree of sequence homology but differ in their ligand binding and functional properties, as well as in their tissue distribution (Hulme et al., 1990). They are predicted to be composed of seven hydrophobic transmembrane domains (TMD; I-VII) connected by alternating cytoplasmic and extracellular loops, an extracellular amino-terminal domain, and an intracellular carboxy-terminal segment. Although they are generally considered closely packed structures, an increasing amount of evidence indicates that they can behave structurally in a fashion analogous to multiple subunit receptors. In previous experiments, Maggio et al. (1993a) showed that when truncated M₂ or M₃ receptors (containing TMDs I-V, named M₂- or M₃-trunc) were coexpressed in African green monkey kidney (COS-7) cells with gene fragments coding for the corresponding carboxy-terminal receptor portions (containing TMDs VI and VII, named M₂- or M₃-tail), muscarinic receptors with ligand-binding properties and functional activity similar to the wild-type receptors were obtained. Because the separate transfection of each fragment individually was not able to show any binding, it was supposed that the two receptor domains interact with each other to form a macromolecular complex with the same characteristics as the wild-type receptor. Similar results have been obtained with bacteriorhodopsin (Kahn and Engelman, 1992), ₂-adrenergic (Kobilka et al., 1988), rhodopsin (Ridge et al., 1995), V₂ vasopressin (Schöneberg et al., 1996), and D₂ dopamine receptors (Barbier et al., 1996). Furthermore, experiments performed with the M₃ muscarinic receptor have shown that it can be split not only at the level of the third cytoplasmic loop (i3 loop), but also in the second intra- and third extracytoplasmic loops, and that it can retain its binding properties (Schöneberg et al., 1995).

As all muscarinic receptors have a long i3 loop that connects TMDs V to VI, it has been speculated that wild-type receptors can actually express the domains constituted by the amino- (containing TMDs I-V) and carboxyl-terminal receptor portions (containing TMDs VI and VII) as separate units capable of interacting with each other. Furthermore, this interaction was thought to occur not only intra- but also intermolecularly. This was demonstrated by creating two chimeric receptor molecules, ₂/M₃ and M₃/₂, in which the carboxy-terminal receptor portions (including TMDs VI and VII) were exchanged between the _2C-adrenergic and the M₃ muscarinic receptors (Maggio et al., 1993b). Although transfection of the two chimeric constructs alone into COS-7 cells did not result in any detectable binding activity, coexpression of the two mutant receptors resulted in a significant number of specific-binding sites for the muscarinic ligand N-[³H]methylscopolamine ([³H]NM) and the adrenergic ligand [³H]rauwolscine. Other experiments performed with functionally impaired muscarinic receptors confirmed these findings (Maggio et al., 1993b).

In a later article, Maggio et al. (1996) demonstrated that receptor interaction is prevented by the shortening of the i3 loop. Chimeric ₂/M₃ and M₃/₂ receptors with the deletion of a large portion of the i3 loop were no longer able to cross-interact and bind M₃ muscarinic and ₂-adrenergic receptor ligands when coexpressed in the same cells. Furthermore, although the activity of a functionally impaired M₃ muscarinic receptor mutant in which 16 amino acids of the amino-terminal part of the i3 loop were substituted with the corresponding M₂ sequence [M₃/M₂(16aa)] was rescued by coexpression with a healthy M₃-trunc fragment bearing the correct sequence, the deletion of 196 amino acids from the i3 loop of M₃/M₂(16aa), although leaving intact the binding characteristics of the receptor, completely prevented the functional rescue operated by M₃-trunc. Experiments performed with a short form of the wild-type M₃ muscarinic receptor (M₃-short), in which 196 amino acids of the i3 loop had been removed, demonstrated that the large deletion of the i3 loop leaves unvaried the binding characteristics of the receptor, the phosphatidyl inositol hydrolysis activity, and the ability of the receptor to internalize.

The property of receptors to cross-interact at the molecular level and the ability of cotransfected fragments to form a macromolecular receptor complex capable of binding with high-affinity muscarinic ligands leads to the hypothesis that when different subtypes of muscarinic receptors are coexpressed in the same cells, they might interact at the molecular level and form a new pharmacological receptor. In agreement with this view, the cotransfection of M₂-trunc with M₃-tail results in a fragmented chimeric M₂-trunc/M₃-tail receptor capable of binding muscarinic ligands with a high affinity (the other chimeric combination, M₃-trunc + M₂-tail did not show any binding; Maggio et al., 1993a). Barbier et al. (1998) studied the antagonist binding characteristics of the chimeric M₂-trunc/M₃-tail receptor with the purpose of identifying compounds capable of discriminating between this chimera and the wild-type M₂ and M₃ muscarinic receptors. The pharmacological profile of this fragmented chimeric receptor was clearly different from that of M₂ and M₃ wild-type receptors; whereas most of the compounds tested showed an intermediate affinity between M₂ and M₃, tripitramine, and pirenzepine showed a peculiar pharmacological profile. Tripitramine, a selective antagonist for the M₂ receptor (Maggio et al., 1994), also displayed a high-affinity binding for the chimeric M₂-trunc/M₃-tail receptor that was more than 100-fold higher than that found for the M₃ muscarinic receptor. Pirenzepine, on the other hand, had a 12- and 3-fold higher affinity for the chimeric receptor than for the M₂ and M₃ wild-type receptors, respectively. In the present work, using the unique pharmacological profile of M₂-trunc/M₃-tail chimeric receptor, we provide evidence that the exchange of the amino- and carboxy-terminal receptor domains between the wild-type M₂ and M₃ receptors leads to the formation of a heterodimeric M₂-trunc/M₃-tail muscarinic receptor.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Preparation of Mutant Receptor Constructs. HM₂pcD and RM₃pcD (Bonner et al., 1987), two mammalian expression vectors containing the entire coding sequence of the human M₂ and the rat M₃ muscarinic receptors, were used to construct the various fragmented (M₂-trunc and M₃-tail) and mutant [M₃-short and M₂ (Asn404  Ser)] muscarinic receptor genes.

pcDM₂-Trunc and pcDM₃-Tail. The construction of these fragmented receptors has been described previously (Maggio et al., 1993a). The encoded M₂-trunc receptor contains an in-frame stop codon after the amino acid codon Ser-283 of the human M₂ sequence, whereas M₃-tail codes for the 202 carboxy-terminal amino acids (from Leu-388 to Leu-589) of the rat M₃ muscarinic receptor (Fig. 1). When transfected together to allow an efficient coexpression of the two muscarinic receptor fragments, a plasmid with the two transcriptional units (pcDM₂-trunc/M₃-tail) was used. The pcDM₂-trunc/M₃-tail plasmid contains the simian virus 40 early region promoter, the M₃-tail coding sequence, and the segment carrying the simian virus 40 late region polyadenylation signal from pcDM₃-tail cloned into the blunt-ended SalI site of the pcDM₂-trunc.

View larger version (48K): [in this window] [in a new window]   Fig. 1.   Schematic representation of the wild-type human M2 and rat M3, the fragments M2-trunc and M3-tail, and the mutants M3-short and M2 (Asn404  Ser) muscarinic receptors. The truncated fragment, M2-trunc, contains the amino-terminal domain, the first five hydrophobic transmembrane regions and the initial portion (56 amino acids) of the i3 loop of the wild-type muscarinic M2 receptor. The M3-tail fragment contains the final portion of the i3 loop (105 amino acids), the last two hydrophobic transmembrane regions, and the carboxy-terminal segment of the wild-type M3 muscarinic receptor. The short construct (M3-short) represents a receptor in which 196 amino acids of the i3 loop have been deleted; the remaining i3 loop is 43 amino acids long. The point mutant M2 (Asn404  Ser) has the asparagine 404 replaced with serine.

pcDM₃-Short. The construction of this mutant receptor has been described previously (Maggio et al., 1996). The encoded M₃-short receptor has a deletion of 196 amino acids in the i3 loop, from Ala-274 to Lys-469 (Fig. 1).

pcDM₂(Asn404  Ser). A 1.8-kilobase ApaI-HindIII and a 4.1-kilobase DraIII-HindIII fragments were removed from HM₂pcD and ligated together with a piece of DNA obtained by cutting with ApaI and DraIII, a fragment derived from the polymerase chain reaction of two partially complementary oligonucleotides (46- and 47-mer). The 47-mer oligonucleotide carried a point mutation that transforms the amino acid codon 404 of the HM₂ sequence from asparagine to serine (AAT  AGC; Fig. 1). The integrity of the coding sequences was confirmed by dideoxy-DNA sequencing of the regions derived from the synthetic oligonucleotides and by restriction endonuclease analysis.

Transient Expression of Mutant Receptors. COS-7 cells were incubated at 37°C in a humidified atmosphere (5% CO₂) and grown in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal bovine serum, 2% (v/v) L-glutamine 200 mM, 1% (v/v) penicillin (10,000 U/ml) and streptomycin (10 mg/ml) solution, and 1% (v/v) minimal essential medium nonessential amino acid solution. The cells were seeded at a density of ~1.5 × 10⁶ cells/100-mm dish and 24 h later transiently transfected with the various receptor constructs (a total of 4 µg of plasmid DNA/dish) by a DEAE-dextran method (Cullen, 1987).

Radioligand-Binding Assays. COS-7 cells were harvested 72 h after transfection. Cells were washed twice with PBS, scraped into ice-cold binding buffer (25 mM sodium phosphate containing 5 mM magnesium chloride at pH 7.3), and homogenized for 30 s using a polytron (setting 5). Membranes were pelleted at 15,000g for 30 min at 4°C. The pellet was resuspended in 5 ml of ice-cold binding buffer and the membranes were rehomogenized. Saturation experiments were performed with 7 different [³H]NM concentrations (12.5-800 pM). Inhibition experiments were carried out with 16 different concentrations of the cold ligand (muscarinic agonist or antagonist), against an [³H]NM concentration of 200 pM. Atropine (1 µM) was used to define nonspecific binding. Incubation was at room temperature for 3 h. The bound ligand was separated on glass fiber filters (Whatman GF/B) with a Brandel cell harvester. The filters were washed three times with 4 ml ice-cold binding buffer. Filters were transferred to vials and counted on a liquid scintillation counter. Results are the mean ± S.E. of at least four experiments, each performed in duplicate.

Statistical Analysis. K_d and B_max values of [³H]NM were determined in direct saturation experiments, whereas IC₅₀ values for all other compounds were calculated in competition curves fitted to one- or two-site binding models using the iterative, nonlinear, least-squares regression analysis of the Kaleidagraph software running on a Macintosh computer. To determine whether the data were best fitted by a one- or two-site model, the residual sums of squares were compared by Scheffé's F test.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Cotransfection of M₂-trunc with M₃-tail (see Fig. 1) results in the formation of a chimeric receptor with a high-affinity binding for [³H]NM. Table 1 shows the affinity constants obtained by Barbier et al. (1998) for tripitramine, pirenzepine, methoctramine, para-fluoro-hexahydrosiladifenidol, and [³H]NM at M₂, M₃, and M₂-trunc/M₃-tail chimeric receptors. These compounds were used in our work to explore the possibility that cross-interaction between the wild-type M₂ and M₃ receptors could result in the formation of the heterodimeric M₂-trunc/M₃-tail receptor. We initially expressed the wild-type M₂ receptor together with the M₃-tail fragment in COS-7 cells. Then we tested pirenzepine in a displacement curve against [³H]NM. The idea was to see whether the M₃-tail fragment (which by itself does not bind muscarinic ligands) was able to recruit the amino-terminal receptor domain of the muscarinic M₂ receptor in the formation of the chimeric receptor. As pirenzepine has a higher affinity for the chimeric M₂-trunc/M₃-tail receptor than for the M₂ receptor (Table 1), the displacement curve should have been shallow and best fitted by a two-site model. In agreement with our anticipation, the displacement curve we obtained with pirenzepine had a Hill coefficient of 0.79 ± 0.02 (significantly lower than unity), and the curve was best fitted by a two-site model (Fig. 2a). Thirty-four percent of receptors were in the high-affinity state, indicating that a fair amount of receptors were recruited in the chimeric receptor form. The IC₅₀s that we calculated were very close to those found with pirenzepine for the wild-type M₂ receptor and the coexpressed M₂-trunc and M₃-tail receptor fragments (compare Tables 1 and 2). In a similar experiment in which pirenzepine was tested on membranes obtained from cells transfected separately with M₂ and M₃-tail and then pooled together, we obtained a Hill coefficient not far from unity (1.01 ± 0.01; Fig. 2a) and an IC₅₀ equal to that of the wild-type M₂ receptor (Table 2), indicating that the cotransfection of M₂ and M₃-tail in the same cells was the condition to observe two affinities.

View this table: [in this window] [in a new window]   TABLE 1 Binding affinities of wild-type muscarinic M2 and M3 receptors and the fragmented chimeric M2-trunc/M3-tail receptor for five different muscarinic receptor antagonists COS-7 cells were transfected in 100-mm dishes with 4 µg of plasmid DNA by a DEAE-dextran method. [3H]NM Kd values were determined in direct binding assays. IC50 values were obtained in competition binding experiments and transformed into Ki by the equation of Cheng and Prusoff (1973). Data are from Barbier et al. (1998).

View larger version (32K): [in this window] [in a new window]   Fig. 2.   Displacement of specific [3H]NM binding to cotransfected M2 and M3 wild-type and mutant muscarinic receptors by pirenzepine and tripitramine. a, pirenzepine inhibition curve () in COS-7 cells cotransfected with the M2 wild-type receptor and the M3-tail fragment. The curve was best fitted by a two-site model (p < .05, Scheffé's F test); the two populations of binding sites showed a 14-fold difference in affinity. b, tripitramine inhibition curve () for the cotransfected M3 wild-type receptor and M2-trunc fragment. Also, this graph was best fitted by a two-site model (p < .05, Scheffé's F test); the two populations of binding sites showed a 181-fold difference in affinity. c, tripitramine inhibition curve () for the cotransfected M3-short receptor and M2-trunc fragment. In this case, a single population of binding sites was identified by the fitting. d, tripitramine inhibition curve () for the cotransfected M2 (Asn404  Ser) and M3 wild-type receptor. The curve was best fitted by a two-site model (p < .05, Scheffé's F test); the two populations of binding sites showed a 346-fold difference in affinity. a, b, and d, , represent curves obtained in membranes from COS-7 cells separately transfected with M2 and M3-tail, M3 and M2-trunc, and M2 (Asn404  Ser) and M3, respectively. c, , curve obtained with the M3-short receptor transfected alone.

View this table: [in this window] [in a new window]   TABLE 2 Antagonist binding parameters of wild-type, fragmented, and mutant muscarinic receptors coexpressed in COS-7 cells COS-7 cells were cotransfected () or separately transfected () with plasmid DNAs coding for M2 and M3 wild-type receptors and for different muscarinic receptor mutants. A total amount of 4 µg of DNA was transfected (the relative amount of each plasmid is given in parentheses). IC50 values were obtained in displacement curves against 200 pM [3H]NM. Competition curves were fitted to one- or two-site binding models; to determine whether the data were best fitted by a one- or two-site model, the residual sums of squares were compared by Scheffé's F test. Data are presented as mean ± S.E. of at least four experiments, each performed in duplicate.

[³H]NM binding to the M₂ receptor was not affected by the presence of the M₃-tail fragment. The saturation curve was best fitted by a one-site model and the K_d value was very similar to that of the M₂ receptor transfected alone (Table 3).

View this table: [in this window] [in a new window]   TABLE 3 [3H]NM binding parameters of wild-type, fragmented, and mutant muscarinic receptors coexpressed in COS-7 cells COS-7 cells were cotransfected () or separately transfected () with plasmid DNAs coding for M2 and M3 wild-type receptors and for different muscarinic receptor mutants. A total amount of 4 µg DNA was transfected (the relative amount of each plasmid is given in parentheses). Binding parameters were calculated in saturation experiments. Data are presented as mean ± S.E. of three experiments, each performed in duplicate.

In another set of experiments, we cotransfected the wild-type M₃ receptor with the M₂-trunc receptor fragment. Then we tested tripitramine in a displacement curve against [³H]NM. As in the previous experiment, the idea was to see whether the M₂-trunc fragment could recruit the carboxy-terminal domain of M₃ to form the chimeric receptor. Due to the big difference in affinity between the M₃ and the chimeric receptor, the inhibition curve obtained with tripitramine was clearly biphasic (Fig. 2b). The curve was best fitted by a two-site model formula and the affinities found were very close to those of the M₃ wild-type receptor and the fragmented M₂-trunc/M₃-tail chimeric receptor (compare Tables 1 and 2). Similar experiments performed with tripitramine on membranes obtained from COS-7 cells separately transfected with M₃ and M₂-trunc and then pooled together gave an inhibition curve indicative of a single population of binding sites (Hill coefficient 0.98 ± 0.03; Fig. 2b) with the affinity of the muscarinic M₃ receptor (Table 2).

As observed with the previous cotransfection, [³H]NM binding to the M₃ receptor was not affected by the presence of the M₂-trunc fragment. The saturation curve was best fitted by a one-site model and the K_d value was very similar to that of the M₃ receptor (Table 3).

Using COS-7 cells cotransfected with M₃-tail + M₂ and M₂-trunc + M₃, we tested two other compounds: para-fluoro-hexahydrosiladifenidol and methoctramine. The data obtained with these two compounds in both cotransfections did not fit with a two-site model at the level of p < .05 by Scheffe's F test, probably because they do not differentiate powerfully between M₂-trunc/M₃-tail on the one hand and M₂ or M₃ on the other (Table 2).

As mentioned in the introduction, receptor interaction is prevented by shortening of the i3 loop. We cotransfected a short form of the wild-type M₃ muscarinic receptor (M₃-short, Fig. 1) in which 196 amino acid of the i3 loop had been removed, with the M₂-trunc receptor fragment, and then we tested tripitramine in a displacement curve against [³H]NM. Although we obtained a biphasic curve (see above) with the cotransfection of the muscarinic M₃ wild-type receptor with the M₂-trunc fragment, in this case we observed a monophasic curve, the expression of a single population of binding sites (the Hill coefficient was not significantly different from unity, Fig. 2C). The affinity of tripitramine was equal to that found when the M₃-short receptor was transfected alone (Table 2). These data clearly confirm that the recruitment of the carboxyl terminal receptor portion of M₃ by the M₂-trunc fragment (a fundamental requirement for the formation of the chimeric receptor) directly depends on the length of the i3 loop. [³H]NM binding parameters for the cotransfection tested above are reported in Table 3.

In the previous experiment we described interactions between receptors as a whole and receptor fragments. We did not test the two wild-type M₂ and M₃ receptors together because we do not have yet a ligand fully selective for the hybrid M₂-trunc/M₃-tail receptor. For this reason, the possibility remains that when the two wild-type M₂ and M₃ receptors are coexpressed together, the steric hindrance of the proteins could prevent receptor interaction. To address this issue, we constructed a point mutant M₂ receptor in which the asparagine 404 in the TMD VI was replaced by serine: M₂ (Asn404  Ser; Fig. 1). This amino acid (conserved in all five muscarinic receptors) has been demonstrated to drastically reduce antagonist binding (Blüml et al., 1994). In line with this finding, [³H]NM up to 2 nM was unable to bind M₂ (Asn404  Ser); nevertheless, when this receptor was transfected with the M₃-tail receptor fragment, [³H]NM specific binding was detected, indicating the correct insertion of the receptor into the membrane; furthermore, the pirenzepine inhibition curve was best fitted by a one-site model (Hill coefficient = 1.05 ± 0.07) and the IC₅₀ value was 135 ± 12 nM. Then we transfected M₂ (Asn404  Ser) together with the wild-type M₃ receptor and we tested tripitramine in a displacement experiment against [³H]NM. As can be seen in Fig. 2d, this cotransfection resulted in the appearance of a high-affinity binding for tripitramine. The best ratio between M₂ (Asn404  Ser) and M₃ in terms of plasmid DNA amount was 3.3 and 0.7 µg, respectively. With this ratio of plasmid DNA we had 17% of the total binding in the high-affinity state, whereas the rest was in a low-affinity state. These results were not observed on membranes obtained from COS-7 cells separately transfected with M₂ (Asn404  Ser) and M₃ and then pooled together (Fig. 2d). We performed this experiment also with a reduced amount of plasmid DNA to have a more physiological level of expression. Plasmid DNA was reduced to 0.94 and 0.2 µg, respectively for M₂ (Asn404  Ser) and M₃ receptors (the difference was made up with 2.86 µg of empty vector). In this condition, the total amount of [³H]NM binding calculated in saturation experiments decreased from 173 ± 36 to 36 ± 7 fmol/mg of proteins; the K_d value was similar in both experiments and was similar to the value obtained with the muscarinic M₃ receptor transfected alone (Table 3). Also, with a lower amount of receptors expressed in the membrane, tripitramine displacement experiments resulted in the appearance of high-affinity binding sites (Table 2).

We can conclude that a molecular interaction occurs between M₂ (Asn404  Ser) and M₃ when they are together in the same cells, and that a receptor with a high affinity for tripitramine is formed. Because M₂ (Asn404  Ser) and M₂ wild-type have only one amino acid difference, they probably have the same steric impediment, and consequently steric hindrance does not seem to play against the formation of the heterodimeric M₂-trunc/M₃-tail receptor.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

In the last few years, an increasing number of articles have been published that support the existence of G protein-coupled receptor dimers. Receptors that show some evidence of dimerization are muscarinic M₂ and M₃ (Potter et al., 1991; Maggio et al., 1993b, 1996; Wreggett and Wells, 1995), ₂-adrenergic (Hebert et al., 1996, 1998), neurokinin NK-1 and -2 (Huang et al., 1994, 1995), adenosine A₁ (Ciruela et al., 1995), dopamine D₂ (Ng et al., 1996), angiotensin II (Monnot et al., 1996), -opioid (Cvejic and Devi, 1997), metabotropic glutamate mGluR5 (Romano et al., 1996), Ca²⁺-sensing (Bai et al., 1998, 1999), and -aminobutyric acid (GABA)_B (Kaupmann et al., 1998; Jones et al., 1998; White et al., 1998). There is currently much debate on the role of dimers in G protein-coupled receptor, but three recent works on GABA_B receptors (Kaupmann et al., 1998; Jones et al., 1998; White et al., 1998) have shed light on the functional significance of this phenomenon. Neither GABA_B receptor 1 (R1) nor GABA_BR2, when expressed individually, activates inwardly rectifying K⁺ channels; however, the combination of GABA_BR1 and GABA_BR2 confer robust stimulation of K⁺ channel activity. This indicates that heteromeric assembly of the two GABA_B receptor subunits is essential to confer function and suggests that intermolecular interaction could be of general importance for G protein-coupled receptor activity. The molecular mechanism underlying dimerization is not known, but recently Gouldson et al. (1997, 1998), using computer simulations, have proposed domain swapping in G protein-coupled receptor dimerization. Domain swapping is a very efficient method of forming dimers, because the interactions within the monomer are reused in the dimer.

Domain swapping has been proposed to be the base of ₂/M₃ and M₃/₂ chimeric receptors heterodimerization. This phenomenon implies that fragment exchange occurs between the two chimeric receptors and that the trunc and tail portions of the ₂ and M₃ receptors (even though they are present in separate proteins) have the ability to recognize each other and reconstitute the wild-type receptor complex. A long i3 loop allows considerable spatial freedom for the fragments belonging to the ₂ and M₃ receptors located in the two chimeric ₂/M₃ and M₃/₂ receptors (Maggio et al., 1996). As all muscarinic receptors have a long i3 loop, it is reasonable to think that fragment exchange can occur also between the wild-type muscarinic receptors. If this proves to be true, interaction between two different muscarinic receptor subtypes could lead to the spontaneous formation of a new heterodimeric receptor.

It has previously been found that the amino-terminal fragment of the M₂ receptor (M₂-trunc) is able to interact with the carboxy-terminal part of the M₃ receptor (M₃-tail) to form a fragmented chimeric receptor (Maggio et al., 1993a). A pharmacological screening with muscarinic antagonists identified several compounds with high affinity for this fragmented M₂-trunc/M₃-tail receptor hybrid (Barbier et al., 1998). In this study, two of these compounds, pirenzepine and tripitramine, were used to show that interaction occurs between the M₂ and the M₃ muscarinic receptors. Displacement experiments performed with these two compounds against [³H]NM indicated that the wild-type M₂ and M₃ receptors are able to complex with the M₃-tail and M₂-trunc fragments, respectively. These receptor fragments are able to recruit a fraction of the wild-type M₂ and M₃ muscarinic receptors in the formation of the chimeric M₂-trunc/M₃-tail receptor. A definitive experiment using both M₂ and M₃ wild-type receptors in the same preparation could not be performed in our work because of the lack of a ligand fully selective for the chimeric M₂-trunc/M₃-tail receptor. For this reason, the use of a point mutant M₂ (Asn404  Ser) was required to demonstrate that two entire receptors are still able to interact with each other. The M₂ (Asn404  Ser) receptor does not bind [³H]NM up to 2 nM, but binding can be rescued by cotransfection with the M₃-tail fragment, indicating that this receptor is correctly inserted into the plasma membrane. Cotransfection of this mutant M₂ (Asn404  Ser) with the wild-type M₃ receptor results in the appearance of a high-affinity binding for tripitramine because of the interaction of the trunc part of M₂ (Asn404  Ser) with the tail part of M₃. This experiment rules out the possibility that steric hindrance can interfere with receptor cross-interaction, and strengthens the concept that wild-type muscarinic receptors are able to interact. The possibility that the M₃-tail fragment or the full-length M₃ could serve only as chaperone to target the M₂ (Asn404  Ser) receptor to the cell surface and that the high-affinity binding for tripitramine in the cotransfection of M₂ (Asn404  Ser) with the wild-type M₃ receptor could be caused by the rescue of M₂ binding and not by the interaction of the trunc part of M₂ (Asn404  Ser) with the tail part of M₃ is unlikely. Although the inhibition curve obtained with pirenzepine in the cotransfection of M₃-tail with M₂ was best fitted by a two-site model, the inhibition curve obtained with the cotransfection of M₃-tail with M₂ (Asn404  Ser) was best fitted by a one-site model; the affinity found was similar to the high-affinity binding obtained with the cotransfection of M₃-tail with M₂ and there was no evidence of a low-affinity binding component, indicative of the rescue of M₂ binding. Nevertheless, it is possible that for the efficient targeting and the correct folding of the M₂ (Asn404  Ser) receptor in the plasma membrane, the presence of M₃-tail or the full-length M₃ receptor is required.

The comparison of these data and those obtained with the chimeric ₂/M₃ and M₃/₂ receptors deserves comment. In the latter case, there is no interaction between the amino- and carboxy-terminal receptor parts inside the chimeric ₂/M₃ and M₃/₂ receptors because they originate from two different receptor types; therefore, all the fragments are in some way forced to cross-interact. In the case of the wild-type M₂ and M₃ receptors, the intramolecular interaction of the amino- and carboxy-terminal parts inside the receptors competes with the intermolecular interaction with the foreign fragments. For example, in the cotransfection of M₂ with M₃-tail, the trunc part of M₂ will have a far greater probability of hitting the tail part of M₂ than the foreign M₃-tail fragment. As we calculated in our experiments that up to 34% of the total binding could be found in the hybrid receptor form, it is likely that the relative affinity between the trunc part of M₂ and the M₃-tail receptor fragment is high enough to overcome this disadvantage. Furthermore, it is possible that ligands, depending on their relative affinities for the hybrid and the wild-type receptors, can favor the intra- or the intermolecular interaction.

In our experiments, the existence of heterodimers is entirely based on pharmacological properties of coexpressed receptors and no biochemical evidence of their existence were presented. For the moment, we do not know the functional implications that the interaction between M₂ and M₃ receptors may have, and additional experiments will be necessary to reveal the biological significance of this phenomenon. As coexpression of different subtypes of muscarinic receptors in the same cells occurs in several tissues (Weiner et al., 1990; Fukamauchi et al., 1993; Eglen et al., 1994), it is possible that heterodimeric receptors could be formed in vivo. Based on the high structural homology found among all G protein-coupled receptors, our findings should be of general importance for the entire class of integral membrane proteins.