Roles of threonine 192 and asparagine 382 in agonist and antagonist interactions with M₁ muscarinic receptors

Abstract

1. Conserved amino acids, such as Thr in transmembrane domains (TM) V and Asn in TM VI of muscarinic receptors, may be important in agonist binding and/or receptor activation. In order to determine the functional roles of Thr192 and Asn382 in human M₁ receptors in ligand binding and receptor activation processes, we created and characterized mutant receptors with Thr192 or Asn382 substituted by Ala.

2. HM₁ wild-type (WT) and mutant receptors [HM₁(Thr192Ala) and HM₁(Asn382Ala)] were stably expressed in A9 L cells. The K_d values for ³H-(R)-QNB and K_i values for other classical muscarinic antagonists were similar at HM₁(WT) and HM₁(Thr192Ala) mutant receptors, yet higher at HM₁(Asn382Ala) mutant receptors. Carbachol exhibited lower potency and efficacy in stimulating PI hydrolysis via HM₁(Thr192Ala) mutant receptors, and intermediate agonist activity at the HM₁(Asn382Ala) mutant receptors.

3. The Asn382 residue in TM VI but not the Thr192 residue in TM V of the human M₁ receptor appears to participate directly in antagonist binding. Both Thr192 and Asn382 residues are involved differentially in agonist binding and/or receptor activation processes, yet the Asn382 residue is less important than Thr192 in agonist activation of M₁ receptors.

4. Molecular modelling studies indicate that substitution of Thr192 or Asn382 results in the loss of hydrogen-bond interactions and changes in the agonist binding mode associated with an increase in hydrophobic interactions between ligand and receptor.

Introduction

Muscarinic acetylcholine receptors are members of the large G protein-coupled receptor family featuring seven transmembrane domains, an extracellular N-terminus and an intracellular C-terminus. In humans, five receptor subtypes (M₁ to M₅) Caulfield & Birdsall, 1998) have been cloned thus far. The M₁, M₃ and M₅ mainly couple to the activation of phospholipase C_β through the pertussis toxin-insensitive G_q/11 family of G proteins; while M₂ and M₄ preferentially couple to the inhibition of adenylyl cyclase through the pertussis toxin-sensitive G_i/0 family of G proteins (Hulme et al., 1990; Caulfield, 1993). The molecular mechanisms involved in ligand binding and receptor activation processes have been intensively investigated for many years, yet remain unclear (for recent reviews, see Wess 1996; 1997).

Previous molecular modelling studies predicted that conserved amino acids in TM domains of muscarinic receptors, such as an Asp residue in TM III, a Thr residue in TM V, and Asn and Tyr residues in TM VI (see Table 1), were involved in agonist binding and/or receptor activation processes (Trumpp-Kallmeyer et al., 1992; Ward et al., 1992; Nordvall & Hacksell, 1993). The Asp residue is highly conserved within all G-protein coupled receptors that bind biogenic amines. The carboxylate side chain of the Asp residue serves as a counterion for the quaternary amine headgroup of biogenic amine ligands (Hulme et al., 1990; Fraser et al., 1994; Schwarz et al., 1995). Binding of the endogenous muscarinic agonist acetylcholine (ACh) is initiated by an ion-ion interaction between the negatively charged Asp residue in TM III and the positively charged quaternary amine (Wess et al., 1991; 1992; Blüml et al., 1994). In addition, other interactions, such as hydrogen-bond and/or hydrophobic interactions between conserved residues and the ester moiety, might be responsible for the selective receptor activation processes. Using site-directed mutagenesis technology and pharmacological studies, Wess and colleagues identified Thr234 in TM V and Tyr506 in TM VI of rat M₃ receptors as important residues in agonist binding and receptor activation processes (Wess et al., 1991; 1992). The Thr234 in rat M₃ receptors is equivalent to the Thr192 in M₁ receptors (Table 1). Using Cys scanning mutagenesis, the Thr192 residue was identified to be involved in ACh binding interactions with the acetyl methyl group of ACh (Allman et al., 1997). In contrast to the predicted roles by molecular modelling studies based on the M₁ subtype, Asn507 in TM VI of rat M₃ receptors was found to be critical for antagonist binding, yet less involved in agonist binding and receptor activation in COS-7 cells (Blüml et al., 1994). The Asn507 in rat M₃ receptors is equivalent to the Asn382 residue in M₁ receptors (Table 1). Previous attempts to determine the functional role of Asn382 in TM VI of M₁ receptors failed, reportedly due to a lack of measurable binding in membrane preparations from COS-7 cells in an Ala scanning mutagenesis study (Ward & Hulme, 1997), raising the possibility that mutation of Asn382 to Ala altered receptor expression and folding. There is a significant discrepancy between the results from three-dimensional molecular modelling studies and from site-directed mutagenesis and pharmacological studies. These discrepancies have received attention in discussions regarding the presumed roles of hydrogen-bond donors/acceptors in TM domains of monoamine receptors in agonist binding and receptor activation processes (Schwartz & Rosenkilde, 1996; Strange, 1996). Thus, further studies are required to investigate the functional role of this highly conserved Asn residue as a potential hydrogen-bond donor and/or acceptor in TM VI in ligand binding and receptor activation processes.

Table 1.

Conserved amino acid residues involved in ligand interactions by molecular modelling studies (Trumpp-Kallmeyer et al., 1992; Nordvall & Hacksell, 1993; 1995)

Selective muscarinic agonists may be useful for the treatment of Alzheimer's disease (Growdon, 1997). Understanding the molecular mechanism(s) involved in ligand binding and receptor activation processes would be helpful in designing and developing selective muscarinic agonists. In order to examine the functional roles of the Thr192 and Asn382 residues in HM₁ receptors in ligand binding and receptor activation processes, and to determine if they function as previously reported in rat M₃ receptors or as predicted in molecular modelling studies, a site-directed mutagenesis approach was used to replace Thr192 and Asn382 with Ala in HM₁ receptors. Stable A9 L cell lines expressing HM₁ wild-type [HM₁(WT)] or HM₁ mutant receptors [HM₁(Thr192Ala) and HM₁(Asn382Ala) receptors] were created, and characterized for their ligand binding properties and functional activities. The data reported here indicate that both Thr192 and Asn382 residues are involved in agonist binding and receptor activation processes as predicted in modelling studies, and that Asn382, but not Thr192, directly participates in antagonist binding as reported previously in rat M₃ receptors.

Methods

Materials

Plasmids HM₁pcD1 encoding the human M₁ muscarinic receptor and pcDneo were kindly provided by Dr Tom I. Bonner (NIH) (Bonner et al., 1988) and Dr Jürgen Wess (NIDDK), respectively. Dulbeco's Modified Eagle's Medium (DMEM) was purchased from GIBCO BRL (Grand Island, NY, U.S.A.). Foetal bovine serum (FBS) was ordered from HyClone (Logan, UT, U.S.A.). L-Glutamine and penicillin/streptomycins solutions were obtained from GIBCO BRL or Fisher (Pittsburgh, PA, U.S.A.). Geneticin (G418) was ordered from Sigma (St. Louis, MO, U.S.A.) or Fisher (Pittsburgh, PA, U.S.A.).

Acetylcholine, oxotremorine, oxotremorine-M, arecaidine propagyl ester hydrobromide (APE), and pirenzepine (Figure 2) were purchased from Research Biochemicals Intl. (RBI) (Natick, MA, U.S.A.). Carbachol, N-methyl-(−)-scopolamine, l-hyoscyamine, lithium chloride, 5′-guanylyl imidodiphosphate (GppNHp) were ordered from Sigma. Poly(ethylenimine) was from Aldrich (Milwaukee, WI, U.S.A.). Both myo-³H-inositol and ³H-(R)-quinuclidinyl benzilate (QNB) were purchased from Du Pont-New England Nuclear (Boston, MA, U.S.A.). All other inorganic chemicals were from Fisher.

Figure 2.

NMS (left) and pirenzepine (right) inhibition binding profiles. Radioligand inhibition binding assays were carried out using membrane homogenates from transfected A9 L cells in the presence of 0.1 nM of ³H-(R)-QNB for HM₁(WT) and HM₁(Thr192Ala) receptors and 1 nM of ³H-(R)-QNB for HM₁(Asn382Ala) receptors. Assays were performed in triplicate sets. Data represent the means±s.e.mean from a minimum of three assays (Table 3).

The QuickChange site-directed mutagenesis kit was purchased from Stratagene (La Jolla, CA, U.S.A.). The Super Separator-24 and T7 Sequenase PCR product sequencing kit were obtained from Amersham (Arlington Heights, IL, U.S.A.). The QIAprep Spin Plasmid Mini- and the QIAGEN Plasmid Maxi-Kits were bought from QIAGEN (Chatsworth, CA, U.S.A). The LIPOFECTIN Reagent was obtained from GIBCO BRL. The SEP-PAK anion exchange cartridges were purchased from Waters (Franklin, MA, U.S.A.). UniverSol ES scintillation cocktail was ordered from ICN Biomedicals (Irvine, CA, U.S.A.). FP-100 Whatman GF/B filters were obtained from Brandel (Gaithersburg, MD, U.S.A.).

Site-directed mutagenesis and stable expression of HM₁(WT) and mutant receptors

Replacement of Thr192Ala in TM V or Asn382Ala in TM VI was carried out using the QuickChange Kit from Stratagene. For the Thr192 to Ala mutation, the sense primer was 5′-ATCACCTTTGGCGCCGCCATGGCTGCC-3′ with changed bases in bold, and the antisense primer was 5′-GGCAGCCATGGCGGCGCCAAAGGTGAT-3′. A unique restriction site (KasI) was incorporated for convenient screening. For the Asn382 to Ala mutation, the sense primer was 5′-GACACCGTACGCCATCATGGTGCTG-3′ with mutated bases in bold, the antisense primer was 5′-GCACCATGATGGCGTACGGTGTCCAG-3. A unique restriction site (BsiWI) was incorporated for convenient screening. The mutations were confirmed by unique restriction digestion and dideoxy nucleotide sequencing. A9 L or CHO cells were cotransfected with HM₁pcD1 [or HM₁(Thr192Ala)-pcD1 or HM₁(Asn382Ala)pcD1] and pcDneo (Jürgen Wess, NIDDK) at a ratio of about 10 : 1 using LIPOFECTIN Reagents from GIBCO following the instructions provided by the manufacturer. Selections were carried out in DMEM (supplemented with 10% FBS, 4 mM L-Glutamine, 50 u ml⁻¹ Penicillin, and 50 μg ml⁻¹ Streptomycin) containing 600–800 μg ml⁻¹ Geneticin. Surviving cells were subcultured for functional tests and radioligand binding assays. The creation of stable A9 L cell lines expressing HM₁(WT) was described previously (Huang et al., 1998).

PI turnover assays

Cells stably expressing wild-type or mutant receptors were seeded into 24-well tissue culture plates and incubated with ³H-inositol (1 μCi per well) for 48 h to about 80–90% confluence. The cells were washed twice with DMEM and incubated with 450 μl DMEM containing 10 mM LiCl for 30 min. Then 50 μl of different concentrations of test ligands were added in duplicate sets. The plates were incubated for 30 additional mins, and the incubation was terminated by removal of the mixture and addition of 0.75 ml of 5% TCA. The procedures for PI assays conducted in Krebs-Henseleit (KH) buffer were described (Huang et al., 1998). The radioactive inositol phosphates in lysate were isolated using SEP-PAK cartridge methods (Wreggett & Irvine, 1987) with minor modifications (Hoss et al., 1990). Briefly, the 5% TCA extracts were transferred to balanced SEP-PAK anion exchange cartridges (formate form). The cartridges were then washed with 6 ml distilled water and 10 ml of 5 mM disodium tetraborate. The radioactive ³H-inositol phosphates were eluted from the cartridges by 1 ml of 0.6 M ammonium formate/0.06 M formic acid/5 mM disodium tetraborate (pH 4.75). Then 0.5 ml of eluate was counted in 6 ml of UniverSol ES scintillation cocktail on a 6895 BetaTrac Liquid Scintillation counter. The growth DMEM containing 10 mM LiCl served for determination of basal levels, and activities were presented as the percentage stimulation above basal levels.

Radioligand binding assays

Membrane homogenates were prepared from A9 L or CHO cells stably expressing wild-type or mutant receptors according to procedures reported previously (Dörje et al., 1991; Huang et al., 1998). Protein concentrations were determined by a modified Lowry method (Lowry et al., 1951; Markwell et al., 1981). Membrane homogenates were aliquoted and stored at −70°C. All binding assays were conducted in triplicate sets with a final volume of 1 ml. Eight different concentrations of ³H-(R)-QNB were used in radioligand saturation binding assays; while 14 different concentrations of test ligand were used in ligand inhibition binding assays. Total binding and nonspecific binding activities were determined in the absence and presence of 1000 fold excess of cold (R)-QNB, respectively. Reactions started with the addition of membrane proteins to mixtures and were incubated at room temperature for 2 h in the binding buffer (25 mM sodium phosphate, pH 7.4, containing 5 mM magnesium chloride). G-protein coupling properties were examined in the presence of 100 μM GppNHp. The incubations were terminated by addition of 5 ml ice-cold binding buffer and rapid transfer to Whatman GF/B filters which were soaked with cold binding buffer containing 0.3% poly(ethylenimine) immediately before filtration.

Molecular modelling studies

The M₁ receptor model accepted here was proposed by Nordvall & Hacksell (1993). Previously we used a simplified approach (Messer et al., 1997) by [cutting-out] nine amino acids, including Asp105 (TM III), Thr192 (TM V) and Asn382 (TM VI), which were considered to form a cavity for the agonist binding. In contrast, the full M₁ receptor model was utilized here. AMBER charges (Weiner et al., 1986) were added to the atoms, and the overall charge of the receptor model was set to zero by neutralizing some side chains on the molecular surface. Asp105 (TM III) in binding cavity bore a unit negative charge.

As described previously in the study with the 9-amino-acid model, molecular modelling for the docking of a ligand within the cavity was performed by using the DOCK procedure of the Sybyl 6.1 modelling package (Tripos, Inc.). The procedure minimizes the energy of the dimer formed of the receptor model and the ligand, by changing their relative positions and allowing geometric changes for both sites. The energy was calculated by using the Tripos molecular mechanics force-field. The minimum energy conformation of each ligand optimized previously was docked to the binding site. Atomic charges for the ligand were obtained from AM1 quantum-chemical calculations (Dewar et al., 1985) as implemented in Sybyl.

Energy terms for the dimer, E_D were broken down as follows:

where E_L and E_R stand for the stretching, bending, torsional, intramolecular electrostatic or van der Waals energy terms for the ligand and the receptor, respectively, at the geometry taken in the dimer. E_LR stands for the ligand-receptor intermolecular energy terms. Applying equation 1, change of a specific term, dE(x), upon dimerization is given by

where [opt] terms refer to the [x] energy type in the optimized ligand and receptor. The intermolecular term, E_LR differs from zero regarding only electrostatic (x=e) and van der Waals (x=W) energies (Table 6). E_Ropt was taken from minimization of the pure receptor model. Because of the complexity of its energy surface the found minimum corresponds most likely to a local one. Accordingly, the (E_R−E_Ropt(x)) terms differ by a constant from their correct values. The differences of these terms, Δ (E_R−E_Ropt (x)), however, are correct with different ligands bound to the M₁ receptor model.

Table 6.

Energy terms for the ligand-receptor interactions at HM₁ receptors

Data analysis

Nonlinear least squares curve-fitting was performed using DeltaGraph® version 4.0.1 for the Macintosh (DeltaPoint Inc. 1997). Dose-response data from functional assays were fitted into a one-site stimulation function to obtain S_max and EC₅₀ values. ³H-(R)-QNB saturation binding data were fitted into a one-site binding model for determining B_max and K_d values. Ligand inhibition binding data were fitted into one-site, two-site and three-site binding models. Statistical comparisons were carried out using an F test with P set at the 0.05 level. K_i values were converted from IC₅₀ values according to the Cheng & Prusoff (1973) formula.

Results

Receptor expression and antagonist binding

As indicated in Table 2, HM₁(WT) receptors were expressed at relatively higher levels (3.5 fold) in CHO cells than in A9 L cells. HM₁(Thr192Ala) receptors were expressed at higher levels (2.4 fold) in A9 L cells than in CHO cells. HM₁(Asn382Ala) receptors were generally expressed at relatively low levels both in A9 L cells and CHO cells. When expressed in either A9 L or CHO cells, HM₁(Thr192Ala) and HM₁(WT) receptors exhibited comparable binding affinities for ³H-(R)-QNB, while HM₁(Asn382Ala) mutant receptors showed a marked reduction in binding affinity for ³H-(R)-QNB. Generally, the Asn382Ala substitution produced much greater reductions in antagonist binding affinities than the Thr234Ala mutation (Table 3). Much greater reductions in binding affinities also were observed for structurally different muscarinic antagonists, such as NMS, trihexyphenidyl, and pirenzepine, at HM₁(Asn-382Ala) receptors expressed in A9 L cells (Figure 2). HM₁(Thr192Ala) receptors generally displayed similar binding profiles for these antagonists as HM₁(WT) receptors (Figure 2).

Table 2.

Stable expression of HM₁(WT) and two mutant receptors, HM₁(Thr192Ala) and HM₁(Asn382Ala), in A9 L and CHO cells

Table 3.

Muscarinic antagonist inhibition binding properties of HM₁(WT), HM₁(Thr192Ala) and HM₁(Asn382Ala) receptors, expressed in A9 L cells

Receptor functional assays

Since A9 L cells provide a suitable system for determining muscarinic agonist activities and G-protein coupling interactions (Brann et al., 1987; Messer et al., 1997), mutant receptors were compared to HM₁(WT) receptors in A9 L cells for their abilities to mediate phosphatidylinositol (PI) metabolism (Figure 3 and Table 4). Generally, substitution of Thr192 or Asn382 with Ala resulted in different degrees of reduced efficacies and/or potencies for muscarinic agonists. Carbachol showed about 50 and 80% of HM₁(WT) activity at HM₁(Thr192Ala) and HM₁(Asn382Ala), respectively, with significant reductions in potency. The Thr192Ala mutation reduced efficacy and potency, while the Asn382Ala mutation decreased the potency of oxotremorine-M by 20 fold. The tetrahydropyrimidine ester derivative CDD-0034 displayed full agonist activity at HM₁(WT) but showed only 40% efficacies at both mutant receptors, together with significantly reduced potencies. The tetrahydropyrimidine oxadiazole derivative CDD-0102 exhibited decreased efficacies yet increased potencies at both mutant receptors as compared with HM₁(WT) receptors.

Figure 3.

Maximum response of PI hydrolysis mediated by a series of muscarinic ligands at HM₁(WT), HM₁(Thr192Ala), and HM₁(Asn-382Ala) receptors stably expressed in A9 L cells. Receptor expression levels are indicated in Table 2. PI assays were carried out in growth media (DMEM) containing 10% FBS or in KH buffer as described in the Methods section. Data are presented as the percentage relative to the maximal carbachol activity at HM₁(WT) receptors (250±46% above basal from five assays).

Table 4.

Potencies of tested ligands at HM1(WT), HM1(Thr192Ala) and HM1(Asn382Ala) receptors expressed in A9 L cells

The muscarinic natural agonist ACh originally did not show significant stimulation at HM₁(WT) receptors. We then tested ACh and carbachol activities at HM₁(WT) and HM₁(Asn-382Ala) receptors using Krebs-Henseleit (KH) buffer as the incubation media. As indicated in Table 4, ACh exhibited a 10 fold reduction in potency yet with only a slightly lower maximum response at HM₁(Asn382Ala) than at HM₁(WT) receptors.

Interestingly, arecoline, a partial agonist at HM₁(WT) receptors, did not display any agonist activity at HM₁(Asn-382Ala) receptors, although it had almost the same binding affinity to HM₁(Asn382Ala) and HM₁(WT) receptors (see Table 5). To determine if arecoline functioned as an antagonist at HM₁(Asn382Ala) receptors, we tested PI hydrolysis stimulated by ACh in the presence of arecoline. ACh displayed unchanged maximal responses in the presence of 1, 3 and 10 μM of arecoline yet showed significant reduction of potency by 13.3, 18.6 and 22.4 fold, respectively. A Schild plot analysis (Figure 4B) yielded a pA₂ value of 2.8 for arecoline with Schild slope of 0.28.

Table 5.

Inhibition binding properties of muscarinic agonists at HM1(WT) and two mutant receptors expressed in A9 L cells

Figure 4.

Pharmacology of HM₁(WT) and two mutant receptors stably expressed in A9 L cells. Data (Table 4 and Figure 3) are fitted into a one-site stimulation model. (A) carbachol dose response profiles. PI hydrolysis assays were conducted in growth media (DMEM) containing 10% FBS. (B) acetylcholine dose response at HM₁(WT) receptors and antagonist activity of arecoline at HM₁(Asn382Ala) receptors. PI hydrolysis assays were conducted KH buffer. Insert is a Schild plot analysis of acetylcholine-mediated PI hydrolysis in the presence of 1, 3 and 10 μM of arecoline at HM₁(Asn382Ala) receptors. DR stands for the ratio of EC₅₀ values of acetylcholine in the presence of arecoline over that in the absence of arecoline.

Agonist inhibition binding properties

Muscarinic agonists used in functional assays were also tested in inhibition binding assays to determine the effects of mutations on agonist binding interactions (Table 5 and Figures 5 and 6). ACh distinguished three binding sites, while carbachol and oxotremorine-M exhibited two binding sites at HM₁(WT) receptors. These agonists all interacted with two binding sites at the two mutant receptors. Tetrahydropyrimidine derivatives, such as CDD-0034 and CDD-0102, showed two-site binding profiles at HM₁(Asn382Ala) receptors, CDD-0098 showed two-site binding profiles at both HM₁(WT) and HM₁(Thr192Ala) receptors. Mutation of Thr192 or Asn382 to Ala resulted in marked decreases in the binding affinity of ACh, especially in GppNHp-sensitive high affinity (K_H) binding, by 70 to 130 fold. Only small reductions (<2.1 fold) in the lowest affinity binding (K_L) site were observed. Carbachol exhibited a greater reduction of K_L than K_H at HM₁(Thr192Ala) mutant receptors, while showing comparable degrees of reduction in both K_H and K_L at HM₁(Asn382Ala) mutant receptors (Table 5). The binding of the structurally-related oxotremorine and oxotremorine-M was more affected by the Asn382Ala mutation than the Thr192Ala mutation. Binding of the structurally related tetrahydropyridine derivatives, arecoline and arecaidine propagyl ester (APE), was not affected significantly by either mutation. All these results indicate that muscarinic agonists with different chemical structures may bind to HM₁(WT) and/or mutant receptors in different modes (geometric positions of the ligand relative to amino acids in the putative binding pocket).Figure 1

Figure 5.

Acetylcholine inhibition binding properties in the absence and presence of 100 μM GppNHp. Radioligand binding experiments were carried out on membrane homogenates in the presence of 0.1 nM of ³H-(R)-QNB for HM₁(WT) and HM₁(Thr192Ala) receptors and 1 nM of ³H-(R)-QNB for HM₁(Asn382Ala) receptors. Data represent the means±s.e.mean from a minimum of three assays (Table 7).

Figure 6.

Carbachol inhibition binding properties in the presence of 0.1 nM of ³H-(R)-QNB for HM₁(WT) and HM₁(Thr192Ala) receptors and 1  nM of 3H-(R)-QNB for HM₁(Asn382Ala) receptors. Assays were conducted on membrane homogenates. Data represent the means±s.e.mean from a minimum of three assays (Table 7).

Figure 1.

Chemical structures of ligands used in this study.

In order to determine if the high-affinity agonist binding was sensitive to guanine nucleotides, ACh binding was assessed in the presence of 100 μM GppNHp. As indicated in Figure 5, the high affinity binding sites at HM₁(WT) and the two mutant receptors were abolished by the addition of GppNHp, indicating the presence of GTP-sensitive coupling between the mutant receptors and G proteins as in HM₁(WT) receptors. In the presence of 100 μM GppNHp, ACh still displayed a 17 fold reduction in binding affinity at HM₁(Thr192Ala) receptors and 9 fold at HM₁(Asn382Ala) receptors.

Molecular modelling studies

In order to understand the binding interactions between receptors and agonists, molecular modelling studies were carried out by docking agonists carbachol and two tetrahydropyrimidine derivatives, CDD-0034 and CDD-0098, into the putative binding pocket containing the critical residues Asp105 (TM III), Thr192 (TM V), and Asn382 (TM VI) of the HM₁ receptors (Figure 7). The binding energies, EB, were obtained by calculating the minimized energy for the dimer and the monomers. Resulting electrostatic and van der Waals interaction energies between the ligand and receptors are summarized in Tables 6 and 7. Mutation of Thr or Asn to Ala resulted in significantly more negative binding energies, indicating stronger ligand binding. Van der Waals interactions, dE(W), appear to provide the dominant stabilizing contributions to the binding of a ligand by both HM₁(WT) and mutant receptors. In HM₁(WT) receptors, Thr192 had consistent hydrogen-bond interactions with carbachol and the two tetrahydropyrimidine derivatives. As compared with carbachol and CDD-0034, CDD-0098 exhibited two additional strong hydrogen-bond interactions with Asp105 and Asn382 residues. This interaction profile is in good agreement with the fact that CDD-0098 has the highest binding affinity among the three agonists (Table 5).

Figure 7.

Docking of CDD-0098 to a model of the M₁ muscarinic receptor (Nordvall & Hacksell, 1993). The backbone of the seven transmembrane domains are shown in white. Key amino acid residues Asp105 (left), Thr192 (right), and Asn382 (top) are coded by atom type (carbon, white; nitrogen, dark blue; oxygen, red; hydrogen, light blue). Colour code for CDD-0098: yellow for the wild-type M₁ receptor, purple for the Thr192Ala mutant receptor, and green for the Asn382Ala mutant receptor.

Table 7.

Hydrogen bond patterns for ligand-receptor interactions at HM₁ wild-type and mutant receptors

Hydrogen-bond patterns for ligand-receptor interactions were not significantly affected in the carbachol binding mode by mutation of Thr192 or Asn382 to Ala. These mutations did alter the binding patterns for CDD-0034 and CDD-0098 (Table 7). At HM₁(Thr192Ala) receptors, CDD-0034 lost a hydrogen-bond interaction due to the Thr192 to Ala mutation but gained a strong hydrogen-bond interaction with the Asp105 residue. CDD-0098 lost two hydrogen-bond interactions due to Thr192 to Ala mutation and also lost a strong hydrogen-bond interaction with the Asp105 residue. The different interactions with the Asp105 residue probably result from different binding modes for different agonists.

Discussion

Previous site-directed mutagenesis and pharmacological studies (Wess et al., 1991; 1992; Blüml et al., 1994) confirmed the role of Thr234 (TM V) but not of Asn507 (TM VI) (number in RM₃ subtype) in agonist binding and receptor activation as predicted by three dimensional modelling studies (Trumpp-Kallmeyer et al., 1992; Ward et al., 1992; Nordvall & Hacksell, 1993; 1995) based on the human M₁ receptor subtype. In this study, the roles of the equivalent Thr and Asn in HM₁ receptors in ligand binding and receptor activation were re-examined by stably expressing mutant HM₁ receptors in A9 L cells. Thr192 or Asn382 was replaced by Ala, which has no functional group in its side chain for potential H-bond interactions. A series of muscarinic ligands were characterized at the mutant receptors and HM₁(WT) receptors.

Mutation of Thr192 to Ala did not dramatically change the binding affinities for tested antagonists belonging to different structural classes, indicating that Thr192 may not be involved in antagonist binding. In contrast, substitution of Asn382 by Ala resulted in much greater reductions in binding affinities for pirenzepine, trihexyphenidyl, and NMS than other antagonists, such as QNB. These data are consistent with previous results on the equivalent mutations in RM₃ receptors (Blüml et al., 1994). Therefore, Asn382 played a critical role in the binding of some antagonists, and Thr192 and Asn382 residues in HM₁ receptors appear to have essentially the same functions in antagonist binding as the equivalent residues (Thr234 and Asn507 respectively) in RM₃ receptors.

The selection of ³H-(R)-QNB as radioligand in binding assays appeared to be critical in characterizing the mutant receptors. It would be difficult to use ³H-NMS as the radioactive ligand in assessing receptor binding, since NMS exhibited a 2300 fold reduction in binding affinity at HM₁(Asn382Ala) receptors. This may account for previous failures to detect expression of the same mutant receptors in COS-7 cells (Bourdon et al., 1997; Ward & Hulme, 1997Ward & Hulme, 1997).

We were unable to demonstrate dose response curves for the endogenous muscarinic agonist ACh in growth media, possibly due to the presence of cholinesterases in animal serum as we and others (Spalding et al., 1995) found previously. In KH buffer, we were able to compare ACh activity at both HM₁(WT) and HM₁(Asn382Ala) receptors and show that substitution of Asn382 with Ala resulted in 10 fold reduction of ACh potency. Interestingly, arecoline, a partial agonist at the HM₁(WT) receptor, can behave as an antagonist at HM₁(Asn382Ala) receptors, by inhibiting ACh stimulation of PI hydrolysis. A Schild plotting analysis indicated that the inhibition may not be competitive.

Substitution of Thr192 or Asn382 with Ala resulted in a greater reduction of binding affinity for classical full agonists (such as ACh, carbachol, oxotremorine-M) than for tetrahydropyridine or tetrahydropyrimidine derivatives (such as arecoline, APE and CDD compounds). The Thr192Ala mutation had a greater effect than the Asn382Ala mutation on reducing binding affinity for acetylcholine and carbachol, but less than the Asn382Ala mutation on reducing binding affinity for oxotremorine-M and oxotremorine. These data indicated that both Thr192 and Asn382 are important for agonist binding.

The natural agonist ACh showed a three-site binding profile to HM₁(WT) receptors, but two binding sites at the mutant receptors. In addition, ACh exhibited much greater reductions in binding affinities at both mutant receptors than any other tested agonists. These data indicate that the mutations had much more significant effects on the binding of the natural ligand than any other classical agonist. Thus, characterization of mutant receptors with the endogenous agonist is valuable.

The greatest reductions of ACh binding at Thr192Ala mutation (71 fold) and Asn382Ala mutation (130 fold) were associated with the high affinity binding sites. This high affinity binding was found to be GTP sensitive and therefore involved in G protein interactions and receptor activation. In contrast, no significant reduction (1.7–2.1 fold) was observed for the low affinity binding sites. The magnitude of the ACh affinity reduction was much lower in the presence of GppNHp than in the absence of GppNHp. This suggests that both Thr192 and Asn382 residues are important for the M₁-receptor/ACh complex to interact with G-proteins, indicating that Thr192 and Asn382 are involved in high affinity agonist binding and the receptor activation processes.

Activity differences between M₂ full and partial agonists in mediating the inhibition of adenylyl cyclase were due to reduced affinity of the receptor/agonist complex for G_i, but not the abilities of the receptor/agonist complex to activate G_i (Tota et al., 1990). It is possible that mutations in HM₁ receptors may not change receptor binding affinity for agonists but reduce the binding affinity of the mutant receptor/agonist complex for G-proteins and/or decrease the ability of the complex to activate G-proteins.

Mutation of Thr192Ala or Asn382Ala reduced receptor response to agonists. For classical full agonists, such as ACh and carbachol, reduced efficacies and/or potencies were in general agreement with their reduced binding affinities at mutant receptors. However, there was a discrepancy between unchanged or increased agonist binding affinities and reduced agonist activities. Several ligands, such as arecoline and CDD-0034, which showed partial and full agonist activities respectively at HM₁(WT) receptors, exhibited reduced activities at the mutant receptors, with similar binding affinities for HM₁(WT) and the mutant receptors. CDD-0102 displayed two binding affinity states and higher binding affinity at HM₁(Asn382Ala) receptors than HM₁(WT) receptors. Its efficacy at HM₁(Asn382Ala) receptors was only half of that at HM₁(WT) receptors.

HM₁(Thr192Ala) and HM₁(Asn382Ala) receptors are not the only mutant muscarinic receptors that exhibit increased agonist binding affinity and decreased agonist activity. Similar discrepancies were shown previously at RM₁(Asp71Asn) and RM₁(Asp122Asn) (Fraser et al., 1989), and RM₃(Pro540Ala) receptors (Wess et al., 1993). All these three amino acids are highly conserved in almost all G protein coupled receptors and are believed to be involved in receptor activation by mediating agonist-induced receptor conformational changes (Fraser et al., 1989; Wess et al., 1993). Our findings support the idea that both Thr192 and Asn382 residues are involved in receptor activation processes.

Molecular modelling studies with carbachol and two tetrahydropyrimidine derivatives, CDD-0034 and CDD-0098, indicated that substitution of Thr192 or Asn382 with Ala removed potential H-bond interactions and changed the agonist binding mode. The changes in binding mode may be energetically beneficial for agonist binding but not effective for receptor activation and G-protein coupling. This is reflected by the discrepancy between reduced efficacies and/or potencies and unchanged binding affinities of muscarinic agonists, such as arecoline.

In conclusion, we have shown here that both Thr192 and Asn382, which are highly conserved within the muscarinic receptor family, are involved in agonist binding and receptor activation processes. Our results are consistent with previous molecular modelling studies. In addition, Asn382 but not Thr192 participated in antagonist binding, presumably by providing H-bond interactions with the ester portion of antagonists (Bourdon et al., 1997). Molecular modelling studies suggest that changes in binding mode may contribute to the discrepancies between agonist binding and receptor activity. Site-directed mutagenesis and molecular modelling are an excellent combination in characterizing functionally important amino acid residues in ligand binding and receptor activation. Identification of highly conserved amino acids important in ligand binding and receptor activation processes could aid in the design and development of novel therapeutic drugs for Alzheimer]s disease and other neurological disorders.

Abbreviations

ACh

acetylcholine

APE

arecaidine propagyl ester hydrobromide

4-DAMP

4-diphenylacetoxy-N-methylpiperidine methiodide

DMEM

Dulbecco's Modified Eagle Media

EB

binding energy

FBS

foetal bovine serum

GppNHp

guanylylimidodiphosphate

p-F-HHSiD

p-fluoro-hexahydro-sila-difenidol hydrochloride

HM₁

human muscarinic acetylcholine receptor subtype 1

HM₁(Asn382Ala)

HM₁ mutant receptor with the mutation of Asn382 to Ala

HM₁(Thr192Ala)

HM₁ mutant receptor with the mutation of Thr192 to Ala

KH buffer

Krebs-Henseleit buffer

NMS

N-methyl-scopolamine

PI

phosphatidylinositol

QNB

3-quinuclidinyl benzilate

TM

Transmembrane

WT

wild-type.