Introduction

Top Abstract Introduction Procedures Results Discussion References

The molecular mechanisms by which agonists bind to and activate G protein-coupled serotonin (5-HT) receptors remain major enigmas for modern pharmacologists. Site-directed mutagenesis studies have highlighted the importance of highly conserved aspartic acid residues (Wang et al., 1993; Sealfon et al., 1995), serines (Johnson et al., 1994) and phenylalanines (Choudhary et al., 1993, 1995) for agonist binding and efficacy at various 5-HT receptors, including 5-HT_1A and 5-HT_2A subtypes. The aspartic acid group (from helix III) is thought to anchor the charged nitrogen from 5-HT, whereas serines (in helix V) and a phenylalanine (in helix VI) may anchor hydroxyl/amine residues and aromatic groups, respectively (Wang et al., 1993; Choudhary et al., 1993; Johnson et al., 1994). These studies are in general accord with several computer-generated three-dimensional models of 5-HT-receptor interactions (Westkaemper and Glennon, 1993; Kristiansen and Dahl, 1996).

In some instances, point mutations of these highly conserved residues also diminish the ability of agonists to activate second messenger production. In the case of the 5-HT_2A receptor, which activates PI hydrolysis (Conn and Sanders Bush, 1984; Roth et al., 1984), a diminished ability to augment PI hydrolysis has been reported with mutations at aspartic acids found in helix II (Asp-120) (Wang et al., 1993) and helix III (Asp-155) (Wang et al., 1993) and a phenylalanine found in helix VI (Phe-340) (Choudhary et al., 1993; Roth et al., 1995). How these mutations, which alter ligand affinity, also affect agonist efficacy is unknown. An assumption of many prior studies has been that high-affinity agonist binding is essential for second messenger production.

In this paper, we directly test the hypothesis that high-affinity agonist binding is essential for second messenger production, using native and mutant 5-HT_2A receptors. Our findings suggest that the mere presence of the high-affinity agonist state of the receptor is not sufficient for receptor-effector coupling. Our findings are consistent with recent models that predict that additional transition states are essential for agonist-induced activation of second messenger production (e.g., modified ternary complex model) (Lefkowitz et al., 1993).

	Experimental Procedures

Top Abstract Introduction Procedures Results Discussion References

Materials. Tissue culture reagents were from GIBCO/BRL (Gaithersburg, MD). [³H]Ketanserin (89 Ci/mmol) was from New England Nuclear (Boston, MA). Molecular biology reagents were from Stratagene (Torry Pines, CA) or United States Biochemicals (Cleveland, OH). The rat 5-HT_2A receptor cDNA was a gift from D. Julius (University of California, San Francisco); the F340L mutation has been previously described (Choudhary et al., 1995). Other chemicals were of the highest grade commercially available. Quipazine, DMT, DOM, bufotenine, psilocybin and 5-OMe-DMT were supplied by the Mental Health Clinical Research Center, Case Western Reserve University Medical School; DOB and -methyltryptamine were from Richard Glennon (Medical College of Virginia). 5-HT creatinine sulfate was from Sigma Chemical Co. (St. Louis, MO), whereas DOI was from Research Biochemicals Inc.

Cell lines. All stable cell lines were grown in DMEM containing 10% fetal calf serum and 600 µg/ml G418. An NIH/3T3 cell line expressing the 5-HT_2A receptor (GF-6) was from D. Julius (University of California, San Francisco) and was used to isolate a clonal cell line (GF-62) that expressed 5 to 7 pmol/mg protein of the 5-HT_2A receptor. For the F340L mutation, the F340L cDNA was subcloned into the eukaryotic expression vector pRc/CMV (Invitrogen) using NotI adaptors and was transfected into 3T3 and human embryonic kidney 293 cells using Lipofectin (GIBCO/BRL), exactly as described by the manufacturer. At 72 hr after transfection, the cells were split and grown in DMEM containing 600 µg/ml G418 and 10% fetal calf serum. Surviving clones were isolated and expanded for assessment of F340L receptor expression, as assessed by radioligand binding. One cell line expressing high levels of receptors in 3T3 cells (5-7 pmol/mg; M1C15) was used for subsequent experiments. Both cell lines (GF-62; M1C15) have been previously characterized (Roth et al., 1995; Berry et al., 1996).

Binding assays. Binding assays were performed in total volumes of 0.5 ml (for ³H-radioligands) at 25°C for 90 min with 5 to 20 µg of membrane protein, as described previously (Roth et al., 1987, 1995), in 50 mM Tris-HCl buffer containing 0.5 mM EDTA, 10 mM MgCl₂, 0.05% ascorbic acid and 10 µM pargyline (pH 7.40). Membranes were harvested with a Brandel SM-24 cell harvester, followed by three ice-cold washes onto polyethyleneimine-pretreated (0.1%) glass fiber filters. Filters were soaked for 18 hr in scintillation fluid before counting, with efficiency determined by the external standard method. Specific binding (determined with 10 µM mianserin) represented 90 to 97% of total binding in the experiments reported here; no more than 10% of total counts/assay tube was bound.

PI hydrolysis. Cells were harvested by trypsinization and split into 24-well plates with complete medium. Cells were washed with inositol-free DMEM 24 hr later and then incubated for an additional 18 hr with inositol-free DMEM containing 1 µCi/ml [³H]inositol and 10% dialyzed fetal calf serum. Cells were then rinsed three times with a Krebs-bicarbonate buffer of the following composition: 118 mM NaCl, 4.7 mM KCl, 1.2 mM CaCl₂, 1.2 mM MgCl₂, 25 mM NaHCO₃ and 11 mM glucose. Before use the buffer was equilibrated with 95% O₂/5% CO₂. Cells were then incubated for 30 min with test agents together with 10 mM LiCl in Krebs-bicarbonate buffer. The reaction was then terminated by aspiration and the addition of 1.2 ml of methanol/water/HCl (25:25:0.1). Cells were then harvested into glass tubes and 0.6 ml of chloroform was added, followed by vigorous vortex-mixing. After phase separation, the upper aqueous phase was removed and [³H]inositol monophosphate was isolated and quantified as previously detailed (Roth et al., 1984). Typically a 20- to 40-fold activation of [³H]inositol monophosphate accumulation was measured for both native and mutant receptors.

Data analysis. Binding data were analyzed using a weighted, nonlinear, least-squares program that determines binding to multiple sites using the law of mass action (LIGAND program), as previously detailed (Munson and Rodbard, 1980). For PI hydrolysis experiments, data were fit to a modified Michaelis-Menton equation (Roth et al., 1995) using a nonlinear, least-squares, curve-fitting routine available in Sigma Plot. In all figures, lines represent the theoretical fits using the parameter estimates calculated by the curve-fitting programs. Protein was determined using a kit from Bio-Rad (Richmond, CA).

	Results

Top Abstract Introduction Procedures Results Discussion References

A point mutation alters the affinity of selected agonists at 5-HT_2A receptors. In previous studies, we found that a mutation of a highly conserved phenylalanine (F340L), but not an adjacent conserved phenylalanine (F339L), drastically altered the affinities of several agonists and ergolines at the 5-HT_2A receptor (Choudhary et al., 1993). We next determined whether other agonists were similarly affected by the F340L mutation. A complication encountered in investigating agonist efficacy and affinity at 5-HT_2A receptors, however, is the observation that high- and low-affinity states of the receptor exist. From our initial data (Choudhary et al., 1993), it was unclear whether the high- or low-affinity agonist state(s) might be altered by the F340L mutation. Thus, we first needed to characterize agonist affinity at the high- and low-affinity states of the 5-HT_2A receptor.

View larger version (19K): [in this window] [in a new window]   Fig. 1.   Competition binding isotherms for bufotenine at native and F340L mutant receptors. Inhibition by bufotenine of [3H]ketanserin-labeled native (top) and F340L mutant (bottom) receptors was measured and then sequentially fit to one-site (dashed line) and two-site (solid line) models. A two-site model significantly improved the fit (P < .01; F test) for both native and F340L mutant receptors.

View larger version (15K): [in this window] [in a new window]   Fig. 2.   Competition binding isotherms for quipazine at native and F340L mutant receptors. Inhibition by quipazine of [3H]ketanserin-labeled native (top) and F340L mutant (bottom) receptors was measured and then fit to a one-site model. No improvement of fit was seen with a two-site model (P > .05; F test).

A point mutation alters the ability of selected agonists to augment PI hydrolysis. As can be seen in figure 3 and table 2, all tested compounds behaved as agonists at the cloned 5-HT_2A receptor. Table 2 lists the K_act and V_max values (relative to 5-HT for the native receptor or DOI for the F340L mutant) for these compounds. As can be seen from table 2, a range of intrinsic activities was seen, with some compounds behaving as full agonists (DOB, DOI, DOM, bufotenine, -methyltryptamine, quipazine and 5-HT) and others behaving as partial agonists (DMT and 5-OMe-DMT).

View larger version (20K): [in this window] [in a new window]   Fig. 3.   Effects of F340L mutation on agonist efficacy. The ability of -methyl-5-HT (), DOI () and bufotenine () to activate PI hydrolysis was evaluated with native (top) and F340L mutant (bottom) receptors stably expressed in NIH/3T3 cells. Data represent mean ± S.E.M. of a typical experiment, which has been replicated at least three times for each compound. IP, inositol phosphates.

High- and low-affinity binding is not correlated with second messenger production. We next investigated whether a correlation might exist between the high-affinity agonist state of the receptor and second messenger production. Figures 4 and 5 show the results of these studies. The percentage of high-affinity sites varied, depending on the agonist used, from 25% (5-OMe-DMT) to 80% (DOI). There was no correlation between the percentage of high-affinity sites and the ability of selected agonists to activate PI hydrolysis with the native receptors (fig. 4).

View larger version (19K): [in this window] [in a new window]   Fig. 4.   Relationship between the relative number of high-affinity sites and efficacy at native and F340L mutant receptors. Shown is the relationship between the mean percentage of high-affinity sites and the percentage relative efficacy for native () and F340L mutant () receptors. For the native receptors, 5-HT was used as 100% relative efficacy; for the F340L mutant, DOI was used as 100% relative efficacy. The relationship was statistically significant (P > .05). 1, DOI; 2, DOB; 3, 5-HT; 4, bufotenine; 5, DOM; 6, quipazine; 7, DMT; 8, 5-OMe-DMT; 9, 5-methoxy-5-HT.

View larger version (18K): [in this window] [in a new window]   Fig. 5.   Lack of correlation between binding affinities and potencies for agonists at 5-HT2A receptors. Top, relationship between mean Ki values for the high-affinity site and Kact values for PI hydrolysis at native receptors. Bottom, relationship between mean Ki values for the low-affinity site and Kact values for PI hydrolysis at native receptors. Neither relationship was statistically significant (P > .05 for both). 1, DOI; 2, DOB; 3, 5-HT; 4, bufotenine; 5, DOM; 6, quipazine; 7, DMT; 8, 5-OMe-DMT; 9, 5-methoxy-5-HT.

	Discussion

Top Abstract Introduction Procedures Results Discussion References

The major finding of this paper is that the ability of an agonist to promote high-affinity agonist binding is not necessarily correlated with agonist efficacy. As demonstrated below, our findings are consistent with a modified ternary complex model of ligand-receptor-G protein interactions (Lefkowitz et al., 1993). We arrived at our conclusions by three mutually supportive lines of evidence, 1) the lack of correlation between the number of high-affinity agonist sites and the efficacy of an agonist in activating PI hydrolysis in cells expressing native and mutant receptors, 2) the preferential loss of agonist efficacy for tryptamine analogs without large alterations in the number of high-affinity agonist sites with the F340L mutation and 3) the maintenance of phenylisopropylamine efficacy despite the reduction in the affinity and number of high-affinity agonist sites with the F340L mutation. Each of these points is discussed, and our results are then interpreted with the aid of a modified ternary complex model of receptor-G protein activation (Lefkowitz et al., 1993).

Our first observation was that there was no clear correlation between the number of high-affinity sites and the ability of an agonist to activate PI hydrolysis (K_act). In cells expressing native receptors, agonists had variable efficacies ranging from 77% to 100% of that of the full agonist 5-HT and differed in the formation of the high-affinity state of the 5-HT_2A receptor from a low of 25% to a high of 80%. A prediction of the simple ternary complex model (fig. 6) is that there should be some direct correlation between agonist efficacy and the relative number of receptors in the high-affinity state. There was no correlation between these two parameters.

View larger version (13K): [in this window] [in a new window]   Fig. 6.   Comparison of the modified ternary complex model and the conventional ternary complex model (adapted from Lefkowitz et al., 1993). The modified ternary complex model proposes the existence of an intermediate state of the receptor (R*) that is essential for G protein (G) binding and activation. Ligands (H) may interact directly with either the R, R* or R*G form of the receptor, whereas the receptor may fluctuate between R and R* conformations. The model differs from the conventional ternary complex model by the introduction of an intermediate state of the receptor (R*).

With mutant receptors, the lack of a correlation between the number of high-affinity agonist sites and K_act was even more striking. 5-HT, DOI and DOB were all able to promote similar degrees of high-affinity binding but showed quite variable degrees of second messenger production. There was no correlation between the number of high-affinity sites and the ability of drugs to activate PI hydrolysis in mutant receptors.

Our second observation was that tryptamine analogs showed a preferential loss of efficacy without a corresponding loss of the number of high-affinity agonist sites at the F340L mutation. Thus, for instance, 5-HT displayed 18% efficacy despite the fact that the relative proportions of high- and low-affinity sites were only marginally altered. Other tryptamine analogs showed even more dramatically diminished efficacies with the F340L mutation. Decreased affinity at the high-affinity site was not, however, associated with diminished efficacy. Thus, for instance, bufotenine, DMT and 5-OMe-DMT displayed nearly 100-fold lower K_i values for the high-affinity site with the F340L mutation, and all had efficacies diminished by at least 90%. On the other hand, 5-HT had only a 6-fold lower affinity for the F340L high-affinity state but displayed an 82% decrease in efficacy, relative to the native receptor.

Our third observation was that phenylisopropylamines showed diminished affinities and numbers of high-affinity sites without corresponding losses of agonist efficacy. Interestingly, the shift in K_i values for the high-affinity state (8-20-fold) was much less than the change in K_act (7,200-36,000-fold). If high-affinity binding was directly correlated with agonist efficacy, we would have expected similar shifts in affinity. Additionally, DOI had many fewer sites in the high-affinity state (80% vs. 26%) with the mutant receptor but displayed equivalent agonist efficacy, relative to the native receptor. Taken together, these three types of observations demonstrate that high-affinity agonist binding cannot be directly connected with agonist efficacy.

A similar lack of correlation was seen with quipazine. Quipazine labeled only a single site at both native and mutant receptors and had no change in apparent binding affinity with the F340L mutation, in agreement with our previous study (Roth et al., 1995). Despite the fact that no apparent change in the K_i of quipazine occurred with the F340L mutation, quipazine was inactive at the F340L mutant and fully active at the native receptor. The simple ternary complex model directly predicts that a loss of agonist efficacy would be accompanied by lower binding affinity.

Earlier studies with other receptors have implied that the high-affinity agonist "state" of the receptor is the state that is coupled to G proteins in the absence of bound GTP or GDP (fig. 6). The receptor cycles from this state to a low-affinity state depending on its coupling to G proteins and the presence of GTP or GDP. This model of receptor coupling is based on a number of mutually supportive lines of evidence (see discussion in Lefkowitz et al., 1993), including point mutations of highly conserved aspartic acid residues that abolish high-affinity agonist binding and that abolish G protein coupling as well. Studies by Hausdorff et al. (1991), however, demonstrated that a deletion mutant of the beta adrenergic receptor was able to bind agonists with high affinity but not be coupled to adenylate cyclase. Those authors concluded that, for beta adrenergic receptors, the processes of high-affinity agonist binding and G protein coupling are likely to be distinct processes and that at least one additional state of the receptor is essential for agonist efficacy. That group subsequently proposed a slightly more complicated scheme of ligand binding and G protein coupling, which has been called a modified ternary complex model (fig. 6).

Using this modified model as a template, then, how can one describe situations in which complex binding kinetics (e.g., multiple affinity states) can be disconnected from activation of G proteins? From the model it is clear that agonists (H) can interact with three separate receptor species, i.e., R, R* and R*G. Furthermore, each binding reaction is described by distinct kinetic parameters, i.e., k, K and k, respectively; only one kinetic parameter (M parameter) describes the process leading to activated G proteins. Additionally, each of these states of the ligand-receptor-G protein complex exists in multiple equilibria, of which six possible combinations are described in the most simple situation. G protein activation, according to this model, depends solely on the final entities (R*G or HR*G), which themselves are dependent on many equilibrium reactions, according to this simplified scheme. It is clear, then, that multiple equilibrium binding sites corresponding to different affinity states of the receptor (HR, HR* and potentially others) can exist separately from the activated state of the receptor, HR*G.

This model also predicts that receptors can directly interact with G proteins but that an activated receptor state (R*) is necessary for this to occur. We have recently found that 5-HT_2A receptors can interact with G_q in the absence of agonists but in the presence of magnesium, presumably via a transitional state of the receptor (E. A. Hyde and B. L. Roth, manuscript in preparation). Additionally, Casey et al. (1996) have provided preliminary findings that constitutively active 5-HT_2A receptors (R*) have higher agonist affinity than native receptors. It is likely, of course, that this simplified model will not ultimately be suitable for describing 5-HT_2A receptor-ligand-G protein interactions and that more complex kinetic schemes will be discovered that more accurately describe ligand-receptor interactions. For rhodopsin, the model G protein-coupled receptor, several transition states before receptor activation have been identified (Stewart et al., 1975; Hamdorf and Kirschfeld, 1980), so it will not be surprising if analogous processes occur with 5-HT_2A receptors and other G protein-coupled receptors.

For 5-HT_2A receptors, only a few other studies have examined the effects of mutations on high- and low-affinity states and signal transduction processes. Wang et al. (1993) demonstrated that a mutation at Asp-120 (in the rat 5-HT_2A receptor) abolished both signal transduction and the ability of the 5-HT_2A receptor to be regulated by guanine nucleotides. Similar results have been obtained for the beta adrenergic receptor (Strader et al., 1988).

Johnson et al. (1994) investigated the effects of several mutations in transmembrane region V that altered the structure-activity relationships of N¹-substituted ergolines and tryptamines. They reported that one mutation (A242V) apparently abolished the high-affinity state but did not alter the ability of 5-HT or 5-methoxytryptamine to augment PI hydrolysis. These results are in accord with our findings that mutations may alter the number of high- and low-affinity sites without affecting second messenger production.

How can this model be used to directly clarify our findings obtained with mutant receptors? According to this model, mutations may be constructed that have discrete effects on ligand affinity for the R and R* forms of the receptor (e.g., K_H and K_L, which correspond to k and k in the model), the equilibrium between these two states (e.g., percentages of high- and low-affinity states; J parameter) and/or the ability of G proteins to directly bind to the receptor (M parameter). Thus, for several tryptamines, second messenger production is abolished (R*R*G transition, governed by the M parameter) by the F340L mutation but the RR* transition is relatively unaffected (e.g., percentages of high- and low-affinity states present; J parameter). In the same manner, the data of Wang et al. (1993) can be explained by suggesting that the D120N mutation selectively abolishes the R*R*G transition (M parameter).

Likewise, for compounds like quipazine, for which only one state of the receptor is measured, interaction (according to this model) can only be with the R* form at the native receptor. An alternative explanation is that the RR* transition for quipazine is rapid, relative to ligand binding, and only one affinity state is measured. The model also predicts that DOM binds to the R* form of the F340L mutant, because only one affinity state was seen and the receptor was still coupled to PI hydrolysis. Additionally, the model predicts that quipazine interacts only with the R* form of the receptor with the F340L mutant but that the F340L mutant interferes with G protein coupling. Our assumption is that the single affinity state of quipazine corresponds to its high intrinsic efficacy, an assumption that will need to be tested in further experiments.

Testing additional predictions of this model is quite difficult, because we have no way of measuring conformational changes of receptor proteins. Kobilka and co-workers, however, were able to demonstrate such transition states with purified beta adrenergic receptors (Gether et al., 1995). Experiments currently underway in which large quantities of 5-HT_2A receptors are being purified may help to clarify these issues; we have obtained preliminary evidence that purified 5-HT_2A receptors can directly interact with purified G_q proteins but that this interaction depends on the presence of magnesium (E. A. Hyde and B. L. Roth, manuscript in preparation).

The results with the phenylisopropylamines are somewhat more difficult to explain, because each compound was differentially affected by the F340L mutation. For DOM and DOB, the F340L mutation appears to alter the R*R*G interaction, whereas for DOI the effect is primarily on ligand binding affinity (k constant), because efficacy was not affected. These results suggest that a single point mutation can have multiple effects on ligand binding and efficacy, depending on the ligand used and the assay conditions selected.

In conclusion, we demonstrate that the relationship between high-affinity agonist binding states and second messenger production is more complicated than previously suggested for 5-HT_2A receptors. Our data are in accord with models that suggest that intermediate state(s) of the receptor (R*) are formed before agonist-induced activation of PI hydrolysis and that single point mutations may independently alter multiple steps of receptor-ligand activation.