Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

The hypothesis of agonist-directed trafficking of receptor signaling (ADTRS; Kenakin, 1995) predicts that when a receptor signals through two or more independent signal transduction pathways, the relative efficacies (R.e.s) of a series of agonists may differ for the pathways. This contrasts with the classic concept that intrinsic efficacy is a solely agonist-dependent pharmacodynamic parameter (Furchgott, 1966). This hypothesis builds on the ideas that a receptor can exist in distinct states (or conformations) and that the ability of those states to activate different G protein types or subtypes may differ.

The idea of ADTRS led to the development of the "N-state receptor models," in which the receptor is assumed to exist in N states that may be "inactive" (R) or "active" (R*, R**, etc.). As a special case of the N-state receptor models, Leff et al. (1997) developed a mathematical model for three receptor states that accommodates the concept of ADTRS. They were able to predict differential R.e.s from this model. In this regard, Berg et al. (1998) recently reported strong evidence for the existence of pathway-dependent R.e.s for a series of five agonists at 5-hydroxytryptamine (5-HT)_2A and 5-HT_2C receptors when looking at the phospholipase C and phospholipase A₂ signal transduction pathways. They also pointed out that for ADTRS to be possible, it is critical that the independent pathways diverge at the receptor/G protein level (e.g., receptor activates two G proteins independently, leading to the transduction of the stimulus through two separate effector pathways) and not downstream from a common G protein (e.g., receptor activates a G protein such as G_q that subsequently activates phospholipase C, which in turn activates two independent effector pathways via release of inositol trisphosphate and diacylglycerol).

According to the three-state model by Leff et al. (1997), two cases can be distinguished. The first is when the two effector pathways result in two independently measurable responses, as in the work of Berg et al. (1998) mentioned earlier. Second, the two pathways can diverge and eventually recombine to modulate one measurable response (e.g., regulation of adenylyl cyclase by both G_s and G_i proteins). If one pathway is isolated by inactivating the other selectively, R.e.s can be determined for the active pathway. It should now be possible to obtain an estimate of the R.e.s for each pathway independently, and theoretically their R.e.s may be different.

_2A-Adrenergic receptors (_2A-ARs) have previously been shown to activate three G proteins: G_i, G_s, and G_q (Eason et al., 1992; Chabre et al., 1994). The former study reported a pertussis toxin-insensitive stimulation of cAMP accumulation in Chinese hamster ovary (CHO) cells expressing high amounts of ₂-AR. In addition to functional studies, direct, agonist-dependent, physical coupling of the _2A-AR to G_s was demonstrated (Eason et al., 1992). For the G_q coupling, Chabre et al. (1994) used human embryonic kidney (HEK) 293 cells transiently transfected with the porcine _2A-AR and murine G_q or rat G_s. They estimated that the efficiency of coupling of the _2A-AR to endogenous G_i was approximately 1000 times higher than that to G_s or G_q.

Although no conclusive evidence was provided to support the hypothesis that ADTRS may occur for ₂-ARs, three previous reports suggested such a possibility. Using CHO cells expressing the human _2A-AR (₂C10), Eason et al. (1994) showed a rank order of intrinsic activities of agonists for the G_s signal transduction pathway to be epinephrine = norepinephrine = UK-14,304 > BHT-933 > BHT-920 > oxymetazoline. For the G_i signal transduction pathway, they found that the intrinsic activities of these agonists were remarkably similar. They did not, however, take into account receptor reserve, so the true relative efficacies of these drugs cannot be determined from their data. Kenakin (1995) found that oxymetazoline selectively activated G_i (compared with G_s) in CHO cells stably transfected with _2A-AR. Yang and Lanier (1999) found differential regulation of G_o and G_i by clonidine and epinephrine. A potential mechanism for ADTRS has been suggested by reports that different amino acid residues of the _2A-AR are required for G_i and G_s activation (Eason and Liggett, 1996; Wade et al., 1999).

In the current study, we asked whether convincing evidence for the ADTRS hypothesis could be found for the regulation of G_i and G_s by the _2A-AR. To account for the difference in spare receptors (for a full agonist) when measuring G_s or G_i responses, we used two CHO-K1 cell lines: one expressing the porcine _2A-AR at high levels and the other expressing the porcine _2A-AR at low levels. We estimated the R.e.s for a series of agonists for the G_s and G_i signal transduction pathways, measuring [³H]cAMP accumulation as the response. Interestingly, l-isoproterenol (ISO), but not oxymetazoline, exhibits ADTRS.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Materials

Radiochemicals. [2-³H]Adenine (21-25 Ci/mmol) was obtained from Amersham Life Science (Piscataway, NJ). [³H]Yohimbine (74.5-78 Ci/mmol) was obtained from DuPont-New England Nuclear (Boston, MA).

Other Chemicals. cAMP, ATP, 3-isobutyl-1-methyl-xanthine (IBMX), l-epinephrine-(+)-bitartrate, ISO, and yohimbine HCl were purchased from Sigma (St. Louis, MO). UK-14,304, p-iodoclonidine HCl, l-norepinephrine bitartrate, BHT-933 dihydrochloride, xylazine HCl, and (S)-()-cyanopindolol hemifumarate were obtained from Research Biochemicals Inc. (Natick, MA). Clonidine HCl and p-aminoclonidine were purchased from Boehringer Ingelheim (Ingelheim, Germany). Guanabenz acetate was obtained from Wyeth Laboratories (Philadelphia, PA). BHT-920 Cl₂ was purchased from Dr. Karl Thaomae Inc. (Biberach, Germany). Oxymetazoline HCl was purchased from Schering Corporation (Bloomfield, NJ). Propranolol HCl was purchased from Ayerst Laboratories Inc. (New York, NY). Benextramine tetrahydrochloride monohydrate was obtained from Aldrich (Milwaukee, WI). Pertussis toxin (PTX) was obtained from List Biological Laboratories (Campbell, CA). Forskolin was obtained from Calbiochem-Novabiochem Corp. (San Diego, CA). LipoFECTAMINE and geneticin (G418) were obtained from Life Technologies (Gaithersburg, MD). Fluorescein-conjugated 12CA5 anti-hemagglutinin monoclonal antibody was purchased from Boehringer Mannheim (Indianapolis, IN).

_2A-AR Expressing Cell Lines

In this study, we used several cell lines. The first are CHO-K1 cell lines expressing high concentrations of either the wild type or one of two mutant porcine _2A-ARs (Wade et al., 1999). All receptor constructs contain an amino-terminal HA tag. The receptor concentrations were estimated as 19 ± 2 pmol/mg membrane protein for the wild-type _2A-AR (clone 1 designated _2A-H in this report), 10 ± 1 pmol/mg for the R3 mutant _2A-AR (mutating RWRGR to AWAGA at residues 361-365 of the receptor, designated _2A-R3 in this report), and 36 ± 3 pmol/mg for the B2 mutant (mutating basic residues 368-371 of the membrane-proximal i3c region to form BXAA, designated _2A-B2 in this report). The _2A-R3 and _2A-B2 mutations disrupt coupling to G_s and G_I, respectively (Wade et al., 1999).

For the purpose of this study, we isolated two additional CHO-K1 cell lines expressing a low concentration of the wild-type porcine _2A-AR. This was done with another flow cytometry sorting selection from the original transformation of Wade et al. (1999) from which the WT (clone 1) had been isolated. The cell line expressing the lowest concentration of the _2A-AR, as determined by [³H]yohimbine binding (B_max ~1 pmol/mg, clone 101), was selected for this study (designated _2A-L). This line exhibited a similar EC₅₀ value for the dose-response curve of UK-14,304 through the G_i pathway as did _2A-H through the G_s pathway, simplifying comparison of the G_s and G_i responses. The Neo cells containing the selection plasmid but no _2A-AR vector were used as controls.

Measurement of Whole-Cell [³H]cAMP Accumulation

CHO-K1 cells were maintained in Ham's F-12 medium with 10% fetal bovine serum, 100 U/ml penicillin, and 100 µg/ml streptomycin at 37°C in 5% CO₂. Selection for stable expression was maintained by the addition of 0.4 mg/ml G418 (active). [³H]cAMP accumulation was determined in whole cells in 24-well plates as described by Wade et al. (1999), adding 1 µCi/well [³H]adenine and, when indicated, 100 ng/ml PTX for at least 18 h before the assay. Cells were then washed once with Dulbecco's modified Eagle's medium (DMEM), after which the assay was initiated by adding DMEM with 1 mM IBMX and 30 µM forskolin and the appropriate drug or drugs. After a 20-min incubation, the medium was aspirated, and the reaction was terminated with 1 ml of 5% trichloroacetic acid (TCA) containing 1 mM ATP and 1 mM cAMP. The acid-soluble nucleotides were separated on Dowex and alumina columns as described by Salomon et al. (1974). The cAMP accumulation was normalized by dividing the [³H]cAMP counts by the total [³H]nucleotide counts. The control percent conversion of ATP to cAMP was 10 to 14% for non-PTX-treated cells (n = 81) and 3.7 to 7.4% for PTX-treated cells (n = 66) and did not vary significantly for high, low, or no _2A-AR expression. This percent conversion value was then divided by the corresponding value obtained with only IBMX and forskolin and no drug (to calculate percent of control).

Ligand Binding Assays

The K_i values of the _2A-AR agonists were determined from competition binding curves in whole cells against 5 nM [³H]yohimbine (K_D = 3 nM). The buffer used in these binding studies was comparable with that used for measuring cAMP accumulation. Cells were plated and incubated as before but without [³H]adenine and PTX. Cells were then washed once with OptiMEM, after which the assay was initiated by adding OptiMEM with 5 nM [³H]yohimbine and different concentrations of the testing drug. After a 30-min incubation, the medium was aspirated, the cells were washed twice, and the reaction was terminated with 1 ml of 5% TCA and allowed to stand for at least 30 min to allow the cells to lyse. The TCA from each well was then transferred directly into scintillation vials using transfer pipettes, and the [³H]yohimbine was counted.

Pharmacological Receptor Inactivation

Benextramine was used as an irreversible competitive antagonist of _2A-ARs. The cells were incubated overnight in 24-well plates as described earlier. The cells were then washed once with DMEM and incubated with 0, 1, 10, or 100 µM benextramine in PBS (containing 0.8% NaCl, 0.02% KCl, 0.09% Na₂HPO₄, and 0.02% KH₂PO₄) for 20 min at room temperature. This was followed by two washes with DMEM. With the second wash, the cells were let to stand in the DMEM for at least 5 min to ensure that all unbound benextramine was removed. The measurement of [³H]cAMP accumulation then proceeded as described earlier.

Data Analysis

Functional data, except where indicated otherwise, were obtained from three or more independent and comparable experiments, each in triplicate, and expressed as mean ± S.E.

Analyses of dose-response curves were made with the nonlinear least-squares method of the computer program Prism (GraphPad Software, San Diego, CA), setting the Hill slope factor at 1. Results are expressed as the mean ± S.E. To verify statistical significance of differences between mean values, the nonparametric Student's t test was used. After the Bonferroni correction for multiple comparison, a value of P < .05 was taken as statistically significant.

The relative intrinsic activity (RIA) was determined from functional data and expressed as the maximal response (E_max) of an agonist relative to that of the full agonist UK-14,304. The apparent dissociation constant of an agonist-receptor complex [K_A(app.)], and the fraction of receptors still functional after partial receptor alkylation (q) was estimated from Furchgott analysis of dose-response curves of UK-14,304 before and after partial receptor alkylation (Furchgott, 1966).

Because there appeared to be a small receptor reserve remaining for the full agonists (i.e., l-epinephrine, l-norepinephrine, UK-14,304) in _2a-L for G_i and _2a-H for G_s responses, we estimated the R.e.s of full agonists as follows:
where E₅₀ is the response at a drug concentration equal to its K_i value. Relative efficacies of partial agonists were determined as:
where S_max and E_max are as defined later.

This analysis for full agonists is similar to that used by Van Rossum (1966) and Van den Brink (1977) to define a "corrected" intrinsic activity constant ^S or by Venter (1997) to define the efficacy-related parameter e^ES. It assumes that 50% of the functionally coupled receptors are occupied at a concentration of agonist equal to the K_i value and that the relationship between receptor occupation and stimulus is linear over the range studied. Because the maximal stimulus should be proportional to twice the stimulus at the K_i, we also define S_max to equal two times the response (linearly related to stimulus) obtained at the E₅₀. Supporting this assumption of linearity, the S_max/E_max value for UK-14,304 determined in this manner is also similar to the S_max/E_max value from the q value estimated from Furchgott analysis before and after partial receptor alkylation (see above), where E_max was obtained before partial receptor alkylation. In this case, S_max was estimated from E_max'/q, where E_max' represents the reduced E_max after partial receptor alkylation.

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

_2A-ARs Expressed in _2A-H and _2A-L Are Comparable Pharmacologically. From saturation binding with [³H]yohimbine, the B_max values in _2A-H and _2A-L cells were 2.0 ± 0.1 × 10⁶ and 0.27 ± 0.09 × 10⁶ _2A-ARs per cell, respectively. Thus, _2A-H cells express ~10 times more of the porcine _2A-AR than _2A-L cells. The pK_D values for yohimbine in _2A-H and _2A-L were similar, being 8.57 ± 0.06 and 8.57 ± 0.24, respectively.

View larger version (26K): [in this window] [in a new window]   Fig. 1.   The 2A-AR couples to Gi and Gs in 2A-H and 2A-L. Whole-cell [3H]cAMP accumulation measurements with increasing concentrations of UK-14,304 were made for a high-expressing cell line, 2A-H (, ), or for a lower-expressing cell line, 2A-L (). Cells were pretreated overnight with (top) or without (bottom) 100 ng/ml PTX. The data are averages of triplicate measurements from at least three experiments in each cell line. They are expressed as percent of control samples with no agonist.

View this table: [in this window] [in a new window]   TABLE 1 The KA(app.) and q values for UK-14,304 as calculated from Furchgott analysis of dose-response curves of UK-14,304 in 2A-H (+PTX) and 2A-L (PTX) before and after partial receptor alkylation by the irreversible inhibitor benextramine Smax/Emax ratios were calculated from the q values and Emax values after partial receptor alkylation and from the submaximal response of UK-14,304 at a 1× Ki concentration, as described in the text (Data Analysis).

8

View larger version (33K): [in this window] [in a new window]   Fig. 2.   Effect of the irreversible 2A-AR inhibitor benextramine on responses in 2A-H and 2A-L cell lines. Whole-cell [3H]cAMP accumulation measurements were made with increasing concentrations of the agonist UK-14,304 after a 20-min treatment with 0 (), 1 µM (), 10 µM (), or 100 µM () of the irreversible inhibitor benextramine. Cells used were 2A-H (+PTX) (A), 2A-H (PTX) (B), and 2A-L (PTX) (C). Data are averages of triplicate measurements from four experiments in 2A-H (+PTX), one experiment in 2A-H (PTX), and two experiments in 2A-L (PTX). Curves are nonlinear least-squares fits.

Binding Data for Selected _2A-AR Agonists. According to classic theory, ~95% receptor occupation, and therefore E_max, is expected at a concentration of 20× K_i of an agonist. Thus, RIAs can be estimated from these single maximal concentrations of a series of agonists. Therefore, we calculated the K_i values for a series of _2A-AR agonists from competition binding against 5 nM [³H]yohimbine (Table 2). All competition binding curves, except for l-epinephrine, l-norepinephrine, and clonidine, were monophasic and could be explained by a single binding site. The competition binding curves for l-epinephrine, l-norepinephrine, and clonidine were biphasic with high-affinity pK_i values being 6.19, 5.25, and 9.18, respectively (and 57, 41, and 23% of the [³H]yohimbine displaced, respectively) and low-affinity pK_i values being 4.07, 3.58, and 7.07, respectively. For l-epinephrine and l-norepinephrine, the biphasic competition binding curves can be explained by their hydrophilic properties, presumably hindering access to nonsurface _2A-ARs. The high-affinity pK_i would therefore be expected to describe binding to the surface receptors. After 1 or 10 µM benextramine treatment, the pEC₅₀ values of the dose-response curves of l-epinephrine (data not shown) corresponded with the high-affinity pK_i value as expected. Also, the high-affinity pK_i values for l-epinephrine and l-norepinephrine were closest to the pEC₅₀ values measured from functional dose-response curves in _2A-H (+PTX) and _2A-L (PTX) (see later). However, for the partial agonist clonidine, the pEC₅₀ values from functional dose-response curves (data not shown) corresponded best with the low-affinity pK_i. We cannot explain the small fraction of high-affinity clonidine binding sites observed here.

View this table: [in this window] [in a new window]   TABLE 2 The Ki values, pEC50 values, R.I.A., and R.e. of selected agonists on 2A-ARs for the Gs and Gi pathways The Ki values were determined from competition binding against 5 nM [3H]yohimbine in 2A-H whole cells, and the pEC50 and R.I.A. values were determined from functional studies measuring [3H]cAMP accumulation. R.I.A.s were obtained at 20× Ki concentrations of the agonists and are expressed relative to the maximal response of the full agonist UK-14,304, which showed no response in control Neo cells (±PTX). R.e.s were estimated as described in the text (Data Analysis). Values are mean ± S.E.

Comparison of G_s- and G_i-Mediated Signaling. _2A-H (+PTX) and _2A-L (PTX) were used to estimate RIAs of a series of full and partial agonists for the G_s- and G_i-mediated modulation of adenylyl cyclase activity. Because the _2A-AR couples more effectively to G_i than to G_s, [³H]cAMP production was measured in the lower receptor expressing _2A-L cells to examine the G_i pathway and in the higher receptor expressing _2A-H cells treated with PTX (+PTX) for the G_s pathway. Neo cells without the porcine _2A-AR were used as controls.

View larger version (35K): [in this window] [in a new window]   Fig. 3.   Dose-response curves of various agonists for the Gs and Gi pathways. Whole-cell [3H]cAMP accumulation measurements with increasing concentrations of the agonists l-epinephrine (), l-norepinephrine (), UK-14,304 (), BHT-920 (), oxymetazoline (*), and l-isoproterenol () were performed on 2A-H cells (+PTX) and 2A-L cells (PTX). A, C, E, and G, Gs responses (2A-H, +PTX). B, D, F, and H, Gi responses (2A-L, PTX). The EC50 values are reported in Table 2. Gs- and Gi-mediated responses show comparable relative intrinsic activities with the exception of l-isoproterenol and oxymetazoline. Data are averages of triplicate measurements from three experiments.

View larger version (27K): [in this window] [in a new window]   Fig. 4.   The Gs response of l-isoproterenol, but not the Gi response of oxymetazoline, is mediated by 2A-ARs. Whole-cell [3H]cAMP accumulation measurements with increasing concentrations of (A and B) oxymetazoline and (C and D) l-isoproterenol were performed on 2A-H (+PTX) and 2A-L (PTX) (, ) and compared with the corresponding responses in Neo (±PTX) (, ). The Gs-mediated stimulation of adenylyl cyclase by l-isoproterenol in 2A-H (+PTX) is mediated by 2A-ARs, because Neo (+PTX) shows no response. However, enhanced oxymetazoline inhibition of adenylyl cyclase is mediated by an endogenous Gi-coupled, non-2-AR (see Neo cells). Data are averages of triplicate measurements from three experiments.

ISO Responses Are Blocked by _2A Antagonist Yohimbine but Not by -Blocker Propranolol. To further verify that the G_s response seen with ISO in _2A-H (+PTX) is mediated by interaction with _2A-ARs, we obtained data for dose-response curves for ISO in the absence and presence of the -AR antagonist propranolol (because ISO is a classic -AR agonist) or the ₂-AR antagonist yohimbine. From Fig. 5A, it can be seen that 1 µM propranolol does not inhibit the response by ISO, whereas 1 µM yohimbine greatly inhibits this response. This would exclude the contribution of any endogenous -ARs to the observed stimulation of adenylyl cyclase by ISO. It also confirms that the G_s response is mediated by _2A-ARs, in agreement with data from the Neo cells. Results for the G_i response suggest the same with regard to the interaction of ISO with _2A-ARs (Fig. 5B).

View larger version (31K): [in this window] [in a new window]   Fig. 5.   Yohimbine, but not propranolol, blocks the response of l-isoproterenol. Whole-cell [3H]cAMP accumulation measurements with increasing concentrations of l-isoproterenol for the Gs (2A-H, +PTX) (A) and Gi (2A-L, PTX) (B) pathways without () and with 1 µM -AR antagonist propranolol () or 1 µM 2-AR antagonist yohimbine (). The data are averages of triplicate measurements from three experiments and indicate that the Gs-mediated response is mediated via 2A-ARs and not by -ARs.

Oxymetazoline but Not UK-14,304 Responses Are Blocked by 5-HT₁ Antagonist ()-Cyanopindolol. Figure 6 shows dose-response curves of UK-14,304 and oxymetazoline in the presence and absence of 100 nM ()-cyanopindolol in _2A-L (PTX). The G_i response of oxymetazoline, but not that of UK-14,304, was antagonized by 100 nM ()-cyanopindolol. The estimated pA₂ value is 8.44, indicating a K_B value of ~3.6 nM for ()-cyanopindolol for the receptor involved. These data suggest that 5-HT₁ receptors may mediate much of the G_i response of oxymetazoline (see Discussion). Also, 100 nM propranolol did not shift the response of oxymetazoline to the right, excluding any contribution of -ARs to the response. Thus, the apparent ADTRS by oxymetazoline is an artifact of endogenous receptors, whereas that of ISO reflects a true property of the _2A-AR.

View larger version (18K): [in this window] [in a new window]   Fig. 6.   Adenylyl cyclase inhibition by oxymetazoline, but not that by UK-14,304, is antagonized by ()-cyanopindolol. Whole-cell [3H]cAMP accumulation measurements with increasing concentrations of UK-14,304 (A) in the absence () or presence of 100 nM ()-cyanopindolol () and oxymetazoline (B) in the absence () or presence of 100 nM ()-cyanopindolol () or 100 nM propranolol (*). The data are averages of triplicate measurements from three experiments and indicate that a non-2A-AR mediates the Gi-response of oxymetazoline.

Estimating Relative Efficacies of Agonists. Because EC₅₀/K_i ratios for l-epinephrine, l-norepinephrine, ISO, and UK-14,304 in _2A-H (+PTX) and for l-epinephrine, l-norepinephrine, and UK-14,304 in _2A-L (PTX) suggest a small degree of receptor reserve for these full agonists, R.e. values cannot be estimated directly from E_max values. This was further confirmed from dose-response curves of UK-14,304 before and after benextramine treatment (Fig. 2). Another complicating factor was the significant stimulation of adenylyl cyclase by l-epinephrine and l-norepinephrine seen in control Neo (±PTX) cells at concentrations of 20 times the K_i value (Table 3).

_2A-AR Effector Site for G_s Coupling as Activated by ISO May Be Similar to that for Classic _2A-AR Agonists. Because ISO is not a classic _2A-AR agonist and was the only agonist to show selectivity for activating the G_s pathway, the question arose of whether ISO induced an atypical G protein contact surface (or effector site) in the receptor for G_s coupling. Wade et al. (1999) recently showed that distinct _2A-AR regions are required for G_i and G_s activation. Using these mutant _2A-ARs that disrupt either G_s coupling (_2A-R3) or G_i coupling (_2A-B2), we investigated whether these mutations to the _2A-AR would disrupt the coupling to G_s and G_i in a similar fashion when ISO was used as agonist.

View larger version (28K): [in this window] [in a new window]   Fig. 7.   UK-14,304 and l-isoproterenol use similar effector regions on the 2A-AR for activation of Gs. Whole-cell [3H]cAMP accumulation measurements were made with increasing concentrations of UK-14,304 and l-isoproterenol. Shown are Gs (A and C, +PTX) and Gi (B and D, PTX) responses with wild-type 2A-AR, , the 2A-R3 mutant , or the 2A-B2 mutant  (see text for full explanation). Note that l-isoproterenol acts as a full agonist in the high 2A-AR expressing 2A-H (PTX) cell line compared with being a partial agonist in 2A-L (PTX) in other figures. Data are averages of at least triplicate measurements from three experiments.

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

The present study was undertaken to investigate agonist-directed trafficking of porcine _2A-AR signaling through the G_s- and G_i-coupled signal transduction pathways.

ISO Selectively Activates G_s Pathway by Interaction with _2A-ARs. ISO was the only agonist tested to show clear selectivity for G_s or G_i at the _2A-AR. Interestingly, ISO had a 9-fold higher R.e. at the _2A-AR for G_s versus G_i, but it has low potency for both. It is clear from the data with both Neo control cells and pharmacological antagonists that this effect is mediated by the _2A-AR. These results support the hypothesis of ADTRS (Kenakin, 1995). We were not able to confirm the conclusion of Eason and Liggett (1996) that BHT-920 and BHT-933 lead to selective activation of G_i. The difference between these two conclusions is probably due to the fact that our analysis takes into account spare receptors, whereas theirs did not. A further difference is their use of human _2A-AR, whereas we have porcine receptors, although this does not seem to be a likely explanation. Other recently published reports of ADTRS include the study of Berg et al. (1998) in which 5-HT receptors couple to phospholipases C and A₂ with efficacies dependent on the specific agonist; Bonhaus et al. (1998) found differential G_i and G_s coupling with cannabinoid (CB₁) agonists, and Yang and Lanier (1999) found differential coupling of rat _2A-ARs to G_o and G_i in NIH 3T3 cells.

Oxymetazoline Activates G_i Pathway by Interaction with Non-_2A-ARs. Enhanced inhibition of adenylyl cyclase by oxymetazoline in CHO-K1 cells as previously reported by Kenakin (1995), appears to be mediated via an endogenous 5-HT receptor (Figs. 4 and 6). Schoeffter and Hoyer (1991) estimated the affinity of oxymetazoline at 5-HT_1B (rat cortex) receptors as a K_D value of 26 nM. They also reported that stimulation of 5-HT_1A, 5-HT_1B, and 5-HT_1D receptors inhibited adenylyl cyclase, whereas activation of 5-HT_1C receptors stimulated adenylyl cyclase activity. Because CHO cells have an endogenous 5-HT_1B receptor (Berg et al., 1994; Giles et al., 1996), that is the likely source of the excess G_i response to oxymetazoline.

Rank Order of Relative Efficacies for G_s and G_i Is Generally Similar. With the exception of ISO, all agonists showed very similar R.e.s at the _2A-AR for the G_s and G_i pathways. There may be modest differences in R.e.s, such as BHT-920 for the G_i and G_s pathways (0.27 ± 0.02 and 0.19 ± 0.02, respectively), but this study did not have sufficient statistical power to define such a small difference. The rank order of R.e.s at the _2A-AR for activation of G_s and G_i pathways are as follows.

i1

i2

i3

Defining _2A-H and _2A-L Pharmacologically. The 120-fold difference in the EC₅₀ value of the dose response curves of UK-14,304 for the G_i pathway in _2A-H (PTX) versus the G_s pathway in _2A-H (+PTX) suggests that the _2A-AR couples much more efficiently to the G_i protein than to the G_s protein. This preferential coupling of the _2A-AR to G_i is in agreement with data from Eason et al. (1992, 1994) and Chabre et al. (1994). However, in HEK 293 cells transfected to transiently coexpress the porcine _2A-AR with either G_s, G_i, or G_q, Chabre et al. (1994) found that the _2A-AR couples ~1000 times more efficiently to G_i (endogenous to HEK 293) than to either G_s (rat G_s) or G_q (murine G_q). The reason for the difference in the G_i/G_s selectivity of the _2A-AR coupling as suggested by the data of Chabre et al. versus our data (1000 versus 120) is not known. It is possible that coupling to endogenous G proteins is more efficient than to transfected G proteins.

Conclusions. We provide additional support for the ADTRS hypothesis in a pharmacologically well-defined system. High _2A-AR expression (higher than found in most mammalian tissue) is necessary for significant G_s coupling, which may limit the relevance of these findings for normal physiological conditions. However, the implications of this important hypothesis for drug design are highly significant. To fully understand the molecular mechanisms of ADTRS, it will be important to extend these results to purified systems with direct measurements of G protein activation.