Differential Coupling of Serotonin 5-HT1A and 5-HT1B Receptors to Activation of ERK2 and Inhibition of Adenylyl Cyclase in Transfected CHO Cells

Authors

  • Jorge Mendez,

    1. Department of Psychiatry, University of Medicine and Dentistry of New Jersey-Robert Wood Johnson Medical School, Piscataway, New Jersey, U.S.A.
    Search for more papers by this author
  • Tapan M. Kadia,

    1. Department of Psychiatry, University of Medicine and Dentistry of New Jersey-Robert Wood Johnson Medical School, Piscataway, New Jersey, U.S.A.
    Search for more papers by this author
  • Ravi K. Somayazula,

    1. Department of Psychiatry, University of Medicine and Dentistry of New Jersey-Robert Wood Johnson Medical School, Piscataway, New Jersey, U.S.A.
    Search for more papers by this author
  • Khaled I. El-Badawi,

    1. Department of Psychiatry, University of Medicine and Dentistry of New Jersey-Robert Wood Johnson Medical School, Piscataway, New Jersey, U.S.A.
    Search for more papers by this author
  • Daniel S. Cowen

    1. Department of Psychiatry, University of Medicine and Dentistry of New Jersey-Robert Wood Johnson Medical School, Piscataway, New Jersey, U.S.A.
    Search for more papers by this author

  • Lippincott Williams & Wilkins, Inc., Philadelphia

  • Abbreviations used: cAMP, cyclic AMP; CHO, Chinese hamster ovary; ERK, extracellular signal-regulated kinase; 5-HT, 5-hydroxytryptamine (serotonin); IBMX, 3-isobutyl-1-methylxanthine; MAP, mitogen-activated protein; p-MPPI, 4-iodo-N-[2-[4-(methoxyphenyl)-1-piperazinyl]ethyl]-N-2-pyridinylbenzamide hydrochloride; 8-OH-DPAT, (±)-8-hydroxy-N,N-dipropyl-2-aminotetralin hydrobromide; RIA, radioimmunoassay.

Address correspondence and reprint requests to Dr. D. S. Cowen at Department of Psychiatry, University of Medicine and Dentistry of New Jersey-Robert Wood Johnson Medical School, 125 Paterson Street, New Brunswick, NJ 08901, U.S.A.

Abstract

Abstract: Although the subtypes of serotonin 5-HT1 receptors have distinct structure and pharmacology, it has not been clear if they also exhibit differences in coupling to cellular signals. We have sought to compare directly the coupling of 5-HT1A and 5-HT1B receptors to adenylyl cyclase and to the mitogen-activated protein kinase ERK2 (extracellular signal-regulated kinase-2). We found that 5-HT1B receptors couple better to activation of ERK2 and inhibition of adenylyl cyclase than do 5-HT1A receptors. 5-HT stimulated a maximal fourfold increase in ERK2 activity in nontransfected cells that express endogenous 5-HT1B receptors at a very low density and a maximal 13-fold increase in transfected cells expressing 230 fmol of 5-HT1B receptor/mg of membrane protein. In contrast, activation of 5-HT1A receptors stimulated only a 2.8-fold maximal activation of ERK2 in transfected cells expressing receptors at 300 fmol/mg of membrane protein but did stimulate a 12-fold increase in activity in cells expressing receptors at 3,000 fmol/mg of membrane protein. Similarly, 5-HT1A, but not 5-HT1B, receptors were found to cause significant inhibition of forskolin-stimulated cyclic AMP accumulation only when expressed at high densities. These findings demonstrate that although both 5-HT1A and 5-HT1B receptors have been shown to couple to G proteins of the Gi class, they exhibit differences in coupling to ERK2 and adenylyl cyclase.

At least 16 types of mammalian receptors for serotonin [5-hydroxytryptamine (5-HT)] have been reported (for reviews, see Hoyer et al., 1994; Scalzitti and Hensler, 1996). All are G protein-coupled receptors with the exception of 5-HT3 receptors, which are ligand-gated ion channels. Those receptors that couple to G proteins of the Gi class and inhibit adenylyl cyclase have been classified as 5-HT1 receptors. They have been further subtyped as 5-HT1A, 5-HT1B, 5-HT1D, 5-HT1E, and 5-HT1F receptors. Although these subtypes of 5-HT1 receptors have clear distinctions in pharmacology and structure, differences in coupling to cellular signals have not been previously demonstrated. In contrast, comparison of results across studies suggests that the receptors may be functionally equivalent, bringing into question the significance of expression of multiple subtypes in brain.

In fact, there has been a tendency to lump together all receptors that couple to G proteins of the Gi class, with the assumption that they all activate identical cellular signals. Clearly, various studies have demonstrated that under the appropriate cellular conditions some of these receptors can exhibit identical function. For example, in addition to coupling negatively to adenylyl cyclase, many Gi-coupled receptors have been reported to activate the mitogen-activated protein (MAP) kinases ERK1 and ERK2 (extracellular signal-regulated kinase-1 and -2, respectively) (Meloche et al., 1992; Koch et al., 1994; Flordellis et al., 1995; Luttrell et al., 1995; Cowen et al., 1996; Garnovskaya et al., 1996; Pullarkat et al., 1998). However, some of these studies have used stable transfected cell lines that overexpress receptor at densities not observed in cells expressing endogenous receptor. Other studies have used cells transiently transfected with cDNA for the receptors. Such cells do not express receptor at a homogeneous density. In contrast, some cells express receptors at extremely high density, whereas other cells express no receptors. Studies using cells over-expressing receptor can correctly demonstrate that a particular receptor can couple to a specific signal when receptors are expressed at the high density. However, such studies may miss the fact that the receptor would not couple at lower (more physiological) densities. When receptor function has been studied under conditions where receptors are expressed at the same densities, differences have been found. For example, using stable transfected cell lines expressing receptors in the picomole per milligram of membrane protein range, Flordellis et al. (1995) demonstrated that α2B- and α2D-adrenergic receptors but not α2C-adrenergic receptors couple effectively to MAP kinase.

The MAP kinases ERK1 and ERK2 are serine/threonine protein kinases. They have been shown to phosphorylate several transcription factors, including c-Jun, p62TCF/Elk-1, c-Fos, and c-Myc and also appear to regulate translation of mRNA (for review, see Denton and Tavare, 1995). Touhara et al. (1995) recently described a model for Gi-coupled receptor stimulation of MAP kinase requiring G protein βγ subunits, an unidentified tyrosine kinase, Shc, phosphatidylinositol 3-kinase, Grb2/SOS, and Ras. Other components of the pathway are thought to include Raf and MAP kinase kinase (MEK/MKK). The pathway for activation of MAP kinase is therefore much different from that for inhibition of adenylyl cyclase, which is directly mediated by Gi, through α and βγ subunits (for review, see Tang and Hurley, 1998).

In the present study we sought to identify differences in the coupling of 5-HT1A and 5-HT1B receptors to cellular signals. Both receptors have been shown in separate studies to inhibit the activity of adenylyl cyclase (Fargin et al., 1989; Unsworth and Molinoff, 1992; Berg et al., 1994, 1996; Harrington et al., 1994; Clarke et al., 1996; Giles et al., 1996) and in some cases to evoke small increases in intracellular Ca2+ level (Fargin et al., 1989; Harrington et al., 1994; Dickenson and Hill, 1995; Giles et al., 1996). However, attempts to make comparisons of receptor function across separate studies are complicated by the use of different cell types and of cells expressing receptors at different densities. In the present studies we used the same parent cell line [Chinese hamster ovary (CHO) cells] for all studies and selected stable transfected cell lines that express receptors at specific densities. We chose to study coupling to inhibition of adenylyl cyclase and activation of ERK2 because very different pathways regulate these signals. In earlier studies we previously demonstrated that 5-HT1B receptors, even when expressed in nontransfected CHO cells at low density (a density below the sensitivity of our binding assay), effectively couple to activation of ERK2 (Pullarkat et al., 1998). In separate studies we and others (Cowen et al., 1996; Garnovskaya et al., 1996) have also shown in transfected CHO cells that activation of 5-HT1A receptors expressed at high density (>1 pmol/mg of membrane protein) similarly stimulates ERK2. However, in the present studies we have directly compared the efficacy of each receptor subtype to activate ERK2 and to inhibit adenylyl cyclase in cells expressing receptors at the same densities. It is significant that we found that coupling by 5-HT1A receptors, but not 5-HT1B receptors, requires expression of receptor at a high density. This suggests that the coupling of 5-HT1A receptors to MAP kinase and to adenylyl cyclase in CNS neurons would be significantly more sensitive to changes in density of receptor expression.

MATERIALS AND METHODS

Materials

(±)-8-Hydroxy-N,N-dipropyl-2-aminotetralin hydrobromide (8-OH-DPAT) and 4-iodo-N-[2-[4-(methoxyphenyl)-1-piperazinyl]ethyl]-N-2-pyridinylbenzamide hydrochloride (p-MPPI) were purchased from Research Biochemicals International (Natick, MA, U.S.A.). Pertussis toxin, forskolin, and 3-isobutyl-1-methylxanthine (IBMX) were purchased from Calbiochem (San Diego, CA, U.S.A.). Protein A-Sepharose and 5-HT were purchased from Sigma (St. Louis, MO, U.S.A.).

Cell culture

CHO-K1 cells were obtained from the American Type Culture Collection (Rockville, MD, U.S.A.). A stable transfected CHO-K1 cell line expressing 5-HT1B receptors at a density of 230 fmol/mg of membrane protein was obtained by transfecting CHO-K1 cells with cDNA for the mouse 5-HT1B receptor and selecting for resistance to hygromycin B (200 μg/ml) (Pullarkat et al., 1998). Two stable clones expressing 5-HT1A receptors at a density of 300 and 3,000 fmol/mg of membrane protein were obtained by transfecting CHO-K1 cells with cDNA for the human 5-HT1A receptor (Fargin et al., 1988; Cowen et al., 1996) and selecting for resistance to geneticin (400 μg/ml). The density of 5-HT1B receptors was measured as described by Unsworth and Molinoff (1992) using the radioligand (±)-[125I]iodocyanopindolol obtained from DuPont NEN (Boston, MA, U.S.A.). The density of 5-HT1A receptors was measured as previously described (Cowen et al., 1997) using the radioligand [125I]p-MPPI obtained from DuPont NEN. Transfected cells were maintained in medium containing Ham’s F-12 nutrient mixture with L-glutamine, 10% dialyzed fetal bovine serum (dialyzed in membranes with 1,000-Da cutoffs against a 100-fold greater volume of 150 mM NaCl to remove endogenous 5-HT), 1% penicillin-streptomycin, and either 400 μg/ml geneticin (for cells expressing 5-HT1A receptors) or 200 μg/ml hygromycin B (for cells expressing 5-HT1B receptors) at 37°C (95% air, 5% CO2).

Assay of MAP kinase activity

MAP kinase activity was measured following immunoprecipitation with rabbit polyclonal IgG recognizing ERK2 (C-14) obtained from Santa Cruz Biotechnology (Santa Cruz, CA, U.S.A.), essentially as previously described (Flordellis et al., 1995; Cowen et al., 1996). The day before use, cells were washed with phosphate-buffered saline and cultured overnight under serum- and geneticin/hygromycin B-free conditions. Cells were stimulated with the specified concentrations of agonists for 5 min and lysed. ERK2 was immunoprecipitated from the cytosol, and activity was measured by incorporation of 32P into a nine-amino-acid MAP kinase substrate peptide corresponding to amino acids 95-98 of myelin basic protein (Upstate Biotechnology, Lake Placid, NY, U.S.A.). Phosphocellulose filters were spotted with the phosphorylated substrate, washed extensively in 0.75% phosphoric acid and acetone, and counted for radioactivity by scintillation spectrometry.

Assay of adenylyl cyclase activity

Cells were cultured in 12-well plates until confluent. The day before use, cells were washed with phosphate-buffered saline and cultured overnight under serum- and geneticin/hygromycin B-free conditions. The appropriate agonists were added to each well, immediately followed by 1 μM forskolin and 100 μM IBMX. The plates were incubated for 5 min at 37°C, and the reaction was stopped by aspiration and addition of 0.1 M hydrochloric acid. The cells were scraped from the plates and lysed with a 26-gauge needle. The lysate was neutralized, and cyclic AMP (cAMP) was quantified, after acetylation, by radioimmunoassay (RIA). The RIA protocol, anti-cAMP, and cAMP standard were obtained from Calbiochem. Adenosine 3′,5′-cyclic phosphoric acid 2′-O-succinyl[125I]iodotyrosine methyl ester was from DuPont NEN.

RESULTS

Differential activation of the MAP kinase ERK2 by 5-HT1A and 5-HT1B receptors

Studies were designed to compare directly the magnitude of activation of ERK2 elicited by 5-HT1A and 5-HT1B receptors. All studies were done in transfected and nontransfected CHO cells. This avoided potential cellular differences that would have complicated interpretation of results had different cell lines expressing endogenous receptors been used. 5-HT1B receptors were studied in nontransfected cells that express endogenous receptor at a very low density and in a stable transfected cell line. The transfected cell line (referred to in this article as CHO-1B-230) expresses receptor with a KD for the radioligand (±)-[125I]iodocyanopindolol of 50 pM, at a density of 230 fmol/mg of membrane protein. The density of endogenous 5-HT1B receptors expressed in nontransfected CHO cells is below the sensitivity of our binding assays—30 fmol/mg of membrane protein. 5-HT1A receptors were studied in two stable transfected cell lines expressing receptor at densities of 300 and 3,000 fmol/mg of membrane protein, referred to as CHO-1A-300 and CHO-1A-3000, respectively. The KD for the radioligand [125I]p-MPPI is ∼0.2 nM in both CHO-1A-300 and CHO-1A-3000 cell lines.

As we have previously reported (Pullarkat et al., 1998), treatment of nontransfected CHO-K1 cells with 5-HT causes stimulation of ERK2 activity through endogenous 5-HT1B receptors. 5-HT stimulated a maximal fourfold increase in activity with an EC50 of 45 nM (Fig. 1A). It is significant that in CHO-1B-230 cells 5-HT stimulated a larger, 13-fold, increase in activity with an EC50 of 10 nM (Fig. 1A).

Figure 1.

Differential activation of ERK2 by 5-HT1A and 5-HT1B receptors. A: Nontransfected (wild) CHO cells and CHO-1B-230 cells were incubated for 5 min with the indicated concentrations of 5-HT. B: Wild CHO, CHO-1A-300, and CHO-1A-3000 cells were incubated for 5 min with the indicated concentrations of 8-OH-DPAT. ERK2 was immunoprecipitated, and its activity was measured as described in MATERIALS AND METHODS. In A data are mean ± SE (bars) values of three separate experiments, each performed in triplicate, with activity expressed as a percentage of unstimulated activity. *p < 0.05 versus wild CHO by two-sided unpaired Student’s t test. In B data are mean ± SE (bars) values of six separate experiments, each performed in triplicate. *p < 0.05 versus CHO-1A-300 by two-sided unpaired Student’s t test.

FIG. 1.

In contrast to the activation of ERK2 by 5-HT1B receptors, which occurred even when the density of receptors was very low, the activation by 5-HT1A receptors required expression of receptor at a high density. Treatment of CHO-1A-300 cells with 8-OH-DPAT, an agonist selective for 5-HT1A receptors, stimulated only a maximal 2.8-fold increase in ERK2 activity with an EC50 of 60 nM (Fig. 1B). Therefore, 5-HT1A receptors expressed at a density of 300 fmol/mg of membrane protein stimulated only a 2.8-fold increase in ERK2 activity, in contrast to the 13-fold increase stimulated by 5-HT1B receptors expressed at 230 fmol/mg of membrane protein in CHO-1B-230 cells. The basal activity of ERK2 was similar in both cell types. Stimulation of ERK2 to a magnitude comparable to that seen by 5-HT1B receptors in CHO-1B-230 cells occurred only in CHO-1A-3000 cells. 8-OH-DPAT stimulated, in those cells, a maximal 12-fold increase in ERK2 activity with an EC50 of 10 nM (Fig. 1B). It is significant that the stimulation of ERK2 elicited by 8-OH-DPAT in CHO-1A-300 cells, although small, was selectively mediated by 5-HT1A receptors. 8-OH-DPAT, at concentrations of <10 μM, did not stimulate ERK2 activity in nontransfected CHO cells that do not express 5-HT1A receptors (Fig. 1B). Therefore, 8-OH-DPAT, at the concentrations used in these studies, did not act nonselectively as an agonist at endogenous 5-HT1B receptors.

Because 8-OH-DPAT stimulated only a small activation of ERK2 in CHO-1A-300 cells, it was important to demonstrate that this did not reflect selection of a mutated clonal cell line that had an impaired pathway for activation of MAP kinase. This was not the case. ERK2 could, in fact, be activated in CHO-1A-300 cells to an extent greater than that stimulated by 8-OH-DPAT. As shown in Fig. 2, treatment with 5-HT caused an eightfold stimulation of ERK2, compared with the 2.8-fold activation with 8-OH-DPAT. This larger stimulation was likely the result of activation of both transfected 5-HT1A receptors and endogenous 5-HT1B receptors. However, an alternative hypothesis for the poor activation by 8-OH-DPAT, relative to 5-HT, in CHO-1A-300 cells could have been that 8-OH-DPAT was not a full agonist for 5-HT1A receptors. This proved also to be not the case. 8-OH-DPAT stimulated a large, 13-fold, activation of MAP kinase in CHO-1A-3000 cells that was similar in magnitude to that stimulated by 5-HT (Fig. 2).

Figure 2.

Differential activation of ERK2 by 5-HT and 8-OH-DPAT (DPAT) in cells expressing transfected 5-HT1A receptors and endogenous 5-HT1B receptors. CHO-1A-300 and CHO-1A-3000 cells were incubated for 5 min with either 1 μM 5-HT or 1 μM DPAT. ERK2 was immunoprecipitated, and its activity was measured as described in MATERIALS AND METHODS. Data are mean ± SE (bars) values of 10 separate CHO-1A-300 experiments and four CHO-1A-3000 experiments, each performed in triplicate, with activity expressed as a percentage of unstimulated activity. *p < 0.05, ***p < 0.001 versus DPAT for each cell type by two-sided paired Student’s t test.

FIG. 2.

We further demonstrated that we were not looking at a partial agonist effect in studies in which 5-HT was used as the agonist. We first showed that the antagonist p-MPPI was selective for 5-HT1A receptors and does not inhibit the activation of ERK2 stimulated in nontransfected cells through 5-HT1B receptors (Fig. 3A). It can therefore be used to inhibit selectively the activation by 5-HT of transfected 5-HT1A receptors without inhibiting the activation by 5-HT of endogenous 5-HT1B receptors. As expected, 1 μM p-MPPI almost completely inhibited the activation of ERK2 stimulated by 8-OH-DPAT in CHO-1A-300 cells (Fig. 3B). However, it only slightly inhibited the activation by 5-HT, consistent with inhibition of a small 5-HT1A (relative to 5-HT1B) receptor-mediated response. In contrast, p-MPPI almost completely inhibited the activation of ERK2 by both 5-HT and 8-OH-DPAT in CHO-1A-3000 cells (Fig. 3C). In CHO-1A-3000 cells 5-HT1A receptors are expressed at sufficient density to mediate most of the activation of ERK2 by 5-HT, with the remaining small effect mediated by endogenous 5-HT1B receptors. Therefore, we conclude that although both human 5-HT1A and rodent 5-HT1B receptors stimulate ERK2, 5-HT1A receptors effectively do so only when expressed at high density. It is significant that, although 5-HT1B receptors were found to couple better to ERK2 than did 5-HT1A receptors, both receptors used pathways requiring Gi-type G proteins. Pretreatment with pertussis toxin caused almost complete inhibition of 5-HT1A receptor- and 5-HT1B receptor-mediated activation of ERK2 (data not shown).

Figure 3.

8-OH-DPAT (DPAT) and 5-HT display similar efficacy for stimulating 5-HT1A receptor-mediated activation of ERK2. A: Nontransfected (wild) CHO cells were incubated for 5 min with 1 μM 5-HT in the presence or absence of 1 μMp-MPPI (MPPI). CHO-1A 300 cells (B) and CHO-1A-3000 cells (C) were incubated for 5 min with 1 μM 5-HT or 1 μM DPAT in the presence or absence of 1 μM MPPI. ERK2 was immunoprecipitated, and its activity was measured as described in MATERIALS AND METHODS. Data are mean ± SE (bars) values of three separate experiments, each performed in triplicate, with activity expressed as a percentage of the activity stimulated by the individual agonist in the absence of antagonist. ***p < 0.001 versus value in the absence of MPPI by two-sided paired Student’s t test.

FIG. 3.

Differential inhibition of adenylyl cyclase activity by 5-HT1A and 5-HT1B receptors

Endogenous 5-HT1B receptors expressed on nontransfected CHO cells have been previously reported to couple negatively to adenylyl cyclase (Berg et al., 1994); Dickenson and Hill, 1995; Giles et al., 1996). We found that 5-HT caused a maximal 40% inhibition of forskolin-stimulated cAMP accumulation in these cells with an EC50 of 100 nM (Fig. 4A). It is significant that 5-HT stimulated a larger, 60% inhibition in CHO-1B-230 cells with an EC50 of 80 nM (Fig. 4A).

Figure 4.

Differential inhibition of forskolin-stimulated cAMP accumulation by 5-HT1A and 5-HT1B receptors. A: Nontransfected (wild) CHO cells and CHO-1B-230 cells were incubated for 5 min with 1 μM forskolin, 100 μM IBMX, and the indicated concentrations of 5-HT. B: CHO-1A-300 and CHO-1A-3000 cells were incubated for 5 min with 1 μM forskolin, 100 μM IBMX, and the indicated concentrations of 8-OH-DPAT. Intracellular cAMP was quantified by RIA, as described in MATERIALS AND METHODS. Data are mean ± SE (bars) values of three separate experiments, each performed in triplicate, with the activity expressed as the percentage of forskolin-induced increases in cAMP accumulation in the absence of receptor agonists. **p < 0.01 versus CHO-1A-3000 by two-sided unpaired Student’s t test.

FIG. 4.

In contrast, 5-HT1A receptors were found to couple effectively to inhibition of adenylyl cyclase only when expressed at a high density. Treatment of CHO-1A-300 cells with 8-OH-DPAT caused a maximal 20% inhibition of forskolin-stimulated cAMP accumulation with an EC50 of 200 nM (Fig. 4B). However, 8-OH-DPAT did inhibit forskolin-stimulated cAMP accumulation by 80% in CHO-1A-3000 cells with an EC50 of 3 nM (Fig. 4B). The accumulation of cAMP stimulated by forskolin was similar in all cell types.

DISCUSSION

Our findings demonstrate that in addition to having distinct structures and pharmacology, individual subtypes of 5-HT1 receptors also exhibit differences in coupling to cellular signals. 5-HT1B receptors were found to couple better to both inhibition of adenylyl cyclase and to activation of ERK2 than do 5-HT1A receptors. Activation of 5-HT1A receptors in CHO-1A-300 cells caused only a 20% inhibition of forskolin-stimulated cAMP accumulation and a 2.8-fold increase in ERK2 activity. In contrast, activation of 5-HT1B receptors in CHO-1B-230 cells caused a 60% inhibition of forskolin-stimulated cAMP accumulation and a 13-fold increase in MAP kinase activity. Even more striking was the 40% inhibition of adenylyl cyclase activity and fourfold activation of ERK2 seen in nontransfected CHO cells. These cells express endogenous 5-HT1B receptors at a density below that detectable by our binding assays and that of others (Giles et al., 1996). Because our binding assay is sensitive enough to measure receptor expression as low as 30 fmol/mg of membrane protein, we can assume that the receptors are expressed at a density below that and consequently at a density at least 10-fold lower than that at which 5-HT1A receptors are expressed on CHO-1A-300 cells. It is significant that although we have not been able to detect expression of 5-HT1B receptors in nontransfected CHO cells, we have previously demonstrated that the pharmacology for agonist-induced activation of ERK2 (Pullarkat et al., 1998) is consistent with their expression. Similarly, the pharmacology for inhibition of adenylyl cyclase (Berg et al., 1994; Dickenson and Hill, 1995; Giles et al., 1996) and activation of p70 S6 kinase (Pullarkat et al., 1998) in nontransfected CHO cells is consistent with mediation by 5-HT1B receptors. Therefore, the more effective coupling by 5-HT1B receptors to activation of ERK2 and inhibition of adenylyl cyclase, relative to 5-HT1A receptors, is striking.

It should be pointed out that our findings cannot be attributed to cellular differences in transfected cell lines resulting from clonal selection. Studies using a different transfected CHO cell line expressing human 5-HT1A receptors at a density of 250 fmol/mg of membrane protein confirmed our results from CHO-1A-300 cells, that agonists for 5-HT1A receptors cause a small activation of ERK2 when receptors were expressed at a low density (data not shown). In contrast, Garnovskaya et al. (1996), using a transfected CHO cell line expressing 5-HT1A receptors at 1,000 fmol of receptor/mg of protein, reported a fivefold increase in ERK2 activity. That magnitude of stimulation fits well with our finding of a 2.8-fold activation in cells expressing 300 fmol of receptor/mg of membrane protein and a 12-fold activation in our CHO-1A-3000 cells expressing 3,000 fmol of receptor/mg of membrane protein. The results of our studies demonstrate the importance of receptor density when comparisons are made of receptor function.

Ideally, studies of receptor function in transfected cell lines should use cells that express receptors at densities similar to those found in vivo. 5-HT1A receptors have been reported to be expressed at ∼200 fmol/mg of membrane protein in rat hippocampus (Butkerait et al., 1995), and 5-HT1B receptors have been reported to be expressed at a similar density in rat cortex (Hoyer et al., 1985). Therefore, our CHO-1A-300 and CHO-1B-230 cells appear to express receptors at a density similar to that found in vivo. However, the densities reported in hippocampus and cortex represent an average of the cellular expression in those regions. Because the density of receptors expressed on individual cells in those regions is not homogeneous, those neurons that do express receptors would be expected to do so at densities of >200 fmol/mg of membrane protein. Therefore, our CHO-1A-300 and CHO-1B-230 cell lines express receptors at levels similar to, or below, that expressed on neurons in hippocampus and cortex.

Our findings that receptor coupling to adenylyl cyclase and MAP kinase required expression of 5-HT1A, but not 5-HT1B, receptors at a high density suggest that the cellular effects elicited by 5-HT1A receptors in brain may be more sensitive to changes in receptor expression. In fact, Yocca et al. (1992) used the irreversible antagonist N-ethoxycarbonyl-2-ethoxy-1,2-dihydroquinoline to examine, in rat hippocampus, the relationship between 5-HT1A receptor occupancy and inhibition of forskolin-stimulated cAMP accumulation. Their findings were consistent with a lack of receptor reserve. Also, 5-HT1A receptors do not universally couple to adenylyl cyclase. Although the activation of postsynaptic 5-HT1A receptors in hippocampus causes inhibition of forskolin-stimulated cAMP accumulation, activation of somatodendritic 5-HT1A receptors in the dorsal raphe does not (Clarke et al., 1996). Although it is not known why 5-HT1A receptors in the dorsal raphe do not couple negatively to adenylyl cyclase, our findings suggest that the density of receptor expression relative to G protein expression may be insufficient.

In our studies of 5-HT1A and 5-HT1B receptors we found that the magnitude of activation of ERK2 was positively correlated with the magnitude of inhibition of forskolin-stimulated cAMP accumulation. However, in previous studies we have demonstrated that inhibition of adenylyl cyclase activity is not required for activation of ERK2 (Cowen et al., 1996). Pretreatment of CHO-1A-3000 cells with the permeable cAMP analogues dibutyryl cAMP or 8-bromo cAMP caused no inhibition of 8-OH-DPAT-stimulated MAP kinase activity. We similarly found that pretreatment with forskolin or IBMX had no effect. Therefore, activation of MAP kinase by 5-HT1A receptors is not the result of decreases in cAMP concentration. Instead, it more likely fits the model described for other Gi-coupled receptors (Touhara et al., 1995), requiring G protein βγ subunits, phosphatidylinositol 3-kinase, Grb2/SOS, and Ras. That there was a correlation in the magnitude of change elicited in ERK2 and adenylyl cyclase activity suggests that both signals may be regulated by the same G protein.

However, it is not currently known which pertussis toxin-sensitive G proteins are required for coupling of 5-HT1A and 5-HT1B receptors to ERK2 and adenylyl cyclase. It is also not known if the receptors use different G proteins. Of the pertussis toxin-sensitive G proteins, CHO-K1 cells express primarily Giα2 and Giα3 (Giα2 > Giα3), but not Giα1, and in some studies have been found to express small amounts of GoαA (Raymond et al., 1993; van Biesen et al., 1996). It is possible that both 5-HT1A and 5-HT1B receptors couple to an identical G protein (or G proteins) but that 5-HT1B receptors couple more effectively. Alternatively, 5-HT1A and 5-HT1B receptors may preferentially use different G proteins. The more effective coupling of 5-HT1B receptors to ERK2 and adenylyl cyclase could then be a result of differences in the efficacy of individual G proteins. The G protein used by 5-HT1B receptors might be more effective in coupling to these signals than that used by 5-HT1B receptors.

Our present study used rodent 5-HT1B receptors to compare the coupling of 5-HT1A and 5-HT1B receptors to cellular signals. The endogenous expression of rodent 5-HT1B receptors in CHO cells allowed studies of endogenous receptors expressed at a very low density, in addition to transfected receptors expressed at a specific higher density. Although there are some differences in receptor pharmacology, the amino acid sequence of rat 5-HT1B receptors is 94% identical to that of human 5-HT1B receptors (Veldman and Bienkowski, 1992). Therefore, rat and murine (98% identical to rat) receptors are often used as models for studying the function of 5-HT1B human receptors. We would expect that human 5-HT1B receptors would be similar to rodent receptors in displaying functional differences compared with human 5-HT1A receptors.

Our results help provide some insight into understanding the significance of expression of multiple subtypes of 5-HT1 receptors in brain. Similar differences in function appear also to be relevant to other receptor subtypes classified within individual families. For example, Flordellis et al. (1995) used stable transfected cell lines expressing similar densities of receptors to demonstrate that the subtypes of α2-adrenergic receptors exhibit differences in coupling to second messengers. α2B- and α2D-adrenergic receptors, but not α2C-adrenergic receptors, were found to couple effectively to MAP kinase. Some differences have also been found in dopamine receptors of the D2 family: D2 and D3 dopamine receptors have been reported to be better than D4 dopamine receptors with respect to inhibiting dopamine release (Tang et al., 1994). In summary, there are accumulating data demonstrating that Gi-coupled receptors are not identical but exhibit functional differences.

Acknowledgements

These studies were supported by an NARSAD Young Investigator Award to D.S.C., a grant from the Foundation of the University of Medicine and Dentistry of New Jersey, and grant GM55145-02 from the National Institutes of Health. We thank Dr. Julie Hensler (University of Texas Health Science Center, San Antonio) for subcloning the 5-HT1A receptor construct into pcDNA3 and Dr. Robert Hamer (University of Medicine and Dentistry of New Jersey-Robert Wood Johnson Medical School) for advice on statistical analyses.

Ancillary