Pharmacological characterization of human recombinant melatonin mt1 and MT2 receptors

Authors


National Institute for Medical Research, The Ridgeway, Mill Hill, London, NW7 1AA.

Abstract

  • We have pharmacologically characterized recombinant human mt1 and MT2 receptors, stably expressed in Chinese hamster ovary cells (CHO-mt1 and CHO-MT2), by measurement of [3H]-melatonin binding and forskolin-stimulated cyclic AMP (cAMP) production.

  • [3H]-melatonin bound to mt1 and MT2 receptors with pKD values of 9.89 and 9.56 and Bmax values of 1.20 and 0.82 pmol mg−1 protein, respectively. Whilst most melatonin receptor agonists had similar affinities for mt1 and MT2 receptors, a number of putative antagonists had substantially higher affinities for MT2 receptors, including luzindole (11 fold), GR128107 (23 fold) and 4-P-PDOT (61 fold).

  • In both CHO-mt1 and CHO-MT2 cells, melatonin inhibited forskolin-stimulated accumulation of cyclic AMP in a concentration-dependent manner (pIC50 9.53 and 9.74, respectively) causing 83 and 64% inhibition of cyclic AMP production at 100 nM, respectively. The potencies of a range of melatonin receptor agonists were determined. At MT2 receptors, melatonin, 2-iodomelatonin and 6-chloromelatonin were essentially equipotent, whilst at the mt1 receptor these agonists gave the rank order of potency of 2-iodomelatonin>melatonin>6-chloromelatonin.

  • In both CHO-mt1 and CHO-MT2 cells, melatonin-induced inhibition of forskolin-stimulated cyclic AMP production was antagonized in a concentration-dependent manner by the melatonin receptor antagonist luzindole, with pA2 values of 5.75 and 7.64, respectively. Melatonin-mediated responses were abolished by pre-treatment of cells with pertussis toxin, consistent with activation of Gi/Go G-proteins.

  • This is the first report of the use of [3H]-melatonin for the characterization of recombinant mt1 and MT2 receptors. Our results demonstrate that these receptor subtypes have distinct pharmacological profiles.

British Journal of Pharmacology (2000) 129, 877–886; doi:10.1038/sj.bjp.0703130

Abbreviations:
cAMP

cyclic adenosine 3′:5′ monophosphate

CHO

Chinese hamster ovary

5-HT

5-hydroxytryptamine

MT

melatonin

8-M-PDOT

8-methoxy-2-propionamidotetraline

4-P-PDOT

4-phenyl-2-propionamidotetraline

Introduction

Melatonin is the principle hormone of the pineal gland and is believed to have a central role in the regulation of the mammalian circadian system and reproductive function in seasonally breeding animals (for a review see Arendt, 1995). To date, two human melatonin receptors have been cloned, termed mt1 and MT2, and have been shown to be seven transmembrane G-protein coupled receptors (Reppert et al., 1994; 1995). When expressed in immortalized mammalian cell lines, both mt1 and MT2 receptors bind melatonin with high affinity and couple to the inhibition of adenylate cyclase (Reppert et al., 1994; 1995).

Radioligand binding studies using 2-[125I]-iodomelatonin have identified high affinity binding sites in rat and human suprachiasmatic nucleus, sheep pars tuberalis and chicken retina (for a review see Morgan et al., 1994). In situ hybridization studies have shown the presence of mt1 mRNA in the human suprachiasmatic nucleus (Weaver & Reppert, 1996), whilst MT2 mRNA is found in human retina and brain (Reppert et al., 1995). Subtype selective agonists and antagonists would greatly assist the elucidation of the relative contributions of melatonin receptor subtypes to the physiological actions of melatonin, such as its effects in the suprachiasmatic nucleus to regulate the circadian system. The expression of mt1 and MT2 receptors as isolated, single populations of receptors allows the investigation of melatonin receptor pharmacology and the search for subtype selective ligands which cannot be achieved in physiological systems where mixed populations of melatonin receptors may exist.

Thus, we have stably expressed mt1 and MT2 receptors in Chinese hamster ovary (CHO) cells and characterized the pharmacology of these receptors using both radioligand binding assays and measurement of the inhibition of forskolin-stimulated cyclic adenosine 3′:5′ monophosphate (cyclic AMP). Previous melatonin receptor binding studies have generally employed 2-[125I]-iodomelatonin as a radioligand. However, high receptor expression levels have enabled us to develop a radioligand binding assay utilising [3H]-melatonin. This radioligand has advantages over 2-[125I]-iodomelatonin, as it is safer, more economical to use, and is chemically identical to the endogenous hormone. We have used a number of melatonin receptor agonists, including the indolene GR196429 (Beresford et al, 1998a) and the napthalenic compound S-20098 (Yous et al., 1992) to characterize CHO-mt1 and CHO-MT2 receptors. In addition, a number of putative melatonin receptor partial agonists and antagonists with selectivity for the MT2 receptor were evaluated, including GR128107, 5-MCA-NAT, 8-M-PDOT and 4-P-PDOT (Dubocovich et al., 1997). The most selective of these compounds, the amidotetraline 4-P-PDOT, was compared to the previously characterized melatonin receptor antagonist luzindole (Dubocovich, 1988) for its ability to antagonize melatonin mediated inhibition of forskolin-stimulated cyclic AMP production. Preliminary accounts of this work have already been presented (Browning et al., 1997; 1998).

The nomenclature and classification of melatonin receptors used here was recently approved by the Nomenclature Committee of the International Union of Pharmacology (Dubocovich et al., 1998a). The denomination ‘mt1’ corresponds to that of the recombinant receptor previously termed Mel1a. MT2 refers to native functional receptors with pharmacological characteristics similar to that of the recombinant receptor mt2, previously termed Mel1B. MT3 corresponds to the pharmacologically defined melatonin receptor subtype, with unknown gene sequence, previously referred to as ML2.

Methods

Generation of CHO-mt1 and CHO-MT2 cells

Clones representing the sequence of the human mt1 receptor were amplified using degenerate primers based on the sequence of the Xenopus laevis melatonin receptor (Ebisawa et al., 1994). These clones defined the genomic sequence flanking the intron within the coding sequence of the receptor and enabled the coding exons to be amplified from genomic DNA independently and reassembled in frame using an engineered site. Cloning was in pBluescript (Stratagene). The human MT2 sequence was cloned in a similar manner using the sequence described by Reppert and co-workers (1995). The nucleotide sequences obtained encoded for the same amino acid sequences as described by Reppert et al. (1994; 1995). The sequences encoding the receptor were cloned into the mammalian expression vector pcDNA3 (Invitrogen) and introduced into CHO cells by conventional calcium phosphate precipitation techniques, which were then placed under G418 selection (1 mg ml−1). Several G418 resistant mt1 and MT2 cell lines were selected for [3H]-melatonin saturation binding studies and measurement of melatonin-mediated inhibition of forskolin-stimulated cyclic AMP production. The cell-lines which gave the highest levels of [3H]-melatonin binding and greatest melatonin-induced inhibition of cyclic AMP accumulation were chosen for further [3H]-melatonin radioligand binding and cyclic AMP studies, respectively.

Preparation of cell membranes

CHO cells stably expressing human mt1 or MT2 receptors were maintained in DMEM-F12 medium, supplemented with 10% (v v−1) foetal calf serum, 2 mML-glutamine, 1 mg ml−1 G418 and 400 μg ml−1 hygromycin. Cells were grown at 37°C in 5% CO2 in air. Membranes were prepared by harvesting CHO cells using Hanks balanced salt solution, containing EDTA (5 mM). The suspension was homogenized and centrifuged at 4500×g for 35 min. The pellet was re-homogenized, resuspended in (mM): Tris-HCl buffer 50, containing MgCl2 2, EDTA 1, 0.1% ascorbic acid, pH 7.4, and aliquots (∼2 mg protein ml−1) stored at −80°C until use.

Radioligand binding assays

The assay was performed in 96-well plates in a final assay volume of 500 μl. Drugs and [3H]-melatonin (0.3 nM) or 2-[125I]-iodomelatonin (50 pM) were incubated with membranes (50 μg protein ml−1) for 120 min at 37°C in a buffer comprising (mM): Tris-HCl 50, MgCl2 2, EDTA 1, 0.1% ascorbic acid, pH 7.4. At the concentration of protein used, less than 10% of added radioligand was bound to membrane (data not shown). Non-specific binding was defined with melatonin (1 μM). For association studies, membranes (50 μg ml−1) were added to [3H]-melatonin (0.3 nM) and incubated for increasing periods of time (2–180 min). For dissociation studies, membranes were incubated with [3H]-melatonin (0.3 nM) for 2 h prior to the addition of cold melatonin (1 μM) to initiate dissociation, and then incubated for increasing periods of time (0–240 min). Bound radioactivity was separated by rapid filtration through GF/B filter paper using a Brandel or Wallac cell harvester and filters were washed with 4×1 ml Tris-HCl (50 mM; pH 7.4). Radioactivity was measured by liquid scintillation spectrometry.

cyclic AMP assays

Cells were cultured as described above. The assay was performed in 96-well plates in a final assay volume of 200 μl. Confluent CHO-mt1 or CHO-MT2 cells were incubated at 37°C with DMEM-F12, containing 300 μM isobutylmethylxanthine (IBMX) to inhibit phosphodiesterase activity. Following 60 min incubation, agonist (0.01 pM–100 μM) was added. Sixty minutes later, forskolin (30 μM) was added and cells were incubated for a further 15 min. For antagonist experiments, melatonin was co-incubated with luzindole (0.1–100 μM) for 60 min prior to addition of forskolin. The reaction was terminated by removal of media and addition of ice-cold ethanol (100 μl) for 30 min at 4°C. Ethanol samples were evaporated to dryness and cyclic AMP concentrations determined by [125I]-cyclic AMP scintillation proximity assay (Amersham).

Data analysis

Saturation binding experiments were analysed by fitting the sum of a hyperbola and a straight line to the total binding data to generate pKD and Bmax values. Competition binding data were analysed by fitting with a four-parameter logistic equation. pIC50 values were converted to pKi values using the equation of Cheng & Prusoff (1973). Association kinetic data were analysed by fitting total binding data to the equation B=C+Bmax.(1–exp-K*t), where B=binding at time t, C=non-specific binding and k=observed association rate constant. Dissociation kinetic data were analysed by fitting total binding data to the equation B=C+Bmax.exp-K*t, where k=dissociation rate constant.

Cyclic AMP data were fitted with a four parameter logistic equation to determine pIC50 values and Hill coefficients. Drug responses were expressed as percentage inhibition of forskolin-stimulated cyclic AMP. The potency ratio was defined as the ratio of the EC50 of the drug relative to that of melatonin determined in the same experiment. The pA2 values for luzindole were determined using a modified form of the Schild equation (Black et al., 1985; Lew & Angus, 1995).

All data are expressed as arithmetic mean±s.e.mean. Statistical comparisons were made using two-tailed Student's t-test. For luzindole studies, the upper and lower asymptotes and Hill slopes of the melatonin concentration-effect curves were compared using one-way analysis of variance (ANOVA) followed by Dunnett's test for significance. Where data are presented as the mean of separate experiments, curves were generated from the mean of the individual curve fitting parameter estimates.

Drugs

GR196429, S20098, 5-MCA-NAT (GR135531), GR128107 and luzindole were synthesized by Medicinal Chemistry, Glaxo Wellcome. Melatonin, 6-hydroxymelatonin, 6-chloromelatonin, N-acetyl-5-HT, 5-HT, forskolin and IBMX were supplied by Sigma. 2-iodomelatonin, 4-phenyl-2-propionamidotetraline (4-P-PDOT) and 8-methoxy-2-propionamidotetraline (8-M-PDOT) were purchased from RBI. [3H]-melatonin (85 Cimmol−1) and 2-[125I]-iodomelatonin (∼2000 Cimmol-1) were supplied by Amersham.

Results

Kinetics and saturation of [3H]-melatonin binding to mt1 and MT2 receptors

At 37°C the specific binding of [3H]-melatonin (0.4 nM) to mt1 receptors reached equilibrium within 15–30 min (Figure 1A; t1/2 3.9±1.2 min; observed association rate constant (kobs) 3.89e8M−1.min−1; n=3) and remained stable for more than 360 min. Binding of [3H]-melatonin (0.4 nM) to MT2 receptors occurred more slowly, taking 60–90 min to reach equilibrium (Figure 1B; t1/2 14±2 min; kobs 9.18e7M−1.min−1; n=3). For both receptors, an incubation time of 120 min was used routinely. Specific [3H]-melatonin binding (0.4 nM) to mt1 receptors dissociated rapidly to non-specific binding levels (Figure 1C), with a dissociation rate constant (k−1) of 0.056±0.003 min−1 and half-time of 12.5±0.8 min (n=3). In contrast, binding of [3H]-melatonin (0.4 nM) to MT2 receptors dissociated more slowly (Figure 1D), with a dissociation rate constant (k−1) of 0.011±0.002 min−1 and half-time of 64±8 min (n=3). Whilst only 3.6±0.6% (n=3) of [3H]-melatonin remained specifically bound to mt1 receptors after 240 min, 22±2% (n=3) of the specific [3H]-melatonin remained bound to MT2 receptors, which was significantly different from non-specific binding (P<0.05). From the kinetic data, pKD values of 9.78±0.16 and 9.92±0.03 (n=3) for mt1 and MT2 receptors, respectively, were determined.

Figure 1.

Time-courses of association (A, B) and dissociation (C, D) of [3H]-melatonin binding to the human recombinant mt1 (A, C) and MT2 (B, D) receptors. Data are the mean of duplicate points and representative of three similar experiments. For association experiments, upper asymptote (mean±s.e.mean)=0.37±0.05 pmol mg−1 (mt1) and 0.39±0.01 pmol mg−1 (MT2). For dissociation experiments, binding at time t0 (mean±s.e.mean)=0.38±0.03 pmol mg−1 (mt1) and 0.40±0.03 pmol mg−1 (MT2).

Saturation binding studies indicated that [3H]-melatonin bound in a specific, concentration-dependent and saturable manner to both mt1 and MT2 receptors (Figure 2A,B), with pKD values of 9.89±0.13 and 9.56±0.03, respectively (n=3). Bmax values were 1.20±0.10 and 0.82±0.06 pmol mg−1 protein, respectively. Hill coefficients were 1.02±0.13 and 1.06±0.06, respectively, suggesting interactions with single populations of binding sites. In the cell lines selected for cyclic AMP assays, [3H]-melatonin saturation binding studies yielded Bmax values of 0.39±0.07 and 0.23±0.06 pmoles mg−1 protein, and pKD values of 9.43±0.06 and 9.04±0.18 at mt1 and MT2 receptors respectively (n=3).

Figure 2.

[3H]-melatonin saturation binding to human recombinant mt1 (A) and MT2 (MT2) receptors. Data are mean of duplicate points and representative of three separate experiments.

Comparison of [3H]-melatonin and 2-[125I]-iodomelatonin binding to mt1 receptors

There was excellent agreement between the abilities of a range of melatonin analogues to compete for [3H]-melatonin and 2-[125I]-iodomelatonin binding to mt1 receptors (Table 1). Linear regression analysis yielded a correlation coefficient of 0.99 and a slope of 1.01±0.05 (n=8).

Table 1. Affinities of melatonin analogues to compete for [3H]-melatonin and 2-[125I]-iodomelatonin binding to human recombinant mt1 and MT2 receptorsThumbnail image of

Pharmacological characterization of [3H]-melatonin binding to mt1 and MT2 receptors

The abilities of fourteen melatonin analogues to compete for [3H]-melatonin binding to mt1 and MT2 receptors were determined (Table 1; Figure 3A,B). All Hill coefficients were not significantly different from unity, with the exception of 2-iodomelatonin at mt1 receptors (1.24±0.07, P<0.05). Most melatonin receptor agonists bound with similar affinities to both receptor subtypes (linear regression analysis yielded a correlation coefficient of 0.87 and a slope of 0.71±0.12). Greatest selectivity was exhibited by 2-iodomelatonin, which had 5 fold higher affinity for mt1 receptors, and 6-chloromelatonin, which had 5 fold higher affinity for MT2 receptors. A number of putative partial and silent receptor antagonists had substantially higher affinities for MT2 than mt1 receptors. Greatest selectivity was demonstrated by 4-P-PDOT, which had a 61 fold higher affinity for MT2 compared to mt1 receptors.

Figure 3.

Competition for [3H]-melatonin binding at human recombinant mt1 (A) and MT2 (B) receptors. Data are mean of duplicate points and representative of at least three separate experiments. Binding in the absence of competitor (mean±s.e.mean)=0.42±0.02 pmol mg−1 (mt1) and 0.35±0.02 pmol mg−1 (MT2).

Inhibition of forskolin-stimulated cyclic AMP in CHO-mt1 and CHO-MT2 cells by melatonin

Melatonin produced a potent, monophasic and concentration-dependent inhibition of forskolin-stimulated cyclic AMP in both CHO-mt1 and CHO-MT2 cells, with pEC50 values of 9.53±0.16 and 9.74±0.05 and maximum inhibitions of 83±4 and 64±3%, respectively (Table 2; Figure 4A,B). Melatonin had no effect on basal levels of cyclic AMP in control or pertussis toxin treated cells, or on forskolin-stimulated cyclic AMP levels in mock transfected CHO cells (data not shown). Overnight incubation with pertussis toxin (100 ng ml−1) completely abolished responses to melatonin in both CHO-mt1 and CHO-MT2 cells (Figure 4A,B). In CHO-mt1 cells, pertussis toxin incubation revealed a small stimulatory response to melatonin in the presence of forskolin, with a pEC50 value of 8.37±0.17 and maximum response of 26±3% (0.1 μM) of forskolin-stimulated cyclic AMP (Figure 4A). This stimulatory response of melatonin was not observed in pertussis toxin-treated CHO-MT2 cells (Figure 4B).

Table 2. Potencies of melatonin receptor agonists to inhibit forskolin stimulated cyclic AMP accumulation in CHO-mt1 and CHO-MT2 cellsThumbnail image of
Figure 4.

The effect of pertussis toxin (100 ng ml−1) on melatonin-induced inhibition of forskolin-stimulated cyclic AMP production in CHO-mt1 (A) and CHO-MT2 (B) cells. Data are mean±s.e.mean of three separate experiments. Mean basal cyclic AMP levels were 0.30 pmol well−1 (range 0.18–0.35); forskolin-stimulated cyclic AMP levels were 27 pmol well−1 (range 14–40). Similar values were obtained in subsequent experiments.

Pharmacological characterization of functional responses in CHO-mt1 and CHO-MT2 cells

The inhibitory effect of melatonin was mimicked by a range of indolic and non-indolic melatonin analogues (Figure 5A,B; Table 2). At the mt1 receptor, the rank order of potency was 2 - iodomelatonin>melatonin=S20098>6-chloromelatonin= GR196429>6-hydroxymelatonin> N-acetyl-5-HT, whilst at the MT2 receptor the rank order was S20098=2-iodomelatonin > melatonin = 6-chloromelatonin∼thinsp;4 GR196429∼thinsp;4 6 - hydroxymelatonin > N - acetyl - 5 - HT.  At both mt1 and MT2 receptors all compounds tested were full agonists with respect to melatonin, with the exception of 6-hydroxymelatonin at the MT2 receptor, which was a partial agonist with an intrinsic activity of 0.82±0.05 (P<0.05 vs melatonin, n=4). 2-iodomelatonin showed the greatest degree of mt1 selectivity relative to melatonin, being 30 fold more potent than melatonin to inhibit forskolin-stimulated cyclic AMP in CHO-mt1 cells, whilst only 2 fold more potent than melatonin in CHO-MT2 cells. 6-chloromelatonin showed the greatest degree of MT2 receptor selectivity relative to melatonin. In CHO-mt1 cells, 6-chloromelatonin was 25 fold less potent than melatonin, whilst in CHO-MT2 cells it was equipotent with melatonin. All other agonists had similar potency ratios, with respect to melatonin, at both mt1 and MT2 receptors.

Figure 5.

Agonist concentration-effect curves for inhibition of forskolin-stimulated cyclic AMP in CHO-mt1 (A) and CHO-MT2 cells. The data are mean of 3–11 separate experiments.

Antagonist studies in CHO-mt1 and CHO-MT2 cells

The putative melatonin receptor antagonist luzindole (0.1–100 μM) produced a concentration-dependent rightward shift of the melatonin concentration-effect curves in both CHO-mt1 and CHO-MT2 cells (Figure 6A,B). Non-linear regression analyses yielded pA2 values of 5.75±0.10 and 7.64±0.11 for mt1 and MT2 respectively (n=3–4). Schild slope parameters for luzindole were 1.00±0.10 and 0.94±0.07 for mt1 and MT2 respectively, which were not significantly different to unity (P>0.05). In CHO-mt1 cells luzindole had no significant effect (P>0.05) on Hill slopes or maximum responses to melatonin (data not shown). However, luzindole caused a concentration-dependent enhancement of the lower asymptote of the melatonin concentration-effect curve of 95±20% (P<0.05) at 30 μM, and 134±36% (P<0.01) at 100 μM, of forskolin-stimulated cyclic AMP (Figure 6A). Luzindole alone produced a small and variable enhancement of forskolin-stimulated cyclic AMP, which was not of the same magnitude as that seen in the presence of melatonin (Figure 6A and 7A). In contrast, in CHO-MT2 cells, luzindole was without effect on forskolin-stimulated cyclic AMP, and had no effect on the maximum response to melatonin (Figure 6B and 7B). However, luzindole reduced the Hill slopes for the melatonin concentration-effect curves in a concentration-dependent manner (e.g. nH=1.65± 0.33 and 0.89±0.19, control and 10 μM luzindole respectively; P<0.05). For these reasons, pA2 rather than pKB values have been quoted.

Figure 6.

Antagonism by luzindole of melatonin-induced inhibition of forskolin-stimulated cyclic AMP production in CHO-mt1 (A) and CHO-MT2 (B) cells. Data are mean±s.e.mean of 3–4 separate experiments. Inset: Schild plots depicting antagonism of melatonin responses by luzindole. The line through the data was plotted using the parameters obtained from the analysis described in the Methods.

Figure 7.

The effect of melatonin, luzindole and 4-P-PDOT on forskolin-stimulated cyclic AMP production in CHO-mt1 (A) and CHO-MT2 (B) cells. Data are mean±s.e.mean of 3–4 separate experiments.

In CHO-mt1 cells the amidotetraline 4-P-PDOT (10 μM) had no effect on forskolin-stimulated cyclic AMP levels, either alone (Figure 7A), or in the presence of melatonin (data not shown). In contrast, in CHO-MT2 cells, 4-P-PDOT was an agonist, producing a concentration-dependent inhibition of forskolin stimulated cyclic AMP, with a pEC50 value of 8.72±0.29 and intrinsic activity of 0.86±0.15 (n=3; Figure 7B). This degree of agonism precluded the determination of a pA2 value for 4-P-PDOT at MT2 receptors.

Discussion

This study has described the pharmacology of human recombinant melatonin mt1 and MT2 receptors stably expressed in CHO cells, using a number of indolic and non-indolic analogues of melatonin. Human recombinant mt1 and MT2 receptors were expressed in identical cell lines using the same expression vectors and achieved similar receptor expression levels. This has allowed the pharmacology of each receptor to be compared in systems with identical cellular backgrounds.

The affinities of melatonin analogues at the mt1 receptor measured using [3H]-melatonin are in good agreement with affinities obtained using 2-[125I]-iodomelatonin. This is the first report of the use of [3H]-melatonin for the pharmacological characterization of melatonin receptor subtypes and validates the use of this radioligand as an alternative to 2-[125I]-iodomelatonin. Similarly, Kennaway et al. (1994) showed comparable pharmacological profiles for the binding of these two radioligands to chicken brain. In addition to its safety benefits, [3H]-melatonin is chemically identical to the endogenous hormone and is therefore the ligand of choice for the generation of structure-activity relationships. Indeed, Kennaway et al. (1994) concluded that 2-[125I]-iodomelatonin may not be an appropriate ligand for melatonin binding studies because of its disproportionately slow dissociation rate.

Studies of association binding kinetics at mt1 receptors showed that [3H]-melatonin associated rapidly and reached equilibrium after 30 min. At the MT2 receptor, [3H]-melatonin associated more slowly and reached equilibrium after approximately 60 min. Interestingly, [3H]-melatonin dissociated from the MT2 receptor over a much longer period of time, and failed to reach non-specific binding levels, with 22% of specifically bound radioligand still present after 240 min. Our data suggests that, over the time frame of the experiment, a component of the [3H]-melatonin binding to the MT2 receptor may be irreversible. This result may be explained using a model of multiple receptor states. Long term exposure of receptor with radioligand may ‘lock’ a proportion of occupied receptor into a super-high affinity state which may appear irreversible over the time frame of the experiment. This behaviour of slowly dissociating radioligands has been reported for a number of 7-transmembrane, G-protein coupled receptors and has been shown to be sensitive to the presence of guanine nucleotides, suggesting the presence of high affinity, agonist occupied, G-protein linked receptors (Cohen et al., 1996). Interestingly, whilst the pKD of 9.78 for [3H]-melatonin at the mt1 receptor determined from kinetic studies was in good agreement with the value of 9.89 estimated from equilibrium saturation binding, at the MT2 receptor the pKD of 9.92 determined from kinetic studies was nearly half a log unit higher than the pKD of 9.56 estimated from equilibrium saturation binding. The irreversible component of the binding of [3H]-melatonin to the MT2 receptor may explain this discrepancy.

The affinities of a range of melatonin analogues at mt1 and MT2 receptors are in broad agreement with those obtained by other groups (Dubocovich et al., 1997; Reppert et al., 1995). Most ligands were either non-selective or showed some degree of MT2 selectivity. Ligands which exhibited the greatest degree of selectivity were 2-iodomelatonin, having 5 fold higher affinity at the mt1 receptor, and the amidotetraline derivative, 4-P-PDOT, with 60 fold higher affinity for the MT2 receptor. Similarly, Dubocovich and co-workers (1997) have reported that 4-P-PDOT had 300 fold higher affinity at the MT2 receptor.

Functional studies demonstrated that melatonin receptor activation resulted in the inhibition of forskolin-stimulated cyclic AMP production. The abrogation of the response to melatonin following pre-treatment with pertussis toxin indicates that this response is mediated via Gi/Go G-proteins, confirming previous reports using both recombinant human receptor systems (Reppert et al., 1994; 1995) and sheep pars tuberalis (Morgan et al., 1990). In CHO-mt1 cells, pertussis toxin treatment revealed a small enhancement by melatonin of forskolin-stimulated cyclic AMP, suggesting that the mt1 receptor can promiscuously couple to G-proteins other than Gi/Go, presumably Gs. Promiscuous coupling of melatonin receptors has been previously reported for cloned Xenopus receptors expressed in human embryonic kidney 293 cells (Yung et al., 1995) and endogenous receptors in sheep pars tuberalis (Morgan et al., 1990; 1995). Interestingly, the stimulatory response to melatonin following pertussis toxin treatment was not observed in CHO-MT2 cells, indicating a difference in the second-messenger coupling characteristics of these receptors.

In both CHO-mt1 and CHO-MT2 cells, luzindole antagonised melatonin responses in a concentration-dependent manner, with approximately 80 fold higher affinity at the MT2 receptor. This is the first direct comparison of the affinity of luzindole for mt1 and MT2 receptors measured in a functional assay and confirms previous reports that luzindole has selectivity for MT2 receptors (Dubocovich et al., 1997). However, the pA2 value of 5.75 for luzindole at the mt1 receptor is somewhat lower than the values of 7.2 and 7.3 determined by Beresford and co-workers (1998b) and Witt-Enderby & Dubocovich (1996) respectively. It is not clear the extent to which the pA2 value for luzindole at the mt1 receptor determined in this study is compromized by its complex action in this system. In CHO-mt1, but not in CHO-MT2 cells, luzindole alone produced a small increase in forskolin-stimulated cyclic AMP, suggesting that luzindole may act as an inverse agonist at the mt1 receptor. In support of this hypothesis, Dubocovich & Masana (1998) have shown that the affinity of luzindole for the mt1 receptor increased in the presence of GTP, a property which is displayed by inverse agonists. Interestingly, in the presence of low concentrations of melatonin, luzindole enhanced forskolin-stimulated cyclic AMP, indicating that the inverse agonist response can be augmented in the presence of melatonin. Alternatively, as our experiments with pertussis toxin suggest that the mt1 receptor can promiscuously couple to non Gi/Go G-proteins, luzindole may potentiate the signalling of melatonin through this alternative G-protein pathway. Further studies are clearly required to elucidate the exact nature of the effects of luzindole and its interaction with melatonin at the mt1 receptor.

In CHO-mt1 cells, the putative MT2 receptor selective antagonist 4-P-PDOT, at concentrations up to 10 μM, had no effect on forskolin-stimulated cyclic AMP levels alone, and did not affect melatonin concentration-effect curves. In contrast, in CHO MT2 cells, 4-P-PDOT was a partial agonist with respect to melatonin, producing a concentration-dependent inhibition of forskolin-stimulated cyclic AMP accumulation with a pEC50 of 8.7 and intrinsic activity of 0.87. The intrinsic activity of 4-P-PDOT at the MT2 receptor precluded the determination of a pA2 value for this compound. Whilst this result supports the claim that 4-P-PDOT is a MT2 receptor selective compound (Dubocovich et al., 1997), 4-P-PDOT did not behave as a neutral antagonist in our assay. In agreement, Nonno and co-workers (1999) recently reported that 4-P-PDOT was a partial agonist (intrinsic activity of 0.37) at the MT2 receptor in a [35S]-GTPγS binding assay, a functional measure of receptor activation. In contrast, 4-P-PDOT has previously been shown to be a silent, simple competitive antagonist with high affinity at the presynaptic MT2 receptor mediating the inhibition of electrically-evoked dopamine release in rabbit retina (Dubocovich et al., 1997). The lack of agonism in the rabbit retina may be explained by the lower level of receptor expression in this tissue (8.5 fmol mg−1 protein; Dubocovich, 1995), compared with 0.2 pmol mg−1 protein in our CHO-MT2 cells and 0.5 pmol mg−1 in the NIH3T3 cells used by Nonno et al. (1999). It is well established that a reduction in receptor number may reduce the sensitivity of a tissue to low efficacy agonists (Furchgott, 1966). Therefore, the rabbit retina may not reveal the agonist activity of low efficacy compounds. Interestingly, it has recently been demonstrated that 4-P-PDOT is a silent antagonist of melatonin responses in an animal model of circadian rhythms (Dubocovich et al., 1998b).

While antagonist affinity estimates provide a theoretically robust method of classifying and characterizing G-protein coupled receptors, functional data from agonist experiments can also be used to characterize receptors (Leff, 1987). Estimates of agonist potency alone are of little value in receptor classification, as agonist potency is dependent on both receptor density and transduction efficiency (Kenakin & Beek, 1980). However, the potency of an agonist, when expressed relative to the potency of a reference agonist, provides a robust parameter for receptor classification, since this value will be tissue-independent for full agonists. Therefore, agonists that yield different potency ratios with respect to melatonin can be used to discriminate between mt1 and MT2 receptors. 2-iodomelatonin was 30 fold more potent than melatonin in CHO-mt1 cells, but had a similar potency to melatonin in CHO-MT2 cells. Conversely, 6-chloromelatonin was 25 fold less potent than melatonin in CHO-mt1 cells, whilst being equipotent with melatonin in CHO-MT2 cells. At MT2 receptors, melatonin, 2-iodomelatonin and 6-chloromelatonin are essentially equipotent, whilst at the mt1 receptor these agonists have the rank order of potency of 2-iodomelatonin>melatonin>6-chloromelatonin. Thus, in addition to antagonist data, these three agonists may be of use in classifying melatonin receptors.

In conclusion, we have expressed human recombinant melatonin receptors in CHO cells using identical expression systems and have characterized these receptors using [3H]-melatonin binding assays and measurement of forskolin stimulated cyclic AMP accumulation. Luzindole is an antagonist at melatonin receptors, with higher affinity at MT2 receptors. In addition to the use of antagonists, the rank order of potency of melatonin, 2-iodomelatonin, and 6-chloromelatonin can be used to classify melatonin receptors.

Acknowledgments

We thank Dr Stephen Foord and Jason Brown for providing mt1 and MT2 receptor cDNAs, Kerri Cartwright for expert technical assistance and Dr David Hall for advice during the preparation of the manuscript.

Ancillary