Convergence of Melatonin and Serotonin (5-HT) Signaling at MT₂/5-HT_2C Receptor Heteromers

Background: There is cross-talk between serotonin and melatonin hormones.

Results: There is evidence for unidirectional transactivation and a heteromer-specific signaling profile for formation of functional melatonin MT₂ and serotonin 5-HT_2C receptor heteromers.

Conclusion: A new potential target of the antidepressant agomelatine is identified.

Significance: The importance of binding of multitarget drugs to GPCR heteromers in psychiatric disorders is demonstrated.

Abstract

Inasmuch as the neurohormone melatonin is synthetically derived from serotonin (5-HT), a close interrelationship between both has long been suspected. The present study reveals a hitherto unrecognized cross-talk mediated via physical association of melatonin MT₂ and 5-HT_2C receptors into functional heteromers. This is of particular interest in light of the “synergistic” melatonin agonist/5-HT_2C antagonist profile of the novel antidepressant agomelatine. A suite of co-immunoprecipitation, bioluminescence resonance energy transfer, and pharmacological techniques was exploited to demonstrate formation of functional MT₂ and 5-HT_2C receptor heteromers both in transfected cells and in human cortex and hippocampus. MT₂/5-HT_2C heteromers amplified the 5-HT-mediated G_q/phospholipase C response and triggered melatonin-induced unidirectional transactivation of the 5-HT_2C protomer of MT₂/5-HT_2C heteromers. Pharmacological studies revealed distinct functional properties for agomelatine, which shows “biased signaling.” These observations demonstrate the existence of functionally unique MT₂/5-HT_2C heteromers and suggest that the antidepressant agomelatine has a distinctive profile at these sites potentially involved in its therapeutic effects on major depression and generalized anxiety disorder. Finally, MT₂/5-HT_2C heteromers provide a new strategy for the discovery of novel agents for the treatment of psychiatric disorders.

Introduction

The monoamine serotonin (5-HT)² is derived from dietary tryptophan, which is transformed into 5-HT in diverse clusters of neurons in the gut and brain. 5-HT exerts its actions via 14 classes of receptors, which are broadly expressed in peripheral tissues and the central nervous system (1). Conversely, although the neurohormone melatonin is derived from 5-HT, it is mainly produced by the pineal gland in a circadian pattern under the control of hypothalamic nuclei, attaining peak levels during the night. Melatonin binds with high affinity to MT₁ and MT₂ receptors and with moderate affinity to the enzyme quinone reductase 2 (2). Both MT₁ and MT₂ receptors as well as all 14 classes of 5-HT receptor (except 5-HT₃) belong to the G protein-coupled receptor (GPCR) superfamily. Despite structural similarities between melatonin and 5-HT, melatonin does not recognize 5-HT receptors, and 5-HT fails to bind MT₁ or MT₂ receptors. Furthermore, to date, there have been only a few reports of functional cross-talk between melatonergic and serotonergic transmission: for example, melatonin inhibits the ability of 5-HT to phase shift the suprachiasmatic circadian clock (3).

Recent studies have demonstrated other more direct modes of potential functional interaction expressed not only among signaling pathways but also operating directly at the level of GPCRs, which can assemble into heteromeric complexes (4). Such complexes frequently display functional properties distinct from those of the corresponding homomers and may even transduce novel and unique cellular responses. Moreover, several classes of GPCR heteromers have been associated with the pathogenesis and control of CNS disorders like 5-HT_2A and metabotropic glutamate-2 receptor (mGluR2) heteromers in frontal cortex implicated in schizophrenia and in the actions of antipsychotics (5) and limbic dopamine D₁ and D₂ receptor heteromers incriminated in depressed states (6).

To date, the possible existence of heteromeric associations of MT₁ or MT₂ receptor with specific classes of 5-HT receptors has not been evaluated. Their putative existence is of particular interest inasmuch as the clinically proven antidepressant agomelatine, the first to possess a non-monoaminergic component of action, behaves as an agonist at G_i-coupled MT₁ and MT₂ receptors but as a neutral antagonist at G_q/11-coupled 5-HT_2C receptors (7, 8). Intriguingly, although the affinity of agomelatine is substantially lower at 5-HT_2C versus MT₁ and MT₂ in vitro, this apparent difference is much less pronounced in vivo, suggesting that it may exert its actions “synergistically” via these sites. Indeed, both 5-HT_2C and MT₁ and MT₂ receptors are necessary for expression of the antidepressant actions of agomelatine, which cannot be reproduced either by melatonin or by selective 5-HT_2C antagonists alone (9). For example, “synergistical” MT₁, MT₂, and 5-HT_2C receptor-transduced actions of agomelatine may account for its induction of neurogenesis and BDNF synthesis as well as its modulation of glutamate release (for a review, see Ref. 9). In light of the above observations, the present studies explored the potential formation of heteromers between melatonin receptors and 5-HT_2C receptors and specifically examined the functional profile of agomelatine at these sites.

EXPERIMENTAL PROCEDURES

Compounds

All chemicals and ligands were purchased from Sigma-Aldrich with the exception of pertussis toxin, which was purchased from Alexis Biochemicals, and 4-PPDOT, luzindole, and SB242084, which were purchased from Tocris. S20928, S21767, and agomelatine were a gift from the Institut de Recherches Servier (France).

Cell Culture and Transfection

HEK293 cells were grown in complete medium (Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 4.5 g/liter glucose, 100 units/ml penicillin, 0.1 mg/ml streptomycin, and 1 mm glutamine) (Invitrogen). Geneticin (G418) was added at 0.4 mg/ml to culture HEK293 cells stably expressing the MT₂ receptor from the pcDNA3-CMV plasmid containing the neomycin resistance gene. Transient transfections were performed using JetPEI (Polyplus Transfection, France) according to the manufacturer's instructions.

DNA Constructs

The pcDNA3-CMV vectors expressing the human MT₂ receptor, the double brilliance Rluc8-βARR2-YPet sensor, and the MT₁-YFP and the 5-HT_4d-YFP fusion proteins were described previously (10–12). The INI isoform of the human 5-HT_2C coding region was fused at its C terminus with the coding region of Renilla luciferase (5-HT_2C-Rluc) or the yellow fluorescent protein (5-HT_2C-YFP) or at its N terminus with three HA tags (3xHA-5-HT_2C). The 5-HT_2C-S138N-Rluc mutant was obtained by mutagenesis from the 5-HT_2C-Rluc plasmid. All constructs were verified by sequencing.

Immunoprecipitation

For co-immunoprecipitation assays, HEK293 cells were seeded in 10-cm dishes and co-transfected with 4 μg of each indicated plasmid. 48 h after transfection, crude membranes were prepared as described previously (13). Membrane proteins were solubilized with 1% digitonin, and receptors were precipitated with the indicated antibodies (14). Immunoprecipitated proteins were eluted with 4× Laemmli buffer and immunoblotted using the indicated primary antibodies. Immunoreactivity was revealed using secondary antibodies coupled to 680 or 770 nm fluorophores using the LI-COR Odyssey infrared fluorescence scanner (ScienceTec, France). For experiments with the pork plexus choroid, membranes were prepared, and melatonin receptors were labeled with 400 pm 2-[¹²⁵I]iodomelatonin as described previously (15). Receptors were solubilized with 1% digitonin and precipitated with a mixture of three anti-5-HT_2C rabbit antibodies (18461-10656, 18461-10657, and 18003-42961, Genway) or a pool of control preimmune serum, and immunoprecipitated radioactivity was determined.

Immunofluorescence

HeLa cells transiently expressing 6xMyc-MT₂ and 3xHA-5-HT_2C were fixed in phosphate-buffered saline containing 2% paraformaldehyde for 15 min. Cells were permeabilized with 0.1% Triton X-100. Monoclonal anti-Myc antibody 9E10 (Santa Cruz Biotechnology) and polyclonal anti-HA antibody (Cell Signaling Technology) were applied followed by TRITC-tagged anti-mouse and fluorescein isothiocyanate-tagged anti-rabbit antibodies. Cells were examined by fluorescence microscopy.

In-cell Western Experiments

Cell surface expression of 5-HT_2C receptors was evaluated by in-cell Western experiments in intact and Triton X-100-permeabilized HEK293 cells transiently expressing the 3xHA-5-HT_2C construct and stably expressing the MT₂ receptor (5-HT_2C expression) or transiently expressing the 6xMyc-MT₂ receptor (MT₂ expression) as described previously (16) using the rabbit monoclonal anti-HA antibody (1:1000; Cell Signaling Technology), anti-Myc 9E10 antibody (1:500) and the LI-COR Odyssey infrared imaging system.

Bioluminescence Resonance Energy Transfer (BRET) Measurement

For BRET donor saturation curves, HEK293 cells seeded in 12-well plates were transiently transfected with 50 ng of 5-HT_2C-Rluc and 50–1950 ng of the corresponding YFP plasmids. 24 h after transfection, cells were transferred into a 96-well white Optiplate (Perkin Elmer Life Sciences) precoated with 10 μg/ml poly-l-lysine (Sigma) and incubated for another 24 h before BRET measurements. Luminescence and fluorescence were measured simultaneously using the lumino/fluorometer Mithras^TM (Berthold) as described previously (11) using optimized filter settings (Rluc filter, 480 ± 10 nm; YFP filter, 540 ± 20 nm).

Intracellular Signaling Assays

HEK293 cells stably expressing 20–30 fmol of MT₂/mg of protein and transiently expressing or not 20–30 fmol of 5-HT_2C or HEK293 cells transiently expressing only 5-HT_2C receptors were used. Inositol 1-phosphate (IP) levels were determined in cells stimulated with the indicated ligands for 1 h at 37 °C by homogeneous time-resolved fluorescence using the Cisbio IP-One Tb kit according to the manufacturer's instructions. Cyclic AMP levels were determined in cells treated with the indicated ligands for 30 min at room temperature in the absence or presence of 2 μm forskolin by homogeneous time-resolved fluorescence using the Cisbio cAMP femto Tb kit according to the manufacturer's instructions. IP and cAMP measurements were performed in triplicates, and experiments were repeated three to eight times. To determine ERK1/2 phosphorylation levels, cells were stimulated for 2, 5, 10, 15, and 30 min with 100 nm melatonin, and ERK1/2 phosphorylation was determined as described previously (16, 17).

Statistical Analysis

Results were analyzed by PRISM (GraphPad Software). Data are expressed as mean ± S.E. of at least three experiments. Student's t test was applied for statistical analysis.

RESULTS

Physical Interaction between 5-HT_2C and MT₁/MT₂ Receptors

We first carried out co-immunoprecipitation experiments to assess a possible interaction between the human 5-HT_2C and MT₁ and MT₂ receptors. These experiments revealed that 5-HT_2C interacts with MT₁ and MT₂ in HEK293 cells co-expressing the respective receptors (Fig. 1, A and B). To further confirm the existence of these heteromeric complexes and to assess the propensity of heteromer formation, BRET experiments were performed. 5-HT_2C was fused at its C terminus to the energy donor Rluc. C-terminal YFP fusion proteins (5-HT_2C-YFP, MT₁-YFP, MT₂-YFP, and 5-HT_4d-YFP) acted as energy acceptors (Fig. 1C). The expected hyperbolic donor saturation curve, reflecting a specific interaction between BRET donor and acceptor pairs, was observed for all receptor combinations except for the negative control 5-HT_4d-YFP fusion protein for which a quasilinear increase in BRET reflecting a nonspecific interaction due to random collision was observed (18, 19). Determination of BRET₅₀ values, corresponding to the saturation of 50% of BRET donors by BRET acceptors, revealed that the relative propensity of 5-HT_2C to form heteromers with MT₂ (BRET₅₀ = 2.2 ± 0.4) is higher than with MT₁ (BRET₅₀ = 6.9 ± 1.5) or with itself (BRET₅₀ = 8.8 ± 1.5), indicating preferential formation of MT₂/5-HT_2C heteromers. As shown by immunofluorescence staining, 5-HT_2C expressed alone was mainly intracellular, and a significant amount of MT₂ and 5-HT_2C was present at the plasma membrane where both receptors colocalized (Fig. 1D). Formation of MT₂/5-HT_2C heteromers was further suggested by co-immunoprecipitation studies in human cortex and hippocampus, two regions shown previously to express melatonin and 5-HT_2C receptors (20, 21) (data not shown). Heat-inactivated anti-MT₂ antibodies were used as a negative control. Further evidence for the formation of MT₂/5-HT_2C heteromers was obtained from choroid plexus membranes, which are known to express significant amounts of 5-HT_2C receptors. Melatonin receptors were labeled with 2-[¹²⁵I]iodomelatonin, and protein complexes were solubilized and immunoprecipitated with anti-5-HT_2C antibodies. As shown in Fig. 1E, significant amounts of radiolabeled melatonin receptors were precipitated with anti-5-HT_2C antibodies as compared with a mixture of irrelevant control antibodies. Overall, our results indicate that 5-HT_2C specifically interacts with MT₂ in transfected HEK293 cells and in the cortex, hippocampus, and choroid plexus.

FIGURE 1.

Formation of MT₁/5-HT_2C and MT₂/5-HT_2C heteromers. A and B, co-immunoprecipitation of MT₁/5-HT_2C and MT₂/5-HT_2C heteromers. Lysates from HEK293T cells expressing the indicated receptors were immunoprecipitated (IP) with antibodies recognizing Myc and FLAG epitopes, and the presence of co-precipitated 5-HT_2C-YFP was detected with antibodies against GFP (lower Western blot (WB)). Expression of fusion proteins in lysates was assessed by immunoblotting with the indicated antibodies (upper Western blots). Data are representative of three experiments. C, BRET donor saturation curves were determined by co-transfecting a fixed amount of 5-HT_2C-Rluc in the presence of increasing amounts of the indicated YFP fusion proteins in HEK293 cells. Curves were normalized to BRET_max values. The saturation curves were obtained from three independent experiments performed in triplicates. D, immunofluorescence staining of permeabilized HeLa cells showing localization of 3xHA-5-HT_2C (FITC) and Myc-MT₂ (TRITC) expressed alone (upper and middle panels, respectively) or together (lower panels) (scale bar, 17 μm). Data are representative of three experiments. E, immunoprecipitation of melatonin receptors with 5-HT_2C receptors in pork choroid plexus; ¹²⁵I-labeled melatonin receptors were immunoprecipitated from solubilized membranes by anti-5-HT_2C antibodies (Ab) or control (Ctrl) rabbit sera (pool of five preimmune rabbit sera). Data represent the mean ± S.E. (error bars) of three independent experiments (*, p < 0.05). Representative results are shown for A, C, and D; similar results were obtained in two additional experiments.

Melatonin but Not 5-HT Activates the G_i/cAMP Pathway through MT₂/5-HT_2c Heteromers

HEK293 cells expressing equivalent and physiologically relevant levels of MT₂ and 5-HT_2C (20–30 fmol each/mg of protein) either alone or together were treated with forskolin and increasing concentrations of melatonin followed by determination of cAMP levels. Melatonin decreased cAMP levels in cells expressing MT₂ alone as expected and in cells co-expressing MT₂ and 5-HT_2C with the same efficiency (EC₅₀ = 0.76 ± 0.46 versus 0.76 ± 0.39 nm, respectively) and potency (30% reduction) (Fig. 2A and Table 1). Melatonin was without effect in cells expressing 5-HT_2C alone. Stimulation with up to 10 μm 5-HT did not modify cAMP levels in any of the three cell types (Fig. 2B). These results indicate that activation of the MT₂ protomer of the MT₂/5-HT_2C heteromer activates the G_i/cAMP pathway but that activation of the 5-HT_2C protomer is unable to transactivate this pathway.

FIGURE 2.

Effect of melatonin- and 5-HT-promoted activation of MT₂/5-HT_2C heteromers on the G_i/cAMP pathway. HEK293 cells expressing 5-HT_2C (▿), MT₂ (●), or 5-HT_2C and MT₂ (○) receptors were either treated with 2 μm forskolin (Fsk) and increasing concentrations of melatonin (MLT) (A) or increasing concentrations of 5-HT (B). Data represent the mean ± S.E. (error bars) of at least three independent experiments performed in triplicates.

TABLE 1.

Compared potencies and properties of melatonin receptor ligands on two signaling pathways in cells expressing MT₂ or MT₂ and 5-HT_2C receptors

The K_i values of MTR antagonists are derived from data shown in Fig. 6 and are defined as K_i = IC₅₀/1 + (S/X) where S and X represent the agonist concentration and EC₅₀, respectively. ND, not determined.

	EC50 ± S.E.
	Inhibition of cAMP production		Stimulation of inositol phosphate production
	MT2	MT2/5-HT2C	MT2	MT2/5-HT2C
	nm
MTR agonists
MLT	0.76 ± 0.46	0.76 ± 0.39	No effect	69 ± 20
Agomelatine	0.29 ± 0.16	0.89 ± 0.35	No effect	Antagonist
MTR antagonists
Luzindole	Antagonist (Ki = 1.12 ± 0.48 nm)	Antagonist (Ki = 7.73 ± 4.98 nm)	No effect	Full agonist (EC50 = 223 ± 97 nm)
4-PPDOT	Antagonist (Ki = 0.14 ± 0.12 nm)	Antagonist (Ki = 0.04 ± 0.01 nm)	No effect	Partial agonist (EC50 = 2.5 ± 1.7 nm)
S20928	Antagonist (Ki = 4.88 ± 1.5 nm)	Antagonist (Ki = 0.66 ± 0.45 nm)	No effect	Antagonist (Ki = ND)

MT₂ Potentiates 5-HT-induced Signaling by Increasing Cell Surface Expression of 5-HT_2C in the MT₂/5-HT_2C Heteromer

We next investigated the capacity of MT₂/5-HT_2C heteromers to activate the G_q/PLC pathway by monitoring 5-HT-induced IP production. Stimulation of HEK293 cells expressing 5-HT_2C alone resulted in the expected dose-dependent increase in IP production (Fig. 3A). This response was potentiated (∼3-fold) in cells expressing similar quantities of 5-HT_2C receptors, but in the presence of MT₂, the response was of similar potency (EC₅₀ = 21 ± 11 nm (5-HT_2C) versus 68 ± 25 nm (MT₂/5-HT_2C)). 5-HT was ineffective in cells expressing MT₂ alone (Fig. 3A). Amplification of the 5-HT-induced response was G_q/11- but not G_i-dependent as determined by pretreating cells with YM254890 or pertussis toxin inhibitors, respectively (Fig. 3B). Expression of the Gβγ scavenger βARK_Cter had no significant effect (Fig. 3C), confirming the predominant role of G_q/11α proteins in the observed amplification. Pretreatment of cells with 5-HT_2C antagonists (RS102221 and SB242084) or an inverse agonist (SB206553) completely blocked the 5-HT-induced IP production in cells expressing 5-HT_2C alone and in the presence of MT₂ as expected (Fig. 3D). Pretreatment with S20928, a melatonin receptor antagonist, showed no cross-reactivity on 5-HT-induced responses (Fig. 3E).

FIGURE 3.

5-HT-promoted potentiation of the G_q/PLC pathway by the MT₂/5-HT_2C heteromer. A, 5-HT-induced IP production was assessed in HEK293 cells expressing 5-HT_2C (▿), MT₂ (●), or 5-HT_2C and MT₂ (○) receptors. B–E, HEK293 cells expressing 5-HT_2C receptors alone or together with MT₂ receptors were treated with heterotrimeric G protein inhibitors (B), 5-HT_2C receptor ligands (D), or the melatonin receptor ligand S20928 (E) or co-expressed with the βARK_Cter Gβγ scavenger (C). F, cell surface expression of 5-HT_2C and MT₂ receptors in cells expressing 5-HT_2C in the absence and presence of MT₂ receptors measured by in-cell Western experiments (*, p < 0.05; n.s., not significantly different). Data represent the mean ± S.E. (error bars) of at least three independent experiments performed in triplicates. Results that are statistically different compared with 5-HT alone for 5-HT_2C cells (●, p < 0.05; ●●, p < 0.01) and MT₂/5-HT_2C cells (**, p < 0.01; ***, p < 0.001) are indicated.

Amplified 5-HT-induced responses in the context of the MT₂/5-HT_2C heteromer might be explained by increased cell surface expression of 5-HT_2C receptors. In agreement with previous reports, only a minor fraction of 5-HT_2C receptors (13.2 ± 1.1%) was expressed at the cell surface when expressed alone as determined by in-cell Western experiments (Fig. 3F). In cells co-expressing 5-HT_2C and MT₂ receptors, the fraction of cell surface-expressed 5-HT_2C almost doubled (22.7 ± 3.0%), whereas the total amount of 5-HT_2C receptors was not modified. The total amount and the fraction of cell surface-expressed MT₂ remained constant irrespective of the presence or absence of 5-HT_2C. Taken together, our results indicate that MT₂ co-expression potentiates 5-HT_2C receptor signaling in the MT₂/5-HT_2C heteromer by increasing the cell surface expression of 5-HT_2C.

Activation of the MT₂ Protomer Transactivates the PLC Pathway and Improves β-Arrestin Recruitment in the MT₂/5-HT_2C Heteromer

To explore the possible effect of MT₂ activation in the MT₂/5-HT_2C heteromer on the G_q/PLC pathway, we stimulated HEK293 cells expressing MT₂ and 5-HT_2C either alone or together with melatonin and determined IP production. Whereas melatonin had no effect in cells expressing either receptor alone, a dose-dependent increase in IP production was observed in cells co-expressing both receptors with an EC₅₀ of 69 ± 20 nm (Fig. 4A). To verify that melatonin-induced IP production is not due to the signaling cross-talk between MT₂ and any G_q-coupled GPCR, we co-expressed MT₂ with the G_q-coupled M1 muscarinic receptor, which does not form heteromers with MT₂ (data not show). In these cells, melatonin did not increase IP production, indicating that melatonin-induced IP production is specific for the MT₂/5-HT_2C heteromer (Fig. 4B). Functional expression of M1 receptors was shown by the expected increase in IP production in the presence of acetylcholine (Fig. 4B).

FIGURE 4.

Melatonin activates the G_q/PLC pathway in the MT₂/5-HT_2C heteromer. A, melatonin (MLT)-induced IP production was assessed in HEK293 cells expressing 5-HT_2C (▿), MT₂ (●), or 5-HT_2C and MT₂ (○) receptors. B, HEK293 cells co-expressing MT₂ receptors and M1 muscarinic receptors were stimulated with melatonin or acetylcholine, and IP production was measured. The influence of the βARK_Cter Gβγ scavenger (C) or heterotrimeric G protein inhibitors (D) on melatonin-promoted IP production of the MT₂/5-HT_2C heteromer is shown. Data represent the mean ± S.E. (error bars) of at least three independent experiments performed in triplicates. Results that are statistically different compared with melatonin alone for MT₂/5-HT_2C cells (*, p < 0.05) are indicated.

Melatonin-induced IP production in cells co-expressing MT₂ and 5-HT_2C receptors was not affected by the expression of the Gβγ scavenger βARK_Cter (Fig. 4C), but it was partially and completely abrogated by pertussis toxin and YM254890 treatment, respectively (Fig. 4D). This indicates the predominant role of G_q/11α proteins, which can be assisted by the presence of G_iα proteins. This result raises the intriguing possibility that melatonin binding to the MT₂ protomer of the MT₂/5-HT_2C heteromer transactivates the 5-HT_2C protomer, which then activates G_q.

To verify this hypothesis, we evaluated whether the melatonin-induced response depended on the activation state of the 5-HT_2C protomer. Pretreatment of cells with 5-HT_2C neutral antagonists (RS102221 and SB242084) prior to melatonin addition had no effect on melatonin-induced IP production, whereas the inverse agonist SB206553 blocked the effect (Fig. 5A). This is consistent with the notion that the melatonin-induced effect is independent of the occupation of the 5-HT_2C receptor binding site but dependent on the constitutive activity of the 5-HT_2C protomer.

FIGURE 5.

Transactivation of the G_q/PLC pathway by melatonin in the MT₂/5-HT_2C heteromer. A, melatonin (MLT)-induced IP production in the presence of 5-HT_2C ligands in HEK293 cells expressing MT₂ receptors alone or co-expressing 5-HT_2C receptors (*, p < 0.05: melatonin versus melatonin + SB206553 on heteromers). B, 5-HT-induced IP production in cells co-expressing the 5-HT_2C-S138N mutant and MT₂ receptor. C, melatonin-induced transactivation of 5-HT_2C wild-type and 5-HT_2C-S138N mutant receptors in the presence of MT₂. D, BRET donor saturation curves of HEK293 cells expressing MT₂-Rluc and 5-HT_2C-Rluc or 5-HT_2C-S138N-Rluc. Melatonin-induced (E) and 5-HT-induced (F) recruitment of β-arrestin 2 measured by BRET in HEK293 cells expressing 5-HT_2C (▿), MT₂ (●), or 5-HT_2C and MT₂ (○) receptors is shown. Data represent the mean ± S.E. (error bars) of at least three independent experiments performed in triplicates. mBu, milli-BRET units.

To further verify the transactivation hypothesis, we used the 5-HT_2C-S138N mutant, which does not bind 5-HT and exhibits decreased constitutive activity due to decreased coupling to G_q proteins compared with the wild-type receptor (22). The absence of 5-HT-promoted IP production was confirmed in cells expressing 5-HT_2C-S138N and MT₂ (Fig. 5B). When cells co-expressing this mutant and MT₂ receptors were stimulated with melatonin, the anticipated reduction in amplitude and efficiency (EC₅₀ = 550 ± 187 versus 69 ± 20 nm for MT₂/5-HT_2C-S138N and MT₂/5-HT_2C, respectively (p < 0.001)) was observed compared with 5-HT_2C wild-type receptor-expressing cells (Fig. 5C). Decreased activity was not due to less efficient heteromerization with MT₂ as comparable BRET₅₀ values (2.2 ± 0.4 versus 3.5 ± 0.6 for MT₂/5-HT_2C and MT₂/5-HT_2C-S138N, respectively) were observed (Fig. 5D). The decreased response to melatonin in the presence of the 5-HT_2C-S138N mutant further supports the transactivation mechanism in which the melatonin-induced conformational change in the MT₂ protomer is transmitted to the 5-HT_2C protomer and then further downstream to the G_q/PLC pathway.

Apart from interacting with heterotrimeric G proteins, GPCRs are also known to recruit β-arrestins. We therefore studied the ability of melatonin to activate the previously described β-arrestin 2 BRET sensor (11, 23) in cells expressing either MT₂ alone or together with 5-HT_2C. As seen in Fig. 5E, MT₂ alone only weakly activated the BRET sensor with an EC₅₀ of 260 ± 170 nm. In contrast, melatonin potently recruited the sensor in cells co-expressing MT₂ and 5-HT_2C (EC₅₀ = 65 ± 23 nm). No effect was seen in cells expressing 5-HT_2C receptors alone as expected. Stimulation with 5-HT activated the sensor in cells expressing 5-HT_2C irrespective of the presence of MT₂ (Fig. 5F). These results show that melatonin-induced β-arrestin recruitment to MT₂ is potentiated in the MT₂/5-HT_2C heteromer.

Agomelatine Exhibits a Unique Functional Profile at MT₂/5-HT_2C Heteromers

To further explore the functional profile of MT₂/5-HT_2C heteromers, we studied the effect of several synthetic melatonin receptors ligands. S20928, a well know melatonin receptor antagonist (24), antagonized the melatonin-induced inhibition of cAMP levels as expected in cells expressing MT₂ alone (K_i = 4.88 ± 1.50 nm) but also in cells co-expressing MT₂ and 5-HT_2C receptors (K_i = 0.66 ± 0.45 nm) (Table 1 and Fig. 6A). Similarly, IP production was also antagonized by S20928 under both conditions (Fig. 6B), suggesting that S20928 behaves as an antagonist of MT₂/5-HT_2C heteromers for both pathways. Some agonistic activity is seen at micromolar concentrations. In contrast, 4-PPDOT and luzindole, two reported antagonists of MT₂ on the G_i/cAMP pathway (7, 25), showed pathway-biased properties on MT₂/5-HT_2C heteromers as they behaved as full antagonists on the G_i/cAMP pathway (K_i = 0.04 ± 0.01 nm and K_i = 7.73 ± 4.98 nm, respectively) but as a partial (4-PPDOT) or full (luzindole) agonist on the G_q/PLC pathway (EC₅₀ = 2.5 ± 1.7 nm and EC₅₀ = 223 ± 97 nm, respectively) (Fig. 6, A and C, and Table 1). Taken together, these data suggest that S20928 is an antagonist at both pathways with some partial agonistic activity at very high concentrations and that 4-PPDOT and luzindole are G_q/PLC pathway-biased ligands of MT₂/5-HT_2C heteromers.

FIGURE 6.

Biased ligands reveal unique functional profile of MT₂/5-HT_2C heteromers. A, cAMP production in HEK293 cells co-expressing MT₂ and 5-HT_2C receptors in the presence of 2 μm forskolin (Fsk) and different concentrations of melatonin (MLT) or 10 nm melatonin and different concentrations of 4-PPDOT, luzindole, or S20928. B, antagonistic effect of S20928 on melatonin-induced IP production in HEK293 cells expressing MT₂ alone or together with 5-HT_2C receptors. C, IP production in HEK293 cells co-expressing MT₂ and 5-HT_2C receptors in the presence of different concentrations of the indicated ligands. Data represent the mean ± S.E. (error bars) of at least three independent experiments performed in triplicates.

We next studied the effect of agomelatine, the antidepressant with agonistic activity at MT₂ receptors and neutral antagonistic properties at 5-HT_2C receptors (7, 8). In cells expressing MT₂ in the absence or presence of 5-HT_2C receptors, agomelatine behaved as an agonist of the G_i/cAMP pathway (EC₅₀ = 0.29 ± 0.16 nm and EC₅₀ = 0.89 ± 0.35 nm, respectively) (Fig. 7A and Table 1) but as an antagonist for the melatonin-induced activation of the G_q/PLC pathway (Fig. 7, B and C). Agomelatine also antagonized the 5-HT response on this pathway, which is compatible with the properties of this compound (Fig. 7D). This clearly shows that agomelatine has a distinct action profile on MT₂/5-HT_2C heteromers as compared with melatonin and 5-HT.

FIGURE 7.

Effect of agomelatine on the MT₂/5-HT_2C heteromer signaling. Effect of melatonin (MLT) and agomelatine on cAMP (A) and IP (B) production in HEK293 cells co-expressing MT₂ and 5-HT_2C receptors. C, effect of agomelatine on melatonin-promoted IP production in cells expressing MT₂ receptors alone or together with 5-HT_2C receptors. D, effect of agomelatine on 5-HT-promoted IP production in cells expressing 5-HT_2C receptors alone or together with MT₂ receptors. Data represent the mean ± S.E. (error bars) of at least three independent experiments performed in triplicates. Results that are statistically different compared with melatonin (C) or 5-HT (D) alone for MT₂/5-HT_2C cells (*, p < 0.05) are indicated. Fsk, forskolin.

DISCUSSION

We describe here a previously unappreciated dimension of cross-talk between melatonin and 5-HT that is mediated by heteromers of MT₂ and 5-HT_2C receptors. MT₂/5-HT_2C heteromers have unique functional properties and are formed preferentially compared with the corresponding homomers. MT₁ and 5-HT_2C receptors also form heteromers in transfected cells, but we focus herein on MT₂/5-HT_2C heteromers. Within these heteromers, melatonin is able to activate distinct cellular cascades: not only the G_i/cAMP pathway as for MT₂ homomers but also the G_q/PLC pathway by transactivation of the 5-HT_2C protomer. This transactivation was unidirectional and not observed for the MT₂ protomer upon 5-HT stimulation. Whereas melatonin activates both pathways, other ligands have a more restricted profile using either the direct activation or transactivation mode. Interestingly, the clinically active antidepressant agomelatine shows functional properties on MT₂/5-HT_2C heteromers that are biased toward the G_i/cAMP pathway and thus distinct from those of melatonin- and 5-HT_2C-specific antagonists.

GPCR heteromers are indeed increasingly recognized as independent pharmacological entities participating in physiological functions and drug action (4). Ligands such as agomelatine are particularly interesting in this context as they have the potential to bind to both protomers. Previous studies established that agomelatine behaves as an agonist at G_i-coupled MT₁ and MT₂ receptors and as a neutral antagonist at G_q/11-coupled 5-HT_2C receptors (7, 8). Our data indicate that agomelatine preserves these properties in MT₂/5-HT_2C heteromers and behaves as a competitive antagonist on the 5-HT binding site and as an agonist on the melatonin binding site. However, not all effects of melatonin are mimicked by agomelatine because it is unable to transactivate 5-HT_2C receptors. Importantly, these properties clearly distinguish agomelatine from melatonin or any other tested compound including 5-HT_2C receptor antagonists.

Interestingly, the MT₂/5-HT_2C heteromer appears to behave in an asymmetric manner as the properties of the two protomers are differently affected by heteromerization. In the case of the 5-HT_2C protomer, no modification of the signaling profile per se was observed but rather an amplification of known 5-HT-promoted responses most likely due to increased surface expression of the 5-HT_2C receptor. Limited surface expression of 5-HT_2C receptors is in agreement with previous studies showing a high level of constitutive internalization for this receptor (26). In the MT₂ protomer, melatonin stimulation not only activates the G_i/cAMP pathway as for MT₂ homomers but also transactivates the G_q/PLC pathway through the 5-HT_2C protomer in a unidirectional manner (not seen upon 5-HT stimulation). Transactivation between protomers is a unique property of GPCR dimers that has been observed in a limited number of other cases such as the GABA_B receptor, which is an obligatory heteromer composed of two subunits with one subunit binding the ligand and the other activating the G protein (27). Another example of this activation mode is the dopamine D₂ receptors (28). Notably, the present functional characterization allowed us to identify biased ligands. Whereas luzindole and 4-PPDOT behave as agonists in the transactivation mode of the 5-HT_2C protomer by the MT₂ protomer, agomelatine is completely inactive in this mode but still able to fully activate the MT₂ protomer.

Simultaneous activation of G_i- and G_q-dependent signaling by melatonin is a distinctive feature of MT₂/5-HT_2C heteromers compared with the corresponding homomers. Notably, the balance between G_i- and G_q-dependent signaling has been recently suggested to be an important parameter determining the ligand action on GPCR heteromers (29). This has been shown for mGluR2/5-HT_2A heteromers, which are composed of the G_q-coupled 5-HT_2A receptor and the G_i-coupled mGluR2 receptor (5) and for which the balance between G_i- and G_q-dependent signaling predicts the anti- or propsychotic activity of drugs targeting mGluR2 and 5-HT_2A receptors. Antipsychotic drugs have a high G_i/G_q activation ratio regardless of which receptor they target, whereas propsychotic drugs have a low ratio. A similar balance between G_i and G_q activation can be proposed for ligands acting on MT₂/5-HT_2C heteromers. According to our current functional characterization, agomelatine is unique inasmuch as it possesses the highest G_i/G_q activation ratio (activation of G_i pathway and allosteric antagonistic effect on 5-HT-induced G_q pathway activation), whereas melatonin would have a more balanced G_i/G_q activation ratio (activation of both pathways), and luzindole, 4-PPDOT, and 5-HT would have a low ratio, exclusively activating the G_q pathway.

In conclusion, this present work revealed the capacity of MT₂ and 5-HT_2C receptors to assemble into functional heteromers. The binding and coupling properties of MT₂/5-HT_2C heteromers and the cellular pathway-biased ligands identified in the present study provide a solid framework for the study of the potential involvement of MT₂/5-HT_2C heteromers in the beneficial action of agomelatine in the treatment of major depression and generalized anxiety disorder. Targeting of GPCR heteromers might be of general importance for the increasing number of multitarget drugs developed in particular to treat psychiatric diseases (30, 31).