Convergence of Melatonin and Serotonin (5-HT) Signaling at MT2/5-HT2C 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-HT2C 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. 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-HT2C receptors into functional heteromers. This is of particular interest in light of the “synergistic” melatonin agonist/5-HT2C 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-HT2C receptor heteromers both in transfected cells and in human cortex and hippocampus. MT2/5-HT2C heteromers amplified the 5-HT-mediated Gq/phospholipase C response and triggered melatonin-induced unidirectional transactivation of the 5-HT2C protomer of MT2/5-HT2C heteromers. Pharmacological studies revealed distinct functional properties for agomelatine, which shows “biased signaling.” These observations demonstrate the existence of functionally unique MT2/5-HT2C 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-HT2C heteromers provide a new strategy for the discovery of novel agents for the treatment of psychiatric disorders.

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 MT 1 and MT 2 receptors and with moderate affinity to the enzyme quinone reductase 2 (2). Both MT 1 and MT 2 receptors as well as all 14 classes of 5-HT receptor (except 5-HT 3 ) 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 MT 1 or MT 2 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 D 1 and D 2 receptor heteromers incriminated in depressed states (6).
To date, the possible existence of heteromeric associations of MT 1 or MT 2 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 ago-melatine, the first to possess a non-monoaminergic component of action, behaves as an agonist at G i -coupled MT 1 and MT 2 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 MT 1 and MT 2 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 MT 1 and MT 2 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" MT 1 , MT 2 , 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 MT 2 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.
Immunoprecipitation-For co-immunoprecipitation assays, HEK293 cells were seeded in 10-cm dishes and co-transfected with 4 g 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 4ϫ 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-[ 125 I]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.
Intracellular Signaling Assays-HEK293 cells stably expressing 20 -30 fmol of MT 2 /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 M 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 Ϯ S.E. of at least three experiments. Student's t test was applied for statistical analysis.

Physical Interaction between 5-HT 2C and MT 1 /MT 2 Receptors-
We first carried out co-immunoprecipitation experiments to assess a possible interaction between the human 5-HT 2C and MT 1 and MT 2 receptors. These experiments revealed that 5-HT 2C interacts with MT 1 and MT 2 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, MT 1 -YFP, MT 2 -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 BRET 50 values, corresponding to the sat- uration of 50% of BRET donors by BRET acceptors, revealed that the relative propensity of 5-HT 2C to form heteromers with MT 2 (BRET 50 ϭ 2.2 Ϯ 0.4) is higher than with MT 1 (BRET 50 ϭ 6.9 Ϯ 1.5) or with itself (BRET 50 ϭ 8.8 Ϯ 1.5), indicating preferential formation of MT 2 /5-HT 2C heteromers. As shown by immunofluorescence staining, 5-HT 2C expressed alone was mainly intracellular, and a significant amount of MT 2 and 5-HT 2C was present at the plasma membrane where both receptors colocalized (Fig. 1D). Formation of MT 2 /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-MT 2 antibodies were used as a negative control. Further evidence for the formation of MT 2 /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-[ 125 I]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 MT 2 in transfected HEK293 cells and in the cortex, hippocampus, and choroid plexus.
Melatonin but Not 5-HT Activates the G i /cAMP Pathway through MT 2 /5-HT 2c Heteromers-HEK293 cells expressing equivalent and physiologically relevant levels of MT 2 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 MT 2 alone as expected and in cells co-expressing MT 2 and 5-HT 2C with the same efficiency (EC 50 ϭ 0.76 Ϯ 0.46 versus 0.76 Ϯ 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 M 5-HT did not modify cAMP levels in any of the three cell types (Fig. 2B). These results indicate that activation of the MT 2 protomer of the MT 2 /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.
MT 2 Potentiates 5-HT-induced Signaling by Increasing Cell Surface Expression of 5-HT 2C in the MT 2 /5-HT 2C Heteromer-We next investigated the capacity of MT 2 /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 (ϳ3-fold) in cells expressing similar quantities of 5-HT 2C receptors, but in

TABLE 1 Compared potencies and properties of melatonin receptor ligands on two signaling pathways in cells expressing MT 2 or MT 2 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 ϭ IC 50 /1 ϩ (S/X) where S and X represent the agonist concentration and EC 50 , respectively. ND, not determined. the presence of MT 2 , the response was of similar potency (EC 50 ϭ 21 Ϯ 11 nM (5-HT 2C ) versus 68 Ϯ 25 nM (MT 2 /5-HT 2C )). 5-HT was ineffective in cells expressing MT 2 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 G␤␥ scavenger ␤ARK Cter had no significant effect (Fig. 3C), confirming the predominant role of G q/11 ␣ 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 MT 2 as expected (Fig. 3D). Pretreatment with S20928, a melatonin receptor antagonist, showed no cross-reactivity on 5-HT-induced responses (Fig. 3E).
Amplified 5-HT-induced responses in the context of the MT 2 /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 Ϯ 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 MT 2 receptors, the fraction of cell surface-expressed 5-HT 2C almost doubled (22.7 Ϯ 3.0%), whereas the total amount of 5-HT 2C receptors was not modified. The total amount and the fraction of cell surface-expressed MT 2 remained constant irrespective of the presence or absence of 5-HT 2C . Taken together, our results indicate that MT 2 co-expression potentiates 5-HT 2C receptor signaling in the MT 2 /5-HT 2C heteromer by increasing the cell surface expression of 5-HT 2C .

Activation of the MT 2 Protomer Transactivates the PLC Pathway and Improves ␤-Arrestin Recruitment in the MT 2 /5-HT 2C
Heteromer-To explore the possible effect of MT 2 activation in the MT 2 /5-HT 2C heteromer on the G q /PLC pathway, we stimulated HEK293 cells expressing MT 2 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 EC 50 of 69 Ϯ 20 nM (Fig. 4A). To verify that melatonin-induced IP production is not due to the signaling cross-talk between MT 2 and any G q -coupled GPCR, we co-expressed MT 2 with the G q -coupled M1 muscarinic receptor, which does not form heteromers with MT 2 (data not show). In these cells, melatonin did not increase IP production, indicating that melatonin-induced IP production is specific for the MT 2 /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).
Melatonin-induced IP production in cells co-expressing MT 2 and 5-HT 2C receptors was not affected by the expression of the G␤␥ scavenger ␤ARK 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/11 ␣ proteins, which can be assisted by the presence of G i ␣ proteins. This result raises the intriguing possibility that melatonin binding to the MT 2 protomer of the MT 2 /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.
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 MT 2 (Fig. 5B). When cells co-expressing this mutant and MT 2 receptors were stimulated with melatonin, the anticipated reduction in amplitude and efficiency (EC 50 ϭ 550 Ϯ 187 versus 69 Ϯ 20 nM for MT 2 /5-HT 2C -S138N and MT 2 /5-HT 2C , respectively (p Ͻ 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 MT 2 as comparable BRET 50 values (2.2 Ϯ 0.4 versus 3.5 Ϯ 0.6 for MT 2 /5-HT 2C and MT 2 /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 MT 2 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 ␤-arrestins. We therefore studied the ability of melatonin to activate the previously described ␤-arrestin 2 BRET sensor (11,23) in cells expressing either MT 2 alone or together with 5-HT 2C . As seen in Fig. 5E, MT 2 alone only weakly activated the BRET sensor with an EC 50 of 260 Ϯ 170 nM. In contrast, melatonin potently recruited the sensor in cells co-expressing MT 2 and 5-HT 2C (EC 50 ϭ 65 Ϯ 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 MT 2 (Fig. 5F). These results show that melatonin-induced ␤-arrestin recruitment to MT 2 is potentiated in the MT 2 /5-HT 2C heteromer.
Agomelatine Exhibits a Unique Functional Profile at MT 2 /5-HT 2C Heteromers-To further explore the functional profile of MT 2 /5-HT 2C heteromers, we studied the effect of several syn-thetic melatonin receptors ligands. S20928, a well know melatonin receptor antagonist (24), antagonized the melatonin-induced inhibition of cAMP levels as expected in cells expressing MT 2 alone (K i ϭ 4.88 Ϯ 1.50 nM) but also in cells co-expressing MT 2 and 5-HT 2C receptors (K i ϭ 0.66 Ϯ 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 MT 2 /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 MT 2 on the G i /cAMP pathway (7,25), showed pathway-biased properties on MT 2 /5-HT 2C heteromers as they behaved as full antagonists on the G i /cAMP pathway (K i ϭ 0.04 Ϯ 0.01 nM and K i ϭ 7.73 Ϯ 4.98 nM, respectively) but as a partial (4-PPDOT) or full (luzindole) agonist on the G q /PLC pathway (EC 50 ϭ 2.5 Ϯ 1.7 nM and EC 50 ϭ 223 Ϯ 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 MT 2 /5-HT 2C heteromers.
We next studied the effect of agomelatine, the antidepressant with agonistic activity at MT 2 receptors and neutral antagonistic properties at 5-HT 2C receptors (7,8). In cells expressing MT 2 in the absence or presence of 5-HT 2C receptors, agomelatine behaved as an agonist of the G i /cAMP pathway (EC 50 ϭ 0.29 Ϯ 0.16 nM and EC 50 ϭ 0.89 Ϯ 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 MT 2 /5-HT 2C heteromers as compared with melatonin and 5-HT.

DISCUSSION
We describe here a previously unappreciated dimension of cross-talk between melatonin and 5-HT that is mediated by heteromers of MT 2 and 5-HT 2C receptors. MT 2 /5-HT 2C heteromers have unique functional properties and are formed preferentially compared with the corresponding homomers. MT 1 and 5-HT 2C receptors also form heteromers in transfected cells, but we focus herein on MT 2 /5-HT 2C heteromers. Within these heteromers, melatonin is able to activate distinct cellular cascades: not only the G i /cAMP pathway as for MT 2 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 MT 2 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 MT 2 /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 MT 1 and MT 2 receptors and as a neutral antagonist at G q/11coupled 5-HT 2C receptors (7,8). Our data indicate that agomelatine preserves these properties in MT 2 /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 MT 2 /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-HTpromoted 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 MT 2 protomer, melatonin stimulation not only activates the G i /cAMP pathway as for MT 2 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 D 2 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 MT 2 protomer, agomelatine is completely inactive in this mode but still able to fully activate the MT 2 protomer.
Simultaneous activation of G i -and G q -dependent signaling by melatonin is a distinctive feature of MT 2 /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 iand 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 MT 2 /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 MT 2 and 5-HT 2C receptors to assemble into functional heteromers. The binding and coupling properties of MT 2 /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 MT 2 /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).