Melatonin MT1 and MT2 Receptors Exhibit Distinct Effects in the Modulation of Body Temperature across the Light/Dark Cycle

Melatonin (MLT) is a neurohormone that regulates many physiological functions including sleep, pain, thermoregulation, and circadian rhythms. MLT acts mainly through two G-protein-coupled receptors named MT1 and MT2, but also through an MLT type-3 receptor (MT3). However, the role of MLT receptor subtypes in thermoregulation is still unknown. We have thus investigated the effects of selective and non-selective MLT receptor agonists/antagonists on body temperature (Tb) in rats across the 12/12-h light–dark cycle. Rectal temperature was measured every 15 min from 4:00 a.m. to 9:30 a.m. and from 4:00 p.m. to 9:30 p.m., following subcutaneous injection of each compound at either 5:00 a.m. or 5:00 p.m. MLT (40 mg/kg) had no effect when injected at 5 a.m., whereas it decreased Tb during the light phase only when injected at 5:00 p.m. This effect was blocked by the selective MT2 receptor antagonist 4P-PDOT and the non-selective MT1/MT2 receptor antagonist, luzindole, but not by the α1/MT3 receptors antagonist prazosin. However, unlike MLT, neither the selective MT1 receptor partial agonist UCM871 (14 mg/kg) nor the selective MT2 partial agonist UCM924 (40 mg/kg) altered Tb during the light phase. In contrast, UCM871 injected at 5:00 p.m. increased Tb at the beginning of the dark phase, whereas UCM924 injected at 5:00 a.m. decreased Tb at the end of the dark phase. These effects were blocked by luzindole and 4P-PDOT, respectively. The MT3 receptor agonist GR135531 (10 mg/kg) did not affect Tb. These data suggest that the simultaneous activation of both MT1 and MT2 receptors is necessary to regulate Tb during the light phase, whereas in a complex but yet unknown manner, they regulate Tb differently during the dark phase. Overall, MT1 and MT2 receptors display complementary but also distinct roles in modulating circadian fluctuations of Tb.


Introduction
The maintenance of body temperature (T b ) in mammalians is critical for survival and internal homeostasis. The main brain structure involved in controlling T b is the hypothalamus, which receives inputs from the thermoreceptors located in both the brain and the periphery. Depending on these inputs, homeostatic changes are subsequently induced, causing sweating or shivering [1]. In particular, the preoptic area (POA) and dorsomedial hypothalamus (DMH) are critical hypothalamic areas for thermoregulation. In these areas, based on the firing rate responses to changes in local brain temperature, electrophysiological recordings have shown three different neuronal populations: warm-sensitive neurons (~30%), cold-sensitive neurons (~6%), and insensitive neurons (~60%), [2,3]. Activation of the thermally-responsive GABAergic and glutamatergic neurons in the ventral part of the lateral preoptic nucleus (vLPO) and the dorsal part of the dorsomedial hypothalamus (DMD), respectively, decreases temperature, physical activity, and metabolic rate. On the contrary, GABAergic neurons in the DMD promote the increase of T b , energy expenditure, and physical activity [4].
Recently, it has also been found that T b can be internally regulated in a circadian manner by endogenous signals from other parts of the hypothalamus [4,5], specifically in the suprachiasmatic nucleus (SCN) [1,6,7]. The SCN regulates the circadian rhythmicity of several physiological responses [1], including the oscillatory decrease of the thermoregulatory threshold of heat production during day time and heat loss during night time in diurnal species [1,8,9]. However, the mechanisms underlying this circadian modulation of T b remain unclear. Interestingly, the hypothalamus, including the SCN, DMH, and POA, are rich in melatonin (MLT) MT 1 and MT 2 receptors [7,[10][11][12][13], and the activation of these MLT receptors modulates numerous physiological effects including the control of T b [14].
MLT is produced in the pineal gland, mostly during the dark phase in both diurnal and nocturnal species [12,15]. The circadian production and release of MLT are controlled by the SCN [16]. Most of the physiological effects of MLT result from the activation of two high-affinity G-protein coupled receptors (GPCRs), MT 1 (pK i = 10.09) and MT 2 (pK i = 9.42), both of which are widely expressed in the mammalian brain [7,10,17]. The specific localization of the two MLT receptor subtypes in different regions of the brain and/or neuronal populations [11,[18][19][20] partially explains the selective and differential functional activity of the two MLT receptor subtypes, such as in sleep [18,[21][22][23], anxiety [24], pain [25,26], circadian rhythms [27], and depression [28]. In addition to these high-affinity MLT receptors, another low-affinity MLT binding site, termed MT 3 (pKi = 6.0), has been reported [29,30]. Given its both hydrophilic and lipophilic nature, MLT can easily pass through the cell membrane and bind nuclear receptors, including retinoic acid receptor-related orphan receptors (RORs) [31].
It is known that MLT decreases T b during the night in diurnal species [1]. Clinical studies have shown that exogenous administration of MLT suppresses the physiological increase in T b observed during daytime [32,33] and has hypothermic properties at the dose of 5 mg/kg [34][35][36][37][38]. Similar results have been demonstrated in preclinical studies in diurnal animals in which administration of MLT acted as a hypothermic agent in the active/light phase in fat sand rats and Marshall broiler chickens [39,40]. However, the neurobiological mechanism through which MLT exerts this hypothermic effect, as well as the selective contribution of the three MLT receptor subtypes, are yet to be investigated. Therefore, we investigated modifications in T b produced by selectively activating the three MLT receptor subtypes across the light-dark cycle in rats. To achieve this aim, we tested the effects of the selective MT 2 receptor partial agonist N-{2-[(3-bromophenyl)-(4-fluorophenyl)amino] ethyl}acetamide (UCM924) (pK iMT1 = 6.76; pK iMT2 = 9.27) [41], the selective MT 1 receptor partial agonist N-(2-{Methyl-[3-(4-phenylbutoxy)phenyl]amino}ethyl) acetamide (UCM871) (pK iMT1 = 8.93; pK iMT2 = 7.04) [42], and the MT 3 receptor agonist 5-Methoxycarbonylamino-N-acetyltryptamine (GR135531) (pK iMT3 = 29.5) on T b . The effects of UCM924, UCM871, and GR135531 were compared to those of MLT. In addition, selective and non-selective MLT receptor antagonists were also tested together with MLT and the other MLT receptor agonists/partial agonists to further dissect the MLT receptor subtypes involved in their thermoregulatory effects.

Results
In physiological conditions, as already known [1,9], there are changes in T b between the light and the dark phase; in particular, T b oscillations mostly occur during the phase shift ( Figure 1). During the shift from the dark to the light phase, the T b drops from an average of 38.45 to 37.5 • C after the light turns on ( Figure 1A). The opposite occurs during the shift from the light to the dark phase when the light is turned off ( Figure 1B).

Results
In physiological conditions, as already known [1,9], there are changes in Tb between the light and the dark phase; in particular, Tb oscillations mostly occur during the phase shift ( Figure 1). During the shift from the dark to the light phase, the Tb drops from an average of 38.45 to 37.5 °C after the light turns on ( Figure 1A). The opposite occurs during the shift from the light to the dark phase when the light is turned off ( Figure 1B).   In contrast, when MLT (40 mg/kg) was injected at the end of the light phase (5:00 p.m.), it induced a significant decrease (p < 0.05) in T b from 5:45 p.m. to 6:45 p.m. that was close to the transition from the light to the dark phase ( Figure 2B; interaction: F 17,408 = 1.908, p = 0.016; treatment: F 1,408 = 1.996, p = 0.171; time of the day: F 17,408 = 10.658, p < 0.001). Importantly, we observed no further effects of MLT on T b after the light-dark transition or during the beginning of the dark phase. The selective MT 2 receptor antagonist 4P-PDOT at a dose not affecting T b ( Figure 2D

Effects of the Selective MT 2 Partial Agonist UCM924 Injected at the End of the Dark and of the Light Phases on T b
As indicated in Figure 3A, the injection of UCM924 (40 mg/kg) at the end of the dark phase (5:00 a.m.) induced a significant decrease (p < 0.05) in T b immediately before the dark-light transition (from 6:45 a.m. to 7:30 a.m.), and did not affect T b during the dark-light transition and at the beginning of the light phase (two-way repeated measures ANOVA; interaction: F 17 s.c. injection of either vehicle, UCM924, UCM924 + 4P-PDOT, or 4P-PDOT.

Effects of the Selective MT1 Partial Agonist UCM871 Injected at the End of the Dark and of the Light Phases on Tb
As indicated in Figure    s.c. injection of either vehicle, UCM871, or UCM871 + luzindole.

Effects of the Selective MT3 Agonist GR135531 and Prazosin Injected at the End of the Light Phase on Tb
As indicated in Figure 5A, the injection of GR135531 (10 mg/kg) at the end of the light phase   Figure 2F; interaction: F 17,408 = 1.144, p = 0.309; treatment: F 1,408 = 0.012, p = 0.912; time of the day: F 17,408 = 9.289, p < 0.001).

Effects of the Selective MT 3 Agonist GR135531 and Prazosin Injected at the End of the Light Phase on T b
As indicated in Figure 5A, the injection of GR135531 (10 mg/kg) at the end of the light phase (5:00 p.m.) did not affect T b during the end of the light phase, the dark-light transition or the beginning of dark phase (two-way repeated measures ANOVA; interaction: F 17,306 = 0.94, p = 0.527; treatment:  5:00 p.m.) did not affect Tb during the end of the light phase, the dark-light transition or the beginning of dark phase (two-way repeated measures ANOVA; interaction: F17,306 = 0.94, p = 0.527; treatment: F1,306 = 0.342, p = 0.566; time of the day: F17,306 = 7.817, p < 0.001). The effects of MLT on Tb, when injected during the light phase, were not mediated by MT3 receptors, since the pre-treatment with the non-selective α1/MT3 antagonist prazosin at a dose not affecting Tb ( Figure 5C, interaction: F17,357 = 1.01, p = 0.446; treatment: F1,357 = 0.022, p = 0.882; time of the day: F17,357 = 7.588, p < 0.001) did not block the effects of MLT ( Figure 5B; interaction: F17,340 = 1.65, p = 0.050; treatment: F1,340 = 3.064, p = 0.095; time of the day: F17,340 = 10.086, p < 0.001). Interestingly, the treatment with prazosin plus MLT induced a further decrease of Tb even during the dark phase at 8:00 p.m. (Figure 5B) that was not observed with MLT ( Figure 2B) or prazosin ( Figure 5C) alone.

Discussion
In this study, we investigated the effects of MLT and its three receptors on Tb during the light and the dark phase for the first time. To achieve this aim, we used a pharmacological approach employing MLT, the selective MT1 receptor partial agonist UCM871, the selective MT2 receptor

Discussion
In this study, we investigated the effects of MLT and its three receptors on T b during the light and the dark phase for the first time. To achieve this aim, we used a pharmacological approach employing MLT, the selective MT 1 receptor partial agonist UCM871, the selective MT 2 receptor partial agonist UCM924, the MT 3 receptor agonist GR135531, and selective/non-selective MLT receptor antagonists, including the MT 2 selective antagonist 4P-PDOT, the MT 1 /MT 2 non-selective antagonist luzindole, and the MT 3 /α 1 antagonist prazosin. The exogenous administration of MLT during the light phase decreased T b immediately after the administration and before the light-dark phase shift, an effect blocked by both 4P-PDOT and luzindole. Interestingly, unlike MLT, neither UCM924 nor UCM871 produced a change in T b during the light phase. In contrast, the selective MT 2 partial agonist UCM924 administered at the end of the dark phase decreased T b during the dark phase, just prior to the dark-light switch, whereas the selective MT 1 partial agonist UCM871 injected at the end of the light phase increased T b during the following dark phase. On the other hand, MT 3 receptors did not seem to be involved in the regulation of T b , since the MT 3 receptor agonist GR135531 and the MT 3 /α 1 antagonist prazosin alone had not produced any effect on T b .
The rat circadian body temperature displays a cosine wave [39], showing a temperature that oscillates around the 35.6-36 • C during the light phase and around 37.8-38 • C during the dark phase [43]. The daily change in T b shows a characteristic deviation at two different times, consistent with the switch between day (light phase) and night (dark phase) hours [44]. The present study replicates the same physiological T b deviation that was previously reported in nocturnal rodents [43,44], with a daily T b peak during the night, which is concomitant with the increase in the activity of the animals [45].
In nocturnal rodents, T b peaks during night time when MLT levels are high, and decreases during the light phase when MLT levels are low. In contrast, in diurnal species, T b regulation follows the reverse direction in relation to MLT levels, showing a T b peak during the light phase [45] when circulating levels of MLT are very low (~10 pg/mL) [12,15,46]. However, the mechanism by which MLT regulates T b has not been established.
The neuronal circuit controlling the regulation of T b involves several structures including the SCN, SON, mPOA, and DMH [4,5]. Notably, the hypothalamus is an area rich in MLT receptors [11], and their expression may vary according to the phase and the time of the day [47][48][49][50][51].
Previous reports have shown that exogenous MLT administration during the light phase induced a decrease in T b in both humans and rodents [32,[35][36][37]39,52], although the active doses in humans are lower than those in rodents due to a significantly faster metabolism and very short half-life of MLT in the latter [53]. Our findings confirm that exogenous MLT influences T b , and its effects are strictly dependent on the time of day: MLT reduces T b only towards the end of the light phase and if administered during the light phase. In regard to its possible mechanism of action during the light phase, we found that both the selective MT 2 antagonist 4P-PDOT and the non-selective MT 1 /MT 2 antagonist luzindole blocked the T b reduction due to MLT, yet neither the selective MT 1 partial agonist UCM871 nor the selective MT 2 partial agonist UCM924 recapitulated the effects of MLT on T b . These findings suggest that during the light phase, MLT needs to simultaneously activate both MT 1 and MT 2 receptors to modulate T b . Indeed, the selective activation/inhibition of only MT 1 or MT 2 receptors did not affect T b during the day. In contrast, UCM871 and UCM924 produced changes in T b at different times of the dark phase and of opposite magnitude: UCM871 enhanced T b just after the light-dark transition, whereas UCM924 decreased T b just before the dark-light transition. Importantly, unlike UCM871 and UCM924, MLT did not induce any change in T b during the dark phase. We previously observed a similar time-of-day-dependent effect of MLT on sleep [23]. However, it is interesting that when α 1 receptors/MT 3 receptors were blocked by prazosin, MLT decreased T b also during the dark phase. These complex findings observed during the dark phase are likely dependent on the fact that during the dark phase there is a significant increase in the endogenous levels of MLT, and thus the expression of the two MLT receptors [19], as well as the involvement of other receptors implicated in thermoregulation, such as α 1 adrenoceptors [54], probably vary.
Interestingly, it is now well recognized that MLT receptors can form MT 1 /MT 2 hetero-oligomers and also heteromers with other receptors, and from a functional point of view, their properties are different from those of the corresponding homomers [55,56]. Since MLT receptors as well as other receptors including α 1 -adrenoceptors are highly expressed in brain regions/nuclei involved in thermoregulation, we cannot exclude that some of the effects of MLT on T b described here were mediated by these oligomers/heteromers. Future studies are needed to investigate the possible circadian variability in the formation and role of oligomers and/or heteromers of MLT receptors in hypothalamic nuclei regulating T b . Similarly, the potential contribution of nuclear receptors, among which RORs that are also activated by MLT [31], is worth investigating.
The phase-dependent response of T b to exogenous MLT may depend not only on the changes in the density of MLT receptors across the light-dark cycle, but also on the relative distribution and function of MT 1 and MT 2 receptors that control unique physiological responses in the brain, for example in sleep [21][22][23], anxiety [18,24], pain [25,26,57], and depression [28,58,59], and in the periphery, for example at the cardiovascular level [60,61].
In conclusion, we have investigated the role of MLT receptors in thermoregulation, and found that during the light phase T b is affected only if both MT 1 and MT 2 receptors are simultaneously activated. Further, during the dark phase, a time-dependent effect was found in that the activation of MT 1 and MT 2 produces an increase and decrease of T b respectively. No effects on T b of the MLT MT 3 receptor subtype were evidenced. However, MT 1 and MT 2 receptors control T b in synergy with other receptors including α 1 adrenoceptors. These data further support the recent findings showing that the MT 1 and MT 2 receptors modulate physio-pathological functions in different and sometimes opposing ways, and in a time-of-day dependent manner. In particular, MT 1 or MT 2 agonists may be further tested for hypothermia or hyperthermia, respectively.

Animals
Male Wistar rats (200-250 g, Charles River) were used for behavioral tests. All animals were housed at constant room temperature (20 • C) and humidity under a 12/12-h light-dark cycle (lights on at 7:30 a.m. and off at 7:30 p.m.) with food and water ad libitum. All experimental procedures were performed between 5:00 a.m. and 9:30 a.m. and between 5:00 p.m. to 9:30 p.m. The experimental protocol was approved by the Animal Ethics Committee (AUP#5253, McGill University, QC, Canada) and followed the ethical guidelines of the Canadian Institute of Health Research for animal care and scientific use.

Assessment of Body Temperature
Body (rectal) temperature (Tb) in awake animals was measured by goosing the animal using a Traceable Snap-in Module with probe (Fisher Science Education, S90862). The probe was inserted to a depth of 2 cm for no more than 10 s, whereas the tested individual was kept in a cotton bag. Animals were handled every day for five days before the experiments with the aim of habituating the animal to the testing procedure and thus minimizing the associated stress.

Statistical Analysis
Data analysis was conducted using the SigmaPlot statistical software version 13 (Systat Software, Inc.). After controlling for the normal distribution of the data, a two-way ANOVA for repeated measures was used to analyze the data using treatments (between) and testing time (within) as factors. Post hoc analyses were performed using the Bonferroni test for multiple comparisons. The effect of vehicle was compared with that of the different agonists/partial agonists, the antagonists alone, or the agonist/partial agonist plus the antagonist. All data were expressed as mean ± SEM. p < 0.05 was considered significant. All figures were made using MATLAB software.
Temperature values were normalized as follow: [Temperature at time X -Average of

Assessment of Body Temperature
Body (rectal) temperature (T b ) in awake animals was measured by goosing the animal using a Traceable Snap-in Module with probe (Fisher Science Education, S90862). The probe was inserted to a depth of 2 cm for no more than 10 s, whereas the tested individual was kept in a cotton bag. Animals were handled every day for five days before the experiments with the aim of habituating the animal to the testing procedure and thus minimizing the associated stress.

Statistical Analysis
Data analysis was conducted using the SigmaPlot statistical software version 13 (Systat Software, Inc.). After controlling for the normal distribution of the data, a two-way ANOVA for repeated measures was used to analyze the data using treatments (between) and testing time (within) as factors. Post hoc analyses were performed using the Bonferroni test for multiple comparisons. The effect of vehicle was compared with that of the different agonists/partial agonists, the antagonists alone, or the agonist/partial agonist plus the antagonist. All data were expressed as mean ± SEM. p < 0.05 was considered significant. All figures were made using MATLAB software.