Dicentrarchus labrax used in this study originated from commercial aquaculture plants (Supplementary Figure 1A). Most of the experiments were performed with fish from “Méditerranée Pisciculture” [Salses-Le-Chateau, France; 42°50′ North (N), E2°55′ east (E)]. At the aquaculture plant the fish were reared outdoors, under natural conditions of photoperiod. The tanks were supplied with a mixture of seawater (from the Mediterranean Sea) and freshwater (from a local resurgence), so that the resulting salinity and temperature were maintained constant all year long (∼7‰ salinity and 16 ± 2°C, respectively). Random samplings indicated that most of the fish were males, in agreement with previous studies showing the sex ratio of farmed sea bass is highly biased, with a high proportion of males (ranging between 75 and 100%: Piferrer et al., 2005; Vandeputte et al., 2012, 2014, 2020), particularly at the water temperature at the “Méditerranée Pisciculture” facilities.

Upon arrival at the lab facilities of Banyuls-sur-Mer (France; 42°28′ N, 3°12′ E) the fish were placed for 3 weeks in 1 m³ indoors tanks (35 fish per tank) supplied with Mediterranean Sea water (∼37‰ salinity along the French Mediterranean coast) provided by an open circuit, at a flow rate of ∼400 l/h (Supplementary Figure 1B). In the experiments investigating the effects of salinity, the seawater was replaced by freshwater (4‰ salinity), all other conditions remaining similar [natural temperature and simulated natural photoperiod (light provided by fluorescent bulbs)] (Supplementary Figure 1B). To minimize stress induced by sampling, the fish were evenly distributed in different tanks, isolated one from each other by a double layer of thick black plastic curtains, so that the sampling was carried out in an alternate manner. Fish were fed ad libitum. Dark sampling was performed under dim red light. Mortality was close to 0 in the lab facilities. From now on these fish are named BW fish as they had been reared in brackish water before their arrival at the lab. As specified in Supplementary Figure 1, these fish were used for the in vivo (annual variations and acclimation to water salinities), and in vitro (melatonin challenges), experiments.

A second group of fish was made available to us, which originated from “Cannes aquaculture” (Cannes, France; 43°33′ N, 7°00′ E) and had been maintained for more than a year at the IFREMER station of Palavas-les-Flots (France; 43°31′ N, 3°55′ E). At the IFREMER station, they were placed in 10,000 l indoors round tanks, under simulated natural photoperiod (light provided by fluorescent bulbs) and natural temperature (see details below and Supplementary Figure 1A). Two major differences distinguished these fish from the BW fish described above: (1) they experienced seawater only (salinity ∼37‰) and (2) the same pool of fish was used from the beginning until the end of the experiment, so that they weighed 172 ± 37 g (mean ± S.E.) at the beginning, and 388 ± 77 g at the end, of the experiment (i.e., 12 months later). When brought to the lab the fish were kept in seawater, and were thus named SW fish (Supplementary Figure 1B); all other lab conditions were similar to those described above for the BW fish. These fish were used to investigate the annual variations in abundance of pituitary hormones and melatonin receptors mRNA (Supplementary Figure 1).

All experiments were performed according to the European Union regulations concerning the protection of experimental animals, and in agreement with the protocols approved by the Animal Care and Use Committee of the Observatoire Océanologique of Banyuls sur Mer (CNRS-UPMC; authorization # A-66-01-601).

The fish were sacrificed by decapitation every 4 h of a 24 h LD cycle (starting at 08:00) (Supplementary Figure 1C). The sampling was performed in April (13°C, water temperature), June (22°C), October (20°C), January (13°C) and March (13°C). After removal, the pituitary glands were dipped in RNAlater ^® (Applied Biosystems, Courtaboeuf, France). At each sampling time, 5 pools of 3 pituitary glands each were obtained for the BW fish. For technical reasons (size and number of fish available) only two 2 pools of 5 pituitary glands each were collected for the SW fish. The pools were frozen and kept at −80°C until mRNA extraction (as indicated below).

After dissection, pituitary glands from BW fish (Supplementary Figure 1C) were cultured individually as detailed elsewhere (Falcón et al., 2003). Briefly, the pituitary glands were rinsed and cultured for 2 days in 750 μL of HEPES-buffered RPMI-1640 culture medium (one pituitary per well of a 24-well culture plate) at the temperature at which the fish had been acclimated. At noon of day 3, the pituitaries were challenged for 24 h with melatonin concentrations ranging from 10^–12 to 10^–8 M, which correspond to the plasma physiological concentrations measured during daytime and night-time in D. labrax (Garcia-Allegue et al., 2001; López-Olmeda et al., 2009). The effects of S20304 (10^–8 M) and luzindole (10^–7 M), respectively, a non-selective agonist and a non-selective antagonist at the mammalian melatonin receptors, were also investigated. At the end of the treatment, pools of 3 glands each were dipped in RNAlater ^® and stored at −80°C. The in vitro experiments investigating the effects of melatonin were carried out in February and August, i.e., during the reproductive and resting seasons, respectively. The experiments conducted in February were performed with pituitaries of fish adapted to salinities of 37 or 4‰ salinities, and those in August with pituitaries of the 37‰, salinity adapted fish. Data are expressed as the mean ± S.E.M. More details are provided in the figure legends and in Supplementary Figure 1.

Total RNA from the pituitary glands was extracted using the TRIZOL reagent (Life technologies SAS; Saint Aubin, France) according to the manufacturer’s instructions. The concentration and quality of the RNA was checked by spectrophotometry (Bio-photometer, Eppendorf, Le Pecq, France) and after migration in a MOPS gel. One μg of RNA was reversed transcribed using the Superscript Reverse transcriptase (Clontech, Saint-Germain-en-Laye, France). The cDNA obtained was stored at −20°C.

Real time quantitative PCR (qPCR) was performed using a Light Cycler 1.5 (Roche Diagnostics, Meylan, France) and the Light Cycler Fast Start DNA Master SYBR Green I kit (Roche Molecular Biochemicals, Meylan, France). The conditions of the reaction were 10 min at 95°C followed by 10 s at 95°C, 5 s at 60°C and 15 s at 72°C, for 30 cycles. The melting curves were analysed for each sample, to check that only a single sequence was amplified. They were obtained after increasing temperature progressively to 95°C, decreasing it to 65°C for 20 s, and finally increasing it to 95°C. Negative controls included cDNA replacement by water and the use of non-transcribed total RNA. In order to ensure the specificity of the amplification, some controls also included running the obtained PCR products in an agarose gel, followed by extraction, sub-cloning and sequencing (Sauzet et al., 2008). The 2^–ΔΔCt method was used to determine the relative mRNA abundance (Livak and Schmittgen, 2001), with the L17 ribosomal protein mRNA as the reference gene (Herrera-Pérez et al., 2011). The primers (Table 1) were designed using the Light Cycler Probe Design software (Roche, Meylan, France). Their efficiency was tested using serial dilutions of cDNA from pituitary extracts; the calibration curves exhibited slopes and efficiencies close to −3.32 and 100%, respectively.

Data were analysed using the one-way or two-way analysis of variance (ANOVA) followed by Holm-Sidak’s post hoc test for multiple comparisons. Individual means were compared using the Student’s t test (Two-tailed). Prism 6.01 (GraphPad™ software, La Jolla, CA, United States) was used to perform the statistics (level of significance set at P < 0.05) and plot the graphs.

Unless otherwise specified, the products were from Sigma Aldrich (Saint Quentin Fallavier, France). S20304 was a kind gift of the ‘Servier Labs’ (Neuilly-sur-Seine, France).

Sampling started in April and ended in March of the following year. Relative abundance of mRNA from all the pituitary hormones investigated displayed significant annual variations.

FSH and LH: The profiles obtained were similar for both hormones and independent of the fish origin (BW or SW fish) (Figure 1 and Table 2). Levels were low from April to October, and then increased to reach a peak value in January and resume low levels in March.

POMC and GH: POMC mRNA abundance displayed a bell-shape response (peak in October) in the SW fish, while no significant annual rhythm was seen in the BW fish (Figure 2 and Table 2). The opposite situation was observed with GH, which exhibited a U-shape profile in the BW fish (high values in October and January), and no significant variation in the SW fish (Figure 2 and Table 2).

PRL, SL and TSH: Two situations were observed after comparing the BW to the SW fish as the annual variations displayed 180° out of phase profiles (Figure 3 and Table 2). In the BW fish, PRL mRNA relative abundance decreased significantly from April to October, and increased from October to March, while SL and TSH levels were high in March and low at the other months investigated.

The three melatonin receptor subtypes cloned in the sea bass displayed daily variations in abundance in the SW fish, significant in March (MT1, MT2, Mel1c), January and October (MT1, Mel1c), but not in June (Figure 4A). They also displayed annual variations. The highest abundance was found in June and October for MT2, October for MT1 and March for Mel1c (Figure 4B).

After acclimation of the seabass to 4 or 37‰ salinities for 3 weeks in the lab, the mRNA abundance of pituitary hormones changed significantly for PRL (more than a 4-fold difference), GH, POMC and FSH (∼3-fold difference) (Figure 5). In the case of PRL and POMC, the levels were high under the 4‰ condition and low under the 37‰ condition, while the opposite held true for GH and FSH mRNA levels. SL levels were also 30% lower under the 4 vs. 37‰ condition, a difference not statistically significant. No difference was seen for LH and TSH (Figure 5).

The effects of melatonin on the pituitary hormones mRNA levels were investigated in February and August, in BW fish acclimated for 3 weeks to a 37 or a 4‰ salinity (see experimental protocol design in Supplementary Figure 1).

-February, 37‰ salinity. Melatonin had no significant effect on PRL, SL and LH mRNA abundance (Figures 6–8). The hormone induced dramatic effects on GH and POMC, inhibitory for the former (3-fold inhibition at 10^–12 M; Figure 6) and stimulatory for the latter (4-fold stimulation at 10^–9 M; Figure 7). The FSH mRNA relative abundance was slightly inhibited by increasing concentrations of melatonin (Figure 8). The effect on TSH was complex as the stimulation observed at 10^–12 and 10^–10 M was reversed at 10^–9 and 10^–8 M (Figure 7). The effects were mimicked by the melatonin analogs S20304 and luzindole used at the concentration used (insets in Figures 6–8).

-February, 4‰ salinity. In the fish adapted to the 4‰ salinity, melatonin had no significant effects on SL, POMC, TSH and LH mRNA abundances (Figures 6–8). A v-shaped dose-response curve was obtained with GH (highest inhibition at 10^–10 M of melatonin; Figure 6), while FSH mRNA levels were increased at the single concentration of 10^–9 M (Figure 8). In a general manner, the effects were mimicked by the melatonin analogs S20304 and luzindole used at one single concentration (insets in Figures 6–8).

-August, 37‰ salinity. While TSH mRNA abundance was not affected by melatonin, POMC, GH, SL mRNA relative levels were inhibited in a dose-dependent manner (Figures 9, 10). PRL and LH mRNA were also affected, however, the response was more complex, stimulatory at the low (10^–12, 10^–10 M), and inhibitory at the high (10^–9 M), concentrations (Figures 9, 10). In a general manner, the effects of the melatonin agonist S20304 (10^–8 M), were in agreement with those obtained with melatonin, except for POMC (inset Figure 9). For comparison the data obtained in February and August (at the salinity of 37‰) are plotted together in Supplementary Figures 2, 3.

Most of the studies reporting on the daily and annual variations of pituitary hormones focus on FSH and LH (Mateos et al., 2003; Cowan et al., 2017). Here we provide the first information on a panel of pituitary hormones in the European sea bass. Previous studies have shown a clear annual rhythm in the abundance of FSH and LH mRNA in the sea bass pituitary, with a peak occurrence in January/February, and lowest abundance from May to October (Mateos et al., 2003). This rhythm was paralleled by variations in the gonado-somatic index and plasma LH protein levels, in keeping with the annual rhythm of reproduction in this species (Kara, 1997; Alvarado et al., 2013). Our data agree fully with these previous findings. In addition, we found that similar profiles were obtained independent of the fish origin and previous rearing conditions, which brings strong support to the conclusion that the seasonal control of reproduction is a tightly locked photoperiod-dependent process in the European sea bass, as suggested earlier (Bayarri et al., 2004; Falcón et al., 2010).

A contrasting situation occurred with the other pituitary hormones, in which the annual rhythms of mRNA abundance differed in phase (TSH, SL, PRL) or amplitude (GH, POMC), in the SW vs. BW fish. The reason for these differences could lie in the fact that the fish originated from two different locations and had different historical backgrounds. The BW fish originated from distinct fertilization processes, and had all similar standard commercial sale size at each sampling month. In contrast, the SW fish were all from a single pool of fish that were followed during whole duration of the experiment. In addition, the SW fish were raised in a 37‰ salinity all life on, while the BW fish experienced a transition from a ∼7‰ to a ∼37‰ salinity 3 weeks before sacrifice. The differences may have impacted differently PRL, SL, GH, TSH, and POMC. These hormones have pleiotropic effects [control of skin pigmentation (POMC, SL), immunity and stress (POMC, GH, PRL), food intake (POMC, GH), growth (GH, TSH), osmotic regulation (GH, PRL), sexual maturation (GH, PRL, TSH)], and this involves a complex network of interactions within the pituitary, as well as the whole neuroendocrine loops (Agustsson and Bjornsson, 2000; Harris and Bird, 2000; Boeuf and Falcón, 2001; Bhandari et al., 2003; Deane and Woo, 2009; Laiz-Carrión et al., 2009; Blanco, 2020); in this regard, their regulation implies more flexibility than is the case for the reproductive hormones.

Part of the annual life cycle of the European sea bass involves salinity changes, which needs rapid adaptive neuroendocrine responses. We indeed observed that a 3 weeks adaptation to either low (4‰) or high (37‰) salinity impacted mRNA levels of GH, FSH, PRL, and POMC, while those of LH, SL, and TSH remained unaffected. In the case of PRL, POMC, SL, and GH, our data agree with those obtained in this (Varsamos et al., 2006) and other (Laiz-Carrión et al., 2009; Lu et al., 2016; Yamaguchi et al., 2016; Yuan et al., 2017; Liu et al., 2020; Vargas-Chacoff et al., 2021) fish species maintained under low or high salinities. They are also in line with the known roles of these hormones, essential in the processes of acclimation to different salinities (PRL, SL, GH, or rather the GH/IGF-I axis); both, PRL and GH/IGF-I interact with cortisol to control ionic balance (Takei and McCormick, 2013). The production of cortisol by the interrenal gland is controlled by ACTH (adrenocorticotropic hormone), a cleavage product of POMC. Here we found that POMC mRNA levels were higher under the freshwater than under the seawater condition, as observed in the ayu, an anadromous species (Lu et al., 2016), but in contrast to the situation in the tilapia, mainly a fresh water fish (Aruna et al., 2015). The case of SL is interesting as we found a slight non-significant increase in mRNA relative abundance, as is the case in the gilthead seabream (Laiz-Carrión et al., 2009); however, in the latter intermediate salinities (12‰) induced a dramatic decrease in SL mRNA compared to the low and high salinities (6 and 38‰, respectively), which highlights the complexity of the mechanisms studied.

Melatonin receptors have been identified in the pituitary of several fish species (Gaildrat and Falcón, 1999, 2000; Gaildrat et al., 2002; Sauzet et al., 2008; Ikegami et al., 2009; Confente et al., 2010; Hong et al., 2014; Ciani et al., 2019; Sakai et al., 2019). Three subtypes have been identified in the sea bass (MT1, MT2, Mel1c), MT2 displaying the higher relative abundance in the pituitary (Sauzet et al., 2008; Herrera-Pérez et al., 2010; and this study). They also exhibited daily and seasonal variations as reported earlier for both, mRNA relative abundance and melatonin binding sites, in the pituitary and brain of other fish species (Falcón et al., 1996; Gaildrat et al., 1998; Amano et al., 2003; Park et al., 2006; Bayarri et al., 2010; Confente et al., 2010; Ciani et al., 2019). These variations are controlled by circadian clocks in the pike (Gaildrat et al., 1998) and circannual clocks in the European sea bass (Bayarri et al., 2010) as they were observed under constant conditions. Thus, a good understanding of the time-dependent effects of melatonin (as discussed below) requires that not only its cyclic production, but also the rhythmic availability and affinity of its receptors to be considered.

The current study supports previous investigations indicating melatonin treatment impacts LH, GH, and PRL productions in fish (Khan and Thomas, 1996; Falcón et al., 2003), and extends to POMC, SL, FSH, and TSH the range of hormones affected. In most cases, the conclusions were supported by the observed effects of the melatonin receptor agonist S20304 and antagonist luzindole. As characterized in mammals, luzindole is supposed to be an antagonist. However, paradoxical effects as the ones described in this study have been reported. In rats, luzindole may also act as a partial agonist at the MT2 receptor subtype (Legros et al., 2020) or exert direct proper effects (i.e., not mediated through melatonin receptors) on plasma membrane K⁺ currents (Zhou et al., 2003) or [Ca²⁺]_i stores (Estaras et al., 2020). In the medaka luzindole acted as a partial antagonist or a full agonist depending on the receptor subtype (Sakai et al., 2019). It has been emphasized that “ligands that target melatonin receptor subtypes in one species does not have the same affinity and specificity in another species” (ViviD and Bentley, 2018), which might be particularly true in fish, in which the receptors display proper structural and functional characteristics different from those of mammals.

In brief, the responses to the melatonin challenges appeared complex, depending for each hormone on several factors, including dose, time of year and previous adaptation to salinity.