Ghrelin induces clock gene expression in the liver of goldfish in vitro via protein kinase C and protein kinase A pathways

The circadian system controls physiological rhythms that let organisms anticipate cyclic environmental changes. In vertebrates, this endogenous timing system consists of multiple coupled central and peripheral oscillators that are entrained by environmental cues (Albrecht, 2012; Tsang et al., 2014). The molecular basis of these oscillators is well conserved among vertebrates (Dunlap, 1999; Panda et al., 2002) and it is based on interlocked auto-regulatory feedback loops of genes known as clock genes. The positive limb of the main loop includes clock and bmal1 genes (circadian locomotor output cycles kaput, and brain and muscle ARNT-like 1, respectively). These genes form the heterodimer CLOCK-BMAL1 that activates the transcription of negative elements, the per (Period) and cry (Cryptochrome) clock genes, whose protein products inhibit CLOCK-BMAL1 transactivation (Hastings et al., 2007; Nader et al., 2010; Schibler et al., 2015). This main auto-regulatory loop is stabilized by an auxiliary loop formed by rev-erbα (V-erbA-related protein EAR-1) and ror (Retinoic acid related Orphan Receptor) genes, which mainly modulates bmal1 expression (Nader et al., 2010; Schibler et al., 2015).

It is well known that the light–dark and feeding cycles may act as potent synchronizers of locomotor activity daily rhythms in vertebrates, including teleosts (Bechtold, 2008; Madrid et al., 2001; Mistlberger, 2011; Sánchez-Vázquez and Madrid, 2001; Spieler, 1992; Stephan, 2002). However, the food-related signals that entrain the molecular clocks (food-entrainable oscillators, FEOs) in the circadian system remain unknown. In mammals, the oscillators of the circadian system respond to feeding inputs with different sensitivities, the liver being one of the most sensitive peripheral oscillators in these vertebrates (Albrecht, 2012; Damiola et al., 2000; Reddy et al., 2007; Schibler et al., 2015; Schmutz et al., 2012; Sujino et al., 2012). In teleosts, the liver is highly sensitive to the feeding/fasting cycle and food-related signals (Costa et al., 2016; del Pozo et al., 2012; Feliciano et al., 2011; López-Olmeda et al., 2010; Vera et al., 2013; Sánchez-Bretaño et al., 2015b). Some studies suggest that the liver in fish may be acting as an oscillator that is synchronized by photoperiod and feeding schedule (del Pozo et al., 2012; Feliciano et al., 2011; López-Olmeda et al., 2010; Martín-Robles et al., 2011; Sánchez-Bretaño et al., 2015a,b; Tinoco et al., 2014; Vera et al., 2013).

Ghrelin is a peripheral orexigenic peptide hormone mainly involved in energy balance by stimulating food intake, carbohydrate utilization and adiposity (Abizaid and Horvath, 2012; Delporte, 2013), although it also exerts a wide variety of physiological functions (Delporte, 2013; Sato et al., 2012). This hormone displays a daily rhythm in terms of expression and content in mouse stomach (LeSauter et al., 2009), and in rat hypothalamus and plasma (Bodosi et al., 2004; Patton et al., 2014). Ghrelin is also rhythmically expressed in the hypothalamus, pituitary and anterior intestine of goldfish, Carassius auratus (Linnaeus 1758) (Sánchez-Bretaño et al., 2015c). Such ghrelin rhythms have been mainly related to the feeding–fasting cycle, and it has been suggested that this hormone may drive food anticipatory activity, acting as an output of the FEOs (LeSauter et al., 2009; Nisembaum et al., 2014; Patton et al., 2014). Additionally, some studies in mouse and goldfish point to ghrelin as an input of circadian clocks, by signalling feeding–fasting rhythms. Indeed, ghrelin induces a phase advance and a delay of the spontaneous firing rhythm and clock gene expression in vivo and in cultured mouse suprachiasmatic nuclei (Yannielli et al., 2007; Zhou et al., 2014). In goldfish, the peripheral administration of ghrelin stimulates per expression in hypothalamus and liver (Nisembaum et al., 2014). While this background is available, the direct action of ghrelin on the liver oscillator is unexplored to date, which would be a requirement to support the role of this hormone as a temporal messenger in the entrainment of circadian liver function.

Ghrelin actions are mediated by G-protein-coupled receptors known as growth hormone secretagogue receptors (GHS-Rs) or ghrelin receptors (Kaiya et al., 2013; Kojima et al., 1999). In Otophysi teleosts, two paralogue ghs-r genes have been identified (GHS-R1 and GHS-R2), which has been tetraploidized in the members of the Cyprininae subfamily (e.g. goldfish), resulting in the presence of four receptor subtypes (GHS-R1a1, GHS-R1a2, GHS-R2a1 and GHS-R2a2; Kaiya et al., 2010). Among the different GHS-R subtypes, GHS-R1a seems to be involved in most of the ghrelin physiological actions (Gnanapavan et al., 2002; Kaiya et al., 2013; Yin et al., 2014). This receptor is mainly coupled to the phospholipase C (PLC)/protein kinase C (PKC) pathway (Kojima et al., 1999; Yin et al., 2014), but it can also trigger alternative intracellular pathways, including the adenylyl cyclase (AC)/protein kinase A (PKA) pathway (Kohno et al., 2003).

Considering the relevance of the liver in synchronizing feeding inputs in both mammals and fish (supporting its role as a food-entrainable oscillator), and the role of ghrelin as a signal of the feeding–fasting cycle, it is plausible that this hormone might link energy status and the circadian system by acting as an input of the hepatic oscillator. The similar anatomical distribution of GHS-R1a and per1b expression in the forebrain and gut of goldfish (Sánchez-Bretaño et al., 2015a,c) supports the possible role of ghrelin as an input of circadian clocks in this teleost. To test this possible role of ghrelin, the present study investigated the possible direct regulatory role of ghrelin on the hepatic molecular clock of goldfish. To achieve this aim, we first verified the expression of the ghrelin receptor GHS-R1a in the liver of this teleost by in situ hybridization. Second, we demonstrated that ghrelin modulates the in vitro expression of hepatic clock genes (gper1a, gper1b, gper2a, gper3, gbmal1a, gclock1a and grev-erbα) in a concentration-dependent manner, and that these effects are counteracted by the ghrelin antagonist [d-Lys3]-growth hormone releasing peptide-6 ([d-Lys3]-GHRP-6). Finally, we investigated the possible involvement of the PLC/PKC and AC/PKA intracellular pathways in ghrelin-induced modulation of clock genes in the goldfish hepatic clock, providing for the first time in vertebrates a putative mechanism by which ghrelin can act as an input to circadian clocks.

For the anatomical experiments, goldfish (2.0±0.5 g; n=7) obtained from a local supplier (Rennes, France) were maintained in 60 l aquaria with filtered and aerated fresh water (22±1°C) under a 12 h light:12 h dark photoperiod (lights on at 09:00 h). Fish were daily fed at 11:00 h (Zeitgeber time 2, ZT2) with food pellets (1% body mass, M_b; Novo GranoMix, JBL, GmbH and Co., Neuhofen, Germany). Goldfish (48 h fasted) were anaesthetized at ZT2 with 1 ml l⁻¹ phenoxyethanol (ICN Biomedicals Inc., Irvine, CA, USA) and killed (overdose of anaesthetic). Then, the whole fish was immersed overnight in 4% paraformaldehyde diluted in 0.1 mol l⁻¹ phosphate-buffered saline (PBS, pH 7.4). The following day, the liver was removed and post-fixed for 3 h in the same solution. Samples were cryoprotected overnight with 30% sucrose (MP Biomedical, LLC, Illkirch, France), included in the frozen section medium (Richard-Allan Scientific™ Neg-50, Thermo Shandon Scientific, Cheshire, UK) and stored at −80°C.

For in vitro studies, goldfish (7.2±0.5 g) obtained from a local supplier (Madrid, Spain) were maintained in 60 l aquaria with filtered and aerated fresh water (22±1°C) under a 12 h light:12 h dark photoperiod (lights on at 08:00 h). Fish were daily fed at ZT2 with food pellets (1% M_b; Bioflakes, Sera Pond, Heidelberg, Germany). On the day of the experiment, non-fed goldfish were anaesthetized in MS-222 (0.175 g l⁻¹, Sigma Aldrich, Carlsbad, CA, USA) at ZT2. Then, animals were killed (overdose of anaesthetic) and the liver was quickly sampled and distributed in the different wells (15 mg liver per well) of sterile 24-well culture plates.

Fish handling procedures complied with International Standards for the Care and Use of Laboratory Animals and were in accordance with the Guidelines of the European Union Council (2010/63/EU) for the use of research animals.

The probes for in situ hybridization were synthesized from plasmids (pCR™4-TOPO^® vector, Invitrogen, Carlsbad, CA, USA) containing 979 bp of goldfish ghs-r1a (Sánchez-Bretaño et al., 2015c). This probe targets a common fragment of goldfish ghs-r1a1 and ghs-r1a2 (GenBank accession numbers AB504275.1 and AB504276.1). Plasmids with the insert were linearized with SpeI and NotI, and antisense and sense mRNA probes were obtained with digoxigenin (DIG) RNA labelling mix (Roche Diagnostics, Mannheim, Germany) by in vitro transcription with T7 and T3 RNA polymerases (Promega, Madison, WI, USA). The specificity of the probes was confirmed with parallel series of slides hybridized with the respective sense RNA probes.

The liver obtained and stored as described above was placed in TissueTek and sectioned at 8 μm using a cryostat. Sections were mounted on superfrost slides. In situ hybridization was performed as previously described (Escobar et al., 2013) with minor modifications. In brief, cryostat sections were washed twice in PBS over a period of 10 min before post-fixing in Antigenfix (DiaPath, Martinengo, Italy) for 20 min. Then, sections were treated for 5 min at 37°C with proteinase K (2 μg ml⁻¹, Sigma, Steinheim, Germany) diluted in PBS, and fixed in 4% paraformaldehyde for 15 min. Sections were rinsed twice in 2× standard saline citrate (SSC). Hybridization was performed at 65°C overnight in a humidified chamber using 100 μl hybridization buffer (50% deionized formamide, 2× SSC, 5× Denhardt's solution, 50 μg ml⁻¹ yeast tRNA, 4 mmol l⁻¹ EDTA, 2.5% dextran sulfate) containing the DIG-labelled probe (3 μg ml⁻¹). After hybridization, slides were washed successively in 2× SSC at 65°C (2×30 min), 2× SSC/50% formamide at 65°C (2×30 min), 0.2× SSC (1×15 min) and 0.1× SSC (1×15 min) at room temperature. Slides were then washed in 100 mmol l⁻¹ Tris-HCl (pH 7.5) containing 150 mmol l⁻¹ NaCl for 10 min, washed in the same buffer containing 0.1% Triton X-100 and 0.5% skimmed milk powder (2×30 min), and incubated overnight at room temperature with anti-DIG alkaline phosphatase Fab fragments (1:2000; Roche Pharma, Mannheim, Germany). The next day, slides were incubated for 4.5 h with an HNPP (2-hydroxy-3-naphtoic acid-2′-phenylanilide phosphate)/FastRED detection kit (Roche Pharma), according to the manufacturer's instructions. Finally, slides were cover slipped with Vectashield mounting medium containing 4′,6-diamidino-2-phenylindole (DAPI; Vector Laboratories, Burlingame, CA, USA). Slides were observed with an epifluorescence microscope (Olympus Provis, equipped with a DP71 digital camera). Images were processed with either Olympus Analysis or Zeiss Cell software. Micrographs were generated in the TIFF format and adjusted linearly for light and contrast using Photoshop CS6 before being assembled on plates.

Liver cultures were prepared as previously described (Sánchez-Bretaño et al., 2016). A portion of liver from a different fish was used in each experimental group (n=6 fish per group). Liver portions were pre-incubated for 2 h in 1 ml of control medium (15 mg liver ml⁻¹ per well, quantified as 15 µl of tissue; Sánchez-Bretaño et al., 2016). The control medium consists of Dulbecco's modified Eagle’s medium (DMEM; 17.3 g l⁻¹ Sigma Aldrich) modified for fish tissues by adding NaHCO₃ (3.7 g l⁻¹) and antibiotics (10 ml l⁻¹ penicillin–streptomycin and 500 mg l⁻¹ gentamicin; Sigma Aldrich). After the 2 h pre-incubation period, medium was replaced by 1 ml of fresh DMEM containing the respective vehicle (control groups) or the corresponding drug (treated groups). Incubation time was either 1 or 5 h depending on the experiment (see figures). The liver cultures were maintained under constant dim light and temperature (21±1°C) conditions. At the end of each culture time, liver samples were collected, quickly frozen in liquid nitrogen and maintained at −80°C until clock gene expression was quantified.

Stock solutions were prepared and stored at 4°C until used. The 17-amino acid isoform of goldfish ghrelin [GTS(octanoyl)FLSPAQKPQGRRPP; Bachem, Bubendorf, Switzerland] and the PKA inhibitor H89 (Sigma Aldrich) were prepared in distilled water at a concentration of 2 and 15 mmol l⁻¹, respectively. Stock solutions of the PLC inhibitor U73122 (Tocris Bioscience, Bristol, UK), the ghrelin antagonist [d-Lys³]-GHRP-6 (Bachem, Bubendorf, Switzerland) and the AC activator forskolin (Sigma Aldrich) were prepared in absolute ethanol at 1, 10 and 15 mmol l⁻¹ concentrations, respectively. Stock solutions of the PKC inhibitor chelerythrine chloride (CHEL; Sigma Aldrich) and the PKC activator phorbol 12-myristate 13-acetate (PMA; Sigma Aldrich) were prepared in dimethyl sulfoxide (DMSO) at 5 and 20 mmol l⁻¹ concentrations, respectively. All stock solutions were diluted in DMEM to reach the required final concentrations just before use. Whenever the experimental design required the use of the antagonist or an inhibitor (i.e. [d-Lys³]-GHRP-6, U73122, CHEL, H89), the drug was added to the culture medium 15 min prior to the addition of the respective activator of gene expression (i.e. ghrelin-17, PMA, forskolin).

Clock gene expression (gper1a, gper1b, gper2a, gper3, gbmal1a, gclock1a and grev-erbα) was quantified by real-time quantitative PCR (RT-qPCR) using gβ-actin as a reference gene, as previously described (Nisembaum et al., 2014). Specific primers and GenBank reference numbers are shown in Table 1. The RNA extraction (TRI^® Reagent method, Sigma Chemical, Madrid, Spain), DNase treatment (Promega), cDNA synthesis (Invitrogen) and RT-qPCR reactions (iTaq™ SYBR^® Green Supermix in a CFX96™ Real-Time System, BioRad Laboratories, Hercules, CA, USA) were carried out following the manufacturer’s instructions with minor modifications (Nisembaum et al., 2014). Total RNA (1 µg) was reverse transcribed and PCR reactions were developed in a final volume of 10 µl (2 µl of cDNA per sample). PCR conditions were 30 s at 95°C, and 40 cycles consisting of 5 s at 95°C and 30 s at 60°C for all genes. Calibration curves were made of serial dilutions of cDNA, exhibiting efficiencies of around 100%. Specificity of amplifications was ensured by melting curves and tested by agarose gels. The relative mRNA expression was determined by the ΔΔCt method (Livak and Schmittgen, 2001).

Data obtained from the ΔΔCt method were logarithmically transformed in order to normalize the variance and to obtain homoscedasticity. A probability level of P<0.05 was considered statistically significant. Analysis of the relative expression changes in the ghrelin concentration–response curves was conducted using one-way ANOVA followed by the post hoc Student–Newman–Keuls (SNK) test. A two-way ANOVA followed by a post hoc SNK was used when the interaction of activators (ghrelin-17, PMA, forskolin) and inhibitors ([d-Lys³]-GHRP-6, U73122, CHEL, H89) was studied (see details in the figure legends).

The ghrelin receptor ghs-r1a was found to be widely expressed in the goldfish liver (Fig. 1). Almost all the hepatic cells showed a strong ghs-r1a signal surrounding the nucleus (Fig. 1A,C), while the sense riboprobes yielded no signal (Fig. 1B,D), supporting the specificity of the obtained signal in the goldfish liver.

Ghrelin modified the expression of some clock genes in the goldfish liver in vitro (Fig. 2). With the exception of gper1b, all the clock genes studied were induced by 0.1–10 nmol l⁻¹ ghrelin at 1 h post-treatment (for gclock1a and gper3, induction was only observed with ghrelin concentrations greater than 1 nmol l⁻¹). The highest induction was found for gper1a transcripts, which were increased around 7-fold, while the induction of gper2a and gper3 expression was smaller (around 2-fold). Expression of genes from the positive limb, i.e. gbmal1a and gclock1a, and from the auxiliary loop, grev-erbα, was also induced by ghrelin (around 4- to 5-fold). All the ghrelin-evoked upregulations of clock gene expression observed at 1 h were diminished after 5 h of ghrelin exposure, and even disappeared in the case of gper3 and grev-erbα. Only in the case of gper1 was exposure of the liver to ghrelin (10 nmol l⁻¹) for 5 h a requirement for a 3-fold increase of transcripts to be observed.

The specificity of the ghrelin-evoked induction of clock gene expression was tested by using a ghrelin antagonist, [d-Lys³]-GHRP-6 (Fig. 3). The presence of this antagonist in the culture medium did not modify the expression of clock genes by itself, with the exception of gper2a, gclock1a and gper3, levels of which were modified by the antagonist after a 5 h incubation (gper2a, gclock1a) or after a 1 h incubation (gper3). Pre-incubation of liver samples with the ghrelin antagonist abolished the stimulatory effect of ghrelin on gper1a, gper2a, gbmal1a and grev-erbα expression. This blocking effect was observed after 1 h of exposure to the antagonist, coincident with the time when the inductions evoked by ghrelin were the highest. This blocking effect of the antagonist was also observed at 5 h in the case of gper1b. In the case of gclock1a and gper2a, the counteraction of the ghrelin-evoked induction was not observed at 5 h, probably because of the significant increase of transcripts observed in the presence of the antagonist alone at this time.

Pre-incubation (15 min) with the PLC inhibitor U73122 prior to the addition of ghrelin totally abolished the induction of hepatic clock gene expression (gper1a, gper1b, gper2a, gbmal1a and grev-erbα) evoked by the presence of ghrelin in the culture medium for 1 h (Fig. 4). In the case of gclock1a and gper3, the direct stimulatory effect of U73122 on mRNA levels found at 1 h post-incubation probably hampered the blocking effects of the PLC inhibitor on ghrelin induction of gene expression. Ghrelin effects at longer exposures (5 h) were blocked by U73122 only in the case of gper1a, gper1b and gper2a, while the inhibitor by itself induced gbmal1a, gclock1a and grev-erbα expression. The basal expression of the remaining clock genes was not affected by U73122 at either of the tested concentrations (1 and 10 µmol l⁻¹).

The possible role of PKC in the regulation of clock genes by ghrelin was assessed by the use of CHEL, a specific inhibitor of this kinase (Fig. 5). Induction of expression by ghrelin in cultured liver was totally blocked by pre-treatment with CHEL for the majority of the studied clock genes (gper1a, gper2a, gper3, gbmal1a and gclock1a), with the exception of grev-erbα, where the induction of mRNA levels produced by ghrelin was only partially blocked. This PKC inhibitor had minor effects on basal clock gene expression at the tested concentrations (1 and 5 μmol l⁻¹), except for a slight reduction of gclock1a transcripts (Fig. 5). As a positive control for the involvement of PKC in clock gene expression, the liver was treated with a direct PKC activator, PMA, for 1 h (Fig. 5). This activator exerted slight effects on clock gene expression at a low concentration (50 nmol l⁻¹), but significantly induced gper1a, gper3, gbmal1a, gclock1a and grev-erbα expression at a higher concentration (200 nmol l⁻¹; Fig. 5). Pre-treatment of liver with CHEL (5 μmol l⁻¹) prior to the addition of PMA diminished the induction of gper1a, gper3, gbmal1a and grev-erbα produced by the activator. Neither ghrelin nor PMA modified gper1b expression after 1 h of treatment.

Forskolin, an AC activator, induced the expression of gper2a, gper3, gclock1a and grev-erbα while it had no effect on the rest of the clock genes analysed after 1 h of treatment (Fig. 6). Pre-treatment with the PKA inhibitor H89 abolished forskolin effects in all cases. High concentrations of H89 (200 μmol l⁻¹) decreased the expression of gper3 in cultured liver by itself. The stimulatory effect of ghrelin on gbmal1a and grev-erbα expression was partially blocked by pre-treatment with H89 (significant interaction in two-way ANOVA), but this PKA inhibitor did not block ghrelin-induced upregulation of gper1a and gper2a expression. In the case of gper3, the down-regulatory effect of H89 and the stimulatory effect of ghrelin on its expression seem to be independent actions.

In the present work, we report evidence for a direct effect of ghrelin on clock gene expression in the liver of a teleost. We found that ghrelin (acyl-ghrelin-17) induces clock gene expression via its receptor located in hepatic cells, and that the intracellular PLC/PKC and AC/PKA pathways are involved in this direct effect. This is the first report that links ghrelin with the molecular basis of functional metabolic oscillators in vertebrates.

First, the present results show that ghrelin induces the mRNA expression of the positive (gbmal1a, gclock1a) and negative (pger1a, gper2a, gper3) elements of the main loop of the molecular clock machinery, as well as mRNA expression of grev-erbα (auxiliary loop) over short time periods (1 h), and of gper1b over longer periods (5 h) in cultured goldfish liver. This induction of hepatic clock gene expression by ghrelin seems to be an acute and fast response that disappears after 5 h of exposure to the hormone (except for gper1b). These results are in accordance with previous studies in goldfish, where the acute intraperitoneal administration of ghrelin-19 induced per gene expression (gper1a, gper2 and gper3) in the liver at 1 h post-injection, an effect that disappeared 3 h later (Nisembaum et al., 2014). The delay observed in gper1b induction by ghrelin could be due to a different sensitivity of this clock gene to this hormone compared with the other per genes present in the goldfish liver.

The ghrelin receptor antagonist [d-Lys³]-GHRP-6 partially blocked the ghrelin-evoked effects on hepatic clock genes in cultured liver, in agreement with previous in vivo results obtained from this teleost (Nisembaum et al., 2014). We therefore suggest the involvement of a hepatic ghrelin receptor in the majority of the observed actions of ghrelin as a modulator of clock gene expression. The specificity of the ghrelin effect on gclock1a and gper3 remains unsolved as the ghrelin receptor antagonist increases basal levels of these two clock genes by itself. In support of such direct actions of ghrelin on liver cells, our results show for the first time in vertebrates a wide distribution of the ghrelin receptor ghs-r1a in hepatic cells. This is in agreement with the previous identification by PCR of the ghrelin receptor in the liver of some teleosts, including goldfish (Cai et al., 2015; Kaiya et al., 2010).

The GHS-R1a ghrelin receptor subtype seems to be linked to the PLC/PKC intracellular transduction pathway (Chen et al., 2009; Grey and Chang, 2011; Yin et al., 2014). A circuitry that includes the activation of PLC and the regulation of different transcriptional factors has previously been suggested as a determinant in the modulation of the circadian system in mammals. PKC is involved in the phase shift of the firing rate of suprachiasmatic nucleus cells in vitro (Schak and Harrington, 1999). The PLC/PKC system also mediates light (Bonsall and Lall, 2013; Lee et al., 2007) and food entrainment (Zhang et al., 2012), and is involved in the effects of melatonin and neuropeptide Y on phase advance in rodents (Biello et al., 1997; McArthur et al., 1997). The in vitro induction of clock gene expression in goldfish liver by the activation of the PLC/PKC pathway (by PMA) indicates that this intracellular pathway is involved in the functionality of liver circadian oscillators in teleosts. Furthermore, the present results show that this intracellular pathway is involved in the induction of hepatic clock genes by ghrelin, as this effect is blocked by pre-treatment with PLC or PKC inhibitors (U73122 and CHEL, respectively). In the case of gclock1a and gper3, the PLC inhibitor (like the ghrelin antagonist) increased the basal expression of these genes, but the inhibitor of PKC (CHEL) totally blocked the ghrelin effects. Thus, we suggest that the PKC is at least one of the mechanisms underlying the ghrelin induction of these clock genes. It should be noted that our results demonstrate the involvement of the PLC/PKC system in ghrelin actions at 1 h, but other mechanisms may be involved at longer times. Overall, these experiments indicate a relationship among the presence of ghrelin, the activation of ghrelin receptors in the hepatocytes and the signal transduction via the PLC/PKC pathway in order to induce clock gene expression. Interestingly, we found that liver clock genes show similar responses to both ghrelin and the PKC activator (PMA), with a high induction of gper1a, intermediate sensitivity for gper2a, gper3, gbmal1a and gclock1a, and an evident insensitivity of gper1b at 1 h post-treatment. From the present results, it seems that the PKC pathway underlies the regulatory effect of ghrelin on clock genes in goldfish liver.

The intracellular Gs/AC/PKA pathway is also involved in the activation of the GHS-R1 ghrelin receptor in mammals (Kohno et al., 2003). In cultured goldfish liver, the direct activation of the AC/PKA pathway by forskolin produced a slight induction of some clock genes (gper2a, gper3 and gclock1a) and a pronounced increase (4-fold) of grev-erbα transcript levels, suggesting a dependence on this intracellular signalling pathway. The specificity of forskolin is evidenced by the blocking of its effects with the PKA inhibitor H89, supporting the involvement of the AC/PKA pathway in the regulation of the liver circadian oscillator in this teleost. In agreement with our findings in goldfish liver, the AC/PKA intracellular pathway has been linked to the molecular functioning of endogenous clocks in some phylogenetically distant species. cAMP, which activates PKA, is a stabilizer and modulator of per gene transcripts in the fruit fly Drosophila melanogaster (Li et al., 2014) and mammals (Hastings et al., 2014; Motzkus et al., 2000; Zmrzljak et al., 2013). Moreover, PKA induces Per1 expression in humans (Motzkus et al., 2007) and adjusts endogenous clocks in the presence of light pulses (Tischkau et al., 2000), and its inhibition delays the mammalian clock (Lee et al., 1999). Our results in fish, in agreement with these previous reports in mammals, suggest that the possible functional role of the AC/PKA pathway (as PLC/PKC) in the regulation of clock genes is conserved throughout phylogeny.

Despite the fact that the AC/PKA pathway seems to be involved in the regulation of hepatic clock gene expression, current results suggest that the effects of ghrelin on clock genes are independent of this intracellular pathway, except for grev-erbα and probably gbmal1a. The involvement of the AC/PKA pathway in the increase in grev-erbα expression is supported by the induction of this gene by both forskolin and ghrelin, and the counteraction of this induction by the PKA inhibitor. The blockade of the ghrelin-evoked increase in gbmal1a levels with the PKA inhibitor needs to be further explored with cAMP analogues or AC inhibitors, given that forskolin did not induce expression of this gene. The relationship between the AC/PKA pathway and the circadian system has previously been suggested in the signal transduction of the light–dark cycle by cAMP response elements (CRE) (Ginty et al., 1993; Motzkus et al., 2007; Travnickova-Bendova et al., 2002), which play a key role in the light-entrainable oscillators. Ghrelin, as a food intake regulator and energy balance signal, is expected to be mainly related to the food-entrainable oscillators (such as the liver). This could justify the lower relevance of the AC/PKA intracellular pathway (compared with the PLC/PKC pathway) in the transduction of this hormonal signal to the hepatic clock.

In the present study, we used ghrelin-17, the biologically active isoform of ghrelin, which exerts orexigenic actions in goldfish (Kang et al., 2011; Miura et al., 2009). The range of ghrelin concentrations used (0.1–10 nmol l⁻¹) has previously been reported as physiologically significant in goldfish cultured pituitary, where different ghrelin isoforms (ghrelin-12 and ghrelin-19) induced luteinizing hormone and growth hormone release (Grey and Chang, 2013; Unniappan and Peter, 2004). The fact that two isoforms, ghrelin-19 (the isoform used in in vivo studies in goldfish; Nisembaum et al., 2014) and ghrelin-17 (current experiments), modulate clock gene expression in liver suggests that both forms of ghrelin might play physiological roles in fish, and emphasizes the relevance of this hormone as an input of the hepatic oscillator. In addition, the observed effect of ghrelin on clock gene expression in cultured liver shows key properties (acute and short time effects) of synchronizing agents. A similar fast and acute effect on per2 expression has been established for light synchronization in zebrafish (Vatine et al., 2011).

Our results strongly suggest that ghrelin modulates the clock machinery in the liver, a key target for the interplay between the circadian system and metabolism. Considering the well-known role of this peptide in the signalling of energy status, it is plausible to suggest that ghrelin may be acting as a link in the regulation of both energy balance and the circadian system in teleosts. The high levels of circulating ghrelin (e.g. during starvation) might modify clock gene expression in the hepatic oscillator in an acute but strong manner. Then, this hormone may be acting as an input to reset the hepatic metabolism via modulation of the hepatic oscillator entrainable by food. The fact that most of the genes that show circadian oscillations in the liver are related to metabolic processes (Oishi et al., 2005; Reddy et al., 2007) supports the cross-talk between signals of nutritional reserves (such as ghrelin) and the circadian system in order to maintain metabolic balance in the liver and even in the whole organism.

In conclusion, the present results demonstrate for the first time in vertebrates the direct effect of ghrelin on the modulation of the molecular machinery of the hepatic oscillator by inducing the expression of some clock genes via the intracellular PKC/PLC and to a lesser extent the AC/PKA pathways. Whether the acute response of clock genes to ghrelin observed in vitro confirms a physiological role of this orexigenic hormone as an endogenous input of the circadian system in fish remains to be elucidated. As ghrelin induced both negative and positive clock genes, it may be that this hormone leads to the disruption of the hepatic rhythmicity, which would make the synchronizing action of ghrelin on the hepatic clock less likely. However, it is also possible that ghrelin sensitizes the liver to other signals (i.e. the liver could respond differentially to other signals in the presence or absence of ghrelin). This interesting but unexplored physiological role of ghrelin deserves to be further studied.