Orally administered fatty acids enhance anorectic potential but do not activate central fatty acid sensing in Senegalese sole post-larvae

In previous studies, we characterized the presence and function of fatty acid (FA)-sensing systems in the hypothalamus of the teleost fish rainbow trout (Oncorhynchus mykiss) (Librán-Pérez et al., 2012, 2013, 2014a,b, 2015) and Senegalese sole (Solea senegalensis Kaup) (Conde-Sieira et al., 2015), which are comparable in many aspects to those in mammals (Blouet and Schwartz, 2010). These systems respond to increased levels of unsaturated long-chain FAs such as oleate (OA, C18:1 n-9) and correlate with the expression of neuropeptides involved in the control of food intake (Librán-Pérez et al., 2012, 2014a). FA-sensing mechanisms are based on: (i) FA metabolism via carnitine palmitoyltransferase 1 (CPT-1) inhibition to import FA-coenzyme-A (CoA) into the mitochondria for oxidation; (ii) binding to FA translocase (FAT/CD36) and modulation of the transcription factors peroxisome proliferator-activated receptor type α (PPARα) and sterol regulatory element-binding protein type 1c (SREBP1c); and (iii) production of reactive oxygen species in the mitochondria, resulting in inhibition of ATP-dependent inward rectifier potassium channel (K_ATP) activity (Soengas, 2014). The activation of these systems relates to the inhibition of the orexigenic factors agouti-related protein (AgRP) and neuropeptide Y (NPY), and the enhancement of the anorexigenic factors cocaine- and amphetamine-related transcript (CART) and pro-opiomelanocortin (POMC) (Librán-Perez et al., 2012, 2014a; Conde-Sieira et al., 2015), ultimately leading to decreased food intake.

In contrast to what is known in mammals, we also provided evidence in Senegalese sole juveniles for the activation of FA-sensing systems by polyunsaturated FAs (PUFAs). Furthermore, hypothalamic FA-sensing systems in Senegalese sole were activated by α-linolenate (ALA, C18:3 n-3) but not by eicosapentaenoate (EPA, C20:5 n-3) (Conde-Sieira et al., 2015), indicating that the response might be specific to certain PUFAs. Therefore, both the level of unsaturation and the chain length of the FA seem to be important factors for the hypothalamic sensing capacity, at least in Senegalese sole juveniles, but are probably not the only factors involved. The finding that FA-sensing systems in Senegalese sole respond to changes in PUFA levels may relate to the general importance of PUFAs (especially of the n-3, but also of the n-6 series) in marine fish (Tocher, 2003).

Studies in fish regarding FA-sensing systems and control of food intake have only been carried out in juveniles and adults so far (Libran-Pérez et al., 2012, 2014a,b; Conde-Sieira et al., 2015). However, it is of great importance to clearly establish when appetite regulation becomes functional in fish larvae, particularly in species of commercial interest to aquaculture. Very little is known concerning the neural regulation of feeding and appetite during early larval stages, with very few studies describing changes in the expression of brain neuropeptides (Kortner et al., 2011; Gomes et al., 2015). As far as we know, there are no available studies assessing the impact of specific nutrients on the regulation of food intake in fish larvae and post-larvae. In a preliminary study with Senegalese sole larvae (Bonacic et al., 2016), we observed differences in the intake of Artemia prey that had been enriched with different oils (olive, soybean, linseed and cod liver oil); specifically, there was a higher ingestion of prey enriched with cod liver oil. However, the expression of neuropeptides in whole larvae could not explain such differences, even though larvae on the cod liver oil-enriched Artemia diet had the most dissimilar gene expression pattern of the four dietary groups. Similar findings were obtained in the same experiment with post-larvae (although effects on food intake were less clear because of a higher variability). At this life stage, even though there was still no correlation between neuropeptide expression and diet intake, gene expression in the body and head of post-metamorphic larvae was generally more in accordance with the putative anorexigenic or orexigenic function of the analysed genes, which could suggest a slightly more developed regulatory system. The oils used were especially rich in specific FAs such as OA in olive oil, linoleate (LA, C18:2 n-6) in soybean oil, ALA in linseed oil and EPA in cod liver oil. However, other nutrients present in the diet could also have interacted to produce the effect, so the observed responses could not be unequivocally attributed to any of these specific FAs.

In the current study, we aimed to elucidate the possible effects of different specific FAs on neural FA-sensing systems and neuropeptides involved in the control of food intake in the early life stages of fish. To achieve these objectives, we orally administered different solutions containing OA, LA, ALA or EPA to Senegalese sole post-larvae. Because of the small size of the animals, which prevented dissection of specific brain regions, we evaluated changes in mRNA abundance of neuropeptides related to the control of food intake, such as agrp2, npy, pomca, pomcb, cart1a, cart1b, cart2b and cart4 in head sections. Furthermore, we also evaluated changes in mRNA abundance of transcripts related to putative FA-sensing systems based on: (1) FA metabolism, such as acetyl-CoA carboxylase (acc), ATP-citrate lyase (acly), cpt-1 and fatty acid synthetase (fas); (2) binding to FAT/CD36, such as fat/cd36, liver X receptor α (lxrα), pparα and srebp1c; and (3) mitochondrial activity, such as inward rectifier K⁺ channel pore type 6.x (kir6.x) and sulfonylurea receptor (sur).

Newly hatched Senegalese sole larvae were obtained from Stolt Sea Farm S.A. (Carnota, A Coruña, Spain) and distributed into two 100 l cylindroconical tanks at a density of 50 larvae l⁻¹. The tanks were connected to a recirculation system (IRTAMAR^®) with 50% daily water renewal and regulated temperature (18–19°C), salinity (35 ppt) and dissolved oxygen (7.5 mg l⁻¹). The photoperiod was 16 h light:8 h dark, with a light intensity of <500 lx at the water surface. Feeding was performed twice a day (at 09:30 h and 17:30 h). Larvae were fed rotifers enriched with Algamac 3050 flake (Aquafauna Bio-Marine, Inc., Hawthorne, CA, USA), according to the manufacturer's instructions, at 10 rotifers ml⁻¹ from 1 to 9 days post-hatching (dph). Algamac-enriched Artemia metanauplii were introduced at 7 dph and fed until 12 dph, in quantities gradually increasing from 0.5 to 3 metanauplii ml⁻¹. From 13 dph onwards, to avoid any influence from the high levels of lipids in the enrichment product, larvae were fed only non-enriched Artemia metanauplii (kept in clean seawater instead of the enrichment emulsion), at 4 metanauplii ml⁻¹ during the planktonic stage (until 20 dph). After settling of the majority of the larvae, frozen non-enriched Artemia were also added in increasing amounts, at 6–12 metanauplii ml⁻¹ in total, up to 40 dph. It is possible to feed Senegalese sole larvae on non-enriched Artemia without significantly affecting growth performance, given the relatively low requirements for lipids and essential fatty acids in this species compared with other marine species, particularly after settlement (Dâmaso-Rodrigues et al., 2010).

The experiments described here comply with the Guidelines of the European Union Council (2010/63/UE) and the Spanish Government (RD53/2013) for the use of animals in research, with protocols (including tube feeding, described below) approved by the ethics committee of IRTA.

Post-larvae were tube-fed one of four FA treatments on four consecutive days, from 36 to 39 dph. Test solutions consisted of the following pure free FAs: OA (≥99%), LA (≥99%), ALA (≥99%) or EPA (≥98.5%), all from Sigma-Aldrich (Madrid, Spain). On each preceding day, a group of post-larvae were taken from the rearing tanks and washed in a submerged 1 mm net (to remove any uneaten Artemia), transferred to a 4 l tray with clean water from the recirculation system and moved to the experimental room (maintained at 18°C), where they were kept with aeration and fasted for 18 h overnight. Post-larvae were gently anesthetized with 1 ml l⁻¹ of tricaine methanesulfonate (MS-222, at 20 g l⁻¹). They were then individually tube-fed either a saline (control) or test FA solution under a dissecting microscope using a plastic capillary (0.19 mm inner diameter, Sigma-Aldrich) attached to a Nanoliter 2000 injector and micromanipulator (World Precision Instruments, Sarasota, FL, USA), to deposit ∼46 nl of the solution into the post-larvae foregut, as described in Rønnestad et al. (2001). Each post-larva was then transferred into an incubation system of plates containing 10 ml wells with clean seawater. Four post-larvae (N=4) were tube-fed per treatment and time point (3 and 6 h incubation, to allow for luminal absorption of FAs; Morais et al., 2005) and another four post-larvae (N=4) were tube-fed a saline solution and then immediately killed (control, reflecting the basal condition), every 4 consecutive days. The whole process took under 2 min per animal but, in order to avoid interference from the time of tube feeding, each day the order of administering the treatments was changed. Hence, each day a larva that was tube-fed a FA treatment in a different order was added to the sample pools, so that every sample (pool of four post-larvae) contained animals tube-fed at slightly different times, equally for each treatment. Larvae were killed by a lethal dose of MS-222, washed and transferred to an Eppendorf tube containing RNAlater stabilization buffer (Ambion, Life Technologies, Madrid, Spain), agitated at 4°C for 24 h, and then stored at −80°C. The post-larvae were later dissected on ice under a dissecting microscope, in order to separate and remove the body from the head section, which was then processed for molecular analysis of selected neuropeptide genes and genes involved in central FA-sensing pathways.

Total RNA was extracted from the head sections by homogenization in TRIzol (Ambion) with 50 mg of 1 mm diameter zirconium glass beads in a Mini-Beadbeater (Biospec Products Inc., Bartlesville, OK, USA). Solvent extraction was performed following the manufacturer's instructions and RNA quality and quantity were assessed by gel electrophoresis and spectrophotometry (NanoDrop2000, Thermo Fisher Scientific, Madrid, Spain). Total RNA (2 µg per sample) was reverse transcribed into cDNA using the High-Capacity cDNA RT kit (Applied Biosystems, Life Technologies, Carlsbad, CA, USA), following the manufacturer's instructions, but using a mixture of random primers (1.5 µl as supplied) and anchored oligo-dT (0.5 µl at 400 ng µl⁻¹, Eurogentec, Cultek, S.L., Madrid, Spain). Negative controls (containing no enzyme) were performed to check for genomic DNA contamination.

Gene expression levels were determined by real-time quantitative RT-PCR (qPCR) using the iCycler iQ (Bio-Rad, Hercules, CA, USA). Analyses were performed on 5 µl of diluted (1/50) cDNA using the SsoAdvanced™ Universal SYBR^® Green Supermix (Bio-Rad), in a total PCR reaction volume of 15 µl, containing 120–500 nmol l⁻¹ of each primer. Sequences of the forward and reverse primers used for each gene expression assay are shown in Table 1. Relative quantification of the target genes was done using elongation factor 1 alpha (ef1α) and ubiquitin (ubq) as reference genes. Thermal cycling was initiated with incubation at 95°C for 2 min; 35 steps of qPCR were performed, each one consisting of heating at 95°C for 15 s, 30 s at each specific annealing temperature (Table 1) and 30 s at 72°C. Following the final PCR cycle, melting curves were systematically monitored (temperature gradient at 0.5°C s⁻¹ from 55 to 94°C) to ensure that only one fragment was amplified. Samples without reverse transcriptase and samples without cDNA were run as negative controls, which were indeed negative, and confirmed no amplification of genomic DNA. Relative expression of the target genes was calculated using the delta-delta CT method (2^−ΔΔCT), following Pfaffl (2001).

Comparisons among groups were carried out using the statistical package SigmaStat (Systat Software, Inc.) with two-way ANOVA, with FA treatment and time as main factors. In cases where a significant effect was noted, post hoc comparisons were carried out by a Student–Newman–Keuls test. Differences were considered statistically significant at P<0.05.

The mRNA abundance of neuropeptides is shown in Fig. 1. The levels of npy (Fig. 1A) decreased after treatment with LA (6 h), ALA (3 and 6 h) and EPA (6 h) compared with the control; the levels in LA-treated fish after 6 h were also significantly reduced compared with the 3 h time point. The expression of pomcb (Fig. 1D) increased postprandially with all FA assessed, compared with the control, and then decreased in OA-treated fish from 3 h post-treatment to 6 h post-treatment. The mRNA levels of cart1b (Fig. 1F) were reduced in the ALA-treated group compared with the remaining groups, 6 h after FA administration. The expression of cart4 (Fig. 1H) increased after treatment with OA (6 h), ALA (6 h), and EPA (3 and 6 h) compared with the control, and further increased in OA- and EPA-treated fish from 3 h post-treatment to 6 h post-treatment. No significant changes were observed for agrp2 (Fig. 1B), pomca (Fig. 1C), cart1a (Fig. 1E) and cart2b (Fig. 1G) mRNA levels.

The expression of genes related to FA transport and metabolism is shown in Fig. 2. The mRNA levels of fat/cd36-pg4l (Fig. 2B) increased 6 h after treatment with OA, and this value was also higher than that for the same treatment at 3 h. The expression of acc (Fig. 2C) decreased after treatment with OA (6 h), ALA (3 h) and EPA (3 and 6 h); in OA-treated fish, acc mRNA levels were reduced from 3 h post-treatment to 6 h post-treatment. The levels of fas (Fig. 2E) increased after treatment with OA (3 and 6 h), LA (3 h), ALA (3 and 6 h) and EPA (3 h), but then decreased from 3 h post-treatment to 6 h post-treatment in all FA treatments. No significant changes were observed for fat/cd36-lmp2 (Fig. 2A) and acly (Fig. 2D).

The mRNA level of genes involved in mitochondrial activity and the K_ATP channel is shown in Fig. 3. The level of cpt1.1 (Fig. 3A) increased 6 h after treatment with OA compared with the control, and was also higher than that in the same treatment at 3 h. The expression of cpt1.2 (Fig. 3B) increased after treatment with ALA (3 and 6 h) and decreased after treatment with EPA (6 h) compared with the control; in OA-treated fish there was an increase from 3 h post-treatment to 6 h post-treatment. The mRNA abundance of cpt1.4 (Fig. 3C) increased in OA- (6 h), LA- (6 h), ALA- (6 h) and EPA-treated (3 and 6 h) fish compared with the control, and also in LA-treated fish from 3 h post-treatment to 6 h post-treatment. The expression of sur (Fig. 3E) increased after treatment with OA (3 h), LA (3 h), ALA (3 and 6 h) and EPA (3 h) compared with the control. No significant changes were observed for kir6.x (Fig. 3D).

The mRNA abundance of three transcription factors is shown in Fig. 4. The levels of pparα (Fig. 4A) increased after treatment with OA (3 h) and ALA (6 h) compared with the control, while its expression was reduced from 3 h post-treatment to 6 h post-treatment in OA- and EPA-treated fish. The expression of lxrα (Fig. 4C) was not different to the control for any of the treatments but was reduced from 3 h post-treatment to 6 h post-treatment in fish treated with OA and LA. In the case of srebp1c (Fig. 4B), there were no significant differences.

Oral treatment with different FAs induced changes in the mRNA abundance of some neuropeptides in sole post-larvae. Despite not all peptides displaying changes in their mRNA abundance, the changes observed generally suggest the existence of an anorectic response in fish post-larvae to the ingestion of different FAs. In particular, enhanced expression of anorexigenic neuropeptides (CART and POMC; at least one homologue – cart4 and pomcb) was observed in parallel with decreased transcript levels of the orexigenic npy neuropeptide (although no effect was observed in agrp2). A similar effect was also suggested in juveniles of the same species after intraperitoneal (i.p.) injection with FA (Conde-Sieira et al., 2015), although there were important differences in the responses of the different neuropeptides and FAs assessed. For instance, in the present experiment there was no response of cart2b or agrp2 to any treatment, in contrast to juvenile sole where cart2b and agrp2 were significantly regulated (increased and decreased, respectively) in the hypothalamus following i.p. treatment with OA or ALA (Conde-Sieira et al., 2015). Interestingly, cart2b was also significantly increased by the EPA treatment in juvenile sole, similar to the cart4 homologue in the present study. Dissimilarities in the results could potentially be due to differences in the maturity of the fish or in the methodology employed, i.e. oral versus i.p. administration of free FAs. In the present study, the orally administered FAs will have been absorbed through the intestinal mucosa, where they can be at least partly metabolized, or reacylated into neutral or polar lipids and incorporated into chylomicrons for transport, reflecting the normal pathways in a fish feeding on a complete diet (only skipping the digestion step). However, an increased anorectic potential was also observed in the hypothalamus of rainbow trout fed a lipid-enriched diet (Librán-Pérez et al., 2015), which could suggest a conserved mechanism in teleosts in response to the ingestion of FAs.

In spite of a few differences, the present results do not seem to indicate major discrepancies between the different FAs tested with respect to their potential to affect the anorectic response of Senegalese sole post-larvae. The treatment with OA induced an increased anorectic potential based on the increased expression of the anorectic peptides pomcb and cart4 postprandially. This increased anorectic potential after OA treatment is similar to that already observed in juvenile fish of the same species (Conde-Sieira et al., 2015), as well as in rainbow trout (Librán-Pérez et al., 2012, 2014a). Treatment with LA also stimulated the anorectic potential based on the increased mRNA abundance of pomcb and the decreased expression of the orexigenic neuropeptide npy. As far as we are aware, this is the first time that the effects of this FA have been assessed in any species. ALA intake similarly induced clear responses in the mRNA abundance of neuropeptides including increased expression of pomcb and cart4 and decreased expression of npy. The resulting anorectic potential is comparable to that observed in the hypothalamus of juvenile fish of the same species when fish were i.p. treated with the same FA (Conde-Sieira et al., 2015). EPA treatment induced basically the same response as ALA, which is very interesting as in juvenile sole i.p. treatment with EPA did not result in relevant changes in the expression of these peptides (only of cart2b; Conde-Sieira et al., 2015).

In a previous experiment, we analysed changes in gene expression of peripheral peptides and neuropeptides in Senegalese sole larvae and post-larvae after feeding with Artemia enriched with different oils, especially rich in the same FAs that were administered in the present study (Bonacic et al., 2016). However, it is difficult to correlate the results of the two experiments given that we previously evaluated more complex (complete) diets, probably with lower total amounts of the tested FA, and fed during the whole larval and post-larval stage, rather than a single meal as in the present study. The previous experiment showed that larvae and post-larvae fed a diet containing higher EPA levels (enriched with cod liver oil) tended to have a higher food intake than those in the remaining treatments, and hence we expected to see differences in neuropeptide expression between treatments. Nevertheless, this did not correlate with a higher orexigenic and/or lower anorexigenic potential in either of the two experiments. In fact, EPA-treated fish showed higher levels of the putatively anorexigenic cart4 than those in the remaining treatments in the present experiment. However, different classes of FA can have diverse effects on metabolism and energy homeostasis, and on the release of peripheral appetite-regulating peptides, which can potentially affect food intake through different routes (French et al., 2000; Lawton et al., 2000; Wang et al., 2002; Tocher, 2003; Relling and Reynolds, 2007; Parra et al., 2008; Soengas, 2014). Furthermore, a peptide such as CART, for instance, is known to have multiple physiological functions besides the regulation of feeding (e.g. regulation of energy metabolism) in mammals (Lau and Herzog, 2014). In teleosts, its roles have not been well documented but the retention of several duplicated genes (in sole, up to seven homologues have been reported) suggests the conservation of multiple functions, or even the development of new ones (Bonacic et al., 2015). In this respect, Bonacic et al. (2016) reported the transcriptional regulation of cart1b, which did not show a response to the ingestion of food in the brain of sole juveniles (Bonacic et al., 2015), but was affected by the FA composition of the diet in a manner consistent with an anorexigenic role (expression reduced in larvae fed the cod liver oil, EPA-rich diet). Similarly, cart1a and cart1b seemed to respond to the FA composition (lipid source) of the diet, correlating with food intake in juvenile sole (Bonacic et al., 2017). In the present study, the transcription of cart1a was not significantly regulated postprandially or by the FA treatment, which suggests that dietary effects may be exerted in these genes via long-term changes in metabolism and energy homeostatic status rather than FA sensing following a meal. In fact, FA-sensing systems did not appear to be activated in the present study (see below), supporting this hypothesis. In the case of cart1b, its expression was not affected 3 h after feeding with any FA treatment but was significantly reduced 6 h after administration of ALA, but not in any of the remaining treatments, including EPA (as might have been expected from Bonacic et al., 2016). The reasons for this response are elusive at the moment, but do not seem to be related to FA sensing. Therefore, although preliminary and still speculative, data are starting to emerge in fish suggesting different roles of CART homologues in modulating the feeding response to not only dietary lipid levels but also FA composition, and this should be further evaluated in future studies.

The presence of several mechanisms involved in FA sensing in the mammalian brain was investigated in Senegalese sole in response to the orally administered FA treatments, for the first time in a teleost at such an early stage of development. Highly variable responses were observed in this study, which might suggest that FA-sensing systems are not active in the post-larval stage, contrary to what was reported previously after i.p. administration of FA in sole juveniles (Conde-Sieira et al., 2015), as well as in rainbow trout (Librán-Pérez et al., 2012, 2013, 2014a). In the hypothalamus of juvenile sole, several FA systems were activated by OA and ALA treatment: (1) those based on FA metabolism (decrease in mRNA levels of acly, acc and fas); (2) mechanisms based on binding of FAs to FAT/CD36 and subsequent modulation of transcription factors (no effect on fat/cd36 but reduction of pparα, sbrepb1c and lxrα expression, although only in OA-treated fish in the case of lxrα); and (3) based on mitochondrial activity (decreased expression of kir6.x and sur). Furthermore, a differential effect of FAs depending on chain length and degree of saturation was described, with a saturated FA such as stearate lacking a response and EPA inducing only a few changes in the transcription of these genes, probably unrelated to a role in FA sensing (Conde-Sieira et al., 2015). In the present study, the only result that is in line with previous observations is the reduction in mRNA levels of acc in OA- (only at 6 h) ALA- (3 h) and EPA-treated (3 h and 6 h) post-larvae, while the remaining genes were either not significantly regulated (fat/cd36-lmp2, acly, kir6.x, srebp1c) or affected in a way inconsistent with FA-sensing mechanisms (fat/cd36-pg4l, fas, cpt1.1, cpt1.2, cpt1.3, sur, pparα and lxrα) (Soengas, 2014). However, we cannot disregard the fact that previous results demonstrating the presence of FA-sensing systems in sole juveniles were obtained after i.p. injection of free FAs, obviating the steps of intestinal absorption, which is likely to result in different timing and bioavailability of circulating FAs.

Interestingly, important changes were measured in the postprandial expression of cpt1 genes after treatment with different FAs but, just as observed in juvenile sole, the response was quite variable depending on the isoform that was measured. However, we would need more information, particularly on the functional characterization of these genes, in order to be able to interpret these results.

Therefore, it appears that oral administration of FAs to sole post-larvae in this study did not activate central FA-sensing systems, even when fish were administered with OA and ALA, which were previously demonstrated to induce the activation of these systems in juvenile sole (Conde-Sieira et al., 2015). The lack of response to any of the tested FAs in Senegalese sole post-larvae could suggest that the metabolic control of food intake involving FA sensing is not yet functional at this early stage of development. The finding that the anorectic potential was enhanced irrespective of the FA treatment also supports this possibility. A gradual development of FA-sensing capacities is likely as the operability of other mechanisms involved in the control of food intake, such as gastrointestinal remodelling, is also dependent on developmental stage (Kortner et al., 2011; Gomes et al., 2015). However, these results raise an interesting question: if FA-sensing systems are not responding to the intake of FAs, how do neuropeptides respond to FA administration? This differential behaviour may rely on the fact that neurons expressing neuropeptides not only integrate metabolic information on specific nutrients in circulation but also integrate information from other sources, including nervous inputs from the gastrointestinal tract and/or levels of peripheral hormones (leptin, insulin and ghrelin, among others) (Wang et al., 2002; Blouet and Schwartz, 2010). Hence, we might speculate that the capacity to integrate this kind of information in brain centres involved in metabolic regulation of food intake might develop earlier than that of FA-sensing mechanisms.

In summary, we have described for the first time in fish post-larvae the effect of feeding long-chain FAs of different nature (including PUFAs), such as OA, LA, ALA and EPA, on the metabolic regulation of food intake in Senegalese sole. All FAs induced changes in the expression of neuropeptides involved in the control of food intake. Although not all the assessed neuropeptides changed their mRNA abundance, the changes observed generally suggest an enhanced anorectic potential irrespective of the FA administered. However, none of the treatments activated central FA-sensing systems. These results are different from those observed in the hypothalamus of juveniles of the same species, where anorectic responses, in parallel with activation of FA-sensing systems, were observed in response to specific FAs injected i.p. (Conde-Sieira et al., 2015). This different behaviour between post-larvae and juveniles allows us to suggest that the metabolic control of food intake involving the function of FA-sensing systems and integration of this metabolic information through changes in neuropeptide expression might not be functional at this early stage of development.

The authors are grateful to Águeda J. Martín-Robles and José A. Muñoz-Cueto (Department of Biology, Faculty of Marine and Environmental Sciences, University of Cádiz, Spain) for supplying us with the qPCR primers for agrp2 and npy and to Almudena Martínez (IRTA) for assistance with fish dissection and RNA extraction.