Long-term fasting in the anadromous Arctic charr is associated with downregulation of metabolic enzyme activity and upregulation of leptin A1 and SOCS expression in the liver

The anadromous life strategy of Arctic charr [Salvelinus alpinus (Linnaeus 1758)] is characterized by substantial seasonal changes in food intake, growth and adiposity. Most of the annual growth and energy accumulation occurs during a short summer seawater residence, whereas overwintering in freshwater is characterized by anorexia and depletion of energy reserves (Dutil, 1986; Boivin and Power, 1990; Jørgensen et al., 1997; Jobling et al., 1998). Captive offspring of anadromous Arctic charr also exhibit pronounced seasonal changes in food intake and growth when held at constant temperature and given food in excess (Tveiten et al., 1998), showing that the seasonal cycle of food intake and growth is under physiological control. In captive offspring of anadromous charr, feeding commences in May (Tveiten et al., 1996; Bottengård and Jørgensen, 2008; Aarseth et al., 2010). In wild anadromous charr, appetite seems to return in early May (Aas-Hansen et al., 2005), whereas seaward migration takes place 3 to 4 weeks later (Aas-Hansen et al., 2005; Jensen et al., 2012). During the pre-migratory period the fish apparently find little food, and their condition factor continues to decrease until the time of downstream migration, when it may be as low as 0.70 (Nilssen et al., 1997; Aas-Hansen et al., 2005). Hence, the anadromous life strategy of Arctic charr incorporates a prolonged, voluntary anorexia with emaciation, indicating high tolerance to food deprivation. As such, the anadromous charr represents an interesting model for investigation of mechanisms associated with long-term regulation of appetite and energy homeostasis.

Since the first identification of a gene encoding a cDNA homologue to mammalian leptin (Lep) in pufferfish (Takifugu rubripes) (Kurokawa et al., 2005), Lep genes have been identified in many fish species, including the Arctic charr (Frøiland et al., 2010). In contrast to mammals, in which Lep is produced mainly in adipose tissue (Masuzaki et al., 1995), Lep is produced in numerous tissues in fish, including the liver (Wong et al., 2007; Kurokawa and Murashita, 2009; Rønnestad et al., 2010; Tinoco et al., 2012; Trombley et al., 2012). Hepatic LepA1 and LepA2 expression increased towards the end of a 7 week period of feed restriction in Atlantic salmon (Salmo salar) (Trombley et al., 2012) and during the latter part of a 75 day period of feed deprivation in Arctic charr (Frøiland et al., 2012). Increased liver LepA1 expression was observed without any change in plasma Lep concentrations (Frøiland et al., 2012; Kling et al., 2012), indicating a paracrine role for Lep in nutrient processing and regulation of endogenous energy reserves in the liver. A paracrine role for Lep in the fish liver is also supported by the presence of the Lep receptor (LepR) gene in this organ in several fish species (Liu et al., 2010; Rønnestad et al., 2010; Tinoco et al., 2012; Trombley et al., 2012).

We have studied the metabolic responses of the liver to long-term fasting in the Arctic charr, and have investigated whether Lep and the suppressor of cytokine stimulation (SOCS) may play a role in regulating these responses. Examination of SOCS gene expression was included in the study because SOCS is known to modulate the signalling of hormones regulating energy homeostasis such as growth hormone (GH) and Lep in mammals (Fujimoto and Naka, 2003).

The experiment was carried out at Tromsø Aquaculture Research Station (TARS), Tromsø, Norway (69°N), with hatchery-reared offspring of anadromous Arctic charr derived from a broodstock captured in Lake Vårfluesjøen, Svalbard (79°N), in 1990. The eggs hatched in winter 2008 and juveniles were held in freshwater at 6°C under continuous light until July 2009, and thereafter at ambient water temperature and natural light (transparent roof) conditions until the start of the experiment. On 8 and 9 March 2010, a total of 360 individually tagged (Floy FTF-69 fingerling tags; MGF, Seattle, WA, USA) fish with an average body mass of 146 g were taken from a stock tank, anaesthetised in benzocaine (60 p.p.m.) and randomly distributed among four 300 litre circular tanks. The tanks were subjected to simulated natural light conditions (69°N) and were supplied with freshwater at ca. 5°C (ranging between 4.7 and 6.3°C). The fish were provided with commercial dry-pellet feed (Nutra Parr; Skretting, Stavanger, Norway) in excess until 24 March, after which the fish in two tanks were feed deprived until the end of the experiment in June (FU fish). The fish in the other two tanks continued to be fed in excess throughout the experiment (FF fish). The experiment was terminated on 17 June, when the condition factor [(mass/length³)×100] of the fasted fish was approaching that (0.70) of wild, anadromous Arctic charr when they descend to the sea.

On 25 March, 6 May and 17 June, a total of five randomly selected fish were quickly netted from each tank and killed with an overdose of benzocaine (120 p.p.m.). Blood was collected from the caudal vein using 2 ml, lithium-heparinized (30 USP units) vacutainers and held on ice until plasma was separated by centrifugation (6000 g for 10 min). Plasma was stored at −80°C until analyses of cortisol and glucose concentrations were carried out. Thereafter, body mass (M_b; g) and fork length (FL; cm) of the fish were measured before each fish was dissected. Liver masses were noted and three small pieces were removed. One piece was put in a tube with RNAlater (Ambion, Life Technologies, Carlsbad, CA, USA) and stored at −20°C until analysis of LepA1, LepA2, IGF-I and SOCS1–3 gene expression, and the other two were wrapped in aluminium foil and frozen at −80°C for later analysis of glycogen and enzyme activities. The fish were then examined for feed in the stomach, sex and maturity status before the body was wrapped in aluminium foil and stored at −80°C for later analysis of proximate chemical composition. After the sampling in March and May, the remainder of the fish in each tank were anaesthetized in benzocaine for body mass and length measurement.

For the FF fish, one out of the 10 fish sampled in March had eaten and in May, four fish had eaten, whereas six had empty stomachs. In June, seven out of the 10 sampled fish had well-filled stomachs. The FF fish were intended to simulate the seasonal feeding cycle of wild, anadromous Arctic charr, so the fish that was recorded as eating in March and those that were anorexic in June were excluded from the analyses. The FF fish sampled in May were also excluded because of the low number of feeding fish and the heterogeneity of feeding behaviour within the group.

The experiment was approved by the Norwegian Committee on Ethics in Animal Experimentation (ID 2442).

Plasma cortisol concentration was determined by radioimmunoassay, using an established protocol (Schulz, 1985) validated for Arctic charr (Tveiten et al., 2010). The detection limit for the assay was 3.0 ng ml⁻¹. Plasma glucose concentration, given as nmol l⁻¹, was measured spectrophotometrically using a commercial enzymatic kit (HL 1611, Randox Laboratories, Crumlin, UK). Liver glycogen concentration was determined by glucose measurements, using the same kit as for plasma glucose measurements, before and after hydrolysis of glycogen with amyloglucosidase (Keppler and Decker, 1974) and expressed as μmol g⁻¹ protein. Liver enzyme activities were measured as described in Aas-Hansen et al. (Aas-Hansen et al., 2005) and expressed as μmol min⁻¹ g⁻¹ protein. Enzyme activities determined were alanine aminotransferase (AlaAT) (EC 2.6.1.2), aspartate aminotransferase (AspAT) (EC 2.6.1.1), glucose-6-phosphate dehydrogenase (G6PDH) (EC 1.1.1.49), 3-hydroxyacylcoenzyme A dehydrogenase (HOAD) (EC 1.1.1.35), lactate dehydrogenase (LDH) (EC 1.1.1.27), phosphoenolpyruvate carboxykinase (PEPCK) (EC 4.1.1.32) and pyruvate kinase (PK) (EC 2.7.1.40).

Whole fish were homogenized by passing them three times through a mincer (Sirman TC22F, Marsango, Italy). Duplicate samples of ca. 10 g homogenate were weighed (±0.1 g), dried for 24 h at 105°C and then re-weighed to calculate water content. Fat extraction was carried out as described by Johansen et al. (Johansen et al., 2002), using a Behrotest extraction system (Behr Labor-technik, Düsseldorf, Germany) and ether (Merck, Darmstadt, Germany) as the solvent. Following fat extraction, the extraction thimbles with tissue samples were dried and weighed. Fat content was calculated as mass loss during extraction, and fat concentration was calculated in relation to sample wet mass and expressed as a percentage of body mass. Finally, samples were ashed at 500°C for 12 h to estimate the percentage of ash in the sample. Protein was estimated as the difference between the dry sample mass and the combined masses of fat and ash, under the assumption of negligible carbohydrate content. The full data set for these results and calculations is given in supplementary material Table S1.

The in vitro study was performed in January 2013, with liver slices obtained from one Arctic charr (345 g) of the same strain as those used in Experiment 1. The fish were not eating at the time of sampling, but had a higher condition factor (1.06) than the fasted fish used in Experiment 1. The fish was killed by a sharp blow to the head and then decapitated. Thereafter the fish was dissected and the liver was perfused with ice-cold PBS via the hepatic portal vein until it turned white. The liver was then excised and placed in ice-cold L15 medium. A liver piece of ca. 0.7×0.7×0.5 cm was then taken and sliced (250 μm thickness) with a Leica VT1200 (Leica Microsystems, Wetzlar, Germany).

Nine liver slices were incubated individually under gentle shaking for 24 h in a 24-well plate at 5°C. Three wells contained 0.5 ml L15 medium with recombinant rainbow trout Lep dissolved in 5 μl phosphate buffered saline (PBS), resulting in a Lep concentration in the medium of 50 nmol l⁻¹ (PBS + Lep). The choice of concentration was based on results from treatment of grass carp (Ctenopharyngodon idellus) hepatocytes with recombinant grass carp leptin (Lu et al., 2012). The recombinant rainbow trout leptin with proven bioactivity was produced by the method described by Murashita et al. (Murashita et al., 2008). The purity of produced leptin was approximately 60%, determined by SDS-PAGE. Three wells contained 0.5 ml L15 medium with 0.5 μl PBS added (PBS) and the final three wells contained 0.5 ml L15 medium alone (untreated). After 24 h the slices were put in RNAlater and stored at −20°C until analysis of SOCS1, SOCS2, SOCS3, G6PDH, AlaAT and IGF-I gene expression. The choices of metabolic enzymes were made in order to investigate representatives of lipid metabolism (G6PDH) and gluconeogenesis (AspAT) that decreased and remained unchanged with fasting, respectively. Liver slice viability was estimated by LDH leakage, using a Cytotoxicity Detection Kit (LDH) from Roche (Roche Diagnostics, Mannerheim, Germany) and a Spetramax plus spectrophotometer (Molecular Devices, Sunnyvale, CA, USA) set at 490 nm.

Reverse transcription of total RNA was performed using the QuantiTect Reverse Transcription Kit (Qiagen, Hilden, Germany) with 1000 ng of RNA per 20 μl cDNA reaction according to the manufacturer's instructions. This kit includes an initial step of gDNA removal, and a reverse transcription test on a selection of samples showed that this was effective. Quantitative real-time PCR (qPCR) was performed, using the comparative C_T method (Applied Biosystems, Warrington, UK), on an ABI 7500 Fast Real-Time PCR System. For real-time estimations, cDNA corresponding to 25 ng RNA was amplified for 40 cycles in a 20 μl PCR mix (Fast SYBR Green Master Mix, Applied Biosystems, Foster City, CA, USA) containing a final concentration of 500 nmol l⁻¹ of each primer (PrimerDesign Ltd, The Mill Yard, Southampton, UK; Table 1). Cycling conditions were as follows: 95°C for 20 s, 40 cycles at 95°C for 3 s, and 60°C for 30 s including melt curve analysis. Duplicate PCR analyses were performed on each cDNA sample; the absence of gDNA was confirmed by performing a no-reverse-transcriptase control for every RNA sample, and absence of contamination was confirmed by including a no-template control in every run. The ΔΔC_T method (Livak and Schmittgen, 2001) was used to determine the relative amount of target gene, normalizing against the average expression of the two reference genes EF1α and β-Actin in Experiment 1 and against EF1α in Experiment 2. The primer sequences used for RTqPCR analyses are listed in Table 1.

All data from Experiment 1 are presented as means ± s.e.m. LepA1 and LepA2 expression are given as relative values in comparison with the mean liver LepA1 expression in FF fish in March. SOCS genes expression are given relative to the mean SOCS1 expression in FF fish in March, and IGF-I gene expression is given relative to the mean liver IGF-I expression in FF fish in March. Only values normalized against EF1α are given for the in vitro data. Condition factor (CF) and hepatosomatic index (HSI) were calculated as (M_b×FL⁻³)×100 and (liver mass×100)×M_b⁻¹, respectively. The masses of body fat and protein present in the fish were calculated by combining the body fat and protein data for the fish sampled in March, May and June with those relating to changes in M_b over time for the fish sampled in May (only FU fish) and June. All data were log-transformed before statistical analyses, with 1 added to the CF, liver HOAD, PEPCK, PK enzyme activity and liver Lep, IGF-I and SOCS expression data in order to achieve positive log numbers. A two-way ANOVA was used to reveal possible effects of time and treatment (replicates pooled) on M_b, CF, percentage body fat and protein, HSI, plasma cortisol and glucose concentrations, liver glycogen concentration and enzyme activities and liver LepA1, LepA2, SOCS1-3 and IGF-I gene expression. A post hoc (Tukey's HSD) test was used to reveal where significant differences occurred. Possible effects of Lep on the mRNA expression of liver SOCS1-3, AspAT, G6PDH and IGF-I genes in the in vitro study were tested using a non-parametric Kruskal–Wallis test. Results were considered significant at P≤0.05.

M_b did not differ between treatments in March (Fig. 1A), but in June FF fish had a much higher M_b than FU fish (P<0.001). A similar pattern was seen for CF, which did not differ between the two treatment groups in March, but was much higher (P<0.001) in FF than in FU fish in June (Fig. 1B). The FF fish sampled in June had increased markedly in M_b (P<0.01) and CF (P<0.01) from March to June (supplementary material Table S2). There was no change in M_b between March and May in FU fish sampled in March, while those sampled in June had only a minor, insignificant decline in M_bfrom March to June (from 155.4±9.3 to 138.1±7.5 g). The latter fish decreased in CF (P<0.001) from 0.88±0.01 to 0.75±0.01 (supplementary material Table S2).

There were no differences in the percentage of body protein between FF and FU fish sampled in March and June. The percentage of total body protein did not differ between FU fish sampled in March and May, but was slightly lower (P<0.05) in June than in March and May (Fig. 1C). There were no differences in the percentage of total body protein in FF fish sampled in March and June. The percentage of total body fat did not differ significantly between FF and FU fish in March (Fig. 1D). In FU fish, the percentage of total body fat did not differ significantly between March and May, but was much lower (P<0.01) in June than in May. In FF fish, the proportion of body fat was lower in March (4.5±0.5%) than in June (10.4±1.5%) and was higher than in FU fish (1.2±0.4%) in June (P<0.001). In mass terms, there was an almost total depletion of body fat in FU fish during the course of the experiment, and a greater loss of protein between May and June than between March and May (Fig. 2B). In FF fish, both total body fat and protein content were much higher in June than in March (Fig. 2A).

Plasma cortisol concentration did not differ between FF and FU fish in March and June (Fig. 1E), but was higher in FU fish in May than in March (P<0.05). The same trend was seen for plasma glucose concentrations (Fig. 1F), which also showed a transient increase in FU fish from March to May (P<0.05).

Liver LepA1 mRNA expression did not differ between FF and FU fish in March (Fig. 3A) and for FF fish there was no change in expression between March and June. In FU fish there was a higher expression in both March and June than in May (P<0.05 and P<0.01, respectively). In June, LepA1 expression was markedly lower in FF fish than in FU fish (P<0.05). The same general pattern was seen for LepA2 expression (Fig. 3B), which was higher in FU fish than in FF fish in June (P<0.01). Liver IGF-I mRNA expression did not differ between FF and FU fish in March. There was no difference in IGF-I expression for FU fish sampled in March and May (Fig. 3C), but there was a decrease in expression from May to June (P<0.05). There was no difference in IGF-I expression in FF fish sampled in March and June.

Liver SOCS gene expression is shown in Fig. 4. There were no differences between FF and FU fish in SOCS1, SOCS2 and SOCS3 in March. In FF fish, SOCS1 mRNA expression was higher in March than in June (P<0.05). There were no differences in SOCS1 expression in FU fish sampled at different times, resulting in a tendency toward a higher SOCS1 expression in FU fish than in FF fish in June (P=0.062). For SOCS2, expression was higher in FF fish sampled in March than in June (P<0.001), whereas the opposite was seen in FU fish (P<0.001). This resulted in a higher SOCS2 expression in FU fish than in FF fish in June (P<0.001). SOCS3 expression was higher in March than June in both FF (P<0.01) and FU fish (P<0.05), with a higher expression in FU than in FF fish in June (P<0.05).

HSI did not differ between FF and FU fish in March, after which there was a decrease in FU fish (P<0.05) and an increase in FF fish (P<0.001), resulting in a higher HSI in FF than in FU fish in June (P<0.001; Table 2). Liver glycogen concentration (Table 2) did not differ between FF and FU fish in March, but was higher in FF fish in June than in March (P<0.05) and lower in FU fish in June than in both March and May (P<0.001). This resulted in a marked difference in liver glycogen concentrations between FF (28.2±4.7 μmol g⁻¹ protein) and FU fish in June (1.8±0.5 μmol g⁻¹ protein; P<0.001).

There were no differences in any enzyme activities between FF and FU fish in March, nor between March and May in FU fish (Table 2). In FF fish, G6PDH (P<0.001), LDH (P<0.01), HOAD (P<0.001) and PEPCK (P<0.05) activities were higher in June than in March. In FU fish, G6PDH (P<0.01), LDH (P<0.01) and HOAD (P<0.01) activities were lower in June than in March. In June, there were higher Asp-AT, G6PDH, LDH, HOAD, PEPCK and PK activities in FF than in FU fish.

There was significantly higher SOCS1 (P<0.05) and SOCS3 (P<0.05) gene expression in Lep-treated liver slices (PBS + Lep) than in untreated and sham (PBS) treated slices, but no effects of Lep were seen for SOCS2, G6PDH, AspAT and IGF-I gene expression (Fig. 5).

The present study was designed to simulate the overwintering fast in wild, anadromous Arctic charr and the emaciation experienced by these fish during the latter part of their overwintering in freshwater. Because the CFs of the fish used in the present study (Fig. 1B) were similar to those of wild anadromous charr sampled in a previous study (Aas-Hansen et al., 2005), in which CF decreased from 0.84 in April to 0.77 in June, we consider this objective to have been achieved. In our study, only one fish sampled in March had food in its stomach, despite access to food, indicating that the fish were in a voluntary anorexic state. Some FF fish had consumed food when sampled on 6 May (see Materials and methods), which agrees with previous findings of a return of appetite in May in both captive anadromous Arctic charr (Tveiten et al., 1996; Bottengård and Jørgensen, 2008; Aarseth et al., 2010) and their wild conspecifics (Aas-Hansen et al., 2005).

In FU fish, fasting resulted in a decrease in M_b, CF, body fat and protein, particularly between the samplings in May and June, whereas there were increases in M_b, CF, body fat and protein in FF fish between March and June (Fig. 1A,B,D, Fig. 2). The low percentage (1.2±0.4%) and low absolute amount (Fig. 2) of body fat in FU fish in June indicates that stored fat was severely depleted at this time. In a previous study, plasma triglyceride concentration was also found to decrease in anadromous Arctic charr deprived of food during late spring/early summer, concomitant with an increase in plasma glucose concentration (Frøiland et al., 2012). Thus it seems that the energy needed during the latter part of overwintering, as well as during migration, must be taken from the body protein store in the anadromous Arctic charr. This conclusion is supported by the reduction in body protein observed in the FU fish from May to June (Fig. 2).

Prolonged fasting in fish is generally associated with reduced plasma glucose, HSI, liver glycogen and liver metabolic capacity (Navarro et al., 1992; Foster and Moon, 1991; Bastrop et al., 1992; Navarro and Gutiérrez, 1995). The lack of any large change in these parameters in FU fish between March and May (Fig. 1F, Table 2) indicates that physiological responses may differ between voluntarily fasting fish and fish subjected to enforced feed deprivation. The reduction in liver glycogen and liver enzyme activities from May to June in FU fish (Table 2) supports the notion that these fish were voluntarily anorexic until May, and that they subsequently experienced starvation stress in the absence of food. The reduced metabolic capacity of the liver, together with the depletion of glycogen in FU fish from May onwards, indicates that they responded by entering a state of metabolic depression, i.e. reduced metabolism under adverse environmental conditions (Guppy and Withers, 1999). Nevertheless, FU fish had plasma glucose concentrations similar to those of FF fish (Fig. 1F). This glucose was probably derived from protein, as evidenced from the decrease in the body protein store from May to June (Fig. 2B) and the maintained activity of enzymes involved in gluconeogenesis and amino acid catabolism (PEPCK and AlaAT; Table 2). In support of this, FU fish had plasma cortisol concentrations similar to those of FF fish at the end of the experiment in June (Fig. 1E), in accordance with the accepted gluconeogenetic action of this hormone (Mommsen et al., 1999). The maintenance of a high de novo glucose synthesis in the liver, despite depleted liver glycogen levels, reflects the importance of this substrate during a long-term fast (Polakof et al., 2012).

The changes in liver LepA1 and LepA2 expression during the course of the experiment resembled those previously reported for anadromous Arctic charr (Frøiland et al., 2012): a decrease in fed fish from March to July and higher LepA1 and LepA2 expression in FU than in FF fish at the end of the fasting period in June. An increase in liver Lep expression with fasting/feed restriction has also been reported in Atlantic salmon (Rønnestad et al., 2010; Trombley et al., 2012), whereas a reduced liver Lep expression with fasting has been reported in goldfish (Carassius auratus) (Tinoco et al., 2012) and striped bass (Morone saxatilis) (Won et al., 2012), and no change with prolonged fasting in common carp (Cyprinus carpio) (Huising et al., 2006). The reason for these differences is not known, but they could relate to the degree of emaciation of the fish used in the different experiments, and interspecific differences in metabolic responses to an imposed fast.

In the present study, the activity of enzymes involved in lipid metabolism (G6PDH and HOAD) and glycolysis (PK) appeared to be inversely related to liver LepA1 and LepA2 expression in FU fish. In addition, the highest activity of these enzymes was recorded concurrently with low liver LepA1 and LepA2 expression in FF fish in June (Fig. 3A,B, Table 2). No such interrelationships were seen for enzymes involved in amino acid metabolism and gluconeogenesis; LDH activity decreased in FU fish and increased in FF fish, but there were no changes in liver AlaAT and AspAT activities with time (Table 2). These results suggest that Lep may be involved in the suppression of lipolytic pathways, but in vitro Lep treatment of liver slices did not affect the gene expression of G6PDH (Fig. 4D). This indicates that Lep was not directly responsible for the changes in liver enzyme activity.

In mammals, SOCS1, 2 and 3 are involved in attenuating GH and Lep signalling by downregulating post-receptor JAK/STAT-induced gene transcription (Fujimoto and Naka, 2003). SOCS1–3 genes have been identified in several fish species (Jin et al., 2007; Xiao et al., 2010; Wang and Secombes, 2008; Studzinski et al., 2009) and their expression may be induced by cytokines in fish, as in mammals (Wang and Secombes, 2008). In the present study, SOCS gene expression was affected by the feeding regime (Fig. 4). The higher expression of the SOCS genes in FU fish than in FF fish in June coincided with a reduced expression of IGF-I in FU fish (Fig. 3C). This situation closely resembles the reduced IGF-I expression seen in fasted salmonid fish, as a result of downregulated hepatic GH signalling (Björnsson, 1997), and that in homozygotic GH transgenic zebrafish (Danio rerio), in which SOCS1 and SOCS3 seemed to be responsible for the downregulation of GH signalling in the liver (Studzinski et al., 2009). In common with the elevated plasma GH concentration in fasting salmonids (Björnsson, 1997), GH transgenic zebrafish overexpress GH in comparison with their hemizygotic counterparts (de Azevedo Figueiredo et al., 2007).

It is interesting to note that liver IGF-I expression in the FU fish used in our study only decreased during the latter part of the enforced fasting period (i.e. from May onwards), despite the fact that the fish most likely had been fasting voluntarily for months when sampled in May. This resembles the situation seen in seasonal mammals, which differ in their responses when undergoing seasonal winter weight loss and when they are subjected to feed deprivation during summer (Adam and Mercer, 2004). The reduction in liver IGF-I expression with fasting seen in the present study (Fig. 3C) was probably not a direct result of the increase in Lep expression because in vitro Lep treatment did not affect liver IGF-I expression (Fig. 5F), but was more likely related to the elevated SOCS1 and SOCS3 expression (Fig. 4A,C). This would comply with the results obtained with GH-transgenic zebrafish, which indicated that reduced GH signalling was caused by SOCS1 and SOCS3 (Studzinski et al., 2009).

The present study showed that Lep stimulated hepatic SOCS1 and SOCS3 gene expression in vitro (Fig. 5A,C). This result provides evidence for a link between the higher liver LepA1 and LepA2 expression in FU fish than in FF fish in June (Fig. 3A,B) and the corresponding difference in liver SOCS1 and SOCS3 gene expression in FU and FF fish (Fig. 4A,C). The higher liver SOCS2 expression in FU fish than in FF fish in June was probably not caused by Lep, which did not affect SOCS2 expression in the in vitro experiment (Fig. 5B). Another candidate could be cortisol, which was recently shown to stimulate SOCS2 expression in rainbow trout hepatocytes in vitro (Philip et al., 2012). However, we did not find differences in plasma cortisol concentrations between FU and FF fish in June (Table 2), even though the FU fish may have been suffering from starvation stress during the latter part of the fasting period.

In summary, the present study has revealed a metabolic suppression in anadromous Arctic charr subjected to fasting after they had entered their natural summer feeding period. Upregulated Lep and SOCS genes and downregulated IGF-I expression in the liver are suggested to be a part of the mechanisms that suppress lipolytic pathways and depress metabolism when fat stores are depleted. We have demonstrated that Lep stimulated hepatic SOCS1 and SOCS3 expression in vitro, and this stimulation may be responsible for downregulation of GH signalling and, consequently, reduced IGF-I expression in the liver. Further studies are needed to elucidate how Lep and SOCS interact to regulate intermediary metabolism in fish that display seasonal feeding cycles.

We would like to thank the staff at Tromsø Aquaculture Research Station for excellent care of the fish, Fredrikke Musæus for help with fat extraction and sampling, Tanja Hanebrekke for help with the cortisol analyses, and Birgit Süssenbach for help during the analyses of liver enzyme activity. We also thank Björn Thrandur Björnsson's Fish Endocrinology Laboratory, University of Gothenburg, for help in producing the trout recombinant leptin. Parts of the work reported here were submitted by M.M. and V.S. as partial requirements for the awarding of the degree in Master of Fisheries Sciences, University of Tromsø.