Social status affects lipid metabolism in rainbow trout, Oncorhynchus mykiss

Abstract

Juvenile rainbow trout (Oncorhynchus mykiss) confined in pairs form social hierarchies in which socially subordinate fish display characteristic traits, including reduced growth rates and altered glucose metabolism. These effects are, in part, mediated by chronically elevated cortisol levels and/or reduced feeding. To determine the effects of social status on lipid metabolism, trout were held in pairs for 4 days, following which organismal and liver-specific indexes of lipid metabolism were measured. At the organismal level, circulating triglycerides were elevated in dominant trout, whereas subordinate trout exhibited elevated concentrations of circulating free fatty acids (FFAs) and lowered plasma total cholesterol levels. At the molecular level, increased expression of lipogenic genes in dominant trout and cpt1a in subordinate trout was identified, suggesting a contribution of increased de novo lipogenesis to circulating triglycerides in dominant trout and reliance on circulating FFAs for β-oxidation in the liver of subordinates. Given the emerging importance of microRNAs (miRNA) in the regulation of hepatic lipid metabolism, candidate miRNAs were profiled, revealing increased expression of the lipogenic miRNA-33 in dominant fish. Because the Akt-TOR-S6-signaling pathway is an important upstream regulator of hepatic lipid metabolism, its signaling activity was quantified. However, the only difference detected among groups was a strong increase in S6 phosphorylation in subordinate trout. In general, the changes observed in lipid metabolism of subordinates were not mimicked by either cortisol treatment or fasting alone, indicating the existence of specific, emergent effects of subordinate social status itself on this fuel.

INTRODUCTION

Among juvenile salmonid fish, linear dominance hierarchies form as a result of competition for feeding territories (1, 23). The dominant fish within these hierarchies monopolize resources, such as food or shelter, displaying high levels of aggression toward their subordinate counterparts (1, 60). These differences in behavior are accompanied by a range of physiological effects, including changes in energy metabolism (21, 33, 35). Previous studies largely have focused on changes in carbohydrate metabolism, revealing an increased potential for hepatic glucose liberation in subordinates. For example, subordinate trout display increased mobilization of stored glycogen compared with dominant trout, as evidenced by lower hepatic glycogen concentrations and higher glycogen phosphorylase activity (35). As well, subordinate trout display enhanced gluconeogenic potential (35), which is further supported by increased hepatic phosphoenol pyruvate carboxykinase (PEPCK) activity and decreased pyruvate kinase activity (21). These changes are largely dependent on cortisol (21, 62, 83), with chronically elevated cortisol levels in subordinate trout leading to increased circulating glucose concentrations (14). However, while circulating glucose represents an important fuel source for specific rainbow trout tissues, such as the brain, gill, and erythrocytes (70), glucose utilization for global energy metabolism is limited in rainbow trout (63).

In contrast, it is well established that lipid metabolism plays a key role in energy homeostasis in rainbow trout. Indeed, basal lipolytic rates in this species are so high that only 13% of liberated free fatty acids (FFAs) are sufficient to support total energy expenditure with 87% being reesterified (52). Another index supporting the important contribution of lipid metabolism to total energy expenditure in rainbow trout is the resting rate of triglyceride (TG) turnover, which is sufficiently high to fuel red muscle during endurance exercise (53). Rainbow trout store excess energy in the form of TGs, which, once synthesized in the liver, are stored in white adipose tissue (WAT) (2, 81). Conversely, in times of energy demand, FFAs are mobilized from WAT and are shuttled by lipoproteins to fuel metabolically active target tissues (2, 81). In line with their high reported lipid turnover rates, high basal and exercise-induced lipoprotein lipase activity in rainbow trout allows lipids to be liberated from shuttling proteins for use in muscle (52, 53). As important contributors to overall energy expenditure at rest, as well as during locomotion, it is not surprising that lipids play a key role in energetically demanding life-history events, such as migration and sexual maturation (10, 11, 31).

Given the importance of lipids in rainbow trout energy metabolism, there is a need to investigate the consequences of social interactions on lipid metabolism. To determine these effects, size-matched juvenile rainbow trout were confined in pairs for 4 days, after which organismal and liver-specific indexes of lipid metabolism of dominant and subordinate fish were compared with those of sham-treated fish that were handled in the same fashion as paired fish but housed individually. Both chronic elevation of cortisol and reduced feeding are key traits of subordinate status in rainbow trout that could mediate changes in lipid metabolism. Therefore, the contribution of each factor individually was examined in a second experiment, where fish were either cortisol-treated or fasted for 4 days.

We tested the hypothesis that social status affects organismal and hepatic lipid metabolism, with the specific prediction that dominant trout would upregulate anabolic pathways, while subordinate trout would upregulate catabolic pathways. A secondary hypothesis was that changes in lipid metabolism were mediated by endocrine and physiological correlates of social status, specifically, cortisol elevation and fasting. While most previous studies have focused on consequences of long-term fasting on lipid metabolism parameters in rainbow trout (71), previous studies revealing a decrease in serum TG after a 3-day fast (44), a decrease in total liver lipid content (26) after a 2-day fast, and decreases in hepatic mRNA abundance of lipogenic genes and increases in hepatic mRNA abundance of genes involved in fatty acid β-oxidation (57) after a 2-day fast led us to formulate the prediction that fasting would significantly inhibit lipid anabolism and stimulate lipid catabolism in our experimental time period.

Both hypotheses and their specific predictions were addressed at the organismal level and, given its importance in regulating organismal lipid metabolism, also at the level of the liver. Specifically, to address effects of social status and associated endocrine and physiological factors on organismal lipid metabolism, plasma concentrations of three classes of lipids, TG, FFAs, and total cholesterol (TC), were determined. To gain insight into hepatic metabolic pathways, which are key mediators in organismal lipid homeostasis, transcript abundance of components involved in lipogenesis [sterol regulatory element-binding protein 1c (srebp1c), fatty acid synthase (fasn), and ATP citrate lyase (acly)], β-oxidation [(carnitine palmitoyltransferase 1A (cpt1a)], and cholesterol biosynthesis and degradation pathways [(hydroxymethylglutaryl-CoA synthase (hmgcs), liver X receptor (lxr), UDP glucuronosyltransferase family 1 member A3 (ugt1a3), and sterol regulatory element-binding protein 2 (srebp2)] were measured. Given the emerging importance of microRNAs (miRNA) in the regulation of hepatic and organismal lipid metabolism in rainbow trout (55–59), the potential involvement of miRNAs was first investigated by measuring hepatic mRNA abundance of drosha. Drosha is a necessary factor in the miRNA biogenesis pathway (48) that exists as a single paralogue in rainbow trout (9). We also measured the hepatic abundance of two specific miRNAs, miRNA-122 and miRNA-33, owing to their roles in teleost lipid metabolism (55–58, 85). Liver-specific miRNA-122 has been shown to stimulate serum triglyceride and cholesterol concentrations in rainbow trout (56), in line with previously reported roles in mammals (24, 25). While miRNA-122 remains the only miRNA to have been studied functionally in rainbow trout (56), we also profiled the hepatic abundance of miRNA-33, first because of its evolutionarily conserved genomic location within the introns of srebp2 suggesting a supporting function in its host gene’s role as a transcription factor mediating cholesterol biosynthesis (65, 72), and second, because a functional role for miRNA-33 in fatty acid biosynthesis was suggested recently in a marine fish species, Siganus canaliculatus (85).

Although the phospholipase C/protein kinase C (PLC/PKC), mitogen-activated protein kinase kinase/extracellular signal-regulated kinase (MEK/ERK) and the Janus tyrosine kinase/signal transducer and activator of transcription (JAK-STAT) pathways recently have been shown to correlate with (6), and indeed functionally regulate lipolytic pathways in the liver (7, 8), the protein kinase B/target of rapamycin/ribsosomal protein S6 (Akt/Tor/S6) pathway remains the only hepatic cell pathway in rainbow trout that has been directly linked to the acute regulation of mRNA abundance and enzyme activity of hepatic fasn/Fasn and cpt1a/Cpt1 in studies employing rapamycin, a Tor inhibitor (15–17, 50). In light of the demonstrated functional importance of hepatic Akt-Tor-S6 cell signaling at the transcriptional level of lipogenic (srebp1c, fasn) and β-oxidation (cpt1a) pathways, also probed in this study, we quantified Akt/Tor/S6 activity to determine its possible contribution to social status-dependent differences in lipid metabolism. An integrative overview of the relevant indexes of hepatic and organismal lipid metabolism profiled in this study is presented in Fig. 1.

Fig. 1.

Schematic of hepatic lipid metabolic pathways investigated in this study. Protein kinase B/target of rapamycin/ribsosomal protein S6 (Akt/Tor/S6) pathway components are highlighted in green, key metabolic enzymes in blue and microRNAs are in purple. Circulating metabolites are highlighted in red, while metabolites present within the liver are highlighted in yellow. In all cases, bold letters indicate variables measured in this study. Established functional relationships between measured parameters are based on previous literature (15, 17, 57), and question mark symbols indicate predicted interactions. FFA, free fatty acid; srebp1c, sterol regulatory element-binding protein 1c; fasn, fatty acid synthase; acly, ATP citrate lyase; cpt1a, carnitine palmitoyltransferase 1A; hmgcs, hydroxymethylglutaryl-CoA synthase; lxr, liver X receptor; ugt1a3, UDP glucuronosyltransferase family 1 member A3; srebp2, sterol regulatory element-binding protein 2.

MATERIALS AND METHODS

Experimental Animals

Female rainbow trout, Oncorhynchus mykiss, purchased from Linwood Acres Trout Farm (Campbellcroft, ON, Canada) were held at the University of Ottawa in 1,275-liter fiberglass tanks. Tanks were supplied with flowing, aerated, dechloraminated city of Ottawa tap water at a temperature of 13°C. Fish were fed a ration of 0.5% body mass daily by scattering commercial trout pellets on the water’s surface. A 12:12-h light-dark photoperiod was maintained. Trout were acclimated to these holding conditions, which acted to minimize hierarchy formation (e.g., use of scatter feeding, homogenous tanks with a mild current), for at least 2 wk before experimentation. All experimental protocols complied with the guidelines of the Canadian Council on Animal Care for the use of animals in research and teaching and were approved by the University of Ottawa’s Animal Care Committee (Protocol No. BL-2118).

Experimental Protocols

Two experimental series were conducted. The first experiment used juvenile trout (mass = 97.2 ± 2.4 g; fork length = 20.3 ± 0.2 cm; means ± SE; n = 37) that had been confined in pairs or handled identically but held individually (sham-treated), for 4 days (n = 12 pairs, 13 shams). The second experiment used groups of 10–12 trout (mass = 187.0 ± 16.4 g; fork length = 25.8 ± 0.6 cm; means ± SE; n = 17). These trout were cortisol treated, fasted, or controls and were held under these conditions for 4 days. These treatments were included to assess the effects of subordinate-associated traits, specifically, elevated cortisol and reduced feeding.

Experiment 1: social interactions and behavioral phenotyping.

Fish were lightly anesthetized to the point of losing equilibrium in a solution of benzocaine (0.05 g/l ethyl-p-aminobenzoate; Sigma-Aldrich, Oakville, ON, Canada), initial mass and fork length were measured, and fin damage was scored. When existing physical differences did not allow for identification of individual fish, a pectoral fin clip was used for identification. Fish were paired based on fork length, with differences not exceeding 5% (fork length difference averaged 33 ± 9 mm or 1.6% of fork length; n = 12 pairs). After initial assessment, members of a pair were placed into a 40-liter flow-through Plexiglas observation tank, separated from one another by an opaque, perforated divider. Tanks were supplied with flowing, aerated 13°C water.

The following morning, the divider was removed and fish were allowed to interact for 4 days. A shelter (t-shaped PVC tube, 11 × 13 cm long, 6-cm diameter) was added at the end of the first day of interaction. Behavioral observations were carried out twice per day, at 900–1100 and 1500–1700, for 5 min per observation period. Sham-treated trout underwent the same handling and treatment as paired fish but were held individually. Fish were offered 0.5% fish mass per tank daily following the final observation period (except on the initial day of interaction), and the mass of food consumed as well as the fish that consumed the food was noted.

Social status was assessed by assigning points to each fish for position within the tank, food acquisition, aggressive acts, and fin damage acquired during the interaction period (as previously described by Refs. 20, 21, 78). This scoring system awards more points for more dominant behaviors; specifically, patrolling the water column in the tank, aggressive behavior, first to feed, and absence of fin damage. A principle components analysis (SigmaPlot v13.0; Systat software, San Jose, CA) was used to calculate behavior scores for each fish based on the mean scores across observation periods of each parameter. Within a pair, the fish with the higher score was assigned dominant status, while the fish with the lower score was subordinate. Behavior scores as well as parameters associated with social status (plasma cortisol levels, specific growth rate, and food intake) are listed in Table 1.

Table 1.

Characterization of dominant, subordinate, and sham-treated rainbow trout (Oncorhynchus mykiss)

	Dominant (n = 12)	Subordinate (n = 12)	Sham (n = 13)	P Value
Behavior score	1.6 ± 0.1	−1.6 ± 0.2	—	—
Plasma cortisol concentration, ng/ml	4.0 ± 0.9A	111.4 ± 25.2B	2.5 ± 0.5A	<0.001
Food intake, g/day	1.12 ± 0.13A	0.04 ± 0.04B	0.62 ± 0.15A	<0.001
SGR, %/day	0.48 ± 0.14A	−0.50 ± 0.24B	−0.15 ± 0.16A,B	0.003

Values are means ± SE. Data were analyzed by ANOVA and ANOVA on ranks for food intake, and treatment groups that share a letter are not significantly different from one another. Specific growth rate (SGR) was calculated as (ln final mass – ln initial mass)/interaction period (day) × 100%.

After the interaction period, both fish in a pair were rapidly euthanized via terminal anesthesia (0.5 g/l ethyl-p-aminobenzoate). Mass and fork length were measured, fin damage was scored, and blood samples were collected via caudal venipuncture into heparin-coated syringes (2,500 IU/ml heparin sodium salt; Sigma-Aldrich). Blood samples were centrifuged (10,000 g for 2 min), and plasma was extracted. Plasma samples were flash frozen in liquid nitrogen before being stored at −80°C. Liver tissue was collected, freeze clamped, and stored at −80°C for later analysis.

Experiment 2: control, cortisol-treated, and fasted groups.

Trout were randomly allocated into one of three 115-liter holding tanks in groups of 10–12 fish per tank. This group size (in combination with the use of scatter-feeding, where appropriate, and homogenous tanks with a mild current) acted to minimize unwanted hierarchy formation. One set of fish was not fed for 4 days (fasted treatment group). Fish in a second group (cortisol-treated) were lightly anesthetized and given an intraperitoneal implant of cocoa butter (5 ml/kg body mass) containing hydrocortisone 21-hemisuccinate (22 mg/ml; Sigma-Aldrich). In previous studies, this dosage (110 mg/kg body mass) raised circulating cortisol levels to those of subordinate fish (33, 45). Lastly, an untreated group of fish was used to control for effects of handling (control treatment group). Sham-implanted fish (fish treated only with cocoa butter) were not used because the injection itself has unpredictable effects on cortisol levels (20). A random subset of five to six fish from each set was sampled as described above (see Experiment 1: social interactions and behavioral phenotyping).

Plasma Cortisol and Metabolite Concentrations

Plasma cortisol levels were measured using a commercially available radioimmunoassay (MP Biomedicals, Santa Ana, CA) that had previously been validated for trout plasma (30). The kit has a detection limit of 0.17 μg/dl. Intra-assay variation was 9.6% and interassay variation was 11.3% (% coefficient variation). All lipid metabolites were assayed in duplicate using commercially available kits according to the manufacturer’s protocols (Cayman Chemical, Ann Arbor, MI). Plasma TG levels were measured using a colorimetric assay with absorbance measurements at 538 nm. The detection limit of this assay was 3.125 mg/dl. Plasma FFA levels were measured by means of a fluorometric assay using an excitation wavelength of 535 nm and an emission wavelength of 590 nm. The detection limit of this assay was 7.04 mg/dl. Plasma TC levels were measured using a fluorometric assay with an excitation wavelength of 535 nm and an emission wavelength of 590 nm. The detection limit for this assay was 77.34 μg/dl. Colorimetric and fluorometric assays were carried out using SpectraMax Plus 384 or SpectraMAX GeminiXS plate readers, respectively (Molecular Devices, Sunnyvale, CA).

Analysis of mRNA Transcript Abundance

Total RNA was extracted from 20 to 100 mg of liver using TRIzol reagent (Invitrogen, Burlington, ON, Canada) following the manufacturer’s protocol. Tissues were homogenized by forcing the solution of TRIzol and tissue through 18- and 23-G needles using a syringe until the solution passed easily through the needle. Extracted RNA was quantified using a NanoDrop 2000c UV-Vis Spectrophotometer (Thermo-Fisher Scientific, Ottawa, ON, Canada). Next, cDNA was generated using a QuantiTech Reverse Transcription Kit (Qiagen, Toronto, ON, Canada) following the manufacturer’s protocol.

Two-step semiquantitative real-time RT-PCR assays were performed on a BioRad CFX96 instrument (Bio-Rad, Mississauga, ON, Canada) to quantify fold changes in relative hepatic mRNA abundances of key transcripts involved in lipogenesis (srebp1, fasn, and acly), fatty acid β-oxidation (cpt1a), TC biosynthesis (srebp2 and hmgcs), TC degradation (lxr, ugt1a3), and miRNA biogenesis (drosha), as well as two reference genes (ef1a and 18s). Briefly, a standard curve consisting of serial dilutions of pooled cDNA, a negative no-RT control consisting of cDNA generated in a reaction that did not include reverse transcriptase, and individual samples were run in duplicate for each experiment. For each individual reaction, the total volume was 20 µl, which consisted of 4 µl of diluted cDNA template, 0.5 µl of 10 nM specific forward and 0.5 µl of 10 nM specific reverse primer (Table 2), 10 µl of SsoAdvanced Universal Inhibitor-Tolerant SYBR Green Supermix (Bio-Rad), and 5 µl of H₂O. For each assay, cycling parameters were a 5-min activation step at 95°C, followed by 40 cycles consisting of a 20-s denaturation step at 95°C and a combined 30-s annealing and extension step at primer specific temperatures (Table 2). After each run, melting curves were produced by gradually increasing temperature and the final curves were monitored for single peaks to confirm the specificity of the reaction and the absence of primer dimers. In cases where primers were newly designed (drosha), pooled samples were sent for sequencing (Ottawa Hospital Research Institute, Ottawa, ON, Canada), followed by BLAST search (National Center for Biotechnology Information), to confirm amplicon specificity. The acceptable range for amplification efficiency calculated from serially diluted standard curves was 90–110%, with R² values >0.95. Assays were subsequently normalized using the NORMA-Gene approach as described by Heckman et al. (39). Finally, mRNA fold changes were calculated relative to the sham group for the social hierarchy experiment (experiment 1), and relative to the control group for the experiment investigating fasting and cortisol treatment (experiment 2).

Table 2.

Primer sequences and annealing temperatures used for mRNA and miRNA quantification by real-time RT-PCR

Target	Forward Primer Sequence (5′–3′)	Reverse Primer Sequences (3′–5′)	Annealing Temperature, °C	Reference No.
srebp1c	GACAAGGTGGTCCCAGTTGCT	CACACGTTAGTCCGCATCAC	58	(68a)
fasn	TGATCTGAAGGCCCGTGTCA	GGGTGACGTTGCGTGGTAT	58	(58)
acly	GCTTTTGCCACGGTGGTCTC	GCTTCCGCTACGCCAATGTC	60	(76a)
cpt1a	TCGATTTTCAAGGGTCTTCG	CACAACGATCAGCAAAGTGG	57	(76a)
srebp2	TAGGCCCCAAAGGGATAAG	TCAGACACGACGAGCACAA	58	(58)
hmgcs	AGTGGCAAAGAGAGGGTGTG	TTCTGGTTGGAGACGAGGAG	60	(58)
lxr	TGCAGCAGCCGTATGTGGA	GCGGCGGGAGCTTCTTGTC	60	(14a)
ugt1a3	CCACCAGCAAGACAGTCTCA	CAACAGCACAGTGGCTGACT	61	(58)
ef1a	CATTGACAAGAGAACCATTGA	CCTTCAGCTTGTCCAGCAC	56	(1a)
18S	GGCGGCGTTATTCCCATGA	TGCCCTTCCGTCAATTCCTTTA	60	(45)
drosha	GAGGAGTCGGTGAAGGAATG	CATGTGGGAGAAGAGGGAGA	60	Newly designed
miR-122	TGGAGTGTGACAATGGTGTTTG	Universal reverse primer	60	(58)
miR-33	GTGCATTGTAGTTGCATTGCAT	Universal reverse primer	58	(58)
Sno-U23	GCCCATGTCTGCTGTGAAACAAT	Universal reverse primer	60	(76)

srebp1c, Sterol regulatory element-binding protein 1c; fasn, fatty acid synthase; acly, ATP citrate lyase; cpt1a, carnitine palmitoyltransferase 1A; hmgcs, hydroxymethylglutaryl-CoA synthase; lxr, liver X receptor; ugt1a3, UDP glucuronosyltransferase family 1 member A3; srebp2, sterol regulatory element-binding protein 2.

Analysis of Mature miRNA Abundance

With the use of the extracted total RNA, small RNAs <17 nucleotides in length were removed using a miRNeasy kit (Qiagen) according to the manufacturer’s instructions. After quantification by NanoDrop (Thermo-Fisher Scientific), 144 ng of the purified RNAs were used to synthesize cDNA with HiFlex buffer using the miScript II RT kit (Qiagen) according to the manufacturer’s instructions. Specific miRNAs were then quantified using the miRScript SYBR Green PCR kit (Qiagen) with miRNA-specific forward primers and a universal reverse primer (Table 2). Reactions were run in duplicate on a CFX96 instrument (Bio-Rad) with a total volume of 25 µl containing 2.5 µl cDNA, 2.5 µl of 10 nM miRNA specific primer, 2.5 µl miScript Universal Primer, 12.5 µl 2× QuantiTect SYBR Green PCR Master Mix (Qiagen), and 5 µl H₂O, according to the manufacturer’s instructions. For each assay, cycling parameters were an initial 15 min 95°C activation step, followed by 40 cycles of 15-s incubation at 94°C, 30 s at 60°C, and 30 s at 70°C. After each run, melting curves were produced by a gradual increase in temperature and the final curves were monitored for single peaks to confirm the specificity of the reaction and the absence of primer dimers. Standard curves and no-RT controls were used to assess efficiency and specificity of amplifications as previously described. Because the abundance of fewer than five miRNAs was analyzed, the NORMA-Gene method was not used. Rather, the ∆∆CT (67) method for normalization was adopted using SnoU23 as a reference gene as previously described for rainbow trout (76). The miRNA fold changes were then calculated relative to sham and control groups, for experiments 1 and 2, respectively.

Analysis of Cell Signaling Pathways

Frozen liver samples (~100 mg) were homogenized on ice with a sonicator model 100 (Thermo-Fisher Scientific). Tissues were homogenized in a buffer (pH 7.4) containing 150 mmol/l NaCl, 10 mmol/l Tris, 1 mmol/l EGTA, 1 mmol/l EDTA, 100 mmol/l sodium fluoride, 4 mmol/l sodium pyrophosphate, 2 mmol/l sodium orthovanadate, 1% (vol/vol) Triton X-100, 0.5% (vol/vol) NP40-IGEPAL (all Sigma-Aldrich), and Pierce protease inhibitor (Thermo-Fisher Scientific). Homogenates were centrifuged at 15,000 g for 30 min at 4°C, and supernatants were stored at −80°C. Protein concentrations were quantified using a BCA (Sigma-Aldrich) assay with BSA standards and a SpectraMax Plus 384 plate reader (Molecular Devices) at a wavelength of 562 nm. Aliquots of 50 μg protein per sample were subjected to SDS-PAGE and Western blotting, using the appropriate primary antibodies (Table 3). All primary antibodies used for analysis of the insulin signaling pathway were obtained from Cell Technologies (Ozyme, Saint Quentin Yvelines, France) and previously have been shown to cross react successfully with rainbow trout proteins of interest and to be regulated in physiological responses to nutritional and endocrine factors (56, 57, 68, 77). Also, activation of downstream components of this pathway has been shown to be inhibited with concurrent utilization of the Tor-specific inhibitor rapamycin (15–17, 50). All primary antibodies used were raised in rabbit, and after final washing, membranes were incubated with an IRDye infrared secondary anti-rabbit antibody raised in goat (LI-COR Biotechnology, Lincoln, NE). Bands were visualized and quantified by infrared fluorescence using the Odyssey Imaging System (LI-COR Biotechnology). Active phosphorylated protein levels were expressed relative to total protein level.

Table 3.

Primary antibody sources and dilutions used for cell signaling quantification by Western blot analysis

Target	Reference ID No.	Supplier	Dilution Factor	Band Size, kDa
Akt-P	9271	Cell Signaling/New England Biolabs	1:10,000	60
Akt	9272	Cell Signaling/New England Biolabs	1:10,000	60
S6-P	2211	Cell Signaling/New England Biolabs	1:10,000	32
S6	2217	Cell Signaling/New England Biolabs	1:10,000	32

Statistical Analysis

All data were tested for normality and homoscedasticity using the Shapiro-Wilk test and Bartlett’s test, respectively. If these assumptions were met, data were tested for single outliers using Grubb’s test and subsequently analyzed using a one-way ANOVA, followed by Tukey’s post hoc test for multiple comparisons. When data did not meet these assumptions, data were either transformed to meet the assumptions or were analyzed using the nonparametric Kruskal-Wallis test followed by Bonferroni-adjusted Mann-Whitney U-tests. To facilitate statistical analyses that included paired and sham fish, the members of a pair were considered to be independent of one another. With the use of this conservative approach, it may be more difficult to reject the null hypothesis of no difference stemming from social status than when the members of a pair are not considered to be independent of one another (13). All statistical analysis and graphing were carried out using Prism, Version 7 (Graphpad software, La Jolla, CA).

RESULTS

Behavioral and Endocrine Correlates of Social Status

The social pairing experiment generated dominant and subordinate phenotypes (Table 1). Plasma cortisol concentration (Table 1, df = 2; F = 75.90, P < 0.01) was significantly higher in subordinate trout compared with both dominant and sham-treated trout (P < 0.01). The elevation of circulating cortisol levels in subordinate trout was accompanied by significantly lower food intake (Table 1, df = 2; F = 7.02, P < 0.01) compared with dominant fish (P < 0.01). Food intake was marginally elevated in dominant fish compared with sham fish (P = 0.052). Specific growth rates (Table 1, df = 2; H = 23.17, P < 0.01) in subordinate fish were significantly lower compared with both dominant and sham fish (P < 0.01 and P < 0.05, respectively). In experiment 2, plasma cortisol concentration (Fig. 2, df = 2; H = 11.93, P < 0.01) was significantly elevated in cortisol-treated fish compared with control and fasted fish (P < 0.01).

Fig. 2.

Plasma cortisol concentrations of control, cortisol-treated, and fasted rainbow trout (Oncorhynchus mykiss). Values are presented as means ± SE with n = 5–7 for all groups; values for individuals included in the means are indicated by the symbols. Bars that share a letter are not significantly different from one another (see text for details).

Circulating Lipid Metabolites

Plasma TG concentrations differed significantly with social status (Fig. 3A, df = 2; F = 5.24, P < 0.05), with plasma TG concentrations in dominant fish being elevated compared with sham fish (P < 0.05) and marginally higher than those of subordinate fish (P = 0.0501). Circulating plasma TG concentrations did not differ among control, cortisol-treated, and fasted fish (Fig. 3B, df = 2; F = 0.58, P > 0.05). Plasma FFA concentrations were significantly influenced by social status (Fig. 3C, df = 2; F = 17.65, P < 0.01), with elevated concentrations in subordinate fish compared with dominant and sham fish (P < 0.05). No differences in plasma FFAs were observed among control, cortisol-treated, and fasted fish (Fig. 3D, df = 2; F = 2.37, P > 0.05). Plasma TC concentration was significantly (Fig. 3E, df = 2; F = 2.99, P < 0.05) lower in subordinate compared with dominant fish (P < 0.05). Circulating TC concentrations were not significantly different among control, cortisol-treated, or fasted trout (Fig. 3F, df = 2; F = 0.60, P > 0.05).

Fig. 3.

Plasma concentrations of triglycerides (TG; A and B), free fatty acids (FFAs; C and D), and total cholesterol (TC; E and F) in sham, dominant, and subordinate (A, C, and E), and control, cortisol-treated, and fasted (B, D, and F) rainbow trout (Oncorhynchus mykiss). Values are presented as means ± SE with n = 5–7 for all groups; values for individuals included in the means are indicated by the symbols. Bars that share a letter are not significantly different from one another (see text for details).

Hepatic mRNA Abundance of Genes Involved in Lipid Metabolism

mRNA abundance of lipogenic genes.

The mRNA abundance of srebp1c was significantly (Fig. 4A, df = 2; F = 11.20, P < 0.01) higher in dominant fish compared with subordinates and shams (P < 0.01). In experiment 2, srebp1c mRNA abundance was significantly elevated (Fig. 4B, df = 2; F = 6.24, P < 0.05) in cortisol-treated fish compared with control and fasted fish (P < 0.05). The mRNA abundance of fasn was affected by social status (Fig. 4C, df = 2; F = 30.02, P < 0.01), with elevated levels in dominant fish compared with subordinates and shams (P < 0.01) but did not differ significantly among control, cortisol-treated and fasted fish (Fig. 4D, df = 2; F = 2.34, P > 0.05). The mRNA abundance of acly was significantly (Fig. 4E, df = 2, F = 16.74, P < 0.01) increased in both dominant and subordinate trout compared with shams (P < 0.01). In addition, a higher abundance of acly mRNA was detected in cortisol-treated fish compared with fasted fish (Fig. 4F, df = 2, F = 4.88, P < 0.05).

Fig. 4.

Abundance of mRNA markers of hepatic lipogenesis pathways. Transcript levels of sterol regulatory element-binding protein 2 (srebp1c; A and B), fatty acid synthase (fasn; C and D), and ATP citrate lyase (acly; E and F) in sham, dominant, and subordinate (A, C, and E) and control, cortisol-treated, and fasted (B, D, and F) rainbow trout (Oncorhynchus mykiss). Values are presented as means ± SE with n = 4–7 for all groups; values for individuals included in the means are indicated by the symbols. Bars that share a letter are not significantly different from one another (see text for details).

mRNA abundance of β-oxidation genes.

The mRNA abundance of cpt1a differed significantly with social status (Fig. 5A, df = 2; F = 10.25, P < 0.01), with increased mRNA abundance in subordinates compared with both sham (P < 0.05) and dominant fish (P < 0.01). Conversely, no significant differences in cpt1a mRNA abundance were observed among control, cortisol-treated, and fasted fish (Fig. 5B, df = 2; F = 1.48, P > 0.05).

Fig. 5.

mRNA abundance of the β-oxidation marker carnitine palmitoyltransferase 1A (cpt1a) in sham, dominant, and subordinate (A) and control, cortisol-treated, and fasted (B) rainbow trout (Oncorhynchus mykiss). Values are presented as means ± SE with n = 3–7 for all groups; values for individuals included in the means are indicated by the symbols. Bars that share a letter are not significantly different from one another (see text for details).

mRNA abundance of TC biosynthesis genes.

Increased mRNA abundance of srebp2 (Fig. 6A, df = 2; H = 6.38, P < 0.05) was detected in subordinate fish compared with sham fish (P < 0.05), while no significant differences among control, cortisol-treated, and fasted trout were observed (Fig. 6B, df = 2, F = 2.02, P < 0.05). The mRNA abundance of hmgcs differed significantly with social status (Fig. 6C, df = 2; F = 16.29, P < 0.01), with dyad-housed trout (dominant and subordinate) exhibiting increased mRNA abundance compared with sham fish (P < 0.01). In addition, hmgcs mRNA abundance was significantly (Fig. 6D, df = 2; F = 4.441, P < 0.05) increased in cortisol-treated trout relative to control trout (P < 0.05).

Fig. 6.

Abundance of mRNA markers of total cholesterol (TC) biosynthesis pathways. Transcript levels of sterol regulatory element-binding protein 2 (srebp2; A and B) and hydroxymethylglutaryl-CoA synthase (hmgcs; C and D) in sham, dominant, and subordinate (A and C) and control, cortisol-treated, and fasted (B and D) rainbow trout (Oncorhynchus mykiss). Values are presented as means ± SE with n = 4–7 for all groups; values for individuals included in the means are indicated by the symbols. Bars that share a letter are not significantly different from one another (see text for details).

mRNA abundance of TC degradation genes.

The mRNA abundance of lxr was significantly (Fig. 7A, df = 2; F = 19.35, P < 0.01) increased in paired fish (dominant and subordinate) compared with shams (P < 0.01). When we investigated cortisol and fasting as factors in regulating lxr abundance (Fig. 7B, df = 2; F = 4.28, P < 0.05), a significantly higher mRNA abundance in cortisol-treated fish compared with control fish (P < 0.05) was detected. Social status also differentially affected ugt1a3 mRNA abundance (Fig. 7C, df = 2; F = 14.05, P < 0.01), with transcript abundances differing significantly among all treatment groups (dominant > sham > subordinate, P < 0.05). Conversely, no significant differences among control, cortisol-treated, and fasted trout were observed (Fig. 7D, df = 2; F = 2.94, P > 0.05).

Fig. 7.

Abundance of mRNA markers of total cholesterol (TC) degradation pathways. Transcript levels of liver X receptor (lxr; A and B) and UDP glucuronosyltransferase family 1 member A3 (ugt1a3; C and D) in sham, dominant, and subordinate (A and C) and control, cortisol-treated, and fasted (B and D) rainbow trout (Oncorhynchus mykiss). Values are presented as means ± SE with n = 4–7 for all groups; values for individuals included in the means are indicated by the symbols. Bars that share a letter are not significantly different from one another (see text for details).

Hepatic mRNA Abundance of Drosha and Abundance of miRNAs Involved in Lipid Metabolism

The mRNA abundance of drosha was significantly elevated (Fig. 8A, df = 2; F = 14.23, P < 0.01) in both dominant (P < 0.05) and subordinate (P < 0.01) trout compared with shams. Similarly, both cortisol treatment and fasting induced a significant increase in drosha mRNA abundance (Fig. 8B, df = 2; F = 20.16, P < 0.01) compared with the control group (P < 0.01).

Fig. 8.

Abundance of mRNA of the rate-limiting enzyme in miRNA biogenesis drosha (A and B) and lipid metabolism-related miRNA-122 (C and D) and miRNA-33 (E and F) in sham, dominant, and subordinate (A, C, and E) and control, cortisol-treated, and fasted (B, D, and F) rainbow trout (Oncorhynchus mykiss). Values are presented as means ± SE with n = 5–7 for all groups; values for individuals included in the means are indicated by the symbols. Bars that share a letter are not significantly different from one another (see text for details).

The abundance of miRNA-122 was unaffected by social status (Fig. 8C, df = 2; F = 2.77, P > 0.05) or cortisol and fasting treatments (Fig. 8D, df = 2; F = 0.65, P > 0.05). However, social interactions (Fig. 8E, df = 2; F = 4.48, P < 0.05) significantly increased miRNA-33 in dominant compared with subordinate fish (P < 0.05). In addition, miRNA-33 abundance was significantly (Fig. 8F, df = 2; F = 9.42, P < 0.01) lower in both cortisol-treated (P < 0.01) and fasted fish (P < 0.05) relative to control fish.

Hepatic Cell Signaling

The ratio of Akt-p/Akt did not differ with social status (Fig. 9A, df = 2; F = 2.14, P > 0.05) or in response to cortisol treatment or fasting (Fig. 9B, df = 2; F = 3.15, P > 0.05). However, social status significantly affected the S6-p /S6 ratio (Fig. 9C, df = 2; H = 12.13, P < 0.01), with a significantly higher ratio in subordinate fish compared with sham (P < 0.01) and dominant fish (P < 0.05). A significant effect of treatment on S6-p/S6 ratio was also identified in experiment 2 (Fig. 9D, df = 2; H = 10.66, P < 0.01), with a lower S6-p/S6 ratio in fasted fish compared with cortisol-treated fish (P < 0.01).

Fig. 9.

Ratio of phosphorylated protein to total protein for Akt (A and B) and S6 (C and D) pathways in sham, dominant, and subordinate (A and C) and control, cortisol-treated, and fasted (B and D) rainbow trout (Oncorhynchus mykiss). Values are presented as means ± SE with n = 5–7 for all groups; values for individuals included in the means are indicated by the symbols. Bars that share a letter are not significantly different from one another (see text for details). Representative images of Western blots are included.

DISCUSSION

Confirmation of Social Status Phenotypes in Juvenile Rainbow Trout

Behavioral observations over a 4-day period allowed differentiation of dominant and subordinate phenotypes as described in previous studies (1, 14, 21, 34, 78). These behavioral distinctions were corroborated by clear differences in endocrine indexes of stress, evidenced by elevated plasma cortisol concentrations in subordinate fish, again as previously described (22, 66, 78). Clear physiological differences were identified between the subordinates and dominants, as evidenced by differences in food intake and specific growth rates (Table 1), again similar to previously reported studies (14, 21, 60). Collectively, these differences distinguish dominant and subordinate phenotypes, allowing the hypothesis that social status affects lipid metabolism to be tested. Because subordinates are characterized by increased circulating cortisol levels coupled with reduced food intake, cortisol-treated and fasted fish were used to address the possible contribution of these individual factors to changes in indexes of lipid metabolism.

Lipid Metabolism Indexes in Dominant Fish Suggest a Shift Toward Energy Storage

Dominant fish displayed elevated levels of circulating TG compared with shams and marginally elevated levels compared with subordinate fish. The molecular signature of increased hepatic mRNA abundance of the lipogenic genes srebp1c and fasn in dominant fish as well as acly relative to shams suggests that hepatic de novo lipogenesis contributes to this effect, at least in part. The srebp1c gene is a well-characterized lipogenic transcription factor that stimulates the hepatic de novo lipogenesis pathway by inducing fasn and acly transcription in response to physiological signals via insulin signaling (19, 42, 43). The enzyme products of fasn and acly catalyze palmitate synthesis and its building block, cytosolic acetyl-CoA, respectively (43). In trout, coinduction of hepatic srebp1c, fasn, and acly has been observed in fish selected for increased lipid content (77), during feeding after a short-term fast (57), and following insulin treatment (50). Because dominant trout monopolize the majority of the food offered to each pair, the increase in molecular markers of de novo lipogenesis suggests that selective activation of this pathway in dominants contributes to the observed increase in circulating TG. However, measurement of enzyme activities would be necessary to extrapolate our findings from mRNA abundance to the functional protein level. In addition to energetically costly de novo lipogenesis (79), other factors, especially intestinal absorption of dietary lipids and release of resynthesized dietary lipids, likely contribute to an increase in circulating TG (4). Indeed, a postprandial time-course study in rainbow trout suggested a peak in plasma TG concentration 12 h after the ingestion of a meal (5), timing that corresponds in the present study to the last feeding in the evening before animals were collected the following morning. Investigation of metabolic and molecular markers of intestinal lipid metabolism in rainbow trout (47) may provide insight into the specific contribution of lipid absorption. Similarly, study of metabolic and molecular changes in WAT (49) would be useful in confirming whether the pattern of changes in hepatic transcript abundance reflects a shift toward lipid synthesis for lipid storage in dominant trout.

Subordinate and sham trout were similar in terms of plasma TG levels and lipogenic mRNA abundances, and neither cortisol-treated nor fasted fish exhibited significant decreases in these variables from control fish. Collectively, these data support the hypothesis of a dominant phenotype in which diet-induced stimulation of circulating TG levels allows the initiation of energetically costly de novo lipogenesis. A comparable effect was proposed to explain elevation of hepatic glycogen stores in dominant trout above those in control fish (35).

Lipid Metabolism Indexes in Subordinate Fish Suggest a Reliance on Circulating FFAs for Hepatic β-Oxidation Pathways

In contrast to TG, circulating FFAs were elevated in subordinate rainbow trout compared with dominant and sham fish, and this increase coincided with increased mRNA abundance of the cpt1a isoform in the liver. Cpt1 is the rate-limiting enzyme of mitochondrial FFA β-oxidation in that it regulates FFA import into mitochondria (64). Together, these responses suggest increased reliance on FFAs for energy metabolism in the liver of subordinate rainbow trout. FFAs increase in response to fasting in rainbow trout within a week or two of cessation of feeding (26, 71), an effect generally attributed to liberation of FFAs from storage in WAT (46). However, although FFAs tended to increase in response to fasting and cortisol treatment, neither factor alone was sufficient to significantly elevate either circulating FFAs or the mRNA abundance of cpt1a. These differences between subordinate fish and fasted or cortisol-treated fish suggest that it is either the combination of both factors, or additional differences related to low social status, that cause FFAs to increase in subordinate trout. Investigation of molecular markers involved in FFA mobilization from WAT and/or muscle tissue is warranted.

Plasma TC is Lowered in Subordinate Rainbow Trout

In subordinate rainbow trout, TC, which serves among other roles as the precursor for cortisol synthesis, was significantly lower than in dominant trout. Examination of the mRNA abundances of genes involved in hepatic TC biosynthesis and degradation pathways did not reveal any significant differences between dominant and subordinate trout, with the exception of ugt1a3. Transcript abundance of ugt1a3 was elevated in dominant fish and reduced in subordinate fish relative to sham-treated trout. In humans, the UGT1A3 enzyme transforms hydrophobic TC metabolites into polar metabolites for biliary excretion (3), so these differences in ugt1a3 mRNA abundance might suggest a counterregulatory response to maintain TC homeostasis, particularly in subordinate fish where presumably there is an increase in TC demand for cortisol synthesis. The decrease in hepatic ugt1a3 mRNA abundance elicited by subordinate social status was not mimicked by cortisol treatment or fasting, suggesting again that additional factors are responsible for the differential regulation of ugt1a3 mRNA abundance observed in dominant vs. subordinate fish.

Hepatic miRNAs as Mediators of Social Status-Dependent Regulation of Lipid Metabolism

Lipid metabolism differs between dominant and subordinate rainbow trout, and while many factors may be responsible for these differences, we examined the role of hepatic miRNAs, because studies in rainbow trout have shown their regulation by nutritional status on the one hand (55, 57, 58) and their function to regulate hepatic and organismal lipid metabolism on the other (56), similar to findings from mammalian models (28, 73). The present study reports, for the first time, social status-dependent regulation of the rate-limiting miRNA biogenesis component Drosha (48), of which a single paralogue exists in rainbow trout (9). Social interaction increased drosha mRNA abundance in both dominant and subordinate trout compared with shams, suggesting that hepatic miRNA biogenesis plays a role in shaping social status-dependent hepatic metabolism in rainbow trout. Both cortisol treatment and fasting increased drosha mRNA abundance compared with controls, suggesting that both factors may contribute to the observed elevation of drosha mRNA abundance in subordinate trout. The factors responsible for increased drosha transcript abundance in dominant fish remain to be determined. In keeping with the focus of the present study on lipid metabolism, the liver-specific miRNA-122 was subsequently targeted because it previously had been shown to be an important regulator of hepatic de novo lipogenesis and circulating TG and TC concentrations in trout (56), as in mice (24, 25, 82) in vivo. However, no significant differences in miRNA-122 abundance were detected with respect to social status or cortisol or fasting treatments, suggesting that miRNA-122 is not a major contributor to the changes in lipid metabolism observed in the present study.

The hepatic mRNA abundance of miRNA-33 was suppressed in subordinate fish relative to dominant fish. Moreover, two hallmarks of the subordinate phenotype, fasting and elevation of cortisol, also significantly reduced miRNA-33 abundance, suggesting that these factors may contribute to the observed decrease in hepatic miRNA-33 abundance in subordinate rainbow trout. In fish as in mammals, miRNA-33 is located in srebp host genes and once transcribed supports the regulatory role of its host gene in hepatic and organismal lipid metabolism (18, 32, 36, 40–43, 65, 72). Functionally, several studies in mammalian models provide evidence for a role of srebp2 intron-encoded miRNA-33 in stimulating hepatic cholesterol abundance by limiting cholesterol efflux via abc transport proteins (32, 36, 41, 65, 72), thus augmenting the principal role of its srebp2 host gene as a key transcription factor in the cholesterol biogenesis pathway (42, 43). These and other studies in mammalian models have also shown inhibitory roles for miRNA-33 in hepatic β-oxidation by targeting enzymes involved in FA oxidation, including cpt1a (18, 32, 36). More recently, studies in mammalian model systems have revealed hepatic de novo lipogenesis as being regulated by miRNA-33 via inhibition of its target srebp1 (40). Conversely, evidence from what is currently the only study to have explored functional roles of miRNA-33 in a teleost, the marine species Siganus canaliculatus (85), points to a stimulatory role of miRNA-33 in Srebp1-mediated de novo lipogenesis by targeting insig1, a repressor blocking Srebp proteolytic activation. With regard to cholesterol metabolism, miRNA-33-dependent regulation of abca1 suggests a conserved role in the regulation of hepatic cholesterol metabolism between teleost fish and mammals (85). Although miRNA-33-dependent regulation of lipid metabolism in conjunction with its srebp host genes appears to be evolutionarily deeply conserved, as evidenced by its role in the negative regulation of β-oxidation in fat bodies of the fruit fly Drosophila melanogaster (18), the extrapolation of its functional role in rainbow trout is currently difficult, particularly because miRNA-33 can be differentially located in introns of srebp1 or srebp2 in teleost fish and other species (85) and because miRNA-target networks between mammals and fish can vary considerably (55). To extrapolate possible roles for miRNA-33 in the absence of functional studies of miRNA-33 in rainbow trout, we determined the genomic location of miRNA-33 in the National Center for Biotechonology Institute-deposited rainbow trout genome (www.salmobase.org) and then used rainbow trout-specific in silico predictions (59) to determine whether characterized miRNA-33 target genes involved in lipid metabolism in mammals are conserved predicted targets in rainbow trout.

With regard to its genomic location, we identified mature miRNA-33 sequences in the intron between exons 16/17 in two srebp2 gene loci, suggesting a synergistic function of miRNA-33 and srebp2 and, additionally, in an intergenic region (Fig. 10). With regard to predicted target genes involved in hepatic lipid metabolism, miRNA-33 targets in rainbow trout include few directly conserved targets identified in mammals, but nevertheless contain several target genes whose Gene Ontology annotation is enriched for lipid metabolism (Supplemental Table 1; Supplemental Material for this article is available online at the Journal website). This observation suggests that the regulation of hepatic lipid metabolism by miRNA-33 in rainbow trout is likely mediated via different target genes compared with mammalian models (54) and functional studies are clearly needed to establish the role of miRNA-33 in the regulation of hepatic lipid metabolism in rainbow trout before its role in mediating effects of social status on lipid metabolism in rainbow trout can be firmly established.

Fig. 10.

Schematic illustrating the genomic loci of miRNA-33 in rainbow trout (Oncorhynchus mykiss) derived from Salmobase (75).

Akt-Tor-S6 Signaling Cascade is Differentially Affected by Social Status

Because the insulin-responsive Akt-Tor-S6 cell signaling cascade upregulates hepatic lipogenesis while downregulating β-oxidation in rainbow trout (15, 50), the potential involvement of these upstream regulators in shaping differences in lipid metabolism between dominant and subordinate rainbow trout was investigated. There are multiple pathways involved in lipid metabolism; however, the Akt-Tor-S6 cell pathway has previously been functionally characterized and been directly linked to the acute regulation of mRNA abundance and enzyme activity of hepatic fas/Fasn and cpt1a/Cpt1 in studies employing rapamycin, a Tor inhibitor (15, 50). No significant differences in the phosphorylation status of Akt (indicative of Akt activity) were observed with social status, cortisol treatment, or fasting, suggesting that other metabolic signaling pathways that have been characterized in rainbow trout, such as PLC/PKC, MEK/ERK, and JAK–STAT (6–8), may be involved in this response. Additionally, it is relevant to note that fish were sampled in the morning, ~16 h after feeding, which may have masked differences that were previously reported to occur soon after feeding, as documented in a previous time-course study (57).

Interestingly, S6 phosphorylation status was elevated in subordinate trout. The regulation of S6P is mainly via S6K, which is also termed p70 RSK (29). In turn, S6K is prominently activated by the Akt-Tor pathway in trout (15) in response to metabolic hormones, such as insulin, as well as macronutrients, especially amino acids (16, 17, 50). Although not investigated in trout, a second Tor-independent pathway utilizing p90 RSK has been described as phosphorylating S6 serine residues 235/236 in mammals, albeit to a lesser extent (61). Given the lack of significant differences in Akt activation among treatment groups, the observed difference in S6 phosphorylation status is unlikely to have been mediated by endocrine factors, such as insulin. Fasted trout in the present study did not exhibit an increase in S6P compared with controls, suggesting that the increase in S6 phosphorylation in subordinate trout was not related to fasting. Future studies should investigate Tor activation status to clarify whether branched-chain AA-dependent p70 RSK or p90 RSK signaling is involved.

The paradigm that S6 phosphorylation is a global regulator of translation recently has been called into question, following pharmacological inhibition and genetic ablation experiments that inhibited S6 phosphorylation yet reported a lack of effect on translation in various conditions (61, 74, 80). Attention has returned to a role for S6 postulated in a pioneer study, which identified S6 as the only ribosomal protein that was phosphorylated in liver regeneration (37). In line with this possibility, recent evidence suggested a role for S6 phosphorylation in cell cycle progression, cell proliferation, and cell size determination (12, 61, 74, 80, 84). Speculatively, the strong activation of S6 in subordinate trout may reflect a physiological response to sustain liver function in the face of reductions in hepatosomatic index (12).

Perspectives and Significance

The current study revealed, for the first time, that social status in juvenile rainbow trout results in clear differences in lipid metabolism, reflected at the organismal level by increased plasma TG in dominant fish versus increased FFAs as well as decreased TC in the plasma of subordinate trout. Molecular signatures in the liver, a key hub of organismal lipid metabolism (69), provided insight into contributions of hepatic de novo lipogenesis to plasma TG concentrations in dominant rainbow trout and, conversely, increased β-oxidation pathways likely to utilize increased circulating FFAs in subordinate rainbow trout. Overall, these trends are suggestive of diet-dependent changes as reported in short-term fasted rainbow trout after acute refeeding (5, 57). In particular, increased TG and lipogenic gene expression in dominant compared with sham fish suggests a physiological signal, because dominant trout monopolize the ration offered to the paired fish, thus resulting in increased caloric intake compared with sham fish that were fed a single ration. Neither fasting nor cortisol treatment, two factors strongly associated with subordinate status, independently mimicked the circulating FFAs and hepatic cpt1a mRNA induction detected in subordinate trout. This observation suggests that either both factors together or additional factors are required to elicit these responses. Given the strong effects of social status on organismal and hepatic lipid metabolism in juvenile rainbow trout described in the present study, two areas of investigation will be of interest for future studies.

First, at the molecular level in the liver, the increase in paired trout of mRNA abundance of drosha, the rate-limiting enzyme in miRNA biogenesis (38), emphasizes a need for further study of possible roles of hepatic miRNAs in rainbow trout in social hierarchies. Our current study investigated the possible involvement of two specific miRNAs, miRNA-122 and miRNA-33, because of their importance in hepatic lipid metabolism in teleost fish (55), but combinatorial effects of miRNAs on lipid metabolism gene mRNAs in the liver can only be assessed by transcriptome level approaches (55) in conjunction with species-specific in silico prediction algorithms (59). Together, transcriptomic and bioinformatic approaches will help to uncover novel rainbow trout miRNA-target networks involved in regulating lipid metabolism (current study) and other metabolic pathways, such as glucose metabolism (35), in dominant and subordinate trout. In addition, the lower abundance of hepatic miRNA-33 in subordinate and cortisol-treated fish warrants investigation, as the role of hepatic miRNAs as possible mediators of metabolic effects of cortisol remains uncharacterized (27). Equally at the molecular level, the contribution of extrahepatic tissues, including the intestine, WAT, and muscle to social status-dependent changes in organismal lipid metabolism is warranted, and such studies should use a combination of lipid flux analysis (52, 53) and measurements of molecular markers of lipid metabolic pathways at the mRNA and enzymatic activity levels (16, 47, 49). With regard to the observed social status-dependent changes in hepatic cell signaling, and the significant activation of S6 in subordinate trout, future studies should address the upstream involvement of cell signaling pathway components, especially p70 RSK or p90 RSK and mTOR. At the tissue level, possible downstream effects S6P on liver plasticity warrant further investigation, as they contribute to the sustainability of altered metabolic demands in subordinate fish. Second, the central role of lipids in trout energy metabolism suggests that social status-dependent changes in lipid metabolism in rainbow trout may have functional consequences at physiological and ecological levels. For example, a fall in FFAs was linked to increased food intake in rainbow trout (51), raising the question of whether the sustained inhibition of feeding observed in subordinate rainbow trout, even following separation from the dominant fish (14), may be mediated at least in part by increased circulating FFAs. The subtle yet significant decrease in circulating TC in subordinate rainbow trout may reflect increased cortisol biogenesis and/or may impact cortisol biogenesis. At the ecological level, lipid metabolism in juvenile trout is predictive of migration strategy (11), and whether social status-dependent changes in lipid metabolism in juvenile fish translate into different life-history strategies warrants further study.

GRANTS

This work was supported by funds from Natural Sciences and Engineering Research Council Discovery (to K. M. Gilmour and J. A. Mennigen) and Research Tools & Instruments (to K. M. Gilmour) grants, John R. Evans Leader Fund from the Canadian Foundation for Innovation (to J. A. Mennigen), and Ontario Research Fund-Research Infrastructure from the Ministry of Research, Innovation and Science (to J. A. Mennigen).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

J.A.M. and K.M.G. conceived and designed research; D.J.K. and B.M.C. performed experiments; D.J.K., B.M.C., J.A.M., and K.M.G. analyzed data; D.J.K., B.M.C., J.A.M., and K.M.G. interpreted results of experiments; D.J.K. prepared figures; D.J.K. and J.A.M. drafted manuscript; D.J.K., B.M.C., J.A.M., and K.M.G. edited and revised manuscript; D.J.K., B.M.C., J.A.M., and K.M.G. approved final version of manuscript.