Regulation of energy metabolism during social interactions in rainbow trout: a role for AMP-activated protein kinase

Abstract

Rainbow trout (Oncorhynchus mykiss) confined in pairs form social hierarchies in which subordinate fish typically experience fasting and high circulating cortisol levels, resulting in low growth rates. The present study investigated the role of AMP-activated protein kinase (AMPK) in mediating metabolic adjustments associated with social status in rainbow trout. After 3 days of social interaction, liver AMPK activity was significantly higher in subordinate than dominant or sham (fish handled in the same fashion as paired fish but held individually) trout. Elevated liver AMPK activity in subordinate fish likely reflected a significantly higher ratio of phosphorylated AMPK (phospho-AMPK) to total AMPK protein, which was accompanied by significantly higher AMPKα₁ relative mRNA abundance. Liver ATP and creatine phosphate concentrations in subordinate fish also were elevated, perhaps as a result of AMPK activity. Sham fish that were fasted for 3 days exhibited effects parallel to those of subordinate fish, suggesting that low food intake was an important trigger of elevated AMPK activity in subordinate fish. Effects on white muscle appeared to be influenced by the physical activity associated with social interaction. Overall, muscle AMPK activity was significantly higher in dominant and subordinate than sham fish. The ratio of phospho-AMPK to total AMPK protein in muscle was highest in subordinate fish, while muscle AMPKα₁ relative mRNA abundance was elevated by social dominance. Muscle ATP and creatine phosphate concentrations were high in dominant and subordinate fish at 6 h of interaction and decreased significantly thereafter. Collectively, the findings of the present study support a role for AMPK in mediating liver and white muscle metabolic adjustments associated with social hierarchy formation in rainbow trout.

rainbow trout (Oncorhynchus mykiss) confined in small groups or pairs form social hierarchies in which dominant fish can be distinguished from subordinate fish on the basis of their behavior (42). Distinctive physiologies accompany these behavioral phenotypes (reviewed in Refs. 13, 19, 40). In particular, subordinate trout experience chronic social stress, leading to persistent elevation of the glucocorticoid stress hormone cortisol (38), and this factor, together with low food intake, reflecting monopolization of food resources by dominant fish (2, 28, 33) and appetite suppression in subordinate fish (7, 9, 33), typically results in low growth rates in subordinate compared with dominant fish (1, 7, 9, 31). Social status also appears to reprogram liver metabolism, particularly glycogen metabolism. In dominant trout, low glycogen phosphorylase activity in conjunction with abundant energy intake from feeding favors accumulation of hepatic glycogen reserves (14). By contrast, livers of subordinate fish exhibit enhanced gluconeogenic potential (14), in keeping with the reliance of these fish on on-board energy reserves. This situation reflects elevated hepatic phosphoenolpyruvate carboxykinase (PEPCK) activity coupled with a reduction in pyruvate kinase (PK) activity and a moderate reduction in glycogen phosphorylase activity (9, 14). Similarly, skeletal muscle PK activity was significantly lower in subordinate than dominant fish (22).

The factors responsible for these changes in metabolism with social status have yet to be fully described, particularly at the cellular level. A role for cortisol is likely, because cortisol treatment increases PEPCK mRNA abundance and activity (10, 44–46), and administration of the glucocorticoid receptor antagonist RU486 eliminated differences in PEPCK activity between dominant and subordinate fish (9). The differences in energy status between dominant and subordinate trout suggest that AMP-activated protein kinase (AMPK) also should play a role in regulating energy metabolism during social interactions. AMPK serves as a cellular energy sensor; once activated by increases in the ratios of AMP to ATP and ADP to ATP, which indicate falling cellular energy status, AMPK conserves ATP by inhibiting anabolic pathways and stimulates catabolic pathways that generate ATP, thereby stabilizing ATP levels and cellular energy status (15). In fish, activation of AMPK has been associated with changes in nutritional status (5, 34), increases in activity (25), and exposure to hypoxia (18, 41), all situations in which changes in energy metabolism are initiated. The available evidence for rainbow trout suggests that, as in mammals, AMPK exists as a heterotrimeric complex, where binding of AMP to the regulatory γ-subunit promotes phosphorylation of Thr¹⁷² in the catalytic α-subunit, thereby activating AMPK (34). Once activated, AMPK suppressed mRNA abundance of glucokinase, glucose-6-phosphatase, and fatty acid synthase in trout liver and hepatocytes (34), effects that are consistent with inhibition of anabolic pathways. In addition, AMPK activation in skeletal muscle from rainbow trout subjected to sustained swimming was associated with increased mRNA abundance of lipoprotein lipase, citrate synthase, and carnitine palmitoyltransferase 1β1b (25), effects that are consistent with stimulation of catabolic pathways.

Given the available evidence suggesting that AMPK in fish, as in mammals, serves as a cellular energy sensor, being activated by phosphorylation of the α-subunit in situations of falling cellular energy status, the present study investigated whether AMPK plays a role in regulating energy metabolism during social interactions in rainbow trout. To focus on the canonical mechanism of AMPK activation, cellular energy status was assessed by measurement of adenylate concentrations. To provide as complete a picture as possible of AMPK activation, particularly given the paucity of AMPK activity measurements in fish (18, 21, 25, 26), AMPK activity was measured together with changes in AMPK protein and transcript abundances. Specifically, AMPK activity, phosphorylated AMPK (phospho-AMPK) protein abundance, and AMPKα₁ mRNA abundance were assessed in liver and skeletal (white) muscle of dominant, subordinate, and sham-treated (handled in the same fashion as paired trout, but held individually) trout after 6 h, 24 h, or 3 days of social interaction. These sampling times were chosen to capture effects associated with hierarchy formation, as well as the consequences of subordinate social status. That is, if AMPK is involved in adjusting energy metabolism in response to the demands of social interactions, then the activity associated with hierarchy establishment would be predicted to activate muscle AMPK in interacting, but not sham, fish, particularly at 6 and 24 h. After 3 days of social interaction, lowered energy intake in conjunction with the energetic demands of chronic social stress would be predicted to elevate liver AMPK activity in subordinate fish relative to dominant or sham-treated fish.

MATERIALS AND METHODS

Experimental Animals

A domesticated strain of rainbow trout was obtained from the Fraser Valley Freshwater Trout Hatchery (Abbottsford, BC, Canada). The fish were transferred to the University of British Columbia, where they were held under a natural photoperiod in two flow-through tanks supplied with aerated, dechlorinated City of Vancouver tap water; water temperature was 12–13°C. The fish used in the present study were diploid but were held with triploid siblings (used in an independent study); visible implant elastomer tags were used to distinguish between diploid and triploid fish. Fish were fed twice daily with 2% body weight Bio Vita fry food. All holding and experimental protocols were approved by an Institutional Animal Care Committee (protocol AUP A13-0024) and were in compliance with the guidelines of the Canadian Council on Animal Care for the use of animals in research and teaching.

Experimental Protocol

Social hierarchies were established in fork-length-matched pairs of rainbow trout (fork-length difference averaged 0.9 ± 0.2% of fork length for 36 pairs in total; mean fork length = 9.6 ± 0.1 cm, mean mass = 9.5 ± 0.4 g, n = 71 fish) that were confined together for 6 h, 24 h, or 3 days. Sham-treated fish were handled identically but housed individually, rather than with a conspecific (mean fork length = 9.5 ± 0.1 cm, mean mass = 9.4 ± 0.4 g, n = 46 fish). For the 3-day interaction period, two groups of sham-treated fish were used: one group was offered food in the same manner as paired fish (see below), whereas the other group was fasted for the duration of the experimental period.

Rainbow trout 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) to allow assessment of fish mass, fork length, and condition of the dorsal and caudal fins. The anal fin was clipped for easy identification of individuals within a pair. Fork-length-matched fish were placed individually on either side of an opaque divider in an observation tank. The observation tanks consisted of 20-gallon (75-liter) glass aquaria held in an environmental chamber at 12°C, with a 12:12-h light-dark photoperiod. Three sides of each tank were covered with black plastic, so that adjacent tanks were visually isolated from each other. The water in each tank was aerated, and partial water changes were carried out every 48 h. After an overnight recovery period, the divider was removed, and fish were observed at least four times, for 5 min each time, during which they were scored for position in the tank and acts of aggression. For 6-h interactions, the four observation periods were separated by 1.5 h. For 24-h interactions, three observation periods were carried out over the course of the day the divider was removed, with assessment of feeding behavior at the end of the day. A final set of observations was collected before the fish were sampled. For pairs used in 3-day interactions, two observation periods per day were carried out, and feeding behavior was assessed at the second observation period each day, after which fish were fed to satiation (without measurement of food intake). Fin condition was reassessed at the end of the interaction period, and points awarded for fin damage accumulated during the interaction period, as well as position, aggressive behavior, and feeding, were combined by principal components analysis (Minitab, v16) to generate an overall behavior score for each fish. The point system, which associates higher scores with more-dominant behaviors, such as patrolling in the water column, carrying out aggressive acts, and monopolizing food resources, was similar to that used in previous studies (17, 29, 38). The fish within a pair that received the higher behavior score was assigned dominant status (Table 1). Pairs with similar scores (4 of 42 pairs) were excluded from further analysis.

Table 1.

Behavior scores for dominant and subordinate rainbow trout (Oncorhynchus mykiss) confined in pairs for 6 h, 24 h, or 3 days

Interaction Period	n	Dominant	Subordinate	ΔBehavior Score
6 h	13	0.9 ± 0.2	−0.9 ± 0.3	1.8 ± 0.4
24 h	12	1.4 ± 0.3	−1.3 ± 0.2	2.7 ± 0.5
3 days	11	1.7 ± 0.3	−1.7 ± 0.2	3.3 ± 0.4

Values are means ± SE; n, number of pairs. ΔBehavior score, difference in behavior score between the dominant and subordinate fish within a pair.

At the end of 6 h, 24 h, or 3 days of social interaction, fish were euthanized by terminal anesthesia (0.5 g/l ethyl p-aminobenzoate), with members of a pair being removed from the tank at the same time, and mass, length, and fin condition were assessed. Blood was collected by caudal severance into heparinized microhematocrit tubes (Thermo Fisher Scientific, Ottawa, ON, Canada), which were centrifuged at 6,000 g (Clay Adams Autocrit Ultra 3, Becton Dickinson, Mississauga, ON, Canada) for 5 min. Separated plasma was flash-frozen in liquid nitrogen and stored at −80°C for later analysis of plasma cortisol concentrations using a commercially available radioimmunoassay (MP Biomedicals, Solon, OH) that has been validated for trout plasma (12). All samples were included in a single assay, where intra-assay variation was 7.3% (% coefficient of variation). The liver and a sample of white muscle tissue were collected into preweighed microcentrifuge tubes, flash-frozen in liquid nitrogen, and stored at −80°C for later analysis of AMPK activity, AMPK protein abundance, or metabolite concentrations; samples for analysis of AMPKα₁ mRNA abundance were added to RNAlater (Life Technologies, Thermo Fisher Scientific) and held at 4°C for 4 h and then stored at −80°C.

Analytical Techniques

AMPK activity.

Liver or white muscle AMPK activity was measured using the SAMS peptide [³²P]ATP approach of Davies et al. (8) as described by Jibb and Richards (18) according to the procedures described by Jibb and Richards (18) with the following exceptions. Liver (~25–50 mg) or muscle (~100–200 mg) tissue samples in two volumes of ice-cold homogenization buffer were disrupted with a Kontes sonicator by five 5-s bursts at the highest setting. After purification was completed, the protein content of the purified, resuspended protein solution was determined using the Bradford protein assay (Sigma-Aldrich), and sample aliquots diluted to a protein concentration of 1 mg/ml were assayed for 5 min (liver) or 10 min (muscle) at 20°C.

Western blotting for AMPK protein abundance.

To measure liver or white muscle AMPK protein abundance by Western blotting, phospho-AMPKα (Thr¹⁷²) (catalog no. 2553) and AMPKα (catalog no. 2532) rabbit monoclonal antibodies (Cell Signaling Technology, New England Biolabs, Whitby, ON, Canada) were used. These antibodies have been used previously with trout tissues (34) and recognize α₁- and α₂-isoforms of the AMPK catalytic (α) subunit, with the phospho-AMPKα (Thr¹⁷²) antibody detecting these isoforms only when they are phosphorylated at Thr¹⁷². Briefly, frozen liver (~25–50 mg) or muscle (~100–200 mg) tissue was homogenized on ice using an Omni TH tissue homogenizer in a buffer containing 150 mM NaCl, 50 mM Tris (pH 8.0), 1.0% Triton X-100, an anti-phosphatase inhibitor cocktail (catalog no. 4906845001, Roche, Mississauga, ON, Canada), and an anti-protease inhibitor cocktail (catalog no. 4693159001, Roche). Homogenates were centrifuged for 15 min at 12,000 g, and the resulting supernatants were stored at −80°C. Protein concentrations were determined using the bicinchoninic acid protein assay (Sigma-Aldrich). Protein lysates (20 µg of protein) were subjected to SDS-PAGE and Western blotting (polyvinylidene difluoride membranes). After a 1-h incubation in blocking solution [5% skim milk in 1× Tris-buffered saline containing Tween 20 (TBST)], the membrane was incubated for 1 h with anti-phospho-AMPKα (Thr¹⁷²) or anti-AMPKα in a solution of 5% bovine serum albumin in 1× TBST. After they were washed, membranes were incubated (1 h) with an anti-rabbit IgG-horseradish peroxidase antibody (catalog no. 7074, Cell Signaling Technology) and washed, and chemiluminescence was developed with the Clarity Western ECL blotting substrate following the manufacturer’s guidelines (Bio-Rad, Mississauga, ON, Canada). Bands were visualized using the ChemiDoc XRS+ imaging system (Bio-Rad), and band intensities were quantified using the Image Lab software system (Bio-Rad). After visualization of bands for phospho-AMPK and AMPK, the polyvinylidene difluoride membranes were stripped [ReBlot Plus mild antibody stripping solution used according to the manufacturer’s instructions (EMD Millipore, Etobicoke, ON, Canada)] and reprobed using a β-actin antibody (catalog no. 4967, Cell Signaling Technology). Chemiluminescence was developed and detected as described above to account for differences in loading.

Real-time RT-PCR for AMPKα₁ relative transcript abundance.

The relative mRNA abundance of AMPKα₁ was evaluated in liver or white muscle tissue using semiquantitative real-time RT-PCR. Total RNA was extracted from liver or muscle tissue using an RNeasy Lipid Tissue Mini Kit (Qiagen, Toronto, ON, Canada) according to the manufacturer’s instructions. Extracted total RNA was quantified using a spectrophotometer (Nanodrop ND-2000c UV-Vis, Thermo Fisher Scientific). cDNA was synthesized using a QuantiTect reverse-transcription kit (Qiagen) according to the manufacturer’s protocol. A Rotor-Gene SYBR Green PCR kit (Qiagen) and a Rotor-Gene Q real-time PCR system (Qiagen) were used to quantify mRNA abundance in duplicate. The primers used for AMPKα₁ were those reported by Craig and Moon (6), ATC TTC TTC ACG CCC CAG TA (forward) and GGG AGC TCA TCT TTG AAC CA (reverse), which give an amplicon size of 131 bp, and are based on the sequence reported by Polakof et al. (34) (GenBank accession no. HQ403672). To account for differences in cDNA production and loading differences, the housekeeping gene 18S was used for normalization, with primers ATG GCC GTT CTT AGT TGG TG (forward) and CTC AAT CTC GTG TGG CTG AA (reverse), which yield an amplicon size of 146 bp (GenBank accession no. FJ710873https://www.ncbi.nlm.nih.gov/nuccore/) (6). Each reaction contained 10 µl of SYBR mix, 1 µl of forward and reverse primers (100 µM), 0.375 µl of ROX reference dye (1:500 dilution), 6.7 µl of RNase/DNase-free water, and 1 µl of 5×-diluted cDNA template. Cycling conditions were 3 min of initial denaturation at 95°C, 40 cycles of 95°C for 20 s, and 60°C for 20 s. Standard curves were constructed using serial dilutions of liver and muscle cDNA to optimize reaction compositions. Negative controls, including no-template controls (where cDNA was replaced with water) and no-reverse-transcriptase controls (where RNA was treated as for other cDNA reactions, but reverse transcriptase was replaced with water), were used to confirm that primers did not bind to genomic DNA. The mRNA abundance of AMPKα₁ relative to the housekeeping gene 18S, which was not affected by experimental treatments, was calculated using the 2^−ΔΔCt method (24).

Metabolite measurements.

Liver and white muscle ATP, CrP, and free creatine (free Cr) concentrations were measured enzymatically using the methods described by Bergmeyer (3) following measurement of intracellular pH (pH_i) using the approach of Pörtner et al. (36). Frozen liver (~25–50 mg) or muscle (~100–200 mg) tissue was disrupted in an ice-cold metabolic inhibitor solution [nitrilotriacetic acid and fluoride (36)] using a cell disrupter (Micro Ultrasonic KT50, Kontes, Vineland, NJ) by three 5-s bursts, and pH_i of the homogenate was measured using a micro combination pH electrode (Accumet 215, Thermo Fisher Scientific) and pH meter (sympHony benchtop meter, VWR, Edmonton, AB, Canada), after which 30% perchloric acid was immediately added to the sample to stop all metabolic reactions and facilitate metabolite extraction. Homogenates were centrifuged at 20,000 g for 5 min at 4°C, the supernatant was adjusted to pH 7.6 using 3 mol/l K₂CO₃, and the neutralized extract was centrifuged at 20,000 g for 5 min at 4°C to yield supernatant that was assayed for ATP, CrP, and free Cr concentrations. To minimize ATP hydrolysis at low pH, samples were processed in small batches and held on ice, so that <10 min elapsed from sample sonication to neutralization of sample pH.

Statistical Analyses

Specific growth rate (SGR) was calculated for the 3-day interaction period as [ln(m_Final) – ln(m_Initial)] × 100/D, where m is the mass of the fish in grams and D is the number of days that elapsed between measurements of mass. Hepatosomatic index (HSI) was calculated as (m_Liver/m) × 100, where m_Liver is the mass of the liver in grams. Free cytosolic ADP and AMP concentrations were calculated from measured concentrations of ATP, CrP, and free Cr, as well as pH_i, using the calculations described by Jibb and Richards (18). These calculations were not carried out for liver, because free Cr concentrations were too low for the data to be meaningful.

Data are reported as means ± SE. Two-way analysis of variance (ANOVA) was used to assess the impact of social status (dominant, subordinate, or sham) and interaction period (6 h, 24 h, or 3 days). These analyses did not include the fasted sham fish, which were incorporated in the experimental design only for 3-day interactions. Thus, in addition, a one-way ANOVA was carried out on all data for 3-day interactions to assess the impact of social status (dominant, subordinate, sham, or fasted sham) on the variable of interest. These analyses did not take into account the likelihood that the members of a pair are not independent of one another (4). This conservative approach facilitates statistical analyses that include both paired and sham fish but may make it more difficult to reject the null hypothesis of no effect of social status. When ANOVA yielded significant effects, the Holm-Sidak test was used for post hoc analyses. When data did not meet assumptions of normality or equal variance, the data were transformed to meet these assumptions, or equivalent nonparametric tests were carried out if the assumptions could not be met. The fiducial limit of significance was 5%, and statistical analyses were carried out using SigmaPlot v13.0 (Systat).

RESULTS

Rainbow trout identified as dominant or subordinate on the basis of their behavior (Table 1) also exhibited physiological differences. Plasma cortisol concentrations were significantly higher in subordinate than dominant or sham fish (Fig. 1A; 2-way ANOVA on log-transformed data, P < 0.001). Plasma cortisol concentrations also were significantly affected by interaction time, with significantly higher values at 6 and 24 h than at 3 days (P = 0.006). This effect appeared to be driven primarily by the fall in cortisol levels in dominant and sham fish over interaction time, although social status × interaction time failed to reach statistical significance (P = 0.086). Cortisol levels in sham fish were higher than typical “unstressed” values of ~10 ng/ml (12), particularly for the 6- and 24-h groups, probably owing to disturbance associated with the need for multiple behavioral observation periods during these short interaction periods. Examination of the data for 3-day interactions revealed significantly elevated plasma cortisol concentrations in subordinate fish, although this group did not differ significantly from the fasted sham fish (ANOVA on log-transformed data, P = 0.026). Trends for HSI were in many respects the opposite of those for plasma cortisol; HSI was significantly lower in subordinate than dominant or sham fish (Fig. 1B; 2-way ANOVA on reciprocal-transformed data, P < 0.001), and HSI increased significantly over interaction period (P = 0.031), with no significant interaction between interaction period and social status (P = 0.578). At 3 days of interaction, values were significantly lower for subordinate than dominant and sham, but not fasted sham, fish (ANOVA on reciprocal-transformed data, P = 0.005). These changes occurred on a background of weight loss in all fish for the 6-h interaction period, in sham fish held for 24 h, and in sham and fasted fish held for 3 days (Table 2). For the 3-day interaction period, SGR was calculated and was significantly influenced by social status (ANOVA, P < 0.001), with dominant fish (1.23 ± 0.58%/day, n = 11) exhibiting the highest growth rates. Interestingly, SGR of dominant fish did not differ significantly from that of subordinate fish (0.14 ± 0.40%/day, n = 10) but was significantly higher than SGR for sham (−0.86 ± 0.28%/day, n = 12) or fasted sham (−2.04 ± 0.35%/day, n = 10) fish.

Fig. 1.

Plasma cortisol concentrations (A) and hepatosomatic indexes (HSI; B) for dominant (dom) and subordinate (sub) rainbow trout (Oncorhynchus mykiss) confined in pairs for 6 h, 24 h, or 3 days and for sham-treated or fasted sham trout over corresponding periods. Values are means ± SE; n = 12–13 pairs and shams at 6 h and 24 h; for 3 days, n = 11 dominant, 10 subordinate and fasted, and 12 sham fish. Data were analyzed by 2-way ANOVA. Uppercase letters indicate a significant main effect of interaction period, with times that share a letter not significantly different from one another, and a significant main effect of social category is indicated by bar fill symbols. A 1-way ANOVA was carried out on treatment groups at 3 days; statistical symbols (lowercase letters) are shown in parentheses, and social status categories that share a letter are not significantly different from one another.

Table 2.

Changes in mass of rainbow trout confined in pairs or held individually for 6 h, 24 h, or 3 days or held individually without feeding for 3 days

Interaction Period	Dominant	Subordinate	Sham	Fasted
6 h	−0.3 ± 0.0* (13)	−0.3 ± 0.0* (13)	−0.4 ± 0.1* (12)
24 h	0.0 ± 0.2 (12)	−0.1 ± 0.1 (12)	−0.4 ± 0.1* (12)
3 days	0.4 ± 0.2 (11)	0.0 ± 0.2 (10)	−0.3 ± 0.1* (12)	−0.6 ± 0.1* (10)

Values (means ± SE of number of pairs in parentheses) are expressed in grams. Sham fish were held individually; fasted fish were held individually and fasted.

^*Significantly different from 0 (1-sample Student’s t-test, P < 0.05).

In liver tissue, social status and interaction time had significant effects on AMPK activity (Fig. 2A; 2-way ANOVA on square-root-transformed data, P = 0.001 for interaction period × social status). Within dominant and subordinate, but not sham, fish, liver AMPK activity increased significantly with interaction period. Significant effects of social status on liver AMPK activity were observed only for 3-day interactions, with significantly higher activity in subordinate than dominant or sham fish. Inclusion of fasted sham fish in the analysis revealed that values in subordinate fish did not differ significantly from those in fasted sham fish (ANOVA, P = 0.002). Measurements of liver phospho-AMPK protein abundance relative to total AMPK protein abundance, which also serves as an index of AMPK activity because phospho-AMPK is the active form of the enzyme, revealed higher ratios in subordinate than dominant and sham fish (Fig. 2B; 2-way ANOVA on log-transformed data, P < 0.001). These differences appeared to reflect differences in the level of phosphorylated protein, because total AMPK protein abundance was not affected by social status (data not shown; 2-way ANOVA on log-transformed data, P = 0.096). At 3 days of interaction, relative phospho-AMPK protein abundance was elevated in subordinate and fasted fish over dominant and sham fish (ANOVA on reciprocal-transformed data, P < 0.001). Total AMPK protein levels in subordinate fish were elevated, suggesting contributions of phosphorylation and increased AMPK protein abundance to the elevated ratio, whereas total AMPK protein levels in fasted sham fish were low, identifying phosphorylation as the main contributor to the elevated ratio (data not shown; ANOVA on log-transformed data, P < 0.001). The main effect of interaction time on liver phospho-AMPK protein abundance relative to total AMPK protein abundance also was significant (P < 0.001), with values peaking at 6 h. Owing to a sample mix-up, no protein measurements were available for sham fish at 6 h of interaction, which prevented analysis of the interaction period × social status term in the two-way ANOVA. To determine whether transcriptional regulation of AMPK could be contributing to the elevated activity in subordinate fish, the relative mRNA abundance of AMPKα₁ was measured. Liver AMPKα₁ relative mRNA abundance was significantly impacted only by social status (Fig. 2C; 2-way ANOVA, P = 0.002 social status, P = 0.246 interaction period, P = 0.166 interaction period × social status), with significantly higher values for subordinate than dominant or sham trout. At 3 days of interaction, values were significantly higher for subordinate than dominant and sham, but not fasted sham, fish (ANOVA, P = 0.007).

Fig. 2.

Effects of social interaction on AMP-activated protein kinase (AMPK) variables in liver of rainbow trout (Oncorhynchus mykiss), specifically AMPK activity (A), protein abundance of phosphorylated AMPK (phospho-AMPK) relative to total AMPK (B), and relative mRNA abundance of AMPKα₁ (C). Trout were confined in pairs for 6 h, 24 h, or 3 days, and sham-treated or fasted sham trout were held over corresponding periods. Values are means ± SE; n = 6 for all liver AMPK activity measurements, except dominant fish at 24 h of interaction and 3-day sham-treated fish, where n = 5. For relative protein and mRNA abundance measurements, n = 4, except protein measurements for sham-treated fish at 24 h and 3 days, where n = 6. Owing to a sample mix-up, no protein measurements were available for sham-treated fish at 6 h of interaction. Data in A and C were analyzed by 2-way ANOVA. In A, uppercase letters indicate a significant effect of interaction time within a social status category and lowercase letters indicate a significant effect of social status category within the 3-day interaction period; in all cases, groups that share a letter are not significantly different from one another. In C, main effect of social status is indicated by bar fill symbols. A 1-way ANOVA was carried out on each interaction time for B as well as the 3-day interaction times for A and C, where statistical symbols (lowercase letters) are shown in parentheses; social status categories that share a letter are not significantly different from one another.

Accompanying the differences in AMPK activity were significant effects of social status and interaction period on liver ATP and CrP concentrations, with similar trends for both variables. Liver ATP concentration decreased significantly over interaction period (Fig. 3A; 2-way ANOVA on square-root-transformed data, P = 0.001), an effect that was likely the result of falling ATP concentration in dominant and sham fish, although this effect did not reach statistical significance (P = 0.055 for interaction period × social status). Liver ATP concentration was significantly higher in subordinate than dominant, but not sham, trout (P = 0.001) or fasted sham fish (ANOVA, P = 0.001) at 3 days of interaction. Liver CrP concentration was significantly higher in subordinate than dominant and sham trout (Fig. 3B; 2-way ANOVA on log-transformed data, P = 0.011), but the main effect of interaction period on liver CrP concentration failed to reach statistical significance (P = 0.05, P = 0.132 for interaction period × social status). At 3 days of interaction, ANOVA revealed significant differences (P = 0.028), but post hoc analyses failed to reveal the origin of these differences.

Fig. 3.

Effects of social interaction on ATP (A) and creatine phosphate (CrP; B) concentrations in liver of rainbow trout (Oncorhynchus mykiss). Trout were confined in pairs for 6 h, 24 h, or 3 days, and sham-treated or fasted sham trout were held over corresponding periods. Values are means ± SE; n = 6 for all 6-h and 3-day interactions, except dominant fish at 3 days, where n = 4–5, and n = 4–5 for 24-h interactions. Data were analyzed by 2-way ANOVA. Uppercase letters indicate a significant main effect of interaction period, with times that share a letter not significantly different from one another, and a significant main effect of social category is indicated by bar fill symbols. A 1-way ANOVA was carried out on the treatment groups at 3 days; statistical symbols (lowercase letters) are shown in parentheses, and social status categories that share a letter are not significantly different from one another. Although 1-way ANOVA on liver CrP concentrations indicated that significant differences were present, post hoc comparisons could not pinpoint the source of these differences.

Skeletal (white) muscle AMPK activity was significantly influenced by social status, but not interaction period (Fig. 4A; 2-way ANOVA on log-transformed data, P = 0.020 for social status, P = 0.111 for interaction period, P = 0.125 for interaction period × social status), with significantly higher values for dominant and subordinate than sham fish. At 3 days of interaction, no significant effects of social status were detected (ANOVA, P = 0.228). By contrast, muscle relative phospho-AMPK abundance was significantly influenced by both social status and interaction period (Fig. 4B; 2-way ANOVA, P = 0.001 interaction period × social status). Muscle relative phospho-AMPK abundance was unaffected by interaction period in sham fish but peaked in dominant and subordinate fish at 24 h of social interaction. For all three interaction periods, subordinate fish exhibited the highest relative phospho-AMPK abundance, although at 3 days of interaction, the difference between dominant and subordinate fish was not significant, and unlike the situation in liver, values were significantly higher for subordinate than fasted sham fish (ANOVA on log-transformed data, P < 0.001). Changes in the abundance of phospho-AMPK appeared to be driven by changes in total AMPK protein abundance and phosphorylation, because analysis of total AMPK protein abundance revealed patterns that generally paralleled those of the ratio (data not shown; 2-way ANOVA, P = 0.021 for interaction period, P < 0.001 for social status, P = 0.005 for interaction period × social status). However, at 3 days of interaction, total AMPK protein levels were not influenced by social status (data not shown; ANOVA on ranks, P = 0.06). Relative mRNA abundance of AMPKα₁ was measured in skeletal muscle to assess whether AMPK was regulated transcriptionally, and significant main effects of interaction period and social status were detected (Fig. 4C; 2-way ANOVA on ranked data, P = 0.682 for interaction period × social status). Muscle AMPKα₁ relative mRNA abundance decreased significantly with interaction period (P < 0.001) and was significantly higher in dominant than sham fish (P = 0.042). At 3 days of social interaction, no effect of social status was apparent (ANOVA, P = 0.079).

Fig. 4.

Effects of social interaction on AMPK variables in white muscle of rainbow trout (Oncorhynchus mykiss), specifically AMPK activity (A), the relative protein abundance of phosphorylated AMPK to total AMPK (B), and relative mRNA abundance of AMPKα₁ (C). Trout were confined in pairs for 6 h, 24 h, or 3 days, and sham-treated or fasted sham trout were held over corresponding periods. Values are means ± SE, with n = 5–6 for muscle AMPK activity measurements and n = 4 for all relative protein and mRNA abundance measurements. Data were analyzed by 2-way ANOVA. In A and C, uppercase letters indicate a significant main effect of interaction time, while the main effect of social status is indicated by bar fill symbols. In B, uppercase letters indicate a significant effect of interaction time within a social status category, and lowercase letters indicate a significant effect of social status category within an interaction period. A 1-way ANOVA was carried out on the treatment groups at 3 days; statistical symbols (lowercase letters) are shown in parentheses. In all cases, groups that share a letter are not significantly different from one another.

Muscle ATP and CrP concentrations were significantly affected by social status and interaction period. For ATP concentrations, there was a significant interaction between these factors (Fig. 5A; 2-way ANOVA on log-transformed data, P = 0.006); in dominant and subordinate, but not sham, fish, muscle ATP levels fell significantly between the 6-h and 24-h interaction periods. Largely as a result of this effect, muscle ATP concentrations were significantly higher in dominant and subordinate than sham fish at 6 h, but not 24 h, of interaction. At 3 days of interaction, differences in muscle ATP concentrations were small but significant, with lower values for subordinate and fasted sham than dominant or sham fish (ANOVA, P = 0.003). Muscle CrP concentrations exhibited similar decreases between the 6-h and longer interaction periods (Fig. 5B; 2-way ANOVA on rank data, P = 0.005). In addition, a significant main effect of social status was detected (P = 0.004), with values for dominant but not subordinate fish being significantly higher than those for sham fish. However, the interaction of social status and interaction period did not reach statistical significance (P = 0.051). No effects of social status were apparent at 3 days of interaction (ANOVA, P = 0.573). By contrast, the ratio of muscle free AMP concentration to ATP concentration increased significantly as a function of interaction period (Fig. 5C; 2-way ANOVA, P < 0.001), with significantly higher values for subordinate and sham than dominant fish (P = 0.007); the interaction term for this analysis was not significant (P = 0.881 interaction period × social status). This trend reflected the corresponding decrease in muscle ATP concentration (Fig. 5A) and increase in muscle free AMP concentration (Table 3) with interaction period. At 3 days of social interaction, the ratio of muscle free AMP concentration to ATP concentration was significantly lower in dominant than fasted sham fish (ANOVA on ranks, P = 0.032).

Fig. 5.

Effects of social interaction on ATP (A) and CrP (B) concentrations in liver of rainbow trout (Oncorhynchus mykiss), as well as the calculated ratio of free AMP to ATP concentrations (C). Trout were confined in pairs for 6 h, 24 h, or 3 days, and sham-treated or fasted sham trout were held over corresponding periods. Values are means ± SE; n = 4–6. Data were analyzed by 2-way ANOVA. In A, uppercase letters indicate a significant effect of interaction time within a social status category, and lowercase letters indicate a significant effect of social status category within an interaction period. In B and C, uppercase letters indicate a significant main effect of interaction time, while the main effect of social status is indicated by bar fill symbols. A 1-way ANOVA was carried out on the treatment groups at 3 days; statistical symbols (lowercase letters) are shown in parentheses. In all cases, groups that share a letter are not significantly different from one another.

Table 3.

White muscle pH_i, free [Cr], [ADP_free], and [AMP_free] in rainbow trout confined in pairs for 6 h, 24 h, or 3 days and for sham-treated and fasted sham fish held for corresponding periods

	6-h Interaction			24-h Interaction			3-Day Interaction
Measure	Dom	Sub	Sham	Dom	Sub	Sham	Dom	Sub	Sham	Fasted
pHi	7.04 ± 0.03 (5)	7.12 ± 0.05 (6)	7.09 ± 0.05 (5)	7.17 ± 0.04 (4)	7.23 ± 0.02 (5)	7.10 ± 0.02 (5)	7.09 ± 0.04 (6)	7.14 ± 0.05 (6)	7.00 ± 0.02 (6)	7.13 ± 0.04 (6)
Free [Cr], µmol/g tissue	2.53 ± 0.21 (5)	2.71 ± 0.24 (4)	2.24 ± 0.20 (5)	2.70 ± 0.39 (6)	2.73 ± 0.52 (5)	2.60 ± 0.39 (6)	4.98 ± 0.95a (6)	7.54 ± 0.54a,b (5)	8.14 ± 0.69b (5)	8.85 ± 0.40b (5)
[ADPfree], µmol/l	14.10 ± 3.64A (5)	12.98 ± 2.17 (4)	6.32 ± 1.11A (5)	6.00 ± 1.36B (6)	6.11 ± 1.99 (5)	6.48 ± 2.18A (6)	9.63 ± 2.39A,B,a,b (6)	7.24 ± 1.58b (5)	17.35 ± 2.75B,a (4)	5.05 ± 2.07b (4)
[AMPfree], nmol/l	89.49 ± 32.47 (5)	109.23 ± 31.78 (4)	65.79 ± 16.68 (5)	70.40 ± 22.45 (6)	76.84 ± 14.57 (5)	92.96 ± 31.14 (6)	285.45 ± 111.42 (6)	256.42 ± 43.09 (5)	577.73 ± 83.96 (4)	267.52 ± 124.68 (4)

Values are means ± SE of number of pairs in parentheses. pH_i, intracellular pH; Cr, creatine. Data were analyzed by 2-way ANOVA. For pH_i, the main effect of social status was significant (P = 0.019); muscle pH_i was significantly higher in subordinate (Sub) than sham, but not dominant (Dom), fish. Main effect of interaction period also was significant (P = 0.041), but post hoc comparisons did not reveal the origin of this difference. For free [Cr] and [AMP_free], the main effect of pairing time was significant (P < 0.001 in both cases), with significantly higher values at 3 days than at 6 h or 24 h. For [ADP_free], a significant interaction between interaction period and social status was detected (P = 0.008). Significant effects of social status were observed at 3 days of interaction, indicated by lowercase letters. Significant effects of interaction period were observed in dominant and sham fish, indicated by uppercase letters. A 1-way ANOVA was used at 3 days of interaction. Significant effects of social status were observed for free [Cr] (P = 0.005) and [ADP_free] (P = 0.015), but not pH_i (P = 0.253) or [AMP_free] (P = 0.130), and are indicated by lowercase letters. In all cases, groups that share a letter are not significantly different from one another.

DISCUSSION

The findings of the present study support the hypothesis that AMPK is activated in liver of rainbow trout by the demands of chronic social stress and in white muscle by the demands of social interactions; therefore, AMPK would be expected to be involved in mediating metabolic adjustments associated with social status. Effects in the liver were most pronounced at 3 days of social interaction, when AMPK activity, relative abundance of phosphorylated AMPKα protein, and AMPKα₁ mRNA abundance in subordinate fish were significantly elevated over those in dominant or sham fish, probably as a response to a situation in which energy consumption would be elevated by the demands of chronic social stress (39) while the animal is likely to be experiencing lowered energy intake (2, 9, 28, 33). The chronic social stress experienced by subordinate fish was evident from their high plasma cortisol concentrations. High circulating glucocorticoid levels, in turn, may have contributed to the increased AMPKα₁ transcript abundance in these fish, because treatment with the synthetic glucocorticoid dexamethasone increased AMPKα₁ and AMPKα₂ mRNA and protein expression in the liver (but not skeletal muscle) of rats and mice (43). Typically, food intake in subordinate fish is restricted because dominant fish monopolize and defend food resources (2, 28, 33) and remains low even after removal of dominant fish, suggesting that physiological suppression of appetite also occurs (7, 9, 33). Using a fasting-re-feeding protocol, Polakof and colleagues (34) reported that liver AMPK protein phosphorylation status in rainbow trout was significantly decreased 1 h after the trout were fed from the value observed in fish that had been fasted for 10 days, and tended to increase by 24 h after the meal. After 3 wk of fasting, an increase in liver AMPKα₁ mRNA abundance was detected in zebrafish, and this increase was eliminated within 24 h of refeeding (5). Similarly, in the present study, sham fish that were fasted for 3 days exhibited increased AMPKα phosphorylation status and trends for higher AMPK activity and AMPKα₁ mRNA abundance than sham fish that were fed daily. These findings are in agreement with the conclusion (5, 34) that liver AMPK is responsive to nutritional status in teleost fish and that increases in AMPK protein levels and AMPK phosphorylation status contribute to elevated AMPK activity under these conditions. Moreover, the marked similarity of responses between subordinate trout and fasted sham fish suggests that low food intake may serve as a key trigger of the AMPK response in subordinate trout.

The mechanism through which low food intake is coupled to AMPK activation in trout remains to be investigated. Liver ATP concentrations in subordinate and fasted sham trout were elevated over values for dominant or sham trout, suggesting that AMPK was not activated by the canonical mechanism of increases in cellular AMP and ADP concentrations (15). In mammals, AMPK serves as a glycogen sensor (23, 27, 35). The β-subunit of AMPK contains a carbohydrate-binding module that interacts with the kinase domain of the catalytic α-subunit in a fashion that is modulated by glycogen binding and phosphorylation status (23). Some (27), but not all, evidence (23, 35) suggests that glycogen binding inhibits AMPK activity. Previous studies in rainbow trout have reported low liver glycogen concentrations in subordinate as well as fasted fish (7, 14, 30), consistent with the low HSI values of subordinate and fasted sham trout in the present study. If AMPK binds and is regulated by glycogen in trout, as it does in mammals, a possibility that requires investigation, then low liver glycogen levels may provide a link between low food intake and AMPK activation. Alternatively or additionally, AMPK activation may be affected by hormones that regulate liver metabolism, as appears to be the case in mammals (16). For example, plasma leptin levels in rainbow trout increase during fasting (20), and several lines of evidence link plasma leptin levels and liver AMPK activity in mammals (16). In fish, a link between leptin and AMPK activity was suggested by concomitant increases in plasma leptin levels and muscle AMPK phosphorylation status in fasting fine flounder (11). The effects of chronic social stress on plasma leptin levels and/or leptin receptor expression clearly warrant investigation.

By contrast with the cohesive pattern of AMPK responses in the liver, the results for white muscle in the present study were more variable. Differences between dominant and subordinate fish were less marked than in the liver, as were similarities between subordinate and fasted sham trout. Skeletal muscle in salmonid fish contains higher energy stores than the liver, particularly with respect to lipids, which are the predominant energy source in fish (47); this effect may help buffer skeletal muscle tissue against increased energy demands, making it more difficult to detect effects of social stress. However, some responses appeared to be present in interacting (i.e., dominant and subordinate) fish and, at least in some cases, were more apparent at shorter interaction periods. For example, muscle AMPK activity was significantly elevated in interacting compared with sham fish, and muscle ATP concentrations were significantly higher in interacting than sham fish, but only for the 6-h interaction period. Interacting effects of nutritional status and physical activity, both of which may influence white muscle AMPK activity (11, 25), may have contributed to these patterns.

The early stages of hierarchy formation in rainbow trout are characterized by displays that rapidly escalate to aggressive behaviors such as circling, chasing, and nipping, with both fish being involved. This stage usually lasts <2 h and ends when the loser of the encounter switches from countering to fleeing the aggressive attacks of the winner, initiating a second phase of the interaction, which features unidirectional aggression of the dominant fish toward the subordinate (1, 32, 38, 48). Whether the high level of physical activity experienced by both fish during hierarchy establishment could serve as a stimulus for AMPK activation in muscle requires investigation. White muscle AMPK activity was stimulated in rainbow trout subjected to sustained swimming at 0.75 body lengths/s for 40 days, as well as in cultured myotubes that were electrically stimulated to contract in vitro for several hours, suggesting that hours, as opposed to days, of activity are sufficient to activate muscle AMPK (25). Moreover, AMPK activation in brown trout (Salmo trutta) cultured myotubes was associated with increased glucose uptake and utilization (25, 26), and sustained swimming in rainbow trout increased the expression of AMPK target genes that are involved in energy use (25), factors that may have contributed to the higher levels of ATP in white muscle of interacting fish after 6 h of interaction. Activity in the dominant fish remains high, as the interaction is extended because the dominant fish patrols the water column and also may continue to harass the subordinate fish, albeit more sporadically over longer interaction periods (33, 48). Such sustained physical activity could maintain AMPK activation in white muscle, and the higher AMPKα₁ transcript abundance observed in dominant fish also could contribute to maintaining elevated AMPK activity. Subordinate fish, on the other hand, limit swimming activity (33) but experience elevated standard metabolic rate (39) and limited food intake (33). In white muscle of fine flounder, as in the liver of rainbow trout (34), AMPK phosphorylation status was increased by fasting and fell within hours of refeeding (11). Thus white muscle AMPK activity in subordinate trout may respond to nutritional cues, rather than swimming activity, at 3 days of interaction. Although the combined effects of activity and fasting may account for the higher AMPK activity in dominant and subordinate than sham fish, AMPK phosphorylation status in the present study was higher in subordinate than dominant fish, except at 3 days of interaction. The apparent differences between AMPK phosphorylation status and AMPK activity were unexpected and remain to be explained. An intriguing possibility is that AMPK activity may be affected by the number and type of subunit isoform combinations present, as appears to be the case in mammals (37).

Perspectives and Significance

Collectively, the findings of the present study support a role for AMPK in mediating the adjustments of liver and white muscle metabolism that are associated with social status in rainbow trout, with AMPK likely responding to changes in food intake and physical activity that occur during social interactions. In dominant trout, AMPK activity and AMPKα₁ transcript abundance may be increased in skeletal muscle in response to the sustained physical activity that accompanies hierarchy establishment and maintenance. Activation of AMPK would be expected to enhance glucose uptake and utilization, as well as lipid mobilization and fatty acid oxidation (25, 26), thereby matching metabolism to the increased energetic demands of physical activity. The elevated white muscle PK activity reported in dominant over subordinate trout (22) could be a downstream effect of AMPK activation, as AMPK activation in trout cultured myotubes resulted in increased PK gene expression (26). In subordinate trout, elevated liver AMPK activity, phospho-AMPK protein abundance, and AMPKα₁ transcript abundance, in conjunction with parallel observations in fasted sham fish, are consistent with the prediction that liver AMPK activity is elevated in subordinate fish to meet the energetic demands of chronic social stress in an animal experiencing lowered energy intake. It is tempting to speculate that cortisol and AMPK interact to remodel liver metabolism, favoring gluconeogenesis [effects of cortisol on PEPCK (9)] and, presumably, inhibiting energy-consuming lipogenic and protein synthesis pathways via AMPK (5, 34). Clearly, additional studies are needed to investigate the mechanisms through which food intake and activity may influence AMPK and to identify the specific downstream targets of activated AMPK during social interactions. The evidence of the present study implicating AMPK in the regulation of liver and white muscle metabolism during hierarchy establishment in rainbow trout adds to the growing body of evidence that AMPK in fish (5, 18, 21, 25, 26, 34, 41), as in mammals (15), serves as a cellular energy sensor and metabolic “master switch.”

GRANTS

This work was supported by Natural Sciences and Engineering Research Council of Canada Discovery and Research Tools & Instruments grants to K. M. Gilmour and J. G. Richards. G. Y. Lau was supported by a Natural Sciences and Engineering Research Council of Canada Graduate Scholarship.

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

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