Fish Physiology and Biochemistry 29: 225–235, 2003. © 2004 Kluwer Academic Publishers. Printed in the Netherlands.
225
Relationship between dietary lipid source, oxidative stress, and the physiological response to stress in sub-yearling chinook salmon (Oncorhynchus tshawytscha) Thomas L. Welker1 and James L. Congleton2 1 Aquatic
Animal Health Research Laboratory, Agricultural Research Service, United States Department of Agriculture, Auburn, AL 36832-0952, USA (Phone: 334-887-3741; Fax: 334-887-2983; E-mail:
[email protected]); 2 Idaho Cooperative Fish and Wildlife Research Unit, United States Geological Survey, College of Natural Resources, University of Idaho, Moscow, ID 83844-1141, USA
Accepted: 7 June 2004
Key words: cortisol, dietary oil, fatty acid, lipid peroxidation, stress response
Abstract Relationships between dietary lipid source, stress, and oxidative stress were examined in juvenile chinook salmon (Oncorhynchus tshawytscha). Four different experimental diets were used: menhaden oil (MHO; elevated 20:5n3 and 22:6n-3), soybean oil (SBO; elevated 18:2n-6), linseed oil (LSO; elevated 18:3n-3), and a mixture of 55% linseed oil and 45% soybean oil (MIX; approximately equal levels of 18:2n-6 and 18:3n-3). Juvenile salmon (initial body weight of 16.0 g) were fed experimental diets for 12 weeks (early March to early June). At the end of feeding, fish subjected to a low-water stressor for 96 h had greater liver and brain lipid peroxidation compared to unstressed controls; peroxidation was not influenced by diet. Diet and stress affected plasma cortisol levels. Stressed fish fed SBO had the greatest cortisol concentrations, followed by MIX, MHO, and LSO (mean concentrations for the SBO and LSO diets differed significantly). The cortisol response to stress may have been influenced by the ratio of prostaglandin 1- and 2-series to prostaglandin 3-series precursor fatty acids provided by the different diets. The results of this study suggest a connection between the physiological response to stress, dietary lipid quality, and oxidative stress. This is the first evidence of such a relationship in fish. Abbreviations: AA – arachidonic acid; ACTH – adrenocorticotropin; BHT – butylated hydroxytoluene; BLPO – brain lipid peroxidation; dGLA – dihomo-γ -linolenic acid; DHA – docosahexanoic acid; EPA – eicosapentanoic acid; FER – feed efficiency ratio; FOX – ferrous oxidation-xylenol orange; GLA – γ -linolenic acid; LA – linoleic acid; LCO3 – long-chain n-3 polyunsaturated fatty acids; LLPO – liver lipid peroxidation; LN – linolenic acid; LPO – lipid peroxidation; LSO – linseed oil; MHO – menhaden oil; MIX – 55% linseed oil + 45% soybean oil; PC – plasma cortisol; PG – prostaglandin(s); PGE2 – prostaglandin E2 ; PUFA – polyunsaturated fatty acid; SBO – soybean oil.
Introduction Recent evidence suggests a connection may exist between the physiological response to stress, tissue fatty acid composition, and oxidative stress. In mammals (Bondarenko et al. 1985; Liu et al. 1996) and fish (Welker and Congleton 2004), the physiological stress response can lead to oxidative stress. This phe-
nomenon has been attributed to catecholamine autoxidation (Bindoli et al. 1992), secondary effects of glucocorticoids (Nishigori et al. 1984), and free radicals generated during the production of prostaglandins (PG) and other eicosanoids by the AA cascade (Liu et al. 1996). Little attention has been focused on the influence of dietary lipids on the stress response, despite
226 evidence that the production of PG is affected by dietary lipids and that PG may play a role in stress hormone release (Gupta et al. 1985; Wales 1988). Prostaglandins are biologically active derivatives of arachidonic acid (AA; 20:4n-6) and (to a lesser degree) dihomo-γ -linolenic acid (dGLA; 20:3n-6) of the n-6 pathway, and of eicosapentanoic acid (EPA; 20:5n-3) of the n-3 pathway (Bell et al. 1994a). Homologues of PG are produced from n-6 and n-3 precursors (e.g., prostaglandin E2 or PGE2 from AA, and PGE3 from EPA) that exert similar biological effects in fish and mammals. Arachidonic acid-derived PG are much more biologically active than EPA-derived PG in both mammals (Calder 1998) and fish (Ashton et al. 1994), and AA is preferred for prostaglandin synthesis, even though EPA is more abundant in fish tissues (Bell et al. 1997). In mammals, PG derived from n-6 fatty acids have been implicated in mediation of the adrenocorticotropic hormone (ACTH) response to psychological and physical stress (Bugajski et al. 1996). Arachidonic acid, PGE2 (Wales 1988), and PGE1 (Gupta et al. 1985) may stimulate cortisol release in fish. Dietary fatty acids can also influence the susceptibility of tissue lipids to peroxidative damage. Increasing dietary n-3 polyunsaturated fatty acids (PUFA) can elevate the unsaturation index of lipids in fish tissues and make them more prone to free radical attack (Stéphan et al. 1995). Fish tissues and commercial diets supplemented with marine fish oils containing high levels of PUFA are highly susceptible to peroxidative damage. The fatty acid composition of dietary oil can influence osmoregulatory functions during the parr-smolt transformation (Bell et al. 1997; Tocher et al. 2000). Bell et al. (1997) suggested that the ability to osmoregulate is regulated by the AA:EPA ratio in tissue phospholipids. Diets with fatty acid compositions that closely resemble those found in freshwater insects (e.g. vegetable oil mixtures; equal levels of linoleic acid and linolenic acid and low levels of EPA and docosahexanoic acid or DHA, 22:6n-3) may be more beneficial to fish undergoing the parr-smolt transformation than diets containing marine fish oils (high EPA and DHA and low LN), which are commonly fed to juvenile salmon (Bell et al. 1994b and 1997). A shift to vegetable oil based diets may provide osmoregulatory benefits to juvenile anadromous salmonids; however, this change is likely to affect other aspects of fish physiology, such as the response to stress, that are as of yet unexplored. We examined the effects of four dietary oil treatments containing different ratios of n-3 and n-6 fatty
Table 1. Composition of experimental diets (dry matter basis). Ingredient1
% of diet
Low-fat fish meal2 Dextrin3 Oil3 (Menhaden, Soybean, Linseed, or 55% Linseed + 45% Soybean oil) Vitamin premix4 Mineral premix5 α-cellulose3 Gelatin3 Choline chloride (60%)3 α-tocopherol6
45.0 22.2 12.9 3.1 4.1 8.2 4.0 0.5 Adjusted to 200 mg kg−1 diet
1 All reagents supplied by Sigma Chemical, Inc., St. Louis, MO
unless otherwise stated. 2 Provided by Ron Hardy, Director, Hagerman Fish Culture Sta-
tion, Hagerman, Idaho: 8.73% moisture; 9.5% fat of dry weight; 72.8% protein of dry weight. 3 ICN Biochemicals, Aurora, OH. 4 Supplied as (mg kg−1 diet): retinol palmitate, 5.0; cholecalciferol, 0.025; menadione sodium bisulfite, 10.0; thiamin, 30.0; riboflavin, 25.0; pyridoxine mono-HCl, 30.0; cyanocobalamin, 0.02; nicotinic acid, 200.0; D-pantothenic acid, 50.0; D-biotin, 1.0; folic acid, 10.0; myo-inositol, 500.0; α-tocopherol acetate, 50.0; ascorbate-2-polyphosphate, 100.0. 5 Supplied as (g kg−1 diet): KCl, 15.0; CaHPO dibasic, 12.0; 4 MgSO4, 2.58; NaCl, 3.0. Trace minerals supplied as mg kg−1 diet: MnSO4 •H2 O, 20.0; CuSO4 •5H2 O, 5.0; ZnSO4 •H2 O, 40.0; KI, 10.0; FeSO4 •7H2 O, 70.0; Na2 SeO3 , 0.01. 6 Determined in the laboratory of Dr. Boon Chew, Dept. of Animal Sciences, Washington State University, Pullman, WA.
acids on the health of juvenile spring chinook salmon (Oncorhynchus tshawytscha). Diets were evaluated for their influence on tissue lipid composition, growth, feed efficiency ratio, oxidative stress, and the physiological stress response.
Materials and methods Fish rearing On 3 March 1999, spring chinook salmon parr (n=40 per aquaria, mean weight of 16.0 g) were randomly allocated to each of 24 glass aquaria (114 l; 45 cm × 61 cm × 45 cm). Four experimental diets were used: ingredients of each were the same, except for the dietary oil source (Table 1). The diets contained menhaden oil (high levels of DHA and EPA), soybean oil (high levels of linoleic acid or LA, 18:2n-6), linseed oil (high levels of linolenic acid or LN, 18:3n-3), or a mixture of 55% linseed oil and 45% soybean oil
227 (approximately equal levels of LA and LN). Feeding of experimental diets began on 8 March 1999, with each diet randomly allocated to six aquaria. The stocking density for each tank was approximately 6.2 kg m−3 initially and 10.7 kg m−3 by study’s end. Tank water inflow was 4.0 l min−1 , and water temperatures ranged from 8 to 12 ◦ C. Feeding rates for experimental diets were based on the caloric content of dry feed (3240 cal kg−1 digestible energy) and were approximately 1.5% body weight day−1. Feeding rates were calculated as described in Klontz et al. (1980). Every three weeks, a random sample of fish (n=10) was removed from each tank and anesthetized with buffered MS-222 (tricaine methanesulfonate, 50 mg l−1 ). Fork length (nearest 1 mm) and wet weight (nearest 0.1 g) were recorded and used to determine growth and to adjust feeding rate calculations. Influence of the diets on growth was determined by comparing the final weights of fish taken at the conclusion of the study. The feed efficiency ratio (kg wet gain kg−1 dry feed fed; FER) was also calculated for each diet. Fish sampling On 9 June 1999, fish in three randomly selected tanks per dietary treatment were subjected to low-water stress for 96 h by lowering the water level in each tank until the dorsal fin of fish just broke the water surface (approximately 6.5–7.5 cm deep); the three remaining tanks for each dietary treatment served as controls. After termination of low-water stress, four fish were randomly sampled from each tank and killed with a lethal dose of MS-222 (200 mg l−1 ), and wet weights and fork lengths were recorded. Whole blood was taken from the caudal vasculature with a 1 ml heparinized (air-dried; 800 U ml−1 ) syringe fitted with a 20-gauge needle. Whole blood samples were immediately placed on ice and subsequently centrifuged at 2000 × g to separate plasma and blood cells. Aliquots of plasma were removed for cortisol analysis; brain and liver samples were also taken for measurement of lipid peroxidation and fatty acid profiles. All samples were immediately frozen on dry ice and stored at −80 ◦ C until assayed. Chemical analyses Polar lipids were extracted by homogenizing frozen tissue samples for 30 s in 10 parts ice cold methanol (wt vol−1 ) and 0.01% butylated hydroxytoluene (BHT; to prevent in vitro peroxidation). This
procedure isolated polar lipids (primarily phospholipid) with minimal contamination by neutral lipids (primarily triglyceride) as evaluated by thin-layer chromatographic separation (data not shown) using a dual solvent system of chloroform:methanol:acetic acid:H2 O (65:25:8:4) and hexane:diethyl ether (4:1; Turunen 1988). Extracted samples were assayed immediately for lipid peroxidation (LPO). The remainder of the four extracts for each tank were pooled, evaporated under N2 , and stored at −80 ◦ C for determination of fatty acid composition at a later date. Feed total lipids were extracted according to Folch et al. (1957), except that dichloromethane was substituted for chloroform (Chen et al. 1981). The ferrous oxidation-xylenol orange (FOX) method was used to quantify polar lipid peroxides (Shantha and Decker 1994; Burat and Bozkurt 1996; E.A. Decker, Dept. of Food Science, University of Massachusetts, Amherst, MA, pers. comm.), with slight modifications. Methanol rather than water was used as the assay solvent system, and lipid peroxidation was expressed as hydrogen peroxide equivalents (per g of polar lipid) by replacing the Fe3+ calibration curve with one utilizing hydrogen peroxide. After incubation of polar lipid extracts with xylenol orange and Fe3+ for 5 min, 200 µl of the reaction mixture was pipetted into the well of a 96-well flat-bottomed microplate, and the absorption at 562 nm was read and recorded (Kinetic Microplate Reader EL312E, BioTek Instruments, Winooski, VT). The lipid content of tissue polar lipid extracts was determined by the sulfo-phospho-vanillin reaction (Frings et al. 1972; Van Handel 1985). Lipid concentrations in the extracts were estimated from a menhaden fish oil (Sigma, Inc., St. Louis, MO) calibration curve. The cortisol content of fish plasma was determined with a cortisol ELISA kit (Oxford Biomedical Research, Inc., Oxford, MI). Analysis of fatty acid methyl esters for feeds and tissues was performed on a gas-liquid chromatograph (Hewlett Packard 6890 Series with auto injector) fitted with a flame ionization detector. Fatty acid profiles were determined by split injection (20:1) onto a CP-Sil 88 fused silica capillary column (100 m × 0.25 mm, Chrompack, Raritan, NJ) using a programmed temperature gradient method. The hydrogen carrier gas pressure was constant at 1.85 kg cm−2 , and the injector and detector temperatures were 255 ◦ C. Initial oven temperature was 70◦ C. Following injection of sample, oven temperature was increased at 1 ◦ C min−1 to 185 ◦ C and held for 20 min. Oven temperature was then increased at 3 ◦ C min−1 to 215 ◦ C, fol-
228 lowed by an increase at 10 ◦ C min−1 to 240 ◦ C for 5 min, after which oven temperature was returned to 70 ◦ C. Individual fatty acids were identified by comparison of retention times to those of pure standards (butter, Matreya Inc., Pleasant Gap, PA; menhaden oil CAS 8002-50-4 and Supelco 37 FAME mixture No. 407885; Supelco Inc., Bellefonte, PA). A response correction factor for each fatty acid methyl ester was used to convert peak area percentage to weight percentage. Correction factors were determined by analyzing butter oil of known fatty acid profile with certified values (CRM 164; European Community Bureau of Reference, Brussels, Belgium). Fatty acid profiles were provided by the laboratory of Dr. Mark McGuire, Dept. of Animal and Veterinary Science, University of Idaho, Moscow, Idaho. Statistical analyses The effect of experimental treatments (diet or stress) on dependent variables was tested with a univariate general linear model (GLM) approach to analysis of variance (ANOVA). Evaluation of statistical significance was based on ANOVA F-test results and associated R2 of the GLM. Because all statistical models had the same number of factors and the goal was not to fit the models, R2 was used as a measure of effect size. Cohen’s criteria (Cohen 1988) for assessing the magnitude of R2 values (small = 0.01; medium = 0.09; large = 0.25; very large = 0.64) were adopted. Growth and FER were analyzed by a one-way ANOVA. Brain and liver lipid peroxidation, tissue fatty acid composition, and plasma cortisol were evaluated with a two-way ANOVA. An independent samples ttest (two-tailed) was used to examine the change in tissue LPO and plasma cortisol concentrations after exposure to low-water stress (control vs. stress) within dietary treatments. If the effect of an experimental treatment was deemed significant according to the criteria outlined above, differences between individual treatment means were determined by examining their contrasts with the associated confidence interval (Bonferroni correction for multiple comparisons: α/c, where c = number of comparisons or groups and α=0.05). This same procedure was used for a priori contrasts of tissue LPO and plasma cortisol means between diets in stressed fish. A significance level of α=0.05 was used for all statistical tests. In some cases, statistical significance at α=0.10 is also noted. Measured dependent
variables from individual fish were averaged for each tank (experimental unit). Results Fatty acid composition Diet had a significant effect on tissue concentrations of n-6 and n-3 fatty acids (Table 2). Generally, the fatty acid composition of tissues reflected that of diet, but liver fatty acids more closely resembled dietary fatty acids than did brain fatty acids (Table 3). Low-water stress did not significantly affect (P>0.05) tissue fatty acid concentrations. Levels of LN and LA in brain and liver reflected concentrations in diet. Tissue concentrations of LN were highest in fish fed linseed oil (LSO) and lowest in those fed soybean oil (SBO); the reverse was true of LA (Table 3). Compared to the SBO and LSO diets, MIX (dietary oil a mixture of 55% LSO and 45% SBO) produced intermediate levels of LA and LN in tissues. Levels of both fatty acids were extremely low in tissues of fish fed the menhaden oil (MHO) diet. Concentrations of LN and LA in brain were noticeably lower than in liver. All betweendiet comparisons for LA and also LN in tissue were significantly different. Concentrations of 20:3n-6, 20:4n-6 (1- and 2series prostaglandin fatty acid precursors), and 20:5n3 (3-series prostaglandin fatty acid precursor) in liver were generally reflective of diet and greater in liver than brain (Table 3). Levels of AA and dGLA in liver were greatest for SBO followed by MIX, LSO, and MHO diets (Table 3). The greatest level of AA was in the MHO diet, but fish fed SBO had significantly greater concentrations in both brain and liver compared to other diets. Similar to AA, liver dGLA was significantly greater in fish fed SBO; however, levels were low in liver and undetectable in brain except in fish fed the SBO diet. Eicosapentanoic acid concentrations in liver were greatest in fish fed MHO, followed by LSO, MIX, and SBO (Table 3). All between-diet mean contrasts for liver EPA were significantly different, except for the MIX and LSO comparison. The AA to EPA ratio was greater in liver and brain of fish fed SBO compared to fish fed other diets. The ratio in liver was approximately twice that in brain. Although the MHO diet had much greater long-chain n-3 PUFA (LCO3; sum of 20:5n-3, 22:5n-3, and 22:6n-3) than the other diets, levels of PUFA in tissue did not generally reflect the magnitude of this difference, except that liver
229 Table 2. General linear model analysis of variance for fatty acid content (mean weight%) of tissues Parameter
Liver 18:3n-3 20:5n-3 22:6n-3 Long-chain n-31 18:2n-6 20:4n-6 20:3n-6 AA:EPA Brain 18:3n-3 20:5n-3 22:6n-3 Long-chain n-31 18:2n-6 20:4n-6 20:3n-6 AA:EPA
Diet
Stress
Interaction
R2
F
P
R2
F
P
R2
F
P
0.96 0.90 0.61 0.67
132 48.0 8.6 11.0
