Fibrate derivatives are widely used as normolipidaemic or hypolipidaemic drugs for the treatment of hypertriacylglycerolaemic and/or hypercholesterolaemic patients(Reference Forcheron, Cachefo, Thevenon, Pinteur and Beylot1–Reference Liamis, Kakafika, Bairaktari, Miltiadous, Tsimihodimos, Goudevenos, Achimastos and Elisaf3). As a lipid-lowering agent, fenofibrate has been shown to lower the concentration of plasma TAG and cholesterol(Reference Klosiewicz-Latoszek and Szostak4, Reference Vakkilainen, Steiner, Ansquer, Perttunen-Nio and Taskinen5). Some mechanisms of clinical effects of fenofibrate have been identified, including interference with fatty acid (FA) synthesis, stimulation of hepatic FA oxidation, increase in lipoprotein lipolysis, inhibition of cholesterol biosynthesis, induction of hepatic uptake of cholesterol from plasma and increased elimination of cholesterol into bile as biliary acids(Reference Guo, Wang, Milot, Ippolito, Hernandez, Burton, Wright and Chao6). Fenofibrate is also known to activate the transcriptional factor PPARα, which binds to the peroxisome proliferator response element in the regulatory region of target genes(Reference Fruchart, Staels and Duriez7, Reference Marx, Duez, Fruchart and Staels8). By this way, PPARα is known to induce enzymes involved in the regulation of lipid metabolism, particularly those related to mitochondrial and peroxisomal FA oxidation(Reference Berthou, Duverger and Emmanuel9–Reference Minnich, Tian, Byan and Bilder12). It has also been suggested that fenofibrate regulates body weight (BW) and energy homeostasis by increasing FA oxidation(Reference Mancini, Lanni, Sabatino, Moreno, Giannino, Contaldo, Colantuoni and Goglia13).
Similarly to fibrate derivatives, fasting also activates PPARα and thereby regulates mitochondrial and peroxisomal FA oxidation activities(Reference Kersten, Seydoux, Peters, Gonzalez, Desvergne and Wahli14, Reference Casas, Pineau, Rochard, Rodier, Daury, Dauca, Cabello and Wrutniak-Cabello15). The regulation of these pathways affects the lipid metabolism and was effectively demonstrated to modulate blood lipid composition in both animal models and humans(Reference Ohama, Matsuki, Saito, Tsukamoto, Kinoshita, Katsuragawa, Okazaki, Yamanaka and Teramoto16–Reference Kochan, Goyke, Karbowska, Slominska and Swierczynski18). The observations regarding blood lipid alteration during fasting are conflicting(Reference Sävendahl and Underwood17–Reference Garcia-Fuentes, Gil-Villarino, Zafra and Garcia-Peregrin21), but hypolipidaemic effects have been reported(Reference Kochan, Goyke, Karbowska, Slominska and Swierczynski18, Reference Andriamampandry, Bnouham, Michard, Gutbier, Le Maho and Leray22).
The hypolipidaemic effects of fenofibrate and fasting have been shown in mammals, but little information is available in fish. Previous studies indicated that, in rainbow trout fed a 22 % fat diet, the level of peroxisomal FA oxidation was increased by fenofibrate with a concomitant decrease in body EPA and DHA(Reference Du, Demizieux, Degrace, Gresti, Moindrot, Liu, Tian, Cao and Clouet23). Contrasting with the carnivorous rainbow trout, grass carp (Ctenopharyngodon idella) is a typical herbivorous finfish. Recent studies showed that the energy requirement of grass carp was relatively lower than that of most carnivorous fish species. High-fat (HF) diets (above 6 % fat), especially with high levels of PUFA, resulted in lowered feed intake, decreased growth, accumulation of lipids in the liver and other organs, alteration of lipoprotein synthesis, high plasma TAG and cholesterol, and elevated LDL-cholesterol (LDL-C)(Reference Du, Liu, Tian, Wang, Wang and Liang24–Reference Du, Clouet, Huang, Degrace, Zheng, Tian and Liu26). Under these conditions, it would be interesting to know whether grass carp possesses sufficient potential capacities to utilise excess dietary lipids. The aim of the present study was therefore to increase the lipid catabolism in grass carp fed an HF diet by using fenofibrate as a potent ligand of PPARα and fasting that represents a usual physiological condition. Further, since fenofibrate is used to treat hyperlipidaemic patients, it could also be capable of curing fish with hyperlipidaemia symptoms. Since fasting is shown to regulate several genes involved in lipid metabolism in a similar manner as fenofibrate in mammals(Reference Casas, Pineau, Rochard, Rodier, Daury, Dauca, Cabello and Wrutniak-Cabello15), comparable effects could be expected in our herbivorous fish model.
In the present study, to evaluate the possible hypolipidaemic effect of fenofibrate and fasting in grass carp, juvenile grass carp were fed an HF diet for 6 weeks and then treated with fenofibrate for 2 weeks, followed by 1 week of fasting to the end of the study. Nutritional and biochemical measurements including body composition, blood lipids, peroxidation products and hepatic FA mitochondrial and peroxisomal β-oxidation were performed to evaluate the extent of biochemical alterations induced by fenofibrate and fasting.
Materials and methods
Fish and diets
Juvenile grass carp, initially weighing (7·83 (se 0·11) g, were distributed into six glass tanks with recirculation (300 litres; thirty fish per tank) maintained at 28°C with a 12 h light–dark cycle. Fish were acclimatised to experimental conditions for 2 weeks during which fish were fed the lipid-free experimental diet (gross energy; 9673 kJ/kg) at 1 % BW/d. An HF diet, whose composition is detailed in Table 1, was prepared as previously described(Reference Du, Liu, Tian, Wang, Wang and Liang24) and freeze dried for further use as the HF diet. Fenofibrate, provided by Professor P. Clouet (UMR 866, Dijon, France), was added to the HF diet (0·34 % DM) as indicated in Table 1. HF and HF+fenofibrate diets were transformed into 1·5 mm pellets (about 10·5 % moisture). All fish were fed the HF diet for 6 weeks (weeks 1–6) at a feeding rate of 3 % BW/d to induce body fat accumulation. After this period, the average fish weight in the two groups was 12·1 (se 0·2) g and did not significantly differ between tanks. Fish of one group were then fed the HF diet containing 0·34 % fenofibrate for the next 2 weeks (weeks 7 and 8) at the same feeding rate, which provided 100 mg fenofibrate/kg BW per d (the usual dose active in rats). During the feeding trial, fish were fed at 09.00 and 18.00 hours with an equal portion of diet, and were weighed once per week. At the end of the 8-week feeding trial, the water in the tanks was totally renewed and the fish of both groups were fasted for 1 week (week 9). The detailed trial protocol is shown in Fig. 1. During the whole experimental period, faeces were siphoned from tanks using a rubber pipe every morning, while dissolved O2, pH and ammonia were maintained at 7·74 (se 0·61) mg/l, 7·21 (se 0·20) and 0·15 (se 0·05) mg/l, respectively.
HF, high-fat.
* The plant oil mixture contained mainly maize oil and linseed oil.
† In this commercial product, butylated hydroxytoluene was added at 1 g/kg to avoid lipid peroxidation.
‡ The vitamin mix (Evergreen Feed Co., Guangzhou, China) contained (mg/kg diet): thiamin, 50; riboflavin, 50; vitamin A, 9; vitamin E, 400; vitamin D3, 6; menadione, 40; pyridoxine HCl, 40; cyanocobalamin, 0·1; biotin, 6; calcium pantothenate, 100; folic acid, 15; niacin, 200; inositol, 2000. Cellulose was used as a carrier.
§ The mineral mix (Evergreen Feed Co., Guangzhou, China) contained (mg/kg diet): calcium biphosphate, 9·8; calcium lactate, 37·9; sodium chloride, 2·6; potassium sulfate, 13·1; potassium chloride, 5·3; ferrous sulfate, 0·9; ferric citrate, 3·1; magnesium sulfate, 3·5; zinc sulfate, 0·04; manganese sulfate, 0·03; cupric sulfate, 0·02; cobalt chloride, 0·03; potassium iodide, 0·002; cellulose, 42.
∥ Cellulose excluded.
Sampling procedure and chemical analysis of organs
After the first 6 weeks of the feeding trial, four fish were randomly captured from each tank, fasted for 24 h and regrouped for whole-body and blood chemical analysis (four fish from six tanks, totalling twenty-four fish). After fenofibrate treatment and the following fasting for 1 week, six fish were randomly captured from each tank and regrouped for whole-body and blood chemical analysis (six fish from three tanks per treatment, totalling eighteen fish per treatment). Another six fish randomly captured from each tank were killed by spinal destruction and used for individual measurements of BW and length, and weights of viscera, liver, mesenteric fat tissue and white muscle (eighteen fish per treatment). Pieces of liver, mesenteric fat tissue and white muscle from both sides of the spine (fillets) without skin were stored at − 20°C until analysis. Total N and carbohydrate of the diets and body samples were measured by the Kjeldahl method using an auto-analyser (Tecator Kjelte model 1030; Tecator AB, Höganäs, Sweden) and the 3′5-dinitrosalicylic acid method(Reference Yu, Olsen and Marcussen27), respectively.
Lipid analysis by GLC
Total lipids of the diet, muscle, liver and mesenteric fat tissue were extracted according to the method of Bligh & Dyer(Reference Bligh and Dyer28). FA from lipid extracts were methylated with 10 % potassium hydroxide in methanol for 1 h at room temperature. FA methyl esters were then analysed and quantified using a Hewlett-Packard HP-5890 gas chromatograph in a cross-linked 5 % phenylmethyl silicone gum phase column (length 25 m, internal diameter 0·32 mm, film thickness 0·25 μm; HP-Ultra 2, with N2 as the carrier gas), equipped with flame ionisation detection. The injector and detector temperatures were 280 and 300°C, respectively. The column temperature set at 190°C was then increased to 260°C (2°C/min) and held for 5 min. Each FA was identified using a mixture of standard FA methyl esters (Cayman Co., Ann Arbor, MI, USA). Results were expressed as the percentage of each FA with respect to total FA.
Plasma analysis
Blood collected from the heart with a heparinised microsyringe was immediately centrifuged. Plasma samples were frozen at − 30°C until analysis. Plasma TAG, total cholesterol, HDL-cholesterol (HDL-C), LDL-C and apo A-I were assayed by enzymic procedures using an automatic biochemical analyser (Hitachi 7170, Tokyo, Japan) and attached kits (Daiichi Pure Chemicals Co., Ltd, Tokyo, Japan). TAG and cholesterol were assessed by the glucose oxidase-peroxidase (GOD-PAP) method(Reference Koditschek and Umbreit29) and the cholesterol oxidase-peroxidase (CHOD-PAP) method(Reference Richmond30), respectively. HDL-C and LDL-C were assessed by enzymic methods(Reference Nakamura, Taniguti, Yamamoto, Hino and Manabe31, Reference Arranz-Pena, Tasende-Mata and Martin-Gil32), and apo A-I by immunoassay(Reference Steinberg, Cooper, Graiser and Rosseneu33). Thiobarbituric acid-reactive substances (TBARS) were quantified as described by Rueda-Jasso et al. (Reference Rueda-Jasso, Conceicao, Dias, De Coen, Gomes, Rees, Soares, Dinis and Sorgeloos34) using the malondialdehyde kit (Jiancheng Biotech Co., Nanjing, China).
Histology
Livers from three fish were sampled from each tank and prepared for histological examination. Pieces of liver were fixed in 10 % formaldehyde and embedded in paraffin. Sections of 5 μm thickness were stained with the Harris haematoxylin–eosin mixture and examined under a light microscope(Reference Bancroft and Stevens35).
Liver homogenates and mitochondrial fraction preparations
After treatments of fenofibrate and fasting, pieces of liver (about 3 g) collected from each group were cut finely in ice-cold 0·25 m-sucrose medium containing 1 mm-EGTA and 10 mm-2-amino-2-hydroxymethyl-propane-1,3-diol-HCl, pH 7·4, rinsed five times in the same medium, blotted with absorbent paper and weighed. The tissue was diluted (1:20, w/v) in the chilled sucrose medium and homogenised by only four strokes of a Teflon pestle rotating at 300 rpm in a Potter-Elvehjem homogeniser (Wheaton Science Products, Millville, NJ, USA). The 1 ml samples of homogenate were kept apart for the immediate measurement of mitochondrial and peroxisomal palmitate oxidation levels (see below) and the delayed measurement of marker enzyme activities. The remaining homogenate was 2-fold diluted with sucrose medium containing 2 % FA-free bovine serum albumin and was centrifuged at 2500 g for 4 min at 4°C. The supernatant fraction was immediately centrifuged at 16 000 g for 6 min and the pellet obtained was re-suspended with the sucrose medium without bovine serum albumin and re-sedimented using the preceding conditions. The procedure was repeated once and the pellet re-suspended in 1 ml of buffered 0·3 m-sucrose was used as the mitochondrial fraction. The protein content of this fraction was roughly estimated by a rapid spectrophotometric method(Reference Clouet, Niot and Bézard36), which allowed the immediate measurement of the activity levels of palmitate oxidation and carnitine palmitoyltransferase I (CPT I). These measurements were corrected later after more accurate determination of the protein content with the bicinchoninic acid method(Reference Smith, Krohn, Hermanson, Mallia, Gartner, Provenzano, Fujimoto, Goeke, Olson and Klenk37).
Mitochondrial and peroxisomal enzyme assays
The presence of mitochondria was assessed by the activity of monoamine oxidase(Reference Weissbach, Smith, Daly, Witkop and Udenfriend38). The mitochondrial protein content per g liver was calculated by dividing the activity of monoamine oxidase per g liver (measured in total homogenate) by the activity expressed per mg mitochondrial protein (measured in the mitochondrial fraction). The presence of peroxisomes was assessed by the activity of catalase(Reference Aebi and Bergmeyer39), and of CN− -insensitive palmitoyl-CoA-dependent NAD+ reduction, a short sequence of enzymes acting at the beginning of the β-oxidation cycle, described as the peroxisomal FA-oxidising system(Reference Bronfman, Inestrosa and Leighton40). CPT I activity measurement was carried out using 3H-labelled l-carnitine and palmitoyl-CoA(Reference Bremer41), and extraction of the palmitoyl-[3H]carnitine produced was performed using butan-1-ol. The associated radioactivity was counted in the presence of a convenient (1:6, v/v) scintillation cocktail(Reference Chen, Li, Liu, Qiu and Lin42) in a liquid scintillation spectrometer (LS3500; Beckman, Fullerton, CA, USA).
Liver mitochondrial and peroxisomal palmitate oxidation
From liver homogenates
Homogenates originated from pieces of liver homogenised in 20 vol. sucrose medium as indicated above and were used as soon as possible. Palmitate oxidation rates were measured at 37°C using two media as already described(Reference Veerkamp, Van Moerkerk, Glatz and Van Hinsbergh43), the first allowing the mitochondrial and peroxisomal activities to occur, the second allowing the peroxisomal activity only. After 30 min, the radioactivity initially held by [1-14C]palmitate was recovered on labelled short molecules released from the β-oxidative cycle and soluble in perchloric acid (acid-soluble products) using 0·45 membrane filters (Millipore, Billerica, MA, USA). The radioactivity of the acid-soluble products was measured as above for that of butan-1-ol extracts.
From liver mitochondria
Isolated mitochondria were used to assess whether activity rates related to mitochondria previously measured in liver homogenates were not altered by the presence of extra-mitochondrial compounds and/or enzyme reactions. The medium contained [1-14C]palmitate bound to bovine serum albumin in a 1·5:1 molar ratio, and the incubation was stopped after 8 min with perchloric acid. The radioactivity of the acid-soluble products was measured as the method indicated above for liver homogenates.
Statistical analysis
Results are expressed as mean values with their standard errors. Significant differences (P < 0·05) of each variable were first detected using the one-way ANOVA test, and then the Duncan's multiple-range test was used to rank the four experimental groups. All analyses were made using the SPSS 9.0 software (SPSS Inc., Chicago, IL, USA).
Results
Body parameters, total tissue lipid and hepatic histology
In grass carp fed the HF diet for 6 weeks, there was no significant difference in mean values between tanks for all the parameters tested (Tables 2 and 3). The whole-body fat mass was nearly 60 % greater than in fish fed a fat-free diet or a low-fat diet as previously shown(Reference Du, Liu, Tian, Wang, Wang and Liang24–Reference Du, Clouet, Huang, Degrace, Zheng, Tian and Liu26, Reference Du, Liu, Tian, Wang, Wang, Guo and Liang44), and was associated with hyperlipidaemia symptoms. Fenofibrate administered during the last 2 weeks of the 8-week HF-diet feeding period did not affect the weight of the whole body, viscera, white muscle and liver, but significantly decreased the weight of mesenteric fat tissue compared with fish fed the HF diet for 8 weeks (Table 2). The final fasting for 1 week applied to both groups significantly decreased whole-body, liver and mesenteric fat tissue weights, but there was still a significant difference in the relative mass of mesenteric fat tissue between the fenofibrate and HF-diet groups. Table 3 shows that the fenofibrate treatment significantly reduced the lipid content of the whole body and liver, relative to the HF-diet group. The final week of fasting resulted in a decrease in lipids of the whole body, white muscle and mesenteric fat tissue, but significant differences were only found in whole-body lipid contents for each group (without or with fenofibrate), and in white muscle lipid contents only for the fenofibrate group. The significant differences in whole-body lipid contents existing between the HF-diet and fenofibrate groups still persisted after fasting. Compared with the lipid content of whole-body, muscle and mesenteric fat tissue, liver lipid content conversely increased significantly after fasting for 1 week, but the significant difference that existed between the two groups was not observed after fasting (Table 3). The histological study of liver showed that, in fish fed the HF diet (Fig. 2 (a)), hepatocytes were swollen with numerous clear round areas of various sizes (corresponding to lipid droplets whose initial content was eliminated during the staining procedure) occupying most space and pushing nuclei aside to cell periphery. In the fenofibrate group (Fig. 2 (b)), steatotic hepatocytes with varying size lipid droplets were also found, but were less numerous than in the liver of the HF-diet group. After 1 week of fasting, the number of swollen hepatocytes appeared to increase in both dietary groups (Fig. 2 (c) and (d)), and there was no apparent visible difference between hepatocytes of the HF-diet and fenofibrate groups.
a,b,c Mean values (excluding those for the HF-diet group weeks 1–6) within a line with unlike superscript letters were significantly different (P < 0·05).
* Condition factor = fish weight (g) × 100/body length3 (cm).
† White muscle index = white muscle weight × 100/fish weight.
‡ Viscera index = viscera weight × 100/fish weight.
§ Hepatosomatic index = liver weight × 100/fish weight.
∥ Mesenteric fat index = mesenteric fat tissue weight × 100/fish weight.
MDA, malondialdehyde.
a,b,c,d Mean values (excluding those for the HF-diet group weeks 1–6) within a line with unlike superscript letters were significantly different (P < 0·05).
Plasma lipids and thiobarbituric acid-reactive substances
As shown in Table 3, except for HDL-C, the plasma contents of TAG, cholesterol, LDL-C and apo A-I were all significantly lowered after fenofibrate treatment. After the final fasting, there was no more difference in plasma lipid indexes between fish previously treated with or not with fenofibrate, except for plasma TAG that were still more reduced in fish previously treated by fenofibrate. Fenofibrate treatment significantly increased the plasma TBARS concentration, but this increase disappeared after the fasting period.
Fatty acid composition in diet and tissues
The dietary lipid source was a mixture of maize, linseed and fish oils, and was characterised by its high contents of oleic acid (17·7 %), linoleic acid (25·6 %), EPA (4·97 %) and DHA (24·7 %). The dietary FA distribution is given for reference in Tables 4–6, which show the FA composition of white muscle, liver and mesenteric fat tissue, respectively. The FA composition was altered by fenofibrate and fasting only in white muscle and liver, but not in mesenteric fat tissue. In the former organs, the contents of SFA were unchanged after fenofibrate treatment, but slightly decreased after fasting. In the same organs, the contents of oleic acid and total MUFA were greater in the fenofibrate group than in the HF-diet group, but after fasting, both groups exhibited a significant decrease in MUFA contents, with the level in the fenofibrate group still greater than in the HF-diet group. Conversely, the contents of total PUFA, n-3 PUFA and EPA+DHA decreased significantly after fenofibrate treatment, but increased after fasting in both groups, with lower values in the fenofibrate group than in the HF-diet group. When the amounts of EPA and DHA were expressed in mg per whole organ for white muscle, liver and mesenteric fat tissue, the total amount of EPA and DHA in the three organs after fenofibrate treatment was 31·7 % less than for the HF-diet group. A similar drop of 34·7 % was also found in both groups after fasting. Furthermore, the amounts of EPA and DHA after fasting in the organs of fish fed the HF diet were 30·3 % less than before fasting, and a comparable drop of 30 % was also found between before and after fasting in the fenofibrate group. However, despite the comparable total loss of EPA+DHA (about 30 %) after fasting in the three organs of each fish group, it was apparent that the loss of EPA+DHA differed between the organs. Indeed in white muscle, fenofibrate treatment caused a loss of 18 mg EPA+DHA by reference to the HF diet, but after 1 week of fasting, the organ lost only 2·6 mg. In the liver, the loss of EPA+DHA due to fenofibrate and fasting was 4 or 3 mg, respectively. By contrast in mesenteric fat tissue, the loss of EPA+DHA due to fenofibrate was 29 mg in continuously fed fish, while the 1 week of fasting allowed the fenofibrate group to lose 24 mg EPA+DHA. Therefore EPA+DHA would be more lowered by fenofibrate than fasting in white muscle, while fasting would be relatively more effective in mesenteric fat tissue according to the decrease in n-3 PUFA which partly resulted from the significant decrease of total fat (Tables 2 and 3). EPA and DHA of liver lipids were maintained at low levels, despite their high content in the dietary lipids, and had similar sensitivity to fenofibrate and fasting.
a,b,c,d Mean values (excluding the dietary lipids values) within a line with unlike superscript letters were significantly different (P < 0·05).
* ΣMUFA or EPA+DHA amount per whole organ = EPA+DHA or ΣMUFA composition (in weight % of total lipids) × total lipid content (in weight % of organ) × organ weight (g).
a,b,c,d Mean values (excluding the dietary lipids values) within a line with unlike superscript letters were significantly different (P < 0·05).
* ΣMUFA or EPA+DHA amount per whole organ = EPA+DHA or ΣMUFA composition (in weight % of total lipids) × total lipid content (in weight % of organ) × organ weight (g).
a,b,c,d Mean values (excluding the dietary lipids values) within a line with unlike superscript letters were significantly different (P < 0·05).
* ΣMUFA or EPA+DHA amount per whole organ = EPA+DHA or ΣMUFA composition (in weight % of total lipids) × total lipid content (in weight % of organ) × organ weight (g).
Parameters related to fatty acid oxidation
The content in mitochondrial protein per g liver was calculated from monoamine oxidase activity expressed per g liver and divided by the activity of 1 mg mitochondrial protein. Its value was 57·6 % more elevated in the fenofibrate group than in the HF-diet group (Table 7). After 1 week of fasting, the mitochondrial protein content of the liver significantly increased in the HF-diet group, but only marginally in the fenofibrate group, so that fenofibrate treatment and fasting appeared equally capable of inducing the similar increase in liver mitochondrial protein per g liver. Among parameters related to mitochondrial FA oxidation, carnitine-dependent palmitate oxidation rates expressed per g liver were significantly greater in the fenofibrate group than in the HF-diet group only when calculated from results obtained from crude liver homogenates (Table 7). Indeed, when activity rates were calculated from mitochondrial fractions (excluding the other parts of cells), there was no more difference between the HF-diet and fenofibrate groups. After fasting, activity rates of carnitine-dependent palmitate oxidation in both groups calculated from homogenates and isolated mitochondria were significantly increased, but with lower values in the fenofibrate group than in the HF-diet group. The activity rate of CPT I, a key enzyme of the mitochondrial FA oxidation pathway, when expressed per g liver, was slightly greater in the fenofibrate group than in the HF-diet group. This activity significantly increased during the fasting period in both groups, but with relatively less extent in the fenofibrate group.
a,b,c Mean values within a line with unlike superscript letters were significantly different (P < 0·05).
* Calculated from activities in mitochondrial fractions (nmol/min per mg protein) multiplied by corresponding mitochondrial protein contents per g liver.
As regards catalase, a peroxisomal marker enzyme, its activity expressed per g liver was significantly greater in the fenofibrate group than in the HF-diet group. However, after fasting, the peroxisomal marker activity only exhibited slightly higher values in the HF-diet group, and the difference between the two groups before fasting disappeared (Table 7). Peroxisomal FA oxidation does not require carnitine, and its activity rates measured with palmitate were significantly more elevated in the fenofibrate group than in the HF-diet group. After fasting, these activity rates were increased in both groups, but with comparable levels (Table 7). The activity rate of the peroxisomal FA-oxidising system, which is a key sequence of the whole peroxisomal FA oxidation pathway, was nearly 20 % increased after fenofibrate treatment. After fasting, this activity was significantly lowered in both groups, but still with a slightly greater value in the fenofibrate group.
Discussion
Hyperlipidaemia and herbivorous grass carp
Chronic hyperlipidaemia represents a major risk of CVD in humans. It may arise from inborn defects, but increasingly results from unbalanced dietary habits with excess fat-rich food consumption(Reference Ellen and McPherson45, Reference Despres46). Hyperlipidaemia is also easily induced in animal models fed HF diets(Reference Xue, Kageyama, Kashiba, Kobayashi, Osaka, Namba, Kimura and Inoue47, Reference Roberts, Barnard, Liang and Vaziri48). This blood symptom may regress in humans and animals via fibrate administration or dietary regimens often based on the reduction of food intake. Comparing these treatments is difficult because of side effects met with fibrates in rodents through apparent liver peroxisomal proliferation, which does not occur in humans, and with partial fasting in humans often associated with sudden contrary reactions. Carnivorous fish fed HF diets were described to exhibit hyperlipidaemia-associated alterations similar to those reported in hum-ans(Reference Kennish, Sharpdahl, Chambers, Thrower and Rice49–Reference Du, Liu, Zheng, Tian and Liang52). The advantage of using the herbivorous grass carp as a model is that this fish is able to support relatively long times of fasting and to be sensitive to dietary lipids(Reference Du, Liu, Tian, Wang, Wang and Liang24–Reference Du, Clouet, Huang, Degrace, Zheng, Tian and Liu26). Previous studies showed that, in normally fed grass carp, plasma TAG, cholesterol, HDL-C and LDL-C were nearly 1·5, 3, 0·4 and 0·3 mmol/l, respectively, but in the present study with fish fed an HF diet, the corresponding values were 3·5, 6·7, 1·2 and 1·1 mmol/l, respectively. Consequently, grass carp is capable of developing hyperlipidaemia, and its biological reactions to fibrates (here used as fenofibrate) and to fasting are discussed below.
Hypolipidaemic effects of fenofibrate and fasting
The fenofibrate-induced reduction of plasma TAG and cholesterol has been suggested to maximally result from alteration of liver lipoprotein synthesis and secretion, inhibition of cholesterol synthesis and increased FA oxidation via PPAR in rodents and hyperlipidaemic patients(Reference Guo, Wang, Milot, Ippolito, Hernandez, Burton, Wright and Chao6–Reference Minnich, Tian, Byan and Bilder12). The present study shows that the hypolipidaemic effects of fenofibrate also occur in grass carp. We observed that plasma TAG, cholesterol and LDL-C decreased after fenofibrate treatment, but also with a decrease in HDL-C and apo A-I. Some studies have shown that low concentrations of HDL-C predict an increased CHD risk at any LDL-C level(Reference Castelli53, Reference Assmann, Schulte, von Eckardstein and Huang54), but fenofibrate could produce a significant increase in HDL-C in patients(Reference Despres46). Contrary to humans, fibrate-treated rats, mice and hamsters exhibited decreased plasma HDL-C and liver apo A-I mRNA levels(Reference Guo, Wang, Milot, Ippolito, Hernandez, Burton, Wright and Chao6, Reference Berthou, Duverger and Emmanuel9, Reference Staels, van Tol, Andreu and Auwerx55). The divergent effects of fenofibrate in rodents and fish v. humans suggest that the sensitivity of PPAR to fibrate and the sequence of the involved peroxisome-proliferator responsive element of apo A-I markedly differ in rodents and fish compared with humans. Indeed, the promoters of the apo A-I gene have been shown to possess distinct sequences in rodents and humans(Reference Vu-Dac, Chopin-Delannoy, Gervois, Bonnelye, Martin, Fruchart, Laudet and Staels56). Compared with fenofibrate, fasting resulted in the reduction of plasma TAG, but not of cholesterol. A similar decrease in plasma TAG during starvation was also reported in some carnivorous fish(Reference Chatzifotis and Takeuchi57, Reference Qian, Cheng and Sun58), but this decrease was associated with an increase in cholesterol. In humans and animal models, similar conflicting data regarding the effects of fasting on plasma TAG and cholesterol have also been found(Reference Sävendahl and Underwood17–Reference Garcia-Fuentes, Gil-Villarino, Zafra and Garcia-Peregrin21). These inconsistencies may originate from sex, age, obesity, hyperlipidaemia, health state, medication, diets and/or physical activity during fasting(Reference Sävendahl and Underwood17). Interestingly, semi-starvation of 6 d did not affect serum TAG and cholesterol in non-obese men and women, but lowered TAG and did not change cholesterol in obese individuals(Reference Vaisman, Sklan and Dayan19). This latter situation was very similar to the results of grass carp in the present study whose excess body fat was due to an HF feeding.
Lipid deposition and mobilisation
Grass carp fed HF diets were previously shown to easily accumulate body fats, especially when ingested lipids were rich in PUFA which exerted deleterious effects(Reference Du, Liu, Tian, Wang, Wang and Liang24–Reference Du, Clouet, Huang, Degrace, Zheng, Tian and Liu26). In the present study, the HF (8 %) diet which also contained high PUFA (62·8 % total FA) was capable of markedly increasing body lipids. In liver cells, fat accumulation was developed with the numerous large lipid droplets filling up nearly all the cytoplasmic volume. Total hepatic lipid content exceeded 20 % in the present study, but normally it is maintained below 15 % in the liver. Further, mesenteric fat tissue that normally represents less than 2 % of BW, reached 3·28 % in the fish fed the HF diet. With the grass carp model, fenofibrate treatment succeeded in significantly reducing whole-body and liver lipids, as well as mesenteric fat tissue mass. Such a reduction should imply an increased use of fat to meet energetic requirements. These requirements are metabolically low in mesenteric fat tissue, compared with those in the liver and muscles; the loss of weight of fat tissues with fenofibrate therefore suggests that the large amounts of NEFA released from the fat stores were used in energy-demanding organs. The liver appeared to be particularly sensitive to fenofibrate because of its capacity to reduce size and number of fat droplets accumulated within hepatocytes, and also to deal with NEFA of peripheral origin. This efficiency has also been recovered in diet-induced obese rats(Reference Mancini, Lanni, Sabatino, Moreno, Giannino, Contaldo, Colantuoni and Goglia13) and in rainbow trout(Reference Du, Demizieux, Degrace, Gresti, Moindrot, Liu, Tian, Cao and Clouet23). Fenofibrate and other fibrates have been demonstrated to alter hepatic FA metabolism(Reference Yamamoto, Fukuda, Zhang and Sakai59) via induction of FA oxidation-related enzymes, such as mitochondrial CPT I, peroxisomal acyl-CoA oxidase and lipoprotein lipase, through PPARα activation. This suggests that lipid deposition in the grass carp model was counterbalanced by higher levels of FA oxidation through specific fibrate effects. After fasting, lipids of the whole body, mesenteric fat tissue and white muscle were decreased, but, surprisingly, lipids of the liver were increased as far as to trigger hepatic steatosis. These data differ from those reported in most fish, in which liver lipids tended to be used first(Reference Silva, Gunasekera and Austin60), but resemble those described in PPARα-null mice(Reference Kersten, Seydoux, Peters, Gonzalez, Desvergne and Wahli14), shrews (Suncus murinus)(Reference Ohama, Matsuki, Saito, Tsukamoto, Kinoshita, Katsuragawa, Okazaki, Yamanaka and Teramoto16) and rats(Reference Degrace, Demizieux, Du, Gresti, Caverot, Djaouti, Jourdan, Moindrot, Guilland, Hocquette and Clouet61) with liver lipid accumulation after fasting. Indeed ligands of PPARα as fibrates are known to up-regulate, as mentioned above, FA oxidation-related enzymes(Reference Auwerx, Schoonjans, Fruchart and Staels62), while those of PPARγ as thiazolidinediones result in stimulating the esterification levels in adipose tissue(Reference Escher, Braissant, Basu-Modak, Michalik, Wahli and Desvergne63). It was effectively shown that PPARγ-null rats were unable to preserve their fat stores(Reference Sharma and Staels64). In this context, the fact that PPARγ is down-regulated by fasting as well as in insulin-deficient diabetes(Reference Vidal-Puig, Jimenez-Linan, Lowell, Hamann, Hu, Spiegelman, Flier and Moller65) would explain in our fish model, associated with the need of energetic substrates, the marked loss of weight of mesenteric fat tissue observed after fasting. The absence of insulin secretion during this period should result in an intense peripheral lipolysis followed by a likely increased uptake of NEFA by liver cells. This would saturate the hepatic FA oxidation capacities, with concomitant esterification of excess FA into TAG, as clearly shown in Table 3. Previous studies have also shown that impaired hepatic FA oxidation may occur concurrently with excess esterification reactions, as developed by the hepatic steatosis setup(Reference Du, Clouet, Zheng, Degrace, Tian and Liu25). Further, food deprivation could be even still more deleterious in steatotic hepatocytes(Reference Grattagliano, Vendemiale, Caraceni, Domenicali, Nardo, Cavallari, Trevisani, Bernardi and Altomare66).
Fatty acid composition in tissues
The total amounts of EPA and DHA in the lipids of liver, white muscle and mesenteric fat tissue were 42, 23 and 40 %, respectively, less in the fenofibrate group than in the HF-diet group (Tables 4–6). This meant that fenofibrate was more effective at decreasing the contents of n-3 highly unsaturated FA in the liver and mesenteric fat tissue than in white muscle. Further, the composition of EPA+DHA (expressed as % of total FA) in the HF-diet v. fenofibrate groups in liver, muscle and fat tissue was 6·9 v. 4·95, 26·2 v. 20·9 and 16·1 v. 15·1, respectively. This implies that the natural mechanisms of lowing n-3 highly unsaturated FA were more active in the liver and fat tissue than in white muscle, and these organ difference did not alter by finofibrate treatment. Data obtained in rodents showed that the levels of FA oxidation are markedly greater in the liver than in muscle, and that DHA is nearly exclusively β-oxidised within peroxisomes, and EPA within both mitochondria and peroxisomes(Reference Willumsen, Hexeberg, Skorve, Lundquist and Berge67, Reference Willumsen, Vaagenes, Lie, Rustan and Berge68). Further, fenofibrate was demonstrated to up-regulate mitochondrial and peroxisomal FA oxidation activities via PPARα(Reference Berthou, Duverger and Emmanuel9–Reference Minnich, Tian, Byan and Bilder12). These data obtained in rodents appeared to meet ours in herbivorous fish because liver activities of grass carp were naturally capable of strongly reducing EPA+DHA, relative to white muscle, and fenofibrate treatment amplified this capability to reduce more n-3 highly unsaturated FA. These results also support those obtained in carnivorous rainbow trout(Reference Du, Demizieux, Degrace, Gresti, Moindrot, Liu, Tian, Cao and Clouet23). On the other hand, 1 week of fasting did not change the amounts of EPA+DHA in white muscles, but decreased those in the liver and mesenteric fat tissue by 30 and 58 %, respectively. Meanwhile the amounts of MUFA lost after fasting in the liver, white muscle and mesenteric fat tissue were 50, 54 and 56 %, respectively. These results meet the results reported in other fish, such as channel catfish(Reference Tidwell, Webster and Clark69) and red tilapia(Reference Silva, Gunasekera and Austin60). Further, the amounts of EPA+DHA lost by the liver+white muscle after fasting or fenofibrate treatment were 5·6 or 22 mg, respectively, while the amounts of MUFA lost under the same conditions were 71 and 4 mg, respectively. The inverse relationship of n-3 PUFA v. MUFA contents first between the liver and white muscle, and second between fasting and fenofibrate treatment may be explained through: (a) the FA oxidation specificity of peroxisomes mentioned above, compared with that of mitochondria in which MUFA from 16 to 18 carbons are among the best FA oxidation substrates(Reference Mannaerts, Debeer, Thomas and De Schepper70); (b) the relatively high mitochondria content of muscle cells compared with that of hepatocytes that also contain non-negligible amounts of peroxisomes(Reference Veerkamp and Van Moerkerk71, Reference Veerkamp and van Moerkerk72); (c) the much higher capacity of peroxisome proliferation in liver cells than in muscle cells after fibrate treatment(Reference Veerkamp and Van Moerkerk71, Reference Veerkamp and van Moerkerk72). Since fish compared with mammals have lower capacities to synthesise EPA and DHA from linolenic acid(Reference Sargent, Tocher, Bell, Halver and Hardy73), these n-3 PUFA were relatively preserved in tissues, more particularly in muscles. In fasting grass carp, the peroxisomal FA oxidation pathway should be weakly active. Inversely, the apparent decrease in MUFA after fasting also suggested that the mitochondrial FA oxidation pathway was significantly activated. These data prompted us to further investigate some FA oxidation-related activities in the liver.
Parameters related to hepatic fatty acid oxidation
Comparing lipid metabolism between mammals and fish in similar experimental designs, some differences have been observed. For example in mammals, fenofibrate has been shown to induce the enlargement of the liver via anti-apoptotic mechanisms(Reference Roberts, James, Woodyatt, Macdonald and Tugwood74) and to increase the number and size of mitochondria and peroxisomes in liver cells(Reference Hawkins, Jones, Bonner and Gibson75, Reference Lock, Mitchell and Elcombe76). The same treatment applied to grass carp did not change the weight of liver, but, however, increased the mitochondria content in this tissue, as has been previously observed in rainbow trout(Reference Du, Demizieux, Degrace, Gresti, Moindrot, Liu, Tian, Cao and Clouet23).
Although CPT I activities in isolated mitochondria were slightly lowered after fenofibrate treatment, the concomitant greater contents of mitochondria in the liver led to a marginal increase of CPT I activity expressed per g tissue, and this partly explained the increased activities of greater palmitate oxidation using liver homogenates in the fenofibrate group (Table 7). Indeed, previous studies reported that fenofibrate up-regulated mitochondrial activities via the nuclear DNA, and also via the mitochondrial genome, which would increase the liver content of mitochondria(Reference Mancini, Lanni, Sabatino, Moreno, Giannino, Contaldo, Colantuoni and Goglia13, Reference Casas, Pineau, Rochard, Rodier, Daury, Dauca, Cabello and Wrutniak-Cabello15). The increased activity of catalase, which is a peroxisome marker, was associated with a significantly increased activity rate of the peroxisomal FA-oxidising system and with a significantly increased carnitine-independent palmitate oxidation rate in the liver of fish treated with fenofibrate (Table 7). This strongly suggests that both the peroxisome induction by fenofibrate in the liver and the specificity of peroxisomal oxidation towards very-long-chain FA were mainly responsible for the marked reduction of EPA and DHA in the tissues studied.
In food-deprived rodents, PPARα was found to play a role in the regulation of mitochondrial and peroxisomal FA oxidation, which suggests that PPARα was involved in the transcriptional response to fasting(Reference Kersten, Seydoux, Peters, Gonzalez, Desvergne and Wahli14). This was supported by the fact that, during the fasting period, PUFA released from peripheral tissues were potential ligands for PPARα in liver cells(Reference Thomas, Schladt, Knehr and Oesch77), but also would imply that fenofibrate and fasting employed the same regulatory pathways(Reference Casas, Pineau, Rochard, Rodier, Daury, Dauca, Cabello and Wrutniak-Cabello15). This assertion was not fully satisfied in grass carp with the differences mentioned above between fasting and fenofibrate treatment, liver and muscle FA metabolism, and disappearance of MUFA v. EPA+DHA content. Further, starvation of Fischer-344 rats for 5 d was found to result in a 2·1-fold increase in liver peroxisomal β-oxidation(Reference Thomas, Schladt, Knehr and Oesch77), while another study reported that the activity of peroxisomal fatty acyl-CoA oxidase was 73 % lower in the liver of fasted rats(Reference Andriamampandry, Bnouham, Michard, Gutbier, Le Maho and Leray22). Interestingly, it has been shown in cultured hepatocytes that CPT I can be up-regulated by a PPARα-dependent mechanism using clofibrate and FA-CoA esters (also allowing peroxisomal proliferation) and by a PPARα-independent mechanism activated by non-esterified PUFA, both systems aiming at two different levels of the gene sequence(Reference Louet, Chatelain, Decaux, Park, Kohl, Pineau, Girard and Pegorier78). If this regulation also applied to grass carp, fenofibrate would induce mainly peroxisomal activities, while NEFA released from fat tissues during fasting and entering liver cells would induce mainly CPT I activities. The former activities would allow the preferential β-oxidation of EPA and DHA, and the latter would allow the preferential β-oxidation of MUFA. However, in fish, the peroxisomal induction by fenofibrate would be of much lesser extent than that in mammals, which suggests that the peroxisomal FA oxidation pathway is very weak in normal fish and would explain the usual accumulation of dietary EPA+DHA in tissues.
Lipid peroxidation
In grass carp fed the HF diet, there was a permanent infiltration of hepatic cell lipids. This metabolic state could result from excess uptake of blood lipids as lipoproteins and NEFA, insufficient use of lipids as energetic substrates and/or impaired lipoprotein secretion. Part of these could originate from the peroxidation of membrane lipids rich in highly unsaturated FA. Gray(Reference Gray79) found a geometrical relationship between sensitivity of FA to peroxidation and degree of unsaturation. Indeed, it has been shown that high fish oil levels in diets increased the susceptibility of turbot to FA peroxidation with a clear correlation between increased malondialdehyde production and decrease in PUFA(Reference Stephan, Guillaume and Lamour80). Further, Grattagliano et al. (Reference Grattagliano, Vendemiale, Caraceni, Domenicali, Nardo, Cavallari, Trevisani, Bernardi and Altomare66) pointed out that fatty livers were often associated with low levels of antioxidants and lipid peroxidation. The present results show that, even before fasting or fenofibrate treatment, there was already a relatively elevated level of plasma malondialdehyde when compared with previous results(Reference Du, Clouet, Zheng, Degrace, Tian and Liu25, Reference Du, Clouet, Huang, Degrace, Zheng, Tian and Liu26). Fasting represents an oxidative stress for animals(Reference Wohaieb and Godin81–Reference Morales, Perez-Jimenez, Hidalgo, Abellan and Cardenete83) and has been assumed to consume antioxidant stores of organs through the immediate generation of free radicals(Reference Pascual, Pedrajas, Toribio, Lopez-Barea and Peinado82). However, in the present experiment, fasting did not change plasma malondialdehyde concentrations despite the high amounts of EPA+DHA released from adipose tissues and transported in the bloodstream (Tables 3 and 6). These data suggest that there was a plateau level of lipoperoxidation (developed by the high TBARS concentration) balanced through convenient antioxidant defences, and also suggest that there was no absolute correlation between TBARS and liver fat contents. Indeed, 1 week of fasting resulted in hepatic steatosis but without any change in plasma TBARS concentrations and, in fenofibrate-treated fish, the decrease in hepatic lipids after fenofibrate treatment was associated with increased TBARS concentrations. Hypolipidaemic agents, such as fenofibrate, are known to trigger hepatomegaly and hepatic peroxisome proliferation in rats and mice(Reference Palma, Garrido, Rodriguez-Garcia and del Rio84). These fibrate derivatives are transformed within cells into compounds known to exhibit hepatocarcinogenic properties in the long term in rodents(Reference Reddy and Lalwani85). This might result from a metabolic imbalance due to a permanent free radical production(Reference Bieri and Lhuguenot86). Fenofibrate and clofibrate have also been shown to induce peroxisomal H2O2-generating reactions and to increase ubiquinol levels in tissues, however, without any change in plasma TBARS concentrations(Reference Elliot and Elcombe87–Reference Lores Arnaiz, Travacio, Monserrat, Cutrin, Llesuy and Boveris90). In grass carp treated by fenofibrate, the deleterious effects of H2O2 produced during the first steps of the peroxisomal FA oxidation pathway may be partly prevented through the increased activity of catalase. On the other hand, the increase in plasma malondialdehyde indicated that the peroxidation of unsaturated FA was insufficiently counteracted by appropriate antioxidant defences.
On the whole, grass carp fed an HF diet were shown to exhibit lipid-related properties which are close to those existing in mammals after fasting, but have a weak response towards PPAR ligands when compared with rodents and humans. This herbivorous fish species appears to represent a useful model for the investigation of mechanisms related to abnormalities of lipid metabolism.
Acknowledgements
The present study was funded by the National Natural Science Foundation of China (project no. 39970576/C020501). Z.-Y. D. designed all experiments, carried out the main experimental work and wrote the draft under the direction of project leader Y.-J. L.; P. C. and P. D. assisted with the experiment design and draft writing; W.-H. Z. provided all data of FA composition; L. F. provided important information to improve the final paper; L.-X. T. supplied healthy fish and did all the aquaculture management. The authors are also grateful to the referees for their comments for improvement. There are no personal or financial conflicts of interest.