Global fisheries are in decline and farmed fish constitute an increasing proportion of fish in the human food basket(Reference Subasinghe, Soto and Jia1). Thus, to assure the continuous growth of aquaculture production, the industry is obliged to find suitable alternatives to fish meal and fish oil in fish feeds. Plant products are the obvious choice, but vegetable oils are devoid of n-3 long-chain PUFA (LC-PUFA), and fillet fatty acid (FA) composition of farmed fish points towards a reduction in EPA (20 : 5n-3), docosapentaenoic acid (22 : 5n-3) and DHA (22 : 6n-3) content. Therefore, the sustainable development of aquaculture and the preservation of health benefits of fish consumption represent a complex trade-off, which is sometimes difficult to reconcile. Indeed, high levels of n-3 LC-PUFA are important quality factors in human foods, and there is increasing pressure to include EPA and DHA in the finishing diets of salmonids and freshwater fish, though they do not have specific n-3 LC-PUFA requirements(Reference Bell, Tocher and Henderson2–Reference Thanuthong, Francis and Senadheera4). This reinforces the interest for reliable FA descriptors linking dietary and muscle FA composition. In particular, the association between dietary and fillet FA composition is stronger in oily fish than in lean fish, and direct and highly predictable effects are likely to be found in marine fish due to their limited capacity to convert C18 PUFA of n-6 and n-3 series to their LC-PUFA(Reference Sargent, Tocher, Bell, Halver and Hardy5). According to this, fillet FA composition of gilthead sea bream juveniles is highly predictable for a given class of fish size and adiposity(Reference Benedito-Palos, Bermejo-Nogales and Karampatos6), and bivariate regression approaches with FA diet composition and fillet lipid content, as independent variables effectively mirror the year-round fillet FA composition of mature and immature fish(Reference Ballester-Lozano, Benedito-Palos and Navarro7).
Ration size triggers significant variations in muscle lipid levels of gilthead sea bream(Reference Velázquez, Zamora and Martínez8–Reference Valente, Cornet and Donnay-Moreno11). In salmonids, this outcome has also been broadly described, and concomitant changes in the FA composition of muscle total lipids (TL) have been reported(Reference Kiessling, Johansson and Storebakken12, Reference Kiessling, Åsgård and Storebakken13). Most of these changes in FA signatures have been primarily attributed to switches in the relative proportion of phospholipids (PL) and TAG(Reference Kiessling, Pickova and Johansson14, Reference Kiessling, Pickova and Eales15). However, the FA composition of these two lipid fractions is differentially regulated by nutritional and/or environmental conditions due to their different roles in cellular and lipid metabolism(Reference Sargent, Tocher, Bell, Halver and Hardy5, Reference Henderson and Tocher16). In fact, the FA composition of TAG usually bears a close resemblance to dietary lipids, whereas that of PL is highly influenced by environmental factors including temperature and osmolarity(Reference Los and Murata17, Reference Ibarz, Blasco and Beltran18), which may be attributed, at least in part, to changes in feeding levels. In this way, the aim of the present study was to evaluate the specific effects of ration size on fillet FA composition and mRNA gene expression of key regulatory enzymes of lipid and lipoprotein metabolism. For that purpose, lipid classes and FA composition of TL, TAG and PL were analysed in juveniles of gilthead sea bream fed from satiety to maintenance ration. In parallel, to gain knowledge of the regulation of lipid metabolism and the nutritional value of flesh, the expression pattern of fourteen selected genes was assessed by means of candidate gene array focused on tissue FA uptake, FA biosynthesis, PL synthesis and remodelling and upstream regulatory transcription factors. This issue was made after exhaustive searches in our gilthead sea bream transcriptome database, which rendered new sequences annotated as lipoprotein lipase-like (LPL-like), lysophosphatidylcholine acyltransferase 1, 2 and 3 (LPCAT1–3), phosphatidylethanolamine N-methyltransferase (PEMT), stearoyl-CoA desaturase 1a (SCD1a) and 1b (SCD1b) and sterol regulatory element-binding protein-1 (SREBP1), and uploaded to GenBank with accession numbers JQ277704–JQ277707, JQ277709, JQ390609 and JQ390612. The study also presented phylogenetic analyses of gene-lineage subfamilies (extracellular TAG lipases and SCD1 enzymes) for which mammalian orthologous genes do not exist yet or remain unclear.
Experimental methods
Animal care and experimental setup
Juvenile gilthead sea bream (Sparus aurata L.) of Atlantic origin (Ferme Marine de Douhet) were acclimatised to laboratory conditions for 25 d before the start of trial (May–July) in the indoor experimental facilities of the Institute of Aquaculture Torre de la Sal (IATS). Following the acclimatisation period, fish of 17 g initial mean body weight were randomly distributed into 500-litre tanks in triplicate groups of fifty fish each. Fish were fed from May to August (11 weeks) with a commercial diet (D-2 Excel 1P; Skretting) twice per d at three different feeding levels: (i) full ration until visual satiety (R100 group), (ii) 70 % of satiation (R70 group) and (iii) 70 % of satiation with the last 2 weeks at the maintenance ration (20 % of the satiation level; R70-20 group). Maintenance ration was conveniently calculated according to fish size and temperature (23–25°C) at this last stage of the experimental trial. The FA composition of the diet is reported in Table 1.
Diet fatty acid (FA) composition (% total FA methyl esters)
The trial was conducted under the natural photoperiod and temperature conditions at IATS latitude (40°5N; 0°10E). Water flow was 20 litres/min and oxygen content of water effluents was always higher than 85 % saturation. At the end of the trial period and following overnight fasting, nine randomly selected fish per dietary treatment were anaesthetised with 3-aminobenzoic acid ethyl ester (MS-222, 100 μg/ml). The entire fillet from fish left side (denuded from skin) was taken, vacuum packed and stored at − 80°C until lipid analyses. A representative portion of the skeletal muscle (dorsal white muscle) from the right side of the fish was taken under RNAse-free conditions, frozen with liquid N2 and stored at − 80°C until gene expression analyses. Nine aditional fish per dietary treatment were used for whole-body composition analyses.
All procedures were carried out according to the national (IATS-CSIC Review Board) and present EU legislation on the handling of experimental animals.
Lipid analyses
Fillets and specimens for whole-body analyses were ground and small aliquots were dried to estimate the moisture content. The moisture content was determined by drying in an oven at 105°C for 24 h. The remaining samples were freeze-dried and the TL content was determined gravimetrically using the Soxhlet 4001046 Auto extraction apparatus (Selecta) with 50 ml diethyl ether at 120°C, according to the manufacturer's instructions.
Fillet lipids for analyses of FA and lipid classes were extracted by the method of Folch et al. (Reference Folch, Lees and Sloane Stanley19), with chloroform–methanol (2:1) containing 0·01 % butylated hydroxytoluene as antioxidant. The analysis of lipid classes was performed by high-performance TLC using a one-dimensional double development, as described by Olsen & Henderson(Reference Olsen and Henderson20). Briefly, TL were separated on TLC plates (10 × 10 cm, silica gel (200 μm); Merck) using methyl acetate–isopropanol–chloroform–methanol–0·25 % aqueous potassium chloride (25:25:25:10:9, by vol.) and hexane–diethyl ether–acetic acid (85:15:1·5, by vol.) as developing solvent mixtures for PL and neutral lipids, respectively. The separated lipid fractions were charred for 10 min at 160°C after spraying the plate with 3 % (w/v) cupric acetate in 8 % (v/v) phosphoric acid(Reference Fewster, Burns and Mead21) and were quantified by calibrated scanning densitometry (Bio-Rad GS670 Imaging densitometer; Bio-Rad). The quantity of each lipid class was expressed on wet matter basis.
The FA from TL were fractionated by TLC (Silica gel G 60, 20 × 20 cm glass plates; Merck), using hexane–diethyl ether–acetic acid (85:15:1·5, by vol.) as a solvent system. The PL bands were scraped and extracted with chloroform–methanol (2:1, v/v) containing 0·01 % butylated hydroxytoluene. The TAG were scraped and extracted with hexane–diethyl ether (1:1, v/v). After the addition of non-adecanoic FA (Sigma) as internal standard, fractions of TL, TAG and PL were subjected to acid-catalysed transmethylation for 16 h at 50°C using toluene and 1 % (v/v) H2SO4 in methanol(Reference Christie22). The FA methyl esters (FAME) were extracted with hexane–diethyl ether (1:1, v/v), and those derived from TL were purified by TLC using hexane–diethyl ether–acetic acid (85:15:1·5, by vol.) as a solvent system. The FAME were then analysed with a gas chromatograph (GC 8000 Series; Fisons Instruments) equipped with a fused-silica 30 m × 0·25 mm open tubular column (film thickness: 0·25 μm; Tracer, TR-WAX; Teknokroma) and a cold on-column injection system. Helium was used as a carrier gas and temperature programming was from 50 to 180°C at 40°C/min and then to 220°C at 3°C/min. Peaks were recorded in a personal computer using a software package (Azur, Datalys). Individual FAME were identified by reference to well-characterised fish oil standards in the analysed sample.
Marinol (sardine oil) from the Fishing Industry Research Institute (Rosebank) and FAME 37 from Supelco were used as fish oil standards. Butylated hydroxytoluene and internal standard (19 : 0) were obtained from Sigma-Aldrich. All lipid solvents were of HPLC grade and were obtained from Merck.
Gene expression analysis
For the two extreme groups (R100 and R70-20), total RNA from skeletal muscle was extracted using the ABI PRISM™ 6100 Nucleic Acid PrepStation (Applied Biosystems) with a DNAse step. The RNA yield was 30–50 μg, with absorbance measures (A 260/280) of 1·9–2·1. Reverse transcription of 500-ng total RNA was performed with random decamers using the High-Capacity cDNA Archive Kit (Applied Biosystems). Quantitative real-time PCR was performed with an iCycler IQ Real-time Detection System (Bio-Rad), as described elsewhere(Reference Calduch-Giner, Mingarro and Vega-Rubín de Celis23). Briefly, diluted reverse transcription reactions were conveniently used for PCR reactions in 25 μl volume, in combination with a SYBR Green Master Mix (Bio-Rad). Specific primers for target genes and β-actin (housekeeping gene) were used at a final concentration of 0·9 μm (Table 2). The ΔΔC t method(Reference Livak and Schmittgen24) was used for the normalisation procedure, and expression values were referred to PEMT as reference gene. The set of primers for LPL, liver X receptor α (LXRα) and PPARα, PPARβ and PPARγ were the same as previously used(Reference Saera-Vila, Calduch-Giner and Gómez-Requeni25, Reference Cruz-García, Saera-Vila and Navarro26). New sets of primers were designed for FA desaturase 2 (FADS2) and additional sequences were generated following the update (November 2011) of the IATS complementary DNA gilthead sea bream database (http://www.sigenae.org/iats). These new sequences were submitted to GenBank as partial (LPCAT1, LPCAT2, LPCAT3, SREBP1) or complete (LPL-like, PEMT, SCD1a, SCD1b) codifying sequences (Table 3). Of note, the best hit for LPL-like is a red sea bream sequence (AB054063) annotated as LPL2 by Oku et al. (Reference Oku, Koizumi and Okumura27). SCD1a (JQ277703) and SCD1b (JQ277704) sequences were annotated according to the revised nomenclature for the stearoyl-CoA desaturase family(Reference Castro, Wilson and Goncalves28).
Forward and reverse primers for quantitative real-time PCR
* Fatty acid desaturases.
† Key enzymes on tissue fatty acid uptake.
‡ Transcription factors.
§ Enzymes involved in phospholipid synthesis and remodelling.
Characteristics of assembled sequences according to basic local alignment search tool (BLAST) searches
nt, Nucleotides; CDS, codifying domain sequence.
* Number of sequences.
† Gene identity determined through BLAST searches.
‡ Best BLAST-X protein sequence match.
§ Expectation value.
Phylogenetic analysis
Phylogenetic analysis was carried out on the basis of amino acid differences (p-distance) with the unweighted pair group method with arithmetic mean algorithm (pairwise deletion) in MEGA software (version 5.0; www.megasoftware.net). A total of thirty-seven sequences from fourteen species were included in the phylogenetic analyses of SCD1 enzymes and extracellular TAG lipases. Reliability of phylogenetic trees was assessed by bootstrapping using 1000 bootstrap replications.
Statistical analysis
Data on growth performance and lipid composition analyses were analysed by one-way ANOVA followed by the Student–Newman–Keuls test (P< 0·05). Data on gene expression were analysed by Student's t test. All analyses were performed using the SPSS 17.0 program (SPSS, Inc.).
Results
Growth performance and lipid analyses
Data on body weight are shown in Fig. 1. Fish of the R100 group grew from 17 (sd 0·12) to 72 (sd 0·34) g, with a feed efficiency (wet weight gain/dry feed intake) of 0·89 (sd 0·07). The 30 % feed restriction (R70 group) had a lesser final body weight (60 (sd 0·02) g) with a slight, but statistically significant (P< 0·05), improvement in feed efficiency (0·97 (sd 0·02)). In the R70-20 group, the feed intake was adjusted in the last 2 weeks to maintenance ration, and the final body weight (48·6 (sd 0·22) g) was accordingly lower. For more details on growth and bioenergetics, see Bermejo-Nogales et al. (Reference Bermejo-Nogales, Benedito-Palos and Calduch-Giner29).
Body weight over the course of the growth trial of fish fed a practical diet with different feeding levels. 
, Full ration until visual satiety (R100); 
, 70 % of satiation (R70); 
, 70 % of satiation with the last 2 weeks at 20 % of satiation level (R70-20).
As shown in Table 4, whole-body lipid content in wet matter basis was significantly lower in R70-20 group (8·8 %) in comparison to the other two experimental groups (12·3–11·7 %). In parallel, the water content increased significantly and progressively with the reduction of ration size from 65·3 % in the R100 group to 69·8 % in the R70-20. Similar trends were found for moisture and lipid content in fillet samples, but in this case, the reduction of lipid content was higher, changing progressively and significantly from 6·5 % in the R100 group to 3·3 % in the R70-20. The reduction of ration size also affected the lipid class composition, mainly due to changes in neutral lipids, with a 5·1 % reduction in the R100 group and 2·9–3·1 % in the other two experimental groups. A slight but significant reduction of phosphatidylethanolamine was found with the intermediate and severe reduction of ration size, but significant changes were not found for PL, considered globally as the sum of phosphatidylethanolamine, phosphatidylcholine (PC) and other minor polar lipid fractions.
Fillet lipid class composition (% wet weight), moisture (%) and lipid content (% fillet wet weight) of whole body and fillets of fish fed at three different feeding levels (Mean values and standard deviations, n 9)
R100, full ration until visual satiety; R70, 70 % of satiation; R70-20, 70 % of satiation with the last 2 weeks at 20 % of satiation level; CE, cholesteryl esters; COL, cholesterol; DG, diacylglycerols; MG, monoacylglycerols; PE, phosphatidyl ethanolamine; PA/CL, phosphatidic acid/cardiolipin; PI, phosphatidyl inositol; PS, phosphatidyl serine; PC, phosphatidyl choline; SM, sphingomyelin; NL, neutral lipids; PL, polar lipids.
a,b,c Mean values with unlike superscript letters were significantly different (P< 0·05).
The changes in ration size modified the fillet FA composition of TL (Table 5), with a statistically significant reduction in %FAME of 14 : 0, 16 : 1n-7, 18 : 1n-9, 18 : 2n-6, 18 : 3n-6 and 18 : 3n-3. This trend was the opposite for DHA, which achieved a maximum of 8·4 % in the R70-20 group and lower levels (7·2–7·3 %) in R70 and R100 groups.
Fatty acid (FA) composition (% total FA methyl esters) of total lipids from fillets of fish fed different feeding levels: full ration until visual satiety (R100 group), 70 % of satiation (R70) and 70 % of satiation with the last 2 weeks at 20 % of satiation level (R70-20) (Mean values and standard deviations, n 9)
a,bMean values within experimental groups with unlike superscript letters were significantly different (P< 0·05).
Experimental data evidenced minor changes in the FA composition of TAG (Table 6). Of note, only four FA (16 : 4, 20 : 1n-9, 22 : 1n-9, 22 : 6n-3) varied significantly and progressively with the reduction of ration size. This contrasted with the observations made in PL, where a high number of FA, including 14 : 0, 17 : 0, monoenes (16 : 1n-7, 18 : 1n-9, 18 : 1n-7, 20 : 1n-9) and most PUFA of n-3 (18 : 3n-3, 18 : 4n-3, 20 : 4n-3, 20 : 5n-3) and n-6 series (18 : 2n-6, 18 : 3n-6, 20 : 2n-6, 20 : 3n-6) decreased significantly in R70-20 fish. This trend was opposite for 20 : 4n-6 (arachidonic acid) and DHA, which varied between two extreme groups from 2·4 to 2·2 % and from 23·7 to 19·2 %, respectively.
Effect of feeding ration on the fillet fatty acid (FA) profile of TAG and phospholipids (% of total FA methyl esters; mean values and standard deviations of individual fish, n 9)
R100, full ration until visual satiety; R70, 70 % of satiation; R70-20, 70 % of satiation with the last 2 weeks at 20 % of satiation level.
a,b,c Mean values for each lipid class with unlike superscript letters were significantly different (P< 0·05).
Phylogenetic and gene expression analyses
Phylogenetic analysis of the endogenous extracellular TAG lipase family was performed with representative sequences of reptiles, mammals and fish, and the resulting tree revealed three major nodes of LPL, hepatic lipase and endothelial lipase (EL), according to the present hierarchy of vertebrates (Fig. 2(a)). A fourth node, closely related to LPL but exclusive of fish lineage, was identified. This additional node, named LPL-like, is presently represented by LPL sequences from zebrafish, tuna and sparid fish (red sea bream and gilthead sea bream).
Phylogenetic analyses of (a) endogenous extracellular TAG lipase family and (b) stearoyl-CoA desaturase 1 family. Bootstrap values are shown at each node. LPL, lipoprotein lipase; LPL-like, lipoprotein lipase-like; HL, hepatic lipase; EL, endothelial lipase; SCD1a, stearoyl-CoA desaturase 1a; SCD1b, stearoyl-CoA desaturase b.
As shown in Fig. 2(b), phylogenetic analysis of SCD1 sequences from modern teleost fish identified two major clades of SCD1a and SCD1b isoforms, according to the present hierarchy of teleosts. The analysis was carried out with sequences derived from model and non-model fish species, including those derived from gilthead sea bream. As expected, representative mammalian SCD1 sequences are fitted as outgroup when they are included in the analysis.
The muscle gene expression profile of the two extreme groups is shown in Fig. 3. Irrespective of experimental condition, the highest mRNA levels for the analysed lipid enzymes were encountered for SCD1b, LPL-like and LPCAT3, whereas lower levels were found for FADS2 and then for PEMT and SCD1a (Fig. 3(a)). As concerns transcriptional regulatory factors (Fig. 3(b)), the highest mRNA expression was found for SREBP1 and the lowest for LXRα, with intermediate values for PPAR isoforms. When the gene expression pattern was compared between R100 and R70-20 groups, no differences were found at the level of PEMT, LPCAT1, LPCAT2, LPCAT3, LPL, FADS2, FAS, PPARγ, LXRα and SREBP1. In contrast, the expression of LPL-like, PPARα and PPARβ was significantly up-regulated in the R70-20 group, whereas that of SCD1a and SCD1b was significantly down-regulated. When data for a given gene in R70-20 fish were referred to R100 fish, the fold-changes for the up-regulated genes were 1·93, 2·02 and 1·92 for LPL-like, PPARα and PPARβ, respectively. In the case of down-regulated genes, the fold-changes were 0·17 for SCD1a and 0·62 for SCD1b.
Relative mRNA expression of (a) lipid enzymes and (b) transcription regulatory factors in the two extreme groups (full ration until visual satiety (R100) and 70 % of satiation with the last 2 weeks at 20 % of satiation level (R70-20)) are plotted against each other in the scatter plot. Genes encoding lipoprotein lipase (LPL-like), PPARα and PPARβ, are significantly up-regulated (P< 0·05) in R70-20 fish. Genes encoding stearoyl-CoA desaturase 1a (SCD1a) and stearoyl-CoA desaturase 1b (SCD1b) are significantly down-regulated (P< 0·05) in R70-20 fish. LPCAT1, lysophosphatidylcholine acyltransferase 1; LPCAT2, lysophosphatidylcholine acyltransferase 2; LPCAT3, lysophosphatidylcholine acyltransferase 3; PEMT, phosphatidylethanolamine N-methyltransferase; FADS2, fatty acid desaturase 2; SREBP1, sterol regulatory element-binding protein-1; LPLα, lipoprotein lipase α.
Discussion
Muscle lipid content has a considerable interest for the farm animal industry, as it strongly affects meat quality(Reference Wood, Enser and Fisher30). Hence, dietary manipulation and genetic selection are presently the main tools used to manage muscle fat content in farmed fish(Reference Kolditz, Borthaire and Richard31–Reference Bell, Pratoomyot and Strachan33). Different flesh lipid deposition rates also modify the nutritional value of the end product and, thereby, the present study was focused on the effect of ration size on the fillet FA profiles of gilthead sea bream. From the present results, it is conclusive that the reduction of ration size entailed the decrease of lipid content of whole body and fillet, which was accompanied by a decrease of water content, whereas that of protein and ash remained unchanged (data not shown). The decrease of TL was closely associated to the reduction of neutral lipid content, whereas the amount of polar lipids remained unchanged. Regarding FA profiles, our previous gilthead sea bream study highlights the fact that the fillet FA composition of PL remains mostly unaltered when the theoretical requirements of essential FA are met through diet(Reference Benedito-Palos, Navarro and Sitjà-Bobadilla34). Moreover, the robustness of PL is apparently greater in tissues with higher dietary requirements in essential FA(Reference Benedito-Palos, Navarro and Kaushik35). This contrasts with the plasticity of TAG, which closely resembles the FA composition of the diet in a wide range of cultured fish, including gilthead sea bream(Reference Benedito-Palos, Navarro and Kaushik35), European sea bass(Reference Skalli and Robin36), Atlantic cod(Reference Jobling, Leknes and Sæther37), turbot(Reference Regost, Arzel and Cardinal38), Atlantic salmon(Reference Jobling and Bendiksen39), sunshine bass(Reference Trushenski, Lewis and Kohler40), pike perch(Reference Schulz, Knaus and Wirth41) and arctic charr(Reference Pettersson, Pickova and Brännäs42). However, ration size and dietary lipid sources affect the FA composition of neutral and polar lipids differentially, at least when evaluated in juvenile gilthead sea bream fed at maintenance ration for 2 weeks in the summer growing period. Indeed, the fillet FA profile of TAG remained mostly unaltered with the reduction in ration size and muscle lipid depots, whereas polar lipids exhibited significant changes that mostly affected arachidonic acid and DHA in the case of LC-PUFA. Earlier studies in rainbow trout(Reference Kiessling, Pickova and Johansson14) and chinook salmon(Reference Kiessling, Pickova and Eales15) do not support this finding, and the authors concluded that the effect of ration on muscle FA composition is primarily due to changes in the relative levels of PL and TAG. However, in the present study, changes in the PL profile contribute directly and/or indirectly to the lean phenotype of R70-20. It is not easy to explain the differences in lipid metabolism of these apparent fish species, but it should be stressed that the overall literature strongly supports the allostatic regulation of PL in response to different stimuli(Reference Los and Murata17).
Several studies in cultured fish with varying feeding levels(Reference Delgado, Estévez and Hortelano43, Reference Kjaer, Vegusdal and Berge44) and diet composition(Reference Torstensen, Froyland and Ørnsrud3, Reference Bendiksen and Jobling45–Reference Pratoomyot, Bendiksen and Campbell48) denote that both arachidonic acid and DHA can be selectively retained to meet the requirements for tissue membrane FA and function. It has also been observed that lean strains of Atlantic salmon accumulate n-3 LC-PUFA more rapidly that fat fish(Reference Bell, Pratoomyot and Strachan33). For instance, intervention studies in human subjects and rodents highlight the fact that hypoenergetic diets facilitate the unsaturation of skeletal muscle PL in association with improved insulin sensitivity(Reference Storlien, Jenkins and Chisholm49–Reference Lombardo and Chicco52). This agrees with the observation that the Pima Indians, a population with a high prevalence of diabetes and obesity, display a low concentration of n-3 LC-PUFA in muscle PL(Reference Pan, Lillioja and Milner53). Conversely, insulin sensitivity in obese subjects can be facilitated by the incorporation of DHA in the muscle membrane to a level that may exceed the level found in a lean healthy background population(Reference Haugaard, Madsbad and Hoy54, Reference Haugaard, Vaag and Mu55). This functional association has not been demonstrated in fish yet, although the increased incorporation of DHA in the muscle PL of R70-20 fish can be viewed as part of the complex regulatory mechanisms that orchestrate the change towards energy storage rather than energy mobilisation when nutrient insulin secretagogues become available above a critical threshold level. This type of regulation also operates on a daily basis, and earlier studies in the European sea bass evidenced that the stimulation of feeding behaviour in fish fed at mid-day is accompanied by the phase advancement of circadian insulin rhythms(Reference Pérez, Zanuy and Carrillo56).
It is well known that PL are first synthesised in the de novo Kennedy pathway(Reference Kennedy and Weiss57), but their FA composition at the sn-2 position is altered in the remodelling pathway (Land's cycle)(Reference Lands58) through the concerted action of acyltransferases and phospholipase A2 enzymes. Both biosynthetic pathways have also been described in teleost fish(Reference Tocher59, Reference Tocher, Bendiksen and Campbell60). The discovery of a number of acyltransferases has recently occurred(Reference Hishikawa, Shindou and Kobayashi61), but it appears that both mammals and fish contain a number of LysoPL acyltransferases that exhibit distinct acyl-CoA and lysoPL acceptor specificities(Reference Tocher and Sargent62). Herein, we focused on LPCAT, given that PC is the principal component of animal cellular membranes(Reference Kazachkov, Chen and Wang63), and interestingly, the three LPCAT reported in human subjects(Reference Pérez-Chacón, Astudillo and Balgoma64) are represented by the corresponding orthologues (LPCAT1–3) in our gilthead sea bream transcriptome database. Besides, all of them were expressed at detectable levels in the skeletal muscle of gilthead sea bream (LPCAT1< LPCAT2< LPCAT3), but none was transcriptionally regulated by ration size in the present experimental model, which limits their role as master regulators of PL FA composition. In this respect, it must also be noted that LPCAT are highly selective for PUFA, but do not discriminate well between n-6 PUFA and n-3 PUFA, which would explain, at least in part, why the membrane FA composition of rat skeletal muscle is highly responsive to the balance between n-6 and n-3 PUFA in the diet(Reference Abbott, Else and Hulbert65).
Some other important regulatory steps of PL synthesis and remodelling are the transmethylation reactions catalysed by the PEMT enzyme to convert phosphatidylethanolamine to PC. Of note, cystidine diphosphate choline is the major source of PC under normal physiological conditions but, interestingly, the expression of PEMT is markedly up-regulated in the liver tissue of human subjects and rodents when dietary choline or PC are deficient(Reference Cui and Vance66–Reference Dong, Wang and Li68). In consequence, as phosphatidylethanolamine unsaturation is generally higher than that of PC(Reference Tocher, Bendiksen and Campbell60, Reference Tocher69), a different contribution of cystidine diphosphate choline and PEMT pathways on PC biosynthesis may alter the final composition of plasma membrane PL. Nevertheless, as reported above for LPCAT, the PEMT enzyme was not transcriptionally regulated by severe energy restriction in the present experimental model, which suggests that the observed changes in PL FA composition were more directly related to switches in FA metabolism than PL biosynthesis and remodelling. In that sense, other authors have suggested that PL biosynthesis pathways are limited or inefficient for several fish species (reviewed by Tocher et al. (Reference Tocher, Bendiksen and Campbell60)).
Obviously, one important issue in FA metabolism is the capability for LC-PUFA biosynthesis of different fish species. This has been widely studied in recent years, and a general statement in marine fish, including gilthead sea bream(Reference Seiliez, Panserat and Corraze70), is that the expression of FADS2 (Δ-6 desaturase) is induced at very low levels. This, together with the blockage of Δ-5 desaturase and C18-C20 elongase steps, greatly limits the flux through LC-PUFA biosynthetic pathways in marine fish(Reference Tocher, Zheng and Schlechtriem71, Reference Zheng, Ding and Xu72). In the present study, this point of view is reinforced by the reduced mRNA expression of FADS2 in comparison to other analysed transcripts. Additionally, FADS2 was little responsive to changes in ration size; therefore, the enhanced incorporation of LC-PUFA (arachidonic acid and DHA) in muscle PL of R70-20 fish cannot be explained by the enhanced biosynthesis of LC-PUFA.
The SCD enzymes with Δ-9 desaturase activity are ubiquitously found in all living organisms and their mRNA expression, in particular that corresponding to SCD1 enzymes, was highly regulated by ration size in the present experimental model. Major products of SCD activity are palmitoleic acid and oleic acid, but the evolutionary history of these rate-limiting enzymes in the biosynthesis of monoenes is a complex process of duplication and loss events. Thus, as recently reviewed by Castro et al. (Reference Castro, Wilson and Goncalves28), a different number of paralogous genes (SCD1–SCD5) have been reported in tetrapods and fish. In particular, SCD enzymes have evolved through the evolution of teleost lineage as duplicated genes of SCD1-type (SCD1a, SCD1b), as was further corroborated by the phylogenetic analyses of SCD sequences, including those from gilthead sea bream. From the present results, it is also conclusive that the two SCD isoforms of gilthead sea bream were significantly expressed in the skeletal muscle, although clearly the most abundant mRNA was that codifying for the SCD1b isoform. A different tissue expression level has also been reported in the adipose tissue for the two SCD1 enzymes of red sea bream(Reference Oku and Umino73), but regardless of this, the two gilthead sea bream SCD1 isoforms were down-regulated by ration size in the skeletal muscle, which was not surprising given that SCD enzymes are strong markers of lipogenesis in human subjects and rodents and low SCD expression and activity have been associated to a reduced risk of obesity and insulin resistance(Reference Ntambi, Miyazaki and Stoehr74–Reference Sampath and Ntambi78).
A third means of regulation of tissue FA composition is the endogenous extracellular TAG lipases, which act as key-limiting enzymes on tissue FA uptake. The overall literature strongly supports a distinct regulation of LPL, hepatic lipase and EL, with distinct substrate specificities to accommodate the full spectrum of circulating lipoproteins(Reference Wong and Schotz79). Hence, LPL displays the highest TAG activity and EL the lowest, whilst at the same time becoming more active as phospholipase(Reference McCoy, Sun and Marchadier80). The expression of lipases is also tissue specific and EL is synthesised by endothelial cells and by other cell types such as macrophages(Reference Yasuda, Hirata and Ishida81), which agrees with the observation that the gilthead sea bream EL sequence (JQ390610) derived from de novo assembly of next-generation sequencing data contains mostly nucleotide reads from head kidney, the equivalent of bone marrow in mammals. Conversely, the expression of hepatic lipase is restricted to liver (A Saera-Vila and J Pérez-Sánchez, unpublished results), whereas LPL is more ubiquitous, and together with the hormone-sensitive lipase drives the reallocation of body fat depots in fish either with lean phenotypes(Reference Cruz-García, Saera-Vila and Navarro26) or with nutritionally mediated changes in lipid storage capacity(Reference Saera-Vila, Calduch-Giner and Gómez-Requeni25, Reference Cruz-García, Sánchez-Gurmaches and Bouraoui82). In mammals, it is also well known that fasting down-regulates the LPL activity of adipose tissue(Reference Lladó, Pons and Palou83, Reference Bergo, Olivecrona and Olivecrona84), with no changes or increases in heart and skeletal muscle(Reference Sugden, Holness and Howard85, Reference Ruge, Svensson and Eriksson86). Likewise, fasting decreases LPL activity and mRNA expression in the adipose tissue of gilthead sea bream, which is reversed by either insulin or arginine (potent insulin secretagogue) treatments(Reference Albalat, Saera-Vila and Capilla87). In the same study, a role of insulin in the regulation of LPL has not been proved in other tissues, which is not surprising given that results from this and previous studies(Reference Saera-Vila, Calduch-Giner and Gómez-Requeni25, Reference Saera-Vila, Calduch-Giner and Navarro88) highlight a constitutive or low expression of LPL in the white skeletal muscle. In contrast, LPL-like is the predominant LPL isoform in the skeletal muscle of gilthead sea bream and its mRNA expression was markedly up-regulated by the reduction of ration size. Experimental evidence also indicates that LPL transcripts are differentially regulated in the liver tissue of Atlantic salmon in response to diet(Reference Morais, Edvardsen and Tocher89). A different tissue distribution has also been reported for LPL and LPL-like in red sea bream(Reference Oku, Koizumi and Okumura27), which suggests that, at least in sparid fish, LPL-like is preferentially expressed in muscle tissues in a well-regulated manner. This offers the possibility of a complex tissue regulation that might contribute to modifying the muscle PL FA composition if LPL and LPL-like enzymes actually have distinct TAG and PL activities.
TNF-α is a mediator of insulin resistance, and important studies with TNF-α-deficient mice evidenced the improvement of insulin sensitivity in hyperphagic models of rodent obesity(Reference Ventre, Doebber and Wu90). In mammals, most of these TNF-α-mediated effects are produced by stimulating lipolysis and inhibiting LPL activity at transcriptional and post-transcriptional levels(Reference Bulló-Bonet, García-Lorda and López-Soriano91). A lipolytic effect of recombinant human TNF-α has been demonstrated in freshly isolated adipocytes of rainbow trout(Reference Albalat, Gómez-Requeni and Rojas92) and gilthead sea bream(Reference Saera-Vila, Calduch-Giner and Navarro88) and, interestingly, sequence analysis of the 5′-flanking region of gilthead sea bream LPL reveals a strong conservation of some regulatory elements (GC box, CCAAT/NF-Y, OCT-1) with a synergistic effect on the promoter activity of avian and mammalian LPL genes. Conserved regulatory elements for LXR, PPARα and SREBP1 have also been reported in the promoter region of SCD enzymes(Reference Mauvoisin and Mounier93), although, unfortunately, SCD promoters remain little studied in fish and in gilthead sea bream, in particular. This makes the interpretation of results arising from upstream regulatory transcriptors more difficult, although it must be noted that PPARα, and also PPARβ, were responsive to changes in ration size in the present experimental model. In the case of PPARα, its increased mRNA expression with the reduction of ration size is not surprising given that this PPAR isoform has been traditionally associated with lipolytic and SCD-deficient phenotypes(Reference Ntambi, Miyazaki and Stoehr74, Reference Sampath and Ntambi78, Reference Miyazaki, Flowers and Sampath94). More difficult to explain is the response of PPARβ, but given that this more ubiquitous PPAR isoform has recently been associated to lipogenic rather than lipolytic roles(Reference Barak, Liao and He95, Reference Peters, Lee and Li96), the reduced but also stimulated expression of PPARβ in R70-20 fish can be viewed as a counter-regulatory response to maintain the muscle lipolytic and lipogenic pathways finely regulated. Indirect evidence for this is provided by the earlier in vitro studies, in which the lipolytic TNF-α cytokine exerts dual effects in primary cultures of gilthead sea bream adipocytes that differ among and between lean and fat fish(Reference Cruz-García, Saera-Vila and Navarro26). However, at least in one set of fat fish, the lipolytic effects of TNF-α were apparently mediated by the down-regulated expression of PPARβ.
In summary, a number of new gilthead sea bream sequences (EL, LPL-like, LPCAT, PEMT, SCD1, SREBP1) with important roles in lipid metabolism have been introduced in public database repositories. Additionally, new insights into muscle FA composition and mRNA expression patterns have been addressed in an experimental model of ration size. Interestingly, the results highlight a distinct regulation of FA composition in neutral and polar lipids, which reinforces the concept of the allostatic regulation of PL. From a mechanistic point of view, it is complex to define the vast array of enzymes and upstream regulatory transcriptors that contribute to modify the membrane FA composition, although PPAR, LPL-like and SCD1 are emerging as reliable target genes to direct marine fish production and selection towards lean muscle fish phenotypes with increased retentions of LC-PUFA.
Acknowledgements
The present study was funded by the Spanish MICINN (AQUAFAT, AGL2009-07797, Predictive modelling of flesh FA composition in cultured fish species with different muscle lipid content; AQUAGENOMICS, CSD2007-00002, Improvement of aquaculture production by the use of biotechnological tools) and the European Union (ARRAINA, FP7-KBBE-2011-5-288925, Advanced research initiatives for nutrition and aquaculture) projects. Additional funding was obtained from the ‘GeneralitatValenciana’ (research grant PROMETEO 2010/006). G. F. B.-L. was recipient of a Spanish PhD fellowship from the Diputación Provincial de Castellón. The authors are grateful to M. A. González for the excellent technical assistance in PCR analysis. The authors declare that there are no conflicts of interest perceived to bias the study. L. B.-P. performed the FA and phylogenetic analyses; J. A. C.-G. supervised gene expression analyses; G. F. B.-L. was involved in fish culture; and J. P.-S. designed and coordinated the work.
