Parental LC-PUFA biosynthesis capacity and nutritional intervention with alpha-linolenic acid affect performance of Sparus aurata progeny

Environmental factors such as nutritional interventions during early developmental stages exert long-lasting metabolic and physiological changes in mammals (Burdge and Lillycrop, 2010) and other vertebrates, including fish (Izquierdo et al., 2015; Turkmen et al., 2017a; Turkmen et al., 2019a; Turkmen et al., 2017b; Xu et al., 2019). Polyunsaturated fatty acids (PUFA) play important roles in programming the metabolism of different organisms (Lillycrop and Burdge, 2018). Gilthead seabream (Sparus aurata), a marine fish, is an interesting animal model to study the effects of alterations of PUFA during the early stages as the yolk sac composition of the offspring depends on the continuous uptake of nutrients and can be modified to some extent through the diets supplied during the spawning period (Fernández-Palacios et al., 1995). In recent years, there has been increasing interest in the replacement (either partial or complete) of fishmeal (FM) and fish oil (FO) with more readily available plant protein sources and vegetable oils (VO) in diets used for aquaculture (Turchini et al., 2009; Vestergren et al., 2012). However, high substitution levels may negatively affect a range of different parameters such as growth or health (Izquierdo et al., 2005; Montero and Izquierdo, 2010; Rosenlund et al., 2010; Torrecillas et al., 2017a; Torrecillas et al., 2017b). While marine oils are rich sources of n-3 long-chain polyunsaturated fatty acids (LC-PUFA), oils extracted from conventional terrestrial oil seeds – namely VO – rarely have ≥20 carbon fatty acids and totally lack essential fatty acids, such as eicosapentaenoic acid (20:5n-3, EPA) and docosahexaenoic acid (22:6n-3, DHA). Commonly used VOs can be rich in precursors of LC-PUFA, such as alpha-linolenic acid (18:3n-3, ALA) and linoleic acid (18:2n-6, LA), and certain oils such as linseed oil (LO) even have a very high content of ALA, which is the primer substrate for the n-3 LC-PUFA. In the LC-PUFA biosynthesis pathway, there are several elongation, desaturation and β-oxidation steps, and in different species of fish, there are differences in the evolution of the synthesis capacity (Castro et al., 2016; Monroig et al., 2018). Therefore, maximizing the capacity of these pathways in a given species is of interest to improve the utilization of VO sources in fish, given the limited availability of marine fish oils.

Recently, studies have looked into the possibility of channelling specific metabolic pathways of fish either through broodstock nutrition or through early nutritional conditioning (Engrola et al., 2018; Izquierdo et al., 2015; Panserat et al., 2019; Turkmen et al., 2017a; Turkmen et al., 2017b; Vagner et al., 2007). Nutritional programming presumes that early environmental clues such as diet or specific nutrients provide the organism with the capacity to forecast environmental challenges in later life and give it the opportunity to adjust its metabolism to better adapt it to the new environment, potentially improving health, reproduction and survival (Burdge and Lillycrop, 2010). Hence, this tool may have applications in the animal production sector, as shown in ruminants, to channel the individuals to better utilization of key nutrients (Gotoh, 2015). In rodents, dietary fatty acid composition during pregnancy and lactation influences growth and glucose metabolism in the offspring (Siemelink et al., 2002). Likewise, maternal phytoestrogens can affect gene expression and alter skin colour as well as susceptibility to obesity in adulthood (Dolinoy et al., 2006). Among marine teleosts, gilthead seabream is a multi-batch spawner whose eggs largely depend on the continuous intake of nutrients by the broodstock during the spawning season when egg composition can be modified – to some extent – by the diets used (Fernandez Palacios et al., 2011). Indeed, previous research in gilthead seabream has established that feeds supplied during the spawning season play a critical role in regulating lipid metabolism even in the offspring (Izquierdo et al., 2015; Turkmen et al., 2017b). For instance, expression of genes related to the fatty acid desaturation and elongation pathways, such as fatty acyl desaturase 2 (fads2) and fatty acid elongase 6 (elovl6), was affected, as well as expression of other genes involved in lipid metabolism, such as carnitine palmitoyltransferase I (cpt1) or lipoprotein lipase (lpl) (Izquierdo et al., 2015). Some stress-related genes were also found to be regulated in the offspring obtained from broodstock fed diets with different VO levels during early larval stages (Turkmen et al., 2019a). Low FM FO feed utilization can also be improved by early dietary interventions in Atlantic salmon (Salmo salar) (Clarkson et al., 2017), through the up-regulation of different genes including those involved in oxidative phosphorylation, pyruvate metabolism, tricarboxylic acid cycle, glycolysis or fatty acid metabolism (Vera et al., 2017). That nutritional interventions by different fatty acids during periods of high developmental plasticity may also regulate dietary lipid utilization in later stages in fish was also shown in European seabass (Vagner et al., 2007). In the gilthead seabream, our own previous studies showed that replacement of different levels of FO (rich in n-3 series fatty acids with 20 or more carbon atoms, n-3 LC-PUFA) by VO (rich in C₁₈ fatty acids) improved growth of offspring challenged with a high vegetable meal (VM) and VO diet (Izquierdo et al., 2015). However, these studies did not show whether this effect was related to the increase in C₁₈ precursor or the reduction on n-3 LC-PUFA in the programming diet used during the spawning period to condition the offspring.

Selective breeding provides the opportunity to improve the economic production efficiency of aquatic livestock (Gjedrem et al., 2012). In fish, critical analysis of data on the response to selection for different species and traits shows that gain per generation can vary between 4% and 40%, depending on the species and the chosen criterion (Chavanne et al., 2016; de Verdal et al., 2018; Gjedrem and Rye, 2018; Janssen et al., 2017). Such selection programmes can be also tailored to address the present needs in aquaculture such as utilization of feeds that are less reliant on marine capture fishery-derived ingredients. Studies undertaken with rainbow trout (Oncorhynchus mykiss) as well as European sea bass (Dicentrarchus labrax) have shown the potential of selection for improved use of 100% VM and VO diets free of marine sources (Le Boucher et al., 2012). One of the approaches in animal breeding programmes is to use biomarkers for a certain expected outcome by identifying genes related to the desirable characteristics (Cassar-Malek et al., 2008), as was shown in rainbow trout (Le Boucher et al., 2013). In this sense, the fads2 gene which codifies the delta-6-desaturase enzyme (delta-6), a rate-limiting step of LC-PUFA biosynthesis, is a very strong candidate as its regulation in response to VO is well documented in a variety of fish species (Vagner and Santigosa, 2011), including gilthead seabream (Izquierdo et al., 2008). In mice, knockout of Fads2 and the absence of LC-PUFA in the diet resulted in failure of reproduction, showing the key role of Fads2 among seven other desaturases (Scd1–5, Fads1, Fads3) in this species (Stoffel et al., 2008). Studies in humans show a very high correlation (up to 70%) between red blood cell fatty acid composition of the parents and its inheritance in the offspring (Lemaitre et al., 2008). In this sense, blood fatty acid composition or fads2 expression could be used as an important trait. Although there are genetic selection programmes in gilthead seabream such as the public Spanish breeding programme PROGENSA^® (Afonso et al., 2012), and other commercial breeding programmes (Fernandes et al., 2017; Thorland et al., 2015) which include selection pressure on growth traits and an absence of body deformities (García-Celdrán et al., 2015; Lee–Montero et al., 2015), up to now there has been no information about the use of fads2 expression as a biological biomarker in fish.

The present study aimed to determine the influence of using broodstock with high fads2 expression in combination with conditioning feeding with a high VO diet on spawning quality and growth of offspring during larval development and juvenile stage when challenged with very low levels of marine fisheries-derived ingredients. Several parameters related to egg and larval quality, growth and fatty acid profiles were studied to ascertain a potential nutritional programming effect in the offspring.

All the experiments described below were conducted according to the European Union Directive (2010/63/EU) on the protection of animals for scientific purposes, at the facilities of Ecoaqua Institute, University of Las Palmas de Gran Canaria (Canary Islands, Spain).

A total of 70 gilthead seabream, Sparus aurata Linnaeus 1758, broodstock fish (42 females and 28 males, aged 2–4 years) were used for identifying individuals with high or low fads2 expression for the current investigation. These individuals were fed a high VO diet (Table S1) at a daily feeding ration equal to 1% biomass at 08:00 h and 14:00 h, 6 days a week for 1 month. After the feeding period, whole blood was taken from the caudal vein of the brood fish for the identification of fads2 expression levels of the individuals. Prior to measurements, all fish were anaesthetized with 10 ppm clove oil/methanol (1:1 v/v) in sea water. A sample of 2 ml blood was taken with 2 ml sterile syringes (Terumo Europe NV, Leuven, Belgium) and transferred to 2.5 ml K3 EDTA tubes (L.P. Italiana, Milan, Italy). Whole-blood samples were kept on ice during sampling and immediately centrifuged at 11,200 g, 4°C for 20 min. Plasma was separated, and erythrocytes were snap frozen with liquid nitrogen and kept at −80°C until RNA extraction. RNA extraction and purification are explained more in detail below (see ‘Biochemistry and gene expression analyses’). Broodstock showing the highest and the lowest fads2 expression from the 70 fish sampled were separated into two groups, HD and LD, respectively. Twelve brood fish of each sex from the HD and 12 from the LD fads2 expression group, with similar length and mass (P>0.05, Table 1), were distributed into 12 experimental tanks in a flow-through system with filtered seawater (mean±s.d. 37.0±0.5‰ salinity, 19.59–21.30°C) at a renewal rate of 100% per hour with proper aeration.

Spawning quality was determined before and after feeding with the experimental diets (Table 2). At the beginning of the trial, the mean body mass for females and males was measured (Table 1). Fish from the HD and LD groups were randomly assigned to experimental tanks using a ratio of 1 female and 1 male per tank. After placing the brood fish in the experimental tanks, fish were fed with commercial diets and spawning quality was monitored daily. Two isocaloric and isonitrogenous diets (Biomar, Aarhus, Denmark) were formulated to contain either high levels of FM and FO (diet F) or 30% FO and 70% VO (diet V; Table 2). Once equal spawning quality parameters were observed between the tanks during the acclimation period of 21 days (P>0.05, data not shown), tanks were randomly assigned to one of the four experimental groups as follows: FHD, high fads2 expression broodfish (HD) fed with 100% FO diet (F); FLD, low fads2 expression broodfish (LD) fed with 100% FO diet (F); VHD, high fads2 expression broodfish (HD) fed with 30% FO–70% VO diet (V); and VLD, low fads2 expression broodfish (LD) fed with 30% FO–70% VO diet (V).

The spawning quality parameters were determined using eggs collected daily from each tank at around 08:00 h and concentrated in 5 l beakers. Immediately, eggs were transferred to the laboratory and aeration was supplied to ensure mixing of the eggs through the water column. Then, a 5 ml sample was taken using a graduated glass pipette and transferred to a Bogorov chamber. Eggs were counted and observed under a binocular microscope (Leica Microsystems, Wetzlar, Germany) in five replicates to calculate the total number of eggs and the percentage of fertilized eggs, and to determine the morphological characteristics. Egg viability rate was determined as the percentage of morphologically normal eggs at the morula stage, described as transparent, perfectly spherical with clear, symmetrical early cleavage (Fernandez Palacios et al., 2011). After that, eggs were randomly transferred to 96-well ELISA microplates using a micropipette (0.7 ml of seawater and one egg per well). Plates were observed under a binocular microscope to ensure that there was a single fertilized egg in each well. These eggs were kept at a controlled temperature of 23°C. By observing the egg from these ELISA plates after 24 and 72 h under a binocular microscope, hatching rate and survival 3 days post-hatching (dph) were calculated as percentages. With these percentage values, the total number of fertilized, viable and hatched eggs and larvae alive at 3 dph was calculated per kg female per spawn.

After 1 month of feeding with the respective diets, eggs were collected and distributed into 500 l tanks for mass production at a density of 100 eggs l⁻¹ from each treatment group. All larvae were reared following the same common rearing protocol, regardless of the origin of the spawn. Water renewal in the tanks was progressively increased from 10% to 40% per hour until 46 dph and the water was continuously aerated (125 ml min⁻¹). Larvae were reared under natural photoperiod and living phytoplankton [Nannochloropsis sp.; mean±s.d. 250(±100)×10³ cells ml⁻¹] was added to the rearing tanks. From 3 to 17 dph, larvae were fed twice a day with rotifers (Brachionus plicatilis, 10 rotifer ml⁻¹) enriched with commercial emulsions (ORI-GREEN, Skretting, Norway). From 15 to 32 dph, Artemia sp. enriched with commercial emulsions (ORI-GREEN) was added to the rearing tanks 3 times a day. From 20 dph, larvae were fed commercial diets according to the suggested diet particle size by the manufacturer (Gemma Micro, Skretting, France). Water was continuously aerated (125 ml min⁻¹) attaining 6.1±0.4 ppm dissolved O₂ (mean±s.d. ). Average water temperature and pH along the trial were 21.1±0.4°C and 7.0±0.6, respectively. Water quality was monitored daily as regards dissolved O₂ and pH. At 3, 15 and 30 dph, growth was determined by measuring the total length of 60 anaesthetized larvae per treatment using a profile projector (V-12A, Nikon, Tokyo, Japan). RNA extraction methods are detailed below (see ‘Biochemistry and gene expression analyses’). At 46 dph, each experimental group was transferred to 10,000 l tanks in duplicate.

Obtained offspring were kept in similar environmental conditions. Fish were fed with commercial diets all through the grow-out period until they reached 6 months of age, and were fed 3 times a day by hand until apparent satiation. Rearing water temperature and photoperiod were natural and equal for all the experimental groups (18.5–21.8°C, 11–13 h light). Triplicate groups of juvenile fish (mean±s.d. initial body mass of all groups: 23.3±0.7 g, n=12) from each broodstock group were assigned to one of 12 tanks (500 l capacity) for the juvenile feeding challenge experiment (n=75 per each treatment). Fish were kept in a flow-through system and had a natural photoperiod (12 h light:12 h dark). Triplicate groups of fish initially coming from one of the four broodstock groups (FHD, FLD, VHD and VLD) were reared with a pelleted low FM and low FO diet (Table 3) and were fed until apparent satiation twice a day at 08:00 h and 14:00 h, 6 days a week over 60 days. Uneaten pellets were collected in a net by slowly opening the water outlet for 10 min after each meal, dried in an oven for 24 h and weighed to estimate net feed intake. The amount of feed given to each tank was recorded daily. Water temperature and oxygen levels were monitored daily after the last feeding. Average water temperature was 22.5±0.7°C, and dissolved oxygen was 6.1±0.3 mg l⁻¹ during the experimental period.

All fish were anaesthetized with 10 ppm clove oil:methanol (1:1 v/v) in sea water prior to measurements for the initial and final samplings. Fish were fasted for 24 h then individually weighed. Growth was determined by measuring wet body mass after 24 h starvation at 30 and 60 days of the experiment. Prior to measurements, all fish were anaesthetized with 10 ppm clove oil:methanol (1:1 v/v) in sea water. The experimental scheme is presented graphically in Fig. 1. Growth and feed utilization parameters were calculated using the following equations: feed conversion ratio (FCR)=(dry mass of consumed feed)/(final biomass−initial biomass), and weight gain (%)=(final biomass−initial biomass)/initial biomass×100, where mass is in g.

Lipid, protein, ash and moisture analysis of the samples was done as previously described (Turkmen et al., 2017b). For each 200 µl of blood cells, 1 ml of TRI Reagent (Sigma-Aldrich, St Louis, MO, USA) was added into 2 ml Eppendorf tubes. To each tube, four pieces of 1 mm diameter zirconium glass beads were added and homogenized using TissueLyzer-II (Qiagen, Hilden, Germany) for 60 s with a frequency of 30 s⁻¹; 250 µl chloroform was added to homogenized samples, which were then centrifuged at 12,000 g for 15 min at 4°C for phase separation. The clear upper aqueous phase containing RNA was mixed with 75% ethanol and transferred into an RNeasy spin column to bind total RNA. Then, RNA was extracted using an RNeasy Mini Kit (Qiagen) with the protocol supplied by the manufacturer. Real-time quantitative PCR was performed in an iQ5 Multi-colour Real-Time PCR detection system (Bio-Rad) using β-actin (acbt) as the housekeeping gene in a final volume of 15 μl per reaction well and with 100 ng of total RNA reverse transcribed to cDNA. Samples, housekeeping gene, cDNA template and reaction blanks were analysed in duplicate. Primer efficiency was tested with serial dilutions of a cDNA pool (1:5, 1:10, 1:100 and 1:1000). Sequences of the primers used in this study were: acbt (GenBank accession no. X89920) 5′–3′ (F) TCT GTC TGG ATC GGA GGC TC, (R) AAG CAT TTG CGG TGG ACG); and fads2 (GenBank accession no. AY055749) 5′–3′ (F) CGA GAG CCA CAG CAG CAG GGA, (R) CGG CCT GCG CCT GAG CAG TT). Gene, primer efficiency and blank samples were analysed in 96-well PCR plates (Multiplate, Bio-Rad). Melting-curve analysis was performed, and amplification of a single product was confirmed after each run. Fold-change in expression of each gene was determined by the delta-delta CT method (2^ΔΔCT) (Livak and Schmittgen, 2001). PCR efficiencies were similar, and no efficiency correction was required (Livak and Schmittgen, 2001; Schmittgen and Livak, 2008).

Results are expressed as means±s.d. (n=3), unless otherwise stated in the tables and figures. The data were compared statistically using analysis of variance (ANOVA), at a significance level of 5%. All variables were checked for normality and homogeneity of variance using the Kolmogorov–Smirnoff and Levene tests, respectively (Sokal and Rohlf, 1969). If significant differences were detected with ANOVA, means were compared by Student’s t-test. All data were analysed using IBM SPSS v23.0.0.2 for Mac (IBM SPSS Inc., Chicago, IL, USA). A Pearson correlation test was performed to identify relationships between parameters using R (http://www.R-project.org/).

Individual broodstock fish showed very different expression levels of fads2 gene after being fed the experimental (conditioning) diet. (VM and VO diet; Table S1). The average fold-change in gene expression was 7.1±13.6, with a maximum of 61.6 and a minimum of 0.02. No relationship was found between fish mass and fads2 expression among individuals (P>0.05, R=0.0061). Mass and fads2 expression of the broodstock fish are presented in Table 1.

During the first part of the spawning season, when broodstock were fed with the same diet, there were no differences in total number of eggs per kg female per spawn, fertilization rate, viability, hatching rate and survival of larvae at 3 dph, which on average were 49,023±1195, 46,339±1862, 33,990±2513, 31,899±3092, 22,960±2114 eggs/larvae (10³/kg female/spawn) (means±s.d., P>0.05), respectively. After feeding the broodstock with either the F or the V diet (Table 2) for 1 month, the total number of eggs produced and survival of 3 dph larvae were lower in broodstock with low fads2 expression (FLD and VLD), regardless of the diet. The difference between the two groups was 37% lower total number of eggs and 10.2% lower number of 3 dph larvae in low fads2 expression groups (FLD and VLD) than in those with high fads2 expression (FHD and VHD) (P<0.05) (Fig. 2). The same trend was observed for the other spawning quality parameters including fertilized, viable or hatched eggs, but without significant differences among groups (P>0.05) (Fig. 2). The percentage ratios of fertilized eggs after feeding broodstock with the F and V diet was 92.6±7.3, 91.8±7.6, 89.3±7.3 and 89.9±9.5 in FHD, FLD, VHD and VLD groups, respectively. The proximate composition of these eggs, obtained after 1 month of broodstock feeding with the diets, was not significantly different among the different broodstock groups (P>0.05) (Table 1). fads2 expression levels, diet and the interaction between these parameters did not affect the proximate composition of the eggs (P>0.05) (Table 1). Irrespective of the fads2 expression levels of the broodfish, replacement of FO by VO in the diet led to significantly increased levels of ALA and LA in the eggs (P>0.05) (Table 4). Analysis of egg fatty acid composition (Table 4) showed that among saturated fatty acids, 14:0 and 17:0 were lower in V diet groups (P<0.05), while the other saturated fatty acids such as 15:0, 16:0, 18:0 and 20:0 were similar among all groups (P>0.05) (Table 4). Additionally, egg content of monoenoic fatty acids such as 14:1n-7, 14:1n-5, 16:1n-7, 18:1n-7, 18:1n-5, 20:1n-9, 20:1n-7, 20:1n-5 and 20:1n-11 was lower in V-diet groups (Table 4). Also, ALA, the first substrate for delta-6 in the n-3 LC-PUFA biosynthesis pathway to EPA and DHA, was up to 4 times higher (P<0.05) in eggs obtained from V-diet groups (Table 4). In contrast, egg content of LA, another substrate for LC-PUFA biosynthesis and high in VOs, was similar among eggs from different groups (P>0.05; Table 4). However, the products of n-3 LC-PUFA biosynthesis such as arachidonic acid (ARA) 20:4n-6, EPA 20:5n-3 and DHA 22:6n-3 were similar among the eggs of different experimental groups (P>0.05) (Table 4).

Despite the fact that larvae were kept under similar conditions and fed with the same diets, growth was higher in FHD larvae in comparison with VHD larvae at 3 dph (P<0.05; Table 5). At 15 dph, among the progeny obtained from broodstock fed the F diet, those from the FLD group had a greater length than those from the FHD group (P<0.05). Between the larval groups from broodstock with low fads2 expression, VLD larvae were smaller in comparison to FLD larvae (P<0.05). There were no differences in larval growth among any of the experimental groups at 30 dph (P>0.05; Table 5).

There was no significant difference (P>0.05) in the initial mass of the 6 month old juveniles at the beginning of the nutritional challenge test (FHD: 23.4±0.1, FLD: 22.1±0.1, VHD: 23.8±0.4 and VLD: 23.7±0.1g). However, after 60 days of feeding juveniles with the high VM and VO diet (Table 3), the FHD group reached the highest final mass (47.4±0.8 g) and had the highest relative gain in mass (99.7±7.3%) (Fig. 3). The two-way ANOVA analysis showed that broodstock with high fads2 expression levels (FHD and VHD) had a positive effect on juvenile offspring gain in mass (P<0.001), a nutritional programming effect of the broodstock diet (P<0.01), and a significant interaction between these two parameters (P>0.05) (Fig. 3). Thus, the parents showing high fads2 levels improved the gain in mass of the offspring, regardless of the nutritional history of the broodstock (F or V diet) (Fig. 3). Progeny obtained from broodstock selected for high fads2 expression and fed with the F diet (FHD) showed significantly higher (P<0.01) growth than progeny obtained from broodstock selected for high fads2 expression and fed with the V diet (VHD) (Fig. 3). Gain in mass of the progeny obtained from broodstock selected for low fads2 expression and fed with the F diet (FLD) or the V diet (VLD) were not significantly different (P>0.05) (Fig. 3). In summary, VO inclusion in the broodstock diets, thus increasing the dietary 18C fatty acid levels, resulted in lower gain in mass in juvenile offspring of broodstock selected for high fads2 (P<0.05) (Fig. 3), but not in those coming from low fads2 broodstock. There was no difference (P>0.05) in feed intake of the fish. Additionally, FCR in juveniles from broodstock selected for high fads2 expression (FHD and VHD groups) was significantly lower than that in low fads2 expression groups (FLD and VLD groups) (P<0.05) regardless of the diet fed to their parents (Fig. 3). In contrast, gain in mass (%) values were significantly higher in progeny from broodstock showing low fads2 expression, regardless of the broodstock diet (P<0.05) (Fig. 3).

Muscle biochemical composition was similar among the experimental groups (P<0.05; Table 6). The type of broodstock diet (F or V) as well as the fads2 expression levels of the parents and their combination led to significant (P<0.05) differences in the muscle fatty acid profiles of offspring; this was particularly relevant for total saturated, monounsaturated and n-6 PUFA levels, whereas total n-3 PUFA contents were similar (Table 7). Total saturated fatty acid levels were higher (P<0.05) in the VLD group than in the VHD group, while the VHD group had higher levels of saturated fatty acids than the FHD group (P<0.05). Among the saturated fatty acids, 14:0 and 16:0 were higher in the VLD group than in the VHD and FLD groups (P<0.05), thus the two-way ANOVA showed a significant effect of broodstock fads2 expression levels (P<0.05), broodstock diet (P<0.01) and the interaction between these parameters (P<0.01) (Table 7). In addition, if LD groups were compared, 18:0 was expressed at a lower level (P<0.05) in VLD than in FLD juveniles (Table 7). Total monounsaturated fatty acids were affected by the diet in LD groups and were higher in the VLD than in the FLD group (P<0.05). n-6 PUFA were changed in response to the broodstock diet (P<0.05) and the interaction between broodstock diet and broodfish fads2 expression levels (P<0.1). LA was higher in the VLD group if compared with the FLD group (P<0.05); however, there was a clear tendency of LA accumulation in progeny from V-diet fed broodstock than in those from the F-diet fed broodstock (Table 7). The same kind of differences were observed in ALA, with a broodstock diet leading to greater accumulation of ALA in the V-diet groups than in those coming from the F-diet broodstock, while the accumulation was significantly higher in the VLD than in the FLD group (P<0.05). However, there were no differences (P>0.05) in EPA, DHA or total n-3 PUFA content among the experimental groups (Table 7). In addition, there were no differences in total PUFA content of the progeny obtained from broodstock with different fads2 expression levels and fed with either the V or F diet (P>0.05).

The importance of the fads2 gene for reproduction in mammals has been demonstrated in Fads2 −/− mouse, where males are not able to produce mature sperm and folliculogenesis is disrupted in females (Stoffel et al., 2008). Knocking out the Fads2 gene inhibits the synthesis of LC-PUFA, and other desaturases (Scd1–5, Fads1, Fads3) are not able to compensate for the deletion of Fads2 in the biosynthesis of ARA, EPA and DHA. In the present study, the broodstock group with higher fads2 expression showed a better reproductive performance than those with low fads2 expression, as denoted by the greater total number of eggs produced per kg female per spawn. To the best of our knowledge, this is the first study showing the positive effects of broodstock fads2 expression levels on spawning quality in fish. The importance of LC-PUFA on fish reproductive success is well documented in a number of teleosts (Izquierdo et al., 2001; Watanabe and Vassallo-Agius, 2003), including gilthead seabream (Fernandez Palacios et al., 2011; Fernández-Palacios et al., 1995; Izquierdo et al., 2001). In addition to these findings, the present study demonstrates the importance of the broodstock's ability to express fads2 on reproductive success, in agreement with a study conducted in mammals (Stoffel et al., 2008).

Feeding broodstock with the F diet resulted in a greater length of larvae at 3 dph among larvae obtained from parents showing high fads2 expression and greater growth in 15 dph larvae obtained from parents showing low fads2 expression. By 30 dph, larvae were able to compensate for early differences in growth, indicating that the common rearing protocols supplied sufficient amounts of nutrients for growth. It has been shown in several species of fish that parental nutritional history and dietary interventions during early ontogenesis can have significant impacts on offspring development, somatic mass, metabolism and immune responses (Izquierdo et al., 2015; Morais et al., 2014; Turkmen et al., 2017b). In the present study, growth differences at 3 and 15 dph may be related to the genetic background of the eggs in relation to the broodstock expression of fads2, as rearing with the same feeding protocol allowed the different larval groups to achieve similar growth performance at 30 dph. Interestingly, when juvenile offspring were challenged with a low FM and low FO diet, those obtained from broodstock with high fads2 expression again showed a positive effect on gain in mass, feed gain and FCR, although the values of the last were slightly high as may be expected from a fish fed a pelleted rather than a extruded diet (Izquierdo et al., 2003; Torno et al., 2018). The use of fads2 gene expression as a selection criterion in broodfish can indicate modifications in juvenile offspring metabolism and improves their ability to cope with high VM VO dietary levels. Recent studies in gilthead seabream show a positive and significant correlation between fads2 expression levels in the peripheral blood cells and liver of broodstock (Ferosekhan et al., 2020). In mammals, a single nucleotide polymorphism in the Fads2 gene occurs if mice are selected for low or high basal metabolism (Czajkowska et al., 2019). In fish, recent studies have shown that one of the underlying mechanisms of these alterations in metabolism appears to be caused by epigenetic modifications such as methylation of the promoter region of fad2 in gilthead seabream (Perera et al., 2019; Turkmen et al., 2019b).

We have previously demonstrated that it is possible to nutritionally programme gilthead seabream offspring by increasing ALA and LA, fatty acid precursors of LC-PUFA biosynthesis, and reducing EPA and DHA, products of the synthesis, in the broodstock diet (Izquierdo et al., 2015; Turkmen et al., 2017b). However, the effect of increasing the dietary supply of precursors without reducing the product fatty acids had not been investigated. In the present study, the broodstock diets were formulated based on previous studies in which four different diets with increasing levels of FO replacement by VO (0%, 60%, 80% and 100% VO) were tested (Izquierdo et al., 2015; Turkmen et al., 2017b). In these studies, 80% replacement of FO by VO had adverse effects on spawning quality parameters, while fish fed diets with 60% replacement of FO performed equally well as those fed a 100% FO diet. In agreement with these findings, in the present study, a 70% replacement of FO by VO did not negatively affect spawning quality and only slightly modified egg fatty acid profile. For instance, FO replacement by VO in the broodstock diet led to an increase in the egg content of saturated fatty acids, particularly 14:0 and 16:0, direct products of lipid biosynthesis, in agreement with the slightly higher lipid content found in these eggs. Regarding egg LC-PUFA content, whereas FO replacement by VO led to a 15% reduction in DHA in the broodstock diet, DHA content in the egg not only was not reduced but also was slightly increased by 2%. DHA, oleic acid (18:1n-9) and 16:0 are the main fatty acids in various fish eggs and, together with EPA and ARA, are recognized as the most important fatty acids during larval development (Izquierdo, 1996). Additionally, whereas replacement of FO by VO in the broodstock diet lead to a 16 times increase in ALA in the diet, ALA content in the eggs only increased 4 times. ALA is not a major fatty acid component of fish eggs; thus, low retention of this fatty acid may be related to selective retention of the essential fatty acids such as DHA. In vitro studies in pig oocytes showed that ALA supplement may enhance the nuclear maturation of oocyte and embryo development; however, excessive ALA could have a negative influence by altering the oxidative status of the oocytes (Lee et al., 2017). Dietary ALA is related to increased peroxidation risk. For instance, inclusion of LO, high in ALA, in place of FO in the diet of juvenile gilthead seabream raises basal and post-acute stress cortisol levels (Ganga et al., 2011). Moreover, excessive dietary ALA levels may cause the displacement of other n-3 PUFA from phospholipids, particularly from the second position of the phospholipids (Izquierdo, 2005; Izquierdo et al., 2000). Indeed, increased FO replacement by LO leads to a deposition of ALA in seabream organs, and ALA content in head kidney is related to increased plasma cortisol levels (Ganga et al., 2011).

In juveniles, replacement of FO by VO in broodstock diets also increased the ALA content in the muscle in comparison to that in juveniles obtained from broodstock fed FO, even though all the juveniles were fed the same low FM and low FO diet. VO inclusion in the broodstock diet also increased muscle saturated fatty acids, particularly 16:0, and monounsaturated fatty acids, together with 18:2n-6, while reducing other PUFA. These significant changes in the fatty acid profiles in juveniles obtained from broodstock fed different feeds, despite the fact that all juveniles were fed the same diet, indicate persistent changes in lipid metabolism by nutritional programming initiated through broodstock nutrition. These results are in agreement with the previous findings that demonstrate that it is possible to nutritionally programme gilthead seabream offspring by modifying the fatty acid profiles of the broodstock diet (Izquierdo et al., 2015; Turkmen et al., 2017b). However, in the present study, FO replacement by LO negatively affected juvenile offspring growth and feed utilization, contrary to the improvement found in previous studies (Izquierdo et al., 2015; Turkmen et al., 2017b). A main difference between those studies and the present one is in the fatty acid profiles of the broodstock diets.

Comparison of broodstock diets from those previous studies (Izquierdo et al., 2015; Turkmen et al., 2017b) shows the ALA content of the VO diet (V diet) was 20 times higher than in the FO diet (F diet), and replacement of FO with VO caused a 40% decrease in n-3 LC-PUFA products. Feeding broodstock during the spawning period with a high ALA content diet in combination with the decrease in n-3 LC-PUFA products altered the lipid metabolism of the progeny, which led to a 20–30% increase of tissue n-3 LC-PUFA content and higher growth in 4 and 16 month old juveniles (Izquierdo et al., 2015; Turkmen et al., 2017b). However, in the present trial, because of the contribution of n-3 LC-PUFA of FM, although the level of ALA was 16 times higher in the V diet, LC-PUFA levels were only reduced by half in comparison to previous studies (Izquierdo et al., 2015; Turkmen et al., 2017b). Comparison of offspring fatty acid composition shows n-3 LC-PUFA content was not significantly different (∼5% change) and juveniles’ growth was also reduced. These results suggest the important role of LC-PUFA dietary levels for nutritional programming of gilthead seabream offspring through broodstock feeding.

Indeed, dietary LC-PUFA are strong modulators of lipid metabolism in seabream and their ability to synthesize LC-PUFA (Izquierdo et al., 2008), as has also been seen in rodents (Gibson et al., 2013). The results are in agreement with the long-lasting effects of a reduction of both LC-PUFA precursors and products in first feeding diets for Atlantic salmon on lipid metabolism at the juvenile stage (Clarkson et al., 2017; Vera et al., 2017).

In summary, this study confirms modifying the fatty acids in the broodstock diet causes metabolic changes in offspring of gilthead seabream, and it also points to the importance of adequate LC-PUFA levels in parental diets to enhance the offspring ability to synthesize LC-PUFA and promote juvenile growth. Additionally, a major finding of the present study was that the broodstock's ability to biosynthesize fatty acids as denoted by their fads2 expression enhances reproductive performance. Similarly, broodstock fads2 expression levels affect offspring larval and juvenile growth rates, suggesting an advantage for offspring performance if fed with a high VM VO diet. Further studies are being conducted to understand the underlying physiological and molecular mechanisms involved in nutritional programming of gilthead seabream.

The authors wish to thank Dr Monica Betancor for her suggestions on the analysis of fatty acids.