Reproductive performance of gilthead seabream (Sparus aurata) broodstock showing different expression of fatty acyl desaturase 2 and fed two dietary fatty acid profiles

Previous studies have shown that it is possible to nutritionally program gilthead seabream offspring through fish oil (FO) replacement by vegetable oils (VO) in the broodstock diet, to improve their ability to grow fast when fed low fish meal (FM) and FO diets during grow-out phase. However, in those studies broodstock performance was reduced by the VO contained diet. Therefore, the present study aimed to determine if it is possible to replace FO by a mixture of FO and rapeseed oil (RO) with a specific fatty acid profile in broodstock diets, without altering gilthead seabream broodstock reproductive performance. Besides, the study also aimed to evaluate the reproductive performance of broodstock with different expression of fatty acid desaturase 2 gene (fads2) a key enzyme in synthesis of long chain polyunsaturated fatty acids. For that purpose, broodfish having either a high (HD) or low (LD) expression of fads2 were fed for three months during the spawning season with two diets containing different fatty acid profiles and their effects on reproductive hormones, fecundity, sperm and egg quality, egg biochemical composition and fads2 expression were studied. The results showed that blood fads2 expression in females, which tended to be higher than in males, was positively related to plasma 17β-estradiol levels. Moreover, broodstock with high blood fads2 expression showed a better reproductive performance, in terms of fecundity and sperm and egg quality, which was correlated with female fads2 expression. Our data also showed that it is feasible to reduce ARA, EPA and DHA down to 0.43, 6.6 and 8.4% total fatty acids, respectively, in broodstock diets designed to induce nutritional programming effects in the offspring without adverse effects on spawning quality. Further studies are being conducted to test the offspring with low FM and FO diets along life span.

. Body weight (kg) of gilthead seabream male (n = 6) and female (n = 3) broodstock selected based on blood cells fads2 expression at the end of Phase I. Different superscripts in a line indicate significant differences among broodfish groups for a given parameter (P < 0.05, one-way ANOVA, Tukey Post-Hoc test).  www.nature.com/scientificreports/ 11-ketotestosterone or 17β-estradiol levels among the different broodstock groups ( Table 2). The highest 11-ketotestosterone levels were found in broodstock with the smallest weight and a significant negative linear regression relation was found between broodstock body weight and 11-ketotestosterone levels (R 2 = 0.517; P < 0.001) (Fig. S2a). The opposite relation was found with the 17β-estradiol levels that showed a significant positive exponential relation with the broodstock body weight (R 2 = 0.464; P < 0.001) (Fig. S2b). No relation was found between plasma steroid hormones and broodstock blood cells fads2 expression when all broodstock data were compared. However, in females with low fads2 expression, there was a positive linear regression relation but no significant difference between plasma 17β-estradiol levels and blood cells fads2 expression (R 2 = 0.502; P = 0.115) was observed.
Sperm quality. After one month of feeding the experimental conditioning diets, sperm concentration or motility duration were not influenced by the fads2 expression in the broodstock neither by the dietary fatty acids profile or their combination as denoted by the two-way ANOVA analysis (P > 0.05) ( Table 3). However, the percentage of sperm showing motility after activation with seawater was around 10% higher in broodstock with higher fads2 expression, indicating the effect of fads2 expression in broodstock on sperm motility (Table 3). Indeed, a highly significant positive linear regression relation was found between fads2 expression levels in broodstock males and sperm motility (R 2 = 0.70; P = 0.003) (Fig. 3). Besides, a positive linear regression relation but no significant difference was observed between the male testosterone levels and the sperm concentration (R 2 = 0.823; P = 0.093). No other relation was found between sperm quality and male hormones plasma levels.
Egg and larval quality. Before feeding the experimental conditioning diets neither broodstock quality nor egg biochemical composition differed among experimental broodstock (Table S1, S2 and S3). However, after feeding the experimental diets, fish fecundity, measured as mean number of eggs/spawn/kg female, was the highest in HDFO broodstock and the lowest in LDRO broodstock ( Fig. 4; Table 4). Thus, the two-way ANOVA analysis showed a significant (P < 0.05) improvement in fecundity in broodstock with higher fads2 expression, whereas feeding RO did not significantly (P > 0.05) affected broodstock fecundity. Other spawning quality parameters such as fertilization, hatching or larval survival rates were not affected by neither broodstock fads2 expression nor by the broodstock conditioning diet or their combination (Table 4). Only egg viability rate was slightly improved in broodstock with higher fads2 expression (P = 0.07) (Table 4). Besides, a significant linear relation (R 2 = 0.397; P = 0.05) was found between sperm motility and egg viability percentage (Fig. 5). A significant linear regression relationship was observed between female broodstock blood cells fads2 expression and all the spawning quality parameters (Fig. 6). No interaction between diet or broodstock fads2 expression was detected by the two-way ANOVA analysis (Table 4). As a consequence, the number of fertilized eggs/spawn/kg female (P < 0.05) Table 2. Plasma steroid hormone levels of male (n = 12) and female (n = 6) gilthead seabream broodstock with high (HD) or low (LD) fads2 expression fed with either FO or RO experimental diets during Phase-III. Different superscripts in a line would indicate significant differences among broodfish groups for a given parameter (P < 0.05, one-way ANOVA, Tukey Post-Hoc).  www.nature.com/scientificreports/   www.nature.com/scientificreports/ and particularly, the numbers of viable eggs/spawn/kg female, hatched larvae/spawn/kg female or larval survival 3 dph/spawn/kg female were the highest in HDFO broodstock and the lowest in LDRO (Fig. 4).
Molecular studies. The study of absolute fads2 expression (mRNA copies/µl) in eggs from the different broodstock showed values that in average were 3 times higher for broodstock fed RO diet than for those fed FO (Fig. 8). However, due to the large standard deviations, there were not significant (P > 0.05) differences in fads2 expression according to both the one-way or two-way ANOVA analysis.

Discussion
Marine fish have a limited ability to synthesise LC-PUFAs, due to insufficient expression of key genes such as fads2 and the inhibition of desaturase enzymes by dietary LC-PUFA 11,18,[36][37][38] . Therefore, large FO replacement by VO in broodstock diets may lead to extremely low dietary LC-PUFA levels and markedly reduce broodstock performance 33,39 . Optimizing FO replacement by VO is desired to induce nutritional programming in the offspring for a better use of low FM and low FO diets 32,33 . Although, it has been found that there are large variations in the fads2 expression showed by different gilthead seabream individuals, there is little information on its relation to fish reproductive performance 29 . The present study thus aimed to determine the reproductive success of gilthead seabream broodstock with different bloods fads2 expression fed diets containing different types of oils. Plasma 17β-estradiol and 11-ketotestosterone levels showed respectively positive and negative correlations to the body weight of broodstock GSB, as is expected from a protandric hermaphrodite species that turns from male to female as it ages 40 . The lack of relation between body weight and plasma testosterone levels as can be attributed to the fact that this hormone is an intermediate metabolite in most protandric fish 41 . Broodstock body weight had no effect on fads2 expression, in males or females, nor with any parameter regarding spawning quality. The higher level of expression of fads2 (30% more in females than in the males) and the significant positive correlation between plasma 17β-estradiol levels and the fads2 expression in females from the low fads2 expression www.nature.com/scientificreports/ groups are noteworthy. These results are in agreement with the increased conversion of LNA to n-3 LC-PUFA promoted by oestrogen in women, claimed to be important to fulfil the essential fatty acid requirements of the foetus and the neonate 42 . In fish, n-3 LC-PUFA are also highly demanded during embryogenesis and larval development to sustain growth, normal development and neural and sensorial organs formation, with EPA and particularly DHA being the major fatty acids found in marine fish eggs and larvae 9,43 . To our knowledge, this is the first study that finds a significant correlation between plasma oestrogen levels and fads2 expression in blood cells of female fish, as observed in pregnant rats 34,44 . Indeed, treatments with 17β-estradiol have shown to increase fads2 expression in female rats 45 . This higher fads2 expression in females seems to be a response to the higher level of requirement of n-3 LC-PUFA fatty acid for the normal gonadal, oocyte and larval development through vitellogenetic processes in different teleosts 9,14,33,43,46,47 . Interestingly, in females showing a very high expression of fads2 such a strong correlation was not found, despite a general trend to increased fads2 expression in females with higher oestrogen levels. These results suggest the influence of a different genetic or epigenetic background between both types of females showing very high or low fads2 expression. Very early, it was reported that stenohaline marine teleosts have lower LC-PUFA bioconversion capacities compared to freshwater teleosts 48 . More recent work found that freshwater fish to possess greater number of copies of Fads2, compared to marine fish and that the Fads2 was a key metabolic gene for overcoming some of the nutritional constraints associated with freshwater environment 49 . We observed that there was a very large variation in the individual fads2 expression values, for males the maximum value being 47 times higher than the minimum one and for females 30 times, in agreement with the large variation found in our previous studies 29 . From the 215 broodfish analysed for fads2 expression, about 10% of the population had fads2 expression values that were higher than 5 mRNA copies μL −1 . In humans, FADS2 expression and delta-6 desaturase activity is linked to different FADS2 genotypes and, in turn to lower DNA methylation in specific CpG sites of the FADS2 promoter 50 . In agreement with human studies, in gilthead seabream, there is an increased methylation of specific CpG sites in the promoter region of this gene in offspring from broodstock with low fads2 expression 29 . Moreover, LC-PUFA biosynthesis ability is affected by single nucleotide polymorphisms (SNPs) in the fatty acid desaturases 51,52 . Further studies are under way to elucidate the genetic and epigenetic mechanisms regulating fads2 expression and LC-PUFA biosynthesis in gilthead seabream.
The improved reproductive performance in terms of sperm and egg quality in as observed here in broodstock with high blood fads2 expression clearly reflects the importance of the end product(s) of fads2 in fish reproduction. Sperm quality, in terms of motility and concentration, has been found to depend on its content in n-3 LC-PUFA, particularly DHA, in gilthead seabream after cryopreservation 53 , as well as in rainbow trout (Oncorhynchus mykiss) 54 , European seabass (Dicentrarchus labrax) 55 , Senegalese sole (Solea senegalensis) 56 or European eel (Anguilla Anguilla) 57 fed different amounts n-3 LC-PUFA. Therefore, a higher fads2 expression in gilthead seabream broodstock would allow a higher n-3 LC-PUFA synthesis to promote a higher incorporation of these fatty acids into sperm that would have improved sperm motility, affecting in turn egg viability. Besides, testis fatty acids profile may also affect seminal plasma, which maintains sperm cells in a quiescent state required to achieve motility 54,58 . Finally, mice and human sperm also have high contents in DHA, which also acts as a precursor of very long chain LC-PUFAs with 26-32 carbons that form sphingolipids in spermatozoa and have been related to sperm quality 59 .
As regards seabream females where those with a higher fads2 expression exhibited 20% higher fecundity in terms of eggs and larvae produced per kg female per spawn, this can also be linked to the production of n-3 LC-PUFA. Indeed, the DHA content in the eggs was effectively increased in females with a high fads2 expression, denoting an efficient n-3 LC-PUFA biosynthesis in these females in comparison to those with low fads2 expression, regardless of the diet fed. Interestingly, among the different LC-PUFA, only those from the n-3 series and with 22 carbons were significantly increased, despite the higher dietary content of 18:2n-6 (5.5-10.6% total fatty acid, TFA) in comparison to 18:3n-3 (1.2-2.7% TFA). Therefore, n-3/n-6 and DHA/ARA ratios were www.nature.com/scientificreports/ increased and 22:4n-6 reduced in females with a high fads2 expression, suggesting the preference of the enzymatic complexes involved in fatty acid deposition in eggs for the n-3 LC-PUFA. The higher DHA/EPA and 22:5n-3 (n-3 DPA) contents in the eggs also suggest the activation of the Sprecher pathway, since fish Fads2 have a Δ6 desaturase activity on 24:5n-3, produced by elongation from 22:5n-3, to synthesise DHA after beta-oxidation from 24:6n-3 60 . Such high DHA and n-3 DPA contents in the egg may also be related to an increased mobilization of DHA to ovaries caused by 17β-estradiol, to a lower beta-oxidation of these fatty acids or to morphological changes in the ovaries. In turn, DHA increase in females would lead to an increased production of docosanoids, which play an important role in the induction of oocyte maturation 61 , improving fecundity in terms of eggs produced. Our observation of a direct positive correlation between n-3 LC-PUFA contents in the egg and egg viability and larval survival confirms that these fatty acids are determinant of embryo and larval development 14,43,46,47 .
That the dietary fatty acid profile is reflected in tissue fatty acid profile is well established and our data show that broodstock fed with a diet rich in rapeseed oil led to increased levels of 18C fatty acids and relatively reduced levels of LC-PUFAs in egg lipids. These changes were however mild, where the 18C fatty acids increased by an average 60% in comparison to those from broodstock fed FO and the LC-PUFA were reduced by 30%, while in the RO diet these values were 200% and 20%, respectively in comparison to fish oil based diet. This lower accumulation of 18C fatty acids together with the relatively lower reduction in LC-PUFA in the eggs, in comparison to the diet, suggests an increased LC-PUFA biosynthesis ability in conformity with the trend for an up-regulation of fads2 expression found in the eggs of broodstock fed RO.
Broodstock diets replacement of FO by VO, namely reduction of dietary LC-PUFA and increase in 18:2n-6 and 18:3n-3, allows the conditioning of gilthead seabream offspring to produce juveniles which can potentially utilize low FM and FO diets better and grow fast 29,[31][32][33] . Total replacement of FO by vegetable oils markedly reduces reproductive performance of gilthead seabream 32,33 , since LC-PUFA are essential for reproduction of this species 30,62 . The partial FO replacement by RO as done here, giving dietary levels for LA + ALA:n-3 LC-PUFA ratio levels of 16:1:16.3 did not negatively affect sperm quality or spawning quality parameters. These results suggest that the LC-PUFA levels in diet RO were able to match the minimum requirements for gilthead seabream broodstock, in agreement with previous studies 30,62 . Thus, the ARA and EPA contents in diet RO were similar and those of DHA even higher than the optimum dietary levels determined for gilthead seabream broodstock 30 . The levels of these LC-PUFA in the present study were also higher than those in broodstock diets used to induce nutritional programming in gilthead seabream, which caused a reduction in all spawning quality parameters 32,33 .
In summary, the results showed that blood fads2 expression in gilthead seabream broodstock females, which tended to be higher than in males, was positively related to plasma oestrogen levels. Moreover, broodstock with high blood fads2 expression showed a better reproductive performance, in terms of fecundity and sperm and egg quality, which was correlated with female fads2 expression. Besides, the present study has demonstrated that it is feasible to reduce ARA, EPA and DHA down to 0.4, 6.6 and 8.4% of total fatty acids, respectively, without affecting spawning quality, in broodstock diets designed to induce nutritional programming effects in the offspring. Further studies are being conducted to test the offspring with low FM and FO diets along life span.

Characterization of broodstock with high or low fads2 expression (Phase-I).
The broodstock utilized were obtained as part of a series of long-term ongoing selection programmes involving multiple criteria 63,64 . They were reared right from larval stages in our own research facilities of the ECOAQUA Institute. In order to identify broodstock with different ability to synthesize LC-PUFA from LA and ALA, seventy one 2-year old males (1.02 ± 0.38 kg body weight) and one hundred fourteen 4-year old females (2.07 ± 0.39 kg body weight) were individually tagged with PIT tags (EID Iberica SA-TROVAN, Madrid, Spain) and maintained in a 40 m 3 (5 × 2.35 m) circular tank. The tanks were supplied with seawater (37 g l −1 salinity, 17.8-19.0 °C) at a water exchange of 600% daily and maintained under natural photoperiod. Three months before the spawning season, broodfish were fed twice a day for one month with a low FM (5%) and FO (3%) diet (Low FM/FO diet) 15 , high in LA and ALA (Table S4 and S5) and low in LC-PUFA to induce the up-regulation of the fatty acid desaturases 2 gene (fads2). After that period, blood samples were collected from all broodfish (n = 185) and centrifuged at 3000×g for 10 min to separate blood cells and plasma. The blood cell samples were stored at − 80 °C until RNA extraction. Seabream broodfish were divided into two categories, namely high (HD) or low (LD), based on their fads2 mRNA copy numbers per µl in blood cells. So, eighteen fish with the highest fads2 expression values and eighteen with the lowest ones were selected for the conditioning trial (Phase III). Blood cell fads2 gene expression level was analysed by droplet digital polymerase chain reaction (ddPCR) as previously described 8,29 . Comparison of broodstock quality before feeding the conditioning diets (Phase-II). The selected males and females from HD and LD broodfish were stocked in twelve (6 HD and 6 LD) 1,000 L fiberglass tanks with a sex ratio of 2 males to 1 female. Broodstock tanks were supplied with 16L min −1 filtered seawater (37 ± 0.5‰ salinity) and strong aeration. At the beginning of the spawning season, from 08 January 2018 to 07 February 2018, fish were fed with a commercial diet (Europa Turbot 18, Skretting, Burgos, Spain) (Tables S6 and S5) to ensure that there were no significant differences in the spawning quality among broodfish from the same category (HD and LD). For the evaluation of spawning quality, the spontaneously spawned eggs from each experimental broodstock group were collected four times per week, following this procedure 8,30,33 . Eggs were also collected at the end of the feeding period and kept at − 80 °C until biochemical analysis. www.nature.com/scientificreports/ (FO) or a mixture of 20% fish oil (FO) and 80% rapeseed oil (RO) and were produced by Skretting ARC (Stavanger, Norway) (Tables S6 and S5). Compared to the FO diet, the RO diet had higher levels of 18:2n-6 and 18:3n-3 fatty acids and reduced levels of saturated, monoenoic and n-3 LC-PUFA (20:5n-3; eicosapentaenoic acid, EPA and 22:6n-3; docosahexaenoic acid, DHA) (Table S5). Fish were fed two times a day (9:00 and 14:00 h) at 1% of their estimated total biomass. Seawater temperature during broodstock spawning period was in the range of 18-22 °C (January-April 2018) and fish were kept under natural photoperiod (12 h light). Egg collection for spawning quality and biochemical composition followed the same protocol described in Phase II. Finally, after 30 days of feeding the two different experimental diets, eggs were collected from all broodfish groups (HDFO, HDRO, LDFO, LDRO) and conserved in 1,000 μl of RNA Later (Sigma-Aldrich) overnight at 4 °C, and then samples were kept at − 80 °C until RNA extraction.

Broodstock nutritional conditioning (Phase-III
Plasma steroid hormones. At the end of Phase III, all the GSB broodstock were fasted overnight and anesthetized with clove oil (10 ppm clove oil:methanol (50:50) in sea water) to collect blood samples. Blood was taken from the caudal vein using sterile syringes (Terumo Europe NV, Leuven, Belgium) and transferred to 3.0 mL K3-EDTA tubes (L.P. Italiana, Milan, Italy). Whole blood samples were centrifuged at 3,000g for 10 min at 4 °C and plasma was separated and stored at − 80 °C for sex steroid hormone analyses. Plasma sex steroids were measured by enzyme immunoassays (EIA) as described for European sea bass for testosterone (T) 65 , 11-ketotestosterone (11-KT) 66 and 17β-estradiol (E2) 67 and later validated for seabream 16  Sperm quality. At the end of Phase III, all the male broodfish were anesthetized as mentioned above and sperm was collected from the blot dried genital pore after a gentle abdominal massage to induce spermiation and taking care to avoid contamination with water, faeces or urine. The collected sperm was stored on ice until transferred to a 4 °C refrigerator. The sperm quality parameters that were evaluated included sperm concentration (number of spermatozoa/ml sperm, 10 9 ml −1 ), spermatocrit percentage, sperm motility percentage (percentage of spermatozoa showing forward motility) and sperm motility duration (Seconds). Sperm concentration was estimated after a 1,000-fold dilution with sperm inactivation media using a Neubauer haematocytometer under 400× magnification. Sperm motility and motility duration were evaluated on a microscope slide (400× magnification) after mixing 1 µl of sperm with 50 μl of seawater 53,68,69 . Egg and larval quality. The collected eggs were placed in 5 l containers provided with aeration, from where 3 randomized 5 ml samples were taken and placed in a Bogorov chamber under the light microscope to calculate the total number of eggs and percentages of fertilized and viable eggs. Egg viability was determined by observing the percentage of morphologically normal eggs after 1-day post fertilization (1 dpf) 8,30 . Then, the viable eggs were individually placed in two replicates in 96-well microtiter plates filled with filtered and sterilized seawater. Eggs were incubated in a controlled temperature incubator at 19-21 °C, to estimate the percentage of hatching (2 dpf) and larval survival rates at 3 days post hatch (dph). From these values, the total numbers of fertilized, viable, hatched and larvae produced per kg female were calculated 8,30 .
Biochemical analysis. One month after feeding the commercial diet in Phase-II and one month after the experimental conditioning diets in Phase-III, egg samples were collected from all the broodstock groups and stored at − 80 °C for analysis of proximate and fatty acid composition. Crude protein content was determined by measuring the N content (N × 6.25) through automated Kjeldahl analysis 70 and crude lipid extraction was carried out with chloroform:methanol 71 . Fatty acids from total lipids were prepared by transmethylation 72 and separated by gas chromatography 73 and identified by comparison with previously characterized standards and GLC-MS (Polaris QTRACETM Ultra; Thermo Fisher Scientific). Moisture contents were obtained after drying the samples in an oven at 110 °C for 24 h and then for 1 h until constant weight. Ash content was determined after incineration at 600 °C for 16 h.
Molecular studies. Total RNA from blood cells (300-400 µl) (Phase I) and egg samples (60-70 mg) (Phase III) was extracted using the RNeasy Mini Kit (Qiagen) and homogenized using the Tissue Lyzer-II (Qiagen, Hilden, Germany) with TRI Reagent (Sigma-Aldrich). Samples were centrifuged with chloroform for phase separation (12,000g, 15 min, 4 °C). The upper aqueous phase containing RNA was mixed with 75% ethanol and transferred into the RNeasy spin column, where total RNA bound to a membrane and RW1 and RPE buffers (Qiagen) were used to wash away contaminants. Purified RNA was eluted with 50 μL of RNase-free water. The quality and quantity of RNA were analysed using the NanoDrop 1000 Spectrophotometer (Thermo Scientific, Wilmington, DE, USA). Synthesis of cDNA was undertaken using the iScript cDNA Synthesis Kit (Bio-Rad)