Effect of different levels of feed restriction and fish oil fatty acid supplementation on fat deposition by using different techniques, plasma levels and mRNA expression of several adipokines in broiler breeder hens

Background Reproductive hens are subjected to a restricted diet to limit the decline in fertility associated with change in body mass. However, endocrine and tissue responses to diet restriction need to be documented. Objective We evaluated the effect of different levels of feed restriction, with or without fish oil supplementation, on metabolic parameters and adipokine levels in plasma and metabolic tissues of reproductive hens. Methods We designed an in vivo protocol involving 4 groups of hens; RNS: restricted (Rt) unsupplemented, ANS: ad libitum (Ad, receiving an amount of feed 1.7 times greater than animals on the restricted diet) unsupplemented, RS: Rt supplemented, and AS: Ad supplemented. The fish oil supplement was used at 1% of the total diet composition. Results Hens fed with the Rt diet had a significantly (P < 0.0001) lower growth than Ad hens, while the fish oil supplementation had no effect on these parameters. Furthermore, the bioelectrical impedance analysis (BIA) and the fat ultrasonographic examinations produced similar results to the other methods that required animals to be killed (carcass analysis and weight of adipose tissue). In addition, the Rt diet significantly (P < 0.05) decreased plasma levels of triglycerides, phospholipids, glucose and ADIPOQ, and fish oil supplementation decreased plasma levels of RARRES2. We also showed a positive correlation between insulin values and ADIPOQ or NAMPT or RARRES2 values, and a negative correlation of fat percentage to RARRES2 values. Moreover, the effects of the Rt diet and fish oil supplementation on the mRNA expression depended on the factors tested and the hen age. Conclusions Rt diet and fish oil supplementation are able to modulate metabolic parameters and the expression of adipokines and their receptors in metabolic tissue.


Introduction
Growth performance of breeder hens has increased spectacularly over the past several decades, mainly due to genetic progress and improvements in nutrition and management strategies. Unfortunately, this high growth rate has been accompanied by increased body fat deposition and high incidence of metabolic, skeletal and reproductive disorders [1]. Moreover, breeders have few non-invasive tools for estimating body fat composition in their animals which comply with ethical and welfare requirements. Consequently, this accentuates the difficulties in detecting metabolic diseases which occur in the case of overfeeding. Restriction of the consumption of ad libitum feed during rearing [2] and production is a common practice for reducing metabolic disorders and improving productivity [3,4]. Indeed, improvements in metabolic deregulation and also in reproductive performance were shown in meat type chickens with restricted diets [5]. However, there is evidence that restricting feed intake leads to physiological stress, stereotypies, aggression, and other abnormal behaviour in poultry [6] [7]. Many studies comparing an ad libitum (Ad) diet versus a restricted (Rt) diet have shown the effects of Rt diet on welfare, growth, body composition and egg production [8] [9] [10]. Whilst much attention has been focused on the differences between restricted-and ad libitumfed birds, less is documented on the endocrine response to long-term feed-restriction programmes. Supplementation with omega-3 polyunsaturated fatty acids (PUFAs) of marine origin was shown to improve zootechnical performance, such as body weight, laying rate, egg quality and fertility, in chickens [11]. Indeed, PUFAs are essential for development and growth in mammals. Omega-3 PUFAs mostly recognised for their beneficial effects on the metabolic and reproductive system are eicosapentanenoic acid (EPA) and docosahexaenoic acid (DHA) [12]. They can be added to the diet or synthetised from endogenous linoleic acid by the animals [13]. However, the effects of supplementation with these PUFAs on growth, fattening, metabolic parameters and fertility in Rt broiler breeder hens have not yet been fully investigated.
The metabolic changes due to nutritional status (Rt diet or omega 3 PUFAs supplementation) partly lead to variations in the regulation of metabolic hormonal profiles. In chickens, fasting reduces circulating 3,3 0 ,5-triiodothyronine (T 3 ), insulin, and insulin-like growth factor-I (IGF-I) levels [14], whereas plasma levels of glucocorticoid, insulin-like growth factor-II (IGF-II) [15], and growth hormone are increased [16]. A well-known effect of feed restriction in broiler breeder is the reduction of adiposity. The link between diet and body composition involves a regulation of the hormones secreted by adipose tissue, called adipokines. The most studied adipokine in mammals is leptin, but its existence has been long debated in birds [17] [18]. However, in recent years, many adipokines have been discovered in poultry, including adiponectin C1Q and collagen domain containing (ADIPOQ) [19], retinoic acid receptor responder 2 (RARRES2) [20] and nicotinamide phosphoribosyl transferase (NAMPT) [21] [22]. In chickens, ADIPOQ is involved in multiple processes, such as appetite, adipocyte differentiation, lipid metabolism and steroidogenesis [23][24][25][26]. Furthermore, NAMPT is mostly considered to be a myokine rather than an adipokine [21] and nothing has been described about chicken RARRES2. In hens and mice, dietary restriction decreases plasma leptin levels [27] [28] and the addition of PUFAs of marine origin leads to high concentrations of adiponectin [28]. Addition of EPA and DHA in mice and humans led to higher secretion of ADIPOQ [28] [29] [30]; only DHA enhanced ADIPOQ mRNA expression [29], whereas EPA increased ADIPOQ protein expression [30]. The effects of EPA and DHA appear to be mediated by the peroxisome proliferator-activated receptor gamma (PPARG). Furthermore, another study demonstrated that in vivo administration of EPA has a direct stimulatory effect on NAMPT gene expression and protein secretion in rat primary visceral adipose tissue cells [31]. The majority of studies on the incorporation of these fatty acids into the diets of chickens focus on their ability to produce meat or eggs for consumption that are richer in omega 3 [32], but less of the studies focus on the regulation of the metabolism of the animal itself. However, the pattern of expression of the recent adipokines in hens can be key indicators to help predict body composition and thus support management decisions in order to optimise production.
Thus, the present study was carried out to evaluate the impact of feeding conditions on metabolic parameters and to explore the plasma and tissue levels of adipokines, considered to be potential indicators of body mass in broiler breeder hens. Therefore, we performed an in vivo protocol to evaluate the effects of different levels of dietary restriction associated or not with fish oil supplementation on the body weight, fattening estimated using non-invasive (bioimpedance, ultrasonography and X-ray scanning analysis) and invasive techniques (carcass composition analysis and weight of abdominal fat) and plasma and tissue concentrations of various metabolic markers including adipokines.

Ethical issues
All experimental procedures were performed in accordance with the French National Guidelines for the care and use of animals for research purposes (certificate of authorisation to experiment on living animals n˚01607.02, Ministry of Agriculture and Fish Products, and favourable notice of ethics committee of Val de Loire N˚19).

Animals
Three hundred and twenty broiler breeder female chicks (Cobb 500) from Hendrix Genetics (Saint Laurent de la Plaine, France) were studied from day 1 to 39 weeks of age. Animals were distributed on the ground on the day of hatching, in homogeneous groups of 10 birds in 32 pens, each pen with an area of 3 m 2 . The animals were reared at 'Pôle Expérimental Avicole de Tours' (INRA, Nouzilly, France) according to the traditional conditions of breeding: 14 h of light per day on arrival, followed by a gradual decrease until reaching 8 h of light per day at the time of laying (week 21), and then a gradual increase until reaching 15 h of light per day at the end of the study (week 39). Animals were killed by electrical stunning and bled out as recommended by the ethical committee.

Diets
From one to 28 days of age (week 4), female breeder chicks received a diet, called a starting diet, ad libitum (free access to food). At 28 days of age (week 4), animals were distributed into two treatment groups: the first group (n = 160 animals) received a restricted growing diet according to Hendrix Genetics recommendation; the second group (ad libitum group; n = 160 animals) received the same diet on a daily basis, but the amount was 1.7 times greater than in restricted animals. In order to adjust the amount of feed consumed by the animals, animals in two control pens (Rt) were weighed, compared and then feed adjusted weekly with respect to the theoretical curve provided by the supplier Cobb. From 63 days (week 9) to 273 days of age (week 39), the two treatment groups were each subdivided into two groups, one with fish oil supplementation and one without fish oil. The resulting four treatment groups were: group RNS (restricted unsupplemented); group ANS (ad libitum unsupplemented); group RS (restricted supplemented); group AS (ad libitum supplemented). During this period, these four groups of animals received three different diets (growing, before laying and during laying diets). The supplement was a protected encapsulated fish oil OMG750 provided by Kemin (Nantes, France) and was composed of refined fish oil (77%) and gelatin (capsule; 23%). The supplement was manually mixed into the diet at 1% of total diet (S1 Table and S2 Table and S1  Fig). Determination of the lipid composition of the fish oil supplement was performed by In Vivo Labs (Vannes, France) and is described in the S3 Table. Fatty acids profile in feed, egg yolk, adipose tissue, liver and muscle The total lipids were extracted from the different diets (starting, growing, before laying and laying), egg yolk (week 25), adipose tissue, liver and muscle (week 39) after homogenization of the samples (n = 10 for each condition) with a chloroform/methanol mixture. Lipids were extracted gravimetrically into methanol:chloroform (1volume: 2 volume) according to Folch et al. [33]. The fatty acid composition was determined by gas chromatography (Autosystem; Perkin Elmer, St Quentin en Yvelines, France) after transmethylation of lipids (Morrisson and Smith, 1964).

Determination of body weight, feed conversion and fattening
The fasted hens (n = 80 for each treatment) were weighed every 3 weeks using an automated balance from Grosseron (B146782-ENTRIS 8201i-1S, Coueron, France) with a precision of 0.1 g. Feed conversion was calculated as the ratio between the total feed intake during a fixed period (6 to 9 weeks, 9 to 18 weeks, 18 to 21 weeks or 23 to 39 weeks) and the total gain in body weight (6 to 9 weeks, 9 to 18 weeks and 18 to 21 weeks) or the total mass of eggs laid (23 to 39 weeks) during the same fixed period. The fat content of the chickens was estimated every 3 weeks by fat ultrasonographic examination (MyLab 30 Gold Vet,Hospimedi France, Saint-Crépin-Ibouvillers, France) and by the bioimpedance analysis (BIA, Quantum II-Body Composition Analyzer, RJL Systems, Clinton Township, Michigan, USA) (n = 80 animals for each treatment). The ultrasonographic examination determined (n = 80 animals per group the depth of adipose tissue on the back of animal and the BIA estimated percentage of fat in collaboration with INZO Laboratories (Argentan, France). The fat percentage of the whole carcass was estimated at 21 weeks (4 animals/group) and 39 weeks of age (4 animals/group). Briefly, animals were killed by cervical dislocation, feathers were removed and carcasses were frozen (−20˚C). Each frozen carcass was first cut into small pieces with an electric band saw and ground with an electric tabletop meat grinder. Ground carcasses were freeze-dried and individually analysed for lipids, with no acidic treatment prior to lipid extraction by petroleum ether [34]. Fat content was estimated at 39 weeks of age by dissection and weighing the abdominal adipose tissue (n = 15 animals per group) and by scanning (n = 12 per group). Scanning was performed using an X-ray Computerized Tomography (CT) scanner (Siemens Somatom Definition AS). The X-ray source was set at 100 kV and 120 mA. Five hundred images were acquired every 0.6 mm, with a pitch of 0.45. The images were reconstructed using a reconstruction filter Safire I26. Phantoms of known fat tissue were used to calibrate the scanning parameters for adipose tissue measurement and phantoms were reconstructed under the same parameters as the animals. Analyses of all data were carried out using an Acquisition Sinogram Image Processing IDL's virtual machine (ASIPro VM, Siemens Medical Solutions).

Adipokine assays
Plasma concentrations of adipokines (n = 8 for each condition) were obtained using chickenspecific kits: E12V0003 (sensitivity 1 ng/mL), E12A0125 (sensitivity 0.1 ng/mL) and E112C0104 (sensitivity 1 pg/mL) were used for NAMPT, ADIPOQ and RARRES2, respectively (Koln, Germany). The measurements were carried out according to the manufacturer's protocol with an intra-assay coefficient of variation < 6%. The absorbance was measured at 450 nm and then compared with reference values.

Measurement of the surface area of the pectoral and oyster muscles after slaughter
The surface area of the major and minor pectoralis muscles (Pecto) and the oysters (located in the hollow on the dorsal side of the ilium bone) were measured at 39 weeks of age (n = 4 animals/group) by ultrasonography using a MyLab 30 Gold Vet ultrasound scanner (Hospimedi France, Saint-Crépin-Ibouvillers, France) equipped with two linear probes (Esaote L332 and L435, Hospimedi France, Saint-Crépin-Ibouvillers, France). The scanner operated at low frequency for the pectoral muscles (5 MHz) and a higher frequency for the oysters (15 MHz).

mRNA expression of lipid metabolism factors, adipokines and their receptors in adipose tissue, liver and pectoralis muscle
Total RNA was extracted from the abdominal adipose tissue (AT abd), subcutaneous adipose tissue (AT sc), liver and Pecto from 39-week-old hens by homogenization in the TRIzol reagent using an Ultraturax, according to the manufacturer's recommendations (Invitrogen by Life Technologies, Villebon sur Yvette, France). The cDNA was generated by reverse transcription (RT) of total RNA (1 μg) in a mixture comprising 0.5 mM of each deoxyribonucleotide triphosphate (dATP, dGTP, dCTP and DTTP), 2 M of RT buffer, 15 μg/μL of oligodT, 0.125 U of ribonuclease inhibitor, and 0.05 U of Moloney murine leukemia virus reverse transcriptase (MMLV) for one hour at 37˚C. Real-time PCR was performed using the MyiQ Cycle device (Bio-Rad, Marnes-la-Coquette, France), in a mixture containing SYBR Green Supermix 1X reagent (Bio-Rad, Marnes la Coquette, France), 250 nM specific primers (Invitrogen by Life Technologies, Villebon sur Yvette, France) (S4 Table) and 5 μL of cDNA (diluted fivefold) for a total volume of 20 μL The samples were duplicated on the same plate and the following PCR procedure used: after an incubation of 2 min at 50˚C and a denaturation step of 10 min at 95˚C, samples were subjected to 40 cycles (30 s at 95˚C, 30 s at 60˚C and 30 s at 72˚C). The levels of expression of messenger RNA were standardised to three reference genes (RPL15, EF1 and β actin). For each gene, the relative abundance of transcription was determined by the calculation of e -ct The relative expression of the gene of interest was then related to the relative expression of the geometric mean of the three reference genes.

Statistical analysis
SAS software (version 9.3) was used for all analyses. A two-way ANOVA test in repeated measurements was used to compare the mean values for body weight, feed intake, fat content and plasma concentrations of biochemical parameters between different treatment groups from the beginning of the experiment to week 39.
The model used was: where Y ijkl is the dependent variable (body weight, feed intake, fat [ultrasound and BIA] and plasma biochemical parameters), μ is the overall mean, Diet i is the fixed effect of diet i (I = Rt, Ad), Supp j is the fixed effect of fish oil j (j = supplemented, unsupplemented), Wk k is the fixed effect of week k (k = 1, 2), Diet i Ã Supp j is the interaction between Diet i and Supp j , Diet i Ã Wk k is the interaction between Diet i and Wk k , and e ijkl is the residual error.
Then, for plasma hormone concentrations, we applied a similar model for different periods (week 3 to 9, week 9 to 18, week 18 to 21 and week 21 to 39) without week effect.
The model used was: where Y ijkl is the dependent variable (triglyceride, phospholipid, cholesterol, glucose, insulin, ADIPOQ, NAMPT and RARRES2), μ is the overall mean, Diet i is the fixed effect of diet i (I = Rt, Ad), Supp j is the fixed effect of fish oil j (j = supplemented, unsupplemented), Diet i Ã Supp j is the interaction between Diet i and Supp j , and e ijk is the residual error. At week 9, we only analysed the effect of the Rt diet, as animals were separated into only two groups (Rt and Ad animals) after blood sampling. For parameters measured only at a specific time point (feed conversion, fat content by carcass analysis and scanner, weight of abdominal adipose tissue and mRNA expression), we used a two way ANOVA. The results are represented as mean ± SEM, with P < 0.05 being considered significant.
A Pearson test was used for the correlation analyses between the different parameters and the correlation coefficient is noted as "r", with P < 0.05 being considered significant.

Effect of restricted diet and fish oil supplementation on fatty acid composition in egg yolk at 39 weeks
As shown in Table 1, the intake of fish oil resulted in a change in the fatty acid composition of egg yolk of hens. The fish oil supplementation (1% of total feed) used is sufficient to cause a significant increase of n-3 fatty acids (P <0.0001), including EPA (C20:5 n-3, P <0.0001), docosapentaenoic acid (DPA, C22:5 n-3, P <0.0001) and DHA (C22:6 n-3, P <0.0001), while we observed a decrease in alpha-linolenic acid (ALA, C18:3, P = 0.004). The fish oil supplementation also significantly decreased n-6 fatty acids (P = 0.05), such as arachidonic acid (AA, C20:4 n-6, P = 0.05) and docosapentaenoic acid (Osbond acid, C22:4 n-6, P <0.0001). These effects induced a decrease in the n-6/n-3 ratio (P < 0.0001). However, no effect of the diet and on variation of LA (C18:2) incorporation were noted between the different groups ( Table 1).

Effect of restricted diet and fish oil supplementation on fatty acid composition in adipose tissue at 39 weeks
Similar to the liver, the incorporation of fatty acids in adipose tissue is dependent on the diet quantity and its fatty acid components. Fish oil supplementation decreased the incorporation of LA (C18:2, P = 0.004) and ALA (C18:3, P < 0.0001), whereas it increased the incorporation of EPA (C20:5 n-3, P = 0.03) and DHA (C22:6 n-3, P = 0.05). These effects were associated with a decrease in total n-3 (P = 0.0002) and n-6 (P = 0.004) fatty acids and an increase in the n-6/ n-3 ratio (P = 0.005). In addition, the Rt diet improved the incorporation of LA (C18:2, P = 0.005) and AA (C20:4 n-6, P = 0.01), leading to an increase of total n-6 fatty acid (P = 0.005). However, the Rt diet decreased the DPA (C22:5 n-3, P = 0.01) incorporation, which was insufficient to affect the total n-3 fatty acid content as well as the n-6/n-3 ratio (Table 3).

Effect of restricted diet and fish oil supplementation on fatty acid composition in muscle tissue at 39 weeks
As shown in Table 4, we found that animals fed with a diet supplemented with fish oil had lower percentage of n-6 fatty acids (P < 0.0001) and higher n-3 fatty acids (P = 0.0004), leading to a lower n-6/n-3 ratio (P < 0.0001) in thoracic limb muscle as compared to animals fed with an unsupplemented diet. These effect were mostly due to a lower incorporation of LA (C18:2, P < 0.0001), AA (C20:4 n-6, P = 0.01) and Osbond acid (C22:4 n-6, P < 0.0001), as well as higher incorporation of EPA (C20: n-3, P < 0.0001), DPA (C22:5 n-3, P = 0.0008) and DHA (C22:6 n-3, P < 0.0001). However, the decreasing effect on ALA (C18:3) was not sufficient to reverse the effect on n-3 fatty acids. Furthermore, we showed an effect of diet on the percentage of LA (C18:2, P = 0.0007), AA (C20:4 n-6, P = 0.006) and Osbond acid (C22:4 n-6, P = 0.01), leading to a higher percentage of n-6 fatty acids (P < 0.0001). We also observed that the Rt diet increased the percentage of DPA (C22:5 n-3, P = 0.03) in thoracic limb muscle, but it did not affect the percentage of total n-3 fatty acids and the n-6/n-3 ratio (Table 4). We found a similar effect of fish oil supplementation on LA, ALA, EPA, Osbond acid, DHA and total n-6 fatty acids in the pelvic limb muscle, but it had no effect on AA, DPA, total n-3 fatty acids and the n-6/n-3 ratio. Concerning the effect of the Rt diet in pelvic limbs, it still affected LA and n-6 fatty acids, it decreased the percentage of EPA and had no effect on the other fatty acids or the n-6/n-3 ratio ( Table 5).

Effect of restricted diet and fish oil supplementation on body weight, fattening, feed conversion from 3 to 39 weeks and muscle surface area at slaughter
From week 3 to week 39, we observed a weak effect (P < 0.0001) on body weight and fattening (ultrasound, BIA) that increased gradually throughout the study. We also observed an effect of the diet (P < 0.0001) on body weight and fattening (ultrasound, BIA), as well as an effect of fish oil supplementation (P < 0.0001) (only fattening by ultrasound) ( Table 6). The effect of diet on body weight and fattening was reported for each period according to the different feed (growing: week 6 to 9 and week 9 to 18, P < 0.0001; before laying: week 18 to 21, P < 0.0001; and laying: week 24 to 39, P < 0.0001; Table 6). More precisely, Rt animals reduced body weight gain as compared to Ad animals from 6 (0.77 ± 0.01 kg vs 1.15 ± 0.008 kg, respectively; P < 0.0001) to 39 weeks (3.56 ± 0.05 kg vs 4.89 ± 0.05 kg, respectively; P < 0.0001) (Fig 1A).
We also observed a decrease in feed conversion in Rt animals compared to Ad animals from week 18 (week 18 to 21: 5.19 ± 1.46 vs 6.36 ± 1.43, respectively; P < 0.0001, and week 23 to 39: 3.60 ± 0.63 vs 4.12 ± 0.52, respectively; P < 0.0001) (Fig 1B and 1C). A similar negative effect of the Rt diet was noted for fattening from the second ultrasound measurement (week 6: Results are presented as lsmeans ± SEM. P values of the effects of diet, supplementation, week and the interaction between diet and supplementation, diet and week, and supplementation and week were considered as significant if P 0.05. Different individual letters in superscript (a,b,c and d) indicate significant differences.
2.5 ± 0.04 mm vs 4.0 ± 0.1 mm, respectively; P < 0.001) (Fig 2A and 2B). Regarding to the different time periods, fish oil supplementation during the growing period increased only the body weight (week 9 to 18: 1.77 ± 0.03 kg vs 2.07 ± 0.03 kg, P < 0.0001) and, during the laying period, decreased only the fattening (ultrasound) (week 24 to 39: 5.76 ± 0.07 mm vs 5.6 ± 0.06 mm; Table 6). However, when we tested the effect of the fish oil supplementation week by week, no significant effect was observed on body weight or fattening anymore (Figs 1A, 2A and 2B).
In addition, we showed a positive correlation between the measurement of fattening using ultrasound and measurements obtained using BIA (r = 0.81, P < 0.0001, Fig 2C and Table 7), by carcass analysis (r = 0.92, P < 0.0001, Fig 2D and Table 7), by X-ray computerised tomography scanner (r = 0.87, P < 0.0001, Fig 2E and Table 7) and by measuring the weight of abdominal fat (r = 0.86, P < 0.0001, Fig 2F and Table 7). This suggests that fat ultrasonographic examination represents a reliable method for evaluating fattening of hens. At week 39, the effect of the diet on animal growth was confirmed by measurement of the surface area of muscles (Pecto: 45.3 ± 5.04 cm 2 vs 53.96 ± 3.23 cm 2 , P = 0.04; oyster muscle: 4.7 ± 0.26 cm 2 vs 6.52 ± 0.23 cm 2 , P = 0.0004) (Fig 3A and 3B). However, no effect of fish oil supplementation was observed on muscle surface area.

Effect of restricted diet and fish oil supplementation on plasma metabolic factor concentrations from 3 to 39 weeks
As shown in Table 8 and Fig 4, all the plasma concentration profiles of biochemical parameters varied significantly with time (P < 0.0001), while plasma concentrations of triglyceride (P < 0.0001), phospholipid (P = 0.0003) and glucose (P = 0.02) were also decreased by the Rt diet and no effect of fish oil supplementation was observed. The Rt diet decreased the plasma triglyceride and glucose concentrations during both growing (triglyceride: 1.78 ± 0.01 g/l vs 1.88 ± 0.03 g/l, P = 0.001; glucose: 2.36 ± 0.03 g/l vs 2.46 ± 0.04 g/l, P = 0.04) and laying periods (triglyceride: 2.94 ± 0.23 g/l vs 3.80 ± 0.31 g/l, P = 0.002; glucose: 2.90 ± 0.05 g/l vs 3.09 ± 0.07 g/l, P = 0.003) (Tables 4 and 5) compared to Ad diet. The effects of the diet appeared more precisely at week 39 for plasma triglyceride concentrations (2.80 ± 0.04 g/l vs 4.25 ± 0.06 g/l, P = 0.003) and at week 32 for plasma glucose concentrations (2.85 ± 0.03 g/l vs 3.25 ± 0.04 g/l, P = 0.04) (Fig 4A and 4E). The effect of the Rt diet on plasma phospholipid concentrations was noted during the laying period (3.56 ± 0.23 g/l vs 4.37 ± 0.31 g/l, P = 0.003) ( Table 8) and more specifically at week 27 (4.07 ± 0.50 g/l vs 5.79 ± 0.85 g/l, P = 0.01) and week 39 (3.40 ± 0.30 g/l vs 4.25 ± 0.45 g/l, P = 0.003) (Fig 4B). Hence, the various diets had different effects depending on the time-point and the biochemical parameters studied.

Effect of restricted diet and fish oil supplementation on plasma adipokine concentrations from 3 to 39 weeks
As shown for plasma concentrations of biochemical parameters, we observed a significant but weak effect (P < 0.0001) on plasma ADIPOQ, NAMPT and RARRES2 concentrations (Table 9). Plasma ADIPOQ concentrations decreased from week 3 to 27 and increased slightly from week 27 to 39 (Fig 5A). Plasma NAMPT concentrations increased from week 3 to 18,  Metabolism and adipokines in broiler hens then decreased from week 18 to 27 and increased from week 27 to 39 (Fig 5B). Plasma RARRES2 concentrations alternately decreased and increased from week 3 to 18, then remained stable from week 18 to 32 and decreased from week 32 to 39 (Fig 5C). Unlike plasma NAMPT and RARRES2 concentrations, we showed that plasma ADIPOQ concentrations were decreased by the Rt diet (3.3 ± 0.14 μg/ml vs 3.43 ± 0.19 μg/ml, P = 0.003) and especially during the growing period (week 3 to 9: 4.25 ± 0.13 μg/ml vs 5.26 ± 0.3 μg/ml, P = 0.004) ( Table 9). Fish oil supplementation did not alter plasma ADIPOQ and NAMPT concentrations, whereas it decreased those of RARRES2 (252.92 ± 6.61 ng/ml vs 238.19 ± 5.88 ng/ml, P = 0.003) ( Table 9). Regarding to the different time periods, plasma concentrations of RARRES2 decreased during the growing period (week 9 to 18: 252.92 ± 6.61 ng/ml vs 238.19 ± 5.88 ng/ml, P = 0.0003), whereas they increased before the laying period (252.92 ± 6.61 ng/ml vs 238.19 ± 5.88 ng/ml, P < 0.0001) in supplemented animals (Table 9).

Correlations between adipokine levels in plasma and adipose tissue at 39 weeks
We did not find any correlation between adipokine expression in plasma and in ATs. However, we found a positive correlation between plasma ADIPOQ concentrations and the mRNA Table 11. mRNA expression of metabolic factors (FASN, FATP1, PPARG, FFAR4, GLUT8 and CD36) in liver, pectoralis major (Pectco), abdominal (abd) and subcutaneous (sc) adipose tissue (AT).   (Table 14).

Discussion
The present study shows that the quantity of food and its lipid composition influences body condition during development through the regulation of metabolic factors and adipokine expression in broiler breeder hens. As expected, animals fed with the Rt diet were thinner, had less fat and smaller muscles than Ad animals. The state of the fattening was evaluated regularly throughout the study using multiple methods, including fat ultrasonographic examinations for the first time and by BIA, a Results are presented as lsmeans ± SEM. P values of the effects of diet, supplementation and the interaction between diet and supplementation were considered as significant if P 0.05. Different letters in superscript (a,b and c) or group (ab, bc) of letters with no letter in common in superscript indicate significant differences.Ad:  non-invasive method. These results were positively correlated with those obtained by traditional methods, such as carcass analysis and weighing of abdominal adipose tissue. However, the calculations of BIA that were used do not yield results for animals less than 12 weeks old and we have no indication of the effect of the electrode clamps on the stress level in the animals. Thus, the use of ultrasound could provide a good technique for breeders to evaluate the fattening of their animals. Furthermore, the arrival on the market of new portable devices with very good image quality and for a reasonable price should favour the emergence of this technique. As compared to mammals, the main site of lipogenesis in chicken is the liver [35] and the late reduction in fat mass in Rt animals was associated with different gene expression depending on the type of metabolic tissue. On the one hand, Rt diet-induced lipogenesis (FASN) was positively correlated with a decrease in RARRES2 and its receptors CMKLR1 in AT abd, and promoted the transport of fat (FATP1) and glucose (GLUT8) in AT sc associated with an increase in ADIPOQ, CMKLR1 and CCRL2 expression. However, in contrast to our findings, Tahmoorespur et al. (2010) found an increasing effect of food restriction on ADIPOQ mRNA expression in AT [36]. These may be explained by a transitory effect which depends on the duration of the food restriction. In the present study, we also observed that plasma RARRES2 was negatively correlated with the percentage of fat determined by BIA or ultrasound. These data are in contrast to those reported in humans. Indeed, plasma RARRES2 levels are increased in obese patients and reduce after bariatric surgery [37]. All of these results indicate that adipokines, especially RARRES2 and ADIPOQ, may be potential markers of fat mass in broiler breeder hens.
In the current study, we observed no fish oil-induced effects on growth, performance or fattening, whereas fish oil supplementation significantly increased the levels of omega 3 PUFAs in the adipose tissue and muscles. Fish oil is one of the major source of EPA, DPA and DHA. Previous studies have reported variable effects of fish oil on performance parameters [38] [39] [40] [41]. The lack of an effect of fish oil in the current study may be attributed to differences in the source of the fish oil, its composition, the proportion of supplement added to the diet and the diet composition. Conversely to plants, animals were not able to produce EPA and DHA despite the multiple studies that show their beneficial effects on inflammation [42] and insulin resistance [43] [44]. However, animals can metabolise EPA and DHA by a series of desaturation and elongation reactions of ALA, which are particularly active in the liver and to a lesser extent in other tissues [45]. Here, the low proportion of fish oil in the diet (1%) was sufficient to decrease the n-6/n-3 ratio (except in adipose tissue and thoracic limb), as previously demonstrated by Koppenol (2014) [11]. Supplementation with fish oil also led to a significant increase of EPA in egg yolks and in liver tissue, but had no effect on DHA levels in the liver. This may be explained by a reduction in the retroconversion of EPA in DHA in vivo, or by an effect of the nutritional status on desaturase enzyme leading to a complex hormonal retrocontrol. Similar results have been described by López-Ferrer et al. investigating the effect of supplying linseed oil in the diet [38] [39] [46]. According to our funding Lopez-Ferrer et al., [39] also showed that intake of fish oils leads to an impairment of LA metabolism and a deficit of n-6 fatty acids.
In our study, we demonstrated for the first time that FFAR4, the main receptor of EPA and DHA [47], was expressed in the peripheral tissues and its expression in the liver and muscles of broiler breeder hen can be regulated by the diet and fish oil supplementation. We found a decreasing effect of fish oil supplementation onthe mRNA expression of FFAR4 in liver, as well as an increase of lipogenesis (FASN) and glucose transporter (GLUT8) associated with an increase of ADIPOQ expression. This inhibitory effect of fish oil supplementation on FFAR4 expression in liver is in contrast to data previously reported for the rat colon in response to chronic supplementation of diets with 10% fish oil for a period of seven weeks [48]. However, Cornall and colleagues reported a tissue-specific upregulation of FFAR4 expression in rats, such that high-fat diets increased FFAR4 expression in cardiac muscle and EDL skeletal muscle, but not in liver or soleus skeletal muscle [49]. The mechanism of EPA and DHA needs to be explored. We can speculate that a higher percentage of fish oil in the diet may increase the levels of EPA and DHA in metabolic tissue and consequently affect their metabolism.
In agreement with Maddineni et al. (2005) and Ramachandran et al. (2013), we showed that ADIPOQ, as well as its receptors ADIPOR1 and APIDOR2, was expressed in chicken adipose tissue, liver and muscles [50] [19]. Food deprivation is known to decrease ADIPOQ expression in adipose tissue and the liver [50] and to decrease expression of ADIPOQ receptors in adipose tissue [51]. However, our findings showed an increase in ADIPOQ mRNA expression in AT sc of Rt animals and no effect on ADIPOR1 and ADIPOR2 in either adipose tissue nor liver. We regard to plasma levels, we confirmed that the Rt diet had no effect on circulating ADIPOQ [50]. In addition, plasma level of the high molecular weight ADIPOQ isoform was inversely correlated to body weight and fat mass [52], while we did not find any correlation between circulating ADIPOQ and fat mass in the present study.
In chickens, NAMPT is mostly recognised as a myokine rather than an adipokine [21]. Herein, we showed its expression in adipose tissue, as well as in muscle and liver tissue, without different between tissues. NAMPT is not differentially expressed in adipose tissue from chickens selected for their high fat mass or in lean chickens [53], and no relationship was found between NAMPT mRNA expression and fat mass (data no shown).
Concerning the adipokine RARRES2 in avian species, only one of our previous studies revealed the presence of circulating RARRES2 and it expression in adipose tissue, liver and muscle of turkeys [20]. Hence, for the first time, in this study we demonstrated that RARRES2 was expressed in adipose tissue, liver and muscle of broiler breeder hens. Moreover, consistent with studies in humans [54] and rodents [55], we found that circulating RARRES2 was related to fattening. Interestingly, all of the adipokines studied were correlated with plasma triglycerides, phospholipids and insulin concentrations, even if they were unaffected by the diet and supplementation. Furthermore, it is well known that insulin is able to regulate glucose and fatty acid release. Therefore, we can hypothesise that changes in metabolic factors and adipokine expression may be controlled by insulin.
In summary, food restriction and fish oil supplementation modify the metabolic parameters, as well as the adipokine concentrations, in the plasma and tissues in the broiler breeder hens. In addition, we propose two additional tools (BIA and plasma analysis of adipokines) that can be used routinely by breeders to detect possible disorders of metabolism.
Supporting information S1 Fig. Description of the in vivo protocol. From one to 28 days of age (week 4), 320 female breeder chicks received an ad libitum diet (free access to food), called a starting diet. At 28 days of age (week 4), animals were distributed into two groups. The first group (n = 160 animals) received a restricted growing diet and the second group (ad libitum group; n = 160 animals) received the same diet on a daily basis, but the amount was 1.7 times greater than in restricted animals. From 63 days (week 9) to 273 days of age (week 39), these two groups were each subdivided into two groups, one with fish oil supplementation and one without fish oil. The four groups were: group RNS (restricted unsupplemented); group ANS (ad libitum unsupplemented); group RS (restricted supplemented); group AS (ad libitum supplemented). During this period, these four groups of animals received three different diets (growing, before laying and laying diets). (TIF) S1 Table. Composition of the diet with or without fish oil supplementation (%). Crtl: control without fish oil supplementation, Supp: supplemented with fish oil supplementation. (DOCX) S2