Long-term programming effect of embryonic hypoxia exposure and high-carbohydrate diet at first feeding on glucose metabolism in juvenile rainbow trout

It is widely recognised that environmental modifications (nutritional or non-nutritional) experienced at critical periods of an organism's life (for instance during early development) can cause long-term changes in metabolism and physiology, which is termed developmental programming, or metabolic programming when modifying metabolism (Lucas, 1998). Such adaptations often manifest through persistent modifications in gene expression patterns or metabolic signalling pathways, and may persist later in life in the absence of the environmental stimulus that initiated them (George et al., 2012; Kongsted et al., 2014; Patel and Srinivasan, 2002).

Metabolic programming studies have attracted broad attention in aquaculture in recent years and shed light on how modulations in early nutrition may confer adaptive advantages to organisms to better cope with their future nutritional environment (Balasubramanian et al., 2016; Fang et al., 2014; Panserat et al., 2017; Rocha et al., 2015; Vagner et al., 2009). In particular, in order to better understand and improve the glucose-intolerant phenotype in rainbow trout (Oncorhynchus mykiss), researchers have examined metabolic programming strategy using a strict nutritional stimulus in this species (Geurden et al., 2007, 2014). Promising results were obtained in some recent studies in trout. For instance, early nutritional stimulus at the first-feeding stage, such as high carbohydrate (HC) intake, was shown to induce persistent modifications in the mRNA levels of glucose-metabolism-related genes at the juvenile stage (Geurden et al., 2014). However, such a long-term programming effect was only noted in the muscle, and no persistent programming effect was observed in the liver of juvenile fish, despite the central role of this organ in glucose metabolism. These facts highlighted the necessity to optimise and review the programming conditions.

With this aim, a short-term metabolic programming assay was conducted with a strict non-nutritional stimulus, i.e. through embryonic hypoxia exposure (Liu et al., 2017). Indeed, hypoxia is a favourable stimulus candidate when investigating glucose-metabolism programming because hypoxia is known to affect hepatic glucose utilisation (Osumek et al., 2014; Zhong and Mostoslavsky, 2010). Moreover, such a stimulus can be easily applied before first feeding. Finally, effects of hypoxia are at least in part mediated by epigenetic mechanisms (Shmakova et al., 2014; Zhong and Mostoslavsky, 2010) such as DNA methylation or histone modifications (H3K4me3, H3K9me3, H3K36me3 and H3K9ac). Epigenetic modifications are considered an important mechanism mediating the long-term metabolic programming effect because they are sensitive to environmental changes and they constitute, by definition, the horizontal cell memory that is transmissible from one cell generation to another (Feil and Fraga, 2012; Szyf, 2009). In addition, the epigenetic modifications at the global level are known to affect chromatin stability, thereby leading to activation or repression of gene transcription (Klose and Bird, 2006; Kurdistani et al., 2004). Abnormal global DNA methylation or histone modification patterns are associated with metabolic disorders; thus, they were considered important biomarkers for health (Robertson, 2005; Seligson et al., 2005).

Considering hypoxia as a promising stimulus in the context of metabolic programming in trout, Liu and collaborators (2017) demonstrated that an acute hypoxic exposure during the embryonic stage affected glucose-metabolism-related gene expression in whole embryos at first feeding (Liu et al., 2017). For instance, when fed an HC diet at first feeding, significant increases in mRNA levels of genes involved in glycolytic and glucose transport pathways were noted in fish previously subjected to hypoxia. However, questions remain to be elucidated: (i) can such programming effects of an early hypoxic stimulus persist at the juvenile stage in trout; (ii) can a hypoxic stimulus applied during the embryonic stage combined with a nutritional stimulus at first feeding more profoundly affect glucose metabolism in trout later in life than a single stimulus; and (iii) if so, do these early stimuli modify the global epigenome in the long term?

The main objective of the present study was to investigate the above questions in juvenile trout that encountered an acute hypoxic stimulus (20% dissolved O₂ for 24 h) at the embryonic stage (152 degree days) alone (hypoxic history) or combined with an HC-diet stimulus at first feeding (HC dietary history) when challenged again with a 5-day high carbohydrate diet (30%) at the juvenile stage. The growth performance, main plasma metabolites, mRNA levels of glucose-metabolism-related genes and global epigenetic modifications were analysed in juvenile trout in the present study.

Investigations were conducted according to the guiding principles for the use and care of laboratory animals and in compliance with French and European regulations on animal welfare (Décret 2001-464, 29 May 2001 and Directive 2010/63/EU, respectively). This protocol and the project as a whole were approved by the French National Consultative Ethics Committee (reference numbers 2015112018112159 and 201511201756973).

Embryonic hypoxia and first-feeding stimuli were previously described and analysed by Liu and collaborators (Liu et al., 2017). Briefly, rainbow trout [Oncorhynchus mykiss (Walbaum 1792)] embryos at 152 degree days [corresponding to setting up of the primitive liver (Vernier, 1969)] were initially subjected to a 24-h acute hypoxic stimulus [20% dissolved O₂ (2.5 mg l⁻¹); 8°C] or maintained at normoxic conditions (11.0 mg l⁻¹ dissolved oxygen; 8°C) (Fig. 1, phase 1, stimulus 1). This first stimulus will be referred to as ‘oxygen history’ hereafter: either hypoxic history or normoxic history, respectively. Hypoxia was obtained and maintained by bubbling nitrogen in the water as previously described by Liu and collaborators (Liu et al., 2017). After the stimulus, embryos were kept in normoxic conditions. After hatching, fish alevins were reared in a flow-through rearing system supplied with natural spring water (18°C) under a natural photoperiod. At first feeding, fish were fed either an HC (∼60% carbohydrate) diet or a no carbohydrate (NC; 0% carbohydrate) diet for 5 days (Fig. 1, phase 1, stimulus 2, and Table S1). This second stimulus will be referred as ‘dietary history’ hereafter. Each group was conducted in triplicate.

After the second stimulus (first feeding), all groups (initial average wet body mass=0.1 g) were subsequently subjected to a 24-week growth trial (Fig. 1, phase 2) with a commercial diet (Skrettring, Fontaine-lès-Vervins, France) followed by a 5-day challenge test (Fig. 1, phase 3) with a challenge diet containing 30% carbohydrate (Table S1). Fish were fed eight times per day at the beginning and the feeding frequency gradually reduced to two times per day (from 18 weeks and thereafter). Each meal, the fish were fed to apparent satiation. Dry feed intake was calculated as the amount of feed supplied minus the amount of unconsumed feed. Residual feed pellets were counted and the corresponding mass was deducted from feed intake. The parameters for growth performance were monitored every 3 weeks and the following variables were calculated: (1) survival (%)=100×final fish number/initial fish number; (2) specific growth rate (SGR; % day⁻¹)=100×(lnM₂–lnM₁)/d; (3) feed intake (% day⁻¹)=100×{dry feed intake (g)/[(M₂+M₁)/2]}/d; and (4) feed efficiency (FE)=(M₂+M₃–M₁)/dry feed intake, where M₁ and M₂ are the initial and final average wet body mass, respectively, M₃ is the mass for dead fish (Li et al., 2010; Silva-Carrillo et al., 2012; Thodesen et al., 1999), and d is the experimental period in days.

At the end of the 5-day challenge test, fish were first anaesthetized in a benzocaine bath at 30 mg l⁻¹ and then killed in a benzocaine bath at 60 mg l⁻¹ 6 h after the last meal. Blood was collected from the caudal vasculature of nine fish per condition (three fish per tank) and centrifuged (3000 g, 5 min). The recovered plasma was frozen and stored at −20°C until analysis. Liver and white muscle tissues of six fish per condition (two fish per tank) were dissected and immediately frozen in liquid nitrogen, and stored at −80°C for further analysis.

The approximate composition of the diets was analysed as follows: dry matter was determined after drying to constant mass at 105°C; crude protein (N×6.25) was determined by the Kjeldahl method after acid digestion; crude lipid was determined by petroleum ether extraction (Soxtherm); gross energy was measured in an adiabatic bomb calorimeter (IKA, Heitersheim Gribeimer, Germany); ash was estimated through incinerating in a muffle furnace for 6 h at 600°C.

Plasma glucose, triglycerides, lactate and NEFA (non-esterified fatty acid) were analysed with Glucose RTU (BioMerieux, Marcy-l'Étoile, France), PAP 150 (BioMerieux), Lactate PAP (BioMerieux) and NEFA C (Wako Chemicals GmbH, Neuss, Germany) kits, respectively, according to the recommendations of each manufacturer.

The analysis of mRNA levels was conducted in liver and white muscle tissues of juvenile fish. Samples were homogenised in Trizol reagent (Invitrogen, Carlsbad, CA, USA) with Precellys^®24 (Bertin Technologies, Montigny-le-Bretonneux, France), and total RNA was then extracted according to the Trizol manufacturer's instructions. Total RNA (1 µg) was subsequently reverse-transcribed to cDNA in duplicate using the SuperScript III RNase H-Reverse Transcriptase kit (Invitrogen) with random primers (Promega, Charbonnières-les-Bains, France).

The primers used in quantitative real-time PCR (qPCR) assays for glucose-metabolism-related genes have previously been published, and the primer sequences used in the present study can be found in the previous studies (Marandel et al., 2015, 2016b; Liu et al., 2017). qPCR assays were performed with the Roche Lightcycler 480 system (Roche Diagnostics, Neuilly-sur-Seine, France). The reaction mix was 6 µl per sample, including 2 µl of diluted cDNA template (1:76), 0.12 µl of each primer (10 µmol l⁻¹), 3 µl of Light Cycler 480 SYBR^® Green I Master mix and 0.76 µl of DNase/RNase-free water (5 Prime GmbH, Hamburg, Germany). The qPCR protocol was initiated at 95°C for 10 min for the initial denaturation of the cDNA and hot-start Taq-polymerase activation, followed by 45 cycles of a two-step amplification programme (15 s at 95°C; 10 s at 60°C). Melting curves were monitored systematically (temperature gradient 0.11°C s^–1 from 65 to 97°C) at the end of the last amplification cycle to confirm the specificity of the amplification reaction. Each qPCR assay included replicate samples (duplicate of reverse transcription and PCR amplification) and negative controls (reverse-transcriptase- and cDNA-template-free samples). The relative quantification of mRNA levels of target genes were normalised to the transcript abundance of reference genes (β-actin for liver: forward 5′-GATGGGCCAGAAAGACAGCTA-3′, reverse 5′-TCGTCCCAGTTGGTGACGAT-3′; ef1α for muscle: forward 5′-TCCTCTTGGTCGTTTCGCTG-3′, reverse-5′-ACCCGAGGGACATCCTGTG-3′) using the E-method on Light Cycler software.

DNA extraction was performed on liver and muscle tissues of six fish per experimental treatment. Ten mg liver or muscle samples were subjected to 37°C overnight digestion under agitation (250 rpm) in 1 ml TNES buffer (pH=8; 125 mmol l⁻¹ NaCl, 10 mmol l⁻¹ EDTA, 0.5% SDS, 4 mol l⁻¹ urea, 10 mmol l⁻¹ Tris-HCl) supplemented with 100 μg proteinase K (P6556, Sigma-Aldrich, St Louis, MO, USA). One ml phenol–chloroform–isoamyl-alcohol (25:24:1) was added to each sample and mixed gently for 15 min followed by a 10,000 g centrifuge at room temperature for 15 min. A total of 100 μl 5 mol l⁻¹ NaCl and 1 ml 100% ice-cold ethanol was added to the supernatant and centrifuged for 15 min at 10,000 g, 4°C. The pellets were washed in 1 ml 75% ice-cold ethanol and centrifuged for 15 min at 10,000 g, 4°C again. Finally, the pellets were dried and re-suspended in DNase-free water and treated with 2.4 μg RNase (A7973, Promega) for 1 h at 37°C. DNA was quantified by NanoDrop (Thermo Fisher Scientific, Waltham, MA, USA) and the quality (high molecular weight) was verified on 1% agarose gel. DNA global methylation pattern [5-methylcytosine (5-mC)] was assessed using the MethylFlash Methylated DNA Quantification kit (EpiGentek, Farmingdale, NY, USA). Each analysis was performed in duplicate with 100 ng liver DNA or 200 ng muscle DNA samples following the manufacturer's instructions.

Histone extraction was performed on liver and muscle tissues of six fish per experimental treatment. One hundred mg tissue was homogenised using Precellys^®24 (Bertin Technologies) in 2 ml tubes containing 1 ml of TEB buffer [1×PBS, 0.5% Triton X-100, 5 mmol l⁻¹ NaBu and one tablet of protease inhibitor (Roche, cat. no. 04693116001)] and four 2.8 mm ceramic beads. The grinding programme was set as 2×10 s (15 s off) at 5000 rpm for liver, and 2×20 s (15 s off) at 5500 rpm for muscle samples. The homogenates were kept on ice for 20 min, transferred to a new 1.5 ml tube and centrifuged for 10 min at 2000 rpm, 4°C. Pellets were then re-suspended in 0.5 N HCl containing 10% glycerol (volume between 100 μl and 1200 μl depending on the size of the pellet) and incubated on ice for 30 min (vortex every 10 min). Samples were then centrifuged for 5 min at 12,000 rpm, 4°C. Three volumes of iced acetone were then added to the supernatant and precipitation was performed at −20°C overnight. Samples were centrifuged at 12,000 rpm, 4°C for 5 min and pellets were washed three times with 100% ice-cold acetone. Pellets were re-suspended in distilled water and incubated at 60°C for 1 h. Total protein concentration was determined using the Bio-Rad protein assay kit (Bio-Rad Laboratories, Munich, Germany). Lysate (5 μg and 14 μg of total protein for liver and muscle samples, respectively) were subjected to SDS-PAGE and western blotting using the appropriate antibody. Samples were loaded on AnykD™ Criterion™ TGX™ Precast Midi Protein Gel for the analysis of H3K4me3, H3K9me3, H3K36me3 and H3K9ac marks. Western blotting was performed with the Trans-Blot^® Turbo™ Transfer Starter System (Bio-Rad Laboratories) using the pre-programmed protocol Mixed MW (Turbo) 5–150 kDa and 7 min for transfer. Anti-H3K4me3 (C15410003), anti-H3K9me3 (C15410056), anti-H3K36me3 (C15410192) and anti-H3K9ac (C15410004) were purchased from Diagenode (Liège, Belgium); anti-H3 (ab1791) was purchased from Abcam (Cambridge, UK).

The results are presented as means±s.d. The effects of oxygen, diet and oxygen×diet on different parameters were tested using R software (v3.1.0)/R Commander by two-way ANOVA with oxygen and diet as independent variables. When oxygen×diet interaction was significant, means of all treatments were compared by a Tukey's post hoc analysis.

In general, there were few significant differences in survival and growth performance among different groups during the 24-week growth trial (Fig. 1, phase 2; Table 1). However, during the last 6 weeks (19–24 weeks), juveniles with the HC dietary history showed significantly lower final body mass and feed efficiency compared with those receiving the NC diet at first feeding. Additionally, fish with a hypoxic history showed a significantly higher feed efficiency compared with those kept under normoxia during early development.

The major plasma metabolites (Table 2) related to hypoxia and carbohydrate intake (glucose, triglycerides, lactate and NEFA) were measured in juveniles at the end of the 5-day challenge test (6 h after the last meal; Fig. 1, phase 3). There was no significant effect of oxygen or dietary history on plasma metabolites.

We then analysed mRNA levels of glucose-metabolism-related genes in both liver and muscle of juvenile trout after the 5-day challenge test (6 h after the last meal; Fig. 1, phase 3). No interaction effect between oxygen and dietary histories were observed on the mRNA levels of all target genes in juvenile fish.

Regarding the effect of dietary history in liver tissue (Table 3), gcka (glucokinase a) and gckb (glucokinase b) mRNA levels were significantly lower in fish that received the HC-diet stimulus at first feeding compared with those fed the NC diet. Concerning the effect of oxygen history, fish subjected to a hypoxic stimulus during embryogenesis displayed significantly higher mRNA levels of gluconeogenic fbp1a (fructose 1,6-bisphosphatase-1a), g6pcb1b (glucose 6-phosphatase-b1 b) and g6pcb2a (glucose 6-phosphatase-b2 a) compared with their normoxic controls. Similarly higher mRNA levels of glucose transporters glut1ba (glucose transporter-1b a) and glut2b (glucose transporter-2b) were found in juvenile trout with the hypoxic history than in those kept under normoxia during early life.

In muscle (Table 4), no significant effect of dietary history was observed in juvenile fish except for the mRNA level of pfkmbb [phosphofructokinase (muscle)-bb], which was significantly lower in fish with the HC dietary history than in fish with the NC dietary history. With respect to the effect of oxygen history (Fig. 1, phase 1, stimulus 1), significantly higher mRNA levels of glucose transport genes glut4a (glucose transporter-4a) and glut4b (glucose transporter-4b) and glycolytic pkmaa [pyruvate kinase (muscle)-aa] were observed in juvenile fish with hypoxic history compared with their controls.

The 5-methyl-cytosine methylation status was analysed in liver and muscle tissues of juvenile trout to test the global DNA methylation level (Table 5). No significant differences in global DNA methylation level in either liver or muscle of juvenile trout were observed regardless of the oxygen or dietary history.

Global levels of selected histone modifications (H3K4me3, H3K9me3, H3K36me3 and H3K4ac) were also measured in both liver and muscle tissues of juvenile trout (Fig. 2). In liver (Fig. 2A), a significant interaction between the effects of oxygen and dietary histories (Fig. 1, phase 1, stimuli 1 and 2, respectively) was observed on H3K9me3 level even though no significant independent effect of oxygen or nutritional history was monitored. By contrast, the global level of all these four histone modifications remained stable in the muscle of juvenile fish (Fig. 2B), whatever the historical stimuli conditions.

Previous studies in mammals extensively documented that early-life history may profoundly affect the metabolic adaptability and risk of developing disease in adults, making metabolic programming an important topic in scientific research (Dunn and Bale, 2011; Hochberg et al., 2010; Lucas, 1998; Romani-Perez et al., 2016; Taylor and Poston, 2007; Vickers et al., 2000). This type of study was introduced into aquaculture a decade ago, raising the prospect of tailoring the metabolic capacities in fish through the manipulation of early nutrition (Fang et al., 2014; Geurden et al., 2007, 2014; Gong et al., 2015; Izquierdo et al., 2015; Lazzarotto et al., 2015; Rocha et al., 2015). Recently, we reported that an acute non-nutritional hypoxic stimulus during early developmental stages can affect glucose metabolism in rainbow trout alevins in the short term (Liu et al., 2017). With the aim of investigating the long-term effects in juvenile fish induced by such an early hypoxic stimulus, the present study further tested the effects of an acute hypoxic stimulus alone or combined with an HC diet at first feeding on the growth performance, levels of plasma metabolites, mRNA level of glucose-metabolism-related genes and epigenetic modifications in juvenile trout.

In the present study, long-term programming effects of an HC stimulus applied at first feeding was tested in juvenile trout after the 24-week growth trial. Significantly lower body mass and feed efficiency were observed at the end of the growth trial in fish with an HC dietary history compared with those fed the NC diet at first feeding. Results indicated that an early HC dietary stimulus may lead to some detrimental effects in long-term growth performance in trout. However, such effects were not reported in the study previously performed by Geurden and collaborators (2014), in which a 5-day HC dietary stimulus at first feeding did not affect growth performance in juvenile trout. This difference can possibly be explained by the different durations of the growth trial and of the challenge period in these two studies. In the present study, the inferior growth performance of fish with the HC dietary history was not observed until the end of the growth trial (19–24 weeks; 133–168 days). In comparison, the corresponding pre-challenge period in the study of Geurden and collaborators was done over 105 days, which may be too short to monitor a significant difference in growth.

Durable adaptive changes in gene expression patterns are one of the most important biological mechanisms that mediate the persistence of programming effects (Lucas, 1998). At the molecular level, results obtained from the present study showed that dietary history significantly affected mRNA levels of some glycolytic genes in juvenile trout. Indeed, the HC dietary stimulus at first feeding seems to decrease glycolysis in juvenile trout, i.e. mRNA levels of hepatic gcka and gckb and muscular pfkmbb were lower in fish with an HC dietary history than those with an NC dietary history. These findings were opposite to what was reported for the short-term effect (i.e. at first feeding) (Liu et al., 2017). Indeed, at first feeding, gcka and gckb mRNA levels were increased in fish that received the HC dietary stimulus compared with the control fish (Liu et al., 2017). Additionally, the lower mRNA level of pfkmbb in the muscle of juvenile fish with the HC dietary history was not detected previously in the short term. All these results revealed the presumable existence of indirect programming regulations (i.e. through altered hormonal levels) of the long-term expression of metabolic genes by early dietary stimulus. In comparison, the decrease in glycolytic genes in juvenile fish fed an HC diet at first feeding was consistent with the previous study performed by Geurden and collaborators (2014), which showed a decrease in hk1 (hexokinase 1) and pkm mRNA levels in the muscle of juvenile trout with an HC dietary history (Geurden et al., 2014).

It should be noted that, similar to previous studies in mammals (Bellinger and Langley-Evans, 2005; de Maredsous et al., 2016; Devaskar and Thamotharan, 2007), the present study balanced the dietary carbohydrate proportion against protein content to keep the same energy level in diets. Therefore, cautions should be taken when considering the exact causal factor for the long-term effect of dietary history: high dietary carbohydrates, low dietary protein, or both.

Apart from perinatal nutrition, fetal hypoxia is one of the most extensively studied environmental factors able to induce long-term metabolic changes in mammals (Myatt, 2006; Seckl and Holmes, 2007). Several studies in both mammals and teleost species showed that hypoxia exposure during early life can affect growth and glucose metabolism in adulthood (Camm et al., 2011; Osumek et al., 2014; Vanderplancke et al., 2015). In the present study, a subtle but significant increase in feed efficiency in weeks 19–24 was found in juveniles with a hypoxic history compared with those with a normoxic history. Such an increase in feed efficiency by early hypoxic stimulus has never been reported in trout, although the detailed mechanism behind this long-term effect is still not clear.

Concerning the long-term effect of oxygen history on glucose metabolism in juvenile fish challenged with an HC diet, our results highlighted a significant activating effect of hypoxic history on mRNA levels of genes involved in gluconeogenesis (fbp1a, g6pcb1b, g6pcb2a) and glucose transport (glut1ba and glut2b in liver, and glut4a, glut4b in muscle) pathways. Meanwhile, the hypoxic stimulus rarely induced long-term modifications in mRNA levels of glycolytic genes, except for pkmaa in the muscle of juvenile fish.

These results were quite in accordance with the modifications induced by hypoxic history in the short term described in the previous study (Liu et al., 2017). Indeed, mRNA levels of fbp1a, glut1ba, glut4a and pkmaa in fish with a hypoxic history were consistently higher compared with those with a normoxic history in both the short (first-feeding alevins) (Liu et al., 2017) and long (juveniles) term. The persistent alteration in the expression pattern of these genes demonstrated that an acute hypoxic stimulus applied during early development can quickly modify the transcription of glucose-metabolism-related genes in trout, and these modifications may remain until the juvenile stage.

It is to be noted that, both in the short-term (Liu et al., 2017) and long-term experiments, we observed a strong and consistent activating effect of hypoxic history on glucose-transport genes in trout. These results were in line with a previous study in mammals that showed that prenatal hypoxia can cause persistent alterations in the expression of glucose-transport genes in adult rat offspring (Camm et al., 2011). However, contradictory results were obtained in a previous study in European sea bass, which reported that the mRNA level of glut2 was increased after a moderate hypoxic episode during the larval stage but, concerning the long-term effect, decreased in the liver of juvenile fish (Vanderplancke et al., 2015). The differences between this study and the present study can be attributed to the different hypoxic intensity and duration, and also the different physiological responses to hypoxia within these two species. Nevertheless, the increased mRNA levels of glucose-transport genes in both liver and muscle of juvenile fish with a hypoxic history in the present study may to some extent prove a better functioning peripheral glucose disposal. Although, after the challenge test, the glycaemic level was similar and ranged between 6.6 and 8.3 mmol l⁻¹ among different groups, which corresponded to the HC-diet-induced hyperglycaemia (>4 mmol l⁻¹) in rainbow trout, it could possibly be explained by the concomitant enhanced gluconeogenesis (fbp1a, g6pcb1b and g6pcb2a) in the liver of fish that experienced hypoxia during development. Globally, we hypothesised that glucose transport activated by hypoxic history might provide some clues to ameliorate the glucose-intolerant phenotype in trout, which was known to be related to low glucose transport in muscle of this species (Blasco et al., 1996; Kamalam et al., 2017; Polakof et al., 2012; West et al., 1993).

Remodelling of the epigenetic landscape is an important mechanism relating to the persistent metabolic adaptations by early environmental stimuli. DNA methylation and histone modifications are the most comprehensively studied forms of epigenetic markers, which are known to affect gene transcription (Klose and Bird, 2006; Kurdistani et al., 2004). Previous studies showed that both dietary carbohydrate levels and a hypoxic stimulus can alter global DNA methylation and histone modification (Craig and Moon, 2013; Marandel et al., 2016a; Chen et al., 2006; Hancock et al., 2015; Honma et al., 2007; Perez-Perri et al., 2011; Tausendschön et al., 2011; Watson et al., 2010). However, results obtained in the present study showed that the global DNA methylation level and histone modifications were barely affected by oxygen and dietary histories in juvenile fish after the challenge test, except for an interactive effect of oxygen and dietary histories on the global H3K9me3 level observed in the liver of juvenile fish. Despite this interaction, no statistically significant difference was found within groups.

In summary, our study showed that an acute hypoxic stimulus during embryonic development alone or an HC-diet stimulus at first feeding can affect growth performance in the long term in rainbow trout. Both stimuli also have a long-term programming effect on the mRNA levels of glucose-metabolism-related genes in juvenile trout.

We thank F. Terrier, P. Maunas, N. Turonnent and F. Vallée for animal care and preparation of diets. We thank I. Geurden for the design of the diet formulae. Thanks to A. Surget, L. Larroquet and E. Arbenoits for their help with sampling and their technical assistance in the laboratory.