Acute stress response in gilthead sea bream (Sparus aurata L.) is time-of-day dependent: Physiological and oxidative stress indicators.

Since fish show daily rhythms in most physiological functions, it should not be surprising that stressors may have different effects depending on the timing of exposure. In this study, we investigated the influence of time of day on the stress responses, at both physiological and cellular levels, in gilthead sea bream (Sparus aurata L.) submitted to air exposure for 30 s and then returned to their tank. One hour after air exposure, blood, hypothalamus and liver samples were taken. Six fish per experimental group (control and stressed) were sampled every 4 h during a 24-h cycle. Fish were fed in the middle of the light cycle (ML) and locomotor activity rhythms were recorded using infrared photocells to determine their daily activity pattern of behaviour, which showed a peak around feeding time in all fish. In the control group, cortisol levels did not show daily rhythmicity, whereas in the stressed fish, a daily rhythm of plasma cortisol was observed, being the average values higher than in the control group, with increased differences during the dark phase. Blood glucose showed daily rhythmicity in the control group but not in the stressed one which also showed higher values at all sampling points. In the hypothalamus of control fish, a daily rhythm of corticotropin-releasing hormone (crh) gene expression was observed, with the acrophase at the beginning of the light phase. However, in the stressed fish, this rhythm was abolished. The expression of crh-binding protein (crhbp) showed a peak at the end of the dark phase in the control group, whereas in the stressed sea bream, this peak was found at ML. Regarding hepatic gene expression of oxidative stress biomarkers: (i) cytochrome c oxidase 4 showed daily rhythmicity in both control and stressed fish, with the acrophases located around ML, (ii) peroxiredoxin (prdx) 3 and 5 (prdx5) only presented daily rhythmicity of expression in the stressed fish, with the acrophase located at the beginning of the light cycle and (iii) uncoupling protein 1 showed significant differences between sampling points only in the control group, with significantly higher expression at the beginning of the dark phase. Taken together, these results indicate that stress response in gilthead sea bream is time-dependent as cortisol level rose higher at night, and that different rhythmic mechanisms interplay in the control of neuroendocrine and cellular stress responses.


INTRODUCTION
Fish in the wild and in aquaculture facilities face a variety of challenges, such as attacks from predators, food competition, disturbance and exposure to poor water quality, which seriously compromise fish welfare (Huntingford et al., 2006). The specimens react to these adverse conditions through cellular, neuroendocrine and behavioural adjustments, although the circadian mechanisms controlling these responses are not fully understood (Kulczykowska & Sánchez-Vázquez, 2010).
The primary physiological response to stress in fish involves two major neuroendocrine pathways: (i) the hypothalamic sympathetic chromaffin cells (HSC) axis and (ii) the hypothalamic-pituitary-interrenal (HPI) axis (Wendelaar Bonga, 1997). In the HPI cascade, corticotropin-releasing hormone (CRH) is synthesised in the hypothalamus and activates the production and release of adrenocorticotropic hormone (ACTH) from the pituitary, which in turn stimulates the production and release of cortisol in the interrenal cells. In the bloodstream, cortisol stimulates glycogenolysis to cope with the increased energy demand (Mommsen et al., 1999). At the hypothalamic level, a CRH-binding protein (CRH-BP) with antagonistic roles to CRH has been also described in fish (Huising et al., 2004;Wunderink et al., 2011). In mammals, daily rhythms in the HPI axis have been reported (Haus, 2007), with cortisol levels rising at the beginning of the active phase of the animal. In fish, plasma cortisol daily rhythms have recently been reviewed (Ellis et al., 2012). In Senegalese sole (Solea senegalensis K.), a nocturnal flatfish, marked daily oscillations in cortisol appeared under light-dark (LD) conditions with a peak in the afternoon, which persisted under continuous light conditions with lower values (Oliveira et al., 2013). In this flatfish, a recent paper revealed that stress responses differed during day or night, so that higher cortisol was registered when the stressor was applied at ''zeitgeber'' time 1 (ZT 1; one hour after lights on) than at ZT 13 (one hour after lights off) (López-Olmeda et al., 2013). In gilthead sea bream (Sparus aurata), plasma cortisol showed a postprandial peak in both fish fed in the middle of the day and in the middle of the night, indicating that feeding time influenced the daily rhythm of cortisol production (Montoya et al., 2010). In rainbow trout, cortisol exhibited a diurnal pattern, which also seemed to be correlated with feeding time, although additional changes associated with the scotophase were observed too (Holloway et al., 1994). However, despite the great interest of this teleost fish for the European aquaculture industry, little is known about the existence of daily rhythms in the HPI axis and the time-dependent response to acute stress.
The role of mitochondria as the first responders to various stress challenging homeostasis of the cell and organism has been extensively evidenced in many organisms (Manoli et al., 2007), including gilthead sea bream Pérez-Sánchez et al., 2013). The mitochondrial DNA of current vertebrates encodes 37 genes, such as cytochrome c oxidase 4 (coxIV), peroxiredoxins (prdx) or uncoupling proteins (ucp), with many of them being involved in the maintenance of balance between the oxidative and antioxidative processes that occur inside the cell (Brown, 2008). Indeed, recent studies have reported changes in the expression of these genes when sea bream were subjected to stress conditions (Bermejo-Nogales et al., 2010Pérez-Sánchez et al., 2011, pointing that these mitochondrial genes could be used as biomarkers of health and welfare in this fish species (Pérez-Sánchez et al., 2011. However, there are no data about the time-dependent differences in their expression when fish are stressed at different times of the day. The aim of this research was to investigate the possible existence of time-dependent stress response in gilthead sea bream. To this end, we recorded the locomotor activity of light-entrained fish and studied the effect of 30 s air exposure at different times of the day and night on physiological stress indicators (cortisol and glucose) as well as hypothalamic expression of genes encoding hormones of HPI axis (crh and crhbp) and mitochondrial oxidative stress biomarkers (coxIV,prdx3,prdx5 and ucp1).

Animals and housing
A total of 72 gilthead sea bream (211 ± 6 g initial body weight) were obtained from a local farm (Culmarex S.A., Aguilas, Murcia) and reared at the marine facilities of the University of Murcia located at the Naval Base of Algameca (E.N.A., Cartagena, Spain). Fish were kept in 150-L tanks supplied with aeration and filtered seawater from an open system. The photoperiod was set at 12:12 h LD and water temperature at 18 C.

Experimental design
Experimental procedure complied with the Guidelines of the European Union (2010/63/UE) and the Spanish legislation (RD 53/2013 and law 32/2007) for the use of animals in research. In addition, the experimental design and methodology followed in this investigation were in accordance with the international ethical standards of Chronobiology International (Portaluppi et al., 2010).
Fish were divided into 12 tanks of 150 L (n ¼ 6/tank). Each tank was equipped with an automatic feeder (EHEIM, model 3581, Germany), which provided the fish with 1% of the biomass once a day (D-4 EXCELL 2-P, Skretting), in the middle of the light phase (ML), at ZT6. Locomotor activity was measured by means of infrared photocells (Omron, mod E3S-AD62, Kyoto, Japan) immersed in each tank under the feeder and 3 cm from the water surface. A computer connected to the photocells counted and stored the number of light beam interruptions in 10-min intervals. This system has been previously used and validated in this species (Sánchez et al., 2009).
Gilthead sea bream were maintained under these experimental conditions for two weeks and, after one day of fasting, blood, hypothalamus and liver samples were collected. Six fish per treatment (stressed and control) were sampled every 4 h during a 24-h cycle, at ZT3,7,11,15,19 and 23. To this end, one hour before each sampling point (ZT2,6,10,14,18 and 22), six fish were removed from their tank and exposed to the air for 30 s. This experimental procedure has been previously reported to elicit an acute stress response in gilthead sea bream (Arends et al., 1999(Arends et al., , 2000. Then, fish were returned to the tank and sampled one hour later (stressed group). Fish from the control group, in contrast, were sampled directly at each sampling time ( Figure 1). Both groups of fish were anesthetised with eugenol (clove oil essence; Guinama, Valencia, Spain) dissolved in water at a concentration of 50 mL/L. Previously, eugenol was diluted in ethanol (1 eugenol:9 ethanol) to facilitate dissolution in water (Cooke et al., 2004). Blood was collected by caudal puncture with heparinised sterile syringes. Blood samples were collected from all fish of each tank in less than 5 min, to avoid the increase of plasma cortisol and glucose levels originated by manipulation (Molinero et al., 1997). Blood was centrifuged at 3000 rpm for 15 min at 4 C, and plasma was separated and frozen at À80 C until analysis. After blood collection, fish were sacrificed by decapitation and hypothalamus, and liver samples were collected and snap frozen and stored at À80 C until further analysis. During the dark phase, a dim red light (4600 nm) was used for sampling.

Plasma cortisol and glucose analyses
Blood glucose concentration was measured immediately after extraction by means of a glucometer (Glucocard G meter, Menarini, Italy). Plasma cortisol levels were measured with a commercial ELISA kit (IBL, Hamburg, Germany). Both analytical techniques had been previously validated for gilthead sea bream (López-Olmeda et al., 2009a,b).

Gene expression analyses
Hypothalamus and liver samples were homogenised in Trizol reagent (Invitrogen, Carlsbad, CA) using a tissue homogeniser (POLYTRON Õ , PT1200, Kinematica, Lucerne, Switzerland). For total RNA isolation, the homogenised tissues were mixed with chloroform and separated by centrifugation. RNA was then precipitated from the aqueous phase with isopropanol. Total RNA concentration was determined by spectrometry (Nanodrop Õ ND-1000, Thermo Fisher Scientific Inc., Wilmington, DE), and 1 mg was treated with DNase I amplification grade (1 unit/mg RNA, Invitrogen) to prevent genomic DNA contamination. cDNA synthesis was carried out with Superscript III Reverse Transcriptase (Invitrogen) and Oligo (dT) 12-18 (Invitrogen) in a 20 mL reaction volume. Real-time PCR was performed using SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA) and ABI Prism 7500 apparatus (Applied Biosystems). The ABI Sequence Detection System 7000 software (Applied Biosystems) was programmed to perform the following protocol: 95 C for 10 min, followed by 40 cycles at 95 C for 15 s and 60 C for 1 min. The final volume of the PCR reaction was 20 mL: 5 mL of cDNA, 10 mL of the qPCR Master Mix and 5 mL of forward and reverse primers (Table 1). All samples were run in triplicate. The primers used to amplify prdx3 and prdx5 genes were previously tested and validated for sea bream (Pérez- Sánchez et al., 2011), as well as those for ucp1 (Bermejo-Nogales et al., 2010), coxIV (Pérez- Sánchez et al., 2013), crh and crhbp (Martos-Sitcha et al., 2014). The amplification efficiency, specificity of primers and the quantity of cDNA per sample were tested by the standard-curve method. Moreover, melting curves were analysed to verify PCR specificity. The relative expression of all genes was calculated by the 2 ÀDDCT method (Livak & Schmittgen, 2001), using S. aurata -actin as the endogenous reference.

Gene
Accession number Acute stress response in gilthead sea bream 1053

Data analysis
Locomotor activity records were analysed and represented as mean waveforms, for which chronobiology software El Temps was used (Version 1,228; Prof. Díez-Noguera, University of Barcelona). Glucose, cortisol and gene expression data were subjected to Cosinor analysis to test the existence of significant daily rhythmicity. Cosinor analysis is based on least squares approximation of time series data with a cosine function of known period of the type where Mesor is the time series mean; Amplitude is a measure of the amount of temporal variability explained by the rhythm; period () is the cycle length of the rhythm, i.e. 24 h for circadian rhythms; and Acrophase is the time of the peak value relative to the designated time scale. Cosinor analysis also provided a statistical value for a null hypothesis of zero amplitude. Therefore, if for a statistical significance of p50.05 this null hypothesis was rejected, the amplitude could be considered as differing from 0, thereby constituting evidence for the existence of a statistically significant rhythm of the given period under consideration. Statistical differences in cortisol, glucose and gene expression levels between sampling points were analysed by a one-way ANOVA (ANOVA I). In addition, at each sampling point, cortisol, glucose and target genes expression levels were compared between treatments (control versus stressed) by means of a t-test, for which a Levene's test was previously used to check for homogeneity of variances. A Univariate General Linear Model (GLM) was carried out to analyse possible interactions between experimental groups and time points. For this, the fixed factors were ''ZT'' and ''treatments''.
All statistical tests were carried out with the SPSS v19.0 program (SPSS Inc., Chicago, IL), being the statistical threshold set at p values 50.05 in all tests. All values are reported as the mean ± SEM.

Locomotor activity rhythms
Gilthead sea bream activity showed an arrhythmic daily pattern of locomotor activity (57% of the total daily activity registered during the light phase), displaying most of activity around meal time (food anticipatory activity, FAA), followed by a gradual decrease after feeding. Fish increased significantly their activity levels 1.5 h before meal time, reaching a peak just before meal time ( Figure 2). FAA was calculated as the time span in which activity increased 50% over the baseline without subsequent inflections until meal time.

Circulating physiological stress indicators
In the control fish, plasma cortisol values did not show significant differences between sampling points. However, a significant daily rhythm was observed in fish subjected to stress (Cosinor, p50.05) with the acrophase being found around the middle of the dark phase (ZT ¼ 18:24) (Table 2; Figure 3). In addition, plasma cortisol levels in the stressed fish were significantly higher than in the control group at all sampling points, except at ZT3, with mean differences between groups being higher during the scotophase (66.1 ± 9.0 ng/mL) than during the photophase (29.8 ± 8.2 ng/mL) (t-test independent samples, p50.05) (Figure 4). A significant daily rhythm of blood glucose was observed in the control group with the acrophase located at ZT ¼ 16:17 (Cosinor, p50.05) but not in the   (Table 2). Overall, blood glucose levels in the stressed fish were significantly higher than in the control ones at all sampling points except at ZT19 (t-test independent samples, p50.05) ( Figure 5). However, in this case, the average increase in the stressed group was similar during the photophase and scotophase ($1.1 mmol/L) (t-test independent samples, p40.05).

Brain HPI axis
Hypothalamic crh showed a significant daily rhythm of expression in the control group (Cosinor, p50.05), with the acrophase at the beginning of the light phase and the lowest levels during the first hours of the night (Table 2; Figure 3). However, in the stressed fish, no daily rhythmicity was observed, though a significant peak of expression was found at ZT23 (ANOVA I, p50.05). Moreover, there was a statistically significant interaction between the effects of ZTs and treatments (control/stressed) (Univariate GLM, p50.01) (Table 3) being crh expression at ZT3 in the control group significantly higher than in the stressed one (t-test independent samples, p50.05) ( Figure 6A). As regards crhbp expression, significant differences were detected in both experimental groups: in the control group maximum expression was observed at ZT23 (ANOVA I, p50.05), whereas in the stressed fish, crhbp expression peaked at ZT7. Furthermore, the Univariate GLM revealed an interaction between treatments and sampling points (p50.01) ( Table 3). Simple main effects analysis showed that crhbp expression was significantly lower in the control sea bream at ZT3, whereas at ZT23, this expression was significantly higher than in the stressed group (t-test independent samples, p50.05) ( Figure 6B). However, the Cosinor analysis failed to reveal significant daily rhythms of crhbp expression in both groups.

Expression of mitochondrial oxidative stress biomarkers in liver
CoxIV expression displayed a significant daily rhythm in both control and stressed fish (Cosinor, p50.05), with the acrophases located around the middle of the day ($1-1.5 h before feeding time) (Table 2; Figure 3). In addition, there was a significant interaction between the effects of ZTs and treatments (Univariate GLM, p50.01) (Table 3). Thus, in the control sea bream, coxIV expression was significantly higher than in the stressed  The acrophase is indicated by a circle, black and white for stressed and control group, respectively. The confidence intervals (set at 95%) are indicated by the lateral bars. White and black bars above the graph represent light and darkness, respectively.
fish at ZT3 and ZT7 (t-test independent samples, p50.05) ( Figure 7A). Ucp1 gene expression showed significant differences between sampling points in the control group, with a peak of expression at ZT15 (ANOVA I, p50.05) ( Figure 7B). However, significant daily rhythmicity was not detected using the Cosinor analysis. On the contrary, in the stressed group, neither significant differences between sampling points nor daily rhythmicity was observed.
In the case of prdx3, no significant rhythmicity or daily differences between sampling points were observed in the control group. However, a significant daily rhythm was detected in the stressed fish (Cosinor, p50.05) with the acrophase located $2 h after lights on (Table 2), peaking at ZT3 (ANOVA I, p50.05) (Figure 2). The Univariate GLM showed an interaction between the fixed factors (ZT and treatments) (p50.01) (Table 3). Hence, expression levels at ZT3 and ZT23 were significantly lower in the control fish (t-test independent samples, p50.05) ( Figure 7C).      Prdx5 expression showed significant differences between sampling points in both control and stressed sea bream: in the control fish, prdx5 expression presented two peaks, at ZT7 and ZT15, whereas in the stressed group, a peak of expression was observed at ZT7 (ANOVA I, p50.05). Furthermore, a significant daily rhythm (Cosinor, p50.05) was found in the stressed fish group, with the acrophase located during the day at ZT ¼ 4:30. In addition, there was a significant interaction between the effects of ZTs and treatments (Univariate GLM, p50.01) (Table 3), with prdx5 expression being higher in the control fish at ZT15, whereas at ZT18, this expression was down-regulated compared with the stressed group (t-test independent samples, p50.05) ( Figure 7D).

DISCUSSION
In vertebrates, cortisol rhythms are tightly related to the species-specific circadian rhythm of behaviour. Thus, the acrophase of cortisol daily rhythm is usually located in the transition from dark to light in diurnal species such as humans, while it is located at the beginning of the dark phase in nocturnal animals, such as the rat (Dickmeis, 2009). In teleosts, daily rhythms of plasma cortisol have been also reported to be species-specific and related to the activity pattern (diurnal/nocturnal) (Ellis et al., 2012).
In Senegalese sole, a nocturnal flatfish, cortisol levels peaked at the beginning of the dark phase (López-Olmeda et al., 2013). In this study, the control sea bream showed a cortisol increase during the first hours of the dark phase, whereas the stressed fish showed a daily rhythm of plasma cortisol with the acrophase at mid-darkness (MD). All gilthead sea bream (control and stressed) were fed at ML, and fish actually showed an activity peak around meal time, suggesting their synchronisation to the feeding cycle. Indeed, feeding entrainment occurs when fish are presented with food on a daily basis and they display locomotor activity in anticipation of the forthcoming meal (López-Olmeda et al., 2009b). In gilthead sea bream, previous results pointed out the role of feeding time in changing the diurnal/nocturnal behavioural pattern of fish and thus their cortisol rhythms: fish fed at MD were nocturnal and had a cortisol peak at ZT23, while fish fed at ML were mostly diurnal and showed a cortisol peak around ZT7 (Montoya et al., 2010). In our trial, however, fish were not strictly diurnal despite displaying FAA at ML. Furthermore, gilthead sea bream has been reported to show dual behavioural patterns, with seasonal inversions from diurnal to nocturnal behaviour (Velázquez et al., 2004). The fact that our experiment was carried out in winter-early spring, when sea bream shows nocturnal behaviour (Velázquez et al., 2004), may explain the shifts in plasma cortisol rhythms.
Glucose levels in the control fish showed a daily rhythm with the acrophase located around 10 h later than mealtime (ZT6), which is consistent with a previous investigation reporting that in this species glucose concentration peaked 8 h after feeding, regardless of mealtime (Montoya et al., 2010). Variations in plasma cortisol and glucose levels are closely related, especially under stress conditions, since after exposure to an acute stressor, cortisol elevation acts as a gluconeogenic signal increasing blood glucose levels. Actually, in our experiment, the stressed sea bream showed higher plasma cortisol and glucose levels than the control fish, which supports previous results obtained in gilthead sea bream subjected to air exposure (Arends et al., 1999). Interestingly, the stress-induced increase in plasma cortisol was higher during the dark phase, suggesting that stress response show daily rhythmicity in this species. Recent investigations carried out in Senegalese sole, a nocturnal fish, pointed also to the existence of daily rhythms in HPI-axis endocrine factors, as well as differences in the stress response depending on the time of day. However, contrasting with sea bream, Senegalese sole showed the highest cortisol and glucose levels when they were stressed in the middle of the light phase (López-Olmeda et al., 2013), indicating that response to acute stress (i.e. air exposure) is species-specific. Furthermore, previous investigations in rainbow trout have showed that the serotonergic system could play a role in triggering the initial steps of the activation of both HPI and HSC axis in fish (Gesto et al., 2013) and that increased cortisol synthesis in head kidney under stress conditions could be linked to the hyperglycaemia elicited by catecholamines (Gesto et al., 2014). Nevertheless, in this study, glucose levels in the stressed fish were higher than in the control ones at all sampling times and did not show daily rhythmicity, whereas the increase of plasma cortisol was higher during the dark phase. In mammals, there is evidence of a rhythm in the sensitivity of the adrenal gland to ACTH, which might be controlled by neural mechanisms (Engeland & Arnhold, 2005). Furthermore, a peripheral clock in the adrenal gland itself also seems to be involved in the circadian control of glucocorticoid secretion (Dickmeis, 2009). So far, in fish species, there are no studies reporting the existence of a daily rhythm in the sensitivity of adenohypophyseal ACTH cells to CRH or the interrenal gland to ACTH. Therefore, further studies will be necessary to fully understand rhythmicity of stress response in gilthead sea bream.
Regarding hypothalamic gene expression of crh, a significant daily rhythm was observed in the control sea bream, with the acrophase located two hours before meal time, coinciding with the peak of locomotor activity. Similarly, in Senegalese sole, crh expression peaked at the beginning of their activity phase -at night - (López-Olmeda et al., 2013). In the stressed sea bream, however, crh expression levels showed daily oscillations, but no daily rhythmicity. It seems there is a rhythm under control conditions (basal rhythm) but not in response to air exposure (induction rhythm). As for crhbp, a peak of expression was observed in the stressed fish at ZT7, coinciding with the acrophase of crh expression. Contrasting, at that time of the day, plasma cortisol levels were lowest, suggesting that CRHBP might be binding and inactivating the ACTH releasing activity of CRH (Huising at al., 2004;Wunderink et al., 2011). Conversely, during the dark phase, the average expression of crhbp decreased, whereas cortisol levels increased. Nonetheless, no differences in crhbp expression were observed between control and stressed groups. Previous studies in rainbow trout (Oncorhynchus mykiss) showed that stress-induced response by the CRH-BP gene differs between brain regions and different stressors. Thus, after 24 h of hypoxic stress, hypothalamic crhbp expression decreased in dominant fish and remained at control levels in subordinate fish, whereas in telencephalon, crhbp expression increased significantly (Alderman et al., 2008). Therefore, further investigations on crhbp expression in different sea bream brain regions would be needed to clarify its role in the regulation of the HPI axis response to stress. Cytochrome c oxidase (COX) is an oligomeric enzymatic complex located in the inner membrane of mitochondria and it is considered to be a major site of regulation of mitochondrial oxidative phosphorylation (Kadenbach et al., 2000). At high intramitochondrial ATP/ADP ratio, COX IV is phosphorylated and therefore ATP synthesis is inhibited. On the contrary, food intake increases the mitochondrial NADH/NAD + ratio and the substrate pressure for COX IV, resulting in relief of ATP inhibition (Arnold & Kadenbach, 1997). In this study, in both control and stressed sea bream, a daily rhythm of coxIV expression was found, with the acrophases located close to meal time, indicating that feeding cycles might entrain the expression of this enzyme, which in turn would improve ATP synthesis following oxidation of reducing equivalents of nutrients. Previous studies have reported changes in hepatic cytochrome oxidase activity in fish exposed to toxic compounds and pathogens (Craig et al., 2007;Sinha et al., 2012;Tiwari & Singh, 2006). In gilthead sea bream liver, coxIV was transiently up-regulated after 24 h of confinement exposure, depending upon the nutritional background (Pérez- Sánchez et al., 2013). Our present results indicate that in sea bream liver, oxidative stress caused by air exposure would also affect coxIV expression in a timedependent manner.
UCPs are mitochondrial transporters that uncouple oxidative phosphorylation by net discharge of the proton gradient (Krauss et al., 2005). In gilthead sea bream, enhancement of metabolic rates after chronic confinement exposure significantly reduced hepatic ucp1 expression (Bermejo-Nogales et al., 2010). In our trials, significant differences between sampling points were found in the control group. Thus, ucp1 expression peaked at the beginning of the night (ZT15), which would result in the uncoupling of oxidative phosphorylation and thereby inhibition of ATP synthesis, in accordance with hepatic coxIV rhythm of expression, which showed lowest levels during the dark phase. However, in the stressed fish, no significant differences in ucp1 expression were observed between sampling points.
Peroxiredoxins are the most recently discovered family of antioxidant enzymes. Initially identified in yeast, they have been found in all kingdoms of life, playing a key role in the organisms defence against oxidative stress (Rhee et al., 2005). Furthermore, a recent study has reported that the oxidation-reduction cycles of prdx proteins constitute a universal marker for circadian rhythms in all domains of life (Edgar et al., 2012). In this regard, it must be noted that the daily differences in prdx5 expression in control fish are parallel to those in ucp1, showing two peaks (at ZT7 and ZT15) and suggesting the existence of temporal coordination between the antioxidant systems and mitochondrial respiration uncoupling to minimise the risk of oxidative stress. However, the physiological relevance of these daily differences remains unclear. In fish exposed to air, a daily rhythm of hepatic prdx3 and prdx5 expression was found, with the acrophase at the beginning of the light phase in both cases and only two hours apart, suggesting a time-dependent response of prdx3 and prdx5 expression to oxidative stress induced by air exposure. There appears to be a daily rhythm in induction, but not in basal prdx3-5 expression. In gilthead sea bream, previous investigations have reported that different stressors can exert an effect on prdx gene expression in liver and head kidney (Pérez-Sánchez et al., 2011. In addition, our results point that stress response shows daily rhythmicity and therefore up-or down-regulation of prdx3 and prdx5 expression show differences between sampling points. In summary, these results indicate that stress response shows daily rhythmicity in gilthead sea bream, although the phase of the rhythm differs among stress indicators (neuroendocrine and mitochondrial oxidative markers). Hence, in the stressed fish, the acrophase of the daily rhythm of plasma cortisol was located at MD, whereas the acrophases of coxIV, prdx3 and prdx5 rhythms were located during the light phase, which suggests that different timing mechanisms may be involved in the control of specific stress response. Taken together, these results indicate that cortisol responses are species-dependent (diurnal/ nocturnal behaviour). Therefore, the time of day should be considered when submitting fish to stressful conditions.