Activation of oxygen-responsive pathways is associated with altered protein metabolism in Arctic char exposed to hypoxia

Fish are regularly exposed to fluctuations in oxygen concentrations in the wild, often associated with increased water temperature and eutrophication. The ability to tolerate hypoxia varies between species and individuals, and depends on a fish's capacity to alter its behaviour and/or physiology when oxygen levels decrease (Farrell and Richards, 2009; Richards, 2010). Different species employ a wide range of strategies to increase survival during hypoxia, including increasing gill ventilation and cardiac output to enhance oxygen extraction from the environment (Farrell and Richards, 2009; Richards, 2010). Hypoxia-tolerant species such as crucian carp (Carassius carassius), oscars (Astronotus ocellatus) and goldfish (Carassius auratus) lower their metabolic rate to limit the use of anaerobic metabolism and the accumulation of toxic end-products (reviewed in Boutilier, 2001; Lewis et al., 2007). This is accomplished by suppressing energetically expensive processes such as reproduction, digestion and growth (Boutilier, 2001; Farrell and Richards, 2009).

Metabolic suppression during hypoxia can be partially accomplished by suppressing protein metabolism, which includes both protein synthesis and protein degradation. Hypoxia-tolerant species, such as the Amazonian oscar and crucian carp can suppress protein synthesis by as much as 95% in liver and 55% in muscle (Cassidy et al., 2018; Lewis et al., 2007; Smith et al., 1996) during hypoxia, accounting for a 20–36% decrease in metabolism (Cassidy et al., 2018). The adjustments of protein metabolism, if any, have not yet been characterized in hypoxia-sensitive species such as salmonids. Regulation of protein synthesis during hypoxia is well documented in mammals and is starting to be characterized in fish, yet there is little information on regulation of protein degradation. In mammals, the autophagy pathway is activated during hypoxia (Wouters and Koritzinsky, 2008), with little information on the other protein degradation pathways (ubiquitin proteasome pathway and calpains).

There are three major cellular signalling pathways involved in the hypoxia response: the hypoxia inducible factor (HIF) pathway, the unfolded protein response (UPR) and mTOR pathways (Cassidy et al., 2018; Fang et al., 2015; Wouters and Koritzinsky, 2008). When activated, these pathways eventually inhibit the rate of protein synthesis (Wouters and Koritzinsky, 2008; summarized in Fig. 1). The responses of these pathways are well characterized in mammalian tumours and are starting to be elucidated in fish exposed to hypoxia. During hypoxia, HIF-1α accumulates in cells and dimerizes with HIF-β which promotes the expression of genes involved in various processes such as angiogenesis and glycolysis (Neufeld et al., 1999; Schofield and Ratcliffe, 2004). HIF may also control the rate of protein synthesis through transcriptional regulation of BNIP3 and REDD1, which inhibit mTOR and protein synthesis (Wouters and Koritzinsky, 2008). In fish, HIF transcripts increase during hypoxia and its function during hypoxia appears conserved in fish (Baptista et al., 2016; Heinrichs-Caldas et al., 2019; Rimoldi et al., 2012; Shen et al., 2010). The UPR is activated by an accumulation of unfolded or misfolded proteins in the endoplasmic reticulum (ER) during hypoxia. Its primary functions include degrading misfolded proteins and inhibiting protein synthesis to limit the accumulation of misfolded proteins in the ER (Zhang and Kaufman, 2006). Briefly, PERK (protein kinase RNA-like endoplasmic reticulum kinase) is activated by autophosphorylation and phosphorylates eIF2α, which directly inhibits translation initiation and, therefore, protein synthesis (Ron and Walter, 2007). The phosphorylation of eiF2α during hypoxia was reported in the Amazonian oscar (Cassidy et al., 2018), and transcriptional activation of its downstream targets were reported in the common sole (Solea solea) (Mazurais et al., 2014), suggesting that this pathway is also conserved in fish. Finally, the mTOR pathway regulates, among other things, both protein synthesis and degradation in response to various effectors such as hypoxia, food availability or pathogens (Fader et al., 2015; Hay and Sonenberg, 2004; Johnston et al., 2011). mTOR phosphorylates its downstream targets, 4EBP1 and P70S6K, which directly regulate global protein synthesis (Hay and Sonenberg, 2004). During hypoxia in mammals, mTOR is regulated by multiple pathways, including HIF-dependent, and HIF-independent pathways that involve AMPK (Liu et al., 2006; Papandreou et al., 2008). AMPK is an energy sensor that regulates multiple cellular processes including glucose and protein metabolism. AMPK is activated during severe hypoxia, and its activation inhibits mTOR activity, thus inhibiting protein synthesis (Cam and Houghton, 2011; Mungai et al., 2011; Papandreou et al., 2008). There is limited information on the regulation of the mTOR pathway during hypoxia in fish. However, AMPK phosphorylation increased in goldfish and rainbow trout exposed to hypoxia, suggesting that the mTOR pathway should also be inhibited (Jibb and Richards, 2008; Williams et al., 2019). In mammals, there is also significant crosstalk between the three hypoxia-response pathways and each one (HIF, UPR or mTOR) can regulate the activity of one of the other pathways (Wouters and Koritzinsky, 2008).

Protein degradation is also regulated by hypoxia-responsive signalling pathways, although less is known on the regulation of the three major protein degradation pathways (autophagy, ubiquitin proteasome pathway and calpains) during hypoxia. Autophagy and the ubiquitin proteasome pathway are both controlled by cell signalling pathways, including the AKT/mTOR pathway (Johnston et al., 2011). During hypoxia, autophagy is activated by all three hypoxia-response pathways in mammals (Wouters and Koritzinsky, 2008). It is not known, however, whether the ubiquitin proteasome pathway or the calpain system are also regulated by hypoxia-response pathways such as HIF or UPR. Autophagy is responsible for degrading organelles and proteins with long half-lives and includes acidic proteases called cathepsins that degrade proteins and organelles inside vacuoles (Ciechanover, 2012). It is not known whether autophagy is regulated during hypoxia in fish, but a recent study linked autophagy and hypoxia in prawn (Macrobrachium nipponense) hepatocytes (Sun et al., 2019). The ubiquitin proteasome pathway is mostly responsible for degrading damaged proteins, and proteins with short half-lives (Glickman and Ciechanover, 2002). This pathway involves three groups of enzymes that tag proteins with a poly-ubiquitin chain that will then be recognized and degraded by the proteasome complex (Ciechanover et al., 1984). In mammals, one study observed a decrease in proteasome activity during hypoxia (Gozal et al., 2003), which is in accordance with a decrease in proteasome subunit transcripts observed in stickleback (Gasterosteus aculeatus) (Leveelahti et al., 2011). The third important degradation pathway is the calpain system. Calpains are small, calcium-dependent, proteases that selectively degrade proteins in the cytosol, especially in muscle (Goll et al., 2003). Calpain activity is thought to increase in mammals during hypoxia (Cui et al., 2015; Nanduri et al., 2009); however, this has yet to be evaluated in fish.

Our objective was to determine whether and how protein metabolism is deregulated in tissues of Arctic char submitted to cumulative, graded hypoxia trials. The hypotheses were that: (1) as observed in hypoxia-tolerant species, protein metabolism is regulated during hypoxia in an attempt to maintain energy homeostasis and that this regulation is carried by the aforementioned hypoxia-responsive pathways; and (2) the reorganization of protein metabolism during hypoxia differs among tissues that have drastically different metabolic rates and functions (i.e. liver, white muscle and cardiac muscle). Unlike many cyprinids that tolerate low-oxygen environments, salmonids require higher oxygen availability (Jones et al., 2008). Arctic char, Salvelinus alpinus, is the northernmost freshwater fish species, and is considered amongst the most hypoxia-sensitive salmonids (Jones et al., 2008). This low hypoxia tolerance makes Arctic char a good model to study the effects of hypoxia on protein metabolism and to study which hypoxia-responsive pathways are involved in fish tissues.

Arctic char were exposed to a protocol of graded decreases in air saturation (3 h at each of 100%, 50%, 30% and then 15% air saturation) and we measured the fractional rate of protein synthesis and indicators of protein degradation in the liver, white muscle and heart at each oxygen concentration. We found profound alterations of the rate of protein synthesis and the abundance of various markers of protein degradation, but these were highly tissue specific.

Juvenile Arctic char [Salvelinus alpinus (Linnaeus 1758) Fraser strain; body mass=60.9±1.6 g, mean±s.e.m.] were obtained from the Coastal Zones Research Institute Inc. (Shippagan, NB, Canada). Five hundred fish were held in two 250 liter tanks supplied with flow-through dechlorinated freshwater. Fish were fed a commercial salmon diet (Corey Nutrition Company Inc., Fredericton, NB, Canada) at 2% body weight per day, every other day. Water temperature varied seasonally and was 16°C in the summer, when the experiment occurred.

The experimental system consisted of three shelves each containing six 9 liter tanks. Each shelf had its own water filtration and biofiltration system. The dissolved oxygen was adjusted by nitrogen injection by means of a programmable proportional–integral–derivative controller (Aquabiotech, Coaticook, QC, Canada). Temperature was maintained at 16±0.1°C. In June 2017, fish were semi-randomly selected (excluding the runts) and transferred to experimental tanks over 5 days. Fish were food deprived for 4 days before the experiment. Twenty fish (5 per group) were exposed to a cumulative, graded hypoxic challenge, with an endpoint of either 100%, 50%, 30% or 15% air saturation (Fig. 2) as in Cassidy et al. (2018). Five fish (1 per tank) were placed in the experimental chambers overnight at 100% air saturation for acclimation and sampled the next day (control group). The next day, 5 additional fish were placed in the chambers overnight at 100% air saturation for acclimation. Air saturation was then dropped to 50%, and maintained at 50% for 3 h before sampling. A third group of 5 fish were then placed in the tanks overnight at 100% air saturation. Air saturation levels were then reduced to 50% for 3 h the following morning, then held at 30% for 3 h before sampling. The final group of fish was placed in the tanks overnight at 100% air saturation. The next morning, air saturation levels dropped to 50% for 3 h, then 30% for 3 h and finally 15% for 3 h before sampling. We did not expose fish to oxygen concentrations lower than 15% since at this level fish were starting to lose equilibrium. Oxygen concentration measured by the system was always confirmed using a handheld oxygen meter (5512-Ft YSI, Yellow Springs, OH, USA). At the end of each hypoxic challenge, fish were killed by severing the spinal cord directly behind the brain. Blood was immediately collected from the caudal vein, centrifuged for 5 min at 2000 g and the plasma frozen. Samples of muscle, liver, heart, gill and blood were collected and immediately frozen in liquid nitrogen. Samples were then stored at −80°C until further analysis. Experiments were conducted with the approval of the animal care committee at the Université de Moncton, New Brunswick, Canada (UdeM17-05).

Lactate was measured in plasma, muscle and liver by first deproteinizing tissues with 6% perchloric acid followed by neutralization with 2 mol l⁻¹ KHCO₃. Lactate was measured in glycine buffer (320 mmol l⁻¹ glycine, 320 mmol l⁻¹ hydrazine, 2.4 mmol l⁻¹ NAD⁺) with subsequent treatment of excess lactate dehydrogenase (2 U ml⁻¹) (Brandt et al., 1980). Absorbance was read with a multimode plate reader (Varioskan Flash, Thermo Fisher Scientific, Inc., Waltham, MA USA) at 340 nm wavelength.

Western blots were used to analyse specific protein levels in liver, muscle and heart. Tissues were homogenized in 9 volumes of lysis buffer (50 mmol l⁻¹ Tris-HCl, 0.1 mmol l⁻¹ EDTA, 1.0 mmol l⁻¹ β-mercaptoethanol, pH 8) with a sonicating homogenizer and centrifuged at 13,000 g for 15 min at 4°C. Protein concentration was then measured (Bradford protein assay, Bio-Rad, Hercules, CA) (Bradford, 1976). Fifteen mg of protein was loaded in each well and resolved in a 10% SDS-PAGE, transferred to PVDF membranes (Millipore, Bedford, MA), and blocked for 1 h 5% BSA dissolved in TBS. Membranes were probed overnight at 4°C in primary antibodies (Cell Signaling Technologies, Beverley, MA, USA): 4EBP1 (no. 9649), p-4EBP1 (Thr37/46; no. 2855), eIF2α (no. 5324) and p-eIF2α (Ser51; no. 3398), AMPKα (no. 5831), p-AMPKα (Thr172; no. 2535). Protein bands were revealed by probing for 1 h with HRP-conjugated goat anti-rabbit secondary antibody (no. 7074) followed by enhanced chemiluminescence (Clarity Western ECL substrate, Bio-Rad, Hercules, CA). Images were taken with a Chemidoc Touch Imaging System (Bio-Rad). The densitometric analyses were performed using Image Lab v.5.2.1 software (Bio-Rad). Levels of phosphorylation were obtained by calculating the ratio between phosphorylated and total proteins (e.g. phospho-eIF2α/total eIF2α). Protein bands were normalized using the stain-free imaging technology (Gürtler et al., 2013). The relative quantity of polyubiquitinated proteins was determined by dot blots using an antibody detecting polyubiquitinated proteins (mAb FK1, Enzo, BML-PW8805) and an anti-mouse IgM HRP-linked antibody (ab97230, Abcam, Cambridge, MA) as previously described (Cassidy et al., 2016).

Transcripts related to the hypoxia response and protein degradation were amplified via quantitative (q)PCR in liver, muscle and heart. Approximately 50 mg of tissue (15 mg for heart) were homogenized in 1 ml of TRI Reagent (Sigma-Aldrich, St Louis, MO, USA) with a TissueLyser (Qiagen,Venlo, The Netherlands) and 5 mm stainless steel beads for 5 min at 27 Hz. RNA was quantified by UV absorption at 260 nm using a NanoVue spectrophotometer (GE Healthcare, Chicago, IL, USA). RNA was treated with a DNA-free kit (Life Technologies, Carlsbad, CA, USA) and then reverse transcribed with qScript cDNA SuperMix (Quanta Biosciences Inc., Gaithersburg, MD, USA). Primers were designed with Primer3 (v.0.4.0) (Koressaar and Remm, 2007; Untergasser et al., 2012) using Atlantic salmon (Salmo salar) gene sequences obtained on NCBI (http://www.ncbi.nlm.nih.gov/) (Table S1). Gene sequences were aligned with zebrafish (Danio rerio) and mouse (Mus musculus) to obtain the most conserved regions of the genes using Mega7 software (Kumar et al., 2016). Primers were validated by sequencing the PCR products (Centre de Recherche du CHUL at Université Laval) and using BLAST (NCBI) searches. qPCR assays were conducted with CFX Connect Real-Time PCR Detection System (Bio-Rad) and respected the minimum information for publication of quantitative PCR experiment guidelines (Bustin et al., 2009). Amplification efficiencies were assessed for each tissue. Triplicates of qPCR reactions contained 5 µl of SsoAdvanced Universal SYBR Green Supermix (Bio-Rad), 4 µl of cDNA, 400 nmol l⁻¹ of forward and reverse primers and RNase-free water to a final volume of 10 µl. Data were normalized by integrating the geometric average of the 3 most stable reference genes into the Pfaffl equation (Pfaffl, 2001; Vandesompele et al., 2002). E1fα, α-tubulin and rs11 were the most stable reference genes in liver; E1fα, β-actin and rs11 were selected in muscle; and in heart, E1fα, β-actin and α-tubulin were selected. Normalized mRNA values are expressed relative to the average of the control group.

Linear models (lm function) with air saturation levels as the fixed factor (one-way ANOVA) and Tukey's post hoc were performed in R (https://www.r-project.org) statistical software. Box-cox transformations were applied when needed, to improve the normality of the residuals. P-values smaller than 0.05 were considered statistically significant. The figures were constructed with Prism 7 software (GraphPad Software Inc.).

No mortalities were observed during the cumulative, graded hypoxia treatments although a few fish started losing equilibrium at 15% air saturation. Lactate concentration in plasma and liver showed a gradual increase (P<0.001, n=5) as % air saturation levels decreased (Fig. 3). In muscle, a higher concentration of lactate occurred only at 15% oxygen saturation (P<0.001, N=5). Lactate concentration was omitted in the heart because of limited availability of tissue.

The fractional rate of protein synthesis was measured in gill, liver, digestive system, white and red muscle, and heart of fish at each stage of hypoxia (Fig. 4). In most tissues, protein synthesis decreased with % air saturation, but this trend was not observed in the heart, where protein synthesis remained more consistent (Fig. 4F). The liver and digestive system were the most responsive tissues, with rates of protein synthesis decreasing after only 3 h at 50% air saturation (Fig. 4B,C).

Markers of the ubiquitin proteasome pathway, calpain and autophagy pathways were measured in liver, muscle and heart (Tables 1, 2 and 3, respectively). Ubiquitin proteasome pathway markers were unaffected in liver and muscle, but in heart, Fbxo25 and UBE2H transcript levels were decreased at 15% air saturation (P<0.01, N=5; Table 3). Polyubiquitinated protein levels were unchanged in all tissues (Fig. S1). In the calpain pathway, calpain7 transcript levels increased at 15% air saturation in liver and decreased in heart (P=0.004, N=5). Calpastatin transcript levels also increased in liver at 50%, 30% and 15% air saturation (P=0.003, N=5), but did not change in muscle and heart. The autophagy-related transcripts Cathepsin D and LC3B increased in liver as air saturation decreased (P<0.001, N=5; Table 1). In heart, transcripts levels of Cathepsin L (P=0.04, N=5), Atg4 (P=0.04, N=5), and Beclin1 (P=0.001, N=5) were lowered at 15% air saturation (Table 3).

In the mTOR pathway, the phosphorylation levels of AMPKα, 4EBP1 were measured in liver, muscle and heart (Fig. 5). AMPKα phosphorylation increased in the liver and heart at 15% air saturation (P<0.001, N=5; Fig. 5A–C). In all three tissues, phosphorylation levels of 4EBP1 decreased at 15% air saturation compared with controls (P<0.001, N=5; Fig. 5D–F). Activation of the unfolded protein response pathway was assessed by measuring eIF2α phosphorylation (Fig. 5G–I) and transcript levels of eIF2α, ATF4 and DDIT3. The phosphorylation levels of eIF2α were higher at 15% air saturation in the liver and muscle (P<0.001, N=5; Fig. 5G,H). In the heart, eIF2α phosphorylation decreased at 50% and 30%, before returning to control levels at 15% (P=0.01, N=5; Fig. 5I). The hypoxia treatments had no effect on eIF2α transcripts in any of the tissues and levels of ATF4 transcripts were only affected in the heart (P=0.02, N=5) where transcript levels decreased at 15% air saturation (Table 3). DDIT3 transcripts showed opposite responses in the liver and muscle. DDIT3 transcript levels gradually increased in the liver as air saturation decreased (P=0.02, N=5; Table 1), while DDIT3 transcripts decreased in muscle with decreasing air saturation (P=0.01, n=5; Table 2). No changes occurred for DDIT3 transcripts in the heart (Table 3). Commercial antibodies for HIF-1 working in salmonids are not available, therefore we used the transcript numbers of genes involved in the pathway. In the liver, transcript levels for HIF1a (P<0.001, N=5) and its target VEGF (P=0.007, N=5) were elevated in fish exposed to 50% air saturation, and this elevation was maintained at 30% and 15% (Table 1). VHL transcript numbers in the liver were not significantly affected although they showed a similar trend (P=0.06, N=5) to VEGF (Table 1). In muscle, HIF1a transcripts remained unchanged, while VEGF transcript levels decreased at 30% and 15% air saturation (P<0.001, N=5), and VHL increased at 50% and 30% air saturation (P=0.004, N=5; Table 2). In the heart, there were no changes for HIF1a and VHL transcript levels, but VEGF transcripts increased at 15% air saturation (P<0.001, n=5; Table 3).

Our objectives were to determine whether protein metabolism was deregulated in different tissues of Arctic char during cumulative, graded hypoxia. The fish were exposed to a graded hypoxia protocol with a first step of hypoxia at 50% air saturation for 3 h. At this concentration, the fish were already in a hypoxic metabolic state, as demonstrated by the elevated plasma and liver lactate concentrations (van Raaij et al., 1996; Via et al., 1997; Wood et al., 2007; Zhou et al., 2005). We should also acknowledge that during each subsequent trial (30% and 15% air saturation) more time was allowed for biochemical responses as well as for gene expression to occur, which could bias the comparisons of the hypoxia response at each level. Nevertheless, our study showed that protein metabolism in Arctic char is altered in a tissue-specific fashion during graded hypoxia, which is in accordance with the responses of the three major hypoxia-sensitive pathways (HIF, UPR and mTOR), as observed in mammalian cells.

The rate of protein synthesis was highly responsive to graded hypoxia in Arctic char. The effects of hypoxia on the rate of protein synthesis is documented in hypoxia-tolerant fish species (Cassidy et al., 2018; Lewis et al., 2007; Smith et al., 1996; Zhou et al., 2005), but to our knowledge, we report for the first time that it is also the case in hypoxia-sensitive species such as Arctic char. At 50% air saturation, some tissues such as the liver and the digestive tract were already responding to hypoxia by decreasing their rate of protein synthesis by ∼20% air saturation, while protein synthesis in the heart seemed to be insensitive to graded hypoxia, even at air saturation of 15%. Overall, the liver and digestive systems showed the most dramatic decreases of protein synthesis rates, reflecting the sensitivity of these tissues to changes in oxygen availability and their triviality for short-term survival. Skeletal muscle constitutes over 50% of the fish's body mass; therefore, suppressing muscle protein synthesis by 50% significantly decreases resting metabolic rate. The only tissue that maintained its rate of protein synthesis throughout the graded hypoxia experiment is the heart, contrasting with our previous findings in the hypoxia-tolerant oscar (Cassidy et al., 2018). The heart of most fish, including oscars, receives oxygen from venous blood (Farrell, 2002); therefore, when oxygen is limiting, cardiac functions are impaired (Farrell, 2002). In oscars, since energy demand during hypoxia is so low, a decrease in heart function could have little effect on survival. In the salmonid heart, about 30% of the myocardium is supplemented by a coronary artery (Farrell, 2002), which probably explains how the Arctic char's heart maintained protein synthesis during hypoxia. This anabolic state may also be fuelled by the increase in plasma lactate, one of the preferred metabolic substrates of salmonid hearts (Milligan and Farrell, 1991).

The effects of hypoxia on protein degradation-related markers have scarcely been studied in fish (Cassidy et al., 2018). Our results provide strong evidence that protein degradation plays an important role in the hypoxia response. Markers of the ubiquitin proteasome pathway, autophagy and calpain system showed different trends among tissues. Overall, muscle protein degradation was maintained during hypoxia, supporting the lack of responses measured in the oxygen-response pathways. However, the liver and heart showed large but opposing effects of hypoxia on protein degradation markers. In mammals, hypoxia is a well-documented inducer of autophagy (Fang et al., 2015; Schaaf et al., 2013), but this is not yet verified in fish. In the liver, autophagy was activated during graded hypoxia, as shown by the elevation of LC3B transcripts, the hallmark marker of autophagy activation. This increase in autophagy is similar to what is observed in mammals, and likely reflects the liver's sensitivity to hypoxia. Perhaps surprisingly, the ubiquitin proteasome pathway markers were unresponsive to hypoxia in the Arctic char liver, contrasting with a downregulation of proteasome-subunit mRNAs, previously reported in hypoxic three-spine stickleback (Leveelahti et al., 2011). It is likely that this discrepancy results from the different hypoxia protocols the two experiments are using but this issue remains to be clarified by including a higher number of genes and looking at protein abundance.

In the heart, graded hypoxia was accompanied by an unexpected downregulation of protein degradation markers of all three pathways. Ubiquitin proteasome and calpain markers (Fbxo25, UBE2H and Calpain7) decreased, suggesting a downregulation of these two pathways. Transcripts from the autophagy pathway also decreased at 15% air saturation, which is opposite to what we observed in Arctic char liver and what is known in mammals (Mazure and Pouysségur, 2010; Wouters and Koritzinsky, 2008). We offer two potential explanations for this. First, decreasing protein degradation might protect the heart's main functions by allocating more energy towards other more vital ones, such as contractions and protein synthesis. It is also possible that during hypoxia, the Arctic char heart is in an anabolic state supplied by an increase in plasma lactate. This hypothesis is also supported by the maintained rates of protein synthesis in the heart. A decrease in protein degradation during an anabolic state would maximize protein retention and thereby increase heart growth in the event of long-term hypoxia exposure, or during multiple exposures (diel cycling). An enlarged heart could then increase cardiac output, to increase the Arctic char's ability to deliver oxygen to tissues (Driedzic et al., 1996; Marques et al., 2008). These results warrant further investigation to understand the role of protein degradation inhibition in the heart during hypoxia.

We measured the activation of the three major hypoxia-response cellular signalling pathways (HIF, UPR and mTOR) in liver, muscle and heart by looking at the phosphorylation of key proteins involved in each pathway and/or the transcript numbers of genes under their control. These pathways promote energy conservation and enhance survival under low oxygen conditions (Liu et al., 2006). Our results confirm that these pathways are also regulated in hypoxia-sensitive fish and in a tissue-specific manner.

In liver, the hypoxia-response pathways were regulated as documented in mammals (Wouters and Koritzinsky, 2008). First, HIF was activated, and later UPR was activated during the more severe graded hypoxia exposure, while the mTOR pathway was inhibited. The HIF pathway was activated early during hypoxia as shown by the rapid elevation of its target gene, VEGF transcripts (Wouters and Koritzinsky, 2008). This might also be supported by the observed increase of HIF1a transcripts. Although measuring HIF1a transcripts is not directly related to HIF accumulation in tissues, transcript levels increased during hypoxia in several fish species and this has been associated with HIF activation (Heinrichs-Caldas et al., 2019; Rimoldi et al., 2012; Ton et al., 2003). The UPR was activated during severe graded hypoxia, as shown by the accumulation of phosphorylated eIF2α. This is similar to what is observed in livers in the oscar (Cassidy et al., 2018). The gradual activation of AMPK as the graded hypoxia treatment progresses, eventually translated into an inhibition of the mTOR pathway (shown by the dephosphorylation of 4EBP1). These results are similar to goldfish that increased AMPK activity in the liver during hypoxia (Jibb and Richards, 2008). The dephosphorylation of 4EBP1 results in a depression of cap-dependent translation consistent with the depression of the rate of protein synthesis we observed (Richter and Sonenberg, 2005). Overall, the hypoxia-response pathways were strongly regulated in the liver, and in accordance with the large decrease in protein synthesis and increase of autophagy markers. These results suggest that when oxygen decreases and blood becomes hypoxemic, oxygen delivery curtails early in the liver.

In muscle, the hypoxia-response pathways were regulated differently than in the liver. The transcript numbers of VEGF decreased as hypoxia progressed which would be consistent with a greater destabilization of HIF in this tissue. Furthermore, the transcript numbers of VHL, which is responsible for the selective degradation of HIF (Turcotte et al., 2008), were rapidly elevated during hypoxia. Downregulation of the HIF pathway in muscle during hypoxia is surprising but suggests that muscle vascularization was impeded. To our knowledge, VEGF transcripts were never measured in the skeletal muscle of fish during hypoxia. The UPR was activated during the most severe hypoxia treatment in muscle, and mTOR was inhibited (decreased 4EBP1 phosphorylation), which is consistent with the decreased rates of protein synthesis in muscle. The regulation of protein synthesis via signalling pathways is important to reduce energy expenditure and the load of proteins in the ER, even in muscle, which otherwise appeared to have low sensitivity to hypoxia.

Out of the three examined tissues, the heart was the least responsive to hypoxia. There was a late activation of the HIF pathway, as VEGF transcripts increased at 15% air saturation and the UPR pathway was not activated. Similarly, AMPK and mTOR pathways responded as if the heart became anabolic during intermediate oxygen concentrations. The pattern of activation of these oxygen-responsive signalling pathways indicates that the heart is one of the last tissues to become hypoxic, probably as a result of its preferential oxygen supply via the coronary artery (Gamperl et al., 1995). The apparent anabolic state of the heart during hypoxia is metabolically fuelled by the circulating muscle-derived lactate as suggested by Lanctin et al. (1980) and Milligan and Farrell (1991).

We show that despite being a hypoxia-intolerant species, Arctic char demonstrate considerable ability to alter protein metabolism, which could be under the control of hypoxia-responsive cellular signalling pathways (HIF, UPR and mTOR). The activation pattern of these pathways and the cellular processes that are under their control varies greatly among tissues, sometimes even going in opposite direction. This study provides new insights on the effects of hypoxia on protein metabolism. The adjustments of these cellular processes probably all contribute to shifting the fish phenotype to a more hypoxia-tolerant one. Here, we measured the response of protein metabolism to a single and cumulative, graded hypoxia event in fish that were naïve to hypoxia. Our results warrant studying these adjustments in fish exposed to long-term and diel cycling hypoxia.

The authors are grateful to André Dumas and Claude Pelletier at the Coastal Zones Research Institute (Shippagan, NB, Canada) for providing the fish.