eIF2α phosphorylation in response to nutritional deficiency and stressors in the aquaculture fish, Rachycentron canadum

: The present study investigates the response of the marine fish cobia, Rachycentron canadum , to stressors as measured by phosphorylation of the α -subunit of the translational initiation factor, eIF2. eIF2α is the target of phosphorylation by a family of kinases that respond to a range of physiological stressors. Phosphorylation of eIF2α not only inhibits overall protein synthesis, but allows cells to reprogram gene expression to adapt to, and recover from, stress. The deduced coding sequence of cobia eIF2α has 94 % identity to both zebrafish ( Danio rerio ) and human eIF2α sequences with identical phosphorylation and kinase docking sites. The present study uses a cobia cell line, Cm cells, derived from muscle, as well as cobia larvae to investigate the response of cobia eIF2α to various stressors. In Cm cells, phosphorylation of eIF2α is increased by nutrient deficiency, leucinol, and ER stress, consistent with the activation of the eIF2 kinases, GCN2, and PERK. In cobia juveniles, diet and water temperature affect the phosphorylation state of eIF2α. We conclude that evaluation of eIF2α phosphorylation could function as an early marker to evaluate diet, aquaculture stressors and disease in cobia and may be of particular use in the optimization of conditions for raising cobia larvae and juveniles. seabass (43), red sea bream (44), turbot (45), flounder (46; 47), sea perch (48) and gilthead seabream (49; 50). The isolation of two cobia cell lines, Cb and Cf from brain and fin, respectively, was previously reported and characterized, but the lines were sub-sequently lost (51). In this study we report the establishment and use of a new cell line from cobia muscle, Cm cells. To facilitate studies on Cm cells and cobia, we have success-fully cloned cobia eIF2α cDNA using degenerate primers based on eIF2α sequences from other fish species and completing the coding sequence using 5’ & 3’ RACE. The current work demonstrates that recombinant cobia eIF2α can be phosphorylated by stress-acti-vated eIF2α -kinases and that Cm cells respond to activators of eIF2α -kinases by increasing eIF2α phosphorylation. Furthermore, preliminary studies show that changes in eIF2α phosphorylation can arise from nutritional deficiencies and temperature changes in cobia


Introduction
R. canadum is a perciform pelagic fish species found in warm-temperate to tropical waters of the West and East Atlantic Ocean, throughout the Caribbean, and in the Indo-Pacific off India, Australia and Japan. It is eurythermal, tolerating a wide range of temperatures, from 1.6 to 32.2 °C. It is also euryhaline, living at salinities of 5 to 44.5 ppt. Cobia is sold commercially and commands a relatively high price for its firm texture and excellent flavor. However, no designated wild fishery exists because it is a solitary species. It is currently farmed in aquaculture (1)(2)(3) and exhibits rapid growth in both off-shore cage systems as well as marine recirculating aquaculture tanks (4). Cobia seed stock used for large scale commercial aquaculture production comes exclusively from hatcheries and fingerling production is still an industry-wide bottleneck despite improvements in cobia larviculture (3; 5). Problems exist in larvae/fingerling culture, including high mortality due to stress and disease in the larval stage and the reliance on unsustainable fish meal diets. Expansion of cobia farming requires the development of sensitive assays to monitor stress during larval culture as well as the optimization of plant-based diets to increase production and sustainability (6; 7). Diet studies are both time consuming and expensive and could be facilitated by the development of faster indicators of diet quality. Similarly, a sensitive indicator of stress has the potential to assist in optimization of larviculture conditions. Phosphorylation of Ser51 of the α-subunit of the translational initiation factor eIF2 is a key cellular response to wide range of environmental stresses such as nutrient deficiency, hypoxia, endoplasmic reticulum stress and viral infection in eukaryotes from protists to vertebrates (8)(9)(10). There are five eIF2α-specific kinases in vertebrates (11), five in teleost fish and amphibians, four in tetrapods (12)(13)(14) that can phosphorylate eIF2α. Each of these is activated by different stressors. HRI (EIF2AK1) is stimulated by heme depletion in erythroid cells, but also responds to oxidative stress, osmotic shock, and heat shock (15)(16)(17). HRI is the only eIF2α kinase activated by arsenite in erythroid cells (15; 18) and is the major eIF2α kinase responsive to heat shock. PKR (EIF2AK2) is stimulated by viral infection (19; 20), but also plays a more general role in cellular physiology (21); it can be activated in response to signals as diverse as oxidative and ER stress (22)(23)(24), as well as cytokine and growth factors signaling (22; 25) and has been implicated in the pathology of obesity (26; 27). PERK (EIF2AK3) is an endoplasmic reticulum (ER) transmembrane protein activated by the accumulation of misfolded proteins in the ER, a phenomenon termed the unfolded protein response (UPR) or ER stress (28). Changes in Ca 2+ levels within the ER negatively affect the ability of the chaperone protein, binding immunoglobulin protein (BiP), also known as 78 kDa glucose-regulated protein (GRP-78), to maintain PERK in its inactive state (29). This allows diverse signals such as inhibition of the SERCA calcium pump, glucose deprivation, and high levels of fatty acids reduce the ER luminal calcium concentration (30; 31). GCN2 is the primary responder to nutritional deprivation and is the only eIF2α-kinase conserved among plants and metazoans. It is activated by reduced amino acid levels in yeast and mammalian cells through uncharged tRNAs and the ribosomal P-stalk (32)(33)(34)(35) and can also be activated by glucose deprivation (36). GCN2 is also involved in a variety of organismal functions in vertebrates such as long-term memory formation and feeding behavior (37). These four members of the eIF2α kinases share extensive homology in their kinase catalytic domains and phosphorylate eIF2α at the same serine residue (Ser51) (reviewed (11)). Teleost fish and amphibians also have a PKR-like kinase, PKZ, that is more closely related to the kinase domain of PKR compared to other three known eIF2α kinases and, like PKR, is involved in the innate immune response (12; 14; 38). Phosphorylation of eIF2α on Ser51 or its equivalent reduces the rate of protein synthesis overall. However, phosphorylation of eIF2α increases the recruitment of a subset of mRNAs resulting in the transient activation of an alternate program of gene expression that enables the cell or organism to respond to a range of stressors (9; 39).
With the known relationship between eIF2α phosphorylation and nutritional status and food choices in mammals, it was of interest to determine whether eIFα phosphorylation could be used as an early marker to evaluate diets and aquaculture conditions in fish. Studies were initiated in zebrafish, a tractable model system, to lay the groundwork for looking at fish of more interest to aquaculture. Zebrafish ZFL cells respond to a variety of stressors that activate a range of eIF2α-kinases (Liu et al m/s in preparation). However, it was of more interest to see if cells from a marine species of aquaculture interest, such as cobia, would respond in the same way. A number of cell lines have been described from economically important marine species. These include cell lines from groupers (40; 41), tilapia (42), seabass (43), red sea bream (44), turbot (45), flounder (46; 47), sea perch (48) and gilthead seabream (49; 50). The isolation of two cobia cell lines, Cb and Cf from brain and fin, respectively, was previously reported and characterized, but the lines were subsequently lost (51). In this study we report the establishment and use of a new cell line from cobia muscle, Cm cells. To facilitate studies on Cm cells and cobia, we have successfully cloned cobia eIF2α cDNA using degenerate primers based on eIF2α sequences from other fish species and completing the coding sequence using 5' & 3' RACE. The current work demonstrates that recombinant cobia eIF2α can be phosphorylated by stress-activated eIF2α-kinases and that Cm cells respond to activators of eIF2α-kinases by increasing eIF2α phosphorylation. Furthermore, preliminary studies show that changes in eIF2α phosphorylation can arise from nutritional deficiencies and temperature changes in cobia found at 28 °C in 10 % FBS. Cells are fibroblast-like in appearance (Figure 1). Figure 1. Cm Cells in culture resembling fibroblast morphology.

Identification of Cm cell origin
Analysis of the sequence of mitochondrial cytochrome oxidase subunit I gene (cox I) was used to verify whether the cell lines were actually derived from cobia. RNA was isolated from Cm cells and fresh cobia muscle using the PureLink® RNA Mini Kit (Ambion) following the manufacturer's protocol. cDNA was synthesized using Revertaid™ M-MULV reverse transcriptase and random hexamer primers to produce first strand cDNA. Amplification of cDNA fragments of the predicted size was confirmed by end-point PCR amplification using the primers described from FDA, (52). PCR products were separated using 1% TAE (tris-acetate-EDTA) agarose gel electrophoresis and stained with ethidium bromide. The cDNA products were recovered using a QIAquick PCR purification Kit (Qiagen) followed by sequencing. CLC and NCBI blast were used to compare sequence identity.
Cobia cell extract samples were fractionated by 15 % SDS-PAGE, electrotransferred to PVDF membranes, and subjected to immunoblot analysis using polyclonal antibody against phosphorylated Ser51 of mammalian eIF2α (CellSignaling, Danvers, MA, USA) followed by goat anti-rabbit secondary antibody coupled to HRP and coupled with chemiluminescence. Chemiluminescence was detected using the ProteinSimple Fluorochem E (Pasadena, CA, USA) with quantification using ImageJ software. Total eIF2α levels were determined by reprobing the stripped blot with rabbit antibody that recognizes equally the phosphorylated and unphosphorylated forms of eIF2α. Values obtained with the anti-eIF2α[P] antibody were normalized for the total amount of eIF2α present in the sample.

Experimental fish and systems
This study was carried out in accordance with the guidelines of the International Animal Care and Use Committee of the University of Maryland Medical School (IACUC protocol # 0610015). Juvenile (~2 g) cobia, R. canadum, were obtained from the Virginia Agricultural Experiment Station, Virginia Tech, Hampton, VA. Juveniles were maintained at 27 °C in a marine recirculating system at the Institute of Marine and Environmental Technology's Aquaculture Research Center, Baltimore, MD. Diet formulations are given in Supplemental Table S2.

Preparation of cobia liver samples
Samples of liver from cobia juveniles maintained on a range of diets and at different temperatures were collected, snap-frozen in liquid N2 immediately and stored at -80°C for later analyses. Tissue extracts were homogenized using lysing matrix D (MP Biomedicals, Solon, OH) (3 lysis buffer: 1 tissue ratio). Tissue was disrupted using FastPrep-24 5G (MP Biomedicals, Solon, OH) for 40 sec, kept on ice for 5 min and microcentrifuged at 12,000 g for 15 min at 4 oC. Supernatants were snap-frozen in liquid N2 immediately and stored at -80°C for later analyses. Phosphorylation of eIF2α was determined by immunoblot analysis of SDS-PAGE fractionated proteins as described (54).

Identification of Cm cell origin
The identification of Cm cells as cobia derived was confirmed by sequence comparison of cox 1 cDNA from Cm cells and fresh muscle. Using each of two primer sets, the PCR-amplified cox 1 cDNA amplicons from Cm cells and fresh muscle were of the same predicted sizes; ~ 650 bp band after agarose gel electrophoresis. DNA sequencing and analysis demonstrated that cox 1 cDNAs amplified from fresh cobia muscle tissue and from Cm cells were identical. Moreover, all cox 1 cDNA sequences amplified from RNA from Cm cells or fresh cobia muscle were 99%~100% identical with (GenBank accession # FJ154956.1) and cobia cytochrome oxidase subunit 1 (COI) gene (GenBank assession # KJ202194.1) (Figure 2). These data confirmed that the origin of the established Cm cells is from cobia.  Table S1. Amplicons were fractionated by 1 % agarose gel electrophoresis. Multiple alignments of amplified cobia COI gene from cobia muscle cell line and cobia muscle in comparison with cobia cytochrome oxidase subunit 1 (COI) gene (GenBank accession # KJ202194.1) by the ClustalW program were show below. The red arrows indicate the primer position.

Cobia have two eIF2α transcripts but identical coding sequences
Two transcripts of eIF2α are found in cobia, although the coding sequence for each is identical (accession # KJ513464). Compared to human and zebrafish eIF2α, the cobia coding sequence is 94 % identical at the amino acid level. Table 1 shows the percent identities in amino acid composition, predicted molecular weights and isoelectric points. The predicted isoelectric points and molecular weights of the eIF2αs are in good agreement with those of human and zebrafish eIF2α. The cobia eIF2α is predicted to be 35.97 kDa, respectively, compared with human eIF2α which has a predicted molecular weight of 36.1 kDa. The deduced isoelectric points are also very similar, 4.84 for cobia eIF2α compared to 5.02 for human eIF2α. Figure 3 shows the multiple alignment of the N-terminal 120 amino acids of cobia eIF2α compared to human and zebrafish eIF2αs. This covers the phosphorylation site and kinase docking domain. (A multiple alignment of the complete amino acid sequences are provided as supplementary Figure S1). Residues in the phosphorylation loop region surrounding the eIF2α phosphorylation site (Ser-51 in human eIF2α) are identical. Similarly, residues in the PKR docking site (residues 79-83) of cobia eIF2α are identical to those in the human and zebrafish sequences, suggesting it is a good substrate for eIF2α-specific protein kinases. In addition, Ser-48, Ile-55, Leu-84, Arg-88 and Val-89, residues critical for interaction with the α-, β-, and δ-subunits of the guanine nucleotide exchange factor, eIF2B (55), are present in cobia eIF2α indicating that the interaction between both eIF2αs and eIF2B is also conserved. The first α-helix of human eIF2α is known to be involved in interactions that fix the orientation of the helical domain with the β-barrel (56). In mammalian eIF2α, this region contains a cysteine residue that is involved in a disulfide bridge and is thought to stabilize the interaction of the two domains. This residue is not found in non-vertebrate eIF2αs and appears to be a characteristic of vertebrate eIF2αs. Each of the four eIF2α-kinases, HRI, PKR, PERK and GCN2 can be activated in mammalian cells by a variety of stressors (reviewed (10)). Thapsigargin is an inhibitor of endoplasmic reticulum Ca 2+ ATPase that blocks the ability of the cell to pump calcium into the sarcoplasmic and endoplasmic reticula which causes these stores to become depleted (57). This causes endoplasmic reticulum (ER) stress and activates the unfolded protein response, including the activation of PERK (58; 59). GCN2 is activated by amino acid and glucose starvation (36; 60). Leucinol is the alcohol formed by total reduction of the carboxylic acid group of leucine. It is an inhibitor of leucyl-tRNA synthetase and so can mimic leucine deficiency in cells (61). N-methylprotoporphyrin (NMPP) is a potent inhibitor of ferrochelatase (Ki∼10 nM), the terminal enzyme of the heme-biosynthetic pathway (62) and is able to activate the heme-sensitive kinase HRI (63). Polyinosinic:polycytidylic acid (poly I:C) is a synthetic dsRNA that is used experimentally to model viral infections in vivo (reviewed (64) and is used in vivo and in vitro to activate PKR.
The response of Cm cells to these known activators of eIF2α-kinases was examined, as shown in Figure 4. Cm cells treated with known activators of eIF2α-kinases; incubation in nutrient limiting medium or in the presence of leucinol (4 µM) for activation of GCN2; thapsigargin (1 µM) for activation of PERK; poly(I):poly(C) (50 µg/ml with 200 µg/ml DEAE dextran) for activation of PKR and NMPP (2 µM) for activation of HRI. Cells were harvested at various time points, as determined by preliminary experiments, and analyzed by immunoblotting.

Effects of nutrient deficiency and leucinol on Cm cells
Nutrient deficiency and leucinol treatment were used to determine whether GCN2 can be activated in cobia cells. Reducing the Cm cell L-15 culture medium to 30 % in phosphate buffered saline (PBS) for 24 h reduced both amino acid and serum levels. After 24 h, eIF2α phosphorylation was increased (Figure 4, Panel A, left) consistent with the activation of GCN2. When leucinol (4 mM) was included in the complete culture medium, an increase in eIF2α phosphorylation could be seen within 1 h and continued to increase over 24 h (Figure 4, Panel A, right). This also strongly suggests activation of GCN2 and is consistent with the response of mouse ES cells to leucinol (65). Mouse ES cells also respond very rapidly to leucinol treatment. However, Gcn2-/-mouse ES cells failed to show any induction in eIF2α phosphorylation during the first 3 h of leucine deprivation demonstrating that the effect of leucine deprivation on eIF2α phosphorylation could be attributed to GCN2. , for up to 24 h, for activation of HRI. Proteins in cell extracts were fractionated by SDS-PAGE and eIF2α was visualized by chemiluminescence following electro-transfer to PVDF, and probed with antibody to phosphorylated eIF2α and (after stripping) antibodies to total eIF2α. Phosphorylated and total eIF2α was visualized by chemiluminescence following electro-transfer to PVDF. Chemiluminescence was detected and image generated (upper panels) using the ProteinSimple Fluorochem E, with quantification using Im-ageJ to estimate level of phosphorylated eIF2α (lower panels).

Effects of thapsigargin on Cm cells
Thapsigargin is non-competitive inhibitor of the sarco/endoplasmic reticulum Ca 2+ ATPase (SERCA; 66). eIF2α phosphorylation increases quickly in Cm cells on incubation with thapsigargin (1 µM). These results indicate that PERK can be activated in cobia cells.

Effects of poly I:C on Cm cells
Cm cells were treated with 50 µg/ml poly I:C and 200 µg/ml DEAE dextran. Figure  4, panel C, shows that treatment for 24 h showed almost no eIF2α phosphorylation level difference between control and treatment group. This is in contrast to the effect of poly I:C in rainbow trout RTG2 cells and zebrafish ZFL cells (53). These results could indicate that PKR is not expressed in Cm cells, is not activated, or that cobia eIF2α is not a good substrate for PKR.

Effects of N-methylprotoporphyrin on Cm cells
N-methylprotoporphyrin (NMPP) is a transition-state analogue and potent inhibitor of ferrochelatase, is used to induce heme deficiency and has been shown to increase eIF2α phosphorylation in rat hepatocytes (63). Tobia Cm cells were treated with NMPP (2 µM) for up to 24 h. Unlike the effect in rat heptocytes which respond rapidly to NMPP, Cm cells show no increases in eIF2α phosphorylation after 24 h (Figure 4, panel D). These results indicate that HRI may not be expressed in Cm cells, is not activated or that cobia eIF2α is not a good substrate for HRI.

Cobia eIF2α can be phosphorylated by eIF2α-kinases in vitro
Although Cm cells responded to activators of GCN2 and PERK with increased eIF2α phosphorylation, they did not respond to activators of PKR or HRI. We took the alternative approach of looking at the capacity of recombinant cobia eIF2α to be phosphorylated by the kinases in vitro as we have previously shown for recombinant zebrafish and rainbow trout eIF2α (53). 35S-labeled cobia eIF2α was synthesized in a reticulocyte translation system to which we later added purified mammalian eIF2α-kinases. Phosphorylated and non-phosphorylated forms were separated by vertical slab gel isoelectric focusing (VSIEF) (Figure 5). This analysis showed that cobia eIF2α can be phosphorylated by eIF2α-kinases and that PKR and HRI are either not well expressed or not activated by poly I.C or NMPP in Cm cells.

Figure 5.
In vitro phosphorylation of cobia eIF2α by purified recombinant mammalian eIF2α-kinases 35 S-radiolabeled cobia eIF2α, eIF2α-S51A were synthesized in the reticulocyte TnT cell-free transcription/translation system by incubation at 30 oC for 30 min after which they were supplemented with x units of purified mammalian recombinant HRI (EIF2AK1), PKR (EIF2AK2), PERK (EIF2AK3) or GCN2 (EIF2AK4) followed by incubation at 30 °C for an additional 10 min. Microcystin (0.5 μM) was included to inhibit phosphatase activity. Each sample was diluted in 40 volumes isoelectric focusing sample buffer prior to analysis. Samples were subjected to vertical slab isoelectric focusing (VSIEF) to separate phosphorylated and non-phosphorylated forms, using a narrow pH range of 4.5-6. Proteins transferred to PVDF and visualized using the Typhoon imager. The red arrow indicates the position of phosphorylated eIF2α.

Comparison of eIF2α phosphorylation in fish fed a fish meal versus an all plant protein diet
In order to be useful in aquaculture situations, it was necessary to demonstrate that eIF2α phosphorylation responds to nutritional differences and stress, not just in Cm cells, but in fish. During the current investigation, the Place lab was conducting comparisons of different dietary formulations for cobia, focusing on the formulation of plant protein diets. We took advantage of this by sampling fish from a few of the diet trials. Our investigation was not intended to be an exhaustive analysis of eIF2α phosphorylation under all dietary regimes, but samples of fish were taken to ask if changes in eIF2α phosphorylation could be documented. For instance, the performance of cobia juveniles maintained for 12 weeks on an all plant protein diet, PP, was compared with those on a mixed fish meal/plant protein diet, FM (6). The all plant protein diet was supplemented with menhaden oil to provide essential omega-3 fatty acids, but not with taurine, known to be essential in cobia diets (6; 7). However, the PP diet did contain 0.02 % taurine. The complete formulations are given in Supplemental Table S3. The specific growth rate (SGR) in cobia juveniles on the FM diet for 8 weeks was 4.72+/-0.02, compared with 0.57 ± 0.12 for fish on the PP diet.
The phosphorylation of eIF2α in livers of these fish at the end of the feeding trial is shown in Figure 6. Higher eIF2α phosphorylation levels can be seen in the fish fed the plant protein diet without taurine supplementation. In fact, the level of eIF2α phosphorylation is as high as that observed in juveniles maintained without feeding for 7 days.  Table S2. Proteins in tissue extracts were fractionated by SDS-PAGE and transferred to PVDF. Phosphorylated and total eIF2α were visualized by enhanced chemiluminescence, using antibody to phosphorylated eIF2α and (after stripping) antibodies to total eIF2α, as described in legend to Figure 4. The bar diagrams in lower panels represent the quantification of the blots corresponding to the respective lanes in the upper panel.

Comparison of eIF2α phosphorylation in fish fed a diet with partial replacement of fish meal/poultry by-product/soy protein concentrate with a non-GMO soybean cultivars
Although fish meal can be replaced at least in part by soybean concentrate, most commercially available soybean products are from genetically modified organisms (GMOs), the use of which is not universally accepted. Schillinger Genetics, Inc. has developed multiple cultivars of soybean with potential as fishmeal replacements in diets for aquaculture. Cultivars have been developed with reduced levels of the anti-nutritional factors raffinose, stachyose, and trypsin inhibitors. A diet was formulated, 3010-50, to replace 50 % of protein supplied by fishmeal with a cooked, solvent-extracted soybean meal formulation 3010 from Schillinger Genetics Inc. Poultry meal, wheat flour, soy protein concentrate, corn, and fish oil were all varied to maintain the 3010-50 diet isonitrogenous, isolipidic, and isocaloric qualities compared with the reference diet, FM, and to contain similar amino acid profiles. The diets differ significantly in menhaden and poultry meal, soy protein concentrate, NPF1-3010, menhaden oil. The 3010-50 diet is also supplemented with 0.15 % taurine. The performance characteristics of 3010-50 in a 12-week trial were compared with that of the reference diet, FM, containing 45.5 % fishmeal (7). The complete formulation is given in Supplemental Table S4. Both diets gave very good performance characteristics; 3010-50 gave slightly better performance characteristics than FM, with an SGR of 3.45 +/-0.08, CF of 0.715+/-0.04 compared to SGR of 3.29+/-0.08, CF of 0.637+/-0.637 (7). This is reflected in the level of eIF2α phosphorylation which is lower in fish fed the 3010-50 diet (Figure 7). Note that levels of phosphorylated eIF2α are low in each condition, with the ratio of phosphorylated to total eIF2α in the reference diet is equivalent to that observed in the FM diet (Diet 6) in Figure 6. Juvenile cobia keeps in tank fed a diet with partial replacement of fish meal, poultry by-product (reference diet) and soy protein concentrate with a non-GMO soybean culti-vars (3010-50) and diet formula given in Table S3. Proteins in tissue extracts were fractionated by SDS-PAGE and transferred to PVDF. Phosphorylated and total eIF2α were visualized by enhanced chemiluminescence, using antibody to phosphorylated eIF2α and (after stripping) antibodies to total eIF2α, as described in legend to Figure 4. The bar diagrams in lower panels represent the quantification of the blots corresponding to the respective lanes in the upper panel.3.2.3. eIF2α phosphorylation during "cold banking" and acclimation "Cold banking" is a carefully employed technique to slow down fish growth rate. It is especially effective with fingerlings, when an investigator is trying to stagger fish production. A comparison was made of "cold banked" juveniles, maintained at 20 °C and fed a maintenance ration 5 days a week at 1.5 % body weight (bw) daily for 6 weeks prior to acclimation, with cold-banking parameters based on Holt et al. (5). Acclimation to 27 °C and 5 % bw feeding daily was established over the course of two weeks, increasing 1 °C per day and 0.5 % bw per day for one week, followed by maintenance at 27 °C for one week. eIF2α phosphorylation was compared in the cold-banked and acclimated fish in Figure 8. The level of eIF2α phosphorylation was higher in the fish acclimated to 27 °C. This surprised us since cold shock activates PERK in mammals (67). However, a similar response is seen in human livers kept at cold temperatures for transportation (68). Cold transportation temperatures activate the IRE-1 pathway component of ER stress early. In contrast, at colder temperatures, ATP levels are lower, so that PERK is not activated (requires autophosphorylation), and levels of eIF2 phosphorylation is low. However, this portion of the ER stress response happens as a second phase once the liver is re-perfused and the temperature increased. Effects of water temp change on eIF2α phosphorylation level in cobia liver. Juvenile cobia keeps in tank that water temperature holds at 20 °C and 27 °C. Proteins in cell extracts were fractionated by SDS-PAGE and transferred to PVDF. Phosphorylated and total eIF2α were visualized by enhanced chemiluminescence, using antibody to phosphorylated eIF2α and (after stripping) antibodies to total eIF2α, as described in legend to Figure  4. The bar diagrams in lower panels represent the quantification of the blots corresponding to the respective lanes in the upper panel. Y-axis is the relative ratio of eIF2α phosphorylation compare to total eIF2α.

Discussion
The present study investigated the response of the marine fish cobia, Rachycentron canadum, to stressors as measured by phosphorylation of the α-subunit of the translational initiation factor, eIF2, the target of a family of protein kinases that respond to physiological stressors. The high conservation of eIF2α throughout vertebrates allowed us to use commercial antibodies to eIF2α and phosphorylated eIF2α for our studies. Also in this study, a new cobia cell line, Cm, has been described and demonstrated to be of use in demonstrating the response to nutrient deficiency and ER stress. The response in Cm cells to nutrient deficiency and ER stress mimic that observed in mammalian and ZFL cells. The lack of response to poly(I:C) and NMPP most likely reflects low level expression of these kinases in the Cm cell line. However, investigation of this will depend on uncovering the sequences of cobia eIF2α kinases to allow antibody development. The Cm cells are likely to be very useful for additional molecular studies to investigate the basis of the dietary requirements of cobia. These cells could also be used for investigation of fatty acyl elongase, a key enzyme in the synthesis of polyunsaturated fatty acids. Similarly, the Cm cells could be used to investigate the defects in the taurine biosynthetic pathway in cobia in more detail as well as responses to added taurine. Because of the lack of response to poly(I:C), it is unclear whether Cm cells will be useful for susceptibility testing for viruses such as iridoviruses and nervous necrosis virus (NVV) important viral pathogens of at the fry and fingerling stages of cobia (51; 69).
Beyond investigations in Cm cells, we asked if eIF2α phosphorylation could be used as a rapid indicator of fish physical condition. There is an accumulation of reports of the eIF2α kinases in fish. The investigations have focused on the dsRNA activated eIF2α kinase, PKR, rather than nutrition because of its role in defense against virus infections. PKR has been identified in zebrafish, three-spined stickleback, Gasterosteus aculeatus, fugu, Takifugu rubripes, puffer fish, Tetraodon nigroviridis, common carp, Cyprinus carpio, crucian carp, Carassius auratus, grass carp, Ctenopharyngodon idellus, fathead minnow, Pimephales promelas, medaka, Oryzias latipes, Atlantic salmon, Salmo salar, and Japanese flounder, Paralichthys olivaceus (14; 70-73). All are interferon-stimulated genes and play roles in interferon-mediated antiviral responses. It has been proposed that PKR may also form part of the adaptive response to prevent further accumulation of energy in the overfed state. PKR is normally activated by double-stranded RNA, and interestingly the RNA binding domain of PKR has been shown to be required for its activation in response to lipids (27). PERK has been identified in Chinese rare minnow, Rhynchocypris oxycephalus and medaka, Oryzias latipes, in which it is involved in ER stress (74; 75), and in zebrafish in which it is involved in the response to tributyl tin and other metal poisoning (76; 77). PERK is the other eIF2α kinase known to be involved in the response to nutritional changes. One of the problems that can arise with unbalanced fish diets is the development of fatty liver disease (FLD) and PERK has been reported to be involved in fatty liver disease zebrafish larvae (78). HRI has only been reported so far in Japanese flounder where it responds to heat shock (79). So far, the results reported here in cobia cells represent the first report of GCN2 activation in fish, although it is implied in the effects of dietary lysine in largemouth bass, Micropterus salmoides (80). Since GCN2 is the primary responder to nutritional deprivation, this opens up the whole area of fish diet and eIF2α phosphorylation. GCN2 is also involved in a variety of organismal functions in vertebrates such as feeding behavior (37; 81). To ensure an adequate supply of nutrients, omnivores choose among available food sources. This process is exemplified by the well-characterized innate aversion of omnivores to otherwise nutritious foods of imbalanced amino acid content. Brain-specific inactivation of GCN2 impairs this aversive response (37). In mice, GCN2 is rapidly activated in the mediobasal hypothalamus after consumption of a leucine-deficient diet (81). Knockdown of GCN2 in this particular region shows that GCN2 activity controls the onset of the aversive response.

Conclusions
Our preliminary studies have shown that in cobia juveniles, responses at the level of eIF2α phosphorylation can be observed in response to diet and water temperature. This highlight the usefulness of monitoring eIF2α phosphorylation in the establishment of optimum diets and aquaculture conditions and provides a means to investigate reasons underlying poorly accepted diets. This proof of principle opens the way to a more systematic investigation of responses at the level of eIF2α phosphorylation in cobia and other species of aquaculture interest. We conclude that eIF2α phosphorylation may be useful as an indicator of fish condition and health in aquaculture situations. The development of eIF2α phosphorylation as a routine assay for fish health is unlikely because of level of technical skill needed to use the method usefully. Nevertheless, the monitoring of eIF2α phosphorylation has the potential to provides a very useful tool for analysis in dissecting out where dietary deficiencies may lie and what molecular processes underlie nutritional pathologies and suboptimal aquaculture conditions.
Supplementary Materials: The following supporting information can be downloaded at: www.mdpi.com/xxx/s1, Table S1. Primer pairs used for eIF2a-RACEprimer, cloning of cobia eIF2α and cobia COX1 gene recognition Primers used to amplify the cobia eIF2α cDNAs were designed from the published coding sequences from five close related fish species, zebra fish (NM_131800.2), catfish (GU588091.1), puffer fish (CR685632.2), Atlantic salmon (NM_001140183) and rainbow trout (NM_001124296.1), selected most conserved sequence as target region and we initially amplified a ~900 bp cDNA. The complete coding sequence was assembled by 5 & 3 RACE. The degenerate primers among the above oligonucleotides incorporate a statistical mix of monomers at the positions labeled V (A, C or G), S (C or G), R (A or G), Y (C or T) or D (A, G or T) [in accordance with IUPAC convention].

Primer
Sequence ( ; writing-original draft preparation, C.LL., R.J.; writing-review and editing, C.LL, A.R.P. and R.J.; supervision, A.R.P., R.J.; project administration, A.R.P., R.J; funding acquisition, A.R.P. and RJ. All authors have read and agreed to the published version of the manuscript. Funding: CLL was supported in part by a scholarship from the Ministry of Education, Republic of China (Taiwan). This work was supported by the NOAA-EPP-funded Living Marine Sciences Cooperative Science Center (LMRCSC), NA11SEC4810002. ED was supported by LMRCSC undergraduate internship funds. This is contribution #xxxx from the University of Maryland Center for Environmental Science (UMCES) and #xx-xxx for the Institute of Marine and Environmental Technology (IMET).

Institutional Review Board Statement:
All experiments involving fish were carried out in accordance with the guidelines of the International Animal Care and Use Committee of the University of Maryland Medical School: IACUC protocol #0610015 for cobia. Fish used for tissue sampling were anesthetized with Tricaine methanosulfonate (MS-222, 70 mg L -1 ) for blood sampling and then euthanized with MS-222 (150 mg L -1 ).

Data Availability Statement:
The cDNA sequence for the Rachycentrum canadum eIF2α has been deposited in GenBank (accession # KJ513464).