Introduction

Unlike carnivorous mammals, carnivorous fish usually exhibit poor utilization of carbohydrates. However, the basis of such apparent glucose intolerance in carnivorous fish remains unknown (Moon 2001; Wilson 1994). One hypothesis is that there is an imbalance between hepatic glucose uptake (glycolysis) and production (gluconeogenesis), leading to inadequate regulation of glucose homeostasis (Panserat et al. 2001c).

In rainbow trout (Oncorhynchus mykiss), a typical carnivorous fish, the poor utilization of dietary carbohydrate is reflected by prolonged postprandial hyperglycemia after high carbohydrate intake (Cowey and Walton 1989; Wilson 1994). Regarding nutritional regulation, there is no specific effect of dietary carbohydrate on hepatic pyruvate kinase (PK) gene expression (Panserat et al. 2001a), as this is not controlled by nutritional status, i.e., being food-deprived or well-fed (Kirchner et al. 2003; Panserat et al. 2001a). In gilthead sea bream (Sparus aurata), short-term refeeding after starvation is not sufficient to recover its earlier hepatic PK activity (Metón et al. 2003). For the gluconeogenic enzyme, phosphoenolpyruvate carboxykinase (PEPCK), in gilthead sea bream and rainbow trout, the hepatic PEPCK gene is expressed at the same level independent of dietary composition and nutritional status (Panserat et al. 2001b, 2002). The absence of the nutritional regulation of PEPCK gene expression leads to speculation that the PEPCK gene cloned in rainbow trout liver codes for a mitochondrial isoform. In mammals, hepatic glycolysis PK and gluconeogenesis PEPCK activity and mRNA abundance are primarily regulated by dietary carbohydrate intake and are major contributors to glucose homeostasis (Hanson and Reshef 1997; Yamada and Noguchi 1999).

Cobia (Rachycentron canadum) is a carnivorous fish species that mainly preys on squid and fish (Franks et al. 1996). Because of their carnivorous nature, juvenile cobia can utilize large amounts of proteins and lipids without any impact on production characteristics. Food given to cultured cobia generally contains more than 45% crude protein and 15–16% crude lipid (Craig et al. 2006; Liao et al. 2004); however, Chou et al. (2001) reported that high levels of lipid (up to 18%) showed no growth or efficiency benefit over a 6% lipid diet. Interestingly, being carnivores, cobia should be unable to utilize high levels of carbohydrates, but research has demonstrated that they can tolerate up to 360 g kg−1 of dietary dextrin (Schwarz et al. 2007). Ren et al. (2011) also reported that the appropriate dietary starch supplementation for juvenile cobia was estimated to be about 20%, along with 42% protein and 10% lipid. These data suggest that it is highly feasible for fish farms to reduce food costs by replacing lipids with cheaper carbohydrates; however, no information on the effect of different dietary carbohydrate-to-lipid ratios (CHO/L) on intermediary metabolism is available for this fish.

Rainbow trout have been found to lack molecular induction of glycolytic enzymes and inhibition of gluconeogenic enzyme gene expressions (such as PK and PEPCK), which can at least partially account for their poor utilization of dietary carbohydrate (Panserat et al. 2001a, b). Given that cobia seem to be able to utilize high levels of dietary carbohydrates, we did not expect to find these same phenomena in this species. Thus, the first step was to clone full-length cDNA for PK since full-length cDNA of cytosolic PEPCK (accession no. FJ645270) has already been cloned in a previous study (Lee 2009). Then, we assessed the resulting changes in the gene expression and activity of key enzymes (PK and PEPCK) in the glucose metabolism pathway in cobia liver in response to various diets (various CHO/L ratios and the starved-to-fed transition).

Materials and methods

Feeding trial, experimental conditions, and sample collection

Dextrin was used as the main carbohydrate source in three isonitrogenous and isoenergetic diets containing different CHO/L ratios (low CHO/L ratio 0.68%, moderate CHO/L ratio 1.91% and high CHO/L ratio 5.64%). Compositions of the experimental diets are shown in Table 1.

Table 1 Formulation and proximate composition of cobia experimental diets (%dry matter)
Full size table

Before the feeding trial, six healthy fish were randomly captured from a pool and were sacrificed to obtain heart, whole kidney, white muscle, liver, and intestine for PK tissue expression tests. Then, triplicate groups (20 fish in each cage) of juvenile cobia with an average of body weight of 23.53 ± 0.39 g were reared in a farm of the South Island, Zhanjiang, Guangdong. They were fed twice a day (08:00 and 16:30) to apparent satiation with formulated dry diets containing different CHO/L ratios during the feeding trial period (8 weeks). After the 8-week feeding experiment, fish were starved for 24 h. Next, four fish (two fish for the measurement of gene expression, the other two for the measurement of enzyme activity) were randomly captured from each cage and were deeply anesthetized with eugenol (200 mg L−1). The fish were then euthanized by immediate spinal dislocation, and liver tissue was promptly removed from the fish and immediately frozen in liquid nitrogen and stored at − 80 °C until required.

In addition, 120 fish of similar size divided into another six cages (20 fish per cage) were starved for a period of 4 weeks. Three of these cages (four fish per cage) were sampled at the end of the food-deprivation period. At the same time, fish in the other three cages were refed over the 4-week period, and the same number fish were then sampled at the end of trial experiment. The fish fed with a moderate CHO/L ratio diet for 8 weeks were used as a control, according to Ren et al. (2011) in comparison with fish during the starved-to-fed transition. During the experiment, the water temperature remained between 26 and 33 °C, and dissolved oxygen was maintained above 6.0 mg/L. The procedures of this study involving animals and their care was conducted in conformity with NIH guidelines (NIH Pub. No.85–23, revised 1996) and were approved by the Animal Care and Use Committee of Guangdong Ocean University.

Analytical methods

The proximate composition of the experimental diets was analyzed by the standard methods of AOAC (AOAC 1995). For the measure of PK and PEPCK in livers, a frozen sample of cobia liver (500 mg) was homogenized (dilution 1/10) in ice-cold buffer (80 mM Tris; 2 mM DTT; 5 mM EDTA;1 mM benzamidine; 1 mM 4-(2-aminoethyl) benzenesulfonyl fluoride, pH 7.6). The homogenate was centrifuged at 10,000×g for 20 min at 4 °C. The resultant supernatant was separated, and the activity of hepatic PK was determined following the procedure described by Kirchner et al. (2003), monitoring the decrease in absorbance (β-NAD, reduced form disappearance) using purified lactate dehydrogenase (LDH, 300 unit/mg protein, Sigma) in excess as the coupling enzyme. The PEPCK enzymatic activity was assessed with the method of Polakof et al. (2008), briefly (in mM): 50 Tris–HCl (pH 7.5), 1 MnCl2·4H2O, 20 NaHCO3, 1.5 PEP, 0.3 NADH, excess malate dehydrogenase (MDH, 400 unit/mg protein, Sigma), and 12 deoxyguanosine-5′-diphosphate was used. The reaction was started with deoxyguanosine-5′-diphosphate, and the absorbance at 340 nm was recorded in a microtiter plate reader (SLT).

RNA isolation and reverse transcription

Total RNA was extracted from liver samples according to the RNAiso Plus Kit manual (TaKaRa, Dalian, China). The tissues were lysed with RNAiso Plus buffer, extracted with chloroform (Sangon, Shanghai, China), and precipitated with 2-propanol (Sangon) to obtain total RNA. The extracted RNA was digested with RNase-free DNase I (TaKaRa). The quantity and purity of the isolated RNA was later determined by absorbance measurements at 260 and 280 nm using a NanoDrop ND-1000 (Thermo Scientific, USA), and its integrity was tested by electrophoresis in 1.5% formaldehyde denaturing agarose gels. Only high-quality samples in which OD 260/OD 280 > 1.8, OD 260/OD 230 > 2.0, and 28S/18S rRNA > 1 were used for subsequent complementary DNA (cDNA) synthesis. cDNA was obtained by annealing 1 μg of total RNA with 1 μg of random primers and incubating with M-MLV reverse transcriptase (Invitrogen, Carlsbad, CA, USA) for 1 h at 42 °C.

Molecular clone of cobia PK genes

PK sequences from Homo sapiens (GenBank accession number CAA39849.1), Rattus norvegicus (GenBank accession numbers AAB93666.1), Danio rerio (GenBank accession number AAH45421.1), and Trachinotus ovatus (GenBank accession numbers AJD07313.1) were compared using the Clustal-W multiple alignment algorithm (Thompson et al. 2002). Degenerate primers were chosen on the basis of the most conserved coding regions of the PK gene. Specific primers were designed using Primer Premier 5.0 software (Premier Biosoft International, Palo Alto, CA, USA) under default parameters. The primers for BLAST searches were conducted to confirm the specificity of the primer sequences. The degenerate (upstream and downstream) primers and specific primers are presented in Table 2. The polymerase chain reaction (PCR) was conducted to amplify PK cDNA with 1 μl cDNA template and 0.5 μM degenerate primers in a reaction mixture containing 50 mmol/L KCl, 2 mmol/L MgCl2, 20 mmol/L Tris–HCl, 0.25 mmol/L dNTP and 2.5 units of Taq polymerase (TransTaq@ DNA Polymerase High Fidelity Boehringer, TransGen Biotech, China). The PCR program contained 33 cycles of denaturation at 94 °C for 30 s, annealing at 56 °C for 30 s followed by an extension at 72 °C for 2 min. PCR products were separated by electrophoresis in 1.5% agarose gels in Tris-acetate (TAE) buffer and photographed after staining with ethidium bromide. A fragment of approximately 1400-bp bright band was purified using an GeneJET Gel Extraction Kit (Fermentas, Vilnius, Lithuania) following the manufacturer’s protocol. The purified fragment was ligated into the pMD18-T Vector (TaKaRa, Dalian, China) and transformed into Escherichia coli Cast & Chalm (strain DH5a) cells (Sangon Biotech, Shanghai, China). Recombinant plasmids were extracted using the Plasmid Mini Kit (QIAGEN, Valencia, CA, USA), and cDNA fragments were sequenced by Shanghai Sangon Biotech Service Co. Ltd. (Shanghai, China).

Table 2 Primers used for PK cDNA cloning by RT-PCR (degenerate primers) and RACE-PCR and for PK and PEPCK gene expression analysis by RT-PCR
Full size table

According to the core PK cDNA fragments sequence, specific primers (Table 2) for RACE-PCR were designed to obtain the full-length cDNAs by using the SMARTer™ RACE cDNA Amplification Kit (Clontech). The first strand of cDNA from the 3′ end was obtained with reverse transcription of total RNA and Oligo (dT)16AP as the primer. This was used as a template for PCR with a gene-specific forward primer 3′ PK-RACE1 and a reverse primer AP. After that, the product was applied to the nested PCR with a specific forward primer 3′ PK-RACE2. For 5′ rapid-amplification of the cDNA ends, reverse transcription of total RNA was conducted with a specific reverse primer 5′ PK-RACE1 to obtain the first-strand cDNA. Then, an Oligo (dA) tail at the 5′ end was added using terminal deoxynucleotidyl transferase. That was used for the first PCR with a universal forward primer Oligo (dT)16AP and a specific reverse primer 5′ PK-RACE2. The resulting product was diluted for the nested PCR with a specific reverse primer, 5′ PK-RACE3, and an anchor forward primer AP. PCR products were separated by electrophoresis in 1.5% agarose gels in Tris-acetate (TAE) buffer and photographed after staining with ethidium bromide. A fragment of approximately 1500-bp bright band was purified using an GeneJET Gel Extraction Kit (Fermentas, Vilnius, Lithuania) following the manufacturer’s protocol. The purified fragment was ligated into the pMD18-T Vector (TaKaRa, Dalian, China) and transformed into Escherichia coli Cast & Chalm (strain DH5a) cells (Sangon Biotech, Shanghai, China). Recombinant plasmids were extracted using the Plasmid Mini Kit (QIAGEN, Valencia, CA, USA), and cDNA fragments were sequenced by Shanghai Sangon Biotech Service Co. Ltd. (Shanghai, China).

Real-time quantitative PCR analysis

The expression of PK mRNA in different tissues was analyzed by using real-time quantitative PCR. Total RNA was isolated from five tissues (heart, whole kidney, white muscle, liver and intestine) of six cobia prior to the trial experiment. The cDNA, obtained as described above, was amplified by PCR using specific primers chosen from the full-length sequence for cobia PK cDNA (Table 2). The RT-PCR analysis (ABI 7500, Applied Biosystems, America) was performed with the following program: 1 cycle at 95 °C for 30 s and 40 cycles of 95 °C for 5 s, 55 °C for 30 s, and 72 °C for 30 s. The melting curves of the qPCR products were performed at the end of each PCR to confirm that only one product was amplified and detected. PK mRNA expression levels were normalized to the equivalent RNA levels of the reference gene β-actin (accession number EU266539) using the optimized comparative Ct (2–ΔΔCt) value method (Livak and Schmittgen 2001). β-actin from cobia was chosen as the reference gene to normalize expression levels of targets between different samples. Among the reference genes tested, β-actin did not reveal changes in expression levels and presented the lowest standard deviation within different tissues (0.79) and conditions studied (0.88) using the BestKeeper. In addition, the Cq values of the candidate reference gene were used for comparison of expression stability in the NormFinder and GeNorm programs. β-actin was also found to be a stable reference gene within different tissues and conditions. The PCR efficiency (E) and correlation coefficients (R2) were determined based on the slopes of the standard curves generated from the Cq values of serial dilutions of the cDNA. The E values of several references ranged from 96.7 to 101.1%, and the R2 values ranged from 0.991 to 0.997. PK (accession number: KM262818) and PEPCK (accession number: FJ645270) gene expressions of cobia in response to different dietary CHO/L ratios and nutritional status were also determined following the procedures described above. There were six fish (N = 6) for each treatment, each fish was analyzed for each sample, and each sample was run in triplicate.

Sequence alignment and phylogenetic analysis

Nucleotide sequences were compared with DNA sequences from the GenBank database using the basic local alignment search tool (BLAST) algorithm. Multiple amino acid sequence alignments and percentages of amino acid conservation were evaluated with the ClustalX 2.1 and Genedoc Package 4.0. The neighbor-joining phylogenic trees were constructed using MEGA 4.0 software based on amino acid sequences.

Statistical analysis and sequence analysis

The results were presented as means ± standard deviation (SD, N = 6). Data in this study were subjected to one-way analysis of variance (ANOVA) after testing the homogeneity of variances using the SPSS 20.0 software package (SPSS Inc. Wacker Drive, Chicago, IL USA). Statistical differences among means were determined using Duncan’s multiple-range test. Differences were considered significant at 5%.

Results

Molecular characterization of cobia PK

The full length of cobia PK (accession no. KM262818) gene cDNA covered 1869 bp with a 1599 bp open reading frame (ORF) encoding 532 amino acids (Fig. 1). This AA sequence contained several conserved functional sites, including an ADP/ATP binding site, a phosphoenolpyruvate binding site, an effector site, cation requirement sites, and an additional N-terminal sequence (Fig. 2). Complete AA sequence alignment analysis showed that cobia PK had similar structural features to those of other species (Fig. 2) and showed the greatest similarity of amino acid residues to zebra fish PKM1 (up to 89%), but lower identity with human, rat, and zebra fish PKL (about 68%), suggesting that cobia PK is likely to be the PKM gene.

Fig. 1
figure 1

Nucleotide and deduced amino acid sequences of cobia PK cDNA. The start and stop codons (ATG, TGA) are indicated with bold letters. The deduced amino acid sequence is displayed above the nucleotide sequence. This nucleotide sequence was submitted to the NCBI GenBank with an accession number of KM262818

Full size image
Fig. 2
figure 2

Comparison of the deduced AA sequence of PK from cobia with that from human, zebra fish, and mice. The AA sequences were aligned using ClustalX and Genedoc. Identical residues are shaded black, and similar residues are shaded gray. The three domains of the PK subunit (A, B, and C), and the N-terminal sequences, are marked by a continuous line in the upper part of the sequences. The residues involved in binding PEP and ADP/ATP are indicated below the sequence with square and triangle, respectively. Circle indicates ligands for the cations. The 16 residues involved in the formation of binding site for F1, 6BP in animals are shown in the black box, in which the residue Cys involved in the allosteric regulation of PKM2 is shown in red, and the residue Glu of the non-allosteric-type human and rat PKM1 is shown in orange

Full size image

Phylogenetic analysis

The evolutionary relationships between the PKs were investigated by the construction of phylogenetic trees (Fig. 3). Alignments were made based on the amino acid sequences studied. MEGA 4.0 neighbor-joining software was used to produce an un-rooted tree. The tree represented a 50% consensus from 1000 bootstrap replicates. The sequences fell into two clusters: vertebrate PKLR and PKM, with a bootstrap value of 100%. In each cluster, mammalian sequences formed a group distinct from the others, as shown by the branch bifurcations. The second cluster comprised a branch with one group formed by Homo sapiens and Rattus norvegicus PKM. For another part of the cluster, a close evolutionary relationship was observed between the zebra fish (PKM1) and cobia PK sequences. The fact that the cobia PK sequence fell into the cluster of PKM suggested that cobia PK was expected to be the PKM gene (Fig. 3).

Fig. 3
figure 3

Phylogenetic tree based on PKs AA sequences was made with MEGA 4.0 software using the neighbor-joining method. The numbers represent bootstrap percentages. The topology was tested using bootstrap analyses (1000 replicates). GenBank accession numbers of the sequences are indicated in the figure

Full size image

Tissue-specific expression of PK mRNA in cobia

Distributions of PK mRNA were detected in a range of tissues, including the heart, intestine, liver, kidney, and white muscle of cobia (Fig. 4). The results showed that the cobia PK gene was predominantly expressed in the liver and heart, followed by white muscle, while significantly lower expression levels were observed in the kidney and intestine (P < 0.05).

Fig. 4
figure 4

Relative expression of PK in various tissues of cobia. For tissue expressions, data are referred to the relative expression value (Relative units, RU) obtained in the kidney. Each data represents the mean ± SD (N = 6). Bars assigned with different superscripts are significantly different (P < 0.05)

Full size image

The effect of dietary CHO:L ratios and nutritional status on mRNA expression and hepatic PK and PEPCK activity in cobia

In mammalian PK enzymes, there are four distinct isozymes, M1, M2, L, and R, encoded by two genes: PKM and PKL. The PEPCK enzyme possesses two distinct isoforms (a constitutive mitochondrial isozyme and an inducible cytosolic isozyme) encoded by two different PEPCK genes (Imamura and Tanaka 1982; Hanson and Reshef 1997). Given that it is unknown how many PEPCK isozymes there are in cobia and which ones are inducible, we decided to only examine total PEPCK activities and the cytosolic PEPCK gene expression in cobia liver.

The relative expression levels of hepatic PK and PEPCK in cobia in response to different dietary CHO/L ratios and nutritional status are presented in Figs. 5 and 7. Significantly induced hepatic PK gene expression was observed when the fish were fed an increasing CHO/L-ratio diet. Such an effect was also observed in fish in the refed diet category in comparison with the food-deprived fish (P < 0.05), and after 4 weeks of refeeding PK mRNA abundance restored to the same level of the control group (Fig. 5). However, there was no significant effect of CHO/L ratio on PK enzyme activity (P > 0.05), and PK activity was significantly higher in food-deprived and refed than fish in control group (Fig. 6).

Fig. 5
figure 5

The expression of liver PK in response to a different dietary carbohydrate-to-lipid (CHO/L) ratios (L: low, M: moderate, H: high) for 8 weeks and b different nutritional status (C: feeding, S: starvation, R: refeeding) in cobia. For different dietary CHO/L ratio, data are referred to the relative expression value obtained in the L-CHO/L ratio diet. For different nutritional status, data are referred to the relative expression values obtained in the control group (M-CHO/L ratio diet). Bars bearing different letters were significantly different (P < 0.05) and values are means ± SD (N = 6)

Full size image
Fig. 6
figure 6

The activity of liver PK in response to a different dietary carbohydrate-to-lipid (CHO/L) ratios (L: low, M: moderate, H: high) for 8 weeks and (b) different nutritional status (C: feeding, S: starvation, R: refeeding) in cobia. Bars bearing different letters were significantly different (P < 0.05) and values are means ± SD (N = 6)

Full size image

As to the gluconeogenic enzyme PEPCK, there was a significant effect of the dietary CHO/L ratio and nutritional status on PEPCK gene expression, with downregulated PEPCK gene expression when the fish were fed an increasing CHO/L-ratio diet. Such effect was also observed in fish when refed the diet in comparison with the food-deprived fish, but after 4 weeks of refeeding, it did not restore the same values from the control group (Fig. 7). In addition, the decrease in PEPCK activity was significant between the L-CHO/L and the H-CHO/L group, whereas the M-CHO/L group showed intermediate values. However, no significant difference was observed in fish with different nutritional status (Fig. 8).

Fig. 7
figure 7

The expression of liver PEPCK in response to a different dietary carbohydrate-to-lipid (CHO/L) ratios (L: low, M: moderate, H: high) for 8 weeks and b different nutritional status (C: feeding, S: starvation, R: refeeding) in cobia. For different dietary CHO/L ratio, data are referred to the relative expression value obtained in the L-CHO/L ratio diet. For different nutritional status, data are referred to the relative expression values obtained in the control group (M-CHO/L ratio diet). Bars bearing different letters were significantly different (P < 0.05) and values are means ± SD (N = 6)

Full size image
Fig. 8
figure 8

The activity of liver PEPCK in response to a different dietary carbohydrate-to-lipid (CHO/L) ratios (L: low, M: moderate, H: high) for 8 weeks and b different nutritional status (C: feeding, S: starvation, R: refeeding) in cobia. Bars bearing different letters were significantly different (P < 0.05) and values are means ± SD (N = 6)

Full size image

Discussion

Glycolysis and gluconeogenesis are opposing metabolic pathways involved in glucose uptake and production and play vital roles in the regulation of glucose homeostasis. In the present study, the full-length PK cDNA was cloned and characterized. The complete AA sequence alignment analysis showed that cobia PK had similar structural features to those of other species, with the greatest similarities with zebra fish PKM1 (up to 89%), but with a lower identity with human, rat, and zebra fish PKL (about 68%), suggesting that the cobia PK is expected to be the PKM gene. The conserved sequence and characteristic motifs of PKs were identified in it, including an ADP/ATP binding site, phosphoenolpyruvate binding site, effector site, and cation requirement sites (Muñoz and Ponce 2003). Based on the results from sequence alignments, these sites were all highly conserved in all species examined in this study, with only minor AA changes observed. In the present study, the amino acid residues Cys-423 (Fig. 2, residue in red) involved in the allosteric regulation of the enzyme were not found in the cobia PK at this position, which was the same in the non-allosteric-type rat and human PKM1 (Ikeda and Noguchi 1998). This indicated that the cobia PK might be a cytosolic enzyme clearly distinct from allosteric-type PKM2. However, the structural function of the replacement of Glu-432 of the non-regulated PKM1 (Fig. 2, residue in orange by Gly) remains unknown (Ikeda and Taniguchi 2000). In addition, the deduced amino acid sequence of cobia PK was organized into three principal domains: A—subdivided into A1 and A2; B and C; and an additional N-terminal domain (Muirhead et al. 1986). Finally, the analysis of PK mRNA tissue distributions showed that high abundance of PK was expressed in the liver, heart, and muscle of cobia, while lower expression levels were observed in the intestine and kidney. The presence of the highly expressed PK gene in the liver was to be expected, given the known tissue expression profiles of PK in other species (Panserat et al. 2001a; Yuan et al. 2013). It is well known that the liver plays a central role in the production and catabolism of glucose to maintain glucose homeostasis (Kamalam et al. 2016). Most enzymes involved in the regulation of glucose metabolism, including PK and PEPCK, are found in the liver. However, the lower expression of PK in the cobia kidney and intestinal tissues may be species-dependent, as a previous study found the PK gene to be highly expressed in both rainbow trout and mammals (Panserat et al. 2001a; Imamura and Tanaka 1982). Currently, limited data on tissue-specific expressions of PK in fish are available, hindering further comparative analysis.

The glycolytic enzyme PK is also associated with the regulation of nutrients and hormones (Gali et al. 2012; Yokozawa et al. 2008). In fish, previous studies have mostly focused on the regulation of PK activity by dietary composition (Dias et al. 2004; Enes et al. 2006; Fernández et al. 2007; Couto et al. 2008), but molecular-level analyses of the regulation of this gene are relatively limited. In the present study, the nutritional regulation of hepatic PK was analyzed at the enzymatic and molecular levels. The results showed that the abundance of PK mRNA was under nutritional regulation. A significantly increased expression of hepatic PK genes was observed when the fish were fed an increasing CHO/L-ratio diet, and such an effect was also found in fish when refed the diet in comparison with the food-deprived fish, and after 4 weeks of refeeding, PK mRNA abundance restored to the same level of control group. This contrasted with rainbow trout, in which no specific effect of dietary carbohydrate on hepatic PK gene expression was observed (Panserat et al. 2001a). In fact, PK is known to occur as four distinct isozymes, M1, M2, L, and R, encoded by two genes: PKL and PKM (Imamura and Tanaka 1982). In rainbow trout, the PK gene is cloned coding for L-type isoform, which is different from our cloning of cobia PK cDNA that coded for a M1 isoform enzyme. However, the discrepancy in nutritional regulation between these two fish species cannot simply be explained by the different isoforms assessed, as it was recently reported that, in grass carp, PKL gene expression and activity are both induced by high dietary carbohydrate levels (Yuan et al. 2013). In addition, an induction of hepatic PK activity by dietary carbohydrates was also observed in gilthead sea bream and European sea bass (Dicentrarchus labrax) (Enes et al. 2006; Fernández et al. 2007; Couto et al. 2008) but not in other fish species (Dias et al. 2004; Enes et al. 2008). In the present study, in contrast to gene expression, there was no significant effect of various CHO/L-ratio diets on PK enzyme activity. In fact, PK activity was significantly higher in food-deprived fish than those fed with the M-CHO/L-ratio diet. An explanation for the increase in gene expression but not in enzymatic activity could be that there is no direct relationship between gene expression and activity, possibly linked to post-transcriptional regulation or to qualitative enzyme modifications, such as the phosphorylation/dephosphorylation of PK (Panserat et al. 2001a). During starvation, no dietary glucose is available, but because animals still require energy, it is possible that they were performing glycolysis at high levels compared to fish that were fed a moderate glucose content. Moreover, these results could be in agreement with the results for PEPCK, the gluconeogenic enzyme, which showed an increase in gene expression in the food-deprived fish but did not present differences in activity, since that would have provided glucose for PK.

As for the gluconeogenesis pathway, previous studies have demonstrated that dietary carbohydrates do not inhibit certain hepatic gluconeogenic enzymes in some teleosts, like PEPCK and G6Pase in rainbow trout (Panserat et al. 20002001c; Coutinho et al. 2016) as well as FBPase in European sea bass (Moreira et al. 2008) and gilthead sea bream (Couto et al. 2008). Here, we observed for the first time that there was a significant effect of the dietary CHO/L ratio and nutritional status on PEPCK (cytosolic isoform) gene expression, which was downregulated when fish were fed increasing CHO/L-ratio diets. Such effects were also observed in fish when refed the diet in comparison with fish that were food-deprived. This result is in agreement with that generally observed in mammals (Hanson and Reshef 1997). In fact, there are two distinct isoforms (mitochondrial and cytosolic) encoded by separate PEPCK genes with different regulation patterns (Hanson and Reshef 1997). The gene for the cytosolic form is acutely under the regulation of hormones and nutritional status, whereas the mitochondrial form is largely constitutively expressed. In Siberian sturgeon (Acipenser baerii) and turbot (Scophthalmus maximus), liver cytosolic PEPCK expression was significantly higher in fish fed a high carbohydrate diet than fed a carbohydrate-free diet, while mitochondrial PEPCK expression was similar in both groups (Nie 2003; Gong et al. 2015). In rainbow trout and gilthead sea bream, the fact that PEPCK gene expression is not regulated by nutritional has led to speculation that the PEPCK gene cloned in these species codes for a mitochondrial isoform (Panserat et al. 2001b, 2002). This has been confirmed by a previous finding of rainbow trout, that the mitochondrial isoform dominates PEPCK activity (> 80%; Walton and Cowey 1979). Thus, nutritional regulation of the expression and activity of cytosolic PEPCK could have been masked in rainbow trout, although the cytosolic form is tightly controlled by nutrients. In the present study, we only determined the total PEPCK activities and the cytosolic PEPCK gene expressions in cobia liver. In contrast to gene expression, the decrease in PEPCK activity was significant between the L-CHO/L and H-CHO/L groups, whereas the M-CHO/L group showed intermediate values. But PEPCK activity remained at the same level independent of the fishes’ nutritional status (fed, unfed, or refed). This result can be interpreted as the PEPCK gene in the present study coding for a cytosolic enzyme that is nutritionally regulated, whereas hepatic total PEPCK activity in cobia might be mainly mitochondrial, so that nutritional regulation of cytosolic PEPCK activity could have been masked.

Overall, in accordance with our hypothesis, at least at the molecular level, the mRNA abundance of the important glycolytic (PK) and gluconeogenic (PEPCK) enzymes are regulated by dietary composition and nutritional status in cobia liver, which is not the case in rainbow trout (Panserat et al. 2001a, b). However, in contrast to the regulation occurring at the molecular level, we did not observe the same nutritional regulation at the enzymatic level. Further studies that include data from other key enzymes involved in hepatic glucose metabolism are necessary to reveal the truth about whether cobia effectively utilize carbohydrates. In addition, the issue of PK and PEPCK isoforms is important in the study of the nutritional regulation of PK and PEPCK gene expression. As the proportions of the PK and PEPCK isoforms in liver cells are species-specific, future studies will need to begin by defining the proportion of each isoform in each species under investigation.