Nutritional regulation and tissue specificity of gene expression for proteins involved in hepatic glucose metabolism in rainbow trout (Oncorhynchus mykiss)

Improvement of dietary carbohydrate utilisation by fish has practical implications in aquaculture. The rainbow trout (Oncorhynchus mykiss) is an interesting model species recognised for its low levels of carbohydrate utilisation (Palmer and Ryman, 1972; Bergot, 1979; Cowey and Walton, 1989; Wilson, 1994; Moon and Foster, 1995). Analysis of glucose utilisation such as glucose phosphorylation (Cowey and Walton, 1989; Moon and Foster, 1995), glucose transporters (Wright et al., 1998) and insulin receptors (Mommsen and Plisetskaya, 1991; Parrizas et al., 1994), in target tissues is important in order to obtain an overall view of low dietary glucose utilisation in this carnivorous fish species. One hypothesis to explain the poor utilisation of dietary glucose by fish is a dysfunction of nutritional regulation between two major metabolic pathways in the liver: (i) a low capacity to store excess glucose in the postprandial stage (glycogen synthesis or lipogenesis) (Cowey and Walton, 1989; Wilson, 1994) and/or (ii) persistent highly active hepatic glucose production when carbohydrates are provided in the diet (Panserat et al., 2000b; Panserat et al., 2001). Indeed, we have recently shown, in rainbow trout, the induction of expression of glucokinase, the first enzyme in the glycolytic pathway, and the absence of inhibition of glucose-6-phosphatase, the last enzyme in hepatic glucose production (Panserat et al., 2000a; Panserat et al., 2000b).

The aim of the present study was to analyse two key hepatic glycolytic enzymes: (i) 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase, otherwise termed the bifunctional enzyme (6PF-2K/F-2,6BPase, EC2.7.1.105/EC3.1.3.46) (Pilkis and Granner, 1992; Pilkis et al., 1995), which produces fructose 2,6-bisphosphate, a key allosteric regulator of glycolysis/gluconeogenesis in the liver (Pilkis et al., 1995), and (ii) l-type pyruvate kinase (PK, EC2.7.1.40), which catalyses the conversion of phosphoenolpyruvate to pyruvate, the last step in glycolysis (Yamada and Noguchi, 1999). We also studied the key gluconeogenic enzyme fructose-1,6-bisphosphatase (FBPase, EC3.1.3.11); this enzyme catalyses the conversion of fructose-1,6-bisphosphate to fructose-6-phosphate (Marcus et al., 1987). Finally, we looked for a type-2 glucose transporter (Glut2) in the liver of rainbow trout. This transporter maintains the same glucose concentration between the extra and intracellular (hepatocyte) media because of its high K_m for glucose (Thorens, 1992).

In mammals, the expression of PK, 6PF-2K/F-2,6BPase and Glut2 proteins is positively controlled by feeding, specifically by dietary carbohydrates (Thorens, 1992; Pilkis et al., 1995; Yamada and Noguchi, 1999); this induction potentiates dietary glucose utilisation in the liver. However, kinase activity of 6PF-2K/F-2,6BPase, which generates the fructose-2,6-bisphosphate, is also positively dependent on protein phosphatase 2A activity. The mechanism of induction of these proteins is mediated largely by induction of their gene expression even though, for the Glut2, the regulation of gene expression in the liver is rather complex (Thorens, 1992; Rencurel et al., 1996; Pilkis et al., 1995; Yamada and Noguchi, 1999). In contrast, feeding and dietary carbohydrates inhibit FBPase gene expression (El-Magrabi et al., 1982; El-Magrabi et al., 1988).

Nutritional regulation of the activities of PK, FBPase and 6PF-2K/F-2,6BPase has been studied previously in fish (Cowey et al., 1977; Cowey et al., 1981; Hilton and Atkinson, 1982; Fideu et al., 1983; Petersen et al., 1987; Lupianez et al., 1989; Garcia de Fructos and Baanante, 1994; Shikata et al., 1994; Shimeno et al., 1995; Meton et al., 1999a; Meton et al., 1999b; Tranulis et al., 1996; Borrebaek and Christophersen, 2000). It seems that there is an increase in the activities of PK and 6PF-2K/F-2,6BPase in response to feeding, although no clear data on the mechanism (transcriptional, post-transcriptional, post-translational) involved in this nutritional regulation are available except for the 6PF-2K/F-2,6BPase in gilthead seabream (Sparus aurata) (Garcia de Fructos and Baanante, 1994; Meton et al., 1999a; Meton et al., 1999b). No specific inhibition of hepatic FBPase activity in response to feeding has been observed in perch (Perca fluviatilis) or salmon (Salmo salar) (Tranulis et al., 1996; Borrebaek and Christophersen, 2000) whereas in European seabass (Dicentrarchus labrax) and gilthead seabream inhibition of FBPase activity in response to feeding has been demonstrated (Garcia-Rejon et al., 1996; Meton et al., 1999b). Finally, the existence of a mammalian-type glucose transporter (Glut) in fish is still a matter of debate; for example, studies using mammalian anti-Glut antibodies suggest, in tilapia (Oreochromis niloticus), a very limited distribution of Glut1 and the complete absence of Glut4 (Wright et al., 1998), whereas recent studies demonstrate the existence of Glut1 and Glut4 cDNAs in rainbow trout and brown trout (Salmo trutta), respectively (Teerijoki et al., 2000; Planas et al., 2000).

In rainbow trout, no molecular data for 6PF-2K/F-2,6BPase, PK, FBPase and Glut2 proteins are yet available. The aim of this study was the partial cloning of PK, GLUT2, FBPase and 6PF-2K/F-2,6BPase cDNAs expressed in the liver, followed by the analysis of their expression in different tissues and in fish under different nutritional conditions.

Juvenile immature rainbow trout (Oncorhynchus mykiss) were grown for 10 weeks at 18°C at the INRA experimental fish farm (Donzacq, France) and fed with (20%) or without (<0.5%) carbohydrates, as described previously (Panserat et al., 2000a; Panserat et al., 2000b). On the sampling day, the fish were fed once and killed 6h and 24h after feeding. Food-deprived fish were deprived for 4 days. Tissues (liver, muscle, heart, kidney, brain) were quickly removed, frozen in liquid nitrogen and stored at −80°C.

Total RNA was extracted from rainbow trout tissues as described by Chomczinski and Sacchi (Chomczinski and Sacchi, 1987). cDNA was obtained by annealing 2μg of total RNA with 1μg of random primers and incubating with AMV reverse transcriptase (Boehringer, Roche Molecular Biochemicals, Germany) for 1h at 42°C.

6PF-2K/F-2,6BPase, PK, FBPase and Glut2 sequences from different species (see Table1) were compared using the Clustal-W multiple alignment algorithm (Higgins and Sharp, 1989). The sequences of the upstream and downstream (degenerate) primers are presented in Table2. cDNA (1μl) was amplified by polymerase chain reaction (PCR) using 100pmol of the degenerate primers in a reaction mixture containing 2 mmoll⁻¹MgCl₂, 50 mmoll⁻¹KCl, 20 mmoll⁻¹Tris-HCl, 0.25 mmoll⁻¹dNTP and 2.5 units of Taq polymerase (Boehringer, Roche Molecular Biochemicals, Germany). 35 cycles of denaturation for 1min at 94°C, annealing at 56°C for 30s, and extension at 72°C for 30s were performed. PCR products were subjected to electrophoresis in 1% agarose gels and fragments of the expected size range were purified (Micropure System, Amicon, USA). The purified DNA fragments were inserted into the pCR™II plasmid and used for transformation of One Shot™competent cells (Invitrogen, Carlsbad, CA, USA). Inserts were detected by EcoRI digestion of the extracted plasmid DNA. Two clones with inserts were sequenced (Cybergène, Evry, France).

Nucleotide sequences (excluding the primer sequences) were compared with DNA sequences from the GenBank database using the basic local alignment search tool (BLAST) algorithm (Altschul et al., 1990). Sequence alignments and percentage of amino acid conservation were assessed with the Clustal-W multiple alignment algorithm using the cloned fish sequence and sequences from other species corresponding to the amplified regions from databases.

Extracted total RNA (20μg) samples were electrophoresed in 1% agarose gels containing 5% formaldehyde and transferred by capillary onto nylon membrane (Hybond-N⁺, Amersham, England). After transfer, RNA blots were stained with Methylene Blue to locate 26S and 16S rRNAs and to determine the amount of loaded RNA. Membranes were hybridized with fish ³²P-labelled DNA probes labelled by random priming (Stratagene, USA) and recognizing partial rainbow trout 6PF-2K/F-2,6BPase, PK, FBPase and Glut2 cDNAs. After stringent washing (2× SSC, 0.1% SDS for 20min; 1× SSC, 0.1% SDS for 20min; 0.2× SSC, 0.1% SDS for 15min), the membranes were exposed to X-ray film and the resulting images were quantified using Visio-Mic II software (Genomic, France).

cDNAs were amplified by PCR using specific primers chosen from the partial sequences for rainbow trout 6PF-2K/F-2,6BPase, PK, FBPase and Glut2 cDNAs (Table2). PCR reactions were carried out in a final volume of 25μl containing 1.5 mmoll⁻¹ MgCl₂ and 4pmol of each primer, 2μl cDNA and 1 unit of Taq polymerase (Boehringer, Roche Molecular Biochemicals, Germany). 35 cycles of 20s for hybridization (at 57°C), 20s for elongation (at 72°C) and 20s for denaturation (at 94°C) were performed. The PCR products were characterized by sequencing (Cybergène, Evry, France).

The results are expressed as means ± s.d. When there were significant differences in variances (one-way analysis of variance, ANOVA), statistical differences between series of data were determined using Tuckey’s post-hoc test (Systat 9 software products, SPSS Inc.). Differences were considered significant at 5%.

The available PK, Glut2, FBPase and 6PF-2K/F-2,6BPase cDNA sequences were aligned and highly conserved regions from different species (Table1) were identified. Four sets of primers were designed (Table2) and this made it possible to amplify partially PK, Glut2, FBPase and 6PF-2K/F-2,6BPase mRNAs. RT-PCR were performed on hepatic total RNA extracted from fish fed with or without carbohydrate (for FBPase cloning only). PCR conditions were optimized, and a major amplification product of the expected size was obtained for PK, Glut2, FBPase and 6PF-2K/F-2,6BPase transcripts (data not shown). The fragments were purified, cloned and sequenced. The cDNA sequences of 366 base pairs (bp) (PK), 293bp (Glut2), 395bp (FBPase) and 268bp (6PF-2K/F-2,6BPase) (Fig.1A, Fig.2A, Fig.3A, Fig.4A) were highly similar to those of ‘mammalian’ genes (Blast algorithm, p=10⁻⁵⁶ to 10⁻⁴⁶, p=10⁻¹² to 10⁻⁸, p=10⁻¹⁰⁶ to 10⁻¹¹, p=10⁻⁷² to 10⁻⁶² for PK, Glut2, FBPase and 6PF-2K/F-2,6BPase, respectively). The corresponding amino acid sequences were deduced from the four cDNA sequences and showed open reading frames of 121 (PK), 97 (Glut2), 131 (FBPase) and 89 (6PF-2K/F-2,6BPase) codons highly homologous to ‘mammalian’ proteins (Blast algorithm, p=10⁻⁵⁶ to 10⁻⁴², p=10⁻⁴⁷ to 10⁻⁴⁵, p=10⁻⁴⁷ to 10⁻³³, p=10⁻⁵⁴ to 10⁻⁴⁵ for PK, Glut2, FBPase and 6PF-2K/F-2,6BPase, respectively) (Fig.1B, Fig.2B, Fig.3B, Fig.4B).

To analyze the tissue specificity of PK, Glut2, FBPase and 6PF-2K/F-2,6BPase gene expression (Fig.5), we performed RT–PCR using specific rainbow trout PK, Glut2, FBPase and 6PF-2K/F-2,6BPase primers (Table2). This RT–PCR was not quantitative, and thus only the presence or absence of a specific mRNA could be determined. PK and Glut2 mRNAs were detected in liver, intestine and kidney but not in muscle, brain or heart (Fig.5). FBPase mRNA was detected in liver, intestine and kidney but also in muscle, brain and heart of fed fish. Finally, 6PF-2K/F-2,6BPase gene expression was ubiquitous, even not much in brain of fed fish.

As growth rates of rainbow trout fed with 20% carbohydrates or without carbohydrates were comparable (Panserat et al., 2000a), comparative analysis of the effects of dietary carbohydrates on the regulation of metabolic enzyme expression between fish groups fed different carbohydrate levels was possible. Expression of PK, Glut2, FBPase and 6PF-2K/F-2,6BPase genes was analyzed in fish livers by northern blotting (Fig.6). Single mRNA species for PK and Glut2 cDNAs of approximately 2.2kb and 2kb, respectively, were found (Fig.6A,B). In contrast, two mRNAs for 6PF-2K/F-2,6BPase and FBPase were found (Fig.6C,D); the major forms were approximately 2kb and 2.7kb for 6PF-2K/F-2,6BPase and FBPase, respectively, whereas the minor mRNAs (very weak) were of approximately 1.5kb and 2.5kb (data not shown). As the expression of the minor fragment is not easy to analyze by northern blotting (level of gene expression is too low), only the major 6PF-2K/F-2,6FBase and FBPase mRNA species were studied in this nutritional analysis.

The effect of fasting and refeeding on gene expression was analyzed. In contrast to expression of PK, FBPase and Glut2 (in which no significant differences were observed between food-deprived and fed fish), it appeared that the level of 6PF-2K/F-2,6BPase gene expression was significantly higher in fish fed with carbohydrates 24h after feeding than in food-deprived fish (P<0.05, Tukey’s test: Fig.7A–D). Our results indicated that dietary carbohydrates did not induce a specific effect; expression of PK, Glut2, FBPase and 6PF-2K/F-2,6BPase mRNAs were elevated at both 6h and 24h after feeding, similar to levels in the livers of fish fed with and without carbohydrates (Fig.7A–D). We also analyzed the effects of the postprandial time on gene expression in fish fed with a specific diet (6h compared with 24h feeding): no significant effect was observed (data not shown).

PK, Glut2, FBPase and 6PF-2K/F-2,6BPase belong to a family of enzymes, such as phosphoenolpyruvate carboxykinase (EC4.1.1.32) and fatty acid synthetase (EC2.3.1.85), whose expression in mammals has been shown to be regulated by dietary carbohydrates (Rencurel and Girard, 1998). In fish, literature data in this area are scarce. Although a formal proof will await the cloning of the full-length cDNA sequence, the high similarity between the cDNA sequences of hepatic PK, Glut2, FBPase and 6PF-2K/F-2,6BPase in rainbow trout and of the PK, Glut2, FBPase and 6PF-2K/F-2,6BPase sequences previously characterized in other vertebrates (up to 72%, 78%, 76% and 95% similarity, respectively) strongly suggests that these sequences correspond to functional proteins (enzymes or transporter). We detected only one species of PK and Glut2 mRNA in trout liver, whereas there are two FBPase and 6PF-2K/F-2,6BPase mRNAs. In gilthead seabream, a major hepatic 6PF-2K/F-2,6BPase RNA band of approximately 4.2kb and two minor bands of approximately 2.1kb and 1.1kb have been detected (Meton et al., 1999a). The existence of different mRNAs may correspond either to distinct 5′ and 3′ untranslated regions or to the existence of distinct genes. Moreover, it is worth noting that trout FBPase mRNAs (2.5kb and 2.7kb) are approximately twice the size of mammalian mRNAs (approximately 1.3 and 1.5kb) (El-Magrabi et al., 1988). Furthermore, the cloning of cDNAs represented here is not exhaustive; the existence of other (different) hepatic PK, Glut2, FBPase and 6PF-2K/F-2,6BPase mRNAs is possible because (i) only two hepatic cDNA clones have been analysed and (ii) our cDNA cloning strategy based on RT–PCR with degenerate primers may amplify only one specific mRNA and not all the different mRNA species.

As in mammals (Marcus et al., 1987; Thorens, 1992; Pilkis et al., 1995; Yamada and Noguchi, 1999), PK, 6PF-2K/F-2,6BPase, FBPase and Glut2 genes are expressed in the liver, intestine and kidney. Moreover, there is no apparent nutritional effect on the regulation of PK, Glut2, FBPase and 6PF-2K/F-2,6BPase gene expression in these different tissues, except for the FBPase gene, which is expressed in non-gluconeogenic tissues such as the muscle and heart in fed fish only. To our knowledge, the biochemical consequences of FBPase gene expression in muscle (white muscle and heart) have not been studied.

PK and 6PF-2K/F-2,6BPase genes code for hepatic glucose metabolism proteins that are induced by feeding in mammals (Pilkis et al., 1995; Yamada and Noguchi, 1999). Induction of hepatic trout 6PF-2K/F-2,6BPase gene expression by feeding (and specifically by dietary carbohydrates) in rainbow trout confirms what has been already shown in mammals and gilthead seabream (Garcia de Fructos and Baanante, 1994; Meton et al., 1999a; Meton et al., 1999b; Meton et al., 2000). However, a significant change in 6PF-2K/F-2,6BPase gene expression may either increase or decrease concentrations of fructose-2,6-bisphosphate; more data on the phosphorylation/dephosphorylation status of the enzyme (induced by protein kinase A or protein phosphatase 2A activities) (Pilkis et al., 1995) would be needed to determine in which direction 6PF-2K/F-2,6BPase affects fructose-2,6-bisphophate concentrations. The trout hepatic PK gene is always highly expressed, irrespective of the nutritional status. Using enzymatic methods, some authors (Cowey et al., 1977; Cowey et al., 1981; Fideu et al., 1983) found that PK activity was induced by feeding in rainbow trout, while others (Hilton and Atkinson, 1982; Shikata et al., 1994) found no such induction. Our data suggest that this induction is not (mainly) due to a transcriptional mechanism, but is more probably linked either to post-transcriptional regulation or to qualitative alterations such as phosphorylation/dephosphorylation of PK, as in mammals (Yamada and Noguchi, 1999). Moreover, we were unable to demonstrate a significant decrease in FBPase gene expression in response to dietary carbohydrates in rainbow trout liver (at least for the major FBPase mRNA species), in accordance with the absence of inhibition of FBPase activity by dietary carbohydrates in other fish species such as perch (Perca fluviatilis) or Atlantic salmon (Salmo salar) (Tranulis et al., 1996; Borrebaek and Christophersen, 2000). It is also worth noting that some individuals (irrespective of the nutritional status) show very low levels of FBPase gene expression; the explanation for this result is unknown.

Our data demonstrate the existence of a glucose transporter very similar to the mammalian Glut2 (cloning of a full-length Glut2 cDNA is in progress in collaboration with Dr A. Krasnov and confirms the existence of a Glut2 protein in trout; our unpublished data). Glut2 seems to be expressed at high levels in rainbow trout, irrespective of nutritional status. This result is expected since Glut2 is important for the entry of glucose into hepatocytes but to allow glucose (produced by gluconeogenesis and glycogenolysis) out of the hepatocytes during fasting (Thorens, 1992).

In conclusion, our data demonstrate the existence of mammalian-type induction of 6PF-2K/F-2,6BPase gene expression by feeding, as has been shown previously for trout glucokinase (Panserat et al., 2000a; Panserat et al., 2000b) in association with high levels of PK gene expression. Taken together, these data suggest the existence of an efficient enzymatic mechanism for storing excess glucose. The existence of a type-2 glucose transporter further supports the possibility that hepatocytes store excess dietary glucose. In contrast, FBPase is not strongly inhibited by dietary carbohydrates. Such a lack of inhibition has been observed previously for two other key hepatic enzymes, phosphoenolpyruvate carboxykinase and glucose-6-phosphatase (Panserat et al., 2000a; Panserat et al., 2000b; Panserat et al., 2001), suggesting an absence of nutritional control of hepatic glucose production in trout. Further analysis of hepatic glucose production in vivo and ex vivo is necessary.

We thank F. Sandres, F. Terrier and Y. Hontang for their assistance during trials in the experimental fish farms (Donzacq, France). We are also grateful to J. Camblong and D. Bazin for their technical assistance. GenBank accession numbers: PK, AF246146; Glut2, AF246147; FBPase, AF333188; 6PF-2K/F-2,6BPase, AF246148.