Rainbow trout glucose transporter (OnmyGLUT1): functional assessment in Xenopus laevis oocytes and expression in fish embryos

Passive glucose transporters (GLUTs) facilitate the passage of monosaccharides across the plasma membrane of animal cells, a key stage of carbohydrate metabolism (Mueckler, 1990; Burant et al., 1991; Baldwin, 1993). In mammals, there are at least five distinct sugar transporters (GLUT1–GLUT5). These proteins are expressed in a tissue-specific manner; they exhibit different kinetics, substrate specificity and sensitivity to inhibitors.

Fish genes encoding putative glucose transporters have been identified recently. We have cloned cDNAs from the rainbow trout Oncorhynchus mykiss and the common carp Cyprinus carpio that are similar to mammalian and avian GLUT1 (Teerijoki et al., 2000; Teerijoki et al., 2001). Planas et al. (Planas et al., 2000) reported molecular identification of the GLUT of the brown trout Salmo trutta, which is structurally related to mammalian GLUT4. We and S. Panserat (INRA, France) have cloned the rainbow trout GLUT2 (GenBank AF321816). From an analysis of the derived amino acid sequence, the residues involved in glucose transport (Barrett et al., 1999) appear to be well conserved in these proteins. However, direct evidence for the functionality of fish GLUTs has hitherto been lacking.

The importance of facilitative glucose transporters for fish remains undefined. To study its functional properties, we expressed the putative glucose transporter OnmyGLUT1 in Xenopus laevis oocytes. This heterologous expression system has been very useful for characterizing mammalian and protozoan hexose transporters. The relative abundance of OnmyGLUT1 transcripts in rainbow trout embryos suggested that this protein might play an important role in development. In mammals, GLUT1 is referred to as an early isoform because it is expressed at high levels in embryos and foetuses, being gradually substituted for other GLUT types during the course of development (Santalucia et al., 1992; Postic et al., 1994). We have analysed the expression of OnmyGLUT1 at a number of developmental stages (blastula, early and late gastrula, and during somitogenesis and the formation of the vitelline plexus) using whole-mount in situ hybridization. We also measured glucose uptake and determined the ratio of transport to hexokinase activity in embryos and dissociated embryonic cells.

OnmyGLUT1 was amplified by reverse transcriptase/polymerase chain reaction (RT-PCR) from RNA extracted from newly hatched rainbow trout [Oncorhynchus mykiss (Walbaum)] alevins. Primers 5′-ACTGATCAACCACCATGGATTCAGGCGGCAAGCAAG-3′ (forward) and 5′-TGTG-ATCATTAGAGTTGAGAGTCAGCCCCCAGG-3′ (reverse) were designed to introduce a BclI restriction site (underlined). The forward primer also included a strong eukaryotic Kozak consensus (CACC). PCR amplification was first carried out with Pfu DNA polymerase (Stratagene) (annealing temperature 68°C). After 20 cycles, Taq DNA polymerase (MBI Fermentas) was added to increase the yield of product, and the reaction was continued for an additional 17 cycles (annealing temperature 60°C). The PCR product was cloned into pcDNA3.1/V5/His-TOPO vector (Invitrogen). The insert was excised by BclI digestion and cloned into the BglII site of pSPGT1. This vector contains 5′- and 3′-untranslated sequences of Xenopus β-globin mRNA (Kayano et al., 1990). The resulting construct was verified by DNA sequence analysis.

Xenopus laevis oocytes were prepared and used for hexose uptake analyses as described elsewhere (Penny et al., 1998). In brief, mRNA encoding OnmyGLUT1 was transcribed (MEGAscript SP6 with cap-analog; Ambion, Austin, TX, USA) from template linearised with XbaI (Promega). Oocytes were injected with either mRNA (15ng) or RNase-free water. All studies were carried out 24–72h after microinjection. Uptake of labelled sugars was measured under zero-trans conditions. Groups of 8–10 oocytes were placed into 600μl of Barth’s medium containing labelled hexose (0.5μCi; 18.5kBq). The reaction was terminated after 20min, and the oocytes were washed three times with ice-cold Barth’s medium. Oocytes were then placed individually into scintillation counter vials. Uptakes were corrected for the uptake by water-injected controls. To verify the functionality of the transporter, d-[U-¹⁴C]glucose, 3-O-methyl-d-[U-¹⁴C]glucose (3-OMG) and d-[U-¹⁴C]fructose (all from Amersham Pharmacia Biotech, Amersham, UK) were used. To study kinetics, substrate selectivity and sensitivity to inhibitors, the uptake of labelled d-glucose was measured. Concentrations of d-glucose, competitors and inhibitors (all from Sigma) are given in the figures and Table1. To measure transport in Na⁺-free medium, Na⁺ was replaced with equimolar choline chloride. Kinetic parameters were estimated by nonlinear regression analysis using a Michaelis–Menten model, and K_i was calculated using a one-site competition model (PRISM Ver.2, GraphPad, San Diego, CA, USA).

Rainbow trout eggs were incubated at 10°C. The time of formation of one somite pair, t_s unit (Gorodilov, 1996), was used to determine the developmental age of embryos. At 10°C, t_s is equal to 167min. Samples were collected at 25t_s (blastula), 36t_s (early gastrula), 42 and 50t_s (mid and late gastrula), 60–62t_s (5–7 somites), 69–70t_s (14–15 somites), 88–90t_s (33–35 somites), 115–117t_s (60–62 somites), 130t_s (early vitelline plexus) and 150t_s (middle vitelline plexus). These developmental stages have been described in detail elsewhere (Gorodilov, 1989; Gorodilov, 1996). At every stage, expression of OnmyGLUT1 was confirmed by RT-PCR. Cytochalasin-B-sensitive uptake of labelled 3-OMG (NEN Life Science Products Boston, MA, USA) was measured as described elsewhere (Teerijoki et al., 2000).

OnmyGLUT1 in pcDNA3.1-TOPO backbone (Invitrogen) was linearised with KpnI (MBI Fermentas) and used to generate the RNA probe. In vitro transcription with T7 RNA polymerase (MBI Fermentas) was carried out in the presence of DIG-oxigenin-11-UTP (Boehringer Mannheim). Embryos were fixed and processed as described previously (Joly et al., 1993) and the transcript was detected using a Boehringer Mannheim kit according to the manufacturer’s recommendations. Stained embryos mounted in glycerol were observed and photographed using an Olympus SZX9 stereomicroscope.

Transport measurements were performed in dissociated embryonic cells using 2-deoxy-d-glucose (2-DOG); this hexose is phosphorylated by hexokinase. Embryos were excised from chorions and washed with phosphate-buffered saline (PBS) to remove yolk. To facilitate dissociation, embryos were incubated in calcium- and magnesium-free PBS for 30min. Embryos were triturated first with a glass Pasteur pipette and then with an automatic pipette equipped with a 1ml plastic tip. Cells were filtered through tissue paper, centrifuged for 30s at 600g and resuspended in PBS. Using the exclusion of erythrosin B as an indicator, over 95% of the cells were found to be alive. Uptake was initiated by adding incubation medium (100μl), which included PBS, 5mmoll⁻¹ 2-DOG and 1.0μCi (37.0kBq) of label (2-deoxy-d-[1-³H]glucose; Amersham Pharmacia Biotech). To account for carrier-independent binding of label, cytochalasin B (50μmoll⁻¹) was used in parallel incubations. Cells were incubated with labelled 2-DOG for 20min with periodic shaking. To complete the reaction, the cells were loaded onto the surface of oil (1-bromododecane; Sigma) and centrifuged briefly at 13000g. The medium was removed by aspiration, and the oil surface was washed three times with PBS. After removal of the oil, the cells were lysed in water with 0.2% Triton X-100 and heated to 90°C for 3min. To measure total and non-phosphorylated 2-DOG, lysates were divided into two samples. Phosphorylated sugar was removed using treatment with Ba(OH)₂ and Zn(SO)₄ (as described by Colville et al., 1993). Equal volumes of water were added to the parallel samples. Protein was determined using a BioRad kit (catalog no. 500–01210).

Embryos were homogenized in 5mmoll⁻¹ K₂HPO₄, 5mmoll⁻¹ KH₂PO₄ (pH7.8 at 20°C), 0.5mmoll⁻¹ EDTA, 250mmoll⁻¹ sucrose and centrifuged at 900g for 2min to remove cell debris. The homogenate was added to a mixture containing 75mmoll⁻¹ Tris (pH7.4), 7.5mmoll⁻¹ MgCl₂, 0.8mmoll⁻¹ EDTA, 1.5mmoll⁻¹ KCl, 0.4mmoll⁻¹ NADP, 2.5mmoll⁻¹ ATP and 0.7i.u.ml⁻¹ glucose-6-phosphate dehydrogenase. The reaction was started by adding glucose to a final concentration of 1mmoll⁻¹. Increases in light absorbance were measured at 340nm.

Carrier-mediated transport of d-glucose was examined over a period of 3 days after microinjection of OnmyGLUT1 mRNA. The rate of d-glucose uptake was 2.3 times greater than that of 3-OMG and 13.9 times greater than that of d-fructose. Accumulation of labelled d-glucose in oocytes was linear over a period of 30min at room temperature (20–22°C). Kinetic analyses and studies with competitors and inhibitors were carried out 1 or 2 days after microinjections, and oocytes were incubated with d-glucose for 20min. The kinetic characteristics of d-glucose transport under zero-trans conditions were determined in three independent experiments. As expected, transport was saturable and followed Michaelis–Menten kinetics: K_m ranged from 8.3 to 14.9mmoll⁻¹. Fig.1 shows the result of a typical experiment. Transport of d-glucose was inhibited by cytochalasin B (1–100μmoll⁻¹) and phloretin (10–150μmoll⁻¹) (Fig.2). The inhibitory effect of phloridzin was detected at concentrations over 500μmoll⁻¹, and there was no decrease in the rate of transport in Na⁺-free medium. K_i for cytochalasin B determined in four experiments ranged from 1.4 to 2.3μmoll⁻¹ (data not shown). The rate of d-glucose transport decreased in the presence of 3-OMG, 1-deoxy-d-glucose (1-DOG), 2-DOG, 3-DOG, 6-DOG and d-mannose, whereas l-glucose, d-fructose, d-galactose, d-xylose, d-ribose, d-allose, l-sorbitol and mannitol did not compete with d-glucose (Table1).

Expression of OnmyGLUT1 was confirmed at all analysed stages using RT-PCR. Spatial patterns of OnmyGLUT1 expression were analysed using whole-mount in situ hybridization. The distribution of transcripts was ubiquitous at early developmental stages. In the early gastrula, OnmyGLUT1 was expressed throughout the embryo; however, hybridization of the probe was most evident on the edges of the blastodisc in the regions of the germ ring and terminal node (Fig.3A). These structures are formed at the beginning of gastrulation and they are the places with the highest concentration of cells. In the mid and late gastrula, OnmyGLUT1 expression was also clearly evident in all cells (data not shown).

Gastrulation in salmonid fish is completed when the dense core that includes the anlagen of the axial organ complex develops in the middle of the embryonic shield; this consists of the notochord, two bands of unsegmented prospective somite mesoderm and the neural plate (Gorodilov, 1989; Gorodilov, 1996). During the course of somitogenesis, 70–72 somite pairs and primordia of almost all organs and functional systems are formed in rainbow trout embryos. At early somitogenesis (5–7 somites), as well as in the preceding developmental stages, OnmyGLUT1 expression was observed throughout the embryo and blastoderm (Fig.3B). The spatial pattern of expression gradually became increasingly restricted over the course of development. Uneven distribution of the probe was seen first at the 14–15 somite stage. An abundance of transcript was observed in the forebrain, especially in the region of the eye vesicles (Fig.3C). OnmyGLUT1 was also expressed in the neural crest. This structure lies at the borders between the neural plate and lateral ectoderm along the entire axis of the embryo. Expression in the neural crest was observed during the whole study period (Fig.3C–F); however, after the completion of somitogenesis, hybridization of the probe was most apparent in the head part of embryo (Fig.3E,F). In addition to the neural crest, OnmyGLUT1 expression was found in a number of other embryonic structures such as the gill arches, pectoral fins, upper jaw, primordia of mouth lips and olfactory organs. No hybridization was observed on the dorsal surface of the brain or in the cranial dorsal ectoderm. Diffuse staining of ectoderm was found below the border of the neural crest and in the ventral part of caudal segments. Over the course of development, somitic expression of OnmyGLUT1 decreased along the rostro-caudal axis. At the 60–62 somite stage, hybridization was weak in the anterior trunk myotomes, in contrast to hybridization in the posterior trunk and tail somites (Fig.3D). No expression of OnmyGLUT1 was evident in the somites of advanced embryos (Fig.3F). We did not detect the transcript in the notochord, kidney, heart, liver or gut.

Cytochalasin B (50μmoll⁻¹) significantly decreased 3-OMG uptake by rainbow trout embryos at all analysed stages except the blastula (data not shown). To assess the contribution of transport by OnmyGLUT1 to hexose uptake, we carried out assays with 2-DOG (Table2). Hexose absorbed by dissociated embryonic cells was separated into phosphorylated and non-phosphorylated moieties, as described in Materials and Methods. Cytochalasin B decreased the content of 2-DOG phosphate (by 75.6%) in cells to a greater extent than that of non-phosphorylated 2-DOG (by 18.1%). This result suggested that transport might be rate-limiting for the delivery of glucose to embryonic fish cells. If the rate of uptake were limited by the rate of phosphorylation, inhibitors of transport (such as cytochalasin B) would have affected the accumulation of 2-DOG to a much greater extent than the accumulation of phosphorylated hexose. Hexokinase activity in rainbow trout embryos (1.127±0.037nmolmin⁻¹μg⁻¹protein, mean ± s.e.m., N=8) was two orders of magnitude greater than that of hexose transport and was therefore not rate-limiting in our experiments.

Three putative facilitative glucose transporters have been identified in salmonid fish, although none of these has been characterized functionally. To obtain direct evidence that one of these transporters, OnmyGLUT1, mediates glucose uptake, we expressed the protein in Xenopus laevis oocytes. This expression system is commonly used for studies of the functional properties of facilitative glucose transporters because the oocyte has low levels of endogenous glucose transport and efficiently translates foreign RNA. Our results showed that OnmyGLUT1 encodes a fully functional Na⁺-independent glucose transporter. The K_m for d-glucose transport was similar to those reported in fish cells. Soengas and Moon (Soengas and Moon, 1995) measured a K_m of 10.4mmoll⁻¹ for 3-OMG uptake by red blood cells of the American eel (Anguilla rostrata). The glucose carrier was not identified in that study. However, transport was probably mediated by a GLUT1-like carrier because mammalian erythrocytes harbour exclusively this isoform. We studied 3-OMG uptake in the carp EPC cell line and estimated that the K_m was 8.5mmoll⁻¹ (Teerijoki et al., 2001). A glucose transporter cloned from these cells has been identified as GLUT1. In human cells that employ GLUT1, the K_m for d-glucose ranged from 4 to 10mmoll⁻¹ (Stein, 1990). Similar values have been determined for mammalian GLUT1 expressed in Xenopus laevis oocytes. The K_m of rat GLUT1 for 2-DOG is 6.9mmoll⁻¹ (Burant and Bell, 1992). Woodrow et al. (Woodrow et al., 2000) reported a slightly higher affinity of mammalian GLUT1 for d-glucose (K_m=2.5mmoll⁻¹). d-Glucose transport mediated by OnmyGLUT1 is inhibited by cytochalasin B and phloretin, the classic inhibitors of facilitative glucose transporters. The K_i for cytochalasin B was similar to that of human GLUT1 expressed in Xenopus laevis oocytes (Woodrow et al., 2000).

Competitive inhibition analyses (Table1) were designed to address the substrate specificity of OnmyGLUT1. Uptake of d-glucose was not inhibited by l-glucose, which suggested stereospecificity of transport. Transport activity for d-glucose decreased in the presence of 1-DOG, 2-DOG, 3-DOG and 6-DOG, suggesting that the removal of hydroxyl groups from these positions did not eliminate the ability to compete with d-glucose. d-Mannose, the C2 epimer of d-glucose, also decreased d-glucose transport, suggesting an insignificant role of equatorial hydroxyl in C2. Substitution or removal of the C3 hydroxyl group reduced transport activity. For example, OnmyGLUT1 showed higher activity with d-glucose than with 3-OMG. The inhibitory effects of 2-DOG and mannose were greater than those of 3-DOG and 3-OMG. d-Fructose, d-galactose, d-xylose, d-ribose, d-allose, l-sorbitol and mannitol did not compete with d-glucose. Similar results were obtained with human (Gould et al., 1991; Woodrow et al., 2000) and carp (Teerijoki et al., 2000; Teerijoki et al., 2001) GLUT1. Of all the vertebrate facilitative glucose transporters, OnmyGLUT1 is most similar to GLUT1, both from an analysis of its primary sequence and from its functional characteristics. Mammalian GLUT3 and GLUT4 are characterised by a higher affinity for d-glucose. The K_m of GLUT2 for d-glucose is greater than that of OnmyGLUT1, and GLUT2 is capable of fructose transport.

We analysed hexose transport activity and expression of OnmyGLUT1 in embryos to investigate its importance for rainbow trout development. Uptake of 3-OMG by embryos was inhibited by cytochalasin B. We carried out assays with 2-DOG to determine the relative contributions of transport and phosphorylation to hexose uptake in embryonic cells. Accumulation of 2-DOG-phosphate was almost completely abolished by cytochalasin B, suggesting that metabolized sugar was absorbed predominantly via the facilitative carrier.

We studied the temporal and spatial patterns of OnmyGLUT1 expression in rainbow trout embryos. In mammals, expression of GLUT1 is markedly increased after fertilization. Both mRNA and GLUT1 protein are accumulated during pre-implantation in accord with an increase in glucose transport activity during development (Hogan et al., 1991; Morita et al., 1992; Dan-Goor et al., 1997). Using RT-PCR, the earliest stage at which OnmyGLUT1 expression was detected was the blastula. In situ hybridization analyses suggested that, at early developmental stages, OnmyGLUT1 was expressed in all cells. Transcripts were found throughout the embryo until the 5–7 somite stage (Fig.3B), but hybridization was most evident in the germ ring and embryo proper. These structures are known for their high density of cells and active morphogenetic processes. Restricted spatial patterns of OnmyGLUT1 expression were observed beginning from the 14–15 somite stage (Fig.3C). In older embryos, expression declined during development along the rostro-caudal axis and was undetectable in the myotomes, which are the derivatives of the somites. Expression of mammalian GLUT1 is also reduced during the differentiation of skeletal muscle. Myogenic differentiation includes the fusion of myoblasts into multinuclear myotubes, with subsequent upregulation of muscle-specific protein expression. Guillet-Deniau et al. (Guillet-Deniau et al., 1994) reported the expression of different GLUT proteins at these stages; GLUT1 was abundant in myoblasts, expression of GLUT3 increased markedly at cell fusion and GLUT4 was found exclusively in myotubes.

Expression of OnmyGLUT1 was stable in embryonic neural tissues of the rainbow trout. The transcripts were abundant at the boundary between the neural plate and lateral ectoderm (Fig.3C–F). In vertebrate embryos, this area harbours the cells of the neural crest (Artinger et al., 1999), which is an exceptionally important structure found exclusively in vertebrates (Baker and Bronner-Fraser, 1997; Gorodilov, 2000). The neural crest gives rise to the connective, skeletal and muscle tissues of the head; moreover, it produces skeleton, pigmented cells and nervous roots in the trunk/tail (Gans and Northcutt, 1983; Langille and Hall, 1989; Couly et al., 1993; Kontges and Lumsden, 1996). The detection of OnmyGLUT1 transcripts in putative derivatives of the neural crest (cartilaginous tissue of the pectoral fins, gill arches and upper jaw) confirmed that OnmyGLUT1 expression is likely to distinguish this cell population. Smoak and Branch (Smoak and Branch, 2000) reported a preponderance of GLUT1 protein in the heart and neural tube in mouse embryos during early organogenesis. An increase in GLUT1 expression in response to hypoglycaemia demonstrated the importance of the enzyme for the embryonic heart. We did not detect OnmyGLUT1 transcripts in the heart of rainbow trout embryos. This finding was unexpected since, in adult fish, this gene is expressed predominantly in the cardiac muscle (Teerijoki et al., 2000). Mammalian GLUT1 is expressed in virtually all embryonic and foetal tissues. In contrast, no OnmyGLUT1 transcripts were found in the notochord, kidney, liver and gut of the rainbow trout embryos.

The distribution of OnmyGLUT1 transcripts in rainbow trout embryos was clearly related to active morphogenetic processes. Expression was most stable in the neural crest cells, which are unique in their totipotent characteristics and extensive migration abilities. It is likely that demand for glucose is increased in differentiating cells because of high metabolic activity. Another possibility is that GLUT is substituted for other isoforms upon differentiation. Recent identification of salmonid GLUT4 and GLUT2 will allow this issue to be addressed.

This study was funded by the Technology Development Centre, Finland (TEKES). H.T. was supported by a grant from the Saastamoinen Foundation.