Regulation of glycogen metabolism in gills and liver of the euryhaline tilapia (Oreochromis mossambicus) during acclimation to seawater

Glucose plays a central role in providing energy for metabolism. It is primarily stored in animal tissues as a long-branching high-molecular-mass polysaccharide called glycogen (Roach et al., 1998). Glycogen metabolism is the principal energy source in both vertebrates and invertebrates, especially during environmental fluctuations (Karlsson, 1979; Hoffman and Katz, 1998; Oliveira et al., 2004; Bacca et al., 2005). The synthesis and degradation of glycogen are mainly initiated by glycogen synthase (GS; EC 2.4.1.11) and glycogen phosphorylase (GP; EC 2.4.1.1),respectively. GS, which constitutes an additional mechanism of control over glycogen metabolism, plays a major role in glycogen storage and is also the rate-limiting step in the glycogen synthesis pathway(Fernandez-Novell et al.,1992; Garcia-Rocha et al.,2001). On the other hand, GP is a rate-limiting enzyme that maintains energy support for various tissues. It is composed of two identical subunits that control the breakdown of glycogen to glucose 1-phosphate(Newgard et al., 1989; Frolow and Milligan, 2004; Greenberg et al., 2006).

Because gills are directly exposed to the outer aquatic environment, they have multiple functions including gas exchange, acid–base balance and ionic/osmotic regulation (Evans et al.,2005). In terms of the mechanisms of ionic/osmotic regulation,mitochondrion-rich (MR) cells in gill epithelia are the main sites responsible for active ion transport functions, which are conducted by the operations of various ion transporters and enzymes. These operations are highly energy-consumptive processes (Hirose et al., 2003; Evans et al.,2005).

It has been well documented that acclimation to hypertonic seawater (SW) in euryhaline teleosts involves a timely and sufficient activation of an ion excretion system. This includes the morphological modifications of gill MR cells and stimulation of both the expression and activity of ion transporters(Hwang, 1987; Hwang et al., 1989; Marshall et al., 1999; Lee et al., 2003). These processes require an additional energy supply, reflecting changes in oxygen consumption in fish upon encountering fluctuations in environmental salinity(Boeuf and Payan, 2001). However, monitoring oxygen consumption, which most studies have emphasized, is not a direct approach for determining energy metabolism in a gill cell that conducts ion regulation (Morgan and Iwama,1991; Morgan and Iwama,1998). Several studies have addressed the major role that carbohydrate plays in energy metabolism for osmoregulation. Energy may be mainly supplied by the oxidation of glucose and lactate obtained from the circulation as a result of carbohydrate metabolism(Perry and Walsh, 1989; Morgan et al., 1997). Indeed,consumption of the glycogen content of the liver has been reported in rainbow trout (Oncorhynchus mykiss) after transfer to SW(Soengas et al., 1991). Hepatic metabolism was found to be related to salinity acclimation in euryhaline fishes (Nakano et al.,1998; Sangiao-Alvarellos et al., 2003).

In earlier studies on teleost gills, glycogen granules/particles were found in MR cells by electron microscopic observations(Philpott and Copeland, 1963);however, this was not further confirmed by molecular evidence. Recently, an antibody against glycogen was used to identify glycogen deposits localized in a group of cells, glycogen-rich (GR) cells, in tilapia gills, and these GR cells express a gill form of GP (Tseng et al., 2007). Moreover, the glycogen content and expression/activity of GP in these cells were affected by environmental salinity(Tseng et al., 2007). These results imply that glycogenesis and glycogenolysis in fish gills may be involved in energy metabolism during acclimation to salinity changes; however,the detailed mechanism is still unclear. No study has clarified the partitioning of energy supplements between the liver and gills in gill energy requirement during acclimation to salinity changes.

In the present study, we attempted to examine glycogen metabolism in fish liver and gills during acute exposure to SW. Mozambique tilapia(Oreochromis mossambicus), a euryhaline cichlid, was selected for study due to its ability to adapt to acute fluctuations in salinity. The following experiments were performed: (1) GS was cloned and sequenced from tilapia, and the expression of GS in various tissues, including the gills, was examined; (2) immunocytochemistry was used to localize GP, GS and glycogen in tilapia gill cells; (3) time-course changes in glycogen content in tilapia liver and gills after exposure to SW were examined; (4) time-course changes in GP protein and/or activity levels in tilapia liver and gills after exposure to SW were examined and (5) time-course changes in GS protein level in tilapia liver and gills after exposure to SW were examined.

Tilapia (Oreochromis mossambicus Peters), 40–60 g in body mass and 10–14 cm in total length, were obtained from stocks of the Institute of Cellular and Organismic Biology, Academia Sinica, Taipei, Taiwan. The fish were kept in local freshwater (FW) circulating in a tank at 27–28°C under a 14 h:10 h light:dark photoperiod and were fed daily with artificial feed pellets (Fu-So, Taipei, Taiwan). The experimental protocols were approved by the Academia Sinica Institutional Animal Care and Utilization Committee (approval no. RFiZOOHP2006083).

Seawater (SW) was prepared by adding appropriate amounts of artificial sea salt (Taikong, Taipei, Taiwan) into the FW. In exposure experiments, FW tilapia were directly transferred to 25‰ SW and sampled at 1, 3, 6, 12,24, 48 and 168 h after the transfer. The fish in the acute exposure experiment were not fed. In total, eight fish were sacrificed at each sampling point:four fish for the experimental treatment (FW to SW) and another four fish for the control (FW transferred to FW). All samples were processed at the same time: 09.30–11.30 h in order to normalize the effects of circadian rhythms on physiological metabolism. Sampled gills and liver were immediately placed in ice-cold 1% phosphate-buffered saline (PBS) and were subsequently subjected to the treatments described below.

Preliminary experiments indicated that blood cells and muscle cells contain considerable amounts of glycogen, GP and GS. Isolation of epithelial cells from gill tissue is necessary in order to exclude the effects of these non-epithelial cells. The isolation method followed a previous paper(Tseng et al., 2007). Gills were carefully rinsed with PBS and then tissue was scraped from the gill filaments and immediately kept in dissociation buffer (0.5 mol l^-1EDTA, 500 μl Percoll in PBS) on ice. After rigorous agitation with a stirring bar on ice for 30 min, gill cells were isolated from the tissue by gently filtering the solution through a 100-μm nylon mesh to remove the larger tissue fragments. The filtered cell suspension was poured into a Percoll (Sigma, St Louis, MO, USA) solution (2:1:1 cell suspension:Percoll:PBS) and centrifuged for 10 min at 2000 gat 4°C. The epithelial cell fraction was collected, washed with PBS and centrifuged. Afterwards, the isolated cells were subjected to western blotting with an anti-tropomyosin (a muscle-specific protein) monoclonal antibody (mAb)(Sigma) to confirm that there was no contamination of muscle cells (data not shown). The isolated epithelial cells were stored at –80°C.

In the process of preparing mRNA, 200–300 mg of gill tissues was homogenized in 3 ml of Trizol reagent (Invitrogen, Carlsbad, CA, USA) and treated according to the manufacturer's protocols. The amount and quality of total RNA were determined by measuring the absorbance at 260 and 280 nm with a spectrophotometer (Hitachi U-2000, Tokyo, Japan) and analysis using RNA-denatured gels. The total amount of RNA was subsequently extracted with a QuickPrep Micro mRNA Purification Kit (Amersham Pharmacia, Piscataway, NJ,USA). Finally, the pellets of mRNA were precipitated with 0.1 mg glycogen,1/10 volume of 3 mol l^-1 NaOAc and 95% ethanol and stored at–20°C.

For cDNA synthesis, 0.36 μg of mRNA was reverse-transcribed in a final volume of 20 μl containing 0.5 mmol l^-1 dNTPs, 2.5 μmol l^–1 oligo (dT)₁₈, 5 mmol l^-1dithiothreitol and 200 units of PowerScript reverse transcriptase (Clontech,Palo Alto, CA, USA) for 1.5 h at 42°C, followed by a 15-min incubation period at 70°C. The cDNA samples were finally stored at–20°C.

Degenerate oligonucleotide primers for PCR were designed based on the conserved sequences of GS from different vertebrates. The primer set for cloning GS consisted of forward (5′-AATGTGAAGAAGTTCTCKGC-3′) and reverse (5′-GCAAAGATGGCYCTCTTCAT-3′) sequences. For PCR amplification, 2 μl cDNA was used as a template in a 50 μl final reaction volume containing 0.25 mmol l^-1 dNTP, 2.5 units EXTaq polymerase (Takara, Shiga, Japan) and 0.2 μmol l^–1 of each primer. The PCR products were subcloned into the pGEM-T Easy vector (Promega,Madison, WI, USA), and the nucleotide sequence was determined with an ABI 377 DNA Sequencer (ABI, Warrington, UK). Sequences were analyzed with the BLASTx program (NCBI).

The 5′ and 3′ RACE (rapid amplification of cDNA ends) cDNAs were cloned with a commercial kit (SMART RACE cDNA Amplification Kit;Clontech) following the manufacturer's protocols. The specific primers of GS for 5′ and 3′ RACE were 5′-CGGCGAACACAGTTGAGGATGGG-3′and 5′-ACAGCAGCGACCCCATCCTCAA-3′, respectively. The RACE PCR products were also subcloned into the pGEM-T Easy vector and sequenced. The entire amino acid sequence deduced by the CLUSTAL program(Higgins and Sharp, 1988; Higgins et al., 1996) was used for multiple sequence alignments and analysis of the phylogenetic tree. The dataset was subjected to a distance analysis using the maximum parsimony (MP analysis) method, and 1000 bootstrap replicates of the analysis were carried out with the MEGA program version 2.1.

Total RNA samples were extracted from the brain, gills, liver, muscle,intestines, heart, spleen and kidneys in FW tilapia. The samples were subjected to RT-PCR analysis, and β-actin gene expression was utilized as an internal control. The primer set for GS consisted of the forward(5′-TGGGATACTGCTCAGACTGTGA-3′) and reverse sequences(5′-TGTCCTCCAGCATGTTGTGAGT-3′) (a 187-bp fragment), and that for the control β-actin consisted of the forward(5′-CGGAATCCACGAAACCACCTA-3′) and reverse sequences(5′-ATCTCCTGCATCCTGTCA-3′) (a 135-bp fragment). Denaturation was performed for 3 min at 95°C, and thereafter for 30 s at 95°C. The annealing time was 30 s at 57°C, and the elongation time was 30 s at 72°C. The reaction was run for 30 cycles. All amplicons were sequenced to confirm that the PCR products were the desired gene fragments.

Both mouse and tilapia tissues (isolated gill epithelial cells and liver tissues) were homogenized with the homogenization solution (100 mmol l^-1 imidazole-HCl, 5 mmol l^-1 Na₂EDTA, 200 mmol l^-1 sucrose and 0.1% sodium deoxycholate; pH 7.6) at 600 revs min^–1 on ice. After centrifugation at 4°C and 9800 g for 30 min, the total protein concentration of the supernatant was measured with a protein assay kit (protein assay dye reagent concentrate kit; Bio-Rad, Hercules, CA, USA), and the supernatant was then stored at –80°C. The protein sample (50 μg) was supplemented with 6× electrophoresis sample buffer (250 mmol l^–1Tris-base, 2 mmol l^–1 Na₂EDTA, 2% SDS and 5%dithiothreitol) and then incubated at 95°C for 10 min. The denatured samples were subjected to SDS–polyacrylamide gel electrophoresis(SDS-PAGE) at 110 V for 2 h. After being transferred to polyvinylidene difluoride membranes, the blots were incubated in 5% nonfat milk for 3 h at room temperature and then washed twice with PBST buffer (0.01 mol l^–1 phosphate, 0.09% NaCl, pH 7.5 and 0.05% Tween 20). The blotted membranes were incubated overnight with a rabbit anti-human GS polyclonal antibody (Rockland, Gilbertsville, PA, USA; diluted 1:750) at 4°C. After being washed twice with PBST buffer, the blotted membranes were reacted for another 2 h with an alkaline-phosphatase-conjugated goat anti-rabbit IgG antibody (Jackson ImmunoResearch, Cambridgeshire, UK, diluted 1:1000). After washing with PBST buffer, immunoreactive proteins were visualized with nitro blue tetrazolium (NBT) and 5-bromo-4-chloro-3-indolyl phosphate (BCIP) in staining buffer. Immunoblots were scanned and exported as JPEG files, and the differences between the band intensities of FW and SW samples were compared using a commercial software package (Image-Pro Plus 4.0;Media Cybernetics, Silver Spring, MD, USA).

Na⁺/K⁺-ATPase activity was determined as described by Hwang et al. (Hwang et al.,1989). Isolated gill epithelial cells were briefly homogenized in the homogenization solution [100 mmol l^–1 imidazole-HCl buffer (pH 7.6), 5 mmol l^–1 Na₂EDTA, 200 mmol l^–1 sucrose and 0.1% sodium deoxycholate]. The Na⁺/K⁺-ATPase activity was assayed in the reaction mixture containing 100 mmol l^–1 imidazole-HCl buffer (pH 7.6), 125 mmol l^–1 NaCl, 75 mmol l^–1 KCl,7.5 mmol l^–1 MgCl₂ and 5 mmol l^–1Na₂ATP. The reaction was run at 37°C for 30 min and then stopped by adding 200 ml of ice-cold 30% trichloroacetic acid. The enzyme activity was defined as the difference between the inorganic phosphates liberated in the presence and absence of 3.75 mmol l^–1ouabain in the reaction mixture. Each sample was assayed in triplicate.

Measurement of gill glycogen phosphorylase activity followed a procedure described by Milligan (Milligan,2003), with some modifications. Isolated gill epithelial cells were homogenized in ice-cold homogenization solution (100 mmol l^–1 imidazole, 100 mmol l^–1 KF, 5 mmol l^–1 EDTA and 1 mmol l^–1phenylmethylsulfonylfluoride). Total GP (GPa+GPb) activity was measured by incubating the samples at 25°C in the presence of 1.6 mmol l^–1 5′AMP, 45 mmol l^–1potassium-phosphate buffer (pH 7.0), 0.2 mg ml^–1 glycogen,0.34 mmol l^–1 NADP, 4 μmol l^–1glucose-1,6-bisphosphate, 0.1 mmol l^–1 EDTA, 15 mmol l^–1 MgCl_2, 1.6 U ml^–1phosphoglucomutase and 12 U ml^–1 glucose-6-phosphate dehydrogenase. The absorbance changes between the reactions with and without glycogen (the substrate) were measured at 340 nm with a Hitachi U-2000 spectrophotometer (Tokyo, Japan). Each sample was assayed in triplicate.

Isolated gill epithelial cells and liver tissues were homogenized in 30%KOH and heated to 100°C for 30 min. The samples were supplemented with two volumes of 100% ethanol and incubated overnight. Glycogen was precipitated by centrifugation after the addition of 2–3 drops of Na₂SO₄. The glycogen pellets were washed with 66%ethanol and then completely dried. The glycogen content was analyzed in 0.2%anthrone reagent dissolved in H₂SO₄ using a Hitachi spectrophotometer.

Fresh gills were fixed with 4% paraformaldehyde at 4°C for 3 h and then gradually immersed in PBS containing 5%, 10% and 20% concentrations of sucrose for 15 min, with each step at room temperature. Finally, they were soaked in a mixed PBS solution (OCT compound:20% sucrose, 1:2) overnight and then embedded in OCT compound embedding medium (Sakura, Tokyo, Japan) at –20°C. Cryosections (5 μm thick) were made with a cryostat (CM 1900; Leica,Heidelberg, Germany) and placed onto poly-l-lysine-coated slides(Erie, Hooksett, NH, USA). Prepared slides were then rinsed in PBS and blocked with 3% bovine serum albumin (BSA) for 30 min. Afterwards, the slides were incubated with the Na⁺/K⁺-ATPase α5 mAb (Hybridoma Bank, University of Iowa, Ames, IA, USA; diluted 1:200), mouse anti-human GP brain form mAb (Biotrend Chemikalien, Cologne, Germany; diluted 1:200) and/or glycogen mAb (Baba, 1993)(diluted 1:200) overnight at 4°C, respectively. The slides were then washed twice with PBS and incubated with anti-mouse IgG conjugated with Texas Red (Jackson ImmunoResearch Lab., West Grove, PA, USA; diluted 1:200) for 2 h at room temperature. After being washed with PBS twice, the slides were incubated again with the anti-human GS polyclonal antibody (diluted 1:200)overnight at 4°C and then washed again before being incubated with anti-rabbit IgG conjugated with fluorescein isothiocyanate (FITC) for 2 h at room temperature. The images were acquired with a Leica TCS-NT confocal laser scanning microscope (Leica Lasertechnik, Heidelberg, Germany). Observations were made in the trailing edge of gills where most of the MR cells locate.

Values are presented as means ± s.d. (N=6). All time-course data were statistically analyzed by two-way analysis of variance (ANOVA), and pairwise multiple comparisons were made by Tukey's test(P<0.05).

Full-length GS cDNA was cloned and sequenced from tilapia gills(Fig. 1) (GenBank Accession No. EF565371). The complete sequence obtained was 2660 nucleotides (nt) long and contained a 2133-nt open reading frame, a 359-nt 5′-untranslated region and a 168-nt 3′-untranslated region with a poly (A) tail. The deduced amino acid sequence encodes a protein of 711 amino acids. Using commercial software (Vector NTI; Invitrogen, San Diego, CA, USA) to analyze the amino acid sequence, we discovered that the GS from tilapia gills (tGS) is about 70%identical to the zebrafish GS muscle form, 65% to the mouse GS muscle form,62% to the mouse GS brain form, 55% to the mouse liver form and 62.5% to the human GS muscle form (data not shown). Analysis of sequence alignments indicated that the majority of the sequences were highly conserved(Pederson et al., 2000). The NH₂ and COOH termini are the most variable regions and are involved in regulating phosphorylation and dephosphorylation. Five important phosphorylation sites in a mammal (rabbit) GS, viz. Ser-650, Ser-654,Ser-658, Ser-662 and Ser-666 (Hardy and Roach, 1993), were also found to be conserved in the COOH termini of tilapia GS (Fig. 1).

In Fig. 2, a phylogenetic tree was generated based on the MP analysis of the full-length amino acid sequences of the GS isoforms (muscle and liver) from eight species (mouse,rat, rabbit, human, monkey, zebrafish, tilapia and Drosophila). The tGS cloned from the gill had the highest homology to the muscle form.

tGS mRNA expressions in different tissues of tilapia were examined by RT-PCR with β-actin as the internal control. Fig. 3 clearly indicates that the PCR products of the 187-bp fragment of tGS were present. The tGS gene was ubiquitously expressed among various tissues, with higher expression levels in the brain, heart and muscle.

Tilapia liver and isolated gill cells, and mouse muscle and kidney cells(as a positive control, following the Product Specification Sheet of Rockland)were subjected to western blotting with an anti-human GS polyclonal antibody. The antibody was raised against a synthetic peptide,A-Q-G-Y-R-Y-P-R-P-A-pS-V-P, which was derived from the 631–643 amino acid sequence of the human muscle GS. The 631–643 amino acid sequence of the tGS and mouse muscle GS showed 77% and 92% identities, respectively, to that of the human muscle GS. As shown in Fig. 4, immunoreactive bands with a similar size were found in all of the tissues examined. The size was about 80.9 kDa in mouse kidney, tilapia liver and gill cells, and 83.9 kDa in mouse muscle. Negative experiments with the pre-immune serum showed no bands(data not shown). These results indicate that the antibody recognizes different iosforms of the tGS and, thus, can be used to detect the tGS protein in gills and liver.

The specificities of anti-GP, anti-glycogen and anti-Na⁺/K⁺-ATPase antibodies have been confirmed in previous studies (Tseng et al.,2007; Hwang et al.,1999). Double immunohistochemical labeling was conducted to localize GS, GP, glycogen and MR cells. Differential interference contrast(DIC) observations showed distinct images of the outline of the afferent edge of a gill filament (Fig. 5A,D,H). FITC signals for GS, and Texas Red signals for Na⁺/K⁺-ATPase, were both found in gill epithelial cells,but they were not colocalized in the same cells(Fig. 5B,C), clearly indicating that GS was localized in a specific type of cells, GR cells(Tseng et al., 2007) (also see below). Based on confocal-microscopic observations from the serial sections in more than 10 individuals, all these GS-expressing GR cells were adjacent to MR cells (Fig. 5B,C). Moreover, GS and GP were colocalized in the same GR cells(Fig. 5E–G), and GS and glycogen were also found in the same type of cells(Fig. 5I–K). As a result,GS, GP and glycogen were all colocalized in the same group of cells, GR cells,which were recently identified because of their abundant deposits of glycogen(Tseng et al., 2007).

FW tilapia were transferred to 25‰ SW and FW (as a control) and then the activity of gill Na⁺/K⁺-ATPase was measured at different time points after the transfer. Gill Na⁺/K⁺-ATPase activity of the FW control group was maintained at 4–6 μmol Pi h^–1 mg^–1protein during the entire experiment (Fig. 6). However, the activity in the 25‰ SW group showed dramatic changes, with two peaks (Fig. 6). The activity exhibited a significant increase (of about 1.9-fold) from the first hour immediately after transfer to 25‰ SW, a high level was maintained during the subsequent 1–6 h, and then the level declined to that of the control. Thereafter, the activity showed another rapid and evident increase 24 h post-transfer, peaked (at about 3.2-fold)again at 48 h, and then maintained a high level until the end of the experiment.

Similar to the above transfer experiment, glycogen content in isolated gill and liver cells was measured at different time points after the transfer. Glycogen content in isolated gill (mainly in GR cells) and liver cells was affected by environmental salinity (Fig. 7A,B). Glycogen deposits in gill GR cells were depleted by about 50% in the first hour immediately after transfer from FW to 25‰ SW,then remained at a low level, and finally recovered to the level of the control (Fig. 7A). On the other hand, it was not until the sixth hour that the glycogen content began to be depleted in the liver, and the content of glycogen in the liver was maintained at a lower level, at ∼40–60% of that of the control FW group, until the end of the experiment (Fig. 7B).

Following the changes in the gill Na⁺/K⁺-ATPase activity and glycogen content, gill GP protein levels and activity and GS protein levels also showed dramatic changes(Fig. 8). In isolated gill cells, the patterns of change in the GP protein amount(Fig. 8B) and activity(Fig. 8A) were reversed compared with those of the GS protein (Fig. 8C). Gill GP activity began to show evident stimulation (of∼2-fold) from the third hour after transfer to 25‰ SW, it decreased thereafter and then recovered to a stable level similar to that of the control FW group from 12 h post-transfer until the end of the experiment(Fig. 8A). Simultaneously, the gill GP protein amount also increased immediately after transfer, peaked (at∼2-fold) in the third hour and returned to the control level from the 12th hour (Fig. 8B); however, the GP protein amount was higher than that in the control FW group at 168 h at the end of the experiment (Fig. 8B). By contrast, the gill GS protein amount in the SW group showed a sudden decline (to ∼60%) in the third hour after transfer, then immediately recovered to the original level thereafter, but it turned out to be higher than the control at the 168th hour at the end of the experiment(Fig. 8C).

The profiles of GP and GS protein amounts in the liver were similar to those in isolated gill cells; however, the changes in the liver appeared more slowly than did those in isolated gill cells (Figs 8, 9). As shown in Fig. 9A, the liver GP protein amount appeared to increase from the sixth hour after the transfer to 25‰ SW and returned to the same level as the control group from the 24th hour after transfer (Fig. 9A). By contrast, the liver GS protein amount showed a sudden decrease after 12 h and then immediately returned to the level of the control group (Fig. 9B).

The notable findings of the present study were as follows: (1) GS was cloned and sequenced from fish gills for the first time; (2) GS, GP and glycogen were co-localized in a specific group of gill cells, i.e. GR cells,which are adjacent to MR cells (the major ionocytes); (3) glycogen content in both gill and liver was significantly depleted after acute transfer from FW to 25‰ SW, but the depletion occurred earlier in gills than in the liver;and (4) gill Na⁺/K⁺-ATPase activity rapidly increased immediately after SW transfer. Following this, gill GP activity and protein expression were upregulated 1–3 h post-transfer and eventually recovered, while at the same time GS protein expression was downregulated. Similar changes in liver GP and GS protein expressions were observed but they occurred more slowly, at 6–12 h post-transfer to 25‰seawater.

The amino acid sequence of the tilapia GS (tGS) from the gills showed the highest identity of 70% with the zebrafish GS muscle form and 55–65%with the mammalian counterparts. Phosphorylation of the C-terminal serine residues of mammalian GS, a key regulator for controlling enzyme activity(Cohen, 1982; Roach, 1990), is conserved in the tGS. Moreover, the tGS also has two Lys residues, Lys38 and Lys300, which have been identified as binding sites for UDP-glucose in the mammalian enzyme(Tagaya et al., 1985; Mahrenholz et al., 1988). Based on these results, the tGS may be a homologue of the mammalian GS muscle form.

The mechanisms of glycogen synthesis and degradation are generally known in mammal tissues, including the liver, muscle and other organs(Smythe and Cohen, 1991; Bollen et al., 1998), and this mechanism seems to be similar in fish gills, an energy-consuming organ. Glycogen was reported to exist in the cytoplasm of fish branchial MR cells according to electron microscopic observations(Philpott and Copeland, 1963; Nakao, 1974). This traditional inference seems unlikely because of the present and previous(Tseng et al., 2007)convincing molecular and cellular evidence of the colocalization of GP, GS and glycogen in a specific group of cells (GR cells) but not in MR cells in fish gills. As addressed by Tseng et al., GR cells and MR cells share the same apical opening to form a multicellular complex(Tseng et al., 2007). More sensitive analysis, like immuno-electron microscopy, is necessary to further confirm whether GR cells are actually previously reported accessory cells or other transporting cells (Tseng et al.,2007). However, the present result raises the possibility that the relationship in energy translocation between mammal astrocytes and neurons may also occur between GR cells and MR cells in fish gills. In the rat brain, GP and glycogen mainly exist in astrocytes, astroglial cells and ependymal cells of ventricles but never in neurons(Pfeiffer-Guglielmi et al.,2003). During energy deprivation in the central nervous system,glycogen is degraded to lactate, which is shuttled from astrocytes to high-energy-requiring neurons (Ransom and Fern, 1997; Brown et al.,2003; Pfeiffer-Guglielmi et al., 2007).

Acclimation to SW in fish was suggested to be involved in activation of ion secretion pathways in gill MR cells (Evans et al., 2005), and this requires additional energy support(Morgan et al., 1997). Immediately after an acute challenge with SW, gill Na⁺/K⁺-ATPase activity showed a rapid and drastic increase from the first hour in order to recover the impaired internal ion levels and osmolalities in tilapia (Hwang et al., 1989), indicating that a prompt energy supply for stimulating ion-secretion mechanisms is critically needed. Indeed, Morgan et al. (Morgan et al., 1997)examined the oxygen consumption in tilapia after transfer to SW and found a significant increase after 4 days of acclimation. Other studies, however,reported a reduction in oxygen consumption in tilapia after 2 weeks to 1 month of acclimation to a higher salinity (Febry and Lutz, 1987; Sardella et al., 2004). Most of those previous studies focused on the energetic changes from several days to 1 month. So far, no observations have been made in the case of very short term (in terms of hours) energetic changes, which is critically important for the rapid stimulation of the gill Na⁺/K⁺-ATPase activity observed in the present study.

Carbohydrates, such as glucose and lactate, have been suggested to be an essential energy source for fish osmoregulation during acclimation to different salinities; hence, changes in glycogen content are another suitable indicator for monitoring energetic metabolism in a specific organ or cell(Assem and Hanke, 1979; Nakano et al., 1998; Sangiao-Alvarellos et al.,2005). Effects of environmental salinities on glycogen content in the liver, gills and/or other organs have been examined in several studies. Nakano et al. found no significant changes in liver glycogen contents in tilapia (O. mossambicus) after transfer from FW to 23‰ SW for 0.5–96 h (Nakano et al.,1998). Likewise, Assem and Hanke found no significant difference in glycogen content of the liver in the same species 6–168 h after transfer from FW to 25‰ SW (Assem and Hanke, 1979). The controversy between these previous data and the present results in tilapia may be partially due to differences in the fish size, the conditions (e.g. temperature) in which the fish were kept, the method of sampling, etc. On the other hand, Sangiao-Alvarellos et al. examined metabolic changes in the liver and gills in gilthead sea bream (Sparus auratus) during acclimation between 38‰ and 55‰ SW and found that glycogen was depleted in the liver on the first day while that in the gills accumulated (Sangiao-Alvarellos et al., 2005). These results imply that mobilization of liver glycogen may provide endogenous carbohydrate fuel to the gills(Sangiao-Alvarellos et al.,2005); however, no further investigations have been done within the first day after transfer, which is a very critical period for a fish exposed to a salinity stress as we addressed above. The present study examines the differential roles of the gills and liver and their spatial and temporal relationships to the energy supply for osmoregulation during acute exposure to a high-salinity environment. The present data of changes in glycogen content indicate that the glycogen in gills may provide prompt energy from the first hour after SW transfer; subsequently, from 6 h post-transfer, liver glycogen may become the major carbohydrate reserve supporting the operation of ion-secretion mechanisms in tilapia gills. Furthermore, the glycogen content in the liver is much greater, about 250–300-fold, than that in the gills, and this also reasonably reflects the different partitions of the two organs in energy supply. SW acclimation stimulates the activities of Na⁺/K⁺-ATPase and other transporters or enzymes in gills, intestine and other osmoregulatory organs(Kelly et al., 1999; Ando et al., 2003; Evans et al., 2005). Liver glycogen may provide energy for all the processes in these organs. It is impossible to distinguish the partitions of liver glycogen to different organs or different transporters; therefore, it is not surprising to find that the profile of liver glycogen (started to deplete at 6 h) was not totally in concert with that of gill Na⁺/K⁺-ATPase (second increase from 24 h).

In both the gills and liver of gilthead sea bream, GP activities and glycogen content changed in parallel immediately after transfer from 38‰ to 55‰ SW, but neither GS activity nor protein expression was studied (Sangiao-Alvarellos et al.,2005). Obviously, these previous findings might not correctly reflect the functional regulation of glycogenesis and glycogenolysis during hyperosmotic acclimation. On the contrary, in the present study, tilapia gill GP and GS protein expressions and/or activities showed opposite changes,increasing (upregulation of glycogenolysis) and decreasing (downregulation of glycogenesis), respectively, from the first to the third hours after transfer to 25‰ SW, and these profiles reasonably reflect depletion of gill glycogen. Similar results were also found in tilapia liver, but stimulation of glycogenolysis in the liver began later, from the sixth to the 12th hour after transfer. In previous studies on rainbow trout (O. mykiss),acclimation to 12‰ or 28‰ SW also stimulated liver glycogenolysis, which was associated with an increase in GP activity, a decrease in GS activity and a decline in glycogen levels(Soengas et al., 1991; Soengas et al., 1993). Based on our data, stimulation of glucogenolysis in the gills and thereafter in the liver probably provides a sequential energy supply for the operation of ion-secretion mechanisms, including Na⁺/K⁺-ATPase and other transporters, in tilapia gills during acute exposure to SW.

Glycogen is stored in the liver as a reserve of glucose for extrahepatic tissues, and the liver, acting as a sensor of blood glucose stores, mobilizes glycogen according to peripheral needs(Bollen et al., 1998). Therefore, it is not surprising to find in the present study and other previous work that mobilization of liver glycogen may provide carbohydrate reserves to fuel glycolysis in gills for operation of ion-regulation mechanisms during acclimation to salinity changes(Assem and Hanke, 1979; Soengas et al., 1991; Soengas et al., 1993; Sangiao-Alvarellos et al.,2005). The notable new finding from the present study is that the gills accumulate glycogen as a local carbohydrate reserve, and upon acute salinity stress, the glycogen in the gills is initially degraded to provide a prompt emergency energy supply for activation of salt secretion mechanisms before glycogenolysis in the liver is stimulated. However, direct evidence to demonstrate the energy shuttle between gill cells is still lacking and requires further studies. The role of the gills themselves in the emergency energy supply for osmoregulation seems to have been overlooked in previous studies that examined metabolic changes in gills of the gilthead sea bream during acclimation to different salinities(Sangiao-Alvarellos et al.,2003; Sangiao-Alvarellos et al., 2005). There are some possible reasons for this: first, those studies did not examine short-term changes (in terms of hours) as discussed above; second, perhaps there are differences in the methodologies for measuring glycogen content and GP expression and activity. Our preliminary experiments showed that blood in gill tissues comprised as much as >60% and>80%, respectively, of the glycogen content and GP protein amount in the whole gill (data not shown). Thus, the present study using isolated gill epithelial cells to examine metabolic changes in gill cells avoided the contaminating effects from blood and other cells. The actual changes in gill cells might have been obscured in previous studies by Sangiao-Alvarellos et al. (Sangiao-Alvarellos et al.,2003; Sangiao-Alvarellos et al., 2005), since they used whole gill tissues without isolating the gill cells.

Taking all this into account, we propose a model for carbohydrate metabolism for osmoregulation in tilapia gills during acute exposure to salinity stress: GR cells accumulate glycogen as a local carbohydrate reserve in gills, glucogenolysis in gill GR cells is initially stimulated to provide prompt energy for neighboring MR cells in order to trigger ion-secretion mechanisms, and several hours later, the liver begins to degrade its glycogen for the subsequent energy supply. This local and systematic partitioning of glycogen metabolism for emergency energy requirements is similar to what has been well documented in mammal brains (neurons and astrocytes) and liver.

This study was financially supported by grants to P.P.H. from the National Science Council and Academia Sinica, Taiwan, Republic of China. We extend our thanks to Ms Y. C. Tung and Mr J. Y. Wang for their assistance during the experiments and also to the Core Facility of the Institute of Cellular and Organismic Biology, Academia Sinica for assistance with confocal microscopy.