Characterization of a Hypoxia-inducible Factor (HIF-1a) from Rainbow Trout ACCUMULATION OF PROTEIN OCCURS AT NORMAL VENOUS OXYGEN TENSION*

The mammalian hypoxia-inducible factor-1 (HIF-1) is a heterodimeric transcription factor that controls the induction of several genes involved in glycolysis, erythropoiesis, and angiogenesis when cells are exposed to hypoxic conditions. Until now, the expression and function of HIF-1a have not been studied in fish, which experience wide fluctuations of oxygen tensions in their natural environment. Using electrophoretic mobility shift assay, we have ascertained that a hypoxia-inducible factor is present in rainbow trout cells. We have also cloned the full-length cDNA (3605 base pairs) of the HIF-1a from rainbow trout with a predicted protein sequence of 766 amino acids that showed a 61% similarity to human and mouse HIF-1a. Polyclonal antibodies against the N-terminal part (amino acids 12–363) and the C-terminal part (amino acids 330–730) of rainbow trout HIF-1a protein recognized rainbow trout and chinook salmon HIF-1a protein in Western blot analysis. Also, the human and mouse HIF-1a proteins were recognized by the N-terminal rainbow trout anti-HIF-1a antibody but not by the C-terminal HIF-1a antibody. The accumulation of HIF-1a was studied by incubating rainbow trout and chinook salmon cells at different oxygen concentrations from 20 to 0.2% O2 for 1 h. The greatest accumulation of HIF-1a protein occurred at 5% O2 (38 torr), a typical oxygen tension of venous blood in normoxic animals. The protein stability experiments in the absence or presence of a proteasome inhibitor, MG-132, demonstrated that the inhibitor is able to stabilize the protein, which normally is degraded via the proteasome pathway both in normoxia and hypoxia. Notably, the hypoxia response element of oxygen-dependent degradation domain is identical in mammalian, Xenopus, and rainbow trout HIF-1a proteins, suggesting a high degree of evolutionary conservation in degradation of HIF-1a protein.

The oxygen content, especially in freshwater environments, varies markedly daily, seasonally, and spatially. Due to the low oxygen capacitance of water, respiration of organisms and breakdown of organic material can cause large decreases in oxygen tensions especially during the night when oxygen-producing photosynthesis does not occur. It is thus not surprising that the large variations in environmental oxygen levels have played a significant role in the evolution of fishes. Therefore, fish have developed various physiological and biochemical adaptations to enable survival in hypoxic and anoxic environment, including air breathing organs (1), specialized metabolic pathways enabling long term anoxic survival (2), and modifications of the hemoglobin molecules to optimize oxygen transport (3).
Due to the variability of oxygen content in water and the pronounced role that oxygen has played in the evolution of structure and function of fishes, they present a unique opportunity to study the evolution, function, and regulation of oxygendependent genes and their role in the environmental adaptation. Up to the present, this possibility has been poorly utilized. In mammals, more than 40 hypoxically regulated genes have been characterized (4), including those for the glucose transporter, several enzymes of the glycolytic pathway, erythropoietin, transferrin, and the vascular endothelial growth factor. In contrast, although up-regulation of the synthesis of several proteins in hypoxic fish has been described (5), the identity of these hypoxia-inducible proteins is not known.
In mammals, oxygen-dependent gene expression is transcriptionally regulated by the hypoxia-inducible factor-1 (HIF-1), 1 a heterodimer that consists of two subunits initially called HIF-1␣ and HIF-1␤ (6,7). HIF-1␣ is unique for the hypoxia response, whereas HIF-1␤ turned out to be identical to the aryl hydrocarbon nuclear translocator (ARNT), which acts as a dimerization partner also for other transcription factors, among them the aryl hydrocarbon receptor (AhR, dioxin receptor (8)). Both HIF-1␣ and ARNT belong to the basic-helix-loophelix (bHLH)-PAS family of proteins. All of these proteins have characteristic N-terminal bHLH and PAS domains. The bHLH domain is required for DNA binding and dimerization, whereas PAS domain is involved in heterodimerization, DNA binding, * This work was supported by the Academy of Finland. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence (s)  transactivation, and probably also in HSP90 ligand binding (9,10). Several other members of bHLH-PAS family of proteins in mammals have recently been cloned that show a high homology to the HIF-1␣ such as HIF-2␣, also termed EPAS (endothelial PAS domain protein (11)), HLF (HIF-1␣-like factor (12)), HRF (HIF-related factor (13)), MOP2 (member of PAS superfamily 2 (14)), and HIF-3␣ (15). Hitherto, there are no reports on a hypoxia-inducible factor in fish. The previously characterized bHLH-PAS family proteins in fish play a role in transcriptional regulation induced by xenobiotics. To date, fish aryl hydrocarbon receptor (AhR) and at least two isoforms of its dimerization partner ARNT have been cloned (see Ref. 16).
Since hypoxia-inducible transcription activators have not been characterized in fish, it is obvious that the mechanisms and conditions by which the up-regulation of protein synthesis occurs in hypoxic fish have not been elucidated. In mammals, the activation of hypoxic gene expression occurs at various levels. Although it appears that the mRNA for HIF-1␣ is constitutively expressed (17,18), the levels of HIF-1␣ protein are markedly higher in hypoxic than in normoxic conditions. Under normoxic conditions the HIF-1␣ protein is rapidly ubiquitinated and degraded by the 26 S proteasome, the half-life being less than 5 min (19,20). However, a shift of cells to hypoxic conditions stabilizes and enables HIF-1␣ protein to translocate from the cytoplasm to the nucleus, where it heterodimerizes with HIF-1␤ (21). In addition, hypoxic conditions enhance the transactivating function of HIF-1␣ (22,23). Although the stabilizing effects of hypoxia on the HIF-1␣ protein have been clearly demonstrated, the exact relationship between oxygen tension and the stability of the protein has not been elucidated. Furthermore, the possible differences in the oxygen thresholds of the hypoxia response between different cell types have not been investigated, and the possible effects on the HIF-1 response of previous acclimation to hypoxic conditions have remained unclarified. These questions are conveniently investigated using fish, which experience large variations in the ambient oxygen levels in their normal life. To enable such investigations, we have in the present study cloned and characterized the first fish HIF-1␣ protein from rainbow trout. Using recombinant fish HIF-1␣ protein, we have generated polyclonal antibodies recognizing fish HIF-1␣ protein. These antibodies have been utilized to investigate the oxygen tension dependence of HIF-1␣ protein levels in several types of fish cells.

EXPERIMENTAL PROCEDURES
Cell Culture and Hypoxia Treatment-Rainbow trout hepatocytes were isolated according to Råbergh et al. (24). The number of cells was calculated using a hemocytometer, and viability was tested with trypan blue solution (Sigma). The cells were diluted to a density of 5 ϫ 10 6 cells/ml. Ten ml of hepatocyte suspension was incubated overnight at 18°C before hypoxia treatments. The human epitheloid carcinoma cell line HeLa was cultured in Dulbecco's modified Eagle's medium (high glucose; Life Technologies, Inc.) supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin. The rain-bow trout gonad (RTG-2 (25)) and chinook salmon embryonic (CHSE-214 (26)) cells were cultured in HEPES-buffered Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, 2 mM L-glutamate, 100 g/ml streptomycin, and 0.075% NaHCO 3 . All fish cells were grown at 18°C under air or air supplemented with 1% CO 2 , whereas human cells were grown at 37°C in a multi-gas incubator under normal gas pressure of 21% O 2 supplemented with 5% CO 2 . In most experiments, fish cells were exposed to hypoxia (1% O 2 , 1% CO 2 , 98% N 2 ) for 4 h at 18°C and HeLa cells at 37°C (1% O 2 , 5% CO 2 , 94% N 2 ). When the effect of oxygen tension on HIF-1␣ stability was studied, the fish cells were exposed to different concentrations of oxygen (20, 10, 5, 2.5, 1, and 0.2%) supplemented with 1% CO 2 for 1 h at 18°C. For studying the degradation of HIF-1␣ in fish cells by the 26 S proteasome, a proteasome inhibitor MG-132 (Peptide Institute, Inc.) was used. Fish cells were first exposed to hypoxia (1% O 2 , 1% CO 2 , 98% N 2 ) for 4 h at 18°C in the absence or presence of 10 M MG-132. For re-oxygenation experiments, a portion of the cells already treated under hypoxia in the presence or absence of MG-132 were transferred to air supplemented with 1% CO 2 and incubated for 30 min.
Cloning of Rainbow Trout HIF-1␣ cDNA-Total RNA was isolated from RTG-2, CHSE-214 cells, and freshly isolated rainbow trout hepatocytes using the RNAzol TM B method (TEL-TEST Inc.) based on the method described by Chomczynski and Sacchi (27). cDNA synthesis was performed with 5 g of total RNA using avian myeloblastosis virus reverse transcriptase (Promega) and oligo(dT) primers (Invitrogen) according to the manufacturer's recommendations. This cDNA was used as a template in reverse transcription-PCR, in which the primers were based on the mouse HIF-1␣ sequence. The primer pair of the ba1 primer within the basic region of HIF-1␣ and the hif375 primer distal to the PAS B region of HIF-1␣ sequence (Table I) produced two 1100-base pair RT-PCR fragments. The nucleotide sequence of these two products (termed rba2 and rba5) differed only by 1%. Both PCR fragments were then cloned into a pGEM-T vector (Promega) according to manufacturer's instructions. The rba5 DNA fragment was used as a probe to screen a juvenile rainbow trout UNI-ZAP TM XR cDNA library (a kind gift of Dr. Thomas Chen, University of Connecticut). To find the correct 5Ј and the 3Ј ends of the cDNAs, we used 5Ј-3Ј rapid amplification of cDNA ends (RACE) according to Frohman et al. (28) using the primers listed in Table I. Briefly, 5 g of total RNA from rainbow trout hepatocytes was reverse-transcribed either with the XSC-dT 17 primer (for 3Ј RACE) or the RAC1 primer (for 5Ј RACE). In 3Ј RACE, the cDNA was then amplified by PCR using the XSC-dT 17 and PAC3 PCR primers. In 5Ј RACE, the RAC1 primer was removed using 30-kDa Centricon concentrators (Millipore Inc.), and the cDNAs were polydedioxyadenylic acidtailed using terminal deoxynucleotidyltransferase. The cDNAs were first amplified with the XSC-dT 17 and RAT5 primer pair followed by a second round of amplification with the XSC and ATF3 primer pair.
Generation of Polyclonal Antibodies against Rainbow Trout HIF-1␣-Two different peptides were constructed to produce antibodies against both the N-terminal and the C-terminal part of HIF-1␣. The rainbow trout HIF-1␣ cDNA, called r33a, cloned into a Bluescript KSϩ vector (Stratagene) was cut with either SacI or HincII. The SacI frag-TABLE I PCR primers used in RT-PCR (ba1 and hif375) and 5Ј-3Ј RACE Primers for RT-PCR were based on mouse HIF-1␣ DNA sequence (GenBank TM accession number x95580). Adaptor primers xsc-dT 17 and XSC have been previously published by Frohman et al. (28). Primers RAC1, PAC3, RAT5, and ATF3 for 5Ј-3Ј RACE were designed based on rainbow trout HIF-1␣ DNA sequence (newly deposited as a result of this work, GenBank TM accession number AF304864).  ment of the rainbow trout HIF-1␣ cDNA spanning the N-terminal amino acids 12-363 was cloned in-frame into the SacI site of the bacterial overexpression vector pET-22b(ϩ) (Novagen). Similarly, the HincII fragment of rainbow trout HIF-1␣ cDNA spanning the C-terminal amino acids 330 -730 was cloned in-frame into the Klenow bluntended NcoI-EcoRI site of the expression vector pET-22b(ϩ). This HincII fragment was also in-frame with the His tag present in the pET-22b(ϩ)vector. The bacterial strain BL21 (DE3) plysS (Novagen) was transformed with the plasmids. Cells were cultured in the presence of 1 mM isopropyl-1-thio-␤-D-galactopyranoside for 3 h and collected by centrifugation. The overexpressed proteins were extracted according to Soncin et al. (29). The N-terminal RT-HIF 12-363 antigen was further purified with preparative SDS-polyacrylamide gel electrophoresis, excised, pressed with a syringe through a hypodermic needle, and lyophilized. Due to the formation of inclusion bodies, the C-terminally Histagged RT-HIF 360 -730 antigen was further purified by first washing with 0.2% Triton X-100 in buffer A (20 mM KCl, 100 mM K 2 HPO 4, pH 7.4). The His 6 -tagged protein was then purified using metal (Ni 2ϩ ) chelate affinity chromatography under denaturating (6 M urea) conditions according to the manufacturer's recommendations (Novagen). The affinity chromatography-purified protein was finally desalted and lyophilized. The lyophilized N-and C-terminal RT-HIF antigens (divided into portions of 0.1 mg) were dissolved in PBS, mixed with an equal volume of complete Freund's adjuvant, and injected subcutaneously into rabbits (immunizations were performed in Cancer Research Center, Russian Academy of Medical Sciences, Moscow, Russia). After 1 month, the immunization was boosted with 0.1 mg of antigen every other week for 3-6 months. The collected serum was tested with Western blot analysis. Sera recognizing rainbow trout HIF-1␣ protein were further purified. The purification was performed using N-hydroxysuccinimide-activated Sepharose TM high performance 1-and 5-ml columns (Amersham Pharmacia Biotech) according to the manufacturer's recommendations. The serum was first negatively purified against total protein extract of the Escherichia coli strain BL21 (DE3) plysS. Further purifications were carried out using C-terminal HIF-1␣ protein. This protein could be used for purification of both antibodies, because the N-and C-terminal HIF-1␣ antigens overlapped by 30 amino acids, thus producing antibodies that partially included the same epitope recognition site.
Preparation of Nuclear Extract and Electrophoretic Mobility Shift Assay (EMSA)-Nuclear extracts of cells treated under normoxic and hypoxic conditions were prepared according to Semenza and Wang (30). Electrophoretic mobility shift assay was performed according to Kvietikova et al. (31). Since no fish sequences were available, the sense and antisense strands for the HIF-1␣ binding sites in the promoter region of the human erythropoietin gene were used (31). For competition experiments, a 100-or 500-fold excess of unlabeled oligonucleotides was added before the addition of the labeled probes. For supershift experiments, 1 l of preimmunization serum or serum containing the polyclonal antibody was added to the completed electrophoretic mobility shift assay reaction mixtures and incubated an additional 30 min on ice before electrophoresis.
Southern and Northern Blot Analysis-Genomic DNA was isolated from rainbow trout liver according to Boyle (32). The DNA was digested with two different restriction enzymes, EcoRI and HindIII. Digested DNA was run on a 0.7% agarose gel, blotted to a Hypond-N (Amersham Pharmacia Biotech) filter, and hybridized with a 32 P-labeled SacI DNA fragment of rainbow trout HIF-1␣ cDNA using standard methods (33).
Northern blot analysis was performed after denaturation of total RNA using glyoxal (33). Denaturated RNA was run on a 1.0% agarose gel, blotted to a Hypond-N filter, and hybridized with a 32 P- labeled   FIG. 1. cDNA cloning of rainbow trout HIF-1␣ HincII DNA fragment of rainbow trout HIF-1␣ cDNA using standard methods (33).

RESULTS AND DISCUSSION
Cloning of Rainbow Trout HIF-1␣ cDNA-RT-PCR was used to generate rainbow trout-specific HIF-1␣ gene fragments. Two isoformic cDNAs (rba2 and rba5) with sizes of 1100 base pairs were identified (Table II). These cDNA fragments had 66 and 67% nucleotide sequence identity to human HIF-1␣, respectively, and were therefore considered to be the first fish HIF-1␣ sequences. The rba5 DNA fragment that had slightly higher sequence homology to human HIF-1␣ cDNA was used as a probe in screening a cDNA library from juvenile rainbow trout. As shown in Fig. 1A, four cDNA clones were obtained that had similar and overlapping cDNA sequences. These cDNA sequences were missing the 5Ј end of the cDNA, including ATG, the start codon. The missing 5Ј end of cDNA sequence was isolated by 5Ј-3Ј RACE. Furthermore, the previously found 3Ј end of the cDNA could be verified. The 3Ј end had an atypical polyadenylation signal (CATAAA) 5Ј to the normal site of polyadenylation. In addition, the 3Ј-untranslated region contained only three AUUUA mRNA instability elements in rainbow trout HIF-1␣, whereas human HIF-1␣ has been shown to contain eight such elements (7).
Total cDNA of rainbow trout (rt) HIF-1␣ was 3605 base pairs long, with an open reading frame of 766 amino acids. Thus, the rainbow trout HIF-1␣ was slightly smaller in size than the corresponding HIF-1␣ in human (826 amino acids), mouse, or rat (810 amino acids). The predicted rtHIF-1␣ amino acid sequence (Table III) had a 61% similarity to the human, mouse, and rat HIF-1␣, about 52% similarity to the human and mouse HIF-2␣, and 46% similarity to the human and mouse HIF-3␣.
Although the overall similarity of the predicted amino acid sequence of rtHIF-1␣ was much closer to human HIF-1␣ than HIF-2␣, the sequence corresponding to the bHLH domain had a similar homology to both human HIF-1␣ and to HIF-2␣ (Fig.  1B). Interestingly, the homology to the human proteins was greater than to the HIF-1␣ from the amphibian Xenopus laevis. However, the bHLH/PAS A/B regions appear to be relatively well conserved, with 70 -90% similarity between the different proteins. Another conserved sequence was the hypoxia re-sponse element of the oxygen-dependent degradation domain, which comprises of amino acids 557-571 in human HIF-1␣ protein (35): it was identical in all the HIF-1␣ proteins examined. In contrast, the transactivation domains (TAD), especially the TAD-N (20,22) sequence, varied considerably between the different proteins; there was less than 40% identity in the TAD-N sequence between the proteins (Figs. 1, B and C).
The cloning of rtHIF-1␣ indicates that the encoding gene is present in rainbow trout genome. Its presence was further ascertained using Southern blot analysis (Fig. 2). The cleavage of rainbow trout genomic DNA with either EcoRI or HindIII restriction enzymes produced a band of 8-kilobase pairs in the rainbow trout genome (Fig. 2).
Rainbow Trout HIF-1␣ mRNA Is Constitutively Expressed-Using Northern blot analysis, the expression pattern of HIF-1␣ was determined in rainbow trout gonad (RTG-2) and chinook salmon embryonic (CHSE-214) cells exposed to hypoxia for 2 and 4 h (Fig. 3). The levels of HIF-1␣ mRNA, normalized to 18 S rRNA levels, showed no change during a 2-h hypoxia treatment and only a slight decrease after a 4-h hypoxia treatment. This is in agreement with the notion that HIF-1 activation during hypoxia is due to posttranslational mechanisms (17,18,36). Increased mRNA levels have also been reported (37)(38)(39), possibly as a result of increased stability of mRNA in hypoxia (4).
The HIF-␣ DNA Binding Complex in Rainbow Trout Is Smaller than in Mammals-Since the hypoxia response elements (HREs) of hypoxia-inducible genes have not been characterized in rainbow trout, we performed electrophoretic mobility shift assay using the HRE of the human erythropoietin gene (Fig. 4A). As a control for HIF-1␣ DNA binding, we used nuclear extracts from hypoxia-treated HeLa cells (Figs. 4, A  and C). In fish RTG-2 cells, the induction of the HIF-1␣ complex in nuclear extracts was detected after 2 h of hypoxia treatment (1% O 2 ), and the amount of HIF-1␣ complex was further increased after a 4-h treatment (Fig. 4A). Interestingly, the HIF-1␣ complex of nuclear extracts isolated from RTG-2 cells migrated faster in a 4% nondenaturing polyacrylamide gel than that isolated from HeLa cells, suggesting a lower molecular weight of the HIF-1 complex in rainbow trout. This is in agreement with data obtained from the HIF-1␣ cDNA cloning, suggesting that HIF-1␣ protein in rainbow trout has smaller molecular weight than the corresponding human HIF-1␣ protein. Both isoforms of HIF-1␤ (ARNT) in rainbow trout are also smaller (70 and 79 kDa (40)) than the corresponding mammalian proteins (91 to 94 kDa (6)). The formation of the HIF-1␣-HRE complex was also observed after 4 h of hypoxia in primary-cultured rainbow trout hepatocytes (Fig. 4B). The specificity of the HIF-1␣ DNA binding was demonstrated for the hepatocytes by adding a 500 M excess of unlabeled eryth-  2. HIF-1␣ gene is present in rainbow trout genomic DNA. A, Southern blot analysis was performed on genomic DNA isolated from rainbow trout using the SacI fragment of rainbow trout cDNA as a probe. The genomic DNA was cleaved using two restriction enzymes, EcoRI and HindIII. The DNA marker indicates the size of genomic DNA fragments in kilo base pairs (kbp).
ropoietin HRE probe before adding the 32 P-end-labeled erythropoietin HRE probe. In the presence of unlabeled HRE, the HIF-1␣ binding to the radiolabeled DNA was inhibited. In the presence of serum containing the rainbow trout HIF-1␣ antibody, the mobility of the HIF-1␣-HRE complex was further slowed down, indicating that the antibody was specific to the C-terminal part of HIF-1␣ bound to the HRE-HIF complex (Fig. 4B).
Characterization of Antibodies Produced against Rainbow Trout HIF-1␣ Protein-Two polyclonal antibodies were generated against the rainbow trout HIF-1␣ protein, one against the N-terminal part and another against the C-terminal part. Initial characterization of these antibodies using crude serum indicated that the antibody produced against the N-terminal part of the recombinant rainbow trout HIF-1␣ protein recognized a hypoxia-inducible protein both in HeLa cells and in rainbow trout hepatocytes (Fig. 5A). In contrast, the antibody produced against the C-terminal part only recognized fish proteins in a hypoxia-dependent manner (Figs. 4B and 5A). This result is expected, since the N-terminal part of the protein shows more homology across the animal groups than the Cterminal part.
To investigate whether the C-terminal rainbow trout anti-HIF-1␣ antibody recognized a HIF-1␣ protein in fish cell lines, two trout cell lines were exposed to hypoxia for 4 h. The antibody recognized a hypoxia-inducible protein of 95 kDa in both rainbow trout gonad (RTG-2) and chinook salmon embryonic (CHSE-214) cells (Fig. 5B) based on calculation of the mobility of HIF-1␣ protein to the standard molecular weights.
Stabilization of HIF-1␣ Protein Occurs under Physiological Oxygen Concentrations-Since the initial screening indicated that the crude serum containing antibodies against both N-and C-terminal parts of the HIF-1␣ recombinant protein reacted with hypoxia-inducible proteins in fish cells, the antisera were further purified in order to study the behavior of HIF-1␣ protein in normoxic and hypoxic conditions (for details, see "Experimental Procedures"). After the purification, the specificity of antibodies was greatly increased, and apart from the hypoxia-specific recognition, only some unspecific bands remained that could be used as markers for equal loading in Western blot analysis after normoxic and hypoxic treatments (Fig. 6).
Consequently, the antibodies were used to investigate how oxygen levels affect the amount of HIF-1␣ protein in rainbow trout gonad (RTG-2) and chinook salmon embryonic (CHSE-214) cells. Since the mRNA levels were not affected by hypoxia (Fig. 3), changes in the HIF-1␣ protein levels would indicate increased stability of the protein. In CHSE cells, the HIF-1␣ protein accumulated already at 10% oxygen (76 torr) (Fig. 6). The accumulation of HIF-1␣ protein was maximal at 5% O 2 (38 torr) in both cell types. Below 5% oxygen (38 torr), the HIF-1␣ protein levels tended to decrease, and there was a pronounced drop in the level of HIF-1␣ at the lowest oxygen level (0.2% ϭ 1.5 torr). Thus, the fish HIF-1␣ appears to accumulate at much higher oxygen levels than mammalian HIF-1␣, which shows a half-maximal response between 1.5 and 2% O 2 and maximal response at 0.5-1% O 2 (41,42). Ambient oxygen levels of 55-60 A, a comparison of HIF-1 complex induction between nuclear extracts isolated from HeLa or rainbow trout gonad (RTG-2) cells after hypoxic (1% O 2 ) treatment of 2 and 4 h. In addition to HIF-1, activating transcription factor-1 (ATF-1), cAMPresponsive element binding-1 (CREB-1), nonspecific band (NS), and free probe are indicated by lines. B and C, to confirm the specificity of HIF-1 binding of rainbow trout to HRE of human erythropoietin gene, competition and supershift experiments were performed using nuclear extracts from primary-cultured rainbow trout hepatocytes (B) or HeLa cells (C) after treatment under normoxia (Ϫ) or hypoxia (ϩ) for 4 h (1). A 500 M excess (rainbow trout hepatocytes) or 100 M excess (HeLa cells) of unlabeled human HRE of erythropoietin gene was used in competition experiment (2). For supershift experiment (3), nuclear extracts of rainbow trout hepatocytes were incubated together with either preimmunization serum (P) or serum containing C-terminal rainbow trout anti-HIF-1␣ antibody (S), whereas nuclear extracts of HeLa cells were incubated together with preimmunization serum or serum containing mouse anti-HIF-1␣ (IgY) antibody (48). A supershifted band is indicated by an asterisk. torr are easily tolerated by the active rainbow trout, resulting in arterial oxygen tension of 35-45 torr and venous oxygen tensions of 15-25 torr (43). In normoxic conditions (140 -150 torr), the arterial and venous oxygen tensions are around 100 and 30 -40 torr, respectively (43). Thus, oxygen levels under which HIF-1␣ protein accumulates during the in vitro incubation are routinely experienced by fish cells in vivo. Consequently, it is possible that oxygen-dependent gene regulation forms an important component of regulation of gene expression in fishes, not only in extreme conditions and during environmental hypoxia but also in more or less normoxic conditions. In mammals, the rapid degradation of HIF-1␣ protein occurs by a ubiquitin-proteasome pathway that is inhibited by hypoxia (19,20,44). To study whether this applies also to fish, we carried out experiments with proteasome inhibitor MG-132. Treatment with MG-132 slowed down the degradation of fish HIF-1␣ protein under normoxia, hypoxia, and during re-oxygenation (Fig. 7). The mechanism of degradation and stabilization of HIF-1␣ protein is therefore most likely the same in man and fish. Interestingly, although the oxygen-dependent degradation domain of HIF-1␣ protein generally shows only 47% similarity between rainbow trout and man, a critical hypoxia response element of oxygen-dependent degradation domain (35) is identical. This element appears to bind von Hippel-Lindau tumor repressor protein (VHL), which functions as a ubiquitin ligase, directing HIF-1␣ protein degradation (45)(46)(47).
In conclusion, our results show that HIF-1␣ is present in fish. The predicted amino acid sequence shows high level of conservation at the bHLH/PAS A/B region, whereas there are large variations in the transactivation domains among vertebrates. The HIF-1␣ levels in fish cells are regulated via the ubiquitin-proteasome pathway as has been shown in mammals, and the protein shows similar oxygen-dependent DNA binding as other known hypoxia-inducible transcription factors. However, the HIF-1␣ of rainbow trout and chinook salmon cells is stabilized at much higher oxygen levels than previously reported for mammals, suggesting a role for oxygen-regulated gene expression in the normal physiology of these fish.
Acknowledgments-We thank Dr. Andrey Mikhailov for technical instructions concerning production and purification of rainbow trout HIF-1␣ antibodies and the Cancer Research Center, Russian Academy of Medical Sciences, Moscow, Russia for immunizations. Dr. Harry Björklund and Tove Johansson are acknowledged for providing the fish cell lines. We also thank Jonna Sonne and Annukka Palomä ki for excellent technical assistance.

FIG. 5. Characterization of antibodies raised against N-or Cterminal part of rainbow trout HIF-1␣ protein.
A, Western blot analysis was performed using nuclear cell extracts from HeLa cells or primary cultured rainbow trout hepatocytes after treatment of cells under normoxia or hypoxia (1% O 2 ) for 4 h. 30 g of nuclear cell extracts was loaded into each well. Purified N-or C-terminal rainbow trout anti-HIF-1␣ antibodies were used as primary antibodies (1:500). B, induction of HIF-1␣ protein in two trout cell lines, rainbow trout gonad (RTG-2) and chinook salmon embryonic (CHSE-214) cells, after hypoxia (ϩ) treatment for 4 h. Unpurified C-terminal rainbow trout anti-HIF-1␣ antibody (1:1000) was used as primary antibody.