Erythropoietin gene from a teleost fish, Fugu rubripes

Erythropoietin (Epo) is a glycoprotein hormone that plays a crucial role in ensuring supply of adequate oxygen to tissues by regulating the production of red blood cells. In mammals, Epo stimulates differentiation and proliferation of erythroid precursor cells in the bone marrow, in response to decreased environmental oxygen concentration or systemic oxygen deficiency caused by anemia.¹  In humans, fetal liver and adult kidney are the primary sites of EPO production.¹  In addition to these tissues, EPO is also produced in the central nervous system, bone marrow, spleen, heart, lung, ovary, testis, and breast cancer cells.^2-7  The production of Epo in the kidney, liver, and the central nervous system is greatly induced by hypoxic conditions.^8,9 

The gene encoding the human EPO was first cloned in 1985 by 2 groups independently.^10,11  Since then, the Epo gene has been cloned from several mammalian species including nonhuman primates, rodents, ruminants, and felines.^12-15  The human EPO gene is transcribed from several start sites in the kidney and liver, and is thus independently regulated in these tissues.¹⁶  Expression of the human EPO gene in transgenic mice has indicated that the cis-elements directing expression to the kidney are located between 6 kb and 14 kb upstream of the basal promoter, whereas the liver-specific elements are located in the 3′ flanking region.¹⁶  A 43-bp cis-element, capable of mediating the hypoxia response in the liver, also resides within the 3′ flanking region, 120 bp downstream of the polyadenylation (polyA) signal.^16-18  Transient transfection studies in human hepatoma cell lines have demonstrated that the 3′ enhancer induces 15- to 50-fold higher expression of a reporter gene in response to hypoxia.

Although an “immunoreactive erythropoietin” that competes with human EPO for reaction with human EPO antibodies has been demonstrated in some teleosts, amphibians, reptiles, and birds,^19,20  the Epo gene has not been cloned from any nonmammalian vertebrates. In teleost fish, the kidney is the primary erythropoietic organ. Since the kidney of teleosts such as the rainbow trout, carp, and eel was found to contain a higher level of the immunoreactive erythropoietin than other tissues, it was concluded that in teleosts, the kidney is the major erythropoietic as well as Epo-producing organ.²⁰  Recently, a “draft” genome sequence of the pufferfish, Fugu rubripes, was completed.²¹  Interestingly, a homology search of the Fugu genome using human protein sequences failed to identify Fugu orthologs for a large number of human cytokines including EPO. It was concluded that these proteins are either absent in fish or have evolved rapidly since the divergence of the mammalian and fish lineages to such an extent that the Fugu proteins are not recognizable by homology search.²¹  In this paper, we report the cloning and characterization of the Epo gene from the Fugu by using a combination of basic local alignment search tool (BLAST) and reverse transcription–polymerase chain reaction (RT-PCR). Our results indicate that the heart is the major Epo-producing organ in the adult Fugu.

The draft Fugu genome sequence²¹  database at http://www.fugu-sg.org (Release 6.1.1; the Second Assembly) was BLAST searched with human and mouse Epo protein sequences using the “tblastn” algorithm. This version of the draft Fugu genome is composed of 12 403 scaffolds (unordered fragments of genomic DNA) that span 332 Megabases (Mb), accounting for about 95% of the nonrepetitive portion of the 365-Mb Fugu genome.

Total RNA was extracted from various Fugu tissues using Trizol reagent (Gibco BRL, Grand Island, NY) by following the manufacturer's protocol. The purified total RNA was reverse transcribed into cDNA using AMV reverse-transcriptase first-strand cDNA synthesis kit (Gibco BRL). The following primers, complementary to putative Fugu Epo exons, were used to clone a fragment of cDNA: FEPOF (5′-CTTGCCTTCCTGTTGATCGTGTTG-3′) and FEPOR (5′-CAAACTCAACTGTTGTTTGGGGAAC-3′). The same primer pair was used to analyze the expression pattern of Epo in various Fugu tissues. The PCR cycles comprised a denaturation step at 95° C for 2 minutes, 35 cycles of 95° C for 30 seconds, 55° C for 1 minute, and 72° C for 1 minute, followed by a final elongation step at 72° C for 5 minutes. Representative RT-PCR products were sequenced to confirm their identity. A fragment of actin cDNA was amplified as an internal control for the quality and quantity of cDNA using the primers ACTF (5′-AACTGGGAYGACATGGAGAA-3′) and ACTR (5′-TTGAAGGTCTCAAACATGAT-3′). The Northern blot analysis was carried out by fractionating Fugu total RNA on a 1.2% agarose gel containing formaldehyde. The RNA was transferred to a nylon membrane and probed with [α-³²P]–labeled Fugu Epo cDNA or actin probe.

The 5′- and 3′-RACE were performed using the SMART RACE cDNA Amplification kit (Clontech, Palo Alto, CA) in a nested PCR. The RACE products were cloned into a T-vector and sequenced completely. The gene-specific primers used for 5′-RACE are FEPOR and FEPOR2 (5′-GAATCTGACAGGGTACATCCCTC-3′), and for 3′-RACE are FEPOF and FEPOF1 (5′-GTACCCTGTCAGAT TCCGTGATTG-3′). The PCR cycles were similar to those used in the RT-PCR except that annealing was done at 60° C for 30 seconds.

A Fugu cosmid, no. c068C11, from Greg Elgar's Fugu cosmid library (HGMP-RC, MRC, Hinxton, United Kingdom), that spans the Epo locus was identified based on the similarity of its end sequences to scaffold no. 7655. The gaps in the scaffold sequence were filled by sequencing cosmid DNA using primers flanking the gaps. DNA sequencing was carried out on an Applied Biosystems 3700 DNA Analyzer (Foster City, CA) using the dye-terminator chemistry. The Fugu Epo sequence has been submitted to GenBank (accession number AY303753).

Protein sequences for the mammalian Epo were retrieved from the National Center for Biotechnology Information (NCBI) database. Alignments of the Fugu and mammalian Epo sequences were generated using the program ClustalX. The hydrophilicity plots of the Epo sequences were generated using the program Protean (DNAStar, Madison, WI) with the default Kyte & Doolittle hydrophilicity parameters.

The upstream intergenic region of Fugu Epo (–5.9 kb) was amplified by PCR using cosmid c068C11 as a template and the following primer pair: 5′-GCATCCGCATTTGCAGTCTTC-3′ and 5′-TCGCGATCAGGTGGCGCATC-3′. The PCR fragment was blunt ended and cloned into the SmaI site of pGL3-basic vector (Promega, Madison, WI). Then, 5′ end deletions were made in this construct by digesting with SalI, EcoRV, and NsiI to generate –3.8 kb, –2.1 kb, and –0.8 kb promoter constructs, respectively. The first intron (3.1 kb, specific to the heart and liver) was amplified by using the primer pair 5′-GAGGTCCGAGATCTCGCGG-3′ and 5′-ACAGGAAGGCAAGCAATCCTG-3′. The PCR product was blunt ended and cloned into the Fugu Epo promoter-luciferase reporter construct at the blunt-ended XhoI site downstream of the promoter sequence. The 3′ intergenic region (1.58 kb, from Epo stop codon to the stop codon of downstream RPP20 gene), designated 3′IGR, was amplified by PCR using the primers 5′-GCGCGGCCGCTGCCGGCTGACATTTCCACTC-3′ and 5′-GCGCGGCCGCCAGTATGAGGCTTCATGTGCTC-3′ containing restriction sites for NotI, and cloned into the NotI site of the Fugu Epo promoter-luciferase reporter construct. All the recombinant Fugu Epo promoter-luciferase reporter constructs were sequenced to confirm the sequence and orientation of the promoter. The 5.9-kb promoter contained 6 single base substitutions compared with the genomic clones, presumably due to the error introduced during PCR.

An 11-kb SalI and XmaI fragment of the cosmid c68C11, containing the Epo gene flanked by 3.8-kb upstream and 0.95-kb downstream sequences, was subcloned into pBluescript and used for transfection.

A 420-bp fragment of the human EPO promoter was amplified from the genomic DNA using the primers 5′-CCCAAGCTTGGAACTCAGCAACCCAGGCATCTCTG-3′ and 5′-CCCAAGCTTGCGGTGGCCCCGGTCCGGCTC-3′, digested with HindIII and cloned into pGL3-basic vector. A 266-bp fragment containing the human EPO enhancer was amplified using the primers 5′-GCGGTACCCTGGGAACCTCCAAATCCCCT-3′ and 5′-GCGGTACCGCGCACTGCAGCCTTGCCC-3′, cloned into the KpnI site upstream of the EPO enhancer to generate the “human EPO promoter-enhancer” construct.

A fish hepatoma cell line (PLHC-1, top minnow hepatoma) and a human hepatoma cell line (HepG2) were obtained from the American Type Culture Collection (Manassas, VA). Both cell lines were grown in Dulbecco modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS), penicillin, and streptomycin in a CO₂ incubator (Forma Scientific, Marietta, OH) with 5% CO₂. PLHC-1 cells were incubated at room temperature, while HepG2 cells were grown at 37° C. DMRIE-C (1,2-dimyristyloxypropyl-N, N-dimethyl-N-hydroxyethylammonium bromide; Gibco BRL) was used as the transfection reagent. DNA construct (2 μg/mL) or Fugu Epo cosmid DNA was mixed with 16 μg/mL DMRIE-C in OptiMEM (Gibco BRL) and transfected into the 80% confluent cell culture, as recommended by the manufacturer. Plasmid cytomegalovirus β-galactosidase (pCMVβ-gal) (0.5 μg/mL) was cotransfected for normalizing the transfection efficiency. Cells were split into desired replicates 6 hours after the transfection and cultured with complete medium for 30 hours, and then grown under normoxic, hypoxic, or anaerobic conditions or in the presence of 100 μM cobalt chloride (Sigma, St Louis, MO), for another 18 hours. Hypoxic conditions were created by filling an air-tight incubator with 1% O₂, 5% CO₂, and 94% N₂. Anaerobic cultures were done in an anaerobic culture jar using GasPak anaerobic system (Becton Dickinson, San Jose, CA), and the depletion of oxygen was monitored by using a disposable anaerobic indicator (Becton Dickinson, Sparks, MD).

Cells were harvested and lysed in passive lysis buffer (Promega). Lysates were assayed for luciferase activity using the Luciferase Assay System kit (Promega) according to the protocol described by the manufacturer. Luciferase activity was measured using a TD-20/20 luminometer (Promega). β-galactosidase activity was determined using the β-galactosidase enzyme assay system (Promega) according to the protocol recommended by the manufacturer and used to normalize levels of luciferase activity in the lysates.

To identify the Fugu Epo gene, we BLAST searched the draft Fugu genome sequence with human and mouse Epo sequences. The BLAST search identified 2 short exons (90 bp and 53 bp) on scaffold no. 7655 (62 kb) that were about 36% identical to the human EPO at the amino acid level. We designed primers complementary to these putative exons and amplified a 193-bp cDNA fragment from the Fugu liver. Comparison of the sequence of this cDNA to genomic DNA identified a 97-bp intron containing the consensus GT and AG splice donor and acceptor sequences. We then obtained the complete sequence of the cDNA by 5′- and 3′-RACE PCR. The full-length Fugu cDNA encodes a protein with 185 amino acids, which is more similar to mammalian Epo sequences than any other protein sequence in the public database.

Since teleosts are known to contain 2 copies of several single-copy mammalian genes due to a whole-genome duplication early during the evolution of teleosts,^22,23  we searched the Fugu genome using the protein sequence of the Fugu Epo cloned by us to determine if there is a second copy. However, our search indicated that Epo is single copy in the Fugu.

We determined the exon-intron structure of the Fugu Epo by mapping the cDNA cloned from the liver to the genomic sequence. The Fugu Epo contains 5 exons and 4 introns similar to the human Epo (Figure 1A). The transcription start site (tss) is located at 237 bp from the first codon (Figure 1A), and a putative TATA box, with an atypical sequence TAATAA, is present at –31 bp. The human Epo also contains an atypical TATA box.¹¹  The 3′ untranslated region of the Fugu Epo does not contain a typical polyadenylation signal (AATAAA), but has an atypical AAGAAA at 17 to 22 bp upstream of the polyA site. Interestingly, the human gene contains a similar atypical signal (AAGAAC) at 11 to 16 bp upstream of the polyA site.¹¹  The Fugu gene spans 5.9 kb from the transcription start site to the polyA signal and is larger than the human Epo, which spans only 2.9 kb. The large size of the Fugu gene is mainly because of its larger first and last introns (Figure 1A).

To determine if the Fugu Epo is transcribed from different sites, we cloned the 5′ end sequence of the Epo transcripts from the heart and brain by 5′-RACE. Interestingly, whereas the 5′ sequence from the heart was identical to that from the liver, the transcript from the brain contained an alternative first exon. This exon codes for only 4 residues (MEFP) and is located 702 bp upstream of the first coding exon of the liver transcripts. The start site of the transcript in the brain is located 60 bp upstream (Figure 1A) and contains a typical TATA box at –49 bp.

We annotated all the genes present in the 62-kb Fugu Epo locus based on their homology to sequences in GenBank. The Fugu locus contains 5 genes besides Epo (Figure 1B). The human orthologs of 2 of these genes, PERQ1 and RPP20, have been mapped to the human EPO locus (University of California, Santa Cruz [UCSC] Human Genome Gateway at http://genome.ucsc.edu) upstream of the EPO gene (Figure 1B). BLAST searching of the human genome sequence with the other Fugu genes from the Epo locus uncovered a new gene at the human EPO locus. This gene has a high similarity to mouse cDNA 2700038N03Rik. It is located downstream of the human EPO gene separated by 5 other genes (Figure 1B). Thus, the synteny of 4 of the genes, PERQ1, 2700038N03Rik, Epo, and RPP20, at the Fugu and human EPO loci is conserved, although their order has been shuffled. This conserved synteny between the human and Fugu Epo loci provides further evidence that the Fugu gene we cloned is indeed the Fugu ortholog of the human EPO.

An alignment of the Fugu and mammalian Epos is shown in Figure S1 (see the Supplemental Figures link at the top of the online article on the Blood website). The Fugu Epo shows an overall identity of only 32% to human EPO and 32% to 34% to other mammalian Epos. However, the hydrophilicity plots of the Fugu and mammalian Epos are very similar (Figure 2). The first 21 residues of the Fugu Epo are highly hydrophobic similar to the signal peptide of the human EPO, indicating that this sequence is the signal peptide of the Fugu Epo (Figure 2). The mature human protein contains 4 cysteine residues that form 2 disulphide bridges, a main bridge linking the amino and carboxyl terminal (Cys7 and Cys161), and a short internal bridge (Cys29 and Cys33).²⁴  The main disulphide bridge is important for the stability and function of the mature protein, whereas the short bridge is not essential for the function of EPO.^14,25  The 4 cysteine residues are conserved in all mammals except rodents, in which 1 of the cysteine residues that forms the internal short bridge is replaced by proline (Figure 2). In the Fugu, however, all 4 cysteine residues are conserved (Figure S1), indicating that the Fugu Epo is folded in the same way as the human EPO.

The human EPO is highly glycosylated; nearly 40% of its molecular weight is due to sugars.²⁵  It carries 3 N-linked sugar chains on Asn24, Asn38, and Asn83, and 1 O-linked sugar chain on Ser126 (highlighted in Figure S1). These N- and O-linked glycosylation sites are conserved in all mammals except rodents, in which the O-linked glycosylation site is absent (Figure S1). Interestingly, the Fugu Epo does not contain the 3 N-glycosylation sites (N-X-S/T), but contains a potential O-linked glycosylation site at the same position as in the human sequence (Figure S1). The carbohydrate moieties of the EPO are known to be important for the synthesis and secretion of the protein, and siacylation of the carbohydrates extends the half-life of EPO. However, siacylation reduces the affinity of EPO to its receptor.²⁶  The presence of only 1 potential glycosylation site in Fugu Epo, in contrast to the 4 glycosylation sites in human EPO, suggests that Fugu Epo may have a higher affinity to the receptor than the human EPO.

Besides the 4 cysteine residues, several other residues are also conserved in the Fugu and mammalian Epos (Figure S1). The conservation of these residues over 450 million years of divergent evolution of fish and mammalian lineages, despite an overall low identity of fish and mammalian Epos, indicates that these residues may be functionally important.

We analyzed the expression pattern of Fugu Epo by RT-PCR and Northern blot analysis. Our results show that in adult Fugu, the heart is the main Epo-producing organ (Figure 3). Besides the heart, Epo production was also detected in the liver and brain. Interestingly, contrary to the previous immunohistochemical studies that suggested the kidney as the main Epo-producing organ in fish,^19,20  we did not detect Epo transcripts in the Fugu kidney. It appears that the previous expression studies done using the human EPO antibody^19,20  were flawed in some way, and the “expression” signals observed were due to some nonspecific binding of the human antibody.

In order to characterize the Fugu Epo promoter, we made a series of promoter-luciferase reporter constructs as well as Fugu genomic fragments and analyzed their activity in a fish and a human hepatoma cell line. The luciferase constructs tested included –5.9 kb, –3.8 kb, –2.1 kb, and –0.8 kb 5′ flanking sequences, designated fEpoP–5.9, fEpoP–3.8, fEpoP–2.1, and fEpoP–0.8, respectively (Figure 4A). There were 2 more constructs, fEpoP–5.9+3′IGR and fEpoP–3.8+3′IGR, generated by inserting 1.5 kb of the 3′ intergenic region (3′IGR) into fEpoP–5.9 and fEpoP–3.8 constructs. Whereas the 0.8-kb and 2.1-kb promoters showed baseline activity (comparable with that of the promoterless pGL3-basic vector), 3.8-kb and 5.9-kb promoters supported 1.5- to 2-fold higher levels of expression than the baseline activity (Figure 4C). Addition of the 3′ intergenic region sequence to 3.8-kb and 5.9-kb promoters increased their expression levels by about 50% and 28%, respectively (Figure 4C). Transient transfection studies in the human HepG2 cell line showed that the expression levels of various Fugu Epo constructs were close to that of the baseline expression level of the promoterless vector (data not shown). These results suggest that the Fugu Epo promoter region between –5.9 kb and –3.8 kb and the 3′ intergenic region contain elements that mediate expression in the fish hepatoma cell line, but not in the human hepatoma cell line.

To determine if the fEpo flanking sequences contain any hypoxia-responsive elements, we analyzed constructs fEpoP–3.8, fEpoP–3.8+3′IGR, fEpoP–5.9, and fEpoP–5.9+3′IGR in the fish hepatoma cell line under hypoxic and anaerobic conditions, or in the presence of CoCl₂. Cobalt is known to mimic hypoxic induction of Epo production in human hepatoma cells by competing with iron for the heme protein.²⁷  Although some of the constructs (fEpoP–3.8, fEpoP–5.9, and fEpoP–5.9+3′IGR) showed 5% to 24% higher levels of expression in the presence of CoCl₂, no marked induction of expression, as reported for the human EPO flanking sequence, was evident under various hypoxic conditions (Figure 4D). Inclusion of the first intron of Fugu Epo into constructs fEpoP–5.9 and fEpoP–3.8 also did not elicit any induction (data not shown). We then repeated these transfection experiments using the human hepatoma cell line. None of the constructs showed any significant induction of reporter gene expression under the hypoxic conditions tested (data not shown).

We then transfected fish and human cell lines with the Fugu cosmid c68C11 DNA or an 11-kb fragment of c68C11, and subjected the cell lines to various hypoxic conditions to see if the Epo gene is hypoxia regulated. The expression of Epo was analyzed by RT-PCR. Our results show that although there was no apparent increase in the overall abundance of transcripts in fish cell lines under hypoxic conditions, the levels of spliced products (193 bp) were found to be higher in cells under hypoxic and anaerobic conditions compared with the control cells, with a concomitant decrease in the unspliced products (Figure 5A). These results suggest that hypoxia may induce higher levels of expression of the Fugu Epo gene in fish cells. In contrast to the fish cell line, there was no difference in the expression levels of the Epo gene in human cells cultured under normal and hypoxic conditions (Figure 5B).

To ascertain if the Fugu Epo transcripts in the fish and human cell lines were generated from appropriate transcription start sites, we mapped the transcription start sites of the transcripts generated from the fEpoP–3.8+3′IGR construct and the Fugu cosmid by 5′-RACE. In both cell lines, the transcription was initiated from the proximal transcription site active in the Fugu liver/heart. A representative chromatogram of the 5′-RACE product is shown in Figure S2. The sequence of the RACE product confirms the transcription start site as well as indicates that the first intron is also correctly spliced in the fish cell line. Further, we confirmed that no transcripts were generated from upstream regions by doing RT-PCR using primers located upstream and downstream of the proximal transcription start site (Figure S2). These results confirm that the Epo transcripts in the fish and human hepatic cell lines were generated appropriately from the proximal transcription start site.

To test whether the fish cell line (PLHC-1) used in this study has the potential to mediate hypoxia response of the human EPO enhancer, we transfected the fish cell line and the human cell line with the human EPO promoter-enhancer construct and cultured them under various hypoxic conditions. Whereas the human EPO construct showed up to a 13-fold increase in expression levels in human cells subjected to hypoxic conditions, no significant increase in the expression levels was observed in the fish cell line (data not shown). Thus, the fish cell line used in this study does not appear to have the potential to mediate the hypoxia response of the human EPO enhancer. This could be due to either the inability of the transcription factors in the fish cells to bind to human enhancer (presumably due to their divergent sequences) or to the lack of some transcription factors in this fish cell line.

In this paper we report the cloning and characterization of the Epo gene from the pufferfish. This is the first nonmammalian Epo gene cloned to date. Several lines of evidence, such as the identical exon-intron structure, conserved synteny, similar primary structure of the protein, and the conserved residues, show that the gene cloned by us is the Fugu ortholog of the human EPO gene. The predicted Fugu Epo sequence exhibits an overall identity of only 32% to the human EPO. The Epo from different mammals also appears to be quite divergent, with the identity of human EPO to Epo from other mammals ranging from 80% to 91%.^14,15  Thus, the coding sequence of Epo appears to be evolving rapidly. Residues that are evolutionarily conserved in such a rapidly evolving gene are likely to be important for the function of the protein. In the present study, we have identified several stretches of residues that are conserved between the Fugu and human Epo. Further characterization of these residues should help to provide new insights into the functions of Epo in vertebrates.

An unexpected finding of our study is that, unlike in mammals in which the kidney is the primary Epo-synthesizing organ in adults,¹  the main Epo-producing organ in adult Fugu is the heart. In primitive vertebrates such as fish, which lack bone marrow, erythropoiesis occurs predominantly in the kidney. It may be that in higher vertebrates, with the invention of the bone marrow as a specialized erythropoietic tissue, shifting of the site of erythropoiesis from kidney to bone marrow was accompanied by a shift in the site of Epo production from the heart to the kidney. The evolutionary pressure(s) that led to such a shift in the site of Epo production is not clear. Cloning and characterization of Epo genes from vertebrates such as amphibians and birds that are intermediary to fish and mammals should help to shed light on this evolutionary shift in the site of Epo production.

In mammals, besides the kidney, Epo is synthesized in the adult liver and brain. Fugu Epo also expresses in the adult liver and brain, indicating that the production of Epo in these organs has been conserved during the evolution of vertebrates, despite a shift in the major Epo-producing organ. Recent studies in mammals show that Epo plays nonerythropoietic functions in the central nervous system (eg, causing the persistence of pluripotent stem cells during neurogenesis of the adult brain and protecting neurons from ischemia-induced apoptosis).^28,29  It is possible that in fish, Epo plays a similar nonerythropoietic function in the brain. Interestingly, the Epo transcripts in the Fugu brain are transcribed from an alternative transcription site and include an alternate first exon. No such brain-specific alternative promoter and transcript have been reported in mammals, although multiple transcription start sites have been identified in the human EPO that are used differentially in the liver, kidney, and brain in response to anemia.³⁰  These alternative transcription start sites are located within a region of 250 bp to 630 bp upstream of the first coding exon, and the transcripts generated from these sites contain the same first coding exon. It remains to be seen if the brain-specific alternative promoter and first exon identified in Fugu is specific to fish, and whether the sequence of the alternate first exon confers a different function to Epo in the brain.

Epo plays an important role in mammals in the supply of oxygen to tissues during exposure to hypoxic conditions such as that encountered at high altitudes and in caves. Hypoxia induces Epo production, which in turn stimulates erythropoiesis. Our transient transfection studies of Fugu Epo promoter constructs in the fish cell line suggest that the Fugu Epo promoter and the 3′ flanking region may not respond to hypoxia. However, when the Fugu Epo cosmid was transfected into the fish cell line, higher levels of spliced transcripts were found in cells subjected to hypoxic and anaerobic conditions. These results suggest that although the Epo promoter is not inducible by hypoxia, the Epo gene may be regulated by hypoxia. However, it should be noted that we were unable to unequivocally demonstrate the hypoxia-response potential of the fish cell line used in this study. Therefore, further in vitro studies using fish cell lines with proven potential for hypoxia response, and in vivo studies using small experimental fish, will be required to confirm if hypoxia induces Epo gene expression and erythropoiesis in fish. Previous studies in gar, a primitive ray-finned fish, showed that erythropoiesis is induced by experimental anemia but not by hypoxia.³¹  This is surprising considering that fish are more frequently exposed to extreme levels of oxygen in the aquatic environment than mammals in the terrestrial environment. The solubility of oxygen in water is just 1/30 of that in the air, and the rate of oxygen diffusion in water is only 1/10 000 of that in air. Thus, oxidation of organic materials and respiration by aquatic organisms can drastically lower the oxygen levels in water, particularly during the night. Therefore, fish would be expected to possess a sensitive and rapid mechanism to regulate the level of erythrocytes. Fish have indeed been shown to maximize oxygen uptake and economize oxygen expenditure under hypoxic conditions by a variety of measures such as increased ventilation, the bradycardia of hypoxia, cardiorespiratory synchrony, constriction of peripheral blood vessels, and increased release of catecholamines from the chromaffin tissue.³²  These measures are probably much more efficient in coping with hypoxic conditions in water than an increased synthesis of erythrocytes, given that the oxygen level in water (∼ 0.0005%) is much lower than that in the atmosphere (21%), and is lower still under hypoxic conditions.

We thank the UK-HGMP Resource Center for providing the Fugu cosmid and the IMCB DNA Sequencing Facility for its help in sequencing. B.V. is an adjunct staff of the Department of Paediatrics, National University of Singapore.