Fish, the lowest vertebrate group, typically possess a weaker acquired immune response as compared to mammals. Therefore, the innate immune response to infectious invaders is generally agreed to be more vital to fish than mammals (Shi and Camus, 2006;Magnadottir, 2010). Antimicrobial peptides (AMPs) play crucial roles as the first line of defense against invading microbes, and diverse AMPs with different structures and charges are involved in the defense pathway in an interactive or coordinated fashion (Cuesta et al., 2008; Rajanbabu and Chen, 2011).

Hepcidin, a recently discovered AMP, is the first identified from human urine (Park et al., 2001).Similar to a number of naturally occurring AMPs,hepcidins are cysteine-rich and form multiple disulfide bonds in the β-sheet structures (Shi and Camus, 2006). Many previous studies have shown that hepcidin is central in the cross talk between innate immunity and iron homeostasis in vertebrates,although the primary role of mammalian hepcidins is now believed to be hormonal, serving as a negative regulator of iron homeostasis (Vyoral and Petrak, 2005; Hugman, 2006; Atanasiu et al., 2007). The function of mammalian hepcidin is believed to be conserved in fish. Previous studies support this hypothesis, showing that fish hepcidin peptides (both natural and synthetic versions) have potent antimicrobial activities (Lauth et al., 2005; Huang et al., 2007); the transcription of fish hepcidin genes are rapidly activated by experimental challenges using live bacteria, lipopolysaccharides (LPS), poly I:C, and viral suspensions (Shike et al., 2002; Douglas et al., 2003a; Hirono et al., 2005; Chiou et al., 2007); and fish hepcidins are significantly stimulated by iron overload and can regulate mammalian ferroportin, a known iron exporter in enterocytes and macrophages(Huang et al., 2007; De Domenico et al., 2008; Cho et al., 2009).

Unlike most mammals, having only one genomic copy of hepcidin (hamp1), many fish species(especially those belonging to the superorder Acanthopterygii) possess multiple hamp gene copies(at least two, hamp1 and hamp2) in which the fish hamp1 is thought to be an ortholog of the mammalian hamp, while fish hamp2 is believed to play a central role in supporting innate immunity of the fish (Hilton and Lambert, 2008). Moreover, certain species(especially belonging to Perciformes or Pleuronectiformes) have multiple diversified hamp2 copies by way of gene duplication, which is the result of positive Darwinian selection driven by a hostpathogen interaction (Padhi and Verghese, 2007). Based on this finding, one can reasonably assume that fish hepcidins experienced a different evolutionary path from that of mammals. Consequently, this suggests that isoform-specific responses or diversification of hepcidin function should be considered when evaluating fish species. The differential regulation of hamp1 and hamp2 in fish species responding to different biological stimulations remains poorly characterized. Also, hormonal or antimicrobial functions of hepcidin isoforms, as well as tissue distribution patterns, are variable among different fish species (Douglas et al., 2003a; Huang et al., 2007; Cho et al., 2009).

The Javanese ricefish, or medaka (Oryzias javanicus: Beloniformes), is a small, egg-laying fish species, similar to Japanese medaka (Oryzias latipes), a popular model organism for vertebrate genetic and genomic research. Like other Oryzias species, these species possess numerous advantages for use in biological experiments, such as efficient laboratory rearing, transparent embryos, and year-round spawning under controlled conditions (Wittbrodt et al., 2002; Song et al., 2010). Also, because of its ability to efficiently osmoregulate, this euryhaline species would be a useful model system to study the interrelationship between environmental salinities and innate immune function of the brackish fishes.Based on our long-term goal to develop O. javanicus as a model fish system to study the functional genomics of AMP-mediated immunity, the objective of this study was to isolate and characterize two hepcidin paralogs, hamp1 and hamp2, from O.javanicus.

Fish specimens used in this study were laboratorypropagated at the Institute of Marine Living Modified Organisms (IMLMO, PKNU). The conditions for general fish culture have been described previously by Cho et al. (2010) and Song et al. (2010). Fish were fed with Artemia nauplii (INVE Aquaculture Inc.,Salt Lake City, UT, USA) and artificial diets of flounder larvae (150-500 ㎛ in diameter, 50% crude protein; Woosung Feed Corp., Daejeon, Korea) on an ad libitum basis. Water temperature was maintained at 25±1°C, and dissolved oxygen was maintained at 5±1 ppm throughout the experiment. The daily water exchange rate was approximately 20%.

Genomic DNA was isolated from fin tissue using conventional sodium dodecyl sulfate/proteinase K digestion, followed by phenol/chloroform extraction and ethanol precipitation. Total RNAs from various tissues were purified using the RNeasy Mini Kit(Qiagen, Hilden, Germany) according to the manufacturer’s instructions, including the DNase treatment step. The cDNA library was constructed from total RNA purified from whole body O.javanicus juveniles using the Lambda Zap cDNA Synthesis and Packaging Kit (Stratagene, La Jolla,CA, USA) according to the manufacturer’s instructions. Plasmid DNA pools were directly prepared from the Escherichia coli SOLR cells(Stratagene) infected with the excised phagemid stock of the O. javanicus cDNA library to prepare the template for vectorette PCR isolation of the hepcidin cDNA copies.

Based on sequence information from previously characterized hamp1 and hamp2 orthologs obtained from BLAST and/or text searches against the NCBI GenBank (http://ncbi.nlm.nih.gov/) and Ensembl genome databases (http://www.ensembl.org/index.html), two pairs of degenerate primers (OJHAMP1 gFW/gRV and OJHAMP2 gFW/gRV) were designated to PCR-amplify the O. javanicus hamp1 and hamp2 genes, respectively. Information on the primers and thermal cycling conditions used in this study are listed in Table 1. Using two sets of PCR primers, PCR was performed independently to isolate putative hamp1 and hamp2 fragments from O.javanicus genomic DNA. Six randomly chosen PCR clones for the putative O. javanicus hamp1 or hamp2 genes were sequenced in both directions. The raw sequence data were edited manually using the sequence editing software Sequencher (Gene Codes Corp., Ann Arbor, MI, USA). Based on the genomic gene sequences, cDNA sequences for hamp1 and

[Table 1.] Oligonucleotide primers used in this study

hamp2 were isolated using vectorette PCR. Two gene-specific forward primers (OJHAMP1 cFW and OJHAMP2 cFW) were included in the vectorette PCR reactions, respectively, to isolate hamp1 and hamp2. The T7 primer was used as a reverse vector primer, of which the binding site was located at one end of multiple cloning site in the phagemid vector,pBluescript SK(-), which was used to construct the O.javanicus cDNA library. Twelve clones for the putative O. javanicus hamp1 and hamp2 genes were sequenced to identify the cDNA copies corresponding to their genomic counterparts.

Sequence homology of O. javanicus HAMP1 and HAMP2 was compared with the orthologs from representative teleosts using a BLAST search against the NCBI GenBank database (http://ncbi.nlm.nih.gov/Blast.cgi). The putative cleavage sites for the preproprotein and proprotein of the O. javanicus HAMP1 and HAMP2 were predicted using the SignalP 3.0 server (http://www.cbs.dtu.dk/services/SignalP/) and the ProP 1.0 server (http://www.cbs.dtu.dk/services/ProP/), respectively. Theoretical isoelectric point (pI) values for different regions of HAMP1 and HAMP2 were estimated using the ExPASy ProtParam tool (http://www.expasy.org/tools/protparam.html). Multiple sequence alignment of mature or premature hepcidin proteins of O.javanicus HAMP1 and HAMP2, along with vertebrate orthologs, were generated using ClustalW(http://align.genome.jp/). Gene structures (exonintron organization) of the O. javanicus hepcidin isoforms were compared with previously characterized vertebrate hepcidin genes (for species and accession codes, see Figs. 3 and 4).

Basal expression patterns of O. javanicus hamp1 and hamp2 mRNA in adult tissues and semiquantitative reverse transcription (RT)-PCR assays were performed. Total RNA was prepared from the brain, eye, fin, gill, intestine, kidney, liver,muscle, ovary, and testis, as described above. Two micrograms of purified total RNA was reverse transcribed using the Omniscript RT Kit (Qiagen)according to the manufacturer’s instruction. To prepare the normalization control, an 18S rRNA reverse primer (OJ18S RV) was included in the RT reaction, as previously described by Cho et al. (2009).The O. javanicus hamp1 transcripts were amplified by RT-PCR using the O. javanicus hamp1 transcriptspecific primers qOJHAMP1 1F/1R (amplicon=244,247 bp). The O. javanicus hamp2 transcripts were amplified using the qOJHAMP2 1F/1R primers(amplicon=286 bp). The normalization control 18S rRNA was amplified by using the qOJ18S 1F/1R primers (amplicon=253 bp). Amplified product was separated on an agarose gel (1.5%), visualized with ethidium bromide) staining, and assessed by Quantity-One^TM implemented in VersaDoc 4000(Bio-Rad, Hercules, CA, USA). Relative expression levels of hamp1 and hamp2 mRNAs across tissues were assessed based on normalization against specific 18S rRNA levels in each tissue. Assays were performed independently in triplicate.

To examine differential modulation of hamp1 and hamp2 in response to immune challenge, an experimental challenge was performed using the Gram-negative pathogenic bacterium Edwardsiella tarda (FSW910410). Healthy adult O. javanicus individuals (n=12; six females and six males) were immersed in a bacterial suspension (1×10⁶ cells/L)contained in a 5 L water bath for 5 h. The bacterial suspension was prepared by diluting freshly grown E.tarda culture (OD_600nm=1.0) in 5 ppt water by using synthetic sea salt (Kent Marine Inc., Franklin, WI,USA). After the immersion treatment, the fish were washed three times with a large volume (>20 L) of clean brackish (5 ppt) water (5 min per wash) and then immediately transferred to the water recirculating tank (30 L). A non-challenged control group was also treated identically, except for not being exposed to E. tarda. At 24 and 48-h posttransfer to the recirculating tanks, six individuals(three females and three males) were randomly chosen from each tank for gene expression assay.From each individual, the liver and kidney were surgically removed and pooled from the six individuals prior to RNA extraction. Total RNA samples and cDNA templates were prepared as described above. The real-time RT-PCR assay was performed using two gene-specific primer pairs that were used for the tissue distribution assay.Amplification was carried out using the 2× SYBR Supermix (Bio-Rad) and iCycler Real-Time Optic Module (Bio-Rad). The normalization control 18S rRNA was also amplified using two PCR primers, as described above. Based on the standard curves for both hepcidins and the 18S rRNA control, the fold changes of hamp1 and hamp2 during bacterial challenge were estimated based on the comparative Ct method in duplicate (Schmittgen and Livak, 2008).

Differences in the basal expression levels of hepcidin transcripts across tissues were assessed by ANOVA, followed by Duncan’s multiple range test.Transcriptional modulation in response to bacterial challenge was expressed as the fold change of hepcidin transcripts in bacteria-challenged groups compared to non-challenged controls. Differences in expression levels between challenged and non-challenged groups were assessed using Student’s t-test.Statistical analysis was carried out the SPSS software version 10.1.3 (SPSS Inc., Chicago, IL,USA) and the difference was considered to be significant when P<0.05.

Based on the genomic PCR isolation, the O.javanicus hamp1 gene (ojhamp1) had a tripartite structure consisting of three exons interrupted by two introns (Fig. 1) (GenBank accession no. JF419525). The ojhamp1 clones possessed variable repeats of poly(A)+ and poly(C)+ in the 5′-untranslated region(UTR) and intron II, respectively. At the mRNA level,the two were homogenous but encode different amino acids. O. javanicus hamp1 cDNA (ojhamp1-1)matched the predicted sequence of the genomic ojhamp1. The representative ojhamp1-1 cDNA consisted of 270 bp of an open reading frame (ORF) encoding 90 aa. In the 3′-UTR, two putative polyadenylation signals (AATAAA) were found either 18 or 90 bp before the poly(A)+ tail. The ojhamp1-2 cDNA had an identical ORF sequence to ojhamp1-1, except for three additional nucleotides(CAG) encoding glutamine (Gln). In both the 5′- and 3′-UTRs, the sequence homology between ojhamp1-1 and ojhamp1-2 was greater than 95%. The ojhamp1-2 transcripts had some sequence variation in the 3′-UTR, including the mRNA species, with different 3′-UTR lengths processed by one of the two polyadenylation signals. Consequently, the deduced polypeptide sequence of the OJHAMP1 preproprotein consisted of 90 or 91 amino acids, in which two cleavage sites for processing the proprotein and mature peptides were predicted at Ala²⁴/Ile²⁵ and Arg^64-65/Gln^65-66, respectively. The whole preproprotein and proprotein of OJHAMP1 were predicted to have moderate cationic charges (pI=7.53 and 7.73, respectively), while the final form of the mature OJHAMP1 peptide had a higher cationic charge(pI=8.52). The mature OJHAMP1 had eight Cys residues (8/26; 30.8%) at conserved positions.

The PCR-isolated O. javanicus hamp2 gene(ojhamp2) had a conserved structure, typical of vertebrate hamp genes (Fig. 2). The lengths of the three exons (87, 78, and 105 bp for exon I-III,respectively) and two introns (78 and 122-bp for introns I and II, respectively) were not significantly different from those of ojhamp1 (GenBank accession no. JF419526). The exon sequences at the genomic level were similar to the ojhamp2 cDNA counterpart.The ojhamp2 cDNA possessed 132 bp of the 5′-UTR,267 bp of a single ORF encoding 89 aa and 106 bp of the 3′-UTR, excluding the poly(A) + tail. A consensus polyadenylation signal (AATAAA) was found 25 bp

[Fig. 1.] Nucleotide and deduced amino acid sequences of Oryzias javanicus hepcidin isoform hamp1. Coding region is indicated by bold uppercase letters while non-coding region by lower case letters. Stop codon (tga) is indicated by an asterisk. Deduced amino acid sequence in the singlet code is provided below the nucleotide sequence. Amino acid number and theoretical isoelectric point (pI) value for each of three regions (signal peptide proregion and mature peptide) are also indicated. The variable repetitive numbers for adenines (in the 5′-UTR) and cytosines (in the intron II) are observed. Two adenine-guanine sequences in the junction site between intron I and exon II are boxed which is a potential target site for alternative splicing to produce 90 or 91 aa-encoded mRNA species. The Gln missed in the 90 aa-encoded mRNA is noted in the parenthesis. In the 3′-UTR an insertion/deletion of five nucleotides (ggata) was found among different cDNA clones and mRNAs processed with two different polyadenylation signals (aataaa; underlined and bolded) were also detected. The sites observed for the poly(A)+ tailing are indicated by vertical arrows. UTR untranslated region.

before the poly(A)+ tail. The amino acid sequence of the OJHAMP2 polypeptide (the whole preproprotein)was predicted to have a negative overall charge(pI=5.70). The preproprotein OJHAMP2 had two potential cleavage sites: one between Ala²⁴ and Val²⁵(cleavage of the signal peptide), and the other between Arg⁶⁴ and Glu⁶⁵ (conversion of the proprotein to the mature peptide). The pI values for the three regions were 5.75 (for the signal peptide; 24 aa), 5.64 (proregion; 40 aa), and 6.03 (mature peptide; 25 aa). The OJHAMP2 mature peptide also had eight cysteine residues.

The mature peptide sequence of OJHAMP1 was aligned with 40 orthologs compiled from nine orders(Fig. 3). Fish HAMP1s possessed either 25 or 26 aa,displaying a cationic pI range from 7.73 to 8.94 (Fig.3). They shared a relatively high degree of sequence homology, with the eight conserved Cys residues in all the active sequences, except one isoform from Atlantic salmon Salmo salar having only seven Cys. Besides these conserved Cys residues, the teleostean HAMP1s possessed four more conserved amino acid

[Fig. 2.] Nucleotide and deduced amino acid sequences of Oryzias javanicus hepcidin isoform hamp2. Coding region is indicated by bold uppercase letters while non-coding region by lower case letters. Stop codon (tga) is indicated by an asterisk. Deduced amino acid sequence in the singlet code is provided below the nucleotide sequence. Amino acid number and theoretical isoelectric point (pI) value for each of three regions (signal peptide proregion and mature peptide) are also indicated. A putative polyadenylation signal (aataaa) is underlined and bolded while the site for poly(A)+ tailing is indicated by vertical arrow.

residues at positions 3 (His), 20 (Gly), 22 (Gly), and 27 (Phe) (alignment positions including gaps in Fig.3). Moreover, the sequence identities across the species were roughly in agreement with the known taxonomic appraisal of the species, although we did not reconstruct the molecular phylogeny in detail. As shown in the alignment, teleostean HAMP1 represented a well conserved N-terminal motif(QSHL/I). Exceptions were the ISHI signals found in a few species belonging to Pleuronectiformes and the QIHL observed in one of the three isoforms recovered from Atlantic salmon. However, these substitutions did not change the pI values of the first 5 aa essential for hormonal function of HAMP1. All the HAMP1s had identical pI values (6.74) in this region.

Alignment of the major HAMP2 sequences (only HAMP2s having eight clearly conserved Cys residues, aligned in Fig. 4) revealed that teleostean HAMP2s were more variable and heterogeneous than HAMP1s in both the length and charge of the mature peptide(Fig. 4). The HAMP2 sequences recovered from species belonging to the superorder Acanthopterygii and many of species from Perciformes possess multiple HAMP2 copies. The fish HAMP2 mature peptides are predicted to have amino acid lengths ranging from 21 aa (empirically identified from the bass species Morone chrysops in Fig. 4) to 26 aa. Unlike fish HAMP1s, which have a conserved length in the N-terminal amino acids prior to the first conserved cysteine, the corresponding HAMP2 region varied in length (1-6 aa). The pI values of the HAMP2 mature sequences were also variable,ranging from negative (the most anionic value was 4.21, estimated in one isoform from Acanthopagrus schlegelii: Perciformes) to positive (the most cationic value was 8.97, observed in several perciform species). Moreover, the pI values of the first five Nterminal amino acids were highly variable among HAMP2s. The lowest values (3.80-4.00) were

[Fig. 3.] Multiple alignment of Oryzias javanicus mature HAMP1 peptide along with its teleostean orthologs.Seven clearly conserved Cys residues are noted by asterisks on the top while four conserved non-cysteine residues by open circle. One Cys residue conserved in most species except one isoform from Atlantic salmon (S.salar) was indicated by a closed square on the top. Predicted amino acid number (aa) and theoretical isoelectric point (pI) value are provided at the end of each sequence. For the accession code for O. niloticus (TH2-3) refer to Huang et al. (2007).

estimated in the two Oryzias species, whereas several perciform species from Notothenioidei represented an extremely positive charge (10.06) in this region.

Using the RT-PCR assay, ojhamp1 and ojhamp2 mRNAs were evaluated in eight somatic and two gonadic tissues (Fig. 5). Basal expression levels of ojhamp1 transcripts were highest in the liver(P<0.05), although many other tissues had detectable levels of ojhamp1 transcripts. The ovary, testis, and kidney showed the next highest expression level,after the liver. Brain, gill, and intestine showed a moderate level of ojhamp1 mRNA, and eye tissue exhibited very low levels of expression (P<0.05). However, in fin and muscle, ojhamp1 transcripts were not detected. The tissue distribution pattern of ojhamp2 transcript was slightly more ubiquitous than that of ojhamp1. The ojhamp2 transcripts were detected in all the tissues examined. Similar to ojhamp1, the expression of ojhamp2 was highest in the liver (P<0.05). The expression levels of ojhamp2 in the ovary and testis were comparable to that of the liver. Again, expression of ojhamp2 was the lowest in the fin and muscle (P<0.05).

After bacterial challenge with E. tarda, the two hepcidin isoforms had different modulation patterns in the kidney and liver (Fig. 6). Expression of ojhamp1 in the kidney was slightly downregulated, but not significantly at either detection time (24 and 48-h post-challenge; HPC) (P>0.05). However,hepatic ojhamp1 transcripts were upregulated 1.5-fold at 24 HPC (P<0.05) and 2-fold at 48 HPC(P<0.05) compared to the nonexposed control. The ojhamp2 transcripts, however, were significantly downregulated in response to bacterial challenge in

[Fig. 4.] Multiple alignment of Oryzias javanicus mature HAMP2 peptide along with its teleostean orthologs in a quantitative mode. Eight clearly conserved Cys residues are noted by asterisks on the top. Putative RXKR or RRXR cleavage sites are also shown and boxed based on the prediction using the ProP Server. Cleavage sites based on the bioinformatic prediction or experimental isolation (from the hybrid bass M. chrysops) are indicated by vertical arrows. Predicted amino acid number (aa) and theoretical isoelectric point (pI) values for whole predicted mature peptide (pI-w) and the five N-terminal amino acids (pI-n) are provided at the end of each sequence.

both tissues. The ojhamp2 transcripts in the kidney were suppressed 5-fold at 24 HPC, and then partially recovered, as compared to the control level (P<0.05). Hepatic ojhamp2 transcripts were markedly suppressed 20-fold at 48 HPC (P<0.05).

Teleosts, which belong to Acanthopterygii, are thought to possess at least two copies of hamp genes. In this study, we isolated and characterized the genetic determinants for two paralog hamp genes, hamp1 and hamp2, from a beloniform species, O.javanicus (ojhamp1 and ojhamp2). Both ojhamp1 and ojhamp2 had a conserved tripartite genomic structure, similar to all the vertebrate hamp genes characterized so far (Douglas et al., 2003a; Cuesta et al., 2008; Hilton and Lambert, 2008; Wang et al., 2009). However, at the mRNA level, ojhamp1 was highly variable. First, the ojhamp1 mRNA could undergo processing, yielding different 3′-UTR lengths due to the presence of the two functional polyadenylation signals. Second, the ojhamp1 could be transcribed into two different mRNA variants, encoding either 90 or 91 aa in the preproprotein, which may have been caused by alternative splicing at the junction site between intron I and exon II (…tttagCAGGTC…).As a result, the splicing machinery may differentiate between the nucleotide sequence CAG as either the first codon for Gln (i.e., 91 aa protein; ojhamp1-2 type) or an intronic AG acceptor site (i.e., 90 aa-protein; ojhamp1-1 type). However, the biological consequences of these alternatively spliced forms are not yet understood. Alternatively, partially spliced transcripts of the immune gene have been previously reported in carp interleukin (Engelsma et al., 2001) and mouse tumor necrosis factor genes (Reddy et al., 2001) under stimulated conditions. More importantly,

[Fig. 5.] Representative reverse transcription-PCR gels showing the tissue distribution of Oryzias javanicus hamp1 and hamp2 transcripts. Abbreviations for tissues are brain (B) eye (E) fin (F) gill (G) intestine (I) kidney (K) liver (L) muscle (M) ovary(O) and testis (T). A gel for 18S rRNA normalization control is also shown on the top.

a similar alternative splicing pattern has been reported in hepcidin isoforms from the rockbream Oplegnathus fasciatus. Notably, the site (between intron I and exon II) targeted for the alternative splicing in the rockbream hepcidin was the same as that in ojhamp1 (Cho et al., 2009), and furthermore,the amino acid (Gln) neglected during the alternative recognition was the same in the two hepcidin genes.This suggests that many other hamp genes with similar junction sequences could be subjected to alternative splicing, producing variant hamp transcripts. Currently, we have not evaluated the effects of alternative splicing on the structure and function of the hepcidins at either the mRNA or proprotein levels, since the presence or absence of Gln (uncharged and polar) did not alter the pI or the protein configuration, as assessed by in silico prediction (data not shown).

Based on the multiple sequence alignments of the mature hepcidin peptides, fish HAMP1s have highly conserved characteristics in their amino acid lengths, as well as in the pI values. However, in contrast to the HAMP1 group, the teleost HAMP2s are highly variable in these parameters across species and isoforms. Previous studies have claimed that fish hamp1s are orthologs of the human counterpart(hamp), and hence fish HAMP1 should function as a negative regulator of iron homeostasis, similar to mammalian HAMPs. The fish hamp2s, however, have been acquired through a different pathway than that of mammalian hamp genes (Rodrigues et al., 2006; Hilton and Lambert, 2008). This claim is based not only on the chromosomal synteny between mammalian hamp and fish hamp1, but also the observation that most fish HAMP1s (mature form)display an anionic charge in the N-terminal region of

[Fig. 6.] Differential expression of Oryzias javanicus hamp1 (A) and hamp2 (B) in the liver and kidney in response to the bacterial challenge using Edwardsiella tarda. Real-time RT-PCR assay was performed at 24 or 48 h post challenge (HPC) and the gene modulations are defined as fold changes of transcript levels relative to non-challenged control.Mean±SDs with different letters (a-c or w-z) are significantly different based on ANOVA (P<0.05) while the significant up- or down-regulations from the control level are noted by asterisks based on the student’s t-test (P<0.05).

the mature peptide, similar to mammalian HAMPs.The negatively charged N-terminal region (usually 5 aa prior to the first conserved Cys) is known to be essential for its interaction with ferroportin (Nemeth et al., 2006). The QSHLS motif of the zebrafish(Danio rerio) HAMP1 is known to internalize ferroportin molecules, similar to the mammalian motif DTHEP (Fraenkel et al., 2005). The OJHAMP1, described in this study, represents a clearly conserved signature in the N-terminal region, and it shared an identical anionic pI value with all other fish HAMP1 orthologs, suggesting that the dual function of mammalian HAMP1s should be conserved in the broad teleostean taxa. In contrast to HAMP1, the fish HAMP2s were considerably heterogeneous in their structures and charges. Recent molecular phylogeny has argued that, unlike hamp1, which likely exists in every fish species, the fish hamp2 copies are recovered only from Acanthopterygii and could play a primary role in the immune response (e.g.,antimicrobial function) rather than iron regulation(Hilton and Lambert, 2008). This argument was based on the cationic N-terminal charge of fish HAMP2 mature peptides, which was different from fish HAMP1s and mammalian HAMPs. However,this study suggests that this argument should be carefully validated with various fish HAMP2 isoforms. The OJHAMP2 isolated from this study displays a highly anionic charge in the predicted Nterminal region (see also the neighbors of OJHAMP2 in Fig. 4). Furthermore, based on bioinformatic predictions, the overall charge of mature OJHAMP2 is not ‘cationic,’ suggesting that extensive expression analyses may be required to determine whether pIdependent functional diversification exists among fish HAMP2 peptides. Experimental purification of the active peptide, rather than prediction, should be completed to provide direct evidence to support this concept because one example of an isolated HAMP2 from the hybrid bass (see MCH in Fig. 4) was shown to be shorter (only 21 aa) than predicted (26 aa)(Shike et al., 2002; Lauth et al., 2005). Also, human hepcidin proprotein is known to undergo some processing to yield mature HAMP with different lengths (Valore and Ganz, 2008).

Tissue distribution patterns of fish hepcidin transcripts are known to be species-specific, unlike the liver-exclusive or predominant patterns found in most mammalian species. In this study, both ojhamp1 and ojhamp2 transcripts had a wide tissue distribution.Although the liver had the highest level of both isoforms, other tissues (such as the gonads and kidney) had comparable amounts of the hepcidin transcript. This finding supports previous reports in many other species, including channel catfish Ictalurus punctatus, black porgy A. schlegelii, red sea bream Chrysophrys major, turbot Scophthalmus maximus, gilthead seabream Sparus aurata, and redbanded seabream Pagrus auriga (Bao et al., 2005; Chen et al., 2005, 2007; Yang et al., 2007; Cuesta et al., 2008; Martin-Antonio et al., 2009). In contrast,our finding is significantly different from the highly liver-predominant expression observed in winter flounder Pseudopleuronectes americanus, sea bass Dicentrarchus labrax, and orange-spotted grouper Epinephelus coioides (Douglas et al., 2003b; Rodrigues et al., 2006; Yin et al., 2006).

During immersion treatment with E. tarda, the two isoforms, ojhamp1 and ojhamp2, revealed different transcriptional responses to bacterial challenge. The ojhamp1 was not significantly responsive to bacterial challenge in the kidney, yet was moderately upregulated in the liver. Moreover, ojhamp2 was consistently downregulated 20-fold in the liver at 48 HPC during the same bacterial challenge. Such an opposing isoform-dependent regulation in a given tissue (i.e., the liver in this study) has not been reported so far, and the mechanism for this phenomenon requires further characterization. Many postmortem evaluations on the upregulation of hepcidin transcripts in fish during bacterial challenges have been based on direct injection (often intraperitoneal or intramuscular) of bacteria (Chen et al., 2007; Cuesta et al., 2008; Yang et al., 2011), while only a few studies have used the immersion technique (Yang et al., 2007). For this reason, performing experimental challenges with a series of bacterial cell numbers may be valuable in future studies to examine the ‘bacterial burden’-dependent regulation of ojhamp1 and ojhamp2. A previous study on zebrafish indicated that only clinically affected individuals, after bacterial challenge, exhibited a significant elevation of hepcidin transcripts, while healthy fish did not respond (Shike et al., 2004). In addition,redbanded seabream (P. auriga) hepcidin isoforms are significantly downregulated in the spleen by immune challenge using LPS (Martin-Antonio et al., 2009).

In summary, two hepcidin paralogs (ojhamp1 and ojhamp2) were isolated and characterized from the Javanese ricefish O. javanicus, a potential model fish species. Both O. javanicus hamp genes have a conserved tripartite gene structure, typical of vertebrate hepcidin genes. However, ojhamp1 could be transcribed into two mRNA species encoding 90 or 91 aa, potentially through an alternative splicing mechanism, whereas ojhamp2 consistently produced a single mRNA encoding 89 aa. Transcripts of both isoforms were expressed in a wide array of adult tissue types, with the highest expression level in the liver. Upon bacterial challenge using E. tarda, the two isoforms exhibited different patterns of transcriptional responses, including significant downregulation, no apparent alteration, or moderate upregulation.