Introduction

Cytokines are intercellular messengers that mediate immune responses. Although often hampered by very poor sequence conservation, a number of fish cytokines have been described (Secombes et al. 2001). Among these are several type I cytokines such as interleukin 12 (IL-12; Yoshiura et al. 2003) and IL-2 (Bird et al. 2005). These recent successes were possible because fish genomic databases became available and identification of genes could be supported by their relative position in the genome (synteny analysis).

Type I cytokines fold into a characteristic “up-up-down-down” bundle of four α-helices, A to D, and share related receptors (Bazan 1990; Rozwarski et al. 1994; Leonard 2003; Huising et al. 2006; Liongue and Ward 2007). They can be further divided into the short-chain and the long-chain subfamilies. The short-chain subfamily members are typically less than 150 amino acids, have α-helices that are less than 20 residues long, and have β-strands positioned between helices A and B and between helices C and D. The short-chain subfamily includes IL-2, IL-3, IL-4, IL-5, IL-7, IL-9, IL-13, IL-15, IL-21, granulocyte-macrophage colony-stimulating factor (GM-CSF), macrophage colony-stimulating factor (M-CSF), stem cell factor (SCF), and thymic stromal lymphopoietin (TSLP). Based on similar receptors, further distinction can be made into the IL-2 subfamily (IL-2, IL-4, IL-7, IL-9, IL-13, IL-15, IL-21, TSLP) and the IL-3 subfamily (IL-3, IL-5, GM-CSF; Liongue and Ward 2007). Except for IL-7, M-CSF, and SCF, the human genes for the short-chain type I molecules are situated on chromosome arms 4q (IL-2, IL-15, IL-21) and 5q (IL-3, IL-4, IL-5, IL-9, IL-13, GM-CSF, TSLP). HuChr4 and huChr5q are paralogues from before teleost fish and tetrapods separated, probably as a result of a whole-genome duplication (WGD) event (Ohno 1970; Pebusque et al. 1998; Morgan et al. 1999, Wraith et al. 2000; Gu et al. 2002; Dehal and Boore 2005; Nakatani et al. 2007). Although the huChr4–huChr5q paralogy is important for the interpretation of the evolution of type I cytokines, to our knowledge, this has not been appreciated in cytokine literature. Fish orthologues of the three short-chain type I cytokine genes on huChr4q have been convincingly identified (Bird et al. 2005; Bei et al. 2006; Wang et al. 2007), whereas that was not done for any of the huChr5q-situated family members. The short-chain type I cytokine genes are very poorly conserved, and previous comparison with huChr5q regions was not extensive enough to pinpoint the expected sites in teleost genomes (Huising et al. 2006).

IL-4 is intensively searched after in current fish immunology science. That is because IL-4 in mammals is typifying and pivotal for Th2 differentiation of stimulated T helper cells (Howard et al. 1982; Murphy and Reiner 2002; Ansel et al. 2006) and because hallmark genes of Th1 responses as IL-12 (Yoshiura et al. 2003) and IFNγ (Zou et al. 2004) have been found in fish already. IL-4 binding to stimulated Th cells silences the IFNγ locus and enhances expression from the IL-4 locus (autocrine loop), whereas IFNγ has the reciprocal effect (Ansel et al. 2006). Functional evidence for Th1 responses in fish is accumulating (Zou et al. 2005; Takizawa et al. 2008), but there are no solid indications for Th2 responses yet.

IL-4 and IL-13 genes are organized in head-to-tail fashion in tetrapod genomes (e.g., Avery et al. 2004) and are closely related structurally and functionally (Minty et al. 1993; Zurawski and de Vries 1994; Chomarat and Banchereau 1998; Moy et al. 2001; Wynn 2003). Both Th2 cytokines uniquely act through binding of a receptor having an IL-4 receptor (IL-4R) α chain, thereby uniquely activating signal transducer and activator of transcription 6 protein. However, besides the IL-4Rα chain, the IL-4R and IL-13R also have distinct components (IL-2Rγc and IL-13Rα1 chain, respectively), and whereas IL-4 can functionally bind either IL-4R or IL-13R, IL-13 function needs IL-13R (Murata et al. 1998; Nelms et al. 1999). T cells express IL-4R but do not express functional amounts of IL-13R, enabling only IL-4 to affect them directly (Zurawski and de Vries 1994; Obiri et al. 1995; de Vries 1998). IL-4 and IL-13 both promote IgE antibody production and inhibit inflammatory cytokine production, but only IL-4 has a pivotal negative effect on IFNγ production, whereas IL-13 is expressed at higher levels and more important for resistance to a number of diseases (Minty et al. 1993; Wynn 2003).

There has been an IL-4 annotation for the fish Tetraodon (Tetraodon nigrovirides; green spotted pufferfish; Li et al. 2007). This IL-4 annotation was based on head-to-tail organization with a RAD50 gene and proximity of a gene of the Shroom family (alias “apical”) as found in chicken and mammals, an intron–exon organization characteristic for short-chain type I cytokines, and an expression pattern agreeing with immune function. Unfortunately, Li and coworkers failed to align the molecules properly with tetrapod IL-4 or to discuss structural and functional motifs, despite such necessity for identification because of the very low homology (<15% amino acid identity between tetrapods and Tetraodon). Li et al. (2007) also failed to compare their detected sequence with IL-13 or other related genes situated in this genomic region of tetrapod species. Therefore, IL-4 identity of the gene was questioned by many researchers, including us. The present article, however, concludes after extensive analysis of the genome history and structural characteristics of the encoded molecules that Li and coworkers found an orthologue of IL-4 and/or IL-13 indeed. In addition, a new IL-4/13 locus was found in a fish-specific paralogous region. The present study constitutes the first detailed analysis of Th2 cytokine loci in fish.

Materials and methods

Database searching and genetic software analysis

Genes were found in the Ensembl genome databases (http://www.ensembl.org/index.html) for chicken (release 2.1), stickleback (first release), Tetraodon (release 7), medaka (first release), and zebrafish (version Zv7), by means of tblastn similarity searches (default settings) using human molecule fragments that were encoded by single (partial) exons. Then, the matching Ensembl genome fragments plus the surrounding region (~200 kb) were analyzed for the presence of genes. Similarly, Fugu genes were found using the Joint Genome Institute Fugu rubripes v4.0 database (http://genome.jgi-psf.org/Takru4/Takru4.info.html). Most genes were only roughly identified by blastx analysis using the National Center for Biotechnology Information database (NCBI; http://www.ncbi.nlm.nih.gov/). However, for the prediction of IL-4/13 molecules and molecules that were compared by phylogenetic analysis (supplementary Fig. 1), a variety of methods was used, including gene prediction software (http://genes.mit.edu/GENSCAN.html) and comparison with expressed sequence tags (ESTs) in the NCBI database as well as with homologous sequences. Sequence alignment for phylogenetic tree analysis was performed by computer software (GENETYX version 7.0.6, Software Development, Tokyo, Japan) and modified by hand, and phylogenetic tree analysis (GENETYX) was performed by the neighbor-joining method based on these alignments. Regions upstream of teleost IL-4/13 open reading frames (ORFs) were screened for promoter elements by TFSEARCH software (http://www.cbrc.jp/research/db/TFSEARCH.html). Only GATA-3 binding motifs recognized by this software were discussed in the present article; although the consensus sequence for GATA-3 binding is (A/T)GATA(A/G), exceptions to this rule were found (Ko and Engel 1993). The prediction of leader peptides of IL-4/13 lineage molecules was performed by SignalP software (http://www.cbs.dtu.dk/services/SignalP/). NCBI Genbank accession numbers used for the analysis of short-helix type I cytokines were the following: human IL-4, M23442; mouse IL-4, NM_021283; chicken IL-4, NM_001007079; human IL-13, AF043334; mouse IL-13, NM_008355; chicken IL-13, AJ621735; human IL-2, U25676; mouse IL-2, NM_008366; chicken IL-2, NC_006091; fugu IL-2, AJ749303; human IL-21, NM_021803; mouse IL-21, NM_021782; chicken IL-21, NC_006091; human IL-5, NM_000879; mouse IL-5, NM_010558; chicken IL-5, NM_001007084; human GM-CSF, NM_000758; mouse GM-CSF, NM_009969; chicken GM-CSF, NM_001007078.

Gene nomenclature

The nomenclature of human genes in Fig. 1 and supplementary Table 1 is in agreement with the GeneCards online database (http://www.genecards.org). Chicken gene nomenclature was based on similarity and synteny with the human genes. Teleost gene nomenclature was based on top matches (lowest expectation value) with human molecules in the NCBI database upon blasting of large (~200 kb) genomic stretches including the entire gene (blastx analysis; matrix BLOSUM62, gap existence cost 11, gap extension cost 1).

Fig. 1
figure 1figure 1

Chromosome maps of genes informative on IL-4/13 region history. a Chromosomes with C1 signatures. b Chromosomes with C2 signatures and without pronounced C1 signatures. Numbers express the chromosome position in megabases and were derived from the databases “Genecards” (human) and “Ensembl” (all others). Similar genes are indicated with identical symbols

Full size image

Experimental fish models

Adult medaka (Oryzias latipes) obtained from a pet shop were maintained at 20°C and fed commercial foods. Three fish were injected intraperitoneally with concanavalin A type IV (Sigma-Aldrich), 10 ng/g body weight. Three hours after stimulation, the kidney was collected, and total ribonucleic acid (RNA) was isolated using “RNAiso” (TAKARA). Total RNA isolated by TRIzol (Invitrogen) from the kidney of adult zebrafish (Danio rerio), 7 days after intraperitoneal injection with 104 PFU of spring viraemia of carp virus strain A-1, was kindly provided by Dr. Sun Jingjing, Sun Yat-Sen University,Guangzhou, China.

cDNA synthesis, PCR, and sequence analysis

Total RNA isolates were reverse-transcribed into complementary deoxyribonucleic acid (cDNA) using oligo-dT RBam24 primer (Table 1) and SuperScript™ III RNase H− Reverse Transcriptase (Invitrogen). Gene-specific polymerase chain reaction (PCR) amplifications were performed in 25-μl reaction mixtures: 0.5 μM of forward and reverse primers, 0.5 μl cDNA template, 2.5 μl 10× buffer, 0.2 mM deoxyribonucleotide triphosphates, 1.5 mM MgCl2, and 0.5 U Platinum Taq DNA polymerase (Invitrogen). Primers are listed in Table 1. The PCR conditions were: 95°C for 2 min, then 35 cycles (30 s at 95°C, 30 s at 60°C, and 30 s at 72°C), followed by 5 min at 72°C. For medaka IL-4/13A2 amplification, initially, 3′ rapid amplification of cDNA ends (RACE) was performed using primer me-IL4/13A2-F1 in conjunction with primer RBam24. Then, the diluted product was used for amplification with primers me-IL4/13A2-F2 plus me-IL4/13A2-R1, and this procedure was repeated using primers me-IL4/13A2-F3 plus me-IL4/13A2-R2. For zebrafish IL-4/13A amplification, a 5′ part was amplified with primers zf-IL4/13A-F1 plus zf-IL4/13A-R1. An overlapping 3′ part was amplified by 3′RACE, using primer zf-IL-4/13A-F2 in conjunction with primer RBam24. For zebrafish IL-4/13B amplification, this gene was amplified by a single 3′RACE reaction using primers zf-IL4/13B-F1 and RBam24. To determine the sequence of PCR products, all PCR products were ligated into pCR®2.TOPO® vector (Invitrogen) and transformed into competent Escherichia coli cells (Invitrogen). Plasmid DNA was purified and sequenced by ABI 3130 Genetic analyzer (Applied Biosystems). GenBank accession numbers of the determined sequences are the following: medaka IL-4/13A2, AB375406; zebrafish IL-4/13A, AB375404; zebrafish IL-4/13B, AB375405.

Table 1 Primer sequences
Full size table

Tertiary structural modeling

The initial prediction of the likelihood of tertiary structures was performed with 3D-PSSM software, which compares sequences against a large database of known protein structures (http://www.sbg.bio.ic.ac.uk/~3dpssm/; Kelley et al. 2000); for zebrafish IL-4/13A, this retrieved human IL-13 as top match. With the tertiary structure of human IL-13 (PDB 1GA3) as the template and an initial sequence alignment as in supplementary Fig. 4, a tertiary structural model of the zebrafish IL-4/13A was constructed by a homology-modeling method using Homology module of Insight II 2000 (Accerlys, Cambridge, UK). The geometry of the models was optimized by Discover 3 module of Insight II 2000 (Accerlys) using a consistent valence force field (Furuhashi et al. 2000). Atomic charges for potentially charged groups were estimated assuming a pH of 7.0. After taking the initial atom coordinate, another restrained molecular dynamics and energy minimization protocol was performed using distance restraints of each pairs of the cystein residues to construct the two predicted disulfide bridges. Then, to search the most stable arrangement of the predicted structure, the system was equilibrated through 5 ps of number/volume/temperature molecular dynamics at 298 K. The program Discover 3 module of Insight II 2000 was used again for energy minimization and the molecular dynamic calculations.

Results and discussion

Initial screening for probable teleost IL-4 sites: comparison of IL-2 and IL-4 genomic regions between human, chicken, and fish

To detect teleost IL-4/13 lineage loci, teleost genomic sequence databases were searched for homologues of human genes neighboring IL-2 and IL-4. The IL-2 regions were included because IL-2, IL-21, IL-4, and IL-13 are related cytokines in paralogous gene clusters (IL-2 family cytokines, FGF, SPRY; Fig. 1), and this paralogy should be helpful to understand the site of teleost IL-4/13 loci. Supplementary Table 1 lists top similarities of selected human Chr4-, Chr5-, and ChrX-encoded molecules with the translated genomic sequences of chicken, stickleback, Tetraodon, medaka, and zebrafish (Ensembl database; tblastn analysis). Figure 1 shows the chromosomal organization of these genes, with the designations of teleost genes based on top similarities with human sequences (NCBI database; blastx analysis) and colors allowing rapid comparison with the human genome map. Beware that top similarity does not necessarily mean orthology and that Fig. 1 is not an attempt to provide the best possible nomenclature for the fish genes. Orthologous gene blocks in Fig. 1 are numbered according to the chicken genome because of the rather conservative genome arrangement in this species (Hillier et al. 2004; Nakatani et al. 2007). Chromosome location or scaffold information of some gene blocks that were not mapped (yet) to the teleost chromosomes depicted in Fig. 1 (e.g., block 3) can readily be retrieved from supplementary Table 1.

The general findings depicted in Fig. 1 are as follows: (1) Stickleback chromosome IV (stChrIV), Tetraodon chromosomes 1 plus 20 (teChr1, teChr20), medaka chromosome 10 (meChr10), and zebrafish chromosome 14 (zeChr14) have segments that are orthologous with huChr4/huChrX/chChr4 (ch = chicken) as well as segments orthologous with huChr5/chChr13; (2) stChrVII, teChr7, meChr14, and zeChr21/zeChr9 only have the huChr5/chChr13 signature. Although lacking the details of Fig. 1 and to our knowledge without the stickleback analysis, these basic patterns of chromosome orthologies have been reported before (Liu et al. 2002; Jaillon et al. 2004; Woods et al. 2005; Kasahara et al. 2007; Nakatani et al. 2007). Separation of chChr4 gene blocks among huChr4 and huChrX (Rabie et al. 2004) and of zeChr14 gene blocks among teChr1 and teChr20 (Woods et al. 2005) have been reported as well, the separations probably being derived traits. It remains to be determined if the zeChr9 cluster depicted in Fig. 1 truly maps to that chromosome, because in an earlier draft of the zebrafish genome sequence (Ensembl Zv6), it mapped to zeChr21, which much better agrees with the findings in other fishes.

Closely related genes are indicated in Fig. 1 with identical symbols and reveal syntenies between gene blocks 2, 5, and 12, between 3 and 12, between 4 and 10, between 6 and 12, between 7 and 12, and between 8 and 9. These syntenies derived from both intrachromosomal and interchromosomal segmental duplications, many of the latter presumably as part of WGD events (see below). Previous reports on paralogy between huChr4 and huChr5q mentioned, among others, the gene couples MSX1/MSX2, DRD5/DRD1, GABRA4/GABRA6, FGF2/FGF1, SPRY1/SPRY4, and IRF2/IRF1 (Fig. 1; Pebusque et al. 1998; Morgan et al. 1999; Wraith et al. 2000; Itoh and Ornitz 2004).

Teleost fish have two paralogous IL-4/13 regions; probable implications of whole-genome duplications

A disproportional large number of gene duplications in vertebrate genomes were mapped to the period between the separation from cephalochordata (amphioxus) and the split between cartilaginous and bony fish, the incidence peaking around 600 million years ago (MYA; Gu et al. 2002; McLysaght et al. 2002; Robinson-Rechavi et al. 2004; Vandepoele et al 2004; Blomme et al. 2006). Although there remains discussion on the relative contribution of various modes of duplication in that period, most researchers now agree on one and probably two WGDs (Ohno 1970; Gu et al. 2002; McLysaght et al. 2002; Vandepoele et al. 2004; Dehal and Boore 2005; Blomme et al. 2006; Nakatani et al. 2007). Quadruplicated syntenies on human chromosomes 4, 5, 8, and 10 serve as prime example for the two rounds of the WGD theory (Wolfe 2001; Dehal and Boore 2005). Nakatani et al. (2007) designated the ancestral chromosomes of huChr4 and huChr5q as “C1” and “C2,” respectively, and below we will follow that nomenclature. Nakatani et al. (2007) hypothesized that the two rounds of WGD in early vertebrates resulted in the four chromosomes C0 (huChr10/13 regions), C1 (huChr4), C2 (huChr5q), and C3 (huChr7/8 regions) from a single C chromosome. Phylogenetic tree analyses (Pebusque et al. 1998; Morgan et al. 1999) suggest that the C1–C2 duplication occurred after their separation from C0/C3.

It is now commonly accepted that in an ancestor of all teleost fish, ~350 MYA, an additional WGD event took place (fish-specific round 3; FS3R; Amores et al. 1998; Taylor et al. 2003; Jaillon et al. 2004; Vandepoele et al. 2004; Woods et al. 2005; Kasahara et al. 2007; Nakatani et al. 2007). Teleosts form the vast majority of extant fish species. For reasons explained below, from hereon, the combination of blocks 9, 11, and 12 on stChrVII/teChr7/meChr14/zeChr21/9 and stChrIV/teChr1/meChr10/zeChr14 are mentioned as teleost IL-4/13 blocks A and B, respectively (gray regions in Fig. 1). The duplicated MSX2, NR3C1, FGF1, and HSPA4 genes in teleost IL-4/13 blocks A and B (Fig. 1) have been listed as argumentative for FS3R (Liu et al. 2002; Taylor et al. 2003; Kasahara et al. 2007). The reciprocal loss of most duplicated genes in teleost IL-4/13 blocks A and B (RAD50, KIF3A, POU4F3, RBM27, ARHGAP26, SPRY4, RNF14, GNPDA1, AFF4, SHROOM, GDF9, FLJ16793, CCNG1, SFXN1, SH3PXD2B, STK10, and SLU7; Fig. 1) agrees with the reported massive loss of FS3R-derived gene duplicates (Jaillon et al. 2004; Woods et al. 2005; Brunet et al. 2006; Sémon and Wolfe 2007). Kasahara et al. (2007) hypothesized that large parts of the FS3R-duplicated C1 and C2 chromosomes fused together, followed by a number of smaller intrachromosomal rearrangements (Fig. 2b). They based this hypothesis solely on syntenies and top similarities, which do not exclude the possibility that stretches such as IL-4/13 block B appear as C2 only because they were lost from the corresponding tetrapod C1 regions. However, phylogenetic analyses by other authors and us support their theory, the general finding being that the proteins encoded by the paralogous genes in teleost IL-4/13 blocks A and B share more similarity with each other than with their tetrapod orthologues (MSX2: Ekker et al. 1997; KCTD16: Gamse et al. 2005; Septin 8: Cao et al. 2007; NR3C1: Stolte et al. 2006; examples in supplementary Fig. 1). The FS3R-history/C2-identity of teleost IL-4/13 block B is even better argued by unique organization features shared with teleost IL-4/13 block A, such as the relative position and orientation of block 11 and the division of block 9 (Fig. 1). Figure 2a schematically depicts the organization of the IL-4/13 region in a hypothetical C2 chromosome of an ancestor to both teleosts and tetrapods. The Fig. 2a model is based on: (1) simplicity; (2) similar genes, such as for the IL-4, IL-13, IL-5, GM-CSF, and IL-3 cytokines, or for the FLJ16793 (alias “similar to cyclin I”) and CCNG1 (cyclin G1) cyclins, often originate from nearby tandem duplications (Read et al. 2004); (3) the hypothetical ancestral C2 organization shares syntenies with C1 blocks 5 (SEPT11–CCNI–CCNG2) and 7 (GABRA4–GNPDA2–KCTD8); (4) the DRD1 and SFXN1 genes are very close in tetrapod block 9. The comparison between neoteleosts (stickleback, Tetraodon, medaka) and zebrafish (a cyprinid fish) indicates that the insertion of gene block 1 into IL-4/13 block A occurred only after these teleosts separated.

Fig. 2
figure 2

IL-4/13 region history. a Hypothetical ancestral C2 organization. Gene blocks are indicated with a few representative genes. Gene families supportive of the history model are indicated with symbols as in Fig. 1. b C1 and C2 history in teleost fish according to Kasahara et al. (2007) and exemplified by medaka. The drawing is only schematic. The white blocks represent several other ancestral chromosomes

Full size image

Teleost IL-4/13A and IL-4/13B loci: location, intron–exon organization, GATA-3 binding, and mRNA instability motifs

Except for medaka, a single IL-4/13 lineage gene was detected in each teleost block 11. MeChr14 block 11 harbors two similar IL-4/13 genes, IL-4/13A1 and IL-4/13A2, whereas no IL-4/13B gene was found in proximity of KIF3A on meChr10. However, the Ensembl database information on the respective stretch of meChr10 is incomplete (many residues are denoted as “n”) and future improvement of the sequence information may reveal the presence of IL-4/13B. The teleost IL-4/13A genes are located downstream of both RAD50 and POU4F3, in the same orientation as RAD50, and the teleost IL-4/13B genes are located downstream of KIF3A and upstream of ATRX or DRD1, in an opposite orientation of KIF3A; the positioning of teleost IL-4/13 relative to RAD50 and KIF3A agrees with the tetrapod situation (e.g., Avery et al. 2004). The four-exon organization of the teleost IL-4/13 genes is typical for the short-helix type I cytokine family (Fig. 3), with exons 2, 3, and 4 each starting at codon position 1 (e.g., Fig. 4). For zebrafish IL-4/13A, but for none of the other teleost IL-4/13 genes, ESTs could be retrieved from the NCBI database (accessions CO922126, CK702857, EH475380, EH449554, EB991904, CK674382). The zebrafish IL-4/13A and IL-4/13B ORFs were confirmed by cDNA analysis. Because of some difficulty in predicting the last exons of the medaka IL-4/13A genes, also an exon 2–4 part of medaka IL-4/13A2 was analyzed from cDNA. However, most teleost IL-4/13 genes discussed in the present article were only predicted from genomic database sequences, and sometimes the exon borders are discussable (not shown).

Fig. 3
figure 3

Intron–exon organization of IL-4/13 lineage genes and several other short-helix type I cytokine genes. Blocks indicate the coding parts of exons. Numbers indicate length. Introns are not drawn to scale

Full size image
Fig. 4
figure 4

Alignment of Tetraodon and Fugu IL-4/13A genomic sequences. Amino acids of the Tetraodon sequence are listed above first nucleotides of codons. TATA boxes and GATA-3 binding motifs are boxed. Red, open reading frames; green, mRNA destabilization motif; purple, poly-adenylation motif. Arrow, from this end the poly-A tail was attached (Li et al. 2007)

Full size image

The IL-4/13A sequence reported by Li et al. (2007) for T. nigrovirides is substantially different from that in the Ensembl T. nigrovirides database with more than 25-nucleotide substitutions and a six-nucleotide length difference between the ORFs. Because the Ensembl Tetraodon IL-4/13A sequence encodes a highly conserved cysteine, which is absent in the molecule predicted by Li and coworkers, the present article uses the Ensembl Tetraodon sequence information for comparison with other fishes. The cDNA analysis by Li et al. (2007) was used, however, for predicting exon borders of pufferfish IL-4/13A.

The pufferfishes T. nigrovirides and F. rubripes separated 18–30 MYA, with an overall genetic divergence rate about twice as fast as observed between human and mouse (Jaillon et al. 2004; http://www.genoscope.cns.fr/). Tetraodon IL-4/13A and Fugu IL-4/13A are situated in identical gene settings, and so are Tetraodon IL-4/13B and Fugu IL-4/13B (not shown). Comparison between the pufferfish IL-4/13A sequences (Fig. 4) or IL-4/13B sequences (supplementary Fig. 2) indicates that relative to the poor sequence conservation of the ORFs (IL-4/13A = 60% nucleotide identity when excluding gaps; IL-4/13B = 61%), fragments upstream and downstream of the ORFs are well conserved (IL-4/13A upstream stretch depicted in Fig. 4 = 73%; IL-4/13A downstream stretch = 72%; IL-4/13B upstream stretch depicted in supplementary Fig. 2: 74%; IL-4/13B downstream stretch = 77%). Better conservation of regulatory sequences than of coding sequences was also reported for cytokine genes in tetrapods (Avery et al. 2004). Between the pufferfishes and the other discussed teleost species, which separated longer ago, or between teleost IL-4/13A and IL-4/13B, the overall sequence similarities of the noncoding regions are very minor, but most teleost IL-4/13 genes share a TATA box preceded by a GATA-3 binding motif in the presumable promoter region and multiple messenger RNA (mRNA) instability motifs (ATTTA) in the immediate downstream noncoding region (Fig. 4; supplementary Figs. 2 and 3). GATA-3 is a critical enhancing transcription factor for Th2 differentiation (Ansel et al. 2006; Ho and Pai 2007). The importance of the TATA box and GATA-3 binding motif for the teleost IL-4/13A genes is underlined by their conservation at very similar sites (Fig. 4 and supplementary Fig. 3), except in medaka IL-4/13A1. The closest GATA-3 binding motif upstream from the medaka IL-4/13A1 ORF is only found at 0.9 kb, whereas these sites also tend to be further upstream from the IL-4/13B ORFs: 2.4 kb from Fugu IL-4/13B, 0.3 kb from Tetraodon IL-4/13B, 1.2 kb from the stickleback IL-4/13B, and 1.0 kb from zebrafish IL-4/13B (not shown). Overall, the conservation pattern suggests that GATA-3 has a more important direct enhancing function for IL-4/13A than for IL-4/13B. It also suggests that medaka IL-4/13A1 has lost the original mode of IL-4/13A regulation. Whereas all teleost IL-4/13A have perfect TATA boxes, the proximal promoter regions of both Fugu and Tetraodon IL-4/13B lack such motifs (supplementary Fig. 2). This, combined with the several ESTs found for zebrafish IL-4/13A whereas none for zebrafish IL-4/13B (see above), suggests that generally the expression of teleost IL-4/13A genes is higher than that of teleost IL-4/13B.

Mammalian and chicken IL-4 and IL-13 promoter regions also have TATA boxes preceded by a GATA-3 motif, but only for mammalian IL-13 and not mammalian IL-4, this GATA-3 motif was found to be important (Ranganath et al. 1998; Lavenu-Bombled et al. 2002; Avery et al. 2004; Ansel et al. 2006). GATA-3, however, does have an important enhancing effect on mammalian IL-4 by inducing epigenetic chromatin changes by binding elsewhere in the region (for an extensive review on transcription regulation in the extended IL-4 region, see Ansel et al. 2006).

Li et al. (2007) determined the full-length 3′ untranslated region of Tetraodon IL-4/13A and already reported the many mRNA instability motifs (ATTTA; Sachs, 1993), which are also found in Fugu IL-4/13A (Fig. 4) and the other teleost IL-4/13 genes (supplementary Fig. 2 and data not shown). Short half-life induced by such motifs improves regulatory control (Sachs 1993; Umland et al. 1998).

Hopefully, we detected all teleost IL-4/13 lineage genes, but with the short exons, the low similarities, and the low expression levels, short-chain type I cytokine genes are truly hard to find.

Other short-chain type I cytokine genes

Detection of teleost IL-3 family members was anticipated because of the detection of zebrafish IL-3Rβc (common component of the receptors for IL-3, IL-5, and GM-CSF; Liongue and Ward 2007), but none was reported so far. We did not search very intensively for this family and found only a single possible gene situated in zeChr14 block 11 in a reverse orientation to IL-4/13B. This zebrafish gene, which was not determined by cDNA analysis but predicted from genomic sequence by computer software (GenBank accession XP_001332104; no annotation), has an intron–exon organization agreeing with other short-helix type I cytokines and encodes a 161-amino acid protein with a probable leader sequence and some motifs reminiscent of IL-3 family members (not shown). For synteny reasons (Fig. 2), this gene was tentatively designated IL-5 (Fig. 1), but further investigation is needed for proper identification.

Tetraodon IL-15 has been mapped to chromosome 18 (Fang et al. 2006), which agrees with the presence of other block 3 homologues (supplementary Table 1) and the reported C1 identity of this Tetraodon chromosome (Jaillon et al. 2004). A fish-specific copy of the IL-15 lineage, IL-15X, did not map to chromosomes with a major C1 or C2 identity (Fang et al. 2006). We did not search for teleost TSLP or IL-9.

Deduced teleost IL-4/13 molecules share important characteristics with tetrapod IL-4 and IL-13 while distinguished by a unique cysteine pair

Figure 5 shows the deduced amino acid sequences of teleost IL-4/13A and IL-4/13B in alignment with full-length tetrapod IL-4 and IL-13 and fragments of a few other IL-2 family members (IL-2 and IL-21) and a few IL-3 family members (IL-5 and GM-CSF). The five fragments represent the consensus structural framework within the short-chain type I cytokines as distinguished by Rozwarski et al. (1994) based on overlapping segments of superimposed IL-2, IL-4, IL-5, GM-CSF, and M-CSF structures. These distinguished fragments basically consist of (major parts of) the α-helix A (αA), β-strand 1 (β1), αB, αC, and the tightly connected β2 plus αD. We resolved to this alignment method because, even between subfamily members, there is an enormous divergence in sequence and variation in helix lengths and positions of disulfide bridges (e.g., Fig. 5; Rozwarski et al. 1994; Eisenmesser et al. 2001; Moy et al. 2001; Avery et al. 2004), making full-length alignment virtually impossible. Within the framework (Fig. 5 blue and red residues), Rozwarski et al. distinguished residues that form the inner core of the proteins (Fig. 5, red residues). They determined that most of the inner core residues were hydrophobic but that 23% of them had a side chain with hydrogen-bonding potential, which mostly folded out of the core but in some cases was involved in buried hydrogen-bonding interactions. We added the IL-13 and IL-21 sequences to Fig. 5 based on structural alignments by Eisenmesser et al. (2001) and Bondensgaard et al. (2007), respectively. The αB helix of mammalian IL-13 is shorter than found in most type I cytokines (Eisenmesser et al. 2001; Moy et al. 2001), whereas chicken IL-13 αB probably has a more usual length (Fig. 5; Avery et al. 2004). The Fig. 5 sequences between the framework fragments predominantly form loop structures, and we aligned them based on sequence similarities and exon mapping (Fig. 5 small font relates to exon borders); like for other short-helix type I cytokines, the IL-4/13 lineage leader+αA, β1, αB+αC, and β2+αD domains are mostly encoded by exons 1, 2, 3, and 4, respectively (Rozwarski et al. 1994). Cysteines forming disulfide bridges in human IL-4 and IL-13 molecules (Walter et al. 1992; Wlodawer et al. 1992; Powers et al. 1993; Eisenmesser et al. 2001; Moy et al. 2001) or expected in the other depicted IL-4/13 lineage molecules are depicted with identical color shading. Other than these cysteines, positions with residues characteristic for the IL-4/13 lineage are depicted in Fig. 5 with yellow shading. A highly conserved cysteine pair in the IL-4/13 lineage, only modified in mammalian IL-13, consists of a cysteine in αB and a cysteine in the CD loop, which is encoded by the beginning of exon 4 (gray pair); a typical setting of the cysteine in αB, found in many IL-4/13 lineage molecules, is the preceding by an aromatic residue and/or the following by a basic residue. The yellow-shaded glutamic acid in αA and arginine in helix αC were shown to be the important sites for binding of human IL-4 to the IL-4Rα receptor chain (Wang et al. 1997;Hage et al. 1999), and other tetrapod IL-4/13 lineage molecules also have acidic and basic residues at these positions, respectively (Fig. 5; Eisenmesser et al. 2001; Moy et al. 2001; Avery et al. 2004). The huIL-4 αA glutamic acid forms part of a hydrogen network including IL-4Rα tyrosine and serine residues (Hage et al. 1999), thus exchange for other groups with hydrogen-bonding potential as observed in most teleost IL-4/13 molecules does not necessarily interfere with the binding mode; however, the hydrophobic residues at this position in Tetraodon and Fugu IL-4/13A do not agree. HuIL-4 αC arginine (Fig. 5, yellow) forms an ion pair with an aspartic acid on IL-4Rα (Hage et al. 1999), and the positive charges of the teleost IL-4/13 residues at this position agree with a similar binding mode. According to Fig. 5, the ancestral setting of this αC arginine probably was (L/I)XRXLXX(Y/F).

Fig. 5
figure 5

Alignment of IL-4/13 lineage molecules with structural framework fragments of several other short-chain type I cytokines. α-Helix and β-strand regions determined for human IL-4 (Powers et al. 1993) and human IL-13 (Eisenmesser et al. 2001) are underlined; the border between human IL-4 β2 and αD is between threonine and leucine. Green font, (predicted) leader sequence; other colors are explained in the main text

Full size image

Two cysteine pairs collectively distinguish tetrapod IL-4 and IL-13 from the teleost IL-4/13 molecules. An ABloop–BCloop pair is only found in the tetrapod molecules (Fig. 5, orange), whereas an ABloop–αD pair (Fig. 5, green) is only found in the teleost molecules. The simplest explanation based on genomic positions (paragraphs above) and positioning of these disulfide pairs is that a single ancestral IL-4/13 gene duplicated independently in the tetrapod and teleost lines into IL-4 and IL-13 versus IL-4/13A and IL-4/13B. However, the sequence motifs are probably too few to be conclusive on this mode of evolution.

Homology comparison of Zebrafish IL-4/13B with the NCBI protein database (blastp analysis) retrieved various tetrapod IL-4 sequences as top matches (top score 39.7 bits, identity ~27%). In contrast, the other teleost IL-4/13 sequences did not retrieve tetrapod IL-4 or IL-13 among their top matches and showed only minor scores with these proteins (<30 bits). We assume that the somewhat higher similarity of zebrafish IL-4/13B with tetrapod IL-4 is only the result of similar fluctuation around a sequence theme and does not relate to their closer common ancestry within the IL-4/13 lineage.

Tertiary structural modeling of teleost IL-4/13 suggests the presence of four stable α-helices in “up-up-down-down” orientation and two disulfide bridges

When the various teleost IL-4/13 molecules were analyzed for the likelihood of tertiary structures by the 3D-PSSM software, many retrieved IL-4 and/or IL-13 sequences among the best matches (not shown), but only zebrafish IL-4/13 sequences retrieved them as a true top match. The top match of zebrafish IL-4/13B was human IL-4 with the highly significant PSMM score 4.25e−0.7. The top match of zebrafish IL-4/13A was human IL-13, with PSMM score 1.44. To investigate the likelihood of a disulfide bridge in ABloop–αD, which is unique for teleost IL-4/13 (Fig. 5, green), we continued 3D modeling of zebrafish IL-4/13A (since, as a teleost exception, this cysteine pair lacks in zebrafish IL-4/13B). To this end, zebrafish IL-4/13A and human IL-13 were aligned inspired by the 3D-PSSM result, which in addition to the expected helix regions αA, αB, αC, and αD, predicted an additional α-helix in the CD loop (supplementary Fig. 4). The so aligned structure was investigated by Insight II 2000 software, and the initial predicted conformation was built to satisfy the requirement of existence of the extra helix. However, when forming the two predicted disulfide bridges in molecular dynamics simulation, the extra α-helix melted away (thick yellow stretch in Fig. 6), while αA, αB, αC, and αD were stably maintained (Fig. 6). The result shows that these four α-helices are energetically stable and suggests reliability of existence of the disulfide bridges C44–C120 (Fig. 5, green) and C50–C92 (Fig. 5, gray). The 3D-PSSM blast results for the various teleost IL-4/13 sequences and the Insight II 2000 modeling result for zebrafish IL-4/13A collectively support that teleost IL-4/13 molecules have a four α-helical “up-up-down-down” structure typical of short-chain type I cytokines and, in most cases, two disulfide bridges.

Fig. 6
figure 6

3D modeling by Insight II 2000 software of zebrafish IL-4/13A (left) in comparison with human IL-13 (right). Stretches with α-helix potential according to 3D-PSSM software are indicated by thick coloring (as in supplementary Fig. 4); upon formation of the disulfide bridges Cys44-Cys120 and Cys50-Cys92 the α-helices A-D are stable but the thick yellow stretch does not form an α-helix. Above, top view from side with N termini of helices A and B and C termini of helices C and D. Below, side view with ascending helices A and B and descending helices C and D. N-term., N terminus of protein; C-term., C terminus of protein; green and red, sulfur and other atoms in cysteines forming disulfide bridges

Full size image

Conclusion

Teleost short-chain type I cytokine genes were found at genomic loci orthologous to tetrapod IL-4 and IL-13. In contrast to the tandemly organized tetrapod IL-4 and IL-13, most teleost IL-4/13 genes are organized as singletons. However, presumably due to the fish-specific WGD (FS3R WGD), teleost fishes acquired two IL-4/13 lineage loci as well, designated IL-4/13A and IL-4/13B. Peculiar is that the duplicated region including teleost IL-4/13 was translocated to a chromosome with which it shares earlier paralogy, probably from 2R WGD. GATA-3 binding motifs in the promoter regions of teleost IL-4/13 support that these genes encode Th2 cytokines indeed. The positioning of the GATA-3 binding motifs and the levels of expression inferred from the representation in the EST database suggest that the expression of teleost IL-4/13A versus teleost IL-4/13B somewhat resembles that of tetrapod IL-13 versus tetrapod IL-4 (the promoter GATA-3 binding motif is more important for IL-13, and this gene is expressed at higher levels than IL-4). The teleost IL-4/13 molecules share typical IL-4/13 lineage motifs determined from tetrapod IL-4 and IL-13, the lineage signature being most pronounced in zebrafish IL-4/13B. The teleost versus tetrapod IL-4/13 lineage molecules are distinguishable by a characteristic cysteine pair each. Probably, there is insufficient information to conclude if tetrapod IL-4 and IL-13 duplicated before or after fish and tetrapods separated. Future studies of teleost IL-4/13 should investigate if T lymphocytes in fish have a Th2 differentiation potential.