The 14-3-3 proteins in the teleost fish rainbow trout (Oncorhynchus mykiss)

The 14-3-3s are small acidic proteins that are found in all eukaryotic organisms (reviewed in Aitken et al.,1992; Burbelo and Hall,1995; Finnie et al.,1999; Baldin, 2000; Tzivion and Avruch, 2002; van Hemert et al., 2001; Ferl et al., 2002; Masters et al., 2002). The homodimers or heterodimers of 14-3-3 proteins interact with client proteins through the amphipathic groove in the middle region, which alters conformation of the bound ligand. This may affect activity of regulated proteins or their ability to interact with other protein partners, modify their intracellular distribution and/or protect from proteolysis and dephosphorylation (reviewed in Tzivion and Avruch, 2002). The number of known partners of 14-3-3 is already greater than 100(van Hemert et al., 2001). Analyses of plant protein sequences for the conserved binding domains suggested that as much as 20% of Arabidopsis proteins are potential clients of 14-3-3s (Sehnke et al.,2002). Interaction with a wide range of partners accounts for the diversified functions of the 14-3-3 proteins. They are involved in signal transduction, checkpoint and cell cycle control, apoptosis, gene transcription, chromosome remodeling, regulation of metabolism, intracellular protein trafficking and stress response.

Taking into account remarkable structural conservation of the 14-3-3 genes (human and yeast proteins share 70% similarity), the existence of a large number of isoforms may seem surprising. Thirteen 14-3-3 genes were found in the genome of Arabidopsis, and expression of 12 isoforms has been confirmed(Ferl et al., 2002). In mammals, there are seven distinct 14-3-3 proteins. One gene belongs to the epsilon type, which is common to plants and animals; the others are denoted as sigma, gamma, eta, beta, zeta and tau. Are the functions of the 14-3-3 gene products redundant or does each gene have a unique function? Microarray analyses and sequencing of cDNA libraries showed that most human tissues harbor several or all of the 14-3-3 proteins, which do not have any marked tissue specificity [expression profiles are deposited in the GeneCards database(http://bioinfo.weizmann.ac.il/cards/)]. In most in vitro studies, 14-3-3 proteins bound ligands with similar affinities. Although several client proteins showed preferential interaction with some of the 14-3-3 isoforms, the biological relevance of subtle differences in affinities remains unknown (reviewed in Fu et al., 2000). At present,there is not much evidence for the functional divergence of animal 14-3-3 isoforms. Studies of lower vertebrate 14-3-3 proteins are important for understanding of evolution and structural and functional diversification of this gene family. To date only three non-mammalian vertebrate 14-3-3genes have been characterized. Two genes were identified in Xenopus laevis tadpoles (Kousteni et al.,1997; Kumagai et al.,1998) and one 14-3-3 gene was characterized in the teleost fish Fundulus heteroclitus(Kultz et al., 2001). In the present study, we present 10 14-3-3 genes from rainbow trout, which are likely to be duplicates of five ancestral genes. Their structure,evolution and expression are reported.

A subtracted cDNA library was prepared from the brain of stressed rainbow trout. Fingerlings (40-50 g) were stressed by netting (2 min) followed by confinement (20 min). This treatment was repeated over five consecutive days. RNA was extracted with Trizol reagent (Invitrogen, Paisley, UK), and mRNA was purified with Dynabeads kit (Dynal, Oslo, Norway). Direct and reverse suppressive subtracted hybridization was performed as described in Diatchenko et al. (1999). PCR-amplified cDNA was cloned into pcdna3.1 vector (Invitrogen), and a 14-3-3 gene was identified among the cDNA sequences. Additional expressed sequence tags(ESTs) having homology to 14-3-3 proteins were identified from a large number of ESTs derived from a single normalized cDNA library constructed from rainbow trout brain, gill, kidney, spleen, liver and muscle mRNA(Rexroad et al., 2003). These clones included complete reading frames.

EST sequences were analyzed with stand-alone blast(Altschul et al., 1997). Multiple sequence alignments were performed with ClustalW(Thompson et al., 1994). Synonymous and nonsynonymous substitutions in genes were determined with Dnasp(Rozas and Rozas, 1999). Phylogenetic analyses were performed with Phylip(Felsenstein, 1989). After alignment of protein sequences, 1000 bootstrap datasets were generated and Dayhoff's PAM matrix was computed. The consensus trees were constructed with the neighbor-joining method. The rainbow trout, Oncorhynchus mykiss(Omy), and zebrafish, Danio rerio (Dr), 14-3-3 proteins were denoted with respect to the result of phylogenetic study.

Expression patterns of Omy14-3-3 genes in embryos and tissues of rainbow trout were analyzed with RT-PCR. The PCR primers (Table S1 in supplementary material) were designed from cDNA sequences to specifically amplify diverged fragments.

Anti-sense probes were prepared for the 3′-UTRs of Omy14-3-3genes. Plasmids were PCR amplified with gene-specific primers aligning to the ends of protein coding sequences (Table S1 in supplementary material) and an universal M13 primer. In vitro transcription was conducted using T7 RNA polymerase (MBI Fermentas, Vilinius, Lithuania) and DIG RNA labeling mix(Roche Bioscience, Indianapolis, IN, USA). Samples were collected at the 40-somite stage and immediately after completion of somitogenesis. Embryos were processed as described in Joly et al.(1993). The probes were detected with AP-conjugated antibodies using NBT/BCIP (nitroblue tetrazolium chloride/5-bromo-4-chloro-3-indolyl-phosphate) substrate (Roche Bioscience). Pictures were taken with an Olympus CK40 microscope.

The Omy14-3-3 genes were selected for inclusion onto a glass microarray containing 1300 rainbow trout and Baltic salmon (Salmo salar) genes, and each clone was spotted in six replicates. Multiple gene expression profiling was performed in experiments that addressed genomic response to stress, toxicity, bacterial antigens and adaptation to cold. RNA was extracted with Trizol reagent (Invitrogen), and four fish were pooled in each sample. Labeling with Cy3- and Cy5-dCTP (Amersham Pharmacia, Little Chalfont, UK) was made using SuperScript III reverse transcriptase(Invitrogen) and oligo(dT) primer; cDNA was purified with Microcon YM30(Millipore, Bedford, MA, USA). We used a dye-swap experimental design(Kerr and Churchill, 2001). Each sample was hybridized to two microarrays. For the first slide, test and control cDNA were labeled with, respectively, Cy5 and Cy3. For the second array, dye assignment was reversed. Slides were prehybridized with 1% bovine serum albumin (BSA), fraction V, 5×SSC and 0.1% sodium dodecyl sulfate(SDS) (30 min at 50°C) and washed with 2×SSC (3 min) and 0.2×SSC (3 min). Arrays were hybridized overnight; hybridization cocktail contained 1.3×Denhardt's, 3×SSC, 0.3% SDS, 0.67 μgμl^-1 polyadenylate and 1.4 μg μl^-1 yeast tRNA(Sigma-Aldrich, St Louis, MO, USA). Slides were washed with 0.5×SSC,0.1% SDS (15 min), 0.5×SSC, 0.01% SDS (15 min) and twice with 0.06×SSC (2 min). Scanning was conducted with ScanArray 5000 (Perkin Elmer-Wallac, Turku, Finland), and images were processed with QuantArray (GSI Luminomics, Munich, Germany). The spot measurements were filtered by criteria I-B>3 and(I-B)/(S_I+S_B)>0.6,where I and B are the mean signal and background intensities, respectively, and S_I and S_B are their standard deviations. After subtraction of mean background, lowess normalization was performed. Divergence of expression profiles of 14-3-3 genes was analyzed using results of 31 microarray experiments. For every pair, we estimated the number of genes that significantly (P<0.05) correlated with both (Pearson r). Then, similarity matrices were analyzed with multidimensional scaling. Genes that significantly correlated with all 14-3-3 isoforms were determined to characterize the stress response of the brain. Enrichment of the functional classifications of these transcripts was observed by analysis with Ease (Hosack et al., 2003);significance was determined with exact Fisher's test (P<0.05).

The first Omy14-3-3 gene was found in the subtracted library prepared from stressed rainbow trout brains and was observed to be similar to the mammalian beta protein. To identify other isoforms, we compared the sequences of 47395 rainbow trout ESTs from the normalized rainbow trout library with human 14-3-3 proteins using stand-alone blastx(Altschul et al., 1997). Blastn distributed the clones encoding putative 14-3-3 genes into 10 clusters, which contained identical sequences of different length. Furthermore, we sequenced 10 clones whose 5′-ends included predicted start codons. We analyzed all salmonid fish ESTs deposited in National Center for Biotechnology Information (NCBI) dbEST(http://www.ncbi.nlm.nih.gov/dbEST/)and The Institute for Genomic Research (TIGR; http://www.tigr.org/tdb/tgi/rtgi/)and did not find any new isoforms. Multiple sequence alignment with ClustalW(Thompson et al., 1994; http://www.ebi.ac.uk/clustalw/)divided 10 Omy14-3-3 genes into five pairs, which were designated as A, B (similar to beta), C, E (similar to epsilon) and G (similar to gamma). The protein coding sequences were highly similar in every pair (Fig. 1),whereas the 5′- and 3′-UTRs were divergent in all genes except Omy14-3-3A.

For phylogenetic analyses, we used 14-3-3 proteins of teleost fish(Fundulus and Danio) and mammals whose sequences were retrieved from Swiss-Prot(http://us.expasy.org/sprot/);the outroot was from ascidia (Ciona intestinalis). In addition to five Dr14-3-3 proteins from Swiss-Prot, five sequences were available from Ensembl(http://www.ensembl.org/)and four more were found by comparison of UniGene EST clusters(http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=unigene)with human 14-3-3 proteins using blastx (Table S2 in supplementary material). The total number of putative Dr14-3-3 genes was 14; however, two short sequences were not included in the analyses. Phylogenetic study provided additional evidence for duplication of ancestral rainbow trout genes, since 10 Omy14-3-3 proteins were divided into five separate clades(Fig. 2). Ten of 12 Dr14-3-3 genes were duplicates, and three Dr14-3-3G genes could arise as a result of two subsequent duplications. Some cyprinids are,like salmonids, tetraploid, and the possibility of ancient genome duplication in zebrafish is debated. All types of Omy14-3-3 proteins were found in zebrafish. In rainbow trout, there was no ortholog to the Fundulus(F) gene; however, this was assigned to the clade including Dr14-3-3F1 and F2. Three of six types of fish 14-3-3 proteins (E, G and B) were clustered with mammalian epsilon, gamma and beta proteins at high bootstrap values. The A, C and F types were specific for fish, whereas zeta, sigma, eta and tau 14-3-3 proteins were found exclusively in mammals. Interestingly, in four of five pairs of Omy14-3-3 proteins (B, C, G and E), one gene was assigned to a clade containing the second gene and either the mammalian or zebrafish ortholog. The 14-3-3 cladogram was markedly different from the organismic tree and this suggested rapid divergence of duplicated Omy14-3-3 genes. To verify this finding, we compared synonymous (K_s) and nonsynonymous (K_a) divergence in rainbow trout and mammalian (human and mouse) 14-3-3 genes(Table 1). The mean K_s in fish genes was 1.65 times greater than in mammalian 14-3-3 genes, whereas K_a was increased 10.5-fold.

Expression of 14-3-3 genes in tissues of rainbow trout was analyzed by RT-PCR. The PCR primers were designed to distinguish between the duplicated genes; however, we were unable to separate the Omy14-3-3A1and A2 isoforms due to their high sequence similarity. Distribution of 14-3-3 genes was ubiquitous, and transcripts of six genes(Omy14-3-3B1, B2, C1, C2, G1 and G2) were found in 10 of 11 analyzed tissues (Fig. 3). Brain, ovary and testis harbored a complete set of Omy14-3-3 genes. The number of expressed isoforms in other tissues ranked from eight (gill and kidney) to two (skin). Expression of Omy14-3-3 genes was also studied in embryos. Omy14-3-3B1 and B2 were expressed at all analyzed developmental stages (Fig. 4). Three more genes (Omy14-3-3C1, C2 and A)were detectable in blastulas. This could be due to persistence of maternal transcripts, because expression of these genes was interrupted at the subsequent developmental stages. Six of nine analyzed Omy14-3-3 genes were already expressed in the late gastrulas. Expression of Omy14-3-3E2 and C2 began at early somitogenesis (15 somites), whereas G2 was the latest isoform (34 somites).

We analysed differential expression of Omy14-3-3 genes in 31 microarray experiments that dealt with response of rainbow trout to stress,environmental pollutants and bacterial antigens. Isoforms were compared by correlation of expression profiles with other genes presented on the slide and the data were analysed with multi-dimensional scaling. Distance metrics were calculated for every pair of duplicated genes and differences of expression profiles were clearly related to the non-synonymous divergence of duplicates(Fig. 5; Table 1). However, it is possible that difference of expression between the A1 and A2isoforms was underestimated. In contrast to other Omy14-3-3 genes,the 3′-UTR sequences of these mRNAs are highly conserved, and cross-hybridization could affect the results of microarray analyses. We observed tight coordination of expression of 10 isoforms in studies of stress response in the rainbow trout brain. All Omy14-3-3 genes were downregulated after 1 day, which was followed by a subsequent increase of expression levels (Fig. 6). We selected genes with significantly similar profiles (Pearson r,P<0.05) and analyzed over-representation of Gene Ontology functional classes (Ashburner et al.,2000) in this group (Table 2). This result suggested that Omy14-3-3 genes could share regulatory mechanisms with nuclear proteins, chaperones and proteins involved in signal transduction, nucleotide binding and metabolism.

Expression of six Omy14-3-3 genes (A1, A2, E1, E2, G1 and G2) in somitic (40 somites) and postsomitic embryos was analyzed with in situ hybridization. To separate closely related isoforms, we used the PCR-amplified 3′-UTRs as templates for preparation of probes. We were unable to find any marked difference in the expression patterns of Omy14-3-3 genes; therefore Omy14-3-3B1 is shown as an example (Fig. 7). In somitic embryos (Fig. 7A,B),transcripts were found in the neural crest, eyes, yolk syncytium, tail bud and caudal somites. Interestingly, expression of 14-3-3 genes in the tail bud and caudal somites was seen in some of the analyzed embryos. In postsomitic embryos (Fig. 7C,D), transcripts were detected in the neural crest, gill covers and gill arches and in pectoral fins; however, there was no expression in the tail and eyes.

The number of 14-3-3 isoforms in complex organisms, such as mammals and seed plants, may seem striking taking into account the exclusively high degree of sequence conservation of these proteins. Identification of 10 14-3-3 genes in bony fish rainbow trout provided additional evidence for diversity of this multi-gene family in vertebrates. Bony fish and mammals have different sets of 14-3-3 proteins. We were unable to find orthologs to four of seven types of mammalian 14-3-3 proteins among more than 527 000 zebrafish and salmonid ESTs. In this study, we identified five types of rainbow trout 14-3-3 proteins and one more gene that was previously characterized in the teleost fish Fundulus(Kultz et al., 2001). Of six fish protein types, one (epsilon) is common to plants and animals and two more(beta and gamma) are shared by fish and mammals. Probably, these were established before segregation of teleosts and tetrapods. Fish C and A proteins were clustered with mammalian tau and eta; however, low bootstrap values did not allow us to regard these genes as orthologs(Fig. 2). Finally, type F protein was clearly distinct from any of the mammalian genes. One may conclude that the appearance of novel 14-3-3 proteins was parallel to the origin of vertebrate classes. All mammalian 14-3-3 proteins were found in different species, whereas six types of fish 14-3-3 proteins were present in at least two phylogenetically remote groups.

Continuous emergence of novel 14-3-3 isoforms in the vertebrate evolution suggested a strong tendency towards functional diversification of this multi-gene family. This was supported by rapid divergence of rainbow trout genes. All Omy14-3-3 genes seem to be duplicates, and at least four of five pairs most likely appeared due to the duplication of the salmonid genome, which took place about 50 million years ago(Bailey et al., 1978). It is possible that duplication of Omy14-3-3A occurred independently and later, since sequence divergence in this pair was markedly lower than in other genes (Fig. 1; Table 1). In organisms of tetraploid origin, functionally redundant duplicated genes are gradually silenced, becoming pseudogenes in a process referred to as diploidisation. Active genes may evolve in a neutral mode or, alternatively, they are selected for conservation or divergence. In rainbow trout, all duplicated 14-3-3 genes are expressed at relatively high level. Phylogenetic analysis suggested that in four of five pairs of Omy14-3-3 genes one was closer to the mammalian or zebrafish ortholog than to the homologous gene. The only exclusion was the least diverged Omy14-3-3A pair. A similar result was reported by Taylor et al.(2001) for nine of 27 analyzed duplicate zebrafish genes. Furthermore, nonsynonymous divergence in the duplicated rainbow trout 14-3-3 genes was one order of magnitude greater than in human and mouse 14-3-3 genes. Structural conservation of orthologous 14-3-3 genes in the phylogenetically remote mammalian species is likely to be due to their indispensable cellular functions. Duplication of genes can relax these constraints, since one isoform is sufficient for preservation of function whereas the second one can acquire novel features. Activity and rapid divergence of the duplicated rainbow trout 14-3-3 genes strongly suggests that appearance of new distinct isoforms was favored.

Rapid divergence of rainbow trout 14-3-3 proteins suggests functional diversification of the duplicated genes. Microarray analyses in yeast showed differential expression in most pairs with nonsynonymous divergence(Gu et al., 2002). Recent study of the zebrafish annexin gene family found that difference in expression patterns of duplicated genes was closely related to the degree of their sequence divergence (Farber et al.,2003). Analyses of tissue distribution and embryonic expression of Omy14-3-3 revealed only minor differences between isoforms. Thus,expression of Omy14-3-3C1 but not of C2 was found in the intestine whereas the opposite was observed in skin(Fig. 3B). One of two Omy14-3-3B isoforms was active in skin (B1) and in muscle(B2). In embryonic development, Omy14-3-3E1 and G1were activated earlier than Omy14-3-3E2 and G2, respectively(Fig. 4). Furthermore, we compared expression profiles of Omy14-3-3 genes in microarray experiments. Distance between isoforms was determined by numbers of genes that showed similar expression profiles with both genes of each pair and it decreased in the range C>E=B>G>A. Noteworthy,the duplicated genes were ranked in nearly the same order by nonsynonymous divergence (Table 1). This result indicated that structural diversification of the duplicated genes could be due to adaptation to different client proteins.

In most microarray experiments, differential expression was shown by one or several Omy14-3-3 isoforms. We demonstrated close concordance of expression profiles of all 10 Omy14-3-3 genes in the brain of stressed rainbow trout (Fig. 6), and similar responses were shown by a large group of genes,which were significantly over-represented by nuclear proteins and proteins involved in cell communication and signal transduction. In line with this finding, we observed coordinated expression of all Omy14-3-3 isoforms in parts of embryos that are known for rapid growth and differentiation. Stable expression was seen in the neural crest, which is characterized by exclusively high morphogenetic potential. This structure gives rise to connective tissue, some muscle, dermal and pigmented tissues and many structures including the cranium, branchial skeleton and sensory ganglia(reviewed in Baker and Bronner-Fraser,1997; Gorodilov,2000). We did not find tail bud expression after completion of somitogenesis. Noteworthy, only some of the analyzed somitic embryos expressed Omy14-3-3 genes in the tail buds, which could indicate cyclic activity of these genes. At the subsequent stages, expression began to decrease in most tissues and the marked differences were observed between early and differentiated mesenchyme.

This study was supported by the National Agency of Technology, Finland. We wish to thank Seppo Kukkonen for analyses of plasma cortisol.