It is increasingly clear that alterations in Na+-K+-ATPase kinetics to fit the demands in specialized cell types is vital for the enzyme to execute its different physiological roles in diverse tissues. In addition to tissue-dependent expression of isoforms of the conventional subunits, α and β, auxiliary FXYD proteins appear to be essential regulatory components. The present study identified genes belonging to this family in Atlantic salmon by analysis of expressed sequence tags. Based on the conserved domain of these small membrane proteins, eight expressed FXYD isoforms were identified. Phylogenetic analysis suggests that six isoforms are homologues to the previously identified FXYD2, FXYD5, FXYD6, FXYD7, FXYD8, and FXYD9, while two additional isoforms were found (FXYD11 and FXYD12). Using quantitative PCR, tissue-dependent expression of the different isoforms was analyzed in gill, kidney, intestine, heart, muscle, brain, and liver. Two isoforms were expressed in several tissues (FXYD5 and FXYD9), while six isoforms were distributed in a discrete manner. In excitable tissues, two isoforms were highly expressed in brain (FXYD6 and FXYD7) and one in skeletal muscle (FXYD8). In osmoregulatory tissues, one isoform was expressed predominantly in gill (FXYD11), one in kidney (FXYD2), and one equally in kidney and intestine (FXYD12). Expression of several FXYD genes in kidney and gill differed between fresh water and seawater salmon, suggesting significance during osmoregulatory adaptations. In addition to identifying novel FXYD isoforms, these studies are the first to show the tissue dependence in their expression and modulation by salinity in any teleosts.
- Atlantic salmon
- Salmo salar
- quantitative polymerase chain reaction
during the last decade, there has been an increasing focus on the role of FXYD proteins in regulation of cellular ion transport. FXYD proteins constitute a family of small proteins with a single transmembrane segment (for review, see Refs. 17, 26, 28). They are named after the conserved extracellular motif, and all have a high degree of homology in a 35-amino acid stretch adjacent to and in the transmembrane domain, whereas homology outside this area is low (55). In mammals, eight different FXYD isoforms have been identified, and association with the Na+-K+-ATPase has been reported to affect the enzyme affinity for substrate (Na+, K+, and ATP) and/or Vmax (e.g., FXYD1, Ref. 39; FXYD2, Ref. 4; FXYD3, Ref. 19; FXYD4, Ref. 6; FXYD5, Ref. 41; FXYD7, Ref. 7). It is well established that three of four mammalian isoforms of the catalytic α-subunit of Na+-K+-ATPase are expressed in a developmental- and tissue-specific manner, suggesting particular roles during development or coupled to specific physiological functions (10, 40). The FXYD proteins are, in many cases, expressed in a tissue-specific manner (6, 12, 15), and, along the collecting duct of the rat kidney, FXYD isoform expression is differentiated among segments and cell types (41, 48). Specific interactions with FXYD proteins may, therefore, add to the versatility of the Na+-K+-ATPase in various tissues, and they can be regarded as tissue-specific modulators of ion transport. In addition, FXYD1, FXYD2, and FXYD10 also seem to operate as regulatory subunits responsive to exterior signals, since the interaction with Na+-K+-ATPase in these isoforms is altered by PKA and PKC phosphorylation (18, 25, 43).
Little is known about the presence and function of FXYD proteins in nonmammalian vertebrates. In spiny dogfish (Squalus acanthias), a FXYD protein has been characterized and cloned (FXYD10, Ref. 44). In teleosts, three FXYD isoforms were identified among expressed sequence tags (ESTs) from zebrafish by Sweadner and Rael (55). In several teleost species, tissue-dependent expression of the α-subunit isoforms of Na+-K+-ATPase has been observed (20, 30, 52). Recently, it was suggested that, in rainbow trout, a shift in salinity between fresh water (FW) and seawater (SW) is associated with a shift in branchial expression of two of these isoforms (α1a and α1b; Ref. 49). FXYD proteins may well have regulatory influence on both cellular and transepithelial ion transport in teleosts, similar to what is known for mammals. For example, during transitions between FW and SW, expression of FXYD proteins may contribute to the switch from hyper- to hypoosmoregulation by altering the kinetic behavior of Na+-K+-ATPase in branchial chloride cells and in epithelial cells of other osmoregulatory tissues. In addition, due to the sensitivity of some FXYD proteins to kinase signaling, they may also be involved in altered tissue responsiveness to endocrine or paracrine signals.
Despite the importance of FXYD proteins in determining ion transport characteristics in mammalian tissues, studies of these potential regulators in teleosts are, however, largely missing. The propositions of the present study are as follows: 1) there may be more teleost FXYD isoforms than the three identified in zebrafish (55); 2) some isoforms may be expressed in a tissue-specific manner; and 3) salinity change may induce altered expression patterns in osmoregulatory tissues. The anadromous Atlantic salmon were used as the choice model to test these hypotheses.
Fish and sampling.
One-year-old Atlantic salmon (Salmo salar, Ätran stock) were obtained in early March from The Danish Centre for Wild Salmon (Randers, Denmark; http://www.vildlaks.dk/). They were kept at a constant photoperiod (12:12-h light-dark) in recirculated FW in indoor 500-liter tanks on the Odense Campus of the University of Southern Denmark until sampling in January, when the fish weight was 100–150 g. SW-acclimated fish were obtained by transferring a batch of fish to 28 parts per thousand (ppt) artificial SW (Red Sea, Eliat, Israel) 2 wk before sampling. The fish were fed ad libitum three times a week with pelleted trout feed. When sampled, fish were stunned with a blow to the head, and blood was collected with a heparinized syringe from the caudal vessels after which the fish was killed. The first gill arch, posterior trunk kidney, anterior intestine, heart, muscle, brain, and liver were dissected and, when needed, trimmed free of cartilage, fat, and connective tissue and then instantly frozen in liquid N2 until analysis. The experimental protocols were approved by the Danish Animal Experiments Inspectorate, in accordance with the European convention for the protection of vertebrate animals used for experiments and other scientific purposes (no. 86/609/EØF).
Extraction of RNA and cDNA synthesis.
Total RNA was extracted by the TRIzol procedure (Invitrogen, Carlsbad, CA) using maximally 100 mg tissue/ml TRIzol, according to manufacturer's recommendations. Total RNA concentrations were determined by measuring ratio of absorbance at 260 nm to that at 280 nm in duplicate with a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE). Purity (ratio of absorbance at 260 nm to that at 280 nm) ranged from 1.9 to 2.2. One-microgram RNA was treated with 1 unit RQ1 DNase (Promega, Madison, WI) for 40 min at 37°C in a total volume of 10 μl, followed by 10 min at 65°C to inactivate RQ1 DNase. Single-stranded cDNA was synthesized by reverse transcription carried out on 1-μg DNase-treated RNA using 0.5 μg oligo(dT)12–18 primers (Invitrogen) and 200 units Superscript II reverse transcriptase (Invitrogen) for 1 h at 42°C in a reaction volume of 25 μl. At the end, the reaction mixture was heated to 70°C for 10 min and then kept at −20°C until use.
Sequences were obtained from GenBank (9). For most of the work, the bioinformatics tools available at the National Center for Biotechnology Information (www.ncbi.nlm.nih.gov) and ExPASy proteomics server of the Swiss Institute of Bioinformatics (http://www.expasy.org/sprot/) (27) were applied. The main search tool was Basic Local Alignment Search Tool (BLAST) (3), and the conserved regions in FXYD proteins previously found were used to search the salmonid EST database. Multiple alignments were performed with ClustalW server at the European Bioinformatics Institute (http://www.ebi.ac.uk/clustalw) (31), and consensus sequences were generated in Jalview 2.2 (16).
The deduced consensus sequences have been submitted to the Third Party Annotation database (BK006241–BK006254). Nucleotide consensus sequences were translated to protein with the translate resource at the ExPASy server, and sequences were aligned with ClustalW and JalView 2.2. The amino acid sequences of the salmon FXYD proteins were used to identify putative membrane domains by means of the HMMTOP server and TMHMM Server version 2.0 for predicting transmembrane helices and topology of proteins [http://www.enzim.hu/hmmtop/ (58) and http://www.cbs.dtu.dk/services/TMHMM-2.0/ (38, 53), respectively]. The NetPhos 2.0 server neural network was used for predictions of kinase-specific serine, threonine, and tyrosine phosphorylation sites (http://www.cbs.dtu.dk/services/NetPhosK/) (11). Potential N-glycosylation was predicted by the NetNglyc server (http://www.cbs.dtu.dk/services/NetNGlyc/) (35). Potential O-glycosylation was predicted by the NetOglyc server (http://www.cbs.dtu.dk/services/NetOGlyc/) (34). The SignalP 3.0 server was used to predict the presence and location of signal peptide cleavage sites in the amino acid sequences (http://www.cbs.dtu.dk/services/SignalP/) (8).
Protein sequences and hypothetical amino acid sequences deduced from nucleotide sequences of nonteleost FXYD proteins were obtained from GenBank. Mouse (Mus musculus) genes were chosen as a mammalian representative with FXYD1 (EDL23964), FXYD2 (EDL25650), FXYD3 (EDL23956), FXYD4 (EDK99588), FXYD5 (EDL23967), FXYD6 (AAH42579), and FXYD7 (EDL23965). Three chicken (Gallus gallus) representatives were found: a chicken protein similar to mouse FXYD2 (expect value 6e-12; BX931806), a chicken protein weakly similar to mouse FXYD5 (expect value 1e-6; BU252825), and chicken FXYD6 (NP001074348). Amphibian representatives were found in the African clawed frog (Xenopus laevis): frog FXYD1 (NP001091365), frog FXYD2 (NP001081551), frog protein similar to mouse FXYD3 (expect value 9e-19; CX132847), a long frog FXYD5-like protein (BC085206), frog FXYD6 (NP001086408), and frog protein weakly similar to mouse FXYD7 (expect value 7e-8; CF548198). A known elasmobranch FXYD protein from spiny dogfish (Squalus Acanthias) was included (FXYD10, AJ556170). A sea lamprey (Petromyzon marinus) protein weakly similar to mouse FXYD1 (expect value 6e-10; CO551377) was used as the outgroup.
The predicted FXYD genes from Atlantic salmon and zebrafish (Danio rerio), together with those of mouse and other vertebrates, were aligned using ClustalW. A majority rule consensus tree was generated using SEQBOOT, PROML, and CONSENSE, all programs in the PHYLIP package (22). Maximum likelihood (ML) was used for the phylogenetic tree construction. A total of 1,000 bootstraps were used to test the consistency of the grouping within the tree. A parsimonious model was used to further evaluate the main nodes found with the ML method, also with the PHYLIP package (SEQBOOT, PROTPARS, and CONSENSE).
Primers specifically detecting the deduced sequences were designed using Primer3 software (50) and checked using NetPrimer software (Premier Biosoft International). Primer sequences are listed in Table 1. They were in all cases designed to detect multiple transcript variants of the specific isoforms, when such were found. Primers were tested for nonspecific product amplification and primer-dimer formation by analysis of melting curve and agarose gel verification of amplicon size. In addition, the PCR product of the specific primers was separated on a low-melting agarose gel, purified on spin columns (Geneclean Turbo kit; QBIOgene, Irvine, CA), and sequenced (DNA Technology A/S; Aarhus, Denmark). In all cases, the products matched the deduced sequences and confirmed the covered consensus fragments. All primers were synthesized by DNA Technology A/S.
Quantitative real-time PCR analysis using SYBRgreen detection was carried out on a Mx3000p instrument (Stratagene, La Jolla, CA) using standard software settings with adaptive baseline for background detection, moving average, and amplification-based threshold settings with built-in FAM/SYBR filter (excitation wavelength: 492 nm; emission wavelength: 516 nm). Reactions were carried out with 1 μl cDNA (40 ng RNA), 4 pmol forward and reverse primer, and 12.5 μl of 2× Brilliant SYBRgreen master mix (Stratagene) in a total volume of 25 μl. Cycling conditions were 95°C for 30 s, 60°C for 60 s, and 95°C for 60 s in 50 cycles. Melting curve analysis was carried out consistently with 30 s for each 1°C interval from 55 to 95°C. For each primer set, cDNA was diluted 2, 4, 8, and 16 times and analyzed by real-time PCR to establish amplification efficiency.
The amplification efficiency for each primer set was used for calculation of relative copy numbers of individual target genes. For normalization of gene expression, elongation factor (EF)-1α was used in accordance with Olsvik and coworkers (46). No significant differences were observed between groups in their expression of EF-1α normalized to total RNA. Relative copy number of the target genes was calculated as Ea(ΔCt), where Ct is the threshold cycle number, and Ea is the amplification efficiency. Normalized units were obtained by dividing relative copy number of FXYD genes with relative copy number of EF-1α.
Data were analyzed by one-way ANOVA when the assumption of homogeneity of variance was met, as tested by Levene's test. In most cases, examination of the data sets was carried out by Kruskal-Wallis nonparametric one-way ANOVA. Differences between means were assessed by Student's t-test or Mann-Whitney U-test, as appropriate, followed by Bonferroni adjustment for a priori chosen number of comparisons. Significance was set at α = 0.05. All tests were performed using Simstat (version 2.5.5 for Windows, by Provalis Research, Montreal, QC, Canada).
Salmonid FXYD genes.
The conserved regions in FXYD protein sequences found in mammals, zebrafish, and spiny dogfish (40, 51) were used to search the salmonid EST database, which is accessible through GenBank. Related protein coding sequences derived from ESTs were found by their homology with the known amino acid sequences. Corresponding nucleotide sequences of ESTs were aligned, and consensus sequences were generated.
The FXYD sequences were found to be organized in six Unigenes for Atlantic salmon (FXYD2: Ssa.25839; FXYD5: Ssa.7814; FXYD6: Ssa.2060; FXYD7: Ssa.23766; FXYD8: Ssa.13359; FXYD9: Ssa.10273), with homologous sequences found in rainbow trout for all but Ssa.25839. In addition, two Unigenes in rainbow trout (FXYD11: Omy.10988; FXYD12: Omy.13072) were also clearly contigs representing FXYD proteins and 97–100% homologous to Atlantic salmon sequences that incorrectly are grouped with 26,033 sequences in Ssa.423. In the present study, the relevant sequences were assembled into separate contigs and dealt with individually. Seven FXYD isoforms were found in the EST sequences available for rainbow trout (Oncorhynchus mykiss) and Japanese medaka (Oryzias latipes) and eight isoforms for zebrafish. Amino acid homology to their apparent Atlantic salmon orthologues is shown in Table 2. The FXYD protein identified in Ssa.7814 differs from the others in having a much longer amino-terminal segment, and, in this case, the nucleotide consensus sequences were extended 5′ and 3′ by means of overlapping ESTs. For some isoforms, two abundant transcript variants with changes in the coding region were found. For three isoforms (FXYD2, FXYD6, and FXYD8), the available number of ESTs is relatively low, and determining differential transcripts is limited by the redundancy. For the other isoforms, “a” and “b” isotypes are based on an approximately equal number of ESTs, and a higher redundancy gives the sequences good quality. In the case of FXYD12, three transcript variants were detected and are denoted “a”, “b”, and “c”.
Homology examination and phylogenetic analysis.
In the present study, a classification of teleost FXYD isoforms is proposed based on their homology to the known isoforms (Table 3) and phylogenetic analysis (Fig. 1) and is supported by specific characters, like tissue distribution and glycosylation domains. In general, conservation of sequence outside the membrane domain is limited, making phylogenetic analysis difficult. The homology of the proposed salmon and zebrafish FXYD2 to its mammalian counterpart is somewhat greater than to other isoforms (Table 3), and phylogenetic analysis includes this teleost isoform in a clade with mammalian, amphibian, and bird isoforms (Fig. 1). The proposed FXYD5 is very weakly similar to mouse FXYD5 (Table 3). It shares a node with the mammalian and the putative chicken FXYD5 and, for the latter, with a bootstrap value of 67 (Fig. 1). The FXYD6 and FXYD8 are both very clearly grouped with their previously described zebrafish homologue (Fig. 1; Ref. 55) and are both most similar to tetrapod FXYD6 (Fig. 1, Table 3). The presumed salmon FXYD7 is weakly similar to the mouse counterpart (expect value = 1e-06, Table 1). Phylogenetic analysis groups teleost FXYD7 with mammalian and amphibian isoforms (Fig. 1). FXYD9 are grouped with the previously described zebrafish homologue (Fig. 1; Ref. 55) and shark FXYD10 and most similar to tetrapod FXYD3 and FXYD4 (Fig. 1, Table 3). The proposed FXYD11 shares highest homology with tetrapod FXYD3 and FXYD4, teleosts FXYD9, and shark FXYD10 and are grouped with these in the phylogenetic tree (Fig. 1). The proposed FXYD12 shares little homology with any known isoform and is isolated in the phylogenetic tree. The annotation of FXYD2, FXYD5, and FXYD7 was all supported by a parsimonious model bootstrap analysis (bootstrap values: 69 for FXYD2, 100 for FXYD5, 83 for FXYD7).
The predicted protein products of the eight genes are aligned in Fig. 2, as translated from consensus nucleotide sequences. For clarity, just “a” forms are used for comparisons when more than one transcript variant was identified. The eight salmon contigs clearly represents FXYD proteins, as judged by their sequential characteristics. As indicated in Fig. 2, a FXYDY domain is shared, with only minor deviations. The fish sequences have two conserved glycine residues within the predicted membrane domains, as found in all prior characterized members of the family. A serine residue at the end of the membrane domain was shared by five isoforms, whereas another polar amino acid was found in the remaining three. Just before the predicted membrane domain, an invariant leucine and arginine residue was found.
Sequence characteristics of the FXYD isoforms.
The deduced protein sequences were examined for the potential presence of membrane domains, phosphorylation sites, glycosylation sites, and signal peptides. All of the FXYD proteins were predicted to be membrane proteins with an extracellular amino terminal and a cytosolic carboxy terminal. They all contained a ∼19-residue membrane-spanning domain (Fig. 2). At least two transcript variants were found in five FXYD proteins in salmon, and these are aligned in Supplemental Table S1. (The online version of this article contains supplemental data.)
The sequence of FXYD2 (accession no. BK006252) was deduced from three ESTs organized in Unigene Ssa.25839. The three ESTs for this isoforms were 100% homologous, apart from one EST (CK880699), which contained an insert in the coding region. Therefore, two splice variants of this FXYD protein may be present. The consensus of FXYD5a (accession no. BK006253) and FXYD5b (accession no. BK006254) are derived from 50 and 32 ESTs, respectively, and organized in Unigene Ssa.7814. They are mostly homologous in the ∼100 amino acids of the carboxy terminal, where they share homology with the other teleosts members of the FXYD family. In the extended amino-terminal domain, they mainly differ in inserts that do not induce frame shifts, but are lacking in the smaller FXYD5b. The shared amino acid sequence has only a few substitutions, but additional regions are found in the “a” variant. However, since substitutions are found, the two forms may not merely be splice variants. None of them contains predicted intracellular phosphorylation sites, and both contains a transmembrane domain. The sequences of FXYD6 are based on four ESTs (accession no. BK006241), and, due to variations in the 5′ end, the deduced consensus does not cover sequence upstream of the start codon. The three ESTs containing a stop codon deviate, so consensus in the 3′ end covering the carboxy terminus of the putative peptide was not established. In Fig. 2, the peptide sequence carboxy terminus is based on DW178692, and amino acids that are not supported by the consensus are shown in lowercase letters. Five putative phosphorylation sites were found in the intracellular domain of FXYD6. Isoform 7 is represented by a series of ESTs organized in Unigene Ssa.2060, and the sequences of two transcript variants were found (FXYD7a, accession no. BK006245; FXYD7b, accession no. BK006246). The two transcript types were 100% homologous, except for a deletion in the coding region. The deletion leads to exclusion of seven amino acids in the cytosolic domain, but no reading frame shift occur. The deletion resulted, in FXYD7b, in a removal of three of the four cytosolic phosphorylation sites predicted for FXYD7a. In the case of FXYD8, the sequences of two transcript variants were recognized in rainbow trout, found in Unigene Omy.16702 and Omy.34505. A long variant (Omy.16702, 5 ESTs) contained an insert in the coding region, but was otherwise similar in amino acid sequence compared with a short variant (Omy.34505; 16 ESTs). In Atlantic salmon, it was possible to established consensus for a similar long variant in Ssa.13359 (accession no. BK006242). However, substitution and deletions are found in two ESTs (EG799673 and EG773650), and they may represent alternative short transcripts, as suggested by the variants found in rainbow trout. The possible FXYD8 variants of salmon and trout distinguished themselves by an extracellular deletion in the peptide sequence, and both contain two putative intracellular phosphorylation sites. The consensus of FXYD9a (accession no. BK006243) and FXYD9b (accession no. BK006244) are based on 73 and 99 ESTs, respectively. The two transcript variants are 96% homologous, and five nucleotide substitutions within the coding region lead to one amino acid substitution toward the amino terminal. Within the great number of ESTs representing both of these transcript variants, minor additional substitutions are found. Both contain three possible cytosolic phosphorylation sites at serine residues 73, 75, and 88, being potential substrates for a series of protein kinases [ribosomal S6 kinase, protein kinase (PK) A, PKB, PKC, and PKG].
Isoform FXYD11 is represented by a great number of ESTs placed in Ssa.473, but clearly constituting a contig of their own. Two transcript types were deduced from a comparable number of sequences (FXYD11a, accession no. BK006247; FXYD11b, accession no. BK006248). The two transcript types were 98% homologous, and five nucleotide substitutions within the coding region lead to five amino acid substitutions in the extracellular domain, three of which are within the putative signal peptide. The substitutions introduce an additional predicted membrane domain in variant “a”, located upstream of the conserved membrane domain. However, this possible additional membrane domain may be part of a putative signal peptide. Both putative proteins have two potential cytosolic phosphorylation sites.
Isoform FXYD12 is represented by a series of ESTs placed in Ssa.473 but constituting their own contig. Nineteen ESTs represent FXYD12a (accession no. BK006249). Two other less numerous transcript types, FXYD12b (accession no. BK006250) and FXYD12c (accession no. BK006251), are found to be mostly homologous, except toward the 3′ end. The two variants contain a deletion leading to a frame shift and a series of substitutions in the nucleotide sequences following. This may result in dissimilar carboxy terminals in the mature peptides. The SignalP 3.0 server was used to analyze for likely signal peptides, and these were found in FXYD5, -6, -8, -9, and -11, but not in FXYD2, -7, and -12. There are no sites for N-glycosylation outside the predicted signal-peptide area, except in FXYD5a and FXYD5b, where eight and five potential N-glycosylation sites were found, respectively. Thirty potential O-glycosylation sites were found in the extracellular domain of salmon FXYD5, and this character is also observed in the isoform of Japanese medaka (31 sites) and zebrafish (22 sites).
Expression of the eight FXYD isoforms was analyzed in osmoregulatory tissues (gill, kidney, intestine), excitable tissues (heart, skeletal muscle, brain), and glandular tissue (liver) from FW-acclimated salmon. The mRNA expression of FXYD2 was strongest in kidney, being 30-fold higher than in intestine and more than 100-fold higher than in any other tissue (Fig. 3A). The relative order between tissues was as follows: kidney > intestine > gill = heart = brain = liver = muscle. The ESTs representing this isoform were only found in kidney and mixed tissue libraries. The FXYD5 isoform does not display major tissue difference in expression (Fig. 3B). The relative expression levels among the tissues were as follows: kidney = gill = heart > intestine > muscle = brain = liver. The ESTs representing this isoform were found in many different libraries, thus indicating a broad and less tissue-specific distribution of this transcript.
The mRNA expression of FXYD6 was higher in brain than in any other tissue (Fig. 3C), with the following order between tissues: brain >> muscle = intestine > gill = kidney = heart > liver. Two of three ESTs representing this isoform were found in brain libraries, thus supporting the expression data. The expression of FXYD7 was 1,000-fold higher in brain than in any other tissue (Fig. 3D), with the following order between tissues: brain >>> muscle = gill = heart = liver > intestine = kidney. In accord with this observation, the ESTs representing this isoform were found in brain libraries. The order of expression of FXYD8 (Fig. 4A) was as follows: muscle > brain = gill > kidney = intestine = heart > liver. The ESTs representing this isoform, and annotated to tissue, were found in muscle, lymphoreticular, and thyroid libraries. The FXYD9 isoform was expressed at a relatively high level in all of the investigated tissues, with the following order between tissues (Fig. 4B): heart > gill = brain = liver > kidney = intestine = muscle. In accord with our data, the ESTs representing this isoform were found in many different tissue libraries. The expression of FXYD11 was 1,000-fold higher in gill than any other examined tissue (Fig. 4C), with the following relative order between tissues: gill >>> kidney = intestine = heart = muscle = brain = liver. The ESTs representing this isoform were from thyroid, thymus, and gill. The expression of FXYD12 was 100-fold higher in kidney and intestine than in any other tissue (Fig. 4D). The relative order between tissues was as follows: kidney = intestine >> heart = muscle = brain > gill = liver. The ESTs representing this isoform belonged to kidney and intestine libraries.
To get an indication of the relative importance of the different FXYD isoforms in osmoregulatory tissues in FW relative to SW, the expression level in fully acclimated salmon was examined. The relative expression of the different isoforms in gill, kidney, and intestine (Figs. 5 and 6) was generally the same as found previously (Figs. 3 and 4). Specificity of expression in SW was the same as in FW, as far as FXYD11 appeared to be gill specific, FXYD12 kidney and intestine specific, and FXYD2 kidney specific. In gill, the levels of FXYD5, FXYD6, FXYD9, FXYD11, and FXYD12 transcripts doubled in SW salmon compared with FW salmon, yet this was only significant for FXYD12. In kidney, FXYD2 and FXYD12 were both expressed at a significantly lower level in SW compared with FW salmon.
It is increasingly clear that, in mammals, FXYD protein isoforms are involved in adjusting the kinetics of Na+-K+-ATPase to specific demands in different tissues and cell types (26). This study classified several novel isoforms of these proteins and demonstrated differential tissue expression and response to salinity in a euryhaline teleost. Two isoforms were expressed consistently in all tissues, whereas differential expression of three isoforms, in particular in osmoregulatory tissues, and three isoforms in excitable tissues is suggested.
Identification and characterization of expressed FXYD genes.
Rather than being poorer, consensus sequences derived from multiple ESTs can be better quality than average sequences in the GenBank nr division. Individual ESTs have the same problems, but analysis of multiple replicate sequences gives the consensus a higher confidence through the attained redundancy. The sequences coding FXYD proteins were picked out based on alignment with the conserved region of sequences formerly described. Separate Unigenes were located for all but two family members. Both FXYD11 and FXYD12 were located in the Ssa.423, annoted β-globin. However, both were 97–100% homologous to sequences found in their own Unigene in rainbow trout (Table 2). Unigene formation is an automated process, and there is little doubt that Ssa.423 has sequences that should not be included. For example, a series of Atlantic salmon ESTs (EG845224, EG828874, EG828875, EG855281, EG855280, EG845223) are located in this cluster, but share 97% nucleotide homology with the α1a-subunit of Na+-K+-ATPase from rainbow trout. Assembled contigs and consensus are able to pick up misplaced ESTs, a process that is, however, difficult to automate because of alternatively spliced forms, contamination, and chimeric entries. In the present study, the relevant sequences were assembled into separate contigs and dealt with individually. Both start and stop codons were found in all isoforms, indicating that they all may be translational competent and suggesting that they are not expressed pseudogenes. Most family members had apparent orthologue ESTs arranged in Unigenes in other teleosts (e.g., zebrafish, Japanese medaka, and rainbow trout; Table 2). As shown in Table 2, very high amino acid homology (82–100%) was found between the orthologues of two salmonids: Atlantic salmon and rainbow trout.
A classification of the teleost FXYD proteins are proposed based on ML phylogenetic analysis (Fig. 1) and homology examination (Table 3) and were supported by parsimonious analysis, tissue distribution, and further characters. The homology of the proposed salmon and zebrafish FXYD2 to mammalian counterpart is only slightly greater than to other isoforms (Table 3). However, phylogenetic analysis includes the teleost isoforms in a clade with mammalian, amphibian, and bird isoforms (Fig. 1), and expression is found predominantly in the kidney, similar to what has been observed in mammalian FXYD2 (49). Similarly, presumed salmon FXYD7 is only weakly similar to the mouse counterpart (Table 3). Again, phylogenetic analysis groups teleost FXYD7 with mammalian and amphibian isoforms (Fig. 1), and expression appears to be brain specific, as seen in mammals (7). The proposed FXYD12 share little homology with the γ-subunit (FXYD2) and is isolated in the phylogenetic tree. It is located upstream of FXYD6 on the zebrafish genome, just like it is seen in FXYD2 in mammals (FXYD6-FXYD2). This could suggest that FXYD12 is a second teleost homologue of the γ-subunit. Interestingly, FXYD12 is expressed predominantly in kidney and intestine, and, like FXYD2 and FXYD7, it has no predicted signal peptide, which parallels what is found for these isoforms in other vertebrates (17). The proposed FXYD5 shares a node with the mammalian and putative chicken homologues (Fig. 1) and is very weakly similar to mouse FXYD5 (expect value < 1e-5). In this FXYD protein isoform, a large extracellular domain is observed in the proposed teleost FXYD5, similar to what is seen in other vertebrates. The domain is in the salmonid protein predicted to be very highly O-glycosylated, as it is seen in mammalian, amphibian, and bird FXYD5. This character is only observed in this isoform and appears to be well conserved. Surprisingly, both salmon FXYD5a and FXYD5b are longer than any other FXYD5, including other nonsalmonid teleost FXYD5, like Japanese medaka (Unigene Ola.17166) and zebrafish (Unigene Dr.133190). The long salmon FXYD5 does not appear to be an intron-derived cloning artifact, since the sequence material is ESTs derived from mRNA. Both FXYD5a and FXYD5b are based on more than 30 ESTs, with a redundancy that is never lower than 11 and 14, respectively. In addition, the rainbow trout counterparts also appear to be extended compared with other teleosts, suggesting that this may be an occurrence specific to the salmonid lineage. The phylogenetic analysis suggests that the new teleost isoform, FXYD11, along with the previously described FXYD9 (55), is grouped with amphibian FXYD3 and mammalian FXYD3 and FXYD4. The analysis does not clearly group them with FXYD3 or FXYD4, and it is possible that FXYD4 has occurred in the mammalian lineage, since no amphibian form was found in the available amphibian database. Shark FXYD10 also belongs to this clade and could be an elasmobranch counterpart to teleost FXYD9 (similarity: 70%; expect value: 1e-21 with blast).
In all of the teleost family members, the FXYD motif was largely conserved, with two exceptions (Fig. 2): 1) the tyrosine is exchanged with phenylalanine, another aromatic amino acid, in FXYD9; 2) the aspartic acid is exchanged with its amide, aspargine, in FXYD11. No such substitution is seen in FXYD1–7 from mouse, rat, or human (55), but is found consistently in FXYD9 and FXYD11 of rainbow trout, zebrafish, and Japanese medaka. The aspartic acid substitution with aspargine is also observed in the newly annoted human FXYD8, identified within the DNA sequence of the human X chromosome (UniProt KB entry: P58550). A tyrosine following the FXYD domain was conserved in all isoforms of the four teleosts inspected. This tyrosine is retained in mammalian FXYD1, -2, -6, and -7. As shown in Fig. 2, an arginine following a leucine and a leucine following a glycine are further conserved motifs, as is a glycine later in the potential membrane domain. These features confirm the association of these genes compared with the FXYD family signature sequence (Ref. 26, or see the Expasy site: http://us.expasy.org/cgi-bin/nicedoc.pl?PDOC01014). The conservation in the transmembrane span of bigger and smaller side chains suggests that this is not merely an anchoring structure, and the two conserved glycines within this span thus may represent points involved in helix packing (1, 33). Outside of the signature motif, there is considerable sequence variation between family members, as is also the case in mammals.
For isoforms 5, 7, 9, and 11, two major transcript variants with changes in the coding region were found, and each is represented by a comparable number of ESTs. For FXYD12, three transcript variants were detected. The transcript variants are aligned in Supplemental Table S1. In FXYD2, FXYD6, and FXYD8, one consensus sequence could be extracted, but additional transcript variant may still escape detection due to the low number of ESTs available for these isoforms. In the case of FXYD7, “a” and “b” variants are most likely splice variants, as their sole difference is a deletion in the coding sequence. Splice variants are also known in mammals where the FXYD transcripts have been shown to be made up of at least six exons (56). Among the remaining of the salmon isoforms, the variants discern themselves by deletions and substitutions for FXYD5 and FXYD12 and only substitutions for FXYD9 and FXYD11. Because genetic changes are more frequent in fishes, their genes have, in several cases, more isoforms than other vertebrates (59). A recent gene duplication event could be involved in the occurrence of isotypes, and, in favor of this explanation, a recent polyploidization has been detected in the salmonid lineage (2).
The sequence of the eight salmon genes contained a putative membrane domain downstream of the FXYDY motif (Fig. 2). In all family members, the carboxy terminal has more positive charge than the NH2-terminal end, consistent with insertion in the membrane with the amino terminal outward. This orientation has been demonstrated experimentally in mammals for FXYD1 and FXYD2 (5, 47). The mammalian family members have residues in the carboxy terminal ends that potentially could be phosphorylation sites. FXYD1 (39), FXYD2 (18), and FXYD10 (44) have been phosphorylated in vitro, leading to altered interaction with the Na+-K+-ATPase. Potential regulatory phosphorylation sites are found in the intracellular domain of all salmon isoforms, except FXYD12a and FXYD5. It is important to stress that the presence of predicted phosphorylation sites does not automatically mean that they are physiologically relevant. For instance, even though phosphorylation of consensus sites in the purified α-subunit is observed, proximity to the plasma membrane or adjacent loops may alter these site to be inaccessible in living cells (58). However, shark FXYD10 contains three regulatory phosphorylation sites in the cytosolic domain (44), and these are also found in FXYD9 (Fig. 1), indicating a conserved physiological importance. In general, the FXYD9 isoform has the highest degree of conservation among the examined teleost species (Table 2), indicating that it may be involved in physiological processes that are not evolving within this group of vertebrates.
Judging from the relative abundance of transcript in the tissues examined, it appears that FXYD6 is expressed mainly in brain (Fig. 3). FXYD6 shares highest homology with its mammalian counterpart (Table 3), which is highly expressed in the brain (36), but also in epithelial cells of the inner ear (21). It is possible that a difference in basal expression in other tissues than brain may result from different degrees of innervation. The expression of FXYD7 was also specific to the brain. FXYD7 shares no obvious homology with any mammalian FXYD family members. The highest homology in the full sequence is found in FXYD7, a family member also exclusively expressed in the brain (7). In skeletal muscle tissue, FXYD8 expression was more than 20-fold higher than in the brain and gill and more than 200-fold of that found in heart muscle. This isoform shared highest homology with the mammalian FXYD6 isoform (Table 3) and may represent a muscle-specific analog. FXYD8 could be expressed in muscle cells or, for instance, nerve cells in the muscle tissue. Three of the salmon isoforms were predominantly expressed in osmoregulatory tissues. FXYD11 was almost exclusively expressed in the gill, FXYD2 in the kidney, and FXYD12 in the kidney and intestine. Two isoforms, FXYD5 and FXYD9, did not display a major tissue-specific expression pattern. Although originally identified in cancer cells, mammalian FXYD5 is expressed in a series of normal tissues (41), and this parallels the current observations. Interestingly, the isoform most similar to teleost FXYD9, shark FXYD10, is highly expressed in rectal gland, but also in heart, brain, intestine, and kidney (44).
In many tissues, FXYD proteins are known to affect cellular ion transport by interacting with Na+-K+-ATPase (17). The Na+-K+-ATPase is the major energizer of transepithelial ion transport in the gill, kidney, and intestine of teleosts, and, therefore, FXYD proteins may be expected to play a central role in the functioning of these tissues as well. During salinity acclimation of euryhaline teleosts, a substantial remodeling of these tissues is observed. For example, major changes in cell type (37) and expression of ion transport proteins are observed in the gill epithelium (e.g., Refs. 51, 57). In addition, Na+-K+-ATPase isoform shifts (49) and lipid composition of the gill cell membrane may play a role during salinity acclimation (13).
The expression of FXYD isoforms appeared at similar levels in the intestine of FW and SW salmon. This is unexpected, considering the major change that appears in water and ion transport and Na+-K+-ATPase expression in the intestine during acclimation to hyperosmotic conditions (see Ref. 29); however, this does not exclude a differential role in FW and SW. Several isoforms responded to salinity change in the gill and kidney, possibly signifying changes in type and function of ion-transporting cells in these tissues. The relative abundance in the gill of the generally expressed FXYD9 and gill-specific FXYD11 isoforms doubled in SW compared with FW salmon; however, this difference was insignificant. Changes in expression pattern should be investigated in the transition phase during salinity change as well, since it is possible that the response may be larger during the transient stress period than in fully acclimated fish. By comparison, induction of gill transcripts with major roles in hypoosmoregulation (Na+-K+-ATPase, Na+-K+-2Cl− cotransporter, and CFTR-like chloride channel) is at maximum level in euryhaline killifish in the initial days after SW transfer (51). In the kidney, the decreased levels of FXYD2 and FXYD12 mRNA in SW could reflect qualitative changes, since the renal Na+-K+-ATPase level itself is known to be unaffected by salinity in other salmonids like rainbow trout (24) and brown trout (42). It is also possible that posttranscriptional regulation is a factor, as in the regulation of FXYD2 in mammalian kidney cells, which has been shown to involve both a transcriptional and a posttranscriptional component (16).
Perspectives and significance.
The present study classified a series of FXYD isoforms in teleosts, and the expression patterns indicated general as well as tissue-specific isoforms among these. In addition, some isoforms appeared to be responsive to major changes in environmental salinity, suggesting an involvement in regulation of ion transport. The next step will be to develop tools to study FXYD protein function at protein level to establish the physiological role. In this context, the interrelationship with the various distinct isoforms of Na+-K+-ATPase in osmoregulatory tissues will be a challenge to investigate. Thus this study has opened a new perspective and a whole new level of complexity in the field of teleost osmoregulation.
The research was supported by a grant from The Danish Natural Research Council (272-06-0526) and a Postdoctoral Fellowship grant by The Carlsberg Foundation (2005-1-311).
Steffen S. Madsen (University of Southern Denmark) is warmly thanked for critical review of the manuscript.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2008 the American Physiological Society