Genomic structure and expression of uncoupling protein 2 genes in rainbow trout (Oncorhynchus mykiss)

Background Uncoupling protein 2 (UCP2) belongs to the superfamily of mitochondrial anion carriers that dissociate the respiratory chain from ATP synthesis. It has been determined that UCP2 plays a role in several physiological processes such as energy expenditure, body weight control and fatty acid metabolism in several vertebrate species. We report the first characterization of UCP2s in rainbow trout (Oncorhynchus mykiss). Results Two UCP2 genes were identified in the rainbow trout genome, UCP2A and UCP2B. These genes are 93% similar in their predicted amino acid sequences and display the same genomic structure as other vertebrates (8 exons and 7 introns) spanning 4.2 kb and 3.2 kb, respectively. UCP2A and UCP2B were widely expressed in all tissues of the study with a predominant level in macrophage-rich tissues and reproductive organs. In fry muscle we observed an increase in UCP2B expression in response to fasting and a decrease after refeeding in agreement with previous studies in human, mouse, rat, and marsupials. The converse expression pattern was observed for UCP2A mRNA which decreased during fasting, suggesting different metabolic roles for UCP2A and UCP2B in rainbow trout muscle. Phylogenetic analysis including other genes from the UCP core family located rainbow trout UCP2A and UCP2B with their orthologs and suggested an early divergence of vertebrate UCPs from a common ancestor gene. Conclusion We characterized two UCP2 genes in rainbow trout with similar genomic structures, amino acid sequences and distribution profiles. These genes appeared to be differentially regulated in response to fasting and refeeding in fry muscle. The genomic organization and phylogeny analysis support the hypothesis of a common ancestry between the vertebrate UCPs.


Background
In living cells, most energy is produced in the mitochondria through oxidative phosphorylation. In this process, the electron flow from reduced substrates to oxygen gen-erates an electrochemical proton gradient across the inner membrane. This force drives the proton back into the matrix and results in ATP synthesis from ADP and P i . Uncoupling proteins (UCPs), which are members of the superfamily of mitochondrial anion-carrier proteins, are capable of dissipating the proton gradient across the inner mitochondrial membrane to generate heat while reducing the efficiency of ATP synthesis [1]. The archetypical UCP1 is expressed in brown adipose tissue of mammals and is involved in non-shivering thermogenesis [2,3]. UCP1 mRNA has been recently found in ectothermic organisms such as carp, zebrafish and pufferfish [4]. Homologues of UCP1 (UCP2, 3,4,5) have been identified from various tissues in vertebrates [5][6][7] and plants [1,8].
The present study was designed to characterize UCP2 genes in the rainbow trout and investigate their potential as candidate genes affecting traits associated with energy balance and nutrition. To this end, we analyzed the genomic structure, phylogenetic relationships with other UCPs, tissue distribution and expression in muscle of UCP2 in response to fasting.

Analysis of cDNA and amino acid sequences
We identified two similar tentative consensus sequences (TC78216 and TC78217) by homology search for UCP2 in the TIGR rainbow trout gene index (RTGI). Both TC78216 and TC78217 were found to contain full-length coding sequences from clones tcad0009a.o21 and tcad0008a.b11, respectively. An additional cDNA clone (1RT84B23) containing EST CA344639 which is assigned to TC78216 was also picked, purified and sequenced. Full sequences of tcad0009a.o21, RT84B23 and tcad0008a.b11 were deposited to GenBank and assigned accession numbers [GenBank: DQ295326, DQ295327 and DQ295328].
The deduced amino acid sequences of UCP2A and UCP2B consist of 304 and 311 amino acid residues, respectively ( Figure 1). The peptide sequence deduced from 1RT84B23 is a truncated form of that obtained from tcad0009a.o21. The deduced protein, which consisted only of transmembrane domain I and one proton carrier signature, most likely lacks the proton dissipation function as it has been demonstrated that the second transmembrane domain of UCP genes is essential for the anion channel formation [25]. This sequence was therefore discarded from further analysis.
We found 93% similarity between the UCP2A and UCP2B peptide sequences with both containing six transmembrane domains and three proton carrier signatures which define the general triplicate structure of mitochondrial uncoupling proteins [26]. The purine-binding domain involved in the control of coupling efficiency was also identified. Over the protein conserved domains rainbow trout UCP2A and UCP2B diverged only by one aminoacid at position 189 in transmembrane domain IV. The switch from alanine (UCP2A) to valine (UCP2B) should not make any difference in the three-dimensional structure of the peptides as these are both hydrophobic amino acids. On average, both rainbow trout UCP2s showed 83% amino acid similarity with zebrafish UCP2 and 78% amino acid similarity with human, rat and mouse UCP2s. UCP2 and UCP3 have been described to be fairly similar in their peptide sequences and their physiological functions [23,27,28]. Our results showed 69% amino acid similarity between rainbow trout UCP2s and mammalian UCP3s.
The identification of two copies of UCP2 genes in rainbow trout genome is consistent with the ancestral whole genome duplication [29,30] and re-diploidization of salmonidae species. Several genes and microsatellite loci in rainbow trout have been previously described to be present in multiple copies [31,32] and Palti et al. estimated that approximately two-thirds of the rainbow trout genes remained duplicated [33].

Phylogenetic relationships
We performed a phylogenetic analysis to classify rainbow trout UCP2A and UCP2B and determine their relationships to UCPs from other species (Figure 2). The tree generated included 27 peptide sequences from the UCP gene family retrieved from public databases. Trees with similar topology to that of Figure 2 were drawn when either Minimum Evolution or Unweighted Pair Group algorithms were performed. Both rainbow trout UCP2 duplicates were grouped together with their UCP2 orthologs in the same clade. Fish (D. rerio, C. carpio and O. mykiss) and frog UCP2 peptides were assigned to a specific cluster, with a high bootstrap (87%), as opposed to mammalian UCP2 genes. Our data showed a close relationship between UCP2 and UCP3 in agreement with previous expression patterns and amino acid sequence similarity suggesting that UCP2 and UCP3 are paralogs [23,27,28]. Moreover both genes have been mapped on the same chromosome in endothermic vertebrates such as human, bovine, porcine and mouse [34-37] and more recently the same evidence was found in ectothermic vertebrates like zebrafish. [4].
Overall, UCP genes were clustered in three major branches at 100% bootstrap value leading to three groups consisting of UCP1-3, UCP-4 and UCP-5. We suggest an early divergence of the UCP family gene into three major clades in agreement with the phylogeny described by Sokolova and Sokolov [38]. These authors also demonstrated that this subdivision could be correlated with a functional specialization among the UCP superfamily.

Genomic organization of rainbow trout UCP2
We identified 32 total BAC clones putatively containing an UCP2 gene of which 16 where specific to UCP2A and 16 were specific to UCP2B. After fingerprinting, most of these BAC clones were assembled into three contigs C1, C2 and C3 containing 12, 7 and 2 clones, respectively. C1 consisted of UCP2A-containing BACs while C2 and C3 consisted of UCP2B-containing BACs. Two representative BACs were selected from each contig for direct sequencing. Two BAC clones, 243P06 and 398G03, were selected from C1 and C3 for further analysis based on their sequence similarities with cDNA clones tcad0009a.o21 (UCP2A) and tcad0008a.b11 (UCP2B), respectively. Genomic sequences for UCP2A and UCP2B genomic sequences have been deposited at NCBI [GenBank: DQ295324, DQ295325].
Alignment between cDNA and BAC genomic sequences revealed the exon-intron structures of UCP2A and UCP2B ( Figure 3). In each alignment a TATA box was identified on genomic DNA 47 and 45 bp upstream of the 5'end of UCP2A and UCP2B, respectively. UCP2A and UCP2B span 4.2 kb and 3.2 kb, respectively. Both rainbow trout UCP2s contain 8 exons and 7 introns. Coding sequences for both genes covered 6 exons as described for mammalian UCP1  [68]. Sequences are presented in single letter code. Gaps, illustrated with a dash, are introduced into sequences to optimize alignment score. Most conserved amino acids are highlighted in black, less conserved ones being in decreasing gray background. Potential transmembrane conserved α-helices domains are boxed and numbered in roman numerals. Proton carrier signatures are underlined once and the potential purine binding domain is underlined twice.
Evolutionary tree of UCP gene family inferred using the neighbor-joining algorithm Figure 2 Evolutionary tree of UCP gene family inferred using the neighbor-joining algorithm. Genetic distance proportional to branch length was calculated based on amino acid difference (p-distance) with complete deletion of gaps. Bootstrap values computed after 1,000 random replications are indicated at tree nodes. Similarity between UCP2A and UCP2B was low in the introns with 60% similarity on average while the average similarity in the exon sequences was 86%. The number and the size of the exons and introns of rainbow trout, human, mouse and zebrafish UCP2s are represented in Table 2. The number of exons was identical in these vertebrate organisms although exon lengths were observed to be somewhat variable. Translated exons 4, 5, 6 and 7 were the most conserved in size between species and within rainbow trout. Exon 7 had the same size in all species. These translated exons showed 93% average sequence similarity between rainbow trout UCP2A and UCP2B while untranslated exons were 79% identical. Mutations apparently occur at a slower rate in the coding region than in the untranslated region of duplicated genes. These results show conservation of the structure of UCP2 genes coding exons within vertebrates confirming the hypothesis of a common ancestry for UCP genes.
We examined introns 1 and 2, untranslated exons 1, 2, untranslated part of exon 3 and approximately 300 bp of genomic sequence upstream of each gene's transcription initiation site as the putative promoter regions. This represented 1.6 and 1.3 kb-long sequence upstream of the translation-initiation site (ATG) of UCP2A and UCP2B, respectively. Similarity was 59% between overlaps suggesting differentiation between regulatory regions of UCP2A and UCP2B. The promoter regions of rainbow trout UCP2A and UCP2B, and the upstream non-coding sequences identified in human (4.5 kb), mouse (3.5 kb) and zebrafish (3.0 kb) UCP2 were searched for regulatory element binding sites using the TFSearch engine version 1.3. The results for ten transcription factors frequently identified are summarized in Tables 3 and 4. The identification of regulatory conserved motifs upstream of the five genes is consistent with the phylogenetic relationships   [47,48]. Our results suggest, in agreement with previous data in the literature [49,50], that some transcription factor binding sites are conserved across species while others are not.

Tissue distribution and response to fasting
The distribution of relative expression of rainbow trout UCP2A and UCP2B was investigated on a panel of 23 tissues from adult fish (Figure 4). Both UCP2A and UCP2B genes appeared ubiquitously expressed in the body as described in marsupials [23], human [6,46], carp [4] and mouse [7] making the genes likely to influence several physiological processes. The highest relative expression was observed in ovary, peripheral blood leucocytes (PBL), gill, testis, spleen and trunk kidney. In all these tissues except ovary, UCP2B mRNA relative amounts were at least twice that of UCP2A. In brain, fat, pituitary and red blood cells, UCP2A mRNA was more abundant than UCP2B mRNA. High amounts of UCP2 mRNAs in reproductive organs and in tissues rich in macrophages like PBLs, spleen, and kidney have been reported in several studies. Arsenijevic et al. [51] and Kizaki et al. [52] suggested a role for UCP2 in immunity or inflammatory responsiveness  through the regulation of reactive oxygen species in macrophages. In mouse ovary, UCP2 mRNA levels have been described to increase during ovulation as a result of the increasing number of macrophages and neutrophils [53].
In trout, ovaries in maturation are characterized by the presence of many macrophages inside the cytoplasm of the oocytes [54]. The ovaries used in our tissue distribution panel were developing. Therefore, we suggest the high amount of UCP2 mRNA in trout ovary could be due to the macrophage-rich developmental stage of the tissue used.
The initial average weight of the fish used in the current study was 231 mg (225 for the fed group and 236 for the fasted group). Following day 14, the fed fish weighed 475 mg and the fasted fish were 167 mg. At day 21 of fasting, termination of the fasted group, the average weight of the fish was 661 mg, 171 mg and 239 mg for the fed, fasted and refed groups, respectively. Rainbow trout UCP2A and UCP2B were differentially expressed in response to fasting and refeeding in fry ( Figure 5). UCP2A mRNA expression was similar between all groups at the beginning of the experiment. Three days after food deprivation, UCP2A expression was three times higher in fed fish than in fasted fish. The amount of gene expression remained significantly higher in fed fish than in fasted fish until day 21 of the experiment. Refeeding of fasted animals started at day 14 and resulted in a 2-fold increase of UCP2A at day 21. At day 28, refed and fed fish had the same amount of UCP2A expression.
At day 0, UCP2B expression is similar in all fry. At day 3, expression was 3-fold augmented in fasted fish while no significant increase was found in the fed fish. UCP2B expression drastically decreased in fasted fish to the level observed in fed fish at day 7 before a significant increase was observed again at day 14. At days 14 and 21, UCP2B was expressed higher in fasted fry than in fed fry. In fed animals, no significant differences were noted between UCP2B expression levels throughout the experiment, except at day 17. Refed fish showed the same level of expression of the gene as the fasted fish at day 14. UCP2B expression then decreased in refed fish to reach the same level expressed in fed animals at days 21 and 28. The expression pattern of UCP2B in response to fasting is consistent with what has been described for other species in the literature. The food deprivation induced-increase in mRNA expression is one of the most spectacular features of UCP2 and UCP3 in rodent and human [20][21][22][55][56][57][58].
The hypothesis is that during fasting when glucose is limiting, upregulation of muscle UCP2 is associated with the shift in muscle substrate utilization primarily towards lipid. On refeeding with a low-fat diet, UCP2 is down regulated allowing the re-use of glucose as major substrate. It has been also demonstrated that a high-fat diet induces up-regulation of UCP2 and UCP3 thus promoting fat utilization over storage [55,57]. The actual mechanism of UCP involvement in fatty acid metabolism is not yet fully understood, but Samec et al. [56] established a close relationship between expression of UCP2 and UCP3 and well-known key regulators of lipid oxidation. Our study, unlike most of experiments in the literature, monitored UCP2 mRNA levels for a relatively long fasting time (four weeks). Our results suggest that increase of UCP2B mRNA is maintained during the whole fasting time, although the higher expression is detected early.
The overall amount of expression of UCP2B genes in fry muscle were much lower than what we observed for UCP2A. Five times more cDNA was included in Real-time PCR for UCP2B primers than was used for UCP2A to detect readable amplification signals. As a result, the highest detected amount of UCP2B gene EF1-normalized expression is 0.023 compared with that of UCP2A which was 18.81. In adult fish, as described above UCP2B mRNA expression was much higher relatively to UCP2A mRNA in some of the tissues. In light of the expression data in response to fasting, UCP2B is likely the homolog of the UCP2 described in other vertebrates. UCP2A is a duplicate form of the gene that has undergone a different evolutionary process leading to a different expression pattern. This is illustrated by the sequence divergence and different cisregulatory elements identified in the promoter regions of both genes. Further functional characterization of the pro-  moter region and identification of regulatory motifs could help address the differences between mRNA expression of UCP2A and UCP2B.

Conclusion
We have identified two UCP2 genes in rainbow trout. Both genes were highly similar at the amino acid level and widely expressed in the body. Analyses of the genomic organization and phylogeny data based on protein sequence similarity suggest a close relationship between members of the UCP superfamily. In rainbow trout fry UCP2A and UCP2B appeared downregulated and upregulated on fasting, respectively. Further characterization of both genes in different physiological and dietary conditions, and developmental stages will assist in better testing and understanding hypothesis related to their specific functions and differentiation.

Identification and fingerprinting of UCP2 gene-containing BAC clones
BACs identified as positive for UCP2 genes were fingerprinted using HindIII to identify sets of overlapping clones as previously described [33]. BAC DNA was obtained using the Qiagen Large Construct Kit protocol. Following digestion and electrophoresis, gel images were captured using a Molecular Dynamics Typhoon 9210 Variable Mode Imager and exported as TIF files. Banding patterns were analyzed using Image 3.10b and FPC (Fingerprinted Contigs) V6 software [61] to assemble overlapping BACs into contigs [62].

Sequence analysis
DNA sequencing was performed using the Big Dye ® Terminator Cycle Sequencing Kit and an ABI3100 automatic sequencer (Applied Biosystems, Foster City, CA). cDNA and genomic sequences were obtained by primer walking. Sequences were analyzed using the Sequencher 4.1.4 software (Gene Codes Corporation, Ann Arbor, MI, USA). Exon-intron boundaries were identified by locating 5' donor GT and 3' acceptor AG consensus sequences in a cDNA-genomic DNA gap alignment.
Genomic sequence upstream of the rainbow trout UCP2 gene was obtained in order to perform a computational prediction of cis-acting transcription factor binding sites using the TFSearch engine [63] that uses the TRANSFAC databases. Also for each human, mouse and zebrafish, we downloaded the genomic and mRNA sequences from GenBank and aligned them to retrieve the gene structure and locate the translation-initiation site. The upstream sequences including 300 bp upstream on the genomic sequence were then selected for the prediction of transcription factors binding sites. The identifiers of human, mouse and zebrafish UCP2 genomic sequence obtained are [GenBank: NC_000011.8, NT_039433, NC_007116.1], respectively. Sequence obtained for promoter analysis was located on human chromosome 11q13 (73,363,064-73,367,806), on mouse chromosome region 7E2 (18,239,242,911) and zebrafish chromosome 5 (60,833,836,877).

Experiment design and tissue collection
Experiments were conducted on rainbow trout fry hatched at the National Center for Cool and Cold Water Aquaculture (Leetown, WV). Thirty three fish at 32 days post-hatching (dph) were placed into each of 18 × 2 4L buckets. The buckets were supplied with flow-through oxygen saturated well water at 12.5 + 0.2°C, and a photoperiod of 12 h light -12 h dark. Fish were fed a commercial diet, Finfish Starter Meal, (57% protein and 15% fat; Ziegler Bros. Inc., Gardner PA) at 2% body weight daily. Fish were divided into three treatment groups: 1) fed fish as controls, 2) fasted fish, and 3) fish fasted the first 14 days of the experiment and re-fed thereafter. The experiment was initiated 37 dph, shortly after the completion of yolk-sac absorption. Muscle tissues were collected on days 0, 3, 7, 14, 17, 21 and 28 of the experiment. Each tank was only sampled once and fish from two tanks for each treatment were sampled. At sampling, fish were anesthetized with 200 mg/L Tricaine and mixed fiber type muscle pools from four individuals were excised and immediately frozen in liquid nitrogen then stored at -80°C until RNA extraction. All animal handling and sampling procedures were reviewed and approved by the NCCCWA Institutional Animal Care and Use Committee.
Tissues were collected from branchial arch, brain, eye, fat, gill, heart, intestines, liver, ovary, pancreas, peripheral blood leukocytes, red and white muscles, pituitary, pyloric ceca, red blood cells, skin, spleen, stomach and testis on healthy 2-year old adult fish (approximately 1 kg in mass) for the survey of UCP2 mRNA expression in the rainbow trout body.

RNA isolation and first strand cDNA synthesis
Total RNA was extracted using Tri-reagent (Sigma, St Louis, Missouri) according to the manufacturer's protocol. Total RNA was precipitated in ethanol, washed and dissolved in nuclease-free water. RNA was treated with DNAse to eliminate genomic DNA contamination and reextracted using Tri-reagent. RNA concentrations were estimated by spectrophotometry and integrity was checked on agarose gels. First strand cDNA synthesis was performed in a 40 µl reaction volume containing 2 µg total RNA, 1 µg of Random Hexamer primers, 1× M-MLV reaction buffer, 500 µM of dNTPs, 200 units of M-MLV reverse transcriptase (Promega, Madison, USA) and 25 units Recombinant Rnasin ® Ribonuclease Inhibitor (Promega, Madison, USA). The mix was incubated at 37°C for 60 min with a final denaturation at 95°C for 5 min. Reactions were stored at -20°C until further use.

Real-time PCR
Specific forward and reverse primers were designed for each target and reference gene cDNA (Table 1) on the boundary of two consecutive exons or in intron flanking regions to prevent amplification of contaminating genomic DNA or to make it detectable on a gel. Either way, amplicons were run on gel to confirm their uniqueness and size expectation prior to real-time PCR runs. Real-time PCR was performed in a 15 µl reaction volume containing 1 µl of cDNA template, 250 nM of each forward and reverse primers and 7.5 µl of SYBR Green PCR Master Mix (Applied Biosystems, Foster City, USA). Amplifications were performed on the ABI PRISM 7900 Sequence Detection System real-time cycler using the following program: 50°C for 10 min, 95°C for 15 s, 40 cycles of 95°C for 15s, 60°C for 30 s, 72°C for 30s and a dissociation stage at 95°C for 15s, 60°C for 15 s and 95°C for 15s. Dissociation curves were analyzed to detect nonspecific amplification. Non-template controls were included in each run and checked for no amplification. The amount of transcript in each sample was quantified using the standard curve method [65]. The standard curve was constructed using 10 dilutions (1 to 1/1000) of a pool of cDNA. Gene expression in tissues was rendered as a ratio of target gene versus rainbow trout Elongation Factor-1α (EF1-α, GenBank Accession Number AF498320.1) as reference gene. We surveyed EF1-α, β-actin, TATA box Binding protein a (TBP-a) and 18S rRNA as potential internal control genes for RT-PCR (data not shown). Of the housekeeping genes tested, EF1-α demonstrated the most consistent Ct values between treatments (average 23.7 ± 1.1). EF1-α has been previously demonstrated to be a reliable housekeeper in several tissues of Atlantic salmon [64].
The real time PCR mRNA expression data were analyzed using the Statistica software version 6 (StatSoft. Inc., Tulsa, OK, USA). Differences in UCP2 mRNA expression between treatments were assessed at each time point using the Kruskal-Wallis non-parametric test. A pair wise post hoc comparison of means was performed when difference between treatments was significant at the 0.05 threshold.

Phylogenetic inference
Twenty seven UCP peptide sequences were identified and retrieved from public databases using keywords and BLAST searches. Multiple sequence alignments were conducted using Clustal W [66]. Sequence alignments were used as an input file to construct a consensus phylogenetic tree by the neighbor joining method using MEGA version 3 [67]. Reliability of tree was assessed by bootstrapping using 1,000 random replications.

Authors' contributions
IC carried out the molecular genetics studies, sequencing, sequence alignment, statistical analysis, and drafted the manuscript. SAG participated in the gene identification process, designed the experiment and extracted the RNA. YP contributed to the BAC fingerprinting. JY and CER participated in the study design and coordination. All authors read and approved the final manuscript.