The fate of the duplicated androgen receptor in fishes: a late neofunctionalization event?
- Véronique Douard†1,
- Frédéric Brunet†2,
- Bastien Boussau3,
- Isabelle Ahrens-Fath4,
- Virginie Vlaeminck-Guillem2,
- Bernard Haendler4,
- Vincent Laudet2Email author and
- Yann Guiguen1
© Douard et al; licensee BioMed Central Ltd. 2008
Received: 21 June 2008
Accepted: 18 December 2008
Published: 18 December 2008
Based on the observation of an increased number of paralogous genes in teleost fishes compared with other vertebrates and on the conserved synteny between duplicated copies, it has been shown that a whole genome duplication (WGD) occurred during the evolution of Actinopterygian fish. Comparative phylogenetic dating of this duplication event suggests that it occurred early on, specifically in teleosts. It has been proposed that this event might have facilitated the evolutionary radiation and the phenotypic diversification of the teleost fish, notably by allowing the sub- or neo-functionalization of many duplicated genes.
In this paper, we studied in a wide range of Actinopterygians the duplication and fate of the androgen receptor (AR, NR3C4), a nuclear receptor known to play a key role in sex-determination in vertebrates. The pattern of AR gene duplication is consistent with an early WGD event: it has been duplicated into two genes AR-A and AR-B after the split of the Acipenseriformes from the lineage leading to teleost fish but before the divergence of Osteoglossiformes. Genomic and syntenic analyses in addition to lack of PCR amplification show that one of the duplicated copies, AR-B, was lost in several basal Clupeocephala such as Cypriniformes (including the model species zebrafish), Siluriformes, Characiformes and Salmoniformes. Interestingly, we also found that, in basal teleost fish (Osteoglossiformes and Anguilliformes), the two copies remain very similar, whereas, specifically in Percomorphs, one of the copies, AR-B, has accumulated substitutions in both the ligand binding domain (LBD) and the DNA binding domain (DBD).
The comparison of the mutations present in these divergent AR-B with those known in human to be implicated in complete, partial or mild androgen insensitivity syndrome suggests that the existence of two distinct AR duplicates may be correlated to specific functional differences that may be connected to the well-known plasticity of sex determination in fish. This suggests that three specific events have shaped the present diversity of ARs in Actinopterygians: (i) early WGD, (ii) parallel loss of one duplicate in several lineages and (iii) putative neofunctionalization of the same duplicate in percomorphs, which occurred a long time after the WGD.
Actinopterygian fishes have provided the first clear demonstration of an ancient whole genome duplication (WGD) in vertebrate evolution . This event was originally suggested based on the finding that zebrafish and medaka possess seven Hox clusters [2–4], compared to four in mammals and one in most invertebrates. It was confirmed later on by comparative mapping  and through the analysis of genome sequences of two pufferfishes [1, 6]. Indeed, many short duplicated groups of linked genes were identified in the Takifugu rubripes and Tetraodon nigroviridis genomes [1, 7]. The duplication event leading to these duplicates was dated by molecular clock to a window between divergence of Actinopterygians from Tetrapods, and diversification of teleost fish [8, 9]. In addition, all chromosomes of Tetraodon nigroviridis were assigned to syntenic groups of duplicated genes, demonstrating the genomic scale of the duplication. It was further shown that each pair of duplicated genes was homologous to one non-duplicated human chromosomal region .
One basic question regarding gene duplication is the fate of duplicated genes. According to the Duplication – Degeneration – Complementation (DDC) model proposed by Force et al. , duplicated genes may have three main fates: the majority of duplicated copies are lost, some duplicated genes are subfunctionalized (i.e. they share the ancestral function of their non duplicated ancestor) and some others undergo neofunctionalization (i.e. they change their function when compared to their ancestor). In most cases the sub- or neofunctionalization events are classically considered to have occurred relatively soon after the duplication event. We recently suggested that a biased subset of genes was retained as duplicates after the genome duplication and that gene retention was biased with regard to biological processes . Most notably, we observed an enrichment of fish genomes in new paralogs implicated in development, supporting the link between genome duplication and fish morphological diversity [15, 18]. In addition, several studies have shown that sub- or neofunctionalization events can be observed at the expression level when specific pairs of duplicated genes are studied (see for example [19–21]). Of note, if there are some examples of neofunctionalization events affecting the biochemical function of genes after the vertebrate WGD, much less numerous specific examples of that sort were shown for fish duplicated genes [22–24].
Androgen receptors, species names and their accession numbers.
European clawed frog
African clawed frog
Spotted green pufferfish
European sea bass
Red sea bream
Indian major carp
Black Widow tetra
The case of AR in fish is particularly interesting to study as sex determination mechanisms are known to be particularly plastic in Actinopterygians [37, 38]. For example, sequential hermaphroditism is common among marine fishes, particularly in tropical and subtropical seas, and can involve females becoming males (protogyny) or males becoming females (protandry), and also bidirectional (repetitive) sex changes [39–42]. Sex changes among species with well organized social and mating systems are controlled by social cues [41, 43–45] and involve complete alterations in gonadal anatomy and function, as well as changes in color and behavior. It is known that sex steroid hormones play important roles in sex change and behavior in many fish species, and androgens have been shown to be crucial for completion of this process in many protogynous hermaphrodites [46–48].
Interestingly, in human, mutations of the AR gene represent the molecular basis of androgen insensitivity syndrome (AIS) . AIS is characterized by defective virilization in 46, XY individuals. The phenotypic spectrum of AIS is extremely large: Complete AIS (CAIS) is characterized by completely female external genitalia. In Partial AIS (PAIS) the phenotype ranges from almost female external genitalia through ambiguous forms to predominantly male external genitalia with hypospadias. Minimal (or Mild) forms of AIS exist which are characterized by impaired spermatogenesis with or without a slight virilization deficit. In addition, the androgen receptor is also implicated in prostate cancer and a specific set of mutations often occurred in patients whose cancer became androgen-independent, an evolution of poor clinical prognosis .
In this paper, we reconstructed the evolutionary history of AR in Actinopterygians. We observed a complex history shaped by three successive events well separated in time: (i) an ancestral duplication event specific to teleost fishes corresponding to the WGD; (ii) a parallel loss of one duplicated copy (AR-B) in basal Clupeocephala and (iii) a major sequence divergence indicative of a change in functional constraints in the AR-B duplicate of Percomorphs. This evolutionary history together with the striking mutation patterns is indicative of a putative neofunctionalization event that took place late during AR-B evolution It is tempting to link this neofunctionalization event to the plasticity of sex determination in Percomorphs.
Results and discussion
The Androgen receptor is duplicated in teleost fishes
Using a combination of RT-PCR with degenerate primers designed in the conserved C and E domains and in silico search against various databases, we were able to characterize 26 different new AR cDNA fragments from 20 different fish species. Along with these AR cDNA fragments, we also identified other steroid receptors (NR3C group) in A. baerii and E. stoutii (Genbank accession numbers ABF50787 and ABF50785). This is probably due to a combination of the high conservation of the C and E domains used to design primers among all steroid nuclear receptors and the low stringency of touchdown PCR procedure that we used. In this study, five new sequences of AR were also identified using database searches and 21 new sequences of AR were isolated by RT-PCR with degenerated primers, the majority of them being teleosteans (20), one being chondrichthyan, one being Dipnoi and one being chondrostean (Table 1).
Of note, one expressed sequence tag in Oryzias latipes was found to match with the 5' end region of the divergent AR in Haplochromis burtoni (AF121257). The corresponding clone was further sequenced and was confirmed to be a medaka AR. No more than one gene was identified in zebrafish by searching both EST databases and the whole genome sequence (see below). Additional file 1 provides an amino acid alignment of a representative choice of these sequences, focusing on complete DBD and LBD sequences. A complete alignment is available upon request to F.B.
AR-B was secondarily lost in basal Clupeocephala, including zebrafish
When analyzing the tree presented in Figure 1, we were puzzled to observe that the AR-B sequence can be found in basal teleosts (Heterotis, Anguilla) as well as in many Percomorphs but is missing in many basal Clupeocephala lineages.
As this observation could be due to an experimental bias linked to a failure to amplify a divergent gene by PCR, we then first checked that this observation was not due to an artifactual lack of detection of the AR-B gene. To this end, we focused our analysis on zebrafish for which a large number of data (complete genome, ESTs, etc...) are available. First, we carried out RT-PCR experiments using different batches of primers and several RNA extracts of zebrafish embryos at various developmental stages, as well as adult organs. In all cases, we detected only one AR sequence whereas our primer batches were able to detect divergent NR3C steroid receptors such as GR, MR or PR. PCR experiments based on DNA amplification of short fragments contained in only one exon also failed. Finally, we intensively screened the release Zv7 (13 July 2007) of the zebrafish genome using various fragments of the AR gene as baits without any significant hit. Of note, no sequence reminiscent of a pseudogene was detected.
Due to the availability of a complete and assembled zebrafish genome sequence, we tried to better understand the fate of the AR-B gene in zebrafish. In Tetraodon, we found two AR genes, AR-A and AR-B (Table 1 and Figure 2). Since we and others previously showed that an extensive synteny persists between Tetraodontiformes and zebrafish genomes [1, 7, 54], we precisely mapped in Tetraodon and zebrafish the syntenic regions containing AR-A and AR-B. Figure 4 clearly shows that AR genes map in a large duplicated region corresponding to chromosomes T1 and T7 in Tetraodon. Chromosome T7 in Tetraodon is syntenic to chromosomes Z5, Z10 and Z21 in zebrafish. Interestingly, the zebrafish AR-A ortholog is present in chromosome Z5, as predicted based on conserved synteny. A detailed map shows that the organization of this region is conserved between tetraodon and zebrafish (data not shown). The Tetraodon AR-B gene map to chromosome T1 and the region encompassing the gene corresponds mainly to the zebrafish chromosome Z14. The mapping of the region containing the Tetraodon AR-B sequence on the zebrafish genome shows that this region has been scrambled during evolution. Many gene orders are not conserved and large fragments are missing or were exchanged (data not shown). The same consideration is reached when the medaka genome is considered (see Additional file 3).
These data indicate clearly that a secondary loss of AR-B occurred in zebrafish. Interestingly, in related Cypriniformes (5 species), Characiformes (1 species) and Siluriformes (1 species) that altogether form the well-supported clade Otophysi , we also found only one AR-A sequence and no AR-B one. We recently screened EST data available for all these species and we could not find any sequence reminiscent of AR-B. Of course, although complete genome sequences are not available and RT-PCR results can artifactually miss a divergent sequence, these data collectively suggest that an unique event of loss of AR-B occurred early on in the Otophysi lineage.
It is clear that loss of duplicated genes is a very common fate after a genome duplication event but the present analysis nicely illustrates a late case of neofunctionalization. Our data suggest that basal teleosts and percomorphs kept two functional copies of AR whereas "intermediate" lineages such as Otophysi and Salmoniformes lost it secondarily (see below). According to the topology of teleost fish phylogeny presented in Figure 1, our results imply two independent losses of AR-B, one at the base of Otophysi and one at the base of Salmoniformes (see also Figure 5). This is based on the assumption that the current topology based essentially on complete mitochondrial DNA analysis is correct in the respective placement of Salmoniformes and Otophysi [51, 55, 57]. If, as suggested by some authors, these two groups form a monophyletic clade, it may be possible that in fact only one ancestral event of loss occurred . In that case, we can predict that AR-B should not be found in Esociformes. In any case, our present data plead for the search of AR-A and AR-B in orders of Actinopterygians located at key positions in the evolutionary tree: it would be interesting for example to see if AR-B is present in other Ostariophysi lineages such as Gonorhynchiformes, or Clupeomorphs  as well as other Protacanthopterygii such as Esociformes, Argentinoidea and Osmeroidea. This will allow a more precise determination of when the events of loss occurred .
It is difficult to speculate with the data available why the Otophysi and the salmonids apparently do not need a second AR-B gene. Given the major function of AR in sex determination and sex organ differentiation, it is tempting to link these events with these processes but given that these data on Otophysi and salmonids are limited to some specific models such as zebrafish, salmon and trout, it is up to now difficult to find an obvious connection. It is striking that zebrafish and salmonids are extremely different regarding sex determination and sex differentiation. In addition, as discussed above, the exact phylogenetic range of this loss of AR-B is still unclear.
Functional shift of AR-B in Euteleosts
In the tree presented in Figure 2, we noticed the presence of a highly divergent terminal group of AR-B sequences. This is confirmed when a larger dataset including partial sequences is used to construct a phylogeny with any of the 4 methods used (Figure 3 and data not shown). In all cases, we found a long terminal branch uniting divergent AR-B sequences. This divergent AR-B subtype unambiguously (bootstrap value: 1000 out of 1000; posterior probabilities: 1.00) clusters AR sequences of fish belonging to the percomorphs, i.e., the seabass D. labrax, the sand goby, P. minutus, the nile tilapia, O. niloticus and A. burtoni, a scorpaeniforme with M. scorpius, the shorthorn sculpin,; a beloniforme with the medaka, O. latipes, a cyprinodontiforme with the mosquitofish, G. affinis, and 2 tetraodontiformes with the tetraodon, T. nigroviridis and the fugu, T. rubripes. Indeed, when we considered the sequence alignment (see Additional file 1), we observed a serie of mutations that are present only in the percomorph AR-B sequences (highlighted in green). The divergence of these sequences corresponds to a transient episode of sequence divergence as the AR-B sequences clustered inside this group are not particularly variable. Thus, all these data suggest that the percomorphs AR-B are connected to the basal teleosts AR-B through a long branch and exhibit some striking sequence divergence at key positions. From the phylogenetic range of species in which these divergent AR-B sequences are found, it is likely that this acceleration occurred specifically in percomorphs, although this remains to be fully established by a broader taxonomic sampling including other Neotelestoi lineages such as basal Acanthomorphs (e.g. Gadiformes; ) as well as Bericyformes . To really assess if this event is found in all Percomorphs, some basal lineages (e.g. Ophidiiformes) of this extremely vast group of fishes should also be studied . In the mean time, given our observation that divergent AR-Bs are found only in percomorphs from our dataset, we will refer to these divergent sequences as "percomorph AR-B".
It is important to insist on the fact that basal teleosts (Anguilla and Heterotis) clearly contained AR-A and AR-B paralogs. For AR-A, this is not difficult to establish given that this gene is present in a wide phylogenetic range of species. For AR-B, the assignment is less obvious since, as discussed above, this gene has been lost in basal Clupeocephala. The fact that the AR-Bs from Anguilla and Heterotis are indeed orthologs of the percomorph AR-B is indicated by several features: (i) these sequences exhibit a few key sequence signatures that represent synapomorphies of AR-B sequences (highlighted in yellow and orange in Additional file 1), this is for example the case of Gly633, Ser861 and Ser928 in the LBD; (ii) the topology of the phylogenetic tree supports this assumption albeit with a moderate support (posterior probability of 0.96, bootstrap value of 459‰; Figure 2). Of note, we constructed trees based on Bayesian analysis which confirm that the topology presented in Figures 2 and 3 is robust (see Additional file 2).
The most likely scenario accounting for the data available concerning Actinopterygian AR evolution is therefore a three step model (Figure 5): (i) ancestral duplication of a unique AR gene during the WGD event specific of teleost fishes. This explains why Anguilla and Heterotis have two AR sequences, AR-A and AR-B; (ii) secondary loss of AR-B in basal Clupeocephala (Otophysi and Salmoniformes) explaining the restricted phylogenetic occurrence of AR-B when compared to AR-A; (iii) a late specific divergence of AR-B. The long branch connecting percomorph AR-B to the basal AR-B sequences is indicative of the accumulation of numerous mutations and we thus proposed that it corresponds to a functional shift that has affected the AR-B protein.
We therefore wanted to test whether the two groups of paralogous genes AR-A and percomorph AR-B were under different selective pressures. We reasoned that if selective pressures differed between the two groups, there should be sites undergoing substitutions in the AR-A subtree and constrained in the percomorph AR-B subtree, and symmetrically sites constrained in the AR-A subtree undergoing substitutions in the AR-B subtree. Patterns of evolutionary rates in one subtree versus the other were compared to answer this question: are they significantly more different than they would be if branches were picked at random among the two subtrees? Expected numbers of substitutions were estimated for all branches of the tree and all sites of the alignment . A non-symmetric correspondence analysis was applied on these numbers of substitutions, and the percentage of variance between branches explained when branches are clustered according to the two subtrees was computed. The significance of this percentage was assessed by a permutation test based on 500 000 replicates, where branches are picked randomly from the two subtrees. Among the 500 000 random clusterings, only 0.36% explained a higher percentage of variance among branches than the clustering according to the paralogous subtrees (See Additional file 4). As branches have been normalized with respect to their lengths by the correspondence analysis, this variance comes from differences in patterns of substitutions, not branch lengths. Therefore, patterns of substitutions are significantly more different between the two groups of paralogous genes than between two random groups of branches. This suggests that selective pressures differ between AR-A and AR-B genes, which is in favor of a possible neofunctionalization. The same conclusion is reached with the use of the PAML software . A significant change in the selective pressure onto the branch specific to the AR-B in percomorphs (p-value = 1.688343e-08) is unequivocally detected, although the test is not sensitive enough to tell whether it is a relaxation of the selective pressure or positive selection that drove this change. It should be made clear that, in the absence of a functional characterization, including a comparison of a basal non-duplicated AR (e.g. sturgeon), duplicated AR with a non-divergent AR-B (e.g. eel) and duplicated AR with a divergent AR-B (e.g. medaka) this neofunctionalization cannot yet be formaly proved and should be regarded as only putative.
This pattern of a late spectacular divergence of a duplicated gene in a precise taxonomic group is an interesting case in which the duplication and the functional shifts are clearly two recognizable events that were decoupled in time. The AR-B gene will thus be a very interesting model to study the precise functional and biological impact of these two events since we have sequences of non-duplicated fish AR (sturgeon), duplicated AR-A (in eel and medaka for example), duplicated and non divergent AR-B (eel) and duplicated and divergent AR-B (medaka). In addition, we have other interesting cases for comparison such as a unique zebrafish AR-A gene with secondary loss of AR-B. The fact that AR is a gene encoding a nuclear hormone receptor with a known ligand, a clear biological role and for which several functional tests are available renders this gene particularly suitable for a precise integrated study of the consequences and respective roles of duplication and evolutionary sequence divergence. For example, it may be very interesting to study if, as proposed recently at a broader scale for nuclear receptors, sequence divergence is correlated to expression divergence .
Analysis of the substitution pattern in relation to human Androgen Insensitivity Syndrome
We thus studied in more detail the 38 mutations found in the divergent AR-B sequences in order to see if some of them could have obvious functional consequences. Of note, and not surprisingly, none of these mutations affect the positions known to directly interact with the ligand or implicated in coactivator binding as determined in the 3D structure of the AR LBD complexed with various ligands [61–63]. Few specific changes are observed such as L744V, M749L, Q783H and M895I (numbering according to the consensual human AR mutations database). Some substitutions are also observed in the AF-2 region: A898 is substituted to a S in most AR-Bs and to a G in most AR-As, as well as I899 is substituted to a V in most AR-Bs. Of note, they are observed in both AR-As and divergent AR-Bs and overall, they are unlikely to account for significant functional consequences.
Then, we scrutinized the positions in the DBD and LBD divergent in AR-B and we checked whether these mutations are affecting amino acids found mutated in human pathologic conditions (Figure 6 and see Additional file 6). We were particularly interested by mutations occurring in Androgen Insensitivity Syndrome (AIS) or prostate cancer since these pathologies affect the ability of the receptor to regulate transcription of target genes in response to ligand binding. AIS is a pathologic condition in humans defined by the eventual occurrence of female differentiation despite the male XY genome and results from germinal mutations in the human AR gene. As discussed above, AIS can be complete (CAIS), partial (PAIS) or mild (MAIS) . We noted that effectively, some divergent AR-B specific mutations are localized in close proximity to functionally relevant residues and may thus impact, in a subtle manner, the function of the receptor. For example the substitution Y739L is observed in all divergent AR-Bs (see Additional file 1). Located close to M742, another amino acid involved in ligand binding, Y739 could influence ligand binding by itself since its substitution to aspartic acid has been described in a CAIS patient . The same question addresses several other amino acids specifically different in divergent AR-Bs as compared to AR-As, such as the F856L substitution which has been observed in patients with CAIS  (see Additional file 6).
Common substitutions in fish AR and AR diseases in human.
AR-A = B
The above analyses suggest that these sequence differences between AR-A and AR-B will affect the functionality of these receptors and are linked to a putative neofunctionalization event. This remains of course to be directly addressed through in vitro and in vivo analysis of the role of AR-A and AR-B in suitable fish models. How this functionality is precisely affected remains therefore an open question. The teleost duality in terms of active androgens involved in reproduction  is of great interest in that context. Of note, AR-A and AR-B have been shown to both bind regular androgens and the fish specific 11-oxygenated androgens (11KT) although no direct comparison between AR-A and AR-B has been carried out in the same species until now. We thus have no clear comparison of the respective affinities and potencies of AR-A and AR-B for DHT and 11KT. Of special interest are the in vivo binding studies carried out in two perciform species, i.e. the Atlantic croaker, Micropogonias undulates and the kelp bass, Paralabrax clathratus, that demonstrated the existence of two different nuclear androgen receptors that may mediate the physiological actions of different androgens [34–36]. The mutation pattern we observed in AR-B is indicative of a neofunctionalization event at the functional level, but it is likely that this may also be coupled to differences at the expression level. Indeed, neofunctionalization of the expression pattern has been suggested in the cichlid fish A. burtoni in which it has been shown that AR-A and AR-B have distinct expression patterns in the brain , with a differential implication of these receptors in the maintenance of social dominance status of male fish . Taking into account the functional shift that we specifically observed in the Percomorph lineage, it is tempting to link this functional shift and the sexual lability that is observed in this lineage as the Percomorphs contain nearly 90% of all the hermaphrodite species known to date . One may thus hypothesize that the existence of two functionally divergent AR genes play a role in the plasticity of sex determination often observed in these fishes. In this sense, the presence of the divergent AR-B gene could be viewed as a permissive factor allowing the evolvability of divergent sex determination in these fishes.
Fish and RNA extraction
Common and scientific names of all fish species used in this study are given in Table 1. Tissues samples (whole fish, ovaries, testis and brain) were obtained from fish specimens collected in the wild (Perca fluviatilis, Lepomis gibbosus, Pomatoschistus minutus, Myoxocephalus scorpius, Protopterus annectens, Labeo rohita and Eptatretus stoutii), bred in captivity in experimental or aquaculture facilities (Dicentrarchus labrax, Ctenopharyngodon idella, Gambusia fluvalis, Tetraodon fluvalis, Danio rerio, Carassius auratus, Clarias gariepinus, and Heterotis niloticus) or from local aquarium fish retailers (Gymnocorymbus ternetzi and Acipenser baerii). For Ginglymostoma cirratum, we used the Epigonal Nurse Shark cDNA Library provided by Dr. M.F. Flajnik at the University of Miami School of Medicine . All animals were anesthetized with 2-phenoxyethanol and then sacrificed by decapitation before the sampling of tissues. Total RNA was prepared after homogenization in TRIZOL® reagent (Invitrogen, Cergy Pontoise, France) following the manufacturer's instructions and mRNA was further purified from 500 μg of total RNA using the OligoTex mRNA kit (Qiagen).
For cDNA synthesis, 2 μg of mRNA were denatured in the presence of oligo (dT) (0.5 μg) for 5 min at 70°C, and then chilled on ice. Reverse transcription (RT) was performed at 37°C for 1 hour using M-MLV reverse transcriptase (Promega, Madison, WI) as described by the manufacturer. Namely, 2 μg of mRNA were reverse transcribed with 200 units of M-MLV reverse transcriptase in the presence of 1.25 μl of each dNTP at 10 mM, 5 μl of M-MLV 5× reaction buffer and 25 units of RNasin® (Promega, Madison, WI, USA), in a total volume of 25 μl.
Degenerate oligonucleotide primers were designed after alignment of various fish, reptile, bird and mammalian AR amino acids sequences. Two degenerate primers, AR.AS: TGY TAY GAR GCI GGI ATG AC [CYEAGM] and AR.AAS: AAI ACC ATI ACI BYC CTC CA [WMGVMVF], were selected respectively in the highly conserved DBD (C domain) and steroid LBD (E domain) regions. The alignment of the divergent H. burtoni AR-B (see Table 1) sequence was used to design another specific degenerate primer, AR.BS: TGC TTY ATG KCG GGN ATG [CFMSGM] in the same region as AR.AS. AR.AS and AR.BS were both used in conjunction with AR.AAS [AR.AS × AR.AAS and AR.BS × AR.AAS]. A second set of degenerate primers outside the PCR fragments produced by AR.AS × AR.AAS or AR.BS × AR.AAS was designed: AR3S: GTI TTY TTY AAR AGR GCI GC [VFFKRAA] and AR3AS: CCA ICC CAT IGC RAA IAA IAC CAT [MVFAMGW] and were used in a nested PCR strategy.
Touchdown RT -PCR and cloning of AR sequences
The RT-PCR strategy was used according to Escriva et al. . PCR reactions were set up using 2 μl of cDNA or 2 μl of Epigonal Nurse Shark cDNA library and 0.5 units of Taq Polymerase (Sigma), 200 μM of dNTPs, 30ρmol of each degenerated primer and 2.5 μl of Taq buffer 10× (Sigma). The total volume of the reaction was 25 μl and the cycling Touchdown PCR conditions were: 94°C for 1 min, 20 cycles of regularly decreasing annealing temperature from 50°C to 40°C for 30 sec and 72°C for 30 sec, and 30 cycles at the annealing temperature of 40°C, ending at 72°C for 30 sec. PCR products were analyzed on agarose gel (1%) and amplified DNA fragments of the anticipated length (average between 370 and 450 bp) were subsequently subcloned into pCR 2.1 plasmid. Bacteria (INVαF', E. coli TOP10 cells, Invitrogen) were transformed by electroporation, spread on LB-ampicillin agar plates and incubated overnight at 37°C. From 10 to 20 randomly selected recombinant colonies were then screened either using PCR with primers amplifying the inserts (T7 and M13 reverse primers) or by hybridization of nitrocellulose membrane lifts with a rainbow trout AR radiolabelled (dCTP32) probe. Positive clones were sequenced using a dideoxy cycle-sequencing method with the Dye Terminator Cycle Sequencing Kit (Applied Biosystems) and reaction sequences were read on an ABI PRISM 310 Genetic Analyzer (Applied Biosystems). A secondary nested PCR was carried out for RNA samples from G. ternezi, D. rerio, P. annectens, and A. baerii species (see Table 1) using as template a first PCR reaction, obtained using the primers [AR3S × AR3AS], at a 1/100 dilution and a second set of nested degenerate primers [AR.AS × AR.AAS] or [AR.Bs × AR.AAS]. PCR conditions, subsequent cloning, clone selection, and sequencing were as described above.
Searching AR in sequence databases
Homologous DNA and protein fish ARs were searched on available public databases (non redundant, Expressed Sequence Tags) using the various BLAST programs available through the network servers at the National Center of Biotechnology Information http://0-www.ncbi.nlm.nih.gov.brum.beds.ac.uk/BLAST/. We also retrieved AR sequences from the whole genome databases at the Ensembl Genome browser http://www.ensembl.org/index.html. From Ensembl v48 (Aug. 2007), we retrieved AR sequences belonging to the Ensembl family ENSF000000000291.
Sequence and structural analysis
Multiple alignments of the deduced amino acid sequences were generated with Muscle using the default parameters . Phylogenetic trees were realized by multiple alignments of deduced amino acid sequences using the neighbor-joining and parcimony methods implemented in PhyloWin . PhyML  was used to generate maximum likelihood phylogenetic trees. Bayesian trees were generated using MrBayes v3 http://mrbayes.csit.fsu.edu/index.php.
To test whether the two groups of paralogous genes AR-A and AR-B were under different selective pressures, patterns of substitutions estimated in the AR-A subtree versus the AR-B one in percomorphs were compared. Expected numbers of substitutions per site and per branch were estimated with the CoMap program  based on the Bio++ library . This produced a matrix containing branches of the tree as rows, and sites of the alignment as columns. Branches belonging to the AR-A and AR-B subtrees were selected, discarding the two eel sequences, so that the number of branches was the same in the two subtrees. A non-symmetric correspondence analysis was applied on the resulting submatrix, and the percentage of variance between branches explained when branches are clustered according to the two subtrees computed. The significance of this percentage was assessed by a permutation test based on 500,000 replicates, where branches were clustered randomly. All these analyses were conducted with the ade4 package  in the R environment (R development core team).
In order to determine any change in the selective pressure along the branch leading to the percomorphs AR-Bs beside of this previous test, we also used PAML version 4 http://abacus.gene.ucl.ac.uk/software/paml.html.
Location of all the substitutions found in the LBD of the human AR was retrieved from Bruce Gottlieb's database at his web site http://androgendb.mcgill.ca/. Amino acid substitutions in this human AR database and those specific to the two main lineages were thus positioned onto the 3D structure using PyMOL (by Warren L. Delano, version 2004, http://pymol.sourceforge.net/).
We are grateful to Marc Robinson-Rechavi, Marie Sémon, Gérard Benoit, Michael Schubert and Hector Escriva for critical reading of the manuscript. This work has been funded by MENRT, CNRS for financial support. The research leading to these results has received funding from the EC Sixth Framework Program under the project X-OMICS, grant agreement n° LSHG-CT-2004-512065.
- Jaillon O, Aury JM, Brunet F, Petit JL, Stange-Thomann N, Mauceli E, Bouneau L, Fischer C, Ozouf-Costaz C, Bernot A, et al: Genome duplication in the teleost fish Tetraodon nigroviridis reveals the early vertebrate proto-karyotype. Nature. 2004, 431 (7011): 946-957.View ArticlePubMedGoogle Scholar
- Amores A, Force A, Yan YL, Joly L, Amemiya C, Fritz A, Ho RK, Langeland J, Prince V, Wang YL, et al: Zebrafish hox clusters and vertebrate genome evolution. Science. 1998, 282 (5394): 1711-1714.View ArticlePubMedGoogle Scholar
- Naruse K, Fukamachi S, Mitani H, Kondo M, Matsuoka T, Kondo S, Hanamura N, Morita Y, Hasegawa K, Nishigaki R, et al: A detailed linkage map of medaka, Oryzias latipes: comparative genomics and genome evolution. Genetics. 2000, 154 (4): 1773-1784.PubMed CentralPubMedGoogle Scholar
- Wittbrodt j, Meyer A, Schartl M: More genes in fishes?. Bioassay. 1998, 20: 511-515.View ArticleGoogle Scholar
- Postlethwait JH, Woods IG, Ngo-Hazelett P, Yan YL, Kelly PD, Chu F, Huang H, Hill-Force A, Talbot WS: Zebrafish comparative genomics and the origins of vertebrate chromosomes. Genome Res. 2000, 10 (12): 1890-1902.View ArticlePubMedGoogle Scholar
- Aparicio S, Chapman J, Stupka E, Putnam N, Chia JM, Dehal P, Christoffels A, Rash S, Hoon S, Smit A, et al: Whole-genome shotgun assembly and analysis of the genome of Fugu rubripes. Science. 2002, 297 (5585): 1301-1310.View ArticlePubMedGoogle Scholar
- Woods IG, Wilson C, Friedlander B, Chang P, Reyes DK, Nix R, Kelly PD, Chu F, Postlethwait JH, Talbot WS: The zebrafish gene map defines ancestral vertebrate chromosomes. Genome Res. 2005, 15 (9): 1307-1314.PubMed CentralView ArticlePubMedGoogle Scholar
- Christoffels A, Koh EG, Chia JM, Brenner S, Aparicio S, Venkatesh B: Fugu genome analysis provides evidence for a whole-genome duplication early during the evolution of ray-finned fishes. Mol Biol Evol. 2004, 21 (6): 1146-1151.View ArticlePubMedGoogle Scholar
- Vandepoele K, De Vos W, Taylor JS, Meyer A, Peer Van de Y: Major events in the genome evolution of vertebrates: paranome age and size differ considerably between ray-finned fishes and land vertebrates. Proc Natl Acad Sci USA. 2004, 101 (6): 1638-1643.PubMed CentralView ArticlePubMedGoogle Scholar
- Jordan IK, Marino-Ramirez L, Wolf YI, Koonin EV: Conservation and coevolution in the scale-free human gene coexpression network. Mol Biol Evol. 2004, 21 (11): 2058-2070.View ArticlePubMedGoogle Scholar
- Robinson-Rechavi M, Laudet V: Evolutionary rates of duplicate genes in fish and mammals. Mol Biol Evol. 2001, 18 (4): 681-683.View ArticlePubMedGoogle Scholar
- Peer Van de Y, Taylor JS, Braasch I, Meyer A: The ghost of selection past: rates of evolution and functional divergence of anciently duplicated genes. J Mol Evol. 2001, 53 (4–5): 436-446.View ArticlePubMedGoogle Scholar
- Vankatesh B, Tay A, Dandona N, Patil J, Brenner S: A compact cartilaginous fish model genome. Curr Biol. 2005, 15 (3): R82-R83.View ArticleGoogle Scholar
- Hoegg S, Brinkmann H, Taylor JS, Meyer A: Phylogenetic timing of the fish-specific genome duplication correlates with the diversification of teleost fish. J Mol Evol. 2004, 59 (2): 190-203.View ArticlePubMedGoogle Scholar
- Hoegg S, Meyer A: Hox clusters as models for vertebrate genome evolution. Trends Genet. 2005, 21 (8): 421-424.View ArticlePubMedGoogle Scholar
- Kasahara M: The 2R hypothesis: an update. Curr Opin Immunol. 2007, 19 (5): 547-552.View ArticlePubMedGoogle Scholar
- Force A, Lynch M, Pickett FB, Amores A, Yan YL, Postlethwait J: Preservation of duplicate genes by complementary, degenerative mutations. Genetics. 1999, 151 (4): 1531-1545.PubMed CentralPubMedGoogle Scholar
- Brunet FG, Crollius HR, Paris M, Aury JM, Gibert P, Jaillon O, Laudet V, Robinson-Rechavi M: Gene loss and evolutionary rates following whole-genome duplication in teleost fishes. Mol Biol Evol. 2006, 23 (9): 1808-1816.View ArticlePubMedGoogle Scholar
- Bertrand S, Thisse B, Tavares R, Sachs L, Chaumot A, Bardet PL, Escriva H, Duffraisse M, Marchand O, Safi R, et al: Unexpected Novel Relational Links Uncovered by Extensive Developmental Profiling of Nuclear Receptor Expression. PLoS Genet. 2007, 3 (11): e188-PubMed CentralView ArticlePubMedGoogle Scholar
- Bruce AE, Oates AC, Prince VE, Ho RK: Additional hox clusters in the zebrafish: divergent expression patterns belie equivalent activities of duplicate hoxB5 genes. Evol Dev. 2001, 3 (3): 127-144.View ArticlePubMedGoogle Scholar
- Hurley IA, Scemama JL, Prince VE: Consequences of hoxb1 duplication in teleost fish. Evol Dev. 2007, 9 (6): 540-554.View ArticlePubMedGoogle Scholar
- Braasch I, Salzburger W, Meyer A: Asymmetric evolution in two fish-specifically duplicated receptor tyrosine kinase paralogons involved in teleost coloration. Mol Biol Evol. 2006, 23 (6): 1192-1202.View ArticlePubMedGoogle Scholar
- Escriva H, Bertrand S, Germain P, Robinson-Rechavi M, Umbhauer M, Cartry J, Duffraisse M, Holland L, Gronemeyer H, Laudet V: Neofunctionalization in vertebrates: the example of retinoic acid receptors. PLoS Genet. 2006, 2 (7): e102-PubMed CentralView ArticlePubMedGoogle Scholar
- Howarth DL, Law SH, Barnes B, Hall JM, Hinton DE, Moore L, Maglich JM, Moore JT, Kullman SW: Paralogous Vdr's in Teleosts: Transition of Nuclear Receptor Function. Endocrinology. 2008Google Scholar
- Liley NR, Stacey NE: Hormones, pheromones and reproductive-behavior in fish. Fish Physiology. Edited by: Hoar DJRaEMD WS. 1983, New York.: Academic Press, 9: 1-63.Google Scholar
- Borg B: Androgens in Teleost fishes. Comparative Biochemistry and Physiology C-pharmacology Toxicology and endocrinology. 1994, 109 (3): 219-245.View ArticleGoogle Scholar
- Laudet V, Gronemeyer H: The nuclear receptors. 2002, Edited by Press A. LondonGoogle Scholar
- Touhata K, Kinoshita M, Tokuda Y, Toyohara H, Sakaguchi M, Yokoyama Y, Yamashita S: Sequence and expression of a cDNA encoding the red seabream androgen receptor. Biochim Biophys Acta. 1999, 1450 (3): 481-485.View ArticlePubMedGoogle Scholar
- Takeo J, Yamashita S: Two distinct isoforms of cDNA encoding rainbow trout androgen receptors. J Biol Chem. 1999, 274 (9): 5674-5680.View ArticlePubMedGoogle Scholar
- Ogino Y, Katoh H, Yamada G: Androgen dependent development of a modified anal fin, gonopodium, as a model to understand the mechanism of secondary sexual character expression in vertebrates. FEBS Lett. 2004, 575 (1–3): 119-126.View ArticlePubMedGoogle Scholar
- Ikeuchi T, Todo T, Kobayashi T, Nagahama Y: cDNA cloning of a novel androgen receptor subtype. J Biol Chem. 1999, 274 (36): 25205-25209.View ArticlePubMedGoogle Scholar
- Todo T, Ikeuchi T, Kobayashi T, Nagahama Y: Fish androgen receptor: cDNA cloning, steroid activation of transcription in transfected mammalian cells, and tissue mRNA levels. Biochem Biophys Res Commun. 1999, 254 (2): 378-383.View ArticlePubMedGoogle Scholar
- Olsson PE, Berg AH, von Hofsten J, Grahn B, Hellqvist A, Larsson A, Karlsson J, Modig C, Borg B, Thomas P: Molecular cloning and characterization of a nuclear androgen receptor activated by 11-ketotestosterone. Reprod Biol Endocrinol. 2005, 3: 37-PubMed CentralView ArticlePubMedGoogle Scholar
- Sperry TS, Thomas P: Characterization of two nuclear androgen receptors in Atlantic croaker: comparison of their biochemical properties and binding specificities. Endocrinology. 1999, 140 (4): 1602-1611.PubMedGoogle Scholar
- Sperry TS, Thomas P: Identification of two nuclear androgen receptors in kelp bass (Paralabrax clathratus) and their binding affinities for xenobiotics: comparison with Atlantic croaker (Micropogonias undulatus) androgen receptors. Biol Reprod. 1999, 61 (4): 1152-1161.View ArticlePubMedGoogle Scholar
- Sperry TS, Thomas P: Androgen binding profiles of two distinct nuclear androgen receptors in Atlantic croaker (Micropogonias undulatus). J Steroid Biochem Mol Biol. 2000, 73 (3–4): 93-103.View ArticlePubMedGoogle Scholar
- Devlin RH, Nagahama Y: Sex determination and sex differentiation in fish: an overview of genetic, physiological, and environmental influences. Aquaculture. 2002, 2008: 191-364.View ArticleGoogle Scholar
- Volff JN: Genome evolution and biodiversity in teleost fish. Heredity. 2005, 94 (3): 280-294.View ArticlePubMedGoogle Scholar
- Atz JW: Hermaphroditic Fish. Science. 1965, 150 (3697): 789-797.View ArticlePubMedGoogle Scholar
- Kuwamura T, Nakashima Y: New aspects of sex change among reef fishes: recent studies in Japan. Environmental Biology of Fishes. 1998, 52: 125-135.View ArticleGoogle Scholar
- Munday PL, Caley MJ, Jones GP: Bi-directional sex change in a coral-dwelling goby. Behav Ecol Sociobiol. 1998, 43: 371-377.View ArticleGoogle Scholar
- Policansky D: Sex choice and the size advantage model in jack-in-the-pulpit (Arisaema triphyllum). Proc Natl Acad Sci USA. 1981, 78 (2): 1306-1308.PubMed CentralView ArticlePubMedGoogle Scholar
- Fricke H, Fricke S: Monogamy and sex change by aggressive dominance in coral reef fish. Nature. 1977, 266 (5605): 830-832.View ArticlePubMedGoogle Scholar
- Shapyro DY: Differentiation and evolution of sex-change in fishes I. Bioscience. 1987, 37 (7): 490-497.View ArticleGoogle Scholar
- Warner RR: Mating-behavior and hermaphroditism in coral-reef fishes. American Scientist. 1984, 72 (2): 128-136.Google Scholar
- Bhandari RK, Alam MA, Soyano K, Nakamura M: Induction of female-to-male sex change in the honeycomb grouper (Epinephelus merra) by 11-ketotestosterone treatments. Zoolog Sci. 2006, 23 (1): 65-69.View ArticlePubMedGoogle Scholar
- Cardwell JR, Liley NR: Hormonal control of sex and color change in the stoplight parrotfish, Sparisoma viride. Gen Comp Endocrinol. 1991, 81 (1): 7-20.View ArticlePubMedGoogle Scholar
- Kroon FJ, Liley NR: The role of steroid hormones in protogynous sex change in the Blackeye goby, Coryphopterus nicholsii (Teleostei: Gobiidae). Gen Comp Endocrinol. 2000, 118 (2): 273-283.View ArticlePubMedGoogle Scholar
- McPhaul MJ: Androgen receptor mutations and androgen insensitivity. Mol Cell Endocrinol. 2002, 198 (1–2): 61-67.View ArticlePubMedGoogle Scholar
- Heinlein CA, Chang C: Androgen receptor in prostate cancer. Endocr Rev. 2004, 25 (2): 276-308.View ArticlePubMedGoogle Scholar
- Ishiguro NB, Miya M, Nishida M: Basal euteleostean relationships: a mitogenomic perspective on the phylogenetic reality of the "Protacanthopterygii". Mol Phylogenet Evol. 2003, 27 (3): 476-488.View ArticlePubMedGoogle Scholar
- Miya M, Takeshima H, Endo H, Ishiguro NB, Inoue JG, Mukai T, Satoh TP, Yamaguchi M, Kawaguchi A, Mabuchi K, et al: Major patterns of higher teleostean phylogenies: a new perspective based on 100 complete mitochondrial DNA sequences. Mol Phylogenet Evol. 2003, 26 (1): 121-138.View ArticlePubMedGoogle Scholar
- Lavoue S, Sullivan JP: Simultaneous analysis of five molecular markers provides a well-supported phylogenetic hypothesis for the living bony-tongue fishes (Osteoglossomorpha: Teleostei). Mol Phylogenet Evol. 2004, 33 (1): 171-185.View ArticlePubMedGoogle Scholar
- Kasahara M, Naruse K, Sasaki S, Nakatani Y, Qu W, Ahsan B, Yamada T, Nagayasu Y, Doi K, Kasai Y, et al: The medaka draft genome and insights into vertebrate genome evolution. Nature. 2007, 447 (7145): 714-719.View ArticlePubMedGoogle Scholar
- Lavoue S, Miya M, Inoue JG, Saitoh K, Ishiguro NB, Nishida M: Molecular systematics of the gonorynchiform fishes (Teleostei) based on whole mitogenome sequences: implications for higher-level relationships within the Otocephala. Mol Phylogenet Evol. 2005, 37 (1): 165-177.View ArticlePubMedGoogle Scholar
- Allendorf FW, Stahl G, Ryman N: Silencing of duplicate genes: a null allele polymorphism for lactate dehydrogenase in brown trout (Salmo trutta). Mol Biol Evol. 1984, 1 (3): 238-248.PubMedGoogle Scholar
- Saitoh K, Miya M, Inoue JG, Ishiguro NB, Nishida M: Mitochondrial genomics of ostariophysan fishes: perspectives on phylogeny and biogeography. J Mol Evol. 2003, 56 (4): 464-472.View ArticlePubMedGoogle Scholar
- Steinke D, Salzburger W, Braasch I, Meyer A: Many genes in fish have species-specific asymmetric rates of molecular evolution. BMC Genomics. 2006, 7: 20-PubMed CentralView ArticlePubMedGoogle Scholar
- Teletchea F, Laudet V, Hanni C: Phylogeny of the Gadidae (sensu Svetovidov, 1948) based on their morphology and two mitochondrial genes. Mol Phylogenet Evol. 2006, 38 (1): 189-199.View ArticlePubMedGoogle Scholar
- Dutheil J, Pupko T, Jean-Marie A, Galtier N: A model-based approach for detecting coevolving positions in a molecule. Mol Biol Evol. 2005, 22 (9): 1919-1928.View ArticlePubMedGoogle Scholar
- Matias PM, Donner P, Coelho R, Thomaz M, Peixoto C, Macedo S, Otto N, Joschko S, Scholz P, Wegg A, et al: Structural evidence for ligand specificity in the binding domain of the human androgen receptor. Implications for pathogenic gene mutations. J Biol Chem. 2000, 275 (34): 26164-26171.View ArticlePubMedGoogle Scholar
- Sack JS, Kish KF, Wang C, Attar RM, Kiefer SE, An Y, Wu GY, Scheffler JE, Salvati ME, Krystek SR, et al: Crystallographic structures of the ligand-binding domains of the androgen receptor and its T877A mutant complexed with the natural agonist dihydrotestosterone. Proc Natl Acad Sci USA. 2001, 98 (9): 4904-4909.PubMed CentralView ArticlePubMedGoogle Scholar
- He B, Minges JT, Lee LW, Wilson EM: The FXXLF motif mediates androgen receptor-specific interactions with coregulators. J Biol Chem. 2002, 277 (12): 10226-10235.View ArticlePubMedGoogle Scholar
- Suzuki K, Fukabori Y, Nakazato H, Hasumi M, Matsui H, Ito K, Kurokawa K, Yamanaka H: Novel amino acid substitutional mutation, tyrosine-739-aspartic acid, in the androgen receptor gene in complete androgen insensitivity syndrome. Int J Androl. 2001, 24 (3): 183-188.View ArticlePubMedGoogle Scholar
- Ahmed SF, Cheng A, Dovey L, Hawkins JR, Martin H, Rowland J, Shimura N, Tait AD, Hughes IA: Phenotypic features, androgen receptor binding, and mutational analysis in 278 clinical cases reported as androgen insensitivity syndrome. J Clin Endocrinol Metab. 2000, 85 (2): 658-665.PubMedGoogle Scholar
- Harbott LK, Burmeister SS, White RB, Vagell M, Fernald RD: Androgen receptors in a cichlid fish, Astatotilapia burtoni: structure, localization, and expression levels. J Comp Neurol. 2007, 504 (1): 57-73.PubMed CentralView ArticlePubMedGoogle Scholar
- Burmeister SS, Kailasanath V, Fernald RD: Social dominance regulates androgen and estrogen receptor gene expression. Horm Behav. 2007, 51 (1): 164-170.PubMed CentralView ArticlePubMedGoogle Scholar
- Rumfelt LL, Lohr RL, Dooley H, Flajnik MF: Diversity and repertoire of IgW and IgM VH families in the newborn nurse shark. BMC Immunol. 2004, 5 (1): 8-PubMed CentralView ArticlePubMedGoogle Scholar
- Escriva H, Robinson M, Laudet V: Evolutionary biology of the nuclear receptor superfamily. Steroid/Nuclear receptor superfamily: Practical Approach. Edited by: Picard D. 1999, Academic Press, 1-28.Google Scholar
- Edgar RC: MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res. 2004, 32 (5): 1792-1797.PubMed CentralView ArticlePubMedGoogle Scholar
- Galtier N, Gouy M, Gautier C: SEAVIEW and PHYLO_WIN: two graphic tools for sequence alignment and molecular phylogeny. Comput Appl Biosci. 1996, 12 (6): 543-548.PubMedGoogle Scholar
- Guindon S, Gascuel O: A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Syst Biol. 2003, 52 (5): 696-704.View ArticlePubMedGoogle Scholar
- Dutheil J, Gaillard S, Bazin E, Glemin S, Ranwez V, Galtier N, Belkhir K: Bio++: a set of C++ libraries for sequence analysis, phylogenetics, molecular evolution and population genetics. BMC Bioinformatics. 2006, 7: 188-PubMed CentralView ArticlePubMedGoogle Scholar
- Thioulouse J, Lobry JR: Co-inertia analysis of amino-acid physico-chemical properties and protein composition with the ADE package. Comput Appl Biosci. 1995, 11 (3): 321-329.PubMedGoogle Scholar
- Gottlieb B, Lehvaslaiho H, Beitel LK, Lumbroso R, Pinsky L, Trifiro M: The Androgen Receptor Gene Mutations Database. Nucleic Acids Res. 1998, 26 (1): 234-238.PubMed CentralView ArticlePubMedGoogle Scholar
- Yang Z: PAML: a program package for phylogenetic analysis by maximum likelihood. Comput Appl Biosci. 1997, 13 (5): 555-6.PubMedGoogle Scholar
- Verrijdt G, Tanner T, Moehren U, Callewaert L, Haelens A, Claessens F: The androgen receptor DNA-binding domain determines androgen selectivity of transcriptional response. Nuclear Receptors: Structure, Mechanisms and Therapeutic Targets. Biochemical Society Transactions. 2006, 34 (6): 1089-1094.Google Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.