A total of eleven vtg transcripts (six full-length and five partials) were cloned from vitellogenic livers of the labrid species investigated (Fig. 1). NCBI BLAST searches of the deduced proteins verified that all sequences are members of the Vtg family. Four distinct vtg cDNA sequences were cloned from rock cook (Centrolabrus exoletus) and goldsinny wrasse (Ctenolabrus rupestris), two of which (cevtgAb1 and crvtgAb1) were entirely novel for vertebrates. Three transcripts (lmvtgAa, lmvtgAb2 and lmvtgC) were obtained from cuckoo wrasse (Labrus mixtus). Repeated attempts using gene-specific primers to extract a second vtgAb-type transcript from livers of cuckoo wrasse did not yield any novel transcripts. The deduced amino acid sequences of the full-length rock cook and cuckoo wrasse Vtgs revealed that the VtgAa and VtgAb products are complete pentapartite type proteins (NH₂-LvH-Pv-LvL-β'-CT-COO^-) while VtgC belongs to the phosvitinless class of Vtg (NH₂-LvH-LvL-COO^-) 25.

Since it is now postulated that the vertebrate vtg gene complement represents a conserved cluster in Sarcopterygii (the tetrapod branch) and Actinopterygii (the fish branch) 7, we aligned the longest partial segment of the labrid sequences to the genomic variants in chicken (Gallus gallus) and medaka (Oryzias latipes), respectively (Fig. 2). Relative homology scores of the aligned amino acids and codons revealed three forms with highest identity to medaka olvtgAa1, olvtgAb and olvtgC, respectively (Fig. 3). The relationship between the teleost and chicken genes was less obvious, with essentially equal identity scores for the ggvtgII and ggvtgIII products. Interestingly, GgvtgI had slightly lower scores to the teleost C-type products compared to the Aa- or Ab-type products over the aligned LvH sub-region (Fig. 2). This suggests that despite their orthology (see below) the minor ggvtgI and vtgC genes have functionally diverged in association with the loss of the phosvitin (Pv) and C-terminal domains in teleosts.

Within the labrid species, sequences had highest identities to their homologs in closely related goldsinny wrasse. Since only a single Ab-type transcript was detected in cuckoo wrasse, and it showed 100% identity to the novel cevtgAb2 and crvtgAb2 sequences over the ~250 aligned aa and ~750 aligned nt (Fig. 3), we named the cuckoo wrasse Ab-type sequence lmvtgAb2.

To further classify the labrid vtgs, we conducted large-scale phylogenetic analyses of the deduced amino acid and codon alignments of available (genbank) and novel genomic (ensembl) variants. For the genomic variants, we identified three genes in 3-spined stickleback (Gasterosteus aculeatus), torafugu (Takifugu rupribes) and spotted green pufferfish (Tetraodon nigriviridis), respectively, and four genes in medaka, of which two have been independently sequenced (see Additional file 1). In zebrafish eight genes are found in the genome, of which several have been fully 26 or partially sequenced 27, and each of which is expressed and deposited in the growing oocyte 28. Validation of the tree topology was achieved through multiple methods of phylogenetic inference. Each method consistently clustered the labrid sequences as three forms: vtgAa, vtgAb and vtgC, respectively (Fig. 4). For Bayesian, maximum parsimony and neighbour-joining analyses the majority of branches were supported by 100% posterior probabilities, and 100% bootstrap values, respectively, with only minor branch rearrangements within gene groups. These data verify that three forms of vtg exist with acanthomorph teleosts, two within protacanthomorph teleosts 29, and three within ostariophysan teleosts. The data further revealed that a fourth novel gene in medaka is an Aa-type product, and we thus classified it as olvtgAa2. In a separate study, we have identified three novel gene variants in a basal clupeocephalan teleost, the Atlantic herring (Kristoffersen et al. unpublished data). These Atlantic herring transcripts clustered as a basal node to all ostariophysan vtgs in full agreement with the clupeocephalan phylogenetic rank.

The present analyses thus fully corroborate our earlier study of vertebrate vtgs 4 wherein the major and minor transcripts cluster according to taxonomic group. We further verified the inferred duplication of the vtgAa and vtgAb clusters using the method of Zmasek and Eddy 30. Hence the tree topology is inconsistent with the syntenic arrangement of vertebrate vtg genes, and suggests that different functions have evolved within the major clades. Specifically for acanthomorph teleosts, the LvH-Aa domains have evolved a sensitivity to acidic degradation following activation of V-class proton pumps during oocyte maturation 19111213151718. The data for labrid teleosts that spawn benthic and pelagic eggs support this view [1920, see below].

In order to understand the disparity between the phylogenetic and syntenic arrangement of vertebrate vtg genes, we increased the phylogenetic data set to include all known sarcopterygian vtgs (108 sequences, containing ~370 k nt), including amphibian, bird, reptile and platypus variants. In addition we examined the syntenic positions of vtg genes within non-metatherian vertebrates. The larger data set did not affect the topology of the teleost branches, but did place ggvtgI and the partial platypus transcript (ENSOANT00000031211) at the base of the tetrapoda and closer to, but on a separate branch to teleost vtgC variants (data not shown). Similarly, two further partial platypus transcripts (ENSOANT00000008462 and a construct of ENSOANT00000013101-ENSOANT00000013100) clustered as a basal node to ggvtgIII and ggvtgII, but after amphibian variants. These findings agree with the recent studies of Brawand et al. 3 and Babin 7 who demonstrated that three genes exist in monotremata and are putative orthologs of the three bird genes. Since these latter platypus genes are not yet localized to any chromosome, but are annotated in contigs 49.51 and 49.49, respectively, it was not possible to discern their true orthology, or syntenic arrangement in relation to chicken. However, by combining the results of the present phylogenetic data with the syntenic positions of the chicken and teleost vtgs, it can be stated that teleost vtgC genes are the putative orthologs of ggvtgI, and that teleost vtgAa and vtgAb genes are the putative orthologs of ggvtgII and ggvtgIII, respectively.

Previously we showed that the evolution of Vtg sub-domains is neither clock-like, nor under strong functional constraint 4. We further highlighted the disparate retention of the sub-domain structures of the mature proteins. Teleost VtgC proteins have all lost the highly phosphorylated polyserine Pv region and the C-terminal domains that are homologous to human VWFD. In chicken, however, all three Vtgs, including GgvtgI, are complete type proteins containing all encoded domains. The same appears to be true for other tetrapod Vtgs, exemplified by amphibia and the platypus, although Brawand et al. 3 have shown that premature stop codons or indels have led to loss of function in ggvtgII/III orthologs in the platypus. The loss of sub-domains in teleosts is not restricted to the VtgC class. Amongst all ostariophysan teleosts, which represent the second largest superorder of fishes 313233, the major gene expressed encodes a truncated form of Vtg that lacks the Vwfd region 2627343536. However, a complete-type gene is present in zebrafish as vtg2, and a novel juxtaposed gene (vtg8) encodes a protein that lacks the Pv and CT domains, i.e. a tripartite protein (NH₂-LvH-LvL-β'-COOH). To address the vtg orthologies in Ostariophysi, we thus included all known zebrafish coding variants in the phylogenetic data set and examined their loci.

Unlike other acanthoperygian and sarcopterygian animals for which genomic data are available, the chromosomal loci of zebrafish vtgs are complex. Seven genes (vtg1, vtg2, vtg4, vtg5, vtg6, vtg7 and vtg8, hereafter called ZfVTG1-8) are located in a tight cluster on Dre22 (bp 23,098,595 – 23,301,096), while vtg3 is located on Dre11 (bp 26,086,024 – 26,105,006). However, none of the upstream or downstream genes that flank the ggvtgII/III or vtgAa/Ab clusters in sarcopterygians or teleosts, respectively, are currently annotated on Dre22. The ZfVTG1-8 cluster thus represents an island of vtg genes that has either lost the flanking genes, or translocated from another chromosome such as Dre11. Alternatively, the chromosomal loci may reflect the remnants of WGD, where differential loss via diploidization 373839, or germline fusion and fission events have only left traces of the ancestral condition 404142. Indeed Kasahara et al. 40 have argued that Tni1 in spotted green pufferfish, which has retained the VGC, is derived from fusion of proto-chromosomes f, g and m, while Ola4 in medaka, which also has retained the VGC, is derived from proto-chromosome m. Interestingly medaka Ola17 which has lost the VGC, but retained the flanking genes, and Ola20 which also is derived from proto-chromosome m, as are Tni6 and Tni15 in spotted green pufferfish are the likely genomic remnants of the lost VGC paralogons. In zebrafish the derivatives of proto-chromosome m have translocated to Dre6, 8, 11 and 22, and fission variants have been retained in Dre2 and 24 40. Thus orthologs of the genes that flank the VGC in chicken should be localized in these chromosomes. This happens to be the case (Fig. 5). However, since none of the orthologs that flank the chicken VGC are annotated on Dre22, deciphering the orientation of the ZfVTG1-8 remains challenging. The synovial sarcoma, X breakpoint 2 interacting protein (Ssx2ip) that is juxtaposed to GgvtgII and OlvtgAa1, as in all other vertebrates 7, is currently annotated in scaffold Zv7_NA1811 in zebrafish. A BLAST search for this protein revealed multiple hits throughout the genome. However, one hit (e = 5.2 × 10^-6) was found on Dre22 at position 7.4 Mb. We have therefore oriented the ZfVGC1-8 as shown in Fig. 5, which also corresponds to the increasing bp positions observed in other vertebrates.

To ascertain the local cis-duplication events, we integrated the results of the phylogentic analyses. The topology of the ZfVTG1-8 shown in Fig. 4 precisely replicates the chromosomal loci of each gene in the genome shown in Fig. 5. This is also true for the meadaka VGC, where olvtgAa2 represents an internal duplication of the independently sequenced olvtgAa1 (Q8UW88_ORYLA). In fact each of the major genes that are expressed in vertebrates are located at the outer edge of the ggvtgII/III and vtgAa/Ab VGC. This arrangement appears to have implications for the differential expression of the cluster, where cis-regulatory elements associated with estrogen induction 4344 and recruitment of the transcription initiation complexes are concerned.

In zebrafish, vtg1 is the major gene, which can be expressed at levels 100 – 1000× higher than other variants 27. Similarly, in other teleosts, expression ratios of the Aa- or Ab-type vtgs vary according to species 1122425454647. This is also true for the major ggvtgII gene in chicken 48. Previously we have shown that in goldsinny wrasse, extreme levels of Aa-type Yps are incorporated in the growing oocyte 19. These Yps are substantially degraded to FAA during pre-ovulatory maturation, leaving only a lipovitellin light chain (LvL-Aa) fragment for embryonic development. Due to the neofunctionalization of VtgAa 4, the degree of Yp proteolysis may depend upon the relative expression levels of each vtg, and it has been predicted that a dynamic expression ratio of vtgAa and vtgAb forms should exist in relation to ambient temperature and salinity, egg size and the presence of oil globules 25. To investigate this possibility, we performed Northern blots of hepatically expressed transcripts in the three labrid species (Fig. 6).

The selected model of labrid teleosts are best known for their cleaning behaviour, a character that is currently being exploited as an environmentally-friendly means of biological control of ectopic parasites in mariculture. Members of this family, the third largest of vertebrates 49 with more than 600 species in 82 genera 50 are also known for their remarkable sex lives 51, where sex change, transvestitism and male-dominated parental care is prevalent. Although only one species spawns pelagic eggs in the temperate coastal waters of our Norwegian study area, it is noteworthy that all labrid teleosts are marine species, with the vast majority spawning pelagic eggs in a subtropical environment 52. An understanding of the differential expression in the non-disrupted vtg-gene complement in this family of fishes is an important step towards development of sex-specific molecular markers for this group.

Probes were designed from the highly conserved N-terminal area of each cDNA sequence and their gene specificity verified by subsequent cloning and sequencing. With the exception of crvtgAb2, a single band of relevant size was detected after hybridization in each species. Although several levels of regulation lie between an expressed hepatic transcript and deposited Yp product in the growing oocyte, it is noteworthy that vtg band intensities were highly correlated to the type of Yp deposited in the oocytes and the pelagic or benthic nature of the spawned egg. For the goldsinny wrasse, which spawns pelagic eggs and generates the largest pool of FAA during oocyte hydration, an exceptional level of vtgAa is expressed in the liver compared to the other vtg transcripts. These data strongly support our earlier findings where virtually all of the oocyte Yps are maturationally proteolysed to FAA that subsequently drive the osmotic flow of water into the highly hydrated egg 19. In the benthophil rock cook, more even band intensities of the four vtg transcripts are found, although higher levels of vtgAa are also expressed in this species. The rock cook is a close relative of the corkwing wrasse (Crenilabrus melops), which generates a small pool of FAA from partial Yp proteolysis during oocyte hydration 20. We have recently conducted a proteomic analysis of the oocyte and egg Yps in the rock cook and cuckoo wrasse and found that moderate oocyte hydration is associated with proteolysis of mainly vtgAa derived Yp products (Kolarevic unpublished data). In cuckoo wrasse, a species that spawns benthic eggs and shows very limited maturational proteolysis, higher levels of VtgAb-type Yps are deposited in the oocyte. These latter observations are further supported by the fact that although greater amounts of RNA were inadvertently loaded in the vtgAa lane, the vtgAb2 band had the highest intensity. The significance of higher expression levels of Ab-type vtgs relates to the fact that these genes have not neofunctionalized 4 and their Yp products remain mostly intact in the hydrated egg as the major protein reserve for the developing embryo. Taken together, these data are in line with the notion that differential expression of non-neofunctionalized and neofunctionalized vtgs in acanthomorph teleosts is related to the benthic or pelagic character of the spawned egg.

Mature female cuckoo wrasse (Labrus mixtus), rock cook (Crenilabrus exoletus) and goldsinny wrasse (Ctenolabrus rupestris) were collected using traps and gill nets in the costal waters near Bergen, Norway. Fish were transported live to the laboratory and maintained in fish tanks. Later they where euthanized in accordance with the International Guiding Principles for Biomedical Research Involving Animals as promulgated by the Society for the Study of Reproduction. Subsequent sampling of livers and ovaries was performed in a cold room (4°C). Pre-hydrated oocytes (PH ooc) and ovulated eggs (OV egg) were dissected from the ovaries and processed as described previously 19.

Total RNA was isolated from vitellogenic livers of three rock cook females using RNAeasy kit (Qiagen). Extracts were subsequently mixed together for single strand 3' and 5'-cDNA synthesis using Smart Race cDNA Amplification kit (Clonetech, http://www.clonetech.com). The alignment of Finn & Kristoffersen 4 was used to select areas that were specific to each form of vtg. Gene specific primers (GSP) (see Additional file 2) subsequently designed from nt sequences of red seabream vtgAa, vtgAb and vtgC (primers P1, P11 and P21) were then used to run 3' and 5'-RACE polymerase chain reactions (PCR) as recommended by the manufacturer.

A PCR product of approximately 4000 bp was amplified using sense primer P1. It was cloned and sequenced as described previously 19. Three sense primers (P2–P4) designed from a partial rock cook sequence were used in addition to M13 vector primers to obtain the sequence of the cloned product. In order to sequence the remaining N-terminal area of this gene, a new antisense GSP (P5) was constructed from the aforementioned sequence. The RACE PCR product (~1300 bp) was cloned and bi-directionally sequenced.

An antisense GSP for red seabream vtgAb (P11) was used in a 5'-RACE PCR together with single stranded rock cook 5'-cDNA giving ~800 bp long PCR product. After cloning and sequencing, two different ESTs were identified to match the N-terminal end of vtgAb in other teleost species using BLAST. To verify that the ESTs represent two novel products, the same experiment was conducted with new total RNAs that were extracted from two females and independently used for single strand cDNA synthesis. PCR products from both reactions were gel-purified, cloned and sequenced giving the same two distinct vtgAb sequences. Full sequence of vtgAb1 was achieved by primer walking with five sense rock cook GSPs (P12–P16). An additional sense GSP (P17) was used to obtain the remaining part of the partial vtgAb2 sequence.

Cloning of vtgC was accomplished using an antisense GSP made from red seabream nt sequence (P21) and a sense primer (P22) designed from rock cook ESTs. A PCR product of approximately 3500 bp was amplified using the latter primer and was sequenced with M13 vector primers and three additional sense primers (P23–P25).

The same extraction procedures and sequencing strategies were employed to retrieve full-length sequences of three different vtg forms in cuckoo wrasse. Cloning of PCR products amplified with the same red seabream GSPs (P1, P11 and P21) as for rock cook, gave partial sequences that were used to construct new vtgAa (P6–P10), vtgAb (P18–P20) and vtgC (P26–P29) primers. Subsequent PCR reactions, cloning and sequencing of new PCR products were done as described above. Despite using GSPs for both vtgAb types in rock cook wrasse (P12 and P13), only a single vtgAb2 transcript was obtained.

Total RNA was extracted from each of the labrid teleost livers, as described above, and electrophoretically fractioned in 1% agarose gels containing formaldehyde (6.7%), stained with ethidium bromide solution to visualize rRNAs and blotted onto a Hybond-N nylon membrane (Amersham) by capillary transfer in 10 × SSC and covalently linked to the membrane by exposure to the UV light. Membranes were prehybridized in hybridization buffer (PerfectHyb™ Plus, Sigma) at 68°C for 60 min before being hybridized with denatured and labeled ³²P-cDNA probes (Strip-EZ DNA, Ambion) overnight as described previously 53.

Four vtg gene specific probes for rock cook and three for cuckoo and goldsinny wrasse ranging is size from 602 to 736 nt were constructed from the N-terminal regions. To verify probe specificity, each was sequenced following amplification, gel purification, and excision. Blots were rinsed two times with 2 × SSC/0.1% SDS for 5 min and once with 1 × SSC/0.1% SDS for 15 min at the room temperature. Additional washing was done with 0.1 × SSC/0.1% SDS twice for 10 min at 68°C. In order to detect the signals, membranes were exposed to Kodak's BioMax MS film for 2 hr.

Multiple sequence alignments of the deduced amino acids were used to generate codon alignments of the sequenced transcripts as described previously 4. In order to determine gene orthologies, vtg genes from each of the currently sequenced teleost genomes were accessed from the ensembl servers (zebrafish, medaka, 3-spined stickleback, torafugu and spotted green pufferfish: ensembl release 49, May 2008). For zebrafish, 8 genes were identified using the graphical view and contiguous alignments of transcripts, of which 2 are located in genbank (see Additional file 1 for accession numbers). For medaka, Babin 7 annoted 5 genes on chromosome 4, however, we only found evidence of 4 vtg genes, 3 of which are located between bp 9,868,166 – 9,974,743. A novel construct (olvtgAa2) was assembled from transcripts: ENSORLESTT00000013001, ENSORLT00000007668, ENSORLESTT00000012994, ENSORLESTT00000012967, ENSORLESTT00000012953 that encoded a protein of 1671 aa. Similarly, a novel construct for 3-spined stickleback (gavtgAb) was assembled from transcripts ENSGACT00000012852 and ENSGACT00000012880 located between bp 12,491,853 – 12,515,240 in group VIII. To provide greater statistical support, the novel sequences from the labrid species were aligned with the genomic variants and other vertebrate taxa for which vtg sequences are known (see Additional file 1).

Phylogenetic reconstruction was performed using Bayesian (Mr Bayes 3.1.2: 4 × mcmc chains, 1,000,00 generations, sample frequency 100, burnin 3500; 54, maximum parsimony and neighbour joining (PAUP 4.0b10: 1,000 bootstraps; 55) analyses of the amino acid and codon alignments, and maximum likelihood analyses of the codon alignments as described by Finn & Kristoffersen 4.