The vertebrate makorin ubiquitin ligase gene family has been shaped by large-scale duplication and retroposition from an ancestral gonad-specific, maternal-effect gene

Mkrn4 is a new member of the vertebrate makorin gene family

In order to identify new potential gonad-specific transcription factors or regulatory proteins expressed in fish gonads, we randomly sequenced clones from a testis cDNA library of the platyfish Xiphophorus maculatus. In silico screening for sequences containing zinc finger domains revealed a clone with similarity to uncharacterized members of the makorin gene family [20] (Genbank accession number HQ377193). Since three functional mkrn genes have been described so far in vertebrates, the newly identified gene was called makorin4 (mkrn4).

We could identify sequences orthologous to platyfish mkrn4 in other fish species (Figures 1 and 2). All fully sequenced fish genomes including zebrafish (Danio rerio), medaka (Oryzias latipes[21], green spotted pufferfish (Tetraodon nigroviridis),[22], fugu (Takifugu rubripes) [23] and three-spined stickleback (Gasterosteus aculeatus) contained mkrn4 as a single copy gene. Mkrn4 ESTs were also identified in databases for other fish species without available genome draft. Mkrn4 is present in birds (on Gga26 in chicken) and amphibians as well as in most mammals, including human (Hsa6q14.1, where it was annotated as a pseudogene, see below). Mkrn4 sequences form a clear monophyletic group within the Makorin molecular phylogeny (Figure 2).

Surprisingly, analysis of genome sequence drafts and EST data failed to detect mkrn4 in two murid rodents, the mouse Mus musculus and rat Rattus norvegicus, even after targeted scrutiny of chromosomal segments syntenic to mkrn4-containing loci from other species (on Mmu17 in mouse and Rno20 in rat). Partial mkrn4 sequences were found in other rodents such as Ord's kangaroo rat (Dipodomys ordii), thirteen-lined ground squirrel (Spermophilus tridecemlineatus) and Guinea pig (Cavia porcellus; for this species a deduced sequence could be included into the Mkrn phylogeny), as well as in the American pika Ochotona princeps (lagomorph). No mkrn4-related sequences were identified in the genome draft of another lagomorph, the rabbit Oryctolagus cuniculus, and in the northern tree shrew Tupaia belangeri (order Scandentia). For all these species, no mkrn4 EST sequence was present in databases. Hence, these results suggest that the mkrn4 gene has been lost in murid rodents (mouse and rat). In some other rodent or lagomorph species, mkrn4 may be also on the way of degradation (as suggested by long branch for Guinea pig sequence; Figure 2) or even might have been already eliminated, but genomic sequences of better quality will be required to confirm this hypothesis.

We noticed that mkrn4 was annotated as a pseudogene in the current Ensembl human genome release, under the name of Mkrnp2[24]. Very similar orthologous sequences were identified in syntenic regions of the genome of chimpanzee (chr. 6, 99% nucleotide identity with the human sequence in the coding region), orangutan (chr. 6, 99% nucleotide identity) and macaque (chr. 4, 96% nucleotide identity), a degree of conservation surprising for a pseudogene evolving under neutral selection. Only gorilla did not show any mkrn4 sequence, despite the presence of a large syntenic region on chromosome 6 including several genes surrounding mkrn4 in other primates. However, the draft sequence of this region contains many gaps, especially at the position where mkrn4/mkrnp2 would be expected. Hence, mkrn4 might be absent from draft but present in the genome of the gorilla.

In human, mkrnp2 was annotated as pseudogene due to a premature stop codon at position 53 in the Ensembl transcript. We suggest that the translation start codon was wrongly annotated in the Ensembl prediction. A start codon in frame +3, using the ATG at position 54 of the predicted transcript, leads to a full length protein without any further interruption of the open reading frame. This ATG is conserved between primate sequences and annotated as the bona fide start codon in GenBank entry EAX03829.1 used in sequence alignments and molecular phylogenies (Figures 1 and 2). Primate sequences were reannotated accordingly (Figure 1).

Using primate sequences as query against EST databases in BLAST analysis, no evidence of transcription could be found. We then compared the rate of non-synonymous (Ka) to the rate of synonymous substitutions (Ks) between mkrn sequences in primates (human, chimpanzee, orangutan and macaque) and other vertebrates. The average Ka/Ks value obtained between primate mkrn4 sequences was 0.12, indicating an excess of synonymous vs. non-synonymous substitutions. This value was similar to those obtained between primate mkrn1 (0.16) and mkrn2 (0.11). The average Ka/Ks ratio for mkrn4 between fish sequences (medaka, platyfish and guppy) was 0.27. Taken together, these observations suggest that mkrn4 evolved under purifying selection in primates and fish, as observed for other makorin genes. Mkrn4 might be a functional protein-coding gene in primates but expressed at a very low level, in a particular type of cells or under very specific conditions. Alternatively, mkrn4 might have lost its expression potential after initial functionality in different primate species.

No mkrn4 sequence was found in the genome of more divergent species, including the elephant shark Callorhinchus milii (Chondrichthyes aka cartilaginous fish, a divergent jawed vertebrate) [25] and the invertebrate chordates Branchiostoma floridae (amphioxus) and Ciona intestinalis/savignyi (sea squirts).

Large-scale duplication and the evolution of the makorin gene family

Mkrn1, mkrn2 and mkrn4 are all present in both ray-finned fish and tetrapod lineages, indicating that these genes have been formed through gene duplications at least 450 million years ago early in the vertebrate lineage. Mkrn1 and mkrn2 but not mkrn4 were detected in the genome of elephant shark Callorhinchus milii, a cartilaginous fish. In contrast, we could identify only one makorin gene in amphioxus and sea squirts (non-vertebrate chordates), as well as in other more divergent non-vertebrate species including yeast and rice. In the fly Drosophila melanogaster, four makorin genes were identified, but three of them clearly corresponded to retrocopies of the only bona fide intron-containing mkrn gene. Hence, the mkrn gene family observed in vertebrates is probably the result of two events of gene duplication having taken place between the emergence of vertebrates and the divergence between tetrapods and ray-finned fish. At least the duplication having led to mkrn1 and mkrn2 occurred before the split between cartilaginous and bony fish.

In order to know if gene duplications at the origin of the mkrn family might be linked to rounds of whole genome duplications having occurred at the basis of the vertebrate lineage [26–29], we tested if mkrn genes might be included in large ancient duplicated regions called paralogons having resulted from such events.

On human chromosome 3, mkrn2 and the raf1 proto-oncogene partially overlap in antisense transcriptional orientation. This syntenic relationship is conserved in most vertebrates but not in teleost fish, with the exception of the zebrafish, where mkrn2 and raf1 are closely linked on chromosome 11 (Figure 3). Strikingly, in medaka raf1 is located close to mkrn4 but not mkrn2. We further analyzed syntenic relationships between genomic regions containing mkrn2 and mkrn4 in human, chicken, medaka and zebrafish genomes (Figure 3). We could identify a set of genes with duplicated copies linked to mkrn2 and mkrn4 in at least one species (for example ppard/pparg in human, chicken and medaka). Other genes like raf1 or cnpb showed only one copy but either linked to mkrn2 in some species or to mkrn4 in others, suggesting differential loss of paralogous copies after duplication.

These results indicate that mkrn2 and mkrn4 are parts of ancient large duplicated regions present in distant vertebrate species. Hence, these paralogons might have been generated by one of the whole genome duplications having occurred in the ancestral vertebrate lineage, the so called 1R and 2R duplications [26–29]. Both mkrn2 and mkrn4 are called ohnologs (paralogous sequences resulting from whole genome duplications). On the other hand, we can not exclude the occurrence of a very large segmental gene duplication having generated these duplicated blocks.

The region containing mkrn1 shows little conservation between mammals and fish (data not shown). We could not find any evidence for synteny between the mkrn1-containing region and other mkrn regions. It has been previously proposed that mkrn1 and mkrn2 are indeed the result of a single gene duplication event [17].

Retroposition and the evolution of the makorin gene family

In addition to mkrn1, mkrn2 and mkrn4, multiple other mkrn-like sequences are present in mammalian genomes. Some of these sequences correspond to intronless retrocopies of mkrn1. Their functional status concerning expression, regulation and mutual interference has been the source of controversial debate [16, 30].

We detected makorin3 and other retrocopies of makorin1 only in mammals and not in chicken, frog and all five fish genomes analyzed. An mkrn3 sequence was detected in opossum, suggesting an event of retroposition that took place before the split of marsupials from therian mammals. No mkrn3 was found in the current genome draft of the platypus.

Retrocopies of mkrn2 were found in the mouse but neither in the human nor in non-mammalian vertebrate genomes (Table 1). These three retropseudogenic copies of mkrn2 are located on three different chromosomes in M. musculus. No retrocopy of mkrn4 was found in all vertebrate genomes analysed, but three potentially functional retrocopies of the single intron-containing mkrn gene were identified in the fruit fly Drosophila melanogaster. Taken together, these observations demonstrate an intriguing tendency of the mkrn gene family to generate duplicated copies by retroposition.

Structural evolutionary dynamics in the Makorin family

The evolution of the nature and number of zinc fingers in the Makorin protein family was investigated (Figure 4). In most plants and invertebrate species (but not in Drosophila), the single Makorin protein contains the RING domain flanked by three C3H zinc fingers on its N-terminal side and one C3H zinc finger on its C-terminal side. An additional motif rich in cysteine and histidine residues (the CysHis motif) is located between the third C3H zinc finger and the RING domain (Figures 1 and 4). This structure, likely to be ancestral in vertebrates, is also found in most, if not all Mkrn1 and Mkrn2 proteins.

In contrast, this otherwise well conserved structure was not found in Mkrn4 proteins (Figures 1 and 4). While the RING domain and the C-terminal C3H zinc finger are found in all Mkrn4 sequences, the third C3H zinc finger and the CysHis motif are missing and have been probably lost in the ancestral mkrn4 gene early after duplication.

Only fish and Xenopus have kept the first two C3H complete domains, which are both absent from mammalian sequences. In birds, a remnant of one of these domains has been retained in chicken and zebra finch under the form of a C2H domain, the second C3H domain having been maintained only in zebra finch but not in chicken.

These results indicate that the evolution of the Mkrn4 protein is associated with the dynamic loss of zinc finger domains otherwise conserved in Mkrn proteins. This is consistent with a to some extent faster evolution also reflected by the longer branches in the Makorin-like group in the phylogenetic tree (Figure 2). Another example of zinc finger domain loss is visible in mkrn3: this mkrn1-derived retrogene encodes a protein having lost the second C3H zinc finger (Figures 1 and 3) in human and mouse.

In the much conserved Mkrn1 and Mkrn2 groups, the platypus sequences present a reduced number of zinc fingers, like the opossum Mkrn2 sequence. This might reflect a biological reality, but might also be an artifact of local poor sequence assembly quality of the genome drafts for these species.

Evolution of makorin expression in vertebrates

We compared the expression of the different vertebrate makorin genes in different species representative of the fish (medaka and zebrafish), amphibian (the African clawed frog Xenopus laevis) and bird (chicken) lineages (published and EST data are generally available for mammals) (Figures 5 and 6). This might allow to assess the functional evolution of mkrn genes in vertebrates and to estimate the ancestral expression pattern and function in the primitive vertebrate before gene duplication.

In adult tissue, the whole mkrn gene family showed a clear gonad-biased expression in fish and frog, with particular high expression in ovary (Figure 5). Accordingly, mkrn1 is upregulated during ovarian maturation in flatfish, with strong expression in previtellogenic ovarian follicles [32]. Such a specificity of expression was not observed in birds and mammals, even if mkrn1 has been reported to be strongly expressed in mouse adult testis [4] and mkrn4 showed a slightly testis-biased expression in chicken.

Taken together, these results suggested that the ancestral mkrn gene was preferentially expressed in the gonads in a primitive vertebrate, with conservation of expression pattern after duplication in fish and amphibians but partial relaxation of this specificity and possible extension of functions in birds and mammals.

Mkrn expression during early embryonic stages in fish largely reflected maternally stored mRNA expressed during oogenesis [33, 34], since mRNA levels were very high at very early stages (Figure 6). RNA levels then decreased throughout development, probably due to dilution effect through continuous cell division, and reached a very low level shortly after midblastula transition (MBT), the state of zygotic transcription activation. MBT takes place around stage 11 in medaka [35] and starts at cycle 10 corresponding to 512 cell stage in zebrafish [36] (between 128 cell stage and dome in our experiments). Maternal expression was observed for all three mkrn genes in both medaka and zebrafish, suggesting here again an ancestral expression pattern. As observed in adults, no tissue-specific expression was observed in chicken embryos one day before hatching (Figure 6).

Fish

Platyfish (Xiphophorus maculatus, population Rio Jamapa WLC1274, closed colony stock derived from Jp163A), zebrafish (Danio rerio, strain AB/TU) and medaka (Oryzias latipes, strain Hd-rR) were kept under standard conditions at the PRECI aquarium facility of the IFR128 Biosciences Gerland-Lyon Sud (Lyon, France). Fish embryos were raised to the required stages of development in E3 embryo media [52] at 26°C for medaka and 28.5°C for zebrafish. Medaka and zebrafish embryos were staged according to Iwamatsu [53] and Kimmel et al.[54], respectively. Chicken samples (embryonic chicken, one day before hatching and male adult tissue) were provided by B. Pain (Institut de Génomique Fonctionnelle de Lyon, ENS de Lyon, France). Adult Xenopus laevis samples were gifts from F. Vollmar (Department of Cell and Developmental Biology, University of Würzburg, Germany) and C. Schultheis (Physiological Chemistry I, University of Würzburg, Germany). Experimental research reported here has been performed following internationally recognized guidelines (DDSV, Direction Départementale des Services Vétérinaires de Lyon, agreement number A 69 387 0602).

In silico analyses

Protein and nucleotide sequences of mkrn genes were identified using BLAST/BLAT searches against GenBank [55], the current Ensembl genome assemblies ([24] version 57, March 2010) of human (Homo sapiens), chimpanzee (Pan troglodytes), gorilla (Gorilla gorilla), orangutan (Pongo pygmaeus), macaque (Macaca mulatta), mouse lemur (Microcebus murinus), mouse (Mus musculus), guinea pig (Cavia porcellus), horse (Equus caballus), rabbit (Oryctolagus cuniculus), cow (Bos taurus), opossum (Monodelphis domestica), wallaby (Macropus eugenii), platypus (Ornithorhynchus anatinus), chicken (Gallus gallus), clawed frog (Xenopus tropicalis), zebrafish (Danio rerio), medaka (Oryzias latipes), green spotted pufferfish (Tetraodon nigroviridis), fugu (Takifugu rubripes), three-spined stickleback (Gasterosteus aculeatus), sea squirts (Ciona intestinalis and Ciona savignyi), the amphioxus genome (Branchiostoma floridae) [56] and the elephant shark genome (Callorhinchus milii) [57]. Automatic genome annotation was refined manually using EST data and related protein sequences with the help of FGENESH [58].

Nucleotide and protein sequences were loaded into BioEdit [59] and aligned using MUSCLE [60, 61] and CLUSTALW2 [62]. Alignments were checked manually and ambiguously aligned regions were removed before phylogenetic analysis. Phylogenetic trees were calculated based on a 131 amino acid alignment including the RING domain and the downstream adjacent C3H zinc finger. Maximum likelihood reconstruction was done using PhyML 3.0 [63, 64], with 1000 bootstrap replicates under the LG model. Similar results were obtained using the neighbor-joining method as implemented in MEGA4 [65] with bootstrap values of 10.000 replicates. MEGA4 was also used to visualize phylogenetic trees. To establish syntenic relationships between genomes we used Genomicus [66], Synteny Database [67] and information provided by Ensembl Gene Trees [24]. Annotation of conserved protein motifs was performed with Motif Scan [68] and NCBI conserved domains [12].

Ka/Ks calculation was peformed using KaKsCalculator under the MYN model [69] on pairwise nucleotide alignments done with LASTZ [70] and MUSCLE [60, 61]. Indicated Ka/Ks values are averages of values obtained for all possible pairwise comparisons.

RT-qPCR expression analysis

Total RNA was extracted from separated male and female tissues (brain, eyes, gills, heart, intestine, kidneys, lung, liver, muscle, ovary, skin, spleen and testis) and from whole embryos using the TRIReagent (Molecular Research Center, Inc.). After DNase treatment, reverse transcription was performed starting from 1 μg of total RNA using the RevertAid First Strand Synthesis kit (Fermentas) and random hexamer primers. Real-time PCR was done on 2 μl of dilutional series of cDNA starting from a dilution of factor 20 using IQTM Custom SYBR^® Green Supermix (Bio-Rad). PCR amplification was monitored with a CFX96 RealtimeSystem (Bio-Rad). After one incubation step at 95°C for 3 min, the thermal cycling protocol was as follows: 95°C for 10 seconds, 55-62°C (primer-dependent annealing temperature) for 20 seconds and 72°C for 15 seconds during 40 cycles. Primers were designed manually and ordered from Invitrogen (Additional file 3). For each primer set, the efficiency of the PCR reaction was measured in duplicate on serial dilutions (factor 10) of the cDNA sample. Real-time PCR efficiencies (E) for each reaction were calculated from the slope of the standard curve using the equation E = 10^(-1/slope)-1 as implemented in the BioRad CFX manager V1.5 software, with 100% efficiency as an indicator of a robust assay. All results reported here are averages of two independent reverse transcription reactions, which have been each tested with two PCR reactions. For quantification, data were analyzed by the 2-DDCT method [71], by normalizing to the housekeeping gene rpl7 (ribosomal protein l7). Expression was normalized to eye tissue (fish) or heart (due to lack of eye samples for female frog and chicken).