Complete Mitogenome of Threespot Flounder Grammatobothus polyophthalmus (Pleuronectiformes: Bothidae) and Study on the Mechanism of Gene Rearrangement in 13 Bothids CURRENT STATUS:

Background: The mitochondrial genomes (mitogenomes) of 12 bothids (Pleuronectiformes) from eight genera have been obtained. From the data, the genomic-scale and various gene rearrangements revealed the high diversity of variation in these mitogenomes. Results: A total of 18170 bp of Grammatobothus polyophthalmus mitogenome was determined including 37 genes and two control regions (CRs). Genes encoded by L-strand were grouped to an eight-genes cluster (Q-A-C-Y-S1-ND6-E-P) except for the tRNA-N, other genes encoded by H-strand were grouped together (F-12S … CytB-T) except for the tRNA-D that was translocated to inside of the eight-genes cluster. The mitogenome of G. polyophthalmus and that of 12 known bothids possessed the similar genomic-scale rearrangements with the only differences in the various combinations of CR, tRNA-D and eight-genes cluster, and the shuffling of tRNA-V. Based on the structure character of all 13 bothid mitogenomes, the Dimer-Mitogenome and Non-Random Loss (DMNR) model was fitted to account for all these rearrangements. And the translocation of tRNA-D occurring after the DMNR process in 10 of 13 bothid mitogenomes was confirmed. The striking finding was that each of degenerated genes existing in the gene rearrangement process in 13 bothids had their counterparts of intergenic spaces. Conclusions: The result of corresponding relationship between degenerated genes and intergenic spaces provided the significant evidence to support the possibility of the DMNR model, as well as, the existing of dimeric mitogenome in mitochondrion. The findings of this study were rare phenomenona in teleost fish, which not only promoted the understanding of mitogenome structural diversity, but also shed light on studying of mitochondrial rearrangement and replication.

Among these genes, most are encoded on the heavy strand (H-strand), only ND6 and eight tRNA genes (N, Q, A, C, Y, S 1 , E and P) are encoded on the light strand (L-strand).
Additionally, two main non-coding regions are also existed, including the origin of replication of L-strand (O L ) as well as the control region (CR) where both of the replication origin of H-strand (O H ) and the transcriptional initiation of two strands are located [1][2][3].
Three types of gene rearrangement were observed in mitogenome of animals, including shuffling, translocation and inversion [4][5][6][7]. Before the gene inversion in tongue fish was firstly discovered [8], only the first two of above gene rearrangement types were reported in fishes [9][10][11]. Since then, an increasing number of rearranged mitogenomes of flatfishes featuring three rearrangement types have been found [12][13][14][15][16]. Among them, one representative case was mitochondrial rearrangement of blue flounder Crossorhombus azureus [14]. In this mitogenome, genes were grouped with identical transcriptional polarities, including eight genes on L-strand grouped to a cluster (8-cluster, Q-A-C-Y-S 1 -ND6-E-P) except for the tRNA-N, and other genes (F-12S … CytB -T) on H-strand were grouped together. Particularly, the order of these genes grouped in each strand was maintained as that in non-rearranged mitogenome of fish, except for the site of tRNA-D.Furthermore, unlike the typical position of CR in fish, CR of this species located between tRNA-D and tRNA -Q, which separated the genes on H-strand and L-strand.
How did this particular order of mitogenome emerge? Four mechanisms had been proposed to account for the gene rearrangements of mitogenome, including duplicationrandom loss [17], tRNA miss-priming model [18], intramitochondrial recombination [19] and duplication-nonrandom loss [20]. However, none of these four mechanisms could explain the rearrangement case occurred in C . azureus well. Therefore, a novel mechanism of Dimer-Mitogenome and Non-Random Loss (DMNR) was put forward to specially account for this rearrangement [14]. The inferred DMNR process started from the mitogenome with ancestral gene order in typical fish (Fig. 1A). First, the dimerized event of two-monomers mitogenomes accidently occurred and formed a functionally dimeric molecule linked headto-tail ( Fig. 1I-C). At this stage, the transcription of genes on dimeric mitochondrial DNA (mtDNA) could be normally initiated by the H-strand promoters (HSP and HSP ′ ) and the Lstrand promoters (LSP andLSP ′ ) in two CRs. The transcription on H-strand would terminate at TAS and TAS ′ in CRs, and that on L-strand particularly at tRNA-L 1 and tRNA-L 1 ′ .
Subsequently, the function of promoters (assumed to be LSPand HSP) in one CR was lost, and the genes controlled by them could not be transcribed, and then these genes degenerated as non-coding sequences or even disappearance. Whereas, the function of promoters (assumed to be LSP ′ and HSP ′ ) in the other CR still worked, the genes controlled by them could be transcribed, and formed the final gene order in the mitogenome of C. azureus ( Fig. 1I-D and I-E).
During this process, the only exception is the transcription of tRNA-N and tRNA-N′. The tRNA-N would not be transcribed but be transcribed and retained, and the tRNA-N′ would be transcribed but not be transcribed and then lost. The reason for this exception is the retained tRNA -N related to the structure and function of O L . The O L is usually located between tRNA -N and tRNA -C of WANCY region that is form by tRNA cluster of tRNA-W, tRNA-A, tRNA-N, tRNA-C and tRNA -Y [21]. Because tRNA-C and tRNA -Y were rearranged, only a 7-bp intergenic space was left between tRNA-N and COI in which failed to form the second structure of the O L . But 26-bp middle portion of tRNA-N could form an O L -like structure ( Fig. 2A), therefore, tRNA-N was inferred to act as the function of O L during L-strand replication [14,22]. Additionally, one unsure event occurred in C. azureus mitogenome is when did the tRNA-D translocate from a site between tRNA-S 1 and COII to between tRNA-T and CR? Based on the feature of mitogenome and process of rearrangement, Shi speculated that the translocation of tRNA-D could occur either before or after the DMNR process as shown in Fig. 1I-B and I-E [14].
So far, mitochondrial genomes of 12 bothid species from eight genera were obtained.
Comparing these genomes, the gene orders among all genomes are same, except for the azureus translocated to the outside of eight-genes cluster, that of B. myriaster was translocated to the inside of eight-genes cluster ( Fig. 1II-B). Whether this translocation occurred before or after the DMNR process was not mentioned [12].
From the data, the genomic-scale and various gene rearrangements revealed the high diversity of variation in 12 bothid mitogenomes. Therefore, to better understand the character of mitochondrial structure of bothids, the mitogenome of threespot flounder Grammatobothus polyophthalmus was determined in this study. This species is one of few bothids featuring one lateral line on both sides of the body. What are mitogenomic characteristics of this species; and whether did rearrangement occur in this mitogenome, if so, what is the rearrangement type? All these questions were addressed in this study, and the result would reveal more mitochondrial diversity in Bothidae, and provide scientific foundation for further research in mitochondrial rearrangement of fish.

Results
Organization and gene rearrangement of the G. polyophthalmus mitogenome A total of 18170 bp in length of G. polyophthalmus mitogenome contained 37 genes, including 13 protein-coding genes, two rRNA genes and 22 tRNA genes. Among these genes, 28 genes were encoded on the H-strand, others of ND6 and eight tRNA genes (N, Q, A, C, Y, S 1 , E and P) on the L-strand (Additional file 2: Table S2 and Additional file 3: Figure S1). In this mitogenome, after tRNA-C and tRNA -Y were rearranged, a 40-bp Comparing with the mitogenomes of 12 bothid species, the gene order of G.
polyophthalmus was identical to that of four of these species, A. tenuis, L . gallus, L.
lanceolata, and P. iijimae. How did genomic-scale rearrangement and duplicated CRs generate in these five species? Although the mitogenomes of above four bothids have been reported for years, the mechanism of their gene rearrangement remains unaddressed. Here, using the G. polyophthalmus mitogenome as a representative, the generated process of gene rearrangement was conjectured via DMNR model based on the similar gene orders with that of C. azureus and B. myriaster mitogenomes.
This process began from the typical mitogenome of fish (Fig. 1A). Firstly, the dimerized event occurred and formed a functionally dimeric mtDNA ( Fig. 1III-C). And then the function of the promoters (assumed to be LSP and HSP) in one of CRs was lost, thus the genes controlled by the disabled promoters of LSP and HSP could not be transcribed and then degenerated as non-coding sequences or even disappearance ( Fig. 1III-D). firstly translocated, it would be between ND5 and ND6 of dimeric mtDNA (Fig. 1II-B′), and then, tRNA-D will be outside the cluster between ND5 and CytB after the DMNR process ( Fig. 1II-F′). To the location inside the cluster, this geneneeded one more translocation ( Fig. 1II-B). Whereas, if the translocation of the tRNA-D occurred after the DMNR process ( Fig. 1II-B), only this step needed to accomplish the final structure. Therefore, the parsimonious way for the translocation of tRNA-D inside the cluster is this translocation occurring after, not before, the DMNR process.
The exception of tRNA-N and tRNA-N′ in DMNR processes and Pelotretis flavilatus (Fig. 4). So many different intergenic spaces raised our interesting to verify do such spaces also occur in other 12 bothid species? The result showed that the total number of the unique or longer spaces in 12 bothids was 121, including most of them with the length from 2 bp to 88 bp, and other six (located at No. 12 intergenic space) from 155 to 511 bp (Fig. 4). Where did the unique or longer spaces come from? And what significance do these mean in rearranged mitogenome? Tracing the DMNR process in each of 13 species, several degenerated genes existed. A striking finding was each of degenerated genes could find their counterparts that were intergenic spaces with shorter length at same location ( Fig. 4A and B). This result confirmed that these spaces were traces for remaining of degenerated genes. The corresponding relationship between the degenerated genes and the intergenic spaces provided the direct evidence to support the possibility of inferred DMNR model, as well as, the existing of dimeric mitogenome in mitochondrion.
In addition, further analyses showed that both degenerated genes and its corresponding intergenic spaces are evolutionarily diverse. The number of unique or longer intergenic spaces in each of 13 bothid species ranged from 5 to 11, which indicated that the evolution states of degenerated genes were the progressive degeneration and complete disappearance. As well as, the number of intergenic spaces corresponding to each of degenerated genes varied widely from 1 to 13, which meant these un-transcribed genes were degenerated at different rate in different species. Such as the degenerated genes . Sequenced fragments were assembled to a complete mitochondrial genome by using CodonCode Aligner v3 and BioEdit v7 [24]. For large fragments and walking sequences, regular manual examinations were made to ensure reliable assembly of the mitogenome.
The complete sequence of G. polyophthalmus mitogenome was submitted to GenBank under the accession number MK770643.

Sequence analysis
Identification and annotation of protein coding genes and rRNA genes were performed by using NCBI-BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi). The tRNA genes and their secondary structures were determined by using tRNAscan-SE 1.21 [25], setting the cut-off values to 1 when necessary. The secondary structure of O L was identified by using the mfold web server (http://unafold.rna.albany.edu/?q=mfold). The gene map of G.
polyophthalmus mitogenome was generated by using CGView [26]. Mitogenomes of eight out of 12 bothid fishes used in this study were determined in our laboratory, including A.

Consent for publication
Not applicable.
Availability of data and materials

Competing interests
The authors declare that they have no competing interests.