The sample of B. taipingensis was collected from Dongguan offshore water, Guangdong, China. A piece of muscle tissue excised from the individual was preserved in 95% ethanol for DNA extraction. Total genomic DNA from muscle tissue was extracted with a standard phenol/chloroform procedure followed by ethanol precipitation and kept at 4 °C for subsequent use.

The complete B. taipingensis mitogenome was amplified using a long-PCR technique (Miya et al. 1999). Six sets of primers (Table 1) were designed based on multiple alignments of the conserved region of the complete mitochondrial DNA sequences of other Sciaenidae fishes: Larimichthys crocea (EU339149), Collichthys niveatus (JN678726), Collichthys lucida (JN857362), Larimichthys polyactis (FJ618559), Miichthys miiuy (HM447240) and Pennahia argentata (HQ890946), as well as previously determined, partial sequences of the 16S rRNA, Cyt b, COI genes and control region. Subsequent sequencing was accomplished by primer walking method. After the sequencing of these fragments, 31 normal PCR primer sets were designed using Premier 5.0 (Primer Biosoft International) to obtain contiguous, overlapping segments of the entire mitogenome. It was necessary that every two contiguous segments overlapped by at least 50 bp to ascertain the accuracy of sequencing.

All PCRs were performed in a Takara thermal cycler. Takara Ex-Taq and LA-Taq polymerase (Takara Biomedical) were used for normal and long-PCR reactions, respectively. Long-PCR reactions were carried out in 25 µl reaction mixture containing 15.25 µl of sterile distilled H₂O, 2.5 µl of LA-Buffer, 4 µl of dNTP, 1 µl of each primer (5 µM), 0.25 µl of LA-Taq polymerase (1 unit/µl, Takara), and 1 µl of DNA template. The long-PCR reactions consisted of an initial denaturing step at 94 °C for 2 min, followed by 30 cycles of denaturing at 94 °C for 30 s, annealing at about 57 °C for 3 min and a final extension at 72 °C for 15 min. The normal PCR was performed following the standard procedure. Negative controls were included in all PCR amplifications to confirm the absence of contaminants. PCR products were cleaned by adding 0.45 µl of Shrimp Alkaline Phosphatase (Biotech Pharmacon), 0.9 µl of Exonuclease I (GE Healthcare) and 1.65 µl of sterile distilled H₂O to 9 µL of PCR product and incubating at 37 °C for 30 min and 80 °C for 20 min. The purified product was then sequenced on ABI Prism 3730 (Applied Biosystems) from both strands with the same primers as those used for PCRs.

Sequence trace files were corrected and aligned with the DNAstar 5.0 software package (DNAstar, Inc.,Wisconsin, USA). The locations of 13 protein-coding genes and 2 rRNA genes were determined by their similarity to published mitogenomes of other Sciaenidae species as shown in Table 2, whereas the tRNA genes were identified using the program tRNAscan-SE 1.21 (Lowe et al. 1997). Some tRNA genes, e.g. tRNA-Ser (AGY) that could not be found by the tRNAscan-SE, were identified by their secondary structure and their position in the mitogenome (Zhang et al. 2009).

The structure of the control region and its conserves motifs were identified by making a comparison with homologous sequences of reported teleost (Lee et al. 1995; Cui et al. 2009; Cheng et al. 2010). The proposed secondary structure of the putative O_L was analyzed with the program Mfold v.3.2 with default setting (Zuker 2003) and visualized using RNAViz (De Rijk and De Wachter 1997).

To clarify the phylogenetic position of B. taipingensis within the family Sciaenidae, the complete mitogenome sequences of 9 fish species with 10 complete mitogenome sequences in Sciaenidae (Table 2) were incorporated together with the presently obtained mitogenome sequence of B. taipingensis for phylogenetic analysis. In addition, possible close outgroups in Percoidei (Table 2) were chosen to root phylogenetic trees (Boger and Kritsky 2003). Sequences were aligned using Clustal W (Thompson et al. 1994), and adjustments were made manually. Phylogenetic analyses were based on the concatenated sequences of 12 protein-coding genes and 2 rRNA. The ND6 gene was excluded because of its heterogeneous base composition and consistently poor performance in phylogenetic analysis (Miya et al. 2003). For protein-coding genes, all stop codons were excluded from the analysis. The possible bias of substitution saturation at each codon position of protein-coding genes and 2 rRNA genes was investigated using DAMBE v.4.5.57 (Xia et al. 2001), and the results suggested that the third codons position were saturated both for transitions and transversions in the plot against with pairwise sequence divergence. Finally, unambiguously aligned sequences were 3630, 3630, 2728 nucleotide positions from first and second codon position of 12 protein-coding genes, 2 rRNA genes, respectively, and thus a total of 9988 bp positions were utilized for phylogenetic analysis.

Two different methods, Bayesian inference (BI) and maximum likelihood (ML), were used to construct the phylogenetic tree. Three partitions (first and second codon positions of protein-coding genes, 2 rRNA genes) were set in the combined data set for partitioned Bayesian analyses using MrBayes 3.1.2 (Ronquist et al. 2003), which allowed different substitution models in individual partitions. Markov Chain Monte Carlo (MCMC) Bayesian analyses were undertaken with MrBayes 3.1.2 setting for the best-fit model of nucleotide evolution selected by Hierarchical Likelihood Ratio Tests (hLRTs) in MrModeltest version 2.3 (Posada et al. 2004). Four Markov chains (one cold and three heated) were used in each of two simultaneous runs starting from different random trees. Analyses were run for 1,000,000 generations, sampled every 100 generations to assess convergence. The distribution of log-likelihood scores was examined to determine stationarity for each search and to determine if extra runs were required to achieve convergence in log likelihoods searches. We discarded initial trees with non-stationary log-likelihood values as part of a burn-in procedure, and combined the remaining trees that resulted in convergent log-likelihood scores from both independent searches. These trees were used to construct a 50% majority rule consensus tree.

Maximum likelihood analysis (ML) was performed in PAUP 4.0 (Swofford 2000), and the GTR+I+G (I=0.45, G=0.88) model of DNA substitution for the analysis was assessed by Modeltest version 3.7 (Posada and Crandall, 1998). The ML analysis was performed with random sequence addition replicates. Heuristic search was undertaken using 10 random addition sequence starting trees and tree bisection reconnection (TBR) branch swapping. The confidence level (Felsenstein, 1985) at each branch was evaluated by performing bootstrapping (BP) with 100 replicates in ML analysis.

The complete mitogenome of B. taipingensis was sequenced to be 16500 bp which consisted of 13 typical vertebrate protein-coding genes, 22 tRNA genes, 2 rRNA genes, and 1 putative control region (CR, Table 3). It had been submitted to GenBank with accession number JX232404. The mitogenome of B. taipingensis had substantially similar patterns on mitogenome structural organization with other vertebrates (Anderson et al. 1981; Miya et al. 1999; Cui et al. 2009). The encoding genes of mitogenome were located on H-strand with the exception of ND6 and 8 tRNA genes that were transcribed from L-strand (Table 3). All genes from B. taipingensis mitogenome were similar in size to most Perciformes species (Kim et al. 2004; Mabuchi et al. 2007; Cui et al. 2009; Cheng et al. 2011a; Cheng et al. 2012a) and the presence length of control region assumed variation in size, because they were prone to undergo the insertion/deletion events in the sequences (Sbisa et al. 1997).

The overall base composition of the B. taipingensis mitogenome was estimated to be 28.2% for A, 31.1% for C, 16.2% for G, and 24.6% for T (Table 4), respectively, indicating an obvious antiguanine bias. Furthermore, the G content of all protein-coding genes presents obviously lower just as found in other bony fishes (Miya et al. 2003; Mabuchi et al. 2007). The most remarkable character of metazoan mitogenomes is the strand-specific bias in nucleotide composition (Reyes et al. 1998; Hassanin et al. 2005), which can be measured as GC-skew (G%-C%)/(G%+C%) and AT-skew (A%-T%)/(A%+T%), respectively (Perna et al. 1995). The overall GC- and AT-skews of the H-strand of B. taipingensis mitogenome were -0.328 and 0.047, respectively, indicating a strand compositional bias characterized by a strong excess of C over G nucleotides and a slight excess of A over T nucleotides on the H-strand.

The B. taipingensis genome contained 13 protein-coding genes encoded on the H-strand excluding ND6 gene that was oriented to L-strand. The 13 protein-coding genes were total 11,436 bp in size, accounting for 69.15% of the whole mitogenome. All protein-coding genes initiated with an ATG codon, just as in most vertebrates. Three open reading frames (ATP8, ND4L and ND6) of B. taipingensis ended with TAA, two open reading frames (ND1 and ND5) with TAG, and one open reading frames (COI) with AGA. The remainder used incomplete stop codons, either TA (ND2, ATP6 and COIII) or T (COII, ND3, ND4 and Cytb), probably completed by post-transcriptional polyadenylation (Ojala et al. 1981). It should be noted that these genes (ND4L with ND4, ATP8 with ATP6 and COI with tRNA^Ser(UUR)) could complete their stopped codons within the overlapping portion of the next genes.

Nucleotide composition and codon using frequencies were calculated from a concatenated sequence of all protein-coding genes on the H-strand, except for ND6 on the L-strand. The base composition of protein-coding genes revealed weak bias against G (14.4%), especially at third codon positions (10.8%, Table 4). For all protein genes, C was the most frequent nucleotide at the first and third positions whereas T was most frequent at the second position as found in other bony fishes (Oh et al. 2007).

Like other mitochondrial genomes (Zardoya et al. 1995; Inoue et al. 2000), twenty-two tRNA genes were identified. The tRNA genes were interspersed among the mitochondrial genome and ranged in size from 66 to 74 bp (Table 3). They showed the typical gene arrangement as found in most vertebrates. Fourteen tRNA genes were transcribed on the H-strand, whereas the remaining eight tRNA genes were oriented on the L-strand (Table 3). These tRNA genes were predicted capable of folding into typical cloverleaf secondary structures with normal base pairing. The B. taipingensis mitogenome also contained a small subunit of rRNA (12S rRNA) and a large subunit of rRNA (16S rRNA) as in other bony fishes (Zardoya et al. 1995; Inoue et al. 2000), which were 947 bp and 1684 bp in length, respectively. As in other vertebrate genomes, these genes were located between the tRNA^Phe and tRNA^Val genes and between tRNA^Val and tRNA^Leu(UUR) genes, respectively.

As shown in Table 3, there were non-coding intergenic spacers from 1 to 8 bp observed in B. taipingensis, spanning the contiguous genes apart from O_L and control region. Furthermore, mitochondrial intergenic spacers were a total of 36 bp in eleven different locations.

As in most vertebrates, the major non-coding region in B. taipingensis mitochondrial genome was located between tRNA-Pro and tRNA-Phe. It was determined to be 826 bp in length, longer than other reported Sciaenidae species, and it had an overall base composition that was rich in A and T (A+T=62.0%). By comparing with the recognition sites in some reported fishes (Lee et al. 1995; Cui et al. 2009; Cheng et al. 2010; Cheng et al. 2011a; Cheng et al. 2012a), three domains were detected in B. taipingensis, namely, the termination associated sequence domain (ETAS), the central conserved sequence block domain (CSB-D, CSB-E and CSB-F) and the conserved sequence block domain (CSB-1, CSB-2 and CSB-3) (Figure 1). The ETAS was thought to act as a signal for the termination of H-strand elongation (Clayton 1991), and this domain was a hypervariable domain that might be useful for population genetic analyses. Furthermore, the motif sequence of ETAS was TACATAT with one palindromic sequence ATGTATA. The control region of mammals contained five blocks (CSB-B, CSB-C, CSB-D, CSB-E and CSB-F) in central conserved sequence blocks, however, only CSB-D, CSB-E and CSB-F were mostly detected in fishes (Brought et al. 1994; Lee et al. 1995). In this study, all these three motifs were identified in the central domain in accordance with M. miiuy (Cheng et al. 2010), and Nibea albiflora (Cheng et al. 2011b) within Sciaenidae, which was not detected in four other species of Pseudosciaeniae (Cui et al. 2009; Cheng et al. 2011a; Cheng et al. 2012a; Cheng et al. 2012b). In addition, the consensus sequence of CSB-F was ATGTAATAAGAACCGACCAT, which distinguished the central conserved sequence block domain from the termination associated sequence domain. CSB-E was located downstream of CSB-F, whose consensus sequence was AGGGACAAGTATTGTGGGGG, characterized by the box GTGGGG. CSB-E was followed by CSB-D with its consensus sequence TATTCCTGGCATTTGGT. Generally, these key sequences were highly conserved and easily recognized. Three conserved sequence blocks (CSB-1, CSB-2 and CSB-3) were determined in the conserved sequence block domain which was thought to be involved in positioning RNA polymerase both for transcription and for priming replication (Shadel et al. 1997). Moreover, the critical central conserved sequences of CSB-1, CSB-2, and CSB-3 were ATTTGGATATCAAGTGCATAAA, ACCCCCCCTACCCCCCC, and AAACCCCCCGTAAA, respectively.

Figure 1. 

The structure of control region about B. taipingensis.

The additional non-coding region, the putative origin of L-strand replication (O_L), was located in a cluster of five tRNA genes (the WANCY region) between the tRNA-Asn and tRNA-Cys genes, almost identical with other Sciaenidae fishes. The putative O_L could form a stable stem-loop secondary structure with 20 bp in the stem and 13 bp in the loop (Figure 2), which was 37 bp in length (CCTTTCCCCCGCCTACTATAGGACTAAAGGCGGGGGA). Furthermore, the conserved stem-loop structures in mitochondrial genomes was thought to play an importance role in conjunction with the origin of mtDNA replication.

Figure 2. 

Potential secondary structure of the origin of L-strand replication (O_L) of B. taipingensismtDNA.

The phylogenetic trees (the 50% majority-rule consensus tree is shown in Figure 3) were highly coincident regardless of the analytic method used, and were statistically supported by high posterior probability and intermediate bootstrap values. This phylogenetic analysis represented the first investigation of relationships of B. taipingensis within the Sciaenidae based on the whole mitogenome. In our analysis, B. taipingensiswas found to be more closely related to Pseudosciaeniae (Collichthys, Larimichthys and Miichthys) than to Pennahia and Nibea, the latter of which was suggested by morphological topology (Zhu et al. 1963; Cheng et al. 1987) and previous molecular study (He et al. 2012). However, phylogenetic analyses showed that Miichthys could not be merged into the Collichthys–Larimichthys clade. On the contrary, Miichthys and Bahaba formed an independent clade well supported by high posterior probability value, and this clade formed the sister group of the Collichthys–Larimichthys clade. Therefore, the relationship between Miichthys and Pseudosciaeniae deservesd to be further studied. The proposed phylogenetic position of B. taipingensis within the Sciaenidae based on the findings of the present study should be accepted with caution due to limited taxon sampling. However, the phylogenetic relationship within the Sciaenidae remains to be resolved, and it is necessary to make further analysis based on more molecular information and extensive taxon sampling.

Figure 3. 

Phylogenetic relationships among Sciaenidae species based on the combined 9988 bp nucleotide positions. The posterior probability value of BI analyses and bootstrap support values of ML analyses (in the order: BI, ML) are indicated near the branches.