Introduction

Rock carp, Procypris rabaudi (Tchang), is an endemic species to China and mainly distributed in upper reaches of the Yangtze River drainage [1]. The fish prefers to inhabit slow flowing deep water with plenty of rocks at bottom, and lives through winter in the holes of rocks at bottom of deep pool or bed of river. As omnivorous fish, it mainly feeds on zoobenthos, such as aquatic insect, Limnoperna, aquatic Oligochaeta, and etc., and secondly on oddment of plant and phytoplankton [1, 2]. It is an important commercial fish due to its good taste, rich nutrition, and potential in aquaculture. Because of heavy fishing, dam construction and water pollution in the Yangtze River drainage, the wild populations of this species have rapidly declined in recent years. Now this species is listed as vulnerable in China [2]. In order to prevent the population decline from anthropogenic impacts, rock carp has been recommended as a second-class state protected animal in China (Wenxuan Cao, personal communication).

Rock carp belongs to the family Cyprinidae, and was further classified to Cyprininae based on its mainly morphological characters [3]. In phylogenetic relationships, Cyprininae is closest to Barbinae within Cyprinidae. Cyprininae is distinguished from Barbinae based on one mainly morphological character, the presence of strong last single ray in anal fin [3]. Some species of Puntioplites and Procypris in Cyprininae, including rock carp, were considered having closer relative to Barbinae based on their other morphological characters, such as pharyngeal tooth [3, 4] and the fist vertebra [5]. Phylogenetic analysis based on RAPD also suggested that Procypris rabaudi is closer to Barbinae than Cyprininae [6]. However, to date, the deeper phylogenetic relationship of rock carp remains ambiguous, and there is no more information available about phylogenetic knowledge for rock carp.

The typical vertebrate mitochondrial genome is circular, ranging in size from approximately 15 to 18 kb and generally containing 37 genes (22 tRNAs, two rRNAs, and 13 proteins) and a control region (D-loop) [7–10]. The gene order of mitochondrial genome is high conserved in fish with a few exception reported so far [8, 10–15]. Because of its maternal inheritance, relative lack of recombination, fast evolutionary rate compared to nuclear DNA, and the ability to provide an abundance of genotype, the mitochondrial DNA is a useful molecular marker for population genetic and phylogenetic studies [16–18]. In fish, numerous phylogenetic analyses have been conducted based on cytochrome b, rRNA, and control region of mitochondrion [19–27]. However, short sequence data may give rise to some misleading conclusions for resolution of deeper evolutionary branches [28–30]. Recent studies have shown that longer sequence, especially the 12 concatenated protein-coding genes, have great potential in phylogenetic inferences of deeper branches [8, 18, 31–36]. There are more than 600 complete mitochondrial DNA sequences of teleosts been deposited in GenBank. However, it is quite insufficient for complete mitochondrial data available to insight into the phylogenetic relationships in the family Cyprinidae. In this paper, we reported the complete sequence of the mitochondrial genome of rock carp and determined the mitochondrial genomic structure, gene order, codon usage and base composition. Based on rock carp’s complete mitochondrial DNA sequence and along with the sequences of 40 other cyprinids, we recovered the phylogenetic relationships to further clarify the relative phylogenetic position of P. rabaudi in Cyprinidae. We hope that the knowledge of the mitochondrial genome sequence of this species could contribute to the phylogeny clarification of cyprinids.

Materials and methods

Fish sample, mitochondrial DNA extraction, PCR amplification and sequencing

The rock carp was collected from Mudong in Chongqing, which was one of the main regions for distribution of this species in the Yangtze River. Total mitochondrial DNA (mtDNA) was extracted from the muscle tissue using the method described by Tapper et al. [37].

We used eight sets of primers to amplify contiguous, overlapping segments of the complete mitochondrial genome in rock carp. The primer sequences were shown in Table 1. The primers were designed from the mitochondrial conservative region based on a result of multiple sequence alignment of complete mitochondrion of Cyprinus carpio, Carassius carassius, Barbus barbus, Puntius ticto, Cyprinella spiloptera, Danio rerio, Ischikauia steenackeri, and Labeo batesii (GenBank accession nos. were shown in Table 2). PCR amplification was conducted on iCycler PCR System (Bio-Rad, USA) in a 25 μl reaction volume containing about 30 ng mt DNA, 1× La PCR buffer II (TaKaRa, China), 1.5 mM MgCl2, 2 μM of each primer, 0.5 mM dNTP and 1.0 U La Taq DNA polymerase (TaKaRa, China). PCR condition was 94°C for 4 min, 30 cycles consisting of 94°C for 30 s, 58°C for 50 s, 68°C extension 2–4 min, with a final extension at 72°C for 10 min. PCR products were electrophoresed on 1.0% agarose gel and sized relative to molecular weight marker D2000 (TIANGEN, China).

Table 1 The eight primer combinations for amplifying the complete mitochondrial DNA of Procypris rabaudi
Full size table
Table 2 The fish species and the GenBank accession nos. of their complete mitochondrial DNA sequences used in this study
Full size table

PCR products were purified using QUIEXII Kit (OMEGA, USA), and then directly sequenced using the primer walking method on ABI 3730 Genetic Analyzer (Applied Biosystems). The self-designed primers used in PCR and BigDye Termination v3.1 Cycle Sequencing Kit (Applied Biosystems) were used for sequencing.

Sequence analysis

DNA sequences were analyzed using the software DNAMAN version 3.0 (Lynnon Biosoft, Quebec, Canada). The locations of protein-coding and rRNA genes were determined by comparison with the corresponding known sequences of other three cyprinid fishes, Cyprinus carpio [38], Carassius carassius [39] and Barbodes gonionotus [34]. The tRNA genes were identified using the program tRNAscan-SE 1.21 [40]. Some tRNA genes, which could not be found by the tRNAscan-SE were identified by their secondary structure [41] and specific anti-codons.

Phylogenetic analysis

In order to acquire some implications about phylogenetic position of Procypris rabaudi within cyprinid, the nucleotide sequence data of 12 heavy-strand protein-coding genes were used for phylogenetic analysis. The ND6 was excluded from the phylogenetic analysis, because it is encoded by the opposite strand with considerably different base composition and codon bias. ND6 might possess a different evolutionary pattern from 12 protein-coding genes on the heavy-strand [30]. After removal of the gaps, all ambiguous sites around the gaps, overlapping region and stop codons, a 10865 nucleotide sequence set was obtained. Twelve concatenated protein-coding gene sequence of mitochondrion from rock carp and 40 other cyprinid fishes (Table 2) were used in the phylogenetic analysis. Leptobotia mantschurica, Vaillantella maassi, Gyrinocheilus aymonieri and Hypentelium nigricans (Table 2) were used as outgroups. Multiple alignments of the 12 concatenated protein-coding gene sequences were conducted using Clustal X [42] with the default settings. Two different methods, Maximum-likelihood (ML) and Bayesian, were used to construct phylogenetic relationship. For ML analysis, the best fitting models of sequence evolution were determined with Modeltest 3.06 [43]; heuristic searches were executed in 100 replicates with all characters unordered and equally weighted, and using tree bisection reconnection (TBR) branch swapping in the program PHYML [44]; bootstrapping proportions with 100 replicates were used for nodal evaluation. The Bayesian analysis was conducted using MrBayes 3.1.2 [45]. The Bayesian posterior probabilities were estimated with 2 million generations, sampling trees every 100 generations. About 20% of sampling trees were discarded (the burnin) after estimating with a conservative approach. Then a consensus tree was calculated using the remaining 16000 trees (whose log-likelihoods converged to stable values). Two separate runs with four Markov chains were performed. The genetic distances among P. rabaudi and other species in Cyprininae and Barbinae were calculated based on the concatenated nucleotide data of 12 heavy-strand protein-coding gene sequences using MEGA 3.1 [46] with the Kimura 2-parameter model.

Results and discussion

Genome organization

The complete mitochondrial genome sequence of rock carp was determined to be 16595 bases in length and was deposited in GenBank (Accession no. EU082030). As shown in Fig. 1, the organization of mitochondrial genome of rock carp is similar to that of typical vertebrate mitochondrial genome, consisting of 13 protein-coding genes, two rRNA genes and 22 tRNA genes. These genes are arranged in line in rock carp mitochondrial genome (Table 3). Also as in other vertebrates, most rock carp mitochondrial genes are encoded on the H-strand, except ND6 and eight tRNA genes (tRNA-Gln, tRNA-Ala, tRNA-Asn, tRNA-Cys, tRNA-Tyr, tRNA-Ser, tRNA-Pro, tRNA-Glu), which are encoded on the L-strand. The overall base composition of H-strand of rock carp mitochondrial genome is A: 32.27%, T: 25.20 %, C: 26.92% and G: 15.60%, with an A + T rich feature as that of other vertebrate mitochondrial genome.

Fig. 1
figure 1

The structure of complete mitochondrial genome of Procypris rabaudi

Full size image
Table 3 The characteristics of genes of Procypris rabaudi mitochondrial genome
Full size table

Protein-coding genes

Among rock carp mitochondrial protein-coding genes, the open reading frames of two pairs of contiguous genes overlap occurred on the same strand: ATPase8-ATPase6 and ND4L-ND4, and they overlap by seven nucleotides, respectively. ND5 and ND6 overlap by four nucleotides as well, whereas they are encoded on the opposition strand. All 13 protein-coding genes in rock carp mitochondrial genome use ATG as the initiation codon except the COI gene, which uses GTG as initiation codon. All COI genes in reported fishes use GTG as initiation codon, thus, the feature that COI uses GTG as initiation codon seems to be prevalent among nontetrapod vertebrates [30]. However, termination codons vary among different fish species [11]. Six protein-coding genes in rock carp mitochondrial genome end with complete stop codons, TAA (ND1, COI, ND4L, ND5, ND6) and TAG (ATPase8), the rest seven genes end with incomplete stop codons, either TA (ATPase6, COIII) or T (ND2, COII, ND3, ND4, Cytb), which are presumably completed as TAA after transcriptions [47]. The codon usage in rock carp mitochondrial genome was given in Table 4. The frequency of CTA (Leu) is the highest (count: 290), and TGT (Cys) and AAG (Lys) are the lowest (count: 6, respectively) among codons used in rock carp mitochondrial genome. This codon usage bias might be associated with the available tRNA in organism.

Table 4 Codon usage in mitochondrial genome of Procypris rabaudi
Full size table

Base composition of rock carp mitochondrial protein-coding genes is given in Table 5. Similar to other vertebrates, the base composition of 12 protein-coding genes on the H-strand is bias against G and strong bias against G at the third codon position. The most frequent nucleotide at the third codon position is A, which is consistent with most cyprinid fish, such as Cyprinus carpio, Carassius carassius, Barbus barbus, Danio rerio, but inconsistent with Opsariichthys bidens [36], where C is the most frequent nucleotide. The most frequent nucleotide at the second codon position is T (40%) and pyrimidine is over-represented (T + C=68%), owing to the hydrophobic character of the proteins [48]. However, the ND6 possesses markedly different base composition and codon bias, having more G than C or A both in total base composition and at the third codon position. To explore the codon evolution of 12 protein-coding genes in teleostean mitochondrion, we investigated the nucleotide frequency of codon using 20 teleostean data sets reported in GenBank. As shown in Table 6, the significant changes of nucleotide frequency at the third codon position indicate that there are different nucleotide preferences for the codon ends among teleosteans. Whereas, there are highly conservative nucleotide frequencies at the second codon position, with extremely minor nucleotide frequency alteration (T: 40–41%, C: 27–29%, A: 18–19%, G: 13–14%). At the second codon position, the frequency of pyrimidine seldom alternated. Most teleosteans have a pyrimidine frequency of 68%, and it ranges from 68% to 69% in 20 investigated teleosteans. These results suggested that the frequency of non-synonymous mutations is very low in 12 mitochondrial protein-coding genes of teleosteans. In order to compare the codon evolution pattern of teleosteans to that of other vertebrates, we further investigated the codon nucleotide frequency of 12 mitochondrial protein-coding genes in four elasmobranches, two amphibians, three reptiles, five birds and five mammals. As shown in Table 6, there is a similar nucleotide frequency variation as that in teleosteans, exhibiting a relatively conservative nucleotide frequency at the second codon position and an obviously various nucleotide frequency at the third codon position among different species. Compared with nucleotide frequency at the second codon position of teleosteans, there is a slightly higher variation among the investigated amniotes as follows: T: 39–42%, C: 27–31%, A: 18–20%, G: 11–12%. This might be associated with diverse habitats. Most of the amniotes have a pyrimidine frequency of 69% or 70%.

Table 5 Base compositions (%) of protein-coding genes in Procypris rabaudi mitochondrial genome
Full size table
Table 6 Base compositions (%) of 12 protein-coding genes on H-strand of mitochondrial genome in 20 teleosteans and 19 other vertebrates
Full size table

Ribosomal and transfer RNA genes

The 12S and 16S rRNA genes of rock carp mitochondrion are 958 and 1683 bp in length, respectively. As in other vertebrates, they are located between tRNA-Phe and tRNA-Leu (UUR) genes and separated by tRNA-Val gene (Fig. 1). The base composition of the two rRNA gene sequences is 35.4% A, 20.37% G, 19.88% T and 24.35% C. The content of A + T (55.28%) is higher than that of C + G (44.72%). Thus, the rock carp mitochondrial rRNA genes also exhibit A+T content-rich like as other bony fishes [49–53]. Rock carp mitochondrial genome contains 22 tRNA genes, which are interspersed between the rRNA and protein-coding genes and range from 67 bp (tRNA-Cys) to 77 bp (tRNA-Lys) in size (Table 3). Twenty-one tRNA genes, which could fold into the typical cloverleaf secondary structure, were identified by tRNAscan-SE v.1.21 [40]. Due to lacking of the complete dihydrouridine arms (D-arms), the tRNA-Ser (AGY) gene was determined by proposed secondary structures [41] and the anti-codon. The anti-codons of the 22 tRNA genes of rock carp mitochondrial genome have no unique characteristics compared to other vertebrates.

Control region

The major non-coding sequence (control region) of the rock carp mitochondrial genome is located between the tRNA-Pro and tRNA-Phe genes and is 943 bp in length. The conserved sequence blocks (CSB1-3), which were thought to be involved in positioning RNA polymerase both for transcription and priming replication [54, 55], were identified in the positions 615–641, 698–715 and 741–759 nt downstream of the 5′ end, respectively. The extended termination associated sequences (ETAS) was found in the position 25–61 nt downstream of the 5′ end, which can form a stable hairpin-loop structure. The sequence ACCAAAAACTTCCAAAAAATA, which is a putative promoter for H-strand transcription (HSP) [49], was found at 55 nt upstream of the 5′ end of tRNA-Phe gene. The HSP has a few nucleotide substitutions at the two underlined positions compared to that of Cyprinus carpio [49]. An AT-repeat microsatellite sequence was also identified at 45 nt upstream of the 5′ end of HSP. The microsatellite has only one repeat variation in 50 tested individuals of rock carp (from 12 to 13 repeats) (Jun Song et al. unpublished data). It also presents at other teleostean mitochondrion control region, but the repeats of AT might be different, for example, Cyprinus carpio: 9 repeats (AP009047), Barbus barbus: 14 repeats (AB238965). This microsatellite sequence might be useful in some interspecies identification. As like as most vertebrates, the origin of L-strand replication (OL) in rock carp mitochondrion is located in a cluster of five tRNA genes (WANCY region). The region is 35 bp in length, overlaps the tRNA-Cys gene by 3 bp and has the potential to fold into a stable stem–loop secondary structure consisting of 22 bp in the stem and 13 bp in the loop. The conserved motif 5′-GGCGGG-3′ also presents in the stem of tRNA-Cys gene. Compared with the OL of Cyprinus carpio, the rock carp OL exhibits five nucleotide substitutions and one nucleotide deletion in the loop, yet they are completely identical in the stem.

Phylogenetic analysis

To investigate the phylogenetic position of rock carp, the concatenated nucleotide sequence of the 12 heavy-strand protein-coding genes were used to construct the phylogenetic relationships by Bayesian and ML methods (Fig. 2). Hierarchical likelihood ratio tests indicate that the (GTR +I + Г) model of substitution and gamma distribution was the best for our data (GTR + I + Г, −lnL = 280184.91; Ts/Tv ratio = 5.66; distribution shape parameter = 0.6044). The yielded Bayesian tree had a nearly same topology as that of ML (Fig. 2). In the clade C of Fig. 2, Barbodes gonionotus (subfamily Barbinae) was placed on the basal position, sister to the clade comprising of P. rabaudi, C. carpio and C. carassius. In addition, as shown in clade B, three other barbines diverged prior to cyprininaes as well. This suggested that the traditional taxonomic barbines possibly originated more early than cyprininaes. Procypris rabaudi diverged after the emergence of B. gonionotus of Barbinae and prior to C. carpio and C. carassius of Cyprininae. Also, the genetic distances of P. rabaudi to C. carpio and C. carassius were 0.1260 and 0.1382, respectively, and that to B. gonionotus, B. barbus, B. trimaculatus and P. ticto were 0.1396, 0.1498, 0.2242 and 0.1872, respectively. Both the phylogenetic tree and genetic distances showed that P. rabaudi had a closer relationship to Cyprininae than Barbinae, which was incongruent with the opinion that P. rabaudi was closer to Barbinae than Cyprininae inferred from RAPD analysis [6]. Our result appeared to be consistent with the results from morphological characters of rock carp compared with that of the Cyprininae and Barbinae [3–5]. However, considering only four species of Barbinae, and two other species of Cyprininae have been incorporated into the phylogenetic analysis, it is needed more evidences rooted in the more species of Barbinae and Cyprininae to confirm the phylogenetic position of rock carp in Cyprinidae in the future.

Fig. 2
figure 2

Phylogenetic relationships of the family Cyprinidae by Bayesian and Maximum likelihood (ML) methods based on the concatenated nucleotide sequence of 12 protein-coding genes on the heavy strand. Numbers in the nodes: posterior probabilities for Bayesian analysis and bootstrap values for ML analysis. Less than 0.95 of Bayesian posterior probabilities and 50% of bootstrap values were omitted. If only less than 0.95 of Bayesian posterior probability, or 50% of bootstrap value in the same clade, both were kept to avoid the confusion between them

Full size image

In the present study, the fishes of Gobinoninaes formed an independent monophyletic group (clade A in Fig. 2), which agreed with the traditional Gobinoninae grouping [1, 3, 4]. In clade B, the Labeo batesii was placed at the basal-most position, sister to the clade consisting of barbines, cyprininaes and a schizothoracine. Several traditional taxonomic subfamilies, such as Cultrinae, Danioninae and Leuciscinae, represented polyphyletic in the phylogenetic tree, which was accordant to the conclusions of Saitoh et al. [34]. This may be due to the result that some similar morphological characters were from either convergent evolution or retained ancestral characters shared across some taxa, which were used to group by the traditional subfamily classification within Cyprinidae [56]. However, the relationships of the genera within Labeoninae, Schizothoracinae, Xenocyprinae and Acheilognathinae were ambiguous because of lack of complete mitochondrial data of various genera in these subfamilies.