| Edited by Toshihiko Shiroishi* Corresponding author. E-mail: msumida@hiroshima-u.ac.jp |
Vertebrate mitochondrial DNA (mtDNA) is a closed circular molecule. The genome organization is highly compact and the genome is approximately 16 kbp in length (Wolstenholme, 1992). This genome typically contains 37 genes for 2 ribosomal (r)RNAs, 22 transfer (t)RNAs and 13 proteins, and a long noncoding region called the D-loop that includes signals required for regulating and initiating mtDNA replication and transcription (Wolstenholme, 1992; Janke et al., 1994).
Mitochondrial gene arrangements are generally conserved in vertebrates. All 37 genes are arranged in the same relative order in almost all vertebrate species from teleost fishes to eutherian mammals (Anderson et al., 1981; Roe et al., 1985; Tzeng et al., 1992; Boore, 1999). Although minor rearrangements have been reported for marsupials (Pääbo et al., 1991), birds (Desjardins and Morais, 1990, 1991; Quinn and Wilson, 1993; Mindell et al., 1998), reptiles (Kumazawa and Nishida, 1995; Quinn and Mindell, 1996; Macey et al., 1997), lampreys (Lee and Kocher, 1995), and teleost fishes (Miya and Nishida, 1999; Inoue et al., 2001; Miya et al., 2001), most of these cases involve only a few rearrangements of tRNA genes and/or the D-loop region.
Mitochondrial gene arrangements have attracted the attention of evolutionary biologists as novel phylogenetic markers (Smith et al., 1993; Kumazawa and Nishida, 1995; Quinn and Mindell, 1996; Macey et al., 1997; Boore and Brown, 1998; Boore, 1999; Kurabayashi and Ueshima, 2000). The complete mtDNA sequences have been published for only six amphibian species, including a caecilian (Zardoya and Meyer, 2000), three salamanders, (Zardoya et al., 2003; Zhang et al., 2003a, b), and two anurans, the clawed frog Xenopus laevis (Roe et al., 1985) and the Japanese pond frog Rana nigromaculata (Sumida et al., 2001). The mitochondrial gene arrangements of five of these six amphibian species are identical with those of typical vertebrates. However, in the R. nigromaculata mtDNA, the positions of four tRNA genes (tRNA-Leu(CUN), tRNA-Thr, tRNA-Pro, and tRNA-Phe) differ from those of typical vertebrates (see Fig. 3). The same gene arrangement is also found in other ranid frogs so far investigated (Rana catesbeiana, Yoneyama, 1987; Rana limnocharis, Macey et al., 1997; Rana porosa, Sumida et al., 2000). A lack of available information on the mitochondrial genomic organization of other anuran family members makes it difficult to determine whether this unique arrangement is shared by other anuran groups besides the ranids.
In order to elucidate various aspects of mitochondrial gene rearrangement, we determined the complete mtDNA sequence of the bell-ring frog Buergeria buergeri, a representative of the family Rhacophoridae (tree frogs). Rhacophoridae is generally considered to have close affinity to the family Ranidae (true frogs), and the genus Buergeria is regarded as the most basal group in the former (Channing, 1989; Jiang et al., 1987; Liem, 1970; Richards and Moore, 1998; Wilkinson and Drewes, 2000; Wilkinson et al., 2002).
In this report, we present the first data on the complete mtDNA sequence of a rhacophorid frog and describe features of the genome. The evolutionary implications of our findings are also discussed.
Bell-ring frogs (Buergeria buergeri) were collected from Ota River, Hiroshima prefecture, western Japan. The total genomic DNAs were extracted from a clipped toe of a living frog using the DNeasy Tissue Kit (QUIAGEN) according to the manufacturer’s protocol.
The total length of B. buergeri mtDNA was amplified by polymerase chain reaction (PCR), beginning with partial mitochondrial segments and finishing with the remaining mtDNA region. Two partial mitochondrial segments were amplified from B. buergeri total DNA by long-and-accurate PCR (LA-PCR) using two primer sets (F20N7 and R16; FR16 and R51) (Fig. 1 and Table 1). PCR mixtures were prepared using a TaKaRa LA Taq™ Kit as recommended by the manufacturer (TaKaRa). LA-PCR reactions consisted of an initial denaturation at 94°C for 1 min, 30 cycles of denaturation at 94°C for 20 sec plus annealing and extension at 60°C for 3 min, and a final extension at 72°C for 10 min. The resultant PCR fragments were electrophoresed on a 0.7% agarose gel, and the DNAs were purified from excised pieces of gel using Wizard SV Gel and the PCR Clean-UP System (Promega) and used for sequencing (see below). The remaining mtDNA fragment was amplified with the primer set F70 and R71 (Fig. 1 and Table 1) designed based on the ND5 gene and 16S rRNA gene sequences determined above. LA-PCR reactions consisted of 1 cycle of 1 min at 94°C, 14 cycles of 20 sec at 98°C followed by 20 min at 68°C, 17 cycles of 20 sec at 98°C followed by 20 min 20 sec at 68°C, and 1 final cycle of 10 min at 72°C. The amplified mtDNA fragment of approximately 16.5 kbp was then purified using MicroSpin™ S-300 HR Columns (Pharmacia Biotech) and used for the sequencing reaction.
 View Details | Fig. 1. Sequencing and cloning strategy for Buergeria buergeri mtDNA. Localizations and directions of primers used in the LA-PCR amplification and DNA sequencing are denoted by arrowheads. The sequences of these primers are available from the www site, http://home.hiroshimau.ac.jp/~amphibia/syukeisei/usedprimers.html. LA-PCR products are shown as bold lines below the gene map. Cloned restriction fragments and their lengths are indicated above the gene map. |
 View Details | Table 1. Primers used for PCR amplification in the present study. |
The entire mtDNA genome of B. buergeri was sequ-enced from both strands. The primer walking method was employed to sequence almost all portions of the mtDNA (Fig. 1). The sequencing was performed using 373A and 3100-Avant automated DNA sequencers (ABI) with DYEnamic ET Terminator cycle sequencing reagent (Amersham). Sequencing primers for internal portions of the long PCR fragment were designed by cloning five restriction fragments (see Fig. 1) into pUC 118 E. coli vector and sequencing them. The D-loop region was difficult to sequence by primer walking due to an abundance of lengthy tandem repeat units. To determine the precise sequence of this region, a series of deleted subclones was made from the clones of an Sma I/EcoR I-digested fragment (see Fig. 1) using the Exonuclease III deletion method (Henikoff, 1987). The nucleotide sequence of the B. buergeri mtDNA was deposited in the DDBJ database accession number AB127977.
For the phylogenetic analysis, we created an alignment dataset using CLUSTAL W (Thompson et al., 1994). The data set consisted of all 37 mitochondrial gene sequences from 9 vertebrates, including 7 amphibians, 1 coelacanth (Latimeria chalumnae), and 1 lungfish (Protopterus dolloi). The latter two were used as outgroups. The alignment was checked by eye, and all positions with gaps and ambiguous sites were excluded. Based on the alignment data (12,865 nucleotide sites), we reconstructed a phylogenetic tree by the maximum likelihood (ML) method. The tree construction was performed with PAUP* ver. 4.10b (Swofford, 2001). The best-fit model of DNA substitution was estimated using ModelTest ver. 3.06 (Posada and Crandall, 1998) and a general-time-reversible + gamma + invariant (GTR + G + I, G = 1.0444, I = 0.3276) model was proposed under AIC consideration.
The complete nucleotide sequence of the mitochondrial genome of B. buergeri was determined. Though the genome was extremely long–19,959 bp, the longest among all vertebrate mtDNAs so far examined, B. buergeri mtDNA included only 37 typical mitochondrial genes, 13 protein genes, 2 rRNA genes and 22 tRNA genes (Fig. 2 and Table 2), most of which were similar in length to their counterpart genes in other amphibians. However, length differences between Buergeria and other amphibians were observed in the Cytb and ND6 genes, and especially in the D-loop region (Table 3). The large genome size of B. buergeri mtDNA was due to the accumulation of lengthy repetitive sequences into the D-loop region (see the section about the noncoding region).
 View Details | Fig. 2. Organization of the B. buergeri mitochondrial genome. All protein-coding genes are encoded by the H strand, with the exception of ND6, which is encoded by the L strand. Transfer RNA genes are represented by the standard one-letter amino acid code and those encoded by the H and L strands are shown outside and inside the circle, respectively. L1, L2, S1, and S2 denote tRNA-Leu(CUN), tRNA-Leu(UUR), tRNA-Ser(UCN), and tRNA-Ser(AGY), respectively. Genes are abbreviated as in Table 2, except for A6 and A8, which denote ATPase6 and ATPase8, respectively. |
 View Details | Table 2. Location of features in the mitochondrial DNA of Buergeria buergeri. |
The base composition of the complete B. buergeri mtDNA was A: 29.9%, T: 30.5%, G: 14.7%, and C: 24.8%. The slightly high A + T content (60.4%) of B. buergeri mtDNA is similar to those of other vertebrates.
The gene arrangement of B. buergeri mtDNA diverged from that of typical vertebrates (Fig. 3). Specifically, the ND5 gene and four tRNA genes (tRNA-Leu(CUN), tRNA-Thr, tRNA-Pro and tRNA-Phe) were located between the D-loop and the 12S rRNA gene. The tRNA cluster upstream of the 12S rRNA gene was also identified in the Japanese pond frog, Rana nigromaculata (Sumida et al., 2001) (Fig. 3) and other known ranid frogs (Yoneyama, 1987; Macey et al., 1997; Sumida et al., 2000). However, the rearrangement of the ND5 gene has not been seen in other amphibians sequenced previously.
 View Details | Fig. 3. Comparison of mitochondrial gene arrangements among amphibians. Arrows indicate the rearranged homologous genes. Closed and shaded boxes indicate the genes whose positions vary among anurans. Genes are abbreviated as in Table 2 and Fig. 2. |
All of the 13 protein-coding genes found in other animals were also present in the B. buergeri mitochondrial genome. The codon usage of B. buergeri mtDNAs was identical to that of the other vertebrates. All but two of these genes started with an ATG or ATA initiation codon (the exceptions, ND1 and ND2, started with an ATT codon) (Table 2). Four protein-coding genes of B. buergeri (COI, COII, ND5 and ND6) ended with the AGR (AGA and AGG) stop codon characteristic of vertebrate mtDNAs. Four protein genes (ATPase8, Cytb, ND2, and ND4L) stopped with the usual TAR codon. The remaining 5 genes (ATPase6, COIII, ND1, ND3 and ND4) had an incomplete stop codon, a single stop nucleotide T, where the post transcriptional polyadenylation could produce a complete TAA stop codon (Ojara et al., 1980).
The lengths of the 12S and 16S rRNA genes in the B. buergeri mitochondrial genome (927 and 1,574 bp, respectively) were similar to those of the corresponding genes in other amphibians (Table 3) and vertebrates. Our sequence showed only minor differences (99% similarity) from the partial nucleotide sequences of the 12S (507 bp) and 16S rRNA (1,383 bp) genes of B. buergeri (previously reported by Wilkinson et al., 2002).
 View Details | Table 3. Comparisons of lengths (bp) of mitochondrial genes of amphibians. |
Twenty-two tRNA genes were identified in the mitochondrial genome of B. buergeri by comparison with homologues of other amphibians and by their ability to fold into a putative secondary structure (Fig. 4). Twenty-one of the 22 tRNAs could be folded into the canonical cloverleaf secondary structure, while tRNA-Ser(AGY), with an unpaired dihydrouridine (D) arm, could not. The unpaired D-arm in tRNA-Ser(AGY) is a common feature of metazoan mtDNAs (Wolstenholme, 1992).
 View Details | Fig. 4. Putative secondary structures of B. buergeri mitochondrial tRNA genes. The cloverleaf structures of 22 tRNA genes identified in B. buergeri mtDNA are shown. Watson-Crick base pairing is indicated by solid bars (–), and the G-T pairs usually observed in animal mtDNAs are shown with plus marks (+). |
A major noncoding region of 4,576 bp was found in the B. buergeri mtDNA between the Cytb and ND5 genes (Fig. 2). This region was thought to correspond to the D-loop region as it contained several components characteristic of the D-loop region: apparent homologues of the termination-associated sequence (TAS), an H-strand origin of replication (OH), and three conserved sequence blocks (CSB-1, CSB-2 and CSB-3) (Fig. 5 and Table 4). Although these notable structural features were conserved in the D-loop region of the B. buergeri mtDNA, the region was extremely long (4,576 bp) compared to those in other vertebrates, including amphibians (Table 3). This unusually large size was due to the presence of two distinct tandem repeat units identified in the 5’- and 3’-sides of the D-loop region (Fig. 5). The 5’-side repeated region consisted of 20 repeat units of 38 bp and one incomplete repeat unit of 22 bp. The 3’-side repeated region consisted of 4 repeat units of 105 bp and 19 repeat units of 96 bp (Table 4).
 View Details | Fig. 5. Schematic diagram of the B. buergeri mtDNA control (D-loop) region. Dark hatched squares show the tandem repeat sequences and the numbers (bp) under the map represent the lengths of the tandem repeats. Abbreviations are follows: TAS, termination associated sequence; OH, H-strand origin of replication; CSB, conserved sequence block. |
 View Details | Table 4. Nucleotide sequences of conserved segments and tandem repeat units in the D-loop region of B. buergeri. |
The putative origin of L-strand replication (OABCDECGH) of the B. buergeri mitochondrial genome was located between tRNA-Asn and tRNA-Cys in the WANCY tRNA gene cluster (Fig. 2). The putative OL was 29 nucleotides in length and the sequence had the potential to fold in a stem-loop secondary structure with a stem formed by 9 paired nucleotides and a loop of 9 nucleotides (Fig. 6). A “GACGG” sequence was present at the base of the stem region in B. buergeri (Fig. 6). This pentanucleotide is very similar to the “GCCGG” sequence motif involved in the transition from RNA to DNA synthesis of the human mtDNA system (Hixson et al., 1986).
 View Details | Fig. 6. Putative secondary structures in the L-strand replication origins of anuran mtDNAs. The pentanucleotides predicted to be involved with the transition of RNA to DNA synthesis are boxed. The sequences and secondary structures of R. nigromaculata and X. laevis were quoted from Sumida et al. (2001) and Roe et al. (1985), respectively. |
We analyzed the phylogenetic relationship among amphibians (Anura, Caudata and Gymnophiona) using the long nucleotide sequences of all the mitochondrial genes. The resultant ML tree (-lnL = 85787.4062) is shown in Fig. 7. The robustness of our result was confirmed by high bootstrap support (≥95) of all nodes. Each amphibian order in our tree split into an independent branch. Gymnophiona was the first to branch away, and Caudata and Anura formed a monophyletic group. In the order Anura, Xenopus laevis (family Pipidae) was the first to branch away, and Rana nigromaculata (Ranidae) and B. buergeri (Rhacophoridae) formed a sister group.
 View Details | Fig. 7. Maximum-likelihood (ML) tree based on all the 37 mitochondrial genes (12,856 nucleotide sites). The following parameters were used for ML analysis: the proportion of invariable sites (I) = 0.3276, the gamma distribution shape parameter (α) = 1.0444, empirical base frequencies (A: 0.3130; C: 0.2577; G: 0.1502; T: 0.2791), and substitution rates ([A-C] = 2.0504, [A-G] = 4.3039, [A-T] = 2.3986, [C-G] = 0.4303, [C-T] = 8.5382, [G-T] = 1). The scale bar shows the genetic distance calculated using the above parameter values. The values of internal branches are the bootstrap values (1,000 replicates). The accession numbers for the sequences used were as follows: Rana nigromaculata (AB043889; Sumida et al., 2001), Xenopus laevis (M10217; Roe et al., 1985), Mertensiella luschani (AF154053; Zardoya and Meyer, 2001), Ranodon sibiricus (AJ419960; Zhang et al., 2003a), Andrias davidianus (AJ492192; Zhang et al., 2003b), Typhlonectes natans (AF154051; Zardoya and Meyer, 2000), Latimeria chalumnae (coelacanth; U82228; Zardoya and Meyer, 1997) and Protopterus dolloi (lungfish; L42813; Zardoya and Meyer, 1996). The latter two were used as outgroups. Arrows indicate the ancestral lineages in which the gene rearrangements occurred (see text). |
The genome size of vertebrate mtDNAs ranges from 15,181 bp in the tuatara (Rest et al., 2003) to 18,978 bp in the deep-sea gulper eel (Inoue et al., 2003). The Buergeria buergeri mtDNA, whose sequence is reported here, has the largest genome size (19,959 bp) among the vertebrates investigated so far.
Vertebrate mtDNAs of large sizes (>18 kbp) generally have multiple long noncoding portions. For example, the mtDNAs of two eel species (Eurypharynx pelecanoides, 18,978 bp, Inoue et al., 2003 and Conger myriaster, 18,705 bp, Inoue et al., 2001) contain three noncoding regions, and those of several bird species (18 kbp ~ 18.7 kbp) possess two noncoding regions (Haring et al., 2001). These additional noncoding regions are the chief factor responsible for the increased genome sizes. On the other hand, B. buergeri mtDNA contains only a single long noncoding region, corresponding to the D-loop. This region contains two series of long repeat units (Fig. 5) that considerably expand the genome. It is interesting that relatively long repetitive sequences in the D-loop region are also found in all reported anuran species, particularly in ranids (1.2 kbp in R. porosa and 1.0 kbp in R. nigromaculata; Sumida et al., 2000) and in rhacophorids (at least 2.0 kbp in Polypedates leucomystax, 1.3 kbp in Rhacophorus schlegelii; Kurabayashi et al., in preparation). The tendency for the repetitive sequences to accumulate in the D-loop region might be a feature of ranid and rhacophorid mtDNAs.
No complete consensus has yet been reached on the phylogenetic relationships among the three orders of living amphibians (Trueb and Cloutier, 1991b; Carroll et al., 1999). Most morphological studies support a close relationship bet--ween anurans (frogs) and caudates (salamanders and newts) generally referred to as the “Batrachia hypothesis” (Trueb and Cloutier, 1991a). On the other hand, several molecular studies based on nuclear and mitochondrial rRNA data have supported the notion that caecilians are a sister taxon of salamanders (Larson and Wilson, 1989; Hedges et al., 1990; Hay et al., 1995; Feller and Hedges, 1998). In attempts to resolve this conflict, larger data sets, i.e., complete mitochondrial genomes, have been employed in several phylogenetic studies, and the results supported the former hypothesis (Zardoya and Meyer, 2001; Zardoya et al., 2003, Zhang et al., 2003b). Furthermore, our phylogenetic analysis (Fig. 7) using the complete mitochondrial gene sequence data of B. buergeri also confirmed the monophyly of anurans and caudates with sufficient statistical significance (BP = 95).
It has been broadly accepted that the Pipidae is one of the basal families in anurans and that Ranidae and Rhacophoridae are members of a derived anuran group, the suborder Neobatrachia (e.g., Duellman, 1975; Ford and Cannatella, 1993). Our phylogenetic tree also supported this traditional phylogenetic relationship.
The present study showed that the gene order of Buergeria buergeri mtDNA differs from that of typical vertebrates by rearrangements of the positions of four tRNA genes and the ND5 gene (Fig. 3). Gene rearrangement in animal mtDNA is generally believed to take place through tandem duplication of gene regions as a result of slipped strand mispairing, followed by multiple deletions of redundant genes (Moritz and Brown, 1986, 1987; Moritz et al., 1987; Boore and Brown, 1998). According to the gene arrangement mechanism, at least two duplication-deletion events through an intermediate arrangement would be needed to generate the B. buergeri gene order from that of typical vertebrates (see Fig. 8). In addition, taking into account the parsimonious principle, two rearrangement pathways are possible, as shown in Fig. 8. In these hypothetical pathways, the first step, the duplication of the region between the tRNA-Leu and D-loop in the typical gene order of vertebrates, is common. However, the intermediate arrangements differ from each other due to differences in the deleted genes. Regarding these pathways, it is remarkable that the gene order of the intermediate arrangement in hypothesis A (i.e., D-loop, tRNA-Leu, tRNA-Thr, tRNA-Pro, tRNA-Phe, 12S) is identical to that found in ranid frogs. This strongly suggests that the pathway illustrated by hypoth-esis A occurred during anuran evolution. This rearrangement pathway is also consistent with the phyl-ogenetic relationships of anurans. Rearrangements of four tRNAs appear to have occurred in a common ancestral lineage of ranids and rhacophorids after the pipid branching, and the rearrangement of the ND5 gene appears to have taken place after ranid divergence (Fig. 7).
 View Details | Fig. 8. Possible gene rearrangement pathways in anuran mtDNAs. Two hypothetical pathways postulated from the duplication-deletion rearrangement model are illustrated. Duplicated regions are indicated by thick lines and the deleted genes and D-loop are indicated by asterisks. Genes are abbreviated as in Table 2 and Fig. 2. The intermediate gene order in hypothesis A is the same as that in ranid frogs. |
In the present report, we showed that the previously known arrangement of four tRNA genes in ranids is also shared by a rhacophorid frog, B. buergeri, and we suggest that the rearrangements of the genes occurred in a lineage leading to the Ranidae and Rhacophoridae after the Pipidae (including Xenopus) divergence. In the recent classification of anurans (e.g., Dubois, 1992; Ford and Cannatella, 1993; Vences and Glaw, 2001), the Ranidae and Rhacophoridae families are grouped in the superfamily (or epifamily) Ranoidea with other closely related families (e.g., Mantellidae, Dendrobatidae, and Microhylidae). In addition, Ranoidea is placed in the suborder Neobatrachia with many other frog families (e.g., Bufonidae, Hylidae) and the neobatrachian clade, in turn, forms a sister group with a primitive group (generally called Archaeobatrachia) that includes the Pipidae. It remains unclear whether the unique arrangement of tRNA genes was acquired in the common ancestor of all neobatrachian frogs or in the ancestor of limited ranoid lineages. Investigations of many more neobatrachian families will be required in order to understand the rearrangement of the mitochondrial tRNA genes in detail.
We also showed that the ND5 gene has a novel position in B. buergeri mtDNA and that the rearrangement of this gene occurred in a lineage leading to the species after the split of ranid frogs. The details of the phylogenetic lin-eage in which the novel ND5 position emerged are unclear. However, our newest finding shows that ND5 occupies the same position in many rhacophorid frogs and some species belonging to another ranoid family, Mantellidae (Kurabayashi et al., in preparation). This suggests that rearrangement of the position of the ND5 gene occurred in an ancestral lineage leading to not only rhacophorids, but also some of the ranoidea taxa. Phylogenetic relationships among ranoid taxa (especially at the family to genus level) have been the subject of intense debate during the past few decades (see Vences and Glaw, 2001 and references therein). The position of the ND5 gene may be one of valuable characteristics for elucidating ranoid phylogeny.
The authors express their sincere thanks to Dr. Shigeru Ohta, Hiroshima Kokusai Gakuin High School, for his kind aid in collecting specimens of Buergeria buergeri. This work was sup-ported by a Grant-in-Aid for Scientific Research (C) (No. 13839012) from the Ministry of Education, Culture, Sports, Science and Technology, Japan (to M. Sumida) and a Grant-in-Aid for the Extensive Research Program of Hiroshima Prefectural Government (to T. Fujii).
|
Index