Introduction

The order Araneae ranks seventh in global animal diversity based on its described and anticipated species (Coddington and Levi 1991), next to the five largest insect orders (Coleoptera, Hymenoptera, Lepidoptera, Diptera, Hemiptera) and Acari among the Arachnida. To date, approximately 38,432 species of spiders have been named and placed in 3542 genera and 110 families (Platnick 2004). Statistical analysis suggests that about 20% of the world fauna has been described, implicating that there may be as many as 190,000 distinct spiders in all (Coddington and Levi 1991).

Mitochondrial genome sequence and gene arrangement comparisons have emerged as powerful tools during the past decade for resolving ancient phy- logenetic relationships. Although partial sequences of rrnS, rrnL, nad1, cox1, and tRNAs from spiders have been documented (Gillespie et al. 1994; Hedin 1997a, b, 2001; Hedin and Maddison 2001; Zehethofer and Sturmbauer 1998; Fang et al. 2000; Masta 2000; Piel and Nutt 2000; Bond et al. 2001; Vink et al. 2002; Vink and Paterson 2003; Hormiga et al. 2003; Maddison and Hedin 2003; Arnedo et al. 2004), the complete mitochondrial sequence has not been reported for any of the spiders. The number of complete mtDNAs of Arthropoda is limited to 40, of which 6 are from Chelicerata. Moreover, the sampling of order in Chelicerata is extremely biased. Of the six sequenced mtDNAs, five are from Acari, and the sixth one is from Xiphosura, the horseshoe crab Limulus.

Morphological and auxiliary parameters were used to form three major monophyletic groups of Araneae: Mesothelae, Mygalomorphae, and Araneomorphae. The viewpoint of Platnick and Gertsch (1976), based on a cladistic analysis using 14 characteristics, is accepted by most in the arachnological field: Mesothelae as a basal suborder and Mygalomorphae and Araneomorphae as two infraorders under the suborder Opisthothelae. This classification is also supported by a recent study using both morphological and molecular characters (Giribet et al. 2002). Two spiders, Heptathela hangzhouensis (suborder Mesothelae, family Liphistiidae) and Ornithoctonus huwena (suborder Opisthothelae, infraorder Mygalomorphae, family Theraphosidae), were selected as the representatives of Araneae for studies of mitochondrial gene structure.

In this article, the gene sequences and arrangements of mitochondrial genomes of the two spiders are compared with those of other chelicerates. The complete mitochondrial sequences of H. hangzhouensis and O. huwena allowed us to analyze nucleotide composition and codon usage patterns and to identify structural features that may be involved in regulating mtDNA replication. We observed truncations in tRNA and rRNA genes and overlaps among genes in the two species, which may reflect a trend of mitochondrial genome minimization. The mismatches in the tRNA aminoacyl acceptor arms of the two species may indicate involvement of editing mechanisms (Yokobori and Pääbo 1995a, b; Lavrov et al. 2000b).

Materials and Methods

Of the two spider species used in this report, H. hangzhouensis was collected from Xianning County, Hubei Province (29.8 N, 114.2 E), and O. huwena was collected from Ningming County, Guangxi Province (22.1 N, 107.0 E), in China. Voucher specimens are stored at Nanjing Normal University at −70°C. Total cellular DNA was prepared using whole bodies of H. hangzhouensis and from leg muscles of O. huwena, as described previously by Hwang et al. (2001).

Prior to long PCR, partial regions of rrnL and cox1 from the total DNAs of the two spiders were amplified (Fig. 1A) and sequenced, using the primer pairs LR-J-12887/LR-N-13398 (Simon et al. 1994) and PMT1/PMT2 (Folmer et al. 1994). The primer pair 16S-aa/16S-bb (Table 1) was designed based on the partial rrnL sequence of O. huwena and the pair of primers (16S-aa and 16S-bb) reported by Hwang et al. (2001). The complete mitochondrial genome of O. huwena was amplified with this primer pair using the Expand Long Template PCR System (Roche) except for 249 bp of the rrnL segment that had been sequenced (Fig. 1B). The PCR product was purified using the Gel Extraction Mini Kit (Watson Biotechnologies). Using the purified long-PCR amplification product as template, the mitochondrial genome of O. huwena was amplified in four pieces (Fig. 1C) with TaKaRa Ex Taq (TaKaRa Biotechnology). Among the four pairs of primers, four primers (16S-H and 16S-L, COI-L and COI-H) (Table 1) were designed based on the sequences of rrnL and cox1 of O. huwena, two primers (CO3-L and ND4-H) were designed based on C3-N-5460 and N4-J-8944 (Simon et al. 1994), and two primers (CO3-H and ND4-L) were designed based on the sequences of cox1–cox3 and nad4–rrnL of O. huwena. For H. hangzhouensis, using total cellular DNA as the template, the entire mitochondrial genome was amplified in four pieces (Fig. 1C) with TaKaRa Ex Taq. Among the four pairs of primers, four primers (16S-H and 16S-M-L, COI-M-L and COI-M-H) (Table 1) were designed based on the sequences of rrnL and cox1 of H. hangzhouensis, two primers (CO3-L and ND4-H) were designed based on C3-N-5460 and N4-J-8944, and two primers (CO3-H and ND4-M-L) were designed based on the sequences of cox1–cox3 and nad4–rrnL of H. hangzhouensis.

Figure 1
figure 1

Schematic of sequencing strategy. Horizontal lines indicate amplified fragments. A The partial portions of the rrnL and cox1 genes of the two species were amplified with two pairs of universal primers: LR-J-12887/LR-N-13398 and PMT1/PMT2. B For Ornithoctonus huwena, the complete mitochondrial genome was amplified with a pair of primers, 16S-aa and 16S-bb, designed based on the sequence of rrnL. C In O. huwena, the rrnL–cox1 region was amplified with the primers 16S-H and COI-L, the cox1–cox3 region was amplified with the primers COI-H and CO3-L, the cox3–nad4 region was amplified with the primers CO3-H and ND4-L, and the nad4–rrnL region was amplified with the primers ND4-H and 16S-L. In Heptathela hangzhouensis, the rrnL–cox1 region was amplified with the primers 16S-H and COI-M-L, the cox1–cox3 region was amplified with the primers COI-M-H and CO3-L, the cox3–nad4 region was amplified with the primers CO3-H and ND4-M-L, and the nad4–rrnL region was amplified with the primers ND4-H and 16S-M-L.

Full size image
Table 1 Primer sequences and their target genes/genome regions
Full size table

The four amplification products (rrnL–cox1, cox1–cox3, cox3–nad4 and nad4–rrnL) of the two spiders were sequenced by primer walking using an ABI 3700 and LI-COR automated DNA sequencer. Sequences were assembled with the DNASTAR software package (DNASTAR, Inc.) and aligned with Clustal_X (Version 1.8) (Thompson et al. 1997), DNAClub (Chen, DNAclub, Inc.), and MEGA (Version 2.1) (Kumar et al. 2001). Protein-coding and ribosomal genes were identified using the NCBI Internet Blast Search function. Transfer RNA genes were identified visually as sequences with potential tRNA secondary structure and specific tRNAs were identified by their anticodon sequences. The complete sequences of the mtDNA of H. hangzhouensis and O. huwena have been submitted to GenBank under accession numbers AY309258 and AY309259, respectively.

Results

Genome Size and Structure

The complete mtDNA sequences of H. hangzhouensis and O. huwena are 14,215 and 13,874 bp, respectively. Both contain the 37 genes typical for animal mtDNA. The size of the mitochondrial genome of O. huwena is relatively short in metazoa but longer than that of the nematode Caenorhabditis elegans (13,794 bp [Okimoto et al. 1992]). The majority of genes on both genomes are either immediately contiguous or overlapping. In addition to the large noncoding region, 71 and 12 noncoding nucleotides are present in H. hangzhouensis and O. huwena, respectively. In H. hangzhouensis, protein-coding genes account for 75.7% (10,759 bp) of the genome, rRNA genes for 12.8% (1,817 bp), tRNA genes for 9.0% (1,285 bp), and noncoding DNA for 2.9% (411 bp). Similarly, in O. huwena, protein-coding genes account for 77.3% (10,719 bp) of the genome, rRNA genes for 12.4% (1,714 bp), tRNA genes for 8.9% (1,240 bp), and noncoding DNA for 2.9% (408 bp). In H. hangzhouensis and O. huwena, 57 and 207 bp, which overlap among genes, were counted twice in the above calculation, respectively.

The gene arrangement of H. hangzhouensis appears identical to that of Limulus (Lavrov et al. 2000a); However, O. huwena and certain species of suborder Opisthothelae, infraorder Araneomorphae (Hedin 1997b; Masta 2000; Hormiga et al. 2003), share a partial mitochondrial gene arrangement that may reflect a derivative status. Compared with the basal arthropod mitochondrial gene arrangement in Limulus, a total of seven tRNA genes were found to be rearranged in O. huwena (Fig. 2).

Figure 2
figure 2

Comparison of gene arrangements in mtDNA of Limulus polyphemus, Heptathela hangzhouensis, and Ornithoctonus huwena. Genes are transcribed from left to right except those underlined, which are transcribed oppositely. Transfer RNA genes are labeled with single-letter amino acid abbreviations except for those encoding leucine and serine, which are labeled L1 (tag), L2 (taa), S1 (gct), and S2 (tga). Abbreviations of protein-coding genes and rRNA genes: cox1–cox3, cytochrome oxidase subunits 1–3; cob, cytochrome b; nad1–nad6, NADH dehydrogenase subunits 1–6; nad4L, NADH dehydrogenase subunit 4L; atp6 and atp8, ATP synthase subunits 6 and 8; rrnL and rrnS, large and small ribosomal subunit RNAs. Long arrows are drawn from the location of each tRNA gene in the ancestral arrangement to its location in O. huwena. The rotating arrow indicates inversion. CR indicates the control region (the large noncoding region).

Full size image

Base Composition

The A+T content of H. hangzhouensis and O. huwena is 72.2 and 69.8%, respectively, lower than that of all other chelicerates (Navajas et al. 2002) except for Limulus, with an A+T content of 67.6% (Lavrov et al. 2000a). We designated the strand bearing more protein-encoding sequence the α strand, and its complementary strand the β strand. H. hangzhouensis has the nucleotide composition of the α strand: A = 5014 (35.27%), T = 5252 (36.95%), C = 2438 (17.15%), and G = 1511 (10.63%); O. huwena has the nucleotide composition of the α strand: A = 4438 (31.99%), T = 5246 (37.81%), C = 1375 (9.91%), and G = 2815 (20.29%). The base compositional difference between the two strands can be measured as GC- and AT-skews, where GC-skew = (G − C)/(G + C) and AT-skew = (A−T)/(A+T) (Perna and Kocher 1995). For the H. hangzhouensis α strand, GC-skew = −0.23 ana AT-skew = −0.02. For the O. huwena α strand, GC-skew = 0.34 and AT-skew = −0.08. Although lower than the two absolute values of Limulus, the absolute GC-skew of the two-species α strand and the AT-skew of the O. huwena α strand are comparatively high for an arthropod mtDNA, reflecting their base compositional differences and an obvious bias between the two strands. The opposite directions of GC-skews of α strands in H. hangzhouensis and O. huwena mitochondrial genomes may suggest that the β strand is replicated first in H. hangzhouensis and that the α strand is replicated first in O. huwena. Because the parental strand is displaced by its daughter strand and remains single-stranded until the lagging strand is synthesized, the strand displaced earlier will presumably be more susceptible to hydrolytic deamination. As replication of DNA is strand-asymmetric due to antiparallel alignment of the complementary strands, mutations are expected to occur in a strand-asymmetric manner (Tanaka and Ozawa 1994; Saccone et al. 1999). Therefore, the difference of nucleotide composition between the two strands may be related to the asynchronous synthesis of the two mitochondrial strands.

Protein-Coding Genes

Size and Sequence Similarity

The lengths of most protein-coding genes in H. hangzhouensis and O. huwena mtDNAs are either identical to or slightly smaller than their Limulus counterparts. Sequence comparison among H. hangzhouensis, O. huwena, and Limulus revealed cox1 as the most conserved gene and atp8, nad2, and nad6 as the least conserved genes (Table 2). The protein-coding genes of the two species have an arrangement commonly observed among arthropods (Crease 1999).

Table 2 Comparison of Mitochondrial Proteins of Heptathela hangzhouensis, Ornithoctonus huwena, and Limulus polyphemus
Full size table

Translation Initiation and Termination Codons

The 5′ ends of protein-coding genes of the two species were inferred to be at the first legitimate start codon (ATN, GTG, TTG, GTT [Wolstenholme 1992]) in the open reading frame (ORF), except that of cox2 in H. hangzhouensis, which appears to use the infrequent start codon CTG (Table 2), and cox1 in O. huwena, which appears to use the nonstandard start codon TTA (Table 2). The unusual start codons were determined for cox1 and cox2 based on the sequence alignment among our two sequences and the 5′ ends of cox1 and cox2 in Limulus. Unusual start codons CTG and TTA were also assigned in a similar fashion in Balanoglossus (Castresana et al. 1998) and Limulus (Lavrov et al. 2000a), respectively. Regardless of the actual initiation codon, all proteins were assumed to start with formyl-Met, as has been demonstrated for other mitochondrial systems (Smith and Marcker 1968; Fearnley and Walker 1986).

Only 10 protein-coding genes in the two species are inferred to have complete termination codons (Table 2). Among them, the termination codons of atp8 and nad1 in both O. huwena and H. hangzhouensis are entirely within the sequences of the downstream genes. But only the protein-coding gene terminus of atp8 was inferred to be located within the sequence of the downstream gene encoded on the same strand. Although four genes (cob, nad2, nad3, and nad4L) in H. hangzhouensis and three genes (cox2, nad2, and nad4) in O. huwena do not have complete termination codons, their next one or two nucleotides of downstream gene constitute a stop codon which could be used as such if translation occurred before the genes adjacent to the 3′ ends of these genes were cleaved from the common transcript (Ojala et al. 1981). In three genes (cox2, nad4, and nad5) of H. hangzhouensis and the four (cox3, cob, nad3, and nad5) of O. huwena, a solitary T at the inferred 3′ end of a gene abuts the 5′ end of a downstream tRNA gene, having no downstream gene to complete truncated stop codons. The mRNA is probably polyadenylated to form a UAA stop codon after the downstream tRNA is cleaved from the polycistronic transcript (Lavrov et al. 2000a). The gene atp6 of H. hangzhouensis and the gene nad4L of O. huwena have a truncated stop codon (T), have no downstream gene to complete them, and do not abut tRNA. In mammals the transcription products of mtDNA are primarily polycistronic. Individual gene transcripts are produced by precise cleavage, and protein-coding gene transcripts ending in U acquire complete termination codons by polyadenylation (Ojala et al. 1981). So these truncated stop codons may also serve as functional stop codons, as similar scenarios have been reported in Limulus (Lavrov et al. 2000a).

A significant degree of physical overlap has been observed in the genes from both species. There are overlaps between many genes of the two species, maybe because of their condensed genomes. However, the overlap between protein-coding genes only happened in atp8–atp6. Among them, a 7-nt (nucleotide) overlap is present between them, which has been reported in all arthropod mtDNA sequences published except that of Apis (Crozier and Crozier 1993), where the overlap is 19 nt. Transcriptional mapping analysis in several species of animals has demonstrated the presence of bicistronic transcripts for the atp8–atp6 gene pairs (Ojala et al. 1980; Berthier et al. 1986). It was also demonstrated that both of these genes are fully translated into proteins (Fearnley and Walker 1986). Taanman (1999) has suggested that the atp8 mRNA, if single, may be too short to be translated efficiently. Therefore there may be some selective advantage in having atp8–atp6 adjacent (Lavrov et al. 2000a).

Codon Usage

The frequencies of nucleotides in all three codon positions for each of the strands are presented in Fig. 3. In H. hangzhouensis, the percentages of T’s and G’s are lower in the α strand than in the β strand, whereas the percentages of A’s and C’s are higher. The more frequent T’s and infrequent A’s in the α strand are an exception in third codon positions. In O. huwena, the percentage of T’s and G’s is higher in the α strand than in the β strand, whereas the percentages of A’s and C’s are lower. The exceptions are the equal frequency of T’s at the first codon position between the two strands and the high percentage of A’s at second codon position in the α strand.

Figure 3
figure 3

Nucleotide composition comparison of the first, second, and third codon positions of protein-coding genes encoded by each strand of mtDNAs of Heptathela hangzhouensis (above) and Ornithoctonus huwena (below). The darker shaded bars show the nucleotide composition of the α strand and the lighter shaded bars show the nucleotide composition of the β strand.

Full size image

Selection may explain the high frequency of T at second codon positions in both strands, since codons with T in such position specify predominantly hydrophobic (nonpolar) amino acids, which are essential for the membrane-associated proteins encoded by mtDNA (Lavrov et al. 2000a). In H. hangzhouensis, the frequencies of G and C at third codon position in the two strands are greatly reduced. In O. huwena, the frequency of C at third codon position in the α strand and G at third codon position in the β strand are greatly reduced. Both reductions appear to reflect the mutational pattern in the genome, as nucleotides at third codon positions are under the least selective pressure.

We compared individual amino acid frequency in each strand and found significant differences for 8 of the 20 amino acids in H. hangzhouensis and in O. huwena (Tables 3 and 4). However, when we grouped amino acids based on the chemical nature of their side chains (nonpolar, polar, acidic, or basic) and compared the frequency of each group in each strand, we found no significant difference in H. hangzhouensis and three of the four groups in O. huwena (Tables 3 and 4). Only the group of acidic amino acids in O. huwena was significantly different in frequency between the strands.

Table 3 Amino Acid Composition of Protein-Coding Genes in Heptathela hangzhouensis
Full size table
Table 4 Amino Acid Composition of Protein-Coding Genes in Ornithoctonus huwena
Full size table

Ribosomal RNA Genes

The putative sizes of the inferred rrnS and rrnL in H. hangzhouensis are 698 and 1119 nt, respectively, while those in O. huwena are 666 and 1048 nt, respectively. Although the entire rRNA genes of the two species are truncated about 100–250 nt with respect to Limulus, H. hangzhouensis and Limulus share 75% sequence similarity for the last 594 nt of the 3′ end of the rrnL and 73% sequence similarity for the last 419 nt of the 3′ end of the rrnS. Similarly, O. huwena and Limulus share 71% sequence similarity for the last 497 nt of the 3′ end of the rrnL and 66% sequence similarity for the last 422 nt of the 3′ end of the rrnS. Furthermore, rrnL of H. hangzhouensis and O. huwena share highly conserved stem-and-loop-region sequence motifs including the peptidyl transferase center, as reported by Masta (2000) and Hedin and Maddison (2001). Masta (2000) also found truncated rrnL in the spiders, which have tRNAs lacking the TΨC arm and suggested that normal functioning of tRNAs with a TV-replacement loop may require changes in ribosomal RNA structures.

Transfer RNA

Based on the sequence similarity with Limulus and the potential secondary structures of these tRNA genes, there are 22 putative tRNA genes in each species, as there are in most other published metazoan mtDNAs. In H. hangzhouensis, 9 tRNA genes are β strand-encoded tRNAs and 13 tRNA genes are α strand-encoded tRNAs, which is identical in Limulus. In O. huwena, however, 10 tRNA genes are encoded by the β strand and 12 tRNA genes are encoded by the α strand, while 7 have been translocated to other positions (Fig. 2), compared with Limulus.

When these tRNA genes of H. hangzhouensis and O. huwena are folded into putative secondary structures (Figs. 4 and 5), the tRNA genes of H. hangzhouensis have fewer mismatches in the aminoacyl acceptor arms compared with those of O. huwena. Most of the tRNAs appear to be truncated and lack one of the arms found in canonical tRNAs. Instead, these arms are replaced by a TV-replacement loop, as previously described for nematodes (Wolstenholme et al. 1987; Okimoto and Wolstenholme 1990; Navajas et al. 2002) and some mollusks (Terrett et al. 1996; Yamazaki et al. 1997). At the same time, the loss of the TΨC arm or DHU arm was also found in the Arthropoda (Mitchell et al. 1993; Hedin 1997b; Masta 2000; Lavrov et al. 2000b; Hormiga et al. 2003). It is most important that the tRNA secondary structures proposed for the two species may not be unusual among spiders. Mitochondrial sequences for tRNALeu(CUN) and tRNA val for Salticidae (Masta 2000) have been reported, and tRNALeu(CUN) for Nesticidae has been published by Hedin (1997b). All of them can be folded into secondary structures lacking the TΨC arm. Both Salticidae and Nesticidae are in the derived suborder Araneomorphae of the order Araneae; we found that the mygalomorph spider and the mesothelid spider share this attribute. Our observations from the mygalomorph spider and the mesothelid spider further enlarged this family of tRNAs.

Figure 4
figure 4

Anticipated secondary structures of the 22 inferred tRNAs of Heptathela hangzhouensis. AT/GC pairs are indicated by dashes, and GT pairs are indicated by dots.

Full size image
Figure 5
figure 5

Anticipated secondary structures of the 22 inferred tRNAs of Ornithoctonus huwena. AT/GC pairs are indicated by dashes, and GT pairs are indicated by dots.

Full size image

Large Noncoding Region

The longest noncoding region in H. hangzhouensis mtDNA is 340 bp. It is located between rrnS and trnI and is significantly more AT-rich (80.6%) than the rest of the genome as previously reported in Limulus. Within this region, a 51-nt sequence close to rrnS has the potential to form a stem-loop structure with a 21-bp stem and a 9-bp loop (Fig. 6). In O. huwena, the largest noncoding region is 396 bp long. It is located between trnQ and trnM and is 68.2% A+T, which is slightly less AT-rich than the rest of the genome. The absence of such a higher A+T content had also been found in four mites (Navajas et al. 2002), the hexapod Triatoma dimidiata (Reduviidae) (Dotson and Beard 2001), and the myriapod Thyropygus sp. (Harpagophoridae) (Lavrov et al. 2002). These comparatively lower A+T contents may imply a higher efficiency of enzymatic complex involved in the DNA denaturation before replication–transcription (Navajas et al. 2002). Within O. huwena, a 26-nt sequence close to rrnS has the potential to form a stem–loop structure with an 8-bp stem and a 10-bp loop (Fig. 6A). The other stem–loop structure may exist in O. huwena (Fig. 6B) with a 14-nt stem and an 8-nt loop, to the 3′ flanking of the 26-nt stem–loop structure. A feature of the sequences of stem–loop structures in the two species is the ability to form a secondary structure that is less stable. The function of the region, if any, is unknown. However, similar stem–loop structures in the region have been shown to initiate replication in vertebrates (Sumida et al. 2001). Whether the large noncoding regions in spiders serving the same function remains elusive.

Figure 6
figure 6

The potential stem–loop structures in the large noncoding regions of Heptathela hangzhouensis and Ornithoctonus huwena. A The potential stem–loop structure close to rrnS in O. huwena. B The other stem–loop structure in O. huwena near the 3′ flanking of the 26-nt stem–loop structure.

Full size image

Discussion

Gene Rearrangement

In general, a large number of possible gene rearrangements make it improbable that the same order would arise independently; shared–derived arrangements most likely indicate common ancestry (Boore et al. 1995). Thus, gene rearrangement can also be used as an effective parameter to establish phylogenetic relationships.

Moritz (1987) proposed the most commonly recognized model for gene order change. Figure 7 shows a model modified for the gene rearrangement (trnN–trnA–trnS1–trnR) of O. huwena, compared with the arrangement in Limulus and H. hangzhouensis. The mechanism of the rearrangement (trnW– trnY– trnC) is similar to what took place in trnN–trnA–trnSl–trnR (genes underlined are transcribed in the reverse direction). First, the portions of the mitochondrial genome may have become duplicated, perhaps by slipped-strand mispairing or by erroneous termination of replication (Boore and Brown 1998). Next, one copy of each of the duplicated genes is lost by deletion. Depending on which copy of these genes is lost, the original arrangement may be restored or a rearrangement may occur. Three transpositions (trnL2, trnT, and trnQ) may take place by a duplication of a large region followed by loss of all but one or a few genes in a single or several deletion events (Lavrov et al. 2002). The inverted trnI may have arisen by local small inversion, a kind of intramitochondrial recombination, necessarily invoking breakage and rejoining of the mitochondrial genome (Dowton and Austin 1999). The finding of subgenomic minicircle in a nematode (Lunt and Hyman 1997) indicates the presence of such intramitochondrial recombination.

Figure 7
figure 7

Proposed mechanism of gene rearrangements in Ornithoctonus huwena, compared with the arrangement in Limulus polyphemus and Heptathela hangzhouensis. The hypothetical duplication of the trnA–trnR–trnN–trnS1 and trnA–trnS1–trnR–trnN that could have predated the rearrangement, and the paths of gene loss from this hypothetical duplication that would lead to a return to the original order or that found in O. huwena.

Full size image

To date, seven complete mtDNA gene arrangements have been published for Chelicerata. The gene arrangements in Ixodes hexagonus (Arachnida: Ixodida: Ixodidae) (AF081828), Ixodes persulcatus (Arachnida: Ixodida: Ixodidae) (AB073725), and Ornithodoros moubata (Arachnida: Ixodida: Argasidae) (AB073679) are identical to those in Limulus and H. hangzhouensis. In an other two species of Ixodidae, Rhipicephalus sanguineus (Arachnida: Ixodida: Ixodidae) (AF081829) and Boophilus microplus (Arachnida: Ixodida: Ixodidae) (AH007623), an eight-gene block has been translocated to the region between nad3 and nad5 (Fig. 8), trnC has reversed to the region between trnL1 and trnM, and trnL1 has been translocated to the region between trnS2 and trnC (Black and Roehrdanz 1998; Campbell and Barker 1998) (Fig. 8). Navajas et al. (2002) have found translocations of rrnS, trnQ, trnY, trnP and trnC and translocations and inversions of trnH and trnS2 in Varroa destructor (Arachnida: Mesostigmata: Varroidae) (AJ493124) (Fig. 8). These results, along with our data, revealed four different mitochondrial gene arrangements in chelicerates. The rearrangement of an eight-gene block in R. sanguineus and B. microplus and the three tRNA rearrangements in O. huwena were found between nad3 and nad5. This may imply that the region between nad3 and nad5 is a hot spot of rearrangement in chelicerates. Another hot spot is around the large noncoding region such as in many vertebrates. As Boore (1999) suggested, the most common changes in arthropods are genes near the large noncoding region or the tRNAs in the region between nad3 and nad5 that is trnA–trnR–trnN–trnS1–trnE–trnF in Drosophila. Because rearrangements often occur near the region where replication is initiated by erroneous termination of replication, provided that the large noncoding region probably initiates replication, we suggest that the second-strand origin in arthropod mtDNA is perhaps near the region between nad3 and nad5 (Boore 1999), while the first-strand origin is like around the large noncoding region.

Figure 8
figure 8

Comparisons of gene arrangements in mtDNAs of six chelicerates (Ixodes hexagonus, Ixodes persulcatus, Ornithodoros moubata, Rhipicephalus sanguineus, Boophilus microplus, and Varroa destructor). Genes are labeled just as in Fig. 2.

Full size image

We also studied the phylogenetic relationship among the three high-level taxa of Araneae using the partial mitochondrial gene rearrangement. The rrnL–trnL1–trnL2–nad1 arrangement shared by Limulus and the mesothelid spider H. hangzhouensis is thought to be ancestral for arthropods (Boore et al. 1995). In the mygalomorph spider O. huwena, trnL2 has been translocated to the position between trnN and nad3 and exhibits an rrnL–trnL1–nad1 arrangement. The latter arrangement is also shared by araneomorph taxa: Habronattus sp. of Salticidae (Masta 2000), Nesticus sp. of Nesticidae (Hedin 1997b), and Orsonwelles sp. of Linyphiidae (Hormiga et al. 2003). The rrnL–trnL1–nad1 arrangement therefore is a putative synapomorphy joining the Mygalomorphae with the Araneomorphae. However, the location of the trnL2 in the two araneomorphs is still unknown; further study on the location trnL2 in the two taxa is needed.

Transfer RNA

Masta (2000) first discovered the modified truncated structure of spiders tRNAs. Presumably, the tRNAs lacking the TψC arm in Araneae have certain tertiary interactions to maintain their functionality (Masta 2000). Kumazawa et al. (1996) proposed that selective pressure to shorten the TψC arms was balanced by selective pressure to maintain TψC loops because of functional interaction with DHU loops. Even for nematode tRNAs completely lacking the TψC arm, Watanabe et al. (1994) had shown evidence of tertiary interactions between the TV-replacement loop and the DHU arm, suggesting that these tRNAs maintain their functionality. On the other hand, Ohtsuki et al. (2001, 2002) found the atypical EF-Tul to be special for nematode mitochondrial tRNAs that lack the TψC arm and the EF-Tu2 to bind only to tRNAs that lack the DHU arm. The two translation elongation factor Tu (EF-Tu) homologues may suggest the co-evolution of a structurally simplified RNA and the cognate RNA-binding protein and may explain the structural divergence in animal mitochondrial tRNAs such as these special tRNAs in Araneae. Based on the co-evolutionary viewpoint, there might be special tRNA-binding proteins in Araneae to make tRNAs lacking the TΨC arm or DHU arm functional.

Among the tRNAs of the two spiders, only 14 inferred tRNA genes in H.hangzhouensis encode a fully paired aminoacyl acceptor stem. The other 30 genes encode tRNAs with up to four mismatches in the aminoacyl acceptor stem, as previously reported for spiders tRNAs, with up to three mismatches (Hedin 1997b; Masta 2000), and centipede tRNAs, with up to five mismatches in the aminoacyl acceptor stem (Lavrov et al. 2000b). Because a well-paired aminoacyl acceptor stem is required for proper tRNA functioning, Lavrov et al. (2000b) studied cDNA sequences from tRNAs of Lithobius forficatus and found the editing of up to 5 nt at their 3’ ends. This editing appears to occur by a novel mechanism, with the 5’ end of the aminoacyl acceptor arm being used as a template for the de novo synthesis of the 3’ end, presumably by an RNA-dependent RNA polymerase. Therefore, tRNAs from the two spiders we have analyzed may also undergo posttranscriptional processing to acquiring their functionality. Such posttranscriptional editing mechanisms (Yokobori and Pääbo 1995a, b; Lavrov et al. 2000b) may correct potential mismatches in the aminoacyl acceptor stems.

The extensive overlaps with the downstream genes in O. huwena also occurred in L. forficatus (Lavrov et al. 2000b), but the overlaps in O. huwena are much longer than those of L. forficatus. The excision of the unedited pre-tRNAs from polycistronic transcript must be an active process and the enzymes involved should recognize the unusual secondary structures formed by these sequences. When overlaps of tRNA genes are excessively long, excisions at different points may yield one or the other functional tRNA in a mutually exclusive manner.

Mitochondrial genome may be particularly susceptible to accumulation of mutations, and mitochondrial tRNAs have been shown to accumulate deleterious mutations relative to their nuclear tRNA counterparts (Lynch 1996). Macey et al. (1997) proposed that replication slippage might cause loss of tRNA arms. Yamazaki et al. (1997) suggested that mt-tRNA genes would have changed their secondary structure due to reduction of their chain lengths under the pressure for genome minimization. The truncated structures of the tRNAs from our study provide additional support to these ideas.

While this paper was under review, a paper reporting complete mitochondrial genome of the spider Habronattus oregonensis was published. In that paper, Masta and Boore (2004) revealed seven rearranged tRNAs, the extremely truncated tRNAs, and the overlaps among many of the mitochondrial gene sequences. The rearrangements in H. oregonensis occurred similarly as in O. huwena, except for trnI being translocated to a different position compared with those of O. huwena. Because both O. huwena and H. oregonensis are mygalomorph spiders and share most rearrangements, their finding appears to be consistent with our findings that the rearrangements join the mygalomorph with the araneomorph and that Mesothelae is basal to opisthothelids. In H. oregonensis, most tRNAs are extremely truncated, and many of the gene sequences overlap with one another and are truncated as in O. huwena.