Introduction

Mitochondrial (mt) genomes of animals are usually circular, 15–20 kb long, and have 37 genes plus a large noncoding region (LNR; Boore 1999). Nucleotide (nt) sequences of mt genomes are often used in evolutionary studies. In recent years, arrangements of mt genes have also been used in evolutionary studies. Early work showed that recombination of mtDNA was extremely rare in mammals (Clayton et al. 1974; Zuckerman et al. 1984; Hayashi et al. 1985); this was generally thought to be the case for other animals too. Indeed, the view that animal mt genomes recombine rarely was a key part of the rationale for using mtDNA sequences in evolutionary biology. Further, whether or not mt genomes of animals recombine, and if they recombine, how they recombine, is critical to our understanding of the mechanisms of gene rearrangements in mt genomes (Boore 2000; Dowton and Campbell 2001).

Recent studies have challenged the view that mt genomes of animals do not recombine or recombine rarely. Lunt and Hyman (1997) reported illegitimate (i.e., nonhomologous) intra-mtDNA recombination in a nematode: two double-strand breaks cleaved out a section of the large noncoding region; the two ends of this section then rejoined to form a minicircle. Tang et al. (2000) showed that in long-term cultured human cell lines the mt genomes with partial duplications could generate wild-type mt genomes and mt genomes with sections deleted, apparently by illegitimate intra-mtDNA recombination. Ladoukakis and Zouros (2001) showed that in heteroplasmic male mussels the male-type and female-type mt genomes could recombine by homologous recombination. The mt genomes of humans may also recombine, but this is controversial. Awadalla et al. (1999) reported that the linkage disequilibrium declined as the physical distance between two polymorphic sites in a mt genome increased. This was interpreted as evidence for homologous recombination in human mt genomes. However, four subsequent papers reported the reanalyses of the data of Awadalla et al. (1999) with alternative methods of measuring linkage disequilibrium (Jorde and Bamshad 2000; Kivisild and Villems 2000; Kumar et al. 2000; Parsons and Irwin 2000). These authors argued that the linkage disequilibrium did not decline as the physical distance between two polymorphic sites increased. More recently, Kraytsberg et al. (2004) showed that in the muscle tissue of a patient with paternal inheritance of mtDNA about 0.7% of the mt genomes contain both maternal and paternal segments. This indicates homologous recombination between the paternal-type and the maternal-type mt genomes in this patient. In summary, these studies indicate that recombination of animal mt genomes may be more common than previously thought but this remains controversial.

The subphylum Chelicerata is one of the four major lineages of the phylum Arthropoda. The Chelicerate comprises the horseshoe crabs, mites, ticks, spiders, scorpions, sea scorpions, sea spiders, and their kin. The nt sequences of entire mt genomes have been reported for 13 chelicerates: 1 horseshoe crab (Lavrov et al. 2000), 1 jumping spider (Masta and Boore 2004), 1 honeybee mite (Evans and Lopez 2002), 3 soft ticks (Shao et al. 2004, 2005), 4 prostriate ticks, and 3 metastriate ticks (Black and Roehrdanz 1998; Shao et al. 2004, 2005). The horsehoe crab, the soft ticks, and two prostriate ticks, Ixodes hexagonus and I. persulcatus, have exactly the same gene content and gene arrangement as the hypothetical ancestor of the arthropods (Lavrov et al. 2000), whereas the jumping spider, the honeybee mite, and the metastriate ticks have novel arrangements of mt genes relative to the hypothetical ancestor of the arthropods. The other two prostriate ticks, I. holocyclus and I. uriae, have the same gene arrangement as the hypothetical ancestor of the arthropods except that these two Australasian Ixodes ticks have duplicate LNRs, i.e., two separated LNRs with identical or highly similar nt sequences. The metastriate ticks also have duplicate LNRs. The presence of duplicate LNRs is thought to be a synapomorphy for the Australasian Ixodes ticks (Shao et al. 2005). The rearrangements of mt genes and the presence of duplicate LNRs are thought to be synapomorphies for the metastriate ticks (Black and Roehrdanz 1998; Campbell and Barker 1998; Murrell et al. 2003).

All of the Chelicerata studied so far, except the horseshoe crab and the jumping spider, are from the Parasitiformes. The Parasitiformes is one of the three major divisions of the Acari (the mites and ticks [Lindquist 1984]). Nucleotide sequences of entire mt genomes or arrangements of mt genes are not available for any mites from the other two divisions of the Acari, the Acariformes and the Opilioacariformes. Yet about two thirds of the 45,000 described species of Acari are in the Acariformes. In order to get a better understanding of the evolution of mt genomes in the Acari, we sequenced the entire mt genome of the chigger mite, L. pallidum, which is from the Acariformes. L. pallidum has a novel mt gene content and differs substantially in gene arrangement from the hypothetical ancestor of the arthropods and the other species of Acari studied. Here, we present the novel features of the mt genome of L. pallidum and discuss their implications for the transcription of mt genes, the replication of mt genomes, and the phylogeny of the Acari. Further, we propose a model of illegitimate inter-mtDNA recombination to account for the novel gene content and gene arrangement of L. pallidum.

Materials and Methods

Source of Mites, DNA Extraction, PCRAmplification, and nt Sequencing

The chigger mite, L. pallidum, was from a laboratory colony in Saitama, Japan (see Takahashi et al. 1994). Three mites were used in DNA extraction and the method of DNA extraction was described by Shao et al. (2004). Fragments of two mt genes, cob and cox1, were amplified with primers that were conserved among arthropods. Nucleotide sequences from these fragments were then used to design four PCR primers that were specific to L. pallidum. The mt genome of L. pallidum was then amplified in two fragments (∼8 kb each) by long-PCR with the following primers: (1) Lp-cob-f (5′-TATGCAATTTTGCGATCAATTCCT-3′) with Lp-cox1-r (5′-ATCCGGGCAAAATAAGAATATACA-3′); and (2) Lp-coxl-f (5′-CTATGTTGTTGCTCATTTCCACTA-3′) with Lp-cob-r (5′-TAGTAATTACTGTTGCACCTCAAA-3′). Each 50-μl PCR contained 1 μl (∼2.5 ng) of total DNA, 400 μM of each dNTP, 1 μM of each prime, and 2.5 units of LA Taq polymerase (Takara Bio). GeneAmp PCR System 9700 (PE Biosystems) was used in PCR and the reaction conditions were 95°C for 1 min, followed by 37 cycles of 95°C for 40 sec, 57°C for 45 sec, 68°C for 10 min, and then 68°C for 10 min. Long-PCR fragments were sheared by sonication and then separated by gel electrophoresis. Fragments of the size 1.5–2.5 kb were excised from gels, purified, blunt-ended, ligated into pUC-18 vectors, and then transformed into competent cells. One hundred clones were picked at random for each long-PCR fragment and sequenced with BigDye Terminator Cycle Sequencing Kit (Applied Biosystems).

Sequence Analysis

The nt sequence of the entire mt genome of L. pallidum was assembled by aligning sequences from the 200 clones (100 clones from each fragment) with AutoAssembler (Applied Biosystem). Both strands of the mt genome were sequenced. On average, each nt was sequenced seven times. Protein-coding genes were identified by BLAST searches of GenBank (Altschul et al. 1990) and confirmed by comparisons of hydrophilicity profiles of putative proteins (Hopp and Woods 1981) with those of the fruit fly, Drosophila yakuba (Clary and Wolstenholme 1985). BLAST searches of GenBank did not identify atp8, nad4L, and nad6 of L. pallidum; these three genes were identified by comparing the hydrophilicity profiles of three “orphan” open reading frames of L. pallidum with the hydrophilicity profiles of ATP8, NAD4L, and NAD6 of D. yakuba. rRNA genes were identified by BLAST searches of GenBank and confirmed by comparison with rRNA genes of D. yakuba. tRNA genes were identified with tRNAscan-SE (Lowe and Eddy 1997) or by eye and confirmed by comparisons with tRNA genes of the horseshoe crab, Limulus polyphemus (Lavrov et al. 2000). The mtDNA sequence of L. pallidum has been deposited in DDBJ (accession number AB180098).

Results and Discussion

General Features of the mt Genome of L. pallidum

The mt genome of L. pallidum is circular and has 16,779 bp. In addition to the 37 genes that are typical of the mt genomes of animals, the mt genome of L. pallidum has an extra gene for large subunit rRNA (rrnL), a pseudogene for small subunit rRNA (PrrnS), and four large noncoding regions (LNR). This genome is the third largest of the 49 mt genomes of arthropods that are in GenBank. Only the mt genomes of Drosophila melanogaster (19,517 bp; accession number NC_001709) and the kissing bug, Triatoma dimidiata (17,019 bp; NC_002609), are larger than the mt genome of L. pallidum. For D. melanogaster and T. dimidiata, the large sizes of the noncoding regions (4,600 and 2,472 bp, respectively) account for the large sizes of these mt genomes. For L. pallidum, however, the extra genes and the noncoding regions together (3,414 bp in total) account for the large size of the mt genome, In contrast to the large size of the mt genome, most mt genes of L. pallidum are shorter than those of other arthropods. The proteins inferred from the 13 protein-coding genes of this genome are shorter or substantially shorter (by 0.4–18.3%) than those of D. yakuba. Comparison of hydrophilicity profiles indicated that some domains of the proteins of D. yakuba are not present in L. pallidum (Table 1). All the 22 tRNA genes and rrnS of L. pallidum are substantially shorter than their counterparts in D. yakuba (see below).

Table 1 Comparison of the 13 putative mitochondrial proteins of the chigger mite, Leptotrombidium pallidum (L.p), and the fruit fly, Drosophilla yakuba (D.y)
Full size table

Novel mt Gene Arrangement of L. pallidum and Its Implications for mt Gene Transcription, mt Genome Replication, and Acari Phylogeny

The arrangement of genes in the mt genome of L. pallidum differs drastically from that of the hypothetical ancestor of the arthropods, which has been retained in the horseshoe crab, the soft ticks, and at least two prostriate ticks, I. hexagonus and I. persulcatus (Fig. 1). Thirty-four of the 43 gene boundaries in L. pallidum are novel for an arthropod. Four features of the mt genome of L. pallidum are particularly noteworthy. First, the two genes for NADH dehydrogenase subunits 4 and 4L (nad4 and nad4L) are not adjacent but are 4,085 bp apart. Thus, these two genes may have their own mature mRNAs rather than share a single mature mRNA as is the case in fruit flies (Berthier et al. 1986) and possibly in most other arthropods in which these two genes overlap. Second, there are two copies of rrnL genes but one copy of rrnS gene. Thus, the large subunit rRNA may be expressed more than the small subunit rRNA. Third, there are four nearly identical LNRs. The transcription of mt genes and the replication of the mt genome of L. pallidum may, therefore, start at multiple sites if each of the four LNRs contains the initiation sites for transcription and/or replication, as is the case for the LNRs in the mt genomes of mammals (Taanman 1999) and fruit flies (Goddard and Wolstenholme 1980). Stem–loop structures in noncoding regions are associated with the initiation of replication or transcription (Wong and Clayton 1985; Clayton 1991; L’Abbe et al. 1991). Such structures are also present on both strands of the four LNRs of L. pallidum (Fig. 2A), Fourth, rrnS is immediately downstream of rrnL whereas in most animals rrnS is upstream of rrnL. In humans the two mt rRNA genes are transcribed together at a rate much higher than other mt genes; the transcription initiates at a site upstream of rrnS in the LNR and terminates at a site downstream of rrnL in a tRNA gene (Taanman 1999). A conserved heptamer, TGGCAGA, was thought to be the termination site of the transcription of the two rRNA genes in all eukaryotes examined (Valverde et al. 1994; Richard et a1. 1998). This heptamer is always present at the 5′-end of the gene that is immediately downstream of rrnL; if the gene immediately downstream of rrnL is a tRNA gene, this heptamer is present at positions 8–14 of the 5′-end of that tRNA gene. In L. pallidum, this heptamer was not present at the 5′-end of rrnS or PrrnS, which is adjacent to the 3′-end of rrnL1 or rrnL2. However, a heptamer, TGGTGTA, was present at positions 8–14 of the 5′-end of trnQ, which is adjacent to rrnS. The first three and the last nts of this heptamer are identical to those of the conserved heptamer that has been found in other eukaryotes examined. We propose that rrnL1 and rrnS are transcribed together in L. pallidum: the transcription initiates at a site upstream of rrnL1 in one of the four LNRs (LNR1) and terminates at the heptamer, which is downstream of rrnS in trnQ. This indicates that there may be functional constraints on the rearrangement of the two mt rRNA genes that keep the two genes to be close to each other and on the same strand, regardless of their relative positions.

Figure 1
figure 1

Comparison of the gene content and gene arrangement of the mt genomes of the chigger mite, Leptotrombidium pallidum, the hypothetical ancestor of the arthropods and other species of the Acari studied to date. Abbreviations of protein-coding and rRNA genes: atp6 and atp8, ATP synthase subunits 6 and 8; cox1–3, cytochrome oxidase subunits 1–3; cob, cytochrome b; nad1-6 and nad4L, NADH dehydrogenase subunits 1-6 and 4L; rrnL and rrnS, large and small rRNA subunits. tRNA genes are shown with the single-letter abbreviations of their corresponding amino acids. The two tRNA genes each for leucine and serine are L 1 (anti-codon sequence nag), L2 (yaa), S1 (nct), and S2 (nga). The circular genomes were linearized at the 5′ end of cox1 to aid comparison. Genes are transcribed from left to right except those underlined, which are transcribed from right to left. Asterisks indicate genes or noncoding regions that are present in L. pallidum and metastriate ticks but not in the hypothetical ancestor of the arthropods. Dark gray–shaded boxes indicate genes which changed locations relative to the arrangement of mt genes of the hypothetical ancestor of the arthropods (Lavrov et al. 2000). Light gray–shaded boxes indicate genes that changed both. location and orientation relative to those of the hypothetical ancestor of the arthropods. Triangles indicate the inversion of rrnS in the chigger mite, L. pallidum, and the honeybee mite, Varroa destructor. Thick arrows show the two 1,852-bp sections which have identical nt sequences but opposite orientations of transcription in L. pallidum.

Full size image
Figure 2
figure 2

(A) Stem–loops in the large noncoding regions (LNR) of the mt genome of the chigger mite, Leptotrombidium pallidum. The nts of the stem–loops are in boldface, whereas the nts that flank the stem–loops are in regular font. (B) Alignment of the nt sequences of the four large noncoding regions. Dashes indicate indels. Periods indicate nts that are identical to LNR1. Sequences that can form stem–loops are highlighted in gray.

Full size image

Comparisons of the mt gene arrangements of the Acari studied so far show that the chigger mite, L. pallidum (Acariformes), and the honeybee mite, Varroa destructor (Parasitiformes), share the inversion of rrnS. In all other Acari studied so far rrnS is in the orientation that is ancestral for arthropods. Further, V. destructor (Parasitiformes) does not share any gene boundaries nor gene rearrangements with the metastriate ticks which are also from the Parasitiformes (Fig. 1). Inversions of rRNA genes in mt genomes are rare and, thus are less likely to be homoplastic (Boore 1999). If the inversion of rrnS that is shared by the chigger mite (Acariformes) and the honeybee mite (Parasitiformes) is a synapomorphy, it would indicate that the honeybee mite (Parasitiformes) is more closely related to the chigger mite (Acariformes) than it is to the ticks (Parasitiformes), i.e., the Parasitiformes would be paraphyletic. This phylogenetic arrangement differs markedly from the current hypothesis of the phylogeny of the Acari that the Parasitiformes is monophyletic (Lindquist 1984). Broad sampling is needed to elucidate whether the inversion of rrnS in the chigger mite and the honeybee mite is a synapomorphy or evolved independently.

Unusual Secondary Structures of the mt tRNAs ofL. pallidum

Like those of most animals, the mt genome of L. pallidum apparently encodes 22 tRNA genes. However, the tRNAs inferred from these genes have unusual secondary structures (Fig. 3). Nine tRNAs of L. pallidum (C, Q, E, H, L2, F, P, T, and W) lack a T-arm and instead have a TV replacement loop. Ten tRNAs (A, R, N, G, I, L1, S1, S2, Y, and V) lack D-arms and have a D-arm replacement loop. Only three tRNAs (D, K, and M) can form cloverleaf-shaped structures but none of these appear to have a T-loop. These changes caused substantial size reduction (by 9–31%) of the tRNAs of L. pallidum compared to those of the horseshoe crab and the fruit fly. Most of the mt tRNAs of L. pallidum are even shorter than the nematode, Ascaris suum, of which all putative mt tRNAs lack either a T-arm or a D-arm (Table 2).

Figure 3
figure 3

Inferred secondary structures of the 22 tRNAs encoded by the mt genome of the chigger mite, Leptotrombidium pallidum. tRNAs are labeled with the abbreviations of their corresponding amino acids. The conventional cloverleaf secondary structures and the numbering of nts are after Sprinzl et al. (1989). Dashes indicate Watson–Crick bonds; dots indicate bonds between U and G. Circled nts are also present in the tRNAs of the horseshoe crab, Limulus polyphemus. The six pairs of nts that form a tertiary structure in tRNA-Phe of the nematode, Ascaris suum, are linked with lines in tRNA-Phe.

Full size image
Table 2 Comparison of the 22 putative tRNA genes of the chigger mite, Leptotrombidium pallidum (L.p), the horseshoe crab, Limulus polyphemus (Li.p), the fruit fly, Drosophila yakuba (D.y), and the nematode, Ascaris suum (A.s)
Full size table

For most animals, all 22 mt tRNAs can form cloverleaf-shaped secondary structures except tRNA-Ser(ncu), which lacks a D-arm and has a D-arm replacement instead (Wolstenholme 1992). In several exceptions, some of the 22 tRNAs, in addition to tRNA-Ser(ncu), lack either a T-arm or a D-arm, whereas the rest can form a cloverleaf-shaped structure (Yamazaki et al, 1997; Lavrov and Brown 2001; Masta and Boore 2004), Changes to the secondary structures of all the 22 mt tRNAs of L. pallidum have not been observed in arthropods before. Changes of this scale have only been observed previously in nematodes of the class Secernentea (Okimoto et a1. 1992; Wolstenholme et al. 1987).

The unusual putative structures and small sizes of the mt tRNAs of L. pallidum beg the question: Are these tRNA genes functional? The analyses below support our view that the mt tRNA genes of L. pallidum encode a complete set of the functional tRNAs. First, there is no evidence to date that tRNAs can be imported from cytoplasm into mitochondrion in arthropods, although we cannot exclude such a possibility in L. pallidum. Second, 21 of the inferred 22 putative tRNAs of L. pallidum have anticodon sequences identical to those of Li. polyphemus and D. yakuba (Table 2). The only difference is the anticodon sequence of tRNA-Lys, which is TTT in L. pallidum but CTT in Li. polyphemus and D. yakuba. However, the use of anticodon TTT in L. pallidum is consistent with the colon composition of the mt proteins of L. pallidum: there are 128 AAA but only 19 AAG. Third, only tRNA-Phe and tRNA-Ser(gct) have three mismatches, whereas the other 20 tRNAs have either no mismatches or one or two mismatches (Fig. 3), Fourth, most tRNAs of L. pallidum share about 50% of their nt sequences with their counterparts in Li. polyphemus, in which all tRNAs can form a conventional cloverleaf-shaped structure except tRNA-Ser(gcu) (see Lavrov et al, 2000). Fifth, it has been shown that in the nematode, A. suum, the mt tRNAs which lack either a T-arm or a D-arm are functional (Watanabe et al. 1994). Further, the six pairs of nts U8–A14, A9–A23, G10–GL2, A22–AL3, U15–AL4, and A26–GL1, which form the functional L-shaped tertiary structure of the tRNA-Phe of A. suum are also present in tRNA-Phe of L. pallidum: the first five pairs of nts in L. pallidum are identical to those in A. suum and the last pair in L. pallidum, A26–AL1, differs from to that in A. suum by one nt (Fig. 3). In both L. pallidum and A. suum the tRNA-Phe has a TV replacement-loop.

rRNA Genes, Large Noncoding Regions, and Illegitimate Inter-mtDNA recombination

There are two identical copies of rrnL (1,282 bp each), one copy of rrnS (601 bp), and a pseudogene of rrnS, PrrnS (212 bp), in the mt genome of L. pallidum. The size of the rrnL genes of L. pallidum is similar to that of Li. polyphemus (1,296 bp) and D. yakuba (1,326 bp), whereas rrnS of L. pallidum is only about 75% of the size of rrnS of Li. polyphemus (799 bp) and D. yakuba (789 bp). The two copies of rrnL in L. pallidum are 6,586 bp apart and identical in size and nt sequence but have opposite orientations of transcription. PrrnS is 5,772 bp away from rrnS and has identical sequence to the 5′-end of rrnS but has the opposite orientation of transcription. The presence of two copies of rrnL genes with identical sequences but opposite orientations in L. pallidum was confirmed by two specifically designed PCR tests. In each PCR test, we used only one rrnL specific primer (a forward primer in one test and a reverse primer in another test). Both PCR tests generated products of expected sizes (6.8 and 8.1 kb, respectively; data not shown). The four LNRs of L. pallidum are 460, 483, 509, and 472 bp, respectively, The four LNRs share 94–98% of the 454 bp in the middle parts (Fig. 2B). LNR1 and 2 have opposite transcription orientations to those of LNR3 and 4.

Two sections of the mt genome of L. pallidum have identical nt sequences but have opposite orientations of transcription. These two sections are 1,852 bp long and are 5,772 by apart (Fig. 1). Section #1 has part of LNR1, the entire rrnL1, and part of rrnS, whereas Section #2 has part of LNR4, the entire rrnL2, and the entire PrrnS. This is the first report of a mt genome of an animal that has two distantly separated sections of identical nt sequences but opposite orientations of transcription. This arrangement cannot be accounted for by homologous recombination (Tsukarnoto and Ikeda 1998; Rokas et al. 2003) or by the other two well-known mechanisms of mt gene rearrangement: (1) tandem duplication followed by deletion (Moritz and Brown 1986; Macey et a1. 1998; Boore 2000) and (2) illegitimate intra-mtDNA recombination (Lunt and Hyman 1997; Dowton and Campbell 2001). Homologous recombination may change the nt sequences of mt genes but would not change the gene content nor gene arrangement in the mt genomes. Tandem duplications may generate two identical sections in a mt genome but these two sections would always be adjacent to each other and in the same orientation. Illegitimate intra-mtDNA may change the position and/or orientation of a section of a mt genome but would not duplicate that section. In theory, tandem duplication and illegitimate intra-mtDNA recombination could act together to give a mt genome with two separated sections of identical or highly similar nt sequences but opposite orientations, like Sections #1 and #2 of L. pallidum. However, the most plausible explanation, in our view, is illegitimate inter-mtDNA recombination. Illegitimate inter-mtDNA recombination has been considered as a possible mechanism for gene rearrangement in animal mt genomes (Boore 2000) but there was little evidence for this mechanism until now. Indeed, L. pallidum is the first species of animal for which we need to invoke this mechanism to account for the novel mt gene content and gene arrangement. Evidence from experiments on recombination in nuclear and mt genomes further leads us to propose that the illegitimate inter-mtDNA recombination in L. pallidum may occur in two steps.

(1) Double-strand breaks: Consider two hypothetical, identical mt genomes, A and B (Fig. 4). Double-strand breaks cause a single break in A and two breaks in B. Thus, a section of genome B, SB, is cleaved out. Double-strand breaks are a type of DNA damage that can be caupdsed by ionizing radiation, chemicals, oxidative stress or routine cellular processes (Tsukamoto and Ikeda 1998; Lakshmipathy and Campbell 1999). Double-strand breaks are harmful to cells and may lead to cell death or uncontrolled cell growth if not properly repaired (Tsukamoto and Ikeda 1998; Pfeiffer et al. 2000).

Figure 4
figure 4

Model of illegitimate inter-mtDNA recombination. A and B are two identical mt genomes. Consider three double-strand breaks: one break in genome A and two breaks in genome B. A section in genome B, SB, is cleaved out. If SB then joins genome A at the break in an orientation opposite to that of its counterpart in genome A, SA, genome A will have two identical sections, SA and SB, which are in opposite orientations (a). This may have occurred in the chigger mite, Leptotrombidium pallidum, which has two identical sections, in opposite orientation (see Fig. 1). If SB joins genome A in the same orientation as that of SA, genome A will have two identical sections, SA and SB, in the same orientation (b). This may have occurred in several animals which have duplicate large noncoding regions (see Shao and Barker 2003). Another way of end-joining, illegitimate intra-mtDNA recombination, will lead to the minicircle SB (c), which was reported in a nematode (Lunt and Hyman 1997).

Full size image

(2) End-joining repair: Double-strand breaks are repaired in two ways in eukaryotes: (1) homologous recombination and (2) illegitimate recombination (non-homologous end-joining; Tsukamoto and Ikeda 1998; Pfeiffer et al. 2000; van Rijk and Bloemendal 2003). Nonhomologous end-joining is apparently the main way that double-strand breaks are repaired in the nuclei of mammalian cells (Tsukamoto and Ikeda 1998). Repair of double-strand breaks in animal mitochondria is still poorly understood. However, it has been shown in mammals that mt protein extracts can catalyze end-joining of linearized DNA molecules. This indicates that end-joining repair of double-strand breaks may also occur in mitochondria (Lakshmipathy and Campbell 1999). If the cleaved section, SB, joins genome A at the break point in an orientation that is opposite to that of its counterpart in genome A, SA, genome A will have two identical sections, SA and SB, in opposite orientations (Fig. 4a). The mt genome of the chigger mite, L. pallidum, may have been formed in this way. There are, however, other possible outcomes of this type of repair. One outcome is SA and SB in the same orientation in genome A (Fig. 4b). This type of gene arrangement was found in several animals which have well-separated duplicate LNRs in the same orientations of transcription (Kumazawa et al. 1998; Shao and Barker 2003). However, tandem duplication followed by deletions of genes could also account for this type of gene arrangement (Kumazawa et al. 1998). Another outcome is the two ends of SB rejoining to form a minicircle (Fig. 4c); this was reported in a nematode (Lunt and Hyman 1997).

Three questions remain unanswered: (1) What is the phylogenetic extent of the novel mt gene content and gene arrangement observed in L. pallidum? (2) Is the 16,779-bp mt genome we sequenced with novel gene content and gene arrangement fixed in L. pallidum, or is it an intermediate type of mt genome present only in some individuals or populations of L. pallidum? and (3) Are the identical nt sequences of the two distantly separated sections (Sections #1 and #2) in the mt genome of L. pallidum attributable to concerted evolution or a very recent event of inter-mtDNA recombination? A broad sample of L. pallidum from different populations, and their kin, is needed to address these questions.