Evolution of extensively fragmented mitochondrial genomes in the lice of humans.

Bilateral animals are featured by an extremely compact mitochondrial (mt) genome with 37 genes on a single circular chromosome. The human body louse, Pediculus humanus, however, has its mt genes on 20 minichromosomes. We sequenced the mt genomes of two other human lice: the head louse, P. capitis, and the pubic louse, Pthirus pubis. Comparison among the three human lice revealed the presence of fragmented mt genomes in their most recent common ancestor, which lived ∼7 Ma. The head louse has exactly the same set of mt minichromosomes as the body louse, indicating that the number of minichromosomes, and the gene content and gene arrangement in each minichromosome have remained unchanged since the body louse evolved from the head louse ∼107,000 years ago. The pubic louse has the same pattern of one protein-coding or rRNA gene per minichromosome (except one minichromosome with two protein-coding genes, atp6 and atp8) as the head louse and the body louse. This pattern is apparently ancestral to all human lice and has been stable for at least 7 Myr. Most tRNA genes of the pubic louse, however, are on different minichromosomes when compared with their counterparts in the head louse and the body louse. It is evident that rearrangement of four tRNA genes (for leucine, arginine and glycine) was due to gene-identity switch by point mutation at the third anticodon position or by homologous recombination, whereas rearrangement of other tRNA genes was by gene translocation between minichromosomes, likely caused by minichromosome split via gene degeneration and deletion.

Introduction

Bilateral animals (Bilateria) are featured by an extremely compact mitochondrial genome that consists of 37 genes and a control region on a single circular chromosome, ∼16 kb in size (Boore 1999; Lavrov 2007). This typical mt genome organization is conserved among most of the bilateral animals known, from worms to insects, fish, and humans (Anderson et al. 1981; Lewis et al. 1995; Cao et al. 1998; Lavrov and Brown 2001). On the other hand, however, deviation from the typical mt genome organization has occurred in many bilateral animals. For instance, most nematodes lack atp8 gene and thus have only 36 mt genes (Lavrov and Brown 2001; but see Gissi et al. 2008 and Breton et al. 2010 for an alternative possibility of “missing” atp8). Duplication of mt genes or the control region occurred in many animals, including snakes (Kumazawa et al. 1998), birds (Eberhard et al. 2001), thrips (Shao and Barker 2003), ticks (Shao et al. 2005a), mites (Shao et al. 2005b), and geckos (Fujita et al. 2007). The mt genomes of the rotifer, Brachionus plicatilis, and the booklouse, Liposcelis bostrychophila, consist of two chromosomes (Suga et al. 2008; Wei et al. 2012). The mt genome of the potato cyst nematode, Globodera pallida, has six chromosomes (Gibson et al. 2007).

The most radical departure to date from the typical mt genome organization of bilateral animals is in the human body louse, Pediculus humanus—an exclusive blood sucker and the vector of three infectious diseases of humans: epidemic typhus, relapsing fever, and trench fever (Raoult and Roux 1999). The mt genome of the body louse consists of 20 types of minichromosomes: each minichromosome is 3–4 kb in size and has 1–3 genes and a control region (Shao et al. 2009). Extensive fragmentation of mt genomes also occurred outside bilateral animals in ichthyosporean protists (Burger et al. 2003), diplonemid protists (Marande et al. 2005), and box jellyfish (Smith et al. 2012). In addition, the chloroplast genomes of dinoflagellates also fragmented extensively (Zhang et al. 1999; Howe et al. 2008). Extensive fragmentation of organelle genomes is a fascinating phenomenon; why organelle genomes became fragmented and how they became fragmented, however, are still poorly understood (Zhang et al. 1999; Landweber 2007; Howe et al. 2008; Rand 2009; Shao et al. 2009).

The parasitic lice of humans present an opportunity for investigating how extensively fragmented mt genomes of bilateral animals evolved. In addition to the body louse, P. humanus, humans are also the hosts of two other blood-sucking lice: the head louse, P. capitis, and the pubic louse, Pthirus pubis (Durden and Musser 1994). Humans share the genus Pediculus with chimpanzees, and share the genus Pthirus with gorillas (Durden and Musser 1994). The head louse often infests school children whereas the pubic louse infests primarily adults and is often transmitted via sexual contact (Burgess 2004; Anderson and Chaney 2009). Of the three human lice, the body louse is the youngest: it evolved when humans adopted frequent use of clothing ∼107,000 years ago (Kittler et al. 2003, 2004) [but see Toups et al. (2011) for an alternative dating]. The head louse is much older than the body louse and has been with humans since the ancestors of humans and chimpanzees split ∼5.4 Ma (Chen and Li 2001; Stauffer et al. 2001). The pubic louse was the result of a host switch of gorilla louse to humans 3–4 Ma (Reed et al. 2007). The pubic louse shares the most recent common ancestor (MRCA) with the head louse and the body louse ∼7 Ma before gorillas split from the lineage that led to humans and chimpanzees (Chen and Li 2001; Stauffer et al. 2001). The three human lice, thus, provide a system for understanding how fragmented mt genomes of bilateral animals changed over a period of ∼7 Myr. We sequenced the mt genomes of the head louse and the pubic louse; comparison among the three human lice provided unprecedented insights into the phenomenon of mt genome fragmentation in bilateral animals.

Materials and Methods

Sample Collection, DNA Extraction, Polymerase Chain Reaction Amplification, Cloning, and Capillary Sequencing

Head lice, P. capitis, were collected in Brisbane, Australia (sample B2135), China (B1453), and Iran (B1572A). Pubic lice, P. pubis, were collected in Ipswich and Brisbane, Australia (B2155 and B2182). Total DNA was extracted from individual louse specimens with DNeasy Tissue Kit (QIAGEN). Because of the high sequence similarity between the head louse and the body louse revealed by previous studies (Kittler et al. 2003; Leo et al. 2005; Reed et al. 2007; Herd et al. 2012), we used, initially, the primers designed from the coding regions of the mt minichromosomes of the body louse (Shao et al. 2009) to amplify by polymerase chain reaction (PCR) each mt minichromosome of the head louse. These primers are next to each other with no gap or a small gap (<10 bp) in between; PCRs with these primers amplified 12 entire or near entire mt minichromosomes of the head louse. These amplicons were sequenced directly (without cloning) with AB3730xl 96-capillary sequencers at the Australian Genome Research Facilities (AGRF) in Brisbane, Australia. A pair of head louse–specific primers, PcF and PcR, was designed from conserved noncoding sequences that flank the coding region of each minichromosome (fig. 1A; supplementary table S1, Supplementary Material online). The PCR with PcF and PcR primers produced a mixture of amplicons ranging from 100 bp to 2 kb in size, expected from the coding regions of the whole set of mt minichromosomes of the head louse (fig. 2A); these amplicons were sequenced with a Roche 454 platform at the AGRF.

F

Alignment of nucleotide sequences in the NCRs of the mitochondrial genomes of the human head louse, Pediculus capitis (A), and the human pubic louse, Pthirus pubis (B).

F

PCR amplicons from the mitochondrial genomes of the human head louse, Pediculus capitis (A) and the human pubic louse, Pthirus pubis (B). (A) Amplicons from PcF and PcR primers that span the entire coding regions of the mt minichromosomes of the head louse. (B) Amplicons from PpF and PpR1 primers that span the entire coding and NCRs of the mt minichromosomes of the pubic louse. (C) Amplicons from PpF and PpR2 primers that span the entire coding regions of the mt minichromosomes of the pubic louse.

For the pubic louse, a 452-bp fragment of mt rrnS gene and a 360-bp fragment of mt rrnL gene were amplified initially by PCR with primer pairs 12SA–12SB and 16SA–16SB (supplementary table S1, Supplementary Material online). These two pairs of primers target sequence motifs that are highly conserved among arthropods. The rrnS and rrnL fragments were sequenced directly; two pairs of pubic louse–specific primers, 12SF–12SR and 16SF–16SR, were designed from rrnS and rrnL genes (supplementary table S1, Supplementary Material online). PCRs with these specific primers amplified a 1.8-kb fragment and a 2.3-kb fragment that spanned nearly the entire trnL–rrnS minichromosome and trnL–rrnL minichromosome, respectively. These fragments were cloned and sequenced; two primers, PpF (forward) and PpR1 (reverse), were designed from conserved noncoding sequences adjacent to the 5′-end of the coding regions of trnL–rrnS minichromosome and trnL–rrnL minichromosome (fig. 1B). The PCR with the primers PpF and PpR1 produced a mixture of amplicons ranging from 1.6 to 2.7 kb (fig. 2B), expected from the whole set of mt minichromosomes of the pubic louse. PCR amplicons with PpF and PpR1 were cloned and sequenced with AB3730xl 96-capillary sequencers. A reverse primer, PpR2, was also designed from conserved noncoding sequences downstream of the coding regions of minichromosomes (fig. 1B). The PCR with primers PpF and PpR2 generated a mixture of amplicons ranging from 1.2 to 1.9 kb in size (fig. 2C), expected from the coding regions of the whole set of mt minichromosomes of the pubic louse; these amplicons were sequenced with a Roche 454 platform at the AGRF (discussed later).

TaKaRa LA Taq kit was used in all PCR amplifications. Each PCR (25 µL) contained 0.25 µL of LA Taq, 2.5 µL of 10× Buffer, 2.5 µL of MgCl2 (25 mM), 2.0–4.0 µL of dNTP mixture (2.5 mM each), 1.0 µL of forward primer (10 µM), 1.0 µL of reverse primer (10 µM), 1.0 µL of DNA template, and 12.75–14.75 µL of Milli-Q water. PCR cycling conditions were: 94°C for 1 min, 30–40 cycles of 98°C for 10 s, 45–65°C (depending on primers) for 30 s, and 68°C for 1–4 min (depending on target size, ∼1 min/kb), followed by 72°C for 2–6 min. Positive and negative controls were always executed with each PCR experiment to help to detect DNA contamination and false positive amplicons. PCR amplicons were checked by agarose gel (1%) electrophoresis. The sizes of PCR amplicons were estimated by comparison with molecular markers. PCR amplicons used for sequencing were purified with Wizard SV Gel/PCR Clean-up System (Promega); cloning was with pGEM-T Easy Vector System (Promega). Sequence-reads were assembled into contigs with Sequencher 5.0 (Gene Codes); the parameters for assembly were minimum match 90% and minimum overlap 100 bp. tRNA genes were identified with tRNA-Scan (Lowe and Eddy 1997) and ARWEN (Laslett and Canback 2008). Protein-coding genes and rRNA genes were identified with BLAST searches of GenBank (Altschul et al. 1990). Identical sequences shared between genes were identified by Wordmatch (Rice et al. 2000). Sequence alignments were with ClustalX (Larkin et al. 2007).

Roche 454 Sequencing and Verification of Individual Mitochondrial Minichromosomes

The contigs generated from capillary sequencing had low coverage (1 to 3×). To ensure the reliability of sequence analyses, we used a Roche GS-FLX (454) platform to sequence, in high coverage (4 to 2,700×), the amplicons obtained from the mt minichromosomes of the head louse and the pubic louse. Roche 454 sequence reads were assembled and analysed as described earlier. We also verified by PCR the size and the circular organization of each mt minichromosome of the head louse and the pubic louse identified by sequence-read assembly. A pair of outbound primers (forward and reverse) was designed from the coding region of each minichromosome; these two primers were next to each other with a small gap or no gap in between. PCRs with these primers amplify the full or near full length of each minichromosome if it has a circular organization. PCR set-up, cycling conditions, agarose gel electrophoresis and size measurement were the same as described earlier. The nucleotide sequences of the mitochondrial genomes of the human head louse and the pubic louse have been deposited in GenBank under accession numbers EU219988-95, HM241895-8, and JX080388-407.

Results

Mitochondrial Genome of the Human Head Louse, P. capitis

We obtained 9,412 sequence reads from the mt genome of the head louse by capillary sequencing and Roche 454 sequencing (table 1). These sequence-reads range from 100 to 700 bp in length. We assembled these sequence reads into contigs and identified all of the 37 mt genes typical of bilateral animals. The 37 mt genes of the head louse are on 20 minichromosomes; the distribution of these genes among the minichromosomes in the head louse is identical to that in the body louse (Shao et al. 2009). Each minichromosome is 1.5–3.0 kb in size and has a circular organization (fig. 3A). Each minichromosome has a coding region and a noncoding region (NCR). The coding region of each minichromosome contains one to three genes, and varies in size from 65 bp for trnM minichromosome to 1,686 bp for nad5 minichromosome (Note: minichromosomes are named after their genes hereafter). With the intriguing exception of nad1 and trnQ, all mt genes of the head louse have the same orientation of transcription relative to the NCRs (fig. 3A).

Table 1

Mitochondrial Minichromosomes of the Head Louse, Pediculus capitis, Identified by Capillary Sequencing and Roche 454 Sequencing

Minichromosome	Size of Coding Region (bp)	Number of Capillary Sequence Reads	Number of Roche 454 Sequence Reads
atp8-atp6	846	9	222
cox1	1,572	16	99
Y-cox2	757	11	222
cox3-A	875	10	182
cob	1,074	13	125
nad1-Q	967	11	70
P-nad2-I	1,129	18	89
R-nad3	428	7	683
K-nad4	1,419	10	49
G-nad4L-V	409	0	556
nad5	1,686	11	15
F-nad6	580	0	499
L1(tag)/L2(taa)-rrnS-C	933	17	411
L1(tag)/L2(taa)-rrnL	1,220	14	188
S1(tct)-N-E	226	0	1,259
T-D-H	227	0	869
M	65	0	992
W-S2(tga)	144	0	2,735
Total	14,557	147	9,265

F

The mitochondrial (mt) genomes of the human head louse, Pediculus capitis (A) and the human pubic louse, Pthirus pubis (B). Each minichromosome has a coding region (with gene name, transcription orientation, and length indicated) and a NCR (in black). The mini-chromosomes are in alphabetical order according to the names of their protein coding and rRNA genes, followed by those with only tRNA genes. Minichromosomes with asterisk symbols (*) are present in all three of the human lice. The gray shading in the trnC minichromosome is the 99 bp unique section that has 90% similarity in sequence to trnL2 and rrnL genes (fig. 4A); the black dot shading is the 751 bp unique section that has 50% similarity in sequence to rrnL gene (fig. 4B). The white dot shading in the trnL–rrnL minichromosome is the 67 bp unique section that has 59% similarity in sequence to trnC gene (fig. 4C). Protein-coding genes are abbreviated as atp6 and atp8 (for ATP synthase subunits 6 and 8), cox1–3 (for cytochrome c oxidase subunits 1–3), cob (for cytochrome b), and nad1-6 and 4L (for NADH dehydrogenase subunits 1–6 and 4L). rrnL and rrnS are for large and small rRNA subunits. tRNA genes are shown with the single-letter abbreviations of their corresponding amino acids. Courtesy: of Vincent Smith (for the head louse image) and the CDC, USA (for the pubic louse image).

F

Alignment of the nucleotide sequences between the two unique sections in the NCR of trnC minichromosome (gray shading and black dot shading in fig. 3B) and the coding region of trnL–rrnL minichromosome (A and B), and between the unique section in the NCR of trnL–rrnL minichromosome (white dot shading in fig. 3B) and the coding region of trnC minichromosome of the pubic louse, Pthirus pubis (C).

We sequenced the full length NCRs of two mt minichromosomes, cox1 and trnK–nad4, of the head louse (fig. 1A). The six copies of the NCR of the cox1 minichromosome of the head louse are ∼1.4 kb in size, that is, ∼0.5 kb shorter than their counterparts in the body louse (Shao et al. 2009). The NCR of the trnK–nad4 minichromosome of the head louse is 1,905 bp, a similar length to those of the body louse (Shao et al. 2009). The NCRs of the head lice have 72–97% pairwise identity to each other. An AT-rich motif (203 bp, 69% A and T) is present in the NCRs of the head louse before the 5′-end of the coding region, whereas a GC-rich motif (63 bp, 60% C and G) is present after the 3′-end of the coding region (fig. 1A). Between the GC-rich motif and the AT-rich motif, there are numerous conserved sequence motifs ranging from 1 to 103 bp in size, and indels ranging from 1 to 116 bp in size.

Mitochondrial Genome of the Human Pubic Louse, P. pubis

The capillary sequencing and Roche 454 sequencing generated 2,663 sequence reads in total, ranging from 100 to 700 bp in length, from the mt genome of the pubic louse (table 2). We assembled these sequence reads into contigs and identified 34 of the 37 mt genes that are typical of bilateral animals; these 34 genes are on 14 minichromosomes (table 2; fig. 3B). Each minichromosome is circular, 1.8–2.7 kb in size, and consists of a coding region and a NCR. The coding regions contain one to five genes each, and vary in size from 64 bp for trnC minichromosome to 1,647 bp for nad5 minichromosome (table 2; fig. 3B). Each of the 34 mt genes we identified in the pubic louse is present on only one minichromosome; there is no overlap in gene content between different minichromosomes. All of the 34 mt genes, including nad1 and trnQ, have the same orientation of transcription relative to the NCRs (fig. 3B). Intriguingly, we did not find nad4, trnK, and trnN in the pubic louse by either capillary sequencing or Roche 454 sequencing. These three genes are present in the fragmented mt genomes of the head louse (discussed earlier), the body louse (Shao et al. 2009), and the chimpanzee louse, P. schaeffi (Herd K, et al., unpublished data). There could be two possibilities. The minichromosomes that contained these three genes were less abundant than the 14 minichromosomes we identified in the pubic louse, and thus, were missed in our sequencing at the current coverage (table 2). Alternatively, the primers we used to amplify the mt genome of the pubic louse might not be conserved in the minichromosomes that contained nad4, trnK, and trnN genes. The second possibility is more likely, in our view, as the expression of mt genes needs to be coordinated and regulated; a substantially decreased frequency of an mt minichromosome would present serious metabolic challenges to cells (Lane 2005).

Table 2

Mitochondrial Minichromosomes of the Pubic Louse, Pthirus pubis, Identified by Capillary Sequencing and Roche 454 Sequencing

Minichromosome	Size of Coding Region (bp)	Number of Clones Sequenced	Number of Capillary Sequence Reads	Number of Roche 454 Sequence Reads
atp8-atp6	821	6	12	138
cox1	1,548	3	37	185
Y-cox2	766	7	21	37
cox3-A	885	8	22	71
cob-S1	1,119	14	37	431
nad1-Q	958	6	21	522
P-nad2-I	1,126	2	4	41
G-nad3-V-W-S2	795	9	33	24
T-D-H-R-nad4L	597	4	14	45
nad5	1,647	2	8	17
F-nad6-E-M	761	2	6	0
L1(tag)-rrnS	830	9	28	28
L2(taa)-rrnL	1,287	8	25	665
C	64	26	62	129
nad4, K, N	NA	0	0	0
Total	13,204	106	330	2,333

Note.—NA, not applicable.

We sequenced the full length NCRs of 10 mt minichromosomes of the pubic louse (fig. 1B). The NCRs of these minichromosomes, with the exception of trnC minichromosome and trnL–rrnL minichromosome, have similar length (978–1,044 bp) and high sequence similarities to each other (91–99%). A perfectly conserved AT-rich motif (111 bp, 76% A and T) is upstream of the 5′-end of the coding region of each minichromosome; a GC-rich motif (77 bp, 58% G and C) is downstream of the 3′-end of the coding region. In addition, there are numerous highly conserved motifs, ranging from 1 to 87 bp, in the NCRs between the GC-rich motif and the AT-rich motif (fig. 1B).

The NCRs of the trnC minichromosome and the trnL–rrnL minichromosome of the pubic louse contain unique sections that are not present in the NCRs of other minichromosomes (fig. 3B). One of the unique sections in the NCR of trnC minichromosome is 99 bp long and has 90% similarity in sequence to trnL and rrnL genes (fig. 4A); the other unique section is 751 bp long and has 50% similarity in sequence to rrnL gene (fig. 4B). On the other hand, the unique section in the trnL–rrnL minichromosome is 67 bp long and has 59% similarity in sequence to trnC gene (fig. 4C). The unique sections in the trnC minichromosome are apparently degenerate trnL and rrnL genes and the unique section in the trnL–rrnL minichromosome a degenerate trnC gene.

Discussion

Recombination between Mitochondrial Genes on Different Minichromosomes

Whether and how animal mt DNA recombine has been a controversial issue in the past decade (Rokas et al. 2003; Kraytsberg et al. 2004; Ladoukakis et al. 2011). An extraordinary feature of the fragmented mt genome of the human body louse is the stretches of identical sequences (26–127 bp long) shared between genes on different minichromosomes; this provides unequivocal evidence for inter-minichromosome recombination (Shao et al. 2009; Shao and Barker 2011). As in the body louse, inter-minichromosome recombination also occurs in the head louse and the pubic louse. Six pairs of mt genes in the head louse share identical sequences, 26–127 bp long, which are much longer than expected by chance (table 3). Two pairs of tRNA genes in the pubic louse, trnL and trnL, and trnR and trnG, share identical sequences that are three to five times longer than expected by chance (table 3). Two other pairs of mt genes of the pubic louse, cox1 and nad4L, and trnI and trnT, share identical sequences two to three times longer than expected by chance. It is noteworthy that in the pubic louse, only the genes of the same type, that is, two tRNA genes or two protein-coding genes, share longer than expected identical sequences. In the body louse and the head louse, however, the large subunit rRNA gene, rrnL, shares longer than expected identical sequences, 26 and 99 bp, respectively, with two protein-coding genes, nad2 and nad5 (table 3). The fact that all of the three human lice have longer than expected stretches of identical sequences shared between different mt genes, whereas no other bilateral animals known appear to have this feature indicates that mt genome fragmentation facilitates recombination between mt genes. Indeed, similar evidence for recombination between minichromosomes was also found recently outside bilateral animals in a box jellyfish that has an extensively fragmented mt genome with eight linear minichromosomes (Smith et al. 2012).

Table 3

The Longest Stretches of Identical Sequence Shared by Mitochondrial Genes of the Three Human Lice, Which Have Fragmented Mitochondrial Genomes, and Six Other Species of Bilateral Animals that Have the Typical Mitochondrial Genomes

Pairs of gene		The Longest Stretches of Identical Sequence (bp)
		Pc	Pp	Ph	Bm	Cb	Hm	Dy	Ce	Hs
nad4	nad5	127, 30	NA	127, 30	13	15	15	16	14	11
rrnL	nad5	99	10	99	12	14	13	15	16	10
trnL1	trnL2	33, 32	35, 32	33, 32	7	6	7	10	6	6
trnR	trnG	28, 14	32, 26	28, 14	5	6	7	6	8	6
atp8	Cob	26	9	26	10	11	11	12	NA	9
rrnL	nad2	26	10	26	13	11	14	13	12	10
cox1	nad4L	10	29	10	13	11	14	13	12	10
trnI	trnT	6	16	6	6	5	7	7	9	6

Note.—The longest stretches of shared identical sequences are indicated in bold. Pc, Pediculus capititis (human head louse); Pp, Pthirus pubis (human pubic louse); Ph, Pediculus humanus (human body louse); Bm, Bothriometopus macrocnemis (screamer louse); Cb, Campanulotes bidentatus (pigeon louse); Hm, Heterodoxus macropus (wallaby louse); Dy, Drosophila yakuba (fruitfly); Ce, Caenorhabditis elegans (roundworm); Hs, Homo sapiens (human); NA, not applicable.

Unusual Secondary Structures of Mitochondrial tRNAs

Most tRNAs of the human lice have the typical clover leaf–shaped secondary structure; four tRNAs, however, have atypical structures (fig. 5). The anticodon loop (AC-loop) of tRNAs almost always has seven nucleotides; the tRNA–alanine of the pubic louse, however, has nine nucleotides in this loop. tRNA–glutamine lacks the D-arm in all three of the human lice—a feature that is not seen in other species of parasitic lice known (Shao et al. 2001; Covacin et al. 2006; Cameron et al. 2007). As in most bilateral animals, tRNA–serine (UCU) also lacks the D-arm in the three human lice. The genes for tRNA–glutamine and tRNA–serine (UCU) were annotated incorrectly for the body louse due to the atypical structures of these two tRNAs (Shao et al. 2009). Comparison among the head louse, the pubic louse, and the body louse allowed us to correct these misannotations (figs. 3A and 5).

F

Inferred secondary structures of the mitochondrial tRNAs of the human lice: the body louse, Pediculus humanus (Ph), the head louse, Pediculus capitis (Pc), and the pubic louse, Pthirus pubis (Pp). Shared identical sequences between tRNA genes are in bold (see text in Discussion and table 3).

For tRNAs with the typical structure, the variable loop (V-loop, fig. 5) usually has four or five nucleotides; tRNA–isoleucine of the pubic louse, however, has seven nucleotides in this loop; the extra nucleotides in this loop make this tRNA gene undetectable by the two commonly used tRNA search programs, ARWEN (Laslett and Canback 2008) and tRNAscan–SE (Lowe and Eddy 1997). It is noteworthy that the V-loop of tRNA–isoleucine is part of the 16 nucleotide identical sequence shared between trnI and trnT genes (see above, fig. 5 and table 3), indicating that recombination between tRNA genes may affect the secondary structure of their corresponding tRNAs. This affect is much evident in the secondary structures of four other tRNAs: 1) tRNA–leucine (UAG) and tRNA–leucine (UAA) of the three human lice have identical secondary structures except for a nucleotide variation at the third anticodon position; and 2) tRNA–arginine and tRNA–glycine of the pubic louse also have identical secondary structures except for the nucleotide at the third anticodon position and three other nucleotides of the anticodon arm (AC-arm; fig. 5).

Fragmentation of Mitochondrial Genomes in the Human Lice

The MRCA of the three human lice lived ∼7 Ma on the MRCA of humans, chimpanzees, and gorillas (Chen and Li 2001; Stauffer et al. 2001; Reed et al. 2007). All three of the human lice have fragmented mt genomes and, furthermore, a common pattern for the distribution of protein-coding and rRNA genes among mt minichromosomes, in which atp6 and atp8 are adjacent on one minichromosome, whereas each of the other protein-coding and rRNA genes has its own minichromosome (fig. 3). This pattern of gene distribution and the fragmented mt genomes, thus, have apparently evolved in the MRCA of the three human lice, and have been retained for at least ∼7 Myr.

Eight of the 14 minichromosomes of the pubic louse have identical gene content and gene arrangement to their counterparts in the head louse and the body louse (fig. 3). It is noteworthy that the transcription orientation of nad1–trnQ relative to the NCR differs between the pubic louse on the one hand, and the body louse and the head louse on the other hand. Thus, nad1–trnQ has inverted in either the lineage that led to the pubic louse or the lineage that led to the body louse and the head louse. Six of the 14 minichromosomes of the pubic louse are not present in the body louse and the head louse (fig. 3). The arrangement of tRNA genes in these six minichromosomes distinguishes the pubic louse from the body louse and the head louse. With the only exception of trnC, all other tRNA genes of the pubic louse are either upstream or downstream of a protein-coding or rRNA gene. In the two extreme cases, four tRNA genes in a cluster, trnT–trnD–trnH–trnR, are upstream of nad4L gene, and three other tRNA genes in another cluster, trnV–trnW–trnS, are downstream of nad3 gene. In the body louse and the head louse, however, the number of tRNA gene upstream or downstream of a protein-coding or rRNA gene does not exceed one; furthermore, nine tRNA genes on four minichromosomes have no connection at all with any protein-coding or rRNA genes (fig. 3A). Clearly, the pace of mt genome fragmentation has been faster in the lineage leading to the body louse and the head louse than in the lineage leading to the pubic louse. In the typical single-chromosome mt genome of bilateral animals, the secondary structure of tRNA sequences serves as punctuation marks in the processing of the polycistronic transcripts into individual mRNAs and rRNAs (Ojala et al. 1981). For the fragmented mt genomes of the human lice, the role of tRNA structures as punctuation marks is likely less important, or not needed, as only the transcript from the atp8–atp6 minichromosome encodes two proteins; all other transcripts encode either only one protein or rRNA, or only tRNAs.

The NCR in the mt genome of bilateral animals contains initiation sites for genome replication and gene transcription and are commonly called the control region (Boore 1999). For instance, in the human mt genome, two of the three transcription-initiation sites and one of the replication-initiation sites are in the NCR (Taanman 1999). In the mt genome of Drosophila melanogaster, two of the five transcription-initiation sites are in the NCR (Berthier et al. 1986). The NCRs of the three human lice vary substantially in both size and nucleotide sequence. Two common features, however, are present in the NCRs of all mt minichromosomes in all three of the human lice: 1) an AT-rich motif (>59% A and T) upstream of each coding region and a CG-rich motif (>60% C and G) downstream of each coding region; and 2) a perfectly conserved motif, ACCAAATAGCTA, between the AT-rich and the CG-rich motifs (fig. 1). The roles of these sequence motifs in the initiation and termination of mt gene transcription and mt minichromosome replication remain to be determined.

Molecular Mechanisms for tRNA Gene Rearrangement between Minichromosomes

Rearrangements of tRNA genes can be powerful phylogenetic markers (Boore et al. 1998). Recognition of tRNA orthologs, however, is not always straightforward. Inferring orthologs solely on anti-codon sequence and deduced secondary structure can be misleading (Rawlings et al. 2003). In the pubic louse, trnR is before nad4L, and trnG is before nad3. In the body louse and the head louse, however, these two genes have different positions: trnR gene is before nad3, and trnG gene is before nad4L (fig. 3). The arrangements, trnR–nad4L and trnG–nad3, in the pubic louse can be inferred to be ancestral to sucking lice (suborder Anoplura, which includes the three human lice) because these arrangements are also present in the screamer louse, Bothriometopus macrocnemis (suborder Ischnocera, sister to the Anoplura + Rhyncopthirina) (Cameron et al. 2007) (fig. 6). Therefore, we infer that trnR and trnG genes had rearranged in the body louse and the head louse. Because of the high sequence similarity between trnR and trnG (fig. 5), the rearrangements of these two tRNA genes in the body louse and the head louse occurred more likely through gene-identity switch, either by point mutation at the third anticodon position or by homologous recombination, than by gene translocation between minichromosomes. In the same way, gene-identity switch between trnL and trnL explains well the variation in the arrangement of these two tRNA genes between the pubic louse on the one hand, and the body louse and the head louse on the other hand. In the pubic louse, trnL is upstream of rrnS gene only, and trnL is upstream of rrnL gene only (fig. 3B). In the body louse and the head louse, however, trnL is upstream of rrnS gene in one minichromosome and upstream of rrnL gene in another minichromosome; so is trnL gene (fig. 3A). Because of the near identical sequence of trnL and trnL genes in all of the three human lice (fig. 5), their identities can swap relatively easily by either point mutation at the third anticodon position or by homologous recombination. Variation in the arrangement of other 10 tRNA genes, however, between the pubic louse on the one hand, and the body louse and the head louse on the other hand, is very likely due to gene translocation from one minichromosome to another; there is no evidence that gene identity-switch could cause variation in the arrangement of these tRNA genes in the human lice. The degenerate gene sequences present in the trnC minichromosome and the trnL–rrnL minichromosome of the pubic louse indicates that gene degeneration followed by deletion is likely responsible for the split of one minichromosome into two, and thus for the translocation of tRNA genes between minichromosomes (fig. 7).

F

Comparison of mitochondrial gene arrangement among the three human lice (suborder Anoplura) and the screamer louse, Bothriometopus macrocnemis (suborder Ischnocera). Genes on the same chromosome are linked by a hyphen (-); different minichromosomes are separated by a comma (,). Genes underlined have an opposite orientation of transcription relative to the NCR, compared with genes on other minichromosomes.

F

A gene-degeneration-followed-by-deletion model for the split of one mitochondrial minichromosome into two, and thus the translocation of tRNA genes between minichromosomes, based on the degenerate gene sequences present in the trnC and trnL–rrnL minichromosomes of the human pubic louse, Pthirus pubis.

Conclusion

We sequenced the mt genomes of the human head louse and the human pubic louse. Comparison among the three human lice revealed that fragmented mt genomes were present in the MRCA of these lice ∼7 Ma. The pace of mt genome fragmentation has been unequal; it is faster in the lineage leading to the body louse and the head louse than in the lineage leading to the pubic louse, evidenced by the fact that there are more minichromosomes and thus fewer genes per minichromosome in the body louse and the head louse than in the pubic louse. Despite substantial variation in the sequence and the size of the NCR, the number of mt minichromosomes, and the gene content and gene arrangement in each minichromosome has been perfectly conserved between the head louse and the body louse for ∼107,000 years. The pubic louse has the same pattern as the body louse and the head louse for the distribution of protein-coding genes and rRNA genes among the mt minichromosomes; this pattern, thus, has been stable for at least 7 Myr. The majority of the tRNA genes of the pubic louse are on different minichromosomes when compared with their counterparts in the body louse and the head louse. It is evident that rearrangement of four tRNA genes (for leucine, arginine, and glycine) is due to gene-identity switch caused by either point mutation at the third anticodon position or homologous recombination. Rearrangement of 10 other tRNA genes was by gene translocation between minichromosomes, very likely caused by minichromosome split via gene degeneration and then deletion.

Supplementary Material

Supplementary table S1 is available at Genome Biology and Evolution online (http://www.gbe.oxfordjournals.org/).