| Edited by Kyoichi Sawamura. Tadashi Yokogawa: Corresponding author. E-mail: t-yokogawa@kyudai.jp |
Scale insects (Hemiptera: Coccoidea) are unique not only in their morphology but also in their chromosome systems. In most species of the group called Neococcoidea, which comprises most of the species of scale insects, paternal chromosomes are eliminated from the male germline and only maternal chromosomes are transmitted by sperm. This extraordinary chromosome system is called PGL (Paternal Genome Loss) or PGE (Paternal Genome Elimination) (Bull, 1979, 1983; Normark, 2003). Three types of PGL have been known in the Neococcoidea (Nur, 1980). In the lecanoid system (L system), paternal chromosomes in males are inactive but nevertheless maintained in diploid somatic tissues and only eliminated later in spermatogenesis. The Comstockiella system (C system) is similar to the L system, but one or more of paternal chromosomes are eliminated just prior to meiosis at spermatogenesis and then all others are eliminated later in spermatogenesis. In the diaspidid system (D system), the paternal genome is eliminated in the early cleavage of eggs and thus developed males are haploids. In all three systems, males are functionally haploids. Thus, PGL is similar to haplodiploidy sensu stricto (arrhenotoky, we refer as haplodiploidy), except in that the paternal genome is transmitted once to the male (Sabelis et al., 2002; Normark, 2003, 2004, 2006; Engelstädter and Hurst, 2006).
PGL is found widely in scale insects. Among 227 species of 19 families examined, 173 species of 14 families exhibited PGL. However, it remains uncertain whether PGL of scale insects evolved only once or repeatedly in some lineages. Nur (1980) proposed the single-origin hypothesis, while Brown (1977) proposed the independent origin hypothesis. A single origin of PGL was inferred from preliminary phylogenetic trees (Cook et al., 2002; Gullan and Cook, 2007) using 495 unambiguously aligned nucleotide sites from nuclear ribosomal DNA sequences that gave only 56% bootstrap supports and 72% Bayesian posterior probability for the hypothesis, respectively.
While haplodiploidy is found in more than 19 lineages of animals, PGL has been known only in seven groups of arthropods. Thus, it has been put forward that PGL may be a transit stage toward haplodiploidy (Schrader and Hughes-Schrader, 1931; Cruickshank and Thomas, 1999; Sabelis et al., 2002). Scale insects provide us an excellent opportunity to examine the evolutionary relationship between haplodiploidy and PGL as some scale insects show haplodiploidy. Cook et al. (2002) and Gullan and Cook (2007) reconstructed phylogeny of scale insects including the genus Icerya of haplodiploidy. However the placement of Icerya remains uncertain due to the low resolution of the phylogenetic tree.
The phylogenetic relationship between the L, C and D systems is also an open question. Brown (1977) hypothesized that the L and C systems derived independently from XX-XO, and the D system arose from the C system repeatedly. On the other hand, Nur (1980) claimed that PGL evolved in the order of the L, C and D, all three systems being derived repeatedly. He also suggested that the L system arose once from XX-XO and also arose reversibly from the C system. The phylogeny of Cook et al. (2002) and Gullan and Cook (2007) again showed low resolution for the evolutionary relationship among the L, C and D systems.
Morphological data is not useful to resolve phylogeny between scale insect families because scale insects have very reduced and highly modified morphology (Foldi, 1997; Gullan and Kosztarab, 1997; Kosztarab, 1996). Cook et al. (2002) were the first to overcome this difficulty by using nuclear DNA data, but their effort and its subsequent extention (Gullan and Cook, 2007) still remain at a preliminary stage. Here we present a mitochondrial DNA phylogeny of scale insects, including 9 families of XX-XO, all three PGL systems and haplodiploidy. By reconstructing a well-resolved and strongly supported phylogeny and mapping chromosome systems onto the tree, we will show that PGL arose once in a common ancestor of the Neococcoidea and haplodiploidy was derived not from PGL but from XX-XO. We also put forward plausible reconstructions for the evolutionary history of the L, C and D systems.
All samples of scale insects and aphids were collected in Japan. We used the classification of Ben-Dov et al. (2006). Insects were fixed and stored in 99.5% ethanol until extraction of total DNA. Three aphid species were used as outgroup taxa. Total DNA was extracted from individual insects for each species using DNeasy Tissue Kit (QIAGEN). To reconstruct phylogeny, we sequenced a region from the mitochondrial cytochrome c oxidase subunit I (COI) gene to cytochrome oxidase c subunit II (COII) (corresponding to positions 2,172 to 3,684 in Drosophila melanogaster mtDNA). In scale insects, to our knowledge, it remains unknown whether mitochondrial DNA is transmitted maternally.
The COI-COII region was amplified by the primers mtD-10 (5’-TTGATTTTTTGGTCATCCAGAAGT-3’) and mtD-18 (5’-CCACAAATTTCTGAACATTGACCA-3’) from the total DNA. Both primers were obtained from Nucleic Acid-Protein Service Unit Biotechnology Laboratory (University of British Columbia). PCR was carried out in 20 μl reaction mixture containing 1x LA PCR buffer II (Mg2+ free, TaKaRa), 2.5 mM MgCl2, each 0.4 mM of dATP and dTTP, each 0.1 mM of dGTP and dCTP, 0.5 U of Ex Taq polymerase (TaKaRa), 0.25 μM of each primer and 1 μl (0.2–2 ng) of total DNA. The following conditions were used: initial denaturation 94°C for 2 min, followed by 30 cycles of 94°C for 20 s, 48°C for 20 s, 65°C for 3 min, and 65°C for 10 min.
The amplified DNA was cloned into plasmid vectors using the TOPO TA Cloning Kit (Invitorogen). Three or more clones having the COI-COII region were isolated using alkali-SDS method and PEG precipitation method, or the automatic nucleic acid isolation system (NA-2000; Kurabo Industories).
We used the shotgun sequencing approach because the region has a paucity of good oligonucleotide priming sites. Shotgun cloning was carried out each of three or more clones having the COI-COII region. For the construction of shotgun libraries, PCR was carried out with the same mixture and condition as above, using isolated plasmid DNA as a template. Then, the PCR product was digested with a nuclease (Afa I, Dra I, EcoR I, Ssp I or DNase I). Each nuclease was added directly to the PCR product, thus nuclease reaction was carried out in 1x PCR buffer containing amplified DNA. The digest was checked on 1% agarose gel to ensure the reaction. The ends of digested DNA are often blunt-ended or frayed. DNase I-digested ends may be especially highly frayed and heterogeneous, due to this nuclease randomly producing nicks in double-stranded DNA in the presence of Mg2+. For TA cloning, we repaired the frayed ends of DNA fragments and added a single A-overhang to each end of the fragment. That is, we added 0.1 U/μl of Ex Taq polymerase to the digested products and carried out PCR amplification by the following conditions: initial denaturation 94°C for 2 min, followed by 2 cycles of 65°C for 10 min and 72°C for 10 min. Finally, shotgun fragments with an adenylated 3’ ends were cloned by ligation to pGEM-T Easy Vector (Promega). Plasmid DNA was isolated as above.
Sequencing of plasmid inserts was conducted using the ABI Prism BigDye Terminator Cycle Sequencing Kit (Perkin Elmer) and ABI 3100 DNA sequencer (Perkin Elmer). SP6 and T7 primers were used to determine the nucleotides.
Sequences were edited and assembled using DNASIS V3.0 (Hitachi Software Engineering Co., Ltd.). Determined sequences were deposited in the DDBJ/EMBL/GenBank nucleotide sequence databases. Accession numbers are listed in Table 1. The region coding tRNA (leu) was removed because it was absent between COI and COII in all analyzed species of Diaspididae. The resulting sequences (parts of COI and COII genes) were aligned manually using Bioedit (Hall, 1999). We further refined the alignment of some amino acid indels in COI and COII genes using protein translation, so that 1,229 bases could be used for phylogenetic analysis. The aligned sequences are deposited in TreeBase under the accession numbers SN4283.
 View Details | Table 1 Coccoidea used in this study |
For phylogenetic analysis, nucleotide substitution models were selected based on the Akaike information criterion using the program Kakusan3.0 (Tanabe, 2007). We allowed it to select the best model in each codon position of each gene. As a result, the following models were selected for the COI and COII genes: GTR + I + G for the first codon positions of COI and COII; GTR + G for the second codon positions of COI and COII; SYM + G for the third codon position of COI, and SYM + G + I for the third codon position of COII. Alignment gaps were treated as missing data.
Both the maximum likelihood method and the Baysian method were employed to reconstruct the phylogenetic tree. Maximum likelihood analyses were conducted with 1,000 bootstrap replicates by using TREEFINDER v. Oct 2008 (Jobb et al., 2004). Bayesian analysis was performed with MrBayes v.3.1.2 (Ronquist and Huelsenbeck, 2003). All analyses were run with two independent runs at 10,000,000 generations, 4 chains (temperature = 0.2), saving trees every 100 generations. Uniform priors between 0 and 1 were used for a gamma shape parameter and the proportion of invariable sites. Consensus trees and posterior probability values were obtained after removing the first 50,000 generations. In order to confirm that the chains had achieved stationarity, we evaluated “burn-in” plots by plotting log-likelihood scores and tree lengths against generation number using the software Tracer v. 1.4 (Rambaut and Drummond, 2007). Bayesian analyses were also performed for the next best models of nucleotide substitutions to examine robustness of the phylogenetic relationship obtained.
Chromosome systems of analyzed species are shown in Table 1. To find the most parsimonious optimization of the evolution of the chromosome systems, the computer program MacClade 3.04 (Maddison and Maddison, 1992) was used. We inferred the evolutionary processes by mapping the chromosome systems onto the scale insect tree. As Kermesidae is not clear in either the L or C system (Nur, 1980), we analyzed each case.
Bayesian and maximum likelihood analyses resulted in highly resolved and strongly supported phylogenies with nearly identical topologies. The consensus Bayesian tree is shown in Fig. 1. The phylogenetic relationship holded under the next best models of nucleotide substitutions. The only difference between the consensus trees by Bayesian and maximum likelihood analyses was the placement of Pulvinaria aurantii, Parthenolecanium corni, Pul. torreyae and Eulecanium kunoense in Coccidae. While the latter three clustered to a group in Fig. 1 (its monophyly is only weakly supported), the four species were polychotomous in the maximum likelihood tree. In the congruent topology, Ortheziidae (a family of the group called the Archaeococcoidea), having the chromosome system of XX-XO or 2N-2N, were sister to all the other scale insect taxa examined. Margarodidae of the Archaeococcoidea, having the chromosome system of XX-XO or haplodiploidy, were sister to the Neococcoidea. This relationship was supported by a Bayesian posterior probability (p) of 0.99 and a maximum likelihood bootstrap probability (BP) of 79.9%. The monophyly of the Neococcoidea was strongly supported (p = 1.00, BP = 99.4%). Five families of the Neococcoidea, in which at least two species were examined, were monophyletic (all, p = 1.00), although in Asterolecaniidae a maximum likelihood BP was not high (64.7%). Two families with the L system (Pseudococcidae and Cerococcidae) were not sister to each other. Also, Eriococcidae, Asterolecaniidae and Coccidae, having the C system, were not a monophyletic group. As for Diaspididae, in which the C and D systems have been reported, we could only examine species with the D system. Those species clustered together and were likely to be sister to Eriococcidae with the C system (p = 0.84, BP = 60%).
 View Details | Fig. 1 Bayesian consensus phylogenetic tree of mitochondrial COI and COII DNA sequences combined. Node support values are Bayesian posterior probabilities on the top and maximum likelihood bootstrap values on the bottom. Scale bar indicates the number of substitutions per site. |
The resulted topology was used to reconstruct the evolutionary sequences of the L, C, and D systems. Two cases were considered; Kermesidae have the L (Fig. 2(a)) or the C system (Fig. 2(b)). In both cases, the haplodiploidy of Icerya (Margarodidae) was derived from the XX-XO system, and PGL was derived once from the XX-XO chromosome system. The ancestral PGL system was L or C. If it was L, the ancestral L system evolved to the C system and then in the lineage of Cerococcidae the C system reverted to the L system. If it was C, the ancestral C system evolved to the L system twice. The D system was derived from the C system.
 View Details | Fig. 2 Most parsimonious reconstructions of the chromosome systems. Reconstructions of chromosome systems were represented whether Kermesidae is the L (a) or C (b) system, respectively. |
Scale insects have been divided to two groups; the Archaeococcoidea and the Neococcoidea. However, most of the morphological traits that characterize the Archaeococcoidea (abdominal spiracles, compound eyes in males for example) are found more widely in the Hemiptera and thus the Archaeococcoidea may be paraphyletic (Foldi, 1997). On the other hand, the Neococcoidea are characterized by such apomorphies as the absence of abdominal spiracles and PGL and are likely to be monophyletic. Cook et al. (2002) was the first to test such phylogenetic hypotheses of scale insects using DNA sequences. They analysed nuclear ribosomal DNA sequences of 495 unambiguously aligned nucleotide sites and supported the monophyly of the Neococcoidea. However, they gave only a 56% bootstrap support for the hypothesis and they themselves admitted their result as ‘preliminary’. Recently, Gullan and Cook (2007) determined sequences of the same DNA region for 42 additional taxa, mainly of the Archaeococcoidea, and gave a 72% Bayesian posterior probability for the hypothesis. This support still remains inconclusive.
Our results, based on 1,229 unambiguously aligned nucleotide sites of the mitochondrial DNA sequences of the cytochrome c oxidase subunit I (COI) gene and the cytochrome oxidase c subunit II (COII) gene, showed a highly improved resolution in the phylogenetic relationship of scale insects. The paraphyly of the Archaeococcoidea and the monophyly of the Neococcoidea are strongly supported by Bayesian posterior probabilities p = 0.99 and 1.00, respectively. These two hypotheses are supported also by maximum likelihood bootstrap probabilities BP = 79.9 and 99.4%, respectively. The relationships among families of the Neococcoidea are also resolved, being supported by Bayesian posterior probabilities of 0.90–1.00 and BP = 58.7–100%. Thus, our phylogenetic tree (Fig. 1) provides a sound basis for reconstructing the evolutionary history of PGL in scale insects.
Two hypotheses have been proposed for the origin of PGL in scale insects; the single-origin hypothesis (Nur, 1980) and the independent origin (multiple-origin) hypothesis (Brown, 1977). In the latter hypothesis, Brown (1977) claimed that the L and C systems arose independently from XX-XO and the D system arose from the C system. Our result did not support this hypothesis but supported the suggestion that the Neococcoidea, which have PGL, are monophyletic. This finding is consistent with morphological evidence that the Neococcoidea are characterized by synapomorphies as needle-like apical setae on the labium (Koteja, 1974, 1996), shared structural and developmental features of the ovaries (Szklarzewicz, 1998), and the absence of abdominal spiracles (Morrison, 1928). Now, based on both molecular and morphological evidence, we can conclude that PGL is of single origin and maintained through the divergence of the Neococcoidea.
PGL is an unusual system known only in 7 groups of the arthropods (Otto and Jarne, 2001). This rarity is in contrast to the wider occurrence of haplodiploidy, known in 19 and more groups of arthropods (Otto and Jarne, 2001; Normark, 2004). Two reasons could explain this rarity; (1) PGL is an unstable state that is prone to change to other systems such as haplodiploidy (Cruickshank and Thomas, 1999; Normark, 2003, 2004) and (2) PGL is a stable state (Haig, 1993; Úbeda and Normark, 2006) but is difficult to evolve.
De Jong et al. (1981), Cruickshank and Thomas (1999) and Sabelis et al. (2002) claimed that PGL is an unstable state in the mesostigmatic mites in which a wide array of chromosome systems are found; XX-XO, XX-XY, haplodiploidy, and PGL that closely resembles the D system of scale insects. They considered that PGL in the mesostigmatic mites is a transit stage to the evolution of haplodiploidy as (1) a clade of haplodiploidy is derived from ancestors with PGL and is sister to a clade with PGL (Cruickshank and Thomas, 1999) and (2) two species of haplodiploidy have a heterochromatic chromosome arm, which is interpreted as the vestige of the heterochromatized complement of a PGL ancestor (de Jong et al., 1981). Evidence from mites does support that haplodiploidy evolved from PGL, as is suggested by Normark (2003, 2004), but does not always support the evolutionary instability of PGL. On the other hand, the result described above suggests that PGL is maintained through the evolutionary history of the Neococcoidea, that diverged to approximately 7,000 species (Ben-Dov et al., 2006). Although our knowledge about chromosome systems in scale insects is limited and observation in more species might reveal additional evolutionary changes such as independent loss of PGL, it may be likely that PGL is a rather stable state in scale insects.
Stability of PGL has been supported by some theoretical studies. There are two major theories for the evolution of PGL; one considers the conflict between the X chromosome and autosomes (Haig, 1993), while another assumes the conflict between the cytoplasmic genome of endosymbionts and the nuclear genome (Hamilton, 1993; Úbeda and Normark, 2006). Haig (1993) argued that XX-XO is a likely starting point for the origin of PGL, and meiotic drive genes on the X chromosome in XO males can spread in a population. Under this condition, autosomal genes would evolve to exploit X-drive by being preferentially included in X-bearing sperm. If all the maternal autosomes in males joined X-drive, those are effectively linked to the X chromosome. Therefore, the elimination of the total paternal genome would evolve, which is equivalent to haplodiploidy. Another hypothesis (Normark, 2004; Úbeda and Normark, 2006) assumes that PGL is the phenotype of maternally transmitted endosymbiont genes acting in male zygotes. Úbeda and Normark (2006) studied the invasion conditions of cytoplasmic male haplodizers that turn male zygotes haploid. The only outcome of a coevolutionary process between male haploidizers and their host would be a perfect transmission of male haplodizers and perfect survivorship of haploidized males. It remains uncertain which hypothesis holds in scale insects. Once PGL is established in either process, however, PGL is likely to be integrated with host’s genetic system and any functional disruption of the system would be deleterious. Thus, these hypotheses provide plausible explanations for apparent stability of PGL in scale insects and critical clues to elucidate its genetic background.
Haplodiploidy is probably derived from PGL in mesostigmatic mites. In scale insects, however, the genus Icerya, having the haplodiploidy system, is related to the other Margarodidae, having the XX-XO system (Fig. 2). Both PGL and haplodiploidy are also known in scolytid beetles. A phylogeny of scolytid beetles, which have diplodiploidy, PGL and haplodiploidy, showed that a haplo-diploid clade had apparently evolved directly from a diplo-diploid species and a PGL clade had not evolved to haplodiploidy (Normark et al., 1999). This provided the same result as was obtained from scale insects. PGL is rare and found only in arthropods but haplodiploidy is found widespread in the animal kingdom; perhaps at least in 19 clades (Otto and Jarne, 2001; Normark, 2004). This fact and results from scale insects and scolytid beetles suggest that the evolution of haplodiploidy directly from diplodiploidy is a more general trend than its evolution through PGL.
Under PGL, the paternal genome is once transmitted to male eggs and thus a male begins his life as a diploid. However, the paternal genome is subsequently inactivated or eliminated and thus adult males are functionally haploids. Therefore, from the view point of the inheritance system, PGL is equivalent to haplodiploidy. Although haplodiploidy is probably derived from PGL in mesostigmatic mites, in scale insects PGL and haplodiploidy independently derived from diplodiploidy, suggesting that the two systems would be alternative stable states.
Only in scale insects, there are three PGL systems; L, C and D. These systems differ in the time at which paternal chromosomes are lost. Two major hypotheses, based on morphological characters and chromosome systems, have been proposed for origin and relationships of PGL systems in scale insects. Nur (1980) assumed that the L system evolved first from the XX-XO sex determination and then the C system evolved from the L system and the D system evolved from the C system. Nur (1980) also suggested a reversal from the C system to the L system. On the other hand, Brown (1977) claimed that the L and C systems arose independently from XX-XO and the D system evolved from the C system.
We have put forward two plausible reconstructions for the evolutionary history of PGL systems (Fig. 2); whether Kermesidae have the L (a) or C (b) system, respectively. Our results did not determine whether the ancestral PGL system was the L or C system. Further studies are required to determine the ancestral PGL system in scale insects. On the other hand, our results did show that the L system arose twice and derived from the C system at least once. Fig. 2 resulted in single origin of the D system because it is known only in Diaspididae and we could only examine four Diaspididae species with the D system. While Morse and Normark (2006) showed that the D system arose at least twice from the C system in Diaspididae, based on the molecular phylogeny of 89 species including our four species in Diaspididae, our result about the relationship of the four species is consistent with their study.
If Kermesidae had the L system (Fig. 2(a)), this system was the ancestral PGL system of the Kermesidae + Cerococcidae + Asterolecaniidae (KCA) clade. On the contrary, if Kermesidae had the C system (Fig. 2(b)), the C was the ancestral system of the KCA clade. Thus, further studies of the chromosome system in Kermesidae are needed to determine which of the PGL systems is ancestral in the KCA clade. In addition, further studies on the chromosome system and phylogeny of scale insects could document more cases of the transitions of the PGL systems.
PGL is one of the most striking examples of genomic imprinting, by which homologous alleles or chromosomes behave differently depending on whether it has passed through sperm or egg. Little is known about the mechanisms underlying PGL. In particular, almost nothing is known at the molecular level. It is suggested that DNA methylation acts as heritable yet reversible marking associated with chromosome inactivation (heterochromatization) because paternal chromosomes are hypomethylated compared with maternal chromosomes (Field et al., 2004).
Several authors have suggested that endosymbiotic bacteria seem to have been involved with the origin of PGL (for review see Normark, 2004; Úbeda and Normark, 2006). An interesting example is known in scale insects: the reversion from PGL to diplodiploidy associated with endosymbiotic bacteria. In Stictococcus, which belongs to the Neococcoidea and is related to the Eriococcidae (Cook et al., 2002, Cook and Gullan, 2004), embryos that receive endosymbiotic bacteria become females and those that receive no bacteria become diploid males (Buchner, 1965). This provides prima facie evidence of a role for endosymbiotic bacteria in PGL. For the present, however, the roles of DNA methylation and endosymbiotic bacteria on the evolution of PGL in scale insects remain an open question. A well-resolved phylogeny, as we have presented here, will provide a base for further studies in scale insects, including challenges for elucidating the roles of DNA methylation and endosymbiotic bacteria.
We are thankful to E. Kasuya for giving theme, introducing the ideas of testing hypothesis based on phylogenetic tree, and helpful comments. We are very grateful to T. Fukatsu, N. Nikoh and Y. Kobayakawa for advice on molecular experiments. Thanks to S. Kawai, M. Shoubu and M. Tanaka for providing useful information about scale insects, specimens of great value and identification, and to the members of Itami City Museum of Insects and Ogasawara Subtropical Branch of Tokyo Metropolitan Agricultural Experiment Station for their assistance with field work. We thank C. Wood for revising the English manuscript. We would like to thank to two anonymous referees and Kyoichi Sawamura as the editor for productive comments. This study was in part supported by Grant-in-Aid, 12304048, and 16370013 for Scientific Research from the Ministry of Education, Science, Sports and Culture, Japan.
|
Index