The order Homoptera, which includes cicadas, planthoppers, aphids, scale insects, psyllids, and other insects, is a group whose members have highly developed endosymbiotic systems (6). Because homopterans live on a nutritionally unbalanced diet consisting of plant sap throughout their lives, it is believed that they need the help of endosymbiotic microorganisms to compensate for the nutritional deficiency. In fact, it has been demonstrated that endosymbiotic microbes of homopterans are involved in metabolic processes, such as synthesis of essential nutrients and recycling of nitrogenous wastes (5, 11–13, 36). These endosymbiotic microorganisms have remained generally unculturable, probably because they are highly adapted to the special environments inside their host organisms and cannot live outside the hosts (4). Therefore, PCR, DNA sequencing, and molecular phylogenetic analyses of microbial genes are powerful approaches that are used to infer the systematic affinities of these fastidious endosymbionts (8, 20, 22, 30–32, 37, 44).

It is commonly found that multiple microbial species coexist in an insect body; these species constitute a complex endosymbiotic microbiota not only in members of the Homoptera but also in members of other insect groups (6, 19, 21, 23). In studies of these organisms the results of a simple PCR and sequencing approach are misleading because they are quite difficult to interpret. A 16S rRNA gene (rDNA) preparation amplified and cloned from whole insect DNA often contains a number of different sequences which might come from multiple endosymbionts, gut microbes, pathogens, occasionally contaminating bacteria, or debris adhering to the insect surface. In addition, possible amplification biases inherent in PCR and DNA cloning can lead to false conclusions. Therefore, the microbial DNA sequences obtained must be interpreted in connection with morphological data by using, for example, in situ hybridization with specifically designed probes (3, 22).

The scale insects in the superfamily Coccoidea include approximately 6,000 species grouped in 15 to 20 families whose delimitation is still controversial (28). There have been a number of histological descriptions of endosymbiosis in scale insects (6, 41–43), and these descriptions have highlighted the amazing diversity and complexity of the endosymbiontic systems. For example, some scale insects harbor bacteria, while others contain yeastlike organisms; and some appear to be monosymbiotic, whereas others harbor several types of microbes. In addition, there is considerable diversity in morphological characteristics, histological distribution, and mode of transmission of the symbionts between families, within a family, or even in the same genus. Therefore, the scale insects are an interesting group with which to investigate the process and dynamics of endosymbiotic evolution. However, there have been only two previous studies on the endosymbionts of members of the family Pseudococcidae in which the researchers used a molecular phylogenetic approach. Munson et al. (31) found 16S rDNA sequences that belonged to members of the β subdivision of the division Proteobacteria (β-Proteobacteria) in Pseudococcus longispinus, Pseudococcus maritimus, and Dysmicoccus neobrevipes, and Kantheti et al. (25) characterized a 16S rDNA sequence from Planococcus lilacinus that was placed in the γ-Proteobacteria and was associated with the mycetocytes of the host insect. The apparent discrepancies in these two reports suggest that careful and detailed analyses are needed to characterize the complex endosymbiotic microbiota of members of the Pseudococcidae.

In this study, we identified three distinct intracellular symbiotic bacteria in the bamboo pseudococcid Antonina crawii by using a molecular phylogenetic approach combined with in situ hybridization.

Materials. The insect species used in this study are listed in Table 1. Female adults of A. crawii were collected several times in June 1997 on the campus of the University of Tokyo and were preserved in acetone (18). The other pseudococcids examined were also collected and kept in acetone.

DNA extraction. The insects preserved in acetone were separated from their waxy secretions and were repeatedly washed with fresh acetone to minimize possible contamination. After the insects were placed on clean tissue paper to remove the preservative, they were individually subjected to a DNA extraction procedure by using a QIAamp tissue kit (QIAGEN).

Molecular biological procedures. Eubacterial 16S rDNA in the whole-insect DNA (length, about 1.5 kb) was amplified by PCR by using primers 16SA1 (5′-AGAGTTTGATCMTGGCTCAG-3′) and 16SB1 (5′-TACGGYTACCTTGTTACGACTT-3′) with the following temperature profile: 94°C for 2 min, followed by 30 cycles of 94°C for 1 min, 50°C for 1 min, and 70°C for 2 min as previously described (22).

Molecular phylogenetic analysis. Multiple alignment of 16S rDNA sequences was accomplished by using the methods of Feng and Doolittle (17) and Gotoh (24). The final alignment was inspected and corrected manually. Ambiguously aligned regions were excluded from the phylogenetic analysis. Nucleotide sites that included alignment gaps were also omitted from the aligned data set. Neighbor-joining trees (35) were constructed by using Kimura's two-parameter distance (26) and the Clustal W program package (40). Maximum-likelihood trees (15) were constructed by using the MORPHY program package (version 2.3) (1). In heuristic searches for an optimal tree with the best log-likelihood score, we used quick add OTU search and local rearrangement search (1). Maximum-parsimony trees were constructed by using the PAUP program package (version 4.0b2) (38). Bootstrap tests (16) were conducted with 1,000 resamplings.

Histology. Histological preparation, in situ hybridization, and enzymatic probe detection were performed as previously described (23). The insects preserved in acetone were transferred to alcoholic formalin (ratio of ethanol to formalin, 3:1), and their lateral cuticles were removed with a razor blade to aid infiltration of reagents. After overnight fixation, the preparations were dehydrated and cleared with an ethanol-xylene series and then embedded in paraffin. Serial tissue sections (thickness, 5 μm) were cut with a rotary microtome and mounted on silane-coated glass slides. The sections were dewaxed with a xylene-ethanol series and air dried prior to in situ hybridization.

In situ hybridization. The sequences of specific oligonucleotide probes DIG-TKSγ, DIG-TKSβ, and DIG-TKSspi, which were used in this study, are shown in Table 2. About 150 μl of hybridization buffer (20 mM Tris-HCl [pH 8.0], 0.9 M NaCl, 0.01% sodium dodecyl sulfate, 30% formamide) containing 70 pmol of probe per ml was applied to a tissue section, which was then covered with a coverslip and incubated in a humidified chamber at room temperature overnight. To remove nonspecifically bound probe, the tissue section was rinsed with washing buffer (20 mM Tris-HCl [pH 8.0], 0.9 M NaCl, 0.01% sodium dodecyl sulfate, 30% formamide) for 10 min at 30°C. After the preparation was washed with 1× SSC (0.15 M NaCl, 0.015 M sodium citrate), the bound probe on the tissue section was detected by using a DIG nucleic acid detection kit (Boehringer Mannheim) as previously described (22). To confirm the specificity of hybridization, the following control experiments were conducted: a no-probe control experiment, an RNase digestion control experiment, and a competitive suppression control experiment with excess unlabelled probe (23). We also performed control experiments with a widely used general eubacterial 16S rRNA probe, digoxigenin-labelled probe DIG-EUB338 (2, 3).

Probe or primer	Sequencea
Probes
DIG-TKSγ	5′-TTTGTATGCACCATTGTA-3′
DIG-TKSβ	5′-TTTGCTCCAGCCATTGTA-3′
DIG-TKSspi	5′-TTTGTATCAGCCATTGTA-3′
Primers
TKSγsp	5′-TAGCCGGTGCTTCTTTTACG-3′
TKSβsp	5′-TAGCCGGTGCTTATTCTTTC-3′
TKSSsp	5′-TAGCCGTGGCTTTCTGGTAA-3′

Diagnostic PCR. Using specific reverse PCR primers TKSγsp, TKSβsp, and TKSSsp (Table 2) in combination with universal forward primer 16SA1, we performed diagnostic PCR experiments to detect 16S rDNA of endosymbiotic bacteria by using the following temperature profile: 94°C for 2 min, followed by 30 cycles consisting of 94°C for 1 min, 55°C for 1 min, and 70°C for 2 min.

Nucleotide sequence accession numbers. The partial 16S rDNA sequences of the γ-Proteobacteria symbiont (γ-symbiont), β-symbiont, and spiroplasma symbiont of A. crawii described in this paper have been deposited in the DDBJ, EMBL, and GenBank nucleotide sequence databases under accession no. AB030020, AB030021, and AB030022, respectively.

Identification of three types of 16S rDNA. Approximately 1.5 kb of eubacterial 16S rDNA was amplified by PCR from the total DNA of A. crawii. Restriction fragment length polymorphism (RFLP) analysis of cloned fragments revealed three major types of sequences, which were tentatively designed types A, B, and C (Fig. 1).

FIG. 1 RFLP analysis of bacterial 16S rDNA amplified and cloned from total DNA of A. crawii. Lanes 1 through 10 contained the cloned 16S rDNA fragments digested with either RsaI or HinfI and resolved on a 2% agarose gel. Lanes 1, 3, 4, 7, and 8, type (more ...)

The nucleotide sequences of the three types of clones were determined. Two clones of each type were subjected to a sequencing analysis. The sequences of the two type A clones were identical. The sequences of the two type B clones were identical except at one nucleotide site, as were the sequences of the two type C clones. The differences may have been due to a PCR error or to a nucleotide substitution in the different 16S rDNA copies. The sequence lengths (not including the regions of amplifying primers) were 1,465 bases for the type A clones, 1,474 bases for the type B clones, and 1,443 bases for the type C clones. The RFLP profiles predicted from the sequences were consistent with the profiles shown in Fig. 1.

Molecular phylogenetic analysis. In the eubacterial phylogenetic analysis, the three sequences were placed in distinct lineages (Fig. 2). The type A organism was a member of the γ-Proteobacteria. No sequence in the DNA databases was closely related to the type A sequence. However, the type A organism formed a cluster with the intracellular symbiotic bacteria of other insects, such as the Y symbiont of psyllids, the Buchnera spp. of aphids, and the Candidatus camponotii of an ant, although the bootstrap support for the group was quite low (Fig. 2A). Below, this putative endosymbiotic bacterium is referred to as the γ-symbiont. The type B organism formed a good monophyletic group, supported by a bootstrap value of 100%, with the endosymbionts of the mealybugs D. neobrevipes, P. longispinus, and P. maritimus, in the β-Proteobacteria (Fig. 2A). This putative endosymbiotic bacterium was designated the β-symbiont. The type C organism was placed in the Spiroplasma-Mycoplasma clade of the Mycoplasmatales (Mollicutes). The type C sequence formed a very compact monophyletic group, supported by a bootstrap value of 100%, with Spiroplasma spp. from ladybird beetles and a tick (Fig. 2B). This bacterium was designed the spiroplasma symbiont.

FIG. 2 Molecular phylogenetic analysis of the endosymbionts of A. crawii based on 16S rDNA sequences. (A) Phylogenetic positions of the γ-symbiont and the β-symbiont in the Proteobacteria. A total of 1,184 unambiguously aligned nucleotide sites (more ...)

In situ hybridization. In order to demonstrate that our 16S rDNA sequences were definitely derived from endosymbiotic bacteria of A. crawii, we designed specific oligonucleotide probes for the sequences (Table 2). Using these digoxigenin-labelled probes, we specifically visualized localization of the three types of 16S rDNA sequences on tissue sections of A. crawii (Fig. 3). The results of a series of control experiments confirmed the specificity of the detection (data not shown). Hybridization with probe DIG-EUB338, which recognizes eubacteria, did not reveal the presence of any bacterial associates other than the symbionts described above in A. crawii (data not shown).

FIG. 3 Specific detection of endosymbiotic bacteria in tissue sections of A. crawii by in situ hybridization. Probes DIG-TKSγ, DIG-TKSβ, and DIG-TKSspi targeted the γ-symbionts, the β-symbionts, and the spiroplasma symbionts in (more ...)

Localization of the γ-symbiont. When probed with DIG-TKSγ, the cytoplasm of the mycetocytes, which formed a large mycetome in the abdomen, was specifically stained (Fig. 3A). In the cytoplasm of the mycetocytes were compartmentlike structures in which a number of tubular or sausagelike bacteria were present (Fig. 3D). In the mycetocytes, the bacterial cells were 2.5 to 3.5 μm wide and up to 30 μm long, although precise estimates were difficult to obtain with tissue sections. In the developing eggs on the lateral oviducts, signals were obtained with probe DIG-TKSγ at the anterior poles of oocytes, where nurse cells and oocytes were connected, around the nutritive cord (Fig. 3G). The γ-symbionts were more pleomorphic in the eggs than in the mycetocytes.

Localization of the β-symbiont. When probe DIG-TKSβ was used, the cytoplasm of mycetocytes was stained (Fig. 3B), as it was when probe DIG-TKSγ was used. However, the intracellular structures were strikingly different at a higher magnification. In the cytoplasm of mycetocytes, the sausagelike γ-symbionts were not stained, but obscure structures around them were stained with probe DIG-TKSβ (Fig. 3E). In both the mycetocytes and the developing eggs, the localization of the β-symbionts was similar to the localization of the γ-symbionts (Fig. 3H).

Localization of the spiroplasma symbiont. In contrast to the γ- and β-symbionts, the spiroplasma symbiont was not detected in the mycetocytes (Fig. 3C). When preparations were probed with DIG-TKSspi, signals were observed intracellularly in various tissues, such as the gut (Fig. 3F and J), fat bodies (Fig. 3I and L), and epithelial tissue (Fig. 3I and K). In the cytoplasm of these tissues, the spiroplasma symbionts occurred singly or in groups at a low density. Some bacterial cells appeared to be coccoid, whereas others were pleomorphic with diameters of 5 to 15 μm, although the exact shapes and sizes were difficult to estimate when tissue sections were examined. The spiroplasma symbionts were occasionally detected in developing eggs, but they did not exhibit specific localization like the γ- and β-symbionts (data not shown).

Diagnostic PCR detection of the symbionts in members of the Pseudococcidae. In addition to A. crawii, four pseudococcids, Dysmicoccus wistariae, Planococcus kraunhiae, Planococcus citri, and Pseudococcus citriculus (Table 1), were examined to determine whether the three types of endosymbionts described above were present by using specific PCR primers for 16S rDNA. When we used TKSβsp, which was specific for the β-symbiont, we detected an amplified band in all of the species examined (Fig. 4A). When we used TKSγsp, which was specific for the γ-symbiont, D. wistariae, P. citriculus, and A. crawii produced an amplified product, whereas the two Planococcus species did not (Fig. 4B). When we used TKSSsp, which was specific for the spiroplasma symbiont, a band was detected only in A. crawii (Fig. 4C).

FIG. 4 Diagnostic PCR detection of the three types of endosymbionts in representatives of the Pseudococcidae. (A) Detection of the β-symbiont with primers 16SA1 and TKSβsp. (B) Detection of the γ-symbiont with primers 16SA1 and TKSγsp. (more ...)

Although workers have recently performed a number of molecular phylogenetic studies with sequences of 16S rDNA or other genes that were putatively derived from endosymbiotic bacteria of insects, in most of these studies the researchers did not definitively characterize the microbes from which the sequences originated. As far as we know, this is the first report that integrates information concerning the morphology, distribution in vivo, and phylogenetic positions of three intracellular endosymbiotic bacteria in an insect. In members of the Pseudococcidae, Munson et al. (31) identified 16S rDNA sequences of members of the β-Proteobacteria, whereas Kantheti et al. (25) characterized members of the γ-Proteobacteria. These seemingly discrepant reports are reconciled by the finding that pseudococcids harbor both β- and γ-symbionts (Fig. 4). In the previous studies the researchers may have failed to detect one of the symbionts because the endosymbiotic population was not adequately sampled.

In the 16S rDNA phylogenetic analysis, the β-symbiont of A. crawii constituted a strongly supported monophyletic group that also contained β-symbionts of other pseudococcids (Fig. 2A). Using diagnostic PCR performed with a primer specific for the β-symbiont, we obtained an amplified product from all five pseudococcids examined (Fig. 4A). Our results suggest that the β-symbiont is the principal intracellular symbiotic bacterium of members of the Pseudococcidae and is conserved in the group, as members of the genus Buchnera appear to be in the Aphididae (29). However, a more extensive survey of pseudococcid symbionts is needed to confirm this.

In contrast to the β-symbionts, the γ-symbionts appeared to be polyphyletic in the Pseudococcidae. In the 16S rDNA phylogenetic analysis, the γ-symbiont of A. crawii did not form a clade with the γ-symbiont of P. lilacinus (Fig. 2C). The results of neighbor-joining, maximum-likelihood, and maximum-parsimony analyses consistently supported this conclusion (data not shown). Using diagnostic PCR performed with a primer specific for the γ-symbiont, we obtained an amplified product from D. wistariae, P. citriculus, and A. crawii but not from P. kraunhiae and P. citri (Fig. 4B). An RFLP analysis of cloned 16S rDNAs from the Planococcus species revealed two major types of clones (unpublished data). These results suggested that in addition to the β-symbiont, many pseudococcids contain the γ-symbiont, which may have multiple evolutionary origins. The endosymbiotic organization of the Pseudococcidae appears to be reminiscent of the endosymbiotic organization of the Aphididae, in which many species, but not all species, contain secondary intracellular symbiotic bacteria in addition to the primary symbiont, Buchnera sp. (19, 21, 23).

In A. crawii, both the β-symbiont and the γ-symbiont were harbored in the same mycetocytes (Fig. 3D and E). A similar localization of symbionts has been found in whiteflies, whose mycetocytes contain two or three morphologically distinct types of bacteria (9, 10). In contrast, it has been reported that in aphids and psyllids two types of symbiotic bacteria are harbored separately in distinct types of mycetocytes (6, 19, 21–23). Within the Sternorrhyncha, however, coccids and aphids are thought to be phylogenetically related, while the positions of whiteflies and psyllids are controversial (7, 45). In addition, the symbionts of whiteflies and pseudococcids did not exhibit significant phylogenetic affinities (Fig. 2). Therefore, the resemblance of the disymbiotic organizations of the pseudococcids and whiteflies does not appear to be due to phylogenetic proximity of the groups. The β- and γ-symbionts also exhibited the same localization in developing eggs of A. crawii (Fig. 3G and H), suggesting that in the process of vertical transmission to the next generation, the two types of endosymbionts may recognize the same signals that specify their localization in tissues and cells of the host, although the nature of the signals and the mechanisms of localization are not understood.

The biological functions of the endosymbionts in pseudococcids have not been investigated. The only relevant study is a report which showed that injection of penicillin into the circulating sap of host plants had a lethal effect on P. citri and P. maritimus, suggesting that some bacterial associates may be essential for these pseudococcids (27). Since intracellular symbiotic bacteria harbored in well-developed mycetomes have been commonly found in the pseudococcids examined so far (6, 41), the endosymbionts may play some essential physiological and nutritional roles in the hosts, as has been demonstrated in aphids, planthoppers, and other insects (5, 11–13, 36). Notably, the β-symbiont was detected in all of the species examined in this study (Fig. 4), suggesting that the β-symbiont may have particularly important functions in the host pseudococcids. At least in A. crawii, the biomass of the γ-symbiont appeared to be comparable to the biomass of the β-symbiont (Fig. 3A, B, D, and E), reserving the possibility that the γ-symbiont also plays important roles in the host.

For the endosymbiotic systems of members of the Pseudococcidae, a number of histological descriptions are available (6, 41). In species of the genera Pseudococcus, Planococcus, Nipaeococcus, Dysmicoccus, Ferrisia, Antonina, and others which are regarded as closely related by taxonomists, the female possesses a voluminous mycetome in which many mycetocytes are enveloped by the epithelial layer. Endosymbiotic bacteria are harbored in the cytoplasm of mycetocytes, where they are embedded in peculiar globular structures composed of a kind of matrix called “mucous spherules.” Transmission to the oocytes takes place via the anterior egg pole, where free mucous spherules penetrate into the “ovariole neck.” These descriptions are totally consistent with our in situ hybridization results for A. crawii. Notably, however, previous researchers consistently regarded the pseudococcids examined to be monosymbiotic. With a number of pseudococcids, it was observed that the symbionts exhibited significant morphological variation depending on the host stage, on transmission to the offspring, and on environmental factors, such as feeding, starvation, and temperature. These observations were interpreted as the result of transformation of a single species of bacterium and were described under the terms such as “symbiont cycles,” “infection forms,” “vegetative forms,” “hungry forms,” etc. (6, 41). Certainly, many microorganisms exhibit remarkable pleomorphism. In the case of mycetocyte symbionts of pseudococcids, however, part of the previously reported pleomorphism was quite likely due to confusion resulting from the presence of two distinct endosymbionts in the same cell.

Even in the pseudococcids for which molecular phylogenetic studies have been conducted, at least three distinct lineages of mycetocyte symbionts, one β-symbiont and two γ-symbionts, have been identified. The diversity of the symbionts may be further expanded when other groups of pseudococcids are investigated in the same way. For example, it has been reported that in members of the genera Phenacoccus, Heliococcus, Centrococcus, Ripersia, and Eumyrmococcus, the symbionts are included in the cytoplasm of mycetocytes without mucous spherules. In members of the genus Rastrococcus, the symbionts are located in fat cells or in syncytial fat tissue (6, 41).

In addition to the two mycetocyte symbionts, we identified a third intracellular symbiotic bacterium that belongs to the genus Spiroplasma. Almost all members of the genus Spiroplasma that have been described are associated in some way with arthropods (47, 48). As far as we know, this is the first time that anyone has identified a spiroplasma in a member of the Coccoidea by using molecular phylogenetic techniques. Also, this is the first detailed study of the distribution of a spiroplasma in host tissues in which a highly specific in situ hybridization technique was used.

In A. crawii, the spiroplasma symbiont was associated with various tissues without any conspicuous structure or localization. The biomass of the spiroplasma symbiont was apparently much smaller than the biomasses of the β- and γ-symbionts (Fig. 3A, B, F, and I). It is unclear whether or how efficiently the spiroplasma symbiont is vertically transmitted to the next generation. Among the pseudococcid species examined, the spiroplasma symbiont was detected only in A. crawii (Fig. 5C). Although these observations are circumstantial, they suggest that the spiroplasma symbiont may be a facultatively endosymbiotic associate with a commensal or parasitic nature. In fact, most spiroplasmas are known or suspected to be parasitic, although the degrees of pathogenicity may be extremely diverse (46, 47).

The origin and life cycle of the spiroplasma symbiont of A. crawii are unclear at this time. In the 16S rDNA phylogeny, this organism was closely related to Spiroplasma spp. from ladybird beetles and a tick (Fig. 3B), suggesting the possibility that horizontal transmission from an unrelated host organism occurs. It may be ecologically significant that ladybird beetles are among the most important predators of coccids (14, 34). Some members of the genus Spiroplasma are maintained in cycles in the phloem of plants and the bodies of sap-sucking insects that vector them (46, 47). It is possible that the spiroplasma symbiont is maintained similarly in a cycle between bamboos and A. crawii.

Interestingly, the spiroplasma symbiont was not detected in the mycetome, which are heavily populated by the β- and γ-symbionts, although it was found in various other tissues and cells (Fig. 4C). It may be presumed that some direct or indirect interactions between different endosymbionts and host cells occur; these interactions are potentially interesting but totally unexplored areas of the microbial ecology of insect endosymbiosis.

We thank T. Arai for pseudococcid samples, A. Sugimura, S. Kumagai, and K. Sato for technical and secretarial assistance, and T. Wilkinson for reading the manuscript.

This research was supported by the Industrial Science and Technology Frontier Program “Technological Development of Biological Resources in Bioconsortia” of the Ministry of International Trade and Industry of Japan and by the Program for Promotion of Basic Research Activities for Innovation Biosciences (ProBRAIN) of the Bio-Oriented Technology Research Advancement Institution.