General observation of midgut crypts
In all of the 14 acanthosomatid species listed in Additional file 1
, well-developed crypts were found in the fourth section of the midgut (Elasmostethus humeralis
and its midgut are shown in Figure ). The midgut crypts were white in color and arranged in two rows, fused into two-dimensional assemblages and forming a butterfly-shaped organ. When the organs of E. humeralis
, Elasmucha signoreti
, Sastragala esakii
, and Acanthosoma giganteum
were subjected to sectioning microscopy, no connection was found between the crypt lumen and the midgut main tract (a section image of the midgut crypt of E. humeralis
is shown in Figure ).
Figure 1 Specialized organs of Elasmostethus humeralis for harboring the symbiotic bacteria. (A) An adult female. (B) A dissected midgut: m1, midgut first section; m2, midgut second section; m3, midgut third section; m4, midgut fourth section with crypts; h, hindgut. (more ...)
Bacterial 16S rRNA gene sequences from midgut crypts
From 14 acanthosomatid species representing 18 populations (Additional file 1
), the crypt-bearing midgut was dissected and subjected to DNA extraction. From the 18 DNA samples, a 1.5 kb segment of bacterial 16S rRNA
gene was amplified by PCR and cloned, and 10 or more clones for each of the samples were subjected to RFLP genotyping. Almost all of the clones derived from a single insect exhibited identical restriction fragment length polymorphism (RFLP) patterns, except for one clone from E. signoreti
and one from Acanthosoma labiduroides
Three or more clones for each of the 18 samples were sequenced. All of the sequences derived not only from a single insect but also from the same species were completely identical to each other. The sequences from 14 acanthosomatid species showed high sequence similarities ranging from 95.4% to 99.7% to each other. DNA database searches with these sequences identified over 90% similarities to 16S rRNA gene sequences of γ-proteobacterial representatives. The exceptional clones identified by RFLP genotyping from E. signoreti and A. labiduroides were also sequenced, which showed the highest similarities to 16S rRNA gene sequences of Wolbachia sp. (supergroup B; M84686; 99.6%) and Spiroplasma citri (X63781; 98.1%), respectively.
In situ hybridization of midgut crypts targeting the bacterial 16S rRNA
To confirm whether the obtained 16S rRNA gene sequences were definitely derived from the gut symbiotic bacteria of the acanthosomatid stinkbugs, we performed 16S rRNA-targeted in situ hybridization with specific oligonucleotide probes. In E. humeralis, in situ hybridization with the specific probes Cy5-EhSym16S and TNKM16S-A555 detected dense signals in the content of the midgut crypts (Figure ). Such signals were observed neither in the midgut main tract nor in the tissue connecting the crypts to the main tract. In Elasmostethus nubilus, S. esakii, E. signoreti, and Acanthosoma denticaudum, in situ hybridization with the group-specific probe Cy5-AcSym16S also detected the symbiont signals specifically in the cavity of the midgut crypts in the same manner (data not shown). A series of control experiments confirmed the specificity of the hybridization signals (data not shown).
Electron microscopy of midgut crypts
Ultrathin sections of the midgut crypts from E. humeralis and E. nubilus were observed by electron microscopy. In both the species, the lumen of the crypts was full of symbiotic bacteria (Figure and ). The rod-shaped bacterial cells, whose cell wall looked very thin, were larger in E. humeralis than in E. nubilus (Figure and ). The cytoplasm of the epithelial cells of the midgut crypt contained a nucleus and many mitochondria but no bacterial cells (Figure and ).
Figure 2 Transmission electron microscopy of the symbiotic organ of acanthosomatid stinkbugs. (A) Midgut crypts of Elasmostethus humeralis. (B) Midgut crypts of Elasmostethus nubilus. (C) Symbiotic bacteria of E. humeralis. (D) Symbiotic bacteria of E. nubilus (more ...)
Phylogenetic analysis of symbiotic bacteria based on 16S rRNA gene
The 16S rRNA gene sequences originating from the gut symbionts of the acanthosomatid stinkbugs, representing 18 populations, 14 species, and 5 genera, were subjected to molecular phylogenetic analyses together with 16S rRNA gene sequences of γ-proteobacterial representatives. The acanthosomatid symbionts formed a monophyletic group with high supporting values (100% in Bayesian, 100% in maximum parsimony (MP), and 95% in maximum likelihood (ML), respectively). The phylogenetic relationship of the symbionts was generally in agreement with the systematics of the host insects: the symbionts from congenic host species, Elasmostethus spp., Elasmucha spp., Sastragala spp., and Acanthosoma spp., formed clades, respectively. The monophyletic group of the acanthosomatid symbionts showed a phylogenetic affinity to the clade of Buchnera, obligate endocellular symbionts of aphids, and also to the clade of Ishikawaella, obligate gut symbionts of plataspid stinkbugs (Figure ).
Figure 3 Phylogenetic placement of the symbiotic bacteria from the acanthosomatid stinkbugs in the γ-Proteobacteria on the basis of 16S rRNA gene sequences. A total of 1271 aligned nucleotide sites were subjected to the analysis. A Bayesian phylogeny is (more ...)
Phylogenetic analysis of symbiotic bacteria based on a protein-coding gene
To further confirm the phylogenetic placement of the symbiotic bacteria, a groEL gene segment was cloned and sequenced from E. nubilus, Elasmucha putoni, and S. esakii. Molecular phylogenetic analyses of the sequences revealed that the sequences of the acanthosomatid symbionts formed a well-supported monophyletic group and clustered with the clade of aphid endosymbionts Buchnera and also the clade of plataspid gut symbionts Ishikawaella in the γ-Proteobacteria (Figure ). The results of groEL gene analyses were generally concordant with the results of 16S rRNA gene analyses (cf. Figure ).
Figure 4 Phylogenetic placement of the symbiotic bacteria from the acanthosomatid stinkbugs in the γ-Proteobacteria on the basis of groEL gene sequences. A total of 1040 aligned nucleotide sites at first and second codon positions were subjected to the (more ...)
Prevalence of symbiotic bacteria in natural host populations
Diagnostic PCR surveys of field-collected acanthosomatid stinkbugs consistently detected 100% infection frequencies of the symbiotic bacteria in all the species wherein multiple samples were examined: 38/38 in E. humeralis; 50/50 in E. nubilus; 6/6 in E. putoni; and 6/6 in S. esakii.
Detection of symbiotic bacteria from dissected tissues
Dissected tissues, including head, flight muscle, foregut, midgut first section, midgut second section, midgut third section, midgut fourth section with crypts, hindgut, abdominal tip, ovary, and egg, were prepared from adult females of E. humeralis, E. nubilus, and E. putoni, and were subjected to diagnostic PCR detection of the symbiotic bacteria. In all the species, positive signals were consistently detected from midgut fourth section, abdominal tip, and egg (Figure ).
Figure 5 Diagnostic PCR detection of the symbiotic bacteria from dissected tissues of acanthosomatid stinkbugs. The results of Elasmostethus humeralis (A and B), Elasmostethus nubilus (C and D), and Elasmucha putoni (E and F) are shown. (A, C, and E) Detection (more ...)
Vertical transmission of symbiotic bacteria via egg surface contamination
Diagnostic PCR surveys of field-collected egg masses consistently detected the symbiotic bacteria from all the eggs examined: 89 eggs from three egg masses of E. humeralis, 23 eggs from one egg mass of E. nubilus, 80 eggs from four egg masses of S. esakii, and 92 eggs from three egg masses of A. giganteum (data not shown). Meanwhile, the symbiotic bacteria were not detected from dissected ovaries of E. humeralis, E. nubilus and E. putoni (Figure ), refuting the possibility of ovarial symbiont transmission. Newborn nymphs of these species exhibited a characteristic behavior, probing of egg surface with their proboscis, soon after hatching (Kikuchi, personal observation), suggesting the possibility of symbiont transmission via egg surface contamination.
Sterilization of egg surface disrupted symbiont transmission to newborn nymphs
In an attempt to experimentally confirm the possibility of symbiont transmission via egg surface contamination, we divided each of the egg masses of E. humeralis into two portions. One of the halves was left untreated, the other half was surface-sterilized, and newborn nymphs from these experimental egg masses were subjected to diagnostic PCR detection of the symbiotic bacteria after hatching. Most of the nymphs from the control egg masses were symbiont-positive, whereas all of the nymphs from the sterilized egg masses were symbiont-negative (Table ).
Effect of surface sterilization of eggs on symbiont acquisition in Elasmostethus humeralis.
Effects of symbiont elimination on fitness and phenotype of host insects
Between the control egg masses and the sterilized egg masses, no significant differences were found in time to hatching (control, 4.1 ± 0.5 days, n = 182; sterilized, 4.1 ± 0.4 days, n = 180) and hatching rate (control, 95.4 ± 6.1%, n = 14; surface sterilized, 97.3 ± 4.0%, n = 14), indicating that surface sterilization of eggs did not affect the embryonic development of the insect. However, when we examined another set of experimental egg masses of E. humeralis for inspection of post-hatch growth and development, drastic differences were detected between the sterilized group and the control group. Adult emergence rate was significantly lower in the sterilized group than in the control group (Figure ). Developmental time to adulthood was significantly longer in the sterilized group (31.8 ± 6.6 days, n = 6) than in the control group (23.9 ± 2.9 days, n = 31) (Figure ). Body size in terms of thorax width was not statistically different between the sterilized group (4.59 ± 0.17 mm, n = 6) and the control group (4.64 ± 0.32 mm, n = 31) (Figure ). However, adult insects from the control egg masses were normal in color (Figure ), whereas adult insects from the sterilized egg masses exhibited abnormal pale coloration (Figure ). Diagnostic PCR detection confirmed that all the adult insects from the sterilized group were symbiont-free, except an individual exhibiting a faint PCR signal (data not shown).
Figure 6 Effects of symbiont elimination on fitness parameters and phenotype of Elasmostethus humeralis. (A) Adult emergence rate (%). Emerged insects per total insects and P values of Fisher's exact probability test are indicated. (B) Developmental time to adulthood (more ...)
Localization of symbiotic bacteria in lubricating organ
In the abdomen of adult females of E. humeralis
, E. nubilus
, and other acanthosomatid species, we identified a pair of characteristic 'lubricating organs' on the ventral side of the body cavity near the abdominal tip (Figure ) which were covered with yellowish membrane (Figure ) and lined with cuticular layer (Figure ), as previously reported by Rosenkranz [28
]. In situ
hybridization identified with certainty strong signals of the symbiotic bacteria in the lubricating organ (Figure ). The symbiont in the lubricating organ was shown to be identical to that in the midgut crypts on the basis of cloning and sequencing of 16S rRNA
gene (data not shown). The lubricating organ consisted of two distinct regions: the sac-like region populated by the symbiotic bacteria and the symbiont-free chitinous ridge region (Figure and ). On the outer surface of the sac-like part, numerous tubulet-like structures were densely arranged into a layer, wherein the symbiont signals were localized (Figure , inset). The organ was not found in adult males (data not shown).
Phylogenetic analysis of host insects based on a mitochondrial gene
From the 14 acanthosomatid species, a mitochondrial COI gene segment was cloned and sequenced. Molecular phylogenetic analyses of the sequences revealed that congenic species, namely Elasmostethus spp., Elasmucha spp., Sastragala spp., and Acanthosoma spp., formed distinct clades, respectively (Figure ), which was in good agreement with the insect systematics.
Figure 7 Phylogenetic relationship of the acanthosomatid stinkbugs on the basis of mitochondrial COI gene sequences. A total of 611 aligned nucleotide sites were subjected to the analysis. A Bayesian phylogeny is shown. Support values for the nodes are indicated (more ...)
Host-symbiont co-evolutionary analysis
Figure contrasts the phylogeny of the acanthosomatid stinkbugs and the phylogeny of their gut symbiotic bacteria. The symbiont phylogeny showed a remarkable similarity to the host phylogeny. The only local discrepancies were seen with the placements of Lindbergicoris gramineus
and Acanthosoma forficula
. The history of the host-symbiont association was inferred by two different algorithms using the programs TreeMap [30
] and TreeFitter [31
]. Both methods identified possible co-evolutionary histories with 10 to 11 co-divergence events (Table ). The consistently-observed 10 co-divergence events are mapped in Figure , based on the result of the TreeMap analysis. Both the TreeMap (Table ) and TreeFitter (Table ) inferred significantly more co-speciation events than expected from a random distribution, regardless of the chosen topologies for the unsolved nodes. The Icong index [32
] also indicated that the topological congruence between the host and symbiont trees was statistically significant (Table ).
Figure 8 Phylogenetic congruence between the acanthosomatid stinkbugs and their symbiotic bacteria. Phylogenetic tree of the host insects (left; cf. Figure 7) and that of the symbiotic bacteria (right; cf. Figure 3) are contrasted. Each symbiont is connected to (more ...)
Results of the co-divergence analyses with fully resolved trees of the acanthosomatid symbionts.
AT-rich genes of acanthosomatid symbionts
When the base compositions of the 16S rRNA gene region from γ-proteobacterial representatives were inspected, free-living bacteria like Escherichia coli and Salmonella typhi exhibited low AT content of around 45%. On the other hand, obligate endocellular insect symbionts such as Buchnera, Wigglesworthia, Blochmannia, and Baumannia exhibited remarkably higher AT content of over 50%. The AT content of the acanthosomatid gut symbionts were, together with those of the plataspid gut symbionts Ishikawaella, consistently over 50% (cf. Figure ), which were equivalent to the values of the obligate insect endosymbionts rather than the values of the free-living bacteria. When the base compositions of the groEL gene region were examined, similar patterns were observed: free-living bacteria like E. coli and S. typhi exhibited AT content of around 45%; obligate endocellular insect symbionts such as Buchnera, Wigglesworthia, Blochmannia, and Baumannia over 60%; the gut symbionts of plataspid stinkbugs Ishikawaella 61.5%; and the gut symbionts of acanthosomatid stinkbugs 62–64% (cf. Figure ).
Accelerated molecular evolution in acanthosomatid symbionts
Table summarizes the results of relative rate tests for the 16S rRNA gene from the lineages of acanthosomatid gut symbionts, obligate aphid endosymbionts Buchnera, plataspid gut symbionts Ishikawaella, and related free-living bacteria. The molecular evolutionary rates in the lineage of acanthosomatid symbionts were significantly higher than those of the free-living bacteria, and were similar to those in the lineages of Buchnera and Ishikawaella. Table shows the results of relative rate tests for the groEL gene sequences, which exhibited similar evolutionary patterns.
Relative rate tests for comparing the molecular evolutionary rate of 16S rRNA gene among the lineages of the symbionts of acanthosomatid stinkbugs, symbionts of plataspid stinkbugs, endocellular symbionts of aphids, and their free-living relatives.
Table 4 Relative rate tests for comparing the molecular evolutionary rate of groEL gene (nucleotide 1st, 2nd positions of codon) among the lineages of the symbionts of acanthosomatid stinkbugs, symbionts of plataspid stinkbugs, endocellular symbionts of aphids, (more ...)
Reduced genome size of acanthosomatid symbionts
Figure shows the pulsed-field gel electrophoresis of the genomic DNA of the acanthosomatid gut symbionts. The symbiont genome sizes were estimated to be 0.93 Mb for E. humeralis, 0.90–0.94 Mb for E. nubilus, and 0.95–0.96 Mb for S. esakii. The genome sizes were much smaller than those of free-living γ-proteobacteria like E. coli (4.6 Mb), S. typhi (4.8 Mb), and V. cholerae (4.0 Mb), and close to those of obligate endocellular insect symbionts such as Buchnera (0.42–0.65 Mb), Wigglesworthia (0.70 Mb), Blochmannia (0.71–0.79 Mb), and Baumannia (0.69 Mb), and also to those of plataspid gut symbionts Ishikawaella (0.82–0.83 Mb).
Figure 9 Pulsed field gel electrophoresis of the symbiont genomic DNA prepared from dissected midgut crypts of the acanthosomatid stinkbugs. (A) Elasmostethus humeralis. (B) Elasmostethus nubilus. (C) Sastragala esakii. Lane M1, size marker (Saccharomyces cerevisiae (more ...)