Evolutionary origin of insect–Wolbachia nutritional mutualism

We carefully dissected 26 adult bedbugs of the monosymbiotic strain JESC infected with wCle only without secondary symbionts (9), collected their bacteriomes, and extracted total DNA from the symbiotic organs. The DNA sample was subjected to shotgun library construction and Sanger sequencing. Of 18,432 reads obtained, 12,491 reads were assembled into 30 major contigs with sequence similarity to known Wolbachia genomes, and gap filling yielded a 1,250,060-bp circular bacterial genome (Fig. 1). The genome of wCle encoded 1,216 putative protein-coding ORFs with an average size of 771.6 bp, which covered 75% of the whole genome (Table S1). The genome size, GC content, coding capacity and density, and abundance of pseudogenes and insertion sequences of wCle were similar to those of facultative insect-associated Wolbachia strains such as wMel, wRi, and wPip (Table S1). The absence of remarkable genome degeneration in wCle suggests that, unlike the ancient aphid–Buchnera and tsetse–Wigglesworthia nutritional mutualisms (3, 4), the bedbug–Wolbachia mutualism is of relatively recent evolutionary origin. Of the 1,216 protein-coding genes, 816 were assigned to putative biological functions, 321 matched hypothetical proteins of unknown function, and 79 were not assigned to any genes in the databases. Overall, cluster of orthologous groups category composition of the wCle genome was similar to those of other insect-associated Wolbachia genomes (Table S2), indicating basically similar metabolic capacities among the different Wolbachia strains.

However, inspection of synthetic pathways for B vitamins revealed a notable peculiarity of the wCle genome. Although all known insect-associated Wolbachia genomes commonly possess a complete pathway for riboflavin (vitamin B2) and partial pathways for pyridoxine (vitamin B6) and folate (vitamin B9), the wCle genome additionally contained a complete pathway for biotin (vitamin B7) and a partial pathway for thiamine (vitamin B1) (Fig. 2 A and B). The biotin synthesis genes bioC, bioH, bioF, bioA, bioD, and bioB formed a compact operon on the wCle genome. Almost the same operon structure was identified on the genome of a facultative endosymbiont Cardinium hertigii (Bacteroidetes) causing cytoplasmic incompatibility in the parasitoid wasp Encarsia pergandiella (10), on the genome of a swine pathogen Lawsonia intracellularis (Deltaproteobacteria) (11), and also on the plasmid of a Rickettsia strain (Alphaproteobacteria) isolated from the tick Ixodes scapularis (12), but not on the genomes of Wolbachia and allied alphaproteobacteria (Fig. 3A). Molecular phylogenetic analyses of these biotin synthesis genes consistently exhibited similar evolutionary patterns: closely allied to the corresponding genes of Cardinium, L. intracellularis, and the Rickettsia plasmid and also related to the corresponding genes from diverse bacterial lineages representing Alphaproteobacteria, Deltaproteobacteria, Gammaproteobacteria, Cyanobacteria, Chlamydiae, and others (Fig. S1 A–F). These patterns suggest that the biotin synthesis genes were acquired as a whole operon by an ancestor of wCle from an unrelated bacterium, which was presumably a facultative endosymbiont (likely either Cardinium or Rickettsia) coinfecting the same insect host.

In addition, three genes involved in a thiamine salvage pathway, tenA1, thiD, and ψthiM (wherein the last one is a pseudogene), formed a compact operon on the wCle genome. Similar operon configuration was identified on the genome a fish francisellosis pathogen Francisella noatunensis (Gammaproteobacteria) (13), but not on the genomes of Wolbachia and allied alphaproteobacteria (Fig. 3B). Molecular phylogenetic analyses of these genes consistently exhibited similar evolutionary patterns: closely allied to the corresponding genes of F. noatunensis, Brachyspira hyodysenteriae, and Legionella drancourtii and also related to the corresponding genes from diverse bacterial lineages representing Gammaproteobacteria, Bacteroidetes, Spirochaetes, and others (Fig. S2 A–C). These patterns suggest that the thiamine synthesis genes tenA1, thiD, and ψthiM were acquired as a partial operon by an ancestor of wCle from an unrelated bacterium. Here it should be noted that, although tenA1 is specific to wCle, its paralogs tenA2 and tenA3 are present in all of the insect-associated Wolbachia strains and likely represent tenA gene copies authentic to the Wolbachia lineage (Fig. S2A). It is also notable that ψtenA3 has been pseudogenized specifically in wCle (Fig. 2A, Fig. S2A, and Table S3), highlighting complicated evolutionary trajectories of the tenA gene families among the Wolbachia strains.

From all these results taken together, it is conceivable, although speculative, that the synthetic pathways for biotin, thiamine, and other B vitamins have been moving across diverse bacterial lineages in a dynamic manner (14), presumably as evolutionarily cohesive functional modules or selfish operons, like antibiotic resistance genes and restriction modification genes, that can confer immediate functional advantage to the recipient organisms (15, 16).

By rearing newborn nymphs of the wCle-infected bedbug strain JESC on normal rabbit blood and rifampicin-supplemented rabbit blood, we generated symbiotic insects and symbiont-deficient insects of the same genetic background. By this antibiotic treatment, wCle in these insects was killed, although remnant Wolbachia DNA tended to be detectable by PCR. These insects were able to grow to adults but suffered significantly impaired growth rate, body size, and fertility (9). When these insects were subjected to extraction and quantification of B vitamins at the fourth instar, the symbiont-deficient insects exhibited significantly lower titers of some B vitamins, including biotin and riboflavin, but not thiamine and pyridoxine, in comparison with the symbiotic insects (Fig. 4 A–D). These results indicate that wCle is capable of provisioning biotin and riboflavin, but not thiamine and pyridoxine, to the host bedbug, which is concordant with the wCle genome data (Fig. 2).

Symbiont-mediated provisioning of riboflavin has been experimentally demonstrated or suggested in aphid–Buchnera, tsetse–Wigglesworthia, louse–Riesia, and other insect–bacterium symbiotic associations (17–19), and the complete synthetic pathway for riboflavin is consistently retained in these endosymbiont genomes (3, 4, 20) and also in most of the Wolbachia genomes determined to date (Fig. 2B). Hence, it is suggested that riboflavin provisioning by these symbionts, including Wolbachia, generally has biological roles in the symbiotic associations. Here, our focal interest is whether the wCle-specific Wolbachia’s capability of biotin provisioning is biologically meaningful for the host bedbug. By rifampicin treatment, we established a wCle-cured bedbug strain JESC^−wCle from the wCle-infected bedbug strain JESC (9). By feeding with rabbit blood supplemented with all B vitamins, the wCle-cured insects restored normal growth and reproduction (9), whereby we were able to maintain the wCle-cured bedbug strain continuously. When biotin was selectively omitted from the B vitamin-supplemented blood meal, the wCle-cured insects exhibited significantly reduced adult emergence rates in comparison with the wCle-cured insects reared on the blood meal supplemented with all B vitamins (Fig. 4E), confirming that the biotin synthetic pathway of wCle plays an important role for the host bedbug. By contrast, when thiamine was selectively omitted from the B vitamin-supplemented blood meal, no significant fitness decline was observed in the wCle-cured insects (Fig. 4F), which probably reflects the incomplete synthetic pathway for thiamine in the wCle genome (Fig. 2A) and consequent absence of the symbiont-derived thiamine supply (Fig. 4D).

These results strongly suggest that the wCle-specific biotin operon, which was presumably acquired via lateral gene transfer from a coinfecting endosymbiont, pivotally underpins the Wolbachia–bedbug nutritional mutualism. Here, our focal interest is the origin of the biotin synthesis genes in the Wolbachia evolution. We surveyed all complete and draft genomes of insect-associated Wolbachia strains available in the DNA databases (SI Materials and Methods), but no biotin synthesis genes were detected. Meanwhile, when we surveyed all genomic data of nematode-associated Wolbachia strains (SI Materials and Methods), a degenerate biotin operon, in which all biotin synthesis genes had been pseudogenized, was identified in the complete genome of wOo from Onchocerca ochengi (HE660029) (21) and the draft Wolbachia genome from O. volvulus (ASM33837v1) (22). Although the biotin operon in the wCle genome was 5.4 kb in size and encoding six intact genes, the biotin operon in the wOo genome was 4.1 kb in size, wherein all six genes were disrupted by a number of stop codons, frame shifts, and deletions (Fig. 3A and Table S3).

Notably, the biotin synthesis pseudogenes ψbioC, ψbioH, ψbioF, ψbioA, ψbioD, and ψbioB on the wOo genome (Fig. 3A) were closely allied to the corresponding genes on the wCle genome (Fig. S3 A–F). Phylogenetic analysis based on 52 ribosomal protein sequences showed that wCle and wOo form a well-supported clade in the Wolbachia phylogeny (Fig. 5), although wCle belonging to the Wolbachia F supergroup and wOo representing the Wolbachia C supergroup diverged early in the Wolbachia diversification (23). These patterns are in favor of the hypothesis that the biotin operon was acquired by the common ancestor of wCle and wOo via lateral gene transfer, which has subsequently been retained in the wCle lineage but disrupted in the wOo lineage (Fig. S4A). On the other hand, the flanking regions of the biotin operon were not conserved between the wCle genome and the wOo genome (Fig. 6A), and the syntenic patterns between the wCle genome and the wOo genome did not favor the notion that the biotin operon of wCle is located at the homologous genomic region of wOo (Fig. 6B). Considering the wandering nature of the biotin operon in the bacterial evolution (Figs. S1 and S3), these patterns may favor the hypothesis that the lineage of wCle and the lineage of wOo acquired the biotin operon from the common bacterial source independently (Fig. S4B). The common bacterial source seems likely a coinfecting endosymbiont lineage on account of the frequent coinfection of Wolbachia with other endosymbionts including Cardinium and Rickettsia (24) and the presence of a closely related biotin operon in the Cardinium genome and the Rickettsia plasmid (10, 11). Meanwhile, it should be noted that the rampant intra- and intergenomic recombinations commonly observed in Wolbachia genomes (25) may account for the apparent divergence of the flanking regions. Fig. S4 C and D depicts, although seemingly less likely, representative alternative hypotheses on the evolutionary origin of the biotin operon in Wolbachia. To address these evolutionary issues conclusively, it is necessary to examine more Wolbachia genomes, in particular those representing the C, D, and F supergroups.

Previous antibiotic curing experiments showed that the Wolbachia strains wBm, wOo, and wCle are essential for their hosts B. malayi, O. ochengi, and C. lectularius, respectively (9, 26, 27). Although wCle was shown to provide B vitamins to its bedbug host (Fig. 4) (9), it has been elusive what mechanisms underlie the essentiality of wBm and wOo for their nematode hosts (21, 28). Most of the synthetic pathways for B vitamins are either absent or eroded in the wBm and wOo genomes (Fig. 2B and Table S3), indicating that the major biological role of wBm and wOo cannot be provisioning of B vitamins. Although genome sizes of wBm (1.08 Mb) and wOo (0.96 Mb) are remarkably smaller than those of insect-associated facultative Wolbachia strains (1.27–1.48 Mb), such a genome reduction is not observed with the wCle genome (1.25 Mb; Table S1). Taken together, the bedbug–wCle mutualism seems to have evolved independently of the B. malayi–wBm and O. ochengi–wOo mutualisms.

In addition to the bedbug strain JESC mainly used in this study, three additional bedbug strains TUA (from Tokyo, Japan), TIH (from Toyama, Japan), and SYDL (from Sydney, Australia), and also an allied cimicid species, the Japanese bat bug Cimex japonicus (from Hokkaido, Japan), were subjected to cloning and sequencing of the insertion site of the biotin operon on the Wolbachia genome. The 7,724-bp region on the wCle genome, which encodes six biotin synthesis genes franked by an ankyrin repeat gene and a purK gene, exhibited 100% (7,724/7,724) nucleotide sequence identity among the Japanese and Australian populations of C. lectularius (Fig. S5A). The biotin operon of the same structure was identified at the same location of the Wolbachia genome associated with C. japonicus, which was 7,721 bp in size and exhibited 98.1% (7,576/7,724 including indels) nucleotide sequence identity to that of wCle (Fig. S5B). These results indicate that the biotin operon laterally transferred to the Wolbachia genome was already present in the common ancestor of C. lectularius and C. japonicus and has stably been maintained in natural populations of C. lectularius. In C. japonicus, notably, frame shift mutations were identified in bioC and bioH genes on the inserted biotin operon (Fig. S5B). The partial erosion of the biotin synthesis genes may be, although speculative, relevant to biological, ecological, and evolutionary differences between the bedbug and the bat bug, such as different nutritional compositions of human blood and bat blood and different facultative/gut microbiotae associated with the bedbug and the bat bug.

To gain further insights into the evolution of the bedbug-Wolbachia mutualism, we analyzed the phylogenetic placement of wCle within the Wolbachia F supergroup on the basis of 16S rRNA gene sequences available in the DNA databases. In the Wolbachia F clade, wCle was the most closely related to the Wolbachia strain associated with the Japanese bat bug C. japonicus and also clustered with Wolbachia strains associated with mite, bat fly, louse fly, grasshopper, ant, and termites (Fig. S6). Some F Wolbachia strains are obligatorily associated with Mansonella spp. and other filarial nematodes (29–31). However, they were placed in distinct lineages from the arthropod-associated F Wolbachia strains including wCle, although statistical supports for the groupings were not necessarily significant (Fig. S6). In the bat fly, the louse fly, the grasshopper, and the termites, the F Wolbachia infections are not fixed in the host populations, suggesting facultative nature of these host–symbiont associations (32–35). These observations suggest that (i) the bedbug–wCle nutritional mutualism evolved independently of the nematode–Wolbachia mutualism within the Wolbachia F supergroup, (ii) the origin of the bedbug–wCle mutualistic association is more recent than the origin of the nematode–Wolbachia mutualistic association, (iii) plausibly, an ancestor of the cimicid bugs acquired wCle from an unrelated arthropod host, (iv) in the donor arthropod, wCle was likely a facultative endosymbiotic associate, and (v) the establishment of the bedbug–wCle association presumably entailed an evolutionary transition from facultative symbiosis to obligate nutritional mutualism for the symbiont side. For the host side, it is elusive whether the acquisition of wCle entailed establishment of a novel nutritional mutualism or replacement of a preexisting nutritional symbiont. In other words, wCle may have been acquired by the common ancestor of cimicid bugs or acquired later in a cimicid lineage via symbiont replacement. To address this evolutionary issue, a comprehensive endosymbiont survey is needed for the family Cimicidae, which embraces more than 22 genera and 74 species of blood-sucking bugs in the world (36).

With all these results taken together, we strongly suggest that acquisition of a single gene cluster consisting of biotin synthesis genes underlies the evolution of the bedbug–Wolbachia nutritional mutualism. This finding provides an impressive case of evolutionary transition from facultative symbiosis to obligate mutualism that was facilitated by lateral gene transfer in an endosymbiont lineage. Here it should be noted that two lateral transfer events at different levels are involved in the evolution of the bedbug–Wolbachia nutritional mutualism: (i) acquisition of the biotin operon by an ancestor of wCle via lateral gene transfer and (ii) acquisition of the biotin operon-bearing Wolbachia strain by an ancestor of the bedbug. Plausibly, the biotin synthetic capability of wCle played no major role in the original nonbedbug host, but, once acquired by the bedbug, it conferred significant fitness advantage to the host that feeds solely on vertebrate blood deficient in biotin and other B vitamins. In this context, it is conceivable, although speculative, that the wCle-allied facultative Wolbachia strains associated with the bat fly and the louse fly (Fig. S6) may have some auxiliary nutritional roles in the blood-sucking insect hosts.

In this study, we elucidated the genomic basis of the exceptional Wolbachia-mediated biotin provisioning in the bedbug, wherein lateral gene transfer underpins the peculiar metabolic capacity of wCle. Notably, another peculiarity of wCle resides in its cellular tropism. Although most of insect-associated facultative Wolbachia strains are sparsely distributed in various host cells and tissues, wCle densely and specifically populates bacteriocytes of the host bedbug (9). Here, the evolutionary transition from facultative symbiosis to obligate mutualism might have occurred in parallel with the transition from systemic infection to bacteriocyte localization, which are consistently directing toward a higher level of host–symbiont integrity. Molecular, cellular, and genetic mechanisms underlying the wCle-specific bacteriocyte localization are, although currently unknown, of evolutionary interest and deserving future studies.

Recent studies have revealed that lateral gene transfers sometimes entail evolutionary consequences to adaptive ecological and physiological traits in insects and other organisms (37–40). Wolbachia endosymbionts occur ubiquitously (7) and rampantly exchange their genetic materials with their cosymbionts and hosts (25, 41). Our results suggest that the biotin synthetic operon of wCle was acquired from a facultative endosymbiont (likely Cardinium or Rickettsia) coinfecting the same insect host. It was recently reported that in the mealybug Planococcus citri, strikingly, the prokaryotic biotin synthesis genes bioA, bioD, and bioB, which are phylogenetically close to those found in wCle, Cardinium, and Rickettsia (Figs. S1 D–F and S3 D–F), are encoded in the host nuclear genome, significantly expressed in the bacteriome, and presumably functioning in the endosymbiotic system (42). These findings highlight the evolutionary importance of endosymbiotic associations, wherein a host genome and multiple symbiont genomes are continuously integrated into a coherent system, as arenas for symbiont–symbiont and symbiont–host exchanges of genetic materials, which potentially lead to biological novelties and innovations including capability of synthesizing essential nutrients. In this context, the origin of Wolbachia–bedbug nutritional mutualism compiles an additional dimension to the dynamic evolutionary perspective of symbiosis.

The wCle-infected bedbug strain JESC was maintained in plastic Petri dishes with a piece of filter paper at 25 °C under constant darkness. The insects were fed with purchased rabbit blood (Kohjin Bio) warmed at 37 °C once per week using an artificial membrane feeding system as previously described (9). For curing of wCle, rifampicin was added to the blood meal at a concentration of 10 μg/mL. For rearing the wCle-cured insects, B vitamins were supplemented to the blood meal as previously described (9).

For wCle genome sequencing, we dissected 26 healthy young adult bedbugs, collected 51 bacteriomes, and extracted DNA from the symbiotic organs. The DNA sample (2 μg) was subjected to whole-genome shotgun sequencing and assembly as previously described (4). Gene prediction and annotation were performed as previously described (43).

Molecular phylogenetic analyses were conducted essentially as previously described (43). Multiple alignment was performed using MAFFT 5.6 (44). Maximum likelihood and Bayesian phylogenies were constructed using RAxML Version 7.7.6 (45) and MrBayes 3.1.2 (46).

Newborn nymphs of the wCle-infected bedbug strain JESC were allocated to two experimental groups: one reared on normal rabbit blood and the other kept on the rabbit blood supplemented with 10 μg/mL rifampicin. Fourth-instar nymphs, which were collected 10 d after the last feeding at the third instar and weighed individually, were homogenized and hydrolyzed in 0.1 N HCl at 100 °C for 30 min. After adjusting to pH 4.5 with 2.5 M sodium acetate, Takadiastase (Sigma-Aldrich) was added to each sample and incubated at 37 °C for 16 h. After adding methanol for protein precipitation, the supernatant was lyophilized, suspended in 0.05% formic acid, 1.25 mM ammonium formate, and 50% methanol, and purified using a polymer-based column (GL-Tip SDB; GL Sciences). The purified sample was lyophilized and quantitatively analyzed using a high-performance LC system (Prominence; Shimadzu) coupled with a mass spectrometer (LCQ Duo; Thermo Fisher Scientific).

First-instar nymphs of the wCle-cured bedbug strain, whose parents had been maintained on rabbit blood supplemented with all B vitamins, were reared to adulthood on nonsupplemented rabbit blood to eliminate transgenerational carryover of B vitamins. These adult insects were allowed to mate and lay eggs, and nymphs from these eggs were randomly allocated to the following experimental groups: (i) all 10 B vitamins were added to the blood meal, (ii) all B vitamins except biotin or thiamine were added to the blood meal, and (iii) no B vitamins were added to the blood meal. These insects were fed once a week and monitored until all of the insects either became adult or died.

See SI Materials and Methods for complete details on the materials and methods.