|Home | About | Journals | Submit | Contact Us | Français|
Bacteria in the genus Rickettsia are intracellular symbionts of disparate groups of organisms. Some Rickettsia strains infect vertebrate animals and plants, where they cause diseases, but most strains are vertically inherited symbionts of invertebrates. In insects Rickettsia symbionts are known to have diverse effects on hosts ranging from influencing host fitness to manipulating reproduction. Here we provide evidence that a Rickettsia symbiont causes thelytokous parthenogenesis (in which mothers produce only daughters from unfertilized eggs) in a parasitoid wasp, Pnigalio soemius (Hymenoptera: Eulophidae). Feeding antibiotics to thelytokous female wasps resulted in production of progeny that were almost all males. Cloning and sequencing of a fragment of the 16S rRNA gene amplified with universal primers, diagnostic PCR screening of symbiont lineages associated with manipulation of reproduction, and fluorescence in situ hybridization (FISH) revealed that Rickettsia is always associated with thelytokous P. soemius and that no other bacteria that manipulate reproduction are present. Molecular analyses and FISH showed that Rickettsia is distributed in the reproductive tissues and is transovarially transmitted from mothers to offspring. Comparison of antibiotic-treated females and untreated females showed that infection had no cost. Phylogenetic analyses of 16S rRNA and gltA gene sequences placed the symbiont of P. soemius in the bellii group and indicated that there have been two separate origins of the parthenogenesis-inducing phenotype in the genus Rickettsia. A possible route for evolution of induction of parthenogenesis in the two distantly related Rickettsia lineages is discussed.
The genus Rickettsia contains a group of obligate intracellular symbionts of eukaryotic cells and belongs to the family Rickettsiaceae in the order Rickettsiales of the Alphaproteobacteria (58, 90). Many species have medical importance as they are pathogens of humans and other vertebrates; pathogenic Rickettsia species infect their hosts through blood-feeding arthropods, including lice, fleas, ticks, and mites (51, 80). In addition to Rickettsia species that cause infectious diseases in vertebrates, symbiotic species have been found in disparate groups of organisms, including arthropods, annelids, amoebae, hydrozoa, and plants (53). Rickettsia appears to be especially common in arthropods, having been found in a wide range of taxa in the classes Entognatha (springtails), Insecta (booklice, lice, bugs, leafhoppers, aphids, whiteflies, fleas, flies, lacewings, moths, beetles, and wasps), and Acarina (ticks and mites) (86). However, in most cases, the effect of Rickettsia on the invertebrate host has not been established yet. In general, Rickettsia bacteria are facultative symbionts, but in the booklouse Liposcelis bostrychophila the association is strictly obligate and Rickettsia has an essential role in oocyte development (54, 92). Facultative symbiotic Rickettsia strains have been reported to negatively affect some aspects of host fitness, causing reductions in body weight, fecundity, and longevity in the pea aphid (16, 60, 64), reductions in viability in some blood-feeding arthropod vectors (5, 46), and increased susceptibility to insecticides in the sweet potato whitefly (41). There is also evidence that Rickettsia has positive effects on host fitness, such as a larger body size in infected leeches (40) and a possible role in the oogenesis of a bark beetle (93). Finally, facultative symbiotic rickettsiae can be reproductive parasites of insects. Rickettsia strains are the causal agents of male killing (infected male embryos die) in some ladybird (79, 88) and buprestid leaf-mining (42) beetles. They are also the cause of thelytokous parthenogenesis (in which mothers produce only daughters from unfertilized eggs) in a parasitoid wasp (32). Both kinds of reproductive manipulation bias the host sex ratio toward females and favor the spread of the transovarially inherited Rickettsia strains in the infected populations. In general, Rickettsia is transmitted primarily vertically to host progeny, but in pathogenic species there is concomitant horizontal transmission via intermediate vertebrate hosts, which plays an important role in maintaining the infection in populations of blood-feeding arthropods (53, 57). An exception is Rickettsia prowazekii, the epidemic typhus agent, which spreads only via horizontal transmission in louse host populations (5). Only one Rickettsia is known to be a plant pathogen, and leafhoppers transfer this pathogen horizontally between plants (20). The fact that Rickettsia can be transmitted horizontally and then perpetuated vertically in host descendants has probably been one of the most important factors determining the enormous diversity of Rickettsia symbiotic associations. This point has been emphasized by analyses that have revealed considerable incongruence between Rickettsia and host phylogenies, indicating that horizontal transfer has occurred multiple times over evolutionary timescales (53, 54, 86).
In addition to Rickettsia, diverse heritable bacteria are known to manipulate host reproduction to enhance their transmission in arthropods (12, 23). Wolbachia (order Rickettsiales, family Anaplasmataceae), a close relative of Rickettsia (90), is able to induce all known forms of manipulation of reproduction, including cytoplasmic incompatibility, feminization of genetic males, male killing, and parthenogenesis (68). Previously, only Cardinium (Sphingobacteria) has been shown to cause a similar range of reproductive phenotypes, except for male killing (35). The emerging diversity of Rickettsia associated with arthropods (53, 86), combined with evidence that it can manipulate host reproduction in more than one way, suggests that this symbiont may also be a master manipulator.
In the Hymenoptera, the dominant mode of reproduction is arrhenotoky; that is, diploid females develop from fertilized eggs, and haploid males develop from unfertilized eggs (76). However, thelytokous parthenogenesis is common, and in some lineages, like the superfamilies Chalcidoidea and Cynipoidea, it is strongly associated with Wolbachia or Cardinium infection (33, 35). Parthenogenesis-inducing (PI) bacteria cause restoration of diploidy in unfertilized haploid eggs, which results in female offspring (28, 50, 69). PI Wolbachia and PI Cardinium also occur in other groups of haplodiploid arthropods, such as mites (82), scale insects (56), and thrips (4). Previously, the only example of PI caused by Rickettsia was found in the eulophid parasitoid wasp Neochrysocharis formosa (1, 32). Besides PI bacteria, uniparental (thelytokous) reproduction in haplodiploid arthropods can also be caused by feminizing bacteria that are able to interact with the host sex determination system and force the development of genotypic males toward functional phenotypic females. To date, only Cardinium has been reported to be a causal agent of feminization in haplodiploid arthropods, and only two examples are known: a mite in which Cardinium causes haploid male embryos to develop as haploid females (18, 83) and a parasitoid wasp in which diploid males are converted to females (27).
In this paper, thelytokous reproduction in a parasitoid wasp, Pnigalio soemius (Hymenoptera: Eulophidae), was studied. This wasp, which is probably a complex of cryptic species (8), is a solitary ectoparasitoid that attacks larvae of many leafminer insect species in the orders Coleoptera, Diptera, Hymenoptera, and Lepidoptera (48), some of which are pests of agricultural crops (37, 61). Female P. soemius wasps paralyze host larvae by injection of venom and subsequently lay a single egg next to the host inside a leaf mine; then the parasitoid larva eats the killed host (7). Commonly, P. soemius reproduces biparentally, and the occurrence of thelytoky has not been reported previously. The aims of this study were to determine whether symbiotic bacteria are involved in manipulating the reproduction of P. soemius and then to determine the taxonomic affiliation and phenotype of the manipulators of reproduction discovered. By using antibiotic treatments and karyological analysis of the insect studied, molecular and phylogenetic characterization of the symbiotic bacteria, and detection of intracellular symbionts by means of fluorescence in situ hybridization, it was demonstrated that a PI Rickettsia causes thelytokous reproduction in P. soemius.
A thelytokous population of P. soemius was obtained from larvae of the leafminer Trypeta artemisiae (Diptera: Tephritidae) collected in the field from plants of the common wormwood (Artemisia vulgaris) in Usseaux, Torino, Italy. A culture of the parasitoid was generated using a single female. As it was difficult to rear the natural host in the laboratory, an alternative host, the leafminer Cosmopterix pulchrimella (Lepidoptera: Cosmopterigidae), was used and maintained on wall pellitory (Parietaria diffusa) plants as described previously (9). Due to high parasitoid mortality during larval development and to rapid degradation of the host larvae stung by the female wasp, it was difficult to establish a permanent colony of thelytokous P. soemius by allowing wasp larvae to develop naturally in leaf mines with single host larvae. Consequently, an artificial rearing system (8) was used, in which several mature host larvae were provided to a single parasitoid during development. Although more time-consuming, this system allowed us to significantly increase production of offspring. Briefly, inside leaf mines mature host larvae were exposed to oviposition by single parasitoids for 24 h, after which the mines were dissected and eggs were singly transferred on a glass slide kept in a humid petri dish. A new host larva, killed by freezing it at −20°C for 3 min (killing was necessary as the parasitoid larvae can eat only immobilized hosts), was added near each egg on the slide. Freshly killed larvae were added daily until the parasitoid pupated. Insects were reared at 25 ± 1°C with a relative humidity of 60% ± 10% using a photoperiod consisting of 12 h of light and 12 h of darkness. To date, after 20 generations of laboratory rearing, male progeny have never been obtained.
In order to determine if bacterial symbionts were involved in parthenogenetic reproduction in P. soemius, the sex ratio of the progeny produced by antibiotic-treated adult females was assessed. Ten newly emerged adult females were individually fed honey containing rifampin (20 mg/ml) for 24 h in a glass vial and then allowed to oviposit for their entire life span as described above. After the initial 24-h treatment with the antibiotic, adults were fed daily using honey streaks until they died. The effects of antibiotic treatment on fecundity and fertility were also examined in order to assess if Rickettsia influences host fitness. The following characteristics were determined for treated and untreated (fed only honey) females: number of eggs laid, percentage of unhatched eggs, percentage of mortality for the progeny during development from larvae to pupae, and number and sex of the adult progeny. To test for antibiotic toxicity, the longevity of the treated females was compared to that of the control females. The experimental conditions were 25 ± 1°C, a relative humidity of 60% ± 10%, and a photoperiod consisting of 12 h of light and 12 h of darkness. Biological data that satisfied conditions of normality and homoscedasticity, both untransformed and after appropriate transformation, were subjected to a one-way analysis of variance (ANOVA). When the assumptions of the ANOVA were violated and could not be met by data transformation, the nonparametric Kruskal-Wallis test was used after having controlled for the data distribution to have the same shape.
Soon after collection all specimens were surface sterilized in 70% ethanol and rinsed five times in sterile water before they were stored in 70% ethanol at −20°C. DNA was extracted from two samples: a pool of five adult P. soemius females and a pool of 15 eggs laid by different females and collected soon after oviposition. For DNA isolation, the protocol of Walsh et al. (81) was used, with a few modifications. Briefly, the two samples were homogenized in DNA extraction buffer consisting of 150 μl Chelex and 25 μl proteinase K (20 mg/ml) for the adult sample and 75 μl Chelex and 15 μl proteinase K (20 mg/ml) for the egg sample. Each sample was incubated for 1 h at 55°C and for 8 min at 100°C and centrifuged at 13,000 rpm for 3 min.
To detect bacterial symbionts, the 16S rRNA gene was amplified using universal primers 27F (5′-AGAGTTTGATCMTGGCTCAG-3′) and 1513R (5′-ACGGYTACCTTGTTACGACTT-3′) (87), cloned, and sequenced. PCRs were performed with 40-μl mixtures containing 1× buffer (Promega), each deoxynucleoside triphosphate at a concentration of 0.2 mM, 3 U of GoTaq DNA polymerase (Promega), and each primer at a concentration of 300 nM. After an initial denaturation for 3 min at 96°C, the temperature profile consisted of 94°C for 30 s, 52°C for 1 min, and 72°C for 2 min for 35 cycles, followed by a final extension for 7 min at 72°C. Five microliters of each reaction mixture was checked on a 1% agarose gel, and samples that yielded amplicons of the expected size (~1,500 bp) were precipitated, ligated into the pGEM-T Easy plasmid vector (Promega), and cloned into Escherichia coli TOP10 competent cells (Invitrogen) according to the manufacturer's instructions. Transformants were screened by performing PCR with universal M13 vector primers, and inserts of the expected size (~1,800 bp) were amplified by nested PCR with universal bacterial primers 341f (5′-CCTACGGGAGGCAGCAG-3′) and 907r (5′-CCGTCAATTCMTTTGAGTTT-3′) (45). The PCR thermal profile was initial denaturation for 3 min at 96°C, 30 cycles of 94°C for 45 s, 52°C for 50 s, and 72°C for 50 s, and then final extension for 5 min at 72°C. All 500- to 600-bp amplicons were directly sequenced with the 341f primer. A 154-bp fragment was obtained, and it included the hypervariable region V3, which has been proven to be useful for studying the composition of bacterial communities (15, 99). As almost all sequences (42 of 44 clones) corresponded to Rickettsia sequences and the other two were sequences of bacteria not associated with manipulation of reproduction, 15 clones were chosen from the clones identified as Rickettsia (6 clones from the adult sample and 9 clones from the egg sample) and were sequenced with the 27F and 1513R primers in order to characterize the Rickettsia strain. Furthermore, a fragment of the citrate synthase gltA gene was sequenced. Rickettsia-specific primers CS409d and CS1273r (59) were used to amplify 864 bp of the gltA gene from the egg sample. Two more sequences were obtained from the adult female sample by using primers CS409d (59) and RicCS-AR (60), which amplified a 600-bp fragment. The PCR conditions used for the gltA gene were the same as those described above for amplification of the 1,500-bp fragment of the 16S rRNA gene, except that the annealing temperature was 48°C; the amplicons obtained were directly sequenced.
Sequencing was performed with an ABI Prism Big Dye terminator cycle sequencing kit and an ABI Prism 310 genetic analyzer (Applied Biosystems, Foster City, CA) at the Dipartimento di Biologia Vegetale, University of Napoli “Federico II,” Italy, sequencing facility. Sequences were assembled using SeqMan in the Lasergene software package (DNASTAR, Madison, WI) and were compared with known sequences in the GenBank database by using BLAST searches (www.ncbi.nlm.nih.gov/blast) (3).
Finally, the possibility that the bacterial symbionts known to cause thelytokous reproduction in arthropods, including Cardinium, Wolbachia, and Rickettsia, were present was checked by performing diagnostic PCR both for the samples used for sequencing and for 10 single P. soemius adult females. The following sets of primers were used: primers CLOf and CLOr1 (85) and primers ChF and ChR (96) for the Cardinium 16S rRNA gene, primers ftsZf1 and ftsZr1 (89) for the Wolbachia cell cycle ftsZ gene, primers 16Swolb76-99f and 16Swolb1012-994r for the Wolbachia 16S rRNA gene (49), and primers Rb-F and Rb-R (29) for the Rickettsia 16S rRNA gene. As only Rickettsia was found in P. soemius females, the infection status of 10 adult males produced by antibiotic-treated females was checked using only primers Rb-F and Rb-R. To confirm that negative results in PCR analyses were not due to poor-quality extracts, PCRs were performed for all samples with primers ITS2F and ITS2R (14). These primers amplify ~500 bp of the internal transcribed spacer 2 region of Hymenoptera. Therefore, only positive samples were considered to assess the infection status.
DNA sequences of the 16S rRNA and gltA genes (1,388 bp and 721 bp, respectively) were aligned with Rickettsia sequences available in the GenBank database. Multiple alignments were constructed with the ClustalW method of MegAlign in the Lasergene software package (DNASTAR, Madison WI). Phylogenies were reconstructed using maximum parsimony (MP) and maximum likelihood (ML) methods as implemented in PAUP 4.0b10 (71). The Rickettsia symbiont of the ciliate Diophrys appendiculata (75) was chosen as the outgroup for the analysis based on the 16S rRNA gene. A Wolbachia strain was chosen as the outgroup for the analysis based on the gltA gene. In the MP analysis, heuristic searches with tree bisection reconnection (TBR) branch swapping, the random addition sequence option, and Maxtrees set to increase without limits were performed. The evolutionary models used in the ML analysis were provided by MODELTEST 3.7 (55). Based on both hierarchical likelihood ratio test (hLRT) and Akaike (AIC) criteria, the models chosen were K81 for the 16S rRNA gene sequences and K81uf+G (Kimura three-parameter model with unequal base frequencies and an estimated gamma shape parameter) for the gltA sequences. All sites were weighted equally. Bootstrap support was evaluated by using 1,000 replicates in the MP analysis and 100 replicates in the ML analysis.
Localization of Rickettsia in the host's reproductive tissues was studied using fluorescence in situ hybridization (FISH). The Rickettsia probe RickPn-Cy3 (′5-Cy3-TCCACGTCGCCGTATTGC-3′) was designed using the 16S rRNA sequences of the P. soemius symbiont. Probe specificity was checked using the Ribosomal Database Project “probe match” analysis tool (http://rdp.cme.msu.edu). Eggs were collected shortly after the oviposition. Ovaries were extracted from adult females in a drop of phosphate-buffered saline (PBS) using a stereomicroscope. Eggs and ovaries were subjected to analysis by the whole-mount FISH method described by Sakurai et al. (60), with slight modifications. Eggs were dechorionated in 50% commercial bleach in PBS for 15 min and then washed in PBS to remove the bleach. Samples were fixed overnight in a 4% paraformaldehyde neutral buffered solution, decolorized in 6% H2O2 in ethanol for 2 h, and hybridized overnight. The hybridization buffer (20 mM Tris-HCl [pH 8.0], 0.9 M NaCl, 0.01% SDS, 30% formamide) contained 10 pmol/ml of fluorescent probe. Stained samples were observed with a Zeiss Axiophot 2 epifluorescence microscope. The specificity of the signals observed was verified using the following controls: no-probe control, RNase-digested control, and Rickettsia-free samples from an uninfected bisexual strain of P. soemius. Nuclei of the host cells were counterstained with 4′,6′-diamidino-2-phenylindole (DAPI) (0.4 μg/ml) in mounting medium. Double FISH was performed with eggs of thelytokous P. soemius using RickPn-Cy3 and the universal probe EUB338-Fluorescein (5′-fluorescein-GCTGCCTCCCGTAGGAGT-3′).
In order to determine if the Rickettsia phenotype resulted from induction of parthenogenesis or feminization, we analyzed chromosome sets of individual female and male wasps produced by untreated and antibiotic-treated adult females of P. soemius, respectively. If induction of parthenogenesis occurred, diploid females and haploid males should have been found, as PI bacteria cause restoration of diploidy in unfertilized haploid eggs (69). If feminization occurred, no difference in the level of ploidy between females and males should have been found (27, 83). Metaphase chromosomes were obtained from single 4-day-old larvae (10 larvae for each sex) processed using the scraping and air drying method described by Baldanza et al. (6), with the following modifications. Each larva was nicked between the head and the thorax in 1.0 ml 0.1% colchicine in Shen solution (0.9 g NaCl, 0.042 g KCl, and 0.025 g CaCl2 in 100 ml distilled water) and incubated in a 1.5-ml tube at room temperature for 30 min. After incubation, the colchicine solution containing the dissected larva was centrifuged at 1,300 rpm for 10 min. The supernatant was removed, 1.0 ml of a hypotonic solution (0.5% sodium citrate) was added, and after 20 min the tube was centrifuged again at 1,300 rpm for 10 min. The supernatant was removed, and 0.4 ml of fixative (glacial acetic acid-methanol, 1:3) was added. After 30 min of incubation at room temperature, the tissues were broken by sucking and pushing them repeatedly with a pipettor equipped with a 0.2-ml tip. The volume of fixative was adjusted to 1 ml, and the sample was pelleted by centrifugation at 1,300 rpm for 10 min. The supernatant was removed, 1.0 ml of fixative was added, and centrifugation at 1,300 rpm for 10 min was performed again. Eventually, the pellet, suspended in 30 μl fixative, was dropped on a slide, air dried, and stained with Giemsa stain (5% in phosphate buffer, pH 6.8) for 30 min. Chromosomes were classified as described by Levan et al. (43).
The sequences of the Rickettsia symbiont from the thelytokous population of P. soemius have been deposited in the GenBank database under accession numbers EU881494 to EU881508 (1,388-bp sequences) and GU121643 to GU121669 (154-bp sequences) for the 16S rRNA gene and GU559856 for the gltA gene.
Nine of 10 rifampin-treated P. soemius females produced only male offspring, and 97.5% of the offspring of treated females were males. Only one treated mother produced progeny consisting of both sexes; it produced 2 sons and 1 daughter, and the latter originated from the first oviposited egg. The single F1 female was fed only honey (no antibiotic treatment), and its progeny were 3 males. In contrast, untreated females produced only female offspring. The numbers of viable adult offspring produced by antibiotic-treated females (mean ± standard error [SE], 4.0 ± 1.11 offspring; n = 10) were significantly lower than the numbers of viable adult offspring produced by untreated females (7.7 ± 1.51 offspring; n = 10) (F = 4.97; df = 1; P = 0.039; ANOVA). There was not a significant difference (F = 1.89; df = 1; P = 0.185; ANOVA) in fecundity (number of eggs laid) between treated females (mean ± SE, 23.0 ± 4.33 eggs) and untreated females (30.7 ± 3.55 eggs) or in the mortality of the progeny that occurred during embryonic and larval development. Indeed, the percentages of unhatched eggs were 17.57% ± 2.37% (mean ± SE) for the eggs laid by treated females and 14.94% ± 2.59% for the eggs laid by untreated females (F = 0.56; df = 1; P = 0.46; ANOVA). The percentages of mortality for the progeny during development from larvae to pupae were 80.1% ± 2.41% (mean ± SE) for treated females and 68.5% ± 6.24% for untreated females (df = 1; P = 0.15; Kruskal-Wallis test). No negative effect of rifampin on the longevity of treated females (mean ± SE, 29.9 ± 2.37 days; n = 10) compared to untreated females (32.9 ± 3.42 days; n = 10) (F = 0.52; df = 1; P = 0.480; ANOVA) was detected.
Using PCR, approximately 1,500 bp of the 16S rRNA gene was amplified and cloned for a pool of adult females and a pool of eggs of the thelytokous P. soemius. Forty-four positive clones were obtained (20 clones from the adult sample and 24 clones from the egg sample) and amplified by nested PCR to sequence a shorter fragment that included the hypervariable V3 region. The 154-bp sequences obtained were checked against sequences available in the GenBank database, and 42 of the 44 clones showed the highest level of similarity to Rickettsia. The sequences of the two remaining clones, which were retrieved from the adult sample, showed the highest levels of similarity to Methylobacterium sp. and Serratia proteamaculans. There is no evidence that these bacteria are involved in any type of reproductive manipulation, and they have been reported to be environmental bacteria, bacteria associated with digestive organs, or pathogens of insects (30, 39, 73, 74). These two bacteria were considered organisms that are not involved in induction of P. soemius parthenogenesis and were not included in further analyses.
In order to characterize the Rickettsia obtained from P. soemius, 15 clones (6 clones from the adult sample and 9 clones from the egg sample) were completely sequenced. Each of the 1,388-bp sequences obtained was subjected to a BLAST search and showed the highest level of sequence similarity (99%) to Rickettsia bellii. The 15 sequences of the P. soemius symbiont exhibited high levels of similarity to one another (range, 99.4% to 100%; average, 99.8%). Also, a fragment of the gltA gene was amplified and directly sequenced. A 721-bp sequence was obtained from the egg sample, and two 500-bp sequences were obtained from the adult female sample. These sequences exhibited 100% nucleotide similarity to one another and exhibited the highest level of similarity (97%) to R. bellii.
Diagnostic PCR was used to screen the pool of adult females, the pool of eggs, and 10 single adult females for infection by known inducers of parthenogenesis of arthropods (Rickettsia, Wolbachia, and Cardinium). Only evidence of infection with Rickettsia was found. PCR analysis of adult males produced by antibiotic-treated females revealed that 9 of 10 insects were still infected by Rickettsia.
MP analysis of the 16S rRNA gene showed that the sequences of 15 Rickettsia clones infecting thelytokous P. soemius form a well-supported lineage in the clade which includes R. bellii and symbionts of non-blood-feeding arthropods (Fig. (Fig.1A).1A). MP analysis of gltA gene sequences produced substantially the same result, placing the P. soemius symbiont in the clade that includes R. bellii (Fig. (Fig.1B).1B). The tree topologies produced by ML analyses of 16S rRNA and gltA gene sequences (data not shown) did not differ consistently and were congruent where nodes were highly supported with trees produced by MP analyses.
The distribution of Rickettsia in the reproductive tissues of thelytokous P. soemius was studied using FISH analysis (Fig. (Fig.22 and and3).3). Inside the ovary, dense clusters of bacteria were observed in the germarium, in the germ line cells, in the nurse cells, and in the developing oocytes (Fig. (Fig.2).2). In the oocytes, bacteria were concentrated at the posterior pole during mid to late oogenesis (Fig. (Fig.2G),2G), but at the end of oocyte development bacteria were also distributed in the ooplasm (Fig. (Fig.2H).2H). Freshly laid eggs appeared to be heavily infected with Rickettsia bacteria that were distributed throughout the ooplasm, following a posterior-anterior gradient; bacteria were still highly concentrated at the posterior pole and sparse in the anterior area (Fig. 3A and B). During embryogenesis, at the syncytial stage Rickettsia bacteria localized at opposite poles of the mitotic spindles in dividing nuclei and were more concentrated in the posterior portion of the embryo, where they formed dense aggregates around nuclei of the pole cells, which are the germ line precursor cells (Fig. 3C and D). Negative controls (no-probe, RNase-digested, and Rickettsia-free sample controls) did not display signals, confirming the specificity of the signals detected (data not shown). Simultaneous probing with the P. soemius symbiont-specific probe RickPn-Cy3 and the universal bacterial probe EUB338-Fluorescein did not reveal bacteria other than Rickettsia (Fig. 4D to F). Eggs oviposited by antibiotic-treated female wasps appeared to be still infected by Rickettsia, but the bacterial density was considerably lower than that in eggs laid by untreated females (Fig. (Fig.4B4B).
In the metaphase plates examined (50 plates for each sex), females had a diploid complement consisting of 12 chromosomes, including five metacentric pairs and one acrocentric pair, while the male karyotype was a haploid complement consisting of 6 chromosomes (Fig. (Fig.55).
Bacteria in the genus Rickettsia are common intracellular symbionts of disparate groups of organisms, yet the effects of infection with Rickettsia on invertebrate hosts are not known for most interactions (53, 86). Here, a Rickettsia that causes thelytokous parthenogenesis in the parasitoid wasp P. soemius was discovered.
Feeding antibiotics to uniparental females resulted in production of almost entirely male progeny, and the single F1 female produced by treated females produced F2 males exclusively (without additional antibiotic treatment). These results are consistent with the hypothesis that the Rickettsia symbiont is involved in determining the wasp reproductive phenotype. The infection status of thelytokous females was investigated by cloning and sequencing a portion of the bacterial 16S rRNA gene amplified with universal primers and by diagnostic PCR screening using primers specific for Rickettsia, Wolbachia, and Cardinium. Rickettsia was always detected in P. soemius samples, whereas no evidence of other bacteria that manipulate reproduction was found. Sequencing of PCR amplicons recovered Rickettsia 16S rRNA and gltA gene sequences for adult females, as well as for parthenogenetic eggs. Furthermore, FISH revealed that this symbiont is distributed in the cells of the ovaries, as well as in eggs. FISH also showed that there was no bacterium other than Rickettsia in the eggs. We concluded that Rickettsia is the causal agent of thelytokous reproduction of P. soemius and is vertically transmitted from mothers to offspring.
Two kinds of bacterium-induced manipulation of reproduction leading to all female progeny and uniparental reproduction of infected host populations have been found previously in haplodiploid arthropods, namely, thelytokous parthenogenesis (development of unfertilized eggs into females occurs through restoration of diploidy) (33) and feminization (genotypic males develop as functional phenotypic females due to interaction of bacteria with their host's sex determination system). Two examples of feminization are known in haplodiploid arthropods, in which haploid male embryos (83) and diploid male embryos (27) develop as functional haploid and diploid females, respectively. Analysis of the ploidy level in P. soemius revealed that females are diploid and males induced by antibiotic treatment are haploid, suggesting that the Rickettsia phenotype reflects genuine induction of thelytokous parthenogenesis. Another Rickettsia has been shown to induce thelytokous reproduction in a different eulophid wasp, N. formosa (1), suggesting that Rickettsia may routinely utilize PI as a mechanism to spread in host populations.
Although feeding antibiotics to infected females induced production of progeny that were almost all male, 90% of the screened males produced by treated mothers were infected, as determined by PCR. The occurrence of Rickettsia in males was confirmed by FISH analysis of freshly laid eggs produced by antibiotic-treated females (putative male eggs); however, in these eggs the bacterial density appeared to be much lower than that in the eggs laid by untreated females. This result suggests that a threshold density of Rickettsia bacteria in eggs is required to trigger development of female embryos. Although there have not been specific studies of Rickettsia, the role of bacterial density in the manipulation of reproduction has been shown in a number of symbiotic associations between insects and cytoplasmic incompatibility-inducing Wolbachia (19, 38, 47, 65). Male killing (36) and induction of parthenogenesis (33, 94) also appear to be positively correlated with the Wolbachia titer.
The pattern of distribution of Rickettsia in the reproductive tissues of P. soemius was studied by using FISH analysis. In the ovary, Rickettsia bacteria are present in the germarium, in the germ line cells, in the nurse cells, and in the developing oocytes. This pattern has also been observed for Wolbachia in several insect hosts (22, 62), as well as for a distantly related lineage of bacteria that manipulate reproduction, Cardinium (44, 95, 97). The high concentration of Rickettsia in the ovary and eggs of P. soemius suggests that the efficiency of transmission of bacteria is high and that there is strong expression of the PI phenotype, as confirmed by infection of all thelytokous females and the production of only female progeny during 20 generations of lab rearing. The mechanism used by Rickettsia symbionts for transovarial transmission in their hosts is not known, but active microtubule-mediated transport and passive transport during cytoplasmic dumping into the oocyte from the nurse cells through the ring canals have been suggested for Wolbachia transmission (72). Although our observations do not elucidate the transport mechanism of the P. soemius symbiont, they suggest that the nurse cells have a role as a reservoir of bacteria that are delivered into the developing oocytes. This type of transport has been observed for symbiotic Rickettsia of Liposcelis bostrycophila (54). Finally, Rickettsia was found to be concentrated at the posterior pole of the oocyte during mid to late and late oogenesis, as well as at the posterior pole of the egg during early embryogenesis. The same pattern has been observed for Wolbachia in many insect host species (13, 22, 63, 70, 91, 98). As the posterior pole is the site where the germ cells form during embryogenesis, concentration of symbionts at this site has been thought to be a mechanism for increasing the probability that bacteria are integrated into the germ cells and then transmitted to host progeny (31, 63, 77).
It has been found that some Rickettsia strains have positive effects on host fitness (92), and in particular, these symbionts have been shown to play an essential role in oogenesis in the booklouse L. bostrycophila (54) and possibly also in the bark beetle Coccotrypes dactyliperda (93). In some cases, infections by reproductive parasites can have beneficial effects on their arthropod hosts. For example, Cardinium infection is associated with an increase in fecundity in a mite (84), while Wolbachia is necessary for oogenesis (22, 67) or for normal egg hatching (21, 66) in some insect hosts. The production of a significantly lower number of viable adult offspring by antibiotic-treated P. soemius females could indicate that the presence of Rickettsia increases fitness, or alternatively, it could be due to a toxic effect resulting from treatment with antibiotics. The finding that rifampin treatment is not detrimental to the longevity of treated adult females suggests that antibiotic toxicity probably does not result in a lower number of viable adult offspring. On the other hand, our results did not show at what level (oogenesis, embryogenesis, larval development) Rickettsia could induce a fitness benefit. In fact, although there was a trend toward a reduction in fitness in treated females, the data for fecundity (total number of eggs oviposited), the percentage of unhatched eggs, and the level of mortality of the progeny during development from larvae to pupae were not significantly different for treated and untreated females. However, the results presented here, although not sufficient to support the hypothesis that there is a clear benefit to the fitness of the host, seem to indicate that infection does not have a detrimental effect on the fecundity, fertility, and longevity of infected P. soemius females.
Analysis of 16S rRNA and gltA gene sequences revealed that the PI Rickettsia of P. soemius showed the highest level of similarity with the tick symbiont R. bellii (99% and 97% nucleotide similarity for 16S rRNA and gltA gene sequences, respectively). Phylogenetic analyses based on 16S rRNA and gltA gene sequences placed the P. soemius symbiont in a well-supported basal clade that includes R. bellii and the symbionts of non-blood-feeding arthropods. On the basis of 16S rRNA gene data, the symbiont of P. soemius appeared to be distantly related to the PI Rickettsia of N. formosa (no data for the gltA gene of this strain were available from public databases). Even though these two PI bacteria both occur in eulophid wasps, they did not form a monophyletic group. The PI Rickettsia of N. formosa was related to pathogenic species in the typhus and spotted fever groups, as shown by previous phylogenetic analyses (53, 54). Using the most recent and well-resolved phylogeny of Rickettsia (86), the PI Rickettsia of P. soemius can be placed in the ancestral bellii group, while the PI Rickettsia of N. formosa is a member of the transitional group, a lineage at the top of the tree that shares immediate ancestry with the spotted fever group and whose members have genotypic and phenotypic characteristics intermediate between those of the typhus and spotted fever groups (25, 26). Recently, three other rickettsiae whose biology is unknown, each of which infects a different species of eulophid wasps, have been found to cluster in the transitional group (86), leaving open the possibility that they and the N. formosa symbiont might form a monophyletic group of transitional PI Rickettsia strains.
One hypothesis to account for the absence of monophyly for the PI rickettsiae of P. soemius and N. formosa is that the ability to manipulate the reproduction (induction of parthenogenesis) of closely related hosts (eulophid wasps) may have been acquired independently by unrelated rickettsiae during evolution. An alternative hypothesis is that parthenogenesis-inducing genes were laterally transferred from PI Rickettsia harbored by a thelytokous wasp species to non-PI Rickettsia harbored by a bisexual wasp species. Recently, it has become evident that lateral gene transfer may play an important role in the evolution of rickettsial genomes (10, 25). In particular, lateral gene transfer appears to have occurred at a high frequency between rickettsiae that harbor plasmids, which was the case for the ancestral species R. bellii and the transitional species R. felis (26). An example of recombination between rickettsiae in the bellii and transitional groups has recently been discovered in the ladybird beetle Coccidula rufa (86). As the PI Rickettsia of P. soemius is closely related to R. bellii and the PI Rickettsia of N. formosa is closely related to R. felis, it is possible that these symbionts exchange genes, including genes involved in PI. However, lateral gene transfer could have occurred only if horizontal transmission of Rickettsia between different parasitoid species took place. Considering the biology of many eulophid parasitoids, horizontal transmission of symbionts might be possible. P. soemius and N. formosa, for example, can parasitize the same leafminer host species (8), and it is likely that they can develop as facultative hyperparasitoids by feeding on a different parasitoid species (2, 11, 52, 78). Then female larvae of a bisexual species (e.g., P. soemius) harboring non-PI Rickettsia could acquire PI Rickettsia during development (hyperparasitization) on larval stages of an infected thelytokous species (e.g., N. formosa). Although there is no experimental evidence of horizontal transmission of Rickettsia between eulophid species so far, interspecies horizontal transmission from host to parasitoid larva has been demonstrated to occur in nature for the Rickettsia symbiont of the whitefly Bemisia tabaci, a member of the bellii group (17). In this case, however, Rickettsia ingested during parasitoid larval development persists in the adult wasp and localizes in the reproductive tissues, but it is not vertically transmitted to the parasitoid progeny. Furthermore, natural interspecies horizontal transmission during hyperparasitization has been shown to occur for Wolbachia, and the recipient parasitoid species lost the symbionts after a few generations (34). If horizontal transmission of Rickettsia occurred between infected eulophid species (e.g., transmission from N. formosa to P. soemius), resident symbionts could have acquired parthenogenesis-inducing genes by lateral transfer from PI Rickettsia. Eventually, even though PI Rickettsia did not become established in the recipient host (a hyperparasitoid [e.g., P. soemius]), its PI genes did become established, giving rise to thelytokous reproduction. Recently, it has been discovered that lateral transfer of novel genes from other intracellular bacteria, like Cardinium, could have contributed to the evolution of Rickettsia genomes (24). Based on this finding, the possibility that a parthenogenesis-inducing gene of PI Rickettsia was acquired by lateral transfer from other manipulators of reproduction that inhabited the same host cannot be excluded. In thelytokous P. soemius, such an event would have been followed by loss of the donor bacterium.
To date, complete sequencing of the Rickettsia genome and comparative analyses have focused on species that are symbionts of blood-feeding arthropods and species that are generally pathogenic to vertebrates (26). Determining the composition of the genomes of nonpathogenic insect-associated rickettsiae, like PI symbionts, and determining whether these organisms can be horizontally transmitted between different host species are important for understanding the evolutionary history of these lineages. It is also important to understand the contributions of mechanisms of lateral gene transfer, such as plasmids and bacteriophages, which may transfer the ability to manipulate reproduction within and among Rickettsia lineages, as well as between Rickettsia and other intracellular bacteria.
We thank Paolo Navone, who first collected and provided the thelytokous parthenogenetic wasp P. soemius, Salvatore Cozzolino for providing sequencing facilities, and Kerry Oliver and Stephan Schmitz-Esser for their useful comments on the manuscript. This work was performed at Istituto per la Protezione delle Piante, CNR, Portici (NA), Italy.
This work was supported by CNR grant AG.P01.013.
Published ahead of print on 19 February 2010.