Spiroplasma Bacteria Enhance Survival of Drosophila hydei Attacked by the Parasitic Wasp Leptopilina heterotoma

Background Maternally-transmitted associations between endosymbiotic bacteria and insects are ubiquitous. While many of these associations are obligate and mutually beneficial, many are facultative, and the mechanism(s) by which these microbes persist in their host lineages remain elusive. Inherited microbes with imperfect transmission are expected to be lost from their host lineages if no other mechanisms increase their persistence (i.e., host reproductive manipulation and/or fitness benefits to host). Indeed numerous facultative heritable endosymbionts are reproductive manipulators. Nevertheless, many do not manipulate reproduction, so they are expected to confer fitness benefits to their hosts, as has been shown in several studies that report defense against natural enemies, tolerance to environmental stress, and increased fecundity. Methodology/Principal Findings We examined whether larval to adult survival of Drosophila hydei against attack by a common parasitoid wasp (Leptopilina heterotoma), differed between uninfected flies and flies that were artificially infected with Spiroplasma, a heritable endosymbiont of Drosophila hydei that does not appear to manipulate host reproduction. Survival was significantly greater for Spiroplasma-infected flies, and the effect of Spiroplasma infection was most evident during the host's pupal stage. We examined whether or not increased survival of Spiroplasma-infected flies was due to reduced oviposition by the wasp (i.e., pre-oviposition mechanism). The number of wasp eggs per fly larva did not differ significantly between Spiroplasma-free and Spiroplasma-infected fly larvae, suggesting that differential fly survival is due to a post-oviposition mechanism. Conclusions/Significance Our results suggest that Spiroplasma confers protection to D. hydei against wasp parasitism. This is to our knowledge the first report of a potential defensive mutualism in the genus Spiroplasma. Whether it explains the persistence and high abundance of this strain in natural populations of D. hydei, as well as the widespread distribution of heritable Spiroplasma in Drosophila and other arthropods, remains to be investigated.


Introduction
Heritable (i.e., maternally-or vertically-transmitted) associations between endosymbiotic bacteria and insects are ubiquitous [1], many of which are obligate because the host cannot reproduce without the symbiont, and thus, mutualistic because the association is beneficial to both partners. However, numerous heritable insect-endosymbiont associations are facultative (i.e., the host can generally reproduce without the symbiont), which renders the bacterial symbiont more prone to loss due to imperfect vertical transmission or due to infection costs to the host [2]. Many inherited facultative symbionts appear to reduce this risk by manipulating their host's reproduction (e.g., cytoplasmic incompatibility and son-killing) [2], but many do not manipulate reproduction [1], so their persistence must depend on other mechanisms [2]. Indeed, recent studies report heritable facultative symbionts that confer fitness benefits to their hosts, which include defense against natural enemies, tolerance to environmental stress, and increased fecundity [1,3]. Despite the broad taxonomic diversity of inherited facultative bacteria [1], most of these mutualistic endosymbionts are clustered within the phylum Proteobacteria-e.g., the intensely-studied bacteria of aphids (class Gamma-proteobacteria; family Enterobacteriaceae) [4][5][6][7][8] and Wolbachia (class Alpha-proteobacteria) of Drosophila melanogaster [9][10][11]-, with a single example outside this group (i.e., phylum Actinobacteria; Streptomyces associated with digging wasps [12]).
Spiroplasma belongs to the class Mollicutes, an ancient wall-less bacterial group within the Gram-positive lineage [13]. Spiroplasma is a diverse genus that associates with many host plants and arthropods, particularly insects. Recent surveys suggest that Spiroplasma-arthropod associations are quite common [14][15][16][17][18][19], and it has been suggested to be one of the most speciose bacterial genera [20]. Within arthropods, some lineages of Spiroplasma transmit horizontally (usually via a plant host) while others are heritable (are transmitted vertically). Among the inherited Spiroplasma, a few strains that associate with flies, butterflies, and beetles are known to kill the sons of infected females [21]; a form of reproductive parasitism known as male-or son-killing. However, numerous heritable strains with undetermined phenotypes exist [15,16,[22][23][24]. Such inherited symbionts are likely candidates for the discovery of new mutualistic associations [3].
The genus Drosophila is prone to associations with inherited Spiroplasma [16]. At least nine strains of Spiroplasma (defined on the basis of DNA sequence divergence and host species) have no known phenotype [24], some of which can reach relatively high frequencies of infection (above 60%) in natural populations [24,25]. Their high prevalence and apparent lack of reproductive manipulation suggests that these endosymbionts might confer net fitness benefits to their hosts. Examination of one of these highly prevalent strains, ''haplotype 1'' in Mateos et al. [23], which is identical to the strain studied by Osaka et al. [19] at the three genes examined to date, suggests that: (a) it has high, but imperfect vertical transmission (95-100% at 25uC) [19]; (b) it does not affect fitness under lab conditions [19], although the fitness measures examined were not stated; and (c) it does not cause obvious malekilling [23,25] or strong cytoplasmic incompatibility [19]. Therefore, prevalence of such an endosymbiont might be explained by alternative mechanisms such as defense against natural enemies.
Here we examine whether a highly prevalent strain of Spiroplasma, ''haplotype 1'' in Mateos et al. [23] protects its natural host Drosophila hydei against parasitism by Leptopilina heterotoma, a common parasitic wasp. Leptopilina heterotoma is a cosmopolitan solitary larval parasitoid of several species of Drosophila [26]. These wasps usually lay eggs on first-and second-instar Drosophila larvae. If the oviposition attack is successful, the developing wasp larva kills its host during the pupal stage, and a single adult wasp emerges from the fly puparium ,23 days after oviposition. Leptopilina heterotoma overcomes the host immune response by actively suppressing host encapsulation of its eggs using Virus Like Particles (VLPs) produced within its venom glands (referred to as long glands). These VLPs enter the larval hemolymph along with the egg and rapidly bind to and enter host lamellocytes, ultimately causing them to lyse [27]. Infection by L. heterotoma also causes apoptosis of pro-hemocytes in the lymph gland, and possibly of the circulating plasmatocytes [28].
Because L. heterotoma routinely probes host larvae to determine whether they have been parasitized previously by a conspecific [29][30][31], it may be capable of detecting whether or not a fly larva is infected with Spiroplasma, as these bacteria reside in the hemolymph and other tissues. Similarly, since Wolbachia has been reported to modify Drosophila's behavior [32], it is possible that Spiroplasma could improve the ability of its host to escape from wasp attacks. For example, enhancement of the rolling behavior response against wasp oviposition attempts [33] or stimulation of a burying response, since wasps attack at the substrate surface, could reduce oviposition rates in Spiroplasma-infected D. hydei larvae. To test this, we examined whether Spiroplasma-infected larvae exhibited fewer oviposition attacks than their Spiroplasma-free counterparts.

Fly strains
We collected nine wild Drosophila hydei females with banana baits in College Station, TX, USA (between April-Nov 2008). Each female was used to establish an independent isofemale line (hereafter fly strain)-i.e., mating only allowed among its descendants. To confirm that these fly strains were free from heritable bacteria, at least three females per fly strain were PCR-screened for infection by Wolbachia, Spiroplasma and other heritable endosymbiotic bacteria. Screening for Wolbachia and Spiroplasma was carried out on whole fly DNA extracts with primers specific to these two genera; wsp for Wolbachia (PCR conditions in Mateos et al. [23]; and p58 [25] for Spiroplasma (newly -designed primers p58IV_F 59-AAAGGTTTACATTCACCAAGTCG-39 and p58IV_R 59-AATTGTTCATTAACTTTATCTTGTGG-39; annealing temperature 53uC). Screening for other heritable bacteria was carried out on ovary DNA extracts with ''universal'' primers for the bacterial 16S rRNA gene (primer pairs 10F-1507R and 27F-1495R) and for a 16S-23S rRNA fragment (primer pair 559F-35R); primer sequences and PCR conditions are described in Mateos et al. [23]. All PCR reactions were carried out with appropriate negative (water or DNA extraction buffer) and positive controls (Escherichia coli DNA extracts for the universal primers sets, and known Spiroplasma-positive and Wolbachia-positive DNA extracts for the Spiroplasmaand Wolbachiaspecific primers; respectively). Throughout this study, flies were maintained on Banana-Opuntia medium at 25uC and 12 h:12 h light:dark regime.

Artificial infection of Spiroplasma
After 4-25 generations in the lab, 3-8 females per strain were artificially infected with Spiroplasma previously isolated from Mexican D. hydei. This strain is genetically identical to Spiroplasma strain in Kageyama et al. [25] based on previously published and new sequences of the 16S rDNA, DnaA, FruA and FruR genes. The donor flies were collected in 2004 and are naturally infected with Spiroplasma hap1 [23] and infection with other heritable bacteria was previously ruled out by PCR screening of ovary DNA extracts with primers and conditions described in Mateos et al. [23]. Artificial infection of uninfected flies was performed via adult-toadult hemolymph microinjection using pulled microcapillaries and a manual microinjector. Infection status of the artificially infected flies was confirmed via Spiroplasma-specific PCR (described above) and/or examination of hemolymph under dark field microscopy.

Antibiotic Treatment
We used antibiotics to cure Spiroplasma-infected flies in two cases: (1) to obtain the uninfected control for the naturally infected isoline (TEN104-102; which also served as the donor of hemolymph for all our artificial infections); and (2) for isoline 6, because our originally uninfected line was lost before completion of our experiments. We added a combination of tetracycline and erythromycin (final concentration = 0.2 and 0.16 mg/ml, respectively) to our standard in Banana-Opuntia media for two generations. The third generation was maintained on antibioticfree food to which we added a solution of crushed dead flies that were naturally Spiroplasma-free (to allow for recovery of the normal gut flora). After the third generation, the flies were maintained on antibiotic-free media with no additives. Spiroplasma-infection status was assessed by PCR as described above.

Fly survival
All experiments were carried out 2-10 generations after artificial infection or antibiotic treatment of Drosophila. Prior to experiments, both infected and uninfected flies were maintained at low-density larval conditions (,30 larvae/vial). For the fly survival experiment, each virgin female (#10-days-old) was placed in a mating cage with two mature ($10 day-old) uninfected males (from its own strain), and allowed to mate and oviposit on Petri dishes with medium. Females and males were then removed, and females were screened for infection status. Approximately 30 first instar larvae were collected from the Petri dishes and transferred to a fresh food vial. Five experienced female wasps (i.e., wasps that had been allowed to oviposit on D. melanogaster larvae prior to experiment), and three male wasps were placed in the vial with larvae and removed three days later. We used a single inbred highly virulent L. heterotoma wasp strain known as Lh14 [34] for all experiments. This strain was tested and confirmed positive for Wolbachia based on PCR assays targeting the wsp gene (primers and methods in Mateos et al. [23]). We recorded the number of larvae introduced to the vial, puparia, emerging flies, and emerging wasps in each vial. Puparia from which neither a fly nor a wasp emerged after 30 days were regarded as inviable. To evaluate the effect of Spiroplasma infection itself on fly survival rate, we also carried out the same experiment in absence of wasps. We performed 2-11 replicates per treatment per fly strain; each replicate corresponded to a separate vial. We used SAS Enterprise Guide version 4.2 statistical package to fit a Generalized Linear Mixed Model (GzLMM) with a binomial distribution of the raw data for the: (a) number of emerging adult flies/number of starting larvae (i.e., larva-to-adult fly survival rate); (b) number of emerging adult flies/total number of puparia (i.e., pupa-to-adult fly survival rate); (c) number of emerging adult flies/number of total emerging adults (flies + wasps) (i.e., adult fly emergence rate); (d) number of pupae/ number of starting larvae (i.e., larva-to-pupa survival rate); and (e) number of emerging adult wasps/number of total emerging adults (flies + wasps) (i.e., adult wasp emergence rate). The independent variables were fly strain (random), Spiroplasma infection status (fixed), and their interaction term (random). Significance tests of random effects were based on the ratio of pseudo-likelihoods (Covtest in SAS).

Differential oviposition
We examined whether Spiroplasma-infected larvae suffer significantly fewer ovipositions than their Spiroplasma-free counterparts. For each of three isolines (isolines 1, 34 and 57 used in experiments described above), three Spiroplasma-free and three Spiroplasma-infected female flies were individually mated with 2 uninfected males each, and allowed to oviposit for two days. Thirty first-to-second instar fly larvae were transferred into a fresh vial (three replicates = vials; one per female) with five parasitoid wasp females as described above. After 1.5 days, 10 out of 30 fly larvae were removed from each vial and dissected in 1 X PBS solution, to determine the number of wasp eggs per fly host. We applied a GzLMM model with a binary distribution and a logit link function to compare the presence (one or more) or absence (zero) of wasp eggs in the Spiroplasma-infected and Spiroplasma-free fly larvae. Fly strain and vial nested within fly strain were included as random factors. We also applied a GzLMM model with a Poisson distribution and a log link function comparing five categories of wasp egg numbers (0, 1, 2, 3, 4; which was the maximum number of wasp eggs observed per larva).

Fly survival
We compared the survival of Spiroplasma-infected and Spiroplasma-free D. hydei flies in the presence and absence of L. heterotoma parasitic wasps. We only included results from replicates in which the expected infection status of the mother was confirmed via PCR. In the absence of wasps, the effect of Spiroplasma infection state was not significant for any of the fly survival measures (Table 1 and Fig. 1), although the interaction between fly strain  and Spiroplasma infection state was significant for two measures: fly larva-to-adult (P = 0.0082) and fly pupa-to-adult survival (P = 0.0002). Closer examination of these interactions ( Fig. 2A) suggests that Spiroplasma increases larval-to-adult survival in four fly strains (7, 10, TEN, 57) and reduces it in four other strains (17,20,23,34). A similar pattern was observed for fly pupa-to-adult survival (Fig. 2B). When the two groups of fly strains (those with increased and those with decreased survival) were analyzed separately, the fly strain x infection state interaction term lost significance, and the effect of infection state remained nonsignificant (results not shown).
In the presence of wasps, fly larva-to-adult survival, fly pupa-toadult survival (Fig. 1A), and fly adult emergence rate (i.e., number of adult flies/number of total adults including wasps) were significantly higher in the Spiroplasma-infected flies than in the Spiroplasma-free flies (Table 1). Similarly, as expected, wasp survival was significantly lower in Spiroplasma-infected treatments (Table 1). A significant interaction between fly strain and Spiroplasmainfection state was observed for fly adult emergence rate (P = 0.0038; Table 1), but all fly strains exhibited an increased survival in the Spiroplasma-infected treatment (Fig. 2C). The protective effect of Spiroplasma is striking, but not complete. In the absence of wasps, fly pupa-to-adult survival averaged ,89.87% for both, Spiroplasma-free and Spiroplasma-infected flies (Fig. 1A). In the presence of wasps, the survival of Spiroplasma-free flies averaged ,7.17%, but survival of Spiroplasma-infected flies was much higher (,47.30%). In contrast, larva-to-pupa survival was not significantly affected by Spiroplasma infection in the presence of wasps, (Table 1; Fig. 1B). Nevertheless, the presence of wasps is correlated with higher fly mortality (lower larva-to-pupa survivorship) during the larva-to-pupa stage [wasp effect P = 0.0065; GzLMM including wasp treatment, infection state, and fly strain (random effect); Fig. 1B]. Therefore, our results suggest that Spiroplasma protects the fly against wasp-induced mortality, but that the effects of this protection are only reflected in differential fly survival during the pupa-to-adult transition.

Differential Oviposition
Our fly and wasp survival measures suggest that Spiroplasma infection confers protection to D. hydei against wasp-induced mortality, and that this effect is not detectable before the fly pupal stage. Nevertheless, we tested whether this protective effect could be also mediated by a pre-oviposition mechanism, in which Spiroplasma-infected larvae suffer significantly fewer ovipositions than their Spiroplasma-free counterparts. The number of wasp eggs found per fly larva did not differ significantly between Spiroplasmafree and Spiroplasma-infected treatments, for either the GzLMM with Poisson distribution (F (1,112) = 0.89; P = 0.3476) or the GzLMM with a binary distribution (i.e., one or more wasp eggs grouped into a single category: F( 1,112 ) = 0.40, P = 0.5248; see Fig. 3). No significant effects of fly strain or vial were detected. These results suggest that Spiroplasma infection does not protect fly larvae from wasp oviposition attacks.

Discussion
Our results indicate that Spiroplasma confers protection to its fly host against wasp-induced mortality during the pupal-to-adult transition. Although there appears to be significant variation among fly strains in the degree of protection induced by Spiroplasma, indicated by the significant Spiroplasma infection state x fly strain interaction for fly adult emergence rate (Table 1), evidence for Spiroplasma protection was found in every fly strain examined. This interaction could be due to genetic variation in ''natural'' (i.e. not Spiroplasma-mediated) resistance, as suggested by the significant effect of fly strain on survival against wasp attack in the absence of Spiroplasma infection (x 2 = 53.16; P,0.0001; d.f. = 1; see Fig. 2C). Indeed, variation in Drosophila natural resistance to parasitoids has been reported [35]. The infection x fly strain interaction could also result from variation in the original transfection itself (e.g., different transfection hemolymph volumes or bacterial titers, or physical damage to transfected female) rather than host genetic background. Despite the effect of host Although infection by Spiroplasma does not restore survival to the levels observed in the absence of wasps, the degree of protection is high and potentially important in nature (although this remains to be tested). Furthermore, it is possible that protection is actually greater than shown here, as we cannot be certain that all of the larvae included in the Spiroplasma-infected treatments were infected, because bacterial transmission is less than 100% efficient [19]. We assessed Spiroplasma-infection status via PCR for a subset of the emerging adult flies (,200 individuals) from the wasptreated Spiroplasma-infected experiments. Indeed, 100% of these flies were Spiroplasma-positive, further supporting the notion that Spiroplasma confers protection against L. heterotoma, since no Spiroplasma-free survivors were observed in these vials.
At this stage the protective mechanism is unknown, but it is exerted after wasp oviposition and is detectable as differential survival during the pupa-to-adult transition. Several post-oviposition mechanisms of protection by Spiroplasma are possible, for example: (a) reduced availability of resources necessary for wasp development (if Spiroplasma consumes resources that would otherwise be available for the parasitoid); (b) the presence of a substance toxic to the parasitoid; and/or (c) an enhanced immune response of the fly larva against the parasitoid (e.g., by countering the encapsulation-suppressive effect of the parasitoid venom or by increasing production of lamellocytes). Presence of a toxic substance encoded by a bacteriophage is the hypothesized protective mechanism provided by the Gamma-proteobacterium Hamiltonella defensa to its aphid host against a parasitoid [36,37]. In this regard, Spiroplasma strains associated with several species of Drosophila, including a non-male killing strain of D. hydei, are  known to possess bacteriophages with lytic activity [22,38,39]. Whether these viruses play a role in the defense against wasp larvae in this system remains to be explored. In this study, we observed no evidence of melanotic capsules in surviving flies, which suggests that the protective mechanism is not through an enhanced fly immune response to the parasitoid. However, although melanotic response is an indication of encapsulation, examples exist of parasitoid larvae that have been killed in the absence of a melanotic response [40], so further work would be needed to study this possible mechanism.
Whether or not the protective effect of Spiroplasma reported here explains the long-term persistence of this Drosophila-Spiroplasma association remains to be tested and depends on, among other things, the degree of D. hydei mortality caused by L. heterotoma under natural conditions, the reproductive fitness of Spiroplasma-infected flies that survive a wasp attack, the degree of protection conferred by Spiroplasma across different host, symbiont and parasitoid backgrounds (e.g., host x parasitoid and symbiont x parasitoid interactions) [41], and any fitness costs associated with Spiroplasma infection. Future studies to resolve this issue are warranted. Nevertheless, the high prevalence (.60%) of Spiroplasma infection in two geographically distinct populations of D. hydei found in Arizona [24] and Japan [25]. the cosmopolitan distribution of L. heterotoma, and the high level of Spiroplasma-free D. hydei mortality when attacked by L. heterotoma in our experiments, strongly suggest that the protection conferred on D. hydei by Spiroplasma in nature may be significant.
Our study has important implications. First, given the rising number of reports of Spiroplasma in arthropods [13][14][15][16][17]23,[42][43][44][45][46][47][48][49], particularly in the genus Drosophila, most of which have undetermined effects [15,16,[22][23][24], there is a high probability of discovering defensive mutualisms in different combinations of host-Spiroplasma lineages. Second, our discovery opens the possibility of using the Drosophila model system to study defensive mutualisms. Third, our findings could have important implications in studies that use the Drosophila-parasitoid model system, such as research on cellular immunity [40], behavior [50], and evolutionary ecology [35]. Fourth, if protection against parasitoids is a general feature of maternally-transmitted Spiroplasma in insects, it could have important implications for biological control, because parasitic wasps are the most successful group of biological control agents [51], and an increasing number of insects are reported to be infected with Spiroplasma symbionts. Fifth, our study significantly expands the taxonomic range of both the inherited defensive symbionts (to include the wall-less class Mollicutes of the phylum Firmicutes) and also the hosts protected against parasitoids by an endosymbiont (to include Diptera).