Insights into Wolbachia Obligatory Symbiosis - Insect Symbiosis - page 205

named w At1, w At2, and w At3. The goal of the study was to determine the respective roles of each
strain in oogenesis (Dedeine et al., in preparation). By using moderate antibiotic treatments followed
by a selection of isofemale lines, the three Wolbachia could be separated and stabilized into different
lines. During the Ýve-generation selection, a strong relationship was found between individual
offspring production and the presence of one strain, w At3. All the females that lost this strain were
unable to produce progeny. Conversely, most of the females in which w At3 was detected produced
offspring. These results have two consequences: Ýrst, they are not in agreement with the presence
of an unknown bacterium. Indeed, there is no reason a priori that such an unknown agent would
be strictly associated with w At3 only. Second, these results strongly suggest that only w At3 is
obligatory for oogenesis.
That w At3 is needed for oogenesis was conÝrmed at the end of the selection procedure. Indeed,
four lines were Ýnally established and stabilized (of the eight combinations normally possible with
three bacterial strains): a line harboring the three Wolbachia strains ÑPi(123),Ò the two bi-infected
lines ÑPi(13)Ò and ÑPi(23),Ò and the monoinfected one ÑPi(3),Ò which is infected only by w At3.
Fecundities of these lines were the same as in untreated control lines, demonstrating that only the
presence of w At3 is obligatory for oogenesis (Dedeine et al., in preparation). Moreover, intercross-
ing these lines demonstrated that variants w At1 and w At2 induced bidirectional CI resulting in
high female mortality among offspring (75%) compared to the compatible crosses (Dedeine et al.,
in preparation). Female mortality was already described as FM CI-phenotype (for ÑFemale Mor-
talityÒ) in haplodiploid species (Breeuwer, 1997; Vavre et al., 2000; Vavre et al., 2002).
In conclusion, triple Wolbachia infection in A. tabida is associated with different relationships
with the host. In each individual, one Wolbachia ( w At3), which is obligatory for oogenesis, coexists
with two reproductive parasitic ones ( w At1 and w At2), which induce bidirectional CI. We know
that Wolbachia can have a wide range of effects on hosts, but the respective contributions of host
and bacteria in this diversity are generally unknown. In the case of A. tabida , the coinfection by
an obligatory strain and two reproductive parasitic strains in the same host individual (i.e., in the
same host genotype) demonstrates the decisive role of bacterial genotype in induction of each
phenotype observed. Thus, different Wolbachia can be genetically specialized in different relation-
ships within the same host.
Two questions have arisen from the studies on A. tabida . First, what mechanisms are responsible
for the dependence on Wolbachia for oogenesis? And what evolutionary scenarios account for it?
A Novel Reproductive Manipulation by Wolbachia ?
Sterility of cured females could result from differences in their own infection status compared to
that of their mothers, leading to some kind of intergeneration incompatibility. Its principle could
be summarized as follows: when Wolbachia is present in the mother, it must also be present in the
daughters, or these will be completely sterilized. Such an effect must consider two independent
products of Wolbachia . First, Wolbachia could act by producing a sort of Ñtime-bomb toxinÒ in
infected mothers. After maternal transmission to offspring, this toxin could be inactivated in infected
daughters by a second bacterial product (Ñdefusing moleculesÒ) but not in uninfected daughters
that are lacking this protective factor and that thus suffer deleterious effects of the toxin that made
them sterile. These modes of action imply three functional constraints. First, the effect of the toxin
must act speciÝcally in oogenesis. Second, the toxin must still be active when the defusing molecule
is absent (the two active molecules could have different turnover kinetics). Third, the effect of the
toxin must be maintained after it is lost from the insect (the toxin could induce epigenetic modi-
Ýcations that persist in the insect). This kind of phenotype has never been shown to be associated
with the presence of symbiotic microorganisms. However, a single nuclear gene has been reported
to induce a similar effect in the beetle T. castaneum (Beeman et al., 1992), showing that this
hypothesis is not unreasonable. In this case, when the selÝsh gene is present in the mother, it must
also be present in the zygote or the zygote will die. In reference to Medea , the Greek mythological
mother who killed her own offspring, this phenotype was called the Medea phenotype.