Recent years have witnessed the revelation of inherited microorganisms as major features of arthropod biology. Among them, the α-Proteobacteria Wolbachia has experienced a remarkable explosion of interest, fascinating biologists by its capacity to drive the evolution of their hosts, for instance inducing postzygotic sterility, altering sex ratios or affecting the host's genome organization (Bandi et al., 2001; Dunning Hotopp et al., 2007). The lifestyle of Wolbachia is a true success story since these endosymbionts became among the most abundant on Earth—they are four or five times as common as any other inherited bacterium (Duron et al., 2008)—by manipulating the reproduction of arthropods. How this manipulation takes place is an intriguing quest for Science but very little is actually known. A key step forward was achieved by studying the behaviour of a variety of Wolbachia strains in the melanogaster group of Drosophila fruit flies. However, in a recent issue of Heredity, Clark et al. (2008) present new insights on Wolbachia biology through the use of understudied host model systems. Although Drosophila reigns supreme as a laboratory system, Clark et al. (2008) establish that conclusions from the Wolbachia–Drosophila interactions may not hold true across all Wolbachia–insect interactions.

The ultimate cause of the reproductive parasitism exerted by Wolbachia can be found at the cytoplasmic level, where there is inheritance only through the female hosts, therefore making males a ‘dead end’ (Bandi et al., 2001). In some cases, this asymmetry has selected for an increased production of daughters—but not of sons—by infected females. More commonly, Wolbachia exerts a form of conditional male sterility termed cytoplasmic incompatibility (CI). Through CI, Wolbachia hamper the reproduction of the uninfected females mated with infected males indirectly. Zygotes produced in this cross are killed in diploid species, and killed or forced into male development in haplodiploid species. This loss of uninfected female progeny confers a reproductive advantage to the infection, and can drive Wolbachia through natural populations to high prevalence. Although the phenotype and population biology of CI are well known, the knowledge of proximate mechanism remains desperately limited. CI is classically considered to result from two bacterial components: a mod—for modification—function that affects sperm, and a resc—for rescue—function provided by the Wolbachia present in the egg that restores compatibility (Werren, 1997). When the spermatozoon enters the egg, CI is expressed by the failure of paternal chromosomes to properly condense, whereas the maternal chromosomes enter normally into mitosis, leading to haploid or aneuploid conditions.

A key requirement for understanding how Wolbachia induce CI is to establish when and how Wolbachia interact with developing sperm. Studies of the interaction of diverse Wolbachia strains during Drosophila spermatogenesis have been the major focus of study to date. Wolbachia are found in the Drosophila testes but are absent from mature sperm, being removed from developing cysts with cytoplasm and most other organelles. Variation of Wolbachia density within cysts for different Wolbachia strains, and across different Drosophila species, correlates with the variation in CI levels between these species, suggesting that an abundance of Wolbachia in the testes is necessary (although not sufficient) to induce CI (Clark et al., 2003; Veneti et al., 2003). Furthermore, Wolbachia density in testes has been shown to decrease with male ageing in Drosophila, which correlates with the reduced strength of CI found in aged males (Clark et al., 2003). These observations form the basis of the Wolbachia-infected spermatocyte/spermatid hypothesis (WISSH), which argues that infected cysts represent the cellular basis of CI (Clark et al., 2003). According to the Drosophila data, Wolbachia infection of spermatocytes is then required for CI expression.

Clark et al. (2008) challenge this hypothesis using two non-Drosophila models: the haplodiploid wasp Nasonia vitripennis and the diploid beetle Chelymorpha alternans. Both species show high levels of CI, resulting in all-male progeny in N. vitripennis and the death of almost all embryos in C. alternans. High CI levels in both host species suggest that all spermatozoa have been successfully modified by Wolbachia. According to WISSH, most of the spermatocytes should be infected by Wolbachia to explain the high CI level observed but a very different pattern of infection is actually observed: only a few developing sperm are seen to be infected in N. vitripennis and none in C. alternans. The convergence of these observations demonstrates that Wolbachia do not modify the male chromosomes during late spermatogenesis in situ. The sperm modification by Wolbachia takes place either across tissue membranes from somatic infected cells or very early in the development of the host before spermatogenesis. The WISSH's scope therefore appears limited to WolbachiaDrosophila interactions, and cannot be generalized across Wolbachia–host interactions.

Many studies that review Wolbachia biology have speculated on how CI is induced on the basis of Drosophila results. We should now raise the question of the adequacy of the other WolbachiaDrosophila results. With respect to the mod function, CI intensity has been shown to decrease with male ageing in Drosophila, but no effect has been found in numerous other hosts, as documented in N. vitripennis by Clark et al., suggesting that Drosophila is a non-typical Wolbachia–host. With respect to the resc function, it is sometimes argued that CI embryos exhibit the same defects as in Drosophila, suggesting a conserved mechanism, but even this assertion must be considered carefully. Embryos blocked at a variety of developmental stages (from recently fertilized to quite well developed) are usually observed in all incompatible cross types in Drosophila, whether females are uninfected or infected by a Wolbachia strain incompatible with the one present in males (Callaini et al., 1996). By contrast, CI crosses in the mosquito Culex pipiens produced embryos that consistently failed to develop beyond a very early stage when mothers are uninfected but, when males and females are infected by incompatible Wolbachia strains, a large number of embryos are blocked at later developmental stages (Duron and Weill, 2006). Maternal Wolbachia present in the eggs allow some morphogenesis in C. pipiens, notwithstanding the infection being different to that in the male, in contrast to the Drosophila case. Such observations highlight another level at which there is variation in Wolbachia–host interactions, and that study beyond Drosophila is essential to gain a complete analysis. There is now no doubt that it is wise to conduct more investigations on a wide range of host–Wolbachia associations before constructing a general model of CI. Furthermore, the relatively recent identification of the Bacteroidetes bacterium Cardinium as another causative agent of CI in arthropods represents a fascinating model for further comparative studies (Hunter et al., 2003).