Wolbachia Transfer from Drosophila melanogaster into D. simulans: Host Effect and Cytoplasmic Incompatibility Relationships

Corresponding author: Denis Poinsot, Institut Jacques Monod, Laboratoire de Dynamique du Génome et Evolution, CNRS - Université Paris 7, 2 place Jussieu 75251 Paris Cedex 05, France., poinsot{at}ccr.jussieu.fr (E-mail).

Communicating editor: A. A. HOFFMANN

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Wolbachia are maternally transmitted endocellular bacteria causing a reproductive incompatibility called cytoplasmic incompatibility (CI) in several arthropod species, including Drosophila. CI results in embryonic mortality in incompatible crosses. The only bacterial strain known to infect Drosophila melanogaster (wDm) was transferred from a D. melanogaster isofemale line into uninfected D. simulans isofemale lines by embryo microinjections. Males from the resulting transinfected lines induce >98% embryonic mortality when crossed with uninfected D. simulans females. In contrast, males from the donor D. melanogaster line induce only 18–32% CI on average when crossed with uninfected D. melanogaster females. Transinfected D. simulans lines do not differ from the D. melanogaster donor line in the Wolbachia load found in the embryo or in the total bacterial load of young males. However, >80% of cysts are infected by Wolbachia in the testes of young transinfected males, whereas only 8% of cysts are infected in young males from the D. melanogaster donor isofemale line. This difference might be caused by physiological differences between hosts, but it might also involve tissue-specific control of Wolbachia density by D. melanogaster. The wDm-transinfected D. simulans lines are unidirectionally incompatible with strains infected by the non-CI expressor Wolbachia strains wKi, wMau, or wAu, and they are bidirectionally incompatible with strains infected by the CI-expressor Wolbachia strains wHa or wNo. However, wDm-infected males do not induce CI toward females infected by the CI-expressor strain wRi, which is found in D. simulans continental populations, while wRi-infected males induce partial CI toward wDm-infected females. This peculiar asymmetrical pattern could reflect an ongoing divergence between the CI mechanisms of wRi and wDm. It would also confirm other results indicating that the factor responsible for CI induction in males is distinct from the factor responsible for CI rescue in females.

WOLBACHIA are maternally transmitted endocellular bacteria infecting numerous species of arthropods (WERREN and O'NEILL 1997 ). Their presence can lead to reproductive alterations (WERREN 1997 ; BOURTZIS and O'NEILL 1998 ), such as feminization (RIGAUD 1997 ), thelytokous parthenogenesis (STOUTHAMER 1997 ), or cytoplasmic incompatibility (CI; HOFFMANN and TURELLI 1997 ). CI is a reproductive incompatibility causing embryo mortality. It appears when a male infected by one (or more) Wolbachia strain(s) is crossed with a female that is either devoid of Wolbachia or does not harbor the strain(s) found in the male. Wolbachia strains can be classified on the basis of their CI type or by using described Wolbachia gene sequences, i.e., 16S rDNA (BREEUWER et al. 1992 ; O'NEILL et al. 1992 ; ROUSSET et al. 1992 ; STOUTHAMER et al. 1993 ), the bacterial cell cycle genes ftsZ (HOLDEN et al. 1993 ; WERREN et al. 1995 ) and dnaA (BOURTZIS et al. 1994 ), and wsp, a very variable surface protein gene (BRAIG et al. 1998 ; ZHOU et al. 1998 ).

Three CI-expressor Wolbachia strains have been described in Drosophila simulans: wRi (HOFFMANN et al. 1986 ), wHa (O'NEILL and KARR 1990 ), and wNo (MERCOT et al. 1995 ). Strains infected by wRi, wHa, or wNo are all bidirectionally incompatible; i.e., CI will occur (typically with 70–100% embryonic mortality) in both directions of crosses. Three more strains that do not cause detectable CI have been found in D. simulans. One (wMa) is restricted to Madagascar (ROUSSET and SOLIGNAC 1995 ), another (wAu) has been found in Australia and America (TURELLI and HOFFMANN 1995 ; HOFFMANN et al. 1996 ), and the most recently discovered, wKi (MERCOT and POINSOT 1998B ), is known from a single population from Mount Kilimanjaro (Tanzania).

In contrast with this diversity, only one Wolbachia CI type has been described in the sibling species D. melanogaster. In this species, the situation is as follows: (i) CI is generally weak, i.e., 20–30% embryonic mortality (HOFFMANN 1988 ; BOURTZIS et al. 1994 , BOURTZIS et al. 1996 ; HOFFMANN et al. 1994 ), although it can vary between strains of hosts, from being undetectable to >70% embryonic mortality (SOLIGNAC et al. 1994 ), and (ii) all infected strains are mutually compatible, regardless of their capacity to induce detectable levels of CI (SOLIGNAC et al. 1994 ). The Wolbachia infecting five different D. melanogaster strains were tested using the highly variable sequence of the Wolbachia surface protein wsp. Four had identical sequences, and one differed from the other by 2 out of 565 bp (ZHOU et al. 1998 ). As a comparison, the same authors find that wsp sequences of D. simulans Wolbachia strains can vary by as much as 162 bp.

The apparent difference of behavior of Wolbachia in these two sibling species of Drosophila prompts the following question: Does the level of CI expression vary because of differences between Wolbachia strains, or is it host dependent? BOYLE et al. 1993 transferred the Wolbachia strain wRi from D. simulans into an uninfected D. melanogaster strain, with a low CI as a result (15–35% in the absence of selection). However, when back transferred into a D. simulans background, wRi expressed >75% embryonic mortality in incompatible crosses. In this case, the host seemed to play a key role in the level of CI. Moreover, these authors noted that the higher CI induced by D. simulans males was associated with higher levels of bacteria in the eggs of infected females from the same stock, compared to the Wolbachia load in D. melanogaster eggs. This brings forth a second question: Could the difference of CI levels be explained simply by different bacterial loads; i.e., could it be caused by only a dosage effect, as suggested by BREEUWER and WERREN 1993 in the microhymenopteran Nasonia vitripennis.?

We have generated transinfected lines by injecting in uninfected D. simulans embryos the Wolbachia strain found in D. melanogaster (in the present work this Wolbachia is referred to as wDm). Our objective was to answer two questions: (i) Would wDm behave differently in a new host as far as CI and bacterial load are concerned? (ii) Would wDm determine a completely new CI type in D. simulans, as compared to the other known infections in this species? Once transferred into D. simulans, the strain wDm induced a very high CI, while CI expression was low in the donor D. melanogaster line. A dot blot assay on DNA extracts from whole flies showed that males of transinfected D. simulans lines did not harbor significantly more Wolbachia than males from the D. melanogaster donor line. However, the observation of testes using a DAPI coloration revealed that the percentage of infected cysts was 10 times higher in D. simulans transinfected males than in males of the D. melanogaster donor line.

When confronted with the natural Wolbachia strains found in D. simulans, wDm exhibited strong unidirectional CI against non-CI-expressor strains wAu and wKi (as well as against the nonexpressor wMau strain found in D. mauritiana). CI was strong and bidirectional between wDm and the CI-expressor strains wHa and wNo. In contrast, we present evidence that wDm is completely compatible with the CI-expressor strain wRi in one direction of cross, with the reciprocal cross exhibiting partial CI. Such an asymmetry (which is not caused by the genome of the host) had not been reported previously between two Wolbachia strains capable of inducing CI.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Strains:
D. melanogaster Wien 5 is an isofemale line that was established in 1994 from a naturally infected population (Vienna, Austria). It is used as the source of wDm Wolbachia. MelO is a naturally uninfected strain from Nasr'allah, Tunisia. It is used in crosses as a standard uninfected control. D. simulans SimO is a naturally uninfected strain from the same location as MelO, i.e., Nasr'allah, Tunisia (MONTCHAMP-MOREAU et al. 1991 ). It is used in crosses as a standard uninfected control. R1A, NHa, and R3A derive from a naturally infected population (Nouméa, New Caledonia). In the wild, Nouméa individuals are either infected simultaneously by wHa and wNo, or they are mono-infected by wHa. The laboratory strain R1A only harbors bi-infected individuals. NHa is a Nouméa stock that was found in the laboratory to be infected by wHa only, and R3A was obtained by substituting the Nouméa genome by that of SimO for 11 generations. It was then found to be infected by wNo only (MERCOT et al. 1995 ). NHaTC is an uninfected stock derived from NHa by an antibiotic treatment (tetracycline) that cured its Wolbachia infection (POINSOT and MERCOT 1997 ). It is used as the uninfected recipient D. simulans strain in the transinfection experiments. ME lines are the transinfected lines obtained after injection of wDm into NHaTC embryos. DSR is a Californian strain infected by wRi (HOFFMANN et al. 1986 ). Coffs Harbour is an Australian strain founded using only infected flies from a 1993 collection. It is carrying the variant wAu (HOFFMANN et al. 1996 ). Kc9 is an isofemale line from the Kilimanjaro 1996 strain (Mount Kilimanjaro, Tanzania) and is infected by the variant wKi (MERCOT and POINSOT 1998B ). DSW(Mau) is an isofemale line obtained by transinfecting wMau from D. mauritiana into the D. simulans Californian uninfected strain Watsonville (GIORDANO et al. 1995 ). The wMau strain is the equivalent in D. mauritiana of the wMa Wolbachia strain found in a Madagascar population of D. simulans (ROUSSET and SOLIGNAC 1995 ).

Rearing conditions:
To ensure the most favorable conditions for the infection, all strains used were maintained at 25° on axenic medium (DAVID 1962 ) as low-density mass cultures (unless otherwise stated) by crossing 20 virgin females aged 3–5 days with 25 virgin males aged 3 days. Low-density rearing is preferable because larval crowding can have a negative effect on the expression of CI (SINKINS et al. 1995 ). Crosses with young males ensure a maximum selection against possible uninfected eggs because CI is highest in young males (HOFFMANN et al. 1986 ; MONTCHAMP-MOREAU et al. 1991 ); Wolbachia is gradually eliminated from the testes as the males become older (BRESSAC and ROUSSET 1993 ).

Microinjections:
Microinjections were carried out as described in SANTAMARIA 1987 . Using a microcapillary needle (Boehringer Femtotips), cytoplasm was drawn from infected Wien 5 embryos and then injected into slightly dehydrated, uninfected NHaTC embryos. Isofemale lines were established after crossing the emerging females with NHaTC males.

PCR conditions:
Total DNA was extracted from the ovaries of individual females, following the method of KOCHER et al. 1989 , or from whole individuals, following the method of O'NEILL et al. 1992 . The Wolbachia 16S ribosomal subunit DNA sequence was amplified using specific primers 99F and 994R as described in O'NEILL et al. 1992 . Alternatively, Wolbachia-specific primers for the dnaA bacterial gene were used as described in BOURTZIS et al. 1994 .

Restriction fragment length polymorphism:
To further ensure that the infection found in ME lines resulted from the microinjection and not from a resurgence of the wHa Wolbachia infecting the NHa strain before tetracycline treatment, PCR amplification products of ME lines were subjected to MwoI digestion at 37° during 3 hr. The 16S rDNA sequence of wDm presents a MwoI restriction site, whereas the sequence of wHa does not, which allows discrimination of the two strains by RFLP.

Wolbachia load in individual males:
Total DNA was extracted by the STE method (O'NEILL et al. 1992 ) from individual males used in incompatible crosses. One microliter (out of 50) of the supernatant was used as a template for PCR analysis with Wolbachia-specific primers for the dnaA gene. The rest was dot blotted on a zeta probe blotting membrane and hybridized with a 480-bp PCR-derived dnaA fragment; the dnaA gene is known to be a single-copy gene in Wolbachia (BOURTZIS et al. 1994 ). The image analysis of the autoradiograms and the calculation of the Wolbachia densities, expressed as bacterial equivalents per male, have been described previously (BOURTZIS et al. 1996 ). Statistical analysis of bacterial densities was carried out using the Tukey test (SOKAL and ROHLF 1995 ) after square root transformation because group variances were proportional to the means.

Wolbachia load in sperm cysts:
Relative Wolbachia densities were estimated in very young males (a few hours old) by DAPI staining of sperm cysts, as described in BRESSAC and ROUSSET 1993 .

Wolbachia load in embryos:
Total DNA was extracted from groups of 50 embryos (0–2 hr old), and a dot blot assay was carried out following the procedure described above.

CI measurements (mass crosses):
Tests were performed at 25°. Fifteen 4- or 5-day-old virgin females were allowed to mate for 8 hr with 25 virgin 3-day-old males in a bottle with standard axenic medium. Flies were transferred for oviposition on fresh axenic medium. After 24 hr, the adults were discarded and the eggs were kept at 25° for at least another 24 hr before the hatch rate was estimated, generally on 200 eggs per line.

CI measurements (individual tests):
In the crosses using F₁ individuals or flies infected by wAu, wKi, or wMau, 3-day-old males were crossed with females aged 3–5 days. All matings were monitored, and inseminated females were placed individually for 48 hr on small petri dishes with fresh medium. Upon removal, the dishes were kept at 25° for at least 24 hr before the hatch rate was estimated on the total egg count. In the experiment using males for which global bacterial density was also assessed, the same protocol was used, except that virgin males were aged 1–2 days and females were aged 1–3 days. In addition, males were frozen immediately after mating for future DNA extraction.

Level of cytoplasmic incompatibility:
When reciprocal crosses with an uninfected strain were available, we used CI_corr, a corrected index of CI. The aim was to minimize the background noise caused by the natural mortality of the cross (which is unrelated to CI). This background noise is estimated by the compatible cross mortality (CCM), i.e., the mortality observed in the cross between standard uninfected males and infected females of the strain under test. CI_corr is then defined as the percentage of eggs that do not hatch among those that would have survived in the absence of CI. Then CI_corr(%) = [] x 100, where CI_obs is the percentage of unhatched eggs observed in the incompatible cross. The CI_corr of a given male was set at 0 whenever CI_obs < CCM. All statistical analyses were carried out after arcsine transformation. By definition, the CI_corr index does not apply in crosses where both directions are incompatible. In such cases, CI in a given direction of cross was simply estimated by the percentage of unhatched eggs.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Establishment of transinfected lines:
We injected 215 eggs of the uninfected D. simulans strain NHaTC with cytoplasm of eggs from the infected D. melanogaster Wien 5 isofemale line. Out of the 33 fertile females recovered, 17 gave a positive PCR signal with 16S rDNA Wolbachia-specific primers. Amplification products obtained from transinfected ME lines were of the wDm type and not of the wHa type, according to their RFLP pattern after MwoI digestion. A similar pattern was found in the MwoI- digested amplification product from Wien 5 females, whereas the MwoI-digested amplification product from a NHa control yielded the expected wHa RFLP pattern (data not shown). The infection found in ME lines is, therefore, a consequence of wDm transfer and not of imperfect elimination of wHa by tetracycline treatment of the NHa strain. As a first step, among the isofemale lines founded by PCR-positive G0 females, 5 were chosen at random for CI testing and were maintained in bottles under the same conditions as Wien 5 and all infected strains during the experiment, i.e., low-density bottles by crossing 20 virgin females to 25 young virgin males for 24 hr. The other 13 lines were initially maintained by mass transfer in vials. In G3, virgin males from the ME lines kept in bottles were tested for CI. At the same time, all 17 ME lines were tested again by PCR. It was then found that 2 of the 5 lines kept in bottles (ME 4 and ME 17) had lost the infection and that 9 out of the 13 lines maintained in vials by mass transfers had also lost the infection. All infected lines kept in vials were then transferred to bottles under the same conditions described above. No further loss of infection was found in the subsequent generations.

CI induced by wDm as a function of the host:
In G3, the five lines maintained in bottles were tested for the expression of cytoplasmic incompatibility. The results (Figure 1) show that all three infected lines expressed moderate to very high levels of CI, while the two uninfected ones did not. CI was also measured in G6. At this time, the infection was clearly established and the expression of CI was very high in all lines, with embryonic mortality >98% (Figure 1). These values are in sharp contrast with the moderate amount of CI expressed by the Wien 5 donor line in mass crosses with the uninfected strain MelO (23.5% CI on average, based on the observation of 1000 eggs). However, it has been shown that naturally infected strains of D. melanogaster can be polymorphic for the infection, with a proportion of the individuals uninfected (HOFFMANN et al. 1994 ; SOLIGNAC et al. 1994 ). Our mass crosses did not allow us to determine whether this effect could explain the difference of CI expression. A comparison was then made between ME lines and the Wien 5 line, using individual crosses for which all males were tested by PCR to exclude the uninfected ones from the analysis. The results (CI being expressed as CI_corr) are presented in Table 1. The ANOVA shows that the line effect is highly significant (F⁷₁₇₀ = 102.8; P < 0.001) , and a Tukey test indicates that this results from the CI_corr of Wien 5 being significantly lower than that of ME lines; the ME lines are not significantly different among themselves.

View larger version (64K): In this window In a new window Download PPT slide   Figure 1. Cytoplasmic incompatibility expressed by males of D. simulans isofemale lines transinfected with wDm. Shaded bars, infected lines; white bars, lines that were founded by a PCR-positive transinfected G0 female but found to be uninfected in G3. CI is estimated by the percentage of unhatched eggs in mass crosses with uninfected SimO females (117–200 eggs per cross). Error bars are 95% confidence intervals.

The Wien 5 line was further tested two generations later and was again found to induce a moderate amount of CI_corr (N = 30 crosses, 2612 eggs, CI_corr = 17.8 ± 2.9%). This figure is even significantly lower than the one presented in Table 1 (t = 2.21; 50 d.f.; P < 0.05). Such a phenomenon is in agreement with previous observations showing that CI can vary significantly in only a few generations in D. melanogaster (SOLIGNAC et al. 1994 ).

Our initial concern regarding a possible infection polymorphism in Wien 5 was not justified; the fraction of uninfected males found in Wien 5 (2 out of 54) was even lower than in the ME lines tested (17 out of 175), but not significantly so (² = 1.96; 1 d.f.; NS).

wDm bacteria load in males as a function of the host:
For a given Wolbachia strain, the level of CI has been correlated to the proportion of infected cysts in the testes in Drosophila (BRESSAC and ROUSSET 1993 ; SOLIGNAC et al. 1994 ). A similar correlation has also been found with the Wolbachia load in eggs laid by infected females in Drosophila (BOYLE et al. 1993 ) and in the parasitoid wasp N. vitripennis (BREEUWER and WERREN 1993 ). To establish whether the large difference of CI between ME lines and Wien 5 could result from a difference of Wolbachia load, we assayed bacteria density using different means. First, dot blots were carried out on total DNA extracts from the 178 infected males used in the individual incompatible crosses presented in Table 1. The results indicate that the overall Wolbachia load (expressed as the number of Wolbachia equivalents per individual) in Wien 5 males at the time of the CI tests was not lower than that of ME males (Table 1). Indeed, Wien 5 obtained the highest load score in this first experiment. The line factor is significant (F⁷₁₇₀ = 66.15; P < 0.001) , but a Tukey test reveals that the only significant differences are among ME lines: ME 14 vs. ME 29, ME 14 vs. ME 22, and ME 22 vs. ME 26.

A further dot blot assay was carried out two generations later using 30 Wien 5 males assayed for CI expression that exhibited very low CI_corr (see the section above). The bacterial load of Wien 5 flies during this second assay was significantly lower than that found previously (28.72 ± 3.83 x 10⁶ bacterial equivalents per male vs. 48.2 ± 6.48 x 10⁶: t = 2.59; 50 d.f.; P < 0.05). However, it was similar to that of some ME lines that exhibited a very high CI (see ME 22 and ME 29 in Table 1).

To ensure that the comparable Wolbachia loads we found in D. melanogaster and D. simulans males did not hide tissue-specific variations, we also assessed the specific load in the testes using a DAPI-staining technique. This time, the results reveal a striking difference: young Wien 5 males exhibit 10 times less infected cysts than ME males of the same age (Table 2). Because CI is caused by the presence of Wolbachia in male reproductive cells, this 10-fold difference alone would seem sufficient to explain the very low CI phenotype exhibited by Wien 5 males in contrast with the high CI exhibited by transinfected ME males. Such patterns of cyst infection are in agreement with previous results regarding the variants wHa and wRi in D. simulans (BRESSAC and ROUSSET 1993 ) and with wDm in D. melanogaster (SOLIGNAC et al. 1994 ).

We also assessed the Wolbachia load in early embryos (0–2 hr old) by dot blot. The results show that the bacterial loads of Wien 5 or ME embryos do not differ significantly (Table 2). Indeed, although the line effect is significant (F⁴₄₅ = 2.67; P < 0.05) , a Tukey test reveals two largely overlapping groups with the only significant difference found between Wien 5 and ME 8. This suggests that the lower load in the testes of Wien 5 males does not result from Wien 5 eggs initially harboring fewer Wolbachia than ME eggs, but would depend on events taking place during later developmental stages. The average number of Wolbachia per mature egg can be estimated from our results at more than 100,000 Wolbachia equivalents per egg (based on 2500 eggs, i.e., 50 groups of 50 eggs). This estimate is lower than but comparable to the 500,000 figure estimated elsewhere by DAPI staining on eggs from the DSR strain (T. KARR in BOURTZIS et al. 1996 ).

CI relationships between wDm and the natural CI-expressor Wolbachia strains of D. simulans:
Infected ME lines were crossed in both directions with the D. simulans strains R1A, NHa, R3A, and DSR. This allowed us to establish the CI relationships between wDm and the three CI-expressor Wolbachia strains (wHa, wNo, and wRi) found in natural populations of D. simulans. ME lines were bidirectionally incompatible with D. simulans strains infected by wHa or wNo or carrying the bi-infection wHa + wNo (Table 3). A strikingly different pattern appeared when we crossed ME lines with the DSR strain carrying the continental Wolbachia strain wRi (Table 4):

1. Males from the seven ME lines infected by wDm induced only a very limited mortality when crossed with wRi-infected females (average of 18.3% unhatched eggs). Although this value is low, it might represent some CI because it is significantly higher than what is observed in crosses involving males from the two uninfected ME lines (average of 7.5%; infected vs. uninfected: t = 2.49; 7 d.f.; P < 0.05). On the other hand, this mortality is comparable to that observed within the DSR strain (which ranges from 13 to 28.5%) or within infected ME lines (Table 4). These latter intraline mortalities might be caused in part by inbreeding: although CI could also play a role, intraline mortality is not significantly higher in the seven infected ME lines compared to the two uninfected ME lines (17.0 vs. 11.25%: t = 1.09; 7 d.f.; NS).

   
   

   View this table: In this window In a new window   Table 3. CI relationshipsa between wDm and the CI-expressors wHa and wNo

   
   

   View this table: In this window In a new window   Table 4. CI relationshipsa between the CI expressors wDm (ME lines) and wRi (DSR strain)

2. In contrast, DSR males clearly induced significant CI against ME females infected by wDm (Table 4; average of 59.8% mortality vs. average of 18.3% in the reciprocal cross: t = 12.64; 9 d.f.; P < 0.001). Yet, this level of CI (59.8%) is significantly reduced compared to the CI induced by DSR males against uninfected ME females (average of 95.5%: t = 15.72; 7 d.f.; P < 0.001).

We then tried to determine whether the partial and possibly unidirectional CI pattern between wRi and wDm could be explained by the number of Wolbachia in the testes of wRi-infected DSR males being too high for CI to be rescued by the number of Wolbachia present in eggs laid by wDm-infected ME females. The analysis of the results, shown in Table 2 (see above), is clearly in opposition with this simple quantitative hypothesis because (i) we did not find any significant difference in the percentage of infected cysts in young males between infected ME lines and the DSR strain, and (ii) the Wolbachia loads of the eggs laid by wDm- or wRi-infected females are not significantly different.

We also tried to rule out the possibility that apparent partial incompatibility in the direction of cross wRi-infected male x wDm-infected female was in fact caused by poor fertility in DSR males or, in general, by the genetic background of the host. The first possibility was suggested by the quite high intrastrain mortality in our DSR strain (up to 28.5% in some mass crosses), which led us to suspect that by inbreeding and drift, this laboratory strain might have fixed male sterility alleles. F₁ individuals were then obtained from both directions of cross between DSR and the infected line ME 29, using old nonvirgin males to minimize selection caused by CI. These individuals are genetically similar but are infected by wRi or wDm, depending on whether their mother was a DSR or a ME female. The individual CI tests carried out using these F₁ individuals are shown in Table 5. Reciprocal crosses confirmed the existence of CI between 3-day-old males infected by wRi (= F_1wRi males) and females infected by wDm (= F_1wDm females). The mean percentage of embryonic mortality induced by F_1wRi males crossed with F_1wDm females is 30.5%. This value is significantly higher than that obtained in the reciprocal cross (8.1%: t = 4.62; 21 d.f.; P < 0.001) and in the cross between F_1wDm individuals (8.6%: t = 4.40; 24 d.f.; P < 0.001). The CI induced by wRi in a F₁ background was significantly lower than that in a DSR background (Table 4; average = 59.8%: t = 5.31; 18 d.f.; P < 0.001). This might result from poor fertility in DSR males. Alternatively, the F₁ background might have depressed CI induction by wRi. However, this second possibility is very unlikely, considering that young F_1wRi males induce a near total CI (99.8% mortality) against uninfected females (Table 5).

The results in Table 5 also allow us to conclude that wDm-infected males are not able to induce CI against wRi-infected females. The egg mortalities found in the crosses F_1wDm male x F_1wRi female and F_1wDm x F_1wDm are not significantly different, regardless whether males are 3 days old (8.1 vs. 8.7%: t = 0.01; 28 d.f.; NS) or 7 days old (6.5 vs. 5.4%: t = 0.37; 24 d.f.; NS). Thus, although wRi and wDm are both able to induce very high levels of CI, they seem to be partially compatible in one direction of cross and fully compatible in the reciprocal direction, a pattern not described before.

The possibility remained that the partial compatibility between F_1wRi males and F_1wDm females was caused only by a quantitative difference, not in bacterial load directly, but in a factor produced by Wolbachia to induce (or rescue from) CI. The results shown in Table 5 suggest that this is not the case. First, it must be noted that 7-day-old (hereafter noted as "old") F_1wRi males induce significantly more egg mortality than old F_1wDm males when crossed with F_1wDm females (12.3 vs. 5.4%: t = 3.07; 26 d.f.; P < 0.01). This is not caused by a lower fertility in aging F_1wRi males because old F_1wRi males do not induce significantly more egg mortality than 3-day-old (hereafter noted "young") F_1wRi males when mated with F_1wRi females (3.4 vs. 4.4%: t = 0.60; 26 d.f.; NS). According to the hypothesis that wRi and wDm differ only by a quantitative factor, the weak but significant CI expressed by old F_1wRi males toward F_1wDm females would imply that the CI capability of old F_1wRi males is at least equivalent to that of young F_1wDm males. This hypothesis would then predict that the CI induced by old F_1wRi males should at least be equivalent to that of young F_1wDm males toward uninfected females. This is clearly not the case: when crossed with uninfected females, old F_1wRi males induce 70.4% egg mortality and young F_1wDm females induce 100% egg mortality (t = 6.96; 9 d.f.; P < 0.001). Accordingly, it would seem that wRi and wDm are qualitatively different as far as their CI mechanisms are concerned.

CI relationships between wDm and non-CI-expressor strains of Wolbachia:
Individual tests were set up between infected ME lines and the D. simulans stocks Kc9, DSW(Mau), and Coffs Harbour. This allowed us to establish the CI relationships between wDm and, respectively, Wolbachia strains wKi, wMau, and wAu. In all cases, males from ME lines infected by wDm were found to be completely incompatible with females infected by wKi, wMau, or wAu (Table 6). The latter finding contradicts the unpublished results reported elsewhere (see DISCUSSION).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Through transinfection experiments, we established D. simulans isofemale lines infected by wDm, the Wolbachia strain found in D. melanogaster. The transinfected lines allowed us to answer two questions, which we will consider in turn: (i) Will wDm behave in the new host as in D. melanogaster, i.e., being able to develop, be transmitted to the offspring by females, and induce only moderate amounts of CI through males? (ii) What will be the CI relationship between wDm and the CI-expressor strains naturally found in D. simulans?

Behavior of wDm in a new host:
The infection was initially detected by PCR in 17 out of 33 surviving G0 females. Seven transinfected ME lines were still infected in G3 and have remained infected ever since. Because no further loss of the infection was found in the next generations, spanning almost 2 yr in the laboratory, we suppose that the apparent massive infection loss (59%) in the first three generations postinjection could be attributed to an initial low density of the symbiont. An alternative hypothesis is that some G0 females were scored as PCR positive while Wolbachia had not reached their germline (although PCR was carried out on ovaries, contamination by Wolbachia from somatic tissues might have occurred during dissection). Therefore, after the early installation phase, it appears that the infection is efficiently maintained in the lines through maternal transmission.

The donor D. melanogaster line Wien 5 shows a moderate level of CI (18–32%), which is typical of many D. melanogaster strains (HOFFMANN 1988 ; BOYLE et al. 1993 ; BOURTZIS et al. 1994 , BOURTZIS et al. 1996 ; HOFFMANN et al. 1994 ; SOLIGNAC et al. 1994 ). In contrast, six generations postinfection, males from transinfected D. simulans ME lines induced >90% CI when crossed with uninfected females. Such high levels of CI are common in natural D. simulans infections (HOFFMANN et al. 1986 ; O'NEILL and KARR 1990 ; MONTCHAMP-MOREAU et al. 1991 ; MERCOT et al. 1995 ; ROUSSET and SOLIGNAC 1995 ), especially with the Wolbachia strain wRi.

Such a host-specific difference in the expression of CI has already been described in Drosophila. BOYLE et al. 1993 successfully injected the wRi strain into an uninfected D. melanogaster strain. The CI induced in transinfected lines was low (maximum of 35% egg mortality), i.e., typical of naturally infected strains of D. melanogaster, yet the Wolbachia caused a high level of CI when it was injected back into an uninfected D. simulans strain (75% egg mortality). These authors noted a correlation between the level of CI induced by males and Wolbachia density in the eggs of the lines under test. They were also able to increase the CI in one of their D. melanogaster transinfected lines through an artificial selection experiment, with a correlative increase in the density of Wolbachia in the eggs of the selected line. Our results show a different situation. As it could be expected, a dot blot assay on total DNA extracts revealed that males from ME transinfected lines harbored high densities of Wolbachia, comparable to the highest loads reported in D. simulans using this technique (i.e., for the wRi strain, see BOURTZIS et al. 1996 ). Yet, surprisingly, the same dot blot assay on total DNA extracts revealed that Wien 5 males, which induced low levels of CI, harbored a bacteria load comparable to that found in ME males. Moreover, Wien 5 embryos were found to harbor Wolbachia loads similar to those of ME embryos. These results suggest that the overall growth of wDm is not inhibited in D. melanogaster compared to its growth in transinfected D. simulans lines. However, we found that the percentage of infected cysts in the testes of young ME males was 10 times higher than that in Wien 5 males. This large difference would seem sufficient to explain the phenotypic gap between the high CI expressed by ME males and the moderate CI exhibited by Wien 5 males. The tissue-specific difference of Wolbachia load found in Wien 5 compared to ME lines (similar density in the eggs and the soma but very lower density in the testes) is of particular interest. Indeed, it is exactly the pattern expected if D. melanogaster had developed some control over the symbiont. This is because the relationship between the host and the Wolbachia depends strongly on the sex of the host. Because males can be sterilized by the Wolbachia present in their testes, any mutation in the genome of the host leading to a reduction of Wolbachia load in the male germline will be favored by natural selection. On the other hand, there will be no such strong selection to eliminate Wolbachia from the soma (unless they represent an important physiological cost) because these somatic bacteria appear to have no bearing on the sterilization of male reproductive cells. A buffering effect caused by Wolbachia present in somatic cells might explain why we could not detect any significant difference in Wolbachia load between Wien 5 and ME lines when using total DNA extract from whole males. In females, on the contrary, any mutation that would eliminate Wolbachia from the germline will be selected against because uninfected eggs laid by such females would die when fertilized by sperm from an infected male. Therefore, a specific decrease of Wolbachia load in the ovaries is not to be expected.

Previous observations indicated that Wolbachia infection of the testes was weak in D. melanogaster (SOLIGNAC et al. 1994 ), whereas it was high in D. simulans (BRESSAC and ROUSSET 1993 ). However, a comparison of CI levels could not be made rigorously because the Wolbachia strains studied were not the same in both species. Our transinfection experiment with a single Wolbachia strain confirms that weak infection of the testes is probably the key to explaining the low CI generally expressed by infected D. melanogaster males because ME lines demonstrate that wDm is clearly able to induce strong CI when present in the testes in sufficient numbers.

Relationships between wDm and the other Wolbachia strains known in D. simulans:
Our results reveal a bidirectional incompatibility between wDm and both CI-expressor strains wHa and wNo. This incompatibility is high (>80% embryonic mortality in both directions of cross), which is similar to the bidirectional incompatibilities already described in D. simulans (O'NEILL and KARR 1990 ; MERCOT et al. 1995 ; MERCOT and POINSOT 1998A ). A very high unidirectional CI was also found between wDm-infected males and females infected by the nonexpressor strains wAu, wKi, or wMau. This is also in line with previous unidirectional CI relationships described in D. simulans between expressor and nonexpressor strains (GIORDANO et al. 1995 ; ROUSSET and SOLIGNAC 1995 ; HOFFMANN et al. 1996 ; MERCOT and POINSOT 1998A , MERCOT and POINSOT 1998B ).

However, our crosses reveal a peculiar CI relationship between the two strong CI expressors wDm and wRi. We found that wDm-infected males from the ME lines were apparently able to induce, at best, very weak CI against wRi-infected females belonging to the DSR strain. In the reverse cross, DSR males were able to induce a clear CI phenotype (albeit at a significantly reduced level of 55–65% egg mortality compared to their usual 95% in crosses against other Wolbachia strains) against ME females. To ensure that this peculiar CI pattern could not be attributed to host factors, we generated F₁ individuals between ME and DSR. Our results confirm that significant CI is induced only in the cross between F_1wRi males and F_1wDm females, although its level is ~30% instead of the 55–65% found when using inbred DSR males. The apparent reduction of CI expressed by F_1wRi males compared to DSR males cannot be attributed to a depressing influence of the F₁ background on Wolbachia expression because young F_1wRi males exhibit complete CI (99.8% egg mortality) when crossed with uninfected females, and we assume it probably reflects the inbreeding of DSR.

In this particular experiment, CI was not detected in the cross F_1wDm males x F_1wRi females. Again, this is probably not explained by a depressing effect of the F₁ genomic background on CI expression by wDm: F_1wDm young males exhibit 100% CI when crossed with uninfected females. We would then assume that the weak mortality induced by ME males might also be caused by inbreeding in these isofemale lines rather than by CI. Therefore, our conclusion is that wDm is fully compatible with wRi in one direction of cross and partially compatible in the other direction.

The unidirectional CI patterns described so far in D. simulans between two Wolbachia strains (GIORDANO et al. 1995 ; ROUSSET and SOLIGNAC 1995 ; TURELLI and HOFFMANN 1995 ; HOFFMANN et al. 1996 ; MERCOT and POINSOT 1998A , MERCOT and POINSOT 1998B ) involved one strain that was totally unable to induce the CI phenotype (i.e., a nonexpressor Wolbachia strain) and one CI-expressor strain (an expressor strain). The pattern found here between wDm and wRi is unique in that both strains are very strong expressors in D. simulans, inducing typically 90% CI or more against uninfected females. This peculiar pattern had been obtained previously in an independent transinfection experiment where the only wDm-transinfected line recovered was unfortunately lost after a few generations (Table 7). It must be noted that both the donor D. melanogaster strain (yw^67C23) and the acceptor D. simulans strain (DSHT) were different from those used in the present work. Yet, early results using the DSR strain were qualitatively similar to those we report with ME lines.

Phylogeny and CI relationships:
Do the above CI relationships agree with molecular evidence bearing on Wolbachia phylogeny? Considering the most recent and discriminant Wolbachia phylogeny available, based on the very variable bacterial surface protein gene wsp (ZHOU et al. 1998 ), we can conclude that the total bidirectional CI between wDm and wHa or wNo is not surprising. Indeed, the divergence between wDm and these two strains is, respectively, 21.4% (i.e., 121 differences out of 565 bp) with wHa and 24.8% (140 differences out of 565 bp) with wNo. Similarly, the unidirectional CI induced by wDm toward wMau was to be expected (the wsp wMau sequence is identical to that of wNo). The unidirectional incompatibility of wDm toward wKi cannot be considered at the molecular level because the wsp sequence of wKi is not yet available.

On the other hand, the relationships of wDm toward the expressor strain wRi and toward the nonexpressor strain wAu are surprising when considering molecular data. We found, in particular, that wDm-infected males were completely incompatible with females infected by wAu (Table 6). This finding contradicts unpublished results reported by BOURTZIS et al. 1998 using the ME16 line described in our present work. From the molecular data presented in ZHOU et al. 1998 , where wMau and wNo share an identical wsp sequence, BOURTZIS et al. 1998 had been able to successfully predict the compatibility between these two strains. Using a similar line of argument, these authors had also assumed that females infected by the non-CI-expressor variant wAu would be compatible with males infected by wDm (a difference of only 1 of 565 bp between their wsp sequences). Preliminary unpublished results that seemed to support this hypothesis were mentioned (BOURTZIS et al. 1998 ). However, the results we present here (and which we have repeated with four ME lines including, of course, ME 16) clearly show that wDm-infected males are totally incompatible with wAu-infected females. Our finding that wDm-infected males are at the same time totally compatible with wRi-infected females is therefore surprising because the wsp sequences of wDm and wRi differ markedly (55 out of 565 bp, i.e., a 9.7% divergence). Although the discovery, cloning, and sequencing of the very variable wsp gene (BRAIG et al. 1998 ) provided a tool of unprecedented precision as far as Wolbachia phylogeny is concerned, it does not seem possible to infer accurately from these molecular data whether or not two Wolbachia strains will be compatible. The total incompatibility of wDm with wAu, despite almost identical wsp sequences, could be explained by a "loss-of-function" mutation of wAu in the mechanism responsible for CI rescue, which would not have been selected against because wAu is a nonexpressor strain. However, other nonexpressor strains such as wMau or wKi have clearly kept their CI rescue capability toward CI induced by wNo intact (BOURTZIS et al. 1998 ; MERCOT and POINSOT 1998B ).

Why is there only unidirectional CI between wDm and wRi?
Two kinds of hypotheses can be proposed to explain the partial and unidirectional CI pattern between two such strong CI expressors: (i) wRi and wDm might have identical CI mechanisms but differ in a quantitative way, with wRi being strong enough to induce partial CI against wDm but not vice versa, and (ii) wRi and wDm might differ qualitatively, but the CI rescue mechanism of wRi might have a broader spectrum of efficiency than that of wDm (i.e., the CI rescue function of wRi would remain functional even when faced with the divergent CI induction mechanism of wDm). In the reverse cross, the wDm rescue mechanism would be only partly efficient because of a lower versatility, hence partial CI in the cross wRi infected male x wDm infected female.

The simplest quantitative hypothesis would assume a difference in bacterial load. Indeed, the level of CI expression has been correlated to Wolbachia load several times (BOYLE et al. 1993 ; BREEUWER and WERREN 1993 ; BRESSAC and ROUSSET 1993 ; ROUSSET and DE STORDEUR 1994 ; SOLIGNAC et al. 1994 ; BOURTZIS et al. 1996 ). Following the load hypothesis, wDm and wRi would be identical as far as CI mechanisms are concerned, but wRi would reach higher densities than wDm. The partial CI observed would then result from the wDm load in the egg being too low to fully rescue the CI phenotype expressed by males heavily infected by wRi. In the other direction of cross, the wRi load in the egg would always be sufficient to rescue the CI phenotype expressed by males infected by wDm. However, none of our density measurements support this hypothesis. Indeed, the Wolbachia load in ME lines or in DSR is similar both in the eggs and in the testes (Table 2). Moreover, it does not seem that wRi would produce more of the unknown factors that allow the induction and rescue of CI. This is suggested by old F_1wRi males being still able to induce detectable CI against F_1wDm females, while their CI capability against uninfected females is considerably weaker than that of young F_1wDm males (Table 5). Finally, young or old F_1wRi males do not induce significantly more CI than F_1wDm males of the same age when mated with uninfected females (Table 5).

The asymmetrical CI pattern we find leads us to suggest that the CI mechanisms of wRi and wDm are qualitatively different, but have not diverged enough for full bidirectional incompatibility to appear (the alternative hypothesis being that their CI mechanisms are partially similar by chance, i.e., convergence). We might then be witnessing one of the evolutionary steps leading to two bidirectionally incompatible strains. The asymmetry of the CI relationship we describe is especially interesting with regard to the unknown biochemical mechanisms underlying CI. Indeed, it suggests that the bacterial function responsible for CI induction through the modification of the male reproductive cell can evolve separately from the bacterial function responsible for CI rescue in the infected embryo. Similar conclusions have already been drawn from the discovery of Wolbachia strains, which can fully rescue CI while being totally unable to induce it (BOURTZIS et al. 1998 ; MERCOT and POINSOT 1998B ; POINSOT and MERCOT 1998 ). It would seem that CI is not a one-protein business.

	FOOTNOTES

¹ These authors contributed equally to this work.

	ACKNOWLEDGMENTS

The authors thank ANDRONIKI NIRGIANAKI for excellent technical assistance, as well as SOPHIA ZABALOU and IOANNIS LIVADARAS for helping with early transinfection experiments with the DSHTM strain. The Wien 5 line originates from a sample collected in Vienna by WOLFGANG MILLER. This work was supported by a grant from the French Ministère de l'Enseignement Supérieur et de la Recherche (MESR) (ACC SV3 no. 9503017), by a John D. and Catherine T. MacArthur Foundation grant and by a grant from the Greek Secretariat for Research and Technology (PENED 15774). D.P. was supported by a Ph.D. grant from the French MESR. K.B. was partially supported by grants from the National Institute of Health (AI34355, AI40620) to SCOTT O'NEILL.

Manuscript received September 23, 1997; Accepted for publication June 8, 1998.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

BOURTZIS, K. and S. L. O'NEILL, 1998  Wolbachia infection and arthropod reproduction. Bioscience 48:287-293.

BOURTZIS, K., A. NIRGIANAKI, P. ONYANGO, and C. SAVAKIS, 1994  A prokaryotic dnaA sequence in Drosophila melanogaster: Wolbachia infection and cytoplasmic incompatibility among laboratory strains. Insect. Mol. Biol. 3:131-142[Medline].

BOURTZIS, K., A. NIRGIANAKI, G. MARKAKIS, and C. SAVAKIS, 1996  Wolbachia infection and cytoplasmic incompatibility in Drosophila species. Genetics 144:1063-1073[Abstract].

BOURTZIS, K., S. L. DOBSON, H. R. BRAIG, and S. L. O'NEILL, 1998  Rescuing Wolbachia have been overlooked. Nature 391:852-853[Medline].

BOYLE, L., S. L. O'NEILL, H. M. ROBERTSON, and T. L. KARR, 1993  Interspecific and intraspecific horizontal transfer of Wolbachia in Drosophila. Science 260:1796-1799[Abstract/Free Full Text].

BRAIG, H. R., W. ZHOU, S. L. DOBSON, and S. L. O'NEILL, 1998  Cloning and characterization of a gene encoding the major surface protein of the bacterial endosymbiont Wolbachia. J. Bacteriol. 180:2373-2378[Abstract/Free Full Text].

BREEUWER, J. A. J. and J. H. WERREN, 1993  Cytoplasmic incompatibility and bacterial density in Nasonia vitripennis. Genetics 135:565-574[Abstract].

BREEUWER, J. A. J., R. STOUTHAMER, S. M. BARNS, D. A. PELLETIER, and W. G. WEISBURG et al., 1992  Phylogeny of cytoplasmic incompatibility micro-organisms in the parasitoid wasp genus Nasonia (Hymenoptera: Pteromalidae) based on 16S ribosomal DNA sequences. Insect. Mol. Biol. 1:25-36[Medline].

BRESSAC, C. and F. ROUSSET, 1993  The reproductive incompatibility system in Drosophila simulans: DAPI-staining analysis of the Wolbachia symbionts in sperm cysts. J. Invert. Pathol. 63:226-230.

DAVID, J., 1962  A new medium for rearing Drosophila in axenic conditions. Dros. Inf. Serv. 93:28.

GIORDANO, R., S. L. O'NEILL, and H. M. ROBERTSON, 1995  Wolbachia infections and the expression of cytoplasmic incompatibility in Drosophila sechellia and D. mauritiana. Genetics 140:1307-1317[Abstract].

HOFFMANN, A. A., 1988  Partial cytoplasmic incompatibility between two Australian populations of Drosophila melanogaster. Entomol. Exp. Appl. 48:61-67.

HOFFMANN, A. A., and M. TURELLI, 1997 Cytoplasmic incompatibility in insects, pp. 42–80 in Influential Passengers: Inherited Microorganisms and Arthropod Reproduction, edited by S. L. O'NEILL, A. A. HOFFMANN and J. H. WERREN. Oxford University Press, Oxford.

HOFFMANN, A. A., M. TURELLI, and G. M. SIMMONS, 1986  Unidirectional incompatibility between populations of Drosophila simulans. Evolution 40:692-701.

HOFFMANN, A. A., D. J. CLANCY, and E. MERTON, 1994  Cytoplasmic incompatibility in Australian populations of Drosophila melanogaster. Genetics 136:993-999[Abstract].

HOFFMANN, A. A., D. J. CLANCY, and J. DUCAN, 1996  Naturally-occurring Wolbachia infection in Drosophila simulans that does not cause cytoplasmic incompatibility. Heredity 76:1-8.

HOLDEN, P. R., J. F. Y. BROOKFIELD, and P. JONES, 1993  Cloning and characterization of an ftsZ cognate from a bacterial symbiont of Drosophila melanogaster. Mol. Gen. Genet. 240:213-220[Medline].

KOCHER, T. D., W. K. THOMAS, A. MEYER, S. V. EDWARDS, and S. PÄÄBO et al., 1989  Dynamics of mitochondrial evolution in animals: amplification and sequencing with conserved primers. Proc. Natl. Acad. Sci. USA 86:6196-6200[Abstract/Free Full Text].

MERÇOT, H. and D. POINSOT, 1998a  Wolbachia transmission in a naturally bi-infected Drosophila simulans strain from New Caledonia. Entomol. Exp. Appl. 86:97-103.

MERÇOT, H. and D. POINSOT, 1998b  Rescuing Wolbachia have been overlooked and discovered on mount Kilimanjaro. Nature 391:853[Medline].

MERÇOT, H., B. LLORENTE, M. JACQUES, A. ATLAN, and C. MONTCHAMP-MOREAU, 1995  Variability within the Seychelles cytoplasmic incompatibility system in Drosophila simulans. Genetics 141:1015-1023[Abstract].

MONTCHAMP-MOREAU, C., J.-F. FERVEUR, and M. JACQUES, 1991  Geographic distribution and inheritance of three cytoplasmic incompatibility types in Drosophila simulans. Genetics 129:399-407[Abstract].

O'NEILL, S. L. and T. L. KARR, 1990  Bidirectional incompatibility between conspecific populations of Drosophila simulans. Nature 348:178-180[Medline].

O'NEILL, S. L., R. GIORDANO, A. M. E. COLBERT, T. L. KARR, and H. M. ROBERTSON, 1992  16S rRNA phylogenetic analysis of the bacterial endosymbionts associated with cytoplasmic incompatibility in insects. Proc. Natl. Acad. Sci. USA 89:2699-2702[Abstract/Free Full Text].

POINSOT, D. and H. MERÇOT, 1997  Wolbachia infection in Drosophila simulans: does the female host bear a physiological cost? Evolution 51:180-186.

POINSOT, D. and H. MERÇOT, 1998  Wolbachia of the third kind. Endocytobiosis Cell Res. 13(Suppl.):99.

RIGAUD, T., 1997 Inherited microorganisms and sex determination of arthropod hosts, pp. 81–101 in Influential Passengers: Inherited Microorganisms and Arthropod Reproduction, edited by S. L. O'NEILL, A. A. HOFFMANN and J. H. WERREN. Oxford University Press, Oxford.

ROUSSET, F. and E. DE STORDEUR, 1994  Properties of Drosophila simulans strains experimentally infected by different clones of the bacterium Wolbachia. Heredity 71:325-331.

ROUSSET, F. and M. SOLIGNAC, 1995  Evolution of single and double Wolbachia symbioses during speciation in the Drosophila simulans complex. Proc. Natl. Acad. Sci. USA 92:6389-6393[Abstract/Free Full Text].

ROUSSET, F., D. VAUTRIN, and M. SOLIGNAC, 1992  Molecular identification of Wolbachia, the agent of cytoplasmic incompatibility in Drosophila simulans, and variability in relation with host mitochondrial types. Proc. R. Soc. Lond. Ser. B 247:163-168[Medline].

SANTAMARIA P., 1987 Injecting eggs, pp. 159–173 in Drosophila: A Practical Approach, edited by D. B. ROBERTS. IRL Press, Oxford.

SINKINS, S. P., H. R. BRAIG, and S. L. O'NEILL, 1995  Wolbachia superinfections and the expression of cytoplasmic incompatibility. Proc. R. Soc. Lond. Ser. B 261:325-330[Medline].

SOKAL, R. R., and ROHLF, 1995 Biometry. The Principles and Practice of Statistics in Biological Research. W. H. Freeman, San Francisco.

SOLIGNAC, M., D. VAUTRIN, and F. ROUSSET, 1994  Widespread occurrence of the proteobacteria Wolbachia and partial cytoplasmic incompatibility in Drosophila melanogaster.. C. R. Acad. Sci. Paris 317:461-470.

STOUTHAMER, R., 1997 Wolbachia-induced parthenogenesis, pp. 102–124 in Influential Passengers: Inherited Microorganisms and Arthropod Reproduction, edited by S. L. O'NEILL, A. A. HOFFMANN and J. H. WERREN. Oxford University Press, Oxford.

STOUTHAMER, R., J. A. J. BREEUWER, R. F. LUCK, and J. H. WERREN, 1993  Molecular identification of microorganisms associated with parthenogenesis. Nature 361:66-68[Medline].

TURELLI, M. and A. A. HOFFMANN, 1995  Cytoplasmic incompatibility in Drosophila simulans: Dynamics and parameter estimates from natural populations. Genetics 140:1319-1338[Abstract].

WERREN, J. H., 1997  Biology of Wolbachia.. Annu. Rev. Entomol. 42:587-609[Medline].

WERREN, J. H., and S. L. O'NEILL, 1997 The evolution of heritable symbionts, pp. 1–41 in Influential Passengers: Inherited Microorganisms and Arthropod Reproduction, edited by S. L. O'NEILL, A. A. HOFFMANN and J. H. WERREN. Oxford University Press, Oxford.

WERREN, J. H., W. ZHANG, and L. R. GUO, 1995  Evolution and phylogeny of Wolbachia: reproductive parasites of arthropods. Proc. R. Soc. Lond. Ser. B 261:55-63[Medline].

ZHOU, W., F. ROUSSET, and S. L. O'NEILL, 1998  Phylogeny and PCR-based classification of Wolbachia strains using wsp gene sequences Proc. R. Soc. Lond. Ser. B 265:509-515.