Research articlesMonitoring long-term evolutionary changes following Wolbachia introduction into a novel host: the Wolbachia popcorn infection in Drosophila simulans
Abstract
Wolbachia may act as a biological control agent for pest management; in particular, the Wolbachia variant wMelPop (popcorn) shortens host longevity and may be useful for dengue suppression. However, long-term changes in the host and Wolbachia genomes can alter Wolbachia spread and/or host effects that suppress disease. Here, we investigate the phenotypic effects of wMelPop in a non-native host, Drosophila simulans, following artificial transinfection approximately 200 generations ago. Long-term rearing and maintenance of the bacteria were at 19°C in the original I-102 genetic background that was transinfected with the popcorn strain. The bacteria were then introgressed into three massbred backgrounds, and tetracycline was used to create uninfected sublines. The effect of wMelPop on longevity in this species appears to have changed; longevity was no longer reduced at 25°C in some nuclear backgrounds, reflecting different geographical origin, selection or drift, although the reduction was still evident for flies held at 30°C. Wolbachia influenced productivity and viability, and development time in some host backgrounds. These findings suggest that long-term attenuation of Wolbachia effects may compromise the effectiveness of this bacterium in pest control. They also emphasize the importance of host nuclear background on Wolbachia phenotypic effects.
1. Introduction
Wolbachia is an intracellular bacterium that has been shown to cause numerous phenotypic effects in its hosts and may potentially be useful as a biological control agent (Sinkins & O'Neill 2000; Brownstein et al. 2003; Rasgon et al. 2003; Rasgon & Scott 2004). One Wolbachia strain that holds particular promise in this respect is popcorn (wMelPop), which has been found to over-replicate within the cells of its host, Drosophila melanogaster, causing a reduction in host longevity (Min & Benzer 1997). Using this infection, the age structure of target host populations might be modified to prevent disease transmission (Brownstein et al. 2003; McMeniman et al. 2009). Apart from influencing longevity, wMelPop also causes other effects. In D. melanogaster, this infection causes cytoplasmic incompatibility (CI) involving a reduction in egg hatch rate following mating between infected males and uninfected females, and also delays egg-to-adult development (Reynolds et al. 2003).
Following identification in D. melanogaster, the popcorn infection was successfully transferred to D. simulans (McGraw et al. 2001) and more recently to Aedes aegypti (McMeniman et al. 2009). In both D. simulans and A. aegypti, the infection reduces the longevity of its host substantially while inducing CI. In D. simulans, wMelPop effects on longevity and other traits appear to have changed over time. McGraw et al. (2002) initially found that while the wMelPop infection caused strong CI, there was also a significant reduction in reproductive fitness owing to a decrease in fecundity and egg hatch rates. However, these effects appeared to attenuate with time; by generation 22 after transinfection, hatch rate in compatible crosses returned to control levels, although there was still a significant effect of the infection on fecundity. McGraw et al. (2002) also showed that egg production in the wMelPop-infected line declined after only 5 days.
For Wolbachia infections, theory predicts a shift towards a decrease in the virulence of the bacteria in its host, while perfect maternal transmission and high levels of CI are expected to be maintained unless there is a cost (Turelli 1994). Attenuation of virulence of the wMelPop infection was recently identified in the native D. melanogaster host, through indirect selection on host longevity (Carrington et al. 2009). This shift in longevity was mainly due to host nuclear background, although interactions with host and bacterial genomes were also identified. There are several other cases where the evolution of phenotypic effects generated by Wolbachia symbionts have been documented. These include the evolution of a fecundity benefit in Californian D. simulans infected with the Wolbachia Riverside (wRi) strain (Weeks et al. 2007), as well as the suppression of the male-killing phenotype induced by the wBol1 strain in the butterfly Hypolimnas bolina (Hornett et al. 2006). Both of these examples involve evolutionary changes that have occurred in natural populations.
In this study, we examine the Drosophila simulans and Wolbachia popcorn (wMelPop) symbiosis several years after transinfection (McGraw et al. 2001), to test for potential attenuation of Wolbachia effects compared with tetracycline-treated controls. We estimate this time to be greater than 200 generations, during which time the infection was maintained in the original background and under laboratory conditions. Longevity effects, levels of CI and several life-history traits are considered. Potential changes in phenotypic effects are assessed in three outcrossed nuclear backgrounds compared with treated controls and, where possible, we further contrast these results to those previously described several generations after transinfection (McGraw et al. 2002). Results are discussed within the context of effective management strategies for biological control of pest species using this Wolbachia strain.
2. Material and methods
The strains used in this experiment are described first, followed by a description of the life-history trait assays.
(a) Strains and maintenance
The original D. simulans I-102 strain, transinfected with Wolbachia popcorn by McGraw et al. (2001), was maintained under laboratory conditions for several years, at 19°C. We then generated three outcrossed infected populations in mid-2007. The wMelPop infection from the I-102-infected females was backcrossed for two generations to males from D. simulans massbred populations originating from Sydney (latitude 33°50′ S) and Sorrell (42°46′ S) in Australia, and Irvine (33°39′ N) in CA, USA. These massbred populations were originally set up from 10 females and 10 males from each of 10 isofemale lines, established from field females collected in April 2007. Backcrossing was conducted within five generations of the massbreds being created. These populations represent genetically distinct backgrounds that will be influenced by both geographical origin and subsequent changes occurring in the laboratory environment (owing to drift and selection). Once generated, all populations were reared at 25°C on cornmeal–yeast–dextrose media under a rapid turnover regime and continuous light, unless otherwise stated (see experimental descriptions for productivity and the second longevity assay).
At the time all populations were placed at 25°C, the planned temperature for the experiments, it was noted that there were difficulties with rearing the infected I-102 line. Multiple attempts were made to rear I-102 at the higher temperature; however, with each generation we observed a significant reduction in adult progeny, and it was then impossible to reach a second generation (data not shown). The inability to rear this line was investigated in the productivity assay (see below), with results clearly confirming that the infected I-102 line has a severe productivity deficit at the higher temperature. We therefore excluded this line from the remainder of the study.
The experiments were undertaken in two stages. The initial stage was conducted seven generations after backcrossing and included all assays except for the second longevity assay, which was conducted in a second stage (21 generations after the first). Immediately prior to each stage, the infected populations were treated with tetracycline to cure the Wolbachia infection. Uninfected subpopulations from each of the massbred populations were created by treating larvae for two generations with tetracycline (0.033% v/v) added to the media. At least 200 flies per background were used to create each subpopulation, which was then cultured for at least two generations without tetracycline, to control for any negative effects of antibiotic treatment and allow natural gut flora to regenerate (Ballard & Melvin 2007). Additionally, infected and treated populations from the same background location were then reciprocally crossed en masse for one generation (using a minimum of 75 males and virgin females), to re-homogenize the host background and control for any differences that might have developed between treated and untreated massbred populations. Because Wolbachia is maternally transmitted, reciprocally crossing infected and treated lines results in progeny that have maintained each infection status. However, reciprocal crossing did not control for nuclear background effects owing to sex linkage. To overcome the incompatibility created in the cross between infected males and uninfected females, we used older males (approx. 6–8 days old), whose ability to induce CI had declined. The resulting uninfected populations are henceforth referred to as I-102-T, Sydney-T, Sorrell-T and Irvine-T. Males of approximately the same age were also used in the reciprocal crosses, and the resulting infected populations are referred to without the -T suffix. At the end of the treatment, flies were confirmed uninfected by PCR (see below).
To generate flies for the experiments, larvae were reared at a low density during development. To control for density, inseminated adult females were allowed to lay eggs onto plastic spoons containing approximately 2 ml of a treacle-based medium, with a light covering of a yeast suspension to enhance oviposition. From these spoons, 40 eggs were picked and placed into a fresh vial with 10 ml of cornmeal medium, where they were allowed to develop. Once eclosed, flies were collected for use in experiments.
(b) Infection status
To test for the presence of Wolbachia, PCR was used to amplify Wolbachia-specific DNA extracted from single flies. To extract DNA, individual females were ground in 5 µl of Proteinase K (4 mg ml−1), and 150 µl of 5 per cent Chelex was subsequently added (McColl et al. 1996). Samples were heated at 37°C for 30 min, and then at 95°C for 4 min to inactivate the Proteinase K. Prior to PCR, the extract was spun at 13 000 r.p.m. for 2 min and 1 µl of the supernatant was added to a 15 µl PCR reaction mix. PCR was conducted with Wolbachia-specific primers (wsp81F and wsp691R; Zhou et al. 1998), as well as primers that amplify a Drosophila-specific gene, Suppressor of Sable (Su(s)), as positive controls for each reaction (Voelker et al. 1991). Infection status was confirmed in lines prior to all experiments.
(c) Strain confirmation
The wsp gene of the Wolbachia in each of the infected outcrossed populations (Irvine, Sorrell and Sydney) was sequenced to verify that the strain of Wolbachia was in fact popcorn. DNA was extracted for sequencing using a phenol–chloroform method from single females (from generation 22), and the wsp gene was amplified. We used exonuclease-Shrimp Albumin Protein (ExoSAP-IT, USB Corporation) for PCR cleanup. DNA sequencing was conducted at the UCDNA Sequencing Facility (University of California at Davis, CA, USA). All strains tested showed 100 per cent sequence identity to the published Wolbachia pipientis wMelPop outer surface protein gene precursor (GenBank accession number AF338346.1).
(d) Productivity and fecundity
Owing to the difficulties in obtaining enough progeny from the I-102-infected line to undertake planned experiments at 25°C, we investigated total productivity for all populations. As we were unable to prepare for the experiment at 25°C, we reared all of the flies at 19°C, so as to control for rearing temperature between populations. When the flies had eclosed, they were immediately placed at 25°C for testing. Pairs of virgin males and females (three per population) were set up in vials, which were changed every 24 h. Productivity was assessed every day for 9 days. Productivity was calculated by counting the number of adult flies that eclosed from each vial at each age. Sexes were counted separately but, as there was no effect of infection/background on sex ratio, only data pooled across sexes are considered further. The results of the assay confirmed the observations and we therefore excluded both I-102 and its treated counterpart, I-102-T, from all subsequent assays.
The fecundity of the remaining populations was assessed over a 10-day period when females were 12–36 h post-eclosion. All flies were reared and tested at 25°C. The number of eggs laid by individual females each day was counted as a measure of the fecundity for each population, with 20 replicate females individually assayed for each. Every female was paired with a male, and males that died were replaced during the assay. Only females that survived the entire 10-day period were included in the fecundity assessment. Eggs were laid on a treacle medium held on a plastic spoon coated with a live yeast suspension to trigger egg laying. Spoons were replaced every 24 h and the eggs counted.
(e) Development time and viability
This was assessed using eggs obtained from adults 3–5 days post-eclosion. After a 2 h laying period, eggs were picked and placed in vials with 12 ml of medium (20 eggs per vial). Egg-to-adult development time at 25°C was scored separately for the sexes every 6 h, starting at 168 h (7 days) after the middle of the laying period. There were 20 replicates for each of six populations (infected and uninfected of each of Sydney, Sorrell and Irvine). Viability was also estimated from this assay.
(f) Longevity
Two longevity assays were conducted. The first assay tested whether the popcorn longevity phenotype was present some 200+ generations after transinfection. All three outcrossed backgrounds were tested at 25°C. The second assay was conducted approximately 20 generations after the first and compared the effect of temperature (30°C versus 25°C) on longevity effects of the popcorn infection, using flies from two of the outcrossed backgrounds (Sydney and Irvine).
The first assay was set up with flies from the development time assay (described above). Longevity was assessed for all combinations of host background, infection status, sex and mating effects. Flies tested were either virgin or once mated, because previous studies have shown a trade-off between mating/reproduction and longevity (Chapman et al. 1993; Sgrò & Partridge 1999). To obtain once-mated individuals, single pair matings were set up when flies were 2 days post-eclosion. Mating pairs were identified over several hours, and sexes were separated after mating was complete. Longevity was scored every 2 days (within an hour of the same time every day), when flies remaining in a vial were placed into a fresh vial. Between three and five replicates for each combination were tested, with a replicate consisting of 10 individuals per vial containing 6 ml of medium.
In the second longevity assay, the effect of the Wolbachia infection at 25°C or 30°C in two outcrossed populations (Sydney and Irvine) was assessed. Owing to the extended period between longevity assays, infected flies were re-treated with tetracycline for two generations. The newly cured and infected populations were reciprocally crossed prior to commencement of the assay as described above. Because male sterility in Drosophila is induced at high temperatures (David et al. 1971), we were unable to rear flies at 30°C; hence, all flies, including those for the 30°C treatment, were reared at 25°C. Newly eclosed virgin flies were collected every 6 h, grouped into replicate vials and placed at the appropriate temperature for subsequent ageing. Again, mortality was scored every 2 days, when the remaining flies were transferred to new vials with fresh medium. For each temperature, there were three to five replicates of each combination of factors (background, infection status and sex), containing 10 individuals. Mating effects were not considered in this second assay because they were not detected in the first assay (see below).
(g) Cytoplasmic incompatibility
CI is expressed when infected males mate with uninfected females. Egg hatch is reduced due to an incompatibility between the sperm and egg. The hatch rate for compatible and incompatible crosses was measured for all three outcrossed backgrounds. The four combinations resulting from crosses between infected (I) and uninfected (U) flies (U♀ × U♂, U♀ × I♂, I♀ × U♂ and I♀ × I♂) were considered, as well as effects of male age on CI.
Males were tested at 1, 3, 5, 7 and 9 days post-eclosion, or until male stocks were depleted (as occurred in the Sorrell background). Background and incompatibility effects were assessed independently for each age class. For ages 1–7 days, mean hatch rates were calculated from 12 replicates per population-cross combination. For age 9 days, 6–12 replicates were used. For each replicate, a virgin female (3–5 days old) was paired with a virgin male of the required age. Males had been aged in a vial with no more than 15 other males. Pairs were placed onto treacle-based medium with a yeast suspension and allowed to mate for 24 h. Males were removed after 24 h, while the female continued to lay eggs for a further 24 h on new medium. At the end of the 48 h laying period, eggs were allowed 24 h to hatch. The total number of eggs laid and the proportion that hatched were recorded. Replicates with fewer than 10 eggs were not included in analyses of hatch rate.
(h) Maternal inheritance
The efficiency of Wolbachia transmission by PCR-positive infected females was assessed for the three outcrossed backgrounds. Males and females were mated in pairs at generation 20 (three pairs per population) and allowed to lay eggs for 7 days. Parents were then frozen at −80°C to check infection status. Between 19 and 25 female progeny from each cross were collected and tested for infection status with the DNA extraction and PCR method described above.
(i) Analysis
All analyses were conducted using SPSS (v. 15). Longevity data were analysed with Kaplan–Meier (KM) log rank tests, to compare rates of mortality of populations and assess effects of host nuclear background, sex and mating status. Mean longevities for each population were also analysed using analysis of variance (ANOVA). Mean longevities were normally distributed based on Kolmogorov–Smirnov tests. For life-history measurements and CI induction at each age class, ANOVAs were carried out to assess the effect of host background and infection status on traits (again, traits were untransformed because they met conditions of normality). The effect of age on CI was independently tested across all backgrounds using a Kruskal–Wallis test. Age and sex were also included as factors in some assays (see below), and viability was included as a covariate when analysing development time, testing for differences in elevation.
3. Results
(a) Productivity and fecundity
For productivity, we initially included I-102 and the three outcrossed backgrounds, but then also re-analysed the results without I-102 because of the low productivity of this background, particularly when flies were infected (figure 1). In both analyses, infection, age and background all affected productivity, and there were also significant interaction terms (table 1). The presence of the infection led to a reduction in productivity in all backgrounds, while productivity in the Sydney background was lower than in the other outcrossed backgrounds (figure 1b–d). Productivity initially increased with age, but then decreased again (figure 1). The significant interaction between background and infection reflected a greater reduction in productivity caused by the infection in the Sorrell and Sydney backgrounds compared to the Irvine background. The interaction between infection and age was due to a relatively larger difference between the infected and uninfected populations at intermediate ages, while the background and age interaction reflected a smaller difference between populations at intermediate ages (figure 1).
| productivity including I-102 |
productivity excluding I-102 |
fecundity excluding I-102 |
||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| terms | d.f. | MS | F | p | d.f. | MS | F | p | d.f. | MS | F | p |
| infection | 1 | 137 959 | 1313.23 | <0.001 | 1 | 110 298 | 892.24 | <0.001 | 1 | 5176.3 | 31.79 | <0.001 |
| background | 3 | 18 568 | 176.75 | <0.001 | 2 | 5779 | 46.75 | <0.001 | 2 | 48.8 | 0.30 | 0.741 |
| age | 8 | 3575 | 34.04 | <0.001 | 8 | 3559 | 28.79 | <0.001 | 9 | 684.8 | 4.21 | <0.001 |
| infection × background | 1 | 1650 | 15.71 | <0.001 | 2 | 2257 | 18.26 | <0.001 | 2 | 162.2 | 1 | 0.370 |
| age × infection | 8 | 1977 | 18.82 | <0.001 | 8 | 1448 | 11.72 | <0.001 | 9 | 306.8 | 1.88 | 0.052 |
| age × background | 24 | 628 | 5.98 | <0.001 | 16 | 362 | 2.93 | <0.001 | 18 | 76.7 | 0.47 | 0.970 |
| age × infection × background | 24 | 323 | 3.08 | <0.001 | 16 | 298 | 2.41 | 0.002 | 16 | 105.6 | 0.65 | 0.844 |
| error | 360 | 105 | — | — | 270 | 124 | — | — | 585 | 162.81 | — | — |
Figure 1. Mean productivity for all populations for each day. (a) I-102; (b) Irvine; (c) Sorrell; (d) Sydney. White bars represent uninfected populations, while grey bars represent infected populations. Error bars indicate s.e. ± 1 of the mean.
In the fecundity assay, the ANOVA indicated significant effects of infection and age, but not background (table 1). The presence of the infection resulted in fewer eggs. Mean eggs per female per day (±s.e.) was 33.26 ± 0.67 for uninfected populations and 23.86 ± 0.79 for their infected counterparts.
(b) Development time and viability
The ANOVA indicated a significant effect of infection status on viability (table 2). Uninfected populations had a higher viability (85.3% ± 2.2) than infected populations (65.5% ± 2.0). A significant background effect (due to geographical origin, selection and/or drift) was also found, with Sydney having the lowest mean viability (64.2% ± 3.5), in contrast to Sorrell (84.3% ± 1.9) and Irvine (78.1% ± 2.6).
| viability |
development time, female |
development time, male |
||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| terms | d.f. | MS | F | p | d.f. | MS | F | p | d.f. | MS | F | p |
| infection | 1 | 1.15 | 55.19 | <0.001 | 1 | 320.99 | 7.57 | 0.006 | 1 | 617.37 | 13.57 | <0.001 |
| background | 2 | 0.41 | 19.94 | <0.001 | 2 | 482.48 | 11.38 | <0.001 | 2 | 487.42 | 10.72 | <0.001 |
| infection × background | 2 | 0.03 | 1.31 | 0.273 | 2 | 663.20 | 15.64 | <0.001 | 2 | 937.78 | 20.62 | <0.001 |
| viability | — | — | — | — | 1 | 1506.83 | 35.53 | <0.001 | 1 | 1795.75 | 39.48 | <0.001 |
| error | 113 | 0.0207 | — | — | 906 | 42.41 | — | — | 914 | 45.49 | — | — |
For development time, we included viability as a covariate in the analysis. Females and males were analysed separately because they had different mean development times; however, for both sexes the effects of infection and background were similar. We found that both host nuclear background and infection influenced development time (table 2). Irvine flies eclosed earlier than those from the other backgrounds, while Wolbachia-infected populations tended to develop faster than their uninfected counterparts (figure 2). An interaction between infection status and background was also observed; this reflected the slower development time of infected populations from Sydney when compared to the uninfected population, but an opposing pattern in the Sorrell background (figure 2). Viability influenced development time; vials with a longer development time tended to come from vials with a higher viability.
Figure 2. Mean number of hours to eclosion from egg laying for each background. Infected populations are represented by grey bars, uninfected populations are shown in white bars. Error bars indicate s.e. ± 1 of the mean.
(c) Longevity
In the first longevity assay, there was no overall effect of the infection on survival curves according to KM analyses (χ2 = 2.62, d.f. = 1, p = 0.105), and also no effect of background (χ2 = 3, d.f. = 2, p = 0.233). Within backgrounds, however, there were differences between the infected and treated populations (Irvine: χ2 = 4.67, d.f. = 1, p = 0.031; Sydney: χ2 = 3.87, d.f. = 1, p = 0.049; Sorrell: χ2 = 6.62, d.f. = 1, p = 0.010). Infected flies from both Sorrell and Irvine had a reduced longevity compared with uninfected populations, as expected based on published data. However, infected flies from Sydney had an increased longevity relative to Sydney-T (figure 3a–c). No overall effect of mating was observed on the longevity curves for all populations (χ2 = 1.81, d.f. = 1, p = 0.178). However, an overall effect of the infection was seen only in virgin flies (χ2 = 7.07, d.f. = 1, p = 0.007), reflecting a reduction in longevity in infected populations. This effect was not seen in mated flies (χ2 = 0.189, d.f. = 1, p = 0.660). Sex influenced longevity, with males living a shorter time than females in both infected and uninfected treatment groups (χ2 = 7.52, d.f. = 1, p = 0.006 and χ2 = 17.46, d.f. = 1, p < 0.001, respectively).
Figure 3. Rates of mortality for each of the three outcrossed backgrounds assessed at 25°C for the first longevity assay. (a) Irvine; (b) Sorrell; (c) Sydney. For each graph, infected flies are represented by black diamonds, and uninfected flies by white squares. All differences between populations were significant to p < 0.05.
Mean longevity for each population was also assessed using ANOVAs. Neither background nor infection status influenced longevity; however, there was an interaction between these two variables (table 3). This reflected the decrease in mean longevity in infected populations from Irvine and Sorrell, compared with an increase in the infected population originating from Sydney (table 4).
| longevity 1 |
longevity 2 |
|||||||
|---|---|---|---|---|---|---|---|---|
| terms | d.f. | MS | F | p | d.f. | MS | F | p |
| infection | 1 | 180.44 | 1.20 | 0.274 | 1 | 49.90 | 0.488 | 0.485 |
| background | 2 | 40.37 | 0.27 | 0.765 | 1 | 389.91 | 3.8133 | 0.051 |
| temperature | — | — | — | — | 1 | 63 670.03 | 622.684 | <0.001 |
| sex | 1 | 2488.57 | 16.54 | <0.001 | 1 | 3621.58 | 35.419 | <0.001 |
| infection × background | 2 | 1407.58 | 9.36 | <0.001 | 1 | 125.79 | 1.2302 | 0.2678 |
| background × temperature | — | — | — | — | 1 | 55.81 | 0.5458 | 0.4603 |
| infection × temperature | — | — | — | — | 1 | 529.62 | 5.1797 | 0.0232 |
| infection × background × temperature | — | — | — | — | 1 | 7.93 | 0.0775 | 0.7808 |
| infection × sex | 1 | 36.66 | 0.24 | 0.622 | 1 | 4.54 | 0.0444 | 0.8332 |
| mating | 1 | 0.59 | 0 | 0.950 | — | — | — | — |
| error | 1097 | 150.42 | — | — | 653 | 102.25 | — | — |
| longevity 1 | longevity 2 |
|||
|---|---|---|---|---|
| background | infection | 25°C | 25°C | 30°C |
| Sydney | infected | 55.9 (52.2, 61.5) | 44.4 (32.7, 53.0) | 22.2 (19.6, 25.3) |
| uninfected | 52.7 (24.8, 62.0) | 44.5 (33.6, 52.5) | 25.4 (20.4, 29.0) | |
| Irvine | infected | 52.5 (25.6, 60.2) | 43.1 (34.6, 47.1) | 22.2 (16.2, 26.8) |
| uninfected | 56.4 (46.4, 66.1) | 40.7 (33.6, 49.2) | 23.6 (21.0, 31.2) | |
| Sorrell | infected | 53.9 (47.8, 60.1) | not done | not done |
| uninfected | 56.2 (47.4, 64.2) | not done | not done | |
In the second longevity assay, there were significant differences between the 25°C and 30°C temperature treatments in both the Sydney and Irvine populations (χ2 = 253.68, d.f. = 1, p < 0.001 and χ2 = 246.74, d.f. = 1, p < 0.001, respectively). The Wolbachia infection influenced longevity at 30°C (χ2 = 7.666, d.f. = 1, p = 0.006), but not at 25°C (χ2 = 0.144, d.f. = 1, p = 0.704). The variation at 30°C was due to a reduction in longevity in the infected Sydney background (χ2 = 13.444, d.f. = 1, p < 0.001), as no difference was seen between infected and uninfected populations from Irvine (χ2 = 0.640, d.f. = 1, p = 0.431). No effect of Wolbachia was identified in the 25°C treatment in either the Sydney or Irvine populations (χ2 = 0.021, d.f. = 1, p = 0.884 and χ2 = 0.059, d.f. = 1, p = 0.807, respectively).
Mean longevity, assessed by ANOVA, showed a highly significant effect of temperature and sex (table 3). There was also an interaction between infection and temperature. Females lived longer than males (table 4), and the effect of the infection on longevity changed for the Sydney population between 25°C and 30°C (figure 4).
Figure 4. Rates of mortality for (a) Sydney 30°C, (b) Irvine 30°C, (c) Sydney 25°C and (d) Irvine 25°C treatments in the second longevity assay. Infected flies are represented by black diamonds, and uninfected populations with white circles.
(d) Cytoplasmic incompatibility
The level of CI induced by virgin males aged 1, 3, 5, 7 and 9 days was measured against young virgin females. An overall effect of age was found to influence hatch rate, identified using a Kruskal–Wallis test (χ2 = 118.89, d.f. = 4, p < 0.001). CI induction decreased as male age increased. As expected, the CI-inducing cross (U♀ × I♂) had the lowest mean hatch rate, which ranged between 0.27 and 0.35 across all backgrounds. Hatch rate was significantly reduced in incompatible crosses at ages 1–5 days, while at ages 7 days and 9 days there was no difference between the compatible and incompatible crosses (figure 5). No effect of background was apparent at any age (table 5), and no interactions were observed between background and compatibility, except at age 7 days.
Figure 5. Mean hatch rate of eggs produced during matings of 3-day old females with males of different ages. Grey bars represent incompatible crosses, white bars compatible crosses. Error bars show s.e. ± 1 of the mean.
| background |
compatibility |
background × compatibility |
error |
|||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| age (days) | d.f. | MS | F | p | d.f. | MS | F | p | d.f. | MS | F | p | d.f. | MS |
| 1 | 2 | 0.01 | 0.26 | 0.773 | 1 | 5.86 | 111.2 | <0.001 | 2 | 0.04 | 0.72 | 0.491 | 132 | 0.053 |
| 3 | 2 | 0.06 | 1.29 | 0.279 | 1 | 3.56 | 79.4 | <0.001 | 2 | 0.07 | 1.51 | 0.225 | 132 | 0.045 |
| 5 | 2 | 0.17 | 2.02 | 0.136 | 1 | 2.94 | 35.35 | <0.001 | 2 | 0.10 | 1.15 | 0.321 | 135 | 0.083 |
| 7 | 2 | 0.03 | 0.25 | 0.778 | 1 | 0.03 | 0.26 | 0.608 | 2 | 0.27 | 2.67 | 0.073 | 132 | 0.102 |
| 9 | 2 | 0.22 | 2.04 | 0.137 | 1 | 0.05 | 0.42 | 0.518 | 1 | 0 | 0 | 0.955 | 85 | 0.109 |
(e) Maternal transmission
Transmission of Wolbachia from single females (in triplicate) into approximately 20 female progeny of the successive generation was tested for each infected population. Perfect transmission was observed for all females tested. The upper 95 per cent binomial confidence limit for this result was 0.016; the maternal leakage rate was therefore likely to be 0 but certainly less than 0.016.
4. Discussion
The original characterization of the D. simulans/wMelPop symbiosis (McGraw et al. 2002) was the point of reference for this study. However, comparisons have turned out to be difficult because of the weakness of the original infected I-102 strain. This line was unable to be reared at 25°C, and was therefore not included in the longevity assessment. Consequently, we have been unable to test whether the longevity phenotype is still present in this background, a limitation in testing for long-term evolutionary change. Nevertheless, longevity-shortening effects appear to have attenuated in the outbred backgrounds, which are likely to be used in any releases with popcorn for the purposes of disease suppression. The infected individuals from the McGraw et al. (2002) study previously had mean longevities of 20.8 days for females and 18.3 days for males. These values were significantly lower than values for the uninfected controls. In addition, Min & Benzer (1997) found that, in the original D. melanogaster host, control lines exhibited 100 per cent mortality after 28 days. In contrast, the mean longevities of the infected outcrossed populations in the present study (at the same temperature) were 53.8 and 51.4 days for females and males, respectively (combined across the two assays), as opposed to uninfected population values of 54.2 and 51.5 days.
Even so, the infection still seems to influence longevity, although not always in the predicted direction and in a way that is strongly dependent on nuclear background. The first longevity assay showed that the popcorn-infected Sydney population had a greater mean longevity than the uninfected population at 25°C, contrary to expectation. This suggests that the presence of the infection provided a benefit to the host, compared with the situation in the Irvine and Sorrell populations, where the infected populations showed reduced longevity relative to their uninfected counterparts. However, in the second assay at 30°C, popcorn negatively influenced the longevity of infected flies from Sydney, suggesting that longevity effects had a strong temperature dependency. Overall, our longevity assays suggest interactions between host, infection and temperature in D. simulans. Interactions between host background and Wolbachia on longevity have been noted in previous studies; both Reynolds et al. (2003) and Carrington et al. (2009) found that host nuclear background influenced the longevity of popcorn-infected D. melanogaster populations. Host-dependent effects of Wolbachia have previously been noted in other systems, including the woodlouse Porcellionides pruinosus (Michel-Salzat et al. 2001) and butterfly Hypolimnas bolina (Hornett et al. 2006).
The present results also indicate that the popcorn infection continues to induce a cost to the reproductive fitness of its D. simulans hosts, consistent with McGraw et al. (2002). In the productivity assay, infected flies produced a significantly lower number of progeny in all backgrounds and at every age tested. The most severe reduction in productivity within populations was in the original infected I-102 strain. Multiple efforts to rear this strain at 25°C for more than one generation failed, owing to inadequate progeny numbers being produced for the successive generation. However, it was possible to rear this strain at lower temperatures (19 and 16°C). In contrast, I-102-T could be reared at all temperatures, suggesting a negative interaction between the host nuclear background, temperature and infection. For the outbred backgrounds, rearing was possible at a range of temperatures.
One of the most important traits induced by Wolbachia in this system is CI, as it is necessary for the intra-population spread of the bacterium. We found that CI induced in D. simulans remained strong, but was age-dependent, with CI effects diminishing within 7 days. This decrease in CI with increasing male age is consistent with results obtained with the popcorn infection in D. melanogaster (Reynolds et al. 2003), where the hatch rate of incompatible crosses is indistinguishable from that of compatible crosses after 7 days. Age effects on CI have also been observed for wRi in D. simulans (Hoffmann et al. 1986, 1990; Weeks et al. 2007) and wMel in D. melanogaster (Reynolds & Hoffmann 2002). It remains to be seen what levels of CI might be present in the field as opposed to laboratory populations for wMelPop. However, under field conditions, CI is typically weaker than in the laboratory, at least in wRi (Hoffmann et al. 1990; Turelli & Hoffmann 1995) and wMel (Hoffmann et al. 1998). Therefore, the CI induction of wMelPop in the field will dictate the likelihood of popcorn spreading in natural populations when coupled with fitness effects and maternal transmission efficiency (Turelli 1994). Maternal transmission does appear to be high and perhaps perfect in the laboratory, although again lower rates of transmission can be found in the field, at least in wRi (Turelli & Hoffmann 1995).
While there are several issues that this laboratory-based study has identified as potentially affecting the success of a control strategy based around longevity reduction by Wolbachia, attenuation in field populations may follow a different trajectory. Conditions in the field will differ from those in the laboratory owing to variable turnover, non-discrete generations and fluctuating environmental conditions. Further, the phenotypic changes seen in the wMelPop infection in D. simulans may also differ from those in infected mosquitoes like A. aegypti. Tissue tropism and density of the bacteria may differ between host species, and assessing the wMelPop infection in D. simulans may not necessarily reflect effects in an alternate host. For instance, the dynamics of wMelPop will differ between flies and A. aegypti, given the evidence of perfect maternal transmission and complete CI not influenced by male age in A. aegypti (McMeniman et al. 2009), in contrast to D. simulans. This may reflect bacterial density and tropism effects; it has recently been shown that higher bacterial density increases the penetrance of a male-killing Wolbachia strain in D. innubila (Unckless et al. 2009). Finally, even if attenuation of the longevity phenotype occurs, it may not influence the effectiveness of a control programme. Given that it has taken around 200 generations for this attenuation to occur, transmission of dengue will probably be eradicated before any attenuation if all other traits associated with the infection are highly penetrant (CI to spread the Wolbachia into the population, maternal transmission to ensure a high infection frequency and sufficient longevity reduction).
Although the fate of the popcorn infection in natural D. simulans populations is likely to depend on the genetic background of the host, our experiments had limitations in predicting the precise trajectories on particular backgrounds. Background effects were noted not only for longevity but also for development time, with Sydney and Sorrell populations showing large differences between infected and treated populations in opposing directions, while there was no difference between the population from Irvine. It is possible that segregating genetic polymorphisms and drift between the generation of the infected/uninfected lines and the experiments (four generations) could have influenced the results, even though we performed a reciprocal cross immediately prior to the experiments to obviate these effects. Prediction about host phenotypic effects or the population dynamics of the infection in a particular background will therefore depend on thorough testing of popcorn in replicated populations with the target background of the host.
In conclusion, we have shown that the wMelPop infection is maintained successfully within D. simulans populations, and also continues to reduce longevity in different host backgrounds. However, host nuclear background influences the Wolbachia-induced phenotypes including longevity and fitness costs/benefits associated with the infection. Phenotypic effects may depend on temperature, as evident for the longevity phenotype in the Sydney background. This makes it pertinent to characterize Wolbachia effects in target backgrounds when developing models for its spread. Attenuation of popcorn's longevity phenotype at 25°C appears to have occurred over the past few years, emphasizing the need for ongoing monitoring of any phenotypic effects. At present, it is not known if these issues will present problems for controlling dengue transmission in A. aegypti mosquitoes. The effect of the infection on longevity of the mosquitoes appears strong (McMeniman et al. 2009), and this also applies when the infection is tested in outbred backgrounds (K. Richardson 2009, personal communication). Nevertheless, attenuation might occur in this system through nuclear changes (Carrington et al. 2009) and interaction effects. This suggests the need for detailed investigations into target populations and their responses to the bacterial infection if long-term predictions are to be developed.
Acknowledgements
L.B.C. would like to thank Clare Doig for help undertaking the experiments. The authors would also like to thank three anonymous reviewers for constructive comments on this manuscript. This research was supported by a grant from the Foundation for the National Institutes of Health through the Grand Challenges in Global Health Initiative of the Bill and Melinda Gates Foundation, and funding from the Australian Research Council via their Special Research Center Program. This work was carried out while A.A.H. held an Australian Federation Fellowship and A.R.W. held an Australian Research Fellowship.
