Maternal environment shapes the life history and susceptibility to malaria of Anopheles gambiae mosquitoes

Background It is becoming generally recognized that an individual's phenotype can be shaped not only by its own genotype and environmental experience, but also by its mother's environment and condition. Maternal environmental factors can influence mosquitoes' population dynamics and susceptibility to malaria, and therefore directly and indirectly the epidemiology of malaria. Methods In a full factorial experiment, the effects of two environmental stressors - food availability and infection with the microsporidian parasite Vavraia culicis - of female mosquitoes (Anopheles gambiae sensu stricto) on their offspring's development, survival and susceptibility to malaria were studied. Results The offspring of A. gambiae s.s. mothers infected with V. culicis developed into adults more slowly than those of uninfected mothers. This effect was exacerbated when mothers were reared on low food. Maternal food availability had no effect on the survival of their offspring up to emergence, and microsporidian infection decreased survival only slightly. Low food availability for mothers increased and V. culicis-infection of mothers decreased the likelihood that the offspring fed on malaria-infected blood harboured malaria parasites (but neither maternal treatment influenced their survival up to dissection). Conclusions Resource availability and infection with V. culicis of A. gambiae s.s. mosquitoes not only acted as direct environmental stimuli for changes in the success of one generation, but could also lead to maternal effects. Maternal V. culicis infection could make offspring more resistant and less likely to transmit malaria, thus enhancing the efficacy of the microsporidian for the biological control of malaria.


Background
The population dynamics of anopheline mosquitoes and the epidemiological dynamics of malaria have long been recognized to depend on environmental variables such as temperature [1][2][3] and its daily variability [4]. In addition to shaping the individuals exposed to them, environmental factors can also have longer-term effects on future generations. It is becoming generally accepted that an individual's phenotype can be influenced not only by its own genotype and environmental experience, but also by its mother's environment and condition [5][6][7]. Such maternal effects can be adaptive [8,9]. In Anopheles stephensi, for example, the daughters of lowfood mothers take up more blood and lay more eggs than the daughters of well-fed females, even if the daughters themselves experience the same environment [10]. Alternatively, maternal effects may reflect the mother's condition [11]. Food-deprived mothers may not be able to compensate for their poor environment and, therefore, have offspring of lower quality [12][13][14][15]. Thus, the maternal environment could influence the population dynamics of mosquitoes.
An individual's susceptibility to parasites may be influenced by its mother's environmental quality [16], or whether she herself was infected [17][18][19][20]. Thus, it is possible, and indeed likely, that malaria transmission could be changed by maternal effects, and that control strategies altering the mosquitoes' environment will affect the main parameters underlying the epidemiology of malaria -the mosquitoes' susceptibility to malaria, longevity and other life-history traits -not only directly, but also indirectly and over a longer period.
Mothers may be exposed to several types of environmental stress simultaneously, including poor resource availability, changing climate, competition with conspecifics, predation and parasitism. Food stress, for example, can exacerbate the harmful effects of infection [21][22][23], so it may also increase the importance of transgenerational effects. Potentially adaptive effects may also switch with the presence of a second stressor. The consequences of a combination of maternal stressors (such as limited food availability and infection) on the transfer of a female's experiences to her offspring could have important and unexpected consequences for the dynamics of the host and the parasite.
The microsporidian parasite Vavraia culicis [24] has been suggested as a potential late-acting control agent of anopheline malaria vectors that will impose little evolutionary pressure for resistance [25,26]. Mosquito larvae orally ingest V. culicis spores and become infected; infectivity rates range between 90-100%. Effects of the microsporidian on Anopheles gambiae sensu stricto (s.s.) include delayed pupation by 10%, decreased fecundity by 23% and reduced adult lifespan by 27% [25], and reduced susceptibility to malaria [27]. Despite the obvious reductions in fitness, resistance mechanisms of mosquitoes to microsporidians are not known [28].
In this study, maternal effects were evaluated for two environmental variables: the food regime available to larvae and infection by V. culicis. Similarly to microsporidian infection, poor nutrition of mosquito larvae increases development time [23,29] and decreases survival [25]. Here, a full factorial experiment where Anopheles gambiae s.s. mosquitoes were exposed to a low or high food regime with or without infection with V. culicis was conducted. It was investigated how these maternal experimental conditions influence the offspring with regard to susceptibility to malaria, which directly determines malaria transmission, and two life-history traits that influence the transmission indirectly by affecting the mosquitoes' population dynamics (larval survival and developmental time). Figure 1 shows the experimental design of this study. The mosquitoes originated from a genetically diverse colony established from A. gambiae s.s. caught in Yaoundé, Cameroon [30]. Mosquitoes were held at 26 ± 1°C and 70 ± 5% relative humidity with 12 hours light: dark cycles at Silwood Park Campus (Imperial College London, UK). Infection of mosquitoes with Plasmodium berghei took place in an insectary kept at 19 ± 1°C and 70 ± 5% relative humidity with 12 hours light: dark cycles. The Vavraia culicis floridensis spores were provided to us by J.J. Becnel (USDA Gainesville, USA). At Silwood Park Campus, the microsporidian parasite has been propagated in large groups of Aedes aegypti and A. gambiae s.s. mosquitoes. For consistency with earlier studies, the parasite is called V. culicis throughout the manuscript, whilst acknowledging the subspecies status of the Florida isolate [24].

Maternal generation
For the maternal generation, 300 larvae were individually reared in 2 ml de-ionized water in 12-well plates for each of four treatment groups: (1) no microsporidian infection, reared on high food (Tetramin fish food: Day 0 (hatching): 0.06 mg, Day 1: 0.12 mg, Day 2: 0.24 mg, Day 3: 0.36 mg, Day 4: 0.48 mg, Day 5 and following days: 0.6 mg per individual); (2) infection with 20,000 V. culicis spores, reared on high food; (3) uninfected, reared on low food (half of the high amount of food increasing in incremental steps to account for larval growth; see above); (4) microsporidian-infected, reared on low food (Figure 1a). Food and infection levels were chosen on the basis of past experience: The standard amount generally lets larvae develop within about 8 days from hatching to adult with low levels of juvenile mortality, whereas half of this amount puts larvae under increased nutritional and developmental stress (JCK, pers. comm.). An intermediate spore concentration affects the host adversely by delaying pupation and decreasing survival, but without killing mosquitoes too quickly [25] to ensure that enough mothers survived and reproduced despite the harmful fitness effects of the microsporidian. Larvae were fed every 24 hours and those to be infected were exposed to the microsporidian when they were two days old (Figure 1b). Larvae pupated after seven to nine days. Pupae were placed into individual 50 ml Falcon tubes to emerge. For each treatment group, 41-46% emerged as female mosquitoes. All females from one treatment group were placed into one cage the day after they emerged. A mix of 66 males from the two microsporidian-free treatments was added to each one of the four cages one or two days after female emergence (Figure 1c). Adult mosquitoes were provided daily with cotton soaked with 6% glucose solution and were allowed to mate for three days, when females were offered a blood meal on JCK's arm for 10 min. One day after the blood meal, the fully engorged female mosquitoes (N = 96) were placed into individual oviposition cups containing a dish lined with filter paper filled with 20 ml de-ionized water (Figure 1d). The eggs from each mother were placed into individual Petri dishes ( Figure 1e). Spores of V. culicis can be attached to the eggs of microsporidian-infected females [31], though the natural occurrence of the frequency and intensity of the trans-ovarial transmission route has not been quantified. Due to this uncertainty, and in order to disentangle physiological maternal effects from external ones, spores were eliminated by bleaching all eggs (irrespective of maternal infection) with 1% household bleach on the day they were laid [32].
After mothers had laid their eggs, they were killed and stored at -20°C. Subsequently, they were individually homogenised in 0.1 ml of de-ionised water, and V. culicis infection was confirmed by counting spores under a phase-contrast microscope (400 × magnification) using a haemacytometer. All 43 mothers from the infection treatments were infected with V. culicis spores.

Offspring generation
For each of the 96 mothers, the susceptibility of her offspring to malaria was investigated. To ensure that each mother was represented equally, one female offspring per mother for each measurement was used. To have a good chance of obtaining the required two females per mother, six larvae per mother were reared. If more than two females survived to adulthood, two were chosen haphazardly. Life-history traits (larval survival and age at pupation) were measured for each of the six larvae.
The larvae were reared individually on the high food level (see above; Figure 1f). Larval mortality and pupation were checked every 24 hours. Pupae were placed into individual 50 ml Falcon tubes and allowed to emerge (Figure 1g). Two cages for each of the four maternal treatments were prepared, each containing one of the two female offspring haphazardly selected per mother to measure two aspects of susceptibility: the parasite's development to 1) the oocyst stage and 2) the sporozoite stage. One of the cages later gave the mosquitoes for oocyst detection; the other the mosquitoes for sporozoite-detection (Figure 1h).
GFP-expressing transgenic P. berghei ookinetes (PbGFP CON strain; [33,34]) were produced by R. Armson at R.E. Sinden's laboratory at Imperial College London according to the laboratory's standard protocol. The ookinete culture was centrifuged at 500 g for 10 minutes at 19°C, the supernatant was removed and the ookinetes were counted under a microscope (400 × magnification) with a haemacytometer. Blood of uninfected mice was added to give a concentration of 800 ookinetes per μl. 400 μl of the mixture were injected into membrane feeders that had been preheated to 37 ± 1°C with a water bath and covered with Parafilm "M" (Pechiney Plastic Figure 1 Schematic representation of the experimental setup. For the parental generation, (a) 600 Anopheles gambiae (s.s.) larvae were reared individually under high and low food conditions, and (b) 300 larvae of each food treatment were exposed to Vavraia culicis spores. (c) After emergence, the females were placed into mating cages according to their treatment, given access to uninfected males, and allowed to blood-feed. (d) Fully engorged females (= mothers) were put into individual egg-laying cups. (e) To start the offspring generation, the eggs of each mother were bleached and placed into Petri dishes for hatching. (f) Six larvae of each mother were reared individually in 12-well plates. (g) The pupae were placed into individual tubes for emergence. (h) Two adult females of each family were moved to two cages (replicates) per treatment, and allowed to feed on malaria-infectious blood. (i) After bloodfeeding the mosquitoes were held individually in cups until dissection. (j) Mosquitoes of cages C1 were dissected for oocysts 12 days after blood-feeding; the mosquitoes of cages C2 were dissected for sporozoites 19 days after blood-feeding.
Packaging). The mosquitoes had access to glucose up to 24 hours before their blood meal. They were blood-fed with the malaria-infectious blood meal for one hour in darkness at 19 ± 1°C five to seven days after emerging. Each group of mosquitoes (two cages per maternal treatment) was provided with two membrane feeders to reduce the effect of possible differences among feeders. One day after the blood meal, fully engorged mosquitoes were placed into cups, which were kept at 19 ± 1°C and were supplied with cotton soaked with 6% glucose solution every 24 hours until dissection (Figure 1i). Twelve days after the blood meal, the mosquitoes from one of the cages from each treatment were dissected (N = 42) and their midguts fixed with 4% formaldehyde in PBS and mounted in antifade mounting fluid (Vectashield, Vector Laboratories Inc., Burlingame). Oocysts on midguts were counted under a fluorescent microscope (100 × magnification). Nineteen days after the blood meal, the mosquitoes from the second cages were dissected (N = 24), their salivary glands mounted on slides and their sporozoites counted under a microscope (400 × magnification). The dissection days were chosen to be 12 and 19 days after the blood meal in order to maximize the malaria parasite detection rate (E. Dawes, pers. comm.; Figure 1j).

Statistical analysis
Full models included maternal food level, maternal microsporidian infection status, their interaction and (when necessary) mother as a random factor. The final models were selected by comparing Akaike Information Criterion (AIC) values, with models with the lowest AIC values chosen as the minimum models [35]. The significance level α was set to 0.05.
The number of eggs produced by mothers from the four treatment groups was analysed with an analysis of variance. To test whether maternal treatment had any effects on the egg hatching rate, a generalized linear model (GLM) with binomial error structure and logit link function was performed.
As the offspring's age at pupation was restricted to seven to nine days after hatching (with most pupating after seven or eight days), mosquitoes with early (pupation on day 7) and late (pupation after day 7) pupation were compared. Generalized linear mixed models with binomial error structures and logit link functions were performed to test whether maternal treatments had effects on offspring survival and pupation. 'Mother' was set as the random factor. When testing for effects on pupation, offspring sex was included as a fixed factor in the analysis.
Offspring mosquitoes from each maternal treatment group were reared in two cages and the mosquitoes from each cage were dissected for malaria at different days (12 and 19 days after the blood meal). In order to determine whether maternal treatment had an effect on the proportion of offspring harbouring malaria parasites whilst accounting for survival until the different dissection days, the following two analyses were performed. First, a GLM with binomial error structure and logit link function was fitted to explain the proportion of mosquitoes surviving until the day of dissection. Cage, maternal food, microsporidian infection and their interactions were set as explanatory variables. Second, a GLM with binomial error structure and logit link function with cage, maternal food, microsporidian infection and their interactions was fitted to explain their effects on the proportion of mosquitoes infected with malaria parasites (oocysts or sporozoites).

Offspring malaria infection
The effects of maternal food and microsporidian infection on the offspring's susceptibility to malaria, measured as the probability that the mosquitoes harboured Plasmodium parasites after feeding on an infected blood meal, were investigated. As neither maternal food (χ 1 2 = 0.9, p = 0.35) nor maternal V. culicis infection (χ 1 2 = 0.1, p = 0.92) had an effect on the survival of offspring (N = 121) up to their day of dissection, and the two cages had similar levels of malaria infection (χ 1 2 = 0.1, p = 0.75), the use of the two cages as replicates to compare malaria-infected with malaria-uninfected mosquitoes was justified.

Discussion
The two types of maternal stress -low food and infection by V. culicis -considered here had different effects on traits that influence the epidemiology of malaria. With both stressors, A. gambiae s.s. had offspring that took longer to pupate. Offspring were more likely to be infected by P. berghei if their mothers were reared on low-food than on high food, whereas offspring from microsporidian-infected mothers were less, rather than more, susceptible to infection by malaria.

Offspring development
Mothers reared in stressful environments often produce smaller and less viable eggs [13,39,40]; examples show that maternal stress leads to weaker offspring are rove beetles (Tachyporus hypnorum, [13]), Hawaiian fruit flies (Drosophila grimshawi, [12]) and locusts (Schistocerca gregaria, [14]). Infection by the microsporidian V. culicis corroborates this pattern: infected Ae. aegypti mosquitoes lay smaller eggs (S. Fellous, pers. comm), and here, microsporidian-infected mothers laid fewer eggs that were less likely to hatch. The larvae that did hatch were less likely to survive as juveniles and pupated later than the offspring of uninfected mothers. This effect of microsporidian infection was exacerbated when mothers were reared on low food ( Figure 2). However, daughters of badly nourished A. stephensi females take larger blood meals and lay more eggs than those of well-fed mothers [10], suggesting that offspring could compensate for expected decreased lifespan in response to poor maternal environments [8,41]. Despite clear maternal effects, it is thus not yet clear how maternal stress would affect the long-term population dynamics of the mosquito.

Offspring malaria infection
Whereas low-food Daphnia magna females produce offspring that are less susceptible to bacterial infection than well-fed females [16], daughters of low-food A. gambiae s.s. in this experiment were more likely to be infected with malaria. A possible explanation for this result is that stressed mothers cannot compensate for their poor environment, and therefore invest fewer resources in their offspring. This would reduce the offspring's ability to mount costly immune responses [14] and lead to greater susceptibility to infection. In contrast, maternal infection by V. culicis led to offspring that were less likely to harbour malaria ( Figure  3). This could be due to trans-generational immunepriming, which occurs in shrimps (Penaeus monodon, [42]), water fleas (D. magna, [43]), bumblebees (Bombus terrestris, [19,44,45]) and yellow mealworm beetles (Tenebrio molitor, [17,46,47]). Although immune-priming is usually specific to the parasite species or strain, immune responses of A. gambiae s.s. against malaria can be activated with bacterial challenges [48][49][50][51][52][53][54][55]. Microsporidian infection impedes the development of malaria [27,[56][57][58][59][60], suggesting that an unspecific component of immune-activation [61] may be extended to trans-generational immune priming of the offspring. Alternatively, the pattern could reflect a trade-off between growth and resistance to infection [62][63][64][65]; mothers could subtly change the development of their offspring so that offspring invest more in their immune system at a cost to their growth. For example, daughters of stressed D. magna mothers increase their growth rates, but are slightly more susceptible to bacterial infection [15]. Trans-generational immune priming in T. molitor, on the other hand, resulted in offspring with longer larval development times [17,47]. In this study, offspring from V. culicis-infected A. gambiae s.s. mothers also took longer to develop into adult mosquitoes, but were less likely to be infected with malaria, thus corroborating a previously demonstrated link between age at pupation and immuno-competence [63]. This association could not explicitly be tested, as all the mosquitoes were pooled after emergence irrespective of their age at pupation.
A lack of sufficient replication limited the power of this study design; each treatment group and sampling of malaria infection was represented by only one mosquito cage. Therefore, maternal treatment, time of dissection and cage were not completely independent variables. However, the use of the two cages as replicates for maternal treatment was still justified as maternal treatments had no effect on offspring survival (maternal food: p = 0.35; maternal infection: p = 0.92), and malaria infection was similar in both cages (p = 0.75). These results add to the confidence that the apparent differences between traits of offspring from different maternal treatment groups represent real differences and transgenerational effects rather than artefacts of cage-effects.

Conclusions
In summary, resource availability and infection with V. culicis of A. gambiae s.s. mosquitoes not only act as direct environmental stimuli for changes in the success of one generation, but can also lead to maternal effects. As A. gambiae s.s. is a major vector of malaria in sub-Saharan Africa, and microsporidia, such as V. culicis and other biopesticides are being considered as potential control agents against this deadly disease [25,66,67], these results may have important social implications. Co-infection of mosquitoes with microsporidia and malaria directly inhibits malaria development [27,59,60]. Maternal V. culicis infection could also make offspring more resistant and less likely to transmit malaria, thus further enhancing the efficacy of the microsporidian as a control agent.