The mechanism involved in achieving this interference is unknown. Possibilities include that the microsporidians use resources required for the development of malaria and that microsporidians block molecular targets used by malaria parasites to invade the mosquito's midgut. In this study we consider the possibility that a microsporidian infection primes the mosquito's immune system in a way that helps it to defend itself against a later infection by Plasmodium. Unlike the adaptive immune system of vertebrates, which possesses antigen-specific lymphocyte populations and memory cells capable of recognising and dealing with previously encountered infections, the innate immune system of invertebrates lacks such memory cells. Nevertheless, pre-exposure to infective agents primes the invertebrates' immune system in such a way that it is more effective at dealing with subsequent infection. This non-specific memory, conferred by the upregulation of generic defence mechanisms is described as immune priming and is involved in several invertebrate-parasite interactions (reviewed in [10]). For example, challenging the beetle Tenebrio molitor with a bacteria-derived elicitor (LPS) decreases the success of a subsequent infection by a fungus (Metarhizium anisopliae) [11]. Immune-priming can also affect the development of malaria parasites. Thus, in a series of experiments, an earlier challenge by bacteria reduced the prevalence of malaria infection in mosquitoes [12]–[17], and mosquitoes treated with antibiotics expressed their immune genes to a lesser degree and were more susceptible to Plasmodium infection than untreated ones [18]–[19].

We investigate the role of immune-priming in the interaction between microsporidians and malaria by exposing larvae of the mosquito Anopheles gambiae to the infective spores of the microsporidian Vavraia culicis and testing the adult females for their immune response and for the success of infection by the malaria parasite Plasmodium berghei. We pose three questions: (i) Does V. culicis stimulate the immune system? (ii) Does the parasite suppress the development of P. berghei? (iii) Do isolates of V. culicis have similar effects and, in particular, are their effects on the mosquito's immune response and the development of malaria correlated?

Of the 159 adult females exposed to the Plasmodium-infected blood meal, 119 (75%) took a full blood meal and survived the 10 days up to being dissected. There was no difference in feeding success between uninfected control mosquitoes (75.4%) and microsporidian-infected mosquitoes (74%) (χ²=0.03, p=0.860), but the proportion of mosquitoes that fed did depend on the microsporidian isolate that infected them (χ²=7.55, p=0.056), ranging from 64% to 87%. Of the 116 fed mosquitoes, 73 harboured oocysts 10 days later. Blocking had no effect on the feeding efficiency of the mosquitoes, except for treatment with isolate 3, where more mosquitoes from block one took a blood meal than block two (χ²=6.046, p=0.0139). Microsporidian-infected mosquitoes were less likely (58.5%) to harbour oocysts than microsporidian-uninfected controls (81.8%) (χ²=4.40, p=0.036) (Fig. 1a). The effect of the microsporidian isolate on the proportion of mosquitoes with oocysts ranged from 54.5% to 65.2%, but this difference was far from statistically significant (χ²=0.65, p=0.885). The mean number of oocysts in the 73 mosquitoes with at least 1 oocyst was 11.9, ranging from 4.8 to 20.7 among the five treatments (Fig. 1b; analysis of square root of oocyst number: F_5,67=7.82, p<0.001). Block had no effect on the number of oocysts harboured by mosquitoes in each treatment group. Mosquitoes infected by a microsporidian harboured an average of 8.9 oocysts; controls harboured 20.7 oocysts (F_1,69=30.93, p<0.001). The mean number of oocysts in mosquitoes infected by different microsporidian isolates ranged from 4.8 to 13.8 (F_1,47=5.45, p=0.003).

Of the 168 adult females inoculated with CM-25 Sephadex beads, 138 (80%) survived. Overall, the five treatments did not differ in their survival (χ²=4.33, p=0.363), though there was a slight difference in survival between blocks one and two for mosquitoes infected with isolate 4 (χ²=3.873, p=0.0491) and 1 (χ²=3.949, p=0.0469). In eleven mosquitoes we could not find the bead, so that we analyzed 127 beads. The degree to which a bead was melanised ranged from 35% in the uninfected control mosquitoes to 73% in the mosquitoes infected with one of the microsporidian isolates (Fig. 2). Infection by any microsporidian isolate led to a stronger melanisation response than in uninfected controls (F_1,123=11.47, p<0.001), and the four isolates differed in the degree to which they stimulated the melanisation response (F_3,95=2.78 p=0.004).

Thus, larval infection by V. culicis enhanced the melanisation response in the adults and decreased the likelihood and intensity of infection by P. berghei (Fig. 3). In contrast to this clear association, there was no correlation within the microsporidian-infected mosquitoes between the melanisation response induced by an isolate and either the likelihood (F_1,5=0.64, p=0.459) or the intensity of infection (F_1,5=1.63, p=0.258) by malaria. Note, however, that with only four isolates in the experiment, the power to detect these correlations was weak (likelihood of infection: power=0.1; intensity of infection: power=0.18).

That V. culicis enhanced the immune (and in particular the melanisation) response of adult A. gambiae is far from trivial. Indeed, as V. culicis depletes the resources of infected larvae, leaving them with fewer lipids, sugars and glycogen reserves than uninfected individuals [22], the body condition of emerging adults is worse and one should therefore expect their immune response to be weak. This, indeed, seems to be the case in Tenebrio molitor, where infection by an (unnamed) microsporidian does not enhance the immune response, and in particular does not stimulate or enhance immune system parameters associated with melanisation [23]. In contrast, and similar to our study, in a proteomic study of V. culicis in another mosquito, Aedes aegypti, the antibacterial response was stimulated up to 15 days after infection [24]. Thus, both the melanisation pathway and the anti-Plasmodium response are primed, suggesting that either both these pathways are used in attempts to clear V.culicis, or that V.culicis infection leads to a general priming of the entire immune system.

That V. culicis impeded the development of malaria in its mosquito vector corroborates several studies, in which microsporidian infection reduced the proportion of mosquitoes developing oocysts and the number of oocysts [4], [9] and decreased the density [4] and quality [7] of sporozoites. This suppression appears to be a general characteristic of microsporidian infection, as similar results were found in all studied systems, including non-human malaria (e.g. P. yoelii [4]) as well as P. falciparum developing in A. stephensi [3] or the main African vector, A. gambiae [6]. What this study adds is that isolates impede development to different degrees, most likely caused by differences in their genetic make-up. This divergence is reflected in the microsporidian's and mosquito's phenotypic traits, e.g. the intensity of microsporidian infection, the mortality of larvae and pupae and the longevity of adults (Lorenz and Koella, unpublished data). Further experiments will investigate whether these traits are correlated with the mosquito's immune response and its reaction to malaria.

Both responses to microsporidian infection — more effective melanisation of a Sephadex bead and less successful infection by malaria — differed among the four microsporidian isolates. While it would be interesting to evaluate whether this variation is due to genetic differences among the isolates or other differences; e.g. maternal effects, this is not within the scope of this study. More importantly, for each isolate, microsporidian infection enhanced the melanisation response (and perhaps other immune responses) and impeded development of malaria, which suggests immune-priming as a mechanism. Stronger support would have been a negative correlation between enhanced melanisation by microsporidian isolates and interference with malaria by the same isolate, which we did not observe. However; one should not over-interpret this lack of association. First, the power to detect any such correlation was low, as we used only four isolates and the variation of the likelihood of malaria infection among the mosquitoes infected by the four isolates was low. Second, it is a prioiri not clear that an association should exist. A certain level of immune capability, for example, may eliminate the Plasmodium parasite or reduce its numbers. Beyond this threshold any increase in immune capability may have little effect on malarial infection, yet still be evident in the increasing melanisation response.

Another possible mechanism for the interference of microsporidian infection with Plasmodium might be competition for healthy gut cells. As microsporidians are intracellular parasites a possible immune response would be to shut down the cell through apoptosis, thus killing the microsporidian within. Plasmodium parasites, on the other hand, must pass through the midgut to form oocysts. It is therefore conceivable, that at this stage in the development process Plasmodium parasites that enter microsporidian-infected cells may be killed through apoptosis. Alternatively, Plasmodium parasites may not be able to develop in the presence of microsporidians, because of competition for a resource required by both parasites. It is possible that a more definitive answer to this question could have been reached by including a second set of controls in the experiment. Immune priming was only achieved with concurrent microsporidian infection, therefore a non-infective treatment such as lipopolysaccharide injections could have been used to induce non-infectious immune priming, allowing the comparison between microsporidian presence and absence to be made. However, these two possibilities do not explain the increased melanisation response associated with less effective infection by malaria. Therefore, immune priming remains the most likely explanation.

Further support for our suggestion that microsporidians impede the development of malaria by priming the immune response comes from a previous study, showing that microsporidian infection leads to an enhanced antibacterial response [24]. This not only corroborates the idea mentioned in the introduction — that the melanisation response reflects more general immune priming — but may be a direct immunological mechanism, as antimicrobial responses appear to be largely responsible for the clearing of malaria parasites [21]. Thus, immune-priming appears to stimulate the mosquito's immune system and prepare it for subsequent Plasmodium challenge, thus making the infected mosquito partially refractory to malarial infection. A similar immune system boost occurs in mosquitoes following bacterial challenge. If mosquitoes are infected with bacteria before being fed on malarious blood, they are less likely to be infected my malaria than if they are not infected by the bacteria [12]–[17], whereas mosquitoes treated with antibiotics display lower expression of immune genes and are more susceptible to Plasmodium infection [18]–[19].