Plasmodium berghei is one of the most commonly studied Plasmodium species, particularly for elucidating the interactions between the parasites and their hosts 12345. The natural mammalian host of P. berghei is the African tree rat Grammomys surdaster and its natural mosquito vector is Anopheles dureni 67. P. berghei is a species of choice for studies employing genetic manipulations due to the relative ease of parasite transfection and the function of many parasite genes has already been investigated in this species 8. Recently, a method has been developed in P. berghei that will now permit the inactivation of essential genes specifically at pre-erythrocytic stages of the parasite 9. Furthermore, the use of fluorescently labeled parasites has given unprecedented insights into the behaviour of these parasites within living insects and mice 10111213. Although there are differences between the rodent and human-infecting Plasmodium parasites at the genomic, antigenic and cellular level 1415, it is nonetheless clear that the rodent parasites are useful for elucidating the molecular basis of the core biology of Plasmodium, which often cannot be addressed with human parasites.

Plasmodium spp that infect humans are transmitted by a range of different mosquito species. While one of the main malaria vectors in Asia is Anopheles stephensi, the main vector in Africa is Anopheles gambiae. Both mosquito species are commonly used in laboratory experiments to study host-parasite interactions 16. Far less common are studies using the main South American malaria vector, Anopheles albimanus 1. This might be partly due to the facts that P. berghei has been reported to not infect A. albimanus 17 and that Plasmodium yoelii sporozoites generated in A. albimanus have been described as being non-infectious to the rodent host 18. Since A. gambiae, A. stephensi and A. albimanus are all amenable to genetic modification 192021, an A. albimanus – P. berghei system would be an interesting addition to the existing laboratory models. In addition, establishment of such a model, or similar models for other parasite species, would be valuable for host-vector-parasite interaction studies in countries where importation of A. gambiae and A. stephensi is not allowed. Here, using a fluorescent parasite line, A. albimanus mosquitoes were screened to check whether it was permissive for P. berghei development. It was found that P. berghei was able to infect A. albimanus and to develop into infectious sporozoites within this mosquito species.

Mosquitoes were screened from day 10 after the infectious blood meal for the presence of fluorescent oocysts in the midgut. In A. stephensi and A. gambiae a fluorescent signal in the abdomen could readily be detected in intact mosquitoes (Figure 1a and data not shown). Dissection of the mosquitoes revealed that the fluorescence originating from even a single oocyst can be detected in intact mosquitoes. Detection was facilitated if the mosquitoes were given a blood meal prior to observation, which expanded the abdomen and thus decreased its opacity. In the vast majority of A. albimanus, there was no fluorescent signal. However, a small number of mosquitoes showed a weak fluorescence signal in their abdomen. To investigate if indeed a small number of A. albimanus were allowing the development of P. berghei, female mosquitoes were first observed intact and then dissected at various times (see methods) after the infecting blood meal. Less than 4% of A. albimanus were infected with one or more (maximum 30) oocysts, although typically all females in a cage had taken a potentially infectious blood meal (Table 1, Figure 1). In control experiments, between 50 and 95% of A. stephensi were infected with several dozen to over one hundred oocysts, and the salivary glands of infected A. stephensi usually contained over 10,000 sporozoites per gland.

Close examination by epi-fluorescence microscopy of immobilized A. albimanus allowed the occasional detection of sporozoites within the haemolymph. Only a few A. albimanus salivary glands were found that contained sporozoites, usually in very small numbers (less than 10). In contrast, in A. stephensi mosquitoes, numerous sporozoites were readily detected in the haemolymph from as early as day 11 after infection (Figure 2a, b). Time-lapse analysis of immobilized A. stephensi mosquitoes showed that the sporozoites were passively flowing within the haemolymph (Figure 2c and Additional file 1 and 2, see also 23). Sporozoites were found in any part of the body that was bathed by the haemolymph including the palps, the labium, the antennae, the legs, wings, abdomen and thorax (Figure 2d, Additional file 1 and 2). Examination of several thousand haemolymph sporozoites did not yield any evidence of active gliding motility by these sporozoites. It is, therefore, concluded that, at least in the A. stephensi – P. berghei system, sporozoites are passively transported by the haemolymph and eventually attach to the salivary glands and that very few P. berghei sporozoites are found in A. albimanus.

It had been previously shown that the efficiency of P. falciparum, P. yoelii and P. berghei development in A. stephensi can be increased by adding para-aminobenzoic acid (PABA) to the sugar water prior to the infectious blood meal 2425. It is also known that, as the presence of bacteria in the midgut of mosquitoes inhibits the infectivity of P. falciparum to A. gambiae, A. stephensi and A. albimanus, Plasmodium infection rates can be increased by adding antibiotics 252627. Therefore, in attempts to increase the efficiency of P. berghei development in A. albimanus, the sucrose solution was supplemented with 0.5 g/l PABA and 0.1 g/l penicillin/streptomycin (Pen/Strep). In this case, a higher number of A. albimanus females were infected by P. berghei and a higher number of oocysts per mosquito gut was found (Table 1). While in the absence of PABA and Pen/Strep, rarely more than one mosquito in a cage of 200 females was seen that had more than five oocysts, infection rates exceeding 20% were regularly achieved in the presence of supplements (Table 1). Of these infected mosquitoes, over 49% had more than five oocysts, while 22% showed more than 20 (Table 1).

A closer examination revealed that P. berghei oocysts developed over a longer period of time (between 5 and 10 days delay) in A. albimanus than in A. stephensi midguts (Figure 1), which resulted in a delayed detection of fluorescent sporozoites in the haemolymph of A. albimanus. In virtually all A. albimanus mosquitoes with at least five oocysts some sporozoites could be detected in their haemolymph (Figure 2d). Nonetheless, the number of sporozoites within the salivary glands of these mosquitoes was usually very low. Occasionally, however, salivary glands were found that were infected by several hundred and sometimes several thousand sporozoites (Figure 3a).

To analyse P. berghei sporozoites in the A. albimanus salivary glands, the latter were incubated in cell culture medium containing foetal calf serum and visualized by time-lapse microscopy. Sporozoites within the glands were able to move (Figure 3a). Their main movement pattern was the "back-and-forth" type of motility previously observed for P. berghei sporozoites in A. stephensi salivary glands, in the absence or presence of serum 11. Next, it was investigated if the sporozoites would be able to glide on a solid substrate, a prerequisite for infectivity to the mammalian host 28. When incubated in medium containing foetal calf serum, P. berghei sporozoites isolated from infected A. albimanus glands moved on glass slides in a manner indistinguishable from P. berghei sporozoites isolated from infected A. stephensi glands (Figure 3b).

Next, the infectivity to the mammalian host of sporozoites isolated from A. albimanus salivary glands was investigated. Injection of 2,000 such sporozoites into the skin of mice caused red blood cell infection as determined by blood smear analysis (Figure 3c). This showed that P. berghei sporozoites, isolated from infected A. albimanus, were capable of invading both mosquito and mammalian tissues and to differentiate into red blood cell invading forms. However, in two separate experiments, intra-dermal injection of 20,000 sporozoites obtained from the haemolymph of A. albimanus mosquitoes failed to induce infection. Mice remained uninfected, which confirms that Plasmodium sporozoites undergo a maturation process in the salivary glands of Anopheles 29. Qualitatively similar results were obtained by an infection of A. albimanus with P. berghei ANKA parasites expressing the green fluorescent protein 30.

Finally, whether P. berghei sporozoites can be transmitted to the mammalian host by the natural bite of A. albimanus mosquitoes during the third week post infection, was tested when sporozoites were present in the salivary glands of A. albimanus. Mouse infection was never induced even when over 10 infected mosquitoes were allowed to bite. Additionally, when artificial salivation was induced in immobilized A. albimanus, no ejection of sporozoites through the proboscis of the mosquitoes could be detected. Whether this reflects a true natural barrier to sporozoites inside A. albimanus or just the small number of sporozoites within the salivary glands remains to be determined. During previous studies using P. berghei infections of A. stephensi, sporozoites were already being ejected at day 11 post infection, when mosquitoes were artificially stimulated to salivate [11]. However, their numbers were very low (less than five in less than 20% of mosquitoes) and sporozoites were never observed within the first minutes during salivation. As those sporozoites ejected early during salivation are likely to be deposited in the skin 1113, it is not surprising that even the combined bites of hundred A. stephensi mosquitoes at day 11 after the infectious blood meal were unable to infect mice. A similar situation might occur during infections with A. albimanus, where only very few sporozoites are taking up residence in the salivary glands and none was observed in the narrow parts of the salivary ducts.

While the difference between the findings described here and those of Vaughan et al. 17 might be due to the different strains of mosquitoes (A-2) and parasites (ANKA) used, it is more likely that the low number of oocysts developing in the absence of supplements escaped detection in the earlier study. Indeed, no difference was seen between fluorescent NK65 and ANKA strains in A. albimanus. This shows the advantage of using fluorescent parasites, as a single oocyst can be detected by careful observation in living mosquitoes and easily in dissected midguts.

The feasibility to screen mosquito species for susceptibility to a malaria parasite species, using fluorescent parasites was demonstrated. Using this methodology, it was found that the rodent malaria parasite P. berghei is able to infect the major South American malaria vector A. albimanus and to develop into infectious sporozoites albeit at low frequency. This suggests that it should be possible to isolate A. albimanus lines that are highly susceptible to P. berghei infection and that such lines may provide useful new tools for studying the interaction between a model malaria parasite and a major malaria-transmitting mosquito.

FF designed the study, FF and BM performed the experiments, FF, BM, IT, CB and RM discussed the experiments, FF, BM and RM analysed the data, IT and CB contributed the mosquitoes, FF and RM wrote the paper.