Malaria infections are not always lethal. One reason for this may be that transmission from mosquitoes creates malaria parasites that trigger a more protective mammalian immune response. See Letter p.228
Malaria parasites can kill people, but death is not inevitable. Most infected individuals recover, some after experiencing relatively mild symptoms or none at all. What accounts for this variability? Host factors such as the expression of sickle-cell genes or acquired immunity are part of the explanation. But it is also well known that malaria parasites themselves can be more or less nasty1,2. In this issue, Spence et al.3 (page 228) report a set of clever experiments in a mouse model of malaria infection that shows that the conditions experienced by parasites before they reach the mammalian bloodstream can determine just how virulent they areFootnote 1.
Malaria parasites transmitted to people by mosquitoes migrate to the liver, where they replicate before entering the bloodstream. For convenience, and because only blood-stage parasites cause disease, most experimental studies of malaria in humans and animals bypass the mosquito and liver stages and inject parasites directly into the bloodstream. Using the malaria parasite Plasmodium chabaudi, which infects rodents, Spence and colleagues compared the blood-stage infections generated by this method with those initiated naturally, by mosquito bite. They found that, compared with directly injected parasites, mosquito-transmitted parasites replicated less well once in the bloodstream and generated lower-grade infections that persisted for longer. Moreover, these parasites did not induce the severe weight loss, hypothermia and liver damage caused by parasites injected directly into the bloodstream.
Why these differences? An important clue came from the authors' finding that, in immunodeficient mice, parasites transmitted by mosquitoes grew just as well as those injected directly. This suggested that there is nothing intrinsically attenuated about parasites derived from mosquitoes. Spence et al. show that mosquito-transmitted parasites elicit a qualitatively different immune response in the mouse — one that better controls parasite replication and relies less on the inflammatory signalling molecules that are associated with severe disease. To try to explain this difference, Spence et al. conducted a genome-wide RNA analysis and found that mosquito transmission modifies the expression of about 10% of the genome of blood-stage parasites. Intriguingly, expression was most intensely regulated for gene families encoding antigenic proteins, against which the host's immune system mounts its response. The hypothesis, then, is that mosquito transmission alters subsequent antigen expression when the parasites are in the bloodstream, and that the induced gene-expression pattern elicits an immune response that more effectively contains the parasites with less collateral damage to the host.
It seems that it is the environment experienced by the parasites during natural transmission that triggers this 'attenuated phenotype'. That environment could be inside the mosquito itself, or it could be something experienced by the parasite in the skin soon after injection, during its journey to the liver or in the liver. Intriguingly, Spence et al. show that the attenuated phenotype also occurs in mice injected with blood-stage parasites isolated from other mice with mosquito-initiated infections. Thus, the phenotype is stable for several cycles of blood-stage parasite replication, although it does gradually decay over subsequent rounds of injecting these parasites into new hosts. It will be interesting to determine whether profiles of the host immune response and of parasite-antigen expression associated with attenuation decay in a similar manner.
Does this discovery mean that all future experimental malaria infections should be initiated by mosquitoes? There is no way to include mosquito transmission in in vitro studies of the most lethal human malaria parasite, Plasmodium falciparum. Immunosuppressed mice with human-cell transplants can support P. falciparum infections4, but it is unclear whether the addition of one aspect of biological reality (mosquito transmission) will make up for the loss of another (the use of human parasites in mice). Mosquito infections are an option in animal models, from which much has already been learned by injecting blood-stage parasites. For example, experiments with P. chabaudi have shown that a powerful contributor to the severity of malaria can be the host immune response itself5, and that competition between different parasite strains can be a potent force shaping the evolution of drug resistance6. The key question is not whether these phenomena still occur if more of the parasite life cycle is incorporated into the experimental work, but whether they occur in nature.
The effects of mosquito transmission on host immune response and parasite antigen expression observed by Spence and colleagues might be the independent outcomes of environmental influences, or they might be causally connected. If the latter is true, the question remains whether immunity triggers the antigenic profile or whether altered antigen expression triggers a more protective immune response. The direction of this causality could have implications for vaccine design. In the first scenario, considering the characteristics of the immune response generated by a vaccine would be important for protecting against severe disease if vaccination does not completely block infection. In the second case, a vaccine that results in exposure to a particular antigenic profile may be crucial for developing an optimally protective response. Spence et al. compared infections initiated by mosquitoes and by blood-stage parasites at just one time in the blood-stage infection, but antigen expression can be highly variable in time and across host tissues7,8, so further assessment of these profiles is needed.
If antigen-expression profiles are indeed a major determinant of malaria-parasite virulence, and if these are not completely constrained by the parasite's developmental requirements, we predict that natural selection will favour different antigen-expression profiles in different epidemiological settings (Fig. 1). If this is the case, then virulence variability due to genetic polymorphisms or phenotypic plasticity will be common in nature. This might explain apparently contrasting experimental results. For example, Spence et al. found that mosquito transmission attenuated parasite replication in two clones of P. chabaudi, but earlier experiments using a different clone found no such effect9. Similarly, physicians who deliberately infected people with P. falciparum to treat neurosyphilis reported the same clinical picture regardless of how the infection was initiated2. Clearly, much is yet to be learned about how malaria parasites make people sick, and about the role of the mosquito vector in modulating the disease it initiates.
*This article and the paper under discussion3 were published online on 29 May 2013.
Mackinnon, M. J. & Marsh, K. Science 328, 866–871 (2010).
Covell, G. & Nicol, W. D. Br. Med. Bull. 8, 51–58 (1951).
Spence, P. J. et al. Nature 498, 228–231 (2013).
Legrand, N. et al. Cell Host Microbe 6, 5–9 (2009).
Langhorne, J., Ndungu, F. M., Sponaas, A.-M. & Marsh, K. Nature Immunol. 9, 725–732 (2008).
Read, A. F., Day, T. & Huijben, S. Proc. Natl Acad. Sci. USA 108, 10871–10877 (2011).
Ebbinghaus, P. & Krücken, J. Malaria J. 10, 272 (2011).
Lawton, J. et al. BMC Genomics 13, 125 (2012).
Mackinnon, M. J., Bell, A. & Read, A. F. Int. J. Parasitol. 35, 145–153 (2005).