Infection of humans with live sporozoites under chloroquine prophylaxis (28) or after immunization with irradiated sporozoites (1, 4) results in a long-lasting expansion of γδT cells with a “memory” phenotype. Although these cells are detected in peripheral blood, it is not known at which stage of the parasite life-cycle, or where, they were induced (skin, draining lymph nodes, infected hepatocytes or peripheral blood/spleen, see Figure 1 and Table 1), whether they recognize antigens expressed uniquely at the sporozoite stage, or even whether they were activated by parasite antigen per se. That they express Vγ9⁺Vδ2⁺ TCRs (1, 29) suggests that they may be circulating γδT cells with access to many tissues. In humans, their activation and effector site is difficult to establish. Although the subpopulations of mouse and human γδT cells are not directly equivalent, the similar preferential tissue locations, eg circulating Vγ4 and Vγ1 mouse γδT cells and Vγ9Vδ2⁺ human cells; tissue-resident Vδ1⁺ cells in humans and Vγ5⁺ and Vγ6⁺ in mice, are such that one could pursue this in mouse models.

Whether and how γδT cells interact with sporozoites in the skin is unknown, but given their appearance after sporozoite immunization, this would be an area of research worth pursuing. Rapidly activated γδT cells could either have some direct effector function or recruit other effector cells to prevent further development of the infection. The use of mice that lack epidermal DETCs or dermal Vγ6⁺ γδT cells (50, 51), may be useful tools to determine the importance of skin-residing γδT cells in malaria.

The association of γδT cells with protection after irradiated sporozoite immunization (1, 4) and the demonstration of the protective effects of γδT cells in irradiated sporozoite vaccination in P. yoelii and P. berghei mouse models (4, 32, 33) also suggest that these cells have an important protective role in the liver. The views are that γδT cells act either as effector cells that operate in the absence of αβT cells, or as accessory cells for appropriate protective responses from other cells (4). With the differences in the experimental approaches and the heterogeneity of γδT cell functions now known, it is likely that γδT cells could be performing both functions in these experimental models. It will also be important to determine whether intrahepatic (possibly Vγ6⁺) or blood Vγ1⁺ or Vγ4⁺ γδT cells are playing the protective role (52). In Zaidi et al.'s mouse model, the Vγ4⁺ γδT cells do not appear to have a role in protection, while the nearest equivalent in humans, blood Vγ9Vδ2⁺ cells are found to associate with protection. We have some clues about γδT cells induced by pre-erythrocytic stages of Plasmodium but many questions remain: which γδT cells? Where are the cells activated? What do they recognize? If they are accessory cells how are they functioning? If they are direct effector cells, what is their mechanism? The mouse models will be a good way to address these issues.

The first observations of γδT cell responses to blood-stage malaria parasites were made more than 25 years ago, and showed that Vγ9Vδ2⁺ γδT cells of malaria-naïve individuals proliferate in response to P. falciparum-infected red blood cells (iRBCs) in vitro (14, 16, 30, 31). Following this, it was demonstrated that Vγ9Vδ2⁺ cells increase to up to 10–30% of total PBMC in P. falciparum- or P. vivax-infected adults with little or no previous exposure to malaria (14–18). More recently, this has also been shown in experimentally infected individuals (19). In regions of high malaria endemicity, or after multiple infections, healthy individuals already have a higher frequency of γδT cells than European populations (53), and there is no further peripheral expansion of Vγ9Vδ2⁺ γδT cells on exposure to the parasite; however, there is a large increase in the proportion of Vδ1⁺ cells in the PBMCs (20–24). The reasons for the relative expansion of Vδ1⁺ cells is not understood but could be due to the retention of active Vδ2⁺ cells in the spleen, thus changing the proportions in peripheral blood, or to the circulation of Vδ1⁺ cells normally activated and residing in tissues, such as liver and skin.

γδT cells also increase in most of the rodent models of malaria studied. They are expanded in spleens of mice infected with different strains of P. chabaudi, P. yoelii, and P. berghei within 1–2 weeks of a blood-stage infection, depending on the mouse/parasite combination, reaching a peak at 3 weeks post-infection in non-lethal infections (13, 19, 34, 35), but with no expansion in a lethal P. yoelii infection (26).

The advantage of mouse models is that they can tell us whether the γδT cell response observed during blood-stage infection plays any protective role. In all the different infections studied—P. yoelii XNL,XL, P. berghei XAT, P. chabaudi AS, AJ, P. chabaudi adami K556A—mice without functioning γδT cells due to in vivo depletion with specific antibodies, or because of targeted deletion of the δ gene, show exacerbated acute parasitemias (although the increase is more pronounced with P. yoelli than with, eg, P. chabaudi) (19, 26, 27, 40, 41, 46–48), and in some cases this results in a lethal infection (26, 42, 48). In P. chabaudi, depletion of γδT cells additionally results in delayed clearance (27, 40, 41, 46, 47), or an increased magnitude of chronic parasitemias (19).

It is important to know which subpopulations of γδT cells in the mouse are responsible for the protective effect, as this could give us clues about where the γδT cells may have been activated as well as the nature of the inducing antigens. The γδT cell response to blood-stage Plasmodium in humans is dominated by cells bearing Vγ9Vδ2⁺ TCR-chains, and thus one might predict that the nearest mouse counterparts are the blood/lymphoid circulating Vγ1⁺ or Vγ4⁺ γδT cells. However, the mouse response, at first glance, appears to be more heterogeneous. Although most reports show the circulating γδT cell bearing Vγ1⁺, and/or Vγ4⁺ TCRs to be expanded with different δ chains (13, 19, 25), there are also reports of Vγ2⁺ T cells (25, 27, 35). This could reflect the very different infections of P. chabaudi, P. yoelii and P. berghei strains in the mouse, and the different mouse strains used. However, some of the differences between different studies may be due to the confusing systems of nomenclature used for designating the γ chain of the mouse γδTCR (49, 54–56). Many of the earlier papers do not define clearly the nomenclature used. Given the contribution of mouse γδT cells to the control of blood-stage infections in mouse models, it would be well worth revisiting this, and reanalysing the γδT cell response in the different blood-stage infections.

All the blood-stage mouse infections described here differ in one major respect from natural infection, or sporozoite vaccination, in that they are not initiated via the bite of an infected mosquito. Not only does this mean that two key sites of potential γδT cell activation, skin and liver, are missing, but also the parasites from serial blood passage may differ in their transcriptional profile and in virulence (57, 58). The lack of the pre-erythrocytic stages of Plasmodium could influence location, specificity and dynamics of γδT cell activation. While the mouse γδT cell subsets are not direct counterparts of human, if we are to use the mouse model to elucidate mechanisms of γδT cell activation and protection, we should approximate as closely as possible to the mode of infection in humans.

In humans and mice, rapidly activated blood/circulating/lymphoid γδT cells produce large amounts of interferon-γ (IFN-γ) after stimulation in vitro with P. falciparum-iRBCs (22, 28, 31, 43, 45) or during early blood-stage infection with Plasmodium (27). IFN-γ can mediate killing of parasites in infected liver cells (72), and more indirectly, activate phagocytes that can eliminate blood-stage parasites by antibody-dependent or -independent mechanisms (73). In the absence of αβT cells, γδT cells are required in some irradiated sporozoite immunization protocols to eliminate liver-stage parasites in mouse models (28, 32, 33, 45). However, it is not clear whether IFN-γ from γδT cells is crucial for these effector mechanisms in malaria, as αβT cells and NK cells also produce this cytokine. On the other hand, loss of γδT cells in most mouse models of malaria does not compromise greatly the ability to remove acute blood-stage infections, although they may be important to control recrudescences (19, 26, 27, 32, 40, 41, 46–48).

Through their cytotoxic activity, it is conceivable that γδT cells can directly kill parasites (5). Both activated human Vδ2⁺ and Vδ1⁺ T cells degranulate when incubated with free merozoites, but not intraerythrocytic parasite stages, and can inhibit P. falciparum replication in vitro in a dose-dependent manner (43, 44, 74). This direct parasiticidal effect is dependent on granulysin rather than perforin, and requires contact or at least close proximity to target cells. Furthermore, patients infected with P. falciparum had elevated granulysin plasma levels and high numbers of granulysin-expressing Vγ9Vδ2⁺ T cells, which degranulated when incubated in the presence of iRBCs (44). How effective these cytotoxic responses are in vivo remains debatable since merozoites only spend a very short time in the extracellular environment between egress and re-invasion. It might be more effective in tissues with a low blood flow, such as the red pulp of the spleen, where both γδT cells and mature iRBCs are highly prevalent and thus would have a higher chance of an encounter (75). Whether γδT cells can directly kill merozoites in mice, and whether this contributes to control of parasites has not yet been demonstrated. Perhaps the use of conditional knock-out mice in which the cytolytic machinery or IFN-γ-signaling has been specifically ablated in γδT cells or CD3DH mice which have reduced numbers of IFN-γ-producing γδT cells (71) would elucidate their roles more clearly.

There is now a wealth of literature about the involvement of different subpopulations of γδT cells in MHC class I presentation, regulation of other immune cells, and production of cytokines important for myeloid cell development. The most widely studied human γδT cells, Vγ9Vδ2⁺ cells, have a wide variety of other functions including follicular helper-like, Th17-like, and Th2-like responses (9). Most of these aspects have not been explored in detail in malaria, but mouse models could offer good insights into how they contribute to the protective host response to Plasmodium. A recent example of an immunoregulatory function of γδT cells is the study by Zaidi et al. (4), where they have shown that γδT cells are important for the protective response induced by irradiated sporozoites, but it seems not as direct effectors. They are required for the development of an effective CD8⁺ T cell response. In this immunization model, γδT cells were required for recruitment of cross-presenting CD8α⁺ dendritic cells, necessary to activate effector CD8⁺ T cells. How this is achieved is currently not known, but a recent paper on the interplay of γδT cells and myeloid cells may offer some clues. γδT cells producing macrophage-colony stimulating factor are important for controlling the chronic phase of P. chabaudi infections, suggesting that γδT cells are interacting with the myeloid cell compartment to control parasitemia (19, 76).

The functional capacities of epithelial γδT cells have not been investigated in great detail in malaria. As mouse skin and liver γδT cells produce IFN-γ and/or IL-17, and human skin Vδ1⁺ cells, in addition to production of IFN-γ, can be cytotoxic, and these cells, when activated, recruit myeloid cells, and enhance phagocytosis (77, 78), such studies would be worthwhile.

Long-term responses, or effective reactivation on second encounter with antigen requires some form of longevity of the cell population, either by constant re-stimulation, or through development of long-lived memory cells. Obviously for harnessing γδT cells in protective immune responses induced by vaccination it would be good to have an expanded population of “memory” cells that give an enhanced and more rapid response. The general view has been that γδT cells, although expressing TCRs encoded by somatically rearranged genes, are innate-like effectors that do not establish antigen-specific memory. It could also be argued that there is no need for the development of memory cells, as γδT cells have a relatively limited repertoire of TCRs which respond rapidly to the same set of antigens without the need for massive expansion, and this would happen on every exposure to appropriate antigens (79). Nevertheless, there are reports of adaptive-type memory responses of γδT cells. Human Vγ9Vδ2⁺ T cell responses to phosphoantigens are increased by prior Mycobacterium bovis BCG vaccination (80). In vivo, there is a long term expansion of effector memory Vδ2⁻ cells in human Cytomegalovirus infections (81) and enhanced “secondary” responses by Vγ9Vδ2⁺ T cells in macaques infected with live Mycobacteria (82). Mouse “memory-like” Vγ6⁺ γδT cells were maintained for more than 5 months in mesenteric lymph nodes after Listeria monocytogenes infection (83) and Vγ4⁺ γδT cells have been found to persist in dermis and draining nodes for more than 3 months in a skin inflammation model (56, 84). Long-term elevation of γδT cells has been observed in peripheral blood of P. falciparum-exposed humans under chloroquine prophylaxis (28) or following irradiated sporozoite vaccination (1, 4), and in humans in malaria-endemic areas (53). Whether these contain true long-lived memory cells able to exist in the absence of antigens is not known. Such studies have not yet been carried out in either human or mouse Plasmodium infections.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.