The complement system is a first line of defense against invading pathogens. It consists of both soluble and membrane-bound proteins, of which many are proteases that are proteolytically cleaved in a sequential cascade during activation [reviewed in (11, 12)]. These proteins can be deposited on the surface of pathogens or on host cells to produce a membrane attack complex (MAC) via three pathways, namely the classical, alternative, and mannose-binding lectin (MBL) [reviewed in (11)]. In malaria, the complement system may be activated in response to whole parasites or parasite-derived proteins in the host circulation [reviewed in (13)] ( Figure 2 ).

The activation of complement is tightly regulated by both soluble and membrane-bound complement regulatory proteins (CRPs) that act at definitive points of the cascade and that are essential to prevent autologous complement attack ( Figure 2 ) [reviewed in (14)]. The membrane-bound CRPs are constitutively expressed on the surface of cells including erythrocytes and cells within organs like the kidney, while the fluid-phase CRPs circulate in the plasma and are recruited onto the cell surface upon requirement [reviewed in (16)].

Alterations in the levels of complement in serum have been reported during malaria (17–22). The studies performed in a P. lophurae infected duck model showed decreased serum levels of initial complement proteins, C1, C2, and C4 during infection (23). Reduced levels of C4 in serum in simian malaria have been reported implicating classical complement activation (19). Rhesus monkeys infected with P. coatneyi showed decreased serum levels of initial complement proteins, C1, C2, and C4 during schizont rupture (24). Human studies showed reduced serum complement levels in patients with cerebral malaria compared to benign infection also indicating classical complement activation (17). Similarly, studies in malaria- infected pregnant women showed increased amounts of C1q, C3d, C4, and C9 in malaria-infected placentas compared to non-infected placentas (25), and deposition of IgG and C3 in some of the P. falciparum-infected placentas (26), although no association was shown between infection severity and the amount of complement deposited on the infected placentas. Genome-wide expression analyses showed an upregulation of C1q, C3, C5aR, and C3aR genes in the placentas of primigravid women with malaria compared to placentas of primigravid women without placental malaria (27), implying a role for classical complement activation in disease pathology.

Both the alternative and the classical complement pathways are activated in malaria, indicated by increased levels of alternative pathway derived components, Bb (a breakdown product of factor B) and classical pathway components like C4d (a split product of inactive C4b) as well as soluble C5b-C9 in natural P. falciparum infection (28). Complement activation is also regarded as one of the earliest immune responses against experimental P. falciparum infection and can be demonstrated when parasitaemia is still undetectable in peripheral blood (22). In studies performed in children with severe malarial anaemia and uncomplicated malaria, the complement system is activated, but a higher level of complement consumption was observed in children with severe malarial anaemia compared to uncomplicated malaria (20).

The parasite components that directly activate complement in malaria include malarial antigens expressed on the surface of IEs (29), and the products of IE rupture like hematin (30). The antigens released by P. falciparum growing in culture including merozoites can activate all three pathways of complement, but they cause greatest activation of the alternative pathway (see Figure 2 ) (31).

The role of lectin complement pathway in malaria is not clearly demonstrated. MBL seems to recognise parasite proteins of P. falciparum-IEs (32, 33) and may activate lectin pathway.

Genetic studies also reveal a role of MBL protein in malaria. The concentration of MBL protein in serum is genetically determined and different haplotype variants of the MBL gene influence the circulating levels of MBL protein (34). In a study from Gabon, MBL gene polymorphisms were associated with reduced serum levels of MBL protein, and these mutations were present at a higher frequency in children with severe malaria compared to those with mild malaria, suggesting that low MBL levels might be a risk factor for severe malaria (35). This observation was further supported by another study from Ghana that showed low levels of MBL associated with the mbl2 gene variant increased both susceptibility to P. falciparum infection and to severe malaria in young children (36).

By contrast, some studies were unable to show any associations between MBL polymorphisms with parasitaemia and severe disease. Multiple variant alleles of the mbl2 gene that predicted low serum levels of MBL were not associated with infection or malaria severity in Ghanaian children (33), asymptomatic P. falciparum infection in Gabonese children (37), or clinical malaria in Gambian children (38). These discrepancies may be a result of the differences in study design as well as the disease manifestations, and age groups of children enrolled in each study. Taken together, all these studies highlight the possible importance of the lectin pathway in malaria severity, but further studies are needed to fully elucidate the role of MBL and its polymorphisms in malaria susceptibility.

Antigen-antibody immune complexes (ICs) activate complement via the classical pathway in malaria [reviewed in (39)]. They are formed when antibodies bind malarial antigens circulating in plasma ( Figure 3A ) or expressed on the surface of sporozoites, merozoites, and IEs ( Figure 3B ). In individuals infected with malaria for the first time, ICs may first form approximately 10-14 days after infection, while in subsequent Plasmodium infections complement appears to be activated earlier (13).

The binding of globular head domains of the complement factor C1q with the antibody constant (Fc) domain regions of IgG hexamers or IgM binding antigen activates the classical complement pathway (40). The ability of IgG or IgM antibodies to activate the classical complement pathway depends on antibody isotype and subclass, with IgM-bound ICs having the highest ability to bind C1q and activate the classical pathway, while IgG4 has the lowest activity, and for the other IgG subclasses, the affinities for C1q are IgG3 > IgG1 > IgG2 [reviewed in (41)].

In the next section, we provide a brief overview of antibody-mediated complement fixation on different stages of P. falciparum within the human host. We discuss the mechanisms of clearance of the parasites via complement-mediated lysis that are brought about by activation of the classical complement pathway in the presence of ICs. We also review the influence of intrinsic and extrinsic properties of both host and parasite that could have a potential impact on complement-dependent lysis of different parasitic stages.

After injection by the mosquito, Plasmodium spp. sporozoites enter the blood vessels and move through the circulation and invade hepatocytes, where they divide to produce merozoites which are released to initiate blood-stage infection ( Figure 1 ) [reviewed in (42)]. Targeting sporozoites is potentially an efficient way of preventing malaria as only a small number of sporozoites are injected by the female mosquito during a blood meal.

Studies in murine models have shown that the passive transfer of monoclonal antibodies against the sporozoites of P. yoelii inhibited liver infection and the progression to blood-stage infection (43, 44) while in humans, antibodies against P. falciparum sporozoites inhibited the movement of sporozoites into hepatocytes in vitro (9).

Recent in vitro studies conducted in humans showed that these naturally-acquired antibodies against sporozoites of P. falciparum can fix complement on the sporozoite surface and are predominantly of cytophilic subclasses, immunoglobulin 1 (IgG1) and IgG3 (9). The hepatocyte transversal inhibitory activity of the naturally acquired anti-sporozoite antibodies was substantially enhanced in the presence of complement, resulting in fixation of C1q on sporozoites and causing their death (9).

Studies show that immunisation with live-attenuated sporozoites of P. falciparum can induce sporozoite-specific IgG and IgM antibodies that can fix complement on the sporozoite surface. These antibodies can inhibit sporozoite traversal and invasion into hepatocytes and enhance sporozoite membrane permeability, resulting in sporozoite lysis (45, 46). Similarly, the RTS,S vaccine is based on the major circumsporozoite protein (CSP) on P. falciparum sporozoites, and anti-CSP antibodies are induced following RTS,S vaccination that are functional and fix complement factor C1q (47).

In summary, induction of complement-fixing antibodies against sporozoite antigens via natural exposure and vaccination can inhibit sporozoite transversal into liver hepatocytes leading to their lysis and death (9), and these antibodies are associated with protection from clinical malaria (9, 46).

Studies have identified merozoite surface proteins (MSP) such as MSP1₁₉, MSP3, and apical membrane antigen-1 [reviewed in (48)], as important targets of protective antibodies, particularly of type IgG (49). The antibodies targeting merozoites limit parasite replication and inhibit invasion of erythrocytes.

Both naturally acquired (50) and vaccine-induced (50, 51) human antibodies against merozoites promote C1q complement deposition on the merozoite surface and activation of the classical complement pathway, inhibiting merozoite invasion and lysing merozoites (50). A longitudinal cohort study performed in older children showed strong associations between complement-fixing antibodies against MSP1 and MSP2 with protection from clinical malaria and high parasitaemia (50). This observation is further supported by a similar study in malaria-exposed children (52) that showed that complement-fixing antibodies against merozoite antigens were a strong predictor of protection against malaria in children.

Gametocyte-IEs are the infective stages of the parasite that enable transmission of infection from human to mosquito [reviewed in (53)]. There is limited recognition of gametocyte-IEs by naturally acquired antibodies within the human host and this low level of recognition may facilitate the evasion of host immunity and transmission of infection to the mosquito (54).

When an Anopheles mosquito takes a blood meal, host serum components like complement proteins and antibodies are taken in along with the gametocyte-IEs ( Figure 1 ) [reviewed in (53)]. In the mosquito midgut, the gametocytes emerge from the erythrocytes, and are exposed to complement proteins and antibodies [reviewed in (53)].

Most studies on immunity to P. falciparum sexual stages revolve around the major gametocyte surface antigen, Pfs230 [reviewed in (55)]. Previous studies showed that the transmission blocking activities of monoclonal antibodies against Pfs230 were complement dependent (56), and in vitro complement-mediated lysis of gametes by immune sera is associated with antibodies towards Pfs230 (57). But in membrane feeding assays, the transmission blocking activity of immune sera has yet to be shown to be complement dependent (58). Pfs230 is a leading candidate for transmission-blocking vaccines (59), and is likely a major antigenic target for complement-fixing antibodies.

The clinical symptoms of malaria are attributable to blood-stage infection. Some of the major targets of acquired immunity to blood-stage infection are the surface antigens on P. falciparum-IEs (60).

Parasite antigens on the surface of IEs play a major role in the pathology of severe malaria via parasite sequestration leading to cytoadhesion of IEs to vascular endothelium [reviewed in (61)] or syncytiotrophoblast of the placenta (62). These surface antigens can undergo antigenic variations (63) and are known as variant surface antigens (VSA) (64). The dominant VSAs that are expressed on the surface of IEs are the P. falciparum erythrocyte membrane protein 1 (PfEMP1) family of proteins (65).

The surface of trophozoite-IE of P. falciparum is a target for antibody-dependent complement activation (29). In the presence of immune sera the classical complement pathway was activated as indicated by formation of complexes of C1s and C1 inhibitor, measured by ELISAs, although antibody-mediated lysis of IEs was not observed visually (29). Spectrometric quantification of the changes in optical absorbance induced by the release of haemoglobin serves as a better option (66).

Later, Weisner et al. showed that the complement cascade can be activated on the surface of IEs, detecting immunoglobulins, C3, C4, and C9 on the surface of IEs (but not uninfected erythrocytes) incubated with immune sera via western blot analyses (67). However, they failed to observe IE lysis by classical complement activation in the presence of immune sera suggesting that IEs are resistant to complement-mediated lysis, discussed in more detail in section 5.

Notwithstanding the resistance of antibody-opsonised IEs to complement-mediated lysis, IEs are susceptible to other mechanisms of antibody-mediated removal that are briefly discussed here. Previous studies have shown that antibody-opsonised IEs are cleared by monocytes (68, 69) and neutrophils (70) by cellular phagocytosis. Antibody-opsonised IEs can also activate NK cells, which resulted in lysis of IEs and inhibition of parasite growth (71).

Under normal physiological conditions, the ICs are efficiently cleared by a functional complement system preventing excess deposition of complement that brings detrimental effects to the host [reviewed in (11)]. The complement receptor 1 (CR1) on the surface of macrophages, B cells, neutrophils, dendritic cells, and erythrocytes in humans can recognise complement fragments C3b, iC3b, and C4b that are bound with ICs [reviewed in (11)]. CR1 binds, transports, and endocytoses C3b-bearing ICs to remove them from circulation [reviewed in (14)]. Additionally, the complement receptor 3 (CR3 or CD11b/CD18 complex) on the surface of leukocytes is involved in C3b-mediated opsonic phagocytosis by monocytes and neutrophils and plays a role in clearance of C3b-containing ICs [reviewed in (14)]. The Fc receptors of innate immune cells like neutrophils and monocytes can directly bind to the Fc domain of the ICs also promoting antibody-mediated opsonic phagocytosis (72).

Mice infected with P. yoelii showed a downregulation of the level of expression of CR1 on monocytes or macrophages, a similar though less pronounced downregulation was reported in patients infected with P. falciparum and P. vivax compared to non-infected controls (73). Decreased CR1 expression on monocytes or macrophages in the mice was also associated with inhibited uptake of immune complexes and was also seen following exposure to lipopolysaccharide (73). Inflammation may contribute to decreased CR1 expression, which then leads to impaired ICs clearance in malaria, and possibly to disease as ICs are associated with increased disease severity (74). But the contribution of ICs to malaria pathogenesis is not fully known.

Both IgG and complement are shown to be deposited on the surface of uninfected erythrocytes in severe malarial anaemia (75, 76). Complement deposition promotes erythrophagocytosis of IC-deposited erythrocytes via CR1 on macrophage surface (77). In Plasmodium infection, erythrophagocytosis by macrophages seems complement-dependent (78), a possible mechanism for severe malarial anaemia not broadly discussed here (see Box 1 ).

Other than erythrophagocytosis, erythrocytes with C3b-containing ICs are taken up by splenic reticuloendothelial cells. This may lead to stripping off of the CR1 from the erythrocyte surface (76, 79, 80). CR1-deficient erythrocytes are recirculated and are susceptible to complement attack, implicating complement deposition as a driver of severe malarial anaemia. Among heavily malaria-exposed Gambian children, about half developed a positive direct antiglobulin (Coombs) test (81). IgG was eluted from the surface of their uninfected erythrocytes, and in many cases it was shown to recognise schizonts (82), although the antigen specificity of the antibodies bound to uninfected erythrocytes has not been studied in detail. More recent studies [reviewed in (83)] confirm the deposition of IgG and C3 on these cells. The antibodies are postulated to be immunologically unrelated to the uninfected erythrocytes (81).

Complement activation generates C5a via the cleavage of C5, by C5 convertase (see Figure 2 for complement pathway). C5a is a potent inflammatory mediator (anaphylatoxin) that is readily cleared from plasma via receptor internalisation under normal physiological conditions [reviewed in (84)]. The ligation of C5 with C5a receptors (CD88 and C5L2) on innate immune cells promotes a cascade of conventional inflammatory events including increased leukocyte extravasation, neutrophil chemotaxis, degranulation, delayed apoptosis, phagocytosis, oxidative burst, and the activation of vascular endothelial cells to upregulate the expression of cell adhesion molecules [reviewed in (84)].

C5a is implicated in the pathogenesis of cerebral malaria (85) and placental malaria (86, 87). C5a is increased in women with placental malaria (87) and elevated C5a was associated with increased risk of birth of ‘small-for-gestational-age’ babies (86). Blocking C5a in a murine model of placental malaria resulted in improved placental and foetal development (86). Murine models have also highlighted a possible role for C5a in cerebral malaria as C5 deficient mice or those treated with antibodies blocking C5a and its receptor respectively did not develop and could be rescued from cerebral malaria (85). Though there is some evidence for a role of the inflammatory complement C5a protein in disease, the anti-C5 monoclonal antibody eculizumab has not been studied in malaria (see Box 1 ).

The inflammatory effector functions mediated by the release of C5a in response to infection and the C3b-mediated opsonic phagocytosis of IEs (and/or other parasite stages, such as sporozoites, merozoites, and intraerythrocytic gametocytes) by innate immune cells are not discussed in detail in this review.