Introduction

With an estimated 250 million cases every year and an annual death toll of one million people, the tropical disease malaria remains a major global health threat and a serious economic burden (reviewed in [1, 2]). Approximately 50% of the world’s population live in affected areas within wide regions of Africa, South-East Asia, Central America, and the Eastern Mediterranean [3]. In most endemic countries, the treatment and control of malaria is undermined by the spread of drug resistant parasites, extreme poverty, and the impact of HIV infections.

Malaria is a vector-borne disease that is caused by protozoan parasites of the genus Plasmodium and transmitted by blood-feeding anopheline mosquitoes. There are four human malaria species, among them P. falciparum, the causative agent of the dangerous malaria tropica. When infecting a human host, malaria parasites initially target liver hepatocytes, causing a primary but asymptomatic infection. The parasites subsequently infect erythrocytes, and the recurrent intraerythrocytic replication cycles cause the typical symptoms of the disease, such as fever and anemia, which in severe cases are followed by respiratory distress and organ failure.

Besides three different periods of asexual replication, which take place in the human hepatocytes and erythrocytes as well as inside the mosquito vector, the malaria life cycle also includes a phase of sexual reproduction that mediates parasite transition from the human to the mosquito and thus plays a crucial part in disease transmission. In P. falciparum, the malaria sexual phase begins with the differentiation of intraerythrocytic parasites to the non-replicative gametocyte lineage. Following maturation to crescent-shaped intraerythrocytic forms, the gametocytes freely circulate in the peripheral circulation in anticipation of being taken up by the mosquito during the blood meal. Environmental stimuli in the mosquito midgut during blood feeding rapidly activate the gametocytes, which within minutes egress from the host erythrocyte and undergo gametogenesis, resulting in the development of female macro- and male microgametes. The motile microgamete fertilizes the sessile macrogamete, and within 20 h, the resulting zygote transforms into a motile form, termed the ookinete. The ookinete exits the midgut lumen and crosses the gut epithelium, which marks the end of the malaria sexual phase (reviewed in [4, 5]).

It was only in the past two decades, when research focused on the sexual phase of the parasite life cycle. This interest arose upon the observation that the host’s immune response to sexual stage antigens can completely inhibit the development of Plasmodium in the insect vector, and such antigens would therefore represent ideal candidates for the development of transmission-blocking vaccines (TBVs) (reviewed in [6]). This type of vaccine is based on human antibodies directed against select sexual stage antigens, which are taken up by the mosquito with the blood meal and which eliminate the parasite sexual stages inside the mosquito midgut, as soon as these lose protection by the enveloping erythrocyte. Hence, TBVs represent a powerful tool in malaria control efforts, particularly together with new combination-therapy drugs, and vaccines targeting morbidity and mortality due to the asexual stages (reviewed in [7, 8]).

Among the first vaccine candidates discovered in the 1980s were Pfs230 and Pfs48/45, which are expressed at the surface of the developing gametocytes and later of male and female gametes (reviewed in [4]). Both proteins share a primary structure composed of several cysteine-rich domains assigning them to the cysteine-rich motif superfamily that is comprised of ten members (Fig. 1a) [9–11]. Two additional antigens, found on the surface of zygotes and ookinetes, are Pfs25 and its paralog Pfs28, both possessing four epidermal growth factor (EGF) domains and glycosylphosphatidylinositol (GPI) anchors. The transmission-blocking potential of antisera targeting Pfs25 and its vivax malaria ortholog Pvs25 in humans was subject of clinical phase I trials [12, 13], further confirming the suitability of this protein for the generation of TBVs.

Fig. 1
figure 1

a Schematic of the domain structure of select Plasmodium falciparum sexual stage proteins containing adhesive motifs, i.e., the PfCCp multi-domain adhesion protein family, the EGF domain-containing protein Pfs25, and the cysteine-rich motif proteins Pfs230 and Pfs48/45. Encircled black-dotted domains indicate newly identified motifs. b Schematic demonstrating the proposed network of interactions between sexual stage adhesion proteins. The multimeric protein complex most likely is retained at the plasma membrane of gametocytes via binding to the GPI-anchored Pfs48/45. After gametocyte activation, Pfs25 also locates to the cell surface of the forming macrogamete, covalently linked by a GPI anchor, thereby providing an additional retention site for the multi-protein complex. PPM parasite plasma membrane

Full size image

Additional molecules were identified in the following years, including the here discussed PfCCp multi-domain adhesion protein family [14]. Noteworthy is that the majority of sexual stage surface proteins have adhesive properties and can be divided into two classes. One class of sexual stage proteins, including Pfs230, Pfs48/45 and the six PfCCp proteins, is expressed within the parasitophorous vacuole (PV) of the developing gametocyte and subsequently present on the gamete surface, but expression of these proteins usually ceases during fertilization. The expression of the second class of surface proteins, among them Pfs25 and Pfs28, is initiated after the parasites have entered the mosquito midgut and persists until the ookinete has formed (reviewed in [4, 5]).

The reasons for the high number of adhesive proteins in the malaria parasite sexual stages and the coordinated expression pattern of these proteins during transmission remain elusive. A recent study from our laboratory indicated that a subset of these proteins assemble to multimeric complexes on the parasite surface. In this review, we highlight our previous findings on the PfCCp protein family and provide evidence for complex formation of multiple surface-associated sexual stage proteins. We will compare these data with findings on other protein complexes of malaria parasites and speculate on the possible function of the sexual stage multimeric complexes.

Domain structure of the PfCCp protein family

The PfCCp protein family was identified via annotation of the P. falciparum genome sequence for proteins encoding multi-domain “animal-like” architectures. Six members were described exhibiting multiple adhesion domains conserved among animals and bacteria (Fig. 1a) [14], five of which share a common LCCL (Limulus coagulation factor C, cochlear protein Coch-5b2, and late gestation lung protein Lgl1) domain, which was eponymous for the proteins thenceforward named PfCCp1 to PfCCp5 (LCCL domain-containing protein). Despite the lack of a LCCL domain, a sixth protein referred to as PfFNPA was assigned to the PfCCp protein family due to its structural similarity to PfCCp5. The predicted signal peptide sequence common to all PfCCp proteins indicated an extracellular location, yet GPI-anchor signal sequences or transmembrane regions were not detected [14]. Significantly, all members of the PfCCp protein family possess orthologs in other organisms of the apicomplexan clade.

PfCCp1 and PfCCp2 have similar architectures and are paralogs arising from gene duplication after domain accretion. PfCCp1 and PfCCp2 possess binding domains, corresponding to ricin, discoidin, and levanase-type lectin domains [15, 16]. C-terminal to these modules is two copies of an apicomplexan-specific cysteine-rich module, termed ApicA. Between the discoidin and levanase lectin domains are two LCCL domains [17] and a NEC module (termed after neurexins [18] and collagens).

PfCCp3 contains in total four LCCL domains, a tandem repeat of a scavenger receptor (SR) domain, a lipoxygenase LH2 domain, and a region similar to pentraxin. PfCCp4 possesses two center LCCL domains, followed by tandem levanase domains. PfCCp5 encodes a region at the amino terminus that is similar to the type 2 domain of fibronectin (FN2), fused to a domain similar to the N-terminal region of anthrax protective antigen (PA) and a single LCCL domain near the C-terminus. As described above, PfFNPA is similar to PfCCp5 in overall architecture but lacks an LCCL domain. Given that PfFNPA has an apparent ortholog in the distantly related apicomplexan, Cryptosporidium parvum, the two genes are likely to have diverged in P. falciparum from a common ancestor that was similar in architecture to PfFNPA, followed by accretion of an LCCL domain in PfCCp5.

Recent re-annotation of the PfCCp protein sequences [I] revealed several additional adhesion domains, i.e., a second discoidin domain in PfCCp1 and PfCCp2, which is located at the N-terminus of the respective protein, and a discoidin domain in PfCCp5, which is located between the anthrax PA N-terminal domain and the LCCL domain (Fig. 1a). Furthermore, two novel N-terminal adhesion domains were identified in PfCCp4, which correspond to ricin domain and anthrax PA N-terminal domain. The various adhesion domains and their potential ligands are discussed as follows.

Expression of PfCCp proteins in sexual stage parasites

Expression profile studies on the PfCCp proteins revealed transcript and protein expression specific to gametocytes within the five developmental stages during maturation (termed stages I–V) [19, 20]. As predicted from the signal peptide sequences, the proteins are secreted and localized to the parasite plasma membrane within the PV. PfCCp1, PfCCp2, and PfCCp3 co-localize in a punctate pattern at the parasite surface, while the surface-associated expression of PfCCp4 is rather homogenous and is thus similar to the expression pattern of Pfs230 and Pfs48/45. PfCCp5 and PfFNPA are less abundant in gametocytes than the other four PfCCp proteins and in mature gametocytes often localize to the cell poles [20].

After gametogenesis in the mosquito midgut, which involves parasite egress from the host erythrocyte, the PfCCp proteins are present on the cell surface of macrogametes. For PfCCp1, PfCCp2, and PfCCp3, partial release of the proteins was postulated [14]. Expression of the six proteins ceases within a few hours after egress; hence, the PfCCp proteins belong to class I of sexual stage proteins.

To gain information on possible protein functions, we constructed gene-disruptant parasite lines for PfCCp2, PfCCp3, PfCCp4, and PfFNPA, resulting from single-crossover recombination, as well as a double-crossover knock-out mutant for PfCCp1 [14, 20, 21]. Expression profile studies in the mutants (herein referred to as PfCCp-KOs) led to the remarkable observation that the six PfCCp proteins are expressed in a co-dependent manner and that the absence of one protein results in the complete or partial loss of all other protein family members. While the transcript abundances of intact PfCCp genes in the knock-out mutants are comparable to those in wild-type parasites, the proteins are not detectable via Western blot analysis and immunofluorescence assay [21, 22]. Thus, the co-dependent expression manifests at the translational or post-translational level rather than at the transcriptional level of gene expression. The observed co-dependent expression of the PfCCp family led to the assumption that the proteins interact in a way, which ensures localization to and/or retention at the correct expression site and prevents a likely degradation and loss of protein.

Intermolecular binding between PfCCp proteins

Based on the co-dependent PfCCp protein expression, we hypothesized that the six proteins interact in a complex and that instability in the complex due to gene ablation results in co-dependent degradation. By means of co-immunoprecipitation on wild-type gametocyte lysates followed by Western blot analysis, numerous protein–protein interactions between the distinct PfCCp proteins were revealed and interplays on the molecular level were observed between all PfCCp proteins (Fig. 1b) [21]. While no interactions were detectable between PfCCp4 and PfCCp5, the lack of signal for the respective precipitated protein by Western blot analysis is possibly due to the low efficiency of the available antisera against PfCCp4 and PfCCp5.

The molecular interactions between the PfCCp proteins were subsequently studied via affinity chromatography co-elution binding assays, using recombinant proteins comprising selected adhesion domains. The eluted protein complexes were screened by Western blot analyses. Out of 33 combinations investigated, 18 showed adhesive interaction, while 15 recombinant protein pairs did not interact. Adhesion domains that were predominant in protein–protein binding include the LCCL domains of all PfCCp proteins and the SR domains of PfCCp3 (with an involvement of 32% (LCCL) and 19% (SR) of all binding events). Moreover, the ApicA, NEC, and discoidin domains of PfCCp1 and PfCCp2 showed involvement in interactions between PfCCp proteins [21]. Subsequent latex beads-mediated cell binding assays confirmed the interaction of recombinant PfCCp proteins with endogenous proteins on the macrogamete surface. Here, fluorescent latex beads, coated with recombinant proteins corresponding to PfCCp1, PfCCp2, or PfCCp3 domains, bound significantly more often to newly emerged (PfCCp protein-positive) macrogametes than to intraerythrocytic gametocytes, older (PfCCp-negative) macrogametes, and PfCCp3-KO (PfCCp-negative) macrogametes [21].

An interaction of sexual stage adhesion proteins at the gamete plasma membrane leading to a stable protein complex has been shown before for the two TBV candidates Pfs48/45 and Pfs230 [23, 24]. Here, the GPI-anchored Pfs48/45 ensures the retention of Pfs230 to the gamete surface. While in the Pfs48/45-deficient gametocytes, Pfs230 is expressed as in wild-type parasites, it is not present on the surface of gametes after egression from the host cell [25].

New data from our laboratory indicate that the molecular interactions of sexual stage adhesion proteins are more complex than originally expected. We recently detected interactions of select PfCCp proteins with Pfs230 and of PfCCp4 with Pfs48/45 by co-immunoprecipitation assays on gametocyte lysates, and we further revealed the binding of these proteins with Pfs25 following gametocyte activation (Fig. 1b) [20] (N. Simon, A. Kuehn, S.M. Scholz, and G. Pradel, unpublished observation). Pfs25 belongs to class II of sexual stage proteins that are translationally repressed in gametocytes, as was shown for Plasmodium berghei Pbs25 [26, 27]. However, small amounts of protein can be found in vesicular structures in the maturing gametocytes, and the protein is relocated to the surface of macrogametes within minutes after activation [20].

CCp orthologous proteins

Select LCCL domain proteins, i.e., CCp1, CCp2, and CCp3, are highly conserved throughout the apicomplexan clade, including Cryptosporidium spp., Toxoplasma gondii, Babesia bovis, and Theileria annulata. FNPA was further described for C. parvum [14, 28] (reviewed in [29, 30]), and CCp5 was identified in T. gondii [II]. A recent genome survey study also identified the CCp family members in the early branching apicomplexan class of gregarines, Ascogregarina taiwanensis [31], and thus, the family is likely widely conserved in apicomplexans. The LCCL domain proteins are absent in Tetrahymena, for which complete genome sequence is available, and thus, it might be concluded that the acquisition and accretion of the domain architectures was relatively ancient and predates the divergence of the apicomplexan clade, but that the domains entered the apicomplexan lineage after the split of the ciliates [30].

For most of the CCp orthologs, sites of expression in the respective parasite species are yet unknown. New studies in Babesia divergens indicate an expression of BdCCp2 in gametocytes within the tick midgut (C. Becker, S. Bonnet, UMR INRA ENVN Nantes, personal communication). In contrast, expression of CpCCp1 of C. parvum (also referred to as Cpa135) starts in oocyst-sporozoites and increases rapidly after excystation [32]. In sporozoites, CpCCp1/Cpa135 is seemingly stored in the micronemes and is excreted upon host cell invasion. Subsequently, the protein localizes to the PV membrane. With the formation of merozoites, new transcript of CpCCp1/Cpa135 is detectable [33]. It is not yet known if the protein is expressed in Cryptosporidium gametocytes. Notably, both Cryptosporidium sporozoites and Plasmodium gametocytes have to resist the gut environment of either mammalian host or mosquito vector.

CCp orthologs are further present in the rodent malaria parasite P. berghei, where they are also termed LAP proteins [34, 35]. Transcript analysis revealed a predominant expression in P. berghei gametocytes for the six proteins, which was confirmed by transgenic parasites expressing GFP driven by select PbCCp/LAP promoters [36]. Genetic cross studies with female-deficient or male-deficient parasites reported that the functional genes are inherited from female gametocytes, indicating that the PbCCp/LAP proteins are female specific [35]. Recent studies on GFP fusions of PbCCp3/LAP1 (also termed PbSR), PbCCp1/LAP2, and PbCCp5/LAP3 showed protein expression in macrogametocytes and accumulation of these proteins in the crystalloids, organelles that are formed in the ookinete and persist until the early oocyst stage [37–39].

Adhesive modules of CCp proteins

A striking feature of all CCp proteins is the presence of numerous modules, most of which were described to possess adhesive properties in other organisms and were likely acquired by the parasite via lateral transfer and converted into multi-domain adhesive molecules. With the exception of the ApicA and the levanase domains, the adhesion domains of the CCp protein family were previously identified in animal cells and bacteria and assigned to diverse adhesive functions, which are described below.

The eponymous LCCL domain was first described in the horseshoe crab Limulus, where it is found in the coagulation factor C, as well as identified in the vertebrate cochlear protein Coch-5b2 and in the mammalian late gestation lung protein Lgl1 [17]. For Limulus factor C and Lgl1, a role in antimicrobial defense via recognition of cell-surface carbohydrates of pathogens has been proposed [17]. The LCCL domain has been found in association with a number of other adhesion domains, e.g., complement B-type domains, C-type lectin domains, von-Willebrand factor A domains, or discoidin domains [17].

The prototypic discoidin domain is derived from the cell adhesion protein discoidin of the slime mold Dictyostelium discoideum. The domain is found in the coagulation factors V and VIII, where it promotes phospholipid binding on the surface of platelets and endothelial cells [15, 40]. Pentraxins, on the other hand, are a family of evolutionarily conserved pattern-recognition proteins that are made up of five identical subunits. Proteins of the pentraxin family are involved in acute immunological responses and often represent serum proteins. They play a role in the binding of lipoproteins and carbohydrates [41–44] and are reported to interact with other pentraxin domains as well as with complement control protein modules (also called sushi domains) [45]. Interestingly, pentraxin 3 is essential for female fertility, participating in the assembly of the cumulus oophorus extracellular matrix [46].

The LH2 lipoxygenase domain is found in a variety of membrane- or lipid-associated proteins [47], and it is suggested that the domain mediates membrane attachment via other protein binding partners. Interactions with lectin domains and other lipoxygenase domains were predicted [III]. Fibronectin, on the other hand, is a multi-domain glycoprotein found in the plasma as well as in basement membranes, where it exhibits functions in wound healing and blood coagulation, as well as in cell adhesion and migration [48]. The FN2 domain is predicted to interact with EGF domains, lectin C, and ricin domains [IV].

Ricin is a legume lectin from seeds of the castor bean plant, Ricinus communis. The ricin B lectin domain, present in CCp1 and CCp2, binds to carbohydrates [49]. Interactions with FN2 domains, lectins, and other ricin domains were predicted [V]. The proteins CCp1, CCp2, and CCp4 also contain a pair of levanase-type lectin domains, which are found in secreted glucosidases and levanases of Bacillus subtilis and proposed to be involved in sugar binding [28].

CCp4, CCp5, and FNPA share a domain termed PA (protective antigen) that is proposed to have affinity with a domain described in Bacillus anthracis. This bacterial pathogen produces a toxin composed of three distinct proteins, i.e., protective antigen, edema factor, and lethal factor. PA is secreted in a precursor form, which can heptamerize and form a channel in membranes, allowing the other two factors to enter the target cell. The anthrax PA N-terminal domain is reported to be involved in carbohydrate binding [50]. Perhaps the best described domain of the CCp family is the scavenger receptor (SR) domain of CCp3. SR is found in several extracellular receptors, like the macrophage scavenger receptor type I. The domain is involved in protein–protein interactions and also interacts with other SR domains [45, 51, 52].

As mentioned above, the ApicA domain is an apicomplexan-specific cysteine-rich module. The NEC domain, on the other hand, appears in a wide range of animal proteins in combination with other cysteine-rich segments [53, 54], such as neurexins [18], fibrillar collagen α globular domain [55] from vertebrates and sponges and the fibrinogen family of proteins [53]. The NEC domain is also known as fibrinogen C domain, which represents the C-termini of fibrinogen ß and γ chains and which is predicted to interact, among others, with fibronectin type 3, laminin, discoidin, and other fibrinogen C domains [VI]. Fibrinogen is a dimeric protein involved in blood clotting and platelet aggregation [54, 56]. When the PfCCp protein family was described in 2004 [14], it was the first instance where the NEC domain was reported as a stand-alone form. Several animal proteins functioning as lectins contain the NEC domain, such as ficolin and intelectin [57, 58], suggesting that it is involved in sugar binding.

Multi-protein complexes of malaria parasites

Multi-protein complexes play important roles in regulating the activity and stability of polypeptides. Moreover, protein complexes function in establishing cell–cell contacts and thus are of great importance for interactions between pathogens and host cells. For intracellular pathogens, such as the majority of apicomplexan parasites, recognition, adhesion, and invasion of host cells are essential steps during infection by zoite stages (sporozoites, merozoites, bradyzoites, tachyzoites) and are often mediated by protein–protein interactions. Examples of apicomplexan multi-protein complexes are summarized in Table 1.

Table 1 Constituents of multi-protein complexes in Plasmodium falciparum, their known or proposed functions, respective orthologs in other apicomplexan species, life-cycle expression, and gene identifier
Full size table

One well-studied example of an apicomplexan multi-protein complex is represented by protein interactions comprising the so-called glideosome that has been described in T. gondii and several Plasmodium species (e.g., [59–63], reviewed in [64, 65]. This striking protein network mediates gliding motility, which is required for the movement of the parasite along host tissue barriers and for host cell invasion. A group of transmembrane proteins belonging to the TRAP family plays a key role within the glideosome due to their transmembrane linkage of the intracellular actin-myosin motor of the infective parasite stages with the host cell surface. The P. falciparum TRAP family includes five members, namely the sporozoite micronemal proteins PfTRAP (thrombospondin-related anonymous protein; [66]), PfTLP (TRAP-like protein; [62, 67]), and PfTREP (TRAP-related protein; [68]) as well as the ookinete-specific micronemal protein PfCTRP (CSP and TRAP-related protein; [69]) and the merozoite-specific protein PfMTRAP [60]). Unifying features of P. falciparum TRAP proteins include a transmembrane domain, a short acidic cytoplasmic tail with one or more aromatic residues near the C-terminus, an extracellularly exposed array of adhesive domains comprised of type-1 thrombospondin-like domains and, with the exception of TLP, von-Willebrand factor A domains. When secreted by the zoite micronemes, the proteins become integrated into the apical surface of the parasite and bind to host cell receptors. For PfTRAP, PfMTRAP, and PfTLP, it was shown that they are connected with the intracellular actin-myosin motor complex of the parasite, which is located between the parasite outer and inner membrane, by interacting with the adapter protein aldolase [59–62] (reviewed in [65]). The motor complex consists of actin and the class XIV myosin MyoA and features several additional proteins; for instance, the glideosome-associated proteins of merozoites, GAP45, and GAP50 that in concert with PfMTIP (myosin tail-interacting protein) link MyoA to the inner membrane complex [59, 60, 70] (reviewed in [65]).

Similar glideosome structures were described in T. gondii, where the MIC (micronemal) proteins represent major key players and, indeed, much of the mechanistics were worked out using the experimental system of this pathogen. The TRAP family protein TgMIC2 is responsible for linking receptors of the host cell surface with the intracellular motor complex of tachyzoites via aldolase binding [71]. TgMIC2 forms a hexameric complex with TgM2AP, which might be necessary for TgMIC2 transport and is considered prototypical of other TRAP family members [72]. Further members of the T. gondii MIC family are the soluble TgMIC1 and TgMIC4. The two proteins interact with TgMIC6 that acts as an escorter to ensure correct targeting of both proteins to the micronemes [73]. TgMIC8 appears to escort the soluble adhesin TgMIC3 [74]. Furthermore, TgMIC6 interacts with aldolase, pointing at an involvement in host cell invasion [75]. Noteworthy, orthologs for most glideosome components were recently also identified in most other apicomplexan parasites (reviewed in [76]), including their recent discovery in Eimeria tenella [77]. The MIC orthologs EtMIC4 and EtMIC5 also aggregate to a multimeric protein complex [78].

Another extensive complex involved in merozoite attachment is constituted by the P. falciparum MSP (merozoite surface protein) group. The majority of these peripheral proteins are secreted into the PV of schizonts and subsequently binds to the surface of developing merozoites via interaction with GPI-anchored proteins such as PfMSP1, which exhibits two EGF-like domains [79]. Surface-associated proteins include the MSP3/6 group, the MSP7 family as well as Pf41, a member of the cysteine-rich motif superfamily (reviewed in [76]). This adhesive MSP-based multi-protein complex, which consists of multiple secreted proteins assembling around a GPI-anchored EGF-domain protein, shows superficial similarities with the here discussed cell surface-associated PfCCp multi-protein complex.

During invasion, a focal adhesion to the host cell, referred to as tight junction or moving junction, is established in order to enable the movement of the apicomplexan parasite toward the interior of the host cell [80]. In T. gondii, this attachment is mediated by a protein complex comprised of the highly conserved micronemal apical membrane antigen 1 (AMA1) [81], which is inserted into the parasite plasma membrane, and of the four rhoptry neck proteins RON2/4/5/8 [82, 83]. Recent data led to the provocative hypothesis that the receptor for TgAMA1 located at the host cell surface is provided by TgRON2, which upon secretion becomes inserted into the host cell plasma membrane, whereas TgRON4, TgRON5, and TgRON8 locate to the cytosolic face of the host cell [83, 84]. Although the intracellular motor complex had been proposed to associate with the site of the tight junction, direct interaction partners have not been identified [85]. An additional role of the moving junction complex might be the exclusion of certain host cell plasma membrane proteins from the newly formed PV [86, 87]. AMA1 and RON orthologs are also present in other apicomplexan parasites [77, 84, VII]. In P. falciparum, the PfRON2, PfRON4, PfRON5, and PfAMA1 have been proposed to form a multimeric complex [88, 89]. PfAMA1 is translocated to the merozoite surface before invasion (reviewed in [76]). A recent study suggests a role after reorientation and establishment of the tight junction, which succeed the initial weak contact between merozoite and red blood cell (RBC). Therefore, PfAMA1 might be involved in the secretion of rhoptry contents which presumably form the nascent PV [89].

Another rhoptry-derived protein complex crucial for parasite survival is the Rhop complex of P. falciparum, which features two subgroups of protein associations, referred to as PfRhopL and PfRhopH complexes (low and high molecular weight rhoptry protein complex, respectively, [90]). PfRhopH complex is formed by the non-covalent-associated polypeptides PfRhopH1 (155 kDa), PfRhopH2 (140 kDa), and PfRhopH3 (110 kDa) [91, 92]. PfRhopH1 is encoded by one gene of the rhoph1/clag gene family (named after the cytoadherence-linked asexual gene, clag9, [93]) comprised of five polymorphic paralogous genes, while PfRhopH2 and PfRhopH3 are encoded by a single gene [94, 95]. The binding of the PfRhopH complex to the erythrocyte membrane is essential for merozoite invasion as in vitro studies showed, employing inhibitory antibodies raised against PfRhopH3 as well as PfRhopH3 peptides masking the putative receptor at the RBC surface [90, 96, 97]. In addition to cytoadherence, the PfRhopH complex is also proposed to participate in PV formation (reviewed in [98]). The low molecular weight PfRhopL complex comprises rhoptry-associated proteins such as PfRAP1 (83 kDa), PfRAP2 (40 kDa), and PfRAP3 (37 kDa) [99], and likewise PfRhopH complex binds to the erythrocyte cell surface [100]. Both PfRhopH and PfRhopL complexes bind non-covalently to the rhoptry-associated membrane antigen (RAMA), which is anchored to the inner face of the rhoptry membrane via a GPI tail, thereby mediating the retention of the protein complexes prior to rhoptry discharge [90, 101]. Both RhopH and RhopL complex constituents have been shown to possess homologs in several Plasmodium species [102–104], but not in other apicomplexan parasites [88]. However, RhopH1/Clag exhibits a region homologous to RON2, suggesting a common function of this region within the respective protein complex [88].

A prominent multimeric protein association modifying a host cell membrane is responsible for the formation of knob-like protuberances of the erythrocyte membrane after infection of malaria parasites (reviewed in [105, 106]). Knob formation in erythrocytes is limited to infections by P. falciparum and underpins virulence mechanisms causing the symptoms of severe malaria. Erythrocyte knobs mediate cytoadherence of these cells to a variety of host cell receptors, among them CD36 and ICAM-1 of endothelial cells, leading to clotting and capillary damage and subsequently resulting in organ failure (reviewed in [107]). Fundamental for knob formation is the knob-associated histidine-rich protein KAHRP, which is deposited at the erythrocyte plasma membrane facing the cytoplasm [108, 109]. KAHRP seems to assist the presentation of PfEMP1 (erythrocyte membrane protein 1) via binding to the PfEMP1 cytosolic region [110]. PfEMP1 is an antigenically highly diverse transmembrane protein comprised of multiple adhesion modules that enable the protein to bind to the different types of host cell receptors (reviewed in [111, 112]). KAHRP further interacts with actin and spectrin, thereby establishing a linkage of the knobs with the erythrocyte cytoskeleton [113, 114]. Components of the erythrocyte cytoskeleton in turn interact with further parasite-derived proteins such as PfEMP3, mature-parasite-infected erythrocyte surface antigen (MESA), and Pf322, which presumably alter the deformability of the cytoskeleton (reviewed in [106]).

Multimeric complexes of sexual stage parasites: which role do they play?

Initially phenotypic studies on PfCCp-KO lines were performed in order to reveal protein function. It was anticipated that a knock-out phenotype would manifest at the gametocyte stages, thus mirroring the stage specificity of PfCCp transcript and protein expression. Surprisingly, despite the loss of all PfCCp proteins in the respective PfCCp-KO lines, all lines exflagellated upon gametocyte activation and apparently normal ookinete and oocyst formation was observed [14, 20, 21] (A. Kuehn, G. Pradel, unpublished observations). In PfCCp2- and PfCCp3-KO mutants, however, no sporozoites were detected in the salivary glands. PfCCp4-KO parasites, on the other hand, did not reveal any differences in the mosquito-specific development compared to wild-type parasite [20], and no phenotype analyses on mosquito-specific parasites stages have so far been performed for PfCCp1-KO and PfFNPA-KO. Hence, PfCCp2-KO and PfCCp3-KO parasites appeared to be blocked in the transition of midgut sporozoites to the salivary glands [14], and thus, the respective knock-out phenotype for these mutants manifested in the blockage of sporogonic mosquito-specific parasite stages, in which PfCCp proteins were not present.

Similar phenotypes were observed, when the orthologous genes were knocked out in P. berghei. Loss of function mutants were generated for PbCCp1/LAP2, PbCCp2/LAP4, PbCCp3/LAP1, and PbCCp4/LAP6 [35] as well as for two double knock-out mutants, PbCCp1/CCp3 and PbCCp1/CCp4 [36]. For all mutants, the formation of sporozoites in the midgut oocysts of parasite-infected mosquitoes was aborted [35, 36]. Interestingly, sporulation of PbCCp3/LAP1 (termed PbSR in this study) appeared to be normal in in vitro assays, pointing to the involvement of mosquito factors in the loss-of-function phenotype of this protein [38].

Up to date, there is no explanation for the obvious difference in the stage of expression and stage of knock-out phenotype for the PfCCp proteins. It might represent a delayed phenotypic effect, in which reduced fertilization events result in a lower number of ookinetes and subsequently oocysts, in turn leading to a diminishing number of salivary gland sporozoites. An alternative explanation is offered by the observation that select PbCCp/LAP proteins of P. berghei associate with the crystalloid, a cytoplasmic aggregation in the developing ookinete [37–39, 115], which has been suggested to serve as storage for protein required during oocyst maturation [116, 117]. PbCCp3/LAP1 (PbSR) seems to play a role in crystalloid formation leading to the assumption that the impaired sporulation of PbCCp-KO mutants results from the absence of crystalloids [38]. Notably in this context, both the LCCL domain protein family and the ookinete crystalloid are inventions specific to the apicomplexan clade.

In view of our current data and despite the observed delayed knock-out phenotypic effect, we presume that the PfCCp functions lie in mediating molecular interplay during sexual reproduction. We hypothesize that the PfCCp proteins assemble to surface-associated multimeric protein complexes via intermolecular binding between select adhesion motifs. The secreted PfCCp proteins, together with Pfs230, are kept to the gametocyte surface by interacting with the GPI-anchored Pfs48/45 and, after activation, with Pfs25 to the macrogamete surface (Fig. 1b). Such assembly of secreted adhesion proteins around a GPI-anchored EGF-domain protein would be analog to the MSP-based multimeric complexes on the surface of merozoites.

The PfCCp complexes might be involved in important adhesive processes of macrogametes during sexual reproduction and transmission. On the one hand, the complexes might play a role in promoting contact between the emerging gametes within the blood meal. The physical contact between mating gametes is a requirement to initiate fertilization but can represent a major challenge for midgut parasites under natural conditions due to the low number of gametocytes taken up by the mosquito [118]. For example, the complexes might support the binding of macrogametes to microgametes due to interaction of select PfCCp proteins with Pfs230 or Pfs48/45 on the microgamete surface. In this context, recent studies described the identification of the microgamete protein GCS1/HAP2 in P. berghei that enables gamete fusion, but not initial binding between the two mating partners, which appears to involve other adhesion proteins [119, 120].

On the other hand, the complexes might mediate the protection of the exposed macrogametes from the aggressive environment of the mosquito midgut by forming a barrier between parasite and midgut content. During egress from the host erythrocyte, the emerging gametes become vulnerable to the factors of the mosquito midgut, resulting in an approximately 300-fold loss in parasite number between gametocytes and ookinetes [121]. Harmful factors of the midgut content include digestive enzymes, midgut bacteria as well as potential immune defense molecules of the mosquito. It was recently shown that a subset of immune genes become upregulated by the mosquito microbial flora, which then reduce the infection level of malaria parasites [122]. Of particular interest is an immune gene family of fibrinogen-related proteins (termed FREP), which is up-regulated after challenge with bacteria or malaria parasites. One family member, FBN 9, interacts with gut bacteria as well as with rodent and human malaria parasites in the mosquito midgut epithelium [123]. Further harmful components of the blood meal include human antibodies and complement proteins. Noteworthy, complement of the alternative pathway can be active for several hours post-feeding, as was shown in rodent and avian malaria models [124–126]. The predicted binding of selected sexual stage protein-specific adhesion domains with complement control protein modules might provide a key to identifying potential protection mechanisms of the sexual stage parasite.

Concluding remarks

Apicomplexan parasites exhibit a variety of life-cycle stage-dependent protein complexes, all of which are inventions specific to the clade. Proteins of these complexes either show lineage specificity, like the Rhop proteins of the malaria parasite erythrocytic stages, or possess adhesion motifs of wider conservation, as found in the TRAP and the CCp families, which were acquired by the pathogen via lateral gene transfer. The multi-protein complexes are utilized by the parasite for the different types of intercellular contact, ranging from attachment to and gliding on target cells to host cell invasion and cytoadherence of infected erythrocytes. While the majority of multimeric complexes are so far described for the infective parasite stages, a new type of complex involving the PfCCp protein family was recently identified in the sexual stages of the malaria parasite P. falciparum. The complex-forming proteins support the adhesiveness of macrogametes, but their detailed functions remain elusive. Follow-up studies will reveal if these complexes mediate intercellular contact, the way it was described for the complexes of other life-cycle stages, and which receptors they would use for these binding events. Another aspect worth investigating is the possible connection of these adhesion proteins to intracellular complexes by yet unknown linker proteins, thereby either participating in motility or in protein trafficking and/or re-cycling. Whichever role the complex-forming PfCCp proteins play for the malaria parasite sexual stages, the evolutionary conservation of the numerous adhesion modules and architectures of those proteins strongly argues that the proteins exhibit similar functions in other apicomplexan species.