The malaria parasite Plasmodium utilizes specialized proteins for adherence to cellular receptors in its mosquito vector and human host. Adherence is critical for parasite development, host cell traversal and invasion, and protection from vector and host immune mechanisms. These vital roles have identified several adhesins as vaccine candidates. A deficiency in current adhesin-based vaccines is induction of antibodies targeting non-conserved, non-functional and decoy epitopes due to the use of full length proteins or binding domains. To alleviate the elicitation of non-inhibitory antibodies, conserved functional regions of proteins must be identified and exploited. Structural biology provides the tools necessary to achieve this goal, and has succeeded in defining biologically functional receptor binding and oligomerization interfaces for a number of promising malaria vaccine candidates. We describe here the current knowledge of Plasmodium adhesin structure and function, and how it has illuminated elements of parasite biology and defined interactions at the host/vector and parasite interface.
Adhesion of Plasmodium malaria parasites to host cells is critical in mediating traversal through cellular barriers, cellular invasion and protection from host clearance. To traverse host cells, the parasite disrupts the host membrane, glides through the cytosol, and exits the cell (Mota et al., 2001). This movement contrasts cellular invasion, during which the parasite engages the host cell and invaginates the membrane to form the parasitophorous vacuole, where the parasite resides as it develops internally (Baum et al., 2008; Cowman et al., 2012). Adhesins also mediate rosetting and cytoadherence of infected erythrocytes within the human host, effectively providing protection from immune clearance by the spleen (Rowe et al., 2009).
Proteins that provide these functions contain defined domains and are organized into families based on domain similarity. Atomic resolution crystal structures of adhesins, alone and in complex with host receptors, have delineated protein folds and structurally conserved segments. More importantly, structures have identified receptor binding interfaces, multimeric contacts and mechanisms of receptor binding. Evaluation of protein ultrastructure and oligomeric state complemented crystallographic studies by illuminating an overall view of protein shape, flexibility and multimeric assembly in solution.
Here, we describe the current knowledge of how protein structures of malaria adhesins provide wide ranging cell adherence functions with defined roles in the parasite life cycle. We begin with families of proteins that are proposed to coat the parasite at multiple life cycle stages, followed by the multiple adhesive roles of the sporozoite surface coat protein. Proteins the parasite utilizes to invade red blood cells (RBC) are then detailed. The role of adhesion during parasite invagination of the host membrane during cellular invasion is discussed, along with proteins that provide a link between extracellular receptors and the parasite's internal actin motor. The review concludes with adhesins exported to the surface of infected RBCs that bind numerous receptors to provide rosetting and cytoadherence capabilities. We link the common structural features that serve similar adhesive purposes (Fig. 1), and highlight multimeric assembly necessary for adhesion.
Proteins functioning as a parasite surface coat
The mosquito blood meal initiates the growth of the parasite in its vector, as gametes enter the proper environment for fertilization. Surface coat adhesins are immediately required, and gametes utilize members of the conserved apicomplexan 6-cysteine family for recognition and attachment during fertilization (van Dijk et al., 2001; 2010). 6-cysteine family members are also expressed on the surface of sporozoites and merozoites and are likely redundant adhesins at cellular invasion interfaces (Ishino et al., 2005; Sanders et al., 2005; Taechalertpaisarn et al., 2012). Members in this family contain a varied number of s48/45 domains, a predominantly beta-sheet fold (Fig. 2A; Arredondo et al., 2012; Tonkin et al., 2013). The crystal structure of Pf12, a merozoite protein, illuminated only minor contacts between the tandem s48/45 domains, and showed that the domain linker lacks conformationally restrictive residues (Tonkin et al., 2013). This suggests that movement between two tandem s48/45 domains is possible, and thus that the link between the two domains is flexible. Furthermore, Pf12 forms heterodimers with another 6-cysteine member, Pf41, suggesting that 6-cysteine family proteins may function as pairs (Taechalertpaisarn et al., 2012).
The domain fold of 6-cysteine family members contrasts the fold of the surface coat proteins on the fertilized zygote and ookinete within the mosquito midgut. After fertilization, zygotes are capable of adhering to one another through tubule extensions coated in the protein p25, an adhesin that contains four tandem evolutionarily conserved epidermal growth factor-like (EGF) domains (Fig. 2B; Saxena et al., 2006; Rupp et al., 2011). p25 and its homologue, p28, are subsequently expressed on the surface of the motile ookinete, the parasite form that results from the developing zygote (Saxena et al., 2007). As ookinete surface proteins, p25 and p28 are thought to provide protection from mosquito proteolytic defence mechanisms and to facilitate adhesion to the midgut membrane (Tomas et al., 2001; Saxena et al., 2007). Adherence likely depends on p25/p28 binding to laminin, which constitutes a large portion of the midgut epithelium (Vlachou et al., 2001). Knocking out both p25 and p28 is required to severely limit midgut crossing, suggesting both proteins are critical for adhesion and are functionally redundant (Tomas et al., 2001).
Crystallization of P. vivax p25 identified extensive contacts between monomers in the crystal lattice (Fig. 2B; Saxena et al., 2006). Contacts between p25 monomers form a ‘triangular prism’ that may link the proteins on the ookinete surface to form a specific coat structure (Tomas et al., 2001; Saxena et al., 2006; 2007). It is important to note that the way in which a protein packs into a crystal can result in artificial contacts that may not be functional in vivo. Crystal packing interfaces and oligomeric states should be assessed in solution to support inferences made from crystal structures. The intermolecular contact residues are conserved among p25 orthologues and monomers can self-interact in solution, supporting a physiological role for multimerization (Siden-Kiamos et al., 2000; Saxena et al., 2006; 2007). Together, these studies suggest that multimeric complexes of p25 and/or p28 form a protective coat and initiate and maintain interactions with the midgut cell wall (Tomas et al., 2001).
In contrast to the surface coat of the ookinete, the sporozoite does not express any characterized EGF containing proteins, but instead utilizes circumsporozoite protein (CSP), discussed below. However, EGF-like domains are used again on the surface of the merozoite after release from the human liver. The merozoite surface protein family (MSP), of which MSP1 is the most abundant, coats the merozoite and provides adhesive functions. MSP1 is a 185–215 kDa protein that undergoes extensive proteolytic processing resulting in multiple subunits non-covalently linked to the C-terminal portion designated MSP1-19 (Fig. 2C). At the point of RBC invasion, all subunits of MSP1 are released except for MSP1-19 (Holder and Freeman, 1984; Kauth et al., 2003). MSP1-19 contains two tandem EGF domains that are compact and rigid, and is linked to the membrane by a GPI anchor (Fig. 2C; Chitarra et al., 1999; Morgan et al., 1999). The structure of MSP1-19 in complex with an antibody that effectively coats the merozoite surface identified a region of MSP1-19 that is exposed on the merozoite (Pizarro et al., 2003). This region may contact Band 3, the erythrocyte receptor for MSP1 (Goel et al., 2003).
The CSP surface coat on sporozoites
CSP comprises the surface coat of the sporozoite during its journey through the mosquito salivary glands to invasion of human hepatocytes (Kappe et al., 2004). CSP possesses a distinct domain architecture: an N-terminal domain, a sequence designated Region I, a stretch of tetra-amino acid repeats, a C-terminus comprised of two regions designated II and III, and an evolutionarily conserved thrombospondin type-I repeat (TSR) domain. A GPI anchor links CSP to the membrane (Wang et al., 2005).
The domains of CSP function in two distinct steps of the sporozoite's life cycle. Initially, the CSP N-terminal domain/Region I binds heparan sulfate on the mosquito salivary glands (Sidjanski et al., 1997; Sinnis et al., 2007; Ghosh and Jacobs-Lorena, 2009; Armistead et al., 2011). Upon initiation of the mosquito blood meal, sporozoites are injected from the salivary glands into the human and migrate to the liver. Within the liver, the second step of CSP host cell recognition depends on proteolytic removal of the N-terminal domain and timed exposure of the C-terminal TSR domain (Coppi et al., 2005; 2011). This exposure is correlated with recognition of increased sulfation of heparan sulfate proteoglycans on hepatocytes (Coppi et al., 2007; 2011). Mutant sporozoites that constitutively express the cleaved version of CSP, exposing the TSR domain, continually migrate within the dermis and do not reach the liver. This suggests that the N-terminal domain prevents the CSP C-terminal region from binding improper receptors in the dermis, allowing the sporozoite to properly target the liver (Coppi et al., 2011). Retaining the N-terminus prior to hepatocyte recognition also shields the functional C-terminal binding region from antibody recognition, representing an in vivo structural mechanism for protection of critical binding domains (Coppi et al., 2011).
Ultrastructural analysis suggests that CSP maintains a flexible rod-like structure (Plassmeyer et al., 2009). Small percentages of CSP appear to form dimers and oligomers in solution, suggesting that intermolecular contacts may function in sporozoite coat formation (Plassmeyer et al., 2009). Extensive interdomain contacts are formed between the TSR domain and Region III (Fig. 2D; Doud et al., 2012). However, the TSR and Region III construct is monomeric in solution. Thus, CSP oligomerization may depend on the N-terminal domain, Region I and/or Region II, and internal repeat regions have been proposed to interact to form a protective coat for the parasite (Godson et al., 1983).
Surface proteins that adhere to cellular receptors during invasion
After release from the liver into the bloodstream, the merozoite recognizes RBCs and must form a tight link with the host cell membrane. Two distinct protein families are involved at this stage: the erythrocyte binding like (EBL) and the reticulocyte binding like protein homologue (RH). Members of both families engage specific RBC receptors for invasion (Sim et al., 1990; Adams et al., 1992).
The EBL family contains a conserved domain architecture (Adams et al., 1992). Receptor binding requires essential adhesive domains unique to Plasmodium termed Duffy binding like (DBL) found in Region II (RII) of EBL proteins. P. falciparum contains multiple EBL family members (PfEBA-175, PfEBA-140, PfEBL-1 and PfEBA-181), each containing two tandem DBL domains. In contrast, P. vivax is thought to be limited to a sole EBL member, Duffy binding protein (PvDBP), which contains a single DBL domain. However, recent sequencing of field isolates of P. vivax have revealed certain isolates carry a duplication of the PvDBP gene (Menard et al., 2013), and others carry a novel EBL ligand that also has a single DBL domain (Hester et al., 2013). The DBL domains have a characteristic boomerang shape stabilized by extensive disulfide bridging (Tolia et al., 2005; Singh et al., 2006; Higgins, 2008; Khunrae et al., 2009; Batchelor et al., 2011; Juillerat et al., 2011; Lin et al., 2012; Vigan-Womas et al., 2012; Malpede et al., 2013). In addition to Region II, EBL proteins contain a segment of uncharacterized structure (regions III–V), a structured C-terminal cysteine rich domain (Region VI), a transmembrane domain, and a cytoplasmic region (Adams et al., 1992; Withers-Martinez et al., 2008).
PfEBA-175-RII contains two tandem DBL domains that bind glycophorin A in a sialic acid dependent manner (Camus and Hadley, 1985; Sim et al., 1990; 1994; Klotz et al., 1992; Orlandi et al., 1992). The crystal structure of PfEBA-175-RII in complex with a sialic acid containing glycan revealed the receptor-binding pockets are formed at the dimer interface of PfEBA-175-RII (Fig. 2E; Tolia et al., 2005). This suggests a dimer of PfEBA-175 assembles around dimeric glycophorin A during invasion. Multimeric assembly of PfEBA-175 enhances binding to glycophorin A and is augmented by regions outside of RII (Salinas and Tolia, 2014; Wanaguru et al., 2013).
The P. vivax EBL member, PvDBP-RII, shares similar molecular structures and functional characteristics with PfEBA-175-RII (Tolia et al., 2005; Batchelor et al., 2011; 2014; Wanaguru et al., 2013). PvDBP-RII is monomeric in the absence of its receptor, the Duffy antigen receptor for chemokines (DARC), and dimerization of PvDBP-RII is driven by receptor binding (Fig. 2F; Batchelor et al., 2011). Crystal structures of PvDBP-RII in complex with the ectodomain of DARC, and complementary solution studies, demonstrated formation of two distinct complexes: a heterotrimer of two PvDBP-RII and one DARC, and a heterotetramer of two PvDBP-RII and two DARCs (Batchelor et al., 2014). These complexes are intermediates in a multi-step binding mechanism. In both structures, DARC is sandwiched by two PvDBP-RII molecules facilitating receptor-induced dimerization, suggesting a conserved binding mechanism between two EBL ligands despite distinct dimeric architectures (Fig. 2E and F). However, the receptor binding sites and dimer interfaces lie on distinct faces of the DBL domains.
Crystal structures of PfEBA-140 Region II also identified receptor binding regions (Fig. 2G; Lin et al., 2012; Malpede et al., 2013). PfEBA-140-RII is monomeric in the absence of its receptor glycophorin C (Lobo et al., 2003; Lin et al., 2012; Malpede et al., 2013) and further studies of receptor-bound PfEBA-140 and other EBL ligands are necessary to determine if multimeric assembly occurs upon receptor binding. The receptor binding pockets in PfEBA-140 are located in a distinct region of the DBL fold compared with the pockets used by either PfEBA-175-RII or PvDBP-RII. This suggests that the DBL fold can create multiple binding pockets to bind a wide variety of receptors. PfEBA-140 also exhibits polymorphisms that affect receptor-specificity and/or binding affinity (Mayer et al., 2002; Maier et al., 2009), and a polymorphism that maps to one receptor-binding pocket suggests a structural basis for altered specificity (Malpede et al., 2013). Additional merozoite surface proteins containing the DBL domain bind RBCs and maintain the conserved DBL architecture (Hodder et al., 2012). The specific role of these DBL containing proteins during invasion is not completely understood.
EBL ligands function primarily in the blood stage, but one unique member, designated Apical membrane antigen/erythrocyte binding like (MAEBL), enables sporozoite entry into the mosquito salivary glands (Blair et al., 2002; Kariu et al., 2002; Fu et al., 2005). Sporozoites lacking MAEBL cannot attach to the salivary glands, but retain normal motility, supporting a specific role in attachment to host cells (Kariu et al., 2002). Unlike most EBL proteins that contain DBL domains, the tandem adhesive domains in MAEBL contain homology to domains observed in apical membrane antigen 1 (AMA-1), described below. MAEBL may be involved in a similar function at the moving junction as AMA-1, and may incorporate functional elements of AMA-1 and the EBL family.
Less is known about RH family structure as there is no sequence similarity to known domains and crystal structures are not available. Initial insight provided by a low resolution surface envelope of the RBC binding element from a RH family member (P. yoelli 235) suggested that this segment may resemble Region II of the EBL family (Gruber et al., 2011). However, the sequence similarity between the RH binding element and EBL Region II is low and further study is necessary to define the structure and function of this family. This is particularly relevant as the receptor Basigin, which is engaged by PfRH5, appears essential for invasion of multiple isolates and strains of P. falciparum (Crosnier et al., 2011).
Adhesion during membrane invagination to form the parasitophorous vacuole
In addition to forming a connection with the host cell membrane, the parasite must maintain this link during membrane invagination and engulfment into the parasitophorous vacuole, where it will develop internally. This membrane invagination requires the formation of the moving junction, a ring-shaped link between parasite and host membranes that begins at the apical end of the merozoite and migrates to the posterior end as the parasite invades. AMA-1 is a unique adhesin that is released during invasion, and is observed at the moving junction (Triglia et al., 2000; Lamarque et al., 2011). AMA-1 is expressed on sporozoites, and antibodies targeting this adhesin inhibit hepatocyte invasion, suggesting a functional role in the liver stage (Silvie et al., 2004). Initial attempts to knockout AMA-1 were unsuccessful, however, more recent studies suggest AMA-1 is dispensable for blood stage growth (Bargieri et al., 2013).
AMA-1 is a type I integral membrane protein, and the extracellular portion is composed of three Domains (I, II and III) defined by disulfide bridging (Hodder et al., 1996). The cytoplasmic region is conserved among Plasmodium species, and contains two cytoplasmic C-terminal tyrosines that are proposed to function in signalling through phosphorylation, potentially providing communication with downstream effectors (Remarque et al., 2008). Domains I and II of AMA-1 adopt the evolutionarily conserved PAN domain fold, which mediates diverse protein-protein and protein-carbohydrate interactions (Fig. 2H; Tordai et al., 1999; Bai et al., 2005; Pizarro et al., 2005). Domain III adopts a novel fold (Pizarro et al., 2005).
AMA-1 is not known to engage a host cell receptor. Instead, it binds the parasite expressed RON2, which is secreted from the merozoite rhoptries into the RBC membrane during invasion (Cao et al., 2009; Srinivasan et al., 2011). Crystallization with a peptide comprising the minimal RON2 binding domain clarified a specific hydrophobic groove of Domain I that engages the receptor (Vulliez-Le Normand et al., 2012). Displacement of a loop in Domain II is required for RON2 binding to AMA-1, and this loop may function to protect the binding site from host immunity until RON2 is available (Vulliez-Le Normand et al., 2012). AMA-1 binds the RON2 peptide with nanomolar affinity, and this interaction is strengthened by high local concentration and clustering of several independent AMA-1/RON2 interactions to facilitate moving junction formation (Vulliez-Le Normand et al., 2012). This clustering of interactions provides necessary avidity to maintain the parasite's link to the host membrane as the parasite propels itself into a host cell.
The link between host cell receptors and the parasite's actin motor
The formation of the moving junction is accompanied by parasite movement that links an external host cellular receptor with the parasite's internal actin motor (Kappe et al., 1999; Baum et al., 2006). The thrombospondin-related anonymous protein (TRAP) family provides this connection. TRAP family members contain a combination of adhesive evolutionarily conserved Von-Willebrand factor type A (VWA) and TSR domains. The cytoplasmic portion of TRAP members links to the actin motor through aldolase providing movement (Fig. 2I; Buscaglia et al., 2003; Kappe et al., 2004; Bosch et al., 2007). Sporozoites deficient in the hallmark member of this family, TRAP, are capable of adhering to host cells, suggesting that TRAP function is required for invasion but not attachment (Sultan et al., 1997).
The VWA and TSR domains of PfTRAP and PvTRAP have been captured in various conformations by X-ray crystallography (Song et al., 2012). The VWA domains appear rigid but can adopt open and closed conformations with the closed conformation correlated with divalent cation binding. In contrast, the TSR domains are highly flexible, evidenced by complete disorder in PfTRAP and two distinct orientations in the open conformations of PvTRAP. This led the authors to suggest that TSR domains become structured upon receptor binding and the observed conformational changes may be necessary to initiate motility into the cell (Song et al., 2012).
An individual TRAP member is proposed to function at different stages of the parasite's life cycle (Baum et al., 2006). Movement through the mosquito midgut cellular barrier is the function of CTRP binding to laminin, as genetic deletion of CTRP severely disrupts this ability (Yuda et al., 1999a,b; Mahairaki et al., 2005). TRAP, present on the sporozoite, is required for movement into the mosquito salivary glands via binding to Saglin and also mediates hepatocyte invasion (Sultan et al., 1997; Wengelnik et al., 1999; Kappe et al., 2004; Ghosh et al., 2009). The parasite must also actively invade RBCs during blood stage growth, and MTRAP provides motor function at this juncture. The receptor for MTRAP has been defined as Semaphorin-7A, and the two appear to bind in a 2:2 complex (Bartholdson et al., 2012). Ultrastructural analysis demonstrates that MTRAP maintains an extended conformation that may provide flexibility that facilitates the active movement of the parasite into a host cell (Uchime et al., 2012). This flexible form may represent the overall shape of the TRAP family.
Adhesive proteins exported to the infected RBC surface
Within the RBC, the parasite actively exports variant surface antigens to the RBC membrane (Leech et al., 1984; Baruch et al., 1995; Smith et al., 1995; Su et al., 1995). The var gene family encodes erythrocyte membrane protein 1 (PfEMP1), which provides critical adhesive properties that provide protection from immune function (Baruch et al., 1995; Smith et al., 1995). The var gene family in P. falciparum contains approximately 60 unique variants in each haploid parasite genome, of which one variant is predominantly expressed at a given time (Chen et al., 1998; Scherf et al., 1998; Gardner et al., 2002). Each member contains an N-terminal domain along with a varied combination of two major Plasmodium specific adhesion domains: the DBL fold, and a cysteine rich interdomain region (CIDR) that is also highly alpha helical (Fig. 2J; Smith et al., 2000; Heddini et al., 2001; Higgins, 2008; Klein et al., 2008; Khunrae et al., 2009). Similar to the DBL domains of merozoite adhesins, the PfEMP1 DBL domains appear to have a wide receptor range. The CIDR domains are also involved in receptor recognition, further increasing the receptor repertoire engaged by PfEMP-1.
Solution structural analysis by small-angle X-ray scattering (SAXS) of a recombinant full length extracellular region of one PfEMP1 variant, IT4VAR13, showed a flexible, extended molecule (Brown et al., 2013). In contrast, examination of a second PfEMP1 variant, VAR2CSA, showed that the DBL-CIDR domains assemble into higher order structures that may involve domain-domain interactions (Srivastava et al., 2010). The full length extracellular region of VAR2CSA was required for highest affinity receptor binding, supporting a crucial role for interdomain interactions (Srivastava et al., 2010). Crystal structures of constructs containing the tandem N-terminus, DBL and adjacent CIDR domain displayed extensive contacts between the individual elements of the PfEMP1 architecture (Juillerat et al., 2011; Vigan-Womas et al., 2012). The functional role for inter-domain contact is not clear, however, the domain-domain contacts represent a view that contrasts the previously invoked model of ‘beads on a string,’ in which all PfEMP1 domains are flexible and capable of binding individual, unique receptors. Although individual domains are proposed to bind unique receptors, the role of each domain as a piece of the whole PfEMP1 protein is not fully understood.
Exposure on the parasite surface marks Plasmodium adhesins as prime targets for host immunity and vaccines. The focus of current anti-malaria vaccines on full length adhesins or complete binding domains grants the immune system with access to decoy and non-inhibitory epitopes (Chen et al., 2013), diminishing the production of inhibitory antibodies. Limiting or eliminating access to decoy and non-functional epitopes is critical towards developing rapid and effective immunity. Structural work has successfully defined conserved receptor binding sites and multimeric interfaces that can be specifically targeted to focus an antibody response. Functional interfaces have been defined by crystallizing the adhesin with receptors, and with characterized inhibitory antibodies. Techniques that focus the immune response to target specific epitopes are currently in development. Specific epitope targeting can be achieved by mutating immune-dominant non-inhibitory epitopes and by shrouding non-inhibitory epitopes with glycosylation (Ntumngia and Adams, 2012; Sampath et al., 2013). Similar approaches have effectively defined broadly neutralizing epitopes to viral antigens, supporting the efficacy of these techniques (Corti and Lanzavecchia, 2013). Continued structural effort to identify critical and conserved contacts between parasite and host proteins provides an excellent opportunity in the development of vaccines that elicit strain-transcendent, highly inhibitory antibodies to malaria parasites.
We are grateful to J.P. Vogel, M.M. Paing and J. Park for advice on the manuscript. This work was supported by National Institutes of Health Grant AI080792 (to N.H.T.) and by a National Science Foundation Graduate Research Fellowship (to B.M.M.) under Grant DGE-1143954. The authors declare that there are no conflicts of interest.