Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
Malaria parasites are obligate intracellular parasites whose invasive stages select and invade the unique host cell in which they can develop with exquisite specificity and efficacy. Most studies aimed at elucidating the molecules and the mechanisms implicated in the selection and invasion processes have been conducted on the merozoite, the stage that invades erythrocytes to perpetuate the pathological cycles of parasite multiplication in the blood. Bioinformatic analysis has helped identify the members of two parasite protein families, the reticulocyte-binding protein homologues (RBL) and erythrocyte binding like (EBL), in recently sequenced genomes of different Plasmodium species. In this article we review data from classical studies and gene disruption experiments that are helping to illuminate the role of these proteins in the selection-invasion processes. The manner in which subsets of proteins from each of the families act in concert suggests a model to explain the ability of the parasites to use alternate pathways of invasion. Future perspectives and implications are discussed.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Plasmodium the causative agent of malaria is an obligate intracellular parasite with a complex life cycle involving both an invertebrate and a vertebrate host. The invasive forms of the malaria parasites are like all members of the phylum Apicomplexa (Levine, 1971) characterized by a unique set of organelles the rhoptries, micronemes and dense granules that play a crucial role in the invasion process. The contents of these organelles are thought to be important in enabling the parasite to recognize the appropriate host cell, penetrate it effectively and develop successfully within it. During the life cycle of Plasmodium the parasite needs to invade a range of different host cells. The ookinete needs to penetrate the mosquito midgut endothelium while the sporozoite is required to both traverse mosquito salivary gland and traverse or invade skin and liver cells in the vertebrate host. The merozoite invades erythrocytes where they develop to generate more merozoites, the cyclical multiplication phase of the life cycle that is the cause of the disease symptoms typically associated with malaria. Variations in the contents of the apical organelles are thought to be responsible for the different invasion properties displayed by the invasive forms of Plasmodium.
The processes by which the invasive forms of the parasite recognize and penetrate a host cell are still not completely understood and most of our current knowledge is based on studies of the merozoite (Ladda et al., 1969; Bannister et al., 1975; Dvorak et al., 1975; Aikawa et al., 1978; Miller et al., 1979; Mitchell and Bannister, 1988; Sherman, 1999). The main steps in the invasion process are: (i) initial merozoite binding, reorientation and deformation of erythrocyte, (ii) formation of a junction (the irreversible commitment of the parasite to invasion) and (iii) parasite entry. The preferred dogma for the initial interaction between a free merozoite and erythrocyte is by a random collision. Once in contact proteins located on the surface of the merozoite have been implicated in mediating a reversible interaction between parasite and erythrocyte. After binding to the erythrocyte, the merozoite reorients itself such that the apical end is in contact with the erythrocyte membrane. Simultaneously, the parasite induces locally the deformation of the erythrocyte. An indentation of the membrane at point of contact with the parasite is observed. Once reorientation has occurred micronemal and rhoptry proteins mediate specific host parasite interactions that lead to tight junction formation. Invasion rapidly proceeds with the discharge of the contents of micronemes, rhoptries and finally dense granules. It is thought that rhoptries are discharged after the micronemes and assist in the formation of the parasitophorous vacuole. To complete the process of invasion, merozoite serine proteases cleave erythrocyte membrane protein Band 3 and cause a localized disruption in the cytoskeleton, thus allowing the parasite to enter the host cell (Roggwiller et al., 1996). A parasitophorous vacuole membrane forms in the junction area, and the junction becomes a small annulus through which the parasite moves to enter the expanding parasitophorous vacuole. Upon completion of the parasite entry, the tight junction disappears and the parasitophorous vacuole membrane and erythrocyte membrane fuse and separate, completing the entry process. The contents of the dense granules are thought to be discharged only after the parasite has completed its entry, and to be implicated in the modification of the host cell (Torii et al., 1989; Culvenor et al., 1991). The invasion process is now complete and the parasite resides and develops in the erythrocyte.
The parasite invades different host cells in the other stages of its life cycle. However, the mechanisms implicated in these experimentally relatively inaccessible stages have been much less well studied than those of the merozoite. In the mosquito vector, the ookinete penetrates the mosquito peritropic matrix and enters the midgut of the mosquito (Huber et al., 1991). Ookinete invasion mechanisms differ from hepatic and erythrocytic invasion in that no recognizable rhoptries are found and no parasitophorous vacuole appears to form around the internalized parasite after invasion. After the oocyst develops in the midgut, it ruptures to release mature sporozoites (Meis et al., 1992) into the haemocoelic cavity, which will migrate to, adhere to and invade the salivary glands (Vanderberg, 1974; Golenda et al., 1990; Vaughan et al., 1992). During a blood meal, mosquitoes deposit sporozoites in the host's skin. The majority of these mobile parasites will traverse skin cells, enter the blood circulation to be carried to the liver where they will traverse Kupffer cells and hepatocytes until a suitable host hepatocyte is invaded. Different parasite proteins have been implicated in these processes, including proteins such as MAEBL, EBA175, and the apical complex protein Py235 (Grüner et al., 2001; Preiser et al., 2002; 2004) that are also involved in merozoite invasion.
Within the mammalian host, the parasite has to overcome two major hurdles in the invasion process. First it has to migrate to or attach quickly to its host cell in order to evade immune attack, and second it has to recognize the hepatocyte or the red blood cell, where it can develop and multiply. Indeed this is the only time during an infection where the parasite is directly exposed to the host immune system. In vitro studies suggest that the sporozoites and merozoites are labile, as they appear to lose infectivity quickly if they do not invade a host cell (Johnson et al., 1980). Whether this is the case in vivo still needs to be confirmed.
From a clinical point of view, the rate of successful invasion of red blood cells by merozoites is a major factor in the pathogenesis of malaria, as it has direct bearing on asexual blood stage parasite densities and dynamics. The process of red cell invasion is relatively easily amenable to laboratory studies, as compared with the invasion of hepatocytes or the invasive processes in the mosquito. Consequently, most data on host cell invasion by Plasmodium are centred on red blood cell invasion. We therefore confined ourselves to the events leading to the invasion of red blood cells by blood stage merozoites.
In normal peripheral blood, the two principal erythrocyte populations encountered are usually reticulocytes and normocytes. The reticulocyte is a submature red blood cell where active protein synthesis, mainly of haemoglobin, still takes place. The ribosomal content, which appears as a fine reticulum, hence the name, diminishes as the cell matures over a couple of days to form the normocyte, the principal blood erythrocyte population. Reticulocytes represent less than 1% of the erythrocyte population in normal peripheral blood. Different Plasmodium species show distinct preferences for the type of erythrocyte they invade (Garnham, 1966). These differences are clearly seen in the case of the four Plasmodium species known to infect humans. Plasmodium vivax only invades reticulocytes that are Duffy blood group positive while Plasmodium falciparum, associated with the most clinically severe forms of malaria, invades all erythrocytes regardless of age. Plasmodium malariae prefers mature erythrocytes and Plasmodium ovale is again restricted to reticulocytes. Similar variations in red cell preferences are also seen in non-human Plasmodium species. The simian Plasmodium knowlesi invades normocytes (Hegner, 1938) while Plasmodium cynomolgi can invade all erythrocytes but has a slight preference for reticulocytes. The rodent malaria parasite Plasmodium berghei invades preferentially reticulocytes, while Plasmodium chabaudi invades both reticulocytes and mature cells and Plasmodium vinckei infects mature erythrocytes. Differences in host cell preferences can even be observed in the same species, for example, the Plasmodium yoelii 17X YA and 17X YM lines (Fahey and Spitalny, 1984). YA infections yield a self-limiting, non-lethal infection in BALB/c mice, where the parasites are restricted to reticulocytes, whereas YM infections yield a fulminating almost always lethal infection with parasites found in erythrocytes of all ages. There seems to be some relationship between the type and age of erythrocytes invaded and parasite virulence in its broadest sense. Parasite loads seen in P. vivax infections are generally low compared with those resulting from P. falciparum infections, possibly because reticulocytes make up only 1% of the total number of host erythrocytes. Similar differences in parasitaemias are seen with the YA and YM lines of P. yoelii.
Ultimately parasites are able to discriminate between different types of red blood cells, probably through specific receptor–ligand interactions. While a subset of proteins such as MSP1 and AMA1 seem to mediate the initial merozoite/erythrocyte interaction (Holder and Freeman, 1982; Peterson et al., 1989), it is during the subsequent steps (i.e. just before junction formation) when the specific host cell recognition observed in the different parasites occurs. Work initially done in P. yoelii, P. vivax and P. knowlesi has identified two sets of parasite proteins the Duffy-binding protein (DBP) and the 235 kDa rhoptry proteins located in the micronemes and rhoptries, respectively, that appear crucial in determining the red cell type invaded by a merozoite. However, recent studies have shown that two merozoite surface proteins, MSP7 (Tewari et al., 2005) and MSP8 (Shi et al., 2005), are also involved in red blood cell selection by the rodent malaria parasites P. berghei and P. yoelii respectively. This suggests a dynamic interplay between merozoite surface proteins, micronemal and rhoptry proteins during merozoite invasion. These different sets of proteins are conserved in different Plasmodium species. In this review we focus on our current understanding on how the EBLs (erythrocyte binding-like proteins/ligands), the family of the DBP, and the RBLs (reticulocyte binding-like proteins), the family of the 235 kDa rhoptry proteins (summarized in Table 1), give the malaria parasite an adaptive advantage enabling it to select alternative red blood cell invasion pathways and to evade host immunity.
Table 1. The superfamilies of reticulocyte-binding protein homologues (RBLs) and the family of erythrocyte-binding proteins/ligands (EBLs) summarized in various Plasmodium spp.
Elegant work using gene knockouts in P. falciparum has shown that members of both these gene families seem to interact with specific receptors on the erythrocyte surface (Kaneko et al., 2000; Reed et al., 2000; Gilberger et al., 2003a). Furthermore, disruption or deletion of a specific member leads to the utilization of a different erythrocyte receptor by the parasite (Kaneko et al., 2000; Gilberger et al., 2003a). This change of receptor use is analogous to the previously observed in vitro selection of P. falciparum on enzymatically treated erythrocytes leading to the selection of parasites that utilize a different invasion pathway (Camus and Hadley, 1985b; Dolan et al., 1990; 1994; Gaur et al., 2003). This ability to change the host cell receptors used for invasion gives the merozoite a significant advantage for finding a suitable host cell. All the work published to date has focused on the specific interaction between a single parasite ligand and host cell receptor leading to the concept of one parasite ligand – one invasion pathway. In the case of P. vivax it is clear that members of both gene families are required for successful invasion as the parasite is only able to invade reticulocytes (mediated by RBP) that are Duffy blood group antigen positive (mediated by the DBP) (Galinski and Barnwell, 1996), and recently the effect of one gene family member on another has been demonstrated in P. falciparum (Stubbs et al., 2005). It is therefore important when discussing invasion mechanisms of Plasmodium species that the role of both members of the two gene families are considered.
The RBL homologues
The first member of the RBL superfamily of Plasmodium proteins involved in recognition of host cell receptors by the parasite was first discovered in P. yoelii (termed Py235), and could be directly involved in specific differentiation between red blood cell subtypes (reviewed by Grüner et al., 2004). Homologous proteins have since been discovered in other Plasmodium species including P. vivax and P. falciparum. These proteins are thought to play a crucial role in erythrocyte recognition and invasion (Table 1).
Plasmodium yoelii and other rodent malaria parasites. The 235 kDa rhoptry protein family of P. yoelii (Py235) was first identified using monoclonal antibodies prepared from mice resistant to the virulent YM strain of P. yoelii (Freeman et al., 1980). Two of these antibodies were able to confer protection in passive transfer experiments by modulating the infection, restricting the YM parasites to predominately reticulocytes, thereby ensuring the survival of the host mouse strain (Freeman et al., 1980). Immunofluorescence microscopy and subsequently immunoelectron microscopy showed that these antibodies recognized a protein located in the rhoptries of the P. yoelii merozoites (Freeman et al., 1980; Oka et al., 1984). Furthermore, vaccination of BALB/c with the immunopurified 235 kDa protein also protected against a subsequent challenge with the P. yoelii YM line, converting fulminating infections to a reticulocyte-restricted self-limiting infection (Holder and Freeman, 1981). This suggested that the antibody to this protein protected against challenge infection by preventing the entry of parasites into mature erythrocytes without affecting reticulocyte invasion. In addition, immunofluorescence reactivity of anti-Py235 antibodies to other malaria parasites indicated that the protein may be conserved in other rodent Plasmodium species (Holder and Freeman, 1984). Screening of an expression library using antibodies specific to Py235 led to the identification of the gene coding for this protein (Keen et al., 1990). Once the gene had been identified it became rapidly apparent that many distinct py235 genes were present in the genome of P. yoelii. The copy number was estimated to be between 20 and 50 based on PCR analysis and hybridization experiments (Borre et al., 1995; Khan et al., 2001a), somewhat higher than the 14 copies identified by the P. yoelii genome-sequencing project (Carlton et al., 2002). Recent work using quantitative PCR has confirmed a copy number closer to that predicted by the genome project in a range of P. yoelii strains (Iyer et al., 2006).
Sequence comparisons between py235 members showed high overall conservation with regions of variability interspersed throughout the coding region (Narum et al., 2001; Khan et al., 2001a) (Fig. 1). Chromosome mapping has shown that all py235 genes are located in the subtelomeric region of some but not all chromosomes (Owen et al., 1999). From the genome sequencing of other rodent parasites it has become apparent that orthologues of this family are present in P. berghei and P. chabaudi. A similar copy number was found in P. berghei while there seemed to be fewer members in P. chabaudi (Sanger Centre sequencing project). The predominant red blood cell-expressed member of Py235 recognized by the monoclonal antibody has been identified by Mass spectrometry (Ogun et al., 2006). Further evidence for the involvement of Py235 in red cell recognition came from the demonstration that the protein specifically bound to neuraminidase-resistant, chymotrypsin- and trypsin-sensitive erythrocyte receptors (Ogun and Holder, 1996; Ogun et al., 2000). The protective effect of anti-Py235 antibodies has given clear evidence that Py235 plays an important role in virulence. Differences in the py235 repertoires and transcription profile observed between the YM and YA lines are consistent with such a role (Preiser and Jarra, 1998; Khan et al., 2001a). Whether these differences account for the different invasion phenotype and virulence observed still needs to be determined. Studies of the transcriptional profile of py235 in individual parasites have provided further insights into the role Py235 play in parasite biology. A new form of clonal phenotypic variation in which a single schizont produces merozoites that transcribe distinct members of py235 suggests that the parasite can adapt to variations in the host cell environment and evade host immunity (Preiser et al., 1999).
Plasmodium vivax. P. vivax only invades reticulocytes that are Duffy positive. There has recently been a report on P. vivax infections in a Duffy-negative population, indicating that alternative pathways may indeed be present for P. vivax (Ryan et al., 2006). While the recognition of the Duffy antigen is mediated by the DBP (see below) the recognition of the reticulocyte is mediated by another set of proteins called reticulocyte-binding proteins (RBP) 1 and 2 (Galinski et al., 1992). The PvRBP proteins have been localized to the apical end of the merozoite but it is not yet clear whether they are located in the micronemes or the rhoptries. PvRBP-1 and 2 have been postulated to form a complex that mediates adhesion and recognition of the reticulocyte and homologues are also found in the closely related simian parasite P. cynomolgi (McCutchan et al., 1984; Okenu et al., 2005). Sequence comparison with the py235 genes of P. yoelii showed that PvRBP-1 and PvRBP-2 shared weak but significant sequence identity as well as structural features, including net charge and hydrophobicity (Galinski et al., 2000). The importance of these conserved features is not yet clear.
Plasmodium falciparum. For P. falciparum, it had been difficult to establish whether members of RBL were also present because of low sequence homology and the lack of any cross-reactive immunological reagents with those known in other Plasmodium species. Once the P. falciparum genome sequence became available, six members of the RBL family were identified, the rhoptry protein homologue 1 (PfRH1) also named normocyte binding protein 1 (NBP1) (Rayner et al., 2001), the reticulocyte-binding protein homologues 2a and 2b (PfRBP-2a and 2b) (Rayner et al., 2000; Triglia et al., 2001a), the rhoptry protein homologue 3 (PfRH3) a possible pseudogene in a number of laboratory cultured parasite isolates (Taylor et al., 2001), the reticulocyte-binding homologue 4 (PfRH4) (Kaneko et al., 2002) and the reticulocyte-binding protein homologue 5 (PfRH5) (Cowman and Crabb, 2006). For this review and for the sake of uniformity we will name them PfRH1, PfRH2a, PfRH2b, PfRH3, PfRH4 and PfRH5 respectively.
All P. falciparum members of the RBL family except PfRH-4 share an approximately 500-amino-acid high-homology region initially identified when comparing Py235 and PvRBP (Keen et al., 1994) (Fig. 1). There is some conservation between P. falciparum and P. vivax RBL proteins (though less so for PvRBP-2) (Rayner et al., 2005). Overall gene sequence homology between the different members is low but they contain a number of conserved blocks of amino acids which clearly identify them as members of this gene family. The expression pattern and apical location of all members are consistent with a role during the invasion process. While PfRH1 as well as PfRH2a and 2b seem to locate to the rhoptries of the merozoite PfRH-4 seems to have a micronemal location (Kaneko et al., 2002), although recent evidence place it in the rhoptries (Stubbs et al., 2005). There is some evidence that at least a small proportion of PfRH2a is moved to the merozoite apical pole after schizont rupture (Triglia et al., 2001a), akin to the behaviour of Py235 (Oka et al., 1984; Sam-Yellowe, 1996). PFRH3 is considered to be a pseudogene and no corresponding protein is expressed during the erythrocytic cycle, although evidence suggests it may be expressed in sporozoites (Florens et al., 2002). Little is known about PfRH5, apart from its gene sequence (Cowman and Crabb, 2006).
In both P. yoelii and P. vivax the RBL are believed to play an important role in the recognition of specific receptors on the red blood cell. Direct binding of the protein to the erythrocyte has been shown, and in the case of P. yoelii antibodies against Py235 have been shown to inhibit invasion (Table 1). PfRH1 has been shown to directly bind a sialic acid containing trypsin-resistant receptor on the erythrocyte and antibodies raised against this protein inhibit merozoite invasion of trypsin-treated erythrocytes (Rayner et al., 2001). So far direct binding of PfRH2a or 2b to erythrocytes has not been demonstrated although there is evidence that antibodies can differentially inhibit invasion in some parasite strains, and certain peptide sequences from these proteins mediate stronger erythrocyte binding than others (see Table 1) (Triglia et al., 2001a; Ocampo et al., 2004). The most convincing evidence came from gene knockout experiments showing that PfRH2b recognizes a unique chymotrypsin-sensitive receptor and that it functions independently of other members (Duraisingh et al., 2003a). Unlike P. vivax, PfRH2a and 2b may not form a complex as they appear to have distinct independent functions. At least one strain of P. falciparum lacks one gene, and disrupting one of the pair does not affect the other (Triglia et al., 2001a; Duraisingh et al., 2003a).
During the study of P. falciparum RBLs it became clear that different parasite lines display sequence variation between the same family members ranging from a few amino acid changes to large deletions (Taylor et al., 2002; Lobo et al., 2006). Furthermore, some members were found to be completely absent in some strains (Duraisingh et al., 2003a). These variations might represent a type of antigenic diversity and/or mediate changes in the binding properties of the protein. The latter is supported by a recent study which associates different polymorphism in the RBL with changes in the invasion pathways used by the parasite (Lobo et al., 2006). Further complexity arises due to the fact that the transcription and expression pattern of PfRBLs vary between different parasite lines (Taylor et al., 2002; Duraisingh et al., 2003a). These variations have implications in the way these proteins are utilized as changes in the expression of PfRBL could lead to the recognition of different receptors on the erythrocyte surface resulting in merozoites with different invasion potentials.
Plasmodium reichenowi. Comparison of the RBL gene family from the chimpanzee parasite P. reichenowi shed further light on the evolution as well as conservation of these parasite ligands. Like its most close relative P. falciparum it has five members of this gene family (Rayner et al., 2004a). Overall the genes are conserved but there are some differences, most notably the orthologue of PfRH1 has a large number of deletions as well as some insertions leading to over 200 stop codons, so that the longest ORF in PrRH1 is only 645 base pairs compared with about 8 kb for the ORF of the P. falciparum gene. The PrRH2b gene, on the other hand, is highly conserved while for PrRH2a only the C-terminal end is conserved. PrRH3 lacks the stop codons found in the PfRH3 orthologue distinguishing it from the P. falciparum pseudogene, while overall conservation of PrRH4 is approximately 86% identity. The diversity between the members in P. reichenowi and P. falciparum indicate a high selection pressure (Rayner et al., 2004a).
The EBL/DBP homologues
A number of Plasmodium proteins involved in adhesion of the parasite to host cell receptors are characterized by a conserved cysteine-rich domain of approximately 35 kDa that has been shown to directly mediate binding to host cell receptors (Fig. 1) (Chitnis et al., 1996). This domain was first described in the DBP of P. vivax and has therefore been named Duffy binding-like (DBL) domain (Fang et al., 1991). DBL domains have been identified in two groups of parasite-encoded ligands, the variant P. falciparum erythrocyte membrane protein 1 (PfEMP-1) (Su et al., 1995) and the superfamily of EBLs expressed in the invasive stages of many Plasmodium species (see Table 1) (Adams et al., 2001).
Erythrocyte binding-like proteins are thought to play a crucial role in erythrocyte recognition, junction formation and invasion. All the EBLs are characterized by conserved exon–intron structures, splicing boundaries and contain two extracellular regions that have conserved cysteines and hydrophobic amino acid residues (Adams et al., 1992). This suggests a conserved three-dimensional structure of this domain, and a common evolutionary origin.
Plasmodium vivax. The unique restriction of P. vivax to Duffy-positive erythrocyte is mediated by a single-copy gene that codes for the 140 kDa DBP (Fang et al., 1991). This parasite-encoded ligand specifically binds the Duffy blood group antigen on human erythrocytes (Wertheimer and Barnwell, 1989). Subsequent work has shown that DBP specifically interacts with a 35-amino-acid-long region of the Duffy receptor (Chitnis et al., 1996) and that this interaction is independent of sialic acid. Competition experiments with purified Duffy glycoprotein or pre-treatment of erythrocytes with a monoclonal antibody against a Duffy determinant (Wertheimer and Barnwell, 1989) indicated that DBP plays a role after merozoite reorientation during the invasion process (Miller et al., 1979). High levels of antibodies against DBP are detected in patients exposed to repeated doses of P. vivax (Fraser et al., 1997), and there is some evidence that there is some protection mediated by these antibodies
The extracellular domain of the PvDBP (and for that matter all members of the EBL gene family) can be divided into six regions (I to VI) based on amino acid homology to PkDBP (below) (Adams et al., 1990). Two cysteine-rich domains (in regions II and VI) contain conserved cysteines and aromatic amino acid residues (Fig. 1). Erythrocyte binding assays have shown that specific sequences in one of the cysteine-rich domains (region II) in the DBP are involved in the binding to the Duffy antigen (Chitnis and Miller, 1994; Chitnis et al., 1996; Ranjan and Chitnis, 1999). Furthermore, recent work using site-directed mutagenesis has given a more detailed map of the essential amino acids that mediate the specific interaction (Vanbuskirk et al., 2004a; Hans et al., 2005). In P. vivax field isolates only limited sequence variability in the DBL domain of DBP has been observed (Tsuboi et al., 1994). None of the observed changes lead to alternate receptor specificity, but rather plays an important role in immune evasion (Vanbuskirk et al., 2004b). Structural motifs in this region have also been elucidated for receptor binding (Singh et al., 2003).
Plasmodium knowlesi and P. cynomolgi. Proteolytic cleavage of Duffy determinant from erythrocytes have been shown to inhibit P. knowlesi invasion of human erythrocytes (Miller et al., 1975; Mason et al., 1977; Wertheimer and Barnwell, 1989), indicating that this antigen is required by this simian parasite to invade human erythrocytes. On the other hand P. knowlesi invasion of rhesus erythrocytes is not dependent on the Duffy determinant demonstrating the ability of the parasite to utilize other simian-specific erythrocyte receptors for invasion. Erythrocyte binding assays identified a 135 kDa DBP (PkDBPs) of P. knowlesi (Miller et al., 1979) that was subsequently shown to be cross-reactive with the PvDBP (Haynes et al., 1988). Unlike the case for P. vivax it soon became clear that there were multiple members of DBPs in P. knowlesi. Antibodies to a PkDBP protein revealed the presence of a total of three PkDBPs, termed α, β and γ (Adams et al., 1990; 1992), all sharing conserved DBL domains with 70% homology to that of P. vivax. Only the DBL domain of PkDBPα has been shown to bind to the Duffy receptor while the DBLs of the other two proteins are thought to bind to so far uncharacterized receptors specific to the simian erythrocytes. Direct disruption of PkDBPα results in parasites that are unable to invade human erythrocytes while at the same time having no measurable effect on the invasion of rhesus red blood cells, confirming the existence of alternative invasion pathways in P. knowlesi (Singh et al., 2005). In the case of the PvDBP, it is region II of PkDBPα that mediated binding to the Duffy receptor. The fact that P. knowlesi has more then one DBP is thought to account for the ability of this parasite to utilize additional receptors. Antibodies raised against region II of PkDBPα effectively inhibit invasion of human erythrocytes as well as partially inhibit invasion of rhesus erythrocytes by P. knowlesi merozoites.
Plasmodium falciparum. Unlike P. vivax, P. falciparum can invade Duffy-negative and -positive erythrocytes equally well and is known to utilize a number of distinct receptors (Miller et al., 1977; Mitchell et al., 1986; Perkins and Holt, 1988; Dolan et al., 1994). The ability of P. falciparum to utilize a wider range of erythrocyte receptors is at least in part due to the expansion of the number of EBL genes. Even before the completion of the P. falciparum genome-sequencing project multiple distinct EBL members had been identified. Currently, six members have been identified: EBA175 (Camus and Hadley, 1985b), BAEBL (EBA140) (Mayer et al., 2001; Lobo et al., 2003; Maier et al., 2003), JESEBL (EBA181) (Adams et al., 2001; Gilberger et al., 2003b; Mayer et al., 2004), PEBL (EBA165) (Triglia et al., 2001a), EBL1 (Peterson and Wellems, 2000) and MAEBL (Kappe et al., 1998). For consistency we will call them EBA175, EBA140, EBA181, EBA165, EBL1 and MAEBL. All members of this EBL superfamily with the exception of MAEBL contain a DBL domain indicating that MAEBL is not involved in DBL-mediated erythrocyte binding. The P. falciparum EBLs are all located in the subtelomeric region of different chromosomes (Gardner et al., 2002), a region that contains many genes associated with parasite virulence and pathology. Like other members of this superfamily the overall exon–intron structure and overall structural characteristics are conserved. While all DBL domains are characterized by 12 conserved cysteine residues, there are a number of differences that may have important functional implications. All P. falciparum EBLs have duplicated DBL domains (termed F1 and F2, Fig. 1), unlike the single copy in PvDBP (Sim et al., 1994), indicating more complex proteins (Adams et al., 2001). The dual DBL domain of EBA175 has been crystallized forming an interdomain channel, and shown to constitute a binding cleft (Tolia et al., 2005), although only the F2 region has been shown to have any binding activity (Sim et al., 1994; Ockenhouse et al., 2001). Based on current evidence EBA165 has to be considered an untranslated pseudogene as it contains a number of frameshift mutations and no evidence for a protein product has been obtained to date (Triglia et al., 2001a).
Receptor binding specificities have been shown for four EBLs of P. falciparum. Using a range of enzyme treatments of erythrocytes it was clearly demonstrated that EBA175 binding to erythrocytes was dependent on the sialic acid components on Glycophorin A (Camus and Hadley, 1985b), indicating that Glycophorin A is a receptor for this protein. This was ultimately confirmed by the direct disruption of EBA175 in the parasite using gene knockouts. In this case the parasite was unable to utilize Glycophorin A as a receptor during the invasion process (Reed et al., 2000; Duraisingh et al., 2003b). Using similar approaches EBA140 which has a 30% homology to EBA175 was shown to bind to Glycophorin C (Mayer et al., 2001; Lobo et al., 2003; Maier et al., 2003). EBA181 which has a 37% similarity to EBA140 binds to an uncharacterized erythrocytic sialoglycoprotein (Gilberger et al., 2003b; Mayer et al., 2004). Additional evidence for the role of EBLs in erythrocyte binding and invasion came from the observation that antibodies against EBA175 and EBA140 can inhibit erythrocyte binding and invasion (Jakobsen et al., 1998; Sim, 1998; Narum et al., 2002). Furthermore, immunization with EBA175 gives some protection in a primate challenge model (Jones et al., 2001).
It seems that without selection pressure cultured parasites mostly use the EBA175/GlyA pathway, with a minority of parasites exhibiting alternate ligand receptor combinations. However, the parasite is able to use other receptors as treatment of erythrocytes with enzymes like Neuraminidase (to remove sialic acid) or trypsin to remove other protein components does not completely ablate invasion (Camus and Hadley, 1985b; Dolan et al., 1990; 1994; Rayner et al., 2001; Gaur et al., 2003). While overall parasite replication is greatly reduced on initial exposure to these treated erythrocytes, the parasite adapts with time and is eventually able to invade the treated erythrocytes with similar efficiency as when grown in untreated erythrocytes. This would indicate that the parasite adapts to utilize a new receptor on the erythrocyte, giving strong evidence that P. falciparum can use multiple different invasion pathways. Interestingly, if selection pressure is subsequently removed the parasite slowly reverts back to the Glycophorin A-dependent pathway. These in vitro observations using cultured parasites somewhat contrasts with some of the observations made recently using parasites directly obtained from infected individuals. Short-term culture-adapted field isolates from India (Okoyeh et al., 1999), Kenya (Nery et al., 2006) and Brazil (Lobo et al., 2004) show that alternate ligand receptor combinations are mostly used for invasion while similar studies performed in the Gambia (Baum et al., 2003a) and Tanzania (Bei et al., 2007) indicate that the conventional EBA175/Glycophorin A is the prevalent pathway utilized. It is clear from this that the parasite can switch invasion pathways depending on the host environment it encounters. Direct evidence for both the switching of the invasion pathway and the role of the EBLs in this process came from some elegant work using parasite transfection. Disruption of EBA175, EBA140 and EBA181 prevents the interaction of the corresponding parasite ligand with its receptor forcing the parasite to use a different receptor. Nevertheless, these parasites are still able to invade erythrocytes efficiently (Reed et al., 2000; Gilberger et al., 2003a; Maier et al., 2003; Stubbs et al., 2005), indicating the flexibility that has evolved in this parasite. In the case of PvDBP it has been shown that any sequence variation of the DBL domain observed in different field isolates did not lead to any change in the receptor specificity but instead led to different antigenic variants (Vanbuskirk et al., 2004b). This may not always be the case in the EBLs of P. falciparum where it had been shown that polymorphism in EBA140 may have arisen in response to mutations in the human erythrocyte receptor (Mayer et al., 2002), and that furthermore changes in EBA140 and EBA181 allow for variations in binding characteristics to different erythrocyte receptors (Mayer et al., 2002; 2004).
Plasmodium reichenowi. Even though P. reichenowi is considered in evolutionary terms most closely related to P. falciparum it cannot invade human erythrocytes. This is despite the fact that orthologue of EBA175 showing 83% predicted amino acid identity to the P. falciparum gene has been identified in P. reichenowi (Ozwara et al., 2001). Modelling the crystal structure of PfEBA175 with the unique residues in the P. reichenowi homologue reveals differences in glycan binding sites that could explain the distinct host cells specificity of these parasites (Chattopadhyay et al., 2006). In addition, P. reichenowi EBA140 (Baum et al., 2003b), EBA181 and EBA165 (Rayner et al., 2004b) homologues have been identified. The EBA140 fragment isolated has a 92% deduced amino acid identity to P. falciparum EBA140 with a duplicated DBL domain (Baum et al., 2003b). Although P. reichenowi EBLs exhibit high sequence identity to its P. falciparum counterparts, these parasites do not bind to and invade human erythrocytes. This indicates that other parasite factors contribute to the specificity of the ligand to the host erythrocyte.
Plasmodium yoelii, P. berghei and P. chabaudi. Invasion assays of P. yoelii in Duffy knockout mice have revealed that the parasite uses the Duffy antigen to invade mature erythrocytes and an as yet unidentified receptor to invade reticulocytes (Swardson-Olver et al., 2002). The fact that in P. yoelii a single ORF containing a DBL domain was identified based on its shared characteristics with PvDBP, PkDBPs and EBA175 (Prasad et al., 2003) makes this a candidate protein for mediating this interaction. This is further supported by observation that it binds to mature erythrocytes using region II of its DBL domain (Prasad et al., 2003). Analysis of the recently obtained P. berghei and P. chabaudi genome databases had also revealed the presence of an EBL1 orthologue in each of these two species (Hall et al., 2005).
Other proteins related to EBL. A highly conserved member of the EBLs, MAEBL has been found in all Plasmodium species analyses so far. As mentioned above, MAEBL is unique in that it does not contain a DBL domain but is instead a hybrid sharing identity to AMA1 within the tandem duplicated cysteine-rich domains and the carboxyl cysteine-rich domain of EBL (Kappe et al., 1998; Blair et al., 2002). The duplicated cysteine-rich domains conserved with the AMA1 regions have previously been shown to have erythrocyte binding activity (Kappe et al., 1998). MAEBL, unlike the other EBLs, is located in the rhoptries of the merozoite indicating that it may have evolved a unique role in erythrocyte invasion (Noe and Adams, 1998).
Red blood cell selection and invasion: a multi-ligand model
The concept that complex ligand/receptor interactions must be implicated in the red cell selection/invasion process was raised by the discovery that P. vivax, a parasite with strong reticulocyte tropism, has an absolute requirement for the Duffy antigen that is present on all red blood cells. It is now known that the Duffy determinant is actually a chemokine receptor (DARC) whose expression is prevented in erythroid but not in non-erythroid cells as a result of mutation of the promoter at the GATA box (Tournamille et al., 1995). The notion that a single ligand–receptor interaction determines the invasive phenotype was further questioned by investigations where red blood cells, whose surface was modified with different proteases, or with blood collected from individuals with specific red blood cell phenotypes, clearly demonstrated that merozoites from clonal parasite lines can adopt alternative invasion pathways (Dolan et al., 1990; 1991). Today it is known that there are many ligands present in the parasite's genome. Functional and molecular analyses of parasites that have been adapted to different red blood cell types, or where specific ligands were genetically disrupted, are revealing that complex modifications in the expression patterns of both the RBL and EBL genes belie the alternative invasion pathways. It should be noted that relatively few host receptors have been identified as compared with the number of known parasite ligands in a given Plasmodium species, although there are strong indications that each ligand binds to a distinct receptor. The cost of maintaining a large number of genes and complex mechanisms to regulate their expression must be compensated by a selective advantage to the parasite. First, the human red blood cell surface exhibits a high degree of diversity between different individuals, and in a single person the red blood cell formula can vary by age and sex, and also in response to infection or other factors like nutrition (Henry, 1996; Berkow, 1997). Second, in endemic areas most individuals would have acquired a certain degree of immunity to the parasite that presumably encompass specific responses to ligands implicated in red blood cell invasion. Thus, a flexible invasive potential would enhance the parasite's reproductive success as this would optimize its ability to establish an infection that will persist and be transmitted in any individual. It should be noted that this reasoning also applies to the other invasive stages of the parasite.
We propose that a mechanism for red blood cell recognition and invasion based on distinct combinations of two or more ligand–receptor interactions would offer a good solution for maintaining a broad invasive potential at a minimal cost in genetic complexity. We further propose that a situation where the combined ligands are selected from the two families (RBL and EBL) would be optimal. This hypothetical scheme has the potential to account for a number of features associated with malaria infections. First, reliance on at least two distinct host receptors for target cell identification would help ensure that the parasite will predominantly invade the cells suitable for its development. Second, a defined combination between two ligands, one from the RBL and the other from the EBL families, is likely to target a different set of host receptors from those targeted by another combination. Thus, each RBL/EBL pair would potentially define a distinct invasion pathway. For instance, in P. falciparum where there are five RBL and five EBL members, one might envisage that the merozoites would have the potential to adopt 25 different invasion pathways or, to be more precise, to identify 25 erythrocyte subsets each expressing one of 25 distinct receptor combinations. Third, a given RBL/EBL combination is likely to be immunologically distinct from others, thereby providing the parasite with a way to evade immune responses that might have arisen against some RBL/EBL combinations.
The signal to invade would then depend on the quality of the interaction between the merozoite ligands and the red blood cell receptors. The avidity of each ligand to its receptor and the number of ligand–receptor combinations would determine the overall strength of the interaction, and the process of invasion will be triggered when this reaches a certain threshold. If this were the case, then the predilection of different Plasmodium species to red blood cells of different ages (from reticulocytes to mature and possibly ageing normocytes) might well be accounted for by variations in the density of some receptors at the surface of the red blood cell. The fact that red blood cells are devoid of DNA does not preclude the appearance and increase of surface receptor as the red cell ages, as it is known that age-related alteration of the membrane leads to the exposure of some proteins that were inaccessible in the young red blood cell (Killmann, 1964; Berlin and Berk, 1975; Ballas et al., 1986; Clark, 1988; Woolley et al., 2000). However, experimental approaches that aim to elucidate the nature of the parasite's red blood cell tropism are actually quite restricted. Observations in natural infections of humans are clearly precluded on ethical grounds. Interpretation of data from in vitro cultivation, in addition to being restricted to P. falciparum, is confounded by the variable characteristics of the red blood cells used. The latter often originate from different donors, and are used after prolonged storage. This might explain why culture-adapted strains of P. falciparum tend to have relatively stable invasion phenotypes that are reflected in the RBL and the EBL expression profiles. Freshly isolated parasites by contrast which had been maintained in culture for a short time were shown to display a much broader range of invasion pathways (Okoyeh et al., 1999).
At present it is not known whether the EBL and RBL interact with their respective receptors in separate independent events, or whether this occurs in the context of a macromolecular complex comprising both ligands. It has been recently proposed that the P. falciparum RBLs have an initial sensing function that serves to characterize the erythrocyte as suitable (Duraisingh et al., 2003a). This initial positive interaction would then lead to the recruitment of the high-affinity ligands, such as EBA175 or another member of the EBL family (Fig. 2), ultimately resulting in the release of micronemal contents and junction formation which initiate the process of red blood cell invasion. This is consistent with current understanding of red cell invasion by P. vivax where reticulocyte identification precedes recruitment of the DBP (Galinski and Barnwell, 1996), and in line with the proposed mode of action of the P. yoelii Py235 protein (Khan et al., 2001b). It must be stated that establishing the timelines of the numerous steps of a process normally completed within minutes is fraught with technical and interpretational hurdles. Irrespective of the finer details, disruption of either the interaction mediated by RBL as well as EBL would significantly inhibit merozoite invasion.
Diversity of invasive potential: control and consequences
The invasive potential of merozoites might differ between Plasmodium species, or genetically distinct lines within a species, not only as a result of difference in the mix of RBL and EBL genes present in the genome, but also through the nature of the mechanisms that control their expression.
Analysis of the Plasmodium genomes sequenced to date reveals that members of the RBL and EBL families differ substantially between species, whether this also occurs for different lines within a species awaits further sequencing. These differences imply that each Plasmodium species has adopted distinct evolutionary paths with respect to host cell invasion. At present it is not possible to conclude that a parasite whose genome has a large number of members in the RBL and/or EBL families would have a higher invasive potential than one with lower numbers, although this would be a logical consequence of the multi-ligand hypothesis. It would be tempting to suggest that the reduced virulence of P. vivax as compared with P. falciparum results from the presence of only 2 RBL + 1 EBL in the first and 5 RBL + 5 EBL in the second of these species. We feel that this should remain a matter of speculation until such a time where all the molecular players implicated in red blood cell selection are known, and when reliable means to measure invasive potential or indeed virulence reliably are devised. To date, there is no direct evidence that indicates that the presence or expression of a particular RBL and/or EBL is responsible for differences in the invasion potential of merozoites. There are nonetheless many studies suggestive of a role for these protein families in the modulation of host cell tropism.
In P. yoelii, where there are 14 RBL members (the Py235 family) but only a single EBL representative, investigations of Py235 expression uncovered a novel molecular strategy to generate merozoite diversity (Preiser et al., 1999). Transcriptional analysis of single parasites revealed that each merozoite expresses one or a restricted set of the Py235 repertoire and merozoites within a single schizont can differ in the Py235 genes they express. If it assumed that differential expression does indeed lead to functional invasive differences, then this mechanism of clonal phenotypic variation would ensure that the parasite population maintains its overall invasive potential with time both in the face of immune pressure and in the face of changes of the host red cell population. Recent investigations in P. yoelii have further linked the total amount of Py235 expressed in merozoites with the red blood cell invasion profile (Iyer et al., 2007). Parasite populations restricted to reticulocytes expressed low amounts of overall erythrocyte stage-expressed Py235, whereas those that invaded all red blood cells expressed high levels of Py235. It is interesting to note that modest increases (two- to threefold) in the expression level of a restricted subset of four Py235 genes were observed in P. yoelii populations restricted to reticulocytes by immunization with MSP8 (Shi et al., 2005). The functional significance of these alterations remains to be elucidated.
Investigations carried out using in vitro cultivated P. falciparum lines whose expression profiles were modified by enzymatically modified red blood cells are consistent with the selection of a parasite population which expresses appropriate ligands that can bind to the resultant erythrocyte receptor repertoire (Fig. 3). These experiments, along with others involving gene disruption of EBA175, uncovered an association between the upregulation of PfRH4 and the loss of sialic acid-dependent invasion (Stubbs et al., 2005). This reinforces the notion that regulating the expression of RBL and/or EBL might be one of the mechanisms used by different Plasmodium species to modulate merozoite invasive potential. Whether differential expression of these ligands, in a clonal phenotypic fashion akin to that observed for Py235 in P. yoelii, occurs in P. falciparum awaits detailed quantitative reverse transcriptase-PCR studies of single parasites. Nonetheless, data from numerous investigations strongly indicate that in most cases each P. falciparum merozoite expresses the whole or a major portion of the RBL and EBL repertoires (Taylor et al., 2002; Duraisingh et al., 2003a; Stubbs et al., 2005; Cowman and Crabb, 2006). Despite this apparent multiplicity of ligands, these merozoite populations appear to utilize only one invasive pathway defined by a distinct subset of the expressed RBL-EBL genes. Assuming that all the transcribed genes lead to the translation of equivalent amounts of protein, this conundrum might be resolved if the subcellular location and/or accessibility of the ligands is taken into account, as previously hypothesized (Duraisingh et al., 2003a; Baum et al., 2005). Thus, it has been proposed that only one or a small subset of the EBL and/or RBL expressed can be exposed in any one merozoite at the site where interactions with red blood cell take place (Duraisingh et al., 2003a; Baum et al., 2005). This ‘biased’ or selective exposure of a particular set of ligands might arise during the genesis of the apical complex. In this way, loss of a particular ligand from the ‘active site’ would liberate the place for another ligand (Fig. 4), and would consequently lead to a switch in the invasion phenotype. The notion of maintaining a limited number of parasite ligands in an accessible position would be consistent with a strategy of immune evasion for the merozoite, a parasite form fully exposed to the host's humoral defences.
Finally, modulation of the invasive potential can also be brought about by genetic diversity of the ligands. Sequence analysis of different parasite isolates has revealed there are extensive polymorphisms for EBL and RBL genes (Taylor et al., 2002; Nery et al., 2006). The functional significance of these polymorphisms, in terms of receptor recognition or invasion profiles, is yet to be demonstrated. Nonetheless, it is has been elegantly shown that mutation of a single amino acid in the binding domains of EBA140 and EBA181 was sufficient to significantly impair binding to a putative host receptor in vitro (Mayer et al., 2002; 2004).
Conclusions and perspectives
It is now clear that members from two protein families are involved in the host cell selection processes that define the red blood cell tropism of the malaria parasite merozoite. The diversity of these proteins within the genomes of different Plasmodium species allows to propose a hypothetical model where the combinations of ligands and the control of their expression could account for the distinct invasive behaviour of merozoites in the various species. However, neither conclusive evidence for this model nor the fine details of the interactions that underpin the selection of red blood cells for invasion are as yet at hand. There is little doubt that the ability to genetically modify the parasite has led to quantum advances in our perception of these invasive mechanisms, and that ongoing systematic analyses of parasites with disrupted RBL and/or EBL genes will yield further knowledge. This approach is unlikely to provide full insight into the nature of red blood cell selection, unless it is associated with studies where the red blood cell receptors are also defined. The red blood cell variants that are naturally found in nature constitute a rich source of material for these studies, but these are likely to be insufficient. For those parasites that can be maintained routinely in vitro (P. falciparum and P. knowlesi) the challenge will be to devise ways to obtain reproducibly homogenous population of red blood cells with defined receptor characteristics. Advances in the genetic manipulation of erythroblasts and in the generation of large numbers of erythrocytes from these progenitor cells might provide a solution.
The demonstration that switches to alternative invasion pathways where different ligands and receptors are implicated are easily obtained for P. falciparum under laboratory conditions, and evidence that this also occurs in parasites circulating in endemic residents, is of concern for malaria vaccines based on RBL or EBL proteins. The fact that parasites that do not express EBA175, the leading candidate for such vaccines, are still able to multiply (Duraisingh et al., 2003b) might translate in the selection of escape variants if this vaccine is deployed. However, pessimism must be tempered as it is not actually known whether these variant parasite lines that grow under laboratory conditions would be viable in vivo. It has nonetheless become necessary to consider inclusion of two or more RBL and EBL ligands in any vaccine intended to prevent interactions between the merozoite and the red blood cell.
Finally, one of the central assumptions concerning the RBL and EBL proteins is that they are solely involved in the invasion of red blood cells. This has primarily arisen from the fact that these proteins were first found associated to merozoites. Moreover, for many of these proteins binding to red blood cells could be demonstrated, and for some a red blood cell receptor was identified. Finally antibodies raised against a number of these proteins can inhibit or alter the tropism and invasive profiles of merozoites. There are two relatively recent studies that question the validity of this assumption. First, the expression of distinct subsets of Py235 genes (RBL family) was demonstrated (both at the transcriptional and protein levels) in the sporozoite and in the hepatic parasite (Preiser et al., 2002). Second, EBA175 (EBL family), an important P. falciparum anti-red blood cell invasion vaccine candidate, was also shown to be expressed on the surface of sporozoites and in the infected hepatocyte (Grüner et al., 2001). Whereas expression of these proteins might be expected in hepatic merozoites that are destined to invade red blood cells, their role in the biology of the sporozoite, a parasite form that interacts with mosquito salivary glands, the cells in the skin where it is deposited by the infective bite before invading hepatocytes, is yet to be explored. Systematic studies to determine which of the other members of the RBL and EBL families are expressed during the pre-erythrocytic stages of Plasmodium are currently underway. This might result in the identification of some ligands that specifically interact with a single host cell type, and others that might be part of the set of parasite proteins implicated in all invasive events.
In conclusion, elucidation of the molecular mechanisms underlying host cell tropism and invasion in Plasmodium parasites present researchers with a formidable challenge, both technically and intellectually. The resources that would be required to achieve this goal are justified by the central role these processes play in the survival of the parasite and in the possibility that the knowledge to be gained might yield novel and efficient strategies to control the infection.
The authors would like to thank Z. Bozdech for critical reading of the manuscript. This work was in part supported by the Biomedical Research Council, Singapore (04/1/22/19/364) (P.R.P.) and by the European Union (MALINV project n°012199) (L.R.). J.I. was a recipient of the Singapore Millennium Foundation Scholarship.