A Conserved Molecular Motor Drives Cell Invasion and Gliding Motility across Malaria Life Cycle Stages and Other Apicomplexan Parasites*

Apicomplexan parasites constitute one of the most significant groups of pathogens infecting humans and animals. The liver stage sporozoites of Plasmodium spp. and tachyzoites of Toxoplasma gondii, the causative agents of malaria and toxoplasmosis, respectively, use a unique mode of locomotion termed gliding motility to invade host cells and cross cell substrates. This amoeboid-like movement uses a parasite adhesin from the thrombospondin-related anonymous protein (TRAP) family and a set of proteins linking the extracellular adhesin, via an actin-myosin motor, to the inner membrane complex. The Plasmodium blood stage merozoite, however, does not exhibit gliding motility. Here we show that homologues of the key proteins that make up the motor complex, including the recently identified glideosome-associated proteins 45 and 50 (GAP40 and GAP50), are present in P. falciparum merozoites and appear to function in erythrocyte invasion. Furthermore, we identify a merozoite TRAP homologue, termed MTRAP, a micronemal protein that shares key features with TRAP, including a thrombospondin repeat domain, a putative rhomboid-protease cleavage site, and a cytoplasmic tail that, in vitro, binds the actin-binding protein aldolase. Analysis of other parasite genomes shows that the components of this motor complex are conserved across diverse Apicomplexan genera. Conservation of the motor complex suggests that a common molecular mechanism underlies all Apicomplexan motility, which, given its unique properties, highlights a number of novel targets for drug intervention to treat major diseases of humans and livestock.

Parasites from the phylum Apicomplexa represent some of the most significant human and agricultural pathogens. Their ranks include Theileria parva and Theileria annulata, parasites that give rise to lymphoproliferative diseases of cattle, the opportunistic pathogens Toxoplasma gondii and Cryptosporidium parvum that can cause life-threatening, prolonged infection in immunocompromised patients, and the most lethal of the group, the genus Plasmodium, in particular Plasmodium falciparum, the cause of millions of human deaths and as many as 500 million infections annually (1).
Apicomplexa are a monophyletic group of obligate intracellular parasites that invade a wide range of host cells but lack the classical means of motility such as a flagellum or cilia. Instead, they move by a unique form of actin-based locomotion called gliding motility (for recent reviews, see Refs. [2][3][4]. Efficient motility and invasion requires the release of proteins from secretory organelles located at the apical prominence, the defining structure of the phylum. These organelles, the micronemes, rhoptries, and dense granules contain many of the key proteins needed for directional attachment, cell invasion, and establishment of the parasitophorous vacuole (PV) 5 within the host cell (5). Much of our understanding of gliding motility comes from studies with the liver stage parasite from Plasmodium spp., the sporozoite, or the morphologically similar tachyzoite of Toxoplasma. Micronemal adhesive proteins are released onto the zoite (motile parasite form) surface prior to invasion and are transported to the posterior end of the cell, being cleaved and released on their way, thus creating a dynamic treadmill (2)(3)(4). Ultimately, the extracellular adhesin is linked internally to an anchor that is attached to a series of membrane structures, underlying the outer plasma membrane of the parasite, called the inner membrane complex (IMC) (6,7). Between the plasma membrane and IMC lies an internal actin-myosin motor that drives cell motility (2)(3)(4). In addition to driving cell motility in sporozoites and tachyzoites, the same molecular components also appear to be used for host cell invasion (8 -11). The thrombospondin-related anonymous protein (TRAP) (12) from Plasmodium is the essential adhesin needed for sporozoite motility and for liver cell invasion (9) and defines a family of related proteins conserved across Apicomplexan genera and parasite life cycle stages (e.g. TgMIC2 in Toxoplasma (reviewed in Ref. 11) and PfCTRP in Plasmodium ookinetes (13)). TRAP is a type I integral membrane protein that binds to sulfated glycosaminoglycans on liver cells (14). Key to its adhesive function is the presence of a thrombospondin repeat (TSR) domain (14), a well characterized protein domain that is present on surface proteins across diverse organisms and is implicated in several biological functions such as cell adhesion and cell motility (15). Two essential Plasmodium proteins that contain a TSR domain, the aforementioned TRAP and liver stage circumsporozoite protein (CSP) (16), are both included in ongoing vaccine trials, suggesting that other TSR proteins may also provide novel avenues for anti-malaria therapeutics.
The cytoplasmic tails of the Plasmodium TRAP and TgMIC2 in Toxoplasma have both been shown to interact with the filamentous (F)actin binding protein aldolase (17,18). This would then constitute the bridge between the extracellular adhesin and F-actin (17,18). F-actin interacts dynamically with the class XIV myosin A (MyoA) (6) and in Toxoplasma is anchored to the IMC via the myosin light chain (MLC) (19) and the membrane-anchored glideosome-associated protein TgGAP50 (20) and its associated TgGAP45 (20). A homologue of MLC has been described in both P. yoelii and P. falciparum (7), referred to as the MyoA tail-interacting protein (MTIP). The chain of interaction of these components (TRAP-aldolase-actin-MyoA-MLC/MTIP-GAP45-GAP50) constitutes what we refer to as the motor complex.
Unlike sporozoites and tachyzoites, merozoites, the blood stage malaria parasites, do not demonstrate gliding motility, and the molecular mechanisms that underlie their invasion of the erythrocyte are still largely unknown. Evidence for the involvement of an actin-myosin motor in merozoites comes from the inhibition of invasion when cultures are pretreated with drugs that target actin (cytochalasin B (21)) or myosin (butane-2,3-dione monoxime (6)). This suggests that, despite the absence of gliding motility, the merozoite may utilize actin polymerization, myosin, and by inference a homologous molecular motor for efficient erythrocyte invasion.
Here we show that the components of the motor complex characterized in sporozoites and tachyzoites are expressed in developing merozoites, localize consistently with their inferred function, and form complexes that support their having a role in erythrocyte invasion. We identify a putatively essential merozoite-specific TRAP homologue (MTRAP) that is localized to the micronemes, released and processed during invasion, and interacts in vitro with aldolase, key features suggesting that it is the blood stage invasion adhesin that links the cell surface to the motor complex. A search for homologues of each of these motor proteins in the recently published T. annulata (22) and C. parvum (23) genomes as well as partial genome sequences for Eimeria and Babesia confirms that the motor complex is universally present and likely to form the foundation for all of the different modes of cell motility and invasion seen across the phylum.

Identification of P. falciparum TRAP Paralogues and Motor Protein
Orthologues-The TSR sequence of TRAP (PlasmoDB ID: PF13_0201 (12)) and CTRP (PFC0640w (13)) were used as BLASTP and TBLASTN queries to search the P. falciparum genome (24)  Reverse Transcriptase (RT)-PCR of TRAP Homologues-Total RNA was isolated from synchronized 3D7 parasites 8,16,24,32,40, and 48 h postinvasion using TRIzol (Invitrogen). RNA was further purified using DNase I digestion by passage over an RNAeasy column (Qiagen) to remove any residual genomic DNA. 5 g of total RNA was reverse transcribed either with or without SuperScript TM II reverse transcriptase using random hexamers (Invitrogen).
Parasite Cultures-P. falciparum asexual parasites were maintained in human erythrocytes (blood group Oϩ) at a hematocrit of 4% with 10% Albumax TM II (Invitrogen) (25). 3D7 strain parasites were originally obtained from David Walliker at Edinburgh University. Cultures were synchronized as previously described (26).
3D7 parasites were transfected as recently described (29) with 100 g of purified plasmid DNA (Qiagen). Positive selection for transfectants was achieved using 10 nM WR99210.
Pull-downs with recombinant TRAP tails (or a GST negative control) and either His-PfAldo or P. falciparum schizont-enriched parasite lysate were carried out essentially as described (17,18). Schizont-enriched lysate was taken from saponin-lysed cultures ϳ40 h postinvasion (representing ϳ10 8 parasites), resuspended in 50 mM KCl, 10 mM HEPES (pH 7.7), 1 mM MgCl 2 , 1 mM EDTA, and 0.2% Tween 20 (buffer A) (17,18). A negative control, using buffer A alone, was included with each pull-down. Bound proteins were loaded in 2ϫ sample buffer and separated by SDS-PAGE. Pull-downs involving His-PfAldo were visualized by Coomassie staining, whereas those with schizont lysate were visualized by Western blot probed with rabbit polyclonal antisera raised against GST-PfAldo (see above) using standard procedures.

Motor Complex Proteins PfGAP45 and PfGAP50 Are Expressed in
Merozoites-Recent studies have identified three proteins that link MyoA to the IMC in Toxoplasma tachyzoites, MLC1 (also called MTIP) (7,19), TgGAP45, and TgGAP50 (20), all of which have orthologues identified in P. falciparum (7,20). We analyzed P. falciparum microarray data through the asexual erythrocytic cycle (34) for expression of PfGAP genes, PfMTIP, and three other components of the actin-myosin motor complex (PfActinI, PfMyoA, and PfAldolase) and found that each of them has a pattern of expression consistent with involvement in invasion, peaking in transcription ϳ32-44 h postinvasion (Fig. 1A). PfAldolase has a flatter expression profile, consistent with a dual role in both glycolysis and invasion (17) (Fig. 1A).
To determine whether the PfGAP proteins are expressed in the erythrocytic stage of P. falciparum, we raised antiserum to recombinant PfGAP50 and PfGAP45. When analyzed by immunoblot, mouse and rabbit antisera against PfGAP50 recognized a single product at ϳ45 kDa, whereas those against PfGAP45 recognized a doublet at ϳ37 to ϳ40 kDa (Fig. 1B). PfGAP50 has a predicted molecular mass of 44.5 kDa, and the immunoblot reactivity was consistent with its predicted size. PfGAP45 has a predicted molecular mass of 24 kDa, but its migration through SDS-PAGE is likely to be affected by its elongated structure or the high content of charged residues as with its T. gondii orthologue (20). The extra band present for PfGAP45 (Fig. 1B) suggests that it undergoes limited processing, consistent with TgGAP45 in T. gondii (20).
PfGAP45 and PfGAP50 Localize to the IMC in Merozoites-Having shown that both PfGAP proteins are expressed, we undertook indirect immunofluorescence analysis (IFA) of schizont preparations (pre-and postrupture). This showed that both PfGAP proteins are present in erythrocyte stage parasites, localizing to a ring of fluorescence around the developing merozoites within schizonts or around free merozoites (Fig. 1C). This pattern is consistent with localization to the IMC that lines the length of the merozoite under the plasma membrane (6,7). In the very rare event of capturing an invading merozoite (Fig. 1C, bottom panels), fluorescence appeared to be concentrated at the point of interface between the merozoite and the erythrocyte membrane (white arrows), consistent, although not unequivocally, with localization to the moving junction that migrates along the length of the invading zoite (35,36). Immunoelectron microscopy of invading merozoites, which to our knowledge has not yet been possible with P. falciparum, would enable confirmation of the localization of these proteins to the moving junction.
In P. yoelii sporozoites, MTIP has previously been shown to localize to the IMC (7). Using antibodies that recognize the P. falciparum orthologue, 6 IFAs showed consistent co-localization between PfMTIP and PfGAP45 and, to a limited extent, with PfGAP50 ( Fig. 2A). Lack of complete co-localization between the components is not unexpected, since the intact motor complex is predicted to form only transiently at the moving junction (20,37). Close inspection of the reactivity with anti-PfMTIP ( Fig. 2A) suggests that there is a reduction in fluorescence at the extreme tip of the invading merozoite (white arrow on inset), which is consistent with an absence of IMC at the apical pole (20). We also observed partial co-localization of the PfGAP proteins with mouse antisera raised against a PfMyoA peptide ( Fig. 2A), which gives a diffuse appearance around the periphery of merozoites forming a gradient with greatest intensity at the apical region ( Fig. 2A), consistent with previous reports (6,38). Again, lack of complete co-localization is not unexpected, given the transient nature of the motor complex. As predicted, we saw no co-localization by IFA with antibodies against SERA5, a PV protein, whose fluorescence is lost following schizont, and therefore PV, rupture (data not shown).
Transmission electron microscopy of P. falciparum schizont sections showed the typical trilaminar appearance predicted from the plasma membrane and two inner membranes of the IMC (Fig. 2B, white arrows). The parasitophorous vacuolar membrane and plasma membrane are likely to be too close together to differentiate at this late stage. Immunoelectron microscopy of these sections with anti-PfGAP45 (25-nm gold particles) localized the protein to an electron-dense band around the periphery of forming merozoites with little to no labeling in the internal cytoplasmic regions (Fig. 2B). This structure has previously been interpreted as the IMC (6, 7). We also attempted double labeling of schizont sections with anti-PfGAP50 (10-nm gold particles); however, these reacted poorly, with few gold particles per merozoite, although when they were found, they consistently located close to the larger gold particles of PfGAP45 (Fig. 2B, white asterisk, and data not shown). PfAldolase has a flatter expression profile as predicted by its dual role in glycolysis and motility (17). B, polyclonal rabbit and mouse antisera raised against recombinant PfGAP50 and PfGAP45 recognize specific protein products. C, rabbit antiserum against recombinant PfGAP50 localizes to the merozoite periphery and remains following rupturing of schizont, consistent with localization to the IMC. The white arrows on putative invading merozoite point to interface with erythrocyte where increased fluorescence is detectable, possibly indicating concentration at the moving junction. Rabbit antiserum against recombinant PfGAP45 similarly suggests localization to the IMC. A second putative invading merozoite also suggests PfGAP45 is concentrated at the interface (white arrows) with the erythrocyte, consistent with the moving junction. Scale bar, 1 m.
To investigate the trafficking of PfGAP proteins through the asexual cycle, we generated parasites that expressed PfGAP45 with a C-terminal GFP tag. Visualization of these parasites at different life cycle stages consistently showed PfGAP45 localizing to the divisions between forming merozoites, giving a circumferential fluorescence in pre-and postruptured merozoites (Fig. 2C) in a pattern very similar to that shown for PfMyoA through the erythrocytic cycle (38). Taken together with the IFA and electron microscopy data, this demonstrates that, like their orthologues in Toxoplasma, the PfGAP proteins localize to the IMC in P. falciparum merozoites and suggests that the inner portion of the motor complex, MyoA-MTIP-GAP45-GAP50, is conserved between the two genera.
The MyoA-MTIP-GAP Complex Is Conserved in Merozoites-Having shown that the components of the inner portion of the motor com-plex are present in merozoites, we undertook immunoprecipitations with radiolabeled schizont preparations harvested from late stage cultures (40 -48 h postinvasion) using polyclonal rabbit antibodies against PfMTIP. Immunoprecipitations consistently gave rise to protein bands at apparent molecular masses of ϳ90, ϳ45, ϳ40, ϳ35 and 25 kDa (Fig. 3A). Western blots with specific antibodies to each subunit reacted with products at the same predicted molecular weight, strongly suggesting that these proteins were PfMyoA, PfGAP50, PfGAP45, and PfMTIP, respectively (Fig. 3B). Additional immunoprecipitations using antibodies against PfGAP45 and PfGAP50 similarly pulled down bands reactive with anti-PfMyoA, PfMTIP, and the two PfGAP proteins (Fig. 3B). The precipitation of minimal amounts of PfGAP50 with anti-PfGAP45 is surprising, given that the reciprocal immunoprecipitation successfully isolated all of the components, and suggests that the complex may not be very stable under our current isolation conditions (Fig. 3B). This may also explain why antibodies raised against PfMyoA did not precipitate any of the components (data not shown). Overall, the interaction between these four components of the motor complex in vivo argues strongly for the functional conservation of the inner portion of the motor complex in P. falciparum merozoites as originally identified in T. gondii.
Identification of a Merozoite-specific TRAP Homologue-Having identified conservation in the inner portion of the motor complex, we sought to identify a merozoite adhesin that might serve as the extracellular link to the internal motor complex. TSR-containing proteins have been shown to be essential for T. gondii tachyzoite and both Plasmodium sporozoite and ookinete motility and invasion (8 -10) and have also been shown to form the extracellular link to the internal motor complex (17,18). To determine whether there was also a merozoitespecific TRAP/MIC2 (micronemal protein 2) homologue that functions for erythrocyte invasion, we searched the genome sequence of P. falciparum with BLASTP and TBLASTN algorithms using the TSR domain of TRAP and CTRP. Eight TSR-containing genes, all with signal peptides, were identified (Fig. 4A). RT-PCR with specific primers for each of the genes suggests that of these, only the circumsporozoite protein (required for development of malaria sporozoites in mosquitoes) (16), a recently described erythrocytic TSR protein PTRAMP (39), and two unidentified genes PF10_0281 and MAL8P1.45 were transcribed in trophozoites and schizonts late in the erythrocytic cycle, consistent with a role in invasion (Fig. 4B). PF10_0281 was transcribed at considerably FIGURE 2. Localization of PfGAP proteins to the inner membrane complex. A, PfGAP50 and PfGAP45 co-localize with MTIP (a protein known to localize to the inner membrane complex) and partially with PfMyoA (known to associate with the inner membrane complex and merozoite apical tip (6,38)). The white arrow in the MTIP inset points to an area of reduced fluorescence at the apical tip in a single z-stack, consistent with the absence of IMC at this position. B, immunoelectron microscopy confirms IMC localization for PfGAP45 (large 25-nm gold particle). The white arrows indicate trilaminar appearance of the IMC and merozoite plasma membrane. i, inner IMC membrane; o, outer IMC membrane; p, plasma membrane R, rhoptry. *, single anti-PfGAP50-conjugated gold particle found in close proximity to PfGAP45 but giving poor reactivity (typically 1 particle/merozoite). Scale bar in electron microscopy, 0.2 m. C, PfGAP45-GFP shows fluorescence that concentrates to the merozoite periphery late in asexual development, consistent with the formation of the IMC as described for T. gondii (20). Scale bar in IFA, 1 m. greater levels than all the other genes through the life cycle (Fig. 4B). Our identification of eight TRAP-related proteins excluded the recently identified secreted protein with altered thrombospondin repeat, PfS-PATR (40), that, despite having a modified TSR domain, does not have a definable transmembrane domain or a cytoplasmic tail.
The presence of a cytoplasmic tail with a concentration of acidic residues and a subterminal tryptophan residue is thought to be central to the interaction of TgMIC2 and TRAP with the internal actin-myosin motor via the glycolytic enzyme aldolase (17,18). Of the eight genes, five have a concentration of acidic residues, of which four have a subterminal Trp residue (PTRAMP has a tyrosine residue) (Fig. 4C). In addition to features of the cytoplasmic tail, intramembrane cleavage of TRAP/ MIC2 also appears to be essential for TRAP function (41), with cleavage thought to occur via an intramembrane rhomboid protease (42)(43)(44)(45). Alignment of the C-terminal regions of the eight TSR-containing genes (with TgMIC2), shows that only four of the eight have a putative rhomboid cleavage site (Fig. 4C) (44).
By a process of elimination, PF10_0281, which peaks in expression late in the erythrocytic cycle and has a putative rhomboid cleavage site within its transmembrane domain and an acidic cytoplasmic tail with recessed terminal Trp residue, is expected to share the greatest levels of functional homology with sporozoite TRAP. We have therefore named it the merozoite-specific TRAP homolog, or MTRAP.
MTRAP Is a Putatively Essential TSR Protein That Localizes to the Micronemes-The MTRAP gene is predicted to encode a protein of 498 amino acids with a predicted molecular mass of 58 kDa. The TSR crystal structure has recently been solved for human TSP-1 (15), consisting of three anti-parallel strands (A, B, and C) (supplemental Fig. S1) containing conserved residues that dictate folding and function of the domain. Alignments with TSR domains from other TRAP homologues shows that MTRAP has a type II-like TSR domain, where the third disulfide bond is absent as it is in the second TSR repeat of human HB-GAM (15) (supplemental Fig. S1). BLASTP and TBLASTN searches of partial shotgun sequences for P. reichenowi (a closely related chimpanzee malaria parasite) and P. berghei (a more distantly related mouse malaria parasite) identified orthologues of MTRAP in both species sharing 81 and 35% identity, respectively, and sharing the type II-like TSR domain structure and sequence conservation in the C terminus, EXEFWXXE (which includes the recessed terminal Trp residue). Both of these features are unique to MTRAP among P. falciparum TRAP homologues (Fig. 4) (supplemental Fig. S2). No direct homologues of MTRAP (aside from homology with the TSR domain) are found outside of the Plasmodium genus.
To investigate whether a functional product of MTRAP is expressed, we raised antiserum to a recombinant GST fusion protein encompassing the MTRAP-TSR domain. Western blot of schizont pellet material (44 -48 h postinvasion) with mouse and rabbit antisera recognized a ϳ70-kDa band with an additional band running at ϳ25 kDa (Fig. 5A). Although the larger product is greater than the predicted size of MTRAP (which is 58 kDa), this may be due to the high concentration of negative residues in the region between the TSR and transmembrane domain, which are likely to affect the running of the full-length protein through SDS-PAGE. The specificity of the rabbit antibody was confirmed by expressing an additional MTRAP copy with a triple hemagglutinin (3*HA) or GFP tag at the end of the cytoplasmic tail (Fig. 5A). Parasite material from cultures expressing MTRAP-3*HA showed an additional band at ϳ80 kDa (Fig. 5A, b) whereas MTRAP-GFP showed a 100-kDa band (Fig. 5A, c). The higher molecular weight bands were confirmed to be tagged MTRAP using monoclonal anti-HA (Fig. 5A,  second panel) or anti-GFP (Fig. 5A, third panel). In addition to the shared high molecular weight bands, both monoclonals detected further products that do not contain the TSR domain, suggesting that MTRAP may be processed (see asterisks in Fig. 5A). There was no crossreactivity seen between the GFP monoclonal and the HA-tagged protein or vice versa (data not shown).
Given that the sporozoite TRAP localizes to the micronemes and is released apically to mediate motility and invasion (46), we carried out IFAs with MTRAP-TSR antisera to investigate whether the same was true for MTRAP. The MTRAP-TSR antibody clearly localizes MTRAP to the apical tip of the merozoite (Fig. 5B). Co-localization with antisera to PfRAP1, a protein that is known to reside in the body of the rhoptries (47), demonstrates that MTRAP is either located in the apical neck of the rhoptry or in the micronemes (Fig. 5B). To definitively differentiate between these two compartments, we carried out immunoelectron microscopy of schizont sections with the MTRAP-TSR antibody (Fig.  5C). Gold particles (10 nm) consistently localized to the micronemes, easily distinguished from the rhoptries as small vesicular bodies at the apical tip of developing merozoites. This confirms that MTRAP is indeed located in the micronemes (Fig. 5C).
To investigate the function of MTRAP, we attempted to disrupt the endogenous MTRAP gene using standard protocols (48). Despite repeated attempts in different parasite strains, we were unable to knock out the gene (data not shown). This suggests that MTRAP may play an essential role in parasite development, which is consistent with its being of equivalent importance to erythrocyte invasion as Plasmodium TRAP and CTRP are to invasion of liver cells and mosquito midgut cells, respectively (9,10).
MTRAP Is Released onto the Merozoite Surface and Processed during Invasion-To visualize the trafficking and investigate whether MTRAP is released apically during merozoite development, we expressed a copy of the full-length MTRAP gene with a C-terminal GFP or triple HA tag. Both tagged proteins localize to the apical tip ( Fig. 6A; data not shown). In the developing schizont and immediately after merozoite release, MTRAP-GFP localizes to a singular point at the extreme apical tip (Fig.  6, A and B), confirming localization seen with the MTRAP-TSR antibody (Fig. 5B). In some free merozoites, an apical capping of the merozoite is also seen (Fig. 6A). This is clearly the apical tip, since it is anterior to the nucleus and overlaps with a darker region that is thought to be the rhoptries (visible under bright field; see 0 h). This capping pattern of fluorescence has been described previously with another microneme protein AMA1 (49) and suggests that MTRAP may also be released onto the surface of the invading merozoite prior to invasion (30). This suggestion is confirmed by co-localization of MTRAP with AMA1, both of which are apical before schizont rupture (Fig. 6B, top panel) and asso-ciate with the surface of the invading merozoite, concentrating at the apical pole (Fig. 6B, middle panel).
Following invasion, both the cytoplasmic tail (Fig. 6, A and B) and TSR domain (data not shown) are detectable by IFA in newly invaded rings. Close inspection of Fig. 6B (bottom panel, inset) shows that there does not appear to be exact co-localization between AMA1 and the GFP-tagged tail. This would be predicted if the tail remains bound to the merozoite plasma membrane, whereas the ectodomain of AMA1 (to which this antibody is raised (30)) would remain in the PV. To further investigate MTRAP processing, we carried out Western blots with highly synchronous 3D7 and 3D7-MTRAP-3*HA cultures using parasite preparations harvested pre-and post-invasion. Immunoblots of wild-type and MTRAP-3*HA schizont preparations (44 -48 h postinvasion), parasite supernatants (postschizont rupture), and early ring stage parasites (Ͻ8 h) show that the full-length MTRAP, MTRAP full , is diminished post-invasion with only a small amount of the unprocessed form being carried through to the ring stages (Fig. 6C, band 1). No MTRAP full was detected in supernatant, consistent with its being membrane-bound at the C terminus (Fig. 6C, band 1). The ϳ25-kDa processed form that contains the TSR domain, MTRAP 25 , was detected in supernatant and was also present in the newly formed ring (Fig. 6C,  band 2). Release of MTRAP 25 into the supernatant is consistent with MTRAP being cleaved in its extracellular domain during invasion. Its presence in very young rings may be due to excess TSR domain remaining in the PV or due to its remaining bound to either an erythrocyte receptor or other merozoite invasion proteins that are carried through into the newly formed ring (as would also appear to apply to AMA1) (Fig. 6B, bottom panel). Immunoblots with 3D7-MTRAP-3*HA parasite material showed three additional bands to MTRAP full at ϳ48, ϳ45, and 10 -20 kDa, respectively (Fig. 6C, bands 3-5). The ϳ48-kDa band may represent the loss of the TSR domain, since it is not recognized by anti-TSR, with further processing leading to the slightly shorter ϳ45-kDa band. The size of the 10 -20-kDa band is consistent with a cleavage event near or within the transmembrane domain of the tagged protein, leaving the cytoplasmic tail. Such a cleavage event is supported by the presence of a putative rhomboid cleavage site in the transmembrane domain of MTRAP (Fig. 4C) and the essential nature that this cleavage has for other TRAP homologues (42,43,45). A ϳ30-kDa band detectable in 3D7-MTRAP-GFP parasite material (Fig. 5A, right panel, lane c) is likely to be the equivalent cytoplasmic tail, factoring in the increased mass given by the GFP tag. Together, this suggests that at least two cleavage events occur in MTRAP proteolysis during invasion (Fig. 6D), leaving a cytoplasmic tail that has a predicted size consistent with its being the product of intramembrane cleavage by a rhomboid protease.
The Cytoplasmic Tail of MTRAP Binds to Recombinant and Native P. falciparum Aldolase-Central to the function of TRAP in sporozoites and TgMIC2 in tachyzoites is the interaction of the cytoplasmic tail with the rest of the motor complex, an interaction that may be mediated by aldolase, a protein known to bind F-actin (17,18). To investigate whether MTRAP also had the ability to bind aldolase, we expressed the cytoplasmic tails of MTRAP and PfTRAP along with a mutant MTRAP tail (MTRAP W/A , having a Trp to Ala mutation in the fourth residue from the C terminus) fused to GST (Fig. 7A) and investigated their binding to recombinant PfAldo expressed with a His 6 tag. GST-MTRAPtail and GST-PfTRAPtail immobilized on GSH-Sepharose beads bound recombinant PfAldo (Fig. 7B). Binding of all tails to PfAldo was not very efficient, with a maximum of 0.1 g bound/g of added tail fusion protein (Fig. 7B; data not shown). This may indicate that the interaction in vivo is not very strong, although it may also be the result of poor refolding of our recombinant PfAldo. Binding of GST- . Asterisks indicate processed products not detected by TSR antiserum. B, immunofluorescence microscopy shows apical concentration of MTRAP different to that of PfRAP1, a known basal rhoptry protein (47). C, immunoelectron microscopy confirms that apical localization is to the micronemes. m, microneme; R, rhoptry. The arrows indicate concentrations of immunogold labeling against MTRAP-TSR. FEBRUARY 24, 2006 • VOLUME 281 • NUMBER 8 MTRAP W/A tail was almost 60% less than that in the wild type (Fig.  7C), a reduction that is consistent with the Trp residue being important for stabilizing the interaction with aldolase (18). As additional controls, immobilized GST was shown to have minimal binding to PfAldo (Fig. 7B), whereas PbTRAP bound to PfAldo to almost the same degree as PfTRAP (Fig. 7, B and C). We also expressed the cytoplasmic tail of another erythrocytic TSR protein, PTRAMP (39). This bound PfAldo at a level between that of MTRAP W/A and GST, suggesting that PTRAMP, despite having acidic residues and a terminal Tyr residue (that is also aromatic) in its cytoplasmic tail, may not be functionally equivalent to that of TRAP (Fig. 7, B and C).

Conservation of the Apicomplexan Actin-Myosin Invasion Motor
To further characterize the in vitro interaction of the MTRAP tail with aldolase, we carried out pull-down assays using the same GSH-Sepharose-immobilized GST-MTRAPtail but this time incubated the recombinant tail with whole parasite lysate taken from schizont-enriched cultures (ϳ40 h postinvasion). Immunoblots of bound parasite proteins probed with rabbit polyclonal antisera raised against recombinant PfAldo recognized a single band at ϳ40 kDa (Fig. 7D). This corre-sponds to the predicted molecular mass of P. falciparum aldolase at 40.1 kDa. A similar band was detectable in a pull-down with the GST-PfTRAPtail and with the GST-MTRAP W/A tail (Fig. 7D). This confirms the affinity of the MTRAP and TRAP tails for native aldolase. Pulldowns using the GST-PTRAMPtail or GST did not bring down any detectable aldolase (Fig. 7D). This further argues against any functional equivalence between TRAP and PTRAMP.
The in vitro interaction of the MTRAP C terminus with aldolase supports its role as the functional homolog of TRAP for the merozoite, although the functional significance of the aldolase interaction is still unknown. The similarities between MTRAP and TRAP and the presence of the other components linking the IMC to an extracellular adhesin suggests that the actin-myosin motor complex is conserved between the liver stage sporozoite and the blood stage merozoite of Plasmodium parasites despite their very different modes of motility.
Conservation of the Motor Complex across Apicomplexa-Given the apparent conservation in the motor complex that is used to drive both sporozoite motility and invasion and merozoite invasion and the known FIGURE 6. MTRAP undergoes extensive processing during invasion. A, episomally expressed MTRAP-GFP followed through the life cycle confirms apical localization (44 -48 h) suggests the release of MTRAP onto the surface of free merozoites (capping, 0 h) and suggests that, despite processing, the cytoplasmic tail is carried through to newly invaded rings (Ͻ8 h). B, this is confirmed by following co-localization with antibodies that recognize AMA1, which have been previously reported as being carried through to rings (49). In early ring stages, there does not appear to be direct co-localization, suggesting that they may be associating with different compartments. C, a processed fragment containing the TSR domain is released into the supernatant (band 2), with some also carried through to the ring stages. The fulllength product is markedly diminished following invasion (band 1). Episomally expressed MTRAP-3*HA shows similar processing to wild type MTRAP, with antiserum against the TSR domain or C-terminal HA tag identifying two processing events in the MTRAP backbone (bands [3][4][5]. The asterisk indicates cross-reactivity with albumin in the medium. The numbers refer to processed fragments (see below). D, a model of MTRAP processing, showing the fragments produced and how they are recognized by antiserum against the TSR domain or HA tag. The colored regions refer to those in Fig. 4A, with the green shaded region indicating the C-terminal triple HA tag.
conservation between Plasmodium zoites and Toxoplasma tachyzoites in the way they move (8), we sought to determine whether each of the subunits now identified in the merozoite is also seen across other Apicomplexan genera. Searches of the publicly available genomes for T. annulata (22), C. parvum (23), E. tenella (available on the World Wide Web at www.GeneDB.org), and B. bigemina (available on the World Wide Web at www.sanger.ac.uk) show that orthologues for all contributing genes (TRAP homologue-aldolase-actin-MyoA-MTIP/MLC-GAP45-GAP50) are present ( Table 1, supplemental Fig. S3). The presence of orthologues to all of the key components supports the notion that the same actin-myosin-based molecular motor drives cell motility and invasion across all Apicomplexan parasites.

DISCUSSION
General conservation in the molecular mechanisms that underlie gliding motility and cell invasion across Apicomplexan parasites has previously been suggested following heterologous replacement of cell surface adhesins (8,50) and by the presence of orthologues to different components of the motor complex in three genera of Apicomplexan parasites (7,19,20). However, until now, this conservation has not been demonstrated for the relatively nonmotile blood stage merozoite of Plasmodium spp. Here we show that the same basic motor complex that drives gliding motility in the amoeboid-like Plasmodium sporozoite is also present and likely to drive cell invasion in the merozoite. Furthermore, we show that the generic complex, consisting of TRAP homologue-aldolase-actin-MyoA-MTIP/MLC-GAP45-GAP50, can be identified in each of the available Apicomplexan genomes, representing six highly divergent genera. This suggests that the motor complex is evolutionarily conserved across the phylum Apicomplexa.
Primary support for the role of the conserved motor in P. falciparum erythrocyte invasion comes from the expression and localization of each of the components of the motor complex and their association together. PfGAP45, PfGAP50, and MTIP all localize to the merozoite periphery by IFA (Figs. 1 and 2), which is confirmed by electron microscopy to be the IMC (Fig. 2B) (7). Although localization by microscopy is not definitive, immunoprecipitation of the complex using antibodies against both PfGAP proteins and MTIP confirms that these components do associate in vivo (Fig. 3). Although this may only involve a small fraction of the different subunits, this is likely to be determined by the transient nature of the moving junction in the invading merozoite and not necessarily a demonstration of low affinity between different subunits. Unlike gliding motility in more motile Apicomplexan life cycle stages, merozoite invasion is a very rapid process, perhaps lasting only a few seconds, making the potential to harvest large quantities of midinvasion parasites technically very challenging.
Conserved Domains in the Motor Complex Subunits-At present, the precise molecular interactions between the GAPs and the motor complex are unknown. Regions that are likely to play a key role in these interactions may be identifiable as conserved domains between Apicomplexan orthologues (from the P. falciparum, T. gondii, C. parvum, T. annulata, E. tenella, and B. bigemina genomes) (see supplemental Fig.  S3). For GAP45, sequence conservation in the extreme N and C terminus highlights these regions as potentially critical to function. This includes the Met-Gly N-myristoylation motif in the first two amino acid positions (20). The addition of a C-terminal GFP tag to PfGAP45 was not detrimental to parasite development here, suggesting that the interaction with the other components may be via the N terminus. It is possible that a functional domain may still be accessible in the GFP mutant (since there is a linker between the C terminus and GFP) or that endogenous PfGAP45 is sufficient to sustain function despite the presence of the episomal GFP copy. Further characterization of the motor complex with the PfGAP45-GFP-expressing line will help to resolve whether the tagged form is indeed functional.
There are long stretches of amino acid identity in the N-terminal region of the GAP50 protein (supplemental Fig. S3). The formation of the Toxoplasma motor complex relies critically on a functional GAP50 C-terminal tail (20), which, along with conservation in the six amino acids that make up the C terminus among orthologues from Plasmodium and Eimeria, was used to argue that the complex with GAP45, MTIP, and MyoA is associated with the short cytoplasmic tail of GAP50 (20). Including orthologues from Cryptosporidium, Theileria, and Babesia reduces the degree of sequence homology in the cytoplasmic tail, suggesting that the conservation may have been overestimated (supplemental Fig. S3). We believe that the high degree of conservation in N-terminal regions between the six Apicomplexan species argues in favor of the complex being assembled using the much larger N terminus and not the short C terminus of the GAP50 anchor.
The level of sequence homology between MTIP and TgMLC1 is well characterized (7) and appears to hold across Apicomplexan genera. Binding of MTIP to PfMyoA was shown previously to be dependent on a large portion of the MTIP protein (7) and a dibasic motif in the PfMyoA tail (51). The dibasic motif is maintained across the six genera, as are three highly conserved amino acid stretches within the C-terminal portion of MTIP, which may constitute the sites that are central to the interaction with MyoA or have other functional roles (positions 97-100, 111-119, and 195-202; supplemental Fig. S3). The CpMTIP ortholog has a very large additional C-terminal tail. Further investigation will determine whether this is an error in the gene annotation or if the long cytoplasmic tail has unique properties or functions.
A TRAP-based Motor Complex-Conservation of the molecular mechanism that underlies all Apicomplexan motility is strengthened by the presence of a TRAP homologue in each of the Apicomplexan genera (Table 1). However, the generality of this assertion has not been extended to erythrocyte invasion by Plasmodium merozoites. Here, we have identified a novel, putatively essential, erythrocytic protein that has a number of features that suggest it may be the functional TRAP homolog for merozoites, which we have named MTRAP. With its identification, there are now at least two P. falciparum merozoite-specific TRAP homologues (MTRAP, PF10_0281; PTRAMP (39), PFL0870w). Four main observations support the notion that MTRAP is a probable merozoite-specific functional homolog of TRAP: (a) MTRAP is expressed at high levels in middle to late stage asexual parasites; (b) it is  TABLE 1 Orthologues of the conserved motor complex from six Apicomplexan genera Entries in boldface type have been shown to be functional and/or present, and nonbold entries are only inferred from sequence similarity. released onto the merozoite surface prior to invasion; (c) MTRAP is processed during invasion in a manner that is consistent with there being intramembrane cleavage by a rhomboid protease; and (d) it is putatively essential as demonstrated by repeated attempts to knock out the endogenous gene. Further support for the functional homology also comes from the ability of the cytoplasmic tail of MTRAP to bind in vitro both recombinant and native aldolase and the reduction in binding following mutation of the terminal Trp residue (Fig. 7B). However, despite the observed interactions of TRAP, TgMIC2, and now MTRAP with aldolase (17,18), the functional significance of this interaction is still unknown. Furthermore, we believe that other proteins are likely to be involved and may be critical for stable interaction between the cytoplasmic tail and F-actin. Studies are currently under way to identify other actin-binding proteins that may link with the extracellular adhesin.

Gene
We have not been able to demonstrate direct binding of the MTRAP TSR domain to the host cell surface, 7 and whereas this may not rule out a direct interaction with an erythrocyte receptor, it is possible that MTRAP binds the erythrocyte indirectly via another parasite protein.
Such an association of MTRAP with a secondary binding partner would not be unprecedented, since TgMIC2 associates closely with M2AP in Toxoplasma (52), disruption of which severely impairs host cell entry (53). A potential secondary partner for MTRAP might be one of the erythrocyte binding antigens, such as EBA-175, which does not have a TSR domain and has not been shown to interact with the motor complex. In support of this, it has been proposed that trafficking of EBA proteins to the micronemes requires an unidentified escorter protein (54). One possibility is that a micronemal protein, such as MTRAP, may function as both the escorter and the link between invasion adhesins and the motor complex in merozoites. The ability of the TSR domain to interact with a conserved domain, such as the 3Ј Cys-rich domain of EBAs (55), would provide a generic link to a variable extracellular invasion ligand. The conserved domain of Rh proteins represents another potential site of interaction (56). The recent observation of an association between proteins from different cellular compartments at the moving junction in Toxoplasma (35) supports such a complex being involved in invasion. The micronemal protein TgAMA1 (whose P. falciparum orthologue does not bind to the erythrocyte surface despite being essential for invasion (28)) is associated with a complex of rhoptry neck proteins RON2 and RON4 (35). Studies are currently under way to determine whether the MTRAP TSR is involved in a similar tight junction complex with other micronemal or rhoptry proteins.
The extracellular domain of MTRAP does not appear to be under positive selection, as are analogous regions of AMA1 (57) and EBA-175 (58). In comparison with its inferred orthologue in P. reichenowi (supplemental Fig. S2), there is no significant evidence for an excess of nonsynonymous polymorphism in PfMTRAP (Fisher's exact p ϭ 0.56) (59). This may suggest that MTRAP is hidden from the host immune response by associating in a complex or, alternatively, that its release in vivo is later than other micronemal proteins and occurs in the tight junction away from host antibodies. In support of this lack of exposure, invasion assays with polyclonal serum raised against the MTRAP TSR domain did not inhibit invasion. 8 The conservation of the motor complex across Apicomplexan parasites is further supported by the recent demonstration that Cryptosporidium uses gliding motility to invade host cells (60) and that invasion or motility can be inhibited in Cryptosporidium (60), Theileria (61), and Babesia (62), following the addition of drugs that inhibit actin or myosin. The presence of the complex in Cryptosporidium is particularly striking, since this genus represents a phylogenetically early branch on the Apicomplexan tree more closely related to gregarines (parasites of invertebrates) than it is to either Plasmodium or Toxoplasma (63). Together, this suggests that a generic motor complex is used across the phylum Apicomplexa and has been adapted to the variety of hosts and tissues targeted and the diversity of cell morphology found among Apicomplexan parasites. The conservation of this complex despite the diversity of Apicomplexan parasites highlights a number of potential targets for chemotherapeutic inhibition that may be applicable in a wide variety of diseases of both humans and livestock.