Origin, composition, organization and function of the inner membrane complex of Plasmodium falciparum gametocytes

Malaria affects more than 240 million people annually and causes about 781,000 deaths (WHO World Malaria Report 2010; http://www.who.int/malaria/world_malaria_report_2010/world_malaria_report_2010.pdf). The apicomplexan parasite Plasmodium falciparum is responsible for the most deadly form of the disease in humans. The parasite has a complex lifecycle involving asexual and sexual reproduction within a human host and an Anopheles mosquito vector. To prepare itself for life in these different intracellular niches, the parasite undergoes a remarkable series of morphological transformations, changing its own shape and remodelling its host cell environment. The link between asexual multiplication in the red blood cells (RBCs) of the human host and sexual reproduction in the midgut of the mosquito is the formation of a specialized sexual blood stage parasite called the gametocyte.

Late stage P. falciparum gametocytes are characterized by their unique crescent or falciform shape. Development of these specialized cells takes at least 7 days and can be divided into five distinct stages (Baton and Ranford-Cartwright, 2005; Carter and Miller, 1979; Dixon et al., 2008; Hawking et al., 1971). Early gametocytes (stage I and IIa) are morphologically indistinguishable from early asexual parasites. Changes in cellular architecture are first observed during late stage II to stage III of development, leading to the characteristic crescent shape in stage IV, with the gametocyte ends becoming more rounded in stage V parasites (Dixon et al., 2008; Sinden, 1982; Sinden et al., 1978).

Like mature asexual stage parasites, stage I–IV gametocytes sequester away from the peripheral circulation (Day et al., 1998; Rogers et al., 1996), thereby avoiding passage through and potential clearance by the spleen. It has been suggested that stage I–II gametocytes adhere to CD36 before undergoing a switch in stage III–IV that permits sequestration in the bone marrow, whereas the mature stage V gametocytes lose the ability to cytoadhere (Rogers et al., 2000). Indeed, stage V gametocytes reappear in the peripheral circulation and this is the only stage that can complete sexual development after ingestion by a mosquito. Gametocyte maturation thus represents a ‘bottle-neck’ in parasite development; inhibition of this process ablates disease transmission. Despite the importance of this stage in efforts to eradicate malaria, relatively little is known of the mechanisms controlling gametocyte shape, form and function.

Coincident with the adoption of the unique crescent shape of the late stage gametocyte is the appearance of a tri-laminar membrane structure known as the sub-pellicular membrane complex. Early electron microscopy studies on mammalian malarial parasites identified this structure as a double-membraned cisternal compartment underlying the parasite plasma membrane (Sinden and Smalley, 1979). In this work, we describe some of the molecular components of the sub-pellicular complex and show that the complex is closely related to the inner membrane complex (IMC) of the invasive stages of the parasite.

In the motile (zoite) stages of Plasmodium and other apicomplexa, the IMC is connected to an actin–myosin motor that drives cellular invasion through a process called ‘gliding motility’. The glideosome proteins include myosin-A (MyoA), myosin-A tail domain interacting protein (MTIP) (Bergman et al., 2003; Herm-Götz et al., 2002) and the glideosome-associated proteins 45 and 50 (GAP45 and GAP50) (Gaskins et al., 2004). P. falciparum GAP50 is an integral membrane protein that anchors pre-complexed GAP45–MTIP–MyoA (Johnson et al., 2007). Host cell invasion is thought to require binding of the actin–myosin complex to adhesive proteins such as the merozoite thrombospondin-related adhesive protein (mTRAP). mTRAP is embedded in the parasite plasma membrane and binds to receptors on the host cell, thereby providing the force that drives invasion (Baum et al., 2006a). Underlying the IMC is a microtubule scaffold that forms a major structural element of the invasive stages of apicomplexan parasites (Kudryashev et al., 2010; Sinden, 1982).

First, we show that the sub-pellicular membranes of gametocytes have the same core components as in the invasive stages and that these components are in complex. Second, we provide evidence that a coordinated recruitment of the IMC to the parasite periphery and assembly of a network of microtubules drives elongation of the P. falciparum gametocyte, confirming at a molecular level suggestions from early electron microscopy studies (Sinden, 1982). This shape change is associated with an increase in cellular deformability and might explain how late stage gametocytes can survive circulation through the blood stream of the host.

The late stage P. falciparum gametocyte is characterized by its unique crescent shape. Coincident with the adoption of this shape is the appearance of a tri-laminar structure known as the sub-pellicular membrane complex, consisting of a cisternal compartment underlying the parasite plasma membrane (Sinden and Smalley, 1979). We performed high-resolution serial section electron tomography of this region in a mature stage gametocyte. Fig. 1A shows a virtual section from the tomogram and a rendered model of the membrane complex. The RBC membrane (red), parasitophorous vacuole membrane (PVM, blue), parasite plasma membrane (PPM, yellow), the sub-pellicular membranes (green) and a layer of sub-pellicular microtubules (MT, gray) are evident.

We were struck by the ultrastructural similarities between this sub-pellicular complex (Fig. 1A) and the IMC of merozoites (Fig. 1B). In the merozoite, the IMC is a cisternal compartment that is flattened against the parasite plasma membrane (Fig. 1Bi). It is supported by two or three longitudinally running microtubules that extend along one side of the merozoite (Fig. 1Bii, arrows), as reported previously (Bannister et al., 2000).

Although previous studies have described the sub-pellicular membrane complex of the P. falciparum gametocyte at the ultrastructural level, there are few reports on the composition or origin of this organelle. In an effort to determine whether the gametocyte membrane structure is indeed related to the IMC, we performed fixed cell immunofluorescence microscopy using antibodies recognizing known components of the merozoite IMC. Fig. 2A–D shows the presence of GAP50, GAP45, MTIP and MyoA at the periphery of stage IV gametocytes, consistent with a location in the sub-pellicular membrane complex. At this level of resolution, the profile was similar to that of the P. falciparum PVM protein, Pfs16, which has been shown to be gametocyte-specific (Baker et al., 1995). Protein solubility and expression profiling by western blotting of gametocytes and schizont stage parasites confirmed the presence of the IMC components (supplementary material Fig. S1, and below).

We have previously generated transfectants expressing a GFP chimera of P. falciparum GAP50 as a tool to assess IMC genesis and reorganization during the development of merozoites (Yeoman et al., 2011). We transfected the P. falciparum GAP50–GFP gene construct into a high gametocyte producing clone of 3D7 and found that the GAP50–GFP chimera is correctly expressed and present at the periphery of mature stage gametocytes, co-locating with endogenous GAP45 (Fig. 3A). To determine whether components of the IMC are in complex, we performed immunoprecipitation experiments. Stage III–IV gametocytes expressing the GAP50–GFP chimera were magnet-enriched and then solubilized using RIPA detergent. The cleared lysate was incubated with rabbit antibodies against P. falciparum GAP45, and immunoprecipitated using Protein A/G Sepharose. Precipitated samples were separated by SDS-PAGE and transferred for western blotting. Blots were probed with anti-mouse GFP, GAP50 or MTIP. Full-length GAP50–GFP (64 kDa) was detected in the precipitated pellet of the transgenic sample, by both the anti-GFP and anti-GAP50 reagents, but not in the 3D7 parent line (Fig. 3B). MTIP (28 kDa) and endogenous GAP50 (42 kDa) were detected in the precipitated pellets of both the transfectant and wild-type samples (Fig. 3B). This demonstrates that both GAP50–GFP and endogenous GAP50 are in complex with endogenous GAP45 and MTIP. These studies validate the use of P. falciparum GAP50–GFP as a marker of the gametocyte IMC.

Western blotting analyses revealed the presence of GAP45, GAP50 and MTIP from stage II to stage V gametocytes (supplementary material Fig. S1B) and immunofluorescence microscopy confirmed that these proteins were present at the gametocyte periphery (supplementary material Fig. S2A–D). Moreover, the solubility profiles for these proteins were similar in gametocytes and merozoites (supplementary material Fig. S1). These data, in conjunction with the immunoprecipitation data, show that the gametocyte sub-pellicular membrane complex is very closely related to the merozoite IMC and we suggest that it be renamed the gametocyte IMC.

P. falciparum GAP50 is recruited from the endoplasmic reticulum (ER) to the nascent IMC during asexual schizogony (Yeoman et al., 2011). To determine the origin and reorganization of proteins during the formation of the gametocyte IMC we examined the location of GAP50–GFP in live gametocytes that were co-labelled with the membrane probe, BODIPY-TR-ceramide. During the initial stages of elongation (stage III), when the gametocyte first becomes morphologically distinguishable, GAP50 was associated with a reticular structure in the parasite as well as along one side of the elongating parasite (Fig. 4Ai). Fixed cell immunofluorescence microscopy confirmed that GAP50–GFP labels internal structures that were also recognized by an antibody against the P. falciparum ER resident protein, ERC (ER calcium binding protein) (Fig. 3C). By contrast, ERC was absent from the peripheral location to which GAP50–GFP is recruited as the gametocyte elongates. 3D-structured illumination microscopy (3D-SIM) can provide enhanced resolution of cellular structures (Schermelleh et al., 2008). Using 3D-SIM we found that the GAP50–GFP-labelled IMC remained closely associated with the intracellular reticular structure (Fig. 3D,E, arrows). A rendered model of the GAP50–GFP fluorescence illustrates the points of associations (Fig. 3F, arrows), as do translational and rotation views of the fluorescence micrographs (supplementary material Movies 1, 2). This is consistent with an early ultrastructural study (Sinden et al., 1978) that reported connectivity between the ER and the sub-pellicular membranes.

Stage III gametocytes undergo a dramatic cellular rearrangement, adopting a ‘hat-like’ appearance. During this stage, the GAP50–GFP is concentrated in the flattened rim region of the ‘hat’, with additional looping structures around the top of the cell (Fig. 4Ai,ii; arrows; see supplementary material Movie 2). As the IMC develops further it adopts a cupped shape along the foot, in addition to the looping structures. Electron microscopy confirmed the presence of the IMC along the flattened surface and around the ‘pinching’ ends of the gametocyte (Fig. 4Bi,ii). Fig. 4Bi depicts a longitudinal slice through a region of the sub-pellicular microtubule layer, with the IMC visible at the periphery. Some of the microtubules that pass through the plane of the section are indicated with arrows. Fig. 4Bii shows a cross-section in which the microtubules are viewed end-on and the IMC is evident as a layer outside the microtubule ‘basket’.

In stage IV of parasite development, the characteristic pointed ends of the P. falciparum gametocyte become apparent. By this stage, the GAP50–GFP extends further around the parasite, the surface of which was delineated by BODIPY-TR-ceramide labelling (Fig. 4Aiii; see supplementary material Movie 2 for a 3D rotation). 3D-SIM imaging of GAP50–GFP revealed a periodic banding pattern running perpendicular to the direction of gametocyte extension (Fig. 5A, arrows). The average width of the GAP50–GFP fluorescence bands was 460±40 nm, with the interleaved regions of low fluorescence being 120±40 nm in width (Fig. 5A; supplementary material Movie 3).

Cryo-electron tomography allows 3D visualization of specimens after plunge-freezing in cryogenic liquids to preserve membrane structures (Cyrklaff et al., 2007; Kudryashev et al., 2010). Tomographic analysis of a stage IV gametocyte permits visualization of virtual sections near the surface of the cell without the need for physical sectioning. Fig. 5Bi highlights the microtubule network running underneath the cell surface and the locations of the IMC and parasite membranes within the cell. The section in Fig. 5Bii was physically located 100 nm above the section in Fig. 5Bi and shows regions of membrane (approximately 400 nm in width) interleaved with connecting bands (about 100 nm in width), layered across the underlying microtubules. A higher magnification view of the microtubules (Fig. 5C) highlights the spacing and arrangement of the microtubule network within the mature stage parasite; the microtubules (25 nm diameter) are spaced at intervals of approximately 10 nm. The spacing, shape and orientation of the flattened membrane structures are reminiscent of the GAP50–GFP-labelled features (Fig. 5A; supplementary material Fig. S3; Movie 5a,b). The intervening bands might represent areas of deposition of proteins that stabilize the IMC and the underlying microtubule network.

In stage V, the ends of the gametocytes become rounded (Fig. 4Aiv, Fig. 5Aiii). This is probably due to dismantling of the supporting microtubule network as described below and in a previous study (Sinden et al., 1978). GAP50–GFP persists at the periphery of the parasite still co-locating with BODIPY-TR-ceramide (Fig. 4Aiv). 3D-SIM revealed the maintenance of the IMC around the cell periphery, and the persistence of the stripes (Fig. 5Aiii, arrows). The gametocyte presented in supplementary material Movie 3 shows the transition from stage IV to stage V with one rounded and one pointed end. In this rotation, the stripes are clearly observed and there is evidence for some remnant GAP50–GFP fluorescence around the parasite nucleus.

We investigated the assembly of microtubules during gametocyte development. The microtubule-labelling reagent, Tubulin Tracker (Invitrogen) started to accumulate in a peripheral region of a stage II gametocyte (Fig. 6Ai; supplementary material Movie 4) before becoming concentrated along one edge of the stage III gametocyte (Fig. 6Aii; supplementary material Movie 4). Extension and elaboration of the microtubule network was observed in stage IV gametocytes (Fig. 6Aiii; supplementary material Movie 4), adopting a pattern analogous to that of the peripheral population of GAP50–GFP (Fig. 4Aii). No microtubule filament bundles were observed in the stage V gametocyte, consistent with the disassembly of the network in mature gametocytes.

We used the enhanced resolution of 3D-SIM to investigate the locations of the microtubules and IMC during gametocyte elongation. Antibody against β-tubulin labelled microtubules within the ‘foot’ of the developing gametocyte (Fig. 6B, white arrows) in addition to a diffuse cytoplasmic staining. β-tubulin in the foot region was co-located with GAP50–GFP (Fig. 6B). The distribution of the microtubules and the IMC can best be appreciated in supplementary material Movie 5a,b, which shows a translational through and a rotation of the same parasite. GAP50–GFP and β-tubulin were also co-located in the loop around the top of the gametocyte (Fig. 6Bii; supplementary material Movie 5a,b). Some GAP50–GFP-labelled regions around the edge of the foot did not have associated microtubules (supplementary material Fig. S3; Movie 5a,b). Electron micrographs of equivalent stages show areas where the IMC had started to expand but without any underlying microtubules (Fig. 4Bii, red arrows). By stage IV there was a complete network of microtubules underlying the peripherally located IMC and forming a similar pattern to that for GAP50. This characteristic shape and the distribution of the microtubules can be best appreciated in the 3D-SIM reconstruction shown in supplementary material Movie 6.

During stage III we saw a marked difference in the locations of GAP45 compared with GAP50–GFP and β-tubulin. This protein was restricted to the foot and was not co-located with the ER-associated population of GAP50–GFP (Fig. 6B,C; supplementary material Fig. S2B).

Gametocytes do not display gliding motility and it is unlikely that the IMC and associated MyoA participate in a motor complex analogous to that operating in the zoite stages of P. falciparum. To test for a potential role of actin, we examined the effect of cytochalasin D on gametocyte morphology. Previous work has shown that merozoites treated with cytochalasin D (0.1–2 μM) are rendered immobile and unable to invade erythrocytes (Miller et al., 1979; Srinivasan et al., 2011). By contrast, we found that maintenance of gametocytes over a period of 6 days in the presence of cytochalasin D (at concentrations up to 10 μM) did not prevent the morphological changes that accompany development to stage IV (data not shown). This indicates that although MyoA interactions might play an important structural role, morphological changes are not driven by an actin–myosin motor.

In an effort to determine a functional role for the remarkable shape changes of gametocytes, we used ektocytometry to measure cellular deformability at different stages of development (Fig. 7). The elongation index (EI) of synchronous enriched gametocytes was measured at sheer stresses of 0–20 Pa. Asexual trophozoite stage parasites showed a significant decrease in ability to elongate under flow (EI=0.13) compared to uninfected RBCs (EI=0.34) at the physiologically relevant sheer stress of 3 Pa (Fig. 7A). Stage III gametocytes showed a similarly low deformability (EI=0.12) (Fig. 7A). By contrast, there was an increase in the EI upon transition to stage IV of gametocyte development (EI=0.16); this value was significantly higher than that for stage III (P=0.0001). A further increase in the EI was observed upon transition to stage V (EI=0.18) (Fig. 7A).

We made a rough quantification of the change in the shape at different stages of development by monitoring length and width in DIC images of live gametocytes. Mature trophozoite stage parasites displayed a roughly circular profile with a mean length of 4.31±0.05 μm and mean width of 4.0±0.2 μm (Fig. 7B,C). The mean length of the gametocytes increased significantly from stage III (6.6±0.4 μm) to stage IV (10.9±0.4 μm) (P=0.0001), with a slight relaxation upon transition to stage V (9.1±0.4 μm) (P=0.005) (Fig. 7B,C). The width of the stage III gametocyte (3.0±0.7 μm) was significantly less (P=0.0001) than that of the asexual stage parasite, with a further significant decrease in the stage IV gametocyte (2.6±0.1 μm, P=0.006), before remaining stable for the remainder of its development (stage V; 2.6±0.2 μm) (Fig. 7B,C). The length-to-width ratio increased from 1.1 (trophozoite) to 2.2 (stage III) to 4.3 (stage IV), and then decreased to 3.5 (stage V).

In 1880, Alphonse Laveran observed gametocytes in the blood smear of an Algerian malaria patient and remarked upon the crescent shape of the parasite (Laveran, 1880). Over a century later we are still battling to understand the molecular basis for this unusual shape and its role in P. falciparum transmission and virulence. Early ultrastructural analyses revealed that the gross morphological restructuring of the gametocyte is coincident with the appearance of a tri-laminar membrane structure subtended by a layer of structural microtubules (Aikawa and Beaudoin, 1969; Sinden, 1982; Sinden et al., 1978). Comprising a double-membrane-bound compartment underneath the parasite plasma membrane, this structure was named the sub-pellicular membrane complex (Sinden and Smalley, 1979). Indeed the presence of alveolar sacks subtending the plasma membrane is a unifying morphological feature of apicomplexa, dinoflagellates and ciliates. Different organism groups have adapted this membrane structure for different cellular functions, with roles ranging from shape maintenance (free-living protozoa), to protective armor (dinoflagellates) to calcium stores (ciliates). In the motile stages of apicomplexan parasites, the sub-pellicular membrane is termed the IMC and plays a vital role as the scaffold for the motor complex that drives parasite motility (Beck et al., 2010; Frénal et al., 2010; Gaskins et al., 2004).

Our work confirms previous studies showing that the gametocyte sub-pellicular membrane complex has ultrastructural similarity to the IMC of the motile stages of P. falciparum (Bannister et al., 2000; Sinden and Smalley, 1979); however, the molecular similarity between these structures has not previously been investigated. In this work, we show that key proteins of the merozoite glideosome complex (GAP50, GAP45, MyoA and MTIP) are present in the gametocyte sub-pellicular membrane complex. We also show that exogenous GAP50–GFP forms a complex with endogenous GAP45, GAP50 and MTIP. These data indicate that the gametocyte membrane complex is closely related to the IMC of the motile stages of P. falciparum.

An interesting issue is the roles that the IMC and the associated glideosome proteins play in gametocyte assembly. Gametocytes do not display gliding motility, suggesting that the MyoA does not actively participate in an actin–myosin motor complex. Furthermore, we showed that treatment with cytochalasin D did not prevent gametocyte shape change. This indicates that the MyoA interaction with glideosome proteins probably plays a structural role rather than driving elongation per se. It seems likely that the microtubular cytoskeleton is attached to (and organized by) the IMC, which in turn is linked to the PPM via glideosome proteins. Thus, our data confirm and support the original suggestion (Sinden, 1982) that gametocyte elongation is driven by the assembly of microtubules underneath the IMC.

The shape adopted by P. falciparum gametocytes is reminiscent of the elongated sporozoite and ookinete. Indeed, the morphological changes that accompany maturation from stage II to stage IV of gametocytogenesis resemble (in reverse) the shape changes that accompany the transformation of the elongated sporozoites into liver stage trophozoites (Jayabalasingham et al., 2010) or ookinetes into oocysts (Carter et al., 2007). Taken together these data suggest that the IMC plays an important structural role in each of the cellular remodelling events that drive parasite elongation. This is consistent with recent work showing that specific IMC proteins can play roles in both motility and cell morphology (Tremp and Dessens, 2011).

Recent work from our laboratory indicates that the merozoite IMC is formed by the redistribution of a sub-compartment of the ER to nascent apical caps (Yeoman et al., 2011). Similarly, we show here that P. falciparum GAP50 is located in the ER in early stage gametocytes and that part of this population is recruited to the periphery to form the IMC. This is in agreement with early ultrastructural study of P. falciparum (Sinden et al., 1978) and with studies from other alveolates (Gould et al., 2008). Recruitment of GAP50 to the gametocyte periphery appears to be coordinated with the laying down of microtubules. Because the microtubule-supported IMC initially develops along one side of the gametocyte, stage III gametocytes have a ‘hat-like’ appearance. The IMC and the microtubule network then expand around the periphery, resulting in a cupped leaf shape. Our electron and fluorescence microscopy data suggest that the IMC extends first, followed by microtubule deposition, such that the leading edges of the nascent IMC are outside the supporting layer of microtubules. Other IMC proteins, GAP45 and MTIP appear to be recruited onto the nascent IMC. This process might be similar to the recruitment of pre-complexed GAP45–MTIP–MyoA onto membrane-embedded GAP50 during the formation of the Toxoplasma IMC (Gaskins et al., 2004).

3D-SIM imaging of the GAP50–GFP-labelled IMC revealed a remarkable architectural feature – bands of fluorescence of ~400 nm interrupted by ~100 nm regions of low fluorescence. Cryo-electron tomography revealed the ultrastructural analogue of these bands as flattened regions of membrane with intervening amorphous material. These structures are probably equivalent to the transverse ‘sutures’ encircling the gametocyte that were identified in an early freeze-fracture study (Meszoely et al., 1987). Similar patchworks of sub-pellicular membrane have been described in the IMC of the invasive stages of P. falciparum, as well as in Toxoplasma and other apicomplexa (Bannister et al., 2000; Raibaud et al., 2001). Another early transmission electron microscopy study identified protein ‘crosslinks’ running perpendicular to the microtubule network (Kaidoh et al., 1993). We suggest that the GAP50–GFP-labelled IMC ‘stripes’ represent IMC cisternae laid down as rectangular plates, whereas the gaps represent proteinaceous material that holds the cisternae and the microtubules in place.

A concurrent large-scale analysis of IMC components in P. falciparum revealed a novel IMC protein that appears to be restricted to the genus Plasmodium (Mal13P1.228). In the early gametocyte, this protein is located in thin stripes along the ‘foot’, which expand and eventually encircle the mature stage gametocyte (Kono et al., 2012). These structures might represent the regions of low fluorescence and proteinaceous crosslinks identified in our 3D-SIM and cryo-electron microscopy studies.

Consistent with previous ultrastructural studies of gametocytes (Sinden, 1982; Sinden et al., 1978), there is no morphologically identifiable apical polar ring in gametocytes to organize the peripheral microtubules [although a microtubule organizing center is formed very quickly after activation (Aikawa, 1977)]. This begs the question: how are the microtubules organized onto the well-ordered cytoskeletal basket? We suggest that the proteinaceous ‘sutures’ between the IMC plates provide a means of stabilizing the microtubule network. Proteins resident in these sutures might function in the targeting of the nascent IMC to the periphery of the cell and in the stabilization of the IMC and the sub-IMC microtubule network.

Analysis of DIC images revealed a significant increase in the length and a reduction in the width of the gametocyte during maturation as compared with asexual stage parasites. A more detailed analysis of gametocyte shape by cryo-X-ray tomography gives similar results (Hanssen et al., 2011). The elongated form of P. falciparum gametocytes is unusual. Gametocytes from other human malaria species and gametocytes from most other species of Plasmodium have a more rounded morphology, although gametocytes of avian plasmodia are also elongate (Sinden et al., 1978). P. falciparum is also unusual among human malaria parasites in that late stage asexual parasites drastically remodel the RBC through the export and insertion of parasite-derived proteins into and under the host RBC membrane (Tilley et al., 2011). These modifications lead to a significant increase in cellular rigidity, which would make mature parasite-infected RBCs vulnerable to recognition and clearance within the spleen (Glenister et al., 2002; Safeukui et al., 2008). The parasite avoids these clearance mechanisms by cytoadhereing within the microvasculature of the host. Early stage P. falciparum gametocytes also sequester; however, upon reaching sexual maturity they release from their sites of adhesion and re-enter the circulation to enable transmission to mosquitoes (Day et al., 1998; Hayward et al., 1999).

In an effort to determine the functional consequences of the gametocyte shape changes we investigated the ability of asexual trophozoites and stage III–V gametocytes to elongate in flow, a surrogate measure of cellular deformability (Groner et al., 1980; Hardeman et al., 1987). In agreement with previous reports, we found that asexual trophozoite-infected RBCs are highly undeformable (Nash et al., 1989). This loss of deformability is probably due to a combination of membrane rigidification and the presence of the large rigid intracellular parasite (Glenister et al., 2002; Nash et al., 1989). We found that stage III gametocytes have similar deformability properties to trophozoite-infected RBCs; however, upon transition to stage IV the gametocytes show a significant increase in the elongation index measured under flow conditions. This is probably due in part to the elongated shape. A further increase in the elongation index in stage V gametocytes might be due to the fact that the microtubule network is disassembled at this point, leaving an already elongated cell that probably extends further under flow pressure.

These changes in rheological properties might help stage V gametocytes survive in the circulation of the host. Indeed, stage V gametocytes might have rheological properties more akin to P. vivax-infected RBCs (Handayani et al., 2009; Suwanarusk et al., 2004). As its name suggests (vivax means lively), P. vivax is more amoeboid than P. falciparum and appears to avoid splenic clearance in spite of an inability to cytoadhere. Stage III–V gametocytes lack knobs and are thought to adhere with lower affinity than asexual stages and to prefer sites of reduced vascular flow (Rogers et al., 1996; Smalley et al., 1981). It is possible that another consequence of the elongated shape of P. falciparum gametocytes is enhanced adhesion to lower affinity receptors by permitting flattening of the cell, which might increase the area in contact with the capillary walls.

In conclusion, our data show that the falciform shape of P. falciparum gametocytes is supported by an IMC with very similar composition to the merozoite IMC, linked to underlying microtubules. A previously unrecognized correlate of gametocyte elongation is an enhanced cellular deformability. We anticipate that disruption of the IMC and its associated microtubules compromises the ability of P. falciparum to adopt the falciform shape; this might in turn affect its ability to survive in the circulation. This could provide a new avenue for interfering with this important human pathogen.

Parasite-infected RBCs were cultured in RPMI-HEPES supplemented with 5% human serum and 0.25% AlbuMAX II as previously described (Foley et al., 1994). Laboratory strains of 3D7 cultured for long periods form gametocytes with low efficiency, therefore we used a 3D7 isolate recovered from a patient (Lawrence et al., 2000) and cultured for only a limited time. A GAP50–GFP transgenic parasite was created by transfecting a high gametocyte producing clone of 3D7 with the pGLUX–GAP50–GFP construct as previously described (Yeoman et al., 2011).

Gametocytes were prepared using a modification of a published protocol (Fivelman et al., 2007). A culture of mainly ring stage parasites (6–8% parasitemia) was treated with 5% sorbitol (Lambros and Vanderberg, 1979), then separated using a Percoll density gradient (Knight and Sinden, 1982) to enrich the ring stage and remove any gametocytes. The parasites were cultured until they reached 8–10% trophozoites then sub-divided into 2% trophozoites (5% hematocrit). The culture was maintained for 10 days in the presence of 62.5 mM N-acetyl glucosamine to inhibit merozoite invasion and thus asexual replication (Hadley et al., 1986). Giemsa-stained slides were used to monitor stage progression. The medium was changed daily but no fresh erythrocytes were added. At the desired stages of development, gametocytes were enriched by magnetic separation (Fivelman et al., 2007). For analysis of the effect of cytochalasin D (Sigma-Aldrich), the drug was added at concentrations of 0–10 μM at day one of gametocyte generation and supplemented daily into the culture media. The parasites were assessed by microscopy at different stages of development.

Live infected RBCs were prepared for fluorescence microscopy as previously described (Dixon et al., 2011). Immunofluorescence was performed on acetone-fixed blood films (Spielmann et al., 2006) using the following primary antibodies in PBS containing 3% BSA: mouse anti-GFP (1:500, Roche), rabbit anti-GFP (1:500), rabbit anti-GAP45 [1:500 (Baum et al., 2006b)], rabbit anti-GAP50 [1:500 (Baum et al., 2006b)], rabbit anti-Pfs16 [1:1000 (Baker et al., 1995)], mouse anti-Pfs16 [1:1000 (Dixon et al., 2009)], mouse anti-MTIP [1:500 (Green et al., 2006)], mouse anti-MyoA (Baum et al., 2006b) followed by anti-mouse or anti-rabbit IgG conjugated to Alexa Fluor 568, Alexa Fluor 647 or FITC. Slides were labelled with 1 μg/ml of DAPI prior to mounting.

BODIPY labelling of parasite-infected cells was performed by adding BODIPY-TR-ceramide at a final concentration of 0.7 μM to the parasite culture. Cells were incubated overnight under standard culture conditions. Labelling with Tubulin Tracker Green was performed according to the manufacturer's instructions, using a final concentration of 250 nM and 30 minutes of incubation at 37°C. Microscopy was performed using an Olympus IX81 wide field microscope, or Leica TCS-SP2 or Zeiss LSM 510/FCS confocal microscopes. Images were processed using NIH ImageJ version 1.42 (www.rsbweb.nih.gov/ij). Samples were prepared and 3D-SIM (Microbial Imaging Facility, University of Technology, Sydney) performed as previously described (Yeoman et al., 2011). Analysis of the dimensions of at least seven cells from DIC images was performed using ImageJ. Isosurface renderings of GAP50–GFP-expressing gametocytes were generated using IMOD software (Kremer et al., 1996; Mastronarde, 1997).

Gametocytes were harvested on days 3, 5, 7, 9 and 11, representing stages I to V. Pellets of magnet-purified gametocytes were lysed in 0.015% saponin on ice for 15 minutes. The saponin pellets were solubilized in SDS loading buffer and equal amounts (10 μg protein) were separated by SDS-PAGE and transferred to nitrocellulose for immunoblotting.

For solubility studies, magnet-purified stage IV gametocytes were lysed in 0.015% saponin and the pellets incubated for 15 minutes with 1% Triton X-100 or RIPA buffer (1% IGEPAL CA-630), 150 mM NaCl and 0.5% sodium deoxycholate on ice, or with 0.1% SDS and 50 mM TRIS, pH 7.4 at room temperature or with 2% SDS at room temperature. The soluble supernatant and insoluble pellet fractions were collected. Equivalent amounts of protein were separated by SDS-PAGE and transferred to nitrocellulose for immunoblotting.

For immunoprecipitation studies, purified parasites were lysed in RIPA buffer for 15 minutes on ice in the presence of protease inhibitor, centrifuged, and the supernatants collected. The precipitating agents were GFP–TRAP reagent (Chromotek) and Protein A Sepharose and anti-rabbit GAP45 (1:500) (Yeoman et al., 2011). The pellets were analyzed by SDS-PAGE and immunoblotting blotting.

Immunoblotting was performed as previously described (Dixon et al., 2011). Briefly, all antibodies were prepared in PBS containing 3.5% skim milk, and both primary and secondary incubations were performed for 1 hour. Three washes of PBS containing 0.1% Tween-20 were performed between each incubation. The following primary antibodies were used: mouse anti-GFP (1:500, Roche), rabbit anti-GFP (1:500), rabbit anti-GAP45 [1:500 (Baum et al., 2006b)], rabbit anti-GAP50 [1:500 (Baum et al., 2006b)], mouse anti-GAP50 [1:500 (Baum et al., 2006b)], mouse anti-Pfs16 [1:1000 (Dixon et al., 2009)], mouse anti-MTIP [1:500 (Green et al., 2006)] and mouse mAb anti-AMA1 [mAb 1F9 1:500 (Coley et al., 2001)]. Goat anti-mouse or anti-rabbit IgG secondary antibodies conjugated to horseradish peroxidase were used. Electrochemiluminescence reagents and autoradiography were used to visualize the immunoblots.

Mature stage asexual (30–36 hour trophozoite) and stage III–V gametocytes were magnet-purified from culture, washed in PBS, then reconstituted with uninfected RBCs to obtain a working parasitemia of 70% (Maier et al., 2008). The cell samples (5 μl) were mixed with 500 μl of polyvinylpyrrolidone (PVP) solution at a viscosity of 25 mPa second (Rheo Meditech). The elongation index was measured in a RheoScan Ektocytometer according to the manufacturer's instructions. Measurements were acquired over the 0–20 Pa range. Values shown are means±s.e.m. from a minimum of three individual measurements. The results shown are from a single experiment. The experiment was repeated showing comparable results (two separate experiments).

Transmission electron microscopy and tomography of asexual stages were performed as previously described (del Pilar Crespo et al., 2008; Hanssen et al., 2010); for merozoites, the cells were post-fixed with potassium-ferricyanide-reduced osmium tetroxide. Gametocytes from selected stages were fixed and embedded using a modification of a published protocol (Kass et al., 1971). Briefly, cells were fixed in 5% glutaraldehyde in PBS, pH 7.3, overnight, then embedded in 3% agarose before undergoing post-fixation of lipids in osmium tetroxide (1% in PBS) for 1 hour followed by progressive dehydration in ethanol. Samples were further dehydrated using progressive ethanol:acetone ratios of 2:1, 1:1 and 1:2, and finally immersed in acetone for 10 minutes. Acetone was gradually replaced with liquid epoxy resin, and samples embedded and stained as described above. For both asexual and gametocyte samples, 70-nm thick sections were observed at 120 keV using a JEOL 2010HC transmission electron microscope (EM Unit, La Trobe University,). For electron tomography, sections of 200 or 300 nm were cut and stained and observed at 200 keV on an FEI Tecnai G2 F30 electron microscope (Bio21 Institute, Melbourne). Tilt series were collected every 2° between −70° and +70° and reconstructed to generate cell models as described previously (Abu Bakar et al., 2010; Yeoman et al., 2011). Tomograms were generated using the IMOD software (Kremer et al., 1996; Mastronarde, 1997). IMOD was also used to generate segmentation models where contours were assigned manually.

Gametocytes were transferred onto carbon-coated grids. After removal of excess liquid, grids were rapidly frozen by plunging into liquid ethane then transferring to liquid nitrogen. Grids were mounted on a Gatan cryo-holder and imaged using a Tecnai G² F30 (300 keV). Tilt series were collected every 2° between −70° and +70°. The total electron dose kept was 6000 e⁻/nm².

We would like to thank Alex Lowdin, Michael Jones and Michael Johnson for assistance with microscopy protocols. We thank Tony Holder (National Institute for Medical Research, London) for donating anti-MTIP antibodies, Robin Anders (La Trobe University) for anti-AMA1 antibodies, Jake Baum (Walter and Eliza Hall Institute) for the anti-GAP50, GAP45 and MyoA antisera and David Baker (London School of Hygiene and Tropical Medicine) for anti-Pfs16 antibodies. We thank Tim Gilberger and Tobias Spielmann (Bernhard Nocht Institute) for useful discussions.