Erythrocyte invasion by merozoites forms of the malaria parasite is a key step in the establishment of human malaria disease. To date, efforts to understand cellular events underpinning entry have been limited to insights from non-human parasites, with no studies at sub-micrometer resolution undertaken using the most virulent human malaria parasite, Plasmodium falciparum. This leaves our understanding of the dynamics of merozoite sub-cellular compartments during infectionincomplete, in particular that of the secretory organelles. Using advances in P. falciparum merozoite isolation and new imaging techniques we present a three-dimensional study of invasion using electron microscopy, cryo-electron tomography and cryo-X-ray tomography. We describe the core architectural features of invasion and identify fusion between rhoptries at the commencement of invasion as a hitherto overlooked event that likely provides a critical step that initiates entry. Given the centrality of merozoite organelle proteins to vaccine development, these insights provide a mechanistic framework to understand therapeutic strategies targeted towards the cellular events of invasion.
Invasion and parasitism of the human erythrocyte is central to the pathology of malaria disease (Cowman et al., 2012). Much of our understanding of the process of red blood cell entry has been gleaned from microscopic dissection of host cell invasion by non-human malaria parasites (Ladda et al., 1969; Bannister et al., 1975; Dvorak et al., 1975; Aikawa et al., 1978) with comparative insights gained from invasion by other apicomplexan zoites, in particular Toxoplasma gondii (Michel et al., 1980; Dobrowolski and Sibley, 1996). Only recently has understanding been extended to the most virulent human malaria parasite Plasmodium falciparum (Gilson and Crabb, 2009; Boyle et al., 2010; Riglar et al., 2011). Although molecular investigations have yielded a partial understanding of invasion, and many participating molecules are now known, exact roles remain unclear for most proteins (Cowman and Crabb, 2006; Cowman et al., 2012). The process of merozoite invasion is very rapid, involving discrete kinetic steps (Dvorak et al., 1975; Gilson and Crabb, 2009). On commencement of invasion merozoites adhere reversibly to blood cells in a non-directional manner then reorient so that the merozoite apex points toward the erythrocyte (Dvorak et al., 1975; Gilson and Crabb, 2009). Once complete, the alignment appears to be stable, and the merozoite commences active invasion.
Key to the penetration of the erythrocyte is formation of an electron dense attachment ring that links the merozoite to the invaded host cell (Bannister et al., 1975). Described as the tight or moving junction (Ladda et al., 1969; Bannister et al., 1975; Aikawa et al., 1978), this structure acts as an organizing nexus around which the core events of invasion take place, including shedding of the merozoite surface coat, activation of the parasite actin–myosin motor and release of apical secretory organelles (Bannister et al., 1975; Aikawa et al., 1978; Miller et al., 1979; Pinder et al., 1998; Boyle et al., 2010; Riglar et al., 2011). Secretion from apical organelles is crucial for the initiation of invasion but also lays the foundation for parasitophorous vacuole (PV) formation, the compartment within which erythrocytic development progresses (reviewed in Lingelbach and Joiner, 1998). The timing of organelle secretion appears to co-ordinate the process of host-cell entry with the three main classes – the rhoptries, micronemes and dense granules – defining much of the invasion biology of apicomplexan parasites (reviewed in Carruthers and Sibley, 1997; Blackman and Bannister, 2001). Many organelle proteins are known to function during invasion and, as such, are major vaccine targets (Cowman and Crabb, 2006; Cowman et al., 2012), but exactly how each compartment secretes from the parasite is still not clear.
A major challenge towards complete molecular and cellular dissection of invasion is the merozoites' extremely small size (∼ 1 μm) and, for P. falciparum in particular, the lack of robust methods for the successful isolation of viable merozoites. Serial section reconstruction of different lifecycle stages has been used successfully to analyse various cellular structures during development (Prensier and Slomianny, 1986; Hopkins et al., 1999; Waller et al., 2000; Bannister et al., 2000a); however, existing reconstructions of merozoites are limited to chemically fixed preparations that have limited resolution in the vertical ‘Z’ plane. Furthermore, no study has applied such technology towards a detailed inspection of the invasion process.
Here, we address prior limitations in imaging merozoite invasion using viable isolated merozoites (Boyle et al., 2010) and three-dimensional reconstructions by cyro-electron and X-ray tomography to image whole cells and sectioned specimens with a wide variety of chemical and cryo-fixation methods. This enables us to build a robust high-resolution four-dimensional reconstruction of the cellular events underlying invasion. A major revelation of this approach is the hitherto overlooked process of fusion between rhoptries, described 40 years ago by Aikawa et al. (1978), as an early defining event in the entry process. Combined with insights into the inner membrane complex and endomembrane system, persistence of heterochromatic nuclear architecture, and the internal compartmentalization of rhoptry organelles, this study provides the most complete cellular overview of merozoite invasion of the erythrocyte to date.
Results and discussion
To examine the ultrastructure of the blood stage malaria parasite infection we isolated viable invasive P. falciparum merozoites (Boyle et al., 2010) and used these in conjunction with several different methods to check for fixation artefacts and explore parasite cell features differentially preserved by each approach. Six different methods were used (Table 1) with samples examined in depth with the aid of X-ray tomography of whole parasites (Hanssen et al., 2011; 2012), electron tomography of whole and serial sectioned parasites (Cyrklaff et al., 2007; Hanssen et al., 2008; Abu Bakar et al., 2010), and traditional reconstruction of transmission electron microscopy (TEM) serial sections (see Experimental procedures). Each fixation method generated specific insights (summarized in Table 1) with three-dimensional reconstruction of merozoites yielding characterization of features that were not apparent in two dimensions. Combined, these imaging approaches define a series of methodological frameworks for exploring the sub-cellular architecture of the invasive malaria parasite blood stage.
Table 1. List of electron microscopy fixation methods
Good overall preservation of merozoite shape and organelles. Allows visualization of membranes – membrane contrast far superior to standard osmium treatment, but with loss of contrast for proteinaceous structure and DNA. Unsuitable for post-embedding immunolabelling.
Seligman et al. (1966); Willingham and Rutherford (1984)
Glutaraldehyde and osmium tetroxide
TEM and electron tomography of serial sections
Good preservation of membranes. Variable contrast for proteinaceous structures. Generally unsuitable for post-embedding immunolabelling.
Methodologically more complex and fixation requires more expensive infrastructure. Excellent preservation of merozoite and organelle structure and shape, lacking ruffles sometimes observed in the chemical fixations (ROTO, GO, GF). Dense granules and mitochondria appeared to be smoother, denser and more turgid. Compatible with post-embedding immunolabelling.
Cryo-preservation by plunge freezing in liquid ethane
Electron tomography of whole cells (individual)
Excellent whole-cell preservation down to potentially molecular detail. Resolves some cytoskeletal elements not discernible in embedded cells. The resolution and contrast degrade with sample thickness, practically limiting the cell thickness to ∼ 0.5–1 μm. No need for staining. Not readily suitable for immunolabelling internal structures
Cryo-preservation in capillaries for X-ray imaging
X-ray tomography of whole cells (multiple)
Excellent whole-cell preservation – no material lost through sectioning. Contrast not altered by staining. Resolution higher than light microscopy but lower than electron microscopy. Membranes not visible. Fast acquisition image and tomography of multiple cells. Not readily suitable for immunolabelling internal structures
High definition cellular dissection of the P. falciparum merozoite
The apical complex
A major improvement obtainable via the different three-dimensional approaches was seen with the apical complex (Fig. 1). Of note, the structure, a defining feature of apicomplexan parasites (Lee et al., 2000), was observed to contain three electron dense concentric rings (known as polar rings), with the smallest directly adjacent to the apical tip as has been previously reported (Morrissette and Sibley, 2002; Bannister et al., 2003). This was visible under chemical fixation conditions (Fig. 1A–C), but was more clearly apparent using cryo-electron tomography (CET), where the rings were visible without chemical contrast (Fig. 1D–F, Movie S1). We observed that the third ring is much thicker and flatter than the two apical cylindrical rings (Fig. 1E and F, Movie S1), a feature apparent in plates and drawings from other studies (e.g. Bannister et al., 2003). In coccidians, two apical polar rings (called preconoidal rings) are separated from the basal ring (Scholtyseck et al., 1970) (inconsistently called the apical ring or the microtubule organizing centre or MTOC) by a conoid structure, an arrangement that is clearly very different in the Haemosporidia, with the third ring serving as anchor point for attachment of the numerous longitudinal microtubules (Morrissette and Sibley, 2002). Such an arrangement is clearly divergent in the case of P. falciparum where only two or three longitudinal microtubules are attached adjacent to each other (see below). Thus, the different structure of the most basal ring in the merozoite remains unexplained.
As previously described, the apical extremity of the merozoite was observed to be devoid of a pellicular network of flattened membrane-bound compartments called the Inner Membrane Complex (IMC) (Aikawa, 1967), which covers the remainder of the internal surface of the merozoite (Fig. 1A and F). Tomographic reconstruction revealed that the apical extremity of the IMC terminates against the external side of the third and most basal of the polar rings (Fig. 1F, Movie S1). Asymmetrically arranged longitudinal microtubules which run most of the length of the merozoite also end at this site (Fig. 1D and E) with cryo-preserved merozoites demonstrating two (Fig. 1E, Movie S1) or three well-preserved filaments (Fig. 1D). This is consistent with previous descriptions of microtubule numbers of either two or three in the merozoite (Aikawa, 1967; Bannister et al., 2003). When imaged by cryo-X-ray (CX), cryo-preserved merozoites facilitated easy imaging of merozoite form and rhoptry morphology (Fig. 1G–I), but no other internal organelles were readily visible using these conditions. Both endosymbiotic organelles, the apicoplast and mitochondrion, were unpredictably preserved in all conditions, but were often positioned very close to one another and found immediately apical to the nucleus to one side of the merozoite (Fig. 1C). A detailed CET description of the merozoite apicoplast is reported elsewhere (Lemgruber et al., 2013).
The nucleus and the endomembrane system
The endomembrane system of Plasmodium parasites is poorly characterized compared with many other organisms, partly because of the poor resolution obtainable from light microscopic analyses of such a small cell, and in part because the endoplasmic reticulum (ER) and Golgi often are very poorly preserved in fixed preparations of P. falciparum ring or trophozoite stage parasites, even using methods that work very well on related apicomplexan parasites (Lemgruber et al., 2010). The small size of the merozoite and simplicity in its overall architecture provides an opportunity to reconstruct the entire complement of the endomembrane system. The endomembrane system was relatively well-preserved using high-pressure freezing/freeze substitution (HPF/FS) (Fig. 2A and B), and the nuclear and ER membranes were readily discernible using the reduced osmium fixation (ROTO) method (Fig. 2C–G). Ribosomes were not resolvable by X-ray tomography or the ROTO fixation method (McDonald, 1984) (Figs 1H and 2E), but they were particularly well visualized by CET (Fig. 1F, Movie S1) or conventional EM using HPF and glutaraldehyde fixation (GF) samples (Fig. 2B and Fig. 3A–F). Typical for Plasmodium fixations, the Golgi ultrastructure was poorly preserved by all of our methods and is not clearly distinguishable from the ER, which is contiguous with the nuclear envelope (Movie S2).
On close inspection of the nuclear membranes, electron tomographic reconstructions demonstrated that merozoite nuclei typically possessed 3–4 nuclear pores, consistent with counts seen in segmented schizont nuclei (Weiner et al., 2011). These were positioned on the apical face of the nucleus (Fig. 2C–E). This may be similar to the situation in sporozoites, where nuclear pores are only found on one side of the nucleus, possibly due to size exclusion as the other side is tightly associated with the IMC (Kudryashev et al., 2010). Heterochromatin and euchromatin-like material in the nucleus were well separated into coherent blocks in HPF/FS merozoites, which is consistent with previous observations in schizont nuclei (reviewed in Ralph and Scherf, 2005), but distinct from that observed in early ring-stage nuclei (Weiner et al., 2011).
The IMC (also called alveoli) is one of the major endomembrane systems within the parasite pellicle (Gould et al., 2008). Although considered as a scaffold that determines cellular structure and anchors proteins required for parasite motility and invasion (Bergman et al., 2003; Khater et al., 2004; Baum et al., 2006; Kono et al., 2012) the full extent of the roles played by the IMC is still unclear. The IMC is believed to develop from a post-Golgi compartment (Bannister et al., 2000b; Agop-Nersesian et al., 2009; 2010). Using ROTO we were able to trace reticulations of the ER and IMC system in several cells (Fig. 2E–G, Fig. S1, Movie S2). In many of these the membranes of the IMC appear to be continuous with the membranes of the ER and nuclear envelope (Fig. 2F, Fig. S1, Movie S2). We attempted to find similar connections in high-pressure frozen samples, which should have fewer membrane artefacts, but the contrast of the relevant areas was poorer for these preparations than for the ROTO prepared cells, and no such connections were obvious.
Tomograms from CET- and ROTO-fixed merozoites revealed that the IMC envelops most of the basal half of the merozoite in a rarely interrupted sheath (Fig. 2F and G, Fig. S2), with the exception of a hole around the single cytostomal ring (Movie S2). Numerous discontinuities in this sheath exist at the midpoint (or waist) of the merozoite (Fig. 2F, G and H, Movie S2, Fig. S2). Towards the apical third of the merozoite, a more continuous IMC was seen, with the exception of the distinct single opening at the apical ring (Fig. 1A–C, Fig. 2F and G). This non-uniform IMC distribution, which is visible (though unnoted) in micrographs presented in previous studies (Aikawa et al., 1978; Mitchell and Bannister, 1988), has clear implications for the spatial organization of the motility apparatus. This apparatus, also known as the glideosome, is connected to proteins embedded in the IMC and in the plasma membrane (Bergman et al., 2003; Khater et al., 2004; Baum et al., 2006; Kono et al., 2012), so gaps in the IMC would suggest that continuity of adhesion and de-adhesion cycles may be disrupted during traction progression towards the merozoite's basal end. In addition, the openings in the IMC may provide points at which invasion organelles like dense granules might fuse with the plasma membrane post invasion (Torii et al., 1989; Riglar et al., 2013). The possibility of continuity of the IMC to internal endomembrane compartments also raises the prospect of a continuum of protein exchange with the ER-Golgi. Some protein markers are clearly distinct between ER lumen and IMC, but the luminal composition of the IMC is only minimally characterized (Yeoman et al., 2011).
High definition dissection of the process of merozoite invasion
A three-dimensional overview of invasion
We extended our electron microscopy analysis of the merozoite to the process of invasion itself, providing the first three-dimensional cellular presentation of erythrocyte entry at this level of resolution (Fig. 3A–E). One of the powers of using different fixation methods is the ability to identify artefacts unique to particular methods. In most previous studies of invasion, the presence of an electron lucent gap is near universally observed in front of the invading merozoite between the parasite plasma membrane and nascent parasitophorous vacuole membrane (PVM), sometimes referred to as the invasion pit (Bannister et al., 1975). Critically, HPF/FS samples displayed no such gap (Fig. 3F). This demonstrates that the space is likely an artefact of the chemical fixation and dehydration process, and suggests that association between the merozoite surface and erythrocyte plasma membrane are likely very close, even beyond the boundaries of the tight junction. A second phenomenon noted was the presence of a ‘hook’ arising from the surface of the invading merozoite that appears embedded into an invaginated plasma membrane of the target erythrocyte (Fig. 3G and H). Seen under GF, glutaraldehyde/osmium fixation (GO) and ROTO fixation methods in parasites that are in contact with the red cell but not yet oriented with their apical end towards it, this structure was sited in zones distal from the parasite apex and, as such, likely arises from the merozoite plasma membrane and not a secretory organelle. These attachment points reached deep into the erythrocyte and were coated with material continuous with the merozoite coat (Fig. 3G and H). Of 16 complete tomograms reconstructed for parasites in contact with red blood cells, six were found in this non-apical orientation, and each of these had hook-like protrusions into the red blood cell. It is uncertain if such structures are part of the process that leads to reorientation, aberrant interactions that would lead to aborted invasion or a consistent artefact of chemical fixation methods. Forceful deformations of the red blood cell are apparent in live cell videos of invasion (Gilson and Crabb, 2009), and it is conceivable that these invaginations at non-apex attachment sites result from dynamic interactions between the merozoite and erythrocyte or underlying changes in the tension of the host cell cytoskeleton. Alternatively, such extended zones of interaction might result as a fixation artefact given the numerous adhesive interactions that govern interactions between the parasite and erythrocyte surface. Irrespective of their origin, and while treating such structures with due caution, either scenario points to a closer degree of interaction between merozoite surface and erythrocyte than has previously been appreciated.
The invasion organelles – micronemes and dense granules
Invasive zoites from across the phylum Apicomplexa are characterized by the presence of at least three secretory organelles associated with motility and invasion, the micronemes, dense granules and rhoptries (Blackman and Bannister, 2001). Micronemes are small, apically oriented, teardrop shaped structures (Scholtyseck and Mehlhorn, 1970) that release major invasion ligands (Sim et al., 1992; Healer et al., 2002; Treeck et al., 2006) at the commencement of active invasion (Singh et al., 2010), possibly prior to or during merozoite egress (Riglar et al., 2011). Previous studies suggested that microneme contents might be exposed to the surface by fusion with the rhoptry neck, whereupon their release is via the apical opening of the rhoptry (Bannister et al., 2003; Ravindran et al., 2009). Under all fixation conditions, micronemes were frequently clustered around the periphery of the rhoptry (Fig. 1B and F, Fig. 2F, Fig. 3C and Fig. 4A) as previously described (Scholtyseck and Mehlhorn, 1970) but no fusion events were apparent under any fixation methods (Fig. 4A–D). Such events are also unclear from analysis of Toxoplasma tachyzoites (Lemgruber et al., 2010; Paredes-Santos et al., 2012). Given the rapidity of the onset of egress through to invasion in P. falciparum (Dvorak et al., 1975; Glushakova et al., 2005; Gilson and Crabb, 2009) our failure to capture this event certainly does not preclude its existence. We are therefore unable at present conclude whether micronemes release via direct fusion with the plasma membrane or to the rhoptry. Given that individual micronemes persisted even into later invasion stages (Fig. 4F), however, these data certainly suggest microneme release appears to be via each individual organelle, and in gradual stages, rather than fusion and release en masse.
Dense granules are generally defined by their, relatively, electron-dense nature and posterior positioning with respect to the rhoptries and micronemes (Bannister et al., 1975) (Fig. 3F and Fig. 4A–D). Most dense granule proteins appear to be involved in modifying the host cell post invasion, rather than playing a role in invasion (Mills et al., 2007; Pei et al., 2007; Bullen et al., 2012; Riglar et al., 2013) consistent with reports that their release is at a late stages of invasion/post-invasion (Bannister et al., 1975; Torii et al., 1989; Culvenor et al., 1991; Riglar et al., 2011). Consistent with this, we only saw dense granules closely apposed to the plasma membrane in invaded parasites (Figs 3F and 4F). We also, however, frequently observed organelles that appear to be intermediate in density and size between micronemes and dense granules (Fig. 3E, Fig. 4B, D and F) that would support the growing appreciation of heterogeneity in these invasion organelles (Zhou et al., 2005; Singh et al., 2007). Future work towards correlative fluorescence-electron microscopy should enable a molecular resolution of this heterogeneity.
The invasion organelles – the rhoptries
The most visibly imposing invasion organelles of merozoites are the rhoptries (Hepler et al., 1966). These are large membrane bound club-shaped compartments with a curved bulb oriented towards the centre of the merozoite (Bloom and Aghajanian, 1966; Hepler et al., 1966; Bannister et al., 1977; 2000b). Plasmodium merozoites each possess two rhoptries, whose narrow extremities are in close proximity to the apical protrusion. Other apicomplexan species possess more than two rhoptries, while sporozoite stages of the rodent malaria parasite Plasmodium berghei, and perhaps other Plasmodium species, have up to four rhoptries (Schrevel et al., 2008; Kudryashev et al., 2010),
Prior to invasion, chemically fixed rhoptries have two distinct zones, with the rhoptry neck being visually distinct from that of the bulb (Bannister et al., 2000b; Lemgruber et al., 2010). This can be seen by electron microscopy using GF, and GO fixation (Fig. 3A–E, Fig. 4G and H) as the interior of the rhoptry neck is considerably more electron dense than the bulb. The transition from bulb to neck is quite distinct, though no membrane separates the two subcompartments. The different zones are known to correlate with protein distribution (Sadak et al., 1988; Sam-Yellowe et al., 1995; Duraisingh et al., 2003; Richard et al., 2010; Zuccala et al., 2012); however, the effect of different chemical treatments, which will extract different components within the rhoptry, and corresponding differences in electron density seen under TEM may not perfectly reflect the native content of proteins, lipids or sugars in each compartment. Following this, exploration of rhoptry morphology using GF, ROTO, HPF, CX and CET fixation methods highlighted several details. First, the key distinction in electron density between contents of rhoptry neck and bulb (Fig. 4G and H, Fig. 5B) dissipated on commencement of merozoite invasion once rhoptry expulsion initiated (Fig. 3C and D, Fig. 5B–E). This supports the intuitive idea that content of the rhoptry neck is expelled first, but also indicates that the shape of the rhoptry is not determined purely by the contents of the lumen – i.e. shape is retained post secretion. Possible alternative explanations for the maintenance of this shape include constraint by an external or internal scaffold (Lemgruber et al., 2010) or even differences in membrane composition of the neck and bulb, ideas that will require further biochemical characterization. Investigation using the CET method demonstrated no detectable difference in electron density between rhoptry bulb and neck (Fig. 4I). This was also observed in the CX prepared micrographs (Fig. 1G–I). These data indicate that differences in density are an artefact of stain affinity for different components rather than a native difference in protein or lipid concentration. Thus the mechanism by which subcompartmentalization is achieved now becomes a priority towards understanding how this key organelle releases key invasion proteins sequentially.
A second major observation related to the nature of rhoptries during the invasion process. In mature merozoites that were still retained within schizonts, and in free merozoites, nearly all merozoites observed possessed two independent rhoptries of similar size and shape, irrespective of fixation conditions (Fig. 5A). Upon invasion we noted that, as previously observed, the membranes at the apical tip of the rhoptry fuse with the apical membrane of the merozoite, so that the rhoptry lumen is now topologically at the exterior of the merozoite (Fig. 5C). However, using a variety of specimen preparation methods, we consistently noted only a single rhoptry fused at this position in invading merozoites (Fig. 5B). Examination of electron micrographs and tomograms of invading merozoites revealed that in early invading merozoites the two rhoptries were clearly in the process of fusing with each other (Fig. 5D and E, Movie S3). Tomograms generated included rhoptries undergoing fusion at the neck, forming a bifurcated rhoptry with two separate bulbs and by the time of active erythrocyte penetration, resolved into only a single rhoptry with a single neck and bulb (Fig. 5E). These latter states were often observed with membranous whorls within the rhoptry (Fig. 5B, D and E) reminiscent of those noted previously in P. knowlesi (Bannister et al., 1986).
The reproducibility of alternate fixation methods permitted an attempt to make quantitative measurements of volumetric indices for the merozoite and the internal organelles. CX-tomography in particular was found to be well suited to rapid generation of reconstructions for a large number of merozoites (Fig. 1G and I). In total, three-dimensional models were generated for 31 free merozoites. The rhoptry pair was clearly identifiable in 27 of these merozoites. From these models, the average merozoite volume was estimated as 1.73 × 109 nm3 with an average volume of each pair of rhoptries estimated as 6.1 × 107 nm3 (comprising ∼ 3.5% of the cellular volume) (Table 2). Volumetric models were also made from electron tomograms of free and invading merozoites preserved using GF, GO and ROTO. Free merozoites from GF and GO were around 40–70% smaller than those measured using ROTO and CX (Table 2), consistent with known shrinkage artefacts of such fixations (Choi and Stoecker, 1989). Nuclear volumes could be more accurately assessed in the electron tomograms, and were found to represent ∼ 25% of the cellular volume (Table 2). The accuracy and degree of confidence of nuclear measurements with CX is low due to the low contrast of this organelle compare with the surrounding cytoplasm. During invasion, volumetric models of the single fused rhoptries demonstrated a volume similar to the combined volume of pairs of rhoptries prior to fusion (Fig. 5F) but as invasion proceeded, rhoptries were found to steadily diminish in size (Fig. 5F). Even before merozoites were completely enveloped, the total rhoptry size shrank to around 20–25% of that of the original rhoptry pair (∼ 1 × 107 nm3 compared with ∼ 4 × 107 nm3) (Fig. 5F). The surface area of the rhoptries diminished more slowly (data not shown) than the change in volume due to changes in rhoptry shape during invasion. Although some of the limiting rhoptry membrane is thought to be incorporated into the PV, a considerable portion of the rhoptry membranes still exists in merozoite once the PVM is fully closed around the merozoite, as observed in other Plasmodium species (Bannister et al., 1975). Of the ten whole invading merozoites fully reconstructed by serial tomography, all ten had fused rhoptries.
Table 2. Whole-cell and organelle-specific volumes for free merozoites
Merozoite volume, (nm3) ± SD
Nuclear volume, (nm3) ± SD
Rhoptry volume, (nm3) ± SD
6 × 108 ± 3.6 × 107, n = 5
1.4 × 108 ± 1.1 × 107, n = 5
3.5 × 107 ± N/A, n = 2
6.4 × 108 ± 8.5 × 107, n = 6
1.4 × 108 ± 2.0 × 107, n = 6
3.8 × 107 ± N/A, n = 2
1.7 × 109 ± 2.3 × 108, n = 6
4.5 × 108 ± 7.5 × 107, n = 6
5.1 × 107 ± 6 × 106, n = 6
1.7 × 109 ± 3.2 × 108, n = 32
2.7 × 108 ± 1.0 × 108, n = 6
6.1 × 107 ± 1.3 × 107, n = 27
The observation of rhoptry fusion is supported by previous work in Toxoplasma gondii in which inspection of individual TEM sections led to the speculation that rhoptries may fuse with each before or during release (Nichols et al., 1983), and some early micrographs by Aikawa et al. (1978) show the tips of Plasmodium rhoptries sharing a common duct. However other work supports a sequential release for the multiple Toxoplasma rhoptries (Paredes-Santos et al., 2012), and full rhoptry fusion has not previously been captured. While it is unclear what purpose the process of fusion serves during invasion we hypothesize that the likely change in ratio of surface area to volume brought about at the start of rhoptry fusion, and the instability caused by mixing of membranes, may provide a physical stimulus for the opening of the rhoptries at the merozoite apex and be a key step in the fusion/inversion of the nascent PVM with erythrocyte membranes that is required towards establishment of the PV. It is noteworthy that blockage of true invasion still demonstrates release of rhoptry contents into the target erythrocyte, which suggests that activation of fusion may underpin release even in the absence of invasion being successful (Riglar et al., 2011).
The detailed interrogation of the cellular events of P. falciparum merozoite invasion in four dimensions and at a nanometer scale provides several profound insights, previously unobserved or lacking replicate support, into this key step in parasite infection. These include demonstration of rhoptry fusion as an integral event during invasion, specific discontinuities in the apical half of the IMC, and maintenance of heterochromatin architecture within the nucleus. Our exploration of different fixation methods also demonstrates the importance of choosing appropriate techniques to answer distinct questions about the protein, lipid or morphological delineation of cellular events underlying invasion. With advances in super resolution imaging of the malaria parasite (Riglar et al., 2011; Yeoman et al., 2011) and development of methods for correlative fluorescence-electron microscopy (Nixon et al., 2009; Kukulski et al., 2011) linkage of these cellular insights with parasite molecules is clearly now a firm priority towards understanding the mechanisms underlying invasion and, ultimately, developing therapeutic strategies to stop it.
Parasite culture and maintenance
The culture of P. falciparum parasites using donated blood from the Australian Red Cross Society has been approved by The Walter and Eliza Hall Institute Human Ethics Committee (HEC 86/17). P. falciparum parasites (from a D10 parental strain; Boyle et al., 2010) were maintained using standard culturing procedures in human O + erythrocytes at 4% haematocrit with 0.5% w/v Albumax II (Life Technologies). Cultures were maintained in synchrony using 5% Sorbitol treatment or via treatment with 30 IU (approximately 230 μg ml−1) heparin (Pfizer) (Boyle et al., 2010) and cultured through to schizogony for merozoite invasion preparation. Merozoites were filtered through a 1.2 μm, 32 mm syringe filter (Sartorius Stedim Biotech) as described (Boyle et al., 2010; Riglar et al., 2011) for free merozoites, or incubated with human erythrocytes for 2 min (invasion) or 10 min (post-invasion) as described (Riglar et al., 2011).
Electron microscopy fixation methods
Reduced osmium fixation (ROTO)
This method is a combination of several published methods (Seligman et al., 1966; Willingham and Rutherford, 1984). Briefly, cells were fixed in 0.1 M sodium cacodylate pH 7.4 containing 5 mM calcium chloride and 2.5% glutaraldehyde for 4 h on ice followed by rinsing in 0.175 M sodium cacodylate. The cells were subsequently post fixed in potassium ferricyanide reduced osmium tetroxide for 1 h on ice then rinse in double distilled water, treated with 1% thiocarbohydrazide for 20 min at room temperature, rinse and further osmicated with 2% non-reduced osmium tetroxide for 30 min at room temperature, further rinsed and en-bloc stained with 1% uranyl acetate followed by dehydration and embedding in epoxy resin.
Glutaraldehyde/osmium fixation (GO)
The samples were fixed in freshly prepared solution of 2.5% glutaraldehyde in 0.1 M phosphate buffer (PB) (pH 7.4) for 1 h on ice. They were washed three times in PB and fixed in 1% osmium tetroxide (ProSciTech, Australia) prepared in 0.1 M PB for 1 h. Following extensive rinsing in distilled water samples were dehydrated with graded series of ethanol and embedded in LR Gold Resin (ProSciTech). Following polymerization by benzoyl peroxide (SPI-Chem, USA) ultrathin sections were cut on a Leica Ultracut R ultramicrotome (Wetzlar). The sections were post-stained with 2% uranyl acetate and 5% lead citrate and examined at 120 kV on a Philips CM120 BioTWIN Transmission Electron Microscope (Advanced Microscopy facility, School of Botany, the University of Melbourne).
Glutaraldehyde fixation (GF)
The samples were fixed in 1% glutaraldehyde (ProSciTech, Australia) on ice for 1 h. Samples were washed three times in water following dehydration in ethanol and embedding in LR Gold Resin (ProSciTech, Australia). After polymerization by benzoyl peroxide (SPI-Chem, USA) samples were sectioned on a Leica Ultracut R ultramicrotome (Wetzlar) and then prepared for imaging with 2% aqueous uranyl-acetate followed by 5% triple lead and observed at 120 kV on a Philips CM120 BioTWIN Transmission Electron Microscope (Advanced Microscopy facility, School of Botany, the University of Melbourne).
Free and invading merozoite samples were loaded into freezer hats (Proscitech), with well depths of 100 μm. The hats were frozen using a high-pressure freezer and quickly transferred into liquid nitrogen where the two halves of freezer hats were separated. Cells were freeze substituted with 1% uranyl acetate at −90°C for 24 h using a Leica AFS automatic freeze substitution machine. Freeze substitution was continued with acetone and samples were infiltrated using graded series in Lowicryl HM20 resin in acetone. Finally the resin was polymerized under UV light for 48 hours at −45°C and additional 48 hours at room temperature. The samples were sectioned, double contrasted with uranyl acetate and lead citrate and imaged with a Philips CM120 BioTWIN transmission electron microscope at 120 kV (Advanced Microscopy facility, School of Botany, the University of Melbourne).
Cryo-electron tomography (CET)
Rupturing schizonts (strain CS2) were placed onto holey carbon-coated quantifoil grids and rapidly plunge-frozen in liquid ethane and stored in liquid nitrogen. Imaging was carried out essentially as previously described (Cyrklaff et al., 2007; Kudryashev et al., 2010). Tilt series were collected using the FEI batch tomography tool on an FEI Titan Krios equipped with a GIF and US1000 post-GIF CCD with a pixel size at specimen level 1.05 nm with a FEI Polara microscope (operating at 300 kV) equipped with a field emission gun. The total dose used for each tomogram was kept below 20 k electrons/nm2, aiming at an angular coverage of 120 degrees. Acquired tilt series were aligned by gold fiducials cross correlation and tomogram reconstruction calculated by weighted-back projection using the Etomo program from an IMOD software package (Kremer et al., 1996). Tomograms were filtered with non-linear anisotropic diffusion (Mastronarde, 1997; Frangakis and Hegerl, 2001). Generation of the three-dimensional model was done using the 3dmod program of the IMOD package.
Cryo-X-ray tomography (CX)
Merozoites were purified as described above, then cryo-preserved and imaged as described previously (Hanssen et al., 2011; 2012).
Datasets were acquired as described previously (Abu Bakar et al., 2010; Hanssen et al., 2010) on a Tecnai F30 (FEI, Eindhoven, NL) running at 300 kV at the Advanced Microscopy Facility (Bio21 Institute, the University of Melbourne). Briefly, dual tilt tomograms from 200 nm serial section were acquired between −70° and +70° every two degrees for each axis, then reconstructed, segmented and measured using the IMOD package (Kremer et al., 1996; Mastronarde, 1997). Volumes were calculated after taking into consideration shrinkage to the beam exposure observed in our samples (200 nm sections gave ∼ 120–140 nm tomograms) and previously described shrinkage (Luther et al., 1988) of 33% in z and 10% in x and y dimensions.
The authors thank Carolyn Larabell, and Mark A. Le Gros (University of California, San Francisco) and Christian Knoechel (Lawrence Berkeley National Laboratory) for assistance and advice with X-ray tomography. X-ray tomography was conducted at the National Center for X-ray Tomography, supported by grants from the National Center for Research Resources (5P41RR019664-08) and the National Institute of General Medical Sciences (8 P41 GM103445-08) from the National Institutes of Health. Human erythrocytes were kindly provided by the Red Cross Blood Bank (Melbourne). This work was made possible through Victorian State Government Operational Infrastructure Support and Australian Government NHMRC IRIISS. Funding for the research came from the National Health and Medical Research Council of Australia (NHMRC Project Grants 637340 and APP1047085 JB & SAR), The Australian Research Council (Discovery Project Grant DP120103161), and Human Frontier Science Organisation (HFSP YI Program Grant RGY0071/2011 J.B. and F.F.), the Australian Synchrotron and the Australian Academy of Science, the US Department of Energy, Office of Biological and Environmental Research (DE-AC02-05CH11231), the National Center for Research Resources of the National Institutes of Health (RR019664). Use of the Advanced Light Source was supported by the US Department of Energy, Office of Science. D.T.R. is supported by a Pratt Foundation PhD scholarship through the University of Melbourne; M.R. was supported by the Australian Academy of Science and a grant from the OzeMalaR Network. L.L. was supported by a Postdoctoral fellowship from the University of Heidelberg cluster of excellence CellNetworks. M.K. acknowledges support of the CINA grant from SystemX.ch and from Henning Stahlberg (University of Basel). J.B. and S.A.R. are supported by Australian Research Council (ARC) Future Fellowships (FT100100112 and FT0990350). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The authors have declared that no conflict of interest exists.