Malaria, which is caused by Plasmodium spp. parasites, is a major infectious disease that affects the developing world, and there are global efforts that are devoted to generating strategies for its control1. Malaria parasites are transmitted through the bite of an infected female Anopheles spp. mosquito. Within the vertebrate host, the parasite undergoes two consecutive stages of schizogonic asexual replication (Box 1). The first stage is clinically silent and takes place in the liver (Box 1; see the figure, steps 2–8). The second stage occurs following parasite release from the liver into the bloodstream (Box 1; see the figure, steps 9–14) and is responsible for the clinical manifestation of the disease.

The biology of Plasmodium spp. parasites, as well as their interactions with human and mosquito hosts, has been a focus of research for more than a century, and microscopy has historically greatly accelerated progress in this field. In the past decade, revolutionary advances in light and electron microscopy have provided invaluable information about parasite biology and host–pathogen interactions (Table 1).

Table 1 Advanced imaging techniques used to study Plasmodium spp. parasites

The imaging of parasite architecture in most stages of the life cycle was mainly advanced by various electron microscopy methods, including cryoelectron tomography and cryo-X-ray tomography, as well as by super-resolution fluorescence imaging2,3,4,5,6,7. Techniques such as spinning-disc microscopy, among others, have been pivotal for the elucidation of sporozoite motility8,9,10. Finally, major breakthroughs in time-lapse microscopy enabled the imaging of Plasmodium spp. parasites during the complete blood11 and liver12 stages of their life cycles in vitro. In this Review, we focus on imaging methods that have advanced our understanding of parasite cell biology and host–pathogen interactions in the mammalian host, and have led to important paradigm shifts in our knowledge of malaria during the past decade.

Transmission and the 'skin phase'

During a blood meal, a Plasmodium-infected mosquito injects sporozoites into a human host (Box 1; see the figure, step 1; Supplementary information S1 (movie)). In turn, human-to-vector transmission occurs when, during blood feeding, a mosquito ingests gametocytes that are circulating in the blood of an infected host (Box 1; see the figure, step 15).

Sporozoite architecture. Mature sporozoites (Fig. 1a) are slender cells that form in oocysts at the mosquito midgut wall. They are approximately 10 μm long, just less than 1 μm thick and are slightly curved13. 3D cryoelectron tomography has increased our understanding of the relevance of sporozoite architecture, in particular with regard to sporozoite migration. This technique enables the examination of native and hydrated structural features, as well as the study of different functional states of molecules within the sample (Table 1). Four key observations in terms of sporozoite architecture are discussed below.

Figure 1: Imaging of Plasmodium spp. sporozoites.
figure 1

a | Architecture of a mature sporozoite, showing organelles and specific architectural features that have been identified by cryoelectron tomography. These organelles include the posterior polar ring (PPR), the mitochondrion, the endoplasmic reticulum (ER), the inner membrane complex (IMC), the nuclear pore complexes (NPCs) and the nucleus, the Golgi apparatus, the apicoplast, the dense granules, the microtubules, the rhoptries, the micronemes and the apical polar ring (APR). b | Schematic (left panel) showing the inclined APR (red), the subpellicular microtubules (green; yellow arrowhead indicates the 'single' microtubule, which defines the dorsal side of the chiral sporozoite), the micronemes (cyan) and the rhoptries (magenta). A volume-rendered representation of an electron tomogram (right panel) clearly demonstrates the tilted APR and the 'single' microtubule (yellow arrowheads). White bars indicate distances between individual microtubules next to the APR and at 800 nm distance from the APR. Scale bar represents 100 nm. c | Electron tomography of a mature sporozoite, showing the projection of a 10-μm-thick section through the tomogram (left panel). The grey-scale intensity plot (bottom middle panel) shows the electron density peaks that correspond to (from the right) the plasma membrane, the IMC and a microtubule. The slope in the plot between the microtubule and IMC peaks corresponds to the thickened subpellicular network. A model of the microtubule (right panel) shows the luminal density (red), representing an unknown protein. Scale bar represents 100 nm. d | Filopodium-like extensions of the plasma membrane at the apical end of the sporozoite. The volume-rendered representation of a sporozoite shows three extensions (yellow arrowhead) of the plasma membrane (blue) with electron-dense material inside (red). The IMC (yellow) and the microtubules (green) are also shown. Scale bar represents 100 nm. e | Traction forces revealed by traction-force microscopy. The sporozoite (green) moves on a polyacrylamide gel loaded with fluorescent beads (red) to measure the traction force (top left panel). Plotting the data over time (three top right panels; red corresponds to high force, and blue corresponds to low force) illustrates the traction force that is exerted as the sporozoite (white outline) moves. The yellow arrow indicates how the contact between the sporozoite and substrate 'zips' backwards during motility, as illustrated in the schematic (bottom panel). The sporozoite depicted is 10 μm long. Part a is adapted with permission from Ref. 138, Wiley. The left panel of part b is adapted from Ref. 16. The right panel of part b is reproduced with permission from Ref. 3, Wiley. The left and middle panels of part c are adapted from Ref. 138, Wiley. The right panel of part c is reproduced from © 2007 Cyrklaff, M. et al. J. Exp. Med. 204: 1281–1287. doi: 10.1084/jem.20062405. Part d is reproduced from Ref. 16. Images in the top panel of part e courtesy of F.F. and J. Kratzer, University Hospital Heidelberg, Germany. The bottom panel of part e is reproduced from Ref. 10, Elsevier.

PowerPoint slide

Many studies that used classic electron microscopy methods showed the overall architecture of sporozoites and similar parasite forms. For example, it was found that the polar ring at the apical end could function as a microtubule-organizing centre14. The so-called subpellicular microtubules are arranged as a 'cage' that encases secretory organelles (Fig. 1b). In sporozoites, unlike in most other parasite forms, most microtubules cover two-thirds of the circumference of the cage, whereas a single microtubule is found in the remaining one-third13. Cryoelectron tomography revealed that the apical polar ring is tilted away from the single microtubule, seemingly being pushed towards the external substrate3. Similar to the invasive forms of other apicomplexan parasites, sporozoites release the contents of their secretory vesicles through the apical polar ring15. Therefore, the tilt of this polar ring might direct the secretion of proteins that are important for adhesion directly onto the substrate.

The second insight into sporozoite architecture came from findings in Plasmodium berghei that revealed the existence of a thickened subpellicular network (SPN) in sporozoites from the mosquito salivary gland but not in sporozoites that were directly isolated from the oocysts in the mosquito midgut3,16 (Fig. 1c). In salivary gland sporozoites, the SPN protein inner membrane complex 1 (IMC1) was shown to have relocated from the cytoplasm, where it is found in midgut sporozoites, to the SPN17, which suggests that the SPN builds up during sporozoite maturation. Salivary gland sporozoites, but not midgut sporozoites, are crescent shaped and capable of gliding motility. These findings suggest that the crescent shape results from the assembly of the SPN. The crescent shape of mature sporozoites might have evolved to lower the energy costs of migrating through tissues or as a result of preferred binding to blood vessels, which have the same curvature as sporozoites18,19.

Two further structural findings that were revealed by 3D cryoelectron tomography of P. berghei sporozoites are the enigmatic presence of filopodium-like extensions of the plasma membrane at the front of some sporozoites16 (Fig. 1d), and the discovery of an unknown protein inside the lumen of the subpellicular microtubules (Fig. 1c) that possibly has a role in stabilizing these microtubules2.

Although 3D cryoelectron tomography has indeed provided key details about sporozoite architecture, it relies on static imaging, which makes it impossible to directly assess the functional relevance of these architectural characteristics. However, the combination of static and dynamic microscopy methods both in vivo and in vitro has become a powerful tool for elucidating the link between structure and function.

Sporozoite migration and the skin phase. Wide-field, confocal or spinning-disk intravital microscopy (IVM)8,19,20,21,22,23,24 were used to visualize mosquito-to-vertebrate transmission of P. berghei parasites in vivo (Box 1; see the figure, step 1; Box 2; see the figure, part b). These studies (reviewed in detail in Refs 25,26) revealed that sporozoites are deposited in the dermis and immediately start migrating at speeds of 1–2 μm s−1 (Ref. 20). Sporozoite movement follows random paths in the ear but more linear paths in the dermis of the mouse tail22. Interestingly, P. berghei sporozoites were shown to migrate in structured microenvironments in vitro, in which they show similar linear or random paths, depending on the environment. This suggests that their motility in vivo is dictated by characteristics of their local environment, such as the presence of collagen fibres22. Sporozoites also transmigrate through cells aided by the secretion of pore-forming proteins23,24. They can enter either blood capillaries or lymph vessels, or they can remain in the skin8,27. A long-standing misconception in the malaria field was that sporozoites are injected directly into blood vessels by the mosquito proboscis. IVM that was carried out on P. berghei-infected mice, together with experiments that removed the bite site and interrupted mosquito feeding, proved that this was not the case. Instead, sporozoites are injected into the connective tissue of the skin28. These findings suggest that the skin phase of Plasmodium spp. infection is the first possible target for prophylaxis29,30.

To actively migrate, sporozoites need to attach to a substrate. Total internal reflection fluorescence microscopy (TIRFM) is a powerful method to visualize single-molecule fluorescence at the immediate interface of substrate and cell. Similarly, reflection interference contrast microscopy (RICM) enables the measurement of inter-surface distances with nanometric precision and millisecond time resolution to study adhesion dynamics. Both techniques showed that P. berghei sporozoites attach to a flat substrate in a stepwise manner10,31. Using RICM, it was also shown that when the sporozoite is motile, the distinct adhesion sites detach and re-attach rapidly in a manner that is dependent on actin dynamics and the surface protein thrombospondin-related anonymous protein (TRAP)10. This suggests that rapid adhesion turnover is crucial for the sporozoite to obtain high speed (Fig. 1e).

The forces that are exerted by the sporozoites onto the substrate to enable movement are small in comparison to those exerted by much larger mammalian cells, such as fibroblasts10. Nevertheless, using a sensitive traction force microscopy method, a study was able to distinguish zones of different force transduction during P. berghei sporozoite motility10. Force generation was also shown to depend on the presence of dynamic actin filaments. To investigate the role of surface proteins, another investigation used nano-patterned substrates that enable sporozoites to attach only to 10 nm-wide gold particles that are spread at large distances from each other32. Remarkably, sporozoites were still moving on substrates for which only about 100 molecules could have attached the parasite to the substrate. One of these molecules, TRAP, is essential for sporozoite gliding motility33: sporozoites and other motile apicomplexan stages move forwards when TRAP family adhesins bind to the substrate. Sporozoites need to be activated to undergo gliding. This activation has been investigated using a fluorescence resonance energy transfer (FRET)-based calcium sensor34 and a fluorescently tagged actin-binding protein35, showing that an increased calcium concentration leads to the secretion of TRAP followed by the organization of actin filaments and, hence, motility. The speed of parasite forward movement was presumed to equal the speed of the retrograde movement of the adhesins along the parasite surface36. However, subsequent studies have used laser traps to identify a distinct function for different members of the TRAP family of membrane proteins37,38,39. To this end, either P. berghei sporozoites were trapped and directly probed37 or small beads were trapped and brought into contact with the sporozoites38. These biophysical studies showed that force production for motility is not linearly coupled to the retrograde flow of the adhesins but requires complex macromolecular interactions. These assays, coupled to high-speed image acquisition, revealed different types of interaction between the TRAP-family adhesins and the actin cytoskeleton, and suggested that they act together to form macromolecular assemblies that are important for applying force. Importantly, these findings imply that sporozoites have evolved to apply force mainly for adhesion and not for the retrograde flow of the adhesins. The detailed interactions between adhesins and the force-producing motor are not known to date.

IVM also revealed that some P. berghei sporozoites injected into mice can successfully develop in cells of the dermis, epidermis and hair follicles27,40. Most strikingly, skin merozoites and the detachment of merosomes27 were observed, possibly a relic of the shared ancestry of these mammalian malaria parasites with their avian counterparts, which develop within the skin (although in different cell types)41. Interestingly, a small proportion of parasites were able to persist for weeks in association with hair follicles27, which are considered an immunoprivileged site. Additional investigation is required to determine whether this constitutes a new niche or source for recurrent malaria infections. Moreover, IVM also showed that 10–20% of transmitted sporozoites accumulate in the draining lymph nodes, where they can induce immune responses8,42. Whether or how these alter an ongoing infection remains to be shown, and further studies are required to investigate sporozoite-induced immune reactions in the lymph nodes.

The liver stages

Within hepatocytes, Plasmodium spp. parasites undergo a clinically silent phase of exponential replication (Box 1; see the figure, steps 2–4; Supplementary information S2 (movie)). As well as merozoite numbers, the duration of the liver stage of infection differs substantially across host and parasite species: this stage typically lasts 2–3 days in rodents and 7–10 days in humans, but some parasite species, such as Plasmodium vivax and Plasmodium ovale, are able to produce dormant stages that lead to relapses as long as several years after the primary infection.

Sporozoite entry into the liver. Sporozoites arrest in the liver probably owing to both a slower circulation speed in liver sinusoids (Box 1; see the figure, step 2) and molecular interactions of the parasitic protein circumsporozoite protein (CSP) with highly sulfated proteoglycans (HSPGs) that are presented by hepatocytes through the fenestrae of endothelial cells43,44. IVM of exposed liver lobes in combination with genetically modified fluorescent parasites revealed, for the first time, the dynamics of the aggressive cell traversal behaviour of the sporozoite in the living host (Box 1; see the figure, steps 3–4; Box 2; see the figure, part b; Supplementary information S3 (movie)). Different IVM modalities have been used, including spinning-disc microscopy45 and high-speed two-photon microscopy, to increase the penetration depth to about 500 μm. Improvements in microscope optical design and the camera technologies that are used for spinning-disc microscopy have enabled the visualization of dynamic events involving the parasite en route to, and within, the liver, in the range of milliseconds.

One of the main findings achieved through the use of IVM was that P. berghei sporozoites are capable of using several pathways to cross the liver sinusoidal barrier en route to the hepatocytes9,25 rather than translocating across the sinusoidal barrier exclusively through Kupffer cells (also known as the gateway model; Box 1; see the figure, step 3)46,47,48. Using fluorescent P. berghei sporozoites that lack a protein that is essential for traversal (sporozoite microneme protein essential for traversal 2 (SPECT2))49, in combination with control parasites and various reporters and dyes for endothelial and Kupffer cell integrity, spinning-disc confocal IVM revealed that sporozoites are able to cross the liver sinusoidal barrier by four different additional routes9: endothelial cell traversal following interactions with a Kupffer cell; endothelial cell traversal that is independent of interactions with Kupffer cells; an invasion–transcytosis variant of the gateway pathway, whereby sporozoites cross the barrier between a Kupffer cell and an adjacent endothelial cell; and crossing at a distance from Kupffer cells and without traversing endothelial cells. To explain this fourth form of crossing, two possibilities have been suggested: a paracellular pathway, whereby sporozoites cross between two endothelial cells, or a transcellular pathway, whereby sporozoites cross through endothelial cells but do not induce a loss of membrane integrity. Such a transcellular pathway is used by lymphocytes to cross the endothelium (reviewed in Ref. 50). These findings not only disproved the previously accepted paradigm that the gateway model was the only method of sporozoite entry to the liver, but also indicated that about one-quarter of all crossing events are independent of cell traversal.

Development of Plasmodium spp. parasites within the liver. As a result of proteolytic processing of surface-exposed CSP, the motile sporozoites switch from the transmigration to the invasion mode43. During the process of invasion, sporozoites induce the formation of a parasitophorous vacuole membrane (PVM; Box 1; see the figure, step 5). The PVM represents the interface between the parasite and its host hepatocyte throughout liver-stage development, and it enables the parasite to scavenge nutrients by both passive diffusion through membrane pores and active transport51,52,53. One of the most relevant findings achieved by a combination of time-lapse imaging and fluorescence recovery after photobleaching (FRAP) was the extremely dynamic nature of the PVM12,52. Fluorescence tagging of the P. berghei proteins IBIS1 (Ref. 52), UIS4 and EXP1 (Refs 12,52), which are transported to the PVM, enabled the characterization of the connectivity and spatiotemporal dynamics of the PVM and the host tubovesicular network (TVN), which is a microtubule-dependent membranous network. Remarkable findings included the observation of dynamic and motile PVM-derived and TVN-derived vesicles that spread throughout the host cell54,55,56,57 (Fig. 2a; Supplementary information S4 (movie)).

Figure 2: Imaging techniques reveal relevant biological events during liver-stage development.
figure 2

a | Schematic images (left panel) of the parasitophorous vacuole membrane (PVM) at early (12 hour) and late (36 hour) time points post-infection. Sporozoites traverse several cells (cell traversal; CT) before undergoing a switch to invade. On invasion, the parasite induces the formation of a PVM. Time-lapse imaging, photo-activation and super-resolution microscopy have shown the PVM and the host tubovesicular network (TVN) to be extremely dynamic, with tubules and vesicles extending large distances into the host cell, away from the parasite body. The parasite plasma membrane (PPM) and parasitophorous vacuole (PV) are indicated. The confocal image (right panel) shows the PVM of a Plasmodium berghei liver-stage parasite 36 hours post-infection; scale bar represents 50 μm. b | Schematic form of host cell microtubule-associated protein 1 light chain 3B (LC3B; purple) dynamics with respect to the P. berghei parasite throughout infection, including heavy accumulation at early time points and lateral localization of LC3B at late time points post-infection. Microscopy shows host cell LC3B accumulations around an invaded P. berghei sporozoite early post-infection (middle panel), and LC3B localization to the lateral side of the PVM (away from the P. berghei schizont; P) and to TVN vesicles later post-infection (right panel). Scale bars represent 10 μm. c | Schematic (left panel) and wide-field fluorescence microscopy images (middle and right panels) of the dynamics of host cell actin clouds around developing P. berghei liver-stage parasites. Asymmetrical and symmetrical actin clouds (AACs and SACs, respectively; actin is labelled red) are present around GFP-tagged P. berghei (green). Scale bars represent 10 μm. d | Schematic of merosome formation during parasite egress (left panel). Spinning-disc confocal intravital imaging of liver tissue infected with GFP-tagged P. berghei shows the parasite undergoing merosome budding and egress into the sinusoidal vessel (middle panel). Red fluorescent BSA was used to label the blood vessels. Scanning electron microscopy shows budding P. berghei merosomes (right panel). Scale bars represent 10 μm. Confocal microscopy image in part a courtesy of V.T.H. and C. Agop-Nersesian, University of Bern, Switzerland. Microscopy images in part b courtesy of V.T.H. (left and right images) and N. Eickel, University of Bern, Switzerland (left image). Microscopy images in part c adapted from Ref. 59. Part d is adapted from Sturm, A. et al. Manipulation of host hepatocytes by the malaria parasite for delivery into liver sinusoids. Science 313, 1287–1290 (2006). Reprinted with permission from AAAS.

PowerPoint slide

Using long-term time-lapse imaging of sporozoite-infected HepG2 and HeLa cells and IVM of sporozoite-infected mouse livers, it was shown that the P. berghei PVM is recognized by the host cell autophagic machinery12,58. The labelling of the PVM by host autophagy markers was confirmed using super-resolution stimulated emission depletion (STED) microscopy12. Compared with electron microscopy-based methods, STED microscopy has the advantage that infected cells can be easily identified and analysed by fluorescence, rather than requiring cell sorting followed by extensive fixing and embedding procedures. Long-term time-lapse microscopy (for up to 72 hours) showed, for the first time, that the autophagic marker microtubule-associated protein 1 light chain 3B (LC3B) targets the PVM early in infection (Fig. 2b; Supplementary information S2 (movie)), followed by the elimination of most parasites, thereby uncovering a previously unknown cytosolic immune response12 (Fig. 2b). By contrast, a surprising discovery was that during P. berghei schizogony (Box 1; see the figure, step 6), successful parasites seem to actively remove autophagy marker proteins from the PVM, potentially by membrane shedding12, and thus avoid continued attack by lysosomes (Fig. 2b, right panel). In contrast to selective autophagy, which can potentially eliminate the invading parasite, canonical autophagy is induced by starvation and can provide nutrients for essential cell functions. Indeed, quantitative confocal microscopy for measuring parasite sizes showed that the rapidly growing Plasmodium spp. parasite exploits canonical autophagy as an additional source of nutrients12,58. In addition to the success of IVM for imaging the P. berghei pre-erythrocytic developmental stages in terms of speed, time lapse and penetration depth, very recent work has obtained subcellular-resolution images of the parasite in vivo, enabling the visualization of previously unknown organelle structures in the developing parasite5.

Another important observation achieved by wide-field fluorescence time-lapse microscopy was the highly dynamic host cell actin reorganization (or actin cloud formation) that takes place around developing P. berghei parasites 10–16 hours post-infection59 (Fig. 2c; Supplementary information S5 (movie)). It has been proposed that gelsolin is necessary for severing actin filaments and, possibly, contributing to the dynamic actin turnover, which in turn might contribute to the mechanical confinement and/or elimination of the parasite. Thus, the observed reorganization of host actin has been suggested to be part of a broader defence strategy of the hepatocyte, such as autophagy.

Parasite egress from the liver. The final step of the pre-erythrocytic stage of Plasmodium spp. infection is egress from the liver hepatocytes and parasite release into the bloodstream (Box 1; see the figure, steps 7–9). This transition from the liver to the blood stage was the final part of the pre-erythrocytic stage to be elucidated, and the process that was uncovered was one of the most exciting microscopy-based revelations of the past decade in the field of Plasmodium spp. liver-stage research. For many years, it was accepted that infected hepatocytes rupture to release progeny merozoites into the bloodstream to infect red blood cells (RBCs). However, this assumption failed to explain how merozoites reach the bloodstream, as even following liver cell rupture, merozoites would still be extravascular. Unlike sporozoites, merozoites cannot use typical gliding motility to transmigrate endothelial cells. Long-term time-lapse live imaging of P. berghei-infected hepatocytes in vivo and in vitro revealed that after PVM rupture, released merozoites move freely in the host cell cytoplasm and this is followed by the initiation of host cell death45,60 (Box 1; see the figure, step 7). The delayed rupture of the host cell plasma membrane enables the formation of merozoite-filled membranous bodies that were hence named merosomes45 (Box 1; see the figure, step 8; Fig. 2d; Supplementary information S6 (movie)). To further characterize host cell death during merosome formation, fluorescence-based detection of phosphatidylserine residues was used45,60, because phosphatidylserine residues act as generic 'eat me' signals of apoptotic cells to attract phagocytic macrophages. In contrast to phosphatidylserine-positive apoptotic cells, the plasma membrane of merosomes remains phosphatidylserine-negative, which indicates that the observed cell death of infected hepatocytes is ordered but is not typical apoptosis. Thus, patrolling macrophages and neutrophils are unable to recognize the migrating phosphatidylserine-negative merosomes. Merosome budding from infected hepatocytes through the endothelial layer into the blood vessel has been demonstrated by versatile quantitative IVM techniques45 (Box 2; see the figure, part c). Importantly, merosome formation seems to be a conserved egress strategy of Plasmodium spp., as similar structures to those identified in P. berghei infections have been observed in Plasmodium yoelii infection47 and in Plasmodium falciparum-infected immunocompromised mice engrafted with human hepatocytes61.

The asexual blood stages

Following the migration of merosomes into the liver vasculature and transport to the lung capillaries47, merozoites are released into the circulation (Box 1; see the figure, step 9), which marks the beginning of the blood stage. Following invasion of an RBC by the merozoite, the intra-erythrocytic parasite life cycle is divided into ring, trophozoite and schizont stages, eventually leading to the release of 16–32 daughter merozoites, which invade new RBCs to begin the cycle again (Box 1; see the figure, step 13). Each cycle results in RBC rupture and the induction of periodic waves of fever in infected patients. Blood-stage parasites modify their host cell in a process that is dependent on the export of numerous parasite proteins into the RBC. Host cell modifications are necessary for the intracellular survival of the parasite. These changes include altered permeability of the RBC membrane to nutrients and the creation of host cell surface structures termed knobs, which facilitate the virulence factor display that is necessary for the sequestration of infected RBCs (reviewed in Ref. 62). One prominent host cell modification in the cytoplasm of infected RBCs is the formation of Maurer's clefts, which are parasite-induced membranous organelles that are involved in the sorting of parasite proteins in the host cell and are key for the delivery of virulence factors to the surface of the infected erythrocyte63 (Box 1; see the figure, steps 11–12). These virulence factors mediate parasite sequestration to the vascular endothelium of several organs of the host (Box 2; see the figure, parts d–f).

Imaging the erythrocytic cycle and host cell remodelling. Previously, phototoxicity and lysis of the acidic food vacuole of the parasites had made it impossible to view the entire P. falciparum blood stage at the level of individual parasites64. However, a recent investigation used confocal fluorescence and differential interference contrast (DIC) microscopy linked with photoconversion to study for the first time the development and concomitant host cell modifications in individual erythrocytic stages of the parasite (Fig. 3a; Supplementary information S7, S8 (movies))11. In this new study, 4D imaging revealed two unexpected findings regarding host cell remodelling, protein export and Maurer's clefts. A large body of research has provided crucial insights into the architecture, function and morphology of Maurer's clefts, as well as their role in protein trafficking. Previous studies proposed that parasite membrane proteins are exported to Maurer's clefts by insertion into the nascent clefts at the PVM, thereby linking the biogenesis of Maurer's clefts to protein export65,66,67,68,69 (recently reviewed in Ref. 70). The first unexpected finding revealed by 4D imaging was that Maurer's clefts do not constantly form from the PVM, as was previously believed, but rather that the total number of Maurer's clefts is already present shortly after parasite invasion and remains constant throughout the entire erythrocytic cycle until shortly before parasite egress11. These findings suggest that proteins that are destined for export are trafficked to already formed Maurer's clefts. Moreover, photoconversion combined with 4D imaging revealed the arrival of new cargo at the Maurer's clefts11. This implies that protein export must occur through soluble intermediates71, or possibly also through vesicles trafficking between the PVM and the Maurer's clefts72, rather than by insertion into the clefts at the PVM, as proposed by the nascent-cleft hypothesis. The second previously unknown phenomenon shown by 4D imaging was that segmentation of progeny merozoites following schizogony (Box 1; see the figure, step 12) coincides with the collapse of Maurer's clefts, which might indicate that the parasite-induced structures in the host cell are dismantled, perhaps to facilitate egress11 (Box 1; see the figure, step 13).

Figure 3: Imaging of Plasmodium spp. erythrocytic stages.
figure 3

a | Graphic representation of the four main blood stages of Plasmodium spp. development (upper panel), and 4D imaging of the developing Plasmodium falciparum parasite (blue) using confocal microscopy (lower panel), showing the complete erythrocytic stage of development in an individual red blood cell (RBC; red). Selected time points in differential interference contrast (DIC) microscopy are shown, from ring stages to schizont stages. The white arrow at 31 h 20 m shows a ring stage derived from reinvasion after rupture. Scale bar represents 2 μm. b | Maurer's clefts (MCs) at the RBC membrane. One study indicates that Maurer's clefts attach to the host cell membrane through stalk-like tethers, as illustrated in the schematic (top left). A volumetric reconstruction and rendering of the immuno-electron tomograms (top right) shows a tether-like structure (grey) connecting a Maurer's cleft (blue) to the RBC membrane (red). Regions that may represent RBC cytoskeleton extensions are indicated with cyan arrows. Thickening of the Maurer's cleft coat is observed in some regions and is indicated with a yellow arrow. Bulge in the RBC membrane is indicated by a red arrow. A second study found evidence that the attachment is instead mediated by host actin filaments, as illustrated (bottom left). A surface-rendered cryoelectron tomogram of a P. falciparum-infected erythrocyte (bottom right) shows a network of long branched actin filaments (yellow) connecting the Maurer's clefts (cyan) to the knobs (red) and the plasma membrane (dark blue); 20–200 nm vesicles (cyan) are also attached to these filaments. c | The three steps of invasion observed by time-lapse imaging, and details of the receptor–ligand interactions as defined by super-resolution microscopy7. During the first step of invasion, merozoite attachment induces perturbations in the RBC membrane. This step ends in the merozoite reorienting to bring its apical end in contact with the RBC membrane. In the second step, a tight junction is formed, mediated by erythrocyte receptors and parasite ligands. Penetration of the parasite into the RBC is mediated by an actomyosin motor. Following complete invasion, the third step involves resealing of the RBC membrane and RBC echinocytosis. d | High-speed DIC time-lapse microscopy enabled the identification of a series of ordered events that are involved in merozoite egress95. In the mature schizont, osmotic pressure increases before rupture of the parasitophorous vacuole membrane (PVM). Following PVM rupture, the merozoites are able to move freely in the RBC while the RBC membrane remains intact. An osmotic pore is then formed in the RBC membrane, during which an osmotic release of 1–2 merozoites occurs in a matter of milliseconds. After this, the RBC membrane curls and then 'buckles', dispersing all remaining merozoites in multiple directions up to 10 μm away, towards neighbouring RBCs. Microscopy images in part a are from Ref. 11, Nature Publishing Group. Top right image in part b is adapted with permission from Ref. 139, Wiley. Bottom right image in part b is adapted from Cyrklaff, M. et al. Hemoglobins S and C interfere with actin remodeling in Plasmodium falciparum-infected erythrocytes. Science 334, 1283–1286 (2011). Reprinted with permission from AAAS. Part c is adapted with permission from Ref. 82, National Academy of Sciences.

PowerPoint slide

In RBCs that contain ring-stage parasites (Box 1; see the figure, step 10; Fig. 3a), the Maurer's clefts show rapid movement in the cytoplasm11. Concomitant with the transition of the parasites to the trophozoite stage (Box 1; see the figure, step 11; Fig. 3a), the Maurer's clefts become anchored to the RBC cytoskeleton11. Cryoelectron tomography and fixed-cell electron tomography imaging techniques have provided two competing but not mutually exclusive ideas of how the Maurer's clefts attach to the RBC plasma membrane and/or cytoskeleton. One research group suggested that attachment is mediated by stalk-like structures called tethers73,74,75,76, whereas a different group proposed that this connection is mediated by host actin filaments4 (Fig. 3b). Both of the structures seen contained attached vesicles and could, in principle, act as cables for virulence factors that traffic from the Maurer's clefts to the host cell surface. Electron tomography of parasites in RBCs from patients with sickle cell disease and of a parasite line lacking a Maurer's cleft protein showed ablated surface display of virulence factors and a disrupted state of the Maurer's clefts–actin organization4,77. These images provided the first molecular insights into how parasite proteins are trafficked over the last hundred nanometres en route to the RBC plasma membrane.

Merozoite attachment to and invasion of host red blood cells. Merozoite invasion and egress (Box 1; see the figure, steps 9, 13; Fig. 3c,d) are widely studied topics in the field of malaria research, but real-time imaging of these events had been hindered by poor merozoite viability, their small size (1–3 μm), their susceptibility to photo-damage and the speed of these processes. Despite their small size, merozoites contain all of the organelles that are necessary to attach to and invade RBCs. Invasion-related organelles that are present in the merozoite include rhoptries, dense granules and micronemes15,78. Various imaging methods, including time-lapse DIC and fluorescence microscopy, optical tweezers, super-resolution microscopy, cryo-electron tomography and cryo-X-ray tomography, have led to a better understanding of the molecular mechanisms and kinetics that mediate merozoite attachment and invasion. Time-lapse microscopy studies revealed three distinct phases of P. falciparum invasion6 (Box 1; see the figure, step 9; Fig. 3c). In the first phase, merozoites make contact with the RBC, triggering waves of deformation in the RBC emanating from the contact site. This phase ends with the reorientation of the merozoite, bringing its apical pole, in which the secretory organelles are located, into contact with the erythrocyte surface. At this point, the merozoite is irreversibly attached to the erythrocyte surface and committed to invasion6. A combination of imaging techniques, including cryoelectron tomography, cryo-X-ray tomography and super-resolution structured illumination microscopy (SIM), have shown that irreversible attachment of the merozoite to the RBC is mediated by erythrocyte-binding ligands (EBLs) and reticulocyte protein-binding homologues (RHs) on the surface of the parasite7,74,79,80,81,82,83. It is noteworthy that cryo-X-ray tomography provided 3D nanoscale information about the localization of molecules within cells in their native state. This is the only technique that can provide information at this resolution about unstained, whole cells thicker than 1 μm and that can combine the power of light, electron and X-ray microscopes.

In the second phase of invasion, the merozoite enters the RBC in a process that is dependent on an actin–myosin (actomyosin) motor and the formation of a tight junction at the interface between the merozoite and erythrocyte plasma membranes7,84,85. SIM demonstrated an RBC receptor-independent mode of host cell adhesion through the parasite ligand apical membrane antigen 1 (AMA1)7. The merozoite secretes the rhoptry neck (RON) complex of proteins, and RON2 acts as an anchor in the erythrocyte membrane for RON complex assembly and AMA1 engagement7,85,86. AMA1 and RON proteins were found to colocalize at the tight junction during invasion. Importantly, inhibition of the AMA1–RON interaction prevents merozoite invasion87. Although these findings are supported by some genetic studies, other genetic data are contradictory, and it is possible that this model for AMA1 function is too simplistic88,89,90,91. In terms of the relevance of imaging for the elucidation of the aforementioned interactions, super-resolution microscopy provides various techniques for imaging other Plasmodium spp. stages, but thus far SIM is the only super-resolution microscopy technique to be regularly used for imaging the blood stages7. SIM is compatible with live cell imaging, given that it requires only low illumination intensities, can rapidly acquire large fields of view and can take advantage of most existing fluorophores92. However, compared with some other super-resolution techniques such as STED, stochastic optical reconstruction microscopy (STORM) or photoactivated localization microscopy (PALM), the resolution gain over conventional fluorescence microscopy is comparably low.

The third phase of invasion occurs 10–20 seconds after the merozoite fully enters the erythrocyte and is characterized by echinocytosis of the RBC, whereby the edges of the RBC curl and fine spicules are formed. It was hypothesized that this process is triggered by the resealing of the erythrocyte plasma membrane behind the merozoite6.

An important finding achieved through the use of optical tweezers was that the different steps in invasion are independent93. Placing dead merozoites close to RBCs using optical tweezers showed that merozoites which had lost the ability to invade 2–3 minutes post-egress nonetheless retain the ability to adhere to RBCs and still induce transient local membrane deformations in the RBC membrane. A contribution of the RBC membrane to merozoite invasion was also suggested using computational modelling94, an approach that is likely to be used more frequently to complement quantitative imaging in future studies.

Egress from host red blood cells. The final step of the erythrocytic cycle is the release of invasive merozoites on rupture of the infected erythrocyte (Box 1; see the figure, step 13; Fig. 3d). Parasite release from the host RBC requires the opening of the PVM and RBC membranes. DIC imaging showed that before egress, the parasites move freely in the RBC following rupture of the PVM, while the RBC membrane is still intact95. Time-lapse microscopy using calcium and fluorescence imaging further demonstrated the involvement of parasite proteases and perforin-like proteins in increased osomolarity96 and revealed that Ca2+-dependent permeabilization and membranolytic activities are necessary for merozoite release. Although it was previously believed that merozoite egress was an explosive event, recent studies using high-speed DIC video microscopy and epifluorescence imaging showed that egress is in fact an ordered process, beginning with the opening and stabilization of an osmotic pore in the RBC membrane95 (Fig. 3d). High-speed imaging indicated that after osmotic swelling of the infected RBC, lasting 400–800 milliseconds95, and rupture of the PVM, three steps lead to egress: an initial phase termed osmotic release, whereby 1 or 2 merozoites are released in the first 100–200 milliseconds, followed by two steps of 'inelastic release', whereby the RBC membrane changes its curvature and buckles, liberating the remaining merozoites in a dispersing manner. This form of egress and dispersion is believed to be necessary to optimize merozoite invasion of neighbouring RBCs up to 10 μm away from the RBC of origin.

The sexual blood stages

Asexual replication of blood-stage parasites results not only in the multiplication of parasite numbers in the blood, but also in the generation of gametocytes (Box 1; see the figure, step 14) for transmission to the definitive host (that is, the mosquito vector; Box 1; see the figure, step 15). Interestingly, only very early (stage I) and mature (stage V) P. falciparum gametocytes are found in human blood circulation (Box 1; see the figure, step 14), whereas immature forms (that is, stages II–IV) are absent from the circulation and instead sequestered in the bone marrow and spleen, as confirmed by ultrastructural analysis97 (Fig. 4a).

Figure 4: Imaging of the architecture, deformability and sublocalization of Plasmodium spp. sexual stages within the host bone marrow.
figure 4

a | Autopsy studies showed that immature gametocytes are found mainly in the extravascular space of the bone marrow, where they are preferentially localized in the vicinity of erythroblastic islands (which are nursing macrophages in which erythroblasts and other erythroid precursors localize) and develop within erythroid precursors. Stages I – IV remain within the extravascular space, whereas stage V gametocytes undergo a deformability switch and are able to re-enter the peripheral circulation, where they can be taken up by mosquitoes. b | Model geometry from a 3D analysis of gametocyte development. Morphological differentiation of the parasite causes substantial deformation of the host red blood cell (RBC), transforming it from a discoidal to an elongated ellipse as the gametocyte matures from stage II to stage V. c | Imaging studies have found unexpected localization of F-actin in Plasmodium falciparum gametocytes. F-Actin (red) accumulates at the gametocyte tips and at the parasite periphery in early-stage gametocytes. The transition to mature stage V gametocytes is accompanied by the coordinated disassembly of the microtubule and F-actin cytoskeletons, leading to a more diffuse labelling pattern107. d | Extracellular vesicles (EVs; such as microvesicles and exosomes) that are derived from infected erythrocytes mediate cellular communication within the parasite population and are thought to stimulate the production of transmission-stage parasites. Part b is adapted with permission from Ref. 100, Wiley.

PowerPoint slide

Whereas the sequestered immature gametocyte stages are rigid, it is thought that a switch in RBC and gametocyte deformability occurs in the final stage of gametocyte maturation, rendering mature gametocytes (and their associated RBCs) more flexible; the presence of mature gametocytes in the circulation is thought to be a direct consequence of this switch. 3D live imaging in capillary assays in vitro and 3D finite-element whole-cell modelling were used to study the circulatory characteristics of P. falciparum gametocytes at different stages of maturation98,99,100 and the role of cellular deformability during circulation in the blood100 (Fig. 4b). The simulations that were generated from these studies led to the hypothesis that the morphological and biophysical changes in the gametocyte-infected RBC are linked to tissue distribution in vivo, as only RBCs that contain either very early (stage I), and thus small, or mature (stage V) gametocytes could easily traverse the splenic interendothelial slits under physiological pressure; RBCs infected with stage II–IV gametocytes would be unable to squeeze through the slits and would be trapped in the spleen. This ensures the survival of the mature gametocytes so that they can be taken up by the mosquito during a blood meal (Box 1; see the figure, step 15). The spleen is generally regarded as a central organ in malaria pathology (reviewed Refs 101,102,103,104) owing to its capacity to eliminate RBCs with reduced deformability, including infected RBCs. The function of the spleen is thought to have driven Plasmodium spp. to evolve mechanisms to avoid splenic passage and sequestration. To date, the presence of gametocytes in the spleen and the effect this might have on transmission and overall parasite biology are being investigated.

Cryo-X-ray microscopy led to the first detailed observation of changes in the subpellicular membrane complex (a membrane structure that is underpinned by a layer of microtubules) as gametocytes mature. The extension of this subpellicular complex drives parasite and host cell elongation and is responsible for the characteristic crescent shape of P. falciparum gametocytes. 3D SIM imaging uncovered new details about the organization and function of the actin cytoskeleton in gametocytes98,105,106,107. Actin polymerization has been previously described as necessary for the motility of apicomplexan parasites, so the discovery of polymerized actin in non-motile gametocytes was very surprising. The structures that contain F-actin in gametocytes constitute a previously unrecognized actin-based cytoskeleton that seems to be assembled very early in gametocyte development. This cytoskeleton provides a template for the positioning of microtubules and stabilizes the tubulin cytoskeleton107. 3D SIM demonstrated the accumulation of such structures at the poles of the gametocyte, radiation into the parasite body and close association with the tubulin cytoskeleton (Fig. 4c).


Over the past decade, technical advances in imaging techniques have enabled extraordinary studies of Plasmodium spp. biology and host–pathogen interactions. It is safe to assume that imaging methods will remain prominent in malaria research in the future, for two reasons. First, the new techniques that have been introduced to date, such as super-resolution microscopy, cryoelectron tomography and IVM, have not yet been fully exploited for malaria research. Second, other recently developed imaging techniques, such as those involving mesoscopic imaging, will surely be applied in the field in the years to come.

One key challenge in malaria research is to dissect the molecular details of the different life cycle stages of the parasite. Resolution down to single molecules has been achieved by cryoelectron tomography, but the thickness of the specimens that can be analysed is currently restricted to less than 1 μm. Hence, cryoelectron tomography has only been used to explore sporozoite architecture and the role of the host cell cytoskeleton at the edge of infected RBCs. Further improvements in this technique108,109 will enable the investigation of thicker parasite forms, such as oocysts, later liver-stage parasites and perhaps schizont-infected RBCs. Correlative light and electron tomography110 provides great opportunities for dynamic imaging, for instance to study the changes in cytoskeleton architecture during gliding motility of sporozoites. Depending on the freezing procedure that is required, it should theoretically be possible to image the same sporozoite in different migration phases.

Other underexplored topics in malaria research are signalling pathways in the parasite and the host cell. These cannot be approached solely by biochemical means, as an understanding of signalling at a single-cell level, in particular during liver-stage development, is necessary. A biochemical approach would be limited because only a small percentage of hepatocytes are infected, and not every parasite develops successfully. As signalling events are often very fast, imaging methods such as spinning-disc microscopy, light sheet microscopy111,112,113 and holographic high-speed imaging114 — in combination with a series of FRET-based signalling reporter constructs that are now available for mammalian cells115 — would be needed to further increase our understanding of these events in infected cells. To this end, optogenetic tools that enable the triggering of signalling processes by light will also prove beneficial116.

Phototoxicity is a major problem in live imaging approaches owing to the energy that is required to excite the fluorescent dyes. Exposure times can be limited through the use of ultrafast imaging techniques in combination with highly sensitive cameras that capture maximum numbers of photons. A different approach that might revolutionize live cell microscopy is lattice light sheet microscopy117. This method combines the advantages of super-resolution SIM microscopy with markedly reduced levels of photo-damage, achieved by the use of a multiple-beam approach applied to light sheet microscopy. This non-invasive high-speed 4D imaging enables single molecules to be followed in living cells. In combination with the nascent technology of small-molecule labelling (which enables the tagging of proteins that are refractory to fusion with large fluorescent proteins), these new techniques should facilitate the dissection of most cellular processes118,119.

Another technique with interesting potential in terms of high resolution and speed is atomic-force microscopy (AFM). AFM has greatly facilitated the study of the infected-erythrocyte surface and in particular has enabled a detailed analysis of the 30–40 nm knobs on the surface of P. falciparum-infected RBCs120,121,122,123. Other applications of AFM have included investigations of RBC cytoadherence and parasite sequestration. These included functionalization of cantilevers to attach host receptors to which erythrocyte membrane protein 1 (PfEMP1) binds, and measure the force of interaction that is relevant to sequestration of the parasite124,125. Surface topography and receptor–ligand interactions have been extremely valuable applications, but, over the past decade, progress in AFM has also enabled real-time high-speed imaging of living cells. Although high-speed AFM is challenging because of the time of collision between the cantilever and the sample, various techniques have been implemented, one of which relies on the attachment of extremely long and thin tips by amorphous carbon to the cantilever126,127. Because of its versatility, AFM is regarded as a 'lab on a tip', capable of simultaneously providing information about the structure and the biological, chemical and physical parameters of whole living cells or individual molecules128. AFM has also been used to identify microvesicles and exosomes generated by infected RBCs129. Microvesicles act as a means of communication among Plasmodium spp. parasites. In 2014, two independent studies showed that following internalization by infected RBCs, these microvesicles stimulate conversion and commitment to the sexual parasite cycle129,130 (Fig. 4d). These findings have important implications for cell–cell communication, as the process is reminiscent of quorum sensing displayed by other organisms and might be relevant for research developing methods to block malarial transmission129,130. The microvesicle field is currently in its infancy but has already attracted major attention owing to its potential application in vaccine development131 and its relevance to disease transmission129,130, and imaging methods to study the generation and the fate of the microvesicles are likely to provide further insights in this emerging field.

Regarding the transmission stages of Plasmodium spp., high-resolution imaging of gametocytes led to the identification of structures similar to those relevant for motility and invasion in sporozoites and merozoites107. This is an interesting aspect, as the preferential localization of gametocytes seems to be the extravascular space of the bone marrow, but it is not known how they cross the endothelium to reach this location. The elucidation of gametocyte migration in the bone marrow is therefore a next step to address in the field.

Although IVM and animal imaging advances are not addressed exhaustively in this Review, an enormous amount of progress has been achieved in this field, some of which has been translated to malaria research using rodent models. In addition, remarkable headway has been made in the past few years in developing humanized mouse models for in vivo studies of human diseases, offering an unprecedented balance between human species-specific tissues and cells, and a small vertebrate model132,133,134,135. Of relevance to malaria research, progression through the complete P. falciparum life cycle is now possible in humanized mice133,134, and the formation of hypnozoites and invasion of reticulocytes by P. vivax have also been achieved using these models135. Together with the suite of imaging techniques and discoveries reviewed here in mice, extrapolation to human infections is now feasible in these models.

Finally, as imaging techniques progress and quantitative imaging becomes more important for the malaria field, advances in image processing and analysis software, as well as improvements in data storage, have become equally necessary. Undoubtedly, in the decades to come, further advances in imaging tools will play a key part in improving our understanding of the malaria parasite in terms of its basic biology, pathology and host–pathogen interactions.