We present here a new model for the transition from the liver to the blood phase of the malaria life cycle (): large merosomes of various sizes bud from infected hepatocytes, enter the hepatic circulation, exit the liver intact, subdivide into smaller more uniform sizes, but otherwise withstand bloodstream shear forces during passage through the right ventricle, and accumulate in the lungs where the merosomes disintegrate and release merozoites to initiate the erythrocytic phase of the malaria cycle. While EEF of avian and reptilian malaria parasites develop in the reticulo-endothelial or hematopoietic systems [43
], a major evolutionary change occurred with the mammalian malaria parasites, whose EEF mature in hepatocytes. Perhaps the nutritionally rich and immunologically privileged hepatic environment offers advantages, but it also presents a problem for merozoites released from EEFs into hepatic sinusoids: unless they invade an erythrocyte very quickly they face a gauntlet of highly phagocytic Kupffer cells. The location of most EEFs in the periportal area of the liver lobule [46
] means they must travel almost the full length of the sinusoid and pass by a large complement of Kupffer cells before escaping into relative safety outside the liver. As proposed previously by us and others [14
], our premise is that evolution produced a countermeasure to this threat: release of merozoites within large packets that are initially hidden from the host's innate immune system by envelopment with a hepatocyte-derived membrane. Here we show that merosomes are delivered to the pulmonary microcirculation where they are released. We propose that release of merozoites into the lung microvasculature rather than into larger blood vessels is advantageous, because the low macrophage density and the reduced blood velocity with reduced shear forces will enhance the ability of merozoites to invade erythrocytes.
Model of Merosome Dissemination and Merozoite Liberation
Merosome disintegration in the lungs appears to be the predominant mechanism of merozoite liberation into the bloodstream for the following reasons: (1) In confirmation of previous reports on the asynchronous nature of EEF maturation [5
], we observed P. yoelii
merosome formation in the liver from 46 h to 56 h after sporozoite infection. Assuming a 10-h window of merosome release, roughly 3 ml total blood volume in a 40 g mouse, and a 100% rate of sporozoite infection and EEF development, 2.5 million sporozoites would generate 4,167 maturing EEFs per minute, corresponding to 1.4 merosome-releasing EEFs per μl blood. (2) Assuming that extrahepatic merosomes contain on average 150 merozoites, the roughly 29 merosomes we found per μl venous liver blood should have contained 4,350 merozoites. Since P. yoelii
EEFs contain 4,200–29,000 merozoites (Table S1
), up to 74% of the total number of merozoites released by 1.4 EEFs per min and μl would have been enclosed in merosomes. (3) A large number of merosomes was arrested in alveolar capillaries suggesting that many merosomes withstand the shear forces inside the central cardiovascular system. Together, these data indicate that a major proportion of the merosome population arrives intact in the lungs and then gradually disintegrates, thus liberating merozoites into the microvasculature. Pulmonary merosomes were detectable in the lungs at least up to 58 h after infection, i.e., beyond the period of release from the liver (46–56 h), suggesting that they remained intact for at least many minutes. Similar to hepatic merosomes, which appeared to be infectious and did not stain with annexin V, YO-PRO-1, or PI, pulmonary merozoites were ultrastructurally well preserved, TUNEL-negative, and did not incorporate PI. Together, these data suggest that merosomal merozoites remain viable until their release into the pulmonary microvasculature. Based on the above assumptions, we propose that merozoite liberation in the lungs represents an integral part of the Plasmodium
Further support for our premise was found in the following observations and suggestions derived from them. The notion that merozoites shuttled out of the liver within merosomes that are protected from phagocytosis by Kupffer cells [8
] was confirmed by demonstrating that murine Kupffer cells do not phagocytoze PyGFP merosomes in vitro (unpublished data), in agreement with the finding that P. berghei
merosomes are not ingested by a murine macrophage cell line in vitro [15
]. Trager and Jensen's finding that P. falciparum
merozoite invasion is enhanced by lack of flow and dense erythrocyte packing [47
] supports our hypothesis that merozoites released within capillary beds have a better chance to invade erythrocytes than those released into larger vessels. We can imagine that capillary occlusion by arrested merosomes could be helpful by causing local stagnation of the pulmonary blood flow. We can also speculate that merosome arrest in lung septal capillaries allows Plasmodium
to exploit the unique microenvironment of the blood-air barrier. Virtually nothing is known about the biology of the first-generation (hepatic) merozoites, but perhaps transient residence in the lungs provides these parasites with time and a suitable microenvironment to gain infectivity for erythrocytes. The well-oxygenated milieu of the terminal airways and the anastomozed nature of the pulmonary microvasculature [49
] likely allow local occlusion of septal capillaries by merosomes without causing the necrotic tissue damage associated with infarction of microvessels in other organs.
Many aspects of the process of merosome formation and release we describe are in agreement with earlier work, but others are not. For example, we found that similar to P. berghei
–infected HepG2 cells, which detach in toto from the culture vessel after merozoite differentiation is complete [15
], merosomes exiting P. yoelii
–infected mouse livers contain viable merozoites and initially do not expose PS on their surface. This confirms earlier predictions [14
] that merozoites are safely shuttled out of the liver disguised as merosomes. The presence of intact mitochondria in mature EEFs indicates that Plasmodium
liver stages are able to manipulate hepatocytes in a way that useful organelles (such as mitochondria as a source of energy) are preserved, even after merosome budding. Our interpretation, namely that Plasmodium
controls certain host cell functions to the last minute, differs from the P. berghei
HepG2 cell model, in which the parasites induce death and detachment of their host cells followed by merosome budding [15
]. Further, the cell membrane of P. yoelii
–infected hepatocytes remains in close apposition to that of neighboring parenchymal and endothelial cells until the very end of EEF differentiation, i.e., up to the onset of merosome budding, as reported [5
]. As merosomes are produced, the host cell gradually decreases in size and loses contact with neighboring cells as reported [15
]. We observed that after releasing merosomes over several hours, the exhausted host cell eventually disintegrates. Some free merozoites still escaped and entered the sinusoidal lumen, thus being exposed to attack by Kupffer cells. In contrast, others proposed that the remaining host cell remnant is rapidly expelled in toto from the tissue with the resulting void immediately filled by neighboring cells [15
]. We find that the necrotic remnant attracts neutrophils and mononuclear phagocytes, which eventually produce a small granuloma. Such granulomata are a frequent observation in P. yoelii
– and particularly in P. berghei
–infected mouse livers [5
]. Rather than the void created by expulsion of an EEF being filled quickly, our in vivo observations suggest that hours, if not days, are required for phagocytic removal of parasite and host cell debris with subsequent repair of the structural damage before normal tissue architecture is restored.
Although we found merosome formation to be the predominant mode of merozoite release from the liver, we observed a less frequent but still common alternative: EEFs undergoing what we interpret as decay. This alternative process of EEF ghost formation was rapid and typically complete within minutes to an hour. In contrast to merozoite release by merosome formation, ghost-forming EEFs did not detach from the surrounding tissue. EEF decay was accompanied by leakage of GFP into the surrounding tissue suggesting damage to the host cell membrane. It occurred in immature EEFs (recognizable by a homogeneous green fluorescent cytoplasm) and also in mature EEFs (containing fully formed merozoites) without merosome formation regardless of maturity. Sometimes it was found as early as 42 h after sporozoite infection, hours before merozoite differentiation begins. The end result of this alternative process was the formation of large faintly fluorescent EEF ghosts containing some cellular debris and a few dead merozoites. We interpret this rapid conversion of EEFs to ghosts as abortive liver stage development.
Merozoite content of EEFs has historically been difficult to estimate due to the large number of parasites and their high packing density. Based on measurements of the size and merozoite content of small merosomes combined with size measurements of EEFs and an appropriate mathematical algorithm [42
], we were able to calculate the number of merozoites in an EEF (Table S2
). Under intravital imaging conditions, mature P. yoelii
EEFs measured 40–75 μm and the calculated space effectively occupied by a merozoite is a sphere of 2.2 μm diameter. Using this effective size, we calculated that individual P. yoelii
sporozoites produce roughly 4,200–29,000 merozoites per EEF. This number is in general agreement with older estimates of EEF merozoite content [5
] (Table S1
). An exception is P. falciparum
, which produces considerably larger numbers of hepatic merozoites, most likely because of the small size of the parasites. As far as we know, our analysis of the number of merozoites produced in hepatocytes is the first such analysis based on actual merozoite counts and host cell measurements. Precision is limited by variations in measurements, but basing calculations on direct in vivo measurements enhances accuracy.
Earlier studies conducted by us and others had suggested that merosome budding may precede completion of merozoite differentiation [14
]. One factor that helped lead to this interpretation is that GFP expressed in the parasite stroma can obscure the parasites in mature EEFs. We now show that prior to merosome formation, the signal of the stromal GFP fluorescence equaled that of the merozoite cytoplasm, thus preventing clear definition of parasites enmeshed in the stroma. At the onset of merosome budding, the stromal GFP emission signal decreased abruptly thus revealing the presence of the already formed fluorescent parasites (A–E and Video 5). Two factors contribute to this reduction in fluorescence of material surrounding the parasites: dilution and loss of cytosolic GFP. Dilution of GFP results from PV disassembly and mixing of fluorescent parasite stroma with non-fluorescent host cytoplasm. Loss of GFP is caused by leakage of the fluorochrome into the environment. In agreement with reports that the hepatocyte membrane becomes permeable at late stages of infection with P. berghei
], we found that merosome-forming EEFs are typically surrounded by a halo of green fluorescence. Optimization of the imaging conditions allowed us to visualize the parasites inside mature EEFs and revealed that merosomes always contain mature merozoites. Thus, merozoites maturation precedes merosome formation.
Depending on the approach used for measurement, the reported diameters of hepatic and pulmonary capillaries vary greatly. For example, when measured in perfusion-fixed liver tissue, the sinusoidal diameter ranged from 4–6 μm to 9–12 μm [58
]. A crucially important factor is the pressure applied during perfusion fixation, because the sinusoidal diameter is known to vary with changes in blood pressure [61
]. To determine the sinusoidal diameter under normal blood pressure conditions, we used live Tie2-GFP mice [32
], whose fluorescent endothelia clearly delineate the boundaries of the sinusoidal lumen [31
]. In agreement with earlier in vivo microscopic studies, which reported a diameter of 6 μm for portal sinusoids and 7 μm for central sinusoids [58
], we found by intravital imaging that liver sinusoids measure 6.7 ± 1.9 μm in diameter. Similar differences between fixed and live specimens were reported for the size of alveolar capillaries. While vascular casts of the lung suggested that alveolar capillaries measure 6.69 ± 1.39 μm in diameter [63
], intravital measurements determined a functional diameter of only 1–4 μm [64
]. Regardless which liver sinusoid and lung capillary measurements are relied upon and regardless of the drastic reduction in merosome size after leaving the liver, merosomes still exceed the size of the lumen of the microvasculature of both liver and lung. Since even the largest merosomes were eventually transported out of the liver, the much smaller extrahepatic merosomes would be expected to be malleable enough to be able to pass though the pulmonary capillary bed. Therefore it is somewhat surprising that the lungs effectively clear the blood of all merosomes, so virtually none were detectable in arterial blood harvested from the left ventricle, in the capillary beds of spleen, brain and kidney, or in tail vein blood. The fact that the velocity in pulmonary capillaries is somewhat higher than hepatic sinusoids [66
] makes this more unexpected. Consequently, the possibility of a receptor-mediated mechanism for pulmonary merosome arrest cannot be excluded.