Morphogenesis of mimivirus and its viral factories: an atomic force microscopy study of infected cells

Abstract

Amoebas infected with mimivirus were disrupted at sequential stages of virus production and were visualized by atomic force microscopy. The development of virus factories proceeded over 3 to 4 h postinfection and resulted from the coalescence of 0.5- to 2-μm vesicles, possibly bearing nucleic acid, derived from either the nuclear membrane or the closely associated rough endoplasmic reticulum. Virus factories actively producing virus capsids on their surfaces were imaged, and this allowed the morphogenesis of the capsids to be delineated. The first feature to appear on a virus factory surface when a new capsid is born is the center of a stargate, which is a pentameric protein oligomer. As the arms of the stargate grow from the pentamer, a rough disk the diameter of a capsid thickens around it. This marks the initial emergence of a protein-coated membrane vesicle. The capsid self-assembles on the vesicle. Hillocks capped by different pentameric proteins spontaneously appear on the emerging vesicle at positions that are ultimately occupied by 5-fold icosahedral vertices. A lattice of coat protein nucleates at each of the 5-fold vertices, but not at the stargate, and then spreads outward from the vertices over the surface, merging seamlessly to complete the icosahedral capsid. Filling with DNA and associated proteins occurs by the transfer of nucleic acid from the interior of the virus factory into the nearly completed capsids. The portal, through which the DNA enters, is sealed by a plug of protein having a diameter of about 40 nm. A layer of integument protein that anchors the surface fibers is acquired by the passage of capsids through a membrane enriched in the protein. The coating of surface fibers is similarly acquired when the integument protein-coated capsids pass through a second membrane that has a forest of surface fibers embedded on one side.

Fig 1

Mimivirus structural review. (a) Intact mimivirus about 750 nm in diameter with a characteristic corona of surface fibers. Its stargate is just beginning to open at the right. (b) Icosahedral capsid of mimivirus, which has been exposed by treatment of the virus with lysozyme and bromelain to remove superior layers. The stargate is at the top. (c) Higher-magnification image of the stargate showing some details of its multilayer construction. The arms of the stargate split along their lengths and then fall away as the stargate opens. The icosahedral protein lattice making up the capsid is apparent. (d and e) Capsid coated in turn with a layer of membrane-associated integument protein (d) that is responsible for anchoring the surface fibers (e). With integument protein added, but without surface fibers, the capsid is about 450 nm in diameter. The surface fibers are 125 to 140 nm in length and are tipped with a globular protein head. (f) AFM image capturing the opening of the stargate of a capsid to release the internal membrane sac, derived from a membrane near the surface of a viral factory, containing the genomic DNA and its associated proteins. Scan areas are 1.5 μm by 1.5 μm (a), 500 nm by 500 nm (b), 250 nm by 250 nm (c), 500 nm by 500 nm (d), 1 μm by 1 μm (e), and 625 nm by 625 nm (f).

Fig 2

Transport vesicles are produced in great abundance at the borders of the nucleus at about 2 h postinfection (p.i.), and they ultimately flood the host cell cytoplasm by 3 to 4 h p.i. The transport vesicles appear smooth but irregular in shape, and they have dimensions ranging from about 1/2 μm to 2 1/2 μm. (a) The transport vesicles contain material that, when dispersed inside, gives them a uniform appearance. The internal material, however, frequently assumes a condensed state that manifests itself as a prominent bulge or protruding mass. (b and c) These bulges, some of which are marked by arrows, are seen most clearly as the intense white areas emerging from the vesicles. (d to f) AFM images of infected cells showing how the cytoplasm becomes increasingly congested with the transport vesicles. Coalescence of transport vesicles and initiation of virus factory formation are seen on the right in panel f. Scan areas are 5 μm by 5 μm (a and b), 2 μm by 2 μm (c), 30 μm by 30 μm (d), 40 μm by 40 μm (e), and 35 μm by 35 μm (f).

Fig 3

AFM images of an infected amoeba at about 2 h p.i. The prominent yolk-like bodies are cell nuclei (Nu). The image in panel b is a higher-magnification scan of the cell nucleus shown in panel a. Apparent in these images is that the transport vesicles (TV) originate at or near the borders of the nucleus, where they are densely clustered. They then stream outward from the nuclear borders to fill the cell. Many of the transport vesicles contain condensed material, while in others, it remains dispersed. A cell with a double nucleus (D-Nu) is shown in panel h. Scan areas are 30 μm by 30 μm (a), 15 μm by 15 μm (b), 30 μm by 30 μm (c), 30 μm by 30 μm (d), 20 μm by 20 μm (e), 20 μm by 20 μm (f), 20 μm by 20 μm (g), and 30 μm by 30 μm (h).

Fig 4

The transport vesicles begin coalescing once their density has reached some limit, and the clusters and larger aggregates take on the appearance of foam. Note the disparity of sizes and shapes of the transport vesicles within the clusters. The large mass seen in the AFM image in panel c contains many hundreds of transport vesicles that have condensed in preparation for virus factory formation. The interior of the central mass is seen at a higher magnification in panel d, and transport vesicles are shown at the bottom right edge in panel e. Note that as the aggregate evolves, many of the transport vesicles have rounded up and exhibit a late-stage globular form. An example where coalescence has advanced is shown in panel f. The transport vesicles have lost their structural individuality and have amalgamated. The body shown in panel f is a nascent virus factory. Scan areas are 10 μm by 10 μm (a), 5 μm by 5 μm (b), 20 μm by 20 μm (c), 10 μm by 10 μm (d and e), and 5 μm by 5 μm (f).

Fig 5

AFM images illustrating that even after virus factories have formed and are actively producing virus (V), transport vesicles (TV) continue to be generated and are presumably still joining productive virus factories (VF). Thus, virus factories and transport vesicles coexist in the same infected cell. In panel d, an active virus factory, seen at a higher magnification in panel e, has been ejected from an infected cell after disruption and lies immediately adjacent to a nucleus (Nu) at whose borders transport vesicles continue to appear. The bright spheres scattered in panels a to e are mature particles or capsids filled with DNA. At the end of the infection cycle, when cells burst as a consequence of virus accumulation, transport vesicles are scarcely present. Panel f shows a small area on the surface of a young virus factory. Seen simultaneously in this AFM image are a portion of the virus factory surface, transport vesicles merging into the virus factory, and assembling capsids (C) already emerging from the surface. Scan areas are 10 μm by 10 μm (a), 30 μm by 30 μm (b and c), 50 μm by 50 μm (d), 15 μm by 15 μm (e), and 5 μm by 5 μm (f).

Fig 6

AFM images showing examples of virus factories arranged more or less according to their stages of development, ranging from rather young virus factories just beginning to produce capsids (a and b) to highly productive virus factories (c to f) and late-stage virus factories that are breaking up and dispersing throughout the cell (g and h). Note that in panel e, virions containing DNA, evidenced by their maintenance of shape, are present along with a large number of capsids that are either incomplete or complete but unfilled with DNA. The unfilled particles are recognizable by their collapse upon drying. Scan areas are 10 μm by 10 μm (a), 5 μm by 5 μm (b), 8 μm by 8 μm (c), 10 μm by 10 μm (d), 5 μm by 5 μm (e), 10 μm by 10 μm (f), 15 μm by 15 μm (g), and 25 μm by 25 μm (h).

Fig 7

Variety of AFM images of areas on the surfaces of virus factories showing the range of activities and stages of development of virus particles. (a) Area where some capsids are about to begin forming but is otherwise free of incomplete particles. Apparent in this image is the rough, granular nature of the surface, which is essentially a continuous blanket of protein molecules resembling gravel at the nanoscale. The variegated pattern of light and dark (representing elevation) areas portends particle genesis. (b to h) Surface cluttered with virus capsids at all stages of development. These range from relatively sparse distributions of developing particles (b) to extremely dense accumulations of particles (h) that completely obscure the virus factory to which they remain attached. Panel g shows a higher-magnification image of several collapsed particles in fairly early stages of development, as indicated by their incomplete stargates, the lack of ordered surface lattice, and the absence of internal DNA. Note that in virtually all of the images, the vast majority of particles have collapsed and therefore have not yet been filled with DNA. This suggests that capsid formation is rapid, while filling with DNA lags substantially. Many of the particles in panel f, however, are still attached to the virus factory but have evidently been filled with DNA, as they have not collapsed. Scan areas are 1 μm by 1 μm (a), 10 μm by 10 μm (b), 5 μm by 5 μm (c), 3 μm by 3 μm (d), 2 μm by 2 μm (e), 2 μm by 2 μm (f), 1 μm by 1 μm (g), and 10 μm by 10 μm (h).

Fig 8

Mimivirus capsids are developed on spherical vesicles that bud out from the virus factory surfaces. The vesicles, which ultimately become both the icosahedral protein capsid and the internal membrane sac containing the viral DNA, consist, at minimum, of a lipid membrane thoroughly coated and likely embedded with structural and auxiliary proteins, including the coat protein or its precursors. The bubble-like vesicles are easily recognized because most are decorated with a stargate that appears at the initiation of capsid morphogenesis. At the contrast levels used to produce these images, the vesicles appear almost transparent, although they are virtually coated with proteins. In panel f, a high contrast level was applied, and this makes evident the rough, disordered protein surface of the vesicles that will subsequently compose itself into the fully formed icosahedral lattice. It is evident from these images that, initially, the proteins on the vesicle surfaces exhibit no geometrical order. Most of the vesicles shown in these images were still attached to fragments of dispersed virus factories. Scan areas are 3.5 μm by 3.5 μm (a), 2 μm by 2 μm (b), 1 μm by 1 μm (c), 1 μm by 1 μm (d), 3 μm by 3 μm (e), and 2 μm by 2 μm (f).

Fig 9

AFM images illustrating early stages of capsid morphogenesis. In virtually all cases, the first event is the appearance of a large pentameric protein assembly, presumably composed of five identical subunits, on the virus factory surface. In panels a to c, the appearance of the stargate center and the initiation of the five arms are shown. The arms do not necessarily all develop at the same rate, implying that their growth is not coordinated. A thickening of the protein layer in a disk shape is just perceptible, but in panel d, it is clearly evident. This provides the outline of an emergent vesicle. Note also that in panel d, only the center of the stargate is visible, but the disk of thickened protein is already prominent. The same is true of emerging vesicles/capsids in panels e and f, suggesting that only the central pentamers of the stargate may be necessary to signal capsid morphogenesis. Scan areas are 1 μm by 1 μm (a), 500 nm by 500 nm (b and c), 2 μm by 2 μm (d and e), and 1 μm by 1 μm (f).

Fig 10

AFM images representing stages in the morphogenesis of the mimivirus capsid. (a to d) As events progress, the outer edge of the thickened disk takes on the appearance of a ridge. By the time the arms of the stargate have achieved almost full length, the disk has become an emerging vesicle/capsid. (e to h) Following or during completion of the stargate, prominent hillocks appear at locations occupied by 5-fold vertices in the completed icosahedral capsid. In panel f, the nuclei of some secondary, stargate-distal sets of 5-fold vertices are indicated by arrows. At the centers of the hillocks, pentameric proteins serve as nuclei for the development of an icosahedral lattice that eventually covers the surface. Scan areas are 1 μm by 1 μm (a and b), 500 nm by 500 nm (c to e), 1 μm by 1 μm (f and g), and 500 nm by 500 nm (h).

Fig 11

When the stargate arms have reached or are approaching their terminal lengths, hillocks capped by a pentameric protein at their peaks, marked by an arrow and a pentagon in panel a, appear spontaneously and precisely at the positions on the vesicle that will exhibit 5-fold vertices on the completed icosahedral capsid. This pentameric protein is not the same as that at the center of the stargate. It is significantly smaller and eventually integrates into the icosahedral capsid protein lattice without prominence. Beginning with this pentameric protein, which serves as a nucleus, the icosahedral lattice begins to appear, and this spreads outward from the hillocks in an approximately isometric manner. All 5-fold vertices do not appear simultaneously. Only the five nearest to the ends of the stargate arms first appear. After some development of the lattice from these vertices, the next set of five secondary vertices, some marked with arrows, emerges from the surface as similar hillocks. From all of the vertices, the respective lattice islands expand, encounter the edges of other lattice islands, and merge in a seamless manner to form the completed icosahedral capsid. As shown in panel f, lattice islands extend toward the stargate and fill in the gaps between the stargate arms, also joining to their edges without any visible discontinuity. Scan areas are 150 nm by 150 nm (a), 500 nm by 500 nm (b), 1 μm by 1 μm (c), 500 nm by 500 nm (d), 700 nm by 700 nm (e), and 500 nm by 500 nm (f).

Fig 12

(a) The lattice islands from two vertices have merged (arrow) without any discontinuity, and the unified lattice front is now proceeding to fill the gap between two stargate arms. This has transpired while the vesicle/capsid has only partially emerged from the virus factory surface. (b) The surface of an emerging, apparently aberrant capsid, lacking any stargate at all, exhibits a large area covered by protein lattice (arrow), while the surface of the remainder of the particle is comprised of still disorganized protein. (c) An anomaly is presented. Fivefold vertices and the initial propagation of the icosahedral lattice (arrow) appeared on the virus factory surface in the absence of any stargate or thickened disk. The images in panels b and c suggest that the 5-fold vertices and icosahedral coat protein lattice can arise independently of a stargate and that their initiation is not necessarily dependent upon the development of a stargate. (d) AFM image of a developing 5-fold hillock made in fluid. The detail shows the pentagonal protein arrangement at the hillock center. Scanning in fluid yields a more realistic, undistorted image of the icosahedral protein lattice and emphasizes the three-dimensional nature of the thick honeycomb mesh. Scan areas are 500 nm by 500 nm for all images.

Fig 13

When the external membrane/protein layers of a virus factory are removed, the DNA matrix in the interior becomes exposed. (a and b) AFM images of the contents of an active virus factory showing the distribution of DNA (a) and the same sample scanned at a higher magnification (b). (c) High-magnification image of the nucleic acid showing the cables of condensed DNA heavily associated with proteins. The scan areas are 15 μm by 15 μm (a), 10 μm by 10 μm (b), and 3 μm by 3 μm (c).

Fig 14

(a and b) Sides or facets of two independent capsids that are opposite the stargate and exhibit the 40-nm-diameter patch or plug that fills the DNA entry portal. No arms of the stargate are seen, confirming that the view is indeed of the side opposite the stargate. The icosahedral lattice is weakly visible in the neighborhood of the closure, validating the resolution of the images. (c and d) Another particle with a similar patch/plug seen at a lower magnification (c) and the closure at a higher magnification (d). Scan areas are 500 nm by 500 nm (a and b), 1 μm by 1 μm (c), and 500 nm by 500 nm (d).

Fig 15

Mimivirus DNA (arrows) is present as highly condensed nucleoprotein masses about 350 nm in diameter within the inner membrane sacs (Mb) of virions. (a) Nucleoprotein-filled capsid that has spontaneously burst and ejected its almost spherical bolus of DNA. (b to d) Capsids that have expelled the internal membrane sacs containing their complements of DNA. The membranous sacs spread over the AFM substrate, leaving a bolus of DNA near their centers. In panel c, the DNA has begun to separate into strands. (e) DNA that is still compacted within a partial capsid, showing its highly convoluted, condensed form. (f) An isolated bolus of genomic DNA beginning to unravel, showing it to be composed of tightly wound, thick cables of the nucleic acid. Scan areas are 1.5 μm by 1.5 μm (a), 2 μm by 2 μm (b and c), 800 nm by 800 nm (d), 500 nm by 500 nm (e), and 3 μm by 3 μm (f).

Fig 16

After the release of a capsid from the virus factory, filled or unfilled with DNA, it quickly passes through a lipid membrane thickly embedded with the integument protein responsible for anchoring the surface fibers. A mass of capsids is seen enmeshed in a membrane network (a), and individual particles or groups are seen passing through the membrane as they acquire a layer of membrane and integument protein (b to h). Arrows indicate trailing integument protein-laden membrane (IPM). The stargate, however, slips through the membrane without being thus coated. A capsid coated with integument protein presents a rather nondescript surface with no obvious geometrical order. In panels g and h, individual integument protein molecules can be seen on the particle surfaces. Note that the stargates of the particles in panels g and h have begun opening, revealing the longitudinal splitting of the stargate arms. Scan areas are 5 μm by 5 μm (a), 1.5 μm by 1.5 μm (b), 1 μm by 1 μm (c), 800 nm by 800 nm (d), 1 μm by 1 μm (e and f), 500 nm by 500 nm (g), and 550 nm by 550 nm (h).

Fig 17

The last layer of structure to be added to a virion is the forest of surface fibers. The distal ends of the fibers are tipped by a globular protein head, with the opposite end fixed to the particle through an integument protein anchor. (a to e) Integument protein-coated particles encountering membranes, indicated by arrows and labeled SFM, whose outer surfaces are decorated with forests of surface fibers tipped with head groups. The surface fibers' tails are embedded in the lipid membrane. (f to h) Groups of virus that have acquired fibers by passing through the surface fiber-decorated membranes. Scan areas are 800 nm by 800 nm (a), 260 nm by 260 nm (b), 500 nm by 500 nm (c), 750 nm by 750 nm (d), 1 μm by 1 μm (e), 3.6 μm by 3.6 μm (f), 2.3 μm by 2.3 μm (g), and 1 μm by 1 μm (h).

Fig 18

(a) The tangle of a long fiber (arrow) near a cluster of surface fibers (SF) (arrow). (b and c) The long fiber shown at a higher magnification. Panel c shows that the fiber exhibits a 7-nm periodicity along its length, as did fibers previously reported to be associated with mimivirus (4). (d) Higher-magnification image of the surface fiber cluster indicated by the arrow in panel a. Scan areas are 1 μm by 1 μm (a), 500 nm by 500 nm (b), 250 nm by 250 nm (c), and 300 nm by 300 nm (d).

Fig 19

Schematic diagrams summarizing the steps most generally observed in capsid formation up to its release from a virus factory (a) and outlining the final steps in mimivirus morphogenesis, the acquisition of an integument layer and coating of surface fibers (b). (a) Capsid morphogenesis begins on the virus factory surface (i) by the formation of a stargate central pentamer (ii), initial growth of stargate arms, thickening of a capsid disk, and initial emergence of a vesicle, shown more fully developed (iii). Pentamer-capped hillocks appear at eventual 5-fold vertices (iv), and these hillocks serve as nuclei for the development of the coat protein, icosahedral net (v). Upon completion of the protein capsid (vi), the DNA is poised to enter the empty capsid through a stargate-distal portal, which it does (vii) as the nucleic acid condenses inside. The entry portal is sealed (viii) to yield a completed capsid. (b) The superior structural layers are acquired when a capsid (i) passes sequentially through a membrane embedded with integument protein (ii) and then through a membrane containing on its distal side a coating of surface fibers (iii and iv). This results in the completed virion (v).