SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Nucleo cytoplasmic large DNA viruses (NCLDVs) are a group of double-stranded DNA viruses that replicate their DNA partly or entirely in the cytoplasm in association with viral factories (VFs). They share about 50 genes suggesting that they are derived from a common ancestor. Using transmission electron microscopy (TEM) and electron tomography (ET) we showed that the NCLDV vaccinia virus (VACV) acquires its membrane from open membrane intermediates, derived from the ER. These open membranes contribute to the formation of a single open membrane of the immature virion, shaped into a sphere by the assembly of the viral scaffold protein on its convex side. We now compare VACV with the NCLDV Mimivirus by TEM and ET and show that the latter also acquires its membrane from open membrane intermediates that accumulate at the periphery of the cytoplasmic VF. In analogy to VACV this membrane is shaped by the assembly of a layer on the convexside of its membrane, likely representing the Mimivirus capsid protein. By quantitative ET we show for both viruses that the open membrane intermediates of assembly adopt an ‘open-eight’ conformation with a characteristic diameter of 90 nm for Mimi- and 50 nm for VACV. We discuss these results with respect to the common ancestry of NCLDVs and propose a hypothesis on the possible origin of this unusual membrane biogenesis.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Being obligatory intracellular parasites of the cell, enveloped viruses acquire their membrane from the host. Membrane acquisition occurs by budding at intracellular membranes or the plasma membrane, a process resembling the formation of cellular vesicles. Viruses may also use cisternal wrapping, acquiring two membranes at once, resembling cellular autophagy. Both budding and wrapping mimic cellular membrane dynamics, controlled by fission and fusion reactions, ensuring the maintenance of closed vesicular structures (Falanga et al., 2009).

The membrane acquisition of the large DNA virus Vaccinia virus (VACV) may be quite different. By transmission electron microscopy (TEM) the first evidence of VACV membrane biogenesis is the formation of half-moon shaped membranes, crescents, at the site of cytoplasmic DNA-replication. These crescents grow to form spherical particles, the immature virion, with a diameter of about 350 nm that after DNA-uptake, mature in to an infectious intracellular mature virus (reviewed in Condit et al., 2006; Roberts and Smith, 2008). Since the crescents seemingly expose open ends in the cytoplasm and are not continuous with cellular membranes, Dales and colleagues proposed that the VACV membrane is made ‘de novo’, creating a single open membrane in the cytoplasm (Dales and Mosbach, 1968). This model does not conform cell biological dogmas, leading others to propose that the crescent membrane is a collapsed cisterna, composed of two, rather than one, membranes (reviewed in Sodeik and Krijnse Locker, 2002). We recently re-addressed the question of one versus two membranes and showed by cryo-EM of vitreous sections that the crescent is composed of a single membrane studded on its convex side with the D13L viral scaffold protein (Heuser, 2005; Chlanda et al., 2009). By electron tomography (ET) the crescent was found to be continuous with small, uncoated, membranes and we proposed that these contributed to the formation of the spherical immature virion. Contrary to cell biological dogmas predicting that cellular membranes are closed compartments, quantitative ET showed that the majority of these connected membranes were open and that the crescent is an open sheet composed of a single open membrane. EM immunolabelling and ET also revealed membranes, which are likely the precursors of the crescents. They are small membrane structures abundantly labelled with antibodies to two major viral membrane proteins, genes products of A14L and A17L (Chlanda et al., 2009). ET showed that they accumulate close to areas that are arranged into a honeycomb-like structure of the viral scaffold protein (the gene product of D13L) and most of these small membranes appeared to be open (Chlanda et al., 2009; Krijnse Locker et al., 2013). They can be labelled with antibodies to ER proteins (Chlanda et al., 2009) suggesting that they are derived from this compartment. The latter is consistent with the fact that the crescent-membrane is also labelled to some extend with ER marker proteins (Salmons et al., 1997; Risco et al., 2002) and with lipid analyses of isolated and purified virions (Sodeik et al., 1993; Krijnse Locker et al., 2013).

The unusual membrane assembly raised many questions; how are the membranes opened, how are they targeted to the crescent, how do they ‘fuse’ with the growing membrane sphere, how are the open ends stabilized in the cytoplasm, what prevents their spontaneous resealing? We have started to address this by analysing recombinants of VACV in which individual viral proteins are conditionally expressed and found no role for the D13L scaffold protein (Chlanda et al., 2009), the major membrane protein A17L (Chlanda et al., 2011) and the F10L kinase (P. Chlanda and J. Krijnse Locker, unpublished) in membrane rupture.

VACV is a member of the nucleo-cytoplasmic large DNA viruses (NCLDVs) and among this group the best characterized both molecularly and morphologically (Roberts and Smith, 2008). NCLDVs are a group of viruses with a large genome and particle size. They encode for 100 to more than 1000 proteins and range from 100 to about 750 nm in size and are thus the largest viruses known (Colson et al., 2012; Yutin and Koonin, 2012; Yutin et al., 2013). Sequence analyses show that they share about 50–60 genes suggesting that they may be derived from a common ancestor (Iyer et al., 2006; Yutin and Koonin, 2012). The majority of these genes encode for proteins involved in transcription and DNA replication enabling these viruses to replicate their genome in the cellular cytoplasm, in association with large cytoplasmic virus factories (VFs). NCLDVs also share about six genes that for VACV are known to be involved in particle assembly. Among these is a gene with homology to the VACV D13L scaffold protein found in all NCLDVs discovered and sequenced so far (reviewed in Xiao and Rossmann, 2011; Yutin and Koonin, 2012). In VACV this protein transiently associates with the convex side of the crescent membrane shaping it into a sphere, whereas it is lost in the mature, brick-shaped, virion (Sodeik et al., 1994; Heuser, 2005). For other NCLDVs analysed, it remains part of the mature virion, shapes the membrane into an icosahedron and is referred to as ‘capsid’ protein (Hyun et al., 2011; Xiao and Rossmann, 2011; Salas and Andrés, 2013) .

The NCLDV Mimivirus was isolated from amoebae growing in a British water tower in 1992 (La Scola et al., 2003). It is an icosahedral-shaped enveloped particle with a size of about 500 nm covered with 125–140 nm long glycosylated protein fibres (Xiao et al., 2005; Klose et al., 2010; Kuznetsov et al., 2010). Its 1.2 Mb genome has the potential to encode for almost 1000 proteins, thus exceeding the genome size of some bacteria (Raoult et al., 2004). In infected Acantamoeba polyphaga Mimivirus creates a large cytoplasmic VF from which virions are assembled and which is likely also the site of cytoplasmic DNA replication (Suzan-Monti et al., 2007; Mutsafi et al., 2010). The immediate periphery of the VF displays partly formed icosahedral-shaped particles as well as membranes potentially involved in the formation of these particles. At some distance from the VF completed icosahedral virions displaying a core structure and a nucleoid, some additionally coated with the long fibres, can be seen (Suzan-Monti et al., 2007).

The present study was initiated to ask whether Mimivirus acquires its membranes similar to VACV using open membrane intermediates. For this we analysed the membranes that accumulate at the periphery of the VF in detail and determined their relationship to the maturing particles using ET. Our data show striking similarities to VACV as the membranes at the VF-periphery appeared open and were connected to growing particles studded on their convex side with a protein coat, likely the equivalent of the VACV scaffold protein.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Thawed cryo-sections confirm open membranes as intermediates of VACV assembly

In our previous study on VACV we used infected cells preserved by high-pressure freezing and freeze substitution embedded in lowicryl HM20. As shown below, however, we found that this embedding did not resolve the membranes associated with Mimivirus assembly whereas these were very well resolved in thawed cryo-sections. Thus, we first used VACV-infected cells and compared thawed cryo-sections with HM20 embedding to make sure that the membranes were well resolved with both methods. VACV-infected cells were prepared for cryo-sectioning or embedded in HM20 as before (Chlanda et al., 2009) and labelled with anti-A14L, a major membrane protein of VACV.

Labelled HM20-embedded samples displayed all of the feature observed before; within the VACV-VF A14L-positive crescent membranes accumulated that in fortuitous sections revealed two layers which we previously identified as the membrane and the viral scaffold (Fig. 1A and B). The ends of the crescent were connected to small, uncoated, membranes that curled in the opposite direction (Fig. 1B). In close proximity to the crescents A14L-positive membranes accumulated. All of these details were also revealed on thawed cryo-sections where membranes typically appear white (Fig. 1C). The crescent-membrane was studded on its convex side with the spike-like scaffold and connected to ends that curved in the opposite direction. As before, A14L-positive membranes, that appeared as small vesicles and membrane curls, accumulated close to the crescent-membrane (Fig. 1C).

figure

Figure 1. Appearance of the VACV membranes in Lowicryl-HM20 embedded cells and in thawed cryo-sections. A and B, HeLa S2 fixed at 12 h post-infection and embedded in HM20 after high-pressure freezing. C adherent HeLa cells fixed at 12 h post-infection and prepared for cryo-sectioning. Thin sections were labelled with anti-A14L and protein A coupled to 10 nm gold. Arrowheads show the membrane ends of crescents (cr). The arrows point to the viral membrane (mb) and scaffold protein (sc) respectively. ‘m’ depicts the A14L-positive membranes that typically accumulate close to the crescents. Note that the membranes in thawed cryo-sections appear white. Bars in A and C – 200 nm, in B – 100 nm. IV, immature virion, mi, mitochondrium.

Download figure to PowerPoint

Figure  2 shows 0.9 nm slices of a representative tomogram of a 200 nm thick cryo-section contrasted with uranyl-acetate and embedded in methyl-cellulose. The images show an immature virion (IV) and a crescent both coated on their convex side with scaffold. The crescent-membrane was connected to small curved membranes that were revealed only in some of the slices of the tomogram (Fig. 2A, compare slices in 1–4, Movies S1 and S2). The tomogram also revealed the small membranes that accumulate in the vicinity of the crescents. As shown before, ET revealed that they were closely associated with a honeycomb-structured scaffold patch (Fig. 2A and B slices 1–4, Movies S1 and S2; Chlanda et al., 2009). These patches were typically surrounded by cisternal membranes reminiscent of the endoplasmic reticulum (ER; Fig. 2A and B).

figure

Figure 2. ET of VACV-crescents and associated membranes. A and B shows several 0.9 nm slices of a tomogram (taken from Movie S1) from 200 nm thick thawed cryo-sections of HeLa cells infected as in Fig. 1C. The images show an immature virion (IV), a crescent (cr) and a small patch of scaffold protein (star). The white arrows point to the small membrane curls that accumulate close to the scaffold patch, the black arrows point to the ends of the crescent that typically curve away from the coated crescent membrane. Note that the crescent ends and the scaffold patch can only be seen in some of the slices of the tomogram. B and C rendering of the crescent and the scaffold patches (taken from Movie S2). The membranes are in green, the scaffold in red. It shows that the crescent ends are open as are the small membranes that accumulate close to scaffold patch and adopt an ‘open-eight’ structure (right image in C). The small cisternae in the vicinity of the scaffold patch in B-4 and C (blue) is closed. Bars in A and B – 100 nm.

Download figure to PowerPoint

Rendering (see Experimental procedures for details) confirmed that the crescent consisted of an open membrane sheet connected to open membrane ends (Fig. 2C; Movies S1 and S2). The small membranes that accumulated on the honeycomb-scaffold looked like curls or small vesicles on 2D sections (Fig. 1, 2A-1–4) but after rendering they appeared as a serpentine-like structure with open ends (Movie S2). In contrast, the membrane cisternae in the vicinity of the honeycomb structures consisted of closed compartments (Fig. 2C). We generated several tomograms and analysed the ends of the crescents as well as the small precursor membranes. When considering those structures that were entirely contained within the volume of the section we found, consistent with our previous results, that the majority (86%) of them were open (Table 1).

Table 1. Quantification of membrane-ends and precursor membranes
 MimivirusVACV
  1. Membrane ends connected to a growing icosahedron (Mimivirus) or crescent (VACV) or membrane precursors not connected to growing virions were analysed by ET and determined whether they were closed or open as explained in Experimental procedures. A membrane structure was only considered for quantification when its entire volume was contained within the tomogram. n.a., not available, because the tomogram did not allow for an unequivocal determination.

  2. The average diameter of the membranes was determined as described in Experimental procedures.

  3. a.n = 49, b.n = 14, c.n = 59, d.n = 9, e.n = 10, f.n = 14, g.n = 10.

Membranes connected to icosahedrona/crescentb
open74%86%
closed6%0%
n.a.20%14%
Average diameter (nm)89.6 ± 24.2d53.2 ± 12e
Membrane not connected to icosahedronc/crescent
open54%
closed29%
n.a.17%
Average diameter (nm)94.7 ± 19.7f46.1 ± 6.7g

Altogether, we were able to confirm all previous observations using thawed cryo-sections to resolve the VACV membranes both by TEM and ET.

Appearance of the Mimivirus VF in thawed cryo-sections

Both in conventional embedded samples as well as in thawed cryo-sections prepared at 10 h post-infection, the Mimivirus-VF occupied about half of the cytoplasmic space (Fig. 3A and C). It consisted of an electron-dense central part surrounded by a layer of lower electron-density, as observed before (Suzan-Monti et al., 2007). In thawed cryo-sections the outer VF-layer displayed many (white) membranes and partly finished icosahedral-shaped particles (Fig. 3C–F). The former were not visible in conventional embedded infected cells (not shown) nor in cells prepared by high-pressure freezing and freeze substitution (figure compared 3A/B). With resin-embedding we found that the VF was heavily contrasted, which likely disguised the underlying membranes. Reducing the concentration of uranyl-acetate during substitution did not resolve the membranes (not shown) and we therefore focused on thawed cryo-sections throughout this study.

figure

Figure 3. Overview of Mimivirus VF. A. polyphaga was infected with Mimivirus and cells preserved by high-pressure freezing and freeze substitution (A and B) or prepared for cryo-sectioning (C–F) at 10 h after addition of virus. A and C are an overview over the VF with a typical electron-dense core surrounded by a layer of lower electron-density. The boxed area in A is shown at higher magnification in B and the boxed areas in C in D and E. The VF is surrounded by virions in various stages of maturation. The immediate periphery of the VF is shown in B, D–F; it shows growing icosahedral forms and in thawed cryo-sections also the accumulation of membranes that appear as small curls and vesicles (white arrows). These membranes are not obviously revealed in B (HM20-embedding). In D–F the forming virions are composed of two layers separated by a small gap. The inner layer is a membrane (black arrow – mb), whereas the outer layer presumably is the mimivirus capsid protein (black arrow – pcp). These features are not well resolved in HM20-embedded samples (B). Bars in A and C – 1 μm; in B, D–F – 200 nm. N, nucleus; fb; glycoprotein fibres; MV, mature virion; m, mitochondrium.

Download figure to PowerPoint

On thawed cryo-sections the partly formed icosahedral particles at the periphery of the VF displayed an inner white layer typical of a membrane and an outer, slightly thinner, layer that seemingly shaped the membrane into an icosahedron (Fig. 3D and E). A thin gap separated the two layers (Fig. 3D–F). Although the outer layer also looked like a membrane, additional evidence described below suggested that it may be the equivalent of the VACV scaffold, the viral capsid protein of Mimivirus (gene product of L425; see e.g. Xiao et al., 2005; Kuznetsov et al., 2010). On thin sections the inner membrane-layer seemed continuous with other membranes that lacked the putative capsid protein. In sharp contrast to the VACV crescent, these membranes curved into the forming particle rather than away from it (compare Fig. 3D–F with Fig. 1A–D). Resin-embedded cells revealed little of these details; the two layers of the forming virions were not readily resolved as were the VF-associated membranes (Fig. 3B).

The VF was surrounded by virions that are in increasing stages of maturation the further they are away from the centre of the factory (Fig. 4A–E). At the periphery of the VF forming icosahedrons and completed icosahedral particles without obvious internal structure accumulated (tentatively called immature virion-1; IV1; Fig. 4A/F and B/G). These particles acquired an additional layer, which in fortuitous sections appeared to be studded by tiny spikes (Fig. 4B/H; tentatively called IV2). At some distance from the VF, particles with a spherical core structure encasing the viral genome (as assessed by anti-DNA labelling) accumulated [tentatively called mature virus (MV); Figure 4A, C, I and J]. The surface of these particles appeared thicker and fuzzier than in IV2 (Fig. 4A, C, E, I and J). In such particles the internal membrane, surrounding the core, seemingly connected to the outer layers at the vertices (Fig. 4C and E) and in some sections displayed two blob-like structures (Fig. 4C). One of these blobs was always located at the putative ‘stargate’ of the particle (Fig. 4A, C and D). Altogether we readily discriminated four different maturation forms summarized in F to J (the particles in I and J are two different views from the same mature virion).

figure

Figure 4. TEM appearance of the different Mimivirus maturation stages in thin thawed cryo-sections. A–E thawed cryo-sections of Mimivirus infected A. polyphaga fixed at 10 h post-infection and labelled with anti-DNA. A shows an overview over the different forms that accumulate around the VF. It shows immature virions (IV1) and mature virions (MV) with a clear spherical core and that are labelled with anti-DNA. The arrowheads point to the two lateral bodies-like structures of the MV shown in more detail in C. The black arrow point to the putative stargate and the white arrow to the putative DNA-entry portal of the virion shown in more detail in C. B shows two IV1s and one IV (IV2) with an additional layer that displays small spikes (arrow). D shows a mature particle (arrow indicates the putative stargate) and a stargate cut from the top (star). E higher magnification view of a mature particle to show the different layers. In some sections the virions displays at least 5 layers: 1. The core surface, 2. The membrane, 3. The putative capsid, 4. The additional layer with small spikes seen in IV2, 5. The fuzzy surface layer of the mature virion likely representing the attachment layer of the glycoprotein fibres. F–J is a summary of the different forms displayed according to increasing maturation. I and J are different views of mature virions; for details see text. Bars – 200 nm. PM, plasma membrane.

Download figure to PowerPoint

The accumulation of membranes in close proximity to the unfinished icosahedral forms suggested that these might contribute to their formation, which was next analysed by ET.

ET to show that the membranes are connected to forming viruses

We used ET to analyse whether the membranes at the periphery of the VF were continuous with the partly formed icosahedral particles as suggested by the images from thin sections. Figure 5 shows several 2.7 nm slices of two different tomograms imaging the forming virus particles and the associated membranes. These confirmed that the inner membrane layer of the forming virions was continuous with membranes that accumulated at the periphery of the VF (Fig. 5A and B; Movies S3 and S6). As also shown in Fig. 3 these membranes were not coated with the additional putative capsid layer and they curved into, rather than away, from the coated membrane (Fig. 5A and B). The outer layer was not continuous with other (VF-associated) structures, similar to what we previously showed for the D13L scaffold assembled on the VACV crescent (Chlanda et al., 2009). This layer seemingly contributed to the icosahedral shape of the membrane, suggesting a scaffolding function (Fig. 5A and B, 1–4 and 5–8). Finally, compared with the underlying membrane it appeared less discrete, of slightly different electron-density and in fortuitous slices showed thin striations (see e.g. Fig. 5A-5–8), suggesting altogether that this layer was not a membrane. Rendering confirmed the observations; it displayed the growing particle as an open membrane sphere coated on its convex side with the capsid putative protein. Whereas the inner membrane layer was connected to other, uncoated membranes, the outer layer was not connected to other structures. Finally, it also showed that the connected membranes had open ends (Fig. 5A and B, I–IV; Movies S4, S5, S7 and S8).

figure

Figure 5. The membranes at the periphery of the VF are connected to forming virions. A and B 2.7 nm slices of two different tomograms made from 300 nm thick thawed cryo-sections. A and B shows four slices (1–4) with three growing viruses in A and one in B. In 5–8 the middle virus in A and the virus in B are shown at higher magnification. The inner layer of the growing particle is continuous with the membranes (white) associated with the periphery of the VF. The outer layer is not continuous with other structures and thus likely represents the viral capsid protein (pcp). In I to IV rendering of the particle shown in 5–8, shown from different angles. The corresponding movies are S3–S5 for A and S6–S8 for B. Bars – 200 nm. mb, membrane.

Download figure to PowerPoint

We generated several dual-axis tomograms and analysed the ends of 55 growing icosahedral particles. We found that of all partly formed particles analysed (n = 55) 89% of the ends were connected to an uncoated membrane 74% of which appeared to be open (Table 1).

ET of the small membranes

In a next step we analysed the membranes that accumulated at the periphery of the VF that were not connected to forming virions. Several tomograms of 300 nm thick thawed cryo-sections were collected and analysed for membrane continuity and 3D organization. Rendering of such membranes showed another striking similarity to VACV membrane biogenesis. Whereas the membranes looked like open curls and small vesicles on 2D sections and in slices of the tomogram (Fig. 3D and E, Fig. 6A) in 3D they formed serpentines, with open ends on both sides (Fig. 6B, Movies S9–S11). Quantifying the membranes imaged in several dual-axis tomograms showed that of all serpentine structure that could be analysed unambiguously and of which the entire volume was contained in the tomogram, 54% were open.

As a control we also collected tilt series of membranes that typically accumulated close to the VF. The tomograms and rendering showed that these membranes formed closed compartments (Movies S12 and S13).

Because of the similarity to the VACV precursor membranes that accumulate on scaffold patches (Fig. 2) we also determined the average diameter of the serpentine turns. The average diameter of the curled membranes was the same irrespective of whether they were connected or not to growing virus particles. Strikingly, however, this diameter was about 90 nm for Mimi- and 50 nm for VACV (Table 1).

Altogether, the direct comparison of VACV to Mimivirus membrane biogenesis showed a number of similarities, in particular the use of open membrane intermediates for viral membrane biogenesis.

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

The discovery of Mimivirus has launched a vivid discussion about the definition of viruses and the evolution of life (Iyer et al., 2006; Boyer et al., 2010; Williams et al., 2011; Colson et al., 2012) . This discussion was rekindled recently with the discovery of yet a bigger virus, Megavirus chilensis of the NCLDV group (Arslan et al., 2011). NCLDVs have been compared extensively at the sequence level both between various members of the group as well as to bacterial, eukaryotic and archaeal genomes. NCLDVs share about 50 genes and most of those encode for proteins involved in DNA replication and RNA transcription (Iyer et al., 2006; Yutin and Koonin, 2012). The presence of these common genes makes it likely that NCLDVs are derived from a common ancestor (Iyer et al., 2006).

For VACV we showed recently that its membrane acquisition involves the formation of open membrane intermediates that create a membrane sheet composed of a single open membrane and is shaped by the assembly of the scaffold on its convex side. Our quantitative TEM and ET analyses of Mimivirus infected A. polyphaga show striking similarities; the viral membrane is formed by membrane-precursors that are open. This membrane is shaped by the assembly of another structure on its convex side, likely composed of the Mimivirus D13L homologue L425. Another striking similarity was the structure of the precursor membranes that in both cases formed serpentine-like structures with a well-defined turn, 90 nm for Mimivirus and 50 nm for VACV. Together with other striking similarities in morphogenesis (Mutsafi et al., 2010), we propose that the similar, unconventional mechanism of membrane acquisition is another strong piece of evidence in favour of a common NCLDV-ancestor. It exemplifies how studying morphogenesis can complement the evolutionary debates based on sequence analyses (Claverie and Abergel, 2010).

Collective data imply that the VACV membrane is derived from the early secretory pathway, the ER (summarized in Krijnse Locker et al., 2013). The VACV membranes that accumulate on the scaffold mostly exclude host proteins and were not found to be continuous with cellular membranes in this and in our previous ET study, even though the scaffold patches were surrounded by cisternal elements typical of the ER. We therefore suggested that they bud off from their donor membrane (the ER) and are transported to the scaffold (Chlanda et al., 2009). Rupture may occur in close association with the scaffold patches; the membranes associated with the patches were found to be open whereas the cisternae that accumulate around them were found to be closed. For Mimivirus the origin of its precursor membranes is virtually unknown and awaits further experiments. We did typically observe, however, the accumulation of membranes at some distance from the factory region, located between the maturing virions, and it is tempting to speculate that these are the source of the open membranes that accumulate at the VF-periphery. ET analyses showed that these membranes are closed and it is therefore likely that rupture occurs at the periphery of the VF.

figure

Figure 6. ET and rendering of the membranes at the periphery of the VF. A shows four 2.7 nm slices (1–4) of a tomogram of membranes associated with the periphery of the VF. This area was chosen to reconstruct membranes not connected to growing particles. 5–8 higher magnification view of the boxed areas in 1–4. The slices display the membranes as curls and sometimes as small vesicles. B rendering of the same area with the membranes in green and the putative capsid in red. In 3D the membranes adopt serpentines, ‘open-eights’ and open curls. The slices were taken from Movies S9–S11.

Download figure to PowerPoint

Compared with the forming viruses at the periphery of the VF the viruses that accumulate further away from the VF are exceedingly more complex. Our thin section TEM analyses discriminated at least four different forms. At the immediate VF-periphery were the forming viruses consisting of a single open membrane shaped by a single layer on its convex side, the putative capsid protein. In the immediate vicinity we observed closed icosahedral particles with an amorphous central part. Such particles then became studded with an additional outer layer displaying small spikes. In a next stage we observed particles with an internal core structure. These also contained two characteristic structures located between the core and the membrane, reminiscent of the VACV lateral bodies. Particles most distantly located from the VF then may acquire their glycoprotein fibres that were not well resolved in thawed cryo-sections. Based on our observations we assume that the mature particle is build up as follows; a spherical core structure encases the viral DNA and is surrounded by a single icosahedral-shaped membrane connected to the outer shell at the vertices and containing two lateral body-like structures. As suggested previously the latter may play two distinct roles; one during closure of the membrane and DNA-uptake, one during release of the viral DNA after entry and disassembly (Zauberman et al., 2008). The membrane of the virion is surrounded by the capsid, which in turn is studded by at least two additional layers (see Fig. 4E). This model is mostly consistent with cryo-EM and atomic force microscopy data by others (Xiao et al., 2005; 2009; Klose et al., 2010; Kuznetsov et al., 2010) with the exception that some proposed that the layer convex to the viral membrane was also a membrane enclosing the particle with two, rather than with one membrane (Xiao et al., 2005; 2009; Xiao and Rossmann, 2011). We think that this layer is not a membrane for several reasons. It shapes the membrane into an icosahedron in analogy to the VACV crescent and, most importantly, it is not continuous with other structures (including membranes) by ET on semi-thin sections. The final proof, however, awaits in our opinion detailed cryo-EM analyses of vitrified sections as shown for VACV (Chlanda et al., 2009).

Membrane dynamics in mammalian cells reveals no mechanism for membrane rupture and the use of open membrane intermediates to create new membranes. However, bacteria and viruses provide examples where cellular membrane rupture may be a crucial part of their intracellular life cycle. Thus, it is not clear how non-enveloped viruses acquire access to the cellular cytoplasm. Such viruses may be taken up by endocytosis and then somehow cross the endosomal membrane. The formation of a pore large enough for transport of the viral capsid or endosomal lysis have been proposed as a mechanism for cytosolic penetration (reviewed in Tsai, 2007; Moyer and Nemerow, 2011). Similarly, some intracellularly replicating bacteria gain access to the cell via phagocytosis and reach the cytosol by phagosomal lysis (reviewed in Ray et al., 2009). The molecular mechanism underlying endosomal or phagosomal escape is to some extent understood and likely involves bacterial or virally encoded proteins activated by cellular cues (Tsai, 2007; Ray et al., 2009; Moyer and Nemerow, 2011). An interesting aspect is the recently proposed idea that amoebae may have served as melting pot of (viral) evolution, in particular the evolution of NCLDVs (Boyer et al., 2009; Moliner et al., 2010; Yutin and Koonin, 2012). By virtue of their large phagocytic capacity amoebae can accommodate the simultaneous uptake, and possibly replication, of bacteria and viruses, thus allowing for exchange of genes. In such a scenario an ancestor NCLDV picked up membrane-rupturing genes from co-infecting bacteria or viruses. This early ancestor could originally have lacked a membrane but the acquisition of membrane remnants of co-infecting other pathogens could have provided an advantage, thus becoming a hallmark of the new ancestor NCLDV. Advantages of such a membrane are obvious; a more stable particle resisting (membrane-impermeable) nucleases and proteases, providing a scaffold for the formation of protein coats and access of the viral DNA to the cellular cytoplasm during particle uncoating by fusion with the phagosomal membrane (Zauberman et al., 2008).

Finally, we have provided evidence that the NCLDV-Mimivirus also uses open membrane intermediates for its membrane assembly. The challenge will be to find common molecules involved in this unusual process, which is the subject of our future studies.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Cells and viruses

HeLa S2 and adherent HeLa cells were maintained in RPMI (Gibco, cat# 618870-010) and DMEM (Gibco, cat# 31885-023), respectively, containing 10% heat-inactivated fetal calf serum, penicillin and streptomycin. HeLa cells were infected as described using the western reserve strain of VACV (Chlanda et al., 2009). VACV-infected cells were fixed with 1% glutaraldehyde (electron microscopy sciences, cat# 16220) in PHEM buffer [60 mM Pipes, 25 mM Hepes, 2 mM MgCl2, 10 mM EGTA (pH 6.9)] added directly to the medium for 1 h at RT. A. polyphaga cells were cultured in tissue-culture coated 75 cm2 flasks (5 cellstar cellculture, cat# 658170) in PYG (Biotechnologie Appliquee. Code 68992) medium at 28°C without CO2 as described (La Scola et al., 2003). A. polyphaga monolayers were infected by taking 3 ml of the supernatant of Mimivirus-infected cultures added directly to the medium and further incubated at 32°C. Fixed VACV-infected and unfixed Mimi-infected cells were processed for high-pressure freezing and freeze substitution as described below.

High pressure-freezing and freeze substitution

Unfixed Mimivirus- or (glutaraldehyde-fixed) VACV-infected HeLa S2 cells were collected by centrifugation (1000 g) at the indicated time post-infection, the pellet mixed with an equal volume of 40% (w/v) dextran (40 kDa; Fluka, Heidelberg, Germany). About 5 μl was loaded in a specimen ‘sandwich’ formed by 0.15/0.15 mm and 0.3 mm (Engineering Office M. Wohlwend GmbH cat# 353 and cat# 242 respectively) carriers adapted to a HPM-010 high-pressure freezer and immediately high-pressure frozen. The specimen carriers were opened and kept in liquid nitrogen during the manipulation steps. Freeze substitution of Mimi-infected cells was performed using Lowicryl HM20 (Polysciences) according to Chlanda et al. (2009) with some modifications. Freeze substitution was for 2 h at −90°C with 1.8% (w/v) uranyl acetate, 9 % (w/v) methanol in dry acetone in an ASFII freeze-substitution machine (Leica) and then warmed up to −50°C (slope 15°C h−1). Samples were washed three times with water-free glass-distilled acetone (EM sciences) and stepwise embedded with increasing concentrations of Lowicryl HM20 mixed with acetone (25% Lowicryl, 50% Lowicryl, 75% Lowicryl and twice 100% Lowicryl; 3 h for each step) over 13 h. The samples were warmed to −40°C (slope 5°C h−1) and HM20 was polymerized at −40°C for 48 h under UV. Polymerized samples were warmed to 20°C with a slope of 5°C h−1 and HM20 further polymerized at 20°C under UV for 48 h. Thin sections were cut with a Leica ultracut S microtome, plastic sections were post-stained for 2 min with 2.66% lead citrate in water and used for conventional EM. In order to reveal the membranes that accumulate at the periphery of the VF different substitution protocols were tried. These consisted of using concentrations of uranyl acetate between 1.8% and 0.1% or using uranyl acetate and osmium tetroxide (at concentrations between 1 and 0.1%) followed by embedding in epoxy resin at room temperature.

Cryo-sections and immunolabelling

For Tokuyasu cryo-sectioning, VAC-infected or Mimi-infected cells were fixed with 4% paraformaldehyde (Electron microscopy sciences, cat# 15710) and 0.1% glutaraldehyde in PHEM buffer for 1 h at RT. Cells were washed three times with 50 mM glycine (Sigma, cat# G8898) in PHEM-buffer. Fixed cells were incubated with 12% gelatin (Merck) in PHEM buffer for 15 min at 37°C, quickly pelleted and the gelatine solidified by incubation on ice for at least 30 min. The pellet was cut into small 1 by 1 mm pieces that were infiltrated overnight at 4°C with 2.3 M sucrose (Affymetrix cat# 21938). The infiltrated pellets were mounted on aluminium pins, flash-frozen in liquid nitrogen, the frozen pellet trimmed and cut at −90 to −120°C into 60–300 nm sections using a UC6 cryomicrotome (Leica). Thawed cryo-sections were immunolabelled as described (Slot et al., 1991) with anti-DNA (Roche) or anti-A14L (Salmons et al., 1997) with the some modifications; grids with thawed sections were incubated on 2% gelatin at 37°C for 15 min and subsequently on PBS for 15 min at 37°C. The sections were washed five times with washing solution (50 mM glycine in PBS) and incubated for 30 min at RT on drops of blocking solution [1.5 % BSA (Sigma, cat# A2153) (w/v), 0.1% (w/v) fish gelatin, 50 mM glycine in PBS]. The sections were labelled for 30 min with primary antibody, diluted in blocking solution at room temperature, washed and incubated with protein-A coupled to 10 nm gold particles for 30 min, washed with washing solution followed at least 10 changes of distilled water. Thin (60 nm) sections were post-contrasted with 1.8% uranyl acetate and 0.8% methylcellulose and semi-thin (200–300 nm) sections with 1.8% uranyl acetate and 1.2% methylcellulose for 8 min on ice.

Transmission-electron microscopy

Thin cryo- or plastic sections were observed with a Zeiss EM10 transmission electron microscope at 60 kV equipped with a Megaview CCD camera (Olympus). Digital images were processed with the Adobe Photoshop and Illustrator.

Electron tomography and image analysis

Semi-thin cryo-sections of VACV-infected cells or Mimi-infected cells were placed on 50 mesh grids coated with pilioform, carbon and protein A coupled to 15 nm gold (for image alignment). Prior to contrasting the sections were coated with another layer of protein A gold. Dual-axis tilt series were acquired using a Fishione dual-axis holder and a Tecnai (FEI) operating at 200 kV (TF20) equipped with an ‘Eagle’ bottom mounted 4K camera (200 kV, HS CCD 4 port read-out; FEI). Acquisition was performed using serial EM software at 1° increment from +60 to −60 degrees. Tilt series were recorded at 19 000 or 25 000 magnification, binning 2, corresponding to a pixel size of 1.13 nm and 0.89 nm respectively. Alignment, 3D reconstruction, merging and modelling of the tilt series were performed with IMOD software (Kremer et al., 1996). Rendering was performed manually, assigning colours to certain structures (membrane, scaffold, capsid) based on their morphology. Movies of tomograms were made using IMOD and Image J. To quantify open versus closed membranes the continuity of the membrane was followed manually in consecutive slices of the tomogram. Only those membranes were considered for quantification when the tomogram contained the entire volume of the membrane structure. This was considered to be the case if the end of the membrane structures was located in the middle of the 200–300 nm section, while those structures where the ends localized to the boarder of the sections were not considered. To quantify the diameter of the membranes connected or not to the viral particle three slice of a tomogram with a known pixel size were considered and the size in nm determined in Image J.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

This work was funded by the priority programme SPP1175 of the German scientific research foundation (DFG). We also thank the electron microscopy core facility of the Heidelberg university and their staff for technical as well as the cluster of excellence CellNetworks for financial support.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
cmi12156-sup-0001-si.mov9741K

(In all rendered tomograms the membranes are green and the (putative) scaffold in red)

Movie S1. Tomogram and rendering of a VACV factory region. Dual-axis tilt series was generated from 200 nm thick thawed cryo-sections from +60 to −60 degrees and tomogram made in iMOD. The movie corresponds to the slices shown in Fig. 2A and B and the crescent, the small scaffold patch, its associated membranes and a cisternal element close by (blue) were rendered. Bar = 200 nm.

cmi12156-sup-0002-si.avi2225K

Movie S2. Three-dimensional model of the crescent shown in the tomogram in Movie S1 obtained after rendering of Movie S1. The membrane is green and the scaffold is red. Bar = 200 nm.

cmi12156-sup-0003-si.mov3927K

Movie S3. Tomogram of a peripheral area of the Mimivirus factory region. Dual-axis tilt series were generated from 300 nm thick thawed cryo-sections from +60 to −60 degrees. The movie corresponds to the slices shown in Fig. 5A. Bar = 200 nm.

cmi12156-sup-0004-si.avi8070K

Movie S4. Tomogram and rendering of the same area as shown in Movie S3. It shows the membranes in green and the putative capsid in red. The movie shows how the factory-associated membranes are connected to the growing virion whereas the putative scaffold on its convex side is not continuous with other structures. Bar = 200 nm.

cmi12156-sup-0005-si.avi1801K

Movie S5. Three dimensional model of the rendered growing particle in Movie S4. The growing particle is connected to open membranes (green) and shaped by the putative capsid (red) on the convex side. Bar = 100 nm.

cmi12156-sup-0006-si.mov5973K

Movie S6. Tomogram of a peripheral area of the Mimivirus factory region. Dual-axis tilt series were generated from 300 nm thick thawed cryo-sections from +60 to −60 degrees. The movie corresponds to the slices shown in Fig. 5B. Bar = 200 nm.

cmi12156-sup-0007-si.avi7596K

Movie S7. Tomogram and rendering of the same area as shown in Movie S6. It shows the membranes in green and the putative capsid in red. The movie shows how the factory-associated membranes are connected to the growing virion whereas the capsid on its convex side is not continuous with other structures. Bar = 100 nm.

cmi12156-sup-0008-si.avi2658K

Movie S8. Three-dimensional model after rendering of one of the growing particles in Movie S7. 3D model of a growing particle showing the connected open membranes and the capsid on the convex side. Bar = 100 nm.

cmi12156-sup-0009-si.mov4145K

Movie S9. Tomogram of membranes that accumulate at the factory region that are not connected to growing particles. Dual-axis tilt series were generated from 300 nm thick thawed cryo-sections from +60 to −60 degrees. The movie corresponds to the slices shown in Fig. 6A. Bar = 200 nm.

cmi12156-sup-0010-si.avi7263K

Movie S10. Tomogram and rendering of membranes shown in Movie S9. Dual-axis tilt series were generated from 300 nm thick thawed cryo-sections from +60 to −60 degrees. The movie corresponds to the slices shown in Fig. 6. Bar = 100 nm.

cmi12156-sup-0011-si.avi5092K

Movie S11. Rendering of the membranes shown in Movie S9. In three dimensions the VF-associated membranes appear as open membrane sheets connected to form an open-eight conformation. Bar = 100 nm.

cmi12156-sup-0012-si.avi5044K

Movie S12. Tomogram and rendering of control membranes. Tilt series and reconstruction of membranes in Mimivirus-infected cells not located at the periphery of the VF (control membranes) It shows that the green membranes are closed compartments. Bar = 100 nm.

cmi12156-sup-0013-si.avi1101K

Movie S13. Rendering of the Movie S12. The 3D model shows the control membranes as closed compartments. Bar = 100 nm.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.