Three-dimensional architecture and biogenesis of membrane structures associated with hepatitis C virus replication

Abstract

All positive strand RNA viruses are known to replicate their genomes in close association with intracellular membranes. In case of the hepatitis C virus (HCV), a member of the family Flaviviridae, infected cells contain accumulations of vesicles forming a membranous web (MW) that is thought to be the site of viral RNA replication. However, little is known about the biogenesis and three-dimensional structure of the MW. In this study we used a combination of immunofluorescence- and electron microscopy (EM)-based methods to analyze the membranous structures induced by HCV in infected cells. We found that the MW is derived primarily from the endoplasmic reticulum (ER) and contains markers of rough ER as well as markers of early and late endosomes, COP vesicles, mitochondria and lipid droplets (LDs). The main constituents of the MW are single and double membrane vesicles (DMVs). The latter predominate and the kinetic of their appearance correlates with kinetics of viral RNA replication. DMVs are induced primarily by NS5A whereas NS4B induces single membrane vesicles arguing that MW formation requires the concerted action of several HCV replicase proteins. Three-dimensional reconstructions identify DMVs as protrusions from the ER membrane into the cytosol, frequently connected to the ER membrane via a neck-like structure. In addition, late in infection multi-membrane vesicles become evident, presumably as a result of a stress-induced reaction. Thus, the morphology of the membranous rearrangements induced in HCV-infected cells resemble those of the unrelated picorna-, corona- and arteriviruses, but are clearly distinct from those of the closely related flaviviruses. These results reveal unexpected similarities between HCV and distantly related positive-strand RNA viruses presumably reflecting similarities in cellular pathways exploited by these viruses to establish their membranous replication factories.

Figure 1. Colocalization of HCV Proteins with organelle-specific markers in HCV infected cells.

Huh7 cells were infected with HCV (strain Jc1) using 30 TCID₅₀/cell and 48 h later cells were fixed and processed for fluorescence microscopy. In case of samples shown in panels B–D, cells were first transfected with expression constructs specified in the left of each panel and 24 h later cells were infected as described above. Samples were analyzed with a Nikon TE2000-E inverted confocal microscope at 60× magnification. (A)–(D) Colocalization of HCV proteins specified in the top of each panel with protein disulphide isomerase (PDI; an ER marker), GFP-Rab21 (marker for early endosomes), GFP-Rab7 (marker for late endosomes) or βCOP-YFP (marker for COP I vesicles). The upper panels represent a low magnification overview; boxed areas are shown as enlargement in the corresponding panel below. The nucleus was stained with DAPI (blue). Scale bars represent 10 µm (top panels) and 2 µm (lower panels). The quantification of the degree of colocalization (Pearson's correlation coefficient) is given at the top of the enlarged pictures.

Figure 2. Correlative light-electron microscopy of cells containing a GFP-tagged subgenomic replicon.

(A) Epifluorescence microscopy of live cells containing a subgenomic replicon with a GFP-tagged NS5A. Huh7-Lunet cells were transfected with replicon RNA and seeded onto carbon-patterned sapphire discs. Twenty-four hours later cells were analyzed by fluorescence microscopy and immediately processed for EM. (a) Fluorescence image; (b) enlarged fluorescence image of the cell of interest; (c) merge of bright field and fluorescence images. Coordinates etched onto the surface of the sapphire disc were used to record the position of the selected cells. White squares in a and c enclose the cell shown in b. (B) EM micrograph of the cell boxed in panel Ab overlapped with the fluorescence image. Areas marked with a green dotted line indicate regions of intense fluorescence. Note that the images do not match perfectly because the fluorescence image corresponds to the complete cell whereas the EM image represents one 60 nm ultrathin section of the same cell. (C) Higher magnification images of the two different regions, labeled 1 and 2 in panel B, corresponding to regions with intense fluorescence (1) or a region corresponding to the intersection of high to low fluorescence (2). Region 1 (top panel) corresponds to a DMV-containing area residing in close proximity of the ER; region 2 (bottom panel) corresponds to a LD-enriched area containing DMVs in very close proximity. Areas marked with white squares in the left images are magnified in the corresponding right panels. LD, lipid droplet; ER, endoplasmic reticulum; DMV, double membrane vesicle; m, mitochondrium; if, intermediate filaments.

Figure 3. Time course of membrane alterations upon HCV infection.

(A) Images of mock-infected high-pressure-frozen and freeze-substituted Huh7.5 cells. (B–E) Micrographs of high pressure frozen freeze-substituted Huh7.5 cells harvested 16, 24, 36 and 48 h after infection with Jc1 (100 TCID₅₀/cell). Yellow squares indicate the areas that are shown at higher magnification on the right of each subpanel. Note the time-dependent increase of complexity of HCV-induced membrane alterations. ER, endoplasmic reticulum; m, mitochondria; MVB, multi-vesicular bodies; LD, lipid droplet; DMV, double membrane vesicle; MMV, multi membrane vesicle; DMT, double membrane tubule (labeled with a black arrow).

Figure 4. Immuno-EM localization of NS5A and NS3 in HCV-infected cells.

Huh7.5 cells were infected with 100 TCID₅₀/cell of Jc1 and fixed 16 h post-infection. (A) Cells shown in panels b, c and e to g were post-fixed with 0.1% OsO₄ providing best membrane preservation, especially in case of DMVs (f and g) and LDs (e). Both NS5A and NS3 localize predominantly to the ER (b and d) and 50–70 nm diameter single-membrane vesicles (SMVs; subpanel a, c and d). NS5A also localizes to lipid droplets (e). (B) Amount of gold particles per µm² in Jc1-infected versus mock-infected cells after immunolabeling with NS3- and NS5A-specific primary antibodies. Note the higher immunolabeling with samples prepared without OsO₄, but also the lower membrane preservation under this condition. (C) Relative labeling distribution of NS3 and NS5A. Thawed cryosections of cells post-fixed or not with OsO₄ were labelled with NS3- or NS5A-specific antibodies by using two different blocks and 3 different labeling experiments. Per labeling experiment two grids were considered, counting ∼100–200 gold particles per grid and attributing the particles to the indicated structures. In the case of uninfected cells only background labeling in the cytoplasm, on mitochondria and undefined structures was seen. Numbers refer to the percent of total gold particles counted per sample. ER, endoplasmic reticulum; SV, small vesicles; Cyto, cytosol; Mito, mitochondria; n.d., non-defined structures; NE, nuclear envelope; EE/LE, early/late endosomes; PM, plasma membrane; DMV, double membrane vesicle; LD, lipid droplet.

Figure 5. Correlation between HCV RNA replication and appearance of double membrane vesicles.

(A) Time course of spread of HCV infection in Huh7.5 cells infected with 100 TCID₅₀/cell of Jc1. Infected cells were detected by immunofluorescence using an NS5A-specific antiserum (upper panels). The graph shown below represents the result of counting ∼200 cells for each time point to determine the percentage of infected cells. Scale bars represent 50 µm. (B) Time course of accumulation of intracellular HCV RNA in infected Huh7.5 cells. The graph shows the result of two independent experiments (3 replicas each). Whiskers indicate the minimum and maximum values. (C) Colocalization of dsRNA and NS5A in cells infected with Jc1 (10 TCID₅₀/cell). Cells were fixed at time points specified in the left of each panel row and NS5A and dsRNA were detected by indirect immunofluorescence microscopy. DNA was stained with DAPI (blue). Boxed areas in the left panels indicate areas that are shown as enlargements in the corresponding right panels. The quantification of the degree of colocalization (Pearson's correlation coefficient) is given in the enlarged pictures. Scale bars represent 10 µm and 2 µm (left and right panels, respectively). (D) Time course of accumulation of DMVs and MMVs. For each time point, 10 cellular profiles were counted. The Mann-Whitney (non-parametric) test was applied to determine statistical significance. Error bars refer to the standard deviation. Note the striking correlation between the increase of intracellular HCV RNA and DMV number between 16 and 24 hpi.

Figure 6. 3D architecture of membrane rearrangements induced 16 h after HCV infection.

(A) Huh7.5 cells were infected with 100 TCID₅₀/cell of Jc1, fixed 16 h later and after HPF-FS processed for ET as described in materials and methods. Left: slice of a dual axis tomogram showing the various membrane alterations. Right: 3D reconstruction of the complete tomogram. Note the high number of DMVs. Panels B, C and D are part of the tomogram displayed in panel A and their position in panel A is highlighted by either a yellow dashed square (B and C) or by a star (D). In the 3D models shown on the right, the ER is depicted in dark brown, the inner membrane of DMVs and DMTs in yellowish brown and their outer membrane in semi-transparent light brown. Single membrane vesicles are colored in pink, intermediate filaments in dark blue and the Golgi apparatus in green. (B) Left: serial single slices through the same tomogram shown in panel (A) displaying a connection between the outer membrane of a DMV and the ER membrane (black arrows). Right: 3D surface model showing the membrane connection. (C) Left: serial single slices through the same tomogram illustrating a lasso-like structure of a DMV that after rendering reveals a pore-like opening that connects the interior of the DMV with the cytosol. The position of this opening in the 2D slice is marked with a black arrow. Right: 3D view of this DMV showing the ‘pore’. (D) Left: serial single slices through the same tomogram showing a DMV with a large inter-membrane space between its inner and outer membranes. Right: 3D view of this DMV. Scale bars represent 100 nm. This tomogram is shown in movie S1.

Figure 7. 3D architecture of membrane rearrangements induced 36 h after HCV infection.

(A) Left: slice of a dual axis tomogram taken from Huh7.5 cells 36 h after infection with 100 TCID₅₀/cell of Jc1. Right: 3D reconstruction of the tomogram. Note the extensive membrane reorganization and the appearance of MMVs predominating at late time points after infection. Black arrows show invaginations of DMVs. The white star indicates a large MMV; due to its complexity only its middle part could be rendered. (B) Left: slices through the same tomogram highlighting a DMT enwrapping a DMV and presumably leading to the formation of a MMV. Right: 3D surface rendering of this structure. (C) Left: slices through the same tomogram highlighting a ‘self-invagination’ event of a DMV, also leading to the formation of a MMV. Right: 3D surface rendering of this late structure, revealing an opening of this MMV towards the cytosol as a result of the self-invagination. Panels B and C are part of the tomogram displayed in panel A and their positions are highlighted by yellow dashed squares in panel A. Scale bars represent 100 nm. For further details see legend to Figure 6. This tomogram is shown in movie S2.

Figure 8. Membrane alterations induced by expression of individual HCV proteins.

Huh7.5 cells transfected with replicon RNAs (A, B) or Huh7-Lunet-T7 cells transfected with pTM-based expression constructs (C–H) specified in the left of each panel, respectively, were high pressure frozen 24 h after transfection, freeze substituted, embedded into epon resin and sections were analyzed by transmission EM. Representative images showing HCV-induced membrane alterations are shown. (I) Average diameter of specific vesicular structures, either DMVs or SMVs, detected in cells that had been transfected with constructs specified in the upper panels. Whiskers represent minimum and maximum values. (J) Number of vesicular structures detected in profiles of 10 transfected cells. Cells transfected with the pTM expression vector without HCV insert were used as reference. The number of vesicular structures per µm² is given; whiskers represent minimum and maximum values. (K) Relative abundance of membranous structures. Note that only upon expression of the NS3-5A polyprotein and NS5A two different structures were observed. In all other cases, only one membranous structure was detected.

Figure 9. Hypothetical models describing the formation of double membrane vesicles and their possible role in viral RNA replication.

(A) By analogy to flaviviruses HCV proteins induce invaginations of the ER membrane. Extensive invagination leads to a local ‘shrinking’ of the ER lumen. This model assumes that enzymatically active HCV replicase (green dots) reside in the lumen of the invagination and remain active as long as the vesicle is linked to the cytosol. Upon closure of the DMV, the replicase would become inactive (grey dots). Alternatively, closed DMVs might be connected to the cytosol via proteinaceous channels. (B) HCV proteins might induce tubulation of ER membranes that undergo secondary invagination and thus double membrane wrapping. These DMVs could initially be open to the cytosol, but might close off as replication/infection progresses. The resulting DMV might stay connected to the ER via a stalk or be released as a ‘free’ DMV (left or right drawing, respectively). (C) Induction of DMVs follows the same pathway as described for panel B, but the viral replicase remains on their cytosolic surface as discussed e.g. for the poliovirus , . (D) HCV RNA replication might occur on SMVs in close proximity of DMVs. In this case, DMVs might be an epiphenomenon or serve some other purpose for the HCV replication cycle. For each model, structures identified in the 3D reconstructions are shown next to or below the corresponding schematic drawing. For further details see text.