The Norovirus NS3 Protein Is a Dynamic Lipid- and Microtubule-Associated Protein Involved in Viral RNA Replication

Abstract

Norovirus (NoV) infections are a significant health burden to society, yet the lack of reliable tissue culture systems has hampered the development of appropriate antiviral therapies. Here we show that the NoV NS3 protein, derived from murine NoV (MNV), is intimately associated with the MNV replication complex and the viral replication intermediate double-stranded RNA (dsRNA). We observed that when expressed individually, MNV NS3 and NS3 encoded by human Norwalk virus (NV) induced the formation of distinct vesicle-like structures that did not colocalize with any particular protein markers to cellular organelles but localized to cellular membranes, in particular those with a high cholesterol content. Both proteins also showed some degree of colocalization with the cytoskeleton marker β-tubulin. Although the distribution of MNV and NV NS3s were similar, NV NS3 displayed a higher level of colocalization with the Golgi apparatus and the endoplasmic reticulum (ER). However, we observed that although both proteins colocalized in membranes counterstained with filipin, an indicator of cholesterol content, MNV NS3 displayed a greater association with flotillin and stomatin, proteins known to associate with sphingolipid- and cholesterol-rich microdomains. Utilizing time-lapse epifluorescence microscopy, we observed that the membrane-derived vesicular structures induced by MNV NS3 were highly motile and dynamic in nature, and their movement was dependent on intact microtubules. These results begin to interrogate the functions of NoV proteins during virus replication and highlight the conserved properties of the NoV NS3 proteins among the seven Norovirus genogroups.

Importance: Many mechanisms involved in the replication of norovirus still remain unclear, including the role for the NS3 protein, one of seven nonstructural viral proteins, which remains to be elucidated. This study reveals that murine norovirus (MNV) NS3 is intimately associated with the viral replication complex and dsRNA. We observed that the NS3 proteins of both MNV and Norwalk virus (NV) induce prominent vesicular structures and that this formation is dependent on microtubules and cellular cholesterol. Thus, this study contributes to our understanding of protein function within different Norovirus genogroups and expands a growing knowledge base on the interaction between positive-strand RNA [(+)RNA] viruses and cellular membranes that contribute to the biogenesis of virus-induced membrane organelles. This study contributes to our understanding of viral protein function and the ability of a viral protein to recruit specific cellular organelles and lipids that enable replication.

Keywords: cellular lipids; cytoskeleton; norovirus; virus replication.

FIG 1

The MNV NS3 protein associates with the viral replication complex. Thawed cryosections of MNV-infected RAW264.7 cells were fixed at 12 h p.i. (A and B), 18 h p.i. (C and D), or 24 h p.i. (E and F) and were immunolabeled with anti-NS3 antibodies and 10-nm protein A gold. Bars, 200 nm in all cases.

FIG 2

The MNV NS3 protein associates with dsRNA within the viral replication complex. Thawed cryosections of MNV-infected RAW264.7 cells fixed at 12 h p.i. (A and B) and 18 h p.i. (C and D) were dually labeled with antibodies to dsRNA (5-nm protein A gold) and NS3 (15-nm protein A gold). Arrowheads indicate areas enriched for both the 5-nm and 10-nm gold particles corresponding to immunolabeling for dsRNA and NS3. RC; replication complex. Bars, 200 nm in all cases.

FIG 3

Comparative analysis of the Norwalk virus and murine norovirus NS3 proteins. (A) ClustalW alignment of the amino acid sequences of NV NS3 (GenBank accession no. NP_786946) and MNV NS3 (accession no. YP_724457), with the dark shading corresponding to identical amino acids and light shading corresponding to conserved amino acids. (B) PROSITE scanning of the NV and MNV NS3 amino acid (aa) sequences reveals a conserved SF3_Helicase motif in both proteins. (C) TMpred prediction of transmembrane regions of each protein indicates a potential transmembrane alpha-helix within the N termini of both proteins.

FIG 4

Both MNV NS3 and NV NS3 show a vesicular pattern upon transfection. Vero cells were transfected with either MNV NS3- or NV NS3-containing plasmids and fixed at 24 h p.t., and their localization was assessed by IF analysis. Both proteins were expressed well in transfected cells and were recognized by antibodies specific to their recombinant tag (MNV NS3-His [a and b] and NV NS3–c-myc [c and d]). Both the MNV and NV NS3 proteins displayed similar localization patterns, and both proteins induced a vesicular staining pattern (indicated by arrows). Additional strong perinuclear staining was observed for NV NS3, and cytoplasmic reticular staining was observed for both proteins. This staining pattern for NV NS3 was consistent with what we previously described for MNV NS3 (23).

FIG 5

NV NS3 displays significant colocalization with PDI (n = 50), GM130 (n = 49), and β-tubulin (n = 33). Vero cells were transfected with an NV NS3-containing plasmid and fixed at 24 h p.t., and localization was assessed by IF analysis. Shown is a comparison of the intracellular staining pattern of NV NS3 with those of cellular markers for specific organelles, namely, PDI (for the rER), GM130 (for the Golgi apparatus), EEA1 (n = 50) (for endosomes), and β-tubulin (for microtubules). MNV NS3, like NV NS3, showed little colocalization with markers for the ER and endosomes (a to d and i to l). However, we observed colocalization of NV NS3 with the Golgi apparatus and microtubules using β-tubulin (e to h and m to p). dapi, 4′,6-diamidino-2-phenylindole.

FIG 6

Both MNV NS3 and NV NS3 vesicular structures colocalize with filipin cholesterol stain upon transfection. Vero cells were transfected with either MNV NS3- or NV NS3-containing plasmids, fixed at 24 h p.t., and stained with filipin. In mock-transfected cells, filipin staining was confined to the plasma membrane and also foci within the cytoplasm and perinuclear region, which are characteristic of the trans-Golgi network and endosomes. Transfection of cells with either NV NS3–c-myc or MNV NS3-His resulted in significant colocalization of NS3 proteins, with filipin surrounding the NS3 vesicles (f and i [n = 15 and n = 20, respectively]). In all cases, the anti-6×His and anti-c-myc antibodies were detected with species-specific antibodies conjugated to Alexa Fluor 488, and the filipin stain naturally fluoresces blue. Pearson's coefficient values are provided in the merged panels, with a value of >0.500 corresponding to colocalization.

FIG 7

Both MNV NS3 and NV NS3 vesicular structures colocalize with the lipid markers stomatin-GFP and flotillin-GFP. Vero cells were transfected with either MNV NS3- or NV NS3-containing plasmids, concurrently transfected with either stomatin-GFP or flotillin-GFP, and fixed at 24 h p.t. Colocalization of both the MNV and NV NS3 proteins with stomatin-GFP and flotillin-GFP was observed. MNV NS3 predominantly associated with stomatin-GFP, particularly within the vesicular structures induced in the cytoplasm (n = 29) (D to F), rather than flotillin-GFP (n = 29) (A to C). In contrast, NV NS3 appeared to associate more with flotillin-GFP in the perinuclear region (n = 20) (G to I), although some colocalization was also observed with stomatin-GFP (n = 24) (J to L).

FIG 8

Colocalization of both the MNV NS3 and NV NS3 vesicular structures with the cholesterol filipin stain is disrupted after treatment with lovastatin (A to C and G to I [n = 14 and 32, respectively]) but not methyl-β-cyclodextrin (MBCD) (D to F and J to L [n = 13 and 26, respectively]). At 6 h p.t., Vero cells were transfected with either MNV or NV, treated with lovastatin (10 μM) or methyl-β-cyclodextrin (10 μg/ml) for 18 h, and fixed at 24 h p.t. Treatment with lovastatin showed a disruption of the vesicle structures and dispersion from the perinuclear regions with which they are usually associated. NS3 labeling appears to represent smaller vesicles (or aggregations). Treatment with MβCD did not show a visible effect on NS3 structures, although a reduction in the diffuse cytoplasmic NS3 labeling signal was apparent in MNV NS3 samples, and a small reduction in colocalization was observed, possibly due to an overall reduction of cellular cholesterol levels.

FIG 9

Both MNV NS3 and NV NS3 vesicular structures are dependent on microtubules and are disrupted upon nocodazole (NOZ) treatment. Vero cells were transfected with recombinant cDNA plasmids encoding MNV NS3 and NV NS3 at 8 h p.t. Cells were subsequently incubated with or without medium containing 10 μM NOZ and fixed at 24 h p.t. Both the MNV and NV NS3 proteins show a disruption of the regular vesicle-like structures in the presence of NOZ and an overall reduction in microtubule association (D to F [n = 17] and J to L [n = 21]) compared to the untreated controls (A to C [n = 15] and G to I [n = 37]).

FIG 10

(A) MNV NS3-GFP vesicular structures display regular movement upon transfection. Vero cells were transfected with an MNV NS3-GFP plasmid and viewed over a 4-h period at 8 to 12 h p.t. NS3-GFP-transfected cells show movement of NS3-GFP vesicles throughout the cell. (A) Still images show the movement of NS3-GFP over time (images shown were those collected every 25 min). The time-lapse merge panel shows tracking over an ∼4-h period at 8 to 12 h p.t., showing highly motile vesicle structures (with images being taken every 5 min) (see Movie S1 in the supplemental material). (B) MNV NS3-GFP vesicular structures show a lack of movement upon NOZ treatment. Vero cells were transfected with an MNV NS3-GFP recombinant cDNA plasmid, and at 7.5 h p.t., cells were subsequently incubated with medium containing 10 μM NOZ and visualized from 8 to 12 h p.t. Still images show the movement of NS3-GFP over time (images shown were those collected every 25 min). The time-lapse merge panel shows tracking over an ∼4-h period at 8 to 12 h p.t. and was analyzed by using IMARIS software, showing immotile GFP structures. NOZ-treated NS3-GFP-transfected cells show a lack of vesicle movement (with images being taken every 5 min) (Movie S2).

FIG 11

IF analysis showing the effect of NOZ treatment on the distribution of MNV NS3, β-tubulin, and dsRNA in infected RAW264.7 cells. Cells were infected only and viewed for the distribution of MNV NS3 (red) and β-tubulin (green) (A to C) or dsRNA (green) (G to I). MNV-infected cells were additionally treated with NOZ at 1 h p.i. and viewed for the distribution of MNV NS3 (red) and β-tubulin (green) (D to F) or dsRNA (green) (J to L). The distributions of NS3 and dsRNA are affected by NOZ treatment, and the associations between NS3 and β-tubulin (Rr of 0.47 ± 0.14 [untreated] [n = 21] versus Rr of 0.46 ± 0.15 [treated] [n = 23]) and between NS3 and dsRNA (Rr of 0.83 ± 0.08 [untreated] [n = 20] versus Rr of 0.74 ± 0.11 [treated] [n = 19]) are unaltered.