Negri bodies are viral factories with properties of liquid organelles

Abstract

Replication of Mononegavirales occurs in viral factories which form inclusions in the host-cell cytoplasm. For rabies virus, those inclusions are called Negri bodies (NBs). We report that NBs have characteristics similar to those of liquid organelles: they are spherical, they fuse to form larger structures, and they disappear upon hypotonic shock. Their liquid phase is confirmed by FRAP experiments. Live-cell imaging indicates that viral nucleocapsids are ejected from NBs and transported along microtubules to form either new virions or secondary viral factories. Coexpression of rabies virus N and P proteins results in cytoplasmic inclusions recapitulating NBs properties. This minimal system reveals that an intrinsically disordered domain and the dimerization domain of P are essential for Negri bodies-like structures formation. We suggest that formation of liquid viral factories by phase separation is common among Mononegavirales and allows specific recruitment and concentration of viral proteins but also the escape to cellular antiviral response.Negative strand RNA viruses, such as rabies virus, induce formation of cytoplasmic inclusions for genome replication. Here, Nikolic et al. show that these so-called Negri bodies (NBs) have characteristics of liquid organelles and they identify the minimal protein domains required for NB formation.

Fig. 1

Characterization of cytoplasmic inclusions in BSR cells infected by RABV. BSR cells were infected with CVS strain at a MOI of 0.5 and fixed at different times p.i. (8 h, 12 h, 16 h, 24 h). a Confocal analysis was performed after staining with a mouse monoclonal anti-N antibody followed by incubation with Alexa-488 donkey anti-mouse IgG. DAPI was used to stain the nuclei. Scale bars correspond to 15 µm. b–d Quantification of cytoplasmic structures labeled with anti-N antibody. At the indicated time p.i. the number of small dots (surface < 0.26 µm²) b, of intermediate inclusions (surface between 0.26 and 3.7 µm²) c, and of large inclusions (surface > 3.7 µm²) d per cell was quantified using the image toolbox of MatLab software as described in the experimental procedures. *p < 0.02; **p < 0.01 (n = 20, two tailed Mann Whitney U test). e Average surface of the largest inclusion in the cell at the indicated time p.i. Surfaces of inclusions were determined using the image toolbox of MatLab software as described in the experimental procedures. The mean is shown with error bars representing the SD. ***p < 10⁻⁴ (n = 20, two tailed Welch’s t-test). f Confocal analysis revealing the basal localization of small dots and the median localization of inclusions in RABV infected cells at 24 h p.i. The analysis was performed after staining as in a. Scale bars correspond to 15 µm. g, h EM characterization of the ultrastructural aspects of BSR cells infected by RABV. Basal sections g reveal the presence of RNPs whereas median sections h reveal the presence of NBs displaying an electron dense granular structure (colored in blue in the bottom row) which lose their spherical shape when they associate with membranes at 24 h p.i

Fig. 2

NBs are liquid organelles. a NBs are close to spherical. Distribution of axial ratios of NBs observed 12 h p.i. (20 cells and 100 NBs). The axial ratio (a/b) was determined by fitting the NB to an ellipse having the same second moments of area using the image toolbox of MatLab software (a, b: long and short axes of the ellipse). b NBs can fuse together. BSR cells infected by rCVSN2C-P-mCherry were imaged at the indicated time (lower left corner). The initial time corresponds to 15 h p.i (top row) and 18 h p.i. (bottom row). Images have been extracted from Supplementary Movies 1 and 2 and are shown at 10-sec intervals (top row) and 5-sec intervals (bottom row). Scale bars: 3 µm. c Spherical bubbles cross NBs. BSR cells infected by rCVSN2C-P-mCherry were imaged at the time indicated (lower left corner). The initial time corresponds to 16 h p.i. Images have been extracted from Supplementary Movie 3 and are shown at 5-s intervals. Scale bar: 3 µm. d NBs are sensitive to a hypotonic shock. BSR cells infected by rCVSN2C-P-mCherry were imaged. A hypotonic shock was applied at indicated t = 0 (corresponding to 18 h p.i.). Images are shown at 2-min intervals. Images have been extracted from Supplementary Movie 4. Scale bar: 15 µm. e–h Fluorescence recovery after photobleaching (FRAP) of P-mCherry in BSR cells at 37 °C. The diameter of the photobleached regions was 2.7 µm. e–g FRAP data were corrected for background, normalized to the minimum and maximum intensity. The mean is shown with error bars representing the SD. Experimental curves were fitted with a two-exponential model (in black). e Cytosolic P-mCherry expressed in BSR-T7/5 cells was photobleached 24 h after transfection of pTit-P-mCherry plasmid. Data were from 11 FRAP events (Supplementary Fig. 2). f Cytosolic P-mCherry expressed in BSR cells infected by rCVSN2CΔG-P-mCherry was photobleached 16 h p.i. Data were from 13 FRAP events (Supplementary Fig. 3). g P-mCherry localized in NBs in BSR cells infected by rCVSN2CΔG-P-mCherry was photobleached 16 h p.i. Data were from 12 FRAP events (Supplementary Fig. 4). h Fluorescence recovery profile along a diameter of a photobleached NB as in g

Fig. 3

NBs and SGs are non-miscible liquid organelles. U373-MG cells were transiently transfected with pG3BP-eGFP (to visualize SGs, top row). 1 h post transfection, they were infected with the recombinant virus rCVSN2C-P-mCherry (medium row). Live-cell time-lapse experiments were performed at 16 h p.i. G3BP-GFP signals (green) and P-mCherry signals (red) are merged in the bottom row. The time post-infection is displayed in the upper left corner of each panel and the scale bars correspond to 10 μm. Images have been extracted from Supplementary Movie 5 and are shown at 5-min intervals. Note also the mobile, punctate G3BP-GFP-containing structures inside NBs

Fig. 4

Effect of cytoskeleton-disrupting drugs on formation and evolution of NBs. Nocodazole (NCZ, 2 µM), Taxol (1.25 nM) and Cytochalasin D (Cyto D, 2.5 µM) were added 1 h before and kept all along infection. BSR cells were infected with CVS strain at a MOI of 0.5 and fixed at 16 h p.i. NT: non treated cells. a Confocal analysis was performed after staining with a mouse monoclonal anti-N antibody followed by incubation with Alexa-488 donkey anti-mouse IgG. DAPI was used to stain the nuclei. Scale bars correspond to 15 µm. b–d Quantification of cytoplasmic structures labeled with anti-N antibody in treated and non-treated cells. The number of small dots (surface < 0.26 µm²) b, of intermediate inclusions (surface between 0.26 and 3.7 µm²) c, and of large inclusions (surface >3.7 µm²) d per cell was quantified using the image toolbox of MatLab software as described in the experimental procedures. **p < 0.01, ns: not significant (n = 20, two tailed Mann Whitney U test). e Average surface of the largest inclusion in the cell 16 h p.i. in non-treated and treated cells. Areas of inclusions were determined using the image toolbox of MatLab software as described in the experimental procedures. The mean is shown with error bars representing the SD. ns: not significant; **p < 2.10⁻³; ***p < 10⁻³ (n = 20, two tailed Welch’s t-test)

Fig. 5

RNPs are ejected from NBs and transported along the microtubule network. Live cell imaging was performed on BSR cells infected by rCVSN2C-P-mCherry at a MOI of 0.5 at 16 h p.i. a Ejection of RNPs from NBs. The time is indicated in the upper right corner. Images have been extracted from Supplementary Movie 6 and are shown at 5-sec intervals. Ejection events are observed on a single image and indicated by arrowheads. The resulting RNPs are indicated by arrows. Scale bar corresponds to 3 µm. b Impact of cytoskeleton-disrupting drugs on RNPs transport in the cytosol. Nocodazole (NCZ, 2 µM), Taxol (1.25 nM) and Cytochalasin D (Cyto D, 2.5 µM) were added 1 h before and kept all along infection (NT: non-treated cells). 120 frames (one frame per 1 s, reflecting 120 s) of time-lapse microscopy (such as Supplementary Movie 7) are displayed as maximal intensity projection in order to visualize RNP trajectories which are indicated by arrows in the magnification shown in the lower row. In the NCZ-treated cell (Supplementary Movie 8), an ejection event is indicated by an arrowhead showing that RNP ejection from NBs occurs in absence of an intact microtubule network. Scale bars correspond to 15 µm. c Velocity of RNPs in the cytosol in non-treated and treated cells. The speed and the trajectories of RNPs were determined as described in the experimental procedures. Only the RNPs that were tracked on four consecutives images were taken into account. The mean is shown with error bars representing the SD. d RNPs are transported along microtubules. BSR cells were co-infected with rCVS N2C-P-mCherry and a modified baculovirus encoding human tubulin-GFP (Cell-light Tubulin-GFP). Images were deconvoluated using the Huygens Imaging software (Supplementary Fig. 5). Scale bar corresponds to 15 µm. Images have been extracted from Supplementary Movie 9 and are shown at 2.5-s intervals (bottom row)

Fig. 6

Co-expression of N and P leads to the formation of inclusion bodies recapitulating NB properties. a BSR-T7/5 cells were co-transfected for 24 h with plasmids pTit-P and pTit-N (in equimolar concentration). N was revealed with a mouse monoclonal anti-N antibody followed by incubation with Alexa-488 donkey anti-mouse IgG and P was revealed with a rabbit polyclonal anti-P antibody followed by incubation with Alexa-568 donkey anti-rabbit IgG. DAPI was used to stain the nuclei. Scale bars correspond to 15 µm. b Fluorescence recovery after photobleaching (FRAP) of P-mCherry localized in inclusion bodies in BSR-T7/5 co-expressing P-mCherry and N. The diameter of the photobleached regions was 2.7 µm. FRAP data were corrected for background, normalized to the minimum and maximum intensity, and the mean is shown with error bars representing the SD. Experimental curves were fitted with a two-exponential model (in black). Data were from 21 FRAP events (Supplementary Fig. 6). c Domain organization of RABV P polypeptide chain. P contains an N-terminal domain which binds to N⁰ (PNTD:P-N⁰), two intrinsically disordered domains (IDD1 and IDD2), a dimerization domain (DD) and a C-terminal domain which binds to RNA-associated N protein (PCTD:P-N^ARN). Phosphorylation sites in position 162, 210 and 271 are indicated. d, e Identification of the P domains involved in inclusion bodies formation. BSR-T7/5 cells were co-transfected with plasmids pTit-N and the indicated construction of pTit-P. N was revealed with a mouse monoclonal anti-N antibody followed by incubation with Alexa-488 donkey anti-mouse IgG and P was revealed with a rabbit polyclonal anti-P antibody followed by incubation with Alexa-568 donkey anti-rabbit IgG. DAPI was used to stain the nuclei. Scale bars correspond to 15 µm

Fig. 7

A model for the dynamics of RNPs and NBs in RABV infected cells treated (+NCZ) or not (−NCZ) by Nocodazole. 1. and 2. The initial NB is formed around an incoming RNP. 3. RNPs are ejected from NB by a process which is microtubule independent and are transported away from the initial NB along the microtubule network (−NCZ) or remain in the vicinity of the initial NB when the cells are treated by Nocodazole (+NCZ). 4. In untreated cells (–NCZ), the newly formed RNPs can give rise either to new virions upon budding at the cell membrane or to new viral factories which form NBs of intermediate size. In Nocodazole-treated cells (+NCZ), the newly formed viral factories are located in the vicinity and rapidly fuse with the initial NB which then becomes much larger