Principles of RNA recruitment to viral ribonucleoprotein condensates in a segmented dsRNA virus

Figures

• 

  Figure 1.. RNA ccumulation during Rotavirus infection. (a) Experimental design for detecting EGFP-tagged viroplasms during early infection (3–5 hpi) stages. (b) Schematics of the FISH probe design for detecting all 11 RV transcripts. A pooled set of 3–4 segment-specific probes labelled with Quasar570 dye were used for visualising RV transcripts in EGFP-tagged viroplasms. (c) RV RNA accumulation in viral factories during early infection. Top: RV transcripts (yellow) visualized by smFISH using a combined set of 3–5 segment-specific probes targeting each genomic transcript. Bottom: NSP5-EGFP-tagged viral factories (green). Images represent maximum intensity Z-projections acquired using identical settings and brightness levels for both channels. DAPI-stained nuclei are shown in blue. Scale bars, 25 µm. (d) RV transcripts and NSP5-EGFP-specific signals shown in the white box in panel (c). Note the low intensity of the RNA and EGFP signals during early infection stages that increase by 5 hpi. Scale bar, 5 µm. (e) Pearson’s correlation coefficients (PCCs) of colocalising NSP5-EGFP-tagged viral factories and RV transcripts, as shown above in panel (d). Dashed and dotted lines represent median and…

• 

  Figure 1—figure supplement 1.. smFISH of GAPDH mRNA and RV transcripts. GAPDH mRNA red, (a) and RV transcripts yellow, (b) imaged using two distinct sets of FISH probes targeting GAPDH mRNA (Quasar 670 dye-labelled probes) and RV transcripts (Quasar 570 dye-labelled probes), as shown in Figure 1a. At 4 hpi, RV transcripts form distinct, high intensity foci. Scale bars, 25 μm. (c) A close-up view of the highlighted area (i). Scale bar, 10 μm. (d) Signal intensity distribution histograms for GAPDH transcripts and RV transcript foci (rvRNA).

• 

  Figure 1—figure supplement 2.. Viroplasms accumulate RV transcripts but not polyadenylated mRNAs. FISH visualisation of polyadenylated transcripts (oligo (dT)30-ATTO647N probes, see Methods, red signal) and RV transcripts (Seg4, yellow) at 4 hpi. NSP5-EGFP-labelled viroplasms contain RV transcripts, while non-viral polyadenylated RNAs seemingly do not partition to viroplasms. Scale bar, 10 μm.

• 

  Figure 2.. Segment-specific transcript accumulation during Rotavirus infection. (a) smFISH of RV transcripts at early infection time points (2–6 hpi) using segment-specific probes. Two sets of 48 gene-specific probes were designed for each Seg3 (magenta, Quasar570) and Seg4 (cyan, Quasar670) transcripts. Top: NSP5-EGFP-tagged viral factories (green), nuclei (blue). Middle: Seg3 (magenta) and Seg4 (cyan) RNA signals in the ROIs shown above, with colocalising Seg3 and Seg4 RNAs (white). As the amount of each transcript increases during the course of infection, brightness settings were adjusted accordingly to reveal single transcripts (Brightness x100 for 2–4 hpi and 1 x for 6 hpi). Scale bars, 25 µm and 5 µm (zoom-in). (b) Colocalisation of Seg3 and Seg4 transcripts. Pearson’s correlation coefficients (PCCs) of colocalising Seg3 and Seg4 transcripts. Dashed and dotted lines represent median and quartile values of PCCs, respectively. Each data point represents the PCC value calculated for a single cell. N=37 (2 hpi), N=62 (4 hpi), N=51 (6 hpi). Statistical analysis of data was performed using Kolmogorov-Smirnov test. At the 0.01 level, the observed distributions are significantly different.…

• 

  Figure 2—figure supplement 1.. Analysis of Seg3 and Seg4 transcript accumulation and localisation in RV-infected cells. (a) Signal intensity distributions of Seg3 and Seg4 transcript foci at 2 hpi and 4 hpi. Raw signal intensity distributions (counts) of smFISH-detected spots were derived for Seg3 and Seg4 transcripts at 2 and 4 hpi. Each spot was detected after applying a constrained deconvolution algorithm (see Methods). (b) Changes in the integrated signal intensities for Seg3 (magenta), Seg4 (cyan) RNAs and EGFP-tagged viroplasms (green) over the course of RV infection. Counts were calculated for single cells separately, N=24 (2 hpi), N=21 (4 hpi), N=30 (6 hpi). (c) Colocalisation of EGFP-tagged viroplasms with Seg3 (magenta) and Seg4 (cyan) transcripts over the course of RV infection. Median and quartile values of the calculated Pearson’s correlation coefficients (PCCs) are shown as dashed and dotted lines, respectively. Each data point represents the PCC value calculated for a single cell. N=37 (2 hpi), N=62 (4 hpi), N=51 (6 hpi). (d) Correlation between Seg4 RNA and viroplasmic EGFP signals vs Seg3 RNA signals at 6 hpi. Integrated intensities per pixel for Seg4 RNA (blue) and NSP5-EGFP (green) are…

• 

  Figure 2—figure supplement 2.. NSP2 knockdown in MA104 cells results in the apparent loss of RV RNA clustering. (a) Wild-type MA104 cells infected with RV (MOI = 1) at 6 hpi, with a schematic showing Seg3 and Seg4 RNA clustering. (b) MA104 cells expressing shRNAs that target NSP2 (Methods) were infected with RVs as described in panel (a). Note that in both cases single Seg3 and Seg4 transcripts accumulate in the cytosol, while the signal intensity of RNAs in case of NSP2 knockdown is lower. Scale bars, 50 μm (left), 10 μm (middle and right panel). (c) Violin plots showing distributions of integrated intensities per pixel for Seg4 RNA (blue) and Seg3 RNA (magenta) from images shown in (a) and (b). Median and quartile values are shown as dashed and dotted lines, respectively. Depletion of NSP2 causes a significant decrease of the overall RNA signal intensity (P<0.0001 assessed by an unpaired t-test with Welch’s correction for both Seg 3 and Seg4 RNA signals). Note that the median value for Seg3 signal remains broadly unchanged upon NSP2 depletion. (d) Western blot of NSP2 from RV-infected cells (WT MA104 and NSP2-specific shRNA-expressing MA104 cells). MA104 cells stably expressing shRNA targetting NSP2…

• 

  Figure 3.. DNA-PAINT analysis of the RV RNA oligomerisation. (a) Schematics of DNA-PAINT imaging of RV transcripts. RNA hybridisation is used to install unique DNA ‘docking’ P1 strands for Seg3 RNA targets. These P1 sequences bind to complementary Cy3B-labelled DNA ‘imager’ strands (denoted P1*). Transient binding of P1* to accessible P1 sites on Seg3 RNAs is used for stochastic super-resolution reconstruction of the RNA targets. Only specific binding events (i.e. fully complementary DNA imager-docking strand) give rise to characteristic ‘on/off’ times, allowing discrimination from non-specific binding events. (b) DNA-PAINT super-resolution reconstruction of Seg3 transcripts at 2, 4, and 6 hpi. (c) Zoomed-in regions of Seg3 RNA structures highlighted in (b). (d) Quantitative PAINT (qPAINT) analysis of the RNA structures shown above. kon values were calculated for each selected structure assuming kon=(τD×c)-1 , where c is the imager strand concentration and τD the dark time between binding events (cimager = 1 nM at 2 hpi; 125 pM at 4 hpi and 100 pM at 6 hpi). Each point represents the apparent kon value for an RNA structure that is directly proportional to the relative number of the FISH…

• 

  Figure 4.. Universal DNA exchange-smFISH approach for single-cell imaging of the RV transcriptome. (a) Schematics of the Universal DNA Exchange-smFISH approach (UDEx-FISH). FISH probes targeting different RNA segments are extended with 12 nucleotide-long orthogonal DNA sequences A1-An (DNA ‘handles’). Cy3 and Cy5-dye-modified DNA sequences (A1–An, ‘Imagers’) complementary to their respective handles can rapidly (<1 min) and stably hybridise with their segment-specific FISH probes RNA. Non-hybridised imagers are washed away and the two RNA targets, as well as NSP5-EGFP tagged viroplasms and nuclei are imaged in each imaging round (see Methods), allowing the next imaging round to take place. The imaging procedure is repeated until all targets of interest are recorded without affecting the quality of the fixed sample or loss of the RNA signal intensity due to photobleaching. (b) RV Transcriptome during early infection stage (2 hpi). Images were taken with same acquisition parameters for each imaging round. Scale bar, 20 µm. (c) Zoomed-in images of the highlighted areas shown in (b). Scale bar, 5 µm. Figure 4—source data 1.The integrated signal intensity after each successive round of imager…

• 

  Figure 4—figure supplement 1.. Validation of the UDEx-FISH approach in RV-infected MA104 cells. (a) Washing and labelling efficiencies of smFISH probes with orthogonal DNA handles designed for Segments 1–9. After each wash step, no residual RNA signal was detected for each segment tested in this experiment. Seg2 target was re-imaged iteratively to ensure that multiple washing steps did not dissociate FISH probes pre-hybridized to a target RNA. (b) The integrated signal density after each successive round of imager dissociation and re-association. Each image plane for each channel was recorded with the same image acquisition parameters. Brightness levels were set identical for each colour channel for all the images shown. Scale bars, 15 µm.

• 

  Figure 5.. Single-cell analysis of the RV transcriptome. (a) In situ analysis of the RV transcriptome at 6 hpi. Scale bar, 20 µm. (b) Zoomed-in images of the highlighted regions containing viroplasms and 11 distinct RNA targets. Scale bar: 5 µm. (b) The proportion of RNA-seq reads assigned to the annotated protein-coding transcriptome of MA104 cells infected with RVA strain RF at 6 hpi. The RVA transcriptome represented ~17% of total reads (Methods), with the remaining 82.99% reads mapped onto MA104 protein-coding transcriptome. Bar plot shows the relative fractions of individual RV transcripts in the RV transcriptome at 6 hpi (Seg1-Seg11). (c) Single-cell analysis of the relative fractions of each segment-specific transcript (N=15 cells) estimated from the UDEx-FISH images data shown in panel (a). (d) Relative RV transcript abundance in viroplasms. Each data point represents the ratio of the viroplasmic signal density (per pixel) to the signal density of the corresponding cell. Higher values indicate stronger viroplasmic accumulation of viral transcripts. Lines, boxes, and whiskers represent the median, quartiles, and 5–95 percentile range of the data points, respectively. Relative…

• 

  Figure 5—figure supplement 1.. Quantification of transcripts by UDEx-FISH. (a) UDEx-FISH and RNA-Seq quantitation of RV transcripts at 6 hpi (MA104 cells, as described in Methods). Relative fractions of RV transcripts quantified by RNA-Seq (e.g. mean values shown in Figure 5b) are plotted as black horizontal bars, and violin plots show distribution of transcripts of Seg1-Seg11 in individual cells quantified by UDEx-FISH. Median and quartile values are shown as dashed and dotted lines, respectively. (b) Pairwise correlation plots of viroplasm-localised transcripts of segment 1 vs segments 2–11. Integrated intensities of detected Seg1 RNAs for individual viroplasms (y-axis) are plotted against the detected intensities for Seg 2–11 transcripts (x-axis). Correlations (R2 values) between the detected transcript signals were calculated by linear fitting to reveal different relative stoichiometries of distinct RV transcripts in viroplasms (N=21).

• 

  Figure 6.. 3’ UTR is required for RV transcript localisation to viroplasms. (a) Schematics of the EGFP transcript construct containing the EGFP-coding sequence flanked by the 5’UTR and 3’UTR sequences derived from Seg1 RNA (5’UTR-EGFP-3’UTR) or only 3’UTR sequence (denoted EGFP-Δ3’UTR) that lacks the 17-nt long 3’UTR. Both RNA constructs were electroporated into MA104-mCherry-NSP2 cells 6 hr prior to being infected with RVs, as described in Methods. (b) Widefield images of RV-infected MA104-mCherry-NSP2 cells electroporated with EGFP-Δ3’UTR (top) and the 5’UTR-EGFP-3’UTR transcript (bottom). Cells were fixed at 6 hpi prior to the hybridisation with FISH probes against EGFP coding region, as described in Methods. Green: NSP2-mCherry-tagged viroplasms, magenta: smFISH for EGFP RNA. Scale bars, 10 μm.