Rotavirus RNA chaperone mediates global transcriptome-wide increase in RNA backbone flexibility

Abstract

Due to genome segmentation, rotaviruses must co-package eleven distinct genomic RNAs. The packaging is mediated by virus-encoded RNA chaperones, such as the rotavirus NSP2 protein. While the activities of distinct RNA chaperones are well studied on smaller RNAs, little is known about their global effect on the entire viral transcriptome. Here, we used Selective 2'-hydroxyl Acylation Analyzed by Primer Extension and Mutational Profiling (SHAPE-MaP) to examine the secondary structure of the rotavirus transcriptome in the presence of increasing amounts of NSP2. SHAPE-MaP data reveals that despite the well-documented helix-unwinding activity of NSP2 in vitro, its incubation with cognate rotavirus transcripts does not induce a significant change in the SHAPE reactivities. However, a quantitative analysis of mutation rates measured by mutational profiling reveals a global 5-fold rate increase in the presence of NSP2. We demonstrate that the normalization procedure used in deriving SHAPE reactivities from mutation rates can mask an important global effect of an RNA chaperone. Analysis of the mutation rates reveals a larger effect on stems rather than loops. Together, these data provide the first experimentally derived secondary structure model of the rotavirus transcriptome and reveal that NSP2 acts by globally increasing RNA backbone flexibility in a concentration-dependent manner.

Figure 1.

The Rotavirus transcriptome exhibits a diverse ensemble of RNA structures. (A) Median SHAPE reactivity of each of the 11 transcripts representing the rotavirus transcriptome averaged over two highly correlated (R² ∼ 0.73–0.98) replicates. SHAPE-MaP experiments were carried out by incubating equimolar amounts of each RV segment transcript and probed with 5NIA, as described in Materials and Methods. A 50-nucleotide window was used to calculate median SHAPE values along each transcript and normalized to the overall median of the individual transcript. Areas of high median SHAPE (i.e. above zero, shown in green) represent conformationally flexible RNA regions. Areas of low median SHAPE (i.e. below zero, shown in blue) represent regions with high propensity to base-pair. Arc plots show pseudo-free energy derived base-pairing probabilities computed using nearest-neighbor RNAStructure parameters constrained by incorporating SHAPE reactivities (18,28). Probabilities of base-pairing are color-coded, with green arcs corresponding to the most probable base-pairs, and grey being the least probable ones. Boxed regions indicate conserved nucleotides. Conformational entropy plots (median Shannon entropy calculated with a 50-nucleotide sliding window) computed for RV transcripts are plotted underneath median SHAPE values. High Shannon entropy regions typically correspond to positions with alternative conformations, while low entropy regions identify sequences that are more prone to adopting a dominant MFE configuration.

Figure 2.

SHAPE-MaP derived models support covariation-based secondary structure models. (A) Median SHAPE reactivity of segment 6 in the presence of the RV transcriptome. Median SHAPE profile and base-pairing probabilities were calculated as described in Figure 1. Grey box indicates two terminal stem-loops that were predicted in segment 6 by Li et al. (5), and are further confirmed by our SHAPE-informed secondary structure model. Beneath the median SHAPE profile is a conservation score heatmap plot calculated using a 10-nt rolling window as described in Materials and Methods. (B) Median SHAPE-MaP profile and base-pairing probabilities of segment 10 in the presence of the RV transcriptome. Median SHAPE profile and base-pairing probabilities were calculated as described in Figure 1. Grey box highlights three stem-loops modelled in the 3′ terminal region of segment 10, as previously predicted (5). Additionally, these three loops are located in a highly-conserved region of segment 10 as indicated by the high conservation score. (C) SHAPE-MaP profile of segment 11 RNA shown along with the conservation heatmap plot (left), and the corresponding models of conserved RNA helices H2 and H3 (right) supported by the SHAPE-MaP data. Highly conserved nucleotides were previously predicted to participate in LRIs between terminal regions and our SHAPE-informed MFE model validates these predictions (5).

Figure 3.

SHAPE-MaP analysis of in vitro transcribed segment 11 RNA versus in-cell transcript probing during RV infection. (A) Top panel: SHAPE reactivities of segment 11 transcript probed with 5NIA along with RV transcriptome. Bottom panel: in-cell SHAPE reactivity profile of segment 11 RNA probed with 5NIA at 6 hours post infection, as described in Materials and Methods. Individual bars represent SHAPE reactivities at single nucleotide resolution, black bars indicate nucleotides with the lowest reactivities (<0.4) and red bars denote nucleotides with the highest reactivities (>0.8). (B) Overlaid arc diagrams show similarities and differences between base-pairing in segment 11 RNA in the context of the complete RV transcriptome in vitro (Top) versus in the context of RV-infected cells (Bottom). Minimum free energy arc diagrams were generated by incorporating SHAPE-MaP data as a pseudo-free energy term in RNAStructure, as described in Figure 1. (C) Left: Scatterplot correlation analysis reveals that segment 11 transcript SHAPE reactivities between the in cell and in vitro transcribed RNA conditions are highly correlated (R²= 0.8). Right: Violin plots showing distribution of SHAPE reactivities of segment 11 RNA when probed in cell or in vitro. Boxes represent the 25th/75th interquartile range, and medians are shown as central bands. SHAPE reactivities are not significantly different as assessed by Kruskal-Wallis test (P < 0.05).

Figure 4.

NSP2 addition does not result in detectable changes in the median SHAPE reactivities of segment 11 RNA. (A) Segment 11 was incubated with increasing amounts of NSP2 (0, 5, 10 and 20 μM) then probed with 5NIA, as described in the Methods section. SHAPE reactivities for each of the listed titration points show no obvious difference due to the normalization protocol. (B) Scatterplots showing the linear relationship between segment 11 SHAPE reactivity (x axis) and segment 11 SHAPE reactivity when incubated with increasing amounts of NSP2 (y-axis). Note high correlation between the data (R² = 0.96, R² = 0.78), and the linear relationship exhibits no significant change between the titration points. (C) Violin plots comparing distribution of SHAPE reactivities of segment 11 RNA upon incubation with NSP2. Boxes represent the 25th/75th interquartile range, and medians are shown as central bands. Addition of 20 μM NSP2 changes the distribution of SHAPE reactivities, while the median SHAPE reactivity does not significantly change, as assessed by the Kruskal-Wallis test (P < 0.05). Note there is no significant difference between the conditions despite a significant global increase in mutation rate. (D) Raw mutation rates of segment 11 RNA structure probing experiment. The red line is representative of segment 11 RNA mutation rates induced by the incubation with 12.5 mM of the modifying reagent, 5-nitroisatoic anhydride (5NIA), and the blue is the measured mutation rate across the segment 11 transcript treated with dimethyl sulfoxide (DMSO) as a control. Normalization of these mutation rates results in the SHAPE profiles shown in panel A. (E) Histogram representation of distribution of nucleotide mutation rates in treated and untreated samples of segment 11 RNA shown in panel D. (F) Scatterplot showing linear relationship of mutation rate data between reproducible replicate experiments. The slope when comparing both replicates is near 1, showing that mutation rate data are highly correlated (R² = 0.74, slope = 1.08 ± 0.03). (G) Scatterplot showing linear relationship of SHAPE-MaP data between reproducible replicate experiments. SHAPE-MaP data are highly correlated (R² = 0.93, slope = 0.94 ± 0.02).

Figure 5.

NSP2 uniformly increases segment 11 RNA mutation rate in a concentration- dependent manner. (A) Segment 11 was incubated with increasing amounts of NSP2 (0, 5, 10 and 20 μM), then probed with 5NIA, as described in Materials and Methods. Higher mutation rates reveal increased flexibility in the RNA backbone, while lower mutation rates indicate less conformationally flexible nucleotides. Note a uniform increase in mutation rates proportional to NSP2 concentration. Zoomed-in: representative regions of segment 11 RNA. (B) Violin plots comparing distribution of mutation rates of segment 11 RNA upon incubation with NSP2. Boxes represent the 25th/75th interquartile range, and medians are shown as central bands. Note, both change in mutation rate distribution and significant change in median mutation rate. Significance values were calculated using Kruskal–Wallis test (P < 0.05). (C) Mutation rates of segment 11 RNA (x-axis) plotted against mutation rates of segment 11 RNA in the presence of increasing amounts of NSP2 to reveal linear relationship with distinct slopes. The slope increase as NSP2 is titrated in, while the linear dependence persists, suggesting a global change in the RNA flexibility in response to saturation with NSP2.

Figure 6.

NSP2 uniformly increases backbone flexibility in RV RNAs. (A–C) Mutation rates of individual RNAs (segment 5, segment 6 and segment 10, respectively) alone (x axis) plotted against the mutation rates of the RNAs incubated with 20 μM NSP2 (y axis). (D) Schematic of structural motifs found within an RNA. Using SHAPE-MaP-informed secondary structure models generated as shown in Figure 1, individual nucleotides (color-coded) were categorized into each of the different structural modalities described: interior bulge (red), multiloop (yellow), stem (green), and hairpin loop (orange) using the Forgi library (Materials and Methods). (E) NSP2-mediated mutation rate change analysis for individual structural motifs as described in panel D. Log₂(FC) denotes log₂(mutation rates induced by 20 μM NSP2 divided by the mutation rates without NSP2). A vast majority of log₂(FC) values are positive. Statistical significance was assessed using Kruskal-Wallis test. Stem-forming nucleotides involved in base-pairing have a higher average log₂(FC) than those located within predicted forming hairpins, multiloops and interior bulges.

Figure 7.

Changes in global backbone flexibility in RV transcriptome mediated by NSP2. Addition of NSP2 to an equimolar mix of 11 distinct RV transcripts results in global increase of mutation rates across all segments. (A–D) Mutation rates of individual RNAs incubated with an RNA mix representing the RV transcriptome in the presence of 20 μM NSP2(y-axis) plotted against mutation rates of the RNA alone in the presence of 20 μM of NSP2 (x-axis) to reveal highly correlated data. (E) Analysis of NSP2-mediated mutation rate changes in the presence or absence of the rotavirus transcriptome. Comparison of distribution of NSP2 mediated mutation rate changes (i.e. the log₂ of the quotient of the mutation rates in the presence of 20 μM NSP2 over the mutation rates for RNA without NSP2) for each the structural motif (stem, hairpin loops, multiloops, interior bulge, as described in Figure 6D) for RNAs alone and RNAs in the context of the transcriptome. Q1 and Q3 represent the 25th and 75th interquartile range, respectively, and the median, Q2, is labeled shown as solid lines. Statistical significance was assessed using Kruskal-Wallis test.