Long-range architecture in a viral RNA genome

Abstract

We have developed a model for the secondary structure of the 1058-nucleotide plus-strand RNA genome of the icosahedral satellite tobacco mosaic virus (STMV) using nucleotide-resolution SHAPE chemical probing of the viral RNA isolated from virions and within the virion, perturbation of interactions distant in the primary sequence, and atomic force microscopy. These data are consistent with long-range base pairing interactions and a three-domain genome architecture. The compact domains of the STMV RNA have dimensions of 10-45 nm. Each of the three domains corresponds to a specific functional component of the virus: The central domain corresponds to the coding sequence of the single (capsid) protein encoded by the virus, whereas the 5' and 3' untranslated domains span signals essential for translation and replication, respectively. This three-domain architecture is compatible with interactions between the capsid protein and short RNA helices previously visualized by crystallography. STMV is among the simplest of the icosahedral viruses but, nonetheless, has an RNA genome with a complex higher-order structure that likely reflects high information content and an evolutionary relationship between RNA domain structure and essential replicative functions.

Figure 1

Models for the organization of RNA helices in the STMV capsid and analysis of native RNA genome structure by SHAPE. (A) Crystallographic model of helices and capsid interactions in STMV (1a34). Three capsid dimers and their respective bound RNA helices are shown. (B, C) Schematic interpretations of the helices visualized by crystallography in terms of (B) linked stem-loop and (C) long-range base pairing models. (D) Approach for analyzing STMV genome structure under native-like conditions by SHAPE.

Figure 2

SHAPE reactivity profiles for ex virio (top) and in virio (middle) STMV RNAs. Difference plot of SHAPE reactivities (bottom). Positive and negative values indicate protection from versus enhanced SHAPE reactivity in virio.

Figure 3

Secondary structure models for the STMV RNA ex virio. (A) SHAPE-directed model. Maximum allowed base pairing distance was 600 nucleotides. The start and stop codons for the capsid protein are boxed. (B) Linked stem-loop model, created using SHAPE data and parameters designed to force formation of short stem-loop motifs by restricting the maximum base pairing distance to ≤50 nucleotides. Nucleotides are colored by SHAPE reactivity (see legend); gray indicates no data were obtained. Calculated lengths of major structural features in each structure are shown (in nanometers).

Figure 4

Analysis of long-range interactions in the STMV RNA genome by LNA-mediated structure disruption. (A) Schematic image showing sites of LNA binding (open boxes) and the resulting SHAPE-detected perturbation (heavy lines). Quantification of LNA-induced structure perturbations for (B) an LNA bound at positions 180-188 and (C) an LNA bound at positions 536-544. The largest observed changes for each LNA (≥ 0.15) are indicated by color.

Figure 5

AFM visualization of STMV RNA structure. (A) Volume distribution for all species in AFM images. The peak at 320 nm³ is consistent with the calculated molecular volume of a 1058-nt RNA. (B) Classification of single RNA molecules by structural features. Central chart shows the fraction of RNAs in each category. Observed species suggest a general unfolding from most condensed to extended conformations of three branches. (C) Lengths of observed features based on branch length and peak-to-peak distance. Feature lengths corresponding to peaks in each histogram are labeled explicitly. A single length was measured for the uniform height, two peaks, one branch, and variable height molecules; two and three lengths, respectively, were measured for the two and three branches molecules. No molecules had a length greater than 100 nm.

Figure 6

Model for the secondary structure and domain architecture of the STMV RNA genome derived from SHAPE measurements performed inside intact virions. Thirty representative helices (numbered in brackets) that constitute plausible sites of interaction with capsid protein dimers are emphasized with colored lines. Three regions with the largest differences relative to the ex virio structure are identified as a, b, and c. The start and stop codons for the capsid protein open reading frame are boxed. Nucleotides are colored by SHAPE reactivity using the scheme shown in Fig. 3. (inset) Schematic map of STMV RNA helices and connectivity superimposed onto icosahedral geometry. Each edge corresponds to an RNA-capsid interaction site.