Modelling the three-dimensional structure of the right-terminal domain of pospiviroids

Abstract

Viroids, the smallest know plant pathogens, consist solely of a circular, single-stranded, non-coding RNA. Thus for all of their biological functions, like replication, processing, and transport, they have to present sequence or structural features to exploit host proteins. Viroid binding protein 1 (Virp1) is indispensable for replication of pospiviroids, the largest genus of viroids, in a host plant as well as in protoplasts. Virp1 is known to bind at two sites in the terminal right (TR) domain of pospiviroids; each site consists of a purine- (R-) and a pyrimidine- (Y-)rich motif that are partially base-paired to each other. Here we model the important structural features of the domain and show that it contains an internal loop of two Y · Y cis Watson-Crick/Watson-Crick (cWW) pairs, an asymmetric internal loop including a cWW and a trans Watson/Hoogsteen pair, and a thermodynamically quite stable hairpin loop with several stacking interactions. These features are discussed in connection to the known biological functions of the TR domain.

Figure 1

Secondary structure of pospiviroids. The consensus sequence and structure presented were predicted for alignments with MAFFT, optimized in ConStruct at 37 °C, excluding lonely base pairs, and drawn with R2R. (a) Structure of circular PSTVd. Borders of the five domains (TL, terminal left; P, pathogenicity-related; C, central; V, variable; TR, terminal right) are marked by gray lines. Loops 6, 7, and 15 (E) are marked. RY motifs critical for binding of Virp1 are outlined and marked by R and Y. (b) TR domain based on an alignment of 234 full-length unique PSTVd sequences. High and low SHAPE reactivity is marked by filled and open triangles, respectively [red; orange, blue and green]; reactivity with RNase S1 (black triangle), RNase T1 (open triangle), and dimethyl sulfate (DMS, circle) is marked. (c) Terminal right stem-loop based on an alignment (see Supplementary Fig. S2) of unique TR sequences of 95 pospiviroids (14 PSTVd, 25 CEVd, 21 columnea latent viroid (CLVd), 8 chrysanthemum stunt viroid (CSVd), 3 iresine viroid (IrVd), 3 Mexican papita viroid (MPVd), 10 pepper chat fruit viroid (PCFVd), 4 tomato apical stunt viroid (TASVd), 7 tomato chlorotic dwarf viroid (TCDVd)). The internal loops are marked by IL1 and IL2, respectively, and the hairpin loop by HP. The color code used in (b) and (c) for the annotation of nucleotides, base pairs and mapping is given in the right box.

Figure 2

Structure logo of terminal right stem-loop. This logo is based on the same alignment (see Supplementary Fig. S2) of 95 pospiviroids as used for Fig. 1c. The height of nucleotide characters is proportional to the observed frequency of nucleotides in the alignment and the expected (a priori) frequency p = 0.25; nucleotide characters that appear less than expected are displayed up-side-down. The mutual information (marked by ‘M’) between base-paired positions, as indicated in the two bottom lines, is calculated as the relative entropy between the fractions of complementary bases and the number of basepairs one would expect by chance from the distribution of nucleotides at the involved positions. The nucleotide numbering is as in Fig. 1c; missing numbers denote that the respective column of the alignment contains mostly gaps. The RY motif critical for binding of Virp1 is outlined and marked by R and Y. The sequences of internal loops are marked by IL1 and IL2, respectively, and the hairpin loop sequence by HP.

Figure 3

Results of Rosetta modelling of IL1 variants. Results of Rosetta modelling are shown for three typical loop variants: CC/UU (a), CA/UU (b), and CU/UU (c). Top: secondary structure of IL1 used as input to Rosetta modelling; that is, the input structure was ((..((+))..)) for all three sequences. Middle: interactions identified by rnaview in Rosetta’s top model; standard cWW base pairs are marked by double lines; all “loop” nucleotides are paired in cWW conformation; for the basepairing nomenclature of Leontis & Westhof see Supplementary Fig. S1. Bottom: Logos produced by RNAredesign for Rosetta’s top model.

Figure 4

Results of Rosetta modelling of IL2 loop. (a) Interactions identified by rnaview in Rosetta’s top model. Loop nucleotides were highly connected by a complex network of base/base hydrogen bonds and stacking interactions. (b) Triple pair C₁₈₇ · A₁₇₂:U₁₈₅ (cWW A₁₇₂:U₁₈₅, trans Watson/Sugar edge (tWS) C₁₈₇ · A₁₇₂). (c) Logo produced by RNAredesign for Rosetta’s top model.

Figure 5

Results of Rosetta modelling of HP loop. Results for a typical loop of most pospiviroids is shown in (a); an exceptional loop of TCDVd is shown in (b). Top: Interactions identified by rnaview and x3dna-dssr. Middle: 3D view produced by PyMOL. Bottom: Logos produced by RNAredesign.

Figure 6

A motif from Thermus thermophilus 16S ribosomal RNA. (a) Nucleotides 932–936 and 1379–1385 are similar in sequence to IL2 of PSTVd; interactions are extracted by x3dna-dssr from PDB 1FJG. (b) A logo of the motif (a) from an alignment of 23,500 bacterial 16S ribosomal RNAs. (c) A logo of the motif (a) for 1850 sequences from the same alignment as used in (b), but sequences identical to the consensus 5′ CGCAC/GUUCCCG 3′ were removed. Columns with less then 1% of nucleotides and sequences not fully spanning the motif or containing nucleotides other then [AUCG] were removed from the alignments used for the logos.

Figure 7

Histone mRNA stem-loop. This hairpin loop including the closing base pair is identical in sequence to PSTVd’s HP (Fig. 5a). (a) Comparison of hairpin loops of PDB entries 1JU7 (red), 1KKS (cyan), and 4QOZ (gold); model 5 of 1JU7 and model 8 of 1KKS are shown, which are annotated as best structures in their respective PDB files. Views from different angles are shown in Supplementary Fig. S4. (b) Hairpin loop (cartoon in gold) from 4QOZ, bound by Exo (cartoon in lightpink) and SLBP (cartoon in darkgreen); a few amino acid side chains, which interact with RNA backbone and bases, are labelled. (c) Logo of 11,099 histone mRNA loops from Rfam-11 (ID: Histone3); three sequences have an additional nucleotide after position 6. (d) Logo of 36 non-redundant histone mRNA loops from Rfam-11.

Figure 8

View on kissing-loop interaction. The nucleotides 1319–1329 and 1362–1370 from PDB entry 3J3W are visualized by rnaview (left) and by RasMol (right). The hairpin loop 1363–1369 is identical in sequence to the standard hairpin loop of CEVd.