Solution structure of the cap-independent translational enhancer and ribosome-binding element in the 3' UTR of turnip crinkle virus

Abstract

The 3(') untranslated region (3(') UTR) of turnip crinkle virus (TCV) genomic RNA contains a cap-independent translation element (CITE), which includes a ribosome-binding structural element (RBSE) that participates in recruitment of the large ribosomal subunit. In addition, a large symmetric loop in the RBSE plays a key role in coordinating the incompatible processes of viral translation and replication, which require enzyme progression in opposite directions on the viral template. To understand the structural basis for the large ribosomal subunit recruitment and the intricate interplay among different parts of the molecule, we determined the global structure of the 102-nt RBSE RNA using solution NMR and small-angle x-ray scattering. This RNA has many structural features that resemble those of a tRNA in solution. The hairpins H1 and H2, linked by a 7-nucleotide linker, form the upper part of RBSE and hairpin H3 is relatively independent from the rest of the structure and is accessible to interactions. This global structure provides insights into the three-dimensional layout for ribosome binding, which may serve as a structural basis for its involvement in recruitment of the large ribosomal subunit and the switch between viral translation and replication. The experimentally determined three-dimensional structure of a functional element in the 3(') UTR of an RNA from any organism has not been previously reported. The RBSE structure represents a prototype structure of a new class of RNA structural elements involved in viral translation/replication processes.

Fig. 1.

The TCV RBSE. (A) Model for cap-independent translation initiation in TCV. The model suggests that the RBSE functions to recruit or recycle 60S ribosomal subunits to the template that then accesses the 5^′ end possibly through interaction with prebound 40S subunits. (B) Schematic drawing of the genome organization of TCV and the secondary structure of the RBSE. The secondary structure of RBSE was verified by imino-NOE walks of this construct and a number of mutants (see SI Text). The hairpins are labeled as H1, H2 and H3, which consists of H3a, H3b and an internal loop (the nomenclature is simplified from previous nomenclature [(7) for this report]. The pseudoknot formed between the residues in the H2 loop and the residues at the 3^′ end was previously determined (10). The cis-acting sequences, external to the RBSE on both the 5^′ and 3^′ ends of the TCV 3^′ UTR are also shown. The italic numbers are those of the genome and smaller numbers are those for RBSE starting from position 1. (C) RBSE ribosome binding competition experiments. Deacylated tRNA^phe can compete with binding of RBSE to the P site of yeast ribosomes (Left), and RBSE does not compete with acylated Phe-tRNA^phe for binding to the A site of ribosomes (Right).

Fig. 2.

The molecular envelope of RBSE, RDC waves of RBSE and the TCV3M mutant. (A) The low-resolution molecular envelope of RBSE with dimension measurements. The envelope is an average of 16 bead models calculated from SAXS data using DAMMIN (34). The top short arm on the left of the envelope has a cylindrical shape, about 20 Å in diameter which is comparable in diameter to an A-form RNA duplex. The long arm of the envelope is bent with an angle of approximately 140° and is twisted outward. (B). The RDC waves of the three duplexes in RBSE with the solid curves calculated using parameters D _a, R, orientation (Θ,Φ), phase ρ ₀ from the simultaneous fit using the program ORIENT and the structural parameters of an A-form duplex. The experimental RDCs are shown as circles. Interestingly, duplex H3a is interrupted at the mismatch A61-G94 but the remaining residues whose imino RDCs are available are still in the A-form conformation as indicated by their imino RDC fits. (C). On the left is the secondary structural drawing of the TCV3M mutant that has an extended H3b (highlighted in red) and a point mutation at A61 to C61. This mutant allowed for a better determination of the angle between H3a and H3b. On the right are the RDC waves of the H3a and H3b. From the simultaneous fit, we obtained the angle between the axes of the H3a and H3b duplexes.

Fig. 3.

A two-dimensional topology drawing (Left) of RBSE, the initial structure (Center) generated with the G2G toolkit, and the structure after regularization that fixes the bond breaks (Right). The regularization was accomplished in Xplor-NIH (35). The orientations and phases in terms of (Θ,Φ,ρ ₀) of H1, H2 and H3a, obtained from the best simultaneous fit, are given in the figure. The angles between hairpins labeled on the left figure were calculated from the hairpin orientations (Θ,Φ) on the figure. The linker residues are represented with broken lines in the topology drawing (Left) and residue numbers are drawn on the regularized structure (Right). The angle between H3a and H3b was determined using the mutant TCV3M (Fig. 2 C). Angles between all pairs of duplexes are given in the text.

Fig. 4.

The ensemble of global structures of the RBSE determined using the G2G “top-down” method and SAXS and PDDF curves comparison. (a) The front (Center) and side (Left and Right) views of the superimposed RBSE backbone structures (top 50% lowest energy) overlaid with the molecular envelope in gray mesh (top) or itself (bottom). The linker between the H1 and H2 is located at a position that is similar to the variable-loop found in canonical tRNA. The region important for large subunit ribosome and RdRp binding is highlighted in red; sequences in LSL involved in canonical basepairing with 5^′- and 3^′ends are colored in green and magenta, respectively. (B) The correlation plot of the back-calculated RDCs based on the starting structure (Fig. 3, Right), where the orientation of the three duplexes were determined using the RDC-structural periodicity correlation. Only RDCs in the duplex regions (the same experimental RDC data as shown in Fig. 2 B) were used the correlation coefficient calculation. The correlation coefficient is approximately 0.97. (C) The correlation plot of the back-calculated RDCs based on the top 10% lowest G2G structures vs. the experimental RDCs. The correlation is near unit. (D) The comparison of experimental (circle) and back-calculated SAXS curves (red) based on the top 10% ensemble. The RMSD between experimental data and the back-calculated curves is 0.20 ± 0.01. RMSD is calculated based on the logarithm of the normalized [i.e., I(q = 0) = 1.0] SAXS intensities. (E) The comparison of PDDFs of the corresponding experimental SAXS (black) and back-calculated SAXS curves (red) in (C).