Folding and finding RNA secondary structure

Abstract

Optimal exploitation of the expanding database of sequences requires rapid finding and folding of RNAs. Methods are reviewed that automate folding and discovery of RNAs with algorithms that couple thermodynamics with chemical mapping, NMR, and/or sequence comparison. New functional noncoding RNAs in genome sequences can be found by combining sequence comparison with the assumption that functional noncoding RNAs will have more favorable folding free energies than other RNAs. When a new RNA is discovered, experiments and sequence comparison can restrict folding space so that secondary structure can be rapidly determined with the help of predicted free energies. In turn, secondary structure restricts folding in three dimensions, which allows modeling of three-dimensional structure. An example from a domain of a retrotransposon is described. Discovery of new RNAs and their structures will provide insights into evolution, biology, and design of therapeutics. Applications to studies of evolution are also reviewed.

Figure 1.

Folding constraints and restraints. Traditional chemical agents that act on bases are applied as folding constraints, i.e., a base accessible to chemical modification cannot be in a base pair flanked by Watson-Crick pairs on each side. In RNAstructure, this is implemented by assigning a large positive free energy to any conformation that violates the constraint. SHAPE reactivity is applied as a folding restraint, i.e., a free energy change bonus or penalty for pairing of a nucleotide. For nucleotides with low SHAPE reactivity, a pairing stabilization is provided and for high reactivity, a pairing penalty is provided. The SHAPE restraint is provided per nucleotide in a base pair stack. Therefore, the free energy change is applied twice per nucleotide buried in a helix and once per nucleotide in a pair at the end of a helix (Deigan et al. 2009).

Figure 2.

Determination of Structured Regions of an RNA. A cartoon of the 5′ region of the silk moth R2 retrotransposon is shown. The conserved structure is organized into four hairpin loops (labeled I and III–V) and a pseudoknot (labeled II). Also shown are three conserved coding regions (A–C) and a putative open reading frame (ORF) start site. The five conserved structures are detailed with data that went into the structural modeling. The sequences shown are for B. mori whereas mutations are those that occur in four other moth species. Mutational data appear next to the main sequence and is color annotated: dark blue are double mutations that maintain base pairing (compensatory), light blue are single point mutations that maintain pairing (consistent), gray are mutations in loops, red disrupt canonical base pairs (inconsistent), green are insertions (green X represents a deletion). Experimental mapping is color annotated on the backbone sequence: red are NMIA only modifications and orange are modifications by both traditional mapping agents (DMS or CMCT) and NMIA. Base pairs are indicated with dashes between nucleotides and are color annotated for probability from partition function calculation: Red, probability (P) ≥99%; Orange, 99% > P ≥ 95%; Yellow, 95% > P ≥ 90%; Dark Green, 90% > P ≥ 80%; Light Green, 80% > P ≥ 70%; Light Blue, 70% > P ≥ 60%; Dark Blue, 60% > P ≥ 50%; Black <50%. Many base pairs in the pseudoknot have low probability because the RNAstructure program does not allow pseudoknots and thus, under-counts them in the partition function.

Figure 3.

Experiment and Sequence Comparison are Used to Model 3D Structure. Homology with known structures was used to propose 3D folds for the B. mori R2 element pseudoknot from MC-Sym (Parisien and Major 2008) which were then screened with respect to experimental data (e.g. solvent accessibility to chemical reagents [Kierzek 2009] and helix stacking from NMR [Hart et al. 2008]). Helical motifs in the 3D model are color coded to match the secondary structural model.