The Unpaved Road of Non-Coding RNA Structure-Function Relationships: Current Knowledge, Available Methodologies, and Future Trends

Abstract

The genomes from complex eukaryotes are enriched in non-coding genes whose transcription products (non-coding RNAs) are involved in the regulation of genomic output at different levels. Non-coding RNA action is predominantly driven by sequence and structural motifs that interact with specific functional partners. Despite the exponential growth in primary RNA sequence data facilitated by next-generation sequencing studies, the availability of tridimensional RNA data is comparatively more limited. The subjacent reasons for this relative lack of information regarding RNA structure are related to the specific chemical nature of RNA molecules and the limitations of the currently available methods for structural characterization of biomolecules. In this review, we describe and analyze the different structural motifs involved in non-coding RNA function and the wet-lab and computational methods used to characterize their structure-function relationships, highlighting the current need for detailed structural studies to explore the molecular determinants of non-coding RNA function.

Keywords: X-ray crystallography; chemical probing; cryo-EM; non-coding RNA; structure–function relationships.

Figure 5

Selected examples of experimentally determined tertiary structures of ncRNAs or complexes of ncRNAs with proteins using X-ray crystallography and cryo-EM methods. (A) Human memRNA; (B) human telomere associated RNA; (C) human 7SK RNA; (D) human pre-mascRNA; (E) complex between lncRNA Rmrp and an RNA-binding protein; (F) human 7SK core ribonucleoprotein; (G) glutamine-tRNA in complex with PUS protein; (H) lncRNA Rmrp in complex with TLR; (I) human spliceosome; (J) human pre-miRNA complex with Dicer endonuclease; (K) human ribonuclease P complex; and (L) Tetrahymena telomerase. All the figures were prepared from the atomic coordinate data extracted from the PDB database and the 3D Protein Imaging application [100].

Figure 1

Different organization levels of RNA structure and their representations, exemplified by the human menRNA. (A) Primary structure of memRNA; (B) secondary structure representation of memRNA using the “dot-bracket” notation; (C) secondary structure of memRNA using the arc representation; (D) secondary structure of memRNA using the 2D diagram format; (E) tridimensional structure of memRNA using coordinate data extracted from the experimentally determined structure (PDB code: 8VT5).

Figure 2

Selected RNA elements generated by secondary and tertiary structure arrangements with relevance in the biogenesis and function of ncRNAs. (A) Kissing hairpins; (B) combined RNA segment that contains a stem loop, and internal loop, a bulge, a three-stem junction and a tetraloop; and (C) pseudoknot.

Figure 3

Time course of the deposition statistics in the PDB database showing the number of released entries by year for protein–nucleic acid complexes, and for DNA and RNA molecules (source: PDB statistics).

Figure 4

Structural organization levels of RNA molecules, and protocols used for their characterization, in vitro, in vivo, and in silico. Note that the different protocols and methods are highly connected, and in many cases are used together as synergistic approaches.

Figure 6

Secondary structures of (A), HOTAIR [108] and (B), MEG3 [56] lncRNAs as determined by chemical probing followed by next-generation sequencing. Sequences and secondary structures were retrieved from the lncRNA-folding repository [109]. Secondary structure maps were generated by the RNArtist software v.1.0 (https://github.com/fjossinet/RNArtist, accessed on 31 January 2025).

Figure 7

Examples of secondary structure predictions applied to the study of the influence of sequence variants in the functions of ncRNAs. (A) Fernández and coworkers used a combination of in silico prediction techniques with experimental validation to demonstrate the differences in the processing efficiency of miRNA precursors induced by single point mutations [123]. This depiction shows the computer predictions of the secondary structure of the human pre-miR-30c canonical precursor (pre-miR-30c) compared to the A52G mutant (pre-miR-30c-G/A, mutation position depicted by an arrow). In the mutated precursor, the base of the loop presents an open structure that exposes the CNNC canonical sequence for SRSF3 binding, ensuring a more efficient processing by the nuclear DGCR8-Drosha complex. (B) Detailed analysis of the computer predictions of miRNA–mRNA hybrids established between let-7 miRNA and lin-41 3′-UTR, showing the differences in the estimated Gibbs function in two miRNA targets according to the presence of sequence bulges generated by sequence variants [124].