RNA secondary structure

The path from a DNA template to a protein is via the so-called transcription and translation. During transcription, an RNA molecule is derived from the DNA template.

In eukaryotes, internal, non-coding parts, so-called introns, are removed by a process called splicing. Finally, a poly-A tail is appended to the 3′ end of the mRNA, and a CAP region at the 5′ end.

Once these processes are finished in the cell’s nucleus, the RNA – which subsequently is namd messenger or mRNA – is exported into the cytoplasm. The mRNA is single stranded. The free bases are therefore capable of forming bonds between adenosine and uracil, guanosince and cytidine and uracil, and guanosine residues, thus leading to a complex, non-linear structure.

Note that virtually all RNA single strands form such secondary structures. Some RNA molecules have evolved to maintain a very specific structure, such as tRNA. Others have a more or less purposeless structure.

Knowledge about the secondary structure can serve two purposes: First it can help to infer or understand the function of an RNA molecule. Second, it can predict whether an mRNA will cause problems during translation: Extended secondary structures can interfere with the translation process at the ribosome.

Typical structures

RNA forms a number of typical structures. The most prominent is the so-called hairpin: Two stretches of complementary nucleotides form a base-paired double helix, ending in a small loop of free, unpaired nucleotides.

The hairpin stem is the double helical part, the hairpin loop is the circular ‘end’. In the example above, the hairpin contains two unpaired nucleotides, a so-called ‘bulge’.

Predicting RNA secondary structure

Predictions of RNA secondary structures are based on a model of the base interactions. Theoretically, each potential interaction between any two nucleotides must be considered. To accelerate the prediction, certain heuristics about base interactions are applied, such as steric limitations of hairpin loops. In addition, the thermodynamic efficiency of different base pairings are usually considered, in order to get a realistic model of secondary structure formation.

For performance reasons, prediction algorithms usually exclude pseudknots. These are structures where two parts of the RNA form a hairpin stem, and another two parts form a second hairpin stem – one of them being located in between the first two parts, the other outside.

Note that the prediction of any RNA secondary structure is just an approximation, using a number of simplifying assumptions. In addition, experimental conditions for RNA folding such as ion concentration or temperature are difficult to control to a degree where absolute reproducibility is possible. Depending on the length and nucleotide sequence of an RNA molecule, the prediction may be close to a structure found in nature or very far off.