Introduction

RNA pseudoknots are RNA structures with diverse functions [1], usually formed by the interaction of the loop of an RNA stem-loop with nucleotides downstream of the stem. These are referred to as H-type pseudoknots. Alternatively, two RNA stem-loops can interact via their loops, a type of pseudoknot known as kissing stem-loops. Among their diverse functions, certain pseudoknots can induce ribosome pausing during translation. In viral mRNA molecules, this mechanism is often combined with a heptanucleotide sequence upstream of the pseudoknot to induce a translation frameshift, which is necessary for the proper expression of certain viral ORFs [2].

For our project, we decided to exploit the ribosome-pausing capabilities of viral RNA pseudoknots to delay the translation of genes transcribed in a polycistronic mRNA molecule. To achieve this, we placed the pseudoknots in between genes rather than within the ORFs of our composite parts.

BioBricks

We have designed a number of BioBricks containing one or more pseudoknots. The rationale of the design was the following: the sequences of the chosen pseudoknots were synthesised, including the 20 bases upstream and downstream of the pseudoknot, based on the original viral genomes. This is to give enough space in between pseudoknots, should one wish to string several pseudoknots together to further delay the translation of a gene. We also included the standard BioBrick prefix and suffix to the parts. Since the goal is to delay the ribosome before it reaches the next gene of the polycistronic mRNA, it is important to avoid creating any new ORFs in the pseudoknot BioBricks. Thus, when necessary, we removed any potential START codons from the parts in our design. In the cases where potential ATG codons existed within the pseudoknot structure, we inverted any of the three base pairs in which the ATG triplet participated in order to keep the new sequence as close to the original as possible. We then conducted a few RNA pseudoknot predictions to verify that the new sequences yielded similar structures. The criteria chosen were to keep the estimated free energy of the new pseudoknots as close as possible to the original, while keeping the same stems and loops as the reported structures. Pseudoknot predictions were achieved with DotKnot [3, 4], while structure visualisation was carried out with PseudoViewer [5].

DotKnot was able to predict the correct stems and loops as reported in [2] for the wild type sequences. One exception was the kissing stem-loops of HCV229E, where DotKnot was unable to predict the full length of the second stem, probably due to a series of mismatches reported in this stem. However, the interaction of the loops was accurately predicted. This increased our confidence in the prediction of the new structures after removal of the START codons.
We chose to synthesise four known pseudoknots to test which, alone or in combination, could cause sufficient delay of the ribosome: three H-type pseudoknots, from the Mouse Mammary Tumor Virus (MMTV), the Infectious Bronchitis Virus (IBV) and the Transmissible Gastroenteritis Virus (TGV); and one kissing stem-loops type, from the Human Coronavirus 229E (HCV229E). These pseudoknots vary in length (34 nucleotides in the MMTV pseudoknot to over 200 nucleotides in the kissing stem-loops) and in the number of paired nucleotides. For example, for the H-type structures, the stem in the TGV pseudoknot is longer and contains no mismatches, whereas the IBV pseudoknot is shorter and contains one mismatch. As such, we expect these different structures to pause the ribosome with varying efficiencies.