Stem–loop structures can effectively substitute for an RNA pseudoknot in -1 ribosomal frameshifting

Yu, Chien-Hung; Noteborn, Mathieu H.; Pleij, Cornelis W. A.; Olsthoorn, René C. L.

In the HIV-1 gag-pol gene Bidou et al. (20) observed a 15-20% decrease in frameshifting in vivo with their most stable hairpin, similar to our results with the 21 bp hairpin. However, in our case, the stability at which this happened was À45 kcal/mol much higher than their most stable hairpin of À22.1 kcal/mol. It is possible that this difference is due to the different experimental systems. Although it has been suggested that too stable stems increase the time for tRNAs to shift back into the 0-frame again (20) we believe that our 21 bp hairpin is less efficient because it has more AU bps in the middle of the stem compared to the 12 and 15 bp hairpins (Figure 2A) . The experiments with hairpins harboring bulges or mismatches halfway through the stem demonstrated that this region is quite important for frameshifting ( Figure 3A and B) . Even though the overall stability of these constructs was comparable to that of a hairpin of 9 or 10 bp, their frameshift activity was equal or lower than that of a 6 bp hairpin of À13.1 kcal/mol: as if the mismatch or bulge after the 6th base pair disconnected the upper part of the stem. This observation is reminiscent of the overall destabilizing effect of mismatches in DNA hairpins. In a pioneer singlemolecule pulling study, it was shown that introducing a mismatch in a 20 bp DNA hairpin shifted its transition state close to the location of the mismatch (36) . Our data also comply with this mechanical study and suggest that mechanical stability may be a better parameter than thermodynamic stability to describe the frameshift efficiency of hairpins.