High-throughput interrogation of programmed ribosomal frameshifting in human cells

Martin Mikl; Yitzhak Pilpel; Eran Segal

Replacing the endogenous downstream region with elements that have different primary sequence, 326 but are predicted to fold into the same secondary structure (Fig 3E, top) abolished frameshifting in 327 most cases, regardless of whether pseudoknots were included or not (Vienna RNA vs. pKiss to 328 determine the structure that served as the input for antaRNA (Kleinkauf et al., 2015a (Kleinkauf et al., , 2015b ; Fig 3F) . including ones where the wild-type sequence did not yield a frameshifting signal in the context of our 332 assay (CCR5 and PEG10). Strikingly, events associated with higher wild-type frameshifting rates 333 could not be "rescued" by introducing a secondary structure variant from a different PRF event (Fig 334 3H ), indicating that more efficient frameshifting sites like SARS seem to be highly optimized, but less 335 tolerant to changes in the type of downstream secondary structure. We therefore repeated the analyses using pKiss (Janssen and Giegerich, 2015) , an algorithm that 350