High-throughput interrogation of programmed ribosomal frameshifting in human cells

Martin Mikl; Yitzhak Pilpel; Eran Segal

In the case of SRV1 (GGGAAAC) replacing the Z position with other options than the endogenous C 259 led to a drastic decrease (U, A) or complete loss (G) of -1 PRF signal (Fig S4E) , while in the case of 260 SARS (UUUAAAU) all other nucleotides apart from the endogenous U at the Z position almost 261 completely abolished the signal. To determine whether there is a universally optimal slippery site, we 262 replaced the slippery sequence in our set of -1 PRF events with all possible variations of the pattern 263 XXXYYYZ (Fig 2D, Fig S4F) . This revealed preferences for specific combinations common in known 264 PRF sites (Fig 2D) , like UUUUUUZ (HIV, SIVmac239) and XXXAAAC (HERV-K10, HTLV, PEG10, 265 SARS), but no universally optimal combination of bases that would lead to maximal frameshifting 266 efficiency and consequently GFP fluorescence across contexts ( Fig S5F) . Notably, PRF sites showed 267 a high degree of optimization when it comes to the identity of the bases, with any substitution (although 268 preserving the canonical XXXYYYZ pattern) typically leading to a more or less severe drop in PRF 269 signal ( Fig 2E, Fig S5EF) . 270