2
SUPPORTING DISCUSSION
Prior experiments on the S4-binding elementThe original double pseudoknot (DPK) model of the S4-binding element (S4E) was developed from early experiments mapping the impact of mutations on S4 binding (Fig. S6) (1). Following these initial experiments, the same library of S4E mutants has been extensively characterized by complementary biophysical and functional experiments (2-4). These data have been interpreted through the lens of the DPK model. However, as detailed below, much of this data is ambiguous and in several cases is inconsistent with the DPK structure.The DPK structure is motivated by two pairs of mutations that disrupt and rescue S4 binding in filter-binding assays (1). Disruption of the proposed PK2 interaction by mutation 17 reduces S4 binding ~10-fold and is rescued by the compensatory mutation 21 ( Fig. S6A). Disruption of PK3 by mutation 18 reduces S4 binding ~2-fold and is rescued by the compensatory mutation 22.However, the significance of these compensatory mutations is less clear when considered in the context of other mutation data. Relative to mutations spanning the kissing loop (KL) structure that uniformly and catastrophically disrupt binding, mutations to the PK2 and PK3 regions have inconsistent and modest impact on binding (Fig. S6A). Notably, multiple mutations that should disrupt PK2 and PK3 have no impact on binding. Subsequent in vivo functional assays also are inconsistent with the DPK model (2). Significantly, what should be compensatory 18+22 mutations do not rescue S4 repressive function in vivo. Finally, biophysical characterization of different S4E mutants also reveals inconsistencies of the DPK model, with the expected compensatory mutations (17+21 and 18+22) failing to rescue S4E tertiary folding (3). Overall, the inconsistent and modest impact of mutations to the proposed PK2 and PK3 interactions and the inconsistency of compensatory rescue argue against direct PK2 and PK3 pairing. Rather, these data are more consistent with these regions interacting indirectly, as would be expected for a kissing-loop type structure where nucleotides adjacent to the KL duplex contribute to tertiary stability and S4 binding but do not directly pair ( Fig. S6D).While most of the mutations tested by prior studies are not expected to impact H3, three mutations do provide evidence supporting H3 and the kissing-loop structure. Mutation 19 (G95àA) converts a GU pair in H3 to an AU pair, and as expected, has no impact on S4 3 binding in vitro or repression function in vivo (2). Mutation 18 modestly disrupts H3 and modestly decreases S4 binding affinity, although has no impact on repression function in vivo (1). Most significantly, mutation of the AGGAG Shine-Dalgarno sequence (SD; Fig. S6A) to its sequence complement, UCCUC, completely disrupts H3 and, as expected, completely abolishes S4 binding (4). By comparison, in the context of the DPK model, the SD mutation occurs in the middle of a 25-nt single-stranded loop and thus would not be expected to have such profound impa...