In vivo DMS mapping reveals a flexible structure for HSUR1. (A) Primer extension analysis of HSUR1 performed on total RNA isolated from HVS-transformed marmoset T cells treated or untreated with DMS. M, marker; A, C, G, and U correspond to sequencing lanes containing, respectively, ddTTP, ddGTP, ddCTP, and ddATP. Green stars depict strong DMS reactivity (>4× average) averaged from three independent experiments. (B,C) In vivo DMS data plotted on the previous model of HSUR1 (B), and the unconstrained MFE model (C). The green arrows indicate nucleotides predicted to be in Watson–Crick pairs flanked by Watson–Crick pairs that have strong DMS reactivity. (D) The HSUR1 MFE model predicted using hard constraints from in vivo DMS probing. (E) The MFE model predicted using soft constraints from in vivo DMS probing. The DMS reactivity averaged from three independent experiments on the structure models obtained with primer A in B–E is represented by the intensity of red color at each base. (F) The partial MFE model (omitting the 3′ terminal hairpin) depicting DMS reactivity averaged from three independent experiments obtained with primer B. Watson–Crick base pairs are represented by solid lines and wobble base pairs by dotted lines. Predicted interactions of HSUR1 with miR-27a are depicted in B–F. The seed sequence of miR-27a is shown in bold. Known interaction sites for HuR and Sm proteins are indicated on the structure models.
