Nat. Methods doi:10.1038/nmeth.3029

Credit: NATURE METHODS

Accurately predicting the secondary structures of long RNAs from their sequences is very challenging, but generating accurate models of RNA structures can be greatly facilitated by incorporating information obtained from chemical probes. Siegfried et al. have modified their previously published 'SHAPE' method—in which 2′-hydroxyl groups in flexible regions of the RNA backbone are selectively acylated with a reactive chemical probe—so that it can be used to rapidly generate highly accurate models of long, structured RNAs. Whereas previous versions of SHAPE identified these modified 2′-hydroxyl groups because they inhibit reverse transcription, this new 'SHAPE-MaP' method records sites of 2′-hydroxyl–modified RNA as mutations introduced during reverse transcription. Next-generation sequencing is then used to directly identify the positions and quantify the relative frequencies of the modified nucleotides. The authors first showed that SHAPE-MaP could replicate existing knowledge about ligand-induced conformational changes in the Escherichia coli thiamine pyrophosphate riboswitch (see image). They then used the strategy to examine the 9,200-nucleotide HIV-1 RNA genome and identified 15 RNA regions that were predicted to have well-defined structures, several of which were not previously known. The authors determined that two newly identified pseudoknots may have an important function in vivo, as the introduction of silent mutations that disrupted the secondary structures of these pseudoknots yielded viruses with reduced fitness in cell-based assays.