The error bars indicate means??s.d.; the value Rabbit polyclonal to ANKRD49 was determined by the two-tailed BrunnerCMunzel test. and SHAPE-MaP dataset of HIV-1 genome5. To design a barcode microarray, we used the datasets of barcodes for the hybridization of nucleic acid17. The data supporting the findings of this study are available from the corresponding authors upon reasonable request.?Source data are provided with this Sclareol paper. The custom scripts for the motif extraction of the terminal motifs, designing the RNA structure library, and generation of the microarray ordering template are available in the Github page with the instruction to order as products (https://github.com/KRK13/FOREST2020/). The other codes used in this study are available from the corresponding authors upon request. Abstract Biochemical assays and computational analyses have discovered RNA structures throughout various transcripts. However, the roles of these structures are mostly unknown. Here we develop folded RNA element profiling with structure library (FOREST), a multiplexed affinity assay system to identify functional interactions from transcriptome-wide RNA structure datasets. We generate an RNA structure library by extracting validated or predicted RNA motifs from gene-annotated RNA regions. The RNA structure library with an affinity enrichment assay allows for the comprehensive identification of target-binding RNA sequences and structures in a high-throughput manner. As a proof-of-concept, FOREST discovers multiple RNA-protein interaction networks with quantitative scores, including translational regulatory elements that function in living cells. Sclareol Moreover, FOREST reveals different binding landscapes of RNA G-quadruplex (rG4) structures-binding proteins and discovers rG4 structures in the terminal loops of precursor microRNAs. Overall, FOREST serves as a versatile platform to investigate RNA structure-function relationships on a large scale. loops using RNA structure library, version 1 (1916 structures). The known loops were categorized into three subclasses according to the presence of a cold-shock domain (CSD)-binding site and a zinc knuckle domain (ZKD)-binding site. The class-1 and class-2. We omitted the score of the loop from the calculation because it did not show a high intensity, unlike other class-2 loops. The numbers of analyzed loops are described in parentheses. The error bars indicate means??s.e.m., and the values were determined by two-tailed tests. Each indicates a data point. To evaluate the affinities of RNP interactions, we performed RNA pull-down assays, quantified the amounts of enriched RNA probes with fluorescence, and calculated enrichment scores by subtracting the amounts of a control sample as?the background noise and represented the scores as binding intensities (Figs.?1 and ?and4a).4a). We first used the human U1A (SNRPA) protein and RNA structure library, v1. The known U1A-binding aptamer was significantly enriched and had the highest binding intensity, whereas its defective mutant did not show a significant score (Supplementary Fig.?7a, b). Sequence motif frequency analysis confirmed that the known U1A binding sequence18,19, 5-GCAC-3, was a vital factor for interacting with U1A (Supplementary Fig.?7c). Further analysis of the structural context revealed that U1A preferred the GCAC sequence on the loop region rather than the GCAC sequence on the stem region (Supplementary Fig.?7d, e). This result can explain why some GCAC-containing RNA structures were detected with low intensities. We validated the results with an electrophoretic mobility shift assay (EMSA) and confirmed that top-ranked RNA structures bound to U1A and that their U1A-binding intensities on FOREST correlated with their affinities (Supplementary Fig.?8). These results demonstrated that FOREST can quantitatively assess RNP interactions using thousands of RNA structures. Next, we used FOREST to analyze the binding properties of LIN28A protein with the pre-miRNA loops. We focused on human (loops interacted with LIN28A with significantly high affinities (Fig.?4b). FOREST and Sclareol EMSA validation found that the loop did not interact efficiently with LIN28A under our conditions to a similar extent as the loop (Fig.?4c, d). Notably, we observed a significant difference in binding intensities depending Sclareol on the presence of the cold shock domain (CSD)-binding sequences (Fig.?4e). These results coincided with a previous study that showed interaction with the CSD domain of LIN28A is a vital component for high-affinity binding to specific loops and the downregulation of miRNA biogenesis in cells21. Collecting these results of human RBPs, we concluded that FOREST can generate the RNA structure library with proper folding and identify biologically functional RNA elements through binding intensities. Analysis of rG4 structure and its binding proteins To investigate whether we could analyze the interaction landscape between highly structured.