Keth-seq for transcriptome-wide RNA structure mapping

Abstract

RNA secondary structure is critical to RNA regulation and function. We report a new N₃-kethoxal reagent that allows fast and reversible labeling of single-stranded guanine bases in live cells. This N₃-kethoxal-based chemistry allows efficient RNA labeling under mild conditions and transcriptome-wide RNA secondary structure mapping.

Figure 1 |. N₃-kethoxal and experimental evaluation of its selectivity, cell permeability and reversibility.

(a) The structure of N₃-kethoxal and the reaction with guanine. (b) Denaturing gel electrophoresis demonstrating N₃-kethoxal only react with single-strand RNA (ssRNA). (c) Upper: Denaturing gel electrophoresis analysis of the labelling reaction of kethoxal and N₃-kethoxal with FAM-RNA oligo (5’-FAM-GAGCAGCUUUAGUUUAGAUCGAGUGUA, lane 1–3) and biotinylation with biotin-DBCO (lane 5, 6). Only N₃-kethoxal labelled RNA can be biotinylated (lane 6). Bottom: Dot blot of RNA after labelling and Biotinylation reactions. Methylene blue dot results are listed as control. (d) Dot blot of isolated total RNA from mES cells which were treated by N₃-kethoxal with different periods, 1, 5, 10, 15, 20 min. (e) Dot blot analysis of reversibility of N₃-kethoxal labelled mRNA in present of 50 mM GTP at 95 °C. The N₃-kethoxal modification in mRNA was removed thoroughly after 10 min incubation. Experiments were independently repeated twice with similar results obtained. Uncropped scans for b, c, d, and e are provided in Supplementary Figure 15.

Figure 2 |. Keth-seq method and the profile around rG4 regions.

(a) Scatter plot of reverse transcription (RT) stop reads distribution between replicates for N₃-kethoxal sample. The inset pie plots show RT stopped base distribution for replicate 2 (upper left, A: 604,222; T: 497,602; C: 481,596; G: 7,204,998) and replicate 1 (bottom right, A: 703,486; T: 586,297; C: 551,962; G: 8,683,824). (b) Accumulation plot of correlation coefficient between Keth-seq and icSHAPE for all transcript. For each common transcript, we calculate the Pearson correlation coefficient for structural signal of guanine bases. The inset plot shows all guanine reactivity between Keth-seq and icSHAPE for Rpl6 (a gene encoding ribosomal protein) transcript with a high correlation (Pearson correlation coefficient R: 0.789). (c) Left: scatter plot of AUC between Keth-seq and icSHAPE for RNAs with known structure model (18S ribosomal RNA from RNA STRAND database and others from Rfam database, 32 RNAs in total). Right: A fragment (240–285) of 18S ribosomal RNA with both Keth-seq and icSHAPE reactivity filled in the structure model. (d) Gini index of known rG4 regions (based on previously identified by Kwok et.al., 2016, Nature method) between +PDS treatment and native sample for in vitro (left) and in vivo (right). Only regions with structural information in both + PDS treatment and native conditions are retained for plotting (extended to 50-nucleotide long). (e) An example of Keth-seq profile around previously identified in vitro rG4 regions.