The reverse transcription signature of N-1-methyladenosine in RNA-Seq is sequence dependent - PubMed (original) (raw)

. 2015 Nov 16;43(20):9950-64.

doi: 10.1093/nar/gkv895. Epub 2015 Sep 13.

Lyudmil Tserovski 1, Katharina Schmid 1, Kathrin Thüring 1, Marie-Luise Winz 2, Sunny Sharma 3, Karl-Dieter Entian 3, Ludivine Wacheul 4, Denis L J Lafontaine 4, James Anderson 5, Juan Alfonzo 6, Andreas Hildebrandt 7, Andres Jäschke 2, Yuri Motorin 8, Mark Helm 9

Affiliations

The reverse transcription signature of N-1-methyladenosine in RNA-Seq is sequence dependent

Ralf Hauenschild et al. Nucleic Acids Res. 2015.

Abstract

The combination of Reverse Transcription (RT) and high-throughput sequencing has emerged as a powerful combination to detect modified nucleotides in RNA via analysis of either abortive RT-products or of the incorporation of mismatched dNTPs into cDNA. Here we simultaneously analyze both parameters in detail with respect to the occurrence of N-1-methyladenosine (m(1)A) in the template RNA. This naturally occurring modification is associated with structural effects, but it is also known as a mediator of antibiotic resistance in ribosomal RNA. In structural probing experiments with dimethylsulfate, m(1)A is routinely detected by RT-arrest. A specifically developed RNA-Seq protocol was tailored to the simultaneous analysis of RT-arrest and misincorporation patterns. By application to a variety of native and synthetic RNA preparations, we found a characteristic signature of m(1)A, which, in addition to an arrest rate, features misincorporation as a significant component. Detailed analysis suggests that the signature depends on RNA structure and on the nature of the nucleotide 3' of m(1)A in the template RNA, meaning it is sequence dependent. The RT-signature of m(1)A was used for inspection and confirmation of suspected modification sites and resulted in the identification of hitherto unknown m(1)A residues in trypanosomal tRNA.

© The Author(s) 2015. Published by Oxford University Press on behalf of Nucleic Acids Research.

PubMed Disclaimer

Figures

Figure 1.

Figure 1.

Principle of generation and analysis of RNA-Seq data for the detection of m1A residues.

Figure 2.

Figure 2.

Detection of m1A signatures in deep sequencing data. The representations illustrate the coverage of a given site in gray, the arrest rate is plotted as a red line, and the mismatch composition is visualized by colored stacks at the m1A sites. For a position p, the arrest rate reflects the relative amount of mapped reads ending at p + 1, i.e. not covering p. (A) Sequencing profiles from single and double methyltransferase knockouts of Saccharomyces cerevisiae's LSU rRNA with m1A sites 645 and 2142. Signatures of m1A residues are clearly apparent in the wild-type, and disappear in the corresponding knockout constructs. (B) Sequencing profiles of tRNAIle_TAT from wild-type and Trm6-knockout strains. The signature clearly disappears in RNA from a knockout strain of the enzyme, which is responsible for synthesis of m1A58 in tRNAs (28). tRNAIle_TAT was chosen as an example out of 37 signatures, which are detailed in Supplementary Figure S4. Positions are labeled according to absolute length of reference sequences, including variable regions. (C) Average signature of said 37 yeast cytosolic tRNAs at m1A58 complemented with absolute standard deviations of signature features among and between groups of isotypes. For the displayed profile, signatures were averaged among isotypes first, before calculating the final means.

Figure 3.

Figure 3.

Revolver assay. Revolver oligonucleotides feature permutation of the four major nucleotides at a position of interest, here the +1 position (3′ to m1A). For a position p, the arrest rate reflects the relative amount of reads ending at p + 1 (i.e. not covering p) out of all reads covering p + 1. (A) Ternary plot of mismatch composition of 41 natural m1A sites (black dots) for base configurations guanosine (yellow), cytidine (blue), uridine (red, T in mapping profile) and adenosine (green) at position +1 w.r.t. m1A. Data points from revolver oligonucleotides are represented as colored letters corresponding to the color code also used in (C) and (D). (B) Twenty-two hierarchically clustered data points derived from initial 41 measurements in (A). (C) Mismatch composition at m1A site for base configurations guanosine (i), cytidine (ii), uridine (iii, T in mapping profile) and adenosine (iv) at position +1 w.r.t. m1A in sequencing profiles of synthetic oligonucleotides. (D) RT signature by modification level. Arrest rates and mismatch contents at different ratios of modified and unmodified equivalents of revolver oligonucleotide are shown: 0% m1A in (D-i), 25% in (D-ii), 50% in (D-iii), 75% in (D-iv) and 100% m1A in (C-i).

Figure 4.

Figure 4.

Quantification of m1A by LC-MS using a biosynthetic internal standard. LC-MS/MS chromatograms showing the m1A and 13C-labeled m1A peaks in the revolver oligonucleotides (A) and in 25S rRNA from wild-type and rrp8/bmt2 knockout yeast (B). Continuous lines represent the peaks of unlabeled m1A, dotted lines those of 13C-labeled m1A added as an internal standard (24). To ensure inter-sample comparability of the m1A peaks, the peak heights were adjusted to the respective 13C-m1A peaks and normalized to the injected amount of oligonucleotide or 25S rRNA. The amount of analyzed oligonucleotide or 25S rRNA was determined by calculating the amount of adenosine in the respective samples using the UV peak of adenosine and dividing the amount by the number of adenosines per molecule. AU—arbitrary units. (C) Plot of RT signature occupancy by m1A content.

Figure 5.

Figure 5.

Homology based confirmation of m1A. For a position p, the arrest rate reflects the relative amount of mapped reads ending at p + 1, i.e. not covering p. (A) Homologous identification of m1A1136 in murine 28S rRNA (i) by alignment to human sequence containing m1A1309 (ii). (B) m1A9 in human mitochondrial tRNA, identified by alignment to identical bovine sequence with published m1A9.

Figure 6.

Figure 6.

Validation outline for supervised prediction. RT signatures (yellow) of m1A and non-m1A (A*) sites and are distributed into subsamples, termed folds, with uniform ratios (stratification) m1A / A*. The system was tuned toward both, sensitivity and specificity by equal abundance of each class, minimizing learning biases due to a priori class probabilities. In each of 10 repetitions (10×), the Random Forest was trained on another four of five possible fold combinations (5×) and tested on the respective left-out fold.

References

    1. Lempereur L., Nicoloso M., Riehl N., Ehresmann C., Ehresmann B., Bachellerie J.P. Conformation of yeast 18S rRNA. Direct chemical probing of the 5′ domain in ribosomal subunits and in deproteinized RNA by reverse transcriptase mapping of dimethyl sulfate-accessible. Nucleic Acids Res. 1985;13:8339–8357. - PMC - PubMed
    1. Motorin Y., Muller S., Behm-Ansmant I., Branlant C. Identification of modified residues in RNAs by reverse transcription-based methods. Methods Enzymol. 2007;425:21–53. - PubMed
    1. Behm-Ansmant I., Helm M., Motorin Y. Use of specific chemical reagents for detection of modified nucleotides in RNA. J. Nucleic Acids. 2011;2011:408053. - PMC - PubMed
    1. Mortimer S.A., Trapnell C., Aviran S., Pachter L., Lucks J.B. SHAPE-Seq: high-throughput RNA structure analysis. Curr. Protoc. Chem. Biol. 2012;4:275–297. - PubMed
    1. Roovers M., Wouters J., Bujnicki J.M., Tricot C., Stalon V., Grosjean H., Droogmans L. A primordial RNA modification enzyme: the case of tRNA (m1A) methyltransferase. Nucleic Acids Res. 2004;32:465–476. - PMC - PubMed

Publication types

MeSH terms

Substances

LinkOut - more resources