Probing RNA structure in vivo

Abstract

RNA structure underpins many essential functions in biology. New chemical reagents and techniques for probing RNA structure in living cells have emerged in recent years. High-throughput, genome-wide techniques such as Structure-seq2 and DMS-MaPseq exploit nucleobase modification by dimethylsulfate (DMS) to obtain complete structuromes, and are applicable to multiple domains of life and conditions. New reagents such as 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC), glyoxal, and nicotinoyl azide (NAz) greatly expand the capabilities of nucleobase probing in cells. Additionally, ribose-targeting reagents in selective 2'-hydroxyl acylation and primer extension (SHAPE) detect RNA flexibility in vivo. These techniques, coupled with crosslinking nucleobases in psoralen analysis of RNA interactions and structures (PARIS), provide new and diverse ways to elucidate RNA secondary and tertiary structure in vivo and genome-wide.

Figure 1.

Comparison of in vivo structure probing methods. (A) DMS (green circles), EDC (blue triangles), and SHAPE reagents (purple diamonds) react with solvent-accessible, single-stranded RNA. Conversely, NAz (pink hexagons) is agnostic to base pairing and reacts with any solvent-exposed purines. (B) Reverse transcription reactions result in RT stops, which lead to truncated cDNAs or mutations that lead to cDNA extension past the modified RNA base.

Figure 2.

In vivo RNA structure probing reagents targeting nucleobases and ribose. Sites of targeting on the nucleobase or ribose for each reagent are shown at an atomic level. Arrow directionality represents electron transfer from the nucleophile to the electrophile. “SHAPE” here represents an assortment of ribose-targeting reagents discussed in this review, as all such reagents exhibit similar chemistry. Reaction is prevented if the sites of reactivity are blocked by base pairing or by protection via protein binding or tertiary structure formation.

Figure 3.

Comparison of a previous Xist structure model (left) generated from DMS probing using targeted Structure-seq data only [28], and the updated Xist model (right) generated from DMS using targeted Structure-seq data combined with DMS-MaPseq and EDC probing data [38]. Differences in base pairing between the two models are shown by magenta-colored dashes between the bases in each structure. Adapted from [28,38].