The chemistry and applications of RNA 2′-OH acylation

Abstract

RNA is a versatile biomolecule with a broad range of biological functions that go far beyond its initially described role as a simple information carrier. The development of chemical methods to control, manipulate and modify RNA has the potential to yield new insights into its many functions and properties. Traditionally, most of these methods involved the chemical modification of RNA structure using solid-state synthesis or enzymatic transformations. However, over the past 15 years, the direct functionalization of RNA by selective acylation of the 2′-hydroxyl (2′-OH) group has emerged as a powerful alternative that enables the simple modification of both synthetic and transcribed RNAs. In this Review, we discuss the chemical properties and design of effective reagents for RNA 2′-OH acylation, highlighting the unique problem of 2′-OH reactivity in the presence of water. We elaborate on how RNA 2′-OH acylation is being exploited to develop selective chemical probes that enable interrogation of RNA structure and function, and describe new developments and applications in the field.

Over the last two decades, the classical view of RNA as a simple information carrier that facilitates the translation of genetic information encoded in the DNA into functional proteins has been replaced by the concept that RNA is a highly versatile molecule with a multitude of biological functions^1,2. Indeed, RNAs are not only directly involved in gene expression but also have important roles in many other cellular processes² and in cell signalling³, and can travel between cells in exosomes⁴. They are dynamic, changing structure and subcellular location in response to stimuli⁵, and have lifespans ranging from minutes to days⁶. RNAs can also function as switches^7,8 and can perform a wide range of catalytic functions^8,9 during, for example, peptide-bond formation by the ribosome¹⁰. Importantly, RNAs can be post-transcriptionally modified with over a hundred base and sugar modifications¹¹, which markedly alter their structure and biological activities^12,13. Furthermore, interest is growing in the many thousands of non-coding RNAs (ncRNAs) in the cell^1,14, which include over a dozen classes of small and large RNAs, ranging from very short transfer RNA (tRNA) fragments of just a dozen nucleotides to large ncRNAs thousands of nucleotides in length¹⁴. These ncRNAs can interact with each other, with messenger RNAs (mRNAs) and with many proteins¹⁵, and are present throughout the cell, both individually and as part of large complexes in specialized cellular compartments, such as stress granules, processing bodies and nucleoli^16,17. Thus, RNAs have complex and significant roles in biology.

Non-coding RNAs.

(ncRNAs). RNA molecules that are not translated into proteins, but often have other biological roles, such as assisting in splicing, gene regulation and DNA replication.

Transfer RNA.

(tRNA). A non-coding RNA molecule that carries an amino acid and helps to decode messenger RNA (mRNA) into protein; tRNAs contain a three-nucleotide sequence (anticodon) that matches to a three-nucleotide sequence on mRNA (codon).

Messenger RNAs.

(mRNAs). Coding RNA molecules that convey the genetic information from DNA to facilitate biosynthesis of functional proteins; the mRNA nucleotide sequence is translated into protein by the ribosome.

This versatility of RNA has been exploited in the field of medicine over recent years. Indeed, patisiran, the first drug to exploit RNA interference to silence disease-associated gene expression, was approved in 2018 to treat polyneuropathy¹⁸, and therapeutic approaches involving CRISPR RNAs¹⁹ and mRNAs²⁰ are under intensive investigation, emphasizing the robust clinical potential of RNAs²¹. In addition, specific classes of RNAs are now under study as potential direct therapeutic drug targets².

Although the complexity of the transcriptome is becoming increasingly appreciated, the roles of many — possibly most — RNAs in the cell remain unknown. Indeed, it seems likely that many RNAs have yet to be discovered or characterized, and might be part of the unrecognized and uncharacterized ‘dark matter’ of the cell²²·²³. In addition, even for most known RNAs, their interactome (that is, the wide range of molecules that they interact with)²⁴ is poorly understood, and their secondary and tertiary structures are just beginning to be characterized²⁵. New tools, such as RNA sequencing²⁶ and epitranscriptomic analysis²⁷, have opened the door to study many unresolved questions, most of which will require new chemical and analytical methods to answer.

To study the numerous functions of RNA, it is essential to be able to manipulate and observe its properties. To this end, many methods have been developed, including crosslinking technologies such as psoralen analysis of RNA interactions (PARIS)²⁸ and crosslinking immunoprecipitation (CLIP)²⁹. Other structural analytical methods include selective 2′-hydroxyl acylation analysed by primer extension (SHAPE)³⁰ and dimethyl sulfate footprinting (DMS footprinting)^31,32. To interrogate RNA localization and folding properties, fluorescence-labelling techniques have been devised, such as the use of fluorescent RNA aptamers³³ and fluorescent nucleobase analogues³⁴, as well as the enzymatic incorporation of fluorophores³⁵·³⁶. Methods have also been developed to enhance the cellular uptake of RNAs, including the use of lipid-RNA conjugates^37,38 and cell-penetrating peptide-RNA conjugates^38,39. Furthermore, innovative means of introducing photocleavable or photoisomerizable bonds into the structure of RNA have been developed to spatio-temporally control the properties of RNAs^40,41. A common feature of the majority of these modification methods is the ability to covalently attach a functional group to the molecular structure of RNA — including crosslinks²⁸, fluorophores⁴², targeting groups or affinity tags⁴³ — under mostly aqueous conditions, which puts constraints on the available synthetic reactions. For example, both DMS footprinting^31,32 and SHAPE^30,44 reagents are powerful for RNA structural analysis, but are susceptible to hydrolysis^30,31, which dictates the experimental setup. Several strategies are available for the covalent modification of RNA, including installing modifications during solid-phase RNA synthesis, post-synthetic treatment with electrophiles and enzymatic transformations⁴⁵. The majority of the modifications involve the nucleobases or the extreme ends of the polymer⁴⁵.

Psoralen analysis of RNA interactions.

(PARIS). A method for mapping RNA structure in cells using the small-molecule psoralen as an RNA crosslinker; crosslinked RNA fragments are analysed with next-generation sequencing and, using informatics methods, duplex regions can be assigned throughout the transcriptome.

Reactions involving the 2′-hydroxyl (2′-OH) group are arguably the most generalized approaches for covalent modification of RNA, given that nearly every nucleotide, regardless of sequence, contains the 2′-OH group. Specialized enzymes, such as aminoacyl tRNA synthetase, have long been known to acylate specific terminal 2′-OH and 3′-OH groups in RNA⁴⁶, and 2′-O methylation of RNA is known to be accomplished by complex assemblies of proteins associated with RNAs⁴⁷, but these enzymatic approaches can not yet be applied to any RNA of interest. Chemical methods that are generally applicable to RNA are beginning to find widespread use and will be the focus of this Review. Following early interest in the 1960s and 1970s^48–50, RNA acylation at the 2′-OH position was largely disregarded for several decades; however, this strategy has experienced a resurgence in the past 15 years due to the experience with the chemical reactivity of this group and the development of specialized and highly tuned reagents.

Crosslinking immunoprecipitation.

(CLIP). A method for studying protein-RNA interactions whereby cells are exposed to high-intensity ultraviolet light, which crosslinks proteins and RNA molecules that are in close proximity; using immuno-precipitation, the complexes can be isolated and RNAs can be identified with sequencing.

Selective 2′-hydroxyl acylation analysed by primer extension.

(SHAPE). A method for analysing RNA structure whereby 2′-OH groups in RNA can be acylated with small-molecule reagents in unpaired, accessible and flexible regions; the acylation groups block reverse transcriptase during primer extension. Subsequent analysis of primer extension products is used to predict accessible regions and secondary structures of RNAs.

In this Review, we summarize the current state of the art of chemical methods for RNA acylation. We first focus on the chemistry of acylating 2′-OH groups in RNA and the requirements for the design of an effective RNA acylation reagent, after which we describe and provide illustrative examples of the applications of RNA acylation.

Dimethyl sulfate footprinting.

(DMS footpinting). A method for determining unpaired regions of nucleic acids using DMS, which can methylate the N1 position of adenine and N3 position of cytosine; methylation occurs selectively on unpaired nucleobases and can block reverse transcriptase. Analysis of reverse-transcription products reveals unpaired regions in nucleic acids.

The chemistry of RNA acylation

The molecular structure of RNA contains multiple potential modification sites. However, the selective modification of the 2′-OH position of RNA through acylation (FIG. 1a) poses a significant challenge.

Fig. 1 |. Selective acylation of 2′-OH groups in RNA with activated carbonyls.

a | Schematic of the acylation reaction involving a 2′-hydroxyl (2′-OH) group and an activated carbonyl, resulting in functionalized RNA. The appended group (green) can be a variable functional handle. b | Gel electrophoretic analysis of single-stranded DNA and analogous single-stranded RNA treated with acylating reagent 25 at a concentration of 100 mM (REF⁵²). A mobility shift is only observed for treated RNA, indicating selective acylation of 2′-OH positions. c | Inductive effect of the neighbouring 3′ and 4′ oxygens and the nucleobase nitrogen on the 2′-OH group. d | Suggested mechanism of 2′-OH nucleophilic attack kinetically assisted by deprotonation by the neighbouring phosphate group⁶⁰. e | Molecular structures of selected acylating reagents (1–18). 1M7, 1-methyl-7-nitroisatoic anhydride; NAI, 2-methylnicotinic acid imidazolide; NMIA, N-methylisatoic anhydride. Part b reproduced with permission from REF⁵², ACS.

RNA as a nucleophile

As RNA is highly soluble in water and poorly soluble in most organic solvents, the majority of acylating reactions are performed under aqueous conditions. Thus, water is assumed to be the de facto solvent in this Review, unless otherwise stated. Given the similar reactivities of water and alcohols, conducting acylation in the presence of water is challenging and raises several considerations.

RNA has several nucleophilic functional groups that could potentially serve as reactive partners in acylating reactions, including the exocyclic amines on cytosine, adenine and guanine nucleobases, the 5′-OH and 3′-OH groups at the ends of the RNA strand and the 2′-OH groups. As the phosphate diesters in the RNA backbone are generally unreactive towards electrophiles and are considered to be the least reactive of all phosphate esters⁵¹, they are not expected to be affected by acylating chemistry. Among the more nucleophilic groups, the exocyclic amines have, thus far, been found to be relatively unreactive to acylation, likely due to resonance delocalization, which decreases electron density on the amines. The amino groups possess p/π electrons that donate electron density into the conjugated system, which reduces amine nucleophilicity⁵⁰. Acylation reactions of acylimidazole reagent 25 with single-stranded DNA and single-stranded RNA of an analogous sequence showed that acylation is highly selective towards RNA⁵² (FIG. 1b), indicating that the acylation reaction mainly occurs at the 2′-OH position of RNA, rather than at the exocyclic amine positions; similar observations have been made for single nucleotides⁵³. Amine acylation in very low yields (<1%) has been observed on treatment of RNA with N-methylisatoic anhydride (NMIA; 2)⁵⁴. However, amine acylation on cytidine residues with a yield of up to 65% has been reported when reactions were carried out with acetic anhydride in organic solvents such as dimethylformamide (DMF)⁵⁵. It seems that water interferes with this amine reactivity, possibly by outcompeting these groups.

The reaction of alcohols in water at neutral pH is challenging because most alcohols (pKa of ~14–15) have similar nucleophilicity to water (pKa of 14)⁵⁶, which, at 55 M, readily outcompetes most alcohols kinetically. However, compared with many alcohols, the 2′-OH group in RNA has a lower pKa (estimated at ~12–14)^57,58, which provides a potential basis for enhanced reactivity. Early studies by Knorre et al.^49,59 revealed that RNA is readily acylated at the 2′-OH position with acetic anhydride in water, with a final anhydride concentration of 0.25 M. This observation provided early evidence that this reaction could have utility for RNA acylation. One contributing factor in this reactivity is likely to be the inductive effect of the neighbouring 3′ and 4′ oxygens and the nucleobase nitrogen, all of which withdraw electron density from the 2′-OH (FIG. 1c), rendering it more acidic. It has also been suggested that the neighbouring 3′ phosphodiester anion kinetically enhances acylation by functioning as a general base to remove the 2′-OH proton during attack⁶⁰ (FIG. 1d). However, the relative contributions of these mechanisms remain to be measured.

In addition to the 2′-OH groups, acylation can potentially also occur at the terminal 5′-OH and 3′-OH groups of RNA, which should be considered when designing an acylation experiment⁴⁸. It should be noted that biologically derived RNAs are often phosphorylated at the terminal 5′-OH and 3′-OH positions, blocking these potential acylation sites, whereas synthetic RNAs are mostly designed without terminal phosphate groups. Interestingly, the observation that DNA is poorly acylated relative to RNA suggests that the reactivity of the terminal 5′-OH and 3′-OH groups is lower than that of the 2′-OH groups⁵².

Design of effective acylating reagents

To design an effective acylating reagent for aqueous use, it is crucial to tune the reactivity of the electrophile. For RNA, the acylating reagent should ideally have high reactivity for the 2′-OH group and low reactivity towards water, which is the solvent in most RNA-acylation reactions. In practice, achieving this RNA selectivity is challenging. Highly reactive reagents tend to react quickly with water, resulting in a rapid decline in concentrations prior to the reaction with the low-concentration RNA. Conversely, reactivity that is too low might result in poor yields on RNA due to insufficient electrophilicity. In addition, the aqueous solubility of the acylating reagent, which can limit the achievable concentrations, should be considered.

The reactivity of an acylating reagent can be assessed by determining its half-life in water. A wide range of half-lives, ranging from seconds to several hours, have been observed for RNA-acylating agents, depending on their intended application^44,61. A short half-life results in rapid hydrolysis of the reagent, leading to minor (but still potentially useful) RNA acylation. For example, Mortimer and Weeks⁶¹ studied the fast-acting compound 1-methyl-7-nitroisatoic anhydride (1M7; 1), which is a more reactive nitro-substituted version of NMIA (2) and has a half-life of 14 s compared with 240 s for the parent compound (FIG. 1e). Using compound 1, they were able to speed up their experiments to analyse RNA secondary and tertiary structure⁶¹. It should be noted that the definition of ‘useful’ levels of RNA acylation largely depends on the application; some applications require stoichiometric levels of reaction, whereas others require only a trace of reaction. Recent studies have shown that both NMIA (2) and 1M7 (1) isatoic anhydrides reacted with an 18-nucleotide RNA and yielded no measurable reaction on a stoichiometric basis as measured by mass spectrometry, consistent with a low yield of ~2–3%⁶². However, it is clear that such reactions do occur at 2′-OH groups in low yields that are observable by primer extension; although not useful for stoichiometric applications, this ‘sparse’ reaction provides the basis for evaluating RNA secondary structure using the SHAPE method.

An overly long half-life in water usually means that the acylating reagent has low reactivity and might also result in low levels of RNA acylation. For example, compound 3 (FIG. 1e) bears an electrophilic carbamate, but the imidazole leaving group is not electron-deficient enough to significantly activate the carbonyl, resulting in a half-life in water of well over 4 h (REF⁵²); little to no RNA acylation was seen in this study. Replacing imidazole with 2-chloroimidazole to afford compound 4 (FIG. 1e) enhanced the leaving-group ability, resulting in a shortened half-life of ~1.5 h and a robust RNA-acylating ability, yielding a superstoichiometric reaction of, on average, three carbonate groups on a 12-nucleotide RNA strand⁵². In general, reagents with intermediate half-lives of ~30–120 min seem to yield stoichiometrically high levels of acylation, as is the case for 2-methylnicotinic acid imidazolide (NAI; 5)⁴⁴ (FIG. 1e), which has a half-life of 33 min in water. This compound can react with RNA in a superstoichiometric fashion, acylating most of the hydroxyls on an RNA strand when reacted at 200 mM for 2 h (REF.⁶³).

As some RNA-acylating strategies are aimed at superstoichiometric levels of reaction, it is important to consider the water solubility of the acylating reagent. For example, the reagent NAI (5), which has a heterocyclic structure and an imidazole leaving group, was designed with enhanced solubility relative to prior isatoic anhydrides⁴⁴. Addressing this latter class of compounds, Ogle and colleagues^64,65 recognized the limited water solubility and short half-lives of early isatoic anhydrides. They designed improved versions of NMIA bearing a quaternary ammonium group to improve water solubility; one such version, compound 6 (FiG. 1e), was shown by primer-extension experiments at a concentration of 15 mM to readily react with 2′-OH groups in RNA and with lysine residues on proteins^64,65. In another approach to enhancing solubility, Velema and Kool⁶⁶ modified the imidazole leaving group in carbamate derivatives through addition of a tertiary amine group, which is expected to be positively charged at neutral pH. This enabled the superstoichiometric functionalization of RNA with highly hydrophobic groups, such as heptyl chains.

RNA-acylating reactions that result in ester and carbonate formation at 2′-OH groups have been reported. To design a reagent that results in ester formation, an activating group with moderate electron-withdrawing properties can be used. For carbonate formation, groups with strong electron-withdrawing properties are necessary to activate the carbonyl of the carbamate acylation reagent. By tuning the electron-withdrawing properties of the activating group, the reactivity of the resulting reagent can be controlled. For example, Park et al.⁶² prepared a range of alkyl oxycarbamate compounds (8-11) (FiG. 1e) that were activated with varied imidazole groups. The strongest electron-withdrawing groups were those shown to decrease the half-life of the acylating reagent in water, underlining the possibility of tuning reagent reactivity by changing electronic properties.

Early work by Kool, Chang and colleagues⁴⁴ involved the screening of nine acylating agents for aqueous stoichiometric reaction with ribonucleotides in water, and they found anhydrides and N-hydroxysuccinimide (NHS) esters to generally be poorly reactive, whereas acyl cyanides were too reactive. However, the acylimidazole derivative of 2-methylbenzoic acid gave good levels of ribose reactions, apparently due to its intermediate reactivity. Similarly, the group of Fossé, Xie and colleagues⁶⁷ investigated several activated carbonyls for 2′-OH acylation of RNA and found that symmetric anhydrides (12 and 13), cyanomethyl ester (14), succinimidyl esters (15 and 16) and acyl fluorides (17 and 18) were less reactive than NMIA (2) (FiG. 1e). Of these compounds, the acyl fluoride 18 was able to acylate RNA but exhibited lower reactivity than NMIA (2). Interestingly, after addition of 4-dimethylaminopyridine (DMAP), compounds 12, 13, 15, 16 and 18 showed substantially increased reactivity, presumably as a result of nucleophilic catalysis by DMAP Levels of reaction were measured by primer extension and analysed by gel electrophoresis, using NMIA as a control. This finding suggests that pyridinium might be a suitable activating group for acylating reagents.

Collectively, these studies suggest that, when designing acylating reagents, the intended application dictates the reactivity of the acylating group. When sparse acylation is desired, highly reactive fast-acting reagents can be useful. However, when stoichiometric or superstoichiometric acylation is to be achieved, reagents with milder reactivity are advisable. Furthermore, water solubility of the reagents is crucial to enable sufficient effective concentrations.

Applications of RNA acylation

RNA acylation has found application in many areas of research, including RNA structural analysis, RNA labelling, nuclease protection, tRNA aminoacylation and caging studies.

Reverse transcriptases.

A class of DNA polymerase enzymes that produces complementary DNA (cDNA) from an RNA template.

Investigation of RNA structure and interactions

The most widespread application of 2′-OH acylation has been in analysis of RNA secondary structure using SHAPE^30,53. The main purpose of SHAPE is to probe RNA secondary structure and interactions; for a comprehensive account of SHAPE, we redirect readers to several excellent reviews that have been published on this topic^30,68–70.

Next-generation sequencing.

(NGS). A term used to describe different modern sequencing technologies, all of which are capable of determining the sequence of millions of DNA fragments in a single reaction volume.

In 2005, pioneering work by Weeks and colleagues⁵³ showed that RNA acylation by isatoic anhydrides stalls reverse transcriptases and that this phenomenon can be exploited to probe the structure of RNA. Key to this method is that 2′-OH acylation shows little sequence selectivity, but mainly occurs on accessible and flexible nucleotides in non-paired (or weakly paired) regions of RNA. Flexible nucleotides can adopt several conformations, some of which sterically favour nucleophilic attack by the 2′-OH position, compared with non-flexible and base-paired nucleotides³⁰. After sparse RNA acylation using a SHAPE reagent, primer extension with reverse transcriptase is performed to yield complementary DNA (cDNA) extension products, which can be analysed to determine the exact positions of the acylated 2′-OH groups. Extension products are analysed by capillary or gel electrophoresis, revealing the relative reactivities at essentially every nucleotide position along an RNA strand. As reverse transcriptase stops one nucleotide prior to an acylated nucleotide, the primer extension step yields cDNAs of varied lengths and, therefore, bands of quantifiable intensity on electrophoresis (FIG. 2). Over the past decade, electrophoresis has been increasingly replaced with next-generation sequencing (NGS); this method is referred to as SHAPE-seq^71,72, and, when used in conjunction with mutational analysis, is termed SHAPE-MaP⁷³. To yield useful data, it is important to employ acylating reagents at concentrations that result in acylation of ~1 out of every 300 nucleotides, or less³⁰. Thus, the optimal concentration depends on the reactivity of the reagent and the amount of RNA used. The main purpose of SHAPE is to provide a ‘snapshot’ of the RNA structure at the time of reagent addition, but it is important to realize that long-term incubation with acylating reagents might alter cellular physiology.

Fig. 2 |. Determination of RNA structure using SHAPE.

a | Schematic flowchart of a standard selective 2′-hydroxyl acylation analysed by primer extension (SHAPE) experiment. RNA with an unknown structure is treated with an acylating reagent (SHAPE reagent) for ~1–30 min, depending on the reagent. Acylation mainly occurs at non-paired or weakly paired accessible sites on RNA. Next, primer extension is performed with reverse transcriptase, which stops at acylation sites, yielding complementary DNAs (cDNAs) of varied length. Finally, sequence analysis is performed using electrophoresis or sequencing methods, and data Quantifying the stop intensity can be presented in a reactivity map. This, in turn, is used in combination with folding algorithms to deduce the structure of the RNA. b | Molecular structures of SHAPE reagents (1, 2, 5, 19 and 20). 1M7, 1-methyl-7-nitroisatoic anhydride; FAI, 2-methyl-3-furoic acid imidazolide; NAI, 2-methylnicotinic acid imidazolide; NMIA, N-methylisatoic anhydride.

Seminal SHAPE studies were performed with NMIA (2) ^53,74, which had previously been shown by Hiratsuka^75,76 to acylate mononucleotides. An alternative version of this compound, 1M7 (1), included a nitro group at the 7-position to increase reactivity and improve the speed of RNA structural analysis, which could be completed in as little as 70 s (REF⁶¹). The initial studies by Weeks and colleagues^53,74 reported on the in vitro structure of several RNAs, including tRNA-Asp, the specificity domain of the Bacillus subtilis enzyme ribonuclease P (RNase P)⁶¹ and the Escherichia coli 16S ribosomal RNA (rRNA)⁷⁷. These methods have proven useful in many laboratories and SHAPE is now an important tool for determining RNA structure^30,68–70.

Methods for analysing RNA structure and interactions in cells are highly valuable as tools for investigating RNA biology⁷⁸. Until 2013, no data existed for acylation-based structure mapping of RNAs in intact cells. To address this challenge, new acylating reagents were designed with improved water solubility and tuned reactivity for cellular experiments⁴⁴. The acylimidazole reagents 2-methyl-3-furoic acid imidazolide (FAI; 19) and NAI (5) (FIG. 2b) exhibited half-lives in water of 73 min and 33 min, respectively, which are several fold longer than that of the isatoic anhydrides NMIA (2) and 1M7 (1). In experiments with intact cells, this resulted in markedly higher RNA acylation than fast-acting isatoic anhydrides, which likely hydrolyse before reaching the cellular RNA targets. Analysis of rRNA structures revealed that the reagent 5 yielded robust acylation data in mammalian, insect, yeast and bacterial cells⁴⁴. Both acylimidazole compounds (19 and 5) were validated for their ability to acylate 5S rRNA in mouse embryonic stem cells. Notably, the compound NAI (5) could be used to map not only cytosolic RNAs but also nuclear RNAs, showing effective penetration of both the plasma and nuclear membranes. Substantial differences were found in RNA structure between in vitro and in vivo SHAPE experiments with NAI (5), which could be attributed to RNA-RNA and RNA-protein interactions occurring in the cell.

In a study aimed at broadening the applicability of the in vivo SHAPE method, the structure of NAI (5) was modified to include an azide moiety, which resulted in NAI-N₃ (20)⁷⁹ (FIG. 2b). This reagent was then used to measure the intracellular structures of 16,000 poly-adenylated RNAs in mammalian cells in a single experiment. After acylation with NAI-N₃ (200 mM, 20 min) in mouse embryonic stem cells, all acylated RNAs were pulled down via biotinylation of the azide handle and analysed using RNA sequencing. This pull-down strategy has the benefit of strongly enhancing the signal-to-noise ratio, isolating the acylation data from the background signal that results from random RNA fragmentation. The utility of this procedure, termed in vivo click SHAPE (icSHAPE), was demonstrated by analysis of the intracellular RNA-binding sites of RNA-binding proteins (RBPs)⁷⁹. Indeed, the RNA-binding protein fox-1 homologue (RBFOX) family of RBPs has a known binding motif (UGCAUG), which could be observed at nucleotide resolution with icSHAPE in mouse cells⁷⁹. Furthermore, comparison with known CLIP and NMR structural data showed that data from the icSHAPE experiments precisely matched crucial RNA residues involved in interaction with RBPs. The studies also revealed positions of N6-methylation of adenines in the mRNAs and revealed a structural change that occurs in vivo on methylation of the RNAs³⁶.

In line with this utility, a number of studies have reported insights into RNA biology attained using SHAPE. In a 2019 study, RNA structure was determined in subcellular compartments⁸⁰. To accomplish this, cells were treated with NAI-N₃ (20) and RNA structure was determined in chromatin, nucleoplasm and cytoplasm using RNA fractionation⁸¹. It was found that RNA structures are mostly stable across all studied subcellular compartments, indicating that they were likely determined at biogenesis.

SHAPE has also been conducted in bacterial cells with the classical SHAPE reagents 1M7 (1) and NMIA (2)^82–84. The group of Weeks and colleagues⁸² studied switching between the active and inactive state of the 30S ribosomal subunit in E. coli cells, which were incubated with 5 mM 1M7 for a brief period of 5 min and lysed. Ribosomal RNA was isolated and reactivities were analysed using capillary electrophoresis. The authors found that, in the 30S ribosomal subunit, 43 out of 45 helices had SHAPE reactivities that were consistent with known RNA secondary structure. Interestingly, two helices showed inconsistencies and were part of the neck region of the 30S subunit, next to the decoding site. These data implied that the conformation of this neck region can be used to differentiate between free and translating 30S subunits. In 2018, the isatoic anhydride reagent 1M7 (1) was employed to map the structure of 194 endogenous mRNA transcripts in E. coli cells⁸³. Cultured cells were treated three times with compound 1 and subsequently lysed, after which isolated RNA was mapped. The study revealed that mRNA structures in E. coli are very diverse in reactivity profiles of coding regions. Importantly, it was found that the structure of mRNAs in living cells is largely different to those in cell-free structures.

A 2017 study directly compared SHAPE reagents for their applicability in cellular experiments⁸⁵. All tested reagents (1, 2, 5, 19 and 20) were capable of robustly acylating U1 small nuclear RNA in vitro. However, relative to isatoic anhydride reagents, NAI (5) and NAI-N₃ (20) produced considerably higher levels of acylation in vivo. The acylimidazole NAI (5) has been shown to provide useful SHAPE mapping in a variety of cell types, including mammalian, insect, yeast and bacterial cells⁷⁹. SHAPE mapping with 1M7 has also been reported in mammalian and bacterial cells^83,86. It has been suggested that highly reactive reagents such as 1M7 might be preferred in bacterial studies, as their half-life is shorter than the timescale of bacterial cell division⁸⁷. However, the duration of reagent exposure to cells (as opposed to reagent half-life) in mapping experiments is relatively short (shorter than the cell-division cycle) for both isatoic anhydride and nicotinate reagents.

In summary, the discovery that 2′-OH acylation stops reverse transcriptase and the subsequent development of SHAPE technology has enabled the high-resolution analysis of RNA structure, and in-cell SHAPE experiments conducted under biologically relevant conditions are improving our understanding of the roles of RNA structure and interactions in biology.

RNA functionalization and labelling

Several strategies for labelling RNA with functional groups using acylation chemistry have been reported (FIG. 3). Early investigators recognized that individual nucleotides and nucleosides could be labelled by acylating the 2′-OH or 3′-OH groups. Indeed, pioneering work was performed in the 1980s by Hiratsuka^75,76, who showed that fluorescent ATP and GTP analogues were readily prepared by reacting the nucleosides in water with an excess of isatoic anhydride (21) or NMIA (2) (FIG. 4a), producing the desired products at yields of 25–60%⁷⁶. The resulting compounds, 22 and 23 (FIG. 4b), exhibited solvatochromic fluorescence with high quantum yields in DMF and moderate quantum yields in water. The N-methylanthraniloyl-labelled nucleotides 23 exhibited slightly red-shifted emission compared with anthraniloyl derivative 22. Interestingly, the modified nucleoside triphosphates maintained considerable substrate activity for their target enzymes such as adenylate kinase and guanylate kinase, myosin ATPase and Na⁺/K⁺-ATPase, with only slightly altered K_m values, enabling the use of these labelled nucleotides in biochemical studies. Subsequent studies used these methods to study kinesin movement along microtubules^42,43.

Fig. 3 |. Overview of functional groups introduced into RNA using acylation chemistry.

RNA has been reported to be functionalized with several strategies, including via acylation with biotin moieties^88,89, hydrophobic groups⁶⁶, photocages⁵² and azide^62,63,79, as well as alkyne handles^64,66. Photocages and azide groups can be removed on light-triggered and reductive deacylation, respectively, to liberate the RNA. Azides can be reduced to stable amine-bearing groups, which can be sequentially modified using standard amide coupling⁶². Azide-containing and alkyne-containing RNA can be further functionalized using Huisgen 1,3-dipolar cycloaddition^62,66,79. R, variable functional group X, O or C.

Fig. 4 |. Functionalization and labelling of (oligo)nucleotides using acylation chemistry.

a | Scheme of labelling nucleotide triphosphates with fluorescent anhydrides 2 and 21, resulting in a mixture of 2′-hydroxyl-acylated and 3′-hydroxyl-acylated triphosphates, b | Molecular structure of fluorescently labelled nucleotide diphosphates and triphosphates (22 and 23), c | Molecular structure of biotin-bearing acylation reagents (24 and 25) and their application for purifying RNA. A mixture of RNA and DNA is incubated with the biotin probe at 65 °C, which selectively reacts with RNA. In turn, RNA can be purified by streptavidin capture. d | Scheme of the synthesis of water-soluble acylating reagents (26, 6 and 27) that include a handle for biorthogonal functionalization. e | Molecular structure of compounds used for hydrophobic functionalization of RNA (28–33). DMF, dimethylformamide.

The successful acylation of single nucleosides and nucleotides at the 2′-OH and 3′-OH positions inspired researchers to employ a similar strategy to label 2′-OH groups in RNA. The group of Fabis, Laurent and colleagues⁸⁸ described a possible method for purifying RNA samples contaminated with DNA. A biotin-labelling reagent was designed based on NMIA (2) and functionalized with a polyethylene glycol linker containing a disulfide bond and a terminal biotin group, resulting in compound 24 (FiG. 4c). In an initial proof-of-concept experiment, a mixture of 27-nucleotide RNA and the analogous ssDNA were incubated with 24 at 65 °C (REF⁸⁸). Using streptavidin beads, labelled oligonucleotides were captured and sequentially digested and analysed by liquid chromatography-mass spectrometry. Gratifyingly, it was found that RNA was selectively labelled as expected with a selectivity ratio of 10.7 over DNA, given that DNA lacks 2′-OH groups. Next, the authors successfully separated a 1,500-nucleotide human immunodeficiency virus RNA from a 20-kb calf genomic DNA by incubation with 24, capture with streptavidin beads and subsequent disulfide cleavage with dithiothreitol. Depending on the concentrations of the acylating reagent, an RNA isolation yield of 27% and selectivity ratio of 57 relative to the DNA could be achieved; this selectivity was attributed to the presence of 2′-OH groups in RNA. Of note, after disulfide cleavage, endogenous RNA was not recovered, but a 2′-OH label remained on the RNA. However, this 2′-OH label did not markedly interfere with reverse transcription, as evaluated with reverse transcriptase PCR. In later work⁸⁹, the authors optimized the structure of the labelling reagent by introducing a pyrimidine group, resulting in compound 25 (FiG. 4c), which could acylate RNA at room temperature instead of at 65 °C. The improved reactivity was attributed to the increased electrophilicity and water solubility of 25 compared with 24. Furthermore, the selectivity ratio of RNA:DNA capture was improved by 50%, and yields of RNA isolation were still good (66%).

The limited solubility of NMIA (2) in water limits its use for acylating RNA at stoichiometric levels. Ogle and colleagues^64,65 addressed this issue by developing water-soluble isatoic anhydrides. Isatoic anhydride was N-alkylated with N,N-dimethylaminoethylene chloride to produce tertiary-amine-containing anhydride 26, which was sequentially alkylated to obtain highly water-soluble quaternary ammoniums 6 and 27 (FIG. 4d). The reactivity of these new compounds was first assessed on proteins and they were shown to readily react with bovine serum albumin with an efficiency of up to 88%, rendering them potentially interesting bioconjugation reagents⁶⁴. By alkylating the tertiary amine with azides or alkynes, reactive handles could be introduced into bovine serum albumin, which could be sequentially further functionalized with a fluorophore, emphasizing the versatility of this reagent. In a later study⁶⁵, the RNA-acylating efficiency of these water-soluble isatoic anhydrides was investigated by performing RNA SHAPE on a synthetic PfU3 small nucleolar RNA in vitro. The results correlated well with those obtained with the established SHAPE reagent NAI (5). Cellular experiments are warranted to fully validate the utility of water-soluble isatoic anhydrides as RNA SHAPE reagents, but the in vitro results are promising.

Another acylation-based labelling strategy makes use of an appended azide group. Compound 28 (FIG. 4e) bears an azide at the end of an alkyl chain, and it was envisioned that this could be used as a bioconjugation handle when reacted with an appropriate alkyne⁶². RNA treated with compound 28 was sequentially incubated with TAMRA-DBCO, a strained alkyne bearing a rhodamine fluorophore. After RNA purification, a strongly emissive band was observed at 580 nm, confirming successful bioconjugation with the dye. In the same study, a related azide-substituted acylating agent with a longer chain (29) (FIG. 4e) was shown to acylate RNA at superstoichiometric levels; reduction of the azide group with water-soluble phosphine resulted in amine groups appended to the RNA. This provides a simple approach to amine functionalization of the biopolymer.

The solubility of acylating agents is especially important for stoichiometric functionalization, which makes achieving acylation of RNA with hydrophobic groups challenging. To address this challenge, Velema and Kool⁶⁶ developed the dimethylamino-substituted imidazole leaving group 7, which was expected to confer improved solubility to less polar acylating groups. It was tested successfully with several non-polar groups (30) (FIG. 4e), although its acylating efficiency was only modest, with ~1–3 groups on a 15-nucleotide RNA. Treatment with compound 31 (100 mM, 16 h) (FIG. 4e) enabled the modification of RNA with several heptyl carbonate groups; the resultant RNA exhibited unusual solubility in organic solvents, with 86% solubility in ethanol and 22% solubility in acetonitrile as compared with water. Similar hydrophobic modifications of RNA might prove useful in cellular delivery and targeting applications, in which increased lipophilicity might assist cellular uptake. Interestingly, heptyl-acylating reagent 31 (FIG. 4e) exhibited improved acylating yields compared with reagents with smaller alkyl groups. A similar trend was observed by Powner and Fernández-García⁹⁰, who found that imidazole-activated compound 33, which bears an octanoyl group, showed markedly higher acylating efficiency than acetyl imidazole 32 when reacting with glycerol-3-phosphocholine. One possible explanation is that the highly amphiphilic nature of these compounds results in the formation of aggregated structures that favour 2′-OH acylation.

These examples clearly showcase the potential for labelling RNA with a wide range of modifications, including fluorescent groups and affinity tags. It is expected that more labelling and conjugation methods based on 2′-OH acylation with high efficiency and selectivity will be devised in the years to come.

Protection of RNA from degradation

RNA is a relatively unstable biopolymer due to the nucleophilicity of the 2′-OH group, which can attack the neighbouring phosphate to form a 2′,3′-cyclic phosphate and cleave the original phosphodiester bond (FIG. 5a). This reaction can occur spontaneously (especially at elevated pH) or can be catalysed by ribonuclease enzymes. One strategy to stabilize RNA is modification of the 2′-OH group to prevent the nucleophilic attack. For example, 2′-O-methyl ethers are introduced into phosphoramidite monomers and are employed during solid-state RNA synthesis, resulting in enhanced stability^91,92. The downside of this strategy is that it is not applicable to large endogenous RNAs. For large RNAs, in vitro transcription via specialized mutant polymerases can be used⁹³.

Fig. 5 |. Acylation protects RNA from degradation.

a | Mechanism of RNA hydrolysis. The 2′-hydroxyl is deprotonated by a base and attacks the neighbouring phosphate, resulting in a cyclic phosphate and strand cleavage. b | Selective 2′-hydroxyl acylation analysed by primer extension (SHAPE) reactivities of 80-nucleotide thiamine pyrophosphate (TPP) riboswitch RNA determined by ribonuclease R (RNAse R)-mediated degradation⁹⁶. A marked change in reactivities was observed in the absence and presence of the TPP ligand. Reactive sites are shown in red (high reactivity), orange (medium reactivity) and black (no reactivity). Part b reproduced with permission from REF.⁹⁶, ACS.

This vulnerability to nucleophilic attack by the 2′-OH could, in principle, be overcome by post-synthetic acylation of this group in RNA. In 1990, Ovodov and Alakhov⁹⁴ explored this possibility on mRNA. Calcitonin precursor mRNA was incubated with 8% acetic anhydride in 1 M sodium acetate buffer at pH 7.0 to generate acetylated RNA, which was sequentially purified by precipitation. The level of acetylation, as assessed by the hydroxam reaction, was reported to be 70–75%. When incubated with pancreatic ribonuclease, the acetylated calcitonin precursor mRNA appeared fully stable on electrophoresis, whereas the non-acetylated control mRNA was digested. Surprisingly, acetylation did not seem to affect translation in a cell-free system from wheat germ; since the position of acylation was not confirmed, it is possible that acylation at amines on bases and/or on 2′-OH groups were present. The rate of product formation, as determined with radiolabelled leucine, appeared to be unaffected by the presence of the acetyl groups, and the translation products displayed correct mobility on gel electrophoresis. These results present an attractive method for stabilizing mRNA, although, to the best of our knowledge, similar experiments have not yet been conducted in cells.

In a patent filed in 2003, Goldsborough⁹⁵ claimed the use of acylation to protect RNA from degradation. It was envisioned that RNA could be temporarily acylated for stabilizing purposes and subsequently deacylated, either chemically or enzymatically. Proposed acylating reagents included anhydrides and acyl chlorides, to be applied in organic solvents, and chemical reversal was suggested to be performed with base or acid under aqueous conditions or using potassium cyanide as a catalyst. No experimental data were presented, but this intriguing concept deserves more exploration.

Kadina et al.⁶³ used NAI-N₃ (20) to temporarily protect RNA from nuclease degradation. When an 18-nucleotide RNA was acylated with, on average, 12 groups, it exhibited protection against ribonuclease H (RNase H)-mediated degradation compared with unmodified RNA; after a 30-min incubation, 35% of acylated RNA was cleaved, whereas 75% of unmodified RNA was cleaved. This moderate result indicates that acylation could potentially be useful to protect against nuclease-mediated degradation, but to reach full protection, it is likely that nearly all 2′-OH groups need to be acylated.

An earlier study from 1966 showed that tRNA and shorter tRNA fragments can be partially protected from nuclease-mediated degradation by acylation⁵⁹. To study this, tRNA was acylated with 1-C¹⁴ acetic anhydride and sequentially incubated with nucleases. After purification, the amount of intact tRNA was determined with a radiometer. It was found that, at 15 °C, viper venom phosphodiesterase hydrolysed acetylated tRNA at a slower rate than the control tRNA; however, this protective effect was not observed at 37 °C. Interestingly, acylation completely protected tRNA and tRNA fragments from digestion by spleen phosphodiesterase, which requires the 5′-OH to be available. To remove potential acetyl groups from the 5′-OH position, acylated tRNA fragments were first incubated with pancreatic RNase and sequentially exposed to spleen phosphodiesterase. The tRNA fragments remained intact after this treatment, indicating that spleen phosphodiesterase requires substantial amounts of 2′-OH groups to be available and that acylation might be a possible strategy for protecting tRNA from degradation.

Taking advantage of the concept that RNA can be protected from nucleases by acylation, Weeks and colleagues^96,97 developed a clever alternative SHAPE method using exoribonuclease in place of primer extension. They recognized that analysis of short RNAs using primer extension can be challenging due to the need for a primer that is often similar in size as the RNA of interest. To overcome this problem, it was envisioned that a SHAPE method that instead analyses nuclease degradation products might be useful for structural analysis of short RNAs. After screening several ribonucleases, one was found that degrades structured RNA but not RNA with 2′-O-methyl modifications, and was used as a model for 2′-OH-acylated RNA. An 80-nucleotide thiamine pyrophosphate (TPP) riboswitch RNA was incubated with compound 1 and sequentially subjected to ribonuclease-mediated degradation. A strong banding pattern was observed when analysed by gel electrophoresis, whereas the untreated control RNA was fully degraded. The advantage of this method was demonstrated by obtaining a single-nucleotide-resolution structure of the TPP riboswitch RNA in the unbound state (that is, not bound to TPP), which was still unknown, and compared with the TPP riboswitch RNA in the TPP-bound state. Three main differences were discovered between both states (FIG. 5b). First, the loops of the TPP riboswitch RNA in the unbound state are fully flexible. Second, with the exception of two base pairs, non-canonical base pairs were not stable in the ligand-free state. Third, the P3 helix was significantly shifted in the unbound form compared with the ligand-bound state (FIG. 5b). This example illustrates how protection of RNAs from nuclease through acylation can be exploited to develop an improved research tool to study RNA structures.

The possibility of protecting the 2′-OH groups in RNA through acylation can be a powerful tool to protect RNA from degradation. This is useful in structural analysis and could potentially be useful for temporary protection and storage, in cases where these acylating groups can be removed on demand to restore the native RNA molecule in high yield.

Synthetic tRNA acylation

There is ongoing interest in chemical methods to reprogram the genetic code⁹⁸. One crucial yet challenging aspect of this endeavour is to charge tRNAs with artificial amino acids. Naturally, tRNAs are aminoacylated with their cognate amino acids, which they carry to the ribosome. Here, mRNA is translated into protein based on the mRNA sequence, which matches to a 3-nucleotide sequence (anticodon) on the tRNA⁹⁹. When tRNAs are charged with artificial amino acids, the resulting protein or peptide will bear these unnatural amino acids. An initial challenge for this research is charging the tRNA with the new amino acid via an acyl linkage. Pioneering work was conducted by Schultz and colleagues¹⁰⁰ in the early 1990s, who showed that the 2′-OH or 3′-OH group of the dinucleotide 5′-phospho-2′-deoxyribocytidylylriboadenosine (pdCpA) could be efficiently acylated with cyanomethyl-activated esters of N-protected amino acids (FIG. 6a) in DMF, without the need for protection of the exocyclic amines. This was an important extension to previous work by Hecht et al.¹⁰¹, who had used 1,1’-carbonyldiimidazole to activate carboxylic acids to acylate the 2′-OH or 3′-OH of dinucleoside diphosphates. Subsequent enzymatic ligation of the acylated pdCpA to abbreviated tRNA resulted in charged tRNAs. In addition, pdCpA was efficiently acylated with cyanomethyl-activated esters in yields of 76–87% with varied amino acids¹⁰⁰.

Fig. 6 |. Charging tRNA with unnatural amino acids using acylation chemistry.

a | Scheme of acylation of the dinucleotide 5′-phospho-2′-deoxyribocytidylylriboadenosine using cyanomethyl-activated esters for use in charging transfer RNA (tRNA) with unnatural amino acids¹⁰⁰. In a subsequent step, the protecting group (PG) can be removed, b | Chemical structures of activated acylating reagents used in combination with flexizyme to amino-acylate tRNAs. DMF, dimethylformamide; R, varied amino acid side chains.

This remains a chief method for synthetically charging tRNAs^102–106. For example, the group of Hecht and colleagues¹⁰⁷ has adopted this method to study the mechanism of type IB topoisomerase-mediated DNA relaxation. A tyrosine group in the active site of this enzyme was replaced with 11 artificial tyrosine analogues, which were charged onto a suppressor tRNA using the acylating method developed by Schultz and colleagues¹⁰⁰ and were subsequently used in a cell-free translation system to yield the mutated topoisomerase enzymes.

In 2014, the group of Kwiatkowski, Forster and colleague¹⁰⁸ developed a method that enables acylation of the natural dinucleotide 5′-phospho-ribocytidylylriboadenosine (pCpA) substrate, instead of the unnatural deoxy pdCpA substrate that had been used previously. One potential concern is that acylation could occur at the internal 2′-OH position, which is not present in the unnatural deoxy pdCpA substrate. The authors showed that they could obtain exclusive acylation at the terminal 2′-OH and 3′-OH groups when not exceeding a 3:1 ratio of activated amino acid to pCpA, which was analysed by mass spectrometry and high-performance liquid chromatography. The selectivity for the terminal 2′-OH and 3′-OH groups was explained by the lower pKa (~12.3) than the internal 2′-OH group (13.0–13.9)¹⁰⁸. One potential advantage of using the natural pCpA substrate over pdCpA could be higher incorporation yields during translation.

Ribozyme.

An RNA molecule that can carry out an enzymatic function, such as ligation or hydrolysis reactions.

An alternative aminoacylating strategy based on a ribozyme was ingeniously devised by Suga and colleagues¹⁰⁹. The bifunctional ribozyme dubbed ‘flexizyme’ was originally generated using directed in vitro evolution and can be selectively aminoacylated at its own 5′-OH group using cyanomethyl-activated esters. A second catalytic domain within the flexizyme recognizes a tRNA and transfers the aminoacyl group from the 5′-OH group of the ribozyme to the 3′-end of the tRNA. Subsequent versions of this ribozyme were able to recognize a multitude of tRNAs¹¹⁰ and unnatural amino acids¹¹¹. The original flexizyme functioned optimally with aromatic amino acids such as activated phenylalanine (34) (FIG. 6b), because the aromaticity was essential for recognition by the ribozyme. To overcome this limitation, Suga and colleagues¹¹¹ cleverly used aromatic leaving groups such as dinitrobenzyl (35) and chlorobenzyl (36) (FIG. 6b) instead of cyanomethyl to activate the carbonyl, and could then use non-aromatic amino acids. Flexizyme has proven to be an important tool for reprogramming the genetic code, and multiple research groups have adopted this strategy^112–115.

Overall, chemical RNA acylation has been an indispensable method for the development of synthetic tRNAs and has been important for obtaining peptides and proteins bearing unnatural amino acids via translation.

Reversible acylation for caging studies

Early studies showed that acylation of RNA might result in the loss of its biological function. For example, Knorre et al.⁴⁹ and Ovodov and Alakhov⁹⁴ showed that tRNA loses its secondary structure and biological recognition, as determined by spleen phosphodiesterase activity in aqueous buffer, when 80–90% of the nucleotides are acetylated. Studies have considered that, if such acylating groups could be designed to be removed with high spatial and/or temporal resolution, RNA structure and function could, in principle, be reactivated on demand, enabling the study of its functions over time. Such ‘caging’ studies have been highly useful for the study of RNA^40,116,117, but have been almost exclusively limited to short, chemically synthesized RNAs, and have not utilized post-transcriptional acylation chemistry. As post-synthetic acylation of RNA is carried out in one step, and functions equally well with short or long RNAs, it has the potential to add considerable utility to the caging field, by making it accessible to laboratories that do not possess synthetic chemistry expertise and to RNAs longer than those that can be synthesized chemically. Acylation of the 2′-OH position of RNA results in ester or carbonate linkages, depending on the acylating reagent used. These relatively stable groups are challenging to remove in a selective manner without degrading the RNA molecule. Thus, milder approaches were devised to selectively reverse the acylation reaction.

In 2018, Kool and colleagues^52,63 reported the design of chemical and optical strategies for reversing the acylation of RNA in a bioorthogonal fashion (FIG. 7a). In a first test of this concept, the SHAPE reagent NAI-N₃ (20)⁷⁹ was used to reversibly acylate RNA⁶³. In this case, the azide group of NAI-N₃ was not used as a ‘click’ label for RNA pull-down (as performed in the aforementioned icSHAPE approach) but rather as a masked amine that can be revealed on azide reduction. The amine undergoes intramolecular lactam formation, which removes the acyl group, liberating the native RNA (FIG. 7a). Using a molecular beacon experiment with a moderately poly-acylated RNA (with an average of five acyls per strand), acylation with NAI-N₃ (20) was shown to inhibit hybridization by >95%. Reversal of this acylation by treatment with the reducing agent diphenylphosphinobenzoic acid (DPBA; 37) (FIG. 7b) (20 mM, 1 h, 37 °C) resulted in complete recovery of hybridization. The utility of this method to chemically control a larger, biologically relevant RNA was demonstrated with the Spinach aptamer³³, a widely used folded RNA that binds the 3,5-difluoro-4-hydroxybenzylidene imidazolinone (DFHBI) fluorophore¹¹⁸. The aptamer must fold correctly to yield a fluorescent signal with the DHFBI dye. When the 105-nucleotide acylated Spinach aptamer was incubated with the fluorogenic DFHBI dye, only 2% of the fluorescence intensity was observed, relative to a non-acylated control aptamer. Treatment with diphenylphosphinoethylamine (DPPEA; 38) (5 mM, 1 h) efficiently removed the acyl groups and the original fluorescence signal was quantitatively recovered (FIG. 7c,d).

Fig. 7 |. Reversible acylation for controlling RNA activity.

a | Scheme of the RNA acylation and phosphine-mediated deacylation reaction. b | Molecular structures of phosphines (37 and 38) used to uncage NAI-N₃-acylated RNA. c | Fluorescence intensity of the Spinach aptamer (540 nm) before and after treatment with NAI-N₃ (100 mM) in water or buffer and after deacylation with diphenylphosphinoethylamine (DPPEA; 5 mM) (38). The error bars represent standard deviation and P values: ***, P < 0.001; ns., not significant. d | Visible emission of the Spinach aptamer in microcentrifuge tubes before and after treatment with NAI-N₃ (100 mM) and DPPEA (5 mM). e | Scheme of the RNA acylation and light-triggered deacylation reaction using compounds 39 and 40. f | HeLa cells were transfected with a Broccoli aptamer (5 μg) and incubated with 3,5- difluoro-4-hydroxybenzylidene imidazolinone (DFHBI; 40 μM), resulting in a strong green fluorescent signal. Cells that were transfected with Broccoli aptamer and treated with compound 40 (100 mM) remain dark. On light exposure, the fluorescence is readily restored. DMSO, dimethyl sulfoxide. Parts c and d reproduced with permission from REF.⁶³, Wiley-VCH. Part f reproduced with permission from REF.⁵², ACS.

Stoichiometric acylation of RNA offers other opportunities for reversible control of structure and function. Photocontrol, if possible, enables the modulation of biomolecules, both in time and in space. With this goal in mind, Velema et al.⁵² tested an acylation strategy to control RNA using photocleavable bonds. Compounds 39 and 40 (FIG. 7e) were designed to have an activated carbamate group that can react with 2′-OH groups in RNA to form carbonate linkages and acylate RNA with photolabile ortho-nitrobenzyl groups. Compound 39 was used to acylate a hammerhead ribozyme, which resulted in complete loss of its hydrolytic activity. Exposure to 365-nm light restored the catalytic activity to 80% of its original activity. The applicability of this method to photochemically control RNA in a cellular environment was tested with a 237-nucleotide RNA construct for the dimeric Broccoli aptamer¹¹⁹. When the aptamer RNA was acylated with compound 40 and transfected into HeLa cells, no fluorescence was observed in the presence of the fluorogenic DHFBI dye. However, when the cells were exposed to 365-nm light, a strong green emissive signal was observed, which was comparable in intensity with that of the control untreated aptamer (FIG. 7f). This experiment established photocontrol over RNA folding in living cells.

The strength of these caging approaches lies in the simplicity of the method. RNA of any length can be acylated in a single, post-synthetic step, avoiding the need for complicated nucleoside and oligonucleotide syntheses and advanced equipment. However, the utility of these caging strategies still remains to be validated with more biologically relevant RNAs in cellular and animal experiments.

Viral RNA.

RNA that defines the genetic material of a virus; this can be single-stranded or double-stranded in structure.

Other applications

In addition to the aforementioned applications of RNA acylation, several other uses have been suggested.

Miyamae¹²⁰ investigated the possibility of acylating viral RNA to inactivate it and applying the acylated product as a vaccine. Sendai virus was treated with acetic anhydride and incubated for two months at 23 °C to establish full viral inactivation, as assessed by inoculation into hen eggs. Mice were nasally inoculated with three doses of inactivated Sendai virus and were sequentially infected with a 50% egg infective dose (EID₅₀) of the virus of 10^6.0 to evoke a viral challenge. Interestingly, the acylated RNA induced complete protection against Sendai virus, as analysed with immunostaining of samples taken from different parts of mouse lungs. In an earlier study, Schell and colleagues¹²¹ had shown that 2′-OH acetylation results in complete loss of antiviral activity of RNA. Oligonucleotides can evoke interferon-induced antiviral reactions and, as such, are potentially interesting antiviral agents¹²². It was found that poly(A), poly(C) and poly(I) ribonucleotides induced viral interference, as measured by plaque formation, in primary rabbit kidney cells, which was caused by vesicular stomatitis virus. By contrast, the acetylated RNA analogues showed up to 78% less plaque reduction. It has been suggested that the decrease in antiviral properties could stem from a reduction in 2′-OH availability, which might have an important role in the molecular mechanism of the capacity of RNA to induce an interferon response¹²¹.

Prebiotic RNA synthesis.

Refers to part of the ‘RNA World’ hypothesis that suggests that RNA molecules proliferated before DNA and proteins and relied on self-replication.

In a study of potential prebiotic RNA synthesis, Sutherland and colleagues^123,124 showed that the 2′-OH group of RNA can be acetylated under prebiotically credible conditions and that this might be exploited to protect the 2′-OH position to yield exclusive 3′/5′ oligomerization. Several acetylation agents were tested, of which acetylimidazole (32) showed desirable acylating properties. When tested as part of a template-mediated ligation reaction, acetylimidazole (32) preferentially acetylated the terminal 2′-OH position of a 10-nucleotide oligoribonucleotide. The terminal 3′ phosphate group was sequentially activated with N-cyanoimidazole (41) and reacted with the 5′-OH of the neighbouring strand to complete the ligation reaction (FIG. 8a). Without the prior 2′-OH acetylation, cyclic 2′/3′ phosphates were readily formed. Finally, acetyl groups were removed in aqueous ammonia, which did not affect the stability of the RNA backbone.

Fig. 8 |. Acylation assists prebiotic RNA synthesis and might explain drug effects.

a | Schematic representation of RNA ligation. A primer and ligator bind to a template and subsequent acylation protects the 2′-hydroxyl (2′-OH) position on the primer strand. Next, Ligation with N-cyanoimidazole results in 5′−3′ coupling. b | Hypothesized reaction of aspirin with RNA. The drug could potentially transfer its acetyl group to 2′-OH positions on RNA, which could possibly block RNA-protein interactions¹²⁵. c | Acylation of RNA can potentially disrupt RNA-protein interactions and might explain certain idiosyncratic drug effects¹²⁵.

In a final example, it has been suggested that some drugs might exhibit their activity through acylating the 2′-OH groups of RNA. Bhat, Spitale and colleagues¹²⁵ described a hypothetical scenario in which the activity of aspirin (42) is affected by its RNA-acylation properties (FIG. 8b). The 2′-OH group is crucial in many biomolecular mechanisms and, accordingly, reducing its availability through acylation could have major consequences. For example, RNA folding can be substantially disturbed by blocking the 2′-OH group, which can result in the deactivation of important RNAs¹²⁵. Furthermore, RNA-protein interactions rely heavily on 2′-OH availability¹²⁶, and acyl transfer from aspirin might alter binding affinity (FIG. 8c), potentially resulting in downstream effects that might explain the desired and/or undesired effects of aspirin¹²⁵. Experimental studies to follow up on this notion would be of interest.

Conclusions and outlook

The modification of oligoribonucleotides through 2′-OH acylation offers many fascinating possibilities, with potential applications in biotechnology, biochemistry and medicine. Although initial experiments showing the possibility of 2′-OH acylation were conducted as early as 1965 (REF.⁴⁹), it is only in the past decade that the broader potential has been recognized, initially with the development of SHAPE technology^30,53,69,74 and more recently with the demonstration of stoichiometric and superstoichiometric acylations with a broader range of acylating agents to functionalize RNA for advanced biochemical studies^52,63. Many more acylating agents and applications are anticipated in the near future.

The continuous discovery of new chemical methods for RNA acylation is likely to result in improvements in the efficiency and selectivity of acylation reagents. For example, it would be of interest to be able to control selectivity for either single-stranded or double-stranded RNA, or for acylation of specific locations of a long RNA strand. One such example is the flexizyme ribozyme, which is capable of selectively acylating the 3′-OH position of tRNAs¹⁰⁹. Extending this technology could potentially enable one to site-selectively acylate any RNA strand of interest. Furthermore, the concept of reversible acylation opens up multiple opportunities in application. For example, the resurgence of RNA as pharmacotherapy has led to increased efforts in the discovery of new technologies for intracellular delivery. The possibility of covalently attaching delivery and targeting scaffolds to RNA through 2′-OH acylation is a potentially attractive strategy to enhance cellular uptake of RNA. The combination of reversible acylation with existing technologies that rely on RNA activity are likely to become powerful methods for biological studies. For example, chemically or optically reversible acylation might serve as a general system to gain temporal control over RNA folding and function. It is envisioned that this strategy could be applied to other RNA-based technologies, such as CRISPR-Cas9 applications¹²⁷, short tandem target mimics¹²⁸ and artificial microRNAs¹²⁹ to obtain spatiotemporal control over gene expression.

In conclusion, RNA acylation has become an indispensable tool in biochemistry and biotechnology, and has enabled researchers to study its many properties, including its structure, function and reactivity. It is expected that RNA acylation will significantly contribute to the discovery of new RNA functions in the future.