Fingerprinting the junctions of RNA structure by an open-paddlewheel diruthenium compound.

RNA function is determined by its structural organization. The RNA structure consists of the combination of distinct secondary structure motifs connected by junctions that play an essential role in RNA folding. Selective 2'-hydroxyl acylation analyzed by primer extension (SHAPE) probing is an established methodology to analyze the secondary structure of long RNA molecules in solution, which provides accurate data about unpaired nucleotides. However, the residues located at the junctions of RNA structures usually remain undetected. Here we report an RNA probing method based on the use of a novel open-paddlewheel diruthenium (OPW-Ru) compound [Ru2Cl2(µ-DPhF)3(DMSO)] (DPhF = N,N'-diphenylformamidinate). This compound has four potential coordination sites in a singular disposition to establish covalent bonds with substrates. As a proof of concept, we have analyzed the reactivity of OPW-Ru toward RNA using two viral internal ribosome entry site (IRES) elements whose function depends on the structural organization of the molecule. Our study suggests that the compound OPW-Ru preferentially attacks at positions located one or two nucleotides away from junctions or bulges of the RNA structure. The OPW-Ru fingerprinting data differ from that obtained by other chemical reagents and provides new information about RNA structure features.

INTRODUCTION

RNA plays a central role in fundamental processes in all kingdoms of life (Sharp 2009). RNAs vary in primary sequence as well as in secondary and tertiary structure. The function of RNA in biology processes relies on its three-dimensional structure, which finally determines the physical shape of the molecule (Cruz and Westhof 2009; Butcher and Pyle 2011). The RNA secondary structure combines multiple structural motifs, typically stems, internal bulges and apical loops, linked by various types of junctions (Lilley 2000; Leontis et al. 2006). RNA junctions are secondary structural elements found in a wide diversity of RNA molecules and play crucial roles in RNA folding, serving as guides to the overall assembly of RNA molecules and, importantly, to RNA function.

Several approaches are available to determine the RNA three-dimensional structure at the atomic level. However, these methodologies are often challenged by the size of the RNA to be studied, or the inherent problems in the preparation of stable ribonucleoprotein complexes (Duss et al. 2015). Alternatively, the local nucleotide conformation of a given RNA can be analyzed in solution, taking advantage of chemical (DMS, DEPC) or enzymatic (RNase T1, T2, V1) reagents that attack specific nucleotides within the folded RNA molecule (Brunel and Romby 2000; Huntzinger et al. 2008; Kwok et al. 2015; Philippe et al. 2015). Selective 2′-hydroxyl acylation analyzed by primer extension (SHAPE) probing represents an advance in the structural analysis of long RNA molecules (Wilkinson et al. 2008). In this approach, RNA is treated with a hydroxyl-selective electrophilic reagent forming 2′-O-adducts with unconstrained nucleotides that can be subsequently identified by primer extension. This methodology offers the advantage of attacking nucleotides irrespective of their identity, with low influence of solvent accessibility. Furthermore, it also allows the determination of the impact of RNA-binding proteins and other RNA ligands on RNA structure (Filbin and Kieft 2011; Pineiro et al. 2013; Romero-Lopez et al. 2014; Lozano et al. 2015).

In spite of the power of SHAPE methodology (Wilkinson et al. 2008), N-methylisatoic anhydride (NMIA) fails to sense the conformation of every single nucleotide within the RNA molecule. Other compounds developed to perform differential SHAPE [benzoyl cyanide (BzCN), 1-methyl-7-nitroisatoic anhydride (1M7), 1-methyl-6-nitroisatoic anhydride (1M6), or isatoic anhydride (IA)] (Mortimer and Weeks 2007, 2009a; Steen et al. 2012) can distinguish distinct local conformations of the RNA molecule. However, none of these reagents can sense all kinds of local conformations. In particular, the residues located at the junction of helices or those joining internal bulges and stems often fail to be detected by these compounds.

Dirhodium compounds with paddlewheel structure have the capacity to form covalent links with DNA (Dunham et al. 2005) and tRNA (Rubin and Sundaralingam 1984). Although their diruthenium counterparts (Fig. 1A, left panel) have been less investigated in biological systems (Aquino 2004; Miyake et al. 2014), their interaction with nucleic acids is thought to be comparable (Chifotides and Dunbar 2005). Additionally, given that the reactivity of open-paddlewheel (Fig. 1A, right panel) is higher than that of paddlewheel diruthenium species (Barral et al. 2007, 2008), we decided to test the reactivity of the compound [Ru2Cl2(µ-DPhF)3(DMSO)] (designated OPW-Ru) (DPhF = N,N′-diphenylformamidinate) toward RNA. Numerous ruthenium compounds have been designed to interact with nucleic acid such as mononuclear ruthenium complexes with labile ligands (Busto et al. 2013; Wu et al. 2013; Adhireksan et al. 2014) or tris(chelate)ruthenium(II) compounds (Song et al. 2012; Li et al. 2015). The existence of labile ligands in the first type of compounds allows establishing direct ruthenium–nucleic acid bonds; the coordination mode implies bond angles of 90°. Tris(chelate)ruthenium(II) compounds are very stable and the interaction with nucleic acids takes place by intercalation (or semi-intercalation) without the formation of a covalent bond between the nucleic acid species and the metal center. However, open-paddlewheel diruthenium compounds allow the formation of two parallel bonds at ∼2.4 Å (Barral et al. 2007), which make possible the coordination of a phosphate or a nucleobase chemical group as a bridge between the ruthenium atoms. Moreover, these compounds have two additional lateral coordination sites that provide a singular disposition to establish covalent bonds with substrates.

FIGURE 1.

(A) Schematic representation of diruthenium paddlewheel (left) and open-paddlewheel (right) compounds. Reactive positions in the open-paddlewheel compound are marked with an asterisk. (B) ORTEP diagram representation of the crystal structure of compound [Ru2Cl2(µ-DPhF)3(DMSO)] (OPW-Ru). For clarity, only non-hydrogen atoms are shown: Ru (dark green), N (cyan), Cl (light green), O (red), S (yellow), C (gray).

Here we have analyzed the OPW-Ru compound as a novel RNA structure-probing reagent using two viral IRES elements differing in RNA structure. Our study shows that under the conditions used here, OPW-Ru preferentially attacks residues located at, or close to, junctions of the RNA structure, while nucleotides reactive toward NMIA and other SHAPE reagents are unpaired. These results support the use of this novel chemical compound as a novel reliable RNA probing compound that provides information complementary to other chemical reagents.

RESULTS AND DISCUSSION

Determination of the open-paddlewheel structure of OPW-Ru

The X-ray crystal structure determination confirms that compound OPW-Ru (Fig. 1B) has an open-paddlewheel structure. Three formamidinates (DPhF) act as bridging ligands between the two ruthenium atoms. The other two equatorial positions are occupied by chloride ligands. A DMSO molecule is coordinated to one of the axial positions while the other axial position remains free of ligands. A particular feature of this compound is the different coordination number of each ruthenium center (6 for Ru1 and 5 for Ru2, Fig. 1B). Other open-paddlewheel compounds such as [Ru2Br2(µ-DPhF)3] or [Ru2(µ-DPhF)3(κ2O,O-NO3)2] present coordination numbers of 5 or 6, respectively, for both metal centers of each diruthenium unit. Although the coordination number of the ruthenium atoms in OPW-Ru is different (6 for Ru1 and 5 for Ru2), the Ru–Ru bond distance [2.3408(4) Å, Supplemental Table 1] is significantly shorter than in the other known open-paddlewheel complexes: 2.4011(6) Å for [Ru2Br2(µ-DPhF)3] and 2.3694(4) Å for [Ru2(µ-DPhF)3(κ2O,O-NO3)2] (Barral et al. 2007). This short Ru–Ru distance indicates a stronger metal–metal bond in OPW-Ru than in the other open-paddlewheel complexes structurally characterized (Barral et al. 2007). However, the chloride and DMSO ligands are sufficiently labile to generate the fragment [Ru2(µ-DPhF)3]2+. This moiety has four vacant coordination sites in a singular disposition that is not possible to get with a mononuclear coordination compound or an organic molecule. More importantly, its disposition might allow the coordination of two or more chemical groups employing not only the equatorial but also the axial positions. This compound is soluble in water following a prior solution in DMSO, a critical feature for its application in biological systems.

Monitoring RNA structure with OPW-Ru compound

As a proof of concept, we have analyzed the structure in solution of two long RNAs, the IRES elements of two genetically distant RNA viruses, foot-and-mouth disease virus (FMDV) and hepatitis C virus (HCV), for which RNA structure is critical for function (Easton et al. 2009; Berry et al. 2011; Fernandez et al. 2011a; Filbin et al. 2013; Lozano and Martinez-Salas 2015). We have found that the compound OPW-Ru attacks specific RNA residues well above the background DMSO (Fig. 2A), as determined by primer extension analysis, a methodology widely applied to RNA structure analysis (McGinnis et al. 2009).

FIGURE 2.

FMDV IRES reactivity toward OPW-Ru analyzed by primer extension. (A) Integrated peak areas for OPW-Ru treated (red) and untreated (green) reactions obtained after QuSHAPE software processing. (B) Normalized RNA reactivity as a function of the nucleotide position. RNA reactivity is colored according to the scale shown on the right. Values correspond to the mean ± SD of three independent experiments.

To investigate the optimal conditions for OPW-Ru to react with RNA in solution, we set up standard reactions (2 pmol of RNA prefolded in 100 mM HEPES, pH 8.0, 0.5 mM MgCl2, 100 mM NaCl) using OPW-Ru in two concentrations (10 and 100 µM) and different reaction times (15, 30, and 60 min) at 37°C. RNA reactivity toward OPW-Ru depends on both concentration and time (Supplemental Figs. 1, 2). The average reactivity (±SD) found under the optimal RNA probing conditions (10 µM, 15 min at 37°C) in three independent experiments (Fig. 2B) demonstrated that this compound recognizes unique positions within the RNA molecule. Longer incubation time (60 min) or higher concentration (100 µM) results in a massive attack to nucleotides around 218–220 positions, reducing the sensitivity of the assay to reveal other attacks after normalization.

Importantly, OPW-Ru reactive positions strongly differ from those reactive toward NMIA. Representation of the OPW-Ru reactive residues on the secondary structure of the FMDV IRES element, which has been determined by using a combination of chemical and enzymatic reagents, mutational analysis, and sequence covariation data (Fernandez-Miragall et al. 2009; Fernandez et al. 2011a; Lozano et al. 2014), indicated that this compound preferentially attacks nucleotides located at, or close to, the junctions of the RNA structure (Fig. 3A, left panel), though not all residues located one or two nucleotides away from junctions are reactive. An exception to the rule is the strong reactivity of C343 toward OPW-Ru located on a loop, although this position is not reactive toward DMS or NMIA. Similarly, three consecutive positions with intermediate reactivity located in the middle of a stem, predicted to form noncanonical base pairs, do not obey the rule. In contrast to OPW-Ru reactivity, SHAPE reactivity determined with NMIA preferentially attack residues located in loops, internal bulges, and spacers between structural domains, as it also occurs with DMS (Fig. 3A, right panel).

FIGURE 3.

(A) FMDV IRES secondary structure showing OPW-Ru (left) and NMIA plus DMS (right) reactive positions. The location of IRES domains (2–5) is indicated. Nucleotide numbers are used as in Lozano and Martinez-Salas (2015). (B) RNA reactivity of the apical region of domain 3 (nts 120–262) toward NMIA (light blue), IA (dark blue), 1M7 (light green), 1M6 (dark green), and OPW-Ru (red). OPW-Ru reactive positions are depicted on A (left panel) and B.

To explore the possibilities of OPW-Ru for RNA probing, we focused our attention on the apical region of domain 3 (nts 120–262), which has been analyzed using multiple approaches aimed to elucidate its structural organization (Fernandez-Miragall and Martinez-Salas 2003; Fernandez-Miragall et al. 2006; Fernandez et al. 2011b; Jung and Schlick 2013; Lozano et al. 2014; Jung and Schlick 2014). The results of RNA reactivity with 1M6, 1M7, IA and NMIA reagents reactive toward unpaired nucleotides although with different sensitivities to distinct local nucleotide dynamics (Gherghe et al. 2008), superimposed with OPW-Ru denote strongly different patterns. In particular, positions U139, C151, A187, G194, G218-C220, and U255 were reactive toward OPW-Ru, but not toward 1M6, 1M7, IA, and NMIA (Fig. 3B).

To reinforce these results and provide further support to the validity of this compound for RNA structural analysis, we investigated the reactivity of a different RNA molecule for which three-dimensional structure has been determined using different approaches alone or in complex with the 80S ribosome (Lukavsky et al. 2000, 2003; Kieft et al. 2002; Melcher et al. 2003; Easton et al. 2009; Berry et al. 2010; Perard et al. 2013; Quade et al. 2015). To this end, the reactivity of the hepatitis C virus (HCV) IRES toward OPW-Ru was analyzed in parallel to NMIA (Fig. 4A). Nucleotides reactive toward NMIA were in agreement with previous data (Berry et al. 2011; Pang et al. 2011). As observed above on the FMDV IRES, the OPW-Ru reactive positions on the HCV IRES secondary structure (Fig. 4B) strongly differ from reactive positions toward NMIA (Fig. 4C). In agreement with the results found for the FMDV IRES, OPW-Ru reactive positions tend to be located close to junctions, within flexible regions of the three-dimensional HCV IRES structure (Fig. 4B, inset) as defined by Perard et al. (2013).

FIGURE 4.

(A) HCV IRES reactivity toward OPW-Ru and NMIA as a function of the nucleotide position. RNA incubated with OPW-Ru (10 µM) for 15 min at 37°C RNA reactivity is colored according to the scale shown on the right. Values correspond to the mean ± SD of three independent experiments. (B) OPW-Ru reactive positions on the HCV IRES secondary structure, compared to NMIA reactivity (C). The inset in B highlights the position of reactive residues on the flexible regions (orange dots) of the RNA structure model adapted from Perard et al. (2013).

Next, to gain information about the type of residues that are reactive toward OPW-Ru, we performed a computational analysis of the base in the reactive position and the type of structural motifs around the reactive positions on the IRES elements studied. These data indicated a higher presence of Gs at the 3′ position of the reactive nucleotide, but no preference for the identity of the base in the reactive positions of the RNAs analyzed was found. The potential correlation between structural features of both RNAs and reactivity toward OPW-Ru was also analyzed (Supplemental Tables 2, 3). We could not conclude that there is a direct correlation among the OPW-Ru reactivity and any of the structural features (solvent accessibility measured by OH-radical reactivity in both the HCV and FMDV IRES elements [Kieft et al. 1999; Lozano et al. 2014]; and interacting edges, sugar conformation, or backbone turn in the case of the HCV IRES element [Kieft et al. 2002; Lukavsky et al. 2003; Berry et al. 2010; Perard et al. 2013]). Concerning the type of secondary structure motif, a preferential attack to positions present in four-way junctions was observed (Figs. 3, 4), followed by internal bulges and three-way junctions. Importantly, molecular dynamics simulations of the conformational variability of FMDV domain 3 which capture transitions of the four-way junction between parallel and antiparallel conformations (Jung and Schlick 2014), suggest that the OPW-Ru reactive positions are located in flexible regions of the RNA structure. Additionally, RNA ligands interacting with the hexaloop (nts 166–171) perturbed the local conformation of the entire apical region of domain 3 (Lozano et al. 2015), in agreement with the flexibility of this region.

Although it is assumed that OPW-Ru forms covalent links with the RNA, the chemical groups involved in the covalent bond remain elusive. In this regard, preliminary results indicate that no interstrand links are detected using the apical region of domain 3 as substrate, suggesting that OPW-Ru forms covalent links with nearest-neighbor chemical groups on the RNA. Efforts toward the elucidation of the OPW-Ru/RNA complex three-dimensional structure are ongoing.

Conclusions

The comparative analysis of two genetically distant IRES elements differing not only in primary sequence but also in RNA structure reinforced the usefulness of the OPW-Ru compound for RNA biology. Our study shows that the reactive positions toward this ruthenium compound on two RNA molecules strongly differ from reactive positions toward NMIA, IA, 1M7, and 1M6 reagents. Under the conditions used here, complex OPW-Ru preferentially attacks residues located at or close to the junctions of the RNA structure, coincident with flexible regions of the three-dimensional RNA structure, while unpaired nucleotides reactive towards NMIA and other SHAPE reagents are mainly located in loops and internal bulges. These results support the use of this novel chemical compound as a reliable RNA probing compound that provides information complementary to other chemical reagents.

MATERIALS AND METHODS

Synthesis of [Ru2Cl2(µ-DPhF)3(DMSO)] (OPW-Ru)

The compound [Ru2Cl2(µ-DPhF)3] was prepared following published procedure (Barral et al. 2007). FTIR spectra (4000–650 cm−1) were recorded with a Perkin-Elmer Spectrum 100 with a universal ATR accessory. Elemental analyses were performed at the Microanalytical Service of the Universidad Complutense de Madrid. All reactants and solvents were obtained from commercial sources and used without further purification unless otherwise indicated.

For synthesis of compound [Ru2Cl2(µ-DPhF)3(DMSO)] (OPW-Ru), water was slowly added to a solution of [Ru2Cl2(µ-DPhF)3] (DPhF = N,N′-diphenylformamidinate) (0.43 g, 0.5 mmol) in DMSO. The solid obtained was washed with water and dried under vacuum. Yield: 0.43 g (91%). IR absorptions (cm−1): 3382w, 3061w, 3023w, 2952w, 1593m, 1523vs, 1484vs, 1451m, 1320s, 1305s, 1214vs, 1075w, 1026m, 974m, 937vs, 923s, 837w, 784m, 764s, 754s, 696vs, 658m (Supplemental Fig. 3). Elemental analysis found, %: C, 51.66; H, 4.20; N, 8.82; S, 3.38. Calcd. for H39C41N6OSCl2Ru2·H2O, %: C, 51.57; H, 4.33; N, 8.80; S, 3.36.

Crystals of compound OPW-Ru suitable for X-ray analysis were obtained from [Ru2Cl2(µ-DPhF)3] by slow evaporation of a solution of the compound in DMSO. IR absorptions (cm−1): 3382w, 3062w, 3023w, 2948w, 1592m, 1581m, 1523vs, 1485vs, 1450m, 1320s, 1304s, 1212vs, 1075w, 1027m, 974m, 937vs, 925s, 837w, 785m, 767s, 753s, 693vs, 686vs, 657m (Supplemental Fig. 4).

Crystal structure determination of compound [Ru2Cl2(µ-DPhF)3(DMSO)] (OPW-Ru)

A single crystal of OPW-Ru was mounted on a glass fiber with glue and transferred to a goniometer head. Data were collected at 293 K. A Bruker Smart CCD diffractometer was employed for crystal screening, unit cell determination, and data collection. The structure was solved by direct methods and refined by full-matrix least-squares techniques using the SHELX system of programs: G.M. Sheldrick, SHELXL-97, program for the solution of crystal structures, University of Göttingen, Göttingen, 1997. The absorption correction program SADABS was employed to correct the data for absorption.

The space group of P21/c (No. 14) was determined from the crystal systematic reflection conditions of the diffraction patterns. All non-hydrogen atoms were refined anisotropically. The hydrogen atoms attached to carbon atoms in the complex were generated and assigned isotropic thermal parameters, riding on their parent carbon atoms. A plot of the crystal structure is shown in Figure 1B. Crystallographic data for compound OPW-Ru are provided in Table 1. Selected bond distances and angles are listed in Supplemental Table 1.

TABLE 1.

Crystallographic data for compound OPW-Ru

RNA synthesis

The constructs expressing the FMDV and HCV IRES RNAs were previously described (Lozano et al. 2015). For RNA structural analysis, RNA was synthesized in vitro using SphI linearized plasmids (Lozano et al. 2014). Transcription was performed using T7 RNA polymerase, and DNA template was removed by RQ1 DNase treatment, followed by phenol extraction and ethanol precipitation. Synthesis of full-length products was verified by gel electrophoresis.

RNA reactivity reactions

Prior to treatment, in vitro synthesized RNA was folded by heating at 95°C for 2 min, snap cooling on ice for 2 min, and subsequently incubated in a final volume of 18 μL of folding mix (100 mM HEPES, pH 8.0, 0.5 mM MgCl2, 100 mM NaCl) for 20 min at 37°C. Then, prefolded RNAs (2 pmol) were incubated with OPW-Ru (10 or 100 µM) dissolved in DMSO, for 15, 30, or 60 min at 37°C. Untreated RNA was incubated with DMSO. Treated and untreated RNAs were precipitated and finally resuspended in 10 μL of 0.5× TE.

In parallel, folded RNA was treated with N-methylisatoic anhydride (NMIA) (Invitrogen) as described previously (Lozano et al. 2014). All SHAPE reagents [NMIA, 1-methyl-7-nitroisatoic anhydride (1M7), 1-methyl-6-nitroisatoic anhydride (1M6), and isatoic anhydride (IA)] (Mortimer and Weeks 2007, 2009b; Steen et al. 2012) were used at 6.5 mM. DMS treatment is described in Fernandez-Miragall and Martinez-Salas 2003.

Primer extension

For primer extension reactions, treated and untreated RNAs (2 pmol) were incubated with an antisense 5′-end fluorescently labeled primer (Lozano et al. 2014) (2 pmol) at 65°C for 5 min, 35°C for 5 min and then chilled on ice for 2 min. Primer extension reactions were conducted in a final volume of 16 µL, containing reverse transcriptase (RT) buffer (50 mM Tris-HCl, pH 8.3, 3 mM MgCl2, 75 mM KCl, 8 mM DTT) and 1 mM each dNTP. The mixture was heated at 52°C for 1 min prior to addition of 100 U of Superscript III RT (Invitrogen) and incubated at 52°C for 30 min. The enzyme was inactivated heating at 70°C for 15 min. A sequencing ladder was generated using the same transcript (1 pmol), 0.1 mM ddCTP and 0.5 mM each dNTP. NED fluorophore (Applied Biosystems) was used for both treated and untreated RNAs while FAM fluorophore was used for the sequencing ladder (Francisco-Velilla et al. 2015). Primer extension products were resolved by capillary electrophoresis.

Data analysis

Electropherograms obtained by capillary electrophoresis were analyzed using QuSHAPE software with default parameters (Karabiber et al. 2013). This process includes preprocessing of traces, signal alignment, sequence alignment, and reactivity estimation by Gaussian peak integration. The reactivity values obtained for the untreated RNA (DMSO) were subtracted from the treated (OPW-Ru or NMIA) samples to obtain the net reactivity for each nucleotide (Francisco-Velilla et al. 2015). Quantitative reactivity for individual data sets were normalized to a scale spanning 0 to 2, in which 0 indicates an unreactive nucleotide and the average intensity at highly reactive residues is set to 1.0. Data from three independent assays were used to calculate the mean (± standard deviation) reactivity.

Hydroxyl radical cleavage intensities were determined as in Lozano et al. (2014), and normalized to a scale from 0 to 1.5, where 1.0 is defined as the average intensity of highly reactive residues (Chakraborty et al. 2012). Nucleotides with reactivity values lower than half the mean reactivity are defined as solvent-inaccessible.

Coordinates

CCDC 1410351 contains the atomic coordinates and the rest of the crystallographic data of compound OPW-Ru. These data can be obtained free of charge from the Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.

SUPPLEMENTAL MATERIAL

Supplemental material is available for this article.