An uncommon [K+(Mg2+)2] metal ion triad imparts stability and selectivity to the Guanidine-I riboswitch.

The widespread ykkC-I riboswitch class exemplifies divergent riboswitch evolution. To analyze how natural selection has diversified its versatile RNA fold, we determined the X-ray crystal structure of the Burkholderia sp. TJI49 ykkC-I subtype-1 (Guanidine-I) riboswitch aptamer domain. Differing from the previously reported structures of orthologs from Dickeya dadantii and Sulfobacillus acidophilus, our Burkholderia structure reveals a chelated K+ ion adjacent to two Mg2+ ions in the guanidine-binding pocket. Thermal melting analysis shows that K+ chelation, which induces localized conformational changes in the binding pocket, improves guanidinium-RNA interactions. Analysis of ribosome structures suggests that the [K+(Mg2+)2] ion triad is uncommon. It is, however, reminiscent of metal ion clusters found in the active sites of ribozymes and DNA polymerases. Previous structural characterization of ykkC-I subtype-2 RNAs, which bind the effector ligands ppGpp and PRPP, indicate that in those paralogs, an adenine responsible for K+ chelation in the Burkholderia Guanidine-I riboswitch is replaced by a pyrimidine. This mutation results in a water molecule and Mg2+ ion binding in place of the K+ ion. Thus, our structural analysis demonstrates how ion and solvent chelation tune divergent ligand specificity and affinity among ykkC-I riboswitches. Published by Cold Spring Harbor Laboratory Press for the RNA Society.

INTRODUCTION

Riboswitches are structured RNA elements that evolved to regulate gene expression through binding of small molecules (Roth and Breaker 2009; Serganov and Nudler 2013). A compelling example highlighting the divergent evolution of riboswitches and their acquisition of different specificities is the ykkC-I riboswitch class (Nelson et al. 2017; Sherlock et al. 2018a,b). ykkC-I is widespread among bacterial phyla (e.g., Actinobacteria, Firmicutes, Proteobacteria, and Cyanobacteria). This riboswitch class is comprised of five subtypes (subtype 1, subtypes 2a–2d), which control disparate metabolic pathways using a nearly identical RNA fold (Battaglia et al. 2017; Reiss et al. 2017; Knappenberger et al. 2018; Peselis and Serganov 2018). The effector ligands for these subtypes vary in both chemical structure and cellular function (subtype 1 [guanidine], 2a [ppGpp] 2b [PRPP], 2c [{d}NDP], 2d [unknown]). The structural dissimilarity between ykkC-I effector ligands has reinforced the notion that this RNA may reside within a privileged class (Hallberg et al. 2017; Page et al. 2018) of RNA molecules. Thus, these RNAs are amenable to mutations in the binding pocket to enable tuning of ligand specificity and affinity while maintaining global architecture and regulatory function. Indeed, paired elements 1 and 3 of the ykkC-I fold (Fig. 1A) can be swapped between subtypes to change ligand specificity with only modest decrease in ligand affinity (Knappenberger et al. 2020).

FIGURE 1.

Overall structure of the Burkholderia Guanidine-I riboswitch. (A) Secondary structure. Thin lines with arrowheads denote connectivity. Base pairs are represented with Leontis–Westhof symbols (Leontis and Westhof 2001). Single hydrogen bonds are represented as dashed lines. Red triangle denotes bound guanidine. Asterisks denote residues that participate in A-minor interaction. Wild-type and crystallization construct sequences of the terminal loop of P2 are boxed and blue, respectively. Disordered residues are shown in gray. (B) Cartoon representation of the three-dimensional structure with van der Waals surface of the RNA. Green, yellow, and red spheres denote Mg2+, K+, and guanidine, respectively. (C) Van der Waals surface of the Burkholderia Guanidine-I riboswitch (gray) superimposed with D. dadantii Guanidine-I riboswitch (cartoon; orange). Dashed and solid lines depict the P2 helical axes of D. dadantii and Burkholderia, respectively.

Structures have been determined for members of each of the ykkC-I riboswitch subtypes, with the exception of the orphan ykkC-I-2d subtype (Battaglia et al. 2017; Reiss et al. 2017; Knappenberger et al. 2018; Peselis and Serganov 2018). Those structures reveal a conserved global architecture with varying sequences in paired elements, and an optional paired element in the vicinity of the ligand binding pocket. The most common effector ligand for ykkC-I riboswitches is guanidine, with subtype-1 (Guanidine-I) comprising 75% of all known ykkC-I examples, and being observed in the most diverse phyla. To further understand how RNA structure is molded by natural selection, we solved the X-ray crystal structure of the Burkholderia sp. TJI49 Guanidine-I riboswitch. Structural comparison to two previously characterized Guanidine-I riboswitch variants from D. dadantii (Battaglia et al. 2017) and S. acidophilus (Reiss et al. 2017) reveal differences in both overall global architecture and core binding-pocket structure. The most notable difference is the chelation of a K+ ion by conserved nucleotides adjacent to the guanidine binding pocket. This K+ ion participates in a metal ion triad with two adjacent Mg2+ ions. Thermal melting analyses demonstrate that the ligand bound Burkholderia sp. TJI49 Guanidine-I riboswitch is preferentially stabilized by the monovalent ion K+ relative to Na+. The [K+(Mg2+)2] metal ion triad appears to be uncommon in RNA, although it has been reported (Nakamura et al. 2012; Freudenthal et al. 2013) in the active site of some DNA polymerases.

RESULTS

Overall structure of the Burkholderia Guanidine-I riboswitch

The crystal structure of the Burkholderia Guanidine-I riboswitch was solved using the molecular replacement method (Materials and Methods, Fig. 1A,B; Supplemental Table 1). Our crystals of this RNA diffract X-rays to ∼2.4 Å resolution (refinement was limited to 2.7 Å), and contain one RNA molecule in the crystallographic asymmetric unit. Overall, the aptamer domain is comprised of three helices (paired elements P1–P3). P1 and P2 form a ∼136 Å-long helical stack (Fig. 1B; Supplemental Fig. 1). A long-range tertiary-interaction is formed through an A-minor motif composed of four conserved adenines in the stem–loop of P3 and three bases in P1. The guanidine binding-pocket is formed at the P1–P3 interface below the A-minor motif. Here, a two-residue bulge in P1 extends into the vicinity of P3 to form the guanidine binding-pocket. The Burkholderia Guanidine-I riboswitch contains an internal loop in P2 which imparts a 36° bend relative to the helical axis (Fig. 1C). The P2 internal loop forms a poorly resolved triplex between residues U65, C66, and G99.

Guanidinium recognition by the Burkholderia riboswitch

An oblong feature in the unbiased residual electron density map near the highly conserved residues G54, A55, G107, and G125 (Fig. 2A,B) was modeled as a bound guanidinium. The guanidinium interacts with the RNA on all three of its edges and one of its two faces. The ligand stacks on the nucleobase of G107. The guanidinium is coplanar with G125, whose Hoogsteen edge makes two hydrogen bonds with amines of the ligand. Two additional hydrogen bonds are formed between the guanidinium and the pro-Rp oxygen and 05′ of G108. Compared to previously reported Guanidine-I riboswitch structures, these four hydrogen bonds are more closely coplanar with the sp2-hybridized ligand, and appear to impose a slight tilt to the guanidinium toward the conserved residues G54 and A55. Here, the edge of guanidinium participates in two dipole–dipole interactions with O6 of G54 and N1 of A55. Overall, these interactions impart selectivity for guanidinium over urea through cation-π, hydrogen bond, and dipole interactions. The conserved binding pocket bases of the RNA would also discriminate sterically against alkylated derivatives of guanidinium, such as arginine.

FIGURE 2.

Guanidine-binding pocket structure. (A) The Burkholderia ykkC-I guanidine-binding pocket, colored as in Figure 1. Gray mesh depicts portion of the |Fo|–|Fc| electron density map, prior to ligand modeling, contoured at 2.0σ. Hydrogen bonds are depicted as black dashed lines. Bound guanidine is shown in yellow ball-and-stick representation. (B) 80° rotation of A. (C) Structural superposition of Burkholderia residues on the corresponding D. dadantii binding pocket (RNA, orange; guanidine, magenta). Residue numbers in parentheses correspond to PDB ID 5U3G (Battaglia et al. 2017). (D) Structural superposition on the S. acidophilus binding pocket (RNA, gray; guanidine, magenta). Residue numbers in parentheses correspond to PDB ID 5T83 (Reiss et al. 2017). Structural superpositions were performed for all non-hydrogen atoms, with residues 20–41 and 63–102 of the Burkholderia RNA omitted. (E) Van der Waals representation of A55 from Burkholderia ykkC-I and G67 from D. dadantii Guanidine-I. (F) Van der Waals representation of A55 from Burkholderia ykkC-I and G72 from S. acidophilus Guanidine-I.

The ligand-binding site of the Burkholderia Guanidine-I riboswitch differs subtly but consistently from those of its D. dadantii and S. acidophilus orthologs. After overall superposition of the RNAs, root-mean-square differences (r.m.s.d.) for all non-hydrogen atoms of the guanidinium and binding-pocket residues G54, A55, G107, and G108 are 1.1 and 0.9 Å versus the corresponding atoms of the D. dadantii and S. acidophilus riboswitches, respectively. While G54 and G125 are in indistinguishable positions in the three RNAs, A55, G107, and G108 (Burkholderia numbering) are not (Fig. 2C,D). In the new structure, the three residues shift relative to the base plane of G125 by ∼1.1 Å, and the base of A55 rotates toward its sugar edge. Concomitantly, the base of G107, with which it is in van der Waals contact, shifts. This is evident in the superimposed structures, where the base planes of G107 (Burkholderia) and the corresponding G67 and G72 (D. dadantii and S. acidophilus numbering, respectively) are offset by 22° and 16°, respectively (Fig. 2C,D) to relieve steric clash (Fig. 2E,F). Because the bound guanidinium stacks on the nucleobase of G107, the plane of the bound ligand differs between the Burkholderia complex and those of D. dadantii and S. acidophilus by 46° and 45°, respectively. This structural rearrangement improves cation-π interaction between the ligand and G107; the interplane distances shrink from 4.1 Å (D. dadantii) and 3.8 Å (S. acidophilus), to 3.4 Å in the Burkholderia structure.

A [K+(Mg2+)2] triad in the ligand binding site

Below the A-minor motif, two bulges from the paired elements P1 and P3 come into close proximity. As observed in the D. dadantii and S. acidophilus aptamer domain cocrystal structures, two metal ions are coordinated at this interface. MA, which is modeled as Mg2+, is coordinated by the pro-Sp nonbridging phosphate oxygen (n.b.p.o) of A53 and the pro-Sp n.b.p.o of C120 with distances of 2.05 and 2.45 Å, respectively (Fig. 3A,B). The coordination network of MA forms an angle of 79°. The second metal ion, MB (also modeled as Mg2+), is coordinated by the pro-Rp n.b.p.o of C120 (1.98 Å), the pro-Sp n.b.p.o of G119 (2.57 Å), and the pro-Rp n.b.p.o of A53 (2.41 Å), with interatomic angles of 91°, 90°, and 74°. All of these values are within the range of distances and angles expected for Mg2+ chelation at a coordinate precision of 0.37 Å (Auffinger et al. 2020) and the assignment of magnesium to MA and MB is consistent with the higher resolution cocrystal structure of the D. dadantii ortholog.

FIGURE 3.

Structural analysis of the metal-ion triad. (A) Metal ion network in Burkholderia Guanidine-I riboswitch. |Fo|–|Fc| electron density map, prior to ion modeling, contoured at 3.2σ (gray mesh). Modeled Mg2+ (green, MA and MB) and K+ (yellow, MC). Orange dashed lines denote metal ion coordination. (B) 160° rotation of A. (C) Structural superposition of D. dadantii Mg2+ cluster (RNA, orange; ion, gray). (D) Structural superposition of S. acidophilus Mg2+ cluster (RNA and ions, gray). (E) Structural superposition of the Burkholderia Guanidine-I riboswitch ion core and binding pocket with the S. acidophilus ppGpp riboswitch binding pocket and ion core (PDB ID:6DMC; [Peselis and Serganov 2018]). Ions from Burkholderia Guanidine-I riboswitch are shown as gray transparent spheres. (F) Structural superposition of the Burkholderia Guanidine-I riboswitch ion-core and binding pocket with the S. lipocalidus PRPP riboswitch binding pocket and ion core (PDB ID: 6DLT [Peselis and Serganov 2018]). Ions from Burkholderia Guanidine-I riboswitch are shown as gray transparent spheres. (G) Ion cluster of the closed DNA polymerase β product complex. Mg2+ and Na+ are shown in green and yellow spheres, respectively (PDB ID: 4KLL [Freudenthal et al. 2013]). (H) Structural superposition of the DNA polymerase β product complex with the Burkholderia Guanidine-I metal-ion triad (translucent purple and gray). For RNAs superpositions between ykkC-I riboswitches, structural superpositions were performed for all non-hydrogen atoms with residues 20–41 and 63–102 of the Burkholderia ykkC-I riboswitch deleted from the search. Structural superposition of Burkholderia ykkC-I and DNA polymerase β product complex is centered on the three metal ions.

Unexpectedly, in the Burkholderia structure, an electron density feature consistent with a third metal ion (MC) is present in the vicinity of MA and MB. MC is within coordination distance from five ligands: the pro-Rp n.b.p.o of A53 (3.1 Å), the pro-Sp n.b.p.o of G119 (2.9 Å), the 2′-OH of A53 (2.5 Å), the O3′ of A53 (3.2 Å), and the N7 of A55 (3.4 Å). The coordination geometry of MC most closely resembles incomplete heptacoordination. On the basis of this, the coordination distances, the refined B-factors (Zheng et al. 2017; Auffinger et al. 2020), and residual difference Fourier synthesis features after alternative refinements modeling the feature as any of the metal ions in the crystallization solution (Supplemental Fig. 2), we have modeled MC as K+. Superposition of the three Guanidine-I riboswitch structures yields good overlay of MA and MB (Fig. 3C,D). However, chelation of MC results in a displacement of the nucleobase of A55 (r.m.s.d. 1.2 Å relative to either of the D. dadantii and S. acidophilus orthologs), altering the binding pocket conformation.

The ykkC-I subtype-2a (ppGpp) and subtype-2b (PRPP) riboswitches differ, in part, from Guanidine-I riboswitches by a transversion of A55 to uridine (U49 in S. acidophilus numbering). Structural studies of the ppGpp and PRPP riboswitches (Knappenberger et al. 2018; Peselis and Serganov 2018) revealed that in both of these riboswitches, U49 coordinates a water molecule in the exact location as the K+ ion in the Burkholderia Guanidine-I structure (Fig. 3E,F). Additionally, the nucleobase of U49 coordinates a Mg2+ ion adjacent to the equivalent location of the guanidine binding pocket in Guanidine-I (Fig. 3E,F). Our structure suggests, therefore, that mutation to uridine prevents guanidine binding to subtype-2 riboswitches not only through RNA-ligand contacts, but also altered solvent and ion interactions.

While clusters of two adjacent cations are commonly observed in structured RNAs such as the ribosome (Conn et al. 1999, 2002; Hsiao and Williams 2009), M-box riboswitch (Dann et al. 2007), NiCo riboswitch (Furukawa et al. 2015), and Mn2+ riboswitch (Price et al. 2015; Bachas and Ferré-D'Amaré 2018), RNA-bound [K+(Mg2+)2] triads are not well documented. Simulations have demonstrated long residence times for symmetric K+ and Mg2+ association sites (Auffinger et al. 2003, 2004) in locations void of metal ions in crystal structures. These reports suggest that metal ion clusters may be more common than generally appreciated. To examine how common [K+(Mg2+)2] triads are, we searched for Mg2+, Na+, and K+ ions that form triads with inter-ion distances shorter than 7.5 Å in a high-resolution structure (1.98 Å) of the 70S ribosome (PDB ID: 7K00 [Watson et al. 2020]). Manual curation of the output of our algorithm for identifying ions that make at least two RNA contacts with electronegative groups yielded no metal ion triads. Biochemical evidence for three-metal ion catalysis has been reported for the Tetrahymena ribozyme (Shan et al. 1999). Despite a lack of structural corroboration for those results, a similar three-metal ion catalytic network has been observed in the product states of DNA polymerases η and β (Nakamura et al. 2012; Freudenthal et al. 2013). Like the Guanidine-I riboswitch, the product state of DNA polymerase β also coordinates two divalent ions and a monovalent ion (Fig. 3G). Structural superposition of the chelated ions and coordinating atoms from DNA Pol. β and the Guanidine-I riboswitch shows striking similarity (Fig. 3H; r.m.s.d. = 0.9 Å). This suggests that [K+(Mg2+)2] triads, while uncommon, are integral structural features of biomolecules.

Thermal melting analysis supports role of K+

Thermal melting can indicate whether specific, chelated, ions stabilize the tertiary structure of an RNA (Shiman and Draper 2000; Lambert et al. 2009). To test if K+ participates in forming the ligand-bound structure of the Burkholderia riboswitch aptamer domain, we determined its melting profiles in the presence of K+ or Na+, with varying concentration of guanidinium or urea (Fig. 4A–E). Folded, structured RNAs exhibit biphasic thermal melting profiles when plotted as the derivative of UV absorbance versus temperature. The first transition corresponds to the unfolding of tertiary structure, while the second corresponds to the melting of the secondary structure. In the absence of ligand or presence of urea, the Guanidine-I riboswitch exhibits only secondary structure melting (Fig. 4A,B). Addition of as little as 50 µM guanidinium produces a biphasic thermal melting profile, but only in the presence of K+. The midpoint of this thermal transition (Tm) increases with increased guanidinium concentration (Fig. 4C–E). While a weak phase transition is suggested by melting in the presence of Na+ at the highest guanidinium concentration (Fig. 4E), a gaussian fit did not converge on a solution using these data, and therefore a Tm could not be determined. These results support the existence of a specific K+ chelation site that is thermodynamically linked to guanidinium binding in the Burkholderia riboswitch aptamer domain.

FIGURE 4.

Thermal melting analysis of Guanidine-I riboswitch. Thermal denaturation of the Burkholderia Guanidine-I riboswitch was performed in the presence of 0.5 mM MgCl2 and either 50 mM NaCl (gray) or 50 mM KCl (yellow) and (A) no ligand, (B) 250 µM urea, (C) 50 µM guanidine, (D) 150 µM guanidine, or (E) 250 µM guanidine. Error bars represent standard error of three independent measurements. Melting temperatures (Tm) for the tertiary structure unfolding transition (K+ data only) are marked with a blue line. Insets show zoomed-in view of 35°C–55°C regions.

DISCUSSION

The Guanidine-I riboswitch is one of four known riboswitch classes that respond to guanidinium (Nelson et al. 2017; Sherlock and Breaker 2017; Sherlock et al. 2017; Lenkeit et al. 2020; Salvail et al. 2020), and is also part of a larger riboswitch class, ykkC-I, members of which recognize their respective, distinct metabolites using a conserved RNA fold (Battaglia et al. 2017; Reiss et al. 2017; Knappenberger et al. 2018; Peselis and Serganov 2018). We now demonstrate that a Guanidine-I riboswitch chelates a K+ ion adjacent to a binuclear Mg2+ cluster in its guanidinium binding pocket. While ion chelation is commonly observed in RNA crystal structures, this K+ chelation site went unrecognized in two previous crystallographic studies of orthologous Guanidine-I riboswitches (Battaglia et al. 2017; Reiss et al. 2017). The cocrystals from which those structures were derived were grown under conditions containing abundant, nonphysiological Na+ as the predominant monovalent ion, and either no K+ (Battaglia et al. 2017), or an unphysiologically low concentration of it (∼10 mM; Reiss et al. 2017). Given the conserved sequence and structure in this region of the Guanidine-I riboswitch, and the similar resolution of all three data sets, it is likely that the differences in crystallization conditions account for the discrepancy in K+ chelation.

Through thermal melting analysis, we have demonstrated that the Burkholderia Guanidine-I riboswitch tertiary fold is preferentially stabilized by the physiologically relevant K+ ion relative to Na+. While RNA tertiary structures are often stabilized to a greater extent by small, diffuse monovalent ions (i.e., Li+ and Na+), K+ chelation can disrupt this trend to enable RNA folds with exceptionally high electrostatic potential (Lambert et al. 2009). Our results establish the stabilizing effect of K+ on Guanidine-I riboswitch structure and ligand recognition. Residue A55, the nucleobase responsible for coordinating K+ and altering the organization of the guanidinium binding pocket upon chelation, is mutated from an adenine to a pyrimidine in all reported ykkC subtype-2 examples. Consistent with our analysis, all reported structures of ykkC subtype-2 examples contain a water molecule in place of the chelated K+ of the Burkholderia riboswitch, and a Mg2+ coordinated proximally to the guanidine pocket. Just as the binding of K+ molds the Burkholderia riboswitch binding pocket to suit guanidinium, water and Mg2+ appear to do the same for ppGpp and PRPP riboswitches. Together, these results show how solvent and counterions can expand the ligand specificity of a single aptamer fold.

While ykkC-I riboswitches are structurally interesting and complex, there is no evidence to suggest that this RNA fold, or any other, is more amenable to divergent evolution than other structured RNAs. Structure-guided in vitro evolution of the glmS (Lau and Ferre-D'Amare 2013; Lau et al. 2017) and purine riboswitches (Porter et al. 2017) has demonstrated that even a small number of mutations can impart dissimilar ligand specificity. With as few as three mutations, the glmS riboswitch-ribozyme shifts selectivity from requiring glucosamine-6-phosphate (GlcN6P) for catalysis, to only requiring divalent ions (glmSAAA and glmSCa). While maintaining the same global fold, glmSAAA repurposes a structural divalent cation as a coenzyme and reorganizes the GlcN6P binding site to chelate a divalent ion. The glmSCa variant results in a shift in the active site location so that the cleavage occurs +1 nucleotide relative to glmS and glmSAAA. Much like the glmS riboswitch-ribozyme, ykkC-I riboswitches alter binding pocket locally to switch specificity. Our new structure of a member of the ykkC-I riboswitch class showcases how evolution can use solvent and cation interactions to tune the small molecule selectivity of an RNA through subtle changes in binding pocket organization.

MATERIALS AND METHODS

RNA preparation

RNA was in vitro transcribed as previously described (Trachman et al. 2020); purified by electrophoresis on a 13% polyacrylamide gel (37:1; acrylamide:bisacrylamide), 1×TBE, 8 M Urea gels; electroeluted from gel slices; washed once with 1 M KCl; and exchanged by ultrafiltration into 20 mM MOPS pH 7.0, 100 mM KCl, and 10 µM EDTA (0.1 µm cutoff, Amicon Ultrafree-MC, Millipore), and stored at −20°C.

Crystallization and diffraction data collection

Burkholderia Guanidine-I riboswitch in 20 mM MOPS-KOH pH 7.0, 100 mM KCl, 10 µM EDTA was heated to 95°C for 3 min, incubated at RT for 10 min, mixed with equimolar guanidine•HCl and 5 mM MgCl2, then incubated at RT for 30 min. For crystallization, 1.0 µL of RNA solution (150 µM) and 1.0 µL reservoir solution (50 mM HEPES pH 7.5, 200 mM NH4SO4, 10 mM SrCl2, 30% PEG 3350, 3% isopropanol) were mixed and equilibrated at 288 K. Crystals were grown by sitting drop vapor diffusion under microbatch (1:1; silicon : paraffin oil). Large hexagonal rod-like crystals grew in 1–3 d to maximum dimensions of 600 × 200 × 200 µm3. Additional cryoprotection was not necessary prior to mounting the crystal in a nylon loop and vitrifying by plunging into liquid nitrogen. Data were collected at 100 K at ALS Bl501 using 12.5 keV X-radiation. Data were reduced in XDS (Kabsch 2010) with 10% of reflections flagged for RFree calculation. Data collection statistics are summarized in Supplemental Table 1.

Structure determination and refinement

The Burkholderia Guanidine-I riboswitch structure was solved by molecular replacement using the program Phaser (McCoy et al. 2007). The S. acidophilus Guanidine-I structure (PDB ID: 5T83), with nonconserved residues, the guanidine ligand, and all ions deleted, was used as the search model. The top solution had a TFZ score of 20.2. Multiple rounds of manual building and refinement were performed in Coot and Phenix.refine, respectively. Electron density for residues 31–33 was poorly defined; however, positive electron density features were observed that were modeled as inorganic phosphate and are likely part of the RNA backbone. Simulated annealing was performed every six refinement cycles at 4000 K until an RFree of 0.32 was reached. Refinement statistics are summarized in Supplemental Table 1.

Small angle X-ray scattering

SAXS experiments were performed at beamline 12-ID-B of The Advanced Photon Source at Argonne National Laboratories. Samples were prepared and purified by size-exclusion chromatography as described previously (Jeng et al. 2021), with the exception of MgCl2 being supplied in 2.5 mM. Samples, ranging in concentration from 0.25–0.9 mg/mL, were passed through a flow cell upon which they were exposed to X-rays (12 keV) for 1.0 sec, followed by a 1.0 sec rest time. Forty data sets were collected per sample using a Pilatus 2M detector positioned at 3.6 m from the sample capillary. Prior to averaging, each data set was examined for radiation damage and aggregation using Igor Pro (WaveMetrics). Guinier analysis was performed using Igor Pro. Indirect Fourier transformation was performed using the programs DATGNOM and GNOM (Petoukhov and Svergun 2007).

Thermal melts

Thermal melts were performed as previously described (Trachman and Draper 2013). Briefly, thermal denaturation was carried out on a Cary 400 spectrophotometer with temperatures monitored within a quartz cuvette. RNA samples were dissolved in 10 mM HEPES pH 7.5, 50 mM monovalent ion (either KCl or NaCl), 0.5 mM MgCl2, and stated amount of ligand (either guanidine or urea). RNA was heated to 65°C and cooled to 15°C prior to data collection. Temperature was ramped from 15°C to 95°C at 0.5°C/min. UV signals were monitored at 260 and 280 nm and collected every 30 sec. Derivative of data was taken over a five data point window and fit using Global Melt Fit (Draper et al. 2001) to determine the melting transition temperature.

DATA DEPOSITION

Atomic coordinates and structure factors amplitudes have been deposited with the PDB under accession code 7MLW.

SUPPLEMENTAL MATERIAL

Supplemental material is available for this article.