Insights into metalloregulation by M-box riboswitch RNAs via structural analysis of manganese-bound complexes

Mn²⁺-induced folding of the Bacillus subtilis M-box aptamer

Divalent metal ions associate closely with RNA, occupying specific sites and serving to neutralize negatively charged pockets in the RNA. This charge neutralization likely stabilizes tertiary contacts and folding of the RNA. The global effect of Mg²⁺ on the conformation of Bacillus subtilis M-box aptamer has been characterized previously⁹ (Fig. 1a-b). However, in this study we wanted to compare the folding characteristics of M-box in the presence of different ions and elaborate on the interactions of other divalent ions with the M-box using structural probing techniques. First, the compaction of the M-box RNA in response to increasing concentrations of Mg²⁺, Mn²⁺ and Ca²⁺ was quantified using analytical ultracentrifugation (Fig. 1c). In the presence of 100 mM monovalent ions, the compacted form of the RNA was induced by Mn²⁺ with a half-maximum change (EC₅₀) occurring at 0.1 mM. This compares to a value of 0.4 mM for Mg²⁺ and Ca²⁺, suggesting a 4-fold higher overall affinity for binding Mn²⁺ as compared to Mg²⁺ and Ca²⁺. The Hill coefficients for these fits, however, showed similar values (ranging from 4.3 to 5.5) regardless of the metal bound. This suggests that the M-box may bind different metal ions with similar degrees of cooperativity but with varying affinity.

To further characterize metal-induced structural changes in the M-box, we employed 2′-hydroxyl acylation analyzed by primer extension (SHAPE)³⁷ (Fig. 1d-e). In this technique, Nmethylisatoic anhydride (NMIA) forms a covalent adduct with 2′-hydroxyl groups in a structure-dependent manner. We measured the NMIA reactivity of M-box riboses in response to increasing concentrations of Mg²⁺, Mn²⁺ or Ca²⁺, and in the presence of physiological (100 mM) or high (2.1 M) monovalent ions. These data show that in the presence of molar concentration of monovalents, all three divalents induce compaction of the M-box RNA at nearly identical concentrations. However, in the presence of 100 mM monovalents, the M-box RNA was compacted at a lower concentration of Mn²⁺ (EC₅₀ of 0.13 mM) as compared to Mg²⁺ and Ca²⁺ (EC₅₀ of 0.4 mM). Together, these results suggest that the M-box compacted state can be induced by multiple divalent ions, albeit at different concentrations. In the cellular context, this difference in affinity may be overcome by the relative abundances of each metal ion, with the M-box functioning as a Mg²⁺-responsive genetic switch due to the elevated levels of Mg²⁺ relative to other metal ions. However, metal-induced folding of the M-box RNA was similar for the different divalents in the context of high monovalent ions, suggesting that a common minimal number of divalent cation sites were required when low affinity metal interactions could be outcompeted by monovalent ions.

Global arrangement of the Mn²⁺-chelated M-box aptamer

Having confirmed that the overall fold of the M-box is similar in the presence of Mn²⁺ and Mg²⁺, we identified conditions for crystallographic analysis of the M-box aptamer in the presence of 5 mM Mn²⁺ instead of Mg²⁺. While this substitution resulted in the formation of crystals, the addition of 50 mM ammonium sulfate as an additive significantly improved diffraction quality. The previously reported 2.7 Å structural model revealed that the M-box forms a three helical arrangement in the Mg²⁺-bound conformation⁹. This compacted native state appeared to be stabilized by multiple long-range nucleoside and base stacking interactions. Six well-ordered Mg²⁺ ions were identified in the structural model, which appeared to assist many of the long-range interactions. Four of these cation sites (Mg1-Mg4) are located within a common region at the base of the molecule where the three helical elements converge. The remaining two cation sites were located within an internal bulge of the P4 helix. A selection for phosphorothioate substitutions that prevented Mg²⁺-mediated compaction revealed that each of these six cation sites are functionally required for the compact conformation³². Also, the general strength of phosphorothioate interferences moderately agreed within these two cation binding “cores”, with Core 1 consisting of four cation pockets (M1-M4) and Core 2 consisting of the region encompassing two such sites (M5-M6). Core 1 appears to stabilize key interhelical interactions at the portion where the three helices converged. Core 2 may assist orientation of the L4 terminal loop to permit key long-range interactions. However, additional phosphorothioate interferences were identified at positions other than within the binding sites for M1-M6. Several of these sites, which exhibited a similar overall strength of phosphorothioate interference, were found to be located within electronegative cavities reminiscent of the other cation binding pockets. From these data we proposed the presence of three additional metal binding sites (M7-M9). These putative metal sites were found to cluster at the apex of the molecule (‘Core 3’) and appeared to be poised to stabilize important inter-helical interactions. However, the sites in Core 3 were unoccupied in the Mg²⁺-bound M-box structural model. It is possible that these electronegative cavities represent relatively low occupancy sites and therefore were not observed in the 2.7 Å structure. To further investigate these additional putative metal sites we solved the crystal structure of the M-box RNA in the presence of Mn²⁺, thereby exploiting two features of Mn²⁺ – its ability to occupy Mg²⁺ sites in RNA structures and its behavior as an anomalous scatterer of X-rays. This latter feature permitted the unequivocal identification of divalent ion binding sites.

Crystals of the M-box RNA, grown in the presence of 5 mM Mn²⁺ diffracted to 1.9 Å, thus providing higher resolution data than the previous Mg²⁺-bound structure. Representative electron density for these data is shown in Fig. 2c. The asymmetric unit was composed of two molecules (hereon referred to as chain A and chain X) that exhibited subtle structural differences. These two molecular conformations also exhibited differences in the total number of Mn²⁺ sites that were occupied. The overall structure of Mn²⁺-bound M-box RNA resembled closely that of the Mg²⁺-chelated complex. Overlay of chain A of the Mn²⁺ bound structure with the Mg²⁺ bound structure (Fig. 2a) suggests an RMSD of 0.9 Ų for the superimposed backbone atoms. Interestingly, superimposition of chain A and chain X of the Mn²⁺-bound M-box (Fig. 2b, RMSD= 0.7 Ų) shows structural differences in the L4 loop, from positions A61 to C68. This region of the M-box is involved in multiple long-range contacts that are centered around M5 and M6 in Core 2. The apparent flexibility of M-box backbone in this region may support the previous observation that metal sites in Core 2 show relatively weaker phosphorothioate interference³². Consistent with this, chains A and X not only show altered backbone conformation in this region, but also different Mn²⁺ occupancies; the M5 and M6 metal sites in Core 2 are only occupied for chain A which shares identical structural arrangement with the Mg²⁺-bound structure.

The P6 helix region of the molecule was disordered in the previous structural model. However, increased resolution and different space group symmetry in the Mn²⁺ crystals allowed visualization of the P6 helix. Intermolecular kissing loop interactions between symmetry related molecules appeared to stabilize the L6 loop (Fig. 2d). The loop-loop interaction involves 6 nucleotides (G130, C131, U132, U133, G134 and U135) where G-U and G-C base pairs flank the central uridines and allow them to be stabilized by N3-O4 and O2-N3 interactions, respectively. While it is likely that the relative orientation of these nucleotides with respect to the native M-box structure may be affected by this crystal contact, the increased order in this region resulted in improved electron density of the entire P6 helix region. The increased resolution extended beyond the 6 nucleotides involved in the kissing loop, thereby permitting visualization of the P6 helix, which was previously not modeled for the Mg²⁺-bound M-box structure from lack of clear electron density.

X-ray diffraction data collected at the absorption edge of Mn²⁺ allowed the calculation of anomalous difference fourier maps. This allowed for the direct placement of Mn²⁺ ions in the M-box crystal (Fig. 3a). Guided by difference density calculated from 1.9 Å resolution data, key water molecules that coordinate Mn²⁺ were modeled into the Mn²⁺-bound M-box structure. The geometry of these sites provided further confirmation for the presence of fully coordinated Mn²⁺ ions in the M-box RNA structure. The different Mn²⁺ sites in the two chains exhibited a range of strengths for positive anomalous difference density. The two chains also differed modestly in the number of occupied metal sites (Fig. 3b-c).

Mn²⁺ occupies Mg²⁺ sites in the Core 1 and Core 2 regions

We compared the Core 1 from the previously solved M-box structure to each of the two chains in the Mn²⁺-bound M-box. Three out of four sites in Core 1 (M1, M2 and M3) showed nearly identical occupancies in both chains. Mn1 was coordinated via interactions to the RNA backbone with the non-bridging phosphate oxygens of G100, C102 and A103, at an average distance of 2.18Å. Similar to the Mg²⁺-bound structure, another key coordination occurred through the nucleobase oxygen of U104. Two water molecules completed the hexacoordinate state of Mn²⁺, at distances of 2.3 Å. Coordination of Mn2 occurred via two inner-sphere interactions with backbone phosphate oxygens of U23 and G100, and was confirmed with Mn²⁺ anomalous density of 13σ. Again, distances from the remaining coordinating water molecules averaged around 2.25 Å, slightly higher than the 2.15 Å seen for the Mg²⁺-bound structure. Mn2 is also coordinated via an outer-sphere interaction with a backbone phosphate oxygen of G157, similar to what was observed for the Mg²⁺-occupied site. This interaction is likely to be functionally significant as previous phosphorothioate interference experiments had inferred the importance of this outer–sphere interaction³². Mn²⁺ occupancy at the M3 cation site, which was coordinated via two inner-sphere interactions with U24 and A25 phosphate oxygens, was confirmed by the presence of Mn²⁺ anomalous density of 16σ. In total, these data agreed well with previous experiments, wherein nucleotides involved in coordinating Mg²⁺ at these three sites exhibited strong phosphorothioate interferences of nearly identical intensities³². Moreover, Mn²⁺ sites 1, 2 and 3 were superimposable with the corresponding Mg²⁺ sites (Fig. 4b), with negligible change in position of not only the metal ion but also the coordinating water molecules. This suggests that certain M-box Mg²⁺ sites can be substituted with Mn²⁺ without distortion of the site geometry and positions of the coordinating ligands from the RNA. Finally, the last remaining Core 1 cation site, Mn4, was coordinated by a single inner-sphere contact to a nonbridging phosphate oxygen of A72, and an outer-sphere contact to the N7 of A72. It is noteworthy that this site is likely to represent the weakest specific cation site in Core 1 since the anomalous density ranged from 3.8σ in one chain to 5.8σ in the other. Consistent with this, position A72 had previously exhibited only moderate phosphorothioate interference.

In contrast to Core 1, we observed moderate structural differences between chain A and chain X in the Mn²⁺-bound M-box within the Core 2 region (Fig. 4c). Whereas chain A permitted substitution of Mg²⁺ sites with Mn²⁺, chain X showed a different backbone conformation in the vicinity of sites M5 and M6, which were unoccupied by Mn²⁺ in the latter molecule. In general, the chain A M5 and M6 sites also showed weaker Mn²⁺anomalous peaks (~5.5σ), relative to M1-M3. Superposition of Mn5 and Mn6 in chain A with those in the Mg²⁺-bound M-box (Fig. 4b) revealed a high degree of structural conservation. Mn5 contacts the RNA though one inner-sphere interaction with the phosphate oxygen of A63 and multiple outer-sphere interactions with nucleobases of C77 and U75. However, while the distance between these atoms is ~6.4 Å (O2P of A63 to O2 of C77) in chain A it is ~8.5 Å in chain X (Fig. 4d.), thereby positioning the metal-chelating ligands too far from one another for formation of the cation pocket. Similarly, Mn6 is coordinated by a backbone phosphate oxygen of A80 and the O4 of U81. The two metal-coordinating atoms are 3.3 Å apart, allowing the formation of a relatively more compact cation-binding pocket. In chain X however, this distance is 4.4 Å, thereby eliminating this cation site.

Mn²⁺ occupies a subset of the putative sites within Core 3

Previously, we had observed phosphorothioate interferences at positions C35, C36, C89, G91 and U119 in the M-box aptamer.³² These positions showed similar degrees of interference and when mapped on to the structure of M-box, were found to cluster around the apex of the three-helical bundle that forms the architectural foundation of the molecule. These positions together appeared to represent three putative metal sites, which we designated M7 through M9. However, in the crystallographic model of the Mg²⁺-chelated M-box aptamer the sites were not occupied by metal. One possible explanation is that they exhibit relatively low-affinity for metal ligand and therefore were unoccupied in the final crystal structure. Alternatively, the putative M7-M9 sites might occupied with metals only during folding. Since the overall affinity of M-box for Mn²⁺ was higher than that for Mg²⁺, we considered the possibility that Mn²⁺ might exhibit higher affinity at specific cation binding sites in the M-box. Therefore, we examined the Mn²⁺-bound M-box structure for Mn²⁺ anomalous density in the predicted metal sites M7 to M9. The M7 and M9 sites exhibited clear anomalous density of 7σ and 11σ respectively (Fig. 3a), while M8 exhibited very weak anomalous density and a metal could not be modeled with enough confidence (Fig. 5b).

Closer examination of Mn7 (Fig. 5a) showed that it is coordinated by three backbone interactions with phosphate oxygens of A117, A118 and U119. Of these, the interaction with U119 alone is via the pro-Rp oxygen and was previously seen as a position of phosphorothioate interference. Metal-ligand distances are in the range of 2.3 Å, as seen for the other Mn²⁺ sites. Three water molecules, at an average distance of 2.5Å, complete the Mn²⁺ coordination. The combined orientations of the A117, A118, and U119 phosphates, towards formation of an electronegative cavity, appeared to help position adjacent nucleosides for key long-range interactions at the apex of the molecule. For example, the nucleobase ring of A118 stacks against the G56-U85 base pair located in the backbone U-turn separating the P4 and P5 helices. Long-range interactions in this region of the M-box are likely to be critical for stabilization of the sharp, 180° turn required between the P4 and P5 helices. Similarly, A117 stacks against the A87-U116 base pair at the apex of the P5 helix and also forms an A-minor motif interaction with the P4 G84-C57 base pair, again stabilizing distant regions of the RNA.

Mn9 is coordinated by inner-sphere interactions with the pro-Rp oxygens of C35 and C36 (Fig. 5c). Outer-sphere interactions with the phosphate oxygens of C143 from the adjacent turn of the P2 helix further stabilize this coordination. Considering the placement of this Mn²⁺ site, it is possible that it influences the turn of the P6 helix, allowing the M-box to fold upon itself and form the three-helical bundle. C35 not only makes direct inner-sphere contacts with Mn9, but also stacks against A88, which is involved in an A-minor interaction to a P2 G:C base pair. These different metal sites exhibit a concentration of electronegative charge as would be expected for specific cation binding sites (Fig. 5d-f).

New metal sites

Having visualized Mn²⁺ anomalous density in the six Mg²⁺ sites observed previously in the Mg²⁺-bound M-box structure and confirmed two of the three putative metal sites predicted by phosphorothioate substitution experiments, we next examined the structure for additional anomalous density. Several new sites of Mn²⁺ occupancy were observed (Fig. 3b). In total we observed six additional Mn²⁺ ions, labeled herein as Mn10-15. Most were observed in both RNA chains, although Mn11 was present only in chain A and Mn14 was only in chain X.

The most interesting of these metal sites, Mn10, is present in the L5 loop and is coordinated via one backbone interaction with the pro-Rp oxygen of A106 and the N7 of G107 (Fig. 6a). Two outer-sphere interactions with pro-Rp oxygen of A105 and G107 further stabilize the coordination. Notably, A105 had previously shown to exhibit moderate phosphorothioate interference. Also, O2 of A106 on the non-metal binding face coordinates a potassium ion. The ribose oxygen of G107 further interacts with N1 of A155 from the P2 helix. The adenine rings of A105 and A106 stack against each other and interact on opposite faces with elements from L4 and P2. While the A105 nucleobase interacts with the C68 ribose oxygen from the L4 loop, the N6 of A106 lies within 2.87Å of the U24 O4, located in the P2 helix. The G107 ribose oxygen further interacts with N1 of A155, also located within the P2 helix. Given the importance of these tertiary interactions, it is possible that Mn10 may represent a functionally important metal ion that was previously not observed in the Mg²⁺ bound form of M-box.

Mn11 is present in the P4 helix of the M-box aptamer, making one inner-sphere contact with the A60 phosphate oxygen (Fig. 6b). The remaining coordinations are satisfied by water molecules. Outer-sphere coordinations are also seen with the N7 groups of A60 and A61 and the phosphate oxygen of A77. As seen with the other Mn²⁺ sites in the P4 helix (Mn 5 and 6), Mn11 is present only in chain A and not in chain X. In chain X, the altered backbone conformation in the P4 helix alters the placement of the nucleotides involved in coordinating Mn11, placing them too far away to be able to create a compact metal-coordinating pocket. Mn11 also shares yet another feature with Mn 5 and 6 in metal Core 2; the strength of anomalous density in all three sites is similar, at around 6σ (Fig. 3a).

Mn12 lies at one end of the P2 helix, close to the L6 turn in the M-box aptamer. Both chains showed occupancy of Mn²⁺ ions in the site. However, the position of this Mn²⁺ ion differed between the two chains. Chain A resembled the Mg²⁺-chelated M-box structure in this region with regard to the arrangement of backbone atoms. Here, the Mn²⁺ ion is coordinated by an inner-sphere interaction with a phosphate oxygen of G45 (distance of 2.2 Å) which makes important contacts with U140. The N7 of G45 also makes an outer sphere contact with Mn12. The remaining coordinations are satisfied via water molecules, though only four of the five metal-chelating water molecules were modeled due to lack of clear density for one water molecule. A46 forms multiple interactions with A139 as well as U138 on the same plane, with a distance of 2.96Å between the N6 of A46 and O2 of U138. U138 is positioned so as to make an outer-sphere contact to Mn12 via a water molecule. In chain X, however, the location of Mn12 is different. The Mn²⁺ is coordinated via inner-sphere interactions with the N7 of G45 (distance of 2.5Å), whereas the phosphate oxygen on G45 makes an outer-sphere contact via a metal-chelating water molecule. This region of the RNA shows one more significant difference with regard to the structure of chain A. A46 is splayed outwards from the P2 helix and instead of interacting with A139 on the plane of the nucleobase, it stacks against the adenine ring. Also, the interaction between A46 and U138 are altered such that the N6 of A46 is positioned as far away as 5.23 Å from the O2 of U138.

Mn13 is present in the P6 helix, proximal to the L6 loop and is coordinated to the pro-Sp phosphate oxygens of G127, A128 and U129 and two water molecules. Hexa-coordination is completed by water from the other chain. Also noteworthy is an outer-sphere interaction with G29 from the other chain via a water molecule. This observation suggests that this site is created due to proximity of the two chains of M-box in the crystal lattice and therefore not likely to be of functional relevance. Mn14 is present in the central region of the P2 helix, coordinated by nucleotides A30 and G31 that stack against each other in the P2 helix. The proximity of a symmetry related molecule in this region, lack of apparent metal-mediated tertiary contacts as well as lack of phosphorothioate interference data for the coordinating nucleotides seems to suggest that this site is not likely to be functionally relevant. Mn 15 is coordinated via the N7 of G15 and is in close proximity to the pro-Sp oxygen of G15. Mn 16 is occupied only in chain X and is coordinated via the N7 of G126 and the ribose oxygen of A46. Moderate anomalous density (~ 6 σ) prompted us to assign these sites as Mn²⁺ sites, however, there is lack of clear density for assignment of metal-coordinating water molecules.

RNA preparation

For AUC and SHAPE reactions, DNA templates were amplified by PCR from B. subtilis IA40 chromosomal DNA (Bacillus Genetic Stock Center, Columbus, OH) using oligos SCB3 (5′ taatacgactcactatagggaacttcgttaggtg) and SCB4 (5′ccttcaatcgtagagc). RNA was prepared by in vitro transcription for 2.5 hours at 37 °C, in 50 μl reactions. These reactions included: 30 pmol DNA template and buffer containing 30 mM Tris-HCl pH 8.0, 10 mM DTT, 0.1% Triton X-100, 0.1 mM spermidine-HCl, 5 mM of each NTP (Roche), 40 mM MgCl₂ and ~50 μg/ml of recombinant T7 RNA polmerase. Reactions were terminated with loading dye containing 8M urea. Following resolution on a urea denaturing 6% polyacrylamide gel, bands were cut into ~1 mm cubes and incubated at 25°C for 2.5 hours in buffer containing 200 mM NaCl, 10 mM Tris-HCl pH 7.5 and 10 mM ethylenediaminetetraacetic acid (pH 8.0). Eluted RNAs were then ethanol precipitated and re-suspended in water.

SHAPE assays were performed with 1 pmol RNA in 9 μl solutions containing 5 μl of 2X SHAPE buffer (10 mM Tris-HCl pH 8.0 and 200 mM KCl) and 1 μl of 10X MgCl₂ or MnCl₂ or CaCl₂. Experiments performed at high concentrations of monovalent ions included 4M KCl in the 2X SHAPE buffer. Solutions were equilibrated for 2 hours at 25 °C following which, 1μl of freshly prepared 65mM N-methylisatoic anhydride (NMIA) solution was added. Samples were then incubated for 45 minutes at 37°C. Reactions were ethanol precipitated and resuspended in 11 μl water with the 5′-³²P-labeled oligonucleotide SCB4 (5′ccttcaatcgtagagc). An extension reaction was then performed using Transcriptor reverse transcriptase (Roche), as described previously³⁷. Samples were subjected to PAGE separation, and all polyacrylamide gels were visualized and quantified as described previously⁹.

Analytical Ultra Centrifugation was performed essentially as described previously⁹ using a Beckman AUC XL-I with absorbance optics. 390 μl sample and reference buffer were loaded into dual-sector charcoal-filled epon centerpieces. These were equilibrated to 20°C for 1 hour. Absorbance measurements at 260 nm were collected during centrifugation in an An60Ti rotor at 42,000 rpm for at least 4 hours. Density and viscosity calculations for each buffer was made using SEDNTERP. Since no density and viscosity values for MnCl₂ are included in the SEDNTERP program, this buffer component was excluded from the analysis. Since MnCl₂ was present at submmillimolar concentrations, data analysis was not significantly impacted by this exclusion. Divalent ions titrations were performed in buffer containing 10 mM Tris-HCl pH 7.5, and 100 mM KCl. Data were analyzed using SedFt v9.4, assuming a partial specific colume (vbar) of 0.53 cm³ gm⁻¹ and a hydration value of 0.59 g/g in order to obtain the sedimentation coefficient. The sedimentation coefficient, R_H, a/b and D were obtained from the analysis and sedimentation coefficients were converted into s_20,w by multiplying with the conversion factor obtained from the following equation:

where η_T,b is the viscosity of the buffer, η_20,w is the viscosity of water at 20 °C, ρ_20,w is the density of water at 20 °C, vbar is the vbar of the RNA at 20 °C, and ρ_T,b is the density of the buffer.

To generate RNA for crystallization, DNA templates were generated by PCR using oligos WCW500 (5′gccagtgaattctaatacgactcactatagggcttcgttaggtgaggc) and AR004 (5′gctctagatggaagatcatgtgatttctc). The template for these reactions included the sequence of the B. subtilismgtE M-box aptamer followed by an adenine nucleotide and the glmS ribozyme, as described previously (Dann et al., 2007). Following phenol:chloroform extraction and ethanol precipitation of the PCR product, in vitro T7 transcription reactions were set up as described above. The resultant RNA products encompassed nucleotides 18-171 of the B. subtilis mgtE 5′ leader region and included three additional G:C base pairs at the terminus of the P1 helix. Immediately downstream of the aptamer was an unpaired adenine nucleotide and the glmS ribozyme (e.g., Dann et al. 2007). 2 mM glucosamine-6-phosphate was supplied in the 10 mL transcription reactions to stimulate glmS self-cleavage and release of the M-box aptamer sequence⁹. Following extraction with phenol:chloroform: isoamylalcohol (25:24:1) and ethanol precipitation, RNA was re-suspended in gel-loading buffer containing 8M urea and subjected to denaturing PAGE on an 8% urea-polyacrylamide gel for 3.5 hours. This resulted in clear separation of the M-box RNA from the glmS RNA fragment. The band corresponding to M-box RNA was excised and the RNA recovered by passive elution.⁹ After ethanol precipitation, the RNA was resuspended in buffer (10 mM HEPES pH 8.0, 100 mM KCl) and dialyzed against 1000X volume of the same buffer for a minimum of two hours with two buffer exchanges. Following dialysis the RNA was concentrated to 4 mg/ml using YM-3 microcon centrifugal filters.

Crystallization and X-ray Data Collection

Irregular crystals for the M-box RNA were obtained using hanging drop vapor diffusion by mixing 1.0 μL RNA with 0.9 μL crystallization buffer (40 mM HEPES pH 7.4, 20 mM KCl, 12 mM Spermine•4HCl, 10 % MPD) and 0.1 μL of 0.1 mM MnCl₂ and equilibrating against a reservoir solution consisting of 35% MPD. To obtain better quality crystals, an optimization was performed using an additive screen (Hampton Research) where 1.0 μL RNA at 4 mg/ml was mixed with 0.8 μL of crystallization buffer, 0.1 μL of 0.1 M MnCl₂ and 0.1 μL of additive. Rod-shaped crystals were obtained in the presence of 50 mM ammonium sulfate as the additive. Crystals were picked from the crystallization drop and directly flash frozen in liquid nitrogen. Diffraction data was collected at the Mn²⁺ absorption edge (6.547keV) at beamline 19BM (Advanced Photon Source, Chicago) with the best crystal diffracting to 3 Å with cell dimensions a=69.80, b=103.30, c=129.84 α=β=γ=90°. Another crystal from the same drop yielded higher resolution data (1.86Å) which was collected at 12.659keV on beamline 19ID and showed cell dimensions a=68.78, b=102.49, c=126.38 α=β=γ=90°. Data statistics are compiled in Table 1. M-box crystals from the same crystallization drop often displayed non-isomorphism. This phenomenon has been noted in the past for crystals of the P4-P6 domain of group I intron⁴².

Structure determination

X-ray diffraction data was indexed, integrated, and scaled using HKL2000⁴³. Molecular replacement was performed against the 2.9 Å data set using the Mg²⁺ bound M-box structure (PDB 2QBZ) as a search model, using Phaser via the CCP4 suite^44,45. The resulting solution showed good statistics with a Z score of 46.2, LLG of 3067 and R-factors of 23% and 29.7%. Using this solution as a search model, molecular replacement was performed against the 1.86 Å data set, yielding a solution with a Z-score of 46 and LLG of 3491. Rigid body refinement followed by simulated annealing was performed using phenix.refine⁴⁶. Density corresponding to the P6 helix was visible at greater than 3σ contour of the F_o-F_c map and this region was built manually using Coot⁴⁷. Iterative model building using Coot along with refinement (phenix.refine) significantly reduced the crystallographic R factor to 24% and R free to 27%. Positive electron density in the F_o-F_c map was used for the assignment of water molecules. Strong positive electron density in the F_o-F_c map and distance and geometry constraints were used to place Mn²⁺ ions. Water molecules that displayed significantly lower B-factors as compared to the neighboring atoms were also an indication of occupancy by Mn²⁺. An anomalous difference fourier map created using sfall⁴⁵ was generated using the model and the 2.9 Å data collected at the Mn²⁺ edge. Positive anomalous peaks as strong as 21σ were observed in this map and were used to further confirm the location of Mn²⁺ ions. The final model was refined to an Rfactor of 19.2% and Rfree of 22.8%. Refinement statistics are shown in Table 1. Electrostatic potential calculations was performed using pdb2pqr⁴⁸ (to prepare input files for APBS) and APBS⁴⁹ and visualization was done using Pymol. Illustrations were made using UCSF Chimera⁵⁰.