A Mn-sensing riboswitch activates expression of a Mn²⁺/Ca²⁺ ATPase transporter in Streptococcus

Abstract

Maintaining manganese (Mn) homeostasis is important for the virulence of numerous bacteria. In the human respiratory pathogen Streptococcus pneumoniae, the Mn-specific importer PsaBCA, exporter MntE, and transcriptional regulator PsaR establish Mn homeostasis. In other bacteria, Mn homeostasis is controlled by yybP-ykoY family riboswitches. Here, we characterize a yybP-ykoY family riboswitch upstream of the mgtA gene encoding a P_II-type ATPase in S. pneumoniae, suggested previously to function in Ca²⁺ efflux. We show that the mgtA riboswitch aptamer domain adopts a canonical yybP-ykoY structure containing a three-way junction that is compacted in the presence of Ca²⁺ or Mn²⁺ at a physiological Mg²⁺ concentration. Although Ca²⁺ binds to the RNA aptamer with higher affinity than Mn²⁺, in vitro activation of transcription read-through of mgtA by Mn²⁺ is much greater than by Ca²⁺. Consistent with this result, mgtA mRNA and protein levels increase ≈5-fold during cellular Mn stress, but only in genetic backgrounds of S. pneumoniae and Bacillus subtilis that exhibit Mn²⁺ sensitivity, revealing that this riboswitch functions as a failsafe ‘on’ signal to prevent Mn²⁺ toxicity in the presence of high cellular Mn²⁺. In addition, our results suggest that the S. pneumoniae yybP-ykoY riboswitch functions to regulate Ca²⁺ efflux under these conditions.

INTRODUCTION

Small non-coding regulatory RNAs (sRNAs) are ubiquitous and found in all domains of life and play myriad roles in regulating gene expression (1,2). In bacteria, these sRNAs exist as short transcripts with lengths of 30–500 nucleotides (3,4). Bacterial sRNAs are broadly classified into two categories: cis-acting (riboswitches) or trans-acting (independent transcripts) RNAs. Trans-acting sRNAs modulate the stability or translation of mRNA transcript(s) by imperfect base-pairing interactions (5,6) at or near the ribosome binding site (RBS) on the target transcript. Riboswitches, on the other hand, are highly structured regulatory segments located within the 5′ untranslated regions (UTRs) of mRNAs that interact with small molecules, such as metals and other ligands, resulting in a change in transcription or translation mediated directly by environmental changes. sRNAs may also interact with RNA-binding proteins thereby modifying their activities (7,8).

Bacterial sRNAs function in many cellular processes, including the response to environmental stresses, e.g. those mediated by the immune system, and maintaining homeostasis (5,9). Accumulating evidence suggests that sRNAs are differentially expressed in bacteria when transitioning from colonization to an active infection which suggests that sRNAs are key players in mediating invasive disease (10–13). Streptococcus pneumoniae is a Gram-positive bacterium that commonly colonizes the nasopharynx of healthy individuals, persisting as a harmless bacterial commensal (14). Despite a largely asymptomatic colonization, S. pneumoniae remains a leading causative agent of sinusitis, otitis media (middle ear infection), and the life-threatening invasive diseases pneumonia, bacteremia, and meningitis globally (15,16). Transition from the harmless commensal to pathogen often occurs after a viral respiratory tract infection and is triggered by numerous factors, many of which are unknown (14). Progression into the lungs and intrusion into the circulatory system, exposes the pneumococcus to numerous stress conditions and environmental changes (17), including transition metal fluctuations, an important mediator in the ‘fight or flight’ response during infection (18,19). The capacity for bacteria to quickly adapt to these different environments may derive in part from sRNAs modulating the expression of virulence factors.

Multiple (>100) sRNAs have been identified in S. pneumoniae (20–25), but the functional role for most of these remains unknown. One particular candidate sRNA, SN44, identified in S. pneumoniae TIGR4 using genome-tilling arrays (21) is of interest here. SN44 is predicted on the basis of sequence homology to be a cis-acting yybP-ykoY family response element (riboswitch) that is conserved among different Streptococcus species (21) (Supplementary Figure S1). The yybP-ykoY motif defines a ubiquitous class of riboswitches that are upstream of genes that encode uncharacterized or poorly characterized proteins many of which have been linked to Mn²⁺ homeostasis. To date, yybP-ykoY-regulated gene products, MntP of Escherichia coli (26), MntX of Neisseria spp. (27), and YaoB (a predicted P-type II ATPase) of Lactococcus lactis (28), have been independently implicated as Mn²⁺ efflux transporters capable of relieving cellular Mn toxicity. Other studies show that several of these associated yybP-ykoY family riboswitches respond to high Mn²⁺in vitro and in vivo modulating transcription and translation of downstream gene(s) within the transcript (28–30). Although yybP-ykoY riboswitches have generally been observed to bind Mn²⁺in vitro with highest affinity among other transition metals examined, none thus far exhibit strict Mn²⁺ specificity (30).

The yybP-ykoY riboswitch aptamer domain adopts a four-helix junction architecture with P2 stacked on P1 and P4 stacked on P3, with the transition metal sensing pocket formed by nucleotides in the L1 loop (in P1) and L3 loop in P3 (Figure 1A). P2 and P4 are not strictly conserved and can be replaced by a short single-stranded connecting region or a longer helical stem, just above the cross-over point (Figure 1). A recent structural analysis of the L. lactis yybP-ykoY riboswitch aptamer domain suggests that the metal site may well be plastic and capable of adopting a range of coordination structures and nuclearity, with binuclear and trinuclear metal (Cd²⁺) complexes observed in that study (30). The binuclear site in a previous structure of the L. lactis yaoB riboswitch (28) was modeled as Mg²⁺ or Mn²⁺ in the M_A site and Mn²⁺ in the M_B site (Figure 2A). The M_B site is thought to provide specificity for Mn²⁺ given the presence of five inner-sphere coordination bonds, including the N7 of A41 in the L3 loop (Figure 2A–C).

Figure 1.

Secondary structure diagrams of the aptamer domains of (A) E. coli alx sequence (30) and (B) Streptococcus pneumoniae D39 mgtA riboswitch domains. In panel A, nucleotides in L1 (blue) and in L3 (red) that correspond to those shown in the structural models (Figure 2) are highlighted in bold, and correspond to the numbering convention of the L. lactis yybP-ykoY riboswitch (see Figure 2A) (28). In panel B, we used the natural nucleotide residue numbers, and those nucleotides targeted for substitution in this work in L1 (blue) and L3 (red) are highlighted in bold. For example, the loop L1 G8-G9-A10 L. lactis system corresponds to the G16-G17-A18 in the S. pneumoniae RNA, while the L. lactis C40-A41 L3 sequence corresponds to the C49-A50 sequence in S. pneumoniae. A41 in the L. lactis RNA (A50 in the mgtA RNA) makes an inner sphere coordination bond with the MnB metal (see Figure 2A). The gray nucleotides in panel B are non-native and were added for in vitro transcription.

Figure 2.

Regulatory metal coordination site region (left) and global ribbon diagrams (right) of the various yybP-ykoY riboswitch aptamer domains. (A) L. lactis yybP-ykoY riboswitch bound to Mg²⁺ (site A) and the Mn²⁺ (site B) (28); (B) chimeric E. coli ykoY^mntP riboswitch bound to Mg²⁺ (site A) and Cd²⁺ (site B) (30) and (C) the three metal (Cd²⁺)-bound form of the L. lactis yaoB riboswitch (30).The nucleotide numbering convention in panels B and C are as adopted for the L. lactis yybP-ykoY riboswitch (panel A). L1 residues are highlighted in blue, and L3 residues are shown in gray (left) or red (right) of panels A-C. The nucleotide base of L3 nucleotide A41 is shaded gray since N7 donates a coordination bond to the Mn²⁺ (A) and Cd²⁺ (B, C) ions in these structures.

Like other annotated yybP-ykoY family riboswitches, the putative SN44 riboswitch overlaps the RBS within the target gene transcript and features a transcription terminator helix just 5′ to the RBS (Supplementary Figure S1). The SN44 RNA in S. pneumoniae D39 serotype 2 is encoded within the 5′ UTR of the mgtA (spd_1383; previously caxP or PacL) transcript, encoding a P_II-type ATPase transporter. MgtA was previously suggested to function in Ca²⁺ efflux (25,31), while in a different report was suggested to be implicated in the response against Zn²⁺ and Mn²⁺ toxicity also (32). The idea that MgtA in S. pneumoniae is not restricted to Ca²⁺ and could also transport Mn²⁺ is also supported by the fact that the MgtA protein is encoded from the negative-sense DNA strand immediately upstream of the mntE gene, which encodes the constitutively expressed Mn²⁺-specific cation diffusion facilitator exporter MntE (33,34). Both MgtA and MntE are important for the virulence of S. pneumoniae (31,33). Despite intensive study, the physiological roles of many genes critical for virulence, including those encoding divalent metal transporters like MgtA, remain unclear.

In this study, we examine the metal binding and metal-induced folding of the mgtA aptamer and the regulation of mgtA in S. pneumoniae D39 strain. We find that the S. pneumoniae D39 mgtA aptamer RNA binds stoichiometric Mn²⁺ and Ca²⁺ in a background of physiological Mg²⁺ and each is capable of stabilizing the same compact conformation; however, Mn²⁺ stimulates in vitro read-through transcription to a far greater extent than Ca²⁺. We establish for the first time that the mgtA mRNA is upregulated during Mn-stress in this organism, and that this increased transcription in vivo requires the mgtA (yybP-ykoY family) riboswitch. Further investigation reveals that MgtA may export Mn²⁺ under conditions of extreme Mn-stress thereby protecting cells from Mn²⁺ toxicity. The mgtA riboswitch may also function to regulate Ca²⁺ export under these conditions.

MATERIALS AND METHODS

See Supplementary Material detailing methods for bacterial strain and plasmid construction, bacterial growth, disk diffusion assay, β-galactosidase activity assay, RNA isolation and quantitative real-time PCR, western blot, yybP-ykoY RNA synthesis, and inductively coupled plasma-mass spectrometry (ICP-MS) for measurement of total cell-associated divalent cationic metals. Bacterial strains and plasmids used in this study are listed in Supplementary Table S1 and relevant DNA oligonucleotide primers in Supplementary Tables S2 and S3.

Native polyacrylamide gel electrophoresis (PAGE) analysis of yybP-ykoY RNAs (35)

Purified RNA (2 μg) was heated at 90°C for 2 min followed by incubation at room temperature for 3 min, then subsequently allowed to refold in 50 mM HEPES [pH 7.5], 50 mM NaCl at room temperature for 10 min. Various divalent metal ion concentrations ranging from 0.01 to 5 mM were added and RNA mixture was left at room temperature for an additional 30 min prior to mixing with loading dye (10% glycerol and 0.01% xylene cyanol). RNA was separated by native PAGE (8% gel prepared with 19:1 acrylamide/bisacrylamide) at 4°C in running buffer containing 34 mM Tris, 66 mM HEPES [pH 7.5], and 3 mM MgCl₂, which was recirculated every hour. RNA was stained with ethidium bromide and observed with a Gbox (Biorad).

SHAPE probing

SHAPE was performed as previously described using 1-methyl-7-nitroisatoic anhydride (1M7) derivatization reagent (36–38). RNA constructs consisted of the wild-type sequence from the S. pneumoniae D39 mgtA 5′ UTR (containing the yybp-ykoY aptamer domain) flanked by SHAPE flanking sequences (37). RNAs were in vitro transcribed from PCR templates and purified similarly to that described above for RNA synthesis. Purified RNAs (10 pmol) was refolded in 50 mM HEPES, pH 7.5, 50 mM NaCl buffer before 1M7 derivatization. Reverse transcription of the derivatized RNA was performed with a FAM 6-labeled reverse primer. Unmodified RNA (2 pmol) used as a sequencing ladder was synthesized using reverse transcriptase, NED-labeled reverser primer, and ddCTP or ddGTP. An equal volume of sequencing ladder and reaction mixture with and without 1M7 was precipitated with 0.3 M sodium acetate, pH 5.2 and 100% ethanol, washed with 70% ethanol, air-dried, and re-dissolved in 15 μl water. Fragment analysis was performed using capillary electrophoresis by GENEWIZ (Plainfield, NJ, USA). Data processing was performed using QuShape following a protocol as previously described (39).

Small angle X-ray scattering (SAXS)

Small angle and wide angle X-ray scattering data were acquired at three different RNA concentrations (0.8, 1.6 and 3.2 mg/ml) of select RNAs folded in buffer containing 50 mM HEPES, pH 7.5, 50 mM NaCl, 3 mM MgCl₂ and 0.5 mM of the metal ion indicated_. Data were collected at beam line 12-ID of the Advanced Photon Source at the Argonne National Laboratory. The wavelength (λ) of X-ray radiation was set to 1.033 Å. Procedures for SAXS/WAXS measurement were similar to those previously described (40,41). Thirty images were collected for each RNA concentration and its corresponding background buffer.

SAXS data were averaged, and the background was subtracted using the NCI-SAXS program package. The averaged scattering profiles of three RNA concentrations were merged using PRIMUS in the ATSAS program package (http://www.embl-hamburg.de/biosaxs/). A GUINIER plot was generated as ln(I(q)) versus q² to check sample quality and to obtain I₀ and the radius of gyration (R_g) within the range of q_max*R_g < 1.3. The data from each RNA concentration were normalized with I₀. Conformation of the RNAs was examined using the Kratky plot for q < 0.3 Å⁻¹. Scattering profiles of RNAs were then Fourier-transformed using GNOM of the ATSAS package to obtain the normalized pair distance distribution graph.

Ab in initio modeling was performed using the program DAMMIN in a slow mode (42). For each RNA, 10 models were obtained, filtered, and averaged using the DAMPUP, DAMFILT and DAMAVER of the ATSAS package (http://www.embl-hamburg.de/bioSAXS). Normalized spatial discrepancy (NSD) between each pair of the models was computed. The model with the lowest NSD value was selected as the reference model for superimposing onto other models. Outliner models (two models) with an NSD above mean +2* variation were removed before averaging.

Isothermal titration calorimetry

ITC experiments were performed following standard protocol previously described (30). Briefly, RNAs were heated at 95°C for 2 min, cooled down at room temperature for 3 min, diluted into ITC buffer (30 mM HEPES [pH 6.8], 150 mM NaCl), and incubated at room temperature for an additional 10 min. MgCl₂ was then added to a final concentration of 3 mM and the RNA was subsequently incubated at room temperature for 30 min. RNAs were dialyzed into ITC buffer containing 3 mM MgCl₂, filtered using a 3 kDa MWCO column, and degassed prior to ITC measurement. Divalent metal cations were dissolved in ITC buffer, filtered, and degassed. Experiments were performed with 50–100 μM RNA and 0.5 to 2 mM metal cation at 25°C, depending on affinity and magnitude of enthalpy. Divalent metal cation was injected with 0.5 to 1 ul volume every 300 s. Data were corrected for heats of dilution by subtracting the enthalpies of cations titrated into the ITC buffer from raw data and analyzed using an independent binding model by Nanoanlyze (TA instrument).

IVT termination assay

In vitro transcription termination assays were performed using standard protocols (28,43). To construct the in vitro transcription plasmid, the S. pneumoniae mgtA 5′-UTR (containing the aptamer domain, intrinsic terminator, and the first 45-nt of the MgtA protein coding sequence) was fused to a B. subtilis glyQS promoter. An ApC dinucleotide sequence was added to the 5′ end of the mgtA aptamer domain with no cytosines present in the next 12 nucleotides. DNA templates were produced by PCR and spin column purified (Qiagen). To assemble stalled RNA polymerase complexes, 20 mM template was mixed with initiation buffer (20 mM Tris–HCl [pH 7.5], 50 mM NaCl, 250 μM MgCl₂, 1 mM DTT and 5% glycerol), 0.15 μCuries/μl ³²P ATP, 20 μM each of unlabeled ATP/CTP/GTP, and 0.01 U/μl Escherichia coli RNA polymerase holoenzyme. Reactions were incubated at 37°C for 15 min for elongation, then transferred to ice for 2 min. For synchronized transcription at each condition, 10 μl of stalled RNA polymerase was mixed with 1.5 μl 10× metal solution, and 1.5 μl 10× elongation buffer (4.5 mg/ml, 650 μM each of unlabeled ATP/CTT/GTP/UTP, 100 mM Tris [pH 7.5], 2 mM DTT, 10% glycerol and 2 mM MgCl₂. Reactions were incubated at 37°C for an additional 15 min, then terminated by addition of 10 μl RNA loading dye (95% formamide, 20 mM EDTA [pH 8.0], supplemented with xylene cyanol). Transcripts (10 μl) were separated by 6% urea denaturing PAGE and analyzed by a phosphorimager. The sizes of the terminated and full-length RNA products were confirmed by RNA ladder (data not shown). Bands were quantified with ImageJ and each reaction was converted to the fraction of full-length product over total RNA transcribed. The data were fit using Origin 8 data analysis software to the Hill equation (n = 1), where x is the concentration of metal and k is the concentration at which the change in read-through is half maximal. Each assay was performed in triplicate.

Statistical analysis

When appropriate, P-values were determined relative to non-stressed parent or WT type strains using unpaired t-tests or a one-way ANOVA with Dunnett's post-test determined by GraphPad Prism software.

RESULTS

The S. pneumoniae mgtA riboswitch is comprised of a three-way helical junction that is conserved among Streptococcus spp.

As described above, the crystallographic structure of the Mn²⁺ bound L. lactis yaoB 5′ UTR (yybP-ykoY family) aptamer domain reveals a four-way helical junction (4WJ) comprising of tandem coaxially stacked helices (P2 on P1 and P4 on P3) with the two ‘legs’ of the H-like structure docked at a conserved interface between loop (L) L1 and L3 (Figure 1A) (28). A single Mn²⁺ ion (M_B) is coordinated by six inner-sphere interactions from nucleotides in loops L1 and L3 including an A41 N7-Mn²⁺ coordination bond; a Mg²⁺ ion is also found nearby in M_A (28) (Figure 2). The binding of a single Mn²⁺ ion influences stability of the L. lactis yaoB riboswitch like that of a fluoride-sensing riboswitch (44) but in a way that is distinct from the Mg²⁺-sensing riboswitch (or M-box), in which multiple Mg²⁺ ions cooperatively impact the stability of the riboswitch (45,46).

The yybP-ykoY family riboswitch located upstream of the mgtA gene in Streptococcus pneumoniae D39 reveals, in contrast to the L. lactis structure, a three-way helical junction (3WJ) with P4 replaced by an invariant UNAAA (N, can be U or G) sequence (Figure 1B; Supplementary Figure S1C) followed by a transcription termination hairpin with a U-rich sequence (Supplementary Figure S1A). This UNAAA sequence is also found close to a 3WJ in the twister ribozyme (AUAAA), where it induces a sharp turn in the backbone at a helical junction (47). The 5′ end of L1 loop is highly conserved, while the 3′ end contains an invariable AGA nucleotide sequence (nucleotides 101–104) preceded by a variable nucleotide at position 101 that is only conserved as a guanosine among Streptococcus spp (Figure 1B; Supplementary Figure S1C). These nucleotides are anticipated to contribute to the metal binding pocket of L1 and L3 in the mgtA riboswitch. The metal coordinating nucleotides C49 and A50 as defined by the previous structural work as C49 and A41 in the L. lactis RNA (Figure 2) (28,30) within the L3 bulge are also invariant. Finally, the C46-G83 base-pair in the P3.1 stem above L3 is anticipated to hydrogen bond to A18 to form a cross-helix A minor interaction as in other yybP-ykoY family riboswitches (Figure 1).

Mn²⁺ and Ca²⁺ to the mgtA riboswitch alters an RNA conformational equilibrium

A comparison of the E. coli mntP riboswitch in the divalent metal-bound and metal-free (Na⁺ only) apo states suggest that the aptamer domain is pre-organized into a conformation that is quite close to the metal-bound form (Supplementary Figure S2), with major differences occurring in L3, which is largely unstructured in the absence of metal (28). The extent to which this is true in solution, however, and for this particular riboswitch specifically, is unclear. To address this, we monitored the folding equilibrium of the S. pneumoniae mgtA riboswitch aptamer domain by native PAGE, evaluating the impact of divalent metal cations Mn²⁺, Ca²⁺ and Mg²⁺ (Figure 3).

Figure 3.

Mn²⁺ or Ca²⁺ binding to S. pneumoniae mgtA riboswitch induces conformational change and promotes compact folding. Global hydrodynamic analysis of the RNA aptamer in the presence of metal ions by native-PAGE. (A) Influence of salt solution, Mg²⁺ and Mn²⁺ on folding of the wild-type RNA aptamer domain. (B) Influence of Ca²⁺ in a background of 0.5 mM NaCl. (C) Effects of key metal-binding nucleotides in L1 and L3 that contribute to the overall global structure and compaction of the RNA in a background of 0.5 mM NaCl. Representative gel is shown from at three independent experiments. E, extended form; C, compact form; * and ** denote the two independent folded conformations for the triple mutant RNA, where * is a new conformation.

In these experiments, the RNA aptamer domain was denatured and refolded at room temperature in a step-wise progression by first adding a monovalent salt (50 mM NaCl), followed by divalent metal cation. Two conformations were resolvable by native PAGE run with Mg²⁺ in the running buffer: an extended or open form (E, upper band) and a more compact form (C, lower band) (Figure 3A). In the absence of added salt, the majority of the RNA (≥80%) is found in the extended form, while inclusion of the 50 mM monovalent salt drives the RNA towards a compacted form. Addition of 1 mM Mg²⁺ to 50 mM Na⁺ shifts the equilibrium further to the compacted form but not completely (Figure 3A). The addition of as little as 0.5 mM Mn²⁺ is efficient at compacting the RNA with ≈90% of the RNA adopting the compact form (Figure 3A and Supplementary Figure S3A). As anticipated from a cation competition model, the aptamer appears to fold less efficiently when 1 mM Mg²⁺ and 0.5 mM Mn²⁺ are simultaneously added into the folding buffer relative to folding in the presence of Mn²⁺ alone (Figure 3A). Indeed, nucleotides in L1 and L3 of the L. lactis yaoB riboswitch can coordinate binding of either Mn²⁺ or Mg²⁺ (28). The degree to which increasing concentrations of Mn²⁺ shifts the conformational equilibrium to a compact conformation can be modelled assuming that the apo form in 3 mM Mg²⁺ exists as a mixture of extended and compact and Mn²⁺-bound form is exclusively compact. This model of the relative concentrations of the two different RNA conformers gives an effective equilibrium dissociation constant (K_d) of 0.14 ± 0.02 mM (Supplementary Figure S3A and B) in 3 mM Mg²⁺, or ≈6-fold weaker than that reported for L. lactis yaoB riboswitch (K_d of 25 μM for Mn²⁺) using isothermal titration calorimetry (ITC) in a background of 5 mM Mg²⁺ (30). Quantitative ITC data for the binding of the pneumococcal mgtA riboswitch aptamer to divalent cations is presented below.

Although the likely cognate metal ion for the S. pneumoniae mgtA riboswitch is Mn²⁺ given its classification as a yybP-ykoY family riboswitch, the mgtA riboswitch incorporates the ribosome binding site of the mgtA gene, which encodes P_II-type ATPase transporter (MgtA) that has been implicated in Ca²⁺ efflux (31); furthermore, biophysical experiments show that the analogous transporter from Listeria monocytogenes (LMCA1) transports Ca²⁺ (48). Ca²⁺ binding to this family of RNAs has not yet been investigated. Using native-PAGE, we show that 0.5 mM Ca²⁺ also affects the global hydrodynamics of the RNA aptamer, driving the RNA to the more compact form that is similar in mobility to that obtained upon incubation with Mn²⁺ (Figure 3B).

Mutant mgtA riboswitch aptamers have altered structures or altered Mn²⁺ sensitivities

To assess which nucleotides coordinate Mn²⁺ in the S. pneumoniae mgtA RNA aptamer, we mutated conserved nucleotides in L1 and L3 that comprise the metal binding site region as detailed in other yybP-ykoY structures (Figure 2) and tested the impact of these substitutions on the RNA aptamer folding equilibrium by native-PAGE (Figure 3C). These include G16, G17 and A18 on the 5′ side of L1, C49 and A50 in L3 and A102, G102 and A104 on the 3′ side of L1 (S. pneumoniae numbering) (see Figure 1B). The G16A/G17A and C49A/A50U RNAs show no change in electrophoretic mobility by native-PAGE over the same range of Mn²⁺ concentration that stabilizes the compacted (C) form of the wild-type RNA (Figure 3C and Supplementary Figure S3C and D), suggesting a loss of Mn²⁺ binding ability. We note that the G16A/G17A mutant RNA appears to migrate slightly faster than the extended (E) form of the WT RNA (Figure 3C). Although the C49A/A50U mutant has lost its ability to bind metal (Supplementary Figure S3D), this mutant RNA migrates similarly with the compacted WT RNA. The triple A102G/G103A/A104G (L1) mutant RNA, on the other hand, simultaneously adopts two RNA conformations that migrate closely with that of the compacted WT RNA, but each is formed independent of Mn²⁺ concentration (Supplementary Figure S3E, Figure 3C). Finally, targeting the predicted cross-strand A-minor interaction (nucleotide A18; Figure 1B) with a cytosine substitution (A18C) shifts the E-to-C equilibrium far toward E, as expected, in a background of 3 mM Mg²⁺ (Supplementary Figure S3F and G) while also increasing the apparent K_d for Mn ≈6.5-fold to 0.90 ± 0.01 mM.

Thus, all four mutant RNAs perturb the RNA conformational ensemble, at least as observed by native-PAGE, and only one of four RNAs (A18C) is responsive to the presence of Mn²⁺, albeit at much higher metal concentrations. These data suggest that Mn²⁺ binding may not be required to globally fold the S. pneumoniae mgtA riboswitch aptamer into an ‘X-like’ or undocked structure in the presence of physiological (low mM) Mg²⁺, as all mutant RNAs adopt an E- or E-like state, a finding consistent with recent single-molecule FRET experiments (49).

SHAPE (selective 2′-hydroxyl acylation analyzed by primer extension) probing of the wild-type aptamer

The impact of divalent metal binding on the structure of the S. pneumoniae mgtA RNA aptamer was further investigated by the structural probing method SHAPE. SHAPE experiments measure the relative rates of reactivity of 1-methyl-7-nitroisatoic anhydride (1M7) to 2′-hydroxyls and generally report on local flexibility of an individual nucleotide. In the absence of divalent metal (50 mM Na⁺ only), nucleotides in L1 (16–21; 102–104) and L3 (residues 48–50) show high to moderate reactivity (Figure 4A), suggesting that no metal-binding pocket is formed. A18 of the A-minor interaction is also moderately flexible in the absence of metal ions. The addition of Mg²⁺ stabilizes base-pairing in the P2 stem (residues 28–29), destabilizing the base of the P3.2 stem, while quenching flexibility of all nucleotides implicated in metal binding, including both sides of L1 and C49 and A50 in L3 (Figure 4B). The modest reduction in flexibility observed for nucleotides A18 in L1 and C49 and A50 in L3 suggests that the A-minor interaction forms in the presence of Mg²⁺ alone. Addition of Mn²⁺ to a background of Mg²⁺ leads only to a small change in the flexibility of nucleotides C49 and A50, which are predicted to directly interact with the Mn²⁺ M_B ion (vide infra), making them more rigid (Figure 4C). The addition of Ca²⁺ results in a SHAPE profile (Figure 4D) that is statistically indistinguishable from that obtained upon addition of Mn²⁺. These data taken together reveal that the RNA conformation obtained in the presence of the 3 mM Mg²⁺ alone, Mg/Mn and Mg/Ca are very similar and that major differences in reactivity relative to the Na⁺-only structure are localized to the metal-binding region. These findings are consistent with native-PAGE analysis which reveals roughly two distinct conformations with distinguishable, yet rather similar, electrophoretic mobilities.

Figure 4.

Mn²⁺ and Ca²⁺ impact nucleotide flexibility in the mgtA RNA aptamer. (A) SHAPE reactivity of the S. pneumoniae mgtA RNA aptamer domain in the presence of various divalent metals. Regions of low reactivity (<0.4) are colored gray, moderate reactivity (0.4–0.85) are yellow, and high reactivity (>0.85) are red. Each bar represents a single nucleotide in the sequence as numbered on x-axis. (B) Summary of the flexibility of nucleotides in the RNA aptamer.

Small angle X-ray scattering (SAXS) analysis of mgtA riboswitch aptamers

Small angle X-ray scattering (SAXS) was next used to obtain additional insights into the global fold of the wild-type mgtA RNA as a function of divalent metal ion status, compared to the G16A/G17A L1 mutant which fails to adopt the compact or C-state. Refolding of the aptamer in buffer with NaCl only and no divalent metal ions results in the largest radius of gyration (R_g) (Supplementary Figure S4; ≈37 Å) with clear evidence of conformational heterogeneity in this state. Addition of Mg²⁺ results in a significant increase in RNA compaction, with the R_g some 10 Å smaller, to 26.7 ± 0.5 Å. Addition of Mn²⁺ to the Mg²⁺-folded structure results in a similarly compacted structure, that is perhaps slightly more extended (R_g = 27.6 ± 0.4 Å) with no significant change observed with the addition of Ca²⁺ to the Mg²⁺ folded RNA. The dimensionless Kratky plots also reveal that the mgtA aptamer folds upon the addition of Mg²⁺, while inclusion of other transition metals tested only change the overall compaction of the RNA slightly (Figure 5A and B). A significant change in the pair-wise distance distribution is observed with Mg²⁺ showing a larger number of shorter distances (20–40 Å) within the RNA (Figure 5C), consistent with a folded form of the mgtA aptamer RNA. Only a minor change in the pair-wise distance distribution results when Mn²⁺ or Ca²⁺ are present in a background of Mg²⁺ (Figure 5C). Taken together these data are consistent with the native-PAGE and SHAPE results, demonstrating that different divalent metal ions can lead to conformational change and compaction of the mgtA RNA aptamer.

Figure 5.

Global structure analysis of the mgtA RNA aptamer with divalent metal ions. (A) Scattering profiles collected at separate RNA concentrations 0.8, 1.6 and 3.2 mg/ml with indicated metal, then merged. (B) Dimensionless Kratky plots. (C) Pair-wise distance distribution plots for all RNAs. (D) Simulated 3D structure of the RNA aptamer by SimRNA embedded in the envelope obtained from SAXS in the presence of Mn²⁺. SAXS simulated envelope models for WT RNA with Ca²⁺ (E), Mg²⁺ (F) and the G16A/G17A mutant RNA that is unable to bind metal (G).

We next used these data to calculate a three-dimensional (3D) envelope (or bead model) of the mgtA RNA aptamer, which could only be determined in the presence of divalent metal ions, a finding consistent with the conformational heterogeneity in the metal-free form of the RNA as evidenced by the non-linear Guinier plot (Supplementary Figure S4A). As expected, the resulting averaged envelopes obtained in the presence of Mg²⁺ plus Mn²⁺ or Mg²⁺ plus Ca²⁺ are similar (Figure 5D and E), but yet distinct from that of Mg²⁺ alone (Figure 5F). A simulated structural model of the aptamer embedded into the SAXS envelope in the presence of Mg²⁺/Mn²⁺ or Mg²⁺/Ca²⁺ fits well (Figure 5D), showing that in the presence of Mn²⁺ or Ca²⁺, the three hairpins likely organize into two coaxially stacked helices (P2 on P1) connected to a third helix (P3) at the three-way junction (3WJ).

In contrast to the WT RNA aptamer, the non-metal binding G16A/G17A mutant RNA that fails to adopt the compact form at any Mn²⁺ concentration by native-PAGE analysis (Supplementary Figure S3C) adopts a more extended average conformation (R_g = 29.3 ± 0.4 Å), independent of Mn²⁺ (Figure 5B and C). The resulting averaged SAXS-derived bead model yields a ‘Y-like’ topology, suggesting that the G16A/G17A mutant RNA is characterized by an open or extended conformation minimally comprising three helices connected by a 3WJ (Figure 5G).

The mgtA riboswitch aptamer binds Ca²⁺ more tightly than Mn²⁺

The genomic location of the S. pneumoniae mgtA riboswitch as well as our own hydrodynamic data (Figures 3–5) suggest that Mn²⁺ and possibly Ca²⁺ may be regulatory. We next used ITC to monitor the binding of Mn²⁺ and Ca²⁺ directly, monitored at what we anticipate is near physiological conditions of Mg²⁺ (3 mM). The binding of Mn²⁺ to the mgtA RNA aptamer is exothermic, whereas the binding of Ca²⁺ is endothermic, where Mn²⁺ binding is more strongly entropically-driven relative to Ca²⁺ (Figure 6 and Table 1). Both metals bind to the mgtA riboswitch with an apparent stoichiometry, n, ≈1, likely reporting on the displacement of Mg²⁺ bound to the M_B site by Ca²⁺ or Mn²⁺ (Figure 2A) (28). The effective K_d for Ca²⁺ is low micromolar (1.72 ± 0.03 μM), which is 50-fold higher affinity than for Mn²⁺ (K_d = 54 ± 26 μM). The K_d for Mn²⁺ obtained from ITC is somewhat tighter than that derived from native-PAGE folding equilibrium analysis, although both are in the 10⁻⁴ M Mn²⁺ range (Supplementary Figure S3A and B).

Figure 6.

Thermograms of the interaction of the mgtA aptamer with Mn²⁺ (A) or Ca²⁺ (B) by isothermal titration calorimetry. For each panel, the upper panel shows the raw ITC data plotted as the change in power (μcal s⁻¹) required to maintain equal temperature between the sample and reference cells as the function of time; lower panel shows intergated heat normalized for mol of injectant added. The thermodynamic parameters for metal binding defined by the continuous lines drawn through the data are summarized in Table 1.

Table 1.

Thermodynamic parameters of cation interactions with the mgtA RNA aptamer

Parametera	Mn2+	Ca2+
K d (μM)	54 ± 26	1.72 ± 0.03
n	1.0 ± 0.0	1.07 ± 0.0
ΔH (kcal/mol)	1.1 ± 0.1	–3.6 ± 0.1
TΔS (kcal/mol)	7.0 ± 0.2	4.3 ± 0.1
ΔG (kcal/mol)	–5.9 ± 0.02	–7.9 ± 0.2

^aMean and standard deviations are obtained from duplicated independent experiments. All experiments were performed in a background of 3 mM MgCl₂ at 25°C. Data reported here are calculated from a one-site random binding model.

A significant number of the structural models of metal-bound yybP-ykoY riboswitches have been reported and are often used to interpret the sensing mechanism of riboswitches (Figure 2B and C). Moreover the recent reported affinities for Cd²⁺, K_d in the sub-mM range depending on the construct (30), suggest that Cd²⁺ could potentially turn on the yybP-ykoY riboswitches. To test the significance of those models in the context of the mgtA riboswitch, we also determined Cd²⁺ binding to the mgtA riboswitch. Although the mgtA RNA aptamer is compacted in the presence of 500 μM Cd²⁺ in a background of 3 mM Mg²⁺ (Supplementary Figure S5A), higher concentrations of Cd²⁺ drive the RNA aptamer into a more slowly migrating conformation(s) by native-PAGE. This suggests that at low concentrations, Cd²⁺ may bind in the same pocket (M_B) (see Figure 2B) as Mn²⁺ and Ca²⁺. Both SHAPE (Supplementary Figure S5C) and SAXS (data not shown) analyses suggest that an extended or more open conformation is formed at high Cd²⁺ concentrations. It is also worth noting that low Cd²⁺ concentrations (≤500 μM) fail to activate transcription read-through and that higher Cd²⁺ concentrations are inhibitory (data not shown). Collectively, these data suggest that Cd²⁺ can bind and affect global change of the mgtA riboswitch, however the functional outcome is not likely physiologically relevant.

The mgtA riboswitch responds more robustly to Mn²⁺in vitro

Previously, it was demonstrated by in vitro transcription (IVT) termination experiments that the L. lactis yaoB riboswitch is activated by 0.5 mM Mn²⁺ but not by other metal ions tested (Fe²⁺, Co²⁺, Ni²⁺ and Ca²⁺); Mg²⁺ also induced transcription read-through, but only when provided at high concentrations (10 mM), suggesting that Mg²⁺ is not a preferred metal (28). Likewise, an L. lactis yaoB riboswitch chimera harboring L3 mutations to match the E. coli mntP and alx riboswitch L3 nucleotides, respond to Mn²⁺ as do other transition metals tested (Cd²⁺, Co²⁺, Cu²⁺, Fe²⁺ and Ni²⁺); however, transcription read-through was not induced by Ca²⁺, Mg²⁺ and Zn²⁺ (30). Other experiments demonstrate that these yybP-ykoY family riboswitches can also mediate Mn²⁺-dependent regulation in vivo (28,29). Thus, it is not possible to predict which metal induce transcription for the mgtA RNA aptamer based solely on it being part of the yybP-ykoY family riboswitches, nor on the selectivity in terms of binding for different metal ions. This observation motivated a series of in vitro transcription termination assays to determine which metal induces transcription for the mgtA RNA.

The binding of Mn²⁺ to the S. pneumoniae RNA aptamer is hypothesized to destabilize the terminator hairpin and permit read-through transcription of full-length mgtA mRNA transcript (Supplementary Figure S1C). Indeed, titration of increasing Mn²⁺ concentrations into the reaction led to the synthesis of full-length transcripts by E. coli RNA polymerase (Figure 7A) in a background of 250 μM Mg²⁺, revealing that Mn²⁺ binding to the aptamer activates transcription read-through. Ca²⁺ is also capable of activating transcription read-through (Figure 7B), but to a much lower extent compared to Mn²⁺ (Figure 7C). The half-maximal transcriptional activation by Mn²⁺ (0.23 ± 0.1 mM; Hill number, n_H = 1) reveals that 5-fold less Mn²⁺ is needed to activate transcription read-through when compared to Ca²⁺, which is half-maximal at 1.1 ± 0.1 mM; Hill n_H = 2.5 ± 0.6. Thus, despite the fact that Ca²⁺ binds more tightly to the RNA aptamer than Mn²⁺ (Figure 6) and appears to yield a similar conformation in solution, it stimulates read-through transcription to a far lesser degree. We found that Mg²⁺ alone is also capable of stimulating transcription read-through of mgtA (Supplementary Figure S6) with the addition of 3 mM Mg²⁺ resulting in ≈50% maximum read-through (Supplementary Figure S6B), yielding an effective K_d of 1.7±0.8 mM (Hill n_H = 1.0 ± 0.8). Thus, Mn²⁺ is the most effective activator of mgtA transcription read-through and is uniquely functional at sub-mM Mn²⁺ in a concentration range comparable to the K_d^Mn.

Figure 7.

Mn²⁺ activates in vitro transcription of mgtA better than Ca. ³²P labeled products produced in presence of Mn (A) and Ca (B) (in a background of 250 μM Mg²⁺) were separated by PAGE. Termination (T) products reflect terminator formation. Read-through (RT) products form as a result of metal-binding and stabilizing the aptamer thereby preventing formation of the terminator. (C) Quantification of in vitro transcription termination assays, with the continuous lines fit to a Hill model of activation (see text for details).

The S. pneumoniae mgtA riboswitch responds more robustly to Mn²⁺in vivo

We next determined if the S. pneumoniae mgtA riboswitch could also mediate Mn²⁺- and Ca²⁺-dependent regulation in vivo. S. pneumoniae cells lacking the Mn-specific exporter MntE (encoded by mntE) were used to induce Mn-stress growth conditions; ΔmntE cells accumulate high levels of intracellular Mn compared to wild type (WT) cells (34,50). During cellular Mn stress, mgtA mRNA and MgtA protein levels increase 5- and 4-fold, respectively, for ΔmntE mutants compared to WT cells (Figure 8 and Supplementary Figure S7). Clotrimazole, known to inhibit Ca²⁺ transport of eukaryotic sarco/endoplasmatic reticulum Ca²⁺-ATPase (SERCA P_II-type ATPase), served as a suitable agent for imposing Ca²⁺-stress in S. pneumoniae WT cells (Supplementary Figure S8A). We note that clotrimazole treatment did not significantly affect intracellular Mn levels (Supplementary Figure S8B). We found that MgtA protein synthesis remained unchanged during Ca²⁺-stress, despite a 4-fold increase in cellular Ca²⁺ levels (Supplementary Figure S8C). No further metal activation studies were performed in S. pneumoniae, since we subsequently discovered that deletion of mgtA and possibly lower cellular MgtA protein levels are detrimental to cell viability (see below).

Figure 8.

mgtA is induced during Mn-stress. Exponentially growing S. pneumoniae cells were diluted into pre-warmed BHI broth with 0 (black) or 200 μM (gray) Mn. Cells were harvested after 3.5 h growth. (A) Relative transcript levels of mgtA RNA. (B) Relative intensities of FLAG-tagged MgtA protein. The mean of at least three independent cultures ± SEM are shown. *P-value ≤ 0.05.

As an alternative method, we fused the S. pneumoniae mgtA leader region containing the mgtA riboswitch to a lacZ gene and monitored metal activation in Bacillus subtilis WT and ΔmntR strains. The latter strain experiences Mn toxicity, because the Mn-repressor MntR coordinates transcriptional regulation of the Mn uptake transporters (MntH and MntABC) and the Mn efflux pumps (MneP and MneS) (51,52). In the B. subtilis ΔmntR strain, the mgtA-lacZ mRNA level increases several-fold with the addition of Mn²⁺ to the growth medium (Figure 9A). Likewise, increasing Mn²⁺ concentrations stimulates expression of β-galactosidase up to 3.5-fold (black solid circles, Figure 9B). Removal of the terminator hairpin led to significantly higher activity (red solid squares), while deletion of the aptamer domain resulted in a minimal change in activity (grey solid triangles, Figure 9B). No response was observed in the absence of the 100 nt region upstream of the predicted mgtA riboswitch (open symbols, Figure 9B).

Figure 9.

The S. pneumoniae mgtA riboswitch senses and responds specifically to intracellular Mn in B. subtilis. Exponentially B. subtilis cells growing in LB treated with indicated metal for 2 h, then harvested. (A) Relative transcription of P_mgtA-lacZ fusion in ΔmntR cells grown with increasing concentrations of Mn. (B) β-Galactosidase activity in ΔmntR carrying various promoter-lacZ fusions in response to increasing concentrations of Mn. (C) β-galactosidase activity in response to various metal cations. (D) β-Galactosidase activity in WT cells containing WT-mgtA leader sequence in response to Mn and Ca. (E) β-galactosidase activity in ΔmntR carrying Pr_full (WT) and mutant forms the mgtA riboswitch in response to various metal cations. The mean of at least three independent cultures ± SEM are shown. Pr_Full, promoter plus intact 5′ UTR (185 nt; see Supplementary Figure S1); Pr_Apt, promoter plus aptamer (nucleotides 1–124); and Pr_Term, promoter plus terminator hairpin (nucleotides 101–185).

In contrast to Mn²⁺, the addition of other metal ions (Zn²⁺, Cu²⁺, Fe²⁺, Ni²⁺, Co²⁺, Mg²⁺ or Ca²⁺) to the growth medium had little to no effect on β-galactosidase expression (Figure 9C). Again, no response was observed in the absence of the 100 nt region upstream of the aptamer (Supplementary Figure S9). Deletion of the transcriptional terminator hairpin led to several-fold higher activation, while loss of the aptamer domain showed little if any change (Figure 9C). In B. subtilis WT cells, ≥400 μM Mn²⁺ was required to observe activation of β-galactosidase expression (Figure 9D). No activation was observed in WT cells with other metal ions tested (Supplementary Figure S9B), including Ca²⁺ (Figure 9D) even when added up to 10 mM (data not shown). Together, these data demonstrate that the S. pneumoniae mgtA riboswitch functions as a Mn²⁺-specific ‘on’ switch in vivo, inducing expression of MgtA.

To further connect aptamer folding with function and confirm metal specificity in vivo, key nucleotides within the S. pneumoniae mgtA RNA aptamer shown to disrupt folding of the aptamer were mutated in the context of the 5′ UTR and tested for functionality in vivo in a heterologous B. subtilis expression system. Both L1 (G16A/G17A) and L3 (C49A/A50U) substitutions reduced expression of β-galactosidase for all metal ions tested (Figure 9E). In contrast, the triple mutation of the 3′ side of L1 (A102C/G103A/G104G) results in high-level expression for all metals tested except for Zn²⁺ and Cu²⁺, which inhibit growth under these conditions, to a level consistent with unregulated expression, i.e. that obtained with aptamer alone (Figure 9C). This mutant 5′ UTR RNA may mis-fold in the cell and lead to a loss or destabilization of the terminator hairpin (see Supplementary Figure S1B). Taken collectively, these data correlate well with our global conformational analyses (Figures 3–5).

The mgtA riboswitch regulates a P_II-type ATPase that protects cells against Mn toxicity

Our results so far indicate that the S. pneumoniae mgtA riboswitch functions as a Mn²⁺-specific sensor that turns on a predicted P_II-type ATPase metal effluxer, MgtA, that we reasoned might play a role in cellular Mn homeostasis. To investigate this, we constructed S. pneumoniae D39 strains harboring clean deletions of mgtA and mntE, singularly or in combination. Transformation efficiency for the ΔmgtA strain appeared typical, despite a two-thirds reduction in colony size, while the ΔmgtA ΔmntE strain had the lowest efficiency, when selected on TSAII+SBA medium containing antibiotic under 5% CO₂ atmosphere. Colony appearance was uniform in size and shape (data not shown). Absence of oxygen did not improve colony size of ΔmgtA mutants during the selection process, suggesting that the colony phenotype was not a result of reactive oxygen species. Furthermore, ΔmgtA isolates failed to reach 0.2 OD in 8–12 h when inoculated in brain-heart infusion (BHI) broth. Prolonged growth (>24 h) in BHI yielded fast-growing suppressors in isolated ΔmgtA harboring strains. Supplementation of BHI with exogenous Mg²⁺, Ca²⁺, Mn²⁺ or Fe²⁺ did not affect the growth pattern observed, nor did treatment with metal chelators EDTA and desferrioxamine (data not shown). These data support the idea that MgtA plays a role in Mn homeostasis in S. pneumoniae, since combined ΔmgtA and ΔmntE mutations showed the most pronounced colony defect during strain construction. We note that a ΔmgtA mutation was reported previously in S. pneumoniae D39 (53); however, we suspect that suppressor mutations were present in this strain, given the discrepancies with other published papers (25,31).

Because of this apparent requirement for pneumococcal MgtA under metal replete conditions, we tested the ability of MgtA to rescue Mn-sensitivity in a B. subtilis ΔmntR mutant. The S. pneumoniae mgtA gene was placed under an IPTG-inducible promoter and integrated into the chromosome of B. subtilis in single copy. Induction of MgtA synthesis has no effect on wild-type cell growth over time in LB with or without Mn (Figure 10A and Supplementary Figure S10). B. subtilis ΔmntR mutants fail to grow in the presence of Mn²⁺ compared to wild-type cells (Figure 10A and Supplementary Figure S10) due to Mn accumulation (Figure 10B). This Mn sensitivity is largely eliminated by induction of MgtA synthesis (Figure 10A). Total cell-associated Mn levels do not rise in ΔmntR mutants when MgtA is overexpressed, despite treatment with exogenous Mn²⁺ (Figure 10A). Interestingly, Ca²⁺ levels are significantly lower only in ΔmntR treated with Mn²⁺ only when MgtA is induced (Figure 10B). All other metals (Fe²⁺, Zn ²⁺, Cu²⁺) examined remained similar to WT levels (Supplementary Figure S11); Ni²⁺ and Co²⁺ were below the level of detection. These data indicate that the MgtA P-type ATPase transporter is capable of transporting both Ca²⁺ and Mn²⁺ under conditions of extreme Mn²⁺ toxicity. Further studies are required to fully understand the physiological importance of Ca²⁺ transport by MgtA in cells.

Figure 10.

S. pneumoniae MgtA protein that is associated with the yybP-ykoY riboswitch rescues Mn-sensitivity in B. subtilis. Sensitivity of Mn²⁺ was monitored in B. subtilis WT and ΔmntR cells harboring the mgtA RNA aptamer sequence or MgtA in LB broth containing 0 (darker bars) or 50 μM (lighter bars) Mn overtime. Isopropyl β-D-thiogalactopyranoside (1mM) was added to induce MgtA protein synthesis. (A) 2-h cell density. (B) Total cell associated Mn and Ca determined by IPC-MS after 2.5 h growth. (C) Disk diffusion assay. The mean of at least three independent cultures ± SEM are shown for each experiment. *P-value ≤ 0.05; **P-value ≤ 0.01; ***P-value ≤ 0.001; ns, no significance.

DISCUSSION

Metal cation homeostasis is essential for colonization and pathogenesis of bacteria. As such, the mammalian host has evolved mechanisms to both restrict or intoxicate invading bacteria with specific metals, depending on the microenvironmental niche. In response, bacteria have evolved complex regulatory strategies to adapt in an effort to maintain homeostasis of essential metal ions. Recent work provides support for the idea that individual metal ions in a bacterial cell are buffered to an extent that is largely determined by the absolute sensitivities or set-points of metallosensors that regulate the expression of downstream genes, which often encode metal transporters (54,55). These metallosensors detect metal activities (free metal concentrations) around a relatively narrow concentration window defined by their K_d^Me (Me, metal), a range in concentration that is often far less than total cell-associated metal (54,55). Of importance here is that Mg²⁺ levels are thought to be maintained in the low mM range, primarily controlled by the Mg²⁺-sensing ‘M-box’ riboswitch (in B. subtilis, for example) (45), while Mn²⁺ concentrations are transcriptionally controlled by Mn²⁺-sensing protein repressors MntR in B. subtilis (56) and a related repressor PsaR, in S. pneumoniae (57). The M-box riboswitch is tuned to ≈mM affinity (55). The K_d^Mn for the regulatory ‘sensing’ site in PsaR is in the low μM range (1.3 μM), while the Salmonella MntR K_d^Mn is about 10-fold lower affinity (13 μM at pH 8.0) (55). This ‘set-point’ model suggests that Mn²⁺ may be buffered in the low μM range, which in S. pneumoniae may be further restricted by the constitutive expression of MntE, to which effluxes Mn²⁺ from the pneumococcal cell (50). In contrast to what is known about Mn²⁺ homeostasis, intracellular bacterial Ca²⁺ homeostasis remains largely unexplored, and to our knowledge a bacterial Ca²⁺-specific metallosensor or two-component response regulatory system has yet to be described.

In this report, we describe the physical and functional characterization of a Mn²⁺-sensing riboswitch that is found upstream of the mgtA gene. Using a combination of native gel mobility experiments, SHAPE probing, SAXS analysis and in vitro transcription experiments, we show that this riboswitch aptamer domain is capable of binding Mn²⁺ and Ca²⁺ in the presence of mM Mg²⁺, and that all three metals, including Mg²⁺ alone stabilize a compact state, relative to an extended or ‘open’ conformation. All three metals stimulate transcription read-through in vitro using E. coli RNA polymerase, but Mn²⁺ is clearly the most potent activating metal in these experiments, both in terms of maximal read-through activity and metal sensitivity (≤100 μM metal). The half-maximal Mn²⁺ concentration required to shift the folding equilibrium of the RNA to the compact form and stimulate in vitro transcription is within a factor of four of the K_d^Mn measured by calorimetry (50–200 μM); furthermore, the response in all three experiments conducted in 3 mM Mg²⁺ fits well to a rectangular hyperbolic function, consistent with a single Mn²⁺-sensing site on the riboswitch, which we speculate is the Mn_B site (see Figure 2A). As anticipated from previous work, mutations in the metal-site pocket abrogate or greatly weaken (A18C) Mn²⁺-dependent folding, and when incorporated into the mgtA 5′ UTR in cells, render them functionally inactive.

These physical characteristics are largely recapitulated in pneumococcal cells and in a heterologous B. subtilis host, with Mn²⁺ the only divalent metal ion capable of read-through transcription in either cellular background, which in S. pneumoniae results in a corresponding increase in the cellular amount of MgtA protein. Ca²⁺ is not a strong mgtA inducer in vivo. However, significant Mn²⁺-dependent mRNA or protein expression in either bacterial host is only observed in a strain that is extremely sensitive to Mn²⁺ toxicity, i.e. in ΔmntR B. subtilis and in ΔmntE S. pneumoniae. Under these conditions where mgtA mRNA and MgtA protein levels are higher, MgtA is capable of lowering total cell-associated Mn²⁺, consistent with a role in Mn²⁺ efflux by MgtA. These physiological findings are entirely consistent with the observed in vitro metal sensitivity of the riboswitch itself, which is in the 50–200 μM Mn²⁺ range, or set ≥5–10-fold higher than the metalloregulator proteins MntR and PsaR; unfortunately, the metal binding affinity of the MntE has not yet been determined but might be expected to be in the low μM range. Thus, if these systems are present and functioning, they would prevent intracellular Mn²⁺ from reaching the ≈100 μM range, and thus this riboswitch and subsequent MgtA protein production would simply not fire. Thus, our data support the hypothesis that the MgtA production functions as a ‘fail-safe’ or back-up system to allow S. pneumoniae to adapt to acute phases of Mn²⁺ toxicity in the host when Mn²⁺ uptake regulated by PsaR and Mn²⁺ efflux by MntE fail to avoid cellular toxicity (Figure 11) (50).

Figure 11.

Threshold model for Mn²⁺ sensing and detoxification in S. pneumoniae. The Mn²⁺ binding affinity of the S. pneumoniae mgtA riboswitch is such that there is sufficient free intracellular Mn²⁺, governed by the relative affinities of the transcriptional Mn uptake repressor PsaR and the constitutively expressed Mn-specific efflux pump MntE (34,57), to ensure MgtA is not induced and that key intracellular Mn-dependent enzymes are active. At concentrations >1 μM Mn²⁺, PsaR will bind Mn²⁺, increasing its affinity to DNA and repress psaBCA transcription, thereby reducing Mn import; excess Mn²⁺ will continue to be effluxed by MntE. Our data suggest that as free Mn²⁺ rises to ≥100 μM, the mgtA riboswitch functions as a failsafe ‘on’ signal inducing MgtA expression to prevent Mn²⁺ toxicity. In addition, the riboswitch also functions to regulate Ca²⁺ efflux under this condition.

The functional significance of Ca²⁺ binding by this riboswitch and Ca²⁺ efflux by MgtA under conditions of extreme Mn²⁺ stress (Figure 10B) are not yet known. MgtA is clearly capable of effluxing Ca²⁺ from cells and clotrimazole and extracellular Ca²⁺ interfere with this process (Supplementary Figure S8), but under what conditions this might occur during the course of a bacterial infection are not known. mgtA is the first gene in a co-transcribed two-gene operon with a downstream, non-essential gene (spd_1382) encoding a glutathione S-transferase (25). Consequently, the defective growth of ΔmgtA mutants, irrespective of Mn²⁺ status, could be due to a polarity on Spd_1382 expression and/or a second role of MgtA in cellular metabolism possibly in maintaining Ca²⁺ homeostasis. The former is less likely as the replacement of the native mgtA gene with a construct encoding MgtA triple-FLAG-tagged protein linked to an antibiotic resistance cassette did not show defective growth (data not shown). The putative essentiality of mgtA was not studied further here, but an implication of this operon arrangement is that expression of the Spd_1382 glutathione S-transferase is also controlled by the mgtA riboswitch aptamer in response to Mn²⁺ and Ca²⁺. One possibility is that under metal-replete conditions, small amounts of MgtA produced via leaky expression function in Ca²⁺ efflux, perhaps required to efficiently metallate an obligatory Ca²⁺-requiring enzyme on the outside of the pneumococcal cell; indeed, such a role for P_1B-type ATPases in metallating periplasmic, extracellular or membrane-anchored client proteins is not without precedent for Cu⁺-specific P_1B-ATPases in other bacteria (58–60). Alternatively, MgtA might play an important role in maintaining Ca²⁺ homeostasis, but total levels of Ca²⁺ simply do not change much under transition metal-replete conditions. This would be consistent with the classification and biochemical characterization of the bacterial MgtA-like transporters as primarily Ca²⁺ transporters (48). In this model, only under conditions of acute Mn²⁺ toxicity does MgtA function as a Mn²⁺ transporter, a condition bolstered by increased accumulation of the MgtA in the membrane to effect efflux of this transition metal.

As a classic P_II-type ATPase, MgtA is predicted to transport two metal cations per reaction cycle, which could in theory involve any combination of Mn²⁺ or Ca²⁺ (48). This observation is not intuitive in terms of the basis of metal ion recognition by the transporter based on the significant chemical differences between Mn²⁺ and Ca²⁺. However, it is not unprecedented. The dual metal specificity of MgtA offers striking functional parallels to the secretory pathway Ca²⁺-ATPases (SPCAs) previously identified in yeast and vertebrates and known to transport both Mn²⁺ and Ca²⁺ into the Golgi from the cytosol. Here, the metal is used to metallate key client enzymes in this compartment or is simply secreted via this route (61–63). MgtA in fact shows significant sequence similarity to SPCAs and may well harbor the same Q747A substitution that enhances Mn²⁺ transport relative to Ca²⁺ (Supplementary Figure S12). Biochemical studies of S. pneumoniae MgtA coupled with more extensive physiological characterization of the functional role of MgtA are clearly required to further elucidate the metal-dependence and functional role of the MgtA transport cycle.

SUPPLEMENTARY DATA

Supplementary Data are available at NAR Online.

FUNDING

Idaho State University start up fund (to J.E.M.); Institutional Development Award from the National Institutes of General Medical Sciences [P20GM103408 to Idaho State University]; National Institutes of General Medical Sciences [R35 GM118157 to D.P.G., R01 GM042569-25S1 to J.E.M, R01 GM127715 to M.E.W.]. Funding for open access charge: Idaho State University start up funds (to J.E.M.).

Conflict of interest statement. The authors certify that they have no affiliations with or involvement in any organization or entity with any financial interests or non-financial interests in the materials discussed in this manuscript.