Bacterial riboswitches cooperatively bind Ni(2+) or Co(2+) ions and control expression of heavy metal transporters

Abstract

Bacteria regularly encounter widely varying metal concentrations in their surrounding environment. As metals become depleted or, conversely, accrue to toxicity, microbes will activate cellular responses that act to maintain metal homeostasis. A suite of metal-sensing regulatory ("metalloregulatory") proteins orchestrate these responses by allosterically coupling the selective binding of target metals to the activity of DNA-binding domains. However, we report here the discovery, validation, and structural details of a widespread class of riboswitch RNAs, whose members selectively and tightly bind the low-abundance transition metals, Ni(2+) and Co(2+). These riboswitches bind metal cooperatively, and with affinities in the low micromolar range. The structure of a Co(2+)-bound RNA reveals a network of molecular contacts that explains how it achieves cooperative binding between adjacent sites. These findings reveal that bacteria have evolved to utilize highly selective metalloregulatory riboswitches, in addition to metalloregulatory proteins, for detecting and responding to toxic levels of heavy metals.

Figure 1. Structure and function of czc motif RNAs

(A) Consensus sequence and structural model for czc motif RNAs. Nucleotides present in greater than 97, 90 and 75% of the representatives are depicted as red, black and gray, respectively. Predicted base-paired substructures (P1 through P4), covarying nucleotides (green shading) and non covarying nucleotides (red shading) are shown. (B) RNA motif 95 Cbo from Clostridium botulinum. 5′ G residues (lowercase) were added to facilitate in vitro RNA transcription. Disruptive (M1) and restorative (M2) mutations are depicted. Sites of ion-induced changes in spontaneous cleavage are derived from C, Figure 2A, and Figure S1A. (C) In-line probing analysis of the 95 Cbo RNA with five divalent metal ions. 5′ ³²P-labeled precursor RNAs were subjected to in-line probing without ligand (−) or with 0.3 mM of the divalent ion denoted for each lane. Non-reacted (lane NR) Partial digestion of RNA with RNase T1 for cleavage after G residues (lane T1) and alkaline conditions for cleavage at every position (lane OH) are shown. Red brackets highlight cleavage products whose yields are altered by ligand addition. (D) Periodic table highlighting different effects of metal ions on the czc RNA. The characteristics of various metal cations (20 μM) were established by in-line probing shown in Figure S1.

Figure 2. Cooperative ligand binding of a czc motif RNA

(A) In-line probing analysis of the 95 Cbo RNA with increasing concentrations of Co²⁺ or Ni²⁺ (0.78 μM to 200 μM) shows five regions of structural changes in the RNA (red brackets numbered 1 to 5). Lanes are marked as in Figure 1. (B) Plot of the normalized fraction of RNA cleavage versus Co²⁺ derived from the data in panel A shows half-maximal structural change at 6.5μM Co²⁺. (C) Hill plot of the binding data observed for Co²⁺, Ni²⁺, and Mn²⁺. Estimated Hill coefficients (n) are derived from the slopes of the lines. R² values for the curve analyses are 0.995, 0.995 and 0.998 for Co²⁺, Ni²⁺, and Mn²⁺ respectively. Also see Figure S2 for Ni²⁺ and Mn²⁺ binding assays.

Figure 3. Crystallographic model of a NiCo in complex with Co²⁺ ions

(A) Secondary structure diagram of Eba NiCo according to the crystallographic model. Seven nucleotides were added at the 5′ and 3′ termini (black) to facilitate crystallization. Two nucleotides in the L4 loop (black) were not modeled due to insufficient electron density. Helices P1-P2 (pink) and P3-P4 (gray) form a “H”-shaped structure stabilized at the junction by four Co²⁺ ions. Antiterminator nucleotides (cyan) and long range interactions (dashed) are indicated. Inner sphere (filled symbols) and water mediated (open symbols) contacts to different metals are shown in key (B) Crystal structure of NiCo shows two sets of coaxially stacked helices P1-P2 (pink) and P3-P4 (grey). Interhelical nucleotides coordinate four Co²⁺ ions (green). Anti-terminator nucleotides 78 to 98 (cyan) are sequestered within P4 and P1 makign direct contacts with Co²⁺ ions. Seven K⁺ (dark blue) and two Mg²⁺ (light blue) ions are bound to NiCo. (C) Nucleotides that undergo modulation by in-line probing (red) lie at or near the Co²⁺ binding sites. (D) A Co²⁺ anomalous difference fourier map contoured at 4.5σ (yellow mesh) reveals four Co²⁺ ions (green). C1, C2 and C3 reside on a plane perpendicular to the long axis of NiCo (black disk). (E–F) Side and top view of the molecule shows a highly splayed out structure with minimum long range contacts. Also see Figure S3 for in line probing of Eba NiCo RNA and Table 1 for crystallography data collection and refinement statistics.

Figure 4. Co²⁺ binding sites in a NiCo aptamer

(A to D) Co²⁺ ions bind NiCo in octahedral coordination sites. Metal coordinating nucleotides and water molecules (blue spheres) are indicated. (E) An experimental SAD electron density map contoured at 2.5σ (blue) corresponding to the NiCo crystallographic model is shown. A cobalt anomalous fourier map contoured at 4.5σ (yellow) helped identify Co²⁺ ions (green sphere) unambiguously in the structure. Also see Figure S4 for more information on cobalt site electron density.

Figure 5. An interconnected network of molecular contacts between Co²⁺ sites in a NiCo RNA

(A) Co²⁺ ions (green) bound to NiCo are coordinated by interhelical nucleotides from diametrically opposite sides of the RNA. G87 coordinates cobalt 1 via N7 and cobalt 2 via its ribose oxygen. G45 coordinates cobalt 2 via water mediated contacts with a ribosyl oxygen and cobalt 3 via its N7. These long range connections (pink dashed lines) extend from A14 to G86 via G45 and G87, connecting three of the Co²⁺ sites. Coordination distances are listed in Table S1. Also see Figure S5 for global structural changes in a NiCo RNA. (B) Secondary structure schematic summarizing the connectivity of Co²⁺ ions with interhelical residues. Co²⁺ coordinating nucleotides are the most tightly conserved in the NiCo sequence.

Figure 6. Effects of functional group changes on Co²⁺ binding by a bimolecular NiCo riboswitch aptamer

(A) The bimolecular aptamer derived from E. bacterium NiCo. Complex formation between RNA Eba 1 (black) and chemically altered RNA Eba 2 (grey) was favored by increasing the base-pairing potential of stems P1 and P4 by the addition of several non-native nucleotides (lowercase letters). Nucleotides G86, G87 and G88 were independently modified as denoted, and circled nucleotides were monitored for spontaneous cleavage using in-line probing assays. (B) Plot of the K_D values for Co²⁺ binding by the bimolecular aptamer formed from various Eba 2 constructs. Values were estimated from the modulation of in-line probing bands at four different nucleotides as denoted. (C) Plot of the in-line probing band intensity for various Co²⁺ concentrations at sites G46 and G40 for the bimolecular aptamer formed using WT Eba 2 RNA or wherein G87 or G88 were replaced with N7-deaza nucleobases.

Figure 7. Control of gene expression by a NiCo riboswitch

(A) Sequence and predicted secondary structure of the E. bacterium NiCo RNA. Nucleotides in P4 and downstream of NiCo (gray shading) can form an altered base paired terminator structure as shown. (B) In vitro transcription in increasing metal concentrations reveals a Ni²⁺- and Co²⁺-dependent increase in antiterminated full length (FL) RNA transcript and a corresponding decrease in terminated (T) product. M3 mutation that disrupts Co²⁺ binding exhibits loss of antitermination. (C) Plot of the yield of full-length transcripts versus Co²⁺ or Ni²⁺ concentration for WT and M3 templates. The metal ion concentrations at which elongation efficiency of WT is half-maximal (T₅₀) are indicated with dashed lines. The R² values for the curve fitting analyses are 0.981 and 0.987 for Co²⁺ or Ni²⁺ with WT templates, respectively. (D) Ni²⁺-induced expression of a gene associated with a NiCo riboswitch in Clostridium scindens. Changes in mRNA level of COG0053 and GAPDH genes in cells exposed to various concentrations of NiCl₂ were measured. MDH mRNA levels, which are expected to be unchanged by Ni²⁺ addition, were used to normalize. Each symbol represents the mean of three independent replicates and error bars represent standard error of the mean. GAPDH, predicted glyceraldehyde-3-phosphate dehydrogenase gene. MDH, predicted malate dehydrogenase gene. Also see Figure S6 for additional in vitro transcription data.