Lithium-sensing riboswitch classes regulate expression of bacterial cation transporter genes

Abstract

Lithium is rare in Earth's crust compared to the biologically relevant alkali metal cations sodium and potassium but can accumulate to toxic levels in some environments. We report the experimental validation of two distinct bacterial riboswitch classes that selectively activate gene expression in response to elevated Li⁺ concentrations. These RNAs commonly regulate the expression of nhaA genes coding for ion transporters that weakly discriminate between Na⁺ and Li⁺. Our findings demonstrated that the primary function of Li⁺ riboswitches and associated NhaA transporters is to prevent Li⁺ toxicity, particularly when bacteria are living at high pH. Additional riboswitch-associated genes revealed how some cells defend against the deleterious effects of Li⁺ in the biosphere, which might become more problematic as its industrial applications increase.

Figure 1

Three distinct classes of riboswitch candidates for alkali metal cations. (Top) Sequence and secondary structure models for the nhaA-I, nhaA-II and DUF1646 motif RNAs. Consensus models for nhaA-I and nhaA-II were generated based on updated sequence alignments (see supplemental files). The consensus model for DUF1646 (Na⁺-I riboswitch aptamer) was prepared as described elsewhere. (Bottom) Annotated functions of proteins encoded by genes (parentheses) commonly associated with the three riboswitch candidates (see also Supplementary Table I in the Supplement). Data are based on the total number of non-redundant representatives as noted.

Figure 2

Li⁺ triggers gene expression mediated by bacterial nhaA-I and nhaA-II motif RNAs. (A) A nhaA-I riboswitch-reporter fusion construct was created by fusing a representative associated with the nhaA gene of Azorhizobium caulinodans to a β-galactosidase gene (lacZ). Red nucleotides correspond to the highly conserved nucleotides characteristic of this RNA motif class (Fig. 1A). Boxed nucleotides identify locations of G-to-A mutations present in construct M1, which is a mutant riboswitch RNA used in several subsequent experiments. (B) Agar-diffusion assays were conducted with E. coli cells (either wild type [WT] or a nhaA gene deletion [ΔnhaA] strain) carrying the nhaA-I reporter construct in A. Cells were spread on LBK (pH 9.1) agar media containing X-gal, and filter disks using 10 μL applications of 5 M chloride salts of various monovalent ions as indicated, except that reduced concentrations of KCl (3 M) and RbCl (0.5 M) were used due to limited solubility. (C) A nhaA-II riboswitch-reporter fusion construct was created by fusing a representative associated with the nhaA gene of Brevundimonas subvibrioides to lacZ. Additional annotations are as described for A. (D) Agar-diffusion assays with various cations were conducted with the nhaA-II reporter as described in C. Additional annotations and details are as described for B.

Figure 3

Quantitation of gene expression mediated by representative Li⁺-I and Li⁺-II riboswitches. (A) Left: E. coli cells carrying the WT nhaA-I (Li⁺-I) riboswitch-reporter fusion construct (see Fig. 2A) grown in low salt LB medium at pH 9 either without or with supplementation with 50 mM LiCl, NaCl, or KCl as indicated. Either X-gal (top) or ONPG (bottom), respectively, was added after overnight incubation of cultures to visualize or quantify β-galactosidase reporter activity. Right: E. coli cells carrying the WT nhaA-I (Li⁺-I) or M1 nhaA-I (Li⁺-I) riboswitch-reporter fusion construct (see Fig. 2A) cultured in low salt LB medium at pH 9 either without or with supplementation of 50 mM LiCl. The mean and standard deviation values are presented for experiments conducted in triplicate (n = 3). (B) E. coli cells carrying the WT nhaA-II (Li⁺-II) riboswitch-reporter fusion construct (Fig. 2C) were cultured in low salt LB medium at pH 9.0 without or with supplementation of 50 mM LiCl, NaCl, or KCl as indicated.

Figure 4

Li⁺ induces structural modulation of a Li⁺-I aptamer. (A) Sequence and secondary structure model for the Li⁺-I riboswitch aptamer construct 73 nhaA derived from the nhaA gene from P. monteilii. Lowercase g letters identify non-native nucleotides added to facilitate production by in vitro transcription and red nucleotides are highly conserved in nhaA-I motif RNAs as depicted in Fig. 1A. Boxed nucleotides at positions 56 and 57 were mutated to A nucleotides in construct M1. Nucleotides circled in red are among those that undergo reduced spontaneous cleavage during in-line probing assays upon the addition of Li⁺, as determined from the autoradiogram depicted in B. (B) PAGE analysis of in-line probing reactions with 5′ ³²P-labeled 73 nhaA RNA in the absence of Li⁺ (–), or in the presence of Li⁺ concentrations ranging from 2 to 200 mM. NR, T1 and ^–OH identify RNAs subjected to no reaction, partial digestion with RNase T1 (cleaves after G nucleotides) and partial digestion with hydroxide (cleaves after each nucleotide). Bands corresponding to RNAs carrying a 3′ G nucleotide are identified according to the numbering system in A. A lower contrast version of the gel image is presented in Supplementary Fig. I in the Supplement. (C) Plot of the estimated fraction of RNAs bound to Li⁺ versus the logarithm of the Li⁺ concentration. Fraction bound values were estimated by quantifying band intensities at sites 1, 2 and 3 in B, which correspond to the nucleotides annotated with red circles in A. Note that fraction bound values were set to 1 at the maximum Li⁺ concentration tested because the band intensities are near zero (maximal possible suppression). The solid line depicts a theoretical 1-to-1 binding curve with a K_D of 30 mM. (D) PAGE analysis of in-line probing assays as described for B wherein reactions were supplemented with 200 mM of the ions indicated, except Rb⁺ was tested at 50 mM. Reduced band intensities at site 3 depicted here are indicative of ion binding. A lower contrast version of the full gel image and comparisons of sites 1, 2 and 3 are presented in Supplementary Fig. IV in the Supplement.

Figure 5

Variants of Li⁺-I riboswitches naturally respond to Na⁺. (A) Sequence and secondary structure model for a Na⁺-sensing riboswitch associated with a gene annotated as “hypothetical” (Supplementary Table I in the Supplement) from D. propionicus. Orange shading indicates alternative pairing that is predicted to form an intrinsic terminator stem. Predicted terminated (T) and full-length (FL) RNA transcripts are denoted with arrowheads. Other annotations are as described for Figs. 2 and 3. (B) Top: PAGE autoradiogram of an example single-round transcription termination assay series with a DNA template producing the D. propionicus construct depicted in A. Bands corresponding to FL and T transcripts identified in A are denoted. Transcription reactions were supplemented with the ions as indicated, and M identifies the lane loaded with a marker approximating FL RNA. Bottom: plot of the fraction of FL transcripts for each transcription reaction conducted in triplicate, where the circle colors matching the panels were derived from the gel shown. The dashed line represents the average fraction FL in the absence of additional ion supplementation above the 4 mM Na⁺ initially present in the transcription reaction. An uncropped version of the gel image is presented as Supplementary Fig. VIII in the Supplement. (C) Top: PAGE autoradiogram of an example transcription termination assay series using RNA construct M4, which carries a C54U mutation that represents the nucleotide at this position most commonly observed with nhaA-I motif RNAs. Bottom: Plot of the fraction of FL transcripts for each transcription reaction conducted in triplicate (n = 3), where the numbers indicate the mean and standard deviation. Additional annotations are as described for B. An uncropped version of the gel image is presented as Supplementary Fig. IX in the Supplement.