Observation of preQ1-II riboswitch dynamics using single-molecule FRET

Abstract

PreQ₁ riboswitches regulate the synthesis of the hypermodified tRNA base queuosine by sensing the pyrrolopyrimidine metabolite preQ₁. Here, we use single-molecule FRET to interrogate the structural dynamics of apo and preQ₁-bound states of the preQ₁-II riboswitch from Lactobacillus rhamnosus. We find that the apo-form of the riboswitch spontaneously samples multiple conformations. Magnesium ions and preQ₁ stabilize conformations that sequester the ribosome-binding site of the mRNA within the pseudoknotted structure, thus inhibiting translation initiation. Our results reveal that folding of the preQ₁-II riboswitch is complex and provide evidence favoring a conformational selection model of effector binding by riboswitches of this class.

Keywords: RNA dynamics; Riboswitch; conformation selection; smFRET.

Figure 1.

Queuosine biosynthesis, the preQ₁-II riboswitch structure and experimental construct for smFRET. (a) The queuosine (Q) base is a hypermodified base that is widespread in prokaryotes and eukaryotes. However, Q is only synthesized de novo in bacteria. Q production starts with GTP and proceeds by several enzymatic steps (arrow) to the last free intermediate preQ₁ (7-aminomethyl-7-deazaguanine), which is then inserted into specific tRNAs followed by additional in situ modifications [16]. (b) Ribbon diagram depicting the global HL_out pseudoknot fold of the preQ₁-II riboswitch from Lactobacillus rhamnosus (Lrh) (PDB entry 4jf2) [13]. The riboswitch comprises three co-axially stacked pairing regions (P1, P2 and P3) with a fourth (P4) flanking the P2-P3 interface. The three-helix junction binds preQ₁ to complete P2-P3 coaxial helical stacking. Two nearby Mg²⁺ ions are shown as purple spheres. The co-crystal structure reveals that the pseudoknot buries the entire Shine-Dalgarno sequence (SDS). The SDS nucleotides are filled (yellow) and preQ₁ is drawn as a CPK model (green). Pairing (P) and (J) junction regions are shown in distinct colors. (c) Secondary structure of the wild-type Lrh preQ₁-II riboswitch construct used for this investigation based on the known crystal structure in b. The P1 helix is dispensable for effector sensing and was modified to hybridize with a biotinylated DNA oligonucleotide (inset box). The 3´-end comprises the wild-type sequence of the naturally occurring, downstream queT gene that is regulated by the riboswitch. The start codon (AUG) is highlighted. FRET (Dy547) and acceptor (Cy3) labels are depicted as green and red stars.

Figure 2.

PreQ₁ stabilizes preQ₁-II riboswitch conformations that sequester the SD sequence into intramolecular core structure. Histograms compiled from hundreds of smFRET traces show the distribution of the FRET values for the riboswitch imaged in the absence of Mg²⁺. PreQ₁ concentration was varied as follows: (a) no preQ₁, (b) 0.1 μM, (c) 0.3 μM, (d) 0.5 μM, (e) 1.0 μM and (f) 10 μM. Yellow lines represent individual Gaussian fits; black lines indicate the sum of Gaussians. The fraction of the riboswitch in each FRET state derived from Gaussian fits is shown next to respective individual Gaussian peaks. N equals the number of single-molecule traces compiled.

Figure 3.

Apo form of the preQ₁-II riboswitch imaged in the absence of Mg²⁺ and preQ₁ spontaneously fluctuates between low (~ 0.35) and high (~ 0.7) FRET states. (a) Representative smFRET trace showing fluctuations between the 0.35 and 0.7 FRET states. Observed intensities of donor and acceptor fluorescence and the calculated apparent FRET efficiency are shown in blue, red, and black, respectively. The Hidden Markov Model fit is shown in magenta. (b) Transition density plot (TDP) analysis of 1,277 fluctuations between different FRET states in 126 HMM-idealized FRET traces. The frequency of transitions from the starting FRET value (x axis) to the ending FRET value (y axis) is represented by a heat map. The range of FRET efficiencies from 0 to 1 was separated into 200 bins. The resulting heat map was normalized to the most populated bin in the plot; the lower- and upper-bound thresholds were set to 10% and 100% of the most populated bin.

Figure 4.

The Lrh preQ₁-II riboswitch senses the preQ₁ concentration to yield a compact fold. The fraction of the riboswitch in a compact conformation, determined by the area under the 0.7 FRET peak in Fig (2a-f), was plotted as a function of preQ₁ concentration and fitted to a hyperbola (red line; y = A*[preQ₁]/(EC₅₀+[preQ₁])+ B, where EC₅₀ is the half maximal effective concentration of preQ₁ and B is the fraction of 0.7 FRET observed in the absence of preQ₁).

Figure 5.

Mg²⁺ stabilizes Lrh preQ₁-II riboswitch conformations that sequester the SD sequence into intramolecular core structure. Histograms compiled from hundreds of smFRET traces show the distribution of the FRET values for the riboswitch imaged in the presence of 0.5 mM Mg²⁺ (a-b) or 6 mM Mg²⁺(c-d). Measurements were taken either in the absence (a,c) or the presence of 1 μM preQ₁ (b,d). Yellow lines represent individual Gaussian fits; black lines indicate the sum of Gaussians. The fraction of the riboswitch in each FRET state derived from Gaussian fits is shown next to respective individual Gaussian peaks. N equals the number of single-molecule traces compiled.