Hidden Structural Modules in a Cooperative RNA Folding Transition. Gracia et al
Description
Quantitative single nucleotide resolution RNA footprinting methods probe structural accessibility at each nucleotide in the RNA structure, i.e. high reactivity indicates unpaired/flexible positions and low reactivity base-paired/rigid positions. By steadily increasing the concentration of free Mg2+, we induce RNA folding and observe the structural transitions. We quantify the Mg2+-dependent changes in reactivity by fitting the observed transitions to a sigmoidal binding equation. We extract the transition midpoint parameter from the fit allowing for determination of the Mg2+ requirement for folding at a given nucleotide (i.e. the concentration required to reach half maximal saturation). The spreadsheets contain nucleotide reactivity's starting and ending with the 5' and 3' hairpin loop, respectively, (GAGUA) of the cassette. The conclusion from the data for WT P5abc was that there are multiple Mg2+-dependent transitions involving structural modules. To test for cooperativity between two of these modules, P5c and the metal core, we performed footprinting experiments using several mutants that either blocked the transition in P5c (U167C or U167C/U177C) or blocked folding of the metal core (A186U or ArichU). A lack of signal change at the mutated region (for example, U168 for mutant U167C) indicates that the module of interest is blocked by the mutation. We observed Mg2+-dependent changes in the unmutated region similar to WT, but the Mg2+ was greater than WT, consistent with cooperativity between the structural modules. We designed mutations that preformed the P5c module (Nat+3) and found that the overall Mg2+ requirement for folding the structure was significantly reduced compared to WT (e.g. nucleotides U135, A139, G163, G164, A186). We also partially restored folding of the blocked mutants by inserting restoration nucleotides that replace the blocked region (U167C_U177C_G_GUins). This restoration mutant decreased the Mg2+ requirement and restored signals associated with folding to WT levels compared with the U167C blocking mutant (e.g. nucleotides U135, A139, G163, GinsG164, GinsG176, A186). We utilized mutations that block the long-range interactions that form between the folded P5c and metal core modules (U168C) and observed that folding of the modules was not perturbed compared to WT (e.g. U135, A139, G163, A186). At concentrations above 10 mM Mg2+, we observed that some positions were perturbed for folding compared to WT (e.g. A184 SHAPE remains moderately reactive in U168C at high Mg2+ concentrations). We used mutations that preform the P5c module (C165A_G175U) to observe the formation of long range interactions between the folded modules. We found that signals associated with the P5c packing interaction (e.g. A173 DMS) were destabilized when the metal core module was blocked (C165A_G175U_A186U) compared to WT, consistent with cooperativity between the structural modules during formation of long-range interactions.
Files
Steps to reproduce
To investigate a complex Mg2+-dependent RNA folding process quantitatively, the P5abc RNA sequence was incorporated into a cassette containing flanking hairpins linked by unpaired nucleotides. RNAs were then generated by in vitro transcription. Each RNA molecule was prefolded at various concentrations of Mg2+ and then modified using chemicals (SHAPE or DMS) under conditions that result in one or fewer modifications per molecule (see METHODS). Modifications cause reverse transcription stops at the modified position resulting in a pool of cDNA products with lengths that correspond to nucleotide positions in the RNA structure. The cDNA primers are fluorescently labeled so that they can be resolved by capillary electrophoresis in 96-well format. Electrophoretic traces at each Mg2+ concentration are aligned, quantified, and converted to a pseudo-gel image using MATLAB and the HiTRACE analysis package (Yoon et. al., 2011 and Kladwang et. al., 2014). Each trace (time vs. fluorescence plots) contains intensity peaks that correspond to the cDNA fragment pool at a given Mg2+ concentration. The area under each peak is quantified by fitting each raw peak to a sum of Gaussian distributions that minimizes the deviation between the observed peak and fit. Each peak is then integrated to calculate the area under the curve for each fit. The deviation between the observed peak and fit for each peak is propagated through the analysis process so that a peak error matrix is also generated for each data point. To correct for signal saturation at the full length cDNA product (which dominates the population), samples were resolved by capillary electrophoresis twice at different dilutions. The true size of these saturated peaks is determined by rescaling the diluted peaks by the observed dilution factor of the two traces. Each peak is then converted to a probability by comparing each cDNA product peak to the sum of all peaks that appear after the given peak (longer cDNAs). This process corrects for signal attenuation that may occur for longer cDNA products by converting all peaks into a probability that does not depend on cDNA peaks that occur before a given position. Reactions that are modified with the chemical reagents are then background subtracted from reactions that are carried out in parallel but with no modification reagent. Last, to normalize the data relative to a known RNA structural motif, the reactivity calculated at each position after background subtraction is divided by the reactivity of a set of unpaired nucleotides in the linker region contained within all RNA cassettes (A201 and A202). This region remains highly reactive across all concentrations of Mg2+ tested, and this structural region is contained within the cassettes of all mutants tested in our investigations. In this context, a quantified relative reactivity of ‘1’ has structural significance because it is equivalent to the reactivity of the unpaired linker at a given concentration of Mg2+.