The separation between the 5'-3' ends in long RNA molecules is short and nearly constant

Abstract

RNA molecules play different roles in coding, decoding and gene expression regulation. Such roles are often associated to the RNA secondary or tertiary structures. The folding dynamics lead to multiple secondary structures of long RNA molecules, since an RNA molecule might fold into multiple distinct native states. Despite an ensemble of different structures, it has been theoretically proposed that the separation between the 5' and 3' ends of long single-stranded RNA molecules (ssRNA) remains constant, independent of their base content and length. Here, we present the first experimental measurements of the end-to-end separation in long ssRNA molecules. To determine this separation, we use single molecule Fluorescence Resonance Energy Transfer of fluorescently end-labeled ssRNA molecules ranging from 500 to 5500 nucleotides in length, obtained from two viruses and a fungus. We found that the end-to-end separation is indeed short, within 5-9 nm. It is remarkable that the separation of the ends of all RNA molecules studied remains small and similar, despite the origin, length and differences in their secondary structure. This implies that the ssRNA molecules are 'effectively circularized' something that might be a general feature of RNAs, and could result in fine-tuning for translation and gene expression regulation.

Figure 1.

DNA calibration curve for the smFRET signals as a function of the separation of the fluorophores used as FRET pair. The error bars represent one standard deviation (±1σ). The solid line is the calibration curve fitted to that gives nm.

Figure 2.

FRET histograms of three RNA molecules. Top (bottom) panel corresponds to data taken in the presence of TM (TE) buffer solution. Black lines are Gaussian fits. To avoid cluttering the rest of the RNA smFRET histograms are shown in Supplementary Figures S2a and b.

Figure 3.

Inter-dye distances for end-labeled mRNA molecules of different lengths. Circles and triangles represent monocistronic fungal and viral mRNAs, respectively. Filled squares represent dicistronic mRNA molecules, whereas empty squares represent the antisense mRNA fgen1. Blue and red data are from TM and TE buffer solution smFRET experiments, respectively. Error bars correspond to ±1σ. The plot includes the linear fits (y = a + bx) with a = 6.8 ± 0.47 nm and b = 7.2 ± 2 × 10⁻⁴ nm/nt for TM buffer and a = 7.2 ± 0.5 nm and b = 7.6 ± 2 × 10⁻⁴ nm/nt for TE buffer with the 1σ band for each fit.

Figure 4.

Secondary structure of mRNA fgen1 predicted by mFOLD. The inset is a zoom of the exterior loop that contains the 5′ and 3′ ends. See text for discussion of the ranges marked in the inset.