Irregular G-quadruplexes Found in the Untranslated Regions of Human mRNAs Influence Translation

Abstract

G-quadruplex structures are composed of coplanar guanines and are found in both DNA and RNA. They are formed by the stacking of two or more G-quartets that are linked together by three loops. The current belief is that RNA G-quadruplexes include loops of l to 7 nucleotides in length, although recent evidence indicates that the central loop (loop 2) can be longer if loops 1 and 3 are limited to a single nucleotide each. With the objective of broadening the definition of irregular RNA G-quadruplexes, a bioinformatic search was performed to find potential G-quadruplexes located in the untranslated regions of human mRNAs (i.e. in the 5' and 3'-UTRs) that contain either a long loop 1 or 3 of up to 40 nucleotides in length. RNA molecules including the potential sequences were then synthesized and examined in vitro by in-line probing for the formation of G-quadruplex structures. The sequences that adopted a G-quadruplex structure were cloned into a luciferase dual vector and examined for their ability to modulate translation in cellulo Some irregular G-quadruplexes were observed to either promote or repress translation regardless of the position or the size of the long loop they possessed. Even if the composition of a RNA G-quadruplex is not quite completely understood, the results presented in this report clearly demonstrate that what defines a RNA G-quadruplex is much broader than what we previously believed.

Keywords: G-quadruplex; RNA; RNA structure; gene expression; in-line probing; translation.

FIGURE 1.

Schematic representation of long loop 1 (A) and long loop 3 (B) RNA G-quadruplexes. The long loops are represented by the red lines. Beside each G4, the expression used for the python script is shown.

FIGURE 2.

In-line probing of the TEF candidate. A, autoradiogram of a 10% denaturing (8 m urea) polyacrylamide gel of the in-line probing of the 5′-labeled TEF WT and G/A-mutant PG4 versions performed in the presence of 100 mm of either LiCl or KCl. The lanes designated OH and T1 are an alkaline hydrolysis and an RNase T1 mapping of the WT version, respectively. The guanines thought to being involved in the G4 formation are indicated by the green boxes. To the right of the gel, the histogram of the K⁺/Li⁺ ratios of the band intensities for each nucleotide of both the TEF wild type and the G/A mutant is shown. The K⁺/Li⁺ ratios are shown in blue for the TEF WT and in red for the TEF G/A mutant. The green-boxed guanines represent the predicted G-tracks. The dotted line represents the 2-fold threshold that denotes a significant gain in flexibility. The sequence is indicated on the y axis. The red Gs on the y axis are those mutated to As in the G/A mutant version. Each bar represents the average of two independent experiments, and the error bars represent the S.D. B, the transcript of TEF used for the in-line probing. This candidate possesses a long loop 3 of 26 nucleotides. The G-tracks are denoted by the green boxes, and the guanines that are mutated to adenines in the G/A mutant version are in red.

FIGURE 3.

List of the PLL1-G4 (A) and PLL3-G4 (B) transcripts tested by in-line probing. The entire sequence of each transcript is shown along with their respective location on the mRNA. The length of each long loop is indicated over black brackets, and the stretches of guanines thought to be involved in the formation of the G4 are indicated by the green boxes. Other G-tracks present in the transcripts are identified by the gray boxes. The red guanines are those that changed to adenosines in the mutant version. The cG/cC scores are also shown (i.e. the value calculated for the transcript), and the result of the in-line probing is denoted by Yes or No. The candidates with a gray background are those retained for the in cellulo luciferase assays.

FIGURE 4.

Gene expression levels of the constructs containing the different PLL1-G4 and PLL3-G4 candidates (located in either the 5′ or the 3′-UTRs) fused to the luciferase gene at the protein (A) and the mRNA levels (B). The x axis identifies the different candidates used, and the y axis identifies the fold difference (i.e. WT result divided by G/A-mutated result). Error bars, mean ± S.D. The p values are p = 0.0116 (*), p = 0.0026 (**), p = 0.0002 (***), and p < 0.0001 (****). The experiments were repeated three times (on three different days), and each time, three technical replicates were performed.