Identification of new high affinity targets for Roquin based on structural conservation

Abstract

Post-transcriptional gene regulation controls the amount of protein produced from a specific mRNA by altering both its decay and translation rates. Such regulation is primarily achieved by the interaction of trans-acting factors with cis-regulatory elements in the untranslated regions (UTRs) of mRNAs. These interactions are guided either by sequence- or structure-based recognition. Similar to sequence conservation, the evolutionary conservation of a UTR's structure thus reflects its functional importance. We used such structural conservation to identify previously unknown cis-regulatory elements. Using the RNA folding program Dynalign, we scanned all UTRs of humans and mice for conserved structures. Characterizing a subset of putative conserved structures revealed a binding site of the RNA-binding protein Roquin. Detailed functional characterization in vivo enabled us to redefine the binding preferences of Roquin and identify new target genes. Many of these new targets are unrelated to the established role of Roquin in inflammation and immune responses and thus highlight additional, unstudied cellular functions of this important repressor. Moreover, the expression of several Roquin targets is highly cell-type-specific. In consequence, these targets are difficult to detect using methods dependent on mRNA abundance, yet easily detectable with our unbiased strategy.

Figure 1.

The 100 nt long structurally conserved region in the 3′UTR of UCP3 codes for a repressive element. (A) Overview of luciferase reporter constructs. Different fragments of the UCP3 3′UTR were fused to firefly luciferase. (B) Luciferase activity of UCP3 3′UTR fusion constructs shown in (A). Firefly luciferase activity was normalized to Renilla luciferase as internal transfection control. Values are normalized to an empty vector control, without UCP3 3′UTR sequences. n = 3. (C) Secondary structure prediction by Dynalign of the structurally conserved regions in the human and mouse UCP3 3′UTRs. (E) Overview of truncations of the 100 nt long window. (D) Luciferase activity of UCP3 truncation constructs shown in (E). Firefly luciferase activity was normalized to Renilla luciferase as internal transfection control. Values are normalized to an empty vector control, without UCP3 3′UTR sequences. Luciferase activity of the 100 nt window is indicated as dashed line. n = 3. (**) P-value < 0.01.

Figure 2.

The UCP3 wt element folds into two hairpins and reduces mRNA half-life by interaction with Roquin. (A) Predicted lowest free energy secondary structure of the UCP3 wt element by RNAstructure. Nucleotides detected by in-line probing [shown in (B)] are circled. (B) In-line probing analysis of UCP3 wt RNA. The RNA was loaded directly (NR, no reaction), subjected to cleavage by RNase T1 or alkaline hydrolysis (_¯OH), or incubated for 40 h at room temperature and pH 8.3 (in-line) prior to Urea PAGE. Paired regions are indicated by identically colored lines. (C) Luciferase activity of UCP3 mutants for the identification of motifs essential for gene regulation. Adenine was mutated to cytosine, guanine to uracil and vice versa. Numbers indicate mutated nucleotide positions. Firefly luciferase activity was normalized to Renilla luciferase as internal transfection control. Values are normalized to an empty vector control, without UCP3 3′UTR sequences. Luciferase activity of the UCP3 wt element is indicated as dashed line. n = 3. (D) GFP fluorescence of UCP3 wt and double mutant (MUTI/II). GFP-UCP3-fusion constructs were stably integrated into the genome of HeLa cells. GFP fluorescence was measured by flow cytometry. (E) Half-life of GFP mRNAs containing the UCP3 wt element (wt) or double mutant (MUTI/II). HeLa cells stably expressing one of the two constructs were treated with 5 μg/μl actinomycin D (ActD). Thereafter, total RNA was isolated at 2 h intervals and GFP mRNA levels quantified by RT-qPCR. GFP values are normalized to the housekeeping gene RPLP0. n = 3. (F) Overview of UCP3 constructs used for RNA affinity purification. (G) Analysis of Roquin binding to the UCP3 constructs shown in (F). For RNA affinity purification HEK293 whole cell lysates were incubated with the different UCP3 RNAs. Roquin-1 and Roquin-2 were visualized by western blot using anti-Roquin antibody. n = 2. (H) Western blot of Roquin-1 and Roquin-2 after siRNA-mediated knockdown. Anti-Roquin was used to verify the respective knockdown. Total lane protein is shown as loading control. n = 3. (I) Luciferase activity of the UCP3 wt element after siRNA-mediated knockdown of Roquin proteins. Firefly luciferase activity was normalized to Renilla luciferase as internal transfection control. Values are normalized to an empty vector control, without UCP3 3′UTR sequences. n = 3. (**) P-value < 0.01. (*) P-value < 0.05.

Figure 3.

Both UCP3 CDEs are required for efficient Roquin binding. (A) Domain organization of mouse Roquin-1 and overview of Roquin fragments used for binding experiments. (B) Overview of UCP3 constructs used for binding experiments. (C) Luciferase activity of UCP3 constructs shown in (B). Firefly luciferase activity was normalized to Renilla luciferase as internal transfection control. Values are normalized to an empty vector control, without UCP3 3′UTR sequences. n = 3. (D and F) Binding of recombinant Roquin-1 to the UCP3 constructs shown in (B). Radiolabeled RNAs were incubated with increasing amounts of Roquin-1 ROQ domain (D) or N-terminus (F). The apparent dissociation constant (K_D) was calculated from two to three independent experiments. (E and G) Representative quantification of EMSA experiments with Roquin-1 ROQ domain (E) or N-terminus (G). (**) P-value < 0.01. (*) P-value < 0.05. n.d. = not determined.

Figure 4.

Mutational analysis of the UCP3 tandem CDE. (A) Overview of UCP3 mutants. (B) Luciferase activity of closing base pair mutants. n = 3. (C) Luciferase activity of stem mutants. n = 3. (D) Luciferase activity of triloop mutants. n = 3. (B–D) Firefly luciferase activity was normalized to Renilla luciferase as internal transfection control. Values are normalized to an empty vector control, without UCP3 3′UTR sequences. Luciferase activity of the UCP3 wt element is indicated as dashed line. (**) P-value < 0.01. (*) P-value < 0.05. n.s. = not significant.

Figure 5.

Identification of new Roquin targets. (A) Consensus used for bioinformatic prediction of new CDE structures by RNAMotif. Nucleotide occurrences in the triloop are shown for all and conserved elements. V = A, G, C. P-value < 0.0001 (Pearson’s chi-square test). (B) Co-occurrence of A or G at position 2 in the triloop with R-Y or Y-R closing base pairs. R = A, G and Y = C, U. P-value < 0.0005 (Pearson’s chi-square test). (C) Luciferase activity of different CDE-encoding 3′UTR regions and mutants. n = 3. CDEs and mutations are shown in Supplementary Figure S13. Firefly luciferase activity was normalized to Renilla luciferase as internal transfection control. Values are normalized to an empty vector control, without UCP3 3′UTR sequences. Influence of predicted CDE-like elements on luciferase activity in HEK293 cells. n = 3. (D) Western blot of Roquin-1 and Roquin-2 after siRNA-mediated knockdown in HEK293 cells and HUVECs. Anti-Roquin was used to verify the respective knockdown. Total lane protein is shown as loading control. n = 3. (E) RT-qPCR quantification of new Roquin target genes encoding CDEs after siRNA-mediated knockdown of Roquin-1 and Roquin-2 in HEK293 cells and HUVECs. Values are normalized to the housekeeping gene RPLP0. n = 4. (F) Consensus used for bioinformatic prediction of new ADE structures by RNAMotif. (G) RT-qPCR quantification of new Roquin target genes encoding ADEs after siRNA-mediated knockdown of Roquin-1 and Roquin-2 in HEK293 cells and HUVECs. Values are normalized to the housekeeping gene RPLP0. n = 4. (**) P-value < 0.01. (*) P-value < 0.05. n.d. = not detected.