Human box C/D snoRNAs with miRNA like functions: expanding the range of regulatory RNAs

Abstract

Small nucleolar RNAs (snoRNAs) and microRNAs are two classes of non-protein-coding RNAs with distinct functions in RNA modification or post-transcriptional gene silencing. In this study, we introduce novel insights to RNA-induced gene activity adjustments in human cells by identifying numerous snoRNA-derived molecules with miRNA-like function, including H/ACA box snoRNAs and C/D box snoRNAs. In particular, we demonstrate that several C/D box snoRNAs give rise to gene regulatory RNAs, named sno-miRNAs here. Our data are complementing the increasing number of studies in the field of small RNAs with regulatory functions. In massively deep sequencing of small RNA fractions we identified high copy numbers of sub-sequences from >30 snoRNAs with lengths of ≥18 nt. RNA secondary structure prediction indicated for a majority of candidates a location in predicted stem regions. Experimental analysis revealed efficient gene silencing for 11 box C/D sno-miRNAs, indicating cytoplasmic processing and recruitment to the RNA silencing machinery. Assays in four different human cell lines indicated variations in both the snoRNA levels and their processing to active sno-miRNAs. In addition we show that box D elements are predominantly flanking at least one of the sno-miRNA strands, while the box C element locates within the sequence of the sno-miRNA guide strand.

Figure 1.

Box C/D snoRNA derived sno-miRnas. Predicted secondary structures of the snoRNAs U48, U15a and U21 were calculated with RNAfold. Secondary structure sequences are given in dot-bracket notation with base pairings represented by two complementary parentheses ‘(’ and ‘)’ and non-pairing bases by dots ‘.’. A multiple sequence alignment of U48 derived sequences that were identified by 454 sequencing displays multiple copies of 23–25 nt of the 5′-region of U48, a minority of sequences was identified as 3′-fragments or full length U48. The positions of high frequency sequences from snoRNAs U15a, U48 and U21 are indicated with red bars, the corresponding low frequency strand positions are marked with blue bars. Box C and D elements are given in bold in the sequences.

Figure 2.

Structural and functional analysis of the box C/D snoRNA HBII-336 sno-miRNA. (A) Secondary structure prediction for HBII-336 was calculated using RNAfold. A dashed red line marks the guide-sno-miRNA strand, the dashed blue line indicates the positions of the passenger strands. (B) The dominant sno-miRNA sequence found in deep sequencing of Jurkat small RNAs is highlighted in red, the minor sequence is marked with a blue bar. The sno-miRNA is located in a very stably base paired stem that can possibly be recognized and processed by Dicer. ClustalW alignments of the guide and passenger HBII-336- sno-miRNAs show homogenous distributions of the sequences found in small RNA libraries. The guide sno-miRNA (red) shows length variability of 21- to 27-nt, while the length distribution for the minor species (or the passenger-strand, blue) is from 17- to 27-nt. (C) Functional investigation of the HBII-336 sno-miRNA in human HeLa cells by the dual luciferase assay shows similar silencing effects when compared to the previously described ACA45 sno-miRNA. Insertion of a scrambled target site sequence into the Renilla luciferase gene did not lead to significant luciferase reduction when compared to the unmanipulated vector. Data represent normalized outcomes of three individual experiments.

Figure 3.

Sno-miRNA analysis in human HeLa and Jurkat cells. Five individual snoRNAs were probed for expression and processing via northern blots with total RNA extracts from HeLa and Jurkat cells as shown in (B). Processed sno-miRNAs can be identified in both cell types. In Hela extracts sno-miRNAs from U3, U78 and HBI-43 can be seen, in Jurkat cell extracts the sno-miRNAs U74 and U78 were identified. The total sizes of full length snoRNAs are indicated. Reporter gene assays with complementary target sites for the sno-miRNAs indicate a correlation of snoRNA processing to gene silencing activity as displayed in (A). A randomized non-cognate target sequence served as negative control, the ACA45 sno-miRNA as a positive control. The empty psiCHECK-2 vector was used for normalization. Data represent the average outcome of four (HeLa, Jurkat) or three (RPMI8866, MCF7) individual experiments.

Figure 4.

Functional analysis of snoRNA generated sdRNAs in different human cell lines. Reporter gene assays were performed on human HeLa (light grey), Jurkat T (black), RPMI8866 B (white) and mammary carcinoma cell line MCF7 (dark grey). The snoRNA ACA45 served as positive control, non-cognate target and empty dual luciferase as negative control and for normalization. General gene silencing can be seen for ACA45, U3 and U3-4. HBII-336 and U83a did not induce any reduction of the Rluc gene in HeLa cells, but were very efficiently silencing the same target in both Jurkat and RPMI8866 cells. Note the differences of some sno-miRNA target sensors on the different cell types (U27, HBII-336, U83a, snR39b). Data for HeLa, Jurkat and RPMI8866 cells are from three individual experiments.

Figure 5.

Box H/ACA originating miRNAs in functional analysis. (A) The predicted secondary structures of three miRbase listed snoRNAs (miR-664/ACA36b, miR-1291/ACA34 and miR-1248/HBI-61, top right) share similar features with ACA45, shown top left. Two rather stable hairpins are connected via a flexible hinge loop, one of the hairpins is giving rise to a sno-miRNA (arrows). Sno-miRNA guide strand positions are marked with dashed red lines and locate to the 3′-stem of ACA45 and ACA36b, but to the 5′-stems of ACA34 and HBI-64. (B) Target sites for the sno-miRNA guide strands were checked in RPMI8866 (grey), Jurkat (black) and Hela (white) cells and compared to silencing activity of ACA45-sdRNA. The most efficient silencing on all cell types was observed for miR-664/ACA36b-sdRNA, miR-1291/ACA34 and miR-1248/HBI-61 displayed only mild silencing, in particular in Jurkat cells. Values represent normalized data from two individual experiments.