Mining small RNA sequencing data: a new approach to identify small nucleolar RNAs in Arabidopsis

Abstract

Small nucleolar RNAs (snoRNAs) are noncoding RNAs that direct 2'-O-methylation or pseudouridylation on ribosomal RNAs or spliceosomal small nuclear RNAs. These modifications are needed to modulate the activity of ribosomes and spliceosomes. A comprehensive repertoire of snoRNAs is needed to expand the knowledge of these modifications. The sequences corresponding to snoRNAs in 18-26-nt small RNA sequencing data have been rarely explored and remain as a hidden treasure for snoRNA annotation. Here, we showed the enrichment of small RNAs at Arabidopsis snoRNA termini and developed a computational approach to identify snoRNAs on the basis of this characteristic. The approach successfully uncovered the full-length sequences of 144 known Arabidopsis snoRNA genes, including some snoRNAs with improved 5'- or 3'-end annotation. In addition, we identified 27 and 17 candidates for novel box C/D and box H/ACA snoRNAs, respectively. Northern blot analysis and sequencing data from parallel analysis of RNA ends confirmed the expression and the termini of the newly predicted snoRNAs. Our study especially expanded on the current knowledge of box H/ACA snoRNAs and snoRNA species targeting snRNAs. In this study, we demonstrated that the use of small RNA sequencing data can increase the complexity and the accuracy of snoRNA annotation.

Figure 1.

Enrichment of small RNAs at both termini of snoRNAs. (A) Definition of 5′-, body and 3′-end regions of a snoRNA. (B) The proportion of the sequence length, distinct small RNAs and total reads of small RNAs mapped to each region. (C) The distribution of small RNAs on a known box C/D snoRNA, U30. (D) The distribution of small RNAs on a known box H/ACA snoRNA, snoR74-2 with an incomplete 5′ end. Capital letters indicate the snoRNA sequences reported previously and lowercase letters indicate the flanking sequences. Short sequences are small RNAs followed by their read numbers. Conserved motifs are highlighted.

Figure 2.

Structures of novel box H/ACA snoRNAs predicted by small RNA sequencing data. The program of mfold was used to predict structures of newly identified snoR135, snoR139, snoR140 and snoR142 as described in ‘Materials and Methods’ section.

Figure 3.

Validation of the predicted snoRNA 5′ ends by PARE data. The positions of predicted 5′ ends of 151 box C/D snoRNAs and 37 box H/ACA snoRNAs were set as 0. Positions upstream of the predicted 5′ ends were set as negative values and downstream as positive values. The read number of 5′ ends starting at each position (–10 to 10) was normalized to the total PARE read number of the 21-nt region. A gradient of red color was used to represent the frequency of 5′ ends starting at each position.

Figure 4.

Northern blot analysis of novel snoRNAs predicted by small RNA sequencing data. (A) Aliquots of 10 μg total RNA for each lane were separated on 15% TBU gel for the northern blot analysis of eight new box C/D snoRNAs. snoR128: 92 nt; U46-1: 79 nt; snoR123b: 79 nt; snoR119: 75 nt; snoR127: 138 nt; snoR129: 88 nt; snoR125: 200 nt; U27-2: 84 nt. (B) Aliquots of 10 μg total RNA for each lane were separated on 6% TBU gels for the northern blot analyses of six new box H/ACA snoRNAs. snoR141: 151 nt; snoR140: 142 nt; snoR143: 171 nt; snoR145: 145 nt; snoR135: 141 nt; snoR139: 150 nt; Spliceosomal snRNA U6 was used as the loading control. S, 10-day-old seedlings; L, rosette leaves; F, flowers.