The discovery of novel noncoding RNAs in 50 bacterial genomes

Abstract

Structured noncoding RNAs (ncRNAs) contribute to many important cellular processes involving chemical catalysis, molecular recognition and gene regulation. Few ncRNA classes are broadly distributed among organisms from all three domains of life, but the list of rarer classes that exhibit surprisingly diverse functions is growing. We previously developed a computational pipeline that enables the near-comprehensive identification of structured ncRNAs expressed from individual bacterial genomes. The regions between protein coding genes are first sorted based on length and the fraction of guanosine and cytidine nucleotides. Long, GC-rich intergenic regions are then examined for sequence and structural similarity to other bacterial genomes. Herein, we describe the implementation of this pipeline on 50 bacterial genomes from varied phyla. More than 4700 candidate intergenic regions with the desired characteristics were identified, which yielded 44 novel riboswitch candidates and numerous other putative ncRNA motifs. Although experimental validation studies have yet to be conducted, this rate of riboswitch candidate discovery is consistent with predictions that many hundreds of novel riboswitch classes remain to be discovered among the bacterial species whose genomes have already been sequenced. Thus, many thousands of additional novel ncRNA classes likely remain to be discovered in the bacterial domain of life.

Graphical Abstract

Figure 1.

Consensus sequence and structural models of predicted strong riboswitch candidates. (A) nqrA-II motif. (B) dauA motif. (C) livJ motif. (D) trpB motif. (E) aspC motif. (F) opuB motif. (G) manA-II motif. (H) mgtE motif. (I) aceE-II motif. (J) idnO motif. Annotations for all motifs are as defined in (A). Shine-Dalgarno (SD) sequence regions are speculative. Sequence alignments for all motifs are provided as supplementary sto files.

Figure 2.

Consensus sequence and structural models of several weak riboswitch candidates. (A) rdhA motif. (B) pntA motif. (C) yjbJ motif. (D) fhsA motif. (E) ykoW motif. (F) purS motif. (G) yjbM motif. (H) nifV motif. (I) gcsH motif. Annotations for all motifs are as defined in Figure 1A. Shine-Dalgarno (SD) sequence regions are speculative.

Figure 3.

Consensus sequence and structural models of selected uORF candidates. (A) ilvB uORFC. (B) fthC uORFC. Annotations for all motifs are as defined in Figure 1A.

Figure 4.

Consensus sequence and structural models of other motif types. (A) asfR motif. Repeat sequences are indicated by gray lines. (B) rpsJ motif. (C) dnaK motif. (D) MRSC-65–2 motif. Annotations for all motifs are as defined in Figure 1A.

Figure 5.

Consensus sequence and structural model the smtA motif. Annotations are as defined in Figure 1A. Gray lines indicate zero-length connectors.