Prediction of Sinorhizobium meliloti sRNA genes and experimental detection in strain 2011

Abstract

Background: Small non-coding RNAs (sRNAs) have emerged as ubiquitous regulatory elements in bacteria and other life domains. However, few sRNAs have been identified outside several well-studied species of gamma-proteobacteria and thus relatively little is known about the role of RNA-mediated regulation in most other bacterial genera. Here we have conducted a computational prediction of putative sRNA genes in intergenic regions (IgRs) of the symbiotic alpha-proteobacterium S. meliloti 1021 and experimentally confirmed the expression of dozens of these candidate loci in the closely related strain S. meliloti 2011.

Results: Our first sRNA candidate compilation was based mainly on the output of the sRNAPredictHT algorithm. A thorough manual sequence analysis of the curated list rendered an initial set of 18 IgRs of interest, from which 14 candidates were detected in strain 2011 by Northern blot and/or microarray analysis. Interestingly, the intracellular transcript levels varied in response to various stress conditions. We developed an alternative computational method to more sensitively predict sRNA-encoding genes and score these predicted genes based on several features to allow identification of the strongest candidates. With this novel strategy, we predicted 60 chromosomal independent transcriptional units that, according to our annotation, represent strong candidates for sRNA-encoding genes, including most of the sRNAs experimentally verified in this work and in two other contemporary studies. Additionally, we predicted numerous candidate sRNA genes encoded in megaplasmids pSymA and pSymB. A significant proportion of the chromosomal- and megaplasmid-borne putative sRNA genes were validated by microarray analysis in strain 2011.

Conclusion: Our data extend the number of experimentally detected S. meliloti sRNAs and significantly expand the list of putative sRNA-encoding IgRs in this and closely related alpha-proteobacteria. In addition, we have developed a computational method that proved useful to predict sRNA-encoding genes in S. meliloti. We anticipate that this predictive approach can be flexibly implemented in many other bacterial species.

Figure 1

Northern blot analysis of putative sRNAs encoded in the chromosome of S. meliloti strain 2011. Total RNA was isolated from S. meliloti 2011 cells grown at 28°C with agitation (120 rpm) in RDM minimal medium and harvested at OD₆₀₀ = 0.5 (Exp) or at OD₆₀₀ = 3.9 (Stat). Total RNA was also isolated from cells subjected to high salt stress (NaCl; 0.3 M NaCl in RDM, OD₆₀₀ = 0.55), membrane stress (EtOH; RDM with 2% v/v ethanol, OD₆₀₀ = 1.6; or SDS; RDM with 0.1% w/v SDS; OD₆₀₀ = 1.0), phosphate starvation (-P; RDM with 0.1 mM phosphate and 10 mM MOPS pH 7.0, OD₆₀₀ = 1.0), oxidative stress (H₂O₂; RDM with 0.1 mM H₂O₂, OD₆₀₀ = 1.1), heat stress (37°; RDM grown at 37°C, OD₆₀₀ = 0.95) and acid stress (pH 5.5; treatment of exponential phase cells at OD₆₀₀ = 0.5 during 90 min at pH 5.5 before harvest). Northern hybridizations were done with PCR-generated digoxigenin-labeled dsDNA probes directed against the entire IgR or an internal fragment (see Figure 2 and Additional file 2 for further details). RNA molecular weight markers (with their sizes indicated in nt with small arrows at the left of each panel) were run with each set of RNA samples for estimation of transcript size. As exposure times were optimized for visualization here, the signal intensity does not indicate relative abundance of detected transcripts between different IgRs. Hybridization signals were quantified with ImageJ software, normalized to the amount of 5S RNA, 4.5S RNA and tRNA bands detected in silver stained gels present in each sample (bottom panel) and plotted in a bar graph shown below each Northern blot. The band intensity units are relative to the normalized amount present in Exp cells, which were set as 1.0.

Figure 2

Organization of novel S. meliloti 1021 chromosomal loci encoding putative sRNAs with detected counterparts in S. meliloti strain 2011. The IgRs encompassing novel putative sRNAs from our first compilation (Table 1) are drawn to scale in the portion between breaks. The chromosomal coordinates of predicted promoters and Rho-independent terminators are indicated next to the corresponding symbols. ORFs flanking each IgR are designated with their annotated codes or gene names. sRNAs are designated according to their position in the output of global scoring for the corresponding IgR (Additional file 17) or to the corresponding IgR from Table 1. Small empty arrowheads indicate the approximate position of the chromosomal target sequences for oligonucleotides used to generate probes for Northern blot. Wavy arrowheads denote the location and orientation of oligonucleotides present in Sm14kOLI microarray that detected the putative sRNA transcripts (Table 1).

Figure 3

Improvement of the predictive strategy of chromosomal S. meliloti sRNAs. From the initial list of 2920 chromosomal IgRs, we retained 778 IgRs longer than 150 bp than did not contain annotated or non-annotated repeats. Next, we introduced a global scoring criterion for each IgR to assign a numerical score taking into account the presence of putative independent transcriptional units or transcriptional signals and their relative distance to flanking ORFs, sequence conservation (BlastN analysis) and secondary structure conservation (QRNA analysis). See text for further details.

Figure 4

Summary of the scoring criteria introduced to weigh the relative position of predicted transcriptional signals in IgRs. An IgR with a co-oriented putative promoter and a terminator separated from by 40–500 bp each other was scored 10. Every promoter-terminator pair matching the previous criterion within a single IgR was rated individually and summed to calculate the global score of that IgR. Orphan promoters were scored 3. Putative promoters were rated 2 if the 5'-end of the co-oriented flanking CDS was farther than 300 bp, 1 if this distance was 200–300 bp and 0, if they were closer than 200 bp. Orphan terminators were scored 3. Predicted terminators co-oriented with flanking ORFs were scored according to their relative distance to the 3'-end of the corresponding annotated gene, so that a score of 2 was assigned if the terminator was farther than 200 bp, 1 if the distance was 100–200 nt, and 0 if it was closer than 100 bp.