A linear and circular dual-conformation noncoding RNA involved in oxidative stress tolerance in Bacillus altitudinis

Abstract

Circular RNAs have been extensively studied in eukaryotes, but their presence and/or biological functionality in bacteria are unclear. Here, we show that a regulatory noncoding RNA (DucS) exists in both linear and circular conformation in Bacillus altitudinis. The linear forms promote B. altitudinis tolerance to H₂O₂ stress, partly through increased translation of a stress-responsive gene, htrA. The 3' end sequences of the linear forms are crucial for RNA circularization, and formation of circular forms can decrease the levels of the regulatory linear cognates. Bioinformatic analysis of available RNA-seq datasets from 30 bacterial species revealed multiple circular RNA candidates, distinct from DucS, for all the examined species. Experiments testing for the presence of selected circular RNA candidates in four species successfully validated 7 out of 9 candidates from B. altitudinis and 4 out of 5 candidates from Bacillus paralicheniformis; However, none of the candidates tested for Bacillus subtilis and Escherichia coli were detected. Our work identifies a dual-conformation regulatory RNA in B. altitutidinis, and indicates that circular RNAs exist in diverse bacteria. However, circularization of specific RNAs does not seem to be conserved across species, and the circularization mechanisms and biological functionality of the circular forms remain unclear.

Fig. 1. DucS produces multiple transcripts.

a, b Northern blotting to detect DucS expression in B. altitudinis SCU11 (wild-type, WT) and its derivative ΔducS carrying either the empty vector (pEV) or a plasmid expressing DucS RNA (pDucS). Probe DucS (200 nt) corresponds to almost the full length of DucS. Total RNA was extracted from the indicated strains from cultures at 8 h (a) or 4, 8, 12, and 24 h (b). c Genomic context of DucS. The sequence of DucS (shown in red) and its 60 nt upstream and 68 nt downstream regions are shown below. The black boxes mark the RNA circularization junctions identified in the subsequent experiments. Four oligonucleotide probes (oligCK and olig1, 2, 3) are underlined. The transcriptional start site (TSS) and two transcriptional termination sites (TTS_1st and TTS_2nd) (arrows) of DucS were determined by cRACE assay, whose sequencing visualization is shown in (d). e Northern blotting to clarify the four transcripts of DucS by oligonucleotide probes. For northern blotting performed in this study, unless otherwise indicated, 5 µg RNA per lane was loaded for strains harboring chromosomal DucS, and 2 µg RNA per lane was loaded for strains carrying plasmid-borne DucS. EB-stained 16S rRNA was used as a loading control. Data are representative of at least two independent experiments (a, b, e). Source data are provided as a Source data file.

Fig. 2. Experimental validation of DucS circular RNAs.

a The working principal diagram for the identification of circular RNA products in RNA-seq data. Read mapping to the reference DNA in a permuted, chiastic order is a hallmark of circular RNA. b Percentage of the reads supporting different circularization junctions compared to the total reads aligned to the chromosomal DucS region. The locations of the S1 and S2 junctions are shown in Fig. 1c. Other junctions represent those 5′ termini that are different from S1 and S2. c RT‒PCR verification of circular RNAs of DucS. Left: Schematic representation of primers for circular RNA verification. Black and blue arrows are convergent and divergent primer pairs, respectively. Right: RT‒PCR results. cDNA templates were obtained by reverse transcription of RNA (WT, 8 h) pretreated with or without RNase R (R⁺ or R⁻). Chromosomal DNA and linear RNA (yaaA mRNA) were used as a negative control. EB-stained rRNA acted as RNase R digestion controls. Similar results were obtained in two biologically independent experiments. d Sequencing the RT‒PCR products from divergent primers. A partial sequence of DucS is shown above. N₁₄₂ and N₃₂ represent 142 and 32 nucleotides not shown in the sequence. Black arrows indicate different 5′ end nucleotides of circularization junctions. e Circular RNAs of DucS are more resistant to RNase R treatment and can be detected by junction-spanning probes. Probes circS1 and circS2 correspond to S1 and S2 junctions, respectively. Similar results were obtained in two biologically independent experiments. f Schematic diagram of the production of DucS transcripts and their new designations. L1 and L2 are linear RNAs. C1 and C2 are circular RNAs. Source data are provided as a Source data file.

Fig. 3. Identification of critical sequences for DucS circularization.

a Schematic diagram of DucS variants with the indicated mutations. The solid red arrow represents DucS, and its 5′- and 3′-end sequences are shown above and below the arrow. Deletion mutations at the 5′-end of DucS (upper) and deletion or point mutations at the 3′-end (lower) were constructed on plasmid-borne DucS. The black horizontal dashed line represents deleted nucleotides. b–e Northern blotting to evaluate the effects of different mutations on the formation of DucS circular RNAs. Total RNA was extracted from the indicated strains from cultures at 8 h (b, c, e) or 4, 8, 12, and 24 h (d). Some samples (c) were treated with RNase R by using rRNA as digestion controls. Probe DucS was used for Northern blotting. EB-stained 16 S rRNA was used as a loading control. Data are representative of at least two independent experiments (b–e). Source data are provided as a Source data file.

Fig. 4. Linear RNAs of DucS promote target htrA translation and bacterial oxidative tolerance, while circular RNAs affect the level of linear cognates.

a The DucS-htrA complex (arrow) formed in a htrA concentration-dependent manner as determined by EMSA (n = 2). b Colony fluorescence of ΔducS strains expressing pHtrA::sfGFP fusion with or without the co-expression of DucS (n = 3). Scale bars = 1 mm. Quantitative analysis of fluorescence intensity by ImageJ. c Western blotting revealed that DucS increases HtrA::3×FLAG fusion protein level. Coomassie blue staining of comparable intracellular proteins acted as loading controls (lower) (n = 3). Quantitative analysis of band intensity by ImageJ. d Effect of circular RNA deficiency (DucSM_C) on the regulation of HtrA::sfGFP fusion. Fluorescence was measured in liquid cultures (8 h) of ΔducS_pHtrA::sfGFP co-expressing DucS, DucSM_C or not (Ctr). Normalized fluorescence (Fluorescence Intensity/OD₆₀₀) was used for comparison (n = 3). e Effects of DucS and htrA on bacterial growth under H₂O₂ stress for 4 h (n = 3). f Stability of DucS linear RNAs was evaluated by rifampicin treatment. EB-stained 16S rRNA was used as a loading control. The half-life (t_1/2) of L1 and L2 was determined relative to the zero-time point and calculated by pixel counting (ImageJ software). The bars represent the mean ± S.D (n = 3). Statistical significance was calculated with a two-tailed unpaired t test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Source data are provided as a Source data file.

Fig. 5. Circular RNAs also exist in other bacteria.

a Numbers of chromosomal loci mapped by more than 5 permuted reads in a chiastic order within a region less than 1000 bp in some species of gram-positive (G⁺) or gram-negative (G^-) bacteria. b Statistics of circular RNA candidate number (upper) and permuted read number of four verified circular RNA loci (lower) in strain B. altitudinis SCU11 at different growth phases. c Verification of circular RNA candidates by RT‒PCR. Blue arrows are divergent primer pairs amplifying circular RNAs. cDNA templates were obtained by reverse transcription of total RNA isolated from B. altitudinis SCU11 (8 h, same as that in Fig. 2c) and B. paralicheniformis (24 h) pretreated with or without RNase R. Chromosomal DNA template was used as a negative control. EB-stained rRNA acted as RNase R digestion controls. d Verification of circular SRP RNA by northern blotting. Total RNA was extracted from B. altitudinis SCU11 (4, 12, 24, 48 h), B. paralicheniformis and B. subtilis 168 (24 h) cultured in LB medium. 500 ng of total RNA was loaded for northern blotting. Arrows connected RNA samples were treated with RNase R, and EB-stained rRNA acted as RNase R digestion controls. L and C indicate linear and circular RNAs, respectively. Similar results were obtained in two biologically independent experiments (c, d). Source data are provided as a Source data file.

Fig. 6. Model of DucS function in B. altitudinis SCU11.

a DucS produces different proportions of linear RNAs and circular RNAs depending on bacterial growth phases or stress conditions. Circular RNAs are formed through the circularization of linear RNA L1. Circular RNAs appear at the mid-logarithmic phase and increase gradually thereafter, and H₂O₂ stress can promote the production of linear RNAs. b Circular RNAs and linear RNAs play different roles. The linear RNAs increase bacterial oxidative tolerance by upregulating htrA mRNA translation and regulating other targets, while the presence of circular RNAs may decrease the levels of linear RNAs.