The opuB (opuBA-opuBB-opuBC-opuBD) and opuC (opuCA-opuCB-opuCC-opuCD) operons of B. subtilis encode closely related ABC transport systems with amino acid identities of their components ranging from 69 to 80% (Kappes et al., 1999). Probes for Northern blot analysis were designed against the 5′-part of the genes encoding the extracellular substrate-binding proteins of the ABC transporters [(opuBC; nucleotides 200–470) and (opuCC; nucleotides 17–519)], as these are the least conserved components of the OpuB and OpuC systems (Figure 1). In order to exclude any undesired cross-hybridization, the probes were tested against RNA samples of B. subtilis mutant strains in which one of the two operons is completely deleted. No hybridization signals were obtained in the respective deletion mutants (data not shown), demonstrating that these Northern blot probes can be used to specifically trace changes in the amounts of opuB and opuC mRNA, respectively.

opuB and opuC expression under conditions of acute and sustained salt stress has previously been assessed in a comprehensive tiling array study (Nicolas et al., 2012). To critically re-evaluate these data, the B. subtilis wild-type strain BSB1 (168 Trp⁺) was grown in Spizizen’s minimal medium (SMM) until an optical density at 600 nm (OD₆₀₀) of 0.3 was reached and then cells were stressed with 0.4 M NaCl for 10 min or grown in SMM in the absence or presence of NaCl (1.2 M) until reaching an OD₆₀₀ of 1. Northern blot analysis with the opuBC- and opuCC-specific probes (Figure 2) detected the expected full-length transcripts (approximately 3.5 Kb). This experiment revealed that 10 min after an osmotic upshift elicited by addition of 0.4 M NaCl, transcription of the opuC operon was induced, whereas the opuB expression level was slightly lower compared to the sample taken before exposure to salt stress. Only prolonged growth in high-salinity medium led to increased expression of opuB. Under these conditions, cells exhibited basal levels of opuC expression. These results are in agreement with the data reported by Nicolas et al., 2012 and confirmed the different patterns of transcriptional induction of the opuB and opuC operons in response to either suddenly imposed or sustained high salinity.

The asRNA S1290 (Figure 1B) was identified by Nicolas et al. (2012) and its transcriptional profile suggested that its synthesis was presumably dependent on the alternative sigma factor SigB. Inspection of the expression pattern of S1290 asRNA in the dataset provided by Nicolas et al. (2012) revealed a strong induction under stress conditions typical for SigB regulated genes (Nannapaneni et al., 2012), including the sudden exposure to salt, ethanol or heat (see B. subtilis Expression Data Browser¹). Following the canonical expression profile of SigB-regulated genes, the S1290 RNA was not detected under control conditions and during prolonged growth in high-salinity medium, but was strongly induced 10 min after imposition of salt stress (Figure 2), one of the strongest inducers of the entire SigB regulon (Petersohn et al., 2001). The 5′-end of S1290 resides in opuBD, the fourth gene of the opuB operon (Figure 1B and B. subtilis Expression Data Browser²). Using a probe complementary to nucleotides 51–351 of S1290, several transcripts were detected, with sizes of 0.3, 0.6, 0.75, 1.15, 1.7, and 3.8 Kb, of which the most prominent one (1.15 Kb) would end near the 5′-extremity of opuBC. The largest transcript most probably extends to the termination site of the downstream yvaV gene. Hence, the S1290 asRNA practically covers the entire coding region of the opuB operon and should interfere with opuB transcription or translation of the opuB mRNA.

Inspection of the opuBD region immediately preceding the 5′-end of the S1290 RNA and the corresponding region of opuCD revealed the presence of a conserved SigB-type promoter (Petersohn et al., 2001) only in the case of the opuBD (Figure 3A). In contrast, the corresponding region of opuCD differs in two conserved positions of the promoter elements (A instead of G at position 1 of the −35 region and A instead of G at position 2 of the −10 region). To prove that SigB is responsible for stress induction of the S1290 promoter, S1290 levels were compared in the B. subtilis wild-type BSB1 and an isogenic sigB deletion mutant. In the strain lacking SigB, no S1290 RNA was observed after imposition of salt stress (Figure 3B). Strikingly, under these conditions opuB expression was induced already after 10 min of salt stress to the same level as that of opuC. This observation provided the first indication that osmotic induction of opuB expression might be affected by the SigB-controlled S1290 asRNA.

In order to analyze the relationship between expression of the opuB operon and the S1290 asRNA, we grew the B. subtilis wild-type strain BSB1 (168 Trp⁺) in SMM and challenged exponentially growing cells with either 0.4 M or 1 M NaCl. Samples for RNA preparation were withdrawn before and 10, 30, 60, 150, and 300 min after addition of NaCl. Because antisense-sense interactions can affect not only mRNA amounts, but also transcript patterns through mRNA processing or partial degradation (Georg and Hess, 2018), expression of S1290 and the opu genes was monitored by Northern blot analysis, allowing for quantitative measurements of relative RNA abundance and detection of individual RNA species with different sizes. This analysis revealed a delayed induction of the opuB operon after imposition of salt stress (Figure 4A). At the first time point analyzed (10 min), opuB transcript levels were more than 1.5-fold lower compared to pre-stress levels (Table 1). The earliest time points at which induced transcript levels could be observed were 30 min in the case of 0.4 M NaCl and 150 min in the case of 1 M NaCl. The S1290 asRNA showed an expression pattern inversely related to that of opuB. S1290 levels peaked at 10 or 30 min, respectively, after the osmotic upshift, which can be explained by the transient activation of SigB. This difference in the timing of S1290 expression, which depends on the severity of the osmotic stress, is in agreement with previous results on the speed and magnitude of SigB activity (Cabeen et al., 2017). In contrast to opuB, opuC expression responded to an osmotic upshift within 10 min and reached a maximum level at 30 min (Figure 4A). At subsequent time points, opuC mRNA gradually declined and returned to pre-stress levels in cells adapted to growth in high-salinity medium. The decline was more pronounced at the higher salt concentration of 1 M NaCl. The results obtained in this experiment suggest that expression of the S1290 asRNA retards the osmotic induction of the opuB operon.

In order to verify that expression of the S1290 asRNA is indeed causative for the time-delayed induction of the opuB operon, we sought to construct a mutant in which only S1290 is no longer expressed. Because the S1290 promoter is located within the coding region of opuBD (Figure 1B), it was not possible to simply delete the promoter region. Therefore, the promoter was inactivated by altering essential positions of the −10 and −35 regions of the presumed SigB-dependent promoter without changing the amino acid sequence of the OpuBD protein (Figure 5A). In the resulting P_S1290 promoter mutant, transcription of S1290 in response to salt stress was abolished. Importantly, Northern Blot analysis of a time-series experiment (samples taken before and 10, 30, 60, 150, and 300 min after addition of 0.4 M NaCl) showed that delayed upregulation of the opuB operon was no longer observed in the absence of S1290 (Figure 5B). In this strain, imposition of salt stress led to a more than two-fold increase in opuB mRNA levels within 10 min, which were further increased at the 30 min time point (Table 1). Comparable levels of opuB mRNA were detected 60 min after stress induction in wild-type and P_S1290 mutant cells.

AsRNAs can regulate their sense mRNAs by various mechanisms, which rely on either base-pairing interactions or transcription interference (for reviews see Thomason and Storz, 2010; Georg and Hess, 2018). Formation of double-stranded RNA can trigger RNA degradation by providing a substrate for the RNase III endoribonuclease. The double-strand specific RNase III (encoded by rnc) was shown to play a role in the regulation of gene expression by asRNAs in Gram-positive bacteria (reviewed in Durand and Condon, 2018). Therefore, we investigated the impact of RNase III inactivation on opuB sense and antisense transcript levels. Because rnc can only be deleted in the absence of two prophage-encoded toxin genes (Durand et al., 2012), we constructed a ΔtxpA ΔyonT Δrnc mutant. Using this strain, we performed the same time-series analyses of opuB and S1290 transcript levels after adding 0.4 M or 1 M NaCl as described above for the wild-type strain (Figure 4A). The amounts of opuB mRNA and S1290 asRNA as well as the relative amounts of the individual S1290 species were not significantly different between the ΔtxpA ΔyonT Δrnc deletion mutant (Figure 4B) and the wild-type strain, ruling out that the asRNA exerts its effect on opuB expression via co-degradation of the RNA duplex by RNase III. This result suggested that antisense regulation of opuB occurs at the level of transcription, in that stress-induced transcription from the SigB-dependent S1290 promoter could suppress transcription of opuB by interference between convergent RNA polymerases. In support of this hypothesis, delayed induction of opuB expression in response to salt stress was also observed in a mutant lacking the transcriptional repressor (GbsR) of the opuB operon (Nau-Wagner et al., 2012), a strain, which accordingly exhibits higher opuB transcript levels even under non-inducing conditions (Figure 6A and Table 1).

Next, we tested whether antisense regulation of opuB is abolished when S1290 is expressed from a different genomic locus. To this end, a sequence containing the S1290 promoter and a 3 Kb region of the asRNA was inserted into the amyE gene of the P_S1290 mutant. When S1290 is expressed in trans, the same antisense transcripts in similar amounts are detected as in wild-type cells (Figure 6B), of which only the largest transcript (3.3 Kb) is slightly shorter because the inserted sequence does not contain the opuBA promoter region and the yvaV gene. In the trans constellation, the increase in opuB mRNA abundance after imposition of salt stress was no longer delayed (Figure 6B) as it was the case in the absence of S1290 expression (Figure 5B). Hence, the S1290 asRNA can be assumed to control gene expression by interfering with transcription of the opuB operon rather than a mechanism that depends on asRNA-mRNA base pairing.

Osmotic induction of opuB transcription and the choline-responsive release of the GbsR repressor from its operator are genetically separable events (Nau-Wagner et al., 2012). As expected, the ΔgbsR mutant exhibited both stronger opuB basal level expression as well as induction after imposition of salt stress as compared to the wild-type, with the highest level reached 30 min after the imposition of the salt shock (Figure 6A). However, salt induction was still delayed in the gbsR mutant and did not occur 10 min after addition of 0.4 M NaCl. Import of choline, the inducer for the GbsR repressor, also relieves GbsR-mediated repression of the opuB and gbsAB operons, thus enabling enhanced choline uptake and synthesis of glycine betaine (Nau-Wagner et al., 2012). Therefore, we expected a similar pattern of opuB induction elicited by salt stress when the wild-type was grown in SMM supplemented with choline as observed in the strain lacking the GbsR repressor. Indeed, 30 min after addition of 0.4 M NaCl, enhanced opuB expression level was detected when 1 mM choline was present in the medium (Figure 6C), data comparable to those obtained with the gbsR mutant. However, in contrast to the gbsR mutant, wild-type cells grown in the presence of choline showed an immediate increase in opuB transcript level already 10 min after salt shock. Strikingly, salt induction of the S1290 asRNA was strongly reduced under these conditions.

These observations led us to speculate that the intracellular accumulation of glycine betaine synthesized from choline reduces S1290 levels. In the presence of extracellular glycine betaine or choline, the intracellular glycine betaine pool of B. subtilis cells increases in tune with the increases in the salt concentration of the growth medium (Hoffmann et al., 2013). As noted before, glycine betaine accumulates to notable levels even in cells grown in SMM without additional NaCl, a process that is largely dependent on the substantial transport activity of the OpuA system even in the absence of notable osmotic stress in a standard minimal growth medium (Hoffmann et al., 2013). Our results (Figure 6C) thus indicate that these levels of glycine betaine in non salt-stressed cells can already affect induction of S1290 expression. Strikingly, addition of glycine betaine to salt-stressed cells completely prevented S1290 induction when added at the time of stress imposition (Figure 7A); this phenomenon is likely due to the fast accumulation of glycine betaine from exogenous sources in salt-stressed cells (Kappes et al., 1996; Hoffmann and Bremer, 2017). If glycine betaine was added 60 min before the osmotic upshift, S1290 RNA was clearly detectable, but – as observed when B. subtilis was grown in the presence of the glycine betaine precursor choline (Figure 6C) – the S1290 level was significantly reduced compared to conditions without added glycine betaine. If indeed the osmoprotectant glycine betaine is responsible for reduced S1290 levels, addition of choline to the culture at different times prior to the salt shock should have different effects on the appearance of the S1290 RNA, because choline needs to be first converted to glycine betaine to confer cellular osmostress protection (Boch et al., 1996). In support of this hypothesis, SigB-dependent S1290 levels were not reduced if choline was added to the cultures at the time of salt stress imposition, but gradually declined with increasing time that had passed (15, 30, 60 min) between the addition of choline prior to the salt shock (Figure 7A). Thus, intracellular accumulation of glycine betaine seems to influence signal perception by SigB, as assessed by S1290 levels under conditions of salt stress.

In order to further probe whether salt-stress dependent activation of SigB is mitigated when cells are pre-loaded with glycine betaine, we analyzed at a genome-wide level salt induction of the SigB regulon in the presence of choline using microarrays. The glycine betaine precursor choline was added to exponentially growing cultures either immediately, or 15, 30, and 60 min prior to the osmotic upshift imposed to all cultures at OD₆₀₀ of 0.3. RNA was then isolated for transcriptome analysis from cells that were stressed with 0.4 M NaCl for 10 min. After sample processing and quantification of the hybridization signals, we first investigated S1290 RNA levels, thereby confirming that induction of S1290 decreased with increasing duration of presence of choline in the growth medium prior to the salt shock (Figure 7B). We then analyzed the expression levels of all known SigB-regulated genes (Nannapaneni et al., 2012; Nicolas et al., 2012; Supplementary Table S1). As depicted in Figure 7C, the SigB regulon showed the same pattern of gradual reduction in transcriptional induction with increasing duration of the presence of choline as observed for the S1290 asRNA. The median expression value of SigB-regulated genes was 1.7-fold higher in cells that received choline immediately before the salt stress compared to cells where choline was added 60 min prior to the salt stress.

Next, we assessed if even reduced S1290 levels observed when B. subtilis is grown in the presence of choline can affect opuB expression after salt shock. To this end, we compared opuB induction in wild-type and P_S1290 mutant cells cultivated in SMM containing 1 mM choline. Indeed, the moderate increase in opuB expression observed in the wild-type 10 min after addition of 0.4 M NaCl (Figures 6C, 8A) was significantly enhanced in the mutant lacking the S1290 asRNA, where the maximum level of opuB induction was already reached at this early time point (Figure 8A). Hence, expression of opuB after salt shock peaked earlier in the S1290 promoter mutant as compared to the wild-type, whereas the magnitude of induction reached under these conditions of choline-mediated release of GbsR remained unaffected.

In response to a sudden osmotic upshift, B. subtilis induces the expression of the opuA, opuC and opuD glycine betaine transporters genes (Nicolas et al., 2012; Hoffmann et al., 2013), thus accomplishing immediate uptake of glycine betaine. Intracellular glycine betaine counteracts the choline-mediated de-repression of the opuB and gbsAB operons through its influence on GbsR (Nau-Wagner et al., 2012). We hypothesized that in the presence of glycine betaine and choline in the medium, delayed opuB expression could result in an intracellular glycine betaine to choline ratio that prevents strong induction of opuB and gbsAB after salt shock. In order to test this hypothesis, we grew the B. subtilis wild-type and the P_S1290 mutant strains in SMM containing 1 mM glycine betaine/1 mM choline and exposed exponentially growing cells to salt shock with 0.4 M NaCl. The lack of the S1290 asRNA in the P_S1290 strain indeed led to stronger induction of the opuB and gbsAB operons (Figure 8B), which can be explained by increased choline import resulting in the removal of GbsR-mediated repression (Nau-Wagner et al., 2012). Thus, the asRNA-mediated delay in choline uptake prevents strong expression of opuB and gbsAB in B. subtilis wild-type cells and thereby reduces synthesis of proteins dedicated to choline uptake and conversion.

Another striking difference between the wild-type and the P_S1290 mutant subjected to salt stress in the presence of 1 mM glycine betaine/1 mM choline was the de-repression of both opuB and gbsAB 300 min after the osmotic upshift in wild-type, but not in mutant cells (Figure 8B). Probably, the pool of glycine betaine provided in the medium had been exhausted by the wild-type cells before that time point, necessitating the uptake and conversion of choline to maintain the intracellular glycine betaine pool required for growth at 0.4 M NaCl. In the case of the P_S1290 mutant, synthesis of glycine betaine from imported choline would reduce the consumption of externally provided glycine betaine, and indeed no de-repression of the GbsR-dependent genes occurred at 300 min, because sufficient amounts of glycine betaine seemed to be still available making choline uptake unnecessary. If this assumption is true, earlier de-repression of opuB and gbsAB should be observed if lower amounts of glycine betaine are provided in the medium und de-repression should be prevented by supply of excess glycine betaine, respectively. Indeed, when salt stress was applied to wild-type cells growing in SMM supplemented with a mixture of 0.2 mM glycine betaine/0.2 mM choline, increased opuBC and gbsA mRNA levels were already observed after 150 min (Figure 8C). Consistently, in SMM containing 5 mM glycine betaine/5 mM choline, recurrence of opuB and gbsAB induction was not observed within 300 min after the osmotic upshift.

All B. subtilis strains used in this study are listed in Table 2 the oligonucleotides are listed in Table 3. The B. subtilis strain BHR006 (ΔtxpA ΔyonT Δrnc) was constructed by sequential deletion of the toxin genes yonT and txpA (Durand et al., 2012) followed by deletion of rnc, using a modified two-step PCR method for synthesis of linear DNA fragments carrying a resistance marker flanked by sequences homologous to the upstream and downstream regions of the respective gene (Wach, 1996). The corresponding fragments were amplified using the primer pairs listed in Table 3 and fused to the respective gene mediating resistance to spectinomycin (spc), phleomycin (phleo) or kanamycin (km). The resulting fragments were used to transform competent B. subtilis BSB1 cells (Harwood and Cutting, 1990; Konkol et al., 2013), with subsequent selection for antibiotic-resistant colonies. The correct insertion of the resistance marker and absence of point mutations in the upstream and downstream regions were verified by sequencing (Eurofins Genomics, Ebersberg, Germany). To avoid potential polar effects, all deletions are in-frame replacements of the gene coding sequences by the respective resistance gene sequence, leaving the chromosomal context intact. The gbsR deletion mutant strain TMB405 was constructed by transforming B. subtilis BSB1 with chromosomal DNA (0.2 μg) of B. subtilis JBB8 (gbsR::neo, Boch et al., 1996).

For inactivation of the S1290 SigB-type promoter, the −35 core promoter region was changed in two positions from GTTTCG to ATTACG that still encodes for arginine (CGT) and asparagine (AAT) and the −10 region was changed in four positions from GGGAAT to GACTGT that still encodes for tyrosine (TAC) and serine (AGT) at the respective positions of opuBD on the opposite strand (see Figure 5A). To generate the P_S1290 mutant, a linear DNA fragment carrying a spectinomycin resistance gene flanked by sequences homologous to upstream and downstream regions of the S1290 promoter was synthesized using the listed primer pairs (Table 3). The upstream sequence corresponds to yvaQ, and the spectinomycin resistance gene with a Shine-Dalgarno sequence was placed behind the yvaQ gene, creating a transcriptional fusion terminated at the original yvaQ terminator to avoid read-through into the opuB-operon. The base substitutions were introduced into the downstream fragment via primer P_S1290_for. The four overlapping fragments were merged by fusion PCR and used for transformation of competent B. subtilis BSB1 cells. Antibiotic-resistant transformants were selected and confirmed by sequencing (Eurofins Genomics, Ebersberg, Germany). The BHR018 mutant was constructed by integrating the opuB region without the opuB promoter into the amyE gene of strain BHR010. Genomic DNA of a B. subtilis strain in which the complete opuB region was integrated into amyE by use of the cloning vector pX (Kim et al., 1996) served as PCR template. Two DNA fragments containing (i) sequences homologous to the upstream region of the integration site and the opuB coding sequence and (ii) the cat gene of the pX vector and sequences homologous to the downstream region of the integration site were synthesized using primer pairs listed in Table 3. The two fragments were combined by fusion PCR (Wach, 1996) and used to transform competent B. subtilis BHR010 cells. Antibiotic-resistant transformants were selected and confirmed by sequencing (Eurofins Genomics, Ebersberg, Germany).

Bacillus subtilis transformants were selected on lysogeny broth (LB) agar plates containing spectinomycin (200 μg ml^–1), phleomycin (2 μg ml^–1) or kanamycin (2 μg ml^–1).

For stress experiments, B. subtilis strains were cultivated in Spizizen’s minimal medium (SMM) (Harwood and Cutting, 1990) with 0.5% (wt/vol) glucose as carbon source. Cultures were inoculated from exponentially growing pre-cultures to an optical density at 600 nm (OD₆₀₀) of 0.1 and propagated at 37°C under vigorous shaking. To induce hyperosmotic stress, osmolarity was increased by addition of NaCl (pre-warmed 4 M stock solution prepared in SMM) at OD₆₀₀ of 0.3. When indicated glycine betaine or choline were added to the growth medium from 1 M stock solutions.

Samples for RNA preparation were collected at different time points after an osmotic upshift as indicated. Cells were harvested by addition of ½ volume of frozen killing buffer (20 mM Tris/HCl [pH 7.5], 5 mM MgCl₂, 20 mM NaN₃) and subsequent centrifugation for 3 min at 8.000 g and 4°C. After discarding the supernatant, cell pellets were frozen in liquid nitrogen and stored at −80°C.

Total RNA was prepared by acid-phenol extraction after mechanical cell disruption as described previously (Nicolas et al., 2012). The quality of the RNA was assessed by means of an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, United States) according to the manufacturer’s instructions. Northern blot analysis was performed as described before (Homuth et al., 1997). Transcript sizes were determined using the RiboRuler High Range Ladder (Thermo Fisher Scientific, Waltham, MA, United States) and the digoxigenin-labeled RNA Molecular Weight Marker II (Roche Holding AG, Basel, Switzerland). The bands of the unlabeled RNA size standard were marked directly on the membrane, thereby becoming visible as negative bands during detection that were then indicated by lines in the northern blot images. Quantification of the 3.5 kb opuB transcript and the 1.2 kb S1290 transcript was performed using Image Studio Lite (version 5.2.5, LI-COR Biosciences, Lincoln, NE, United States).

35 μg of total RNA from two biological replicates per condition were DNase-treated using the RNase-Free DNase Set (Qiagen, Hilden, Germany) and purified using the RNA Clean-Up and Concentration Kit (Norgen, Biotek Corp., Thorold, ON, Canada). After quality control (Agilent 2100 Bioanalyzer), 5 μg of the purified RNA were subjected to microarray analysis. Synthesis and fluorescence labeling of cDNA followed a strand-specific method using the FairPlay III Microarray Labeling Kit (Agilent Technologies, Santa Clara, CA, United States) and actinomycin D (Calbiochem, Merck KGaA, Darmstadt, Germany) (Nicolas et al., 2012). 100 ng of Cy3-labeled cDNA were hybridized to the microarray following Agilent’s hybridization, washing and scanning protocol (One-Color Microarray-based Gene Expression Analysis, version 5.5). Data were extracted and processed using the Feature Extraction software (version 12.1). For each gene, the median of the individual probe intensities was calculated and further data analysis was performed using Genedata Analyst^TM software (Genedata AG, Switzerland) and R 3.5.0 (R Core Team, 2018). The microarray data set is available from NCBI’s Gene Expression Omnibus (GEO) database (accession number GSE141882).