A σ^D-dependent antisense transcript modulates expression of the cyclic-di-AMP hydrolase GdpP in Bacillus subtilis

Abstract

Cyclic-di-AMP (c-di-AMP) is an essential second messenger in Bacillus subtilis, and depletion leads to defects in the integrity of the cell wall. Levels of c-di-AMP are regulated by both the rates of synthesis (by diadenylate cyclases) and the rates of degradation (by the GdpP phosphodiesterase, formerly YybT). Little is known about the regulation of gdpP expression or GdpP activity, but mutations that inactivate GdpP lead to high-level resistance to β-lactam antibiotics. Here we demonstrate that expression of gdpP is regulated by a cis-acting antisense RNA (gdpP_as) in vivo. Transcription of this antisense RNA is initiated in the middle of the gdp gene and is dependent on an alternative sigma factor, σ^D, previously associated with the expression of late flagellar genes, chemotaxis proteins and cell wall autolytic enzymes. Changes in σ^D activity can modulate GdpP protein levels by ~2.5-fold, which may provide a mechanism for the cell to upregulate c-di-AMP levels in coordination with the activation of autolytic enzymes.

Introduction

Cyclic-di-AMP (c-di-AMP) is a recently recognized second messenger molecule in bacteria. It was first identified as an endogenous metabolite in the crystal structure of DisA, which catalyses its synthesis at its DAC domain (diadenylate cyclase domain; previously DUF147 domain) (Witte et al., 2008). DAC domain-containing proteins can be found in many other (predominantly Gram-positive) bacteria and archaea (Römling, 2008). Many species only harbour one DAC domain protein, and null mutation of the sole DAC appears to be lethal in Listeria monocytogenes, Staphyloccocus (Staph.) aureus, Streptoccocus pneumoniae, Mycoplasma pulmonis and Mycoplasma genitalium (Chaudhuri et al., 2009; Corrigan et al., 2011; French et al., 2008; Glass et al., 2006; Song et al., 2005; Woodward et al., 2010). The genome of Bacillus subtilis encodes three DAC-containing proteins, DisA, YbbP and YojJ. Although single mutants of these DAC proteins are viable, the double mutant lacking both DisA and YbbP is non-viable (Luo & Helmann, 2012). The essential roles of c-di-AMP are not well understood, but recent results suggest that it is, directly or indirectly, involved in peptidoglycan (PG) homeostasis (Corrigan et al., 2011; Luo & Helmann, 2012).

Cellular levels of c-di-AMP are regulated by rates of synthesis and degradation. The DAC domain in the cyclase catalyses the synthesis of c-di-AMP from two ATP molecules, while the DHH domain in the hydrolase catalyses cleavage of c-di-AMP to linear 5′-pApA. In B. subtilis, the three c-di-AMP cyclases (DisA, YbbP and YojJ) and one hydrolase (GdpP) are subject to both transcriptional and post-transcriptional regulation. Expression of disA is regulated by both σ^A and σ^M (Eiamphungporn & Helmann, 2008), and the cyclase activity of DisA is influenced by DNA integrity. DisA forms a large octamer that moves along intact chromosomal DNA. Upon encountering a DNA double-strand break, the DisA complex pauses at the lesion site and ceases c-di-AMP synthesis, thus delaying sporulation (Bejerano-Sagie et al., 2006; Oppenheimer-Shaanan et al., 2011; Witte et al., 2008). The second cyclase, YbbP, is orthologous to the essential DAC proteins of L. monocytogenes and Staph. aureus. Its transcript is mainly dependent on σ^A, with possible read through from an upstream σ^W promoter (Cao et al., 2002; Luo & Helmann, 2012). YbbP is a membrane-localized protein that responds to cell envelope stress. Mutation of ybbP results in increased susceptibility to β-lactam antibiotics such as cefuroxime (Luo & Helmann, 2012). The third enzyme, YojJ, is a cytosolic protein. In vegetatively growing cells, YojJ is able to restore growth to a normally lethal double disA ybbP mutant only if artificially overexpressed (Luo & Helmann, 2012).

GdpP (formerly YybT) is the only known c-di-AMP phosphodiesterase (PDE) in B. subtilis. This transmembrane protein contains three functional domains: a haem-binding PAS domain, a degenerate GGDEF domain and a DHH/DHHA1 PDE domain (Rao et al., 2010, 2011). The PDE activity of GdpP can be inhibited by the alarmone ppGpp and by haem in vitro. The haem-dependent PAS inhibition can be partially relieved by nitric oxide (NO). However, the biological relevance of these haem, NO and ppGpp effects on PDE activity has not been studied in vivo. In this work, we characterize an antisense RNA transcribed from within gdpP. This cis-acting RNA is dependent on σ^D and it can modulate the cellular levels of GdpP.

Methods

Bacterial strains and growth conditions.

Strains used in this study are listed in Table 1. B. subtilis strains are derivatives of strain 168 or NCIB 3610. Escherichia coli strain DH5α was used for standard cloning procedures. Unless noted otherwise, all cultures were grown in Luria–Bertani (LB) broth at 37 °C with vigorous shaking. Antibiotics were added to the growth medium when appropriate: 100 µg ampicillin ml⁻¹ for E. coli, and 1 µg erythromycin ml⁻¹ plus 25 µg lincomycin ml⁻¹ (MLS; macrolide-lincomycin-streptogramin B resistance), 10 µg chloramphenicol ml⁻¹, 100 µg spectinomycin ml⁻¹, 5 µg tetracycline ml⁻¹ and 10 µg kanamycin ml⁻¹ for B. subtilis. OD₆₀₀ readings were taken on a Spectronic 21 spectrophotometer.

Table 1. Strains used in this study.

Strain	Genotype	Source or reference
168	trpC2	Laboratory stock
NCIB3610	Prototrophic, undomesticated parent of 168	Laboratory stock
HB15826	168 amyE : : PyybS-lacZ cat	This study
HB15827	168 amyE : : PgdpPas-lacZ cat	This study
HB15829	168 flgM : : spc amyE : : PyybS-lacZ cat	This study
HB15830	168 flgM : : spc amyE : : PgdpPas-lacZ cat	This study
HB15833	168 sigD : : kan amyE : : PyybS-lacZ cat	This study
HB15834	168 sigD : : kan amyE : : PgdpPas-lacZ cat	This study
HB15835	168 sigD : : kan flgM : : spc amyE : : PgdpPas-lacZ cat	This study
HB15885	168 amyE : : PD*-lacZ cat	This study
HB15886	168 sigD : : kan amyE : : PD*-lacZ cat	This study
HB15887	168 flgM : : spc amyE : : PD*-lacZ cat	This study
HB13056	168 thrC : : PfabHaF-fabHa-fabF-FLAG mls	Kingston et al. (2011)
HB15843	168 gdpP : : pMUTIN4-gdpP-FLAG mls	This study
HB15845	168 gdpP-FLAG	This study
HB15857	168 gdpP-FLAG thrC : : PfabHaF-fabHa-fabF-FLAG mls	This study
HB15858	168 sigD:kan gdpP-FLAG thrC : : PfabHaF-fabHa-fabF-FLAG mls	This study
HB15859	168 flgM : : spc gdpP-FLAG thrC : : PfabHaF-fabHa-fabF-FLAG mls	This study
HB15860	168 sigD:kan flgM : : spc gdpP-FLAG thrC : : PfabHaF-fabHa-fabF-FLAG mls	This study
HB15836	168 gdpP : : pMUTIN4-PD* mls	This study
HB15837	168 PD*	This study
HB15901	3610 PD*	This study
HB15902	3610 flgM : : cat	This study
HB15903	3610 flgM : : cat PD*	This study
HB15904	3610 flgM : : cat amyE : : Physpank-sigD kan	This study
HB15905	3610 flgM : : cat amyE : : Physpank-sigD kan PD*	This study
HB15906	3610 flgM : : cat amyE : : Physpank-sigD kan lytABC : : spc lytD : : mls lytF : : tet	This study
HB15907	3610 flgM : : cat amyE : : Physpank-sigD kan lytABC : : spc lytD : : mls lytF : : tet PD*	This study
HB15908	3610 disA : : spc flgM : : cat	This study
HB15909	3610 disA : : spc flgM : : cat PD*	This study
HB15910	3610 ybbP : : tet flgM : : spc amy : : Phag-cat-lacZ	This study
HB15911	3610 ybbP : : tet flgM : : spc PD* amy : : Phag-cat-lacZ	This study
HB15912	3610 sigM : : kan ybbP : : tet flgM : : spc amy : : Phag-cat-lacZ	This study
HB15913	3610 sigM : : kan ybbP : : tet flgM : : spc PD* amy : : Phag-cat-lacZ	This study
HB15914	3610 disA : : spc amyE : : Pspac(hy)-gdpP cat	This study
HB15915	3610 ybbP : : tet amyE : : Pspac(hy)-gdpP cat	This study
HB15916	3610 yojJ : : kan amyE : : Pspac(hy)-gdpP cat	This study
HB15917	3610 gdpP : : mls amyE : : Pspac(hy)-gdpP cat	This study
HB15918	3610 amy : : Pspac(hy)-gdpP cat	This study

Strain construction.

All strains were generated by transformation of strain 168 to antibiotic resistance using chromosomal DNA, PCR products or plasmids as described elsewhere (Harwood & Cutting, 1990). When required, mutations in the strain 168 background were transferred to the strain NCIB 3610 background by SPP1-mediated generalized transduction, as described elsewhere (Kearns & Losick, 2005). Unless stated otherwise, all PCR products were generated using chromosomal DNA from strain 168 as a template and all constructs were verified by sequence analysis (Cornell University Life Sciences Core Laboratories Center). The oligonucleotides used in this study are listed in Table 2.

Table 2. Oligonucleotides used in this study.

No.	Oligonucleotide	Sequence (5′–3′)*
5586	yybTanti-GSP1	CGATGAGTGAGGTTGAGGCT
5587	yybTanti-GSP2	GGAATACTTCGATCAAGTGCTGA
5588	yybTanti-GSP3	ATCAATATGATTGAAGCAACAGC
5589	yybTanti-GSP4	ACAGCGGAATTGGTGACAGA
5590	yybTanti-GSP5	AGCCGTCACTCGTCATGGA
5591	yybTanti-GSP6	CGCCTGAAGAAGCAATGGA
5592	yybTanti-GSP7	GTCACCGAAAGCAGCAATGT
5593	yybTanti-GSP8	AACCCAATGGAGAAACGAACA
5594	yybTanti-GSP9	CGCAATCCAGCTTGGACTT
5595	yybTanti-GSP10	CTGACGTTAAGCGTCGGTGT
4549	AAP	GGCCACGCGTCGACTAGTACGGGIIGGGIIGGGIIG
5628	sigD-up-for	CTGATGGAGCTCAGTCAGGT
5629	sigD-up-rev (kan)	CCTATCACCTCAAATGGTTCGCTGCCTGATCTTCATAATTCAAGGA
5630	sigD-do-for (kan)	CGAGCGCCTACGAGGAATTTGTATCGTCAGATCCATTCAAAGGCATT
5631	sigD-do-rev	GCTTCATAGAAATGACTGACATGT
5632	flgM-up-for	TTCAGCAGAAGGTATGAATATCA
5633	flgM-up-rev(spc)	CGTTACGTTATTAGCGAGCCAGTCGATATGGATTAACGGATTGTGT
5634	flgM-do-for(spc)	CAATAAACCCTTGCCCTCGCTACGTCATACAAAGTAGACGCAAATCA
5635	flgM-do-rev	CCGCTGTCTTGTATAACCATCA
5879	flgM-up-rev(cat)	CTTGATAATAAGGGTAACTATTGCCGATATGGATTAACGGATTGTGT
5880	flgM-do-for(cat)	GGGTAACTAGCCTCGCCGGTCCACGTCATACAAAGTAGACGCAAATCA
5565	PyybS-for (EcoRI)	GAGGAATTCACAAAGAGGTGAAACACAATG
5566	PyybS-rev (BamHI)	GAGGGATCCCAATCACAGGAACATAAACGA
5598	PyybTanti-for (EcoRI)	GAGGAATTCACGAGTGACGGCTTATGTGT
5619	PyybTanti-rev2 (BamHI)	GAGGGATCCGTCACCGAAAGCAGCAATGT
5620	yybT-int-for (EcoRI)	GAGGAATTCATCAAACGGGATGAGCGTCTCT
5621	yybT-PDmut-up-rev	TCGCTGCGCCTATGGAATCCATGTCGGGGAATTTATG
5622	yybT-PDmut-do-for	CATAAATTCCCCGACATGGATTCCATAGGCGCAGCGATCGGGATTTTAAAG
5623	yybT-rev (BamHI)	GAAGGATCCTCATCTCTGTACGCCTCCCT
5684	yybS-int-up-for-(EcoRI)	GAGGAATTCTTGATATTGTAGAAACTGTAGCGA
5685	yybS-up-rev-(FLAG)	CATCCGCGGTTTATCATCATCATCTTTATAATCCATTTCTATCACTCCCCACCATGT
5246	yybT-for N-flag 1	ATGGATTATAAAGATGATGATGATAAACCGCGGATGCCAAGCTTTTATGAAAAAC
5686	yybT-int-do-rev (BamHI)	GAGGGATCCATCCAATCCTTGTGTCACATCA
5298	yybT-for	AGTGATAGAAATGCCAAGCT

^* Restriction sites are underlined.

To generate promoter–lacZ fusions, a DNA fragment containing P_yybS or P_gdpPas was PCR-amplified with primer pairs 5565/5566 or 5598/5619, respectively, and cloned into vector pDG1661 (Guérout-Fleury et al., 1996). The resulting plasmids were linearized by digestion with ScaI and integrated into strain 168 at the amyE locus. To create the P_D*-lacZ fusion, the same protocol was used except that the DNA fragment was amplified using strain HB15837 (168 P_D*) as template.

To generate the at-locus marker-less mutation P_D* (HB15837), we utilized an unstable integrative plasmid pMUTIN4, which harbours MLS resistance and lacZ genes (Vagner et al., 1998). A DNA fragment containing P_D* was first generated using overlap-extension PCR as described previously (Gaballa et al., 1998; Ho et al., 1989) with some modifications. Briefly, the up-fragment (including 620 bp upstream of the mutation site) and down-fragment (including 900 bp downstream of the mutation) were amplified using PCR with primer pairs 5620/5621 and 5622/5623, respectively. Primers 5621 and 5622 introduce the desired P_D* mutation and are complementary to each other. These up- and down-fragments were then joined using PCR with primers 5620/5623. The resulting PCR product was cloned into pMUTIN4, which was transformed into strain 168. The transformants were plated on LB agar plates supplemented with MLS and X-Gal (50 µg ml⁻¹). The resulting strain with plasmid pMUTIN4 integrated at the gdpP locus (strain HB15836) was resistant to MLS and blue on X-Gal plates. This at-locus integration of pMUTIN4 is not stable, and is capable of looping out from the chromosome, leaving behind either the wild-type (WT) or mutant sequence in the chromosome. To loop out pMUTIN4, cells of strain HB15836 were grown overnight in LB broth (without antibiotic selection), reinoculated into LB diluted to 1 : 100, grown to OD₆₀₀ 0.4 and diluted to 1 : 10 000, and 100 µl of cells was plated on LB agar supplemented with X-Gal. Cells that had lost the pMUTIN4 plasmid appeared as white colonies on X-Gal plates, and were sensitive to MLS. The strain harbouring the P_D* mutation was verified by PCR amplification using primers 5298/5598 and DNA sequencing. To generate P_D* in the strain NCIB 3610 background, the gdpP : : pMUTIN4-P_D* construct in strain HB15836 was transferred to 3610 by SPP1 transduction, followed by the pMUTIN4 loop-out assay as above.

To generate an at-locus, marker-less N-terminal FLAG-tagged gdpP strain (HB15857), a similar protocol was used as for P_D* construction. A DNA fragment containing gdpP-flag was constructed by overlap extension using up-fragment primers 5684/5685 and down-fragment primers 5246/5686. The flag sequence harbours a PsiI restriction site. After looping out plasmid pMUTIN4, colonies that were white on X-Gal plates and sensitive to MLS were screened by PCR using primers 5684/5686 followed by PsiI digestion. The correct construct was verified by DNA sequencing.

Gene deletions (including constructs of sigD : : kan, flgM : : spc, flgM : : cat) were generated by replacing the coding region with an antibiotic resistance cassette using long flanking homology PCR (LFH-PCR) followed by DNA transformation as described previously (Mascher et al., 2003).

β-Galactosidase activity measurements.

Strains harbouring promoter–lacZ fusions were grown overnight in 5 ml LB broth at 30 °C with vigorous shaking. Cells from 0.5 ml culture were harvested and β-galactosidase assays were performed as described by Miller (1972). Each strain was tested in biological triplicates and repeated three times. Data are reported as the mean and sem.

5′ Rapid amplification of cDNA ends (5′-RACE).

The transcriptional start site of antisense gdpP was determined using 5′-RACE. Five pairs of primers (5586/5587, 5588/5589, 5590/5591, 5592/5593, 5594/5595) were used to map antisense transcripts initiated at the +800 to 1980 bp region relative to the gdpP start codon. For each 5′-RACE, 5 µl total RNA from a mid-exponential-phase LB culture of strain 168 cells was reverse-transcribed to cDNA using TaqMan reverse transcription reagents (Roche) and the first primer (5586, 5588, 5590, 5592 or 5594). The 3′ end of cDNA was tailed with poly-dCTP using terminal deoxynucleotidyltransferase (New England Biolabs). The tailed cDNAs were then amplified by PCR with primer AAP (4549) and the second primer (5587, 5589, 5591, 5593 or 5595, respectively). The PCR products were subject to DNA sequencing.

Immunoprecipitation and Western blotting.

The FLAG-tagged GdpP and FabF were purified using immunoprecipitation with Anti-FLAG M2 Affinity Gel (Sigma-Aldrich) and quantified by Western blotting analysis. An overnight LB culture of B. subtilis cells harbouring both gdpP-flag and/or fabF-flag constructs was inoculated into 300 ml LB broth at an initial OD₆₀₀ of 0.01, and grown to OD₆₀₀ 0.5–0.6 at 37 °C with vigorous shaking. Cells were harvested by centrifugation and resuspended in 4 ml lysis buffer [50 mM Tris/HCl (pH 7.4), 150 mM NaCl, 1 % Triton X-100, 1 mM EDTA and 2 mg lysozyme ml⁻¹]. The cell suspension was incubated at 37 °C for 20 min, followed by sonication and centrifugation at 9000 g for 5 min at 4 °C. Anti-FLAG resins were added to the supernatant, and incubated at 4 °C overnight with gentle agitation. The anti-FLAG resins were washed three times with TBS [50 mM Tris/HCl (pH 7.4), 150 mM NaCl], and proteins bound to the resins were eluted by resuspending in 50 µl SDS-sample buffer [62.5 mM Tris/HCl (pH 6.8), 2 % SDS, 10 % (v/v) glycerol, 50 mM DTT and 0.002 % bromophenol blue] and boiled for 5 min. The samples were briefly centrifuged, and proteins in the supernatant were resolved on a 10 % SDS-PAGE gel and electrophoretically transferred to an Immunblot PVDF membrane (Bio-Rad). Membranes were blocked in TBS with 0.5 % Tween 20 (TBST) supplemented with 5 % non-fat milk for 0.5 h at room temperature, followed by incubation with anti-FLAG rabbit antibody (Sigma-Aldrich) in TBST supplemented with 0.5 % non-fat milk overnight at room temperature. The membrane was washed three times for 10 min each in TBST, and incubated with alkaline phosphatase-coupled secondary goat anti-rabbit antibody for 1 h (Sigma-Aldrich). After being washed three times for 10 min each in TBST, the membrane was cut into two pieces to separate the GdpP-FLAG- and FabF-FLAG-containing sections, and developed for 10 and 1 min, respectively. The developing reagent contained 100 mM Tris/HCl (pH 9.5), 100 mM NaCl, 5 mM MgCl₂, 300 µg BCIP (5-bromo-4-chloro-3-indolylphosphate) ml⁻¹ and 300 µg ml⁻¹ Nitro Blue Tetrazolium substrate (Bio-Rad). The relative level of GdpP-FLAG in each strain was normalized to the internal control FabF-FLAG using densitometry analysis with ImageJ (Girish & Vijayalakshmi, 2004).

Disc diffusion and MIC assays.

Disc diffusion assays were performed as described previously (Luo et al., 2010), with minor modifications. Mueller–Hinton (MH) broth (Sigma-Aldrich) and a MOPS-based glucose minimal medium (MM) (Bsat et al., 1996) were used for these assays. The bottom agar was 15 ml MH or MM broth supplemented with 1.5 % agar, and the top agar was 4 ml MH or MM broth supplemented with 0.75 % agar. Chemical discs were prepared using Whatman filter paper discs (7 mm diameter) and freshly made chemical stocks. For each chemical test, three different amounts were used: aztreonam 6, 30 and 60 µg; cefuroxime 3, 6 and 12 µg; cefixime 5, 10 and 15 µg; and sodium azide 5, 10 and 30 mmol. The zone of growth inhibition was measured after overnight growth at 37 °C. MIC tests were performed as described previously (Luo & Helmann, 2012), except that the media used were MH and MM broths.

Cell lysis assay.

Fresh colonies were first grown in LB, MH or MM broth to OD₆₀₀ 0.4, diluted 1 : 100 in fresh LB, MH or MM, respectively, and inoculated in Bioscreen microtitre plates with a total volume of 200 µl. Growth was measured spectrophotometrically (OD₆₀₀) using a Bioscreen incubator (Growth Curves USA) at 37 °C with vigorous shaking. For cefuroxime-induced lysis, five concentrations of cefuroxime (0.5, 1, 2, 4 or 8 µg ml⁻¹, final) were added to the cell culture at two growth phases (exponential phase at OD₆₀₀ 0.3–0.4 and transition phase at OD₆₀₀ 0.6–0.7).

Motility and biofilm formation.

Swimming and swarming motility tests were performed as described by Kearns & Losick (2005) and Patrick & Kearns (2009), with some modification. Freshly made LB agar was cooled to 55 °C, poured into Petri dishes (25 ml per plate) and dried in a laminar flow hood for 20 min before use. Cells were grown to mid-exponential phase (OD₆₀₀ 0.4) or transition phase (OD₆₀₀ 0.7), harvested and resuspended in phosphate buffer to OD₆₀₀ 10. Ten microlitres of cells was spotted on the swimming plate (LB supplemented with 0.3 % agar) or swarming plate (LB supplemented with 0.7 % agar and 5 µg tetrazolium violet ml⁻¹) and dried in a laminar flow hood for another 10 min. For swimming tests, the plates were incubated at 37 °C, and the swimming zones were measured every 30 min until they reached the edge of the plate (~5 h). For swarming tests, plates were incubated at room temperature (22 °C) overnight. Tetrazolium violet is a redox dye that is imported into the cell and reduced to a purple-coloured formazan. It does not affect cell growth or motility, but helps to mark the swarming zone. Biofilm formation tests were performed as described by Branda et al. (2006) and Romero et al. (2010). We tested pellicle formation in Msgg liquid medium and colony morphology on Msgg-supplemented 1.5 % agar plates.

Results and Discussion

An antisense transcript is encoded within gdpP

The gdpP gene (formerly yybT) is the second gene in a three-gene operon (yybS-gdpP-rplI) (Fig. 1a). Recent transcriptomic surveys using high-density cDNA tiling arrays and RNA-Seq have suggested that there are at least three transcripts emanating from this locus (Irnov et al., 2010; Nicolas et al., 2012; Rasmussen et al., 2009). The major transcript is initiated from a σ^A-dependent promoter located 121 bp upstream of yybS. Within the gdpP coding region, there is a putative σ^A-dependent promoter located upstream of the essential gene rplI (Nicolas et al., 2012). The third transcript is a putative antisense RNA encoded within gdpP. As this antisense transcript could potentially affect the expression of GdpP we investigated its expression and function.

Fig. 1.

A σ^D-dependent promoter in gdpP. (a) Schematic map of the yybS-gdpP-rplI operon. Promoter sites are indicated by bent arrows with a subscript to indicate the relevant holoenzyme. The −35, −10 and +1 elements of P_yybS and the start codon of yybS are in bold type. The balloon indicates a transcriptional terminator. The DNA regions included in the P_yybS-lacZ and P_gdpPas-lacZ fusions are also illustrated. (b) Promoter sequence of gdpP_as. The sense gdpP start codon and the −35, −10 and +1 elements of the antisense P_gdpPas are shown in bold type. P_D* mutations are underlined.

The antisense gdpP (gdpP_as) was first annotated as transcript shd124 by Rasmussen et al. (2009), and more recently as S1559 by Nicolas et al. (2012). However, assignment of the start site and length for this antisense RNA differed in those two studies. While Rasmussen and co-workers suggested a 1429 nt transcript starting from genome nucleotide position 4164574 (GenBank accession no. AL009126.3), Nicolas and co-workers assigned its start site to position 4164565 and predicted its length as 667 nt. To verify expression of this antisense transcript and define its transcriptional start site, we performed 5′-RACE with total RNA isolated from exponential phase cells grown in LB broth. Five pairs of primers were used to map any antisense transcripts initiated within a region of 800–1980 bp downstream of the gdpP start codon. One transcript, designated gdpP_as, was identified and its start site was mapped to 1048 bp downstream of the start codon, corresponding to the genome nucleotide location of 4164576. The start site of gdpP_as is proximal to a candidate σ^D promoter (Fig. 1) that is positionally conserved within the genomes of closely related bacilli (Fig. 2).

Fig. 2.

Alignment of P_gdpPas elements from various bacilli species. The gdpP sequence of B. subtilis 168 (Bsub) was aligned with the corresponding sequences from Bacillus atrophaeus 1942 (Batr), Bacillus amyloliquefaciens FZB42 (Bam), Bacillus licheniformis ATCC 14580 (Blic) and Bacillus pumilus SAFR-032 (Bpum) using clustalw2 (http://www.ebi.ac.uk/Tools/msa/clustalw2/). The WebLogo of the σ^D promoter consensus from B. subtilis is shown above the alignment (Sierro et al., 2008). The −35 and −10 elements are indicated.

We next tried to determine the size and abundance of the sense gdpP transcript and antisense gdpP_as transcript using Northern blotting and radioactively labelled oligonucleotide probes. However, we were unable to detect signals corresponding to either transcript (data not shown). It is possible that the expression levels of these transcripts are below the detection threshold, or perhaps more likely that the transcripts are unstable due to the formation of an RNA–RNA duplex. By examining the tiling data in detail, we noticed the transcript signals fading away after about 600 nt (Nicolas et al., 2012; Rasmussen et al., 2009). No terminators were found in the vicinity of this region, nor was an extended RNA transcript detected in RNA prepared from a rho mutant as monitored using tiling arrays (Nicolas et al., 2012). The termination mechanism for the antisense transcript therefore remains unknown. For simplicity reasons, here we adopt the annotation of 667 nt as suggested by Nicolas et al. (2012), which is a conservative estimate.

The gdpP_as transcript is dependent on σ^D

σ^D is an alternative σ factor that mainly regulates the expression of autolysins and genes involved in flagella biosynthesis and chemotaxis. The activity of σ^D is mediated primarily by the anti-σ factor FlgM, which sequesters σ^D and inhibits its activity. An flgM mutant displays increased sigD activity and hence increased expression of genes within the σ^D regulon (Caramori et al., 1996; Fredrick & Helmann, 1996). To determine whether the predicted gdpP_as promoter is active, we constructed an ectopic promoter–lacZ reporter (amyE : : P_gdpPas-lacZ) and introduced it into a WT strain (strain 168) and into isogenic strains possessing sigD and/or flgM mutations. This promoter fusion includes sequences from −230 to +60 bp relative to the transcription start site. The sense yybS promoter–lacZ fusion was also constructed and used as a control (Fig. 1a). Expression of P_gdpPas-lacZ was completely eliminated in a sigD mutant or a sigD flgM double mutant, but was highly induced in an flgM mutant (Fig. 3). In contrast, transcription of P_yybS-lacZ was not affected by mutations in either sigD or flgM. This result suggests that the antisense promoter is active and that it is σ^D-dependent.

Fig. 3.

β-Galactosidase activity of P_yybS-lacZ, P_gdpPas-lacZ and P_D*-lacZ in different strain backgrounds in LB overnight cultures. This experiment was performed in biological triplicate and repeated three times. Error bars, sem.

To further verify that the gdpP_as promoter activity is due to σ^D, we constructed a P_D*-lacZ fusion in which the −10 element GCCGACA sequence was changed to GCCTATG (mutated nucleotides underlined). These two nucleotide mutations alter two of the most conserved residues in the σ^D promoter consensus, yet maintain the sense GdpP amino acid sequence (Figs 1b and 2). As expected, P_D*-lacZ expressed no β-galactosidase in the WT strain or the sigD/flgM mutant backgrounds (Fig. 3).

gdpP_as transcription reduces the level of GdpP protein in the cell

As we could not detect any ORFs within the gdpP_as transcript, and as gdpP_as shares perfect complementarity with the sense gdpP mRNA, we hypothesized that gdpP_as functions as a cis-acting regulatory RNA. Base-pairing between a cis-acting RNA and its target mRNA typically alters target RNA stability and/or modulates sense translation (Georg & Hess, 2011), both of which alter expression levels of the sense-encoded protein. To measure the expression level of GdpP and the effects of gdpP_as transcription, we constructed an at-locus gdpP allele encoding an N-terminal FLAG-tag for immunodetection. This tagged allele of gdpP was introduced into WT and isogenic mutants of sigD, flgM and sigD flgM. Direct attempts to monitor GdpP-FLAG accumulation by Western blotting using crude cell lysates and anti-FLAG antibodies were unsuccessful, probably due to the low abundance of GdpP in all strain backgrounds. We therefore utilized immunoprecipitation to enrich GdpP-FLAG and then used Western blotting to quantify the recovered protein. As an internal control we used strains also expressing a FabF-FLAG protein (Kingston et al., 2011). Using this strategy, we determined that GdpP-FLAG accumulation increased in a sigD or a sigD flgM mutant, but decreased in an flgM mutant (relative to levels in the WT strain background) (Fig. 4). As σ^D activity is known to be quite heterogeneous in growing cell populations (Kearns & Losick, 2005), perhaps the most biologically relevant comparison of GfpP-FLAG accumulation is between the flgM mutant (wherein gdp_as is highly expressed) and the sigD mutant (wherein gdp_as is not expressed). We estimate an overall change in GfpP-FLAG accumulation of about 2.5- to threefold when comparing the sigD and flgM mutant strains. Thus, the gdpP_as transcript can serve as a negative regulator of GdpP expression.

Fig. 4.

Detection of GdpP-FLAG and FabF-FLAG by immunoprecipitation followed by SDS-PAGE and Western blotting with anti-FLAG antibodies. The sections of the membrane containing GdpP-FLAG and FabF-FLAG were separated after antibody hybridization, and developed separately using a chromogenic assay. Strains used were: lane 1, fabF-FLAG (HB13056); 2, gdpP-FLAG (HB15845); 3, gdpP-FLAG fabF-FLAG (HB15857); 4, sigD gdpP-FLAG fabF-FLAG (HB15858); 5, flgM gdpP-FLAG fabF-FLAG (HB15859); 6, sigD flgM gdpP-FLAG fabF-FLAG (HB15860). The lane between lanes 2 and 3 is a protein ladder, where the 75 and 37 kDa markers are visible at the sections of GdpP-FLAG and FabF-FLAG proteins, respectively. This experiment was repeated three times, and one representative experiment is shown. The intensities of GdpP-FLAG bands were normalized with the internal protein control FabF-FLAG. The numbers below each band represent the fold change (mean±sem) of normalized GdpP-FLAG relative to strain HB15857 (lane 3).

Lack of gdpP_as transcription does not lead to obvious cell lysis or motility phenotypes

We next investigated the possible biological effects of gdpP_as transcription by monitoring functions known or postulated to be related to GdpP function. Previously, we showed that c-di-AMP is involved in PG homeostasis (Luo & Helmann, 2012). A gdpP mutant (exhibiting increased c-di-AMP levels) is more resistant to PG-targeting antibiotics such as the β-lactam drugs and, conversely, overexpression of gdpP confers β-lactam sensitivity. As demonstrated above, transcription of gdpP_as is dependent on σ^D, and σ^D is known to regulate the expression of the major vegetative autolysins (LytC, LytD and LytF) which degrade PG (Blackman et al., 1998; Helmann et al., 1988; Lazarevic et al., 1992; Margot et al., 1994, 1999; Serizawa et al., 2004). We therefore hypothesized that σ^D downregulates GdpP expression (thereby leading to elevated c-di-AMP levels) as a mechanism to upregulate PG biosynthesis concomitant with the upregulation of autolysins under σ^D control. To test this idea, we mutated the gdpP_as promoter and generated an at-locus marker-less P_D* mutant, which eliminates σ^D-dependent gdpP_as transcription without affecting the expression of other σ^D regulon genes (Figs 1b and 3). We predicted that cells carrying P_D* would be unable to downregulate GdpP synthesis when σ^D is activated, and that this might lead to an imbalance between PG biosynthetic and autolytic functions. In an attempt to detect such an imbalance we treated both a WT strain (P_D^wt, strain 168) and the strain carrying the mutant promoter (P_D*, HB15837) with the cell autolysis inducers sodium azide and three β-lactam antibiotics (cefuroxime, aztreonam and cefixime) and monitored cell lysis rates by OD₆₀₀ and c.f.u. counts. We also examined chemical susceptibility by disc diffusion and MIC assays. However, the response differences between P_D^wt and P_D* strains were either not significant or not reproducible with large standard errors (data not shown). We reasoned that this is probably due to the temporary and heterogeneous nature of σ^D activity in the our laboratory strain (strain 168) background.

The heterogeneity of σ^D is due to multiple factors, including the magnitude of processive transcription through the long fla/che (sigD) operon, expression of the anti-σ factor FlgM, and transcriptional regulators SwrA, SwrB, DegU and SlrA/SinR/SlrR (Amati et al., 2004; Calvio et al., 2008; Cozy & Kearns, 2010; Cozy et al., 2012; Fredrick & Helmann, 1996; Hsueh et al., 2011; Kearns & Losick, 2005; Tsukahara & Ogura, 2008). As a result, a growing B. subtilis population contains both σ^D active (ON) and inactive (OFF) cells. The fraction of σ^D ON cells is influenced by growth phase, nutrient levels and the parental strain background: exponential phase, the presence of amino acids in the growth medium and domestic strains (e.g. strain 168) are associated with low σ^D activity, whereas transition phase, the absence of amino acids and the undomesticated strain (strain NCIB 3610) appear to have a relatively high fraction of σ^D ON cells. There is no known specific growth condition that can universally induce σ^D activity in a population. To test the effect of the P_D* mutation in a population with relatively high levels of σ^D activity, we conducted the same experiments using strain NCIB 3610 in a variety of different growth media (LB, MH and minimal medium) and growth phases (exponential, transition, early stationary and late stationary phases). However, no significant differences were observed between the P_D^wt and P_D* strains, at least with respect to cell autolysis and β-lactam susceptibility. It has been reported that a full σ^D ON population can be achieved by simultaneously mutating flgM and overexpressing sigD (Cozy & Kearns, 2010). In this case, however, the highly expressed autolysins in this strain background caused rapid cell lysis and may have masked the effect of gdpP_as, as no phenotypes were attributable to the P_D* mutation (data not shown). To separate the effects of loss of gdpP_as transcription from autolysin activity, we mutated all three major autolysins, lytC, lytD and lytF, in this σ^D ON strain background (flgM P_spank-sigD). Again, we did not observe differences between P_D* and P_D^wt strains in terms of cell lysis and β-lactam susceptibility. It is important to note that this autolysin-defective strain grows mainly as chains of un-separated cells, which may obscure the analysis of cell lysis using OD₆₀₀ as an indicator. We conclude that the detection of a gdpP_as phenotype may require other factors besides high cellular σ^D activity.

We next considered the possibility that modulation of GdpP levels by gdpP_as is most important under conditions where c-di-AMP synthesis is reduced. Therefore, we also introduced disA, sigM (encoding a disA activator) and ybbP mutations into NCIB 3610 strains carrying flgM and flgM P_D* mutations. These three mutations are known to reduce c-di-AMP levels and result in increased sensitivity to cefuroxime (Luo & Helmann, 2012). However, we could not detect any differences between P_D* and P_D^wt alleles in these strains in terms of either cell lysis rates or β-lactam susceptibility. Despite our various efforts, we were not able to associate gdpP_as with an observable phenotype. Nevertheless, the presence of a conserved, σ^D-dependent gdpP_as among closely related Bacillus species (Fig. 2) suggests that it likely plays a (still elusive) biological role.

Besides autolysis and β-lactam susceptibility, we also tested whether c-di-AMP affects cell motility for two reasons: (1) σ^D directs expression of genes involved in flagella biosynthesis and chemotaxis, and is therefore essential for cell motility; and (2) c-di-GMP, a related yet distinct secondary messenger molecule in bacteria, is known to regulate the transition between a motile and a sessile lifestyle, where cells with high c-di-GMP are non-motile and form robust biofilms (Hengge, 2009). c-di-GMP has not been detected in B. subtilis. However, a gdpP mutant (with increased c-di-AMP levels) in Staph. aureus has been reported to form about threefold more biofilm than the WT, which is reminiscent of the effect of c-di-GMP (Corrigan et al., 2011). To test cell motility and biofilm formation, strains harbouring mutations in the c-di-AMP cyclase genes (disA, ybbP and yojJ) or the hydrolase gene gdpP were constructed. An IPTG-inducible gdpP construct (amyE : : P_spac(hy)-gdpP) was introduced into these strains to reduce the level of c-di-AMP. We did not observe significant changes under these genetic conditions when compared with the WT strain NCIB 3610, suggesting that c-di-AMP may not be involved in cell motility (swimming and swarming) or biofilm formation, at least under our test conditions.

Concluding remarks

c-di-AMP is an essential second messenger in B. subtilis. Here, we show that the c-di-AMP hydrolase GdpP is subject to post-transcriptional regulation via an antisense RNA, gdpP_as, which thereby represents a novel means of regulating c-di-AMP levels in the cell. Antisense RNAs are a widespread regulatory mechanism and include both trans- and cis-acting RNAs. In both cases, the regulatory RNA typically acts by annealing to its target and affecting either translation or mRNA stability (Thomason & Storz, 2010). Recent genome-wide studies have revealed over 100 potential regulatory RNA transcripts in B. subtilis (Irnov et al., 2010; Nicolas et al., 2012; Rasmussen et al., 2009). Only a few antisense RNA molecules have been studied in detail, including three cis-acting (ratA, SR4 and antisense of yabE) and two trans-acting (fsrA and bsrF) sRNAs (Eiamphungporn & Helmann, 2009; Gaballa et al., 2008; Jahn et al., 2012; Preis et al., 2009; Silvaggi et al., 2005). Antisense RNA typically base-pairs with its target RNA, forming an RNA–RNA duplex, which alters the half-life of the target RNA, and therefore influences protein product accumulation. Transcription of gdpP_as is initiated in the middle of the gdpP gene and is dependent on σ^D. As demonstrated here, this cis-acting regulatory RNA can modulate the cellular levels of GdpP, although the biological role of this regulation remains elusive. Further studies are needed to better understand the regulatory network that links σ^D, the gdpP_as regulatory transcript, gdpP expression and c-di-AMP levels with effects on PG homeostasis.

Abbreviations:

c-di-AMP

cyclic-di-AMP

DAC

diadenylate cyclase

MLS

macrolide-lincomycin-streptogramin

PG

peptidoglycan

5′-RACE

5′ rapid amplification of cDNA ends

WT

wild-type