Background

Cyclodipeptides and their derivatives constitute a large class of secondary metabolites that are mainly synthesized by microorganisms. The biological role of theses peptide natural products is mostly unknown but such compounds are of interest due to their potential antibacterial, antifungal, antitumoral or antiinflammatory activities. Cyclodipeptides from Pseudomonas aeruginosa exhibit cytotoxic properties towards human tumor cells [1], albonoursin produced by Actinomyces tumemacerans present antibacterial activities [2–4] whereas mycocyclosin may be essential for Mycobacterium tuberculosis viability [5, 6]. Some yeasts and bacterial Bacillus species synthesize pulcherriminic acid that is derived from cyclo-L-leucyl-L-leucyl (cLL) [7–10]. Pulcherriminic acid is secreted in the growth medium and chelates ferric ions by a non-enzymatic reaction to form an extracellular red pigment, the pulcherrimin. In Metschnikowia pulcherrimina strains, pulcherrimin appears to be involved in their antagonistic effects on the other microorganisms by depletion of iron in the growth medium [11–13]. In Bacillus subtilis, the enzymes involved in pulcherriminic acid synthesis were more recently identified. The cyclodipeptide synthase YvmC utilizes charged leucyl tRNAs as substrate to catalyze the formation of cLL [14, 15]. The gene downstream of yvmC, cypX, encodes a cytochrome P450, which is implicated in transformation of cLL into pulcherriminic acid [16].

The extended bacterial MarR family of transcriptional regulators is involved in the regulation of various cellular processes such as regulation of response to antibiotics, aromatic compounds and oxidative stress agents [17–22]. They can also play a crucial role in virulence factor production, enabling the pathogenic bacteria to adapt to host environments [23–26]. The majority of MarR family members have been characterized as transcriptional repressors although a few examples of transcriptional activators have also been reported [27–29]. They function as homodimers and bind inverted repeat nucleic sequence [30–32]. A genomic locus generally consists of divergently oriented genes encoding the MarR homolog and the gene(s) under its control. By binding to a specific site in the intergenic regions between divergently transcribed genes, the MarR homolog represses transcription of both [32]. Another defining feature of MarR homologs is their response to specific ligands. In absence of ligand, specific DNA binding occurs, most often resulting in repression of the transcription, while binding of the ligand results in attenuated DNA binding and hence de-repressed gene expression [31]. For many MarR-type homologues, as well as for many transcription factors, the natural ligand is still unknown.

The objective of this study was to identify the mechanisms of regulation of the pulcherriminic acid biosynthetic pathway in the soil bacterium B. subtilis. We demonstrated that the uncharacterized B. subtilis YvmB MarR-type regulator is the major regulator controlling expression of the yvmCcypX operon. In combination with genetic approaches, gel mobility shift assays allowed us to define a palindromic motif that constitutes the YvmB-binding site. We also characterized the YvmB regulon, which is composed of four transcriptional units. In addition, a comprehensive quantitative proteomic analysis showed significant repercussions of yvmB deletion for proteins associated to various cellular processes such as metabolism, translation, stress response and biosynthesis of cell wall.

Results

Identification of YvmB as repressor of yvmC-cypX and yvmBA

Pairwise alignment of the amino acid sequence of YvmB with proteins from databases, reveals similarity in the secondary structures between YvmB and other characterized MarR regulators (Additional file 1: Figure S1).

The yvmB gene is divergently transcribed from the yvmC-cypX operon (Fig. 1) [33]. We observed that cells carrying a yvmB deletion overproduced a red pigment after entry into the stationary growth phase, in the iron-rich MS medium (Fig. 2a). To test whether YvmB controls yvmC-cypX expression, a transcriptional fusion between cypX and the lacZ reporter gene was constructed and studied in the wild-type and ΔyvmB cells (Table 1). In the wild-type, measurement of β-galactosidase activity revealed a slight increased of cypX expression during entry into stationary phase (Fig. 2b). In the ΔyvmB deletion mutant, cypX::lacZ was 100-fold up-regulated (Fig. 2c), suggesting that YvmB is the transcriptional repressor of yvmC-cypX operon expression. In addition, inactivation of the cypX gene in ΔyvmB cells abolished red pigment production confirming that the observed red color was probably due to the presence of pulcherrimin (Fig. 2a).

Fig. 1
figure 1

Involvement of the yvmC-cypX operon in pulcherriminic acid biosynthesis. a Genetic organisation of the yvnB, yvnA, yvmC-cypX and yvmBA genes in the B. subtilis genome. The function of the corresponding encoded proteins is indicated. b Scheme of the pulcherriminic acid biosynthetic pathway in B. subtilis. The cyclodipeptide synthase YvmC utilizes charged leucyl tRNAs as substrate to catalyze the formation of cyclo-L-leucyl-L-leucyl (cLL). The cytochrome P450 CypX is implicated in the transformation of cLL into pulcherriminic acid. Pulcherriminic acid is excreted from the cells by an unknown mechanism and chelates free Fe(III) by a non-enzymatic reaction to form an extracellular red pigment, the pulcherrimin

Full size image
Fig. 2
figure 2

Involvement of YvmB as repressor in the control of yvmC-cypX expression. a The strains were cultivated in MS medium containing 100 μM FeCl3 for 16 h. at 37 °C in flasks with continuous agitation at 200 rpm. BSB1 wild-type strain; BSAS82 ΔyvmB::aphA3; BSAS147 ΔyvmB::aphA3 cypX::lacZ. Cultures were transferred in tubes for the pictures. b and c Expression of a cypX::lacZ transcriptional fusion was compared in wild-type and ΔyvmB cells. Strains were grown in LB medium. Growth was monitored by measuring OD600: grey curves. β-galactosidase activity of the cypX::lacZ fusion is indicated: black circles, in the wild-type (strain BFA815); red squares, in ΔyvmB cells (strain BSAS147). Values are the means from at least four independent experiments

Full size image
Table 1 Bacillus subtilis strains used in this study
Full size table

The yvmB gene is coexpressed with yvmA leading to a yvmBA transcript [33]. To ask whether yvmB might be subject to autoregulation like other MarR-type repressors, transcription analyses were performed. A transcriptional fusion between the yvmB promoter region (from nucleotide −291 to +9 relative to the translational start site) and the lacZ reporter gene was constructed and integrated into the amyE locus of the wild-type and ΔyvmB cells. Analysis of the β-galactosidase activity revealed that expression of yvmB increased during entry into stationary phase (Fig. 3a). The 30-fold up-regulation of yvmB-promoter-lacZ acitivity in ΔyvmB cells indicated that yvmB is subject to autorepression (Fig. 3b).

Fig. 3
figure 3

Involvement of YvmB CcpA as repressors in the control of yvmB expression. a, b and c) Expression of a pyvmB’-lacZ transcriptional fusion was compared in wild-type (strain BSAS108), ΔyvmB cells (strain BSAS109) and ΔccpA cells (strain BSAS153). Strains were grown in LB medium. Growth was monitored by measuring OD600: grey curves. β-galactosidase activity of the pyvmB’-lacZ fusion is indicated: black circles, in the wild-type; red squares, in ΔyvmB cells; blue triangles, in ΔccpA cells. Values are the means from at least four independent experiments

Full size image

Expression of yvmB is repressed by CcpA in stationary growth phase

Expression of yvmB was proposed to be subject to repression by glucose by direct binding of the carbon metabolism regulator CcpA to a cre site, which overlaps the −35 region [34]. Therefore, we investigated expression of the yvmB-promoter-lacZ fusion in a ccpA mutant. Expression of yvmB was up-regulated in ΔccpA cells upon the stationary growth phase (Fig. 3c). This result confirmed that CcpA acts as negative regulator of yvmB expression. Therefore, yvmB expression undergoes a dual regulation by YvmB and CcpA depending on growth phase.

Transcription of yvmB is induced by iron starvation

Pulcherriminic acid is known to be secreted in the growth medium and to chelate ferric ions. Here, we examined the effect of iron starvation on cypX::lacZ expression. The BFS815 strain (Table 1) was cultivated in LB medium and a sample was spread onto solid medium containing X-gal. A drop of 10 mM bipyridyl, a specific iron chelator, was deposited at the center of the plate. After incubation, a white ring was observed around the bipyridyl drop (Additional file 2: Figure S2) which indicates that iron starvation does not induce yvmC-cypX expression. We further tested the effect of iron starvation on the yvmB-promoter-lacZ fusion (strain BSAS108) using the same experimental procedure. Remarkably, a characteristic blue ring was observed around the bipyridyl drop (Additional file 2: Figure S2) indicating that iron deficiency induces yvmB expression.

Effect of yvmB deletion on B. subtilis proteome

To understand the physiological state of cells overproducing pulcherriminic acid, we performed a comparative analysis of the cytosolic and membrane proteome of the wild-type and ΔyvmB cells. The optimized analyses of four technical replicates resolved more than 1700 proteins (Additional file 3: Table S1). Only the statistically significant (p < 0.05; Kruskal-Wallis test and one way ANOVA) differentially expressed proteins were considered (Additional file 4: Table S2). Among the 32 differentially abundant proteins (17 up-regulated and 15 down-regulated), 4 were common between the membrane and the cytosolic fractions. The results are summarized in the Table 2.

Table 2 Proteins up- or down-regulated in a B. subtilis yvmB mutant
Full size table

Proteomic analysis of a yvmB mutant confirmed a 24- and 13-fold enhanced production of the YvmB primary targets, CypX and YvmA. It also revealed a wide and prominent effect on proteins associated with carbon metabolism, translation, stress response and biosynthesis of cell wall. Remarkably, several of these proteins have a function related to metal ions metabolism. Shifts in metabolic pathways particularly occur in response to iron availability [35, 36]. CitB and CitZ are down-regulated in a yvmB mutant. Both proteins are known to be less synthesized under conditions of iron limitation [37–39]. Moreover, the down-regulated ribosomal proteins RplL and RpsH belong to the stringent response, which could be initiated by iron limitation [40]. Iron participates in many metabolic processes related to electron transfer and redox state of the cell, as well as to pH homeostasis [36, 41]. Interestingly, proteins involved in FeS cluster assembly and in respiration, like SufD and Ndh, respectively, are more abundant in a yvmB mutant. In the same line, a yvmB mutant shows increased level of the ferredoxin-dependent reductase YumC, which is involved in electron transport. The up-regulated protein KatA is part of the iron-dependent peroxide stress response mediated by PerR, which is activated when aerobically growing cells enter the stationary phase [42]. Most of the PerR regulon was shown to be repressed when cells are exposed to high levels of iron [36]. Up-regulated MurAA, YkuQ and PgcA are related to cell wall synthesis and lipid biosynthesis. Membrane structure remodeling can result from interaction of ferric ion with external membrane phospholipids bilayer [43–45]. Therefore, deletion of yvmB induces disruptions of proteins involved in cellular processes that depend on iron availability.

Determination of the YvmB recognition site in the yvmC promoter

In order to identify genes directly regulated by YvmB, we first attempted to identify the YvmB DNA-binding site. The DNA sequence of the yvmC promoter region is presented in Fig. 4. The −10 (TAAAAT) and −35 (TTGACG) σA-dependent regions are indicated according to the genome reannotation data [33]. To identify the DNA regions involved in yvmC regulation, four different parts of the promoter region of the yvmC gene were each fused with the lacZ reporter gene and integrated at the amyE locus of the wild-type strain (Fig. 5a) (Table 1). β-galactosidase activities were measured during growth in LB medium (Fig. 5b). The pAyvmC and pCyvmC fusions were 4-fold more expressed in ΔyvmB cells than in the wild-type. The absence of pByvmC expression was consistent with the position of the −10 and −35 σA-dependent regions. Remarkably, a 42-bp palindromic sequence lies between nucleotides −189 to −149 relative to the yvmC translational start site (Fig. 5c) and was called palindrome I. Three point mutations were introduced in this palindromic sequence. The A, C and A nucleotides at position −162, −163 and −164 were replaced by three nucleotides G (Fig. 5c). The resulting pA* fusion was constitutively expressed in the wild-type strain (Fig. 5b), indicating that palindrome I is the major cis-regulatory sequence involved in yvmC repression by YvmB.

Fig. 4
figure 4

Promoter region of the yvmC-yvmB divergent genes. The −10 and −35 regions of yvmC (in purple) and the −10 and −35 regions of yvmB (in brown) are indicated according to Nicolas et al. [33]. The potential RBS (Ribosomal Binding Site) of the yvmB gene is indicated. The bold letters correspond to the cre site. The blue arrows and the blue letters indicate the palindromic sequence I. The green arrows and the green letters indicate the palindromic sequence II

Full size image
Fig. 5
figure 5

Expression of various yvmC promoter lacZ fusions in wild-type and ΔyvmB strains. a Structure of the yvmC operator region. The −10 and −35 regions similar to the consensus sequence of σA-dependent promoters are indicated: white boxes. Location of the palindromic sequence, palindrome I, is indicated in blue. The DNA fragments used to construct pyvmC’-lacZ transcriptional fusions are diagrammed. They are numbered with respect to the translational start site of yvmC. Stars indicated the location of three point mutations introduced in the pA*yvmC’-lacZ fusion. The −10 and −35 regions located in the yvmB promoter region are indicated: grey boxes. b Cells were grown in LB medium. Growth was monitored by measuring OD600: grey line, wild-type; purple line, ΔyvmB. Expression of the four pyvmC’-lacZ transcriptional fusions was measured: black circles, in the wild-type; red squares, in ΔyvmB cells. Values are the means from at least three independent experiments. c Nucleotide sequence of the palindromic motif (blue arrows), palindrome I. The nucleotides in bold belong to the inverted repeat nucleic sequence. The A, C and A nucleotides at position −162, −163 and −164 (relative to the yvmC translational start site) were replaced by three nucleotides G to construct the pA*yvmC’-lacZ fusion

Full size image

Detection of several regions involved in the control of yvmB by YvmB

The DNA sequence of the yvmB promoter region is shown in Fig. 4. The −10 (TAAAAT) and −35 (TTGACA) σA-dependent regions are indicated [33]. To determine the regions involved in yvmB regulation, strains carrying transcriptional luc fusions using different promoter sub-regions of yvmB at the amyE locus were constructed (Fig. 6a) (Table 1). Expression was very low for the large pAyvmB fusion (Fig. 6b). The pByvmB and pCyvmB fusions were 6- and 5-fold up-regulated during entry in the stationary phase indicating that the −291 to - 228 region was involved in yvmB repression. This corresponded to the location of palindrome I identified above, which lies from nucleotides −278 to −236 relative to the yvmB translational start site (Fig. 4). Thus, YvmB is involved in the control of its own transcription via palindrome I.

Fig. 6
figure 6

Expression of various yvmB promoter luc fusions. a Structure of the yvmB operator region. The −10 and −35 regions similar to the consensus sequence of σA-dependent promoters are indicated: white boxes. The location of the cre site (Blencke et al. [34]) is indicated: grey box. Location of the palindromic sequence, palindrome I, is indicated in blue. Location of the palindromic sequence, palindrome II, is indicated in green. The different fusions with the luc gene are numbered with respect to the translational start site of yvmB. The −10 and −35 regions located in the yvmC promoter region are indicated: grey boxes. b Cells were grown in LB medium. Growth was monitored by measuring OD600: grey lines. Expression of the six pyvmB’-luc transcriptional fusions was measured in the wild-type: black lines. For each strain, one representative curve, out of three independent replicates realized, is shown. Note that the X-axe is different for the pF fusion to get a better view of the data. c Nucleotide sequence of the palindromic motif (green arrows), palindrome II. The nucleotides in bold belong to an inverted repeat nucleic sequence

Full size image

A second set of luc fusions was constructed. Expression of the pDyvmB fusion was not detected whereas the pEyvmB fusion was 10-fold up-regulated in entry to stationary phase (Fig. 6b). Therefore, the −60 to −23 region was also involved in yvmB repression. Remarkably, this region overlaps a 60-bp imperfect palindromic sequence that we named palindrome II (Fig. 6c). The short pFyvmB fusion, without palindrome I nor palindrome II, was fully depressed in entry to stationary phase (Fig. 6b) showing a cumulative repressive effect of palindromes I and II. Remarkably, the pFyvmB fusion yielded a very high and transient induction probably due to the repressive effect of CcpA in stationary phase, as shown above (Fig. 3c).

Definition of a YvmB box consensus and prediction of YvmB boxes within the whole-genome sequence

Browsing the genome sequence of A-T-rich Firmicutes spp., we detected YvmA, YvmC and CypX-like proteins in the genome of Bacillus licheniformis ATCC 14580. A sequence similar to palindrome I is present upstream of yvmA (Additional file 5: Figure S3). Inspection of the promoter regions of B. subtilis yvmB and B. licheniformis yvmA genes revealed conservation of a 14-bp core region containing an inverted repeat (Fig. 7a). We showed above that mutation of the conserved nucleotides in position 13 and 14 resulted in full up-regulation of yvmC expression (Fig. 5). This motif could be viewed as the YvmB binding site (hereafter called the YvmB box). Indeed, three YvmB boxes were identified in the yvmB promoter region: (i) one in palindrome I; (ii) one in palindrome II; (iii) the third one, 41 bp upstream of the yvmB translational start site (Fig. 7a). This allowed us to propose a first YvmB box consensus sequence, GTTYMMYMGTAAAC.

Fig. 7
figure 7

Identification of a 14 bp YvmB-box consensus sequence. a Alignment of potential 14 bp YvmB box motifs identified in the promoter region of the B. subtilis yvmB gene and in the promoter region of the B. licheniformis yvmA gene. Position of the boxes are indicated relative to the translational start sites. b Identification of additional putative YvmB boxes. Position of the boxes are indicated relative to the translational start sites: yvnB; similar to metallophosphatase; yisI, Spo0A-P phosphatase; yxnA, similar to glucose 1-dehydrogenase; ndhF, NADH dehydrogenase (subunit 5); yckD, unknown. c The 14-nt palindromic consensus sequence of the YvmB box motif. The size of the nucleotides at each position indicates its relative prevalence in sequences used as training set for MEME algorithm

Full size image

We used this YvmB box consensus to search putative YvmB boxes within the B. subtilis whole-genome sequence, less than 500 bp upstream of a translational start and in a define promoter region according to Nicolas et al. [33]. This allowed the identification of 5 highly similar motifs preceding the yckD, yisI, yvnB, yxnA and ndhF genes (Fig. 7b). The corresponding proteins were not detected in our proteomic analysis. However, we assumed that YvmB could be involved in the transcriptional regulation of these 5 genes. By contrast, no evident YvmB box-like motif was detected upstream of the transcriptional units encoding proteins differentially abundant in the yvmB mutant suggesting an indirect effect of yvmB deletion.

Interaction of YvmB with the yvmB, yvnB and yisI promoter regions

His-tagged YvmB protein was overproduced in E. coli and purified in a single step by Ni-chelate affinity chromatography to greater than 95 % homogeneity, as revealed by SDS-PAGE (Additional file 6: Figure S4). A sample of the purified protein was fractionated by chromatography on a gel-filtration column. The YvmBHis-tag protein eluted as a major peak with an apparent molecular mass of 45 ± 5 kDa (Additional file 6: Figure S4). As the YvmBHis-tag polypeptide has a molecular mass of 19.2 kDa, this means that the native protein is most likely a dimer.

Ability of YvmBHis-tag to bind to the yvmB-yvmC intergenic region was tested by gel retardation assays using three different DNA fragments as probes (Additional file 7: Table S3). The P1 fragment contains palindrome I, the P2 fragment contains palindrome II while the large P3 fragment covers the entire DNA region located between the yvmB and yvmC genes (Fig. 8a). One YvmB-DNA complex, with an apparent dissociation constant of 5 nM, was detected with P1 probe (Fig. 8b) in agreement with prediction of one YvmB recognition motif in this region. Two YvmB-DNA complexes were detected with P2 probe (Fig. 8b) in agreement with detection of two YvmB box motifs. Finally, gel retardation assays with P3 probe revealed three YvmB-DNA complexes, confirming the presence of three YvmB binding sites (Fig. 8b).

Fig. 8
figure 8

YvmB binding to the yvmB-yvmC intergenic region and to the yvnB and yisI promoter regions. a Location of the three identified YvmB boxes in the promoter region of yvmB. The −10 and −35 regions similar to the consensus sequence of σA-dependent promoters are indicated: white boxes. Location of the palindromic sequence, palindrome I, is indicated in blue. Location of the palindromic sequence, palindrome II, is indicated in green. Location of the three YvmB boxes is indicated in red. The DNA fragments P1, P2 and P3 used to perform gel shift assays are diagrammed. The DNA probes used were: (b) probe P1, probe P2 and probe P3; (c) probe P-yvnB and (d) prove P-yisI. Increasing concentrations of purified YvmBHis-tag were used (nM)

Full size image

Gel shift assays were also performed with a DNA fragment corresponding to the yvnB, yisI, yxnA, ndhF, yckD promoter regions (Additional file 7: Table S3). Fragments (110 bp long) carrying putative YvmB boxes in the centre were subjected to gel retardation analysis with YvmBHis-tag. No interaction was detected with probes corresponding to yxnA, ndhF and yckD promoter regions (data not shown). Binding of YvmBHis-tag to the P-yvnB and P-yisI probes resulted in shifts, for which we determined the apparent dissociation constant at 25 nM and 2.5 nM, respectively (Fig. 8c and d). It should be noted that a minor second protein-DNA complex formed in the presence of 100 nM YvmBHis-tag and probe P-yvnB. This could be due to the presence of a degenerated YvmB motif (GTTTACTGGGTAAT) located 119 bp upstream of yvnB. To conclude, the predicted YvmB box upstream of yvnB and yisI appears functional for YvmB binding in vitro. We assumed that YvmB could be involved in the transcriptional regulation of yisI and yvnB.

Dual role of YvmB as repressor of yvnB and as activator of yisI

The DNA sequence of the yvnB and yisI promoter regions is shown in Additional file 8: Figure S5. The yvnB gene is located less than 1 kb upstream of the yvmCcypX operon (Fig. 1). We then tested the correlation between the presence of a YvmB box and YvmB dependent expression. For this purpose, we constructed a transcriptional fusion between the promoter region of yvnB and the luciferase gene in wild-type and ΔyvmB cells. Accordingly to Nicolas et al. [33], yvnB gene is highly expressed during sporulation condition. Luciferase activity was recorded during growth in standard LB medium and sporulation medium. Expression of the pyvnB fusion showed 10-fold up-regulation in ΔyvmB cells in LB and DSM media (Fig. 9a and b). This result validated the YvmB-dependent repression of the yvnB gene.

Fig. 9
figure 9

Involvement of YvmB in the control of yvnB and yisI expression. Growth was monitored by measuring OD600 every 5 min: grey line, wild-type; purple line, ΔyvmB. a Promoter activity (RLU/OD) of the transcriptional fusion yvnB’-luc in LB medium is indicated: black line, wild-type; red line, ΔyvmB. b Promoter activity (RLU/OD) of the transcriptional fusion yvnB’-luc in DSM sprorulation medium is indicated: black line, wild-type; red line, ΔyvmB. c Promoter activity (RLU/OD) of the transcriptional fusion yisI’-luc in LB medium is indicated: black line, wild-type; red line, ΔyvmB. d Promoter activity (RLU/OD) of the transcriptional fusion yisI’-luc in DSM sporulation medium is indicated: black line, wild-type; red line, ΔyvmB. One representative curve, out of three independent replicates realized, is shown

Full size image

We also tested the activity of YvmB as transcriptional regulator of the yisI gene, which encodes a protein involved in the sporulation process. We constructed a transcriptional fusion between the promoter region of yisI and the luciferase gene in wild-type and ΔyvmB cells. Expression of the yisI fusion is 2-fold down-regulated in ΔyvmB cells in the stationary growth phase in LB and DSM media (Fig. 9c and d). Thus, YvmB positively regulates yisI expression.

Subsequently, we used the MEME standard bioinformatic method [46] to redefine the YvmB box consensus from the YvmB box sequences upstream of the yvmB, yvmC, yvnB and yisI promoter regions. The resulting new YvmB consensus sequence exhibited an inverted repeat sequence with highly conserved nucleotides at the external positions 1, 2, 3, 12, 13 and 14 and in the mid-positions 5 and 6 (Fig. 7c).

Discussion

In this study, we identified the B. subtilis YvmB uncharacterized MarR-like protein as the main negative regulator of the pulcherriminic acid biosynthetic pathway. To our knowledge, this is the first report of the molecular and genetic mechanisms involved in the regulation of this pathway in microorganisms. Our genetic analysis, gel-shift assays and computational analysis highlight a 14-bp core perfect inverted repeat, which likely defines the YvmB binding site in the yvmC-cypX operator region. This motif overlaps the −10 promoter element of the yvmC target gene, as previously described for MarR-type regulators DNA-binding sites [32]. In addition, our work led to the identification of the YvmB regulon, which extends to yvmBA (divergent from yvmC), yvnB and yisI genes. A sequence analysis of the YvmB boxes using the MEME suite produced a consensus composed of highly conserved nucleotides at the external positions 1, 2, 3, 12, 13 and 14 and at the mid-positions 5 and 6 (Fig. 7c). As these nucleotides are also present in the unbound YvmB boxes preceding the yckD, yxnA and ndhF genes, the local sequence environment might also be important for the specific DNA binding of YvmB.

The regulation of yvmB expression is complex since the activity of its promoter is modulated by at least two transcriptional regulators, CcpA and YvmB. CcpA seems to act as repressor of yvmB expression in the stationary growth phase while YvmB represses its own expression. Autoregulation is a characteristic of the MarR-like repressor [32]. We showed that YvmB binds specifically to three regions containing a YvmB box. The three YvmB boxes play a cumulative effect in yvmB repression. In addition, preliminary analysis of the purified YvmBHis-tag protein by analytical chromatography suggests that YvmB is a dimer in the absence of the effector. The nature of the metabolite controlling YvmB repressor activity by binding to DNA remains to be clarified. Remarkably, expression of yvmB is also up-regulated in response to iron starvation. Our results support the existence of an auxiliary transcriptional mechanism, independent of YvmB, to control yvmB expression in response to iron availability.

An unexpected finding is the activation of the yisI gene by YvmB during entry into the stationary growth phase. To our knowledge there are few examples of MarR-like regulator showing a dual role as a repressor and an activator [29]. Remarkably, the YisI protein is an aspartyl-phosphate phosphatase, which specifically dephosphorylates the sporulation transcription factor Spo0A-P and negatively regulates the sporulation initiation pathway in order to control the proper timing of sporulation [47]. It is known that the yisI gene is expressed mainly during the transition phase between exponential and stationary phase, its induction is dependent upon the presence of Spo0A [47]. By our proteomic analysis, two proteins of the Spo0A regulon are induced in a yvmB mutant. These are represented by the RocD ornithine aminotransferase and the Tkt transketolase, which might be due to a difference in phosphorylated/unphosphorylated form of the Spo0A regulator in a yvmB mutant. Therefore, YvmB appears an additional regulatory element into the cell’s decision to grow or sporulate. A database search revealed that yvmBA, yvmC-cypX and yisI genes are present in B. licheniformis and Bacillus thuringiensis, suggesting the functional conservation of the role of YvmB in pulcherriminic acid synthesis and in sporulation in other Bacilli.

The exact role of pulcherriminic acid produced by B. subtilis is still unclear. Expression of yvmC-cypX is not up-regulated in response to iron deficiency in agreement with Kupfer et al., who showed that similar amounts of pulcherriminic acid were detected in the presence of both low and high concentrations of ferric ions [48]. Since pulcherrimin is a large nondiffusible complex [8, 9, 49], mechanism of iron depletion is different from the mechanisms operating in microbes that release siderophores into the environment for iron acquisition. Our comprehensive quantitative proteomic analysis revealed that yvmB deletion induces changes of proteins involved in cellular processes depending on iron availability. However, we did not identify evident YvmB-box motifs upstream of the transcriptional units encoding these proteins. We assume that chelation of external iron by pulcherriminic acid could disrupt iron-dependent processes that can explain the wide indirect cellular changes into ΔyvmB cells.

Here, we demonstrated that YvmB controls expression of the yvmA and yvnB genes, which encodes a putative permease and a putative metallophosphatase, respectively. In M. pulcherrimina, pulcherrimin is involved in antimicrobial effects by depletion of iron in the growth medium [11–13]. M. pulcherrimina also secretes lytic enzymes such as chitinase that may contribute to the overall antagonistic effects of related M. pulcherrimina strains [50]. The role of YvmA and YvnB in potential antagonist effects of pulcherrimin produced by B. subtilis deserve further studies.

Conclusions

This study allowed to identify the mechanism of regulation of the genes involved in the biosynthesis of pulcherriminic acid in B. subtilis. We identified the uncharacterized YvmB MarR-like protein as the main regulator of the yvmCcypX operon. Additionally, we defined the YvmB regulon comprising at least four transcriptional units. Our results also highlight a complex impact of YvmB on cellular processes depending on iron availability and on the regulation of the initiation of sporulation. Further studies will be required to investigate the role of the genes controlled by YvmB.

Methods

Bacterial strains and growth conditions

The B. subtilis strains used in this work are listed in Table 1. B. subtilis strains are derivatives of the BaSysBio reference strain BSB1 which is a tryptophan-prototrophic (trp+) 168 strain [33]. Escherichia coli strains used were TG1 (Lab strain) for plasmid construction and ER2566 (New England Biolabs) for protein expression and purification. B. subtilis cells were grown in Luria-Bertani (LB) medium or in MS medium containing 62 mM K2HPO4, 44 mM KH2PO4, 17 mM trisodium citrate, 11 mM K2SO4, 0.4 % glucose, 0.06 % L-glutamine, 0.01 % L-tryptophane, 0.1 % casamino acids, 2 mM MgSO4, 1 mM CaCl2, 100 μM FeCl3 citrate, 112 μM ZnCl2; 5 μM MnCl2; 2.5 μM CuCl2. E. coli cells were propagated in LB medium. Antibiotics were added at the following concentrations when required: 100 μg ampicillin ml−1; 5 μg chloramphenicol ml−1; 100 μg spectinomycin ml−1; 5 μg kanamycin ml−1; 10 μg erythromycin ml−1. Solid media were prepared by addition of 20 g Agar noble l−1 (Difco). Standard procedures were used to transform E. coli [51] and B. subtilis [52].

The loss of amylase activity was detected as described by Stülke et al. [53]. β-galactosidase specific activity was measured as described by Miller [54] with cell extracts obtained by lysozyme treatment. One unit of β-galactosidase activity was defined as the amount of enzyme that produces 1 nmol o-nitrophenol min−1 at 28 °C. The mean values of three independent experiments are presented. Standard deviations were less than 20 % of the mean.

DNA manipulations

DNA manipulations and cloning procedures were performed as described elsewhere [51]. Restriction enzymes, Pfu DNA polymerase and phage T4 DNA ligase were used as recommended by the manufacturer (Biolabs). DNA fragments were purified from agarose gels using the QIAquick kit (Qiagen).

Construction of plasmids and strains

The BSAS82 yvmB mutant was constructed by homologous replacement of the YvmB coding sequence with a kanamycin resistance gene (aphA3) using a joining PCR technique [55]. The aphA3 gene was first amplified. The region upstream of the yvmB gene (nucleotide −990 to +52 relative to the translational start site) was amplified by PCR with a 24 bp aphA3 fragment at its 3′ end. The region downstream of yvmB (nucleotides +421 to +1495) was amplified with a 24 bp aphA3 fragment at its 5′ end. The three DNA fragments were combined and then a PCR reaction was performed with the two external oligonucleotides. The final product, corresponding to the two regions flanking yvmB with the inserted aphA3 cassette in between, was purified from a gel and used to transform B. subtilis. Integration and deletion were confirmed by PCR and verified by DNA sequencing.

The inactivation of the cypX locus while creating a lacZ fusion (strain BFS815) was done within the framework of a European project on the functional analysis of the genome of B. subtilis (see http://genome.jouy.inra.fr/cgi-bin/micado/index.cgi).

Plasmid pAC6 [53] allowed the construction of a transcriptional fusion between the promoter region of yvmB (nucleotides −291 to +9 relative to the translational start site) and the promoterless lacZ gene. The yvmB promoter region was amplified by PCR, with the creation of EcoRI and BamHI sites. The PCR product was inserted into pAC6. The resulting plasmid was linearized with ScaI, which allowed the insertion of the transcriptional lacZ fusion as a single copy at the amyE locus (Table 1). The same procedure was used to construct a series of transcriptional fusions between the promoter region of yvmC and the lacZ gene.

To construct transcriptional fusions between the promoter region of yvmB, yvnB and yisI and the luc reporter gene, we used the assembly Gibson’s procedure [56] as previously described [57]. The PUC18cm-luc plasmid was used as template to amplify the luc reporter gene. The sequences of the resulting constructs were verified by DNA sequencing.

Luciferase assay

For the detection of luciferase activity, strains were first grown in LB medium to an optical density at 600 nm (OD600) of 2. Cells were then centrifuged and resuspended in fresh minimal medium, adjusting all the cultures to an OD600 of 2. These pre-cultures were then diluted 20 fold in fresh minimal medium and 200 μl was distributed in each of two wells in a 96-well black plate (Corning). 10 μl of luciferin were added to each well to reach a final concentration of 1.5 mg/ml (4.7 mM). The cultures were incubated at 37 °C with agitation in a PerkinElmer Envision 2104 Multilabel Reader equipped with an enhanced sensitivity photomultiplier for luminometry. The temperature of the clear plastic lid was maintained at 38 °C to avoid condensation. Relative Luminescence Unit (RLU) and OD600 were measured at 5 min intervals.

Protein extraction

Four independent cultures of each B. subtilis strains were grown in 500 mL of LB medium to an OD600 of 0.8. Cultures were harvested by centrifugation (6000 × g, 10 min, 4 °C) and washed twice with a cold low-salt buffer (150 mM NaCl, 10 mM Tris, 0.5 mM EDTA, pH 7.5). Resuspended bacteria were mechanically disrupted by a passage through a One Shot Cell Disrupter (Constant Systems Ltd., Warwickshire, UK) at 2.7 Kbar. Cell lysates were centrifuged (27 000 × g, 20 min, 4 °C) and resulting supernatants were treated to ultracentrifugation (100 000 × g, 1 h, 4 °C). Cytoplasmic fractions were considered as the soluble parts after a single ultracentrifugation step, while the remaining pellets were designated as crude membrane fraction. After resuspension in Bis-Tris NaCl buffer (50 mM bis-tris, 50 mM NaCl, pH7.2) followed by ultra-centrifugation (100 000 × g, 1 h, 4 °C), pellets were resuspended overnight in 150 μL resuspension buffer (20 mM Tris–HCl, 10 mM EDTA, pH7.5) and then added of 1 volume of SDS-PAGE sample buffer (125 mM Tris–HCl pH6.8, 20 % v/v glycerol, 10 % w/v SDS, 5 % β-mercaptoethanol). Protein concentration was determined by NanoDrop. All the samples were loaded on a 10 % NuPAGE Bis-Tris Gels (Invitrogen).

Tryptic digestion in gel

All Coomassie-stained protein sample detected on SDS-PAGE were cut in small pieces and washed twice with 0.2 % TFA-50 % acetonitrile. The protein disulfide bridges were reduced by DTT 10 mM for one hour at 56 °C and resulting reduced cysteine were alkylated by iodoacetamide (50 mM) for 1 h at room temperature into darkness before the step of enzymatic digestion performed by adding 500 ng of sequencing grade modified trypsin (Promega) diluted in 25 mM NH4HCO3 for 18 h at 37 °C for each protein sample. Tryptic peptides were recovered by washing the gel pieces twice in 0.2 % TFA-50 % acetonitrile and once in 100 % acetonitrile and the supernatant was evaporated to dryness. The peptides were then resuspended in 25 μL of precolumn loading buffer (0.05 % trifluoroacetic acid (TFA) and 5 % acetonitrile (ACN) in H2O), prior to LC-MS/MS analysis.

Peptide and protein identification

LC-MS/MS analysis was performed NanoLC-Ultra Eksigent (SCIEX) system connected to Q-Exactive mass spectrometer (ThermoFisher) by nanoelectrospray ion source. Tryptic peptide mixtures (4 μl) were loaded at flow rate 7.5 μl min−1 onto precolumn Biosphere C18, 5 μm, 20 mm, 100 μm i.d.; NanoSeparations, Nieuwkoop, NL). After 3 min, the precolumn was connected with the separating nanocolumn PepMap100C18, 3 μm, 500 mm, 75 μm i.d.; Dionex), and the linear gradient was started from 5 to 35 % of buffer B (0.1 % formic acid, 100 % acetonitrile) in buffer A (0.1 % formic acid, 5 % acetonitrile) at 300 nl min−1 during 120 min for a running time of 130 min with washing and equilibration steps. Ionization was performed on liquid junction with a spray voltage of 1.5 kV applied to an uncoated capillary probe (PicoTip EMITER 10-μm tip inner diameter; New Objective). Peptides ions were automatically analyzed in positive mode by the data-dependent method as follows: full MS scan (m/z 400 to 1400, resolution 70,000 at m/z 400) and MS/MS on the 12 most abundant precursors (resolution 17,500 at m/z 400). In the present study only +2 and +3 charged peptides were subjected to MS/MS experiments with an exclusion window of 40 sec, in HCD fragmentation mode with a normalized collision energy fixed to 26 %. The lock mass option was activated on m/z dimethylcyclosiloxan 445.12003.

The raw data produced on Q-Exactive mass spectrometer were first converted in mzXML file with msconvert (http://proteowizard.sourceforge.net, version 3.0.3706), and protein identification was performed with X!Tandem software (X! Tandem Sledgehammer 2013.09.01.1; http://www.thegpm.org) against a protein database of B. subtilis 168 (downloaded to ftp://ftp.ebi.ac.uk/pub/databases/integr8/last_release/fasta/proteomes contain 4253 proteins associated to a classical proteomic contaminant database). The X!Tandem search parameters were as follows: trypsin specificity with two missed cleavage, fixed alkylation of cysteine (+57.0215), and variable oxidation of methionine (+15.9949). The mass tolerance was fixed to 10 ppm for precursor ions and 0.02 Da for fragment ions. For all proteins identified with a protein E-value of < 0.01 in the first step, we searched for additional peptides to reinforce identification using similar parameters except that semitryptic peptides and protein N-terminal acetylations were accepted. All results for each analysis were merged with an homemade program written in java: (X!tandempipeline version 3.3.1; http://pappso.inra.fr/bioinfo/xtandempipeline/). The final search results were filtered by using a multiple threshold filter applied at the protein level and consisting of a Log10 protein E-value lower than −2.6 identified with a minimum of two different peptides sequences, detected in at least one analysis, with a peptide E-value lower than 0.05.

Relative quantification of peptides and proteins

Control quality of data, normalization, filtration and statistical analysis were performed by using MassChroqR (http://pappso.inra.fr/bioinfo/masschroqr/). The peaks showing a width greater than 100 s or instability in retention time (RT standard deviation greater than 20 s) were filtered out. Only proteins quantified with at least two specific and repeatable peptides were analyzed. Unspecific peptides were discarded. Repeatable peptides were those which were presents in at least 7 of 8 samples. Missing values were imputed by linear regression according to the values of the other peptides of the same protein. Data were normalized to compensate for global variations between LC-MS: for each LC–MS the normalization factor was the median value of the ratios of peptide intensities to their intensity in a reference LC-MS. Protein values were computed by adding the normalized values of their specific and repeatable peptides. The proteins whose number of peaks was significantly different (with a minimum difference of 5 peaks between the mutant and the wild type) were determined by using the Kruskal-Wallis test (Additional file 9: Figure S6, Additional file 10: Figure S7, Additional file 11: Figure S8 and Additional file 12: Figure S9). A one-way ANOVA model was used to analyze changes, with the genotype as a fixed effect. A protein was considered as significantly variable when the p-value was < 0.05 (Additional file 4: Table S2).

Purification of YvmB6His

His-tagged YvmB was over-produced as a soluble protein and purified using plasmid pJ411 (DNA2.0) and E. coli strain ER2566. ER2566 carries a chromosomal copy of the T7 RNA polymerase gene inserted into the lacZ gene, and thus is under the control of the lac promoter. The yvmB gene was amplified by PCR using a forward primer (5′-gggcatatgtctgatttgacaaaacagatg-3′) which contains a 5′ NdeI site and a reverse primer (5′-gggctcgagttaatggtgatggtgatggtgctttacaggtttgtctggagt-3′) which contains a His-tag followed by a XhoI site. The amplified fragment was digested with NdeI/XhoI and ligated NdeI/XhoI digested-pJ411 vector to construct plasmid pJ411-yvmB, which encodes a fusion protein containing YvmB followed by a hexa-his-tag. This plasmid was introduced into E. coli strain ER2566. Transformants were grown in 30 μg kanamycin ml−1 LB medium at 30 °C to an OD600 of 0.8. The expression of the yvmB gene was induced by adding 0.5 mM IPTG (isopropyl-β-D-thiogalactopyranoside). The cells were then incubated for an additional 3 h before they were harvested. The cell pellet was resuspended in 1 M NaCl-50 mM Tris–HCl (pH 8.0) and was broken by sonication (Bioblock Sientific, Vibra-CellTM 72408). The cell lysate was clarified by centrifugation for 90 min. at 40 000 rpm to remove cell debris. Purification was performed by stepwise elution with linear imidazole gradient (20 to 400 mM) from a Ni-affinity column. After sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis of the fractions, the fraction containing the protein was dialyzed against against 50 mM Tris–HCl (pH 8.0)-0.4 M NaCl-1 mM DTT-50 % glycerol, and stored at −20 °C.

Denaturing gel electrophoresis

The SDS-polyacrylamide gels electrophoresis (PAGE) were 13.5 % acrylamide. The samples were loaded after being mixed with the sample buffer (125 mM Tris–HCl, 10 % glycerol, 0.1 % bromophenol blue, 2 % SDS, 1 mM DTT) with 10 min. boiling at 90 °C. Electrophoresis were done in Tris-Glycine buffer for 4 to 6 h at 15 mA.

Gel mobility shift assay

Sequences of PCR primer pairs are shown in Additional file 7: Table S3. The Cy5 5′-labeled oligodeoxynucleotides were purchased from Eurofins Genomics (Ebersberg, Germany). PCRs were performed on DNA using Ex Taq Polimerase (Takara). Amplified fragments were gel-purified using Wizard Kit (Promega). A further PCR was run using the purified fragment as template. Electrophoretic mobility shift assay was carried out by mixing 50 ng of the desired Cy5-labelled purified DNA with a 5× binding buffer (100 mM Tris pH 8, 0.5 M KCl, 2.5 mM dithiothreitol, 25 mM MgCl2, 0.25 mg/ml BSA, 0.25 mg/ml poly(dIdC)) to a final volume of 20 μL. Different amounts of purified His-tagged YvmB were used in the reactions to obtain a clear shifted DNA band compared to the control without protein. After incubation of the reaction mixture for 20 min at 30 °C, the shift reaction was loaded onto a 6 or 10 % native polyacrylamide gel (depending on the fragment size) and run for 1 h 45 min. at 150 V. The migration of the DNA bands was visualized by a Bio-Rad Chemidoc imager with Image Lab software (Cy5-based blot protocol).