In Bacillus subtilis, the yoxA and dacC genes were proposed to form an operon. The yoxA gene was overexpressed in Escherichia coli and its product fused to a polyhistidine tag was purified. An aldose-1-epimerase or mutarotase activity was measured with the YoxA protein that we propose to rename as GalM by analogy with its counterpart in E. coli. The peptide d-Glu-δ-m-A2pm-d-Ala-m-A2pm-d-Ala mimicking the B. subtilis and E. coli interpeptide bridge was synthesized and incubated with the purified dacC product, the PBP4a. A clear dd-endopeptidase activity was obtained with this penicillin-binding protein, or PBP. The possible role of this class of PBP, present in almost all bacteria, is discussed.
In Bacillus subtilis, the yoxA and dacC genes are surrounded by transcriptional terminators and were proposed to form an operon (Tognoni et al., 1995). The first gene codes for a 37-kDa protein of unknown function, but is homologous to aldose-1-epimerases or galactose mutarotases. In this study, we describe the purification of the yoxA-encoded protein overexpressed in Escherichia coli. Enzymatic characterization was also performed. The second gene, initially referred to as pbp, was suggested to encode a dd-carboxypeptidase and was renamed dacC (Pedersen et al., 1998). Its product was referred to as PBP4a. It was overproduced in E. coli (Pedersen et al., 1998) and its dd-carboxypeptidase activity was verified in vitro (Duez et al., 2001). The crystal structures of PBP4a (PBP, penicillin-binding protein) alone or in complex with a peptidoglycan mimetic peptide are available (PDB accession code 1W5D and 2J9P, respectively) (Sauvage et al., 2007). It belongs to class C1 of PBPs, also named type 4-PBP in reference to the E. coli PBP4 (Sauvage et al., 2008). The latter is able to cleave the d-alanyl-γ-meso-2,6-diaminopimelyl (d-Ala-γ-m-A2pm) in the peptidoglycan cross-bridges formed by the dd-transpeptidases (Keck & Schwarz, 1979; Korat et al., 1991). This facet of a dd-carboxypeptidase activity is commonly named a ‘dd-endopeptidase’ activity. In this study, we clearly show that the B. subtilis PBP4a exhibits such an activity, the first to be described in this organism. The yoxA and dacC genes are cotranscribed at the end of their exponential growth from a σH-dependent promoter located immediately upstream of yoxA and dacC was first considered as a stationary phase or early-sporulation-specific gene (Pedersen et al., 1998). However, the sporulation of a B. subtilis dacC insertional mutant was not affected when compared with parental cells (Pedersen et al., 1998). A second promoter, which is σB-dependent, overlaps the first one. In B. subtilis, the sigma factor B is necessary to induce the expression of the general stress proteins that respond to difficult environmental conditions, nutrient depletion or low temperature (Hecker et al., 2007). Inside the yoxA gene, a short sequence 5′-AGGAAACGCTTCCT-3′ beginning at −499 bp of the dacC gene corresponds to a catabolite-response element. In the presence of glucose in the culture medium, the expression of dacC is repressed and this repression appears to be under the control of CcpA, the major catabolite mediator in B. subtilis (Lorca et al., 2005). We may reasonably postulate that glucose depletion induces the dacC expression while that of yoxA is induced by ammonium (as a nitrogen source) or l-tryptophan starvation (Thi Tam et al., 2007). Under laboratory growth conditions, the B. subtilis PBP4a is dispensable (Pedersen et al., 1998). However, this class of PBPs is present in almost all bacteria (except in Gram-positive cocci with a low G+C DNA content and in Listeria) and seems to be part of the minimal arsenal of PBPs as suggested by the case of Neisseria gonorrhoeae, which has only four PBPs, one class A (PBP 1), one class B (PBP2), one class C3 (PBP 4) and PBP3 that belongs to class C1. This conservation may indicate an essential role in natural niches.
Materials and methods
Bacterial strains and culture media
The laboratory B. subtilis 168 1A1 strain (Bacillus Genetic stock centre) or E. coli strains were grown at 37 °C in Luria–Bertani (LB) medium. Bacillus subtilis 168 1A1 and its isogenic BFS2613 or BFS2619 mutants (from the MICrobial Advanced Database Organization, the mutant collection of B. subtilis at http://genome.jouy.inra.fr) were also grown in M9 minimal medium supplemented in l-Trp and containing 5 mM d-galactose and 1 mM l-arabinose as carbon sources (Krispin & Allmansberger, 1998).
Recombinant DNA techniques
The genomic DNA was extracted using the Illustra™ Bacteria Genomic Prep Mini Spin kit from GE Healthcare. PCRs were performed with oligonucleotides purchased from Eurogentec (Seraing, Belgium). PCR products or DNA fragments isolated from agarose gels were purified using the Illustra™ GFX PCR DNA and gel band purification kit (GE Healthcare). DNA fragments resulting from PCRs were cloned into the pJET1.2/Blunt plasmid using the CloneJET™ PCR Cloning Kit from Fermentas Life Sciences (Vilnius, Lithuania). Escherichia coli DH5α was used for cloning purposes and the plasmids were extracted using the Genejet™ plasmid miniprep kit (Fermentas Life Sciences). The expression vector was pET20b (Novagen) doubly digested by NdeI and XhoI, and E. coli KRX (Promega Benelux, Leiden, the Netherlands) served as a host to produce the recombinant yoxA product.
An indirect enzymatic assay was performed according to Gatz et al. (1986), Thoden et al. (2002). The principle is as follows: in the presence of a mutarotase, a rapid and reversible conversion of glucose from α to β configuration occurs. A second enzyme, glucose dehydrogenase, utilizes the β-anomer to reduce NAD+. The appearance of NADH can be monitored at 340 mm. The reference reaction mixture consisted of 447.5 μL of 100 mM HEPES, pH 7.4, 12.5 μL of glucose dehydrogenase at 1535 U mL−1 (Sigma), 15 μL of 100 mM NAD+ (Sigma) and 25 μL of 33.3 mM anhydrous alpha-d-(+)-glucose from Aldrich. The reaction mixtures contained 1–5 μL of purified YoxA-His6 (1.7–6.9 μM). In all cases, the glucose solution was added last, and after a short mixing in the cuvettes of a Specord 200 spectrophotometer (Analytik Jena AG, Germany), the absorbance was monitored for 10–15 min at 30 °C. The rate of mutarotation in the samples supplemented with YoxA-His6 was automatically corrected by subtracting the spontaneous rate of α-glucose conversion to the β-form measured in the reference mixture.
For this purpose, the peptide d-Glu-δ-m-A2pm-d-Ala-m-A2pm-d-Ala (Mr: 633) mimicking the peptidoglycan interpeptide bridge was synthesized (details on the synthesis can be obtained from A. Luxen). The assay was performed at 25 °C for 60 min in 10 mM ammonium hydrogen carbonate of pH 7.7 with 1.5 mM peptide and 2 μg of purified B. subtilis PBP4a or E. coli PBP5 (total volume: 20 μL). The reaction products were 50-fold diluted with methanol and analysed by MS using a TSQ 7000 Thermoquest Finnigan apparatus equipped with an electrospray source. The cosolvent (injected at 200 μL min−1) was a 50 : 50 mixture of water and methanol containing 0.1% acetic acid.
Cloning and overexpression of the yoxA gene
The yoxA gene was amplified from the B. subtilis 168 1A1 genomic DNA by PCR using the Pfx DNA polymerase (Invitrogen) and the oligonucleotides: NdeI-yoxA 5′-CGCATATGGCAAACTTTATTGAGAAAATCACG-3′ and yoxA-XhoI 5′-CGCTCGAGTTGATGATTCAGTTCAATGGTGATAC-3′ in which the underlined sequences correspond to the NdeI and XhoI restriction sites, respectively. The ATG translation initiation codon is present in the NdeI site. The fragment was cloned into the pJet1.2 vector and its sequence was completely determined on both strands (GIGA sequencing platform, University of Liege, Belgium). This sequence is available in the EMBL database under the accession number FN430773. The coding sequence was recovered by double digestion with NdeI and XhoI and subcloned into the expression plasmid pET20b in phase with a polyhistine-encoding sequence in the 3′ of yoxA. The recombinant plasmid was used to transform E. coli KRX cells. The production of YoxA-His6 was performed under the following conditions: 1 L of LB medium (in four 1-L flasks) was inoculated (1/25) with an overnight preculture and shaken at 37 °C to an A600 nm of 0.5. The culture was cooled at 18 °C before addition of the inducer (0.1% rhamnose). After 16 h of culture at 18 °C, the cells were recovered by centrifugation (15 min at 4000 g), resuspended in 40 mL of 50 mM potassium phosphate, pH 7.5, 150 mM NaCl (buffer A) and frozen overnight at −20 °C. The cells were thawed and disrupted by sonication on ice (five cycles of 30 s and amplitude of 18 μ peak to peak with an MSE sonicator). The soluble proteins were recovered by a 30-min centrifugation at 30 000 g. A sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis confirmed the production of YoxA-His6 almost entirely in the soluble form.
Purification of YoxA-His6
The soluble protein fraction was subjected to IMAC purification by FPLC using an ÅKTA prime system (GE Healthcare). After filtration on a 0.45 μm membrane it was loaded on a 5-mL His-Trap column (GE Healthcare) equilibrated in buffer A. The column was washed with 100 mL of 50 mM potassium phosphate, pH 7.5, 500 mM NaCl and 10 mM imidazole (buffer B), and a 50 mL linear gradient between buffer B and buffer C (50 mM potassium phosphate, pH 7.5, 500 mM NaCl and 500 mM imidazole) was automatically applied (1 mL min−1) to elute the target protein. The purity of the recombinant protein was confirmed by SDS-PAGE and the more concentrated fractions were pooled and dialysed against a large volume of 10 mM HEPES, pH 8, 200 mM NaCl and 1 mM reduced glutathione. A fraction of the production was aliquoted and frozen at −20 °C while 0.02% sodium azide was added to the remaining part before storage at 4 °C. The final yield was 250 mg of pure YoxA-His6 from a 1-L culture.
Characterization of the YoxA-His6 enzymatic activity
As shown in Fig. 1, the mutarotase assay revealed an increase in A340 nm in the mixtures containing the recombinant YoxA-His6 protein and this increase was proportional to the amount of purified recombinant protein (rYoxA) added. The yoxA product can therefore be classified as a mutarotase or aldose-1-epimerase.
Growth of B. subtilis 168 1A1, BFS2613 and BFS2619 strains in minimal medium with d-galactose as the main carbon source
While possessing all proteins necessary to degrade d-galactose, B. subtilis 168 1A1 is unable to grow in minimal medium with this sugar as the sole carbon source, owing to the lack of a specific transporter. However, d-galactose is imported when l-arabinose is also added to the minimum medium (Krispin & Allmansberger, 1998). Bacillus subtilis 168 1A1 and two isogenic mutants, BFS2613 or BFS2619 (in which yoxA or dacC had been disrupted, respectively), were grown in M9 medium with 5 mM d-galactose as the main carbon source and 1 mM l-arabinose to induce the AraE transporter synthesis. As shown in Fig. 2, the growth of BFS2613, the strain deficient in YoxA, is significantly delayed (>48 h) in comparison with the parental or BFS2619 strains. After 72 h, however, the turbidities of the three cultures were equivalent. From these data, YoxA appears to be important for the utilization of d-galactose as the main C source at least during the first 2 days of culture in minimal medium.
Characterization of the PBP4a dd-endopeptidase activity
The d-Glu-δ-m-A2pm-d-Ala-m-A2pm-d-Ala peptide (Fig. 3) mimicking the E. coli and B. subtilis peptidoglycan cross-bridges was used to detect a dd-endopeptidase activity associated with PBP4a. The E. coli PBP5 served as a negative control. The MS analysis of the products resulting from the hydrolysis of the peptide by PBP4a yielded the following values: MS (ES+): m/z=262 corresponding to the mass of m-A2pm-d-Ala+1 and MS (ES+): m/z=391 corresponding to the mass of d-Glu-δ-m-A2pm-d-Ala+1. This result is in excellent agreement with the expected theoretical values after cleavage by PBP4a at the position indicated by the undulated line in Fig. 3. To our knowledge, this is the first report in which a dd-endopeptidase activity is associated with a B. subtilis PBP.
The yoxA product
In this study, we have expressed the B. subtilis yoxA gene in E. coli. The yoxA product is clearly homologous to mutarotases or aldose-1-epimerases. Indeed, the B. subtilis yoxA product is able to anomerize glucose, but the physiological substrate remains to be identified. Mutarotases do not show amino acid similarities to other proteins in the data banks, but are widely conserved in bacteria, fungi, yeast, plants and even mammals. Such enzymes are necessary for a rapid conversion of aldoses from β to α configurations, the latter being the only form recognized by aldose kinases performing the first step in the Leloir pathway (Frey, 1996; Holden et al., 2003). On the basis of clear homology (38% of similarity) with the E. colid-galactose mutarotase, named GalM, we propose to give the same appellation to the B. subtilis yoxA product. The protein sequence data reported in this paper will appear in the UniProt Knowledgebase under the accession number P39840.
The reason for the yoxA-dacC organization in a two-gene operon is not clear. In Bacillus amyloliquefaciens, a similar situation is found and the direct upstream and downstream genes are homologous to those found in B. subtilis on both sides of this small transcription unit. This is not the case in Bacillus licheniformis in which yoxA is between ydhH (encoding a protein of unknown function) and nhaC coding for the NhaC protein, a Na+/H+antiporter. In E. coli, the galM gene (encoding GalM, the protein homologous to the yoxA product) belongs to the four-gene gal operon involved in lactose and galactose metabolism (Bouffard et al., 1994).
PBP4a, the dacC product
A large set of data concerning PBP4a, encoded by dacC, have recently been published in a review relative to PBPs (Sauvage et al., 2008). In our study, we show that besides an in vitrodd-carboxypeptidase activity, PBP4a is able to cleave a synthetic peptide mimicking the natural peptidoglycan cross-bridge and therefore behaves as a so-called dd-endopeptidase. To our knowledge, this is the first report of this activity associated with a B. subtilis PBP. Nevertheless, other dd-endopeptidase(s) must be present in this rod bacterium because PBP4a appears to be dispensable, at least under laboratory growth conditions (Pedersen et al., 1998). Polyclonal antibodies allowed the detection of PBP4a in extracts of surface proteins and even in culture supernatants, likely due to peptidoglycan turnover (data not shown). These results bring solid arguments for the localization of PBP4a in the external layers of peptidoglycan, where it could exert its activity to reduce the stress generated by the expansion of the cell wall. The dd-endopeptidase activity associated with domain I, the penicillin-binding domain of PBP4a, could correspond to the physiological role of the protein. Of note, a gene (pscA) encoding a functional PBP homologous to E. coli PBP4 has been discovered in the chromosome of a eukaryotic organism, the ameba Dictyostelium discoideum that phagocytes bacteria (Yasukawa et al., 2003). The stabilization of PBP4a in the outer layers of peptidoglycan could result from interactions between polymers bearing negative charges such as teichoic acids and domain II of PBP4a characterized by an external crown of positive charges arising from protruding lysine residues (Sauvage et al., 2007). According to its surface location and its dd-endopeptidase activity, the B. subtilis PBP4a must be envisioned as a member of the autolysin pool as proposed for E. coli PBP4 (Höltje, 1995), but the dispensable nature of PBP4a under laboratory culture conditions and the absence of dacC expression in the exponential phase of growth or in the presence of a rapidly digestible carbon source suggest another role for this PBP. This type of PBP (and others) could be involved in the biofilm formation (Gallant et al., 2005), a common way of life for most bacteria in natural niches, especially under hostile conditions. Several studies comparing the gene expression of B. subtilis in the planktonic or in sessile (biofilm) states mentioned an overexpression of yoxA, dacC (encoding PBP4a) and pbpE (encoding PBP4*) (Stanley et al., 2003; Ren et al., 2004). The presence of a PBP-C1 (also named type-4 PBP) in almost all bacteria strengthens this hypothesis, but an increased expression of the yoxA-dacC operon could also be a side effect resulting from N and/or C source(s) depletion encountered in a sessile community and particularly in mature biofilms. Another role for the PBP4a when B. subtilis is subjected to C source starvation could be the participation of its dd-endopeptidase activity in the predation of other bacteria present in a biofilm or even in cannibalistic behavior, a means to recover nutrients and to retard the sporulation process (Nandy et al., 2007; Nandy & Venkatesh, 2008). Finally, the yoxA and dacC genes could participate in the B. subtilis survival.
We thank A. Galizzi (University of Pavia, Italy) for the gift of the BFS2613 and BFS2619 strains. We are grateful to E. Sauvage and S. Rigali for numerous and very fruitful discussions. This work was supported by the Belgian Program on Interuniversity Poles of Attraction initiated by the Belgian State, Prime Minister's office, Science Policy Programming (PAI6/19) and FRFC (grant 2.4.511.06) FRS-FNRS, Brussels, Belgium. A.Z. was supported by the European Commission LSHM-CT-2004-512138 program (EUR-INTAFAR). N.T. was supported by ARC 03/08-297. C.D. is Chercheur Qualifié of the Fonds de la Recherche Scientifique (Brussels, Belgium).