Bacillus subtilis strain deficient for the protein-tyrosine kinase PtkA exhibits impaired DNA replication

Authors


E-mail im@biocentrum.dtu.dk; Tel. (+45) 45 25 24 95; Fax (+45) 45 93 28 09.

Summary

Bacillus subtilis has recently come into the focus of research on bacterial protein-tyrosine phosphorylation, with several proteins kinases, phosphatases and their substrates identified in this Gram-positive model organism. B. subtilis protein-tyrosine phosphorylation system PtkA/PtpZ was previously shown to regulate the phosphorylation state of UDP-glucose dehydrogenases and single-stranded DNA-binding proteins. This promiscuity towards substrates is reminiscent of eukaryal kinases and has prompted us to investigate possible physiological effects of ptkA and ptpZ gene inactivations in this study. We were unable to identify any striking phenotypes related to control of UDP-glucose dehydrogenases, natural competence and DNA lesion repair; however, a very strong phenotype of ΔptkA emerged with respect to DNA replication and cell cycle control, as revealed by flow cytometry and fluorescent microscopy. B. subtilis cells lacking the kinase PtkA accumulated extra chromosome equivalents, exhibited aberrant initiation mass for DNA replication and an unusually long D period.

Introduction

Protein phosphorylation on serine, threonine and tyrosine residue, a hallmark of eukaryal signal transduction, was not traditionally considered of great importance in bacteria. The evidence of serine, threonine and tyrosine phosphorylation of bacterial proteins slowly accumulated over the past decades. Today, these post-translational modifications are recognized as important parts of bacterial arsenal for regulating protein activity (Deutscher and Saier, 2006). Protein-tyrosine phosphorylation, the most recently discovered of the three types, was documented in several Gram-positive and Gram-negative bacteria (Cozzone et al., 2004), and this study is a part of our ongoing research on tyrosine phosphorylation in Bacillus subtilis. This bacterium possesses a four-gene operon, previously known as ywqCDEF, that was recently shown to encode a protein-tyrosine phosphorylation system (Mijakovic et al., 2003). The components of this system have been extensively analysed in vitro and subsequently renamed according to their functions: the transmembrane modulator of kinase activity TkmA, the soluble protein-tyrosine kinase PtkA and the phosphotyrosine-protein phosphatase PtpZ (Mijakovic et al., 2005a). The kinase PtkA exhibits an interesting feature: it requires activation by a transmembrane adapter protein. It has been shown on a homologous system in Staphylococcus aureus that the kinase activation by the adapter triggers the increase of ATP-binding affinity of the kinase, and is mediated by a specific structural motif in the adapter, pin-pointed in that study (Soulat et al., 2006). Our previous work established that the PtkA/PtpZ protein-tyrosine phosphorylation system can modify the phosphorylation state of several B. subtilis proteins in vitro: the UDP-glucose dehydrogenases Ugd and TuaD (Mijakovic et al., 2003; 2004; 2005b), and single-stranded DNA-binding proteins SsbA and SsbB (YwpH) (Mijakovic et al., 2006). As these previous analyses were mainly biochemical, in this study we investigated the physiological role of the PtkA/PtpZ phosphorylation system by a series of in vivo assays. We previously hypothesized that this system might be involved in two general physiological processes: (i) the phosphate starvation response via TuaD and Ugd, as the tuaD gene is a part of the Pho regulon (activated under phosphate starvation) and its function is the synthesis of UDP-glucuronate, the first precursor in the biosynthesis of taichuronic acid (Pagni et al., 1999; Bhavsar et al., 2004; Mijakovic et al., 2004) and (ii) DNA metabolism via single-stranded DNA-binding proteins SsbA and SsbB (YwpH) (Mijakovic et al., 2006). The larger B. subtilis single-stranded DNA-binding protein, SsbA, is essential and presumably involved in DNA replication (Lindner et al., 2004) while the smaller one, SsbB (YwpH), is involved in natural competence of B. subtilis (Hahn et al., 2005). In the present study we tried to assess the phenotypes of ptkA- and ptpZ-deficient strains with respect to uronic acid content, competence, survival of gamma radiation-induced DNA lesions and DNA replication. The most striking phenotype observed was that of the ΔptkA strain which exhibited impaired cell cycle resulting in large cells with extra chromosomes that could be related to a defect in DNA replication.

Results

Growth of B. subtilis ΔptkA, ΔptpZ and Δugd strains is not affected by phosphate starvation

Our individual gene knockouts were constructed using the vector pMUTIN-2 (Vagner et al., 1998). Three different growth conditions were used: rich medium (LB), minimal medium with high phosphate content (2 mM) and minimal medium depleted for phosphate (75 μM). Growth rate of all strains was assayed in batch growth and in a bioscreen (Bioscreen C, Labsystems), which allows continuous monitoring of growth in thermostated 100 μl wells. All strains exhibited generation times close to the wild type under all employed conditions (data not shown).

Functional analysis of the B. subtilis genome revealed that the first gene, encoding the kinase adapter TkmA, is constitutively expressed in the exponential phase during growth in the minimal medium [data available at the MICADO database (http://locus.jouy.inra.fr/cgi-bin/dev/chiapell/result-old.operl?STRAIN=BFS1325&NAME=). Our pMUTIN-derived constructs allowed us to indirectly monitor the expression of the remaining genes in the ywq operon by measuring the β-galactosidase activity, consequently revealing that all ywq genes were expressed under employed experimental conditions, and the expression levels were not affected by phosphate in the medium (D. Petranovic, unpubl. data).

Uronic acid pool in B. subtilis strains ΔptkA, ΔptpZ and Δugd

The tua operon is responsible for synthesis of taichuronic acid, an acidic exopolysaccharide that partially replaces the phosphate-containing teichoic acid as the major acidic exopolysaccharide in vegetative B. subtilis grown in a phosphate-poor medium (Lahooti and Harwood, 1999; Soldo et al., 1999; Bhavsar et al., 2004). The precursor of taichuronic acid is UDP-glucuronic acid, obtained by oxidation of UDP-glucose that is catalysed by a class of enzymes called UDP-glucose dehydrogenases. It was previously reported that TuaD is the only UDP-glucose dehydrogenase responsible for synthesis of UDP-glucuronic acid in phosphate-starved B. subtilis (Pagni et al., 1999; Soldo et al., 1999). Although B. subtilis possesses two highly homologous proteins, Ugd and YtcA, these are apparently not involved in taichuronic acid synthesis. We have previously shown that Ugd can also oxidize UDP-glucose in vitro and PtkA can phosphorylate tyrosine(s) in Ugd and TuaD in vitro, thereby activating both enzymes (Mijakovic et al., 2003). The B. subtilis strain deficient for the phosphatase PtpZ was shown to possess elevated UDP-glucose dehydrogenase cytoplasmic activity during exponential growth in LB (Mijakovic et al., 2003), further strengthening the observation that tyrosine phosphorylation regulates Ugd (tuaD gene is not transcribed under these conditions). The expression pattern of the gene ytcA is presently not known. We presumed that Ugd might contribute directly to the intracellular pool of glucuronic acid, and PtkA could modulate that activity. Previous quantifications of uronic acids concentrated on the exopolysaccharide fraction, have shown very clearly that uronic acid incorporation is concomitant to phosphate starvation, and depends on the tua operon (Soldo et al., 1999). These quantifications were performed essentially as described by Blumenkrantz and Asboe-Hansen (1973), using a method that releases glucuronic acid from the polysaccharide.

In this study we applied an adapted method with Tollens reagent (naphtoresorcinol) that includes cell lysis in phosphoric acid and subsequent processing at 70°C. This treatment does not release uronic acids from polymers, so our readings corresponded to the free cytosolic uronic acid monomers. We focused on the two extreme conditions, minimal medium with 2 mM phosphate and 75 μM phosphate. B. subtilis samples were taken in the exponential phase and in transition towards the stationary phase (Fig. 1A) and the uronic acids were quantified as described in Experimental procedures. As expected, the pool of uronic acids was about threefold increased in phosphate-starved cells. Knockouts of genes ptkA, ptpZ and ugd did not affect the uronic acid pool. However, in phosphate abundant conditions ΔptkA and Δugd strains reproducibly exhibited a 30–40% decrease in uronic acid content (Fig. 1A). This indicated that PtkA might regulate UDP-glucose dehydrogenase activity in these conditions. Therefore, we compared the UDP-glucose dehydrogenase activity of the ΔptkA strain with that of the wild type in phosphate-rich medium (2 mM phosphate), discovering that inactivation of the kinase PtkA reduced the specific enzyme activity of the extract (Fig. 1B).

Figure 1.

A. Quantification of uronic acids in different B. subtilis strains. The left panel presents the results obtained in minimal medium with abundant phosphate (2 mM), and the right panel with limiting phosphate (75 μM). Samples were collected in the exponential phase (A600 = 0.4–0.6, grey bars), and in transition towards stationary phase (A600 = 2.0–2.5 for phosphate-replete and A600 = 1.0–1.1 for phosphate starved cultures, white bars). Standard deviations from three independent experiments are indicated with error bars.
B. UDP-glucose dehydrogenase activity of B. subtilis crude protein extracts. Wild-type and ΔptkA strain grown in phosphate-replete medium, protein extracts were standardized with respect to protein content (Bradford assay), and UDP-glucose dehydrogenase activity was measured as described by Pagni et al. (1999). Increase in A340 corresponds to reduction of NAD+. Two independent experiments are shown for each strain.

Competence and double-strand break repair in B. subtilis strains ΔptkA and ΔptpZ

In addition to UDP-glucose dehydrogenases, PtkA and PtpZ were shown to control the phosphorylation level of single-stranded DNA-binding proteins in vitro, and with respect to SsbA this observation was also documented in vivo (Mijakovic et al., 2006). Another conclusion from that study was that the in vivo level of SsbA phosphorylation is very low. Tyrosine phosphorylation of SsbA increased its ability to bind single-stranded DNA in gel-shift assays, so we assumed it might affect the steps of DNA metabolism where DNA undergoes a single-stranded stage. That prompted us to examine possible phenotypes of PtkA- and PtpZ-deficient B. subtilis in such stages of DNA metabolism, natural competence being one of them. An elegant study by Hahn et al. (2005) has recently demonstrated the implication of SsbB (YwpH) in this process. We therefore tested the competence of ΔptkA and ΔptpZ strains and established that it does not differ considerably from the wild type. Competence assays were performed as described in Experimental procedures, in three independent experiments, and the respective transformabilities of ΔptkA and ΔptpZ were determined at 95% and 87% of the wild type, neither of which seems like a significant phenotype.

We have also previously reported that the ΔptkA strain exhibited a moderate (about twofold) increase in resistance towards a DNA-damaging agent mitomycin in the exponential growth phase (Mijakovic et al., 2006). SsbA is recruited to single-stranded DNA regions in the initial stages of repair in Escherichia coli (Morimatsu and Kowalczykowski, 2003) and we hypothesized that the observed phenotype might be due to a slight decrease in SsbA DNA-binding affinity, which would facilitate its displacement from the lesion site in order to facilitate the access of the repair machinery (Mijakovic et al., 2006). One would naturally expect the opposite effect in the ΔptpZ strain, but we were unable to demonstrate it reproducibly. In this study we examined the behaviour of ΔptkA and ΔptpZ strains under a different type of DNA-damaging conditions, namely gamma radiation which produces double-strand breaks. Cells were exposed to different doses of gamma radiation and the survivors were counted on plates (Fig. 2). The survival of ΔptkA and ΔptpZ samples that were irradiated in the exponential phase did not exhibit any significant difference from the wild type (Fig. 2A), but the cells in the transition phase showed some differences (Fig. 2B). Wild-type cells were most resistant of the three strains, and ΔptkA and ΔptpZ were two to three times more sensitive. This differences appear to be more significant at larger doses (standard deviations from three independent experiments are presented as error bars in Fig. 2).

Figure 2.

Survival of B. subtilis strains exposed to different doses of gamma radiation (doses indicated on the horizontal axis). Samples were taken in the exponential growth phase (A600 = 0.5–0.6) (A) and transition to stationary phase (A600 = 2.6–3) (B). Survival is plotted on a logarithmic scale, with 100% defined as the number of colony-forming units with zero dose. The plotted data are the mean values from three independent experiments. Standard deviations are indicated with error bars.

DNA replication is affected in B. subtilis ΔptkA strain

Single-stranded DNA-binding proteins are important participants in the process of DNA replication in all organisms, eukarya and bacteria alike (Allen and Kornberg, 1993; Zou et al., 2006). Eukaryal single-stranded DNA-binding proteins are known to be phosphorylated on serine/threonine residues by several different kinases, and interestingly, at least one type of this phosphorylation affects the interaction of eukaryal single-stranded DNA-binding proteins with DNA replication origins (Szuts et al., 2003; Vassin et al., 2004). Bacterial single-stranded DNA-binding proteins also have a prominent role at the origin of DNA replication (Krause and Messer, 1999), so we were compelled to examine a possible effect of ΔptkA mutation on DNA replication in B. subtilis. To analyse DNA replication, flow cytometry experiments were performed with slowly growing balanced cultures (Michelsen et al., 2003), in the minimal medium with succinate as carbon source. This allowed us to grow B. subtilis wild type and ΔptkA with a generation time of over 100 min, but we also encountered some problems with spores in such cultures, which presented a technical hindrance for the flow cytometer. To overcome this problem, we switched to a sporulation-deficient strain ΔspoIIAC. In Figs 3 and 4ΔspoIIAC is referred to as wild type, and ΔspoIIACΔptkA simply as ΔptkA, as both strains are isogenic except for the ΔptkA mutation. Figure 3A displays the flow cytometry profiles of wild type and ΔptkA strains, with distribution of cell size (measured as forward light scatter) and cellular DNA content (measured as fluorescence). The right panel illustrates a considerable subset of ΔptkA cells that are unusually large, and contain excess DNA. Flow cytometer cell samples were examined under a microscope, ruling out artefacts resulting from several cells sticking together (Fig. 4A). The ΔptkA cell population also has over twofold larger initiation mass, indicating that these cells might have a delay in initiating DNA replication (Fig. 3B). We further quantified the aberration of ΔptkA cells by examining the average distribution of DNA content per cell (Fig. 3C). Wild-type cells are found in one of the two usual states: either replicating the chromosome or possessing two complete chromosome equivalents (Fig. 3C, top). The ΔptkA strain had a completely different phenotype. The major peak in the histogram comes from newborn cells with two non-replicating chromosomes. These cells after replication should give rise to cells with four chromosomes, which are also found in the histogram. In between, a peak of cells with three chromosome equivalent is seen, which presumably contains both cells with three chromosomes (resulting from replication initiation on only one chromosome in a two-cromosome progenitor cell) and two-chromosome cells in the process of replicating both their chromosomes (as indicated by a cartoon in Fig. 3C, bottom). In front of the two chromosome cells, a peak of cells with less than two chromosomes is seen, originating from division of three chromosome cells that would give rise to cells with one or two chromosomes. An examination of these ΔptkA cells in the fluorescence microscope revealed a chaotic situation, with cells containing two and four nucleoids together with cells with different numbers of nucleoids sticking together (Fig. 4). The presence of these abnormal cells challenged the interpretation of the flow cytographic histogram, but they did not contradict the main conclusion that the ΔptkA cells are largely born with two non-replicating chromosomes thus being diploid as a result of a very prolonged D period. The three chromosome cells seen in the histogram might thus be cells with one and two chromosomes sticking together, although the presence of a fraction of cells with three non-replicating chromosomes is difficult to explain in a culture of exponentially growing cells as cells sticking together – especially when there were only very few cells with one chromosome present and all of these appeared engaged in replication. The flow cytometric cytogram revealed that initiation in the ΔptkA is asynchronious within the cell cycle in contrast to the wild type. This observation is supported by the broad distribution of the initiation masses of this strain (Fig. 3B). The D period of the ΔptkA strain was found to be longer than the generation time in accordance with the finding that the newborn cells had two chromosomes (see legend to Fig. 3).

Figure 3.

Flow cytometry of B. subtilis strains wild type and ΔptkA (both strains have a ΔspoIIAC background, and are isogenic except for the ΔptkA mutation).
A. The overall distribution of cell size (forward light scatter) and DNA content (fluorescence) in the two strains.
B. The initiation mass of the wild type (open squares) and ΔptkA (grey diamonds) strains.
C. DNA content per cell in the population of two strains. Black diamonds represent actual experimental data and the grey line is a curve fit. The position of chromosome equivalents on the diagram is indicated with cartoons. Calculated cell cycle parameters: generation time 104 min for the wild type, 106 min for ΔptkA, C period: 69 min for the wild type, 56 min for ΔptkA, D period: 42 min for the wild type, 135 min for ΔptkA.

Figure 4.

Fluorescent microscopy with B. subtilisΔptkA (A) and wild-type (B) cells used in the flow cytometry experiment. Pictured cells are representative of a larger population of examined cells.

Using fluorescence and phase contrast microscopy, we confirmed that the observed ΔptkA phenotype also persists in the LB medium, and in cells free of the ΔspoIIAC background, thus ascertaining it was independent of the slow growth experimental set-up we used for flow cytometry (Fig. 5, compare A with B). To confirm that the observed phenotype was indeed caused by the absence of PtkA function, and not a polar effect in the ywq operon, we introduced ectopically a wild-type copy of the ptkA gene in the ΔptkA strain, which restored the wild type appearance of cells under the microscope (Fig. 5C).

Figure 5.

Complementation experiment of ΔptkA strain with the wild-type copy of ptkA inserted in the amyE locus and placed under control of a synthetic promoter. The three strains, wild type (A), ΔptkA (B) and ΔptkA amyE::SP-ptkA (C), were grown in LB medium and examined by phase contrast and fluorescent microscopy. Phase contrast images are on the left, and fluorescent images showing DNA content are on the right.

Discussion

Bacillus subtilis tyrosine phosphorylation system PtkA/PtpZ was previously shown to control the phosphorylation state and activity of UDP-glucose dehydrogenases (Mijakovic et al., 2003) and single-stranded DNA-binding proteins (Mijakovic et al., 2006). In this study we have performed a series of in vivo experiments, looking for phenotypes of B. subtilis strains with inactivated ptkA and ptpZ genes. Previous studies in the field (Pagni et al., 1999; Soldo et al., 1999) have clearly established that TuaD is the only UDP-glucose dehydrogenase responsible for glucuronic acid synthesis in phosphate-starved B. subtilis. Unlike TuaD, the proteins Ugd, PtkA and PtpZ seem to influence the uronic acid pool only in cells growing under phosphate-replete conditions. The knockouts of ugd and ptkA exhibited a slightly reduced pool, supporting our hypothesis that Ugd, phosphorylated by PtkA, produces UDP-glucuronate in these conditions. This was further confirmed by our finding that ΔptkA strain had a reduced UDP-glucose dehydrogenase cytoplasmic activity compared with the wild-type strain. The fact that the uronic acid pool did not fall to zero in the ugd and ptkA knockout mutants suggests the existence of a backup mechanism, possibly the uncharacterized UDP-sugar dehydrogenase YtcA. The cytosolic pool of uronic acid is considerably lower under phosphate-replete conditions compared with phosphate starvation, but the purpose of maintaining this pool is not clear.

The second line of investigation of PtkA/PtpZ was directed towards the other class of their substrates: the single-stranded DNA-binding proteins. The absence of phenotype in competence indicates that SsbB (YwpH) phosphorylation by PtkA, previously detected in vitro (Mijakovic et al., 2006), is probably not relevant for SsbB involvement in competence. However, one cannot rule out the possibility that a specific adapter (or a signal) for PtkA activity might be needed in order to facilitate its effect on competence, and that we lacked such a signal in our experiments. The effects of gamma radiation-induced DNA lesions on different knockout strains were not very pronounced, and they were partially inconsistent with the ones obtained earlier with mitomycin (Mijakovic et al., 2006), where ΔptkA was slightly more resistant than the wild type. We have previously offered an explanation of PtkA and PtpZ effects based on modifying SsbA DNA-binding affinity that could influence the access of repair machinery to the lesion sites (Morimatsu and Kowalczykowski, 2003). The data from the present study indicate that in double-strand break repair, kinase knockout bears a slightly negative effect, identical to the phosphatase knockout. This might indicate that PtkA and PtpZ are required to co-ordinate a ‘cycle’ of SsbA coming on and off the single-stranded DNA template during repair, thus rendering both of their phenotypes equally damaging. Further genetic and biochemical characterization is required to elucidate this effect.

The most striking phenotype presented in this study is the accumulation of extra DNA material in ΔptkA cells, evidenced by our flow cytometry and microscopy analyses. Unfortunately, ΔptpZ exhibited a severe reduction in growth rate in the flow cytometry experimental set-up (growth on succinate) (data not shown), and therefore rendered all comparison of cell cycle parameters to wild type and ΔptkA meaningless. There are two observable defects that explain the accumulation of extra chromosome equivalents in ΔptkA cells. This mutant exhibits problems in initiating DNA replication, which is illustrated by aberrant initiation mass. The observed asynchrony of initiation of replication in the cell cycle is also consistent with difficulties in the initiation process, and this can be directly related to the function of SsbA. Phosphorylation of eukaryal single-stranded DNA-binding proteins affects their interaction with the origin of replication (Szuts et al., 2003), and it is possible that tyrosine phosphorylation of SsbA plays a similar role in B. subtilis. Another defect becomes visible upon analysing the cell cycle parameters. The ΔptkA cells have an unusually long D period, and prolonged D period was previously shown to result in accumulation of extra chromosomes in the bacterial cell (Huls et al., 1999).

Although PtkA is a Walker-motif type of bacterial kinase (Walker et al., 1982), sharing no homology to eukaryal kinases, it exhibits several features similar to its eukaryal counterparts. PtkA-like kinases from Gram-positive bacteria require a protein adapter to activate them (Soulat et al., 2006). As the activation of the S. aureus tyrosine kinase was narrowed down to a small region that interacts with the adaptor (Soulat et al., 2006), one could assume that presently non-identified adapters could exist, such ones that do not share the overall structural features of TkmA and TkmB, but possess a similar activating helix. This would resemble the mode of action of some eukaryal kinases, notably cyclin-dependant kinases involved in cell cycle control (Woo and Poon, 2003). PtkA is also relatively promiscuous for a bacterial kinase. In this respect it is also reminiscent of eukaryal kinases that are known to act upon more than one target protein substrate. For example, it was shown that Saccharomyces cerevisiae possesses on average 15 protein substrates per kinase (Ptacek et al., 2005). Naturally, the promiscuity also complicates studying the role of PtkA, as it is involved in different signalling pathways in the cell.

It is entirely possible that apart from SsbA, PtkA can have other, yet unidentified protein substrates, also participating in regulation of the cell cycle. The regulation of the cell cycle in eukarya relies on cyclin-dependant kinases that phosphorylate different players in the cell cycle, according to the specific cyclin present at that point (Dutta and Stillman, 1992). PtkA could assume a similar role, interacting with different adapter proteins that could direct its activity towards different substrates. Although our results open a possibility that PtkA might play a role in regulating the cell cycle, the examination of individual cells under the microscope reveals a complex and very heterogenous picture, from which it is hard to draw any final conclusions. PtkA mutant may have a complex pleotrophic phenotype related both to the exopolysaccharide synthesis and to the DNA metabolism.

Experimental procedures

Bacillus subtilis strains and growth conditions

Bacillus subtilis strain 168 was used as the wild type. Previously constructed knockout strains included ΔptkA (Mijakovic et al., 2006) and ptpZ (Mijakovic et al., 2003). In this study we constructed Δugd and ΔtuaD using the same method for gene inactivation, depending on single cross-overs with the pMUTIN-2 vector (Vagner et al., 1998). Sporulation-deficient strain ΔspoIIAC was kindly provided by Marta Perego (requires kanamycin at 2 μg ml−1). Derivatives thereof were created by pMUTIN-2-driven inactivations of genes ptkA and ptpZ. For complementation purposes, ptkA gene was inserted in the amyE locus of the ΔptkA strain, thus creating ΔptkA amyE::SP-ptkA. In this strain, ptkA gene was placed under control of a moderately strong synthetic promoter in an integrative vector pDG-268neo, procedure described in (Mijakovic et al., 2006), and was maintained by adding 5 μg ml−1 neomycin to the medium. All pMUTIN-2-derived constructs were maintained with 1 μg ml−1 erythromycin, and were grown with vigorous shaking at 37°C. Growth media included LB, minimal medium (Antelmann et al., 1997) with 2 mM phosphate for phosphate-replete and minimal medium with 75 μM phosphate for phosphate-limiting conditions. When slow growth was needed for flow cytometry experiments, glucose was replaced by succinate in the minimal medium.

Quantification of uronic acids

Bacillus subtilis cells were grown in 400 ml of the appropriate medium (minimal medium with different phosphate concentrations). Samples were taken at different stages of growth (transition phase: A600 = 1.1 for minimal medium + 75 μM phosphate, A600 = 2.8 for minimal medium + 2 mM phosphate, exponential phase: A600 = 0.4 for minimal medium + 75 μM phosphate, A600 = 0.6 for minimal medium + 2 mM phosphate). Cells were harvested by centrifugation and immediately placed on ice. Cell pellets were re-suspended in 2 ml of deionized water and mixed with 4 ml of 30% H3PO4 and 8 ml of 1% solution of naphtoresorcinol in acetic acid. To create optimal conditions for the reaction of naphtoresorcinol with uronic acids (producing aromatic coloured compounds), samples were incubated at 70°C for 75 min. After this incubation, the samples were thoroughly mixed with 3 ml of toluene (extracts the coloured reaction products) and placed on ice for 10 min to allow phase separation. One millilitre of toluene phase (purple/violet upper phase) was centrifuged at 15 000 g for 1 min at 4°C, and then A570 was measured in a glass cuvette. Concentration of uronic acids was calculated from a calibration diagram obtained with known concentrations of glucuronic acid (Fig. 3A). Intracellular concentrations were calculated from the relationship between OD600 of B. subtilis culture and the intracellular volume (Fujita and Freese, 1979).

UDP-glucose dehydrogenase assay

Wild-type and ΔptkA strains were grown in phosphate-replete minimal medium (2 mM phosphate). Dialysed crude protein extracts were prepared as described by Pagni et al. (1999), protein content was quantified by Bradford, and approximately 100 μg of total protein were used in the enzyme assay. Measurements of UDP-glucose dehydrogenase activity were performed in thermostated cuvettes, at 37°C, as described by Pagni et al. (1999).

Competence assay

Transformability of B. subtilis strains (wild type, ΔptkA and ΔptpZ) was assayed according to Yasbin et al. (1975). B. subtilis cells were transformed with 1 μg of the plasmid pDG-268neo (Mijakovic et al., 2006), and different dilutions were plated on LB plates (supplemented with appropriate antibiotics) for colony counting.

Gamma irradiation experiments

Bacillus subtilis strains (wild type, ΔptkA and ΔptpZ) were grown from a single colony in LB medium (supplemented with appropriate antibiotic when necessary) at 37°C with vigorous shaking, and the growth was followed continuously. Samples were taken in the exponential phase (A600 of 0.5–0.6) and transition phase (A600 of 2.5–3); cells were pelleted by centrifugation, re-suspended in 66 mM phosphate buffer, and irradiated on ice with a 60Co gamma ray source at a dose rate of 11 Gy s−1. The number of viable cells was estimated by plating serial dilutions onto LB plates and the colonies were counted after 24 h incubation at 37°C.

Flow cytometry

Flow cytometry experiments were performed as described previously (Michelsen et al., 2003). The computer simulation of DNA distribution was performed by software developed in our laboratory (Michelsen et al., 2003), which is available upon request. Slow exponential growth of all strains was achieved in the minimal medium, where glucose was replaced by succinate. To avoid technical problems with spore formation in the flow cytometer, ΔspoIIAC B. subtilis strain was used instead of wild type. The experiments were therefore performed with B. subtilisΔspoIIAC and ΔspoIIACΔptkA, in slowly growing balanced cultures.

Phase contrast and fluorescent microscopy

Bacillus subtilis cells from the culture used for flow cytometry were also examined by phase contrast and fluorescence microscopy to assess size and DNA content. Living, unfixated cells were stained with the Hoechst 33 342 fluorescence dye (0.5 μg ml−1) and examined first by phase contrast and then by fluorescence microscopy (excitation at 340 nm, emission at 450 nm). In the complementation experiment, wild-type B. subtilis, ΔptkA and ΔptkA amyE::SP-ptkA were grown in the LB medium, and samples were collected at A600 = 0.3.

Acknowledgements

We are grateful to Bjarne Albrektsen and Violeta Djekic for technical assistance, to Marta Perego for donation of the B. subtilis strain ΔspoIIAC, to Inge Lund for precious advice on uronic acid quantification and Boumediene Soufi for critical reading of the manuscript. This work was funded by the Danish National Research Council FNU (to I.M. and P.R.J.), and the Croatian Ministry of Science (to M.P.).

Ancillary