Bacillus subtilis BY-kinase PtkA controls enzyme activity and localization of its protein substrates

Authors


E-mail ivan.mijakovic@grignon.inra.fr; Tel. +33(0)1 30 81 45 40; Fax +33(0)1 30 81 54 57.

Summary

Bacillus subtilis BY-kinase PtkA was previously shown to phosphorylate, and thereby regulate the activity of two classes of protein substrates: UDP-glucose dehydrogenases and single-stranded DNA-binding proteins. Our recent phosphoproteome study identified nine new tyrosine-phosphorylated proteins in B. subtilis. We found that the majority of these proteins could be phosphorylated by PtkA in vitro. Among these new substrates, single-stranded DNA exonuclease YorK, and aspartate semialdehyde dehydrogenase Asd were activated by PtkA-dependent phosphorylation. Because enzyme activity was not affected in other cases, we used fluorescent protein tags to study the impact of PtkA on localization of these proteins in vivo. For several substrates colocalization with PtkA was observed, and more importantly, the localization pattern of the proteins enolase, YjoA, YnfE, YvyG, Ugd and SsbA was dramatically altered in ΔptkA background. Our results confirm that PtkA can control enzyme activity of its substrates in some cases, but also reveal a new mode of action for PtkA, namely ensuring correct cellular localization of its targets.

Introduction

Bacterial tyrosine kinases (BY-kinases) are autophosphorylating bacterial enzymes, with no orthologues in Eukarya, which are involved in many facets of cellular regulation (Grangeasse et al., 2007). BY-kinases autophosphorylate on a number of tyrosine residues in a C-terminal tyrosine cluster, and more notably, they have been found to control a number of key enzyme activities by means of phosphorylating their protein substrates (Klein et al., 2003; Mijakovic et al., 2003; 2006; Minic et al., 2007; Lacour et al., 2008). In vivo, BY-kinase activities are countered by those of cognate protein-tyrosine phosphatases that dephosphorylate BY-kinases and their substrates (Vincent et al., 1999; Mijakovic et al., 2005a). Among several roles of BY-kinases, the best characterized is their involvement in production of extracellular polysaccharides, where they directly participate in export and polymerization of sugar units (dependent on kinase autophosphorylation) and also regulate some key enzymes by means of phosphorylation (Whitfield, 2006). Escherichia coli Wzc (Vincent et al., 1999; Grangeasse et al., 2003) and PtkA from Bacillus subtilis (Mijakovic et al., 2003) are arguably the two most extensively characterized BY-kinases from the biochemical perspective. These enzymes represent two different BY-kinase architectures (Grangeasse et al., 2007), characteristic, respectively, for Proteobacteria and Firmicutes, with the former consisting of a single membrane-spanning polypeptide and the latter being split into a cytosolic kinase and a transmembrane activator. PtkA is thus a soluble cytosolic protein, activated by a specific interaction with its transmembrane modulator TkmA (Mijakovic et al., 2003). Once activated, PtkA was found to phosphorylate a number of protein substrates that belong to two major classes: UDP-glucose dehydrogenases [Ugd (old name YwqF) and TuaD] (Mijakovic et al., 2003; 2004; Petranovic et al., 2009) and single-stranded DNA-binding proteins [SsbA and SsbB (old name YwpH)] (Mijakovic et al., 2006; Petranovic et al., 2007). Both classes of enzymes were found to be activated by PtkA-dependent phosphorylation, and inactivated upon dephosphorylation by the cognate phosphatase PtpZ (Mijakovic et al., 2005b). Historically, most bacterial protein kinases were found to be specific for one protein substrate, therefore substrate promiscuity of PtkA was initially surprising to some degree. However, taking into account that eukaryal protein kinases usually phosphorylate a large number of proteins (Ptacek et al., 2005), and the fact that the number of phosphoproteins in bacterial phosphoproteomes (Macek et al., 2007; 2008; Soufi et al., 2008) surpasses the number of known kinases by a factor of 10, relaxed kinase specificity is not entirely unexpected. Recently, the first structure of a BY-kinase, a PtkA homologue CapB from Staphylococcus aureus, was resolved (Olivares-Illana et al., 2008). This Firmicute-type BY-kinase is an octamer anchored to the membrane via its interaction with the transmembrane modulator CapA (analogue of B. subtilis TkmA). Upon autophosphorylation, CapA/B octamer was suggested to dissociate, with each CapB monomer staying attached to its corresponding CapA partner. Assuming that PtkA is also found in an octameric complex, we speculated that PtkA, unlike CapB, might dissociate from its modulator TkmA under certain conditions, which would allow it to ‘seek out’ cytosolic substrates and potentially also other modulators. The recently published phosphoproteome of B. subtilis (Macek et al., 2007) reported the existence of nine new tyrosine-phosphorylated proteins, and we set out to examine whether PtkA might phosphorylate any of them. We have thus documented a number of new proteins phosphorylated by PtkA in vitro, two of which were activated by phosphorylation. More interestingly, a number of the new PtkA substrates changed their cellular localization in a PtkA-dependant manner, indicating for the first time that tyrosine phosphorylation can influence protein targeting in addition to modulating enzyme activities in bacteria.

Results

PtkA phosphorylates a number of new substrates in vitro

In order to examine whether any of the nine tyrosine-phosphorylated proteins identified in B. subtilis by site-specific phosphoproteomics (Macek et al., 2007) could be substrates of PtkA, we purified them as 6xHis-tagged fusions by affinity chromatography. We then performed in vitro phosphorylation assays, incubating these proteins with PtkA and its adapter TkmA-NCter in the presence of 32P-γ-ATP. As shown previously (Mijakovic et al., 2003), PtkA autophosphorylates weakly even in the absence of TkmA (Fig. 1, lanes 1). Under the employed experimental conditions none of the substrates autophosphorylated (data not shown), and none were phosphorylated by PtkA in the absence of the transmembrane modulator TkmA (Fig. 1, lanes 1). All nine proteins were phosphorylated by PtkA in the presence of its modulator, with various efficiencies (Fig. 1, lanes 3). YnfE was found to be heavily phosphorylated, while proteins Asd, InfA, YjoA, YorK and YvyG were moderately phosphorylated and Ldh and OppA were weakly phosphorylated. Enolase incorporated a very low level of radioactivity, making it unclear whether it could indeed be considered a substrate of PtkA. In order to confirm these finding, and verify whether the phosphorylation sites correspond to the ones detected in vivo (Macek et al., 2007), we performed a mass spectrometry analysis of non-radioactive in vitro phosphorylation mixtures. All substrate proteins were treated by the alkaline phosphatase prior to PtkA-dependent phosphorylation, to remove any potential phosphorylation originating during protein synthesis in E. coli. All nine proteins were phosphorylated on tyrosine and in case of InfA, Ldh, OppA, YjoA, YorK and YvyG the previously reported phosphorylation sites were confirmed (Table 1). Interestingly, we also identified a number of new tyrosine phosphorylation sites, whose physiological relevance remains questionable as they have only been identified in vitro. Our next question was the importance of the dissociation capability of the kinase PtkA from its adapter TkmA for substrate phosphorylation. We constructed a chimeric kinase resembling BY-kinases of Proteobacteria, by fusing the C-terminal part of the modulator TkmA to the N-terminus of PtkA and tested its phosphorylation properties in vitro. Phosphorylation assays demonstrated autophosphorylation of the chimeric kinase to be equivalent to that of wild-type PtkA activated by its modulator (Fig. 1, lanes 2). As expected, the addition of TkmA-NCter could not further stimulate the activity of the chimeric kinase (Fig. 1, lanes 4), as the PtkA region suspected to interact with TkmA was blocked by the fusion with the C-terminal part of TkmA. However, the ability of the chimeric kinase to phosphorylate its substrates was diminished compared with the wild-type PtkA. It was capable of phosphorylating InfA, YjoA, YnfE and YvyG to some extent, but phosphorylation of the other substrates was not detected (Fig. 1, lanes 2 and 4). Overproduction and purification of the chimeric kinase from E. coli resulted in a low yield and co-purification of two proteins that were also strongly phosphorylated in vitro. Whether these could be, for example, dimers of the chimeric kinase and degradation products remains to be established.

Figure 1.

In vitro phosphorylation assays. Purified protein substrates were incubated with either PtkA alone (lanes 1), chimeric kinase (lanes 2), PtkA and TkmA NCter (lanes 3) or chimeric kinase with TkmA NCter (lanes 4). Proteins were incubated with 32P-γ-ATP and 5 mM MgCl2 for 1 h, separated by SDS-PAGE and visualized by STORM PhosphorImager (GE Healthcare). For each gel the positions of PtkA and the chimeric kinase (chim) are indicated with a line, and the protein substrate position is indicated with an arrow to the right of each gel.

Table 1.  Phosphorylation sites on substrates phosphorylated in vitro by PtkA, determined by mass spectrometry.
ProteinPreviously identified sitesaConfirmedNew sites from this study
InfAY60YesNone
LdhY224YesY66, Y69, Y207, Y235
OppAY301 or Y303Yes (both)Y123, Y212, Y395, Y538
YjoAY150YesY30, Y32
AsdY146NoY289
YnfEY12NoY24, Y53, Y55
YorKY473YesY3, Y11, Y168, Y220, Y368, Y473
EnoY281NoY8, Y46, Y249, Y256, Y403, Y419, Y424, Y426
YvyGY49YesNone

Asd and YorK are activated by PtkA-dependent phosphorylation in vitro

Having demonstrated that PtkA could, at least to some extent, phosphorylate all nine new tyrosine-phosphorylated proteins, we wanted to test whether phosphorylation would influence their enzyme activities, as demonstrated for the previously characterized PtkA substrates (Mijakovic et al., 2003; 2006). To this end, in vitro assays to test the primary function of the proteins were set up and activities of unphosphorylated and phosphorylated proteins were assayed. In several cases no impact of PtkA-dependent phosphorylation on enzyme activity could be demonstrated. Such was the case of the enzymes involved in the central carbon metabolism, Ldh and enolase, where we performed simple colorimetric assays. For YvyG, a putative flagellar chaperone (Pallen et al., 2005), we assayed its ATPase activity as an indirect measure of chaperone activity. OppA is a component of the only tripeptide uptake system in B. subtilis (Koide and Hoch, 1994) and, as expected, a ΔoppA strain did not take up tripeptides. We tested peptide uptake in vivo in wild-type and ΔptkA cells but no effect on peptide uptake was observed. In each case we could detect basal enzyme activity, but no effect of phosphorylation was detected. Quantitative overview of experimental data for non-activated substrates is given in Table S2. In the case of proteins of unknown function, YjoA and Ynfe, and the translation initiation factor InfA no activity experiments were conducted. Finally, effects of PtkA-dependent phosphorylation were observed with Asd and YorK. Asd converts aspartyl phosphate to aspartyl semialdehyde and inorganic phosphate, with concomitant oxidation of nicotinamide adenine dinucleotide phosphate (NADPH). We compared the processivity of the PtkA-phosphorylated enzyme and non-phosphorylated enzyme and demonstrated a more than threefold activation of phosphorylated Asd under the conditions employed (Fig. 2). YorK is annotated as a putative ssDNA-specific exonuclease. We therefore initially tested if YorK would exhibit exonuclease activity in the presence of different divalent cations. The enzyme demonstrated ssDNA exonuclease activity in the presence of Mg2+ and Mn2+ but not Ca2+ and Zn2+ (data not shown). We then tested whether phosphorylation of YorK would affect its activity, and here it became apparent that PtkA-mediated phosphorylation activates this otherwise rather inefficient enzyme in our experimental conditions (Fig. 3). No DNA degradation products were observed on the gel, indicating that YorK acted in a processive manner, degrading the entire oligonucleotide. Control reactions confirmed that PtkA alone exhibited no exonuclease activity (Fig. S1).

Figure 2.

Activation of Asd by PtkA-dependent phosphorylation in vitro. Asd was incubated alone, with modulator TkmA, with kinase PtkA or with TkmA and PtkA for 2 h in the presence of 5 mM ATP. Asd reaction was measured as the oxidation of NADPH at 340 nm and initial reaction rates were recorded. The results represent the average of two independent measurements and standard deviations are indicated with error bars.

Figure 3.

Activation of YorK by PtkA-dependent phosphorylation in vitro. YorK was preincubated either alone (A) or with TkmA and PtkA (B) for 5 h in the presence of ATP. Exonuclease activity was thereafter measured on an 82-mer oligonucleotide substrate at the following time points: 0.5 h (lane 1), 4 h (lane 2), 6 h (lane 3), 8 h (lane 4), 10 h (lane 5), 14 h (lane 6), 19 h (lane 7), 24 h (lane 8), 48 h (lane 9). Lane 10 contained a control without YorK.

Cellular localization of PtkA substrates changes in ΔptkA background

Because the effect of PtkA on enzyme activity was observed for only two of the nine new substrates, we wondered whether PtkA might have other kinds of regulatory effects on the rest of its substrates. To this end, we devised an in vivo system to study the impact of PtkA on cellular localization of its substrates in B. subtilis using fluorescent protein tags. PtkA, TkmA and all known PtkA substrates were produced as fluorescent protein fusions [with either yellow fluorescent protein (YFP) or cyan fluorescent protein (CFP)] in wild-type B. subtilis and ΔptkA background. Initially we confirmed by microscopy that there was no signal from CFP in the yellow filter and YFP in the cyan filter (data not shown). We then coexpressed ptkA-cfp with individual target genes fused with yfp, and examined the cells with yellow and cyan filter in exponential and stationary phase. First we examined the transmembrane modulator TkmA, which was found to be associated with the membrane where it localized in patches (Fig. 4C and D). Interestingly, PtkA colocalized with the modulator in exponential phase (Fig. 4E), whereas it seemed to be found primarily in the cytosol in stationary phase (Fig. 4F). Next, the two previously characterized substrates of PtkA, SsbA (Mijakovic et al., 2006) and Ugd (Mijakovic et al., 2003), were examined. SsbA and Ugd both showed a discrete localization profile, no colocalization with PtkA was observed, but the localization pattern changed in a ΔptkA background. SsbA was located in multiple foci in wild-type cells, with the number of foci higher in rapidly dividing cells in exponential phase (Fig. 5C and D). In a ΔptkA background, a notable difference was observed in exponential phase, where a sub-population of cells showed a substantially decreased number of foci (Fig. 5I). Ugd localized to the pole of wild-type cells in both phases of growth. In the ΔptkA background the protein lost the polar localization in exponential phase and accumulated in multiple foci in some cells during stationary phase (Fig. S2). Next, we examined the new PtkA substrates and concluded that they could be categorized in four groups based on their localization profiles. The proteins Asd, YorK and InfA exhibited a diffuse localization profile in both growth phases (Figs S3–S5 respectively) and were not affected by ptkA knockout. Enolase (Fig. 6) and YjoA (Fig. S6) exhibited localization profiles that varied according to the growth phase, and were PtkA-dependent, but did not colocalize with PtkA. Enolase exhibited a diffuse profile in the exponential phase (Fig. 6C) but localized very strongly to one pole in stationary phase cells (Fig. 6D). In ΔptkA cells, enolase foci either disappeared entirely, accompanied by cytosolic distribution of the enzyme, or appeared as aberrant foci, deformed and with sub-polar localization (Fig. 6J). YjoA was localized in grains in a sub-population of cells during exponential phase and was uniformly distributed in the stationary phase (Fig. S6C and D). In the ptkA mutant there was no change in exponential phase, while in stationary phase YjoA localized in grains in a sub-population of cells (Fig. S6J). Ldh (Fig. S7), YnfE (Fig. S8) and YvyG (Fig. 7) exhibited a PtkA-dependent localization profile that varied with the growth phase, and further colocalized with PtkA in stationary but not exponential growth phase. YvyG in particular localized to the poles in the exponential phase (Fig. 7C) and in several grains spread out in the cell in stationary phase (Fig. 7D), while some cells showed a diffuse profile. This localization profile disappeared in most cells in a ΔptkA background (Fig. 7I and J). Finally, in the case of OppA the protein signal was weak and uniformly distributed in the exponential phase (Fig. 8C) while it seemed to disappear in the stationary phase (Fig. 8D), when OppA is expected to be exported and employed in peptide transport. In a ptkA mutant OppA was observed in the cytosol in the stationary phase, which might indicate accumulation due to a defect in its export (Fig. 8J). An overview of the results of this localization screening is summarized in Table 2.

Figure 4.

Localization of CFP-PtkA and TkmA-YFP. Wild-type cells visualized by phase contrast in exponential (A) and stationary phase (B). TkmA-YFP visualized in exponential (C) and stationary phase (D). CFP-PtkA visualized in exponential (E) and stationary phase (F). ΔptkA cells visualized by phase contrast in exponential (G) and stationary phase (H). TkmA-YFP visualized in ΔptkA cells in exponential (I) and stationary phase (J).

Figure 5.

Localization of CFP-PtkA and SsbA-YFP. Wild-type cells visualized by phase contrast in exponential (A) and stationary phase (B). SsbA-YFP visualized in exponential (C) and stationary phase (D). CFP-PtkA visualized in exponential (E) and stationary phase (F). ΔptkA cells visualized by phase contrast in exponential (G) and stationary phase (H). SsbA-YFP visualized in ΔptkA cells in exponential (I) and stationary phase (J).

Figure 6.

Localization of CFP-PtkA and Eno-YFP. Wild-type cells visualized by phase contrast in exponential (A) and stationary phase (B). Eno-YFP visualized in exponential (C) and stationary phase (D). CFP-PtkA visualized in exponential (E) and stationary phase (F). ΔptkA cells visualized by phase contrast in exponential (G) and stationary phase (H). Eno-YFP visualized in ΔptkA cells in exponential (I) and stationary phase (J).

Figure 7.

Localization of CFP-PtkA and YvyG-YFP. Wild-type cells visualized by phase contrast in exponential (A) and stationary phase (B). YvyG-YFP visualized in exponential (C) and stationary phase (D). CFP-PtkA visualized in exponential (E) and stationary phase (F). ΔptkA cells visualized by phase contrast in exponential (G) and stationary phase (H). YvyG-YFP visualized in ΔptkA cells in exponential (I) and stationary phase (J).

Figure 8.

Localization of CFP-PtkA and OppA-YFP. Wild-type cells visualized by phase contrast in exponential (A) and stationary phase (B). OppA-YFP visualized in exponential (C) and stationary phase (D). CFP-PtkA visualized in exponential (E) and stationary phase (F). ΔptkA cells visualized by phase contrast in exponential (G) and stationary phase (H). OppA-YFP visualized in ΔptkA cells in exponential (I) and stationary phase (J).

Table 2.  Summary of experimental results.
ProteinPhosphorylation in vitroEffect on activityColocalization with PtkAPtkA-dependent localization
TkmA+ (exp. phase)
Asd+++Diffuse
YorK+++Diffuse
Enolase(+)+
Ldh++ (stat. phase)+
YjoA++not tested+
YnfE+++not tested+ (stat. phase)+
YvyG+++ (stat. phase)+
InfA++not testedDiffuse
OppA+NANA
Ugd+++a+a+
Ssb++b+b+

Discussion

Unlike single polypeptide chain-membrane spanning BY-kinases in Proteobacteria, Firmicutes possess BY-kinases that are separated into two polypeptides: a transmembrane modulator and a cytosolic kinase. The solved structure of S. aureus BY-kinase CapB (Olivares-Illana et al., 2008) revealed an octameric BY-kinase ring structure anchored to the membrane via interaction with the transmembrane modulators. Upon autophosphorylation, the BY-kinase octamer dissociates, however, the structural data suggested that kinase monomers are likely to remain associated with monomers of the transmembrane modulator. If the BY-kinase is constantly associated to its modulator, why would Firmicutes evolve a split polypeptide chain? Data from our study are beginning to shed some light on this question. We constructed a Proteobacteria-like kinase, a chimera consisting of the C-terminus of the modulator TkmA fused to the N-terminus of PtkA. This chimeric kinase was capable of autophosphorylating to similar levels as the wild-type PtkA in interaction with TkmA; however, its capacity to phosphorylate protein substrates was severely impaired. This might suggest that the conformation of the artificial fusion kinase was not flexible enough to accommodate various substrates, but flexibility itself is unlikely to be the reason for splitting the Proteobacteria-like kinase in two. Interestingly, our localization data showed that PtkA colocalized with TkmA at the membrane in the exponential growth phase as expected, but not in the stationary phase, where it was released entirely in the cytosol. Our data suggest that PtkA plays a particularly decisive role in ensuring correct localization of several of its substrates in the stationary phase, where PtkA also colocalized with some of them. It is presently unclear whether PtkA influences substrate localization via phosphorylation, protein–protein interaction or other less direct means. However, these results clearly suggest a new model whereby PtkA can dissociate from the transmembrane modulator under appropriate conditions, thus freeing itself to phosphorylate or otherwise interact with its cytosolic substrates. Our PtkA interactome data obtained with yeast two-hybrid also support this hypothesis, suggesting that PtkA could have soluble activity modulators other than TkmA, related to its particular functions in the cytosol (M.–F. Noirot-Gros & I. Mijakovic, unpubl. results). A particularly pertinent question for further research is the identity of signal(s) that control the activity and the interaction pattern of PtkA.

We have previously shown that PtkA phosphorylates two classes of substrates, UDP-glucose dehydrogenases and single-stranded DNA-binding proteins, and plays important roles in DNA replication and cell cycle control (Mijakovic et al., 2003; 2006; Petranovic et al., 2007). These findings were further supported by our localization data, in particular pertaining to SsbA. SsbA was localized in multiple foci in the exponential phase cells. The overall number of foci we observed was larger than in previous studies (Berkmen and Grossman, 2006; Meile et al., 2006). This might be due to a higher level of ssbA expression in our system, which could be classified as a technical issue. Nevertheless, the number of SsbA foci decreased dramatically in PtkA-deficient cells. Assuming that foci ensue from SsbA binding to single-stranded regions of the chromosome, this finding concurs with the reduction of SsbA DNA-binding affinity in ΔptkA background, as observed in our previous studies (Mijakovic et al., 2006; Petranovic et al., 2007).

Besides PtkA, B. subtilis possesses a truncated BY-kinase EpsB, which is most likely non-functional (Mijakovic et al., 2003) and a kinase McsB that was initially reported as a tyrosine kinase, but has recently been recognized as an arginine kinase (Fuhrmann et al., 2009). With a relative shortage of tyrosine kinases, and the identification of nine new proteins phosphorylated on tyrosine (Macek et al., 2007), we decided to test whether PtkA would phosphorylate any of them. Here we demonstrated that PtkA could phosphorylate all nine proteins to various degrees in vitro. In order to establish the proteins as true substrates of PtkA we wanted to test if these phosphorylation events would have any regulatory role. The previously characterized substrates were shown to be activated by phosphorylation, and in this study we demonstrated that two of the new protein substrates, Asd and YorK, were also activated by PtkA-mediated phosphorylation (or the presence of PtkA) in vitro. Asd intervenes in the aspartate biosynthesis pathway, but interestingly its product aspartate semialdehyde is a precursor for diaminopimelic acid, a constituent of bacterial cell wall peptidoglycan, possibly linking this regulatory event to cell wall metabolism. YorK is one of two RecJ homologues in B. subtilis (Sutera et al., 1999), but its function has not been confirmed experimentally prior to this study. Here we initially confirmed exonuclease activity on ssDNA dependent on the requirement of Mg2+ or Mn2+ as reported for RecJ in some bacteria. Interestingly, RecJ has been shown to interact with single-stranded DNA-binding protein in Haemophilus influenzae (Sharma and Rao, 2009), and because both interactants are substrates for PtkA in B. subtilis, this phosphorylation event further highlights the importance of PtkA in ssDNA metabolism.

Surprisingly, the activity of most of the newly identified PtkA substrates was not affected by phosphorylation. We reasoned that phosphorylation could instead play a role in mediating protein interactions and we therefore examined their localization in wild-type and ΔptkA background. This approach allowed us to divide the studied proteins into several classes. The proteins that were affected at the activity level, Asd, YorK (and InfA) showed a diffuse localization profile that was not affected by PtkA inactivation. By contrast, the majority of the proteins showed a growth phase- and PtkA-dependent localization profile. Ldh, YnfE and YvyG in addition colocalized with PtkA, and enolase and YjoA did not. In the case of Ldh, YnfE, YvyG and YjoA the functional roles of observed localization profiles are difficult to interpret, but the very strong localization of enolase to one pole in the cell could indicate a role in sporulation. The multiply phosphorylated protein enolase has been shown to be multifunctional; besides its initially recognized role in glycolysis, it is implicated in heat shock response (Miller et al., 1991), RNA degradation (Commichau et al., 2009), DNA replication (Jannière et al., 2007; Commichau et al., 2009) and is also secreted upon automodification with its substrate 2-PG (Boël et al., 2004). It is also one of the most abundant proteins in Bacilli spores (Delvecchio et al., 2006). In one particular case, that of OppA, the protein signal that was absent in the wild-type stationary cells appeared in ptkA mutant cells. This protein is part of the Opp permease system specific for peptides of 3–5 amino acids length. OppA is the ligand-binding protein, and is attached to the outside of the cell via a lipid anchor. The fact that it was retained in the cytosol in the ptkA mutant could point towards phosphorylation of OppA playing a role in its export. However, the import of peptides in vivo was not affected in ΔptkA background, which cautions us against forming any final conclusions in this case.

Previous studies on BY-kinase-mediated substrate phosphorylation have mainly studied the effect on protein activity. In Eukarya, however, several examples of proteins showing a tyrosine phosphorylation-dependent localization have been reported (Madeo et al., 1998; Lukong et al., 2005) and the results presented here indicate for the first time that BY-kinases can also act in terms of ensuring correct cellular location of their substrates.

Experimental procedures

Bacterial strains and growth conditions

Escherichia coli NM522 was used for plasmid propagation in cloning experiments. The chaperone overproducing strain E. coli M15 carrying pREP4-GroESL (Amrein et al., 1995) was used for overproduction of protein. B. subtilis 168 was used in localixation experiments and in vivo assays. E. coli and B. subtilis strains were grown in LB medium shaking at 37°C. B. subtilis was grown in C-medium (Martin-Verstraete et al. 1990) with 5 g l−1 glucose (CG-medium). When relevant 100 µg ml−1 ampicillin, 25 µg ml−1 kanamycin and 8 µg ml−1 tetracycline for E. coli and 5 µg ml−1 erythromycin and 15 µg ml−1 tetracycline for B. subtilis were added to the medium.

DNA manipulations and strain construction

Genes asd, eno, infA, ldh, oppA, yjoA, ynfE, yorK, yvyG and yclM were PCR-amplified using B. subtilis 168 genomic DNA and specific primers with restriction sites (Table S1) and inserted between the BamHI and Cfr9I sites of pQE30-Xa (Qiagen). Vectors for expression of PtkA and TkmA-NCter were described previously (Mijakovic et al., 2003). pQE30-strep encoding a Strep-tag in place of a His-tag was constructed by restricting pQE30 with BamHI and EcoRI, annealing the 5′ phosphorylated oligos pQE-str+ and pQE-str- and ligating the fragments. For production of strep-tagged PtkA, the BamHI-PstI fragment from pQE30-ptkA (Mijakovic et al., 2003) was inserted in pQE30-strep. The chimeric kinase gene encoding the 50 C-terminal amino acids of TkmA fused to the N-terminus of PtkA was constructed by PCR using an overlapping primer and inserted between the BamHI and PstI sites of pQE30. The vector pG+Host8 (Maguin et al., 1996) that contains a temperature-sensitive origin of replication and a tetracycline resistance cassette was used to construct B. subtilisΔptkA by deletion of the middle part of ptkA. The upstream and downstream parts of ptkA were amplified with primer pairs ΔptkA up fwd and rev and ΔptkA down fwd and rev respectively. Vector and PCR products were restricted with appropriate enzymes and ligated. This construct was used to transform B. subtilis 168 and transformants were incubated at 37°C (non-replicative temperature) on LB plates containing 15 µg ml−1 tetracycline for chromosome integration. The integrants were re-streaked and incubated at 28°C (replicative temperature) without selective pressure to induce excision followed by a shift to 37°C for loss of the vector. The deletion was confirmed by PCR on genomic DNA. The oppA gene was inactivated using the vector pMUTIN2 (Vagner et al., 1998). For examination of protein localization, eYFP (Clontech) and CFP+ were used. Gene yfp was PCR-amplified using primers yfpC fwd and yfpC rev and a miniTn7-eyfp delivery plasmid (Lambertsen et al., 2004) as template and inserted between the BamHI and Cfr9I sites of pHT315 (Arantes and Lereclus, 1991) to generate pHT315-yfpC. In order to avoid possible low CFP signal due to slow translation initiation the first eight codons of comGA were encoded on the primer cfpN fwd (Veening et al., 2004). cfp+ (Andersen et al., 2006) was amplified using primers cfpN fwd and cfpN rev and inserted between the KpnI and EcoRI sites of pHT315-yfpC to generate pHT315-yfpC-cfpN. Gene ptkA was amplified using primers pktAN fwd and ptkAN rev and inserted between the Cfr9I and EcoRI sites of pHT315-yfpC-cfpN to generate pHT315-yfpC-cfp:ptkA. Genes encoding the phosphotyrosine proteins and TkmA were PCR-amplified with relevant primers and inserted between the BamHI and AvrII sites of pHT315-yfpC and pHT-yfpC-cfp:ptkA to generate pHT315-′;gene of interest:yfp and pHT-′gene of interest:yfp-cfp:ptkA respectively. The vector pHT315-cfp:ptkA was constructed by restricting pHT315-yfpC-cfp:ptkA with EcoRI and KpnI and inserting the cfp:ptkA fragment in pHT315. B. subtilisΔptkA was transformed with pHT315-′gene of interest:yfp and B. subtilis 168 was transformed with pHT315-gene of interest:yfp-cfp:ptkA and pHT315-cfp:ptkA.

Production and purification of 6xHis-tagged proteins

His-tagged proteins were synthesized in the chaperone overproducing strain E. coli M15 carrying pREP4-groESL. Cultures were grown shaking at 37°C to OD600 0.5, induced with 1 mM IPTG and grown an additional 3 h. His-tagged proteins were purified on Ni-NTA columns (Qiagen) as described previously (Mijakovic et al., 2003), desalted with PD-10 columns and stored in a buffer containing 50 mM Tris-Cl pH 7.5, 100 mM NaCl and 10% glycerol. Proteins for mass spectrometry analysis were further dephosphorylated with shrimp alkaline phosphatase (Fermentas). Crude extract from 1 l culture were incubated with 30 units at 37°C for 1 h.

In vitro phosphorylation assays

Phosphorylation assays were performed essentially as described previously (Mijakovic et al., 2003), with 4.5 nM Ldh, 0.48 nM OppA, 17 nM YjoA, 18 nM YvyG, 26 nM InfA, 45 nM YnfE and 0.17 nM chimeric kinase or 0.23 nM PtkA and TkmA-NCter. For YorK, 2.0 nM, 43 nM and 18 nM Asd and enolase were used, respectively, with 0.55 nM chimeric kinase or 0.75 nM PtkA and TkmA-NCter. After 1 h incubation at 37°C, proteins were separated by electrophoresis and radioactive signals were visualized with a Storm 860 PhosphorImager. For non-radioactive assays, destined for mass spectrometry analysis, the same procedure was applied but with non-radioactive ATP.

Mass spectrometry analysis of in vitro phosphorylation sites

Proteins from in vitro phosphorylation reactions (25–100 µg) were diluted with denaturation buffer (6 M urea, 2 M thiourea in 10 mM Tris, pH 7.0) to a final concentration of 1–2 µg µl−1, reduced, alkylated and digested with endoproteinase Lys-C and trypsin as described previously (Miller et al., 2009). A part (10%) of the sample was taken and the peptides were desalted directly using C18 StageTips (Rappsilber et al., 2007). The rest was subjected to phosphopeptide enrichment using TiO2 beads (GL Sciences) after adding acetonitrile (ACN) to a final concentration of 30%. The beads were preincubated with a solution containing 20 mg ml−1 dihydrobenzoic acid in 80% ACN and 5 mg of beads were added to each sample. Following the incubation for 1 h with end-over-end rotation, the beads were washed twice in a solution containing 60% ACN and 0.1% trifluoroacetic acid and eluted with 150 µl of 40% ammonia solution in 60% ACN (pH 10.5). Eluates were prepared for LC-MS by reducing their volumes to 5 µl and adding an equal volume of a solution containing 2% ACN and 1% trifluoroacetic acid. The peptide mixtures were analysed using a Proxeon Easy-LC system (Proxeon Biosystems) coupled to a LTQ-Orbitrap-XL mass spectrometer (ThermoFisher) equipped with a nanoelectrospray ion source (Proxeon Biosystems). Chromatographic separation and mass spectrometry were performed essentially as described previously (Miller et al., 2009). An inclusion list containing the m/z-values of peptides carrying previously identified phosphorylation sites (Macek et al., 2007) was used. Mass spectra were analysed using the software suite MaxQuant, version 1.0.14.3 (Cox et al., 2009). The data were searched against database of B. subtilis (forward primary annotation database was downloaded from http://cmr.jcvi.org) supplemented with tagged versions of used targets and frequently observed contaminants and concatenated with reversed copies of all sequences using MASCOT (version 2.2.0, Matrix Science, London, UK). Carbamidomethylation of cysteine was set as fixed modification and oxidation of methionine, N-terminal acetylation, and phosphorylation as variable modifications. Initial mass tolerance was set to a maximum of 7 ppm and a maximum of two missed cleavages was allowed. Maximum false discovery rates were set to 1% for both, peptide and protein levels. Phosphorylation events were considered to be localized to a specific site if the calculated localization probability was above 0.75.

Enolase activity assay

Enolase catalyses the conversion of 2-phospho-D-glycerate to phosphoenolpyruvate (2-PGA) and H2O. To phosphorylate enolase, 0.84 µM enolase was incubated with 1.5 µM PtkA, 1.1 µM TkmA-NCter, 5 mM ATP, 5 mM MgCl, 50 mM Tris-Cl (pH 7.5), 100 mM NaCl and 10% glycerol for about 16 h at 37°C. The sample without PtkA and TkmA-NCter was prepared similarly. 25 µl of samples was assayed in total volume of 1 ml comprising 50 mM Tris-HCl (pH 7.4), 100 mM KCl, 1 mM MgSO4, 0.01 mM EDTA and 2 mM 2-PGA, as described previously (Boël et al., 2004). Reactions were initiated by addition of 2-PGA, followed spectrophotometrically at 240 nm at 37°C, and initial reaction rates were recorded.

Ldh activity assay

Ldh catalyses the conversion of pyruvate and NADH to lactate and NAD+. NADP+ can be used as cofactor (Romero et al., 2007) and fructose bisphosphate (FBP) has been shown to activate Ldh in some species (Fushinobu et al., 1996). Effect of phosphorylation was tested with both NADH and NADPH as cofactor, with and without FBP present. In phosphorylation reactions Ldh was incubated with PtkA, TkmA-NCter, 2.5 mM ATP, 2.5 mM MgCl2, 1 g l−1 BSA, 50 mM Tris-Cl (pH 7.5), 100 mM NaCl and 10% glycerol either with or without 0.1 mM FBP for 5 h at 37°C. For NADH experiments 0.56 µM Ldh, 0.47 µM PtkA and 0.47 µM TkmA-NCter and for NADPH experiments were used 8.4 µM Ldh, 7.5 µM PtkA and 7.5 µM TkmA-NCter. Samples without PtkA and TkmA-NCter were prepared similarly. To assay activity, 5.6 nM Ldh for NADH experiments and 84 nM Ldh for NADPH experiments were mixed with 1 mM pyruvate and 1 mM NADH/NADPH with or without 0.1 mM FBP in a buffer containing 50 mM Tris-Cl (pH 7.5), 100 mM NaCl and 10% glycerol. Reactions were started by addition of pyruvate, followed spectrophotometrically at 340 nm at 37°C and initial reaction rates were recorded.

OppA assay

OppA is part of an ABC transporter that imports peptides of 3–5 amino acids length. To test if peptide import was affected by PtkA-mediated phosphorylation we tested sensitivity to the tripeptide antibiotic bialaphos in B. subtilis wild-type and ΔptkA strains essentially as described previously (Koide and Hoch, 1994). Cultures were grown in CG-minimal medium to OD600 of 0.4 (exponential phase) and 1.5 (transition phase). Inhibition zones were measured and results are an average of three experiments.

YvyG chaperone activity assay

YvyG is a putative flagellar chaperone with a putative GTPase domain. Because chaperone activity is ATP-dependent, we attempted to quantify ATP hydrolysis activity as an indirect measure of chaperone activity. Phosphorylation reactions contained 11 µM YvyG, 3.0 µM Strep-tagged PtkA, 2.3 µM TkmA-NCter, 0.5 mM ATP, 5 mM MgCl2, 50 mM Tris-Cl (pH 7.5), 100 mM NaCl and 10% glycerol. Reactions with no PtkA and TkmA were prepared similarly and reactions were incubated for 3 h at 37°C. To remove basal ATP hydrolysis activity from PtkA, Strep-tagged PtkA was removed by mixing 95 µl of the phosphorylation reactions with 150 µl Streptactin resin (IBA). ATP hydrolysis was assayed by adding 32P-γ-ATP to a final concentration of 300 µM. Samples were incubated at 37°C for 1, 2, 4, and 6 h, 1 µl was spotted on a polyethylenimine cellulose sheet (Machery-Nagel) and separated by thin-layer chromatography using a 0.3 M potassium phosphate buffer (Mijakovic et al., 2002). Radioactive signals were visualized using the STORM PhosphorImager and quantified with ImageQuant (GE Healthcare).

Single-stranded DNA exonuclease assay

To test the effect of phosphorylation on YorK exonuclease activity 0.15 µM YorK was phosphorylated with 0.75 µM PtkA and 0.75 µM TkmA-NCter in a buffer containing 50 mM Tris (pH 7.5), 100 mM NaCl, 8 mM MgCl2, 2 mM ATP and 10% glycerol for 5 h. Unphosphorylated YorK was treated in the same buffer without kinase and modulator. The two controls included a phosphorylation reaction without YorK, and one with 115 µM BSA added instead of PtkA. To 12.5 µl phosphorylation reaction, 1.5 µl reaction buffer [20 mM Tris-HCl (pH 8.0), 10 mM MgCl2, 10 mM NaCl, 0.67 mM dithiothreitol and 1 mg ml−1 BSA] and 1 µl of 0.6 pmol of 82-mer oligonucleotide in water. The reactions were incubated at 37°C for indicated times (see figure legend) and terminated by heat inactivation for 10 min at 70°C. Samples were subjected to gel electrophoresis on 1.5% agarose gel with ethidium bromide and visualized by UV light.

Asd assay

Asd catalyses the conversion of aspartyl phosphate and NADPH to aspartyl semialdehyde, inorganic phosphate and NADP+. Aspartyl phosphate was produced by incubating 0.99 µM aspartate kinase III (YclM) with 25 mM aspartate-KOH, 10 mM ATP, 100 mM Tris (pH 8), 20 mM MgCl2 and 150 mM KCl for 50 min at 37°C and subsequently kept on ice. To phosphorylate Asd, 0.78 µM was incubated with 0.50 µM PtkA and 0.75 µM TkmA-NCter in a reaction buffer containing 37.5 mM Tris (pH 7.5), 5 mM MgCl2, 1 mM MnCl2, 5 mM ATP, 10 mg ml−1 BSA, 75 mM NaCl and 7.5% glycerol in a total volume of 30 µl for 2 h at 37°C. Samples without PtkA, TkmA-NCter or both were prepared similarly. To measure Asd activity, 900 µl aspartyl phosphate reaction was preheated to 37°C for 5 min before reaction was started by addition of 30 µl Asd preparation and NADPH to a final concentration of 0.5 mM in a 1 ml volume and absorbance at 340 nm was recorded.

Microscopy

For examination of localization of phospho-tyrosine proteins, TkmA and PtkA, cells were grown in LB at 37°C with shaking and samples were taken at OD600 0.5 (exponential) and after overnight growth of about 18 h (stationary). Samples were concentrated tenfold and 5 µl were deposited on a polylysine-coated glass slide (Thermo Scientific) and examined using a Xeiss Axioplan microscope equipped with a Kappa ACC 1 condenser, a Zeiss Plan Neofluor 100 × objective and a Kappa DX2 HC-FW camera. Images were acquired using Kappa Imagebase Control 2.7.2 software. For each condition presented in figures, about 200 individual cells were examined and a representative sample was chosen.

Acknowledgements

This work was supported by grants from the Danish National Research Council (FNU), the Lundbeckfonden and the Institut National de Recherche Agronomique (INRA) to IM, the Landesstiftung BW to BM and a PhD stipend from the Technical University of Denmark (DTU) to CJ. We are grateful to Prof Flemming G. Hansen for help with microscopy, Dr Sünje Pamp for providing vectors coding for fluorescent proteins, to Peter Boldsen Knudsen and René Jonsgaard Larsen for producing the chimeric kinase during their student project and to Silke Wahl for outstanding technical support.

Ancillary