Antagonistic effects of dual PTS-catalysed phosphorylation on the Bacillus subtilis transcriptional activator LevR

Authors

  • Isabelle Martin-Verstraete,

    1. Unité de Biochimie Microbienne, Institut Pasteur, URA 1300 du CNRS, 25, rue du Docteur Roux, 75724 Paris Cedex 15, France.,
    Search for more papers by this author
    • †The two first authors contributed equally to this work

  • Véronique Charrier,

    1. Institut de Biologie et Chimie des Protéines, UPR 412 du CNRS, 7, Passage du Vercors, 69367 Lyon Cedex 07, France.,
    Search for more papers by this author
    • †The two first authors contributed equally to this work

  • Jörg Stülke,

    1. Unité de Biochimie Microbienne, Institut Pasteur, URA 1300 du CNRS, 25, rue du Docteur Roux, 75724 Paris Cedex 15, France.,
    Search for more papers by this author
  • Anne Galinier,

    1. Institut de Biologie et Chimie des Protéines, UPR 412 du CNRS, 7, Passage du Vercors, 69367 Lyon Cedex 07, France.,
    Search for more papers by this author
  • Bernhard Erni,

    1. Department für Chemie und Biochemie, Universität Bern, Friestrasse 3, CH-3012 Bern, Switzerland.
    Search for more papers by this author
  • Georges Rapoport,

    1. Unité de Biochimie Microbienne, Institut Pasteur, URA 1300 du CNRS, 25, rue du Docteur Roux, 75724 Paris Cedex 15, France.,
    Search for more papers by this author
  • Josef Deutscher

    1. Institut de Biologie et Chimie des Protéines, UPR 412 du CNRS, 7, Passage du Vercors, 69367 Lyon Cedex 07, France.,
    Search for more papers by this author

Isabelle Martin-Verstraete. E-mail iverstra@pasteur.fr; Tel. (1) 4568 8809; Fax (1) 4568 8938.

Abstract

LevR, which controls the expression of the lev operon of Bacillus subtilis, is a regulatory protein containing an N-terminal domain similar to the NifA/NtrC transcriptional activator family and a C-terminal domain similar to the regulatory part of bacterial anti-terminators, such as BglG and LicT. Here, we demonstrate that the activity of LevR is regulated by two phosphoenolpyruvate (PEP)-dependent phosphorylation reactions catalysed by the phosphotransferase system (PTS), a transport system for sugars, polyols and other sugar derivatives. The two general components of the PTS, enzyme I and HPr, and the two soluble, sugar-specific proteins of the lev-PTS, LevD and LevE, form a signal transduction chain allowing the PEP-dependent phosphorylation of LevR, presumably at His-869. This phosphorylation seems to inhibit LevR activity and probably regulates the induction of the lev operon. Mutants in which His-869 of LevR has been replaced with a non-phosphorylatable alanine residue exhibited constitutive expression from the lev promoter, as do levD or levE mutants. In contrast, PEP-dependent phosphorylation of LevR in the presence of only the general components of the PTS, enzyme I and HPr, regulates LevR activity positively. This phosphorylation most probably occurs at His-585. Mutants in which His-585 has been replaced with an alanine had lost stimulation of LevR activity and PEP-dependent phosphorylation by enzyme I and HPr. This second phosphorylation of LevR at His-585 is presumed to play a role in carbon catabolite repression.

Introduction

The phosphoenolpyruvate–carbohydrate phosphotransferase system (PTS) is responsible for the uptake and concomitant phosphorylation of a number of sugars in both Gram-negative and Gram-positive bacteria (for a review, see Postma et al. (1993). The phosphorylation cascade from phosphoenolpyruvate (PEP) to the sugar involves enzyme I (EI), a histidine-containing phosphocarrier protein (HPr), and a sugar-specific multidomain enzyme II (EII). The EII is either a single polypeptide with several domains or several different polypeptides (Reizer and Saier, 1997). The PTS is involved in several regulatory mechanisms, such as chemotaxis, transcriptional regulation and carbon catabolite repression (Saier, 1989; Postma et al., 1993). The glucose-specific EIIA is central to carbon catabolite repression in Escherichia coli, whereas HPr plays this role in Bacillus subtilis. The HPr protein of Gram-positive bacteria can be phosphorylated at two different sites: (i) the catalytic histidine-15 is phosphorylated by EI in the presence of PEP necessary for sugar transport; and (ii) the regulatory seryl residue 46 is phosphorylated by an ATP-dependent protein kinase whose activity is stimulated by glycolytic intermediates such as fructose-1,6-bisphosphate (for a review, see Reizer et al., 1993).

Phosphorylation of HPr at Ser-46 is required for carbon catabolite repression (Deutscher et al., 1994). The pleiotropic catabolite repressor CcpA is able to form a complex with P-Ser-HPr, allowing CcpA to bind to the catabolite responsive element found in many catabolic genes of Gram-positive bacteria (Deutscher et al., 1995; Fujita et al., 1995; Hueck and Hillen, 1995; Gösseringer et al., 1997).

The PTS also controls the expression of genes encoding enzymes of PTS-dependent metabolic pathways, such as the β-glucoside operons in E. coli and B. subtilis, the ptsG gene encoding IICBAGlc, the sucrose regulon and the levanase (lev ) operon of B. subtilis (Mahadevan and Wright, 1987; Crutz et al., 1990; Martin-Verstraete et al., 1990; Arnaud et al., 1992; Le Coq et al., 1995; Stülke et al., 1997). These operons are regulated by two classes of positive regulators: the expression of the β-glucoside, glucose and sucrose systems is controlled by the antiterminators BglG, LicT, GlcT, SacT and SacY, whereas the expression of the lev operon is activated by a NifA/NtrC type regulator, LevR (Débarbouilléet al., 1991; Schnetz et al., 1996; Rutberg, 1997; Stülke et al., 1997). Interestingly, all these regulators whose activity is modulated by the PTS contain a conserved domain (Tortosa et al., 1997).

The PTS exerts its regulatory effect at three different levels: (i) the inducers of all these operons are transported via the PTS; (ii) the phosphorylated sugar-specific components of the PTS act as negative regulators of the expression of the respective operons, modulating transcription in response to the availability of the inducer (Mahadevan and Wright, 1987; Crutz et al., 1990; Martin-Verstraete et al., 1990; Arnaud et al., 1992; Le Coq et al., 1995); and (iii) the activity of the antiterminators SacT and LicT and the transcriptional activator LevR are positively regulated by EI and HPr, which are the general components of the PTS (Arnaud et al., 1992; Stülke et al., 1995; Krüger et al., 1996). Positive regulation by HPr may be involved in carbon catabolite repression, as proposed for LevR and LicT.

Positive and negative controls are mediated via PEP-dependent phosphorylation. Activation of the transcriptional regulators requires phosphorylation by EI and HPr, whereas phosphorylation via EI, HPr and the corresponding sugar-specific EII leads to inhibition of the regulator. The EIIBgl (BglF) phosphorylates the antiterminator BglG reversibly, thereby mediating induction by β-glucosides (Amster-Choder et al., 1989; Schnetz and Rak, 1990). Biochemical evidence for the phosphorylation of SacT, SacY, LevR and LicT by HPr has also been obtained (Arnaud et al., 1996; Deutscher et al., 1997; Stülke et al., 1995; Tortosa et al., 1997). The regulation of SacY activity is different, as the phosphorylation of SacY by HPr at His-99 probably has a negative regulatory role (Tortosa et al., 1997).

The levanase operon of B. subtilis (levDEFG sacC ) encodes a fructose-specific PTS (lev-PTS) and the extracellular levanase, which hydrolyses fructose polymers and sucrose (Martin et al., 1987; Martin-Verstraete et al., 1990). Together with EI and HPr, LevD and LevE have been shown to form a phosphorylation chain. LevD is phosphorylated by HPr, and P-LevD transfers the phosphoryl group to LevE (Charrier et al., 1997). The expression of this operon is induced by fructose and repressed by glucose (Martin-Verstraete et al., 1995). Fructose induction depends on the NifA-type activator LevR (Débarbouilléet al., 1991), which is negatively regulated by the fructose-specific enzymes IIA/IIB of the PTS encoded by levD and levE and positively controlled by HPr (Martin-Verstraete et al., 1990; Stülke et al., 1995; Charrier et al., 1997). Phosphorylation of LevR by EI and HPr has been demonstrated previously (Stülke et al., 1995). In the LevR protein, the regulatory part of the protein appears to have evolved by duplication and can be divided into domains B and C, which are similar to each other. Domains B and C of LevR contain the duplicated P1 and P2 domains found in all the antiterminators (Tortosa et al., 1997). Domain B is necessary for the stimulation by HPr, and domain C was thought to be the target for negative regulation by the lev-PTS (Stülke et al., 1995).

To understand the complex mechanisms of the regulatory interactions between PTS components and the LevR protein, we synthesized domains B and C of LevR separately and demonstrated their phosphorylation. Using genetic and biochemical approaches, we identified the phosphorylation sites of LevR. Phosphorylation at His-585 in domain B requires only EI and HPr. In contrast, His-869 in domain C is only phosphorylated if the fructose-specific PTS components LevD and LevE are also present.

Results

PEP and the general PTS components phosphorylate domain B of LevR

Genetic data indicate that LevR activity is controlled by the PTS via two mechanisms. Positive control is mediated by EI and HPr, whereas negative control also requires two components of the lev-PTS (Stülke et al., 1995). LevR was phosphorylated in vitro by PEP, EI and HPr. On the basis of data obtained with truncated LevR proteins, it has been suggested previously that the regulatory domain B (amino acids 411–689) is necessary for HPr-mediated stimulation of LevR activity, whereas domain C (amino acids 785–938) of LevR (Fig. 1) was predicted to be the target for lev-PTS-mediated induction by fructose (Stülke et al., 1995). To confirm the role of domain B in HPr-dependent regulation of LevR, a set of truncated LevR polypeptides was tested for phosphorylation in vitro by labelled PEP, EI and HPr (Fig. 2). LevR798 and LevR723 containing part or none of domain C were still phosphorylated by EI and HPr (Fig. 2, lanes 3 and 4). LevR407, which contains neither domain B nor domain C, was not phosphorylated (Fig. 2, lane 6). This confirms the genetic data obtained by Stülke et al. (1995) and indicates that domain B is the target of HPr-dependent regulation. However, PEP-dependent phosphorylation of domain C by EI, HPr, LevD and LevE remained to be demonstrated.

Figure 1.

. Alignment of the regions around the conserved histidines in LevR and members of the BglG family of bacterial antiterminators. The map indicates the location of the domains of the LevR protein. Domain A is similar to the NifA/NtrC-like regulators. Domains B and C resemble a family of bacterial antiterminator proteins. Bars indicate the location of the histidyl residues modified by site-directed mutagenesis in this work. Alignment of amino acid sequences around His-506, His-567, His-585 and His-869 of LevR and of the corresponding sequences in the antiterminators is shown. Identical amino acids are boxed. The conserved histidyl residues are indicated by an arrow. The modification of amino acids indicated by asterisks led to constitutive expression of the operons controlled by the corresponding antiterminators (Crutz et al., 1990; Débarbouilléet al., 1990).

Figure 2.

. In vitro [32P]-PEP-dependent phosphorylation of truncated LevR by EI and HPr. Autoradiogram after a 6% SDS–PAGE of samples containing the following proteins: lane 1, EI and HPr; lane 2, EI, HPr and wild-type LevR; lane 3, EI, HPr and LevR798; lane 4, EI, HPr and LevR723; lane 5, EI, HPr and a crude extract without LevR; lane 6, EI, HPr and LevR407. The detailed phosphorylation conditions are described in Experimental procedures and in Charrier et al. (1997). After a 20 min incubation at 37°C, sample buffer was added and proteins were separated by SDS–PAGE followed by autoradiography. The positions of phosphorylated EI, HPr, LevR and truncated LevR are indicated. The electrophoresis conditions (6% polyacrylamide) allowed separation of EI from the variously truncated LevR. In the case of LevR723 (lane 4), only one-fifth of the reaction mixture was loaded on the gel to get it well separated from EI. The band migrating below EI is caused by impurity present in the LevR preparations (see also Figs 4 and 5). Under the experimental conditions used, HPr had migrated out of the gel.

EI, HPr, LevD and LevE are required for the phosphorylation of domain C of LevR

To demonstrate that EI and HPr phosphorylate domain B and to investigate the role of domain C in PTS-dependent regulation, we synthesized the domains separately. Plasmids pRL56 and pRL78 encoding GST fused to domain B or domain C, respectively, were constructed. The corresponding polypeptides were purified as described in Experimental procedures.

The purified domain B of LevR was phosphorylated in the presence of [32P]-PEP, EI and HPr. This phosphorylation was not affected by the presence of LevD or LevE (data not shown). Purified GST–domain C was not phosphorylated by [32P]-PEP in the presence of EI and HPr (Fig. 3, lane 2) or EI, HPr and LevD (Fig. 3, lane 3). The presence of all the four PTS proteins, EI, HPr, LevD and LevE, was required for phosphorylation of the GST–domain C hybrid (Fig. 3, lane 4). These four PTS proteins have previously been shown to form a phosphorylation chain in vitro (Charrier et al., 1997). Thus, domain B is phosphorylated by EI and HPr, whereas phosphorylation of domain C requires the presence of LevD and LevE.

Figure 3.

. In vitro [32P]-PEP-dependent phosphorylation of purified GST–domain B and GST–domain C of LevR. Autoradiogram after a 10% SDS–PAGE of samples containing the following proteins: lane 1, EI and HPr; lane 2, EI, HPr and GST–domain C; lane 3, EI, HPr, LevD(His)6 and GST–domain C; lane 4, EI, HPr, LevD(His)6, LevE(His)6 and GST–domain C; lane 5, EI, HPr and GST–domain C H869A; lane 6, EI, HPr, LevD(His)6 and GST–domain C H869A; lane 7, EI, HPr, LevD(His)6, LevE(His)6 and GST–domain C H869A. As the main purpose of these experiments was to demonstrate a phosphorylation of GST–domain C, experimental conditions that clearly allowed the separation of EI from GST–domain C but barely separated HPr, LevD and LevE were chosen. The phosphorylation chain formed by the four PTS proteins has been described explicitly previously (Charrier et al., 1997), and the optimized conditions (0.2 μg EI, 0.1 μg HPr, 0.5 μg LevD and 5 μg LevE) were used to test the phosphorylation of 20 μg of purified GST–domain C polypeptide.

Mutation of His-869 of LevR leads to the loss of lev-PTS-catalysed phosphorylation

The phosphate bonds in phosphorylated LevR were rapidly hydrolysed under acidic conditions but not under alkaline conditions (pH higher than 10), suggesting that the phosphoryl groups in this protein are bound to a histidine or a cysteine forming either a phosphoamidate or thiophosphate bond (Mattoo et al., 1984; Postma et al., 1993). As the conserved regulatory domain of the PTS-regulated proteins, including the BglG family of antiterminators and LevR, does not contain a conserved cysteine, the amino acid phosphorylated by the PTS in LevR is probably a histidine. Domain C of LevR contains four histidines. However, only His-869 is conserved among domain C and members of the SacT/BglG family of antiterminators (Fig. 1). Conserved histidine is also phosphorylated in SacY, LicT and BglG (Chen et al., 1997; Deutscher et al., 1997; Tortosa et al., 1997).

To test whether His-869 of LevR is phosphorylated, a mutation was introduced into the GST–domain C hybrid protein, replacing His-869 with a non-phosphorylatable alanine residue. The protein was purified and tested for phosphorylation. Unlike the wild-type GST–domain C hybrid, the corresponding H869A modified protein was not phosphorylated by EI, HPr, LevD and LevE (Fig. 3, lanes 4 and 7), suggesting that His-869 is the site of negative regulation by the lev-PTS.

Mutation of His-869 of LevR leads to constitutive expression of the lev operon

To test the involvement of His-869 in the regulation of LevR activity, the levR allele containing the H869A mutation was integrated into the amyE locus of strain QB5182, which contains a deletion of levR and a sacC′–lacZ transcriptional fusion. Expression of the lacZ fusion in the levR H869A mutant was compared with that in the isogenic levR+ and levR8 (encoding LevR truncated at position 798) strains as controls. The different levR alleles were also introduced into isogenic strains carrying either a ΔptsGHI or a levE7 mutation. Both negative and positive control by the PTS was abolished in the ΔptsGHI strain, whereas only negative regulation by the lev-PTS was absent in the levE7 mutant containing a nonsense mutation in levE (Martin-Verstraete et al., 1990). β-Galactosidase activities of the levR mutants were assayed after growth in CSK medium in the presence or in the absence of 0.2% fructose (Table 1). Like the levR8 mutation, the levR H869A mutation led to high constitutive expression of the sacC′–lacZ fusion. The activity of the LevR H869A polypeptide was not altered significantly in a levE7 mutant (Table 1). These results confirm that histidine-869 of LevR is the site of negative regulation by the lev-PTS. Nevertheless, LevR H869A and LevR798 were positively regulated by HPr: fivefold and 300-fold lower β-galactosidase activities, respectively, were measured in a ΔptsGHI background. In these mutants, LevR contains the presumed site of HPr-catalysed phosphorylation in domain B, as both proteins were phosphorylated by EI and HPr (Fig. 2, lane 3; data not shown).

Table 1. . Regulation of transcription of the levanase operon by the PTS in the presence of the levR gene containing point mutations. a. The modified levR alleles were introduced at the amyE locus of strains containing a levR gene disruption, a sacC′–lacZ transcriptional fusion in a wild-type ΔptsGHI and levE7 background. The corresponding strains are listed in Table 2.b. The β-galactosidase activities were measured in extracts prepared from exponentially growing cells (A600 0.7–1).Thumbnail image of

Replacement of His-506 affects LevR activity

His-506 and His-567, both located in domain B of LevR, are conserved in the regulatory domain of the antiterminators (Fig. 1) (Tortosa et al., 1997). Mutations replacing His-506 or His-567 with an alanine were, therefore, introduced in the levR gene. EI and HPr phosphorylated the LevR H506A and LevR H567A polypeptides (data not shown), suggesting that neither His-506 nor His-567 is the phosphorylation site in domain B. The levR H506A allele was also integrated into the amyE locus of a ΔlevR, a ΔlevRΔptsGHI and a ΔlevR levE7 strain, all carrying a sacC′–lacZ transcriptional fusion. The phenotype of the levR H506A mutant is complex. LevR H506A activity, measured after growth in CSK, was 13-fold higher than that observed for wild-type LevR and was induced only threefold by the presence of 0.2% fructose (Table 1). In a levE7 background, β-galactosidase activity only reached the level obtained after induction by fructose in the levR H506A mutant. Although these results suggest that His-506 is involved in the regulation of LevR, the basal activity of LevR H506A, 13-fold higher than in the wild-type LevR, argued against His-506 being the site of positive control by HPr.

We studied further the role of His-506 in the regulation of LevR activity. A double mutant, levR H506A H869A, was constructed and integrated at the amyE locus of ΔlevR, ΔlevRΔptsGHI and ΔlevR levE7 mutants. Expression of the sacC′–lacZ transcriptional fusion was measured in the resulting strains tested after growth in CSK medium, in the presence and in the absence of 0.2% fructose (Table 1). The levR H506A H869A mutant exhibited constitutive expression of the lev operon and remained threefold positively regulated by HPr (Table 1). LevR proteins truncated at position 798 or 723 and containing the H506A mutation exhibited a 65- and 18-fold lower activity in the Δpts strain than in the wild-type strain, respectively (data not shown), indicating the existence of a site in domain B other than His-506 for positive regulation by HPr.

His-585 of LevR is the site of HPr-dependent phosphorylation

To identify the site of regulation by HPr, we compared the sequence of LevR and CelR, a putative regulatory protein of cellobiose catabolism in Bacillus stearothermophilus exhibiting the strongest similarity to LevR (Lai and Ingram, 1993). CelR contains a His at the position corresponding to His-585 in LevR, unlike the BglG family of antiterminators. To test whether histidine-585 is phosphorylated by HPr, levR H585A and levR H585A H869A mutants were constructed as described in Experimental procedures.

PEP-dependent phosphorylation of partially purified LevR H585A and LevR H585A H869A was tested in vitro (Fig. 4). In these two polypeptides, His-585 is replaced by an alanyl residue. Neither was phosphorylated by EI and HPr (Fig. 4, lane 3; data not shown), suggesting that His-585 is the site of HPr-catalysed phosphorylation in LevR.

Figure 4.

. In vitro [32P]-PEP-dependent phosphorylation of mutated LevR. Autoradiogram after a 10% SDS–PAGE of samples containing the following proteins: lane 1, EI, HPr and wild-type LevR; lane 2, EI, HPr, LevD(His)6, LevE(His)6 and wild-type LevR; lane 3, EI, HPr and LevR H585A; lane 4, EI, HPr, LevD(His)6, LevE(His)6 and LevR H585A; lane 5, EI, HPr, LevD(His)6, LevE(His)6 and LevR H585A H869A; lane 6, EI, HPr, LevD(His)6 and LevR H585A; lane 7, EI, HPr, LevE(His)6 and LevR H585A; lane 8, EI, HPr, LevD(His)6, LevE(His)6 and LevR H585A.

We also tested whether the addition of LevD and LevE to EI and HPr would allow phosphorylation of LevR H585A and LevR H585A H869A. EI, HPr and LevD are not sufficient to phosphorylate these proteins, but the addition of LevE restored the phosphorylation of LevR H585A (Fig. 4, lanes 6–8). However, EI, HPr, LevD and LevE failed to phosphorylate LevR H585A H869A (Fig. 4, lane 5), indicating that His-869 is most probably the site of phosphorylation by the lev-PTS. The levR H585A allele was also integrated into the amyE site of the ΔlevR, ΔlevRΔptsGHI and ΔlevR levE7 strains, and the effect of the levR H585A mutation on the expression of the sacC′–lacZ transcriptional fusion was studied (Table 1). Being negatively regulated by the lev-PTS, but presumably missing the positive effect of EI and HPr, LevR H585A exhibited low basal activity. The activity was sevenfold and threefold higher in the Δpts and levE7 background, respectively, probably because of the loss of the negative effect of the lev-PTS on LevR H585A activity.

Direct phosphorylation of LevR by P-LevE

Although P-LevD and P-LevE are necessary for the phosphorylation of LevR at His-869, there is no evidence that the phosphoryl group is not transferred from P-HPr to His-869. P-LevD and P-LevE may only modulate HPr-dependent phosphorylation of LevR at His-869. To assess the role of LevD and LevE, P-LevE was separated from the proteins necessary for its phosphorylation as follows. EI, HPr and LevD, each carrying a polyhistidine tag, were used to phosphorylate LevE in the presence of [32P]-PEP, and P-LevE was then separated from the other proteins by binding them to Ni–NTA agarose (Fig. 5B, lanes 1 and 2). P-LevE was used in vitro to phosphorylate LevR+, LevR H869A, LevR H585A, LevR H506A H585A and LevR H506A H585A H869A. All LevR proteins containing His-869 were phosphorylated by P-LevE (Fig. 5A, lanes 3–5), whereas LevR H869A and LevR H506A H585A H869A were not (Fig. 5A, lane 6 and Fig. 5B, lane 3). These results confirm that His-869 is necessary for the phosphorylation of domain C. The absence of phosphorylation of domain B in LevR H869A by P-LevE rules out a substantial contamination of the P-LevE preparation with HPr. Phosphorylation of the LevR polypeptide presented in Fig. 5 must, therefore, be caused by direct transfer of the phosphate group from P-LevE to His-869 of LevR.

Figure 5.

. In vitro phosphorylation of LevR and mutated LevR by [32P]-P-LevE. Purification of [32P]-P-LevE and the detailed phosphorylation conditions are described in Experimental procedures. Autoradiogram after a 10% SDS–PAGE of samples containing the following proteins: A. Lane 1, purified radioactive P-LevE(His)6; lane 2, EI, HPr and wild-type LevR; lane 3, purified P-LevE(His)6 and wild-type LevR; lane 4, purified P-LevE(His)6 and LevR H585A; lane 5, purified P-LevE(His)6 and LevR H506A H585A; lane 6, purified P-LevE(His)6 and LevR H506A H585A H869A. B. Lane 1, P-LevE(His)6 before purification; lane 2, purified P-LevE(His)6; lane 3, purified P-LevE(His)6 and LevR H869A; lane 4, purified P-LevE(His)6 and LevR H585A.

Discussion

Repressor-controlled gene expression is only one of several mechanisms used by bacteria to induce catabolic genes. Other induction mechanisms described in bacteria include transcriptional activators and antiterminators (Rutberg, 1997). The two latter mechanisms were found to be associated mainly with the induction of operons encoding sugar-specific components of the PTS. The activity of such transcriptional activators and antiterminators is often regulated by PEP-dependent phosphorylation catalysed by proteins of the PTS (Deutscher et al., 1997). The antiterminator BglG is phosphorylated by the corresponding sugar-specific PTS component EIIBgl at a single histidyl residue (Chen et al., 1997), whereas several histidyl residues in LicT and SacY are phosphorylated by HPr (Deutscher et al., 1997; Tortosa et al., 1997).

We studied the transcriptional activator LevR, which controls the expression of a fructose-specific PTS and contains regulatory domains similar to those present in the antiterminators. We found that it was phosphorylated twice, once by HPr and once by the sugar-specific component LevE. Phosphorylation of LevR in domain C requires the presence of PEP, EI, HPr, LevD and LevE and substantially reduced LevR activity. Mutants defective in levD or levE (Martin-Verstraete et al., 1990) or expressing LevR, in which the presumed phosphorylation site His-869 had been replaced with an alanine, exhibit constitutive expression from the lev promoter. In all these mutants, LevR is not phosphorylated in domain C. Although we could not completely exclude that modification at His-869 prevents the phosphorylation of another histidyl residue of domain C, both in vivo and in vitro data give strong evidence that His-869 is the site of phosphorylation by the lev-PTS. Phosphorylation of His-869 therefore seems to regulate induction of the lev operon in response to the availability of a substrate of the lev-PTS (Fig. 6). In the absence of a substrate of the lev-PTS, fructose or mannose (Martin-Verstraete et al., 1990; 1996), the sugar-specific components LevD and LevE, although produced in low amounts, are assumed to be present mainly in the phosphorylated form. This allows phosphorylation and inactivation of LevR by P-LevE, and the lev operon will be poorly expressed. In contrast, in the presence of fructose, P-LevE is assumed to transfer its phosphate group mainly to the incoming sugar, leading to dephosphorylation and activation of LevR. Interestingly, like the IIB domain of EIIBgl (Chen et al., 1997), P-LevE seems to be able to transfer its phosphate group either to a histidyl residue of a regulatory protein or to a hydroxyl group of a sugar. To demonstrate unequivocally that P-LevE is the phosphate donor for the phosphorylation of LevR and to exclude the possibility that one of the other PTS proteins present in the phosphorylation assay can phosphorylate domain C of LevR when P-LevE is present, we separated P-LevE from the other three PTS proteins necessary for its phosphorylation (Charrier et al., 1997). The results obtained with purified P-LevE and several LevR variants confirmed that P-LevE is the phosphate donor for phosphorylation of the domain C of LevR at His-869.

Figure 6.

. Proposed model for LevR regulation by dual PTS-catalysed phosphorylation. Phosphorylation at His-585 stimulates LevR activity and is catalysed by P-His-HPr. It competes with the sugar-specific enzyme II of the PTS for the common phosphoryl donor P-His-HPr. The presence of rapidly metabolizable PTS sugars will, therefore, lower the proportion of molecules phosphorylated at His-585, and LevR will be less active. This modification of LevR is therefore assumed to play a role in carbon catabolite repression of the lev operon exerted by PTS sugars. Phosphorylation at His-869 inhibits LevR activity and is catalysed by P-LevE. It competes with fructose for the common phosphoryl donor P-LevE. The presence of fructose will lead to dephosphorylation and activation of LevR. P-LevE-catalysed phosphorylation of His-869 of LevR seems, therefore, to be involved in the induction of the lev operon. In agreement with this model, mutations affecting one of the four proteins forming the lev-PTS phosphorylation chain or His-869 of LevR lead to constitutive expression of the lev operon.

Phosphate transfer between proteins of the PTS is usually considered to be reversible (Postma et al., 1993). If this was the case for the lev-PTS, dephosphorylation of the PTS components caused by any rapidly metabolizable PTS substrate would lead to dephosphorylation of LevR and, consequently, to induction of the lev operon. Interestingly, it has been observed that, although P-His-HPr can phosphorylate LevD rapidly, the phosphate group transfer from P-LevD to HPr is very slow (Charrier et al., 1997). This unidirectional phosphoryl group transfer, possibly resulting from steric hindrance of the interaction of P-LevD with HPr, could allow LevD, LevE and LevR to remain phosphorylated even when a rapidly metabolizable PTS substrate, such as glucose or mannitol, is present in the growth medium.

The regulatory domain of LevR seems to have evolved by gene duplication, as the two domains (B and C) exhibit extended similarity. Interestingly, we could not demonstrate that His-506 in LevR (corresponding to His-869 in domain B) was phosphorylated by the PTS, although mutations replacing His-506 with an alanine affected LevR activity. These effects could be caused by structural changes accompanying the mutation at His-506. However, we cannot exclude the possibility that His-506 is phosphorylated by intramolecular transfer of a phosphate group from either His-585 or His-869. The regulatory domain of LevR is also similar to the regulatory domains of antiterminators (Tortosa et al., 1997). Anti-terminators usually contain two histidyl residues corresponding to His-869, one located in subdomain P1 and one in subdomain P2. In the case of LicT and SacY, these two histidyl residues, His-100 (His-99 for SacY) and His-207, are phosphorylated by P-His-HPr (Deutscher et al., 1997; Tortosa et al., 1997). As HPr and LevE interact with the same PTS protein, LevD, it is possible that the active sites of these two proteins have a similar secondary structure that allows them to interact with the same conserved phosphorylation sites in transcriptional activators and antiterminators.

PEP-dependent phosphorylation of domain B of LevR requires the presence of EI and HPr, but not of LevD and LevE. The absence of phosphorylation of LevR H585A by EI and HPr strongly suggests that this histidyl residue is the site of phosphorylation by HPr-His-P. According to genetic data, phosphorylation in domain B stimulates LevR activity. Mutants defective in EI or HPr or expressing LevR, in which His-585 has been replaced with an alanine, exhibited LevR activity that is lower than that in levD or levE mutants (Stülke et al., 1995). PEP-dependent phosphorylation in domain B is probably a mechanism for catabolite repression of the lev operon. In the presence of rapidly metabolizable PTS sugars, such as glucose or mannitol, LevR has to compete with the corresponding EIIs for the phosphate group on P-His-HPr, which will be transferred mainly via the sugar-specific PTS components to the incoming sugar. As a consequence, domain B of LevR will not be phosphorylated and, hence, will be present in a less active form, even when fructose is present in the growth medium. Genetic data have been used to propose a very similar carbon catabolite repression mechanism for the bgl operon of B. subtilis (Krüger et al., 1996). The activity of LicT, the antiterminator controlling the expression of the B. subtilis bgl operon, is completely dependent on a functional PTS. LicT is phosphorylated by PEP, EI and HPr at histidyl residues (Deutscher et al., 1997), and this phosphorylation reaction has been suggested to be used as a carbon catabolite repression mechanism for the bgl operon. Interestingly, alignment of the regulatory domains of LevR and the antiterminators indicates that His-585 corresponds to any of the histidyl residues phosphorylated in the antiterminators (Tortosa et al., 1997). However, there is a histidine corresponding to His-585 in the transcriptional regulator CelR that is believed to control expression of the B. stearothermophilus cel operon (Lai and Ingram, 1993).

In summary, we have demonstrated that the transcriptional activator LevR is phosphorylated by two PEP-dependent phosphorylation reactions, according to the model presented in Fig. 6. His-585 has been identified as the probable phosphorylation site for EI and HPr, whereas phosphorylation by LevE-P most probably occurs at His-869. Phosphorylation of domain B by P-His-HPr stimulates LevR activity and is probably involved in a CcpA-independent carbon catabolite repression mechanism. Phosphorylation at His-869 by P-LevE substantially reduces LevR activity and plays a role in the induction of the B. subtilis lev operon. Further detailed studies will be necessary to elucidate the effects of the two PEP-dependent PTS-catalysed phosphorylation reactions on the activation of transcription of the lev promoter by LevR.

Experimental procedures

Bacterial strains and culture media

Bacterial strains used in this work are listed in Table 2. E. coli TGI [K-12 Δ(lac pro ) supE thi hsd5/F′traD36 proA+B +lacI qlacZΔM15] was used for cloning experiments, E. coli CJ236 (dut ung thi relA/pJC105 [CmR]) for oligonucleotide-directed mutagenesis and E. coli BL21 [ompT hsdSB (mbrb) gal dcm] for overproduction of proteins. E. coli was grown in LB broth and B. subtilis in SP medium and C minimal medium. CSK is C medium supplemented with potassium succinate (6 g l−1) and potassium glutamate (8 g l−1) (Martin-Verstraete et al., 1990). LB and SP plates were prepared by the addition of 17 g l−1 Bacto agar (Difco).

Table 2. . B. subtilis strains (Martin-Verstraete et al. (1995); (1994); Stülke et al. (1995)). b. Arrows indicate construction by transformation.Thumbnail image of

Transformation and phenotypic characterization

Standard procedures were used to transform E. coli (Sambrook et al., 1989), and transformants were selected on LB plates supplemented with ampicillin (100 μg ml−1). B. subtilis was transformed as described previously with plasmid or chromosomal DNA (Kunst and Rapoport, 1995), and transformants were selected on SP plates containing chloramphenicol (Cm; 5 μg ml−1), kanamycin (Km; 5 μg ml−1), erythromycin (Em; 1 μg ml−1) plus lincomycin (Lin; 10 μg ml−1), spectinomycin (Spc; 60 μg ml−1) or tetracycline (Tc; 15 μg ml−1).

Amylase activity of B. subtilis was detected after growth on tryptose blood agar base (TBAB; Difco) supplemented with 10 g l−1 hydrolysed starch (Connaught). Starch degradation was detected by sublimating iodine onto the plates.

lacZ expression in liquid medium was studied as follows. Cells were grown in CSK medium supplemented with 2 g l−1 fructose. Cells were harvested at an optical density at 600 nm (OD600) of 0.6–0.8 for cultures grown in CSK and an OD600 of 1 for cultures grown in CSK with sugar. β-Galactosidase activities were determined according to the method of Miller (1972) with cell extracts obtained by lysozyme treatment. One unit of β-galactosidase activity is defined as the amount of enzyme that produces 1 nmol of o-nitrophenol per min at 28°C. The specific activities are expressed as mean values derived from three to five independent experiments.

DNA manipulations

Standard procedures were used to extract plasmids from E. coli (Sambrook et al., 1989). Restriction enzymes, phage T4 DNA polymerase, phage T4 DNA ligase and T4 polynucleotide kinase were used, as recommended by the manufacturers. DNA fragments were purified from agarose gel with a Prep-A-Gene kit (Bio-Rad).

To perform site-directed mutagenesis, the Muta-Gene M13 in vitro mutagenesis kit (Bio-Rad) was used, based on the method described by Kunkel et al. (1987). M13mp18C1 phage DNA (Martin-Verstraete et al., 1994), containing a 4.8 kb PstI–SacI fragment of the levR gene, was used to introduce point mutations into the codons of histidine residues conserved in the BglG/SacT family of regulators (Fig. 1) and of His-585. Double and triple mutants in the levR gene were obtained by the same method using derivatives of the M13mp18C1 carrying one or two mutations in levR as the template for the mutagenesis. The presence of the mutations was confirmed by DNA sequencing by the dideoxy chain termination method (Sanger et al., 1977).

Plasmid constructions

Vector pAC5 (Martin-Verstraete et al., 1992) carries the pC194 chloramphenicol resistance gene cat and a lacZ gene between two fragments of the B. subtilis amyE gene. This vector was used to introduce single copies of the different alleles of the levR gene into the chromosome of B. subtilis. The following constructions were performed: M13 replicative forms corresponding to phages carrying the different levR alleles were extracted and digested with SacI and BamHI. The SacI–BamHI fragments containing the levR alleles were ligated into pAC5 leading to a partial deletion of lacZ. The pAC5 derivatives are listed in Table 3.

Table 3. . pAC5 and PBQ200 derivatives containing different levR alleles.Thumbnail image of

The levR genes containing various point mutations were integrated at the amyE locus of the B. subtilis chromosome of strain QB5182 (trpC2ΔlevR ::aphA3 sacC′–lacZ +erm) or QB5075 (trpC2 levE7ΔlevR ::aphA3 sacC′–lacZ +erm) (see Table 2).

The levR alleles were also inserted into pBQ200 containing the degQ36 promoter (Martin-Verstraete et al., 1994). pAC5 derivatives containing different levR alleles were digested with Bst BI, made blunt using the Klenow fragment of DNA polymerase I and digested with BamHI. The Bst BI–BamHI fragments were ligated between the SmaI and BamHI sites of pBQ200 to give the plasmids listed in Table 3.

The domains B and C of LevR were fused to the GST protein as follows: the part of the levR gene corresponding to amino acids 407–723 (domain B) or 723–938 (domain C) was amplified by polymerase chain reaction (PCR) using oligonucleotides creating BamHI and EcoRI restriction sites. The BamHI–EcoRI fragments were inserted into the expression vector pGexKT (Hakes and Dixon, 1992). The resulting plasmids, pRL56 and pRL78, encode domains B and C, respectively, of LevR fused to the GST protein. Plasmid pRL80 encodes a GST fusion to domain C carrying the H869A mutation.

Protein purification

Enzyme I of the PTS was purified from Staphylococcus carnosus, and HPr was purified from B. subtilis as described previously (Eiserman et al., 1988; Kohlbrecher et al., 1992). The B. subtilis proteins EI(His)6, HPr(His)6, LevD(His)6 and LevE(His)6 were also purified as described previously (Charrier et al., 1997). LevE without a His-tag was prepared as described in Seip et al. (1997).

For the partial purification of LevR, the pBQ200 derivative plasmids carrying the various levR alleles were introduced into strain QB5277 (Table 2; Stülke et al., 1995) containing both deletions of the levR gene and the genes encoding the general and lev-specific components of the PTS. The resulting strains overproduce LevR during stationary phase after growth in LB medium. The various modified LevR polypeptides were partially purified from crude extracts by ammonium sulphate precipitation as described previously (Martin-Verstraete et al., 1994). The modified LevR polypeptides made up about 5% of the total proteins.

To purify GST–domain B, GST–domain C and GST–domain C H869A, plasmids pRL56, pRL78 and pRL80 were introduced into E. coli strain BL21 (Novagen) containing the plasmid pREP4 groESL and, thus, overproducing the bacterial chaperones GroES and GroEL (Amrein et al., 1995). Transformants were grown at 37°C in LB medium until the culture reached an OD600 of 1. The fusion genes present in pGexKT were induced by the addition of 0.15 mM IPTG and incubation for a further 3 h at room temperature.

The cells were pelleted by centrifugation, resuspended in TPBS buffer (140 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 1.8 mM KH2PO4) and disrupted by sonication. Proteins were purified as described by Hakes and Dixon (1992). Fractions containing GST–domain B or GST–domain C were pooled. Protein concentrations were determined spectrophotometrically (Bio-Rad protein assay). Protein solutions were stored at −20°C.

To prepare P-LevE without a His-tag, the same phosphorylation conditions were used as for LevE(His)6 (Charrier et al., 1997). After phosphorylation of LevE, the phosphorylation mixture was treated three times with Ni–NTA resin to separate phosphorylated LevE from EI(His)6, HPr(His)6 and LevD(His)6.

In vitro phosphorylation of LevR

[32P]-PEP was prepared from [γ-32P]-ATP as described previously (Roossien et al., 1983). [32P]-PEP was separated from [γ-32P]-ATP by ion exchange chromatography (Mattoo and Waygood, 1983). Partially purified LevR (30 μg) or purified GST–domain polypeptides (20 μg) were phosphorylated by incubation in a 25 μl reaction mixture containing EI (0.2 μg), HPr (0.1 μg), MgCl2 (10 mM), Tris-HCl (50 mM, pH 7.4) and 1 μM [32P]-PEP (0.5 μCi) for 20 min at 37°C. For the phosphorylation of LevR or GST–domain C by LevE, LevD(His)6 (0.5 μg) and LevE(His)6 (5 μg) were added to this reaction mixture. The optimal conditions of phosphorylation of LevD and LevE by EI, HPr and PEP have been studied extensively (Charrier et al., 1997). The reaction was stopped by adding electrophoresis sample buffer containing 0.1% SDS. Electrophoresis of proteins was carried out on SDS–polyacrylamide gels. Gels were dried and exposed for 2–16 h to autoradiography (Kodak Biomax).

Footnotes

  1. †The two first authors contributed equally to this work

Acknowledgements

We are grateful to W. Hengsterberg for providing us with enzyme I from S. carnosus, to M. Débarbouillé and A. Klier for helpful discussions and to J. Bignon for excellent technical assistance. C. Dugast is acknowledged for typing the manuscript, A. Bosch, C. van Herrewege and Alex Edelman for their assistance in preparing this manuscript. This research was supported by grants from the Centre National de la Recherche Scientifique, The Pasteur Institute, The University Paris 7 and by the European Community Biotech Program BIO4-CT96-0380.

Ancillary