Control of the Bacillus subtilis Antiterminator Protein GlcT by Phosphorylation

Bacillus subtilis transports glucose by the phosphotransferase system (PTS). The genes for this system are encoded in the ptsGHI operon, which is induced by glucose and depends on a termination/antitermination mechanism involving a riboswitch and the RNA-binding antitermination protein GlcT. In the absence of glucose, GlcT is inactive, and a terminator is formed in the leader region of the ptsG mRNA. If glucose is present, GlcT can bind to its RNA target and prevent transcription termination. The GlcT protein is composed of three domains, an N-terminal RNA binding domain and two PTS regulation domains, PTS regulation domain (PRD) I and PRD-II. In this work, we demonstrate that GlcT can be phosphorylated by two PTS proteins, HPr and the glucose-specific enzyme II (EIIGlc). HPr-dependent phosphorylation occurs on PRD-II and has a slight stimulatory effect on GlcT activity. In contrast, EIIGlc phosphorylates the PRD-I of GlcT, and this phosphorylation inactivates GlcT. This latter phosphorylation event links the availability of glucose to the expression of the ptsGHI operon via the phosphorylation state of EIIGlc and GlcT. This is the first in vitro demonstration of a direct phosphorylation of an antiterminator of the BglG family by the corresponding PTS permease.

EII), which may exist as individual proteins or fused in a single polypeptide. In E. coli, the glucose-specific EII is composed of a membrane-bound protein comprising the actual transporter domain EIIC and the phosphotransfer domain EIIB. In addition, the EIIA domain is present as a cytoplasmic protein. In B. subtilis, all domains of the glucose permease are fused to form a single polypeptide with the domain arrangement EIICBA (1,2).
It was long considered that the genes encoding the components of the glucose PTS are constitutively expressed in bacteria. Although this is the case for the genes encoding the general proteins, ptsI and ptsH, the gene encoding EII Glc , ptsG, is induced by glucose in both E. coli and B. subtilis (1)(2)(3). In E. coli, regulation of ptsG expression is accomplished by the transcriptional repressor Mlc. In the presence of glucose, this repressor is sequestered by the non-phosphorylated EIICB Glc and, thus, is unable to repress ptsG transcription (3). In B. subtilis, glucose induction of ptsG expression is mediated by transcriptional antitermination. In the absence of glucose, transcription initiated at the ptsG promoter is terminated in the leader region of the mRNA. If glucose is present, an antitermination protein, GlcT, is active and prevents transcription termination by binding to the RNA antiterminator (RAT), which overlaps the terminator. Binding of GlcT to the RAT is thought to stabilize the RAT structure and to prevent formation of the terminator (4,5).
GlcT belongs to the BglG family of transcriptional antiterminators. These proteins are composed of an N-terminal RNA binding domain (about 60 amino acids) and two reiterated PTS regulation domains (PRD) that modulate the regulatory output of the protein in response to the availability of the inducer (6,7). In the E. coli BglG and the B. subtilis LicT protein, the second PRD (PRD-II) is phosphorylated on conserved histidine residues by the HPr protein of the PTS in the absence of glucose. This phosphorylation results in a structural change of the protein and stimulates the activity of these antiterminator proteins. In contrast, phosphorylation of PRD-I by the corresponding EII is thought to inactivate these antiterminator proteins in the absence of the inducer, salicin (8 -11). The direct phosphorylation of antiterminators of the BglG family by HPr is well documented for several proteins (8,9,12,13). In contrast, the regulatory pathway leading from absence of inducer via phosphorylation of EII and PRD-I to the inactivity of the antiterminators is much less understood. The phosphotransfer from EII to an antiterminator has never been demonstrated unequivocally in vitro. Phosphorylation of antiterminators of the BglG family was proposed to affect the dimerization of the protein, which is required for interacting with the RNA (14,15).
Activity of GlcT is also controlled by PTS components. How-ever, in contrast to BglG and LicT, HPr is not required to stimulate the activity of GlcT. Constitutive activity of GlcT was observed in mutants affecting HPr or EII Glc . However, mutations of the phosphorylation sites of EII Glc did not completely abolish regulation of GlcT activity. The mechanism by which EII Glc controls GlcT activity in response to the availability of glucose has, therefore, remained a matter of speculation. It was proposed that either EII Glc phosphorylates and inactivates GlcT directly in the absence of glucose or that HPr is the actual phosphate donor in a ternary complex with GlcT and EII Glc (16).
In this work, we studied the pathways of phosphorylation of GlcT by HPr and EII Glc . Although HPr efficiently phosphorylates the PRD-II, EII Glc does directly and specifically phosphorylate PRD-I of GlcT. This is the first time that direct phosphorylation of an antiterminator of the BglG family by an EII was demonstrated. Phosphorylation in PRD-I inactivates GlcT, whereas phosphorylation of PRD-II has no significant impact on GlcT antitermination activity.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Growth Conditions-The E. coli and B. subtilis strains used in this study are listed in Table I. All B. subtilis strains are derivatives of the wild type strain 168. B. subtilis was grown in CSE minimal medium (70 mM K 2 HPO 4 , 30 mM KH 2 PO 4 , 25 mM (NH 4 ) 2 SO 4 , 0.5 mM MgSO 4 , 10 M MnSO 4 , 22 mg ferric ammonium citrate/liter, sodium succinate (6 g l Ϫ1 ) potassium glutamate (8 g l Ϫ1 )) supplemented with auxotrophic requirements (at 50 mg liter Ϫ1 ) (17). Carbon sources were added as indicated. E. coli was grown in LB medium, and transformants were selected on plates containing ampicillin (100 g ml Ϫ1 ). LB and SP plates were prepared by the addition of 17 g liter Ϫ1 Bacto agar (Difco) to the medium.
DNA Manipulation-Transformation of E. coli and plasmid DNA extraction were performed using standard procedures (18). Restriction enzymes, T4 DNA ligase, and DNA polymerases were used as recommended by the manufacturers. DNA fragments were purified from agarose gels using the Nucleospin extract kit (Macherey and Nagel). Pfu DNA polymerase was used for the PCR as recommended by the manufacturer. DNA sequences were determined using the dideoxy chain termination method (18). Chromosomal DNA of B. subtilis was isolated as described (19).
Transformation and Characterization of the Phenotype-B. subtilis was transformed with plasmid DNA and chromosomal DNA according to the two-step protocol (19). Transformants were selected on SP plates containing kanamycin (5 g ml Ϫ1 ), chloramphenicol (5 g ml Ϫ1 ), or erythromycin plus lincomycin (2 g ml Ϫ1 and 25 g ml Ϫ1 , respectively). The presence of point mutations in glcT or ptsG was verified by sequencing of the relevant alleles. Quantitative assays of lacZ expression in B. subtilis were performed with cell extracts using o-nitrophenyl ␤-galactopyranoside as the substrate (19). One unit of ␤-galactosidase is defined as the amount of enzyme that produces 1 nmol min Ϫ1 of onitrophenol at 28°C.
Expression and Mutagenesis-For overexpression of His-tagged Enzyme I and HPr of B. subtilis, the plasmids pAG3 and pAG2, respectively, were used (20). HPr carrying am N-terminal Strep tag was constructed as follows. In a first step, the ptsH gene was amplified using the oligonucleotides SB51 (5Ј-AAAGGCGCCATGGCACAAAAAACATT-TAAA) and SB46 (5Ј-AAAGGATCCTTACTCGCCGAGTCCTTCGCTT-TTCAT). The resulting PCR product was digested with KasI and BamHI (the restriction sites were introduced with the primers (underlined)) and cloned into the vector pASK-IBA5 (IBA, Göttingen, Germany) cut with the same enzymes. This plasmid, pGP151, contains the ptsH gene fused to a sequence encoding a Strep tag. In a next step the fusion was amplified using primers SB52 (5Ј CGAGGGCAACATATG-GCTAGCTGGAGC; an NdeI restriction site is underlined) and SB46. This PCR product was cloned between the NdeI and BamHI sites of the expression vector pET3c (Novagen, Inc., Madison, WI). The resulting plasmid was pGP152.
The EIIBA domains of the glucose permease were fused to an Nterminal His tag by amplifying the part of ptsG encoding these domains with the oligonucleotides SB25 (5Ј-AAAGTCGACGGTGAAGCAG-GAGATCTTCCT; a SalI site is underlined) and SB2 (5Ј-CCGG-TACGCTTTTGCATTCGC) and plasmid pTS22 (22) as a template. The PCR product was digested with SalI and HindIII and cloned into pWH844. The resulting plasmid was pGP123.
Plasmid pBQ200 was used for the expression of cloned genes in B. subtilis under control of the strong degQ36 promoter (23). The ptsG fragment encoding EIIBA was amplified using the primers SB24 (5Ј-A-AAGGATCCAGAGGAGGTCAATTCTTATGGGTGAAGCAGGAGATC-TTCTTCCTTATGAG) and SB2 (24). SB24 introduces a BamHI restriction site (underlined in the sequence), a ribosomal binding site identical to the one found upstream of the ptsG coding region, and a ATG start codon. The resulting PCR product was cloned as a BamHI/HindIII fragment into pBQ200 treated with the same enzymes. The resulting plasmid was pGP122.
Protein Purification-E. coli DH5␣, NM522, and FT1 were used as hosts for the overexpression of recombinant proteins. The cultures were grown in one liter of LB medium at 37°C (for the purification of wild type GlcT, HPr, and EI) or 23°C (mutant forms of GlcT). Expression was induced by the addition of isopropyl-1-thio-␤-D-galactopyranoside (final concentration 1 mM) to logarithmically growing cultures (A 600 of 0.8). Cells were harvested 1 h after induction. The pellets were resuspended in 30 ml of disruption buffer (10 mM Tris/HCl, 200 ml NaCl, pH 7.5). The cells were disrupted by using a French press (20,000 p.s.i., 138,000 kilopascals; Spectronic Instruments, UK). For proteins carrying a His tag, the crude extracts were passed over a 3-ml Ni 2ϩ -nitrilotriacetic acid column. The column was washed with 30 ml of disruption buffer followed by elution with an imidazole gradient. Seven different concentration of imidazole were used: 10, 25, 50, 100, 200, 300, and 500 mM imidazole in 10 ml of disruption buffer, pH 7.5. For HPr carrying a Strep tag, the crude extracts were prepared as described above. The extract was passed over a 3-ml Streptactin column (IBA, Göttingen, Germany) followed by washing with 20 ml of disruption buffer. The recombinant protein was eluted with 5 ml of disruption buffer containing desthiobiotin (Sigma, final concentration 2.5 mM). The Bio-Rad dye Laboratory collection QB5448 trpC2 amyE:: trpC2 glcT-H104D amyE::(⌬LA ptsG-lacZ aphA3) trpC2 ptsG ⌬EIIBA amyE::(⌬LA ptsGЈ-ЈlacZ aphA3) trpC2 glcT-H104D amyE:: trpC2 glcT-H104A amyE:: trpC2 glcT-H163D amyE:: trpC2 glcT-H210D amyE:: binding assay was used to determine protein concentrations. Bovine serum albumin was used as the standard. Western Blot Analysis-For Western blot analyses of HPr, proteins were separated by SDS-PAGE and transferred to a polyvinylidene difluoride membrane (Bio-Rad) by electroblotting. HPr was detected with rabbit polyclonal antiserum raised against HPr of B. subtilis (25). The antibodies were visualized by using anti-rabbit IgG-AP secondary antibodies (Chemikon International, Temecula, CA) and the CDP* detection system (Roche Diagnostics).
Protein Phosphorylation and Stability Assays of the Phosphorylated Proteins-In vitro phosphorylation of GlcT and of EIIBA Glc by EI and HPr were performed with 0.5 M [ 32 P]PEP (0.1 Ci) in a total volume of 20 l essentially as described previously (9). [ 32 P]PEP was synthesized in an enzymatic reaction using pyruvate kinase (26). [ 32 P]PEP was separated from [␥-32 P]ATP by ion-exchange chromatography (35). The stabilities of GlcT and EIIBA Glc phosphate were studied under acidic and basic conditions (27).

Phosphorylation of GlcT by Enzyme I and HPr-The
HPr protein of the PTS is capable of interacting with and of phosphorylating several other proteins in B. subtilis, including sugar specific Enzymes II of the PTS, the glycerol kinase, and PRD containing regulators (7,28,29). We asked therefore whether HPr would also directly phosphorylate GlcT. Phosphorylation assays were performed with [ 32 P]PEP in the presence of EI, HPr, or both proteins. As shown in Fig. 1, the His-tagged general PTS proteins were able to autophosphorylate (EI) and to accept the phosphate residue (HPr). Phosphorylation of GlcT was observed in the presence of both general PTS proteins, whereas no GlcT phosphorylation was detected in the absence of HPr (Fig. 1). Thus, GlcT was phosphorylated by HPr.
The PRDs of GlcT contain four strongly conserved histidine residues. In other PRD-containing regulators, these histidine residues are the sites of phosphorylation (9,27,30). To test whether histidine residues serve as phosphorylation sites in GlcT, we studied the stability of the phosphate bound. As a control, EIIBA Glc was used; it is known to be phosphorylated on a histidine and a cysteine residue in its EIIA and EIIB domains, respectively (1,24). In both proteins the phosphate bounds were stable in the presence of sodium hydroxide (Fig.  2). In contrast, the GlcT phosphorylation signal disappeared completely after incubation with hydrochloric acid, whereas some phosphorylation of EIIBA Glc was still detectable under these conditions. The strong reduction of signal intensity of phosphorylated EIIBA Glc is in good agreement with the fact that one phosphate is present as a phosphoamidate (which is acid-labile), whereas the other is bound to a cysteine (this phosphate is stable under strongly acidic conditions) (31). Thus, we may assume that a phosphoamidate is present in GlcT. Phosphorylated EI exhibited the same stability pattern as GlcT. This observation is in good agreement with the idea that the phosphate in both EI and GlcT is present as a phosphoamidate and, therefore, most likely bound to a histidyl residue.
HPr Specifically Phosphorylates PRD-II of GlcT-GlcT contains two conserved histidyl residues in each PRD. HPr might therefore phosphorylate sites in both PRDs as proposed for LicT (9) or it might specifically phosphorylate one of the two PRDs. To distinguish between these two possibilities, we purified a series of GlcT proteins with mutations at the positions of the conserved histidine residues. These mutant proteins were used for phosphorylation assays with [ 32 P]PEP in the presence of EI and HPr. GlcT proteins with mutations affecting the two histidine residues in PRD-I (His-104, His-163) were phosphorylated as efficiently as the wild type protein (Fig. 3). We may, thus, conclude that HPr does not phosphorylate the PRD-I. In contrast, a replacement of His-210 by a aspartyl residue resulted in strongly reduced phosphorylation of GlcT. Unfortunately, we were not able to test the effect of a replacement of His-272 since this mutant protein could not be purified. To verify the specific phosphorylation of PRD-II by HPr, we constructed plasmids encoding GlcT variants with double mutations of His-210 and either of the three other potential phosphorylation sites. As observed before (see Fig. 3), the replacement of His-210 resulted in a strong decrease of phosphorylation. A quantitative analysis of phosphorylation signals revealed that the H210D mutation reduced the phosphorylation to about 12%. Similarly, the H210D,H272Q double mutant protein exhibited about 10% residual phosphorylation (Fig. 3). In contrast, double mutant proteins in which one of the conserved histidyl residues of the PRD-I was replaced in addition to His-210 were no longer phosphorylated by EI and HPr (Fig.  3). We may, thus, conclude that HPr phosphorylates mainly PRD-II, whereas PRD-I is a minor substrate for HPr. However, our results do not allow us to decide whether HPr phosphorylates His-210 or His-272 or both amino acids in PRD-II. Similarly, the weak HPr-dependent phosphorylation of PRD-I may occur on either or both histidine residues (see "Discussion").
Regulatory Functions of PRD-II-The two PRDs of transcriptional antiterminators are specialized in their functions; PRD-I is thought to be involved in the response to the specific inducer, whereas PRD-II was implicated in carbon catabolite repression (7,8,32). The role of HPr-dependent phosphorylation of GlcT in the PRD-II was studied by analyzing the antitermination activity of mutant GlcT variants in vivo. For this purpose we constructed a series of strains containing a fusion of the ptsG promoter region to a promoterless lacZ gene and different glcT alleles. The glcT mutations were transferred into the chromosome of B. subtilis by transformation as described previously (16). The strains were grown in CSE minimal medium in the presence and absence of glucose, and their ␤-galactosidase activities were determined (Table II). In the wild type strain QB5448, expression of ptsG was low in the absence of glucose and strongly induced if glucose was present in the medium. Strain GP128 was isogenic with QB5448, but its glcT gene specified a protein with a H210D replacement. As observed with the wild type, this GlcT variant was inactive in the absence of glucose, suggesting that it was still inactivated as long as no inducer was added to the medium. In the presence of glucose, expression of ptsG was induced in GP128. However, the expression level was somewhat reduced in this strain as compared with the wild type (210 versus 600 units, see Table  II), suggesting that GlcT was not fully active. This may indicate that GlcT requires HPr-dependent phosphorylation on PRD-II for full activity or may result from the amino acid exchange. Because GlcT activity is reduced in a ptsH mutant devoid of HPr as compared with induced conditions in the wild type (4), we may conclude that some further activation of GlcT by HPr is required for full antitermination activity (see "Discussion").
The Role of EIIBA Glc in the Control of GlcT Activity-In contrast to HPr, which exerts a weak stimulating effect on GlcT activity, the glucose-specific EII of the PTS acts as a negative regulator of GlcT activity. A deletion of the phosphotransfer domains of EII Glc results in constitutive GlcT activity (16). To study the phosphorylation of GlcT by EII Glc , we intended to purify the soluble part of the glucose permease, EIIBA Glc , and to test whether it could phosphorylate GlcT. Before studying the activity of EIIBA Glc in vitro, we analyzed the regulatory effect of the isolated phosphotransfer domains in vivo. For this purpose, we expressed a truncated ptsG gene encoding these domains from plasmid pGP122. The effect of EIIBA Glc on GlcT activity was assayed by measuring the expression of a ptsG-lacZ fusion in the wild type strain QB5448 and in the ⌬ptsG mutant strain GP113. In this latter strain, the part of ptsG encoding the EIIBA domains was deleted from the chromosome (16). Although the expression of ptsG was inducible by glucose in the wild type strain QB5448, constitutive expression was observed in the ⌬ptsG mutant GP113. This is in good agreement with previous observations (Ref. 16; Table III). In contrast, expression of EIIBA Glc had a severe negative effect on the induction of ptsG in both genetic backgrounds (Table III). EIIBA Glc overexpression prevented ptsG induction in the wild type strain and reversed the constitutive expression of the ⌬ptsG. We may, therefore, conclude that isolated EIIBA Glc inactivates GlcT even in the presence of glucose. This may be due to the fact that this protein is not linked to the EIIC  domain that transports the glucose. Thus, large amounts of phosphorylated EIIBA Glc may accumulate in the cell, which in turn may result in a permanent inactivation of GlcT. In addition, this experiment provides evidence that EIIBA Glc alone is sufficient to productively interact with GlcT and to control its activity.
Phosphorylation of GlcT by EIIBA Glc -The data presented above demonstrate that phosphorylated EIIBA Glc is sufficient to control the activity of GlcT. Because phosphorylation of the PRD-containing antiterminators by their cognate EII was proposed to negatively control their activity (10, 33, 34), we studied phosphorylation of GlcT by EIIBA Glc . For this purpose, we incubated wild type and mutant forms of GlcT with EIIBA Glc in the presence of [ 32 P]PEP, EI, and HPr. As shown in Fig. 4, wild type GlcT was phosphorylated irrespective of the presence of EIIBA Glc . However, in the absence of EIIBA Glc , GlcT phosphorylation was reduced by about 50%. This residual phosphorylation of GlcT is due to direct HPr-dependent phosphotransfer to PRD-II (see above in "Results" under "HPr Specifically Phosphorylates PRD-II of GlcT "). The GlcT-H210D mutant protein was barely phosphorylated in the absence of EIIBA Glc , but phosphorylation was significantly increased in its presence. Thus, we may assume that EIIBA Glc may be involved in the phosphorylation of GlcT in PRD-I. This idea is supported by the finding that no GlcT phosphorylation was detectable in GlcT-H104A/H210D and GlcT-H163D/H210D double mutant proteins (Fig. 4). Thus, mutations affecting one conserved histidine residue in each PRD are sufficient to abolish phosphorylation by both HPr and EIIBA Glc .
Mutations in PRD-I of GlcT Prevent Inactivation of the Protein-It has previously been shown that a mutation affecting the conserved His-104 in PRD-I of GlcT results in constitutive antitermination activity of the protein, and it was proposed that the GlcT-H104D protein was not longer inactivated by EII Glc (16). To study the role of the conserved histidine residues in PRD-I more rigorously, we constructed strains encoding GlcT variants with single amino acid substitutions. The antitermination activity of the mutant GlcT proteins was assayed by analyzing the expression of the ptsG-lacZ fusion present in these strains after growth in CSE minimal medium in the absence or presence of glucose (Table II). As observed previously, expression of the fusion was constitutive in B. subtilis GP103 (GlcT-H104D). Similarly, constitutive expression was recorded for the H104A protein, suggesting that any mutation that replaces His-104 results in constitutive GlcT activity. The in vitro phosphorylation assays with EIIBA Glc indicated that mutations of His-104 or His-163 prevented phosphorylation of PRD-I (Fig. 4). It was, therefore, very interesting to study the effect of a H163D exchange on GlcT activity in vivo. This protein exhibited constitutive antitermination activity in the absence or presence of glucose (Table II). This in vivo finding is in perfect agreement with the in vitro phosphorylation results and supports the idea that the presence of both conserved histidyl residues in PRD-I is required for phosphorylation and, thus, inactivation of GlcT by EII Glc .
EIIBA Glc Phosphorylates GlcT Directly-Previous genetic and biochemical studies with GlcT and other antiterminators of the BglG family could not identify the source of the phosphate that inactivates the antiterminator. It was proposed that GlcT is phosphorylated either in a ternary complex of HPr, phosphorylated EII Glc , and GlcT or that a direct phosphotransfer from EIIBA Glc may occur (16). To distinguish between these two possibilities, we intended to use a preparation of phosphorylated EIIBA Glc that was devoid of any HPr for in vitro phosphorylation experiments. For this purpose we constructed, expressed, and purified an HPr variant with a N-terminal Strep tag. This protein was readily phosphorylated by EI and was able to phosphorylate both EIIBA Glc and the PRD-II of GlcT (see Fig. 5, for GlcT: data not shown). After incubation of (His 6 )EIIBA Glc with [ 32 P]PEP, (His 6 )EI, and (Strep)HPr, the proteins were passed consecutively over a Streptactin and a Ni 2ϩ column. After elution from the Ni 2ϩ column, phosphorylated EIIBA Glc was free of any HPr (Fig. 5). This was also confirmed by Western blot analysis using polyclonal antiserum raised against HPr from B. subtilis (data not shown).
Purified phospho-EIIBA Glc was used for an in vitro phosphorylation assay with GlcT. As shown in Fig. 6, efficient phosphorylation of GlcT with EIIBA Glc was observed. Thus, EIIBA Glc alone was sufficient to phosphorylate GlcT. If non-phosphorylated HPr was added to the phosphorylation mix, all three proteins were detected in their phosphorylated forms. To confirm EIIBA Glc -dependent phosphorylation of PRD-I, the phosphorylation reaction was performed with mutant GlcT proteins. GlcT-H104A was not a substrate of phosphorylated EIIAB Glc (Fig. 6). Only if HPr was added, a phosphotransfer to the mutant GlcT protein was observed. Two conclusions can be drawn from this result. (i) EIIBA Glc can phosphorylate HPr. Reversibility of phosphorylation reactions is known to be a general feature of the PTS (1). (ii) HPr can phosphorylate the mutant GlcT-H104A as demonstrated before (see Fig. 3). The mutant protein GlcT-H210D was readily phosphorylated by EIIBA Glc , and no additional effect was observed upon addition of HPr (Fig. 6). Thus, the phosphorylation sites in PRD-II are not required for EIIBA Glc -dependent phosphorylation of GlcT. These findings are in good agreement with the genetic evidence that PRD-I is the target of EIIBA Glc -dependent negative control of GlcT. DISCUSSION The activity of GlcT is controlled by phosphorylation events; GlcT can be phosphorylated by HPr(HisϳP) at its PRD-II and by EII Glc on PRD-I. These phosphorylations serve to control the regulatory output of the protein, RNA binding, and transcription antitermination. If the RNA binding domain is expressed a ptsG expression is indicated in units/mg of protein. Table II shows representative values of lacZ expression. All measurements were performed at least twice. FIG. 4. Phosphorylation of GlcT variants in the absence or presence of Enzyme IIBA Glc . Shown is an autoradiograph of a SDSpolyacrylamide gel showing phosphorylation of GlcT . The phosphorylation assays were performed as described in Fig. 1. 100 pmol of each GlcT variant were used as shown above the corresponding lanes. In addition to the GlcT proteins, 100 pmol of EIIBA Glc were present in the assays as indicated. The positions of GlcT and EIIBA Glc are marked with arrows. wt, wild type. without the PRDs, it is able to bind RNA and to exert antitermination activity constitutively (5,16). As shown in this work, the two phosphorylation events have different consequences; HPr-dependent phosphorylation has a slight stimulating effect on GlcT activity, whereas EII Glc -dependent phosphorylation inactivates the protein. Opposing effects of EII-and HPr-dependent phosphorylations have also been observed for other PRD-containing transcriptional regulators; the antiterminators LicT and BglG from B. subtilis and E. coli, respectively, are inactivated by their cognate EII Bgl , and they need HPr-dependent phosphorylation to be active (8,9,10,32,33). Similarly, the transcriptional activator LevR from B. subtilis is positively and negatively controlled by HPr-and EII Lev -dependent phosphorylation events, respectively. However, although EII Lev phosphorylates a PRD of LevR, HPr transfers its phosphate to a EIIA-like domain in LevR (30,35,36). In the MtlR and LicR proteins from Bacillus stearothermophilus and B. subtilis, respectively, the EIIA-like domains are targets of negative control by the cognate EII. The PRDs of these proteins are phosphorylated by HPr and are implicated in positive control of DNA binding activity (37,38).
In each PRD-containing regulator studied so far, the cognate EII negatively affects the activity of the regulator in the absence of the inducer (7,15). In contrast, the role of HPr varies among the different proteins; there is an absolute requirement for HPr-dependent phosphorylation of LicT, BglG, or the B. subtilis antiterminator SacT, whereas the SacY antiterminator from B. subtilis is fully active in the absence of a functional HPr (8, 39 -41). Similarly, GlcT is active in the absence of HPr. However, full activity is observed only in the presence of both HPr and the histidyl residues in PRD-II that are the targets of HPr-dependent phosphorylation (Ref. 4; see Table II and Fig. 3). We may, thus, assume that evolution of the PRD-II gave rise to one type that is positively controlled by HPr, whereas the second type is independent of this phosphorylation. The PRD-II of GlcT may be an intermediate between these two extremes. The isolation of HPr-independent variants of LicT demonstrates that single amino acid exchanges in PRD-II are sufficient to convert a HPr-dependent to a HPr-independent form without affecting substrate induction (32).
HPr phosphorylates several PRD-containing regulators but also enzymes of Gram-positive bacteria such as glycerol kinase in B. subtilis and the lactose transporter LacS in Streptococcus thermophilus (7,29,42). This phosphorylation occurs in the absence of glucose if HPr itself is present as HPr(HisϳP) and is part of the general phenomenon of carbon catabolite repression (7,8,32,35,43,44). It is, thus, not surprising that GlcT, which needs to be active in the presence of glucose, does not depend on HPr-dependent phosphorylation. The residual slight stimulation by HPr may reflect the evolutionary adaptation of an originally HPr-dependent antiterminator to a HPr-independent protein.
The role of the sugar-specific EII in negative control of antiterminators of the BglG family has been discovered long before the concept of the PRDs had been proposed (7,33,34,41,45). However, in the experiments aimed at the investigation of EII-dependent phosphorylation of the antiterminators, there were always EI and HPr present, or their absence was not demonstrated. The genetic evidence suggested a direct role of the EII in the negatively acting phosphorylation, but it was not possible to rule out HPr-dependent phosphorylation that is stimulated in the presence of phosphorylated EII. It was, therefore, crucial to set up experiments that allow us to address this problem more directly. As shown in this study, the use of different affinity tags for the PTS proteins enabled us to separate phosphorylated EIIBA Glc from HPr and to demonstrate that EIIBA Glc alone is sufficient to phosphorylate GlcT and that this phosphorylation specifically affects PRD-I (see Fig. 6). Moreover, we provided evidence that the intracellular accumu- showing samples (20 l) of a phosphorylation assay at different stages of separation. 10 nmol of (His 6 )EIIBA were phosphorylated with [ 32 P]PEP. The assay contained 8 nmol of (His 6 )EI and 10 nmol of (Strep)HPr in a total volume of 2 ml of phosphorylation buffer and was incubated at 30°C for 15 min. After phosphorylation, the assay mixture (P) was passed over a 3-ml Streptactin and a 3-ml Ni 2ϩ -nitrilotriacetic acid column (FT). After washing (W) with 10 ml of phosphorylation buffer, the (His 6 )EII bound to the column was eluted by using phosphorylation buffer containing 100 mM of imidazole (elution fraction). 2 ml of elution buffer were used, with each elution fraction 200 l. Protein concentration was determined by using the Bio-Rad protein assay. The positions of the PTS proteins are indicated by arrows. S refers to the protein standard. lation of phosphorylated EIIBA Glc leads to GlcT inactivation (see Table III). Similarly, the PRD-containing transcriptional activator LevR is directly phosphorylated by its cognate EIIB Lev , LevE (30). However, although the PRDs of the antiterminators of the BglG family and of LevR are similar to each other, the EII are members of different families and do not exhibit similarity at the sequence level (36).
The two PRDs of GlcT are similar to each other and share 34% identical amino acids. For the PRDs of LicT, it was shown that they adopt very similar structures (11). Yet the PRDs are specialized in the choice of their phosphorylation partners. This indicates that some specificity determinants in the PRDs determine whether they are target for HPr-or EII-dependent phosphorylation. Interestingly, HPr and the EIIB, which are thought to phosphorylate the PRDs (16,30,46), share some structural similarity that may reflect their common property to reversibly transfer phosphate to/from EIIA (47-50). An evolutionary analysis suggests that PRD-I and PRD-II diverged in a primordial regulator before the emergence of the individual antiterminators (36). It will be most interesting to identify the specificity determinants in the PRDs.
Another interesting problem is the identification of the actual phosphorylation site(s) in the PRDs. Genetic and biochemical evidence presented here for GlcT and the data for other PRD regulators indicate that the conserved histidine residues are the targets of phosphorylation (Refs. 9, 13, 27, 30, 37, 38, and 51; see Figs. [2][3][4]. Mutations of each of the two conserved histidine residues resulted in a complete loss of phosphorylation of that PRD (see Figs. [3][4][5][6]. Moreover, the mutation of either histidine of PRD-I resulted in constitutive GlcT activity. Identical effects of mutations affecting one of the two histidine residues in one PRD were also observed for LicT and LacT from Lactobacillus casei (10,51). One could imagine several scenarios for phosphorylation. (i) The two histidine residues in each PRD may be both phosphorylated individually. In agreement with this suggestion is the observation of multiple phosphoryl- GlcT is present in different forms in the cell due to presence or absence of glucose. In the presence of glucose the phosphoryl group is transferred from HPr(HisϳP) to the incoming sugar via EII Glc . Under these conditions GlcT is not phosphorylated and binds to the RAT, thus allowing transcription of the ptsGHI operon. In the absence of glucose, phosphorylated EII Glc accumulates in the cell. In this case, the phosphate is transferred to the PRD-I of GlcT (see Fig. 6), leading to inactivation of GlcT (see Table II). Inactive GlcT is thought to be monomeric and may be sequestered by the EII Glc as proposed for the homologous BglG protein from E. coli (54). CM, cytoplasmatic membrane.
ations in the PRDs of LicT and SacY (9,13). However, our data suggest that these phosphorylation events would not occur independently from each other since mutation of one site prevents phosphorylation on both sites (see Figs. 3, 4, and 6). (ii) There might be an intra-PRD phosphate transfer. In that case one would expect complete loss of phosphorylation for the mutant affecting the primary phosphorylation target and a significant residual phosphorylation for the mutant in which the second histidine was replaced. However, such a result was not observed, arguing against an intra-PRD phosphotransfer. (iii) Finally, one of the histidine residues might be phosphorylated, whereas the other would be required for the phosphorylation reaction. Such an involvement of two histidine residues was demonstrated for EIIA Crr from E. coli; His-90 is the site of phosphorylation by HPr, but the nearby His-75 is important for phosphorylation and phosphotransfer to EIIB. These two histidine residues are very close to each other (52,53). Similarly, the conserved histidines of each PRD of LicT are located in close proximity, and phosphorylation of one site may, thus, depend on the presence of the second active site histidine. However, phosphorylation of two sites in one PRD was described for LicT (9). In that study the strongest phosphorylation by HPr was detected on His-207 in PRD-II, whereas phosphorylation on His-269 was close to the background. These findings are, thus, compatible with the third model, and so are our results presented in this work. However, more work is required to unequivocally solve this problem.
Based on the findings presented here we may propose that in the presence of glucose GlcT is present in its non-phosphorylated form, which is capable of forming dimers and causing transcription antitermination upon binding to the ptsG RAT sequence (see Fig. 7A). In the absence of glucose GlcT becomes phosphorylated on its PRD-I by EII Glc . This phosphorylation may prevent dimerization as proposed for BglG and LicT (14,15). This monomeric form of GlcT would be inactive in RNA binding and antitermination (Fig. 7B).