Several operon-specific transcriptional regulators, including antiterminators and activators, contain a duplicated conserved domain, the PTS regulation domain (PRD). These duplicated domains modify the activity of the transcriptional regulators both positively and negatively. PRD-containing regulators are very common in Gram-positive bacteria. In contrast, antiterminators controlling β-glucoside utilization are the only functionally characterized members of this family from Gram-negative bacteria. PRD-containing regulators are controlled by PTS-dependent phosphorylation with different consequences: (i) In the absence of inducer, the phosphorylated EIIB component of the sugar permease donates its phosphate to a PRD, thereby inactivating the regulator. In the presence of the substrate, the regulator is dephosphorylated, and the phosphate is transferred to the sugar, resulting in induction of the operon. (ii) In Gram-positive bacteria, a novel mechanism of carbon catabolite repression mediated by PRD-containing regulators has been demonstrated. In the absence of PTS substrates, the HPr protein is phosphorylated by enzyme I at His-15. This form of HPr can, in turn, phosphorylate PRD-containing regulators and stimulate their activity. In the presence of rapidly metabolizable carbon sources, ATP-dependent phosphorylation of HPr at Ser-46 by HPr kinase inhibits phosphorylation by enzyme I, and PRD-containing regulators cannot, therefore, be stimulated and are inactive. All regulators of this family contain two copies of PRD, which are functionally specialized in either induction or catabolite repression.
Bacteria are capable of using many different carbohydrates as their only source of carbon and energy. The genes and operons encoding the enzymes involved in the transport and catabolism of carbohydrates are, in most cases, only expressed if (i) the corresponding carbohydrate is present in the medium and (ii) the preferred carbon sources are absent. The underlying regulatory mechanisms are referred to as induction and carbon catabolite repression respectively.
Catabolic operons can be induced by increasing the rate of transcription initiation (via transcriptional repressors or activators) or of transcript elongation (dependent on transcriptional antiterminators). In each case, the regulator proteins respond to the availability of the inducer in the medium.
Carbon catabolite repression (CCR) has been studied intensively in both Gram-negative and Gram-positive bacteria. In all bacteria studied so far, the phosphoenolpyruvate:sugar phosphotransferase system (PTS), which transports and phosphorylates a variety of sugars, is involved in CCR. The PTS consists of two general soluble proteins, enzyme I (EI) and HPr, and of membrane-bound sugar specific permeases (enzymes II). The enzymes II (EII) are made up of three or four components, which may exist as independent proteins or as fused domains in a single protein. The domain EIIC is an integral membrane component, whereas EIIA and EIIB are localized in the cytoplasm or at the inner surface of the membrane. The phosphate moiety is transferred from PEP to the sugar via EI, HPr, EIIA and EIIB (Postma et al., 1993). In enteric bacteria, the soluble component of the glucose permease (Crr or EIIAGlc) is the key regulator of CCR. In the absence of glucose, the phosphorylated form of EIIAGlc stimulates the activity of adenylate cyclase. cAMP forms a complex with its receptor protein, allowing the transcriptional activation of catabolic operons. In the presence of glucose, unphosphorylated EIIAGlc binds to several sugar permeases and to glycerol kinase, thereby inhibiting them. This latter process is called inducer exclusion (for a review, see Saier, 1989). Recent studies on CCR of the E. coli lac operon suggest that (i) cAMP levels are not correlated to the presence or absence of glucose and (ii) cAMP-dependent regulation of the lac operon results from inducer exclusion mediated by the PTS (Inada et al., 1996).
In Gram-positive bacteria, the HPr protein of the PTS contains a regulatory phosphorylation site (Ser-46) in addition to the residue (His-15) phosphorylated by EI (for a review, see Hueck and Hillen, 1995). In the presence of fructose-1.6-bisphosphate, an HPr kinase phosphorylates HPr at the expense of ATP (Reizer et al., 1998). HPr(Ser-P) can form a complex with a repressor protein, CcpA (Deutscher et al., 1995). The binding of this complex to catabolite-responsive elements (cre) in front of catabolic genes prevents transcription of the target operons in the presence of glucose and other glycolytically metabolizable carbon sources (Fujita et al., 1995). Recent data indicate that CcpA might alternatively interact with Crh(Ser-P) or glucose 6-phosphate (Galinier et al., 1997; Miwa et al., 1997). Crh is a protein similar to HPr that is phosphorylated at Ser-46 by HPr kinase. In addition, several operon-specific regulators have recently been shown to be involved in CCR. The LevR activator and the LicT and SacT antiterminators are positively regulated by phosphorylation by HPr(His-P) in the absence of glucose (Stülke et al., 1995; Arnaud et al., 1996; Deutscher et al., 1997).
In this review, we discuss recent data concerning a class of positive regulators that is a target for specific induction and/or CCR. These regulations are mediated by PTS-dependent phosphorylation of the regulators, which modulates their activity. All of these positive regulators include a conserved domain, called PRD (PTS regulation domain), which is the target for both positive and negative control by PTS components.
Proteins containing a PRD and their structure
The known and putative proteins containing a PRD identified in bacteria up to January 1998 are listed in Table 1. Such proteins are found in both Gram-negative and Gram-positive bacteria. Most of them are positive transcriptional regulators, either activators or antiterminators, controlling the synthesis of enzymes that are either involved in the catabolism of sugars transported by the PTS or generate PTS substrates.
Antiterminators containing a PRD seem to be very common elements in systems controlling β-glucoside catabolism in both Gram-negative and Gram-positive bacteria. Five antiterminators regulating bgl operons have been sequenced, and two incomplete sequences are available (Rutberg, 1997). Only one tentative antiterminator not involved in β-glucoside utilization has been identified in Gram-negative bacteria. This antiterminator is found in a pathogenicity island of a pyelonephritogenic Escherichia coli strain in front of a gene encoding an enzyme II of unknown function (Kao et al., 1997). Sucrose catabolism in bacilli is also controlled by transcriptional antiterminators (Rutberg, 1997). Interestingly, Bacillus subtilis contains two sucrose utilization systems that are controlled by two different antiterminators, SacT and SacY, which exhibit cross-talk with their respective RNA targets (Steinmetz et al., 1989; Arnaud et al., 1992; Aymerich and Steinmetz, 1992). Recently, the involvement of an antiterminator containing a PRD has also been demonstrated in the regulation of the glucose-specific PTS component in B. subtilis (Stülke et al., 1997), and a very similar gene is present in Staphylococcus carnosus upstream from a gene encoding a glucose permease (Christiansen and Hengstenberg, 1996; GenBank accession no. Y14029). Moreover, lactose utilization in Lactobacillus casei is regulated by an antiterminator belonging to this family (Alpert and Siebers, 1997; Gosalbes et al., 1997).
The transcription of genes and operons controlled by antiterminators is initiated constitutively, but elongation is blocked by transcriptional terminators. The 5′ parts of these terminators overlap conserved inverted repeats, called ribonucleic antiterminators (RAT; Aymerich and Steinmetz, 1992). The binding of BglG, SacT and SacY to their respective RAT sequences has been demonstrated in vitro (Houman et al., 1990; Arnaud et al., 1996). Upon binding, the antiterminator is assumed to stabilize the RAT secondary structure, preventing the formation of a structure acting as the terminator and thus allowing transcription of the target gene. Transcriptional antitermination has been demonstrated in vitro for BglG, SacT and SacY (Houman et al., 1990; Arnaud et al., 1996) and in vivo for LicT (Krüger et al., 1996; for review, see Rutberg, 1997). The antiterminators of this family contain an N-terminal RNA-binding domain of about 55 amino acids, as demonstrated for BglG of E. coli and SacY, SacT, LicT and GlcT of B. subtilis. The isolated N-terminal domains have constitutive antitermination activity (Manival et al., 1997; S. Bachem and J. Stülke, unpublished results). The NMR solution and crystal structures of the RNA binding domain reveal a new RNA-binding motif (Manival et al., 1997; van Tilbeurgh et al., 1997). The activity of these antiterminators is modulated by the PTS via two PRDs (Fig. 1; see below).
Transcriptional activators containing a PRD have so far only been found in Gram-positive bacteria. The B. subtilis LevR protein controls the expression of the levanase operon involved in the degradation of fructose polymers and fructose uptake (Martin-Verstraete et al., 1990). The promoter of this operon is recognized by RNA polymerase associated with the sigma54-like factor σL (Débarbouilléet al., 1991). LevR contains an N-terminal helix–turn–helix motif, a domain homologous to the central domain of NifA-like regulators necessary for the interaction with the RNA polymerase associated with σ54, and two PRDs (Fig. 1; Débarbouilléet al., 1991; Stülke et al., 1995). A truncated LevR polypeptide containing the helix–turn–helix motif and the NifA-like domain is sufficient for binding to the DNA target and the constitutive activation of transcription (Martin-Verstraete et al., 1994). As in the antiterminators, the PRDs are necessary for the modulation of LevR activity in response to fructose and glucose availability. The LicR protein positively controls the B. subtilis licBCAH operon required for the transport and degradation of oligomeric β-glucosides, which are the products of the action of the extracellular enzyme β-glucanase on substrates such as lichenan (Schnetz et al., 1996; Tobisch et al., 1997). The lic operon is transcribed from a σA-dependent promoter. The LicR regulator contains two potential helix–turn–helix motifs, two PRDs and a C-terminal domain similar to mannitol-specific EIIA of the PTS (Tobisch et al., 1997; Fig. 1). A protein with a similar domain structure was identified in the course of the B. subtilis genome sequencing project. The corresponding gene, yjdC, is situated upstream of a gene presumably encoding an EII of the mannitol/fructose family and a phosphomannoisomerase (Kunst et al., 1997). Therefore, this regulator, designated ManR, is probably involved in the control of genes necessary for mannose utilization (see Table 1). In addition, PRD-containing regulators probably involved in the control of mannitol utilization are also found in B. subtilis, Bacillus stearothermophilus and Clostridium acetobutylicum (Kunst et al., 1997; Henstra et al., 1996; S. Behrens and H. Bahl, personal communication).
Thus, based on the complete nucleotide sequences of the genomes, B. subtilis encodes eight PRD-containing regulators including four antiterminators, whereas wild-type E. coli contains only the BglG antiterminator as a member of this family of regulators (Kunst et al., 1997).
The E. coli CsiE protein contains a single PRD, the function of which is, however, unknown. The csiE gene is under the control of the stationary phase σ-factor σS and of the cAMP/CRP complex, suggesting a role in survival under adverse conditions (Marschall and Hengge-Aronis, 1995).
An alignment of the 39 known PTS regulation domains is shown in Fig. 2. With the exception of CsiE, all regulators contain a duplication of the PRD. The PRD located closer to the N-terminus will hereafter be referred to as PRD-I, and to the C-terminus as PRD-II. All PRD sequences contain a histidine at position 7 of the alignment. This histidine is a target for either positive or negative regulation by PTS-dependent phosphorylation, as shown for the antiterminators SacY, LicT and BglG and the LevR activator (see below). Downstream from this regulatory histidine there is a well-conserved arginine (position 14 of the alignment in Fig. 2). It is interesting to note that the region around the regulatory histidine is best conserved in the PRD-I in the transcriptional antiterminators. This might reflect a conserved interaction with a regulatory partner or the RNA-binding domain. Moreover, a glutamate (position 63) is strongly conserved. This glutamate is absent from PRD-II of the BglR antiterminator of Lactococcus lactis. However, the sequence of the C-terminus of BglR may be erroneous, as proposed by Tortosa et al. (1997). In addition to the histidine at position 7, another histidine is conserved, although not in all PRDs, at position 70. Thus, each PRD regulator contains two to four conserved histidines. The involvement of these residues in the activity of the PRD-containing proteins is discussed in the following sections.
Negative regulation at the PRD and substrate-specific induction
Substrate-specific induction of any catabolic gene depends on the ability of the specific regulator to monitor the presence of the substrate. The sugar-specific PTS permeases, whose genes are controlled by PRD-containing regulators, transfer the information on substrate availability to the regulators. The permeases consist of three or four domains or proteins, two of which (EIIA and EIIB) are involved in phosphate transfer from HPr(His-P) to the sugars, whereas EIIC (and EIID, if present) is membrane bound and implicated in sugar transport (Postma et al., 1993). Mutations in the genes encoding the constituents of the sugar permeases have two distinct phenotypes with respect to the expression of their own genes: (i) Expression can be lost. Examples include a deletion of domain EIIC of SacP and mutations affecting the membrane-bound component LevG or EIIC of PtsG of B. subtilis (Steinmetz et al., 1989; Martin-Verstraete et al., 1990; Stülke et al., 1997). (ii) Expression can become constitutive, as is the case for mutations affecting EIIA or EIIB, involved in phosphate transfer, of LevD, LevE and PtsG of B. subtilis and BglF of E. coli (Martin-Verstraete et al., 1990; Chen et al., 1997a; Stülke et al., 1997). Similarly, inactivation of entire sacX or bglP permease genes in B. subtilis results in constitutive expression (Crutz et al., 1990; Le Coq et al., 1995).
The EIIAB components of the PTS permeases act as negative regulators. However, constitutive expression in permease mutants depends on the presence of the positively acting PRD-containing regulators, as demonstrated for SacX/SacY, BglP/LicT, PtsG/GlcT and LevDE/LevR (Crutz et al., 1990; Le Coq et al., 1995; Martin-Verstraete et al., 1995; Stülke et al., 1997). Taken together, these data suggest a link between the phosphorylation status of the permeases and the activity of the corresponding PRD-containing regulator: (i) under conditions in which the permeases cannot phosphorylate their sugar substrates (i.e. in the absence of the inducer or in mutants affecting EIIC), regulators containing a PRD are not active; (ii) if the permeases are not phosphorylated (i.e. in the presence of inducer or in EIIA or EIIB mutants), the regulators are active and the respective catabolic operons are expressed. This modulation of regulator activity may be mediated by PTS-dependent phosphorylation. The phosphate group is transferred from HPr to the sugar in the following sequence: HPr → EIIA → EIIB → sugar (Postma et al., 1993). The fact that EIIB is the last component in the phosphorylation cascade required for both sugar transport and modulation of PRD-containing regulator activity strongly suggests that phosphorylated EIIBs indeed negatively control the PRD-containing regulators.
The mode of negative control of a PRD-containing regulator by an EII has only been studied in any detail for the BglG antiterminator from E. coli and the B. subtilis LevR activator. The E. coliβ-glucoside permease BglF transports and phosphorylates salicin, arbutin and β-methylglucoside. As discussed above, mutations in EIIA and EIIB lead to constitutive expression of the bgl operon, whereas mutations in EIIC result in loss of expression (Schnetz and Rak, 1990; B. Rak, personal communication; and see above). As phosphorylated EII is necessary for the negative regulation of BglG activity, direct phosphorylation of BglG by BglF has been assayed. BglF effectively phosphorylates the antiterminator only in the absence of β-glucosides. Moreover, phosphorylated BglG is efficiently dephosphorylated by BglF if salicin is available as substrate (Amster-Choder et al., 1989; Schnetz and Rak, 1990). Interestingly, EIIAGlc can substitute for EIIABgl for both sugar and BglG phosphorylation, again suggesting that EIIBBgl interacts specifically with and phosphorylates BglG. Recently, the catalytic residue of EIIBBgl, Cys-24, has been shown to be the source of phosphate for both sugar and BglG phosphorylation (Chen et al., 1997a). Three constitutive variants of BglG have also been tested for phosphorylation by BglF. All of these mutations (affecting His-160, His-208, and Asp-100 adjacent to the conserved His-101) severely inhibit phosphorylation by BglF (Amster-Choder et al., 1989; Chen et al., 1997b). The finding that a histidine residue in BglG is the target of phosphorylation (Amster-Choder and Wright, 1997) suggests that all three conserved histidines are necessary for the efficient phosphorylation of BglG by EIIBgl.
The induction of the B. subtilis levanase operon by fructose depends on the activator protein LevR and on the fructose-specific PTS encoded by the operon. Genetic studies indicate that phosphorylated EIIBLev (LevE) is a negative regulator of LevR activity (Charrier et al., 1997) and that PRD-II in LevR, also called domain C, is the target for negative regulation (Stülke et al., 1995; Fig. 1). Using GST–domain C fusion polypeptides, it has been shown recently that domain C is phosphorylated by LevE in the presence of PEP, EI, HPr and LevD. No phosphorylation of domain C was observed in the absence of LevE. In addition, LevE-P separated from the proteins required for its phosphorylation is sufficient to phosphorylate LevR (Martin-Verstraete et al., 1998).
To identify the site of negative regulation by the PTS, mutants conferring constitutive activity to the regulators have been isolated. For the antiterminators SacY, SacT and GlcT, most of these mutations change a conserved histidine residue in PRD-I (His-99 in SacY) or residues close to this position (Fig. 1; Crutz et al., 1990; Débarbouilléet al., 1990; M. Arnaud, unpublished results; S. Bachem and J. Stülke, unpublished results). It seems, therefore, that PRD-I is involved in negative control by the specific sugar permeases. In addition, a deletion of the EIIC domain of SacP, which normally prevents expression of the sacPA operon, has no effect in the presence of the constitutive antiterminator SacT30 (Arnaud et al., 1992). In contrast, phosphorylation of the LevR activator by LevE, and thereby negative regulation of its activity, occurs at PRD-II (Stülke et al., 1995; Charrier et al., 1997). Modification of the conserved histidine residue in domain C (His-869) of LevR by site-directed mutagenesis led to constitutive expression of the levanase operon. In addition, the LevR-H869A and GST-domain C H869A polypeptides are not phosphorylated by LevE-P. His-869 is, therefore, the site of negative regulation of LevR activity (Martin-Verstraete et al., 1998).
The structural consequences of BglG phosphorylation by BglF have been studied. Active unphosphorylated BglG antiterminator is found as a dimer. Upon phosphorylation by EIIBgl, BglG is unable to dimerize, and activity is lost. This was demonstrated by analysing the ability of monomeric and dimeric BglG to bind its RNA target, RAT: only dimers are capable of interacting with the RNA. In addition, a fusion of the DNA-binding segment of the λ repressor to BglG is able to dimerize and is active as a transcriptional repressor (Amster-Choder and Wright, 1992). Thus, substrate availability regulates the antiterminator's activity via modulation of PRD phosphorylation and, probably, of its oligomerization state.
Positive regulation at the PRD and its implication in carbon catabolite repression
In view of the observations discussed above, any mutation preventing phosphate transfer from the PTS to the PRD-containing regulators would be expected to result in constitutive expression of the operons controlled. This is indeed the case for sacB and ptsG regulated by SacY and GlcT, respectively, and partially for the levanase operon activated by LevR, all from B. subtilis (Crutz et al., 1990; Martin-Verstraete et al., 1990; Arnaud et al., 1992; Stülke et al., 1995; 1997). The finding that the SacT antiterminator depends on the presence of the general proteins of the PTS (enzyme I and HPr) to be active (Arnaud et al., 1992) does not fit with this model and suggests a more sophisticated mode of control of PRD-containing regulators. Subsequently, the antiterminator LicT and the activators LevR and LicR of B. subtilis have also been shown to be completely or partially dependent on the general components of the PTS (Le Coq et al., 1995; Stülke et al., 1995; Tobisch et al., 1997).
Genetic studies revealed that HPr is the component of the PTS required for full activity of SacT, LicT and LevR. A replacement of the PEP-dependent phosphorylation site in HPr results in a loss of stimulation of LevR and LicT (Stülke et al., 1995; Krüger et al., 1996). Thus, PEP-dependent phosphorylation of HPr by EI is necessary for full activity of the regulators positively controlled by the PTS. Phosphorylation of LevR, SacT and LicT in a PEP-, EI- and HPr-dependent manner has now been demonstrated (Stülke et al., 1995; Arnaud et al., 1996; Deutscher et al., 1997). The LacT antiterminator from L. casei and the LicR activator from B. subtilis are also subject to positive control by the general components of the PTS (Gosalbes et al., 1997; Tobisch et al., 1997).
HPr-dependent phosphorylation of PRD-containing regulators has been suggested as being involved in CCR (Stülke et al., 1995). The sacPA, bglPH, licBCAH and levanase operons of B. subtilis are all subject to CCR. Repression of the bgl and lev operons is mediated by the repressor protein CcpA and its DNA target, cre (Martin-Verstraete et al., 1995; Krüger et al., 1996). In both cases, however, ccpA-independent CCR was also observed. Expression of the bglPH operon is only partially relieved from glucose repression in a ccpA mutant. Residual repression is lost if the terminator in front of the operon, the target of LicT, is deleted. Northern blot analysis revealed that transcription initiation of the bglPH operon is fully derepressed in a ccpA mutant; however, there was no elongation beyond the terminator in the presence of glucose (Krüger et al., 1996). It was concluded that the operon-specific regulators, LevR and LicT, are involved in CCR, and ccpA-independent CCR of these operons was suggested as being linked to the dependence of the regulators on the general components of the PTS. CcpA-dependent and CcpA-independent mechanisms might also be involved in CCR of the L. casei lactose operon. The RNA target of LacT was implicated in the CcpA-independent mechanism, suggesting regulation of LacT activity by HPr as discussed above (Gosalbes et al., 1997).
In the presence of repressing sugars, the HPr protein in Gram-positive bacteria is phosphorylated by the HPr kinase at Ser-46 (Reizer et al., 1998). This ATP-dependent phosphorylation of HPr inhibits the phosphorylation by EI at His-15 about 600-fold (Saier, 1989). In addition, HPr(His-P) is dephosphorylated when the sugars are taken up by their PTS permease. Therefore, no phosphorylation and no stimulation of the activity of SacT, LicT and LevR are possible if repressing sugars are present in the growth medium. Indeed, when HPr(Ser-P) was used in an in vitro phosphorylation assay of LicT, no phosphorylation was observed (Deutscher et al., 1997), confirming the proposed link between CcpA-independent CCR and HPr(His-P)-dependent stimulation of transcriptional regulators containing a PRD.
The targets of HPr-dependent phosphorylation in the antiterminator LicT and the transcriptional activator LevR have been studied. In both cases, His residues are phosphorylated by HPr. The sites of LicT phosphorylation by HPr have been determined. Four strongly conserved histidine residues are phosphorylated by HPr in vitro, two in each copy of the PRD with His-207 being the most strongly phosphorylated residue (Deutscher et al., 1997). Site-directed mutagenesis studies with the corresponding histidine residue of SacT indicate that His-207 is necessary for SacT activity (M. Arnaud, unpublished results).
Truncated LevR polypeptides lacking PRD-II are completely dependent on the presence of HPr for activity. Coincidentally, a deletion of the ccpA gene has no effect on CCR of the levanase operon in the presence of these truncated LevR polypeptides, indicating the possibility of a regulatory link between PRD phosphorylation by HPr and CCR. These results suggest that PRD-I (contained in domain B of LevR and still present in the truncated LevR polypeptides) is essential for positive regulation by HPr and, thus, that it is the target for HPr-dependent phosphorylation (Martin-Verstraete et al., 1995; Stülke et al., 1995). In vitro phosphorylation experiments with domain B of LevR demonstrated that this domain was indeed phosphorylated by HPr, while PRD-II (domain C) was not (see above). Studies with mutant LevR proteins indicate that His-585 of LevR is the target of HPr-dependent phosphorylation (Martin-Verstraete et al., 1998). Mutants defective in EI, HPr or producing truncated LevR lacking PRD-I or LevR carrying a modification at His-585 exhibited reduced LevR activity. Surprisingly, this histidine does not correspond to His-506 that is conserved in all PRDs. His-585 is, however, present in the CelR protein, a putative regulatory protein of cellobiose catabolism in B. stearothermophilus (Table 1; Lai and Ingram, 1993).
The B. subtilis sacB gene encoding levansucrase is controlled by the SacY antiterminator. The expression of sacB is induced by high concentrations of sucrose and is not subject to CCR (Steinmetz et al., 1989). A mechanism for CCR via the antiterminator SacY is therefore not required. SacY is active even in the absence of a functional PTS (Crutz et al., 1990). Interestingly, the SacY antiterminator is also phosphorylated by HPr on three different histidine residues (His-99 in PRD-I and His-207 and His-269 in PRD-II). Genetic analysis revealed that only His-99 is directly involved in regulation of the negative control of activity of SacY (Tortosa et al., 1997). It seems, therefore, that direct phosphorylation by HPr can also play a negative regulatory role. The formation of a SacX/SacY complex modulating the phosphorylation kinetics of SacY by HPr at His-99 was proposed by Tortosa et al. (1997) to explain both the negative role of SacX (see above) and the specific modulation of this phosphorylation in response to sucrose availability. Alternatively, in view of the results obtained with LicT and LevR, direct phosphorylation of SacY by HPr may only be a minor event, and the regulatory phosphorylation of SacY may be performed by SacX. Direct phosphorylation of SacY by HPr might possibly be an evolutionary remnant that has lost its functional significance. The activity of GlcT, the antiterminator of the B. subtilis ptsGHI operon, like the activity of SacY, is independent of the presence of the general proteins of the PTS (Stülke et al., 1997).
The finding of HPr-dependent phosphorylation of PRD-containing transcriptional regulators raises the question of how this phosphorylation might influence the activity of the proteins. It appears that phosphorylation by HPr enables these proteins to dimerize and that dimerization is necessary for the activity of the E. coli BglG antiterminator (Amster-Choder and Wright, 1992). Overproduction can overcome or reduce the dependence of LicT and LevR on phosphorylation by HPr (Stülke et al., 1995; Krüger et al., 1996). Although the activity of the SacT antiterminator is completely dependent on stimulation by HPr in vivo, no phosphorylation is required for its antitermination activity in vitro. This might result from a relatively high concentration of SacT in the assay mixture, which might circumvent the need for phosphorylation (Arnaud et al., 1996).
So far, positive control of PRD-containing regulators has been described only for regulators from Gram-positive bacteria. However, B. subtilis LevR can be positively regulated by the E. coli PTS, demonstrating that this organism has the capacity to control the activity of PRD-containing regulators positively (Stülke et al., 1995). Moreover, unpublished data suggest that BglG depends on positive control by HPr to be active, as does its B. subtilis counterpart, LicT, and HPr-dependent phosphorylation of BglG has been observed (B. Görke and B. Rak, personal communication). It will be interesting to learn whether this novel, HPr-dependent mechanism of CCR exists even in Gram-negative bacteria.
Regulation at PRDs is achieved by different types of PTS proteins: (i) phosphorylation and positive regulation by HPr; and (ii) negative regulation by sugar-specific enzymes EIIB, which belong to different families and are unrelated to each other (see Table 1). Moreover, as HPr-dependent phosphorylation stimulates the activity of the regulators in many cases, phosphorylation by EII is inhibitory (Fig. 3). The LevR protein from B. subtilis is the only example for which both mechanisms of regulation have been elucidated in detail. The domains mediating positive and negative regulation seem to be inverted in LevR compared with the antiterminator proteins. It would be most interesting to analyse the flow of phosphate from the different PTS components to the antiterminators and its functional consequences.
Undoubtedly, further analysis of regulators containing PTS regulation domains will substantially improve our knowledge about substrate induction and CCR in bacteria.
We are grateful to Steffi Bachem, Michel Débarbouillé, Josef Deutscher and Cordula Lindner for helpful discussions. Wolfgang Hillen is acknowledged for critical reading of this manuscript and for his interest in our work. We would like to thank H. Bahl, S. Behrens, B. Görke, S. Henstra, B. Rak and G. Robillard for providing information before publication. Work in the authors' laboratories is supported by the Deutsche Forschungsgemeinschaft through SFB 473, the Centre National de la Recherche Scientifique, the Université Paris 7 and the Institut Pasteur.