The Bacterial Phosphoenolpyruvate:Carbohydrate Phosphotransferase System: Regulation by Protein Phosphorylation and Phosphorylation-Dependent Protein-Protein Interactions

Abstract

SUMMARY

The bacterial phosphoenolpyruvate (PEP):carbohydrate phosphotransferase system (PTS) carries out both catalytic and regulatory functions. It catalyzes the transport and phosphorylation of a variety of sugars and sugar derivatives but also carries out numerous regulatory functions related to carbon, nitrogen, and phosphate metabolism, to chemotaxis, to potassium transport, and to the virulence of certain pathogens. For these different regulatory processes, the signal is provided by the phosphorylation state of the PTS components, which varies according to the availability of PTS substrates and the metabolic state of the cell. PEP acts as phosphoryl donor for enzyme I (EI), which, together with HPr and one of several EIIA and EIIB pairs, forms a phosphorylation cascade which allows phosphorylation of the cognate carbohydrate bound to the membrane-spanning EIIC. HPr of firmicutes and numerous proteobacteria is also phosphorylated in an ATP-dependent reaction catalyzed by the bifunctional HPr kinase/phosphorylase. PTS-mediated regulatory mechanisms are based either on direct phosphorylation of the target protein or on phosphorylation-dependent interactions. For regulation by PTS-mediated phosphorylation, the target proteins either acquired a PTS domain by fusing it to their N or C termini or integrated a specific, conserved PTS regulation domain (PRD) or, alternatively, developed their own specific sites for PTS-mediated phosphorylation. Protein-protein interactions can occur with either phosphorylated or unphosphorylated PTS components and can either stimulate or inhibit the function of the target proteins. This large variety of signal transduction mechanisms allows the PTS to regulate numerous proteins and to form a vast regulatory network responding to the phosphorylation state of various PTS components.

INTRODUCTION

PTS-Mediated Sugar Transport and Phosphorylation

Many bacteria (1, 2) as well as some archaea (3) take up sugars and sugar derivatives, such as sugar alcohols, amino sugars, glycuronic acids, disaccharides, and numerous other carbon sources, via the phosphoenolpyruvate (PEP):carbohydrate phosphotransferase system (PTS). The PTS is usually composed of one membrane-spanning protein and four soluble proteins. Enzyme I (EI) and HPr are the general cytoplasmic PTS components, which in most organisms are involved in the uptake of all PTS carbohydrates; their genes are usually organized in the ptsHI operon, with ptsH coding for HPr and ptsI for EI. In contrast, the EIIA, EIIB, and EIIC (and in mannose-type PTSs also EIID) proteins are usually specific for one substrate or, in a few cases, for a small group of closely related carbohydrates (1, 2, 4). To indicate their substrate specificity, a three letter code is used, which is added as superscript to the corresponding protein name (5). For example, EIIA^Glc stands for the glucose-specific EIIA component, EIIB^Fru for the fructose-specific EIIB, EIID^Man for the mannose-specific EIID, etc. The genes encoding the PTS components specific for a certain sugar are normally organized in an operon, which frequently also contains the genes required for the catabolism of the transported substrate.

With the exception of several actinobacteria, such as Agromyces italicus, Cellulomonas flavigena, or Actinomyces turicensis, which possess a regular HPr in addition to a DhaM protein composed of an HPr-like domain and an EIIA^Man-like domain, Gram-positive bacteria usually contain only one EI and one HPr. In contrast, Enterobacteriaceae usually produce several EI and HPr homologues or paralogues, such as nitrogen-related EI^Ntr and NPr and the fructose-specific FPr. Both types of bacteria normally contain several operons encoding the sugar-specific PTS components. According to the sequence and composition of the membrane components, seven families of PTS can be distinguished: the glucose, fructose, lactose, glucitol, galactitol, mannose, and ascorbate families (6). Interestingly, numerous alpha-, beta-, gamma-, and deltaproteobacteria also contain an incomplete PTS lacking any known EIIB and EIIC component. These PTSs are therefore probably not involved in carbohydrate transport but carry out only regulatory functions. Finally, in certain bacteria the phosphorylation of intracellular dihydroxyacetone requires EI, HPr, and an EIIA of the mannose family (7).

PTS proteins are often fused to each other, thus forming multifunctional polypeptides made up of two or more domains. Frequently, one or two of the soluble PTS components are fused to the N or C terminus of the membrane-spanning transport protein and are therefore located at the cytosolic side of the membrane. For example, the oligo-β-glucoside-specific PTS of Bacillus subtilis is composed of three distinct proteins, whereas in the mannitol-specific PTS permease of the same organism, the EIIB component is fused to the C terminus of the membrane-spanning EIIC and only EIIA^Mtl is a distinct protein (Fig. 1). The mannose-type PTSs are an exception because they possess two transmembrane proteins, as is the case for most mannose/glucose-specific PTSs of firmicutes as well as the low-efficiency fructose-specific levan PTS of B. subtilis (Fig. 1).

FIG 1.

Schematic presentation of the phosphorylation cascade formed by the B. subtilis PTS components necessary for the uptake of fructose (PTS^Lev), mannitol (PTS^Mtl), and cellobiose (PTS^Lic). The two general PTS components EI and HPr phosphorylate the EIIAs, which are specific for a certain carbohydrate. B. subtilis contains nine entire PTSs, six PTSs lacking an EIIA component, and one PTS lacking EIIA and EIIB components (49). For the seven incomplete PTSs, EIIA and EIIB components of other PTS, most likely from the same family, probably complement the transport and phosphorylation functions. Nevertheless, the PTSs usually exhibit different sugar specificities. The P∼EIIAs transfer the phosphoryl group to their cognate EIIB, which finally phosphorylates the carbohydrate bound to the corresponding membrane-integral EIIC or, for the fructose-specific levan PTS, to EIIC and EIID. The phosphorylated carbohydrate is subsequently released into the cytoplasm. While the PTS phosphorylation cascades for cellobiose and fructose are formed by EI, HPr, and two distinct EIIA and EIIB proteins, the EIIB component of PTS^Mtl is fused to the C terminus of the EIIC domain and is therefore attached to the cytoplasmic side of the membrane. Shown are also the fructose-1,6-bisphosphate (FBP)-stimulated and ATP-requiring phosphorylation of HPr at Ser-46 as well as the dephosphorylation of P-Ser-HPr, which follows a phosphorolysis reaction with P-Ser-HPr and P_i being the substrates and HPr and pyrophosphate (PP_i) the products (22). The ATP-dependent phosphorylation of HPr occurs in firmicutes but also in many proteobacteria containing HprK/P and the HPr paralogue NPr.

The PTS uses PEP as an energy source for the uptake of its substrates and as a phosphoryl donor for their phosphorylation. In order to phosphorylate the carbohydrates during their transport, the soluble PTS components form a phosphorylation cascade beginning with EI, which autophosphorylates at the N-ε3 position of a conserved histidyl residue at the expense of PEP (Fig. 1) (8). Phosphorylated EI (P∼His-EI) transfers the phosphoryl group to the N-δ1 position of His-15 in HPr (9), and histidyl-phosphorylated HPr (P∼His-HPr) passes the phosphoryl group on to one of several sugar-specific EIIAs usually present in a bacterium. (The symbol “∼” indicates an energy-rich phosphoryl bond, in contrast to, for example, seryl-phosphorylated proteins, which are written as P-Ser throughout this article.) EIIAs are also phosphorylated at the N-ε3 position of a histidyl residue (10, 11). P∼EIIAs, however, phosphorylate their cognate EIIB at a cysteyl residue (12), except the EIIBs of the mannose PTS family, which are phosphorylated at the N-δ1 position of a conserved histidine (13). In the last step, P∼EIIB transfers its phosphoryl group to a carbohydrate molecule bound to the cognate EIIC. There is so far only one carbohydrate, fucosyl-α-1,3-N-acetylglucosamine, which is transported by a PTS (that of Lactobacillus casei) without being phosphorylated (14). All other carbohydrates are phosphorylated during their transport and subsequently converted into phosphorylated intermediates of either the Embden-Meyerhof-Parnas, pentose phosphate, or Entner-Doudoroff pathway. Only maltose taken up via a PTS by Enterococcus faecalis cells was reported to be first dephosphorylated to maltose by the maltose-6′-phosphate (maltose-6′-P)-specific phosphatase MapP (15). Intracellular maltose is subsequently cleaved by the enzyme maltose phosphorylase into glucose-1-P and glucose, which are converted into glycolytic intermediates.

The crystal structure of the N,N′-diacetylchitobiose-specific EIIC component of Bacillus cereus with the disaccharide bound to the active site has been determined and indeed confirmed that the phosphorylatable hydroxyl group at the C-6′ position of the disaccharide is located close to the cytoplasm and can therefore be accessed by the P∼EIIB component (16). Phosphorylation of the carbohydrate is thought to lower its affinity for EIIC, and the phosphorylated carbohydrate is therefore released into the cytosol.

Among the about 230 archaea for which the genome sequence has been determined, more than 50 contain the general PTS components EI and HPr; most of these belong to the genus Haloferax. It seems that these organisms use the PTS mainly to transport and phosphorylate fructose (3, 17) or to phosphorylate dihydroxyacetone (18). A previous deduction that archaea are devoid of any PTS is therefore no longer valid (6). Methanopyrus kandleri AV19 possesses a protein (MK1512) corresponding to the short form of HprK/P present in alphaproteobacteria (19). As will be explained below, HprK/P phosphorylates HPr at Ser-46 (Fig. 1). The N-terminal part of MK1512 exhibits 55% sequence identity to a ca. 85-amino-acid-long region of HprK/P from firmicutes comprising the active site with the P loop. This organism also contains an HPr-like domain present in a protein of unknown function (MK0716). Interestingly, in the HPr-like domain the phosphorylatable His is replaced with a Glu, whereas the ATP-dependent phosphorylation site around Ser-46 is conserved. It is therefore likely that in some archaea PTS proteins also carry out regulatory functions, which are associated with phosphorylation of this serine residue.

Regulatory Functions of the PTS

Soon after the discovery of the PTS in Escherichia coli (20), it was realized that it not only transports and phosphorylates carbohydrates but also carries out regulatory functions. In Enterobacteriaceae, EIIA^Glc (also called Crr, for catabolite repression resistance) was found to be the central regulator of carbon metabolism. In firmicutes, this function is carried out by HPr, which in these organisms becomes phosphorylated not only at His-15 by PEP and EI but also at Ser-46 by ATP and the metabolite-controlled bifunctional kinase/phosphorylase HprK/P (Fig. 1) (21, 22). The latter HPr modification plays no role in sugar uptake and phosphorylation but serves exclusively regulatory functions. In addition to EIIA^Glc and HPr, many other PTS proteins also play regulatory roles. Intriguingly, numerous proteobacteria possess EI, HPr, and one or two EIIAs but lack any known EIIB and EIIC proteins. Although E. coli and other Enterobacteriaceae are devoid of HprK/P and the region around Ser-46, the ATP-dependent phosphorylation site in HPr, is not conserved, HPr proteins of proteobacteria with an incomplete PTS usually possess an HprK/P, and the region around Ser-46 strongly resembles that in firmicutes. These observations suggest that in these proteobacteria, the incomplete PTS is not involved in carbohydrate transport but carries out only regulatory functions by interacting with or by phosphorylating non-PTS target proteins.

Indeed, in most cases phosphorylation of non-PTS proteins by PTS components serves regulatory functions. There are nevertheless at least two reported cases where PTS proteins phosphorylate a non-PTS protein or its bound cofactor for enzymatic purposes. First, the phosphoryl group of P∼EI from Salmonella enterica serovar Typhimurium can be transferred to the intermediately phosphorylated active-site residue of acetate kinase (23). Second, P∼EIIA of the dihydroxyacetone phosphorylation system of numerous bacteria and several archaea (24) transfers its phosphoryl group to an ADP molecule tightly bound to the active site of the dihydroxyacetone kinase subunit DhaL, which subsequently phosphorylates dihydroxyacetone bound to DhaK (18). Some enterobacteria possess a more complex dihydroxyacetone-specific multidomain protein composed of an EIIA-like domain and an HPr-like domain followed by a truncated EI-like domain containing the conserved phosphorylatable histidine. This protein is phosphorylated by PEP, EI, and HPr at the truncated EI-like domain, and after primarily intramolecular transfer of the phosphoryl group to the HPr- and EIIA-like domains, it phosphorylates the ADP bound to DhaL (7).

In most cases, however, phosphorylation of non-PTS proteins by PTS components serves regulatory purposes. To allow the PTS to carry out its regulatory functions, several different mechanisms evolved. They include direct phosphorylation of the target proteins by PTS components. Phosphorylation can occur at PTS domains fused to the target proteins or at specific phosphorylation sites or domains that evolved in or were fused to the target protein. Alternatively, PTS components, in either phosphorylated or unphosphorylated form, interact with their target proteins. PTS-catalyzed phosphorylation of a target protein can stimulate or inhibit its activity, and this also holds for the interaction of phosphorylated and unphosphorylated PTS components with their target proteins. We describe these different scenarios in detail below.

PTS-MEDIATED REGULATION BY PHOSPHORYLATION

Since all cytoplasmic PTS proteins or membrane-associated hydrophilic PTS domains undergo transient phosphorylation, it is not surprising that numerous regulatory functions of the PTS are mediated by phosphoryl group transfer to the target proteins. Identical to the case for PTS proteins, all presently known PTS-regulated target proteins also become phosphorylated at histidine (phosphoamidate) or cysteine (thiophosphate) residues. This renders all phosphorylation steps reversible, because the standard free energies of the various high-energy phosphate bonds (P∼Nδ-1 or P∼Nε-3 in P∼histidines and P∼S in cysteines) are very similar (25–27). Consequently, if the PTS proteins are present primarily in phosphorylated form, the PTS-regulated target proteins are also mainly phosphorylated.

Two major factors were so far reported to affect the phosphorylation state of the PTS proteins in Enterobacteriaceae and probably most other bacteria. First, the uptake of an efficiently metabolized PTS sugar, such as glucose, and its concomitant phosphorylation lead to dephosphorylation of the PTS components. The phosphoryl group transfer from most P∼EIIBs to their cognate substrates seems to be more rapid than rephosphorylation via the PTS cascade. Second, changes in the ratio of PEP to pyruvate, the two compounds which are the substrate and product of the autophosphorylation of EI (Fig. 1), also affect the phosphorylation state of the PTS proteins (28, 29). The PEP-to-pyruvate ratio changes in response to the metabolic state of the cell. Starving cells have a high PEP-to-pyruvate ratio, and under resting conditions the PTS proteins are therefore mainly phosphorylated (28, 30) and the cells are primed to take up any PTS substrate they might encounter. In metabolically active cells, the PEP-to-pyruvate ratio is low and the PTS proteins are barely phosphorylated at His and Cys residues (30). Very low phosphorylation of the PTS proteins is therefore usually observed in cells efficiently metabolizing PTS sugars, such as glucose (28, 31). However, growth on non-PTS sugars can also lower the PEP-to-pyruvate ratio and therefore the phosphorylation state of the PTS proteins (28). Finally, α-ketoglutarate, which accumulates in E. coli cells exposed to nitrogen-limiting conditions, inhibits the autophosphorylation activity of EI. Nitrogen limitation therefore inhibits glucose uptake. In contrast, sudden nitrogen availability almost immediately increases glucose uptake and consumption without significantly altering the concentration of glycolytic intermediates (32).

In firmicutes, the metabolite-controlled HprK/P-mediated phosphorylation of HPr at Ser-46 (33) during the rapid metabolism of a carbohydrate slows the phosphoryl group transfer within the PTS phosphorylation cascade (34). The kinase function of HprK/P is enhanced during growth on efficiently utilized carbon sources (30, 35), and the resulting seryl-phosphorylated HPr (P-Ser-HPr) is a poor substrate for the PEP-dependent phosphorylation by P∼EI (36, 37). This mechanism further lowers the poor PEP-requiring phosphorylation of HPr and the EIIA and EIIB components in firmicutes growing on an efficiently metabolizable carbon source. The alterations of the phosphorylation state of the general and sugar-specific PTS components in response to the metabolic state of the cell and/or to the presence of a distinct carbohydrate are used to control numerous cellular functions.

Proteins Containing Their Own PTS-Recognized Phosphorylation Sites

One might expect that the development of a phosphorylation site recognized by PTS components would be the most straightforward and simplest way to render a protein “PTS controlled.” However, there is so far only one well-established example in which a protein developed a specific regulatory site (as opposed to the enzymatic sites described above) for phosphorylation by a PTS protein. This is glycerol kinase (GlpK) from firmicutes (38). GlpKs from firmicutes possess a surface-exposed flexible loop close to the dimer interface, which contains a histidyl residue surrounded by three aromatic amino acids, Aro-His-Aro-Aro (with Aro being Tyr or Phe in arbitrary order, depending on the organism) (39). This sequence does not resemble any known phosphorylation site in PTS proteins. The histidine residue in the above tetrapeptide (His-230 in GlpK from B. subtilis) (40) becomes phosphorylated at the N-ε3 position by PEP, EI, and HPr (38, 41). Despite a usually 60% amino acid sequence identity between GlpKs from firmicutes and other bacterial phyla, the phosphorylatable His is present only in GlpKs of firmicutes. Accordingly, GlpKs of no other phylum were so far found to be phosphorylated by PTS proteins, suggesting that this regulatory mechanism developed relatively late in evolution. In fact, GlpKs from Enterobacteriaceae possess at the same position a similar but histidine-less flexible loop for the binding of fructose-1,6,-bisphosphate (FBP), which leads to the formation of inactive tetramers (42). Proteins strongly resembling GlpKs from firmicutes but phosphorylating other carbohydrates, such as the presumed fuculose kinase Lmo1034 of Listeria monocytogenes (43), also lack the phosphorylation loop with the conserved histidine. The phosphorylation of GlpKs from firmicutes is reversible, and in contrast to the inhibitory effect of FBP binding on GlpKs of Enterobacteriaceae, it stimulates the GlpK-catalyzed ATP-dependent phosphorylation of glycerol 10- to 15-fold (44). Addition of an excess of HPr to purified phosphorylated GlpK leads to rapid dephosphorylation of the kinase and consequently lowers its activity (38). The uptake of efficiently metabolized PTS sugars, which leads to the dephosphorylation of PTS components and therefore also of GlpK, inhibits the activity of the kinase. Indeed, firmicutes in which EI or HPr have been inactivated not only are unable to utilize PTS carbon sources but also are unable to grow on glycerol as the sole carbon source (40, 45). The absence of GlpK phosphorylation leads to inducer exclusion, one of the regulatory mechanisms contributing to carbon catabolite repression, which will be discussed in detail in the section “Inducer exclusion in Enterobacteriaceae.” The regulatory His of GlpK is located about 20 Å from the active site. PTS-mediated GlpK stimulation therefore follows a long-range activation mechanism. Studies with an E. faecalis GlpK mutant protein in which the regulatory His-232 had been replaced with an arginine, which leads to a fully active enzyme without phosphorylation (46), suggested that phosphorylation induces structural rearrangements along the dimer interface that allow an optimal positioning of a crucial arginine residue (Arg-18) in the active site (47).

Proteins Containing a PTS Component Fused to the N or C Terminus

Another mechanism of PTS-mediated regulation that developed during bacterial evolution was the fusion of a PTS component to a target protein. In most cases the phosphorylatable His or Cys is conserved in the regulatory PTS domain and can therefore become phosphorylated by the proteins of the PTS phosphorylation cascade. For some proteins it was not an entire PTS domain that was fused to the target polypeptide but only a relatively small fragment containing the phosphorylatable histidine or cysteine. Phosphorylation or dephosphorylation of the PTS domain induces structural changes in the target protein, which lower or stimulate its activity. Proteins possessing a regulatory PTS domain include non-PTS transporters (48), various types of transcription activators (49–51), and two-component system response regulators (52). Several proteins and protein families containing a clearly identifiable PTS domain are listed in Table 1. There are also numerous proteins of unknown function that contain a PTS domain (6). The PTS-mediated regulation of the lactose transporters from Streptococcus thermophilus (53) and Streptococcus salivarius (54) and of several transcription activators from firmicutes (1) and some Enterobacteriaceae, such as FrzR of E. coli (55), has been studied in detail.

TABLE 1.

Proteins fused to PTS domains with proven or presumed regulatory functions

Organism(s)	Protein function	Fused PTS domain	Location of PTS domain	Conserved phosphorylation site	Effect of phosphorylation
Clostridium acetobutylicum	NtrC-type response regulator HprR CA_C3088	HPr	N terminus	His-15	?
Eubacterium limosum	BkdR-like transcription regulator	HPr	N terminus	His-16	?
Tepidanaerobacter acetatoxydans Re1	BkdR-like transcription regulator	HPr	N terminus	His-15	?
Streptococcus thermophilus	Lactose/galactose antiporter LacS	EIIAGlc	C terminus	His-552	+
Streptococcus salivarius	Lactose/galactose antiporter LacS	EIIAGlc	C terminus	His-552	?
Several Borrelia species	Na+/H+ symporter	EIIANtr	C terminus	His-622	?
Pasteurella multocida	Triosephosphate isomerase	EIIBGlc	C terminus	Cys-304	?
Actinobacillus succinogenes	Triosephosphate isomerase	EIIBGlc	C terminus	Cys-297	?
Several Vibrio species	Cyclic diguanylate phosphodiesterase-like	EIICLac	C terminus	171–432a	?
Firmicutes, actinobacteria, and a few proteobacteria	LevR-type transcription activators	EIIAMan	Fourth domain	His-585b	+
		EIIBGat	Penultimate domain	Cys-718	−
Numerous Clostridiales, Clostridium beijerinckii NCIMB 8052	LevR-type transcription activator, YP_001309609.1, YP_001307365.1, YP_001307684.1	EIIAMan	Fourth domain	His-629c	?
		EIIBGat	Penultimate domain	Cys-760	?
		EIIAMtl	C terminus	His-919	?
Numerous Clostridiales, Thermoanaerobacter ethanolicus JW200	LevR-type transcription activator, EGD53072.1	EIIAMan	Fourth domain	His-606	?
		EIIBGat	Penultimate domain	Cys-740	?
		EIIAMtl	C terminus	His-896	?
Several Selenomonadales and other firmicutes, Pelosinus fermentans B3	LevR-type transcription activator, EIW31722.1	EIIAMan	Fourth domain	His-604	?
		EIIBGat	Penultimate domain	Cys-744	?
		EIIAMtl	C terminus	His-898	?
Firmicutes, actinobacteria, and a few proteobacteria	MtlR/LicR-type transcription activators	EIIBGat	Penultimate domain	Cys-419d	−
		EIIAMtl	C terminus	His-599	−

^aThe number refers to the region corresponding to the EIIC^Lac domain.

^bThe numbers refer to the B. subtilis LevR sequence.

^cThe numbers refer to the C. beijerinckii LevR-like protein with the ID YP_001309609.1.

^dThe numbers refer to the B. subtilis MtlR sequence.

Transcriptional activators.

The PTS domain-containing transcription activators are mainly found in firmicutes and in actinobacteria but are less frequent in proteobacteria. Some of these bacteria possess transcription activators that can contain up to three EIIA and EIIB domains (Table 1). In fact, two types of EII domain-containing transcription activators can be distinguished. The LevR-type activators possess an N-terminal DNA binding helix-turn-helix motif followed by a homologue of the central ATP binding cassette-containing domain of NtrC-type regulators and four regulatory domains; two or, in a few cases, three of the regulatory domains are EIIA and EIIB domains of the mannose/glucose, galactitol, and mannitol/fructose PTS classes (1). These transcription activators contain other regulatory domains (Fig. 2), which are discussed below (see “Proteins Containing a Specific PTS-Recognized Phosphorylation Domain, the PRD”). The NtrC-type regulators containing three EII domains belong primarily to the Clostridiales, including Clostridia, Thermoanaerobacter, Carboxydibrachia, and a few other firmicutes (Table 1). The members of the second type of EII domain-containing transcription activators are composed of a DeoR-like helix-turn-helix motif followed by a winged helix-turn-helix domain resembling that in the Streptococcus pyogenes virulence gene regulator Mga and four regulatory domains, one of which is an EIIB of the galactitol-type PTS and one an EIIA of the mannitol/fructose class PTS (Fig. 2). According to the first members described for this family of transcription activators, they are called MtlR/LicR-type regulators (51, 56).

FIG 2.

Schematic presentation of the different regulatory domains in PRD-containing proteins. Shown are the B. subtilis antiterminator LicT (which binds to RNA) and several transcription activators (which bind to DNA), including the B. subtilis regulators MtlR and LevR. Antiterminators are composed of two PRDs fused to the RNA binding domain. The two PRDs usually contain four potential sites of PTS-mediated phosphorylation (73). Similarly, MtlR-like transcription activators contain two PRDs fused to the DNA binding and Mga-like domains (71). However, in MtlR-like transcription activators, the PRDs are followed by an EIIB^Gat-like domain and an EIIA^Mtl-like domain. B. subtilis MtlR needs to be activated both by phosphorylation at His-342 and by the interaction of its C-terminal EIIB^Gat- and EIIA^Mtl-like domains (marked with a bracket) with the EIIB^Mtl domain of the mannitol-specific PTS permease MtlA (146) (Fig. 1). The domain order in the B. subtilis transcription activator LevR is different from that in MtlR: the DNA binding and NtrC-like domains are followed by PRD1, EIIA^Man- and EIIB^Gat-like domains, and finally a truncated PRD2 (which contains only one conserved His). In all of the presented proteins the known stimulating phosphorylation sites (by P∼His-HPr) are indicated by red numbers, and inhibitory phosphorylation sites (by P∼EIIA or P∼EIIB) are written in blue (pale blue in MtlR indicates slight phosphorylation). Also presented in this figure is the LevR-like transcription activator ManR from L. monocytogenes, which, similar to the case for MtlR from B. subtilis, needs to be activated by both phosphorylation by P∼His-HPr at His-585 in the EIIA^Man-like domain (A. Zébré, E. Milohanic, and J. Deutscher, unpublished results) and interaction with the EIIB component MpoB (78). It should be noted that the phosphorylation sites are not always conserved. For example, ManR from L. innocua, which is almost identical to L. monocytogenes ManR, was reported to become phosphorylated by P∼His-HPr at His-506 in the PRD1 domain (77). Finally, in some LevR-like regulators of the order Clostridiales, the truncated PRD2 can be replaced with an EIIA^Mtl-like domain, as is the case for the C. beijerinckii protein with ID number YP_001309609.1.

Depending on the phosphoryl donor, phosphorylation of the EII domains in these transcription activators either stimulates their activity (phosphorylation by P∼His-HPr) or inhibits it (phosphorylation by P∼EIIA or P∼EIIB). For example, P∼His-HPr-mediated phosphorylation of the EIIA^Man-like domain of LevR from B. subtilis (57) and L. casei (50), the first intensively studied members of the EII and NtrC domain-containing regulators, activates the expression of the lev operon, which encodes the proteins for the extracellular degradation of the polysaccharide levan to fructose monomers and the components for a low-efficiency PTS specific for fructose and mannose. The absence of P∼His-HPr-mediated phosphorylation during the uptake of an efficiently metabolized carbon source represents one of the carbon catabolite repression (CCR) mechanisms operative in firmicutes. Nevertheless, in many firmicutes synthesis of the PTS transport components for the major repressing carbohydrates, glucose and mannose (ManLMN), is also regulated by a LevR-like protein (ManR), which in L. monocytogenes needs to be activated by P∼His-HPr-mediated phosphorylation at the EIIA^Man domain (A. Zébré, E. Milohanic, and J. Deutscher, unpublished results). An autoregulation mechanism for glucose transport therefore seems to be operative. In contrast, phosphorylation by the cognate P∼EIIB component usually inhibits the activity of transcription activators, and its absence is used for induction of the corresponding PTS operon. Such a regulatory mechanism has been reported, for example, for B. subtilis LicR, which controls the expression of an operon encoding the components of a cellobiose-specific PTS. The uptake of cellobiose by the PTS^Cel is assumed to lead to dephosphorylation of the corresponding EIIB component and therefore to prevent the EIIB^Cel-mediated phosphorylation of the C-terminal EIIA^Mtl-like domain of the transcription activator LicR, thereby allowing the expression of the cel (or lic) operon. A similar regulation mechanism is probably operative for those LevR-like transcription activators that contain a C-terminal EIIA^Mtl-like domain (Fig. 2). In contrast, induction of the mannitol operon of B. subtilis, which is controlled by the transcription activator MtlR, occurs only when the P∼EIIA^Mtl-mediated phosphorylation of MtlR at the conserved cysteyl residue in the penultimate EIIB^Gat-like domain is prevented by the presence of mannitol in the growth medium (58). So far, B. subtilis MtlR is the only example of P∼EIIA-mediated phosphorylation of a transcription activator at the conserved Cys in the EIIB^Gat-like domain (Fig. 2).

The lactose-specific transporter LacS.

LacS from S. thermophilus contains an EIIA^Glc-like domain fused to its C terminus (Table 1 and Fig. 3). This natural hybrid protein catalyzes slow H⁺/lactose symport as well as fast lactose/galactose counterflow (59). Intracellular lactose is cleaved into glucose and galactose. The latter hexose cannot be metabolized by S. thermophilus and is therefore secreted. The export of galactose is coupled to the uptake of lactose via a counterflow mechanism. Phosphorylation of the EIIA^Glc-like domain of LacS was demonstrated for S. thermophilus (60) and S. salivarius (61). It increases the affinity of LacS for its substrate and the speed of the counterflow reaction (62). Efficient uptake of lactose therefore requires a functional β-galactosidase and phosphorylation of LacS by PEP, EI, and HPr (Fig. 3). By carrying out in vitro and in vivo complementation studies with truncated LacS lacking the EIIA^Glc-like domain, it could be established that the unphosphorylated EIIA^Glc-like domain does not bind to truncated LacS. In contrast, the phosphorylated EIIA^Glc-like domain interacted with truncated LacS and stimulated its lactose/galactose counterflow activity (63).

FIG 3.

PTS-catalyzed glucose uptake and the HPr/Crh “regulon” in firmicutes. HPr, the central regulator of carbon metabolism in firmicutes, exists in four different forms: unphosphorylated, phosphorylated at His-15, phosphorylated at Ser-46, and doubly phosphorylated. B. subtilis contains in addition the HPr paralogue Crh, which lacks His-15 and therefore can be phosphorylated only at Ser-46. Histidyl-phosphorylated HPr prevails when a less favorable carbon source is utilized, whereas P-Ser-HPr and P-Ser-Crh are formed when glucose or other preferred sugars are metabolized. The utilization of glucose leads to an increase of the concentration of the glycolytic intermediate FBP, which stimulates the kinase function of HprK/P. Either the different forms of HPr (and B. subtilis Crh) interact with their target proteins (YesS, MG synthase, CcpA, and MalK) or P∼His-HPr phosphorylates them (glycerol kinase, PRD-containing transcription activators and antiterminators, and LacS). The following regulatory functions of HPr and Crh and their various phosphorylated forms are presented. (i) Unphosphorylated Crh of B. subtilis interacts with methylglyoxal synthase (MG synthase) and inhibits its activity. Methylglyoxal synthase is an enzyme at the entry point of the methylglyoxal bypass of glycolysis that catalyzes the transformation of dihydroxyacetone-P into methylglyoxal. (ii) P-Ser-HPr as well as P-Ser-Crh interacts with the B. subtilis glycolytic enzyme glyceraldehyde-3-P DH (GAP DH) and inhibits its activity. (iii) P-Ser-HPr and in B. subtilis also P-Ser-Crh interact with CcpA and stimulate its repressor function for CCR by binding to the cre operator sites of numerous catabolic genes. (iv) P-Ser-HPr of lactobacilli and lactococci inhibits maltose uptake by an inducer exclusion mechanism by probably directly interacting with a component of the ABC transporter. (v) P∼His-HPr interacts with and stimulates the B. subtilis transcription activator YesS, which controls the expression of the pectin/rhamnogalacturonan genes. (vi) P∼His-HPr phosphorylates and activates several PRD-containing antiterminators (AT) and transcription activators (TA), and the absence of their phosphorylation during glucose metabolism is used as a CcpA-independent CCR mechanism. (vii) P∼His-HPr also phosphorylates and activates glycerol kinase (GlpK). The absence of GlpK phosphorylation leads to inducer exclusion. (viii) Finally, in streptococci, P∼His-HPr phosphorylates the EIIA^Glc-like domain of LacS and stimulates the lactose/galactose exchange reaction catalyzed by this protein.

HPr domain-containing proteins.

Several proteins were found to contain an HPr-like domain fused either to their N or C termini (Table 1). The presence of an HPr-like domain at the N terminus of an NtrC-type regulator with a C-terminal DNA binding domain (but without EII domains) in Clostridium acetobutylicum ATCC 824 suggested a possible cross talk between this transcription regulator (called HprR) and the PTS (52), with the HPr-like domain functioning as the receiver module (Fig. 4). His-15, the PEP-dependent phosphorylation site for P∼EI, is nicely conserved in the HPr-like domain, whereas the surrounding of Ser-46 differs from that usually present in HPr of firmicutes. Indeed, a gene encoding a paralogue of EI (with 54% sequence identity to B. subtilis EI) is located just upstream from hprR and oriented in the opposite direction (Fig. 4). This is the only ptsI gene present in C. acetobutylicum, although a second ptsH gene is located elsewhere in the genome (unpublished observation). HPr encoded by the second ptsH also contains a well-conserved Ser-46. It is likely that the EI protein not only phosphorylates the second HPr but also transfers the phosphoryl group to His-15 in the HPr domain of HprR. Phosphorylation probably leads to structural changes affecting the affinity of HprR for its DNA target site. HprR does not contain the receiver module with the Asp phosphorylation site normally present in NtrC-like regulators. Activation of HprR by inter- or intramolecular transfer of the phosphoryl group from the HPr-like domain to a second phosphorylation site in the NtrC-like regulator is therefore unlikely. Similar to the case for other NtrC-type regulators, HprR probably controls the expression of RpoN (σ⁵⁴)-dependent transcription units. Interestingly, the gene encoding the EI paralogue is preceded by a typical RpoN-dependent promoter, and its expression was therefore proposed to be controlled by HprR (52); HprR might nevertheless control the expression of other genes.

FIG 4.

Gene arrangements around the hprR gene, encoding a σ⁵⁴-dependent regulator, in different organisms. The hprR gene is represented by the yellow arrows; when it is fused to ptsH, we call the gene ptsH-hprR and the protein accordingly PtsH-HprR. The hprR gene fused to ptsH was discovered in C. acetobutylicum, where a ptsI gene oriented in opposite direction is located upstream from ptsH-hprR. The ptsI gene is expressed from a σ⁵⁴-dependent promoter and therefore probably controlled by PtsH-HprR (CA_C3088). In other organisms, ptsH-hprR is part of the dhaMKL (A. woodii, Awo_c08990) or dhaKLM (C. beijerinckii, Cbei_2147) operon. The latter order of the dha genes is more common and is also found in H. modesticaldum and B. cereus. However, HprR of these two organisms (HM1_0841 and II5_05639, respectively) does not contain an HPr domain fused to its N terminus. Instead, a separate ptsH gene is located downstream from the dhaKLM genes, which is followed by a ptsI gene in H. modesticaldum and by a glpF-like gene in B. cereus. In the latter organism, a gene encoding a glycerol dehydrogenase-like protein (gldA) is located upstream from hprR and oriented in the opposite direction.

Genome sequencing revealed that HprR is present not only in C. acetobutylicum but also in many other bacteria of the order Clostridiales. In addition, an identical organization of the genes encoding HprR and the EI paralogue is found in several of them. In other Clostridiales the hprR gene is followed by the genes encoding the proteins for dihydroxyacetone phosphorylation, in the order dhaMKL or dhaKLM (Fig. 4). Interestingly, in a few species, such as Halobacteroides halobius and Heliobacterium modesticaldum, HPr is not fused to HprR, but ptsH and ptsI genes are located downstream from hprR, with the dhaKLM genes inserted between hprR and ptsHI (Fig. 4). Because in certain bacteria HprR exists without an HPr domain, we suggest that the form without the PTS protein be called HprR and that it be named PtsH-HprR when the PTS domain is fused to it. Accordingly, the gene names should be hprR and ptsH-hprR (Fig. 4). H. halobius and H. modesticaldum contain only the ptsH and ptsI genes located next to hprR. The gene arrangement in these organisms suggests that HprR might regulate the expression of the dha operon in response to dihydroxyacetone or glycerol availability. This hypothesis is further supported by our observation that in Bacillus thuringiensis IBL 200 and Bacillus cereus MSX-A1, the hprR gene is preceded by a gene encoding a glycerol dehydrogenase and followed by dhaK, dhaL, dhaM, ptsH, and glpF, with the last gene encoding a glycerol facilitator-like protein (Fig. 4). A ptsHI operon in these two organisms is located elsewhere on the genomes of the two bacilli. Glycerol metabolism via the enzymes encoded by the dha operon and the adjacent genes in the two bacilli thus resembles the glycerol utilization systems present in L. monocytogenes (64) and E. faecalis (65), which first oxidize intracellular glycerol to dihydroxyacetone, which is subsequently phosphorylated by the DhaKLM system.

HprR and PtsH-HprR probably control the expression of the glycerol/dihydroxyacetone system, depending on the phosphorylation state of the HPr-like domain or protein encoded by the associated or fused ptsH gene. In fact, in organisms possessing only the ptsH gene fused to hprR, such as Clostridium beijerinckii, the HPr-like domain of HprR probably participates in dihydroxyacetone phosphorylation by transferring the phosphoryl group received from P∼EI to DhaM (EIIA^Dha), which in turn passes it on to the ADP molecule bound to DhaL. As already mentioned, the resulting ATP is used by DhaL to phosphorylate dihydroxyacetone bound to DhaK. It is therefore likely that the presence of dihydroxyacetone will lead to dephosphorylation of the HPr-like domain of HprR. Therefore, either the unphosphorylated general PTS component stimulates the transcription activator function of HprR or the phosphorylated HPr domain inhibits it. A similar activation or inhibition of HprR by HPr or P∼His-HPr, respectively, probably occurs in bacteria where the phosphocarrier protein is encoded by a distinct protein. Finally, it should be noted that C. acetobutylicum does not possess a dha operon in its genome. The ptsI gene located upstream from hprR (Fig. 4) therefore remains the only presumed target for the σ⁵⁴-dependent regulator in this organism.

Interestingly, a protein belonging to the Fis family of transcriptional regulators from the two firmicutes Eubacterium limosum and Tepidanaerobacter acetatoxydans Re1 also contains an HPr-like domain (unpublished observation). These proteins exhibit significant sequence identity to BkdR of B. subtilis; however, B. subtilis BkdR, which controls the expression of the genes necessary for the utilization of the branched-chain amino acids valine and isoleucine (66), lacks an HPr-like domain. Unfortunately, for none of the HPr domain-containing proteins listed in Table 1 has the presumed regulatory role of the PTS domain been studied in detail.

Pyruvate kinase of firmicutes is a well-known example of proteins which do not contain an entire PTS domain fused to them but only a relatively small fragment, including the phosphorylation site with the conserved histidine or cysteine. In the case of pyruvate kinase, a peptide composed of about 50 amino acids exhibiting strong similarity to the surrounding of the EI autophosphorylation site is fused to its C terminus. The phosphorylatable His is located at position 539 in the sequence of B. subtilis pyruvate kinase, and the surrounding sequence (GGLTSHAAV) is nicely conserved compared to the phosphorylation site of B. subtilis EI (GGRTSHSAI). The role of the EI fragment is not known. Pyruvate kinase binds PEP to its active center but does not autophosphorylate at the conserved histidine in the EI fragment. The B. subtilis enzyme is also not phosphorylated by P∼EI or P∼His-HPr (G. Boël and J. Deutscher, unpublished results). It is therefore tempting to assume that HPr in phosphorylated or unphosphorylated form might bind to the EI-like fragment and thus affect the activity of pyruvate kinase.

Proteins fused to other PTS domains.

In Pasteurella multocida Pm70 (6) and several other P. multocida strains as well as in Actinobacillus succinogenes 130Z, an entire EIIB^Glc-like domain is fused to another glycolytic enzyme, namely, triosephosphate isomerase. The phosphorylatable Cys is conserved in all these proteins, suggesting that their activity might be controlled by PTS-mediated phosphorylation in response to the presence or absence of glucose, which in most Pasteuriaceae is taken up via a PTS. The Na⁺/H⁺ symporter of Borrelia burgdorferi, the causative agent of borreliosis, and of other Borrelia strains carries an EIIA^Ntr-like domain fused to its C terminus (6). The phosphorylatable His residue is well conserved in the EIIA^Ntr-like domain. EIIA^Ntr is a PTS protein assumed to be involved in the regulation of nitrogen metabolism (for a recent review, see reference 67). Finally, several Vibrio species possess an EAL domain characteristic of cyclic diguanylate phosphodiesterases (68), which contains fused to its C terminus an EIIC^Lac-like domain lacking about 170 amino acids at the N terminus. An EIIB^Lac-like domain usually fused to EIIC^Lac is also absent. This protein therefore cannot be regulated by phosphorylation, because the EIIC domain does not contain a PTS phosphorylation site, but rather might respond to the binding of an extracellular substrate or interact with an EIIB^Lac protein or domain of another PTS.

Another example of a protein containing a fragment of a PTS component is the C4-dicarboxylate transporter DcuA of Enterobacteriaceae, which contains at its C terminus a region of about 45 amino acids strongly resembling (45% sequence identity) the central part of EIIA^Glc, including the phosphorylation site (11). Although the surrounding of the phosphorylatable His of EIIA^Glc is nicely conserved in DcuA (GVELFVHFG in EIIA^Glc and GVALAVCFG in DcuA), the His itself is replaced with a Cys. However, Cys residues are also sometimes used as phosphorylation sites in PTS proteins (EIIB domain) and also in at least one PTS-controlled transcription activator (58). It is therefore tempting to assume that the DcuA transport activity might be controlled via either phosphorylation by or interaction with HPr or the EIIB^Glc domain of the glucose-specific PTS permease. However, it should be noted that according to topology predictions, the C-terminal part of DcuA is quite hydrophobic and is thought to be embedded in the membrane.

There are numerous other proteins which contain clearly identifiable regulatory PTS domains, but their physiological role remains obscure. A list of some of these PTS domain-containing proteins of unknown function was presented by Barabote and Saier (6). Several of them contain a presumed helix-turn-helix motif fused to EIIA^Ntr; others were found to contain an EIIA^Fru-like domain at the N terminus and an FPr-like domain (fructose-specific HPr) at the C terminus. It will be interesting to unravel the role of some of these presumed regulatory PTS domains.

Proteins Containing a Specific PTS-Recognized Phosphorylation Domain, the PRD

Structures of PRD and PRD-containing proteins.

In the course of evolution, bacteria must have produced a peptide composed of about 85 amino acids forming principally two slightly twisted antiparallel α-helices connected via an unstructured loop. The second α-helix contains a histidine, which is arranged in such a way that it can be phosphorylated by at least two different PTS proteins. The DNA region encoding this small peptide was probably duplicated in order to encode a protein fragment of about 170 amino acids. This polypeptide forms dimers (69, 70), thus finally providing what is known as the PTS regulation domain (PRD). Owing to duplication and dimerization, a PRD usually contains four conserved potential PTS phosphorylation sites. PRDs are frequently found to be fused to transcription regulators in firmicutes and actinobacteria but are less abundant in gammaproteobacteria (mostly in the family Enterobacteriaceae) and seem to be absent from alpha-, beta-, delta-, and epsilonproteobacteria. It is therefore likely that the PRDs evolved before the separation of firmicutes and actinobacteria and that some gammaproteobacteria acquired them later by horizontal gene transfer. Horizontal gene transfer probably also explains the presence of PRD-containing regulators in a few spirochaetes (treponemae and brachyspirae). Interestingly, in several pathogenic E. coli strains, genes encoding PRD-containing regulators are sometimes located in pathogenicity islands (55), which also supports the concept of horizontal gene transfer.

PRDs have so far been found in antiterminators and two types of transcription activators. These transcription activators were discussed in the preceding section, because they also contain EIIA and EIIB domains. Two PRDs are usually fused to the RNA binding (coantiterminator) domains of transcription antiterminators (called BglG/SacY-type antiterminators according to the first discovered members). The schematic presentation of LicT (a well-studied BglG/SacY family representative) in Fig. 2 shows only the LicT monomer. However, the crystal structure revealed that LicT forms dimers in which the two LicT subunits are aligned in parallel. PRD1 and PRD2 of one subunit interact with PRD1 and PRD2, respectively, of the second subunit (70).

Two PRDs are also fused to the DNA binding domain and the Mga-like domain of MtlR/LicR type transcription activators (Fig. 2) (1). Again, only a monomer of the B. subtilis MtlR is presented in Fig. 2. However, the transcription activator is likely to form dimers similar to the model proposed in reference 71, with PRD1 and PRD2 of each subunit interacting with PRD1 and PRD2, respectively, of the second subunit. Finally, two PRDs are usually also present in LevR-type transcription activators, which contain a helix-turn-helix motif followed by an NtrC-like central domain and the first PRD (72) (Fig. 2). In these transcription activators, an EIIA^Man-like domain and an EIIB^Gat-like domain separate PRD1 from the usually truncated PRD2 containing only one phosphorylatable His. As already mentioned, in LevR-like proteins of several Clostridiales and a few other firmicutes, the second PRD is sometimes replaced with an EIIA^Mtl-like domain (Table 1) (Fig. 2).

PRD-containing antiterminators.

The well-studied antiterminator LicT of B. subtilis controls the expression of the bglPH and bglS genes, which encode the PTS permease and catabolic enzymes for the uptake and metabolism of β-glucosides. The second PRD of LicT drastically alters its structure in response to specific mutations (His-Asp replacements) rendering the regulator constitutively active (69, 70). Similar structural changes are assumed to occur when PRD2 is phosphorylated by P-His-HPr or when other activating mutations are introduced into PRD2 (73, 74). These changes are probably transmitted to the RNA binding domain via PRD1 and finally lead to enhanced affinity of LicT for its target site, the ribonucleotidic antitermination target (RAT), on the corresponding nascent mRNA (1).

Unfortunately, LicT is so far the only PRD-containing protein for which the crystal structures of the two regulatory PRDs in active and inactive forms have been solved, but it is likely that similar phosphorylation-induced structural changes occur in PRDs of other antiterminators and probably also of the transcription activators. PRDs become phosphorylated by P∼His-HPr or P∼EIIBs. In most antiterminators, P∼His-HPr phosphorylates one or two histidine residues in PRD2 (73) and P∼EIIB phosphorylates one or two histidine residues in PRD1 (75).

PRD-containing transcription activators.

The phosphorylation of PRDs in transcription activators is more variable. In some transcription activators, such as MtlR, only one PRD becomes phosphorylated by EI and HPr (58, 76), whereas in others, such as LicR, all four histidines of the two PRDs are phosphorylated by the general PTS components and mutation of either one of these histidines leads to a loss of function (56). In most LevR proteins (50, 57) and LevR-like transcription activators (77, 78), the C-terminal PRD2 is phosphorylated by EI, HPr, and the cognate EIIA and EIIB components. Surprisingly, the LevR-like ManR protein of Listeria innocua was reported to become phosphorylated by EI and HPr at PRD1 (77), whereas the almost identical ManR of L. monocytogenes was found to be phosphorylated by EI and HPr in the EIIA^Man-like domain (Fig. 2) (A. Zébré, F. Aké, E. Milohanic, and J. Deutscher, unpublished results). The general rule seems to be that phosphorylation of a PRD by P∼His-HPr stimulates the activity of the regulator, whereas phosphorylation by P∼EIIB inhibits it. There is at least one antiterminator, BglG from E. coli, which might not follow this rule. BglG was reported to be stimulated by the general PTS components EI and HPr in a phosphorylation-independent manner (79). However, recently BglG from E. coli was reported to become phosphorylated also by PEP, EI, and HPr as well as by PEP, EI, and FruB, a hybrid protein composed of the fructose-specific FPr and EIIA^Fru (80). There is so far no explanation for these conflicting results.

In transcription activators, such as MtlR from B. subtilis, the stimulating signal has to be transmitted from amino acid His-342 in PRD2 across PRD1 and the Mga-like domain to the N-terminal DNA binding domain (71). The inactivating signal for LevR-type transcription activators is possibly transmitted even over a longer distance: from His-869 either to the active site for ATP hydrolysis in the NtrC-like central domain around amino acid 150 or all the way down to the N-terminal DNA binding domain (Fig. 2). As described above for the phosphorylation of EII domains in transcription activators, the absence of phosphorylation by P∼His-HPr owing to the uptake of an efficiently utilized carbon source is used as a CCR mechanism. In contrast, the absence of phosphorylation by the cognate P∼EIIB owing to the presence of the corresponding PTS substrate in the environment leads to the induction of the respective operon (1). A few antiterminators, such as B. subtilis GlcT, which controls the expression of the gene encoding the glucose-specific PTS permease and of ptsHI, do not require P∼His-HPr-mediated activation. In addition, P∼His-HPr-independent mutants affected in PRD1 or PRD2 could be obtained for antiterminators, such as LicT. These mutants allowed a conclusive confirmation that P∼His-HPr-mediated phosphorylation of these transcription regulators plays a role in CCR (74). The residual CCR observed for certain operons in firmicutes, in which the gene encoding the LacI/GalR-like catabolite control protein A (CcpA), the major player in CCR, had been deleted, disappeared when the corresponding regulator was mutated to a PTS-independent antiterminator (Pia) or when the formation of P-Ser-HPr was prevented. In the latter case, the absence of P-Ser-HPr allows an efficient phosphorylation of the antiterminator by P∼His-HPr, even when glucose or other repressing sugars are present (74). It should be noted that in certain firmicutes, other pleiotropic CCR mechanisms exist, such as glucose kinase-mediated CCR in Staphylococcus xylosus (81). Even in a ccpA background, pia mutants might therefore not always cause a complete relief from CCR.

The Mga regulator of S. pyogenes.

Mga, one of the regulators of Streptococcus pyogenes virulence genes, was also reported to be phosphorylated by EI and HPr (82). Phosphorylation was found to occur at a specific histidine in the central part of Mga, which was claimed to resemble PRDs. However, there does not seem to be significant sequence similarity between Mga and PRDs, and the alignment presented by Hondorp et al. (82) is not convincing. Significant sequence similarity between Mga and MtlR/LicR-type transcription activators was observed for only a small fragment of about 70 amino acids from the N-terminal part, which precedes PRD1 in the transcription activators (71). Hondorp et al. (82) also mention structural similarities between an E. faecalis Mga-like protein, for which the crystal structure has been solved, and the PRDs of LicT (69, 70). However, the tertiary structure of the dimers formed by these two proteins is completely different. The two subunits of unphosphorylated wild-type LicT are aligned in parallel, allowing the four histidines in the PRD1 and PRD2 dimers to face each other. In contrast, the two subunits of the E. faecalis Mga-like protein are oriented in an antiparallel fashion with mainly the C-terminal domain making contact to the other subunit. The regions supposed to represent PRDs and containing the presumed phosphorylatable His are therefore located on opposite ends of the dimer. Again, only about 70 amino acids from the N-terminal part of the E. faecalis Mga-like protein exhibit structural similarity to the region preceding PRD1 in PRD-containing transcription activators (71), which might contribute to DNA binding. It is therefore likely that what the authors report in reference 82 might be more important than just another PRD. Similar to the case for GlpK, Mga and related transcription regulators, such as AtxA (83), might have developed their own PTS-specific phosphorylation sites.

Independently from the question whether Mga contains PRDs or not, Hondorp et al. unequivocally show that PEP-dependent phosphorylation of S. pyogenes Mga requires the general PTS components EI and HPr (82). PEP-dependent phosphorylation of Mga strongly inhibited in vitro expression of emm, one of the Mga-regulated virulence genes. However, there remain some uncertainties concerning the site of phosphorylation in Mga (82). Replacement of His-270, one of the histidines proposed to be located in the presumed PRD1, did not significantly diminish Mga phosphorylation. Surprisingly, Hondorp et al. did not change the two other histidines proposed to resemble the phosphorylatable His in PRDs based on their sequence alignment (His-175 and His-301) but instead replaced His-204 with an alanine. This mutation also had no significant effect on Mga phosphorylation compared to the wild-type protein. Accordingly, doubly mutated Mga(His-204-Ala,His-270-Ala) was also still phosphorylated. Finally, triply mutated Mga(His-204-Ala,His-270-Ala,His-324-Ala), with His-324 also not being one of the presumed conserved PRD-related histidines, was only slightly phosphorylated. However, several radioactive bands of lower molecular weight appeared, suggesting that the triply mutated protein might be unstable and be degraded. It remains to be determined whether replacement of the presumed phosphorylatable His-175 and His-301 has an effect on Mga phosphorylation. Nevertheless, the results presented in reference 82 suggest that PTS-mediated phosphorylation does not occur in the C-terminal part of Mga, because truncated Mga lacking the last 139 amino acids was as strongly phosphorylated by PEP, EI, and HPr as the full-length protein.

The utilization of glucose and other rapidly metabolizable carbon sources by S. pyogenes stimulates the expression of its virulence genes (84). Two regulatory mechanisms are probably responsible for this effect. First, expression of mga was found to be stimulated by carbon catabolite activation. The elevated amounts of glucose-6-P and FBP present in cells utilizing glucose or other efficiently metabolized carbon sources probably allow binding of CcpA to a specific operator site preceding mga (84). Second, in the absence of an efficiently transported PTS sugar, the PTS components are predominantly phosphorylated at their conserved His and Cys residues. The results reported previously (82) suggest that under these conditions Mga is also phosphorylated, leading to its inhibition and therefore to poor expression of virulence genes. In contrast, in the presence of glucose, firmicutes convert most of their HPr into P-Ser-HPr, and little P∼His-HPr is therefore present. Mga is probably also barely phosphorylated and remains active, thus leading to strong virulence gene expression. Mutants producing Mga proteins with the above-described His-Ala replacement, but also additional mutants with presumed phosphomimetic His-Asp replacements, were tested as to whether they exhibit altered expression of Mga-regulated genes (82). The Ala replacements had only slight effects on the expression of the Mga-controlled arp and sof genes. In contrast, the presumed phosphomimetic mutants producing Mga(His-204-Asp,His-270-Asp) and Mga(His-204-Asp,His-270-Asp,His-324-Asp) exhibited very low Mga activity similar to that of an mga deletion mutant. However, owing to the uncertainty about the PTS phosphorylation site(s), it remains to be confirmed that there is a correlation between PTS phosphorylation and Mga activity.

REGULATION BY PROTEIN-PROTEIN INTERACTION

PTS components control the activity of their target proteins not only via phosphorylation but also via direct interaction. Interaction-mediated regulation seems to be even more frequent than regulation by phosphorylation. As described above for PTS-catalyzed phosphorylation of proteins at His or Cys residues, interactions with target proteins have been reported for HPr, EIIA, and EIIB components and in one case also for EI. Interactions occur with the unphosphorylated or phosphorylated forms of these proteins. The target proteins carry out various cellular functions, with the majority acting either as transport proteins or as transcription regulators.

Interaction with Unphosphorylated PTS Proteins

The interaction with a PTS protein and the accompanying activation or inhibition of the non-PTS target protein usually depend on the phosphorylation state of the PTS component. In fact, the phosphorylatable amino acids of PTS proteins and their surrounding always seem to be part of the interface of the protein-protein complex (85). If phosphorylation of a PTS component prevents the interaction with its target protein, the phosphorylatable amino acid is usually facing a negatively charged region. The phosphorylatable His, Cys, or Ser (in seryl-phosphorylated HPr) frequently adds to the stability of the complex by forming hydrogen bonds with amino acids of the target protein, which, of course, is prevented when the PTS protein is phosphorylated.

Inducer exclusion in Enterobacteriaceae.

In most established PTS-mediated regulation mechanisms, PTS components interact in their unphosphorylated forms with non-PTS proteins. The first PTS components shown to interact with a non-PTS protein were the Crr proteins from E. coli and S. enterica serovar Typhimurium, which were later identified as glucose-specific EIIAs of the PTS (EIIA^Glc) (86). E. coli mutants defective in EI and/or HPr had lost the capacity to synthesize the proteins necessary for the uptake and metabolism of several non-PTS carbon sources, including glycerol, lactose, and maltose (87). The repressive effect of the ptsHI mutation could be overcome by a second mutation located in a gene close to the ptsHI operon. This gene was called crr for catabolite repression resistant, because crr mutations also prevented the repressive effect of efficiently metabolizable carbohydrates on the synthesis of the enzymes necessary for the transport and metabolism of less favorable carbon sources.

Inactivation of ptsHI prevents the phosphorylation of EIIA^Glc, and it was therefore thought that under conditions where EIIA^Glc would be present mainly in unphosphorylated form, it would inhibit the synthesis of the catabolic enzymes required for the metabolism of non-PTS carbon sources. Unphosphorylated EIIA^Glc indeed prevails when an efficiently metabolized carbon source, such as glucose, is taken up (28). This concept was also in agreement with the observation that inactivation of the crr gene could overcome the repressive effect of the ptsHI mutation. However, mutants in which the repression caused by the ptsHI mutation was relieved only for one specific carbon source and not the others could be isolated. These mutations mapped not within crr but within the genes encoding the corresponding transporter, which therefore suggested that EIIA^Glc might interact with and inhibit proteins necessary for the uptake of the various carbon sources affected by the ptsHI mutation. Indeed, it was subsequently demonstrated that in E. coli and S. enterica serovar Typhimurium EIIA^Glc interacts with glycerol kinase (GlpK) (88) and the ATP binding protein of the maltose/maltodextrin ABC transport system (MalK) (89, 90) and that in E. coli EIIA^Glc interacts also with the lactose permease (LacY) (91, 92). In all cases, the interaction with unphosphorylated EIIA^Glc leads to inhibition of the non-PTS proteins (93, 94), which are usually transporters or transporter-associated components. Their inhibition prevents the uptake of the corresponding carbon source and consequently the formation of the necessary inducer; this regulation mechanism was therefore called inducer exclusion. In subsequent studies it was found that in Enterobacteriaceae, melibiose (95, 96), raffinose (97), galactose (98), and arabinose (99) also belong to the set of non-PTS sugars subject to inducer exclusion. The different carbohydrate utilization systems controlled by EIIA^Glc in E. coli are summarized in Fig. 5. EIIA^Glc-mediated inducer exclusion seems to be restricted to the family Enterobacteriaceae, indicating that this regulatory mechanism evolved relatively recently. It nevertheless represents the major CCR mechanism operative in Enterobacteriaceae (100). The term inducer exclusion is also used for EIIA^Glc-mediated GlpK inhibition, although GlpK is not a transport protein. Nevertheless, GlpK catalyzes the ATP-dependent phosphorylation of glycerol to glycerol-3-P (Fig. 5), the inducer of the glycerol operon in proteobacteria and most other bacteria. Because glycerol enters bacteria by facilitated diffusion, GlpK activity is important for efficient glycerol uptake. The transport-related activity of GlpK also explains why GlpK inhibition in firmicutes, which is not mediated via interaction with EIIA^Glc but occurs when the conserved phosphorylatable histidine in GlpK is not efficiently phosphorylated by P∼His-HPr (see “Proteins Containing a Specific PTS-Recognized Phosphorylation Domain, the PRD,” above), is also considered inducer exclusion (40). It is interesting to note that regarding GlpK regulation, two completely different inducer exclusion mechanisms evolved in firmicutes and Enterobacteriaceae, both of which nevertheless are mediated by PTS proteins: activation by P∼His-HPr-mediated phosphorylation in firmicutes and inhibition by interaction with unphosphorylated EIIA^Glc in Enterobacteriaceae.

FIG 5.

PTS-catalyzed glucose uptake and the EIIA^Glc inducer exclusion “regulon” of E. coli. Unphosphorylated EIIA^Glc but not phosphorylated EIIA^Glc interacts with several transport proteins or catabolic enzymes and inhibits their activity. Direct interactions have been shown for the catabolic enzyme glycerol kinase GlpK (88), the ATP binding subunit MalK of the maltose/maltodextrin-specific ABC transport system (89, 90), and the lactose permease LacY (91, 92). The crystal structures of the complexes formed between EIIA^Glc and the ATP-hydrolyzing MalK protein (106) as well as EIIA^Glc and GlpK, which catalyzes the ATP-dependent phosphorylation of glycerol to glycerol-3-P, have been solved (85). The uptake of the other four carbohydrates, l-arabinose, d-galactose, melibiose, and raffinose, is also subject to inducer exclusion, which is prevented when the EIIA^Glc-encoding crr gene is deleted. For the transporters labeled with a question mark, a direct interaction with EIIA^Glc is suggested by genetic data but has so far not been established by using biochemical or immunological methods.

Other EIIA^Glc interaction partners.

A regulatory role of EIIA^Glc completely different from that in inducer exclusion was reported for this PTS component in Vibrio vulnificus. EIIA^Glc as well as P∼EIIA^Glc was found to interact with a bacterial 100-kDa protein which exhibits strong similarity to mammalian insulysins (101). Insulysins are peptidases that degrade insulin, and the V. vulnificus protein was therefore called vIDE for Vibrio insulin-degrading enzyme. Although both unphosphorylated and phosphorylated EIIA^Glc interact with vIDE, only the unphosphorylated PTS protein stimulated its insulin-degrading activity. Deletion of the vIDE-encoding ideV gene significantly lowered the survival and virulence of V. vulnificus in mice. It was therefore concluded that the vIDE/EIIA^Glc complex plays an important role in the survival of the bacterium in the host cell by sensing glucose (101).

In a more general study, it was attempted to detect interaction partners of wild-type and unphosphorylatable (with replacement of the phosphorylatable His-90 with Ala) EIIA^Glc (11) of the pathogen Vibrio cholerae in planktonic and biofilm cells by carrying out a tandem affinity analysis. Several EIIA^Glc interaction partners were identified, some of which had already previously been shown to bind to the PTS protein in Enterobacteriaceae, such as glycerol kinase (see above) and adenylate cyclase (see “Interaction with phosphorylated EIIA components” below). However, several other proteins were also found to interact with EIIA^Glc. Interaction partners in planktonic cells include the glycolytic enzyme 6-phosphofructokinase, the purine repressor PurR, and a protein containing the GGDEF motif characteristic of diguanylate cyclases (VC0900) (102). In biofilm cells, wild-type and His-91-Ala mutant EIIA^Glc interacted with the gluconeogenic enzyme PEP carboxykinase. The two forms of EIIA^Glc also interacted with MshH, a homologue of the carbon storage regulator CsrD (103), a Mg²⁺ transporter, and a protein of unknown function (VC1291). Because for these three proteins the number of peptides found to interact with wild-type EIIA^Glc was significantly higher than the number of peptides binding to His-91-Ala mutant EIIA^Glc, it is likely that they interact primarily with P∼EIIA^Glc.

Among the proteins so far identified as EIIA^Glc interaction partners, the E. coli fermentation/respiration switch protein FrsA seems to have the highest affinity for unphosphorylated EIIA^Glc. A dissociation constant as low as 2 × 10⁻⁷ M was measured for the FrsA (previously called YafA) complex with EIIA^Glc (104). This protein is present in most facultative anaerobic gammaproteobacteria and is also found in a few alpha- and betaproteobacteria, such as Thauera selenatis and Hirschia maritima. Its deletion in E. coli leads to increased cellular respiration, whereas its overexpression causes elevated fermentation. FrsA was therefore proposed to regulate the partitioning of the flux through either the respiration or fermentation pathways in response to the phosphorylation state of EIIA^Glc (104).

The EIIA^Glc interface in protein-protein interactions.

The surface of EIIA^Glc interacting with the various target proteins is almost always the same. For the complex formed between E. coli EIIA^Glc and GlpK, the crystal structure has been solved (85). The phosphorylatable His-90 of EIIA^Glc (11) is part of the interface and is exposed to a negatively charged surface of GlpK, including Glu-478. This explains why phosphorylation of EIIA^Glc strongly lowers its affinity for GlpK. In contrast, the affinity of EIIA^Glc for GlpK was strongly enhanced by the presence of a Zn²⁺ ion in the complex, which was found to be liganded by amino acid residues from both proteins: His-90 of EIIA^Glc and Glu-478 of GlpK (105). Recently, the structure of the maltose-specific ABC transporter with bound EIIA^Glc has also been determined (106). Each of the two EIIA^Glc molecules bound to the ABC transporter interacts with both MalK subunits. The surface of EIIA^Glc containing the phosphorylatable His-90 (11) interacts with the nucleotide binding domain of MalK, and a second region centered around Lys-130 interacts with the dimerization domain of the other subunit. Binding of EIIA^Glc prevents MalK closure by stabilizing a resting state of the ABC transporter. MalK closure is required for the interaction of the maltose-loaded periplasmic maltose binding protein with the transmembrane proteins MalF and MalG and for the binding of ATP to MalK (Fig. 5). These two events induce the outward-facing conformation of the membrane proteins, which is essential for transferring maltose from the maltose binding protein to the binding site in the transmembrane proteins (106). Interestingly, out of the 17 residues of EIIA^Glc making contacts with the nucleotide binding domain of MalK, 11 are also involved in the interaction with GlpK (85), HPr (107), and EIIB^Glc (108), thus confirming that EIIA^Glc uses nearly identical surfaces for the interaction with its partner proteins in order to exert its catalytic and regulatory functions.

PTS^Ntr.

The other EIIA component that has been shown to interact with numerous non-PTS proteins in an unphosphorylated form is EIIA^Ntr, an EIIA of the fructose/mannitol PTS family, which exhibits neither sequence nor structure similarity to EIIA^Glc (109). In addition, the phosphorylation sites of the two proteins are quite different (11, 19, 110). In alpha-, beta-, and gammaproteobacteria, the ptsN gene, which encodes EIIA^Ntr, is frequently located downstream from rpoN (111). The RpoN protein, also known as σ⁵⁴, is required for the expression of several genes encoding enzymes for the utilization of nitrogen sources, but RpoN also regulates the expression of many other genes and operons. Inactivation of the ptsN gene from Klebsiella pneumoniae lowered the expression of several RpoN-controlled transcription units (112), suggesting that EIIA^Ntr might directly or indirectly affect RpoN activity. Also sometimes associated with the rpoN region is the npr gene (also called ptsO), which encodes NPr, a paralogue of HPr containing a phosphorylatable His-15 and, with the exception of Enterobacteriaceae, also a phosphorylatable Ser-46. In Enterobacteriaceae, a paralogue of EI is encoded by the ptsP gene (usually located distant from rpoN, ptsO, and ptsN). Together the three proteins EI^Ntr (PtsP), NPr (PtsO), and EIIA^Ntr (PtsN) form what is referred to as the nitrogen PTS (PTS^Ntr) (113). In Enterobacteriaceae, phosphorylation of NPr at His-15 is catalyzed mainly by EI^Ntr. Nevertheless, in bacteria also possessing EI, HPr, and sugar-specific EIIB and EIIC components, such as most Enterobacteriaceae, a cross talk between the two types of PTS exists (114, 115). The activity of EI^Ntr was recently shown to be antagonistically regulated by the metabolic intermediates situated at the intersection between carbon and nitrogen metabolism, namely, α-ketoglutarate and glutamine (116). Components of the PTS^Ntr are found not only in Enterobacteriaceae but also in many proteobacteria that are devoid of any known EIIB and EIIC components. In these organisms, the proteins of the PTS^Ntr are sometimes called EI, HPr, and EIIA^Fru. An HPr kinase able to phosphorylate NPr at Ser-46 and an EIIA of the mannose class PTS are usually also found in organisms containing the incomplete PTS^Ntr (111), except in Enterobacteriaceae, which are devoid of HprK/P. The genes encoding these five proteins are usually organized in two operons, as was shown for Brucella melitensis (19) and Ralstonia eutropha (110). In most alphaproteobacteria, such as B. melitensis and Agrobacterium tumefaciens, the hprK and ptsO genes are located downstream from the genes encoding a two-component system called chvI/chvG in A. tumefaciens (for chromosomal virulence) (111). In most betaproteobacteria, such as R. eutropha and Neisseria meningitidis, the EIIA^Man-encoding gene as well as ptsO and ptsP seem to form an operon, whereas ptsN and hprK are located in a second operon with hprK being followed by the rapZ (former yhbJ) gene (110). In E. coli, RapZ acts as adaptor protein allowing RNase E to degrade the small RNA GlmZ, which is required for the translation of the glucosamine-6-P synthase-encoding RNA (117).

Surprisingly, only EIIA^Ntr and not EIIA^Man of R. eutropha could be phosphorylated with PEP and purified EI^Ntr and NPr (110), although the phosphorylatable His residue is conserved in EIIA^Man. Owing to the absence of an entire PTS able to transport carbohydrates, it is not clear what exactly controls the phosphorylation states of the usually four soluble PTS components in these organisms. It is likely that the PEP/pyruvate ratio (28) and also the α-ketoglutarate/glutamine ratio (116) affect the phosphorylation states of these PTS proteins. In E. coli, the two metabolites α-ketoglutarate and glutamine bind to EI^Ntr and exert antagonistic effects on its activity. In contrast, in Sinorhizobium meliloti only glutamine was found to bind to the N-terminal GAF domain of EI^Ntr and to stimulate its autophosphorylation activity (118). In addition, it seems that the ATP-dependent phosphorylation of NPr affects the PTS phosphorylation cascade, because deletion of the hprK gene in B. melitensis led to significantly increased phosphorylation of EIIA^Ntr compared to that in the wild-type strain (19). In R. eutropha, an hprK mutant could be isolated only in an npr (ptsO) background, suggesting that elevated phosphorylation of HPr at His-15 owing to the absence of phosphorylation at Ser-46 negatively affects growth of this organism (110). Indeed, the npr mutant could easily be complemented with wild-type npr and the npr(His15Ala) allele, whereas no transformants could be obtained with the npr(Ser46Ala) allele.

Regulation of K⁺ transport.

The components of the PTS^Ntr are involved not only in the regulation of nitrogen-related genes and proteins (113). EIIA^Ntr also seems to play an important role in the control of low- and high-affinity potassium (K⁺) transport systems. Indeed, the PTS component was shown to interact with TrkA (119), a soluble protein required for K⁺ uptake via the transporter TrkH (120). TrkA forms a tetrameric ring which needs to bind ATP in order to exert its positive effect on the single-channel activity of TrkH (121). It is therefore tempting to assume that the interaction of EIIA^Ntr with TrkA prevents its stimulating effect on TrkH and therefore on the uptake of K⁺. Indeed, a ptsN mutant accumulates high concentrations of K⁺ inside the cells. K⁺ differentially affects the binding of σ⁷⁰ and σ^S to apo RNA polymerase and thus influences σ factor selectivity (122). In contrast to EIIA^Ntr, phosphorylated EIIA^Ntr does not interact with TrkA.

In addition to the low-affinity Trk system, E. coli also possesses a high-affinity transporter for K⁺ composed of the KdpFABC components, which form a P-type ATPase (123). The expression of the genes encoding these proteins is regulated by the two-component system KdpD/KdpE, with KdpD being a kinase sensing extracellular K⁺ probably via aspartyl residues located in its periplasmic loops (124) and KdpE being the response regulator. The two regulator genes kdpD and kdpE are located downstream from kdpFABC. When the K⁺ concentration in the medium is low, KdpD autophosphorylates and subsequently passes the phosphoryl group on to KdpE. The phosphorylated response regulator promotes transcription initiation of the kdpFABCDE genes (125). Unphosphorylated EIIA^Ntr was shown to bind to KdpD and to stimulate its kinase activity, thereby increasing the amount of phosphorylated KdpE and hence the expression of the kdpFABCDE genes (126). In agreement with this result, inactivation of the ptsN gene caused strongly reduced expression of the kdpFABCDE genes. As observed for other EIIA^Ntr interaction partners, phosphorylation of the PTS component prevents its interaction with KdpD. Owing to the already-mentioned cross talk in E. coli between the components of the regular PTS and the PTS^Ntr, growth on glucose also leads to dephosphorylation of EIIA^Ntr and consequently to elevated expression of the kdpFABCDE genes (126). The opposing regulation of the two K⁺ transport systems by EIIA^Ntr allows E. coli to shut down the low-affinity transporter TrkA when the bacterium is exposed to low K⁺ concentrations and to shut down KdpFABC when facing high K⁺ concentrations.

Interestingly, mutants of Rhizobium leguminosarum lacking both EIIA^Ntr proteins present in this organism were unable to grow at low K⁺ concentrations (127). Unphosphorylated EIIA^Ntr was assumed to interact with KdpD/E and to stimulate the expression of the genes encoding the high-affinity K⁺ transporter. Bacterial two-hybrid experiments indeed revealed an interaction of EIIA^Ntr with the transcription regulator proteins KdpD/E. In agreement with this concept, a ptsP mutant containing only unphosphorylated EIIA^Ntr due to the absence of EI^Ntr grew normally at low K⁺ concentrations. Interestingly, EIIA^Ntr was found to regulate several other ABC transporters in R. leguminosarum (127).

Other EIIA^Ntr interaction partners.

Another E. coli sensor kinase that was reported to be regulated by EIIA^Ntr is PhoR (128). Together with the response regulator PhoB, PhoR controls the expression of the pho regulon, which includes more than 30 genes organized in nine transcription units. When E. coli cells are exposed to low phosphate concentrations, the kinase activity of PhoR increases, which results in an elevated amount of P∼PhoB. The phosphorylated response regulator binds to specific sites on the DNA, the so-called Pho boxes, and stimulates the expression of the genes of the pho regulon. Mutants devoid of EIIA^Ntr exhibit reduced expression of the pho regulon. Deletion of ptsN lowers the expression of the pho regulon because in order to be fully active, the sensor kinase PhoR needs to interact with EIIA^Ntr. This activation seems to occur with unphosphorylated EIIA^Ntr, because mutants unable to phosphorylate the EIIA component also exhibited increased expression of the pho regulon (128).

Unphosphorylated EIIA^Ntr was also found to interact with the E1 subunit of Pseudomonas putida pyruvate dehydrogenase and to inhibit its activity (129). This enzyme produces acetyl coenzyme A (acetyl-CoA) from pyruvate, and the presence of EIIA^Ntr therefore inhibits the metabolism through the tricarboxylic acid cycle. Interestingly, through yeast two-hybrid experiments the EIIA^Man component of the incomplete B. melitensis PTS was demonstrated to interact with the E1 subunit (SucA) of 2-oxoglutarate dehydrogenase, which has a subunit composition similar to that of pyruvate dehydrogenase (19). In addition, a fusion protein composed of SucA and DivIVA sequestered EIIA^Man to the two cell poles, thus confirming the interaction of EIIA^Man with SucA. 2-Oxoglutarate dehydrogenase is part of the tricarboxylic acid cycle and catalyzes the transformation of 2-oxoglutarate into succinyl-CoA. The effect of phosphorylation of EIIA^Man on this protein-protein interaction and its physiological role have not been studied.

The virulence of the pathogen S. enterica serovar Typhimurium is also controlled by PTS proteins. First, it has been reported that inactivation of the general PTS components EI and HPr prevents the replication of this bacterium in macrophages (130). Second, deletion of EIIA^Ntr was found to indirectly affect the virulence of this pathogen. EIIA^Ntr was reported to interact with the response regulator SsrB, which is required for the expression of numerous genes located within pathogenicity island 2 of this bacterium (131). The interaction with EIIA^Ntr prevents SsrB from stimulating the expression of the genes in pathogenicity island 2.

Unphosphorylated EIIA^Ntr was also reported to be necessary for the derepression of the E. coli ilvBN genes, which encode acetohydroxy acid synthase I, the first common enzyme for the synthesis of branched-chain amino acids (132). Mutants with ptsN deletions were extremely sensitive to leucine-containing peptides, whereas ptsP and npr (ptsO) mutants were more resistant than the wild-type strain. However, the interaction partner of EIIA^Ntr for the regulation of the ilvBN genes has not yet been identified.

In the betaproteobacterium R. eutropha, EIIA^Ntr was found to interact with the bifunctional ppGpp synthase/hydrolase (SpoT1), an enzyme involved in the stringent response. Bacterial two-hybrid assays indicated that this interaction occurs only with the unphosphorylated form of EIIA^Ntr (133).

Interaction with EIIB components.

There are also several examples where EIIB components (as distinct proteins or as domains) interact with transcription regulators. When the EIIB component is fused to the cognate transmembrane permease, the target proteins are sequestered to the membrane. The first such example was the E. coli repressor Mlc, which was reported at nearly the same time by three different groups (134–136) to interact with the EIIB^Glc domain of PtsG, the glucose-specific PTS permease. Mlc is a repressor of the ROK (repressors, open reading frames, and kinases) family, and it was discovered because transformation of an E. coli wild-type strain with a plasmid into which the mlc gene had been inserted led to the formation of colonies with increased size when the transformants were grown on glucose-containing solid medium (137). The name mlc (making large colonies) was deduced from this phenotype. The mlc gene turned out to be allelic with the previously described dgsA (2-deoxy-d-glucose sensitive) locus (138).

E. coli Mlc is a regulator which represses the expression of several genes, including mlc, ptsG (ptsG encodes the EIIB^Glc and EIIC^Glc domains of the glucose-specific PTS permease), ptsHI-crr, manXY, and malT (1, 2, 4). The interaction of Mlc with the unphosphorylated EIIB^Glc domain inhibits its repressor function (for reviews, see references 1, 2, and 4). Deletion of the ptsG gene therefore prevented the expression of all Mlc-controlled genes (139, 140).

An interesting aspect of this regulation mechanism is that Mlc needs to be sequestered to the membrane by the EIIB^Glc domain in order to become inactive. When EIIB^Glc was produced as a distinct cytoplasmic protein instead of being fused to the membrane-integral EIIC^Glc, it was still able to bind Mlc but no longer inhibited its activity (141). Possibly, cytoplasmic EIIB^Glc has a lower affinity for Mlc and might therefore have lost its inhibiting effect. However, fusing the EIIB^Glc domain to another membrane protein prevented Mlc activity, suggesting that the interaction with the membrane environment is important for Mlc regulation (141, 142). The crystal structure of the complex formed by the EIIB^Glc domain and Mlc revealed that Mlc forms tetramers with each subunit binding an EIIB^Glc domain. In fact, EIIB^Glc binds to Mlc in the central part of the protein between the N-terminal DNA binding motif and the C-terminal dimerization domain. This interaction lowers the affinity of the repressor for its target sites (142). However, interaction with the EIIB^Glc domain alone does not seem to be sufficient for Mlc inactivation. Interaction with the hydrophobic membrane environment also seems to be required (143). EIIB^Glc-mediated membrane sequestration was proposed to cause structural restrictions leading to reduced flexibility and therefore to lower affinity for its DNA targets (142). Arg-424 located close to the phosphorylatable Cys-421 in PtsG makes polar contacts with the carboxylate of the C-terminal glycine of Mlc (136, 144). Unphosphorylated EIIB^Glc prevails when the cells take up glucose. As a consequence, the uptake of glucose leads to EIIB^Glc-mediated sequestration of Mlc to the membrane and therefore to increased expression of Mlc-repressed genes (140). It should be noted that Mlc activity is also inactivated by the interaction with the cytoplasmic protein MtfA (Mlc titration factor A) when this protein is overproduced (145).

Other proteins regulated by interaction with unphosphorylated EIIB domains or distinct EIIB proteins include PRD-containing transcription activators. The first example was MtlR from B. subtilis, the DeoR-type transcription activator of the mannitol operon, which is also regulated via phosphorylation by P∼His-HPr and P∼EIIA^Mtl (see “Proteins Containing a PTS Component Fused to the N or C Terminus” and “Proteins Containing a Specific PTS-Recognized Phosphorylation Domain, the PRD” above). In addition, the EIIB^Mtl domain of the mannitol-specific PTS permease interacts with the two C-terminal regulatory domains of MtlR and thereby stimulates its transcription activation function (146). EIIB^Mtl fused to the tyrosine kinase modulator YwqC (147), a transmembrane protein not related to mannitol transport, also interacted with MtlR and stimulated its activity. However, similar to what was reported for Mlc, interaction with the EIIB^Mtl domain is not sufficient for MtlR activation, because overproduction of EIIB^Mtl as a distinct cytoplasmic protein in the B. subtilis wild-type strain 168 prevented induction of the mtl operon. In this strain, soluble and membrane-associated EIIB^Mtl probably compete for binding MtlR, and owing to the excess of soluble EIIB^Mtl, little MtlR is sequestered to the membrane and activated. The interaction with the membrane environment is therefore also required for MtlR stimulation. This interaction seems to occur only with unphosphorylated EIIB^Mtl, because first, the Cys-Asp replacement in EIIB^Mtl prevented the interaction with MtlR in yeast-two hybrid experiments (146) and second, replacement of the phosphorylatable Cys with a Ser in soluble EIIB^Mtl prevented the inhibitory effect on PmtlA expression observed with wild-type EIIB^Mtl (71). In contrast to cysteyl-phosphorylated EIIB^Mtl, seryl-phosphorylated EIIB^Mtl cannot transfer its phosphoryl group to mannitol (148) and, owing to the low-energy phosphate bond, probably cannot pass it back to EIIB^Mtl. As a consequence, P-Ser-EIIB^Mtl accumulates in the cytoplasm, and the phosphorylated mutant EIIB^Mtl apparently no longer competes with the EIIB^Mtl domain of MtlA for binding MtlR and rendering it inactive (71).

Several results contradictory to the above model have been published by the group of Altenbuchner. For example, it was reported that the simultaneous deletion of mtlA and mtlF causes strong constitutive expression from the PmtlA promoter (149), whereas according to the results reported in reference 58, deletion of both mtlA and mtlF led to MtlR inactivation and only the single deletion of mtlF caused constitutive MtlR activity. In a more recent study from the Altenbuchner laboratory, deletion of the entire mtl operon, including mtlD, was found to inhibit expression from PmtlA (76). Although MtlR regulation in other bacterial species, such as Geobacillus stearothermophilus and L. casei (P. Joyet and J. Deutscher, unpublished results), significantly differs from that observed in B. subtilis, it is unlikely that the use of different B. subtilis strains (3NA and 168) might account for the observed differences in MtlR regulation. A more likely explanation might be the use of a plasmid-borne reporter gene fusion with an optimized PmtlA promoter in one study and a chromosome-integrated reporter gene fusion with wild-type PmtlA in the other. Heravi and Altenbuchner also recently reported that a B. subtilis mutant MtlR in which His-342 was replaced with an Asp and Cys-419 with an Ala was active even when produced in a mutant lacking MtlA (EIICB^Mtl) (76). These results were also thought to be contradictory to the above model of EIIB^Mtl-mediated MtlR activation. A possible explanation for this EIIB^Mtl-independent MtlR activity could be that the mtlR(His-342-Ala,Cys-419-Ala) double mutation not only prevents the inactivating phosphorylation at Cys-419 but also induces structural changes in the transcription activator resembling those caused by EIIB^Mtl-mediated membrane sequestration.

The second PRD-containing transcription activator regulated by the interaction with an EIIB domain is the LevR-like ManR from L. monocytogenes. This protein controls the expression of the manLMNO operon, which encodes the major glucose transporter of this human pathogen (78, 150). However, ManR activity is regulated not by the components of the mannose-type PTS ManLMN but by the proteins of another mannose-type PTS, MpoABCD, a low-affinity glucose/mannose-specific PTS that functions mainly as glucose sensor (78). An interaction between EIIB^Mpo (MpoB) and ManR was suggested by results from genetic experiments. While deletion of EIIA^Mpo led to constitutive expression of the man operon, deletion of EIIB^Mpo or of both EIIA^Mpo and EIIB^Mpo completely inhibited it, suggesting that EIIB^Mpo is required for ManR activity (78). An interaction of EIIB^Mpo with ManR could indeed be established by carrying out yeast two-hybrid experiments. Similar to the interaction of EIIB^Mtl with MtlR, EIIB^Mpo also interacts with two of the four regulatory domains of ManR (A. C. Zébré, M. Ventroux, M.-F. Noirot-Gros, J. Deutscher and E. Milohanic, unpublished results). However, in contrast to EIIB^Mtl, EIIB^Mpo is not fused to one of the two membrane components MpoC or MpoD but is a distinct cytoplasmic protein. As a consequence, it is unlikely that membrane sequestration plays a role in listerial ManR regulation.

PTS components also play a role in biofilm formation, as has been shown for EI and HPr of V. cholerae (151). In a recent study, the utilization of mannitol by this organism was reported to activate biofilm formation and the expression of the vps genes, which encode the proteins required for the synthesis of the exopolysaccharides forming the biofilm matrix (152). Even when grown on other carbon sources, the mannitol-specific PTS permease MtlA, an EIICBA protein, was found to be necessary for biofilm formation. In fact, it turned out that the EIIB^Mtl domain was sufficient for the stimulating effect. Synthesis of this domain from an IPTG (isopropyl-β-d-thiogalactopyranoside)-inducible allele in cells growing in the absence of mannitol also caused elevated biofilm formation. Phosphorylation of the EIIB^Mtl domain is not required for this effect, because a mutant overproducing EIIB^Mtl, in which the phosphorylatable His of the EIIA domain was replaced with an alanine, also exhibited elevated biofilm formation (152).

Finally, an interesting presumably phosphorylation-independent interaction occurs between the peptide SgrT and the EIIC-EIIB domains of the glucose/N-acetylglucosamine family of PTS proteins. SgrT is encoded by the 5′ end of the small RNA SgrS, whereas its 3′ part is complementary to the 5′ end of the ptsG mRNA and thus allows the formation of an RNA-RNA hybrid, which inhibits translation of ptsG and is a target for RNase E (153). The small peptide SgrT inhibits the transport function of the PTS transporter probably by binding to the linker region connecting the EIIC^Glc and EIIB^Glc domains. Indeed, replacement of a proline (Pro-384) present in a conserved region (-L-K-T-P-G-R-E-D-) of the linker in E. coli PtsG with an arginine prevented the repressive effect of SgrT on glucose utilization (154). Surprisingly, the conserved linker motif is also found in the PTS permeases PtsG, GamP, and NagP of B. subtilis, although there is no evidence for the presence of an SgrT homologue in firmicutes. The conserved linker motif might therefore carry out additional functions.

Interaction with HPr and its paralogues Crh and NPr.

A highly interesting interaction of HPr from enterococci with the response regulator of the CroRS two-component system was recently detected by carrying out a protein fragment complementation assay (155). CroR is required for intrinsic cephalosporin resistance in E. faecalis. It was therefore not surprising that deletion of HPr causes a cephalosporin hyperresistance phenotype, confirming that the interaction of HPr with CroR diminishes the intrinsic cephalosporin resistance of E. faecalis. Hyperresistance toward cephalosporins was also observed when either one of the two phosphorylatable amino acids of HPr, His-15 or Ser-46, was replaced with alanine. This result suggests that phosphorylation of HPr by either PEP and EI or ATP and HprK/P prevents the interaction with CroR. In agreement with these results, cephalosporin resistance of E. faecalis was also influenced by the utilization of carbon sources (155). While growth on the non-PTS carbon sources pyruvate and citrate, which leads to high levels of P∼His-HPr, caused enhanced cephalosporin resistance, growth on the PTS substrates glucose and N-acetylglucosamine, which leads to low levels of P∼His-HPr, had no such effect. The nutrient effect disappeared in a croR mutant, suggesting that it depends on the interaction between HPr and CroR.

Unphosphorylated HPr from E. coli and the two unphosphorylated HPr paralogues Crh from B. subtilis and NPr from E. coli were also found to interact with non-PTS target proteins. HPr from E. coli interacts with a regulatory protein and an enzyme involved in carbon storage. In the latter case, unphosphorylated as well as phosphorylated E. coli HPr was reported to bind with high affinity to glycogen phosphorylase (156). However, only the unphosphorylated PTS protein has a significant stimulating effect on glycogen phosphorylase activity (157). Neither NPr nor FPr (an E. coli HPr homologue specific for the transport of fructose) binds to glycogen phosphorylase. Interaction with HPr increases the V_max and lowers the K_m of glycogen phosphorylase for its substrate. It was proposed that under conditions where the PEP level drops and the PTS proteins are barely phosphorylated, the interaction with HPr will favor the breakdown of glycogen. Recently, HPr from E. coli was also reported to interact with the anti-σ⁷⁰ factor Rsd (31). Only unphosphorylated HPr formed a tight complex with Rsd and thus prevented the inhibitory effect of Rsd on σ⁷⁰-dependent transcription, as was shown in in vivo and in vitro experiments.

B. subtilis Crh contains a well-conserved Ser-46 phosphorylation site, but lacks His-15, which is replaced with a glutamine. Crh is therefore not phosphorylated by PEP and EI. However, similar to the case for B. subtilis HPr, this protein is phosphorylated by ATP and HprK/P. Unphosphorylated Crh was reported to interact with the methylglyoxal synthase MgsA, an enzyme of the methylglyoxal bypass of glycolysis, and to inhibit its activity (158). Phosphorylation of Crh at Ser-46 by ATP and HprK/P prevents this interaction. As a consequence, flux through the methylglyoxal bypass is probably low when no efficiently metabolized carbohydrate is taken up via the PTS. Under these conditions, B. subtilis cells contain high levels of unphosphorylated Crh (35), which inhibits MgsA. Interestingly, regulation of MgsA is a Crh-specific phenomenon because no interaction occurs between MgsA and HPr (158). It is therefore likely that Gln-15 of Crh, which distinguishes it from the His-15-containing HPr, plays an important role in this interaction.

Unphosphorylated NPr from E. coli was reported to regulate the production of lipid A, a component of the lipopolysaccharide layer. For that purpose, NPr interacts with the enzyme LpxD (159), an acetyltransferase essential for the synthesis of lipid A (160). This interaction seems to inhibit the activity of LpxD, because an npr mutant showed increased lipopolysaccharide synthesis. Phosphorylated NPr as well as unphosphorylated HPr did not interact with the acetyltransferase. The physiological role of PTS-mediated regulation of the production of the bacterial endotoxin lipid A is not yet understood. By carrying out yeast two-hybrid experiments, it was found that NPr from B. melitensis interacts with a pyrophosphate phosphatase (Ppa) (19). No further studies were carried out to demonstrate the potential physiological relevance of this protein-protein interaction.

Interaction with EI and EI^Ntr.

The PTS is also involved in carbohydrate-mediated chemotaxis, a regulatory phenomenon in which the frequency of the change of the direction of flagellar rotation (change between tumbling and swimming) determines the speed at which a bacterium approaches an attracting carbon source or moves away from a repellent (161). A central protein in the chemotactic signaling pathway is the autophosphorylating sensor kinase CheA, which passes its phosphoryl group on to the CheY protein. P∼CheY interacts with the lower part of the switch complex, which is composed of the FliG, FliM, and FliN proteins and associated with the flagella rotor, thus inducing clockwise rotation and therefore tumbling (162). In vitro studies with purified enzymes showed that E. coli EI, but not P∼EI, inhibits the autophosphorylation activity of CheA (163). Similar experiments with the B. subtilis proteins revealed an opposite mechanism, i.e., inhibition of CheA activity by P∼EI and not EI (164). These contradictory results are not surprising since in B. subtilis CheY exerts an effect on flagellar rotation antagonistic to that in E. coli. However, for neither the E. coli nor the B. subtilis system has a direct interaction between CheA and EI or P∼EI, respectively, been established by using biochemical or immunological methods, although the inhibiting effect of EI on CheA activity in in vitro experiments suggests a direct interaction of the two proteins (163).

EI^Ntr and aspartokinase from Bradyrhizobium japonicum were also reported to interact with each other (165). This interaction, which was confirmed by in vitro pulldown experiments, inhibits the autophosphorylation activity of EI^Ntr. Aspartokinase is encoded by the lysC gene, and its deletion prevented the uptake of δ-aminolevulinic acid and the utilization of prolyl-glycyl-glycine as a proline source. This phenotype could be compensated for by expressing ptsP in trans from a multicopy plasmid. The result suggests that aspartokinase exerts an effect on EI^Ntr that can be mimicked by the overproduction of the PTS protein.

Interaction with Phosphorylated PTS Proteins

Interactions with phosphorylated PTS components seem to be less frequent than interactions with the unphosphorylated proteins. One reason might be that they are more difficult to detect. Nevertheless, several interactions with phosphorylated HPr and HPr paralogues as well as phosphorylated EIIA and EIIB components have been reported. HPr of firmicutes as well as its paralogue Crh (catabolite repression HPr) present in bacilli is also phosphorylated at Ser-46 by ATP and HprK/P. This modification also occurs with the HPr paralogue NPr of HprK/P-containing proteobacteria (19, 110). The seryl-phosphorylated PTS components also carry out important regulatory functions by interacting with several specific target proteins. The phosphorylation sites of the PTS components are usually part of the interface, and the phosphoryl group faces positively charged amino acids in the target proteins, mostly arginine, which increases the stability of the protein-protein complex (166).

Interaction with phosphorylated EIIA components.

Genetic data had suggested that in Enterobacteriaceae phosphorylated EIIA^Glc stimulates the activity of adenylate cyclase (167). This enzyme converts ATP into cyclic AMP (cAMP), which is an important second messenger in prokaryotes and eukaryotes. Low intracellular levels of cAMP are observed in E. coli and S. enterica serovar Typhimurium cells grown on an efficiently metabolized carbon source or in mutants preventing the formation of P∼EIIA^Glc. In Enterobacteriaceae, cAMP binds to the cAMP receptor protein (Crp), and the resulting complex functions as transcription activator for numerous catabolic genes and operons (for a review, see reference 1). These results therefore strongly suggested that P∼EIIA^Glc interacts with adenylate cyclase and stimulates its catalytic activity. More specifically, cAMP formation seems to affect the lag phase characteristic for diauxie, which is observed when cells grow on a mixture of a preferred and a less favorable carbon source. Under these conditions, bacteria utilize the less favorable carbon source only when the preferred carbon source is exhausted (168), usually leading to two growth phases (hence the name diauxie) separated by a lag phase, during which the enzymes required for the transport and metabolism of the less-favorable carbon source are synthesized. Addition of extracellular cAMP either shortens or entirely eliminates the lag phase (169). However, addition of extracellular cAMP cannot prevent CCR, i.e., the preferred use of, for example, glucose over a less favorable carbon source, such as lactose (170). The interaction of EIIA^Glc with adenylate cyclase could indeed be confirmed by fusing adenylate cyclase to the E. coli integral membrane protein Tsr. Unphosphorylated EIIA^Glc as well as P∼EIIA^Glc was found to bind to the C-terminal regulatory domain of membrane-anchored adenylate cyclase. However, when adenylate cyclase activity assays were carried out in the presence of E. coli crude extracts, only P∼EIIA^Glc was able to stimulate the enzyme activity (171). An interaction of unphosphorylated EIIA^Glc with adenylate cyclase could also be demonstrated by in vivo experiments with the His-91-Ala EIIA^Glc mutant from V. cholerae by using the tandem affinity purification method (102).

Interaction with phosphorylated EIIB components.

While some PRD-containing transcription activators are regulated via interaction with unphosphorylated EIIB domains or proteins, antiterminators seem to be regulated rather by the interaction with phosphorylated EIIB. A well-established example of the interaction of an antiterminator with a phosphorylated EIIB component is the E. coli antiterminator BglG. Similar to the case for other PRD-containing antiterminators, BglG activity is inhibited by phosphorylation catalyzed by the P∼EIIB^Bgl domain of BglF (see “Proteins Containing a Specific PTS-Recognized Phosphorylation Domain, the PRD” above). In addition, BglG binds to the P∼EIIB^Bgl domain of the β-glucoside-specific PTS permease BglF (172). In the absence of a BglF-recognized substrate (salicin or arbutin), the EIIB^Bgl domain of BglF is mainly phosphorylated and interacts with the antiterminator BglG. While EIIB^Mtl-mediated membrane sequestration leads to activation of B. subtilis MtlR, EIIB^Bgl-mediated membrane sequestration inhibits the antiterminator activity of BglG. As soon as a substrate for the PTS^Bgl (salicin or arbutin) becomes available, the EIIB^Bgl domain of BglF will be dephosphorylated, which weakens its affinity for BglG. The antiterminator is therefore released into the cytoplasm, thereby regaining its activity and allowing efficient expression of the bgl operon.

In a recent study, the B. subtilis antiterminator LicT was also found to change its cellular localization in response to substrate availability. In the absence of a substrate, LicT is phosphorylated in its PRD1 by the P∼EIIB^Bgl domain, and the antiterminator is equally distributed in the cell. In the presence of a substrate, LicT is no longer phosphorylated by the P∼EIIB^Bgl domain and is located in subpolar regions (173). Although there is no evidence that the presence or absence of EIIB^Bgl or P∼EIIB^Bgl affects the activity of LicT by direct interaction, phosphorylation of the PRD1 of LicT by P∼EIIB^Bgl seems to play an important role in the cellular localization of the antiterminator.

Interaction with histidyl-phosphorylated HPr.

Presently, there is only one reported example where P∼His-HPr interacts with and controls a target protein. This is the B. subtilis transcription activator YesS, which belongs to the AraC/XylS family of regulators (174). YesS contains a C-terminal DNA binding domain, and it controls the expression of the pectin/rhamnogalacturonan utilization genes of B. subtilis (175). A ptsH deletion mutant exhibited significantly reduced YesS activity. Yeast two-hybrid and “three-hybrid” experiments revealed that YesS interacts with HPr and P∼His-HPr but not with P-Ser-HPr. In three-hybrid experiments, the additional synthesis of HprK/P prevented the interaction of HPr with YesS, whereas the synthesis of EI had no effect (175). The interaction with HPr and P∼His-HPr occurs within a central regulatory domain of YesS (amino acids 406 to 510), whose sequence seems to be unique to YesS. Although HPr and P∼His-HPr bind to YesS, only phosphorylated HPr stimulates its activity (Fig. 3). This was inferred from experiments with a ptsI mutant and a ptsH(H15A) mutant, neither of which allowed formation of P∼His-HPr and which exhibited significantly reduced YesS activity compared to that of the wild-type strain. In contrast, inactivation of the HprK/P-encoding hprK gene or of crh, which encodes the HPr paralogue Crh, did not lower YesS activity (175).

Interactions with P-Ser-HPr and P-Ser-Crh.

As mentioned above, HPr of firmicutes becomes phosphorylated not only at His-15 by PEP and EI but also at Ser-46 by ATP and HprK/P (21). This is also true for the HPr paralogue Crh present in bacilli, in which His-15 is replaced with a Gln, whereas Ser-46 and its surrounding region are well conserved compared to those in HPr of firmicutes (41). Homologues of Crh are found in bacilli, geobacilli, and oceanobacilli (1). HprK/P is a bifunctional enzyme also capable of dephosphorylating P-Ser-HPr and P-Ser-Crh by catalyzing a phosphorolysis reaction producing unphosphorylated protein and pyrophosphate (22). Numerous proteobacteria contain the HPr paralogue NPr (PtsO), in which Ser-46 and its surrounding region are well conserved. They also contain an HprK/P homologue, and in vitro phosphorylation of NPr has been established in the two alphaproteobacteria B. melitensis (19) and Gluconobacter oxydans (176) and the betaproteobacterium R. eutropha (110). A physiological function of P-Ser-NPr has so far not been established, but this posttranslational modification seems to be essential for growth of R. eutropha (110). In contrast, the role of P-Ser-HPr as a corepressor in CCR in firmicutes is well established and has been extensively reviewed (1, 177, 178). We therefore provide here only a brief description of this regulation mechanism.

The uptake of an efficiently metabolized carbon source by firmicutes leads to an increase of the FBP and a decrease of the phosphate concentration. These conditions favor the kinase activity of the bifunctional HprK/P (Fig. 1) and therefore cause an increase of the amount of P-Ser-HPr in the cells. P-Ser-HPr interacts with the catabolite control protein A (CcpA) (179) and allows this LacI-type repressor to bind to the catabolite response elements (cre), its target sites on the DNA (Fig. 3) (180). The cre site either can overlap the promoter region, in which case CcpA/P-Ser-HPr prevents binding of the RNA polymerase holoenzyme, or can be located downstream from the transcription initiation site with CcpA/P-Ser-HPr functioning as a roadblock. Mutants in which ccpA or hprK is deleted or Ser-46 of HPr is replaced with an alanine (181) are relieved from CcpA-mediated carbon catabolite repression. As already mentioned, carbon catabolite repression in firmicutes can be mediated by other mechanisms based on PRD-containing regulators, on inducer exclusion (1), or, in some bacteria, on glucose kinase (81). For certain catabolic genes or operons, the CcpA/P-Ser-HPr complex binds upstream from the promoter, which leads to catabolite activation instead of catabolite repression (182). In bacilli and a few other firmicutes, P-Ser-Crh can also bind to CcpA and function as a corepressor or coactivator for certain catabolic genes and operons during the utilization of an efficient carbon source (Fig. 3) (183, 184).

P-Ser-HPr of B. subtilis as well as its paralogue P-Ser-Crh has also been reported to interact with the glycolytic enzyme glyceraldehyde-3-P dehydrogenase (GapA) (185). In fact, both proteins interact with GapA in their unphosphorylated and seryl-phosphorylated forms, but only the seryl-phosphorylated forms of the two proteins were able to inhibit GapA (Fig. 3). P-Ser-HPr probably also plays an indirect role in S. pyogenes (84, 186) and Bacillus anthracis (187) virulence regulation, because in these pathogens CcpA controls the expression of several virulence genes, one being the streptolysin S-encoding gene of S. pyogenes (186).

There is strong evidence that in certain lactic acid bacteria P-Ser-HPr is also involved in inducer exclusion. Similar to what has been reported for Enterobacteriaceae, the uptake of the non-PTS sugar maltose by L. casei or Lactococcus lactis is immediately arrested when glucose or another efficiently metabolized carbon source is added (188–190). A similar observation was also made for the transport of ribose by L. lactis (190). These non-PTS sugars are taken up by ABC transport systems (191). Mutations in the hprK or ptsH gene preventing the formation of P-Ser-HPr also prevented the inhibitory effect of glucose on the uptake of the non-PTS sugars (188, 190). In contrast, mutations which caused an increase of the intracellular amount of P-Ser-HPr permanently inhibited the uptake of the non-PTS sugar (192). These results strongly suggest that similar to inducer exclusion in Enterobacteriaceae, where EIIA^Glc interacts with several non-PTS permeases, inducer exclusion in firmicutes is mediated by the interaction of P-Ser-HPr with a component of the ABC transporters (Fig. 3); unfortunately, this protein has not yet been identified.

Phosphorylation of HPr was discovered in connection with another regulatory phenomenon called inducer expulsion, which occurs in several firmicutes, including streptococci, lactococci, and lactobacilli (21, 193). Most of these bacteria transport the nonmetabolizable carbohydrate derivatives 2-deoxy-d-glucose (2-DG) and thiomethyl-β-d-galactopyranoside (TMG) via a PTS and therefore accumulate them as P derivatives in the cell. Addition of glucose or another efficiently metabolizable carbon source to cells preloaded with P-2DG or P-TMG led to rapid expulsion of the nonmetabolizable carbon source (194, 195). In fact, inducer expulsion was found to be a two-step process (193). In the first step, accumulated P-2DG or P-TMG is dephosphorylated by a P-sugar phosphatase, and in the second step, 2-DG or TMG is expelled. The conditions triggering inducer expulsion (metabolism of an efficiently utilized carbon source) also lead to the formation of P-Ser-HPr, and it was therefore suspected that P-Ser-HPr might be involved in inducer expulsion. Indeed, several P-sugar phosphatases of streptococci, lactococci, and enterococci were reported to be regulated by P-Ser-HPr (196–198). However, L. casei and L. lactis mutants unable to form P-Ser-HPr (because of ptsH1 mutation or hprK deletion) still exhibited inducer expulsion, establishing that P-Ser-HPr is not essential for this regulatory process. Interestingly, a second type of inducer expulsion was reported for some heterofermentative lactobacilli, such as Lactobacillus brevis. P-Ser-HPr was reported to be involved in this second type of inducer expulsion, which affects only non-PTS carbohydrates (199). The galactose/H⁺ symporter of L. brevis catalyzes not only galactose transport but also the uptake of the nonmetabolizable TMG, which in this organism is accumulated in the unphosphorylated form. The presence of glucose prevents the accumulation of TMG and causes the efflux (expulsion) of already-accumulated thio-sugar. Binding of P-Ser-HPr to the non-PTS permease was assumed to transform the symporter into a uniporter (facilitator) by uncoupling sugar transport from proton symport, thereby allowing the efflux of accumulated substrates. This concept was supported by the reconstitution of the HPr regulatory system (HPr, EI, and HprK/P) of L. brevis in a B. subtilis mutant devoid of HprK/P but still containing HPr. The strain also produced the L. brevis galactose permease GalP (200). When this strain was transformed with a plasmid encoding wild-type L. brevis HPr, it was not able to accumulate TMG when glucose was present in the medium. In contrast, glucose could not prevent TMG accumulation by a strain producing Ser-46-Ala mutant HPr. A strain transformed with a plasmid encoding L. brevis Ser-46-Asp mutant HPr was not able to accumulate TMG even when glucose was absent.

CONCLUSIONS AND PERSPECTIVES

The PTS was discovered 50 years ago in the laboratory of Saul Roseman at the University of Michigan, Ann Arbor, with the first article describing a PTS component, the HPr from E. coli, and its role in hexose phosphorylation being published in 1964 (20). Three years later, evidence for the presence of a PTS in a firmicute was obtained in the laboratory of Melvine Laurance Morse at the University of Colorado, Denver. They identified a PTS catalyzing the transport and phosphorylation of lactose in Staphylococcus aureus (201). In the following years, the proteins forming the glucose- and mannose-specific PTSs in E. coli (202, 203) and the lactose-specific PTS in S. aureus (204) were identified, and their role in transport and phosphorylation of the two hexoses and the disaccharide was established. Since then, a huge number of PTSs transporting a large variety of substrates, including hexoses, 6-deoxy-hexoses (14), amino sugars, N-acetyl-amino sugars, gluconic acids (205), pentitols (206, 207), ascorbate (208), and disaccharides, have been identified. We recently obtained evidence that the E. faecalis maltose-specific EIICBA^Mal (MalT) (15) also transports and phosphorylates the trisaccharide maltotriose and the tetrasaccharide maltotetraose (J. Deutscher, A. Hartke, J. Thompson, C. Magni, C. Henry, V. Blancato, G. Repizo, N. Sauvageot, A. Pikis, T. Kentache, and A. Mokhtari, unpublished results). A previous report had already suggested that the homologous maltose transporter MalT from Streptococcus mutans might take up maltotriose and maltotetraose (209).

PTS proteins are also involved in the phosphorylation of intracellular dihydroxyacetone (7), which enters the cells by facilitated diffusion or is formed from glycerol in a reaction catalyzed by glycerol dehydrogenase (64). Soon after its discovery, it was realized that the PTS not only transports and phosphorylates carbohydrates but also carries out regulatory functions related to carbon metabolism and sugar transport, such as catabolite repression and inducer exclusion. As described in the preceding sections, in the last 25 years PTS proteins were found also to regulate numerous cellular functions not or only indirectly related to carbon metabolism and transport, and this number increases steadily. PTS components control nitrogen and phosphate metabolism as well as potassium transport, antibiotic resistance, biofilm formation, and endotoxin production and also regulate the virulence of several pathogens. Given the high physiological impact of its regulatory functions, the PTS can no longer be considered merely a sugar transport and phosphorylation system that also plays some regulatory roles. The two functions rather seem to be of equal importance. In addition, in view of the tight connection between carbohydrate transport and metabolism and the regulatory functions of the PTS, it seems inappropriate to ask which of the two activities of the PTS is more important. Nevertheless, from an evolutionary point of view it would be interesting to understand how the multicomponent or multidomain PTS developed its different activities. Was the PTS originally a transport system that acquired regulatory functions, or was it an early regulatory system responding to environmental signals that later developed carbohydrate transport and phosphorylation activities, or, finally, did the transport and regulation systems evolve separately and became connected at a later stage in evolution in order to allow a correlation of transport and phosphorylation of carbohydrates with the different regulatory functions? The occurrence of an “incomplete PTS” composed of EI, HPr, and two EIIAs in numerous proteobacteria seems to favor the last hypothesis. In addition, many regulatory functions of the PTS seem to have developed late in evolution.

Concerning the PEP-requiring dihydroxyacetone phosphorylation system, we might be witnessing the creation of a novel PTS. At the present stage, transport and phosphorylation of dihydroxyacetone are two distinct steps, with the second being catalyzed by EI, HPr, EIIA, and DhaL and the first by facilitated diffusion. To make the system resemble a classical PTS and thus provide it with high transport efficiency, the dihydroxyacetone facilitator might integrate the dihydroxyacetone binding domain of DhaK. DhaL then needs to transform its ADP binding site into an EIIA-recognized Cys or His phosphorylation site in order to make it an EIIB-like protein. Given the slow pace of evolution, we will certainly not witness whether one day indeed a classical dihydroxyacetone-specific PTS with the proposed characteristics will be formed.

Whatever the course of evolution of the PTS was, regulatory functions probably played an important role in early stages of the PTS development. The fact that all phosphoryl group transfer steps, with the exception of the last one, i.e., the phosphorylation of the carbohydrate substrate, are reversible makes the PTS an efficient sensor and a fast signal transduction system. In addition, the PTS responds to the most significant signal for bacterial cell growth and proliferation, i.e., the availability of carbon sources in the environment and their efficient metabolism. Together these characteristics explain why the PTS controls so many cellular functions and developed so many different regulatory mechanisms, each probably optimally adapted to the needs of different bacterial species. By developing an additional signal entry via the metabolite-controlled HprK/P in firmicutes and numerous proteobacteria, the complexity of PTS-mediated regulation grew even larger. HprK/P probably evolved from an ancestral kinase and retained the capacity to utilize pyrophosphate as a phosphoryl donor for its kinase activity. Oligophosphates were probably already abundant when life developed on our planet. By reversing the phosphorylation reaction, HprK/P uses P_i for the dephosphorylation of P-Ser-HPr, thereby producing pyrophosphate (22) (Fig. 1). HprK/P probably gained the ability also to utilize ATP when this nucleoside triphosphate became available.

Despite extensive efforts in numerous laboratories, we understand only a few PTS-mediated regulation systems. Little is known about the role of HprK/P in proteobacteria, and the different functions of the PTS^Ntr in Enterobacteriaceae are barely understood (113). In addition, what is the role of the N-terminal domain of HprK/P, which is absent in HprK/Ps of alphaproteobacteria and which can be deleted from HprK/Ps from firmicutes without affecting their known activities (210)? In the crystal structure, this domain binds P_i to conserved arginine residues, but P_i does not seem to be a physiological effector. PTS domains are found in numerous proteins, and the list of Table 1 is far from being complete; there are numerous other proteins containing a domain of a PTS protein or a small fragment of it composed of about 50 amino acids. Only in a few cases has the regulatory function of these PTS domains been determined.

Phosphoproteome analyses revealed that numerous PTS components become phosphorylated at Ser or Thr residues, including the general PTS proteins EI and HPr as well as many sugar-specific PTS components (211–217). In firmicutes, HPr was found to be phosphorylated not only at Ser-46 but also at position 12, which can be Ser or Thr (212, 213). Other HPr phosphorylation sites were also found (213). Other frequent targets of phosphorylation are the EII components of the mannose PTS (211, 213, 215–217). With the exception of P-Ser46-HPr and P-Ser46-Crh, nothing is known about the enzymes catalyzing these protein phosphorylations and the physiological role of Ser/Thr-phosphorylated PTS components.

Highly interesting is the role of the PTS in certain pathogens, such as S. enterica serovar Typhimurium, V. cholerae, V. vulnificus, S. pyogenes, or L. monocytogenes. In L. monocytogenes, the utilization of an efficient carbon source, such as cellobiose, glucose, fructose, mannose, etc., inhibits the expression of the virulence genes, which are controlled by PrfA, a Crp-like transcription activator (218, 219). The detailed mechanism is not yet understood, but it is likely that a PTS component is involved (220, 221). Many other PTS-related regulatory systems exist for which the mechanism and/or function are not understood, and new systems are constantly being discovered, clearly showing that 50 years after the discovery of the PTS, its regulatory functions are far from being fully understood.