INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

In Escherichia coli K-12, D-glucose (Glc) is taken up and concomitantly phosphorylated either by the glucose-specific enzyme II (EII) transporter (II^Glc) or the mannose-specific EII transporter (II^Man) (genes manXYZ) of the phosphoenolpyruvate-dependent carbohydrate phosphotransferase system (PTS) (for reviews, see references 10 and 21). As for most other PTS carbohydrates, the phosphoryl groups are sequentially transferred from PEP through two common intermediates, enzyme I (EI; gene: ptsI) and the phosphohistidine carrier protein (HPr; gene: ptsH), to sugar-specific EII (IICB^Glc; see below) and to glucose (for a review see reference 41).

II^Glc consists of two subunits, IIA^Glc (crr [catabolite repression resistance]) and membrane-bound IICB^Glc (ptsG) (8). The crr gene is part of the ptsHI crr operon (46) separated from the ptsG gene, which maps at 25.0 min (4). IIA^Glc is a small hydrophilic protein which has, in addition to its transport function, a central regulatory role in carbon catabolite repression and inducer exclusion (for a review, see reference 22). The IICB^Glc subunit is composed of an amino-terminal, hydrophobic IIC^Glc domain, which largely determines substrate specificity, and a carboxy-terminal, hydrophilic IIB^Glc domain, which is phosphorylated at the Cys421 residue (32). The system normally recognizes glucose as well as methyl--D-glucoside (MG), 5-thio-D-glucoside, L-sorbose and, with a low affinity, 2-deoxyglucose (2DG) (for a review, see reference 40).

Only little attention has been paid to the regulation of ptsG expression, although IICB^Glc has an outstanding regulatory function in establishing glucose as a favored carbon source. Moreover, for both E. coli (45) and Salmonella enterica serovar Typhimurium (55), it was demonstrated that the activity of IICB^Glc is the rate-limiting step in glucose utilization. Both ptsG expression (19, 37, 43) and manXYZ expression (36) are positively regulated by the cyclic AMP (cAMP)-cAMP receptor protein (CrpA) complex and negatively controlled by the DgsA (Mlc) protein. The dgsA locus (deoxyglucose sensitive) at 35.9 min on the E. coli chromosome was discovered as a suppressor mutation that enables ptsG-negative mutants to grow anaerobically on glucose via a constitutively expressed II^Man system and enhanced sensitivity to 2DG, a major substrate of this transport system (44). The DgsA protein was rediscovered recently and renamed Mlc (making large colonies) (16). Plumbridge (36) demonstrated that dgsA and mlc are the same gene; for priority reasons and according to Berlyn (4), we call this gene dgsA. The DgsA protein represses its own synthesis as well as the expression of the ptsHI crr operon (18, 38) and the mal regulon (7). It may represent a novel global repressor and may counteract the global regulator cAMP-CrpA to ensure the expression of those genes, which are linked to glucose metabolism (19, 38, 39, 44). The inducer for DgsA, however, has not been identified.

A different type of ptsG regulatory mutation, called umgC (uptake of MG control; umg is the former name of ptsG), was described by Jones-Mortimer and Kornberg (17). This mutation, which was claimed to map close to but not in ptsG, enabled E. coli cells with inactive II^Man to grow on mannose (Man) and glucosamine (GlcN or Glm). The authors concluded that mannose and glucosamine are not inducers of the glucose PTS, that the umgC mutation causes constitutive ptsG expression, and that umgC encodes a repressor, UmgC. In this paper, we describe the isolation and characterization of UmgC-like mutants which were selected as described by Jones-Mortimer and Kornberg (17). Moreover, we reinvestigated one of their UmgC mutants and showed that umgC mutations map within the ptsG allele and alter characteristically the ptsG induction pattern and IICB^Glc transporter activity.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, plasmids, and media. The E. coli K-12 strains and plasmids used in this study are listed in Table 1. LJ110 (Fnr⁺) was obtained from a cross of K-12(P1) with W3110 and by selection for growth on minimal agar plates supplemented with 0.2% glycerol and 10 mM KNO₃ under anaerobic conditions. Cells were routinely grown either in standard phosphate minimal medium (54) supplemented with 0.2% various carbon sources, in Lennox broth without glucose and calcium ions, or in 2xTY medium as described by Ausubel et al. (2). The utilization of various carbohydrates was screened on MacConkey agar plates (Difco) containing 1% of the indicated carbon source. Antibiotics were used at the following concentrations: tetracycline, 10 mg/liter; ampicillin, 50 mg/liter; chloramphenicol, 25 mg/liter; kanamycin, 25 mg/liter; and spectinomycin, 1,500 mg/liter for the multicopy plasmid system or 100 mg/liter in rich medium and 500 mg/liter in minimal medium for the single-copy system. Transductions were carried out with P1 vir essentially as described by Arber (1).

Construction of single-copy (ptsGop-lacZ) and (ptsHp-lacZ) fusions. For the construction of a single-copy (ptsGop-lacZ) fusion, the promoter-operator region was amplified from genomic DNA of LJ110 by PCR using the oligonucleotides ptsG4 (5'-AATCAACCTGCGATGGTTCC-3'; hybridizing to bp 337 to 318 upstream of the ptsG start codon) and ptsG3 (5'-AATACCTGCGATAGGCAGTACGGATACCGG-3', hybridizing to codons 19 to 28 for IICB^Glc). The product was treated with Klenow DNA polymerase to produce blunt ends and was inserted into the EcoRV restriction site of pTIM101. Vector pTIM101 essentially is a derivative of plasmid pIC-19H (26) in which the multiple cloning site was deleted. It carries a truncated Tn1721 transposon (52) which consists of the inverted repeat regions, the multiple cloning site of pBluescript II SK(+) (HaeII box) (2), and the so-called  element of plasmid pHP45 (42). The  element provides transcriptional and translational stop signals to prevent read-through from any potential upstream promoter and an spc gene for spectinomycin and streptomycin resistance. The orientation of the PCR insert was controlled by DNA sequencing. Downstream of the ptsG promoter region, the promoterless lacZ gene from plasmid pRU869 (53) was inserted into the HindIII/SalI restriction sites to produce plasmid pTIM103.

Isolation of plasmid DNA, restriction analysis, and cloning procedures. All manipulations with recombinant DNA were carried out using standard procedures as described previously (2). Plasmid DNA was prepared either by using standard phenol extraction protocols as described previously (48) or by using the JETstar DNA purification system (Genomed, Bad Oeynhausen, Germany). Restriction enzymes were purchased from New England Biolabs (Schwalbach, Germany). They were used according to the recommendations of the supplier. Oligonucleotides for sequencing or PCR were purchased from Interactiva (Ulm, Germany).

Mutation analysis. DNA amplification of the ptsG alleles was done as described by Saiki et al. (47) using Taq DNA polymerase from Roche Diagnostics, Mannheim, Germany, or Goldstar polymerase from Eurogentec, Seraing, Belgium. The forward PCR primer ptsG+ (5'-AACTGCAGGTGTTTAAGAATGCATTTGCT-3') for the amplification of ptsG contained an engineered PstI restriction site (underlined) upstream of an artificial GTG start codon (boldface; the original ATG was changed to GTG to lower the levels of expression of ptsG after subcloning into pSU19 [see below]). The reverse PCR primer ptsG (5'-CTTAAAGCTTAGTGGTTACGGATGTA-3') introduced an engineered HindIII restriction site (underlined) immediately downstream of the TAA stop codon (Fig. 1). The reaction profile consisted of 32 cycles of denaturing at 94°C for 1 s, annealing at 50°C for 1 s, and extension at 72°C for 45 s in an Air Thermo-Cycler 1605 from Idaho Technology Inc., Idaho Falls, Idaho. PCR products were directly purified using a Wizard PCR Preps DNA purification system (Promega Corp., Mannheim, Germany). All DNA sequencing reactions were performed by the dideoxy chain termination method (2) using an ALFexpress AutoRead or dATP labeling mix sequencing kit from Amersham-Pharmacia Biotech, Freiburg, Germany. The nucleotide sequences of both strands were determined after subcloning into vector pSU19 (27) using 5' cyanine fluorescent dye (Cy-5)-labeled universal and reverse primers or unlabeled internal ptsG sequencing oligonucleotides priming about every 250 bp within the gene. Computer analysis was done with DNASIS sequencing analysis software (Hitachi) and by using the BLAST programs and database services provided by the National Center for Biotechnology Information, Bethesda, Md.

View larger version (20K): [in this window] [in a new window]   FIG. 1.   Construction of ptsG expression plasmids. The ptsG alleles from various strains were amplified by PCR as described in Materials and Methods, introducing artificial PstI (upstream) and HindIII (downstream) restriction sites. The original ATG start codon of ptsG was changed to GTG (boldface). Fragments were subcloned into pTM30 cut with PstI and HindIII in frame with an artificial ATG start codon provided by the expression vector. All constructs provided three additional, amino-terminal amino acid residues. The expression vector also provided a strictly regulated tac promoter-operator (tacpo), a lacIq gene, and a ribosome binding site (rbs) in an optimal position with respect to the start codon.

Construction of defined carboxy-terminal deletions in IICB^Glc. Plasmid pSTJ30320 is a derivative of pTM110 which was digested with restriction enzymes StuI (bp 957 to 963 in ptsG) and HindIII (in the polylinker region downstream of ptsG) (Fig. 1). DNA was treated with Klenow DNA polymerase to generate blunt ends and was religated. The open reading frame encoded a protein which consisted of up to amino acid 320 of IICB^Glc and six additional, plasmid-encoded amino acids (AWYLTN). All other pSTJ30 deletion plasmids were generated using pTM110 as a template, the ptsG+ primer as an upstream primer, and an appropriate downstream primer in a PCR. The following reverse PCR primers were used for the different deletion plasmids: for pSTJ30396, ptsG21 (5'-CCTGAAGCTTTTGCATCTTCAGTCG-3'); for pSTJ30436, ptsG22 (5'-TGATAAGCTTTAGACACATCAGCAA-3'); and for pSTJ30459, ptsG23 (5'-GAAAAGCTTCTGAACACCAGAACC-3'). All primers created an artificial HindIII restriction site (underlined). The last number in each plasmid name indicates the last original PtsG amino acid encoded by the particular construct. PCR fragments were treated with HindIII and PstI and cloned into pTM30. All plasmid constructs were confirmed by DNA sequencing. The addition of at least 50 µM isopropyl--D-thiogalactopyranoside (IPTG) to cells carrying these deletion plasmids led to complete growth inhibition, indicating that truncated proteins were expressed from these plasmids and, like wild-type PtsG, were lethal for the cells (data not shown).

Transport and enzyme assays. Transport of -D-[methyl-¹⁴C]glucopyranoside (final concentration, 25 µM) was measured in exponentially grown cells as described previously (49). Samples were taken after 10, 20, and 30 s. Transport activities were calculated from the initial uptake rates. The -galactosidase assay was performed as described by Pardee and Prestige (34). All enzyme activities are in nanomoles per milligram of protein per minute.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Isolation and characterization of UmgC mutants of E. coli. Our isogenic E. coli derivatives which lack the II^Man transporter due to a mutation in the manA locus (genes manXYZ) (9) are unable to utilize D-glucosamine (GlcN or Glm) or D-mannose (Man) as a sole carbon source (Table 2). Cells of the ManA mutants LJ130 and LJ132 had to be incubated for 3 days at 37°C on minimal glucosamine plates before small colonies appeared. One hundred of these were isolated, purified, and tested for their ability to grow on various carbohydrates. The majority (80%) of the isolated colonies had a pleiotropic phenotype, i.e., they became Glm⁺ Man⁺ and at the same time became sensitive to D-arabinitol (Atl^s) and ribitol (Rtl^s). Each of these four carbohydrates has been discussed to be a minor gratuitous substrate for IICB^Glc, and transport occurs only when the transporter is expressed constitutively (reviewed in references 39 and 40). No degradation pathways for Atl-5-phosphate and Rtl-5-phosphate are present in E. coli K-12, thus explaining the sensitivity (14). Two mutants, LZ1 and LJ132-3, derived from LJ130 and LJ132, respectively, were further characterized, compared to an authentic UmgC mutant (HK727), and used to map the new mutations (Table 2).

Induction of ptsG expression in various isogenic mutants. To test the ptsG expression levels, various uninduced and induced strains were tested for MG uptake (Table 3). IICB^Glc activity in the wild type (LJ130) was induced about threefold by glucose, while LJ138 exhibited the expected constitutive transport activity. LZ1 and LJ132-3 showed increased and decreased uninduced transport activities, respectively. Both strains could be induced by the addition of glucose but, interestingly and in contrast to LJ130, also by the addition of glucosamine, mannose, or ribitol. The induced UmgC mutant LZ1 always exhibited higher transport activity than LJ130, thus resembling the DgsA mutant LJ138. Introduction of the dgsA::Tn10kan mutation into LZ1 (producing strain LZ170) led to fully constitutive uptake activity which resembled the fully constitutive ptsG expression of LZ727 and that of other authentic UmgC strains, e.g., JM1110 (17). Interestingly, the induction levels in the wild type and the mutants LZ1 and LJ132-3 were always lower than those in the fully constitutive mutants LZ170 and LZ727.

View this table: [in this window] [in a new window]   TABLE 3.   Induction by the uptake of MG in isogenic strainsa

View larger version (22K): [in this window] [in a new window]   FIG. 2.   Kinetics of induction of (ptsGop-lacZ) in strains LZ110 (wild type), LZ100 (ptsG1), and LZ138 (dgsA::kan). Cells harboring a single copy of the (ptsGop-lacZ) translational fusion on F'8 were pregrown overnight in minimal medium with 0.2% glycerol and used to inoculate fresh medium with 0.2% glycerol. Glucose (0.2%) was added in the early exponential growth phase at 0 min. Samples were harvested and analyzed for -galactosidase activity (SpAc). Symbols: , uninduced LZ110; , induced LZ110; , uninduced LZ100; , induced LZ100; , uninduced LZ138; , induced LZ138.

View this table: [in this window] [in a new window]   TABLE 4.   Induction tested by (ptsGop-lacZ) expression in various isogenic mutantsa

Subcloning and sequencing of ptsG alleles from various IICB^Glc mutants. All carbohydrates (Table 4) that induce ptsG expression in LZ1 have been postulated to be substrates of either wild type (40, 41) or mutant (3) IICB^Glc. We speculated that a mutation in this strain and in the other mutants changing the induction and substrate specificity of IICB^Glc and not overexpression of the transporter per se was responsible for the UmgC phenotype. This hypothesis implied that there may be no distinct gene for a UmgC repressor protein, thus explaining our failure to detect a umgC gene in the vicinity of ptsG. To test this hypothesis, the ptsG alleles from all the mutants were amplified by a PCR. Additionally, strain JWL184-1, which was described as being semiconstitutive for ptsG expression (23), was included in this approach. It exhibited a strong Glc⁺ phenotype (55-min generation time) and weak Man⁺ and Glm Atl^r Rtl^r phenotypes. The introduction of defined ptsG or manXYZ mutations revealed that JWL184-1 carries a thus-far-unidentified manXYZ mutation and perhaps a umgC-like mutation in ptsG (Table 2).

View this table: [in this window] [in a new window]   TABLE 5.   Influence of various ptsG alleles on ptsG induction as measured in LZ150 by the F'8::Tn (ptsGop-lacZ) test systema

View larger version (15K): [in this window] [in a new window]   FIG. 3.   Construction of carboxy-terminal deletions of IICBGlc and testing of trans-activation of a (ptsGop-lacZ) fusion. The functional domains of IICBGlc are indicated as follows: the amino-terminal, hydrophobic IICGlc domain is shown as a stippled box; the carboxy-terminal, hydrophilic IIBGlc domain with the phosphorylated Cys421 residue is shown as a hatched box; and the linker with the conserved 382-KTPGRED-388 motif is shown as a thin line. Deletion plasmids are shown as thick lines, and the associated numbers refer to the first and the last original amino acids of the truncated IICBGlc protein. Plasmids pJCH and pJBH (kindly provided by B. Erni) encode six additional His residues. LZ150/F'8::Tn (ptsGop-lacZ) harboring various plasmids was grown in liquid minimal medium with 0.2% glycerol, 0.1% Casamino Acids, and 10 mg of ampicillin per liter. Glucose (0.2%) was added after one cell division cycle where indicated (+Glc). Cells were harvested during exponential growth. -Galactosidase activities are given as nanomoles of protein per milligram per minute. The mean values of at least two measurements are given.

Effects of plasmid-encoded and truncated IICB^Glc on the level of expression of a (ptsGop-lacZ) fusion in a (ptsG::cat) strain. Deletion of ptsG completely prevented the induction of a (ptsGop-lacZ) fusion by the addition of glucose, even in a Man⁺ strain, or any other tested carbohydrate (Table 4). We previously reported (T. Zeppenfeld, C. Larisch, J. W. Lengeler, and K. Jahreis, Abstr. Int. Meet. Fachgr. Biochem. GDCh and SF431, p. 68, 1999) that IIC^Glc expression in the presence of pJCH alone (6), which encodes the IIC^Glc domain and parts of the linker region (amino acids 1 to 386 plus six additional His residues), resulted in strong constitutive -galactosidase activity (Fig. 3) in LZ150/F'8::Tn (ptsGop-lacZ). The truncated protein was not capable of transporting glucose; i.e., there was no complementation for glucose uptake in the ptsG manXYZ double mutant LJ121 (data not shown). Plumbridge (38) suggested that the IIB^Glc domain of PtsG may be responsible for generating the inducing signal for ptsG expression. Therefore, we also tested plasmid pJBH (6), which encodes the IIB^Glc domain (amino acids 390 to 477 plus six additional His residues), for transactivation in this assay. Cells harboring pJBH exhibited only slightly, if any, increased -galactosidase activity from the (ptsGop-lacZ) fusion compared to the wild-type control. These results correspond to the results obtained by Plumbridge (38), who also observed only a very weak positive regulatory effect of overproduction of the IIB^Glc domain.

View this table: [in this window] [in a new window]   TABLE 6.   Induction of (ptsHp-lacZ) expression in LZ110 (ptsG+) and LZ100 (ptsG1) by different carbon sourcesa

Effects of mutated IICB^Glc from LZ1 on the level of expression of a (ptsHp-lacZ) fusion. Besides regulating ptsG expression, DgsA is also involved in the regulation of the pts operon (18, 38). To test whether the ptsG₁ allele has an effect on the pts operon, a single-copy (ptsHp-lacZ) operon fusion was constructed. In wild-type ptsG⁺ strain LZ110, ptsH expression was induced about 3.3-fold by the addition of glucose (Table 6), slightly by mannose and mannitol, but not at all by ribitol. In the ptsG₁ mutant LZ100, the induction pattern changed as described for the (ptsGop-lacZ) fusion; i.e., the basal expression level and the level of expression with glucose were increased, and the mutant was clearly inducible by the addition of mannose, ribitol, and mannitol (three- to fivefold). These changes, caused by alteration of the IICB^Glc transporter, are consistent with the hypothesis that the substrate binding and/or transport activities of IICB^Glc directly modulate DgsA binding activity for all DgsA-regulated operons (see Discussion).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

In this study, we were able to show the mechanism of previously described umuC mutations in E. coli K-12 (17). We provide evidence for the nonexistence of a postulated distinct umgC regulatory gene for ptsG expression and show that mutations causing the UmgC phenotype map within ptsG. From the analysis of existing and newly isolated UmgC mutants emerges a novel regulatory model for ptsG expression. It contains, besides classical regulatory proteins, new elements of a glucose sensory system which correspond to parts of the glucose PTS. Essential components of this new model have been described by Kimata et al. (19) and by Plumbridge (38).

Central regulatory elements for the ptsG promoter-operator are the repressor DgsA (also known as Mlc), the PTS-dependent phosphorylation state of the IICB^Glc complex, and the global cAMP-CrpA activator. As has been shown by direct in vitro binding assays, ptsG transcription is repressed by DgsA binding to the operator and activated by cAMP-CrpA (19, 37). In accordance with the new model, the expression of IICB^Glc cannot be induced above the basal level in a cyaA background, whereas it is constitutively high in a dgsA cyaA⁺ background (Table 4). No molecular inducer for DgsA could be found in gel retardation assays which included glucose and glucose-6-phosphate (18, 19, 37). The new model postulates that IICB^Glc is the glucose sensor responsible for the induction of ptsG and that IIC^Glc occurs in two conformations: one (IIC) with a low affinity and the other (IIC*) with a high affinity for DgsA. IIC* prevents DgsA from binding to the operator, either by competition for a limited amount of DgsA or by allosterically reducing the affinity of DgsA for the operator. In the absence of IICB^Glc (in a ptsG strain), DgsA remains permanently bound to its operator, resulting in complete repression of transcription. In the absence of the DgsA repressor (in a dgsA strain), transcription becomes constitutive, independent of the presence of IICB^Glc but still dependent on activation by cAMP-CrpA. We speculate that phosphorylated IIB^Glc either shifts the equilibrium toward the low-affinity IIC^Glc conformation or that phosphorylated IIB^Glc binds to the IIC^Glc domain and thereby excludes DgsA from binding. As predicted by the model, mutants with defects in EI (ptsI), HPr (ptsH), or IIA^Glc (crr) all show constitutive expression of ptsG, provided an intact or a mutated IICB^Glc complex is also present (Table 4). Thus, ptsG pts mutants are noninducible. Final support for the model comes from the observation that mutants expressing DgsA and the IIC^Glc domain in the absence of the IIB^Glc and/or IIA^Glc domain, i.e., the putative IIC* state, show constitutive ptsG expression (Table 4 and Fig. 3). The failure of the C-terminal deletion proteins of IICB^Glc encoded by the pSTJ30 deletion derivatives (Fig. 3) to generate (like pJCH) enhanced constitutive ptsG expression might be caused by a permanent "locked-off" conformation of the IIC^Glc domain or by reduced protein stability.

In contrast to the bgl and other operons controlled through a PTS-dependent antitermination system (reviewed in reference 51), no antiterminator protein is involved in ptsG control in E. coli, and there is no evidence thus far for PTS-dependent phosphorylation or dephosphorylation of either the repressor DgsA (18) or the global activator CrpA. This observation again corroborates the new model in which the PTS-dependent phosphorylation of IICB^Glc acts through alteration of the conformation of the inducer domain IIC^Glc rather than through phosphotransfer to a cognate regulator. Based on the new model, the UmgC and other ptsG mutations can now be understood in the following way.

(i) The fully constitutive allele ptsG₇₂₇ from the original HK727 mutant is a "locked-in" mutation which blocks IIC permanently in the IIC* state (Table 5). Moreover, all mutated ptsG alleles exhibit trans-dominance over the wild-type ptsG allele. The corresponding E387G mutation affects a linker motif postulated before as being essential for the communication between IIB^Glc and IIC^Glc (5, 6, 24). In E. coli, there is no efficient ATP-dependent glucosamine kinase which could generate the first obligatory intermediate, glucosamine-6-phosphate. Furthermore, glucosamine, normally transported and phosphorylated through II^Man, is a noninducing and poor substrate for IICB^Glc. Therefore, selecting for Glm⁺ derivatives in a ManA mutant selects for mutations allowing a high level of expression of IICB^Glc and efficient phosphorylation of glucosamine by the glucose PTS. Thus, the constitutive DgsA mutant (LJ138 in Table 2) remains unable to take up and phosphorylate this substrate efficiently, while all UmgC-like mutants accept this substrate fairly well. In contrast, other mutants are able to grow on D-ribose and D-xylose via the IICB^Glc transporter but in the absence of concomitant substrate phosphorylation (33). Such mutations always arise in combination with a dgsA mutation, indicating that substrates transported through IICB^Glc by facilitated diffusion do not induce ptsG due to the lack of dephosphorylation of IIB^Glc during the process.

(ii) Besides the fully constitutive type of UmgC mutations represented by ptsG₇₂₇, there are semiconstitutive and inducible UmgC mutations; i.e., they require the presence of a substrate which is taken up and phosphorylated by the IICB^Glc transporter for full induction. They range from efficiently accepting all four substrates and having a clearly increased (2-fold) basal induction level (S169F) to allowing only glucosamine (P238L; data not shown) or mannose (F195L) and not arabinitol and ribitol transport and phosphorylation. Similar UmgC mutations, but in different amino acid residues (G176D, A288V, G320S, and P384R), have been isolated by Notley-McRobb and Ferenci (31). It is tempting to speculate that all of these mutations lead to a relaxed substrate specificity of the IICB^Glc transporter. In accordance with this hypothesis, this group isolated another class of UmgC-like mutants (V12G, V12F, and G13C) with increased growth on glucosamine and enhanced growth on glucose and mannose (31).

It is more difficult to explain how mutations (boldface) in the hydrophilic linker motif (382-KTPGRED-388) cause changes in substrate specificity relative to mutations located in the membrane-bound IIC^Glc domain, which is supposed to carry central parts of the substrate binding and catalytic site. The function of this interdomain linker in transport was systematically investigated by alanine-scanning mutagenesis (20). Amino acid substitutions T383A and G385A caused a strong reduction in phosphotransfer activity. The linker perhaps plays an important role in the conformational changes of the protein, precisely coupling the interaction between the IIB^Glc and the IIC^Glc domains in the process of phosphotransfer to the substrate bound in IIC^Glc (5, 6, 24). Substantial changes in the conformation of a PTS transporter during the process of substrate binding and phosphorylation have been reported for IICBA^Mtl (28). Using isothermal titration calorimetry, these authors found that approximately 50 to 60 residues are involved in the binding and phosphorylation of the substrate mannitol and the interaction between IIC^Mtl and IIB^Mtl. This interaction seems to be necessary for phosphotransfer from IIB^Mtl to the IIC^Mtl-bound substrate and the release of the phosphorylated substrate into the cytoplasm. The fact that the E387G mutant exhibits altered substrate specificity, enhanced transport activity, and a locked-in conformation for the interaction with DgsA seems to indicate that substrate binding, dephosphorylation of IIB^Glc, and a change into the inducing conformation (IIC*) are related processes. The observation that the S169F mutation caused an increased level of basal expression also fits this hypothesis.

The unexpected finding that our laboratory strain JWL184-1 carries a mutation in ptsG indicates that other laboratory strains of E. coli K-12 also may carry unidentified ptsG mutations. Plumbridge (38), for example, reported that in her JM101 copy, a (ptsGop-lacZ) fusion was inducible by glucose, N-acetylglucosamine, mannitol, trehalose and, to a lesser extent, glucosamine and mannose. This finding is in contrast to our finding that the E. coli K-12 reference strain W3110 and its isogenic derivatives can be induced only by glucose. Characteristically, older transport studies had indicated D-glucose, MG, and 5-thio-D-glucoside as the only substrates for the IICB^Glc transporter (39).

An interesting question is how DgsA could be titrated by IICB^Glc if the dgsA gene itself is autoregulated. DgsA binding sites cloned on low-copy-number plasmids indeed do not titrate DgsA and therefore do not lead to constitutive expression of DgsA-controlled genes (unpublished results). Cloning of the same DgsA binding sites on a multicopy-number vector, however, led to titration of DgsA and to constitutive (ptsGop-lacZ) expression (38). This result indicates that dgsA expression is rather limiting and that titration of DgsA by IICB^Glc present at about 2,000 molecules per cell might be possible.

The results presented here extend and contribute to understanding of the function of IICB^Glc not only in glucose transport but also in glucose sensing and response. In this process, the transport of glucose triggers a dual-signal transduction pathway. One branch consists of the glucose-dependent modulation of the level of phosphorylation of IIA^Glc. Unphosphorylated IIA^Glc binds to and reversibly inhibits non-PTS transporters, e.g., for lactose, maltose, or glycerol (inducer exclusion); in its phosphorylated form, IIA^Glc activates adenylate cyclase to synthesize cAMP (reviewed in reference 22). In the second branch, the level of phosphorylation of IICB^Glc directly modulates the activity of the anticatabolite repressor DgsA. DgsA, in interaction with cAMP-CrpA, can be used by E. coli to precisely regulate carbon catabolite gene expression.