A Mutated PtsG, the Glucose Transporter, Allows Uptake ofd-Ribose*

Mutations arose from an Escherichia coli strain defective in the high (Rbs/ribose) and low (Als/allose and Xyl/xylose) affinity d-ribose transporters, which allow cells to grow on d-ribose. Genetic tagging and mapping of the mutations revealed that two loci in the E. coli linkage map are involved in creating a novel ribose transport mechanism. One mutation was found in ptsG, the glucose-specific transporter of phosphoenolpyruvate:carbohydrate phosphotransferase system and the other in mlc, recently reported to be involved in the regulation of ptsG. Five different mutations in ptsG were characterized, whose growth on d-ribose medium was about 80% that of the high affinity system (Rbs+). Two of them were found in the predicted periplasmic loops, whereas three others are in the transmembrane region. Ribose uptakes in the mutants, competitively inhibited by d-glucose, d-xylose, ord-allose, were much lower than that of the high affinity transporter but higher than those of the Als and Xyl systems. Further analyses of the mutants revealed that the rbsK (ribokinase) and rbsD (function unknown) genes are involved in the ribose transport through PtsG, indicating that the phosphorylation of ribose is not mediated by PtsG and that some unknown metabolic function mediated by RbsD is required. It was also found thatd-xylose, another sugar not involved in phosphorylation, was efficiently transported through the wild-type or mutant PtsG inmlc-negative background. The efficiencies of xylose and glucose transports are variable in the PtsG mutants, depending on their locations, either in the periplasm or in the membrane. In an extreme case of the transmembrane change (I283T), xylose transport is virtually abolished, indicating that the residue is directly involved in determining sugar specificity. We propose that there are at least two domains for substrate specificity in PtsG with slightly altered recognition properties.

D-Ribose is a pentose sugar commonly found as a component of nucleic acids. It can be utilized by many bacteria as a sole carbon source. In Escherichia coli, the sugar is transported via the high affinity RbsACB transporter, a member of the ABC 1type permeases (1,2), and phosphorylated by ribokinase (3).
Lopilato et al. (4) showed genetically that at least one low affinity ribose transport system exists which shares the kinase activity with the high affinity ribose transporter. Therefore, it has been believed that a specific low affinity transporter for ribose may exist, similar to xylE for D-xylose and araE for L-arabinose. On the other hand, it was recently reported that D-ribose can be transported through another high affinity transport system for D-xylose (5) and D-allose (6), structurally related to D-ribose, by a mutational derepression allowing an elevated level of transport components.
The bacterial phosphoenolpyruvate:carbohydrate phosphotransferase system (PTS) couples translocation and phosphorylation for the PTS sugars, e.g. glucose, mannose, and mannitol. Permeases of the PTS system differ in their substrate specificity but share the two general cytoplasmic PTS proteins, the enzymes I encoded by ptsI and HPr by ptsH, which sequentially transfer phosphoryl groups from phosphoenolpyruvate to their specific sugars. The glucose transporter consists of two subunits, IIA Glc encoded by the crr gene and IIBC Glc encoded by ptsG. IIA Glc is a hydrophilic protein phosphorylated in its His-90 residue (7) and is a central regulatory molecule in catabolite repression. IIBC Glc (PtsG) consists of the N-terminal domain with eight transmembrane peptides (8) and the hydrophilic C-terminal domain containing a phosphorylation site at Cys-421 (9). The sequence of PtsG protein is homologous to that of N-acetylglucosamine transporter, which has three domains corresponding to IIBC Glc and IIA Glc in a single polypeptide, and that of malX gene product, which can complement ptsG when expressed constitutively (10).
In addition to glucose, PtsG is involved in the transport of other sugars that are structurally related to glucose, i.e. methyl ␣-glucoside, 2-deoxyglucose, glucosamine, mannose, L-sorbose, and 5-thioglucose. Even though transport does not occur in the PTS system without phosphorylation, there are mutations in ptsG that allow facilitated diffusion of glucose without phosphorylation (11) or vice versa (12). Furthermore, mutation that alters the substrate specificity to allow transport of mannitol, another PTS sugar, was mapped in one of the transmembrane helices of PtsG (13). However, nothing is known about the coupling between phosphorylation and translocation or about the structure determining substrate specificity. The Mlc protein was at first reported as a factor affecting glucose transport, causing an enlarged colony size when overproduced because of a decrease in acetate production as a result of a reduced glucose utilization (14). It was recently reported that it is involved in the regulation of several sugar operons, i.e. manXYZ (14), malT (15), and ptsG (16,17). Mlc is a 406-amino acid protein containing a putative helix-turn-helix motif at its C-terminal residue that seems to be required for DNA binding involved in a repression.
In searching for secondary D-ribose transporters different from the high affinity system, we discovered that two mutations were simultaneously required for the utilization of ribose, one at the ptsG gene, a glucose-specific PTS permease, and the other at mlc, a negative regulator of ptsG (16,17). The mutations in ptsG alter specificity of substrate to allow D-ribose transport, whereas the mlc mutation renders ptsG expression constitutive. As part of characterizing the mutant phenotype, we first examined the effect of the mlc mutation on ptsG expression. We further analyzed the ptsG mutations allowing ribose transport in terms of changes in their substrate specificity to glucose, xylose, and allose. Patterns of sugar specificities and locations of the mutations provide insight into an organization of the sugar-binding domains and their coupling with the phosphorylation domain of PtsG.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, and Phages-All strains used are E. coli K12 derivatives that are listed in Table I. MC4100 or ZSC112⌬G was used as a parent for most strains. Cells were grown in Luria-Bertani medium, tryptone broth modified from the preparation according to Miller (18), and H1 minimal medium (19) containing 0.05% ribose, 0.2% glucose, or 0.2% xylose. Antibiotics were added to the growth medium when needed (ampicillin, 100 g/ml; kanamycin, 25 g/ml; chloramphenicol, 30 g/ml; tetracycline, 15 g/ml), and for minimal medium, they were supplied at 50%. The transposon hopping phage, ::TnphoA132, was obtained from B. L. Wanner (20).
DNA Manipulation-The recombinant DNA techniques were performed as described by Sambrook et al. (21). A 2.1-kilobase DNA fragment of the chromosome containing the mlc gene was amplified by polymerase chain reaction. The primer complementary to the 5Ј end of mlc created an XbaI restriction site at 50 bp upstream of the mlc start codon. The 3Ј end region was generated with PvuI, which cuts 600 bp downstream of the mlc stop codon. The resulting 1.9-kilobase DNA fragment was inserted into the pUC19 vector previously cleaved with XbaI and SmaI. The nucleotide sequence was confirmed by DNA sequencing. The clones were named pOH106 (mlc ϩ ) and pOH108 (mlc-101).
Genetic Procedures-The P1vir phage was used for transduction experiments (18). To characterize the ptsG mutation, the TnphoA132 transposon was used to obtain an insertion linked to the mutation. Random insertion of the transposon was generated with ::TnphoA132 for the MC4100 strain from which P1 lysate was prepared to obtain a pool of 20,000 independent insertions. Using the P1 lysate, transduction was carried out for CP1046 showing enhanced growth on ribose, and small colonies that resulted from an exchange of the mutated gene with the wild type were isolated on 0.05% ribose minimal plate containing tetracycline. Auxotrophic mutations were excluded by examining growth on 0.2% glucose minimal medium. The candidate insertions were tested for their cotransduction frequencies (Tet r ) with the ribose growth phenotype to measure their linkages. For localization of the insertion sites, the transposons were transferred to the polA ts strain (22). After appropriate recombination, the flanking regions of the insertions were characterized by DNA sequencing and through a data base search with BLAST. All sequencing reactions were carried out with the DNA sequenase version 2.0 (U. S. Biochemical Corp.) and ABI PRISM TM 377 DNA sequencer (Perkin-Elmer).
Construction of the ptsG-lacZ Transcriptional Fusion-A 1074-bp fragment with KpnI ends, comprising part of the ptsG structural gene and the 455-bp upstream sequence, was obtained as a polymerase chain reaction product whose sequence was confirmed by DNA sequencing. The fragment containing the promoter region of ptsG was cloned into pUC19 using KpnI, obtained by EcoRI and BamHI digestions, and Residues in open circles are mutations allowing facilitated diffusion without phosphorylation (11). Residues in open boxes are mutations that result in poor translocation with phosphorylation (12). The residue in the shaded box has a change with altered substrate specificity toward D-mannitol (13). cloned into EcoRI and BamHI sites of pRS415, a lacZ operon fusion plasmid (23), yielding pOH115. The promoter region was transferred to RS45 and lysogenized into an appropriate host. A single-copy prophage was confirmed by polymerase chain reaction with three primers recognizing the attB, attP, and int genes (24). Cells were grown to an optical density of 0.4 -0.8 at A 600 in H1 minimal medium (19) containing 0.4% glycerol and 0.2% glucose. ␤-Galactosidase activity was measured according to the method of Miller (18). All data were averaged from at least three independent experiments. DNA Mobility Shift Assay-CP1035/pOH106 (mlc ϩ ) and CP1036/ pOH108 (mlc-101) were grown to late log phase in Luria-Bertani medium containing 100 g/ml ampicillin. Cells were harvested by centrifugation (5,000 ϫ g) for 10 min, resuspended in 0.05 volume of 10 mM Tris-HCl (pH 8.0), 10% (v/v) glycerol, and 1 mM EDTA and sonicated in ice. The mixture was then centrifuged to remove insoluble fraction. Protein concentration in the soluble fraction was measured using the bicinchoninic acid protein assay reagent (Pierce). The supernatant was kept frozen at Ϫ20°C. The ptsG promoter region (568 bp from the KpnI to PvuII sites) was inserted into pUC19 with KpnI and HincII. The target DNA fragment containing the ptsG promoter was isolated by EcoRI and HindIII digestions, which were dephosphorylated with calf intestinal phosphatase and labeled with [␥-32 P]ATP using T4 polynucleotide kinase. The assay was carried out in a 20-l solution of 20 mM Tris-HCl (pH 8.0), 50 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, and 10% glycerol with crude extract (0.5-5 g of protein) mixed with about 0.3 ng of the radiolabeled ptsG promoter DNA and 1 g of sonicated salmon sperm DNA as a competitor. The reaction mixtures were incubated for 20 min at 25°C and subjected to electrophoresis on 5% polyacrylamide gel (60:1) with 0.5ϫ TBE buffer (45 mM Tris borate, 1 mM EDTA) at room temperature. The gels were exposed to x-ray film (Eastman Kodak Co.) at Ϫ70°C.
Uptake Assay-Cultures grown to late log phase were harvested and washed three times with equal volumes of KEP buffer (10 mM KH 2 PO 4 (pH 7.0), 0.1 mM EDTA) at 4°C. Cell density was then adjusted to 2.

Screening of Mutations Enhancing Ribose Growth-Even
with a defect in the high affinity ribose transport, the ribokinase-proficient E. coli K12 strain can use ribose as a sole carbon source, albeit less efficiently (4). Previous studies demonstrated that this growth deficiency could be partially recovered by a regulatory mutation in the allose (6) or the xylose (5) operons, both of which enhance expression of the low affinity transporter for ribose. When the high affinity transport system as well as the low affinity transporter genes (⌬als(R-K) and xylG::TnphoAЈ-2) were inactivated in OW1, an E. coli K12 strain, a mutation-enhancing ribose growth still appeared spontaneously at a frequency of 10 Ϫ8 -10 Ϫ9 . This indicates that there are some other loci involved in ribose transport.
To characterize those loci, transposon insertions with TnphoA132 were generated to isolate a tag associated with the mutations. About 20,000 independent insertions in MC4100 were pooled and transduced into the OW1 mutant showing enhanced growth on ribose. Analysis of the mutational locations linked to insertions revealed that two chromosomal loci, one at 36 min and the other at 25 min, are involved in the ribose growth. The high frequency (10 Ϫ8 -10 Ϫ9 ) of the original mutation, which lowers the possibility of a double mutation, led us to suspect that one of the mutations might already be present in the parental strain. Indeed, the mutation at 36 min was resident in the OW1 strain, although it was not found in other E. coli strains such as MC4100 and W3110.
Further characterization of the mutations revealed that the mutation at 25 min was found in ptsG, encoding the glucose transporter of the PTS system. A total of seven independent mutants was sequenced, in which five different mutational changes were observed: one for F37Y, G176D, and G281D and two for I283T and L289Q. Repeated occurrence of the mutations indicates an apparent saturation of the changes. F37Y and G176D were found in the predicted periplasmic loops, whereas G281D, I283T, and L289Q are in the transmembrane region (Fig. 1). The mutation at 36 min was characterized to have a change in mlc that has been implicated in the regulation of ptsG (16,17). This allele in OW1 was named mlc-101 and was sequenced to have C instead of T at the first base of Gln-369, causing an introduction of the UAG stop codon.
Effect of the mlc-101 Mutation on ptsG Expression-To test whether ribose transport through the mutated PtsGs results from an altered level of the transporter, as reported in the Als and Xyl systems (5, 6), we examined a transcriptional effect of Mlc-101 on ptsG expression by using DNA mobility shift assay and the ptsG-lacZ fusion. The 568-bp KpnI/PvuII fragment containing the ptsG promoter region was obtained from pOH116 and mixed with crude extract prepared from CP1035/ pOH106 (mlc ϩ ) and CP1036/pOH108 (mlc-101). A band shift was observed only with the wild-type Mlc but not with the truncated Mlc-101 (Fig. 2). This observation is consistent with the ptsG-lacZ fusion analysis and uptake assay of D-[ 14 C]glucose in which the ptsG transcription was derepressed at about 7-fold, whereas the uptake rates were increased about 33-fold in mlc-101 background (data not shown). Because Mlc-101 lacks the C-terminal peptide containing a putative DNA binding motif, it seems likely that the protein loses its function as a repressor.
Effect of the ptsG Mutations on Ribose Transport-Growths of the various ptsG mutants were examined on minimal plate with 0.05% ribose, rates of which were comparable with that of the Rbs ϩ strain and found to be correlated with the uptakes of [ 14 C]ribose (Table II). Among them, I283T, located in the transmembrane region, exhibited the highest growth and uptake rates. Although the uptake rates through the mutated PtsG were far less than that of the Rbs transport system, they were considerably better than the transport through the Als or Xyl systems, low affinity transporters for D-ribose (5,6).
To investigate the effect of PtsG mutations on glucose transport, growth and uptake rates for glucose were measured. Growths showed little differences from that of wild-type PtsG (data not shown), whereas the uptake rates were reduced (Ta- ble II). Because the tested strain contained the manZ and glk mutations, the coupling between transport and phosphorylation must occur in the mutant proteins (26). In general, the three transmembrane mutations reduce the specificity toward glucose more than the two periplasmic mutations, which is inversely correlated with the affinities to ribose. The I283T mutation with the highest ribose specificity reduced the uptake of glucose at about 50% of the wild-type PtsG, suggesting that at least the I283 residue, and perhaps other mutated residues, plays a key role in determining the substrate specificity.
Unlike glucose, ribose transport through PtsG occurs without phosphorylation. Thus, it seems possible that PTS and non-PTS sugars might use different mechanisms for their transports. However, the ribose transport through PtsG is inhibited by the presence of glucose, xylose, and allose as shown in Table II. The glucose effect is fairly strong, whereas D-xylose exerts a weaker inhibition. Unexpectedly, D-allose shows some inhibitory effects on ribose transport although the sugar was not transported by the mutated PtsG (data not shown). The apparent diversity of the inhibition pattern may reflect a multiplicity of substrate binding specificity of the transporter.
RbsD and Ribokinase Are Indispensable for D-Ribose Transport through PtsG-To test whether D-ribose transport through PtsG is coupled to phosphorylation, the requirement of ribokinase was assessed for F37Y. When the whole rbs operon was deleted (⌬rbsD-R), the mutant no longer maintained its enhanced growth phenotype on ribose (Table III), indicating that some rbs component(s) is required. An addition of rbsK on the pYP60 plasmid, however, did not entirely restore the enhanced growth phenotype. After testing other components of the rbs operon, we discovered that the maximal level of enhanced growth and transport was observed when both rbsD and rbsK were expressed. The rbsD gene is a part of the rbs operon, which codes for a protein with unknown function. The results indicate that the transported ribose is phosphorylated not by the PTS system but by ribokinase and that RbsD somehow plays a critical role in the PtsG-mediated ribose transport. The requirement of rbsD was also demonstrated in other ptsG mutants and in other low affinity transporters, such as the Xyl and Als systems (data not shown).
D-Xylose Can Be Transported through PtsG in the mlc Ϫ Background-Because it was demonstrated that ribose transport can be mediated by the Als and Xyl systems, we tested whether D-xylose is recognized by the mutated PtsG. The ribose uptake through these transporters appears to be based on the structural similarities of the sugars to D-ribose as shown in Fig. 3. Interestingly, it was found that xylose can be transported even through the wild-type PtsG only when there is a mutation in mlc, indicating that the xylose transport through PtsG is dosedependent in terms of the level of transporter (Table IV). The experiment was done in the strain with xylF::TnphoAЈ-1, an insertion in the gene coding for the xylose-binding protein that inactivates the high affinity transporter for xylose. The OW1 strain that has xylA (xylose isomerase) and mlc mutations did not grow on xylose minimal plate in the absence of xylFGH (data not shown), indicating that as in ribose transport, xylose is phosphorylated not by the PTS system but by xylulose kinase encoded by xylB. The presence of the ribose-specific mutations tends to reduce the xylose uptake rate in both the periplasmic and transmembrane mutants. The impairments of glucose uptake (Table II) are more notable in the membrane mutants than in the periplasmic ones. The degree of reduction, I283T Ͼ G281D Ͼ L289Q for both xylose and glucose, appears to be correlated with the specificity changes to ribose. DISCUSSION We report here that in addition to Rbs, Als, and Xyl transporters, D-ribose can be transported through PtsG when mu- b Measured on H1 minimal plates with 0.05% of D-ribose. Symbols represent relative levels of growth.
c Cells were grown in 0.05% ribose minimal media at 37°C, and A 600 was measured.
d Midlog phase culture grown in 0.05% H1 ribose minimal medium was used as described under "Experimental Procedures." e Not determined.  (13) a Strains for the ribose experiments, except wild type (CP1042), have the mlc-101 and ⌬rbsB mutations (CP1041 derivatives). For glucose uptake assay, strains (mlc ϩ ) with the manZ and glk mutations (ZSC112⌬G derivatives) were used.
b Growth on 0.05% ribose minimal plate was assessed after incubation for 2 days at 37°C. c Generation time was measured in H1 minimal medium containing 0.05% ribose by determining A 600 . d Cells were grown at 37°C in H1 minimal medium containing 0.05% ribose. Uptake assay was performed with 20 M D-[ 14 C]ribose as described under "Experimental Procedures." The transport rate was determined using [ 14 C]ribose and expressed in pmol/min/10 7 cells.
e Calculated as (ribose uptake rate with sugar competitor/ribose uptake rate) ϫ 100%. f The transport activities rely only on the presence of PtsG because of an introduction of the manZ and glk mutations. Transport rate was determined with D-[U-14 C]glucose and expressed in pmol/min/10 7 cells. g ND, not determined. tated specifically and derepressed by an inactivation of mlc, encoding a negative regulator of ptsG (16,17). Moreover, we found that D-xylose, structurally analogous to D-glucose and D-ribose, is transported through PtsG in the absence of Mlc. D-Ribose exists primarily as a pyranose form in solution as shown in Fig. 3. It is the one interacting with the periplasmic binding protein (27). This form shows similarity to D-allose, D-glucose, and D-xylose, explaining the cross-specificity between these sugars in sharing the transporters. Besides glucose, PtsG is involved in the transport of other sugars that are structurally related to glucose such as mannose, 2-deoxyglucose, glucosamine, ␣-methylglucoside, L-sorbose, and 5-thioglucose (Fig. 3). Among these, the first three sugars differ from glucose at their C-2 positions, indicating flexibility in recognizing the configuration around C-2. The same is true for ␣-methylglucoside at the C-1 position and 5-thioglucose at C-5. D-Mannitol (1-deoxymannose) was reported to be taken up through PtsG when Gly-320 is mutated to valine and thus modified to recognize the change of D-mannose at C-1 (13). Despite the differences in structures, the glucose analogs have some common features; they can be transported and phosphorylated by PtsG, which is not the case in D-ribose and D-xylose, which require their own kinases for recognizing a furanose form of sugar. The lack of phosphorylation coupling in transport of these sugars might lie in the fact that the hydroxyl group at C-5, normally exposed in the furanose form, is hidden in the transported sugar. The necessity of mlc mutation for the xylose transport, not found in the transport of other phosphorylated glucose analogs, may indicate that an uptake of nonphosphorylated sugar through PtsG requires a higher level of gene expression.
In addition to the C-6 hydroxyl group accepting phosphorylation, D-ribose differs from D-glucose in the hydroxyl group at C-3. Unlike D-ribose, the pyranose form of D-xylose has exactly the same configuration as D-glucose at all carbon positions except for C-6, a site for phosphorylation. Therefore, it seems likely that the configuration at C-3 is a major specificity determinant, more so than the ones at C-1, C-2, or C-6, which is consistent with the fact that an additional mutation is required for the transport of D-ribose and not for D-xylose. The tendency of the ribose-specific mutations toward losing specificities to xylose and glucose is exemplified in the extreme case of I283T. A slight deviation from this general trend is illustrated in the mutations localized in the periplasmic domain in which the glucose specificities were not substantially reduced. This may reflect the differences between the mutations in two locations, which might form two different pores with altered specificities. It is conceivable that a unique feature in the periplasmic pore lies in its discrimination of the C-6 residue. In fact, even though both glucose and xylose are transported, the xylose uptake rate of ptsG ϩ was about half that of glucose, even under a derepressed condition (Table II), which may indicate that the phosphorylation of sugar affects transport, perhaps through an intramolecular signaling. In other words, there might be a coupling between transport and phosphorylation which is likely to be manifested by a structural change in PtsG during sugar phosphorylation.
The possibility of two pores in PtsG could also be discussed in terms of its structural organization. Unlike the ABC transporter, the PTS system has no periplasmic sugar-binding protein, which may explain the fact that PTS has broader substrate specificity than the ABC transporters. It seems likely that in PtsG, the periplasmic region as well as the transmembrane region confers substrate specificities, and the absence of a binding protein may be functionally compensated by the presence of the periplasmic specificity domain, which might be explained by a model proposed here as the "two-sieve mechanism." A similar mechanism was reported to occur in the potassium channel, in which both the narrow selectivity filter recognizing K ϩ located on the outer surface and the wider, hydrophobic inner pore structure stabilizing a cation exist (28).
It is of interest to note that the D-mannitol-specific mutation of G320V was found in helix VIII (13), and three other mutations involved in facilitated diffusion, including ours, were found in helices VI and VII (11). These observations suggest that these helices form a channel for sugar specificity and transport. It was thought that the hydrophobic pocket plays an important role in sliding of sugar during translocation through the channel of PtsG. Also in the potassium channel, the lining of hydrophobic amino acids in the channel was suggested to  c Cells were grown at 37°C to midlog phase in H1 minimal medium containing 0.2% D-xylose. The uptake assay was performed with 0.2 M D-[ 14 C]xylose as described under "Experimental Procedures." The xylose transport rate was measured and expressed in pmol/min/10 7 cells. facilitate a final release of ligand by loosening an association with the substrate (28). In this regard, a change into hydrophilic amino acid as found in our mutations, especially ones located in the transmembrane region, may elicit a negative effect on phosphorylation-mediated glucose transport but a positive effect on ribose transport, perhaps occurring as facilitated diffusion.
During the study of the Rbs component involved in ribose transport though PtsG, we found that RbsD is required in addition to ribokinase. RbsD was originally proposed as a member of the membrane permease associating with RbsC (29). However, it apparently lacks a region predicted to be in the membrane. Furthermore, a mutation in RbsD does not abolish ribose transport, 2 suggesting that it is not directly involved in the permease function. There is no functional homolog found for RbsD based on its sequence similarity. The only clue about its function is that it enhances the utilization of ribose when the sugar is transported through a low affinity transporter. In other words, the role of RbsD is independent of specific types of transporters, implying that RbsD is involved in the step after membrane transport, such as acceleration of the ribose metabolism. The function of RbsD is currently under investigation.