Function of the duplicated IIB domain and oligomeric structure of the fructose permease of Escherichia coli.

The fructose permease of Escherichia coli, the fructose-specific Enzyme II of the phosphoenolpyruvate-dependent phosphotransferase system (PTS), contains a duplicated IIB domain. The protein therefore consists of three distinct domains, B', B, and C (N-terminal to C-terminal), joined by flexible linkers and is thus designated FruB'BC. The N-terminal B' domain was removed using molecular genetic techniques, and the truncated Enzyme II (FruBC) was characterized relative to the wild-type enzyme both in vivo and in vitro. In vivo, FruBC exhibited depressed fermentation characteristics at low fructose concentrations. [14C]Fructose uptake measurements revealed reduced rates only when the permease was rate-limiting for transport. In vitro, FruBC exhibited a 10-fold lower affinity for its phosphoryl donating protein, the IIA-FPr diphosphoryl transfer protein (DTP), than was observed with the wild-type enzyme, and the maximal velocity of fructose phosphorylation was 7-fold depressed. Because the fructose-1-phosphate:[14C]fructose transphosphorylation reaction appeared normal, we conclude that the loss of the B' domain primarily affected phosphoryl transfer between the IIA and IIB domains of the permease. A mutant FruBC derivative with cysteine 112 replaced by serine (C112S FruBC) was inactive as a phosphoryl carrier and a sugar transport protein. Expression of the plasmid-encoded mutant protein inhibited the in vivo activity of the chromosomally encoded wild-type fructose permease, but it did not observably affect the activities of the mannitol or glucitol PTS permeases or of non-PTS sugar permeases. Further, the presence of the detergent extracted mutant protein inhibited the activity of the detergent solubilized wild-type or FruBC enzyme. In contrast, the wild-type FruB BC permease was apparently epistatic over the truncated FruBC permease in vivo. The experiments reported 1) show that the B' domain of the fructose permease functions to facilitate phosphoryl transfer between DTP and the permease, 2) establish the essentiality of cysteine 112 in the B domain of the permease, 3) provide evidence that a functional fructose permease consists of an oligomer in which both IIB domains must be active for the enzyme to catalyze normal rates of phosphoryl transfer and transport, 4) suggest that a single B' domain in the oligomeric Enzyme II is sufficient to allow high efficiency phosphoryl transfer between the IIA domain of DTP and the IIB domain of the permease, and 5) show that the B' domain is not important for oligomerization.

The permeases of the bacterial phosphotransferase system (PTS) 1 are multidomain Enzyme II complexes that mediate the detection, transport, and phosphorylation of their sugar substrates (Robillard and Lolkema, 1988;Lolkema and Robillard, 1992;Postma et al., 1993). They consist minimally of three distinct domains termed IIA, IIB, and IIC, which may be found fused or dissociated in various combinations and various orders with each other and with other PTS protein domains Reizer, 1992, 1994). They accept a phosphoryl group from HPr(his-P), which in turn is phosphorylated with phosphoenolpyruvate (PEP) as the ultimate phosphoryl donor in a reaction catalyzed by Enzyme I of the PTS. The phosphoryl transfer chain of the PTS can be generalized as shown in Scheme I, where S1 and S2 represent two different sugar substrates of the PTS. The upper scheme is representative of the major class of PTS permeases found in almost all eubacteria, which includes the glucose, fructose, mannitol, and glucitol permeases of Escherichia coli, whereas the lower scheme is representative of a minor class of PTS permeases, which includes the mannose permease of E. coli (Lolkema and Robillard, 1992;Meins et al., 1993;Buhr et al., 1994).
Of the PTS permeases, the fructose-1-phosphate-forming fructose permeases are both the most widespread in nature and the most complex. Many Gram-negative and Gram-positive bacteria utilize only fructose via the PTS Romano and Saier, 1992;Mitchell et al., 1993;Titgemeyer et al., 1994Titgemeyer et al., , 1995, and it has consequently been postulated that a fructosespecific permease may have been the primordial system that gave rise to the major class of PTS permeases (Saier et al., 1985;Saier, 1990a, 1990b).
Sequence analyses of fructose permeases in Gram-negative bacteria have revealed that the proteins that comprise these permeases have unique combinations of PTS protein domains linked together. For example, the fru operon in E. coli encodes a tridomain, diphosphoryl transfer protein (DTP) in which IIA fru is linked to a fructose-inducible HPr-like domain called FPr via a central domain of unknown function (Geerse et al., 1989;Wu et al., 1990;Reizer et al., 1994b). The membranal Enzyme II possesses three domains, two B domains and a C domain, with the domain order: BЈBC (Prior and Kornberg, 1988;Wu and Saier, 1990b). E. coli also possesses cryptic operons encoding fructose-like PTS permeases (Reizer et al., 1994a. Internally duplicated IIB domains are found in the fructose permeases of Rhodobacter capsulatus (Wu et al., 1990) and Xanthomonas campestris (de Crécy-Lagard et al., 1991) but not in other sequenced PTS permeases.
The N-terminal BЈ domains of the fructose permeases of E. coli, R. capsulatus, and X. campestris have diverged from the primordial IIB domains of these permeases much more than the central B domains, and they lack the conserved cysteyl residue, which on the basis of analogy with the nonhomologous mannitol and glucose permeases of E. coli are believed to provide the site of phosphorylation (Lolkema and Robillard, 1992). As a result of these observations, it is believed that the BЈ domains cannot participate in phosphoryl transfer (Wu et al., 1990).
In the two relevant cases that have been examined, the PTS Enzyme II complexes appear to occur in the membrane as oligomers. The mannitol permease of E. coli Boer et al., 1994) and the glucose permease of E. coli (Erni, 1986) are both thought to be dimeric. No information is available as to whether or not the fructose Enzyme II complex is also oligomeric. The minimal degree of sequence similarity observed between the fructose, mannitol, and glucose permeases renders the question of the fructose permease's oligomeric structure uncertain.
In this paper we use molecular genetic approaches to probe the structure/function relationships of the fructose permease (FruBЈBC) of E. coli. We first construct a FruBC protein lacking the BЈ domain and show that it is functional for fructose transport in vivo as well as fructose phosphorylation in vitro. This domain, however, is required for normal high affinity recognition of the DTP by the fructose-specific Enzyme II as well as for normal rates of phosphoryl transfer between the IIA and IIB domains of DTP and FruBЈBC, respectively. We further show that a FruBC protein with the putative active site cysteyl residue (Cys 112 ) replaced with serine (C112S FruBC) is completely inactive for fructose transport and that it inhibits the wild-type enzyme, both in vivo and in vitro. Because the mutant fructose permease did not observably inhibit the activities of other PTS and non-PTS permeases, we suggest that the fructose permease is present in the membrane as a functional oligomer in which the functionality of the transport protein complex requires that both B domains within the two monomers be active for phosphoryl transfer. Surprisingly, FruBЈBC appears to be positively dominant over FruBC in vivo, suggesting that a single BЈ domain in the oligomeric permease is sufficient for normal interaction with DTP. EXPERIMENTAL PROCEDURES Bacterial Strains, Media, and Chemicals-E. coli (lac-proAB)) was used as a primary recipient for the pROK1derived plasmids. E. coli strain HK1376 (fruA cysA thr his met srlA::Tn10 recA) was the final recipient for all of the pROK1-derived plasmids. HK1376 transformed with plasmid pROK1 without insert (Clontech, Palo Alto, CA) was used as the FruA-negative control strain.
MacConkey and LB solid media were prepared as described previously (Sambrook et al., 1989). A final concentration of 1% fructose was usually used for MacConkey fructose solid fermentation media except when otherwise noted. Other sugars, when present, were included at a concentration of 0.5% (mannitol, maltose, and arabinose) or 1% (melibiose and glucitol). Ampicillin was added to a final concentration of 100 g/ml. Minimal liquid medium 63 and complete LB medium were prepared as described previously (Saier et al., 1970;Sambrook et al., 1989).

Overproduction and Purification of Enteric Bacterial PTS Proteins-
The overproduction and purification of E. coli Enzyme I and HPr have been described . The DTP encoded within the fructose (fru) operon of Salmonella typhimurium (Geerse et al., 1989) was overproduced as follows: an NdeI restriction site was introduced in the initiation codon of the fruB gene encoding DTP by site-specific mutagenesis as described previously (Reizer et al., 1989). The fruB gene was digested with the NdeI and SalI restriction enzymes and was ligated into the NdeI-XhoI-digested pET19B plasmid (Novagen, Madison, WI). This procedure results in attachment to DTP of an N-terminal polyhistidyl sequence.
Overproduction of his-tagged DTP was accomplished following transfer of the resultant plasmid (pJRFruB) to E. coli strain BL21(DE3) and growth of this plasmid-bearing strain in LB medium (10 liters) in the presence of 0.15 mM isopropyl ␤-thiogalactoside (4 h at 37°C). Cells were harvested, washed three times by centrifugation, and resuspended in 20 mM Tris-HCl, pH 7.5, containing 0.1 mM phenylmethylsulfonyl fluoride. They were then ruptured by three passages through a French pressure cell at 10,000 psi. Cell debris was removed by centrifugation (10,000 ϫ g for 10 min at 4°C), and the clarified extract was loaded onto an immobilized Ni 2ϩ column (Novagen, Madison, WI; column height, 5 cm; column diameter, 2 cm). The column was washed with 20 mM Tris-HCl buffer, pH 7.9, containing 0.5 M NaCl and 60 mM imidazole. DTP was eluted with the same buffer with the imidazole concentration increased to 1 M. DTP was obtained with an estimated purity of 80 -90% based on SDS-polyacrylamide gel electrophoresis analyses. The enzyme was found to complement a fruB-negative mutant of E. coli in vivo (Chin et al., 1989) and proved to catalyze [ 14 C]fructose phosphorylation in the presence of Enzyme I and membranes isolated from fructose grown S. typhimurium strain LT-2 cells in in vitro assays. It was phosphorylated with [ 32 P]PEP and purified Enzyme I and was capable of transferring its phosphoryl moiety to homogeneous IIA glc , IIA ntr , and NPr (Powell et al., 1995).
The two primers used to amplify the gene encoding the truncated FruA permease with a cysteine to serine substitution at site 112 (designated C112S FruBC) were: left primer (50-mer, coding strand), 5Ј-GGCGAATTCAT(ATG)GGTCCGAAACGCGTAGTTGCGGTGACTG-CTTCC*CCG-3Ј (The asterisk indicates that the original TGC codon encoding cysteine 112 has been changed to a TCC codon encoding a serine residue); right primer, same as the right primer used for the full-length gene. In these two constructions an additional ATG codon was placed upstream of the first codon of the B domain (which determines glycine residue 102) in order to provide a translational initiation codon (underlined and parentheses).
Two-step Subcloning of the Amplified DNA Fragments into the pROK1 Expression Vector-The amplified double-stranded DNA fragments were first subcloned into the pCR cloning vector using the Invitrogen TA cloning kit (Invitrogen Corporation, San Diego, CA). The clones with the inserted polymerase chain reaction products were selected by restriction analysis of their plasmid DNAs using standard procedures (Sambrook et al., 1989). Plasmid DNAs from the selected pCR recombinant clones were prepared using Qiagen kits (Qiagen Inc., Chatsworth, CA). The double-stranded DNA fragments encompassing the respective fruBЈBC and fruBC genes were obtained by cutting the pCR recombinant plasmids simultaneously with EcoRI and HindIII restriction enzymes (Biolabs NEN, Beverly, MA). In the three constructs, the gene is present between the EcoRI (upstream of the initiation codon) and HindIII (downstream of the stop codon) sites. After purification of the bands on 1% agarose gels, the EcoRI-HindIII fragments were cloned into the EcoRI-HindIII site of the expression vector pROK1. The ligation mixtures were transformed into the E. coli recipient strain TG1, and the pROK1 recombinant plasmids were tested by restriction enzyme analysis. The pROK1 recombinant plasmid carrying the wild-type fruA gene (encoding the FruBЈBC wild-type permease) was designated pBЈBC. pROK1 recombinant plasmids carrying SCHEME I Function and Structure of Fructose Permease of E. coli truncated fruA genes (encoding FruBC and C112S FruBC mutant permeases) were designated pBC and pC112SBC, respectively.
Expression of the Recombinant Plasmids in HK1376 -The recombinant pROK1 plasmids were transferred by electroporation into the recipient fruA strain HK1376 for expression. These plasmids were again subjected to restriction analysis to verify their correct identity. The N-terminal sequences of FruBЈBC and FruBC were also checked by DNA sequence analysis of plasmids pBЈBC and pBC using the M13 reverse primer (Ϫ48) (Biolabs NEN), and the USB Sequenase version 2.0 DNA sequencing kit (Amersham Corp.) In Vivo Fructose Uptake Measurements-Bacteria were grown overnight at 37°C with agitation in minimal Medium 63 supplemented with 1 g/ml thiamine, 0.2% casamino acids, 0.2% fructose, and 100 g/ml ampicillin. The same medium (5 ml) was then inoculated with an aliquot of the overnight culture (initial A 600 ϭ 0.05). Cultures were grown at 37°C with agitation to an A 600 of 0.6. Bacteria were then washed twice with Medium 63 containing 40 g/ml chloramphenicol, resuspended in the same medium in aliquots of 0.5 ml at A 600 ϭ 0.2, and placed on ice. Cell suspensions were pre-equilibrated at room temperature for 15 min prior to the addition of [ 14 C]fructose (10 M; 20 Ci/mol). At t ϭ 30, 60, and 90 s, aliquots (150 l) of each bacterial suspension were applied to Millipore filters (pore size, 0.45 m), washed twice with minimal medium, dried, and submerged in Biosafe NAscintillation fluid (3 ml) before counting in a Beckman LS230 liquid scintillation counter.
In Vitro Fructose Phosphorylation Assays-Bacteria were grown overnight at 37°C with agitation in LB medium supplemented with various carbon sources and inducers as indicated. Cells were pelleted by centrifugation and washed three times. For 1-liter cultures, the washed pellets were resuspended in 20 ml of 20 mM Tris-HCl, pH 7.5, containing 1 mM dithiothreitol and 0.1 mM phenylmethylsulfonyl fluoride. The resuspended bacteria were disrupted by passage through a French pressure cell at 10,000 psi. Urea-butanol extracted membranes were prepared and dialyzed as described by .
PTS-dependent sugar phosphorylation assays were performed essentially as described previously (Reizer et al., 1989. The final volume was 50 l, the PEP concentration was 5 mM, and the concentration of [ 14 C]fructose (5 Ci/mol) was 10 M unless otherwise specified. When urea-butanol extracted membranes were used, purified components of the PTS (E. coli Enzyme I and DTP) were added as detailed in the figure legends. Transphosphorylation assays were performed essentially as described previously with rate-limiting amounts of urea-butanol washed membranes  and assay mixtures (final volume, 200 l) containing 10 mM fructose-1-phosphate, 10 M [ 14 C]fructose (5 Ci/mol), and other constituents as described previously .
Membrane proteins were solubilized with 0.5% deoxycholate by shaking at 20°C for 30 min in a buffer solution consisting of 20 mM Tris-HCl, pH 8.4, 0.2 M NaCl, and 1 mM dithiothreitol in a total volume of 25 ml. The suspension was then centrifuged at 100,000 ϫ g for 90 min at 4°C. The supernatant was finally dialyzed against 20 volumes of 20 mM Tris-HCl, pH 8.4, 1 mM dithiothreitol, and 0.5% Lubrol PX (TDL buffer). Dialysis was continued for 40 h with two changes of buffer. Fig. 1A presents a topological model of the E. coli fructose permease protein (FruBЈBC). This model is based on TOP-PRED (Sipos and von Heijne, 1993) analyses for the protein members of the fructose IIC family . The TOP-PRED program predicted nine transmembrane helical segments for the E. coli Enzyme IIC fru , for the demonstrably homologous IIC domains of the fructose PTS permeases of R. capsulatus and X. campestris, and for the IIC protein domains encoded within the frw and frv operons of E. coli (Reizer et al., 1994a. The linear domain structure of wild-type FruBЈBC is presented in Fig. 1B where the relative domain sizes are portrayed to scale. The two mutant forms of the fructose permease (FruBC and C112S FruBC) that were constructed for this study are diagramed in Fig. 1, C and D, respectively.

Topological Model of the E. coli Fructose Permease-
In Vivo Properties of FruBЈBC, FruBC, and C112S FruBC-As summarized in Table I, cells expressing FruBЈBC or FruBC were capable of efficient fructose fermentation if the fructose concentration was Ն0.5%. At a fructose concentration of 0.2%, however, FruBC bearing cells fermented fructose weakly, whereas cells possessing FruBЈBC gave a strong response. The C112S FruBC protein appeared nonfunctional at all fructose concentrations tested based on these fermentation criteria ( Table I). The same strains were tested for growth in liquid minimal salts medium containing 0.2% casamino acids as well as 0.2% fructose. The growth rates were the same for all three strains, showing that fructose was not growth inhibitory (data not shown).
Fructose uptake rates were measured with 10 M [ 14 C]fructose after growth under inducing ( Fig. 2A) or noninducing (Fig.  2B) conditions. After growth under fructose-induced conditions, where the fructose permease proteins were present in excess, there was no significant difference in the rate or extent of fructose uptake by cells bearing FruBЈBC or FruBC. By contrast, C112S FruBC was essentially inactive. Glycerol grown cells (uninduced for fru operon expression) took up [ 14 C]fructose much more slowly than did the fructose-induced cells, and the truncated FruBC permease was detectably less FIG. 1. A, proposed topological model of the wild-type E. coli fructose permease (FruBЈBC). Numbers indicate residues that serve as boundaries of the BЈ, B, and C domains in the E. coli protein deduced from sequence alignments of the different fructose and fructose-like Enzyme IIBЈBC protein sequences available (Reizer et al., 1994a(Reizer et al., , 1994b. 2 C112 corresponds to the putative phosphorylation site in the IIB domain. The open rectangles in the IIC domain correspond to the nine putative membrane spanning helices predicted using the TOP-PRED program (Sipos and von Heijne, 1993). B-D, schematic representation of the wild-type FruBЈBC permease (B) and its two mutant derivatives, FruBC (C) and C112S FruBC (D). Domain designations are provided above the linear depictions with shading as follows: IIBЈ, light shading; IIB, medium shading; IIC, dark shading. The residues noted below each linear depiction indicate the initiation methionine (M1), the boundaries of the various domains (G102 and T204), and the terminal aminoacyl residue (A563). active than the wild-type permease. These results show that loss of the BЈ domain has a perceptible effect on fructose transport and metabolism under the conditions employed only when the fructose permease is rate-limiting for uptake.
In Vitro Properties of FruBЈBC and FruBC- Fig. 3 shows the results of an in vitro experiment in which crude extracts from cells possessing the wild-type (FruBЈBC) and truncated (FruBC) proteins were examined for [ 14 C]fructose phosphorylation after growth of the cells in the presence of fructose. The wild-type FruBЈBC enzyme phosphorylated [ 14 C]fructose efficiently with endogenous DTP, and its activity was stimulated slightly by the addition of excess purified DTP. In contrast, the truncated FruBC enzyme exhibited low endogenous activity, but the activity of this preparation was greatly stimulated, essentially to the wild-type level, by the addition of excess DTP. The control strain (HK1376 (pROK1)) was completely inactive either with or without added DTP. These results show that the FruBC protein exhibits appreciable levels of activity only in the presence of excess DTP.
The kinetic properties of FruBC were compared with those of FruBЈBC when studied with excess purified Enzyme I and variable concentrations of purified DTP in the presence of a constant fructose concentration (Fig. 4A). The truncated protein lacking the BЈ domain exhibited a 7-fold depressed maximal phosphoryl transfer velocity, and it exhibited a 10-fold higher K m value for DTP (300 M versus 30 M). When the same two enzymes were examined as a function of fructose concentration (1-10 M) in the presence of constant amounts of DTP, straight lines were obtained on the double reciprocal plot (Fig.  4B). The calculated K m values for fructose (assuming a pingpong bi bi mechanism) (Segal et al., 1975) were 5.4 M and 1.6 M for FruBЈBC and FruBC, respectively.
The fructose-1-phosphate:[ 14 C]fructose transphosphorylation reaction  was also studied. For this experiment, urea-butanol extracted membrane preparations were used (see "Experimental Procedures"). The two enzymes (present in rate-limiting amounts) gave the same rate of transphosphorylation (42 pmol of fructose phosphorylated per min per milligram protein Ϯ 10%). Because transphosphorylation does not involve DTP and is a reflection of the affinities of the enzyme for fructose and fructose-1-phosphate , the results suggest that the primary defect resulting from the loss of the BЈ domain involves phosphoryl transfer from DTP to the B domain of the fructose permease.
Negative Dominant Phenotype of the C112S FruBC Permease over the Wild-type Permease and Positive Dominant Phenotype of the Wild-type Permease over the FruBC Permease in Vivo-Plasmids pBC and pC112SBC encoding the truncated permease lacking the BЈ domain and the truncated C112S mutant permease, respectively, were individually transferred to E. coli strain TG1, which encodes the wild-type fructose permease on the chromosome. Fermentation responses were recorded on MacConkey agar plates in the presence of varying concentrations of fructose either with or without 1 mM IPTG to induce synthesis of the plasmid-encoded permeases. The results are summarized in Table II. Strain TG1 exhibited a strong fermentation response regardless of the fructose concentration used, and the same was observed when pBC was expressed. However, when pC112SBC was present, the fermentation response was somewhat diminished at fructose concentrations of 0.2 and 0.5%. Inclusion of 1 mM IPTG in the agar medium, which enhanced expression of the plasmid-encoded mutant permease, strongly inhibited the fermentation response at all sugar concentrations (see Table II).
The three strains (Table II) were examined for fermentation

. Fructose uptake by intact cells grown under inducing and noninducing conditions. [ 14 C]Fructose uptake by cells grown in the presence of fructose (A) or glycerol (B)
. The experiments were conducted as described under "Experimental Procedures" with 10 M [ 14 C]fructose as the uptake substrate. FruBЈBC corresponds to the full-length permease, FruBC corresponds to the permease deleted for the IIBЈ domain, and C112S FruBC corresponds to the permease deleted for the IIBЈ domain and containing a C112S site directed mutation in the IIB domain. Control, the uptake measurements observed for strain HK1376 transformed with a mock plasmid (pROK1 without an insert). Note the different scales in A and B.
of two other PTS sugars (mannitol and glucitol) as well as three non-PTS sugars (L-arabinose, maltose, and melibiose) (see description in Table II). In no case did high level expression (IPTG-induced) of pC112SBC give rise to fermentation inhibition. Thus, the inhibitory effect of pC112SBC was specific to the fructose permease.
As shown in Fig. 5, inhibition of [ 14 C]fructose uptake was also observed following growth of the pC112SBC bearing strain in the presence of IPTG, although no inhibition was observed when the pBC bearing strain was grown under the same conditions. The transport results agree with the fermentation responses recorded in Table II, showing that C112S FruBC exhibits a negative dominant phenotype over the wild-type allele.
Negative Dominance of C112S FruBC over FruBЈBC and FruBC in Vitro-The wild-type fructose permease (FruBЈBC) from strain HK1376 (pBЈBC), the truncated permease (FruBC) from strain HK1376 (pBC), and the mutant fructose permease from strain HK1376(pC112SBC) were extracted with deoxycholate from isolated membranes and transferred to 0.5% lubrol as outlined under "Experimental Procedures" (Jacobson et al., 1983). The dialyzed extracts were then mixed as specified in the legend to Fig. 6. As can be seen from the data reproduced in Fig. 6, the presence of the C112S mutant fructose permease inhibited the activity of FruBC. The same was observed for the wild-type permease (data not shown). These results are in agreement with the negative dominant phenotype exhibited by C112S FruBC in vivo.

DISCUSSION
The fructose permease of E. coli is unique among PTS permeases in several respects: 1) It is the only known permease in E. coli to possess its own HPr-like protein, FPr (Saier et al., 1970); 2) it is the only one that has its IIA domain covalently linked to one of the general energy coupling proteins of the PTS (Geerse et al., 1989;Wu et al., 1990;Reizer et al., 1994aReizer et al., , 1994b; and 3) it is the only one that has an Enzyme II complex with an internally repeated IIB domain (Wu and Saier, 1990a). These unique features of the fructose PTS and other features discussed previously Saier et al., 1985) have led to the hypothesis that an ancient fructose permease evolved into the first PTS and that this primordial PTS functioned to initiate the glycolytic cycle.
Little is known regarding the mechanism of action of the fructose permease. To correct this deficiency, we first removed the IIBЈ domain of this permease and found that although the truncated protein could still transport and phosphorylate fructose, it exhibited a 10-fold decrease in affinity and a 7-fold decrease in the rate of phosphoryl acceptance from DTP. Because the fructose-1-phosphate:[ 14 C]fructose transphosphorylation reaction was not detectably altered by the loss of the BЈ domain, we conclude that the primary function of the BЈ domain is to control the interaction and reactivity of the FruBЈBC enzyme with DTP.
An apparent quantitative anomaly arose when comparing the activity of the wild-type and truncated mutant permeases (FruBЈBC versus FruBC) in vivo and in vitro. Although the truncated protein exhibited low affinity for DTP and a low rate of phosphoryl transfer with DTP in vitro (Fig. 4), fermentation responses and transport rates by cells bearing only this truncated permease were nearly normal after growth in fructosecontaining medium (Fig. 2 and Table I). It seemed likely that in vivo, growth on fructose results in the synthesis of a large excess of the fructose operon gene products and that DTP probably forms a tight association complex with the fructose permease (FruBЈBC or FruBC) in the membranes of growing cells. The former conclusion was supported by the sugar phosphorylation results obtained with unsupplemented crude extracts (Fig. 3), which revealed almost no difference between the two permeases in the presence of excess DTP. Under these conditions, Enzyme I is probably rate-limiting. 2 Evidence for tight binding PTS enzyme complexes in vivo has been published previously (Gachelin, 1969;Saier and Staley, 1977;Saier et al., 1982). Because duplicated BЈ domains are found in the R. capsulatus and X. campestris fructose permeases (Wu et al., 1990;de Crécy-Lagard et al., 1991) and a detached BЈ domain is encoded within the frw gene cluster of E. coli , the findings reported here may prove applicable to several fructose permease systems.
Construction of the mutant C112S FruBC permease, lacking the putative phosphorylation site residue (Cys 112 ) in the IIB domain, allowed us to show that this protein is inactive as expected (Lolkema and Robillard, 1992;Lengeler et al., 1994;Meins et al., 1993). They also allowed us to provide evidence that like the mannitol and glucose PTS permeases (see Introduction), the fructose permease is probably an oligomer. Because fructose utilization and the activity of the wild-type fructose permease were specifically inhibited by the presence of the C112S FruBC mutant permease, both in vivo and in vitro, whereas mannitol, glucitol, maltose, arabinose, and melibiose utilization was not affected, it must be concluded that high level expression of C112S FruBC did not interfere with membrane protein insertion generally but instead inhibited the permease directly. The fact that the C112S FruBC encoding gene was found to be epistatic over the wild-type gene (i.e. exhibited a negative-dominant phenotype both in vivo and in vitro) leads to the tentative conclusion that the individual subunits within the oligomeric Enzyme II are poorly active. These findings are in agreement with a previously reported observation with the X. campestris fructose permease (de Crécy-Lagard et al., 1991). In this case, a C-terminal truncation, eliminating the second half of the IIC domain, exhibited a negative dominant phenotype over the wild-type enzyme. However, they contrast with results reported in analogous experiments conducted with the mannitol permease of E. coli (van Weeghel et al., 1991;Boer et al., 1994).
Although the FruBC enzyme did not allow efficient fermentation of fructose at 0.2% sugar, the heterooligomeric protein (FruBЈBC/FruBC) in the merodiploid (TG1 (pBC)) apparently was capable of fermentation at low fructose concentrations as efficiently as the wild-type strain (TG1) (compare Tables I and  II and see description of Table II). It therefore appears that although the heterooligomer, C112SFruBC/FruBЈBC, exhibits depressed activity, the heterooligomer, FruBC/FruBЈBC, is as active as the wild-type homooligomer and substantially more active than the truncated homooligomer. All subunits must therefore possess active phosphorylation sites in their B domains in order for the permease to exhibit normal activity, but dominance of chromosomally encoded wild-type FruBЈBC over plasmid-encoded FruBC Fermentation responses of strain TG1 were recorded on fructose-MacConkey agar plates with fructose present at 0.2, 0.5, or 1% as indicated with or without 1 mM IPTG. Fermentation responses were recorded as follows: ϩϩ, wild-type response; ϩ, a less strong response; Ϯ, a weak response; Ϫ, no response. In addition to the responses recorded, all three strains were examined for their fermentation responses with and without IPTG on MacConkey fermentation plates containing 0.5% mannitol, arabinose, or maltose or 1% glucitol or melibiose. Wild-type fermentation responses were recorded for all three strains on all five carbon sources regardless of the presence of IPTG. Overexpression of the truncated permease lacking the BЈ domain was also examined at lower concentrations of fructose (0.15 and 0.10%) with and without IPTG. In no case did expression of the truncated permease noticeably inhibit the fructose fermentation response of strain TG1. Strains used were: none, TG1; pC112SBC, TG1 (pC112SBC); pBC, TG1 (pBC).  5. Competition between FruBBC and C112S FruBC for in vivo fructose transport. Plasmid pBC or pC112SBC was transferred into E. coli strain TG1 containing the wild-type chromosomal fruA gene, and cells were prepared for [ 14 C]fructose uptake measurements as described under "Experimental Procedures" after growth of strain TG1(pBC) without (Ç) or with (Ⅺ) IPTG or of strain TG1(pC112SBC) without (᭛) or with (E) IPTG. [ 14 C]Fructose uptake was measured as a function of time as described under "Experimental Procedures." FIG. 6. Negative dominance of the C112S FruBC mutant fructose permease over the wild-type permease in vitro. PEP-dependent [ 14 C]fructose phosphorylation by the lubrol-soluble truncated permease (FruBC) was measured in the absence (᭛) or in the presence (Ⅺ) of increasing amounts of mutant permease (C112S FruBC). The reaction volume was 50 l. The concentration of [ 14 C]fructose was 10 M (specific activity, 5 Ci/mol). The amount of FruBC was the same in all tubes (32 g), and increasing amounts of the C112S FruBC preparation in TDL buffer were added. In the control, the activity of FruBC was measured in the absence of C112S FruBC by the addition of corresponding amounts of TDL buffer. Purified Enzyme I and DTP were present in the assay mixture in excess. Essentially the same results were obtained when the FruBЈBC enzyme was used instead of the FruBC enzyme. The values represent the average of duplicate determinations. the BЈ domain may be required in only one of the subunits. This conclusion suggests that transport requires fully active B domains in the identical subunits of the Enzyme II complex but that association of DTP with the permease does not. The molecular basis for these interesting observations remains to be elucidated.