Regulation of the Maltose Transport System of Escherichia coli by the Glucose-specific Enzyme III of the Phosphoenolpyruvate-Sugar Phosphotransferase System

Maltose transport in Escherichia coli is regulated at the protein level by the glucose-specific enzyme III (III”‘) of the phosphoenolpyruvate-sugar phosphotransferase system, by a mechanism known as inducer exclusion. We have isolated and characterized four mutants in the maltose transport system, all of which are in m&K, which are resistant to inducer exclusion. The mutations in three of these mutants fall within the COOH-terminal domain of MalK and suggest the first reported function for this domain. Two of these are in a region which shows sequence similarity to lacy and melB, both of which are also regulated by IIIgl”, and thus may define a III g’c-binding domain. We have also reconstituted inducer exclusion in proteoliposomes made from membranes overexpressing the maltose permease. Maltose transport is inhibited by 50-60% when IIIgLc is included in the intravesicular space. The inhibition is due to a decrease in the V,,, of transport by a factor of 2. IIIg’” does not affect the coupling of ATP hydrolysis to maltose transport, since the ratio of ATP hydrolyzed/maltose transported remained constant in the presence and absence of III”‘. Finally, the Ki for IIIg’” was 40 FM, roughly the same as the in viuo concentration of III”“.

Maltose transport in Escherichia coli is regulated at the protein level by the glucose-specific enzyme III (III"') of the phosphoenolpyruvate-sugar phosphotransferase system, by a mechanism known as inducer exclusion.
We have isolated and characterized four mutants in the maltose transport system, all of which are in m&K, which are resistant to inducer exclusion. The mutations in three of these mutants fall within the COOH-terminal domain of MalK and suggest the first reported function for this domain. Two of these are in a region which shows sequence similarity to lacy and melB, both of which are also regulated by IIIgl", and thus may define a III g'c-binding domain.
We have also reconstituted inducer exclusion in proteoliposomes made from membranes overexpressing the maltose permease.
Maltose transport is inhibited by 50-60% when IIIgLc is included in the intravesicular space. The inhibition is due to a decrease in the V,,, of transport by a factor of 2. IIIg'" does not affect the coupling of ATP hydrolysis to maltose transport, since the ratio of ATP hydrolyzed/maltose transported remained constant in the presence and absence of III"'. Finally, the Ki for IIIg'" was 40 FM, roughly the same as the in viuo concentration of III"".
The phosphoenolpyruvate-sugar phosphotransferase system (PTS)' of Escherichia coli regulates the uptake of a number of non-PTS sugars, including maltose, by both transcriptional and post-transcriptional mechanisms (for a recent review, see . Transcriptional regulation of target operons involves both catabolite repression and inducer exclusion (Magasanik, 1970). Catabolite repression is largely mediated by regulatory interactions believed to involve the cyclic AMP biosynthetic enzyme, adenylate cyclase, and the central regulatory protein of the PTS, the glucose-specific enzyme III (IIIp'"). In the phosphorylated state, 111~'" is believed to function as an allosteric activator of adenylate cyclase. On the other hand, inducer exclusion involves direct allosteric inhibition, by the free (dephosphorylated) form of IIIK'", of the target permeases and catabolic enzymes that * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ' The abbreviations used are: PTS, phosphoenolpyruvate-sugar phosphotransferase system; IIF", glucose-specific Enzyme III; MBP, maltose-binding protein; aMeGlc, methyl-wglucoside.
generate endogenous inducers of non-PTS operons. Thus, when IIIg" is not phosphorylated, as is true in the wild-type, energy-proficient cell supplied with a PTS sugar in the extracellular medium, this regulatory protein binds to and inhibits the various target permeases and catabolic enzymes which generate cytoplasmic inducers. Under these same conditions, adenylate cyclase is in its inactive (or less active) form. Conversely, when III"" is phosphorylated, as is observed in the wild-type, energy-proficient cell when a PTS sugar is lacking from the extracellular medium, 1118" does not bind to the permeases and catabolic enzymes, and the inhibition of their activities is relieved. Under these conditions, adenylate cyclase is activated . Hence, cyclic AMP synthesis and the cytoplasmic accumulation of non-PTS inducers are coordinately regulated (Saier and Feucht, 1975). Demonstration of direct binding of III"" to the lactose permease and inhibition of transport activity in membrane vesicles have led to general acceptance of the model described above (Dills et al., 1982;Misko et al., 1987;Nelson et al., 1983;Osumi and Saier, 1982;Saier et al., 1983). The demonstration that IIIg'" interacts with glycerol kinase to inhibit its activity has also provided confirmation of this model (de Boer et al., 1986;Novotny et al., 1985;Postma et al., 1984).
Maltose and maltooligosaccharides are transported into E. coli by a binding protein-dependent transport system, consisting of a periplasmic maltose binding-protein (MBP or MalE), two hydrophobic, integral inner membrane proteins, MalF and MalG, and a peripheral inner membrane protein, MalK (for a review see Schwartz, 1987). The maltose regulon, composed of several different operons, is positively regulated by the product of the malT gene. To date, maltotriose is the only known inducer of the maltose regulon, and it binds to MalT as a coactivator. Several of the mu1 operons, including the two which encode the proteins of the transport system, are also subject to control by the CAMP/CAMP receptor protein (CAP) transcriptional activator complex. Recent work by Boos and colleagues further suggests that the mu1 regulon is osmotically regulated (Bukau et al., 1986). These investigators have identified a gene encoding a L&-like repressor protein, Mali, which acts as a repressor of at least some components of the regulon (Reid1 et al., 1989). MalK has also been implicated in the regulation of maltose regulon expression (Schwartz, 1987), but the mechanism by which it functions in regulation is as yet unknown. MalK is believed to be the energy-transducing protein of the maltose transport system. As initially suggested by sequence analyses, MalK and the MalK homologs in other binding protein-dependent transport systems appear to con-21005 shown to bind both ATP and ATP analogues (Hobson et al., 1984).* Recently, using well-defined cell-free systems, ATP has been shown to be the energy source driving transport via two of these permease systems, those specific for maltose and histidine (Ames, 1990;Ames et al., 1989;Bishop et al., 1989;Davidson and Nikaido, 1990;Dean et al., 1989;Dean et al., 1990;Higgins, 1990).
The consequences of PTS-mediated control of maltose transport in whole cells were first observed by Monod in the 1940s (Monod, 1942). However, due to the complexity of the maltose transport system, this regulation has not been studied extensively.
Several years ago we isolated mutations which mapped to the malK gene that rendered the maltose transport system resistant to inducer exclusion (Saier, 1985;Saier et al., 1978). Similar mutants have been obtained in the lactose permease, the melibiose permease and glycerol kinase (Novotny et al., 1985;Saier et al., 1978). In this report we describe the isolation and molecular characterization of such mutants in the maltose permease. We also utilize the recently developed technique of maltose transport reconstitution  to demonstrate the regulation of the maltose transport system by purified IIP"" and to gain information about the mechanism of regulation.  Table I. To isolate malK' mutants, LJ143 was spread onto maltose (0.2%) minimal plates containing 0.1% methyl-a-glucoside (cuMeGlc) and grown at 37 "C for 2 days. Colonies were restreaked on the same plates and subsequently streaked on a variety of plates to ensure that the mutations were specific for the maltose transport system; ptsH revertants and fruR mutants (Chin et al., 1987) fermented mannitol on EMB mannitol (1%) plates, and err mutants fermented lactose on EMB lactose (1%) plates containing 0.1% cuMeGlc.
The remaining mutations were mapped to malK and confirmed to be in malK by sequence analysis. malK' mutants were grown in medium 63 (Miller, 1972) containing 0.4% maltose and 0.1 pg/ml thiamine at 37 "C with aeration by shaking for whole cell experiments. Preparation of Proteoliposomes-Membranes containing overexpressed maltose transport proteins were prepared from HN597 containing pFG23 and pMRl1 grown in 2 x LB (20 g of tryptone, 20 g of yeast extract, 5 g of NaCl/liter) containing the appropriate antibiotics and induced for 3 h with 0.1 mM isopropyl-/3-D-thiogalactoside. Cells from 2 liters were washed in 0.1 M potassium phosphate buffer (KP,), pH 7.0, resuspended in 10 ml of the same, and passed twice through a French pressure cell at 10,000 psi. Whole cells were removed, and membranes were collected by centrifugation at 100,000 X g and stored in portions at -70 "C. Proteoliposomes were prepared as described

Isolation and Characterization of ma&V Mutants-To
facilitate demonstration of the regulatory interaction between the maltose permease and IIP" of the PTS, we isolated mutants in the maltose transport system that rendered it resistant to inducer exclusion. This was accomplished by selecting mutants from an E. coli ptsH mutant which fermented maltose in the presence of aMeGlc, a non-metabolizable glucose analog. Since III"" cannot be phosphorylated in the absence of HPr, only mutants in the maltose permease which are no longer sensitive to inhibition by unphosphorylated IIP'" will be able to grow. Since the ptsH315 mutation is slightly leaky, the addition of aMeGlc ensures that any phosphorylated IIP'" will be dephosphorylated.
We isolated four independent mutants all of which mapped to the malK gene.5 Whole cell transport assays designed to measure the extent of catabolite repression and inducer exclusion confirmed that the mutants we isolated were no longer sensitive to PTS-mediated inducer exclusion (Table II). Maltose transport was inhibited 46% by the presence of aMeGlc in strain LJ143, wild type for the maltose transport system, while the ma@? mutants showed no inhibition by aMeGlc. Addition of glucose to the cultures for several generations decreased the maltose uptake rate relative to controls by approximately .50% in each strain, indicating that catabolite repression was still operative.
We cloned the malK' mutant genes by polymerase chain reaction using primers directed against sequences both upstream and downstream of the malK gene. The polymerase chain reaction inserts were cloned into pKK223-2 under the control of the trc promoter. Transformants containing inserts were isolated in AD121 by selecting for growth on minimal maltose plates containing the appropriate antibiotic.
To ensure that we had cloned the malK' genes, we transformed ' M. Schwartz, personal communication.
LJ143AmalK with plasmids isolated in AD121 and screened for growth on minimal maltose plates containing 1 mM oMeGlc. 25 of 25 transformants carrying each of the cloned, mutant malK alleles grew in the presence of aMeGlc while none of the 25 transformants carrying the wild-type malK gene was capable of growth. The mutations were identified as described under "Experimental Procedures" and are summarized in Table III. We also cloned and sequenced the wildtype malK gene from LJ143 and found that the sequence was identical to that published (Dahl et al., 1989). Properties of the B. subtilis III%ike Domain Expressed in E. coli-We have found that the ZIP'"-like carboxyl terminus of the B. subtilis enzyme II"'" can function as an independent III""-like domain, during both glucose and sucrose uptake in B. subtilis and in the regulation of non-PTS permeases when transferred to E. coli.3,4 In order to characterize the regulatory interaction between IIP'" and the maltose permease, we transformed E. coli strains LJ288 and JLV86 (AptsHZcrr and err, respectively) with a plasmid carrying the IIP'"-like domain, ~Bs33.~ JLV86 alone does not grow on glucose, but when pBS33 is present, the cells can utilize this sugar. LJ288 ferments maltose, lactose, and melibiose as expected for a AptsHI strain lacking the entire pts operon including part of the err gene encoding IIIg'". However, when the cells are transformed with pBS33, they no longer ferment any of these sugars." These results show that the B. subtilis III@-like domain is able to complement an E. coli err mutant with respect to both glucose transport and regulation of other permeases in E. coli. Consequently, we could use the B. subtilis IIIg'" for the biochemical experiments described below.

Maltose Uptake in Reconstituted
Proteoliposomes-Proteoliposomes were prepared as described by Davidson and Nikaido (1990) from membranes isolated from a strain which overproduces the maltose transport proteins MalF, MalG, and MalK lo-20-fold.
Maltose transport was dependent on intravesicular ATP and extravesicular MBP. We were able to reconstitute inducer exclusion by adding purified IIF'" during the solubilization step and trapping it in the vesicles upon dilution (Fig. 1). The concentration of III"" typically used in the solubilization step was 15 PM, and the protein was then diluted to 0.6 PM upon formation of proteoliposome vesicles. Control samples containing either no added protein or soybean trypsin inhibitor (at the same concentration as III@") showed essentially the same transport rate. We attempted to reverse the inhibition by phosphorylating IIF" with Enzyme I, HPr, and phosphoenolpyruvate, as has been previously demonstrated for the lactose permease (Dills et al., 1982). We were able to relieve 40% of the BP"-mediated inhibition of maltose uptake by phosphorylation of III"'" (data not shown). Under the conditions of this experiment, IIIP" appeared to be largely phosphorylated as determined spectrophotometrically by the method of Meadow and Roseman (1982). Elimination of HPr and Enzyme I from the phosphorylation reaction prevented relief of inhibition, and the presence of phosphoenolpyruvate alone did not stimulate maltose transport.
Effect of IIP' on the Kinetics of Maltose Transport-Using equimolar concentrations of maltose and MBP, we followed the kinetics of maltose uptake into proteoliposomes made with or without IIIp'" (Fig. 2). When transport activity was studied as a function of the maltose-MBP concentration, the K, for liganded MBP was 9 ~.LM both in the presence and absence of IIP". It can be seen (Fig. 2) that the inhibition by IIP" is due to depression of the V,,, value, in this case by a factor of 2, from 5.4 nmol of maltose accumulated/min/mg were both added to 10 pM to measure maltose transport.
To determine the amount of ATP remaining in the proteoliposomes, proteoliposomes were diluted into buffer containing unlabeled maltose (10 PM), samples were withdrawn, and ATP levels were determined as described under "Experimental Procedures." The ATP hydrolysis assay was initiated by either the addition of buffer or MBP (10 PM), and the net transport-dependent hydrolysis is the difference between these two conditions. All assays and ATP determinations were performed in duplicate. Symbols: A, A, net ATP hydrolyzed, 0, 0, maltose accumulated.
protein in the absence of III"'" to 2.8 nmol/min/mg protein in its presence.
Effect of IIIR" on the Ratio of ATP Hydrolyzed/Maltose Transported-Since the MalK protein is thought to act as the energy-coupling protein of the maltose transport system, and since the mutations that we isolated mapped to within malK, we decided to look at the effect of IIP'" on the energetics of maltose transport. We had demonstrated previously that ATP is hydrolyzed concomitantly with maltose transport in both membrane vesicles and proteoliposomes Dean et al,, 1989). In both systems, the stoichiometry of ATP hydrolyzed/maltose transported ranges from approximately 1:l to lO:l, but it remains constant for a given vesicle or proteoliposome preparation. Fig. 3 shows the effect of the inclusion of IIP'" in proteoliposomes on the ratios of ATP hydrolysis to maltose transport.
The inclusion of IIP" decreased the amount of ATP hydrolyzed in parallel with the amount of maltose transported.
Thus, the ratio of ATP hydrolyzed to maltose transported remained constant with or without IIP", in this experiment at a ratio of about 15. Stoichiometry of ZZIgk-mediated Inhibition of Maltose Uptake-Proteoliposomes containing increasing amounts of IIF'" were prepared by varying the amount of IIIg'" added to solubilized membrane protein. The amount of III@ trapped within the washed proteoliposomes was determined and correlated with the amount of transport inhibition (Fig. 4). Using an intravesicular volume of 15 pl/mg protein , the internal concentration of IIIp'" was determined. The maximal inhibition that we could achieve was 65%, with half-maximal inhibition at 40 FM IIIg'" (12 pg/mg membrane protein).
Since the maltose permease constitutes approximately 35% of the membrane protein in these proteoliposomes and the molecular weight of the complex is 171,000 daltons,6 the inhibition by IIIp'" appears to be stoichiometric: the ratio of IIIp'" to maltose permease at 40 PM IIIp'" is -0.4% not far from the theoretical value of 0 The proteoliposomes were washed to remove-extravesicular III"'", and the amount of intravesicular IIIg" was determined as described.
A volume of 15 pl/mg of membrane protein was used as the internal volume of the proteoliposomes  to determine the concentration of III"'.
The control proteoliposomes made in the absence of III"" transported maltose at a rate of 4.62 nmol/min/mg membrane protein.
The intracellular concentration of III"" in E. coli is around 50 pM (Scholte et al., 1981), a value similar to that which gave half-maximal inhibition in our in vitro experiments.
Since the subunit stoichiometry of the maltose permease is FGKZ,' the concentration of III"'" that gave half-maximal inhibition (40 pM) would correspond to a ratio of 0.4 molecules IIIR"/makose permease in our system. In the wild-type E. coli cell, the ratio would be expected to be much higher since the maltose permease is expressed maximally at about 1000 copies/cell (Schwartz, 1987). Postma and colleagues (1988) have demonstrated binding, by cosedimentation, of IIIp'c to membranes containing overexpressed amounts of the maltose permease. While the degree of overexpression is not reported, if we assume that it is the same as in our system (5-10% of total membrane protein), the ratio between III"" and MalFGK* would be between 1 and 2 at an external IIP" concentration of 350 @M. This is in accordance with our results. Although we attempted to demonstrate binding between IIP" and MalFGK, by cross-linking using dithiobis (succinimidy1  propionate) and formaldehyde, we were unable to detect an interaction (data not shown). The mutations in malK that we isolated and mapped by sequencing occurred within two domains of the MalK protein.
The Ala to Thr change at residue 124 is between the two putative ATP-binding domains. While the region around the two ATP-binding domains shows significant homology to other energy coupling proteins of the binding protein-dependent permease systems, the region between the two sites shows very little similarity among the homologous proteins. The bold residues,  in the lactose permease sequence are also mutations which render lactose uptake independent of inducer exc1usion.r Identical residues are marked with a colon and conservative replacements are marked with a period. The alignment of Lacy and MelB is from Yazyu et al. (1984).
The remaining three mutations fall within the carboxyl terminus of MalK. MalK is approximately 100 amino acids larger than most of t,he other MalK homologs. These hundred residues are in the carboxyl terminus and are thought to comprise a separate domain, perhaps involved in regulation of maltose transport. The fact that three of the mutations lie within this region provides the first evidence that this domain may serve as the III""-binding domain.

DISCUSSION
The genetic results presented in this paper clearly suggest that the target for the control of the maltose transport system in E. coli by UP" is the product of the malK gene. The ability of the IIP"-like domain from B. subtitis to regulate maltose transport in E. coli both in viva and in vitro suggests that the regulatory properties of this protein have been conserved in these two organisms over 2 billion years of evolutionary time. Depression of the V,,,,, of transport resembles the inhibition of glycerol kinase and lactose permease by IIIg'". The maximal velocity of glycerol phosphorylation was depressed approximately 50% with respect to both ATP concentration and glycerol concentration, as was that of lactose transport by the lactose permease (Dills et al., 1982;Novotny et al., 1985). We were able to achieve 65% inhibition of maltose transport by the inclusion of III"" in the proteoliposomes. This is very similar to the inhibition seen in whole cells where the addition of crMeGlc to LJ143 causes a 46% inhibition of maltose uptake. Thus, we appear to have faithfully reconstituted in- The region around residues 278 and 284 (mutated in two of the mutants, Table III) was compared with the sequences of other proteins known to be subject to inducer exclusion (Fig.  5). Striking similarity was found between a region of Lacy and the intervening sequence of MalK, and some similarity was also found with GlpK and MelB, the genes encoding glycerol kinase and the melibiose transporter.
Noteworthy is the fact that the region of Lacy showing this sequence similarity is the central, putative cytoplasmically localized loop in which two independent mutations have been isolated that abolish inducer exclusion of lactose transport.7 The region of similarity in MelB is also within the central loop of this permease. Sequence similarity between very different proteins, all of which bind III@', suggests that these regions of the three permeases function as enzyme III"'-binding sites. It is anticipated that this will prove to provide an example of convergent evolution to accommodate a common regulatory mechanism, superimposed on evolutionarily divergent types of sugar transport proteins (Saier, 1990).